Index: projects/clang400-import/contrib/bsnmp/lib/snmpclient.c =================================================================== --- projects/clang400-import/contrib/bsnmp/lib/snmpclient.c (revision 311459) +++ projects/clang400-import/contrib/bsnmp/lib/snmpclient.c (revision 311460) @@ -1,1959 +1,1962 @@ /* * Copyright (c) 2004-2005 * Hartmut Brandt. * All rights reserved. * Copyright (c) 2001-2003 * Fraunhofer Institute for Open Communication Systems (FhG Fokus). * All rights reserved. * * Author: Harti Brandt * Kendy Kutzner * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $Begemot: bsnmp/lib/snmpclient.c,v 1.36 2005/10/06 07:14:58 brandt_h Exp $ * * Support functions for SNMP clients. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef HAVE_STDINT_H #include #elif defined(HAVE_INTTYPES_H) #include #endif #include #ifdef HAVE_ERR_H #include #endif #include "support.h" #include "asn1.h" #include "snmp.h" #include "snmpclient.h" #include "snmppriv.h" /* global context */ struct snmp_client snmp_client; /* List of all outstanding requests */ struct sent_pdu { int reqid; struct snmp_pdu *pdu; struct timeval time; u_int retrycount; snmp_send_cb_f callback; void *arg; void *timeout_id; LIST_ENTRY(sent_pdu) entries; }; LIST_HEAD(sent_pdu_list, sent_pdu); static struct sent_pdu_list sent_pdus; /* * Prototype table entry. All C-structure produced by the table function must * start with these two fields. This relies on the fact, that all TAILQ_ENTRY * are compatible with each other in the sense implied by ANSI-C. */ struct entry { TAILQ_ENTRY(entry) link; uint64_t found; }; TAILQ_HEAD(table, entry); /* * working list entry. This list is used to hold the Index part of the * table row's. The entry list and the work list parallel each other. */ struct work { TAILQ_ENTRY(work) link; struct asn_oid index; }; TAILQ_HEAD(worklist, work); /* * Table working data */ struct tabwork { const struct snmp_table *descr; struct table *table; struct worklist worklist; uint32_t last_change; int first; u_int iter; snmp_table_cb_f callback; void *arg; struct snmp_pdu pdu; }; /* * Set the error string */ static void seterr(struct snmp_client *sc, const char *fmt, ...) { va_list ap; va_start(ap, fmt); vsnprintf(sc->error, sizeof(sc->error), fmt, ap); va_end(ap); } /* * Free the entire table and work list. If table is NULL only the worklist * is freed. */ static void table_free(struct tabwork *work, int all) { struct work *w; struct entry *e; const struct snmp_table_entry *d; u_int i; while ((w = TAILQ_FIRST(&work->worklist)) != NULL) { TAILQ_REMOVE(&work->worklist, w, link); free(w); } if (all == 0) return; while ((e = TAILQ_FIRST(work->table)) != NULL) { for (i = 0; work->descr->entries[i].syntax != SNMP_SYNTAX_NULL; i++) { d = &work->descr->entries[i]; if (d->syntax == SNMP_SYNTAX_OCTETSTRING && (e->found & ((uint64_t)1 << i))) free(*(void **)(void *) ((u_char *)e + d->offset)); } TAILQ_REMOVE(work->table, e, link); free(e); } } /* * Find the correct table entry for the given variable. If non exists, * create one. */ static struct entry * table_find(struct tabwork *work, const struct asn_oid *var) { struct entry *e, *e1; struct work *w, *w1; u_int i, p, j; size_t len; u_char *ptr; struct asn_oid oid; /* get index */ asn_slice_oid(&oid, var, work->descr->table.len + 2, var->len); e = TAILQ_FIRST(work->table); w = TAILQ_FIRST(&work->worklist); while (e != NULL) { if (asn_compare_oid(&w->index, &oid) == 0) return (e); e = TAILQ_NEXT(e, link); w = TAILQ_NEXT(w, link); } /* Not found create new one */ if ((e = malloc(work->descr->entry_size)) == NULL) { seterr(&snmp_client, "no memory for table entry"); return (NULL); } if ((w = malloc(sizeof(*w))) == NULL) { seterr(&snmp_client, "no memory for table entry"); free(e); return (NULL); } w->index = oid; memset(e, 0, work->descr->entry_size); /* decode index */ p = work->descr->table.len + 2; for (i = 0; i < work->descr->index_size; i++) { switch (work->descr->entries[i].syntax) { case SNMP_SYNTAX_INTEGER: if (var->len < p + 1) { seterr(&snmp_client, "bad index: need integer"); goto err; } if (var->subs[p] > INT32_MAX) { seterr(&snmp_client, "bad index: integer too large"); goto err; } *(int32_t *)(void *)((u_char *)e + work->descr->entries[i].offset) = var->subs[p++]; break; case SNMP_SYNTAX_OCTETSTRING: if (var->len < p + 1) { seterr(&snmp_client, "bad index: need string length"); goto err; } len = var->subs[p++]; if (var->len < p + len) { seterr(&snmp_client, "bad index: string too short"); goto err; } if ((ptr = malloc(len + 1)) == NULL) { seterr(&snmp_client, "no memory for index string"); goto err; } for (j = 0; j < len; j++) { if (var->subs[p] > UCHAR_MAX) { seterr(&snmp_client, "bad index: char too large"); free(ptr); goto err; } ptr[j] = var->subs[p++]; } ptr[j] = '\0'; *(u_char **)(void *)((u_char *)e + work->descr->entries[i].offset) = ptr; *(size_t *)(void *)((u_char *)e + work->descr->entries[i].offset + sizeof(u_char *)) = len; break; case SNMP_SYNTAX_OID: if (var->len < p + 1) { seterr(&snmp_client, "bad index: need oid length"); goto err; } oid.len = var->subs[p++]; if (var->len < p + oid.len) { seterr(&snmp_client, "bad index: oid too short"); goto err; } for (j = 0; j < oid.len; j++) oid.subs[j] = var->subs[p++]; *(struct asn_oid *)(void *)((u_char *)e + work->descr->entries[i].offset) = oid; break; case SNMP_SYNTAX_IPADDRESS: if (var->len < p + 4) { seterr(&snmp_client, "bad index: need ip-address"); goto err; } for (j = 0; j < 4; j++) { if (var->subs[p] > 0xff) { seterr(&snmp_client, "bad index: ipaddress too large"); goto err; } ((u_char *)e + work->descr->entries[i].offset)[j] = var->subs[p++]; } break; case SNMP_SYNTAX_GAUGE: if (var->len < p + 1) { seterr(&snmp_client, "bad index: need unsigned"); goto err; } if (var->subs[p] > UINT32_MAX) { seterr(&snmp_client, "bad index: unsigned too large"); goto err; } *(uint32_t *)(void *)((u_char *)e + work->descr->entries[i].offset) = var->subs[p++]; break; case SNMP_SYNTAX_COUNTER: case SNMP_SYNTAX_TIMETICKS: case SNMP_SYNTAX_COUNTER64: case SNMP_SYNTAX_NULL: case SNMP_SYNTAX_NOSUCHOBJECT: case SNMP_SYNTAX_NOSUCHINSTANCE: case SNMP_SYNTAX_ENDOFMIBVIEW: abort(); } e->found |= (uint64_t)1 << i; } /* link into the correct place */ e1 = TAILQ_FIRST(work->table); w1 = TAILQ_FIRST(&work->worklist); while (e1 != NULL) { if (asn_compare_oid(&w1->index, &w->index) > 0) break; e1 = TAILQ_NEXT(e1, link); w1 = TAILQ_NEXT(w1, link); } if (e1 == NULL) { TAILQ_INSERT_TAIL(work->table, e, link); TAILQ_INSERT_TAIL(&work->worklist, w, link); } else { TAILQ_INSERT_BEFORE(e1, e, link); TAILQ_INSERT_BEFORE(w1, w, link); } return (e); err: /* * Error happend. Free all octet string index parts and the entry * itself. */ for (i = 0; i < work->descr->index_size; i++) { if (work->descr->entries[i].syntax == SNMP_SYNTAX_OCTETSTRING && (e->found & ((uint64_t)1 << i))) free(*(void **)(void *)((u_char *)e + work->descr->entries[i].offset)); } free(e); free(w); return (NULL); } /* * Assign the value */ static int table_value(const struct snmp_table *descr, struct entry *e, const struct snmp_value *b) { u_int i; u_char *ptr; for (i = descr->index_size; descr->entries[i].syntax != SNMP_SYNTAX_NULL; i++) if (descr->entries[i].subid == b->var.subs[descr->table.len + 1]) break; if (descr->entries[i].syntax == SNMP_SYNTAX_NULL) return (0); /* check syntax */ if (b->syntax != descr->entries[i].syntax) { seterr(&snmp_client, "bad syntax (%u instead of %u)", b->syntax, descr->entries[i].syntax); return (-1); } switch (b->syntax) { case SNMP_SYNTAX_INTEGER: *(int32_t *)(void *)((u_char *)e + descr->entries[i].offset) = b->v.integer; break; case SNMP_SYNTAX_OCTETSTRING: if ((ptr = malloc(b->v.octetstring.len + 1)) == NULL) { seterr(&snmp_client, "no memory for string"); return (-1); } memcpy(ptr, b->v.octetstring.octets, b->v.octetstring.len); ptr[b->v.octetstring.len] = '\0'; *(u_char **)(void *)((u_char *)e + descr->entries[i].offset) = ptr; *(size_t *)(void *)((u_char *)e + descr->entries[i].offset + sizeof(u_char *)) = b->v.octetstring.len; break; case SNMP_SYNTAX_OID: *(struct asn_oid *)(void *)((u_char *)e + descr->entries[i].offset) = b->v.oid; break; case SNMP_SYNTAX_IPADDRESS: memcpy((u_char *)e + descr->entries[i].offset, b->v.ipaddress, 4); break; case SNMP_SYNTAX_COUNTER: case SNMP_SYNTAX_GAUGE: case SNMP_SYNTAX_TIMETICKS: *(uint32_t *)(void *)((u_char *)e + descr->entries[i].offset) = b->v.uint32; break; case SNMP_SYNTAX_COUNTER64: *(uint64_t *)(void *)((u_char *)e + descr->entries[i].offset) = b->v.counter64; break; case SNMP_SYNTAX_NULL: case SNMP_SYNTAX_NOSUCHOBJECT: case SNMP_SYNTAX_NOSUCHINSTANCE: case SNMP_SYNTAX_ENDOFMIBVIEW: abort(); } e->found |= (uint64_t)1 << i; return (0); } /* * Initialize the first PDU to send */ static void table_init_pdu(const struct snmp_table *descr, struct snmp_pdu *pdu) { if (snmp_client.version == SNMP_V1) snmp_pdu_create(pdu, SNMP_PDU_GETNEXT); else { snmp_pdu_create(pdu, SNMP_PDU_GETBULK); pdu->error_index = 10; } if (descr->last_change.len != 0) { pdu->bindings[pdu->nbindings].syntax = SNMP_SYNTAX_NULL; pdu->bindings[pdu->nbindings].var = descr->last_change; pdu->nbindings++; if (pdu->version != SNMP_V1) pdu->error_status++; } pdu->bindings[pdu->nbindings].var = descr->table; pdu->bindings[pdu->nbindings].syntax = SNMP_SYNTAX_NULL; pdu->nbindings++; } /* * Return code: * 0 - End Of Table * -1 - Error * -2 - Last change changed - again * +1 - ok, continue */ static int table_check_response(struct tabwork *work, const struct snmp_pdu *resp) { const struct snmp_value *b; struct entry *e; if (resp->error_status != SNMP_ERR_NOERROR) { if (snmp_client.version == SNMP_V1 && resp->error_status == SNMP_ERR_NOSUCHNAME && resp->error_index == (work->descr->last_change.len == 0) ? 1 : 2) /* EOT */ return (0); /* Error */ seterr(&snmp_client, "error fetching table: status=%d index=%d", resp->error_status, resp->error_index); return (-1); } for (b = resp->bindings; b < resp->bindings + resp->nbindings; b++) { if (work->descr->last_change.len != 0 && b == resp->bindings) { if (!asn_is_suboid(&work->descr->last_change, &b->var) || b->var.len != work->descr->last_change.len + 1 || b->var.subs[work->descr->last_change.len] != 0) { seterr(&snmp_client, "last_change: bad response"); return (-1); } if (b->syntax != SNMP_SYNTAX_TIMETICKS) { seterr(&snmp_client, "last_change: bad syntax %u", b->syntax); return (-1); } if (work->first) { work->last_change = b->v.uint32; work->first = 0; } else if (work->last_change != b->v.uint32) { if (++work->iter >= work->descr->max_iter) { seterr(&snmp_client, "max iteration count exceeded"); return (-1); } table_free(work, 1); return (-2); } continue; } if (!asn_is_suboid(&work->descr->table, &b->var) || b->syntax == SNMP_SYNTAX_ENDOFMIBVIEW) return (0); if ((e = table_find(work, &b->var)) == NULL) return (-1); if (table_value(work->descr, e, b)) return (-1); } return (+1); } /* * Check table consistency */ static int table_check_cons(struct tabwork *work) { struct entry *e; TAILQ_FOREACH(e, work->table, link) if ((e->found & work->descr->req_mask) != work->descr->req_mask) { if (work->descr->last_change.len == 0) { if (++work->iter >= work->descr->max_iter) { seterr(&snmp_client, "max iteration count exceeded"); return (-1); } return (-2); } seterr(&snmp_client, "inconsistency detected %llx %llx", e->found, work->descr->req_mask); return (-1); } return (0); } /* * Fetch a table. Returns 0 if ok, -1 on errors. * This is the synchronous variant. */ int snmp_table_fetch(const struct snmp_table *descr, void *list) { struct snmp_pdu resp; struct tabwork work; int ret; work.descr = descr; work.table = (struct table *)list; work.iter = 0; TAILQ_INIT(work.table); TAILQ_INIT(&work.worklist); work.callback = NULL; work.arg = NULL; again: /* * We come to this label when the code detects that the table * has changed while fetching it. */ work.first = 1; work.last_change = 0; table_init_pdu(descr, &work.pdu); for (;;) { if (snmp_dialog(&work.pdu, &resp)) { table_free(&work, 1); return (-1); } if ((ret = table_check_response(&work, &resp)) == 0) { snmp_pdu_free(&resp); break; } if (ret == -1) { snmp_pdu_free(&resp); table_free(&work, 1); return (-1); } if (ret == -2) { snmp_pdu_free(&resp); goto again; } work.pdu.bindings[work.pdu.nbindings - 1].var = resp.bindings[resp.nbindings - 1].var; snmp_pdu_free(&resp); } if ((ret = table_check_cons(&work)) == -1) { table_free(&work, 1); return (-1); } if (ret == -2) { table_free(&work, 1); goto again; } /* * Free index list */ table_free(&work, 0); return (0); } /* * Callback for table */ static void table_cb(struct snmp_pdu *req __unused, struct snmp_pdu *resp, void *arg) { struct tabwork *work = arg; int ret; if (resp == NULL) { /* timeout */ seterr(&snmp_client, "no response to fetch table request"); table_free(work, 1); work->callback(work->table, work->arg, -1); free(work); return; } if ((ret = table_check_response(work, resp)) == 0) { /* EOT */ snmp_pdu_free(resp); if ((ret = table_check_cons(work)) == -1) { /* error happend */ table_free(work, 1); work->callback(work->table, work->arg, -1); free(work); return; } if (ret == -2) { /* restart */ again: table_free(work, 1); work->first = 1; work->last_change = 0; table_init_pdu(work->descr, &work->pdu); if (snmp_pdu_send(&work->pdu, table_cb, work) == -1) { work->callback(work->table, work->arg, -1); free(work); return; } return; } /* * Free index list */ table_free(work, 0); work->callback(work->table, work->arg, 0); free(work); return; } if (ret == -1) { /* error */ snmp_pdu_free(resp); table_free(work, 1); work->callback(work->table, work->arg, -1); free(work); return; } if (ret == -2) { /* again */ snmp_pdu_free(resp); goto again; } /* next part */ work->pdu.bindings[work->pdu.nbindings - 1].var = resp->bindings[resp->nbindings - 1].var; snmp_pdu_free(resp); if (snmp_pdu_send(&work->pdu, table_cb, work) == -1) { table_free(work, 1); work->callback(work->table, work->arg, -1); free(work); return; } } int snmp_table_fetch_async(const struct snmp_table *descr, void *list, snmp_table_cb_f func, void *arg) { struct tabwork *work; if ((work = malloc(sizeof(*work))) == NULL) { seterr(&snmp_client, "%s", strerror(errno)); return (-1); } work->descr = descr; work->table = (struct table *)list; work->iter = 0; TAILQ_INIT(work->table); TAILQ_INIT(&work->worklist); work->callback = func; work->arg = arg; /* * Start by sending the first PDU */ work->first = 1; work->last_change = 0; table_init_pdu(descr, &work->pdu); - if (snmp_pdu_send(&work->pdu, table_cb, work) == -1) + if (snmp_pdu_send(&work->pdu, table_cb, work) == -1) { + free(work); + work = NULL; return (-1); + } return (0); } /* * Append an index to an oid */ int snmp_oid_append(struct asn_oid *oid, const char *fmt, ...) { va_list va; int size; char *nextptr; const u_char *str; size_t len; struct in_addr ina; int ret; va_start(va, fmt); size = 0; ret = 0; while (*fmt != '\0') { switch (*fmt++) { case 'i': /* just an integer more */ if (oid->len + 1 > ASN_MAXOIDLEN) { warnx("%s: OID too long for integer", __func__); ret = -1; break; } oid->subs[oid->len++] = va_arg(va, asn_subid_t); break; case 'a': /* append an IP address */ if (oid->len + 4 > ASN_MAXOIDLEN) { warnx("%s: OID too long for ip-addr", __func__); ret = -1; break; } ina = va_arg(va, struct in_addr); ina.s_addr = ntohl(ina.s_addr); oid->subs[oid->len++] = (ina.s_addr >> 24) & 0xff; oid->subs[oid->len++] = (ina.s_addr >> 16) & 0xff; oid->subs[oid->len++] = (ina.s_addr >> 8) & 0xff; oid->subs[oid->len++] = (ina.s_addr >> 0) & 0xff; break; case 's': /* append a null-terminated string, * length is computed */ str = (const u_char *)va_arg(va, const char *); len = strlen((const char *)str); if (oid->len + len + 1 > ASN_MAXOIDLEN) { warnx("%s: OID too long for string", __func__); ret = -1; break; } oid->subs[oid->len++] = len; while (len--) oid->subs[oid->len++] = *str++; break; case '(': /* the integer value between ( and ) is stored * in size */ size = strtol(fmt, &nextptr, 10); if (*nextptr != ')') abort(); fmt = ++nextptr; break; case 'b': /* append `size` characters */ str = (const u_char *)va_arg(va, const char *); if (oid->len + size > ASN_MAXOIDLEN) { warnx("%s: OID too long for string", __func__); ret = -1; break; } while (size--) oid->subs[oid->len++] = *str++; break; case 'c': /* get size and the octets from the arguments */ size = va_arg(va, size_t); str = va_arg(va, const u_char *); if (oid->len + size + 1 > ASN_MAXOIDLEN) { warnx("%s: OID too long for string", __func__); ret = -1; break; } oid->subs[oid->len++] = size; while (size--) oid->subs[oid->len++] = *str++; break; default: abort(); } } va_end(va); return (ret); } /* * Initialize a client structure */ void snmp_client_init(struct snmp_client *c) { memset(c, 0, sizeof(*c)); c->version = SNMP_V2c; c->trans = SNMP_TRANS_UDP; c->chost = NULL; c->cport = NULL; strcpy(c->read_community, "public"); strcpy(c->write_community, "private"); c->security_model = SNMP_SECMODEL_USM; strcpy(c->cname, ""); c->timeout.tv_sec = 3; c->timeout.tv_usec = 0; c->retries = 3; c->dump_pdus = 0; c->txbuflen = c->rxbuflen = 10000; c->fd = -1; c->max_reqid = INT32_MAX; c->min_reqid = 0; c->next_reqid = 0; c->engine.max_msg_size = 1500; /* XXX */ } /* * Open UDP client socket */ static int open_client_udp(const char *host, const char *port) { int error; char *ptr; struct addrinfo hints, *res0, *res; /* copy host- and portname */ if (snmp_client.chost == NULL) { if ((snmp_client.chost = malloc(1 + sizeof(DEFAULT_HOST))) == NULL) { seterr(&snmp_client, "%s", strerror(errno)); return (-1); } strcpy(snmp_client.chost, DEFAULT_HOST); } if (host != NULL) { if ((ptr = malloc(1 + strlen(host))) == NULL) { seterr(&snmp_client, "%s", strerror(errno)); return (-1); } free(snmp_client.chost); snmp_client.chost = ptr; strcpy(snmp_client.chost, host); } if (snmp_client.cport == NULL) { if ((snmp_client.cport = malloc(1 + sizeof(DEFAULT_PORT))) == NULL) { seterr(&snmp_client, "%s", strerror(errno)); return (-1); } strcpy(snmp_client.cport, DEFAULT_PORT); } if (port != NULL) { if ((ptr = malloc(1 + strlen(port))) == NULL) { seterr(&snmp_client, "%s", strerror(errno)); return (-1); } free(snmp_client.cport); snmp_client.cport = ptr; strcpy(snmp_client.cport, port); } /* open connection */ memset(&hints, 0, sizeof(hints)); hints.ai_flags = AI_CANONNAME; hints.ai_family = AF_INET; hints.ai_socktype = SOCK_DGRAM; hints.ai_protocol = 0; error = getaddrinfo(snmp_client.chost, snmp_client.cport, &hints, &res0); if (error != 0) { seterr(&snmp_client, "%s: %s", snmp_client.chost, gai_strerror(error)); return (-1); } res = res0; for (;;) { if ((snmp_client.fd = socket(res->ai_family, res->ai_socktype, res->ai_protocol)) == -1) { if ((res = res->ai_next) == NULL) { seterr(&snmp_client, "%s", strerror(errno)); freeaddrinfo(res0); return (-1); } } else if (connect(snmp_client.fd, res->ai_addr, res->ai_addrlen) == -1) { if ((res = res->ai_next) == NULL) { seterr(&snmp_client, "%s", strerror(errno)); freeaddrinfo(res0); return (-1); } } else break; } freeaddrinfo(res0); return (0); } static void remove_local(void) { (void)remove(snmp_client.local_path); } /* * Open local socket */ static int open_client_local(const char *path) { struct sockaddr_un sa; char *ptr; int stype; if (snmp_client.chost == NULL) { if ((snmp_client.chost = malloc(1 + sizeof(DEFAULT_LOCAL))) == NULL) { seterr(&snmp_client, "%s", strerror(errno)); return (-1); } strcpy(snmp_client.chost, DEFAULT_LOCAL); } if (path != NULL) { if ((ptr = malloc(1 + strlen(path))) == NULL) { seterr(&snmp_client, "%s", strerror(errno)); return (-1); } free(snmp_client.chost); snmp_client.chost = ptr; strcpy(snmp_client.chost, path); } if (snmp_client.trans == SNMP_TRANS_LOC_DGRAM) stype = SOCK_DGRAM; else stype = SOCK_STREAM; if ((snmp_client.fd = socket(PF_LOCAL, stype, 0)) == -1) { seterr(&snmp_client, "%s", strerror(errno)); return (-1); } snprintf(snmp_client.local_path, sizeof(snmp_client.local_path), "%s", SNMP_LOCAL_PATH); if (mktemp(snmp_client.local_path) == NULL) { seterr(&snmp_client, "%s", strerror(errno)); (void)close(snmp_client.fd); snmp_client.fd = -1; return (-1); } sa.sun_family = AF_LOCAL; sa.sun_len = sizeof(sa); strcpy(sa.sun_path, snmp_client.local_path); if (bind(snmp_client.fd, (struct sockaddr *)&sa, sizeof(sa)) == -1) { seterr(&snmp_client, "%s", strerror(errno)); (void)close(snmp_client.fd); snmp_client.fd = -1; (void)remove(snmp_client.local_path); return (-1); } atexit(remove_local); sa.sun_family = AF_LOCAL; sa.sun_len = offsetof(struct sockaddr_un, sun_path) + strlen(snmp_client.chost); strncpy(sa.sun_path, snmp_client.chost, sizeof(sa.sun_path) - 1); sa.sun_path[sizeof(sa.sun_path) - 1] = '\0'; if (connect(snmp_client.fd, (struct sockaddr *)&sa, sa.sun_len) == -1) { seterr(&snmp_client, "%s", strerror(errno)); (void)close(snmp_client.fd); snmp_client.fd = -1; (void)remove(snmp_client.local_path); return (-1); } return (0); } /* * SNMP_OPEN */ int snmp_open(const char *host, const char *port, const char *readcomm, const char *writecomm) { struct timeval tout; /* still open ? */ if (snmp_client.fd != -1) { errno = EBUSY; seterr(&snmp_client, "%s", strerror(errno)); return (-1); } /* copy community strings */ if (readcomm != NULL) strlcpy(snmp_client.read_community, readcomm, sizeof(snmp_client.read_community)); if (writecomm != NULL) strlcpy(snmp_client.write_community, writecomm, sizeof(snmp_client.write_community)); switch (snmp_client.trans) { case SNMP_TRANS_UDP: if (open_client_udp(host, port)) return (-1); break; case SNMP_TRANS_LOC_DGRAM: case SNMP_TRANS_LOC_STREAM: if (open_client_local(host)) return (-1); break; default: seterr(&snmp_client, "bad transport mapping"); return (-1); } tout.tv_sec = 0; tout.tv_usec = 0; if (setsockopt(snmp_client.fd, SOL_SOCKET, SO_SNDTIMEO, &tout, sizeof(struct timeval)) == -1) { seterr(&snmp_client, "%s", strerror(errno)); (void)close(snmp_client.fd); snmp_client.fd = -1; if (snmp_client.local_path[0] != '\0') (void)remove(snmp_client.local_path); return (-1); } /* initialize list */ LIST_INIT(&sent_pdus); return (0); } /* * SNMP_CLOSE * * closes connection to snmp server * - function cannot fail * - clears connection * - clears list of sent pdus * * input: * void * return: * void */ void snmp_close(void) { struct sent_pdu *p1; if (snmp_client.fd != -1) { (void)close(snmp_client.fd); snmp_client.fd = -1; if (snmp_client.local_path[0] != '\0') (void)remove(snmp_client.local_path); } while(!LIST_EMPTY(&sent_pdus)){ p1 = LIST_FIRST(&sent_pdus); if (p1->timeout_id != NULL) snmp_client.timeout_stop(p1->timeout_id); LIST_REMOVE(p1, entries); free(p1); } free(snmp_client.chost); free(snmp_client.cport); } /* * initialize a snmp_pdu structure */ void snmp_pdu_create(struct snmp_pdu *pdu, u_int op) { memset(pdu, 0, sizeof(struct snmp_pdu)); if (op == SNMP_PDU_SET) strlcpy(pdu->community, snmp_client.write_community, sizeof(pdu->community)); else strlcpy(pdu->community, snmp_client.read_community, sizeof(pdu->community)); pdu->type = op; pdu->version = snmp_client.version; pdu->error_status = 0; pdu->error_index = 0; pdu->nbindings = 0; if (snmp_client.version != SNMP_V3) return; pdu->identifier = ++snmp_client.identifier; pdu->engine.max_msg_size = snmp_client.engine.max_msg_size; pdu->flags = 0; pdu->security_model = snmp_client.security_model; if (snmp_client.security_model == SNMP_SECMODEL_USM) { memcpy(&pdu->engine, &snmp_client.engine, sizeof(pdu->engine)); memcpy(&pdu->user, &snmp_client.user, sizeof(pdu->user)); snmp_pdu_init_secparams(pdu); } else seterr(&snmp_client, "unknown security model"); if (snmp_client.clen > 0) { memcpy(pdu->context_engine, snmp_client.cengine, snmp_client.clen); pdu->context_engine_len = snmp_client.clen; } else { memcpy(pdu->context_engine, snmp_client.engine.engine_id, snmp_client.engine.engine_len); pdu->context_engine_len = snmp_client.engine.engine_len; } strlcpy(pdu->context_name, snmp_client.cname, sizeof(pdu->context_name)); } /* add pairs of (struct asn_oid, enum snmp_syntax) to an existing pdu */ /* added 10/04/02 by kek: check for MAX_BINDINGS */ int snmp_add_binding(struct snmp_v1_pdu *pdu, ...) { va_list ap; const struct asn_oid *oid; u_int ret; va_start(ap, pdu); ret = pdu->nbindings; while ((oid = va_arg(ap, const struct asn_oid *)) != NULL) { if (pdu->nbindings >= SNMP_MAX_BINDINGS){ va_end(ap); return (-1); } pdu->bindings[pdu->nbindings].var = *oid; pdu->bindings[pdu->nbindings].syntax = va_arg(ap, enum snmp_syntax); pdu->nbindings++; } va_end(ap); return (ret); } static int32_t snmp_next_reqid(struct snmp_client * c) { int32_t i; i = c->next_reqid; if (c->next_reqid >= c->max_reqid) c->next_reqid = c->min_reqid; else c->next_reqid++; return (i); } /* * Send request and return request id. */ static int32_t snmp_send_packet(struct snmp_pdu * pdu) { u_char *buf; struct asn_buf b; ssize_t ret; if ((buf = calloc(1, snmp_client.txbuflen)) == NULL) { seterr(&snmp_client, "%s", strerror(errno)); return (-1); } pdu->request_id = snmp_next_reqid(&snmp_client); b.asn_ptr = buf; b.asn_len = snmp_client.txbuflen; if (snmp_pdu_encode(pdu, &b)) { seterr(&snmp_client, "%s", strerror(errno)); free(buf); return (-1); } if (snmp_client.dump_pdus) snmp_pdu_dump(pdu); if ((ret = send(snmp_client.fd, buf, b.asn_ptr - buf, 0)) == -1) { seterr(&snmp_client, "%s", strerror(errno)); free(buf); return (-1); } free(buf); return (pdu->request_id); } /* * to be called when a snmp request timed out */ static void snmp_timeout(void * listentry_ptr) { struct sent_pdu *listentry = listentry_ptr; #if 0 warnx("snmp request %i timed out, attempt (%i/%i)", listentry->reqid, listentry->retrycount, snmp_client.retries); #endif listentry->retrycount++; if (listentry->retrycount > snmp_client.retries) { /* there is no answer at all */ LIST_REMOVE(listentry, entries); listentry->callback(listentry->pdu, NULL, listentry->arg); free(listentry); } else { /* try again */ /* new request with new request ID */ listentry->reqid = snmp_send_packet(listentry->pdu); listentry->timeout_id = snmp_client.timeout_start(&snmp_client.timeout, snmp_timeout, listentry); } } int32_t snmp_pdu_send(struct snmp_pdu *pdu, snmp_send_cb_f func, void *arg) { struct sent_pdu *listentry; int32_t id; if ((listentry = malloc(sizeof(struct sent_pdu))) == NULL) { seterr(&snmp_client, "%s", strerror(errno)); return (-1); } /* here we really send */ if ((id = snmp_send_packet(pdu)) == -1) { free(listentry); return (-1); } /* add entry to list of sent PDUs */ listentry->pdu = pdu; if (gettimeofday(&listentry->time, NULL) == -1) warn("gettimeofday() failed"); listentry->reqid = pdu->request_id; listentry->callback = func; listentry->arg = arg; listentry->retrycount=1; listentry->timeout_id = snmp_client.timeout_start(&snmp_client.timeout, snmp_timeout, listentry); LIST_INSERT_HEAD(&sent_pdus, listentry, entries); return (id); } /* * Receive an SNMP packet. * * tv controls how we wait for a packet: if tv is a NULL pointer, * the receive blocks forever, if tv points to a structure with all * members 0 the socket is polled, in all other cases tv specifies the * maximum time to wait for a packet. * * Return: * -1 on errors * 0 on timeout * +1 if packet received */ static int snmp_receive_packet(struct snmp_pdu *pdu, struct timeval *tv) { int dopoll, setpoll; int flags; int saved_errno; u_char *buf; int ret; struct asn_buf abuf; int32_t ip; #ifdef bsdi int optlen; #else socklen_t optlen; #endif if ((buf = calloc(1, snmp_client.rxbuflen)) == NULL) { seterr(&snmp_client, "%s", strerror(errno)); return (-1); } dopoll = setpoll = 0; flags = 0; if (tv != NULL) { /* poll or timeout */ if (tv->tv_sec != 0 || tv->tv_usec != 0) { /* wait with timeout */ if (setsockopt(snmp_client.fd, SOL_SOCKET, SO_RCVTIMEO, tv, sizeof(*tv)) == -1) { seterr(&snmp_client, "setsockopt: %s", strerror(errno)); free(buf); return (-1); } optlen = sizeof(*tv); if (getsockopt(snmp_client.fd, SOL_SOCKET, SO_RCVTIMEO, tv, &optlen) == -1) { seterr(&snmp_client, "getsockopt: %s", strerror(errno)); free(buf); return (-1); } /* at this point tv_sec and tv_usec may appear * as 0. This happens for timeouts lesser than * the clock granularity. The kernel rounds these to * 0 and this would result in a blocking receive. * Instead of an else we check tv_sec and tv_usec * again below and if this rounding happens, * switch to a polling receive. */ } if (tv->tv_sec == 0 && tv->tv_usec == 0) { /* poll */ dopoll = 1; if ((flags = fcntl(snmp_client.fd, F_GETFL, 0)) == -1) { seterr(&snmp_client, "fcntl: %s", strerror(errno)); free(buf); return (-1); } if (!(flags & O_NONBLOCK)) { setpoll = 1; flags |= O_NONBLOCK; if (fcntl(snmp_client.fd, F_SETFL, flags) == -1) { seterr(&snmp_client, "fcntl: %s", strerror(errno)); free(buf); return (-1); } } } } ret = recv(snmp_client.fd, buf, snmp_client.rxbuflen, 0); saved_errno = errno; if (tv != NULL) { if (dopoll) { if (setpoll) { flags &= ~O_NONBLOCK; (void)fcntl(snmp_client.fd, F_SETFL, flags); } } else { tv->tv_sec = 0; tv->tv_usec = 0; (void)setsockopt(snmp_client.fd, SOL_SOCKET, SO_RCVTIMEO, tv, sizeof(*tv)); } } if (ret == -1) { free(buf); if (errno == EAGAIN || errno == EWOULDBLOCK) return (0); seterr(&snmp_client, "recv: %s", strerror(saved_errno)); return (-1); } if (ret == 0) { /* this happens when we have a streaming socket and the * remote side has closed it */ free(buf); seterr(&snmp_client, "recv: socket closed by peer"); errno = EPIPE; return (-1); } abuf.asn_ptr = buf; abuf.asn_len = ret; memset(pdu, 0, sizeof(*pdu)); if (snmp_client.security_model == SNMP_SECMODEL_USM) { memcpy(&pdu->engine, &snmp_client.engine, sizeof(pdu->engine)); memcpy(&pdu->user, &snmp_client.user, sizeof(pdu->user)); snmp_pdu_init_secparams(pdu); } if (SNMP_CODE_OK != (ret = snmp_pdu_decode(&abuf, pdu, &ip))) { seterr(&snmp_client, "snmp_decode_pdu: failed %d", ret); free(buf); return (-1); } free(buf); if (snmp_client.dump_pdus) snmp_pdu_dump(pdu); snmp_client.engine.engine_time = pdu->engine.engine_time; snmp_client.engine.engine_boots = pdu->engine.engine_boots; return (+1); } static int snmp_deliver_packet(struct snmp_pdu * resp) { struct sent_pdu *listentry; if (resp->type != SNMP_PDU_RESPONSE) { warn("ignoring snmp pdu %u", resp->type); return (-1); } LIST_FOREACH(listentry, &sent_pdus, entries) if (listentry->reqid == resp->request_id) break; if (listentry == NULL) return (-1); LIST_REMOVE(listentry, entries); listentry->callback(listentry->pdu, resp, listentry->arg); snmp_client.timeout_stop(listentry->timeout_id); free(listentry); return (0); } int snmp_receive(int blocking) { int ret; struct timeval tv; struct snmp_pdu * resp; memset(&tv, 0, sizeof(tv)); resp = malloc(sizeof(struct snmp_pdu)); if (resp == NULL) { seterr(&snmp_client, "no memory for returning PDU"); return (-1) ; } if ((ret = snmp_receive_packet(resp, blocking ? NULL : &tv)) <= 0) { free(resp); return (ret); } ret = snmp_deliver_packet(resp); snmp_pdu_free(resp); free(resp); return (ret); } /* * Check a GETNEXT response. Here we have three possible outcomes: -1 an * unexpected error happened. +1 response is ok and is within the table 0 * response is ok, but is behind the table or error is NOSUCHNAME. The req * should point to a template PDU which contains the base OIDs and the * syntaxes. This is really only useful to sweep non-sparse tables. */ static int ok_getnext(const struct snmp_pdu * req, const struct snmp_pdu * resp) { u_int i; if (resp->version != req->version) { warnx("SNMP GETNEXT: response has wrong version"); return (-1); } if (resp->error_status == SNMP_ERR_NOSUCHNAME) return (0); if (resp->error_status != SNMP_ERR_NOERROR) { warnx("SNMP GETNEXT: error %d", resp->error_status); return (-1); } if (resp->nbindings != req->nbindings) { warnx("SNMP GETNEXT: bad number of bindings in response"); return (-1); } for (i = 0; i < req->nbindings; i++) { if (!asn_is_suboid(&req->bindings[i].var, &resp->bindings[i].var)) { if (i != 0) warnx("SNMP GETNEXT: inconsistent table " "response"); return (0); } if (resp->version != SNMP_V1 && resp->bindings[i].syntax == SNMP_SYNTAX_ENDOFMIBVIEW) return (0); if (resp->bindings[i].syntax != req->bindings[i].syntax) { warnx("SNMP GETNEXT: bad syntax in response"); return (0); } } return (1); } /* * Check a GET response. Here we have three possible outcomes: -1 an * unexpected error happened. +1 response is ok. 0 NOSUCHNAME The req should * point to a template PDU which contains the OIDs and the syntaxes. This * is only useful for SNMPv1 or single object GETS. */ static int ok_get(const struct snmp_pdu * req, const struct snmp_pdu * resp) { u_int i; if (resp->version != req->version) { warnx("SNMP GET: response has wrong version"); return (-1); } if (resp->error_status == SNMP_ERR_NOSUCHNAME) return (0); if (resp->error_status != SNMP_ERR_NOERROR) { warnx("SNMP GET: error %d", resp->error_status); return (-1); } if (resp->nbindings != req->nbindings) { warnx("SNMP GET: bad number of bindings in response"); return (-1); } for (i = 0; i < req->nbindings; i++) { if (asn_compare_oid(&req->bindings[i].var, &resp->bindings[i].var) != 0) { warnx("SNMP GET: bad OID in response"); return (-1); } if (snmp_client.version != SNMP_V1 && (resp->bindings[i].syntax == SNMP_SYNTAX_NOSUCHOBJECT || resp->bindings[i].syntax == SNMP_SYNTAX_NOSUCHINSTANCE)) return (0); if (resp->bindings[i].syntax != req->bindings[i].syntax) { warnx("SNMP GET: bad syntax in response"); return (-1); } } return (1); } /* * Check the response to a SET PDU. We check: - the error status must be 0 - * the number of bindings must be equal in response and request - the * syntaxes must be the same in response and request - the OIDs must be the * same in response and request */ static int ok_set(const struct snmp_pdu * req, const struct snmp_pdu * resp) { u_int i; if (resp->version != req->version) { warnx("SNMP SET: response has wrong version"); return (-1); } if (resp->error_status == SNMP_ERR_NOSUCHNAME) { warnx("SNMP SET: error %d", resp->error_status); return (0); } if (resp->error_status != SNMP_ERR_NOERROR) { warnx("SNMP SET: error %d", resp->error_status); return (-1); } if (resp->nbindings != req->nbindings) { warnx("SNMP SET: bad number of bindings in response"); return (-1); } for (i = 0; i < req->nbindings; i++) { if (asn_compare_oid(&req->bindings[i].var, &resp->bindings[i].var) != 0) { warnx("SNMP SET: wrong OID in response to SET"); return (-1); } if (resp->bindings[i].syntax != req->bindings[i].syntax) { warnx("SNMP SET: bad syntax in response"); return (-1); } } return (1); } /* * Simple checks for response PDUs against request PDUs. Return values: 1=ok, * 0=nosuchname or similar, -1=failure, -2=no response at all */ int snmp_pdu_check(const struct snmp_pdu *req, const struct snmp_pdu *resp) { if (resp == NULL) return (-2); switch (req->type) { case SNMP_PDU_GET: return (ok_get(req, resp)); case SNMP_PDU_SET: return (ok_set(req, resp)); case SNMP_PDU_GETNEXT: return (ok_getnext(req, resp)); } errx(1, "%s: bad pdu type %i", __func__, req->type); } int snmp_dialog(struct snmp_v1_pdu *req, struct snmp_v1_pdu *resp) { struct timeval tv = snmp_client.timeout; struct timeval end; struct snmp_pdu pdu; int ret; int32_t reqid; u_int i; /* * Make a copy of the request and replace the syntaxes by NULL * if this is a GET,GETNEXT or GETBULK. */ pdu = *req; if (pdu.type == SNMP_PDU_GET || pdu.type == SNMP_PDU_GETNEXT || pdu.type == SNMP_PDU_GETBULK) { for (i = 0; i < pdu.nbindings; i++) pdu.bindings[i].syntax = SNMP_SYNTAX_NULL; } for (i = 0; i <= snmp_client.retries; i++) { (void)gettimeofday(&end, NULL); timeradd(&end, &snmp_client.timeout, &end); if ((reqid = snmp_send_packet(&pdu)) == -1) return (-1); for (;;) { (void)gettimeofday(&tv, NULL); if (timercmp(&end, &tv, <=)) break; timersub(&end, &tv, &tv); if ((ret = snmp_receive_packet(resp, &tv)) == 0) /* timeout */ break; if (ret > 0) { if (reqid == resp->request_id) return (0); /* not for us */ (void)snmp_deliver_packet(resp); } if (ret < 0 && errno == EPIPE) /* stream closed */ return (-1); } } errno = ETIMEDOUT; seterr(&snmp_client, "retry count exceeded"); return (-1); } int snmp_discover_engine(char *passwd) { char cname[SNMP_ADM_STR32_SIZ]; enum snmp_authentication cap; enum snmp_privacy cpp; struct snmp_pdu req, resp; if (snmp_client.version != SNMP_V3) seterr(&snmp_client, "wrong version"); strlcpy(cname, snmp_client.user.sec_name, sizeof(cname)); cap = snmp_client.user.auth_proto; cpp = snmp_client.user.priv_proto; snmp_client.engine.engine_len = 0; snmp_client.engine.engine_boots = 0; snmp_client.engine.engine_time = 0; snmp_client.user.auth_proto = SNMP_AUTH_NOAUTH; snmp_client.user.priv_proto = SNMP_PRIV_NOPRIV; memset(snmp_client.user.sec_name, 0, sizeof(snmp_client.user.sec_name)); snmp_pdu_create(&req, SNMP_PDU_GET); if (snmp_dialog(&req, &resp) == -1) return (-1); if (resp.version != req.version) { seterr(&snmp_client, "wrong version"); return (-1); } if (resp.error_status != SNMP_ERR_NOERROR) { seterr(&snmp_client, "Error %d in responce", resp.error_status); return (-1); } snmp_client.engine.engine_len = resp.engine.engine_len; snmp_client.engine.max_msg_size = resp.engine.max_msg_size; memcpy(snmp_client.engine.engine_id, resp.engine.engine_id, resp.engine.engine_len); strlcpy(snmp_client.user.sec_name, cname, sizeof(snmp_client.user.sec_name)); snmp_client.user.auth_proto = cap; snmp_client.user.priv_proto = cpp; if (snmp_client.user.auth_proto == SNMP_AUTH_NOAUTH) return (0); if (passwd == NULL || snmp_passwd_to_keys(&snmp_client.user, passwd) != SNMP_CODE_OK || snmp_get_local_keys(&snmp_client.user, snmp_client.engine.engine_id, snmp_client.engine.engine_len) != SNMP_CODE_OK) return (-1); if (resp.engine.engine_boots != 0) snmp_client.engine.engine_boots = resp.engine.engine_boots; if (resp.engine.engine_time != 0) { snmp_client.engine.engine_time = resp.engine.engine_time; return (0); } snmp_pdu_free(&req); snmp_pdu_create(&req, SNMP_PDU_GET); req.engine.engine_boots = 0; req.engine.engine_time = 0; if (snmp_dialog(&req, &resp) == -1) return (-1); if (resp.version != req.version) { seterr(&snmp_client, "wrong version"); return (-1); } if (resp.error_status != SNMP_ERR_NOERROR) { seterr(&snmp_client, "Error %d in responce", resp.error_status); return (-1); } snmp_client.engine.engine_boots = resp.engine.engine_boots; snmp_client.engine.engine_time = resp.engine.engine_time; snmp_pdu_free(&req); snmp_pdu_free(&resp); return (0); } int snmp_client_set_host(struct snmp_client *cl, const char *h) { char *np; if (h == NULL) { if (cl->chost != NULL) free(cl->chost); cl->chost = NULL; } else { if ((np = malloc(strlen(h) + 1)) == NULL) return (-1); strcpy(np, h); if (cl->chost != NULL) free(cl->chost); cl->chost = np; } return (0); } int snmp_client_set_port(struct snmp_client *cl, const char *p) { char *np; if (p == NULL) { if (cl->cport != NULL) free(cl->cport); cl->cport = NULL; } else { if ((np = malloc(strlen(p) + 1)) == NULL) return (-1); strcpy(np, p); if (cl->cport != NULL) free(cl->cport); cl->cport = np; } return (0); } /* * parse a server specification * * [trans::][community@][server][:port] */ int snmp_parse_server(struct snmp_client *sc, const char *str) { const char *p, *s = str; /* look for a double colon */ for (p = s; *p != '\0'; p++) { if (*p == '\\' && p[1] != '\0') { p++; continue; } if (*p == ':' && p[1] == ':') break; } if (*p != '\0') { if (p > s) { if (p - s == 3 && strncmp(s, "udp", 3) == 0) sc->trans = SNMP_TRANS_UDP; else if (p - s == 6 && strncmp(s, "stream", 6) == 0) sc->trans = SNMP_TRANS_LOC_STREAM; else if (p - s == 5 && strncmp(s, "dgram", 5) == 0) sc->trans = SNMP_TRANS_LOC_DGRAM; else { seterr(sc, "unknown SNMP transport '%.*s'", (int)(p - s), s); return (-1); } } s = p + 2; } /* look for a @ */ for (p = s; *p != '\0'; p++) { if (*p == '\\' && p[1] != '\0') { p++; continue; } if (*p == '@') break; } if (*p != '\0') { if (p - s > SNMP_COMMUNITY_MAXLEN) { seterr(sc, "community string too long"); return (-1); } strncpy(sc->read_community, s, p - s); sc->read_community[p - s] = '\0'; strncpy(sc->write_community, s, p - s); sc->write_community[p - s] = '\0'; s = p + 1; } /* look for a colon */ for (p = s; *p != '\0'; p++) { if (*p == '\\' && p[1] != '\0') { p++; continue; } if (*p == ':') break; } if (*p == ':') { if (p > s) { /* host:port */ free(sc->chost); if ((sc->chost = malloc(p - s + 1)) == NULL) { seterr(sc, "%s", strerror(errno)); return (-1); } strncpy(sc->chost, s, p - s); sc->chost[p - s] = '\0'; } /* port */ free(sc->cport); if ((sc->cport = strdup(p + 1)) == NULL) { seterr(sc, "%s", strerror(errno)); return (-1); } } else if (p > s) { /* host */ free(sc->chost); if ((sc->chost = strdup(s)) == NULL) { seterr(sc, "%s", strerror(errno)); return (-1); } } return (0); } Index: projects/clang400-import/contrib/bsnmp/snmp_usm/usm_snmp.c =================================================================== --- projects/clang400-import/contrib/bsnmp/snmp_usm/usm_snmp.c (revision 311459) +++ projects/clang400-import/contrib/bsnmp/snmp_usm/usm_snmp.c (revision 311460) @@ -1,614 +1,618 @@ /*- * Copyright (c) 2010 The FreeBSD Foundation * All rights reserved. * * This software was developed by Shteryana Sotirova Shopova 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. * * $FreeBSD$ */ #include #include #include #include #include #include #include #include #include #include "asn1.h" #include "snmp.h" #include "snmpmod.h" #include "usm_tree.h" #include "usm_oid.h" static struct lmodule *usm_module; /* For the registration. */ static const struct asn_oid oid_usm = OIDX_snmpUsmMIB; static const struct asn_oid oid_usmNoAuthProtocol = OIDX_usmNoAuthProtocol; static const struct asn_oid oid_usmHMACMD5AuthProtocol = \ OIDX_usmHMACMD5AuthProtocol; static const struct asn_oid oid_usmHMACSHAAuthProtocol = \ OIDX_usmHMACSHAAuthProtocol; static const struct asn_oid oid_usmNoPrivProtocol = OIDX_usmNoPrivProtocol; static const struct asn_oid oid_usmDESPrivProtocol = OIDX_usmDESPrivProtocol; static const struct asn_oid oid_usmAesCfb128Protocol = OIDX_usmAesCfb128Protocol; static const struct asn_oid oid_usmUserSecurityName = OIDX_usmUserSecurityName; /* The registration. */ static uint reg_usm; static int32_t usm_lock; static struct usm_user * usm_get_user(const struct asn_oid *, uint); static struct usm_user * usm_get_next_user(const struct asn_oid *, uint); static void usm_append_userindex(struct asn_oid *, uint, const struct usm_user *); static int usm_user_index_decode(const struct asn_oid *, uint, uint8_t *, uint32_t *, char *); int op_usm_stats(struct snmp_context *ctx __unused, struct snmp_value *val, uint32_t sub __unused, uint32_t iidx __unused, enum snmp_op op) { struct snmpd_usmstat *usmstats; if (op == SNMP_OP_SET) return (SNMP_ERR_NOT_WRITEABLE); if ((usmstats = bsnmpd_get_usm_stats()) == NULL) return (SNMP_ERR_GENERR); if (op == SNMP_OP_GET) { switch (val->var.subs[sub - 1]) { case LEAF_usmStatsUnsupportedSecLevels: val->v.uint32 = usmstats->unsupported_seclevels; break; case LEAF_usmStatsNotInTimeWindows: val->v.uint32 = usmstats->not_in_time_windows; break; case LEAF_usmStatsUnknownUserNames: val->v.uint32 = usmstats->unknown_users; break; case LEAF_usmStatsUnknownEngineIDs: val->v.uint32 = usmstats->unknown_engine_ids; break; case LEAF_usmStatsWrongDigests: val->v.uint32 = usmstats->wrong_digests; break; case LEAF_usmStatsDecryptionErrors: val->v.uint32 = usmstats->decrypt_errors; break; default: return (SNMP_ERR_NOSUCHNAME); } return (SNMP_ERR_NOERROR); } abort(); } int op_usm_lock(struct snmp_context *ctx __unused, struct snmp_value *val, uint32_t sub, uint32_t iidx __unused, enum snmp_op op) { if (val->var.subs[sub - 1] != LEAF_usmUserSpinLock) return (SNMP_ERR_NOSUCHNAME); switch (op) { case SNMP_OP_GET: if (++usm_lock == INT32_MAX) usm_lock = 0; val->v.integer = usm_lock; break; case SNMP_OP_GETNEXT: abort(); case SNMP_OP_SET: if (val->v.integer != usm_lock) return (SNMP_ERR_INCONS_VALUE); break; case SNMP_OP_ROLLBACK: /* FALLTHROUGH */ case SNMP_OP_COMMIT: break; } return (SNMP_ERR_NOERROR); } int op_usm_users(struct snmp_context *ctx, struct snmp_value *val, uint32_t sub, uint32_t iidx __unused, enum snmp_op op) { uint32_t elen; struct usm_user *uuser, *clone; char uname[SNMP_ADM_STR32_SIZ]; uint8_t eid[SNMP_ENGINE_ID_SIZ]; switch (op) { case SNMP_OP_GET: if ((uuser = usm_get_user(&val->var, sub)) == NULL) return (SNMP_ERR_NOSUCHNAME); break; case SNMP_OP_GETNEXT: if ((uuser = usm_get_next_user(&val->var, sub)) == NULL) return (SNMP_ERR_NOSUCHNAME); usm_append_userindex(&val->var, sub, uuser); break; case SNMP_OP_SET: if ((uuser = usm_get_user(&val->var, sub)) == NULL && val->var.subs[sub - 1] != LEAF_usmUserStatus && val->var.subs[sub - 1] != LEAF_usmUserCloneFrom) - return (SNMP_ERR_NOSUCHNAME); + return (SNMP_ERR_NOSUCHNAME); + /* + * XXX (ngie): need to investigate the MIB to determine how + * this is possible given some of the transitions below. + */ if (community != COMM_INITIALIZE && - uuser->type == StorageType_readOnly) + uuser != NULL && uuser->type == StorageType_readOnly) return (SNMP_ERR_NOT_WRITEABLE); switch (val->var.subs[sub - 1]) { case LEAF_usmUserSecurityName: return (SNMP_ERR_NOT_WRITEABLE); case LEAF_usmUserCloneFrom: if (uuser != NULL || usm_user_index_decode(&val->var, sub, eid, &elen, uname) < 0 || !(asn_is_suboid(&oid_usmUserSecurityName, &val->v.oid))) return (SNMP_ERR_WRONG_VALUE); if ((clone = usm_get_user(&val->v.oid, sub)) == NULL) return (SNMP_ERR_INCONS_VALUE); if ((uuser = usm_new_user(eid, elen, uname)) == NULL) return (SNMP_ERR_GENERR); uuser->status = RowStatus_notReady; if (community != COMM_INITIALIZE) uuser->type = StorageType_volatile; else uuser->type = StorageType_readOnly; uuser->suser.auth_proto = clone->suser.auth_proto; uuser->suser.priv_proto = clone->suser.priv_proto; memcpy(uuser->suser.auth_key, clone->suser.auth_key, sizeof(uuser->suser.auth_key)); memcpy(uuser->suser.priv_key, clone->suser.priv_key, sizeof(uuser->suser.priv_key)); ctx->scratch->int1 = RowStatus_createAndWait; break; case LEAF_usmUserAuthProtocol: ctx->scratch->int1 = uuser->suser.auth_proto; if (asn_compare_oid(&oid_usmNoAuthProtocol, &val->v.oid) == 0) uuser->suser.auth_proto = SNMP_AUTH_NOAUTH; else if (asn_compare_oid(&oid_usmHMACMD5AuthProtocol, &val->v.oid) == 0) uuser->suser.auth_proto = SNMP_AUTH_HMAC_MD5; else if (asn_compare_oid(&oid_usmHMACSHAAuthProtocol, &val->v.oid) == 0) uuser->suser.auth_proto = SNMP_AUTH_HMAC_SHA; else return (SNMP_ERR_WRONG_VALUE); break; case LEAF_usmUserAuthKeyChange: case LEAF_usmUserOwnAuthKeyChange: if (val->var.subs[sub - 1] == LEAF_usmUserOwnAuthKeyChange && (usm_user == NULL || strcmp(uuser->suser.sec_name, usm_user->suser.sec_name) != 0)) return (SNMP_ERR_NO_ACCESS); if (val->v.octetstring.len > SNMP_AUTH_KEY_SIZ) return (SNMP_ERR_INCONS_VALUE); ctx->scratch->ptr1 = malloc(SNMP_AUTH_KEY_SIZ); if (ctx->scratch->ptr1 == NULL) return (SNMP_ERR_GENERR); memcpy(ctx->scratch->ptr1, uuser->suser.auth_key, SNMP_AUTH_KEY_SIZ); memcpy(uuser->suser.auth_key, val->v.octetstring.octets, val->v.octetstring.len); break; case LEAF_usmUserPrivProtocol: ctx->scratch->int1 = uuser->suser.priv_proto; if (asn_compare_oid(&oid_usmNoPrivProtocol, &val->v.oid) == 0) uuser->suser.priv_proto = SNMP_PRIV_NOPRIV; else if (asn_compare_oid(&oid_usmDESPrivProtocol, &val->v.oid) == 0) uuser->suser.priv_proto = SNMP_PRIV_DES; else if (asn_compare_oid(&oid_usmAesCfb128Protocol, &val->v.oid) == 0) uuser->suser.priv_proto = SNMP_PRIV_AES; else return (SNMP_ERR_WRONG_VALUE); break; case LEAF_usmUserPrivKeyChange: case LEAF_usmUserOwnPrivKeyChange: if (val->var.subs[sub - 1] == LEAF_usmUserOwnPrivKeyChange && (usm_user == NULL || strcmp(uuser->suser.sec_name, usm_user->suser.sec_name) != 0)) return (SNMP_ERR_NO_ACCESS); if (val->v.octetstring.len > SNMP_PRIV_KEY_SIZ) return (SNMP_ERR_INCONS_VALUE); ctx->scratch->ptr1 = malloc(SNMP_PRIV_KEY_SIZ); if (ctx->scratch->ptr1 == NULL) return (SNMP_ERR_GENERR); memcpy(ctx->scratch->ptr1, uuser->suser.priv_key, sizeof(uuser->suser.priv_key)); memcpy(uuser->suser.priv_key, val->v.octetstring.octets, val->v.octetstring.len); break; case LEAF_usmUserPublic: if (val->v.octetstring.len > SNMP_ADM_STR32_SIZ) return (SNMP_ERR_INCONS_VALUE); if (uuser->user_public_len > 0) { ctx->scratch->ptr2 = malloc(uuser->user_public_len); if (ctx->scratch->ptr2 == NULL) return (SNMP_ERR_GENERR); memcpy(ctx->scratch->ptr2, uuser->user_public, uuser->user_public_len); ctx->scratch->int2 = uuser->user_public_len; } if (val->v.octetstring.len > 0) { memcpy(uuser->user_public, val->v.octetstring.octets, val->v.octetstring.len); uuser->user_public_len = val->v.octetstring.len; } else { memset(uuser->user_public, 0, sizeof(uuser->user_public)); uuser->user_public_len = 0; } break; case LEAF_usmUserStorageType: return (SNMP_ERR_INCONS_VALUE); case LEAF_usmUserStatus: if (uuser == NULL) { if (val->v.integer != RowStatus_createAndWait || usm_user_index_decode(&val->var, sub, eid, &elen, uname) < 0) return (SNMP_ERR_INCONS_VALUE); uuser = usm_new_user(eid, elen, uname); if (uuser == NULL) return (SNMP_ERR_GENERR); uuser->status = RowStatus_notReady; if (community != COMM_INITIALIZE) uuser->type = StorageType_volatile; else uuser->type = StorageType_readOnly; } else if (val->v.integer != RowStatus_active && val->v.integer != RowStatus_destroy) return (SNMP_ERR_INCONS_VALUE); uuser->status = val->v.integer; break; } return (SNMP_ERR_NOERROR); case SNMP_OP_COMMIT: switch (val->var.subs[sub - 1]) { case LEAF_usmUserAuthKeyChange: case LEAF_usmUserOwnAuthKeyChange: case LEAF_usmUserPrivKeyChange: case LEAF_usmUserOwnPrivKeyChange: free(ctx->scratch->ptr1); break; case LEAF_usmUserPublic: if (ctx->scratch->ptr2 != NULL) free(ctx->scratch->ptr2); break; case LEAF_usmUserStatus: if (val->v.integer != RowStatus_destroy) break; if ((uuser = usm_get_user(&val->var, sub)) == NULL) return (SNMP_ERR_GENERR); usm_delete_user(uuser); break; default: break; } return (SNMP_ERR_NOERROR); case SNMP_OP_ROLLBACK: if ((uuser = usm_get_user(&val->var, sub)) == NULL) return (SNMP_ERR_GENERR); switch (val->var.subs[sub - 1]) { case LEAF_usmUserAuthProtocol: uuser->suser.auth_proto = ctx->scratch->int1; break; case LEAF_usmUserAuthKeyChange: case LEAF_usmUserOwnAuthKeyChange: memcpy(uuser->suser.auth_key, ctx->scratch->ptr1, sizeof(uuser->suser.auth_key)); free(ctx->scratch->ptr1); break; case LEAF_usmUserPrivProtocol: uuser->suser.priv_proto = ctx->scratch->int1; break; case LEAF_usmUserPrivKeyChange: case LEAF_usmUserOwnPrivKeyChange: memcpy(uuser->suser.priv_key, ctx->scratch->ptr1, sizeof(uuser->suser.priv_key)); free(ctx->scratch->ptr1); break; case LEAF_usmUserPublic: if (ctx->scratch->ptr2 != NULL) { memcpy(uuser->user_public, ctx->scratch->ptr2, ctx->scratch->int2); uuser->user_public_len = ctx->scratch->int2; free(ctx->scratch->ptr2); } else { memset(uuser->user_public, 0, sizeof(uuser->user_public)); uuser->user_public_len = 0; } break; case LEAF_usmUserCloneFrom: case LEAF_usmUserStatus: if (ctx->scratch->int1 == RowStatus_createAndWait) usm_delete_user(uuser); break; default: break; } return (SNMP_ERR_NOERROR); default: abort(); } switch (val->var.subs[sub - 1]) { case LEAF_usmUserSecurityName: return (string_get(val, uuser->suser.sec_name, -1)); case LEAF_usmUserCloneFrom: memcpy(&val->v.oid, &oid_zeroDotZero, sizeof(oid_zeroDotZero)); break; case LEAF_usmUserAuthProtocol: switch (uuser->suser.auth_proto) { case SNMP_AUTH_HMAC_MD5: memcpy(&val->v.oid, &oid_usmHMACMD5AuthProtocol, sizeof(oid_usmHMACMD5AuthProtocol)); break; case SNMP_AUTH_HMAC_SHA: memcpy(&val->v.oid, &oid_usmHMACSHAAuthProtocol, sizeof(oid_usmHMACSHAAuthProtocol)); break; default: memcpy(&val->v.oid, &oid_usmNoAuthProtocol, sizeof(oid_usmNoAuthProtocol)); break; } break; case LEAF_usmUserAuthKeyChange: case LEAF_usmUserOwnAuthKeyChange: return (string_get(val, (char *)uuser->suser.auth_key, 0)); case LEAF_usmUserPrivProtocol: switch (uuser->suser.priv_proto) { case SNMP_PRIV_DES: memcpy(&val->v.oid, &oid_usmDESPrivProtocol, sizeof(oid_usmDESPrivProtocol)); break; case SNMP_PRIV_AES: memcpy(&val->v.oid, &oid_usmAesCfb128Protocol, sizeof(oid_usmAesCfb128Protocol)); break; default: memcpy(&val->v.oid, &oid_usmNoPrivProtocol, sizeof(oid_usmNoPrivProtocol)); break; } break; case LEAF_usmUserPrivKeyChange: case LEAF_usmUserOwnPrivKeyChange: return (string_get(val, (char *)uuser->suser.priv_key, 0)); case LEAF_usmUserPublic: return (string_get(val, uuser->user_public, uuser->user_public_len)); case LEAF_usmUserStorageType: val->v.integer = uuser->type; break; case LEAF_usmUserStatus: val->v.integer = uuser->status; break; } return (SNMP_ERR_NOERROR); } static int usm_user_index_decode(const struct asn_oid *oid, uint sub, uint8_t *engine, uint32_t *elen, char *uname) { uint32_t i, nlen; int uname_off; if (oid->subs[sub] > SNMP_ENGINE_ID_SIZ) return (-1); for (i = 0; i < oid->subs[sub]; i++) engine[i] = oid->subs[sub + i + 1]; *elen = i; uname_off = sub + oid->subs[sub] + 1; if ((nlen = oid->subs[uname_off]) >= SNMP_ADM_STR32_SIZ) return (-1); for (i = 0; i < nlen; i++) uname[i] = oid->subs[uname_off + i + 1]; uname[nlen] = '\0'; return (0); } static void usm_append_userindex(struct asn_oid *oid, uint sub, const struct usm_user *uuser) { uint32_t i; oid->len = sub + uuser->user_engine_len + strlen(uuser->suser.sec_name); oid->len += 2; oid->subs[sub] = uuser->user_engine_len; for (i = 1; i < uuser->user_engine_len + 1; i++) oid->subs[sub + i] = uuser->user_engine_id[i - 1]; sub += uuser->user_engine_len + 1; oid->subs[sub] = strlen(uuser->suser.sec_name); for (i = 1; i <= oid->subs[sub]; i++) oid->subs[sub + i] = uuser->suser.sec_name[i - 1]; } static struct usm_user * usm_get_user(const struct asn_oid *oid, uint sub) { uint32_t enginelen; char username[SNMP_ADM_STR32_SIZ]; uint8_t engineid[SNMP_ENGINE_ID_SIZ]; if (usm_user_index_decode(oid, sub, engineid, &enginelen, username) < 0) return (NULL); return (usm_find_user(engineid, enginelen, username)); } static struct usm_user * usm_get_next_user(const struct asn_oid *oid, uint sub) { uint32_t enginelen; char username[SNMP_ADM_STR32_SIZ]; uint8_t engineid[SNMP_ENGINE_ID_SIZ]; struct usm_user *uuser; if (oid->len - sub == 0) return (usm_first_user()); if (usm_user_index_decode(oid, sub, engineid, &enginelen, username) < 0) return (NULL); if ((uuser = usm_find_user(engineid, enginelen, username)) != NULL) return (usm_next_user(uuser)); return (NULL); } /* * USM snmp module initialization hook. * Returns 0 on success, < 0 on error. */ static int usm_init(struct lmodule * mod, int argc __unused, char *argv[] __unused) { usm_module = mod; usm_lock = random(); bsnmpd_reset_usm_stats(); return (0); } /* * USM snmp module finalization hook. */ static int usm_fini(void) { usm_flush_users(); or_unregister(reg_usm); return (0); } /* * USM snmp module start operation. */ static void usm_start(void) { reg_usm = or_register(&oid_usm, "The MIB module for managing SNMP User-Based Security Model.", usm_module); } static void usm_dump(void) { struct usm_user *uuser; struct snmpd_usmstat *usmstats; const char *const authstr[] = { "noauth", "md5", "sha", NULL }; const char *const privstr[] = { "nopriv", "des", "aes", NULL }; if ((usmstats = bsnmpd_get_usm_stats()) != NULL) { syslog(LOG_ERR, "UnsupportedSecLevels\t\t%u", usmstats->unsupported_seclevels); syslog(LOG_ERR, "NotInTimeWindows\t\t%u", usmstats->not_in_time_windows); syslog(LOG_ERR, "UnknownUserNames\t\t%u", usmstats->unknown_users); syslog(LOG_ERR, "UnknownEngineIDs\t\t%u", usmstats->unknown_engine_ids); syslog(LOG_ERR, "WrongDigests\t\t%u", usmstats->wrong_digests); syslog(LOG_ERR, "DecryptionErrors\t\t%u", usmstats->decrypt_errors); } syslog(LOG_ERR, "USM users"); for (uuser = usm_first_user(); uuser != NULL; (uuser = usm_next_user(uuser))) syslog(LOG_ERR, "user %s\t\t%s, %s", uuser->suser.sec_name, authstr[uuser->suser.auth_proto], privstr[uuser->suser.priv_proto]); } const char usm_comment[] = \ "This module implements SNMP User-based Security Model defined in RFC 3414."; const struct snmp_module config = { .comment = usm_comment, .init = usm_init, .fini = usm_fini, .start = usm_start, .tree = usm_ctree, .dump = usm_dump, .tree_size = usm_CTREE_SIZE, }; Index: projects/clang400-import/contrib/bsnmp/snmpd/main.c =================================================================== --- projects/clang400-import/contrib/bsnmp/snmpd/main.c (revision 311459) +++ projects/clang400-import/contrib/bsnmp/snmpd/main.c (revision 311460) @@ -1,3091 +1,3090 @@ /* * Copyright (c) 2001-2003 * Fraunhofer Institute for Open Communication Systems (FhG Fokus). * All rights reserved. * * Author: Harti Brandt * * Copyright (c) 2010 The FreeBSD Foundation * All rights reserved. * * Portions of this software were developed by Shteryana Sotirova Shopova * 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 AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $Begemot: bsnmp/snmpd/main.c,v 1.100 2006/02/14 09:04:20 brandt_h Exp $ * * SNMPd main stuff. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef USE_TCPWRAPPERS #include #include #endif #include "support.h" #include "snmpmod.h" #include "snmpd.h" #include "tree.h" #include "oid.h" #define PATH_PID "/var/run/%s.pid" #define PATH_CONFIG "/etc/%s.config" #define PATH_ENGINE "/var/%s.engine" uint64_t this_tick; /* start of processing of current packet (absolute) */ uint64_t start_tick; /* start of processing */ struct systemg systemg = { NULL, { 8, { 1, 3, 6, 1, 4, 1, 1115, 7352 }}, NULL, NULL, NULL, 64 + 8 + 4, 0 }; struct debug debug = { 0, /* dump_pdus */ LOG_DEBUG, /* log_pri */ 0, /* evdebug */ }; struct snmpd snmpd = { 2048, /* txbuf */ 2048, /* rxbuf */ 0, /* comm_dis */ 0, /* auth_traps */ {0, 0, 0, 0}, /* trap1addr */ VERS_ENABLE_ALL,/* version_enable */ }; struct snmpd_stats snmpd_stats; struct snmpd_usmstat snmpd_usmstats; /* snmpEngine */ struct snmp_engine snmpd_engine; /* snmpSerialNo */ int32_t snmp_serial_no; struct snmpd_target_stats snmpd_target_stats; /* search path for config files */ const char *syspath = PATH_SYSCONFIG; /* list of all loaded modules */ struct lmodules lmodules = TAILQ_HEAD_INITIALIZER(lmodules); /* list of loaded modules during start-up in the order they were loaded */ static struct lmodules modules_start = TAILQ_HEAD_INITIALIZER(modules_start); /* list of all known communities */ struct community_list community_list = TAILQ_HEAD_INITIALIZER(community_list); /* list of all known USM users */ static struct usm_userlist usm_userlist = SLIST_HEAD_INITIALIZER(usm_userlist); /* A list of all VACM users configured, including v1, v2c and v3 */ static struct vacm_userlist vacm_userlist = SLIST_HEAD_INITIALIZER(vacm_userlist); /* A list of all VACM groups */ static struct vacm_grouplist vacm_grouplist = SLIST_HEAD_INITIALIZER(vacm_grouplist); static struct vacm_group vacm_default_group = { .groupname = "", }; /* The list of configured access entries */ static struct vacm_accesslist vacm_accesslist = TAILQ_HEAD_INITIALIZER(vacm_accesslist); /* The list of configured views */ static struct vacm_viewlist vacm_viewlist = SLIST_HEAD_INITIALIZER(vacm_viewlist); /* The list of configured contexts */ static struct vacm_contextlist vacm_contextlist = SLIST_HEAD_INITIALIZER(vacm_contextlist); /* list of all installed object resources */ struct objres_list objres_list = TAILQ_HEAD_INITIALIZER(objres_list); /* community value generator */ static u_int next_community_index = 1; /* list of all known ranges */ struct idrange_list idrange_list = TAILQ_HEAD_INITIALIZER(idrange_list); /* identifier generator */ u_int next_idrange = 1; /* list of all current timers */ struct timer_list timer_list = LIST_HEAD_INITIALIZER(timer_list); /* list of file descriptors */ struct fdesc_list fdesc_list = LIST_HEAD_INITIALIZER(fdesc_list); /* program arguments */ static char **progargs; static int nprogargs; /* current community */ u_int community; static struct community *comm; /* current USM user */ struct usm_user *usm_user; /* file names */ static char config_file[MAXPATHLEN + 1]; static char pid_file[MAXPATHLEN + 1]; char engine_file[MAXPATHLEN + 1]; #ifndef USE_LIBBEGEMOT /* event context */ static evContext evctx; #endif /* signal mask */ static sigset_t blocked_sigs; /* signal handling */ static int work; #define WORK_DOINFO 0x0001 #define WORK_RECONFIG 0x0002 /* oids */ static const struct asn_oid oid_snmpMIB = OIDX_snmpMIB, oid_begemotSnmpd = OIDX_begemotSnmpd, oid_coldStart = OIDX_coldStart, oid_authenticationFailure = OIDX_authenticationFailure; const struct asn_oid oid_zeroDotZero = { 2, { 0, 0 }}; const struct asn_oid oid_usmUnknownEngineIDs = { 11, { 1, 3, 6, 1, 6, 3, 15, 1, 1, 4, 0}}; const struct asn_oid oid_usmNotInTimeWindows = { 11, { 1, 3, 6, 1, 6, 3, 15, 1, 1, 2, 0}}; /* request id generator for traps */ u_int trap_reqid; /* help text */ static const char usgtxt[] = "\ Begemot simple SNMP daemon. Copyright (c) 2001-2002 Fraunhofer Institute for\n\ Open Communication Systems (FhG Fokus). All rights reserved.\n\ Copyright (c) 2010 The FreeBSD Foundation. All rights reserved.\n\ usage: snmpd [-dh] [-c file] [-D options] [-e file] [-I path]\n\ [-l prefix] [-m variable=value] [-p file]\n\ options:\n\ -d don't daemonize\n\ -h print this info\n\ -c file specify configuration file\n\ -D options debugging options\n\ -e file specify engine id file\n\ -I path system include path\n\ -l prefix default basename for pid and config file\n\ -m var=val define variable\n\ -p file specify pid file\n\ "; /* hosts_access(3) request */ #ifdef USE_TCPWRAPPERS static struct request_info req; #endif /* transports */ extern const struct transport_def udp_trans; extern const struct transport_def lsock_trans; struct transport_list transport_list = TAILQ_HEAD_INITIALIZER(transport_list); /* forward declarations */ static void snmp_printf_func(const char *fmt, ...); static void snmp_error_func(const char *err, ...); static void snmp_debug_func(const char *err, ...); static void asn_error_func(const struct asn_buf *b, const char *err, ...); /* * Allocate rx/tx buffer. We allocate one byte more for rx. */ void * buf_alloc(int tx) { void *buf; if ((buf = malloc(tx ? snmpd.txbuf : snmpd.rxbuf)) == NULL) { syslog(LOG_CRIT, "cannot allocate buffer"); if (tx) snmpd_stats.noTxbuf++; else snmpd_stats.noRxbuf++; return (NULL); } return (buf); } /* * Return the buffer size. */ size_t buf_size(int tx) { return (tx ? snmpd.txbuf : snmpd.rxbuf); } /* * Prepare a PDU for output */ void snmp_output(struct snmp_pdu *pdu, u_char *sndbuf, size_t *sndlen, const char *dest) { struct asn_buf resp_b; enum snmp_code code; resp_b.asn_ptr = sndbuf; resp_b.asn_len = snmpd.txbuf; if ((code = snmp_pdu_encode(pdu, &resp_b)) != SNMP_CODE_OK) { syslog(LOG_ERR, "cannot encode message (code=%d)", code); abort(); } if (debug.dump_pdus) { snmp_printf("%s <- ", dest); snmp_pdu_dump(pdu); } *sndlen = (size_t)(resp_b.asn_ptr - sndbuf); } /* * Check USM PDU header credentials against local SNMP Engine & users. */ static enum snmp_code snmp_pdu_auth_user(struct snmp_pdu *pdu) { usm_user = NULL; /* un-authenticated snmpEngineId discovery */ if (pdu->engine.engine_len == 0 && strlen(pdu->user.sec_name) == 0) { pdu->engine.engine_len = snmpd_engine.engine_len; memcpy(pdu->engine.engine_id, snmpd_engine.engine_id, snmpd_engine.engine_len); update_snmpd_engine_time(); pdu->engine.engine_boots = snmpd_engine.engine_boots; pdu->engine.engine_time = snmpd_engine.engine_time; pdu->flags |= SNMP_MSG_AUTODISCOVER; return (SNMP_CODE_OK); } if ((usm_user = usm_find_user(pdu->engine.engine_id, pdu->engine.engine_len, pdu->user.sec_name)) == NULL || usm_user->status != 1 /* active */) return (SNMP_CODE_BADUSER); if (usm_user->user_engine_len != snmpd_engine.engine_len || memcmp(usm_user->user_engine_id, snmpd_engine.engine_id, snmpd_engine.engine_len) != 0) return (SNMP_CODE_BADENGINE); pdu->user.priv_proto = usm_user->suser.priv_proto; memcpy(pdu->user.priv_key, usm_user->suser.priv_key, sizeof(pdu->user.priv_key)); /* authenticated snmpEngineId discovery */ if ((pdu->flags & SNMP_MSG_AUTH_FLAG) != 0) { update_snmpd_engine_time(); pdu->user.auth_proto = usm_user->suser.auth_proto; memcpy(pdu->user.auth_key, usm_user->suser.auth_key, sizeof(pdu->user.auth_key)); if (pdu->engine.engine_boots == 0 && pdu->engine.engine_time == 0) { update_snmpd_engine_time(); pdu->flags |= SNMP_MSG_AUTODISCOVER; return (SNMP_CODE_OK); } if (pdu->engine.engine_boots != snmpd_engine.engine_boots || abs(pdu->engine.engine_time - snmpd_engine.engine_time) > SNMP_TIME_WINDOW) return (SNMP_CODE_NOTINTIME); } if (((pdu->flags & SNMP_MSG_PRIV_FLAG) != 0 && (pdu->flags & SNMP_MSG_AUTH_FLAG) == 0) || ((pdu->flags & SNMP_MSG_AUTH_FLAG) == 0 && usm_user->suser.auth_proto != SNMP_AUTH_NOAUTH) || ((pdu->flags & SNMP_MSG_PRIV_FLAG) == 0 && usm_user->suser.priv_proto != SNMP_PRIV_NOPRIV)) return (SNMP_CODE_BADSECLEVEL); return (SNMP_CODE_OK); } /* * Check whether access to each of var bindings in the PDU is allowed based * on the user credentials against the configured User groups & VACM views. */ enum snmp_code snmp_pdu_auth_access(struct snmp_pdu *pdu, int32_t *ip) { const char *uname; int32_t suboid, smodel; uint32_t i; struct vacm_user *vuser; struct vacm_access *acl; struct vacm_context *vacmctx; struct vacm_view *view; /* * At least a default context exists if the snmpd_vacm(3) module is * running. */ if (SLIST_EMPTY(&vacm_contextlist) || (pdu->flags & SNMP_MSG_AUTODISCOVER) != 0) return (SNMP_CODE_OK); switch (pdu->version) { case SNMP_V1: if ((uname = comm_string(community)) == NULL) return (SNMP_CODE_FAILED); smodel = SNMP_SECMODEL_SNMPv1; break; case SNMP_V2c: if ((uname = comm_string(community)) == NULL) return (SNMP_CODE_FAILED); smodel = SNMP_SECMODEL_SNMPv2c; break; case SNMP_V3: uname = pdu->user.sec_name; if ((smodel = pdu->security_model) != SNMP_SECMODEL_USM) return (SNMP_CODE_FAILED); /* Compare the PDU context engine id against the agent's */ if (pdu->context_engine_len != snmpd_engine.engine_len || memcmp(pdu->context_engine, snmpd_engine.engine_id, snmpd_engine.engine_len) != 0) return (SNMP_CODE_FAILED); break; default: abort(); } SLIST_FOREACH(vuser, &vacm_userlist, vvu) if (strcmp(uname, vuser->secname) == 0 && vuser->sec_model == smodel) break; if (vuser == NULL || vuser->group == NULL) return (SNMP_CODE_FAILED); /* XXX: shteryana - recheck */ TAILQ_FOREACH_REVERSE(acl, &vacm_accesslist, vacm_accesslist, vva) { if (acl->group != vuser->group) continue; SLIST_FOREACH(vacmctx, &vacm_contextlist, vcl) if (memcmp(vacmctx->ctxname, acl->ctx_prefix, acl->ctx_match) == 0) goto match; } return (SNMP_CODE_FAILED); match: switch (pdu->type) { case SNMP_PDU_GET: case SNMP_PDU_GETNEXT: case SNMP_PDU_GETBULK: if ((view = acl->read_view) == NULL) return (SNMP_CODE_FAILED); break; case SNMP_PDU_SET: if ((view = acl->write_view) == NULL) return (SNMP_CODE_FAILED); break; case SNMP_PDU_TRAP: case SNMP_PDU_INFORM: case SNMP_PDU_TRAP2: case SNMP_PDU_REPORT: if ((view = acl->notify_view) == NULL) return (SNMP_CODE_FAILED); break; case SNMP_PDU_RESPONSE: /* NOTREACHED */ return (SNMP_CODE_FAILED); default: abort(); } for (i = 0; i < pdu->nbindings; i++) { /* XXX - view->mask*/ suboid = asn_is_suboid(&view->subtree, &pdu->bindings[i].var); if ((!suboid && !view->exclude) || (suboid && view->exclude)) { *ip = i + 1; return (SNMP_CODE_FAILED); } } return (SNMP_CODE_OK); } /* * SNMP input. Start: decode the PDU, find the user or community. */ enum snmpd_input_err snmp_input_start(const u_char *buf, size_t len, const char *source, struct snmp_pdu *pdu, int32_t *ip, size_t *pdulen) { struct asn_buf b; enum snmp_code code; enum snmpd_input_err ret; int sret; /* update uptime */ this_tick = get_ticks(); b.asn_cptr = buf; b.asn_len = len; ret = SNMPD_INPUT_OK; /* look whether we have enough bytes for the entire PDU. */ switch (sret = snmp_pdu_snoop(&b)) { case 0: return (SNMPD_INPUT_TRUNC); case -1: snmpd_stats.inASNParseErrs++; return (SNMPD_INPUT_FAILED); } b.asn_len = *pdulen = (size_t)sret; memset(pdu, 0, sizeof(*pdu)); if ((code = snmp_pdu_decode_header(&b, pdu)) != SNMP_CODE_OK) goto decoded; if (pdu->version == SNMP_V3) { if (pdu->security_model != SNMP_SECMODEL_USM) { code = SNMP_CODE_FAILED; goto decoded; } if ((code = snmp_pdu_auth_user(pdu)) != SNMP_CODE_OK) goto decoded; if ((code = snmp_pdu_decode_secmode(&b, pdu)) != SNMP_CODE_OK) goto decoded; } code = snmp_pdu_decode_scoped(&b, pdu, ip); decoded: snmpd_stats.inPkts++; switch (code) { case SNMP_CODE_FAILED: snmpd_stats.inASNParseErrs++; return (SNMPD_INPUT_FAILED); case SNMP_CODE_BADVERS: bad_vers: snmpd_stats.inBadVersions++; return (SNMPD_INPUT_FAILED); case SNMP_CODE_BADLEN: if (pdu->type == SNMP_OP_SET) ret = SNMPD_INPUT_VALBADLEN; break; case SNMP_CODE_OORANGE: if (pdu->type == SNMP_OP_SET) ret = SNMPD_INPUT_VALRANGE; break; case SNMP_CODE_BADENC: if (pdu->type == SNMP_OP_SET) ret = SNMPD_INPUT_VALBADENC; break; case SNMP_CODE_BADSECLEVEL: snmpd_usmstats.unsupported_seclevels++; return (SNMPD_INPUT_FAILED); case SNMP_CODE_NOTINTIME: snmpd_usmstats.not_in_time_windows++; return (SNMPD_INPUT_FAILED); case SNMP_CODE_BADUSER: snmpd_usmstats.unknown_users++; return (SNMPD_INPUT_FAILED); case SNMP_CODE_BADENGINE: snmpd_usmstats.unknown_engine_ids++; return (SNMPD_INPUT_FAILED); case SNMP_CODE_BADDIGEST: snmpd_usmstats.wrong_digests++; return (SNMPD_INPUT_FAILED); case SNMP_CODE_EDECRYPT: snmpd_usmstats.decrypt_errors++; return (SNMPD_INPUT_FAILED); case SNMP_CODE_OK: switch (pdu->version) { case SNMP_V1: if (!(snmpd.version_enable & VERS_ENABLE_V1)) goto bad_vers; break; case SNMP_V2c: if (!(snmpd.version_enable & VERS_ENABLE_V2C)) goto bad_vers; break; case SNMP_V3: if (!(snmpd.version_enable & VERS_ENABLE_V3)) goto bad_vers; break; case SNMP_Verr: goto bad_vers; } break; } if (debug.dump_pdus) { snmp_printf("%s -> ", source); snmp_pdu_dump(pdu); } /* * Look, whether we know the community or user */ if (pdu->version != SNMP_V3) { TAILQ_FOREACH(comm, &community_list, link) if (comm->string != NULL && strcmp(comm->string, pdu->community) == 0) break; if (comm == NULL) { snmpd_stats.inBadCommunityNames++; snmp_pdu_free(pdu); if (snmpd.auth_traps) snmp_send_trap(&oid_authenticationFailure, (struct snmp_value *)NULL); ret = SNMPD_INPUT_BAD_COMM; } else community = comm->value; } else if (pdu->nbindings == 0) { /* RFC 3414 - snmpEngineID Discovery */ if (strlen(pdu->user.sec_name) == 0) { asn_append_oid(&(pdu->bindings[pdu->nbindings++].var), &oid_usmUnknownEngineIDs); pdu->context_engine_len = snmpd_engine.engine_len; memcpy(pdu->context_engine, snmpd_engine.engine_id, snmpd_engine.engine_len); } else if (pdu->engine.engine_boots == 0 && pdu->engine.engine_time == 0) { asn_append_oid(&(pdu->bindings[pdu->nbindings++].var), &oid_usmNotInTimeWindows); update_snmpd_engine_time(); pdu->engine.engine_boots = snmpd_engine.engine_boots; pdu->engine.engine_time = snmpd_engine.engine_time; } } else if (usm_user->suser.auth_proto != SNMP_AUTH_NOAUTH && (pdu->engine.engine_boots == 0 || pdu->engine.engine_time == 0)) { snmpd_usmstats.not_in_time_windows++; ret = SNMPD_INPUT_FAILED; } if ((code = snmp_pdu_auth_access(pdu, ip)) != SNMP_CODE_OK) ret = SNMPD_INPUT_FAILED; return (ret); } /* * Will return only _OK or _FAILED */ enum snmpd_input_err snmp_input_finish(struct snmp_pdu *pdu, const u_char *rcvbuf, size_t rcvlen, u_char *sndbuf, size_t *sndlen, const char *source, enum snmpd_input_err ierr, int32_t ivar, void *data) { struct snmp_pdu resp; struct asn_buf resp_b, pdu_b; enum snmp_ret ret; resp_b.asn_ptr = sndbuf; resp_b.asn_len = snmpd.txbuf; pdu_b.asn_cptr = rcvbuf; pdu_b.asn_len = rcvlen; if (ierr != SNMPD_INPUT_OK) { /* error decoding the input of a SET */ if (pdu->version == SNMP_V1) pdu->error_status = SNMP_ERR_BADVALUE; else if (ierr == SNMPD_INPUT_VALBADLEN) pdu->error_status = SNMP_ERR_WRONG_LENGTH; else if (ierr == SNMPD_INPUT_VALRANGE) pdu->error_status = SNMP_ERR_WRONG_VALUE; else pdu->error_status = SNMP_ERR_WRONG_ENCODING; pdu->error_index = ivar; if (snmp_make_errresp(pdu, &pdu_b, &resp_b) == SNMP_RET_IGN) { syslog(LOG_WARNING, "could not encode error response"); snmpd_stats.silentDrops++; return (SNMPD_INPUT_FAILED); } if (debug.dump_pdus) { snmp_printf("%s <- ", source); snmp_pdu_dump(pdu); } *sndlen = (size_t)(resp_b.asn_ptr - sndbuf); return (SNMPD_INPUT_OK); } switch (pdu->type) { case SNMP_PDU_GET: ret = snmp_get(pdu, &resp_b, &resp, data); break; case SNMP_PDU_GETNEXT: ret = snmp_getnext(pdu, &resp_b, &resp, data); break; case SNMP_PDU_SET: ret = snmp_set(pdu, &resp_b, &resp, data); break; case SNMP_PDU_GETBULK: ret = snmp_getbulk(pdu, &resp_b, &resp, data); break; default: ret = SNMP_RET_IGN; break; } switch (ret) { case SNMP_RET_OK: /* normal return - send a response */ if (debug.dump_pdus) { snmp_printf("%s <- ", source); snmp_pdu_dump(&resp); } *sndlen = (size_t)(resp_b.asn_ptr - sndbuf); snmp_pdu_free(&resp); return (SNMPD_INPUT_OK); case SNMP_RET_IGN: /* error - send nothing */ snmpd_stats.silentDrops++; return (SNMPD_INPUT_FAILED); case SNMP_RET_ERR: /* error - send error response. The snmp routine has * changed the error fields in the original message. */ resp_b.asn_ptr = sndbuf; resp_b.asn_len = snmpd.txbuf; if (snmp_make_errresp(pdu, &pdu_b, &resp_b) == SNMP_RET_IGN) { syslog(LOG_WARNING, "could not encode error response"); snmpd_stats.silentDrops++; return (SNMPD_INPUT_FAILED); } else { if (debug.dump_pdus) { snmp_printf("%s <- ", source); snmp_pdu_dump(pdu); } *sndlen = (size_t)(resp_b.asn_ptr - sndbuf); return (SNMPD_INPUT_OK); } } abort(); } /* * Insert a port into the right place in the transport's table of ports */ void trans_insert_port(struct transport *t, struct tport *port) { struct tport *p; TAILQ_FOREACH(p, &t->table, link) { if (asn_compare_oid(&p->index, &port->index) > 0) { TAILQ_INSERT_BEFORE(p, port, link); return; } } port->transport = t; TAILQ_INSERT_TAIL(&t->table, port, link); } /* * Remove a port from a transport's list */ void trans_remove_port(struct tport *port) { TAILQ_REMOVE(&port->transport->table, port, link); } /* * Find a port on a transport's list */ struct tport * trans_find_port(struct transport *t, const struct asn_oid *idx, u_int sub) { return (FIND_OBJECT_OID(&t->table, idx, sub)); } /* * Find next port on a transport's list */ struct tport * trans_next_port(struct transport *t, const struct asn_oid *idx, u_int sub) { return (NEXT_OBJECT_OID(&t->table, idx, sub)); } /* * Return first port */ struct tport * trans_first_port(struct transport *t) { return (TAILQ_FIRST(&t->table)); } /* * Iterate through all ports until a function returns a 0. */ struct tport * trans_iter_port(struct transport *t, int (*func)(struct tport *, intptr_t), intptr_t arg) { struct tport *p; TAILQ_FOREACH(p, &t->table, link) if (func(p, arg) == 0) return (p); return (NULL); } /* * Register a transport */ int trans_register(const struct transport_def *def, struct transport **pp) { u_int i; char or_descr[256]; if ((*pp = malloc(sizeof(**pp))) == NULL) return (SNMP_ERR_GENERR); /* construct index */ (*pp)->index.len = strlen(def->name) + 1; (*pp)->index.subs[0] = strlen(def->name); for (i = 0; i < (*pp)->index.subs[0]; i++) (*pp)->index.subs[i + 1] = def->name[i]; (*pp)->vtab = def; if (FIND_OBJECT_OID(&transport_list, &(*pp)->index, 0) != NULL) { free(*pp); return (SNMP_ERR_INCONS_VALUE); } /* register module */ snprintf(or_descr, sizeof(or_descr), "%s transport mapping", def->name); if (((*pp)->or_index = or_register(&def->id, or_descr, NULL)) == 0) { free(*pp); return (SNMP_ERR_GENERR); } INSERT_OBJECT_OID((*pp), &transport_list); TAILQ_INIT(&(*pp)->table); return (SNMP_ERR_NOERROR); } /* * Unregister transport */ int trans_unregister(struct transport *t) { if (!TAILQ_EMPTY(&t->table)) return (SNMP_ERR_INCONS_VALUE); or_unregister(t->or_index); TAILQ_REMOVE(&transport_list, t, link); return (SNMP_ERR_NOERROR); } /* * File descriptor support */ #ifdef USE_LIBBEGEMOT static void input(int fd, int mask __unused, void *uap) #else static void input(evContext ctx __unused, void *uap, int fd, int mask __unused) #endif { struct fdesc *f = uap; (*f->func)(fd, f->udata); } void fd_suspend(void *p) { struct fdesc *f = p; #ifdef USE_LIBBEGEMOT if (f->id >= 0) { poll_unregister(f->id); f->id = -1; } #else if (evTestID(f->id)) { (void)evDeselectFD(evctx, f->id); evInitID(&f->id); } #endif } int fd_resume(void *p) { struct fdesc *f = p; int err; #ifdef USE_LIBBEGEMOT if (f->id >= 0) return (0); if ((f->id = poll_register(f->fd, input, f, POLL_IN)) < 0) { err = errno; syslog(LOG_ERR, "select fd %d: %m", f->fd); errno = err; return (-1); } #else if (evTestID(f->id)) return (0); if (evSelectFD(evctx, f->fd, EV_READ, input, f, &f->id)) { err = errno; syslog(LOG_ERR, "select fd %d: %m", f->fd); errno = err; return (-1); } #endif return (0); } void * fd_select(int fd, void (*func)(int, void *), void *udata, struct lmodule *mod) { struct fdesc *f; int err; if ((f = malloc(sizeof(struct fdesc))) == NULL) { err = errno; syslog(LOG_ERR, "fd_select: %m"); errno = err; return (NULL); } f->fd = fd; f->func = func; f->udata = udata; f->owner = mod; #ifdef USE_LIBBEGEMOT f->id = -1; #else evInitID(&f->id); #endif if (fd_resume(f)) { err = errno; free(f); errno = err; return (NULL); } LIST_INSERT_HEAD(&fdesc_list, f, link); return (f); } void fd_deselect(void *p) { struct fdesc *f = p; LIST_REMOVE(f, link); fd_suspend(f); free(f); } static void fd_flush(struct lmodule *mod) { struct fdesc *t, *t1; t = LIST_FIRST(&fdesc_list); while (t != NULL) { t1 = LIST_NEXT(t, link); if (t->owner == mod) fd_deselect(t); t = t1; } } /* * Consume a message from the input buffer */ static void snmp_input_consume(struct port_input *pi) { if (!pi->stream) { /* always consume everything */ pi->length = 0; return; } if (pi->consumed >= pi->length) { /* all bytes consumed */ pi->length = 0; return; } memmove(pi->buf, pi->buf + pi->consumed, pi->length - pi->consumed); pi->length -= pi->consumed; } /* * Input from a socket */ int snmpd_input(struct port_input *pi, struct tport *tport) { u_char *sndbuf; size_t sndlen; struct snmp_pdu pdu; enum snmpd_input_err ierr, ferr; enum snmpd_proxy_err perr; ssize_t ret, slen; int32_t vi; #ifdef USE_TCPWRAPPERS char client[16]; #endif ret = tport->transport->vtab->recv(tport, pi); if (ret == -1) return (-1); #ifdef USE_TCPWRAPPERS /* * In case of AF_INET{6} peer, do hosts_access(5) check. */ if (pi->peer->sa_family != AF_LOCAL && inet_ntop(pi->peer->sa_family, &((const struct sockaddr_in *)(const void *)pi->peer)->sin_addr, client, sizeof(client)) != NULL) { request_set(&req, RQ_CLIENT_ADDR, client, 0); if (hosts_access(&req) == 0) { syslog(LOG_ERR, "refused connection from %.500s", eval_client(&req)); return (-1); } } else if (pi->peer->sa_family != AF_LOCAL) syslog(LOG_ERR, "inet_ntop(): %m"); #endif /* * Handle input */ ierr = snmp_input_start(pi->buf, pi->length, "SNMP", &pdu, &vi, &pi->consumed); if (ierr == SNMPD_INPUT_TRUNC) { /* need more bytes. This is ok only for streaming transports. * but only if we have not reached bufsiz yet. */ if (pi->stream) { if (pi->length == buf_size(0)) { snmpd_stats.silentDrops++; return (-1); } return (0); } snmpd_stats.silentDrops++; return (-1); } /* can't check for bad SET pdus here, because a proxy may have to * check the access first. We don't want to return an error response * to a proxy PDU with a wrong community */ if (ierr == SNMPD_INPUT_FAILED) { /* for streaming transports this is fatal */ if (pi->stream) return (-1); snmp_input_consume(pi); return (0); } if (ierr == SNMPD_INPUT_BAD_COMM) { snmp_input_consume(pi); return (0); } /* * If that is a module community and the module has a proxy function, * the hand it over to the module. */ if (comm != NULL && comm->owner != NULL && comm->owner->config->proxy != NULL) { perr = (*comm->owner->config->proxy)(&pdu, tport->transport, &tport->index, pi->peer, pi->peerlen, ierr, vi, !pi->cred || pi->priv); switch (perr) { case SNMPD_PROXY_OK: snmp_input_consume(pi); return (0); case SNMPD_PROXY_REJ: break; case SNMPD_PROXY_DROP: snmp_input_consume(pi); snmp_pdu_free(&pdu); snmpd_stats.proxyDrops++; return (0); case SNMPD_PROXY_BADCOMM: snmp_input_consume(pi); snmp_pdu_free(&pdu); snmpd_stats.inBadCommunityNames++; if (snmpd.auth_traps) snmp_send_trap(&oid_authenticationFailure, (struct snmp_value *)NULL); return (0); case SNMPD_PROXY_BADCOMMUSE: snmp_input_consume(pi); snmp_pdu_free(&pdu); snmpd_stats.inBadCommunityUses++; if (snmpd.auth_traps) snmp_send_trap(&oid_authenticationFailure, (struct snmp_value *)NULL); return (0); } } /* * Check type */ if (pdu.type == SNMP_PDU_RESPONSE || pdu.type == SNMP_PDU_TRAP || pdu.type == SNMP_PDU_TRAP2) { snmpd_stats.silentDrops++; snmpd_stats.inBadPduTypes++; snmp_pdu_free(&pdu); snmp_input_consume(pi); return (0); } /* * Check community */ if (pdu.version < SNMP_V3 && ((pi->cred && !pi->priv && pdu.type == SNMP_PDU_SET) || (community != COMM_WRITE && (pdu.type == SNMP_PDU_SET || community != COMM_READ)))) { snmpd_stats.inBadCommunityUses++; snmp_pdu_free(&pdu); snmp_input_consume(pi); if (snmpd.auth_traps) snmp_send_trap(&oid_authenticationFailure, (struct snmp_value *)NULL); return (0); } /* * Execute it. */ if ((sndbuf = buf_alloc(1)) == NULL) { snmpd_stats.silentDrops++; snmp_pdu_free(&pdu); snmp_input_consume(pi); return (0); } ferr = snmp_input_finish(&pdu, pi->buf, pi->length, sndbuf, &sndlen, "SNMP", ierr, vi, NULL); if (ferr == SNMPD_INPUT_OK) { slen = tport->transport->vtab->send(tport, sndbuf, sndlen, pi->peer, pi->peerlen); if (slen == -1) syslog(LOG_ERR, "send*: %m"); else if ((size_t)slen != sndlen) syslog(LOG_ERR, "send*: short write %zu/%zu", sndlen, (size_t)slen); } snmp_pdu_free(&pdu); free(sndbuf); snmp_input_consume(pi); return (0); } /* * Send a PDU to a given port */ void snmp_send_port(void *targ, const struct asn_oid *port, struct snmp_pdu *pdu, const struct sockaddr *addr, socklen_t addrlen) { struct transport *trans = targ; struct tport *tp; u_char *sndbuf; size_t sndlen; ssize_t len; TAILQ_FOREACH(tp, &trans->table, link) if (asn_compare_oid(port, &tp->index) == 0) break; if (tp == 0) return; if ((sndbuf = buf_alloc(1)) == NULL) return; snmp_output(pdu, sndbuf, &sndlen, "SNMP PROXY"); len = trans->vtab->send(tp, sndbuf, sndlen, addr, addrlen); if (len == -1) syslog(LOG_ERR, "sendto: %m"); else if ((size_t)len != sndlen) syslog(LOG_ERR, "sendto: short write %zu/%zu", sndlen, (size_t)len); free(sndbuf); } /* * Close an input source */ void snmpd_input_close(struct port_input *pi) { if (pi->id != NULL) fd_deselect(pi->id); if (pi->fd >= 0) (void)close(pi->fd); if (pi->buf != NULL) free(pi->buf); } /* * Dump internal state. */ #ifdef USE_LIBBEGEMOT static void info_func(void) #else static void info_func(evContext ctx __unused, void *uap __unused, const void *tag __unused) #endif { struct lmodule *m; u_int i; char buf[10000]; syslog(LOG_DEBUG, "Dump of SNMPd %lu\n", (u_long)getpid()); for (i = 0; i < tree_size; i++) { switch (tree[i].type) { case SNMP_NODE_LEAF: sprintf(buf, "LEAF: %s %s", tree[i].name, asn_oid2str(&tree[i].oid)); break; case SNMP_NODE_COLUMN: sprintf(buf, "COL: %s %s", tree[i].name, asn_oid2str(&tree[i].oid)); break; } syslog(LOG_DEBUG, "%s", buf); } TAILQ_FOREACH(m, &lmodules, link) if (m->config->dump) (*m->config->dump)(); } /* * Re-read configuration */ #ifdef USE_LIBBEGEMOT static void config_func(void) #else static void config_func(evContext ctx __unused, void *uap __unused, const void *tag __unused) #endif { struct lmodule *m; if (read_config(config_file, NULL)) { syslog(LOG_ERR, "error reading config file '%s'", config_file); return; } TAILQ_FOREACH(m, &lmodules, link) if (m->config->config) (*m->config->config)(); } /* * On USR1 dump actual configuration. */ static void onusr1(int s __unused) { work |= WORK_DOINFO; } static void onhup(int s __unused) { work |= WORK_RECONFIG; } static void onterm(int s __unused) { /* allow clean-up */ exit(0); } static void init_sigs(void) { struct sigaction sa; sa.sa_handler = onusr1; sa.sa_flags = SA_RESTART; sigemptyset(&sa.sa_mask); if (sigaction(SIGUSR1, &sa, NULL)) { syslog(LOG_ERR, "sigaction: %m"); exit(1); } sa.sa_handler = onhup; if (sigaction(SIGHUP, &sa, NULL)) { syslog(LOG_ERR, "sigaction: %m"); exit(1); } sa.sa_handler = onterm; sa.sa_flags = 0; sigemptyset(&sa.sa_mask); if (sigaction(SIGTERM, &sa, NULL)) { syslog(LOG_ERR, "sigaction: %m"); exit(1); } if (sigaction(SIGINT, &sa, NULL)) { syslog(LOG_ERR, "sigaction: %m"); exit(1); } } static void block_sigs(void) { sigset_t set; sigfillset(&set); if (sigprocmask(SIG_BLOCK, &set, &blocked_sigs) == -1) { syslog(LOG_ERR, "SIG_BLOCK: %m"); exit(1); } } static void unblock_sigs(void) { if (sigprocmask(SIG_SETMASK, &blocked_sigs, NULL) == -1) { syslog(LOG_ERR, "SIG_SETMASK: %m"); exit(1); } } /* * Shut down */ static void term(void) { (void)unlink(pid_file); } static void trans_stop(void) { struct transport *t; TAILQ_FOREACH(t, &transport_list, link) (void)t->vtab->stop(1); } /* * Define a macro from the command line */ static void do_macro(char *arg) { char *eq; int err; if ((eq = strchr(arg, '=')) == NULL) err = define_macro(arg, ""); else { *eq++ = '\0'; err = define_macro(arg, eq); } if (err == -1) { syslog(LOG_ERR, "cannot save macro: %m"); exit(1); } } /* * Re-implement getsubopt from scratch, because the second argument is broken * and will not compile with WARNS=5. */ static int getsubopt1(char **arg, const char *const *options, char **valp, char **optp) { static const char *const delim = ",\t "; u_int i; char *ptr; *optp = NULL; /* skip leading junk */ for (ptr = *arg; *ptr != '\0'; ptr++) if (strchr(delim, *ptr) == NULL) break; if (*ptr == '\0') { *arg = ptr; return (-1); } *optp = ptr; /* find the end of the option */ while (*++ptr != '\0') if (strchr(delim, *ptr) != NULL || *ptr == '=') break; if (*ptr != '\0') { if (*ptr == '=') { *ptr++ = '\0'; *valp = ptr; while (*ptr != '\0' && strchr(delim, *ptr) == NULL) ptr++; if (*ptr != '\0') *ptr++ = '\0'; } else *ptr++ = '\0'; } *arg = ptr; for (i = 0; *options != NULL; options++, i++) if (strcmp(*optp, *options) == 0) return (i); return (-1); } int main(int argc, char *argv[]) { int opt; FILE *fp; int background = 1; struct tport *p; const char *prefix = "snmpd"; struct lmodule *m; char *value = NULL, *option; /* XXX */ struct transport *t; #define DBG_DUMP 0 #define DBG_EVENTS 1 #define DBG_TRACE 2 static const char *const debug_opts[] = { "dump", "events", "trace", NULL }; snmp_printf = snmp_printf_func; snmp_error = snmp_error_func; snmp_debug = snmp_debug_func; asn_error = asn_error_func; while ((opt = getopt(argc, argv, "c:dD:e:hI:l:m:p:")) != EOF) switch (opt) { case 'c': strlcpy(config_file, optarg, sizeof(config_file)); break; case 'd': background = 0; break; case 'D': while (*optarg) { switch (getsubopt1(&optarg, debug_opts, &value, &option)) { case DBG_DUMP: debug.dump_pdus = 1; break; case DBG_EVENTS: debug.evdebug++; break; case DBG_TRACE: if (value == NULL) syslog(LOG_ERR, "no value for 'trace'"); else snmp_trace = strtoul(value, NULL, 0); break; case -1: if (suboptarg) syslog(LOG_ERR, "unknown debug flag '%s'", option); else syslog(LOG_ERR, "missing debug flag"); break; } } break; case 'e': strlcpy(engine_file, optarg, sizeof(engine_file)); break; case 'h': fprintf(stderr, "%s", usgtxt); exit(0); case 'I': syspath = optarg; break; case 'l': prefix = optarg; break; case 'm': do_macro(optarg); break; case 'p': strlcpy(pid_file, optarg, sizeof(pid_file)); break; } openlog(prefix, LOG_PID | (background ? 0 : LOG_PERROR), LOG_USER); setlogmask(LOG_UPTO(debug.logpri - 1)); if (background && daemon(0, 0) < 0) { syslog(LOG_ERR, "daemon: %m"); exit(1); } argc -= optind; argv += optind; progargs = argv; nprogargs = argc; srandomdev(); snmp_serial_no = random(); #ifdef USE_TCPWRAPPERS /* * Initialize hosts_access(3) handler. */ request_init(&req, RQ_DAEMON, "snmpd", 0); sock_methods(&req); #endif /* * Initialize the tree. */ if ((tree = malloc(sizeof(struct snmp_node) * CTREE_SIZE)) == NULL) { syslog(LOG_ERR, "%m"); exit(1); } memcpy(tree, ctree, sizeof(struct snmp_node) * CTREE_SIZE); tree_size = CTREE_SIZE; /* * Get standard communities */ (void)comm_define(1, "SNMP read", NULL, NULL); (void)comm_define(2, "SNMP write", NULL, NULL); community = COMM_INITIALIZE; trap_reqid = reqid_allocate(512, NULL); if (config_file[0] == '\0') snprintf(config_file, sizeof(config_file), PATH_CONFIG, prefix); init_actvals(); init_snmpd_engine(); this_tick = get_ticks(); start_tick = this_tick; /* start transports */ if (atexit(trans_stop) == -1) { syslog(LOG_ERR, "atexit failed: %m"); exit(1); } if (udp_trans.start() != SNMP_ERR_NOERROR) syslog(LOG_WARNING, "cannot start UDP transport"); if (lsock_trans.start() != SNMP_ERR_NOERROR) syslog(LOG_WARNING, "cannot start LSOCK transport"); #ifdef USE_LIBBEGEMOT if (debug.evdebug > 0) rpoll_trace = 1; #else if (evCreate(&evctx)) { syslog(LOG_ERR, "evCreate: %m"); exit(1); } if (debug.evdebug > 0) evSetDebug(evctx, 10, stderr); #endif if (engine_file[0] == '\0') snprintf(engine_file, sizeof(engine_file), PATH_ENGINE, prefix); if (read_config(config_file, NULL)) { syslog(LOG_ERR, "error in config file"); exit(1); } TAILQ_FOREACH(t, &transport_list, link) TAILQ_FOREACH(p, &t->table, link) t->vtab->init_port(p); init_sigs(); if (pid_file[0] == '\0') snprintf(pid_file, sizeof(pid_file), PATH_PID, prefix); if ((fp = fopen(pid_file, "w")) != NULL) { fprintf(fp, "%u", getpid()); fclose(fp); if (atexit(term) == -1) { syslog(LOG_ERR, "atexit failed: %m"); (void)remove(pid_file); exit(0); } } if (or_register(&oid_snmpMIB, "The MIB module for SNMPv2 entities.", NULL) == 0) { syslog(LOG_ERR, "cannot register SNMPv2 MIB"); exit(1); } if (or_register(&oid_begemotSnmpd, "The MIB module for the Begemot SNMPd.", NULL) == 0) { syslog(LOG_ERR, "cannot register begemotSnmpd MIB"); exit(1); } while ((m = TAILQ_FIRST(&modules_start)) != NULL) { m->flags &= ~LM_ONSTARTLIST; TAILQ_REMOVE(&modules_start, m, start); lm_start(m); } snmp_send_trap(&oid_coldStart, (struct snmp_value *)NULL); for (;;) { #ifndef USE_LIBBEGEMOT evEvent event; #endif struct lmodule *mod; TAILQ_FOREACH(mod, &lmodules, link) if (mod->config->idle != NULL) (*mod->config->idle)(); #ifndef USE_LIBBEGEMOT if (evGetNext(evctx, &event, EV_WAIT) == 0) { if (evDispatch(evctx, event)) syslog(LOG_ERR, "evDispatch: %m"); } else if (errno != EINTR) { syslog(LOG_ERR, "evGetNext: %m"); exit(1); } #else poll_dispatch(1); #endif if (work != 0) { block_sigs(); if (work & WORK_DOINFO) { #ifdef USE_LIBBEGEMOT info_func(); #else if (evWaitFor(evctx, &work, info_func, NULL, NULL) == -1) { syslog(LOG_ERR, "evWaitFor: %m"); exit(1); } #endif } if (work & WORK_RECONFIG) { #ifdef USE_LIBBEGEMOT config_func(); #else if (evWaitFor(evctx, &work, config_func, NULL, NULL) == -1) { syslog(LOG_ERR, "evWaitFor: %m"); exit(1); } #endif } work = 0; unblock_sigs(); #ifndef USE_LIBBEGEMOT if (evDo(evctx, &work) == -1) { syslog(LOG_ERR, "evDo: %m"); exit(1); } #endif } } return (0); } uint64_t get_ticks(void) { struct timeval tv; uint64_t ret; if (gettimeofday(&tv, NULL)) abort(); ret = tv.tv_sec * 100ULL + tv.tv_usec / 10000ULL; return (ret); } /* * Timer support */ /* * Trampoline for the non-repeatable timers. */ #ifdef USE_LIBBEGEMOT static void tfunc(int tid __unused, void *uap) #else static void tfunc(evContext ctx __unused, void *uap, struct timespec due __unused, struct timespec inter __unused) #endif { struct timer *tp = uap; LIST_REMOVE(tp, link); tp->func(tp->udata); free(tp); } /* * Trampoline for the repeatable timers. */ #ifdef USE_LIBBEGEMOT static void trfunc(int tid __unused, void *uap) #else static void trfunc(evContext ctx __unused, void *uap, struct timespec due __unused, struct timespec inter __unused) #endif { struct timer *tp = uap; tp->func(tp->udata); } /* * Start a one-shot timer */ void * timer_start(u_int ticks, void (*func)(void *), void *udata, struct lmodule *mod) { struct timer *tp; #ifndef USE_LIBBEGEMOT struct timespec due; #endif if ((tp = malloc(sizeof(struct timer))) == NULL) { syslog(LOG_CRIT, "out of memory for timer"); exit(1); } #ifndef USE_LIBBEGEMOT due = evAddTime(evNowTime(), evConsTime(ticks / 100, (ticks % 100) * 10000)); #endif tp->udata = udata; tp->owner = mod; tp->func = func; LIST_INSERT_HEAD(&timer_list, tp, link); #ifdef USE_LIBBEGEMOT if ((tp->id = poll_start_timer(ticks * 10, 0, tfunc, tp)) < 0) { syslog(LOG_ERR, "cannot set timer: %m"); exit(1); } #else if (evSetTimer(evctx, tfunc, tp, due, evConsTime(0, 0), &tp->id) == -1) { syslog(LOG_ERR, "cannot set timer: %m"); exit(1); } #endif return (tp); } /* * Start a repeatable timer. When used with USE_LIBBEGEMOT the first argument * is currently ignored and the initial number of ticks is set to the * repeat number of ticks. */ void * timer_start_repeat(u_int ticks __unused, u_int repeat_ticks, void (*func)(void *), void *udata, struct lmodule *mod) { struct timer *tp; #ifndef USE_LIBBEGEMOT struct timespec due; struct timespec inter; #endif if ((tp = malloc(sizeof(struct timer))) == NULL) { syslog(LOG_CRIT, "out of memory for timer"); exit(1); } #ifndef USE_LIBBEGEMOT due = evAddTime(evNowTime(), evConsTime(ticks / 100, (ticks % 100) * 10000)); inter = evConsTime(repeat_ticks / 100, (repeat_ticks % 100) * 10000); #endif tp->udata = udata; tp->owner = mod; tp->func = func; LIST_INSERT_HEAD(&timer_list, tp, link); #ifdef USE_LIBBEGEMOT if ((tp->id = poll_start_timer(repeat_ticks * 10, 1, trfunc, tp)) < 0) { syslog(LOG_ERR, "cannot set timer: %m"); exit(1); } #else if (evSetTimer(evctx, trfunc, tp, due, inter, &tp->id) == -1) { syslog(LOG_ERR, "cannot set timer: %m"); exit(1); } #endif return (tp); } /* * Stop a timer. */ void timer_stop(void *p) { struct timer *tp = p; LIST_REMOVE(tp, link); #ifdef USE_LIBBEGEMOT poll_stop_timer(tp->id); #else if (evClearTimer(evctx, tp->id) == -1) { syslog(LOG_ERR, "cannot stop timer: %m"); exit(1); } #endif free(p); } static void timer_flush(struct lmodule *mod) { struct timer *t, *t1; t = LIST_FIRST(&timer_list); while (t != NULL) { t1 = LIST_NEXT(t, link); if (t->owner == mod) timer_stop(t); t = t1; } } static void snmp_printf_func(const char *fmt, ...) { va_list ap; static char *pend = NULL; char *ret, *new; va_start(ap, fmt); vasprintf(&ret, fmt, ap); va_end(ap); if (ret == NULL) return; if (pend != NULL) { if ((new = realloc(pend, strlen(pend) + strlen(ret) + 1)) == NULL) { free(ret); return; } pend = new; strcat(pend, ret); free(ret); } else pend = ret; while ((ret = strchr(pend, '\n')) != NULL) { *ret = '\0'; syslog(LOG_DEBUG, "%s", pend); if (strlen(ret + 1) == 0) { free(pend); pend = NULL; break; } strcpy(pend, ret + 1); } } static void snmp_error_func(const char *err, ...) { char errbuf[1000]; va_list ap; if (!(snmp_trace & LOG_SNMP_ERRORS)) return; va_start(ap, err); snprintf(errbuf, sizeof(errbuf), "SNMP: "); vsnprintf(errbuf + strlen(errbuf), sizeof(errbuf) - strlen(errbuf), err, ap); va_end(ap); syslog(LOG_ERR, "%s", errbuf); } static void snmp_debug_func(const char *err, ...) { char errbuf[1000]; va_list ap; va_start(ap, err); snprintf(errbuf, sizeof(errbuf), "SNMP: "); vsnprintf(errbuf+strlen(errbuf), sizeof(errbuf)-strlen(errbuf), err, ap); va_end(ap); syslog(LOG_DEBUG, "%s", errbuf); } static void asn_error_func(const struct asn_buf *b, const char *err, ...) { char errbuf[1000]; va_list ap; u_int i; if (!(snmp_trace & LOG_ASN1_ERRORS)) return; va_start(ap, err); snprintf(errbuf, sizeof(errbuf), "ASN.1: "); vsnprintf(errbuf + strlen(errbuf), sizeof(errbuf) - strlen(errbuf), err, ap); va_end(ap); if (b != NULL) { snprintf(errbuf + strlen(errbuf), sizeof(errbuf) - strlen(errbuf), " at"); for (i = 0; b->asn_len > i; i++) snprintf(errbuf + strlen(errbuf), sizeof(errbuf) - strlen(errbuf), " %02x", b->asn_cptr[i]); } syslog(LOG_ERR, "%s", errbuf); } /* * Create a new community */ u_int comm_define(u_int priv, const char *descr, struct lmodule *owner, const char *str) { struct community *c, *p; u_int ncomm; /* generate an identifier */ do { if ((ncomm = next_community_index++) == UINT_MAX) next_community_index = 1; TAILQ_FOREACH(c, &community_list, link) if (c->value == ncomm) break; } while (c != NULL); if ((c = malloc(sizeof(struct community))) == NULL) { syslog(LOG_ERR, "comm_define: %m"); return (0); } c->owner = owner; c->value = ncomm; c->descr = descr; c->string = NULL; c->private = priv; if (str != NULL) { if((c->string = malloc(strlen(str)+1)) == NULL) { free(c); return (0); } strcpy(c->string, str); } /* make index */ if (c->owner == NULL) { c->index.len = 1; c->index.subs[0] = 0; } else { c->index = c->owner->index; } c->index.subs[c->index.len++] = c->private; /* * Insert ordered */ TAILQ_FOREACH(p, &community_list, link) { if (asn_compare_oid(&p->index, &c->index) > 0) { TAILQ_INSERT_BEFORE(p, c, link); break; } } if (p == NULL) TAILQ_INSERT_TAIL(&community_list, c, link); return (c->value); } const char * comm_string(u_int ncomm) { struct community *p; TAILQ_FOREACH(p, &community_list, link) if (p->value == ncomm) return (p->string); return (NULL); } /* * Delete all communities allocated by a module */ static void comm_flush(struct lmodule *mod) { struct community *p, *p1; p = TAILQ_FIRST(&community_list); while (p != NULL) { p1 = TAILQ_NEXT(p, link); if (p->owner == mod) { free(p->string); TAILQ_REMOVE(&community_list, p, link); free(p); } p = p1; } } /* * Request ID handling. * * Allocate a new range of request ids. Use a first fit algorithm. */ u_int reqid_allocate(int size, struct lmodule *mod) { u_int type; struct idrange *r, *r1; if (size <= 0 || size > INT32_MAX) { syslog(LOG_CRIT, "%s: size out of range: %d", __func__, size); return (0); } /* allocate a type id */ do { if ((type = next_idrange++) == UINT_MAX) next_idrange = 1; TAILQ_FOREACH(r, &idrange_list, link) if (r->type == type) break; } while(r != NULL); /* find a range */ if (TAILQ_EMPTY(&idrange_list)) r = NULL; else { r = TAILQ_FIRST(&idrange_list); if (r->base < size) { while((r1 = TAILQ_NEXT(r, link)) != NULL) { if (r1->base - (r->base + r->size) >= size) break; r = r1; } r = r1; } if (r == NULL) { r1 = TAILQ_LAST(&idrange_list, idrange_list); if (INT32_MAX - size + 1 < r1->base + r1->size) { syslog(LOG_ERR, "out of id ranges (%u)", size); return (0); } } } /* allocate structure */ if ((r1 = malloc(sizeof(struct idrange))) == NULL) { syslog(LOG_ERR, "%s: %m", __FUNCTION__); return (0); } r1->type = type; r1->size = size; r1->owner = mod; if (TAILQ_EMPTY(&idrange_list) || r == TAILQ_FIRST(&idrange_list)) { r1->base = 0; TAILQ_INSERT_HEAD(&idrange_list, r1, link); } else if (r == NULL) { r = TAILQ_LAST(&idrange_list, idrange_list); r1->base = r->base + r->size; TAILQ_INSERT_TAIL(&idrange_list, r1, link); } else { r = TAILQ_PREV(r, idrange_list, link); r1->base = r->base + r->size; TAILQ_INSERT_AFTER(&idrange_list, r, r1, link); } r1->next = r1->base; return (type); } int32_t reqid_next(u_int type) { struct idrange *r; int32_t id; TAILQ_FOREACH(r, &idrange_list, link) if (r->type == type) break; if (r == NULL) { syslog(LOG_CRIT, "wrong idrange type"); abort(); } if ((id = r->next++) == r->base + (r->size - 1)) r->next = r->base; return (id); } int32_t reqid_base(u_int type) { struct idrange *r; TAILQ_FOREACH(r, &idrange_list, link) if (r->type == type) return (r->base); syslog(LOG_CRIT, "wrong idrange type"); abort(); } u_int reqid_type(int32_t reqid) { struct idrange *r; TAILQ_FOREACH(r, &idrange_list, link) if (reqid >= r->base && reqid <= r->base + (r->size - 1)) return (r->type); return (0); } int reqid_istype(int32_t reqid, u_int type) { return (reqid_type(reqid) == type); } /* * Delete all communities allocated by a module */ static void reqid_flush(struct lmodule *mod) { struct idrange *p, *p1; p = TAILQ_FIRST(&idrange_list); while (p != NULL) { p1 = TAILQ_NEXT(p, link); if (p->owner == mod) { TAILQ_REMOVE(&idrange_list, p, link); free(p); } p = p1; } } /* * Merge the given tree for the given module into the main tree. */ static int compare_node(const void *v1, const void *v2) { const struct snmp_node *n1 = v1; const struct snmp_node *n2 = v2; return (asn_compare_oid(&n1->oid, &n2->oid)); } static int tree_merge(const struct snmp_node *ntree, u_int nsize, struct lmodule *mod) { struct snmp_node *xtree; u_int i; xtree = realloc(tree, sizeof(*tree) * (tree_size + nsize)); if (xtree == NULL) { syslog(LOG_ERR, "tree_merge: %m"); return (-1); } tree = xtree; memcpy(&tree[tree_size], ntree, sizeof(*tree) * nsize); for (i = 0; i < nsize; i++) tree[tree_size + i].tree_data = mod; tree_size += nsize; qsort(tree, tree_size, sizeof(tree[0]), compare_node); return (0); } /* * Remove all nodes belonging to the loadable module */ static void tree_unmerge(struct lmodule *mod) { u_int s, d; for(s = d = 0; s < tree_size; s++) if (tree[s].tree_data != mod) { if (s != d) tree[d] = tree[s]; d++; } tree_size = d; } /* * Loadable modules */ struct lmodule * lm_load(const char *path, const char *section) { struct lmodule *m; int err; int i; char *av[MAX_MOD_ARGS + 1]; int ac; u_int u; if ((m = malloc(sizeof(*m))) == NULL) { syslog(LOG_ERR, "lm_load: %m"); return (NULL); } m->handle = NULL; m->flags = 0; - strcpy(m->section, section); + strlcpy(m->section, section, sizeof(m->section)); - if ((m->path = malloc(strlen(path) + 1)) == NULL) { + if ((m->path = strdup(path)) == NULL) { syslog(LOG_ERR, "lm_load: %m"); goto err; } - strcpy(m->path, path); /* * Make index */ m->index.subs[0] = strlen(section); m->index.len = m->index.subs[0] + 1; for (u = 0; u < m->index.subs[0]; u++) m->index.subs[u + 1] = section[u]; /* * Load the object file and locate the config structure */ if ((m->handle = dlopen(m->path, RTLD_NOW|RTLD_GLOBAL)) == NULL) { syslog(LOG_ERR, "lm_load: open %s", dlerror()); goto err; } if ((m->config = dlsym(m->handle, "config")) == NULL) { syslog(LOG_ERR, "lm_load: no 'config' symbol %s", dlerror()); goto err; } /* * Insert it into the right place */ INSERT_OBJECT_OID(m, &lmodules); /* preserve order */ if (community == COMM_INITIALIZE) { m->flags |= LM_ONSTARTLIST; TAILQ_INSERT_TAIL(&modules_start, m, start); } /* * make the argument vector. */ ac = 0; for (i = 0; i < nprogargs; i++) { if (strlen(progargs[i]) >= strlen(section) + 1 && strncmp(progargs[i], section, strlen(section)) == 0 && progargs[i][strlen(section)] == ':') { if (ac == MAX_MOD_ARGS) { syslog(LOG_WARNING, "too many arguments for " "module '%s", section); break; } av[ac++] = &progargs[i][strlen(section)+1]; } } av[ac] = NULL; /* * Run the initialization function */ if ((err = (*m->config->init)(m, ac, av)) != 0) { syslog(LOG_ERR, "lm_load: init failed: %d", err); TAILQ_REMOVE(&lmodules, m, link); goto err; } return (m); err: if ((m->flags & LM_ONSTARTLIST) != 0) TAILQ_REMOVE(&modules_start, m, start); if (m->handle) dlclose(m->handle); free(m->path); free(m); return (NULL); } /* * Start a module */ void lm_start(struct lmodule *mod) { const struct lmodule *m; /* * Merge tree. If this fails, unload the module. */ if (tree_merge(mod->config->tree, mod->config->tree_size, mod)) { lm_unload(mod); return; } /* * Read configuration */ if (read_config(config_file, mod)) { syslog(LOG_ERR, "error in config file"); lm_unload(mod); return; } if (mod->config->start) (*mod->config->start)(); mod->flags |= LM_STARTED; /* * Inform other modules */ TAILQ_FOREACH(m, &lmodules, link) if (m->config->loading) (*m->config->loading)(mod, 1); } /* * Unload a module. */ void lm_unload(struct lmodule *m) { int err; const struct lmodule *mod; TAILQ_REMOVE(&lmodules, m, link); if (m->flags & LM_ONSTARTLIST) TAILQ_REMOVE(&modules_start, m, start); tree_unmerge(m); if ((m->flags & LM_STARTED) && m->config->fini && (err = (*m->config->fini)()) != 0) syslog(LOG_WARNING, "lm_unload(%s): fini %d", m->section, err); comm_flush(m); reqid_flush(m); timer_flush(m); fd_flush(m); dlclose(m->handle); free(m->path); /* * Inform other modules */ TAILQ_FOREACH(mod, &lmodules, link) if (mod->config->loading) (*mod->config->loading)(m, 0); free(m); } /* * Register an object resource and return the index (or 0 on failures) */ u_int or_register(const struct asn_oid *or, const char *descr, struct lmodule *mod) { struct objres *objres, *or1; u_int idx; /* find a free index */ idx = 1; for (objres = TAILQ_FIRST(&objres_list); objres != NULL; objres = TAILQ_NEXT(objres, link)) { if ((or1 = TAILQ_NEXT(objres, link)) == NULL || or1->index > objres->index + 1) { idx = objres->index + 1; break; } } if ((objres = malloc(sizeof(*objres))) == NULL) return (0); objres->index = idx; objres->oid = *or; strlcpy(objres->descr, descr, sizeof(objres->descr)); objres->uptime = (uint32_t)(get_ticks() - start_tick); objres->module = mod; INSERT_OBJECT_INT(objres, &objres_list); systemg.or_last_change = objres->uptime; return (idx); } void or_unregister(u_int idx) { struct objres *objres; TAILQ_FOREACH(objres, &objres_list, link) if (objres->index == idx) { TAILQ_REMOVE(&objres_list, objres, link); free(objres); return; } } /* * RFC 3414 User-based Security Model support */ struct snmpd_usmstat * bsnmpd_get_usm_stats(void) { return (&snmpd_usmstats); } void bsnmpd_reset_usm_stats(void) { memset(&snmpd_usmstats, 0, sizeof(snmpd_usmstats)); } struct usm_user * usm_first_user(void) { return (SLIST_FIRST(&usm_userlist)); } struct usm_user * usm_next_user(struct usm_user *uuser) { if (uuser == NULL) return (NULL); return (SLIST_NEXT(uuser, up)); } struct usm_user * usm_find_user(uint8_t *engine, uint32_t elen, char *uname) { struct usm_user *uuser; SLIST_FOREACH(uuser, &usm_userlist, up) if (uuser->user_engine_len == elen && memcmp(uuser->user_engine_id, engine, elen) == 0 && strlen(uuser->suser.sec_name) == strlen(uname) && strcmp(uuser->suser.sec_name, uname) == 0) break; return (uuser); } static int usm_compare_user(struct usm_user *u1, struct usm_user *u2) { uint32_t i; if (u1->user_engine_len < u2->user_engine_len) return (-1); if (u1->user_engine_len > u2->user_engine_len) return (1); for (i = 0; i < u1->user_engine_len; i++) { if (u1->user_engine_id[i] < u2->user_engine_id[i]) return (-1); if (u1->user_engine_id[i] > u2->user_engine_id[i]) return (1); } if (strlen(u1->suser.sec_name) < strlen(u2->suser.sec_name)) return (-1); if (strlen(u1->suser.sec_name) > strlen(u2->suser.sec_name)) return (1); for (i = 0; i < strlen(u1->suser.sec_name); i++) { if (u1->suser.sec_name[i] < u2->suser.sec_name[i]) return (-1); if (u1->suser.sec_name[i] > u2->suser.sec_name[i]) return (1); } return (0); } struct usm_user * usm_new_user(uint8_t *eid, uint32_t elen, char *uname) { int cmp; struct usm_user *uuser, *temp, *prev; for (uuser = usm_first_user(); uuser != NULL; (uuser = usm_next_user(uuser))) { if (uuser->user_engine_len == elen && strlen(uname) == strlen(uuser->suser.sec_name) && strcmp(uname, uuser->suser.sec_name) == 0 && memcmp(eid, uuser->user_engine_id, elen) == 0) return (NULL); } if ((uuser = (struct usm_user *)malloc(sizeof(*uuser))) == NULL) return (NULL); memset(uuser, 0, sizeof(*uuser)); strlcpy(uuser->suser.sec_name, uname, SNMP_ADM_STR32_SIZ); memcpy(uuser->user_engine_id, eid, elen); uuser->user_engine_len = elen; if ((prev = SLIST_FIRST(&usm_userlist)) == NULL || usm_compare_user(uuser, prev) < 0) { SLIST_INSERT_HEAD(&usm_userlist, uuser, up); return (uuser); } SLIST_FOREACH(temp, &usm_userlist, up) { if ((cmp = usm_compare_user(uuser, temp)) <= 0) break; prev = temp; } if (temp == NULL || cmp < 0) SLIST_INSERT_AFTER(prev, uuser, up); else if (cmp > 0) SLIST_INSERT_AFTER(temp, uuser, up); else { syslog(LOG_ERR, "User %s exists", uuser->suser.sec_name); free(uuser); return (NULL); } return (uuser); } void usm_delete_user(struct usm_user *uuser) { SLIST_REMOVE(&usm_userlist, uuser, usm_user, up); free(uuser); } void usm_flush_users(void) { struct usm_user *uuser; while ((uuser = SLIST_FIRST(&usm_userlist)) != NULL) { SLIST_REMOVE_HEAD(&usm_userlist, up); free(uuser); } SLIST_INIT(&usm_userlist); } /* * RFC 3415 View-based Access Control Model support */ struct vacm_user * vacm_first_user(void) { return (SLIST_FIRST(&vacm_userlist)); } struct vacm_user * vacm_next_user(struct vacm_user *vuser) { if (vuser == NULL) return (NULL); return (SLIST_NEXT(vuser, vvu)); } static int vacm_compare_user(struct vacm_user *v1, struct vacm_user *v2) { uint32_t i; if (v1->sec_model < v2->sec_model) return (-1); if (v1->sec_model > v2->sec_model) return (1); if (strlen(v1->secname) < strlen(v2->secname)) return (-1); if (strlen(v1->secname) > strlen(v2->secname)) return (1); for (i = 0; i < strlen(v1->secname); i++) { if (v1->secname[i] < v2->secname[i]) return (-1); if (v1->secname[i] > v2->secname[i]) return (1); } return (0); } struct vacm_user * vacm_new_user(int32_t smodel, char *uname) { int cmp; struct vacm_user *user, *temp, *prev; SLIST_FOREACH(user, &vacm_userlist, vvu) if (strcmp(uname, user->secname) == 0 && smodel == user->sec_model) return (NULL); if ((user = (struct vacm_user *)malloc(sizeof(*user))) == NULL) return (NULL); memset(user, 0, sizeof(*user)); user->group = &vacm_default_group; SLIST_INSERT_HEAD(&vacm_default_group.group_users, user, vvg); user->sec_model = smodel; strlcpy(user->secname, uname, sizeof(user->secname)); if ((prev = SLIST_FIRST(&vacm_userlist)) == NULL || vacm_compare_user(user, prev) < 0) { SLIST_INSERT_HEAD(&vacm_userlist, user, vvu); return (user); } SLIST_FOREACH(temp, &vacm_userlist, vvu) { if ((cmp = vacm_compare_user(user, temp)) <= 0) break; prev = temp; } if (temp == NULL || cmp < 0) SLIST_INSERT_AFTER(prev, user, vvu); else if (cmp > 0) SLIST_INSERT_AFTER(temp, user, vvu); else { syslog(LOG_ERR, "User %s exists", user->secname); free(user); return (NULL); } return (user); } int vacm_delete_user(struct vacm_user *user) { if (user->group != NULL && user->group != &vacm_default_group) { SLIST_REMOVE(&user->group->group_users, user, vacm_user, vvg); if (SLIST_EMPTY(&user->group->group_users)) { SLIST_REMOVE(&vacm_grouplist, user->group, vacm_group, vge); free(user->group); } } SLIST_REMOVE(&vacm_userlist, user, vacm_user, vvu); free(user); return (0); } int vacm_user_set_group(struct vacm_user *user, u_char *octets, u_int len) { struct vacm_group *group; if (len >= SNMP_ADM_STR32_SIZ) return (-1); SLIST_FOREACH(group, &vacm_grouplist, vge) if (strlen(group->groupname) == len && memcmp(octets, group->groupname, len) == 0) break; if (group == NULL) { if ((group = (struct vacm_group *)malloc(sizeof(*group))) == NULL) return (-1); memset(group, 0, sizeof(*group)); memcpy(group->groupname, octets, len); group->groupname[len] = '\0'; SLIST_INSERT_HEAD(&vacm_grouplist, group, vge); } SLIST_REMOVE(&user->group->group_users, user, vacm_user, vvg); SLIST_INSERT_HEAD(&group->group_users, user, vvg); user->group = group; return (0); } void vacm_groups_init(void) { SLIST_INSERT_HEAD(&vacm_grouplist, &vacm_default_group, vge); } struct vacm_access * vacm_first_access_rule(void) { return (TAILQ_FIRST(&vacm_accesslist)); } struct vacm_access * vacm_next_access_rule(struct vacm_access *acl) { if (acl == NULL) return (NULL); return (TAILQ_NEXT(acl, vva)); } static int vacm_compare_access_rule(struct vacm_access *v1, struct vacm_access *v2) { uint32_t i; if (strlen(v1->group->groupname) < strlen(v2->group->groupname)) return (-1); if (strlen(v1->group->groupname) > strlen(v2->group->groupname)) return (1); for (i = 0; i < strlen(v1->group->groupname); i++) { if (v1->group->groupname[i] < v2->group->groupname[i]) return (-1); if (v1->group->groupname[i] > v2->group->groupname[i]) return (1); } if (strlen(v1->ctx_prefix) < strlen(v2->ctx_prefix)) return (-1); if (strlen(v1->ctx_prefix) > strlen(v2->ctx_prefix)) return (1); for (i = 0; i < strlen(v1->ctx_prefix); i++) { if (v1->ctx_prefix[i] < v2->ctx_prefix[i]) return (-1); if (v1->ctx_prefix[i] > v2->ctx_prefix[i]) return (1); } if (v1->sec_model < v2->sec_model) return (-1); if (v1->sec_model > v2->sec_model) return (1); if (v1->sec_level < v2->sec_level) return (-1); if (v1->sec_level > v2->sec_level) return (1); return (0); } struct vacm_access * vacm_new_access_rule(char *gname, char *cprefix, int32_t smodel, int32_t slevel) { struct vacm_group *group; struct vacm_access *acl, *temp; TAILQ_FOREACH(acl, &vacm_accesslist, vva) { if (acl->group == NULL) continue; if (strcmp(gname, acl->group->groupname) == 0 && strcmp(cprefix, acl->ctx_prefix) == 0 && acl->sec_model == smodel && acl->sec_level == slevel) return (NULL); } /* Make sure the group exists */ SLIST_FOREACH(group, &vacm_grouplist, vge) if (strcmp(gname, group->groupname) == 0) break; if (group == NULL) return (NULL); if ((acl = (struct vacm_access *)malloc(sizeof(*acl))) == NULL) return (NULL); memset(acl, 0, sizeof(*acl)); acl->group = group; strlcpy(acl->ctx_prefix, cprefix, sizeof(acl->ctx_prefix)); acl->sec_model = smodel; acl->sec_level = slevel; if ((temp = TAILQ_FIRST(&vacm_accesslist)) == NULL || vacm_compare_access_rule(acl, temp) < 0) { TAILQ_INSERT_HEAD(&vacm_accesslist, acl, vva); return (acl); } TAILQ_FOREACH(temp, &vacm_accesslist, vva) if (vacm_compare_access_rule(acl, temp) < 0) { TAILQ_INSERT_BEFORE(temp, acl, vva); return (acl); } TAILQ_INSERT_TAIL(&vacm_accesslist, acl, vva); return (acl); } int vacm_delete_access_rule(struct vacm_access *acl) { TAILQ_REMOVE(&vacm_accesslist, acl, vva); free(acl); return (0); } struct vacm_view * vacm_first_view(void) { return (SLIST_FIRST(&vacm_viewlist)); } struct vacm_view * vacm_next_view(struct vacm_view *view) { if (view == NULL) return (NULL); return (SLIST_NEXT(view, vvl)); } static int vacm_compare_view(struct vacm_view *v1, struct vacm_view *v2) { uint32_t i; if (strlen(v1->viewname) < strlen(v2->viewname)) return (-1); if (strlen(v1->viewname) > strlen(v2->viewname)) return (1); for (i = 0; i < strlen(v1->viewname); i++) { if (v1->viewname[i] < v2->viewname[i]) return (-1); if (v1->viewname[i] > v2->viewname[i]) return (1); } return (asn_compare_oid(&v1->subtree, &v2->subtree)); } struct vacm_view * vacm_new_view(char *vname, struct asn_oid *oid) { int cmp; struct vacm_view *view, *temp, *prev; SLIST_FOREACH(view, &vacm_viewlist, vvl) if (strcmp(vname, view->viewname) == 0) return (NULL); if ((view = (struct vacm_view *)malloc(sizeof(*view))) == NULL) return (NULL); memset(view, 0, sizeof(*view)); strlcpy(view->viewname, vname, sizeof(view->viewname)); asn_append_oid(&view->subtree, oid); if ((prev = SLIST_FIRST(&vacm_viewlist)) == NULL || vacm_compare_view(view, prev) < 0) { SLIST_INSERT_HEAD(&vacm_viewlist, view, vvl); return (view); } SLIST_FOREACH(temp, &vacm_viewlist, vvl) { if ((cmp = vacm_compare_view(view, temp)) <= 0) break; prev = temp; } if (temp == NULL || cmp < 0) SLIST_INSERT_AFTER(prev, view, vvl); else if (cmp > 0) SLIST_INSERT_AFTER(temp, view, vvl); else { syslog(LOG_ERR, "View %s exists", view->viewname); free(view); return (NULL); } return (view); } int vacm_delete_view(struct vacm_view *view) { SLIST_REMOVE(&vacm_viewlist, view, vacm_view, vvl); free(view); return (0); } struct vacm_context * vacm_first_context(void) { return (SLIST_FIRST(&vacm_contextlist)); } struct vacm_context * vacm_next_context(struct vacm_context *vacmctx) { if (vacmctx == NULL) return (NULL); return (SLIST_NEXT(vacmctx, vcl)); } struct vacm_context * vacm_add_context(char *ctxname, int regid) { int cmp; struct vacm_context *ctx, *temp, *prev; SLIST_FOREACH(ctx, &vacm_contextlist, vcl) if (strcmp(ctxname, ctx->ctxname) == 0) { syslog(LOG_ERR, "Context %s exists", ctx->ctxname); return (NULL); } if ((ctx = (struct vacm_context *)malloc(sizeof(*ctx))) == NULL) return (NULL); memset(ctx, 0, sizeof(*ctx)); strlcpy(ctx->ctxname, ctxname, sizeof(ctx->ctxname)); ctx->regid = regid; if ((prev = SLIST_FIRST(&vacm_contextlist)) == NULL || strlen(ctx->ctxname) < strlen(prev->ctxname) || strcmp(ctx->ctxname, prev->ctxname) < 0) { SLIST_INSERT_HEAD(&vacm_contextlist, ctx, vcl); return (ctx); } SLIST_FOREACH(temp, &vacm_contextlist, vcl) { if (strlen(ctx->ctxname) < strlen(temp->ctxname) || strcmp(ctx->ctxname, temp->ctxname) < 0) { cmp = -1; break; } prev = temp; } if (temp == NULL || cmp < 0) SLIST_INSERT_AFTER(prev, ctx, vcl); else if (cmp > 0) SLIST_INSERT_AFTER(temp, ctx, vcl); else { syslog(LOG_ERR, "Context %s exists", ctx->ctxname); free(ctx); return (NULL); } return (ctx); } void vacm_flush_contexts(int regid) { struct vacm_context *ctx, *temp; SLIST_FOREACH_SAFE(ctx, &vacm_contextlist, vcl, temp) if (ctx->regid == regid) { SLIST_REMOVE(&vacm_contextlist, ctx, vacm_context, vcl); free(ctx); } } Index: projects/clang400-import/contrib/bsnmp/snmpd/trans_lsock.c =================================================================== --- projects/clang400-import/contrib/bsnmp/snmpd/trans_lsock.c (revision 311459) +++ projects/clang400-import/contrib/bsnmp/snmpd/trans_lsock.c (revision 311460) @@ -1,674 +1,669 @@ /* * Copyright (c) 2003 * Fraunhofer Institute for Open Communication Systems (FhG Fokus). * All rights reserved. * * Author: Harti Brandt * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $Begemot: bsnmp/snmpd/trans_lsock.c,v 1.6 2005/02/25 11:50:25 brandt_h Exp $ * * Local domain socket transport */ #include #include #include #include #include #include #include #include #include #include #include #include #include "snmpmod.h" #include "snmpd.h" #include "trans_lsock.h" #include "tree.h" #include "oid.h" static const struct asn_oid oid_begemotSnmpdLocalPortTable = OIDX_begemotSnmpdLocalPortTable; static int lsock_start(void); static int lsock_stop(int); static void lsock_close_port(struct tport *); static int lsock_init_port(struct tport *); static ssize_t lsock_send(struct tport *, const u_char *, size_t, const struct sockaddr *, size_t); static ssize_t lsock_recv(struct tport *, struct port_input *); /* exported */ const struct transport_def lsock_trans = { "lsock", OIDX_begemotSnmpdTransLsock, lsock_start, lsock_stop, lsock_close_port, lsock_init_port, lsock_send, lsock_recv }; static struct transport *my_trans; static int lsock_remove(struct tport *tp, intptr_t arg __unused) { struct lsock_port *port = (struct lsock_port *)tp; (void)remove(port->name); return (-1); } static int lsock_stop(int force) { if (my_trans != NULL) { if (!force && trans_first_port(my_trans) != NULL) return (SNMP_ERR_GENERR); trans_iter_port(my_trans, lsock_remove, 0); return (trans_unregister(my_trans)); } return (SNMP_ERR_NOERROR); } static int lsock_start(void) { return (trans_register(&lsock_trans, &my_trans)); } /* * Open a local port. If this is a datagram socket create also the * one and only peer. */ static int lsock_open_port(u_char *name, size_t namelen, struct lsock_port **pp, int type) { struct lsock_port *port; struct lsock_peer *peer = NULL; int is_stream, need_cred; size_t u; int err; struct sockaddr_un sa; if (namelen == 0 || namelen + 1 > sizeof(sa.sun_path)) return (SNMP_ERR_BADVALUE); switch (type) { case LOCP_DGRAM_UNPRIV: is_stream = 0; need_cred = 0; break; case LOCP_DGRAM_PRIV: is_stream = 0; need_cred = 1; break; case LOCP_STREAM_UNPRIV: is_stream = 1; need_cred = 0; break; case LOCP_STREAM_PRIV: is_stream = 1; need_cred = 1; break; default: return (SNMP_ERR_BADVALUE); } - if ((port = malloc(sizeof(*port))) == NULL) + if ((port = calloc(1, sizeof(*port))) == NULL) return (SNMP_ERR_GENERR); - memset(port, 0, sizeof(*port)); if (!is_stream) { - if ((peer = malloc(sizeof(*peer))) == NULL) { + if ((peer = calloc(1, sizeof(*peer))) == NULL) { free(port); return (SNMP_ERR_GENERR); } - memset(peer, 0, sizeof(*peer)); } if ((port->name = malloc(namelen + 1)) == NULL) { free(port); if (!is_stream) free(peer); return (SNMP_ERR_GENERR); } strncpy(port->name, name, namelen); port->name[namelen] = '\0'; port->type = type; port->str_sock = -1; LIST_INIT(&port->peers); port->tport.index.len = namelen + 1; port->tport.index.subs[0] = namelen; for (u = 0; u < namelen; u++) port->tport.index.subs[u + 1] = name[u]; if (peer != NULL) { LIST_INSERT_HEAD(&port->peers, peer, link); peer->port = port; peer->input.fd = -1; peer->input.id = NULL; peer->input.stream = is_stream; peer->input.cred = need_cred; peer->input.peer = (struct sockaddr *)&peer->peer; } trans_insert_port(my_trans, &port->tport); if (community != COMM_INITIALIZE && (err = lsock_init_port(&port->tport)) != SNMP_ERR_NOERROR) { lsock_close_port(&port->tport); return (err); } *pp = port; return (SNMP_ERR_NOERROR); } /* * Close a local domain peer */ static void lsock_peer_close(struct lsock_peer *peer) { LIST_REMOVE(peer, link); snmpd_input_close(&peer->input); free(peer); } /* * Close a local port */ static void lsock_close_port(struct tport *tp) { struct lsock_port *port = (struct lsock_port *)tp; struct lsock_peer *peer; if (port->str_id != NULL) fd_deselect(port->str_id); if (port->str_sock >= 0) (void)close(port->str_sock); (void)remove(port->name); trans_remove_port(tp); while ((peer = LIST_FIRST(&port->peers)) != NULL) lsock_peer_close(peer); free(port->name); free(port); } /* * Input on a local socket (either datagram or stream) */ static void lsock_input(int fd __unused, void *udata) { struct lsock_peer *peer = udata; struct lsock_port *p = peer->port; peer->input.peerlen = sizeof(peer->peer); if (snmpd_input(&peer->input, &p->tport) == -1 && peer->input.stream) /* framing or other input error */ lsock_peer_close(peer); } /* * A UNIX domain listening socket is ready. This means we have a peer * that we need to accept */ static void lsock_listen_input(int fd, void *udata) { struct lsock_port *p = udata; struct lsock_peer *peer; - if ((peer = malloc(sizeof(*peer))) == NULL) { + if ((peer = calloc(1, sizeof(*peer))) == NULL) { syslog(LOG_WARNING, "%s: peer malloc failed", p->name); (void)close(accept(fd, NULL, NULL)); return; } - memset(peer, 0, sizeof(*peer)); peer->port = p; peer->input.stream = 1; peer->input.cred = (p->type == LOCP_DGRAM_PRIV || p->type == LOCP_STREAM_PRIV); peer->input.peerlen = sizeof(peer->peer); peer->input.peer = (struct sockaddr *)&peer->peer; peer->input.fd = accept(fd, peer->input.peer, &peer->input.peerlen); if (peer->input.fd == -1) { syslog(LOG_WARNING, "%s: accept failed: %m", p->name); free(peer); return; } if ((peer->input.id = fd_select(peer->input.fd, lsock_input, peer, NULL)) == NULL) { close(peer->input.fd); free(peer); return; } LIST_INSERT_HEAD(&p->peers, peer, link); } /* * Create a local socket */ static int lsock_init_port(struct tport *tp) { struct lsock_port *p = (struct lsock_port *)tp; struct sockaddr_un sa; if (p->type == LOCP_STREAM_PRIV || p->type == LOCP_STREAM_UNPRIV) { if ((p->str_sock = socket(PF_LOCAL, SOCK_STREAM, 0)) < 0) { syslog(LOG_ERR, "creating local socket: %m"); return (SNMP_ERR_RES_UNAVAIL); } - strcpy(sa.sun_path, p->name); + strlcpy(sa.sun_path, p->name, sizeof(sa.sun_path)); sa.sun_family = AF_LOCAL; - sa.sun_len = strlen(p->name) + - offsetof(struct sockaddr_un, sun_path); + sa.sun_len = SUN_LEN(&sa); (void)remove(p->name); if (bind(p->str_sock, (struct sockaddr *)&sa, sizeof(sa))) { if (errno == EADDRNOTAVAIL) { close(p->str_sock); p->str_sock = -1; return (SNMP_ERR_INCONS_NAME); } syslog(LOG_ERR, "bind: %s %m", p->name); close(p->str_sock); p->str_sock = -1; return (SNMP_ERR_GENERR); } if (chmod(p->name, 0666) == -1) syslog(LOG_WARNING, "chmod(%s,0666): %m", p->name); if (listen(p->str_sock, 10) == -1) { syslog(LOG_ERR, "listen: %s %m", p->name); (void)remove(p->name); close(p->str_sock); p->str_sock = -1; return (SNMP_ERR_GENERR); } p->str_id = fd_select(p->str_sock, lsock_listen_input, p, NULL); if (p->str_id == NULL) { (void)remove(p->name); close(p->str_sock); p->str_sock = -1; return (SNMP_ERR_GENERR); } } else { struct lsock_peer *peer; const int on = 1; peer = LIST_FIRST(&p->peers); if ((peer->input.fd = socket(PF_LOCAL, SOCK_DGRAM, 0)) < 0) { syslog(LOG_ERR, "creating local socket: %m"); return (SNMP_ERR_RES_UNAVAIL); } if (setsockopt(peer->input.fd, 0, LOCAL_CREDS, &on, sizeof(on)) == -1) { syslog(LOG_ERR, "setsockopt(LOCAL_CREDS): %m"); close(peer->input.fd); peer->input.fd = -1; return (SNMP_ERR_GENERR); } - strcpy(sa.sun_path, p->name); + strlcpy(sa.sun_path, p->name, sizeof(sa.sun_path)); sa.sun_family = AF_LOCAL; - sa.sun_len = strlen(p->name) + - offsetof(struct sockaddr_un, sun_path); + sa.sun_len = SUN_LEN(&sa); (void)remove(p->name); if (bind(peer->input.fd, (struct sockaddr *)&sa, sizeof(sa))) { if (errno == EADDRNOTAVAIL) { close(peer->input.fd); peer->input.fd = -1; return (SNMP_ERR_INCONS_NAME); } syslog(LOG_ERR, "bind: %s %m", p->name); close(peer->input.fd); peer->input.fd = -1; return (SNMP_ERR_GENERR); } if (chmod(p->name, 0666) == -1) syslog(LOG_WARNING, "chmod(%s,0666): %m", p->name); peer->input.id = fd_select(peer->input.fd, lsock_input, peer, NULL); if (peer->input.id == NULL) { (void)remove(p->name); close(peer->input.fd); peer->input.fd = -1; return (SNMP_ERR_GENERR); } } return (SNMP_ERR_NOERROR); } /* * Send something */ static ssize_t lsock_send(struct tport *tp, const u_char *buf, size_t len, const struct sockaddr *addr, size_t addrlen) { struct lsock_port *p = (struct lsock_port *)tp; struct lsock_peer *peer; if (p->type == LOCP_DGRAM_PRIV || p->type == LOCP_DGRAM_UNPRIV) { peer = LIST_FIRST(&p->peers); } else { /* search for the peer */ LIST_FOREACH(peer, &p->peers, link) if (peer->input.peerlen == addrlen && memcmp(peer->input.peer, addr, addrlen) == 0) break; if (peer == NULL) { errno = ENOTCONN; return (-1); } } return (sendto(peer->input.fd, buf, len, 0, addr, addrlen)); } static void check_priv_stream(struct port_input *pi) { struct xucred ucred; socklen_t ucredlen; /* obtain the accept time credentials */ ucredlen = sizeof(ucred); if (getsockopt(pi->fd, 0, LOCAL_PEERCRED, &ucred, &ucredlen) == 0 && ucredlen >= sizeof(ucred) && ucred.cr_version == XUCRED_VERSION) pi->priv = (ucred.cr_uid == 0); else pi->priv = 0; } /* * Receive something */ static ssize_t lsock_recv(struct tport *tp __unused, struct port_input *pi) { struct msghdr msg; struct iovec iov[1]; ssize_t len; msg.msg_control = NULL; msg.msg_controllen = 0; if (pi->buf == NULL) { /* no buffer yet - allocate one */ if ((pi->buf = buf_alloc(0)) == NULL) { /* ups - could not get buffer. Return an error * the caller must close the transport. */ return (-1); } pi->buflen = buf_size(0); pi->consumed = 0; pi->length = 0; } /* try to get a message */ msg.msg_name = pi->peer; msg.msg_namelen = pi->peerlen; msg.msg_iov = iov; msg.msg_iovlen = 1; msg.msg_control = NULL; msg.msg_controllen = 0; msg.msg_flags = 0; iov[0].iov_base = pi->buf + pi->length; iov[0].iov_len = pi->buflen - pi->length; len = recvmsg(pi->fd, &msg, 0); if (len == -1 || len == 0) /* receive error */ return (-1); pi->length += len; if (pi->cred) check_priv_stream(pi); return (0); } /* * Dependency to create a lsock port */ struct lsock_dep { struct snmp_dependency dep; /* index (path name) */ u_char *path; size_t pathlen; /* the port */ struct lsock_port *port; /* which of the fields are set */ u_int set; /* type of the port */ int type; /* status */ int status; }; #define LD_TYPE 0x01 #define LD_STATUS 0x02 #define LD_CREATE 0x04 /* rollback create */ #define LD_DELETE 0x08 /* rollback delete */ /* * dependency handler for lsock ports */ static int lsock_func(struct snmp_context *ctx, struct snmp_dependency *dep, enum snmp_depop op) { struct lsock_dep *ld = (struct lsock_dep *)(void *)dep; int err = SNMP_ERR_NOERROR; switch (op) { case SNMP_DEPOP_COMMIT: if (!(ld->set & LD_STATUS)) err = SNMP_ERR_BADVALUE; else if (ld->port == NULL) { if (!ld->status) err = SNMP_ERR_BADVALUE; else { /* create */ err = lsock_open_port(ld->path, ld->pathlen, &ld->port, ld->type); if (err == SNMP_ERR_NOERROR) ld->set |= LD_CREATE; } } else if (!ld->status) { /* delete - hard to roll back so defer to finalizer */ ld->set |= LD_DELETE; } else /* modify - read-only */ err = SNMP_ERR_READONLY; return (err); case SNMP_DEPOP_ROLLBACK: if (ld->set & LD_CREATE) { /* was create */ lsock_close_port(&ld->port->tport); } return (SNMP_ERR_NOERROR); case SNMP_DEPOP_FINISH: if ((ld->set & LD_DELETE) && ctx->code == SNMP_RET_OK) lsock_close_port(&ld->port->tport); free(ld->path); return (SNMP_ERR_NOERROR); } abort(); } /* * Local port table */ int op_lsock_port(struct snmp_context *ctx, struct snmp_value *value, u_int sub, u_int iidx, enum snmp_op op) { asn_subid_t which = value->var.subs[sub-1]; struct lsock_port *p; u_char *name; size_t namelen; struct lsock_dep *ld; struct asn_oid didx; switch (op) { case SNMP_OP_GETNEXT: if ((p = (struct lsock_port *)trans_next_port(my_trans, &value->var, sub)) == NULL) return (SNMP_ERR_NOSUCHNAME); index_append(&value->var, sub, &p->tport.index); break; case SNMP_OP_GET: if ((p = (struct lsock_port *)trans_find_port(my_trans, &value->var, sub)) == NULL) return (SNMP_ERR_NOSUCHNAME); break; case SNMP_OP_SET: p = (struct lsock_port *)trans_find_port(my_trans, &value->var, sub); if (index_decode(&value->var, sub, iidx, &name, &namelen)) return (SNMP_ERR_NO_CREATION); asn_slice_oid(&didx, &value->var, sub, value->var.len); if ((ld = (struct lsock_dep *)(void *)snmp_dep_lookup(ctx, &oid_begemotSnmpdLocalPortTable, &didx, sizeof(*ld), lsock_func)) == NULL) { free(name); return (SNMP_ERR_GENERR); } if (ld->path == NULL) { ld->path = name; ld->pathlen = namelen; } else { free(name); } ld->port = p; switch (which) { case LEAF_begemotSnmpdLocalPortStatus: if (ld->set & LD_STATUS) return (SNMP_ERR_INCONS_VALUE); if (!TRUTH_OK(value->v.integer)) return (SNMP_ERR_WRONG_VALUE); ld->status = TRUTH_GET(value->v.integer); ld->set |= LD_STATUS; break; case LEAF_begemotSnmpdLocalPortType: if (ld->set & LD_TYPE) return (SNMP_ERR_INCONS_VALUE); if (value->v.integer < 1 || value->v.integer > 4) return (SNMP_ERR_WRONG_VALUE); ld->type = value->v.integer; ld->set |= LD_TYPE; break; } return (SNMP_ERR_NOERROR); case SNMP_OP_ROLLBACK: case SNMP_OP_COMMIT: return (SNMP_ERR_NOERROR); default: abort(); } /* * Come here to fetch the value */ switch (which) { case LEAF_begemotSnmpdLocalPortStatus: value->v.integer = 1; break; case LEAF_begemotSnmpdLocalPortType: value->v.integer = p->type; break; default: abort(); } return (SNMP_ERR_NOERROR); } Index: projects/clang400-import/contrib/netbsd-tests/fs/tmpfs/h_tools.c =================================================================== --- projects/clang400-import/contrib/netbsd-tests/fs/tmpfs/h_tools.c (revision 311459) +++ projects/clang400-import/contrib/netbsd-tests/fs/tmpfs/h_tools.c (revision 311460) @@ -1,319 +1,321 @@ /* $NetBSD: h_tools.c,v 1.4 2011/06/11 18:03:17 christos Exp $ */ /* * Copyright (c) 2005, 2006 The NetBSD Foundation, Inc. * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE * POSSIBILITY OF SUCH DAMAGE. */ /* * Helper tools for several tests. These are kept in a single file due * to the limitations of bsd.prog.mk to build a single program in a * given directory. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef __FreeBSD__ #include #endif /* --------------------------------------------------------------------- */ static int getfh_main(int, char **); static int kqueue_main(int, char **); static int rename_main(int, char **); static int sockets_main(int, char **); static int statvfs_main(int, char **); /* --------------------------------------------------------------------- */ int getfh_main(int argc, char **argv) { int error; void *fh; size_t fh_size; if (argc < 2) return EXIT_FAILURE; #ifdef __FreeBSD__ fh_size = sizeof(fhandle_t); #else fh_size = 0; #endif fh = NULL; for (;;) { if (fh_size) { fh = malloc(fh_size); if (fh == NULL) { fprintf(stderr, "out of memory"); return EXIT_FAILURE; } } /* * The kernel provides the necessary size in fh_size - * but it may change if someone moves things around, * so retry untill we have enough memory. */ #ifdef __FreeBSD__ error = getfh(argv[1], fh); #else error = getfh(argv[1], fh, &fh_size); #endif if (error == 0) { break; } else { if (fh != NULL) free(fh); if (errno != E2BIG) { warn("getfh"); return EXIT_FAILURE; } } } error = write(STDOUT_FILENO, fh, fh_size); if (error == -1) { warn("write"); return EXIT_FAILURE; } free(fh); return 0; } /* --------------------------------------------------------------------- */ int kqueue_main(int argc, char **argv) { char *line; int i, kq; size_t len; struct kevent *changes, event; if (argc < 2) return EXIT_FAILURE; argc--; argv++; changes = malloc(sizeof(struct kevent) * argc); if (changes == NULL) errx(EXIT_FAILURE, "not enough memory"); for (i = 0; i < argc; i++) { int fd; fd = open(argv[i], O_RDONLY); if (fd == -1) err(EXIT_FAILURE, "cannot open %s", argv[i]); EV_SET(&changes[i], fd, EVFILT_VNODE, EV_ADD | EV_ENABLE | EV_ONESHOT, NOTE_ATTRIB | NOTE_DELETE | NOTE_EXTEND | NOTE_LINK | NOTE_RENAME | NOTE_REVOKE | NOTE_WRITE, 0, 0); } kq = kqueue(); if (kq == -1) err(EXIT_FAILURE, "kqueue"); while ((line = fgetln(stdin, &len)) != NULL) { int ec, nev; struct timespec to; to.tv_sec = 0; to.tv_nsec = 100000; (void)kevent(kq, changes, argc, &event, 1, &to); assert(len > 0); assert(line[len - 1] == '\n'); line[len - 1] = '\0'; ec = system(line); if (ec != EXIT_SUCCESS) errx(ec, "%s returned %d", line, ec); do { nev = kevent(kq, changes, argc, &event, 1, &to); if (nev == -1) err(EXIT_FAILURE, "kevent"); else if (nev > 0) { for (i = 0; i < argc; i++) if (event.ident == changes[i].ident) break; if (event.fflags & NOTE_ATTRIB) printf("%s - NOTE_ATTRIB\n", argv[i]); if (event.fflags & NOTE_DELETE) printf("%s - NOTE_DELETE\n", argv[i]); if (event.fflags & NOTE_EXTEND) printf("%s - NOTE_EXTEND\n", argv[i]); if (event.fflags & NOTE_LINK) printf("%s - NOTE_LINK\n", argv[i]); if (event.fflags & NOTE_RENAME) printf("%s - NOTE_RENAME\n", argv[i]); if (event.fflags & NOTE_REVOKE) printf("%s - NOTE_REVOKE\n", argv[i]); if (event.fflags & NOTE_WRITE) printf("%s - NOTE_WRITE\n", argv[i]); } } while (nev > 0); } for (i = 0; i < argc; i++) close(changes[i].ident); free(changes); return EXIT_SUCCESS; } /* --------------------------------------------------------------------- */ int rename_main(int argc, char **argv) { if (argc < 3) return EXIT_FAILURE; if (rename(argv[1], argv[2]) == -1) { warn("rename"); return EXIT_FAILURE; } return EXIT_SUCCESS; } /* --------------------------------------------------------------------- */ int sockets_main(int argc, char **argv) { int error, fd; struct sockaddr_un addr; if (argc < 2) return EXIT_FAILURE; fd = socket(PF_LOCAL, SOCK_STREAM, 0); if (fd == -1) { warn("socket"); return EXIT_FAILURE; } #ifdef __FreeBSD__ - addr.sun_len = sizeof(addr.sun_path); - (void)strlcpy(addr.sun_path, argv[1], addr.sun_len); -#else - (void)strlcpy(addr.sun_path, argv[1], sizeof(addr.sun_path)); + memset(&addr, 0, sizeof(addr)); #endif + (void)strlcpy(addr.sun_path, argv[1], sizeof(addr.sun_path)); addr.sun_family = PF_UNIX; +#ifdef __FreeBSD__ + error = bind(fd, (struct sockaddr *)&addr, SUN_LEN(&addr)); +#else error = bind(fd, (struct sockaddr *)&addr, sizeof(addr)); +#endif if (error == -1) { warn("connect"); #ifdef __FreeBSD__ (void)close(fd); #endif return EXIT_FAILURE; } close(fd); return EXIT_SUCCESS; } /* --------------------------------------------------------------------- */ int statvfs_main(int argc, char **argv) { int error; struct statvfs buf; if (argc < 2) return EXIT_FAILURE; error = statvfs(argv[1], &buf); if (error != 0) { warn("statvfs"); return EXIT_FAILURE; } (void)printf("f_bsize=%lu\n", buf.f_bsize); (void)printf("f_blocks=%" PRId64 "\n", buf.f_blocks); (void)printf("f_bfree=%" PRId64 "\n", buf.f_bfree); (void)printf("f_files=%" PRId64 "\n", buf.f_files); return EXIT_SUCCESS; } /* --------------------------------------------------------------------- */ int main(int argc, char **argv) { int error; if (argc < 2) return EXIT_FAILURE; argc -= 1; argv += 1; if (strcmp(argv[0], "getfh") == 0) error = getfh_main(argc, argv); else if (strcmp(argv[0], "kqueue") == 0) error = kqueue_main(argc, argv); else if (strcmp(argv[0], "rename") == 0) error = rename_main(argc, argv); else if (strcmp(argv[0], "sockets") == 0) error = sockets_main(argc, argv); else if (strcmp(argv[0], "statvfs") == 0) error = statvfs_main(argc, argv); else error = EXIT_FAILURE; return error; } Index: projects/clang400-import/contrib/netbsd-tests =================================================================== --- projects/clang400-import/contrib/netbsd-tests (revision 311459) +++ projects/clang400-import/contrib/netbsd-tests (revision 311460) Property changes on: projects/clang400-import/contrib/netbsd-tests ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head/contrib/netbsd-tests:r311306-311459 Index: projects/clang400-import/contrib/tcp_wrappers/scaffold.c =================================================================== --- projects/clang400-import/contrib/tcp_wrappers/scaffold.c (revision 311459) +++ projects/clang400-import/contrib/tcp_wrappers/scaffold.c (revision 311460) @@ -1,272 +1,262 @@ /* * Routines for testing only. Not really industrial strength. * * Author: Wietse Venema, Eindhoven University of Technology, The Netherlands. * * $FreeBSD$ */ #ifndef lint static char sccs_id[] = "@(#) scaffold.c 1.6 97/03/21 19:27:24"; #endif /* System libraries. */ #include #include #include #include #include #include #include #include #include #include #ifndef INADDR_NONE #define INADDR_NONE (-1) /* XXX should be 0xffffffff */ #endif #ifndef INET6 extern char *malloc(); #endif /* Application-specific. */ #include "tcpd.h" #include "scaffold.h" /* * These are referenced by the options module and by rfc931.c. */ int allow_severity = SEVERITY; int deny_severity = LOG_WARNING; int rfc931_timeout = RFC931_TIMEOUT; #ifndef INET6 /* dup_hostent - create hostent in one memory block */ static struct hostent *dup_hostent(hp) struct hostent *hp; { struct hostent_block { struct hostent host; char *addr_list[1]; }; struct hostent_block *hb; int count; char *data; char *addr; for (count = 0; hp->h_addr_list[count] != 0; count++) /* void */ ; if ((hb = (struct hostent_block *) malloc(sizeof(struct hostent_block) + (hp->h_length + sizeof(char *)) * count)) == 0) { fprintf(stderr, "Sorry, out of memory\n"); exit(1); } memset((char *) &hb->host, 0, sizeof(hb->host)); hb->host.h_length = hp->h_length; hb->host.h_addr_list = hb->addr_list; hb->host.h_addr_list[count] = 0; data = (char *) (hb->host.h_addr_list + count + 1); for (count = 0; (addr = hp->h_addr_list[count]) != 0; count++) { hb->host.h_addr_list[count] = data + hp->h_length * count; memcpy(hb->host.h_addr_list[count], addr, hp->h_length); } return (&hb->host); } #endif /* find_inet_addr - find all addresses for this host, result to free() */ #ifdef INET6 struct addrinfo *find_inet_addr(host) char *host; { struct addrinfo hints, *res; memset(&hints, 0, sizeof(hints)); hints.ai_family = PF_UNSPEC; hints.ai_socktype = SOCK_STREAM; hints.ai_flags = AI_PASSIVE | AI_NUMERICHOST; if (getaddrinfo(host, NULL, &hints, &res) == 0) return (res); memset(&hints, 0, sizeof(hints)); hints.ai_family = PF_UNSPEC; hints.ai_socktype = SOCK_STREAM; hints.ai_flags = AI_PASSIVE | AI_CANONNAME; if (getaddrinfo(host, NULL, &hints, &res) != 0) { tcpd_warn("%s: host not found", host); return (0); } if (res->ai_family != AF_INET6 && res->ai_family != AF_INET) { tcpd_warn("%d: not an internet host", res->ai_family); freeaddrinfo(res); return (0); } if (!res->ai_canonname) { tcpd_warn("%s: hostname alias", host); tcpd_warn("(cannot obtain official name)", res->ai_canonname); } else if (STR_NE(host, res->ai_canonname)) { tcpd_warn("%s: hostname alias", host); tcpd_warn("(official name: %.*s)", STRING_LENGTH, res->ai_canonname); } return (res); } #else struct hostent *find_inet_addr(host) char *host; { struct in_addr addr; struct hostent *hp; static struct hostent h; static char *addr_list[2]; /* * Host address: translate it to internal form. */ if ((addr.s_addr = dot_quad_addr(host)) != INADDR_NONE) { h.h_addr_list = addr_list; h.h_addr_list[0] = (char *) &addr; h.h_length = sizeof(addr); return (dup_hostent(&h)); } /* * Map host name to a series of addresses. Watch out for non-internet * forms or aliases. The NOT_INADDR() is here in case gethostbyname() has * been "enhanced" to accept numeric addresses. Make a copy of the * address list so that later gethostbyXXX() calls will not clobber it. */ if (NOT_INADDR(host) == 0) { tcpd_warn("%s: not an internet address", host); return (0); } if ((hp = gethostbyname(host)) == 0) { tcpd_warn("%s: host not found", host); return (0); } if (hp->h_addrtype != AF_INET) { tcpd_warn("%d: not an internet host", hp->h_addrtype); return (0); } if (STR_NE(host, hp->h_name)) { tcpd_warn("%s: hostname alias", host); tcpd_warn("(official name: %.*s)", STRING_LENGTH, hp->h_name); } return (dup_hostent(hp)); } #endif /* check_dns - give each address thorough workout, return address count */ int check_dns(host) char *host; { struct request_info request; #ifdef INET6 struct sockaddr_storage sin; struct addrinfo *hp, *res; #else struct sockaddr_in sin; struct hostent *hp; #endif int count; char *addr; if ((hp = find_inet_addr(host)) == 0) return (0); request_init(&request, RQ_CLIENT_SIN, &sin, 0); sock_methods(&request); #ifndef INET6 memset((char *) &sin, 0, sizeof(sin)); sin.sin_family = AF_INET; #endif #ifdef INET6 for (res = hp, count = 0; res; res = res->ai_next, count++) { memcpy(&sin, res->ai_addr, res->ai_addrlen); #else for (count = 0; (addr = hp->h_addr_list[count]) != 0; count++) { memcpy((char *) &sin.sin_addr, addr, sizeof(sin.sin_addr)); #endif /* * Force host name and address conversions. Use the request structure * as a cache. Detect hostname lookup problems. Any name/name or * name/address conflicts will be reported while eval_hostname() does * its job. */ request_set(&request, RQ_CLIENT_ADDR, "", RQ_CLIENT_NAME, "", 0); if (STR_EQ(eval_hostname(request.client), unknown)) tcpd_warn("host address %s->name lookup failed", eval_hostaddr(request.client)); } #ifdef INET6 freeaddrinfo(hp); #else free((char *) hp); #endif return (count); } /* dummy function to intercept the real shell_cmd() */ /* ARGSUSED */ void shell_cmd(command) char *command; { if (hosts_access_verbose) printf("command: %s", command); } /* dummy function to intercept the real clean_exit() */ /* ARGSUSED */ void clean_exit(request) struct request_info *request; { exit(0); } -/* dummy function to intercept the real rfc931() */ - -/* ARGSUSED */ - -void rfc931(request) -struct request_info *request; -{ - strcpy(request->user, unknown); -} - /* check_path - examine accessibility */ int check_path(path, st) char *path; struct stat *st; { struct stat stbuf; char buf[BUFSIZ]; if (stat(path, st) < 0) return (-1); #ifdef notdef if (st->st_uid != 0) tcpd_warn("%s: not owned by root", path); if (st->st_mode & 020) tcpd_warn("%s: group writable", path); #endif if (st->st_mode & 002) tcpd_warn("%s: world writable", path); if (path[0] == '/' && path[1] != 0) { strrchr(strcpy(buf, path), '/')[0] = 0; (void) check_path(buf[0] ? buf : "/", &stbuf); } return (0); } Index: projects/clang400-import/contrib/tcp_wrappers/tcpd.h =================================================================== --- projects/clang400-import/contrib/tcp_wrappers/tcpd.h (revision 311459) +++ projects/clang400-import/contrib/tcp_wrappers/tcpd.h (revision 311460) @@ -1,226 +1,219 @@ /* * @(#) tcpd.h 1.5 96/03/19 16:22:24 * * Author: Wietse Venema, Eindhoven University of Technology, The Netherlands. * * $FreeBSD$ */ +#ifdef INET6 +#define TCPD_SOCKADDR struct sockaddr +#else +#define TCPD_SOCKADDR struct sockaddr_in +#endif + /* Structure to describe one communications endpoint. */ #define STRING_LENGTH 128 /* hosts, users, processes */ struct host_info { char name[STRING_LENGTH]; /* access via eval_hostname(host) */ char addr[STRING_LENGTH]; /* access via eval_hostaddr(host) */ -#ifdef INET6 - struct sockaddr *sin; /* socket address or 0 */ -#else - struct sockaddr_in *sin; /* socket address or 0 */ -#endif + TCPD_SOCKADDR *sin; /* socket address or 0 */ struct t_unitdata *unit; /* TLI transport address or 0 */ struct request_info *request; /* for shared information */ }; /* Structure to describe what we know about a service request. */ struct request_info { int fd; /* socket handle */ char user[STRING_LENGTH]; /* access via eval_user(request) */ char daemon[STRING_LENGTH]; /* access via eval_daemon(request) */ char pid[10]; /* access via eval_pid(request) */ struct host_info client[1]; /* client endpoint info */ struct host_info server[1]; /* server endpoint info */ void (*sink) (int); /* datagram sink function or 0 */ void (*hostname) (struct host_info *); /* address to printable hostname */ void (*hostaddr) (struct host_info *); /* address to printable address */ void (*cleanup) (struct request_info *); /* cleanup function or 0 */ struct netconfig *config; /* netdir handle */ }; /* Common string operations. Less clutter should be more readable. */ #define STRN_CPY(d,s,l) { strncpy((d),(s),(l)); (d)[(l)-1] = 0; } #define STRN_EQ(x,y,l) (strncasecmp((x),(y),(l)) == 0) #define STRN_NE(x,y,l) (strncasecmp((x),(y),(l)) != 0) #define STR_EQ(x,y) (strcasecmp((x),(y)) == 0) #define STR_NE(x,y) (strcasecmp((x),(y)) != 0) /* * Initially, all above strings have the empty value. Information that * cannot be determined at runtime is set to "unknown", so that we can * distinguish between `unavailable' and `not yet looked up'. A hostname * that we do not believe in is set to "paranoid". */ #define STRING_UNKNOWN "unknown" /* lookup failed */ #define STRING_PARANOID "paranoid" /* hostname conflict */ extern char unknown[]; extern char paranoid[]; #define HOSTNAME_KNOWN(s) (STR_NE((s),unknown) && STR_NE((s),paranoid)) #define NOT_INADDR(s) (s[strspn(s,"01234567890./")] != 0) /* Global functions. */ #if defined(TLI) || defined(PTX) || defined(TLI_SEQUENT) -extern void fromhost(); /* get/validate client host info */ +extern void fromhost(struct request_info *); /* get/validate client host info */ #else -#define fromhost sock_host /* no TLI support needed */ +#define fromhost sock_host /* no TLI support needed */ #endif -extern int hosts_access(); /* access control */ -extern int hosts_ctl(); /* wrapper around request_init() */ -extern void shell_cmd(); /* execute shell command */ -extern char *percent_x(); /* do % expansion */ -extern void rfc931(); /* client name from RFC 931 daemon */ -extern void clean_exit(); /* clean up and exit */ -extern void refuse(); /* clean up and exit */ -extern char *xgets(); /* fgets() on steroids */ -extern char *split_at(); /* strchr() and split */ -extern unsigned long dot_quad_addr(); /* restricted inet_addr() */ +extern int hosts_access(struct request_info *); /* access control */ +extern int hosts_ctl(char *, char *, char *, char *); /* wrapper around request_init() */ +extern void shell_cmd(char *); /* execute shell command */ +extern char *percent_x(char *, int, char *, struct request_info *); /* do % expansion */ +extern void rfc931(TCPD_SOCKADDR *, TCPD_SOCKADDR *, char *); /* client name from RFC 931 daemon */ +extern void clean_exit(struct request_info *); /* clean up and exit */ +extern void refuse(struct request_info *); /* clean up and exit */ +extern char *xgets(char *, int, FILE *); /* fgets() on steroids */ +extern char *split_at(char *, int); /* strchr() and split */ +extern unsigned long dot_quad_addr(char *); /* restricted inet_addr() */ + /* Global variables. */ extern int allow_severity; /* for connection logging */ extern int deny_severity; /* for connection logging */ extern char *hosts_allow_table; /* for verification mode redirection */ extern char *hosts_deny_table; /* for verification mode redirection */ extern int hosts_access_verbose; /* for verbose matching mode */ extern int rfc931_timeout; /* user lookup timeout */ extern int resident; /* > 0 if resident process */ /* * Routines for controlled initialization and update of request structure * attributes. Each attribute has its own key. */ -#ifdef __STDC__ -extern struct request_info *request_init(struct request_info *,...); -extern struct request_info *request_set(struct request_info *,...); -#else -extern struct request_info *request_init(); /* initialize request */ -extern struct request_info *request_set(); /* update request structure */ -#endif +extern struct request_info *request_init(struct request_info *,...); /* initialize request */ +extern struct request_info *request_set(struct request_info *,...); /* update request structure */ #define RQ_FILE 1 /* file descriptor */ #define RQ_DAEMON 2 /* server process (argv[0]) */ #define RQ_USER 3 /* client user name */ #define RQ_CLIENT_NAME 4 /* client host name */ #define RQ_CLIENT_ADDR 5 /* client host address */ #define RQ_CLIENT_SIN 6 /* client endpoint (internal) */ #define RQ_SERVER_NAME 7 /* server host name */ #define RQ_SERVER_ADDR 8 /* server host address */ #define RQ_SERVER_SIN 9 /* server endpoint (internal) */ /* * Routines for delayed evaluation of request attributes. Each attribute * type has its own access method. The trivial ones are implemented by * macros. The other ones are wrappers around the transport-specific host * name, address, and client user lookup methods. The request_info and * host_info structures serve as caches for the lookup results. */ -extern char *eval_user(); /* client user */ -extern char *eval_hostname(); /* printable hostname */ -extern char *eval_hostaddr(); /* printable host address */ -extern char *eval_hostinfo(); /* host name or address */ -extern char *eval_client(); /* whatever is available */ -extern char *eval_server(); /* whatever is available */ +extern char *eval_user(struct request_info *); /* client user */ +extern char *eval_hostname(struct host_info *); /* printable hostname */ +extern char *eval_hostaddr(struct host_info *); /* printable host address */ +extern char *eval_hostinfo(struct host_info *); /* host name or address */ +extern char *eval_client(struct request_info *); /* whatever is available */ +extern char *eval_server(struct request_info *); /* whatever is available */ #define eval_daemon(r) ((r)->daemon) /* daemon process name */ #define eval_pid(r) ((r)->pid) /* process id */ /* Socket-specific methods, including DNS hostname lookups. */ -extern void sock_host(); /* look up endpoint addresses */ -extern void sock_hostname(); /* translate address to hostname */ -extern void sock_hostaddr(); /* address to printable address */ +extern void sock_host(struct request_info *); /* look up endpoint addresses */ +extern void sock_hostname(struct host_info *); /* translate address to hostname */ +extern void sock_hostaddr(struct host_info *); /* address to printable address */ #define sock_methods(r) \ { (r)->hostname = sock_hostname; (r)->hostaddr = sock_hostaddr; } /* The System V Transport-Level Interface (TLI) interface. */ #if defined(TLI) || defined(PTX) || defined(TLI_SEQUENT) -extern void tli_host(); /* look up endpoint addresses etc. */ +extern void tli_host(struct request_info *); /* look up endpoint addresses etc. */ #endif /* * Problem reporting interface. Additional file/line context is reported * when available. The jump buffer (tcpd_buf) is not declared here, or * everyone would have to include . */ -#ifdef __STDC__ extern void tcpd_warn(char *, ...); /* report problem and proceed */ extern void tcpd_jump(char *, ...); /* report problem and jump */ -#else -extern void tcpd_warn(); -extern void tcpd_jump(); -#endif struct tcpd_context { char *file; /* current file */ int line; /* current line */ }; extern struct tcpd_context tcpd_context; /* * While processing access control rules, error conditions are handled by * jumping back into the hosts_access() routine. This is cleaner than * checking the return value of each and every silly little function. The * (-1) returns are here because zero is already taken by longjmp(). */ #define AC_PERMIT 1 /* permit access */ #define AC_DENY (-1) /* deny_access */ #define AC_ERROR AC_DENY /* XXX */ /* * In verification mode an option function should just say what it would do, * instead of really doing it. An option function that would not return * should clear the dry_run flag to inform the caller of this unusual * behavior. */ -extern void process_options(); /* execute options */ -extern int dry_run; /* verification flag */ +extern void process_options(char *, struct request_info *); /* execute options */ +extern int dry_run; /* verification flag */ /* Bug workarounds. */ #ifdef INET_ADDR_BUG /* inet_addr() returns struct */ #define inet_addr fix_inet_addr -extern long fix_inet_addr(); +extern long fix_inet_addr(char *); #endif #ifdef BROKEN_FGETS /* partial reads from sockets */ #define fgets fix_fgets -extern char *fix_fgets(); +extern char *fix_fgets(char *, int, FILE *); #endif #ifdef RECVFROM_BUG /* no address family info */ #define recvfrom fix_recvfrom -extern int fix_recvfrom(); +extern int fix_recvfrom(int, char *, int, int, struct sockaddr *, int *); #endif #ifdef GETPEERNAME_BUG /* claims success with UDP */ #define getpeername fix_getpeername -extern int fix_getpeername(); +extern int fix_getpeername(int, struct sockaddr *, int *); #endif #ifdef SOLARIS_24_GETHOSTBYNAME_BUG /* lists addresses as aliases */ #define gethostbyname fix_gethostbyname -extern struct hostent *fix_gethostbyname(); +extern struct hostent *fix_gethostbyname(char *); #endif #ifdef USE_STRSEP /* libc calls strtok() */ #define strtok fix_strtok -extern char *fix_strtok(); +extern char *fix_strtok(char *, char *); #endif #ifdef LIBC_CALLS_STRTOK /* libc calls strtok() */ #define strtok my_strtok -extern char *my_strtok(); +extern char *my_strtok(char *, char *); #endif Index: projects/clang400-import/etc/mtree/BSD.tests.dist =================================================================== --- projects/clang400-import/etc/mtree/BSD.tests.dist (revision 311459) +++ projects/clang400-import/etc/mtree/BSD.tests.dist (revision 311460) @@ -1,696 +1,696 @@ # $FreeBSD$ # # Please see the file src/etc/mtree/README before making changes to this file. # /set type=dir uname=root gname=wheel mode=0755 . bin cat .. date .. dd .. expr .. ls .. mv .. pax .. pkill .. sh builtins .. errors .. execution .. expansion .. parameters .. parser .. set-e .. .. sleep .. test .. .. cddl lib .. sbin .. usr.bin .. usr.sbin dtrace common aggs .. arithmetic .. arrays .. assocs .. begin .. bitfields .. buffering .. builtinvar .. cg .. clauses .. cpc .. decls .. drops .. dtraceUtil .. end .. enum .. error .. exit .. fbtprovider .. funcs .. grammar .. include .. inline .. io .. ip .. java_api .. json .. lexer .. llquantize .. mdb .. mib .. misc .. multiaggs .. offsetof .. operators .. pid .. plockstat .. pointers .. pragma .. predicates .. preprocessor .. print .. printa .. printf .. privs .. probes .. proc .. profile-n .. providers .. raise .. rates .. safety .. scalars .. sched .. scripting .. sdt .. sizeof .. speculation .. stability .. stack .. stackdepth .. stop .. strlen .. strtoll .. struct .. sugar .. syscall .. sysevent .. tick-n .. trace .. tracemem .. translators .. typedef .. types .. uctf .. union .. usdt .. ustack .. vars .. version .. .. .. zfsd .. .. .. etc rc.d .. .. games .. gnu lib .. usr.bin diff .. .. .. lib atf libatf-c detail .. .. libatf-c++ detail .. .. test-programs .. .. libarchive .. libc c063 .. db .. gen execve .. posix_spawn .. .. hash data .. .. iconv .. inet .. locale .. net getaddrinfo data .. .. .. nss .. regex data .. .. resolv .. rpc .. ssp .. setjmp .. stdio .. stdlib .. string .. sys .. time .. tls dso .. .. termios .. ttyio .. .. libcasper services cap_dns .. cap_grp .. cap_pwd .. cap_sysctl .. .. .. libcrypt .. libdevdctl .. libmp .. libnv .. libpam .. libproc .. librt .. libthr dlopen .. .. libutil .. libxo .. msun .. .. libexec atf atf-check .. atf-sh .. .. rtld-elf .. .. sbin dhclient .. devd .. growfs .. ifconfig .. mdconfig .. .. secure lib .. libexec .. usr.bin .. usr.sbin .. .. share examples tests atf .. plain .. .. .. .. sys acl .. aio .. fifo .. file .. fs tmpfs .. .. geom class concat .. eli .. gate .. gpt .. mirror .. nop .. raid3 .. shsec .. stripe .. uzip etalon .. .. .. .. kern acct .. execve .. pipe .. .. kqueue libkqueue .. .. mac bsdextended .. portacl .. .. mqueue .. netinet .. opencrypto .. pjdfstest chflags .. chmod .. chown .. ftruncate .. granular .. link .. mkdir .. mkfifo .. mknod .. open .. rename .. rmdir .. symlink .. truncate .. unlink .. .. posixshm .. sys .. vfs .. vm .. .. usr.bin apply .. basename .. bmake archives fmt_44bsd .. fmt_44bsd_mod .. fmt_oldbsd .. .. basic t0 .. t1 .. t2 .. t3 .. .. execution ellipsis .. empty .. joberr .. plus .. .. shell builtin .. meta .. path .. path_select .. replace .. select .. .. suffixes basic .. src_wild1 .. src_wild2 .. .. syntax directive-t0 .. enl .. funny-targets .. semi .. .. sysmk t0 2 1 .. .. mk .. .. t1 2 1 .. .. mk .. .. t2 2 1 .. .. mk .. .. .. variables modifier_M .. modifier_t .. opt_V .. t0 .. .. .. - bsdcat - .. + bsdcat + .. calendar .. cmp .. cpio .. col .. comm .. cut .. dirname .. file2c .. grep .. gzip .. ident .. join .. jot .. lastcomm .. limits .. m4 .. mkimg .. ncal .. opensm .. printf .. sdiff .. sed regress.multitest.out .. .. soelim .. tar .. timeout .. tr .. truncate .. units .. uudecode .. uuencode .. xargs .. xinstall .. xo .. yacc yacc .. .. .. usr.sbin chown .. etcupdate .. extattr .. fstyp .. makefs .. newsyslog .. nmtree .. pw .. rpcbind .. sa .. .. .. # vim: set expandtab ts=4 sw=4: Index: projects/clang400-import/etc/mtree/BSD.var.dist =================================================================== --- projects/clang400-import/etc/mtree/BSD.var.dist (revision 311459) +++ projects/clang400-import/etc/mtree/BSD.var.dist (revision 311460) @@ -1,112 +1,112 @@ # $FreeBSD$ # # Please see the file src/etc/mtree/README before making changes to this file. # /set type=dir uname=root gname=wheel mode=0755 . account .. at /set uname=daemon jobs tags=package=at .. spool tags=package=at .. /set uname=root .. /set mode=0750 /set gname=audit audit dist uname=auditdistd gname=audit mode=0770 .. remote uname=auditdistd gname=wheel mode=0700 .. .. authpf uname=root gname=authpf mode=0770 .. /set gname=wheel backups .. - cache mode=0755 + cache mode=0755 .. crash .. cron tabs mode=0700 .. .. /set mode=0755 db entropy uname=operator gname=operator mode=0700 .. freebsd-update mode=0700 .. hyperv mode=0700 .. ipf mode=0700 .. ntp mode=0700 .. pkg .. ports .. portsnap .. zfsd cases .. .. .. empty mode=0555 flags=schg .. games gname=games mode=0775 .. heimdal mode=0700 .. log .. mail gname=mail mode=0775 .. msgs uname=daemon .. preserve .. run ppp gname=network mode=0770 .. wpa_supplicant .. .. rwho gname=daemon mode=0775 .. spool dma uname=root gname=mail mode=0770 .. lock uname=uucp gname=dialer mode=0775 .. /set gname=daemon lpd .. mqueue .. opielocks mode=0700 .. output lpd .. .. /set gname=wheel .. tmp mode=01777 vi.recover mode=01777 .. .. unbound uname=unbound gname=unbound mode=0755 tags=package=unbound conf.d uname=unbound gname=unbound mode=0755 tags=package=unbound .. .. yp .. .. Index: projects/clang400-import/lib/libproc/proc_sym.c =================================================================== --- projects/clang400-import/lib/libproc/proc_sym.c (revision 311459) +++ projects/clang400-import/lib/libproc/proc_sym.c (revision 311460) @@ -1,658 +1,660 @@ /*- * Copyright (c) 2016 Mark Johnston * Copyright (c) 2010 The FreeBSD Foundation * Copyright (c) 2008 John Birrell (jb@freebsd.org) * All rights reserved. * * Portions of this software were developed by Rui Paulo under sponsorship * from the FreeBSD Foundation. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #include #ifndef NO_CTF #include #include #endif #include #include #include #include #include #include #include #include #include #ifndef NO_CTF #include #endif #include #include "crc32.h" #include "_libproc.h" #define PATH_DEBUG_DIR "/usr/lib/debug" #ifdef NO_CTF typedef struct ctf_file ctf_file_t; #endif #ifndef NO_CXA_DEMANGLE extern char *__cxa_demangle(const char *, char *, size_t *, int *); #endif /* NO_CXA_DEMANGLE */ static int crc32_file(int fd, uint32_t *crc) { uint8_t buf[PAGE_SIZE], *p; size_t n; *crc = ~0; while ((n = read(fd, buf, sizeof(buf))) > 0) { p = &buf[0]; while (n-- > 0) *crc = crc32_tab[(*crc ^ *p++) & 0xff] ^ (*crc >> 8); } *crc = ~*crc; return (n); } static void demangle(const char *symbol, char *buf, size_t len) { #ifndef NO_CXA_DEMANGLE char *dembuf; if (symbol[0] == '_' && symbol[1] == 'Z' && symbol[2]) { dembuf = __cxa_demangle(symbol, NULL, NULL, NULL); if (!dembuf) goto fail; strlcpy(buf, dembuf, len); free(dembuf); return; } fail: #endif /* NO_CXA_DEMANGLE */ strlcpy(buf, symbol, len); } static int symvalcomp(void *thunk, const void *a1, const void *a2) { struct symtab *symtab; GElf_Sym sym1, sym2; u_int i1, i2; int ret; i1 = *(const u_int *)a1; i2 = *(const u_int *)a2; symtab = thunk; (void)gelf_getsym(symtab->data, i1, &sym1); (void)gelf_getsym(symtab->data, i2, &sym2); if (sym1.st_value < sym2.st_value) ret = -1; else if (sym1.st_value == sym2.st_value) ret = 0; else ret = 1; return (ret); } static int load_symtab(Elf *e, struct symtab *symtab, u_long sh_type) { GElf_Ehdr ehdr; GElf_Shdr shdr; Elf_Scn *scn; u_int nsyms; if (gelf_getehdr(e, &ehdr) == NULL) return (-1); scn = NULL; while ((scn = elf_nextscn(e, scn)) != NULL) { (void)gelf_getshdr(scn, &shdr); if (shdr.sh_type == sh_type) break; } if (scn == NULL) return (-1); - if ((symtab->data = elf_getdata(scn, NULL)) == NULL) + nsyms = shdr.sh_size / shdr.sh_entsize; + if (nsyms > (1 << 20)) return (-1); - nsyms = shdr.sh_size / shdr.sh_entsize; + if ((symtab->data = elf_getdata(scn, NULL)) == NULL) + return (-1); symtab->index = calloc(nsyms, sizeof(u_int)); if (symtab->index == NULL) return (-1); for (u_int i = 0; i < nsyms; i++) symtab->index[i] = i; qsort_r(symtab->index, nsyms, sizeof(u_int), symtab, symvalcomp); symtab->nsyms = nsyms; symtab->stridx = shdr.sh_link; return (0); } static void load_symtabs(struct file_info *file) { file->symtab.nsyms = file->dynsymtab.nsyms = 0; (void)load_symtab(file->elf, &file->symtab, SHT_SYMTAB); (void)load_symtab(file->elf, &file->dynsymtab, SHT_DYNSYM); } static int open_debug_file(char *path, const char *debugfile, uint32_t crc) { size_t n; uint32_t compcrc; int fd; fd = -1; if ((n = strlcat(path, "/", PATH_MAX)) >= PATH_MAX) return (fd); if (strlcat(path, debugfile, PATH_MAX) >= PATH_MAX) goto out; if ((fd = open(path, O_RDONLY | O_CLOEXEC)) < 0) goto out; if (crc32_file(fd, &compcrc) != 0 || crc != compcrc) { DPRINTFX("ERROR: CRC32 mismatch for %s", path); (void)close(fd); fd = -1; } out: path[n] = '\0'; return (fd); } /* * Obtain an ELF descriptor for the specified mapped object. If a GNU debuglink * section is present, a descriptor for the corresponding debug file is * returned. */ static int open_object(struct map_info *mapping) { char path[PATH_MAX]; GElf_Shdr shdr; Elf *e, *e2; Elf_Data *data; Elf_Scn *scn; struct file_info *file; prmap_t *map; const char *debugfile, *scnname; size_t ndx; uint32_t crc; int fd, fd2; if (mapping->map.pr_mapname[0] == '\0') return (-1); /* anonymous object */ if (mapping->file->elf != NULL) return (0); /* already loaded */ file = mapping->file; map = &mapping->map; if ((fd = open(map->pr_mapname, O_RDONLY | O_CLOEXEC)) < 0) { DPRINTF("ERROR: open %s failed", map->pr_mapname); return (-1); } if ((e = elf_begin(fd, ELF_C_READ, NULL)) == NULL) { DPRINTFX("ERROR: elf_begin() failed: %s", elf_errmsg(-1)); goto err; } if (gelf_getehdr(e, &file->ehdr) != &file->ehdr) { DPRINTFX("ERROR: elf_getehdr() failed: %s", elf_errmsg(-1)); goto err; } scn = NULL; while ((scn = elf_nextscn(e, scn)) != NULL) { if (gelf_getshdr(scn, &shdr) != &shdr) { DPRINTFX("ERROR: gelf_getshdr failed: %s", elf_errmsg(-1)); goto err; } if (shdr.sh_type != SHT_PROGBITS) continue; if (elf_getshdrstrndx(e, &ndx) != 0) { DPRINTFX("ERROR: elf_getshdrstrndx failed: %s", elf_errmsg(-1)); goto err; } if ((scnname = elf_strptr(e, ndx, shdr.sh_name)) == NULL) continue; if (strcmp(scnname, ".gnu_debuglink") == 0) break; } if (scn == NULL) goto internal; if ((data = elf_getdata(scn, NULL)) == NULL) { DPRINTFX("ERROR: elf_getdata failed: %s", elf_errmsg(-1)); goto err; } /* * The data contains a null-terminated file name followed by a 4-byte * CRC. */ if (data->d_size < sizeof(crc) + 1) { DPRINTFX("ERROR: debuglink section is too small (%zd bytes)", data->d_size); goto internal; } if (strnlen(data->d_buf, data->d_size) >= data->d_size - sizeof(crc)) { DPRINTFX("ERROR: no null-terminator in gnu_debuglink section"); goto internal; } debugfile = data->d_buf; memcpy(&crc, (char *)data->d_buf + data->d_size - sizeof(crc), sizeof(crc)); /* * Search for the debug file using the algorithm described in the gdb * documentation: * - look in the directory containing the object, * - look in the subdirectory ".debug" of the directory containing the * object, * - look in the global debug directories (currently /usr/lib/debug). */ (void)strlcpy(path, map->pr_mapname, sizeof(path)); (void)dirname(path); if ((fd2 = open_debug_file(path, debugfile, crc)) >= 0) goto external; if (strlcat(path, "/.debug", sizeof(path)) < sizeof(path) && (fd2 = open_debug_file(path, debugfile, crc)) >= 0) goto external; (void)snprintf(path, sizeof(path), PATH_DEBUG_DIR); if (strlcat(path, map->pr_mapname, sizeof(path)) < sizeof(path)) { (void)dirname(path); if ((fd2 = open_debug_file(path, debugfile, crc)) >= 0) goto external; } internal: /* We didn't find a debug file, just return the object's descriptor. */ file->elf = e; file->fd = fd; load_symtabs(file); return (0); external: if ((e2 = elf_begin(fd2, ELF_C_READ, NULL)) == NULL) { DPRINTFX("ERROR: elf_begin failed: %s", elf_errmsg(-1)); (void)close(fd2); goto err; } (void)elf_end(e); (void)close(fd); file->elf = e2; file->fd = fd2; load_symtabs(file); return (0); err: if (e != NULL) (void)elf_end(e); (void)close(fd); return (-1); } char * proc_objname(struct proc_handle *p, uintptr_t addr, char *objname, size_t objnamesz) { prmap_t *map; size_t i; for (i = 0; i < p->nmappings; i++) { map = &p->mappings[i].map; if (addr >= map->pr_vaddr && addr < map->pr_vaddr + map->pr_size) { strlcpy(objname, map->pr_mapname, objnamesz); return (objname); } } return (NULL); } int proc_iter_objs(struct proc_handle *p, proc_map_f *func, void *cd) { char last[MAXPATHLEN], path[MAXPATHLEN], *base; prmap_t *map; size_t i; int error; if (p->nmappings == 0) if (proc_rdagent(p) == NULL) return (-1); error = 0; memset(last, 0, sizeof(last)); for (i = 0; i < p->nmappings; i++) { map = &p->mappings[i].map; strlcpy(path, map->pr_mapname, sizeof(path)); base = basename(path); /* * We shouldn't call the callback twice with the same object. * To do that we are assuming the fact that if there are * repeated object names (i.e. different mappings for the * same object) they occur next to each other. */ if (strcmp(base, last) == 0) continue; if ((error = (*func)(cd, map, base)) != 0) break; strlcpy(last, path, sizeof(last)); } return (error); } static struct map_info * _proc_addr2map(struct proc_handle *p, uintptr_t addr) { struct map_info *mapping; size_t i; if (p->nmappings == 0) if (proc_rdagent(p) == NULL) return (NULL); for (i = 0; i < p->nmappings; i++) { mapping = &p->mappings[i]; if (addr >= mapping->map.pr_vaddr && addr < mapping->map.pr_vaddr + mapping->map.pr_size) return (mapping); } return (NULL); } prmap_t * proc_addr2map(struct proc_handle *p, uintptr_t addr) { return (&_proc_addr2map(p, addr)->map); } /* * Look up the symbol at addr using a binary search, returning a copy of the * symbol and its name. */ static int lookup_symbol_by_addr(Elf *elf, struct symtab *symtab, uintptr_t addr, const char **namep, GElf_Sym *sym) { Elf_Data *data; const char *s; int min, max, mid; data = symtab->data; min = 0; max = symtab->nsyms - 1; while (min <= max) { mid = (max + min) / 2; (void)gelf_getsym(data, symtab->index[mid], sym); if (addr >= sym->st_value && addr < sym->st_value + sym->st_size) { s = elf_strptr(elf, symtab->stridx, sym->st_name); if (s != NULL && namep != NULL) *namep = s; return (0); } if (addr < sym->st_value) max = mid - 1; else min = mid + 1; } return (ENOENT); } int proc_addr2sym(struct proc_handle *p, uintptr_t addr, char *name, size_t namesz, GElf_Sym *symcopy) { struct file_info *file; struct map_info *mapping; const char *s; uintptr_t off; int error; if ((mapping = _proc_addr2map(p, addr)) == NULL) { DPRINTFX("ERROR: proc_addr2map failed to resolve 0x%jx", addr); return (-1); } if (open_object(mapping) != 0) { DPRINTFX("ERROR: failed to open object %s", mapping->map.pr_mapname); return (-1); } file = mapping->file; off = file->ehdr.e_type == ET_DYN ? mapping->map.pr_vaddr : 0; if (addr < off) return (ENOENT); addr -= off; error = lookup_symbol_by_addr(file->elf, &file->dynsymtab, addr, &s, symcopy); if (error == ENOENT) error = lookup_symbol_by_addr(file->elf, &file->symtab, addr, &s, symcopy); if (error == 0) { symcopy->st_value += off; demangle(s, name, namesz); } return (error); } static struct map_info * _proc_name2map(struct proc_handle *p, const char *name) { char path[MAXPATHLEN], *base; struct map_info *mapping; size_t i, len; if ((len = strlen(name)) == 0) return (NULL); if (p->nmappings == 0) if (proc_rdagent(p) == NULL) return (NULL); for (i = 0; i < p->nmappings; i++) { mapping = &p->mappings[i]; (void)strlcpy(path, mapping->map.pr_mapname, sizeof(path)); base = basename(path); if (strcmp(base, name) == 0) return (mapping); } /* If we didn't find a match, try matching prefixes of the basename. */ for (i = 0; i < p->nmappings; i++) { strlcpy(path, p->mappings[i].map.pr_mapname, sizeof(path)); base = basename(path); if (strncmp(base, name, len) == 0) return (&p->mappings[i]); } if (strcmp(name, "a.out") == 0) return (_proc_addr2map(p, p->exec_map->pr_vaddr)); return (NULL); } prmap_t * proc_name2map(struct proc_handle *p, const char *name) { return (&_proc_name2map(p, name)->map); } /* * Look up the symbol with the given name and return a copy of it. */ static int lookup_symbol_by_name(Elf *elf, struct symtab *symtab, const char *symbol, GElf_Sym *symcopy, prsyminfo_t *si) { GElf_Sym sym; Elf_Data *data; char *s; int i; if (symtab->nsyms == 0) return (ENOENT); data = symtab->data; for (i = 0; gelf_getsym(data, i, &sym) != NULL; i++) { s = elf_strptr(elf, symtab->stridx, sym.st_name); if (s != NULL && strcmp(s, symbol) == 0) { memcpy(symcopy, &sym, sizeof(*symcopy)); if (si != NULL) si->prs_id = i; return (0); } } return (ENOENT); } int proc_name2sym(struct proc_handle *p, const char *object, const char *symbol, GElf_Sym *symcopy, prsyminfo_t *si) { struct file_info *file; struct map_info *mapping; uintptr_t off; int error; if ((mapping = _proc_name2map(p, object)) == NULL) { DPRINTFX("ERROR: proc_name2map failed to resolve %s", object); return (-1); } if (open_object(mapping) != 0) { DPRINTFX("ERROR: failed to open object %s", mapping->map.pr_mapname); return (-1); } file = mapping->file; off = file->ehdr.e_type == ET_DYN ? mapping->map.pr_vaddr : 0; error = lookup_symbol_by_name(file->elf, &file->dynsymtab, symbol, symcopy, si); if (error == ENOENT) error = lookup_symbol_by_name(file->elf, &file->symtab, symbol, symcopy, si); if (error == 0) symcopy->st_value += off; return (error); } ctf_file_t * proc_name2ctf(struct proc_handle *p, const char *name) { #ifndef NO_CTF ctf_file_t *ctf; prmap_t *map; int error; if ((map = proc_name2map(p, name)) == NULL) return (NULL); ctf = ctf_open(map->pr_mapname, &error); return (ctf); #else (void)p; (void)name; return (NULL); #endif } int proc_iter_symbyaddr(struct proc_handle *p, const char *object, int which, int mask, proc_sym_f *func, void *cd) { GElf_Sym sym; struct file_info *file; struct map_info *mapping; struct symtab *symtab; const char *s; int error, i; if ((mapping = _proc_name2map(p, object)) == NULL) { DPRINTFX("ERROR: proc_name2map failed to resolve %s", object); return (-1); } if (open_object(mapping) != 0) { DPRINTFX("ERROR: failed to open object %s", mapping->map.pr_mapname); return (-1); } file = mapping->file; symtab = which == PR_SYMTAB ? &file->symtab : &file->dynsymtab; if (symtab->nsyms == 0) return (-1); error = 0; for (i = 0; gelf_getsym(symtab->data, i, &sym) != NULL; i++) { if (GELF_ST_BIND(sym.st_info) == STB_LOCAL && (mask & BIND_LOCAL) == 0) continue; if (GELF_ST_BIND(sym.st_info) == STB_GLOBAL && (mask & BIND_GLOBAL) == 0) continue; if (GELF_ST_BIND(sym.st_info) == STB_WEAK && (mask & BIND_WEAK) == 0) continue; if (GELF_ST_TYPE(sym.st_info) == STT_NOTYPE && (mask & TYPE_NOTYPE) == 0) continue; if (GELF_ST_TYPE(sym.st_info) == STT_OBJECT && (mask & TYPE_OBJECT) == 0) continue; if (GELF_ST_TYPE(sym.st_info) == STT_FUNC && (mask & TYPE_FUNC) == 0) continue; if (GELF_ST_TYPE(sym.st_info) == STT_SECTION && (mask & TYPE_SECTION) == 0) continue; if (GELF_ST_TYPE(sym.st_info) == STT_FILE && (mask & TYPE_FILE) == 0) continue; s = elf_strptr(file->elf, symtab->stridx, sym.st_name); if (file->ehdr.e_type == ET_DYN) sym.st_value += mapping->map.pr_vaddr; if ((error = (*func)(cd, &sym, s)) != 0) break; } return (error); } Index: projects/clang400-import/sys/arm/conf/ARMADA38X =================================================================== --- projects/clang400-import/sys/arm/conf/ARMADA38X (revision 311459) +++ projects/clang400-import/sys/arm/conf/ARMADA38X (revision 311460) @@ -1,73 +1,76 @@ # # Kernel configuration for Marvell Armada38x # # $FreeBSD$ # include "../mv/armada38x/std.armada38x" include "std.armv6" ident ARMADA38X options SOC_MV_ARMADA38X makeoptions WERROR="-Werror" options MD_ROOT #makeoptions MFS_IMAGE=/path/to/miniroot #options ROOTDEVNAME=\"ufs:md0\" options ROOTDEVNAME=\"/dev/da0s1a\" options SCHED_ULE # ULE scheduler options SMP # Pseudo devices device random device pty device loop device md # Serial ports device uart device uart_snps # Network device ether device vlan device mii device bpf device re +device mdio +device etherswitch +device e6000sw # PCI device pci # Interrupt controllers device gic options INTRNG # Timers device mpcore_timer # USB device usb device ehci device xhci device umass device scbus device pass device da # I2C device iic device iicbus device twsi # CESA device cesa device crypto device cryptodev #FDT options FDT options FDT_DTB_STATIC makeoptions FDT_DTS_FILE=armada-388-gp.dts Index: projects/clang400-import/sys/boot/fdt/dts/arm/armada-385-db-ap.dts =================================================================== --- projects/clang400-import/sys/boot/fdt/dts/arm/armada-385-db-ap.dts (nonexistent) +++ projects/clang400-import/sys/boot/fdt/dts/arm/armada-385-db-ap.dts (revision 311460) @@ -0,0 +1,271 @@ +/* + * Device Tree file for Marvell Armada 385 Access Point Development board + * (DB-88F6820-AP) + * + * Copyright (C) 2014 Marvell + * + * Nadav Haklai + * + * This file is dual-licensed: you can use it either under the terms + * of the GPL or the X11 license, at your option. Note that this dual + * licensing only applies to this file, and not this project as a + * whole. + * + * a) This file is licensed under the terms of the GNU General Public + * License version 2. This program is licensed "as is" without + * any warranty of any kind, whether express or implied. + * + * Or, alternatively, + * + * b) Permission is hereby granted, free of charge, to any person + * obtaining a copy of this software and associated documentation + * files (the "Software"), to deal in the Software without + * restriction, including without limitation the rights to use, + * copy, modify, merge, publish, distribute, sublicense, and/or + * sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following + * conditions: + * + * The above copyright notice and this permission notice shall be + * included in all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES + * OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT + * HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * $FreeBSD$ + */ + +/dts-v1/; +#include "armada-385.dtsi" + +#include + +/ { + model = "Marvell Armada 385 Access Point Development Board"; + compatible = "marvell,a385-db-ap", "marvell,armada385", "marvell,armada380"; + + chosen { + stdout-path = "serial1"; + }; + + memory { + device_type = "memory"; + reg = <0x00000000 0x80000000>; /* 2GB */ + }; + + soc { + ranges = ; + + internal-regs { + i2c0: i2c@11000 { + pinctrl-names = "default"; + pinctrl-0 = <&i2c0_pins>; + status = "okay"; + + /* + * This bus is wired to two EEPROM + * sockets, one of which holding the + * board ID used by the bootloader. + * Erasing this EEPROM's content will + * brick the board. + * Use this bus with caution. + */ + }; + + mdio@72004 { + pinctrl-names = "default"; + pinctrl-0 = <&mdio_pins>; + + phy0: ethernet-phy@1 { + reg = <1>; + }; + + phy1: ethernet-phy@4 { + reg = <4>; + }; + + phy2: ethernet-phy@6 { + reg = <6>; + }; + }; + + /* UART0 is exposed through the JP8 connector */ + uart0: serial@12000 { + pinctrl-names = "default"; + pinctrl-0 = <&uart0_pins>; + status = "okay"; + }; + + /* + * UART1 is exposed through a FTDI chip + * wired to the mini-USB connector + */ + uart1: serial@12100 { + pinctrl-names = "default"; + pinctrl-0 = <&uart1_pins>; + status = "okay"; + }; + + pinctrl@18000 { + xhci0_vbus_pins: xhci0-vbus-pins { + marvell,pins = "mpp44"; + marvell,function = "gpio"; + }; + }; + + /* CON3 */ + ethernet@30000 { + status = "okay"; + phy = <&phy2>; + phy-mode = "sgmii"; + buffer-manager = <&bm>; + bm,pool-long = <1>; + bm,pool-short = <3>; + }; + + /* CON2 */ + ethernet@34000 { + status = "okay"; + phy = <&phy1>; + phy-mode = "sgmii"; + buffer-manager = <&bm>; + bm,pool-long = <2>; + bm,pool-short = <3>; + }; + + usb@58000 { + status = "okay"; + }; + + /* CON4 */ + ethernet@70000 { + pinctrl-names = "default"; + + /* + * The Reference Clock 0 is used to + * provide a clock to the PHY + */ + pinctrl-0 = <&ge0_rgmii_pins>, <&ref_clk0_pins>; + status = "okay"; + phy = <&phy0>; + phy-mode = "rgmii-id"; + buffer-manager = <&bm>; + bm,pool-long = <0>; + bm,pool-short = <3>; + }; + + crypto@90000 { + status = "okay"; + }; + + crypto@92000 { + status = "okay"; + }; + + bm@c8000 { + status = "okay"; + }; + + nfc: flash@d0000 { + status = "okay"; + num-cs = <1>; + nand-ecc-strength = <4>; + nand-ecc-step-size = <512>; + marvell,nand-keep-config; + marvell,nand-enable-arbiter; + nand-on-flash-bbt; + + partitions { + compatible = "fixed-partitions"; + #address-cells = <1>; + #size-cells = <1>; + + partition@0 { + label = "U-Boot"; + reg = <0x00000000 0x00800000>; + read-only; + }; + + partition@800000 { + label = "uImage"; + reg = <0x00800000 0x00400000>; + read-only; + }; + + partition@c00000 { + label = "Root"; + reg = <0x00c00000 0x3f400000>; + }; + }; + }; + + usb3@f0000 { + status = "okay"; + usb-phy = <&usb3_phy>; + }; + }; + + bm-bppi { + status = "okay"; + }; + + pcie-controller { + status = "okay"; + + /* + * The three PCIe units are accessible through + * standard mini-PCIe slots on the board. + */ + pcie@1,0 { + /* Port 0, Lane 0 */ + status = "okay"; + }; + + pcie@2,0 { + /* Port 1, Lane 0 */ + status = "okay"; + }; + + pcie@3,0 { + /* Port 2, Lane 0 */ + status = "okay"; + }; + }; + }; + + usb3_phy: usb3_phy { + compatible = "usb-nop-xceiv"; + vcc-supply = <®_xhci0_vbus>; + }; + + reg_xhci0_vbus: xhci0-vbus { + compatible = "regulator-fixed"; + pinctrl-names = "default"; + pinctrl-0 = <&xhci0_vbus_pins>; + regulator-name = "xhci0-vbus"; + regulator-min-microvolt = <5000000>; + regulator-max-microvolt = <5000000>; + enable-active-high; + gpio = <&gpio1 12 GPIO_ACTIVE_HIGH>; + }; +}; + +&spi1 { + pinctrl-names = "default"; + pinctrl-0 = <&spi1_pins>; + status = "okay"; + + spi-flash@0 { + #address-cells = <1>; + #size-cells = <1>; + compatible = "st,m25p128", "jedec,spi-nor"; + reg = <0>; /* Chip select 0 */ + spi-max-frequency = <54000000>; + }; +}; Property changes on: projects/clang400-import/sys/boot/fdt/dts/arm/armada-385-db-ap.dts ___________________________________________________________________ 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/clang400-import/sys/boot/fdt/dts/arm/armada-388-clearfog.dts =================================================================== --- projects/clang400-import/sys/boot/fdt/dts/arm/armada-388-clearfog.dts (nonexistent) +++ projects/clang400-import/sys/boot/fdt/dts/arm/armada-388-clearfog.dts (revision 311460) @@ -0,0 +1,463 @@ +/* + * Device Tree file for SolidRun Clearfog revision A1 rev 2.0 (88F6828) + * + * Copyright (C) 2015 Russell King + * + * This board is in development; the contents of this file work with + * the A1 rev 2.0 of the board, which does not represent final + * production board. Things will change, don't expect this file to + * remain compatible info the future. + * + * This file is dual-licensed: you can use it either under the terms + * of the GPL or the X11 license, at your option. Note that this dual + * licensing only applies to this file, and not this project as a + * whole. + * + * a) This file is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License + * version 2 as published by the Free Software Foundation. + * + * This file is distributed in the hope that it will be useful + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * Or, alternatively + * + * b) Permission is hereby granted, free of charge, to any person + * obtaining a copy of this software and associated documentation + * files (the "Software"), to deal in the Software without + * restriction, including without limitation the rights to use + * copy, modify, merge, publish, distribute, sublicense, and/or + * sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following + * conditions: + * + * The above copyright notice and this permission notice shall be + * included in all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED , WITHOUT WARRANTY OF ANY KIND + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES + * OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT + * HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * $FreeBSD$ + */ + +/dts-v1/; +#include "armada-388.dtsi" +#include "armada-38x-solidrun-microsom.dtsi" + +/ { + model = "SolidRun Clearfog A1"; + compatible = "solidrun,clearfog-a1", "marvell,armada388", + "marvell,armada385", "marvell,armada380"; + + aliases { + /* So that mvebu u-boot can update the MAC addresses */ + ethernet1 = ð0; + ethernet2 = ð1; + ethernet3 = ð2; + }; + + chosen { + stdout-path = "serial0:115200n8"; + }; + + reg_3p3v: regulator-3p3v { + compatible = "regulator-fixed"; + regulator-name = "3P3V"; + regulator-min-microvolt = <3300000>; + regulator-max-microvolt = <3300000>; + regulator-always-on; + }; + + soc { + internal-regs { + ethernet@30000 { + phy-mode = "sgmii"; + buffer-manager = <&bm>; + bm,pool-long = <2>; + bm,pool-short = <1>; + status = "okay"; + + fixed-link { + speed = <1000>; + full-duplex; + }; + }; + + ethernet@34000 { + phy-mode = "sgmii"; + buffer-manager = <&bm>; + bm,pool-long = <3>; + bm,pool-short = <1>; + status = "okay"; + + fixed-link { + speed = <1000>; + full-duplex; + }; + }; + + i2c@11000 { + /* Is there anything on this? */ + clock-frequency = <100000>; + pinctrl-0 = <&i2c0_pins>; + pinctrl-names = "default"; + status = "okay"; + + /* + * PCA9655 GPIO expander, up to 1MHz clock. + * 0-CON3 CLKREQ# + * 1-CON3 PERST# + * 2-CON2 PERST# + * 3-CON3 W_DISABLE + * 4-CON2 CLKREQ# + * 5-USB3 overcurrent + * 6-USB3 power + * 7-CON2 W_DISABLE + * 8-JP4 P1 + * 9-JP4 P4 + * 10-JP4 P5 + * 11-m.2 DEVSLP + * 12-SFP_LOS + * 13-SFP_TX_FAULT + * 14-SFP_TX_DISABLE + * 15-SFP_MOD_DEF0 + */ + expander0: gpio-expander@20 { + /* + * This is how it should be: + * compatible = "onnn,pca9655", + * "nxp,pca9555"; + * but you can't do this because of + * the way I2C works. + */ + compatible = "nxp,pca9555"; + gpio-controller; + #gpio-cells = <2>; + reg = <0x20>; + + pcie1_0_clkreq { + gpio-hog; + gpios = <0 GPIO_ACTIVE_LOW>; + input; + line-name = "pcie1.0-clkreq"; + }; + pcie1_0_w_disable { + gpio-hog; + gpios = <3 GPIO_ACTIVE_LOW>; + output-low; + line-name = "pcie1.0-w-disable"; + }; + pcie2_0_clkreq { + gpio-hog; + gpios = <4 GPIO_ACTIVE_LOW>; + input; + line-name = "pcie2.0-clkreq"; + }; + pcie2_0_w_disable { + gpio-hog; + gpios = <7 GPIO_ACTIVE_LOW>; + output-low; + line-name = "pcie2.0-w-disable"; + }; + usb3_ilimit { + gpio-hog; + gpios = <5 GPIO_ACTIVE_LOW>; + input; + line-name = "usb3-current-limit"; + }; + usb3_power { + gpio-hog; + gpios = <6 GPIO_ACTIVE_HIGH>; + output-high; + line-name = "usb3-power"; + }; + m2_devslp { + gpio-hog; + gpios = <11 GPIO_ACTIVE_HIGH>; + output-low; + line-name = "m.2 devslp"; + }; + sfp_los { + /* SFP loss of signal */ + gpio-hog; + gpios = <12 GPIO_ACTIVE_HIGH>; + input; + line-name = "sfp-los"; + }; + sfp_tx_fault { + /* SFP laser fault */ + gpio-hog; + gpios = <13 GPIO_ACTIVE_HIGH>; + input; + line-name = "sfp-tx-fault"; + }; + sfp_tx_disable { + /* SFP transmit disable */ + gpio-hog; + gpios = <14 GPIO_ACTIVE_HIGH>; + output-low; + line-name = "sfp-tx-disable"; + }; + sfp_mod_def0 { + /* SFP module present */ + gpio-hog; + gpios = <15 GPIO_ACTIVE_LOW>; + input; + line-name = "sfp-mod-def0"; + }; + }; + + /* The MCP3021 is 100kHz clock only */ + mikrobus_adc: mcp3021@4c { + compatible = "microchip,mcp3021"; + reg = <0x4c>; + }; + + /* Also something at 0x64 */ + }; + + i2c@11100 { + /* + * Routed to SFP, mikrobus, and PCIe. + * SFP limits this to 100kHz, and requires + * an AT24C01A/02/04 with address pins tied + * low, which takes addresses 0x50 and 0x51. + * Mikrobus doesn't specify beyond an I2C + * bus being present. + * PCIe uses ARP to assign addresses, or + * 0x63-0x64. + */ + clock-frequency = <100000>; + pinctrl-0 = <&clearfog_i2c1_pins>; + pinctrl-names = "default"; + status = "okay"; + }; + + pinctrl@18000 { + clearfog_dsa0_clk_pins: clearfog-dsa0-clk-pins { + marvell,pins = "mpp46"; + marvell,function = "ref"; + }; + clearfog_dsa0_pins: clearfog-dsa0-pins { + marvell,pins = "mpp23", "mpp41"; + marvell,function = "gpio"; + }; + clearfog_i2c1_pins: i2c1-pins { + /* SFP, PCIe, mSATA, mikrobus */ + marvell,pins = "mpp26", "mpp27"; + marvell,function = "i2c1"; + }; + clearfog_sdhci_cd_pins: clearfog-sdhci-cd-pins { + marvell,pins = "mpp20"; + marvell,function = "gpio"; + }; + clearfog_sdhci_pins: clearfog-sdhci-pins { + marvell,pins = "mpp21", "mpp28", + "mpp37", "mpp38", + "mpp39", "mpp40"; + marvell,function = "sd0"; + }; + clearfog_spi1_cs_pins: spi1-cs-pins { + marvell,pins = "mpp55"; + marvell,function = "spi1"; + }; + mikro_pins: mikro-pins { + /* int: mpp22 rst: mpp29 */ + marvell,pins = "mpp22", "mpp29"; + marvell,function = "gpio"; + }; + mikro_spi_pins: mikro-spi-pins { + marvell,pins = "mpp43"; + marvell,function = "spi1"; + }; + mikro_uart_pins: mikro-uart-pins { + marvell,pins = "mpp24", "mpp25"; + marvell,function = "ua1"; + }; + rear_button_pins: rear-button-pins { + marvell,pins = "mpp34"; + marvell,function = "gpio"; + }; + }; + + sata@a8000 { + /* pinctrl? */ + status = "okay"; + }; + + sata@e0000 { + /* pinctrl? */ + status = "okay"; + }; + + sdhci@d8000 { + bus-width = <4>; + cd-gpios = <&gpio0 20 GPIO_ACTIVE_LOW>; + no-1-8-v; + pinctrl-0 = <&clearfog_sdhci_pins + &clearfog_sdhci_cd_pins>; + pinctrl-names = "default"; + status = "okay"; + vmmc = <®_3p3v>; + wp-inverted; + }; + + serial@12100 { + /* mikrobus uart */ + pinctrl-0 = <&mikro_uart_pins>; + pinctrl-names = "default"; + status = "okay"; + }; + + usb@58000 { + /* CON3, nearest power. */ + status = "okay"; + }; + + crypto@90000 { + status = "okay"; + }; + + crypto@92000 { + status = "okay"; + }; + + usb3@f0000 { + /* CON2, nearest CPU, USB2 only. */ + status = "okay"; + }; + + usb3@f8000 { + /* CON7 */ + status = "okay"; + }; + }; + + pcie-controller { + status = "okay"; + /* + * The two PCIe units are accessible through + * the mini-PCIe connectors on the board. + */ + pcie@2,0 { + /* Port 1, Lane 0. CON3, nearest power. */ + reset-gpios = <&expander0 1 GPIO_ACTIVE_LOW>; + status = "okay"; + }; + pcie@3,0 { + /* Port 2, Lane 0. CON2, nearest CPU. */ + reset-gpios = <&expander0 2 GPIO_ACTIVE_LOW>; + status = "okay"; + }; + }; + }; + + dsa@0 { + compatible = "marvell,dsa"; + dsa,ethernet = <ð1>; + dsa,mii-bus = <&mdio>; + pinctrl-0 = <&clearfog_dsa0_clk_pins &clearfog_dsa0_pins>; + pinctrl-names = "default"; + #address-cells = <2>; + #size-cells = <0>; + + switch@0 { + #address-cells = <1>; + #size-cells = <0>; + reg = <4 0>; + + port@0 { + reg = <0>; + label = "lan5"; + vlangroup = <0>; + }; + + port@1 { + reg = <1>; + label = "lan4"; + vlangroup = <0>; + }; + + port@2 { + reg = <2>; + label = "lan3"; + vlangroup = <0>; + }; + + port@3 { + reg = <3>; + label = "lan2"; + vlangroup = <0>; + }; + + port@4 { + reg = <4>; + label = "lan1"; + vlangroup = <0>; + }; + + port@5 { + reg = <5>; + label = "cpu"; + vlangroup = <0>; + }; + + port@6 { + /* 88E1512 external phy */ + reg = <6>; + label = "lan6"; + vlangroup = <0>; + fixed-link { + speed = <1000>; + full-duplex; + }; + }; + }; + }; + + gpio-keys { + compatible = "gpio-keys"; + pinctrl-0 = <&rear_button_pins>; + pinctrl-names = "default"; + + button_0 { + /* The rear SW3 button */ + label = "Rear Button"; + gpios = <&gpio1 2 GPIO_ACTIVE_LOW>; + linux,can-disable; + linux,code = ; + }; + }; +}; + +&spi1 { + /* + * We don't seem to have the W25Q32 on the + * A1 Rev 2.0 boards, so disable SPI. + * CS0: W25Q32 (doesn't appear to be present) + * CS1: + * CS2: mikrobus + */ + pinctrl-0 = <&spi1_pins + &clearfog_spi1_cs_pins + &mikro_spi_pins>; + pinctrl-names = "default"; + status = "okay"; + + spi-flash@0 { + #address-cells = <1>; + #size-cells = <0>; + compatible = "w25q32", "jedec,spi-nor"; + reg = <0>; /* Chip select 0 */ + spi-max-frequency = <3000000>; + status = "disabled"; + }; +}; Property changes on: projects/clang400-import/sys/boot/fdt/dts/arm/armada-388-clearfog.dts ___________________________________________________________________ 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/clang400-import/sys/boot/fdt/dts/arm/armada-38x-solidrun-microsom.dtsi =================================================================== --- projects/clang400-import/sys/boot/fdt/dts/arm/armada-38x-solidrun-microsom.dtsi (nonexistent) +++ projects/clang400-import/sys/boot/fdt/dts/arm/armada-38x-solidrun-microsom.dtsi (revision 311460) @@ -0,0 +1,126 @@ +/* + * Device Tree file for SolidRun Armada 38x Microsom + * + * Copyright (C) 2015 Russell King + * + * This board is in development; the contents of this file work with + * the A1 rev 2.0 of the board, which does not represent final + * production board. Things will change, don't expect this file to + * remain compatible info the future. + * + * This file is dual-licensed: you can use it either under the terms + * of the GPL or the X11 license, at your option. Note that this dual + * licensing only applies to this file, and not this project as a + * whole. + * + * a) This file is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License + * version 2 as published by the Free Software Foundation. + * + * This file is distributed in the hope that it will be useful + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * Or, alternatively + * + * b) Permission is hereby granted, free of charge, to any person + * obtaining a copy of this software and associated documentation + * files (the "Software"), to deal in the Software without + * restriction, including without limitation the rights to use + * copy, modify, merge, publish, distribute, sublicense, and/or + * sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following + * conditions: + * + * The above copyright notice and this permission notice shall be + * included in all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED , WITHOUT WARRANTY OF ANY KIND + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES + * OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT + * HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * $FreeBSD$ + */ +#include +#include + +/ { + memory { + device_type = "memory"; + reg = <0x00000000 0x10000000>; /* 256 MB */ + }; + + soc { + ranges = ; + + internal-regs { + ethernet@70000 { + pinctrl-0 = <&ge0_rgmii_pins>; + pinctrl-names = "default"; + phy = <&phy_dedicated>; + phy-mode = "rgmii-id"; + buffer-manager = <&bm>; + bm,pool-long = <0>; + bm,pool-short = <1>; + status = "okay"; + }; + + mdio@72004 { + /* + * Add the phy clock here, so the phy can be + * accessed to read its IDs prior to binding + * with the driver. + */ + pinctrl-0 = <&mdio_pins µsom_phy_clk_pins>; + pinctrl-names = "default"; + + phy_dedicated: ethernet-phy@0 { + /* + * Annoyingly, the marvell phy driver + * configures the LED register, rather + * than preserving reset-loaded setting. + * We undo that rubbish here. + */ + marvell,reg-init = <3 16 0 0x101e>; + reg = <0>; + }; + }; + + pinctrl@18000 { + microsom_phy_clk_pins: microsom-phy-clk-pins { + marvell,pins = "mpp45"; + marvell,function = "ref"; + }; + }; + + rtc@a3800 { + /* + * If the rtc doesn't work, run "date reset" + * twice in u-boot. + */ + status = "okay"; + }; + + serial@12000 { + pinctrl-0 = <&uart0_pins>; + pinctrl-names = "default"; + status = "okay"; + }; + + bm@c8000 { + status = "okay"; + }; + }; + + bm-bppi { + status = "okay"; + }; + + }; +}; Property changes on: projects/clang400-import/sys/boot/fdt/dts/arm/armada-38x-solidrun-microsom.dtsi ___________________________________________________________________ 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/clang400-import/sys/boot/fdt/dts/arm/armada-38x.dtsi =================================================================== --- projects/clang400-import/sys/boot/fdt/dts/arm/armada-38x.dtsi (revision 311459) +++ projects/clang400-import/sys/boot/fdt/dts/arm/armada-38x.dtsi (revision 311460) @@ -1,661 +1,681 @@ /* * Device Tree Include file for Marvell Armada 38x family of SoCs. * * Copyright (C) 2014 Marvell * * Lior Amsalem * Gregory CLEMENT * Thomas Petazzoni * * This file is dual-licensed: you can use it either under the terms * of the GPL or the X11 license, at your option. Note that this dual * licensing only applies to this file, and not this project as a * whole. * * a) This file is free software; you can redistribute it and/or * modify it under the terms of the GNU General Public License as * published by the Free Software Foundation; either version 2 of the * License, or (at your option) any later version. * * This file is distributed in the hope that it will be useful * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the * GNU General Public License for more details. * * Or, alternatively * * b) Permission is hereby granted, free of charge, to any person * obtaining a copy of this software and associated documentation * files (the "Software"), to deal in the Software without * restriction, including without limitation the rights to use * copy, modify, merge, publish, distribute, sublicense, and/or * sell copies of the Software, and to permit persons to whom the * Software is furnished to do so, subject to the following * conditions: * * The above copyright notice and this permission notice shall be * included in all copies or substantial portions of the Software. * * THE SOFTWARE IS PROVIDED , WITHOUT WARRANTY OF ANY KIND * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES * OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT * HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR * OTHER DEALINGS IN THE SOFTWARE. * * $FreeBSD$ */ #include "skeleton.dtsi" #include #include #define MBUS_ID(target,attributes) (((target) << 24) | ((attributes) << 16)) / { model = "Marvell Armada 38x family SoC"; compatible = "marvell,armada380"; aliases { gpio0 = &gpio0; gpio1 = &gpio1; serial0 = &uart0; serial1 = &uart1; sram0 = &SRAM0; sram1 = &SRAM1; }; pmu { compatible = "arm,cortex-a9-pmu"; interrupts-extended = <&mpic 3>; }; SRAM0: sram@f1100000 { compatible = "mrvl,cesa-sram"; reg = <0xf1100000 0x0010000>; }; SRAM1: sram@f1110000 { compatible = "mrvl,cesa-sram"; reg = <0xf1110000 0x0010000>; }; soc { compatible = "marvell,armada380-mbus", "simple-bus"; #address-cells = <2>; #size-cells = <1>; controller = <&mbusc>; interrupt-parent = <&gic>; pcie-mem-aperture = <0xe0000000 0x8000000>; pcie-io-aperture = <0xe8000000 0x100000>; bootrom { compatible = "marvell,bootrom"; reg = ; }; devbus-bootcs { compatible = "marvell,mvebu-devbus"; reg = ; ranges = <0 MBUS_ID(0x01, 0x2f) 0 0xffffffff>; #address-cells = <1>; #size-cells = <1>; clocks = <&coreclk 0>; status = "disabled"; }; devbus-cs0 { compatible = "marvell,mvebu-devbus"; reg = ; ranges = <0 MBUS_ID(0x01, 0x3e) 0 0xffffffff>; #address-cells = <1>; #size-cells = <1>; clocks = <&coreclk 0>; status = "disabled"; }; devbus-cs1 { compatible = "marvell,mvebu-devbus"; reg = ; ranges = <0 MBUS_ID(0x01, 0x3d) 0 0xffffffff>; #address-cells = <1>; #size-cells = <1>; clocks = <&coreclk 0>; status = "disabled"; }; devbus-cs2 { compatible = "marvell,mvebu-devbus"; reg = ; ranges = <0 MBUS_ID(0x01, 0x3b) 0 0xffffffff>; #address-cells = <1>; #size-cells = <1>; clocks = <&coreclk 0>; status = "disabled"; }; devbus-cs3 { compatible = "marvell,mvebu-devbus"; reg = ; ranges = <0 MBUS_ID(0x01, 0x37) 0 0xffffffff>; #address-cells = <1>; #size-cells = <1>; clocks = <&coreclk 0>; status = "disabled"; }; internal-regs { compatible = "simple-bus"; #address-cells = <1>; #size-cells = <1>; ranges = <0 MBUS_ID(0xf0, 0x01) 0 0x100000>; crypto@90000 { compatible = "mrvl,cesa"; - reg = <0x90000 0x10000>; + reg = <0x90000 0x1000 /* tdma base reg chan 0 */ + 0x9D000 0x1000>; /* cesa base reg chan 0 */ interrupts = ; interrupt-parent = <&gic>; sram-handle = <&SRAM0>; status = "disabled"; }; crypto@92000 { compatible = "mrvl,cesa"; reg = <0x92000 0x1000 /* tdma base reg chan 1 */ 0x9F000 0x1000>; /* cesa base reg chan 1 */ interrupts = ; interrupt-parent = <&gic>; sram-handle = <&SRAM1>; status = "disabled"; }; L2: cache-controller@8000 { compatible = "arm,pl310-cache"; reg = <0x8000 0x1000>; cache-unified; cache-level = <2>; }; scu@c000 { compatible = "arm,cortex-a9-scu"; reg = <0xc000 0x58>; }; timer@c200 { compatible = "arm,cortex-a9-global-timer"; reg = <0xc200 0x20>; interrupts = ; clock-frequency = <800000000>; clocks = <&coreclk 2>; }; timer@c600 { compatible = "arm,cortex-a9-twd-timer"; reg = <0xc600 0x20>; interrupts = ; clock-frequency = <800000000>; clocks = <&coreclk 2>; }; gic: interrupt-controller@d000 { compatible = "arm,cortex-a9-gic"; #interrupt-cells = <3>; #size-cells = <0>; interrupt-controller; reg = <0xd000 0x1000>, <0xc100 0x100>; }; spi0: spi@10600 { compatible = "marvell,armada-380-spi", "marvell,orion-spi"; reg = <0x10600 0x50>; #address-cells = <1>; #size-cells = <0>; cell-index = <0>; interrupts = ; clocks = <&coreclk 0>; status = "disabled"; }; spi1: spi@10680 { compatible = "marvell,armada-380-spi", "marvell,orion-spi"; reg = <0x10680 0x50>; #address-cells = <1>; #size-cells = <0>; cell-index = <1>; interrupts = ; clocks = <&coreclk 0>; status = "disabled"; }; i2c0: i2c@11000 { compatible = "marvell,mv64xxx-i2c"; reg = <0x11000 0x20>; #address-cells = <1>; #size-cells = <0>; interrupts = ; timeout-ms = <1000>; clocks = <&coreclk 0>; status = "disabled"; }; i2c1: i2c@11100 { compatible = "marvell,mv64xxx-i2c"; reg = <0x11100 0x20>; #address-cells = <1>; #size-cells = <0>; interrupts = ; timeout-ms = <1000>; clocks = <&coreclk 0>; status = "disabled"; }; uart0: serial@12000 { compatible = "snps,dw-apb-uart"; reg = <0x12000 0x100>; reg-shift = <2>; interrupts = ; reg-io-width = <1>; clocks = <&coreclk 0>; status = "disabled"; }; uart1: serial@12100 { compatible = "snps,dw-apb-uart"; reg = <0x12100 0x100>; reg-shift = <2>; interrupts = ; reg-io-width = <1>; clocks = <&coreclk 0>; status = "disabled"; }; pinctrl: pinctrl@18000 { reg = <0x18000 0x20>; ge0_rgmii_pins: ge-rgmii-pins-0 { marvell,pins = "mpp6", "mpp7", "mpp8", "mpp9", "mpp10", "mpp11", "mpp12", "mpp13", "mpp14", "mpp15", "mpp16", "mpp17"; marvell,function = "ge0"; }; ge1_rgmii_pins: ge-rgmii-pins-1 { marvell,pins = "mpp21", "mpp27", "mpp28", "mpp29", "mpp30", "mpp31", "mpp32", "mpp37", "mpp38", "mpp39", "mpp40", "mpp41"; marvell,function = "ge1"; }; i2c0_pins: i2c-pins-0 { marvell,pins = "mpp2", "mpp3"; marvell,function = "i2c0"; }; mdio_pins: mdio-pins { marvell,pins = "mpp4", "mpp5"; marvell,function = "ge"; }; ref_clk0_pins: ref-clk-pins-0 { marvell,pins = "mpp45"; marvell,function = "ref"; }; ref_clk1_pins: ref-clk-pins-1 { marvell,pins = "mpp46"; marvell,function = "ref"; }; spi0_pins: spi-pins-0 { marvell,pins = "mpp22", "mpp23", "mpp24", "mpp25"; marvell,function = "spi0"; }; spi1_pins: spi-pins-1 { marvell,pins = "mpp56", "mpp57", "mpp58", "mpp59"; marvell,function = "spi1"; }; uart0_pins: uart-pins-0 { marvell,pins = "mpp0", "mpp1"; marvell,function = "ua0"; }; uart1_pins: uart-pins-1 { marvell,pins = "mpp19", "mpp20"; marvell,function = "ua1"; }; sdhci_pins: sdhci-pins { marvell,pins = "mpp48", "mpp49", "mpp50", "mpp52", "mpp53", "mpp54", "mpp55", "mpp57", "mpp58", "mpp59"; marvell,function = "sd0"; }; sata0_pins: sata-pins-0 { marvell,pins = "mpp20"; marvell,function = "sata0"; }; sata1_pins: sata-pins-1 { marvell,pins = "mpp19"; marvell,function = "sata1"; }; sata2_pins: sata-pins-2 { marvell,pins = "mpp47"; marvell,function = "sata2"; }; sata3_pins: sata-pins-3 { marvell,pins = "mpp44"; marvell,function = "sata3"; }; }; gpio0: gpio@18100 { compatible = "marvell,orion-gpio"; reg = <0x18100 0x40>; ngpios = <32>; gpio-controller; #gpio-cells = <2>; interrupt-controller; #interrupt-cells = <2>; interrupts = , , , ; }; gpio1: gpio@18140 { compatible = "marvell,orion-gpio"; reg = <0x18140 0x40>; ngpios = <28>; gpio-controller; #gpio-cells = <2>; interrupt-controller; #interrupt-cells = <2>; interrupts = , , , ; }; system-controller@18200 { compatible = "marvell,armada-380-system-controller", "marvell,armada-370-xp-system-controller"; reg = <0x18200 0x100>; }; gateclk: clock-gating-control@18220 { compatible = "marvell,armada-380-gating-clock"; reg = <0x18220 0x4>; clocks = <&coreclk 0>; #clock-cells = <1>; }; coreclk: mvebu-sar@18600 { compatible = "marvell,armada-380-core-clock"; reg = <0x18600 0x04>; #clock-cells = <1>; }; mbusc: mbus-controller@20000 { compatible = "marvell,mbus-controller"; reg = <0x20000 0x100>, <0x20180 0x20>; }; mpic: interrupt-controller@20a00 { compatible = "marvell,mpic"; reg = <0x20a00 0x2d0>, <0x21070 0x58>; #interrupt-cells = <1>; #size-cells = <1>; interrupt-controller; msi-controller; interrupts = ; }; timer@20300 { compatible = "marvell,armada-380-timer", "marvell,armada-xp-timer"; reg = <0x20300 0x30>, <0x21040 0x30>; interrupts-extended = <&gic GIC_SPI 8 IRQ_TYPE_LEVEL_HIGH>, <&gic GIC_SPI 9 IRQ_TYPE_LEVEL_HIGH>, <&gic GIC_SPI 10 IRQ_TYPE_LEVEL_HIGH>, <&gic GIC_SPI 11 IRQ_TYPE_LEVEL_HIGH>, <&mpic 5>, <&mpic 6>; clocks = <&coreclk 2>, <&refclk>; clock-names = "nbclk", "fixed"; }; watchdog@20300 { compatible = "marvell,armada-380-wdt"; reg = <0x20300 0x34>, <0x20704 0x4>, <0x18260 0x4>; clocks = <&coreclk 2>, <&refclk>; clock-names = "nbclk", "fixed"; }; cpurst@20800 { compatible = "marvell,armada-370-cpu-reset"; reg = <0x20800 0x10>; }; mpcore-soc-ctrl@20d20 { compatible = "marvell,armada-380-mpcore-soc-ctrl"; reg = <0x20d20 0x6c>; }; coherency-fabric@21010 { compatible = "marvell,armada-380-coherency-fabric"; reg = <0x21010 0x1c>; }; pmsu@22000 { compatible = "marvell,armada-380-pmsu"; reg = <0x22000 0x1000>; }; eth1: ethernet@30000 { compatible = "marvell,armada-370-neta"; reg = <0x30000 0x4000>; interrupts-extended = <&mpic 10>; clocks = <&gateclk 3>; status = "disabled"; }; eth2: ethernet@34000 { compatible = "marvell,armada-370-neta"; reg = <0x34000 0x4000>; interrupts-extended = <&mpic 12>; clocks = <&gateclk 2>; status = "disabled"; }; usb@58000 { compatible = "marvell,orion-ehci"; reg = <0x58000 0x500>; interrupts = ; clocks = <&gateclk 18>; status = "disabled"; }; xor@60800 { compatible = "marvell,orion-xor"; reg = <0x60800 0x100 0x60a00 0x100>; clocks = <&gateclk 22>; status = "okay"; xor00 { interrupts = ; dmacap,memcpy; dmacap,xor; }; xor01 { interrupts = ; dmacap,memcpy; dmacap,xor; dmacap,memset; }; }; xor@60900 { compatible = "marvell,orion-xor"; reg = <0x60900 0x100 0x60b00 0x100>; clocks = <&gateclk 28>; status = "okay"; xor10 { interrupts = ; dmacap,memcpy; dmacap,xor; }; xor11 { interrupts = ; dmacap,memcpy; dmacap,xor; dmacap,memset; }; }; eth0: ethernet@70000 { compatible = "marvell,armada-370-neta"; reg = <0x70000 0x4000>; interrupts-extended = <&mpic 8>; clocks = <&gateclk 4>; status = "disabled"; }; mdio: mdio@72004 { #address-cells = <1>; #size-cells = <0>; compatible = "marvell,orion-mdio"; reg = <0x72004 0x4>; clocks = <&gateclk 4>; }; rtc@a3800 { compatible = "marvell,armada-380-rtc"; reg = <0xa3800 0x20>, <0x184a0 0x0c>; reg-names = "rtc", "rtc-soc"; interrupts = ; }; sata@a8000 { compatible = "marvell,armada-380-ahci"; reg = <0xa8000 0x2000>; interrupts = ; clocks = <&gateclk 15>; status = "disabled"; }; + bm: bm@c8000 { + compatible = "marvell,armada-380-neta-bm"; + reg = <0xc8000 0xac>; + clocks = <&gateclk 13>; + internal-mem = <&bm_bppi>; + status = "disabled"; + }; + sata@e0000 { compatible = "marvell,armada-380-ahci"; reg = <0xe0000 0x2000>; interrupts = ; clocks = <&gateclk 30>; status = "disabled"; }; coredivclk: clock@e4250 { compatible = "marvell,armada-380-corediv-clock"; reg = <0xe4250 0xc>; #clock-cells = <1>; clocks = <&mainpll>; clock-output-names = "nand"; }; thermal@e8078 { compatible = "marvell,armada380-thermal"; reg = <0xe4078 0x4>, <0xe4074 0x4>; status = "okay"; }; flash@d0000 { compatible = "marvell,armada370-nand"; reg = <0xd0000 0x54>; #address-cells = <1>; #size-cells = <1>; interrupts = ; clocks = <&coredivclk 0>; status = "disabled"; }; sdhci@d8000 { compatible = "marvell,armada-380-sdhci"; reg-names = "sdhci", "mbus", "conf-sdio3"; reg = <0xd8000 0x1000>, <0xdc000 0x100>, <0x18454 0x4>; interrupts = ; clocks = <&gateclk 17>; mrvl,clk-delay-cycles = <0x1F>; status = "disabled"; }; usb3@f0000 { compatible = "marvell,armada-380-xhci"; reg = <0xf0000 0x4000>,<0xf4000 0x4000>; interrupts = ; clocks = <&gateclk 9>; status = "disabled"; }; usb3@f8000 { compatible = "marvell,armada-380-xhci"; reg = <0xf8000 0x4000>,<0xfc000 0x4000>; interrupts = ; clocks = <&gateclk 10>; status = "disabled"; }; + }; + + bm_bppi: bm-bppi { + compatible = "mmio-sram"; + reg = ; + ranges = <0 MBUS_ID(0x0c, 0x04) 0 0x100000>; + #address-cells = <1>; + #size-cells = <1>; + clocks = <&gateclk 13>; + no-memory-wc; + status = "disabled"; }; }; pci0: pcie@f1080000 { compatible = "mrvl,pcie"; status = "disabled"; device_type = "pci"; #interrupt-cells = <3>; #size-cells = <2>; #address-cells = <3>; reg = <0xf1080000 0x2000>; bus-range = <0 255>; ranges = <0x42000000 0x0 0xf1200000 0xf1200000 0x0 0x00100000 0x41000000 0x0 0x00000000 0xf1300000 0x0 0x00100000>; interrupt-parent = <&gic>; interrupts = ; interrupt-map-mask = <0xf800 0x0 0x0 0x7>; interrupt-map = < 0x0000 0x0 0x0 0x1 &gic GIC_SPI 29 IRQ_TYPE_LEVEL_HIGH >; }; clocks { /* 2 GHz fixed main PLL */ mainpll: mainpll { compatible = "fixed-clock"; #clock-cells = <0>; clock-frequency = <1000000000>; }; /* 25 MHz reference crystal */ refclk: oscillator { compatible = "fixed-clock"; #clock-cells = <0>; clock-frequency = <25000000>; }; }; }; Index: projects/clang400-import/sys/boot/mips/beri/boot2/Makefile =================================================================== --- projects/clang400-import/sys/boot/mips/beri/boot2/Makefile (revision 311459) +++ projects/clang400-import/sys/boot/mips/beri/boot2/Makefile (revision 311460) @@ -1,89 +1,89 @@ #- # Copyright (c) 2013-2014 Robert N. M. Watson # All rights reserved. # # This software was developed by SRI International and the University of # Cambridge Computer Laboratory under DARPA/AFRL contract (FA8750-10-C-0237) # ("CTSRD"), as part of the DARPA CRASH research programme. # # 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. # # $FreeBSD$ BINDIR?= /boot INSTALLFLAGS= -b LOADERS= flashboot jtagboot FILES= ${LOADERS} ${LOADERS:S/$/.md5/} SRCS= relocate.S \ start.S \ boot2.c \ altera_jtag_uart.c \ cfi.c \ sdcard.c MAN= AFLAGS= -G0 CFLAGS= -ffreestanding \ -I${.CURDIR} \ -I${.CURDIR}/../../../common \ -I${.CURDIR}/../../../.. \ -D_KERNEL \ -Wall \ -G0 -Xassembler -G0 \ -fno-pic -mno-abicalls \ -msoft-float \ -g # where to get libstand from CFLAGS+= -I${.CURDIR}/../../../../../lib/libstand/ LIBSTAND= ${.OBJDIR}/../../../../../lib/libstand/libstand.a LDFLAGS= -nostdlib \ -static \ -N \ -G0 \ -L${.CURDIR} .PATH: ${.CURDIR}/../common CFLAGS+= -I${.CURDIR}/../common flashboot.elf: relocate.o start.o boot2.o altera_jtag_uart.o cfi.o sdcard.o - ${LD} ${_LDFLAGS} -T ${.CURDIR}/flashboot.ldscript -o ${.TARGET} \ + ${CC} ${_LDFLAGS} -T ${.CURDIR}/flashboot.ldscript -o ${.TARGET} \ ${.ALLSRC} ${LIBSTAND} flashboot: flashboot.elf ${OBJCOPY} -S -O binary ${.TARGET}.elf ${.TARGET} flashboot.md5: flashboot md5 flashboot > flashboot.md5 jtagboot: start.o boot2.o altera_jtag_uart.o cfi.o sdcard.o - ${LD} ${_LDFLAGS} -T ${.CURDIR}/jtagboot.ldscript -o ${.TARGET} \ + ${CC} ${_LDFLAGS} -T ${.CURDIR}/jtagboot.ldscript -o ${.TARGET} \ ${.ALLSRC} ${LIBSTAND} jtagboot.md5: jtagboot md5 jtagboot > jtagboot.md5 CLEANFILES+= flashboot.elf .include Index: projects/clang400-import/sys/boot/mips/beri/common/common.ldscript =================================================================== --- projects/clang400-import/sys/boot/mips/beri/common/common.ldscript (revision 311459) +++ projects/clang400-import/sys/boot/mips/beri/common/common.ldscript (revision 311460) @@ -1,77 +1,76 @@ /*- * Copyright (c) 2011-2014 Robert N. M. Watson * All rights reserved. * * This software was developed by SRI International and the University of * Cambridge Computer Laboratory under DARPA/AFRL contract (FA8750-10-C-0237) * ("CTSRD"), as part of the DARPA CRASH research programme. * * 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. * * $FreeBSD$ */ /* * MIPS segment definitions. */ __mips_ckseg_cached__ = 0xffffffff80000000; /* BSD kernel here. */ __mips64_xkphys_cached__ = 0x9800000000000000; /* Device memory here. */ __mips64_xkphys_uncached__ = 0x9000000000000000; /* Device I/O here. */ /* * Physical addresses of various peripherals. */ __cheri_flash_base__ = 0x74000000; __cheri_sdcard_base__ = 0x7f008000; /* * Location of boot2 in flash. */ __cheri_flash_boot_loader_base_ = 0x03fe0000; __cheri_flash_boot_loader_vaddr__ = __mips64_xkphys_cached__ + __cheri_flash_base__ + __cheri_flash_boot_loader_base_; /* * Location of boot file system in flash. */ __cheri_flash_bootfs_base__ = 0x1820000; __cheri_flash_bootfs_len__ = 0x27c0000; __cheri_flash_bootfs_vaddr__ = __mips64_xkphys_cached__ + __cheri_flash_base__ + __cheri_flash_bootfs_base__; /* * Location of SD card controller. */ __cheri_sdcard_vaddr__ = __mips64_xkphys_uncached__ + __cheri_sdcard_base__; /* * Location where the production kernel gets put. This must agree with other * definitions, such as in the kernel's own linker script. * * (As it happens, in the short run, we also place boot2 here, as Miniboot * expects to find an ELF binary there -- but that will change.) */ __kernel_base__ = 0x100000; __kernel_vaddr__ = __mips64_xkphys_cached__ + __kernel_base__; -OUTPUT_FORMAT("elf64-tradbigmips"); OUTPUT_ARCH(mips) Index: projects/clang400-import/sys/boot/mips/beri/loader/loader.ldscript =================================================================== --- projects/clang400-import/sys/boot/mips/beri/loader/loader.ldscript (revision 311459) +++ projects/clang400-import/sys/boot/mips/beri/loader/loader.ldscript (revision 311460) @@ -1,85 +1,84 @@ /*- * Copyright (c) 2011-2014 Robert N. M. Watson * All rights reserved. * * This software was developed by SRI International and the University of * Cambridge Computer Laboratory under DARPA/AFRL contract (FA8750-10-C-0237) * ("CTSRD"), as part of the DARPA CRASH research programme. * * 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 ../common/common.ldscript /* * Location where loader will execute. */ __loader_base__ = 0x20000; __loader_base_vaddr__ = __mips64_xkphys_cached__ + __loader_base__; /* * Highest address the loader is allowed to use below the kernel. */ __loader_end__ = 0x100000; __loader_end_vaddr__ = __mips64_xkphys_cached__ + __loader_end__; -OUTPUT_FORMAT("elf64-tradbigmips"); OUTPUT_ARCH(mips) ENTRY(start) SECTIONS { /* * We rely on boot2 having (a) configured a stack, and (b) loaded us * to an appropriate bit of physical/virtual memory such that no * self-relocating code is required here. */ . = __loader_base_vaddr__; . += SIZEOF_HEADERS; .text ALIGN(0x10): { start.o(.text*) *(EXCLUDE_FILE (start.o) .text*) *(.rodata*) __start_set_Xcommand_set = .; KEEP(*(set_Xcommand_set)) __stop_set_Xcommand_set = .; __start_set_Xficl_compile_set = .; KEEP(*(set_Xficl_compile_set)) __stop_set_Xficl_compile_set = .; } .data ALIGN(0x10): { *(.data*)} .bss ALIGN(0x10): { *(.bss*) } __heap = ALIGN(0x8); /* 64-bit aligned heap pointer */ __data_end = .; __boot_loader_len__ = . - __loader_base_vaddr__; __bss_start = ADDR(.bss); __bss_end = ALIGN(__bss_start + SIZEOF(.bss), 0x8); __heap_start = .; __heap_end = __loader_end_vaddr__; __heap_len = __heap_end - __heap_start; } Index: projects/clang400-import/sys/cam/ctl/ctl_ha.c =================================================================== --- projects/clang400-import/sys/cam/ctl/ctl_ha.c (revision 311459) +++ projects/clang400-import/sys/cam/ctl/ctl_ha.c (revision 311460) @@ -1,1029 +1,1030 @@ /*- * Copyright (c) 2015 Alexander Motin * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer, * without modification, immediately at the beginning of the file. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #if (__FreeBSD_version < 1100000) struct mbufq { struct mbuf *head; struct mbuf *tail; }; static void mbufq_init(struct mbufq *q, int limit) { q->head = q->tail = NULL; } static void mbufq_drain(struct mbufq *q) { struct mbuf *m; while ((m = q->head) != NULL) { q->head = m->m_nextpkt; m_freem(m); } q->tail = NULL; } static struct mbuf * mbufq_dequeue(struct mbufq *q) { struct mbuf *m; m = q->head; if (m) { if (q->tail == m) q->tail = NULL; q->head = m->m_nextpkt; m->m_nextpkt = NULL; } return (m); } static void mbufq_enqueue(struct mbufq *q, struct mbuf *m) { m->m_nextpkt = NULL; if (q->tail) q->tail->m_nextpkt = m; else q->head = m; q->tail = m; } static u_int sbavail(struct sockbuf *sb) { return (sb->sb_cc); } #if (__FreeBSD_version < 1000000) #define mtodo(m, o) ((void *)(((m)->m_data) + (o))) #endif #endif struct ha_msg_wire { uint32_t channel; uint32_t length; }; struct ha_dt_msg_wire { ctl_ha_dt_cmd command; uint32_t size; uint8_t *local; uint8_t *remote; }; struct ha_softc { struct ctl_softc *ha_ctl_softc; ctl_evt_handler ha_handler[CTL_HA_CHAN_MAX]; char ha_peer[128]; struct sockaddr_in ha_peer_in; struct socket *ha_lso; struct socket *ha_so; struct mbufq ha_sendq; struct mbuf *ha_sending; struct mtx ha_lock; int ha_connect; int ha_listen; int ha_connected; int ha_receiving; int ha_wakeup; int ha_disconnect; int ha_shutdown; eventhandler_tag ha_shutdown_eh; TAILQ_HEAD(, ctl_ha_dt_req) ha_dts; } ha_softc; static void ctl_ha_conn_wake(struct ha_softc *softc) { mtx_lock(&softc->ha_lock); softc->ha_wakeup = 1; mtx_unlock(&softc->ha_lock); wakeup(&softc->ha_wakeup); } static int ctl_ha_lupcall(struct socket *so, void *arg, int waitflag) { struct ha_softc *softc = arg; ctl_ha_conn_wake(softc); return (SU_OK); } static int ctl_ha_rupcall(struct socket *so, void *arg, int waitflag) { struct ha_softc *softc = arg; wakeup(&softc->ha_receiving); return (SU_OK); } static int ctl_ha_supcall(struct socket *so, void *arg, int waitflag) { struct ha_softc *softc = arg; ctl_ha_conn_wake(softc); return (SU_OK); } static void ctl_ha_evt(struct ha_softc *softc, ctl_ha_channel ch, ctl_ha_event evt, int param) { int i; if (ch < CTL_HA_CHAN_MAX) { if (softc->ha_handler[ch]) softc->ha_handler[ch](ch, evt, param); return; } for (i = 0; i < CTL_HA_CHAN_MAX; i++) { if (softc->ha_handler[i]) softc->ha_handler[i](i, evt, param); } } static void ctl_ha_close(struct ha_softc *softc) { struct socket *so = softc->ha_so; int report = 0; if (softc->ha_connected || softc->ha_disconnect) { softc->ha_connected = 0; mbufq_drain(&softc->ha_sendq); m_freem(softc->ha_sending); softc->ha_sending = NULL; report = 1; } if (so) { SOCKBUF_LOCK(&so->so_rcv); soupcall_clear(so, SO_RCV); while (softc->ha_receiving) { wakeup(&softc->ha_receiving); msleep(&softc->ha_receiving, SOCKBUF_MTX(&so->so_rcv), 0, "ha_rx exit", 0); } SOCKBUF_UNLOCK(&so->so_rcv); SOCKBUF_LOCK(&so->so_snd); soupcall_clear(so, SO_SND); SOCKBUF_UNLOCK(&so->so_snd); softc->ha_so = NULL; if (softc->ha_connect) pause("reconnect", hz / 2); soclose(so); } if (report) { ctl_ha_evt(softc, CTL_HA_CHAN_MAX, CTL_HA_EVT_LINK_CHANGE, (softc->ha_connect || softc->ha_listen) ? CTL_HA_LINK_UNKNOWN : CTL_HA_LINK_OFFLINE); } } static void ctl_ha_lclose(struct ha_softc *softc) { if (softc->ha_lso) { SOCKBUF_LOCK(&softc->ha_lso->so_rcv); soupcall_clear(softc->ha_lso, SO_RCV); SOCKBUF_UNLOCK(&softc->ha_lso->so_rcv); soclose(softc->ha_lso); softc->ha_lso = NULL; } } static void ctl_ha_rx_thread(void *arg) { struct ha_softc *softc = arg; struct socket *so = softc->ha_so; struct ha_msg_wire wire_hdr; struct uio uio; struct iovec iov; int error, flags, next; bzero(&wire_hdr, sizeof(wire_hdr)); while (1) { if (wire_hdr.length > 0) next = wire_hdr.length; else next = sizeof(wire_hdr); SOCKBUF_LOCK(&so->so_rcv); while (sbavail(&so->so_rcv) < next || softc->ha_disconnect) { if (softc->ha_connected == 0 || softc->ha_disconnect || so->so_error || (so->so_rcv.sb_state & SBS_CANTRCVMORE)) { goto errout; } so->so_rcv.sb_lowat = next; msleep(&softc->ha_receiving, SOCKBUF_MTX(&so->so_rcv), 0, "-", 0); } SOCKBUF_UNLOCK(&so->so_rcv); if (wire_hdr.length == 0) { iov.iov_base = &wire_hdr; iov.iov_len = sizeof(wire_hdr); uio.uio_iov = &iov; uio.uio_iovcnt = 1; uio.uio_rw = UIO_READ; uio.uio_segflg = UIO_SYSSPACE; uio.uio_td = curthread; uio.uio_resid = sizeof(wire_hdr); flags = MSG_DONTWAIT; error = soreceive(softc->ha_so, NULL, &uio, NULL, NULL, &flags); if (error != 0) { printf("%s: header receive error %d\n", __func__, error); SOCKBUF_LOCK(&so->so_rcv); goto errout; } } else { ctl_ha_evt(softc, wire_hdr.channel, CTL_HA_EVT_MSG_RECV, wire_hdr.length); wire_hdr.length = 0; } } errout: softc->ha_receiving = 0; wakeup(&softc->ha_receiving); SOCKBUF_UNLOCK(&so->so_rcv); ctl_ha_conn_wake(softc); kthread_exit(); } static void ctl_ha_send(struct ha_softc *softc) { struct socket *so = softc->ha_so; int error; while (1) { if (softc->ha_sending == NULL) { mtx_lock(&softc->ha_lock); softc->ha_sending = mbufq_dequeue(&softc->ha_sendq); mtx_unlock(&softc->ha_lock); if (softc->ha_sending == NULL) { so->so_snd.sb_lowat = so->so_snd.sb_hiwat + 1; break; } } SOCKBUF_LOCK(&so->so_snd); if (sbspace(&so->so_snd) < softc->ha_sending->m_pkthdr.len) { so->so_snd.sb_lowat = softc->ha_sending->m_pkthdr.len; SOCKBUF_UNLOCK(&so->so_snd); break; } SOCKBUF_UNLOCK(&so->so_snd); error = sosend(softc->ha_so, NULL, NULL, softc->ha_sending, NULL, MSG_DONTWAIT, curthread); softc->ha_sending = NULL; if (error != 0) { printf("%s: sosend() error %d\n", __func__, error); return; } } } static void ctl_ha_sock_setup(struct ha_softc *softc) { struct sockopt opt; struct socket *so = softc->ha_so; int error, val; val = 1024 * 1024; error = soreserve(so, val, val); if (error) printf("%s: soreserve failed %d\n", __func__, error); SOCKBUF_LOCK(&so->so_rcv); so->so_rcv.sb_lowat = sizeof(struct ha_msg_wire); soupcall_set(so, SO_RCV, ctl_ha_rupcall, softc); SOCKBUF_UNLOCK(&so->so_rcv); SOCKBUF_LOCK(&so->so_snd); so->so_snd.sb_lowat = sizeof(struct ha_msg_wire); soupcall_set(so, SO_SND, ctl_ha_supcall, softc); SOCKBUF_UNLOCK(&so->so_snd); bzero(&opt, sizeof(struct sockopt)); opt.sopt_dir = SOPT_SET; opt.sopt_level = SOL_SOCKET; opt.sopt_name = SO_KEEPALIVE; opt.sopt_val = &val; opt.sopt_valsize = sizeof(val); val = 1; error = sosetopt(so, &opt); if (error) printf("%s: KEEPALIVE setting failed %d\n", __func__, error); opt.sopt_level = IPPROTO_TCP; opt.sopt_name = TCP_NODELAY; val = 1; error = sosetopt(so, &opt); if (error) printf("%s: NODELAY setting failed %d\n", __func__, error); opt.sopt_name = TCP_KEEPINIT; val = 3; error = sosetopt(so, &opt); if (error) printf("%s: KEEPINIT setting failed %d\n", __func__, error); opt.sopt_name = TCP_KEEPIDLE; val = 1; error = sosetopt(so, &opt); if (error) printf("%s: KEEPIDLE setting failed %d\n", __func__, error); opt.sopt_name = TCP_KEEPINTVL; val = 1; error = sosetopt(so, &opt); if (error) printf("%s: KEEPINTVL setting failed %d\n", __func__, error); opt.sopt_name = TCP_KEEPCNT; val = 5; error = sosetopt(so, &opt); if (error) printf("%s: KEEPCNT setting failed %d\n", __func__, error); } static int ctl_ha_connect(struct ha_softc *softc) { struct thread *td = curthread; struct sockaddr_in sa; struct socket *so; int error; /* Create the socket */ error = socreate(PF_INET, &so, SOCK_STREAM, IPPROTO_TCP, td->td_ucred, td); if (error != 0) { printf("%s: socreate() error %d\n", __func__, error); return (error); } softc->ha_so = so; ctl_ha_sock_setup(softc); memcpy(&sa, &softc->ha_peer_in, sizeof(sa)); error = soconnect(so, (struct sockaddr *)&sa, td); - if (error != 0 && bootverbose) { - printf("%s: soconnect() error %d\n", __func__, error); + if (error != 0) { + if (bootverbose) + printf("%s: soconnect() error %d\n", __func__, error); goto out; } return (0); out: ctl_ha_close(softc); return (error); } static int ctl_ha_accept(struct ha_softc *softc) { struct socket *so; struct sockaddr *sap; int error; ACCEPT_LOCK(); if (softc->ha_lso->so_rcv.sb_state & SBS_CANTRCVMORE) softc->ha_lso->so_error = ECONNABORTED; if (softc->ha_lso->so_error) { error = softc->ha_lso->so_error; softc->ha_lso->so_error = 0; ACCEPT_UNLOCK(); printf("%s: socket error %d\n", __func__, error); goto out; } so = TAILQ_FIRST(&softc->ha_lso->so_comp); if (so == NULL) { ACCEPT_UNLOCK(); return (EWOULDBLOCK); } KASSERT(!(so->so_qstate & SQ_INCOMP), ("accept1: so SQ_INCOMP")); KASSERT(so->so_qstate & SQ_COMP, ("accept1: so not SQ_COMP")); /* * Before changing the flags on the socket, we have to bump the * reference count. Otherwise, if the protocol calls sofree(), * the socket will be released due to a zero refcount. */ SOCK_LOCK(so); /* soref() and so_state update */ soref(so); /* file descriptor reference */ TAILQ_REMOVE(&softc->ha_lso->so_comp, so, so_list); softc->ha_lso->so_qlen--; so->so_state |= SS_NBIO; so->so_qstate &= ~SQ_COMP; so->so_head = NULL; SOCK_UNLOCK(so); ACCEPT_UNLOCK(); sap = NULL; error = soaccept(so, &sap); if (error != 0) { printf("%s: soaccept() error %d\n", __func__, error); if (sap != NULL) free(sap, M_SONAME); goto out; } if (sap != NULL) free(sap, M_SONAME); softc->ha_so = so; ctl_ha_sock_setup(softc); return (0); out: ctl_ha_lclose(softc); return (error); } static int ctl_ha_listen(struct ha_softc *softc) { struct thread *td = curthread; struct sockaddr_in sa; struct sockopt opt; int error, val; /* Create the socket */ if (softc->ha_lso == NULL) { error = socreate(PF_INET, &softc->ha_lso, SOCK_STREAM, IPPROTO_TCP, td->td_ucred, td); if (error != 0) { printf("%s: socreate() error %d\n", __func__, error); return (error); } bzero(&opt, sizeof(struct sockopt)); opt.sopt_dir = SOPT_SET; opt.sopt_level = SOL_SOCKET; opt.sopt_name = SO_REUSEADDR; opt.sopt_val = &val; opt.sopt_valsize = sizeof(val); val = 1; error = sosetopt(softc->ha_lso, &opt); if (error) { printf("%s: REUSEADDR setting failed %d\n", __func__, error); } bzero(&opt, sizeof(struct sockopt)); opt.sopt_dir = SOPT_SET; opt.sopt_level = SOL_SOCKET; opt.sopt_name = SO_REUSEPORT; opt.sopt_val = &val; opt.sopt_valsize = sizeof(val); val = 1; error = sosetopt(softc->ha_lso, &opt); if (error) { printf("%s: REUSEPORT setting failed %d\n", __func__, error); } SOCKBUF_LOCK(&softc->ha_lso->so_rcv); soupcall_set(softc->ha_lso, SO_RCV, ctl_ha_lupcall, softc); SOCKBUF_UNLOCK(&softc->ha_lso->so_rcv); } memcpy(&sa, &softc->ha_peer_in, sizeof(sa)); error = sobind(softc->ha_lso, (struct sockaddr *)&sa, td); if (error != 0) { printf("%s: sobind() error %d\n", __func__, error); goto out; } error = solisten(softc->ha_lso, 1, td); if (error != 0) { printf("%s: solisten() error %d\n", __func__, error); goto out; } return (0); out: ctl_ha_lclose(softc); return (error); } static void ctl_ha_conn_thread(void *arg) { struct ha_softc *softc = arg; int error; while (1) { if (softc->ha_disconnect || softc->ha_shutdown) { ctl_ha_close(softc); if (softc->ha_disconnect == 2 || softc->ha_shutdown) ctl_ha_lclose(softc); softc->ha_disconnect = 0; if (softc->ha_shutdown) break; } else if (softc->ha_so != NULL && (softc->ha_so->so_error || softc->ha_so->so_rcv.sb_state & SBS_CANTRCVMORE)) ctl_ha_close(softc); if (softc->ha_so == NULL) { if (softc->ha_lso != NULL) ctl_ha_accept(softc); else if (softc->ha_listen) ctl_ha_listen(softc); else if (softc->ha_connect) ctl_ha_connect(softc); } if (softc->ha_so != NULL) { if (softc->ha_connected == 0 && softc->ha_so->so_error == 0 && (softc->ha_so->so_state & SS_ISCONNECTING) == 0) { softc->ha_connected = 1; ctl_ha_evt(softc, CTL_HA_CHAN_MAX, CTL_HA_EVT_LINK_CHANGE, CTL_HA_LINK_ONLINE); softc->ha_receiving = 1; error = kproc_kthread_add(ctl_ha_rx_thread, softc, &softc->ha_ctl_softc->ctl_proc, NULL, 0, 0, "ctl", "ha_rx"); if (error != 0) { printf("Error creating CTL HA rx thread!\n"); softc->ha_receiving = 0; softc->ha_disconnect = 1; } } ctl_ha_send(softc); } mtx_lock(&softc->ha_lock); if (softc->ha_so != NULL && (softc->ha_so->so_error || softc->ha_so->so_rcv.sb_state & SBS_CANTRCVMORE)) ; else if (!softc->ha_wakeup) msleep(&softc->ha_wakeup, &softc->ha_lock, 0, "-", hz); softc->ha_wakeup = 0; mtx_unlock(&softc->ha_lock); } mtx_lock(&softc->ha_lock); softc->ha_shutdown = 2; wakeup(&softc->ha_wakeup); mtx_unlock(&softc->ha_lock); kthread_exit(); } static int ctl_ha_peer_sysctl(SYSCTL_HANDLER_ARGS) { struct ha_softc *softc = (struct ha_softc *)arg1; struct sockaddr_in *sa; int error, b1, b2, b3, b4, p, num; char buf[128]; strlcpy(buf, softc->ha_peer, sizeof(buf)); error = sysctl_handle_string(oidp, buf, sizeof(buf), req); if ((error != 0) || (req->newptr == NULL) || strncmp(buf, softc->ha_peer, sizeof(buf)) == 0) return (error); sa = &softc->ha_peer_in; mtx_lock(&softc->ha_lock); if ((num = sscanf(buf, "connect %d.%d.%d.%d:%d", &b1, &b2, &b3, &b4, &p)) >= 4) { softc->ha_connect = 1; softc->ha_listen = 0; } else if ((num = sscanf(buf, "listen %d.%d.%d.%d:%d", &b1, &b2, &b3, &b4, &p)) >= 4) { softc->ha_connect = 0; softc->ha_listen = 1; } else { softc->ha_connect = 0; softc->ha_listen = 0; if (buf[0] != 0) { buf[0] = 0; error = EINVAL; } } strlcpy(softc->ha_peer, buf, sizeof(softc->ha_peer)); if (softc->ha_connect || softc->ha_listen) { memset(sa, 0, sizeof(*sa)); sa->sin_len = sizeof(struct sockaddr_in); sa->sin_family = AF_INET; sa->sin_port = htons((num >= 5) ? p : 999); sa->sin_addr.s_addr = htonl((b1 << 24) + (b2 << 16) + (b3 << 8) + b4); } softc->ha_disconnect = 2; softc->ha_wakeup = 1; mtx_unlock(&softc->ha_lock); wakeup(&softc->ha_wakeup); return (error); } ctl_ha_status ctl_ha_msg_register(ctl_ha_channel channel, ctl_evt_handler handler) { struct ha_softc *softc = &ha_softc; KASSERT(channel < CTL_HA_CHAN_MAX, ("Wrong CTL HA channel %d", channel)); softc->ha_handler[channel] = handler; return (CTL_HA_STATUS_SUCCESS); } ctl_ha_status ctl_ha_msg_deregister(ctl_ha_channel channel) { struct ha_softc *softc = &ha_softc; KASSERT(channel < CTL_HA_CHAN_MAX, ("Wrong CTL HA channel %d", channel)); softc->ha_handler[channel] = NULL; return (CTL_HA_STATUS_SUCCESS); } /* * Receive a message of the specified size. */ ctl_ha_status ctl_ha_msg_recv(ctl_ha_channel channel, void *addr, size_t len, int wait) { struct ha_softc *softc = &ha_softc; struct uio uio; struct iovec iov; int error, flags; if (!softc->ha_connected) return (CTL_HA_STATUS_DISCONNECT); iov.iov_base = addr; iov.iov_len = len; uio.uio_iov = &iov; uio.uio_iovcnt = 1; uio.uio_rw = UIO_READ; uio.uio_segflg = UIO_SYSSPACE; uio.uio_td = curthread; uio.uio_resid = len; flags = wait ? 0 : MSG_DONTWAIT; error = soreceive(softc->ha_so, NULL, &uio, NULL, NULL, &flags); if (error == 0) return (CTL_HA_STATUS_SUCCESS); /* Consider all errors fatal for HA sanity. */ mtx_lock(&softc->ha_lock); if (softc->ha_connected) { softc->ha_disconnect = 1; softc->ha_wakeup = 1; wakeup(&softc->ha_wakeup); } mtx_unlock(&softc->ha_lock); return (CTL_HA_STATUS_ERROR); } /* * Send a message of the specified size. */ ctl_ha_status ctl_ha_msg_send2(ctl_ha_channel channel, const void *addr, size_t len, const void *addr2, size_t len2, int wait) { struct ha_softc *softc = &ha_softc; struct mbuf *mb, *newmb; struct ha_msg_wire hdr; size_t copylen, off; if (!softc->ha_connected) return (CTL_HA_STATUS_DISCONNECT); newmb = m_getm2(NULL, sizeof(hdr) + len + len2, wait, MT_DATA, M_PKTHDR); if (newmb == NULL) { /* Consider all errors fatal for HA sanity. */ mtx_lock(&softc->ha_lock); if (softc->ha_connected) { softc->ha_disconnect = 1; softc->ha_wakeup = 1; wakeup(&softc->ha_wakeup); } mtx_unlock(&softc->ha_lock); printf("%s: Can't allocate mbuf chain\n", __func__); return (CTL_HA_STATUS_ERROR); } hdr.channel = channel; hdr.length = len + len2; mb = newmb; memcpy(mtodo(mb, 0), &hdr, sizeof(hdr)); mb->m_len += sizeof(hdr); off = 0; for (; mb != NULL && off < len; mb = mb->m_next) { copylen = min(M_TRAILINGSPACE(mb), len - off); memcpy(mtodo(mb, mb->m_len), (const char *)addr + off, copylen); mb->m_len += copylen; off += copylen; if (off == len) break; } KASSERT(off == len, ("%s: off (%zu) != len (%zu)", __func__, off, len)); off = 0; for (; mb != NULL && off < len2; mb = mb->m_next) { copylen = min(M_TRAILINGSPACE(mb), len2 - off); memcpy(mtodo(mb, mb->m_len), (const char *)addr2 + off, copylen); mb->m_len += copylen; off += copylen; } KASSERT(off == len2, ("%s: off (%zu) != len2 (%zu)", __func__, off, len2)); newmb->m_pkthdr.len = sizeof(hdr) + len + len2; mtx_lock(&softc->ha_lock); if (!softc->ha_connected) { mtx_unlock(&softc->ha_lock); m_freem(newmb); return (CTL_HA_STATUS_DISCONNECT); } mbufq_enqueue(&softc->ha_sendq, newmb); softc->ha_wakeup = 1; mtx_unlock(&softc->ha_lock); wakeup(&softc->ha_wakeup); return (CTL_HA_STATUS_SUCCESS); } ctl_ha_status ctl_ha_msg_send(ctl_ha_channel channel, const void *addr, size_t len, int wait) { return (ctl_ha_msg_send2(channel, addr, len, NULL, 0, wait)); } ctl_ha_status ctl_ha_msg_abort(ctl_ha_channel channel) { struct ha_softc *softc = &ha_softc; mtx_lock(&softc->ha_lock); softc->ha_disconnect = 1; softc->ha_wakeup = 1; mtx_unlock(&softc->ha_lock); wakeup(&softc->ha_wakeup); return (CTL_HA_STATUS_SUCCESS); } /* * Allocate a data transfer request structure. */ struct ctl_ha_dt_req * ctl_dt_req_alloc(void) { return (malloc(sizeof(struct ctl_ha_dt_req), M_CTL, M_WAITOK | M_ZERO)); } /* * Free a data transfer request structure. */ void ctl_dt_req_free(struct ctl_ha_dt_req *req) { free(req, M_CTL); } /* * Issue a DMA request for a single buffer. */ ctl_ha_status ctl_dt_single(struct ctl_ha_dt_req *req) { struct ha_softc *softc = &ha_softc; struct ha_dt_msg_wire wire_dt; ctl_ha_status status; wire_dt.command = req->command; wire_dt.size = req->size; wire_dt.local = req->local; wire_dt.remote = req->remote; if (req->command == CTL_HA_DT_CMD_READ && req->callback != NULL) { mtx_lock(&softc->ha_lock); TAILQ_INSERT_TAIL(&softc->ha_dts, req, links); mtx_unlock(&softc->ha_lock); ctl_ha_msg_send(CTL_HA_CHAN_DATA, &wire_dt, sizeof(wire_dt), M_WAITOK); return (CTL_HA_STATUS_WAIT); } if (req->command == CTL_HA_DT_CMD_READ) { status = ctl_ha_msg_send(CTL_HA_CHAN_DATA, &wire_dt, sizeof(wire_dt), M_WAITOK); } else { status = ctl_ha_msg_send2(CTL_HA_CHAN_DATA, &wire_dt, sizeof(wire_dt), req->local, req->size, M_WAITOK); } return (status); } static void ctl_dt_event_handler(ctl_ha_channel channel, ctl_ha_event event, int param) { struct ha_softc *softc = &ha_softc; struct ctl_ha_dt_req *req; ctl_ha_status isc_status; if (event == CTL_HA_EVT_MSG_RECV) { struct ha_dt_msg_wire wire_dt; uint8_t *tmp; int size; size = min(sizeof(wire_dt), param); isc_status = ctl_ha_msg_recv(CTL_HA_CHAN_DATA, &wire_dt, size, M_WAITOK); if (isc_status != CTL_HA_STATUS_SUCCESS) { printf("%s: Error receiving message: %d\n", __func__, isc_status); return; } if (wire_dt.command == CTL_HA_DT_CMD_READ) { wire_dt.command = CTL_HA_DT_CMD_WRITE; tmp = wire_dt.local; wire_dt.local = wire_dt.remote; wire_dt.remote = tmp; ctl_ha_msg_send2(CTL_HA_CHAN_DATA, &wire_dt, sizeof(wire_dt), wire_dt.local, wire_dt.size, M_WAITOK); } else if (wire_dt.command == CTL_HA_DT_CMD_WRITE) { isc_status = ctl_ha_msg_recv(CTL_HA_CHAN_DATA, wire_dt.remote, wire_dt.size, M_WAITOK); mtx_lock(&softc->ha_lock); TAILQ_FOREACH(req, &softc->ha_dts, links) { if (req->local == wire_dt.remote) { TAILQ_REMOVE(&softc->ha_dts, req, links); break; } } mtx_unlock(&softc->ha_lock); if (req) { req->ret = isc_status; req->callback(req); } } } else if (event == CTL_HA_EVT_LINK_CHANGE) { CTL_DEBUG_PRINT(("%s: Link state change to %d\n", __func__, param)); if (param != CTL_HA_LINK_ONLINE) { mtx_lock(&softc->ha_lock); while ((req = TAILQ_FIRST(&softc->ha_dts)) != NULL) { TAILQ_REMOVE(&softc->ha_dts, req, links); mtx_unlock(&softc->ha_lock); req->ret = CTL_HA_STATUS_DISCONNECT; req->callback(req); mtx_lock(&softc->ha_lock); } mtx_unlock(&softc->ha_lock); } } else { printf("%s: Unknown event %d\n", __func__, event); } } ctl_ha_status ctl_ha_msg_init(struct ctl_softc *ctl_softc) { struct ha_softc *softc = &ha_softc; int error; softc->ha_ctl_softc = ctl_softc; mtx_init(&softc->ha_lock, "CTL HA mutex", NULL, MTX_DEF); mbufq_init(&softc->ha_sendq, INT_MAX); TAILQ_INIT(&softc->ha_dts); error = kproc_kthread_add(ctl_ha_conn_thread, softc, &ctl_softc->ctl_proc, NULL, 0, 0, "ctl", "ha_tx"); if (error != 0) { printf("error creating CTL HA connection thread!\n"); mtx_destroy(&softc->ha_lock); return (CTL_HA_STATUS_ERROR); } softc->ha_shutdown_eh = EVENTHANDLER_REGISTER(shutdown_pre_sync, ctl_ha_msg_shutdown, ctl_softc, SHUTDOWN_PRI_FIRST); SYSCTL_ADD_PROC(&ctl_softc->sysctl_ctx, SYSCTL_CHILDREN(ctl_softc->sysctl_tree), OID_AUTO, "ha_peer", CTLTYPE_STRING | CTLFLAG_RWTUN, softc, 0, ctl_ha_peer_sysctl, "A", "HA peer connection method"); if (ctl_ha_msg_register(CTL_HA_CHAN_DATA, ctl_dt_event_handler) != CTL_HA_STATUS_SUCCESS) { printf("%s: ctl_ha_msg_register failed.\n", __func__); } return (CTL_HA_STATUS_SUCCESS); }; void ctl_ha_msg_shutdown(struct ctl_softc *ctl_softc) { struct ha_softc *softc = &ha_softc; /* Disconnect and shutdown threads. */ mtx_lock(&softc->ha_lock); if (softc->ha_shutdown < 2) { softc->ha_shutdown = 1; softc->ha_wakeup = 1; wakeup(&softc->ha_wakeup); while (softc->ha_shutdown < 2) { msleep(&softc->ha_wakeup, &softc->ha_lock, 0, "shutdown", hz); } } mtx_unlock(&softc->ha_lock); }; ctl_ha_status ctl_ha_msg_destroy(struct ctl_softc *ctl_softc) { struct ha_softc *softc = &ha_softc; if (softc->ha_shutdown_eh != NULL) { EVENTHANDLER_DEREGISTER(shutdown_pre_sync, softc->ha_shutdown_eh); softc->ha_shutdown_eh = NULL; } ctl_ha_msg_shutdown(ctl_softc); /* Just in case. */ if (ctl_ha_msg_deregister(CTL_HA_CHAN_DATA) != CTL_HA_STATUS_SUCCESS) printf("%s: ctl_ha_msg_deregister failed.\n", __func__); mtx_destroy(&softc->ha_lock); return (CTL_HA_STATUS_SUCCESS); }; Index: projects/clang400-import/sys/compat/freebsd32/freebsd32_misc.c =================================================================== --- projects/clang400-import/sys/compat/freebsd32/freebsd32_misc.c (revision 311459) +++ projects/clang400-import/sys/compat/freebsd32/freebsd32_misc.c (revision 311460) @@ -1,3162 +1,3172 @@ /*- * Copyright (c) 2002 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_inet.h" #include "opt_inet6.h" #define __ELF_WORD_SIZE 32 #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include /* Must come after sys/malloc.h */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include /* Must come after sys/selinfo.h */ #include /* Must come after sys/selinfo.h */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef INET #include #endif #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include FEATURE(compat_freebsd_32bit, "Compatible with 32-bit FreeBSD"); #ifndef __mips__ CTASSERT(sizeof(struct timeval32) == 8); CTASSERT(sizeof(struct timespec32) == 8); CTASSERT(sizeof(struct itimerval32) == 16); #endif CTASSERT(sizeof(struct statfs32) == 256); #ifndef __mips__ CTASSERT(sizeof(struct rusage32) == 72); #endif CTASSERT(sizeof(struct sigaltstack32) == 12); CTASSERT(sizeof(struct kevent32) == 20); CTASSERT(sizeof(struct iovec32) == 8); CTASSERT(sizeof(struct msghdr32) == 28); #ifndef __mips__ CTASSERT(sizeof(struct stat32) == 96); #endif CTASSERT(sizeof(struct sigaction32) == 24); static int freebsd32_kevent_copyout(void *arg, struct kevent *kevp, int count); static int freebsd32_kevent_copyin(void *arg, struct kevent *kevp, int count); void freebsd32_rusage_out(const struct rusage *s, struct rusage32 *s32) { TV_CP(*s, *s32, ru_utime); TV_CP(*s, *s32, ru_stime); CP(*s, *s32, ru_maxrss); CP(*s, *s32, ru_ixrss); CP(*s, *s32, ru_idrss); CP(*s, *s32, ru_isrss); CP(*s, *s32, ru_minflt); CP(*s, *s32, ru_majflt); CP(*s, *s32, ru_nswap); CP(*s, *s32, ru_inblock); CP(*s, *s32, ru_oublock); CP(*s, *s32, ru_msgsnd); CP(*s, *s32, ru_msgrcv); CP(*s, *s32, ru_nsignals); CP(*s, *s32, ru_nvcsw); CP(*s, *s32, ru_nivcsw); } int freebsd32_wait4(struct thread *td, struct freebsd32_wait4_args *uap) { int error, status; struct rusage32 ru32; struct rusage ru, *rup; if (uap->rusage != NULL) rup = &ru; else rup = NULL; error = kern_wait(td, uap->pid, &status, uap->options, rup); if (error) return (error); if (uap->status != NULL) error = copyout(&status, uap->status, sizeof(status)); if (uap->rusage != NULL && error == 0) { freebsd32_rusage_out(&ru, &ru32); error = copyout(&ru32, uap->rusage, sizeof(ru32)); } return (error); } int freebsd32_wait6(struct thread *td, struct freebsd32_wait6_args *uap) { struct wrusage32 wru32; struct __wrusage wru, *wrup; struct siginfo32 si32; struct __siginfo si, *sip; int error, status; if (uap->wrusage != NULL) wrup = &wru; else wrup = NULL; if (uap->info != NULL) { sip = &si; bzero(sip, sizeof(*sip)); } else sip = NULL; error = kern_wait6(td, uap->idtype, PAIR32TO64(id_t, uap->id), &status, uap->options, wrup, sip); if (error != 0) return (error); if (uap->status != NULL) error = copyout(&status, uap->status, sizeof(status)); if (uap->wrusage != NULL && error == 0) { freebsd32_rusage_out(&wru.wru_self, &wru32.wru_self); freebsd32_rusage_out(&wru.wru_children, &wru32.wru_children); error = copyout(&wru32, uap->wrusage, sizeof(wru32)); } if (uap->info != NULL && error == 0) { siginfo_to_siginfo32 (&si, &si32); error = copyout(&si32, uap->info, sizeof(si32)); } return (error); } #ifdef COMPAT_FREEBSD4 static void copy_statfs(struct statfs *in, struct statfs32 *out) { statfs_scale_blocks(in, INT32_MAX); bzero(out, sizeof(*out)); CP(*in, *out, f_bsize); out->f_iosize = MIN(in->f_iosize, INT32_MAX); CP(*in, *out, f_blocks); CP(*in, *out, f_bfree); CP(*in, *out, f_bavail); out->f_files = MIN(in->f_files, INT32_MAX); out->f_ffree = MIN(in->f_ffree, INT32_MAX); CP(*in, *out, f_fsid); CP(*in, *out, f_owner); CP(*in, *out, f_type); CP(*in, *out, f_flags); out->f_syncwrites = MIN(in->f_syncwrites, INT32_MAX); out->f_asyncwrites = MIN(in->f_asyncwrites, INT32_MAX); strlcpy(out->f_fstypename, in->f_fstypename, MFSNAMELEN); strlcpy(out->f_mntonname, in->f_mntonname, min(MNAMELEN, FREEBSD4_MNAMELEN)); out->f_syncreads = MIN(in->f_syncreads, INT32_MAX); out->f_asyncreads = MIN(in->f_asyncreads, INT32_MAX); strlcpy(out->f_mntfromname, in->f_mntfromname, min(MNAMELEN, FREEBSD4_MNAMELEN)); } #endif #ifdef COMPAT_FREEBSD4 int -freebsd4_freebsd32_getfsstat(struct thread *td, struct freebsd4_freebsd32_getfsstat_args *uap) +freebsd4_freebsd32_getfsstat(struct thread *td, + struct freebsd4_freebsd32_getfsstat_args *uap) { struct statfs *buf, *sp; struct statfs32 stat32; size_t count, size, copycount; int error; count = uap->bufsize / sizeof(struct statfs32); size = count * sizeof(struct statfs); error = kern_getfsstat(td, &buf, size, &count, UIO_SYSSPACE, uap->mode); if (size > 0) { sp = buf; copycount = count; while (copycount > 0 && error == 0) { copy_statfs(sp, &stat32); error = copyout(&stat32, uap->buf, sizeof(stat32)); sp++; uap->buf++; copycount--; } - free(buf, M_TEMP); + free(buf, M_STATFS); } if (error == 0) td->td_retval[0] = count; return (error); } #endif #ifdef COMPAT_FREEBSD10 int freebsd10_freebsd32_pipe(struct thread *td, struct freebsd10_freebsd32_pipe_args *uap) { return (freebsd10_pipe(td, (struct freebsd10_pipe_args*)uap)); } #endif int freebsd32_sigaltstack(struct thread *td, struct freebsd32_sigaltstack_args *uap) { struct sigaltstack32 s32; struct sigaltstack ss, oss, *ssp; int error; if (uap->ss != NULL) { error = copyin(uap->ss, &s32, sizeof(s32)); if (error) return (error); PTRIN_CP(s32, ss, ss_sp); CP(s32, ss, ss_size); CP(s32, ss, ss_flags); ssp = &ss; } else ssp = NULL; error = kern_sigaltstack(td, ssp, &oss); if (error == 0 && uap->oss != NULL) { PTROUT_CP(oss, s32, ss_sp); CP(oss, s32, ss_size); CP(oss, s32, ss_flags); error = copyout(&s32, uap->oss, sizeof(s32)); } return (error); } /* * Custom version of exec_copyin_args() so that we can translate * the pointers. */ int freebsd32_exec_copyin_args(struct image_args *args, char *fname, enum uio_seg segflg, u_int32_t *argv, u_int32_t *envv) { char *argp, *envp; u_int32_t *p32, arg; size_t length; int error; bzero(args, sizeof(*args)); if (argv == NULL) return (EFAULT); /* * Allocate demand-paged memory for the file name, argument, and * environment strings. */ error = exec_alloc_args(args); if (error != 0) return (error); /* * Copy the file name. */ if (fname != NULL) { args->fname = args->buf; error = (segflg == UIO_SYSSPACE) ? copystr(fname, args->fname, PATH_MAX, &length) : copyinstr(fname, args->fname, PATH_MAX, &length); if (error != 0) goto err_exit; } else length = 0; args->begin_argv = args->buf + length; args->endp = args->begin_argv; args->stringspace = ARG_MAX; /* * extract arguments first */ p32 = argv; for (;;) { error = copyin(p32++, &arg, sizeof(arg)); if (error) goto err_exit; if (arg == 0) break; argp = PTRIN(arg); error = copyinstr(argp, args->endp, args->stringspace, &length); if (error) { if (error == ENAMETOOLONG) error = E2BIG; goto err_exit; } args->stringspace -= length; args->endp += length; args->argc++; } args->begin_envv = args->endp; /* * extract environment strings */ if (envv) { p32 = envv; for (;;) { error = copyin(p32++, &arg, sizeof(arg)); if (error) goto err_exit; if (arg == 0) break; envp = PTRIN(arg); error = copyinstr(envp, args->endp, args->stringspace, &length); if (error) { if (error == ENAMETOOLONG) error = E2BIG; goto err_exit; } args->stringspace -= length; args->endp += length; args->envc++; } } return (0); err_exit: exec_free_args(args); return (error); } int freebsd32_execve(struct thread *td, struct freebsd32_execve_args *uap) { struct image_args eargs; struct vmspace *oldvmspace; int error; error = pre_execve(td, &oldvmspace); if (error != 0) return (error); error = freebsd32_exec_copyin_args(&eargs, uap->fname, UIO_USERSPACE, uap->argv, uap->envv); if (error == 0) error = kern_execve(td, &eargs, NULL); post_execve(td, error, oldvmspace); return (error); } int freebsd32_fexecve(struct thread *td, struct freebsd32_fexecve_args *uap) { struct image_args eargs; struct vmspace *oldvmspace; int error; error = pre_execve(td, &oldvmspace); if (error != 0) return (error); error = freebsd32_exec_copyin_args(&eargs, NULL, UIO_SYSSPACE, uap->argv, uap->envv); if (error == 0) { eargs.fd = uap->fd; error = kern_execve(td, &eargs, NULL); } post_execve(td, error, oldvmspace); return (error); } int freebsd32_mprotect(struct thread *td, struct freebsd32_mprotect_args *uap) { struct mprotect_args ap; ap.addr = PTRIN(uap->addr); ap.len = uap->len; ap.prot = uap->prot; #if defined(__amd64__) if (i386_read_exec && (ap.prot & PROT_READ) != 0) ap.prot |= PROT_EXEC; #endif return (sys_mprotect(td, &ap)); } int freebsd32_mmap(struct thread *td, struct freebsd32_mmap_args *uap) { struct mmap_args ap; vm_offset_t addr = (vm_offset_t) uap->addr; vm_size_t len = uap->len; int prot = uap->prot; int flags = uap->flags; int fd = uap->fd; off_t pos = PAIR32TO64(off_t,uap->pos); #if defined(__amd64__) if (i386_read_exec && (prot & PROT_READ)) prot |= PROT_EXEC; #endif ap.addr = (void *) addr; ap.len = len; ap.prot = prot; ap.flags = flags; ap.fd = fd; ap.pos = pos; return (sys_mmap(td, &ap)); } #ifdef COMPAT_FREEBSD6 int freebsd6_freebsd32_mmap(struct thread *td, struct freebsd6_freebsd32_mmap_args *uap) { struct freebsd32_mmap_args ap; ap.addr = uap->addr; ap.len = uap->len; ap.prot = uap->prot; ap.flags = uap->flags; ap.fd = uap->fd; ap.pos1 = uap->pos1; ap.pos2 = uap->pos2; return (freebsd32_mmap(td, &ap)); } #endif int freebsd32_setitimer(struct thread *td, struct freebsd32_setitimer_args *uap) { struct itimerval itv, oitv, *itvp; struct itimerval32 i32; int error; if (uap->itv != NULL) { error = copyin(uap->itv, &i32, sizeof(i32)); if (error) return (error); TV_CP(i32, itv, it_interval); TV_CP(i32, itv, it_value); itvp = &itv; } else itvp = NULL; error = kern_setitimer(td, uap->which, itvp, &oitv); if (error || uap->oitv == NULL) return (error); TV_CP(oitv, i32, it_interval); TV_CP(oitv, i32, it_value); return (copyout(&i32, uap->oitv, sizeof(i32))); } int freebsd32_getitimer(struct thread *td, struct freebsd32_getitimer_args *uap) { struct itimerval itv; struct itimerval32 i32; int error; error = kern_getitimer(td, uap->which, &itv); if (error || uap->itv == NULL) return (error); TV_CP(itv, i32, it_interval); TV_CP(itv, i32, it_value); return (copyout(&i32, uap->itv, sizeof(i32))); } int freebsd32_select(struct thread *td, struct freebsd32_select_args *uap) { struct timeval32 tv32; struct timeval tv, *tvp; int error; if (uap->tv != NULL) { error = copyin(uap->tv, &tv32, sizeof(tv32)); if (error) return (error); CP(tv32, tv, tv_sec); CP(tv32, tv, tv_usec); tvp = &tv; } else tvp = NULL; /* * XXX Do pointers need PTRIN()? */ return (kern_select(td, uap->nd, uap->in, uap->ou, uap->ex, tvp, sizeof(int32_t) * 8)); } int freebsd32_pselect(struct thread *td, struct freebsd32_pselect_args *uap) { struct timespec32 ts32; struct timespec ts; struct timeval tv, *tvp; sigset_t set, *uset; int error; if (uap->ts != NULL) { error = copyin(uap->ts, &ts32, sizeof(ts32)); if (error != 0) return (error); CP(ts32, ts, tv_sec); CP(ts32, ts, tv_nsec); TIMESPEC_TO_TIMEVAL(&tv, &ts); tvp = &tv; } else tvp = NULL; if (uap->sm != NULL) { error = copyin(uap->sm, &set, sizeof(set)); if (error != 0) return (error); uset = &set; } else uset = NULL; /* * XXX Do pointers need PTRIN()? */ error = kern_pselect(td, uap->nd, uap->in, uap->ou, uap->ex, tvp, uset, sizeof(int32_t) * 8); return (error); } /* * Copy 'count' items into the destination list pointed to by uap->eventlist. */ static int freebsd32_kevent_copyout(void *arg, struct kevent *kevp, int count) { struct freebsd32_kevent_args *uap; struct kevent32 ks32[KQ_NEVENTS]; int i, error = 0; KASSERT(count <= KQ_NEVENTS, ("count (%d) > KQ_NEVENTS", count)); uap = (struct freebsd32_kevent_args *)arg; for (i = 0; i < count; i++) { CP(kevp[i], ks32[i], ident); CP(kevp[i], ks32[i], filter); CP(kevp[i], ks32[i], flags); CP(kevp[i], ks32[i], fflags); CP(kevp[i], ks32[i], data); PTROUT_CP(kevp[i], ks32[i], udata); } error = copyout(ks32, uap->eventlist, count * sizeof *ks32); if (error == 0) uap->eventlist += count; return (error); } /* * Copy 'count' items from the list pointed to by uap->changelist. */ static int freebsd32_kevent_copyin(void *arg, struct kevent *kevp, int count) { struct freebsd32_kevent_args *uap; struct kevent32 ks32[KQ_NEVENTS]; int i, error = 0; KASSERT(count <= KQ_NEVENTS, ("count (%d) > KQ_NEVENTS", count)); uap = (struct freebsd32_kevent_args *)arg; error = copyin(uap->changelist, ks32, count * sizeof *ks32); if (error) goto done; uap->changelist += count; for (i = 0; i < count; i++) { CP(ks32[i], kevp[i], ident); CP(ks32[i], kevp[i], filter); CP(ks32[i], kevp[i], flags); CP(ks32[i], kevp[i], fflags); CP(ks32[i], kevp[i], data); PTRIN_CP(ks32[i], kevp[i], udata); } done: return (error); } int freebsd32_kevent(struct thread *td, struct freebsd32_kevent_args *uap) { struct timespec32 ts32; struct timespec ts, *tsp; struct kevent_copyops k_ops = { uap, freebsd32_kevent_copyout, freebsd32_kevent_copyin}; int error; if (uap->timeout) { error = copyin(uap->timeout, &ts32, sizeof(ts32)); if (error) return (error); CP(ts32, ts, tv_sec); CP(ts32, ts, tv_nsec); tsp = &ts; } else tsp = NULL; error = kern_kevent(td, uap->fd, uap->nchanges, uap->nevents, &k_ops, tsp); return (error); } int freebsd32_gettimeofday(struct thread *td, struct freebsd32_gettimeofday_args *uap) { struct timeval atv; struct timeval32 atv32; struct timezone rtz; int error = 0; if (uap->tp) { microtime(&atv); CP(atv, atv32, tv_sec); CP(atv, atv32, tv_usec); error = copyout(&atv32, uap->tp, sizeof (atv32)); } if (error == 0 && uap->tzp != NULL) { rtz.tz_minuteswest = tz_minuteswest; rtz.tz_dsttime = tz_dsttime; error = copyout(&rtz, uap->tzp, sizeof (rtz)); } return (error); } int freebsd32_getrusage(struct thread *td, struct freebsd32_getrusage_args *uap) { struct rusage32 s32; struct rusage s; int error; error = kern_getrusage(td, uap->who, &s); if (error) return (error); if (uap->rusage != NULL) { freebsd32_rusage_out(&s, &s32); error = copyout(&s32, uap->rusage, sizeof(s32)); } return (error); } static int freebsd32_copyinuio(struct iovec32 *iovp, u_int iovcnt, struct uio **uiop) { struct iovec32 iov32; struct iovec *iov; struct uio *uio; u_int iovlen; int error, i; *uiop = NULL; if (iovcnt > UIO_MAXIOV) return (EINVAL); iovlen = iovcnt * sizeof(struct iovec); uio = malloc(iovlen + sizeof *uio, M_IOV, M_WAITOK); iov = (struct iovec *)(uio + 1); for (i = 0; i < iovcnt; i++) { error = copyin(&iovp[i], &iov32, sizeof(struct iovec32)); if (error) { free(uio, M_IOV); return (error); } iov[i].iov_base = PTRIN(iov32.iov_base); iov[i].iov_len = iov32.iov_len; } uio->uio_iov = iov; uio->uio_iovcnt = iovcnt; uio->uio_segflg = UIO_USERSPACE; uio->uio_offset = -1; uio->uio_resid = 0; for (i = 0; i < iovcnt; i++) { if (iov->iov_len > INT_MAX - uio->uio_resid) { free(uio, M_IOV); return (EINVAL); } uio->uio_resid += iov->iov_len; iov++; } *uiop = uio; return (0); } int freebsd32_readv(struct thread *td, struct freebsd32_readv_args *uap) { struct uio *auio; int error; error = freebsd32_copyinuio(uap->iovp, uap->iovcnt, &auio); if (error) return (error); error = kern_readv(td, uap->fd, auio); free(auio, M_IOV); return (error); } int freebsd32_writev(struct thread *td, struct freebsd32_writev_args *uap) { struct uio *auio; int error; error = freebsd32_copyinuio(uap->iovp, uap->iovcnt, &auio); if (error) return (error); error = kern_writev(td, uap->fd, auio); free(auio, M_IOV); return (error); } int freebsd32_preadv(struct thread *td, struct freebsd32_preadv_args *uap) { struct uio *auio; int error; error = freebsd32_copyinuio(uap->iovp, uap->iovcnt, &auio); if (error) return (error); error = kern_preadv(td, uap->fd, auio, PAIR32TO64(off_t,uap->offset)); free(auio, M_IOV); return (error); } int freebsd32_pwritev(struct thread *td, struct freebsd32_pwritev_args *uap) { struct uio *auio; int error; error = freebsd32_copyinuio(uap->iovp, uap->iovcnt, &auio); if (error) return (error); error = kern_pwritev(td, uap->fd, auio, PAIR32TO64(off_t,uap->offset)); free(auio, M_IOV); return (error); } int freebsd32_copyiniov(struct iovec32 *iovp32, u_int iovcnt, struct iovec **iovp, int error) { struct iovec32 iov32; struct iovec *iov; u_int iovlen; int i; *iovp = NULL; if (iovcnt > UIO_MAXIOV) return (error); iovlen = iovcnt * sizeof(struct iovec); iov = malloc(iovlen, M_IOV, M_WAITOK); for (i = 0; i < iovcnt; i++) { error = copyin(&iovp32[i], &iov32, sizeof(struct iovec32)); if (error) { free(iov, M_IOV); return (error); } iov[i].iov_base = PTRIN(iov32.iov_base); iov[i].iov_len = iov32.iov_len; } *iovp = iov; return (0); } static int freebsd32_copyinmsghdr(struct msghdr32 *msg32, struct msghdr *msg) { struct msghdr32 m32; int error; error = copyin(msg32, &m32, sizeof(m32)); if (error) return (error); msg->msg_name = PTRIN(m32.msg_name); msg->msg_namelen = m32.msg_namelen; msg->msg_iov = PTRIN(m32.msg_iov); msg->msg_iovlen = m32.msg_iovlen; msg->msg_control = PTRIN(m32.msg_control); msg->msg_controllen = m32.msg_controllen; msg->msg_flags = m32.msg_flags; return (0); } static int freebsd32_copyoutmsghdr(struct msghdr *msg, struct msghdr32 *msg32) { struct msghdr32 m32; int error; m32.msg_name = PTROUT(msg->msg_name); m32.msg_namelen = msg->msg_namelen; m32.msg_iov = PTROUT(msg->msg_iov); m32.msg_iovlen = msg->msg_iovlen; m32.msg_control = PTROUT(msg->msg_control); m32.msg_controllen = msg->msg_controllen; m32.msg_flags = msg->msg_flags; error = copyout(&m32, msg32, sizeof(m32)); return (error); } #ifndef __mips__ #define FREEBSD32_ALIGNBYTES (sizeof(int) - 1) #else #define FREEBSD32_ALIGNBYTES (sizeof(long) - 1) #endif #define FREEBSD32_ALIGN(p) \ (((u_long)(p) + FREEBSD32_ALIGNBYTES) & ~FREEBSD32_ALIGNBYTES) #define FREEBSD32_CMSG_SPACE(l) \ (FREEBSD32_ALIGN(sizeof(struct cmsghdr)) + FREEBSD32_ALIGN(l)) #define FREEBSD32_CMSG_DATA(cmsg) ((unsigned char *)(cmsg) + \ FREEBSD32_ALIGN(sizeof(struct cmsghdr))) static int freebsd32_copy_msg_out(struct msghdr *msg, struct mbuf *control) { struct cmsghdr *cm; void *data; socklen_t clen, datalen; int error; caddr_t ctlbuf; int len, maxlen, copylen; struct mbuf *m; error = 0; len = msg->msg_controllen; maxlen = msg->msg_controllen; msg->msg_controllen = 0; m = control; ctlbuf = msg->msg_control; while (m && len > 0) { cm = mtod(m, struct cmsghdr *); clen = m->m_len; while (cm != NULL) { if (sizeof(struct cmsghdr) > clen || cm->cmsg_len > clen) { error = EINVAL; break; } data = CMSG_DATA(cm); datalen = (caddr_t)cm + cm->cmsg_len - (caddr_t)data; /* Adjust message length */ cm->cmsg_len = FREEBSD32_ALIGN(sizeof(struct cmsghdr)) + datalen; /* Copy cmsghdr */ copylen = sizeof(struct cmsghdr); if (len < copylen) { msg->msg_flags |= MSG_CTRUNC; copylen = len; } error = copyout(cm,ctlbuf,copylen); if (error) goto exit; ctlbuf += FREEBSD32_ALIGN(copylen); len -= FREEBSD32_ALIGN(copylen); if (len <= 0) break; /* Copy data */ copylen = datalen; if (len < copylen) { msg->msg_flags |= MSG_CTRUNC; copylen = len; } error = copyout(data,ctlbuf,copylen); if (error) goto exit; ctlbuf += FREEBSD32_ALIGN(copylen); len -= FREEBSD32_ALIGN(copylen); if (CMSG_SPACE(datalen) < clen) { clen -= CMSG_SPACE(datalen); cm = (struct cmsghdr *) ((caddr_t)cm + CMSG_SPACE(datalen)); } else { clen = 0; cm = NULL; } } m = m->m_next; } msg->msg_controllen = (len <= 0) ? maxlen : ctlbuf - (caddr_t)msg->msg_control; exit: return (error); } int freebsd32_recvmsg(td, uap) struct thread *td; struct freebsd32_recvmsg_args /* { int s; struct msghdr32 *msg; int flags; } */ *uap; { struct msghdr msg; struct msghdr32 m32; struct iovec *uiov, *iov; struct mbuf *control = NULL; struct mbuf **controlp; int error; error = copyin(uap->msg, &m32, sizeof(m32)); if (error) return (error); error = freebsd32_copyinmsghdr(uap->msg, &msg); if (error) return (error); error = freebsd32_copyiniov(PTRIN(m32.msg_iov), m32.msg_iovlen, &iov, EMSGSIZE); if (error) return (error); msg.msg_flags = uap->flags; uiov = msg.msg_iov; msg.msg_iov = iov; controlp = (msg.msg_control != NULL) ? &control : NULL; error = kern_recvit(td, uap->s, &msg, UIO_USERSPACE, controlp); if (error == 0) { msg.msg_iov = uiov; if (control != NULL) error = freebsd32_copy_msg_out(&msg, control); else msg.msg_controllen = 0; if (error == 0) error = freebsd32_copyoutmsghdr(&msg, uap->msg); } free(iov, M_IOV); if (control != NULL) m_freem(control); return (error); } /* * Copy-in the array of control messages constructed using alignment * and padding suitable for a 32-bit environment and construct an * mbuf using alignment and padding suitable for a 64-bit kernel. * The alignment and padding are defined indirectly by CMSG_DATA(), * CMSG_SPACE() and CMSG_LEN(). */ static int freebsd32_copyin_control(struct mbuf **mp, caddr_t buf, u_int buflen) { struct mbuf *m; void *md; u_int idx, len, msglen; int error; buflen = FREEBSD32_ALIGN(buflen); if (buflen > MCLBYTES) return (EINVAL); /* * Iterate over the buffer and get the length of each message * in there. This has 32-bit alignment and padding. Use it to * determine the length of these messages when using 64-bit * alignment and padding. */ idx = 0; len = 0; while (idx < buflen) { error = copyin(buf + idx, &msglen, sizeof(msglen)); if (error) return (error); if (msglen < sizeof(struct cmsghdr)) return (EINVAL); msglen = FREEBSD32_ALIGN(msglen); if (idx + msglen > buflen) return (EINVAL); idx += msglen; msglen += CMSG_ALIGN(sizeof(struct cmsghdr)) - FREEBSD32_ALIGN(sizeof(struct cmsghdr)); len += CMSG_ALIGN(msglen); } if (len > MCLBYTES) return (EINVAL); m = m_get(M_WAITOK, MT_CONTROL); if (len > MLEN) MCLGET(m, M_WAITOK); m->m_len = len; md = mtod(m, void *); while (buflen > 0) { error = copyin(buf, md, sizeof(struct cmsghdr)); if (error) break; msglen = *(u_int *)md; msglen = FREEBSD32_ALIGN(msglen); /* Modify the message length to account for alignment. */ *(u_int *)md = msglen + CMSG_ALIGN(sizeof(struct cmsghdr)) - FREEBSD32_ALIGN(sizeof(struct cmsghdr)); md = (char *)md + CMSG_ALIGN(sizeof(struct cmsghdr)); buf += FREEBSD32_ALIGN(sizeof(struct cmsghdr)); buflen -= FREEBSD32_ALIGN(sizeof(struct cmsghdr)); msglen -= FREEBSD32_ALIGN(sizeof(struct cmsghdr)); if (msglen > 0) { error = copyin(buf, md, msglen); if (error) break; md = (char *)md + CMSG_ALIGN(msglen); buf += msglen; buflen -= msglen; } } if (error) m_free(m); else *mp = m; return (error); } int freebsd32_sendmsg(struct thread *td, struct freebsd32_sendmsg_args *uap) { struct msghdr msg; struct msghdr32 m32; struct iovec *iov; struct mbuf *control = NULL; struct sockaddr *to = NULL; int error; error = copyin(uap->msg, &m32, sizeof(m32)); if (error) return (error); error = freebsd32_copyinmsghdr(uap->msg, &msg); if (error) return (error); error = freebsd32_copyiniov(PTRIN(m32.msg_iov), m32.msg_iovlen, &iov, EMSGSIZE); if (error) return (error); msg.msg_iov = iov; if (msg.msg_name != NULL) { error = getsockaddr(&to, msg.msg_name, msg.msg_namelen); if (error) { to = NULL; goto out; } msg.msg_name = to; } if (msg.msg_control) { if (msg.msg_controllen < sizeof(struct cmsghdr)) { error = EINVAL; goto out; } error = freebsd32_copyin_control(&control, msg.msg_control, msg.msg_controllen); if (error) goto out; msg.msg_control = NULL; msg.msg_controllen = 0; } error = kern_sendit(td, uap->s, &msg, uap->flags, control, UIO_USERSPACE); out: free(iov, M_IOV); if (to) free(to, M_SONAME); return (error); } int freebsd32_recvfrom(struct thread *td, struct freebsd32_recvfrom_args *uap) { struct msghdr msg; struct iovec aiov; int error; if (uap->fromlenaddr) { error = copyin(PTRIN(uap->fromlenaddr), &msg.msg_namelen, sizeof(msg.msg_namelen)); if (error) return (error); } else { msg.msg_namelen = 0; } msg.msg_name = PTRIN(uap->from); msg.msg_iov = &aiov; msg.msg_iovlen = 1; aiov.iov_base = PTRIN(uap->buf); aiov.iov_len = uap->len; msg.msg_control = NULL; msg.msg_flags = uap->flags; error = kern_recvit(td, uap->s, &msg, UIO_USERSPACE, NULL); if (error == 0 && uap->fromlenaddr) error = copyout(&msg.msg_namelen, PTRIN(uap->fromlenaddr), sizeof (msg.msg_namelen)); return (error); } int freebsd32_settimeofday(struct thread *td, struct freebsd32_settimeofday_args *uap) { struct timeval32 tv32; struct timeval tv, *tvp; struct timezone tz, *tzp; int error; if (uap->tv) { error = copyin(uap->tv, &tv32, sizeof(tv32)); if (error) return (error); CP(tv32, tv, tv_sec); CP(tv32, tv, tv_usec); tvp = &tv; } else tvp = NULL; if (uap->tzp) { error = copyin(uap->tzp, &tz, sizeof(tz)); if (error) return (error); tzp = &tz; } else tzp = NULL; return (kern_settimeofday(td, tvp, tzp)); } int freebsd32_utimes(struct thread *td, struct freebsd32_utimes_args *uap) { struct timeval32 s32[2]; struct timeval s[2], *sp; int error; if (uap->tptr != NULL) { error = copyin(uap->tptr, s32, sizeof(s32)); if (error) return (error); CP(s32[0], s[0], tv_sec); CP(s32[0], s[0], tv_usec); CP(s32[1], s[1], tv_sec); CP(s32[1], s[1], tv_usec); sp = s; } else sp = NULL; return (kern_utimesat(td, AT_FDCWD, uap->path, UIO_USERSPACE, sp, UIO_SYSSPACE)); } int freebsd32_lutimes(struct thread *td, struct freebsd32_lutimes_args *uap) { struct timeval32 s32[2]; struct timeval s[2], *sp; int error; if (uap->tptr != NULL) { error = copyin(uap->tptr, s32, sizeof(s32)); if (error) return (error); CP(s32[0], s[0], tv_sec); CP(s32[0], s[0], tv_usec); CP(s32[1], s[1], tv_sec); CP(s32[1], s[1], tv_usec); sp = s; } else sp = NULL; return (kern_lutimes(td, uap->path, UIO_USERSPACE, sp, UIO_SYSSPACE)); } int freebsd32_futimes(struct thread *td, struct freebsd32_futimes_args *uap) { struct timeval32 s32[2]; struct timeval s[2], *sp; int error; if (uap->tptr != NULL) { error = copyin(uap->tptr, s32, sizeof(s32)); if (error) return (error); CP(s32[0], s[0], tv_sec); CP(s32[0], s[0], tv_usec); CP(s32[1], s[1], tv_sec); CP(s32[1], s[1], tv_usec); sp = s; } else sp = NULL; return (kern_futimes(td, uap->fd, sp, UIO_SYSSPACE)); } int freebsd32_futimesat(struct thread *td, struct freebsd32_futimesat_args *uap) { struct timeval32 s32[2]; struct timeval s[2], *sp; int error; if (uap->times != NULL) { error = copyin(uap->times, s32, sizeof(s32)); if (error) return (error); CP(s32[0], s[0], tv_sec); CP(s32[0], s[0], tv_usec); CP(s32[1], s[1], tv_sec); CP(s32[1], s[1], tv_usec); sp = s; } else sp = NULL; return (kern_utimesat(td, uap->fd, uap->path, UIO_USERSPACE, sp, UIO_SYSSPACE)); } int freebsd32_futimens(struct thread *td, struct freebsd32_futimens_args *uap) { struct timespec32 ts32[2]; struct timespec ts[2], *tsp; int error; if (uap->times != NULL) { error = copyin(uap->times, ts32, sizeof(ts32)); if (error) return (error); CP(ts32[0], ts[0], tv_sec); CP(ts32[0], ts[0], tv_nsec); CP(ts32[1], ts[1], tv_sec); CP(ts32[1], ts[1], tv_nsec); tsp = ts; } else tsp = NULL; return (kern_futimens(td, uap->fd, tsp, UIO_SYSSPACE)); } int freebsd32_utimensat(struct thread *td, struct freebsd32_utimensat_args *uap) { struct timespec32 ts32[2]; struct timespec ts[2], *tsp; int error; if (uap->times != NULL) { error = copyin(uap->times, ts32, sizeof(ts32)); if (error) return (error); CP(ts32[0], ts[0], tv_sec); CP(ts32[0], ts[0], tv_nsec); CP(ts32[1], ts[1], tv_sec); CP(ts32[1], ts[1], tv_nsec); tsp = ts; } else tsp = NULL; return (kern_utimensat(td, uap->fd, uap->path, UIO_USERSPACE, tsp, UIO_SYSSPACE, uap->flag)); } int freebsd32_adjtime(struct thread *td, struct freebsd32_adjtime_args *uap) { struct timeval32 tv32; struct timeval delta, olddelta, *deltap; int error; if (uap->delta) { error = copyin(uap->delta, &tv32, sizeof(tv32)); if (error) return (error); CP(tv32, delta, tv_sec); CP(tv32, delta, tv_usec); deltap = δ } else deltap = NULL; error = kern_adjtime(td, deltap, &olddelta); if (uap->olddelta && error == 0) { CP(olddelta, tv32, tv_sec); CP(olddelta, tv32, tv_usec); error = copyout(&tv32, uap->olddelta, sizeof(tv32)); } return (error); } #ifdef COMPAT_FREEBSD4 int freebsd4_freebsd32_statfs(struct thread *td, struct freebsd4_freebsd32_statfs_args *uap) { struct statfs32 s32; - struct statfs s; + struct statfs *sp; int error; - error = kern_statfs(td, uap->path, UIO_USERSPACE, &s); - if (error) - return (error); - copy_statfs(&s, &s32); - return (copyout(&s32, uap->buf, sizeof(s32))); + sp = malloc(sizeof(struct statfs), M_STATFS, M_WAITOK); + error = kern_statfs(td, uap->path, UIO_USERSPACE, sp); + if (error == 0) { + copy_statfs(sp, &s32); + error = copyout(&s32, uap->buf, sizeof(s32)); + } + free(sp, M_STATFS); + return (error); } #endif #ifdef COMPAT_FREEBSD4 int freebsd4_freebsd32_fstatfs(struct thread *td, struct freebsd4_freebsd32_fstatfs_args *uap) { struct statfs32 s32; - struct statfs s; + struct statfs *sp; int error; - error = kern_fstatfs(td, uap->fd, &s); - if (error) - return (error); - copy_statfs(&s, &s32); - return (copyout(&s32, uap->buf, sizeof(s32))); + sp = malloc(sizeof(struct statfs), M_STATFS, M_WAITOK); + error = kern_fstatfs(td, uap->fd, sp); + if (error == 0) { + copy_statfs(sp, &s32); + error = copyout(&s32, uap->buf, sizeof(s32)); + } + free(sp, M_STATFS); + return (error); } #endif #ifdef COMPAT_FREEBSD4 int freebsd4_freebsd32_fhstatfs(struct thread *td, struct freebsd4_freebsd32_fhstatfs_args *uap) { struct statfs32 s32; - struct statfs s; + struct statfs *sp; fhandle_t fh; int error; if ((error = copyin(uap->u_fhp, &fh, sizeof(fhandle_t))) != 0) return (error); - error = kern_fhstatfs(td, fh, &s); - if (error) - return (error); - copy_statfs(&s, &s32); - return (copyout(&s32, uap->buf, sizeof(s32))); + sp = malloc(sizeof(struct statfs), M_STATFS, M_WAITOK); + error = kern_fhstatfs(td, fh, sp); + if (error == 0) { + copy_statfs(sp, &s32); + error = copyout(&s32, uap->buf, sizeof(s32)); + } + free(sp, M_STATFS); + return (error); } #endif int freebsd32_pread(struct thread *td, struct freebsd32_pread_args *uap) { struct pread_args ap; ap.fd = uap->fd; ap.buf = uap->buf; ap.nbyte = uap->nbyte; ap.offset = PAIR32TO64(off_t,uap->offset); return (sys_pread(td, &ap)); } int freebsd32_pwrite(struct thread *td, struct freebsd32_pwrite_args *uap) { struct pwrite_args ap; ap.fd = uap->fd; ap.buf = uap->buf; ap.nbyte = uap->nbyte; ap.offset = PAIR32TO64(off_t,uap->offset); return (sys_pwrite(td, &ap)); } #ifdef COMPAT_43 int ofreebsd32_lseek(struct thread *td, struct ofreebsd32_lseek_args *uap) { struct lseek_args nuap; nuap.fd = uap->fd; nuap.offset = uap->offset; nuap.whence = uap->whence; return (sys_lseek(td, &nuap)); } #endif int freebsd32_lseek(struct thread *td, struct freebsd32_lseek_args *uap) { int error; struct lseek_args ap; off_t pos; ap.fd = uap->fd; ap.offset = PAIR32TO64(off_t,uap->offset); ap.whence = uap->whence; error = sys_lseek(td, &ap); /* Expand the quad return into two parts for eax and edx */ pos = td->td_uretoff.tdu_off; td->td_retval[RETVAL_LO] = pos & 0xffffffff; /* %eax */ td->td_retval[RETVAL_HI] = pos >> 32; /* %edx */ return error; } int freebsd32_truncate(struct thread *td, struct freebsd32_truncate_args *uap) { struct truncate_args ap; ap.path = uap->path; ap.length = PAIR32TO64(off_t,uap->length); return (sys_truncate(td, &ap)); } int freebsd32_ftruncate(struct thread *td, struct freebsd32_ftruncate_args *uap) { struct ftruncate_args ap; ap.fd = uap->fd; ap.length = PAIR32TO64(off_t,uap->length); return (sys_ftruncate(td, &ap)); } #ifdef COMPAT_43 int ofreebsd32_getdirentries(struct thread *td, struct ofreebsd32_getdirentries_args *uap) { struct ogetdirentries_args ap; int error; long loff; int32_t loff_cut; ap.fd = uap->fd; ap.buf = uap->buf; ap.count = uap->count; ap.basep = NULL; error = kern_ogetdirentries(td, &ap, &loff); if (error == 0) { loff_cut = loff; error = copyout(&loff_cut, uap->basep, sizeof(int32_t)); } return (error); } #endif int freebsd32_getdirentries(struct thread *td, struct freebsd32_getdirentries_args *uap) { long base; int32_t base32; int error; error = kern_getdirentries(td, uap->fd, uap->buf, uap->count, &base, NULL, UIO_USERSPACE); if (error) return (error); if (uap->basep != NULL) { base32 = base; error = copyout(&base32, uap->basep, sizeof(int32_t)); } return (error); } #ifdef COMPAT_FREEBSD6 /* versions with the 'int pad' argument */ int freebsd6_freebsd32_pread(struct thread *td, struct freebsd6_freebsd32_pread_args *uap) { struct pread_args ap; ap.fd = uap->fd; ap.buf = uap->buf; ap.nbyte = uap->nbyte; ap.offset = PAIR32TO64(off_t,uap->offset); return (sys_pread(td, &ap)); } int freebsd6_freebsd32_pwrite(struct thread *td, struct freebsd6_freebsd32_pwrite_args *uap) { struct pwrite_args ap; ap.fd = uap->fd; ap.buf = uap->buf; ap.nbyte = uap->nbyte; ap.offset = PAIR32TO64(off_t,uap->offset); return (sys_pwrite(td, &ap)); } int freebsd6_freebsd32_lseek(struct thread *td, struct freebsd6_freebsd32_lseek_args *uap) { int error; struct lseek_args ap; off_t pos; ap.fd = uap->fd; ap.offset = PAIR32TO64(off_t,uap->offset); ap.whence = uap->whence; error = sys_lseek(td, &ap); /* Expand the quad return into two parts for eax and edx */ pos = *(off_t *)(td->td_retval); td->td_retval[RETVAL_LO] = pos & 0xffffffff; /* %eax */ td->td_retval[RETVAL_HI] = pos >> 32; /* %edx */ return error; } int freebsd6_freebsd32_truncate(struct thread *td, struct freebsd6_freebsd32_truncate_args *uap) { struct truncate_args ap; ap.path = uap->path; ap.length = PAIR32TO64(off_t,uap->length); return (sys_truncate(td, &ap)); } int freebsd6_freebsd32_ftruncate(struct thread *td, struct freebsd6_freebsd32_ftruncate_args *uap) { struct ftruncate_args ap; ap.fd = uap->fd; ap.length = PAIR32TO64(off_t,uap->length); return (sys_ftruncate(td, &ap)); } #endif /* COMPAT_FREEBSD6 */ struct sf_hdtr32 { uint32_t headers; int hdr_cnt; uint32_t trailers; int trl_cnt; }; static int freebsd32_do_sendfile(struct thread *td, struct freebsd32_sendfile_args *uap, int compat) { struct sf_hdtr32 hdtr32; struct sf_hdtr hdtr; struct uio *hdr_uio, *trl_uio; struct file *fp; cap_rights_t rights; struct iovec32 *iov32; off_t offset, sbytes; int error; offset = PAIR32TO64(off_t, uap->offset); if (offset < 0) return (EINVAL); hdr_uio = trl_uio = NULL; if (uap->hdtr != NULL) { error = copyin(uap->hdtr, &hdtr32, sizeof(hdtr32)); if (error) goto out; PTRIN_CP(hdtr32, hdtr, headers); CP(hdtr32, hdtr, hdr_cnt); PTRIN_CP(hdtr32, hdtr, trailers); CP(hdtr32, hdtr, trl_cnt); if (hdtr.headers != NULL) { iov32 = PTRIN(hdtr32.headers); error = freebsd32_copyinuio(iov32, hdtr32.hdr_cnt, &hdr_uio); if (error) goto out; #ifdef COMPAT_FREEBSD4 /* * In FreeBSD < 5.0 the nbytes to send also included * the header. If compat is specified subtract the * header size from nbytes. */ if (compat) { if (uap->nbytes > hdr_uio->uio_resid) uap->nbytes -= hdr_uio->uio_resid; else uap->nbytes = 0; } #endif } if (hdtr.trailers != NULL) { iov32 = PTRIN(hdtr32.trailers); error = freebsd32_copyinuio(iov32, hdtr32.trl_cnt, &trl_uio); if (error) goto out; } } AUDIT_ARG_FD(uap->fd); if ((error = fget_read(td, uap->fd, cap_rights_init(&rights, CAP_PREAD), &fp)) != 0) goto out; error = fo_sendfile(fp, uap->s, hdr_uio, trl_uio, offset, uap->nbytes, &sbytes, uap->flags, td); fdrop(fp, td); if (uap->sbytes != NULL) copyout(&sbytes, uap->sbytes, sizeof(off_t)); out: if (hdr_uio) free(hdr_uio, M_IOV); if (trl_uio) free(trl_uio, M_IOV); return (error); } #ifdef COMPAT_FREEBSD4 int freebsd4_freebsd32_sendfile(struct thread *td, struct freebsd4_freebsd32_sendfile_args *uap) { return (freebsd32_do_sendfile(td, (struct freebsd32_sendfile_args *)uap, 1)); } #endif int freebsd32_sendfile(struct thread *td, struct freebsd32_sendfile_args *uap) { return (freebsd32_do_sendfile(td, uap, 0)); } static void copy_stat(struct stat *in, struct stat32 *out) { CP(*in, *out, st_dev); CP(*in, *out, st_ino); CP(*in, *out, st_mode); CP(*in, *out, st_nlink); CP(*in, *out, st_uid); CP(*in, *out, st_gid); CP(*in, *out, st_rdev); TS_CP(*in, *out, st_atim); TS_CP(*in, *out, st_mtim); TS_CP(*in, *out, st_ctim); CP(*in, *out, st_size); CP(*in, *out, st_blocks); CP(*in, *out, st_blksize); CP(*in, *out, st_flags); CP(*in, *out, st_gen); TS_CP(*in, *out, st_birthtim); } #ifdef COMPAT_43 static void copy_ostat(struct stat *in, struct ostat32 *out) { CP(*in, *out, st_dev); CP(*in, *out, st_ino); CP(*in, *out, st_mode); CP(*in, *out, st_nlink); CP(*in, *out, st_uid); CP(*in, *out, st_gid); CP(*in, *out, st_rdev); CP(*in, *out, st_size); TS_CP(*in, *out, st_atim); TS_CP(*in, *out, st_mtim); TS_CP(*in, *out, st_ctim); CP(*in, *out, st_blksize); CP(*in, *out, st_blocks); CP(*in, *out, st_flags); CP(*in, *out, st_gen); } #endif int freebsd32_stat(struct thread *td, struct freebsd32_stat_args *uap) { struct stat sb; struct stat32 sb32; int error; error = kern_statat(td, 0, AT_FDCWD, uap->path, UIO_USERSPACE, &sb, NULL); if (error) return (error); copy_stat(&sb, &sb32); error = copyout(&sb32, uap->ub, sizeof (sb32)); return (error); } #ifdef COMPAT_43 int ofreebsd32_stat(struct thread *td, struct ofreebsd32_stat_args *uap) { struct stat sb; struct ostat32 sb32; int error; error = kern_statat(td, 0, AT_FDCWD, uap->path, UIO_USERSPACE, &sb, NULL); if (error) return (error); copy_ostat(&sb, &sb32); error = copyout(&sb32, uap->ub, sizeof (sb32)); return (error); } #endif int freebsd32_fstat(struct thread *td, struct freebsd32_fstat_args *uap) { struct stat ub; struct stat32 ub32; int error; error = kern_fstat(td, uap->fd, &ub); if (error) return (error); copy_stat(&ub, &ub32); error = copyout(&ub32, uap->ub, sizeof(ub32)); return (error); } #ifdef COMPAT_43 int ofreebsd32_fstat(struct thread *td, struct ofreebsd32_fstat_args *uap) { struct stat ub; struct ostat32 ub32; int error; error = kern_fstat(td, uap->fd, &ub); if (error) return (error); copy_ostat(&ub, &ub32); error = copyout(&ub32, uap->ub, sizeof(ub32)); return (error); } #endif int freebsd32_fstatat(struct thread *td, struct freebsd32_fstatat_args *uap) { struct stat ub; struct stat32 ub32; int error; error = kern_statat(td, uap->flag, uap->fd, uap->path, UIO_USERSPACE, &ub, NULL); if (error) return (error); copy_stat(&ub, &ub32); error = copyout(&ub32, uap->buf, sizeof(ub32)); return (error); } int freebsd32_lstat(struct thread *td, struct freebsd32_lstat_args *uap) { struct stat sb; struct stat32 sb32; int error; error = kern_statat(td, AT_SYMLINK_NOFOLLOW, AT_FDCWD, uap->path, UIO_USERSPACE, &sb, NULL); if (error) return (error); copy_stat(&sb, &sb32); error = copyout(&sb32, uap->ub, sizeof (sb32)); return (error); } #ifdef COMPAT_43 int ofreebsd32_lstat(struct thread *td, struct ofreebsd32_lstat_args *uap) { struct stat sb; struct ostat32 sb32; int error; error = kern_statat(td, AT_SYMLINK_NOFOLLOW, AT_FDCWD, uap->path, UIO_USERSPACE, &sb, NULL); if (error) return (error); copy_ostat(&sb, &sb32); error = copyout(&sb32, uap->ub, sizeof (sb32)); return (error); } #endif int freebsd32_sysctl(struct thread *td, struct freebsd32_sysctl_args *uap) { int error, name[CTL_MAXNAME]; size_t j, oldlen; uint32_t tmp; if (uap->namelen > CTL_MAXNAME || uap->namelen < 2) return (EINVAL); error = copyin(uap->name, name, uap->namelen * sizeof(int)); if (error) return (error); if (uap->oldlenp) { error = fueword32(uap->oldlenp, &tmp); oldlen = tmp; } else { oldlen = 0; } if (error != 0) return (EFAULT); error = userland_sysctl(td, name, uap->namelen, uap->old, &oldlen, 1, uap->new, uap->newlen, &j, SCTL_MASK32); if (error && error != ENOMEM) return (error); if (uap->oldlenp) suword32(uap->oldlenp, j); return (0); } int freebsd32_jail(struct thread *td, struct freebsd32_jail_args *uap) { uint32_t version; int error; struct jail j; error = copyin(uap->jail, &version, sizeof(uint32_t)); if (error) return (error); switch (version) { case 0: { /* FreeBSD single IPv4 jails. */ struct jail32_v0 j32_v0; bzero(&j, sizeof(struct jail)); error = copyin(uap->jail, &j32_v0, sizeof(struct jail32_v0)); if (error) return (error); CP(j32_v0, j, version); PTRIN_CP(j32_v0, j, path); PTRIN_CP(j32_v0, j, hostname); j.ip4s = htonl(j32_v0.ip_number); /* jail_v0 is host order */ break; } case 1: /* * Version 1 was used by multi-IPv4 jail implementations * that never made it into the official kernel. */ return (EINVAL); case 2: /* JAIL_API_VERSION */ { /* FreeBSD multi-IPv4/IPv6,noIP jails. */ struct jail32 j32; error = copyin(uap->jail, &j32, sizeof(struct jail32)); if (error) return (error); CP(j32, j, version); PTRIN_CP(j32, j, path); PTRIN_CP(j32, j, hostname); PTRIN_CP(j32, j, jailname); CP(j32, j, ip4s); CP(j32, j, ip6s); PTRIN_CP(j32, j, ip4); PTRIN_CP(j32, j, ip6); break; } default: /* Sci-Fi jails are not supported, sorry. */ return (EINVAL); } return (kern_jail(td, &j)); } int freebsd32_jail_set(struct thread *td, struct freebsd32_jail_set_args *uap) { struct uio *auio; int error; /* Check that we have an even number of iovecs. */ if (uap->iovcnt & 1) return (EINVAL); error = freebsd32_copyinuio(uap->iovp, uap->iovcnt, &auio); if (error) return (error); error = kern_jail_set(td, auio, uap->flags); free(auio, M_IOV); return (error); } int freebsd32_jail_get(struct thread *td, struct freebsd32_jail_get_args *uap) { struct iovec32 iov32; struct uio *auio; int error, i; /* Check that we have an even number of iovecs. */ if (uap->iovcnt & 1) return (EINVAL); error = freebsd32_copyinuio(uap->iovp, uap->iovcnt, &auio); if (error) return (error); error = kern_jail_get(td, auio, uap->flags); if (error == 0) for (i = 0; i < uap->iovcnt; i++) { PTROUT_CP(auio->uio_iov[i], iov32, iov_base); CP(auio->uio_iov[i], iov32, iov_len); error = copyout(&iov32, uap->iovp + i, sizeof(iov32)); if (error != 0) break; } free(auio, M_IOV); return (error); } int freebsd32_sigaction(struct thread *td, struct freebsd32_sigaction_args *uap) { struct sigaction32 s32; struct sigaction sa, osa, *sap; int error; if (uap->act) { error = copyin(uap->act, &s32, sizeof(s32)); if (error) return (error); sa.sa_handler = PTRIN(s32.sa_u); CP(s32, sa, sa_flags); CP(s32, sa, sa_mask); sap = &sa; } else sap = NULL; error = kern_sigaction(td, uap->sig, sap, &osa, 0); if (error == 0 && uap->oact != NULL) { s32.sa_u = PTROUT(osa.sa_handler); CP(osa, s32, sa_flags); CP(osa, s32, sa_mask); error = copyout(&s32, uap->oact, sizeof(s32)); } return (error); } #ifdef COMPAT_FREEBSD4 int freebsd4_freebsd32_sigaction(struct thread *td, struct freebsd4_freebsd32_sigaction_args *uap) { struct sigaction32 s32; struct sigaction sa, osa, *sap; int error; if (uap->act) { error = copyin(uap->act, &s32, sizeof(s32)); if (error) return (error); sa.sa_handler = PTRIN(s32.sa_u); CP(s32, sa, sa_flags); CP(s32, sa, sa_mask); sap = &sa; } else sap = NULL; error = kern_sigaction(td, uap->sig, sap, &osa, KSA_FREEBSD4); if (error == 0 && uap->oact != NULL) { s32.sa_u = PTROUT(osa.sa_handler); CP(osa, s32, sa_flags); CP(osa, s32, sa_mask); error = copyout(&s32, uap->oact, sizeof(s32)); } return (error); } #endif #ifdef COMPAT_43 struct osigaction32 { u_int32_t sa_u; osigset_t sa_mask; int sa_flags; }; #define ONSIG 32 int ofreebsd32_sigaction(struct thread *td, struct ofreebsd32_sigaction_args *uap) { struct osigaction32 s32; struct sigaction sa, osa, *sap; int error; if (uap->signum <= 0 || uap->signum >= ONSIG) return (EINVAL); if (uap->nsa) { error = copyin(uap->nsa, &s32, sizeof(s32)); if (error) return (error); sa.sa_handler = PTRIN(s32.sa_u); CP(s32, sa, sa_flags); OSIG2SIG(s32.sa_mask, sa.sa_mask); sap = &sa; } else sap = NULL; error = kern_sigaction(td, uap->signum, sap, &osa, KSA_OSIGSET); if (error == 0 && uap->osa != NULL) { s32.sa_u = PTROUT(osa.sa_handler); CP(osa, s32, sa_flags); SIG2OSIG(osa.sa_mask, s32.sa_mask); error = copyout(&s32, uap->osa, sizeof(s32)); } return (error); } int ofreebsd32_sigprocmask(struct thread *td, struct ofreebsd32_sigprocmask_args *uap) { sigset_t set, oset; int error; OSIG2SIG(uap->mask, set); error = kern_sigprocmask(td, uap->how, &set, &oset, SIGPROCMASK_OLD); SIG2OSIG(oset, td->td_retval[0]); return (error); } int ofreebsd32_sigpending(struct thread *td, struct ofreebsd32_sigpending_args *uap) { struct proc *p = td->td_proc; sigset_t siglist; PROC_LOCK(p); siglist = p->p_siglist; SIGSETOR(siglist, td->td_siglist); PROC_UNLOCK(p); SIG2OSIG(siglist, td->td_retval[0]); return (0); } struct sigvec32 { u_int32_t sv_handler; int sv_mask; int sv_flags; }; int ofreebsd32_sigvec(struct thread *td, struct ofreebsd32_sigvec_args *uap) { struct sigvec32 vec; struct sigaction sa, osa, *sap; int error; if (uap->signum <= 0 || uap->signum >= ONSIG) return (EINVAL); if (uap->nsv) { error = copyin(uap->nsv, &vec, sizeof(vec)); if (error) return (error); sa.sa_handler = PTRIN(vec.sv_handler); OSIG2SIG(vec.sv_mask, sa.sa_mask); sa.sa_flags = vec.sv_flags; sa.sa_flags ^= SA_RESTART; sap = &sa; } else sap = NULL; error = kern_sigaction(td, uap->signum, sap, &osa, KSA_OSIGSET); if (error == 0 && uap->osv != NULL) { vec.sv_handler = PTROUT(osa.sa_handler); SIG2OSIG(osa.sa_mask, vec.sv_mask); vec.sv_flags = osa.sa_flags; vec.sv_flags &= ~SA_NOCLDWAIT; vec.sv_flags ^= SA_RESTART; error = copyout(&vec, uap->osv, sizeof(vec)); } return (error); } int ofreebsd32_sigblock(struct thread *td, struct ofreebsd32_sigblock_args *uap) { sigset_t set, oset; OSIG2SIG(uap->mask, set); kern_sigprocmask(td, SIG_BLOCK, &set, &oset, 0); SIG2OSIG(oset, td->td_retval[0]); return (0); } int ofreebsd32_sigsetmask(struct thread *td, struct ofreebsd32_sigsetmask_args *uap) { sigset_t set, oset; OSIG2SIG(uap->mask, set); kern_sigprocmask(td, SIG_SETMASK, &set, &oset, 0); SIG2OSIG(oset, td->td_retval[0]); return (0); } int ofreebsd32_sigsuspend(struct thread *td, struct ofreebsd32_sigsuspend_args *uap) { sigset_t mask; OSIG2SIG(uap->mask, mask); return (kern_sigsuspend(td, mask)); } struct sigstack32 { u_int32_t ss_sp; int ss_onstack; }; int ofreebsd32_sigstack(struct thread *td, struct ofreebsd32_sigstack_args *uap) { struct sigstack32 s32; struct sigstack nss, oss; int error = 0, unss; if (uap->nss != NULL) { error = copyin(uap->nss, &s32, sizeof(s32)); if (error) return (error); nss.ss_sp = PTRIN(s32.ss_sp); CP(s32, nss, ss_onstack); unss = 1; } else { unss = 0; } oss.ss_sp = td->td_sigstk.ss_sp; oss.ss_onstack = sigonstack(cpu_getstack(td)); if (unss) { td->td_sigstk.ss_sp = nss.ss_sp; td->td_sigstk.ss_size = 0; td->td_sigstk.ss_flags |= (nss.ss_onstack & SS_ONSTACK); td->td_pflags |= TDP_ALTSTACK; } if (uap->oss != NULL) { s32.ss_sp = PTROUT(oss.ss_sp); CP(oss, s32, ss_onstack); error = copyout(&s32, uap->oss, sizeof(s32)); } return (error); } #endif int freebsd32_nanosleep(struct thread *td, struct freebsd32_nanosleep_args *uap) { struct timespec32 rmt32, rqt32; struct timespec rmt, rqt; int error; error = copyin(uap->rqtp, &rqt32, sizeof(rqt32)); if (error) return (error); CP(rqt32, rqt, tv_sec); CP(rqt32, rqt, tv_nsec); if (uap->rmtp && !useracc((caddr_t)uap->rmtp, sizeof(rmt), VM_PROT_WRITE)) return (EFAULT); error = kern_nanosleep(td, &rqt, &rmt); if (error && uap->rmtp) { int error2; CP(rmt, rmt32, tv_sec); CP(rmt, rmt32, tv_nsec); error2 = copyout(&rmt32, uap->rmtp, sizeof(rmt32)); if (error2) error = error2; } return (error); } int freebsd32_clock_gettime(struct thread *td, struct freebsd32_clock_gettime_args *uap) { struct timespec ats; struct timespec32 ats32; int error; error = kern_clock_gettime(td, uap->clock_id, &ats); if (error == 0) { CP(ats, ats32, tv_sec); CP(ats, ats32, tv_nsec); error = copyout(&ats32, uap->tp, sizeof(ats32)); } return (error); } int freebsd32_clock_settime(struct thread *td, struct freebsd32_clock_settime_args *uap) { struct timespec ats; struct timespec32 ats32; int error; error = copyin(uap->tp, &ats32, sizeof(ats32)); if (error) return (error); CP(ats32, ats, tv_sec); CP(ats32, ats, tv_nsec); return (kern_clock_settime(td, uap->clock_id, &ats)); } int freebsd32_clock_getres(struct thread *td, struct freebsd32_clock_getres_args *uap) { struct timespec ts; struct timespec32 ts32; int error; if (uap->tp == NULL) return (0); error = kern_clock_getres(td, uap->clock_id, &ts); if (error == 0) { CP(ts, ts32, tv_sec); CP(ts, ts32, tv_nsec); error = copyout(&ts32, uap->tp, sizeof(ts32)); } return (error); } int freebsd32_ktimer_create(struct thread *td, struct freebsd32_ktimer_create_args *uap) { struct sigevent32 ev32; struct sigevent ev, *evp; int error, id; if (uap->evp == NULL) { evp = NULL; } else { evp = &ev; error = copyin(uap->evp, &ev32, sizeof(ev32)); if (error != 0) return (error); error = convert_sigevent32(&ev32, &ev); if (error != 0) return (error); } error = kern_ktimer_create(td, uap->clock_id, evp, &id, -1); if (error == 0) { error = copyout(&id, uap->timerid, sizeof(int)); if (error != 0) kern_ktimer_delete(td, id); } return (error); } int freebsd32_ktimer_settime(struct thread *td, struct freebsd32_ktimer_settime_args *uap) { struct itimerspec32 val32, oval32; struct itimerspec val, oval, *ovalp; int error; error = copyin(uap->value, &val32, sizeof(val32)); if (error != 0) return (error); ITS_CP(val32, val); ovalp = uap->ovalue != NULL ? &oval : NULL; error = kern_ktimer_settime(td, uap->timerid, uap->flags, &val, ovalp); if (error == 0 && uap->ovalue != NULL) { ITS_CP(oval, oval32); error = copyout(&oval32, uap->ovalue, sizeof(oval32)); } return (error); } int freebsd32_ktimer_gettime(struct thread *td, struct freebsd32_ktimer_gettime_args *uap) { struct itimerspec32 val32; struct itimerspec val; int error; error = kern_ktimer_gettime(td, uap->timerid, &val); if (error == 0) { ITS_CP(val, val32); error = copyout(&val32, uap->value, sizeof(val32)); } return (error); } int freebsd32_clock_getcpuclockid2(struct thread *td, struct freebsd32_clock_getcpuclockid2_args *uap) { clockid_t clk_id; int error; error = kern_clock_getcpuclockid2(td, PAIR32TO64(id_t, uap->id), uap->which, &clk_id); if (error == 0) error = copyout(&clk_id, uap->clock_id, sizeof(clockid_t)); return (error); } int freebsd32_thr_new(struct thread *td, struct freebsd32_thr_new_args *uap) { struct thr_param32 param32; struct thr_param param; int error; if (uap->param_size < 0 || uap->param_size > sizeof(struct thr_param32)) return (EINVAL); bzero(¶m, sizeof(struct thr_param)); bzero(¶m32, sizeof(struct thr_param32)); error = copyin(uap->param, ¶m32, uap->param_size); if (error != 0) return (error); param.start_func = PTRIN(param32.start_func); param.arg = PTRIN(param32.arg); param.stack_base = PTRIN(param32.stack_base); param.stack_size = param32.stack_size; param.tls_base = PTRIN(param32.tls_base); param.tls_size = param32.tls_size; param.child_tid = PTRIN(param32.child_tid); param.parent_tid = PTRIN(param32.parent_tid); param.flags = param32.flags; param.rtp = PTRIN(param32.rtp); param.spare[0] = PTRIN(param32.spare[0]); param.spare[1] = PTRIN(param32.spare[1]); param.spare[2] = PTRIN(param32.spare[2]); return (kern_thr_new(td, ¶m)); } int freebsd32_thr_suspend(struct thread *td, struct freebsd32_thr_suspend_args *uap) { struct timespec32 ts32; struct timespec ts, *tsp; int error; error = 0; tsp = NULL; if (uap->timeout != NULL) { error = copyin((const void *)uap->timeout, (void *)&ts32, sizeof(struct timespec32)); if (error != 0) return (error); ts.tv_sec = ts32.tv_sec; ts.tv_nsec = ts32.tv_nsec; tsp = &ts; } return (kern_thr_suspend(td, tsp)); } void siginfo_to_siginfo32(const siginfo_t *src, struct siginfo32 *dst) { bzero(dst, sizeof(*dst)); dst->si_signo = src->si_signo; dst->si_errno = src->si_errno; dst->si_code = src->si_code; dst->si_pid = src->si_pid; dst->si_uid = src->si_uid; dst->si_status = src->si_status; dst->si_addr = (uintptr_t)src->si_addr; dst->si_value.sival_int = src->si_value.sival_int; dst->si_timerid = src->si_timerid; dst->si_overrun = src->si_overrun; } int freebsd32_sigtimedwait(struct thread *td, struct freebsd32_sigtimedwait_args *uap) { struct timespec32 ts32; struct timespec ts; struct timespec *timeout; sigset_t set; ksiginfo_t ksi; struct siginfo32 si32; int error; if (uap->timeout) { error = copyin(uap->timeout, &ts32, sizeof(ts32)); if (error) return (error); ts.tv_sec = ts32.tv_sec; ts.tv_nsec = ts32.tv_nsec; timeout = &ts; } else timeout = NULL; error = copyin(uap->set, &set, sizeof(set)); if (error) return (error); error = kern_sigtimedwait(td, set, &ksi, timeout); if (error) return (error); if (uap->info) { siginfo_to_siginfo32(&ksi.ksi_info, &si32); error = copyout(&si32, uap->info, sizeof(struct siginfo32)); } if (error == 0) td->td_retval[0] = ksi.ksi_signo; return (error); } /* * MPSAFE */ int freebsd32_sigwaitinfo(struct thread *td, struct freebsd32_sigwaitinfo_args *uap) { ksiginfo_t ksi; struct siginfo32 si32; sigset_t set; int error; error = copyin(uap->set, &set, sizeof(set)); if (error) return (error); error = kern_sigtimedwait(td, set, &ksi, NULL); if (error) return (error); if (uap->info) { siginfo_to_siginfo32(&ksi.ksi_info, &si32); error = copyout(&si32, uap->info, sizeof(struct siginfo32)); } if (error == 0) td->td_retval[0] = ksi.ksi_signo; return (error); } int freebsd32_cpuset_setid(struct thread *td, struct freebsd32_cpuset_setid_args *uap) { struct cpuset_setid_args ap; ap.which = uap->which; ap.id = PAIR32TO64(id_t,uap->id); ap.setid = uap->setid; return (sys_cpuset_setid(td, &ap)); } int freebsd32_cpuset_getid(struct thread *td, struct freebsd32_cpuset_getid_args *uap) { struct cpuset_getid_args ap; ap.level = uap->level; ap.which = uap->which; ap.id = PAIR32TO64(id_t,uap->id); ap.setid = uap->setid; return (sys_cpuset_getid(td, &ap)); } int freebsd32_cpuset_getaffinity(struct thread *td, struct freebsd32_cpuset_getaffinity_args *uap) { struct cpuset_getaffinity_args ap; ap.level = uap->level; ap.which = uap->which; ap.id = PAIR32TO64(id_t,uap->id); ap.cpusetsize = uap->cpusetsize; ap.mask = uap->mask; return (sys_cpuset_getaffinity(td, &ap)); } int freebsd32_cpuset_setaffinity(struct thread *td, struct freebsd32_cpuset_setaffinity_args *uap) { struct cpuset_setaffinity_args ap; ap.level = uap->level; ap.which = uap->which; ap.id = PAIR32TO64(id_t,uap->id); ap.cpusetsize = uap->cpusetsize; ap.mask = uap->mask; return (sys_cpuset_setaffinity(td, &ap)); } int freebsd32_nmount(struct thread *td, struct freebsd32_nmount_args /* { struct iovec *iovp; unsigned int iovcnt; int flags; } */ *uap) { struct uio *auio; uint64_t flags; int error; /* * Mount flags are now 64-bits. On 32-bit archtectures only * 32-bits are passed in, but from here on everything handles * 64-bit flags correctly. */ flags = uap->flags; AUDIT_ARG_FFLAGS(flags); /* * Filter out MNT_ROOTFS. We do not want clients of nmount() in * userspace to set this flag, but we must filter it out if we want * MNT_UPDATE on the root file system to work. * MNT_ROOTFS should only be set by the kernel when mounting its * root file system. */ flags &= ~MNT_ROOTFS; /* * check that we have an even number of iovec's * and that we have at least two options. */ if ((uap->iovcnt & 1) || (uap->iovcnt < 4)) return (EINVAL); error = freebsd32_copyinuio(uap->iovp, uap->iovcnt, &auio); if (error) return (error); error = vfs_donmount(td, flags, auio); free(auio, M_IOV); return error; } #if 0 int freebsd32_xxx(struct thread *td, struct freebsd32_xxx_args *uap) { struct yyy32 *p32, s32; struct yyy *p = NULL, s; struct xxx_arg ap; int error; if (uap->zzz) { error = copyin(uap->zzz, &s32, sizeof(s32)); if (error) return (error); /* translate in */ p = &s; } error = kern_xxx(td, p); if (error) return (error); if (uap->zzz) { /* translate out */ error = copyout(&s32, p32, sizeof(s32)); } return (error); } #endif int syscall32_register(int *offset, struct sysent *new_sysent, struct sysent *old_sysent, int flags) { if ((flags & ~SY_THR_STATIC) != 0) return (EINVAL); if (*offset == NO_SYSCALL) { int i; for (i = 1; i < SYS_MAXSYSCALL; ++i) if (freebsd32_sysent[i].sy_call == (sy_call_t *)lkmnosys) break; if (i == SYS_MAXSYSCALL) return (ENFILE); *offset = i; } else if (*offset < 0 || *offset >= SYS_MAXSYSCALL) return (EINVAL); else if (freebsd32_sysent[*offset].sy_call != (sy_call_t *)lkmnosys && freebsd32_sysent[*offset].sy_call != (sy_call_t *)lkmressys) return (EEXIST); *old_sysent = freebsd32_sysent[*offset]; freebsd32_sysent[*offset] = *new_sysent; atomic_store_rel_32(&freebsd32_sysent[*offset].sy_thrcnt, flags); return (0); } int syscall32_deregister(int *offset, struct sysent *old_sysent) { if (*offset == 0) return (0); freebsd32_sysent[*offset] = *old_sysent; return (0); } int syscall32_module_handler(struct module *mod, int what, void *arg) { struct syscall_module_data *data = (struct syscall_module_data*)arg; modspecific_t ms; int error; switch (what) { case MOD_LOAD: error = syscall32_register(data->offset, data->new_sysent, &data->old_sysent, SY_THR_STATIC_KLD); if (error) { /* Leave a mark so we know to safely unload below. */ data->offset = NULL; return error; } ms.intval = *data->offset; MOD_XLOCK; module_setspecific(mod, &ms); MOD_XUNLOCK; if (data->chainevh) error = data->chainevh(mod, what, data->chainarg); return (error); case MOD_UNLOAD: /* * MOD_LOAD failed, so just return without calling the * chained handler since we didn't pass along the MOD_LOAD * event. */ if (data->offset == NULL) return (0); if (data->chainevh) { error = data->chainevh(mod, what, data->chainarg); if (error) return (error); } error = syscall32_deregister(data->offset, &data->old_sysent); return (error); default: error = EOPNOTSUPP; if (data->chainevh) error = data->chainevh(mod, what, data->chainarg); return (error); } } int syscall32_helper_register(struct syscall_helper_data *sd, int flags) { struct syscall_helper_data *sd1; int error; for (sd1 = sd; sd1->syscall_no != NO_SYSCALL; sd1++) { error = syscall32_register(&sd1->syscall_no, &sd1->new_sysent, &sd1->old_sysent, flags); if (error != 0) { syscall32_helper_unregister(sd); return (error); } sd1->registered = 1; } return (0); } int syscall32_helper_unregister(struct syscall_helper_data *sd) { struct syscall_helper_data *sd1; for (sd1 = sd; sd1->registered != 0; sd1++) { syscall32_deregister(&sd1->syscall_no, &sd1->old_sysent); sd1->registered = 0; } return (0); } register_t * freebsd32_copyout_strings(struct image_params *imgp) { int argc, envc, i; u_int32_t *vectp; char *stringp; uintptr_t destp; u_int32_t *stack_base; struct freebsd32_ps_strings *arginfo; char canary[sizeof(long) * 8]; int32_t pagesizes32[MAXPAGESIZES]; size_t execpath_len; int szsigcode; /* * Calculate string base and vector table pointers. * Also deal with signal trampoline code for this exec type. */ if (imgp->execpath != NULL && imgp->auxargs != NULL) execpath_len = strlen(imgp->execpath) + 1; else execpath_len = 0; arginfo = (struct freebsd32_ps_strings *)curproc->p_sysent-> sv_psstrings; if (imgp->proc->p_sysent->sv_sigcode_base == 0) szsigcode = *(imgp->proc->p_sysent->sv_szsigcode); else szsigcode = 0; destp = (uintptr_t)arginfo; /* * install sigcode */ if (szsigcode != 0) { destp -= szsigcode; destp = rounddown2(destp, sizeof(uint32_t)); copyout(imgp->proc->p_sysent->sv_sigcode, (void *)destp, szsigcode); } /* * Copy the image path for the rtld. */ if (execpath_len != 0) { destp -= execpath_len; imgp->execpathp = destp; copyout(imgp->execpath, (void *)destp, execpath_len); } /* * Prepare the canary for SSP. */ arc4rand(canary, sizeof(canary), 0); destp -= sizeof(canary); imgp->canary = destp; copyout(canary, (void *)destp, sizeof(canary)); imgp->canarylen = sizeof(canary); /* * Prepare the pagesizes array. */ for (i = 0; i < MAXPAGESIZES; i++) pagesizes32[i] = (uint32_t)pagesizes[i]; destp -= sizeof(pagesizes32); destp = rounddown2(destp, sizeof(uint32_t)); imgp->pagesizes = destp; copyout(pagesizes32, (void *)destp, sizeof(pagesizes32)); imgp->pagesizeslen = sizeof(pagesizes32); destp -= ARG_MAX - imgp->args->stringspace; destp = rounddown2(destp, sizeof(uint32_t)); /* * If we have a valid auxargs ptr, prepare some room * on the stack. */ if (imgp->auxargs) { /* * 'AT_COUNT*2' is size for the ELF Auxargs data. This is for * lower compatibility. */ imgp->auxarg_size = (imgp->auxarg_size) ? imgp->auxarg_size : (AT_COUNT * 2); /* * The '+ 2' is for the null pointers at the end of each of * the arg and env vector sets,and imgp->auxarg_size is room * for argument of Runtime loader. */ vectp = (u_int32_t *) (destp - (imgp->args->argc + imgp->args->envc + 2 + imgp->auxarg_size + execpath_len) * sizeof(u_int32_t)); } else { /* * The '+ 2' is for the null pointers at the end of each of * the arg and env vector sets */ vectp = (u_int32_t *)(destp - (imgp->args->argc + imgp->args->envc + 2) * sizeof(u_int32_t)); } /* * vectp also becomes our initial stack base */ stack_base = vectp; stringp = imgp->args->begin_argv; argc = imgp->args->argc; envc = imgp->args->envc; /* * Copy out strings - arguments and environment. */ copyout(stringp, (void *)destp, ARG_MAX - imgp->args->stringspace); /* * Fill in "ps_strings" struct for ps, w, etc. */ suword32(&arginfo->ps_argvstr, (u_int32_t)(intptr_t)vectp); suword32(&arginfo->ps_nargvstr, argc); /* * Fill in argument portion of vector table. */ for (; argc > 0; --argc) { suword32(vectp++, (u_int32_t)(intptr_t)destp); while (*stringp++ != 0) destp++; destp++; } /* a null vector table pointer separates the argp's from the envp's */ suword32(vectp++, 0); suword32(&arginfo->ps_envstr, (u_int32_t)(intptr_t)vectp); suword32(&arginfo->ps_nenvstr, envc); /* * Fill in environment portion of vector table. */ for (; envc > 0; --envc) { suword32(vectp++, (u_int32_t)(intptr_t)destp); while (*stringp++ != 0) destp++; destp++; } /* end of vector table is a null pointer */ suword32(vectp, 0); return ((register_t *)stack_base); } int freebsd32_kldstat(struct thread *td, struct freebsd32_kldstat_args *uap) { struct kld_file_stat stat; struct kld32_file_stat stat32; int error, version; if ((error = copyin(&uap->stat->version, &version, sizeof(version))) != 0) return (error); if (version != sizeof(struct kld32_file_stat_1) && version != sizeof(struct kld32_file_stat)) return (EINVAL); error = kern_kldstat(td, uap->fileid, &stat); if (error != 0) return (error); bcopy(&stat.name[0], &stat32.name[0], sizeof(stat.name)); CP(stat, stat32, refs); CP(stat, stat32, id); PTROUT_CP(stat, stat32, address); CP(stat, stat32, size); bcopy(&stat.pathname[0], &stat32.pathname[0], sizeof(stat.pathname)); return (copyout(&stat32, uap->stat, version)); } int freebsd32_posix_fallocate(struct thread *td, struct freebsd32_posix_fallocate_args *uap) { int error; error = kern_posix_fallocate(td, uap->fd, PAIR32TO64(off_t, uap->offset), PAIR32TO64(off_t, uap->len)); return (kern_posix_error(td, error)); } int freebsd32_posix_fadvise(struct thread *td, struct freebsd32_posix_fadvise_args *uap) { int error; error = kern_posix_fadvise(td, uap->fd, PAIR32TO64(off_t, uap->offset), PAIR32TO64(off_t, uap->len), uap->advice); return (kern_posix_error(td, error)); } int convert_sigevent32(struct sigevent32 *sig32, struct sigevent *sig) { CP(*sig32, *sig, sigev_notify); switch (sig->sigev_notify) { case SIGEV_NONE: break; case SIGEV_THREAD_ID: CP(*sig32, *sig, sigev_notify_thread_id); /* FALLTHROUGH */ case SIGEV_SIGNAL: CP(*sig32, *sig, sigev_signo); PTRIN_CP(*sig32, *sig, sigev_value.sival_ptr); break; case SIGEV_KEVENT: CP(*sig32, *sig, sigev_notify_kqueue); CP(*sig32, *sig, sigev_notify_kevent_flags); PTRIN_CP(*sig32, *sig, sigev_value.sival_ptr); break; default: return (EINVAL); } return (0); } int freebsd32_procctl(struct thread *td, struct freebsd32_procctl_args *uap) { void *data; union { struct procctl_reaper_status rs; struct procctl_reaper_pids rp; struct procctl_reaper_kill rk; } x; union { struct procctl_reaper_pids32 rp; } x32; int error, error1, flags; switch (uap->com) { case PROC_SPROTECT: case PROC_TRACE_CTL: case PROC_TRAPCAP_CTL: error = copyin(PTRIN(uap->data), &flags, sizeof(flags)); if (error != 0) return (error); data = &flags; break; case PROC_REAP_ACQUIRE: case PROC_REAP_RELEASE: if (uap->data != NULL) return (EINVAL); data = NULL; break; case PROC_REAP_STATUS: data = &x.rs; break; case PROC_REAP_GETPIDS: error = copyin(uap->data, &x32.rp, sizeof(x32.rp)); if (error != 0) return (error); CP(x32.rp, x.rp, rp_count); PTRIN_CP(x32.rp, x.rp, rp_pids); data = &x.rp; break; case PROC_REAP_KILL: error = copyin(uap->data, &x.rk, sizeof(x.rk)); if (error != 0) return (error); data = &x.rk; break; case PROC_TRACE_STATUS: case PROC_TRAPCAP_STATUS: data = &flags; break; default: return (EINVAL); } error = kern_procctl(td, uap->idtype, PAIR32TO64(id_t, uap->id), uap->com, data); switch (uap->com) { case PROC_REAP_STATUS: if (error == 0) error = copyout(&x.rs, uap->data, sizeof(x.rs)); break; case PROC_REAP_KILL: error1 = copyout(&x.rk, uap->data, sizeof(x.rk)); if (error == 0) error = error1; break; case PROC_TRACE_STATUS: case PROC_TRAPCAP_STATUS: if (error == 0) error = copyout(&flags, uap->data, sizeof(flags)); break; } return (error); } int freebsd32_fcntl(struct thread *td, struct freebsd32_fcntl_args *uap) { long tmp; switch (uap->cmd) { /* * Do unsigned conversion for arg when operation * interprets it as flags or pointer. */ case F_SETLK_REMOTE: case F_SETLKW: case F_SETLK: case F_GETLK: case F_SETFD: case F_SETFL: case F_OGETLK: case F_OSETLK: case F_OSETLKW: tmp = (unsigned int)(uap->arg); break; default: tmp = uap->arg; break; } return (kern_fcntl_freebsd(td, uap->fd, uap->cmd, tmp)); } int freebsd32_ppoll(struct thread *td, struct freebsd32_ppoll_args *uap) { struct timespec32 ts32; struct timespec ts, *tsp; sigset_t set, *ssp; int error; if (uap->ts != NULL) { error = copyin(uap->ts, &ts32, sizeof(ts32)); if (error != 0) return (error); CP(ts32, ts, tv_sec); CP(ts32, ts, tv_nsec); tsp = &ts; } else tsp = NULL; if (uap->set != NULL) { error = copyin(uap->set, &set, sizeof(set)); if (error != 0) return (error); ssp = &set; } else ssp = NULL; return (kern_poll(td, uap->fds, uap->nfds, tsp, ssp)); } Index: projects/clang400-import/sys/compat/linux/linux_stats.c =================================================================== --- projects/clang400-import/sys/compat/linux/linux_stats.c (revision 311459) +++ projects/clang400-import/sys/compat/linux/linux_stats.c (revision 311460) @@ -1,736 +1,744 @@ /*- * Copyright (c) 1994-1995 Søren Schmidt * 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 * in this position and unchanged. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. The name of the author may not be used to endorse or promote products * derived from this software without specific prior written permission * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #include "opt_compat.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef COMPAT_LINUX32 #include #include #else #include #include #endif #include #include static void translate_vnhook_major_minor(struct vnode *vp, struct stat *sb) { int major, minor; if (vp->v_type == VCHR && vp->v_rdev != NULL && linux_driver_get_major_minor(devtoname(vp->v_rdev), &major, &minor) == 0) { sb->st_rdev = (major << 8 | minor); } } static int linux_kern_statat(struct thread *td, int flag, int fd, char *path, enum uio_seg pathseg, struct stat *sbp) { return (kern_statat(td, flag, fd, path, pathseg, sbp, translate_vnhook_major_minor)); } static int linux_kern_stat(struct thread *td, char *path, enum uio_seg pathseg, struct stat *sbp) { return (linux_kern_statat(td, 0, AT_FDCWD, path, pathseg, sbp)); } static int linux_kern_lstat(struct thread *td, char *path, enum uio_seg pathseg, struct stat *sbp) { return (linux_kern_statat(td, AT_SYMLINK_NOFOLLOW, AT_FDCWD, path, pathseg, sbp)); } /* * XXX: This was removed from newstat_copyout(), and almost identical * XXX: code was in stat64_copyout(). findcdev() needs to be replaced * XXX: with something that does lookup and locking properly. * XXX: When somebody fixes this: please try to avoid duplicating it. */ #if 0 static void disk_foo(struct somestat *tbuf) { struct cdevsw *cdevsw; struct cdev *dev; /* Lie about disk drives which are character devices * in FreeBSD but block devices under Linux. */ if (S_ISCHR(tbuf.st_mode) && (dev = findcdev(buf->st_rdev)) != NULL) { cdevsw = dev_refthread(dev); if (cdevsw != NULL) { if (cdevsw->d_flags & D_DISK) { tbuf.st_mode &= ~S_IFMT; tbuf.st_mode |= S_IFBLK; /* XXX this may not be quite right */ /* Map major number to 0 */ tbuf.st_dev = minor(buf->st_dev) & 0xf; tbuf.st_rdev = buf->st_rdev & 0xff; } dev_relthread(dev); } } } #endif static void translate_fd_major_minor(struct thread *td, int fd, struct stat *buf) { struct file *fp; struct vnode *vp; cap_rights_t rights; int major, minor; /* * No capability rights required here. */ if ((!S_ISCHR(buf->st_mode) && !S_ISBLK(buf->st_mode)) || fget(td, fd, cap_rights_init(&rights), &fp) != 0) return; vp = fp->f_vnode; if (vp != NULL && vp->v_rdev != NULL && linux_driver_get_major_minor(devtoname(vp->v_rdev), &major, &minor) == 0) { buf->st_rdev = (major << 8 | minor); } else if (fp->f_type == DTYPE_PTS) { struct tty *tp = fp->f_data; /* Convert the numbers for the slave device. */ if (linux_driver_get_major_minor(devtoname(tp->t_dev), &major, &minor) == 0) { buf->st_rdev = (major << 8 | minor); } } fdrop(fp, td); } static int newstat_copyout(struct stat *buf, void *ubuf) { struct l_newstat tbuf; bzero(&tbuf, sizeof(tbuf)); tbuf.st_dev = minor(buf->st_dev) | (major(buf->st_dev) << 8); tbuf.st_ino = buf->st_ino; tbuf.st_mode = buf->st_mode; tbuf.st_nlink = buf->st_nlink; tbuf.st_uid = buf->st_uid; tbuf.st_gid = buf->st_gid; tbuf.st_rdev = buf->st_rdev; tbuf.st_size = buf->st_size; tbuf.st_atim.tv_sec = buf->st_atim.tv_sec; tbuf.st_atim.tv_nsec = buf->st_atim.tv_nsec; tbuf.st_mtim.tv_sec = buf->st_mtim.tv_sec; tbuf.st_mtim.tv_nsec = buf->st_mtim.tv_nsec; tbuf.st_ctim.tv_sec = buf->st_ctim.tv_sec; tbuf.st_ctim.tv_nsec = buf->st_ctim.tv_nsec; tbuf.st_blksize = buf->st_blksize; tbuf.st_blocks = buf->st_blocks; return (copyout(&tbuf, ubuf, sizeof(tbuf))); } int linux_newstat(struct thread *td, struct linux_newstat_args *args) { struct stat buf; char *path; int error; LCONVPATHEXIST(td, args->path, &path); #ifdef DEBUG if (ldebug(newstat)) printf(ARGS(newstat, "%s, *"), path); #endif error = linux_kern_stat(td, path, UIO_SYSSPACE, &buf); LFREEPATH(path); if (error) return (error); return (newstat_copyout(&buf, args->buf)); } int linux_newlstat(struct thread *td, struct linux_newlstat_args *args) { struct stat sb; char *path; int error; LCONVPATHEXIST(td, args->path, &path); #ifdef DEBUG if (ldebug(newlstat)) printf(ARGS(newlstat, "%s, *"), path); #endif error = linux_kern_lstat(td, path, UIO_SYSSPACE, &sb); LFREEPATH(path); if (error) return (error); return (newstat_copyout(&sb, args->buf)); } int linux_newfstat(struct thread *td, struct linux_newfstat_args *args) { struct stat buf; int error; #ifdef DEBUG if (ldebug(newfstat)) printf(ARGS(newfstat, "%d, *"), args->fd); #endif error = kern_fstat(td, args->fd, &buf); translate_fd_major_minor(td, args->fd, &buf); if (!error) error = newstat_copyout(&buf, args->buf); return (error); } #if defined(__i386__) || (defined(__amd64__) && defined(COMPAT_LINUX32)) static int stat_copyout(struct stat *buf, void *ubuf) { struct l_stat lbuf; bzero(&lbuf, sizeof(lbuf)); lbuf.st_dev = buf->st_dev; lbuf.st_ino = buf->st_ino; lbuf.st_mode = buf->st_mode; lbuf.st_nlink = buf->st_nlink; lbuf.st_uid = buf->st_uid; lbuf.st_gid = buf->st_gid; lbuf.st_rdev = buf->st_rdev; if (buf->st_size < (quad_t)1 << 32) lbuf.st_size = buf->st_size; else lbuf.st_size = -2; lbuf.st_atim.tv_sec = buf->st_atim.tv_sec; lbuf.st_atim.tv_nsec = buf->st_atim.tv_nsec; lbuf.st_mtim.tv_sec = buf->st_mtim.tv_sec; lbuf.st_mtim.tv_nsec = buf->st_mtim.tv_nsec; lbuf.st_ctim.tv_sec = buf->st_ctim.tv_sec; lbuf.st_ctim.tv_nsec = buf->st_ctim.tv_nsec; lbuf.st_blksize = buf->st_blksize; lbuf.st_blocks = buf->st_blocks; lbuf.st_flags = buf->st_flags; lbuf.st_gen = buf->st_gen; return (copyout(&lbuf, ubuf, sizeof(lbuf))); } int linux_stat(struct thread *td, struct linux_stat_args *args) { struct stat buf; char *path; int error; LCONVPATHEXIST(td, args->path, &path); #ifdef DEBUG if (ldebug(stat)) printf(ARGS(stat, "%s, *"), path); #endif error = linux_kern_stat(td, path, UIO_SYSSPACE, &buf); if (error) { LFREEPATH(path); return (error); } LFREEPATH(path); return (stat_copyout(&buf, args->up)); } int linux_lstat(struct thread *td, struct linux_lstat_args *args) { struct stat buf; char *path; int error; LCONVPATHEXIST(td, args->path, &path); #ifdef DEBUG if (ldebug(lstat)) printf(ARGS(lstat, "%s, *"), path); #endif error = linux_kern_lstat(td, path, UIO_SYSSPACE, &buf); if (error) { LFREEPATH(path); return (error); } LFREEPATH(path); return (stat_copyout(&buf, args->up)); } #endif /* __i386__ || (__amd64__ && COMPAT_LINUX32) */ struct l_statfs { l_long f_type; l_long f_bsize; l_long f_blocks; l_long f_bfree; l_long f_bavail; l_long f_files; l_long f_ffree; l_fsid_t f_fsid; l_long f_namelen; l_long f_spare[6]; }; #define LINUX_CODA_SUPER_MAGIC 0x73757245L #define LINUX_EXT2_SUPER_MAGIC 0xEF53L #define LINUX_HPFS_SUPER_MAGIC 0xf995e849L #define LINUX_ISOFS_SUPER_MAGIC 0x9660L #define LINUX_MSDOS_SUPER_MAGIC 0x4d44L #define LINUX_NCP_SUPER_MAGIC 0x564cL #define LINUX_NFS_SUPER_MAGIC 0x6969L #define LINUX_NTFS_SUPER_MAGIC 0x5346544EL #define LINUX_PROC_SUPER_MAGIC 0x9fa0L #define LINUX_UFS_SUPER_MAGIC 0x00011954L /* XXX - UFS_MAGIC in Linux */ #define LINUX_DEVFS_SUPER_MAGIC 0x1373L #define LINUX_SHMFS_MAGIC 0x01021994 static long bsd_to_linux_ftype(const char *fstypename) { int i; static struct {const char *bsd_name; long linux_type;} b2l_tbl[] = { {"ufs", LINUX_UFS_SUPER_MAGIC}, {"cd9660", LINUX_ISOFS_SUPER_MAGIC}, {"nfs", LINUX_NFS_SUPER_MAGIC}, {"ext2fs", LINUX_EXT2_SUPER_MAGIC}, {"procfs", LINUX_PROC_SUPER_MAGIC}, {"msdosfs", LINUX_MSDOS_SUPER_MAGIC}, {"ntfs", LINUX_NTFS_SUPER_MAGIC}, {"nwfs", LINUX_NCP_SUPER_MAGIC}, {"hpfs", LINUX_HPFS_SUPER_MAGIC}, {"coda", LINUX_CODA_SUPER_MAGIC}, {"devfs", LINUX_DEVFS_SUPER_MAGIC}, {"tmpfs", LINUX_SHMFS_MAGIC}, {NULL, 0L}}; for (i = 0; b2l_tbl[i].bsd_name != NULL; i++) if (strcmp(b2l_tbl[i].bsd_name, fstypename) == 0) return (b2l_tbl[i].linux_type); return (0L); } static int bsd_to_linux_statfs(struct statfs *bsd_statfs, struct l_statfs *linux_statfs) { #if defined(__i386__) || (defined(__amd64__) && defined(COMPAT_LINUX32)) uint64_t tmp; #define LINUX_HIBITS 0xffffffff00000000ULL tmp = bsd_statfs->f_blocks | bsd_statfs->f_bfree | bsd_statfs->f_files | bsd_statfs->f_bsize; if ((bsd_statfs->f_bavail != -1 && (bsd_statfs->f_bavail & LINUX_HIBITS)) || (bsd_statfs->f_ffree != -1 && (bsd_statfs->f_ffree & LINUX_HIBITS)) || (tmp & LINUX_HIBITS)) return (EOVERFLOW); #undef LINUX_HIBITS #endif linux_statfs->f_type = bsd_to_linux_ftype(bsd_statfs->f_fstypename); linux_statfs->f_bsize = bsd_statfs->f_bsize; linux_statfs->f_blocks = bsd_statfs->f_blocks; linux_statfs->f_bfree = bsd_statfs->f_bfree; linux_statfs->f_bavail = bsd_statfs->f_bavail; linux_statfs->f_ffree = bsd_statfs->f_ffree; linux_statfs->f_files = bsd_statfs->f_files; linux_statfs->f_fsid.val[0] = bsd_statfs->f_fsid.val[0]; linux_statfs->f_fsid.val[1] = bsd_statfs->f_fsid.val[1]; linux_statfs->f_namelen = MAXNAMLEN; return (0); } int linux_statfs(struct thread *td, struct linux_statfs_args *args) { struct l_statfs linux_statfs; - struct statfs bsd_statfs; + struct statfs *bsd_statfs; char *path; int error; LCONVPATHEXIST(td, args->path, &path); #ifdef DEBUG if (ldebug(statfs)) printf(ARGS(statfs, "%s, *"), path); #endif - error = kern_statfs(td, path, UIO_SYSSPACE, &bsd_statfs); + bsd_statfs = malloc(sizeof(struct statfs), M_STATFS, M_WAITOK); + error = kern_statfs(td, path, UIO_SYSSPACE, bsd_statfs); LFREEPATH(path); - if (error) + if (error == 0) + error = bsd_to_linux_statfs(bsd_statfs, &linux_statfs); + free(bsd_statfs, M_STATFS); + if (error != 0) return (error); - error = bsd_to_linux_statfs(&bsd_statfs, &linux_statfs); - if (error) - return (error); return (copyout(&linux_statfs, args->buf, sizeof(linux_statfs))); } #if defined(__i386__) || (defined(__amd64__) && defined(COMPAT_LINUX32)) static void bsd_to_linux_statfs64(struct statfs *bsd_statfs, struct l_statfs64 *linux_statfs) { linux_statfs->f_type = bsd_to_linux_ftype(bsd_statfs->f_fstypename); linux_statfs->f_bsize = bsd_statfs->f_bsize; linux_statfs->f_blocks = bsd_statfs->f_blocks; linux_statfs->f_bfree = bsd_statfs->f_bfree; linux_statfs->f_bavail = bsd_statfs->f_bavail; linux_statfs->f_ffree = bsd_statfs->f_ffree; linux_statfs->f_files = bsd_statfs->f_files; linux_statfs->f_fsid.val[0] = bsd_statfs->f_fsid.val[0]; linux_statfs->f_fsid.val[1] = bsd_statfs->f_fsid.val[1]; linux_statfs->f_namelen = MAXNAMLEN; } int linux_statfs64(struct thread *td, struct linux_statfs64_args *args) { struct l_statfs64 linux_statfs; - struct statfs bsd_statfs; + struct statfs *bsd_statfs; char *path; int error; if (args->bufsize != sizeof(struct l_statfs64)) return EINVAL; LCONVPATHEXIST(td, args->path, &path); #ifdef DEBUG if (ldebug(statfs64)) printf(ARGS(statfs64, "%s, *"), path); #endif - error = kern_statfs(td, path, UIO_SYSSPACE, &bsd_statfs); + bsd_statfs = malloc(sizeof(struct statfs), M_STATFS, M_WAITOK); + error = kern_statfs(td, path, UIO_SYSSPACE, bsd_statfs); LFREEPATH(path); - if (error) + if (error == 0) + bsd_to_linux_statfs64(bsd_statfs, &linux_statfs); + free(bsd_statfs, M_STATFS); + if (error != 0) return (error); - bsd_to_linux_statfs64(&bsd_statfs, &linux_statfs); return (copyout(&linux_statfs, args->buf, sizeof(linux_statfs))); } int linux_fstatfs64(struct thread *td, struct linux_fstatfs64_args *args) { struct l_statfs64 linux_statfs; - struct statfs bsd_statfs; + struct statfs *bsd_statfs; int error; #ifdef DEBUG if (ldebug(fstatfs64)) printf(ARGS(fstatfs64, "%d, *"), args->fd); #endif if (args->bufsize != sizeof(struct l_statfs64)) return (EINVAL); - error = kern_fstatfs(td, args->fd, &bsd_statfs); - if (error) - return error; - bsd_to_linux_statfs64(&bsd_statfs, &linux_statfs); + bsd_statfs = malloc(sizeof(struct statfs), M_STATFS, M_WAITOK); + error = kern_fstatfs(td, args->fd, bsd_statfs); + if (error == 0) + bsd_to_linux_statfs64(bsd_statfs, &linux_statfs); + free(bsd_statfs, M_STATFS); + if (error != 0) + return (error); return (copyout(&linux_statfs, args->buf, sizeof(linux_statfs))); } #endif /* __i386__ || (__amd64__ && COMPAT_LINUX32) */ int linux_fstatfs(struct thread *td, struct linux_fstatfs_args *args) { struct l_statfs linux_statfs; - struct statfs bsd_statfs; + struct statfs *bsd_statfs; int error; #ifdef DEBUG if (ldebug(fstatfs)) printf(ARGS(fstatfs, "%d, *"), args->fd); #endif - error = kern_fstatfs(td, args->fd, &bsd_statfs); - if (error) - return (error); - error = bsd_to_linux_statfs(&bsd_statfs, &linux_statfs); - if (error) + bsd_statfs = malloc(sizeof(struct statfs), M_STATFS, M_WAITOK); + error = kern_fstatfs(td, args->fd, bsd_statfs); + if (error == 0) + error = bsd_to_linux_statfs(bsd_statfs, &linux_statfs); + free(bsd_statfs, M_STATFS); + if (error != 0) return (error); return (copyout(&linux_statfs, args->buf, sizeof(linux_statfs))); } struct l_ustat { l_daddr_t f_tfree; l_ino_t f_tinode; char f_fname[6]; char f_fpack[6]; }; int linux_ustat(struct thread *td, struct linux_ustat_args *args) { #ifdef DEBUG if (ldebug(ustat)) printf(ARGS(ustat, "%ju, *"), (uintmax_t)args->dev); #endif return (EOPNOTSUPP); } #if defined(__i386__) || (defined(__amd64__) && defined(COMPAT_LINUX32)) static int stat64_copyout(struct stat *buf, void *ubuf) { struct l_stat64 lbuf; bzero(&lbuf, sizeof(lbuf)); lbuf.st_dev = minor(buf->st_dev) | (major(buf->st_dev) << 8); lbuf.st_ino = buf->st_ino; lbuf.st_mode = buf->st_mode; lbuf.st_nlink = buf->st_nlink; lbuf.st_uid = buf->st_uid; lbuf.st_gid = buf->st_gid; lbuf.st_rdev = buf->st_rdev; lbuf.st_size = buf->st_size; lbuf.st_atim.tv_sec = buf->st_atim.tv_sec; lbuf.st_atim.tv_nsec = buf->st_atim.tv_nsec; lbuf.st_mtim.tv_sec = buf->st_mtim.tv_sec; lbuf.st_mtim.tv_nsec = buf->st_mtim.tv_nsec; lbuf.st_ctim.tv_sec = buf->st_ctim.tv_sec; lbuf.st_ctim.tv_nsec = buf->st_ctim.tv_nsec; lbuf.st_blksize = buf->st_blksize; lbuf.st_blocks = buf->st_blocks; /* * The __st_ino field makes all the difference. In the Linux kernel * it is conditionally compiled based on STAT64_HAS_BROKEN_ST_INO, * but without the assignment to __st_ino the runtime linker refuses * to mmap(2) any shared libraries. I guess it's broken alright :-) */ lbuf.__st_ino = buf->st_ino; return (copyout(&lbuf, ubuf, sizeof(lbuf))); } int linux_stat64(struct thread *td, struct linux_stat64_args *args) { struct stat buf; char *filename; int error; LCONVPATHEXIST(td, args->filename, &filename); #ifdef DEBUG if (ldebug(stat64)) printf(ARGS(stat64, "%s, *"), filename); #endif error = linux_kern_stat(td, filename, UIO_SYSSPACE, &buf); LFREEPATH(filename); if (error) return (error); return (stat64_copyout(&buf, args->statbuf)); } int linux_lstat64(struct thread *td, struct linux_lstat64_args *args) { struct stat sb; char *filename; int error; LCONVPATHEXIST(td, args->filename, &filename); #ifdef DEBUG if (ldebug(lstat64)) printf(ARGS(lstat64, "%s, *"), args->filename); #endif error = linux_kern_lstat(td, filename, UIO_SYSSPACE, &sb); LFREEPATH(filename); if (error) return (error); return (stat64_copyout(&sb, args->statbuf)); } int linux_fstat64(struct thread *td, struct linux_fstat64_args *args) { struct stat buf; int error; #ifdef DEBUG if (ldebug(fstat64)) printf(ARGS(fstat64, "%d, *"), args->fd); #endif error = kern_fstat(td, args->fd, &buf); translate_fd_major_minor(td, args->fd, &buf); if (!error) error = stat64_copyout(&buf, args->statbuf); return (error); } int linux_fstatat64(struct thread *td, struct linux_fstatat64_args *args) { char *path; int error, dfd, flag; struct stat buf; if (args->flag & ~LINUX_AT_SYMLINK_NOFOLLOW) return (EINVAL); flag = (args->flag & LINUX_AT_SYMLINK_NOFOLLOW) ? AT_SYMLINK_NOFOLLOW : 0; dfd = (args->dfd == LINUX_AT_FDCWD) ? AT_FDCWD : args->dfd; LCONVPATHEXIST_AT(td, args->pathname, &path, dfd); #ifdef DEBUG if (ldebug(fstatat64)) printf(ARGS(fstatat64, "%i, %s, %i"), args->dfd, path, args->flag); #endif error = linux_kern_statat(td, flag, dfd, path, UIO_SYSSPACE, &buf); if (!error) error = stat64_copyout(&buf, args->statbuf); LFREEPATH(path); return (error); } #else /* __amd64__ && !COMPAT_LINUX32 */ int linux_newfstatat(struct thread *td, struct linux_newfstatat_args *args) { char *path; int error, dfd, flag; struct stat buf; if (args->flag & ~LINUX_AT_SYMLINK_NOFOLLOW) return (EINVAL); flag = (args->flag & LINUX_AT_SYMLINK_NOFOLLOW) ? AT_SYMLINK_NOFOLLOW : 0; dfd = (args->dfd == LINUX_AT_FDCWD) ? AT_FDCWD : args->dfd; LCONVPATHEXIST_AT(td, args->pathname, &path, dfd); #ifdef DEBUG if (ldebug(newfstatat)) printf(ARGS(newfstatat, "%i, %s, %i"), args->dfd, path, args->flag); #endif error = linux_kern_statat(td, flag, dfd, path, UIO_SYSSPACE, &buf); if (error == 0) error = newstat_copyout(&buf, args->statbuf); LFREEPATH(path); return (error); } #endif /* __i386__ || (__amd64__ && COMPAT_LINUX32) */ int linux_syncfs(struct thread *td, struct linux_syncfs_args *args) { cap_rights_t rights; struct mount *mp; struct vnode *vp; int error, save; error = fgetvp(td, args->fd, cap_rights_init(&rights, CAP_FSYNC), &vp); if (error != 0) /* * Linux syncfs() returns only EBADF, however fgetvp() * can return EINVAL in case of file descriptor does * not represent a vnode. XXX. */ return (error); mp = vp->v_mount; mtx_lock(&mountlist_mtx); error = vfs_busy(mp, MBF_MNTLSTLOCK); if (error != 0) { /* See comment above. */ mtx_unlock(&mountlist_mtx); goto out; } if ((mp->mnt_flag & MNT_RDONLY) == 0 && vn_start_write(NULL, &mp, V_NOWAIT) == 0) { save = curthread_pflags_set(TDP_SYNCIO); vfs_msync(mp, MNT_NOWAIT); VFS_SYNC(mp, MNT_NOWAIT); curthread_pflags_restore(save); vn_finished_write(mp); } vfs_unbusy(mp); out: vrele(vp); return (error); } Index: projects/clang400-import/sys/compat/svr4/svr4_misc.c =================================================================== --- projects/clang400-import/sys/compat/svr4/svr4_misc.c (revision 311459) +++ projects/clang400-import/sys/compat/svr4/svr4_misc.c (revision 311460) @@ -1,1668 +1,1680 @@ /*- * Copyright (c) 1998 Mark Newton * Copyright (c) 1994 Christos Zoulas * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. The name of the author may not be used to endorse or promote products * derived from this software without specific prior written permission * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ /* * SVR4 compatibility module. * * SVR4 system calls that are implemented differently in BSD are * handled here. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include /* Must come after sys/malloc.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 #if defined(__FreeBSD__) #include #include #endif #if defined(NetBSD) # if defined(UVM) # include # endif #endif #define BSD_DIRENT(cp) ((struct dirent *)(cp)) static int svr4_mknod(struct thread *, register_t *, char *, svr4_mode_t, svr4_dev_t); static __inline clock_t timeval_to_clock_t(struct timeval *); static int svr4_setinfo (pid_t , struct rusage *, int, svr4_siginfo_t *); struct svr4_hrtcntl_args; static int svr4_hrtcntl (struct thread *, struct svr4_hrtcntl_args *, register_t *); static void bsd_statfs_to_svr4_statvfs(const struct statfs *, struct svr4_statvfs *); static void bsd_statfs_to_svr4_statvfs64(const struct statfs *, struct svr4_statvfs64 *); static struct proc *svr4_pfind(pid_t pid); /* BOGUS noop */ #if defined(BOGUS) int svr4_sys_setitimer(td, uap) struct thread *td; struct svr4_sys_setitimer_args *uap; { td->td_retval[0] = 0; return 0; } #endif int svr4_sys_wait(td, uap) struct thread *td; struct svr4_sys_wait_args *uap; { int error, st, sig; error = kern_wait(td, WAIT_ANY, &st, 0, NULL); if (error) return (error); if (WIFSIGNALED(st)) { sig = WTERMSIG(st); if (sig >= 0 && sig < NSIG) st = (st & ~0177) | SVR4_BSD2SVR4_SIG(sig); } else if (WIFSTOPPED(st)) { sig = WSTOPSIG(st); if (sig >= 0 && sig < NSIG) st = (st & ~0xff00) | (SVR4_BSD2SVR4_SIG(sig) << 8); } /* * It looks like wait(2) on svr4/solaris/2.4 returns * the status in retval[1], and the pid on retval[0]. */ td->td_retval[1] = st; if (uap->status) error = copyout(&st, uap->status, sizeof(st)); return (error); } int svr4_sys_execv(td, uap) struct thread *td; struct svr4_sys_execv_args *uap; { struct image_args eargs; struct vmspace *oldvmspace; char *path; int error; CHECKALTEXIST(td, uap->path, &path); error = pre_execve(td, &oldvmspace); if (error != 0) { free(path, M_TEMP); return (error); } error = exec_copyin_args(&eargs, path, UIO_SYSSPACE, uap->argp, NULL); free(path, M_TEMP); if (error == 0) error = kern_execve(td, &eargs, NULL); post_execve(td, error, oldvmspace); return (error); } int svr4_sys_execve(td, uap) struct thread *td; struct svr4_sys_execve_args *uap; { struct image_args eargs; struct vmspace *oldvmspace; char *path; int error; CHECKALTEXIST(td, uap->path, &path); error = pre_execve(td, &oldvmspace); if (error != 0) { free(path, M_TEMP); return (error); } error = exec_copyin_args(&eargs, path, UIO_SYSSPACE, uap->argp, uap->envp); free(path, M_TEMP); if (error == 0) error = kern_execve(td, &eargs, NULL); post_execve(td, error, oldvmspace); return (error); } int svr4_sys_time(td, v) struct thread *td; struct svr4_sys_time_args *v; { struct svr4_sys_time_args *uap = v; int error = 0; struct timeval tv; microtime(&tv); if (uap->t) error = copyout(&tv.tv_sec, uap->t, sizeof(*(uap->t))); td->td_retval[0] = (int) tv.tv_sec; return error; } /* * Read SVR4-style directory entries. We suck them into kernel space so * that they can be massaged before being copied out to user code. * * This code is ported from the Linux emulator: Changes to the VFS interface * between FreeBSD and NetBSD have made it simpler to port it from there than * to adapt the NetBSD version. */ int svr4_sys_getdents64(td, uap) struct thread *td; struct svr4_sys_getdents64_args *uap; { struct dirent *bdp; struct vnode *vp; caddr_t inp, buf; /* BSD-format */ int len, reclen; /* BSD-format */ caddr_t outp; /* SVR4-format */ int resid, svr4reclen=0; /* SVR4-format */ cap_rights_t rights; struct file *fp; struct uio auio; struct iovec aiov; off_t off; struct svr4_dirent64 svr4_dirent; int buflen, error, eofflag, nbytes, justone; u_long *cookies = NULL, *cookiep; int ncookies; DPRINTF(("svr4_sys_getdents64(%d, *, %d)\n", uap->fd, uap->nbytes)); error = getvnode(td, uap->fd, cap_rights_init(&rights, CAP_READ), &fp); if (error != 0) return (error); if ((fp->f_flag & FREAD) == 0) { fdrop(fp, td); return (EBADF); } vp = fp->f_vnode; if (vp->v_type != VDIR) { fdrop(fp, td); return (EINVAL); } nbytes = uap->nbytes; if (nbytes == 1) { nbytes = sizeof (struct svr4_dirent64); justone = 1; } else justone = 0; off = fp->f_offset; #define DIRBLKSIZ 512 /* XXX we used to use ufs's DIRBLKSIZ */ buflen = max(DIRBLKSIZ, nbytes); buflen = min(buflen, MAXBSIZE); buf = malloc(buflen, M_TEMP, M_WAITOK); vn_lock(vp, LK_SHARED | LK_RETRY); again: aiov.iov_base = buf; aiov.iov_len = buflen; auio.uio_iov = &aiov; auio.uio_iovcnt = 1; auio.uio_rw = UIO_READ; auio.uio_segflg = UIO_SYSSPACE; auio.uio_td = td; auio.uio_resid = buflen; auio.uio_offset = off; if (cookies) { free(cookies, M_TEMP); cookies = NULL; } #ifdef MAC error = mac_vnode_check_readdir(td->td_ucred, vp); if (error) goto out; #endif error = VOP_READDIR(vp, &auio, fp->f_cred, &eofflag, &ncookies, &cookies); if (error) { goto out; } inp = buf; outp = (caddr_t) uap->dp; resid = nbytes; if ((len = buflen - auio.uio_resid) <= 0) { goto eof; } cookiep = cookies; if (cookies) { /* * When using cookies, the vfs has the option of reading from * a different offset than that supplied (UFS truncates the * offset to a block boundary to make sure that it never reads * partway through a directory entry, even if the directory * has been compacted). */ while (len > 0 && ncookies > 0 && *cookiep <= off) { bdp = (struct dirent *) inp; len -= bdp->d_reclen; inp += bdp->d_reclen; cookiep++; ncookies--; } } while (len > 0) { if (cookiep && ncookies == 0) break; bdp = (struct dirent *) inp; reclen = bdp->d_reclen; if (reclen & 3) { DPRINTF(("svr4_readdir: reclen=%d\n", reclen)); error = EFAULT; goto out; } if (bdp->d_fileno == 0) { inp += reclen; if (cookiep) { off = *cookiep++; ncookies--; } else off += reclen; len -= reclen; continue; } svr4reclen = SVR4_RECLEN(&svr4_dirent, bdp->d_namlen); if (reclen > len || resid < svr4reclen) { outp++; break; } svr4_dirent.d_ino = (long) bdp->d_fileno; if (justone) { /* * old svr4-style readdir usage. */ svr4_dirent.d_off = (svr4_off_t) svr4reclen; svr4_dirent.d_reclen = (u_short) bdp->d_namlen; } else { svr4_dirent.d_off = (svr4_off_t)(off + reclen); svr4_dirent.d_reclen = (u_short) svr4reclen; } strlcpy(svr4_dirent.d_name, bdp->d_name, sizeof(svr4_dirent.d_name)); if ((error = copyout((caddr_t)&svr4_dirent, outp, svr4reclen))) goto out; inp += reclen; if (cookiep) { off = *cookiep++; ncookies--; } else off += reclen; outp += svr4reclen; resid -= svr4reclen; len -= reclen; if (justone) break; } if (outp == (caddr_t) uap->dp) goto again; fp->f_offset = off; if (justone) nbytes = resid + svr4reclen; eof: td->td_retval[0] = nbytes - resid; out: VOP_UNLOCK(vp, 0); fdrop(fp, td); if (cookies) free(cookies, M_TEMP); free(buf, M_TEMP); return error; } int svr4_sys_getdents(td, uap) struct thread *td; struct svr4_sys_getdents_args *uap; { struct dirent *bdp; struct vnode *vp; caddr_t inp, buf; /* BSD-format */ int len, reclen; /* BSD-format */ caddr_t outp; /* SVR4-format */ int resid, svr4_reclen; /* SVR4-format */ cap_rights_t rights; struct file *fp; struct uio auio; struct iovec aiov; struct svr4_dirent idb; off_t off; /* true file offset */ int buflen, error, eofflag; u_long *cookiebuf = NULL, *cookie; int ncookies = 0, *retval = td->td_retval; if (uap->nbytes < 0) return (EINVAL); error = getvnode(td, uap->fd, cap_rights_init(&rights, CAP_READ), &fp); if (error != 0) return (error); if ((fp->f_flag & FREAD) == 0) { fdrop(fp, td); return (EBADF); } vp = fp->f_vnode; if (vp->v_type != VDIR) { fdrop(fp, td); return (EINVAL); } buflen = min(MAXBSIZE, uap->nbytes); buf = malloc(buflen, M_TEMP, M_WAITOK); vn_lock(vp, LK_SHARED | LK_RETRY); off = fp->f_offset; again: aiov.iov_base = buf; aiov.iov_len = buflen; auio.uio_iov = &aiov; auio.uio_iovcnt = 1; auio.uio_rw = UIO_READ; auio.uio_segflg = UIO_SYSSPACE; auio.uio_td = td; auio.uio_resid = buflen; auio.uio_offset = off; #ifdef MAC error = mac_vnode_check_readdir(td->td_ucred, vp); if (error) goto out; #endif /* * First we read into the malloc'ed buffer, then * we massage it into user space, one record at a time. */ error = VOP_READDIR(vp, &auio, fp->f_cred, &eofflag, &ncookies, &cookiebuf); if (error) { goto out; } inp = buf; outp = uap->buf; resid = uap->nbytes; if ((len = buflen - auio.uio_resid) == 0) goto eof; for (cookie = cookiebuf; len > 0; len -= reclen) { bdp = (struct dirent *)inp; reclen = bdp->d_reclen; if (reclen & 3) panic("svr4_sys_getdents64: bad reclen"); if (cookie) off = *cookie++; /* each entry points to the next */ else off += reclen; if ((off >> 32) != 0) { uprintf("svr4_sys_getdents64: dir offset too large for emulated program"); error = EINVAL; goto out; } if (bdp->d_fileno == 0) { inp += reclen; /* it is a hole; squish it out */ continue; } svr4_reclen = SVR4_RECLEN(&idb, bdp->d_namlen); if (reclen > len || resid < svr4_reclen) { /* entry too big for buffer, so just stop */ outp++; break; } /* * Massage in place to make a SVR4-shaped dirent (otherwise * we have to worry about touching user memory outside of * the copyout() call). */ idb.d_ino = (svr4_ino_t)bdp->d_fileno; idb.d_off = (svr4_off_t)off; idb.d_reclen = (u_short)svr4_reclen; strlcpy(idb.d_name, bdp->d_name, sizeof(idb.d_name)); if ((error = copyout((caddr_t)&idb, outp, svr4_reclen))) goto out; /* advance past this real entry */ inp += reclen; /* advance output past SVR4-shaped entry */ outp += svr4_reclen; resid -= svr4_reclen; } /* if we squished out the whole block, try again */ if (outp == uap->buf) goto again; fp->f_offset = off; /* update the vnode offset */ eof: *retval = uap->nbytes - resid; out: VOP_UNLOCK(vp, 0); fdrop(fp, td); if (cookiebuf) free(cookiebuf, M_TEMP); free(buf, M_TEMP); return error; } int svr4_sys_mmap(td, uap) struct thread *td; struct svr4_sys_mmap_args *uap; { struct mmap_args mm; int *retval; retval = td->td_retval; #define _MAP_NEW 0x80000000 /* * Verify the arguments. */ if (uap->prot & ~(PROT_READ | PROT_WRITE | PROT_EXEC)) return EINVAL; /* XXX still needed? */ if (uap->len == 0) return EINVAL; mm.prot = uap->prot; mm.len = uap->len; mm.flags = uap->flags & ~_MAP_NEW; mm.fd = uap->fd; mm.addr = uap->addr; mm.pos = uap->pos; return sys_mmap(td, &mm); } int svr4_sys_mmap64(td, uap) struct thread *td; struct svr4_sys_mmap64_args *uap; { struct mmap_args mm; void *rp; #define _MAP_NEW 0x80000000 /* * Verify the arguments. */ if (uap->prot & ~(PROT_READ | PROT_WRITE | PROT_EXEC)) return EINVAL; /* XXX still needed? */ if (uap->len == 0) return EINVAL; mm.prot = uap->prot; mm.len = uap->len; mm.flags = uap->flags & ~_MAP_NEW; mm.fd = uap->fd; mm.addr = uap->addr; mm.pos = uap->pos; rp = (void *) round_page((vm_offset_t)(td->td_proc->p_vmspace->vm_daddr + maxdsiz)); if ((mm.flags & MAP_FIXED) == 0 && mm.addr != 0 && (void *)mm.addr < rp) mm.addr = rp; return sys_mmap(td, &mm); } int svr4_sys_fchroot(td, uap) struct thread *td; struct svr4_sys_fchroot_args *uap; { cap_rights_t rights; struct vnode *vp; struct file *fp; int error; if ((error = priv_check(td, PRIV_VFS_FCHROOT)) != 0) return error; /* XXX: we have the chroot priv... what cap might we need? all? */ if ((error = getvnode(td, uap->fd, cap_rights_init(&rights), &fp)) != 0) return error; vp = fp->f_vnode; VREF(vp); fdrop(fp, td); vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); error = change_dir(vp, td); if (error) goto fail; #ifdef MAC error = mac_vnode_check_chroot(td->td_ucred, vp); if (error) goto fail; #endif VOP_UNLOCK(vp, 0); error = pwd_chroot(td, vp); vrele(vp); return (error); fail: vput(vp); return (error); } static int svr4_mknod(td, retval, path, mode, dev) struct thread *td; register_t *retval; char *path; svr4_mode_t mode; svr4_dev_t dev; { char *newpath; int error; CHECKALTEXIST(td, path, &newpath); if (S_ISFIFO(mode)) { error = kern_mkfifoat(td, AT_FDCWD, newpath, UIO_SYSSPACE, mode); } else { error = kern_mknodat(td, AT_FDCWD, newpath, UIO_SYSSPACE, mode, dev); } free(newpath, M_TEMP); return (error); } int svr4_sys_mknod(td, uap) struct thread *td; struct svr4_sys_mknod_args *uap; { int *retval = td->td_retval; return svr4_mknod(td, retval, uap->path, uap->mode, (svr4_dev_t)svr4_to_bsd_odev_t(uap->dev)); } int svr4_sys_xmknod(td, uap) struct thread *td; struct svr4_sys_xmknod_args *uap; { int *retval = td->td_retval; return svr4_mknod(td, retval, uap->path, uap->mode, (svr4_dev_t)svr4_to_bsd_dev_t(uap->dev)); } int svr4_sys_vhangup(td, uap) struct thread *td; struct svr4_sys_vhangup_args *uap; { return 0; } int svr4_sys_sysconfig(td, uap) struct thread *td; struct svr4_sys_sysconfig_args *uap; { int *retval; retval = &(td->td_retval[0]); switch (uap->name) { case SVR4_CONFIG_NGROUPS: *retval = ngroups_max; break; case SVR4_CONFIG_CHILD_MAX: *retval = maxproc; break; case SVR4_CONFIG_OPEN_FILES: *retval = maxfiles; break; case SVR4_CONFIG_POSIX_VER: *retval = 198808; break; case SVR4_CONFIG_PAGESIZE: *retval = PAGE_SIZE; break; case SVR4_CONFIG_CLK_TCK: *retval = 60; /* should this be `hz', ie. 100? */ break; case SVR4_CONFIG_XOPEN_VER: *retval = 2; /* XXX: What should that be? */ break; case SVR4_CONFIG_PROF_TCK: *retval = 60; /* XXX: What should that be? */ break; case SVR4_CONFIG_NPROC_CONF: *retval = 1; /* Only one processor for now */ break; case SVR4_CONFIG_NPROC_ONLN: *retval = 1; /* And it better be online */ break; case SVR4_CONFIG_AIO_LISTIO_MAX: case SVR4_CONFIG_AIO_MAX: case SVR4_CONFIG_AIO_PRIO_DELTA_MAX: *retval = 0; /* No aio support */ break; case SVR4_CONFIG_DELAYTIMER_MAX: *retval = 0; /* No delaytimer support */ break; case SVR4_CONFIG_MQ_OPEN_MAX: *retval = msginfo.msgmni; break; case SVR4_CONFIG_MQ_PRIO_MAX: *retval = 0; /* XXX: Don't know */ break; case SVR4_CONFIG_RTSIG_MAX: *retval = 0; break; case SVR4_CONFIG_SEM_NSEMS_MAX: *retval = seminfo.semmni; break; case SVR4_CONFIG_SEM_VALUE_MAX: *retval = seminfo.semvmx; break; case SVR4_CONFIG_SIGQUEUE_MAX: *retval = 0; /* XXX: Don't know */ break; case SVR4_CONFIG_SIGRT_MIN: case SVR4_CONFIG_SIGRT_MAX: *retval = 0; /* No real time signals */ break; case SVR4_CONFIG_TIMER_MAX: *retval = 3; /* XXX: real, virtual, profiling */ break; #if defined(NOTYET) case SVR4_CONFIG_PHYS_PAGES: #if defined(UVM) *retval = uvmexp.free; /* XXX: free instead of total */ #else *retval = vm_cnt.v_free_count; /* XXX: free instead of total */ #endif break; case SVR4_CONFIG_AVPHYS_PAGES: #if defined(UVM) *retval = uvmexp.active; /* XXX: active instead of avg */ #else *retval = vm_cnt.v_active_count;/* XXX: active instead of avg */ #endif break; #endif /* NOTYET */ case SVR4_CONFIG_COHERENCY: *retval = 0; /* XXX */ break; case SVR4_CONFIG_SPLIT_CACHE: *retval = 0; /* XXX */ break; case SVR4_CONFIG_ICACHESZ: *retval = 256; /* XXX */ break; case SVR4_CONFIG_DCACHESZ: *retval = 256; /* XXX */ break; case SVR4_CONFIG_ICACHELINESZ: *retval = 64; /* XXX */ break; case SVR4_CONFIG_DCACHELINESZ: *retval = 64; /* XXX */ break; case SVR4_CONFIG_ICACHEBLKSZ: *retval = 64; /* XXX */ break; case SVR4_CONFIG_DCACHEBLKSZ: *retval = 64; /* XXX */ break; case SVR4_CONFIG_DCACHETBLKSZ: *retval = 64; /* XXX */ break; case SVR4_CONFIG_ICACHE_ASSOC: *retval = 1; /* XXX */ break; case SVR4_CONFIG_DCACHE_ASSOC: *retval = 1; /* XXX */ break; case SVR4_CONFIG_MAXPID: *retval = PID_MAX; break; case SVR4_CONFIG_STACK_PROT: *retval = PROT_READ|PROT_WRITE|PROT_EXEC; break; default: return EINVAL; } return 0; } /* ARGSUSED */ int svr4_sys_break(td, uap) struct thread *td; struct svr4_sys_break_args *uap; { struct obreak_args ap; ap.nsize = uap->nsize; return (sys_obreak(td, &ap)); } static __inline clock_t timeval_to_clock_t(tv) struct timeval *tv; { return tv->tv_sec * hz + tv->tv_usec / (1000000 / hz); } int svr4_sys_times(td, uap) struct thread *td; struct svr4_sys_times_args *uap; { struct timeval tv, utime, stime, cutime, cstime; struct tms tms; struct proc *p; int error; p = td->td_proc; PROC_LOCK(p); PROC_STATLOCK(p); calcru(p, &utime, &stime); PROC_STATUNLOCK(p); calccru(p, &cutime, &cstime); PROC_UNLOCK(p); tms.tms_utime = timeval_to_clock_t(&utime); tms.tms_stime = timeval_to_clock_t(&stime); tms.tms_cutime = timeval_to_clock_t(&cutime); tms.tms_cstime = timeval_to_clock_t(&cstime); error = copyout(&tms, uap->tp, sizeof(tms)); if (error) return (error); microtime(&tv); td->td_retval[0] = (int)timeval_to_clock_t(&tv); return (0); } int svr4_sys_ulimit(td, uap) struct thread *td; struct svr4_sys_ulimit_args *uap; { int *retval = td->td_retval; int error; switch (uap->cmd) { case SVR4_GFILLIM: *retval = lim_cur(td, RLIMIT_FSIZE) / 512; if (*retval == -1) *retval = 0x7fffffff; return 0; case SVR4_SFILLIM: { struct rlimit krl; krl.rlim_cur = uap->newlimit * 512; krl.rlim_max = lim_max(td, RLIMIT_FSIZE); error = kern_setrlimit(td, RLIMIT_FSIZE, &krl); if (error) return error; *retval = lim_cur(td, RLIMIT_FSIZE); if (*retval == -1) *retval = 0x7fffffff; return 0; } case SVR4_GMEMLIM: { struct vmspace *vm = td->td_proc->p_vmspace; register_t r; r = lim_cur(td, RLIMIT_DATA); if (r == -1) r = 0x7fffffff; r += (long) vm->vm_daddr; if (r < 0) r = 0x7fffffff; *retval = r; return 0; } case SVR4_GDESLIM: *retval = lim_cur(td, RLIMIT_NOFILE); if (*retval == -1) *retval = 0x7fffffff; return 0; default: return EINVAL; } } static struct proc * svr4_pfind(pid) pid_t pid; { struct proc *p; /* look in the live processes */ if ((p = pfind(pid)) == NULL) /* look in the zombies */ p = zpfind(pid); return p; } int svr4_sys_pgrpsys(td, uap) struct thread *td; struct svr4_sys_pgrpsys_args *uap; { int *retval = td->td_retval; struct proc *p = td->td_proc; switch (uap->cmd) { case 1: /* setpgrp() */ /* * SVR4 setpgrp() (which takes no arguments) has the * semantics that the session ID is also created anew, so * in almost every sense, setpgrp() is identical to * setsid() for SVR4. (Under BSD, the difference is that * a setpgid(0,0) will not create a new session.) */ sys_setsid(td, NULL); /*FALLTHROUGH*/ case 0: /* getpgrp() */ PROC_LOCK(p); *retval = p->p_pgrp->pg_id; PROC_UNLOCK(p); return 0; case 2: /* getsid(pid) */ if (uap->pid == 0) PROC_LOCK(p); else if ((p = svr4_pfind(uap->pid)) == NULL) return ESRCH; /* * This has already been initialized to the pid of * the session leader. */ *retval = (register_t) p->p_session->s_sid; PROC_UNLOCK(p); return 0; case 3: /* setsid() */ return sys_setsid(td, NULL); case 4: /* getpgid(pid) */ if (uap->pid == 0) PROC_LOCK(p); else if ((p = svr4_pfind(uap->pid)) == NULL) return ESRCH; *retval = (int) p->p_pgrp->pg_id; PROC_UNLOCK(p); return 0; case 5: /* setpgid(pid, pgid); */ { struct setpgid_args sa; sa.pid = uap->pid; sa.pgid = uap->pgid; return sys_setpgid(td, &sa); } default: return EINVAL; } } struct svr4_hrtcntl_args { int cmd; int fun; int clk; svr4_hrt_interval_t * iv; svr4_hrt_time_t * ti; }; static int svr4_hrtcntl(td, uap, retval) struct thread *td; struct svr4_hrtcntl_args *uap; register_t *retval; { switch (uap->fun) { case SVR4_HRT_CNTL_RES: DPRINTF(("htrcntl(RES)\n")); *retval = SVR4_HRT_USEC; return 0; case SVR4_HRT_CNTL_TOFD: DPRINTF(("htrcntl(TOFD)\n")); { struct timeval tv; svr4_hrt_time_t t; if (uap->clk != SVR4_HRT_CLK_STD) { DPRINTF(("clk == %d\n", uap->clk)); return EINVAL; } if (uap->ti == NULL) { DPRINTF(("ti NULL\n")); return EINVAL; } microtime(&tv); t.h_sec = tv.tv_sec; t.h_rem = tv.tv_usec; t.h_res = SVR4_HRT_USEC; return copyout(&t, uap->ti, sizeof(t)); } case SVR4_HRT_CNTL_START: DPRINTF(("htrcntl(START)\n")); return ENOSYS; case SVR4_HRT_CNTL_GET: DPRINTF(("htrcntl(GET)\n")); return ENOSYS; default: DPRINTF(("Bad htrcntl command %d\n", uap->fun)); return ENOSYS; } } int svr4_sys_hrtsys(td, uap) struct thread *td; struct svr4_sys_hrtsys_args *uap; { int *retval = td->td_retval; switch (uap->cmd) { case SVR4_HRT_CNTL: return svr4_hrtcntl(td, (struct svr4_hrtcntl_args *) uap, retval); case SVR4_HRT_ALRM: DPRINTF(("hrtalarm\n")); return ENOSYS; case SVR4_HRT_SLP: DPRINTF(("hrtsleep\n")); return ENOSYS; case SVR4_HRT_CAN: DPRINTF(("hrtcancel\n")); return ENOSYS; default: DPRINTF(("Bad hrtsys command %d\n", uap->cmd)); return EINVAL; } } static int svr4_setinfo(pid, ru, st, s) pid_t pid; struct rusage *ru; int st; svr4_siginfo_t *s; { svr4_siginfo_t i; int sig; memset(&i, 0, sizeof(i)); i.svr4_si_signo = SVR4_SIGCHLD; i.svr4_si_errno = 0; /* XXX? */ i.svr4_si_pid = pid; if (ru) { i.svr4_si_stime = ru->ru_stime.tv_sec; i.svr4_si_utime = ru->ru_utime.tv_sec; } if (WIFEXITED(st)) { i.svr4_si_status = WEXITSTATUS(st); i.svr4_si_code = SVR4_CLD_EXITED; } else if (WIFSTOPPED(st)) { sig = WSTOPSIG(st); if (sig >= 0 && sig < NSIG) i.svr4_si_status = SVR4_BSD2SVR4_SIG(sig); if (i.svr4_si_status == SVR4_SIGCONT) i.svr4_si_code = SVR4_CLD_CONTINUED; else i.svr4_si_code = SVR4_CLD_STOPPED; } else { sig = WTERMSIG(st); if (sig >= 0 && sig < NSIG) i.svr4_si_status = SVR4_BSD2SVR4_SIG(sig); if (WCOREDUMP(st)) i.svr4_si_code = SVR4_CLD_DUMPED; else i.svr4_si_code = SVR4_CLD_KILLED; } DPRINTF(("siginfo [pid %ld signo %d code %d errno %d status %d]\n", i.svr4_si_pid, i.svr4_si_signo, i.svr4_si_code, i.svr4_si_errno, i.svr4_si_status)); return copyout(&i, s, sizeof(i)); } int svr4_sys_waitsys(td, uap) struct thread *td; struct svr4_sys_waitsys_args *uap; { struct rusage ru; pid_t pid; int nfound, status; int error, *retval = td->td_retval; struct proc *p, *q; DPRINTF(("waitsys(%d, %d, %p, %x)\n", uap->grp, uap->id, uap->info, uap->options)); q = td->td_proc; switch (uap->grp) { case SVR4_P_PID: pid = uap->id; break; case SVR4_P_PGID: PROC_LOCK(q); pid = -q->p_pgid; PROC_UNLOCK(q); break; case SVR4_P_ALL: pid = WAIT_ANY; break; default: return EINVAL; } /* Hand off the easy cases to kern_wait(). */ if (!(uap->options & (SVR4_WNOWAIT)) && (uap->options & (SVR4_WEXITED | SVR4_WTRAPPED))) { int options; options = 0; if (uap->options & SVR4_WSTOPPED) options |= WUNTRACED; if (uap->options & SVR4_WCONTINUED) options |= WCONTINUED; if (uap->options & SVR4_WNOHANG) options |= WNOHANG; error = kern_wait(td, pid, &status, options, &ru); if (error) return (error); if (uap->options & SVR4_WNOHANG && *retval == 0) error = svr4_setinfo(*retval, NULL, 0, uap->info); else error = svr4_setinfo(*retval, &ru, status, uap->info); *retval = 0; return (error); } /* * Ok, handle the weird cases. Either WNOWAIT is set (meaning we * just want to see if there is a process to harvest, we don't * want to actually harvest it), or WEXIT and WTRAPPED are clear * meaning we want to ignore zombies. Either way, we don't have * to handle harvesting zombies here. We do have to duplicate the * other portions of kern_wait() though, especially for WCONTINUED * and WSTOPPED. */ loop: nfound = 0; sx_slock(&proctree_lock); LIST_FOREACH(p, &q->p_children, p_sibling) { PROC_LOCK(p); if (pid != WAIT_ANY && p->p_pid != pid && p->p_pgid != -pid) { PROC_UNLOCK(p); DPRINTF(("pid %d pgid %d != %d\n", p->p_pid, p->p_pgid, pid)); continue; } if (p_canwait(td, p)) { PROC_UNLOCK(p); continue; } nfound++; PROC_SLOCK(p); /* * See if we have a zombie. If so, WNOWAIT should be set, * as otherwise we should have called kern_wait() up above. */ if ((p->p_state == PRS_ZOMBIE) && ((uap->options & (SVR4_WEXITED|SVR4_WTRAPPED)))) { PROC_SUNLOCK(p); KASSERT(uap->options & SVR4_WNOWAIT, ("WNOWAIT is clear")); /* Found a zombie, so cache info in local variables. */ pid = p->p_pid; status = KW_EXITCODE(p->p_xexit, p->p_xsig); ru = p->p_ru; PROC_STATLOCK(p); calcru(p, &ru.ru_utime, &ru.ru_stime); PROC_STATUNLOCK(p); PROC_UNLOCK(p); sx_sunlock(&proctree_lock); /* Copy the info out to userland. */ *retval = 0; DPRINTF(("found %d\n", pid)); return (svr4_setinfo(pid, &ru, status, uap->info)); } /* * See if we have a stopped or continued process. * XXX: This duplicates the same code in kern_wait(). */ if ((p->p_flag & P_STOPPED_SIG) && (p->p_suspcount == p->p_numthreads) && (p->p_flag & P_WAITED) == 0 && (p->p_flag & P_TRACED || uap->options & SVR4_WSTOPPED)) { PROC_SUNLOCK(p); if (((uap->options & SVR4_WNOWAIT)) == 0) p->p_flag |= P_WAITED; sx_sunlock(&proctree_lock); pid = p->p_pid; status = W_STOPCODE(p->p_xsig); ru = p->p_ru; PROC_STATLOCK(p); calcru(p, &ru.ru_utime, &ru.ru_stime); PROC_STATUNLOCK(p); PROC_UNLOCK(p); if (((uap->options & SVR4_WNOWAIT)) == 0) { PROC_LOCK(q); sigqueue_take(p->p_ksi); PROC_UNLOCK(q); } *retval = 0; DPRINTF(("jobcontrol %d\n", pid)); return (svr4_setinfo(pid, &ru, status, uap->info)); } PROC_SUNLOCK(p); if (uap->options & SVR4_WCONTINUED && (p->p_flag & P_CONTINUED)) { sx_sunlock(&proctree_lock); if (((uap->options & SVR4_WNOWAIT)) == 0) p->p_flag &= ~P_CONTINUED; pid = p->p_pid; ru = p->p_ru; status = SIGCONT; PROC_STATLOCK(p); calcru(p, &ru.ru_utime, &ru.ru_stime); PROC_STATUNLOCK(p); PROC_UNLOCK(p); if (((uap->options & SVR4_WNOWAIT)) == 0) { PROC_LOCK(q); sigqueue_take(p->p_ksi); PROC_UNLOCK(q); } *retval = 0; DPRINTF(("jobcontrol %d\n", pid)); return (svr4_setinfo(pid, &ru, status, uap->info)); } PROC_UNLOCK(p); } if (nfound == 0) { sx_sunlock(&proctree_lock); return (ECHILD); } if (uap->options & SVR4_WNOHANG) { sx_sunlock(&proctree_lock); *retval = 0; return (svr4_setinfo(0, NULL, 0, uap->info)); } PROC_LOCK(q); sx_sunlock(&proctree_lock); if (q->p_flag & P_STATCHILD) { q->p_flag &= ~P_STATCHILD; error = 0; } else error = msleep(q, &q->p_mtx, PWAIT | PCATCH, "svr4_wait", 0); PROC_UNLOCK(q); if (error) return error; goto loop; } static void bsd_statfs_to_svr4_statvfs(bfs, sfs) const struct statfs *bfs; struct svr4_statvfs *sfs; { sfs->f_bsize = bfs->f_iosize; /* XXX */ sfs->f_frsize = bfs->f_bsize; sfs->f_blocks = bfs->f_blocks; sfs->f_bfree = bfs->f_bfree; sfs->f_bavail = bfs->f_bavail; sfs->f_files = bfs->f_files; sfs->f_ffree = bfs->f_ffree; sfs->f_favail = bfs->f_ffree; sfs->f_fsid = bfs->f_fsid.val[0]; memcpy(sfs->f_basetype, bfs->f_fstypename, sizeof(sfs->f_basetype)); sfs->f_flag = 0; if (bfs->f_flags & MNT_RDONLY) sfs->f_flag |= SVR4_ST_RDONLY; if (bfs->f_flags & MNT_NOSUID) sfs->f_flag |= SVR4_ST_NOSUID; sfs->f_namemax = MAXNAMLEN; memcpy(sfs->f_fstr, bfs->f_fstypename, sizeof(sfs->f_fstr)); /* XXX */ memset(sfs->f_filler, 0, sizeof(sfs->f_filler)); } static void bsd_statfs_to_svr4_statvfs64(bfs, sfs) const struct statfs *bfs; struct svr4_statvfs64 *sfs; { sfs->f_bsize = bfs->f_iosize; /* XXX */ sfs->f_frsize = bfs->f_bsize; sfs->f_blocks = bfs->f_blocks; sfs->f_bfree = bfs->f_bfree; sfs->f_bavail = bfs->f_bavail; sfs->f_files = bfs->f_files; sfs->f_ffree = bfs->f_ffree; sfs->f_favail = bfs->f_ffree; sfs->f_fsid = bfs->f_fsid.val[0]; memcpy(sfs->f_basetype, bfs->f_fstypename, sizeof(sfs->f_basetype)); sfs->f_flag = 0; if (bfs->f_flags & MNT_RDONLY) sfs->f_flag |= SVR4_ST_RDONLY; if (bfs->f_flags & MNT_NOSUID) sfs->f_flag |= SVR4_ST_NOSUID; sfs->f_namemax = MAXNAMLEN; memcpy(sfs->f_fstr, bfs->f_fstypename, sizeof(sfs->f_fstr)); /* XXX */ memset(sfs->f_filler, 0, sizeof(sfs->f_filler)); } int svr4_sys_statvfs(td, uap) struct thread *td; struct svr4_sys_statvfs_args *uap; { struct svr4_statvfs sfs; - struct statfs bfs; + struct statfs *bfs; char *path; int error; CHECKALTEXIST(td, uap->path, &path); - error = kern_statfs(td, path, UIO_SYSSPACE, &bfs); + bfs = malloc(sizeof(struct statfs), M_STATFS, M_WAITOK); + error = kern_statfs(td, path, UIO_SYSSPACE, bfs); free(path, M_TEMP); - if (error) + if (error == 0) + bsd_statfs_to_svr4_statvfs(bfs, &sfs); + free(bfs, M_STATFS); + if (error != 0) return (error); - bsd_statfs_to_svr4_statvfs(&bfs, &sfs); return copyout(&sfs, uap->fs, sizeof(sfs)); } int svr4_sys_fstatvfs(td, uap) struct thread *td; struct svr4_sys_fstatvfs_args *uap; { struct svr4_statvfs sfs; - struct statfs bfs; + struct statfs *bfs; int error; - error = kern_fstatfs(td, uap->fd, &bfs); - if (error) + bfs = malloc(sizeof(struct statfs), M_STATFS, M_WAITOK); + error = kern_fstatfs(td, uap->fd, bfs); + if (error == 0) + bsd_statfs_to_svr4_statvfs(bfs, &sfs); + free(bfs, M_STATFS); + if (error != 0) return (error); - bsd_statfs_to_svr4_statvfs(&bfs, &sfs); return copyout(&sfs, uap->fs, sizeof(sfs)); } int svr4_sys_statvfs64(td, uap) struct thread *td; struct svr4_sys_statvfs64_args *uap; { struct svr4_statvfs64 sfs; - struct statfs bfs; + struct statfs *bfs; char *path; int error; CHECKALTEXIST(td, uap->path, &path); - error = kern_statfs(td, path, UIO_SYSSPACE, &bfs); + bfs = malloc(sizeof(struct statfs), M_STATFS, M_WAITOK); + error = kern_statfs(td, path, UIO_SYSSPACE, bfs); free(path, M_TEMP); - if (error) + if (error == 0) + bsd_statfs_to_svr4_statvfs64(bfs, &sfs); + free(bfs, M_STATFS); + if (error != 0) return (error); - bsd_statfs_to_svr4_statvfs64(&bfs, &sfs); return copyout(&sfs, uap->fs, sizeof(sfs)); } int svr4_sys_fstatvfs64(td, uap) struct thread *td; struct svr4_sys_fstatvfs64_args *uap; { struct svr4_statvfs64 sfs; - struct statfs bfs; + struct statfs *bfs; int error; - error = kern_fstatfs(td, uap->fd, &bfs); - if (error) + bfs = malloc(sizeof(struct statfs), M_STATFS, M_WAITOK); + error = kern_fstatfs(td, uap->fd, bfs); + if (error == 0) + bsd_statfs_to_svr4_statvfs64(bfs, &sfs); + free(bfs, M_STATFS); + if (error != 0) return (error); - bsd_statfs_to_svr4_statvfs64(&bfs, &sfs); return copyout(&sfs, uap->fs, sizeof(sfs)); } int svr4_sys_alarm(td, uap) struct thread *td; struct svr4_sys_alarm_args *uap; { struct itimerval itv, oitv; int error; timevalclear(&itv.it_interval); itv.it_value.tv_sec = uap->sec; itv.it_value.tv_usec = 0; error = kern_setitimer(td, ITIMER_REAL, &itv, &oitv); if (error) return (error); if (oitv.it_value.tv_usec != 0) oitv.it_value.tv_sec++; td->td_retval[0] = oitv.it_value.tv_sec; return (0); } int svr4_sys_gettimeofday(td, uap) struct thread *td; struct svr4_sys_gettimeofday_args *uap; { if (uap->tp) { struct timeval atv; microtime(&atv); return copyout(&atv, uap->tp, sizeof (atv)); } return 0; } int svr4_sys_facl(td, uap) struct thread *td; struct svr4_sys_facl_args *uap; { int *retval; retval = td->td_retval; *retval = 0; switch (uap->cmd) { case SVR4_SYS_SETACL: /* We don't support acls on any filesystem */ return ENOSYS; case SVR4_SYS_GETACL: return copyout(retval, &uap->num, sizeof(uap->num)); case SVR4_SYS_GETACLCNT: return 0; default: return EINVAL; } } int svr4_sys_acl(td, uap) struct thread *td; struct svr4_sys_acl_args *uap; { /* XXX: for now the same */ return svr4_sys_facl(td, (struct svr4_sys_facl_args *)uap); } int svr4_sys_auditsys(td, uap) struct thread *td; struct svr4_sys_auditsys_args *uap; { /* * XXX: Big brother is *not* watching. */ return 0; } int svr4_sys_memcntl(td, uap) struct thread *td; struct svr4_sys_memcntl_args *uap; { switch (uap->cmd) { case SVR4_MC_SYNC: { struct msync_args msa; msa.addr = uap->addr; msa.len = uap->len; msa.flags = (int)uap->arg; return sys_msync(td, &msa); } case SVR4_MC_ADVISE: { struct madvise_args maa; maa.addr = uap->addr; maa.len = uap->len; maa.behav = (int)uap->arg; return sys_madvise(td, &maa); } case SVR4_MC_LOCK: case SVR4_MC_UNLOCK: case SVR4_MC_LOCKAS: case SVR4_MC_UNLOCKAS: return EOPNOTSUPP; default: return ENOSYS; } } int svr4_sys_nice(td, uap) struct thread *td; struct svr4_sys_nice_args *uap; { struct setpriority_args ap; int error; ap.which = PRIO_PROCESS; ap.who = 0; ap.prio = uap->prio; if ((error = sys_setpriority(td, &ap)) != 0) return error; /* the cast is stupid, but the structures are the same */ if ((error = sys_getpriority(td, (struct getpriority_args *)&ap)) != 0) return error; return 0; } int svr4_sys_resolvepath(td, uap) struct thread *td; struct svr4_sys_resolvepath_args *uap; { struct nameidata nd; int error, *retval = td->td_retval; unsigned int ncopy; NDINIT(&nd, LOOKUP, NOFOLLOW | SAVENAME, UIO_USERSPACE, uap->path, td); if ((error = namei(&nd)) != 0) return (error); NDFREE(&nd, NDF_NO_FREE_PNBUF); ncopy = min(uap->bufsiz, strlen(nd.ni_cnd.cn_pnbuf) + 1); if ((error = copyout(nd.ni_cnd.cn_pnbuf, uap->buf, ncopy)) != 0) goto bad; *retval = ncopy; bad: NDFREE(&nd, NDF_ONLY_PNBUF); return error; } Index: projects/clang400-import/sys/conf/files =================================================================== --- projects/clang400-import/sys/conf/files (revision 311459) +++ projects/clang400-import/sys/conf/files (revision 311460) @@ -1,4663 +1,4664 @@ # $FreeBSD$ # # 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 $S/dev/acpica/acpi_quirks" \ no-obj no-implicit-rule before-depend \ clean "acpi_quirks.h" bhnd_nvram_map.h optional bhnd \ dependency "$S/dev/bhnd/tools/nvram_map_gen.sh $S/dev/bhnd/tools/nvram_map_gen.awk $S/dev/bhnd/nvram/nvram_map" \ compile-with "sh $S/dev/bhnd/tools/nvram_map_gen.sh $S/dev/bhnd/nvram/nvram_map -h" \ no-obj no-implicit-rule before-depend \ clean "bhnd_nvram_map.h" bhnd_nvram_map_data.h optional bhnd \ dependency "$S/dev/bhnd/tools/nvram_map_gen.sh $S/dev/bhnd/tools/nvram_map_gen.awk $S/dev/bhnd/nvram/nvram_map" \ compile-with "sh $S/dev/bhnd/tools/nvram_map_gen.sh $S/dev/bhnd/nvram/nvram_map -d" \ no-obj no-implicit-rule before-depend \ clean "bhnd_nvram_map_data.h" # # The 'fdt_dtb_file' target covers an actual DTB file name, which is derived # from the specified source (DTS) file: .dts -> .dtb # fdt_dtb_file optional fdt fdt_dtb_static \ compile-with "sh -c 'MACHINE=${MACHINE} $S/tools/fdt/make_dtb.sh $S ${FDT_DTS_FILE} ${.CURDIR}'" \ no-obj no-implicit-rule before-depend \ clean "${FDT_DTS_FILE:R}.dtb" fdt_static_dtb.h optional fdt fdt_dtb_static \ compile-with "sh -c 'MACHINE=${MACHINE} $S/tools/fdt/make_dtbh.sh ${FDT_DTS_FILE} ${.CURDIR}'" \ dependency "fdt_dtb_file" \ no-obj no-implicit-rule before-depend \ clean "fdt_static_dtb.h" feeder_eq_gen.h optional sound \ dependency "$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" kbdmuxmap.h optional kbdmux_dflt_keymap \ compile-with "kbdcontrol -P ${S:S/sys$/share/}/vt/keymaps -P ${S:S/sys$/share/}/syscons/keymaps -L ${KBDMUX_DFLT_KEYMAP} | sed -e 's/^static keymap_t.* = /static keymap_t key_map = /' -e 's/^static accentmap_t.* = /static accentmap_t accent_map = /' > kbdmuxmap.h" \ no-obj no-implicit-rule before-depend \ clean "kbdmuxmap.h" teken_state.h optional sc | vt \ dependency "$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 $S/dev/usb/usbdevs -d" \ no-obj no-implicit-rule before-depend \ clean "usbdevs_data.h" cam/cam.c optional scbus cam/cam_compat.c optional scbus cam/cam_iosched.c optional scbus cam/cam_periph.c optional scbus cam/cam_queue.c optional scbus cam/cam_sim.c optional scbus cam/cam_xpt.c optional scbus cam/ata/ata_all.c optional scbus cam/ata/ata_xpt.c optional scbus cam/ata/ata_pmp.c optional scbus cam/nvme/nvme_all.c optional scbus nvme !nvd cam/nvme/nvme_da.c optional scbus nvme da !nvd cam/nvme/nvme_xpt.c optional scbus nvme !nvd cam/scsi/scsi_xpt.c optional scbus cam/scsi/scsi_all.c optional scbus cam/scsi/scsi_cd.c optional cd cam/scsi/scsi_ch.c optional ch cam/ata/ata_da.c optional ada | da cam/ctl/ctl.c optional ctl cam/ctl/ctl_backend.c optional ctl cam/ctl/ctl_backend_block.c optional ctl cam/ctl/ctl_backend_ramdisk.c optional ctl cam/ctl/ctl_cmd_table.c optional ctl cam/ctl/ctl_frontend.c optional ctl cam/ctl/ctl_frontend_cam_sim.c optional ctl cam/ctl/ctl_frontend_ioctl.c optional ctl cam/ctl/ctl_frontend_iscsi.c optional ctl cam/ctl/ctl_ha.c optional ctl cam/ctl/ctl_scsi_all.c optional ctl cam/ctl/ctl_tpc.c optional ctl cam/ctl/ctl_tpc_local.c optional ctl cam/ctl/ctl_error.c optional ctl cam/ctl/ctl_util.c optional ctl cam/ctl/scsi_ctl.c optional ctl cam/scsi/scsi_da.c optional da cam/scsi/scsi_low.c optional ct | ncv | nsp | stg cam/scsi/scsi_pass.c optional pass cam/scsi/scsi_pt.c optional pt cam/scsi/scsi_sa.c optional sa cam/scsi/scsi_enc.c optional ses cam/scsi/scsi_enc_ses.c optional ses cam/scsi/scsi_enc_safte.c optional ses cam/scsi/scsi_sg.c optional sg cam/scsi/scsi_targ_bh.c optional targbh cam/scsi/scsi_target.c optional targ cam/scsi/smp_all.c optional scbus # shared between zfs and dtrace cddl/compat/opensolaris/kern/opensolaris.c optional zfs | dtrace compile-with "${CDDL_C}" cddl/compat/opensolaris/kern/opensolaris_cmn_err.c optional zfs | dtrace compile-with "${CDDL_C}" cddl/compat/opensolaris/kern/opensolaris_kmem.c optional zfs | dtrace compile-with "${CDDL_C}" cddl/compat/opensolaris/kern/opensolaris_misc.c optional zfs | dtrace compile-with "${CDDL_C}" cddl/compat/opensolaris/kern/opensolaris_proc.c optional zfs | dtrace compile-with "${CDDL_C}" cddl/compat/opensolaris/kern/opensolaris_sunddi.c optional zfs | dtrace compile-with "${CDDL_C}" cddl/compat/opensolaris/kern/opensolaris_taskq.c optional zfs | dtrace compile-with "${CDDL_C}" # zfs specific cddl/compat/opensolaris/kern/opensolaris_acl.c optional zfs compile-with "${ZFS_C}" cddl/compat/opensolaris/kern/opensolaris_dtrace.c optional zfs compile-with "${ZFS_C}" cddl/compat/opensolaris/kern/opensolaris_kobj.c optional zfs compile-with "${ZFS_C}" cddl/compat/opensolaris/kern/opensolaris_kstat.c optional zfs compile-with "${ZFS_C}" cddl/compat/opensolaris/kern/opensolaris_lookup.c optional zfs compile-with "${ZFS_C}" cddl/compat/opensolaris/kern/opensolaris_policy.c optional zfs compile-with "${ZFS_C}" cddl/compat/opensolaris/kern/opensolaris_string.c optional zfs compile-with "${ZFS_C}" cddl/compat/opensolaris/kern/opensolaris_sysevent.c optional zfs compile-with "${ZFS_C}" cddl/compat/opensolaris/kern/opensolaris_uio.c optional zfs compile-with "${ZFS_C}" cddl/compat/opensolaris/kern/opensolaris_vfs.c optional zfs compile-with "${ZFS_C}" cddl/compat/opensolaris/kern/opensolaris_vm.c optional zfs compile-with "${ZFS_C}" cddl/compat/opensolaris/kern/opensolaris_zone.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/common/acl/acl_common.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/common/avl/avl.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/common/nvpair/opensolaris_fnvpair.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/common/nvpair/opensolaris_nvpair.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/common/nvpair/opensolaris_nvpair_alloc_fixed.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/common/unicode/u8_textprep.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/common/zfs/zfeature_common.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/common/zfs/zfs_comutil.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/common/zfs/zfs_deleg.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/common/zfs/zfs_fletcher.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/common/zfs/zfs_ioctl_compat.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/common/zfs/zfs_namecheck.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/common/zfs/zfs_prop.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/common/zfs/zpool_prop.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/common/zfs/zprop_common.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/gfs.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/vnode.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/arc.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/blkptr.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/bplist.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/bpobj.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/bptree.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/bqueue.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/dbuf.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/ddt.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/ddt_zap.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/dmu.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/dmu_diff.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/dmu_object.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/dmu_objset.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/dmu_send.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/dmu_traverse.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/dmu_tx.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/dmu_zfetch.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/dnode.c optional zfs compile-with "${ZFS_C}" \ warning "kernel contains CDDL licensed ZFS filesystem" cddl/contrib/opensolaris/uts/common/fs/zfs/dnode_sync.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/dsl_bookmark.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/dsl_dataset.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/dsl_deadlist.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/dsl_deleg.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/dsl_destroy.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/dsl_dir.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/dsl_pool.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/dsl_prop.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/dsl_scan.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/dsl_userhold.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/dsl_synctask.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/gzip.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/lz4.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/lzjb.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/metaslab.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/multilist.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/range_tree.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/refcount.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/rrwlock.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/sa.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/sha256.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/skein_zfs.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/spa.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/spa_config.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/spa_errlog.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/spa_history.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/spa_misc.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/space_map.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/space_reftree.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/trim_map.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/txg.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/uberblock.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/unique.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/vdev.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_cache.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_file.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_geom.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_label.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_mirror.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_missing.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_queue.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_raidz.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_root.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/zap.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/zap_leaf.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/zap_micro.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/zfeature.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_acl.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_byteswap.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_ctldir.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_debug.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_dir.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_fm.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_fuid.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_ioctl.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_log.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_onexit.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_replay.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_rlock.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_sa.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_vfsops.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_vnops.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_znode.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/zil.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/zio.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/zio_checksum.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/zio_compress.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/zio_inject.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/zle.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/zrlock.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/zvol.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/os/callb.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/os/fm.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/os/list.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/os/nvpair_alloc_system.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/zmod/adler32.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/zmod/deflate.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/zmod/inffast.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/zmod/inflate.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/zmod/inftrees.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/zmod/opensolaris_crc32.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/zmod/trees.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/zmod/zmod.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/zmod/zmod_subr.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/zmod/zutil.c optional zfs compile-with "${ZFS_C}" # dtrace specific cddl/contrib/opensolaris/uts/common/dtrace/dtrace.c optional dtrace compile-with "${DTRACE_C}" \ warning "kernel contains CDDL licensed DTRACE" cddl/dev/dtmalloc/dtmalloc.c optional dtmalloc | dtraceall compile-with "${CDDL_C}" cddl/dev/profile/profile.c optional dtrace_profile | dtraceall compile-with "${CDDL_C}" cddl/dev/sdt/sdt.c optional dtrace_sdt | dtraceall compile-with "${CDDL_C}" cddl/dev/fbt/fbt.c optional dtrace_fbt | dtraceall compile-with "${FBT_C}" cddl/dev/systrace/systrace.c optional dtrace_systrace | dtraceall compile-with "${CDDL_C}" cddl/dev/prototype.c optional dtrace_prototype | dtraceall compile-with "${CDDL_C}" fs/nfsclient/nfs_clkdtrace.c optional dtnfscl nfscl | dtraceall nfscl compile-with "${CDDL_C}" compat/cloudabi/cloudabi_clock.c optional compat_cloudabi32 | compat_cloudabi64 compat/cloudabi/cloudabi_errno.c optional compat_cloudabi32 | compat_cloudabi64 compat/cloudabi/cloudabi_fd.c optional compat_cloudabi32 | compat_cloudabi64 compat/cloudabi/cloudabi_file.c optional compat_cloudabi32 | compat_cloudabi64 compat/cloudabi/cloudabi_futex.c optional compat_cloudabi32 | compat_cloudabi64 compat/cloudabi/cloudabi_mem.c optional compat_cloudabi32 | compat_cloudabi64 compat/cloudabi/cloudabi_proc.c optional compat_cloudabi32 | compat_cloudabi64 compat/cloudabi/cloudabi_random.c optional compat_cloudabi32 | compat_cloudabi64 compat/cloudabi/cloudabi_sock.c optional compat_cloudabi32 | compat_cloudabi64 compat/cloudabi/cloudabi_thread.c optional compat_cloudabi32 | compat_cloudabi64 compat/cloudabi/cloudabi_vdso.c optional compat_cloudabi32 | compat_cloudabi64 compat/cloudabi32/cloudabi32_fd.c optional compat_cloudabi32 compat/cloudabi32/cloudabi32_module.c optional compat_cloudabi32 compat/cloudabi32/cloudabi32_poll.c optional compat_cloudabi32 compat/cloudabi32/cloudabi32_sock.c optional compat_cloudabi32 compat/cloudabi32/cloudabi32_syscalls.c optional compat_cloudabi32 compat/cloudabi32/cloudabi32_sysent.c optional compat_cloudabi32 compat/cloudabi32/cloudabi32_thread.c optional compat_cloudabi32 compat/cloudabi64/cloudabi64_fd.c optional compat_cloudabi64 compat/cloudabi64/cloudabi64_module.c optional compat_cloudabi64 compat/cloudabi64/cloudabi64_poll.c optional compat_cloudabi64 compat/cloudabi64/cloudabi64_sock.c optional compat_cloudabi64 compat/cloudabi64/cloudabi64_syscalls.c optional compat_cloudabi64 compat/cloudabi64/cloudabi64_sysent.c optional compat_cloudabi64 compat/cloudabi64/cloudabi64_thread.c optional compat_cloudabi64 compat/freebsd32/freebsd32_capability.c optional compat_freebsd32 compat/freebsd32/freebsd32_ioctl.c optional compat_freebsd32 compat/freebsd32/freebsd32_misc.c optional compat_freebsd32 compat/freebsd32/freebsd32_syscalls.c optional compat_freebsd32 compat/freebsd32/freebsd32_sysent.c optional compat_freebsd32 contrib/ck/src/ck_array.c standard compile-with "${NORMAL_C} -I$S/contrib/ck/include" contrib/ck/src/ck_barrier_centralized.c standard compile-with "${NORMAL_C} -I$S/contrib/ck/include" contrib/ck/src/ck_barrier_combining.c standard compile-with "${NORMAL_C} -I$S/contrib/ck/include" contrib/ck/src/ck_barrier_dissemination.c standard compile-with "${NORMAL_C} -I$S/contrib/ck/include" contrib/ck/src/ck_barrier_mcs.c standard compile-with "${NORMAL_C} -I$S/contrib/ck/include" contrib/ck/src/ck_barrier_tournament.c standard compile-with "${NORMAL_C} -I$S/contrib/ck/include" contrib/ck/src/ck_epoch.c standard compile-with "${NORMAL_C} -I$S/contrib/ck/include" contrib/ck/src/ck_hp.c standard compile-with "${NORMAL_C} -I$S/contrib/ck/include" contrib/ck/src/ck_hs.c standard compile-with "${NORMAL_C} -I$S/contrib/ck/include" contrib/ck/src/ck_ht.c standard compile-with "${NORMAL_C} -I$S/contrib/ck/include" contrib/ck/src/ck_rhs.c standard compile-with "${NORMAL_C} -I$S/contrib/ck/include" contrib/dev/acpica/common/ahids.c optional acpi acpi_debug contrib/dev/acpica/common/ahuuids.c optional acpi acpi_debug contrib/dev/acpica/components/debugger/dbcmds.c optional acpi acpi_debug contrib/dev/acpica/components/debugger/dbconvert.c optional acpi acpi_debug contrib/dev/acpica/components/debugger/dbdisply.c optional acpi acpi_debug contrib/dev/acpica/components/debugger/dbexec.c optional acpi acpi_debug contrib/dev/acpica/components/debugger/dbhistry.c optional acpi acpi_debug contrib/dev/acpica/components/debugger/dbinput.c optional acpi acpi_debug contrib/dev/acpica/components/debugger/dbmethod.c optional acpi acpi_debug contrib/dev/acpica/components/debugger/dbnames.c optional acpi acpi_debug contrib/dev/acpica/components/debugger/dbobject.c optional acpi acpi_debug contrib/dev/acpica/components/debugger/dbstats.c optional acpi acpi_debug contrib/dev/acpica/components/debugger/dbtest.c optional acpi acpi_debug contrib/dev/acpica/components/debugger/dbutils.c optional acpi acpi_debug contrib/dev/acpica/components/debugger/dbxface.c optional acpi acpi_debug contrib/dev/acpica/components/disassembler/dmbuffer.c optional acpi acpi_debug contrib/dev/acpica/components/disassembler/dmcstyle.c optional acpi acpi_debug contrib/dev/acpica/components/disassembler/dmdeferred.c optional acpi acpi_debug contrib/dev/acpica/components/disassembler/dmnames.c optional acpi acpi_debug contrib/dev/acpica/components/disassembler/dmopcode.c optional acpi acpi_debug contrib/dev/acpica/components/disassembler/dmresrc.c optional acpi acpi_debug contrib/dev/acpica/components/disassembler/dmresrcl.c optional acpi acpi_debug contrib/dev/acpica/components/disassembler/dmresrcl2.c optional acpi acpi_debug contrib/dev/acpica/components/disassembler/dmresrcs.c optional acpi acpi_debug contrib/dev/acpica/components/disassembler/dmutils.c optional acpi acpi_debug contrib/dev/acpica/components/disassembler/dmwalk.c optional acpi acpi_debug contrib/dev/acpica/components/dispatcher/dsargs.c optional acpi contrib/dev/acpica/components/dispatcher/dscontrol.c optional acpi contrib/dev/acpica/components/dispatcher/dsdebug.c optional acpi contrib/dev/acpica/components/dispatcher/dsfield.c optional acpi contrib/dev/acpica/components/dispatcher/dsinit.c optional acpi contrib/dev/acpica/components/dispatcher/dsmethod.c optional acpi contrib/dev/acpica/components/dispatcher/dsmthdat.c optional acpi contrib/dev/acpica/components/dispatcher/dsobject.c optional acpi contrib/dev/acpica/components/dispatcher/dsopcode.c optional acpi contrib/dev/acpica/components/dispatcher/dsutils.c optional acpi contrib/dev/acpica/components/dispatcher/dswexec.c optional acpi contrib/dev/acpica/components/dispatcher/dswload.c optional acpi contrib/dev/acpica/components/dispatcher/dswload2.c optional acpi contrib/dev/acpica/components/dispatcher/dswscope.c optional acpi contrib/dev/acpica/components/dispatcher/dswstate.c optional acpi contrib/dev/acpica/components/events/evevent.c optional acpi contrib/dev/acpica/components/events/evglock.c optional acpi contrib/dev/acpica/components/events/evgpe.c optional acpi contrib/dev/acpica/components/events/evgpeblk.c optional acpi contrib/dev/acpica/components/events/evgpeinit.c optional acpi contrib/dev/acpica/components/events/evgpeutil.c optional acpi contrib/dev/acpica/components/events/evhandler.c optional acpi contrib/dev/acpica/components/events/evmisc.c optional acpi contrib/dev/acpica/components/events/evregion.c optional acpi contrib/dev/acpica/components/events/evrgnini.c optional acpi contrib/dev/acpica/components/events/evsci.c optional acpi contrib/dev/acpica/components/events/evxface.c optional acpi contrib/dev/acpica/components/events/evxfevnt.c optional acpi contrib/dev/acpica/components/events/evxfgpe.c optional acpi contrib/dev/acpica/components/events/evxfregn.c optional acpi contrib/dev/acpica/components/executer/exconcat.c optional acpi contrib/dev/acpica/components/executer/exconfig.c optional acpi contrib/dev/acpica/components/executer/exconvrt.c optional acpi contrib/dev/acpica/components/executer/excreate.c optional acpi contrib/dev/acpica/components/executer/exdebug.c optional acpi contrib/dev/acpica/components/executer/exdump.c optional acpi contrib/dev/acpica/components/executer/exfield.c optional acpi contrib/dev/acpica/components/executer/exfldio.c optional acpi contrib/dev/acpica/components/executer/exmisc.c optional acpi contrib/dev/acpica/components/executer/exmutex.c optional acpi contrib/dev/acpica/components/executer/exnames.c optional acpi contrib/dev/acpica/components/executer/exoparg1.c optional acpi contrib/dev/acpica/components/executer/exoparg2.c optional acpi contrib/dev/acpica/components/executer/exoparg3.c optional acpi contrib/dev/acpica/components/executer/exoparg6.c optional acpi contrib/dev/acpica/components/executer/exprep.c optional acpi contrib/dev/acpica/components/executer/exregion.c optional acpi contrib/dev/acpica/components/executer/exresnte.c optional acpi contrib/dev/acpica/components/executer/exresolv.c optional acpi contrib/dev/acpica/components/executer/exresop.c optional acpi contrib/dev/acpica/components/executer/exstore.c optional acpi contrib/dev/acpica/components/executer/exstoren.c optional acpi contrib/dev/acpica/components/executer/exstorob.c optional acpi contrib/dev/acpica/components/executer/exsystem.c optional acpi contrib/dev/acpica/components/executer/extrace.c optional acpi contrib/dev/acpica/components/executer/exutils.c optional acpi contrib/dev/acpica/components/hardware/hwacpi.c optional acpi contrib/dev/acpica/components/hardware/hwesleep.c optional acpi contrib/dev/acpica/components/hardware/hwgpe.c optional acpi contrib/dev/acpica/components/hardware/hwpci.c optional acpi contrib/dev/acpica/components/hardware/hwregs.c optional acpi contrib/dev/acpica/components/hardware/hwsleep.c optional acpi contrib/dev/acpica/components/hardware/hwtimer.c optional acpi contrib/dev/acpica/components/hardware/hwvalid.c optional acpi contrib/dev/acpica/components/hardware/hwxface.c optional acpi contrib/dev/acpica/components/hardware/hwxfsleep.c optional acpi contrib/dev/acpica/components/namespace/nsaccess.c optional acpi contrib/dev/acpica/components/namespace/nsalloc.c optional acpi contrib/dev/acpica/components/namespace/nsarguments.c optional acpi contrib/dev/acpica/components/namespace/nsconvert.c optional acpi contrib/dev/acpica/components/namespace/nsdump.c optional acpi contrib/dev/acpica/components/namespace/nseval.c optional acpi contrib/dev/acpica/components/namespace/nsinit.c optional acpi contrib/dev/acpica/components/namespace/nsload.c optional acpi contrib/dev/acpica/components/namespace/nsnames.c optional acpi contrib/dev/acpica/components/namespace/nsobject.c optional acpi contrib/dev/acpica/components/namespace/nsparse.c optional acpi contrib/dev/acpica/components/namespace/nspredef.c optional acpi contrib/dev/acpica/components/namespace/nsprepkg.c optional acpi contrib/dev/acpica/components/namespace/nsrepair.c optional acpi contrib/dev/acpica/components/namespace/nsrepair2.c optional acpi contrib/dev/acpica/components/namespace/nssearch.c optional acpi contrib/dev/acpica/components/namespace/nsutils.c optional acpi contrib/dev/acpica/components/namespace/nswalk.c optional acpi contrib/dev/acpica/components/namespace/nsxfeval.c optional acpi contrib/dev/acpica/components/namespace/nsxfname.c optional acpi contrib/dev/acpica/components/namespace/nsxfobj.c optional acpi contrib/dev/acpica/components/parser/psargs.c optional acpi contrib/dev/acpica/components/parser/psloop.c optional acpi contrib/dev/acpica/components/parser/psobject.c optional acpi contrib/dev/acpica/components/parser/psopcode.c optional acpi contrib/dev/acpica/components/parser/psopinfo.c optional acpi contrib/dev/acpica/components/parser/psparse.c optional acpi contrib/dev/acpica/components/parser/psscope.c optional acpi contrib/dev/acpica/components/parser/pstree.c optional acpi contrib/dev/acpica/components/parser/psutils.c optional acpi contrib/dev/acpica/components/parser/pswalk.c optional acpi contrib/dev/acpica/components/parser/psxface.c optional acpi contrib/dev/acpica/components/resources/rsaddr.c optional acpi contrib/dev/acpica/components/resources/rscalc.c optional acpi contrib/dev/acpica/components/resources/rscreate.c optional acpi contrib/dev/acpica/components/resources/rsdump.c optional acpi acpi_debug contrib/dev/acpica/components/resources/rsdumpinfo.c optional acpi contrib/dev/acpica/components/resources/rsinfo.c optional acpi contrib/dev/acpica/components/resources/rsio.c optional acpi contrib/dev/acpica/components/resources/rsirq.c optional acpi contrib/dev/acpica/components/resources/rslist.c optional acpi contrib/dev/acpica/components/resources/rsmemory.c optional acpi contrib/dev/acpica/components/resources/rsmisc.c optional acpi contrib/dev/acpica/components/resources/rsserial.c optional acpi contrib/dev/acpica/components/resources/rsutils.c optional acpi contrib/dev/acpica/components/resources/rsxface.c optional acpi contrib/dev/acpica/components/tables/tbdata.c optional acpi contrib/dev/acpica/components/tables/tbfadt.c optional acpi contrib/dev/acpica/components/tables/tbfind.c optional acpi contrib/dev/acpica/components/tables/tbinstal.c optional acpi contrib/dev/acpica/components/tables/tbprint.c optional acpi contrib/dev/acpica/components/tables/tbutils.c optional acpi contrib/dev/acpica/components/tables/tbxface.c optional acpi contrib/dev/acpica/components/tables/tbxfload.c optional acpi contrib/dev/acpica/components/tables/tbxfroot.c optional acpi contrib/dev/acpica/components/utilities/utaddress.c optional acpi contrib/dev/acpica/components/utilities/utalloc.c optional acpi contrib/dev/acpica/components/utilities/utascii.c optional acpi contrib/dev/acpica/components/utilities/utbuffer.c optional acpi contrib/dev/acpica/components/utilities/utcache.c optional acpi contrib/dev/acpica/components/utilities/utcopy.c optional acpi contrib/dev/acpica/components/utilities/utdebug.c optional acpi contrib/dev/acpica/components/utilities/utdecode.c optional acpi contrib/dev/acpica/components/utilities/utdelete.c optional acpi contrib/dev/acpica/components/utilities/uterror.c optional acpi contrib/dev/acpica/components/utilities/uteval.c optional acpi contrib/dev/acpica/components/utilities/utexcep.c optional acpi contrib/dev/acpica/components/utilities/utglobal.c optional acpi contrib/dev/acpica/components/utilities/uthex.c optional acpi contrib/dev/acpica/components/utilities/utids.c optional acpi contrib/dev/acpica/components/utilities/utinit.c optional acpi contrib/dev/acpica/components/utilities/utlock.c optional acpi contrib/dev/acpica/components/utilities/utmath.c optional acpi contrib/dev/acpica/components/utilities/utmisc.c optional acpi contrib/dev/acpica/components/utilities/utmutex.c optional acpi contrib/dev/acpica/components/utilities/utnonansi.c optional acpi contrib/dev/acpica/components/utilities/utobject.c optional acpi contrib/dev/acpica/components/utilities/utosi.c optional acpi contrib/dev/acpica/components/utilities/utownerid.c optional acpi contrib/dev/acpica/components/utilities/utpredef.c optional acpi contrib/dev/acpica/components/utilities/utresrc.c optional acpi contrib/dev/acpica/components/utilities/utstate.c optional acpi contrib/dev/acpica/components/utilities/utstring.c optional acpi contrib/dev/acpica/components/utilities/utstrtoul64.c optional acpi contrib/dev/acpica/components/utilities/utuuid.c optional acpi acpi_debug contrib/dev/acpica/components/utilities/utxface.c optional acpi contrib/dev/acpica/components/utilities/utxferror.c optional acpi contrib/dev/acpica/components/utilities/utxfinit.c optional acpi #contrib/dev/acpica/components/utilities/utxfmutex.c optional acpi contrib/ipfilter/netinet/fil.c optional ipfilter inet \ compile-with "${NORMAL_C} ${NO_WSELF_ASSIGN} -Wno-unused -I$S/contrib/ipfilter" contrib/ipfilter/netinet/ip_auth.c optional ipfilter inet \ compile-with "${NORMAL_C} -Wno-unused -I$S/contrib/ipfilter" contrib/ipfilter/netinet/ip_fil_freebsd.c optional ipfilter inet \ compile-with "${NORMAL_C} -Wno-unused -I$S/contrib/ipfilter" contrib/ipfilter/netinet/ip_frag.c optional ipfilter inet \ compile-with "${NORMAL_C} -Wno-unused -I$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 "${NORMAL_C} -Wno-unused -I$S/contrib/ipfilter" contrib/ipfilter/netinet/ip_proxy.c optional ipfilter inet \ compile-with "${NORMAL_C} ${NO_WSELF_ASSIGN} -Wno-unused -I$S/contrib/ipfilter" contrib/ipfilter/netinet/ip_state.c optional ipfilter inet \ compile-with "${NORMAL_C} -Wno-unused -I$S/contrib/ipfilter" contrib/ipfilter/netinet/ip_lookup.c optional ipfilter inet \ compile-with "${NORMAL_C} ${NO_WSELF_ASSIGN} -Wno-unused -Wno-error -I$S/contrib/ipfilter" contrib/ipfilter/netinet/ip_pool.c optional ipfilter inet \ compile-with "${NORMAL_C} -Wno-unused -I$S/contrib/ipfilter" contrib/ipfilter/netinet/ip_htable.c optional ipfilter inet \ compile-with "${NORMAL_C} -Wno-unused -I$S/contrib/ipfilter" contrib/ipfilter/netinet/ip_sync.c optional ipfilter inet \ compile-with "${NORMAL_C} -Wno-unused -I$S/contrib/ipfilter" contrib/ipfilter/netinet/mlfk_ipl.c optional ipfilter inet \ compile-with "${NORMAL_C} -I$S/contrib/ipfilter" contrib/ipfilter/netinet/ip_nat6.c optional ipfilter inet \ compile-with "${NORMAL_C} -Wno-unused -I$S/contrib/ipfilter" contrib/ipfilter/netinet/ip_rules.c optional ipfilter inet \ compile-with "${NORMAL_C} -I$S/contrib/ipfilter" contrib/ipfilter/netinet/ip_scan.c optional ipfilter inet \ compile-with "${NORMAL_C} -Wno-unused -I$S/contrib/ipfilter" contrib/ipfilter/netinet/ip_dstlist.c optional ipfilter inet \ compile-with "${NORMAL_C} -Wno-unused -I$S/contrib/ipfilter" contrib/ipfilter/netinet/radix_ipf.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/libnv/cnvlist.c standard contrib/libnv/dnvlist.c standard contrib/libnv/nvlist.c standard contrib/libnv/nvpair.c standard 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" 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 | ekcd | geom_bde | \ ipsec | random !random_loadable | wlan_ccmp crypto/rijndael/rijndael-api-fst.c optional ekcd | geom_bde | random !random_loadable crypto/rijndael/rijndael-api.c optional crypto | ipsec | wlan_ccmp crypto/sha1.c optional carp | crypto | ipsec | \ netgraph_mppc_encryption | sctp crypto/sha2/sha256c.c optional crypto | ekcd | geom_bde | ipsec | random !random_loadable | \ sctp | zfs crypto/sha2/sha512c.c optional crypto | geom_bde | ipsec | zfs crypto/skein/skein.c optional crypto | zfs crypto/skein/skein_block.c optional crypto | zfs crypto/siphash/siphash.c optional inet | inet6 crypto/siphash/siphash_test.c optional inet | inet6 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/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/aacraid/aacraid.c optional aacraid dev/aacraid/aacraid_cam.c optional aacraid scbus dev/aacraid/aacraid_debug.c optional aacraid dev/aacraid/aacraid_linux.c optional aacraid compat_linux dev/aacraid/aacraid_pci.c optional aacraid 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_asus_wmi.c optional acpi_asus_wmi 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_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_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 dev/ahci/ahciem.c optional ahci dev/ahci/ahci_pci.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 \ compile-with "${NORMAL_C} ${NO_WCONSTANT_CONVERSION}" dev/aic7xxx/ahd_pci.c optional ahd pci \ compile-with "${NORMAL_C} ${NO_WCONSTANT_CONVERSION}" dev/aic7xxx/aic7770.c optional ahc dev/aic7xxx/aic79xx.c optional ahd pci dev/aic7xxx/aic79xx_osm.c optional ahd pci dev/aic7xxx/aic79xx_pci.c optional ahd pci dev/aic7xxx/aic79xx_reg_print.c optional ahd pci ahd_reg_pretty_print dev/aic7xxx/aic7xxx.c optional ahc dev/aic7xxx/aic7xxx_93cx6.c optional ahc dev/aic7xxx/aic7xxx_osm.c optional ahc dev/aic7xxx/aic7xxx_pci.c optional ahc pci dev/aic7xxx/aic7xxx_reg_print.c optional ahc ahc_reg_pretty_print dev/al_eth/al_eth.c optional al_eth fdt \ no-depend \ compile-with "${CC} -c -o ${.TARGET} ${CFLAGS} -I$S/contrib/alpine-hal -I$S/contrib/alpine-hal/eth ${PROF} ${.IMPSRC}" dev/al_eth/al_init_eth_lm.c optional al_eth fdt \ no-depend \ compile-with "${CC} -c -o ${.TARGET} ${CFLAGS} -I$S/contrib/alpine-hal -I$S/contrib/alpine-hal/eth ${PROF} ${.IMPSRC}" dev/al_eth/al_init_eth_kr.c optional al_eth fdt \ no-depend \ compile-with "${CC} -c -o ${.TARGET} ${CFLAGS} -I$S/contrib/alpine-hal -I$S/contrib/alpine-hal/eth ${PROF} ${.IMPSRC}" contrib/alpine-hal/al_hal_iofic.c optional al_iofic \ no-depend \ compile-with "${CC} -c -o ${.TARGET} ${CFLAGS} -I$S/contrib/alpine-hal -I$S/contrib/alpine-hal/eth ${PROF} ${.IMPSRC}" contrib/alpine-hal/al_hal_serdes_25g.c optional al_serdes \ no-depend \ compile-with "${CC} -c -o ${.TARGET} ${CFLAGS} -I$S/contrib/alpine-hal -I$S/contrib/alpine-hal/eth ${PROF} ${.IMPSRC}" contrib/alpine-hal/al_hal_serdes_hssp.c optional al_serdes \ no-depend \ compile-with "${CC} -c -o ${.TARGET} ${CFLAGS} -I$S/contrib/alpine-hal -I$S/contrib/alpine-hal/eth ${PROF} ${.IMPSRC}" contrib/alpine-hal/al_hal_udma_config.c optional al_udma \ no-depend \ compile-with "${CC} -c -o ${.TARGET} ${CFLAGS} -I$S/contrib/alpine-hal -I$S/contrib/alpine-hal/eth ${PROF} ${.IMPSRC}" contrib/alpine-hal/al_hal_udma_debug.c optional al_udma \ no-depend \ compile-with "${CC} -c -o ${.TARGET} ${CFLAGS} -I$S/contrib/alpine-hal -I$S/contrib/alpine-hal/eth ${PROF} ${.IMPSRC}" contrib/alpine-hal/al_hal_udma_iofic.c optional al_udma \ no-depend \ compile-with "${CC} -c -o ${.TARGET} ${CFLAGS} -I$S/contrib/alpine-hal -I$S/contrib/alpine-hal/eth ${PROF} ${.IMPSRC}" contrib/alpine-hal/al_hal_udma_main.c optional al_udma \ no-depend \ compile-with "${CC} -c -o ${.TARGET} ${CFLAGS} -I$S/contrib/alpine-hal -I$S/contrib/alpine-hal/eth ${PROF} ${.IMPSRC}" contrib/alpine-hal/al_serdes.c optional al_serdes \ no-depend \ compile-with "${CC} -c -o ${.TARGET} ${CFLAGS} -I$S/contrib/alpine-hal -I$S/contrib/alpine-hal/eth ${PROF} ${.IMPSRC}" contrib/alpine-hal/eth/al_hal_eth_kr.c optional al_eth \ no-depend \ compile-with "${CC} -c -o ${.TARGET} ${CFLAGS} -I$S/contrib/alpine-hal -I$S/contrib/alpine-hal/eth ${PROF} ${.IMPSRC}" contrib/alpine-hal/eth/al_hal_eth_main.c optional al_eth \ no-depend \ compile-with "${CC} -c -o ${.TARGET} ${CFLAGS} -I$S/contrib/alpine-hal -I$S/contrib/alpine-hal/eth ${PROF} ${.IMPSRC}" dev/alc/if_alc.c optional alc pci dev/ale/if_ale.c optional ale pci dev/alpm/alpm.c optional alpm pci dev/altera/avgen/altera_avgen.c optional altera_avgen dev/altera/avgen/altera_avgen_fdt.c optional altera_avgen fdt dev/altera/avgen/altera_avgen_nexus.c optional altera_avgen dev/altera/sdcard/altera_sdcard.c optional altera_sdcard dev/altera/sdcard/altera_sdcard_disk.c optional altera_sdcard dev/altera/sdcard/altera_sdcard_io.c optional altera_sdcard dev/altera/sdcard/altera_sdcard_fdt.c optional altera_sdcard fdt dev/altera/sdcard/altera_sdcard_nexus.c optional altera_sdcard dev/altera/pio/pio.c optional altera_pio dev/altera/pio/pio_if.m optional altera_pio dev/amdpm/amdpm.c optional amdpm pci | nfpm pci dev/amdsmb/amdsmb.c optional amdsmb pci 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/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-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-acard.c optional ata pci | ataacard dev/ata/chipsets/ata-acerlabs.c optional ata pci | ataacerlabs 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 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/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$S/dev/ath" dev/ath/if_ath_alq.c optional ath \ compile-with "${NORMAL_C} -I$S/dev/ath" dev/ath/if_ath_beacon.c optional ath \ compile-with "${NORMAL_C} -I$S/dev/ath" dev/ath/if_ath_btcoex.c optional ath \ compile-with "${NORMAL_C} -I$S/dev/ath" dev/ath/if_ath_btcoex_mci.c optional ath \ compile-with "${NORMAL_C} -I$S/dev/ath" dev/ath/if_ath_debug.c optional ath \ compile-with "${NORMAL_C} -I$S/dev/ath" dev/ath/if_ath_descdma.c optional ath \ compile-with "${NORMAL_C} -I$S/dev/ath" dev/ath/if_ath_keycache.c optional ath \ compile-with "${NORMAL_C} -I$S/dev/ath" dev/ath/if_ath_ioctl.c optional ath \ compile-with "${NORMAL_C} -I$S/dev/ath" dev/ath/if_ath_led.c optional ath \ compile-with "${NORMAL_C} -I$S/dev/ath" dev/ath/if_ath_lna_div.c optional ath \ compile-with "${NORMAL_C} -I$S/dev/ath" dev/ath/if_ath_tx.c optional ath \ compile-with "${NORMAL_C} -I$S/dev/ath" dev/ath/if_ath_tx_edma.c optional ath \ compile-with "${NORMAL_C} -I$S/dev/ath" dev/ath/if_ath_tx_ht.c optional ath \ compile-with "${NORMAL_C} -I$S/dev/ath" dev/ath/if_ath_tdma.c optional ath \ compile-with "${NORMAL_C} -I$S/dev/ath" dev/ath/if_ath_sysctl.c optional ath \ compile-with "${NORMAL_C} -I$S/dev/ath" dev/ath/if_ath_rx.c optional ath \ compile-with "${NORMAL_C} -I$S/dev/ath" dev/ath/if_ath_rx_edma.c optional ath \ compile-with "${NORMAL_C} -I$S/dev/ath" dev/ath/if_ath_spectral.c optional ath \ compile-with "${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} ${NO_WSHIFT_COUNT_NEGATIVE} ${NO_WSHIFT_COUNT_OVERFLOW} -I$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$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5416/ar5416_btcoex.c \ optional ath_hal | ath_ar5416 | ath_ar9160 | ath_ar9280 | ath_ar9285 | \ ath_ar9287 \ compile-with "${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$S/dev/ath/ath_hal" dev/ath/ath_hal/ar5416/ar5416_radar.c \ optional ath_hal | ath_ar5416 | ath_ar9160 | ath_ar9280 | ath_ar9285 | \ ath_ar9287 \ compile-with "${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_spectral.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 # # Since this is an embedded MAC SoC, there's no need to compile it into the # default HAL. dev/ath/ath_hal/ar9001/ar9130_attach.c optional ath_ar9130 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar9001/ar9130_phy.c optional ath_ar9130 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal" dev/ath/ath_hal/ar9001/ar9130_eeprom.c optional ath_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$S/dev/ath/ath_hal" dev/ath/ath_hal/ar9002/ar9285_btcoex.c optional ath_hal | ath_ar9285 \ compile-with "${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" # ar9300 contrib/dev/ath/ath_hal/ar9300/ar9300_ani.c optional ath_hal | ath_ar9300 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal -I$S/contrib/dev/ath/ath_hal" contrib/dev/ath/ath_hal/ar9300/ar9300_attach.c optional ath_hal | ath_ar9300 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal -I$S/contrib/dev/ath/ath_hal" contrib/dev/ath/ath_hal/ar9300/ar9300_beacon.c optional ath_hal | ath_ar9300 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal -I$S/contrib/dev/ath/ath_hal" contrib/dev/ath/ath_hal/ar9300/ar9300_eeprom.c optional ath_hal | ath_ar9300 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal -I$S/contrib/dev/ath/ath_hal ${NO_WCONSTANT_CONVERSION}" contrib/dev/ath/ath_hal/ar9300/ar9300_freebsd.c optional ath_hal | ath_ar9300 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal -I$S/contrib/dev/ath/ath_hal" contrib/dev/ath/ath_hal/ar9300/ar9300_gpio.c optional ath_hal | ath_ar9300 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal -I$S/contrib/dev/ath/ath_hal" contrib/dev/ath/ath_hal/ar9300/ar9300_interrupts.c optional ath_hal | ath_ar9300 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal -I$S/contrib/dev/ath/ath_hal" contrib/dev/ath/ath_hal/ar9300/ar9300_keycache.c optional ath_hal | ath_ar9300 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal -I$S/contrib/dev/ath/ath_hal" contrib/dev/ath/ath_hal/ar9300/ar9300_mci.c optional ath_hal | ath_ar9300 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal -I$S/contrib/dev/ath/ath_hal" contrib/dev/ath/ath_hal/ar9300/ar9300_misc.c optional ath_hal | ath_ar9300 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal -I$S/contrib/dev/ath/ath_hal" contrib/dev/ath/ath_hal/ar9300/ar9300_paprd.c optional ath_hal | ath_ar9300 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal -I$S/contrib/dev/ath/ath_hal" contrib/dev/ath/ath_hal/ar9300/ar9300_phy.c optional ath_hal | ath_ar9300 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal -I$S/contrib/dev/ath/ath_hal" contrib/dev/ath/ath_hal/ar9300/ar9300_power.c optional ath_hal | ath_ar9300 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal -I$S/contrib/dev/ath/ath_hal" contrib/dev/ath/ath_hal/ar9300/ar9300_radar.c optional ath_hal | ath_ar9300 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal -I$S/contrib/dev/ath/ath_hal" contrib/dev/ath/ath_hal/ar9300/ar9300_radio.c optional ath_hal | ath_ar9300 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal -I$S/contrib/dev/ath/ath_hal" contrib/dev/ath/ath_hal/ar9300/ar9300_recv.c optional ath_hal | ath_ar9300 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal -I$S/contrib/dev/ath/ath_hal" contrib/dev/ath/ath_hal/ar9300/ar9300_recv_ds.c optional ath_hal | ath_ar9300 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal -I$S/contrib/dev/ath/ath_hal" contrib/dev/ath/ath_hal/ar9300/ar9300_reset.c optional ath_hal | ath_ar9300 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal -I$S/contrib/dev/ath/ath_hal ${NO_WSOMETIMES_UNINITIALIZED} -Wno-unused-function" contrib/dev/ath/ath_hal/ar9300/ar9300_stub.c optional ath_hal | ath_ar9300 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal -I$S/contrib/dev/ath/ath_hal" contrib/dev/ath/ath_hal/ar9300/ar9300_stub_funcs.c optional ath_hal | ath_ar9300 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal -I$S/contrib/dev/ath/ath_hal" contrib/dev/ath/ath_hal/ar9300/ar9300_spectral.c optional ath_hal | ath_ar9300 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal -I$S/contrib/dev/ath/ath_hal" contrib/dev/ath/ath_hal/ar9300/ar9300_timer.c optional ath_hal | ath_ar9300 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal -I$S/contrib/dev/ath/ath_hal" contrib/dev/ath/ath_hal/ar9300/ar9300_xmit.c optional ath_hal | ath_ar9300 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal -I$S/contrib/dev/ath/ath_hal" contrib/dev/ath/ath_hal/ar9300/ar9300_xmit_ds.c optional ath_hal | ath_ar9300 \ compile-with "${NORMAL_C} -I$S/dev/ath -I$S/dev/ath/ath_hal -I$S/contrib/dev/ath/ath_hal" # rf backends dev/ath/ath_hal/ar5212/ar2316.c optional ath_rf2316 \ compile-with "${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_ar9130 | ath_ar9160 | ath_ar9280 \ 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/bhnd/bhnd.c optional bhnd dev/bhnd/bhnd_erom.c optional bhnd dev/bhnd/bhnd_erom_if.m optional bhnd dev/bhnd/bhnd_nexus.c optional bhnd siba_nexus | \ bhnd bcma_nexus dev/bhnd/bhnd_subr.c optional bhnd dev/bhnd/bhnd_bus_if.m optional bhnd dev/bhnd/bhndb/bhnd_bhndb.c optional bhndb bhnd dev/bhnd/bhndb/bhndb.c optional bhndb bhnd dev/bhnd/bhndb/bhndb_bus_if.m optional bhndb bhnd dev/bhnd/bhndb/bhndb_hwdata.c optional bhndb bhnd dev/bhnd/bhndb/bhndb_if.m optional bhndb bhnd dev/bhnd/bhndb/bhndb_pci.c optional bhndb bhnd pci dev/bhnd/bhndb/bhndb_pci_hwdata.c optional bhndb bhnd pci dev/bhnd/bhndb/bhndb_pci_sprom.c optional bhndb bhnd pci dev/bhnd/bhndb/bhndb_subr.c optional bhndb bhnd dev/bhnd/bcma/bcma.c optional bcma bhnd dev/bhnd/bcma/bcma_bhndb.c optional bcma bhnd bhndb dev/bhnd/bcma/bcma_erom.c optional bcma bhnd dev/bhnd/bcma/bcma_nexus.c optional bcma_nexus bcma bhnd dev/bhnd/bcma/bcma_subr.c optional bcma bhnd dev/bhnd/cores/chipc/bhnd_chipc_if.m optional bhnd dev/bhnd/cores/chipc/bhnd_sprom_chipc.c optional bhnd dev/bhnd/cores/chipc/bhnd_pmu_chipc.c optional bhnd dev/bhnd/cores/chipc/chipc.c optional bhnd dev/bhnd/cores/chipc/chipc_cfi.c optional bhnd cfi dev/bhnd/cores/chipc/chipc_slicer.c optional bhnd cfi | bhnd spibus dev/bhnd/cores/chipc/chipc_spi.c optional bhnd spibus dev/bhnd/cores/chipc/chipc_subr.c optional bhnd dev/bhnd/cores/chipc/pwrctl/bhnd_pwrctl.c optional bhnd dev/bhnd/cores/chipc/pwrctl/bhnd_pwrctl_subr.c optional bhnd dev/bhnd/cores/pci/bhnd_pci.c optional bhnd pci dev/bhnd/cores/pci/bhnd_pci_hostb.c optional bhndb bhnd pci dev/bhnd/cores/pci/bhnd_pcib.c optional bhnd_pcib bhnd pci dev/bhnd/cores/pcie2/bhnd_pcie2.c optional bhnd pci dev/bhnd/cores/pcie2/bhnd_pcie2_hostb.c optional bhndb bhnd pci dev/bhnd/cores/pcie2/bhnd_pcie2b.c optional bhnd_pcie2b bhnd pci dev/bhnd/cores/pmu/bhnd_pmu.c optional bhnd dev/bhnd/cores/pmu/bhnd_pmu_core.c optional bhnd dev/bhnd/cores/pmu/bhnd_pmu_if.m optional bhnd dev/bhnd/cores/pmu/bhnd_pmu_subr.c optional bhnd dev/bhnd/nvram/bhnd_nvram_data.c optional bhnd dev/bhnd/nvram/bhnd_nvram_data_bcm.c optional bhnd dev/bhnd/nvram/bhnd_nvram_data_bcmraw.c optional bhnd dev/bhnd/nvram/bhnd_nvram_data_btxt.c optional bhnd dev/bhnd/nvram/bhnd_nvram_data_sprom.c optional bhnd dev/bhnd/nvram/bhnd_nvram_data_sprom_subr.c optional bhnd dev/bhnd/nvram/bhnd_nvram_data_tlv.c optional bhnd dev/bhnd/nvram/bhnd_nvram_if.m optional bhnd dev/bhnd/nvram/bhnd_nvram_io.c optional bhnd dev/bhnd/nvram/bhnd_nvram_iobuf.c optional bhnd dev/bhnd/nvram/bhnd_nvram_ioptr.c optional bhnd dev/bhnd/nvram/bhnd_nvram_iores.c optional bhnd dev/bhnd/nvram/bhnd_nvram_plist.c optional bhnd dev/bhnd/nvram/bhnd_nvram_store.c optional bhnd dev/bhnd/nvram/bhnd_nvram_store_subr.c optional bhnd dev/bhnd/nvram/bhnd_nvram_subr.c optional bhnd dev/bhnd/nvram/bhnd_nvram_value.c optional bhnd dev/bhnd/nvram/bhnd_nvram_value_fmts.c optional bhnd dev/bhnd/nvram/bhnd_nvram_value_prf.c optional bhnd dev/bhnd/nvram/bhnd_nvram_value_subr.c optional bhnd dev/bhnd/nvram/bhnd_sprom.c optional bhnd dev/bhnd/siba/siba.c optional siba bhnd dev/bhnd/siba/siba_bhndb.c optional siba bhnd bhndb dev/bhnd/siba/siba_erom.c optional siba bhnd dev/bhnd/siba/siba_nexus.c optional siba_nexus siba bhnd dev/bhnd/siba/siba_subr.c optional siba bhnd # 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/bnxt/bnxt_hwrm.c optional bnxt iflib pci dev/bnxt/bnxt_sysctl.c optional bnxt iflib pci dev/bnxt/bnxt_txrx.c optional bnxt iflib pci dev/bnxt/if_bnxt.c optional bnxt iflib 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 # XXX Work around clang warnings, until maintainer approves fix. dev/bwn/if_bwn.c optional bwn siba_bwn \ compile-with "${NORMAL_C} ${NO_WSOMETIMES_UNINITIALIZED}" dev/bwn/if_bwn_pci.c optional bwn pci bhnd dev/bwn/if_bwn_phy_common.c optional bwn siba_bwn dev/bwn/if_bwn_phy_g.c optional bwn siba_bwn \ compile-with "${NORMAL_C} ${NO_WSOMETIMES_UNINITIALIZED} ${NO_WCONSTANT_CONVERSION}" dev/bwn/if_bwn_phy_lp.c optional bwn siba_bwn \ compile-with "${NORMAL_C} ${NO_WSOMETIMES_UNINITIALIZED}" dev/bwn/if_bwn_phy_n.c optional bwn siba_bwn dev/bwn/if_bwn_util.c optional bwn siba_bwn dev/bwn/bwn_mac.c optional bwn bhnd dev/cardbus/cardbus.c optional cardbus dev/cardbus/cardbus_cis.c optional cardbus dev/cardbus/cardbus_device.c optional cardbus dev/cas/if_cas.c optional cas dev/cfi/cfi_bus_fdt.c optional cfi fdt dev/cfi/cfi_bus_nexus.c optional cfi dev/cfi/cfi_core.c optional cfi dev/cfi/cfi_dev.c optional cfi dev/cfi/cfi_disk.c optional cfid dev/chromebook_platform/chromebook_platform.c optional chromebook_platform dev/ciss/ciss.c optional ciss dev/cm/smc90cx6.c optional cm dev/cmx/cmx.c optional cmx dev/cmx/cmx_pccard.c optional cmx pccard dev/cpufreq/ichss.c optional cpufreq pci dev/cs/if_cs.c optional cs dev/cs/if_cs_isa.c optional cs isa dev/cs/if_cs_pccard.c optional cs pccard dev/cxgb/cxgb_main.c optional cxgb pci \ compile-with "${NORMAL_C} -I$S/dev/cxgb" dev/cxgb/cxgb_sge.c optional cxgb pci \ compile-with "${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$S/dev/cxgb" dev/cxgbe/t4_if.m optional cxgbe pci dev/cxgbe/t4_iov.c optional cxgbe pci \ compile-with "${NORMAL_C} -I$S/dev/cxgbe" dev/cxgbe/t4_mp_ring.c optional cxgbe pci \ compile-with "${NORMAL_C} -I$S/dev/cxgbe" dev/cxgbe/t4_main.c optional cxgbe pci \ compile-with "${NORMAL_C} -I$S/dev/cxgbe" dev/cxgbe/t4_netmap.c optional cxgbe pci \ compile-with "${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$S/dev/cxgbe" dev/cxgbe/t4_tracer.c optional cxgbe pci \ compile-with "${NORMAL_C} -I$S/dev/cxgbe" dev/cxgbe/t4_vf.c optional cxgbev pci \ compile-with "${NORMAL_C} -I$S/dev/cxgbe" dev/cxgbe/common/t4_hw.c optional cxgbe pci \ compile-with "${NORMAL_C} -I$S/dev/cxgbe" dev/cxgbe/common/t4vf_hw.c optional cxgbev pci \ compile-with "${NORMAL_C} -I$S/dev/cxgbe" t4fw_cfg.c optional cxgbe \ compile-with "${AWK} -f $S/tools/fw_stub.awk t4fw_cfg.fw:t4fw_cfg t4fw_cfg_uwire.fw:t4fw_cfg_uwire t4fw.fw:t4fw -mt4fw_cfg -c${.TARGET}" \ no-implicit-rule before-depend local \ clean "t4fw_cfg.c" t4fw_cfg.fwo optional cxgbe \ dependency "t4fw_cfg.fw" \ compile-with "${NORMAL_FWO}" \ no-implicit-rule \ clean "t4fw_cfg.fwo" t4fw_cfg.fw optional cxgbe \ dependency "$S/dev/cxgbe/firmware/t4fw_cfg.txt" \ compile-with "${CP} ${.ALLSRC} ${.TARGET}" \ no-obj no-implicit-rule \ clean "t4fw_cfg.fw" t4fw_cfg_uwire.fwo optional cxgbe \ dependency "t4fw_cfg_uwire.fw" \ compile-with "${NORMAL_FWO}" \ no-implicit-rule \ clean "t4fw_cfg_uwire.fwo" t4fw_cfg_uwire.fw optional cxgbe \ dependency "$S/dev/cxgbe/firmware/t4fw_cfg_uwire.txt" \ compile-with "${CP} ${.ALLSRC} ${.TARGET}" \ no-obj no-implicit-rule \ clean "t4fw_cfg_uwire.fw" t4fw.fwo optional cxgbe \ dependency "t4fw.fw" \ compile-with "${NORMAL_FWO}" \ no-implicit-rule \ clean "t4fw.fwo" t4fw.fw optional cxgbe \ dependency "$S/dev/cxgbe/firmware/t4fw-1.16.26.0.bin.uu" \ compile-with "${NORMAL_FW}" \ no-obj no-implicit-rule \ clean "t4fw.fw" t5fw_cfg.c optional cxgbe \ compile-with "${AWK} -f $S/tools/fw_stub.awk t5fw_cfg.fw:t5fw_cfg t5fw.fw:t5fw -mt5fw_cfg -c${.TARGET}" \ no-implicit-rule before-depend local \ clean "t5fw_cfg.c" t5fw_cfg.fwo optional cxgbe \ dependency "t5fw_cfg.fw" \ compile-with "${NORMAL_FWO}" \ no-implicit-rule \ clean "t5fw_cfg.fwo" t5fw_cfg.fw optional cxgbe \ dependency "$S/dev/cxgbe/firmware/t5fw_cfg.txt" \ compile-with "${CP} ${.ALLSRC} ${.TARGET}" \ no-obj no-implicit-rule \ clean "t5fw_cfg.fw" t5fw.fwo optional cxgbe \ dependency "t5fw.fw" \ compile-with "${NORMAL_FWO}" \ no-implicit-rule \ clean "t5fw.fwo" t5fw.fw optional cxgbe \ dependency "$S/dev/cxgbe/firmware/t5fw-1.16.26.0.bin.uu" \ compile-with "${NORMAL_FW}" \ no-obj no-implicit-rule \ clean "t5fw.fw" t6fw_cfg.c optional cxgbe \ compile-with "${AWK} -f $S/tools/fw_stub.awk t6fw_cfg.fw:t6fw_cfg t6fw.fw:t6fw -mt6fw_cfg -c${.TARGET}" \ no-implicit-rule before-depend local \ clean "t6fw_cfg.c" t6fw_cfg.fwo optional cxgbe \ dependency "t6fw_cfg.fw" \ compile-with "${NORMAL_FWO}" \ no-implicit-rule \ clean "t6fw_cfg.fwo" t6fw_cfg.fw optional cxgbe \ dependency "$S/dev/cxgbe/firmware/t6fw_cfg.txt" \ compile-with "${CP} ${.ALLSRC} ${.TARGET}" \ no-obj no-implicit-rule \ clean "t6fw_cfg.fw" t6fw.fwo optional cxgbe \ dependency "t6fw.fw" \ compile-with "${NORMAL_FWO}" \ no-implicit-rule \ clean "t6fw.fwo" t6fw.fw optional cxgbe \ dependency "$S/dev/cxgbe/firmware/t6fw-1.16.26.0.bin.uu" \ compile-with "${NORMAL_FW}" \ no-obj no-implicit-rule \ clean "t6fw.fw" dev/cy/cy.c optional cy dev/cy/cy_isa.c optional cy isa dev/cy/cy_pci.c optional cy pci dev/cyapa/cyapa.c optional cyapa iicbus 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/dme/if_dme.c optional dme 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 dev/drm/mga_warp.c optional mgadrm dev/drm/r128_cce.c optional r128drm \ compile-with "${NORMAL_C} ${NO_WCONSTANT_CONVERSION}" dev/drm/r128_drv.c optional r128drm dev/drm/r128_irq.c optional r128drm dev/drm/r128_state.c optional r128drm dev/drm/r300_cmdbuf.c optional radeondrm dev/drm/r600_blit.c optional radeondrm dev/drm/r600_cp.c optional radeondrm \ compile-with "${NORMAL_C} ${NO_WCONSTANT_CONVERSION}" dev/drm/radeon_cp.c optional radeondrm \ compile-with "${NORMAL_C} ${NO_WCONSTANT_CONVERSION}" 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/drm2/drm_agpsupport.c optional drm2 dev/drm2/drm_auth.c optional drm2 dev/drm2/drm_bufs.c optional drm2 dev/drm2/drm_buffer.c optional drm2 dev/drm2/drm_context.c optional drm2 dev/drm2/drm_crtc.c optional drm2 dev/drm2/drm_crtc_helper.c optional drm2 dev/drm2/drm_dma.c optional drm2 dev/drm2/drm_dp_helper.c optional drm2 dev/drm2/drm_dp_iic_helper.c optional drm2 dev/drm2/drm_drv.c optional drm2 dev/drm2/drm_edid.c optional drm2 dev/drm2/drm_fb_helper.c optional drm2 dev/drm2/drm_fops.c optional drm2 dev/drm2/drm_gem.c optional drm2 dev/drm2/drm_gem_names.c optional drm2 dev/drm2/drm_global.c optional drm2 dev/drm2/drm_hashtab.c optional drm2 dev/drm2/drm_ioctl.c optional drm2 dev/drm2/drm_irq.c optional drm2 dev/drm2/drm_linux_list_sort.c optional drm2 dev/drm2/drm_lock.c optional drm2 dev/drm2/drm_memory.c optional drm2 dev/drm2/drm_mm.c optional drm2 dev/drm2/drm_modes.c optional drm2 dev/drm2/drm_pci.c optional drm2 dev/drm2/drm_platform.c optional drm2 dev/drm2/drm_scatter.c optional drm2 dev/drm2/drm_stub.c optional drm2 dev/drm2/drm_sysctl.c optional drm2 dev/drm2/drm_vm.c optional drm2 dev/drm2/drm_os_freebsd.c optional drm2 dev/drm2/ttm/ttm_agp_backend.c optional drm2 dev/drm2/ttm/ttm_lock.c optional drm2 dev/drm2/ttm/ttm_object.c optional drm2 dev/drm2/ttm/ttm_tt.c optional drm2 dev/drm2/ttm/ttm_bo_util.c optional drm2 dev/drm2/ttm/ttm_bo.c optional drm2 dev/drm2/ttm/ttm_bo_manager.c optional drm2 dev/drm2/ttm/ttm_execbuf_util.c optional drm2 dev/drm2/ttm/ttm_memory.c optional drm2 dev/drm2/ttm/ttm_page_alloc.c optional drm2 dev/drm2/ttm/ttm_bo_vm.c optional drm2 dev/drm2/ati_pcigart.c optional drm2 agp pci 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/efidev/efidev.c optional efirt 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$S/dev/e1000" dev/e1000/e1000_i210.c optional em | igb \ compile-with "${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/esp_pci.c optional esp pci dev/esp/ncr53c9x.c optional esp dev/etherswitch/arswitch/arswitch.c optional arswitch dev/etherswitch/arswitch/arswitch_reg.c optional arswitch dev/etherswitch/arswitch/arswitch_phy.c optional arswitch dev/etherswitch/arswitch/arswitch_8216.c optional arswitch dev/etherswitch/arswitch/arswitch_8226.c optional arswitch dev/etherswitch/arswitch/arswitch_8316.c optional arswitch dev/etherswitch/arswitch/arswitch_8327.c optional arswitch dev/etherswitch/arswitch/arswitch_7240.c optional arswitch dev/etherswitch/arswitch/arswitch_9340.c optional arswitch dev/etherswitch/arswitch/arswitch_vlans.c optional arswitch dev/etherswitch/etherswitch.c optional etherswitch dev/etherswitch/etherswitch_if.m optional etherswitch dev/etherswitch/ip17x/ip17x.c optional ip17x dev/etherswitch/ip17x/ip175c.c optional ip17x dev/etherswitch/ip17x/ip175d.c optional ip17x dev/etherswitch/ip17x/ip17x_phy.c optional ip17x dev/etherswitch/ip17x/ip17x_vlans.c optional ip17x dev/etherswitch/miiproxy.c optional miiproxy dev/etherswitch/rtl8366/rtl8366rb.c optional rtl8366rb +dev/etherswitch/e6000sw/e6000sw.c optional e6000sw dev/etherswitch/ukswitch/ukswitch.c optional ukswitch dev/evdev/cdev.c optional evdev dev/evdev/evdev.c optional evdev dev/evdev/evdev_mt.c optional evdev dev/evdev/evdev_utils.c optional evdev dev/evdev/uinput.c optional evdev uinput 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/extres/clk/clk.c optional ext_resources clk fdt dev/extres/clk/clkdev_if.m optional ext_resources clk fdt dev/extres/clk/clknode_if.m optional ext_resources clk fdt dev/extres/clk/clk_bus.c optional ext_resources clk fdt dev/extres/clk/clk_div.c optional ext_resources clk fdt dev/extres/clk/clk_fixed.c optional ext_resources clk fdt dev/extres/clk/clk_gate.c optional ext_resources clk fdt dev/extres/clk/clk_mux.c optional ext_resources clk fdt dev/extres/phy/phy.c optional ext_resources phy fdt dev/extres/phy/phy_if.m optional ext_resources phy fdt dev/extres/hwreset/hwreset.c optional ext_resources hwreset fdt dev/extres/hwreset/hwreset_if.m optional ext_resources hwreset fdt dev/extres/regulator/regdev_if.m optional ext_resources regulator fdt dev/extres/regulator/regnode_if.m optional ext_resources regulator fdt dev/extres/regulator/regulator.c optional ext_resources regulator fdt dev/extres/regulator/regulator_bus.c optional ext_resources regulator fdt dev/extres/regulator/regulator_fixed.c optional ext_resources regulator fdt dev/fatm/if_fatm.c optional fatm pci dev/fb/fbd.c optional fbd | vt dev/fb/fb_if.m standard dev/fb/splash.c optional sc splash dev/fdt/fdt_clock.c optional fdt fdt_clock dev/fdt/fdt_clock_if.m optional fdt fdt_clock dev/fdt/fdt_common.c optional fdt dev/fdt/fdt_pinctrl.c optional fdt fdt_pinctrl dev/fdt/fdt_pinctrl_if.m optional fdt fdt_pinctrl dev/fdt/fdt_slicer.c optional fdt cfi | fdt nand | fdt mx25l dev/fdt/fdt_static_dtb.S optional fdt fdt_dtb_static \ dependency "fdt_dtb_file" dev/fdt/simplebus.c optional fdt dev/fe/if_fe.c optional fe dev/fe/if_fe_pccard.c optional fe pccard dev/filemon/filemon.c optional filemon 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/fxp/inphy.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/gpiobacklight.c optional gpiobacklight fdt dev/gpio/gpiokeys.c optional gpiokeys fdt dev/gpio/gpiokeys_codes.c optional gpiokeys fdt 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 !fdt dev/gpio/gpioled_fdt.c optional gpioled fdt dev/gpio/gpiopower.c optional gpiopower fdt dev/gpio/gpioregulator.c optional gpioregulator fdt ext_resources dev/gpio/gpiospi.c optional gpiospi dev/gpio/gpio_if.m optional gpio dev/gpio/gpiobus_if.m optional gpio dev/gpio/gpiopps.c optional gpiopps dev/gpio/ofw_gpiobus.c optional fdt 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/hwpmc/hwpmc_soft.c optional hwpmc dev/ichiic/ig4_acpi.c optional ig4 acpi iicbus dev/ichiic/ig4_iic.c optional ig4 iicbus dev/ichiic/ig4_pci.c optional ig4 pci iicbus 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/iicbus/ad7418.c optional ad7418 dev/iicbus/ds1307.c optional ds1307 dev/iicbus/ds133x.c optional ds133x dev/iicbus/ds1374.c optional ds1374 dev/iicbus/ds1672.c optional ds1672 dev/iicbus/ds3231.c optional ds3231 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/iicbus/iicoc.c optional iicoc dev/iicbus/lm75.c optional lm75 dev/iicbus/ofw_iicbus.c optional fdt iicbus dev/iicbus/pcf8563.c optional pcf8563 dev/iicbus/s35390a.c optional s35390a dev/iir/iir.c optional iir dev/iir/iir_ctrl.c optional iir dev/iir/iir_pci.c optional iir pci dev/intpm/intpm.c optional intpm pci # XXX Work around clang warning, until maintainer approves fix. dev/ips/ips.c optional ips \ compile-with "${NORMAL_C} ${NO_WSOMETIMES_UNINITIALIZED}" dev/ips/ips_commands.c optional ips dev/ips/ips_disk.c optional ips dev/ips/ips_ioctl.c optional ips dev/ips/ips_pci.c optional ips pci dev/ipw/if_ipw.c optional ipw ipwbssfw.c optional ipwbssfw | ipwfw \ compile-with "${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 "${NORMAL_FWO}" \ no-implicit-rule \ clean "ipw_bss.fwo" ipw_bss.fw optional ipwbssfw | ipwfw \ dependency "$S/contrib/dev/ipw/ipw2100-1.3.fw.uu" \ compile-with "${NORMAL_FW}" \ no-obj no-implicit-rule \ clean "ipw_bss.fw" ipwibssfw.c optional ipwibssfw | ipwfw \ compile-with "${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 "${NORMAL_FWO}" \ no-implicit-rule \ clean "ipw_ibss.fwo" ipw_ibss.fw optional ipwibssfw | ipwfw \ dependency "$S/contrib/dev/ipw/ipw2100-1.3-i.fw.uu" \ compile-with "${NORMAL_FW}" \ no-obj no-implicit-rule \ clean "ipw_ibss.fw" ipwmonitorfw.c optional ipwmonitorfw | ipwfw \ compile-with "${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 "${NORMAL_FWO}" \ no-implicit-rule \ clean "ipw_monitor.fwo" ipw_monitor.fw optional ipwmonitorfw | ipwfw \ dependency "$S/contrib/dev/ipw/ipw2100-1.3-p.fw.uu" \ compile-with "${NORMAL_FW}" \ no-obj no-implicit-rule \ clean "ipw_monitor.fw" dev/iscsi/icl.c optional iscsi | ctl dev/iscsi/icl_conn_if.m optional iscsi | ctl dev/iscsi/icl_soft.c optional iscsi | ctl dev/iscsi/icl_soft_proxy.c optional iscsi | ctl dev/iscsi/iscsi.c optional iscsi scbus dev/iscsi_initiator/iscsi.c optional iscsi_initiator scbus dev/iscsi_initiator/iscsi_subr.c optional iscsi_initiator scbus dev/iscsi_initiator/isc_cam.c optional iscsi_initiator scbus dev/iscsi_initiator/isc_soc.c optional iscsi_initiator scbus dev/iscsi_initiator/isc_sm.c optional iscsi_initiator scbus dev/iscsi_initiator/isc_subr.c optional iscsi_initiator scbus dev/ismt/ismt.c optional ismt dev/isl/isl.c optional isl iicbus dev/isp/isp.c optional isp dev/isp/isp_freebsd.c optional isp dev/isp/isp_library.c optional isp dev/isp/isp_pci.c optional isp pci dev/isp/isp_sbus.c optional isp sbus dev/isp/isp_target.c optional isp dev/ispfw/ispfw.c optional ispfw dev/iwi/if_iwi.c optional iwi iwibssfw.c optional iwibssfw | iwifw \ compile-with "${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 "${NORMAL_FWO}" \ no-implicit-rule \ clean "iwi_bss.fwo" iwi_bss.fw optional iwibssfw | iwifw \ dependency "$S/contrib/dev/iwi/ipw2200-bss.fw.uu" \ compile-with "${NORMAL_FW}" \ no-obj no-implicit-rule \ clean "iwi_bss.fw" iwiibssfw.c optional iwiibssfw | iwifw \ compile-with "${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 "${NORMAL_FWO}" \ no-implicit-rule \ clean "iwi_ibss.fwo" iwi_ibss.fw optional iwiibssfw | iwifw \ dependency "$S/contrib/dev/iwi/ipw2200-ibss.fw.uu" \ compile-with "${NORMAL_FW}" \ no-obj no-implicit-rule \ clean "iwi_ibss.fw" iwimonitorfw.c optional iwimonitorfw | iwifw \ compile-with "${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 "${NORMAL_FWO}" \ no-implicit-rule \ clean "iwi_monitor.fwo" iwi_monitor.fw optional iwimonitorfw | iwifw \ dependency "$S/contrib/dev/iwi/ipw2200-sniffer.fw.uu" \ compile-with "${NORMAL_FW}" \ no-obj no-implicit-rule \ clean "iwi_monitor.fw" dev/iwm/if_iwm.c optional iwm dev/iwm/if_iwm_binding.c optional iwm dev/iwm/if_iwm_led.c optional iwm dev/iwm/if_iwm_mac_ctxt.c optional iwm dev/iwm/if_iwm_pcie_trans.c optional iwm dev/iwm/if_iwm_phy_ctxt.c optional iwm dev/iwm/if_iwm_phy_db.c optional iwm dev/iwm/if_iwm_power.c optional iwm dev/iwm/if_iwm_scan.c optional iwm dev/iwm/if_iwm_time_event.c optional iwm dev/iwm/if_iwm_util.c optional iwm iwm3160fw.c optional iwm3160fw | iwmfw \ compile-with "${AWK} -f $S/tools/fw_stub.awk iwm3160.fw:iwm3160fw -miwm3160fw -c${.TARGET}" \ no-implicit-rule before-depend local \ clean "iwm3160fw.c" iwm3160fw.fwo optional iwm3160fw | iwmfw \ dependency "iwm3160.fw" \ compile-with "${NORMAL_FWO}" \ no-implicit-rule \ clean "iwm3160fw.fwo" iwm3160.fw optional iwm3160fw | iwmfw \ dependency "$S/contrib/dev/iwm/iwm-3160-16.fw.uu" \ compile-with "${NORMAL_FW}" \ no-obj no-implicit-rule \ clean "iwm3160.fw" iwm7260fw.c optional iwm7260fw | iwmfw \ compile-with "${AWK} -f $S/tools/fw_stub.awk iwm7260.fw:iwm7260fw -miwm7260fw -c${.TARGET}" \ no-implicit-rule before-depend local \ clean "iwm7260fw.c" iwm7260fw.fwo optional iwm7260fw | iwmfw \ dependency "iwm7260.fw" \ compile-with "${NORMAL_FWO}" \ no-implicit-rule \ clean "iwm7260fw.fwo" iwm7260.fw optional iwm7260fw | iwmfw \ dependency "$S/contrib/dev/iwm/iwm-7260-16.fw.uu" \ compile-with "${NORMAL_FW}" \ no-obj no-implicit-rule \ clean "iwm7260.fw" iwm7265fw.c optional iwm7265fw | iwmfw \ compile-with "${AWK} -f $S/tools/fw_stub.awk iwm7265.fw:iwm7265fw -miwm7265fw -c${.TARGET}" \ no-implicit-rule before-depend local \ clean "iwm7265fw.c" iwm7265fw.fwo optional iwm7265fw | iwmfw \ dependency "iwm7265.fw" \ compile-with "${NORMAL_FWO}" \ no-implicit-rule \ clean "iwm7265fw.fwo" iwm7265.fw optional iwm7265fw | iwmfw \ dependency "$S/contrib/dev/iwm/iwm-7265-16.fw.uu" \ compile-with "${NORMAL_FW}" \ no-obj no-implicit-rule \ clean "iwm7265.fw" iwm8000Cfw.c optional iwm8000Cfw | iwmfw \ compile-with "${AWK} -f $S/tools/fw_stub.awk iwm8000C.fw:iwm8000Cfw -miwm8000Cfw -c${.TARGET}" \ no-implicit-rule before-depend local \ clean "iwm8000Cfw.c" iwm8000Cfw.fwo optional iwm8000Cfw | iwmfw \ dependency "iwm8000C.fw" \ compile-with "${NORMAL_FWO}" \ no-implicit-rule \ clean "iwm8000Cfw.fwo" iwm8000C.fw optional iwm8000Cfw | iwmfw \ dependency "$S/contrib/dev/iwm/iwm-8000C-16.fw.uu" \ compile-with "${NORMAL_FW}" \ no-obj no-implicit-rule \ clean "iwm8000C.fw" dev/iwn/if_iwn.c optional iwn iwn1000fw.c optional iwn1000fw | iwnfw \ compile-with "${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 "${NORMAL_FWO}" \ no-implicit-rule \ clean "iwn1000fw.fwo" iwn1000.fw optional iwn1000fw | iwnfw \ dependency "$S/contrib/dev/iwn/iwlwifi-1000-39.31.5.1.fw.uu" \ compile-with "${NORMAL_FW}" \ no-obj no-implicit-rule \ clean "iwn1000.fw" iwn100fw.c optional iwn100fw | iwnfw \ compile-with "${AWK} -f $S/tools/fw_stub.awk iwn100.fw:iwn100fw -miwn100fw -c${.TARGET}" \ no-implicit-rule before-depend local \ clean "iwn100fw.c" iwn100fw.fwo optional iwn100fw | iwnfw \ dependency "iwn100.fw" \ compile-with "${NORMAL_FWO}" \ no-implicit-rule \ clean "iwn100fw.fwo" iwn100.fw optional iwn100fw | iwnfw \ dependency "$S/contrib/dev/iwn/iwlwifi-100-39.31.5.1.fw.uu" \ compile-with "${NORMAL_FW}" \ no-obj no-implicit-rule \ clean "iwn100.fw" iwn105fw.c optional iwn105fw | iwnfw \ compile-with "${AWK} -f $S/tools/fw_stub.awk iwn105.fw:iwn105fw -miwn105fw -c${.TARGET}" \ no-implicit-rule before-depend local \ clean "iwn105fw.c" iwn105fw.fwo optional iwn105fw | iwnfw \ dependency "iwn105.fw" \ compile-with "${NORMAL_FWO}" \ no-implicit-rule \ clean "iwn105fw.fwo" iwn105.fw optional iwn105fw | iwnfw \ dependency "$S/contrib/dev/iwn/iwlwifi-105-6-18.168.6.1.fw.uu" \ compile-with "${NORMAL_FW}" \ no-obj no-implicit-rule \ clean "iwn105.fw" iwn135fw.c optional iwn135fw | iwnfw \ compile-with "${AWK} -f $S/tools/fw_stub.awk iwn135.fw:iwn135fw -miwn135fw -c${.TARGET}" \ no-implicit-rule before-depend local \ clean "iwn135fw.c" iwn135fw.fwo optional iwn135fw | iwnfw \ dependency "iwn135.fw" \ compile-with "${NORMAL_FWO}" \ no-implicit-rule \ clean "iwn135fw.fwo" iwn135.fw optional iwn135fw | iwnfw \ dependency "$S/contrib/dev/iwn/iwlwifi-135-6-18.168.6.1.fw.uu" \ compile-with "${NORMAL_FW}" \ no-obj no-implicit-rule \ clean "iwn135.fw" iwn2000fw.c optional iwn2000fw | iwnfw \ compile-with "${AWK} -f $S/tools/fw_stub.awk iwn2000.fw:iwn2000fw -miwn2000fw -c${.TARGET}" \ no-implicit-rule before-depend local \ clean "iwn2000fw.c" iwn2000fw.fwo optional iwn2000fw | iwnfw \ dependency "iwn2000.fw" \ compile-with "${NORMAL_FWO}" \ no-implicit-rule \ clean "iwn2000fw.fwo" iwn2000.fw optional iwn2000fw | iwnfw \ dependency "$S/contrib/dev/iwn/iwlwifi-2000-18.168.6.1.fw.uu" \ compile-with "${NORMAL_FW}" \ no-obj no-implicit-rule \ clean "iwn2000.fw" iwn2030fw.c optional iwn2030fw | iwnfw \ compile-with "${AWK} -f $S/tools/fw_stub.awk iwn2030.fw:iwn2030fw -miwn2030fw -c${.TARGET}" \ no-implicit-rule before-depend local \ clean "iwn2030fw.c" iwn2030fw.fwo optional iwn2030fw | iwnfw \ dependency "iwn2030.fw" \ compile-with "${NORMAL_FWO}" \ no-implicit-rule \ clean "iwn2030fw.fwo" iwn2030.fw optional iwn2030fw | iwnfw \ dependency "$S/contrib/dev/iwn/iwnwifi-2030-18.168.6.1.fw.uu" \ compile-with "${NORMAL_FW}" \ no-obj no-implicit-rule \ clean "iwn2030.fw" iwn4965fw.c optional iwn4965fw | iwnfw \ compile-with "${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 "${NORMAL_FWO}" \ no-implicit-rule \ clean "iwn4965fw.fwo" iwn4965.fw optional iwn4965fw | iwnfw \ dependency "$S/contrib/dev/iwn/iwlwifi-4965-228.61.2.24.fw.uu" \ compile-with "${NORMAL_FW}" \ no-obj no-implicit-rule \ clean "iwn4965.fw" iwn5000fw.c optional iwn5000fw | iwnfw \ compile-with "${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 "${NORMAL_FWO}" \ no-implicit-rule \ clean "iwn5000fw.fwo" iwn5000.fw optional iwn5000fw | iwnfw \ dependency "$S/contrib/dev/iwn/iwlwifi-5000-8.83.5.1.fw.uu" \ compile-with "${NORMAL_FW}" \ no-obj no-implicit-rule \ clean "iwn5000.fw" iwn5150fw.c optional iwn5150fw | iwnfw \ compile-with "${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 "${NORMAL_FWO}" \ no-implicit-rule \ clean "iwn5150fw.fwo" iwn5150.fw optional iwn5150fw | iwnfw \ dependency "$S/contrib/dev/iwn/iwlwifi-5150-8.24.2.2.fw.uu"\ compile-with "${NORMAL_FW}" \ no-obj no-implicit-rule \ clean "iwn5150.fw" iwn6000fw.c optional iwn6000fw | iwnfw \ compile-with "${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 "${NORMAL_FWO}" \ no-implicit-rule \ clean "iwn6000fw.fwo" iwn6000.fw optional iwn6000fw | iwnfw \ dependency "$S/contrib/dev/iwn/iwlwifi-6000-9.221.4.1.fw.uu" \ compile-with "${NORMAL_FW}" \ no-obj no-implicit-rule \ clean "iwn6000.fw" iwn6000g2afw.c optional iwn6000g2afw | iwnfw \ compile-with "${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 "${NORMAL_FWO}" \ no-implicit-rule \ clean "iwn6000g2afw.fwo" iwn6000g2a.fw optional iwn6000g2afw | iwnfw \ dependency "$S/contrib/dev/iwn/iwlwifi-6000g2a-18.168.6.1.fw.uu" \ compile-with "${NORMAL_FW}" \ no-obj no-implicit-rule \ clean "iwn6000g2a.fw" iwn6000g2bfw.c optional iwn6000g2bfw | iwnfw \ compile-with "${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 "${NORMAL_FWO}" \ no-implicit-rule \ clean "iwn6000g2bfw.fwo" iwn6000g2b.fw optional iwn6000g2bfw | iwnfw \ dependency "$S/contrib/dev/iwn/iwlwifi-6000g2b-18.168.6.1.fw.uu" \ compile-with "${NORMAL_FW}" \ no-obj no-implicit-rule \ clean "iwn6000g2b.fw" iwn6050fw.c optional iwn6050fw | iwnfw \ compile-with "${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 "${NORMAL_FWO}" \ no-implicit-rule \ clean "iwn6050fw.fwo" iwn6050.fw optional iwn6050fw | iwnfw \ dependency "$S/contrib/dev/iwn/iwlwifi-6050-41.28.5.1.fw.uu" \ compile-with "${NORMAL_FW}" \ no-obj no-implicit-rule \ clean "iwn6050.fw" dev/ixgb/if_ixgb.c optional ixgb dev/ixgb/ixgb_ee.c optional ixgb dev/ixgb/ixgb_hw.c optional ixgb dev/ixgbe/if_ix.c optional ix inet \ compile-with "${NORMAL_C} -I$S/dev/ixgbe -DSMP" dev/ixgbe/if_ixv.c optional ixv inet \ compile-with "${NORMAL_C} -I$S/dev/ixgbe -DSMP" dev/ixgbe/ix_txrx.c optional ix inet | ixv inet \ compile-with "${NORMAL_C} -I$S/dev/ixgbe" dev/ixgbe/ixgbe_osdep.c optional ix inet | ixv inet \ compile-with "${NORMAL_C} -I$S/dev/ixgbe" dev/ixgbe/ixgbe_phy.c optional ix inet | ixv inet \ compile-with "${NORMAL_C} -I$S/dev/ixgbe" dev/ixgbe/ixgbe_api.c optional ix inet | ixv inet \ compile-with "${NORMAL_C} -I$S/dev/ixgbe" dev/ixgbe/ixgbe_common.c optional ix inet | ixv inet \ compile-with "${NORMAL_C} -I$S/dev/ixgbe" dev/ixgbe/ixgbe_mbx.c optional ix inet | ixv inet \ compile-with "${NORMAL_C} -I$S/dev/ixgbe" dev/ixgbe/ixgbe_vf.c optional ix inet | ixv inet \ compile-with "${NORMAL_C} -I$S/dev/ixgbe" dev/ixgbe/ixgbe_82598.c optional ix inet | ixv inet \ compile-with "${NORMAL_C} -I$S/dev/ixgbe" dev/ixgbe/ixgbe_82599.c optional ix inet | ixv inet \ compile-with "${NORMAL_C} -I$S/dev/ixgbe" dev/ixgbe/ixgbe_x540.c optional ix inet | ixv inet \ compile-with "${NORMAL_C} -I$S/dev/ixgbe" dev/ixgbe/ixgbe_x550.c optional ix inet | ixv inet \ compile-with "${NORMAL_C} -I$S/dev/ixgbe" dev/ixgbe/ixgbe_dcb.c optional ix inet | ixv inet \ compile-with "${NORMAL_C} -I$S/dev/ixgbe" dev/ixgbe/ixgbe_dcb_82598.c optional ix inet | ixv inet \ compile-with "${NORMAL_C} -I$S/dev/ixgbe" dev/ixgbe/ixgbe_dcb_82599.c optional ix inet | ixv inet \ compile-with "${NORMAL_C} -I$S/dev/ixgbe" dev/jedec_ts/jedec_ts.c optional jedec_ts smbus dev/jme/if_jme.c optional jme pci dev/joy/joy.c optional joy dev/joy/joy_isa.c optional joy isa dev/kbd/kbd.c optional atkbd | pckbd | sc | ukbd | vt 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/md/md.c optional md dev/mdio/mdio_if.m optional miiproxy | mdio dev/mdio/mdio.c optional miiproxy | mdio 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_syspd.c optional mfi dev/mfi/mfi_tbolt.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 dev/mii/gentbi.c optional miibus | gentbi dev/mii/icsphy.c optional miibus | icsphy dev/mii/ip1000phy.c optional miibus | ip1000phy dev/mii/jmphy.c optional miibus | jmphy dev/mii/lxtphy.c optional miibus | lxtphy dev/mii/micphy.c optional miibus fdt | micphy fdt dev/mii/mii.c optional miibus | mii dev/mii/mii_bitbang.c optional miibus | mii_bitbang 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 dev/mii/smcphy.c optional miibus | smcphy dev/mii/smscphy.c optional miibus | smscphy 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/mpr/mpr.c optional mpr dev/mpr/mpr_config.c optional mpr # XXX Work around clang warning, until maintainer approves fix. dev/mpr/mpr_mapping.c optional mpr \ compile-with "${NORMAL_C} ${NO_WSOMETIMES_UNINITIALIZED}" dev/mpr/mpr_pci.c optional mpr pci dev/mpr/mpr_sas.c optional mpr \ compile-with "${NORMAL_C} ${NO_WUNNEEDED_INTERNAL_DECL}" dev/mpr/mpr_sas_lsi.c optional mpr dev/mpr/mpr_table.c optional mpr dev/mpr/mpr_user.c optional mpr dev/mps/mps.c optional mps dev/mps/mps_config.c optional mps # XXX Work around clang warning, until maintainer approves fix. dev/mps/mps_mapping.c optional mps \ compile-with "${NORMAL_C} ${NO_WSOMETIMES_UNINITIALIZED}" dev/mps/mps_pci.c optional mps pci dev/mps/mps_sas.c optional mps \ compile-with "${NORMAL_C} ${NO_WUNNEEDED_INTERNAL_DECL}" dev/mps/mps_sas_lsi.c optional mps dev/mps/mps_table.c optional mps dev/mps/mps_user.c optional mps dev/mpt/mpt.c optional mpt dev/mpt/mpt_cam.c optional mpt dev/mpt/mpt_debug.c optional mpt dev/mpt/mpt_pci.c optional mpt pci dev/mpt/mpt_raid.c optional mpt dev/mpt/mpt_user.c optional mpt dev/mrsas/mrsas.c optional mrsas dev/mrsas/mrsas_cam.c optional mrsas dev/mrsas/mrsas_ioctl.c optional mrsas dev/mrsas/mrsas_fp.c optional mrsas dev/msk/if_msk.c optional msk dev/mvs/mvs.c optional mvs dev/mvs/mvs_if.m optional mvs dev/mvs/mvs_pci.c optional mvs pci dev/mwl/if_mwl.c optional mwl dev/mwl/if_mwl_pci.c optional mwl pci dev/mwl/mwlhal.c optional mwl mwlfw.c optional mwlfw \ compile-with "${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 "${NORMAL_FWO}" \ no-implicit-rule \ clean "mw88W8363.fwo" mw88W8363.fw optional mwlfw \ dependency "$S/contrib/dev/mwl/mw88W8363.fw.uu" \ compile-with "${NORMAL_FW}" \ no-obj no-implicit-rule \ clean "mw88W8363.fw" mwlboot.fwo optional mwlfw \ dependency "mwlboot.fw" \ compile-with "${NORMAL_FWO}" \ no-implicit-rule \ clean "mwlboot.fwo" mwlboot.fw optional mwlfw \ dependency "$S/contrib/dev/mwl/mwlboot.fw.uu" \ compile-with "${NORMAL_FW}" \ no-obj no-implicit-rule \ clean "mwlboot.fw" dev/mxge/if_mxge.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/nand/nand.c optional nand dev/nand/nand_bbt.c optional nand dev/nand/nand_cdev.c optional nand dev/nand/nand_generic.c optional nand dev/nand/nand_geom.c optional nand dev/nand/nand_id.c optional nand dev/nand/nandbus.c optional nand dev/nand/nandbus_if.m optional nand dev/nand/nand_if.m optional nand dev/nand/nandsim.c optional nandsim nand dev/nand/nandsim_chip.c optional nandsim nand dev/nand/nandsim_ctrl.c optional nandsim nand dev/nand/nandsim_log.c optional nandsim nand dev/nand/nandsim_swap.c optional nandsim nand dev/nand/nfc_if.m optional nand dev/ncr/ncr.c optional ncr pci dev/ncv/ncr53c500.c optional ncv dev/ncv/ncr53c500_pccard.c optional ncv pccard dev/netmap/if_ptnet.c optional netmap inet dev/netmap/netmap.c optional netmap dev/netmap/netmap_freebsd.c optional netmap dev/netmap/netmap_generic.c optional netmap dev/netmap/netmap_mbq.c optional netmap dev/netmap/netmap_mem2.c optional netmap dev/netmap/netmap_monitor.c optional netmap dev/netmap/netmap_offloadings.c optional netmap dev/netmap/netmap_pipe.c optional netmap dev/netmap/netmap_pt.c optional netmap dev/netmap/netmap_vale.c optional netmap # compile-with "${NORMAL_C} -Wconversion -Wextra" dev/nfsmb/nfsmb.c optional nfsmb pci dev/nge/if_nge.c optional nge dev/nxge/if_nxge.c optional nxge \ compile-with "${NORMAL_C} ${NO_WSELF_ASSIGN}" dev/nxge/xgehal/xgehal-device.c optional nxge \ compile-with "${NORMAL_C} ${NO_WSELF_ASSIGN}" dev/nxge/xgehal/xgehal-mm.c optional nxge dev/nxge/xgehal/xge-queue.c optional nxge dev/nxge/xgehal/xgehal-driver.c optional nxge \ compile-with "${NORMAL_C} ${NO_WSELF_ASSIGN}" dev/nxge/xgehal/xgehal-ring.c optional nxge \ compile-with "${NORMAL_C} ${NO_WSELF_ASSIGN}" dev/nxge/xgehal/xgehal-channel.c optional nxge \ compile-with "${NORMAL_C} ${NO_WSELF_ASSIGN}" dev/nxge/xgehal/xgehal-fifo.c optional nxge \ compile-with "${NORMAL_C} ${NO_WSELF_ASSIGN}" dev/nxge/xgehal/xgehal-stats.c optional nxge \ compile-with "${NORMAL_C} ${NO_WSELF_ASSIGN}" dev/nxge/xgehal/xgehal-config.c optional nxge dev/nxge/xgehal/xgehal-mgmt.c optional nxge \ compile-with "${NORMAL_C} ${NO_WSELF_ASSIGN}" 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/oce/oce_hw.c optional oce pci dev/oce/oce_if.c optional oce pci dev/oce/oce_mbox.c optional oce pci dev/oce/oce_queue.c optional oce pci dev/oce/oce_sysctl.c optional oce pci dev/oce/oce_util.c optional oce pci dev/ofw/ofw_bus_if.m optional fdt dev/ofw/ofw_bus_subr.c optional fdt dev/ofw/ofw_cpu.c optional fdt dev/ofw/ofw_fdt.c optional fdt dev/ofw/ofw_if.m optional fdt dev/ofw/ofw_subr.c optional fdt dev/ofw/ofwbus.c optional fdt dev/ofw/openfirm.c optional fdt dev/ofw/openfirmio.c optional fdt dev/ow/ow.c optional ow \ dependency "owll_if.h" \ dependency "own_if.h" dev/ow/owll_if.m optional ow dev/ow/own_if.m optional ow dev/ow/ow_temp.c optional ow_temp dev/ow/owc_gpiobus.c optional owc gpio 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_iov.c optional pci pci_iov dev/pci/pci_iov_if.m standard dev/pci/pci_iov_schema.c optional pci pci_iov dev/pci/pci_pci.c optional pci dev/pci/pci_subr.c optional pci dev/pci/pci_user.c optional pci dev/pci/pcib_if.m standard dev/pci/pcib_support.c 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/pms/freebsd/driver/ini/src/agtiapi.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/sallsdk/spc/sadisc.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/sallsdk/spc/mpi.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/sallsdk/spc/saframe.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/sallsdk/spc/sahw.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/sallsdk/spc/sainit.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/sallsdk/spc/saint.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/sallsdk/spc/sampicmd.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/sallsdk/spc/sampirsp.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/sallsdk/spc/saphy.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/sallsdk/spc/saport.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/sallsdk/spc/sasata.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/sallsdk/spc/sasmp.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/sallsdk/spc/sassp.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/sallsdk/spc/satimer.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/sallsdk/spc/sautil.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/sallsdk/spc/saioctlcmd.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/sallsdk/spc/mpidebug.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/discovery/dm/dminit.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/discovery/dm/dmsmp.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/discovery/dm/dmdisc.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/discovery/dm/dmport.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/discovery/dm/dmtimer.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/discovery/dm/dmmisc.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/sat/src/sminit.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/sat/src/smmisc.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/sat/src/smsat.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/sat/src/smsatcb.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/sat/src/smsathw.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/sat/src/smtimer.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/tisa/sassata/common/tdinit.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/tisa/sassata/common/tdmisc.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/tisa/sassata/common/tdesgl.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/tisa/sassata/common/tdport.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/tisa/sassata/common/tdint.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/tisa/sassata/common/tdioctl.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/tisa/sassata/common/tdhw.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/tisa/sassata/common/ossacmnapi.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/tisa/sassata/common/tddmcmnapi.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/tisa/sassata/common/tdsmcmnapi.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/tisa/sassata/common/tdtimers.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/tisa/sassata/sas/ini/itdio.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/tisa/sassata/sas/ini/itdcb.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/tisa/sassata/sas/ini/itdinit.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/tisa/sassata/sas/ini/itddisc.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/tisa/sassata/sata/host/sat.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/tisa/sassata/sata/host/ossasat.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/pms/RefTisa/tisa/sassata/sata/host/sathw.c optional pmspcv \ compile-with "${NORMAL_C} -Wunused-variable -Woverflow -Wparentheses -w" dev/ppbus/if_plip.c optional plip dev/ppbus/immio.c optional vpo dev/ppbus/lpbb.c optional lpbb dev/ppbus/lpt.c optional lpt dev/ppbus/pcfclock.c optional pcfclock dev/ppbus/ppb_1284.c optional ppbus dev/ppbus/ppb_base.c optional ppbus dev/ppbus/ppb_msq.c optional ppbus dev/ppbus/ppbconf.c optional ppbus dev/ppbus/ppbus_if.m optional ppbus dev/ppbus/ppi.c optional ppi dev/ppbus/pps.c optional pps dev/ppbus/vpo.c optional vpo dev/ppbus/vpoio.c optional vpo dev/ppc/ppc.c optional ppc dev/ppc/ppc_acpi.c optional ppc acpi dev/ppc/ppc_isa.c optional ppc isa dev/ppc/ppc_pci.c optional ppc pci dev/ppc/ppc_puc.c optional ppc puc dev/proto/proto_bus_isa.c optional proto acpi | proto isa dev/proto/proto_bus_pci.c optional proto pci dev/proto/proto_busdma.c optional proto dev/proto/proto_core.c optional proto dev/pst/pst-iop.c optional pst dev/pst/pst-pci.c optional pst pci dev/pst/pst-raid.c optional pst dev/pty/pty.c optional pty dev/puc/puc.c optional puc dev/puc/puc_cfg.c optional puc dev/puc/puc_pccard.c optional puc pccard dev/puc/puc_pci.c optional puc pci dev/puc/pucdata.c optional puc pci dev/quicc/quicc_core.c optional quicc dev/ral/rt2560.c optional ral dev/ral/rt2661.c optional ral dev/ral/rt2860.c optional ral dev/ral/if_ral_pci.c optional ral pci rt2561fw.c optional rt2561fw | ralfw \ compile-with "${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 "${NORMAL_FWO}" \ no-implicit-rule \ clean "rt2561fw.fwo" rt2561.fw optional rt2561fw | ralfw \ dependency "$S/contrib/dev/ral/rt2561.fw.uu" \ compile-with "${NORMAL_FW}" \ no-obj no-implicit-rule \ clean "rt2561.fw" rt2561sfw.c optional rt2561sfw | ralfw \ compile-with "${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 "${NORMAL_FWO}" \ no-implicit-rule \ clean "rt2561sfw.fwo" rt2561s.fw optional rt2561sfw | ralfw \ dependency "$S/contrib/dev/ral/rt2561s.fw.uu" \ compile-with "${NORMAL_FW}" \ no-obj no-implicit-rule \ clean "rt2561s.fw" rt2661fw.c optional rt2661fw | ralfw \ compile-with "${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 "${NORMAL_FWO}" \ no-implicit-rule \ clean "rt2661fw.fwo" rt2661.fw optional rt2661fw | ralfw \ dependency "$S/contrib/dev/ral/rt2661.fw.uu" \ compile-with "${NORMAL_FW}" \ no-obj no-implicit-rule \ clean "rt2661.fw" rt2860fw.c optional rt2860fw | ralfw \ compile-with "${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 "${NORMAL_FWO}" \ no-implicit-rule \ clean "rt2860fw.fwo" rt2860.fw optional rt2860fw | ralfw \ dependency "$S/contrib/dev/ral/rt2860.fw.uu" \ compile-with "${NORMAL_FW}" \ no-obj no-implicit-rule \ clean "rt2860.fw" dev/random/random_infra.c optional random dev/random/random_harvestq.c optional random dev/random/randomdev.c optional random random_yarrow | \ random !random_yarrow !random_loadable dev/random/yarrow.c optional random random_yarrow dev/random/fortuna.c optional random !random_yarrow !random_loadable dev/random/hash.c optional random random_yarrow | \ random !random_yarrow !random_loadable dev/rc/rc.c optional rc dev/rccgpio/rccgpio.c optional rccgpio gpio dev/re/if_re.c optional re dev/rl/if_rl.c optional rl pci 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/rtwn/if_rtwn.c optional rtwn dev/rtwn/if_rtwn_beacon.c optional rtwn dev/rtwn/if_rtwn_calib.c optional rtwn dev/rtwn/if_rtwn_cam.c optional rtwn dev/rtwn/if_rtwn_efuse.c optional rtwn dev/rtwn/if_rtwn_fw.c optional rtwn dev/rtwn/if_rtwn_rx.c optional rtwn dev/rtwn/if_rtwn_task.c optional rtwn dev/rtwn/if_rtwn_tx.c optional rtwn # dev/rtwn/pci/rtwn_pci_attach.c optional rtwn_pci pci dev/rtwn/pci/rtwn_pci_reg.c optional rtwn_pci pci dev/rtwn/pci/rtwn_pci_rx.c optional rtwn_pci pci dev/rtwn/pci/rtwn_pci_tx.c optional rtwn_pci pci # dev/rtwn/usb/rtwn_usb_attach.c optional rtwn_usb dev/rtwn/usb/rtwn_usb_ep.c optional rtwn_usb dev/rtwn/usb/rtwn_usb_reg.c optional rtwn_usb dev/rtwn/usb/rtwn_usb_rx.c optional rtwn_usb dev/rtwn/usb/rtwn_usb_tx.c optional rtwn_usb # RTL8188E dev/rtwn/rtl8188e/r88e_beacon.c optional rtwn dev/rtwn/rtl8188e/r88e_calib.c optional rtwn dev/rtwn/rtl8188e/r88e_chan.c optional rtwn dev/rtwn/rtl8188e/r88e_fw.c optional rtwn dev/rtwn/rtl8188e/r88e_init.c optional rtwn dev/rtwn/rtl8188e/r88e_led.c optional rtwn dev/rtwn/rtl8188e/r88e_tx.c optional rtwn dev/rtwn/rtl8188e/r88e_rf.c optional rtwn dev/rtwn/rtl8188e/r88e_rom.c optional rtwn dev/rtwn/rtl8188e/r88e_rx.c optional rtwn dev/rtwn/rtl8188e/usb/r88eu_attach.c optional rtwn_usb dev/rtwn/rtl8188e/usb/r88eu_init.c optional rtwn_usb dev/rtwn/rtl8188e/usb/r88eu_rx.c optional rtwn_usb # RTL8192C dev/rtwn/rtl8192c/r92c_attach.c optional rtwn dev/rtwn/rtl8192c/r92c_beacon.c optional rtwn dev/rtwn/rtl8192c/r92c_calib.c optional rtwn dev/rtwn/rtl8192c/r92c_chan.c optional rtwn dev/rtwn/rtl8192c/r92c_fw.c optional rtwn dev/rtwn/rtl8192c/r92c_init.c optional rtwn dev/rtwn/rtl8192c/r92c_rf.c optional rtwn dev/rtwn/rtl8192c/r92c_rom.c optional rtwn dev/rtwn/rtl8192c/r92c_rx.c optional rtwn dev/rtwn/rtl8192c/r92c_tx.c optional rtwn dev/rtwn/rtl8192c/pci/r92ce_attach.c optional rtwn_pci pci dev/rtwn/rtl8192c/pci/r92ce_calib.c optional rtwn_pci pci dev/rtwn/rtl8192c/pci/r92ce_fw.c optional rtwn_pci pci dev/rtwn/rtl8192c/pci/r92ce_init.c optional rtwn_pci pci dev/rtwn/rtl8192c/pci/r92ce_led.c optional rtwn_pci pci dev/rtwn/rtl8192c/pci/r92ce_rx.c optional rtwn_pci pci dev/rtwn/rtl8192c/pci/r92ce_tx.c optional rtwn_pci pci dev/rtwn/rtl8192c/usb/r92cu_attach.c optional rtwn_usb dev/rtwn/rtl8192c/usb/r92cu_init.c optional rtwn_usb dev/rtwn/rtl8192c/usb/r92cu_led.c optional rtwn_usb dev/rtwn/rtl8192c/usb/r92cu_rx.c optional rtwn_usb dev/rtwn/rtl8192c/usb/r92cu_tx.c optional rtwn_usb # RTL8812A dev/rtwn/rtl8812a/r12a_beacon.c optional rtwn dev/rtwn/rtl8812a/r12a_calib.c optional rtwn dev/rtwn/rtl8812a/r12a_caps.c optional rtwn dev/rtwn/rtl8812a/r12a_chan.c optional rtwn dev/rtwn/rtl8812a/r12a_fw.c optional rtwn dev/rtwn/rtl8812a/r12a_init.c optional rtwn dev/rtwn/rtl8812a/r12a_led.c optional rtwn dev/rtwn/rtl8812a/r12a_rf.c optional rtwn dev/rtwn/rtl8812a/r12a_rom.c optional rtwn dev/rtwn/rtl8812a/r12a_rx.c optional rtwn dev/rtwn/rtl8812a/r12a_tx.c optional rtwn dev/rtwn/rtl8812a/usb/r12au_attach.c optional rtwn_usb dev/rtwn/rtl8812a/usb/r12au_init.c optional rtwn_usb dev/rtwn/rtl8812a/usb/r12au_rx.c optional rtwn_usb dev/rtwn/rtl8812a/usb/r12au_tx.c optional rtwn_usb # RTL8821A dev/rtwn/rtl8821a/r21a_beacon.c optional rtwn dev/rtwn/rtl8821a/r21a_calib.c optional rtwn dev/rtwn/rtl8821a/r21a_chan.c optional rtwn dev/rtwn/rtl8821a/r21a_fw.c optional rtwn dev/rtwn/rtl8821a/r21a_init.c optional rtwn dev/rtwn/rtl8821a/r21a_led.c optional rtwn dev/rtwn/rtl8821a/r21a_rom.c optional rtwn dev/rtwn/rtl8821a/r21a_rx.c optional rtwn dev/rtwn/rtl8821a/usb/r21au_attach.c optional rtwn_usb dev/rtwn/rtl8821a/usb/r21au_init.c optional rtwn_usb rtwn-rtl8188eufw.c optional rtwn-rtl8188eufw | rtwnfw \ compile-with "${AWK} -f $S/tools/fw_stub.awk rtwn-rtl8188eufw.fw:rtwn-rtl8188eufw:111 -mrtwn-rtl8188eufw -c${.TARGET}" \ no-implicit-rule before-depend local \ clean "rtwn-rtl8188eufw.c" rtwn-rtl8188eufw.fwo optional rtwn-rtl8188eufw | rtwnfw \ dependency "rtwn-rtl8188eufw.fw" \ compile-with "${NORMAL_FWO}" \ no-implicit-rule \ clean "rtwn-rtl8188eufw.fwo" rtwn-rtl8188eufw.fw optional rtwn-rtl8188eufw | rtwnfw \ dependency "$S/contrib/dev/rtwn/rtwn-rtl8188eufw.fw.uu" \ compile-with "${NORMAL_FW}" \ no-obj no-implicit-rule \ clean "rtwn-rtl8188eufw.fw" rtwn-rtl8192cfwE.c optional rtwn-rtl8192cfwE | rtwnfw \ compile-with "${AWK} -f $S/tools/fw_stub.awk rtwn-rtl8192cfwE.fw:rtwn-rtl8192cfwE:111 -mrtwn-rtl8192cfwE -c${.TARGET}" \ no-implicit-rule before-depend local \ clean "rtwn-rtl8192cfwE.c" rtwn-rtl8192cfwE.fwo optional rtwn-rtl8192cfwE | rtwnfw \ dependency "rtwn-rtl8192cfwE.fw" \ compile-with "${NORMAL_FWO}" \ no-implicit-rule \ clean "rtwn-rtl8192cfwE.fwo" rtwn-rtl8192cfwE.fw optional rtwn-rtl8192cfwE | rtwnfw \ dependency "$S/contrib/dev/rtwn/rtwn-rtl8192cfwE.fw.uu" \ compile-with "${NORMAL_FW}" \ no-obj no-implicit-rule \ clean "rtwn-rtl8192cfwE.fw" rtwn-rtl8192cfwE_B.c optional rtwn-rtl8192cfwE_B | rtwnfw \ compile-with "${AWK} -f $S/tools/fw_stub.awk rtwn-rtl8192cfwE_B.fw:rtwn-rtl8192cfwE_B:111 -mrtwn-rtl8192cfwE_B -c${.TARGET}" \ no-implicit-rule before-depend local \ clean "rtwn-rtl8192cfwE_B.c" rtwn-rtl8192cfwE_B.fwo optional rtwn-rtl8192cfwE_B | rtwnfw \ dependency "rtwn-rtl8192cfwE_B.fw" \ compile-with "${NORMAL_FWO}" \ no-implicit-rule \ clean "rtwn-rtl8192cfwE_B.fwo" rtwn-rtl8192cfwE_B.fw optional rtwn-rtl8192cfwE_B | rtwnfw \ dependency "$S/contrib/dev/rtwn/rtwn-rtl8192cfwE_B.fw.uu" \ compile-with "${NORMAL_FW}" \ no-obj no-implicit-rule \ clean "rtwn-rtl8192cfwE_B.fw" rtwn-rtl8192cfwT.c optional rtwn-rtl8192cfwT | rtwnfw \ compile-with "${AWK} -f $S/tools/fw_stub.awk rtwn-rtl8192cfwT.fw:rtwn-rtl8192cfwT:111 -mrtwn-rtl8192cfwT -c${.TARGET}" \ no-implicit-rule before-depend local \ clean "rtwn-rtl8192cfwT.c" rtwn-rtl8192cfwT.fwo optional rtwn-rtl8192cfwT | rtwnfw \ dependency "rtwn-rtl8192cfwT.fw" \ compile-with "${NORMAL_FWO}" \ no-implicit-rule \ clean "rtwn-rtl8192cfwT.fwo" rtwn-rtl8192cfwT.fw optional rtwn-rtl8192cfwT | rtwnfw \ dependency "$S/contrib/dev/rtwn/rtwn-rtl8192cfwT.fw.uu" \ compile-with "${NORMAL_FW}" \ no-obj no-implicit-rule \ clean "rtwn-rtl8192cfwT.fw" rtwn-rtl8192cfwU.c optional rtwn-rtl8192cfwU | rtwnfw \ compile-with "${AWK} -f $S/tools/fw_stub.awk rtwn-rtl8192cfwU.fw:rtwn-rtl8192cfwU:111 -mrtwn-rtl8192cfwU -c${.TARGET}" \ no-implicit-rule before-depend local \ clean "rtwn-rtl8192cfwU.c" rtwn-rtl8192cfwU.fwo optional rtwn-rtl8192cfwU | rtwnfw \ dependency "rtwn-rtl8192cfwU.fw" \ compile-with "${NORMAL_FWO}" \ no-implicit-rule \ clean "rtwn-rtl8192cfwU.fwo" rtwn-rtl8192cfwU.fw optional rtwn-rtl8192cfwU | rtwnfw \ dependency "$S/contrib/dev/rtwn/rtwn-rtl8192cfwU.fw.uu" \ compile-with "${NORMAL_FW}" \ no-obj no-implicit-rule \ clean "rtwn-rtl8192cfwU.fw" rtwn-rtl8812aufw.c optional rtwn-rtl8812aufw | rtwnfw \ compile-with "${AWK} -f $S/tools/fw_stub.awk rtwn-rtl8812aufw.fw:rtwn-rtl8812aufw:111 -mrtwn-rtl8812aufw -c${.TARGET}" \ no-implicit-rule before-depend local \ clean "rtwn-rtl8812aufw.c" rtwn-rtl8812aufw.fwo optional rtwn-rtl8812aufw | rtwnfw \ dependency "rtwn-rtl8812aufw.fw" \ compile-with "${NORMAL_FWO}" \ no-implicit-rule \ clean "rtwn-rtl8812aufw.fwo" rtwn-rtl8812aufw.fw optional rtwn-rtl8812aufw | rtwnfw \ dependency "$S/contrib/dev/rtwn/rtwn-rtl8812aufw.fw.uu" \ compile-with "${NORMAL_FW}" \ no-obj no-implicit-rule \ clean "rtwn-rtl8812aufw.fw" rtwn-rtl8821aufw.c optional rtwn-rtl8821aufw | rtwnfw \ compile-with "${AWK} -f $S/tools/fw_stub.awk rtwn-rtl8821aufw.fw:rtwn-rtl8821aufw:111 -mrtwn-rtl8821aufw -c${.TARGET}" \ no-implicit-rule before-depend local \ clean "rtwn-rtl8821aufw.c" rtwn-rtl8821aufw.fwo optional rtwn-rtl8821aufw | rtwnfw \ dependency "rtwn-rtl8821aufw.fw" \ compile-with "${NORMAL_FWO}" \ no-implicit-rule \ clean "rtwn-rtl8821aufw.fwo" rtwn-rtl8821aufw.fw optional rtwn-rtl8821aufw | rtwnfw \ dependency "$S/contrib/dev/rtwn/rtwn-rtl8821aufw.fw.uu" \ compile-with "${NORMAL_FW}" \ no-obj no-implicit-rule \ clean "rtwn-rtl8821aufw.fw" 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/sdhci/sdhci.c optional sdhci dev/sdhci/sdhci_if.m optional sdhci dev/sdhci/sdhci_pci.c optional sdhci pci dev/sf/if_sf.c optional sf pci dev/sge/if_sge.c optional sge 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/smc/if_smc_fdt.c optional smc fdt 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 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 dev/sound/pci/emu10kx.c optional snd_emu10kx pci dev/sound/pci/emu10kx-pcm.c optional snd_emu10kx pci dev/sound/pci/emu10kx-midi.c optional snd_emu10kx pci 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 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/hdaa.c optional snd_hda pci dev/sound/pci/hda/hdaa_patches.c optional snd_hda pci dev/sound/pci/hda/hdac.c optional snd_hda pci dev/sound/pci/hda/hdac_if.m optional snd_hda pci dev/sound/pci/hda/hdacc.c optional snd_hda pci dev/sound/pci/hdspe.c optional snd_hdspe pci dev/sound/pci/hdspe-pcm.c optional snd_hdspe 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/ofw_spibus.c optional fdt spibus dev/spibus/spibus.c optional spibus \ dependency "spibus_if.h" dev/spibus/spigen.c optional spigen 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/tws/tws.c optional tws dev/tws/tws_cam.c optional tws dev/tws/tws_hdm.c optional tws dev/tws/tws_services.c optional tws dev/tws/tws_user.c optional tws 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_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 | uart uart_snps dev/uart/uart_dev_pl011.c optional uart pl011 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_snps.c optional uart uart_snps fdt 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/dwc_otg.c optional dwcotg dev/usb/controller/dwc_otg_fdt.c optional dwcotg fdt 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_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/saf1761_otg.c optional saf1761otg dev/usb/controller/saf1761_otg_fdt.c optional saf1761otg fdt 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_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_axge.c optional axge 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_smsc.c optional smsc dev/usb/net/if_udav.c optional udav dev/usb/net/if_ure.c optional ure dev/usb/net/if_usie.c optional usie dev/usb/net/if_urndis.c optional urndis dev/usb/net/ruephy.c optional rue dev/usb/net/usb_ethernet.c optional uether | aue | axe | axge | cdce | \ cue | ipheth | kue | mos | rue | \ smsc | udav | ure | urndis dev/usb/net/uhso.c optional uhso # # USB WLAN drivers # dev/usb/wlan/if_rsu.c optional rsu rsu-rtl8712fw.c optional rsu-rtl8712fw | rsufw \ compile-with "${AWK} -f $S/tools/fw_stub.awk rsu-rtl8712fw.fw:rsu-rtl8712fw:120 -mrsu-rtl8712fw -c${.TARGET}" \ no-implicit-rule before-depend local \ clean "rsu-rtl8712fw.c" rsu-rtl8712fw.fwo optional rsu-rtl8712fw | rsufw \ dependency "rsu-rtl8712fw.fw" \ compile-with "${NORMAL_FWO}" \ no-implicit-rule \ clean "rsu-rtl8712fw.fwo" rsu-rtl8712fw.fw optional rsu-rtl8712.fw | rsufw \ dependency "$S/contrib/dev/rsu/rsu-rtl8712fw.fw.uu" \ compile-with "${NORMAL_FW}" \ no-obj no-implicit-rule \ clean "rsu-rtl8712fw.fw" dev/usb/wlan/if_rum.c optional rum dev/usb/wlan/if_run.c optional run runfw.c optional runfw \ compile-with "${AWK} -f $S/tools/fw_stub.awk run.fw:runfw -mrunfw -c${.TARGET}" \ no-implicit-rule before-depend local \ clean "runfw.c" runfw.fwo optional runfw \ dependency "run.fw" \ compile-with "${NORMAL_FWO}" \ no-implicit-rule \ clean "runfw.fwo" run.fw optional runfw \ dependency "$S/contrib/dev/run/rt2870.fw.uu" \ compile-with "${NORMAL_FW}" \ no-obj no-implicit-rule \ clean "run.fw" 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 | usie | \ uslcom | uvisor | uvscom # # USB misc drivers # dev/usb/misc/ufm.c optional ufm dev/usb/misc/udbp.c optional udbp dev/usb/misc/ugold.c optional ugold dev/usb/misc/uled.c optional uled # # 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 dev/usb/input/wsp.c optional wsp # # 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_audio.c optional usb_template dev/usb/template/usb_template_cdce.c optional usb_template dev/usb/template/usb_template_kbd.c optional usb_template dev/usb/template/usb_template_modem.c optional usb_template dev/usb/template/usb_template_mouse.c optional usb_template dev/usb/template/usb_template_msc.c optional usb_template dev/usb/template/usb_template_mtp.c optional usb_template dev/usb/template/usb_template_phone.c optional usb_template dev/usb/template/usb_template_serialnet.c optional usb_template dev/usb/template/usb_template_midi.c optional usb_template # # USB video drivers # dev/usb/video/udl.c optional udl # # USB END # dev/videomode/videomode.c optional videomode dev/videomode/edid.c optional videomode dev/videomode/pickmode.c optional videomode dev/videomode/vesagtf.c optional videomode 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/viapm/viapm.c optional viapm pci dev/virtio/virtio.c optional virtio dev/virtio/virtqueue.c optional virtio dev/virtio/virtio_bus_if.m optional virtio dev/virtio/virtio_if.m optional virtio dev/virtio/pci/virtio_pci.c optional virtio_pci dev/virtio/mmio/virtio_mmio.c optional virtio_mmio fdt dev/virtio/mmio/virtio_mmio_if.m optional virtio_mmio fdt dev/virtio/network/if_vtnet.c optional vtnet dev/virtio/block/virtio_blk.c optional virtio_blk dev/virtio/balloon/virtio_balloon.c optional virtio_balloon dev/virtio/scsi/virtio_scsi.c optional virtio_scsi dev/virtio/random/virtio_random.c optional virtio_random dev/virtio/console/virtio_console.c optional virtio_console dev/vkbd/vkbd.c optional vkbd dev/vr/if_vr.c optional vr pci dev/vt/colors/vt_termcolors.c optional vt dev/vt/font/vt_font_default.c optional vt dev/vt/font/vt_mouse_cursor.c optional vt dev/vt/hw/efifb/efifb.c optional vt_efifb dev/vt/hw/fb/vt_fb.c optional vt dev/vt/hw/vga/vt_vga.c optional vt vt_vga dev/vt/logo/logo_freebsd.c optional vt splash dev/vt/logo/logo_beastie.c optional vt splash dev/vt/vt_buf.c optional vt dev/vt/vt_consolectl.c optional vt dev/vt/vt_core.c optional vt dev/vt/vt_cpulogos.c optional vt splash dev/vt/vt_font.c optional vt dev/vt/vt_sysmouse.c optional vt 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/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/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 "${NORMAL_FWO}" \ no-implicit-rule \ clean "wpifw.fwo" wpi.fw optional wpifw \ dependency "$S/contrib/dev/wpi/iwlwifi-3945-15.32.2.9.fw.uu" \ compile-with "${NORMAL_FW}" \ no-obj no-implicit-rule \ clean "wpi.fw" dev/xdma/xdma.c optional xdma dev/xdma/xdma_if.m optional xdma dev/xdma/xdma_fdt_test.c optional xdma xdma_test fdt dev/xe/if_xe.c optional xe dev/xe/if_xe_pccard.c optional xe pccard dev/xen/balloon/balloon.c optional xenhvm dev/xen/blkfront/blkfront.c optional xenhvm dev/xen/blkback/blkback.c optional xenhvm dev/xen/console/xen_console.c optional xenhvm dev/xen/control/control.c optional xenhvm dev/xen/grant_table/grant_table.c optional xenhvm dev/xen/netback/netback.c optional xenhvm dev/xen/netfront/netfront.c optional xenhvm dev/xen/xenpci/xenpci.c optional xenpci dev/xen/timer/timer.c optional xenhvm dev/xen/pvcpu/pvcpu.c optional xenhvm dev/xen/xenstore/xenstore.c optional xenhvm dev/xen/xenstore/xenstore_dev.c optional xenhvm dev/xen/xenstore/xenstored_dev.c optional xenhvm dev/xen/evtchn/evtchn_dev.c optional xenhvm dev/xen/privcmd/privcmd.c optional xenhvm dev/xen/gntdev/gntdev.c optional xenhvm dev/xen/debug/debug.c optional xenhvm dev/xl/if_xl.c optional xl pci dev/xl/xlphy.c optional xl pci fs/autofs/autofs.c optional autofs fs/autofs/autofs_vfsops.c optional autofs fs/autofs/autofs_vnops.c optional autofs 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/cuse/cuse.c optional cuse fs/fuse/fuse_device.c optional fuse fs/fuse/fuse_file.c optional fuse fs/fuse/fuse_internal.c optional fuse fs/fuse/fuse_io.c optional fuse fs/fuse/fuse_ipc.c optional fuse fs/fuse/fuse_main.c optional fuse fs/fuse/fuse_node.c optional fuse fs/fuse/fuse_vfsops.c optional fuse fs/fuse/fuse_vnops.c optional fuse 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/nandfs/bmap.c optional nandfs fs/nandfs/nandfs_alloc.c optional nandfs fs/nandfs/nandfs_bmap.c optional nandfs fs/nandfs/nandfs_buffer.c optional nandfs fs/nandfs/nandfs_cleaner.c optional nandfs fs/nandfs/nandfs_cpfile.c optional nandfs fs/nandfs/nandfs_dat.c optional nandfs fs/nandfs/nandfs_dir.c optional nandfs fs/nandfs/nandfs_ifile.c optional nandfs fs/nandfs/nandfs_segment.c optional nandfs fs/nandfs/nandfs_subr.c optional nandfs fs/nandfs/nandfs_sufile.c optional nandfs fs/nandfs/nandfs_vfsops.c optional nandfs fs/nandfs/nandfs_vnops.c optional nandfs 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_fha_new.c optional nfsd inet 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/nullfs/null_subr.c optional nullfs fs/nullfs/null_vfsops.c optional nullfs fs/nullfs/null_vnops.c optional nullfs 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_hmac.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_part_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_flashmap.c optional fdt cfi | fdt nand | fdt mx25l 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_gpt 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_label_gpt geom/label/g_label_disk_ident.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_bsd64.c optional geom_part_bsd64 geom/part/g_part_ebr.c optional geom_part_ebr geom/part/g_part_gpt.c optional geom_part_gpt geom/part/g_part_ldm.c optional geom_part_ldm 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_ddf.c 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/raid/tr_raid5.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 contrib/xz-embedded/freebsd/xz_malloc.c \ optional xz_embedded | geom_uzip \ compile-with "${NORMAL_C} -I$S/contrib/xz-embedded/freebsd/ -I$S/contrib/xz-embedded/linux/lib/xz/ -I$S/contrib/xz-embedded/linux/include/linux/" contrib/xz-embedded/linux/lib/xz/xz_crc32.c \ optional xz_embedded | geom_uzip \ compile-with "${NORMAL_C} -I$S/contrib/xz-embedded/freebsd/ -I$S/contrib/xz-embedded/linux/lib/xz/ -I$S/contrib/xz-embedded/linux/include/linux/" contrib/xz-embedded/linux/lib/xz/xz_dec_bcj.c \ optional xz_embedded | geom_uzip \ compile-with "${NORMAL_C} -I$S/contrib/xz-embedded/freebsd/ -I$S/contrib/xz-embedded/linux/lib/xz/ -I$S/contrib/xz-embedded/linux/include/linux/" contrib/xz-embedded/linux/lib/xz/xz_dec_lzma2.c \ optional xz_embedded | geom_uzip \ compile-with "${NORMAL_C} -I$S/contrib/xz-embedded/freebsd/ -I$S/contrib/xz-embedded/linux/lib/xz/ -I$S/contrib/xz-embedded/linux/include/linux/" contrib/xz-embedded/linux/lib/xz/xz_dec_stream.c \ optional xz_embedded | geom_uzip \ compile-with "${NORMAL_C} -I$S/contrib/xz-embedded/freebsd/ -I$S/contrib/xz-embedded/linux/lib/xz/ -I$S/contrib/xz-embedded/linux/include/linux/" geom/uzip/g_uzip.c optional geom_uzip geom/uzip/g_uzip_lzma.c optional geom_uzip geom/uzip/g_uzip_wrkthr.c optional geom_uzip geom/uzip/g_uzip_zlib.c optional geom_uzip geom/vinum/geom_vinum.c optional geom_vinum geom/vinum/geom_vinum_create.c optional geom_vinum geom/vinum/geom_vinum_drive.c optional geom_vinum geom/vinum/geom_vinum_plex.c optional geom_vinum geom/vinum/geom_vinum_volume.c optional geom_vinum geom/vinum/geom_vinum_subr.c optional geom_vinum geom/vinum/geom_vinum_raid5.c optional geom_vinum geom/vinum/geom_vinum_share.c optional geom_vinum geom/vinum/geom_vinum_list.c optional geom_vinum geom/vinum/geom_vinum_rm.c optional geom_vinum geom/vinum/geom_vinum_init.c optional geom_vinum geom/vinum/geom_vinum_state.c optional geom_vinum geom/vinum/geom_vinum_rename.c optional geom_vinum geom/vinum/geom_vinum_move.c optional geom_vinum geom/vinum/geom_vinum_events.c optional geom_vinum 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_extents.c optional ext2fs fs/ext2fs/ext2_inode.c optional ext2fs fs/ext2fs/ext2_inode_cnv.c optional ext2fs fs/ext2fs/ext2_hash.c optional ext2fs fs/ext2fs/ext2_htree.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 # 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_binmisc.c optional imagact_binmisc kern/imgact_elf.c standard kern/imgact_elf32.c optional compat_freebsd32 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_dump.c standard 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_ffclock.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_numa.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_procctl.c standard kern/kern_prot.c standard kern/kern_racct.c standard kern/kern_rangelock.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_sendfile.c standard kern/kern_sharedpage.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 optional ufs_acl | zfs kern/subr_acl_posix1e.c optional ufs_acl kern/subr_autoconf.c standard kern/subr_blist.c standard kern/subr_bus.c standard kern/subr_bus_dma.c standard kern/subr_bufring.c standard kern/subr_capability.c standard kern/subr_clock.c standard kern/subr_counter.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_gtaskqueue.c standard 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_pctrie.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_terminal.c optional vt kern/subr_trap.c standard kern/subr_turnstile.c standard kern/subr_uio.c standard kern/subr_unit.c standard kern/subr_vmem.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_procdesc.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 standard kern/uipc_debug.c optional ddb kern/uipc_domain.c standard kern/uipc_mbuf.c standard kern/uipc_mbuf2.c standard kern/uipc_mbufhash.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 standard 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 "$S/kgssapi/gssd.x" \ compile-with "RPCGEN_CPP='${CPP}' 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_CPP='${CPP}' 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_CPP='${CPP}' 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/asprintf.c standard libkern/bcd.c standard libkern/bsearch.c standard libkern/crc32.c standard libkern/explicit_bzero.c standard libkern/fnmatch.c standard libkern/iconv.c optional libiconv libkern/iconv_converter_if.m optional libiconv libkern/iconv_ucs.c optional libiconv libkern/iconv_xlat.c optional libiconv libkern/iconv_xlat16.c optional libiconv libkern/inet_aton.c standard libkern/inet_ntoa.c standard libkern/inet_ntop.c standard libkern/inet_pton.c standard libkern/jenkins_hash.c standard libkern/murmur3_32.c standard libkern/mcount.c optional profiling-routine libkern/memcchr.c standard libkern/memchr.c standard libkern/memcmp.c standard libkern/memmem.c optional gdb libkern/qsort.c standard libkern/qsort_r.c standard libkern/random.c standard libkern/scanc.c standard libkern/strcasecmp.c standard libkern/strcat.c standard libkern/strchr.c standard libkern/strcmp.c standard libkern/strcpy.c standard libkern/strcspn.c standard libkern/strdup.c standard libkern/strndup.c standard libkern/strlcat.c standard libkern/strlcpy.c standard libkern/strlen.c standard libkern/strncat.c standard libkern/strncmp.c standard libkern/strncpy.c standard libkern/strnlen.c standard libkern/strrchr.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 libkern/timingsafe_bcmp.c standard libkern/zlib.c optional crypto | geom_uzip | ipsec | \ mxge | netgraph_deflate | \ ddb_ctf | gzio net/altq/altq_cbq.c optional altq net/altq/altq_cdnr.c optional altq net/altq/altq_codel.c optional altq net/altq/altq_hfsc.c optional altq net/altq/altq_fairq.c optional altq net/altq/altq_priq.c optional altq net/altq/altq_red.c optional altq net/altq/altq_rio.c optional altq net/altq/altq_rmclass.c optional altq net/altq/altq_subr.c optional altq 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_enc.c optional enc inet | enc inet6 net/if_epair.c optional epair net/if_ethersubr.c optional ether net/if_fddisubr.c optional fddi net/if_fwsubr.c optional fwip net/if_gif.c optional gif inet | gif inet6 | \ netgraph_gif inet | netgraph_gif inet6 net/if_gre.c optional gre inet | gre inet6 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_me.c optional me inet 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/if_vxlan.c optional vxlan inet | vxlan inet6 net/ifdi_if.m optional ether pci net/iflib.c optional ether pci net/mp_ring.c optional ether 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/rss_config.c optional inet rss | inet6 rss net/rtsock.c standard net/slcompress.c optional netgraph_vjc | sppp | \ netgraph_sppp net/toeplitz.c optional inet rss | inet6 rss net/vnet.c optional vimage 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 \ compile-with "${NORMAL_C} -Wno-unused-function" net80211/ieee80211_ageq.c optional wlan net80211/ieee80211_amrr.c optional wlan | wlan_amrr net80211/ieee80211_crypto.c optional wlan \ compile-with "${NORMAL_C} -Wno-unused-function" net80211/ieee80211_crypto_ccmp.c optional wlan wlan_ccmp net80211/ieee80211_crypto_none.c optional wlan net80211/ieee80211_crypto_tkip.c optional wlan wlan_tkip net80211/ieee80211_crypto_wep.c optional wlan wlan_wep net80211/ieee80211_ddb.c optional wlan ddb net80211/ieee80211_dfs.c optional wlan net80211/ieee80211_freebsd.c optional wlan net80211/ieee80211_hostap.c optional wlan \ compile-with "${NORMAL_C} -Wno-unused-function" 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 \ compile-with "${NORMAL_C} -Wno-unused-function" net80211/ieee80211_monitor.c optional wlan net80211/ieee80211_node.c optional wlan net80211/ieee80211_output.c optional wlan net80211/ieee80211_phy.c optional wlan net80211/ieee80211_power.c optional wlan net80211/ieee80211_proto.c optional wlan net80211/ieee80211_radiotap.c optional wlan net80211/ieee80211_ratectl.c optional wlan net80211/ieee80211_ratectl_none.c optional wlan net80211/ieee80211_regdomain.c optional wlan net80211/ieee80211_rssadapt.c optional wlan wlan_rssadapt net80211/ieee80211_scan.c optional wlan net80211/ieee80211_scan_sta.c optional wlan net80211/ieee80211_sta.c optional wlan \ compile-with "${NORMAL_C} -Wno-unused-function" net80211/ieee80211_superg.c optional wlan ieee80211_support_superg net80211/ieee80211_scan_sw.c optional wlan 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 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_frame_relay.c optional netgraph_frame_relay netgraph/ng_gif.c optional netgraph_gif inet6 | netgraph_gif inet 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/in_kdtrace.c optional inet | inet6 netinet/ip_carp.c optional inet carp | inet6 carp netinet/in_fib.c optional inet 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_jail.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_prot.c optional inet | inet6 netinet/in_proto.c optional inet | inet6 netinet/in_rmx.c optional inet netinet/in_rss.c optional inet rss netinet/ip_divert.c optional inet ipdivert ipfirewall netinet/ip_ecn.c optional inet | inet6 netinet/ip_encap.c optional inet | inet6 netinet/ip_fastfwd.c optional inet 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/ip_reass.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_syscalls.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/siftr.c optional inet siftr alq | inet6 siftr alq netinet/tcp_debug.c optional tcpdebug netinet/tcp_fastopen.c optional inet tcp_rfc7413 | inet6 tcp_rfc7413 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 tcp_offload inet | tcp_offload inet6 netinet/tcp_pcap.c optional inet tcppcap | inet6 tcppcap 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_fib.c optional inet6 netinet6/in6_gif.c optional gif inet6 | netgraph_gif inet6 netinet6/in6_ifattach.c optional inet6 netinet6/in6_jail.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_rss.c optional inet6 rss netinet6/in6_src.c optional inet6 netinet6/ip6_fastfwd.c optional inet6 netinet6/ip6_forward.c optional inet6 netinet6/ip6_gre.c optional gre 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_tcp.c optional ipsec inet tcp_signature | \ ipsec inet6 tcp_signature netnatm/natm.c optional natm netnatm/natm_pcb.c optional natm netnatm/natm_proto.c optional natm netpfil/ipfw/dn_aqm_codel.c optional inet dummynet netpfil/ipfw/dn_aqm_pie.c optional inet dummynet netpfil/ipfw/dn_heap.c optional inet dummynet netpfil/ipfw/dn_sched_fifo.c optional inet dummynet netpfil/ipfw/dn_sched_fq_codel.c optional inet dummynet netpfil/ipfw/dn_sched_fq_pie.c optional inet dummynet netpfil/ipfw/dn_sched_prio.c optional inet dummynet netpfil/ipfw/dn_sched_qfq.c optional inet dummynet netpfil/ipfw/dn_sched_rr.c optional inet dummynet netpfil/ipfw/dn_sched_wf2q.c optional inet dummynet netpfil/ipfw/ip_dummynet.c optional inet dummynet netpfil/ipfw/ip_dn_io.c optional inet dummynet netpfil/ipfw/ip_dn_glue.c optional inet dummynet netpfil/ipfw/ip_fw2.c optional inet ipfirewall netpfil/ipfw/ip_fw_bpf.c optional inet ipfirewall netpfil/ipfw/ip_fw_dynamic.c optional inet ipfirewall netpfil/ipfw/ip_fw_eaction.c optional inet ipfirewall netpfil/ipfw/ip_fw_log.c optional inet ipfirewall netpfil/ipfw/ip_fw_pfil.c optional inet ipfirewall netpfil/ipfw/ip_fw_sockopt.c optional inet ipfirewall netpfil/ipfw/ip_fw_table.c optional inet ipfirewall netpfil/ipfw/ip_fw_table_algo.c optional inet ipfirewall netpfil/ipfw/ip_fw_table_value.c optional inet ipfirewall netpfil/ipfw/ip_fw_iface.c optional inet ipfirewall netpfil/ipfw/ip_fw_nat.c optional inet ipfirewall_nat netpfil/ipfw/nat64/ip_fw_nat64.c optional inet inet6 ipfirewall \ ipfirewall_nat64 netpfil/ipfw/nat64/nat64lsn.c optional inet inet6 ipfirewall \ ipfirewall_nat64 netpfil/ipfw/nat64/nat64lsn_control.c optional inet inet6 ipfirewall \ ipfirewall_nat64 netpfil/ipfw/nat64/nat64stl.c optional inet inet6 ipfirewall \ ipfirewall_nat64 netpfil/ipfw/nat64/nat64stl_control.c optional inet inet6 ipfirewall \ ipfirewall_nat64 netpfil/ipfw/nat64/nat64_translate.c optional inet inet6 ipfirewall \ ipfirewall_nat64 netpfil/ipfw/nptv6/ip_fw_nptv6.c optional inet inet6 ipfirewall \ ipfirewall_nptv6 netpfil/ipfw/nptv6/nptv6.c optional inet inet6 ipfirewall \ ipfirewall_nptv6 netpfil/pf/if_pflog.c optional pflog pf inet netpfil/pf/if_pfsync.c optional pfsync pf inet netpfil/pf/pf.c optional pf inet netpfil/pf/pf_if.c optional pf inet netpfil/pf/pf_ioctl.c optional pf inet netpfil/pf/pf_lb.c optional pf inet netpfil/pf/pf_norm.c optional pf inet netpfil/pf/pf_osfp.c optional pf inet netpfil/pf/pf_ruleset.c optional pf inet netpfil/pf/pf_table.c optional pf inet netpfil/pf/in4_cksum.c optional pf inet 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 nfscl nfs/krpc_subr.c optional bootp nfscl nfs/nfs_diskless.c optional nfscl nfs_root nfs/nfs_fha.c optional nfsd nfs/nfs_lock.c optional nfscl | nfslockd | nfsd nfs/nfs_nfssvc.c optional 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 # Linux Kernel Programming Interface compat/linuxkpi/common/src/linux_kmod.c optional compat_linuxkpi \ compile-with "${LINUXKPI_C}" compat/linuxkpi/common/src/linux_compat.c optional compat_linuxkpi \ compile-with "${LINUXKPI_C}" compat/linuxkpi/common/src/linux_pci.c optional compat_linuxkpi pci \ compile-with "${LINUXKPI_C}" compat/linuxkpi/common/src/linux_idr.c optional compat_linuxkpi \ compile-with "${LINUXKPI_C}" compat/linuxkpi/common/src/linux_radix.c optional compat_linuxkpi \ compile-with "${LINUXKPI_C}" compat/linuxkpi/common/src/linux_usb.c optional compat_linuxkpi usb \ compile-with "${LINUXKPI_C}" # OpenFabrics Enterprise Distribution (Infiniband) ofed/drivers/infiniband/core/addr.c optional ofed \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/agent.c optional ofed \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/cache.c optional ofed \ 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 \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/cm.c optional ofed \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/ -Wno-unused-function" ofed/drivers/infiniband/core/cma.c optional ofed \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/device.c optional ofed \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/fmr_pool.c optional ofed \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/iwcm.c optional ofed \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/mad_rmpp.c optional ofed \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/multicast.c optional ofed \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/packer.c optional ofed \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/peer_mem.c optional ofed \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/sa_query.c optional ofed \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/smi.c optional ofed \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/sysfs.c optional ofed \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/ucm.c optional ofed \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/ucma.c optional ofed \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/ud_header.c optional ofed \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/umem.c optional ofed \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/user_mad.c optional ofed \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/uverbs_cmd.c optional ofed \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/uverbs_main.c optional ofed \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/uverbs_marshall.c optional ofed \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/core/verbs.c optional ofed \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/core/" ofed/drivers/infiniband/ulp/ipoib/ipoib_cm.c optional ipoib \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/ulp/ipoib/" #ofed/drivers/infiniband/ulp/ipoib/ipoib_fs.c optional ipoib \ # compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/ulp/ipoib/" ofed/drivers/infiniband/ulp/ipoib/ipoib_ib.c optional ipoib \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/ulp/ipoib/" ofed/drivers/infiniband/ulp/ipoib/ipoib_main.c optional ipoib \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/ulp/ipoib/" ofed/drivers/infiniband/ulp/ipoib/ipoib_multicast.c optional ipoib \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/ulp/ipoib/" ofed/drivers/infiniband/ulp/ipoib/ipoib_verbs.c optional ipoib \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/ulp/ipoib/" #ofed/drivers/infiniband/ulp/ipoib/ipoib_vlan.c optional ipoib \ # compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/ulp/ipoib/" ofed/drivers/infiniband/ulp/sdp/sdp_bcopy.c optional sdp inet \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/ulp/sdp/" ofed/drivers/infiniband/ulp/sdp/sdp_main.c optional sdp inet \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/ulp/sdp/" ofed/drivers/infiniband/ulp/sdp/sdp_rx.c optional sdp inet \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/ulp/sdp/" ofed/drivers/infiniband/ulp/sdp/sdp_cma.c optional sdp inet \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/ulp/sdp/" ofed/drivers/infiniband/ulp/sdp/sdp_tx.c optional sdp inet \ compile-with "${OFED_C} -I$S/ofed/drivers/infiniband/ulp/sdp/" dev/mlx4/mlx4_ib/mlx4_ib_alias_GUID.c optional mlx4ib pci ofed \ compile-with "${OFED_C}" dev/mlx4/mlx4_ib/mlx4_ib_mcg.c optional mlx4ib pci ofed \ compile-with "${OFED_C}" dev/mlx4/mlx4_ib/mlx4_ib_sysfs.c optional mlx4ib pci ofed \ compile-with "${OFED_C}" dev/mlx4/mlx4_ib/mlx4_ib_cm.c optional mlx4ib pci ofed \ compile-with "${OFED_C}" dev/mlx4/mlx4_ib/mlx4_ib_ah.c optional mlx4ib pci ofed \ compile-with "${OFED_C}" dev/mlx4/mlx4_ib/mlx4_ib_cq.c optional mlx4ib pci ofed \ compile-with "${OFED_C}" dev/mlx4/mlx4_ib/mlx4_ib_doorbell.c optional mlx4ib pci ofed \ compile-with "${OFED_C}" dev/mlx4/mlx4_ib/mlx4_ib_mad.c optional mlx4ib pci ofed \ compile-with "${OFED_C}" dev/mlx4/mlx4_ib/mlx4_ib_main.c optional mlx4ib pci ofed \ compile-with "${OFED_C}" dev/mlx4/mlx4_ib/mlx4_ib_exp.c optional mlx4ib pci ofed \ compile-with "${OFED_C}" dev/mlx4/mlx4_ib/mlx4_ib_mr.c optional mlx4ib pci ofed \ compile-with "${OFED_C}" dev/mlx4/mlx4_ib/mlx4_ib_qp.c optional mlx4ib pci ofed \ compile-with "${OFED_C}" dev/mlx4/mlx4_ib/mlx4_ib_srq.c optional mlx4ib pci ofed \ compile-with "${OFED_C}" dev/mlx4/mlx4_ib/mlx4_ib_wc.c optional mlx4ib pci ofed \ compile-with "${OFED_C}" dev/mlx4/mlx4_core/mlx4_alloc.c optional mlx4 pci \ compile-with "${OFED_C}" dev/mlx4/mlx4_core/mlx4_catas.c optional mlx4 pci \ compile-with "${OFED_C}" dev/mlx4/mlx4_core/mlx4_cmd.c optional mlx4 pci \ compile-with "${OFED_C}" dev/mlx4/mlx4_core/mlx4_cq.c optional mlx4 pci \ compile-with "${OFED_C}" dev/mlx4/mlx4_core/mlx4_eq.c optional mlx4 pci \ compile-with "${OFED_C}" dev/mlx4/mlx4_core/mlx4_fw.c optional mlx4 pci \ compile-with "${OFED_C}" dev/mlx4/mlx4_core/mlx4_icm.c optional mlx4 pci \ compile-with "${OFED_C}" dev/mlx4/mlx4_core/mlx4_intf.c optional mlx4 pci \ compile-with "${OFED_C}" dev/mlx4/mlx4_core/mlx4_main.c optional mlx4 pci \ compile-with "${OFED_C}" dev/mlx4/mlx4_core/mlx4_mcg.c optional mlx4 pci \ compile-with "${OFED_C}" dev/mlx4/mlx4_core/mlx4_mr.c optional mlx4 pci \ compile-with "${OFED_C}" dev/mlx4/mlx4_core/mlx4_pd.c optional mlx4 pci \ compile-with "${OFED_C}" dev/mlx4/mlx4_core/mlx4_port.c optional mlx4 pci \ compile-with "${OFED_C}" dev/mlx4/mlx4_core/mlx4_profile.c optional mlx4 pci \ compile-with "${OFED_C}" dev/mlx4/mlx4_core/mlx4_qp.c optional mlx4 pci \ compile-with "${OFED_C}" dev/mlx4/mlx4_core/mlx4_reset.c optional mlx4 pci \ compile-with "${OFED_C}" dev/mlx4/mlx4_core/mlx4_sense.c optional mlx4 pci \ compile-with "${OFED_C}" dev/mlx4/mlx4_core/mlx4_srq.c optional mlx4 pci \ compile-with "${OFED_C}" dev/mlx4/mlx4_core/mlx4_resource_tracker.c optional mlx4 pci \ compile-with "${OFED_C}" dev/mlx4/mlx4_core/mlx4_sys_tune.c optional mlx4 pci \ compile-with "${OFED_C}" dev/mlx4/mlx4_en/mlx4_en_cq.c optional mlx4en pci inet inet6 \ compile-with "${OFED_C}" dev/mlx4/mlx4_en/mlx4_en_main.c optional mlx4en pci inet inet6 \ compile-with "${OFED_C}" dev/mlx4/mlx4_en/mlx4_en_netdev.c optional mlx4en pci inet inet6 \ compile-with "${OFED_C}" dev/mlx4/mlx4_en/mlx4_en_port.c optional mlx4en pci inet inet6 \ compile-with "${OFED_C}" dev/mlx4/mlx4_en/mlx4_en_resources.c optional mlx4en pci inet inet6 \ compile-with "${OFED_C}" dev/mlx4/mlx4_en/mlx4_en_rx.c optional mlx4en pci inet inet6 \ compile-with "${OFED_C}" dev/mlx4/mlx4_en/mlx4_en_tx.c optional mlx4en pci inet inet6 \ compile-with "${OFED_C}" dev/mlx5/mlx5_core/mlx5_alloc.c optional mlx5 pci \ compile-with "${OFED_C}" dev/mlx5/mlx5_core/mlx5_cmd.c optional mlx5 pci \ compile-with "${OFED_C}" dev/mlx5/mlx5_core/mlx5_cq.c optional mlx5 pci \ compile-with "${OFED_C}" dev/mlx5/mlx5_core/mlx5_eq.c optional mlx5 pci \ compile-with "${OFED_C}" dev/mlx5/mlx5_core/mlx5_flow_table.c optional mlx5 pci \ compile-with "${OFED_C}" dev/mlx5/mlx5_core/mlx5_fw.c optional mlx5 pci \ compile-with "${OFED_C}" dev/mlx5/mlx5_core/mlx5_health.c optional mlx5 pci \ compile-with "${OFED_C}" dev/mlx5/mlx5_core/mlx5_mad.c optional mlx5 pci \ compile-with "${OFED_C}" dev/mlx5/mlx5_core/mlx5_main.c optional mlx5 pci \ compile-with "${OFED_C}" dev/mlx5/mlx5_core/mlx5_mcg.c optional mlx5 pci \ compile-with "${OFED_C}" dev/mlx5/mlx5_core/mlx5_mr.c optional mlx5 pci \ compile-with "${OFED_C}" dev/mlx5/mlx5_core/mlx5_pagealloc.c optional mlx5 pci \ compile-with "${OFED_C}" dev/mlx5/mlx5_core/mlx5_pd.c optional mlx5 pci \ compile-with "${OFED_C}" dev/mlx5/mlx5_core/mlx5_port.c optional mlx5 pci \ compile-with "${OFED_C}" dev/mlx5/mlx5_core/mlx5_qp.c optional mlx5 pci \ compile-with "${OFED_C}" dev/mlx5/mlx5_core/mlx5_srq.c optional mlx5 pci \ compile-with "${OFED_C}" dev/mlx5/mlx5_core/mlx5_transobj.c optional mlx5 pci \ compile-with "${OFED_C}" dev/mlx5/mlx5_core/mlx5_uar.c optional mlx5 pci \ compile-with "${OFED_C}" dev/mlx5/mlx5_core/mlx5_vport.c optional mlx5 pci \ compile-with "${OFED_C}" dev/mlx5/mlx5_core/mlx5_wq.c optional mlx5 pci \ compile-with "${OFED_C}" dev/mlx5/mlx5_en/mlx5_en_ethtool.c optional mlx5en pci inet inet6 \ compile-with "${OFED_C}" dev/mlx5/mlx5_en/mlx5_en_main.c optional mlx5en pci inet inet6 \ compile-with "${OFED_C}" dev/mlx5/mlx5_en/mlx5_en_tx.c optional mlx5en pci inet inet6 \ compile-with "${OFED_C}" dev/mlx5/mlx5_en/mlx5_en_flow_table.c optional mlx5en pci inet inet6 \ compile-with "${OFED_C}" dev/mlx5/mlx5_en/mlx5_en_rx.c optional mlx5en pci inet inet6 \ compile-with "${OFED_C}" dev/mlx5/mlx5_en/mlx5_en_txrx.c optional mlx5en pci inet inet6 \ compile-with "${OFED_C}" ofed/drivers/infiniband/hw/mthca/mthca_allocator.c optional mthca \ compile-with "${OFED_C}" ofed/drivers/infiniband/hw/mthca/mthca_av.c optional mthca \ compile-with "${OFED_C}" ofed/drivers/infiniband/hw/mthca/mthca_catas.c optional mthca \ compile-with "${OFED_C}" ofed/drivers/infiniband/hw/mthca/mthca_cmd.c optional mthca \ compile-with "${OFED_C}" ofed/drivers/infiniband/hw/mthca/mthca_cq.c optional mthca \ compile-with "${OFED_C}" ofed/drivers/infiniband/hw/mthca/mthca_eq.c optional mthca \ compile-with "${OFED_C}" ofed/drivers/infiniband/hw/mthca/mthca_mad.c optional mthca \ compile-with "${OFED_C}" ofed/drivers/infiniband/hw/mthca/mthca_main.c optional mthca \ compile-with "${OFED_C}" ofed/drivers/infiniband/hw/mthca/mthca_mcg.c optional mthca \ compile-with "${OFED_C}" ofed/drivers/infiniband/hw/mthca/mthca_memfree.c optional mthca \ compile-with "${OFED_C}" ofed/drivers/infiniband/hw/mthca/mthca_mr.c optional mthca \ compile-with "${OFED_C}" ofed/drivers/infiniband/hw/mthca/mthca_pd.c optional mthca \ compile-with "${OFED_C}" ofed/drivers/infiniband/hw/mthca/mthca_profile.c optional mthca \ compile-with "${OFED_C}" ofed/drivers/infiniband/hw/mthca/mthca_provider.c optional mthca \ compile-with "${OFED_C}" ofed/drivers/infiniband/hw/mthca/mthca_qp.c optional mthca \ compile-with "${OFED_C}" ofed/drivers/infiniband/hw/mthca/mthca_reset.c optional mthca \ compile-with "${OFED_C}" ofed/drivers/infiniband/hw/mthca/mthca_srq.c optional mthca \ compile-with "${OFED_C}" ofed/drivers/infiniband/hw/mthca/mthca_uar.c optional mthca \ compile-with "${OFED_C}" # crypto support opencrypto/cast.c optional crypto | ipsec opencrypto/criov.c optional crypto | ipsec opencrypto/crypto.c optional crypto | ipsec opencrypto/cryptodev.c optional cryptodev opencrypto/cryptodev_if.m optional crypto | ipsec opencrypto/cryptosoft.c optional crypto | ipsec opencrypto/cryptodeflate.c optional crypto | ipsec opencrypto/gmac.c optional crypto | ipsec opencrypto/gfmult.c optional crypto | ipsec opencrypto/rmd160.c optional crypto | ipsec opencrypto/skipjack.c optional crypto | ipsec opencrypto/xform.c optional crypto | ipsec rpc/auth_none.c optional krpc | nfslockd | nfscl | nfsd rpc/auth_unix.c optional krpc | nfslockd | nfscl | nfsd rpc/authunix_prot.c optional krpc | nfslockd | nfscl | nfsd rpc/clnt_bck.c optional krpc | nfslockd | nfscl | nfsd rpc/clnt_dg.c optional krpc | nfslockd | nfscl | nfsd rpc/clnt_rc.c optional krpc | nfslockd | nfscl | nfsd rpc/clnt_vc.c optional krpc | nfslockd | nfscl | nfsd rpc/getnetconfig.c optional krpc | nfslockd | nfscl | nfsd rpc/replay.c optional krpc | nfslockd | nfscl | nfsd rpc/rpc_callmsg.c optional krpc | nfslockd | nfscl | nfsd rpc/rpc_generic.c optional krpc | nfslockd | nfscl | nfsd rpc/rpc_prot.c optional krpc | nfslockd | nfscl | nfsd rpc/rpcb_clnt.c optional krpc | nfslockd | nfscl | nfsd rpc/rpcb_prot.c optional krpc | nfslockd | nfscl | nfsd rpc/svc.c optional krpc | nfslockd | nfscl | nfsd rpc/svc_auth.c optional krpc | nfslockd | nfscl | nfsd rpc/svc_auth_unix.c optional krpc | nfslockd | nfscl | nfsd rpc/svc_dg.c optional krpc | nfslockd | nfscl | nfsd rpc/svc_generic.c optional krpc | nfslockd | nfscl | nfsd rpc/svc_vc.c optional krpc | nfslockd | 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_klib.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/audit/bsm_domain.c optional audit security/audit/bsm_errno.c optional audit security/audit/bsm_fcntl.c optional audit security/audit/bsm_socket_type.c optional audit security/audit/bsm_token.c optional audit 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 | vt 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 ffs directio ufs/ffs/ffs_suspend.c optional ffs 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/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_radix.c standard vm/vm_reserv.c standard vm/vm_domain.c standard vm/vm_unix.c standard vm/vnode_pager.c standard xen/features.c optional xenhvm xen/xenbus/xenbus_if.m optional xenhvm xen/xenbus/xenbus.c optional xenhvm xen/xenbus/xenbusb_if.m optional xenhvm xen/xenbus/xenbusb.c optional xenhvm xen/xenbus/xenbusb_front.c optional xenhvm xen/xenbus/xenbusb_back.c optional xenhvm xen/xenmem/xenmem_if.m optional xenhvm xdr/xdr.c optional krpc | nfslockd | nfscl | nfsd xdr/xdr_array.c optional krpc | nfslockd | nfscl | nfsd xdr/xdr_mbuf.c optional krpc | nfslockd | nfscl | nfsd xdr/xdr_mem.c optional krpc | nfslockd | nfscl | nfsd xdr/xdr_reference.c optional krpc | nfslockd | nfscl | nfsd xdr/xdr_sizeof.c optional krpc | nfslockd | nfscl | nfsd Index: projects/clang400-import/sys/dev/advansys/adwcam.c =================================================================== --- projects/clang400-import/sys/dev/advansys/adwcam.c (revision 311459) +++ projects/clang400-import/sys/dev/advansys/adwcam.c (revision 311460) @@ -1,1504 +1,1504 @@ /*- * CAM SCSI interface for the Advanced Systems Inc. * Second Generation SCSI controllers. * * Product specific probe and attach routines can be found in: * * adw_pci.c ABP[3]940UW, ABP950UW, ABP3940U2W * * Copyright (c) 1998, 1999, 2000 Justin Gibbs. * 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. 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 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. */ /* * Ported from: * advansys.c - Linux Host Driver for AdvanSys SCSI Adapters * * Copyright (c) 1995-1998 Advanced System Products, Inc. * All Rights Reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that redistributions of source * code retain the above copyright notice and this comment without * modification. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include /* Definitions for our use of the SIM private CCB area */ #define ccb_acb_ptr spriv_ptr0 #define ccb_adw_ptr spriv_ptr1 static __inline struct acb* adwgetacb(struct adw_softc *adw); static __inline void adwfreeacb(struct adw_softc *adw, struct acb *acb); static void adwmapmem(void *arg, bus_dma_segment_t *segs, int nseg, int error); static struct sg_map_node* adwallocsgmap(struct adw_softc *adw); static int adwallocacbs(struct adw_softc *adw); static void adwexecuteacb(void *arg, bus_dma_segment_t *dm_segs, int nseg, int error); static void adw_action(struct cam_sim *sim, union ccb *ccb); static void adw_intr_locked(struct adw_softc *adw); static void adw_poll(struct cam_sim *sim); static void adw_async(void *callback_arg, u_int32_t code, struct cam_path *path, void *arg); static void adwprocesserror(struct adw_softc *adw, struct acb *acb); static void adwtimeout(void *arg); static void adw_handle_device_reset(struct adw_softc *adw, u_int target); static void adw_handle_bus_reset(struct adw_softc *adw, int initiated); static __inline struct acb* adwgetacb(struct adw_softc *adw) { struct acb* acb; if (!dumping) mtx_assert(&adw->lock, MA_OWNED); if ((acb = SLIST_FIRST(&adw->free_acb_list)) != NULL) { SLIST_REMOVE_HEAD(&adw->free_acb_list, links); } else if (adw->num_acbs < adw->max_acbs) { adwallocacbs(adw); acb = SLIST_FIRST(&adw->free_acb_list); if (acb == NULL) device_printf(adw->device, "Can't malloc ACB\n"); else { SLIST_REMOVE_HEAD(&adw->free_acb_list, links); } } return (acb); } static __inline void adwfreeacb(struct adw_softc *adw, struct acb *acb) { if (!dumping) mtx_assert(&adw->lock, MA_OWNED); if ((acb->state & ACB_ACTIVE) != 0) LIST_REMOVE(&acb->ccb->ccb_h, sim_links.le); if ((acb->state & ACB_RELEASE_SIMQ) != 0) acb->ccb->ccb_h.status |= CAM_RELEASE_SIMQ; else if ((adw->state & ADW_RESOURCE_SHORTAGE) != 0 && (acb->ccb->ccb_h.status & CAM_RELEASE_SIMQ) == 0) { acb->ccb->ccb_h.status |= CAM_RELEASE_SIMQ; adw->state &= ~ADW_RESOURCE_SHORTAGE; } acb->state = ACB_FREE; SLIST_INSERT_HEAD(&adw->free_acb_list, acb, links); } static void adwmapmem(void *arg, bus_dma_segment_t *segs, int nseg, int error) { bus_addr_t *busaddrp; busaddrp = (bus_addr_t *)arg; *busaddrp = segs->ds_addr; } static struct sg_map_node * adwallocsgmap(struct adw_softc *adw) { struct sg_map_node *sg_map; sg_map = malloc(sizeof(*sg_map), M_DEVBUF, M_NOWAIT); if (sg_map == NULL) return (NULL); /* Allocate S/G space for the next batch of ACBS */ if (bus_dmamem_alloc(adw->sg_dmat, (void **)&sg_map->sg_vaddr, BUS_DMA_NOWAIT, &sg_map->sg_dmamap) != 0) { free(sg_map, M_DEVBUF); return (NULL); } SLIST_INSERT_HEAD(&adw->sg_maps, sg_map, links); bus_dmamap_load(adw->sg_dmat, sg_map->sg_dmamap, sg_map->sg_vaddr, PAGE_SIZE, adwmapmem, &sg_map->sg_physaddr, /*flags*/0); bzero(sg_map->sg_vaddr, PAGE_SIZE); return (sg_map); } /* * Allocate another chunk of CCB's. Return count of entries added. */ static int adwallocacbs(struct adw_softc *adw) { struct acb *next_acb; struct sg_map_node *sg_map; bus_addr_t busaddr; struct adw_sg_block *blocks; int newcount; int i; next_acb = &adw->acbs[adw->num_acbs]; sg_map = adwallocsgmap(adw); if (sg_map == NULL) return (0); blocks = sg_map->sg_vaddr; busaddr = sg_map->sg_physaddr; newcount = (PAGE_SIZE / (ADW_SG_BLOCKCNT * sizeof(*blocks))); for (i = 0; adw->num_acbs < adw->max_acbs && i < newcount; i++) { int error; error = bus_dmamap_create(adw->buffer_dmat, /*flags*/0, &next_acb->dmamap); if (error != 0) break; next_acb->queue.scsi_req_baddr = acbvtob(adw, next_acb); next_acb->queue.scsi_req_bo = acbvtobo(adw, next_acb); next_acb->queue.sense_baddr = acbvtob(adw, next_acb) + offsetof(struct acb, sense_data); next_acb->sg_blocks = blocks; next_acb->sg_busaddr = busaddr; next_acb->state = ACB_FREE; callout_init_mtx(&next_acb->timer, &adw->lock, 0); SLIST_INSERT_HEAD(&adw->free_acb_list, next_acb, links); blocks += ADW_SG_BLOCKCNT; busaddr += ADW_SG_BLOCKCNT * sizeof(*blocks); next_acb++; adw->num_acbs++; } return (i); } static void adwexecuteacb(void *arg, bus_dma_segment_t *dm_segs, int nseg, int error) { struct acb *acb; union ccb *ccb; struct adw_softc *adw; acb = (struct acb *)arg; ccb = acb->ccb; adw = (struct adw_softc *)ccb->ccb_h.ccb_adw_ptr; if (!dumping) mtx_assert(&adw->lock, MA_OWNED); if (error != 0) { if (error != EFBIG) device_printf(adw->device, "Unexepected error 0x%x " "returned from bus_dmamap_load\n", error); if (ccb->ccb_h.status == CAM_REQ_INPROG) { xpt_freeze_devq(ccb->ccb_h.path, /*count*/1); ccb->ccb_h.status = CAM_REQ_TOO_BIG|CAM_DEV_QFRZN; } adwfreeacb(adw, acb); xpt_done(ccb); return; } if (nseg != 0) { bus_dmasync_op_t op; acb->queue.data_addr = dm_segs[0].ds_addr; acb->queue.data_cnt = ccb->csio.dxfer_len; if (nseg > 1) { struct adw_sg_block *sg_block; struct adw_sg_elm *sg; bus_addr_t sg_busaddr; u_int sg_index; bus_dma_segment_t *end_seg; end_seg = dm_segs + nseg; sg_busaddr = acb->sg_busaddr; sg_index = 0; /* Copy the segments into our SG list */ for (sg_block = acb->sg_blocks;; sg_block++) { u_int i; sg = sg_block->sg_list; for (i = 0; i < ADW_NO_OF_SG_PER_BLOCK; i++) { if (dm_segs >= end_seg) break; sg->sg_addr = dm_segs->ds_addr; sg->sg_count = dm_segs->ds_len; sg++; dm_segs++; } sg_block->sg_cnt = i; sg_index += i; if (dm_segs == end_seg) { sg_block->sg_busaddr_next = 0; break; } else { sg_busaddr += sizeof(struct adw_sg_block); sg_block->sg_busaddr_next = sg_busaddr; } } acb->queue.sg_real_addr = acb->sg_busaddr; } else { acb->queue.sg_real_addr = 0; } if ((ccb->ccb_h.flags & CAM_DIR_MASK) == CAM_DIR_IN) op = BUS_DMASYNC_PREREAD; else op = BUS_DMASYNC_PREWRITE; bus_dmamap_sync(adw->buffer_dmat, acb->dmamap, op); } else { acb->queue.data_addr = 0; acb->queue.data_cnt = 0; acb->queue.sg_real_addr = 0; } /* * Last time we need to check if this CCB needs to * be aborted. */ if (ccb->ccb_h.status != CAM_REQ_INPROG) { if (nseg != 0) bus_dmamap_unload(adw->buffer_dmat, acb->dmamap); adwfreeacb(adw, acb); xpt_done(ccb); return; } acb->state |= ACB_ACTIVE; ccb->ccb_h.status |= CAM_SIM_QUEUED; LIST_INSERT_HEAD(&adw->pending_ccbs, &ccb->ccb_h, sim_links.le); callout_reset_sbt(&acb->timer, SBT_1MS * ccb->ccb_h.timeout, 0, adwtimeout, acb, 0); adw_send_acb(adw, acb, acbvtob(adw, acb)); } static void adw_action(struct cam_sim *sim, union ccb *ccb) { struct adw_softc *adw; CAM_DEBUG(ccb->ccb_h.path, CAM_DEBUG_TRACE, ("adw_action\n")); adw = (struct adw_softc *)cam_sim_softc(sim); if (!dumping) mtx_assert(&adw->lock, MA_OWNED); switch (ccb->ccb_h.func_code) { /* Common cases first */ case XPT_SCSI_IO: /* Execute the requested I/O operation */ { struct ccb_scsiio *csio; struct acb *acb; int error; csio = &ccb->csio; /* Max supported CDB length is 12 bytes */ if (csio->cdb_len > 12) { ccb->ccb_h.status = CAM_REQ_INVALID; xpt_done(ccb); return; } if ((acb = adwgetacb(adw)) == NULL) { adw->state |= ADW_RESOURCE_SHORTAGE; xpt_freeze_simq(sim, /*count*/1); ccb->ccb_h.status = CAM_REQUEUE_REQ; xpt_done(ccb); return; } /* Link acb and ccb so we can find one from the other */ acb->ccb = ccb; ccb->ccb_h.ccb_acb_ptr = acb; ccb->ccb_h.ccb_adw_ptr = adw; acb->queue.cntl = 0; acb->queue.target_cmd = 0; acb->queue.target_id = ccb->ccb_h.target_id; acb->queue.target_lun = ccb->ccb_h.target_lun; acb->queue.mflag = 0; acb->queue.sense_len = MIN(csio->sense_len, sizeof(acb->sense_data)); acb->queue.cdb_len = csio->cdb_len; if ((ccb->ccb_h.flags & CAM_TAG_ACTION_VALID) != 0) { switch (csio->tag_action) { case MSG_SIMPLE_Q_TAG: acb->queue.scsi_cntl = ADW_QSC_SIMPLE_Q_TAG; break; case MSG_HEAD_OF_Q_TAG: acb->queue.scsi_cntl = ADW_QSC_HEAD_OF_Q_TAG; break; case MSG_ORDERED_Q_TAG: acb->queue.scsi_cntl = ADW_QSC_ORDERED_Q_TAG; break; default: acb->queue.scsi_cntl = ADW_QSC_NO_TAGMSG; break; } } else acb->queue.scsi_cntl = ADW_QSC_NO_TAGMSG; if ((ccb->ccb_h.flags & CAM_DIS_DISCONNECT) != 0) acb->queue.scsi_cntl |= ADW_QSC_NO_DISC; acb->queue.done_status = 0; acb->queue.scsi_status = 0; acb->queue.host_status = 0; acb->queue.sg_wk_ix = 0; if ((ccb->ccb_h.flags & CAM_CDB_POINTER) != 0) { if ((ccb->ccb_h.flags & CAM_CDB_PHYS) == 0) { bcopy(csio->cdb_io.cdb_ptr, acb->queue.cdb, csio->cdb_len); } else { /* I guess I could map it in... */ ccb->ccb_h.status = CAM_REQ_INVALID; adwfreeacb(adw, acb); xpt_done(ccb); return; } } else { bcopy(csio->cdb_io.cdb_bytes, acb->queue.cdb, csio->cdb_len); } error = bus_dmamap_load_ccb(adw->buffer_dmat, acb->dmamap, ccb, adwexecuteacb, acb, /*flags*/0); if (error == EINPROGRESS) { /* * So as to maintain ordering, freeze the controller * queue until our mapping is returned. */ xpt_freeze_simq(sim, 1); acb->state |= CAM_RELEASE_SIMQ; } break; } case XPT_RESET_DEV: /* Bus Device Reset the specified SCSI device */ { adw_idle_cmd_status_t status; status = adw_idle_cmd_send(adw, ADW_IDLE_CMD_DEVICE_RESET, ccb->ccb_h.target_id); if (status == ADW_IDLE_CMD_SUCCESS) { ccb->ccb_h.status = CAM_REQ_CMP; if (bootverbose) { xpt_print_path(ccb->ccb_h.path); printf("BDR Delivered\n"); } } else ccb->ccb_h.status = CAM_REQ_CMP_ERR; xpt_done(ccb); break; } case XPT_ABORT: /* Abort the specified CCB */ /* XXX Implement */ ccb->ccb_h.status = CAM_REQ_INVALID; xpt_done(ccb); break; case XPT_SET_TRAN_SETTINGS: { struct ccb_trans_settings_scsi *scsi; struct ccb_trans_settings_spi *spi; struct ccb_trans_settings *cts; u_int target_mask; cts = &ccb->cts; target_mask = 0x01 << ccb->ccb_h.target_id; scsi = &cts->proto_specific.scsi; spi = &cts->xport_specific.spi; if (cts->type == CTS_TYPE_CURRENT_SETTINGS) { u_int sdtrdone; sdtrdone = adw_lram_read_16(adw, ADW_MC_SDTR_DONE); if ((spi->valid & CTS_SPI_VALID_DISC) != 0) { u_int discenb; discenb = adw_lram_read_16(adw, ADW_MC_DISC_ENABLE); if ((spi->flags & CTS_SPI_FLAGS_DISC_ENB) != 0) discenb |= target_mask; else discenb &= ~target_mask; adw_lram_write_16(adw, ADW_MC_DISC_ENABLE, discenb); } if ((scsi->valid & CTS_SCSI_VALID_TQ) != 0) { if ((scsi->flags & CTS_SCSI_FLAGS_TAG_ENB) != 0) adw->tagenb |= target_mask; else adw->tagenb &= ~target_mask; } if ((spi->valid & CTS_SPI_VALID_BUS_WIDTH) != 0) { u_int wdtrenb_orig; u_int wdtrenb; u_int wdtrdone; wdtrenb_orig = adw_lram_read_16(adw, ADW_MC_WDTR_ABLE); wdtrenb = wdtrenb_orig; wdtrdone = adw_lram_read_16(adw, ADW_MC_WDTR_DONE); switch (spi->bus_width) { case MSG_EXT_WDTR_BUS_32_BIT: case MSG_EXT_WDTR_BUS_16_BIT: wdtrenb |= target_mask; break; case MSG_EXT_WDTR_BUS_8_BIT: default: wdtrenb &= ~target_mask; break; } if (wdtrenb != wdtrenb_orig) { adw_lram_write_16(adw, ADW_MC_WDTR_ABLE, wdtrenb); wdtrdone &= ~target_mask; adw_lram_write_16(adw, ADW_MC_WDTR_DONE, wdtrdone); /* Wide negotiation forces async */ sdtrdone &= ~target_mask; adw_lram_write_16(adw, ADW_MC_SDTR_DONE, sdtrdone); } } if (((spi->valid & CTS_SPI_VALID_SYNC_RATE) != 0) || ((spi->valid & CTS_SPI_VALID_SYNC_OFFSET) != 0)) { u_int sdtr_orig; u_int sdtr; u_int sdtrable_orig; u_int sdtrable; sdtr = adw_get_chip_sdtr(adw, ccb->ccb_h.target_id); sdtr_orig = sdtr; sdtrable = adw_lram_read_16(adw, ADW_MC_SDTR_ABLE); sdtrable_orig = sdtrable; if ((spi->valid & CTS_SPI_VALID_SYNC_RATE) != 0) { sdtr = adw_find_sdtr(adw, spi->sync_period); } if ((spi->valid & CTS_SPI_VALID_SYNC_OFFSET) != 0) { if (spi->sync_offset == 0) sdtr = ADW_MC_SDTR_ASYNC; } if (sdtr == ADW_MC_SDTR_ASYNC) sdtrable &= ~target_mask; else sdtrable |= target_mask; if (sdtr != sdtr_orig || sdtrable != sdtrable_orig) { adw_set_chip_sdtr(adw, ccb->ccb_h.target_id, sdtr); sdtrdone &= ~target_mask; adw_lram_write_16(adw, ADW_MC_SDTR_ABLE, sdtrable); adw_lram_write_16(adw, ADW_MC_SDTR_DONE, sdtrdone); } } } ccb->ccb_h.status = CAM_REQ_CMP; xpt_done(ccb); break; } case XPT_GET_TRAN_SETTINGS: /* Get default/user set transfer settings for the target */ { struct ccb_trans_settings_scsi *scsi; struct ccb_trans_settings_spi *spi; struct ccb_trans_settings *cts; u_int target_mask; cts = &ccb->cts; target_mask = 0x01 << ccb->ccb_h.target_id; cts->protocol = PROTO_SCSI; cts->protocol_version = SCSI_REV_2; cts->transport = XPORT_SPI; cts->transport_version = 2; scsi = &cts->proto_specific.scsi; spi = &cts->xport_specific.spi; if (cts->type == CTS_TYPE_CURRENT_SETTINGS) { u_int mc_sdtr; spi->flags = 0; if ((adw->user_discenb & target_mask) != 0) spi->flags |= CTS_SPI_FLAGS_DISC_ENB; if ((adw->user_tagenb & target_mask) != 0) scsi->flags |= CTS_SCSI_FLAGS_TAG_ENB; if ((adw->user_wdtr & target_mask) != 0) spi->bus_width = MSG_EXT_WDTR_BUS_16_BIT; else spi->bus_width = MSG_EXT_WDTR_BUS_8_BIT; mc_sdtr = adw_get_user_sdtr(adw, ccb->ccb_h.target_id); spi->sync_period = adw_find_period(adw, mc_sdtr); if (spi->sync_period != 0) spi->sync_offset = 15; /* XXX ??? */ else spi->sync_offset = 0; } else { u_int targ_tinfo; spi->flags = 0; if ((adw_lram_read_16(adw, ADW_MC_DISC_ENABLE) & target_mask) != 0) spi->flags |= CTS_SPI_FLAGS_DISC_ENB; if ((adw->tagenb & target_mask) != 0) scsi->flags |= CTS_SCSI_FLAGS_TAG_ENB; targ_tinfo = adw_lram_read_16(adw, ADW_MC_DEVICE_HSHK_CFG_TABLE + (2 * ccb->ccb_h.target_id)); if ((targ_tinfo & ADW_HSHK_CFG_WIDE_XFR) != 0) spi->bus_width = MSG_EXT_WDTR_BUS_16_BIT; else spi->bus_width = MSG_EXT_WDTR_BUS_8_BIT; spi->sync_period = adw_hshk_cfg_period_factor(targ_tinfo); spi->sync_offset = targ_tinfo & ADW_HSHK_CFG_OFFSET; if (spi->sync_period == 0) spi->sync_offset = 0; if (spi->sync_offset == 0) spi->sync_period = 0; } spi->valid = CTS_SPI_VALID_SYNC_RATE | CTS_SPI_VALID_SYNC_OFFSET | CTS_SPI_VALID_BUS_WIDTH | CTS_SPI_VALID_DISC; scsi->valid = CTS_SCSI_VALID_TQ; ccb->ccb_h.status = CAM_REQ_CMP; xpt_done(ccb); break; } case XPT_CALC_GEOMETRY: { /* * XXX Use Adaptec translation until I find out how to * get this information from the card. */ cam_calc_geometry(&ccb->ccg, /*extended*/1); xpt_done(ccb); break; } case XPT_RESET_BUS: /* Reset the specified SCSI bus */ { int failure; failure = adw_reset_bus(adw); if (failure != 0) { ccb->ccb_h.status = CAM_REQ_CMP_ERR; } else { if (bootverbose) { xpt_print_path(adw->path); printf("Bus Reset Delivered\n"); } ccb->ccb_h.status = CAM_REQ_CMP; } xpt_done(ccb); break; } case XPT_TERM_IO: /* Terminate the I/O process */ /* XXX Implement */ ccb->ccb_h.status = CAM_REQ_INVALID; xpt_done(ccb); break; case XPT_PATH_INQ: /* Path routing inquiry */ { struct ccb_pathinq *cpi = &ccb->cpi; cpi->version_num = 1; cpi->hba_inquiry = PI_WIDE_16|PI_SDTR_ABLE|PI_TAG_ABLE; cpi->target_sprt = 0; cpi->hba_misc = 0; cpi->hba_eng_cnt = 0; cpi->max_target = ADW_MAX_TID; cpi->max_lun = ADW_MAX_LUN; cpi->initiator_id = adw->initiator_id; cpi->bus_id = cam_sim_bus(sim); cpi->base_transfer_speed = 3300; strlcpy(cpi->sim_vid, "FreeBSD", SIM_IDLEN); strlcpy(cpi->hba_vid, "AdvanSys", HBA_IDLEN); strlcpy(cpi->dev_name, cam_sim_name(sim), DEV_IDLEN); cpi->unit_number = cam_sim_unit(sim); - cpi->transport = XPORT_SPI; - cpi->transport_version = 2; - cpi->protocol = PROTO_SCSI; - cpi->protocol_version = SCSI_REV_2; + cpi->transport = XPORT_SPI; + cpi->transport_version = 2; + cpi->protocol = PROTO_SCSI; + cpi->protocol_version = SCSI_REV_2; cpi->ccb_h.status = CAM_REQ_CMP; xpt_done(ccb); break; } default: ccb->ccb_h.status = CAM_REQ_INVALID; xpt_done(ccb); break; } } static void adw_poll(struct cam_sim *sim) { adw_intr_locked(cam_sim_softc(sim)); } static void adw_async(void *callback_arg, u_int32_t code, struct cam_path *path, void *arg) { } struct adw_softc * adw_alloc(device_t dev, struct resource *regs, int regs_type, int regs_id) { struct adw_softc *adw; adw = device_get_softc(dev); LIST_INIT(&adw->pending_ccbs); SLIST_INIT(&adw->sg_maps); mtx_init(&adw->lock, "adw", NULL, MTX_DEF); adw->device = dev; adw->regs_res_type = regs_type; adw->regs_res_id = regs_id; adw->regs = regs; return(adw); } void adw_free(struct adw_softc *adw) { switch (adw->init_level) { case 9: { struct sg_map_node *sg_map; while ((sg_map = SLIST_FIRST(&adw->sg_maps)) != NULL) { SLIST_REMOVE_HEAD(&adw->sg_maps, links); bus_dmamap_unload(adw->sg_dmat, sg_map->sg_dmamap); bus_dmamem_free(adw->sg_dmat, sg_map->sg_vaddr, sg_map->sg_dmamap); free(sg_map, M_DEVBUF); } bus_dma_tag_destroy(adw->sg_dmat); } case 8: bus_dmamap_unload(adw->acb_dmat, adw->acb_dmamap); case 7: bus_dmamem_free(adw->acb_dmat, adw->acbs, adw->acb_dmamap); case 6: bus_dma_tag_destroy(adw->acb_dmat); case 5: bus_dmamap_unload(adw->carrier_dmat, adw->carrier_dmamap); case 4: bus_dmamem_free(adw->carrier_dmat, adw->carriers, adw->carrier_dmamap); case 3: bus_dma_tag_destroy(adw->carrier_dmat); case 2: bus_dma_tag_destroy(adw->buffer_dmat); case 1: bus_dma_tag_destroy(adw->parent_dmat); case 0: break; } if (adw->regs != NULL) bus_release_resource(adw->device, adw->regs_res_type, adw->regs_res_id, adw->regs); if (adw->irq != NULL) bus_release_resource(adw->device, adw->irq_res_type, 0, adw->irq); if (adw->sim != NULL) { if (adw->path != NULL) { xpt_async(AC_LOST_DEVICE, adw->path, NULL); xpt_free_path(adw->path); } xpt_bus_deregister(cam_sim_path(adw->sim)); cam_sim_free(adw->sim, /*free_devq*/TRUE); } mtx_destroy(&adw->lock); } int adw_init(struct adw_softc *adw) { struct adw_eeprom eep_config; u_int tid; u_int i; u_int16_t checksum; u_int16_t scsicfg1; checksum = adw_eeprom_read(adw, &eep_config); bcopy(eep_config.serial_number, adw->serial_number, sizeof(adw->serial_number)); if (checksum != eep_config.checksum) { u_int16_t serial_number[3]; adw->flags |= ADW_EEPROM_FAILED; device_printf(adw->device, "EEPROM checksum failed. Restoring Defaults\n"); /* * Restore the default EEPROM settings. * Assume the 6 byte board serial number that was read * from EEPROM is correct even if the EEPROM checksum * failed. */ bcopy(adw->default_eeprom, &eep_config, sizeof(eep_config)); bcopy(adw->serial_number, eep_config.serial_number, sizeof(serial_number)); adw_eeprom_write(adw, &eep_config); } /* Pull eeprom information into our softc. */ adw->bios_ctrl = eep_config.bios_ctrl; adw->user_wdtr = eep_config.wdtr_able; for (tid = 0; tid < ADW_MAX_TID; tid++) { u_int mc_sdtr; u_int16_t tid_mask; tid_mask = 0x1 << tid; if ((adw->features & ADW_ULTRA) != 0) { /* * Ultra chips store sdtr and ultraenb * bits in their seeprom, so we must * construct valid mc_sdtr entries for * indirectly. */ if (eep_config.sync1.sync_enable & tid_mask) { if (eep_config.sync2.ultra_enable & tid_mask) mc_sdtr = ADW_MC_SDTR_20; else mc_sdtr = ADW_MC_SDTR_10; } else mc_sdtr = ADW_MC_SDTR_ASYNC; } else { switch (ADW_TARGET_GROUP(tid)) { case 3: mc_sdtr = eep_config.sync4.sdtr4; break; case 2: mc_sdtr = eep_config.sync3.sdtr3; break; case 1: mc_sdtr = eep_config.sync2.sdtr2; break; default: /* Shut up compiler */ case 0: mc_sdtr = eep_config.sync1.sdtr1; break; } mc_sdtr >>= ADW_TARGET_GROUP_SHIFT(tid); mc_sdtr &= 0xFF; } adw_set_user_sdtr(adw, tid, mc_sdtr); } adw->user_tagenb = eep_config.tagqng_able; adw->user_discenb = eep_config.disc_enable; adw->max_acbs = eep_config.max_host_qng; adw->initiator_id = (eep_config.adapter_scsi_id & ADW_MAX_TID); /* * Sanity check the number of host openings. */ if (adw->max_acbs > ADW_DEF_MAX_HOST_QNG) adw->max_acbs = ADW_DEF_MAX_HOST_QNG; else if (adw->max_acbs < ADW_DEF_MIN_HOST_QNG) { /* If the value is zero, assume it is uninitialized. */ if (adw->max_acbs == 0) adw->max_acbs = ADW_DEF_MAX_HOST_QNG; else adw->max_acbs = ADW_DEF_MIN_HOST_QNG; } scsicfg1 = 0; if ((adw->features & ADW_ULTRA2) != 0) { switch (eep_config.termination_lvd) { default: device_printf(adw->device, "Invalid EEPROM LVD Termination Settings.\n"); device_printf(adw->device, "Reverting to Automatic LVD Termination\n"); /* FALLTHROUGH */ case ADW_EEPROM_TERM_AUTO: break; case ADW_EEPROM_TERM_BOTH_ON: scsicfg1 |= ADW2_SCSI_CFG1_TERM_LVD_LO; /* FALLTHROUGH */ case ADW_EEPROM_TERM_HIGH_ON: scsicfg1 |= ADW2_SCSI_CFG1_TERM_LVD_HI; /* FALLTHROUGH */ case ADW_EEPROM_TERM_OFF: scsicfg1 |= ADW2_SCSI_CFG1_DIS_TERM_DRV; break; } } switch (eep_config.termination_se) { default: device_printf(adw->device, "Invalid SE EEPROM Termination Settings.\n"); device_printf(adw->device, "Reverting to Automatic SE Termination\n"); /* FALLTHROUGH */ case ADW_EEPROM_TERM_AUTO: break; case ADW_EEPROM_TERM_BOTH_ON: scsicfg1 |= ADW_SCSI_CFG1_TERM_CTL_L; /* FALLTHROUGH */ case ADW_EEPROM_TERM_HIGH_ON: scsicfg1 |= ADW_SCSI_CFG1_TERM_CTL_H; /* FALLTHROUGH */ case ADW_EEPROM_TERM_OFF: scsicfg1 |= ADW_SCSI_CFG1_TERM_CTL_MANUAL; break; } device_printf(adw->device, "SCSI ID %d, ", adw->initiator_id); /* DMA tag for mapping buffers into device visible space. */ if (bus_dma_tag_create( /* parent */ adw->parent_dmat, /* alignment */ 1, /* boundary */ 0, /* lowaddr */ BUS_SPACE_MAXADDR_32BIT, /* highaddr */ BUS_SPACE_MAXADDR, /* filter */ NULL, /* filterarg */ NULL, /* maxsize */ DFLTPHYS, /* nsegments */ ADW_SGSIZE, /* maxsegsz */ BUS_SPACE_MAXSIZE_32BIT, /* flags */ BUS_DMA_ALLOCNOW, /* lockfunc */ busdma_lock_mutex, /* lockarg */ &adw->lock, &adw->buffer_dmat) != 0) { return (ENOMEM); } adw->init_level++; /* DMA tag for our ccb carrier structures */ if (bus_dma_tag_create( /* parent */ adw->parent_dmat, /* alignment */ 0x10, /* boundary */ 0, /* lowaddr */ BUS_SPACE_MAXADDR_32BIT, /* highaddr */ BUS_SPACE_MAXADDR, /* filter */ NULL, /* filterarg */ NULL, /* maxsize */ (adw->max_acbs + ADW_NUM_CARRIER_QUEUES + 1) * sizeof(struct adw_carrier), /* nsegments */ 1, /* maxsegsz */ BUS_SPACE_MAXSIZE_32BIT, /* flags */ 0, /* lockfunc */ NULL, /* lockarg */ NULL, &adw->carrier_dmat) != 0) { return (ENOMEM); } adw->init_level++; /* Allocation for our ccb carrier structures */ if (bus_dmamem_alloc(adw->carrier_dmat, (void **)&adw->carriers, BUS_DMA_NOWAIT, &adw->carrier_dmamap) != 0) { return (ENOMEM); } adw->init_level++; /* And permanently map them */ bus_dmamap_load(adw->carrier_dmat, adw->carrier_dmamap, adw->carriers, (adw->max_acbs + ADW_NUM_CARRIER_QUEUES + 1) * sizeof(struct adw_carrier), adwmapmem, &adw->carrier_busbase, /*flags*/0); /* Clear them out. */ bzero(adw->carriers, (adw->max_acbs + ADW_NUM_CARRIER_QUEUES + 1) * sizeof(struct adw_carrier)); /* Setup our free carrier list */ adw->free_carriers = adw->carriers; for (i = 0; i < adw->max_acbs + ADW_NUM_CARRIER_QUEUES; i++) { adw->carriers[i].carr_offset = carriervtobo(adw, &adw->carriers[i]); adw->carriers[i].carr_ba = carriervtob(adw, &adw->carriers[i]); adw->carriers[i].areq_ba = 0; adw->carriers[i].next_ba = carriervtobo(adw, &adw->carriers[i+1]); } /* Terminal carrier. Never leaves the freelist */ adw->carriers[i].carr_offset = carriervtobo(adw, &adw->carriers[i]); adw->carriers[i].carr_ba = carriervtob(adw, &adw->carriers[i]); adw->carriers[i].areq_ba = 0; adw->carriers[i].next_ba = ~0; adw->init_level++; /* DMA tag for our acb structures */ if (bus_dma_tag_create( /* parent */ adw->parent_dmat, /* alignment */ 1, /* boundary */ 0, /* lowaddr */ BUS_SPACE_MAXADDR, /* highaddr */ BUS_SPACE_MAXADDR, /* filter */ NULL, /* filterarg */ NULL, /* maxsize */ adw->max_acbs * sizeof(struct acb), /* nsegments */ 1, /* maxsegsz */ BUS_SPACE_MAXSIZE_32BIT, /* flags */ 0, /* lockfunc */ NULL, /* lockarg */ NULL, &adw->acb_dmat) != 0) { return (ENOMEM); } adw->init_level++; /* Allocation for our ccbs */ if (bus_dmamem_alloc(adw->acb_dmat, (void **)&adw->acbs, BUS_DMA_NOWAIT, &adw->acb_dmamap) != 0) return (ENOMEM); adw->init_level++; /* And permanently map them */ bus_dmamap_load(adw->acb_dmat, adw->acb_dmamap, adw->acbs, adw->max_acbs * sizeof(struct acb), adwmapmem, &adw->acb_busbase, /*flags*/0); /* Clear them out. */ bzero(adw->acbs, adw->max_acbs * sizeof(struct acb)); /* DMA tag for our S/G structures. We allocate in page sized chunks */ if (bus_dma_tag_create( /* parent */ adw->parent_dmat, /* alignment */ 1, /* boundary */ 0, /* lowaddr */ BUS_SPACE_MAXADDR, /* highaddr */ BUS_SPACE_MAXADDR, /* filter */ NULL, /* filterarg */ NULL, /* maxsize */ PAGE_SIZE, /* nsegments */ 1, /* maxsegsz */ BUS_SPACE_MAXSIZE_32BIT, /* flags */ 0, /* lockfunc */ NULL, /* lockarg */ NULL, &adw->sg_dmat) != 0) { return (ENOMEM); } adw->init_level++; /* Allocate our first batch of ccbs */ mtx_lock(&adw->lock); if (adwallocacbs(adw) == 0) { mtx_unlock(&adw->lock); return (ENOMEM); } if (adw_init_chip(adw, scsicfg1) != 0) { mtx_unlock(&adw->lock); return (ENXIO); } printf("Queue Depth %d\n", adw->max_acbs); mtx_unlock(&adw->lock); return (0); } /* * Attach all the sub-devices we can find */ int adw_attach(struct adw_softc *adw) { struct ccb_setasync csa; struct cam_devq *devq; int error; /* Hook up our interrupt handler */ error = bus_setup_intr(adw->device, adw->irq, INTR_TYPE_CAM | INTR_ENTROPY | INTR_MPSAFE, NULL, adw_intr, adw, &adw->ih); if (error != 0) { device_printf(adw->device, "bus_setup_intr() failed: %d\n", error); return (error); } /* Start the Risc processor now that we are fully configured. */ adw_outw(adw, ADW_RISC_CSR, ADW_RISC_CSR_RUN); /* * Create the device queue for our SIM. */ devq = cam_simq_alloc(adw->max_acbs); if (devq == NULL) return (ENOMEM); /* * Construct our SIM entry. */ adw->sim = cam_sim_alloc(adw_action, adw_poll, "adw", adw, device_get_unit(adw->device), &adw->lock, 1, adw->max_acbs, devq); if (adw->sim == NULL) return (ENOMEM); /* * Register the bus. */ mtx_lock(&adw->lock); if (xpt_bus_register(adw->sim, adw->device, 0) != CAM_SUCCESS) { cam_sim_free(adw->sim, /*free devq*/TRUE); error = ENOMEM; goto fail; } if (xpt_create_path(&adw->path, /*periph*/NULL, cam_sim_path(adw->sim), CAM_TARGET_WILDCARD, CAM_LUN_WILDCARD) == CAM_REQ_CMP) { xpt_setup_ccb(&csa.ccb_h, adw->path, /*priority*/5); csa.ccb_h.func_code = XPT_SASYNC_CB; csa.event_enable = AC_LOST_DEVICE; csa.callback = adw_async; csa.callback_arg = adw; xpt_action((union ccb *)&csa); } fail: mtx_unlock(&adw->lock); return (error); } void adw_intr(void *arg) { struct adw_softc *adw; adw = arg; mtx_lock(&adw->lock); adw_intr_locked(adw); mtx_unlock(&adw->lock); } void adw_intr_locked(struct adw_softc *adw) { u_int int_stat; if ((adw_inw(adw, ADW_CTRL_REG) & ADW_CTRL_REG_HOST_INTR) == 0) return; /* Reading the register clears the interrupt. */ int_stat = adw_inb(adw, ADW_INTR_STATUS_REG); if ((int_stat & ADW_INTR_STATUS_INTRB) != 0) { u_int intrb_code; /* Async Microcode Event */ intrb_code = adw_lram_read_8(adw, ADW_MC_INTRB_CODE); switch (intrb_code) { case ADW_ASYNC_CARRIER_READY_FAILURE: /* * The RISC missed our update of * the commandq. */ if (LIST_FIRST(&adw->pending_ccbs) != NULL) adw_tickle_risc(adw, ADW_TICKLE_A); break; case ADW_ASYNC_SCSI_BUS_RESET_DET: /* * The firmware detected a SCSI Bus reset. */ device_printf(adw->device, "Someone Reset the Bus\n"); adw_handle_bus_reset(adw, /*initiated*/FALSE); break; case ADW_ASYNC_RDMA_FAILURE: /* * Handle RDMA failure by resetting the * SCSI Bus and chip. */ #if 0 /* XXX */ AdvResetChipAndSB(adv_dvc_varp); #endif break; case ADW_ASYNC_HOST_SCSI_BUS_RESET: /* * Host generated SCSI bus reset occurred. */ adw_handle_bus_reset(adw, /*initiated*/TRUE); break; default: printf("adw_intr: unknown async code 0x%x\n", intrb_code); break; } } /* * Run down the RequestQ. */ while ((adw->responseq->next_ba & ADW_RQ_DONE) != 0) { struct adw_carrier *free_carrier; struct acb *acb; union ccb *ccb; #if 0 printf("0x%x, 0x%x, 0x%x, 0x%x\n", adw->responseq->carr_offset, adw->responseq->carr_ba, adw->responseq->areq_ba, adw->responseq->next_ba); #endif /* * The firmware copies the adw_scsi_req_q.acb_baddr * field into the areq_ba field of the carrier. */ acb = acbbotov(adw, adw->responseq->areq_ba); /* * The least significant four bits of the next_ba * field are used as flags. Mask them out and then * advance through the list. */ free_carrier = adw->responseq; adw->responseq = carrierbotov(adw, free_carrier->next_ba & ADW_NEXT_BA_MASK); free_carrier->next_ba = adw->free_carriers->carr_offset; adw->free_carriers = free_carrier; /* Process CCB */ ccb = acb->ccb; callout_stop(&acb->timer); if ((ccb->ccb_h.flags & CAM_DIR_MASK) != CAM_DIR_NONE) { bus_dmasync_op_t op; if ((ccb->ccb_h.flags & CAM_DIR_MASK) == CAM_DIR_IN) op = BUS_DMASYNC_POSTREAD; else op = BUS_DMASYNC_POSTWRITE; bus_dmamap_sync(adw->buffer_dmat, acb->dmamap, op); bus_dmamap_unload(adw->buffer_dmat, acb->dmamap); ccb->csio.resid = acb->queue.data_cnt; } else ccb->csio.resid = 0; /* Common Cases inline... */ if (acb->queue.host_status == QHSTA_NO_ERROR && (acb->queue.done_status == QD_NO_ERROR || acb->queue.done_status == QD_WITH_ERROR)) { ccb->csio.scsi_status = acb->queue.scsi_status; ccb->ccb_h.status = 0; switch (ccb->csio.scsi_status) { case SCSI_STATUS_OK: ccb->ccb_h.status |= CAM_REQ_CMP; break; case SCSI_STATUS_CHECK_COND: case SCSI_STATUS_CMD_TERMINATED: bcopy(&acb->sense_data, &ccb->csio.sense_data, ccb->csio.sense_len); ccb->ccb_h.status |= CAM_AUTOSNS_VALID; ccb->csio.sense_resid = acb->queue.sense_len; /* FALLTHROUGH */ default: ccb->ccb_h.status |= CAM_SCSI_STATUS_ERROR | CAM_DEV_QFRZN; xpt_freeze_devq(ccb->ccb_h.path, /*count*/1); break; } adwfreeacb(adw, acb); xpt_done(ccb); } else { adwprocesserror(adw, acb); } } } static void adwprocesserror(struct adw_softc *adw, struct acb *acb) { union ccb *ccb; ccb = acb->ccb; if (acb->queue.done_status == QD_ABORTED_BY_HOST) { ccb->ccb_h.status = CAM_REQ_ABORTED; } else { switch (acb->queue.host_status) { case QHSTA_M_SEL_TIMEOUT: ccb->ccb_h.status = CAM_SEL_TIMEOUT; break; case QHSTA_M_SXFR_OFF_UFLW: case QHSTA_M_SXFR_OFF_OFLW: case QHSTA_M_DATA_OVER_RUN: ccb->ccb_h.status = CAM_DATA_RUN_ERR; break; case QHSTA_M_SXFR_DESELECTED: case QHSTA_M_UNEXPECTED_BUS_FREE: ccb->ccb_h.status = CAM_UNEXP_BUSFREE; break; case QHSTA_M_SCSI_BUS_RESET: case QHSTA_M_SCSI_BUS_RESET_UNSOL: ccb->ccb_h.status = CAM_SCSI_BUS_RESET; break; case QHSTA_M_BUS_DEVICE_RESET: ccb->ccb_h.status = CAM_BDR_SENT; break; case QHSTA_M_QUEUE_ABORTED: /* BDR or Bus Reset */ xpt_print_path(adw->path); printf("Saw Queue Aborted\n"); ccb->ccb_h.status = adw->last_reset; break; case QHSTA_M_SXFR_SDMA_ERR: case QHSTA_M_SXFR_SXFR_PERR: case QHSTA_M_RDMA_PERR: ccb->ccb_h.status = CAM_UNCOR_PARITY; break; case QHSTA_M_WTM_TIMEOUT: case QHSTA_M_SXFR_WD_TMO: { /* The SCSI bus hung in a phase */ xpt_print_path(adw->path); printf("Watch Dog timer expired. Resetting bus\n"); adw_reset_bus(adw); break; } case QHSTA_M_SXFR_XFR_PH_ERR: ccb->ccb_h.status = CAM_SEQUENCE_FAIL; break; case QHSTA_M_SXFR_UNKNOWN_ERROR: break; case QHSTA_M_BAD_CMPL_STATUS_IN: /* No command complete after a status message */ ccb->ccb_h.status = CAM_SEQUENCE_FAIL; break; case QHSTA_M_AUTO_REQ_SENSE_FAIL: ccb->ccb_h.status = CAM_AUTOSENSE_FAIL; break; case QHSTA_M_INVALID_DEVICE: ccb->ccb_h.status = CAM_PATH_INVALID; break; case QHSTA_M_NO_AUTO_REQ_SENSE: /* * User didn't request sense, but we got a * check condition. */ ccb->csio.scsi_status = acb->queue.scsi_status; ccb->ccb_h.status = CAM_SCSI_STATUS_ERROR; break; default: panic("%s: Unhandled Host status error %x", device_get_nameunit(adw->device), acb->queue.host_status); /* NOTREACHED */ } } if ((acb->state & ACB_RECOVERY_ACB) != 0) { if (ccb->ccb_h.status == CAM_SCSI_BUS_RESET || ccb->ccb_h.status == CAM_BDR_SENT) ccb->ccb_h.status = CAM_CMD_TIMEOUT; } if (ccb->ccb_h.status != CAM_REQ_CMP) { xpt_freeze_devq(ccb->ccb_h.path, /*count*/1); ccb->ccb_h.status |= CAM_DEV_QFRZN; } adwfreeacb(adw, acb); xpt_done(ccb); } static void adwtimeout(void *arg) { struct acb *acb; union ccb *ccb; struct adw_softc *adw; adw_idle_cmd_status_t status; int target_id; acb = (struct acb *)arg; ccb = acb->ccb; adw = (struct adw_softc *)ccb->ccb_h.ccb_adw_ptr; xpt_print_path(ccb->ccb_h.path); printf("ACB %p - timed out\n", (void *)acb); mtx_assert(&adw->lock, MA_OWNED); if ((acb->state & ACB_ACTIVE) == 0) { xpt_print_path(ccb->ccb_h.path); printf("ACB %p - timed out CCB already completed\n", (void *)acb); return; } acb->state |= ACB_RECOVERY_ACB; target_id = ccb->ccb_h.target_id; /* Attempt a BDR first */ status = adw_idle_cmd_send(adw, ADW_IDLE_CMD_DEVICE_RESET, ccb->ccb_h.target_id); if (status == ADW_IDLE_CMD_SUCCESS) { device_printf(adw->device, "BDR Delivered. No longer in timeout\n"); adw_handle_device_reset(adw, target_id); } else { adw_reset_bus(adw); xpt_print_path(adw->path); printf("Bus Reset Delivered. No longer in timeout\n"); } } static void adw_handle_device_reset(struct adw_softc *adw, u_int target) { struct cam_path *path; cam_status error; error = xpt_create_path(&path, /*periph*/NULL, cam_sim_path(adw->sim), target, CAM_LUN_WILDCARD); if (error == CAM_REQ_CMP) { xpt_async(AC_SENT_BDR, path, NULL); xpt_free_path(path); } adw->last_reset = CAM_BDR_SENT; } static void adw_handle_bus_reset(struct adw_softc *adw, int initiated) { if (initiated) { /* * The microcode currently sets the SCSI Bus Reset signal * while handling the AscSendIdleCmd() IDLE_CMD_SCSI_RESET * command above. But the SCSI Bus Reset Hold Time in the * microcode is not deterministic (it may in fact be for less * than the SCSI Spec. minimum of 25 us). Therefore on return * the Adv Library sets the SCSI Bus Reset signal for * ADW_SCSI_RESET_HOLD_TIME_US, which is defined to be greater * than 25 us. */ u_int scsi_ctrl; scsi_ctrl = adw_inw(adw, ADW_SCSI_CTRL) & ~ADW_SCSI_CTRL_RSTOUT; adw_outw(adw, ADW_SCSI_CTRL, scsi_ctrl | ADW_SCSI_CTRL_RSTOUT); DELAY(ADW_SCSI_RESET_HOLD_TIME_US); adw_outw(adw, ADW_SCSI_CTRL, scsi_ctrl); /* * We will perform the async notification when the * SCSI Reset interrupt occurs. */ } else xpt_async(AC_BUS_RESET, adw->path, NULL); adw->last_reset = CAM_SCSI_BUS_RESET; } MODULE_DEPEND(adw, cam, 1, 1, 1); Index: projects/clang400-import/sys/dev/agp/agp_i810.c =================================================================== --- projects/clang400-import/sys/dev/agp/agp_i810.c (revision 311459) +++ projects/clang400-import/sys/dev/agp/agp_i810.c (revision 311460) @@ -1,2359 +1,2357 @@ /*- * Copyright (c) 2000 Doug Rabson * Copyright (c) 2000 Ruslan Ermilov * Copyright (c) 2011 The FreeBSD Foundation * All rights reserved. * * Portions of this software were developed by Konstantin Belousov * 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. */ /* * Fixes for 830/845G support: David Dawes * 852GM/855GM/865G support added by David Dawes * * This is generic Intel GTT handling code, morphed from the AGP * bridge code. */ #include __FBSDID("$FreeBSD$"); #if 0 #define KTR_AGP_I810 KTR_DEV #else #define KTR_AGP_I810 0 #endif #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include MALLOC_DECLARE(M_AGP); struct agp_i810_match; static int agp_i810_check_active(device_t bridge_dev); static int agp_i830_check_active(device_t bridge_dev); static int agp_i915_check_active(device_t bridge_dev); static void agp_82852_set_desc(device_t dev, const struct agp_i810_match *match); static void agp_i810_set_desc(device_t dev, const struct agp_i810_match *match); static void agp_i810_dump_regs(device_t dev); static void agp_i830_dump_regs(device_t dev); static void agp_i855_dump_regs(device_t dev); static void agp_i915_dump_regs(device_t dev); static void agp_i965_dump_regs(device_t dev); static int agp_i810_get_stolen_size(device_t dev); static int agp_i830_get_stolen_size(device_t dev); static int agp_i915_get_stolen_size(device_t dev); static int agp_i810_get_gtt_mappable_entries(device_t dev); static int agp_i830_get_gtt_mappable_entries(device_t dev); static int agp_i915_get_gtt_mappable_entries(device_t dev); static int agp_i810_get_gtt_total_entries(device_t dev); static int agp_i965_get_gtt_total_entries(device_t dev); static int agp_gen5_get_gtt_total_entries(device_t dev); static int agp_i810_install_gatt(device_t dev); static int agp_i830_install_gatt(device_t dev); static int agp_i965_install_gatt(device_t dev); static int agp_g4x_install_gatt(device_t dev); static void agp_i810_deinstall_gatt(device_t dev); static void agp_i830_deinstall_gatt(device_t dev); static void agp_i810_install_gtt_pte(device_t dev, u_int index, vm_offset_t physical, int flags); static void agp_i830_install_gtt_pte(device_t dev, u_int index, vm_offset_t physical, int flags); static void agp_i915_install_gtt_pte(device_t dev, u_int index, vm_offset_t physical, int flags); static void agp_i965_install_gtt_pte(device_t dev, u_int index, vm_offset_t physical, int flags); static void agp_g4x_install_gtt_pte(device_t dev, u_int index, vm_offset_t physical, int flags); static void agp_i810_write_gtt(device_t dev, u_int index, uint32_t pte); static void agp_i915_write_gtt(device_t dev, u_int index, uint32_t pte); static void agp_i965_write_gtt(device_t dev, u_int index, uint32_t pte); static void agp_g4x_write_gtt(device_t dev, u_int index, uint32_t pte); static u_int32_t agp_i810_read_gtt_pte(device_t dev, u_int index); static u_int32_t agp_i915_read_gtt_pte(device_t dev, u_int index); static u_int32_t agp_i965_read_gtt_pte(device_t dev, u_int index); static u_int32_t agp_g4x_read_gtt_pte(device_t dev, u_int index); static vm_paddr_t agp_i810_read_gtt_pte_paddr(device_t dev, u_int index); static vm_paddr_t agp_i915_read_gtt_pte_paddr(device_t dev, u_int index); static int agp_i810_set_aperture(device_t dev, u_int32_t aperture); static int agp_i830_set_aperture(device_t dev, u_int32_t aperture); static int agp_i915_set_aperture(device_t dev, u_int32_t aperture); static int agp_i810_chipset_flush_setup(device_t dev); static int agp_i915_chipset_flush_setup(device_t dev); static int agp_i965_chipset_flush_setup(device_t dev); static void agp_i810_chipset_flush_teardown(device_t dev); static void agp_i915_chipset_flush_teardown(device_t dev); static void agp_i965_chipset_flush_teardown(device_t dev); static void agp_i810_chipset_flush(device_t dev); static void agp_i830_chipset_flush(device_t dev); static void agp_i915_chipset_flush(device_t dev); enum { CHIP_I810, /* i810/i815 */ CHIP_I830, /* 830M/845G */ CHIP_I855, /* 852GM/855GM/865G */ CHIP_I915, /* 915G/915GM */ CHIP_I965, /* G965 */ CHIP_G33, /* G33/Q33/Q35 */ CHIP_IGD, /* Pineview */ CHIP_G4X, /* G45/Q45 */ }; /* The i810 through i855 have the registers at BAR 1, and the GATT gets * allocated by us. The i915 has registers in BAR 0 and the GATT is at the * start of the stolen memory, and should only be accessed by the OS through * BAR 3. The G965 has registers and GATT in the same BAR (0) -- first 512KB * is registers, second 512KB is GATT. */ static struct resource_spec agp_i810_res_spec[] = { { SYS_RES_MEMORY, AGP_I810_MMADR, RF_ACTIVE | RF_SHAREABLE }, { -1, 0 } }; static struct resource_spec agp_i915_res_spec[] = { { SYS_RES_MEMORY, AGP_I915_MMADR, RF_ACTIVE | RF_SHAREABLE }, { SYS_RES_MEMORY, AGP_I915_GTTADR, RF_ACTIVE | RF_SHAREABLE }, { -1, 0 } }; static struct resource_spec agp_i965_res_spec[] = { { SYS_RES_MEMORY, AGP_I965_GTTMMADR, RF_ACTIVE | RF_SHAREABLE }, { SYS_RES_MEMORY, AGP_I965_APBASE, RF_ACTIVE | RF_SHAREABLE }, { -1, 0 } }; struct agp_i810_softc { struct agp_softc agp; u_int32_t initial_aperture; /* aperture size at startup */ struct agp_gatt *gatt; u_int32_t dcache_size; /* i810 only */ u_int32_t stolen; /* number of i830/845 gtt entries for stolen memory */ u_int stolen_size; /* BIOS-reserved graphics memory */ u_int gtt_total_entries; /* Total number of gtt ptes */ u_int gtt_mappable_entries; /* Number of gtt ptes mappable by CPU */ device_t bdev; /* bridge device */ void *argb_cursor; /* contigmalloc area for ARGB cursor */ struct resource *sc_res[2]; const struct agp_i810_match *match; int sc_flush_page_rid; struct resource *sc_flush_page_res; void *sc_flush_page_vaddr; int sc_bios_allocated_flush_page; }; static device_t intel_agp; struct agp_i810_driver { int chiptype; int gen; int busdma_addr_mask_sz; struct resource_spec *res_spec; int (*check_active)(device_t); void (*set_desc)(device_t, const struct agp_i810_match *); void (*dump_regs)(device_t); int (*get_stolen_size)(device_t); int (*get_gtt_total_entries)(device_t); int (*get_gtt_mappable_entries)(device_t); int (*install_gatt)(device_t); void (*deinstall_gatt)(device_t); void (*write_gtt)(device_t, u_int, uint32_t); void (*install_gtt_pte)(device_t, u_int, vm_offset_t, int); u_int32_t (*read_gtt_pte)(device_t, u_int); vm_paddr_t (*read_gtt_pte_paddr)(device_t , u_int); int (*set_aperture)(device_t, u_int32_t); int (*chipset_flush_setup)(device_t); void (*chipset_flush_teardown)(device_t); void (*chipset_flush)(device_t); }; static struct { struct intel_gtt base; } intel_private; static const struct agp_i810_driver agp_i810_i810_driver = { .chiptype = CHIP_I810, .gen = 1, .busdma_addr_mask_sz = 32, .res_spec = agp_i810_res_spec, .check_active = agp_i810_check_active, .set_desc = agp_i810_set_desc, .dump_regs = agp_i810_dump_regs, .get_stolen_size = agp_i810_get_stolen_size, .get_gtt_mappable_entries = agp_i810_get_gtt_mappable_entries, .get_gtt_total_entries = agp_i810_get_gtt_total_entries, .install_gatt = agp_i810_install_gatt, .deinstall_gatt = agp_i810_deinstall_gatt, .write_gtt = agp_i810_write_gtt, .install_gtt_pte = agp_i810_install_gtt_pte, .read_gtt_pte = agp_i810_read_gtt_pte, .read_gtt_pte_paddr = agp_i810_read_gtt_pte_paddr, .set_aperture = agp_i810_set_aperture, .chipset_flush_setup = agp_i810_chipset_flush_setup, .chipset_flush_teardown = agp_i810_chipset_flush_teardown, .chipset_flush = agp_i810_chipset_flush, }; static const struct agp_i810_driver agp_i810_i815_driver = { .chiptype = CHIP_I810, .gen = 2, .busdma_addr_mask_sz = 32, .res_spec = agp_i810_res_spec, .check_active = agp_i810_check_active, .set_desc = agp_i810_set_desc, .dump_regs = agp_i810_dump_regs, .get_stolen_size = agp_i810_get_stolen_size, .get_gtt_mappable_entries = agp_i830_get_gtt_mappable_entries, .get_gtt_total_entries = agp_i810_get_gtt_total_entries, .install_gatt = agp_i810_install_gatt, .deinstall_gatt = agp_i810_deinstall_gatt, .write_gtt = agp_i810_write_gtt, .install_gtt_pte = agp_i810_install_gtt_pte, .read_gtt_pte = agp_i810_read_gtt_pte, .read_gtt_pte_paddr = agp_i810_read_gtt_pte_paddr, .set_aperture = agp_i810_set_aperture, .chipset_flush_setup = agp_i810_chipset_flush_setup, .chipset_flush_teardown = agp_i810_chipset_flush_teardown, .chipset_flush = agp_i830_chipset_flush, }; static const struct agp_i810_driver agp_i810_i830_driver = { .chiptype = CHIP_I830, .gen = 2, .busdma_addr_mask_sz = 32, .res_spec = agp_i810_res_spec, .check_active = agp_i830_check_active, .set_desc = agp_i810_set_desc, .dump_regs = agp_i830_dump_regs, .get_stolen_size = agp_i830_get_stolen_size, .get_gtt_mappable_entries = agp_i830_get_gtt_mappable_entries, .get_gtt_total_entries = agp_i810_get_gtt_total_entries, .install_gatt = agp_i830_install_gatt, .deinstall_gatt = agp_i830_deinstall_gatt, .write_gtt = agp_i810_write_gtt, .install_gtt_pte = agp_i830_install_gtt_pte, .read_gtt_pte = agp_i810_read_gtt_pte, .read_gtt_pte_paddr = agp_i810_read_gtt_pte_paddr, .set_aperture = agp_i830_set_aperture, .chipset_flush_setup = agp_i810_chipset_flush_setup, .chipset_flush_teardown = agp_i810_chipset_flush_teardown, .chipset_flush = agp_i830_chipset_flush, }; static const struct agp_i810_driver agp_i810_i855_driver = { .chiptype = CHIP_I855, .gen = 2, .busdma_addr_mask_sz = 32, .res_spec = agp_i810_res_spec, .check_active = agp_i830_check_active, .set_desc = agp_82852_set_desc, .dump_regs = agp_i855_dump_regs, .get_stolen_size = agp_i915_get_stolen_size, .get_gtt_mappable_entries = agp_i915_get_gtt_mappable_entries, .get_gtt_total_entries = agp_i810_get_gtt_total_entries, .install_gatt = agp_i830_install_gatt, .deinstall_gatt = agp_i830_deinstall_gatt, .write_gtt = agp_i810_write_gtt, .install_gtt_pte = agp_i830_install_gtt_pte, .read_gtt_pte = agp_i810_read_gtt_pte, .read_gtt_pte_paddr = agp_i810_read_gtt_pte_paddr, .set_aperture = agp_i830_set_aperture, .chipset_flush_setup = agp_i810_chipset_flush_setup, .chipset_flush_teardown = agp_i810_chipset_flush_teardown, .chipset_flush = agp_i830_chipset_flush, }; static const struct agp_i810_driver agp_i810_i865_driver = { .chiptype = CHIP_I855, .gen = 2, .busdma_addr_mask_sz = 32, .res_spec = agp_i810_res_spec, .check_active = agp_i830_check_active, .set_desc = agp_i810_set_desc, .dump_regs = agp_i855_dump_regs, .get_stolen_size = agp_i915_get_stolen_size, .get_gtt_mappable_entries = agp_i915_get_gtt_mappable_entries, .get_gtt_total_entries = agp_i810_get_gtt_total_entries, .install_gatt = agp_i830_install_gatt, .deinstall_gatt = agp_i830_deinstall_gatt, .write_gtt = agp_i810_write_gtt, .install_gtt_pte = agp_i830_install_gtt_pte, .read_gtt_pte = agp_i810_read_gtt_pte, .read_gtt_pte_paddr = agp_i810_read_gtt_pte_paddr, .set_aperture = agp_i915_set_aperture, .chipset_flush_setup = agp_i810_chipset_flush_setup, .chipset_flush_teardown = agp_i810_chipset_flush_teardown, .chipset_flush = agp_i830_chipset_flush, }; static const struct agp_i810_driver agp_i810_i915_driver = { .chiptype = CHIP_I915, .gen = 3, .busdma_addr_mask_sz = 32, .res_spec = agp_i915_res_spec, .check_active = agp_i915_check_active, .set_desc = agp_i810_set_desc, .dump_regs = agp_i915_dump_regs, .get_stolen_size = agp_i915_get_stolen_size, .get_gtt_mappable_entries = agp_i915_get_gtt_mappable_entries, .get_gtt_total_entries = agp_i810_get_gtt_total_entries, .install_gatt = agp_i830_install_gatt, .deinstall_gatt = agp_i830_deinstall_gatt, .write_gtt = agp_i915_write_gtt, .install_gtt_pte = agp_i915_install_gtt_pte, .read_gtt_pte = agp_i915_read_gtt_pte, .read_gtt_pte_paddr = agp_i915_read_gtt_pte_paddr, .set_aperture = agp_i915_set_aperture, .chipset_flush_setup = agp_i915_chipset_flush_setup, .chipset_flush_teardown = agp_i915_chipset_flush_teardown, .chipset_flush = agp_i915_chipset_flush, }; static const struct agp_i810_driver agp_i810_g33_driver = { .chiptype = CHIP_G33, .gen = 3, .busdma_addr_mask_sz = 36, .res_spec = agp_i915_res_spec, .check_active = agp_i915_check_active, .set_desc = agp_i810_set_desc, .dump_regs = agp_i965_dump_regs, .get_stolen_size = agp_i915_get_stolen_size, .get_gtt_mappable_entries = agp_i915_get_gtt_mappable_entries, .get_gtt_total_entries = agp_i965_get_gtt_total_entries, .install_gatt = agp_i830_install_gatt, .deinstall_gatt = agp_i830_deinstall_gatt, .write_gtt = agp_i915_write_gtt, .install_gtt_pte = agp_i915_install_gtt_pte, .read_gtt_pte = agp_i915_read_gtt_pte, .read_gtt_pte_paddr = agp_i915_read_gtt_pte_paddr, .set_aperture = agp_i915_set_aperture, .chipset_flush_setup = agp_i965_chipset_flush_setup, .chipset_flush_teardown = agp_i965_chipset_flush_teardown, .chipset_flush = agp_i915_chipset_flush, }; static const struct agp_i810_driver agp_i810_igd_driver = { .chiptype = CHIP_IGD, .gen = 3, .busdma_addr_mask_sz = 36, .res_spec = agp_i915_res_spec, .check_active = agp_i915_check_active, .set_desc = agp_i810_set_desc, .dump_regs = agp_i915_dump_regs, .get_stolen_size = agp_i915_get_stolen_size, .get_gtt_mappable_entries = agp_i915_get_gtt_mappable_entries, .get_gtt_total_entries = agp_i965_get_gtt_total_entries, .install_gatt = agp_i830_install_gatt, .deinstall_gatt = agp_i830_deinstall_gatt, .write_gtt = agp_i915_write_gtt, .install_gtt_pte = agp_i915_install_gtt_pte, .read_gtt_pte = agp_i915_read_gtt_pte, .read_gtt_pte_paddr = agp_i915_read_gtt_pte_paddr, .set_aperture = agp_i915_set_aperture, .chipset_flush_setup = agp_i965_chipset_flush_setup, .chipset_flush_teardown = agp_i965_chipset_flush_teardown, .chipset_flush = agp_i915_chipset_flush, }; static const struct agp_i810_driver agp_i810_g965_driver = { .chiptype = CHIP_I965, .gen = 4, .busdma_addr_mask_sz = 36, .res_spec = agp_i965_res_spec, .check_active = agp_i915_check_active, .set_desc = agp_i810_set_desc, .dump_regs = agp_i965_dump_regs, .get_stolen_size = agp_i915_get_stolen_size, .get_gtt_mappable_entries = agp_i915_get_gtt_mappable_entries, .get_gtt_total_entries = agp_i965_get_gtt_total_entries, .install_gatt = agp_i965_install_gatt, .deinstall_gatt = agp_i830_deinstall_gatt, .write_gtt = agp_i965_write_gtt, .install_gtt_pte = agp_i965_install_gtt_pte, .read_gtt_pte = agp_i965_read_gtt_pte, .read_gtt_pte_paddr = agp_i915_read_gtt_pte_paddr, .set_aperture = agp_i915_set_aperture, .chipset_flush_setup = agp_i965_chipset_flush_setup, .chipset_flush_teardown = agp_i965_chipset_flush_teardown, .chipset_flush = agp_i915_chipset_flush, }; static const struct agp_i810_driver agp_i810_g4x_driver = { .chiptype = CHIP_G4X, .gen = 5, .busdma_addr_mask_sz = 36, .res_spec = agp_i965_res_spec, .check_active = agp_i915_check_active, .set_desc = agp_i810_set_desc, .dump_regs = agp_i965_dump_regs, .get_stolen_size = agp_i915_get_stolen_size, .get_gtt_mappable_entries = agp_i915_get_gtt_mappable_entries, .get_gtt_total_entries = agp_gen5_get_gtt_total_entries, .install_gatt = agp_g4x_install_gatt, .deinstall_gatt = agp_i830_deinstall_gatt, .write_gtt = agp_g4x_write_gtt, .install_gtt_pte = agp_g4x_install_gtt_pte, .read_gtt_pte = agp_g4x_read_gtt_pte, .read_gtt_pte_paddr = agp_i915_read_gtt_pte_paddr, .set_aperture = agp_i915_set_aperture, .chipset_flush_setup = agp_i965_chipset_flush_setup, .chipset_flush_teardown = agp_i965_chipset_flush_teardown, .chipset_flush = agp_i915_chipset_flush, }; /* For adding new devices, devid is the id of the graphics controller * (pci:0:2:0, for example). The placeholder (usually at pci:0:2:1) for the * second head should never be added. The bridge_offset is the offset to * subtract from devid to get the id of the hostb that the device is on. */ static const struct agp_i810_match { int devid; char *name; const struct agp_i810_driver *driver; } agp_i810_matches[] = { { .devid = 0x71218086, .name = "Intel 82810 (i810 GMCH) SVGA controller", .driver = &agp_i810_i810_driver }, { .devid = 0x71238086, .name = "Intel 82810-DC100 (i810-DC100 GMCH) SVGA controller", .driver = &agp_i810_i810_driver }, { .devid = 0x71258086, .name = "Intel 82810E (i810E GMCH) SVGA controller", .driver = &agp_i810_i810_driver }, { .devid = 0x11328086, .name = "Intel 82815 (i815 GMCH) SVGA controller", .driver = &agp_i810_i815_driver }, { .devid = 0x35778086, .name = "Intel 82830M (830M GMCH) SVGA controller", .driver = &agp_i810_i830_driver }, { .devid = 0x25628086, .name = "Intel 82845M (845M GMCH) SVGA controller", .driver = &agp_i810_i830_driver }, { .devid = 0x35828086, .name = "Intel 82852/855GM SVGA controller", .driver = &agp_i810_i855_driver }, { .devid = 0x25728086, .name = "Intel 82865G (865G GMCH) SVGA controller", .driver = &agp_i810_i865_driver }, { .devid = 0x25828086, .name = "Intel 82915G (915G GMCH) SVGA controller", .driver = &agp_i810_i915_driver }, { .devid = 0x258A8086, .name = "Intel E7221 SVGA controller", .driver = &agp_i810_i915_driver }, { .devid = 0x25928086, .name = "Intel 82915GM (915GM GMCH) SVGA controller", .driver = &agp_i810_i915_driver }, { .devid = 0x27728086, .name = "Intel 82945G (945G GMCH) SVGA controller", .driver = &agp_i810_i915_driver }, { .devid = 0x27A28086, .name = "Intel 82945GM (945GM GMCH) SVGA controller", .driver = &agp_i810_i915_driver }, { .devid = 0x27AE8086, .name = "Intel 945GME SVGA controller", .driver = &agp_i810_i915_driver }, { .devid = 0x29728086, .name = "Intel 946GZ SVGA controller", .driver = &agp_i810_g965_driver }, { .devid = 0x29828086, .name = "Intel G965 SVGA controller", .driver = &agp_i810_g965_driver }, { .devid = 0x29928086, .name = "Intel Q965 SVGA controller", .driver = &agp_i810_g965_driver }, { .devid = 0x29A28086, .name = "Intel G965 SVGA controller", .driver = &agp_i810_g965_driver }, { .devid = 0x29B28086, .name = "Intel Q35 SVGA controller", .driver = &agp_i810_g33_driver }, { .devid = 0x29C28086, .name = "Intel G33 SVGA controller", .driver = &agp_i810_g33_driver }, { .devid = 0x29D28086, .name = "Intel Q33 SVGA controller", .driver = &agp_i810_g33_driver }, { .devid = 0xA0018086, .name = "Intel Pineview SVGA controller", .driver = &agp_i810_igd_driver }, { .devid = 0xA0118086, .name = "Intel Pineview (M) SVGA controller", .driver = &agp_i810_igd_driver }, { .devid = 0x2A028086, .name = "Intel GM965 SVGA controller", .driver = &agp_i810_g965_driver }, { .devid = 0x2A128086, .name = "Intel GME965 SVGA controller", .driver = &agp_i810_g965_driver }, { .devid = 0x2A428086, .name = "Intel GM45 SVGA controller", .driver = &agp_i810_g4x_driver }, { .devid = 0x2E028086, .name = "Intel Eaglelake SVGA controller", .driver = &agp_i810_g4x_driver }, { .devid = 0x2E128086, .name = "Intel Q45 SVGA controller", .driver = &agp_i810_g4x_driver }, { .devid = 0x2E228086, .name = "Intel G45 SVGA controller", .driver = &agp_i810_g4x_driver }, { .devid = 0x2E328086, .name = "Intel G41 SVGA controller", .driver = &agp_i810_g4x_driver }, { .devid = 0x00428086, .name = "Intel Ironlake (D) SVGA controller", .driver = &agp_i810_g4x_driver }, { .devid = 0x00468086, .name = "Intel Ironlake (M) SVGA controller", .driver = &agp_i810_g4x_driver }, { .devid = 0, } }; static const struct agp_i810_match* agp_i810_match(device_t dev) { int i, devid; if (pci_get_class(dev) != PCIC_DISPLAY || (pci_get_subclass(dev) != PCIS_DISPLAY_VGA && pci_get_subclass(dev) != PCIS_DISPLAY_OTHER)) return (NULL); devid = pci_get_devid(dev); for (i = 0; agp_i810_matches[i].devid != 0; i++) { if (agp_i810_matches[i].devid == devid) break; } if (agp_i810_matches[i].devid == 0) return (NULL); else return (&agp_i810_matches[i]); } /* * Find bridge device. */ static device_t agp_i810_find_bridge(device_t dev) { return (pci_find_dbsf(0, 0, 0, 0)); } static void agp_i810_identify(driver_t *driver, device_t parent) { if (device_find_child(parent, "agp", -1) == NULL && agp_i810_match(parent)) device_add_child(parent, "agp", -1); } static int agp_i810_check_active(device_t bridge_dev) { u_int8_t smram; smram = pci_read_config(bridge_dev, AGP_I810_SMRAM, 1); if ((smram & AGP_I810_SMRAM_GMS) == AGP_I810_SMRAM_GMS_DISABLED) return (ENXIO); return (0); } static int agp_i830_check_active(device_t bridge_dev) { int gcc1; gcc1 = pci_read_config(bridge_dev, AGP_I830_GCC1, 1); if ((gcc1 & AGP_I830_GCC1_DEV2) == AGP_I830_GCC1_DEV2_DISABLED) return (ENXIO); return (0); } static int agp_i915_check_active(device_t bridge_dev) { int deven; deven = pci_read_config(bridge_dev, AGP_I915_DEVEN, 4); if ((deven & AGP_I915_DEVEN_D2F0) == AGP_I915_DEVEN_D2F0_DISABLED) return (ENXIO); return (0); } static void agp_82852_set_desc(device_t dev, const struct agp_i810_match *match) { switch (pci_read_config(dev, AGP_I85X_CAPID, 1)) { case AGP_I855_GME: device_set_desc(dev, "Intel 82855GME (855GME GMCH) SVGA controller"); break; case AGP_I855_GM: device_set_desc(dev, "Intel 82855GM (855GM GMCH) SVGA controller"); break; case AGP_I852_GME: device_set_desc(dev, "Intel 82852GME (852GME GMCH) SVGA controller"); break; case AGP_I852_GM: device_set_desc(dev, "Intel 82852GM (852GM GMCH) SVGA controller"); break; default: device_set_desc(dev, "Intel 8285xM (85xGM GMCH) SVGA controller"); break; } } static void agp_i810_set_desc(device_t dev, const struct agp_i810_match *match) { device_set_desc(dev, match->name); } static int agp_i810_probe(device_t dev) { device_t bdev; const struct agp_i810_match *match; int err; if (resource_disabled("agp", device_get_unit(dev))) return (ENXIO); match = agp_i810_match(dev); if (match == NULL) return (ENXIO); bdev = agp_i810_find_bridge(dev); if (bdev == NULL) { if (bootverbose) printf("I810: can't find bridge device\n"); return (ENXIO); } /* * checking whether internal graphics device has been activated. */ err = match->driver->check_active(bdev); if (err != 0) { if (bootverbose) printf("i810: disabled, not probing\n"); return (err); } match->driver->set_desc(dev, match); return (BUS_PROBE_DEFAULT); } static void agp_i810_dump_regs(device_t dev) { struct agp_i810_softc *sc = device_get_softc(dev); device_printf(dev, "AGP_I810_PGTBL_CTL: %08x\n", bus_read_4(sc->sc_res[0], AGP_I810_PGTBL_CTL)); device_printf(dev, "AGP_I810_MISCC: 0x%04x\n", pci_read_config(sc->bdev, AGP_I810_MISCC, 2)); } static void agp_i830_dump_regs(device_t dev) { struct agp_i810_softc *sc = device_get_softc(dev); device_printf(dev, "AGP_I810_PGTBL_CTL: %08x\n", bus_read_4(sc->sc_res[0], AGP_I810_PGTBL_CTL)); device_printf(dev, "AGP_I830_GCC1: 0x%02x\n", pci_read_config(sc->bdev, AGP_I830_GCC1, 1)); } static void agp_i855_dump_regs(device_t dev) { struct agp_i810_softc *sc = device_get_softc(dev); device_printf(dev, "AGP_I810_PGTBL_CTL: %08x\n", bus_read_4(sc->sc_res[0], AGP_I810_PGTBL_CTL)); device_printf(dev, "AGP_I855_GCC1: 0x%02x\n", pci_read_config(sc->bdev, AGP_I855_GCC1, 1)); } static void agp_i915_dump_regs(device_t dev) { struct agp_i810_softc *sc = device_get_softc(dev); device_printf(dev, "AGP_I810_PGTBL_CTL: %08x\n", bus_read_4(sc->sc_res[0], AGP_I810_PGTBL_CTL)); device_printf(dev, "AGP_I855_GCC1: 0x%02x\n", pci_read_config(sc->bdev, AGP_I855_GCC1, 1)); device_printf(dev, "AGP_I915_MSAC: 0x%02x\n", pci_read_config(sc->bdev, AGP_I915_MSAC, 1)); } static void agp_i965_dump_regs(device_t dev) { struct agp_i810_softc *sc = device_get_softc(dev); device_printf(dev, "AGP_I965_PGTBL_CTL2: %08x\n", bus_read_4(sc->sc_res[0], AGP_I965_PGTBL_CTL2)); device_printf(dev, "AGP_I855_GCC1: 0x%02x\n", pci_read_config(sc->bdev, AGP_I855_GCC1, 1)); device_printf(dev, "AGP_I965_MSAC: 0x%02x\n", pci_read_config(sc->bdev, AGP_I965_MSAC, 1)); } static int agp_i810_get_stolen_size(device_t dev) { struct agp_i810_softc *sc; sc = device_get_softc(dev); sc->stolen = 0; sc->stolen_size = 0; return (0); } static int agp_i830_get_stolen_size(device_t dev) { struct agp_i810_softc *sc; unsigned int gcc1; sc = device_get_softc(dev); gcc1 = pci_read_config(sc->bdev, AGP_I830_GCC1, 1); switch (gcc1 & AGP_I830_GCC1_GMS) { case AGP_I830_GCC1_GMS_STOLEN_512: sc->stolen = (512 - 132) * 1024 / 4096; sc->stolen_size = 512 * 1024; break; case AGP_I830_GCC1_GMS_STOLEN_1024: sc->stolen = (1024 - 132) * 1024 / 4096; sc->stolen_size = 1024 * 1024; break; case AGP_I830_GCC1_GMS_STOLEN_8192: sc->stolen = (8192 - 132) * 1024 / 4096; sc->stolen_size = 8192 * 1024; break; default: sc->stolen = 0; device_printf(dev, "unknown memory configuration, disabling (GCC1 %x)\n", gcc1); return (EINVAL); } return (0); } static int agp_i915_get_stolen_size(device_t dev) { struct agp_i810_softc *sc; unsigned int gcc1, stolen, gtt_size; sc = device_get_softc(dev); /* * Stolen memory is set up at the beginning of the aperture by * the BIOS, consisting of the GATT followed by 4kb for the * BIOS display. */ switch (sc->match->driver->chiptype) { case CHIP_I855: gtt_size = 128; break; case CHIP_I915: gtt_size = 256; break; case CHIP_I965: switch (bus_read_4(sc->sc_res[0], AGP_I810_PGTBL_CTL) & AGP_I810_PGTBL_SIZE_MASK) { case AGP_I810_PGTBL_SIZE_128KB: gtt_size = 128; break; case AGP_I810_PGTBL_SIZE_256KB: gtt_size = 256; break; case AGP_I810_PGTBL_SIZE_512KB: gtt_size = 512; break; case AGP_I965_PGTBL_SIZE_1MB: gtt_size = 1024; break; case AGP_I965_PGTBL_SIZE_2MB: gtt_size = 2048; break; case AGP_I965_PGTBL_SIZE_1_5MB: gtt_size = 1024 + 512; break; default: device_printf(dev, "Bad PGTBL size\n"); return (EINVAL); } break; case CHIP_G33: gcc1 = pci_read_config(sc->bdev, AGP_I855_GCC1, 2); switch (gcc1 & AGP_G33_MGGC_GGMS_MASK) { case AGP_G33_MGGC_GGMS_SIZE_1M: gtt_size = 1024; break; case AGP_G33_MGGC_GGMS_SIZE_2M: gtt_size = 2048; break; default: device_printf(dev, "Bad PGTBL size\n"); return (EINVAL); } break; case CHIP_IGD: case CHIP_G4X: gtt_size = 0; break; default: device_printf(dev, "Bad chiptype\n"); return (EINVAL); } /* GCC1 is called MGGC on i915+ */ gcc1 = pci_read_config(sc->bdev, AGP_I855_GCC1, 1); switch (gcc1 & AGP_I855_GCC1_GMS) { case AGP_I855_GCC1_GMS_STOLEN_1M: stolen = 1024; break; case AGP_I855_GCC1_GMS_STOLEN_4M: stolen = 4 * 1024; break; case AGP_I855_GCC1_GMS_STOLEN_8M: stolen = 8 * 1024; break; case AGP_I855_GCC1_GMS_STOLEN_16M: stolen = 16 * 1024; break; case AGP_I855_GCC1_GMS_STOLEN_32M: stolen = 32 * 1024; break; case AGP_I915_GCC1_GMS_STOLEN_48M: stolen = sc->match->driver->gen > 2 ? 48 * 1024 : 0; break; case AGP_I915_GCC1_GMS_STOLEN_64M: stolen = sc->match->driver->gen > 2 ? 64 * 1024 : 0; break; case AGP_G33_GCC1_GMS_STOLEN_128M: stolen = sc->match->driver->gen > 2 ? 128 * 1024 : 0; break; case AGP_G33_GCC1_GMS_STOLEN_256M: stolen = sc->match->driver->gen > 2 ? 256 * 1024 : 0; break; case AGP_G4X_GCC1_GMS_STOLEN_96M: if (sc->match->driver->chiptype == CHIP_I965 || sc->match->driver->chiptype == CHIP_G4X) stolen = 96 * 1024; else stolen = 0; break; case AGP_G4X_GCC1_GMS_STOLEN_160M: if (sc->match->driver->chiptype == CHIP_I965 || sc->match->driver->chiptype == CHIP_G4X) stolen = 160 * 1024; else stolen = 0; break; case AGP_G4X_GCC1_GMS_STOLEN_224M: if (sc->match->driver->chiptype == CHIP_I965 || sc->match->driver->chiptype == CHIP_G4X) stolen = 224 * 1024; else stolen = 0; break; case AGP_G4X_GCC1_GMS_STOLEN_352M: if (sc->match->driver->chiptype == CHIP_I965 || sc->match->driver->chiptype == CHIP_G4X) stolen = 352 * 1024; else stolen = 0; break; default: device_printf(dev, "unknown memory configuration, disabling (GCC1 %x)\n", gcc1); return (EINVAL); } gtt_size += 4; sc->stolen_size = stolen * 1024; sc->stolen = (stolen - gtt_size) * 1024 / 4096; return (0); } static int agp_i810_get_gtt_mappable_entries(device_t dev) { struct agp_i810_softc *sc; uint32_t ap; uint16_t miscc; sc = device_get_softc(dev); miscc = pci_read_config(sc->bdev, AGP_I810_MISCC, 2); if ((miscc & AGP_I810_MISCC_WINSIZE) == AGP_I810_MISCC_WINSIZE_32) ap = 32; else ap = 64; sc->gtt_mappable_entries = (ap * 1024 * 1024) >> AGP_PAGE_SHIFT; return (0); } static int agp_i830_get_gtt_mappable_entries(device_t dev) { struct agp_i810_softc *sc; uint32_t ap; uint16_t gmch_ctl; sc = device_get_softc(dev); gmch_ctl = pci_read_config(sc->bdev, AGP_I830_GCC1, 2); if ((gmch_ctl & AGP_I830_GCC1_GMASIZE) == AGP_I830_GCC1_GMASIZE_64) ap = 64; else ap = 128; sc->gtt_mappable_entries = (ap * 1024 * 1024) >> AGP_PAGE_SHIFT; return (0); } static int agp_i915_get_gtt_mappable_entries(device_t dev) { struct agp_i810_softc *sc; uint32_t ap; sc = device_get_softc(dev); ap = AGP_GET_APERTURE(dev); sc->gtt_mappable_entries = ap >> AGP_PAGE_SHIFT; return (0); } static int agp_i810_get_gtt_total_entries(device_t dev) { struct agp_i810_softc *sc; sc = device_get_softc(dev); sc->gtt_total_entries = sc->gtt_mappable_entries; return (0); } static int agp_i965_get_gtt_total_entries(device_t dev) { struct agp_i810_softc *sc; uint32_t pgetbl_ctl; int error; sc = device_get_softc(dev); error = 0; pgetbl_ctl = bus_read_4(sc->sc_res[0], AGP_I810_PGTBL_CTL); switch (pgetbl_ctl & AGP_I810_PGTBL_SIZE_MASK) { case AGP_I810_PGTBL_SIZE_128KB: sc->gtt_total_entries = 128 * 1024 / 4; break; case AGP_I810_PGTBL_SIZE_256KB: sc->gtt_total_entries = 256 * 1024 / 4; break; case AGP_I810_PGTBL_SIZE_512KB: sc->gtt_total_entries = 512 * 1024 / 4; break; /* GTT pagetable sizes bigger than 512KB are not possible on G33! */ case AGP_I810_PGTBL_SIZE_1MB: sc->gtt_total_entries = 1024 * 1024 / 4; break; case AGP_I810_PGTBL_SIZE_2MB: sc->gtt_total_entries = 2 * 1024 * 1024 / 4; break; case AGP_I810_PGTBL_SIZE_1_5MB: sc->gtt_total_entries = (1024 + 512) * 1024 / 4; break; default: device_printf(dev, "Unknown page table size\n"); error = ENXIO; } return (error); } static void agp_gen5_adjust_pgtbl_size(device_t dev, uint32_t sz) { struct agp_i810_softc *sc; uint32_t pgetbl_ctl, pgetbl_ctl2; sc = device_get_softc(dev); /* Disable per-process page table. */ pgetbl_ctl2 = bus_read_4(sc->sc_res[0], AGP_I965_PGTBL_CTL2); pgetbl_ctl2 &= ~AGP_I810_PGTBL_ENABLED; bus_write_4(sc->sc_res[0], AGP_I965_PGTBL_CTL2, pgetbl_ctl2); /* Write the new ggtt size. */ pgetbl_ctl = bus_read_4(sc->sc_res[0], AGP_I810_PGTBL_CTL); pgetbl_ctl &= ~AGP_I810_PGTBL_SIZE_MASK; pgetbl_ctl |= sz; bus_write_4(sc->sc_res[0], AGP_I810_PGTBL_CTL, pgetbl_ctl); } static int agp_gen5_get_gtt_total_entries(device_t dev) { struct agp_i810_softc *sc; uint16_t gcc1; sc = device_get_softc(dev); gcc1 = pci_read_config(sc->bdev, AGP_I830_GCC1, 2); switch (gcc1 & AGP_G4x_GCC1_SIZE_MASK) { case AGP_G4x_GCC1_SIZE_1M: case AGP_G4x_GCC1_SIZE_VT_1M: agp_gen5_adjust_pgtbl_size(dev, AGP_I810_PGTBL_SIZE_1MB); break; case AGP_G4x_GCC1_SIZE_VT_1_5M: agp_gen5_adjust_pgtbl_size(dev, AGP_I810_PGTBL_SIZE_1_5MB); break; case AGP_G4x_GCC1_SIZE_2M: case AGP_G4x_GCC1_SIZE_VT_2M: agp_gen5_adjust_pgtbl_size(dev, AGP_I810_PGTBL_SIZE_2MB); break; default: device_printf(dev, "Unknown page table size\n"); return (ENXIO); } return (agp_i965_get_gtt_total_entries(dev)); } static int agp_i810_install_gatt(device_t dev) { struct agp_i810_softc *sc; sc = device_get_softc(dev); /* Some i810s have on-chip memory called dcache. */ if ((bus_read_1(sc->sc_res[0], AGP_I810_DRT) & AGP_I810_DRT_POPULATED) != 0) sc->dcache_size = 4 * 1024 * 1024; else sc->dcache_size = 0; /* According to the specs the gatt on the i810 must be 64k. */ sc->gatt->ag_virtual = (void *)kmem_alloc_contig(kernel_arena, 64 * 1024, M_NOWAIT | M_ZERO, 0, ~0, PAGE_SIZE, 0, VM_MEMATTR_WRITE_COMBINING); if (sc->gatt->ag_virtual == NULL) { if (bootverbose) device_printf(dev, "contiguous allocation failed\n"); return (ENOMEM); } sc->gatt->ag_physical = vtophys((vm_offset_t)sc->gatt->ag_virtual); /* Install the GATT. */ bus_write_4(sc->sc_res[0], AGP_I810_PGTBL_CTL, sc->gatt->ag_physical | 1); return (0); } static void agp_i830_install_gatt_init(struct agp_i810_softc *sc) { uint32_t pgtblctl; /* * The i830 automatically initializes the 128k gatt on boot. * GATT address is already in there, make sure it's enabled. */ pgtblctl = bus_read_4(sc->sc_res[0], AGP_I810_PGTBL_CTL); pgtblctl |= 1; bus_write_4(sc->sc_res[0], AGP_I810_PGTBL_CTL, pgtblctl); sc->gatt->ag_physical = pgtblctl & ~1; } static int agp_i830_install_gatt(device_t dev) { struct agp_i810_softc *sc; sc = device_get_softc(dev); agp_i830_install_gatt_init(sc); return (0); } static int agp_gen4_install_gatt(device_t dev, const vm_size_t gtt_offset) { struct agp_i810_softc *sc; sc = device_get_softc(dev); pmap_change_attr((vm_offset_t)rman_get_virtual(sc->sc_res[0]) + gtt_offset, rman_get_size(sc->sc_res[0]) - gtt_offset, VM_MEMATTR_WRITE_COMBINING); agp_i830_install_gatt_init(sc); return (0); } static int agp_i965_install_gatt(device_t dev) { return (agp_gen4_install_gatt(dev, 512 * 1024)); } static int agp_g4x_install_gatt(device_t dev) { return (agp_gen4_install_gatt(dev, 2 * 1024 * 1024)); } static int agp_i810_attach(device_t dev) { struct agp_i810_softc *sc; int error; sc = device_get_softc(dev); sc->bdev = agp_i810_find_bridge(dev); if (sc->bdev == NULL) return (ENOENT); sc->match = agp_i810_match(dev); agp_set_aperture_resource(dev, sc->match->driver->gen <= 2 ? AGP_APBASE : AGP_I915_GMADR); error = agp_generic_attach(dev); if (error) return (error); if (ptoa((vm_paddr_t)Maxmem) > (1ULL << sc->match->driver->busdma_addr_mask_sz) - 1) { device_printf(dev, "agp_i810 does not support physical " "memory above %ju.\n", (uintmax_t)(1ULL << sc->match->driver->busdma_addr_mask_sz) - 1); return (ENOENT); } if (bus_alloc_resources(dev, sc->match->driver->res_spec, sc->sc_res)) { agp_generic_detach(dev); return (ENODEV); } sc->initial_aperture = AGP_GET_APERTURE(dev); sc->gatt = malloc(sizeof(struct agp_gatt), M_AGP, M_WAITOK); sc->gatt->ag_entries = AGP_GET_APERTURE(dev) >> AGP_PAGE_SHIFT; if ((error = sc->match->driver->get_stolen_size(dev)) != 0 || (error = sc->match->driver->install_gatt(dev)) != 0 || (error = sc->match->driver->get_gtt_mappable_entries(dev)) != 0 || (error = sc->match->driver->get_gtt_total_entries(dev)) != 0 || (error = sc->match->driver->chipset_flush_setup(dev)) != 0) { bus_release_resources(dev, sc->match->driver->res_spec, sc->sc_res); free(sc->gatt, M_AGP); agp_generic_detach(dev); return (error); } intel_agp = dev; device_printf(dev, "aperture size is %dM", sc->initial_aperture / 1024 / 1024); if (sc->stolen > 0) printf(", detected %dk stolen memory\n", sc->stolen * 4); else printf("\n"); if (bootverbose) { sc->match->driver->dump_regs(dev); device_printf(dev, "Mappable GTT entries: %d\n", sc->gtt_mappable_entries); device_printf(dev, "Total GTT entries: %d\n", sc->gtt_total_entries); } return (0); } static void agp_i810_deinstall_gatt(device_t dev) { struct agp_i810_softc *sc; sc = device_get_softc(dev); bus_write_4(sc->sc_res[0], AGP_I810_PGTBL_CTL, 0); kmem_free(kernel_arena, (vm_offset_t)sc->gatt->ag_virtual, 64 * 1024); } static void agp_i830_deinstall_gatt(device_t dev) { struct agp_i810_softc *sc; unsigned int pgtblctl; sc = device_get_softc(dev); pgtblctl = bus_read_4(sc->sc_res[0], AGP_I810_PGTBL_CTL); pgtblctl &= ~1; bus_write_4(sc->sc_res[0], AGP_I810_PGTBL_CTL, pgtblctl); } static int agp_i810_detach(device_t dev) { struct agp_i810_softc *sc; sc = device_get_softc(dev); agp_free_cdev(dev); /* Clear the GATT base. */ sc->match->driver->deinstall_gatt(dev); sc->match->driver->chipset_flush_teardown(dev); /* Put the aperture back the way it started. */ AGP_SET_APERTURE(dev, sc->initial_aperture); free(sc->gatt, M_AGP); bus_release_resources(dev, sc->match->driver->res_spec, sc->sc_res); agp_free_res(dev); return (0); } static int agp_i810_resume(device_t dev) { struct agp_i810_softc *sc; sc = device_get_softc(dev); AGP_SET_APERTURE(dev, sc->initial_aperture); /* Install the GATT. */ bus_write_4(sc->sc_res[0], AGP_I810_PGTBL_CTL, sc->gatt->ag_physical | 1); return (bus_generic_resume(dev)); } /** * Sets the PCI resource size of the aperture on i830-class and below chipsets, * while returning failure on later chipsets when an actual change is * requested. * * This whole function is likely bogus, as the kernel would probably need to * reconfigure the placement of the AGP aperture if a larger size is requested, * which doesn't happen currently. */ static int agp_i810_set_aperture(device_t dev, u_int32_t aperture) { struct agp_i810_softc *sc; u_int16_t miscc; sc = device_get_softc(dev); /* * Double check for sanity. */ if (aperture != 32 * 1024 * 1024 && aperture != 64 * 1024 * 1024) { device_printf(dev, "bad aperture size %d\n", aperture); return (EINVAL); } miscc = pci_read_config(sc->bdev, AGP_I810_MISCC, 2); miscc &= ~AGP_I810_MISCC_WINSIZE; if (aperture == 32 * 1024 * 1024) miscc |= AGP_I810_MISCC_WINSIZE_32; else miscc |= AGP_I810_MISCC_WINSIZE_64; pci_write_config(sc->bdev, AGP_I810_MISCC, miscc, 2); return (0); } static int agp_i830_set_aperture(device_t dev, u_int32_t aperture) { struct agp_i810_softc *sc; u_int16_t gcc1; sc = device_get_softc(dev); if (aperture != 64 * 1024 * 1024 && aperture != 128 * 1024 * 1024) { device_printf(dev, "bad aperture size %d\n", aperture); return (EINVAL); } gcc1 = pci_read_config(sc->bdev, AGP_I830_GCC1, 2); gcc1 &= ~AGP_I830_GCC1_GMASIZE; if (aperture == 64 * 1024 * 1024) gcc1 |= AGP_I830_GCC1_GMASIZE_64; else gcc1 |= AGP_I830_GCC1_GMASIZE_128; pci_write_config(sc->bdev, AGP_I830_GCC1, gcc1, 2); return (0); } static int agp_i915_set_aperture(device_t dev, u_int32_t aperture) { return (agp_generic_set_aperture(dev, aperture)); } static int agp_i810_method_set_aperture(device_t dev, u_int32_t aperture) { struct agp_i810_softc *sc; sc = device_get_softc(dev); return (sc->match->driver->set_aperture(dev, aperture)); } /** * Writes a GTT entry mapping the page at the given offset from the * beginning of the aperture to the given physical address. Setup the * caching mode according to flags. * * For gen 1, 2 and 3, GTT start is located at AGP_I810_GTT offset * from corresponding BAR start. For gen 4, offset is 512KB + * AGP_I810_GTT, for gen 5 and 6 it is 2MB + AGP_I810_GTT. * * Also, the bits of the physical page address above 4GB needs to be * placed into bits 40-32 of PTE. */ static void agp_i810_install_gtt_pte(device_t dev, u_int index, vm_offset_t physical, int flags) { uint32_t pte; pte = (u_int32_t)physical | I810_PTE_VALID; if (flags == AGP_DCACHE_MEMORY) pte |= I810_PTE_LOCAL; else if (flags == AGP_USER_CACHED_MEMORY) pte |= I830_PTE_SYSTEM_CACHED; agp_i810_write_gtt(dev, index, pte); } static void agp_i810_write_gtt(device_t dev, u_int index, uint32_t pte) { struct agp_i810_softc *sc; sc = device_get_softc(dev); bus_write_4(sc->sc_res[0], AGP_I810_GTT + index * 4, pte); CTR2(KTR_AGP_I810, "810_pte %x %x", index, pte); } static void agp_i830_install_gtt_pte(device_t dev, u_int index, vm_offset_t physical, int flags) { uint32_t pte; pte = (u_int32_t)physical | I810_PTE_VALID; if (flags == AGP_USER_CACHED_MEMORY) pte |= I830_PTE_SYSTEM_CACHED; agp_i810_write_gtt(dev, index, pte); } static void agp_i915_install_gtt_pte(device_t dev, u_int index, vm_offset_t physical, int flags) { uint32_t pte; pte = (u_int32_t)physical | I810_PTE_VALID; if (flags == AGP_USER_CACHED_MEMORY) pte |= I830_PTE_SYSTEM_CACHED; pte |= (physical & 0x0000000f00000000ull) >> 28; agp_i915_write_gtt(dev, index, pte); } static void agp_i915_write_gtt(device_t dev, u_int index, uint32_t pte) { struct agp_i810_softc *sc; sc = device_get_softc(dev); bus_write_4(sc->sc_res[1], index * 4, pte); CTR2(KTR_AGP_I810, "915_pte %x %x", index, pte); } static void agp_i965_install_gtt_pte(device_t dev, u_int index, vm_offset_t physical, int flags) { uint32_t pte; pte = (u_int32_t)physical | I810_PTE_VALID; if (flags == AGP_USER_CACHED_MEMORY) pte |= I830_PTE_SYSTEM_CACHED; pte |= (physical & 0x0000000f00000000ull) >> 28; agp_i965_write_gtt(dev, index, pte); } static void agp_i965_write_gtt(device_t dev, u_int index, uint32_t pte) { struct agp_i810_softc *sc; sc = device_get_softc(dev); bus_write_4(sc->sc_res[0], index * 4 + (512 * 1024), pte); CTR2(KTR_AGP_I810, "965_pte %x %x", index, pte); } static void agp_g4x_install_gtt_pte(device_t dev, u_int index, vm_offset_t physical, int flags) { uint32_t pte; pte = (u_int32_t)physical | I810_PTE_VALID; if (flags == AGP_USER_CACHED_MEMORY) pte |= I830_PTE_SYSTEM_CACHED; pte |= (physical & 0x0000000f00000000ull) >> 28; agp_g4x_write_gtt(dev, index, pte); } static void agp_g4x_write_gtt(device_t dev, u_int index, uint32_t pte) { struct agp_i810_softc *sc; sc = device_get_softc(dev); bus_write_4(sc->sc_res[0], index * 4 + (2 * 1024 * 1024), pte); CTR2(KTR_AGP_I810, "g4x_pte %x %x", index, pte); } static int agp_i810_bind_page(device_t dev, vm_offset_t offset, vm_offset_t physical) { struct agp_i810_softc *sc = device_get_softc(dev); u_int index; if (offset >= (sc->gatt->ag_entries << AGP_PAGE_SHIFT)) { device_printf(dev, "failed: offset is 0x%08jx, " "shift is %d, entries is %d\n", (intmax_t)offset, AGP_PAGE_SHIFT, sc->gatt->ag_entries); return (EINVAL); } index = offset >> AGP_PAGE_SHIFT; if (sc->stolen != 0 && index < sc->stolen) { device_printf(dev, "trying to bind into stolen memory\n"); return (EINVAL); } sc->match->driver->install_gtt_pte(dev, index, physical, 0); return (0); } static int agp_i810_unbind_page(device_t dev, vm_offset_t offset) { struct agp_i810_softc *sc; u_int index; sc = device_get_softc(dev); if (offset >= (sc->gatt->ag_entries << AGP_PAGE_SHIFT)) return (EINVAL); index = offset >> AGP_PAGE_SHIFT; if (sc->stolen != 0 && index < sc->stolen) { device_printf(dev, "trying to unbind from stolen memory\n"); return (EINVAL); } sc->match->driver->install_gtt_pte(dev, index, 0, 0); return (0); } static u_int32_t agp_i810_read_gtt_pte(device_t dev, u_int index) { struct agp_i810_softc *sc; u_int32_t pte; sc = device_get_softc(dev); pte = bus_read_4(sc->sc_res[0], AGP_I810_GTT + index * 4); return (pte); } static u_int32_t agp_i915_read_gtt_pte(device_t dev, u_int index) { struct agp_i810_softc *sc; u_int32_t pte; sc = device_get_softc(dev); pte = bus_read_4(sc->sc_res[1], index * 4); return (pte); } static u_int32_t agp_i965_read_gtt_pte(device_t dev, u_int index) { struct agp_i810_softc *sc; u_int32_t pte; sc = device_get_softc(dev); pte = bus_read_4(sc->sc_res[0], index * 4 + (512 * 1024)); return (pte); } static u_int32_t agp_g4x_read_gtt_pte(device_t dev, u_int index) { struct agp_i810_softc *sc; u_int32_t pte; sc = device_get_softc(dev); pte = bus_read_4(sc->sc_res[0], index * 4 + (2 * 1024 * 1024)); return (pte); } static vm_paddr_t agp_i810_read_gtt_pte_paddr(device_t dev, u_int index) { struct agp_i810_softc *sc; u_int32_t pte; vm_paddr_t res; sc = device_get_softc(dev); pte = sc->match->driver->read_gtt_pte(dev, index); res = pte & ~PAGE_MASK; return (res); } static vm_paddr_t agp_i915_read_gtt_pte_paddr(device_t dev, u_int index) { struct agp_i810_softc *sc; u_int32_t pte; vm_paddr_t res; sc = device_get_softc(dev); pte = sc->match->driver->read_gtt_pte(dev, index); res = (pte & ~PAGE_MASK) | ((pte & 0xf0) << 28); return (res); } /* * Writing via memory mapped registers already flushes all TLBs. */ static void agp_i810_flush_tlb(device_t dev) { } static int agp_i810_enable(device_t dev, u_int32_t mode) { return (0); } static struct agp_memory * agp_i810_alloc_memory(device_t dev, int type, vm_size_t size) { struct agp_i810_softc *sc; struct agp_memory *mem; vm_page_t m; sc = device_get_softc(dev); if ((size & (AGP_PAGE_SIZE - 1)) != 0 || sc->agp.as_allocated + size > sc->agp.as_maxmem) return (0); if (type == 1) { /* * Mapping local DRAM into GATT. */ if (sc->match->driver->chiptype != CHIP_I810) return (0); if (size != sc->dcache_size) return (0); } else if (type == 2) { /* * Type 2 is the contiguous physical memory type, that hands * back a physical address. This is used for cursors on i810. * Hand back as many single pages with physical as the user * wants, but only allow one larger allocation (ARGB cursor) * for simplicity. */ if (size != AGP_PAGE_SIZE) { if (sc->argb_cursor != NULL) return (0); /* Allocate memory for ARGB cursor, if we can. */ sc->argb_cursor = contigmalloc(size, M_AGP, 0, 0, ~0, PAGE_SIZE, 0); if (sc->argb_cursor == NULL) return (0); } } mem = malloc(sizeof *mem, M_AGP, M_WAITOK); mem->am_id = sc->agp.as_nextid++; mem->am_size = size; mem->am_type = type; if (type != 1 && (type != 2 || size == AGP_PAGE_SIZE)) mem->am_obj = vm_object_allocate(OBJT_DEFAULT, atop(round_page(size))); else mem->am_obj = 0; if (type == 2) { if (size == AGP_PAGE_SIZE) { /* * Allocate and wire down the page now so that we can * get its physical address. */ VM_OBJECT_WLOCK(mem->am_obj); m = vm_page_grab(mem->am_obj, 0, VM_ALLOC_NOBUSY | VM_ALLOC_WIRED | VM_ALLOC_ZERO); VM_OBJECT_WUNLOCK(mem->am_obj); mem->am_physical = VM_PAGE_TO_PHYS(m); } else { /* Our allocation is already nicely wired down for us. * Just grab the physical address. */ mem->am_physical = vtophys(sc->argb_cursor); } } else mem->am_physical = 0; mem->am_offset = 0; mem->am_is_bound = 0; TAILQ_INSERT_TAIL(&sc->agp.as_memory, mem, am_link); sc->agp.as_allocated += size; return (mem); } static int agp_i810_free_memory(device_t dev, struct agp_memory *mem) { struct agp_i810_softc *sc; vm_page_t m; if (mem->am_is_bound) return (EBUSY); sc = device_get_softc(dev); if (mem->am_type == 2) { if (mem->am_size == AGP_PAGE_SIZE) { /* * Unwire the page which we wired in alloc_memory. */ VM_OBJECT_WLOCK(mem->am_obj); m = vm_page_lookup(mem->am_obj, 0); vm_page_lock(m); vm_page_unwire(m, PQ_INACTIVE); vm_page_unlock(m); VM_OBJECT_WUNLOCK(mem->am_obj); } else { contigfree(sc->argb_cursor, mem->am_size, M_AGP); sc->argb_cursor = NULL; } } sc->agp.as_allocated -= mem->am_size; TAILQ_REMOVE(&sc->agp.as_memory, mem, am_link); if (mem->am_obj) vm_object_deallocate(mem->am_obj); free(mem, M_AGP); return (0); } static int agp_i810_bind_memory(device_t dev, struct agp_memory *mem, vm_offset_t offset) { struct agp_i810_softc *sc; vm_offset_t i; /* Do some sanity checks first. */ if ((offset & (AGP_PAGE_SIZE - 1)) != 0 || offset + mem->am_size > AGP_GET_APERTURE(dev)) { device_printf(dev, "binding memory at bad offset %#x\n", (int)offset); return (EINVAL); } sc = device_get_softc(dev); if (mem->am_type == 2 && mem->am_size != AGP_PAGE_SIZE) { mtx_lock(&sc->agp.as_lock); if (mem->am_is_bound) { mtx_unlock(&sc->agp.as_lock); return (EINVAL); } /* The memory's already wired down, just stick it in the GTT. */ for (i = 0; i < mem->am_size; i += AGP_PAGE_SIZE) { sc->match->driver->install_gtt_pte(dev, (offset + i) >> AGP_PAGE_SHIFT, mem->am_physical + i, 0); } mem->am_offset = offset; mem->am_is_bound = 1; mtx_unlock(&sc->agp.as_lock); return (0); } if (mem->am_type != 1) return (agp_generic_bind_memory(dev, mem, offset)); /* * Mapping local DRAM into GATT. */ if (sc->match->driver->chiptype != CHIP_I810) return (EINVAL); for (i = 0; i < mem->am_size; i += AGP_PAGE_SIZE) bus_write_4(sc->sc_res[0], AGP_I810_GTT + (i >> AGP_PAGE_SHIFT) * 4, i | 3); return (0); } static int agp_i810_unbind_memory(device_t dev, struct agp_memory *mem) { struct agp_i810_softc *sc; vm_offset_t i; sc = device_get_softc(dev); if (mem->am_type == 2 && mem->am_size != AGP_PAGE_SIZE) { mtx_lock(&sc->agp.as_lock); if (!mem->am_is_bound) { mtx_unlock(&sc->agp.as_lock); return (EINVAL); } for (i = 0; i < mem->am_size; i += AGP_PAGE_SIZE) { sc->match->driver->install_gtt_pte(dev, (mem->am_offset + i) >> AGP_PAGE_SHIFT, 0, 0); } mem->am_is_bound = 0; mtx_unlock(&sc->agp.as_lock); return (0); } if (mem->am_type != 1) return (agp_generic_unbind_memory(dev, mem)); if (sc->match->driver->chiptype != CHIP_I810) return (EINVAL); for (i = 0; i < mem->am_size; i += AGP_PAGE_SIZE) { sc->match->driver->install_gtt_pte(dev, i >> AGP_PAGE_SHIFT, 0, 0); } return (0); } static device_method_t agp_i810_methods[] = { /* Device interface */ DEVMETHOD(device_identify, agp_i810_identify), DEVMETHOD(device_probe, agp_i810_probe), DEVMETHOD(device_attach, agp_i810_attach), DEVMETHOD(device_detach, agp_i810_detach), DEVMETHOD(device_suspend, bus_generic_suspend), DEVMETHOD(device_resume, agp_i810_resume), /* AGP interface */ DEVMETHOD(agp_get_aperture, agp_generic_get_aperture), DEVMETHOD(agp_set_aperture, agp_i810_method_set_aperture), DEVMETHOD(agp_bind_page, agp_i810_bind_page), DEVMETHOD(agp_unbind_page, agp_i810_unbind_page), DEVMETHOD(agp_flush_tlb, agp_i810_flush_tlb), DEVMETHOD(agp_enable, agp_i810_enable), DEVMETHOD(agp_alloc_memory, agp_i810_alloc_memory), DEVMETHOD(agp_free_memory, agp_i810_free_memory), DEVMETHOD(agp_bind_memory, agp_i810_bind_memory), DEVMETHOD(agp_unbind_memory, agp_i810_unbind_memory), DEVMETHOD(agp_chipset_flush, agp_intel_gtt_chipset_flush), { 0, 0 } }; static driver_t agp_i810_driver = { "agp", agp_i810_methods, sizeof(struct agp_i810_softc), }; static devclass_t agp_devclass; DRIVER_MODULE(agp_i810, vgapci, agp_i810_driver, agp_devclass, 0, 0); MODULE_DEPEND(agp_i810, agp, 1, 1, 1); MODULE_DEPEND(agp_i810, pci, 1, 1, 1); -extern vm_page_t bogus_page; - void agp_intel_gtt_clear_range(device_t dev, u_int first_entry, u_int num_entries) { struct agp_i810_softc *sc; u_int i; sc = device_get_softc(dev); for (i = 0; i < num_entries; i++) sc->match->driver->install_gtt_pte(dev, first_entry + i, VM_PAGE_TO_PHYS(bogus_page), 0); sc->match->driver->read_gtt_pte(dev, first_entry + num_entries - 1); } void agp_intel_gtt_insert_pages(device_t dev, u_int first_entry, u_int num_entries, vm_page_t *pages, u_int flags) { struct agp_i810_softc *sc; u_int i; sc = device_get_softc(dev); for (i = 0; i < num_entries; i++) { MPASS(pages[i]->valid == VM_PAGE_BITS_ALL); MPASS(pages[i]->wire_count > 0); sc->match->driver->install_gtt_pte(dev, first_entry + i, VM_PAGE_TO_PHYS(pages[i]), flags); } sc->match->driver->read_gtt_pte(dev, first_entry + num_entries - 1); } struct intel_gtt agp_intel_gtt_get(device_t dev) { struct agp_i810_softc *sc; struct intel_gtt res; sc = device_get_softc(dev); res.stolen_size = sc->stolen_size; res.gtt_total_entries = sc->gtt_total_entries; res.gtt_mappable_entries = sc->gtt_mappable_entries; res.do_idle_maps = 0; res.scratch_page_dma = VM_PAGE_TO_PHYS(bogus_page); if (sc->agp.as_aperture != NULL) res.gma_bus_addr = rman_get_start(sc->agp.as_aperture); else res.gma_bus_addr = 0; return (res); } static int agp_i810_chipset_flush_setup(device_t dev) { return (0); } static void agp_i810_chipset_flush_teardown(device_t dev) { /* Nothing to do. */ } static void agp_i810_chipset_flush(device_t dev) { /* Nothing to do. */ } static void agp_i830_chipset_flush(device_t dev) { struct agp_i810_softc *sc; uint32_t hic; int i; sc = device_get_softc(dev); pmap_invalidate_cache(); hic = bus_read_4(sc->sc_res[0], AGP_I830_HIC); bus_write_4(sc->sc_res[0], AGP_I830_HIC, hic | (1U << 31)); for (i = 0; i < 20000 /* 1 sec */; i++) { hic = bus_read_4(sc->sc_res[0], AGP_I830_HIC); if ((hic & (1U << 31)) == 0) break; DELAY(50); } } static int agp_i915_chipset_flush_alloc_page(device_t dev, uint64_t start, uint64_t end) { struct agp_i810_softc *sc; device_t vga; sc = device_get_softc(dev); vga = device_get_parent(dev); sc->sc_flush_page_rid = 100; sc->sc_flush_page_res = BUS_ALLOC_RESOURCE(device_get_parent(vga), dev, SYS_RES_MEMORY, &sc->sc_flush_page_rid, start, end, PAGE_SIZE, RF_ACTIVE); if (sc->sc_flush_page_res == NULL) { device_printf(dev, "Failed to allocate flush page at 0x%jx\n", (uintmax_t)start); return (EINVAL); } sc->sc_flush_page_vaddr = rman_get_virtual(sc->sc_flush_page_res); if (bootverbose) { device_printf(dev, "Allocated flush page phys 0x%jx virt %p\n", (uintmax_t)rman_get_start(sc->sc_flush_page_res), sc->sc_flush_page_vaddr); } return (0); } static void agp_i915_chipset_flush_free_page(device_t dev) { struct agp_i810_softc *sc; device_t vga; sc = device_get_softc(dev); vga = device_get_parent(dev); if (sc->sc_flush_page_res == NULL) return; BUS_DEACTIVATE_RESOURCE(device_get_parent(vga), dev, SYS_RES_MEMORY, sc->sc_flush_page_rid, sc->sc_flush_page_res); BUS_RELEASE_RESOURCE(device_get_parent(vga), dev, SYS_RES_MEMORY, sc->sc_flush_page_rid, sc->sc_flush_page_res); } static int agp_i915_chipset_flush_setup(device_t dev) { struct agp_i810_softc *sc; uint32_t temp; int error; sc = device_get_softc(dev); temp = pci_read_config(sc->bdev, AGP_I915_IFPADDR, 4); if ((temp & 1) != 0) { temp &= ~1; if (bootverbose) device_printf(dev, "Found already configured flush page at 0x%jx\n", (uintmax_t)temp); sc->sc_bios_allocated_flush_page = 1; /* * In the case BIOS initialized the flush pointer (?) * register, expect that BIOS also set up the resource * for the page. */ error = agp_i915_chipset_flush_alloc_page(dev, temp, temp + PAGE_SIZE - 1); if (error != 0) return (error); } else { sc->sc_bios_allocated_flush_page = 0; error = agp_i915_chipset_flush_alloc_page(dev, 0, 0xffffffff); if (error != 0) return (error); temp = rman_get_start(sc->sc_flush_page_res); pci_write_config(sc->bdev, AGP_I915_IFPADDR, temp | 1, 4); } return (0); } static void agp_i915_chipset_flush_teardown(device_t dev) { struct agp_i810_softc *sc; uint32_t temp; sc = device_get_softc(dev); if (sc->sc_flush_page_res == NULL) return; if (!sc->sc_bios_allocated_flush_page) { temp = pci_read_config(sc->bdev, AGP_I915_IFPADDR, 4); temp &= ~1; pci_write_config(sc->bdev, AGP_I915_IFPADDR, temp, 4); } agp_i915_chipset_flush_free_page(dev); } static int agp_i965_chipset_flush_setup(device_t dev) { struct agp_i810_softc *sc; uint64_t temp; uint32_t temp_hi, temp_lo; int error; sc = device_get_softc(dev); temp_hi = pci_read_config(sc->bdev, AGP_I965_IFPADDR + 4, 4); temp_lo = pci_read_config(sc->bdev, AGP_I965_IFPADDR, 4); if ((temp_lo & 1) != 0) { temp = ((uint64_t)temp_hi << 32) | (temp_lo & ~1); if (bootverbose) device_printf(dev, "Found already configured flush page at 0x%jx\n", (uintmax_t)temp); sc->sc_bios_allocated_flush_page = 1; /* * In the case BIOS initialized the flush pointer (?) * register, expect that BIOS also set up the resource * for the page. */ error = agp_i915_chipset_flush_alloc_page(dev, temp, temp + PAGE_SIZE - 1); if (error != 0) return (error); } else { sc->sc_bios_allocated_flush_page = 0; error = agp_i915_chipset_flush_alloc_page(dev, 0, ~0); if (error != 0) return (error); temp = rman_get_start(sc->sc_flush_page_res); pci_write_config(sc->bdev, AGP_I965_IFPADDR + 4, (temp >> 32) & UINT32_MAX, 4); pci_write_config(sc->bdev, AGP_I965_IFPADDR, (temp & UINT32_MAX) | 1, 4); } return (0); } static void agp_i965_chipset_flush_teardown(device_t dev) { struct agp_i810_softc *sc; uint32_t temp_lo; sc = device_get_softc(dev); if (sc->sc_flush_page_res == NULL) return; if (!sc->sc_bios_allocated_flush_page) { temp_lo = pci_read_config(sc->bdev, AGP_I965_IFPADDR, 4); temp_lo &= ~1; pci_write_config(sc->bdev, AGP_I965_IFPADDR, temp_lo, 4); } agp_i915_chipset_flush_free_page(dev); } static void agp_i915_chipset_flush(device_t dev) { struct agp_i810_softc *sc; sc = device_get_softc(dev); *(uint32_t *)sc->sc_flush_page_vaddr = 1; } int agp_intel_gtt_chipset_flush(device_t dev) { struct agp_i810_softc *sc; sc = device_get_softc(dev); sc->match->driver->chipset_flush(dev); return (0); } void agp_intel_gtt_unmap_memory(device_t dev, struct sglist *sg_list) { } int agp_intel_gtt_map_memory(device_t dev, vm_page_t *pages, u_int num_entries, struct sglist **sg_list) { struct agp_i810_softc *sc; struct sglist *sg; int i; #if 0 int error; bus_dma_tag_t dmat; #endif if (*sg_list != NULL) return (0); sc = device_get_softc(dev); sg = sglist_alloc(num_entries, M_WAITOK /* XXXKIB */); for (i = 0; i < num_entries; i++) { sg->sg_segs[i].ss_paddr = VM_PAGE_TO_PHYS(pages[i]); sg->sg_segs[i].ss_len = PAGE_SIZE; } #if 0 error = bus_dma_tag_create(bus_get_dma_tag(dev), 1 /* alignment */, 0 /* boundary */, 1ULL << sc->match->busdma_addr_mask_sz /* lowaddr */, BUS_SPACE_MAXADDR /* highaddr */, NULL /* filtfunc */, NULL /* filtfuncarg */, BUS_SPACE_MAXADDR /* maxsize */, BUS_SPACE_UNRESTRICTED /* nsegments */, BUS_SPACE_MAXADDR /* maxsegsz */, 0 /* flags */, NULL /* lockfunc */, NULL /* lockfuncarg */, &dmat); if (error != 0) { sglist_free(sg); return (error); } /* XXXKIB */ #endif *sg_list = sg; return (0); } void agp_intel_gtt_insert_sg_entries(device_t dev, struct sglist *sg_list, u_int first_entry, u_int flags) { struct agp_i810_softc *sc; vm_paddr_t spaddr; size_t slen; u_int i, j; sc = device_get_softc(dev); for (i = j = 0; j < sg_list->sg_nseg; j++) { spaddr = sg_list->sg_segs[i].ss_paddr; slen = sg_list->sg_segs[i].ss_len; for (; slen > 0; i++) { sc->match->driver->install_gtt_pte(dev, first_entry + i, spaddr, flags); spaddr += AGP_PAGE_SIZE; slen -= AGP_PAGE_SIZE; } } sc->match->driver->read_gtt_pte(dev, first_entry + i - 1); } void intel_gtt_clear_range(u_int first_entry, u_int num_entries) { agp_intel_gtt_clear_range(intel_agp, first_entry, num_entries); } void intel_gtt_insert_pages(u_int first_entry, u_int num_entries, vm_page_t *pages, u_int flags) { agp_intel_gtt_insert_pages(intel_agp, first_entry, num_entries, pages, flags); } struct intel_gtt * intel_gtt_get(void) { intel_private.base = agp_intel_gtt_get(intel_agp); return (&intel_private.base); } int intel_gtt_chipset_flush(void) { return (agp_intel_gtt_chipset_flush(intel_agp)); } void intel_gtt_unmap_memory(struct sglist *sg_list) { agp_intel_gtt_unmap_memory(intel_agp, sg_list); } int intel_gtt_map_memory(vm_page_t *pages, u_int num_entries, struct sglist **sg_list) { return (agp_intel_gtt_map_memory(intel_agp, pages, num_entries, sg_list)); } void intel_gtt_insert_sg_entries(struct sglist *sg_list, u_int first_entry, u_int flags) { agp_intel_gtt_insert_sg_entries(intel_agp, sg_list, first_entry, flags); } device_t intel_gtt_get_bridge_device(void) { struct agp_i810_softc *sc; sc = device_get_softc(intel_agp); return (sc->bdev); } vm_paddr_t intel_gtt_read_pte_paddr(u_int entry) { struct agp_i810_softc *sc; sc = device_get_softc(intel_agp); return (sc->match->driver->read_gtt_pte_paddr(intel_agp, entry)); } u_int32_t intel_gtt_read_pte(u_int entry) { struct agp_i810_softc *sc; sc = device_get_softc(intel_agp); return (sc->match->driver->read_gtt_pte(intel_agp, entry)); } void intel_gtt_write(u_int entry, uint32_t val) { struct agp_i810_softc *sc; sc = device_get_softc(intel_agp); return (sc->match->driver->write_gtt(intel_agp, entry, val)); } Index: projects/clang400-import/sys/dev/aha/aha.c =================================================================== --- projects/clang400-import/sys/dev/aha/aha.c (revision 311459) +++ projects/clang400-import/sys/dev/aha/aha.c (revision 311460) @@ -1,1824 +1,1824 @@ /* * Generic register and struct definitions for the Adaptech 154x/164x * SCSI host adapters. Product specific probe and attach routines can * be found in: * aha 1542A/1542B/1542C/1542CF/1542CP aha_isa.c * aha 1640 aha_mca.c */ /*- * Copyright (c) 1998 M. Warner Losh. * 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. * * Derived from bt.c written by: * * Copyright (c) 1998 Justin T. Gibbs. * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions, and the following disclaimer, * without modification, immediately at the beginning of the file. * 2. 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 AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE FOR * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #define PRVERB(x) do { if (bootverbose) device_printf x; } while (0) /* Macro to determine that a rev is potentially a new valid one * so that the driver doesn't keep breaking on new revs as it * did for the CF and CP. */ #define PROBABLY_NEW_BOARD(REV) (REV > 0x43 && REV < 0x56) /* MailBox Management functions */ static __inline void ahanextinbox(struct aha_softc *aha); static __inline void ahanextoutbox(struct aha_softc *aha); #define aha_name(aha) device_get_nameunit(aha->dev) static __inline void ahanextinbox(struct aha_softc *aha) { if (aha->cur_inbox == aha->last_inbox) aha->cur_inbox = aha->in_boxes; else aha->cur_inbox++; } static __inline void ahanextoutbox(struct aha_softc *aha) { if (aha->cur_outbox == aha->last_outbox) aha->cur_outbox = aha->out_boxes; else aha->cur_outbox++; } #define ahautoa24(u,s3) \ (s3)[0] = ((u) >> 16) & 0xff; \ (s3)[1] = ((u) >> 8) & 0xff; \ (s3)[2] = (u) & 0xff; #define aha_a24tou(s3) \ (((s3)[0] << 16) | ((s3)[1] << 8) | (s3)[2]) /* CCB Management functions */ static __inline uint32_t ahaccbvtop(struct aha_softc *aha, struct aha_ccb *accb); static __inline struct aha_ccb* ahaccbptov(struct aha_softc *aha, uint32_t ccb_addr); static __inline uint32_t ahaccbvtop(struct aha_softc *aha, struct aha_ccb *accb) { return (aha->aha_ccb_physbase + (uint32_t)((caddr_t)accb - (caddr_t)aha->aha_ccb_array)); } static __inline struct aha_ccb * ahaccbptov(struct aha_softc *aha, uint32_t ccb_addr) { return (aha->aha_ccb_array + + ((struct aha_ccb*)(uintptr_t)ccb_addr - (struct aha_ccb*)(uintptr_t)aha->aha_ccb_physbase)); } static struct aha_ccb* ahagetccb(struct aha_softc *aha); static __inline void ahafreeccb(struct aha_softc *aha, struct aha_ccb *accb); static void ahaallocccbs(struct aha_softc *aha); static bus_dmamap_callback_t ahaexecuteccb; static void ahadone(struct aha_softc *aha, struct aha_ccb *accb, aha_mbi_comp_code_t comp_code); static void aha_intr_locked(struct aha_softc *aha); /* Host adapter command functions */ static int ahareset(struct aha_softc* aha, int hard_reset); /* Initialization functions */ static int ahainitmboxes(struct aha_softc *aha); static bus_dmamap_callback_t ahamapmboxes; static bus_dmamap_callback_t ahamapccbs; static bus_dmamap_callback_t ahamapsgs; /* Transfer Negotiation Functions */ static void ahafetchtransinfo(struct aha_softc *aha, struct ccb_trans_settings *cts); /* CAM SIM entry points */ #define ccb_accb_ptr spriv_ptr0 #define ccb_aha_ptr spriv_ptr1 static void ahaaction(struct cam_sim *sim, union ccb *ccb); static void ahapoll(struct cam_sim *sim); /* Our timeout handler */ static void ahatimeout(void *arg); /* Exported functions */ void aha_alloc(struct aha_softc *aha) { SLIST_INIT(&aha->free_aha_ccbs); LIST_INIT(&aha->pending_ccbs); SLIST_INIT(&aha->sg_maps); aha->ccb_sg_opcode = INITIATOR_SG_CCB_WRESID; aha->ccb_ccb_opcode = INITIATOR_CCB_WRESID; mtx_init(&aha->lock, "aha", NULL, MTX_DEF); } void aha_free(struct aha_softc *aha) { switch (aha->init_level) { default: case 8: { struct sg_map_node *sg_map; while ((sg_map = SLIST_FIRST(&aha->sg_maps))!= NULL) { SLIST_REMOVE_HEAD(&aha->sg_maps, links); bus_dmamap_unload(aha->sg_dmat, sg_map->sg_dmamap); bus_dmamem_free(aha->sg_dmat, sg_map->sg_vaddr, sg_map->sg_dmamap); free(sg_map, M_DEVBUF); } bus_dma_tag_destroy(aha->sg_dmat); } case 7: bus_dmamap_unload(aha->ccb_dmat, aha->ccb_dmamap); case 6: bus_dmamem_free(aha->ccb_dmat, aha->aha_ccb_array, aha->ccb_dmamap); case 5: bus_dma_tag_destroy(aha->ccb_dmat); case 4: bus_dmamap_unload(aha->mailbox_dmat, aha->mailbox_dmamap); case 3: bus_dmamem_free(aha->mailbox_dmat, aha->in_boxes, aha->mailbox_dmamap); case 2: bus_dma_tag_destroy(aha->buffer_dmat); case 1: bus_dma_tag_destroy(aha->mailbox_dmat); case 0: break; } mtx_destroy(&aha->lock); } /* * Probe the adapter and verify that the card is an Adaptec. */ int aha_probe(struct aha_softc* aha) { u_int status; u_int intstat; int error; board_id_data_t board_id; /* * See if the three I/O ports look reasonable. * Touch the minimal number of registers in the * failure case. */ status = aha_inb(aha, STATUS_REG); if ((status == 0) || (status & (DIAG_ACTIVE|CMD_REG_BUSY | STATUS_REG_RSVD)) != 0) { PRVERB((aha->dev, "status reg test failed %x\n", status)); return (ENXIO); } intstat = aha_inb(aha, INTSTAT_REG); if ((intstat & INTSTAT_REG_RSVD) != 0) { PRVERB((aha->dev, "Failed Intstat Reg Test\n")); return (ENXIO); } /* * Looking good so far. Final test is to reset the * adapter and fetch the board ID and ensure we aren't * looking at a BusLogic. */ if ((error = ahareset(aha, /*hard_reset*/TRUE)) != 0) { PRVERB((aha->dev, "Failed Reset\n")); return (ENXIO); } /* * Get the board ID. We use this to see if we're dealing with * a buslogic card or an aha card (or clone). */ error = aha_cmd(aha, AOP_INQUIRE_BOARD_ID, NULL, /*parmlen*/0, (uint8_t*)&board_id, sizeof(board_id), DEFAULT_CMD_TIMEOUT); if (error != 0) { PRVERB((aha->dev, "INQUIRE failed %x\n", error)); return (ENXIO); } aha->fw_major = board_id.firmware_rev_major; aha->fw_minor = board_id.firmware_rev_minor; aha->boardid = board_id.board_type; /* * The Buslogic cards have an id of either 0x41 or 0x42. So * if those come up in the probe, we test the geometry register * of the board. Adaptec boards that are this old will not have * this register, and return 0xff, while buslogic cards will return * something different. * * It appears that for reasons unknow, for the for the * aha-1542B cards, we need to wait a little bit before trying * to read the geometry register. I picked 10ms since we have * reports that a for loop to 1000 did the trick, and this * errs on the side of conservatism. Besides, no one will * notice a 10mS delay here, even the 1542B card users :-) * * Some compatible cards return 0 here. Some cards also * seem to return 0x7f. * * XXX I'm not sure how this will impact other cloned cards * * This really should be replaced with the esetup command, since * that appears to be more reliable. This becomes more and more * true over time as we discover more cards that don't read the * geometry register consistently. */ if (aha->boardid <= 0x42) { /* Wait 10ms before reading */ DELAY(10000); status = aha_inb(aha, GEOMETRY_REG); if (status != 0xff && status != 0x00 && status != 0x7f) { PRVERB((aha->dev, "Geometry Register test failed %#x\n", status)); return (ENXIO); } } return (0); } /* * Pull the boards setup information and record it in our softc. */ int aha_fetch_adapter_info(struct aha_softc *aha) { setup_data_t setup_info; config_data_t config_data; uint8_t length_param; int error; struct aha_extbios extbios; switch (aha->boardid) { case BOARD_1540_16HEAD_BIOS: snprintf(aha->model, sizeof(aha->model), "1540 16 head BIOS"); break; case BOARD_1540_64HEAD_BIOS: snprintf(aha->model, sizeof(aha->model), "1540 64 head BIOS"); break; case BOARD_1542: snprintf(aha->model, sizeof(aha->model), "1540/1542 64 head BIOS"); break; case BOARD_1640: snprintf(aha->model, sizeof(aha->model), "1640"); break; case BOARD_1740: snprintf(aha->model, sizeof(aha->model), "1740A/1742A/1744"); break; case BOARD_1542C: snprintf(aha->model, sizeof(aha->model), "1542C"); break; case BOARD_1542CF: snprintf(aha->model, sizeof(aha->model), "1542CF"); break; case BOARD_1542CP: snprintf(aha->model, sizeof(aha->model), "1542CP"); break; default: snprintf(aha->model, sizeof(aha->model), "Unknown"); break; } /* * If we are a new type of 1542 board (anything newer than a 1542C) * then disable the extended bios so that the * mailbox interface is unlocked. * This is also true for the 1542B Version 3.20. First Adaptec * board that supports >1Gb drives. * No need to check the extended bios flags as some of the * extensions that cause us problems are not flagged in that byte. */ if (PROBABLY_NEW_BOARD(aha->boardid) || (aha->boardid == 0x41 && aha->fw_major == 0x31 && aha->fw_minor >= 0x34)) { error = aha_cmd(aha, AOP_RETURN_EXT_BIOS_INFO, NULL, /*paramlen*/0, (u_char *)&extbios, sizeof(extbios), DEFAULT_CMD_TIMEOUT); if (error != 0) { device_printf(aha->dev, "AOP_RETURN_EXT_BIOS_INFO - Failed."); return (error); } error = aha_cmd(aha, AOP_MBOX_IF_ENABLE, (uint8_t *)&extbios, /*paramlen*/2, NULL, 0, DEFAULT_CMD_TIMEOUT); if (error != 0) { device_printf(aha->dev, "AOP_MBOX_IF_ENABLE - Failed."); return (error); } } if (aha->boardid < 0x41) device_printf(aha->dev, "Warning: aha-1542A won't work.\n"); aha->max_sg = 17; /* Need >= 17 to do 64k I/O */ aha->diff_bus = 0; aha->extended_lun = 0; aha->extended_trans = 0; aha->max_ccbs = 16; /* Determine Sync/Wide/Disc settings */ length_param = sizeof(setup_info); error = aha_cmd(aha, AOP_INQUIRE_SETUP_INFO, &length_param, /*paramlen*/1, (uint8_t*)&setup_info, sizeof(setup_info), DEFAULT_CMD_TIMEOUT); if (error != 0) { device_printf(aha->dev, "aha_fetch_adapter_info - Failed " "Get Setup Info\n"); return (error); } if (setup_info.initiate_sync != 0) { aha->sync_permitted = ALL_TARGETS; } aha->disc_permitted = ALL_TARGETS; /* We need as many mailboxes as we can have ccbs */ aha->num_boxes = aha->max_ccbs; /* Determine our SCSI ID */ error = aha_cmd(aha, AOP_INQUIRE_CONFIG, NULL, /*parmlen*/0, (uint8_t*)&config_data, sizeof(config_data), DEFAULT_CMD_TIMEOUT); if (error != 0) { device_printf(aha->dev, "aha_fetch_adapter_info - Failed Get Config\n"); return (error); } aha->scsi_id = config_data.scsi_id; return (0); } /* * Start the board, ready for normal operation */ int aha_init(struct aha_softc* aha) { /* Announce the Adapter */ device_printf(aha->dev, "AHA-%s FW Rev. %c.%c (ID=%x) ", aha->model, aha->fw_major, aha->fw_minor, aha->boardid); if (aha->diff_bus != 0) printf("Diff "); printf("SCSI Host Adapter, SCSI ID %d, %d CCBs\n", aha->scsi_id, aha->max_ccbs); /* * Create our DMA tags. These tags define the kinds of device * accessible memory allocations and memory mappings we will * need to perform during normal operation. * * Unless we need to further restrict the allocation, we rely * on the restrictions of the parent dmat, hence the common * use of MAXADDR and MAXSIZE. */ /* DMA tag for mapping buffers into device visible space. */ if (bus_dma_tag_create( /* parent */ aha->parent_dmat, /* alignment */ 1, /* boundary */ 0, /* lowaddr */ BUS_SPACE_MAXADDR, /* highaddr */ BUS_SPACE_MAXADDR, /* filter */ NULL, /* filterarg */ NULL, /* maxsize */ DFLTPHYS, /* nsegments */ AHA_NSEG, /* maxsegsz */ BUS_SPACE_MAXSIZE_24BIT, /* flags */ BUS_DMA_ALLOCNOW, /* lockfunc */ busdma_lock_mutex, /* lockarg */ &aha->lock, &aha->buffer_dmat) != 0) { goto error_exit; } aha->init_level++; /* DMA tag for our mailboxes */ if (bus_dma_tag_create( /* parent */ aha->parent_dmat, /* alignment */ 1, /* boundary */ 0, /* lowaddr */ BUS_SPACE_MAXADDR, /* highaddr */ BUS_SPACE_MAXADDR, /* filter */ NULL, /* filterarg */ NULL, /* maxsize */ aha->num_boxes * (sizeof(aha_mbox_in_t) + sizeof(aha_mbox_out_t)), /* nsegments */ 1, /* maxsegsz */ BUS_SPACE_MAXSIZE_24BIT, /* flags */ 0, /* lockfunc */ NULL, /* lockarg */ NULL, &aha->mailbox_dmat) != 0) { goto error_exit; } aha->init_level++; /* Allocation for our mailboxes */ if (bus_dmamem_alloc(aha->mailbox_dmat, (void **)&aha->out_boxes, BUS_DMA_NOWAIT, &aha->mailbox_dmamap) != 0) goto error_exit; aha->init_level++; /* And permanently map them */ bus_dmamap_load(aha->mailbox_dmat, aha->mailbox_dmamap, aha->out_boxes, aha->num_boxes * (sizeof(aha_mbox_in_t) + sizeof(aha_mbox_out_t)), ahamapmboxes, aha, /*flags*/0); aha->init_level++; aha->in_boxes = (aha_mbox_in_t *)&aha->out_boxes[aha->num_boxes]; ahainitmboxes(aha); /* DMA tag for our ccb structures */ if (bus_dma_tag_create( /* parent */ aha->parent_dmat, /* alignment */ 1, /* boundary */ 0, /* lowaddr */ BUS_SPACE_MAXADDR, /* highaddr */ BUS_SPACE_MAXADDR, /* filter */ NULL, /* filterarg */ NULL, /* maxsize */ aha->max_ccbs * sizeof(struct aha_ccb), /* nsegments */ 1, /* maxsegsz */ BUS_SPACE_MAXSIZE_24BIT, /* flags */ 0, /* lockfunc */ NULL, /* lockarg */ NULL, &aha->ccb_dmat) != 0) { goto error_exit; } aha->init_level++; /* Allocation for our ccbs */ if (bus_dmamem_alloc(aha->ccb_dmat, (void **)&aha->aha_ccb_array, BUS_DMA_NOWAIT, &aha->ccb_dmamap) != 0) goto error_exit; aha->init_level++; /* And permanently map them */ bus_dmamap_load(aha->ccb_dmat, aha->ccb_dmamap, aha->aha_ccb_array, aha->max_ccbs * sizeof(struct aha_ccb), ahamapccbs, aha, /*flags*/0); aha->init_level++; /* DMA tag for our S/G structures. We allocate in page sized chunks */ if (bus_dma_tag_create( /* parent */ aha->parent_dmat, /* alignment */ 1, /* boundary */ 0, /* lowaddr */ BUS_SPACE_MAXADDR, /* highaddr */ BUS_SPACE_MAXADDR, /* filter */ NULL, /* filterarg */ NULL, /* maxsize */ PAGE_SIZE, /* nsegments */ 1, /* maxsegsz */ BUS_SPACE_MAXSIZE_24BIT, /* flags */ 0, /* lockfunc */ NULL, /* lockarg */ NULL, &aha->sg_dmat) != 0) goto error_exit; aha->init_level++; /* Perform initial CCB allocation */ bzero(aha->aha_ccb_array, aha->max_ccbs * sizeof(struct aha_ccb)); ahaallocccbs(aha); if (aha->num_ccbs == 0) { device_printf(aha->dev, "aha_init - Unable to allocate initial ccbs\n"); goto error_exit; } /* * Note that we are going and return (to probe) */ return (0); error_exit: return (ENXIO); } int aha_attach(struct aha_softc *aha) { int tagged_dev_openings; struct cam_devq *devq; /* * We don't do tagged queueing, since the aha cards don't * support it. */ tagged_dev_openings = 0; /* * Create the device queue for our SIM. */ devq = cam_simq_alloc(aha->max_ccbs - 1); if (devq == NULL) return (ENOMEM); /* * Construct our SIM entry */ aha->sim = cam_sim_alloc(ahaaction, ahapoll, "aha", aha, device_get_unit(aha->dev), &aha->lock, 2, tagged_dev_openings, devq); if (aha->sim == NULL) { cam_simq_free(devq); return (ENOMEM); } mtx_lock(&aha->lock); if (xpt_bus_register(aha->sim, aha->dev, 0) != CAM_SUCCESS) { cam_sim_free(aha->sim, /*free_devq*/TRUE); mtx_unlock(&aha->lock); return (ENXIO); } if (xpt_create_path(&aha->path, /*periph*/NULL, cam_sim_path(aha->sim), CAM_TARGET_WILDCARD, CAM_LUN_WILDCARD) != CAM_REQ_CMP) { xpt_bus_deregister(cam_sim_path(aha->sim)); cam_sim_free(aha->sim, /*free_devq*/TRUE); mtx_unlock(&aha->lock); return (ENXIO); } mtx_unlock(&aha->lock); return (0); } static void ahaallocccbs(struct aha_softc *aha) { struct aha_ccb *next_ccb; struct sg_map_node *sg_map; bus_addr_t physaddr; aha_sg_t *segs; int newcount; int i; next_ccb = &aha->aha_ccb_array[aha->num_ccbs]; sg_map = malloc(sizeof(*sg_map), M_DEVBUF, M_NOWAIT); if (sg_map == NULL) return; /* Allocate S/G space for the next batch of CCBS */ if (bus_dmamem_alloc(aha->sg_dmat, (void **)&sg_map->sg_vaddr, BUS_DMA_NOWAIT, &sg_map->sg_dmamap) != 0) { free(sg_map, M_DEVBUF); return; } SLIST_INSERT_HEAD(&aha->sg_maps, sg_map, links); bus_dmamap_load(aha->sg_dmat, sg_map->sg_dmamap, sg_map->sg_vaddr, PAGE_SIZE, ahamapsgs, aha, /*flags*/0); segs = sg_map->sg_vaddr; physaddr = sg_map->sg_physaddr; newcount = (PAGE_SIZE / (AHA_NSEG * sizeof(aha_sg_t))); for (i = 0; aha->num_ccbs < aha->max_ccbs && i < newcount; i++) { int error; next_ccb->sg_list = segs; next_ccb->sg_list_phys = physaddr; next_ccb->flags = ACCB_FREE; callout_init_mtx(&next_ccb->timer, &aha->lock, 0); error = bus_dmamap_create(aha->buffer_dmat, /*flags*/0, &next_ccb->dmamap); if (error != 0) break; SLIST_INSERT_HEAD(&aha->free_aha_ccbs, next_ccb, links); segs += AHA_NSEG; physaddr += (AHA_NSEG * sizeof(aha_sg_t)); next_ccb++; aha->num_ccbs++; } /* Reserve a CCB for error recovery */ if (aha->recovery_accb == NULL) { aha->recovery_accb = SLIST_FIRST(&aha->free_aha_ccbs); SLIST_REMOVE_HEAD(&aha->free_aha_ccbs, links); } } static __inline void ahafreeccb(struct aha_softc *aha, struct aha_ccb *accb) { if (!dumping) mtx_assert(&aha->lock, MA_OWNED); if ((accb->flags & ACCB_ACTIVE) != 0) LIST_REMOVE(&accb->ccb->ccb_h, sim_links.le); if (aha->resource_shortage != 0 && (accb->ccb->ccb_h.status & CAM_RELEASE_SIMQ) == 0) { accb->ccb->ccb_h.status |= CAM_RELEASE_SIMQ; aha->resource_shortage = FALSE; } accb->flags = ACCB_FREE; SLIST_INSERT_HEAD(&aha->free_aha_ccbs, accb, links); aha->active_ccbs--; } static struct aha_ccb* ahagetccb(struct aha_softc *aha) { struct aha_ccb* accb; if (!dumping) mtx_assert(&aha->lock, MA_OWNED); if ((accb = SLIST_FIRST(&aha->free_aha_ccbs)) != NULL) { SLIST_REMOVE_HEAD(&aha->free_aha_ccbs, links); aha->active_ccbs++; } else if (aha->num_ccbs < aha->max_ccbs) { ahaallocccbs(aha); accb = SLIST_FIRST(&aha->free_aha_ccbs); if (accb == NULL) device_printf(aha->dev, "Can't malloc ACCB\n"); else { SLIST_REMOVE_HEAD(&aha->free_aha_ccbs, links); aha->active_ccbs++; } } return (accb); } static void ahaaction(struct cam_sim *sim, union ccb *ccb) { struct aha_softc *aha; CAM_DEBUG(ccb->ccb_h.path, CAM_DEBUG_TRACE, ("ahaaction\n")); aha = (struct aha_softc *)cam_sim_softc(sim); mtx_assert(&aha->lock, MA_OWNED); switch (ccb->ccb_h.func_code) { /* Common cases first */ case XPT_SCSI_IO: /* Execute the requested I/O operation */ case XPT_RESET_DEV: /* Bus Device Reset the specified SCSI device */ { struct aha_ccb *accb; struct aha_hccb *hccb; /* * Get an accb to use. */ if ((accb = ahagetccb(aha)) == NULL) { aha->resource_shortage = TRUE; xpt_freeze_simq(aha->sim, /*count*/1); ccb->ccb_h.status = CAM_REQUEUE_REQ; xpt_done(ccb); return; } hccb = &accb->hccb; /* * So we can find the ACCB when an abort is requested */ accb->ccb = ccb; ccb->ccb_h.ccb_accb_ptr = accb; ccb->ccb_h.ccb_aha_ptr = aha; /* * Put all the arguments for the xfer in the accb */ hccb->target = ccb->ccb_h.target_id; hccb->lun = ccb->ccb_h.target_lun; hccb->ahastat = 0; hccb->sdstat = 0; if (ccb->ccb_h.func_code == XPT_SCSI_IO) { struct ccb_scsiio *csio; struct ccb_hdr *ccbh; int error; csio = &ccb->csio; ccbh = &csio->ccb_h; hccb->opcode = aha->ccb_ccb_opcode; hccb->datain = (ccb->ccb_h.flags & CAM_DIR_IN) != 0; hccb->dataout = (ccb->ccb_h.flags & CAM_DIR_OUT) != 0; hccb->cmd_len = csio->cdb_len; if (hccb->cmd_len > sizeof(hccb->scsi_cdb)) { ccb->ccb_h.status = CAM_REQ_INVALID; ahafreeccb(aha, accb); xpt_done(ccb); return; } hccb->sense_len = csio->sense_len; if ((ccbh->flags & CAM_CDB_POINTER) != 0) { if ((ccbh->flags & CAM_CDB_PHYS) == 0) { bcopy(csio->cdb_io.cdb_ptr, hccb->scsi_cdb, hccb->cmd_len); } else { /* I guess I could map it in... */ ccbh->status = CAM_REQ_INVALID; ahafreeccb(aha, accb); xpt_done(ccb); return; } } else { bcopy(csio->cdb_io.cdb_bytes, hccb->scsi_cdb, hccb->cmd_len); } /* * If we have any data to send with this command, * map it into bus space. */ error = bus_dmamap_load_ccb( aha->buffer_dmat, accb->dmamap, ccb, ahaexecuteccb, accb, /*flags*/0); if (error == EINPROGRESS) { /* * So as to maintain ordering, freeze the * controller queue until our mapping is * returned. */ xpt_freeze_simq(aha->sim, 1); csio->ccb_h.status |= CAM_RELEASE_SIMQ; } } else { hccb->opcode = INITIATOR_BUS_DEV_RESET; /* No data transfer */ hccb->datain = TRUE; hccb->dataout = TRUE; hccb->cmd_len = 0; hccb->sense_len = 0; ahaexecuteccb(accb, NULL, 0, 0); } break; } case XPT_EN_LUN: /* Enable LUN as a target */ case XPT_TARGET_IO: /* Execute target I/O request */ case XPT_ACCEPT_TARGET_IO: /* Accept Host Target Mode CDB */ case XPT_CONT_TARGET_IO: /* Continue Host Target I/O Connection*/ case XPT_ABORT: /* Abort the specified CCB */ /* XXX Implement */ ccb->ccb_h.status = CAM_REQ_INVALID; xpt_done(ccb); break; case XPT_SET_TRAN_SETTINGS: /* XXX Implement */ ccb->ccb_h.status = CAM_PROVIDE_FAIL; xpt_done(ccb); break; case XPT_GET_TRAN_SETTINGS: /* Get default/user set transfer settings for the target */ { struct ccb_trans_settings *cts = &ccb->cts; u_int target_mask = 0x01 << ccb->ccb_h.target_id; struct ccb_trans_settings_scsi *scsi = &cts->proto_specific.scsi; struct ccb_trans_settings_spi *spi = &cts->xport_specific.spi; cts->protocol = PROTO_SCSI; cts->protocol_version = SCSI_REV_2; cts->transport = XPORT_SPI; cts->transport_version = 2; if (cts->type == CTS_TYPE_USER_SETTINGS) { spi->flags = 0; if ((aha->disc_permitted & target_mask) != 0) spi->flags |= CTS_SPI_FLAGS_DISC_ENB; spi->bus_width = MSG_EXT_WDTR_BUS_8_BIT; if ((aha->sync_permitted & target_mask) != 0) { if (aha->boardid >= BOARD_1542CF) spi->sync_period = 25; else spi->sync_period = 50; } else { spi->sync_period = 0; } if (spi->sync_period != 0) spi->sync_offset = 15; spi->valid = CTS_SPI_VALID_SYNC_RATE | CTS_SPI_VALID_SYNC_OFFSET | CTS_SPI_VALID_BUS_WIDTH | CTS_SPI_VALID_DISC; scsi->valid = CTS_SCSI_VALID_TQ; } else { ahafetchtransinfo(aha, cts); } ccb->ccb_h.status = CAM_REQ_CMP; xpt_done(ccb); break; } case XPT_CALC_GEOMETRY: { struct ccb_calc_geometry *ccg; uint32_t size_mb; uint32_t secs_per_cylinder; ccg = &ccb->ccg; size_mb = ccg->volume_size / ((1024L * 1024L) / ccg->block_size); if (size_mb >= 1024 && (aha->extended_trans != 0)) { if (size_mb >= 2048) { ccg->heads = 255; ccg->secs_per_track = 63; } else { ccg->heads = 128; ccg->secs_per_track = 32; } } else { ccg->heads = 64; ccg->secs_per_track = 32; } secs_per_cylinder = ccg->heads * ccg->secs_per_track; ccg->cylinders = ccg->volume_size / secs_per_cylinder; ccb->ccb_h.status = CAM_REQ_CMP; xpt_done(ccb); break; } case XPT_RESET_BUS: /* Reset the specified SCSI bus */ ahareset(aha, /*hardreset*/TRUE); ccb->ccb_h.status = CAM_REQ_CMP; xpt_done(ccb); break; case XPT_TERM_IO: /* Terminate the I/O process */ /* XXX Implement */ ccb->ccb_h.status = CAM_REQ_INVALID; xpt_done(ccb); break; case XPT_PATH_INQ: /* Path routing inquiry */ { struct ccb_pathinq *cpi = &ccb->cpi; cpi->version_num = 1; /* XXX??? */ cpi->hba_inquiry = PI_SDTR_ABLE; cpi->target_sprt = 0; cpi->hba_misc = 0; cpi->hba_eng_cnt = 0; cpi->max_target = 7; cpi->max_lun = 7; cpi->initiator_id = aha->scsi_id; cpi->bus_id = cam_sim_bus(sim); cpi->base_transfer_speed = 3300; strlcpy(cpi->sim_vid, "FreeBSD", SIM_IDLEN); strlcpy(cpi->hba_vid, "Adaptec", HBA_IDLEN); strlcpy(cpi->dev_name, cam_sim_name(sim), DEV_IDLEN); cpi->unit_number = cam_sim_unit(sim); - cpi->transport = XPORT_SPI; - cpi->transport_version = 2; - cpi->protocol = PROTO_SCSI; - cpi->protocol_version = SCSI_REV_2; + cpi->transport = XPORT_SPI; + cpi->transport_version = 2; + cpi->protocol = PROTO_SCSI; + cpi->protocol_version = SCSI_REV_2; cpi->ccb_h.status = CAM_REQ_CMP; xpt_done(ccb); break; } default: ccb->ccb_h.status = CAM_REQ_INVALID; xpt_done(ccb); break; } } static void ahaexecuteccb(void *arg, bus_dma_segment_t *dm_segs, int nseg, int error) { struct aha_ccb *accb; union ccb *ccb; struct aha_softc *aha; uint32_t paddr; accb = (struct aha_ccb *)arg; ccb = accb->ccb; aha = (struct aha_softc *)ccb->ccb_h.ccb_aha_ptr; if (error != 0) { if (error != EFBIG) device_printf(aha->dev, "Unexepected error 0x%x returned from " "bus_dmamap_load\n", error); if (ccb->ccb_h.status == CAM_REQ_INPROG) { xpt_freeze_devq(ccb->ccb_h.path, /*count*/1); ccb->ccb_h.status = CAM_REQ_TOO_BIG|CAM_DEV_QFRZN; } ahafreeccb(aha, accb); xpt_done(ccb); return; } if (nseg != 0) { aha_sg_t *sg; bus_dma_segment_t *end_seg; bus_dmasync_op_t op; end_seg = dm_segs + nseg; /* Copy the segments into our SG list */ sg = accb->sg_list; while (dm_segs < end_seg) { ahautoa24(dm_segs->ds_len, sg->len); ahautoa24(dm_segs->ds_addr, sg->addr); sg++; dm_segs++; } if (nseg > 1) { accb->hccb.opcode = aha->ccb_sg_opcode; ahautoa24((sizeof(aha_sg_t) * nseg), accb->hccb.data_len); ahautoa24(accb->sg_list_phys, accb->hccb.data_addr); } else { bcopy(accb->sg_list->len, accb->hccb.data_len, 3); bcopy(accb->sg_list->addr, accb->hccb.data_addr, 3); } if ((ccb->ccb_h.flags & CAM_DIR_MASK) == CAM_DIR_IN) op = BUS_DMASYNC_PREREAD; else op = BUS_DMASYNC_PREWRITE; bus_dmamap_sync(aha->buffer_dmat, accb->dmamap, op); } else { accb->hccb.opcode = INITIATOR_CCB; ahautoa24(0, accb->hccb.data_len); ahautoa24(0, accb->hccb.data_addr); } /* * Last time we need to check if this CCB needs to * be aborted. */ if (ccb->ccb_h.status != CAM_REQ_INPROG) { if (nseg != 0) bus_dmamap_unload(aha->buffer_dmat, accb->dmamap); ahafreeccb(aha, accb); xpt_done(ccb); return; } accb->flags = ACCB_ACTIVE; ccb->ccb_h.status |= CAM_SIM_QUEUED; LIST_INSERT_HEAD(&aha->pending_ccbs, &ccb->ccb_h, sim_links.le); callout_reset_sbt(&accb->timer, SBT_1MS * ccb->ccb_h.timeout, 0, ahatimeout, accb, 0); /* Tell the adapter about this command */ if (aha->cur_outbox->action_code != AMBO_FREE) { /* * We should never encounter a busy mailbox. * If we do, warn the user, and treat it as * a resource shortage. If the controller is * hung, one of the pending transactions will * timeout causing us to start recovery operations. */ device_printf(aha->dev, "Encountered busy mailbox with %d out of %d " "commands active!!!", aha->active_ccbs, aha->max_ccbs); callout_stop(&accb->timer); if (nseg != 0) bus_dmamap_unload(aha->buffer_dmat, accb->dmamap); ahafreeccb(aha, accb); aha->resource_shortage = TRUE; xpt_freeze_simq(aha->sim, /*count*/1); ccb->ccb_h.status = CAM_REQUEUE_REQ; xpt_done(ccb); return; } paddr = ahaccbvtop(aha, accb); ahautoa24(paddr, aha->cur_outbox->ccb_addr); aha->cur_outbox->action_code = AMBO_START; aha_outb(aha, COMMAND_REG, AOP_START_MBOX); ahanextoutbox(aha); } void aha_intr(void *arg) { struct aha_softc *aha; aha = arg; mtx_lock(&aha->lock); aha_intr_locked(aha); mtx_unlock(&aha->lock); } void aha_intr_locked(struct aha_softc *aha) { u_int intstat; uint32_t paddr; while (((intstat = aha_inb(aha, INTSTAT_REG)) & INTR_PENDING) != 0) { if ((intstat & CMD_COMPLETE) != 0) { aha->latched_status = aha_inb(aha, STATUS_REG); aha->command_cmp = TRUE; } aha_outb(aha, CONTROL_REG, RESET_INTR); if ((intstat & IMB_LOADED) != 0) { while (aha->cur_inbox->comp_code != AMBI_FREE) { paddr = aha_a24tou(aha->cur_inbox->ccb_addr); ahadone(aha, ahaccbptov(aha, paddr), aha->cur_inbox->comp_code); aha->cur_inbox->comp_code = AMBI_FREE; ahanextinbox(aha); } } if ((intstat & SCSI_BUS_RESET) != 0) { ahareset(aha, /*hardreset*/FALSE); } } } static void ahadone(struct aha_softc *aha, struct aha_ccb *accb, aha_mbi_comp_code_t comp_code) { union ccb *ccb; struct ccb_scsiio *csio; ccb = accb->ccb; csio = &accb->ccb->csio; if ((accb->flags & ACCB_ACTIVE) == 0) { device_printf(aha->dev, "ahadone - Attempt to free non-active ACCB %p\n", (void *)accb); return; } if ((ccb->ccb_h.flags & CAM_DIR_MASK) != CAM_DIR_NONE) { bus_dmasync_op_t op; if ((ccb->ccb_h.flags & CAM_DIR_MASK) == CAM_DIR_IN) op = BUS_DMASYNC_POSTREAD; else op = BUS_DMASYNC_POSTWRITE; bus_dmamap_sync(aha->buffer_dmat, accb->dmamap, op); bus_dmamap_unload(aha->buffer_dmat, accb->dmamap); } if (accb == aha->recovery_accb) { /* * The recovery ACCB does not have a CCB associated * with it, so short circuit the normal error handling. * We now traverse our list of pending CCBs and process * any that were terminated by the recovery CCBs action. * We also reinstate timeouts for all remaining, pending, * CCBs. */ struct cam_path *path; struct ccb_hdr *ccb_h; cam_status error; /* Notify all clients that a BDR occurred */ error = xpt_create_path(&path, /*periph*/NULL, cam_sim_path(aha->sim), accb->hccb.target, CAM_LUN_WILDCARD); if (error == CAM_REQ_CMP) { xpt_async(AC_SENT_BDR, path, NULL); xpt_free_path(path); } ccb_h = LIST_FIRST(&aha->pending_ccbs); while (ccb_h != NULL) { struct aha_ccb *pending_accb; pending_accb = (struct aha_ccb *)ccb_h->ccb_accb_ptr; if (pending_accb->hccb.target == accb->hccb.target) { pending_accb->hccb.ahastat = AHASTAT_HA_BDR; ccb_h = LIST_NEXT(ccb_h, sim_links.le); ahadone(aha, pending_accb, AMBI_ERROR); } else { callout_reset_sbt(&pending_accb->timer, SBT_1MS * ccb_h->timeout, 0, ahatimeout, pending_accb, 0); ccb_h = LIST_NEXT(ccb_h, sim_links.le); } } device_printf(aha->dev, "No longer in timeout\n"); return; } callout_stop(&accb->timer); switch (comp_code) { case AMBI_FREE: device_printf(aha->dev, "ahadone - CCB completed with free status!\n"); break; case AMBI_NOT_FOUND: device_printf(aha->dev, "ahadone - CCB Abort failed to find CCB\n"); break; case AMBI_ABORT: case AMBI_ERROR: /* An error occurred */ if (accb->hccb.opcode < INITIATOR_CCB_WRESID) csio->resid = 0; else csio->resid = aha_a24tou(accb->hccb.data_len); switch(accb->hccb.ahastat) { case AHASTAT_DATARUN_ERROR: { if (csio->resid <= 0) { csio->ccb_h.status = CAM_DATA_RUN_ERR; break; } /* FALLTHROUGH */ } case AHASTAT_NOERROR: csio->scsi_status = accb->hccb.sdstat; csio->ccb_h.status |= CAM_SCSI_STATUS_ERROR; switch(csio->scsi_status) { case SCSI_STATUS_CHECK_COND: case SCSI_STATUS_CMD_TERMINATED: csio->ccb_h.status |= CAM_AUTOSNS_VALID; /* * The aha writes the sense data at different * offsets based on the scsi cmd len */ bcopy((caddr_t) &accb->hccb.scsi_cdb + accb->hccb.cmd_len, (caddr_t) &csio->sense_data, accb->hccb.sense_len); break; default: break; case SCSI_STATUS_OK: csio->ccb_h.status = CAM_REQ_CMP; break; } break; case AHASTAT_SELTIMEOUT: csio->ccb_h.status = CAM_SEL_TIMEOUT; break; case AHASTAT_UNEXPECTED_BUSFREE: csio->ccb_h.status = CAM_UNEXP_BUSFREE; break; case AHASTAT_INVALID_PHASE: csio->ccb_h.status = CAM_SEQUENCE_FAIL; break; case AHASTAT_INVALID_ACTION_CODE: panic("%s: Inavlid Action code", aha_name(aha)); break; case AHASTAT_INVALID_OPCODE: if (accb->hccb.opcode < INITIATOR_CCB_WRESID) panic("%s: Invalid CCB Opcode %x hccb = %p", aha_name(aha), accb->hccb.opcode, &accb->hccb); device_printf(aha->dev, "AHA-1540A compensation failed\n"); xpt_freeze_devq(ccb->ccb_h.path, /*count*/1); csio->ccb_h.status = CAM_REQUEUE_REQ; break; case AHASTAT_LINKED_CCB_LUN_MISMATCH: /* We don't even support linked commands... */ panic("%s: Linked CCB Lun Mismatch", aha_name(aha)); break; case AHASTAT_INVALID_CCB_OR_SG_PARAM: panic("%s: Invalid CCB or SG list", aha_name(aha)); break; case AHASTAT_HA_SCSI_BUS_RESET: if ((csio->ccb_h.status & CAM_STATUS_MASK) != CAM_CMD_TIMEOUT) csio->ccb_h.status = CAM_SCSI_BUS_RESET; break; case AHASTAT_HA_BDR: if ((accb->flags & ACCB_DEVICE_RESET) == 0) csio->ccb_h.status = CAM_BDR_SENT; else csio->ccb_h.status = CAM_CMD_TIMEOUT; break; } if (csio->ccb_h.status != CAM_REQ_CMP) { xpt_freeze_devq(csio->ccb_h.path, /*count*/1); csio->ccb_h.status |= CAM_DEV_QFRZN; } if ((accb->flags & ACCB_RELEASE_SIMQ) != 0) ccb->ccb_h.status |= CAM_RELEASE_SIMQ; ahafreeccb(aha, accb); xpt_done(ccb); break; case AMBI_OK: /* All completed without incident */ /* XXX DO WE NEED TO COPY SENSE BYTES HERE???? XXX */ /* I don't think so since it works???? */ ccb->ccb_h.status |= CAM_REQ_CMP; if ((accb->flags & ACCB_RELEASE_SIMQ) != 0) ccb->ccb_h.status |= CAM_RELEASE_SIMQ; ahafreeccb(aha, accb); xpt_done(ccb); break; } } static int ahareset(struct aha_softc* aha, int hard_reset) { struct ccb_hdr *ccb_h; u_int status; u_int timeout; uint8_t reset_type; if (hard_reset != 0) reset_type = HARD_RESET; else reset_type = SOFT_RESET; aha_outb(aha, CONTROL_REG, reset_type); /* Wait 5sec. for Diagnostic start */ timeout = 5 * 10000; while (--timeout) { status = aha_inb(aha, STATUS_REG); if ((status & DIAG_ACTIVE) != 0) break; DELAY(100); } if (timeout == 0) { PRVERB((aha->dev, "ahareset - Diagnostic Active failed to " "assert. status = %#x\n", status)); return (ETIMEDOUT); } /* Wait 10sec. for Diagnostic end */ timeout = 10 * 10000; while (--timeout) { status = aha_inb(aha, STATUS_REG); if ((status & DIAG_ACTIVE) == 0) break; DELAY(100); } if (timeout == 0) { panic("%s: ahareset - Diagnostic Active failed to drop. " "status = 0x%x\n", aha_name(aha), status); return (ETIMEDOUT); } /* Wait for the host adapter to become ready or report a failure */ timeout = 10000; while (--timeout) { status = aha_inb(aha, STATUS_REG); if ((status & (DIAG_FAIL|HA_READY|DATAIN_REG_READY)) != 0) break; DELAY(100); } if (timeout == 0) { device_printf(aha->dev, "ahareset - Host adapter failed to " "come ready. status = 0x%x\n", status); return (ETIMEDOUT); } /* If the diagnostics failed, tell the user */ if ((status & DIAG_FAIL) != 0 || (status & HA_READY) == 0) { device_printf(aha->dev, "ahareset - Adapter failed diag\n"); if ((status & DATAIN_REG_READY) != 0) device_printf(aha->dev, "ahareset - Host Adapter " "Error code = 0x%x\n", aha_inb(aha, DATAIN_REG)); return (ENXIO); } /* If we've attached to the XPT, tell it about the event */ if (aha->path != NULL) xpt_async(AC_BUS_RESET, aha->path, NULL); /* * Perform completion processing for all outstanding CCBs. */ while ((ccb_h = LIST_FIRST(&aha->pending_ccbs)) != NULL) { struct aha_ccb *pending_accb; pending_accb = (struct aha_ccb *)ccb_h->ccb_accb_ptr; pending_accb->hccb.ahastat = AHASTAT_HA_SCSI_BUS_RESET; ahadone(aha, pending_accb, AMBI_ERROR); } /* If we've allocated mailboxes, initialize them */ /* Must be done after we've aborted our queue, or aha_cmd fails */ if (aha->init_level > 4) ahainitmboxes(aha); return (0); } /* * Send a command to the adapter. */ int aha_cmd(struct aha_softc *aha, aha_op_t opcode, uint8_t *params, u_int param_len, uint8_t *reply_data, u_int reply_len, u_int cmd_timeout) { u_int timeout; u_int status; u_int saved_status; u_int intstat; u_int reply_buf_size; int cmd_complete; int error; /* No data returned to start */ reply_buf_size = reply_len; reply_len = 0; intstat = 0; cmd_complete = 0; saved_status = 0; error = 0; /* * All commands except for the "start mailbox" and the "enable * outgoing mailbox read interrupt" commands cannot be issued * while there are pending transactions. Freeze our SIMQ * and wait for all completions to occur if necessary. */ timeout = 10000; while (LIST_FIRST(&aha->pending_ccbs) != NULL && --timeout) { /* Fire the interrupt handler in case interrupts are blocked */ aha_intr(aha); DELAY(10); } if (timeout == 0) { device_printf(aha->dev, "aha_cmd: Timeout waiting for adapter idle\n"); return (ETIMEDOUT); } aha->command_cmp = 0; /* * Wait up to 10 sec. for the adapter to become * ready to accept commands. */ timeout = 100000; while (--timeout) { status = aha_inb(aha, STATUS_REG); if ((status & HA_READY) != 0 && (status & CMD_REG_BUSY) == 0) break; /* * Throw away any pending data which may be * left over from earlier commands that we * timedout on. */ if ((status & DATAIN_REG_READY) != 0) (void)aha_inb(aha, DATAIN_REG); DELAY(100); } if (timeout == 0) { device_printf(aha->dev, "aha_cmd: Timeout waiting for adapter" " ready, status = 0x%x\n", status); return (ETIMEDOUT); } /* * Send the opcode followed by any necessary parameter bytes. */ aha_outb(aha, COMMAND_REG, opcode); /* * Wait for up to 1sec to get the parameter list sent */ timeout = 10000; while (param_len && --timeout) { DELAY(100); status = aha_inb(aha, STATUS_REG); intstat = aha_inb(aha, INTSTAT_REG); if ((intstat & (INTR_PENDING|CMD_COMPLETE)) == (INTR_PENDING|CMD_COMPLETE)) { saved_status = status; cmd_complete = 1; break; } if (aha->command_cmp != 0) { saved_status = aha->latched_status; cmd_complete = 1; break; } if ((status & DATAIN_REG_READY) != 0) break; if ((status & CMD_REG_BUSY) == 0) { aha_outb(aha, COMMAND_REG, *params++); param_len--; timeout = 10000; } } if (timeout == 0) { device_printf(aha->dev, "aha_cmd: Timeout sending parameters, " "status = 0x%x\n", status); error = ETIMEDOUT; } /* * For all other commands, we wait for any output data * and the final comand completion interrupt. */ while (cmd_complete == 0 && --cmd_timeout) { status = aha_inb(aha, STATUS_REG); intstat = aha_inb(aha, INTSTAT_REG); if (aha->command_cmp != 0) { cmd_complete = 1; saved_status = aha->latched_status; } else if ((intstat & (INTR_PENDING|CMD_COMPLETE)) == (INTR_PENDING|CMD_COMPLETE)) { /* * Our poll (in case interrupts are blocked) * saw the CMD_COMPLETE interrupt. */ cmd_complete = 1; saved_status = status; } if ((status & DATAIN_REG_READY) != 0) { uint8_t data; data = aha_inb(aha, DATAIN_REG); if (reply_len < reply_buf_size) { *reply_data++ = data; } else { device_printf(aha->dev, "aha_cmd - Discarded reply data " "byte for opcode 0x%x\n", opcode); } /* * Reset timeout to ensure at least a second * between response bytes. */ cmd_timeout = MAX(cmd_timeout, 10000); reply_len++; } DELAY(100); } if (cmd_timeout == 0) { device_printf(aha->dev, "aha_cmd: Timeout: status = 0x%x, " "intstat = 0x%x, reply_len = %d\n", status, intstat, reply_len); return (ETIMEDOUT); } /* * Clear any pending interrupts. Block interrupts so our * interrupt handler is not re-entered. */ aha_intr(aha); if (error != 0) return (error); /* * If the command was rejected by the controller, tell the caller. */ if ((saved_status & CMD_INVALID) != 0) { PRVERB((aha->dev, "Invalid Command 0x%x\n", opcode)); /* * Some early adapters may not recover properly from * an invalid command. If it appears that the controller * has wedged (i.e. status was not cleared by our interrupt * reset above), perform a soft reset. */ DELAY(1000); status = aha_inb(aha, STATUS_REG); if ((status & (CMD_INVALID|STATUS_REG_RSVD|DATAIN_REG_READY| CMD_REG_BUSY|DIAG_FAIL|DIAG_ACTIVE)) != 0 || (status & (HA_READY|INIT_REQUIRED)) != (HA_READY|INIT_REQUIRED)) ahareset(aha, /*hard_reset*/FALSE); return (EINVAL); } if (param_len > 0) { /* The controller did not accept the full argument list */ PRVERB((aha->dev, "Controller did not accept full argument " "list (%d > 0)\n", param_len)); return (E2BIG); } if (reply_len != reply_buf_size) { /* Too much or too little data received */ PRVERB((aha->dev, "data received mismatch (%d != %d)\n", reply_len, reply_buf_size)); return (EMSGSIZE); } /* We were successful */ return (0); } static int ahainitmboxes(struct aha_softc *aha) { int error; init_24b_mbox_params_t init_mbox; bzero(aha->in_boxes, sizeof(aha_mbox_in_t) * aha->num_boxes); bzero(aha->out_boxes, sizeof(aha_mbox_out_t) * aha->num_boxes); aha->cur_inbox = aha->in_boxes; aha->last_inbox = aha->in_boxes + aha->num_boxes - 1; aha->cur_outbox = aha->out_boxes; aha->last_outbox = aha->out_boxes + aha->num_boxes - 1; /* Tell the adapter about them */ init_mbox.num_mboxes = aha->num_boxes; ahautoa24(aha->mailbox_physbase, init_mbox.base_addr); error = aha_cmd(aha, AOP_INITIALIZE_MBOX, (uint8_t *)&init_mbox, /*parmlen*/sizeof(init_mbox), /*reply_buf*/NULL, /*reply_len*/0, DEFAULT_CMD_TIMEOUT); if (error != 0) printf("ahainitmboxes: Initialization command failed\n"); return (error); } /* * Update the XPT's idea of the negotiated transfer * parameters for a particular target. */ static void ahafetchtransinfo(struct aha_softc *aha, struct ccb_trans_settings* cts) { setup_data_t setup_info; u_int target; u_int targ_offset; u_int sync_period; int error; uint8_t param; targ_syncinfo_t sync_info; struct ccb_trans_settings_spi *spi = &cts->xport_specific.spi; target = cts->ccb_h.target_id; targ_offset = (target & 0x7); /* * Inquire Setup Information. This command retreives * the sync info for older models. */ param = sizeof(setup_info); error = aha_cmd(aha, AOP_INQUIRE_SETUP_INFO, ¶m, /*paramlen*/1, (uint8_t*)&setup_info, sizeof(setup_info), DEFAULT_CMD_TIMEOUT); if (error != 0) { device_printf(aha->dev, "ahafetchtransinfo - Inquire Setup Info Failed %d\n", error); return; } sync_info = setup_info.syncinfo[targ_offset]; if (sync_info.sync == 0) spi->sync_offset = 0; else spi->sync_offset = sync_info.offset; spi->bus_width = MSG_EXT_WDTR_BUS_8_BIT; if (aha->boardid >= BOARD_1542CF) sync_period = 1000; else sync_period = 2000; sync_period += 500 * sync_info.period; /* Convert ns value to standard SCSI sync rate */ if (spi->sync_offset != 0) spi->sync_period = scsi_calc_syncparam(sync_period); else spi->sync_period = 0; spi->valid = CTS_SPI_VALID_SYNC_RATE | CTS_SPI_VALID_SYNC_OFFSET | CTS_SPI_VALID_BUS_WIDTH; xpt_async(AC_TRANSFER_NEG, cts->ccb_h.path, cts); } static void ahamapmboxes(void *arg, bus_dma_segment_t *segs, int nseg, int error) { struct aha_softc* aha; aha = (struct aha_softc*)arg; aha->mailbox_physbase = segs->ds_addr; } static void ahamapccbs(void *arg, bus_dma_segment_t *segs, int nseg, int error) { struct aha_softc* aha; aha = (struct aha_softc*)arg; aha->aha_ccb_physbase = segs->ds_addr; } static void ahamapsgs(void *arg, bus_dma_segment_t *segs, int nseg, int error) { struct aha_softc* aha; aha = (struct aha_softc*)arg; SLIST_FIRST(&aha->sg_maps)->sg_physaddr = segs->ds_addr; } static void ahapoll(struct cam_sim *sim) { aha_intr_locked(cam_sim_softc(sim)); } static void ahatimeout(void *arg) { struct aha_ccb *accb; union ccb *ccb; struct aha_softc *aha; uint32_t paddr; struct ccb_hdr *ccb_h; accb = (struct aha_ccb *)arg; ccb = accb->ccb; aha = (struct aha_softc *)ccb->ccb_h.ccb_aha_ptr; mtx_assert(&aha->lock, MA_OWNED); xpt_print_path(ccb->ccb_h.path); printf("CCB %p - timed out\n", (void *)accb); if ((accb->flags & ACCB_ACTIVE) == 0) { xpt_print_path(ccb->ccb_h.path); printf("CCB %p - timed out CCB already completed\n", (void *)accb); return; } /* * In order to simplify the recovery process, we ask the XPT * layer to halt the queue of new transactions and we traverse * the list of pending CCBs and remove their timeouts. This * means that the driver attempts to clear only one error * condition at a time. In general, timeouts that occur * close together are related anyway, so there is no benefit * in attempting to handle errors in parallel. Timeouts will * be reinstated when the recovery process ends. */ if ((accb->flags & ACCB_DEVICE_RESET) == 0) { if ((accb->flags & ACCB_RELEASE_SIMQ) == 0) { xpt_freeze_simq(aha->sim, /*count*/1); accb->flags |= ACCB_RELEASE_SIMQ; } ccb_h = LIST_FIRST(&aha->pending_ccbs); while (ccb_h != NULL) { struct aha_ccb *pending_accb; pending_accb = (struct aha_ccb *)ccb_h->ccb_accb_ptr; callout_stop(&pending_accb->timer); ccb_h = LIST_NEXT(ccb_h, sim_links.le); } } if ((accb->flags & ACCB_DEVICE_RESET) != 0 || aha->cur_outbox->action_code != AMBO_FREE) { /* * Try a full host adapter/SCSI bus reset. * We do this only if we have already attempted * to clear the condition with a BDR, or we cannot * attempt a BDR for lack of mailbox resources. */ ccb->ccb_h.status = CAM_CMD_TIMEOUT; ahareset(aha, /*hardreset*/TRUE); device_printf(aha->dev, "No longer in timeout\n"); } else { /* * Send a Bus Device Reset message: * The target that is holding up the bus may not * be the same as the one that triggered this timeout * (different commands have different timeout lengths), * but we have no way of determining this from our * timeout handler. Our strategy here is to queue a * BDR message to the target of the timed out command. * If this fails, we'll get another timeout 2 seconds * later which will attempt a bus reset. */ accb->flags |= ACCB_DEVICE_RESET; callout_reset(&accb->timer, 2 * hz, ahatimeout, accb); aha->recovery_accb->hccb.opcode = INITIATOR_BUS_DEV_RESET; /* No Data Transfer */ aha->recovery_accb->hccb.datain = TRUE; aha->recovery_accb->hccb.dataout = TRUE; aha->recovery_accb->hccb.ahastat = 0; aha->recovery_accb->hccb.sdstat = 0; aha->recovery_accb->hccb.target = ccb->ccb_h.target_id; /* Tell the adapter about this command */ paddr = ahaccbvtop(aha, aha->recovery_accb); ahautoa24(paddr, aha->cur_outbox->ccb_addr); aha->cur_outbox->action_code = AMBO_START; aha_outb(aha, COMMAND_REG, AOP_START_MBOX); ahanextoutbox(aha); } } int aha_detach(struct aha_softc *aha) { mtx_lock(&aha->lock); xpt_async(AC_LOST_DEVICE, aha->path, NULL); xpt_free_path(aha->path); xpt_bus_deregister(cam_sim_path(aha->sim)); cam_sim_free(aha->sim, /*free_devq*/TRUE); mtx_unlock(&aha->lock); /* XXX: Drain all timers? */ return (0); } MODULE_DEPEND(aha, cam, 1, 1, 1); Index: projects/clang400-import/sys/dev/ahb/ahb.c =================================================================== --- projects/clang400-import/sys/dev/ahb/ahb.c (revision 311459) +++ projects/clang400-import/sys/dev/ahb/ahb.c (revision 311460) @@ -1,1315 +1,1315 @@ /*- * CAM SCSI device driver for the Adaptec 174X SCSI Host adapter * * Copyright (c) 1998 Justin T. Gibbs * 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 immediately at the beginning of the file, without modification, * this list of conditions, and the following disclaimer. * 2. 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 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 #include #include #include #include #include #include #include #include #include #include #include #include #include #define ccb_ecb_ptr spriv_ptr0 #define ccb_ahb_ptr spriv_ptr1 #define ahb_inb(ahb, port) \ bus_read_1((ahb)->res, port) #define ahb_inl(ahb, port) \ bus_read_4((ahb)->res, port) #define ahb_outb(ahb, port, value) \ bus_write_1((ahb)->res, port, value) #define ahb_outl(ahb, port, value) \ bus_write_4((ahb)->res, port, value) static const char *ahbmatch(eisa_id_t type); static struct ahb_softc *ahballoc(device_t dev, struct resource *res); static void ahbfree(struct ahb_softc *ahb); static int ahbreset(struct ahb_softc *ahb); static void ahbmapecbs(void *arg, bus_dma_segment_t *segs, int nseg, int error); static int ahbxptattach(struct ahb_softc *ahb); static void ahbhandleimmed(struct ahb_softc *ahb, u_int32_t mbox, u_int intstat); static void ahbcalcresid(struct ahb_softc *ahb, struct ecb *ecb, union ccb *ccb); static __inline void ahbdone(struct ahb_softc *ahb, u_int32_t mbox, u_int intstat); static void ahbintr(void *arg); static void ahbintr_locked(struct ahb_softc *ahb); static bus_dmamap_callback_t ahbexecuteecb; static void ahbaction(struct cam_sim *sim, union ccb *ccb); static void ahbpoll(struct cam_sim *sim); /* Our timeout handler */ static void ahbtimeout(void *arg); static __inline struct ecb* ahbecbget(struct ahb_softc *ahb); static __inline void ahbecbfree(struct ahb_softc* ahb, struct ecb* ecb); static __inline u_int32_t ahbecbvtop(struct ahb_softc *ahb, struct ecb *ecb); static __inline struct ecb* ahbecbptov(struct ahb_softc *ahb, u_int32_t ecb_addr); static __inline u_int32_t ahbstatuspaddr(u_int32_t ecb_paddr); static __inline u_int32_t ahbsensepaddr(u_int32_t ecb_paddr); static __inline u_int32_t ahbsgpaddr(u_int32_t ecb_paddr); static __inline void ahbqueuembox(struct ahb_softc *ahb, u_int32_t mboxval, u_int attn_code); static __inline struct ecb* ahbecbget(struct ahb_softc *ahb) { struct ecb* ecb; if (!dumping) mtx_assert(&ahb->lock, MA_OWNED); if ((ecb = SLIST_FIRST(&ahb->free_ecbs)) != NULL) SLIST_REMOVE_HEAD(&ahb->free_ecbs, links); return (ecb); } static __inline void ahbecbfree(struct ahb_softc* ahb, struct ecb* ecb) { if (!dumping) mtx_assert(&ahb->lock, MA_OWNED); ecb->state = ECB_FREE; SLIST_INSERT_HEAD(&ahb->free_ecbs, ecb, links); } static __inline u_int32_t ahbecbvtop(struct ahb_softc *ahb, struct ecb *ecb) { return (ahb->ecb_physbase + (u_int32_t)((caddr_t)ecb - (caddr_t)ahb->ecb_array)); } static __inline struct ecb* ahbecbptov(struct ahb_softc *ahb, u_int32_t ecb_addr) { return (ahb->ecb_array + ((struct ecb*)(uintptr_t)ecb_addr - (struct ecb*)(uintptr_t)ahb->ecb_physbase)); } static __inline u_int32_t ahbstatuspaddr(u_int32_t ecb_paddr) { return (ecb_paddr + offsetof(struct ecb, status)); } static __inline u_int32_t ahbsensepaddr(u_int32_t ecb_paddr) { return (ecb_paddr + offsetof(struct ecb, sense)); } static __inline u_int32_t ahbsgpaddr(u_int32_t ecb_paddr) { return (ecb_paddr + offsetof(struct ecb, sg_list)); } static __inline void ahbqueuembox(struct ahb_softc *ahb, u_int32_t mboxval, u_int attn_code) { u_int loopmax = 300; while (--loopmax) { u_int status; status = ahb_inb(ahb, HOSTSTAT); if ((status & (HOSTSTAT_MBOX_EMPTY|HOSTSTAT_BUSY)) == HOSTSTAT_MBOX_EMPTY) break; DELAY(20); } if (loopmax == 0) panic("%s: adapter not taking commands\n", device_get_nameunit(ahb->dev)); ahb_outl(ahb, MBOXOUT0, mboxval); ahb_outb(ahb, ATTN, attn_code); } static const char * ahbmatch(eisa_id_t type) { switch(type & 0xfffffe00) { case EISA_DEVICE_ID_ADAPTEC_1740: return ("Adaptec 174x SCSI host adapter"); break; default: break; } return (NULL); } static int ahbprobe(device_t dev) { const char *desc; u_int32_t iobase; u_int32_t irq; u_int8_t intdef; int shared; desc = ahbmatch(eisa_get_id(dev)); if (!desc) return (ENXIO); device_set_desc(dev, desc); iobase = (eisa_get_slot(dev) * EISA_SLOT_SIZE) + AHB_EISA_SLOT_OFFSET; eisa_add_iospace(dev, iobase, AHB_EISA_IOSIZE, RESVADDR_NONE); intdef = inb(INTDEF + iobase); switch (intdef & 0x7) { case INT9: irq = 9; break; case INT10: irq = 10; break; case INT11: irq = 11; break; case INT12: irq = 12; break; case INT14: irq = 14; break; case INT15: irq = 15; break; default: printf("Adaptec 174X at slot %d: illegal " "irq setting %d\n", eisa_get_slot(dev), (intdef & 0x7)); irq = 0; break; } if (irq == 0) return ENXIO; shared = (inb(INTDEF + iobase) & INTLEVEL) ? EISA_TRIGGER_LEVEL : EISA_TRIGGER_EDGE; eisa_add_intr(dev, irq, shared); return 0; } static int ahbattach(device_t dev) { /* * find unit and check we have that many defined */ struct ahb_softc *ahb; struct ecb* next_ecb; struct resource *io; struct resource *irq; int rid; void *ih; irq = NULL; rid = 0; io = bus_alloc_resource_any(dev, SYS_RES_IOPORT, &rid, RF_ACTIVE); if (io == NULL) { device_printf(dev, "No I/O space?!\n"); return ENOMEM; } ahb = ahballoc(dev, io); if (ahbreset(ahb) != 0) goto error_exit; rid = 0; irq = bus_alloc_resource_any(dev, SYS_RES_IRQ, &rid, RF_ACTIVE); if (irq == NULL) { device_printf(dev, "Can't allocate interrupt\n"); goto error_exit; } /* * Create our DMA tags. These tags define the kinds of device * accessible memory allocations and memory mappings we will * need to perform during normal operation. */ /* DMA tag for mapping buffers into device visible space. */ if (bus_dma_tag_create( /* parent */ bus_get_dma_tag(dev), /* alignment */ 1, /* boundary */ 0, /* lowaddr */ BUS_SPACE_MAXADDR_32BIT, /* highaddr */ BUS_SPACE_MAXADDR, /* filter */ NULL, /* filterarg */ NULL, /* maxsize */ DFLTPHYS, /* nsegments */ AHB_NSEG, /* maxsegsz */ BUS_SPACE_MAXSIZE_32BIT, /* flags */ BUS_DMA_ALLOCNOW, /* lockfunc */ busdma_lock_mutex, /* lockarg */ &ahb->lock, &ahb->buffer_dmat) != 0) goto error_exit; ahb->init_level++; /* DMA tag for our ccb structures and ha inquiry data */ if (bus_dma_tag_create( /* parent */ bus_get_dma_tag(dev), /* alignment */ 1, /* boundary */ 0, /* lowaddr */ BUS_SPACE_MAXADDR_32BIT, /* highaddr */ BUS_SPACE_MAXADDR, /* filter */ NULL, /* filterarg */ NULL, /* maxsize */ (AHB_NECB * sizeof(struct ecb)) + sizeof(*ahb->ha_inq_data), /* nsegments */ 1, /* maxsegsz */ BUS_SPACE_MAXSIZE_32BIT, /* flags */ 0, /* lockfunc */ NULL, /* lockarg */ NULL, &ahb->ecb_dmat) != 0) goto error_exit; ahb->init_level++; /* Allocation for our ccbs */ if (bus_dmamem_alloc(ahb->ecb_dmat, (void **)&ahb->ecb_array, BUS_DMA_NOWAIT, &ahb->ecb_dmamap) != 0) goto error_exit; ahb->ha_inq_data = (struct ha_inquiry_data *)&ahb->ecb_array[AHB_NECB]; ahb->init_level++; /* And permanently map them */ bus_dmamap_load(ahb->ecb_dmat, ahb->ecb_dmamap, ahb->ecb_array, AHB_NSEG * sizeof(struct ecb), ahbmapecbs, ahb, /*flags*/0); ahb->init_level++; /* Allocate the buffer dmamaps for each of our ECBs */ bzero(ahb->ecb_array, (AHB_NECB * sizeof(struct ecb)) + sizeof(*ahb->ha_inq_data)); next_ecb = ahb->ecb_array; while (ahb->num_ecbs < AHB_NECB) { u_int32_t ecb_paddr; if (bus_dmamap_create(ahb->buffer_dmat, /*flags*/0, &next_ecb->dmamap)) break; callout_init_mtx(&next_ecb->timer, &ahb->lock, 0); ecb_paddr = ahbecbvtop(ahb, next_ecb); next_ecb->hecb.status_ptr = ahbstatuspaddr(ecb_paddr); next_ecb->hecb.sense_ptr = ahbsensepaddr(ecb_paddr); ahb->num_ecbs++; ahbecbfree(ahb, next_ecb); next_ecb++; } ahb->init_level++; /* * Now that we know we own the resources we need, register * our bus with the XPT. */ if (ahbxptattach(ahb)) goto error_exit; /* Enable our interrupt */ if (bus_setup_intr(dev, irq, INTR_TYPE_CAM|INTR_ENTROPY|INTR_MPSAFE, NULL, ahbintr, ahb, &ih) != 0) goto error_exit; return (0); error_exit: /* * The board's IRQ line will not be left enabled * if we can't initialize correctly, so its safe * to release the irq. */ ahbfree(ahb); if (irq != NULL) bus_release_resource(dev, SYS_RES_IRQ, 0, irq); bus_release_resource(dev, SYS_RES_IOPORT, 0, io); return (-1); } static struct ahb_softc * ahballoc(device_t dev, struct resource *res) { struct ahb_softc *ahb; ahb = device_get_softc(dev); SLIST_INIT(&ahb->free_ecbs); LIST_INIT(&ahb->pending_ccbs); ahb->res = res; ahb->disc_permitted = ~0; ahb->tags_permitted = ~0; ahb->dev = dev; mtx_init(&ahb->lock, "ahb", NULL, MTX_DEF); return (ahb); } static void ahbfree(struct ahb_softc *ahb) { switch (ahb->init_level) { default: case 4: bus_dmamap_unload(ahb->ecb_dmat, ahb->ecb_dmamap); case 3: bus_dmamem_free(ahb->ecb_dmat, ahb->ecb_array, ahb->ecb_dmamap); case 2: bus_dma_tag_destroy(ahb->ecb_dmat); case 1: bus_dma_tag_destroy(ahb->buffer_dmat); case 0: break; } mtx_destroy(&ahb->lock); } /* * reset board, If it doesn't respond, return failure */ static int ahbreset(struct ahb_softc *ahb) { int wait = 1000; /* 1 sec enough? */ int test; if ((ahb_inb(ahb, PORTADDR) & PORTADDR_ENHANCED) == 0) { printf("ahb_reset: Controller not in enhanced mode\n"); return (-1); } ahb_outb(ahb, CONTROL, CNTRL_HARD_RST); DELAY(1000); ahb_outb(ahb, CONTROL, 0); while (--wait) { DELAY(1000); if ((ahb_inb(ahb, HOSTSTAT) & HOSTSTAT_BUSY) == 0) break; } if (wait == 0) { printf("ahbreset: No answer from aha1742 board\n"); return (-1); } if ((test = ahb_inb(ahb, MBOXIN0)) != 0) { printf("ahb_reset: self test failed, val = 0x%x\n", test); return (-1); } while (ahb_inb(ahb, HOSTSTAT) & HOSTSTAT_INTPEND) { ahb_outb(ahb, CONTROL, CNTRL_CLRINT); DELAY(10000); } return (0); } static void ahbmapecbs(void *arg, bus_dma_segment_t *segs, int nseg, int error) { struct ahb_softc* ahb; ahb = (struct ahb_softc*)arg; ahb->ecb_physbase = segs->ds_addr; /* * Space for adapter inquiry information is on the * tail of the ecb array. */ ahb->ha_inq_physbase = ahbecbvtop(ahb, &ahb->ecb_array[AHB_NECB]); } static int ahbxptattach(struct ahb_softc *ahb) { struct cam_devq *devq; struct ecb *ecb; u_int i; mtx_lock(&ahb->lock); /* Remember who are we on the scsi bus */ ahb->scsi_id = ahb_inb(ahb, SCSIDEF) & HSCSIID; /* Use extended translation?? */ ahb->extended_trans = ahb_inb(ahb, RESV1) & EXTENDED_TRANS; /* Fetch adapter inquiry data */ ecb = ahbecbget(ahb); /* Always succeeds - no outstanding commands */ ecb->hecb.opcode = ECBOP_READ_HA_INQDATA; ecb->hecb.flag_word1 = FW1_SUPPRESS_URUN_ERR|FW1_ERR_STATUS_BLK_ONLY; ecb->hecb.data_ptr = ahb->ha_inq_physbase; ecb->hecb.data_len = sizeof(struct ha_inquiry_data); ecb->hecb.sense_ptr = 0; ecb->state = ECB_ACTIVE; /* Tell the adapter about this command */ ahbqueuembox(ahb, ahbecbvtop(ahb, ecb), ATTN_STARTECB|ahb->scsi_id); /* Poll for interrupt completion */ for (i = 1000; ecb->state != ECB_FREE && i != 0; i--) { ahbintr_locked(ahb); DELAY(1000); } ahb->num_ecbs = MIN(ahb->num_ecbs, ahb->ha_inq_data->scsi_data.spc2_flags); device_printf(ahb->dev, "%.8s %s SCSI Adapter, FW Rev. %.4s, ID=%d, %d ECBs\n", ahb->ha_inq_data->scsi_data.product, (ahb->ha_inq_data->scsi_data.flags & 0x4) ? "Differential" : "Single Ended", ahb->ha_inq_data->scsi_data.revision, ahb->scsi_id, ahb->num_ecbs); /* Restore sense paddr for future CCB clients */ ecb->hecb.sense_ptr = ahbsensepaddr(ahbecbvtop(ahb, ecb)); ahbecbfree(ahb, ecb); /* * Create the device queue for our SIM. */ devq = cam_simq_alloc(ahb->num_ecbs); if (devq == NULL) { mtx_unlock(&ahb->lock); return (ENOMEM); } /* * Construct our SIM entry */ ahb->sim = cam_sim_alloc(ahbaction, ahbpoll, "ahb", ahb, device_get_unit(ahb->dev), &ahb->lock, 2, ahb->num_ecbs, devq); if (ahb->sim == NULL) { cam_simq_free(devq); mtx_unlock(&ahb->lock); return (ENOMEM); } if (xpt_bus_register(ahb->sim, ahb->dev, 0) != CAM_SUCCESS) { cam_sim_free(ahb->sim, /*free_devq*/TRUE); mtx_unlock(&ahb->lock); return (ENXIO); } if (xpt_create_path(&ahb->path, /*periph*/NULL, cam_sim_path(ahb->sim), CAM_TARGET_WILDCARD, CAM_LUN_WILDCARD) != CAM_REQ_CMP) { xpt_bus_deregister(cam_sim_path(ahb->sim)); cam_sim_free(ahb->sim, /*free_devq*/TRUE); mtx_unlock(&ahb->lock); return (ENXIO); } /* * Allow the board to generate interrupts. */ ahb_outb(ahb, INTDEF, ahb_inb(ahb, INTDEF) | INTEN); mtx_unlock(&ahb->lock); return (0); } static void ahbhandleimmed(struct ahb_softc *ahb, u_int32_t mbox, u_int intstat) { struct ccb_hdr *ccb_h; u_int target_id; if (ahb->immed_cmd == 0) { device_printf(ahb->dev, "Immediate Command complete with no " " pending command\n"); return; } target_id = intstat & INTSTAT_TARGET_MASK; ccb_h = LIST_FIRST(&ahb->pending_ccbs); while (ccb_h != NULL) { struct ecb *pending_ecb; union ccb *ccb; pending_ecb = (struct ecb *)ccb_h->ccb_ecb_ptr; ccb = pending_ecb->ccb; ccb_h = LIST_NEXT(ccb_h, sim_links.le); if (ccb->ccb_h.target_id == target_id || target_id == ahb->scsi_id) { callout_stop(&pending_ecb->timer); LIST_REMOVE(&ccb->ccb_h, sim_links.le); if ((ccb->ccb_h.flags & CAM_DIR_MASK) != CAM_DIR_NONE) bus_dmamap_unload(ahb->buffer_dmat, pending_ecb->dmamap); if (pending_ecb == ahb->immed_ecb) ccb->ccb_h.status = CAM_CMD_TIMEOUT|CAM_RELEASE_SIMQ; else if (target_id == ahb->scsi_id) ccb->ccb_h.status = CAM_SCSI_BUS_RESET; else ccb->ccb_h.status = CAM_BDR_SENT; ahbecbfree(ahb, pending_ecb); xpt_done(ccb); } else if (ahb->immed_ecb != NULL) { /* Re-instate timeout */ callout_reset_sbt(&pending_ecb->timer, SBT_1MS * ccb->ccb_h.timeout, 0, ahbtimeout, pending_ecb, 0); } } if (ahb->immed_ecb != NULL) { ahb->immed_ecb = NULL; device_printf(ahb->dev, "No longer in timeout\n"); } else if (target_id == ahb->scsi_id) device_printf(ahb->dev, "SCSI Bus Reset Delivered\n"); else device_printf(ahb->dev, "Bus Device Reset Delivered to target %d\n", target_id); ahb->immed_cmd = 0; } static void ahbcalcresid(struct ahb_softc *ahb, struct ecb *ecb, union ccb *ccb) { if (ecb->status.data_overrun != 0) { /* * Overrun Condition. The hardware doesn't * provide a meaningful byte count in this case * (the residual is always 0). Tell the XPT * layer about the error. */ ccb->ccb_h.status = CAM_DATA_RUN_ERR; } else { ccb->csio.resid = ecb->status.resid_count; if ((ecb->hecb.flag_word1 & FW1_SG_ECB) != 0) { /* * For S/G transfers, the adapter provides a pointer * to the address in the last S/G element used and a * residual for that element. So, we need to sum up * the elements that follow it in order to get a real * residual number. If we have an overrun, the residual * reported will be 0 and we already know that all S/G * segments have been exhausted, so we can skip this * step. */ ahb_sg_t *sg; int num_sg; num_sg = ecb->hecb.data_len / sizeof(ahb_sg_t); /* Find the S/G the adapter was working on */ for (sg = ecb->sg_list; num_sg != 0 && sg->addr != ecb->status.resid_addr; num_sg--, sg++) ; /* Skip it */ num_sg--; sg++; /* Sum the rest */ for (; num_sg != 0; num_sg--, sg++) ccb->csio.resid += sg->len; } /* Underruns are not errors */ ccb->ccb_h.status = CAM_REQ_CMP; } } static void ahbprocesserror(struct ahb_softc *ahb, struct ecb *ecb, union ccb *ccb) { struct hardware_ecb *hecb; struct ecb_status *status; hecb = &ecb->hecb; status = &ecb->status; switch (status->ha_status) { case HS_OK: ccb->csio.scsi_status = status->scsi_status; if (status->scsi_status != 0) { ccb->ccb_h.status = CAM_SCSI_STATUS_ERROR; if (status->sense_stored) { ccb->ccb_h.status |= CAM_AUTOSNS_VALID; ccb->csio.sense_resid = ccb->csio.sense_len - status->sense_len; bcopy(&ecb->sense, &ccb->csio.sense_data, status->sense_len); } } break; case HS_TARGET_NOT_ASSIGNED: ccb->ccb_h.status = CAM_PATH_INVALID; break; case HS_SEL_TIMEOUT: ccb->ccb_h.status = CAM_SEL_TIMEOUT; break; case HS_DATA_RUN_ERR: ahbcalcresid(ahb, ecb, ccb); break; case HS_UNEXPECTED_BUSFREE: ccb->ccb_h.status = CAM_UNEXP_BUSFREE; break; case HS_INVALID_PHASE: ccb->ccb_h.status = CAM_SEQUENCE_FAIL; break; case HS_REQUEST_SENSE_FAILED: ccb->ccb_h.status = CAM_AUTOSENSE_FAIL; break; case HS_TAG_MSG_REJECTED: { struct ccb_trans_settings neg; struct ccb_trans_settings_scsi *scsi = &neg.proto_specific.scsi; xpt_print_path(ccb->ccb_h.path); printf("refuses tagged commands. Performing " "non-tagged I/O\n"); memset(&neg, 0, sizeof (neg)); neg.protocol = PROTO_SCSI; neg.protocol_version = SCSI_REV_2; neg.transport = XPORT_SPI; neg.transport_version = 2; scsi->flags = CTS_SCSI_VALID_TQ; xpt_setup_ccb(&neg.ccb_h, ccb->ccb_h.path, /*priority*/1); xpt_async(AC_TRANSFER_NEG, ccb->ccb_h.path, &neg); ahb->tags_permitted &= ~(0x01 << ccb->ccb_h.target_id); ccb->ccb_h.status = CAM_MSG_REJECT_REC; break; } case HS_FIRMWARE_LOAD_REQ: case HS_HARDWARE_ERR: /* * Tell the system that the Adapter * is no longer functional. */ ccb->ccb_h.status = CAM_NO_HBA; break; case HS_CMD_ABORTED_HOST: case HS_CMD_ABORTED_ADAPTER: case HS_ATN_TARGET_FAILED: case HS_SCSI_RESET_ADAPTER: case HS_SCSI_RESET_INCOMING: ccb->ccb_h.status = CAM_SCSI_BUS_RESET; break; case HS_INVALID_ECB_PARAM: device_printf(ahb->dev, "opcode 0x%02x, flag_word1 0x%02x, flag_word2 0x%02x\n", hecb->opcode, hecb->flag_word1, hecb->flag_word2); ccb->ccb_h.status = CAM_SCSI_BUS_RESET; break; case HS_DUP_TCB_RECEIVED: case HS_INVALID_OPCODE: case HS_INVALID_CMD_LINK: case HS_PROGRAM_CKSUM_ERROR: panic("%s: Can't happen host status %x occurred", device_get_nameunit(ahb->dev), status->ha_status); break; } if (ccb->ccb_h.status != CAM_REQ_CMP) { xpt_freeze_devq(ccb->ccb_h.path, /*count*/1); ccb->ccb_h.status |= CAM_DEV_QFRZN; } } static void ahbdone(struct ahb_softc *ahb, u_int32_t mbox, u_int intstat) { struct ecb *ecb; union ccb *ccb; ecb = ahbecbptov(ahb, mbox); if ((ecb->state & ECB_ACTIVE) == 0) panic("ecb not active"); ccb = ecb->ccb; if (ccb != NULL) { callout_stop(&ecb->timer); LIST_REMOVE(&ccb->ccb_h, sim_links.le); if ((ccb->ccb_h.flags & CAM_DIR_MASK) != CAM_DIR_NONE) { bus_dmasync_op_t op; if ((ccb->ccb_h.flags & CAM_DIR_MASK) == CAM_DIR_IN) op = BUS_DMASYNC_POSTREAD; else op = BUS_DMASYNC_POSTWRITE; bus_dmamap_sync(ahb->buffer_dmat, ecb->dmamap, op); bus_dmamap_unload(ahb->buffer_dmat, ecb->dmamap); } if ((intstat & INTSTAT_MASK) == INTSTAT_ECB_OK) { ccb->ccb_h.status = CAM_REQ_CMP; ccb->csio.resid = 0; } else { ahbprocesserror(ahb, ecb, ccb); } ahbecbfree(ahb, ecb); xpt_done(ccb); } else { /* Non CCB Command */ if ((intstat & INTSTAT_MASK) != INTSTAT_ECB_OK) { device_printf(ahb->dev, "Command 0%x Failed %x:%x:%x\n", ecb->hecb.opcode, *((u_int16_t*)&ecb->status), ecb->status.ha_status, ecb->status.resid_count); } /* Client owns this ECB and will release it. */ } } /* * Catch an interrupt from the adaptor */ static void ahbintr(void *arg) { struct ahb_softc *ahb; ahb = arg; mtx_lock(&ahb->lock); ahbintr_locked(ahb); mtx_unlock(&ahb->lock); } static void ahbintr_locked(struct ahb_softc *ahb) { u_int intstat; u_int32_t mbox; while (ahb_inb(ahb, HOSTSTAT) & HOSTSTAT_INTPEND) { /* * Fetch information about this interrupt. */ intstat = ahb_inb(ahb, INTSTAT); mbox = ahb_inl(ahb, MBOXIN0); /* * Reset interrupt latch. */ ahb_outb(ahb, CONTROL, CNTRL_CLRINT); /* * Process the completed operation */ switch (intstat & INTSTAT_MASK) { case INTSTAT_ECB_OK: case INTSTAT_ECB_CMPWRETRY: case INTSTAT_ECB_CMPWERR: ahbdone(ahb, mbox, intstat); break; case INTSTAT_AEN_OCCURED: if ((intstat & INTSTAT_TARGET_MASK) == ahb->scsi_id) { /* Bus Reset */ xpt_print_path(ahb->path); switch (mbox) { case HS_SCSI_RESET_ADAPTER: printf("Host Adapter Initiated " "Bus Reset occurred\n"); break; case HS_SCSI_RESET_INCOMING: printf("Bus Reset Initiated " "by another device occurred\n"); break; } /* Notify the XPT */ xpt_async(AC_BUS_RESET, ahb->path, NULL); break; } printf("Unsupported initiator selection AEN occurred\n"); break; case INTSTAT_IMMED_OK: case INTSTAT_IMMED_ERR: ahbhandleimmed(ahb, mbox, intstat); break; case INTSTAT_HW_ERR: panic("Unrecoverable hardware Error Occurred\n"); } } } static void ahbexecuteecb(void *arg, bus_dma_segment_t *dm_segs, int nseg, int error) { struct ecb *ecb; union ccb *ccb; struct ahb_softc *ahb; u_int32_t ecb_paddr; ecb = (struct ecb *)arg; ccb = ecb->ccb; ahb = (struct ahb_softc *)ccb->ccb_h.ccb_ahb_ptr; mtx_assert(&ahb->lock, MA_OWNED); if (error != 0) { if (error != EFBIG) device_printf(ahb->dev, "Unexepected error 0x%x returned from " "bus_dmamap_load\n", error); if (ccb->ccb_h.status == CAM_REQ_INPROG) { xpt_freeze_devq(ccb->ccb_h.path, /*count*/1); ccb->ccb_h.status = CAM_REQ_TOO_BIG|CAM_DEV_QFRZN; } ahbecbfree(ahb, ecb); xpt_done(ccb); return; } ecb_paddr = ahbecbvtop(ahb, ecb); if (nseg != 0) { ahb_sg_t *sg; bus_dma_segment_t *end_seg; bus_dmasync_op_t op; end_seg = dm_segs + nseg; /* Copy the segments into our SG list */ sg = ecb->sg_list; while (dm_segs < end_seg) { sg->addr = dm_segs->ds_addr; sg->len = dm_segs->ds_len; sg++; dm_segs++; } if (nseg > 1) { ecb->hecb.flag_word1 |= FW1_SG_ECB; ecb->hecb.data_ptr = ahbsgpaddr(ecb_paddr); ecb->hecb.data_len = sizeof(ahb_sg_t) * nseg; } else { ecb->hecb.data_ptr = ecb->sg_list->addr; ecb->hecb.data_len = ecb->sg_list->len; } if ((ccb->ccb_h.flags & CAM_DIR_MASK) == CAM_DIR_IN) { /* ecb->hecb.flag_word2 |= FW2_DATA_DIR_IN; */ op = BUS_DMASYNC_PREREAD; } else { op = BUS_DMASYNC_PREWRITE; } /* ecb->hecb.flag_word2 |= FW2_CHECK_DATA_DIR; */ bus_dmamap_sync(ahb->buffer_dmat, ecb->dmamap, op); } else { ecb->hecb.data_ptr = 0; ecb->hecb.data_len = 0; } /* * Last time we need to check if this CCB needs to * be aborted. */ if (ccb->ccb_h.status != CAM_REQ_INPROG) { if (nseg != 0) bus_dmamap_unload(ahb->buffer_dmat, ecb->dmamap); ahbecbfree(ahb, ecb); xpt_done(ccb); return; } ecb->state = ECB_ACTIVE; ccb->ccb_h.status |= CAM_SIM_QUEUED; LIST_INSERT_HEAD(&ahb->pending_ccbs, &ccb->ccb_h, sim_links.le); /* Tell the adapter about this command */ ahbqueuembox(ahb, ecb_paddr, ATTN_STARTECB|ccb->ccb_h.target_id); callout_reset_sbt(&ecb->timer, SBT_1MS * ccb->ccb_h.timeout, 0, ahbtimeout, ecb, 0); } static void ahbaction(struct cam_sim *sim, union ccb *ccb) { struct ahb_softc *ahb; CAM_DEBUG(ccb->ccb_h.path, CAM_DEBUG_TRACE, ("ahbaction\n")); ahb = (struct ahb_softc *)cam_sim_softc(sim); mtx_assert(&ahb->lock, MA_OWNED); switch (ccb->ccb_h.func_code) { /* Common cases first */ case XPT_SCSI_IO: /* Execute the requested I/O operation */ { struct ecb *ecb; struct hardware_ecb *hecb; int error; /* * get an ecb to use. */ if ((ecb = ahbecbget(ahb)) == NULL) { /* Should never occur */ panic("Failed to get an ecb"); } /* * So we can find the ECB when an abort is requested */ ecb->ccb = ccb; ccb->ccb_h.ccb_ecb_ptr = ecb; ccb->ccb_h.ccb_ahb_ptr = ahb; /* * Put all the arguments for the xfer in the ecb */ hecb = &ecb->hecb; hecb->opcode = ECBOP_INITIATOR_SCSI_CMD; hecb->flag_word1 = FW1_AUTO_REQUEST_SENSE | FW1_ERR_STATUS_BLK_ONLY; hecb->flag_word2 = ccb->ccb_h.target_lun | FW2_NO_RETRY_ON_BUSY; if ((ccb->ccb_h.flags & CAM_TAG_ACTION_VALID) != 0) { hecb->flag_word2 |= FW2_TAG_ENB | ((ccb->csio.tag_action & 0x3) << FW2_TAG_TYPE_SHIFT); } if ((ccb->ccb_h.flags & CAM_DIS_DISCONNECT) != 0) hecb->flag_word2 |= FW2_DISABLE_DISC; hecb->sense_len = ccb->csio.sense_len; hecb->cdb_len = ccb->csio.cdb_len; if ((ccb->ccb_h.flags & CAM_CDB_POINTER) != 0) { if ((ccb->ccb_h.flags & CAM_CDB_PHYS) == 0) { bcopy(ccb->csio.cdb_io.cdb_ptr, hecb->cdb, hecb->cdb_len); } else { /* I guess I could map it in... */ ccb->ccb_h.status = CAM_REQ_INVALID; ahbecbfree(ahb, ecb); xpt_done(ccb); return; } } else { bcopy(ccb->csio.cdb_io.cdb_bytes, hecb->cdb, hecb->cdb_len); } error = bus_dmamap_load_ccb( ahb->buffer_dmat, ecb->dmamap, ccb, ahbexecuteecb, ecb, /*flags*/0); if (error == EINPROGRESS) { /* * So as to maintain ordering, freeze the controller * queue until our mapping is returned. */ xpt_freeze_simq(ahb->sim, 1); ccb->ccb_h.status |= CAM_RELEASE_SIMQ; } break; } case XPT_EN_LUN: /* Enable LUN as a target */ case XPT_TARGET_IO: /* Execute target I/O request */ case XPT_ACCEPT_TARGET_IO: /* Accept Host Target Mode CDB */ case XPT_CONT_TARGET_IO: /* Continue Host Target I/O Connection*/ case XPT_ABORT: /* Abort the specified CCB */ /* XXX Implement */ ccb->ccb_h.status = CAM_REQ_INVALID; xpt_done(ccb); break; case XPT_SET_TRAN_SETTINGS: { ccb->ccb_h.status = CAM_FUNC_NOTAVAIL; xpt_done(ccb); break; } case XPT_GET_TRAN_SETTINGS: /* Get default/user set transfer settings for the target */ { struct ccb_trans_settings *cts = &ccb->cts; u_int target_mask = 0x01 << ccb->ccb_h.target_id; struct ccb_trans_settings_scsi *scsi = &cts->proto_specific.scsi; struct ccb_trans_settings_spi *spi = &cts->xport_specific.spi; if (cts->type == CTS_TYPE_USER_SETTINGS) { cts->protocol = PROTO_SCSI; cts->protocol_version = SCSI_REV_2; cts->transport = XPORT_SPI; cts->transport_version = 2; scsi->flags &= ~CTS_SCSI_FLAGS_TAG_ENB; spi->flags &= ~CTS_SPI_FLAGS_DISC_ENB; if ((ahb->disc_permitted & target_mask) != 0) spi->flags |= CTS_SPI_FLAGS_DISC_ENB; if ((ahb->tags_permitted & target_mask) != 0) scsi->flags |= CTS_SCSI_FLAGS_TAG_ENB; spi->bus_width = MSG_EXT_WDTR_BUS_8_BIT; spi->sync_period = 25; /* 10MHz */ if (spi->sync_period != 0) spi->sync_offset = 15; spi->valid = CTS_SPI_VALID_SYNC_RATE | CTS_SPI_VALID_SYNC_OFFSET | CTS_SPI_VALID_BUS_WIDTH | CTS_SPI_VALID_DISC; scsi->valid = CTS_SCSI_VALID_TQ; ccb->ccb_h.status = CAM_REQ_CMP; } else { ccb->ccb_h.status = CAM_FUNC_NOTAVAIL; } xpt_done(ccb); break; } case XPT_RESET_DEV: /* Bus Device Reset the specified SCSI device */ { int i; ahb->immed_cmd = IMMED_RESET; ahbqueuembox(ahb, IMMED_RESET, ATTN_IMMED|ccb->ccb_h.target_id); /* Poll for interrupt completion */ for (i = 1000; ahb->immed_cmd != 0 && i != 0; i--) { DELAY(1000); ahbintr_locked(cam_sim_softc(sim)); } break; } case XPT_CALC_GEOMETRY: { cam_calc_geometry(&ccb->ccg, ahb->extended_trans); xpt_done(ccb); break; } case XPT_RESET_BUS: /* Reset the specified SCSI bus */ { int i; ahb->immed_cmd = IMMED_RESET; ahbqueuembox(ahb, IMMED_RESET, ATTN_IMMED|ahb->scsi_id); /* Poll for interrupt completion */ for (i = 1000; ahb->immed_cmd != 0 && i != 0; i--) DELAY(1000); ccb->ccb_h.status = CAM_REQ_CMP; xpt_done(ccb); break; } case XPT_TERM_IO: /* Terminate the I/O process */ /* XXX Implement */ ccb->ccb_h.status = CAM_REQ_INVALID; xpt_done(ccb); break; case XPT_PATH_INQ: /* Path routing inquiry */ { struct ccb_pathinq *cpi = &ccb->cpi; cpi->version_num = 1; /* XXX??? */ cpi->hba_inquiry = PI_SDTR_ABLE|PI_TAG_ABLE; cpi->target_sprt = 0; cpi->hba_misc = 0; cpi->hba_eng_cnt = 0; cpi->max_target = 7; cpi->max_lun = 7; cpi->initiator_id = ahb->scsi_id; cpi->bus_id = cam_sim_bus(sim); cpi->base_transfer_speed = 3300; strlcpy(cpi->sim_vid, "FreeBSD", SIM_IDLEN); strlcpy(cpi->hba_vid, "Adaptec", HBA_IDLEN); strlcpy(cpi->dev_name, cam_sim_name(sim), DEV_IDLEN); cpi->unit_number = cam_sim_unit(sim); - cpi->transport = XPORT_SPI; - cpi->transport_version = 2; - cpi->protocol = PROTO_SCSI; - cpi->protocol_version = SCSI_REV_2; + cpi->transport = XPORT_SPI; + cpi->transport_version = 2; + cpi->protocol = PROTO_SCSI; + cpi->protocol_version = SCSI_REV_2; cpi->ccb_h.status = CAM_REQ_CMP; xpt_done(ccb); break; } #if 0 /* Need these??? */ case XPT_IMMED_NOTIFY: /* Notify Host Target driver of event */ case XPT_NOTIFY_ACK: /* Acknowledgement of event */ #endif default: ccb->ccb_h.status = CAM_REQ_INVALID; xpt_done(ccb); break; } } static void ahbpoll(struct cam_sim *sim) { ahbintr(cam_sim_softc(sim)); } static void ahbtimeout(void *arg) { struct ecb *ecb; union ccb *ccb; struct ahb_softc *ahb; ecb = (struct ecb *)arg; ccb = ecb->ccb; ahb = (struct ahb_softc *)ccb->ccb_h.ccb_ahb_ptr; mtx_assert(&ahb->lock, MA_OWNED); xpt_print_path(ccb->ccb_h.path); printf("ECB %p - timed out\n", (void *)ecb); if ((ecb->state & ECB_ACTIVE) == 0) { xpt_print_path(ccb->ccb_h.path); printf("ECB %p - timed out ECB already completed\n", (void *)ecb); return; } /* * In order to simplify the recovery process, we ask the XPT * layer to halt the queue of new transactions and we traverse * the list of pending CCBs and remove their timeouts. This * means that the driver attempts to clear only one error * condition at a time. In general, timeouts that occur * close together are related anyway, so there is no benefit * in attempting to handle errors in parallel. Timeouts will * be reinstated when the recovery process ends. */ if ((ecb->state & ECB_DEVICE_RESET) == 0) { struct ccb_hdr *ccb_h; if ((ecb->state & ECB_RELEASE_SIMQ) == 0) { xpt_freeze_simq(ahb->sim, /*count*/1); ecb->state |= ECB_RELEASE_SIMQ; } LIST_FOREACH(ccb_h, &ahb->pending_ccbs, sim_links.le) { struct ecb *pending_ecb; pending_ecb = (struct ecb *)ccb_h->ccb_ecb_ptr; callout_stop(&pending_ecb->timer); } /* Store for our interrupt handler */ ahb->immed_ecb = ecb; /* * Send a Bus Device Reset message: * The target that is holding up the bus may not * be the same as the one that triggered this timeout * (different commands have different timeout lengths), * but we have no way of determining this from our * timeout handler. Our strategy here is to queue a * BDR message to the target of the timed out command. * If this fails, we'll get another timeout 2 seconds * later which will attempt a bus reset. */ xpt_print_path(ccb->ccb_h.path); printf("Queuing BDR\n"); ecb->state |= ECB_DEVICE_RESET; callout_reset(&ecb->timer, 2 * hz, ahbtimeout, ecb); ahb->immed_cmd = IMMED_RESET; ahbqueuembox(ahb, IMMED_RESET, ATTN_IMMED|ccb->ccb_h.target_id); } else if ((ecb->state & ECB_SCSIBUS_RESET) != 0) { /* * Try a SCSI bus reset. We do this only if we * have already attempted to clear the condition with a BDR. */ xpt_print_path(ccb->ccb_h.path); printf("Attempting SCSI Bus reset\n"); ecb->state |= ECB_SCSIBUS_RESET; callout_reset(&ecb->timer, 2 * hz, ahbtimeout, ecb); ahb->immed_cmd = IMMED_RESET; ahbqueuembox(ahb, IMMED_RESET, ATTN_IMMED|ahb->scsi_id); } else { /* Bring out the hammer... */ ahbreset(ahb); /* Simulate the reset complete interrupt */ ahbhandleimmed(ahb, 0, ahb->scsi_id|INTSTAT_IMMED_OK); } } static device_method_t ahb_eisa_methods[] = { /* Device interface */ DEVMETHOD(device_probe, ahbprobe), DEVMETHOD(device_attach, ahbattach), { 0, 0 } }; static driver_t ahb_eisa_driver = { "ahb", ahb_eisa_methods, sizeof(struct ahb_softc), }; static devclass_t ahb_devclass; DRIVER_MODULE(ahb, eisa, ahb_eisa_driver, ahb_devclass, 0, 0); MODULE_DEPEND(ahb, eisa, 1, 1, 1); MODULE_DEPEND(ahb, cam, 1, 1, 1); Index: projects/clang400-import/sys/dev/aic/aic.c =================================================================== --- projects/clang400-import/sys/dev/aic/aic.c (revision 311459) +++ projects/clang400-import/sys/dev/aic/aic.c (revision 311460) @@ -1,1598 +1,1598 @@ /*- * Copyright (c) 1999 Luoqi 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("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include static void aic_action(struct cam_sim *sim, union ccb *ccb); static void aic_execute_scb(void *arg, bus_dma_segment_t *dm_segs, int nseg, int error); static void aic_intr_locked(struct aic_softc *aic); static void aic_start(struct aic_softc *aic); static void aic_select(struct aic_softc *aic); static void aic_selected(struct aic_softc *aic); static void aic_reselected(struct aic_softc *aic); static void aic_reconnect(struct aic_softc *aic, int tag); static void aic_cmd(struct aic_softc *aic); static void aic_msgin(struct aic_softc *aic); static void aic_handle_msgin(struct aic_softc *aic); static void aic_msgout(struct aic_softc *aic); static void aic_datain(struct aic_softc *aic); static void aic_dataout(struct aic_softc *aic); static void aic_done(struct aic_softc *aic, struct aic_scb *scb); static void aic_poll(struct cam_sim *sim); static void aic_timeout(void *arg); static void aic_scsi_reset(struct aic_softc *aic); static void aic_chip_reset(struct aic_softc *aic); static void aic_reset(struct aic_softc *aic, int initiate_reset); devclass_t aic_devclass; static struct aic_scb * aic_get_scb(struct aic_softc *aic) { struct aic_scb *scb; if (!dumping) mtx_assert(&aic->lock, MA_OWNED); if ((scb = SLIST_FIRST(&aic->free_scbs)) != NULL) SLIST_REMOVE_HEAD(&aic->free_scbs, link); return (scb); } static void aic_free_scb(struct aic_softc *aic, struct aic_scb *scb) { if (!dumping) mtx_assert(&aic->lock, MA_OWNED); if ((aic->flags & AIC_RESOURCE_SHORTAGE) != 0 && (scb->ccb->ccb_h.status & CAM_RELEASE_SIMQ) == 0) { scb->ccb->ccb_h.status |= CAM_RELEASE_SIMQ; aic->flags &= ~AIC_RESOURCE_SHORTAGE; } scb->flags = 0; SLIST_INSERT_HEAD(&aic->free_scbs, scb, link); } static void aic_action(struct cam_sim *sim, union ccb *ccb) { struct aic_softc *aic; CAM_DEBUG(ccb->ccb_h.path, CAM_DEBUG_TRACE, ("aic_action\n")); aic = (struct aic_softc *)cam_sim_softc(sim); mtx_assert(&aic->lock, MA_OWNED); switch (ccb->ccb_h.func_code) { case XPT_SCSI_IO: /* Execute the requested I/O operation */ case XPT_RESET_DEV: /* Bus Device Reset the specified SCSI device */ { struct aic_scb *scb; if ((scb = aic_get_scb(aic)) == NULL) { aic->flags |= AIC_RESOURCE_SHORTAGE; xpt_freeze_simq(aic->sim, /*count*/1); ccb->ccb_h.status = CAM_REQUEUE_REQ; xpt_done(ccb); return; } scb->ccb = ccb; ccb->ccb_h.ccb_scb_ptr = scb; ccb->ccb_h.ccb_aic_ptr = aic; scb->target = ccb->ccb_h.target_id; scb->lun = ccb->ccb_h.target_lun; if (ccb->ccb_h.func_code == XPT_SCSI_IO) { scb->cmd_len = ccb->csio.cdb_len; if (ccb->ccb_h.flags & CAM_CDB_POINTER) { if (ccb->ccb_h.flags & CAM_CDB_PHYS) { ccb->ccb_h.status = CAM_REQ_INVALID; aic_free_scb(aic, scb); xpt_done(ccb); return; } scb->cmd_ptr = ccb->csio.cdb_io.cdb_ptr; } else { scb->cmd_ptr = ccb->csio.cdb_io.cdb_bytes; } if ((ccb->ccb_h.flags & CAM_DIR_MASK) != CAM_DIR_NONE) { if ((ccb->ccb_h.flags & CAM_DATA_MASK) != CAM_DATA_VADDR) { ccb->ccb_h.status = CAM_REQ_INVALID; aic_free_scb(aic, scb); xpt_done(ccb); return; } scb->data_ptr = ccb->csio.data_ptr; scb->data_len = ccb->csio.dxfer_len; } else { scb->data_ptr = NULL; scb->data_len = 0; } aic_execute_scb(scb, NULL, 0, 0); } else { scb->flags |= SCB_DEVICE_RESET; aic_execute_scb(scb, NULL, 0, 0); } break; } case XPT_SET_TRAN_SETTINGS: { struct ccb_trans_settings *cts = &ccb->cts; struct aic_tinfo *ti = &aic->tinfo[ccb->ccb_h.target_id]; struct ccb_trans_settings_scsi *scsi = &cts->proto_specific.scsi; struct ccb_trans_settings_spi *spi = &cts->xport_specific.spi; if ((spi->valid & CTS_SPI_VALID_DISC) != 0 && (aic->flags & AIC_DISC_ENABLE) != 0) { if ((spi->flags & CTS_SPI_FLAGS_DISC_ENB) != 0) ti->flags |= TINFO_DISC_ENB; else ti->flags &= ~TINFO_DISC_ENB; } if ((scsi->valid & CTS_SCSI_VALID_TQ) != 0) { if ((scsi->flags & CTS_SCSI_FLAGS_TAG_ENB) != 0) ti->flags |= TINFO_TAG_ENB; else ti->flags &= ~TINFO_TAG_ENB; } if ((spi->valid & CTS_SPI_VALID_SYNC_RATE) != 0) { ti->goal.period = spi->sync_period; if (ti->goal.period > aic->min_period) { ti->goal.period = 0; ti->goal.offset = 0; } else if (ti->goal.period < aic->max_period) ti->goal.period = aic->max_period; } if ((spi->valid & CTS_SPI_VALID_SYNC_OFFSET) != 0) { ti->goal.offset = spi->sync_offset; if (ti->goal.offset == 0) ti->goal.period = 0; else if (ti->goal.offset > AIC_SYNC_OFFSET) ti->goal.offset = AIC_SYNC_OFFSET; } if ((ti->goal.period != ti->current.period) || (ti->goal.offset != ti->current.offset)) ti->flags |= TINFO_SDTR_NEGO; ccb->ccb_h.status = CAM_REQ_CMP; xpt_done(ccb); break; } case XPT_GET_TRAN_SETTINGS: { struct ccb_trans_settings *cts = &ccb->cts; struct aic_tinfo *ti = &aic->tinfo[ccb->ccb_h.target_id]; struct ccb_trans_settings_scsi *scsi = &cts->proto_specific.scsi; struct ccb_trans_settings_spi *spi = &cts->xport_specific.spi; cts->protocol = PROTO_SCSI; cts->protocol_version = SCSI_REV_2; cts->transport = XPORT_SPI; cts->transport_version = 2; scsi->flags &= ~CTS_SCSI_FLAGS_TAG_ENB; spi->flags &= ~CTS_SPI_FLAGS_DISC_ENB; if ((ti->flags & TINFO_DISC_ENB) != 0) spi->flags |= CTS_SPI_FLAGS_DISC_ENB; if ((ti->flags & TINFO_TAG_ENB) != 0) scsi->flags |= CTS_SCSI_FLAGS_TAG_ENB; if (cts->type == CTS_TYPE_CURRENT_SETTINGS) { spi->sync_period = ti->current.period; spi->sync_offset = ti->current.offset; } else { spi->sync_period = ti->user.period; spi->sync_offset = ti->user.offset; } spi->bus_width = MSG_EXT_WDTR_BUS_8_BIT; spi->valid = CTS_SPI_VALID_SYNC_RATE | CTS_SPI_VALID_SYNC_OFFSET | CTS_SPI_VALID_BUS_WIDTH | CTS_SPI_VALID_DISC; scsi->valid = CTS_SCSI_VALID_TQ; ccb->ccb_h.status = CAM_REQ_CMP; xpt_done(ccb); break; } case XPT_CALC_GEOMETRY: { cam_calc_geometry(&ccb->ccg, /*extended*/1); xpt_done(ccb); break; } case XPT_RESET_BUS: /* Reset the specified SCSI bus */ aic_reset(aic, /*initiate_reset*/TRUE); ccb->ccb_h.status = CAM_REQ_CMP; xpt_done(ccb); break; case XPT_PATH_INQ: /* Path routing inquiry */ { struct ccb_pathinq *cpi = &ccb->cpi; cpi->version_num = 1; /* XXX??? */ cpi->hba_inquiry = PI_SDTR_ABLE | PI_TAG_ABLE; cpi->target_sprt = 0; cpi->hba_misc = 0; cpi->hba_eng_cnt = 0; cpi->max_target = 7; cpi->max_lun = 7; cpi->initiator_id = aic->initiator; cpi->bus_id = cam_sim_bus(sim); - cpi->base_transfer_speed = 3300; + cpi->base_transfer_speed = 3300; strlcpy(cpi->sim_vid, "FreeBSD", SIM_IDLEN); strlcpy(cpi->hba_vid, "Adaptec", HBA_IDLEN); strlcpy(cpi->dev_name, cam_sim_name(sim), DEV_IDLEN); cpi->unit_number = cam_sim_unit(sim); cpi->transport = XPORT_SPI; cpi->transport_version = 2; cpi->protocol = PROTO_SCSI; cpi->protocol_version = SCSI_REV_2; cpi->ccb_h.status = CAM_REQ_CMP; xpt_done(ccb); break; } default: ccb->ccb_h.status = CAM_REQ_INVALID; xpt_done(ccb); break; } } static void aic_execute_scb(void *arg, bus_dma_segment_t *dm_segs, int nseg, int error) { struct aic_scb *scb = (struct aic_scb *)arg; union ccb *ccb = scb->ccb; struct aic_softc *aic = (struct aic_softc *)ccb->ccb_h.ccb_aic_ptr; if (!dumping) mtx_assert(&aic->lock, MA_OWNED); if (ccb->ccb_h.status != CAM_REQ_INPROG) { aic_free_scb(aic, scb); xpt_done(ccb); return; } scb->flags |= SCB_ACTIVE; ccb->ccb_h.status |= CAM_SIM_QUEUED; TAILQ_INSERT_TAIL(&aic->pending_ccbs, &ccb->ccb_h, sim_links.tqe); callout_reset_sbt(&scb->timer, SBT_1MS * ccb->ccb_h.timeout, 0, aic_timeout, scb, 0); aic_start(aic); } /* * Start another command if the controller is not busy. */ static void aic_start(struct aic_softc *aic) { struct ccb_hdr *ccb_h; struct aic_tinfo *ti; if (aic->state != AIC_IDLE) return; TAILQ_FOREACH(ccb_h, &aic->pending_ccbs, sim_links.tqe) { ti = &aic->tinfo[ccb_h->target_id]; if ((ti->lubusy & (1 << ccb_h->target_lun)) == 0) { TAILQ_REMOVE(&aic->pending_ccbs, ccb_h, sim_links.tqe); aic->nexus = (struct aic_scb *)ccb_h->ccb_scb_ptr; aic_select(aic); return; } } CAM_DEBUG_PRINT(CAM_DEBUG_TRACE, ("aic_start: idle\n")); aic_outb(aic, SIMODE0, ENSELDI); aic_outb(aic, SIMODE1, ENSCSIRST); aic_outb(aic, SCSISEQ, ENRESELI); } /* * Start a selection. */ static void aic_select(struct aic_softc *aic) { struct aic_scb *scb = aic->nexus; CAM_DEBUG(scb->ccb->ccb_h.path, CAM_DEBUG_TRACE, ("aic_select - ccb %p\n", scb->ccb)); aic->state = AIC_SELECTING; aic_outb(aic, DMACNTRL1, 0); aic_outb(aic, SCSIID, aic->initiator << OID_S | scb->target); aic_outb(aic, SXFRCTL1, STIMO_256ms | ENSTIMER | (aic->flags & AIC_PARITY_ENABLE ? ENSPCHK : 0)); aic_outb(aic, SIMODE0, ENSELDI|ENSELDO); aic_outb(aic, SIMODE1, ENSCSIRST|ENSELTIMO); aic_outb(aic, SCSISEQ, ENRESELI|ENSELO|ENAUTOATNO); } /* * We have successfully selected a target, prepare for the information * transfer phases. */ static void aic_selected(struct aic_softc *aic) { struct aic_scb *scb = aic->nexus; union ccb *ccb = scb->ccb; struct aic_tinfo *ti = &aic->tinfo[scb->target]; CAM_DEBUG(ccb->ccb_h.path, CAM_DEBUG_TRACE, ("aic_selected - ccb %p\n", ccb)); aic->state = AIC_HASNEXUS; if (scb->flags & SCB_DEVICE_RESET) { aic->msg_buf[0] = MSG_BUS_DEV_RESET; aic->msg_len = 1; aic->msg_outq = AIC_MSG_MSGBUF; } else { aic->msg_outq = AIC_MSG_IDENTIFY; if ((ti->flags & TINFO_TAG_ENB) != 0 && (ccb->ccb_h.flags & CAM_TAG_ACTION_VALID) != 0) aic->msg_outq |= AIC_MSG_TAG_Q; else ti->lubusy |= 1 << scb->lun; if ((ti->flags & TINFO_SDTR_NEGO) != 0) aic->msg_outq |= AIC_MSG_SDTR; } aic_outb(aic, CLRSINT0, CLRSELDO); aic_outb(aic, CLRSINT1, CLRBUSFREE); aic_outb(aic, SCSISEQ, ENAUTOATNP); aic_outb(aic, SIMODE0, 0); aic_outb(aic, SIMODE1, ENSCSIRST|ENBUSFREE|ENREQINIT); aic_outb(aic, SCSIRATE, ti->scsirate); } /* * We are re-selected by a target, save the target id and wait for the * target to further identify itself. */ static void aic_reselected(struct aic_softc *aic) { u_int8_t selid; CAM_DEBUG_PRINT(CAM_DEBUG_TRACE, ("aic_reselected\n")); /* * If we have started a selection, it must have lost out in * the arbitration, put the command back to the pending queue. */ if (aic->nexus) { TAILQ_INSERT_HEAD(&aic->pending_ccbs, &aic->nexus->ccb->ccb_h, sim_links.tqe); aic->nexus = NULL; } selid = aic_inb(aic, SELID) & ~(1 << aic->initiator); if (selid & (selid - 1)) { /* this should never have happened */ printf("aic_reselected: invalid selid %x\n", selid); aic_reset(aic, /*initiate_reset*/TRUE); return; } aic->state = AIC_RESELECTED; aic->target = ffs(selid) - 1; aic->lun = -1; aic_outb(aic, CLRSINT0, CLRSELDI); aic_outb(aic, CLRSINT1, CLRBUSFREE); aic_outb(aic, SIMODE0, 0); aic_outb(aic, SIMODE1, ENSCSIRST|ENBUSFREE|ENREQINIT); aic_outb(aic, SCSISEQ, ENAUTOATNP); aic_outb(aic, SCSIRATE, aic->tinfo[aic->target].scsirate); } /* * Raise ATNO to signal the target that we have a message for it. */ static __inline void aic_sched_msgout(struct aic_softc *aic, u_int8_t msg) { if (msg) { aic->msg_buf[0] = msg; aic->msg_len = 1; } aic->msg_outq |= AIC_MSG_MSGBUF; aic_outb(aic, SCSISIGO, aic_inb(aic, SCSISIGI) | ATNO); } /* * Wait for SPIORDY (SCSI PIO ready) flag, or a phase change. */ static __inline int aic_spiordy(struct aic_softc *aic) { while (!(aic_inb(aic, DMASTAT) & INTSTAT) && !(aic_inb(aic, SSTAT0) & SPIORDY)) ; return !(aic_inb(aic, DMASTAT) & INTSTAT); } /* * Reestablish a disconnected nexus. */ static void aic_reconnect(struct aic_softc *aic, int tag) { struct aic_scb *scb; struct ccb_hdr *ccb_h; CAM_DEBUG_PRINT(CAM_DEBUG_TRACE, ("aic_reconnect\n")); /* Find the nexus */ scb = NULL; TAILQ_FOREACH(ccb_h, &aic->nexus_ccbs, sim_links.tqe) { scb = (struct aic_scb *)ccb_h->ccb_scb_ptr; if (scb->target == aic->target && scb->lun == aic->lun && (tag == -1 || scb->tag == tag)) break; } /* ABORT if nothing is found */ if (!ccb_h) { if (tag == -1) aic_sched_msgout(aic, MSG_ABORT); else aic_sched_msgout(aic, MSG_ABORT_TAG); xpt_async(AC_UNSOL_RESEL, aic->path, NULL); return; } /* Reestablish the nexus */ TAILQ_REMOVE(&aic->nexus_ccbs, ccb_h, sim_links.tqe); aic->nexus = scb; scb->flags &= ~SCB_DISCONNECTED; aic->state = AIC_HASNEXUS; } /* * Read messages. */ static void aic_msgin(struct aic_softc *aic) { int msglen; CAM_DEBUG_PRINT(CAM_DEBUG_TRACE, ("aic_msgin\n")); aic_outb(aic, SIMODE1, ENSCSIRST|ENPHASEMIS|ENBUSFREE); aic_outb(aic, SXFRCTL0, CHEN|SPIOEN); aic->flags &= ~AIC_DROP_MSGIN; aic->msg_len = 0; do { /* * If a parity error is detected, drop the remaining * bytes and inform the target so it could resend * the messages. */ if (aic_inb(aic, SSTAT1) & SCSIPERR) { aic_outb(aic, CLRSINT1, CLRSCSIPERR); aic->flags |= AIC_DROP_MSGIN; aic_sched_msgout(aic, MSG_PARITY_ERROR); } if ((aic->flags & AIC_DROP_MSGIN)) { aic_inb(aic, SCSIDAT); continue; } /* read the message byte without ACKing on it */ aic->msg_buf[aic->msg_len++] = aic_inb(aic, SCSIBUS); if (aic->msg_buf[0] == MSG_EXTENDED) { if (aic->msg_len < 2) { (void) aic_inb(aic, SCSIDAT); continue; } switch (aic->msg_buf[2]) { case MSG_EXT_SDTR: msglen = MSG_EXT_SDTR_LEN; break; case MSG_EXT_WDTR: msglen = MSG_EXT_WDTR_LEN; break; default: msglen = 0; break; } if (aic->msg_buf[1] != msglen) { aic->flags |= AIC_DROP_MSGIN; aic_sched_msgout(aic, MSG_MESSAGE_REJECT); } msglen += 2; } else if (aic->msg_buf[0] >= 0x20 && aic->msg_buf[0] <= 0x2f) msglen = 2; else msglen = 1; /* * If we have a complete message, handle it before the final * ACK (in case we decide to reject the message). */ if (aic->msg_len == msglen) { aic_handle_msgin(aic); aic->msg_len = 0; } /* ACK on the message byte */ (void) aic_inb(aic, SCSIDAT); } while (aic_spiordy(aic)); aic_outb(aic, SXFRCTL0, CHEN); aic_outb(aic, SIMODE1, ENSCSIRST|ENBUSFREE|ENREQINIT); } /* * Handle a message. */ static void aic_handle_msgin(struct aic_softc *aic) { struct aic_scb *scb; struct ccb_hdr *ccb_h; struct aic_tinfo *ti; struct ccb_trans_settings neg; struct ccb_trans_settings_spi *spi = &neg.xport_specific.spi; if (aic->state == AIC_RESELECTED) { if (!MSG_ISIDENTIFY(aic->msg_buf[0])) { aic_sched_msgout(aic, MSG_MESSAGE_REJECT); return; } aic->lun = aic->msg_buf[0] & MSG_IDENTIFY_LUNMASK; if (aic->tinfo[aic->target].lubusy & (1 << aic->lun)) aic_reconnect(aic, -1); else aic->state = AIC_RECONNECTING; return; } if (aic->state == AIC_RECONNECTING) { if (aic->msg_buf[0] != MSG_SIMPLE_Q_TAG) { aic_sched_msgout(aic, MSG_MESSAGE_REJECT); return; } aic_reconnect(aic, aic->msg_buf[1]); return; } switch (aic->msg_buf[0]) { case MSG_CMDCOMPLETE: { struct ccb_scsiio *csio; scb = aic->nexus; ccb_h = &scb->ccb->ccb_h; csio = &scb->ccb->csio; if ((scb->flags & SCB_SENSE) != 0) { /* auto REQUEST SENSE command */ scb->flags &= ~SCB_SENSE; csio->sense_resid = scb->data_len; if (scb->status == SCSI_STATUS_OK) { ccb_h->status |= CAM_SCSI_STATUS_ERROR|CAM_AUTOSNS_VALID; /*scsi_sense_print(csio);*/ } else { ccb_h->status |= CAM_AUTOSENSE_FAIL; printf("ccb %p sense failed %x\n", ccb_h, scb->status); } } else { csio->scsi_status = scb->status; csio->resid = scb->data_len; if (scb->status == SCSI_STATUS_OK) { /* everything goes well */ ccb_h->status |= CAM_REQ_CMP; } else if ((ccb_h->flags & CAM_DIS_AUTOSENSE) == 0 && (csio->scsi_status == SCSI_STATUS_CHECK_COND || csio->scsi_status == SCSI_STATUS_CMD_TERMINATED)) { /* try to retrieve sense information */ scb->flags |= SCB_SENSE; aic->flags |= AIC_BUSFREE_OK; return; } else ccb_h->status |= CAM_SCSI_STATUS_ERROR; } aic_done(aic, scb); aic->flags |= AIC_BUSFREE_OK; break; } case MSG_EXTENDED: switch (aic->msg_buf[2]) { case MSG_EXT_SDTR: scb = aic->nexus; ti = &aic->tinfo[scb->target]; if (ti->flags & TINFO_SDTR_SENT) { ti->current.period = aic->msg_buf[3]; ti->current.offset = aic->msg_buf[4]; } else { ti->current.period = aic->msg_buf[3] = max(ti->goal.period, aic->msg_buf[3]); ti->current.offset = aic->msg_buf[4] = min(ti->goal.offset, aic->msg_buf[4]); /* * The target initiated the negotiation, * send back a response. */ aic_sched_msgout(aic, 0); } ti->flags &= ~(TINFO_SDTR_SENT|TINFO_SDTR_NEGO); ti->scsirate = ti->current.offset ? ti->current.offset | ((ti->current.period * 4 + 49) / 50 - 2) << 4 : 0; aic_outb(aic, SCSIRATE, ti->scsirate); memset(&neg, 0, sizeof (neg)); neg.protocol = PROTO_SCSI; neg.protocol_version = SCSI_REV_2; neg.transport = XPORT_SPI; neg.transport_version = 2; spi->sync_period = ti->goal.period = ti->current.period; spi->sync_offset = ti->goal.offset = ti->current.offset; spi->valid = CTS_SPI_VALID_SYNC_RATE | CTS_SPI_VALID_SYNC_OFFSET; ccb_h = &scb->ccb->ccb_h; xpt_setup_ccb(&neg.ccb_h, ccb_h->path, 1); xpt_async(AC_TRANSFER_NEG, ccb_h->path, &neg); break; case MSG_EXT_WDTR: default: aic_sched_msgout(aic, MSG_MESSAGE_REJECT); break; } break; case MSG_DISCONNECT: scb = aic->nexus; ccb_h = &scb->ccb->ccb_h; TAILQ_INSERT_TAIL(&aic->nexus_ccbs, ccb_h, sim_links.tqe); scb->flags |= SCB_DISCONNECTED; aic->flags |= AIC_BUSFREE_OK; aic->nexus = NULL; CAM_DEBUG(ccb_h->path, CAM_DEBUG_TRACE, ("disconnected\n")); break; case MSG_MESSAGE_REJECT: switch (aic->msg_outq & -aic->msg_outq) { case AIC_MSG_TAG_Q: scb = aic->nexus; ti = &aic->tinfo[scb->target]; ti->flags &= ~TINFO_TAG_ENB; ti->lubusy |= 1 << scb->lun; break; case AIC_MSG_SDTR: scb = aic->nexus; ti = &aic->tinfo[scb->target]; ti->current.period = ti->goal.period = 0; ti->current.offset = ti->goal.offset = 0; ti->flags &= ~(TINFO_SDTR_SENT|TINFO_SDTR_NEGO); ti->scsirate = 0; aic_outb(aic, SCSIRATE, ti->scsirate); memset(&neg, 0, sizeof (neg)); neg.protocol = PROTO_SCSI; neg.protocol_version = SCSI_REV_2; neg.transport = XPORT_SPI; neg.transport_version = 2; spi->sync_period = ti->current.period; spi->sync_offset = ti->current.offset; spi->valid = CTS_SPI_VALID_SYNC_RATE | CTS_SPI_VALID_SYNC_OFFSET; ccb_h = &scb->ccb->ccb_h; xpt_setup_ccb(&neg.ccb_h, ccb_h->path, 1); xpt_async(AC_TRANSFER_NEG, ccb_h->path, &neg); break; default: break; } break; case MSG_SAVEDATAPOINTER: break; case MSG_RESTOREPOINTERS: break; case MSG_NOOP: break; default: aic_sched_msgout(aic, MSG_MESSAGE_REJECT); break; } } /* * Send messages. */ static void aic_msgout(struct aic_softc *aic) { struct aic_scb *scb; union ccb *ccb; struct aic_tinfo *ti; int msgidx = 0; CAM_DEBUG_PRINT(CAM_DEBUG_TRACE, ("aic_msgout\n")); aic_outb(aic, SIMODE1, ENSCSIRST|ENPHASEMIS|ENBUSFREE); aic_outb(aic, SXFRCTL0, CHEN|SPIOEN); /* * If the previous phase is also the message out phase, * we need to retransmit all the messages, probably * because the target has detected a parity error during * the past transmission. */ if (aic->prev_phase == PH_MSGOUT) aic->msg_outq = aic->msg_sent; do { int q = aic->msg_outq; if (msgidx > 0 && msgidx == aic->msg_len) { /* complete message sent, start the next one */ q &= -q; aic->msg_sent |= q; aic->msg_outq ^= q; q = aic->msg_outq; msgidx = 0; } if (msgidx == 0) { /* setup the message */ switch (q & -q) { case AIC_MSG_IDENTIFY: scb = aic->nexus; ccb = scb->ccb; ti = &aic->tinfo[scb->target]; aic->msg_buf[0] = MSG_IDENTIFY(scb->lun, (ti->flags & TINFO_DISC_ENB) && !(ccb->ccb_h.flags & CAM_DIS_DISCONNECT)); aic->msg_len = 1; break; case AIC_MSG_TAG_Q: scb = aic->nexus; ccb = scb->ccb; aic->msg_buf[0] = ccb->csio.tag_action; aic->msg_buf[1] = scb->tag; aic->msg_len = 2; break; case AIC_MSG_SDTR: scb = aic->nexus; ti = &aic->tinfo[scb->target]; aic->msg_buf[0] = MSG_EXTENDED; aic->msg_buf[1] = MSG_EXT_SDTR_LEN; aic->msg_buf[2] = MSG_EXT_SDTR; aic->msg_buf[3] = ti->goal.period; aic->msg_buf[4] = ti->goal.offset; aic->msg_len = MSG_EXT_SDTR_LEN + 2; ti->flags |= TINFO_SDTR_SENT; break; case AIC_MSG_MSGBUF: /* a single message already in the buffer */ if (aic->msg_buf[0] == MSG_BUS_DEV_RESET || aic->msg_buf[0] == MSG_ABORT || aic->msg_buf[0] == MSG_ABORT_TAG) aic->flags |= AIC_BUSFREE_OK; break; } } /* * If this is the last message byte of all messages, * clear ATNO to signal transmission complete. */ if ((q & (q - 1)) == 0 && msgidx == aic->msg_len - 1) aic_outb(aic, CLRSINT1, CLRATNO); /* transmit the message byte */ aic_outb(aic, SCSIDAT, aic->msg_buf[msgidx++]); } while (aic_spiordy(aic)); aic_outb(aic, SXFRCTL0, CHEN); aic_outb(aic, SIMODE1, ENSCSIRST|ENBUSFREE|ENREQINIT); } /* * Read data bytes. */ static void aic_datain(struct aic_softc *aic) { struct aic_scb *scb = aic->nexus; u_int8_t dmastat, dmacntrl0; int n; CAM_DEBUG_PRINT(CAM_DEBUG_TRACE, ("aic_datain\n")); aic_outb(aic, SIMODE1, ENSCSIRST|ENPHASEMIS|ENBUSFREE); aic_outb(aic, SXFRCTL0, SCSIEN|DMAEN|CHEN); dmacntrl0 = ENDMA; if (aic->flags & AIC_DWIO_ENABLE) dmacntrl0 |= DWORDPIO; aic_outb(aic, DMACNTRL0, dmacntrl0); while (scb->data_len > 0) { for (;;) { /* wait for the fifo to fill up or a phase change */ dmastat = aic_inb(aic, DMASTAT); if (dmastat & (INTSTAT|DFIFOFULL)) break; } if (dmastat & DFIFOFULL) { n = FIFOSIZE; } else { /* * No more data, wait for the remaining bytes in * the scsi fifo to be transfer to the host fifo. */ while (!(aic_inb(aic, SSTAT2) & SEMPTY)) ; n = aic_inb(aic, FIFOSTAT); } n = imin(scb->data_len, n); if (aic->flags & AIC_DWIO_ENABLE) { if (n >= 12) { aic_insl(aic, DMADATALONG, scb->data_ptr, n>>2); scb->data_ptr += n & ~3; scb->data_len -= n & ~3; n &= 3; } } else { if (n >= 8) { aic_insw(aic, DMADATA, scb->data_ptr, n >> 1); scb->data_ptr += n & ~1; scb->data_len -= n & ~1; n &= 1; } } if (n) { aic_outb(aic, DMACNTRL0, ENDMA|B8MODE); aic_insb(aic, DMADATA, scb->data_ptr, n); scb->data_ptr += n; scb->data_len -= n; aic_outb(aic, DMACNTRL0, dmacntrl0); } if (dmastat & INTSTAT) break; } aic_outb(aic, SXFRCTL0, CHEN); aic_outb(aic, SIMODE1, ENSCSIRST|ENBUSFREE|ENREQINIT); } /* * Send data bytes. */ static void aic_dataout(struct aic_softc *aic) { struct aic_scb *scb = aic->nexus; u_int8_t dmastat, dmacntrl0, sstat2; int n; CAM_DEBUG_PRINT(CAM_DEBUG_TRACE, ("aic_dataout\n")); aic_outb(aic, SIMODE1, ENSCSIRST|ENPHASEMIS|ENBUSFREE); aic_outb(aic, SXFRCTL0, SCSIEN|DMAEN|CHEN); dmacntrl0 = ENDMA|WRITE; if (aic->flags & AIC_DWIO_ENABLE) dmacntrl0 |= DWORDPIO; aic_outb(aic, DMACNTRL0, dmacntrl0); while (scb->data_len > 0) { for (;;) { /* wait for the fifo to clear up or a phase change */ dmastat = aic_inb(aic, DMASTAT); if (dmastat & (INTSTAT|DFIFOEMP)) break; } if (dmastat & INTSTAT) break; n = imin(scb->data_len, FIFOSIZE); if (aic->flags & AIC_DWIO_ENABLE) { if (n >= 12) { aic_outsl(aic, DMADATALONG, scb->data_ptr,n>>2); scb->data_ptr += n & ~3; scb->data_len -= n & ~3; n &= 3; } } else { if (n >= 8) { aic_outsw(aic, DMADATA, scb->data_ptr, n >> 1); scb->data_ptr += n & ~1; scb->data_len -= n & ~1; n &= 1; } } if (n) { aic_outb(aic, DMACNTRL0, ENDMA|WRITE|B8MODE); aic_outsb(aic, DMADATA, scb->data_ptr, n); scb->data_ptr += n; scb->data_len -= n; aic_outb(aic, DMACNTRL0, dmacntrl0); } } for (;;) { /* wait until all bytes in the fifos are transmitted */ dmastat = aic_inb(aic, DMASTAT); sstat2 = aic_inb(aic, SSTAT2); if ((dmastat & DFIFOEMP) && (sstat2 & SEMPTY)) break; if (dmastat & INTSTAT) { /* adjust for untransmitted bytes */ n = aic_inb(aic, FIFOSTAT) + (sstat2 & 0xf); scb->data_ptr -= n; scb->data_len += n; /* clear the fifo */ aic_outb(aic, SXFRCTL0, CHEN|CLRCH); aic_outb(aic, DMACNTRL0, RSTFIFO); break; } } aic_outb(aic, SXFRCTL0, CHEN); aic_outb(aic, SIMODE1, ENSCSIRST|ENBUSFREE|ENREQINIT); } /* * Send the scsi command. */ static void aic_cmd(struct aic_softc *aic) { struct aic_scb *scb = aic->nexus; struct scsi_request_sense sense_cmd; CAM_DEBUG_PRINT(CAM_DEBUG_TRACE, ("aic_cmd\n")); if (scb->flags & SCB_SENSE) { /* autosense request */ sense_cmd.opcode = REQUEST_SENSE; sense_cmd.byte2 = scb->lun << 5; sense_cmd.length = scb->ccb->csio.sense_len; sense_cmd.control = 0; sense_cmd.unused[0] = 0; sense_cmd.unused[1] = 0; scb->cmd_ptr = (u_int8_t *)&sense_cmd; scb->cmd_len = sizeof(sense_cmd); scb->data_ptr = (u_int8_t *)&scb->ccb->csio.sense_data; scb->data_len = scb->ccb->csio.sense_len; } aic_outb(aic, SIMODE1, ENSCSIRST|ENPHASEMIS|ENBUSFREE); aic_outb(aic, DMACNTRL0, ENDMA|WRITE); aic_outb(aic, SXFRCTL0, SCSIEN|DMAEN|CHEN); aic_outsw(aic, DMADATA, (u_int16_t *)scb->cmd_ptr, scb->cmd_len >> 1); while ((aic_inb(aic, SSTAT2) & SEMPTY) == 0 && (aic_inb(aic, DMASTAT) & INTSTAT) == 0) ; aic_outb(aic, SXFRCTL0, CHEN); aic_outb(aic, SIMODE1, ENSCSIRST|ENBUSFREE|ENREQINIT); } /* * Finish off a command. The caller is responsible to remove the ccb * from any queue. */ static void aic_done(struct aic_softc *aic, struct aic_scb *scb) { union ccb *ccb = scb->ccb; CAM_DEBUG(ccb->ccb_h.path, CAM_DEBUG_TRACE, ("aic_done - ccb %p status %x resid %d\n", ccb, ccb->ccb_h.status, ccb->csio.resid)); callout_stop(&scb->timer); if ((scb->flags & SCB_DEVICE_RESET) != 0 && ccb->ccb_h.func_code != XPT_RESET_DEV) { struct cam_path *path; struct ccb_hdr *ccb_h; cam_status error; error = xpt_create_path(&path, /*periph*/NULL, cam_sim_path(aic->sim), scb->target, CAM_LUN_WILDCARD); if (error == CAM_REQ_CMP) { xpt_async(AC_SENT_BDR, path, NULL); xpt_free_path(path); } ccb_h = TAILQ_FIRST(&aic->pending_ccbs); while (ccb_h != NULL) { struct aic_scb *pending_scb; pending_scb = (struct aic_scb *)ccb_h->ccb_scb_ptr; if (ccb_h->target_id == scb->target) { ccb_h->status |= CAM_BDR_SENT; ccb_h = TAILQ_NEXT(ccb_h, sim_links.tqe); TAILQ_REMOVE(&aic->pending_ccbs, &pending_scb->ccb->ccb_h, sim_links.tqe); aic_done(aic, pending_scb); } else { callout_reset_sbt(&pending_scb->timer, SBT_1MS * ccb_h->timeout, 0, aic_timeout, pending_scb, 0); ccb_h = TAILQ_NEXT(ccb_h, sim_links.tqe); } } ccb_h = TAILQ_FIRST(&aic->nexus_ccbs); while (ccb_h != NULL) { struct aic_scb *nexus_scb; nexus_scb = (struct aic_scb *)ccb_h->ccb_scb_ptr; if (ccb_h->target_id == scb->target) { ccb_h->status |= CAM_BDR_SENT; ccb_h = TAILQ_NEXT(ccb_h, sim_links.tqe); TAILQ_REMOVE(&aic->nexus_ccbs, &nexus_scb->ccb->ccb_h, sim_links.tqe); aic_done(aic, nexus_scb); } else { callout_reset_sbt(&nexus_scb->timer, SBT_1MS * ccb_h->timeout, 0, aic_timeout, nexus_scb, 0); ccb_h = TAILQ_NEXT(ccb_h, sim_links.tqe); } } } if (aic->nexus == scb || scb->flags & SCB_DISCONNECTED) aic->tinfo[scb->target].lubusy &= ~(1 << scb->lun); if (aic->nexus == scb) { aic->nexus = NULL; } aic_free_scb(aic, scb); xpt_done(ccb); } static void aic_poll(struct cam_sim *sim) { aic_intr_locked(cam_sim_softc(sim)); } static void aic_timeout(void *arg) { struct aic_scb *scb = (struct aic_scb *)arg; union ccb *ccb = scb->ccb; struct aic_softc *aic = (struct aic_softc *)ccb->ccb_h.ccb_aic_ptr; mtx_assert(&aic->lock, MA_OWNED); xpt_print_path(ccb->ccb_h.path); printf("ccb %p - timed out", ccb); if (aic->nexus && aic->nexus != scb) printf(", nexus %p", aic->nexus->ccb); printf(", phase 0x%x, state %d\n", aic_inb(aic, SCSISIGI), aic->state); if ((scb->flags & SCB_ACTIVE) == 0) { xpt_print_path(ccb->ccb_h.path); printf("ccb %p - timed out already completed\n", ccb); return; } if ((scb->flags & SCB_DEVICE_RESET) == 0 && aic->nexus == scb) { struct ccb_hdr *ccb_h = &scb->ccb->ccb_h; struct aic_scb *pending_scb; if ((ccb_h->status & CAM_RELEASE_SIMQ) == 0) { xpt_freeze_simq(aic->sim, /*count*/1); ccb_h->status |= CAM_RELEASE_SIMQ; } TAILQ_FOREACH(ccb_h, &aic->pending_ccbs, sim_links.tqe) { pending_scb = ccb_h->ccb_scb_ptr; callout_stop(&pending_scb->timer); } TAILQ_FOREACH(ccb_h, &aic->nexus_ccbs, sim_links.tqe) { pending_scb = ccb_h->ccb_scb_ptr; callout_stop(&pending_scb->timer); } scb->flags |= SCB_DEVICE_RESET; callout_reset(&scb->timer, 5 * hz, aic_timeout, scb); aic_sched_msgout(aic, MSG_BUS_DEV_RESET); } else { if (aic->nexus == scb) { ccb->ccb_h.status |= CAM_CMD_TIMEOUT; aic_done(aic, scb); } aic_reset(aic, /*initiate_reset*/TRUE); } } void aic_intr(void *arg) { struct aic_softc *aic = (struct aic_softc *)arg; mtx_lock(&aic->lock); aic_intr_locked(aic); mtx_unlock(&aic->lock); } void aic_intr_locked(struct aic_softc *aic) { u_int8_t sstat0, sstat1; union ccb *ccb; struct aic_scb *scb; if (!(aic_inb(aic, DMASTAT) & INTSTAT)) return; aic_outb(aic, DMACNTRL0, 0); sstat0 = aic_inb(aic, SSTAT0); sstat1 = aic_inb(aic, SSTAT1); if ((sstat1 & SCSIRSTI) != 0) { /* a device-initiated bus reset */ aic_outb(aic, CLRSINT1, CLRSCSIRSTI); aic_reset(aic, /*initiate_reset*/FALSE); return; } if ((sstat1 & SCSIPERR) != 0) { aic_outb(aic, CLRSINT1, CLRSCSIPERR); aic_sched_msgout(aic, MSG_PARITY_ERROR); aic_outb(aic, DMACNTRL0, INTEN); return; } if (aic_inb(aic, SSTAT4)) { aic_outb(aic, CLRSERR, CLRSYNCERR|CLRFWERR|CLRFRERR); aic_reset(aic, /*initiate_reset*/TRUE); return; } if (aic->state <= AIC_SELECTING) { if ((sstat0 & SELDI) != 0) { aic_reselected(aic); aic_outb(aic, DMACNTRL0, INTEN); return; } if ((sstat0 & SELDO) != 0) { aic_selected(aic); aic_outb(aic, DMACNTRL0, INTEN); return; } if ((sstat1 & SELTO) != 0) { scb = aic->nexus; ccb = scb->ccb; ccb->ccb_h.status = CAM_SEL_TIMEOUT; aic_done(aic, scb); while ((sstat1 & BUSFREE) == 0) sstat1 = aic_inb(aic, SSTAT1); aic->flags |= AIC_BUSFREE_OK; } } if ((sstat1 & BUSFREE) != 0) { aic_outb(aic, SCSISEQ, 0); aic_outb(aic, CLRSINT0, sstat0); aic_outb(aic, CLRSINT1, sstat1); if ((scb = aic->nexus)) { if ((aic->flags & AIC_BUSFREE_OK) == 0) { ccb = scb->ccb; ccb->ccb_h.status = CAM_UNEXP_BUSFREE; aic_done(aic, scb); } else if (scb->flags & SCB_DEVICE_RESET) { ccb = scb->ccb; if (ccb->ccb_h.func_code == XPT_RESET_DEV) { xpt_async(AC_SENT_BDR, ccb->ccb_h.path, NULL); ccb->ccb_h.status |= CAM_REQ_CMP; } else ccb->ccb_h.status |= CAM_CMD_TIMEOUT; aic_done(aic, scb); } else if (scb->flags & SCB_SENSE) { /* autosense request */ aic->flags &= ~AIC_BUSFREE_OK; aic->tinfo[scb->target].lubusy &= ~(1 << scb->lun); aic_select(aic); aic_outb(aic, DMACNTRL0, INTEN); return; } } aic->flags &= ~AIC_BUSFREE_OK; aic->state = AIC_IDLE; aic_start(aic); aic_outb(aic, DMACNTRL0, INTEN); return; } if ((sstat1 & REQINIT) != 0) { u_int8_t phase = aic_inb(aic, SCSISIGI) & PH_MASK; aic_outb(aic, SCSISIGO, phase); aic_outb(aic, CLRSINT1, CLRPHASECHG); switch (phase) { case PH_MSGOUT: aic_msgout(aic); break; case PH_MSGIN: aic_msgin(aic); break; case PH_STAT: scb = aic->nexus; ccb = scb->ccb; aic_outb(aic, DMACNTRL0, 0); aic_outb(aic, SXFRCTL0, CHEN|SPIOEN); scb->status = aic_inb(aic, SCSIDAT); aic_outb(aic, SXFRCTL0, CHEN); break; case PH_CMD: aic_cmd(aic); break; case PH_DATAIN: aic_datain(aic); break; case PH_DATAOUT: aic_dataout(aic); break; } aic->prev_phase = phase; aic_outb(aic, DMACNTRL0, INTEN); return; } printf("aic_intr: unexpected intr sstat0 %x sstat1 %x\n", sstat0, sstat1); aic_outb(aic, DMACNTRL0, INTEN); } /* * Reset ourselves. */ static void aic_chip_reset(struct aic_softc *aic) { /* * Doc. recommends to clear these two registers before * operations commence */ aic_outb(aic, SCSITEST, 0); aic_outb(aic, TEST, 0); /* Reset SCSI-FIFO and abort any transfers */ aic_outb(aic, SXFRCTL0, CHEN|CLRCH|CLRSTCNT); /* Reset HOST-FIFO */ aic_outb(aic, DMACNTRL0, RSTFIFO); aic_outb(aic, DMACNTRL1, 0); /* Disable all selection features */ aic_outb(aic, SCSISEQ, 0); aic_outb(aic, SXFRCTL1, 0); /* Disable interrupts */ aic_outb(aic, SIMODE0, 0); aic_outb(aic, SIMODE1, 0); /* Clear interrupts */ aic_outb(aic, CLRSINT0, 0x7f); aic_outb(aic, CLRSINT1, 0xef); /* Disable synchronous transfers */ aic_outb(aic, SCSIRATE, 0); /* Haven't seen ant errors (yet) */ aic_outb(aic, CLRSERR, 0x07); /* Set our SCSI-ID */ aic_outb(aic, SCSIID, aic->initiator << OID_S); aic_outb(aic, BRSTCNTRL, EISA_BRST_TIM); } /* * Reset the SCSI bus */ static void aic_scsi_reset(struct aic_softc *aic) { aic_outb(aic, SCSISEQ, SCSIRSTO); DELAY(500); aic_outb(aic, SCSISEQ, 0); DELAY(50); } /* * Reset. Abort all pending commands. */ static void aic_reset(struct aic_softc *aic, int initiate_reset) { struct ccb_hdr *ccb_h; CAM_DEBUG_PRINT(CAM_DEBUG_TRACE, ("aic_reset\n")); if (initiate_reset) aic_scsi_reset(aic); aic_chip_reset(aic); xpt_async(AC_BUS_RESET, aic->path, NULL); while ((ccb_h = TAILQ_FIRST(&aic->pending_ccbs)) != NULL) { TAILQ_REMOVE(&aic->pending_ccbs, ccb_h, sim_links.tqe); ccb_h->status |= CAM_SCSI_BUS_RESET; aic_done(aic, (struct aic_scb *)ccb_h->ccb_scb_ptr); } while ((ccb_h = TAILQ_FIRST(&aic->nexus_ccbs)) != NULL) { TAILQ_REMOVE(&aic->nexus_ccbs, ccb_h, sim_links.tqe); ccb_h->status |= CAM_SCSI_BUS_RESET; aic_done(aic, (struct aic_scb *)ccb_h->ccb_scb_ptr); } if (aic->nexus) { ccb_h = &aic->nexus->ccb->ccb_h; ccb_h->status |= CAM_SCSI_BUS_RESET; aic_done(aic, aic->nexus); } aic->state = AIC_IDLE; aic_outb(aic, DMACNTRL0, INTEN); } static char *aic_chip_names[] = { "AIC6260", "AIC6360", "AIC6370", "GM82C700", }; static struct { int type; char *idstring; } aic_chip_ids[] = { { AIC6360, IDSTRING_AIC6360 }, { AIC6370, IDSTRING_AIC6370 }, { GM82C700, IDSTRING_GM82C700 }, }; static void aic_init(struct aic_softc *aic) { struct aic_scb *scb; struct aic_tinfo *ti; u_int8_t porta, portb; char chip_id[33]; int i; TAILQ_INIT(&aic->pending_ccbs); TAILQ_INIT(&aic->nexus_ccbs); SLIST_INIT(&aic->free_scbs); aic->nexus = NULL; aic->state = AIC_IDLE; aic->prev_phase = -1; aic->flags = 0; aic_chip_reset(aic); aic_scsi_reset(aic); /* determine the chip type from its ID string */ aic->chip_type = AIC6260; aic_insb(aic, ID, chip_id, sizeof(chip_id) - 1); chip_id[sizeof(chip_id) - 1] = '\0'; for (i = 0; i < nitems(aic_chip_ids); i++) { if (!strcmp(chip_id, aic_chip_ids[i].idstring)) { aic->chip_type = aic_chip_ids[i].type; break; } } porta = aic_inb(aic, PORTA); portb = aic_inb(aic, PORTB); aic->initiator = PORTA_ID(porta); if (PORTA_PARITY(porta)) aic->flags |= AIC_PARITY_ENABLE; if (PORTB_DISC(portb)) aic->flags |= AIC_DISC_ENABLE; if (PORTB_DMA(portb)) aic->flags |= AIC_DMA_ENABLE; /* * We can do fast SCSI (10MHz clock rate) if bit 4 of portb * is set and we've got a 6360. The 6260 can only do standard * 5MHz SCSI. */ if (aic->chip_type > AIC6260 || aic_inb(aic, REV)) { if (PORTB_FSYNC(portb)) aic->flags |= AIC_FAST_ENABLE; aic->flags |= AIC_DWIO_ENABLE; } if (aic->flags & AIC_FAST_ENABLE) aic->max_period = AIC_FAST_SYNC_PERIOD; else aic->max_period = AIC_SYNC_PERIOD; aic->min_period = AIC_MIN_SYNC_PERIOD; for (i = 255; i >= 0; i--) { scb = &aic->scbs[i]; scb->tag = i; callout_init_mtx(&scb->timer, &aic->lock, 0); aic_free_scb(aic, scb); } for (i = 0; i < 8; i++) { if (i == aic->initiator) continue; ti = &aic->tinfo[i]; bzero(ti, sizeof(*ti)); ti->flags = TINFO_TAG_ENB; if (aic->flags & AIC_DISC_ENABLE) ti->flags |= TINFO_DISC_ENB; ti->user.period = aic->max_period; ti->user.offset = AIC_SYNC_OFFSET; ti->scsirate = 0; } aic_outb(aic, DMACNTRL0, INTEN); } int aic_probe(struct aic_softc *aic) { int i; /* Remove aic6360 from possible powerdown mode */ aic_outb(aic, DMACNTRL0, 0); #define STSIZE 16 aic_outb(aic, DMACNTRL1, 0); /* Reset stack pointer */ for (i = 0; i < STSIZE; i++) aic_outb(aic, STACK, i); /* See if we can pull out the same sequence */ aic_outb(aic, DMACNTRL1, 0); for (i = 0; i < STSIZE && aic_inb(aic, STACK) == i; i++) ; if (i != STSIZE) return (ENXIO); #undef STSIZE return (0); } int aic_attach(struct aic_softc *aic) { struct cam_devq *devq; /* * Create the device queue for our SIM. */ devq = cam_simq_alloc(256); if (devq == NULL) return (ENOMEM); /* * Construct our SIM entry */ aic->sim = cam_sim_alloc(aic_action, aic_poll, "aic", aic, device_get_unit(aic->dev), &aic->lock, 2, 256, devq); if (aic->sim == NULL) { cam_simq_free(devq); return (ENOMEM); } mtx_lock(&aic->lock); if (xpt_bus_register(aic->sim, aic->dev, 0) != CAM_SUCCESS) { cam_sim_free(aic->sim, /*free_devq*/TRUE); mtx_unlock(&aic->lock); return (ENXIO); } if (xpt_create_path(&aic->path, /*periph*/NULL, cam_sim_path(aic->sim), CAM_TARGET_WILDCARD, CAM_LUN_WILDCARD) != CAM_REQ_CMP) { xpt_bus_deregister(cam_sim_path(aic->sim)); cam_sim_free(aic->sim, /*free_devq*/TRUE); mtx_unlock(&aic->lock); return (ENXIO); } aic_init(aic); device_printf(aic->dev, "%s", aic_chip_names[aic->chip_type]); if (aic->flags & AIC_DMA_ENABLE) printf(", dma"); if (aic->flags & AIC_DISC_ENABLE) printf(", disconnection"); if (aic->flags & AIC_PARITY_ENABLE) printf(", parity check"); if (aic->flags & AIC_FAST_ENABLE) printf(", fast SCSI"); printf("\n"); mtx_unlock(&aic->lock); return (0); } int aic_detach(struct aic_softc *aic) { struct aic_scb *scb; int i; mtx_lock(&aic->lock); xpt_async(AC_LOST_DEVICE, aic->path, NULL); xpt_free_path(aic->path); xpt_bus_deregister(cam_sim_path(aic->sim)); cam_sim_free(aic->sim, /*free_devq*/TRUE); mtx_unlock(&aic->lock); for (i = 255; i >= 0; i--) { scb = &aic->scbs[i]; callout_drain(&scb->timer); } return (0); } Index: projects/clang400-import/sys/dev/ath/ath_hal/ah_regdomain.c =================================================================== --- projects/clang400-import/sys/dev/ath/ath_hal/ah_regdomain.c (revision 311459) +++ projects/clang400-import/sys/dev/ath/ath_hal/ah_regdomain.c (revision 311460) @@ -1,944 +1,996 @@ /* * Copyright (c) 2002-2009 Sam Leffler, Errno Consulting * Copyright (c) 2005-2006 Atheros Communications, Inc. * All rights reserved. * * Permission to use, copy, modify, and/or 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. * * $FreeBSD$ */ #include "opt_ah.h" #include "ah.h" #include #include #include "ah_internal.h" #include "ah_eeprom.h" #include "ah_devid.h" #include "ah_regdomain.h" /* * XXX this code needs a audit+review */ /* used throughout this file... */ #define N(a) nitems(a) #define HAL_MODE_11A_TURBO HAL_MODE_108A #define HAL_MODE_11G_TURBO HAL_MODE_108G /* * Mask to check whether a domain is a multidomain or a single domain */ #define MULTI_DOMAIN_MASK 0xFF00 /* * Enumerated Regulatory Domain Information 8 bit values indicate that * the regdomain is really a pair of unitary regdomains. 12 bit values * are the real unitary regdomains and are the only ones which have the * frequency bitmasks and flags set. */ #include "ah_regdomain/ah_rd_regenum.h" #define WORLD_SKU_MASK 0x00F0 #define WORLD_SKU_PREFIX 0x0060 /* * THE following table is the mapping of regdomain pairs specified by * an 8 bit regdomain value to the individual unitary reg domains */ #include "ah_regdomain/ah_rd_regmap.h" /* * The following tables are the master list for all different freqeuncy * bands with the complete matrix of all possible flags and settings * for each band if it is used in ANY reg domain. */ #define COUNTRY_ERD_FLAG 0x8000 #define WORLDWIDE_ROAMING_FLAG 0x4000 /* * This table maps country ISO codes from net80211 into regulatory * domains which the ath regulatory domain code understands. */ #include "ah_regdomain/ah_rd_ctry.h" /* * The frequency band collections are a set of frequency ranges * with shared properties - max tx power, max antenna gain, channel width, * channel spacing, DFS requirements and passive scanning requirements. * * These are represented as entries in a frequency band bitmask. * Each regulatory domain entry in ah_regdomain_domains.h uses one * or more frequency band entries for each of the channel modes * supported (11bg, 11a, half, quarter, turbo, etc.) * */ #include "ah_regdomain/ah_rd_freqbands.h" /* * This is the main regulatory database. It defines the supported * set of features and requirements for each of the defined regulatory * zones. It uses combinations of frequency ranges - represented in * a bitmask - to determine the requirements and limitations needed. */ #include "ah_regdomain/ah_rd_domains.h" static const struct cmode modes[] = { { HAL_MODE_TURBO, IEEE80211_CHAN_ST, ®Dmn5GhzTurboFreq[0] }, { HAL_MODE_11A, IEEE80211_CHAN_A, ®Dmn5GhzFreq[0] }, { HAL_MODE_11B, IEEE80211_CHAN_B, ®Dmn2GhzFreq[0] }, { HAL_MODE_11G, IEEE80211_CHAN_G, ®Dmn2Ghz11gFreq[0] }, { HAL_MODE_11G_TURBO, IEEE80211_CHAN_108G, ®Dmn2Ghz11gTurboFreq[0] }, { HAL_MODE_11A_TURBO, IEEE80211_CHAN_108A, ®Dmn5GhzTurboFreq[0] }, { HAL_MODE_11A_QUARTER_RATE, IEEE80211_CHAN_A | IEEE80211_CHAN_QUARTER, ®Dmn5GhzFreq[0] }, { HAL_MODE_11A_HALF_RATE, IEEE80211_CHAN_A | IEEE80211_CHAN_HALF, ®Dmn5GhzFreq[0] }, { HAL_MODE_11G_QUARTER_RATE, IEEE80211_CHAN_G | IEEE80211_CHAN_QUARTER, ®Dmn2Ghz11gFreq[0] }, { HAL_MODE_11G_HALF_RATE, IEEE80211_CHAN_G | IEEE80211_CHAN_HALF, ®Dmn2Ghz11gFreq[0] }, { HAL_MODE_11NG_HT20, IEEE80211_CHAN_G | IEEE80211_CHAN_HT20, ®Dmn2Ghz11gFreq[0] }, { HAL_MODE_11NG_HT40PLUS, IEEE80211_CHAN_G | IEEE80211_CHAN_HT40U, ®Dmn2Ghz11gFreq[0] }, { HAL_MODE_11NG_HT40MINUS, IEEE80211_CHAN_G | IEEE80211_CHAN_HT40D, ®Dmn2Ghz11gFreq[0] }, { HAL_MODE_11NA_HT20, IEEE80211_CHAN_A | IEEE80211_CHAN_HT20, ®Dmn5GhzFreq[0] }, { HAL_MODE_11NA_HT40PLUS, IEEE80211_CHAN_A | IEEE80211_CHAN_HT40U, ®Dmn5GhzFreq[0] }, { HAL_MODE_11NA_HT40MINUS, IEEE80211_CHAN_A | IEEE80211_CHAN_HT40D, ®Dmn5GhzFreq[0] }, }; static void ath_hal_update_dfsdomain(struct ath_hal *ah); static OS_INLINE uint16_t getEepromRD(struct ath_hal *ah) { return AH_PRIVATE(ah)->ah_currentRD &~ WORLDWIDE_ROAMING_FLAG; } /* * Test to see if the bitmask array is all zeros */ static HAL_BOOL isChanBitMaskZero(const uint64_t *bitmask) { #if BMLEN > 2 #error "add more cases" #endif #if BMLEN > 1 if (bitmask[1] != 0) return AH_FALSE; #endif return (bitmask[0] == 0); } /* * Return whether or not the regulatory domain/country in EEPROM * is acceptable. */ static HAL_BOOL isEepromValid(struct ath_hal *ah) { uint16_t rd = getEepromRD(ah); int i; if (rd & COUNTRY_ERD_FLAG) { uint16_t cc = rd &~ COUNTRY_ERD_FLAG; for (i = 0; i < N(allCountries); i++) if (allCountries[i].countryCode == cc) return AH_TRUE; } else { for (i = 0; i < N(regDomainPairs); i++) if (regDomainPairs[i].regDmnEnum == rd) return AH_TRUE; } if (rd == FCC_UBNT) { return AH_TRUE; } HALDEBUG(ah, HAL_DEBUG_REGDOMAIN, "%s: invalid regulatory domain/country code 0x%x\n", __func__, rd); return AH_FALSE; } /* * Find the pointer to the country element in the country table * corresponding to the country code */ static COUNTRY_CODE_TO_ENUM_RD* findCountry(HAL_CTRY_CODE countryCode) { int i; for (i = 0; i < N(allCountries); i++) { if (allCountries[i].countryCode == countryCode) return &allCountries[i]; } return AH_NULL; } static REG_DOMAIN * findRegDmn(int regDmn) { int i; for (i = 0; i < N(regDomains); i++) { if (regDomains[i].regDmnEnum == regDmn) return ®Domains[i]; } return AH_NULL; } static REG_DMN_PAIR_MAPPING * findRegDmnPair(int regDmnPair) { int i; if (regDmnPair != NO_ENUMRD) { for (i = 0; i < N(regDomainPairs); i++) { if (regDomainPairs[i].regDmnEnum == regDmnPair) return ®DomainPairs[i]; } } return AH_NULL; } /* * Calculate a default country based on the EEPROM setting. */ static HAL_CTRY_CODE getDefaultCountry(struct ath_hal *ah) { REG_DMN_PAIR_MAPPING *regpair; uint16_t rd; rd = getEepromRD(ah); if (rd & COUNTRY_ERD_FLAG) { COUNTRY_CODE_TO_ENUM_RD *country; uint16_t cc = rd & ~COUNTRY_ERD_FLAG; country = findCountry(cc); if (country != AH_NULL) return cc; } /* * Check reg domains that have only one country */ regpair = findRegDmnPair(rd); return (regpair != AH_NULL) ? regpair->singleCC : CTRY_DEFAULT; } static HAL_BOOL IS_BIT_SET(int bit, const uint64_t bitmask[]) { int byteOffset, bitnum; uint64_t val; byteOffset = bit/64; bitnum = bit - byteOffset*64; val = ((uint64_t) 1) << bitnum; return (bitmask[byteOffset] & val) != 0; } static HAL_STATUS getregstate(struct ath_hal *ah, HAL_CTRY_CODE cc, HAL_REG_DOMAIN regDmn, COUNTRY_CODE_TO_ENUM_RD **pcountry, REG_DOMAIN **prd2GHz, REG_DOMAIN **prd5GHz) { COUNTRY_CODE_TO_ENUM_RD *country; REG_DOMAIN *rd5GHz, *rd2GHz; if (cc == CTRY_DEFAULT && regDmn == SKU_NONE) { /* * Validate the EEPROM setting and setup defaults */ if (!isEepromValid(ah)) { /* * Don't return any channels if the EEPROM has an * invalid regulatory domain/country code setting. */ HALDEBUG(ah, HAL_DEBUG_REGDOMAIN, "%s: invalid EEPROM contents\n",__func__); return HAL_EEBADREG; } cc = getDefaultCountry(ah); country = findCountry(cc); if (country == AH_NULL) { HALDEBUG(ah, HAL_DEBUG_REGDOMAIN, "NULL Country!, cc %d\n", cc); return HAL_EEBADCC; } regDmn = country->regDmnEnum; HALDEBUG(ah, HAL_DEBUG_REGDOMAIN, "%s: EEPROM cc %u rd 0x%x\n", __func__, cc, regDmn); if (country->countryCode == CTRY_DEFAULT) { /* * Check EEPROM; SKU may be for a country, single * domain, or multiple domains (WWR). */ uint16_t rdnum = getEepromRD(ah); if ((rdnum & COUNTRY_ERD_FLAG) == 0 && (findRegDmn(rdnum) != AH_NULL || findRegDmnPair(rdnum) != AH_NULL)) { regDmn = rdnum; HALDEBUG(ah, HAL_DEBUG_REGDOMAIN, "%s: EEPROM rd 0x%x\n", __func__, rdnum); } } } else { country = findCountry(cc); if (country == AH_NULL) { HALDEBUG(ah, HAL_DEBUG_REGDOMAIN, "unknown country, cc %d\n", cc); return HAL_EINVAL; } if (regDmn == SKU_NONE) regDmn = country->regDmnEnum; HALDEBUG(ah, HAL_DEBUG_REGDOMAIN, "%s: cc %u rd 0x%x\n", __func__, cc, regDmn); } /* * Setup per-band state. */ if ((regDmn & MULTI_DOMAIN_MASK) == 0) { REG_DMN_PAIR_MAPPING *regpair = findRegDmnPair(regDmn); if (regpair == AH_NULL) { HALDEBUG(ah, HAL_DEBUG_REGDOMAIN, "%s: no reg domain pair %u for country %u\n", __func__, regDmn, country->countryCode); return HAL_EINVAL; } rd5GHz = findRegDmn(regpair->regDmn5GHz); if (rd5GHz == AH_NULL) { HALDEBUG(ah, HAL_DEBUG_REGDOMAIN, "%s: no 5GHz reg domain %u for country %u\n", __func__, regpair->regDmn5GHz, country->countryCode); return HAL_EINVAL; } rd2GHz = findRegDmn(regpair->regDmn2GHz); if (rd2GHz == AH_NULL) { HALDEBUG(ah, HAL_DEBUG_REGDOMAIN, "%s: no 2GHz reg domain %u for country %u\n", __func__, regpair->regDmn2GHz, country->countryCode); return HAL_EINVAL; } } else { rd5GHz = rd2GHz = findRegDmn(regDmn); if (rd2GHz == AH_NULL) { HALDEBUG(ah, HAL_DEBUG_REGDOMAIN, "%s: no unitary reg domain %u for country %u\n", __func__, regDmn, country->countryCode); return HAL_EINVAL; } } if (pcountry != AH_NULL) *pcountry = country; *prd2GHz = rd2GHz; *prd5GHz = rd5GHz; return HAL_OK; } static uint64_t * getchannelBM(u_int mode, REG_DOMAIN *rd) { switch (mode) { case HAL_MODE_11B: return (rd->chan11b); case HAL_MODE_11G_QUARTER_RATE: return (rd->chan11g_quarter); case HAL_MODE_11G_HALF_RATE: return (rd->chan11g_half); case HAL_MODE_11G: case HAL_MODE_11NG_HT20: case HAL_MODE_11NG_HT40PLUS: case HAL_MODE_11NG_HT40MINUS: return (rd->chan11g); case HAL_MODE_11G_TURBO: return (rd->chan11g_turbo); case HAL_MODE_11A_QUARTER_RATE: return (rd->chan11a_quarter); case HAL_MODE_11A_HALF_RATE: return (rd->chan11a_half); case HAL_MODE_11A: case HAL_MODE_11NA_HT20: case HAL_MODE_11NA_HT40PLUS: case HAL_MODE_11NA_HT40MINUS: return (rd->chan11a); case HAL_MODE_TURBO: return (rd->chan11a_turbo); case HAL_MODE_11A_TURBO: return (rd->chan11a_dyn_turbo); default: return (AH_NULL); } } static void setchannelflags(struct ieee80211_channel *c, REG_DMN_FREQ_BAND *fband, REG_DOMAIN *rd) { if (fband->usePassScan & rd->pscan) c->ic_flags |= IEEE80211_CHAN_PASSIVE; if (fband->useDfs & rd->dfsMask) c->ic_flags |= IEEE80211_CHAN_DFS; if (IEEE80211_IS_CHAN_5GHZ(c) && (rd->flags & DISALLOW_ADHOC_11A)) c->ic_flags |= IEEE80211_CHAN_NOADHOC; if (IEEE80211_IS_CHAN_TURBO(c) && (rd->flags & DISALLOW_ADHOC_11A_TURB)) c->ic_flags |= IEEE80211_CHAN_NOADHOC; if (rd->flags & NO_HOSTAP) c->ic_flags |= IEEE80211_CHAN_NOHOSTAP; if (rd->flags & LIMIT_FRAME_4MS) c->ic_flags |= IEEE80211_CHAN_4MSXMIT; if (rd->flags & NEED_NFC) c->ic_flags |= CHANNEL_NFCREQUIRED; } static int addchan(struct ath_hal *ah, struct ieee80211_channel chans[], u_int maxchans, int *nchans, uint16_t freq, uint32_t flags, REG_DMN_FREQ_BAND *fband, REG_DOMAIN *rd) { struct ieee80211_channel *c; if (*nchans >= maxchans) return (HAL_ENOMEM); + HALDEBUG(ah, HAL_DEBUG_REGDOMAIN, + "%s: %d: freq=%d, flags=0x%08x\n", + __func__, *nchans, (int) freq, flags); + c = &chans[(*nchans)++]; c->ic_freq = freq; c->ic_flags = flags; setchannelflags(c, fband, rd); c->ic_maxregpower = fband->powerDfs; ath_hal_getpowerlimits(ah, c); c->ic_maxantgain = fband->antennaMax; return (0); } static int copychan_prev(struct ath_hal *ah, struct ieee80211_channel chans[], - u_int maxchans, int *nchans, uint16_t freq) + u_int maxchans, int *nchans, uint16_t freq, uint32_t flags) { struct ieee80211_channel *c; if (*nchans == 0) return (HAL_EINVAL); if (*nchans >= maxchans) return (HAL_ENOMEM); + HALDEBUG(ah, HAL_DEBUG_REGDOMAIN, + "%s: %d: freq=%d, flags=0x%08x\n", + __func__, *nchans, (int) freq, flags); + c = &chans[(*nchans)++]; c[0] = c[-1]; c->ic_freq = freq; /* XXX is it needed here? */ ath_hal_getpowerlimits(ah, c); return (0); } static int add_chanlist_band(struct ath_hal *ah, struct ieee80211_channel chans[], int maxchans, int *nchans, uint16_t freq_lo, uint16_t freq_hi, int step, uint32_t flags, REG_DMN_FREQ_BAND *fband, REG_DOMAIN *rd) { uint16_t freq = freq_lo; int error; if (freq_hi < freq_lo) return (0); + HALDEBUG(ah, HAL_DEBUG_REGDOMAIN, + "%s: freq=%d..%d, flags=0x%08x, step=%d\n", __func__, + (int) freq_lo, (int) freq_hi, flags, step); + error = addchan(ah, chans, maxchans, nchans, freq, flags, fband, rd); for (freq += step; freq <= freq_hi && error == 0; freq += step) - error = copychan_prev(ah, chans, maxchans, nchans, freq); + error = copychan_prev(ah, chans, maxchans, nchans, freq, flags); return (error); } static void adj_freq_ht40(u_int mode, int *low_adj, int *hi_adj, int *channelSep) { *low_adj = *hi_adj = *channelSep = 0; switch (mode) { case HAL_MODE_11NA_HT40PLUS: *channelSep = 40; /* FALLTHROUGH */ case HAL_MODE_11NG_HT40PLUS: *hi_adj = -20; break; case HAL_MODE_11NA_HT40MINUS: *channelSep = 40; /* FALLTHROUGH */ case HAL_MODE_11NG_HT40MINUS: *low_adj = 20; break; } } static void add_chanlist_mode(struct ath_hal *ah, struct ieee80211_channel chans[], u_int maxchans, int *nchans, const struct cmode *cm, REG_DOMAIN *rd, HAL_BOOL enableExtendedChannels) { uint64_t *channelBM; uint16_t freq_lo, freq_hi; int b, error, low_adj, hi_adj, channelSep; if (!ath_hal_getChannelEdges(ah, cm->flags, &freq_lo, &freq_hi)) { /* channel not supported by hardware, skip it */ HALDEBUG(ah, HAL_DEBUG_REGDOMAIN, "%s: channels 0x%x not supported by hardware\n", __func__, cm->flags); return; } channelBM = getchannelBM(cm->mode, rd); if (isChanBitMaskZero(channelBM)) return; /* * Setup special handling for HT40 channels; e.g. * 5G HT40 channels require 40Mhz channel separation. */ adj_freq_ht40(cm->mode, &low_adj, &hi_adj, &channelSep); for (b = 0; b < 64*BMLEN; b++) { REG_DMN_FREQ_BAND *fband; uint16_t bfreq_lo, bfreq_hi; int step; if (!IS_BIT_SET(b, channelBM)) continue; fband = &cm->freqs[b]; if ((fband->usePassScan & IS_ECM_CHAN) && !enableExtendedChannels) { HALDEBUG(ah, HAL_DEBUG_REGDOMAIN, "skip ecm channels\n"); continue; } #if 0 if ((fband->useDfs & rd->dfsMask) && (cm->flags & IEEE80211_CHAN_HT40)) { /* NB: DFS and HT40 don't mix */ HALDEBUG(ah, HAL_DEBUG_REGDOMAIN, "skip HT40 chan, DFS required\n"); continue; } #endif - /* * XXX TODO: handle REG_EXT_FCC_CH_144. * * Figure out which instances/uses cause us to not * be allowed to use channel 144 (pri or sec overlap.) */ bfreq_lo = MAX(fband->lowChannel + low_adj, freq_lo); bfreq_hi = MIN(fband->highChannel + hi_adj, freq_hi); + + /* + * Don't start the 5GHz channel list at 5120MHz. + * + * Unfortunately (sigh) the HT40 channel creation + * logic will create HT40U channels at 5120, 5160, 5200. + * This means that 36 (5180) isn't considered as a + * HT40 channel, and everything goes messed up from there. + */ + if ((cm->flags & IEEE80211_CHAN_5GHZ) && + (cm->flags & IEEE80211_CHAN_HT40U)) { + if (bfreq_lo < 5180) + bfreq_lo = 5180; + } + + /* + * Same with HT40D - need to start at 5200 or the low + * channels are all wrong again. + */ + if ((cm->flags & IEEE80211_CHAN_5GHZ) && + (cm->flags & IEEE80211_CHAN_HT40D)) { + if (bfreq_lo < 5200) + bfreq_lo = 5200; + } + if (fband->channelSep >= channelSep) step = fband->channelSep; else step = roundup(channelSep, fband->channelSep); + + HALDEBUG(ah, HAL_DEBUG_REGDOMAIN, + "%s: freq_lo=%d, freq_hi=%d, low_adj=%d, hi_adj=%d, " + "bandlo=%d, bandhi=%d, bfreqlo=%d, bfreqhi=%d, step=%d, " + "flags=0x%08x\n", + __func__, + (int) freq_lo, + (int) freq_hi, + (int) low_adj, + (int) hi_adj, + (int) fband->lowChannel, + (int) fband->highChannel, + (int) bfreq_lo, + (int) bfreq_hi, + step, + (int) cm->flags); error = add_chanlist_band(ah, chans, maxchans, nchans, bfreq_lo, bfreq_hi, step, cm->flags, fband, rd); if (error != 0) { HALDEBUG(ah, HAL_DEBUG_REGDOMAIN, "%s: too many channels for channel table\n", __func__); return; } } } static u_int getmodesmask(struct ath_hal *ah, REG_DOMAIN *rd5GHz, u_int modeSelect) { #define HAL_MODE_11A_ALL \ (HAL_MODE_11A | HAL_MODE_11A_TURBO | HAL_MODE_TURBO | \ HAL_MODE_11A_QUARTER_RATE | HAL_MODE_11A_HALF_RATE) u_int modesMask; /* get modes that HW is capable of */ modesMask = ath_hal_getWirelessModes(ah); modesMask &= modeSelect; /* optimize work below if no 11a channels */ if (isChanBitMaskZero(rd5GHz->chan11a) && (modesMask & HAL_MODE_11A_ALL)) { HALDEBUG(ah, HAL_DEBUG_REGDOMAIN, "%s: disallow all 11a\n", __func__); modesMask &= ~HAL_MODE_11A_ALL; } return (modesMask); #undef HAL_MODE_11A_ALL } /* * Construct the channel list for the specified regulatory config. */ static HAL_STATUS getchannels(struct ath_hal *ah, struct ieee80211_channel chans[], u_int maxchans, int *nchans, u_int modeSelect, HAL_CTRY_CODE cc, HAL_REG_DOMAIN regDmn, HAL_BOOL enableExtendedChannels, COUNTRY_CODE_TO_ENUM_RD **pcountry, REG_DOMAIN **prd2GHz, REG_DOMAIN **prd5GHz) { REG_DOMAIN *rd5GHz, *rd2GHz; u_int modesMask; const struct cmode *cm; HAL_STATUS status; HALDEBUG(ah, HAL_DEBUG_REGDOMAIN, "%s: cc %u regDmn 0x%x mode 0x%x%s\n", __func__, cc, regDmn, modeSelect, enableExtendedChannels ? " ecm" : ""); status = getregstate(ah, cc, regDmn, pcountry, &rd2GHz, &rd5GHz); if (status != HAL_OK) return status; modesMask = getmodesmask(ah, rd5GHz, modeSelect); /* XXX error? */ if (modesMask == 0) goto done; for (cm = modes; cm < &modes[N(modes)]; cm++) { REG_DOMAIN *rd; if ((cm->mode & modesMask) == 0) { HALDEBUG(ah, HAL_DEBUG_REGDOMAIN, "%s: skip mode 0x%x flags 0x%x\n", __func__, cm->mode, cm->flags); continue; } if (cm->flags & IEEE80211_CHAN_5GHZ) rd = rd5GHz; else if (cm->flags & IEEE80211_CHAN_2GHZ) rd = rd2GHz; else { ath_hal_printf(ah, "%s: Unkonwn HAL flags 0x%x\n", __func__, cm->flags); return HAL_EINVAL; } add_chanlist_mode(ah, chans, maxchans, nchans, cm, rd, enableExtendedChannels); if (*nchans >= maxchans) goto done; } done: /* NB: pcountry set above by getregstate */ if (prd2GHz != AH_NULL) *prd2GHz = rd2GHz; if (prd5GHz != AH_NULL) *prd5GHz = rd5GHz; return HAL_OK; } /* * Retrieve a channel list without affecting runtime state. */ HAL_STATUS ath_hal_getchannels(struct ath_hal *ah, struct ieee80211_channel chans[], u_int maxchans, int *nchans, u_int modeSelect, HAL_CTRY_CODE cc, HAL_REG_DOMAIN regDmn, HAL_BOOL enableExtendedChannels) { return getchannels(ah, chans, maxchans, nchans, modeSelect, cc, regDmn, enableExtendedChannels, AH_NULL, AH_NULL, AH_NULL); } /* * Handle frequency mapping from 900Mhz range to 2.4GHz range * for GSM radios. This is done when we need the h/w frequency * and the channel is marked IEEE80211_CHAN_GSM. */ static int ath_hal_mapgsm(int sku, int freq) { if (sku == SKU_XR9) return 1520 + freq; if (sku == SKU_GZ901) return 1544 + freq; if (sku == SKU_SR9) return 3344 - freq; if (sku == SKU_XC900M) return 1517 + freq; HALDEBUG(AH_NULL, HAL_DEBUG_ANY, "%s: cannot map freq %u unknown gsm sku %u\n", __func__, freq, sku); return freq; } /* * Setup the internal/private channel state given a table of * net80211 channels. We collapse entries for the same frequency * and record the frequency for doing noise floor processing * where we don't have net80211 channel context. */ static HAL_BOOL assignPrivateChannels(struct ath_hal *ah, struct ieee80211_channel chans[], int nchans, int sku) { HAL_CHANNEL_INTERNAL *ic; int i, j, next, freq; next = 0; for (i = 0; i < nchans; i++) { struct ieee80211_channel *c = &chans[i]; for (j = i-1; j >= 0; j--) if (chans[j].ic_freq == c->ic_freq) { c->ic_devdata = chans[j].ic_devdata; break; } if (j < 0) { /* new entry, assign a private channel entry */ if (next >= N(AH_PRIVATE(ah)->ah_channels)) { HALDEBUG(ah, HAL_DEBUG_ANY, "%s: too many channels, max %zu\n", __func__, N(AH_PRIVATE(ah)->ah_channels)); return AH_FALSE; } /* * Handle frequency mapping for 900MHz devices. * The hardware uses 2.4GHz frequencies that are * down-converted. The 802.11 layer uses the * true frequencies. */ freq = IEEE80211_IS_CHAN_GSM(c) ? ath_hal_mapgsm(sku, c->ic_freq) : c->ic_freq; HALDEBUG(ah, HAL_DEBUG_REGDOMAIN, "%s: private[%3u] %u/0x%x -> channel %u\n", __func__, next, c->ic_freq, c->ic_flags, freq); ic = &AH_PRIVATE(ah)->ah_channels[next]; /* * NB: This clears privFlags which means ancillary * code like ANI and IQ calibration will be * restarted and re-setup any per-channel state. */ OS_MEMZERO(ic, sizeof(*ic)); ic->channel = freq; c->ic_devdata = next; next++; } } AH_PRIVATE(ah)->ah_nchan = next; HALDEBUG(ah, HAL_DEBUG_ANY, "%s: %u public, %u private channels\n", __func__, nchans, next); return AH_TRUE; } /* * Setup the channel list based on the information in the EEPROM. */ HAL_STATUS ath_hal_init_channels(struct ath_hal *ah, struct ieee80211_channel chans[], u_int maxchans, int *nchans, u_int modeSelect, HAL_CTRY_CODE cc, HAL_REG_DOMAIN regDmn, HAL_BOOL enableExtendedChannels) { COUNTRY_CODE_TO_ENUM_RD *country; REG_DOMAIN *rd5GHz, *rd2GHz; HAL_STATUS status; status = getchannels(ah, chans, maxchans, nchans, modeSelect, cc, regDmn, enableExtendedChannels, &country, &rd2GHz, &rd5GHz); if (status == HAL_OK && assignPrivateChannels(ah, chans, *nchans, AH_PRIVATE(ah)->ah_currentRD)) { AH_PRIVATE(ah)->ah_rd2GHz = rd2GHz; AH_PRIVATE(ah)->ah_rd5GHz = rd5GHz; ah->ah_countryCode = country->countryCode; HALDEBUG(ah, HAL_DEBUG_REGDOMAIN, "%s: cc %u\n", __func__, ah->ah_countryCode); /* Update current DFS domain */ ath_hal_update_dfsdomain(ah); } else status = HAL_EINVAL; return status; } /* * Set the channel list. */ HAL_STATUS ath_hal_set_channels(struct ath_hal *ah, struct ieee80211_channel chans[], int nchans, HAL_CTRY_CODE cc, HAL_REG_DOMAIN rd) { COUNTRY_CODE_TO_ENUM_RD *country; REG_DOMAIN *rd5GHz, *rd2GHz; HAL_STATUS status; switch (rd) { case SKU_SR9: case SKU_XR9: case SKU_GZ901: case SKU_XC900M: /* * Map 900MHz sku's. The frequencies will be mapped * according to the sku to compensate for the down-converter. * We use the FCC for these sku's as the mapped channel * list is known compatible (will need to change if/when * vendors do different mapping in different locales). */ status = getregstate(ah, CTRY_DEFAULT, SKU_FCC, &country, &rd2GHz, &rd5GHz); break; default: status = getregstate(ah, cc, rd, &country, &rd2GHz, &rd5GHz); rd = AH_PRIVATE(ah)->ah_currentRD; break; } if (status == HAL_OK && assignPrivateChannels(ah, chans, nchans, rd)) { AH_PRIVATE(ah)->ah_rd2GHz = rd2GHz; AH_PRIVATE(ah)->ah_rd5GHz = rd5GHz; ah->ah_countryCode = country->countryCode; HALDEBUG(ah, HAL_DEBUG_REGDOMAIN, "%s: cc %u\n", __func__, ah->ah_countryCode); } else status = HAL_EINVAL; if (status == HAL_OK) { /* Update current DFS domain */ (void) ath_hal_update_dfsdomain(ah); } return status; } #ifdef AH_DEBUG /* * Return the internal channel corresponding to a public channel. * NB: normally this routine is inline'd (see ah_internal.h) */ HAL_CHANNEL_INTERNAL * ath_hal_checkchannel(struct ath_hal *ah, const struct ieee80211_channel *c) { HAL_CHANNEL_INTERNAL *cc = &AH_PRIVATE(ah)->ah_channels[c->ic_devdata]; if (c->ic_devdata < AH_PRIVATE(ah)->ah_nchan && (c->ic_freq == cc->channel || IEEE80211_IS_CHAN_GSM(c))) return cc; if (c->ic_devdata >= AH_PRIVATE(ah)->ah_nchan) { HALDEBUG(ah, HAL_DEBUG_ANY, "%s: bad mapping, devdata %u nchans %u\n", __func__, c->ic_devdata, AH_PRIVATE(ah)->ah_nchan); HALASSERT(c->ic_devdata < AH_PRIVATE(ah)->ah_nchan); } else { HALDEBUG(ah, HAL_DEBUG_ANY, "%s: no match for %u/0x%x devdata %u channel %u\n", __func__, c->ic_freq, c->ic_flags, c->ic_devdata, cc->channel); HALASSERT(c->ic_freq == cc->channel || IEEE80211_IS_CHAN_GSM(c)); } return AH_NULL; } #endif /* AH_DEBUG */ #define isWwrSKU(_ah) \ ((getEepromRD((_ah)) & WORLD_SKU_MASK) == WORLD_SKU_PREFIX || \ getEepromRD(_ah) == WORLD) /* * Return the test group for the specific channel based on * the current regulatory setup. */ u_int ath_hal_getctl(struct ath_hal *ah, const struct ieee80211_channel *c) { u_int ctl; if (AH_PRIVATE(ah)->ah_rd2GHz == AH_PRIVATE(ah)->ah_rd5GHz || (ah->ah_countryCode == CTRY_DEFAULT && isWwrSKU(ah))) ctl = SD_NO_CTL; else if (IEEE80211_IS_CHAN_2GHZ(c)) ctl = AH_PRIVATE(ah)->ah_rd2GHz->conformanceTestLimit; else ctl = AH_PRIVATE(ah)->ah_rd5GHz->conformanceTestLimit; if (IEEE80211_IS_CHAN_B(c)) return ctl | CTL_11B; if (IEEE80211_IS_CHAN_G(c)) return ctl | CTL_11G; if (IEEE80211_IS_CHAN_108G(c)) return ctl | CTL_108G; if (IEEE80211_IS_CHAN_TURBO(c)) return ctl | CTL_TURBO; if (IEEE80211_IS_CHAN_A(c)) return ctl | CTL_11A; return ctl; } /* * Update the current dfsDomain setting based on the given * country code. * * Since FreeBSD/net80211 allows the channel set to change * after the card has been setup (via ath_hal_init_channels()) * this function method is needed to update ah_dfsDomain. */ void ath_hal_update_dfsdomain(struct ath_hal *ah) { const REG_DOMAIN *rd5GHz = AH_PRIVATE(ah)->ah_rd5GHz; HAL_DFS_DOMAIN dfsDomain = HAL_DFS_UNINIT_DOMAIN; if (rd5GHz->dfsMask & DFS_FCC3) dfsDomain = HAL_DFS_FCC_DOMAIN; if (rd5GHz->dfsMask & DFS_ETSI) dfsDomain = HAL_DFS_ETSI_DOMAIN; if (rd5GHz->dfsMask & DFS_MKK4) dfsDomain = HAL_DFS_MKK4_DOMAIN; AH_PRIVATE(ah)->ah_dfsDomain = dfsDomain; HALDEBUG(ah, HAL_DEBUG_REGDOMAIN, "%s ah_dfsDomain: %d\n", __func__, AH_PRIVATE(ah)->ah_dfsDomain); } /* * Return the max allowed antenna gain and apply any regulatory * domain specific changes. * * NOTE: a negative reduction is possible in RD's that only * measure radiated power (e.g., ETSI) which would increase * that actual conducted output power (though never beyond * the calibrated target power). */ u_int ath_hal_getantennareduction(struct ath_hal *ah, const struct ieee80211_channel *chan, u_int twiceGain) { int8_t antennaMax = twiceGain - chan->ic_maxantgain*2; return (antennaMax < 0) ? 0 : antennaMax; } Index: projects/clang400-import/sys/dev/ciss/ciss.c =================================================================== --- projects/clang400-import/sys/dev/ciss/ciss.c (revision 311459) +++ projects/clang400-import/sys/dev/ciss/ciss.c (revision 311460) @@ -1,4731 +1,4731 @@ /*- * Copyright (c) 2001 Michael Smith * Copyright (c) 2004 Paul Saab * 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. * * $FreeBSD$ */ /* * Common Interface for SCSI-3 Support driver. * * CISS claims to provide a common interface between a generic SCSI * transport and an intelligent host adapter. * * This driver supports CISS as defined in the document "CISS Command * Interface for SCSI-3 Support Open Specification", Version 1.04, * Valence Number 1, dated 20001127, produced by Compaq Computer * Corporation. This document appears to be a hastily and somewhat * arbitrarlily cut-down version of a larger (and probably even more * chaotic and inconsistent) Compaq internal document. Various * details were also gleaned from Compaq's "cciss" driver for Linux. * * We provide a shim layer between the CISS interface and CAM, * offloading most of the queueing and being-a-disk chores onto CAM. * Entry to the driver is via the PCI bus attachment (ciss_probe, * ciss_attach, etc) and via the CAM interface (ciss_cam_action, * ciss_cam_poll). The Compaq CISS adapters are, however, poor SCSI * citizens and we have to fake up some responses to get reasonable * behaviour out of them. In addition, the CISS command set is by no * means adequate to support the functionality of a RAID controller, * and thus the supported Compaq adapters utilise portions of the * control protocol from earlier Compaq adapter families. * * Note that we only support the "simple" transport layer over PCI. * This interface (ab)uses the I2O register set (specifically the post * queues) to exchange commands with the adapter. Other interfaces * are available, but we aren't supposed to know about them, and it is * dubious whether they would provide major performance improvements * except under extreme load. * * Currently the only supported CISS adapters are the Compaq Smart * Array 5* series (5300, 5i, 532). Even with only three adapters, * Compaq still manage to have interface variations. * * * Thanks must go to Fred Harris and Darryl DeVinney at Compaq, as * well as Paul Saab at Yahoo! for their assistance in making this * driver happen. * * More thanks must go to John Cagle at HP for the countless hours * spent making this driver "work" with the MSA* series storage * enclosures. Without his help (and nagging), this driver could not * be used with these enclosures. */ #include #include #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 MALLOC_DEFINE(CISS_MALLOC_CLASS, "ciss_data", "ciss internal data buffers"); /* pci interface */ static int ciss_lookup(device_t dev); static int ciss_probe(device_t dev); static int ciss_attach(device_t dev); static int ciss_detach(device_t dev); static int ciss_shutdown(device_t dev); /* (de)initialisation functions, control wrappers */ static int ciss_init_pci(struct ciss_softc *sc); static int ciss_setup_msix(struct ciss_softc *sc); static int ciss_init_perf(struct ciss_softc *sc); static int ciss_wait_adapter(struct ciss_softc *sc); static int ciss_flush_adapter(struct ciss_softc *sc); static int ciss_init_requests(struct ciss_softc *sc); static void ciss_command_map_helper(void *arg, bus_dma_segment_t *segs, int nseg, int error); static int ciss_identify_adapter(struct ciss_softc *sc); static int ciss_init_logical(struct ciss_softc *sc); static int ciss_init_physical(struct ciss_softc *sc); static int ciss_filter_physical(struct ciss_softc *sc, struct ciss_lun_report *cll); static int ciss_identify_logical(struct ciss_softc *sc, struct ciss_ldrive *ld); static int ciss_get_ldrive_status(struct ciss_softc *sc, struct ciss_ldrive *ld); static int ciss_update_config(struct ciss_softc *sc); static int ciss_accept_media(struct ciss_softc *sc, struct ciss_ldrive *ld); static void ciss_init_sysctl(struct ciss_softc *sc); static void ciss_soft_reset(struct ciss_softc *sc); static void ciss_free(struct ciss_softc *sc); static void ciss_spawn_notify_thread(struct ciss_softc *sc); static void ciss_kill_notify_thread(struct ciss_softc *sc); /* request submission/completion */ static int ciss_start(struct ciss_request *cr); static void ciss_done(struct ciss_softc *sc, cr_qhead_t *qh); static void ciss_perf_done(struct ciss_softc *sc, cr_qhead_t *qh); static void ciss_intr(void *arg); static void ciss_perf_intr(void *arg); static void ciss_perf_msi_intr(void *arg); static void ciss_complete(struct ciss_softc *sc, cr_qhead_t *qh); static int _ciss_report_request(struct ciss_request *cr, int *command_status, int *scsi_status, const char *func); static int ciss_synch_request(struct ciss_request *cr, int timeout); static int ciss_poll_request(struct ciss_request *cr, int timeout); static int ciss_wait_request(struct ciss_request *cr, int timeout); #if 0 static int ciss_abort_request(struct ciss_request *cr); #endif /* request queueing */ static int ciss_get_request(struct ciss_softc *sc, struct ciss_request **crp); static void ciss_preen_command(struct ciss_request *cr); static void ciss_release_request(struct ciss_request *cr); /* request helpers */ static int ciss_get_bmic_request(struct ciss_softc *sc, struct ciss_request **crp, int opcode, void **bufp, size_t bufsize); static int ciss_user_command(struct ciss_softc *sc, IOCTL_Command_struct *ioc); /* DMA map/unmap */ static int ciss_map_request(struct ciss_request *cr); static void ciss_request_map_helper(void *arg, bus_dma_segment_t *segs, int nseg, int error); static void ciss_unmap_request(struct ciss_request *cr); /* CAM interface */ static int ciss_cam_init(struct ciss_softc *sc); static void ciss_cam_rescan_target(struct ciss_softc *sc, int bus, int target); static void ciss_cam_action(struct cam_sim *sim, union ccb *ccb); static int ciss_cam_action_io(struct cam_sim *sim, struct ccb_scsiio *csio); static int ciss_cam_emulate(struct ciss_softc *sc, struct ccb_scsiio *csio); static void ciss_cam_poll(struct cam_sim *sim); static void ciss_cam_complete(struct ciss_request *cr); static void ciss_cam_complete_fixup(struct ciss_softc *sc, struct ccb_scsiio *csio); static int ciss_name_device(struct ciss_softc *sc, int bus, int target); /* periodic status monitoring */ static void ciss_periodic(void *arg); static void ciss_nop_complete(struct ciss_request *cr); static void ciss_disable_adapter(struct ciss_softc *sc); static void ciss_notify_event(struct ciss_softc *sc); static void ciss_notify_complete(struct ciss_request *cr); static int ciss_notify_abort(struct ciss_softc *sc); static int ciss_notify_abort_bmic(struct ciss_softc *sc); static void ciss_notify_hotplug(struct ciss_softc *sc, struct ciss_notify *cn); static void ciss_notify_logical(struct ciss_softc *sc, struct ciss_notify *cn); static void ciss_notify_physical(struct ciss_softc *sc, struct ciss_notify *cn); /* debugging output */ static void ciss_print_request(struct ciss_request *cr); static void ciss_print_ldrive(struct ciss_softc *sc, struct ciss_ldrive *ld); static const char *ciss_name_ldrive_status(int status); static int ciss_decode_ldrive_status(int status); static const char *ciss_name_ldrive_org(int org); static const char *ciss_name_command_status(int status); /* * PCI bus interface. */ static device_method_t ciss_methods[] = { /* Device interface */ DEVMETHOD(device_probe, ciss_probe), DEVMETHOD(device_attach, ciss_attach), DEVMETHOD(device_detach, ciss_detach), DEVMETHOD(device_shutdown, ciss_shutdown), { 0, 0 } }; static driver_t ciss_pci_driver = { "ciss", ciss_methods, sizeof(struct ciss_softc) }; static devclass_t ciss_devclass; DRIVER_MODULE(ciss, pci, ciss_pci_driver, ciss_devclass, 0, 0); MODULE_DEPEND(ciss, cam, 1, 1, 1); MODULE_DEPEND(ciss, pci, 1, 1, 1); /* * Control device interface. */ static d_open_t ciss_open; static d_close_t ciss_close; static d_ioctl_t ciss_ioctl; static struct cdevsw ciss_cdevsw = { .d_version = D_VERSION, .d_flags = 0, .d_open = ciss_open, .d_close = ciss_close, .d_ioctl = ciss_ioctl, .d_name = "ciss", }; /* * This tunable can be set at boot time and controls whether physical devices * that are marked hidden by the firmware should be exposed anyways. */ static unsigned int ciss_expose_hidden_physical = 0; TUNABLE_INT("hw.ciss.expose_hidden_physical", &ciss_expose_hidden_physical); static unsigned int ciss_nop_message_heartbeat = 0; TUNABLE_INT("hw.ciss.nop_message_heartbeat", &ciss_nop_message_heartbeat); /* * This tunable can force a particular transport to be used: * <= 0 : use default * 1 : force simple * 2 : force performant */ static int ciss_force_transport = 0; TUNABLE_INT("hw.ciss.force_transport", &ciss_force_transport); /* * This tunable can force a particular interrupt delivery method to be used: * <= 0 : use default * 1 : force INTx * 2 : force MSIX */ static int ciss_force_interrupt = 0; TUNABLE_INT("hw.ciss.force_interrupt", &ciss_force_interrupt); /************************************************************************ * CISS adapters amazingly don't have a defined programming interface * value. (One could say some very despairing things about PCI and * people just not getting the general idea.) So we are forced to * stick with matching against subvendor/subdevice, and thus have to * be updated for every new CISS adapter that appears. */ #define CISS_BOARD_UNKNWON 0 #define CISS_BOARD_SA5 1 #define CISS_BOARD_SA5B 2 #define CISS_BOARD_NOMSI (1<<4) #define CISS_BOARD_SIMPLE (1<<5) static struct { u_int16_t subvendor; u_int16_t subdevice; int flags; char *desc; } ciss_vendor_data[] = { { 0x0e11, 0x4070, CISS_BOARD_SA5|CISS_BOARD_NOMSI|CISS_BOARD_SIMPLE, "Compaq Smart Array 5300" }, { 0x0e11, 0x4080, CISS_BOARD_SA5B|CISS_BOARD_NOMSI, "Compaq Smart Array 5i" }, { 0x0e11, 0x4082, CISS_BOARD_SA5B|CISS_BOARD_NOMSI, "Compaq Smart Array 532" }, { 0x0e11, 0x4083, CISS_BOARD_SA5B|CISS_BOARD_NOMSI, "HP Smart Array 5312" }, { 0x0e11, 0x4091, CISS_BOARD_SA5, "HP Smart Array 6i" }, { 0x0e11, 0x409A, CISS_BOARD_SA5, "HP Smart Array 641" }, { 0x0e11, 0x409B, CISS_BOARD_SA5, "HP Smart Array 642" }, { 0x0e11, 0x409C, CISS_BOARD_SA5, "HP Smart Array 6400" }, { 0x0e11, 0x409D, CISS_BOARD_SA5, "HP Smart Array 6400 EM" }, { 0x103C, 0x3211, CISS_BOARD_SA5, "HP Smart Array E200i" }, { 0x103C, 0x3212, CISS_BOARD_SA5, "HP Smart Array E200" }, { 0x103C, 0x3213, CISS_BOARD_SA5, "HP Smart Array E200i" }, { 0x103C, 0x3214, CISS_BOARD_SA5, "HP Smart Array E200i" }, { 0x103C, 0x3215, CISS_BOARD_SA5, "HP Smart Array E200i" }, { 0x103C, 0x3220, CISS_BOARD_SA5, "HP Smart Array" }, { 0x103C, 0x3222, CISS_BOARD_SA5, "HP Smart Array" }, { 0x103C, 0x3223, CISS_BOARD_SA5, "HP Smart Array P800" }, { 0x103C, 0x3225, CISS_BOARD_SA5, "HP Smart Array P600" }, { 0x103C, 0x3230, CISS_BOARD_SA5, "HP Smart Array" }, { 0x103C, 0x3231, CISS_BOARD_SA5, "HP Smart Array" }, { 0x103C, 0x3232, CISS_BOARD_SA5, "HP Smart Array" }, { 0x103C, 0x3233, CISS_BOARD_SA5, "HP Smart Array" }, { 0x103C, 0x3234, CISS_BOARD_SA5, "HP Smart Array P400" }, { 0x103C, 0x3235, CISS_BOARD_SA5, "HP Smart Array P400i" }, { 0x103C, 0x3236, CISS_BOARD_SA5, "HP Smart Array" }, { 0x103C, 0x3237, CISS_BOARD_SA5, "HP Smart Array E500" }, { 0x103C, 0x3238, CISS_BOARD_SA5, "HP Smart Array" }, { 0x103C, 0x3239, CISS_BOARD_SA5, "HP Smart Array" }, { 0x103C, 0x323A, CISS_BOARD_SA5, "HP Smart Array" }, { 0x103C, 0x323B, CISS_BOARD_SA5, "HP Smart Array" }, { 0x103C, 0x323C, CISS_BOARD_SA5, "HP Smart Array" }, { 0x103C, 0x323D, CISS_BOARD_SA5, "HP Smart Array P700m" }, { 0x103C, 0x3241, CISS_BOARD_SA5, "HP Smart Array P212" }, { 0x103C, 0x3243, CISS_BOARD_SA5, "HP Smart Array P410" }, { 0x103C, 0x3245, CISS_BOARD_SA5, "HP Smart Array P410i" }, { 0x103C, 0x3247, CISS_BOARD_SA5, "HP Smart Array P411" }, { 0x103C, 0x3249, CISS_BOARD_SA5, "HP Smart Array P812" }, { 0x103C, 0x324A, CISS_BOARD_SA5, "HP Smart Array P712m" }, { 0x103C, 0x324B, CISS_BOARD_SA5, "HP Smart Array" }, { 0x103C, 0x3350, CISS_BOARD_SA5, "HP Smart Array P222" }, { 0x103C, 0x3351, CISS_BOARD_SA5, "HP Smart Array P420" }, { 0x103C, 0x3352, CISS_BOARD_SA5, "HP Smart Array P421" }, { 0x103C, 0x3353, CISS_BOARD_SA5, "HP Smart Array P822" }, { 0x103C, 0x3354, CISS_BOARD_SA5, "HP Smart Array P420i" }, { 0x103C, 0x3355, CISS_BOARD_SA5, "HP Smart Array P220i" }, { 0x103C, 0x3356, CISS_BOARD_SA5, "HP Smart Array P721m" }, { 0x103C, 0x1920, CISS_BOARD_SA5, "HP Smart Array P430i" }, { 0x103C, 0x1921, CISS_BOARD_SA5, "HP Smart Array P830i" }, { 0x103C, 0x1922, CISS_BOARD_SA5, "HP Smart Array P430" }, { 0x103C, 0x1923, CISS_BOARD_SA5, "HP Smart Array P431" }, { 0x103C, 0x1924, CISS_BOARD_SA5, "HP Smart Array P830" }, { 0x103C, 0x1926, CISS_BOARD_SA5, "HP Smart Array P731m" }, { 0x103C, 0x1928, CISS_BOARD_SA5, "HP Smart Array P230i" }, { 0x103C, 0x1929, CISS_BOARD_SA5, "HP Smart Array P530" }, { 0x103C, 0x192A, CISS_BOARD_SA5, "HP Smart Array P531" }, { 0x103C, 0x21BD, CISS_BOARD_SA5, "HP Smart Array P244br" }, { 0x103C, 0x21BE, CISS_BOARD_SA5, "HP Smart Array P741m" }, { 0x103C, 0x21BF, CISS_BOARD_SA5, "HP Smart Array H240ar" }, { 0x103C, 0x21C0, CISS_BOARD_SA5, "HP Smart Array P440ar" }, { 0x103C, 0x21C1, CISS_BOARD_SA5, "HP Smart Array P840ar" }, { 0x103C, 0x21C2, CISS_BOARD_SA5, "HP Smart Array P440" }, { 0x103C, 0x21C3, CISS_BOARD_SA5, "HP Smart Array P441" }, { 0x103C, 0x21C5, CISS_BOARD_SA5, "HP Smart Array P841" }, { 0x103C, 0x21C6, CISS_BOARD_SA5, "HP Smart Array H244br" }, { 0x103C, 0x21C7, CISS_BOARD_SA5, "HP Smart Array H240" }, { 0x103C, 0x21C8, CISS_BOARD_SA5, "HP Smart Array H241" }, { 0x103C, 0x21CA, CISS_BOARD_SA5, "HP Smart Array P246br" }, { 0x103C, 0x21CB, CISS_BOARD_SA5, "HP Smart Array P840" }, { 0x103C, 0x21CC, CISS_BOARD_SA5, "HP Smart Array TBD" }, { 0x103C, 0x21CD, CISS_BOARD_SA5, "HP Smart Array P240nr" }, { 0x103C, 0x21CE, CISS_BOARD_SA5, "HP Smart Array H240nr" }, { 0, 0, 0, NULL } }; /************************************************************************ * Find a match for the device in our list of known adapters. */ static int ciss_lookup(device_t dev) { int i; for (i = 0; ciss_vendor_data[i].desc != NULL; i++) if ((pci_get_subvendor(dev) == ciss_vendor_data[i].subvendor) && (pci_get_subdevice(dev) == ciss_vendor_data[i].subdevice)) { return(i); } return(-1); } /************************************************************************ * Match a known CISS adapter. */ static int ciss_probe(device_t dev) { int i; i = ciss_lookup(dev); if (i != -1) { device_set_desc(dev, ciss_vendor_data[i].desc); return(BUS_PROBE_DEFAULT); } return(ENOENT); } /************************************************************************ * Attach the driver to this adapter. */ static int ciss_attach(device_t dev) { struct ciss_softc *sc; int error; debug_called(1); #ifdef CISS_DEBUG /* print structure/union sizes */ debug_struct(ciss_command); debug_struct(ciss_header); debug_union(ciss_device_address); debug_struct(ciss_cdb); debug_struct(ciss_report_cdb); debug_struct(ciss_notify_cdb); debug_struct(ciss_notify); debug_struct(ciss_message_cdb); debug_struct(ciss_error_info_pointer); debug_struct(ciss_error_info); debug_struct(ciss_sg_entry); debug_struct(ciss_config_table); debug_struct(ciss_bmic_cdb); debug_struct(ciss_bmic_id_ldrive); debug_struct(ciss_bmic_id_lstatus); debug_struct(ciss_bmic_id_table); debug_struct(ciss_bmic_id_pdrive); debug_struct(ciss_bmic_blink_pdrive); debug_struct(ciss_bmic_flush_cache); debug_const(CISS_MAX_REQUESTS); debug_const(CISS_MAX_LOGICAL); debug_const(CISS_INTERRUPT_COALESCE_DELAY); debug_const(CISS_INTERRUPT_COALESCE_COUNT); debug_const(CISS_COMMAND_ALLOC_SIZE); debug_const(CISS_COMMAND_SG_LENGTH); debug_type(cciss_pci_info_struct); debug_type(cciss_coalint_struct); debug_type(cciss_coalint_struct); debug_type(NodeName_type); debug_type(NodeName_type); debug_type(Heartbeat_type); debug_type(BusTypes_type); debug_type(FirmwareVer_type); debug_type(DriverVer_type); debug_type(IOCTL_Command_struct); #endif sc = device_get_softc(dev); sc->ciss_dev = dev; mtx_init(&sc->ciss_mtx, "cissmtx", NULL, MTX_DEF); callout_init_mtx(&sc->ciss_periodic, &sc->ciss_mtx, 0); /* * Do PCI-specific init. */ if ((error = ciss_init_pci(sc)) != 0) goto out; /* * Initialise driver queues. */ ciss_initq_free(sc); ciss_initq_notify(sc); /* * Initialize device sysctls. */ ciss_init_sysctl(sc); /* * Initialise command/request pool. */ if ((error = ciss_init_requests(sc)) != 0) goto out; /* * Get adapter information. */ if ((error = ciss_identify_adapter(sc)) != 0) goto out; /* * Find all the physical devices. */ if ((error = ciss_init_physical(sc)) != 0) goto out; /* * Build our private table of logical devices. */ if ((error = ciss_init_logical(sc)) != 0) goto out; /* * Enable interrupts so that the CAM scan can complete. */ CISS_TL_SIMPLE_ENABLE_INTERRUPTS(sc); /* * Initialise the CAM interface. */ if ((error = ciss_cam_init(sc)) != 0) goto out; /* * Start the heartbeat routine and event chain. */ ciss_periodic(sc); /* * Create the control device. */ sc->ciss_dev_t = make_dev(&ciss_cdevsw, device_get_unit(sc->ciss_dev), UID_ROOT, GID_OPERATOR, S_IRUSR | S_IWUSR, "ciss%d", device_get_unit(sc->ciss_dev)); sc->ciss_dev_t->si_drv1 = sc; /* * The adapter is running; synchronous commands can now sleep * waiting for an interrupt to signal completion. */ sc->ciss_flags |= CISS_FLAG_RUNNING; ciss_spawn_notify_thread(sc); error = 0; out: if (error != 0) { /* ciss_free() expects the mutex to be held */ mtx_lock(&sc->ciss_mtx); ciss_free(sc); } return(error); } /************************************************************************ * Detach the driver from this adapter. */ static int ciss_detach(device_t dev) { struct ciss_softc *sc = device_get_softc(dev); debug_called(1); mtx_lock(&sc->ciss_mtx); if (sc->ciss_flags & CISS_FLAG_CONTROL_OPEN) { mtx_unlock(&sc->ciss_mtx); return (EBUSY); } /* flush adapter cache */ ciss_flush_adapter(sc); /* release all resources. The mutex is released and freed here too. */ ciss_free(sc); return(0); } /************************************************************************ * Prepare adapter for system shutdown. */ static int ciss_shutdown(device_t dev) { struct ciss_softc *sc = device_get_softc(dev); debug_called(1); mtx_lock(&sc->ciss_mtx); /* flush adapter cache */ ciss_flush_adapter(sc); if (sc->ciss_soft_reset) ciss_soft_reset(sc); mtx_unlock(&sc->ciss_mtx); return(0); } static void ciss_init_sysctl(struct ciss_softc *sc) { SYSCTL_ADD_INT(device_get_sysctl_ctx(sc->ciss_dev), SYSCTL_CHILDREN(device_get_sysctl_tree(sc->ciss_dev)), OID_AUTO, "soft_reset", CTLFLAG_RW, &sc->ciss_soft_reset, 0, ""); } /************************************************************************ * Perform PCI-specific attachment actions. */ static int ciss_init_pci(struct ciss_softc *sc) { uintptr_t cbase, csize, cofs; uint32_t method, supported_methods; int error, sqmask, i; void *intr; debug_called(1); /* * Work out adapter type. */ i = ciss_lookup(sc->ciss_dev); if (i < 0) { ciss_printf(sc, "unknown adapter type\n"); return (ENXIO); } if (ciss_vendor_data[i].flags & CISS_BOARD_SA5) { sqmask = CISS_TL_SIMPLE_INTR_OPQ_SA5; } else if (ciss_vendor_data[i].flags & CISS_BOARD_SA5B) { sqmask = CISS_TL_SIMPLE_INTR_OPQ_SA5B; } else { /* * XXX Big hammer, masks/unmasks all possible interrupts. This should * work on all hardware variants. Need to add code to handle the * "controller crashed" interrupt bit that this unmasks. */ sqmask = ~0; } /* * Allocate register window first (we need this to find the config * struct). */ error = ENXIO; sc->ciss_regs_rid = CISS_TL_SIMPLE_BAR_REGS; if ((sc->ciss_regs_resource = bus_alloc_resource_any(sc->ciss_dev, SYS_RES_MEMORY, &sc->ciss_regs_rid, RF_ACTIVE)) == NULL) { ciss_printf(sc, "can't allocate register window\n"); return(ENXIO); } sc->ciss_regs_bhandle = rman_get_bushandle(sc->ciss_regs_resource); sc->ciss_regs_btag = rman_get_bustag(sc->ciss_regs_resource); /* * Find the BAR holding the config structure. If it's not the one * we already mapped for registers, map it too. */ sc->ciss_cfg_rid = CISS_TL_SIMPLE_READ(sc, CISS_TL_SIMPLE_CFG_BAR) & 0xffff; if (sc->ciss_cfg_rid != sc->ciss_regs_rid) { if ((sc->ciss_cfg_resource = bus_alloc_resource_any(sc->ciss_dev, SYS_RES_MEMORY, &sc->ciss_cfg_rid, RF_ACTIVE)) == NULL) { ciss_printf(sc, "can't allocate config window\n"); return(ENXIO); } cbase = (uintptr_t)rman_get_virtual(sc->ciss_cfg_resource); csize = rman_get_end(sc->ciss_cfg_resource) - rman_get_start(sc->ciss_cfg_resource) + 1; } else { cbase = (uintptr_t)rman_get_virtual(sc->ciss_regs_resource); csize = rman_get_end(sc->ciss_regs_resource) - rman_get_start(sc->ciss_regs_resource) + 1; } cofs = CISS_TL_SIMPLE_READ(sc, CISS_TL_SIMPLE_CFG_OFF); /* * Use the base/size/offset values we just calculated to * sanity-check the config structure. If it's OK, point to it. */ if ((cofs + sizeof(struct ciss_config_table)) > csize) { ciss_printf(sc, "config table outside window\n"); return(ENXIO); } sc->ciss_cfg = (struct ciss_config_table *)(cbase + cofs); debug(1, "config struct at %p", sc->ciss_cfg); /* * Calculate the number of request structures/commands we are * going to provide for this adapter. */ sc->ciss_max_requests = min(CISS_MAX_REQUESTS, sc->ciss_cfg->max_outstanding_commands); /* * Validate the config structure. If we supported other transport * methods, we could select amongst them at this point in time. */ if (strncmp(sc->ciss_cfg->signature, "CISS", 4)) { ciss_printf(sc, "config signature mismatch (got '%c%c%c%c')\n", sc->ciss_cfg->signature[0], sc->ciss_cfg->signature[1], sc->ciss_cfg->signature[2], sc->ciss_cfg->signature[3]); return(ENXIO); } /* * Select the mode of operation, prefer Performant. */ if (!(sc->ciss_cfg->supported_methods & (CISS_TRANSPORT_METHOD_SIMPLE | CISS_TRANSPORT_METHOD_PERF))) { ciss_printf(sc, "No supported transport layers: 0x%x\n", sc->ciss_cfg->supported_methods); } switch (ciss_force_transport) { case 1: supported_methods = CISS_TRANSPORT_METHOD_SIMPLE; break; case 2: supported_methods = CISS_TRANSPORT_METHOD_PERF; break; default: /* * Override the capabilities of the BOARD and specify SIMPLE * MODE */ if (ciss_vendor_data[i].flags & CISS_BOARD_SIMPLE) supported_methods = CISS_TRANSPORT_METHOD_SIMPLE; else supported_methods = sc->ciss_cfg->supported_methods; break; } setup: if ((supported_methods & CISS_TRANSPORT_METHOD_PERF) != 0) { method = CISS_TRANSPORT_METHOD_PERF; sc->ciss_perf = (struct ciss_perf_config *)(cbase + cofs + sc->ciss_cfg->transport_offset); if (ciss_init_perf(sc)) { supported_methods &= ~method; goto setup; } } else if (supported_methods & CISS_TRANSPORT_METHOD_SIMPLE) { method = CISS_TRANSPORT_METHOD_SIMPLE; } else { ciss_printf(sc, "No supported transport methods: 0x%x\n", sc->ciss_cfg->supported_methods); return(ENXIO); } /* * Tell it we're using the low 4GB of RAM. Set the default interrupt * coalescing options. */ sc->ciss_cfg->requested_method = method; sc->ciss_cfg->command_physlimit = 0; sc->ciss_cfg->interrupt_coalesce_delay = CISS_INTERRUPT_COALESCE_DELAY; sc->ciss_cfg->interrupt_coalesce_count = CISS_INTERRUPT_COALESCE_COUNT; #ifdef __i386__ sc->ciss_cfg->host_driver |= CISS_DRIVER_SCSI_PREFETCH; #endif if (ciss_update_config(sc)) { ciss_printf(sc, "adapter refuses to accept config update (IDBR 0x%x)\n", CISS_TL_SIMPLE_READ(sc, CISS_TL_SIMPLE_IDBR)); return(ENXIO); } if ((sc->ciss_cfg->active_method & method) == 0) { supported_methods &= ~method; if (supported_methods == 0) { ciss_printf(sc, "adapter refuses to go into available transports " "mode (0x%x, 0x%x)\n", supported_methods, sc->ciss_cfg->active_method); return(ENXIO); } else goto setup; } /* * Wait for the adapter to come ready. */ if ((error = ciss_wait_adapter(sc)) != 0) return(error); /* Prepare to possibly use MSIX and/or PERFORMANT interrupts. Normal * interrupts have a rid of 0, this will be overridden if MSIX is used. */ sc->ciss_irq_rid[0] = 0; if (method == CISS_TRANSPORT_METHOD_PERF) { ciss_printf(sc, "PERFORMANT Transport\n"); if ((ciss_force_interrupt != 1) && (ciss_setup_msix(sc) == 0)) { intr = ciss_perf_msi_intr; } else { intr = ciss_perf_intr; } /* XXX The docs say that the 0x01 bit is only for SAS controllers. * Unfortunately, there is no good way to know if this is a SAS * controller. Hopefully enabling this bit universally will work OK. * It seems to work fine for SA6i controllers. */ sc->ciss_interrupt_mask = CISS_TL_PERF_INTR_OPQ | CISS_TL_PERF_INTR_MSI; } else { ciss_printf(sc, "SIMPLE Transport\n"); /* MSIX doesn't seem to work in SIMPLE mode, only enable if it forced */ if (ciss_force_interrupt == 2) /* If this fails, we automatically revert to INTx */ ciss_setup_msix(sc); sc->ciss_perf = NULL; intr = ciss_intr; sc->ciss_interrupt_mask = sqmask; } /* * Turn off interrupts before we go routing anything. */ CISS_TL_SIMPLE_DISABLE_INTERRUPTS(sc); /* * Allocate and set up our interrupt. */ if ((sc->ciss_irq_resource = bus_alloc_resource_any(sc->ciss_dev, SYS_RES_IRQ, &sc->ciss_irq_rid[0], RF_ACTIVE | RF_SHAREABLE)) == NULL) { ciss_printf(sc, "can't allocate interrupt\n"); return(ENXIO); } if (bus_setup_intr(sc->ciss_dev, sc->ciss_irq_resource, INTR_TYPE_CAM|INTR_MPSAFE, NULL, intr, sc, &sc->ciss_intr)) { ciss_printf(sc, "can't set up interrupt\n"); return(ENXIO); } /* * Allocate the parent bus DMA tag appropriate for our PCI * interface. * * Note that "simple" adapters can only address within a 32-bit * span. */ if (bus_dma_tag_create(bus_get_dma_tag(sc->ciss_dev),/* PCI parent */ 1, 0, /* alignment, boundary */ BUS_SPACE_MAXADDR, /* lowaddr */ BUS_SPACE_MAXADDR, /* highaddr */ NULL, NULL, /* filter, filterarg */ BUS_SPACE_MAXSIZE_32BIT, /* maxsize */ BUS_SPACE_UNRESTRICTED, /* nsegments */ BUS_SPACE_MAXSIZE_32BIT, /* maxsegsize */ 0, /* flags */ NULL, NULL, /* lockfunc, lockarg */ &sc->ciss_parent_dmat)) { ciss_printf(sc, "can't allocate parent DMA tag\n"); return(ENOMEM); } /* * Create DMA tag for mapping buffers into adapter-addressable * space. */ if (bus_dma_tag_create(sc->ciss_parent_dmat, /* parent */ 1, 0, /* alignment, boundary */ BUS_SPACE_MAXADDR, /* lowaddr */ BUS_SPACE_MAXADDR, /* highaddr */ NULL, NULL, /* filter, filterarg */ (CISS_MAX_SG_ELEMENTS - 1) * PAGE_SIZE, /* maxsize */ CISS_MAX_SG_ELEMENTS, /* nsegments */ BUS_SPACE_MAXSIZE_32BIT, /* maxsegsize */ BUS_DMA_ALLOCNOW, /* flags */ busdma_lock_mutex, &sc->ciss_mtx, /* lockfunc, lockarg */ &sc->ciss_buffer_dmat)) { ciss_printf(sc, "can't allocate buffer DMA tag\n"); return(ENOMEM); } return(0); } /************************************************************************ * Setup MSI/MSIX operation (Performant only) * Four interrupts are available, but we only use 1 right now. If MSI-X * isn't avaialble, try using MSI instead. */ static int ciss_setup_msix(struct ciss_softc *sc) { int val, i; /* Weed out devices that don't actually support MSI */ i = ciss_lookup(sc->ciss_dev); if (ciss_vendor_data[i].flags & CISS_BOARD_NOMSI) return (EINVAL); /* * Only need to use the minimum number of MSI vectors, as the driver * doesn't support directed MSIX interrupts. */ val = pci_msix_count(sc->ciss_dev); if (val < CISS_MSI_COUNT) { val = pci_msi_count(sc->ciss_dev); device_printf(sc->ciss_dev, "got %d MSI messages]\n", val); if (val < CISS_MSI_COUNT) return (EINVAL); } val = MIN(val, CISS_MSI_COUNT); if (pci_alloc_msix(sc->ciss_dev, &val) != 0) { if (pci_alloc_msi(sc->ciss_dev, &val) != 0) return (EINVAL); } sc->ciss_msi = val; if (bootverbose) ciss_printf(sc, "Using %d MSIX interrupt%s\n", val, (val != 1) ? "s" : ""); for (i = 0; i < val; i++) sc->ciss_irq_rid[i] = i + 1; return (0); } /************************************************************************ * Setup the Performant structures. */ static int ciss_init_perf(struct ciss_softc *sc) { struct ciss_perf_config *pc = sc->ciss_perf; int reply_size; /* * Create the DMA tag for the reply queue. */ reply_size = sizeof(uint64_t) * sc->ciss_max_requests; if (bus_dma_tag_create(sc->ciss_parent_dmat, /* parent */ 1, 0, /* alignment, boundary */ BUS_SPACE_MAXADDR_32BIT, /* lowaddr */ BUS_SPACE_MAXADDR, /* highaddr */ NULL, NULL, /* filter, filterarg */ reply_size, 1, /* maxsize, nsegments */ BUS_SPACE_MAXSIZE_32BIT, /* maxsegsize */ 0, /* flags */ NULL, NULL, /* lockfunc, lockarg */ &sc->ciss_reply_dmat)) { ciss_printf(sc, "can't allocate reply DMA tag\n"); return(ENOMEM); } /* * Allocate memory and make it available for DMA. */ if (bus_dmamem_alloc(sc->ciss_reply_dmat, (void **)&sc->ciss_reply, BUS_DMA_NOWAIT, &sc->ciss_reply_map)) { ciss_printf(sc, "can't allocate reply memory\n"); return(ENOMEM); } bus_dmamap_load(sc->ciss_reply_dmat, sc->ciss_reply_map, sc->ciss_reply, reply_size, ciss_command_map_helper, &sc->ciss_reply_phys, 0); bzero(sc->ciss_reply, reply_size); sc->ciss_cycle = 0x1; sc->ciss_rqidx = 0; /* * Preload the fetch table with common command sizes. This allows the * hardware to not waste bus cycles for typical i/o commands, but also not * tax the driver to be too exact in choosing sizes. The table is optimized * for page-aligned i/o's, but since most i/o comes from the various pagers, * it's a reasonable assumption to make. */ pc->fetch_count[CISS_SG_FETCH_NONE] = (sizeof(struct ciss_command) + 15) / 16; pc->fetch_count[CISS_SG_FETCH_1] = (sizeof(struct ciss_command) + sizeof(struct ciss_sg_entry) * 1 + 15) / 16; pc->fetch_count[CISS_SG_FETCH_2] = (sizeof(struct ciss_command) + sizeof(struct ciss_sg_entry) * 2 + 15) / 16; pc->fetch_count[CISS_SG_FETCH_4] = (sizeof(struct ciss_command) + sizeof(struct ciss_sg_entry) * 4 + 15) / 16; pc->fetch_count[CISS_SG_FETCH_8] = (sizeof(struct ciss_command) + sizeof(struct ciss_sg_entry) * 8 + 15) / 16; pc->fetch_count[CISS_SG_FETCH_16] = (sizeof(struct ciss_command) + sizeof(struct ciss_sg_entry) * 16 + 15) / 16; pc->fetch_count[CISS_SG_FETCH_32] = (sizeof(struct ciss_command) + sizeof(struct ciss_sg_entry) * 32 + 15) / 16; pc->fetch_count[CISS_SG_FETCH_MAX] = (CISS_COMMAND_ALLOC_SIZE + 15) / 16; pc->rq_size = sc->ciss_max_requests; /* XXX less than the card supports? */ pc->rq_count = 1; /* XXX Hardcode for a single queue */ pc->rq_bank_hi = 0; pc->rq_bank_lo = 0; pc->rq[0].rq_addr_hi = 0x0; pc->rq[0].rq_addr_lo = sc->ciss_reply_phys; return(0); } /************************************************************************ * Wait for the adapter to come ready. */ static int ciss_wait_adapter(struct ciss_softc *sc) { int i; debug_called(1); /* * Wait for the adapter to come ready. */ if (!(sc->ciss_cfg->active_method & CISS_TRANSPORT_METHOD_READY)) { ciss_printf(sc, "waiting for adapter to come ready...\n"); for (i = 0; !(sc->ciss_cfg->active_method & CISS_TRANSPORT_METHOD_READY); i++) { DELAY(1000000); /* one second */ if (i > 30) { ciss_printf(sc, "timed out waiting for adapter to come ready\n"); return(EIO); } } } return(0); } /************************************************************************ * Flush the adapter cache. */ static int ciss_flush_adapter(struct ciss_softc *sc) { struct ciss_request *cr; struct ciss_bmic_flush_cache *cbfc; int error, command_status; debug_called(1); cr = NULL; cbfc = NULL; /* * Build a BMIC request to flush the cache. We don't disable * it, as we may be going to do more I/O (eg. we are emulating * the Synchronise Cache command). */ if ((cbfc = malloc(sizeof(*cbfc), CISS_MALLOC_CLASS, M_NOWAIT | M_ZERO)) == NULL) { error = ENOMEM; goto out; } if ((error = ciss_get_bmic_request(sc, &cr, CISS_BMIC_FLUSH_CACHE, (void **)&cbfc, sizeof(*cbfc))) != 0) goto out; /* * Submit the request and wait for it to complete. */ if ((error = ciss_synch_request(cr, 60 * 1000)) != 0) { ciss_printf(sc, "error sending BMIC FLUSH_CACHE command (%d)\n", error); goto out; } /* * Check response. */ ciss_report_request(cr, &command_status, NULL); switch(command_status) { case CISS_CMD_STATUS_SUCCESS: break; default: ciss_printf(sc, "error flushing cache (%s)\n", ciss_name_command_status(command_status)); error = EIO; goto out; } out: if (cbfc != NULL) free(cbfc, CISS_MALLOC_CLASS); if (cr != NULL) ciss_release_request(cr); return(error); } static void ciss_soft_reset(struct ciss_softc *sc) { struct ciss_request *cr = NULL; struct ciss_command *cc; int i, error = 0; for (i = 0; i < sc->ciss_max_logical_bus; i++) { /* only reset proxy controllers */ if (sc->ciss_controllers[i].physical.bus == 0) continue; if ((error = ciss_get_request(sc, &cr)) != 0) break; if ((error = ciss_get_bmic_request(sc, &cr, CISS_BMIC_SOFT_RESET, NULL, 0)) != 0) break; cc = cr->cr_cc; cc->header.address = sc->ciss_controllers[i]; if ((error = ciss_synch_request(cr, 60 * 1000)) != 0) break; ciss_release_request(cr); } if (error) ciss_printf(sc, "error resetting controller (%d)\n", error); if (cr != NULL) ciss_release_request(cr); } /************************************************************************ * Allocate memory for the adapter command structures, initialise * the request structures. * * Note that the entire set of commands are allocated in a single * contiguous slab. */ static int ciss_init_requests(struct ciss_softc *sc) { struct ciss_request *cr; int i; debug_called(1); if (bootverbose) ciss_printf(sc, "using %d of %d available commands\n", sc->ciss_max_requests, sc->ciss_cfg->max_outstanding_commands); /* * Create the DMA tag for commands. */ if (bus_dma_tag_create(sc->ciss_parent_dmat, /* parent */ 32, 0, /* alignment, boundary */ BUS_SPACE_MAXADDR_32BIT, /* lowaddr */ BUS_SPACE_MAXADDR, /* highaddr */ NULL, NULL, /* filter, filterarg */ CISS_COMMAND_ALLOC_SIZE * sc->ciss_max_requests, 1, /* maxsize, nsegments */ BUS_SPACE_MAXSIZE_32BIT, /* maxsegsize */ 0, /* flags */ NULL, NULL, /* lockfunc, lockarg */ &sc->ciss_command_dmat)) { ciss_printf(sc, "can't allocate command DMA tag\n"); return(ENOMEM); } /* * Allocate memory and make it available for DMA. */ if (bus_dmamem_alloc(sc->ciss_command_dmat, (void **)&sc->ciss_command, BUS_DMA_NOWAIT, &sc->ciss_command_map)) { ciss_printf(sc, "can't allocate command memory\n"); return(ENOMEM); } bus_dmamap_load(sc->ciss_command_dmat, sc->ciss_command_map,sc->ciss_command, CISS_COMMAND_ALLOC_SIZE * sc->ciss_max_requests, ciss_command_map_helper, &sc->ciss_command_phys, 0); bzero(sc->ciss_command, CISS_COMMAND_ALLOC_SIZE * sc->ciss_max_requests); /* * Set up the request and command structures, push requests onto * the free queue. */ for (i = 1; i < sc->ciss_max_requests; i++) { cr = &sc->ciss_request[i]; cr->cr_sc = sc; cr->cr_tag = i; cr->cr_cc = (struct ciss_command *)((uintptr_t)sc->ciss_command + CISS_COMMAND_ALLOC_SIZE * i); cr->cr_ccphys = sc->ciss_command_phys + CISS_COMMAND_ALLOC_SIZE * i; bus_dmamap_create(sc->ciss_buffer_dmat, 0, &cr->cr_datamap); ciss_enqueue_free(cr); } return(0); } static void ciss_command_map_helper(void *arg, bus_dma_segment_t *segs, int nseg, int error) { uint32_t *addr; addr = arg; *addr = segs[0].ds_addr; } /************************************************************************ * Identify the adapter, print some information about it. */ static int ciss_identify_adapter(struct ciss_softc *sc) { struct ciss_request *cr; int error, command_status; debug_called(1); cr = NULL; /* * Get a request, allocate storage for the adapter data. */ if ((error = ciss_get_bmic_request(sc, &cr, CISS_BMIC_ID_CTLR, (void **)&sc->ciss_id, sizeof(*sc->ciss_id))) != 0) goto out; /* * Submit the request and wait for it to complete. */ if ((error = ciss_synch_request(cr, 60 * 1000)) != 0) { ciss_printf(sc, "error sending BMIC ID_CTLR command (%d)\n", error); goto out; } /* * Check response. */ ciss_report_request(cr, &command_status, NULL); switch(command_status) { case CISS_CMD_STATUS_SUCCESS: /* buffer right size */ break; case CISS_CMD_STATUS_DATA_UNDERRUN: case CISS_CMD_STATUS_DATA_OVERRUN: ciss_printf(sc, "data over/underrun reading adapter information\n"); default: ciss_printf(sc, "error reading adapter information (%s)\n", ciss_name_command_status(command_status)); error = EIO; goto out; } /* sanity-check reply */ if (!(sc->ciss_id->controller_flags & CONTROLLER_FLAGS_BIG_MAP_SUPPORT)) { ciss_printf(sc, "adapter does not support BIG_MAP\n"); error = ENXIO; goto out; } #if 0 /* XXX later revisions may not need this */ sc->ciss_flags |= CISS_FLAG_FAKE_SYNCH; #endif /* XXX only really required for old 5300 adapters? */ sc->ciss_flags |= CISS_FLAG_BMIC_ABORT; /* * Earlier controller specs do not contain these config * entries, so assume that a 0 means its old and assign * these values to the defaults that were established * when this driver was developed for them */ if (sc->ciss_cfg->max_logical_supported == 0) sc->ciss_cfg->max_logical_supported = CISS_MAX_LOGICAL; if (sc->ciss_cfg->max_physical_supported == 0) sc->ciss_cfg->max_physical_supported = CISS_MAX_PHYSICAL; /* print information */ if (bootverbose) { ciss_printf(sc, " %d logical drive%s configured\n", sc->ciss_id->configured_logical_drives, (sc->ciss_id->configured_logical_drives == 1) ? "" : "s"); ciss_printf(sc, " firmware %4.4s\n", sc->ciss_id->running_firmware_revision); ciss_printf(sc, " %d SCSI channels\n", sc->ciss_id->scsi_chip_count); ciss_printf(sc, " signature '%.4s'\n", sc->ciss_cfg->signature); ciss_printf(sc, " valence %d\n", sc->ciss_cfg->valence); ciss_printf(sc, " supported I/O methods 0x%b\n", sc->ciss_cfg->supported_methods, "\20\1READY\2simple\3performant\4MEMQ\n"); ciss_printf(sc, " active I/O method 0x%b\n", sc->ciss_cfg->active_method, "\20\2simple\3performant\4MEMQ\n"); ciss_printf(sc, " 4G page base 0x%08x\n", sc->ciss_cfg->command_physlimit); ciss_printf(sc, " interrupt coalesce delay %dus\n", sc->ciss_cfg->interrupt_coalesce_delay); ciss_printf(sc, " interrupt coalesce count %d\n", sc->ciss_cfg->interrupt_coalesce_count); ciss_printf(sc, " max outstanding commands %d\n", sc->ciss_cfg->max_outstanding_commands); ciss_printf(sc, " bus types 0x%b\n", sc->ciss_cfg->bus_types, "\20\1ultra2\2ultra3\10fibre1\11fibre2\n"); ciss_printf(sc, " server name '%.16s'\n", sc->ciss_cfg->server_name); ciss_printf(sc, " heartbeat 0x%x\n", sc->ciss_cfg->heartbeat); ciss_printf(sc, " max logical logical volumes: %d\n", sc->ciss_cfg->max_logical_supported); ciss_printf(sc, " max physical disks supported: %d\n", sc->ciss_cfg->max_physical_supported); ciss_printf(sc, " max physical disks per logical volume: %d\n", sc->ciss_cfg->max_physical_per_logical); ciss_printf(sc, " JBOD Support is %s\n", (sc->ciss_id->uiYetMoreControllerFlags & YMORE_CONTROLLER_FLAGS_JBOD_SUPPORTED) ? "Available" : "Unavailable"); ciss_printf(sc, " JBOD Mode is %s\n", (sc->ciss_id->PowerUPNvramFlags & PWR_UP_FLAG_JBOD_ENABLED) ? "Enabled" : "Disabled"); } out: if (error) { if (sc->ciss_id != NULL) { free(sc->ciss_id, CISS_MALLOC_CLASS); sc->ciss_id = NULL; } } if (cr != NULL) ciss_release_request(cr); return(error); } /************************************************************************ * Helper routine for generating a list of logical and physical luns. */ static struct ciss_lun_report * ciss_report_luns(struct ciss_softc *sc, int opcode, int nunits) { struct ciss_request *cr; struct ciss_command *cc; struct ciss_report_cdb *crc; struct ciss_lun_report *cll; int command_status; int report_size; int error = 0; debug_called(1); cr = NULL; cll = NULL; /* * Get a request, allocate storage for the address list. */ if ((error = ciss_get_request(sc, &cr)) != 0) goto out; report_size = sizeof(*cll) + nunits * sizeof(union ciss_device_address); if ((cll = malloc(report_size, CISS_MALLOC_CLASS, M_NOWAIT | M_ZERO)) == NULL) { ciss_printf(sc, "can't allocate memory for lun report\n"); error = ENOMEM; goto out; } /* * Build the Report Logical/Physical LUNs command. */ cc = cr->cr_cc; cr->cr_data = cll; cr->cr_length = report_size; cr->cr_flags = CISS_REQ_DATAIN; cc->header.address.physical.mode = CISS_HDR_ADDRESS_MODE_PERIPHERAL; cc->header.address.physical.bus = 0; cc->header.address.physical.target = 0; cc->cdb.cdb_length = sizeof(*crc); cc->cdb.type = CISS_CDB_TYPE_COMMAND; cc->cdb.attribute = CISS_CDB_ATTRIBUTE_SIMPLE; cc->cdb.direction = CISS_CDB_DIRECTION_READ; cc->cdb.timeout = 30; /* XXX better suggestions? */ crc = (struct ciss_report_cdb *)&(cc->cdb.cdb[0]); bzero(crc, sizeof(*crc)); crc->opcode = opcode; crc->length = htonl(report_size); /* big-endian field */ cll->list_size = htonl(report_size - sizeof(*cll)); /* big-endian field */ /* * Submit the request and wait for it to complete. (timeout * here should be much greater than above) */ if ((error = ciss_synch_request(cr, 60 * 1000)) != 0) { ciss_printf(sc, "error sending %d LUN command (%d)\n", opcode, error); goto out; } /* * Check response. Note that data over/underrun is OK. */ ciss_report_request(cr, &command_status, NULL); switch(command_status) { case CISS_CMD_STATUS_SUCCESS: /* buffer right size */ case CISS_CMD_STATUS_DATA_UNDERRUN: /* buffer too large, not bad */ break; case CISS_CMD_STATUS_DATA_OVERRUN: ciss_printf(sc, "WARNING: more units than driver limit (%d)\n", sc->ciss_cfg->max_logical_supported); break; default: ciss_printf(sc, "error detecting logical drive configuration (%s)\n", ciss_name_command_status(command_status)); error = EIO; goto out; } ciss_release_request(cr); cr = NULL; out: if (cr != NULL) ciss_release_request(cr); if (error && cll != NULL) { free(cll, CISS_MALLOC_CLASS); cll = NULL; } return(cll); } /************************************************************************ * Find logical drives on the adapter. */ static int ciss_init_logical(struct ciss_softc *sc) { struct ciss_lun_report *cll; int error = 0, i, j; int ndrives; debug_called(1); cll = ciss_report_luns(sc, CISS_OPCODE_REPORT_LOGICAL_LUNS, sc->ciss_cfg->max_logical_supported); if (cll == NULL) { error = ENXIO; goto out; } /* sanity-check reply */ ndrives = (ntohl(cll->list_size) / sizeof(union ciss_device_address)); if ((ndrives < 0) || (ndrives > sc->ciss_cfg->max_logical_supported)) { ciss_printf(sc, "adapter claims to report absurd number of logical drives (%d > %d)\n", ndrives, sc->ciss_cfg->max_logical_supported); error = ENXIO; goto out; } /* * Save logical drive information. */ if (bootverbose) { ciss_printf(sc, "%d logical drive%s\n", ndrives, (ndrives > 1 || ndrives == 0) ? "s" : ""); } sc->ciss_logical = malloc(sc->ciss_max_logical_bus * sizeof(struct ciss_ldrive *), CISS_MALLOC_CLASS, M_NOWAIT | M_ZERO); if (sc->ciss_logical == NULL) { error = ENXIO; goto out; } for (i = 0; i < sc->ciss_max_logical_bus; i++) { sc->ciss_logical[i] = malloc(sc->ciss_cfg->max_logical_supported * sizeof(struct ciss_ldrive), CISS_MALLOC_CLASS, M_NOWAIT | M_ZERO); if (sc->ciss_logical[i] == NULL) { error = ENXIO; goto out; } for (j = 0; j < sc->ciss_cfg->max_logical_supported; j++) sc->ciss_logical[i][j].cl_status = CISS_LD_NONEXISTENT; } for (i = 0; i < sc->ciss_cfg->max_logical_supported; i++) { if (i < ndrives) { struct ciss_ldrive *ld; int bus, target; bus = CISS_LUN_TO_BUS(cll->lun[i].logical.lun); target = CISS_LUN_TO_TARGET(cll->lun[i].logical.lun); ld = &sc->ciss_logical[bus][target]; ld->cl_address = cll->lun[i]; ld->cl_controller = &sc->ciss_controllers[bus]; if (ciss_identify_logical(sc, ld) != 0) continue; /* * If the drive has had media exchanged, we should bring it online. */ if (ld->cl_lstatus->media_exchanged) ciss_accept_media(sc, ld); } } out: if (cll != NULL) free(cll, CISS_MALLOC_CLASS); return(error); } static int ciss_init_physical(struct ciss_softc *sc) { struct ciss_lun_report *cll; int error = 0, i; int nphys; int bus, target; debug_called(1); bus = 0; target = 0; cll = ciss_report_luns(sc, CISS_OPCODE_REPORT_PHYSICAL_LUNS, sc->ciss_cfg->max_physical_supported); if (cll == NULL) { error = ENXIO; goto out; } nphys = (ntohl(cll->list_size) / sizeof(union ciss_device_address)); if (bootverbose) { ciss_printf(sc, "%d physical device%s\n", nphys, (nphys > 1 || nphys == 0) ? "s" : ""); } /* * Figure out the bus mapping. * Logical buses include both the local logical bus for local arrays and * proxy buses for remote arrays. Physical buses are numbered by the * controller and represent physical buses that hold physical devices. * We shift these bus numbers so that everything fits into a single flat * numbering space for CAM. Logical buses occupy the first 32 CAM bus * numbers, and the physical bus numbers are shifted to be above that. * This results in the various driver arrays being indexed as follows: * * ciss_controllers[] - indexed by logical bus * ciss_cam_sim[] - indexed by both logical and physical, with physical * being shifted by 32. * ciss_logical[][] - indexed by logical bus * ciss_physical[][] - indexed by physical bus * * XXX This is getting more and more hackish. CISS really doesn't play * well with a standard SCSI model; devices are addressed via magic * cookies, not via b/t/l addresses. Since there is no way to store * the cookie in the CAM device object, we have to keep these lookup * tables handy so that the devices can be found quickly at the cost * of wasting memory and having a convoluted lookup scheme. This * driver should probably be converted to block interface. */ /* * If the L2 and L3 SCSI addresses are 0, this signifies a proxy * controller. A proxy controller is another physical controller * behind the primary PCI controller. We need to know about this * so that BMIC commands can be properly targeted. There can be * proxy controllers attached to a single PCI controller, so * find the highest numbered one so the array can be properly * sized. */ sc->ciss_max_logical_bus = 1; for (i = 0; i < nphys; i++) { if (cll->lun[i].physical.extra_address == 0) { bus = cll->lun[i].physical.bus; sc->ciss_max_logical_bus = max(sc->ciss_max_logical_bus, bus) + 1; } else { bus = CISS_EXTRA_BUS2(cll->lun[i].physical.extra_address); sc->ciss_max_physical_bus = max(sc->ciss_max_physical_bus, bus); } } sc->ciss_controllers = malloc(sc->ciss_max_logical_bus * sizeof (union ciss_device_address), CISS_MALLOC_CLASS, M_NOWAIT | M_ZERO); if (sc->ciss_controllers == NULL) { ciss_printf(sc, "Could not allocate memory for controller map\n"); error = ENOMEM; goto out; } /* setup a map of controller addresses */ for (i = 0; i < nphys; i++) { if (cll->lun[i].physical.extra_address == 0) { sc->ciss_controllers[cll->lun[i].physical.bus] = cll->lun[i]; } } sc->ciss_physical = malloc(sc->ciss_max_physical_bus * sizeof(struct ciss_pdrive *), CISS_MALLOC_CLASS, M_NOWAIT | M_ZERO); if (sc->ciss_physical == NULL) { ciss_printf(sc, "Could not allocate memory for physical device map\n"); error = ENOMEM; goto out; } for (i = 0; i < sc->ciss_max_physical_bus; i++) { sc->ciss_physical[i] = malloc(sizeof(struct ciss_pdrive) * CISS_MAX_PHYSTGT, CISS_MALLOC_CLASS, M_NOWAIT | M_ZERO); if (sc->ciss_physical[i] == NULL) { ciss_printf(sc, "Could not allocate memory for target map\n"); error = ENOMEM; goto out; } } ciss_filter_physical(sc, cll); out: if (cll != NULL) free(cll, CISS_MALLOC_CLASS); return(error); } static int ciss_filter_physical(struct ciss_softc *sc, struct ciss_lun_report *cll) { u_int32_t ea; int i, nphys; int bus, target; nphys = (ntohl(cll->list_size) / sizeof(union ciss_device_address)); for (i = 0; i < nphys; i++) { if (cll->lun[i].physical.extra_address == 0) continue; /* * Filter out devices that we don't want. Level 3 LUNs could * probably be supported, but the docs don't give enough of a * hint to know how. * * The mode field of the physical address is likely set to have * hard disks masked out. Honor it unless the user has overridden * us with the tunable. We also munge the inquiry data for these * disks so that they only show up as passthrough devices. Keeping * them visible in this fashion is useful for doing things like * flashing firmware. */ ea = cll->lun[i].physical.extra_address; if ((CISS_EXTRA_BUS3(ea) != 0) || (CISS_EXTRA_TARGET3(ea) != 0) || (CISS_EXTRA_MODE2(ea) == 0x3)) continue; if ((ciss_expose_hidden_physical == 0) && (cll->lun[i].physical.mode == CISS_HDR_ADDRESS_MODE_MASK_PERIPHERAL)) continue; /* * Note: CISS firmware numbers physical busses starting at '1', not * '0'. This numbering is internal to the firmware and is only * used as a hint here. */ bus = CISS_EXTRA_BUS2(ea) - 1; target = CISS_EXTRA_TARGET2(ea); sc->ciss_physical[bus][target].cp_address = cll->lun[i]; sc->ciss_physical[bus][target].cp_online = 1; } return (0); } static int ciss_inquiry_logical(struct ciss_softc *sc, struct ciss_ldrive *ld) { struct ciss_request *cr; struct ciss_command *cc; struct scsi_inquiry *inq; int error; int command_status; cr = NULL; bzero(&ld->cl_geometry, sizeof(ld->cl_geometry)); if ((error = ciss_get_request(sc, &cr)) != 0) goto out; cc = cr->cr_cc; cr->cr_data = &ld->cl_geometry; cr->cr_length = sizeof(ld->cl_geometry); cr->cr_flags = CISS_REQ_DATAIN; cc->header.address = ld->cl_address; cc->cdb.cdb_length = 6; cc->cdb.type = CISS_CDB_TYPE_COMMAND; cc->cdb.attribute = CISS_CDB_ATTRIBUTE_SIMPLE; cc->cdb.direction = CISS_CDB_DIRECTION_READ; cc->cdb.timeout = 30; inq = (struct scsi_inquiry *)&(cc->cdb.cdb[0]); inq->opcode = INQUIRY; inq->byte2 = SI_EVPD; inq->page_code = CISS_VPD_LOGICAL_DRIVE_GEOMETRY; scsi_ulto2b(sizeof(ld->cl_geometry), inq->length); if ((error = ciss_synch_request(cr, 60 * 1000)) != 0) { ciss_printf(sc, "error getting geometry (%d)\n", error); goto out; } ciss_report_request(cr, &command_status, NULL); switch(command_status) { case CISS_CMD_STATUS_SUCCESS: case CISS_CMD_STATUS_DATA_UNDERRUN: break; case CISS_CMD_STATUS_DATA_OVERRUN: ciss_printf(sc, "WARNING: Data overrun\n"); break; default: ciss_printf(sc, "Error detecting logical drive geometry (%s)\n", ciss_name_command_status(command_status)); break; } out: if (cr != NULL) ciss_release_request(cr); return(error); } /************************************************************************ * Identify a logical drive, initialise state related to it. */ static int ciss_identify_logical(struct ciss_softc *sc, struct ciss_ldrive *ld) { struct ciss_request *cr; struct ciss_command *cc; struct ciss_bmic_cdb *cbc; int error, command_status; debug_called(1); cr = NULL; /* * Build a BMIC request to fetch the drive ID. */ if ((error = ciss_get_bmic_request(sc, &cr, CISS_BMIC_ID_LDRIVE, (void **)&ld->cl_ldrive, sizeof(*ld->cl_ldrive))) != 0) goto out; cc = cr->cr_cc; cc->header.address = *ld->cl_controller; /* target controller */ cbc = (struct ciss_bmic_cdb *)&(cc->cdb.cdb[0]); cbc->log_drive = CISS_LUN_TO_TARGET(ld->cl_address.logical.lun); /* * Submit the request and wait for it to complete. */ if ((error = ciss_synch_request(cr, 60 * 1000)) != 0) { ciss_printf(sc, "error sending BMIC LDRIVE command (%d)\n", error); goto out; } /* * Check response. */ ciss_report_request(cr, &command_status, NULL); switch(command_status) { case CISS_CMD_STATUS_SUCCESS: /* buffer right size */ break; case CISS_CMD_STATUS_DATA_UNDERRUN: case CISS_CMD_STATUS_DATA_OVERRUN: ciss_printf(sc, "data over/underrun reading logical drive ID\n"); default: ciss_printf(sc, "error reading logical drive ID (%s)\n", ciss_name_command_status(command_status)); error = EIO; goto out; } ciss_release_request(cr); cr = NULL; /* * Build a CISS BMIC command to get the logical drive status. */ if ((error = ciss_get_ldrive_status(sc, ld)) != 0) goto out; /* * Get the logical drive geometry. */ if ((error = ciss_inquiry_logical(sc, ld)) != 0) goto out; /* * Print the drive's basic characteristics. */ if (bootverbose) { ciss_printf(sc, "logical drive (b%dt%d): %s, %dMB ", CISS_LUN_TO_BUS(ld->cl_address.logical.lun), CISS_LUN_TO_TARGET(ld->cl_address.logical.lun), ciss_name_ldrive_org(ld->cl_ldrive->fault_tolerance), ((ld->cl_ldrive->blocks_available / (1024 * 1024)) * ld->cl_ldrive->block_size)); ciss_print_ldrive(sc, ld); } out: if (error != 0) { /* make the drive not-exist */ ld->cl_status = CISS_LD_NONEXISTENT; if (ld->cl_ldrive != NULL) { free(ld->cl_ldrive, CISS_MALLOC_CLASS); ld->cl_ldrive = NULL; } if (ld->cl_lstatus != NULL) { free(ld->cl_lstatus, CISS_MALLOC_CLASS); ld->cl_lstatus = NULL; } } if (cr != NULL) ciss_release_request(cr); return(error); } /************************************************************************ * Get status for a logical drive. * * XXX should we also do this in response to Test Unit Ready? */ static int ciss_get_ldrive_status(struct ciss_softc *sc, struct ciss_ldrive *ld) { struct ciss_request *cr; struct ciss_command *cc; struct ciss_bmic_cdb *cbc; int error, command_status; /* * Build a CISS BMIC command to get the logical drive status. */ if ((error = ciss_get_bmic_request(sc, &cr, CISS_BMIC_ID_LSTATUS, (void **)&ld->cl_lstatus, sizeof(*ld->cl_lstatus))) != 0) goto out; cc = cr->cr_cc; cc->header.address = *ld->cl_controller; /* target controller */ cbc = (struct ciss_bmic_cdb *)&(cc->cdb.cdb[0]); cbc->log_drive = CISS_LUN_TO_TARGET(ld->cl_address.logical.lun); /* * Submit the request and wait for it to complete. */ if ((error = ciss_synch_request(cr, 60 * 1000)) != 0) { ciss_printf(sc, "error sending BMIC LSTATUS command (%d)\n", error); goto out; } /* * Check response. */ ciss_report_request(cr, &command_status, NULL); switch(command_status) { case CISS_CMD_STATUS_SUCCESS: /* buffer right size */ break; case CISS_CMD_STATUS_DATA_UNDERRUN: case CISS_CMD_STATUS_DATA_OVERRUN: ciss_printf(sc, "data over/underrun reading logical drive status\n"); default: ciss_printf(sc, "error reading logical drive status (%s)\n", ciss_name_command_status(command_status)); error = EIO; goto out; } /* * Set the drive's summary status based on the returned status. * * XXX testing shows that a failed JBOD drive comes back at next * boot in "queued for expansion" mode. WTF? */ ld->cl_status = ciss_decode_ldrive_status(ld->cl_lstatus->status); out: if (cr != NULL) ciss_release_request(cr); return(error); } /************************************************************************ * Notify the adapter of a config update. */ static int ciss_update_config(struct ciss_softc *sc) { int i; debug_called(1); CISS_TL_SIMPLE_WRITE(sc, CISS_TL_SIMPLE_IDBR, CISS_TL_SIMPLE_IDBR_CFG_TABLE); for (i = 0; i < 1000; i++) { if (!(CISS_TL_SIMPLE_READ(sc, CISS_TL_SIMPLE_IDBR) & CISS_TL_SIMPLE_IDBR_CFG_TABLE)) { return(0); } DELAY(1000); } return(1); } /************************************************************************ * Accept new media into a logical drive. * * XXX The drive has previously been offline; it would be good if we * could make sure it's not open right now. */ static int ciss_accept_media(struct ciss_softc *sc, struct ciss_ldrive *ld) { struct ciss_request *cr; struct ciss_command *cc; struct ciss_bmic_cdb *cbc; int command_status; int error = 0, ldrive; ldrive = CISS_LUN_TO_TARGET(ld->cl_address.logical.lun); debug(0, "bringing logical drive %d back online", ldrive); /* * Build a CISS BMIC command to bring the drive back online. */ if ((error = ciss_get_bmic_request(sc, &cr, CISS_BMIC_ACCEPT_MEDIA, NULL, 0)) != 0) goto out; cc = cr->cr_cc; cc->header.address = *ld->cl_controller; /* target controller */ cbc = (struct ciss_bmic_cdb *)&(cc->cdb.cdb[0]); cbc->log_drive = ldrive; /* * Submit the request and wait for it to complete. */ if ((error = ciss_synch_request(cr, 60 * 1000)) != 0) { ciss_printf(sc, "error sending BMIC ACCEPT MEDIA command (%d)\n", error); goto out; } /* * Check response. */ ciss_report_request(cr, &command_status, NULL); switch(command_status) { case CISS_CMD_STATUS_SUCCESS: /* all OK */ /* we should get a logical drive status changed event here */ break; default: ciss_printf(cr->cr_sc, "error accepting media into failed logical drive (%s)\n", ciss_name_command_status(command_status)); break; } out: if (cr != NULL) ciss_release_request(cr); return(error); } /************************************************************************ * Release adapter resources. */ static void ciss_free(struct ciss_softc *sc) { struct ciss_request *cr; int i, j; debug_called(1); /* we're going away */ sc->ciss_flags |= CISS_FLAG_ABORTING; /* terminate the periodic heartbeat routine */ callout_stop(&sc->ciss_periodic); /* cancel the Event Notify chain */ ciss_notify_abort(sc); ciss_kill_notify_thread(sc); /* disconnect from CAM */ if (sc->ciss_cam_sim) { for (i = 0; i < sc->ciss_max_logical_bus; i++) { if (sc->ciss_cam_sim[i]) { xpt_bus_deregister(cam_sim_path(sc->ciss_cam_sim[i])); cam_sim_free(sc->ciss_cam_sim[i], 0); } } for (i = CISS_PHYSICAL_BASE; i < sc->ciss_max_physical_bus + CISS_PHYSICAL_BASE; i++) { if (sc->ciss_cam_sim[i]) { xpt_bus_deregister(cam_sim_path(sc->ciss_cam_sim[i])); cam_sim_free(sc->ciss_cam_sim[i], 0); } } free(sc->ciss_cam_sim, CISS_MALLOC_CLASS); } if (sc->ciss_cam_devq) cam_simq_free(sc->ciss_cam_devq); /* remove the control device */ mtx_unlock(&sc->ciss_mtx); if (sc->ciss_dev_t != NULL) destroy_dev(sc->ciss_dev_t); /* Final cleanup of the callout. */ callout_drain(&sc->ciss_periodic); mtx_destroy(&sc->ciss_mtx); /* free the controller data */ if (sc->ciss_id != NULL) free(sc->ciss_id, CISS_MALLOC_CLASS); /* release I/O resources */ if (sc->ciss_regs_resource != NULL) bus_release_resource(sc->ciss_dev, SYS_RES_MEMORY, sc->ciss_regs_rid, sc->ciss_regs_resource); if (sc->ciss_cfg_resource != NULL) bus_release_resource(sc->ciss_dev, SYS_RES_MEMORY, sc->ciss_cfg_rid, sc->ciss_cfg_resource); if (sc->ciss_intr != NULL) bus_teardown_intr(sc->ciss_dev, sc->ciss_irq_resource, sc->ciss_intr); if (sc->ciss_irq_resource != NULL) bus_release_resource(sc->ciss_dev, SYS_RES_IRQ, sc->ciss_irq_rid[0], sc->ciss_irq_resource); if (sc->ciss_msi) pci_release_msi(sc->ciss_dev); while ((cr = ciss_dequeue_free(sc)) != NULL) bus_dmamap_destroy(sc->ciss_buffer_dmat, cr->cr_datamap); if (sc->ciss_buffer_dmat) bus_dma_tag_destroy(sc->ciss_buffer_dmat); /* destroy command memory and DMA tag */ if (sc->ciss_command != NULL) { bus_dmamap_unload(sc->ciss_command_dmat, sc->ciss_command_map); bus_dmamem_free(sc->ciss_command_dmat, sc->ciss_command, sc->ciss_command_map); } if (sc->ciss_command_dmat) bus_dma_tag_destroy(sc->ciss_command_dmat); if (sc->ciss_reply) { bus_dmamap_unload(sc->ciss_reply_dmat, sc->ciss_reply_map); bus_dmamem_free(sc->ciss_reply_dmat, sc->ciss_reply, sc->ciss_reply_map); } if (sc->ciss_reply_dmat) bus_dma_tag_destroy(sc->ciss_reply_dmat); /* destroy DMA tags */ if (sc->ciss_parent_dmat) bus_dma_tag_destroy(sc->ciss_parent_dmat); if (sc->ciss_logical) { for (i = 0; i < sc->ciss_max_logical_bus; i++) { for (j = 0; j < sc->ciss_cfg->max_logical_supported; j++) { if (sc->ciss_logical[i][j].cl_ldrive) free(sc->ciss_logical[i][j].cl_ldrive, CISS_MALLOC_CLASS); if (sc->ciss_logical[i][j].cl_lstatus) free(sc->ciss_logical[i][j].cl_lstatus, CISS_MALLOC_CLASS); } free(sc->ciss_logical[i], CISS_MALLOC_CLASS); } free(sc->ciss_logical, CISS_MALLOC_CLASS); } if (sc->ciss_physical) { for (i = 0; i < sc->ciss_max_physical_bus; i++) free(sc->ciss_physical[i], CISS_MALLOC_CLASS); free(sc->ciss_physical, CISS_MALLOC_CLASS); } if (sc->ciss_controllers) free(sc->ciss_controllers, CISS_MALLOC_CLASS); } /************************************************************************ * Give a command to the adapter. * * Note that this uses the simple transport layer directly. If we * want to add support for other layers, we'll need a switch of some * sort. * * Note that the simple transport layer has no way of refusing a * command; we only have as many request structures as the adapter * supports commands, so we don't have to check (this presumes that * the adapter can handle commands as fast as we throw them at it). */ static int ciss_start(struct ciss_request *cr) { struct ciss_command *cc; /* XXX debugging only */ int error; cc = cr->cr_cc; debug(2, "post command %d tag %d ", cr->cr_tag, cc->header.host_tag); /* * Map the request's data. */ if ((error = ciss_map_request(cr))) return(error); #if 0 ciss_print_request(cr); #endif return(0); } /************************************************************************ * Fetch completed request(s) from the adapter, queue them for * completion handling. * * Note that this uses the simple transport layer directly. If we * want to add support for other layers, we'll need a switch of some * sort. * * Note that the simple transport mechanism does not require any * reentrancy protection; the OPQ read is atomic. If there is a * chance of a race with something else that might move the request * off the busy list, then we will have to lock against that * (eg. timeouts, etc.) */ static void ciss_done(struct ciss_softc *sc, cr_qhead_t *qh) { struct ciss_request *cr; struct ciss_command *cc; u_int32_t tag, index; debug_called(3); /* * Loop quickly taking requests from the adapter and moving them * to the completed queue. */ for (;;) { tag = CISS_TL_SIMPLE_FETCH_CMD(sc); if (tag == CISS_TL_SIMPLE_OPQ_EMPTY) break; index = tag >> 2; debug(2, "completed command %d%s", index, (tag & CISS_HDR_HOST_TAG_ERROR) ? " with error" : ""); if (index >= sc->ciss_max_requests) { ciss_printf(sc, "completed invalid request %d (0x%x)\n", index, tag); continue; } cr = &(sc->ciss_request[index]); cc = cr->cr_cc; cc->header.host_tag = tag; /* not updated by adapter */ ciss_enqueue_complete(cr, qh); } } static void ciss_perf_done(struct ciss_softc *sc, cr_qhead_t *qh) { struct ciss_request *cr; struct ciss_command *cc; u_int32_t tag, index; debug_called(3); /* * Loop quickly taking requests from the adapter and moving them * to the completed queue. */ for (;;) { tag = sc->ciss_reply[sc->ciss_rqidx]; if ((tag & CISS_CYCLE_MASK) != sc->ciss_cycle) break; index = tag >> 2; debug(2, "completed command %d%s\n", index, (tag & CISS_HDR_HOST_TAG_ERROR) ? " with error" : ""); if (index < sc->ciss_max_requests) { cr = &(sc->ciss_request[index]); cc = cr->cr_cc; cc->header.host_tag = tag; /* not updated by adapter */ ciss_enqueue_complete(cr, qh); } else { ciss_printf(sc, "completed invalid request %d (0x%x)\n", index, tag); } if (++sc->ciss_rqidx == sc->ciss_max_requests) { sc->ciss_rqidx = 0; sc->ciss_cycle ^= 1; } } } /************************************************************************ * Take an interrupt from the adapter. */ static void ciss_intr(void *arg) { cr_qhead_t qh; struct ciss_softc *sc = (struct ciss_softc *)arg; /* * The only interrupt we recognise indicates that there are * entries in the outbound post queue. */ STAILQ_INIT(&qh); ciss_done(sc, &qh); mtx_lock(&sc->ciss_mtx); ciss_complete(sc, &qh); mtx_unlock(&sc->ciss_mtx); } static void ciss_perf_intr(void *arg) { struct ciss_softc *sc = (struct ciss_softc *)arg; /* Clear the interrupt and flush the bridges. Docs say that the flush * needs to be done twice, which doesn't seem right. */ CISS_TL_PERF_CLEAR_INT(sc); CISS_TL_PERF_FLUSH_INT(sc); ciss_perf_msi_intr(sc); } static void ciss_perf_msi_intr(void *arg) { cr_qhead_t qh; struct ciss_softc *sc = (struct ciss_softc *)arg; STAILQ_INIT(&qh); ciss_perf_done(sc, &qh); mtx_lock(&sc->ciss_mtx); ciss_complete(sc, &qh); mtx_unlock(&sc->ciss_mtx); } /************************************************************************ * Process completed requests. * * Requests can be completed in three fashions: * * - by invoking a callback function (cr_complete is non-null) * - by waking up a sleeper (cr_flags has CISS_REQ_SLEEP set) * - by clearing the CISS_REQ_POLL flag in interrupt/timeout context */ static void ciss_complete(struct ciss_softc *sc, cr_qhead_t *qh) { struct ciss_request *cr; debug_called(2); /* * Loop taking requests off the completed queue and performing * completion processing on them. */ for (;;) { if ((cr = ciss_dequeue_complete(sc, qh)) == NULL) break; ciss_unmap_request(cr); if ((cr->cr_flags & CISS_REQ_BUSY) == 0) ciss_printf(sc, "WARNING: completing non-busy request\n"); cr->cr_flags &= ~CISS_REQ_BUSY; /* * If the request has a callback, invoke it. */ if (cr->cr_complete != NULL) { cr->cr_complete(cr); continue; } /* * If someone is sleeping on this request, wake them up. */ if (cr->cr_flags & CISS_REQ_SLEEP) { cr->cr_flags &= ~CISS_REQ_SLEEP; wakeup(cr); continue; } /* * If someone is polling this request for completion, signal. */ if (cr->cr_flags & CISS_REQ_POLL) { cr->cr_flags &= ~CISS_REQ_POLL; continue; } /* * Give up and throw the request back on the free queue. This * should never happen; resources will probably be lost. */ ciss_printf(sc, "WARNING: completed command with no submitter\n"); ciss_enqueue_free(cr); } } /************************************************************************ * Report on the completion status of a request, and pass back SCSI * and command status values. */ static int _ciss_report_request(struct ciss_request *cr, int *command_status, int *scsi_status, const char *func) { struct ciss_command *cc; struct ciss_error_info *ce; debug_called(2); cc = cr->cr_cc; ce = (struct ciss_error_info *)&(cc->sg[0]); /* * We don't consider data under/overrun an error for the Report * Logical/Physical LUNs commands. */ if ((cc->header.host_tag & CISS_HDR_HOST_TAG_ERROR) && ((ce->command_status == CISS_CMD_STATUS_DATA_OVERRUN) || (ce->command_status == CISS_CMD_STATUS_DATA_UNDERRUN)) && ((cc->cdb.cdb[0] == CISS_OPCODE_REPORT_LOGICAL_LUNS) || (cc->cdb.cdb[0] == CISS_OPCODE_REPORT_PHYSICAL_LUNS) || (cc->cdb.cdb[0] == INQUIRY))) { cc->header.host_tag &= ~CISS_HDR_HOST_TAG_ERROR; debug(2, "ignoring irrelevant under/overrun error"); } /* * Check the command's error bit, if clear, there's no status and * everything is OK. */ if (!(cc->header.host_tag & CISS_HDR_HOST_TAG_ERROR)) { if (scsi_status != NULL) *scsi_status = SCSI_STATUS_OK; if (command_status != NULL) *command_status = CISS_CMD_STATUS_SUCCESS; return(0); } else { if (command_status != NULL) *command_status = ce->command_status; if (scsi_status != NULL) { if (ce->command_status == CISS_CMD_STATUS_TARGET_STATUS) { *scsi_status = ce->scsi_status; } else { *scsi_status = -1; } } if (bootverbose) ciss_printf(cr->cr_sc, "command status 0x%x (%s) scsi status 0x%x\n", ce->command_status, ciss_name_command_status(ce->command_status), ce->scsi_status); if (ce->command_status == CISS_CMD_STATUS_INVALID_COMMAND) { ciss_printf(cr->cr_sc, "invalid command, offense size %d at %d, value 0x%x, function %s\n", ce->additional_error_info.invalid_command.offense_size, ce->additional_error_info.invalid_command.offense_offset, ce->additional_error_info.invalid_command.offense_value, func); } } #if 0 ciss_print_request(cr); #endif return(1); } /************************************************************************ * Issue a request and don't return until it's completed. * * Depending on adapter status, we may poll or sleep waiting for * completion. */ static int ciss_synch_request(struct ciss_request *cr, int timeout) { if (cr->cr_sc->ciss_flags & CISS_FLAG_RUNNING) { return(ciss_wait_request(cr, timeout)); } else { return(ciss_poll_request(cr, timeout)); } } /************************************************************************ * Issue a request and poll for completion. * * Timeout in milliseconds. */ static int ciss_poll_request(struct ciss_request *cr, int timeout) { cr_qhead_t qh; struct ciss_softc *sc; int error; debug_called(2); STAILQ_INIT(&qh); sc = cr->cr_sc; cr->cr_flags |= CISS_REQ_POLL; if ((error = ciss_start(cr)) != 0) return(error); do { if (sc->ciss_perf) ciss_perf_done(sc, &qh); else ciss_done(sc, &qh); ciss_complete(sc, &qh); if (!(cr->cr_flags & CISS_REQ_POLL)) return(0); DELAY(1000); } while (timeout-- >= 0); return(EWOULDBLOCK); } /************************************************************************ * Issue a request and sleep waiting for completion. * * Timeout in milliseconds. Note that a spurious wakeup will reset * the timeout. */ static int ciss_wait_request(struct ciss_request *cr, int timeout) { int error; debug_called(2); cr->cr_flags |= CISS_REQ_SLEEP; if ((error = ciss_start(cr)) != 0) return(error); while ((cr->cr_flags & CISS_REQ_SLEEP) && (error != EWOULDBLOCK)) { error = msleep_sbt(cr, &cr->cr_sc->ciss_mtx, PRIBIO, "cissREQ", SBT_1MS * timeout, 0, 0); } return(error); } #if 0 /************************************************************************ * Abort a request. Note that a potential exists here to race the * request being completed; the caller must deal with this. */ static int ciss_abort_request(struct ciss_request *ar) { struct ciss_request *cr; struct ciss_command *cc; struct ciss_message_cdb *cmc; int error; debug_called(1); /* get a request */ if ((error = ciss_get_request(ar->cr_sc, &cr)) != 0) return(error); /* build the abort command */ cc = cr->cr_cc; cc->header.address.mode.mode = CISS_HDR_ADDRESS_MODE_PERIPHERAL; /* addressing? */ cc->header.address.physical.target = 0; cc->header.address.physical.bus = 0; cc->cdb.cdb_length = sizeof(*cmc); cc->cdb.type = CISS_CDB_TYPE_MESSAGE; cc->cdb.attribute = CISS_CDB_ATTRIBUTE_SIMPLE; cc->cdb.direction = CISS_CDB_DIRECTION_NONE; cc->cdb.timeout = 30; cmc = (struct ciss_message_cdb *)&(cc->cdb.cdb[0]); cmc->opcode = CISS_OPCODE_MESSAGE_ABORT; cmc->type = CISS_MESSAGE_ABORT_TASK; cmc->abort_tag = ar->cr_tag; /* endianness?? */ /* * Send the request and wait for a response. If we believe we * aborted the request OK, clear the flag that indicates it's * running. */ error = ciss_synch_request(cr, 35 * 1000); if (!error) error = ciss_report_request(cr, NULL, NULL); ciss_release_request(cr); return(error); } #endif /************************************************************************ * Fetch and initialise a request */ static int ciss_get_request(struct ciss_softc *sc, struct ciss_request **crp) { struct ciss_request *cr; debug_called(2); /* * Get a request and clean it up. */ if ((cr = ciss_dequeue_free(sc)) == NULL) return(ENOMEM); cr->cr_data = NULL; cr->cr_flags = 0; cr->cr_complete = NULL; cr->cr_private = NULL; cr->cr_sg_tag = CISS_SG_MAX; /* Backstop to prevent accidents */ ciss_preen_command(cr); *crp = cr; return(0); } static void ciss_preen_command(struct ciss_request *cr) { struct ciss_command *cc; u_int32_t cmdphys; /* * Clean up the command structure. * * Note that we set up the error_info structure here, since the * length can be overwritten by any command. */ cc = cr->cr_cc; cc->header.sg_in_list = 0; /* kinda inefficient this way */ cc->header.sg_total = 0; cc->header.host_tag = cr->cr_tag << 2; cc->header.host_tag_zeroes = 0; bzero(&(cc->sg[0]), CISS_COMMAND_ALLOC_SIZE - sizeof(struct ciss_command)); cmdphys = cr->cr_ccphys; cc->error_info.error_info_address = cmdphys + sizeof(struct ciss_command); cc->error_info.error_info_length = CISS_COMMAND_ALLOC_SIZE - sizeof(struct ciss_command); } /************************************************************************ * Release a request to the free list. */ static void ciss_release_request(struct ciss_request *cr) { struct ciss_softc *sc; debug_called(2); sc = cr->cr_sc; /* release the request to the free queue */ ciss_requeue_free(cr); } /************************************************************************ * Allocate a request that will be used to send a BMIC command. Do some * of the common setup here to avoid duplicating it everywhere else. */ static int ciss_get_bmic_request(struct ciss_softc *sc, struct ciss_request **crp, int opcode, void **bufp, size_t bufsize) { struct ciss_request *cr; struct ciss_command *cc; struct ciss_bmic_cdb *cbc; void *buf; int error; int dataout; debug_called(2); cr = NULL; buf = NULL; /* * Get a request. */ if ((error = ciss_get_request(sc, &cr)) != 0) goto out; /* * Allocate data storage if requested, determine the data direction. */ dataout = 0; if ((bufsize > 0) && (bufp != NULL)) { if (*bufp == NULL) { if ((buf = malloc(bufsize, CISS_MALLOC_CLASS, M_NOWAIT | M_ZERO)) == NULL) { error = ENOMEM; goto out; } } else { buf = *bufp; dataout = 1; /* we are given a buffer, so we are writing */ } } /* * Build a CISS BMIC command to get the logical drive ID. */ cr->cr_data = buf; cr->cr_length = bufsize; if (!dataout) cr->cr_flags = CISS_REQ_DATAIN; cc = cr->cr_cc; cc->header.address.physical.mode = CISS_HDR_ADDRESS_MODE_PERIPHERAL; cc->header.address.physical.bus = 0; cc->header.address.physical.target = 0; cc->cdb.cdb_length = sizeof(*cbc); cc->cdb.type = CISS_CDB_TYPE_COMMAND; cc->cdb.attribute = CISS_CDB_ATTRIBUTE_SIMPLE; cc->cdb.direction = dataout ? CISS_CDB_DIRECTION_WRITE : CISS_CDB_DIRECTION_READ; cc->cdb.timeout = 0; cbc = (struct ciss_bmic_cdb *)&(cc->cdb.cdb[0]); bzero(cbc, sizeof(*cbc)); cbc->opcode = dataout ? CISS_ARRAY_CONTROLLER_WRITE : CISS_ARRAY_CONTROLLER_READ; cbc->bmic_opcode = opcode; cbc->size = htons((u_int16_t)bufsize); out: if (error) { if (cr != NULL) ciss_release_request(cr); } else { *crp = cr; if ((bufp != NULL) && (*bufp == NULL) && (buf != NULL)) *bufp = buf; } return(error); } /************************************************************************ * Handle a command passed in from userspace. */ static int ciss_user_command(struct ciss_softc *sc, IOCTL_Command_struct *ioc) { struct ciss_request *cr; struct ciss_command *cc; struct ciss_error_info *ce; int error = 0; debug_called(1); cr = NULL; /* * Get a request. */ while (ciss_get_request(sc, &cr) != 0) msleep(sc, &sc->ciss_mtx, PPAUSE, "cissREQ", hz); cc = cr->cr_cc; /* * Allocate an in-kernel databuffer if required, copy in user data. */ mtx_unlock(&sc->ciss_mtx); cr->cr_length = ioc->buf_size; if (ioc->buf_size > 0) { if ((cr->cr_data = malloc(ioc->buf_size, CISS_MALLOC_CLASS, M_NOWAIT)) == NULL) { error = ENOMEM; goto out_unlocked; } if ((error = copyin(ioc->buf, cr->cr_data, ioc->buf_size))) { debug(0, "copyin: bad data buffer %p/%d", ioc->buf, ioc->buf_size); goto out_unlocked; } } /* * Build the request based on the user command. */ bcopy(&ioc->LUN_info, &cc->header.address, sizeof(cc->header.address)); bcopy(&ioc->Request, &cc->cdb, sizeof(cc->cdb)); /* XXX anything else to populate here? */ mtx_lock(&sc->ciss_mtx); /* * Run the command. */ if ((error = ciss_synch_request(cr, 60 * 1000))) { debug(0, "request failed - %d", error); goto out; } /* * Check to see if the command succeeded. */ ce = (struct ciss_error_info *)&(cc->sg[0]); if ((cc->header.host_tag & CISS_HDR_HOST_TAG_ERROR) == 0) bzero(ce, sizeof(*ce)); /* * Copy the results back to the user. */ bcopy(ce, &ioc->error_info, sizeof(*ce)); mtx_unlock(&sc->ciss_mtx); if ((ioc->buf_size > 0) && (error = copyout(cr->cr_data, ioc->buf, ioc->buf_size))) { debug(0, "copyout: bad data buffer %p/%d", ioc->buf, ioc->buf_size); goto out_unlocked; } /* done OK */ error = 0; out_unlocked: mtx_lock(&sc->ciss_mtx); out: if ((cr != NULL) && (cr->cr_data != NULL)) free(cr->cr_data, CISS_MALLOC_CLASS); if (cr != NULL) ciss_release_request(cr); return(error); } /************************************************************************ * Map a request into bus-visible space, initialise the scatter/gather * list. */ static int ciss_map_request(struct ciss_request *cr) { struct ciss_softc *sc; int error = 0; debug_called(2); sc = cr->cr_sc; /* check that mapping is necessary */ if (cr->cr_flags & CISS_REQ_MAPPED) return(0); cr->cr_flags |= CISS_REQ_MAPPED; bus_dmamap_sync(sc->ciss_command_dmat, sc->ciss_command_map, BUS_DMASYNC_PREWRITE); if (cr->cr_data != NULL) { if (cr->cr_flags & CISS_REQ_CCB) error = bus_dmamap_load_ccb(sc->ciss_buffer_dmat, cr->cr_datamap, cr->cr_data, ciss_request_map_helper, cr, 0); else error = bus_dmamap_load(sc->ciss_buffer_dmat, cr->cr_datamap, cr->cr_data, cr->cr_length, ciss_request_map_helper, cr, 0); if (error != 0) return (error); } else { /* * Post the command to the adapter. */ cr->cr_sg_tag = CISS_SG_NONE; cr->cr_flags |= CISS_REQ_BUSY; if (sc->ciss_perf) CISS_TL_PERF_POST_CMD(sc, cr); else CISS_TL_SIMPLE_POST_CMD(sc, cr->cr_ccphys); } return(0); } static void ciss_request_map_helper(void *arg, bus_dma_segment_t *segs, int nseg, int error) { struct ciss_command *cc; struct ciss_request *cr; struct ciss_softc *sc; int i; debug_called(2); cr = (struct ciss_request *)arg; sc = cr->cr_sc; cc = cr->cr_cc; for (i = 0; i < nseg; i++) { cc->sg[i].address = segs[i].ds_addr; cc->sg[i].length = segs[i].ds_len; cc->sg[i].extension = 0; } /* we leave the s/g table entirely within the command */ cc->header.sg_in_list = nseg; cc->header.sg_total = nseg; if (cr->cr_flags & CISS_REQ_DATAIN) bus_dmamap_sync(sc->ciss_buffer_dmat, cr->cr_datamap, BUS_DMASYNC_PREREAD); if (cr->cr_flags & CISS_REQ_DATAOUT) bus_dmamap_sync(sc->ciss_buffer_dmat, cr->cr_datamap, BUS_DMASYNC_PREWRITE); if (nseg == 0) cr->cr_sg_tag = CISS_SG_NONE; else if (nseg == 1) cr->cr_sg_tag = CISS_SG_1; else if (nseg == 2) cr->cr_sg_tag = CISS_SG_2; else if (nseg <= 4) cr->cr_sg_tag = CISS_SG_4; else if (nseg <= 8) cr->cr_sg_tag = CISS_SG_8; else if (nseg <= 16) cr->cr_sg_tag = CISS_SG_16; else if (nseg <= 32) cr->cr_sg_tag = CISS_SG_32; else cr->cr_sg_tag = CISS_SG_MAX; /* * Post the command to the adapter. */ cr->cr_flags |= CISS_REQ_BUSY; if (sc->ciss_perf) CISS_TL_PERF_POST_CMD(sc, cr); else CISS_TL_SIMPLE_POST_CMD(sc, cr->cr_ccphys); } /************************************************************************ * Unmap a request from bus-visible space. */ static void ciss_unmap_request(struct ciss_request *cr) { struct ciss_softc *sc; debug_called(2); sc = cr->cr_sc; /* check that unmapping is necessary */ if ((cr->cr_flags & CISS_REQ_MAPPED) == 0) return; bus_dmamap_sync(sc->ciss_command_dmat, sc->ciss_command_map, BUS_DMASYNC_POSTWRITE); if (cr->cr_data == NULL) goto out; if (cr->cr_flags & CISS_REQ_DATAIN) bus_dmamap_sync(sc->ciss_buffer_dmat, cr->cr_datamap, BUS_DMASYNC_POSTREAD); if (cr->cr_flags & CISS_REQ_DATAOUT) bus_dmamap_sync(sc->ciss_buffer_dmat, cr->cr_datamap, BUS_DMASYNC_POSTWRITE); bus_dmamap_unload(sc->ciss_buffer_dmat, cr->cr_datamap); out: cr->cr_flags &= ~CISS_REQ_MAPPED; } /************************************************************************ * Attach the driver to CAM. * * We put all the logical drives on a single SCSI bus. */ static int ciss_cam_init(struct ciss_softc *sc) { int i, maxbus; debug_called(1); /* * Allocate a devq. We can reuse this for the masked physical * devices if we decide to export these as well. */ if ((sc->ciss_cam_devq = cam_simq_alloc(sc->ciss_max_requests - 2)) == NULL) { ciss_printf(sc, "can't allocate CAM SIM queue\n"); return(ENOMEM); } /* * Create a SIM. * * This naturally wastes a bit of memory. The alternative is to allocate * and register each bus as it is found, and then track them on a linked * list. Unfortunately, the driver has a few places where it needs to * look up the SIM based solely on bus number, and it's unclear whether * a list traversal would work for these situations. */ maxbus = max(sc->ciss_max_logical_bus, sc->ciss_max_physical_bus + CISS_PHYSICAL_BASE); sc->ciss_cam_sim = malloc(maxbus * sizeof(struct cam_sim*), CISS_MALLOC_CLASS, M_NOWAIT | M_ZERO); if (sc->ciss_cam_sim == NULL) { ciss_printf(sc, "can't allocate memory for controller SIM\n"); return(ENOMEM); } for (i = 0; i < sc->ciss_max_logical_bus; i++) { if ((sc->ciss_cam_sim[i] = cam_sim_alloc(ciss_cam_action, ciss_cam_poll, "ciss", sc, device_get_unit(sc->ciss_dev), &sc->ciss_mtx, 2, sc->ciss_max_requests - 2, sc->ciss_cam_devq)) == NULL) { ciss_printf(sc, "can't allocate CAM SIM for controller %d\n", i); return(ENOMEM); } /* * Register bus with this SIM. */ mtx_lock(&sc->ciss_mtx); if (i == 0 || sc->ciss_controllers[i].physical.bus != 0) { if (xpt_bus_register(sc->ciss_cam_sim[i], sc->ciss_dev, i) != 0) { ciss_printf(sc, "can't register SCSI bus %d\n", i); mtx_unlock(&sc->ciss_mtx); return (ENXIO); } } mtx_unlock(&sc->ciss_mtx); } for (i = CISS_PHYSICAL_BASE; i < sc->ciss_max_physical_bus + CISS_PHYSICAL_BASE; i++) { if ((sc->ciss_cam_sim[i] = cam_sim_alloc(ciss_cam_action, ciss_cam_poll, "ciss", sc, device_get_unit(sc->ciss_dev), &sc->ciss_mtx, 1, sc->ciss_max_requests - 2, sc->ciss_cam_devq)) == NULL) { ciss_printf(sc, "can't allocate CAM SIM for controller %d\n", i); return (ENOMEM); } mtx_lock(&sc->ciss_mtx); if (xpt_bus_register(sc->ciss_cam_sim[i], sc->ciss_dev, i) != 0) { ciss_printf(sc, "can't register SCSI bus %d\n", i); mtx_unlock(&sc->ciss_mtx); return (ENXIO); } mtx_unlock(&sc->ciss_mtx); } return(0); } /************************************************************************ * Initiate a rescan of the 'logical devices' SIM */ static void ciss_cam_rescan_target(struct ciss_softc *sc, int bus, int target) { union ccb *ccb; debug_called(1); if ((ccb = xpt_alloc_ccb_nowait()) == NULL) { ciss_printf(sc, "rescan failed (can't allocate CCB)\n"); return; } if (xpt_create_path(&ccb->ccb_h.path, NULL, cam_sim_path(sc->ciss_cam_sim[bus]), target, CAM_LUN_WILDCARD) != CAM_REQ_CMP) { ciss_printf(sc, "rescan failed (can't create path)\n"); xpt_free_ccb(ccb); return; } xpt_rescan(ccb); /* scan is now in progress */ } /************************************************************************ * Handle requests coming from CAM */ static void ciss_cam_action(struct cam_sim *sim, union ccb *ccb) { struct ciss_softc *sc; struct ccb_scsiio *csio; int bus, target; int physical; sc = cam_sim_softc(sim); bus = cam_sim_bus(sim); csio = (struct ccb_scsiio *)&ccb->csio; target = csio->ccb_h.target_id; physical = CISS_IS_PHYSICAL(bus); switch (ccb->ccb_h.func_code) { /* perform SCSI I/O */ case XPT_SCSI_IO: if (!ciss_cam_action_io(sim, csio)) return; break; /* perform geometry calculations */ case XPT_CALC_GEOMETRY: { struct ccb_calc_geometry *ccg = &ccb->ccg; struct ciss_ldrive *ld; debug(1, "XPT_CALC_GEOMETRY %d:%d:%d", cam_sim_bus(sim), ccb->ccb_h.target_id, ccb->ccb_h.target_lun); ld = NULL; if (!physical) ld = &sc->ciss_logical[bus][target]; /* * Use the cached geometry settings unless the fault tolerance * is invalid. */ if (physical || ld->cl_geometry.fault_tolerance == 0xFF) { u_int32_t secs_per_cylinder; ccg->heads = 255; ccg->secs_per_track = 32; secs_per_cylinder = ccg->heads * ccg->secs_per_track; ccg->cylinders = ccg->volume_size / secs_per_cylinder; } else { ccg->heads = ld->cl_geometry.heads; ccg->secs_per_track = ld->cl_geometry.sectors; ccg->cylinders = ntohs(ld->cl_geometry.cylinders); } ccb->ccb_h.status = CAM_REQ_CMP; break; } /* handle path attribute inquiry */ case XPT_PATH_INQ: { struct ccb_pathinq *cpi = &ccb->cpi; int sg_length; debug(1, "XPT_PATH_INQ %d:%d:%d", cam_sim_bus(sim), ccb->ccb_h.target_id, ccb->ccb_h.target_lun); cpi->version_num = 1; cpi->hba_inquiry = PI_TAG_ABLE; /* XXX is this correct? */ cpi->target_sprt = 0; cpi->hba_misc = 0; cpi->max_target = sc->ciss_cfg->max_logical_supported; cpi->max_lun = 0; /* 'logical drive' channel only */ cpi->initiator_id = sc->ciss_cfg->max_logical_supported; strlcpy(cpi->sim_vid, "FreeBSD", SIM_IDLEN); - strlcpy(cpi->hba_vid, "CISS", HBA_IDLEN); - strlcpy(cpi->dev_name, cam_sim_name(sim), DEV_IDLEN); - cpi->unit_number = cam_sim_unit(sim); - cpi->bus_id = cam_sim_bus(sim); + strlcpy(cpi->hba_vid, "CISS", HBA_IDLEN); + strlcpy(cpi->dev_name, cam_sim_name(sim), DEV_IDLEN); + cpi->unit_number = cam_sim_unit(sim); + cpi->bus_id = cam_sim_bus(sim); cpi->base_transfer_speed = 132 * 1024; /* XXX what to set this to? */ cpi->transport = XPORT_SPI; cpi->transport_version = 2; cpi->protocol = PROTO_SCSI; cpi->protocol_version = SCSI_REV_2; if (sc->ciss_cfg->max_sg_length == 0) { sg_length = 17; } else { /* XXX Fix for ZMR cards that advertise max_sg_length == 32 * Confusing bit here. max_sg_length is usually a power of 2. We always * need to subtract 1 to account for partial pages. Then we need to * align on a valid PAGE_SIZE so we round down to the nearest power of 2. * Add 1 so we can then subtract it out in the assignment to maxio. * The reason for all these shenanigans is to create a maxio value that * creates IO operations to volumes that yield consistent operations * with good performance. */ sg_length = sc->ciss_cfg->max_sg_length - 1; sg_length = (1 << (fls(sg_length) - 1)) + 1; } cpi->maxio = (min(CISS_MAX_SG_ELEMENTS, sg_length) - 1) * PAGE_SIZE; ccb->ccb_h.status = CAM_REQ_CMP; break; } case XPT_GET_TRAN_SETTINGS: { struct ccb_trans_settings *cts = &ccb->cts; int bus, target; struct ccb_trans_settings_spi *spi = &cts->xport_specific.spi; struct ccb_trans_settings_scsi *scsi = &cts->proto_specific.scsi; bus = cam_sim_bus(sim); target = cts->ccb_h.target_id; debug(1, "XPT_GET_TRAN_SETTINGS %d:%d", bus, target); /* disconnect always OK */ cts->protocol = PROTO_SCSI; cts->protocol_version = SCSI_REV_2; cts->transport = XPORT_SPI; cts->transport_version = 2; spi->valid = CTS_SPI_VALID_DISC; spi->flags = CTS_SPI_FLAGS_DISC_ENB; scsi->valid = CTS_SCSI_VALID_TQ; scsi->flags = CTS_SCSI_FLAGS_TAG_ENB; cts->ccb_h.status = CAM_REQ_CMP; break; } default: /* we can't do this */ debug(1, "unspported func_code = 0x%x", ccb->ccb_h.func_code); ccb->ccb_h.status = CAM_REQ_INVALID; break; } xpt_done(ccb); } /************************************************************************ * Handle a CAM SCSI I/O request. */ static int ciss_cam_action_io(struct cam_sim *sim, struct ccb_scsiio *csio) { struct ciss_softc *sc; int bus, target; struct ciss_request *cr; struct ciss_command *cc; int error; sc = cam_sim_softc(sim); bus = cam_sim_bus(sim); target = csio->ccb_h.target_id; debug(2, "XPT_SCSI_IO %d:%d:%d", bus, target, csio->ccb_h.target_lun); /* check that the CDB pointer is not to a physical address */ if ((csio->ccb_h.flags & CAM_CDB_POINTER) && (csio->ccb_h.flags & CAM_CDB_PHYS)) { debug(3, " CDB pointer is to physical address"); csio->ccb_h.status = CAM_REQ_CMP_ERR; } /* abandon aborted ccbs or those that have failed validation */ if ((csio->ccb_h.status & CAM_STATUS_MASK) != CAM_REQ_INPROG) { debug(3, "abandoning CCB due to abort/validation failure"); return(EINVAL); } /* handle emulation of some SCSI commands ourself */ if (ciss_cam_emulate(sc, csio)) return(0); /* * Get a request to manage this command. If we can't, return the * ccb, freeze the queue and flag so that we unfreeze it when a * request completes. */ if ((error = ciss_get_request(sc, &cr)) != 0) { xpt_freeze_simq(sim, 1); sc->ciss_flags |= CISS_FLAG_BUSY; csio->ccb_h.status |= CAM_REQUEUE_REQ; return(error); } /* * Build the command. */ cc = cr->cr_cc; cr->cr_data = csio; cr->cr_length = csio->dxfer_len; cr->cr_complete = ciss_cam_complete; cr->cr_private = csio; /* * Target the right logical volume. */ if (CISS_IS_PHYSICAL(bus)) cc->header.address = sc->ciss_physical[CISS_CAM_TO_PBUS(bus)][target].cp_address; else cc->header.address = sc->ciss_logical[bus][target].cl_address; cc->cdb.cdb_length = csio->cdb_len; cc->cdb.type = CISS_CDB_TYPE_COMMAND; cc->cdb.attribute = CISS_CDB_ATTRIBUTE_SIMPLE; /* XXX ordered tags? */ if ((csio->ccb_h.flags & CAM_DIR_MASK) == CAM_DIR_OUT) { cr->cr_flags = CISS_REQ_DATAOUT | CISS_REQ_CCB; cc->cdb.direction = CISS_CDB_DIRECTION_WRITE; } else if ((csio->ccb_h.flags & CAM_DIR_MASK) == CAM_DIR_IN) { cr->cr_flags = CISS_REQ_DATAIN | CISS_REQ_CCB; cc->cdb.direction = CISS_CDB_DIRECTION_READ; } else { cr->cr_data = NULL; cr->cr_flags = 0; cc->cdb.direction = CISS_CDB_DIRECTION_NONE; } cc->cdb.timeout = (csio->ccb_h.timeout / 1000) + 1; if (csio->ccb_h.flags & CAM_CDB_POINTER) { bcopy(csio->cdb_io.cdb_ptr, &cc->cdb.cdb[0], csio->cdb_len); } else { bcopy(csio->cdb_io.cdb_bytes, &cc->cdb.cdb[0], csio->cdb_len); } /* * Submit the request to the adapter. * * Note that this may fail if we're unable to map the request (and * if we ever learn a transport layer other than simple, may fail * if the adapter rejects the command). */ if ((error = ciss_start(cr)) != 0) { xpt_freeze_simq(sim, 1); csio->ccb_h.status |= CAM_RELEASE_SIMQ; if (error == EINPROGRESS) { error = 0; } else { csio->ccb_h.status |= CAM_REQUEUE_REQ; ciss_release_request(cr); } return(error); } return(0); } /************************************************************************ * Emulate SCSI commands the adapter doesn't handle as we might like. */ static int ciss_cam_emulate(struct ciss_softc *sc, struct ccb_scsiio *csio) { int bus, target; u_int8_t opcode; target = csio->ccb_h.target_id; bus = cam_sim_bus(xpt_path_sim(csio->ccb_h.path)); opcode = (csio->ccb_h.flags & CAM_CDB_POINTER) ? *(u_int8_t *)csio->cdb_io.cdb_ptr : csio->cdb_io.cdb_bytes[0]; if (CISS_IS_PHYSICAL(bus)) { if (sc->ciss_physical[CISS_CAM_TO_PBUS(bus)][target].cp_online != 1) { csio->ccb_h.status |= CAM_SEL_TIMEOUT; xpt_done((union ccb *)csio); return(1); } else return(0); } /* * Handle requests for volumes that don't exist or are not online. * A selection timeout is slightly better than an illegal request. * Other errors might be better. */ if (sc->ciss_logical[bus][target].cl_status != CISS_LD_ONLINE) { csio->ccb_h.status |= CAM_SEL_TIMEOUT; xpt_done((union ccb *)csio); return(1); } /* if we have to fake Synchronise Cache */ if (sc->ciss_flags & CISS_FLAG_FAKE_SYNCH) { /* * If this is a Synchronise Cache command, typically issued when * a device is closed, flush the adapter and complete now. */ if (((csio->ccb_h.flags & CAM_CDB_POINTER) ? *(u_int8_t *)csio->cdb_io.cdb_ptr : csio->cdb_io.cdb_bytes[0]) == SYNCHRONIZE_CACHE) { ciss_flush_adapter(sc); csio->ccb_h.status |= CAM_REQ_CMP; xpt_done((union ccb *)csio); return(1); } } /* * A CISS target can only ever have one lun per target. REPORT_LUNS requires * at least one LUN field to be pre created for us, so snag it and fill in * the least significant byte indicating 1 LUN here. Emulate the command * return to shut up warning on console of a CDB error. swb */ if (opcode == REPORT_LUNS && csio->dxfer_len > 0) { csio->data_ptr[3] = 8; csio->ccb_h.status |= CAM_REQ_CMP; xpt_done((union ccb *)csio); return(1); } return(0); } /************************************************************************ * Check for possibly-completed commands. */ static void ciss_cam_poll(struct cam_sim *sim) { cr_qhead_t qh; struct ciss_softc *sc = cam_sim_softc(sim); debug_called(2); STAILQ_INIT(&qh); if (sc->ciss_perf) ciss_perf_done(sc, &qh); else ciss_done(sc, &qh); ciss_complete(sc, &qh); } /************************************************************************ * Handle completion of a command - pass results back through the CCB */ static void ciss_cam_complete(struct ciss_request *cr) { struct ciss_softc *sc; struct ciss_command *cc; struct ciss_error_info *ce; struct ccb_scsiio *csio; int scsi_status; int command_status; debug_called(2); sc = cr->cr_sc; cc = cr->cr_cc; ce = (struct ciss_error_info *)&(cc->sg[0]); csio = (struct ccb_scsiio *)cr->cr_private; /* * Extract status values from request. */ ciss_report_request(cr, &command_status, &scsi_status); csio->scsi_status = scsi_status; /* * Handle specific SCSI status values. */ switch(scsi_status) { /* no status due to adapter error */ case -1: debug(0, "adapter error"); csio->ccb_h.status |= CAM_REQ_CMP_ERR; break; /* no status due to command completed OK */ case SCSI_STATUS_OK: /* CISS_SCSI_STATUS_GOOD */ debug(2, "SCSI_STATUS_OK"); csio->ccb_h.status |= CAM_REQ_CMP; break; /* check condition, sense data included */ case SCSI_STATUS_CHECK_COND: /* CISS_SCSI_STATUS_CHECK_CONDITION */ debug(0, "SCSI_STATUS_CHECK_COND sense size %d resid %d\n", ce->sense_length, ce->residual_count); bzero(&csio->sense_data, SSD_FULL_SIZE); bcopy(&ce->sense_info[0], &csio->sense_data, ce->sense_length); if (csio->sense_len > ce->sense_length) csio->sense_resid = csio->sense_len - ce->sense_length; else csio->sense_resid = 0; csio->resid = ce->residual_count; csio->ccb_h.status |= CAM_SCSI_STATUS_ERROR | CAM_AUTOSNS_VALID; #ifdef CISS_DEBUG { struct scsi_sense_data *sns = (struct scsi_sense_data *)&ce->sense_info[0]; debug(0, "sense key %x", scsi_get_sense_key(sns, csio->sense_len - csio->sense_resid, /*show_errors*/ 1)); } #endif break; case SCSI_STATUS_BUSY: /* CISS_SCSI_STATUS_BUSY */ debug(0, "SCSI_STATUS_BUSY"); csio->ccb_h.status |= CAM_SCSI_BUSY; break; default: debug(0, "unknown status 0x%x", csio->scsi_status); csio->ccb_h.status |= CAM_REQ_CMP_ERR; break; } /* handle post-command fixup */ ciss_cam_complete_fixup(sc, csio); ciss_release_request(cr); if (sc->ciss_flags & CISS_FLAG_BUSY) { sc->ciss_flags &= ~CISS_FLAG_BUSY; if (csio->ccb_h.status & CAM_RELEASE_SIMQ) xpt_release_simq(xpt_path_sim(csio->ccb_h.path), 0); else csio->ccb_h.status |= CAM_RELEASE_SIMQ; } xpt_done((union ccb *)csio); } /******************************************************************************** * Fix up the result of some commands here. */ static void ciss_cam_complete_fixup(struct ciss_softc *sc, struct ccb_scsiio *csio) { struct scsi_inquiry_data *inq; struct ciss_ldrive *cl; uint8_t *cdb; int bus, target; cdb = (csio->ccb_h.flags & CAM_CDB_POINTER) ? (uint8_t *)csio->cdb_io.cdb_ptr : csio->cdb_io.cdb_bytes; if (cdb[0] == INQUIRY && (cdb[1] & SI_EVPD) == 0 && (csio->ccb_h.flags & CAM_DIR_MASK) == CAM_DIR_IN && csio->dxfer_len >= SHORT_INQUIRY_LENGTH) { inq = (struct scsi_inquiry_data *)csio->data_ptr; target = csio->ccb_h.target_id; bus = cam_sim_bus(xpt_path_sim(csio->ccb_h.path)); /* * If the controller is in JBOD mode, there are no logical volumes. * Let the disks be probed and dealt with via CAM. Else, mask off * the physical disks and setup the parts of the inq structure for * the logical volume. swb */ if( !(sc->ciss_id->PowerUPNvramFlags & PWR_UP_FLAG_JBOD_ENABLED)){ if (CISS_IS_PHYSICAL(bus)) { if (SID_TYPE(inq) == T_DIRECT) inq->device = (inq->device & 0xe0) | T_NODEVICE; return; } cl = &sc->ciss_logical[bus][target]; padstr(inq->vendor, "HP", SID_VENDOR_SIZE); padstr(inq->product, ciss_name_ldrive_org(cl->cl_ldrive->fault_tolerance), SID_PRODUCT_SIZE); padstr(inq->revision, ciss_name_ldrive_status(cl->cl_lstatus->status), SID_REVISION_SIZE); } } } /******************************************************************************** * Name the device at (target) * * XXX is this strictly correct? */ static int ciss_name_device(struct ciss_softc *sc, int bus, int target) { struct cam_periph *periph; struct cam_path *path; int status; if (CISS_IS_PHYSICAL(bus)) return (0); status = xpt_create_path(&path, NULL, cam_sim_path(sc->ciss_cam_sim[bus]), target, 0); if (status == CAM_REQ_CMP) { xpt_path_lock(path); periph = cam_periph_find(path, NULL); xpt_path_unlock(path); xpt_free_path(path); if (periph != NULL) { sprintf(sc->ciss_logical[bus][target].cl_name, "%s%d", periph->periph_name, periph->unit_number); return(0); } } sc->ciss_logical[bus][target].cl_name[0] = 0; return(ENOENT); } /************************************************************************ * Periodic status monitoring. */ static void ciss_periodic(void *arg) { struct ciss_softc *sc; struct ciss_request *cr = NULL; struct ciss_command *cc = NULL; int error = 0; debug_called(1); sc = (struct ciss_softc *)arg; /* * Check the adapter heartbeat. */ if (sc->ciss_cfg->heartbeat == sc->ciss_heartbeat) { sc->ciss_heart_attack++; debug(0, "adapter heart attack in progress 0x%x/%d", sc->ciss_heartbeat, sc->ciss_heart_attack); if (sc->ciss_heart_attack == 3) { ciss_printf(sc, "ADAPTER HEARTBEAT FAILED\n"); ciss_disable_adapter(sc); return; } } else { sc->ciss_heartbeat = sc->ciss_cfg->heartbeat; sc->ciss_heart_attack = 0; debug(3, "new heartbeat 0x%x", sc->ciss_heartbeat); } /* * Send the NOP message and wait for a response. */ if (ciss_nop_message_heartbeat != 0 && (error = ciss_get_request(sc, &cr)) == 0) { cc = cr->cr_cc; cr->cr_complete = ciss_nop_complete; cc->cdb.cdb_length = 1; cc->cdb.type = CISS_CDB_TYPE_MESSAGE; cc->cdb.attribute = CISS_CDB_ATTRIBUTE_SIMPLE; cc->cdb.direction = CISS_CDB_DIRECTION_WRITE; cc->cdb.timeout = 0; cc->cdb.cdb[0] = CISS_OPCODE_MESSAGE_NOP; if ((error = ciss_start(cr)) != 0) { ciss_printf(sc, "SENDING NOP MESSAGE FAILED\n"); } } /* * If the notify event request has died for some reason, or has * not started yet, restart it. */ if (!(sc->ciss_flags & CISS_FLAG_NOTIFY_OK)) { debug(0, "(re)starting Event Notify chain"); ciss_notify_event(sc); } /* * Reschedule. */ callout_reset(&sc->ciss_periodic, CISS_HEARTBEAT_RATE * hz, ciss_periodic, sc); } static void ciss_nop_complete(struct ciss_request *cr) { struct ciss_softc *sc; static int first_time = 1; sc = cr->cr_sc; if (ciss_report_request(cr, NULL, NULL) != 0) { if (first_time == 1) { first_time = 0; ciss_printf(sc, "SENDING NOP MESSAGE FAILED (not logging anymore)\n"); } } ciss_release_request(cr); } /************************************************************************ * Disable the adapter. * * The all requests in completed queue is failed with hardware error. * This will cause failover in a multipath configuration. */ static void ciss_disable_adapter(struct ciss_softc *sc) { cr_qhead_t qh; struct ciss_request *cr; struct ciss_command *cc; struct ciss_error_info *ce; int i; CISS_TL_SIMPLE_DISABLE_INTERRUPTS(sc); pci_disable_busmaster(sc->ciss_dev); sc->ciss_flags &= ~CISS_FLAG_RUNNING; for (i = 1; i < sc->ciss_max_requests; i++) { cr = &sc->ciss_request[i]; if ((cr->cr_flags & CISS_REQ_BUSY) == 0) continue; cc = cr->cr_cc; ce = (struct ciss_error_info *)&(cc->sg[0]); ce->command_status = CISS_CMD_STATUS_HARDWARE_ERROR; ciss_enqueue_complete(cr, &qh); } for (;;) { if ((cr = ciss_dequeue_complete(sc, &qh)) == NULL) break; /* * If the request has a callback, invoke it. */ if (cr->cr_complete != NULL) { cr->cr_complete(cr); continue; } /* * If someone is sleeping on this request, wake them up. */ if (cr->cr_flags & CISS_REQ_SLEEP) { cr->cr_flags &= ~CISS_REQ_SLEEP; wakeup(cr); continue; } } } /************************************************************************ * Request a notification response from the adapter. * * If (cr) is NULL, this is the first request of the adapter, so * reset the adapter's message pointer and start with the oldest * message available. */ static void ciss_notify_event(struct ciss_softc *sc) { struct ciss_request *cr; struct ciss_command *cc; struct ciss_notify_cdb *cnc; int error; debug_called(1); cr = sc->ciss_periodic_notify; /* get a request if we don't already have one */ if (cr == NULL) { if ((error = ciss_get_request(sc, &cr)) != 0) { debug(0, "can't get notify event request"); goto out; } sc->ciss_periodic_notify = cr; cr->cr_complete = ciss_notify_complete; debug(1, "acquired request %d", cr->cr_tag); } /* * Get a databuffer if we don't already have one, note that the * adapter command wants a larger buffer than the actual * structure. */ if (cr->cr_data == NULL) { if ((cr->cr_data = malloc(CISS_NOTIFY_DATA_SIZE, CISS_MALLOC_CLASS, M_NOWAIT)) == NULL) { debug(0, "can't get notify event request buffer"); error = ENOMEM; goto out; } cr->cr_length = CISS_NOTIFY_DATA_SIZE; } /* re-setup the request's command (since we never release it) XXX overkill*/ ciss_preen_command(cr); /* (re)build the notify event command */ cc = cr->cr_cc; cc->header.address.physical.mode = CISS_HDR_ADDRESS_MODE_PERIPHERAL; cc->header.address.physical.bus = 0; cc->header.address.physical.target = 0; cc->cdb.cdb_length = sizeof(*cnc); cc->cdb.type = CISS_CDB_TYPE_COMMAND; cc->cdb.attribute = CISS_CDB_ATTRIBUTE_SIMPLE; cc->cdb.direction = CISS_CDB_DIRECTION_READ; cc->cdb.timeout = 0; /* no timeout, we hope */ cnc = (struct ciss_notify_cdb *)&(cc->cdb.cdb[0]); bzero(cr->cr_data, CISS_NOTIFY_DATA_SIZE); cnc->opcode = CISS_OPCODE_READ; cnc->command = CISS_COMMAND_NOTIFY_ON_EVENT; cnc->timeout = 0; /* no timeout, we hope */ cnc->synchronous = 0; cnc->ordered = 0; cnc->seek_to_oldest = 0; if ((sc->ciss_flags & CISS_FLAG_RUNNING) == 0) cnc->new_only = 1; else cnc->new_only = 0; cnc->length = htonl(CISS_NOTIFY_DATA_SIZE); /* submit the request */ error = ciss_start(cr); out: if (error) { if (cr != NULL) { if (cr->cr_data != NULL) free(cr->cr_data, CISS_MALLOC_CLASS); ciss_release_request(cr); } sc->ciss_periodic_notify = NULL; debug(0, "can't submit notify event request"); sc->ciss_flags &= ~CISS_FLAG_NOTIFY_OK; } else { debug(1, "notify event submitted"); sc->ciss_flags |= CISS_FLAG_NOTIFY_OK; } } static void ciss_notify_complete(struct ciss_request *cr) { struct ciss_command *cc; struct ciss_notify *cn; struct ciss_softc *sc; int scsi_status; int command_status; debug_called(1); cc = cr->cr_cc; cn = (struct ciss_notify *)cr->cr_data; sc = cr->cr_sc; /* * Report request results, decode status. */ ciss_report_request(cr, &command_status, &scsi_status); /* * Abort the chain on a fatal error. * * XXX which of these are actually errors? */ if ((command_status != CISS_CMD_STATUS_SUCCESS) && (command_status != CISS_CMD_STATUS_TARGET_STATUS) && (command_status != CISS_CMD_STATUS_TIMEOUT)) { /* XXX timeout? */ ciss_printf(sc, "fatal error in Notify Event request (%s)\n", ciss_name_command_status(command_status)); ciss_release_request(cr); sc->ciss_flags &= ~CISS_FLAG_NOTIFY_OK; return; } /* * If the adapter gave us a text message, print it. */ if (cn->message[0] != 0) ciss_printf(sc, "*** %.80s\n", cn->message); debug(0, "notify event class %d subclass %d detail %d", cn->class, cn->subclass, cn->detail); /* * If the response indicates that the notifier has been aborted, * release the notifier command. */ if ((cn->class == CISS_NOTIFY_NOTIFIER) && (cn->subclass == CISS_NOTIFY_NOTIFIER_STATUS) && (cn->detail == 1)) { debug(0, "notifier exiting"); sc->ciss_flags &= ~CISS_FLAG_NOTIFY_OK; ciss_release_request(cr); sc->ciss_periodic_notify = NULL; wakeup(&sc->ciss_periodic_notify); } else { /* Handle notify events in a kernel thread */ ciss_enqueue_notify(cr); sc->ciss_periodic_notify = NULL; wakeup(&sc->ciss_periodic_notify); wakeup(&sc->ciss_notify); } /* * Send a new notify event command, if we're not aborting. */ if (!(sc->ciss_flags & CISS_FLAG_ABORTING)) { ciss_notify_event(sc); } } /************************************************************************ * Abort the Notify Event chain. * * Note that we can't just abort the command in progress; we have to * explicitly issue an Abort Notify Event command in order for the * adapter to clean up correctly. * * If we are called with CISS_FLAG_ABORTING set in the adapter softc, * the chain will not restart itself. */ static int ciss_notify_abort(struct ciss_softc *sc) { struct ciss_request *cr; struct ciss_command *cc; struct ciss_notify_cdb *cnc; int error, command_status, scsi_status; debug_called(1); cr = NULL; error = 0; /* verify that there's an outstanding command */ if (!(sc->ciss_flags & CISS_FLAG_NOTIFY_OK)) goto out; /* get a command to issue the abort with */ if ((error = ciss_get_request(sc, &cr))) goto out; /* get a buffer for the result */ if ((cr->cr_data = malloc(CISS_NOTIFY_DATA_SIZE, CISS_MALLOC_CLASS, M_NOWAIT)) == NULL) { debug(0, "can't get notify event request buffer"); error = ENOMEM; goto out; } cr->cr_length = CISS_NOTIFY_DATA_SIZE; /* build the CDB */ cc = cr->cr_cc; cc->header.address.physical.mode = CISS_HDR_ADDRESS_MODE_PERIPHERAL; cc->header.address.physical.bus = 0; cc->header.address.physical.target = 0; cc->cdb.cdb_length = sizeof(*cnc); cc->cdb.type = CISS_CDB_TYPE_COMMAND; cc->cdb.attribute = CISS_CDB_ATTRIBUTE_SIMPLE; cc->cdb.direction = CISS_CDB_DIRECTION_READ; cc->cdb.timeout = 0; /* no timeout, we hope */ cnc = (struct ciss_notify_cdb *)&(cc->cdb.cdb[0]); bzero(cnc, sizeof(*cnc)); cnc->opcode = CISS_OPCODE_WRITE; cnc->command = CISS_COMMAND_ABORT_NOTIFY; cnc->length = htonl(CISS_NOTIFY_DATA_SIZE); ciss_print_request(cr); /* * Submit the request and wait for it to complete. */ if ((error = ciss_synch_request(cr, 60 * 1000)) != 0) { ciss_printf(sc, "Abort Notify Event command failed (%d)\n", error); goto out; } /* * Check response. */ ciss_report_request(cr, &command_status, &scsi_status); switch(command_status) { case CISS_CMD_STATUS_SUCCESS: break; case CISS_CMD_STATUS_INVALID_COMMAND: /* * Some older adapters don't support the CISS version of this * command. Fall back to using the BMIC version. */ error = ciss_notify_abort_bmic(sc); if (error != 0) goto out; break; case CISS_CMD_STATUS_TARGET_STATUS: /* * This can happen if the adapter thinks there wasn't an outstanding * Notify Event command but we did. We clean up here. */ if (scsi_status == CISS_SCSI_STATUS_CHECK_CONDITION) { if (sc->ciss_periodic_notify != NULL) ciss_release_request(sc->ciss_periodic_notify); error = 0; goto out; } /* FALLTHROUGH */ default: ciss_printf(sc, "Abort Notify Event command failed (%s)\n", ciss_name_command_status(command_status)); error = EIO; goto out; } /* * Sleep waiting for the notifier command to complete. Note * that if it doesn't, we may end up in a bad situation, since * the adapter may deliver it later. Also note that the adapter * requires the Notify Event command to be cancelled in order to * maintain internal bookkeeping. */ while (sc->ciss_periodic_notify != NULL) { error = msleep(&sc->ciss_periodic_notify, &sc->ciss_mtx, PRIBIO, "cissNEA", hz * 5); if (error == EWOULDBLOCK) { ciss_printf(sc, "Notify Event command failed to abort, adapter may wedge.\n"); break; } } out: /* release the cancel request */ if (cr != NULL) { if (cr->cr_data != NULL) free(cr->cr_data, CISS_MALLOC_CLASS); ciss_release_request(cr); } if (error == 0) sc->ciss_flags &= ~CISS_FLAG_NOTIFY_OK; return(error); } /************************************************************************ * Abort the Notify Event chain using a BMIC command. */ static int ciss_notify_abort_bmic(struct ciss_softc *sc) { struct ciss_request *cr; int error, command_status; debug_called(1); cr = NULL; error = 0; /* verify that there's an outstanding command */ if (!(sc->ciss_flags & CISS_FLAG_NOTIFY_OK)) goto out; /* * Build a BMIC command to cancel the Notify on Event command. * * Note that we are sending a CISS opcode here. Odd. */ if ((error = ciss_get_bmic_request(sc, &cr, CISS_COMMAND_ABORT_NOTIFY, NULL, 0)) != 0) goto out; /* * Submit the request and wait for it to complete. */ if ((error = ciss_synch_request(cr, 60 * 1000)) != 0) { ciss_printf(sc, "error sending BMIC Cancel Notify on Event command (%d)\n", error); goto out; } /* * Check response. */ ciss_report_request(cr, &command_status, NULL); switch(command_status) { case CISS_CMD_STATUS_SUCCESS: break; default: ciss_printf(sc, "error cancelling Notify on Event (%s)\n", ciss_name_command_status(command_status)); error = EIO; goto out; } out: if (cr != NULL) ciss_release_request(cr); return(error); } /************************************************************************ * Handle rescanning all the logical volumes when a notify event * causes the drives to come online or offline. */ static void ciss_notify_rescan_logical(struct ciss_softc *sc) { struct ciss_lun_report *cll; struct ciss_ldrive *ld; int i, j, ndrives; /* * We must rescan all logical volumes to get the right logical * drive address. */ cll = ciss_report_luns(sc, CISS_OPCODE_REPORT_LOGICAL_LUNS, sc->ciss_cfg->max_logical_supported); if (cll == NULL) return; ndrives = (ntohl(cll->list_size) / sizeof(union ciss_device_address)); /* * Delete any of the drives which were destroyed by the * firmware. */ for (i = 0; i < sc->ciss_max_logical_bus; i++) { for (j = 0; j < sc->ciss_cfg->max_logical_supported; j++) { ld = &sc->ciss_logical[i][j]; if (ld->cl_update == 0) continue; if (ld->cl_status != CISS_LD_ONLINE) { ciss_cam_rescan_target(sc, i, j); ld->cl_update = 0; if (ld->cl_ldrive) free(ld->cl_ldrive, CISS_MALLOC_CLASS); if (ld->cl_lstatus) free(ld->cl_lstatus, CISS_MALLOC_CLASS); ld->cl_ldrive = NULL; ld->cl_lstatus = NULL; } } } /* * Scan for new drives. */ for (i = 0; i < ndrives; i++) { int bus, target; bus = CISS_LUN_TO_BUS(cll->lun[i].logical.lun); target = CISS_LUN_TO_TARGET(cll->lun[i].logical.lun); ld = &sc->ciss_logical[bus][target]; if (ld->cl_update == 0) continue; ld->cl_update = 0; ld->cl_address = cll->lun[i]; ld->cl_controller = &sc->ciss_controllers[bus]; if (ciss_identify_logical(sc, ld) == 0) { ciss_cam_rescan_target(sc, bus, target); } } free(cll, CISS_MALLOC_CLASS); } /************************************************************************ * Handle a notify event relating to the status of a logical drive. * * XXX need to be able to defer some of these to properly handle * calling the "ID Physical drive" command, unless the 'extended' * drive IDs are always in BIG_MAP format. */ static void ciss_notify_logical(struct ciss_softc *sc, struct ciss_notify *cn) { struct ciss_ldrive *ld; int ostatus, bus, target; debug_called(2); bus = cn->device.physical.bus; target = cn->data.logical_status.logical_drive; ld = &sc->ciss_logical[bus][target]; switch (cn->subclass) { case CISS_NOTIFY_LOGICAL_STATUS: switch (cn->detail) { case 0: ciss_name_device(sc, bus, target); ciss_printf(sc, "logical drive %d (%s) changed status %s->%s, spare status 0x%b\n", cn->data.logical_status.logical_drive, ld->cl_name, ciss_name_ldrive_status(cn->data.logical_status.previous_state), ciss_name_ldrive_status(cn->data.logical_status.new_state), cn->data.logical_status.spare_state, "\20\1configured\2rebuilding\3failed\4in use\5available\n"); /* * Update our idea of the drive's status. */ ostatus = ciss_decode_ldrive_status(cn->data.logical_status.previous_state); ld->cl_status = ciss_decode_ldrive_status(cn->data.logical_status.new_state); if (ld->cl_lstatus != NULL) ld->cl_lstatus->status = cn->data.logical_status.new_state; /* * Have CAM rescan the drive if its status has changed. */ if (ostatus != ld->cl_status) { ld->cl_update = 1; ciss_notify_rescan_logical(sc); } break; case 1: /* logical drive has recognised new media, needs Accept Media Exchange */ ciss_name_device(sc, bus, target); ciss_printf(sc, "logical drive %d (%s) media exchanged, ready to go online\n", cn->data.logical_status.logical_drive, ld->cl_name); ciss_accept_media(sc, ld); ld->cl_update = 1; ld->cl_status = ciss_decode_ldrive_status(cn->data.logical_status.new_state); ciss_notify_rescan_logical(sc); break; case 2: case 3: ciss_printf(sc, "rebuild of logical drive %d (%s) failed due to %s error\n", cn->data.rebuild_aborted.logical_drive, ld->cl_name, (cn->detail == 2) ? "read" : "write"); break; } break; case CISS_NOTIFY_LOGICAL_ERROR: if (cn->detail == 0) { ciss_printf(sc, "FATAL I/O ERROR on logical drive %d (%s), SCSI port %d ID %d\n", cn->data.io_error.logical_drive, ld->cl_name, cn->data.io_error.failure_bus, cn->data.io_error.failure_drive); /* XXX should we take the drive down at this point, or will we be told? */ } break; case CISS_NOTIFY_LOGICAL_SURFACE: if (cn->detail == 0) ciss_printf(sc, "logical drive %d (%s) completed consistency initialisation\n", cn->data.consistency_completed.logical_drive, ld->cl_name); break; } } /************************************************************************ * Handle a notify event relating to the status of a physical drive. */ static void ciss_notify_physical(struct ciss_softc *sc, struct ciss_notify *cn) { } /************************************************************************ * Handle a notify event relating to the status of a physical drive. */ static void ciss_notify_hotplug(struct ciss_softc *sc, struct ciss_notify *cn) { struct ciss_lun_report *cll = NULL; int bus, target; switch (cn->subclass) { case CISS_NOTIFY_HOTPLUG_PHYSICAL: case CISS_NOTIFY_HOTPLUG_NONDISK: bus = CISS_BIG_MAP_BUS(sc, cn->data.drive.big_physical_drive_number); target = CISS_BIG_MAP_TARGET(sc, cn->data.drive.big_physical_drive_number); if (cn->detail == 0) { /* * Mark the device offline so that it'll start producing selection * timeouts to the upper layer. */ if ((bus >= 0) && (target >= 0)) sc->ciss_physical[bus][target].cp_online = 0; } else { /* * Rescan the physical lun list for new items */ cll = ciss_report_luns(sc, CISS_OPCODE_REPORT_PHYSICAL_LUNS, sc->ciss_cfg->max_physical_supported); if (cll == NULL) { ciss_printf(sc, "Warning, cannot get physical lun list\n"); break; } ciss_filter_physical(sc, cll); } break; default: ciss_printf(sc, "Unknown hotplug event %d\n", cn->subclass); return; } if (cll != NULL) free(cll, CISS_MALLOC_CLASS); } /************************************************************************ * Handle deferred processing of notify events. Notify events may need * sleep which is unsafe during an interrupt. */ static void ciss_notify_thread(void *arg) { struct ciss_softc *sc; struct ciss_request *cr; struct ciss_notify *cn; sc = (struct ciss_softc *)arg; mtx_lock(&sc->ciss_mtx); for (;;) { if (STAILQ_EMPTY(&sc->ciss_notify) != 0 && (sc->ciss_flags & CISS_FLAG_THREAD_SHUT) == 0) { msleep(&sc->ciss_notify, &sc->ciss_mtx, PUSER, "idle", 0); } if (sc->ciss_flags & CISS_FLAG_THREAD_SHUT) break; cr = ciss_dequeue_notify(sc); if (cr == NULL) panic("cr null"); cn = (struct ciss_notify *)cr->cr_data; switch (cn->class) { case CISS_NOTIFY_HOTPLUG: ciss_notify_hotplug(sc, cn); break; case CISS_NOTIFY_LOGICAL: ciss_notify_logical(sc, cn); break; case CISS_NOTIFY_PHYSICAL: ciss_notify_physical(sc, cn); break; } ciss_release_request(cr); } sc->ciss_notify_thread = NULL; wakeup(&sc->ciss_notify_thread); mtx_unlock(&sc->ciss_mtx); kproc_exit(0); } /************************************************************************ * Start the notification kernel thread. */ static void ciss_spawn_notify_thread(struct ciss_softc *sc) { if (kproc_create((void(*)(void *))ciss_notify_thread, sc, &sc->ciss_notify_thread, 0, 0, "ciss_notify%d", device_get_unit(sc->ciss_dev))) panic("Could not create notify thread\n"); } /************************************************************************ * Kill the notification kernel thread. */ static void ciss_kill_notify_thread(struct ciss_softc *sc) { if (sc->ciss_notify_thread == NULL) return; sc->ciss_flags |= CISS_FLAG_THREAD_SHUT; wakeup(&sc->ciss_notify); msleep(&sc->ciss_notify_thread, &sc->ciss_mtx, PUSER, "thtrm", 0); } /************************************************************************ * Print a request. */ static void ciss_print_request(struct ciss_request *cr) { struct ciss_softc *sc; struct ciss_command *cc; int i; sc = cr->cr_sc; cc = cr->cr_cc; ciss_printf(sc, "REQUEST @ %p\n", cr); ciss_printf(sc, " data %p/%d tag %d flags %b\n", cr->cr_data, cr->cr_length, cr->cr_tag, cr->cr_flags, "\20\1mapped\2sleep\3poll\4dataout\5datain\n"); ciss_printf(sc, " sg list/total %d/%d host tag 0x%x\n", cc->header.sg_in_list, cc->header.sg_total, cc->header.host_tag); switch(cc->header.address.mode.mode) { case CISS_HDR_ADDRESS_MODE_PERIPHERAL: case CISS_HDR_ADDRESS_MODE_MASK_PERIPHERAL: ciss_printf(sc, " physical bus %d target %d\n", cc->header.address.physical.bus, cc->header.address.physical.target); break; case CISS_HDR_ADDRESS_MODE_LOGICAL: ciss_printf(sc, " logical unit %d\n", cc->header.address.logical.lun); break; } ciss_printf(sc, " %s cdb length %d type %s attribute %s\n", (cc->cdb.direction == CISS_CDB_DIRECTION_NONE) ? "no-I/O" : (cc->cdb.direction == CISS_CDB_DIRECTION_READ) ? "READ" : (cc->cdb.direction == CISS_CDB_DIRECTION_WRITE) ? "WRITE" : "??", cc->cdb.cdb_length, (cc->cdb.type == CISS_CDB_TYPE_COMMAND) ? "command" : (cc->cdb.type == CISS_CDB_TYPE_MESSAGE) ? "message" : "??", (cc->cdb.attribute == CISS_CDB_ATTRIBUTE_UNTAGGED) ? "untagged" : (cc->cdb.attribute == CISS_CDB_ATTRIBUTE_SIMPLE) ? "simple" : (cc->cdb.attribute == CISS_CDB_ATTRIBUTE_HEAD_OF_QUEUE) ? "head-of-queue" : (cc->cdb.attribute == CISS_CDB_ATTRIBUTE_ORDERED) ? "ordered" : (cc->cdb.attribute == CISS_CDB_ATTRIBUTE_AUTO_CONTINGENT) ? "auto-contingent" : "??"); ciss_printf(sc, " %*D\n", cc->cdb.cdb_length, &cc->cdb.cdb[0], " "); if (cc->header.host_tag & CISS_HDR_HOST_TAG_ERROR) { /* XXX print error info */ } else { /* since we don't use chained s/g, don't support it here */ for (i = 0; i < cc->header.sg_in_list; i++) { if ((i % 4) == 0) ciss_printf(sc, " "); printf("0x%08x/%d ", (u_int32_t)cc->sg[i].address, cc->sg[i].length); if ((((i + 1) % 4) == 0) || (i == (cc->header.sg_in_list - 1))) printf("\n"); } } } /************************************************************************ * Print information about the status of a logical drive. */ static void ciss_print_ldrive(struct ciss_softc *sc, struct ciss_ldrive *ld) { int bus, target, i; if (ld->cl_lstatus == NULL) { printf("does not exist\n"); return; } /* print drive status */ switch(ld->cl_lstatus->status) { case CISS_LSTATUS_OK: printf("online\n"); break; case CISS_LSTATUS_INTERIM_RECOVERY: printf("in interim recovery mode\n"); break; case CISS_LSTATUS_READY_RECOVERY: printf("ready to begin recovery\n"); break; case CISS_LSTATUS_RECOVERING: bus = CISS_BIG_MAP_BUS(sc, ld->cl_lstatus->drive_rebuilding); target = CISS_BIG_MAP_BUS(sc, ld->cl_lstatus->drive_rebuilding); printf("being recovered, working on physical drive %d.%d, %u blocks remaining\n", bus, target, ld->cl_lstatus->blocks_to_recover); break; case CISS_LSTATUS_EXPANDING: printf("being expanded, %u blocks remaining\n", ld->cl_lstatus->blocks_to_recover); break; case CISS_LSTATUS_QUEUED_FOR_EXPANSION: printf("queued for expansion\n"); break; case CISS_LSTATUS_FAILED: printf("queued for expansion\n"); break; case CISS_LSTATUS_WRONG_PDRIVE: printf("wrong physical drive inserted\n"); break; case CISS_LSTATUS_MISSING_PDRIVE: printf("missing a needed physical drive\n"); break; case CISS_LSTATUS_BECOMING_READY: printf("becoming ready\n"); break; } /* print failed physical drives */ for (i = 0; i < CISS_BIG_MAP_ENTRIES / 8; i++) { bus = CISS_BIG_MAP_BUS(sc, ld->cl_lstatus->drive_failure_map[i]); target = CISS_BIG_MAP_TARGET(sc, ld->cl_lstatus->drive_failure_map[i]); if (bus == -1) continue; ciss_printf(sc, "physical drive %d:%d (%x) failed\n", bus, target, ld->cl_lstatus->drive_failure_map[i]); } } #ifdef CISS_DEBUG #include "opt_ddb.h" #ifdef DDB #include /************************************************************************ * Print information about the controller/driver. */ static void ciss_print_adapter(struct ciss_softc *sc) { int i, j; ciss_printf(sc, "ADAPTER:\n"); for (i = 0; i < CISSQ_COUNT; i++) { ciss_printf(sc, "%s %d/%d\n", i == 0 ? "free" : i == 1 ? "busy" : "complete", sc->ciss_qstat[i].q_length, sc->ciss_qstat[i].q_max); } ciss_printf(sc, "max_requests %d\n", sc->ciss_max_requests); ciss_printf(sc, "flags %b\n", sc->ciss_flags, "\20\1notify_ok\2control_open\3aborting\4running\21fake_synch\22bmic_abort\n"); for (i = 0; i < sc->ciss_max_logical_bus; i++) { for (j = 0; j < sc->ciss_cfg->max_logical_supported; j++) { ciss_printf(sc, "LOGICAL DRIVE %d: ", i); ciss_print_ldrive(sc, &sc->ciss_logical[i][j]); } } /* XXX Should physical drives be printed out here? */ for (i = 1; i < sc->ciss_max_requests; i++) ciss_print_request(sc->ciss_request + i); } /* DDB hook */ DB_COMMAND(ciss_prt, db_ciss_prt) { struct ciss_softc *sc; devclass_t dc; int maxciss, i; dc = devclass_find("ciss"); if ( dc == NULL ) { printf("%s: can't find devclass!\n", __func__); return; } maxciss = devclass_get_maxunit(dc); for (i = 0; i < maxciss; i++) { sc = devclass_get_softc(dc, i); ciss_print_adapter(sc); } } #endif #endif /************************************************************************ * Return a name for a logical drive status value. */ static const char * ciss_name_ldrive_status(int status) { switch (status) { case CISS_LSTATUS_OK: return("OK"); case CISS_LSTATUS_FAILED: return("failed"); case CISS_LSTATUS_NOT_CONFIGURED: return("not configured"); case CISS_LSTATUS_INTERIM_RECOVERY: return("interim recovery"); case CISS_LSTATUS_READY_RECOVERY: return("ready for recovery"); case CISS_LSTATUS_RECOVERING: return("recovering"); case CISS_LSTATUS_WRONG_PDRIVE: return("wrong physical drive inserted"); case CISS_LSTATUS_MISSING_PDRIVE: return("missing physical drive"); case CISS_LSTATUS_EXPANDING: return("expanding"); case CISS_LSTATUS_BECOMING_READY: return("becoming ready"); case CISS_LSTATUS_QUEUED_FOR_EXPANSION: return("queued for expansion"); } return("unknown status"); } /************************************************************************ * Return an online/offline/nonexistent value for a logical drive * status value. */ static int ciss_decode_ldrive_status(int status) { switch(status) { case CISS_LSTATUS_NOT_CONFIGURED: return(CISS_LD_NONEXISTENT); case CISS_LSTATUS_OK: case CISS_LSTATUS_INTERIM_RECOVERY: case CISS_LSTATUS_READY_RECOVERY: case CISS_LSTATUS_RECOVERING: case CISS_LSTATUS_EXPANDING: case CISS_LSTATUS_QUEUED_FOR_EXPANSION: return(CISS_LD_ONLINE); case CISS_LSTATUS_FAILED: case CISS_LSTATUS_WRONG_PDRIVE: case CISS_LSTATUS_MISSING_PDRIVE: case CISS_LSTATUS_BECOMING_READY: default: return(CISS_LD_OFFLINE); } } /************************************************************************ * Return a name for a logical drive's organisation. */ static const char * ciss_name_ldrive_org(int org) { switch(org) { case CISS_LDRIVE_RAID0: return("RAID 0"); case CISS_LDRIVE_RAID1: return("RAID 1(1+0)"); case CISS_LDRIVE_RAID4: return("RAID 4"); case CISS_LDRIVE_RAID5: return("RAID 5"); case CISS_LDRIVE_RAID51: return("RAID 5+1"); case CISS_LDRIVE_RAIDADG: return("RAID ADG"); } return("unknown"); } /************************************************************************ * Return a name for a command status value. */ static const char * ciss_name_command_status(int status) { switch(status) { case CISS_CMD_STATUS_SUCCESS: return("success"); case CISS_CMD_STATUS_TARGET_STATUS: return("target status"); case CISS_CMD_STATUS_DATA_UNDERRUN: return("data underrun"); case CISS_CMD_STATUS_DATA_OVERRUN: return("data overrun"); case CISS_CMD_STATUS_INVALID_COMMAND: return("invalid command"); case CISS_CMD_STATUS_PROTOCOL_ERROR: return("protocol error"); case CISS_CMD_STATUS_HARDWARE_ERROR: return("hardware error"); case CISS_CMD_STATUS_CONNECTION_LOST: return("connection lost"); case CISS_CMD_STATUS_ABORTED: return("aborted"); case CISS_CMD_STATUS_ABORT_FAILED: return("abort failed"); case CISS_CMD_STATUS_UNSOLICITED_ABORT: return("unsolicited abort"); case CISS_CMD_STATUS_TIMEOUT: return("timeout"); case CISS_CMD_STATUS_UNABORTABLE: return("unabortable"); } return("unknown status"); } /************************************************************************ * Handle an open on the control device. */ static int ciss_open(struct cdev *dev, int flags, int fmt, struct thread *p) { struct ciss_softc *sc; debug_called(1); sc = (struct ciss_softc *)dev->si_drv1; /* we might want to veto if someone already has us open */ mtx_lock(&sc->ciss_mtx); sc->ciss_flags |= CISS_FLAG_CONTROL_OPEN; mtx_unlock(&sc->ciss_mtx); return(0); } /************************************************************************ * Handle the last close on the control device. */ static int ciss_close(struct cdev *dev, int flags, int fmt, struct thread *p) { struct ciss_softc *sc; debug_called(1); sc = (struct ciss_softc *)dev->si_drv1; mtx_lock(&sc->ciss_mtx); sc->ciss_flags &= ~CISS_FLAG_CONTROL_OPEN; mtx_unlock(&sc->ciss_mtx); return (0); } /******************************************************************************** * Handle adapter-specific control operations. * * Note that the API here is compatible with the Linux driver, in order to * simplify the porting of Compaq's userland tools. */ static int ciss_ioctl(struct cdev *dev, u_long cmd, caddr_t addr, int32_t flag, struct thread *p) { struct ciss_softc *sc; IOCTL_Command_struct *ioc = (IOCTL_Command_struct *)addr; #ifdef __amd64__ IOCTL_Command_struct32 *ioc32 = (IOCTL_Command_struct32 *)addr; IOCTL_Command_struct ioc_swab; #endif int error; debug_called(1); sc = (struct ciss_softc *)dev->si_drv1; error = 0; mtx_lock(&sc->ciss_mtx); switch(cmd) { case CCISS_GETQSTATS: { union ciss_statrequest *cr = (union ciss_statrequest *)addr; switch (cr->cs_item) { case CISSQ_FREE: case CISSQ_NOTIFY: bcopy(&sc->ciss_qstat[cr->cs_item], &cr->cs_qstat, sizeof(struct ciss_qstat)); break; default: error = ENOIOCTL; break; } break; } case CCISS_GETPCIINFO: { cciss_pci_info_struct *pis = (cciss_pci_info_struct *)addr; pis->bus = pci_get_bus(sc->ciss_dev); pis->dev_fn = pci_get_slot(sc->ciss_dev); pis->board_id = (pci_get_subvendor(sc->ciss_dev) << 16) | pci_get_subdevice(sc->ciss_dev); break; } case CCISS_GETINTINFO: { cciss_coalint_struct *cis = (cciss_coalint_struct *)addr; cis->delay = sc->ciss_cfg->interrupt_coalesce_delay; cis->count = sc->ciss_cfg->interrupt_coalesce_count; break; } case CCISS_SETINTINFO: { cciss_coalint_struct *cis = (cciss_coalint_struct *)addr; if ((cis->delay == 0) && (cis->count == 0)) { error = EINVAL; break; } /* * XXX apparently this is only safe if the controller is idle, * we should suspend it before doing this. */ sc->ciss_cfg->interrupt_coalesce_delay = cis->delay; sc->ciss_cfg->interrupt_coalesce_count = cis->count; if (ciss_update_config(sc)) error = EIO; /* XXX resume the controller here */ break; } case CCISS_GETNODENAME: bcopy(sc->ciss_cfg->server_name, (NodeName_type *)addr, sizeof(NodeName_type)); break; case CCISS_SETNODENAME: bcopy((NodeName_type *)addr, sc->ciss_cfg->server_name, sizeof(NodeName_type)); if (ciss_update_config(sc)) error = EIO; break; case CCISS_GETHEARTBEAT: *(Heartbeat_type *)addr = sc->ciss_cfg->heartbeat; break; case CCISS_GETBUSTYPES: *(BusTypes_type *)addr = sc->ciss_cfg->bus_types; break; case CCISS_GETFIRMVER: bcopy(sc->ciss_id->running_firmware_revision, (FirmwareVer_type *)addr, sizeof(FirmwareVer_type)); break; case CCISS_GETDRIVERVER: *(DriverVer_type *)addr = CISS_DRIVER_VERSION; break; case CCISS_REVALIDVOLS: /* * This is a bit ugly; to do it "right" we really need * to find any disks that have changed, kick CAM off them, * then rescan only these disks. It'd be nice if they * a) told us which disk(s) they were going to play with, * and b) which ones had arrived. 8( */ break; #ifdef __amd64__ case CCISS_PASSTHRU32: ioc_swab.LUN_info = ioc32->LUN_info; ioc_swab.Request = ioc32->Request; ioc_swab.error_info = ioc32->error_info; ioc_swab.buf_size = ioc32->buf_size; ioc_swab.buf = (u_int8_t *)(uintptr_t)ioc32->buf; ioc = &ioc_swab; /* FALLTHROUGH */ #endif case CCISS_PASSTHRU: error = ciss_user_command(sc, ioc); break; default: debug(0, "unknown ioctl 0x%lx", cmd); debug(1, "CCISS_GETPCIINFO: 0x%lx", CCISS_GETPCIINFO); debug(1, "CCISS_GETINTINFO: 0x%lx", CCISS_GETINTINFO); debug(1, "CCISS_SETINTINFO: 0x%lx", CCISS_SETINTINFO); debug(1, "CCISS_GETNODENAME: 0x%lx", CCISS_GETNODENAME); debug(1, "CCISS_SETNODENAME: 0x%lx", CCISS_SETNODENAME); debug(1, "CCISS_GETHEARTBEAT: 0x%lx", CCISS_GETHEARTBEAT); debug(1, "CCISS_GETBUSTYPES: 0x%lx", CCISS_GETBUSTYPES); debug(1, "CCISS_GETFIRMVER: 0x%lx", CCISS_GETFIRMVER); debug(1, "CCISS_GETDRIVERVER: 0x%lx", CCISS_GETDRIVERVER); debug(1, "CCISS_REVALIDVOLS: 0x%lx", CCISS_REVALIDVOLS); debug(1, "CCISS_PASSTHRU: 0x%lx", CCISS_PASSTHRU); error = ENOIOCTL; break; } mtx_unlock(&sc->ciss_mtx); return(error); } Index: projects/clang400-import/sys/dev/dpt/dpt_scsi.c =================================================================== --- projects/clang400-import/sys/dev/dpt/dpt_scsi.c (revision 311459) +++ projects/clang400-import/sys/dev/dpt/dpt_scsi.c (revision 311460) @@ -1,2523 +1,2523 @@ /*- * Copyright (c) 1997 by Simon Shapiro * All Rights Reserved * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions, and the following disclaimer, * without modification, immediately at the beginning of the file. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. The name of the author may not be used to endorse or promote products * derived from this software without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR 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("$FreeBSD$"); /* * dpt_scsi.c: SCSI dependent code for the DPT driver * * credits: Assisted by Mike Neuffer in the early low level DPT code * Thanx to Mark Salyzyn of DPT for his assistance. * Special thanx to Justin Gibbs for invaluable help in * making this driver look and work like a FreeBSD component. * Last but not least, many thanx to UCB and the FreeBSD * team for creating and maintaining such a wonderful O/S. * * TODO: * Add ISA probe code. * * Add driver-level RAID-0. This will allow interoperability with * NiceTry, M$-Doze, Win-Dog, Slowlaris, etc., in recognizing RAID * arrays that span controllers (Wow!). */ #define _DPT_C_ #include "opt_dpt.h" #include "opt_eisa.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include /* dpt_isa.c, dpt_eisa.c, and dpt_pci.c need this in a central place */ devclass_t dpt_devclass; #define microtime_now dpt_time_now() #define dpt_inl(dpt, port) \ bus_read_4((dpt)->io_res, (dpt)->io_offset + port) #define dpt_inb(dpt, port) \ bus_read_1((dpt)->io_res, (dpt)->io_offset + port) #define dpt_outl(dpt, port, value) \ bus_write_4((dpt)->io_res, (dpt)->io_offset + port, value) #define dpt_outb(dpt, port, value) \ bus_write_1((dpt)->io_res, (dpt)->io_offset + port, value) /* * These will have to be setup by parameters passed at boot/load time. For * performance reasons, we make them constants for the time being. */ #define dpt_min_segs DPT_MAX_SEGS #define dpt_max_segs DPT_MAX_SEGS /* Definitions for our use of the SIM private CCB area */ #define ccb_dccb_ptr spriv_ptr0 #define ccb_dpt_ptr spriv_ptr1 /* ================= Private Inline Function declarations ===================*/ static __inline int dpt_just_reset(dpt_softc_t * dpt); static __inline int dpt_raid_busy(dpt_softc_t * dpt); #ifdef DEV_EISA static __inline int dpt_pio_wait (u_int32_t, u_int, u_int, u_int); #endif static __inline int dpt_wait(dpt_softc_t *dpt, u_int bits, u_int state); static __inline struct dpt_ccb* dptgetccb(struct dpt_softc *dpt); static __inline void dptfreeccb(struct dpt_softc *dpt, struct dpt_ccb *dccb); static __inline bus_addr_t dptccbvtop(struct dpt_softc *dpt, struct dpt_ccb *dccb); static __inline int dpt_send_immediate(dpt_softc_t *dpt, eata_ccb_t *cmd_block, u_int32_t cmd_busaddr, u_int retries, u_int ifc, u_int code, u_int code2); /* ==================== Private Function declarations =======================*/ static void dptmapmem(void *arg, bus_dma_segment_t *segs, int nseg, int error); static struct sg_map_node* dptallocsgmap(struct dpt_softc *dpt); static int dptallocccbs(dpt_softc_t *dpt); static int dpt_get_conf(dpt_softc_t *dpt, dpt_ccb_t *dccb, u_int32_t dccb_busaddr, u_int size, u_int page, u_int target, int extent); static void dpt_detect_cache(dpt_softc_t *dpt, dpt_ccb_t *dccb, u_int32_t dccb_busaddr, u_int8_t *buff); static void dpt_poll(struct cam_sim *sim); static void dpt_intr_locked(dpt_softc_t *dpt); static void dptexecuteccb(void *arg, bus_dma_segment_t *dm_segs, int nseg, int error); static void dpt_action(struct cam_sim *sim, union ccb *ccb); static int dpt_send_eata_command(dpt_softc_t *dpt, eata_ccb_t *cmd, u_int32_t cmd_busaddr, u_int command, u_int retries, u_int ifc, u_int code, u_int code2); static void dptprocesserror(dpt_softc_t *dpt, dpt_ccb_t *dccb, union ccb *ccb, u_int hba_stat, u_int scsi_stat, u_int32_t resid); static void dpttimeout(void *arg); static void dptshutdown(void *arg, int howto); /* ================= Private Inline Function definitions ====================*/ static __inline int dpt_just_reset(dpt_softc_t * dpt) { if ((dpt_inb(dpt, 2) == 'D') && (dpt_inb(dpt, 3) == 'P') && (dpt_inb(dpt, 4) == 'T') && (dpt_inb(dpt, 5) == 'H')) return (1); else return (0); } static __inline int dpt_raid_busy(dpt_softc_t * dpt) { if ((dpt_inb(dpt, 0) == 'D') && (dpt_inb(dpt, 1) == 'P') && (dpt_inb(dpt, 2) == 'T')) return (1); else return (0); } #ifdef DEV_EISA static __inline int dpt_pio_wait (u_int32_t base, u_int reg, u_int bits, u_int state) { int i; u_int c; for (i = 0; i < 20000; i++) { /* wait 20ms for not busy */ c = inb(base + reg) & bits; if (!(c == state)) return (0); else DELAY(50); } return (-1); } #endif static __inline int dpt_wait(dpt_softc_t *dpt, u_int bits, u_int state) { int i; u_int c; for (i = 0; i < 20000; i++) { /* wait 20ms for not busy */ c = dpt_inb(dpt, HA_RSTATUS) & bits; if (c == state) return (0); else DELAY(50); } return (-1); } static __inline struct dpt_ccb* dptgetccb(struct dpt_softc *dpt) { struct dpt_ccb* dccb; if (!dumping) mtx_assert(&dpt->lock, MA_OWNED); if ((dccb = SLIST_FIRST(&dpt->free_dccb_list)) != NULL) { SLIST_REMOVE_HEAD(&dpt->free_dccb_list, links); dpt->free_dccbs--; } else if (dpt->total_dccbs < dpt->max_dccbs) { dptallocccbs(dpt); dccb = SLIST_FIRST(&dpt->free_dccb_list); if (dccb == NULL) device_printf(dpt->dev, "Can't malloc DCCB\n"); else { SLIST_REMOVE_HEAD(&dpt->free_dccb_list, links); dpt->free_dccbs--; } } return (dccb); } static __inline void dptfreeccb(struct dpt_softc *dpt, struct dpt_ccb *dccb) { if (!dumping) mtx_assert(&dpt->lock, MA_OWNED); if ((dccb->state & DCCB_ACTIVE) != 0) LIST_REMOVE(&dccb->ccb->ccb_h, sim_links.le); if ((dccb->state & DCCB_RELEASE_SIMQ) != 0) dccb->ccb->ccb_h.status |= CAM_RELEASE_SIMQ; else if (dpt->resource_shortage != 0 && (dccb->ccb->ccb_h.status & CAM_RELEASE_SIMQ) == 0) { dccb->ccb->ccb_h.status |= CAM_RELEASE_SIMQ; dpt->resource_shortage = FALSE; } dccb->state = DCCB_FREE; SLIST_INSERT_HEAD(&dpt->free_dccb_list, dccb, links); ++dpt->free_dccbs; } static __inline bus_addr_t dptccbvtop(struct dpt_softc *dpt, struct dpt_ccb *dccb) { return (dpt->dpt_ccb_busbase + (u_int32_t)((caddr_t)dccb - (caddr_t)dpt->dpt_dccbs)); } static __inline struct dpt_ccb * dptccbptov(struct dpt_softc *dpt, bus_addr_t busaddr) { return (dpt->dpt_dccbs + ((struct dpt_ccb *)busaddr - (struct dpt_ccb *)dpt->dpt_ccb_busbase)); } /* * Send a command for immediate execution by the DPT * See above function for IMPORTANT notes. */ static __inline int dpt_send_immediate(dpt_softc_t *dpt, eata_ccb_t *cmd_block, u_int32_t cmd_busaddr, u_int retries, u_int ifc, u_int code, u_int code2) { return (dpt_send_eata_command(dpt, cmd_block, cmd_busaddr, EATA_CMD_IMMEDIATE, retries, ifc, code, code2)); } /* ===================== Private Function definitions =======================*/ static void dptmapmem(void *arg, bus_dma_segment_t *segs, int nseg, int error) { bus_addr_t *busaddrp; busaddrp = (bus_addr_t *)arg; *busaddrp = segs->ds_addr; } static struct sg_map_node * dptallocsgmap(struct dpt_softc *dpt) { struct sg_map_node *sg_map; sg_map = malloc(sizeof(*sg_map), M_DEVBUF, M_NOWAIT); if (sg_map == NULL) return (NULL); /* Allocate S/G space for the next batch of CCBS */ if (bus_dmamem_alloc(dpt->sg_dmat, (void **)&sg_map->sg_vaddr, BUS_DMA_NOWAIT, &sg_map->sg_dmamap) != 0) { free(sg_map, M_DEVBUF); return (NULL); } (void)bus_dmamap_load(dpt->sg_dmat, sg_map->sg_dmamap, sg_map->sg_vaddr, PAGE_SIZE, dptmapmem, &sg_map->sg_physaddr, /*flags*/0); SLIST_INSERT_HEAD(&dpt->sg_maps, sg_map, links); return (sg_map); } /* * Allocate another chunk of CCB's. Return count of entries added. */ static int dptallocccbs(dpt_softc_t *dpt) { struct dpt_ccb *next_ccb; struct sg_map_node *sg_map; bus_addr_t physaddr; dpt_sg_t *segs; int newcount; int i; if (!dumping) mtx_assert(&dpt->lock, MA_OWNED); next_ccb = &dpt->dpt_dccbs[dpt->total_dccbs]; if (next_ccb == dpt->dpt_dccbs) { /* * First time through. Re-use the S/G * space we allocated for initialization * CCBS. */ sg_map = SLIST_FIRST(&dpt->sg_maps); } else { sg_map = dptallocsgmap(dpt); } if (sg_map == NULL) return (0); segs = sg_map->sg_vaddr; physaddr = sg_map->sg_physaddr; newcount = (PAGE_SIZE / (dpt->sgsize * sizeof(dpt_sg_t))); for (i = 0; dpt->total_dccbs < dpt->max_dccbs && i < newcount; i++) { int error; error = bus_dmamap_create(dpt->buffer_dmat, /*flags*/0, &next_ccb->dmamap); if (error != 0) break; callout_init_mtx(&next_ccb->timer, &dpt->lock, 0); next_ccb->sg_list = segs; next_ccb->sg_busaddr = htonl(physaddr); next_ccb->eata_ccb.cp_dataDMA = htonl(physaddr); next_ccb->eata_ccb.cp_statDMA = htonl(dpt->sp_physaddr); next_ccb->eata_ccb.cp_reqDMA = htonl(dptccbvtop(dpt, next_ccb) + offsetof(struct dpt_ccb, sense_data)); next_ccb->eata_ccb.cp_busaddr = dpt->dpt_ccb_busend; next_ccb->state = DCCB_FREE; next_ccb->tag = dpt->total_dccbs; SLIST_INSERT_HEAD(&dpt->free_dccb_list, next_ccb, links); segs += dpt->sgsize; physaddr += (dpt->sgsize * sizeof(dpt_sg_t)); dpt->dpt_ccb_busend += sizeof(*next_ccb); next_ccb++; dpt->total_dccbs++; } return (i); } #ifdef DEV_EISA dpt_conf_t * dpt_pio_get_conf (u_int32_t base) { static dpt_conf_t * conf; u_int16_t * p; int i; /* * Allocate a dpt_conf_t */ if (!conf) { conf = (dpt_conf_t *)malloc(sizeof(dpt_conf_t), M_DEVBUF, M_NOWAIT | M_ZERO); } /* * If we didn't get one then we probably won't ever get one. */ if (!conf) { printf("dpt: unable to allocate dpt_conf_t\n"); return (NULL); } /* * Reset the controller. */ outb((base + HA_WCOMMAND), EATA_CMD_RESET); /* * Wait for the controller to become ready. * For some reason there can be -no- delays after calling reset * before we wait on ready status. */ if (dpt_pio_wait(base, HA_RSTATUS, HA_SBUSY, 0)) { printf("dpt: timeout waiting for controller to become ready\n"); return (NULL); } if (dpt_pio_wait(base, HA_RAUXSTAT, HA_ABUSY, 0)) { printf("dpt: timetout waiting for adapter ready.\n"); return (NULL); } /* * Send the PIO_READ_CONFIG command. */ outb((base + HA_WCOMMAND), EATA_CMD_PIO_READ_CONFIG); /* * Read the data into the struct. */ p = (u_int16_t *)conf; for (i = 0; i < (sizeof(dpt_conf_t) / 2); i++) { if (dpt_pio_wait(base, HA_RSTATUS, HA_SDRQ, 0)) { if (bootverbose) printf("dpt: timeout in data read.\n"); return (NULL); } (*p) = inw(base + HA_RDATA); p++; } if (inb(base + HA_RSTATUS) & HA_SERROR) { if (bootverbose) printf("dpt: error reading configuration data.\n"); return (NULL); } #define BE_EATA_SIGNATURE 0x45415441 #define LE_EATA_SIGNATURE 0x41544145 /* * Test to see if we have a valid card. */ if ((conf->signature == BE_EATA_SIGNATURE) || (conf->signature == LE_EATA_SIGNATURE)) { while (inb(base + HA_RSTATUS) & HA_SDRQ) { inw(base + HA_RDATA); } return (conf); } return (NULL); } #endif /* * Read a configuration page into the supplied dpt_cont_t buffer. */ static int dpt_get_conf(dpt_softc_t *dpt, dpt_ccb_t *dccb, u_int32_t dccb_busaddr, u_int size, u_int page, u_int target, int extent) { eata_ccb_t *cp; u_int8_t status; int ndx; int result; mtx_assert(&dpt->lock, MA_OWNED); cp = &dccb->eata_ccb; bzero((void *)(uintptr_t)(volatile void *)dpt->sp, sizeof(*dpt->sp)); cp->Interpret = 1; cp->DataIn = 1; cp->Auto_Req_Sen = 1; cp->reqlen = sizeof(struct scsi_sense_data); cp->cp_id = target; cp->cp_LUN = 0; /* In the EATA packet */ cp->cp_lun = 0; /* In the SCSI command */ cp->cp_scsi_cmd = INQUIRY; cp->cp_len = size; cp->cp_extent = extent; cp->cp_page = page; cp->cp_channel = 0; /* DNC, Interpret mode is set */ cp->cp_identify = 1; cp->cp_datalen = htonl(size); /* * This could be a simple for loop, but we suspected the compiler To * have optimized it a bit too much. Wait for the controller to * become ready */ while (((status = dpt_inb(dpt, HA_RSTATUS)) != (HA_SREADY | HA_SSC) && (status != (HA_SREADY | HA_SSC | HA_SERROR)) && (status != (HA_SDRDY | HA_SERROR | HA_SDRQ))) || (dpt_wait(dpt, HA_SBUSY, 0))) { /* * RAID Drives still Spinning up? (This should only occur if * the DPT controller is in a NON PC (PCI?) platform). */ if (dpt_raid_busy(dpt)) { device_printf(dpt->dev, "WARNING: Get_conf() RSUS failed.\n"); return (0); } } DptStat_Reset_BUSY(dpt->sp); /* * XXXX We might want to do something more clever than aborting at * this point, like resetting (rebooting) the controller and trying * again. */ if ((result = dpt_send_eata_command(dpt, cp, dccb_busaddr, EATA_CMD_DMA_SEND_CP, 10000, 0, 0, 0)) != 0) { device_printf(dpt->dev, "WARNING: Get_conf() failed (%d) to send " "EATA_CMD_DMA_READ_CONFIG\n", result); return (0); } /* Wait for two seconds for a response. This can be slow */ for (ndx = 0; (ndx < 20000) && !((status = dpt_inb(dpt, HA_RAUXSTAT)) & HA_AIRQ); ndx++) { DELAY(50); } /* Grab the status and clear interrupts */ status = dpt_inb(dpt, HA_RSTATUS); /* * Check the status carefully. Return only if the * command was successful. */ if (((status & HA_SERROR) == 0) && (dpt->sp->hba_stat == 0) && (dpt->sp->scsi_stat == 0) && (dpt->sp->residue_len == 0)) return (0); if (dpt->sp->scsi_stat == SCSI_STATUS_CHECK_COND) return (0); return (1); } /* Detect Cache parameters and size */ static void dpt_detect_cache(dpt_softc_t *dpt, dpt_ccb_t *dccb, u_int32_t dccb_busaddr, u_int8_t *buff) { eata_ccb_t *cp; u_int8_t *param; int bytes; int result; int ndx; u_int8_t status; mtx_assert(&dpt->lock, MA_OWNED); /* * Default setting, for best performance.. * This is what virtually all cards default to.. */ dpt->cache_type = DPT_CACHE_WRITEBACK; dpt->cache_size = 0; cp = &dccb->eata_ccb; bzero((void *)(uintptr_t)(volatile void *)dpt->sp, sizeof(dpt->sp)); bzero(buff, 512); /* Setup the command structure */ cp->Interpret = 1; cp->DataIn = 1; cp->Auto_Req_Sen = 1; cp->reqlen = sizeof(struct scsi_sense_data); cp->cp_id = 0; /* who cares? The HBA will interpret.. */ cp->cp_LUN = 0; /* In the EATA packet */ cp->cp_lun = 0; /* In the SCSI command */ cp->cp_channel = 0; cp->cp_scsi_cmd = EATA_CMD_DMA_SEND_CP; cp->cp_len = 56; cp->cp_extent = 0; cp->cp_page = 0; cp->cp_identify = 1; cp->cp_dispri = 1; /* * Build the EATA Command Packet structure * for a Log Sense Command. */ cp->cp_cdb[0] = 0x4d; cp->cp_cdb[1] = 0x0; cp->cp_cdb[2] = 0x40 | 0x33; cp->cp_cdb[7] = 1; cp->cp_datalen = htonl(512); result = dpt_send_eata_command(dpt, cp, dccb_busaddr, EATA_CMD_DMA_SEND_CP, 10000, 0, 0, 0); if (result != 0) { device_printf(dpt->dev, "WARNING: detect_cache() failed (%d) to send " "EATA_CMD_DMA_SEND_CP\n", result); return; } /* Wait for two seconds for a response. This can be slow... */ for (ndx = 0; (ndx < 20000) && !((status = dpt_inb(dpt, HA_RAUXSTAT)) & HA_AIRQ); ndx++) { DELAY(50); } /* Grab the status and clear interrupts */ status = dpt_inb(dpt, HA_RSTATUS); /* * Sanity check */ if (buff[0] != 0x33) { return; } bytes = DPT_HCP_LENGTH(buff); param = DPT_HCP_FIRST(buff); if (DPT_HCP_CODE(param) != 1) { /* * DPT Log Page layout error */ device_printf(dpt->dev, "NOTICE: Log Page (1) layout error\n"); return; } if (!(param[4] & 0x4)) { dpt->cache_type = DPT_NO_CACHE; return; } while (DPT_HCP_CODE(param) != 6) { param = DPT_HCP_NEXT(param); if ((param < buff) || (param >= &buff[bytes])) { return; } } if (param[4] & 0x2) { /* * Cache disabled */ dpt->cache_type = DPT_NO_CACHE; return; } if (param[4] & 0x4) { dpt->cache_type = DPT_CACHE_WRITETHROUGH; } /* XXX This isn't correct. This log parameter only has two bytes.... */ #if 0 dpt->cache_size = param[5] | (param[6] << 8) | (param[7] << 16) | (param[8] << 24); #endif } static void dpt_poll(struct cam_sim *sim) { dpt_intr_locked(cam_sim_softc(sim)); } static void dptexecuteccb(void *arg, bus_dma_segment_t *dm_segs, int nseg, int error) { struct dpt_ccb *dccb; union ccb *ccb; struct dpt_softc *dpt; dccb = (struct dpt_ccb *)arg; ccb = dccb->ccb; dpt = (struct dpt_softc *)ccb->ccb_h.ccb_dpt_ptr; if (!dumping) mtx_assert(&dpt->lock, MA_OWNED); if (error != 0) { if (error != EFBIG) device_printf(dpt->dev, "Unexepected error 0x%x returned from " "bus_dmamap_load\n", error); if (ccb->ccb_h.status == CAM_REQ_INPROG) { xpt_freeze_devq(ccb->ccb_h.path, /*count*/1); ccb->ccb_h.status = CAM_REQ_TOO_BIG|CAM_DEV_QFRZN; } dptfreeccb(dpt, dccb); xpt_done(ccb); return; } if (nseg != 0) { dpt_sg_t *sg; bus_dma_segment_t *end_seg; bus_dmasync_op_t op; end_seg = dm_segs + nseg; /* Copy the segments into our SG list */ sg = dccb->sg_list; while (dm_segs < end_seg) { sg->seg_len = htonl(dm_segs->ds_len); sg->seg_addr = htonl(dm_segs->ds_addr); sg++; dm_segs++; } if (nseg > 1) { dccb->eata_ccb.scatter = 1; dccb->eata_ccb.cp_dataDMA = dccb->sg_busaddr; dccb->eata_ccb.cp_datalen = htonl(nseg * sizeof(dpt_sg_t)); } else { dccb->eata_ccb.cp_dataDMA = dccb->sg_list[0].seg_addr; dccb->eata_ccb.cp_datalen = dccb->sg_list[0].seg_len; } if ((ccb->ccb_h.flags & CAM_DIR_MASK) == CAM_DIR_IN) op = BUS_DMASYNC_PREREAD; else op = BUS_DMASYNC_PREWRITE; bus_dmamap_sync(dpt->buffer_dmat, dccb->dmamap, op); } else { dccb->eata_ccb.cp_dataDMA = 0; dccb->eata_ccb.cp_datalen = 0; } /* * Last time we need to check if this CCB needs to * be aborted. */ if (ccb->ccb_h.status != CAM_REQ_INPROG) { if (nseg != 0) bus_dmamap_unload(dpt->buffer_dmat, dccb->dmamap); dptfreeccb(dpt, dccb); xpt_done(ccb); return; } dccb->state |= DCCB_ACTIVE; ccb->ccb_h.status |= CAM_SIM_QUEUED; LIST_INSERT_HEAD(&dpt->pending_ccb_list, &ccb->ccb_h, sim_links.le); callout_reset_sbt(&dccb->timer, SBT_1MS * ccb->ccb_h.timeout, 0, dpttimeout, dccb, 0); if (dpt_send_eata_command(dpt, &dccb->eata_ccb, dccb->eata_ccb.cp_busaddr, EATA_CMD_DMA_SEND_CP, 0, 0, 0, 0) != 0) { ccb->ccb_h.status = CAM_NO_HBA; /* HBA dead or just busy?? */ if (nseg != 0) bus_dmamap_unload(dpt->buffer_dmat, dccb->dmamap); dptfreeccb(dpt, dccb); xpt_done(ccb); } } static void dpt_action(struct cam_sim *sim, union ccb *ccb) { struct dpt_softc *dpt; CAM_DEBUG(ccb->ccb_h.path, CAM_DEBUG_TRACE, ("dpt_action\n")); dpt = (struct dpt_softc *)cam_sim_softc(sim); mtx_assert(&dpt->lock, MA_OWNED); if ((dpt->state & DPT_HA_SHUTDOWN_ACTIVE) != 0) { xpt_print_path(ccb->ccb_h.path); printf("controller is shutdown. Aborting CCB.\n"); ccb->ccb_h.status = CAM_NO_HBA; xpt_done(ccb); return; } switch (ccb->ccb_h.func_code) { /* Common cases first */ case XPT_SCSI_IO: /* Execute the requested I/O operation */ { struct ccb_scsiio *csio; struct ccb_hdr *ccbh; struct dpt_ccb *dccb; struct eata_ccb *eccb; csio = &ccb->csio; ccbh = &ccb->ccb_h; /* Max CDB length is 12 bytes */ if (csio->cdb_len > 12) { ccb->ccb_h.status = CAM_REQ_INVALID; xpt_done(ccb); return; } if ((dccb = dptgetccb(dpt)) == NULL) { dpt->resource_shortage = 1; xpt_freeze_simq(sim, /*count*/1); ccb->ccb_h.status = CAM_REQUEUE_REQ; xpt_done(ccb); return; } eccb = &dccb->eata_ccb; /* Link dccb and ccb so we can find one from the other */ dccb->ccb = ccb; ccb->ccb_h.ccb_dccb_ptr = dccb; ccb->ccb_h.ccb_dpt_ptr = dpt; /* * Explicitly set all flags so that the compiler can * be smart about setting them. */ eccb->SCSI_Reset = 0; eccb->HBA_Init = 0; eccb->Auto_Req_Sen = (ccb->ccb_h.flags & CAM_DIS_AUTOSENSE) ? 0 : 1; eccb->scatter = 0; eccb->Quick = 0; eccb->Interpret = ccb->ccb_h.target_id == dpt->hostid[cam_sim_bus(sim)] ? 1 : 0; eccb->DataOut = (ccb->ccb_h.flags & CAM_DIR_OUT) ? 1 : 0; eccb->DataIn = (ccb->ccb_h.flags & CAM_DIR_IN) ? 1 : 0; eccb->reqlen = csio->sense_len; eccb->cp_id = ccb->ccb_h.target_id; eccb->cp_channel = cam_sim_bus(sim); eccb->cp_LUN = ccb->ccb_h.target_lun; eccb->cp_luntar = 0; eccb->cp_dispri = (ccb->ccb_h.flags & CAM_DIS_DISCONNECT) ? 0 : 1; eccb->cp_identify = 1; if ((ccb->ccb_h.flags & CAM_TAG_ACTION_VALID) != 0 && csio->tag_action != CAM_TAG_ACTION_NONE) { eccb->cp_msg[0] = csio->tag_action; eccb->cp_msg[1] = dccb->tag; } else { eccb->cp_msg[0] = 0; eccb->cp_msg[1] = 0; } eccb->cp_msg[2] = 0; if ((ccb->ccb_h.flags & CAM_CDB_POINTER) != 0) { if ((ccb->ccb_h.flags & CAM_CDB_PHYS) == 0) { bcopy(csio->cdb_io.cdb_ptr, eccb->cp_cdb, csio->cdb_len); } else { /* I guess I could map it in... */ ccb->ccb_h.status = CAM_REQ_INVALID; dptfreeccb(dpt, dccb); xpt_done(ccb); return; } } else { bcopy(csio->cdb_io.cdb_bytes, eccb->cp_cdb, csio->cdb_len); } /* * If we have any data to send with this command, * map it into bus space. */ /* Only use S/G if there is a transfer */ if ((ccbh->flags & CAM_DIR_MASK) != CAM_DIR_NONE) { int error; error = bus_dmamap_load_ccb(dpt->buffer_dmat, dccb->dmamap, ccb, dptexecuteccb, dccb, /*flags*/0); if (error == EINPROGRESS) { /* * So as to maintain ordering, * freeze the controller queue * until our mapping is * returned. */ xpt_freeze_simq(sim, 1); dccb->state |= CAM_RELEASE_SIMQ; } } else { /* * XXX JGibbs. * Does it want them both on or both off? * CAM_DIR_NONE is both on, so this code can * be removed if this is also what the DPT * exptects. */ eccb->DataOut = 0; eccb->DataIn = 0; dptexecuteccb(dccb, NULL, 0, 0); } break; } case XPT_RESET_DEV: /* Bus Device Reset the specified SCSI device */ case XPT_ABORT: /* Abort the specified CCB */ /* XXX Implement */ ccb->ccb_h.status = CAM_REQ_INVALID; xpt_done(ccb); break; case XPT_SET_TRAN_SETTINGS: { ccb->ccb_h.status = CAM_FUNC_NOTAVAIL; xpt_done(ccb); break; } case XPT_GET_TRAN_SETTINGS: /* Get default/user set transfer settings for the target */ { struct ccb_trans_settings *cts = &ccb->cts; struct ccb_trans_settings_scsi *scsi = &cts->proto_specific.scsi; struct ccb_trans_settings_spi *spi = &cts->xport_specific.spi; cts->protocol = PROTO_SCSI; cts->protocol_version = SCSI_REV_2; cts->transport = XPORT_SPI; cts->transport_version = 2; if (cts->type == CTS_TYPE_USER_SETTINGS) { spi->flags = CTS_SPI_FLAGS_DISC_ENB; spi->bus_width = (dpt->max_id > 7) ? MSG_EXT_WDTR_BUS_8_BIT : MSG_EXT_WDTR_BUS_16_BIT; spi->sync_period = 25; /* 10MHz */ if (spi->sync_period != 0) spi->sync_offset = 15; scsi->flags = CTS_SCSI_FLAGS_TAG_ENB; spi->valid = CTS_SPI_VALID_SYNC_RATE | CTS_SPI_VALID_SYNC_OFFSET | CTS_SPI_VALID_BUS_WIDTH | CTS_SPI_VALID_DISC; scsi->valid = CTS_SCSI_VALID_TQ; ccb->ccb_h.status = CAM_REQ_CMP; } else { ccb->ccb_h.status = CAM_FUNC_NOTAVAIL; } xpt_done(ccb); break; } case XPT_CALC_GEOMETRY: { /* * XXX Use Adaptec translation until I find out how to * get this information from the card. */ cam_calc_geometry(&ccb->ccg, /*extended*/1); xpt_done(ccb); break; } case XPT_RESET_BUS: /* Reset the specified SCSI bus */ { /* XXX Implement */ ccb->ccb_h.status = CAM_REQ_CMP; xpt_done(ccb); break; } case XPT_TERM_IO: /* Terminate the I/O process */ /* XXX Implement */ ccb->ccb_h.status = CAM_REQ_INVALID; xpt_done(ccb); break; case XPT_PATH_INQ: /* Path routing inquiry */ { struct ccb_pathinq *cpi = &ccb->cpi; cpi->version_num = 1; cpi->hba_inquiry = PI_SDTR_ABLE|PI_TAG_ABLE; if (dpt->max_id > 7) cpi->hba_inquiry |= PI_WIDE_16; cpi->target_sprt = 0; cpi->hba_misc = 0; cpi->hba_eng_cnt = 0; cpi->max_target = dpt->max_id; cpi->max_lun = dpt->max_lun; cpi->initiator_id = dpt->hostid[cam_sim_bus(sim)]; cpi->bus_id = cam_sim_bus(sim); cpi->base_transfer_speed = 3300; strlcpy(cpi->sim_vid, "FreeBSD", SIM_IDLEN); strlcpy(cpi->hba_vid, "DPT", HBA_IDLEN); strlcpy(cpi->dev_name, cam_sim_name(sim), DEV_IDLEN); cpi->unit_number = cam_sim_unit(sim); - cpi->transport = XPORT_SPI; - cpi->transport_version = 2; - cpi->protocol = PROTO_SCSI; - cpi->protocol_version = SCSI_REV_2; + cpi->transport = XPORT_SPI; + cpi->transport_version = 2; + cpi->protocol = PROTO_SCSI; + cpi->protocol_version = SCSI_REV_2; cpi->ccb_h.status = CAM_REQ_CMP; xpt_done(ccb); break; } default: ccb->ccb_h.status = CAM_REQ_INVALID; xpt_done(ccb); break; } } /* * This routine will try to send an EATA command to the DPT HBA. * It will, by default, try 20,000 times, waiting 50us between tries. * It returns 0 on success and 1 on failure. */ static int dpt_send_eata_command(dpt_softc_t *dpt, eata_ccb_t *cmd_block, u_int32_t cmd_busaddr, u_int command, u_int retries, u_int ifc, u_int code, u_int code2) { u_int loop; if (!retries) retries = 20000; /* * I hate this polling nonsense. Wish there was a way to tell the DPT * to go get commands at its own pace, or to interrupt when ready. * In the mean time we will measure how many itterations it really * takes. */ for (loop = 0; loop < retries; loop++) { if ((dpt_inb(dpt, HA_RAUXSTAT) & HA_ABUSY) == 0) break; else DELAY(50); } if (loop < retries) { #ifdef DPT_MEASURE_PERFORMANCE if (loop > dpt->performance.max_eata_tries) dpt->performance.max_eata_tries = loop; if (loop < dpt->performance.min_eata_tries) dpt->performance.min_eata_tries = loop; #endif } else { #ifdef DPT_MEASURE_PERFORMANCE ++dpt->performance.command_too_busy; #endif return (1); } /* The controller is alive, advance the wedge timer */ #ifdef DPT_RESET_HBA dpt->last_contact = microtime_now; #endif if (cmd_block == NULL) cmd_busaddr = 0; #if (BYTE_ORDER == BIG_ENDIAN) else { cmd_busaddr = ((cmd_busaddr >> 24) & 0xFF) | ((cmd_busaddr >> 16) & 0xFF) | ((cmd_busaddr >> 8) & 0xFF) | (cmd_busaddr & 0xFF); } #endif /* And now the address */ dpt_outl(dpt, HA_WDMAADDR, cmd_busaddr); if (command == EATA_CMD_IMMEDIATE) { if (cmd_block == NULL) { dpt_outb(dpt, HA_WCODE2, code2); dpt_outb(dpt, HA_WCODE, code); } dpt_outb(dpt, HA_WIFC, ifc); } dpt_outb(dpt, HA_WCOMMAND, command); return (0); } /* ==================== Exported Function definitions =======================*/ void dpt_alloc(device_t dev) { dpt_softc_t *dpt = device_get_softc(dev); int i; mtx_init(&dpt->lock, "dpt", NULL, MTX_DEF); SLIST_INIT(&dpt->free_dccb_list); LIST_INIT(&dpt->pending_ccb_list); for (i = 0; i < MAX_CHANNELS; i++) dpt->resetlevel[i] = DPT_HA_OK; #ifdef DPT_MEASURE_PERFORMANCE dpt_reset_performance(dpt); #endif /* DPT_MEASURE_PERFORMANCE */ return; } void dpt_free(struct dpt_softc *dpt) { switch (dpt->init_level) { default: case 5: bus_dmamap_unload(dpt->dccb_dmat, dpt->dccb_dmamap); case 4: bus_dmamem_free(dpt->dccb_dmat, dpt->dpt_dccbs, dpt->dccb_dmamap); case 3: bus_dma_tag_destroy(dpt->dccb_dmat); case 2: bus_dma_tag_destroy(dpt->buffer_dmat); case 1: { struct sg_map_node *sg_map; while ((sg_map = SLIST_FIRST(&dpt->sg_maps)) != NULL) { SLIST_REMOVE_HEAD(&dpt->sg_maps, links); bus_dmamap_unload(dpt->sg_dmat, sg_map->sg_dmamap); bus_dmamem_free(dpt->sg_dmat, sg_map->sg_vaddr, sg_map->sg_dmamap); free(sg_map, M_DEVBUF); } bus_dma_tag_destroy(dpt->sg_dmat); } case 0: break; } mtx_destroy(&dpt->lock); } int dpt_alloc_resources (device_t dev) { dpt_softc_t * dpt; int error; dpt = device_get_softc(dev); dpt->io_res = bus_alloc_resource_any(dev, dpt->io_type, &dpt->io_rid, RF_ACTIVE); if (dpt->io_res == NULL) { device_printf(dev, "No I/O space?!\n"); error = ENOMEM; goto bad; } dpt->irq_res = bus_alloc_resource_any(dev, SYS_RES_IRQ, &dpt->irq_rid, RF_ACTIVE); if (dpt->irq_res == NULL) { device_printf(dev, "No IRQ!\n"); error = ENOMEM; goto bad; } return (0); bad: return(error); } void dpt_release_resources (device_t dev) { struct dpt_softc * dpt; dpt = device_get_softc(dev); if (dpt->ih) bus_teardown_intr(dev, dpt->irq_res, dpt->ih); if (dpt->io_res) bus_release_resource(dev, dpt->io_type, dpt->io_rid, dpt->io_res); if (dpt->irq_res) bus_release_resource(dev, SYS_RES_IRQ, dpt->irq_rid, dpt->irq_res); if (dpt->drq_res) bus_release_resource(dev, SYS_RES_DRQ, dpt->drq_rid, dpt->drq_res); return; } static u_int8_t string_sizes[] = { sizeof(((dpt_inq_t*)NULL)->vendor), sizeof(((dpt_inq_t*)NULL)->modelNum), sizeof(((dpt_inq_t*)NULL)->firmware), sizeof(((dpt_inq_t*)NULL)->protocol), }; int dpt_init(struct dpt_softc *dpt) { dpt_conf_t conf; struct sg_map_node *sg_map; dpt_ccb_t *dccb; u_int8_t *strp; int index; int i; int retval; dpt->init_level = 0; SLIST_INIT(&dpt->sg_maps); mtx_lock(&dpt->lock); #ifdef DPT_RESET_BOARD device_printf(dpt->dev, "resetting HBA\n"); dpt_outb(dpt, HA_WCOMMAND, EATA_CMD_RESET); DELAY(750000); /* XXX Shouldn't we poll a status register or something??? */ #endif /* DMA tag for our S/G structures. We allocate in page sized chunks */ if (bus_dma_tag_create( /* parent */ dpt->parent_dmat, /* alignment */ 1, /* boundary */ 0, /* lowaddr */ BUS_SPACE_MAXADDR, /* highaddr */ BUS_SPACE_MAXADDR, /* filter */ NULL, /* filterarg */ NULL, /* maxsize */ PAGE_SIZE, /* nsegments */ 1, /* maxsegsz */ BUS_SPACE_MAXSIZE_32BIT, /* flags */ 0, /* lockfunc */ NULL, /* lockarg */ NULL, &dpt->sg_dmat) != 0) { goto error_exit; } dpt->init_level++; /* * We allocate our DPT ccbs as a contiguous array of bus dma'able * memory. To get the allocation size, we need to know how many * ccbs the card supports. This requires a ccb. We solve this * chicken and egg problem by allocating some re-usable S/G space * up front, and treating it as our status packet, CCB, and target * memory space for these commands. */ sg_map = dptallocsgmap(dpt); if (sg_map == NULL) goto error_exit; dpt->sp = (volatile dpt_sp_t *)sg_map->sg_vaddr; dccb = (struct dpt_ccb *)(uintptr_t)(volatile void *)&dpt->sp[1]; bzero(dccb, sizeof(*dccb)); dpt->sp_physaddr = sg_map->sg_physaddr; dccb->eata_ccb.cp_dataDMA = htonl(sg_map->sg_physaddr + sizeof(dpt_sp_t) + sizeof(*dccb)); dccb->eata_ccb.cp_busaddr = ~0; dccb->eata_ccb.cp_statDMA = htonl(dpt->sp_physaddr); dccb->eata_ccb.cp_reqDMA = htonl(dpt->sp_physaddr + sizeof(*dccb) + offsetof(struct dpt_ccb, sense_data)); /* Okay. Fetch our config */ bzero(&dccb[1], sizeof(conf)); /* data area */ retval = dpt_get_conf(dpt, dccb, sg_map->sg_physaddr + sizeof(dpt_sp_t), sizeof(conf), 0xc1, 7, 1); if (retval != 0) { device_printf(dpt->dev, "Failed to get board configuration\n"); goto error_exit; } bcopy(&dccb[1], &conf, sizeof(conf)); bzero(&dccb[1], sizeof(dpt->board_data)); retval = dpt_get_conf(dpt, dccb, sg_map->sg_physaddr + sizeof(dpt_sp_t), sizeof(dpt->board_data), 0, conf.scsi_id0, 0); if (retval != 0) { device_printf(dpt->dev, "Failed to get inquiry information\n"); goto error_exit; } bcopy(&dccb[1], &dpt->board_data, sizeof(dpt->board_data)); dpt_detect_cache(dpt, dccb, sg_map->sg_physaddr + sizeof(dpt_sp_t), (u_int8_t *)&dccb[1]); switch (ntohl(conf.splen)) { case DPT_EATA_REVA: dpt->EATA_revision = 'a'; break; case DPT_EATA_REVB: dpt->EATA_revision = 'b'; break; case DPT_EATA_REVC: dpt->EATA_revision = 'c'; break; case DPT_EATA_REVZ: dpt->EATA_revision = 'z'; break; default: dpt->EATA_revision = '?'; } dpt->max_id = conf.MAX_ID; dpt->max_lun = conf.MAX_LUN; dpt->irq = conf.IRQ; dpt->dma_channel = (8 - conf.DMA_channel) & 7; dpt->channels = conf.MAX_CHAN + 1; dpt->state |= DPT_HA_OK; if (conf.SECOND) dpt->primary = FALSE; else dpt->primary = TRUE; dpt->more_support = conf.MORE_support; if (strncmp(dpt->board_data.firmware, "07G0", 4) >= 0) dpt->immediate_support = 1; else dpt->immediate_support = 0; dpt->broken_INQUIRY = FALSE; dpt->cplen = ntohl(conf.cplen); dpt->cppadlen = ntohs(conf.cppadlen); dpt->max_dccbs = ntohs(conf.queuesiz); if (dpt->max_dccbs > 256) { device_printf(dpt->dev, "Max CCBs reduced from %d to " "256 due to tag algorithm\n", dpt->max_dccbs); dpt->max_dccbs = 256; } dpt->hostid[0] = conf.scsi_id0; dpt->hostid[1] = conf.scsi_id1; dpt->hostid[2] = conf.scsi_id2; if (conf.SG_64K) dpt->sgsize = 8192; else dpt->sgsize = ntohs(conf.SGsiz); /* We can only get 64k buffers, so don't bother to waste space. */ if (dpt->sgsize < 17 || dpt->sgsize > 32) dpt->sgsize = 32; if (dpt->sgsize > dpt_max_segs) dpt->sgsize = dpt_max_segs; /* DMA tag for mapping buffers into device visible space. */ if (bus_dma_tag_create( /* parent */ dpt->parent_dmat, /* alignment */ 1, /* boundary */ 0, /* lowaddr */ BUS_SPACE_MAXADDR, /* highaddr */ BUS_SPACE_MAXADDR, /* filter */ NULL, /* filterarg */ NULL, /* maxsize */ DFLTPHYS, /* nsegments */ dpt->sgsize, /* maxsegsz */ BUS_SPACE_MAXSIZE_32BIT, /* flags */ BUS_DMA_ALLOCNOW, /* lockfunc */ busdma_lock_mutex, /* lockarg */ &dpt->lock, &dpt->buffer_dmat) != 0) { device_printf(dpt->dev, "bus_dma_tag_create(...,dpt->buffer_dmat) failed\n"); goto error_exit; } dpt->init_level++; /* DMA tag for our ccb structures and interrupt status packet */ if (bus_dma_tag_create( /* parent */ dpt->parent_dmat, /* alignment */ 1, /* boundary */ 0, /* lowaddr */ BUS_SPACE_MAXADDR, /* highaddr */ BUS_SPACE_MAXADDR, /* filter */ NULL, /* filterarg */ NULL, /* maxsize */ (dpt->max_dccbs * sizeof(struct dpt_ccb)) + sizeof(dpt_sp_t), /* nsegments */ 1, /* maxsegsz */ BUS_SPACE_MAXSIZE_32BIT, /* flags */ 0, /* lockfunc */ NULL, /* lockarg */ NULL, &dpt->dccb_dmat) != 0) { device_printf(dpt->dev, "bus_dma_tag_create(...,dpt->dccb_dmat) failed\n"); goto error_exit; } dpt->init_level++; /* Allocation for our ccbs and interrupt status packet */ if (bus_dmamem_alloc(dpt->dccb_dmat, (void **)&dpt->dpt_dccbs, BUS_DMA_NOWAIT, &dpt->dccb_dmamap) != 0) { device_printf(dpt->dev, "bus_dmamem_alloc(dpt->dccb_dmat,...) failed\n"); goto error_exit; } dpt->init_level++; /* And permanently map them */ bus_dmamap_load(dpt->dccb_dmat, dpt->dccb_dmamap, dpt->dpt_dccbs, (dpt->max_dccbs * sizeof(struct dpt_ccb)) + sizeof(dpt_sp_t), dptmapmem, &dpt->dpt_ccb_busbase, /*flags*/0); /* Clear them out. */ bzero(dpt->dpt_dccbs, (dpt->max_dccbs * sizeof(struct dpt_ccb)) + sizeof(dpt_sp_t)); dpt->dpt_ccb_busend = dpt->dpt_ccb_busbase; dpt->sp = (dpt_sp_t*)&dpt->dpt_dccbs[dpt->max_dccbs]; dpt->sp_physaddr = dpt->dpt_ccb_busbase + (dpt->max_dccbs * sizeof(dpt_ccb_t)); dpt->init_level++; /* Allocate our first batch of ccbs */ if (dptallocccbs(dpt) == 0) { device_printf(dpt->dev, "dptallocccbs(dpt) == 0\n"); mtx_unlock(&dpt->lock); return (2); } /* Prepare for Target Mode */ dpt->target_mode_enabled = 1; /* Nuke excess spaces from inquiry information */ strp = dpt->board_data.vendor; for (i = 0; i < sizeof(string_sizes); i++) { index = string_sizes[i] - 1; while (index && (strp[index] == ' ')) strp[index--] = '\0'; strp += string_sizes[i]; } device_printf(dpt->dev, "%.8s %.16s FW Rev. %.4s, ", dpt->board_data.vendor, dpt->board_data.modelNum, dpt->board_data.firmware); printf("%d channel%s, ", dpt->channels, dpt->channels > 1 ? "s" : ""); if (dpt->cache_type != DPT_NO_CACHE && dpt->cache_size != 0) { printf("%s Cache, ", dpt->cache_type == DPT_CACHE_WRITETHROUGH ? "Write-Through" : "Write-Back"); } printf("%d CCBs\n", dpt->max_dccbs); mtx_unlock(&dpt->lock); return (0); error_exit: mtx_unlock(&dpt->lock); return (1); } int dpt_attach(dpt_softc_t *dpt) { struct cam_devq *devq; int i; /* * Create the device queue for our SIM. */ devq = cam_simq_alloc(dpt->max_dccbs); if (devq == NULL) return (0); mtx_lock(&dpt->lock); for (i = 0; i < dpt->channels; i++) { /* * Construct our SIM entry */ dpt->sims[i] = cam_sim_alloc(dpt_action, dpt_poll, "dpt", dpt, device_get_unit(dpt->dev), &dpt->lock, /*untagged*/2, /*tagged*/dpt->max_dccbs, devq); if (dpt->sims[i] == NULL) { if (i == 0) cam_simq_free(devq); else printf( "%s(): Unable to attach bus %d " "due to resource shortage\n", __func__, i); break; } if (xpt_bus_register(dpt->sims[i], dpt->dev, i) != CAM_SUCCESS){ cam_sim_free(dpt->sims[i], /*free_devq*/i == 0); dpt->sims[i] = NULL; break; } if (xpt_create_path(&dpt->paths[i], /*periph*/NULL, cam_sim_path(dpt->sims[i]), CAM_TARGET_WILDCARD, CAM_LUN_WILDCARD) != CAM_REQ_CMP) { xpt_bus_deregister(cam_sim_path(dpt->sims[i])); cam_sim_free(dpt->sims[i], /*free_devq*/i == 0); dpt->sims[i] = NULL; break; } } mtx_unlock(&dpt->lock); if (i > 0) EVENTHANDLER_REGISTER(shutdown_final, dptshutdown, dpt, SHUTDOWN_PRI_DEFAULT); return (i); } int dpt_detach (device_t dev) { struct dpt_softc * dpt; int i; dpt = device_get_softc(dev); mtx_lock(&dpt->lock); for (i = 0; i < dpt->channels; i++) { #if 0 xpt_async(AC_LOST_DEVICE, dpt->paths[i], NULL); #endif xpt_free_path(dpt->paths[i]); xpt_bus_deregister(cam_sim_path(dpt->sims[i])); cam_sim_free(dpt->sims[i], /*free_devq*/TRUE); } mtx_unlock(&dpt->lock); dptshutdown((void *)dpt, SHUTDOWN_PRI_DEFAULT); dpt_release_resources(dev); dpt_free(dpt); return (0); } /* * This is the interrupt handler for the DPT driver. */ void dpt_intr(void *arg) { dpt_softc_t *dpt; dpt = arg; mtx_lock(&dpt->lock); dpt_intr_locked(dpt); mtx_unlock(&dpt->lock); } void dpt_intr_locked(dpt_softc_t *dpt) { dpt_ccb_t *dccb; union ccb *ccb; u_int status; u_int aux_status; u_int hba_stat; u_int scsi_stat; u_int32_t residue_len; /* Number of bytes not transferred */ /* First order of business is to check if this interrupt is for us */ while (((aux_status = dpt_inb(dpt, HA_RAUXSTAT)) & HA_AIRQ) != 0) { /* * What we want to do now, is to capture the status, all of it, * move it where it belongs, wake up whoever sleeps waiting to * process this result, and get out of here. */ if (dpt->sp->ccb_busaddr < dpt->dpt_ccb_busbase || dpt->sp->ccb_busaddr >= dpt->dpt_ccb_busend) { device_printf(dpt->dev, "Encountered bogus status packet\n"); status = dpt_inb(dpt, HA_RSTATUS); return; } dccb = dptccbptov(dpt, dpt->sp->ccb_busaddr); dpt->sp->ccb_busaddr = ~0; /* Ignore status packets with EOC not set */ if (dpt->sp->EOC == 0) { device_printf(dpt->dev, "ERROR: Request %d received with " "clear EOC.\n Marking as LOST.\n", dccb->transaction_id); /* This CLEARS the interrupt! */ status = dpt_inb(dpt, HA_RSTATUS); continue; } dpt->sp->EOC = 0; /* * Double buffer the status information so the hardware can * work on updating the status packet while we decifer the * one we were just interrupted for. * According to Mark Salyzyn, we only need few pieces of it. */ hba_stat = dpt->sp->hba_stat; scsi_stat = dpt->sp->scsi_stat; residue_len = dpt->sp->residue_len; /* Clear interrupts, check for error */ if ((status = dpt_inb(dpt, HA_RSTATUS)) & HA_SERROR) { /* * Error Condition. Check for magic cookie. Exit * this test on earliest sign of non-reset condition */ /* Check that this is not a board reset interrupt */ if (dpt_just_reset(dpt)) { device_printf(dpt->dev, "HBA rebooted.\n" " All transactions should be " "resubmitted\n"); device_printf(dpt->dev, ">>---->> This is incomplete, " "fix me.... <<----<<"); panic("DPT Rebooted"); } } /* Process CCB */ ccb = dccb->ccb; callout_stop(&dccb->timer); if ((ccb->ccb_h.flags & CAM_DIR_MASK) != CAM_DIR_NONE) { bus_dmasync_op_t op; if ((ccb->ccb_h.flags & CAM_DIR_MASK) == CAM_DIR_IN) op = BUS_DMASYNC_POSTREAD; else op = BUS_DMASYNC_POSTWRITE; bus_dmamap_sync(dpt->buffer_dmat, dccb->dmamap, op); bus_dmamap_unload(dpt->buffer_dmat, dccb->dmamap); } /* Common Case inline... */ if (hba_stat == HA_NO_ERROR) { ccb->csio.scsi_status = scsi_stat; ccb->ccb_h.status = 0; switch (scsi_stat) { case SCSI_STATUS_OK: ccb->ccb_h.status |= CAM_REQ_CMP; break; case SCSI_STATUS_CHECK_COND: case SCSI_STATUS_CMD_TERMINATED: bcopy(&dccb->sense_data, &ccb->csio.sense_data, ccb->csio.sense_len); ccb->ccb_h.status |= CAM_AUTOSNS_VALID; /* FALLTHROUGH */ default: ccb->ccb_h.status |= CAM_SCSI_STATUS_ERROR; /* XXX Freeze DevQ */ break; } ccb->csio.resid = residue_len; dptfreeccb(dpt, dccb); xpt_done(ccb); } else { dptprocesserror(dpt, dccb, ccb, hba_stat, scsi_stat, residue_len); } } } static void dptprocesserror(dpt_softc_t *dpt, dpt_ccb_t *dccb, union ccb *ccb, u_int hba_stat, u_int scsi_stat, u_int32_t resid) { ccb->csio.resid = resid; switch (hba_stat) { case HA_ERR_SEL_TO: ccb->ccb_h.status = CAM_SEL_TIMEOUT; break; case HA_ERR_CMD_TO: ccb->ccb_h.status = CAM_CMD_TIMEOUT; break; case HA_SCSIBUS_RESET: case HA_HBA_POWER_UP: /* Similar effect to a bus reset??? */ ccb->ccb_h.status = CAM_SCSI_BUS_RESET; break; case HA_CP_ABORTED: case HA_CP_RESET: /* XXX ??? */ case HA_CP_ABORT_NA: /* XXX ??? */ case HA_CP_RESET_NA: /* XXX ??? */ if ((ccb->ccb_h.status & CAM_STATUS_MASK) == CAM_REQ_INPROG) ccb->ccb_h.status = CAM_REQ_ABORTED; break; case HA_PCI_PARITY: case HA_PCI_MABORT: case HA_PCI_TABORT: case HA_PCI_STABORT: case HA_BUS_PARITY: case HA_PARITY_ERR: case HA_ECC_ERR: ccb->ccb_h.status = CAM_UNCOR_PARITY; break; case HA_UNX_MSGRJCT: ccb->ccb_h.status = CAM_MSG_REJECT_REC; break; case HA_UNX_BUSPHASE: ccb->ccb_h.status = CAM_SEQUENCE_FAIL; break; case HA_UNX_BUS_FREE: ccb->ccb_h.status = CAM_UNEXP_BUSFREE; break; case HA_SCSI_HUNG: case HA_RESET_STUCK: /* * Dead??? Can the controller get unstuck * from these conditions */ ccb->ccb_h.status = CAM_NO_HBA; break; case HA_RSENSE_FAIL: ccb->ccb_h.status = CAM_AUTOSENSE_FAIL; break; default: device_printf(dpt->dev, "Undocumented Error %x\n", hba_stat); printf("Please mail this message to shimon@simon-shapiro.org\n"); ccb->ccb_h.status = CAM_REQ_CMP_ERR; break; } dptfreeccb(dpt, dccb); xpt_done(ccb); } static void dpttimeout(void *arg) { struct dpt_ccb *dccb; union ccb *ccb; struct dpt_softc *dpt; dccb = (struct dpt_ccb *)arg; ccb = dccb->ccb; dpt = (struct dpt_softc *)ccb->ccb_h.ccb_dpt_ptr; mtx_assert(&dpt->lock, MA_OWNED); xpt_print_path(ccb->ccb_h.path); printf("CCB %p - timed out\n", (void *)dccb); /* * Try to clear any pending jobs. FreeBSD will lose interrupts, * leaving the controller suspended, and commands timed-out. * By calling the interrupt handler, any command thus stuck will be * completed. */ dpt_intr_locked(dpt); if ((dccb->state & DCCB_ACTIVE) == 0) { xpt_print_path(ccb->ccb_h.path); printf("CCB %p - timed out CCB already completed\n", (void *)dccb); return; } /* Abort this particular command. Leave all others running */ dpt_send_immediate(dpt, &dccb->eata_ccb, dccb->eata_ccb.cp_busaddr, /*retries*/20000, EATA_SPECIFIC_ABORT, 0, 0); ccb->ccb_h.status = CAM_CMD_TIMEOUT; } /* * Shutdown the controller and ensure that the cache is completely flushed. * Called from the shutdown_final event after all disk access has completed. */ static void dptshutdown(void *arg, int howto) { dpt_softc_t *dpt; dpt = (dpt_softc_t *)arg; device_printf(dpt->dev, "Shutting down (mode %x) HBA. Please wait...\n", howto); /* * What we do for a shutdown, is give the DPT early power loss warning */ mtx_lock(&dpt->lock); dpt_send_immediate(dpt, NULL, 0, EATA_POWER_OFF_WARN, 0, 0, 0); mtx_unlock(&dpt->lock); DELAY(1000 * 1000 * 5); device_printf(dpt->dev, "Controller was warned of shutdown and is now " "disabled\n"); } /*============================================================================*/ #if 0 #ifdef DPT_RESET_HBA /* ** Function name : dpt_reset_hba ** ** Description : Reset the HBA and properly discard all pending work ** Input : Softc ** Output : Nothing */ static void dpt_reset_hba(dpt_softc_t *dpt) { eata_ccb_t *ccb; dpt_ccb_t dccb, *dccbp; int result; struct scsi_xfer *xs; mtx_assert(&dpt->lock, MA_OWNED); /* Prepare a control block. The SCSI command part is immaterial */ dccb.xs = NULL; dccb.flags = 0; dccb.state = DPT_CCB_STATE_NEW; dccb.std_callback = NULL; dccb.wrbuff_callback = NULL; ccb = &dccb.eata_ccb; ccb->CP_OpCode = EATA_CMD_RESET; ccb->SCSI_Reset = 0; ccb->HBA_Init = 1; ccb->Auto_Req_Sen = 1; ccb->cp_id = 0; /* Should be ignored */ ccb->DataIn = 1; ccb->DataOut = 0; ccb->Interpret = 1; ccb->reqlen = htonl(sizeof(struct scsi_sense_data)); ccb->cp_statDMA = htonl(vtophys(&ccb->cp_statDMA)); ccb->cp_reqDMA = htonl(vtophys(&ccb->cp_reqDMA)); ccb->cp_viraddr = (u_int32_t) & ccb; ccb->cp_msg[0] = HA_IDENTIFY_MSG | HA_DISCO_RECO; ccb->cp_scsi_cmd = 0; /* Should be ignored */ /* Lock up the submitted queue. We are very persistent here */ while (dpt->queue_status & DPT_SUBMITTED_QUEUE_ACTIVE) { DELAY(100); } dpt->queue_status |= DPT_SUBMITTED_QUEUE_ACTIVE; /* Send the RESET message */ if ((result = dpt_send_eata_command(dpt, &dccb.eata_ccb, EATA_CMD_RESET, 0, 0, 0, 0)) != 0) { device_printf(dpt->dev, "Failed to send the RESET message.\n" " Trying cold boot (ouch!)\n"); if ((result = dpt_send_eata_command(dpt, &dccb.eata_ccb, EATA_COLD_BOOT, 0, 0, 0, 0)) != 0) { panic("%s: Faild to cold boot the HBA\n", device_get_nameunit(dpt->dev)); } #ifdef DPT_MEASURE_PERFORMANCE dpt->performance.cold_boots++; #endif /* DPT_MEASURE_PERFORMANCE */ } #ifdef DPT_MEASURE_PERFORMANCE dpt->performance.warm_starts++; #endif /* DPT_MEASURE_PERFORMANCE */ device_printf(dpt->dev, "Aborting pending requests. O/S should re-submit\n"); while ((dccbp = TAILQ_FIRST(&dpt->completed_ccbs)) != NULL) { struct scsi_xfer *xs = dccbp->xs; /* Not all transactions have xs structs */ if (xs != NULL) { /* Tell the kernel proper this did not complete well */ xs->error |= XS_SELTIMEOUT; xs->flags |= SCSI_ITSDONE; scsi_done(xs); } dpt_Qremove_submitted(dpt, dccbp); /* Remember, Callbacks are NOT in the standard queue */ if (dccbp->std_callback != NULL) { (dccbp->std_callback)(dpt, dccbp->eata_ccb.cp_channel, dccbp); } else { dpt_Qpush_free(dpt, dccbp); } } device_printf(dpt->dev, "reset done aborting all pending commands\n"); dpt->queue_status &= ~DPT_SUBMITTED_QUEUE_ACTIVE; } #endif /* DPT_RESET_HBA */ /* * Build a Command Block for target mode READ/WRITE BUFFER, * with the ``sync'' bit ON. * * Although the length and offset are 24 bit fields in the command, they cannot * exceed 8192 bytes, so we take them as short integers andcheck their range. * If they are sensless, we round them to zero offset, maximum length and * complain. */ static void dpt_target_ccb(dpt_softc_t * dpt, int bus, u_int8_t target, u_int8_t lun, dpt_ccb_t * ccb, int mode, u_int8_t command, u_int16_t length, u_int16_t offset) { eata_ccb_t *cp; mtx_assert(&dpt->lock, MA_OWNED); if ((length + offset) > DPT_MAX_TARGET_MODE_BUFFER_SIZE) { device_printf(dpt->dev, "Length of %d, and offset of %d are wrong\n", length, offset); length = DPT_MAX_TARGET_MODE_BUFFER_SIZE; offset = 0; } ccb->xs = NULL; ccb->flags = 0; ccb->state = DPT_CCB_STATE_NEW; ccb->std_callback = (ccb_callback) dpt_target_done; ccb->wrbuff_callback = NULL; cp = &ccb->eata_ccb; cp->CP_OpCode = EATA_CMD_DMA_SEND_CP; cp->SCSI_Reset = 0; cp->HBA_Init = 0; cp->Auto_Req_Sen = 1; cp->cp_id = target; cp->DataIn = 1; cp->DataOut = 0; cp->Interpret = 0; cp->reqlen = htonl(sizeof(struct scsi_sense_data)); cp->cp_statDMA = htonl(vtophys(&cp->cp_statDMA)); cp->cp_reqDMA = htonl(vtophys(&cp->cp_reqDMA)); cp->cp_viraddr = (u_int32_t) & ccb; cp->cp_msg[0] = HA_IDENTIFY_MSG | HA_DISCO_RECO; cp->cp_scsi_cmd = command; cp->cp_cdb[1] = (u_int8_t) (mode & SCSI_TM_MODE_MASK); cp->cp_lun = lun; /* Order is important here! */ cp->cp_cdb[2] = 0x00; /* Buffer Id, only 1 :-( */ cp->cp_cdb[3] = (length >> 16) & 0xFF; /* Buffer offset MSB */ cp->cp_cdb[4] = (length >> 8) & 0xFF; cp->cp_cdb[5] = length & 0xFF; cp->cp_cdb[6] = (length >> 16) & 0xFF; /* Length MSB */ cp->cp_cdb[7] = (length >> 8) & 0xFF; cp->cp_cdb[8] = length & 0xFF; /* Length LSB */ cp->cp_cdb[9] = 0; /* No sync, no match bits */ /* * This could be optimized to live in dpt_register_buffer. * We keep it here, just in case the kernel decides to reallocate pages */ if (dpt_scatter_gather(dpt, ccb, DPT_RW_BUFFER_SIZE, dpt->rw_buffer[bus][target][lun])) { device_printf(dpt->dev, "Failed to setup Scatter/Gather for " "Target-Mode buffer\n"); } } /* Setup a target mode READ command */ static void dpt_set_target(int redo, dpt_softc_t * dpt, u_int8_t bus, u_int8_t target, u_int8_t lun, int mode, u_int16_t length, u_int16_t offset, dpt_ccb_t * ccb) { mtx_assert(&dpt->lock, MA_OWNED); if (dpt->target_mode_enabled) { if (!redo) dpt_target_ccb(dpt, bus, target, lun, ccb, mode, SCSI_TM_READ_BUFFER, length, offset); ccb->transaction_id = ++dpt->commands_processed; #ifdef DPT_MEASURE_PERFORMANCE dpt->performance.command_count[ccb->eata_ccb.cp_scsi_cmd]++; ccb->command_started = microtime_now; #endif dpt_Qadd_waiting(dpt, ccb); dpt_sched_queue(dpt); } else { device_printf(dpt->dev, "Target Mode Request, but Target Mode is OFF\n"); } } /* * Schedule a buffer to be sent to another target. * The work will be scheduled and the callback provided will be called when * the work is actually done. * * Please NOTE: ``Anyone'' can send a buffer, but only registered clients * get notified of receipt of buffers. */ int dpt_send_buffer(int unit, u_int8_t channel, u_int8_t target, u_int8_t lun, u_int8_t mode, u_int16_t length, u_int16_t offset, void *data, buff_wr_done callback) { dpt_softc_t *dpt; dpt_ccb_t *ccb = NULL; /* This is an external call. Be a bit paranoid */ dpt = devclass_get_device(dpt_devclass, unit); if (dpt == NULL) return (INVALID_UNIT); mtx_lock(&dpt->lock); if (dpt->target_mode_enabled) { if ((channel >= dpt->channels) || (target > dpt->max_id) || (lun > dpt->max_lun)) { mtx_unlock(&dpt->lock); return (INVALID_SENDER); } if ((dpt->rw_buffer[channel][target][lun] == NULL) || (dpt->buffer_receiver[channel][target][lun] == NULL)) { mtx_unlock(&dpt->lock); return (NOT_REGISTERED); } /* Process the free list */ if ((TAILQ_EMPTY(&dpt->free_ccbs)) && dpt_alloc_freelist(dpt)) { device_printf(dpt->dev, "ERROR: Cannot allocate any more free CCB's.\n" " Please try later\n"); mtx_unlock(&dpt->lock); return (NO_RESOURCES); } /* Now grab the newest CCB */ if ((ccb = dpt_Qpop_free(dpt)) == NULL) { mtx_unlock(&dpt->lock); panic("%s: Got a NULL CCB from pop_free()\n", device_get_nameunit(dpt->dev)); } bcopy(dpt->rw_buffer[channel][target][lun] + offset, data, length); dpt_target_ccb(dpt, channel, target, lun, ccb, mode, SCSI_TM_WRITE_BUFFER, length, offset); ccb->std_callback = (ccb_callback) callback; /* Potential trouble */ ccb->transaction_id = ++dpt->commands_processed; #ifdef DPT_MEASURE_PERFORMANCE dpt->performance.command_count[ccb->eata_ccb.cp_scsi_cmd]++; ccb->command_started = microtime_now; #endif dpt_Qadd_waiting(dpt, ccb); dpt_sched_queue(dpt); mtx_unlock(&dpt->lock); return (0); } mtx_unlock(&dpt->lock); return (DRIVER_DOWN); } static void dpt_target_done(dpt_softc_t * dpt, int bus, dpt_ccb_t * ccb) { eata_ccb_t *cp; cp = &ccb->eata_ccb; /* * Remove the CCB from the waiting queue. * We do NOT put it back on the free, etc., queues as it is a special * ccb, owned by the dpt_softc of this unit. */ dpt_Qremove_completed(dpt, ccb); #define br_channel (ccb->eata_ccb.cp_channel) #define br_target (ccb->eata_ccb.cp_id) #define br_lun (ccb->eata_ccb.cp_LUN) #define br_index [br_channel][br_target][br_lun] #define read_buffer_callback (dpt->buffer_receiver br_index ) #define read_buffer (dpt->rw_buffer[br_channel][br_target][br_lun]) #define cb(offset) (ccb->eata_ccb.cp_cdb[offset]) #define br_offset ((cb(3) << 16) | (cb(4) << 8) | cb(5)) #define br_length ((cb(6) << 16) | (cb(7) << 8) | cb(8)) /* Different reasons for being here, you know... */ switch (ccb->eata_ccb.cp_scsi_cmd) { case SCSI_TM_READ_BUFFER: if (read_buffer_callback != NULL) { /* This is a buffer generated by a kernel process */ read_buffer_callback(device_get_unit(dpt->dev), br_channel, br_target, br_lun, read_buffer, br_offset, br_length); } else { /* * This is a buffer waited for by a user (sleeping) * command */ wakeup(ccb); } /* We ALWAYS re-issue the same command; args are don't-care */ dpt_set_target(1, 0, 0, 0, 0, 0, 0, 0, 0); break; case SCSI_TM_WRITE_BUFFER: (ccb->wrbuff_callback) (device_get_unit(dpt->dev), br_channel, br_target, br_offset, br_length, br_lun, ccb->status_packet.hba_stat); break; default: device_printf(dpt->dev, "%s is an unsupported command for target mode\n", scsi_cmd_name(ccb->eata_ccb.cp_scsi_cmd)); } dpt->target_ccb[br_channel][br_target][br_lun] = NULL; dpt_Qpush_free(dpt, ccb); } /* * Use this function to register a client for a buffer read target operation. * The function you register will be called every time a buffer is received * by the target mode code. */ dpt_rb_t dpt_register_buffer(int unit, u_int8_t channel, u_int8_t target, u_int8_t lun, u_int8_t mode, u_int16_t length, u_int16_t offset, dpt_rec_buff callback, dpt_rb_op_t op) { dpt_softc_t *dpt; dpt_ccb_t *ccb = NULL; int ospl; dpt = devclass_get_device(dpt_devclass, unit); if (dpt == NULL) return (INVALID_UNIT); mtx_lock(&dpt->lock); if (dpt->state & DPT_HA_SHUTDOWN_ACTIVE) { mtx_unlock(&dpt->lock); return (DRIVER_DOWN); } if ((channel > (dpt->channels - 1)) || (target > (dpt->max_id - 1)) || (lun > (dpt->max_lun - 1))) { mtx_unlock(&dpt->lock); return (INVALID_SENDER); } if (dpt->buffer_receiver[channel][target][lun] == NULL) { if (op == REGISTER_BUFFER) { /* Assign the requested callback */ dpt->buffer_receiver[channel][target][lun] = callback; /* Get a CCB */ /* Process the free list */ if ((TAILQ_EMPTY(&dpt->free_ccbs)) && dpt_alloc_freelist(dpt)) { device_printf(dpt->dev, "ERROR: Cannot allocate any more free CCB's.\n" " Please try later\n"); mtx_unlock(&dpt->lock); return (NO_RESOURCES); } /* Now grab the newest CCB */ if ((ccb = dpt_Qpop_free(dpt)) == NULL) { mtx_unlock(&dpt->lock); panic("%s: Got a NULL CCB from pop_free()\n", device_get_nameunit(dpt->dev)); } /* Clean up the leftover of the previous tenant */ ccb->status = DPT_CCB_STATE_NEW; dpt->target_ccb[channel][target][lun] = ccb; dpt->rw_buffer[channel][target][lun] = malloc(DPT_RW_BUFFER_SIZE, M_DEVBUF, M_NOWAIT); if (dpt->rw_buffer[channel][target][lun] == NULL) { device_printf(dpt->dev, "Failed to allocate " "Target-Mode buffer\n"); dpt_Qpush_free(dpt, ccb); mtx_unlock(&dpt->lock); return (NO_RESOURCES); } dpt_set_target(0, dpt, channel, target, lun, mode, length, offset, ccb); mtx_unlock(&dpt->lock); return (SUCCESSFULLY_REGISTERED); } else { mtx_unlock(&dpt->lock); return (NOT_REGISTERED); } } else { if (op == REGISTER_BUFFER) { if (dpt->buffer_receiver[channel][target][lun] == callback) { mtx_unlock(&dpt->lock); return (ALREADY_REGISTERED); } else { mtx_unlock(&dpt->lock); return (REGISTERED_TO_ANOTHER); } } else { if (dpt->buffer_receiver[channel][target][lun] == callback) { dpt->buffer_receiver[channel][target][lun] = NULL; dpt_Qpush_free(dpt, ccb); free(dpt->rw_buffer[channel][target][lun], M_DEVBUF); mtx_unlock(&dpt->lock); return (SUCCESSFULLY_REGISTERED); } else { mtx_unlock(&dpt->lock); return (INVALID_CALLBACK); } } } mtx_unlock(&dpt->lock); } /* Return the state of the blinking DPT LED's */ u_int8_t dpt_blinking_led(dpt_softc_t * dpt) { int ndx; u_int32_t state; u_int32_t previous; u_int8_t result; mtx_assert(&dpt->lock, MA_OWNED); result = 0; for (ndx = 0, state = 0, previous = 0; (ndx < 10) && (state != previous); ndx++) { previous = state; state = dpt_inl(dpt, 1); } if ((state == previous) && (state == DPT_BLINK_INDICATOR)) result = dpt_inb(dpt, 5); return (result); } /* * Execute a command which did not come from the kernel's SCSI layer. * The only way to map user commands to bus and target is to comply with the * standard DPT wire-down scheme: */ int dpt_user_cmd(dpt_softc_t * dpt, eata_pt_t * user_cmd, caddr_t cmdarg, int minor_no) { dpt_ccb_t *ccb; void *data; int channel, target, lun; int huh; int result; int submitted; mtx_assert(&dpt->lock, MA_OWNED); data = NULL; channel = minor2hba(minor_no); target = minor2target(minor_no); lun = minor2lun(minor_no); if ((channel > (dpt->channels - 1)) || (target > dpt->max_id) || (lun > dpt->max_lun)) return (ENXIO); if (target == dpt->sc_scsi_link[channel].adapter_targ) { /* This one is for the controller itself */ if ((user_cmd->eataID[0] != 'E') || (user_cmd->eataID[1] != 'A') || (user_cmd->eataID[2] != 'T') || (user_cmd->eataID[3] != 'A')) { return (ENXIO); } } /* Get a DPT CCB, so we can prepare a command */ /* Process the free list */ if ((TAILQ_EMPTY(&dpt->free_ccbs)) && dpt_alloc_freelist(dpt)) { device_printf(dpt->dev, "ERROR: Cannot allocate any more free CCB's.\n" " Please try later\n"); return (EFAULT); } /* Now grab the newest CCB */ if ((ccb = dpt_Qpop_free(dpt)) == NULL) { panic("%s: Got a NULL CCB from pop_free()\n", device_get_nameunit(dpt->dev)); } else { /* Clean up the leftover of the previous tenant */ ccb->status = DPT_CCB_STATE_NEW; } bcopy((caddr_t) & user_cmd->command_packet, (caddr_t) & ccb->eata_ccb, sizeof(eata_ccb_t)); /* We do not want to do user specified scatter/gather. Why?? */ if (ccb->eata_ccb.scatter == 1) return (EINVAL); ccb->eata_ccb.Auto_Req_Sen = 1; ccb->eata_ccb.reqlen = htonl(sizeof(struct scsi_sense_data)); ccb->eata_ccb.cp_datalen = htonl(sizeof(ccb->eata_ccb.cp_datalen)); ccb->eata_ccb.cp_dataDMA = htonl(vtophys(ccb->eata_ccb.cp_dataDMA)); ccb->eata_ccb.cp_statDMA = htonl(vtophys(&ccb->eata_ccb.cp_statDMA)); ccb->eata_ccb.cp_reqDMA = htonl(vtophys(&ccb->eata_ccb.cp_reqDMA)); ccb->eata_ccb.cp_viraddr = (u_int32_t) & ccb; if (ccb->eata_ccb.DataIn || ccb->eata_ccb.DataOut) { /* Data I/O is involved in this command. Alocate buffer */ if (ccb->eata_ccb.cp_datalen > PAGE_SIZE) { data = contigmalloc(ccb->eata_ccb.cp_datalen, M_TEMP, M_WAITOK, 0, ~0, ccb->eata_ccb.cp_datalen, 0x10000); } else { data = malloc(ccb->eata_ccb.cp_datalen, M_TEMP, M_WAITOK); } if (data == NULL) { device_printf(dpt->dev, "Cannot allocate %d bytes " "for EATA command\n", ccb->eata_ccb.cp_datalen); return (EFAULT); } #define usr_cmd_DMA (caddr_t)user_cmd->command_packet.cp_dataDMA if (ccb->eata_ccb.DataIn == 1) { if (copyin(usr_cmd_DMA, data, ccb->eata_ccb.cp_datalen) == -1) return (EFAULT); } } else { /* No data I/O involved here. Make sure the DPT knows that */ ccb->eata_ccb.cp_datalen = 0; data = NULL; } if (ccb->eata_ccb.FWNEST == 1) ccb->eata_ccb.FWNEST = 0; if (ccb->eata_ccb.cp_datalen != 0) { if (dpt_scatter_gather(dpt, ccb, ccb->eata_ccb.cp_datalen, data) != 0) { if (data != NULL) free(data, M_TEMP); return (EFAULT); } } /** * We are required to quiet a SCSI bus. * since we do not queue comands on a bus basis, * we wait for ALL commands on a controller to complete. * In the mean time, sched_queue() will not schedule new commands. */ if ((ccb->eata_ccb.cp_cdb[0] == MULTIFUNCTION_CMD) && (ccb->eata_ccb.cp_cdb[2] == BUS_QUIET)) { /* We wait for ALL traffic for this HBa to subside */ dpt->state |= DPT_HA_QUIET; while ((submitted = dpt->submitted_ccbs_count) != 0) { huh = mtx_sleep((void *) dpt, &dpt->lock, PCATCH | PRIBIO, "dptqt", 100 * hz); switch (huh) { case 0: /* Wakeup call received */ break; case EWOULDBLOCK: /* Timer Expired */ break; default: /* anything else */ break; } } } /* Resume normal operation */ if ((ccb->eata_ccb.cp_cdb[0] == MULTIFUNCTION_CMD) && (ccb->eata_ccb.cp_cdb[2] == BUS_UNQUIET)) { dpt->state &= ~DPT_HA_QUIET; } /** * Schedule the command and submit it. * We bypass dpt_sched_queue, as it will block on DPT_HA_QUIET */ ccb->xs = NULL; ccb->flags = 0; ccb->eata_ccb.Auto_Req_Sen = 1; /* We always want this feature */ ccb->transaction_id = ++dpt->commands_processed; ccb->std_callback = (ccb_callback) dpt_user_cmd_done; ccb->result = (u_int32_t) & cmdarg; ccb->data = data; #ifdef DPT_MEASURE_PERFORMANCE ++dpt->performance.command_count[ccb->eata_ccb.cp_scsi_cmd]; ccb->command_started = microtime_now; #endif dpt_Qadd_waiting(dpt, ccb); dpt_sched_queue(dpt); /* Wait for the command to complete */ (void) mtx_sleep((void *) ccb, &dpt->lock, PCATCH | PRIBIO, "dptucw", 100 * hz); /* Free allocated memory */ if (data != NULL) free(data, M_TEMP); return (0); } static void dpt_user_cmd_done(dpt_softc_t * dpt, int bus, dpt_ccb_t * ccb) { u_int32_t result; caddr_t cmd_arg; mtx_unlock(&dpt->lock); /** * If Auto Request Sense is on, copyout the sense struct */ #define usr_pckt_DMA (caddr_t)(intptr_t)ntohl(ccb->eata_ccb.cp_reqDMA) #define usr_pckt_len ntohl(ccb->eata_ccb.cp_datalen) if (ccb->eata_ccb.Auto_Req_Sen == 1) { if (copyout((caddr_t) & ccb->sense_data, usr_pckt_DMA, sizeof(struct scsi_sense_data))) { mtx_lock(&dpt->lock); ccb->result = EFAULT; dpt_Qpush_free(dpt, ccb); wakeup(ccb); return; } } /* If DataIn is on, copyout the data */ if ((ccb->eata_ccb.DataIn == 1) && (ccb->status_packet.hba_stat == HA_NO_ERROR)) { if (copyout(ccb->data, usr_pckt_DMA, usr_pckt_len)) { mtx_lock(&dpt->lock); dpt_Qpush_free(dpt, ccb); ccb->result = EFAULT; wakeup(ccb); return; } } /* Copyout the status */ result = ccb->status_packet.hba_stat; cmd_arg = (caddr_t) ccb->result; if (copyout((caddr_t) & result, cmd_arg, sizeof(result))) { mtx_lock(&dpt->lock); dpt_Qpush_free(dpt, ccb); ccb->result = EFAULT; wakeup(ccb); return; } mtx_lock(&dpt->lock); /* Put the CCB back in the freelist */ ccb->state |= DPT_CCB_STATE_COMPLETED; dpt_Qpush_free(dpt, ccb); /* Free allocated memory */ return; } #endif Index: projects/clang400-import/sys/dev/etherswitch/e6000sw/e6000sw.c =================================================================== --- projects/clang400-import/sys/dev/etherswitch/e6000sw/e6000sw.c (revision 311459) +++ projects/clang400-import/sys/dev/etherswitch/e6000sw/e6000sw.c (revision 311460) @@ -1,976 +1,1182 @@ /*- * Copyright (c) 2015 Semihalf * Copyright (c) 2015 Stormshield * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include +#include +#include +#include + #include "e6000swreg.h" #include "etherswitch_if.h" #include "miibus_if.h" #include "mdio_if.h" MALLOC_DECLARE(M_E6000SW); MALLOC_DEFINE(M_E6000SW, "e6000sw", "e6000sw switch"); #define E6000SW_LOCK(_sc) \ sx_xlock(&(_sc)->sx) #define E6000SW_UNLOCK(_sc) \ sx_unlock(&(_sc)->sx) #define E6000SW_LOCK_ASSERT(_sc, _what) \ sx_assert(&(_sc)->sx, (_what)) #define E6000SW_TRYLOCK(_sc) \ sx_tryxlock(&(_sc)->sx) typedef struct e6000sw_softc { device_t dev; + phandle_t node; struct sx sx; - struct ifnet *ifp[E6000SW_NUM_PHYS]; - char *ifname[E6000SW_NUM_PHYS]; - device_t miibus[E6000SW_NUM_PHYS]; - struct mii_data *mii[E6000SW_NUM_PHYS]; + struct ifnet *ifp[E6000SW_MAX_PORTS]; + char *ifname[E6000SW_MAX_PORTS]; + device_t miibus[E6000SW_MAX_PORTS]; + struct mii_data *mii[E6000SW_MAX_PORTS]; struct callout tick_callout; uint32_t cpuports_mask; + uint32_t fixed_mask; + int sw_addr; + int num_ports; + boolean_t multi_chip; int vid[E6000SW_NUM_VGROUPS]; int members[E6000SW_NUM_VGROUPS]; - int vgroup[E6000SW_NUM_PORTS]; + int vgroup[E6000SW_MAX_PORTS]; } e6000sw_softc_t; static etherswitch_info_t etherswitch_info = { - .es_nports = E6000SW_NUM_PORTS, + .es_nports = 0, .es_nvlangroups = E6000SW_NUM_VGROUPS, .es_name = "Marvell 6000 series switch" }; static void e6000sw_identify(driver_t *driver, device_t parent); static int e6000sw_probe(device_t dev); static int e6000sw_attach(device_t dev); static int e6000sw_detach(device_t dev); static int e6000sw_readphy(device_t dev, int phy, int reg); static int e6000sw_writephy(device_t dev, int phy, int reg, int data); static etherswitch_info_t* e6000sw_getinfo(device_t dev); static void e6000sw_lock(device_t dev); static void e6000sw_unlock(device_t dev); static int e6000sw_getport(device_t dev, etherswitch_port_t *p); static int e6000sw_setport(device_t dev, etherswitch_port_t *p); static int e6000sw_readreg_wrapper(device_t dev, int addr_reg); static int e6000sw_writereg_wrapper(device_t dev, int addr_reg, int val); static int e6000sw_readphy_wrapper(device_t dev, int phy, int reg); static int e6000sw_writephy_wrapper(device_t dev, int phy, int reg, int data); static int e6000sw_getvgroup_wrapper(device_t dev, etherswitch_vlangroup_t *vg); static int e6000sw_setvgroup_wrapper(device_t dev, etherswitch_vlangroup_t *vg); static int e6000sw_setvgroup(device_t dev, etherswitch_vlangroup_t *vg); static int e6000sw_getvgroup(device_t dev, etherswitch_vlangroup_t *vg); static void e6000sw_setup(device_t dev, e6000sw_softc_t *sc); static void e6000sw_port_vlan_conf(e6000sw_softc_t *sc); static void e6000sw_tick(void *arg); static void e6000sw_set_atustat(device_t dev, e6000sw_softc_t *sc, int bin, int flag); static int e6000sw_atu_flush(device_t dev, e6000sw_softc_t *sc, int flag); static __inline void e6000sw_writereg(e6000sw_softc_t *sc, int addr, int reg, int val); static __inline uint32_t e6000sw_readreg(e6000sw_softc_t *sc, int addr, int reg); static int e6000sw_ifmedia_upd(struct ifnet *ifp); static void e6000sw_ifmedia_sts(struct ifnet *ifp, struct ifmediareq *ifmr); static int e6000sw_atu_mac_table(device_t dev, e6000sw_softc_t *sc, struct atu_opt *atu, int flag); static int e6000sw_get_pvid(e6000sw_softc_t *sc, int port, int *pvid); static int e6000sw_set_pvid(e6000sw_softc_t *sc, int port, int pvid); -static __inline int e6000sw_cpuport(e6000sw_softc_t *sc, int port); +static __inline int e6000sw_is_cpuport(e6000sw_softc_t *sc, int port); +static __inline int e6000sw_is_fixedport(e6000sw_softc_t *sc, int port); +static __inline int e6000sw_is_phyport(e6000sw_softc_t *sc, int port); static __inline struct mii_data *e6000sw_miiforphy(e6000sw_softc_t *sc, unsigned int phy); static struct proc *e6000sw_kproc; static device_method_t e6000sw_methods[] = { /* device interface */ DEVMETHOD(device_identify, e6000sw_identify), DEVMETHOD(device_probe, e6000sw_probe), DEVMETHOD(device_attach, e6000sw_attach), DEVMETHOD(device_detach, e6000sw_detach), /* bus interface */ DEVMETHOD(bus_add_child, device_add_child_ordered), /* mii interface */ DEVMETHOD(miibus_readreg, e6000sw_readphy), DEVMETHOD(miibus_writereg, e6000sw_writephy), /* etherswitch interface */ DEVMETHOD(etherswitch_getinfo, e6000sw_getinfo), DEVMETHOD(etherswitch_lock, e6000sw_lock), DEVMETHOD(etherswitch_unlock, e6000sw_unlock), DEVMETHOD(etherswitch_getport, e6000sw_getport), DEVMETHOD(etherswitch_setport, e6000sw_setport), DEVMETHOD(etherswitch_readreg, e6000sw_readreg_wrapper), DEVMETHOD(etherswitch_writereg, e6000sw_writereg_wrapper), DEVMETHOD(etherswitch_readphyreg, e6000sw_readphy_wrapper), DEVMETHOD(etherswitch_writephyreg, e6000sw_writephy_wrapper), DEVMETHOD(etherswitch_setvgroup, e6000sw_setvgroup_wrapper), DEVMETHOD(etherswitch_getvgroup, e6000sw_getvgroup_wrapper), DEVMETHOD_END }; static devclass_t e6000sw_devclass; DEFINE_CLASS_0(e6000sw, e6000sw_driver, e6000sw_methods, sizeof(e6000sw_softc_t)); DRIVER_MODULE(e6000sw, mdio, e6000sw_driver, e6000sw_devclass, 0, 0); DRIVER_MODULE(etherswitch, e6000sw, etherswitch_driver, etherswitch_devclass, 0, 0); DRIVER_MODULE(miibus, e6000sw, miibus_driver, miibus_devclass, 0, 0); MODULE_DEPEND(e6000sw, mdio, 1, 1, 1); +#define SMI_CMD 0 +#define SMI_CMD_BUSY (1<<15) +#define SMI_CMD_OP_READ ((2<<10)|SMI_CMD_BUSY|(1<<12)) +#define SMI_CMD_OP_WRITE ((1<<10)|SMI_CMD_BUSY|(1<<12)) +#define SMI_DATA 1 + +#define MDIO_READ(dev, addr, reg) MDIO_READREG(device_get_parent(dev), (addr), (reg)) +#define MDIO_WRITE(dev, addr, reg, val) MDIO_WRITEREG(device_get_parent(dev), (addr), (reg), (val)) static void e6000sw_identify(driver_t *driver, device_t parent) { if (device_find_child(parent, "e6000sw", -1) == NULL) BUS_ADD_CHILD(parent, 0, "e6000sw", -1); } static int e6000sw_probe(device_t dev) { e6000sw_softc_t *sc; const char *description; unsigned int id; + uint16_t dev_addr; + phandle_t dsa_node, switch_node; + dsa_node = fdt_find_compatible(OF_finddevice("/"), + "marvell,dsa", 0); + switch_node = OF_child(dsa_node); + + if (switch_node == 0) + return (ENXIO); + sc = device_get_softc(dev); bzero(sc, sizeof(e6000sw_softc_t)); sc->dev = dev; + sc->node = switch_node; + + /* Read ADDR[4:1]n using indirect access */ + MDIO_WRITE(dev, REG_GLOBAL2, SCR_AND_MISC_REG, + SCR_AND_MISC_PTR_CFG); + dev_addr = MDIO_READ(dev, REG_GLOBAL2, SCR_AND_MISC_REG) & + SCR_AND_MISC_DATA_CFG_MASK; + if (dev_addr != 0) { + sc->multi_chip = true; + device_printf(dev, "multi-chip addresing mode\n"); + } else { + device_printf(dev, "single-chip addressing mode\n"); + } + + if (OF_getencprop(sc->node, "reg", &sc->sw_addr, + sizeof(sc->sw_addr)) < 0) + return (ENXIO); + /* Lock is necessary due to assertions. */ sx_init(&sc->sx, "e6000sw"); E6000SW_LOCK(sc); id = e6000sw_readreg(sc, REG_PORT(0), SWITCH_ID); switch (id & 0xfff0) { case 0x3520: description = "Marvell 88E6352"; break; case 0x1720: description = "Marvell 88E6172"; break; case 0x1760: description = "Marvell 88E6176"; break; default: E6000SW_UNLOCK(sc); sx_destroy(&sc->sx); - device_printf(dev, "Unrecognized device.\n"); + device_printf(dev, "Unrecognized device, id 0x%x.\n", id); return (ENXIO); } device_set_desc(dev, description); E6000SW_UNLOCK(sc); return (BUS_PROBE_DEFAULT); } static int +e6000sw_parse_child_fdt(device_t dev, phandle_t child, uint32_t *fixed_mask, + uint32_t *cpu_mask, int *pport, int *pvlangroup) +{ + char portlabel[100]; + uint32_t port, vlangroup; + boolean_t fixed_link; + + if (fixed_mask == NULL || cpu_mask == NULL || pport == NULL) + return (ENXIO); + + OF_getprop(child, "label", (void *)portlabel, 100); + OF_getencprop(child, "reg", (void *)&port, sizeof(port)); + + if (OF_getencprop(child, "vlangroup", (void *)&vlangroup, + sizeof(vlangroup)) > 0) { + if (vlangroup >= E6000SW_NUM_VGROUPS) + return (ENXIO); + *pvlangroup = vlangroup; + } else { + *pvlangroup = -1; + } + + if (port >= E6000SW_MAX_PORTS) + return (ENXIO); + *pport = port; + + if (strncmp(portlabel, "cpu", 3) == 0) { + device_printf(dev, "CPU port at %d\n", port); + *cpu_mask |= (1 << port); + return (0); + } + + fixed_link = OF_child(child); + if (fixed_link) { + *fixed_mask |= (1 << port); + device_printf(dev, "fixed port at %d\n", port); + } else { + device_printf(dev, "PHY at %d\n", port); + } + + return (0); +} + +static int +e6000sw_init_interface(e6000sw_softc_t *sc, int port) +{ + char name[IFNAMSIZ]; + + snprintf(name, IFNAMSIZ, "%sport", device_get_nameunit(sc->dev)); + + sc->ifp[port] = if_alloc(IFT_ETHER); + if (sc->ifp[port] == NULL) + return (ENOMEM); + sc->ifp[port]->if_softc = sc; + sc->ifp[port]->if_flags |= IFF_UP | IFF_BROADCAST | + IFF_DRV_RUNNING | IFF_SIMPLEX; + sc->ifname[port] = malloc(strlen(name) + 1, M_E6000SW, M_WAITOK); + if (sc->ifname[port] == NULL) + return (ENOMEM); + memcpy(sc->ifname[port], name, strlen(name) + 1); + if_initname(sc->ifp[port], sc->ifname[port], port); + + return (0); +} + +static int +e6000sw_attach_miibus(e6000sw_softc_t *sc, int port) +{ + int err; + + err = mii_attach(sc->dev, &sc->miibus[port], sc->ifp[port], + e6000sw_ifmedia_upd, e6000sw_ifmedia_sts, BMSR_DEFCAPMASK, + port, MII_OFFSET_ANY, 0); + if (err != 0) + return (err); + + sc->mii[port] = device_get_softc(sc->miibus[port]); + return (0); +} + +static int e6000sw_attach(device_t dev) { e6000sw_softc_t *sc; - int phy, err, port; - char name[IFNAMSIZ]; + phandle_t child; + int err, port, vlangroup; + int member_ports[E6000SW_NUM_VGROUPS]; + etherswitch_vlangroup_t vg; err = 0; sc = device_get_softc(dev); + E6000SW_LOCK(sc); - sc->cpuports_mask = E6000SW_CPUPORTS_MASK; - for (port = 0; port < E6000SW_NUM_PORTS; port++) - sc->vgroup[port] = E6000SW_PORT_NO_VGROUP; e6000sw_setup(dev, sc); + bzero(member_ports, sizeof(member_ports)); - snprintf(name, IFNAMSIZ, "%sport", device_get_nameunit(sc->dev)); - for (phy = 0; phy < E6000SW_NUM_PHYS; phy++) { - sc->ifp[phy] = if_alloc(IFT_ETHER); - if (sc->ifp[phy] == NULL) + for (child = OF_child(sc->node); child != 0; child = OF_peer(child)) { + err = e6000sw_parse_child_fdt(dev, child, &sc->fixed_mask, + &sc->cpuports_mask, &port, &vlangroup); + if (err != 0) { + device_printf(sc->dev, "failed to parse DTS\n"); goto out_fail; - sc->ifp[phy]->if_softc = sc; - sc->ifp[phy]->if_flags |= IFF_UP | IFF_BROADCAST | - IFF_DRV_RUNNING | IFF_SIMPLEX; - sc->ifname[phy] = malloc(strlen(name) + 1, M_E6000SW, M_WAITOK); - if (sc->ifname[phy] == NULL) + } + + if (vlangroup != -1) + member_ports[vlangroup] |= (1 << port); + + sc->num_ports++; + + err = e6000sw_init_interface(sc, port); + if (err != 0) { + device_printf(sc->dev, "failed to init interface\n"); goto out_fail; - bcopy(name, sc->ifname[phy], strlen(name) + 1); - if_initname(sc->ifp[phy], sc->ifname[phy], phy); - err = mii_attach(sc->dev, &sc->miibus[phy], sc->ifp[phy], - e6000sw_ifmedia_upd, e6000sw_ifmedia_sts, BMSR_DEFCAPMASK, - phy, MII_OFFSET_ANY, 0); + } + + /* Don't attach miibus at CPU/fixed ports */ + if (!e6000sw_is_phyport(sc, port)) + continue; + + err = e6000sw_attach_miibus(sc, port); if (err != 0) { - device_printf(sc->dev, - "attaching PHY %d failed\n", - phy); + device_printf(sc->dev, "failed to attach miibus\n"); goto out_fail; } - sc->mii[phy] = device_get_softc(sc->miibus[phy]); } + + etherswitch_info.es_nports = sc->num_ports; + for (port = 0; port < sc->num_ports; port++) + sc->vgroup[port] = E6000SW_PORT_NO_VGROUP; + + /* Set VLAN configuration */ + e6000sw_port_vlan_conf(sc); + + /* Set vlangroups */ + for (vlangroup = 0; vlangroup < E6000SW_NUM_VGROUPS; vlangroup++) + if (member_ports[vlangroup] != 0) { + vg.es_vlangroup = vg.es_vid = vlangroup; + vg.es_member_ports = vg.es_untagged_ports = + member_ports[vlangroup]; + e6000sw_setvgroup(dev, &vg); + } + E6000SW_UNLOCK(sc); bus_generic_probe(dev); - bus_enumerate_hinted_children(dev); bus_generic_attach(dev); kproc_create(e6000sw_tick, sc, &e6000sw_kproc, 0, 0, "e6000sw tick kproc"); return (0); out_fail: e6000sw_detach(dev); - return (ENXIO); + return (err); } static __inline void e6000sw_poll_done(e6000sw_softc_t *sc) { while (e6000sw_readreg(sc, REG_GLOBAL2, PHY_CMD) & (1 << PHY_CMD_SMI_BUSY)) continue; } - /* * PHY registers are paged. Put page index in reg 22 (accessible from every * page), then access specific register. */ static int e6000sw_readphy(device_t dev, int phy, int reg) { e6000sw_softc_t *sc; uint32_t val; sc = device_get_softc(dev); val = 0; - if (phy >= E6000SW_NUM_PHYS || reg >= E6000SW_NUM_PHY_REGS) { + if (!e6000sw_is_phyport(sc, phy) || reg >= E6000SW_NUM_PHY_REGS) { device_printf(dev, "Wrong register address.\n"); return (EINVAL); } E6000SW_LOCK_ASSERT(sc, SA_XLOCKED); e6000sw_poll_done(sc); val |= 1 << PHY_CMD_SMI_BUSY; val |= PHY_CMD_MODE_MDIO << PHY_CMD_MODE; val |= PHY_CMD_OPCODE_READ << PHY_CMD_OPCODE; val |= (reg << PHY_CMD_REG_ADDR) & PHY_CMD_REG_ADDR_MASK; val |= (phy << PHY_CMD_DEV_ADDR) & PHY_CMD_DEV_ADDR_MASK; e6000sw_writereg(sc, REG_GLOBAL2, SMI_PHY_CMD_REG, val); e6000sw_poll_done(sc); val = e6000sw_readreg(sc, REG_GLOBAL2, SMI_PHY_DATA_REG) & PHY_DATA_MASK; return (val); } static int e6000sw_writephy(device_t dev, int phy, int reg, int data) { e6000sw_softc_t *sc; uint32_t val; sc = device_get_softc(dev); val = 0; - if (phy >= E6000SW_NUM_PHYS || reg >= E6000SW_NUM_PHY_REGS) { + if (!e6000sw_is_phyport(sc, phy) || reg >= E6000SW_NUM_PHY_REGS) { device_printf(dev, "Wrong register address.\n"); return (EINVAL); } E6000SW_LOCK_ASSERT(sc, SA_XLOCKED); e6000sw_poll_done(sc); val |= PHY_CMD_MODE_MDIO << PHY_CMD_MODE; val |= 1 << PHY_CMD_SMI_BUSY; val |= PHY_CMD_OPCODE_WRITE << PHY_CMD_OPCODE; val |= (reg << PHY_CMD_REG_ADDR) & PHY_CMD_REG_ADDR_MASK; val |= (phy << PHY_CMD_DEV_ADDR) & PHY_CMD_DEV_ADDR_MASK; e6000sw_writereg(sc, REG_GLOBAL2, SMI_PHY_DATA_REG, data & PHY_DATA_MASK); e6000sw_writereg(sc, REG_GLOBAL2, SMI_PHY_CMD_REG, val); e6000sw_poll_done(sc); return (0); } static int e6000sw_detach(device_t dev) { int phy; e6000sw_softc_t *sc; sc = device_get_softc(dev); bus_generic_detach(dev); sx_destroy(&sc->sx); - for (phy = 0; phy < E6000SW_NUM_PHYS; phy++) { + for (phy = 0; phy < sc->num_ports; phy++) { if (sc->miibus[phy] != NULL) device_delete_child(dev, sc->miibus[phy]); if (sc->ifp[phy] != NULL) if_free(sc->ifp[phy]); if (sc->ifname[phy] != NULL) free(sc->ifname[phy], M_E6000SW); } return (0); } static etherswitch_info_t* e6000sw_getinfo(device_t dev) { return (ðerswitch_info); } static void e6000sw_lock(device_t dev) { struct e6000sw_softc *sc; sc = device_get_softc(dev); E6000SW_LOCK_ASSERT(sc, SA_UNLOCKED); E6000SW_LOCK(sc); } static void e6000sw_unlock(device_t dev) { struct e6000sw_softc *sc; sc = device_get_softc(dev); E6000SW_LOCK_ASSERT(sc, SA_XLOCKED); E6000SW_UNLOCK(sc); } static int e6000sw_getport(device_t dev, etherswitch_port_t *p) { struct mii_data *mii; int err; struct ifmediareq *ifmr; err = 0; e6000sw_softc_t *sc = device_get_softc(dev); E6000SW_LOCK_ASSERT(sc, SA_UNLOCKED); E6000SW_LOCK(sc); - if (p->es_port >= E6000SW_NUM_PORTS || + if (p->es_port >= sc->num_ports || p->es_port < 0) { err = EINVAL; goto out; } e6000sw_get_pvid(sc, p->es_port, &p->es_pvid); - if (e6000sw_cpuport(sc, p->es_port)) { + if (e6000sw_is_cpuport(sc, p->es_port)) { p->es_flags |= ETHERSWITCH_PORT_CPU; ifmr = &p->es_ifmr; ifmr->ifm_status = IFM_ACTIVE | IFM_AVALID; ifmr->ifm_count = 0; ifmr->ifm_current = ifmr->ifm_active = IFM_ETHER | IFM_1000_T | IFM_FDX; ifmr->ifm_mask = 0; + } else if (e6000sw_is_fixedport(sc, p->es_port)) { + ifmr = &p->es_ifmr; + ifmr->ifm_status = IFM_ACTIVE | IFM_AVALID; + ifmr->ifm_count = 0; + ifmr->ifm_current = ifmr->ifm_active = + IFM_ETHER | IFM_1000_T | IFM_FDX; + ifmr->ifm_mask = 0; } else { mii = e6000sw_miiforphy(sc, p->es_port); err = ifmedia_ioctl(mii->mii_ifp, &p->es_ifr, &mii->mii_media, SIOCGIFMEDIA); } out: E6000SW_UNLOCK(sc); return (err); } static int e6000sw_setport(device_t dev, etherswitch_port_t *p) { e6000sw_softc_t *sc; int err; struct mii_data *mii; err = 0; sc = device_get_softc(dev); E6000SW_LOCK_ASSERT(sc, SA_UNLOCKED); E6000SW_LOCK(sc); - if (p->es_port >= E6000SW_NUM_PORTS || + if (p->es_port >= sc->num_ports || p->es_port < 0) { err = EINVAL; goto out; } if (p->es_pvid != 0) e6000sw_set_pvid(sc, p->es_port, p->es_pvid); - if (!e6000sw_cpuport(sc, p->es_port)) { + if (!e6000sw_is_cpuport(sc, p->es_port)) { mii = e6000sw_miiforphy(sc, p->es_port); err = ifmedia_ioctl(mii->mii_ifp, &p->es_ifr, &mii->mii_media, SIOCSIFMEDIA); } out: E6000SW_UNLOCK(sc); return (err); } /* * Registers in this switch are divided into sections, specified in * documentation. So as to access any of them, section index and reg index * is necessary. etherswitchcfg uses only one variable, so indexes were * compressed into addr_reg: 32 * section_index + reg_index. */ static int e6000sw_readreg_wrapper(device_t dev, int addr_reg) { if ((addr_reg > (REG_GLOBAL2 * 32 + REG_NUM_MAX)) || (addr_reg < (REG_PORT(0) * 32))) { device_printf(dev, "Wrong register address.\n"); return (EINVAL); } return (e6000sw_readreg(device_get_softc(dev), addr_reg / 32, addr_reg % 32)); } static int e6000sw_writereg_wrapper(device_t dev, int addr_reg, int val) { if ((addr_reg > (REG_GLOBAL2 * 32 + REG_NUM_MAX)) || (addr_reg < (REG_PORT(0) * 32))) { device_printf(dev, "Wrong register address.\n"); return (EINVAL); } e6000sw_writereg(device_get_softc(dev), addr_reg / 5, addr_reg % 32, val); return (0); } /* * These wrappers are necessary because PHY accesses from etherswitchcfg * need to be synchronized with locks, while miibus PHY accesses do not. */ static int e6000sw_readphy_wrapper(device_t dev, int phy, int reg) { e6000sw_softc_t *sc; int ret; sc = device_get_softc(dev); E6000SW_LOCK_ASSERT(sc, SA_UNLOCKED); E6000SW_LOCK(sc); ret = e6000sw_readphy(dev, phy, reg); E6000SW_UNLOCK(sc); return (ret); } static int e6000sw_writephy_wrapper(device_t dev, int phy, int reg, int data) { e6000sw_softc_t *sc; int ret; sc = device_get_softc(dev); E6000SW_LOCK_ASSERT(sc, SA_UNLOCKED); E6000SW_LOCK(sc); ret = e6000sw_writephy(dev, phy, reg, data); E6000SW_UNLOCK(sc); return (ret); } /* * setvgroup/getvgroup called from etherswitchfcg need to be locked, * while internal calls do not. */ static int e6000sw_setvgroup_wrapper(device_t dev, etherswitch_vlangroup_t *vg) { e6000sw_softc_t *sc; int ret; sc = device_get_softc(dev); E6000SW_LOCK_ASSERT(sc, SA_UNLOCKED); E6000SW_LOCK(sc); ret = e6000sw_setvgroup(dev, vg); E6000SW_UNLOCK(sc); return (ret); } static int e6000sw_getvgroup_wrapper(device_t dev, etherswitch_vlangroup_t *vg) { e6000sw_softc_t *sc; int ret; sc = device_get_softc(dev); E6000SW_LOCK_ASSERT(sc, SA_UNLOCKED); E6000SW_LOCK(sc); ret = e6000sw_getvgroup(dev, vg); E6000SW_UNLOCK(sc); return (ret); } static __inline void e6000sw_flush_port(e6000sw_softc_t *sc, int port) { uint32_t reg; reg = e6000sw_readreg(sc, REG_PORT(port), PORT_VLAN_MAP); reg &= ~PORT_VLAN_MAP_TABLE_MASK; reg &= ~PORT_VLAN_MAP_FID_MASK; e6000sw_writereg(sc, REG_PORT(port), PORT_VLAN_MAP, reg); if (sc->vgroup[port] != E6000SW_PORT_NO_VGROUP) { /* * If port belonged somewhere, owner-group * should have its entry removed. */ sc->members[sc->vgroup[port]] &= ~(1 << port); sc->vgroup[port] = E6000SW_PORT_NO_VGROUP; } } static __inline void e6000sw_port_assign_vgroup(e6000sw_softc_t *sc, int port, int fid, int vgroup, int members) { uint32_t reg; reg = e6000sw_readreg(sc, REG_PORT(port), PORT_VLAN_MAP); reg &= ~PORT_VLAN_MAP_TABLE_MASK; reg &= ~PORT_VLAN_MAP_FID_MASK; reg |= members & ~(1 << port); reg |= (fid << PORT_VLAN_MAP_FID) & PORT_VLAN_MAP_FID_MASK; e6000sw_writereg(sc, REG_PORT(port), PORT_VLAN_MAP, reg); sc->vgroup[port] = vgroup; } static int e6000sw_setvgroup(device_t dev, etherswitch_vlangroup_t *vg) { e6000sw_softc_t *sc; int port, fid; sc = device_get_softc(dev); E6000SW_LOCK_ASSERT(sc, SA_XLOCKED); if (vg->es_vlangroup >= E6000SW_NUM_VGROUPS) return (EINVAL); if (vg->es_member_ports != vg->es_untagged_ports) { device_printf(dev, "Tagged ports not supported.\n"); return (EINVAL); } vg->es_untagged_ports &= PORT_VLAN_MAP_TABLE_MASK; fid = vg->es_vlangroup + 1; - for (port = 0; port < E6000SW_NUM_PORTS; port++) { + for (port = 0; port < sc->num_ports; port++) { if ((sc->members[vg->es_vlangroup] & (1 << port)) || (vg->es_untagged_ports & (1 << port))) e6000sw_flush_port(sc, port); if (vg->es_untagged_ports & (1 << port)) e6000sw_port_assign_vgroup(sc, port, fid, vg->es_vlangroup, vg->es_untagged_ports); } sc->vid[vg->es_vlangroup] = vg->es_vid; sc->members[vg->es_vlangroup] = vg->es_untagged_ports; return (0); } static int e6000sw_getvgroup(device_t dev, etherswitch_vlangroup_t *vg) { e6000sw_softc_t *sc; sc = device_get_softc(dev); E6000SW_LOCK_ASSERT(sc, SA_XLOCKED); if (vg->es_vlangroup >= E6000SW_NUM_VGROUPS) return (EINVAL); vg->es_untagged_ports = vg->es_member_ports = sc->members[vg->es_vlangroup]; vg->es_vid = ETHERSWITCH_VID_VALID; return (0); } static __inline struct mii_data* e6000sw_miiforphy(e6000sw_softc_t *sc, unsigned int phy) { - if (phy >= E6000SW_NUM_PHYS) + if (!e6000sw_is_phyport(sc, phy)) return (NULL); return (device_get_softc(sc->miibus[phy])); } static int e6000sw_ifmedia_upd(struct ifnet *ifp) { e6000sw_softc_t *sc; struct mii_data *mii; sc = ifp->if_softc; mii = e6000sw_miiforphy(sc, ifp->if_dunit); if (mii == NULL) return (ENXIO); mii_mediachg(mii); return (0); } static void e6000sw_ifmedia_sts(struct ifnet *ifp, struct ifmediareq *ifmr) { e6000sw_softc_t *sc; struct mii_data *mii; sc = ifp->if_softc; mii = e6000sw_miiforphy(sc, ifp->if_dunit); if (mii == NULL) return; mii_pollstat(mii); ifmr->ifm_active = mii->mii_media_active; ifmr->ifm_status = mii->mii_media_status; } + +static int +e6000sw_smi_waitready(e6000sw_softc_t *sc, int phy) +{ + int i; + + for (i = 0; i < E6000SW_SMI_TIMEOUT; i++) { + if ((MDIO_READ(sc->dev, phy, SMI_CMD) + & SMI_CMD_BUSY) == 0) + return 0; + } + + return 1; +} + static __inline uint32_t e6000sw_readreg(e6000sw_softc_t *sc, int addr, int reg) { E6000SW_LOCK_ASSERT(sc, SA_XLOCKED); - return (MDIO_READREG(device_get_parent(sc->dev), addr, reg)); + if (!sc->multi_chip) + return (MDIO_READ(sc->dev, addr, reg) & 0xffff); + + if (e6000sw_smi_waitready(sc, sc->sw_addr)) { + printf("e6000sw: readreg timeout\n"); + return (0xffff); + } + MDIO_WRITE(sc->dev, sc->sw_addr, SMI_CMD, SMI_CMD_OP_READ | + (addr << 5) | reg); + if (e6000sw_smi_waitready(sc, sc->sw_addr)) { + printf("e6000sw: readreg timeout\n"); + return (0xffff); + } + + return (MDIO_READ(sc->dev, sc->sw_addr, SMI_DATA) & 0xffff); } static __inline void e6000sw_writereg(e6000sw_softc_t *sc, int addr, int reg, int val) { E6000SW_LOCK_ASSERT(sc, SA_XLOCKED); - MDIO_WRITEREG(device_get_parent(sc->dev), addr, reg, val); + if (!sc->multi_chip) { + MDIO_WRITE(sc->dev, addr, reg, val); + return; + } + + if (e6000sw_smi_waitready(sc, sc->sw_addr)) { + printf("e6000sw: readreg timeout\n"); + return; + } + MDIO_WRITE(sc->dev, sc->sw_addr, SMI_DATA, val); + MDIO_WRITE(sc->dev, sc->sw_addr, SMI_CMD, SMI_CMD_OP_WRITE | + (addr << 5) | reg); + if (e6000sw_smi_waitready(sc, sc->sw_addr)) { + printf("e6000sw: readreg timeout\n"); + return; + } + + return; } static __inline int -e6000sw_cpuport(e6000sw_softc_t *sc, int port) +e6000sw_is_cpuport(e6000sw_softc_t *sc, int port) { return (sc->cpuports_mask & (1 << port)); } static __inline int +e6000sw_is_fixedport(e6000sw_softc_t *sc, int port) +{ + + return (sc->fixed_mask & (1 << port)); +} + +static __inline int +e6000sw_is_phyport(e6000sw_softc_t *sc, int port) +{ + uint32_t phy_mask; + phy_mask = ~(sc->fixed_mask | sc->cpuports_mask); + + return (phy_mask & (1 << port)); +} + +static __inline int e6000sw_set_pvid(e6000sw_softc_t *sc, int port, int pvid) { e6000sw_writereg(sc, REG_PORT(port), PORT_VID, pvid & PORT_VID_DEF_VID_MASK); return (0); } static __inline int e6000sw_get_pvid(e6000sw_softc_t *sc, int port, int *pvid) { if (pvid == NULL) return (ENXIO); *pvid = e6000sw_readreg(sc, REG_PORT(port), PORT_VID) & PORT_VID_DEF_VID_MASK; return (0); } static void e6000sw_tick (void *arg) { e6000sw_softc_t *sc; struct mii_softc *miisc; - int i; + int port; sc = arg; E6000SW_LOCK_ASSERT(sc, SA_UNLOCKED); for (;;) { E6000SW_LOCK(sc); - for (i = 0; i < E6000SW_NUM_PHYS; i++) { - mii_tick(sc->mii[i]); - LIST_FOREACH(miisc, &sc->mii[i]->mii_phys, mii_list) { - if (IFM_INST(sc->mii[i]->mii_media.ifm_cur->ifm_media) + for (port = 0; port < sc->num_ports; port++) { + /* Tick only on PHY ports */ + if (!e6000sw_is_phyport(sc, port)) + continue; + mii_tick(sc->mii[port]); + LIST_FOREACH(miisc, &sc->mii[port]->mii_phys, mii_list) { + if (IFM_INST(sc->mii[port]->mii_media.ifm_cur->ifm_media) != miisc->mii_inst) continue; mii_phy_update(miisc, MII_POLLSTAT); } } E6000SW_UNLOCK(sc); pause("e6000sw tick", 1000); } } static void e6000sw_setup(device_t dev, e6000sw_softc_t *sc) { uint16_t atu_ctrl, atu_age; /* Set aging time */ e6000sw_writereg(sc, REG_GLOBAL, ATU_CONTROL, (E6000SW_DEFAULT_AGETIME << ATU_CONTROL_AGETIME) | (1 << ATU_CONTROL_LEARN2ALL)); /* Send all with specific mac address to cpu port */ e6000sw_writereg(sc, REG_GLOBAL2, MGMT_EN_2x, MGMT_EN_ALL); e6000sw_writereg(sc, REG_GLOBAL2, MGMT_EN_0x, MGMT_EN_ALL); /* Disable Remote Management */ e6000sw_writereg(sc, REG_GLOBAL, SWITCH_GLOBAL_CONTROL2, 0); /* Disable loopback filter and flow control messages */ e6000sw_writereg(sc, REG_GLOBAL2, SWITCH_MGMT, SWITCH_MGMT_PRI_MASK | (1 << SWITCH_MGMT_RSVD2CPU) | SWITCH_MGMT_FC_PRI_MASK | (1 << SWITCH_MGMT_FORCEFLOW)); - /* Set VLAN configuration */ - e6000sw_port_vlan_conf(sc); - e6000sw_atu_flush(dev, sc, NO_OPERATION); e6000sw_atu_mac_table(dev, sc, NULL, NO_OPERATION); e6000sw_set_atustat(dev, sc, 0, COUNT_ALL); /* Set ATU AgeTime to 15 seconds */ atu_age = 1; atu_ctrl = e6000sw_readreg(sc, REG_GLOBAL, ATU_CONTROL); /* Set new AgeTime field */ atu_ctrl &= ~ATU_CONTROL_AGETIME_MASK; e6000sw_writereg(sc, REG_GLOBAL, ATU_CONTROL, atu_ctrl | (atu_age << ATU_CONTROL_AGETIME)); } static void e6000sw_port_vlan_conf(e6000sw_softc_t *sc) { int port, ret; - etherswitch_vlangroup_t vg; device_t dev; dev = sc->dev; /* Disable all ports */ - for (port = 0; port < E6000SW_NUM_PORTS; port++) { + for (port = 0; port < sc->num_ports; port++) { ret = e6000sw_readreg(sc, REG_PORT(port), PORT_CONTROL); e6000sw_writereg(sc, REG_PORT(port), PORT_CONTROL, (ret & ~PORT_CONTROL_ENABLE)); } /* Set port priority */ - for (port = 0; port < E6000SW_NUM_PORTS; port++) { + for (port = 0; port < sc->num_ports; port++) { ret = e6000sw_readreg(sc, REG_PORT(port), PORT_VID); ret &= ~PORT_VID_PRIORITY_MASK; e6000sw_writereg(sc, REG_PORT(port), PORT_VID, ret); } - vg.es_vlangroup = 0; - vg.es_vid = 0; - vg.es_member_ports = vg.es_untagged_ports = E6000SW_DEF_VLANGROUP0; - e6000sw_setvgroup(dev, &vg); - vg.es_vlangroup = 1; - vg.es_vid = 1; - vg.es_member_ports = vg.es_untagged_ports = E6000SW_DEF_VLANGROUP1; - e6000sw_setvgroup(dev, &vg); - - device_printf(dev, "Default vlangroups set.\n"); /* Set VID map */ - for (port = 0; port < E6000SW_NUM_PORTS; port++) { + for (port = 0; port < sc->num_ports; port++) { ret = e6000sw_readreg(sc, REG_PORT(port), PORT_VID); ret &= ~PORT_VID_DEF_VID_MASK; ret |= (port + 1); e6000sw_writereg(sc, REG_PORT(port), PORT_VID, ret); } /* Enable all ports */ - for (port = 0; port < E6000SW_NUM_PORTS; port++) { + for (port = 0; port < sc->num_ports; port++) { ret = e6000sw_readreg(sc, REG_PORT(port), PORT_CONTROL); e6000sw_writereg(sc, REG_PORT(port), PORT_CONTROL, (ret | PORT_CONTROL_ENABLE)); } } static void e6000sw_set_atustat(device_t dev, e6000sw_softc_t *sc, int bin, int flag) { uint16_t ret; ret = e6000sw_readreg(sc, REG_GLOBAL2, ATU_STATS); e6000sw_writereg(sc, REG_GLOBAL2, ATU_STATS, (bin << ATU_STATS_BIN ) | (flag << ATU_STATS_FLAG)); } static int e6000sw_atu_mac_table(device_t dev, e6000sw_softc_t *sc, struct atu_opt *atu, int flag) { uint16_t ret_opt; uint16_t ret_data; int retries; if (flag == NO_OPERATION) return (0); else if ((flag & (LOAD_FROM_FIB | PURGE_FROM_FIB | GET_NEXT_IN_FIB | GET_VIOLATION_DATA | CLEAR_VIOLATION_DATA)) == 0) { device_printf(dev, "Wrong Opcode for ATU operation\n"); return (EINVAL); } ret_opt = e6000sw_readreg(sc, REG_GLOBAL, ATU_OPERATION); if (ret_opt & ATU_UNIT_BUSY) { device_printf(dev, "ATU unit is busy, cannot access" "register\n"); return (EBUSY); } else { if(flag & LOAD_FROM_FIB) { ret_data = e6000sw_readreg(sc, REG_GLOBAL, ATU_DATA); e6000sw_writereg(sc, REG_GLOBAL2, ATU_DATA, (ret_data & ~ENTRY_STATE)); } e6000sw_writereg(sc, REG_GLOBAL, ATU_MAC_ADDR01, atu->mac_01); e6000sw_writereg(sc, REG_GLOBAL, ATU_MAC_ADDR23, atu->mac_23); e6000sw_writereg(sc, REG_GLOBAL, ATU_MAC_ADDR45, atu->mac_45); e6000sw_writereg(sc, REG_GLOBAL, ATU_FID, atu->fid); e6000sw_writereg(sc, REG_GLOBAL, ATU_OPERATION, (ret_opt | ATU_UNIT_BUSY | flag)); retries = E6000SW_RETRIES; while (--retries & (e6000sw_readreg(sc, REG_GLOBAL, ATU_OPERATION) & ATU_UNIT_BUSY)) DELAY(1); if (retries == 0) device_printf(dev, "Timeout while flushing\n"); else if (flag & GET_NEXT_IN_FIB) { atu->mac_01 = e6000sw_readreg(sc, REG_GLOBAL, ATU_MAC_ADDR01); atu->mac_23 = e6000sw_readreg(sc, REG_GLOBAL, ATU_MAC_ADDR23); atu->mac_45 = e6000sw_readreg(sc, REG_GLOBAL, ATU_MAC_ADDR45); } } return (0); } static int e6000sw_atu_flush(device_t dev, e6000sw_softc_t *sc, int flag) { uint16_t ret; int retries; if (flag == NO_OPERATION) return (0); ret = e6000sw_readreg(sc, REG_GLOBAL, ATU_OPERATION); if (ret & ATU_UNIT_BUSY) { device_printf(dev, "Atu unit is busy, cannot flush\n"); return (EBUSY); } else { e6000sw_writereg(sc, REG_GLOBAL, ATU_OPERATION, (ret | ATU_UNIT_BUSY | flag)); retries = E6000SW_RETRIES; while (--retries & (e6000sw_readreg(sc, REG_GLOBAL, ATU_OPERATION) & ATU_UNIT_BUSY)) DELAY(1); if (retries == 0) device_printf(dev, "Timeout while flushing\n"); } return (0); } Index: projects/clang400-import/sys/dev/etherswitch/e6000sw/e6000swreg.h =================================================================== --- projects/clang400-import/sys/dev/etherswitch/e6000sw/e6000swreg.h (revision 311459) +++ projects/clang400-import/sys/dev/etherswitch/e6000sw/e6000swreg.h (revision 311460) @@ -1,185 +1,185 @@ /*- * Copyright (c) 2015 Semihalf * Copyright (c) 2015 Stormshield * 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 _E6000SWREG_H_ #define _E6000SWREG_H_ struct atu_opt { uint16_t mac_01; uint16_t mac_23; uint16_t mac_45; uint16_t fid; }; /* * Definitions for the Marvell 88E6000 series Ethernet Switch. */ -#define CPU_PORT 0x5 - /* * Switch Registers */ #define REG_GLOBAL 0x1b #define REG_GLOBAL2 0x1c #define REG_PORT(p) (0x10 + (p)) #define REG_NUM_MAX 31 /* * Per-Port Switch Registers */ #define PORT_STATUS 0x0 #define PSC_CONTROL 0x1 #define SWITCH_ID 0x3 #define PORT_CONTROL 0x4 #define PORT_CONTROL_1 0x5 #define PORT_VLAN_MAP 0x6 #define PORT_VID 0x7 #define PORT_ASSOCIATION_VECTOR 0xb #define PORT_ATU_CTRL 0xc #define RX_COUNTER 0x12 #define TX_COUNTER 0x13 #define PORT_VID_DEF_VID 0 #define PORT_VID_DEF_VID_MASK 0xfff #define PORT_VID_PRIORITY_MASK 0xc00 #define PORT_CONTROL_ENABLE 0x3 /* PORT_VLAN fields */ #define PORT_VLAN_MAP_TABLE_MASK 0x7f #define PORT_VLAN_MAP_FID 12 #define PORT_VLAN_MAP_FID_MASK 0xf000 /* * Switch Global Register 1 accessed via REG_GLOBAL_ADDR */ #define SWITCH_GLOBAL_STATUS 0 #define SWITCH_GLOBAL_CONTROL 4 #define SWITCH_GLOBAL_CONTROL2 28 #define MONITOR_CONTROL 26 /* ATU operation */ #define ATU_FID 1 #define ATU_CONTROL 10 #define ATU_OPERATION 11 #define ATU_DATA 12 #define ATU_MAC_ADDR01 13 #define ATU_MAC_ADDR23 14 #define ATU_MAC_ADDR45 15 #define ATU_UNIT_BUSY (1 << 15) #define ENTRY_STATE 0xf /* ATU_CONTROL fields */ #define ATU_CONTROL_AGETIME 4 #define ATU_CONTROL_AGETIME_MASK 0xff0 #define ATU_CONTROL_LEARN2ALL 3 /* ATU opcode */ #define NO_OPERATION (0 << 0) #define FLUSH_ALL (1 << 0) #define FLUSH_NON_STATIC (1 << 1) #define LOAD_FROM_FIB (3 << 0) #define PURGE_FROM_FIB (3 << 0) #define GET_NEXT_IN_FIB (1 << 2) #define FLUSH_ALL_IN_FIB (5 << 0) #define FLUSH_NON_STATIC_IN_FIB (3 << 1) #define GET_VIOLATION_DATA (7 << 0) #define CLEAR_VIOLATION_DATA (7 << 0) /* ATU Stats */ #define COUNT_ALL (0 << 0) /* * Switch Global Register 2 accessed via REG_GLOBAL2_ADDR */ #define MGMT_EN_2x 2 #define MGMT_EN_0x 3 #define SWITCH_MGMT 5 #define ATU_STATS 14 #define MGMT_EN_ALL 0xffff /* SWITCH_MGMT fields */ #define SWITCH_MGMT_PRI 0 #define SWITCH_MGMT_PRI_MASK 7 #define SWITCH_MGMT_RSVD2CPU 3 #define SWITCH_MGMT_FC_PRI 4 #define SWITCH_MGMT_FC_PRI_MASK (7 << 4) #define SWITCH_MGMT_FORCEFLOW 7 /* ATU_STATS fields */ #define ATU_STATS_BIN 14 #define ATU_STATS_FLAG 12 /* * PHY registers accessed via 'Switch Global Registers' (REG_GLOBAL2). */ #define SMI_PHY_CMD_REG 0x18 #define SMI_PHY_DATA_REG 0x19 #define PHY_CMD 0x18 #define PHY_DATA 0x19 #define PHY_DATA_MASK 0xffff #define PHY_CMD_SMI_BUSY 15 #define PHY_CMD_MODE 12 #define PHY_CMD_MODE_MDIO 1 #define PHY_CMD_MODE_XMDIO 0 #define PHY_CMD_OPCODE 10 #define PHY_CMD_OPCODE_WRITE 1 #define PHY_CMD_OPCODE_READ 2 #define PHY_CMD_DEV_ADDR 5 #define PHY_CMD_DEV_ADDR_MASK 0x3e0 #define PHY_CMD_REG_ADDR 0 #define PHY_CMD_REG_ADDR_MASK 0x1f #define PHY_PAGE_REG 22 -#define E6000SW_NUM_PHYS 5 +/* + * Scratch and Misc register accessed via + * 'Switch Global Registers' (REG_GLOBAL2) + */ +#define SCR_AND_MISC_REG 0x1a + +#define SCR_AND_MISC_PTR_CFG 0x7000 +#define SCR_AND_MISC_DATA_CFG_MASK 0xf0 + #define E6000SW_NUM_PHY_REGS 29 -#define E6000SW_CPUPORTS_MASK ((1 << 5) | (1 << 6)) #define E6000SW_NUM_VGROUPS 8 -#define E6000SW_NUM_PORTS 7 +#define E6000SW_MAX_PORTS 10 #define E6000SW_PORT_NO_VGROUP -1 #define E6000SW_DEFAULT_AGETIME 20 #define E6000SW_RETRIES 100 - - -/* Default vlangroups */ -#define E6000SW_DEF_VLANGROUP0 (1 | (1 << 1) | (1 << 2) | (1 << 3) | \ - (1 << 6)) -#define E6000SW_DEF_VLANGROUP1 ((1 << 4) | (1 << 5)) +#define E6000SW_SMI_TIMEOUT 16 #endif /* _E6000SWREG_H_ */ Index: projects/clang400-import/sys/dev/firewire/sbp.c =================================================================== --- projects/clang400-import/sys/dev/firewire/sbp.c (revision 311459) +++ projects/clang400-import/sys/dev/firewire/sbp.c (revision 311460) @@ -1,2833 +1,2833 @@ /*- * Copyright (c) 2003 Hidetoshi Shimokawa * Copyright (c) 1998-2002 Katsushi Kobayashi and Hidetoshi Shimokawa * 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 acknowledgement as bellow: * * This product includes software developed by K. Kobayashi and H. Shimokawa * * 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. * * $FreeBSD$ * */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #define ccb_sdev_ptr spriv_ptr0 #define ccb_sbp_ptr spriv_ptr1 #define SBP_NUM_TARGETS 8 /* MAX 64 */ /* * Scan_bus doesn't work for more than 8 LUNs * because of CAM_SCSI2_MAXLUN in cam_xpt.c */ #define SBP_NUM_LUNS 64 #define SBP_MAXPHYS MIN(MAXPHYS, (512*1024) /* 512KB */) #define SBP_DMA_SIZE PAGE_SIZE #define SBP_LOGIN_SIZE sizeof(struct sbp_login_res) #define SBP_QUEUE_LEN ((SBP_DMA_SIZE - SBP_LOGIN_SIZE) / sizeof(struct sbp_ocb)) #define SBP_NUM_OCB (SBP_QUEUE_LEN * SBP_NUM_TARGETS) /* * STATUS FIFO addressing * bit *----------------------- * 0- 1( 2): 0 (alignment) * 2- 7( 6): target * 8-15( 8): lun * 16-31( 8): reserved * 32-47(16): SBP_BIND_HI * 48-64(16): bus_id, node_id */ #define SBP_BIND_HI 0x1 #define SBP_DEV2ADDR(t, l) \ (((u_int64_t)SBP_BIND_HI << 32) \ | (((l) & 0xff) << 8) \ | (((t) & 0x3f) << 2)) #define SBP_ADDR2TRG(a) (((a) >> 2) & 0x3f) #define SBP_ADDR2LUN(a) (((a) >> 8) & 0xff) #define SBP_INITIATOR 7 static char *orb_fun_name[] = { ORB_FUN_NAMES }; static int debug = 0; static int auto_login = 1; static int max_speed = -1; static int sbp_cold = 1; static int ex_login = 1; static int login_delay = 1000; /* msec */ static int scan_delay = 500; /* msec */ static int use_doorbell = 0; static int sbp_tags = 0; SYSCTL_DECL(_hw_firewire); static SYSCTL_NODE(_hw_firewire, OID_AUTO, sbp, CTLFLAG_RD, 0, "SBP-II Subsystem"); SYSCTL_INT(_debug, OID_AUTO, sbp_debug, CTLFLAG_RWTUN, &debug, 0, "SBP debug flag"); SYSCTL_INT(_hw_firewire_sbp, OID_AUTO, auto_login, CTLFLAG_RWTUN, &auto_login, 0, "SBP perform login automatically"); SYSCTL_INT(_hw_firewire_sbp, OID_AUTO, max_speed, CTLFLAG_RWTUN, &max_speed, 0, "SBP transfer max speed"); SYSCTL_INT(_hw_firewire_sbp, OID_AUTO, exclusive_login, CTLFLAG_RWTUN, &ex_login, 0, "SBP enable exclusive login"); SYSCTL_INT(_hw_firewire_sbp, OID_AUTO, login_delay, CTLFLAG_RWTUN, &login_delay, 0, "SBP login delay in msec"); SYSCTL_INT(_hw_firewire_sbp, OID_AUTO, scan_delay, CTLFLAG_RWTUN, &scan_delay, 0, "SBP scan delay in msec"); SYSCTL_INT(_hw_firewire_sbp, OID_AUTO, use_doorbell, CTLFLAG_RWTUN, &use_doorbell, 0, "SBP use doorbell request"); SYSCTL_INT(_hw_firewire_sbp, OID_AUTO, tags, CTLFLAG_RWTUN, &sbp_tags, 0, "SBP tagged queuing support"); #define NEED_RESPONSE 0 #define SBP_SEG_MAX rounddown(0xffff, PAGE_SIZE) #ifdef __sparc64__ /* iommu */ #define SBP_IND_MAX howmany(SBP_MAXPHYS, SBP_SEG_MAX) #else #define SBP_IND_MAX howmany(SBP_MAXPHYS, PAGE_SIZE) #endif struct sbp_ocb { STAILQ_ENTRY(sbp_ocb) ocb; union ccb *ccb; bus_addr_t bus_addr; uint32_t orb[8]; #define IND_PTR_OFFSET (8*sizeof(uint32_t)) struct ind_ptr ind_ptr[SBP_IND_MAX]; struct sbp_dev *sdev; int flags; /* XXX should be removed */ bus_dmamap_t dmamap; struct callout timer; }; #define OCB_ACT_MGM 0 #define OCB_ACT_CMD 1 #define OCB_MATCH(o,s) ((o)->bus_addr == ntohl((s)->orb_lo)) struct sbp_dev { #define SBP_DEV_RESET 0 /* accept login */ #define SBP_DEV_LOGIN 1 /* to login */ #if 0 #define SBP_DEV_RECONN 2 /* to reconnect */ #endif #define SBP_DEV_TOATTACH 3 /* to attach */ #define SBP_DEV_PROBE 4 /* scan lun */ #define SBP_DEV_ATTACHED 5 /* in operation */ #define SBP_DEV_DEAD 6 /* unavailable unit */ #define SBP_DEV_RETRY 7 /* unavailable unit */ uint8_t status:4, timeout:4; uint8_t type; uint16_t lun_id; uint16_t freeze; #define ORB_LINK_DEAD (1 << 0) #define VALID_LUN (1 << 1) #define ORB_POINTER_ACTIVE (1 << 2) #define ORB_POINTER_NEED (1 << 3) #define ORB_DOORBELL_ACTIVE (1 << 4) #define ORB_DOORBELL_NEED (1 << 5) #define ORB_SHORTAGE (1 << 6) uint16_t flags; struct cam_path *path; struct sbp_target *target; struct fwdma_alloc dma; struct sbp_login_res *login; struct callout login_callout; struct sbp_ocb *ocb; STAILQ_HEAD(, sbp_ocb) ocbs; STAILQ_HEAD(, sbp_ocb) free_ocbs; struct sbp_ocb *last_ocb; char vendor[32]; char product[32]; char revision[10]; char bustgtlun[32]; }; struct sbp_target { int target_id; int num_lun; struct sbp_dev **luns; struct sbp_softc *sbp; struct fw_device *fwdev; uint32_t mgm_hi, mgm_lo; struct sbp_ocb *mgm_ocb_cur; STAILQ_HEAD(, sbp_ocb) mgm_ocb_queue; struct callout mgm_ocb_timeout; struct callout scan_callout; STAILQ_HEAD(, fw_xfer) xferlist; int n_xfer; }; struct sbp_softc { struct firewire_dev_comm fd; struct cam_sim *sim; struct cam_path *path; struct sbp_target targets[SBP_NUM_TARGETS]; struct fw_bind fwb; bus_dma_tag_t dmat; struct timeval last_busreset; #define SIMQ_FREEZED 1 int flags; struct mtx mtx; }; #define SBP_LOCK(sbp) mtx_lock(&(sbp)->mtx) #define SBP_UNLOCK(sbp) mtx_unlock(&(sbp)->mtx) #define SBP_LOCK_ASSERT(sbp) mtx_assert(&(sbp)->mtx, MA_OWNED) static void sbp_post_explore (void *); static void sbp_recv (struct fw_xfer *); static void sbp_mgm_callback (struct fw_xfer *); #if 0 static void sbp_cmd_callback (struct fw_xfer *); #endif static void sbp_orb_pointer (struct sbp_dev *, struct sbp_ocb *); static void sbp_doorbell(struct sbp_dev *); static void sbp_execute_ocb (void *, bus_dma_segment_t *, int, int); static void sbp_free_ocb (struct sbp_dev *, struct sbp_ocb *); static void sbp_abort_ocb (struct sbp_ocb *, int); static void sbp_abort_all_ocbs (struct sbp_dev *, int); static struct fw_xfer * sbp_write_cmd (struct sbp_dev *, int, int); static struct sbp_ocb * sbp_get_ocb (struct sbp_dev *); static struct sbp_ocb * sbp_enqueue_ocb (struct sbp_dev *, struct sbp_ocb *); static struct sbp_ocb * sbp_dequeue_ocb (struct sbp_dev *, struct sbp_status *); static void sbp_cam_detach_sdev(struct sbp_dev *); static void sbp_free_sdev(struct sbp_dev *); static void sbp_cam_detach_target (struct sbp_target *); static void sbp_free_target (struct sbp_target *); static void sbp_mgm_timeout (void *arg); static void sbp_timeout (void *arg); static void sbp_mgm_orb (struct sbp_dev *, int, struct sbp_ocb *); static MALLOC_DEFINE(M_SBP, "sbp", "SBP-II/FireWire"); /* cam related functions */ static void sbp_action(struct cam_sim *sim, union ccb *ccb); static void sbp_poll(struct cam_sim *sim); static void sbp_cam_scan_lun(struct cam_periph *, union ccb *); static void sbp_cam_scan_target(void *arg); static char *orb_status0[] = { /* 0 */ "No additional information to report", /* 1 */ "Request type not supported", /* 2 */ "Speed not supported", /* 3 */ "Page size not supported", /* 4 */ "Access denied", /* 5 */ "Logical unit not supported", /* 6 */ "Maximum payload too small", /* 7 */ "Reserved for future standardization", /* 8 */ "Resources unavailable", /* 9 */ "Function rejected", /* A */ "Login ID not recognized", /* B */ "Dummy ORB completed", /* C */ "Request aborted", /* FF */ "Unspecified error" #define MAX_ORB_STATUS0 0xd }; static char *orb_status1_object[] = { /* 0 */ "Operation request block (ORB)", /* 1 */ "Data buffer", /* 2 */ "Page table", /* 3 */ "Unable to specify" }; static char *orb_status1_serial_bus_error[] = { /* 0 */ "Missing acknowledge", /* 1 */ "Reserved; not to be used", /* 2 */ "Time-out error", /* 3 */ "Reserved; not to be used", /* 4 */ "Busy retry limit exceeded(X)", /* 5 */ "Busy retry limit exceeded(A)", /* 6 */ "Busy retry limit exceeded(B)", /* 7 */ "Reserved for future standardization", /* 8 */ "Reserved for future standardization", /* 9 */ "Reserved for future standardization", /* A */ "Reserved for future standardization", /* B */ "Tardy retry limit exceeded", /* C */ "Conflict error", /* D */ "Data error", /* E */ "Type error", /* F */ "Address error" }; static void sbp_identify(driver_t *driver, device_t parent) { SBP_DEBUG(0) printf("sbp_identify\n"); END_DEBUG if (device_find_child(parent, "sbp", -1) == NULL) BUS_ADD_CHILD(parent, 0, "sbp", -1); } /* * sbp_probe() */ static int sbp_probe(device_t dev) { SBP_DEBUG(0) printf("sbp_probe\n"); END_DEBUG device_set_desc(dev, "SBP-2/SCSI over FireWire"); #if 0 if (bootverbose) debug = bootverbose; #endif return (0); } /* * Display device characteristics on the console */ static void sbp_show_sdev_info(struct sbp_dev *sdev) { struct fw_device *fwdev; fwdev = sdev->target->fwdev; device_printf(sdev->target->sbp->fd.dev, "%s: %s: ordered:%d type:%d EUI:%08x%08x node:%d " "speed:%d maxrec:%d\n", __func__, sdev->bustgtlun, (sdev->type & 0x40) >> 6, (sdev->type & 0x1f), fwdev->eui.hi, fwdev->eui.lo, fwdev->dst, fwdev->speed, fwdev->maxrec); device_printf(sdev->target->sbp->fd.dev, "%s: %s '%s' '%s' '%s'\n", __func__, sdev->bustgtlun, sdev->vendor, sdev->product, sdev->revision); } static struct { int bus; int target; struct fw_eui64 eui; } wired[] = { /* Bus Target EUI64 */ #if 0 {0, 2, {0x00018ea0, 0x01fd0154}}, /* Logitec HDD */ {0, 0, {0x00018ea6, 0x00100682}}, /* Logitec DVD */ {0, 1, {0x00d03200, 0xa412006a}}, /* Yano HDD */ #endif {-1, -1, {0,0}} }; static int sbp_new_target(struct sbp_softc *sbp, struct fw_device *fwdev) { int bus, i, target=-1; char w[SBP_NUM_TARGETS]; bzero(w, sizeof(w)); bus = device_get_unit(sbp->fd.dev); /* XXX wired-down configuration should be gotten from tunable or device hint */ for (i = 0; wired[i].bus >= 0; i++) { if (wired[i].bus == bus) { w[wired[i].target] = 1; if (wired[i].eui.hi == fwdev->eui.hi && wired[i].eui.lo == fwdev->eui.lo) target = wired[i].target; } } if (target >= 0) { if (target < SBP_NUM_TARGETS && sbp->targets[target].fwdev == NULL) return (target); device_printf(sbp->fd.dev, "target %d is not free for %08x:%08x\n", target, fwdev->eui.hi, fwdev->eui.lo); target = -1; } /* non-wired target */ for (i = 0; i < SBP_NUM_TARGETS; i++) if (sbp->targets[i].fwdev == NULL && w[i] == 0) { target = i; break; } return target; } static void sbp_alloc_lun(struct sbp_target *target) { struct crom_context cc; struct csrreg *reg; struct sbp_dev *sdev, **newluns; struct sbp_softc *sbp; int maxlun, lun, i; sbp = target->sbp; SBP_LOCK_ASSERT(sbp); crom_init_context(&cc, target->fwdev->csrrom); /* XXX shoud parse appropriate unit directories only */ maxlun = -1; while (cc.depth >= 0) { reg = crom_search_key(&cc, CROM_LUN); if (reg == NULL) break; lun = reg->val & 0xffff; SBP_DEBUG(0) printf("target %d lun %d found\n", target->target_id, lun); END_DEBUG if (maxlun < lun) maxlun = lun; crom_next(&cc); } if (maxlun < 0) device_printf(target->sbp->fd.dev, "%d no LUN found\n", target->target_id); maxlun++; if (maxlun >= SBP_NUM_LUNS) maxlun = SBP_NUM_LUNS; /* Invalidiate stale devices */ for (lun = 0; lun < target->num_lun; lun++) { sdev = target->luns[lun]; if (sdev == NULL) continue; sdev->flags &= ~VALID_LUN; if (lun >= maxlun) { /* lost device */ sbp_cam_detach_sdev(sdev); sbp_free_sdev(sdev); target->luns[lun] = NULL; } } /* Reallocate */ if (maxlun != target->num_lun) { newluns = (struct sbp_dev **) realloc(target->luns, sizeof(struct sbp_dev *) * maxlun, M_SBP, M_NOWAIT | M_ZERO); if (newluns == NULL) { printf("%s: realloc failed\n", __func__); newluns = target->luns; maxlun = target->num_lun; } /* * We must zero the extended region for the case * realloc() doesn't allocate new buffer. */ if (maxlun > target->num_lun) bzero(&newluns[target->num_lun], sizeof(struct sbp_dev *) * (maxlun - target->num_lun)); target->luns = newluns; target->num_lun = maxlun; } crom_init_context(&cc, target->fwdev->csrrom); while (cc.depth >= 0) { int new = 0; reg = crom_search_key(&cc, CROM_LUN); if (reg == NULL) break; lun = reg->val & 0xffff; if (lun >= SBP_NUM_LUNS) { printf("too large lun %d\n", lun); goto next; } sdev = target->luns[lun]; if (sdev == NULL) { sdev = malloc(sizeof(struct sbp_dev), M_SBP, M_NOWAIT | M_ZERO); if (sdev == NULL) { printf("%s: malloc failed\n", __func__); goto next; } target->luns[lun] = sdev; sdev->lun_id = lun; sdev->target = target; STAILQ_INIT(&sdev->ocbs); callout_init_mtx(&sdev->login_callout, &sbp->mtx, 0); sdev->status = SBP_DEV_RESET; new = 1; snprintf(sdev->bustgtlun, 32, "%s:%d:%d", device_get_nameunit(sdev->target->sbp->fd.dev), sdev->target->target_id, sdev->lun_id); } sdev->flags |= VALID_LUN; sdev->type = (reg->val & 0xff0000) >> 16; if (new == 0) goto next; fwdma_malloc(sbp->fd.fc, /* alignment */ sizeof(uint32_t), SBP_DMA_SIZE, &sdev->dma, BUS_DMA_NOWAIT | BUS_DMA_COHERENT); if (sdev->dma.v_addr == NULL) { printf("%s: dma space allocation failed\n", __func__); free(sdev, M_SBP); target->luns[lun] = NULL; goto next; } sdev->login = (struct sbp_login_res *) sdev->dma.v_addr; sdev->ocb = (struct sbp_ocb *) ((char *)sdev->dma.v_addr + SBP_LOGIN_SIZE); bzero((char *)sdev->ocb, sizeof(struct sbp_ocb) * SBP_QUEUE_LEN); STAILQ_INIT(&sdev->free_ocbs); for (i = 0; i < SBP_QUEUE_LEN; i++) { struct sbp_ocb *ocb; ocb = &sdev->ocb[i]; ocb->bus_addr = sdev->dma.bus_addr + SBP_LOGIN_SIZE + sizeof(struct sbp_ocb) * i + offsetof(struct sbp_ocb, orb[0]); if (bus_dmamap_create(sbp->dmat, 0, &ocb->dmamap)) { printf("sbp_attach: cannot create dmamap\n"); /* XXX */ goto next; } callout_init_mtx(&ocb->timer, &sbp->mtx, 0); sbp_free_ocb(sdev, ocb); } next: crom_next(&cc); } for (lun = 0; lun < target->num_lun; lun++) { sdev = target->luns[lun]; if (sdev != NULL && (sdev->flags & VALID_LUN) == 0) { sbp_cam_detach_sdev(sdev); sbp_free_sdev(sdev); target->luns[lun] = NULL; } } } static struct sbp_target * sbp_alloc_target(struct sbp_softc *sbp, struct fw_device *fwdev) { int i; struct sbp_target *target; struct crom_context cc; struct csrreg *reg; SBP_DEBUG(1) printf("sbp_alloc_target\n"); END_DEBUG i = sbp_new_target(sbp, fwdev); if (i < 0) { device_printf(sbp->fd.dev, "increase SBP_NUM_TARGETS!\n"); return NULL; } /* new target */ target = &sbp->targets[i]; target->fwdev = fwdev; target->target_id = i; /* XXX we may want to reload mgm port after each bus reset */ /* XXX there might be multiple management agents */ crom_init_context(&cc, target->fwdev->csrrom); reg = crom_search_key(&cc, CROM_MGM); if (reg == NULL || reg->val == 0) { printf("NULL management address\n"); target->fwdev = NULL; return NULL; } target->mgm_hi = 0xffff; target->mgm_lo = 0xf0000000 | (reg->val << 2); target->mgm_ocb_cur = NULL; SBP_DEBUG(1) printf("target:%d mgm_port: %x\n", i, target->mgm_lo); END_DEBUG STAILQ_INIT(&target->xferlist); target->n_xfer = 0; STAILQ_INIT(&target->mgm_ocb_queue); callout_init_mtx(&target->mgm_ocb_timeout, &sbp->mtx, 0); callout_init_mtx(&target->scan_callout, &sbp->mtx, 0); target->luns = NULL; target->num_lun = 0; return target; } static void sbp_probe_lun(struct sbp_dev *sdev) { struct fw_device *fwdev; struct crom_context c, *cc = &c; struct csrreg *reg; bzero(sdev->vendor, sizeof(sdev->vendor)); bzero(sdev->product, sizeof(sdev->product)); fwdev = sdev->target->fwdev; crom_init_context(cc, fwdev->csrrom); /* get vendor string */ crom_search_key(cc, CSRKEY_VENDOR); crom_next(cc); crom_parse_text(cc, sdev->vendor, sizeof(sdev->vendor)); /* skip to the unit directory for SBP-2 */ while ((reg = crom_search_key(cc, CSRKEY_VER)) != NULL) { if (reg->val == CSRVAL_T10SBP2) break; crom_next(cc); } /* get firmware revision */ reg = crom_search_key(cc, CSRKEY_FIRM_VER); if (reg != NULL) snprintf(sdev->revision, sizeof(sdev->revision), "%06x", reg->val); /* get product string */ crom_search_key(cc, CSRKEY_MODEL); crom_next(cc); crom_parse_text(cc, sdev->product, sizeof(sdev->product)); } static void sbp_login_callout(void *arg) { struct sbp_dev *sdev = (struct sbp_dev *)arg; SBP_LOCK_ASSERT(sdev->target->sbp); sbp_mgm_orb(sdev, ORB_FUN_LGI, NULL); } static void sbp_login(struct sbp_dev *sdev) { struct timeval delta; struct timeval t; int ticks = 0; microtime(&delta); timevalsub(&delta, &sdev->target->sbp->last_busreset); t.tv_sec = login_delay / 1000; t.tv_usec = (login_delay % 1000) * 1000; timevalsub(&t, &delta); if (t.tv_sec >= 0 && t.tv_usec > 0) ticks = (t.tv_sec * 1000 + t.tv_usec / 1000) * hz / 1000; SBP_DEBUG(0) printf("%s: sec = %jd usec = %ld ticks = %d\n", __func__, (intmax_t)t.tv_sec, t.tv_usec, ticks); END_DEBUG callout_reset(&sdev->login_callout, ticks, sbp_login_callout, (void *)(sdev)); } #define SBP_FWDEV_ALIVE(fwdev) (((fwdev)->status == FWDEVATTACHED) \ && crom_has_specver((fwdev)->csrrom, CSRVAL_ANSIT10, CSRVAL_T10SBP2)) static void sbp_probe_target(void *arg) { struct sbp_target *target = (struct sbp_target *)arg; struct sbp_softc *sbp = target->sbp; struct sbp_dev *sdev; int i, alive; alive = SBP_FWDEV_ALIVE(target->fwdev); SBP_DEBUG(1) device_printf(sbp->fd.dev, "%s %d%salive\n", __func__, target->target_id, (!alive) ? " not " : ""); END_DEBUG sbp = target->sbp; SBP_LOCK_ASSERT(sbp); sbp_alloc_lun(target); /* XXX untimeout mgm_ocb and dequeue */ for (i=0; i < target->num_lun; i++) { sdev = target->luns[i]; if (sdev == NULL) continue; if (alive && (sdev->status != SBP_DEV_DEAD)) { if (sdev->path != NULL) { xpt_freeze_devq(sdev->path, 1); sdev->freeze++; } sbp_probe_lun(sdev); sbp_show_sdev_info(sdev); sbp_abort_all_ocbs(sdev, CAM_SCSI_BUS_RESET); switch (sdev->status) { case SBP_DEV_RESET: /* new or revived target */ if (auto_login) sbp_login(sdev); break; case SBP_DEV_TOATTACH: case SBP_DEV_PROBE: case SBP_DEV_ATTACHED: case SBP_DEV_RETRY: default: sbp_mgm_orb(sdev, ORB_FUN_RCN, NULL); break; } } else { switch (sdev->status) { case SBP_DEV_ATTACHED: SBP_DEBUG(0) /* the device has gone */ device_printf(sbp->fd.dev, "%s: lost target\n", __func__); END_DEBUG if (sdev->path) { xpt_freeze_devq(sdev->path, 1); sdev->freeze++; } sdev->status = SBP_DEV_RETRY; sbp_cam_detach_sdev(sdev); sbp_free_sdev(sdev); target->luns[i] = NULL; break; case SBP_DEV_PROBE: case SBP_DEV_TOATTACH: sdev->status = SBP_DEV_RESET; break; case SBP_DEV_RETRY: case SBP_DEV_RESET: case SBP_DEV_DEAD: break; } } } } static void sbp_post_busreset(void *arg) { struct sbp_softc *sbp; sbp = (struct sbp_softc *)arg; SBP_DEBUG(0) printf("sbp_post_busreset\n"); END_DEBUG SBP_LOCK(sbp); if ((sbp->sim->flags & SIMQ_FREEZED) == 0) { xpt_freeze_simq(sbp->sim, /*count*/1); sbp->sim->flags |= SIMQ_FREEZED; } microtime(&sbp->last_busreset); SBP_UNLOCK(sbp); } static void sbp_post_explore(void *arg) { struct sbp_softc *sbp = (struct sbp_softc *)arg; struct sbp_target *target; struct fw_device *fwdev; int i, alive; SBP_DEBUG(0) printf("sbp_post_explore (sbp_cold=%d)\n", sbp_cold); END_DEBUG /* We need physical access */ if (!firewire_phydma_enable) return; if (sbp_cold > 0) sbp_cold--; SBP_LOCK(sbp); #if 0 /* * XXX don't let CAM the bus rest. * CAM tries to do something with freezed (DEV_RETRY) devices. */ xpt_async(AC_BUS_RESET, sbp->path, /*arg*/ NULL); #endif /* Garbage Collection */ for (i = 0; i < SBP_NUM_TARGETS; i++) { target = &sbp->targets[i]; STAILQ_FOREACH(fwdev, &sbp->fd.fc->devices, link) if (target->fwdev == NULL || target->fwdev == fwdev) break; if (fwdev == NULL) { /* device has removed in lower driver */ sbp_cam_detach_target(target); sbp_free_target(target); } } /* traverse device list */ STAILQ_FOREACH(fwdev, &sbp->fd.fc->devices, link) { SBP_DEBUG(0) device_printf(sbp->fd.dev,"%s:: EUI:%08x%08x %s attached, state=%d\n", __func__, fwdev->eui.hi, fwdev->eui.lo, (fwdev->status != FWDEVATTACHED) ? "not" : "", fwdev->status); END_DEBUG alive = SBP_FWDEV_ALIVE(fwdev); for (i = 0; i < SBP_NUM_TARGETS; i++) { target = &sbp->targets[i]; if (target->fwdev == fwdev) { /* known target */ break; } } if (i == SBP_NUM_TARGETS) { if (alive) { /* new target */ target = sbp_alloc_target(sbp, fwdev); if (target == NULL) continue; } else { continue; } } sbp_probe_target((void *)target); if (target->num_lun == 0) sbp_free_target(target); } xpt_release_simq(sbp->sim, /*run queue*/TRUE); sbp->sim->flags &= ~SIMQ_FREEZED; SBP_UNLOCK(sbp); } #if NEED_RESPONSE static void sbp_loginres_callback(struct fw_xfer *xfer) { struct sbp_dev *sdev; sdev = (struct sbp_dev *)xfer->sc; SBP_DEBUG(1) device_printf(sdev->target->sbp->fd.dev,"%s\n", __func__); END_DEBUG /* recycle */ SBP_LOCK(sdev->target->sbp); STAILQ_INSERT_TAIL(&sdev->target->sbp->fwb.xferlist, xfer, link); SBP_UNLOCK(sdev->target->sbp); return; } #endif static __inline void sbp_xfer_free(struct fw_xfer *xfer) { struct sbp_dev *sdev; sdev = (struct sbp_dev *)xfer->sc; fw_xfer_unload(xfer); SBP_LOCK_ASSERT(sdev->target->sbp); STAILQ_INSERT_TAIL(&sdev->target->xferlist, xfer, link); } static void sbp_reset_start_callback(struct fw_xfer *xfer) { struct sbp_dev *tsdev, *sdev = (struct sbp_dev *)xfer->sc; struct sbp_target *target = sdev->target; int i; if (xfer->resp != 0) { device_printf(sdev->target->sbp->fd.dev, "%s: %s failed: resp=%d\n", __func__, sdev->bustgtlun, xfer->resp); } SBP_LOCK(target->sbp); for (i = 0; i < target->num_lun; i++) { tsdev = target->luns[i]; if (tsdev != NULL && tsdev->status == SBP_DEV_LOGIN) sbp_login(tsdev); } SBP_UNLOCK(target->sbp); } static void sbp_reset_start(struct sbp_dev *sdev) { struct fw_xfer *xfer; struct fw_pkt *fp; SBP_DEBUG(0) device_printf(sdev->target->sbp->fd.dev, "%s:%s\n", __func__,sdev->bustgtlun); END_DEBUG xfer = sbp_write_cmd(sdev, FWTCODE_WREQQ, 0); xfer->hand = sbp_reset_start_callback; fp = &xfer->send.hdr; fp->mode.wreqq.dest_hi = 0xffff; fp->mode.wreqq.dest_lo = 0xf0000000 | RESET_START; fp->mode.wreqq.data = htonl(0xf); fw_asyreq(xfer->fc, -1, xfer); } static void sbp_mgm_callback(struct fw_xfer *xfer) { struct sbp_dev *sdev; int resp; sdev = (struct sbp_dev *)xfer->sc; SBP_DEBUG(1) device_printf(sdev->target->sbp->fd.dev, "%s:%s\n", __func__, sdev->bustgtlun); END_DEBUG resp = xfer->resp; SBP_LOCK(sdev->target->sbp); sbp_xfer_free(xfer); SBP_UNLOCK(sdev->target->sbp); } static struct sbp_dev * sbp_next_dev(struct sbp_target *target, int lun) { struct sbp_dev **sdevp; int i; for (i = lun, sdevp = &target->luns[lun]; i < target->num_lun; i++, sdevp++) if (*sdevp != NULL && (*sdevp)->status == SBP_DEV_PROBE) return (*sdevp); return (NULL); } #define SCAN_PRI 1 static void sbp_cam_scan_lun(struct cam_periph *periph, union ccb *ccb) { struct sbp_target *target; struct sbp_dev *sdev; sdev = (struct sbp_dev *) ccb->ccb_h.ccb_sdev_ptr; target = sdev->target; SBP_LOCK_ASSERT(target->sbp); SBP_DEBUG(0) device_printf(sdev->target->sbp->fd.dev, "%s:%s\n", __func__, sdev->bustgtlun); END_DEBUG if ((ccb->ccb_h.status & CAM_STATUS_MASK) == CAM_REQ_CMP) { sdev->status = SBP_DEV_ATTACHED; } else { device_printf(sdev->target->sbp->fd.dev, "%s:%s failed\n", __func__, sdev->bustgtlun); } sdev = sbp_next_dev(target, sdev->lun_id + 1); if (sdev == NULL) { free(ccb, M_SBP); return; } /* reuse ccb */ xpt_setup_ccb(&ccb->ccb_h, sdev->path, SCAN_PRI); ccb->ccb_h.ccb_sdev_ptr = sdev; xpt_action(ccb); xpt_release_devq(sdev->path, sdev->freeze, TRUE); sdev->freeze = 1; } static void sbp_cam_scan_target(void *arg) { struct sbp_target *target = (struct sbp_target *)arg; struct sbp_dev *sdev; union ccb *ccb; SBP_LOCK_ASSERT(target->sbp); sdev = sbp_next_dev(target, 0); if (sdev == NULL) { printf("sbp_cam_scan_target: nothing to do for target%d\n", target->target_id); return; } SBP_DEBUG(0) device_printf(sdev->target->sbp->fd.dev, "%s:%s\n", __func__, sdev->bustgtlun); END_DEBUG ccb = malloc(sizeof(union ccb), M_SBP, M_NOWAIT | M_ZERO); if (ccb == NULL) { printf("sbp_cam_scan_target: malloc failed\n"); return; } xpt_setup_ccb(&ccb->ccb_h, sdev->path, SCAN_PRI); ccb->ccb_h.func_code = XPT_SCAN_LUN; ccb->ccb_h.cbfcnp = sbp_cam_scan_lun; ccb->ccb_h.flags |= CAM_DEV_QFREEZE; ccb->crcn.flags = CAM_FLAG_NONE; ccb->ccb_h.ccb_sdev_ptr = sdev; /* The scan is in progress now. */ xpt_action(ccb); xpt_release_devq(sdev->path, sdev->freeze, TRUE); sdev->freeze = 1; } static __inline void sbp_scan_dev(struct sbp_dev *sdev) { sdev->status = SBP_DEV_PROBE; callout_reset_sbt(&sdev->target->scan_callout, SBT_1MS * scan_delay, 0, sbp_cam_scan_target, (void *)sdev->target, 0); } static void sbp_do_attach(struct fw_xfer *xfer) { struct sbp_dev *sdev; struct sbp_target *target; struct sbp_softc *sbp; sdev = (struct sbp_dev *)xfer->sc; target = sdev->target; sbp = target->sbp; SBP_LOCK(sbp); SBP_DEBUG(0) device_printf(sdev->target->sbp->fd.dev, "%s:%s\n", __func__, sdev->bustgtlun); END_DEBUG sbp_xfer_free(xfer); if (sdev->path == NULL) xpt_create_path(&sdev->path, NULL, cam_sim_path(target->sbp->sim), target->target_id, sdev->lun_id); /* * Let CAM scan the bus if we are in the boot process. * XXX xpt_scan_bus cannot detect LUN larger than 0 * if LUN 0 doesn't exist. */ if (sbp_cold > 0) { sdev->status = SBP_DEV_ATTACHED; SBP_UNLOCK(sbp); return; } sbp_scan_dev(sdev); SBP_UNLOCK(sbp); } static void sbp_agent_reset_callback(struct fw_xfer *xfer) { struct sbp_dev *sdev; sdev = (struct sbp_dev *)xfer->sc; SBP_DEBUG(1) device_printf(sdev->target->sbp->fd.dev, "%s:%s\n", __func__, sdev->bustgtlun); END_DEBUG if (xfer->resp != 0) { device_printf(sdev->target->sbp->fd.dev, "%s:%s resp=%d\n", __func__, sdev->bustgtlun, xfer->resp); } SBP_LOCK(sdev->target->sbp); sbp_xfer_free(xfer); if (sdev->path) { xpt_release_devq(sdev->path, sdev->freeze, TRUE); sdev->freeze = 0; } SBP_UNLOCK(sdev->target->sbp); } static void sbp_agent_reset(struct sbp_dev *sdev) { struct fw_xfer *xfer; struct fw_pkt *fp; SBP_LOCK_ASSERT(sdev->target->sbp); SBP_DEBUG(0) device_printf(sdev->target->sbp->fd.dev, "%s:%s\n", __func__, sdev->bustgtlun); END_DEBUG xfer = sbp_write_cmd(sdev, FWTCODE_WREQQ, 0x04); if (xfer == NULL) return; if (sdev->status == SBP_DEV_ATTACHED || sdev->status == SBP_DEV_PROBE) xfer->hand = sbp_agent_reset_callback; else xfer->hand = sbp_do_attach; fp = &xfer->send.hdr; fp->mode.wreqq.data = htonl(0xf); fw_asyreq(xfer->fc, -1, xfer); sbp_abort_all_ocbs(sdev, CAM_BDR_SENT); } static void sbp_busy_timeout_callback(struct fw_xfer *xfer) { struct sbp_dev *sdev; sdev = (struct sbp_dev *)xfer->sc; SBP_DEBUG(1) device_printf(sdev->target->sbp->fd.dev, "%s:%s\n", __func__, sdev->bustgtlun); END_DEBUG SBP_LOCK(sdev->target->sbp); sbp_xfer_free(xfer); sbp_agent_reset(sdev); SBP_UNLOCK(sdev->target->sbp); } static void sbp_busy_timeout(struct sbp_dev *sdev) { struct fw_pkt *fp; struct fw_xfer *xfer; SBP_DEBUG(0) device_printf(sdev->target->sbp->fd.dev, "%s:%s\n", __func__, sdev->bustgtlun); END_DEBUG xfer = sbp_write_cmd(sdev, FWTCODE_WREQQ, 0); xfer->hand = sbp_busy_timeout_callback; fp = &xfer->send.hdr; fp->mode.wreqq.dest_hi = 0xffff; fp->mode.wreqq.dest_lo = 0xf0000000 | BUSY_TIMEOUT; fp->mode.wreqq.data = htonl((1 << (13 + 12)) | 0xf); fw_asyreq(xfer->fc, -1, xfer); } static void sbp_orb_pointer_callback(struct fw_xfer *xfer) { struct sbp_dev *sdev; sdev = (struct sbp_dev *)xfer->sc; SBP_DEBUG(2) device_printf(sdev->target->sbp->fd.dev, "%s:%s\n", __func__, sdev->bustgtlun); END_DEBUG if (xfer->resp != 0) { /* XXX */ printf("%s: xfer->resp = %d\n", __func__, xfer->resp); } SBP_LOCK(sdev->target->sbp); sbp_xfer_free(xfer); sdev->flags &= ~ORB_POINTER_ACTIVE; if ((sdev->flags & ORB_POINTER_NEED) != 0) { struct sbp_ocb *ocb; sdev->flags &= ~ORB_POINTER_NEED; ocb = STAILQ_FIRST(&sdev->ocbs); if (ocb != NULL) sbp_orb_pointer(sdev, ocb); } SBP_UNLOCK(sdev->target->sbp); return; } static void sbp_orb_pointer(struct sbp_dev *sdev, struct sbp_ocb *ocb) { struct fw_xfer *xfer; struct fw_pkt *fp; SBP_DEBUG(1) device_printf(sdev->target->sbp->fd.dev, "%s:%s 0x%08x\n", __func__, sdev->bustgtlun, (uint32_t)ocb->bus_addr); END_DEBUG SBP_LOCK_ASSERT(sdev->target->sbp); if ((sdev->flags & ORB_POINTER_ACTIVE) != 0) { SBP_DEBUG(0) printf("%s: orb pointer active\n", __func__); END_DEBUG sdev->flags |= ORB_POINTER_NEED; return; } sdev->flags |= ORB_POINTER_ACTIVE; xfer = sbp_write_cmd(sdev, FWTCODE_WREQB, 0x08); if (xfer == NULL) return; xfer->hand = sbp_orb_pointer_callback; fp = &xfer->send.hdr; fp->mode.wreqb.len = 8; fp->mode.wreqb.extcode = 0; xfer->send.payload[0] = htonl(((sdev->target->sbp->fd.fc->nodeid | FWLOCALBUS) << 16)); xfer->send.payload[1] = htonl((uint32_t)ocb->bus_addr); if (fw_asyreq(xfer->fc, -1, xfer) != 0) { sbp_xfer_free(xfer); ocb->ccb->ccb_h.status = CAM_REQ_INVALID; xpt_done(ocb->ccb); } } static void sbp_doorbell_callback(struct fw_xfer *xfer) { struct sbp_dev *sdev; sdev = (struct sbp_dev *)xfer->sc; SBP_DEBUG(1) device_printf(sdev->target->sbp->fd.dev, "%s:%s\n", __func__, sdev->bustgtlun); END_DEBUG if (xfer->resp != 0) { /* XXX */ device_printf(sdev->target->sbp->fd.dev, "%s: xfer->resp = %d\n", __func__, xfer->resp); } SBP_LOCK(sdev->target->sbp); sbp_xfer_free(xfer); sdev->flags &= ~ORB_DOORBELL_ACTIVE; if ((sdev->flags & ORB_DOORBELL_NEED) != 0) { sdev->flags &= ~ORB_DOORBELL_NEED; sbp_doorbell(sdev); } SBP_UNLOCK(sdev->target->sbp); } static void sbp_doorbell(struct sbp_dev *sdev) { struct fw_xfer *xfer; struct fw_pkt *fp; SBP_DEBUG(1) device_printf(sdev->target->sbp->fd.dev, "%s:%s\n", __func__, sdev->bustgtlun); END_DEBUG if ((sdev->flags & ORB_DOORBELL_ACTIVE) != 0) { sdev->flags |= ORB_DOORBELL_NEED; return; } sdev->flags |= ORB_DOORBELL_ACTIVE; xfer = sbp_write_cmd(sdev, FWTCODE_WREQQ, 0x10); if (xfer == NULL) return; xfer->hand = sbp_doorbell_callback; fp = &xfer->send.hdr; fp->mode.wreqq.data = htonl(0xf); fw_asyreq(xfer->fc, -1, xfer); } static struct fw_xfer * sbp_write_cmd(struct sbp_dev *sdev, int tcode, int offset) { struct fw_xfer *xfer; struct fw_pkt *fp; struct sbp_target *target; int new = 0; SBP_LOCK_ASSERT(sdev->target->sbp); target = sdev->target; xfer = STAILQ_FIRST(&target->xferlist); if (xfer == NULL) { if (target->n_xfer > 5 /* XXX */) { printf("sbp: no more xfer for this target\n"); return (NULL); } xfer = fw_xfer_alloc_buf(M_SBP, 8, 0); if (xfer == NULL) { printf("sbp: fw_xfer_alloc_buf failed\n"); return NULL; } target->n_xfer++; if (debug) printf("sbp: alloc %d xfer\n", target->n_xfer); new = 1; } else { STAILQ_REMOVE_HEAD(&target->xferlist, link); } if (new) { xfer->recv.pay_len = 0; xfer->send.spd = min(sdev->target->fwdev->speed, max_speed); xfer->fc = sdev->target->sbp->fd.fc; } if (tcode == FWTCODE_WREQB) xfer->send.pay_len = 8; else xfer->send.pay_len = 0; xfer->sc = (caddr_t)sdev; fp = &xfer->send.hdr; fp->mode.wreqq.dest_hi = sdev->login->cmd_hi; fp->mode.wreqq.dest_lo = sdev->login->cmd_lo + offset; fp->mode.wreqq.tlrt = 0; fp->mode.wreqq.tcode = tcode; fp->mode.wreqq.pri = 0; fp->mode.wreqq.dst = FWLOCALBUS | sdev->target->fwdev->dst; return xfer; } static void sbp_mgm_orb(struct sbp_dev *sdev, int func, struct sbp_ocb *aocb) { struct fw_xfer *xfer; struct fw_pkt *fp; struct sbp_ocb *ocb; struct sbp_target *target; int nid; target = sdev->target; nid = target->sbp->fd.fc->nodeid | FWLOCALBUS; SBP_LOCK_ASSERT(target->sbp); if (func == ORB_FUN_RUNQUEUE) { ocb = STAILQ_FIRST(&target->mgm_ocb_queue); if (target->mgm_ocb_cur != NULL || ocb == NULL) { return; } STAILQ_REMOVE_HEAD(&target->mgm_ocb_queue, ocb); goto start; } if ((ocb = sbp_get_ocb(sdev)) == NULL) { /* XXX */ return; } ocb->flags = OCB_ACT_MGM; ocb->sdev = sdev; bzero((void *)ocb->orb, sizeof(ocb->orb)); ocb->orb[6] = htonl((nid << 16) | SBP_BIND_HI); ocb->orb[7] = htonl(SBP_DEV2ADDR(target->target_id, sdev->lun_id)); SBP_DEBUG(0) device_printf(sdev->target->sbp->fd.dev, "%s:%s %s\n", __func__,sdev->bustgtlun, orb_fun_name[(func >> 16) & 0xf]); END_DEBUG switch (func) { case ORB_FUN_LGI: ocb->orb[0] = ocb->orb[1] = 0; /* password */ ocb->orb[2] = htonl(nid << 16); ocb->orb[3] = htonl(sdev->dma.bus_addr); ocb->orb[4] = htonl(ORB_NOTIFY | sdev->lun_id); if (ex_login) ocb->orb[4] |= htonl(ORB_EXV); ocb->orb[5] = htonl(SBP_LOGIN_SIZE); fwdma_sync(&sdev->dma, BUS_DMASYNC_PREREAD); break; case ORB_FUN_ATA: ocb->orb[0] = htonl((0 << 16) | 0); ocb->orb[1] = htonl(aocb->bus_addr & 0xffffffff); /* fall through */ case ORB_FUN_RCN: case ORB_FUN_LGO: case ORB_FUN_LUR: case ORB_FUN_RST: case ORB_FUN_ATS: ocb->orb[4] = htonl(ORB_NOTIFY | func | sdev->login->id); break; } if (target->mgm_ocb_cur != NULL) { /* there is a standing ORB */ STAILQ_INSERT_TAIL(&sdev->target->mgm_ocb_queue, ocb, ocb); return; } start: target->mgm_ocb_cur = ocb; callout_reset(&target->mgm_ocb_timeout, 5 * hz, sbp_mgm_timeout, (caddr_t)ocb); xfer = sbp_write_cmd(sdev, FWTCODE_WREQB, 0); if (xfer == NULL) { return; } xfer->hand = sbp_mgm_callback; fp = &xfer->send.hdr; fp->mode.wreqb.dest_hi = sdev->target->mgm_hi; fp->mode.wreqb.dest_lo = sdev->target->mgm_lo; fp->mode.wreqb.len = 8; fp->mode.wreqb.extcode = 0; xfer->send.payload[0] = htonl(nid << 16); xfer->send.payload[1] = htonl(ocb->bus_addr & 0xffffffff); fw_asyreq(xfer->fc, -1, xfer); } static void sbp_print_scsi_cmd(struct sbp_ocb *ocb) { struct ccb_scsiio *csio; csio = &ocb->ccb->csio; printf("%s:%d:%jx XPT_SCSI_IO: " "cmd: %02x %02x %02x %02x %02x %02x %02x %02x %02x %02x" ", flags: 0x%02x, " "%db cmd/%db data/%db sense\n", device_get_nameunit(ocb->sdev->target->sbp->fd.dev), ocb->ccb->ccb_h.target_id, (uintmax_t)ocb->ccb->ccb_h.target_lun, csio->cdb_io.cdb_bytes[0], csio->cdb_io.cdb_bytes[1], csio->cdb_io.cdb_bytes[2], csio->cdb_io.cdb_bytes[3], csio->cdb_io.cdb_bytes[4], csio->cdb_io.cdb_bytes[5], csio->cdb_io.cdb_bytes[6], csio->cdb_io.cdb_bytes[7], csio->cdb_io.cdb_bytes[8], csio->cdb_io.cdb_bytes[9], ocb->ccb->ccb_h.flags & CAM_DIR_MASK, csio->cdb_len, csio->dxfer_len, csio->sense_len); } static void sbp_scsi_status(struct sbp_status *sbp_status, struct sbp_ocb *ocb) { struct sbp_cmd_status *sbp_cmd_status; struct scsi_sense_data_fixed *sense; sbp_cmd_status = (struct sbp_cmd_status *)sbp_status->data; sense = (struct scsi_sense_data_fixed *)&ocb->ccb->csio.sense_data; SBP_DEBUG(0) sbp_print_scsi_cmd(ocb); /* XXX need decode status */ printf("%s: SCSI status %x sfmt %x valid %x key %x code %x qlfr %x len %d\n", ocb->sdev->bustgtlun, sbp_cmd_status->status, sbp_cmd_status->sfmt, sbp_cmd_status->valid, sbp_cmd_status->s_key, sbp_cmd_status->s_code, sbp_cmd_status->s_qlfr, sbp_status->len); END_DEBUG switch (sbp_cmd_status->status) { case SCSI_STATUS_CHECK_COND: case SCSI_STATUS_BUSY: case SCSI_STATUS_CMD_TERMINATED: if (sbp_cmd_status->sfmt == SBP_SFMT_CURR) { sense->error_code = SSD_CURRENT_ERROR; } else { sense->error_code = SSD_DEFERRED_ERROR; } if (sbp_cmd_status->valid) sense->error_code |= SSD_ERRCODE_VALID; sense->flags = sbp_cmd_status->s_key; if (sbp_cmd_status->mark) sense->flags |= SSD_FILEMARK; if (sbp_cmd_status->eom) sense->flags |= SSD_EOM; if (sbp_cmd_status->ill_len) sense->flags |= SSD_ILI; bcopy(&sbp_cmd_status->info, &sense->info[0], 4); if (sbp_status->len <= 1) /* XXX not scsi status. shouldn't be happened */ sense->extra_len = 0; else if (sbp_status->len <= 4) /* add_sense_code(_qual), info, cmd_spec_info */ sense->extra_len = 6; else /* fru, sense_key_spec */ sense->extra_len = 10; bcopy(&sbp_cmd_status->cdb, &sense->cmd_spec_info[0], 4); sense->add_sense_code = sbp_cmd_status->s_code; sense->add_sense_code_qual = sbp_cmd_status->s_qlfr; sense->fru = sbp_cmd_status->fru; bcopy(&sbp_cmd_status->s_keydep[0], &sense->sense_key_spec[0], 3); ocb->ccb->csio.scsi_status = sbp_cmd_status->status; ocb->ccb->ccb_h.status = CAM_SCSI_STATUS_ERROR | CAM_AUTOSNS_VALID; /* { uint8_t j, *tmp; tmp = sense; for (j = 0; j < 32; j += 8) { printf("sense %02x%02x %02x%02x %02x%02x %02x%02x\n", tmp[j], tmp[j + 1], tmp[j + 2], tmp[j + 3], tmp[j + 4], tmp[j + 5], tmp[j + 6], tmp[j + 7]); } } */ break; default: device_printf(ocb->sdev->target->sbp->fd.dev, "%s:%s unknown scsi status 0x%x\n", __func__, ocb->sdev->bustgtlun, sbp_cmd_status->status); } } static void sbp_fix_inq_data(struct sbp_ocb *ocb) { union ccb *ccb; struct sbp_dev *sdev; struct scsi_inquiry_data *inq; ccb = ocb->ccb; sdev = ocb->sdev; if (ccb->csio.cdb_io.cdb_bytes[1] & SI_EVPD) return; SBP_DEBUG(1) device_printf(sdev->target->sbp->fd.dev, "%s:%s\n", __func__, sdev->bustgtlun); END_DEBUG inq = (struct scsi_inquiry_data *) ccb->csio.data_ptr; switch (SID_TYPE(inq)) { case T_DIRECT: #if 0 /* * XXX Convert Direct Access device to RBC. * I've never seen FireWire DA devices which support READ_6. */ if (SID_TYPE(inq) == T_DIRECT) inq->device |= T_RBC; /* T_DIRECT == 0 */ #endif /* fall through */ case T_RBC: /* * Override vendor/product/revision information. * Some devices sometimes return strange strings. */ #if 1 bcopy(sdev->vendor, inq->vendor, sizeof(inq->vendor)); bcopy(sdev->product, inq->product, sizeof(inq->product)); bcopy(sdev->revision + 2, inq->revision, sizeof(inq->revision)); #endif break; } /* * Force to enable/disable tagged queuing. * XXX CAM also checks SCP_QUEUE_DQUE flag in the control mode page. */ if (sbp_tags > 0) inq->flags |= SID_CmdQue; else if (sbp_tags < 0) inq->flags &= ~SID_CmdQue; } static void sbp_recv1(struct fw_xfer *xfer) { struct fw_pkt *rfp; #if NEED_RESPONSE struct fw_pkt *sfp; #endif struct sbp_softc *sbp; struct sbp_dev *sdev; struct sbp_ocb *ocb; struct sbp_login_res *login_res = NULL; struct sbp_status *sbp_status; struct sbp_target *target; int orb_fun, status_valid0, status_valid, t, l, reset_agent = 0; uint32_t addr; /* uint32_t *ld; ld = xfer->recv.buf; printf("sbp %x %d %d %08x %08x %08x %08x\n", xfer->resp, xfer->recv.len, xfer->recv.off, ntohl(ld[0]), ntohl(ld[1]), ntohl(ld[2]), ntohl(ld[3])); printf("sbp %08x %08x %08x %08x\n", ntohl(ld[4]), ntohl(ld[5]), ntohl(ld[6]), ntohl(ld[7])); printf("sbp %08x %08x %08x %08x\n", ntohl(ld[8]), ntohl(ld[9]), ntohl(ld[10]), ntohl(ld[11])); */ sbp = (struct sbp_softc *)xfer->sc; SBP_LOCK_ASSERT(sbp); if (xfer->resp != 0) { printf("sbp_recv: xfer->resp = %d\n", xfer->resp); goto done0; } if (xfer->recv.payload == NULL) { printf("sbp_recv: xfer->recv.payload == NULL\n"); goto done0; } rfp = &xfer->recv.hdr; if (rfp->mode.wreqb.tcode != FWTCODE_WREQB) { printf("sbp_recv: tcode = %d\n", rfp->mode.wreqb.tcode); goto done0; } sbp_status = (struct sbp_status *)xfer->recv.payload; addr = rfp->mode.wreqb.dest_lo; SBP_DEBUG(2) printf("received address 0x%x\n", addr); END_DEBUG t = SBP_ADDR2TRG(addr); if (t >= SBP_NUM_TARGETS) { device_printf(sbp->fd.dev, "sbp_recv1: invalid target %d\n", t); goto done0; } target = &sbp->targets[t]; l = SBP_ADDR2LUN(addr); if (l >= target->num_lun || target->luns[l] == NULL) { device_printf(sbp->fd.dev, "sbp_recv1: invalid lun %d (target=%d)\n", l, t); goto done0; } sdev = target->luns[l]; ocb = NULL; switch (sbp_status->src) { case 0: case 1: /* check mgm_ocb_cur first */ ocb = target->mgm_ocb_cur; if (ocb != NULL) { if (OCB_MATCH(ocb, sbp_status)) { callout_stop(&target->mgm_ocb_timeout); target->mgm_ocb_cur = NULL; break; } } ocb = sbp_dequeue_ocb(sdev, sbp_status); if (ocb == NULL) { device_printf(sdev->target->sbp->fd.dev, "%s:%s No ocb(%x) on the queue\n", __func__,sdev->bustgtlun, ntohl(sbp_status->orb_lo)); } break; case 2: /* unsolicit */ device_printf(sdev->target->sbp->fd.dev, "%s:%s unsolicit status received\n", __func__, sdev->bustgtlun); break; default: device_printf(sdev->target->sbp->fd.dev, "%s:%s unknown sbp_status->src\n", __func__, sdev->bustgtlun); } status_valid0 = (sbp_status->src < 2 && sbp_status->resp == ORB_RES_CMPL && sbp_status->dead == 0); status_valid = (status_valid0 && sbp_status->status == 0); if (!status_valid0 || debug > 2) { int status; SBP_DEBUG(0) device_printf(sdev->target->sbp->fd.dev, "%s:%s ORB status src:%x resp:%x dead:%x" " len:%x stat:%x orb:%x%08x\n", __func__, sdev->bustgtlun, sbp_status->src, sbp_status->resp, sbp_status->dead, sbp_status->len, sbp_status->status, ntohs(sbp_status->orb_hi), ntohl(sbp_status->orb_lo)); END_DEBUG device_printf(sdev->target->sbp->fd.dev, "%s\n", sdev->bustgtlun); status = sbp_status->status; switch (sbp_status->resp) { case 0: if (status > MAX_ORB_STATUS0) printf("%s\n", orb_status0[MAX_ORB_STATUS0]); else printf("%s\n", orb_status0[status]); break; case 1: printf("Obj: %s, Error: %s\n", orb_status1_object[(status >> 6) & 3], orb_status1_serial_bus_error[status & 0xf]); break; case 2: printf("Illegal request\n"); break; case 3: printf("Vendor dependent\n"); break; default: printf("unknown respose code %d\n", sbp_status->resp); } } /* we have to reset the fetch agent if it's dead */ if (sbp_status->dead) { if (sdev->path) { xpt_freeze_devq(sdev->path, 1); sdev->freeze++; } reset_agent = 1; } if (ocb == NULL) goto done; switch (ntohl(ocb->orb[4]) & ORB_FMT_MSK) { case ORB_FMT_NOP: break; case ORB_FMT_VED: break; case ORB_FMT_STD: switch (ocb->flags) { case OCB_ACT_MGM: orb_fun = ntohl(ocb->orb[4]) & ORB_FUN_MSK; reset_agent = 0; switch (orb_fun) { case ORB_FUN_LGI: fwdma_sync(&sdev->dma, BUS_DMASYNC_POSTREAD); login_res = sdev->login; login_res->len = ntohs(login_res->len); login_res->id = ntohs(login_res->id); login_res->cmd_hi = ntohs(login_res->cmd_hi); login_res->cmd_lo = ntohl(login_res->cmd_lo); if (status_valid) { SBP_DEBUG(0) device_printf(sdev->target->sbp->fd.dev, "%s:%s login: len %d, ID %d, cmd %08x%08x, recon_hold %d\n", __func__, sdev->bustgtlun, login_res->len, login_res->id, login_res->cmd_hi, login_res->cmd_lo, ntohs(login_res->recon_hold)); END_DEBUG sbp_busy_timeout(sdev); } else { /* forgot logout? */ device_printf(sdev->target->sbp->fd.dev, "%s:%s login failed\n", __func__, sdev->bustgtlun); sdev->status = SBP_DEV_RESET; } break; case ORB_FUN_RCN: login_res = sdev->login; if (status_valid) { SBP_DEBUG(0) device_printf(sdev->target->sbp->fd.dev, "%s:%s reconnect: len %d, ID %d, cmd %08x%08x\n", __func__, sdev->bustgtlun, login_res->len, login_res->id, login_res->cmd_hi, login_res->cmd_lo); END_DEBUG if (sdev->status == SBP_DEV_ATTACHED) sbp_scan_dev(sdev); else sbp_agent_reset(sdev); } else { /* reconnection hold time exceed? */ SBP_DEBUG(0) device_printf(sdev->target->sbp->fd.dev, "%s:%s reconnect failed\n", __func__, sdev->bustgtlun); END_DEBUG sbp_login(sdev); } break; case ORB_FUN_LGO: sdev->status = SBP_DEV_RESET; break; case ORB_FUN_RST: sbp_busy_timeout(sdev); break; case ORB_FUN_LUR: case ORB_FUN_ATA: case ORB_FUN_ATS: sbp_agent_reset(sdev); break; default: device_printf(sdev->target->sbp->fd.dev, "%s:%s unknown function %d\n", __func__, sdev->bustgtlun, orb_fun); break; } sbp_mgm_orb(sdev, ORB_FUN_RUNQUEUE, NULL); break; case OCB_ACT_CMD: sdev->timeout = 0; if (ocb->ccb != NULL) { union ccb *ccb; ccb = ocb->ccb; if (sbp_status->len > 1) { sbp_scsi_status(sbp_status, ocb); } else { if (sbp_status->resp != ORB_RES_CMPL) { ccb->ccb_h.status = CAM_REQ_CMP_ERR; } else { ccb->ccb_h.status = CAM_REQ_CMP; } } /* fix up inq data */ if (ccb->csio.cdb_io.cdb_bytes[0] == INQUIRY) sbp_fix_inq_data(ocb); xpt_done(ccb); } break; default: break; } } if (!use_doorbell) sbp_free_ocb(sdev, ocb); done: if (reset_agent) sbp_agent_reset(sdev); done0: xfer->recv.pay_len = SBP_RECV_LEN; /* The received packet is usually small enough to be stored within * the buffer. In that case, the controller return ack_complete and * no respose is necessary. * * XXX fwohci.c and firewire.c should inform event_code such as * ack_complete or ack_pending to upper driver. */ #if NEED_RESPONSE xfer->send.off = 0; sfp = (struct fw_pkt *)xfer->send.buf; sfp->mode.wres.dst = rfp->mode.wreqb.src; xfer->dst = sfp->mode.wres.dst; xfer->spd = min(sdev->target->fwdev->speed, max_speed); xfer->hand = sbp_loginres_callback; sfp->mode.wres.tlrt = rfp->mode.wreqb.tlrt; sfp->mode.wres.tcode = FWTCODE_WRES; sfp->mode.wres.rtcode = 0; sfp->mode.wres.pri = 0; fw_asyreq(xfer->fc, -1, xfer); #else /* recycle */ STAILQ_INSERT_TAIL(&sbp->fwb.xferlist, xfer, link); #endif } static void sbp_recv(struct fw_xfer *xfer) { struct sbp_softc *sbp; sbp = (struct sbp_softc *)xfer->sc; SBP_LOCK(sbp); sbp_recv1(xfer); SBP_UNLOCK(sbp); } /* * sbp_attach() */ static int sbp_attach(device_t dev) { struct sbp_softc *sbp; struct cam_devq *devq; struct firewire_comm *fc; int i, error; if (DFLTPHYS > SBP_MAXPHYS) device_printf(dev, "Warning, DFLTPHYS(%dKB) is larger than " "SBP_MAXPHYS(%dKB).\n", DFLTPHYS / 1024, SBP_MAXPHYS / 1024); if (!firewire_phydma_enable) device_printf(dev, "Warning, hw.firewire.phydma_enable must be 1 " "for SBP over FireWire.\n"); SBP_DEBUG(0) printf("sbp_attach (cold=%d)\n", cold); END_DEBUG if (cold) sbp_cold++; sbp = device_get_softc(dev); sbp->fd.dev = dev; sbp->fd.fc = fc = device_get_ivars(dev); mtx_init(&sbp->mtx, "sbp", NULL, MTX_DEF); if (max_speed < 0) max_speed = fc->speed; error = bus_dma_tag_create(/*parent*/fc->dmat, /* XXX shoud be 4 for sane backend? */ /*alignment*/1, /*boundary*/0, /*lowaddr*/BUS_SPACE_MAXADDR_32BIT, /*highaddr*/BUS_SPACE_MAXADDR, /*filter*/NULL, /*filterarg*/NULL, /*maxsize*/0x100000, /*nsegments*/SBP_IND_MAX, /*maxsegsz*/SBP_SEG_MAX, /*flags*/BUS_DMA_ALLOCNOW, /*lockfunc*/busdma_lock_mutex, /*lockarg*/&sbp->mtx, &sbp->dmat); if (error != 0) { printf("sbp_attach: Could not allocate DMA tag " "- error %d\n", error); return (ENOMEM); } devq = cam_simq_alloc(/*maxopenings*/SBP_NUM_OCB); if (devq == NULL) return (ENXIO); for (i = 0; i < SBP_NUM_TARGETS; i++) { sbp->targets[i].fwdev = NULL; sbp->targets[i].luns = NULL; sbp->targets[i].sbp = sbp; } sbp->sim = cam_sim_alloc(sbp_action, sbp_poll, "sbp", sbp, device_get_unit(dev), &sbp->mtx, /*untagged*/ 1, /*tagged*/ SBP_QUEUE_LEN - 1, devq); if (sbp->sim == NULL) { cam_simq_free(devq); return (ENXIO); } SBP_LOCK(sbp); if (xpt_bus_register(sbp->sim, dev, /*bus*/0) != CAM_SUCCESS) goto fail; if (xpt_create_path(&sbp->path, NULL, cam_sim_path(sbp->sim), CAM_TARGET_WILDCARD, CAM_LUN_WILDCARD) != CAM_REQ_CMP) { xpt_bus_deregister(cam_sim_path(sbp->sim)); goto fail; } SBP_UNLOCK(sbp); /* We reserve 16 bit space (4 bytes X 64 targets X 256 luns) */ sbp->fwb.start = ((u_int64_t)SBP_BIND_HI << 32) | SBP_DEV2ADDR(0, 0); sbp->fwb.end = sbp->fwb.start + 0xffff; /* pre-allocate xfer */ STAILQ_INIT(&sbp->fwb.xferlist); fw_xferlist_add(&sbp->fwb.xferlist, M_SBP, /*send*/ 0, /*recv*/ SBP_RECV_LEN, SBP_NUM_OCB/2, fc, (void *)sbp, sbp_recv); fw_bindadd(fc, &sbp->fwb); sbp->fd.post_busreset = sbp_post_busreset; sbp->fd.post_explore = sbp_post_explore; if (fc->status != -1) { sbp_post_busreset((void *)sbp); sbp_post_explore((void *)sbp); } SBP_LOCK(sbp); xpt_async(AC_BUS_RESET, sbp->path, /*arg*/ NULL); SBP_UNLOCK(sbp); return (0); fail: SBP_UNLOCK(sbp); cam_sim_free(sbp->sim, /*free_devq*/TRUE); return (ENXIO); } static int sbp_logout_all(struct sbp_softc *sbp) { struct sbp_target *target; struct sbp_dev *sdev; int i, j; SBP_DEBUG(0) printf("sbp_logout_all\n"); END_DEBUG SBP_LOCK_ASSERT(sbp); for (i = 0; i < SBP_NUM_TARGETS; i++) { target = &sbp->targets[i]; if (target->luns == NULL) continue; for (j = 0; j < target->num_lun; j++) { sdev = target->luns[j]; if (sdev == NULL) continue; callout_stop(&sdev->login_callout); if (sdev->status >= SBP_DEV_TOATTACH && sdev->status <= SBP_DEV_ATTACHED) sbp_mgm_orb(sdev, ORB_FUN_LGO, NULL); } } return 0; } static int sbp_shutdown(device_t dev) { struct sbp_softc *sbp = ((struct sbp_softc *)device_get_softc(dev)); SBP_LOCK(sbp); sbp_logout_all(sbp); SBP_UNLOCK(sbp); return (0); } static void sbp_free_sdev(struct sbp_dev *sdev) { struct sbp_softc *sbp; int i; if (sdev == NULL) return; sbp = sdev->target->sbp; SBP_UNLOCK(sbp); callout_drain(&sdev->login_callout); for (i = 0; i < SBP_QUEUE_LEN; i++) { callout_drain(&sdev->ocb[i].timer); bus_dmamap_destroy(sbp->dmat, sdev->ocb[i].dmamap); } fwdma_free(sbp->fd.fc, &sdev->dma); free(sdev, M_SBP); SBP_LOCK(sbp); } static void sbp_free_target(struct sbp_target *target) { struct sbp_softc *sbp; struct fw_xfer *xfer, *next; int i; if (target->luns == NULL) return; sbp = target->sbp; SBP_LOCK_ASSERT(sbp); SBP_UNLOCK(sbp); callout_drain(&target->mgm_ocb_timeout); callout_drain(&target->scan_callout); SBP_LOCK(sbp); for (i = 0; i < target->num_lun; i++) sbp_free_sdev(target->luns[i]); STAILQ_FOREACH_SAFE(xfer, &target->xferlist, link, next) { fw_xfer_free_buf(xfer); } STAILQ_INIT(&target->xferlist); free(target->luns, M_SBP); target->num_lun = 0; target->luns = NULL; target->fwdev = NULL; } static int sbp_detach(device_t dev) { struct sbp_softc *sbp = ((struct sbp_softc *)device_get_softc(dev)); struct firewire_comm *fc = sbp->fd.fc; int i; SBP_DEBUG(0) printf("sbp_detach\n"); END_DEBUG SBP_LOCK(sbp); for (i = 0; i < SBP_NUM_TARGETS; i++) sbp_cam_detach_target(&sbp->targets[i]); xpt_async(AC_LOST_DEVICE, sbp->path, NULL); xpt_free_path(sbp->path); xpt_bus_deregister(cam_sim_path(sbp->sim)); cam_sim_free(sbp->sim, /*free_devq*/ TRUE); sbp_logout_all(sbp); SBP_UNLOCK(sbp); /* XXX wait for logout completion */ pause("sbpdtc", hz/2); SBP_LOCK(sbp); for (i = 0; i < SBP_NUM_TARGETS; i++) sbp_free_target(&sbp->targets[i]); SBP_UNLOCK(sbp); fw_bindremove(fc, &sbp->fwb); fw_xferlist_remove(&sbp->fwb.xferlist); bus_dma_tag_destroy(sbp->dmat); mtx_destroy(&sbp->mtx); return (0); } static void sbp_cam_detach_sdev(struct sbp_dev *sdev) { if (sdev == NULL) return; if (sdev->status == SBP_DEV_DEAD) return; if (sdev->status == SBP_DEV_RESET) return; SBP_LOCK_ASSERT(sdev->target->sbp); sbp_abort_all_ocbs(sdev, CAM_DEV_NOT_THERE); if (sdev->path) { xpt_release_devq(sdev->path, sdev->freeze, TRUE); sdev->freeze = 0; xpt_async(AC_LOST_DEVICE, sdev->path, NULL); xpt_free_path(sdev->path); sdev->path = NULL; } } static void sbp_cam_detach_target(struct sbp_target *target) { int i; SBP_LOCK_ASSERT(target->sbp); if (target->luns != NULL) { SBP_DEBUG(0) printf("sbp_detach_target %d\n", target->target_id); END_DEBUG callout_stop(&target->scan_callout); for (i = 0; i < target->num_lun; i++) sbp_cam_detach_sdev(target->luns[i]); } } static void sbp_target_reset(struct sbp_dev *sdev, int method) { int i; struct sbp_target *target = sdev->target; struct sbp_dev *tsdev; SBP_LOCK_ASSERT(target->sbp); for (i = 0; i < target->num_lun; i++) { tsdev = target->luns[i]; if (tsdev == NULL) continue; if (tsdev->status == SBP_DEV_DEAD) continue; if (tsdev->status == SBP_DEV_RESET) continue; xpt_freeze_devq(tsdev->path, 1); tsdev->freeze++; sbp_abort_all_ocbs(tsdev, CAM_CMD_TIMEOUT); if (method == 2) tsdev->status = SBP_DEV_LOGIN; } switch (method) { case 1: printf("target reset\n"); sbp_mgm_orb(sdev, ORB_FUN_RST, NULL); break; case 2: printf("reset start\n"); sbp_reset_start(sdev); break; } } static void sbp_mgm_timeout(void *arg) { struct sbp_ocb *ocb = (struct sbp_ocb *)arg; struct sbp_dev *sdev = ocb->sdev; struct sbp_target *target = sdev->target; SBP_LOCK_ASSERT(target->sbp); device_printf(sdev->target->sbp->fd.dev, "%s:%s request timeout(mgm orb:0x%08x)\n", __func__, sdev->bustgtlun, (uint32_t)ocb->bus_addr); target->mgm_ocb_cur = NULL; sbp_free_ocb(sdev, ocb); #if 0 /* XXX */ printf("run next request\n"); sbp_mgm_orb(sdev, ORB_FUN_RUNQUEUE, NULL); #endif device_printf(sdev->target->sbp->fd.dev, "%s:%s reset start\n", __func__, sdev->bustgtlun); sbp_reset_start(sdev); } static void sbp_timeout(void *arg) { struct sbp_ocb *ocb = (struct sbp_ocb *)arg; struct sbp_dev *sdev = ocb->sdev; device_printf(sdev->target->sbp->fd.dev, "%s:%s request timeout(cmd orb:0x%08x) ... ", __func__, sdev->bustgtlun, (uint32_t)ocb->bus_addr); SBP_LOCK_ASSERT(sdev->target->sbp); sdev->timeout++; switch (sdev->timeout) { case 1: printf("agent reset\n"); xpt_freeze_devq(sdev->path, 1); sdev->freeze++; sbp_abort_all_ocbs(sdev, CAM_CMD_TIMEOUT); sbp_agent_reset(sdev); break; case 2: case 3: sbp_target_reset(sdev, sdev->timeout - 1); break; #if 0 default: /* XXX give up */ sbp_cam_detach_target(target); if (target->luns != NULL) free(target->luns, M_SBP); target->num_lun = 0; target->luns = NULL; target->fwdev = NULL; #endif } } static void sbp_action(struct cam_sim *sim, union ccb *ccb) { struct sbp_softc *sbp = (struct sbp_softc *)sim->softc; struct sbp_target *target = NULL; struct sbp_dev *sdev = NULL; if (sbp != NULL) SBP_LOCK_ASSERT(sbp); /* target:lun -> sdev mapping */ if (sbp != NULL && ccb->ccb_h.target_id != CAM_TARGET_WILDCARD && ccb->ccb_h.target_id < SBP_NUM_TARGETS) { target = &sbp->targets[ccb->ccb_h.target_id]; if (target->fwdev != NULL && ccb->ccb_h.target_lun != CAM_LUN_WILDCARD && ccb->ccb_h.target_lun < target->num_lun) { sdev = target->luns[ccb->ccb_h.target_lun]; if (sdev != NULL && sdev->status != SBP_DEV_ATTACHED && sdev->status != SBP_DEV_PROBE) sdev = NULL; } } SBP_DEBUG(1) if (sdev == NULL) printf("invalid target %d lun %jx\n", ccb->ccb_h.target_id, (uintmax_t)ccb->ccb_h.target_lun); END_DEBUG switch (ccb->ccb_h.func_code) { case XPT_SCSI_IO: case XPT_RESET_DEV: case XPT_GET_TRAN_SETTINGS: case XPT_SET_TRAN_SETTINGS: case XPT_CALC_GEOMETRY: if (sdev == NULL) { SBP_DEBUG(1) printf("%s:%d:%jx:func_code 0x%04x: " "Invalid target (target needed)\n", device_get_nameunit(sbp->fd.dev), ccb->ccb_h.target_id, (uintmax_t)ccb->ccb_h.target_lun, ccb->ccb_h.func_code); END_DEBUG ccb->ccb_h.status = CAM_DEV_NOT_THERE; xpt_done(ccb); return; } break; case XPT_PATH_INQ: case XPT_NOOP: /* The opcodes sometimes aimed at a target (sc is valid), * sometimes aimed at the SIM (sc is invalid and target is * CAM_TARGET_WILDCARD) */ if (sbp == NULL && ccb->ccb_h.target_id != CAM_TARGET_WILDCARD) { SBP_DEBUG(0) printf("%s:%d:%jx func_code 0x%04x: " "Invalid target (no wildcard)\n", device_get_nameunit(sbp->fd.dev), ccb->ccb_h.target_id, (uintmax_t)ccb->ccb_h.target_lun, ccb->ccb_h.func_code); END_DEBUG ccb->ccb_h.status = CAM_DEV_NOT_THERE; xpt_done(ccb); return; } break; default: /* XXX Hm, we should check the input parameters */ break; } switch (ccb->ccb_h.func_code) { case XPT_SCSI_IO: { struct ccb_scsiio *csio; struct sbp_ocb *ocb; int speed; void *cdb; csio = &ccb->csio; mtx_assert(sim->mtx, MA_OWNED); SBP_DEBUG(2) printf("%s:%d:%jx XPT_SCSI_IO: " "cmd: %02x %02x %02x %02x %02x %02x %02x %02x %02x %02x" ", flags: 0x%02x, " "%db cmd/%db data/%db sense\n", device_get_nameunit(sbp->fd.dev), ccb->ccb_h.target_id, (uintmax_t)ccb->ccb_h.target_lun, csio->cdb_io.cdb_bytes[0], csio->cdb_io.cdb_bytes[1], csio->cdb_io.cdb_bytes[2], csio->cdb_io.cdb_bytes[3], csio->cdb_io.cdb_bytes[4], csio->cdb_io.cdb_bytes[5], csio->cdb_io.cdb_bytes[6], csio->cdb_io.cdb_bytes[7], csio->cdb_io.cdb_bytes[8], csio->cdb_io.cdb_bytes[9], ccb->ccb_h.flags & CAM_DIR_MASK, csio->cdb_len, csio->dxfer_len, csio->sense_len); END_DEBUG if (sdev == NULL) { ccb->ccb_h.status = CAM_DEV_NOT_THERE; xpt_done(ccb); return; } #if 0 /* if we are in probe stage, pass only probe commands */ if (sdev->status == SBP_DEV_PROBE) { char *name; name = xpt_path_periph(ccb->ccb_h.path)->periph_name; printf("probe stage, periph name: %s\n", name); if (strcmp(name, "probe") != 0) { ccb->ccb_h.status = CAM_REQUEUE_REQ; xpt_done(ccb); return; } } #endif if ((ocb = sbp_get_ocb(sdev)) == NULL) { ccb->ccb_h.status = CAM_RESRC_UNAVAIL; if (sdev->freeze == 0) { xpt_freeze_devq(sdev->path, 1); sdev->freeze++; } xpt_done(ccb); return; } ocb->flags = OCB_ACT_CMD; ocb->sdev = sdev; ocb->ccb = ccb; ccb->ccb_h.ccb_sdev_ptr = sdev; ocb->orb[0] = htonl(1U << 31); ocb->orb[1] = 0; ocb->orb[2] = htonl(((sbp->fd.fc->nodeid | FWLOCALBUS) << 16)); ocb->orb[3] = htonl(ocb->bus_addr + IND_PTR_OFFSET); speed = min(target->fwdev->speed, max_speed); ocb->orb[4] = htonl(ORB_NOTIFY | ORB_CMD_SPD(speed) | ORB_CMD_MAXP(speed + 7)); if ((ccb->ccb_h.flags & CAM_DIR_MASK) == CAM_DIR_IN) { ocb->orb[4] |= htonl(ORB_CMD_IN); } if (csio->ccb_h.flags & CAM_CDB_POINTER) cdb = (void *)csio->cdb_io.cdb_ptr; else cdb = (void *)&csio->cdb_io.cdb_bytes; bcopy(cdb, (void *)&ocb->orb[5], csio->cdb_len); /* printf("ORB %08x %08x %08x %08x\n", ntohl(ocb->orb[0]), ntohl(ocb->orb[1]), ntohl(ocb->orb[2]), ntohl(ocb->orb[3])); printf("ORB %08x %08x %08x %08x\n", ntohl(ocb->orb[4]), ntohl(ocb->orb[5]), ntohl(ocb->orb[6]), ntohl(ocb->orb[7])); */ if (ccb->csio.dxfer_len > 0) { int error; error = bus_dmamap_load_ccb(/*dma tag*/sbp->dmat, /*dma map*/ocb->dmamap, ccb, sbp_execute_ocb, ocb, /*flags*/0); if (error) printf("sbp: bus_dmamap_load error %d\n", error); } else sbp_execute_ocb(ocb, NULL, 0, 0); break; } case XPT_CALC_GEOMETRY: { struct ccb_calc_geometry *ccg; ccg = &ccb->ccg; if (ccg->block_size == 0) { printf("sbp_action: block_size is 0.\n"); ccb->ccb_h.status = CAM_REQ_INVALID; xpt_done(ccb); break; } SBP_DEBUG(1) printf("%s:%d:%d:%jx:XPT_CALC_GEOMETRY: " "Volume size = %jd\n", device_get_nameunit(sbp->fd.dev), cam_sim_path(sbp->sim), ccb->ccb_h.target_id, (uintmax_t)ccb->ccb_h.target_lun, (uintmax_t)ccg->volume_size); END_DEBUG cam_calc_geometry(ccg, /*extended*/1); xpt_done(ccb); break; } case XPT_RESET_BUS: /* Reset the specified SCSI bus */ { SBP_DEBUG(1) printf("%s:%d:XPT_RESET_BUS: \n", device_get_nameunit(sbp->fd.dev), cam_sim_path(sbp->sim)); END_DEBUG ccb->ccb_h.status = CAM_REQ_INVALID; xpt_done(ccb); break; } case XPT_PATH_INQ: /* Path routing inquiry */ { struct ccb_pathinq *cpi = &ccb->cpi; SBP_DEBUG(1) printf("%s:%d:%jx XPT_PATH_INQ:.\n", device_get_nameunit(sbp->fd.dev), ccb->ccb_h.target_id, (uintmax_t)ccb->ccb_h.target_lun); END_DEBUG cpi->version_num = 1; /* XXX??? */ cpi->hba_inquiry = PI_TAG_ABLE; cpi->target_sprt = 0; cpi->hba_misc = PIM_NOBUSRESET | PIM_NO_6_BYTE; cpi->hba_eng_cnt = 0; cpi->max_target = SBP_NUM_TARGETS - 1; cpi->max_lun = SBP_NUM_LUNS - 1; cpi->initiator_id = SBP_INITIATOR; cpi->bus_id = sim->bus_id; cpi->base_transfer_speed = 400 * 1000 / 8; strlcpy(cpi->sim_vid, "FreeBSD", SIM_IDLEN); strlcpy(cpi->hba_vid, "SBP", HBA_IDLEN); strlcpy(cpi->dev_name, sim->sim_name, DEV_IDLEN); cpi->unit_number = sim->unit_number; - cpi->transport = XPORT_SPI; /* XX should have a FireWire */ - cpi->transport_version = 2; - cpi->protocol = PROTO_SCSI; - cpi->protocol_version = SCSI_REV_2; + cpi->transport = XPORT_SPI; /* XX should have a FireWire */ + cpi->transport_version = 2; + cpi->protocol = PROTO_SCSI; + cpi->protocol_version = SCSI_REV_2; cpi->ccb_h.status = CAM_REQ_CMP; xpt_done(ccb); break; } case XPT_GET_TRAN_SETTINGS: { struct ccb_trans_settings *cts = &ccb->cts; struct ccb_trans_settings_scsi *scsi = &cts->proto_specific.scsi; struct ccb_trans_settings_spi *spi = &cts->xport_specific.spi; cts->protocol = PROTO_SCSI; cts->protocol_version = SCSI_REV_2; cts->transport = XPORT_SPI; /* should have a FireWire */ cts->transport_version = 2; spi->valid = CTS_SPI_VALID_DISC; spi->flags = CTS_SPI_FLAGS_DISC_ENB; scsi->valid = CTS_SCSI_VALID_TQ; scsi->flags = CTS_SCSI_FLAGS_TAG_ENB; SBP_DEBUG(1) printf("%s:%d:%jx XPT_GET_TRAN_SETTINGS:.\n", device_get_nameunit(sbp->fd.dev), ccb->ccb_h.target_id, (uintmax_t)ccb->ccb_h.target_lun); END_DEBUG cts->ccb_h.status = CAM_REQ_CMP; xpt_done(ccb); break; } case XPT_ABORT: ccb->ccb_h.status = CAM_UA_ABORT; xpt_done(ccb); break; case XPT_SET_TRAN_SETTINGS: /* XXX */ default: ccb->ccb_h.status = CAM_REQ_INVALID; xpt_done(ccb); break; } return; } static void sbp_execute_ocb(void *arg, bus_dma_segment_t *segments, int seg, int error) { int i; struct sbp_ocb *ocb; struct sbp_ocb *prev; bus_dma_segment_t *s; if (error) printf("sbp_execute_ocb: error=%d\n", error); ocb = (struct sbp_ocb *)arg; SBP_DEBUG(2) printf("sbp_execute_ocb: seg %d", seg); for (i = 0; i < seg; i++) printf(", %jx:%jd", (uintmax_t)segments[i].ds_addr, (uintmax_t)segments[i].ds_len); printf("\n"); END_DEBUG if (seg == 1) { /* direct pointer */ s = &segments[0]; if (s->ds_len > SBP_SEG_MAX) panic("ds_len > SBP_SEG_MAX, fix busdma code"); ocb->orb[3] = htonl(s->ds_addr); ocb->orb[4] |= htonl(s->ds_len); } else if (seg > 1) { /* page table */ for (i = 0; i < seg; i++) { s = &segments[i]; SBP_DEBUG(0) /* XXX LSI Logic "< 16 byte" bug might be hit */ if (s->ds_len < 16) printf("sbp_execute_ocb: warning, " "segment length(%zd) is less than 16." "(seg=%d/%d)\n", (size_t)s->ds_len, i + 1, seg); END_DEBUG if (s->ds_len > SBP_SEG_MAX) panic("ds_len > SBP_SEG_MAX, fix busdma code"); ocb->ind_ptr[i].hi = htonl(s->ds_len << 16); ocb->ind_ptr[i].lo = htonl(s->ds_addr); } ocb->orb[4] |= htonl(ORB_CMD_PTBL | seg); } if (seg > 0) bus_dmamap_sync(ocb->sdev->target->sbp->dmat, ocb->dmamap, (ntohl(ocb->orb[4]) & ORB_CMD_IN) ? BUS_DMASYNC_PREREAD : BUS_DMASYNC_PREWRITE); prev = sbp_enqueue_ocb(ocb->sdev, ocb); fwdma_sync(&ocb->sdev->dma, BUS_DMASYNC_PREWRITE); if (use_doorbell) { if (prev == NULL) { if (ocb->sdev->last_ocb != NULL) sbp_doorbell(ocb->sdev); else sbp_orb_pointer(ocb->sdev, ocb); } } else { if (prev == NULL || (ocb->sdev->flags & ORB_LINK_DEAD) != 0) { ocb->sdev->flags &= ~ORB_LINK_DEAD; sbp_orb_pointer(ocb->sdev, ocb); } } } static void sbp_poll(struct cam_sim *sim) { struct sbp_softc *sbp; struct firewire_comm *fc; sbp = (struct sbp_softc *)sim->softc; fc = sbp->fd.fc; fc->poll(fc, 0, -1); return; } static struct sbp_ocb * sbp_dequeue_ocb(struct sbp_dev *sdev, struct sbp_status *sbp_status) { struct sbp_ocb *ocb; struct sbp_ocb *next; int order = 0; SBP_DEBUG(1) device_printf(sdev->target->sbp->fd.dev, "%s:%s 0x%08x src %d\n", __func__, sdev->bustgtlun, ntohl(sbp_status->orb_lo), sbp_status->src); END_DEBUG SBP_LOCK_ASSERT(sdev->target->sbp); STAILQ_FOREACH_SAFE(ocb, &sdev->ocbs, ocb, next) { if (OCB_MATCH(ocb, sbp_status)) { /* found */ STAILQ_REMOVE(&sdev->ocbs, ocb, sbp_ocb, ocb); if (ocb->ccb != NULL) callout_stop(&ocb->timer); if (ntohl(ocb->orb[4]) & 0xffff) { bus_dmamap_sync(sdev->target->sbp->dmat, ocb->dmamap, (ntohl(ocb->orb[4]) & ORB_CMD_IN) ? BUS_DMASYNC_POSTREAD : BUS_DMASYNC_POSTWRITE); bus_dmamap_unload(sdev->target->sbp->dmat, ocb->dmamap); } if (!use_doorbell) { if (sbp_status->src == SRC_NO_NEXT) { if (next != NULL) sbp_orb_pointer(sdev, next); else if (order > 0) { /* * Unordered execution * We need to send pointer for * next ORB */ sdev->flags |= ORB_LINK_DEAD; } } } else { /* * XXX this is not correct for unordered * execution. */ if (sdev->last_ocb != NULL) { sbp_free_ocb(sdev, sdev->last_ocb); } sdev->last_ocb = ocb; if (next != NULL && sbp_status->src == SRC_NO_NEXT) sbp_doorbell(sdev); } break; } else order++; } SBP_DEBUG(0) if (ocb && order > 0) { device_printf(sdev->target->sbp->fd.dev, "%s:%s unordered execution order:%d\n", __func__, sdev->bustgtlun, order); } END_DEBUG return (ocb); } static struct sbp_ocb * sbp_enqueue_ocb(struct sbp_dev *sdev, struct sbp_ocb *ocb) { struct sbp_ocb *prev, *prev2; SBP_LOCK_ASSERT(sdev->target->sbp); SBP_DEBUG(1) device_printf(sdev->target->sbp->fd.dev, "%s:%s 0x%08jx\n", __func__, sdev->bustgtlun, (uintmax_t)ocb->bus_addr); END_DEBUG prev2 = prev = STAILQ_LAST(&sdev->ocbs, sbp_ocb, ocb); STAILQ_INSERT_TAIL(&sdev->ocbs, ocb, ocb); if (ocb->ccb != NULL) { callout_reset_sbt(&ocb->timer, SBT_1MS * ocb->ccb->ccb_h.timeout, 0, sbp_timeout, ocb, 0); } if (use_doorbell && prev == NULL) prev2 = sdev->last_ocb; if (prev2 != NULL && (ocb->sdev->flags & ORB_LINK_DEAD) == 0) { SBP_DEBUG(1) printf("linking chain 0x%jx -> 0x%jx\n", (uintmax_t)prev2->bus_addr, (uintmax_t)ocb->bus_addr); END_DEBUG /* * Suppress compiler optimization so that orb[1] must be written first. * XXX We may need an explicit memory barrier for other architectures * other than i386/amd64. */ *(volatile uint32_t *)&prev2->orb[1] = htonl(ocb->bus_addr); *(volatile uint32_t *)&prev2->orb[0] = 0; } return prev; } static struct sbp_ocb * sbp_get_ocb(struct sbp_dev *sdev) { struct sbp_ocb *ocb; SBP_LOCK_ASSERT(sdev->target->sbp); ocb = STAILQ_FIRST(&sdev->free_ocbs); if (ocb == NULL) { sdev->flags |= ORB_SHORTAGE; printf("ocb shortage!!!\n"); return NULL; } STAILQ_REMOVE_HEAD(&sdev->free_ocbs, ocb); ocb->ccb = NULL; return (ocb); } static void sbp_free_ocb(struct sbp_dev *sdev, struct sbp_ocb *ocb) { ocb->flags = 0; ocb->ccb = NULL; SBP_LOCK_ASSERT(sdev->target->sbp); STAILQ_INSERT_TAIL(&sdev->free_ocbs, ocb, ocb); if ((sdev->flags & ORB_SHORTAGE) != 0) { int count; sdev->flags &= ~ORB_SHORTAGE; count = sdev->freeze; sdev->freeze = 0; xpt_release_devq(sdev->path, count, TRUE); } } static void sbp_abort_ocb(struct sbp_ocb *ocb, int status) { struct sbp_dev *sdev; sdev = ocb->sdev; SBP_LOCK_ASSERT(sdev->target->sbp); SBP_DEBUG(0) device_printf(sdev->target->sbp->fd.dev, "%s:%s 0x%jx\n", __func__, sdev->bustgtlun, (uintmax_t)ocb->bus_addr); END_DEBUG SBP_DEBUG(1) if (ocb->ccb != NULL) sbp_print_scsi_cmd(ocb); END_DEBUG if (ntohl(ocb->orb[4]) & 0xffff) { bus_dmamap_sync(sdev->target->sbp->dmat, ocb->dmamap, (ntohl(ocb->orb[4]) & ORB_CMD_IN) ? BUS_DMASYNC_POSTREAD : BUS_DMASYNC_POSTWRITE); bus_dmamap_unload(sdev->target->sbp->dmat, ocb->dmamap); } if (ocb->ccb != NULL) { callout_stop(&ocb->timer); ocb->ccb->ccb_h.status = status; xpt_done(ocb->ccb); } sbp_free_ocb(sdev, ocb); } static void sbp_abort_all_ocbs(struct sbp_dev *sdev, int status) { struct sbp_ocb *ocb, *next; STAILQ_HEAD(, sbp_ocb) temp; STAILQ_INIT(&temp); SBP_LOCK_ASSERT(sdev->target->sbp); STAILQ_CONCAT(&temp, &sdev->ocbs); STAILQ_INIT(&sdev->ocbs); STAILQ_FOREACH_SAFE(ocb, &temp, ocb, next) { sbp_abort_ocb(ocb, status); } if (sdev->last_ocb != NULL) { sbp_free_ocb(sdev, sdev->last_ocb); sdev->last_ocb = NULL; } } static devclass_t sbp_devclass; static device_method_t sbp_methods[] = { /* device interface */ DEVMETHOD(device_identify, sbp_identify), DEVMETHOD(device_probe, sbp_probe), DEVMETHOD(device_attach, sbp_attach), DEVMETHOD(device_detach, sbp_detach), DEVMETHOD(device_shutdown, sbp_shutdown), { 0, 0 } }; static driver_t sbp_driver = { "sbp", sbp_methods, sizeof(struct sbp_softc), }; DRIVER_MODULE(sbp, firewire, sbp_driver, sbp_devclass, 0, 0); MODULE_VERSION(sbp, 1); MODULE_DEPEND(sbp, firewire, 1, 1, 1); MODULE_DEPEND(sbp, cam, 1, 1, 1); Index: projects/clang400-import/sys/dev/mly/mly.c =================================================================== --- projects/clang400-import/sys/dev/mly/mly.c (revision 311459) +++ projects/clang400-import/sys/dev/mly/mly.c (revision 311460) @@ -1,3014 +1,3014 @@ /*- * Copyright (c) 2000, 2001 Michael Smith * Copyright (c) 2000 BSDi * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $FreeBSD$ */ #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 static int mly_probe(device_t dev); static int mly_attach(device_t dev); static int mly_pci_attach(struct mly_softc *sc); static int mly_detach(device_t dev); static int mly_shutdown(device_t dev); static void mly_intr(void *arg); static int mly_sg_map(struct mly_softc *sc); static void mly_sg_map_helper(void *arg, bus_dma_segment_t *segs, int nseg, int error); static int mly_mmbox_map(struct mly_softc *sc); static void mly_mmbox_map_helper(void *arg, bus_dma_segment_t *segs, int nseg, int error); static void mly_free(struct mly_softc *sc); static int mly_get_controllerinfo(struct mly_softc *sc); static void mly_scan_devices(struct mly_softc *sc); static void mly_rescan_btl(struct mly_softc *sc, int bus, int target); static void mly_complete_rescan(struct mly_command *mc); static int mly_get_eventstatus(struct mly_softc *sc); static int mly_enable_mmbox(struct mly_softc *sc); static int mly_flush(struct mly_softc *sc); static int mly_ioctl(struct mly_softc *sc, struct mly_command_ioctl *ioctl, void **data, size_t datasize, u_int8_t *status, void *sense_buffer, size_t *sense_length); static void mly_check_event(struct mly_softc *sc); static void mly_fetch_event(struct mly_softc *sc); static void mly_complete_event(struct mly_command *mc); static void mly_process_event(struct mly_softc *sc, struct mly_event *me); static void mly_periodic(void *data); static int mly_immediate_command(struct mly_command *mc); static int mly_start(struct mly_command *mc); static void mly_done(struct mly_softc *sc); static void mly_complete(struct mly_softc *sc); static void mly_complete_handler(void *context, int pending); static int mly_alloc_command(struct mly_softc *sc, struct mly_command **mcp); static void mly_release_command(struct mly_command *mc); static void mly_alloc_commands_map(void *arg, bus_dma_segment_t *segs, int nseg, int error); static int mly_alloc_commands(struct mly_softc *sc); static void mly_release_commands(struct mly_softc *sc); static void mly_map_command(struct mly_command *mc); static void mly_unmap_command(struct mly_command *mc); static int mly_cam_attach(struct mly_softc *sc); static void mly_cam_detach(struct mly_softc *sc); static void mly_cam_rescan_btl(struct mly_softc *sc, int bus, int target); static void mly_cam_action(struct cam_sim *sim, union ccb *ccb); static int mly_cam_action_io(struct cam_sim *sim, struct ccb_scsiio *csio); static void mly_cam_poll(struct cam_sim *sim); static void mly_cam_complete(struct mly_command *mc); static struct cam_periph *mly_find_periph(struct mly_softc *sc, int bus, int target); static int mly_name_device(struct mly_softc *sc, int bus, int target); static int mly_fwhandshake(struct mly_softc *sc); static void mly_describe_controller(struct mly_softc *sc); #ifdef MLY_DEBUG static void mly_printstate(struct mly_softc *sc); static void mly_print_command(struct mly_command *mc); static void mly_print_packet(struct mly_command *mc); static void mly_panic(struct mly_softc *sc, char *reason); static void mly_timeout(void *arg); #endif void mly_print_controller(int controller); static d_open_t mly_user_open; static d_close_t mly_user_close; static d_ioctl_t mly_user_ioctl; static int mly_user_command(struct mly_softc *sc, struct mly_user_command *uc); static int mly_user_health(struct mly_softc *sc, struct mly_user_health *uh); #define MLY_CMD_TIMEOUT 20 static device_method_t mly_methods[] = { /* Device interface */ DEVMETHOD(device_probe, mly_probe), DEVMETHOD(device_attach, mly_attach), DEVMETHOD(device_detach, mly_detach), DEVMETHOD(device_shutdown, mly_shutdown), { 0, 0 } }; static driver_t mly_pci_driver = { "mly", mly_methods, sizeof(struct mly_softc) }; static devclass_t mly_devclass; DRIVER_MODULE(mly, pci, mly_pci_driver, mly_devclass, 0, 0); MODULE_DEPEND(mly, pci, 1, 1, 1); MODULE_DEPEND(mly, cam, 1, 1, 1); static struct cdevsw mly_cdevsw = { .d_version = D_VERSION, .d_open = mly_user_open, .d_close = mly_user_close, .d_ioctl = mly_user_ioctl, .d_name = "mly", }; /******************************************************************************** ******************************************************************************** Device Interface ******************************************************************************** ********************************************************************************/ static struct mly_ident { u_int16_t vendor; u_int16_t device; u_int16_t subvendor; u_int16_t subdevice; int hwif; char *desc; } mly_identifiers[] = { {0x1069, 0xba56, 0x1069, 0x0040, MLY_HWIF_STRONGARM, "Mylex eXtremeRAID 2000"}, {0x1069, 0xba56, 0x1069, 0x0030, MLY_HWIF_STRONGARM, "Mylex eXtremeRAID 3000"}, {0x1069, 0x0050, 0x1069, 0x0050, MLY_HWIF_I960RX, "Mylex AcceleRAID 352"}, {0x1069, 0x0050, 0x1069, 0x0052, MLY_HWIF_I960RX, "Mylex AcceleRAID 170"}, {0x1069, 0x0050, 0x1069, 0x0054, MLY_HWIF_I960RX, "Mylex AcceleRAID 160"}, {0, 0, 0, 0, 0, 0} }; /******************************************************************************** * Compare the provided PCI device with the list we support. */ static int mly_probe(device_t dev) { struct mly_ident *m; debug_called(1); for (m = mly_identifiers; m->vendor != 0; m++) { if ((m->vendor == pci_get_vendor(dev)) && (m->device == pci_get_device(dev)) && ((m->subvendor == 0) || ((m->subvendor == pci_get_subvendor(dev)) && (m->subdevice == pci_get_subdevice(dev))))) { device_set_desc(dev, m->desc); return(BUS_PROBE_DEFAULT); /* allow room to be overridden */ } } return(ENXIO); } /******************************************************************************** * Initialise the controller and softc */ static int mly_attach(device_t dev) { struct mly_softc *sc = device_get_softc(dev); int error; debug_called(1); sc->mly_dev = dev; mtx_init(&sc->mly_lock, "mly", NULL, MTX_DEF); callout_init_mtx(&sc->mly_periodic, &sc->mly_lock, 0); #ifdef MLY_DEBUG callout_init_mtx(&sc->mly_timeout, &sc->mly_lock, 0); if (device_get_unit(sc->mly_dev) == 0) mly_softc0 = sc; #endif /* * Do PCI-specific initialisation. */ if ((error = mly_pci_attach(sc)) != 0) goto out; /* * Initialise per-controller queues. */ mly_initq_free(sc); mly_initq_busy(sc); mly_initq_complete(sc); /* * Initialise command-completion task. */ TASK_INIT(&sc->mly_task_complete, 0, mly_complete_handler, sc); /* disable interrupts before we start talking to the controller */ MLY_MASK_INTERRUPTS(sc); /* * Wait for the controller to come ready, handshake with the firmware if required. * This is typically only necessary on platforms where the controller BIOS does not * run. */ if ((error = mly_fwhandshake(sc))) goto out; /* * Allocate initial command buffers. */ if ((error = mly_alloc_commands(sc))) goto out; /* * Obtain controller feature information */ MLY_LOCK(sc); error = mly_get_controllerinfo(sc); MLY_UNLOCK(sc); if (error) goto out; /* * Reallocate command buffers now we know how many we want. */ mly_release_commands(sc); if ((error = mly_alloc_commands(sc))) goto out; /* * Get the current event counter for health purposes, populate the initial * health status buffer. */ MLY_LOCK(sc); error = mly_get_eventstatus(sc); /* * Enable memory-mailbox mode. */ if (error == 0) error = mly_enable_mmbox(sc); MLY_UNLOCK(sc); if (error) goto out; /* * Attach to CAM. */ if ((error = mly_cam_attach(sc))) goto out; /* * Print a little information about the controller */ mly_describe_controller(sc); /* * Mark all attached devices for rescan. */ MLY_LOCK(sc); mly_scan_devices(sc); /* * Instigate the first status poll immediately. Rescan completions won't * happen until interrupts are enabled, which should still be before * the SCSI subsystem gets to us, courtesy of the "SCSI settling delay". */ mly_periodic((void *)sc); MLY_UNLOCK(sc); /* * Create the control device. */ sc->mly_dev_t = make_dev(&mly_cdevsw, 0, UID_ROOT, GID_OPERATOR, S_IRUSR | S_IWUSR, "mly%d", device_get_unit(sc->mly_dev)); sc->mly_dev_t->si_drv1 = sc; /* enable interrupts now */ MLY_UNMASK_INTERRUPTS(sc); #ifdef MLY_DEBUG callout_reset(&sc->mly_timeout, MLY_CMD_TIMEOUT * hz, mly_timeout, sc); #endif out: if (error != 0) mly_free(sc); return(error); } /******************************************************************************** * Perform PCI-specific initialisation. */ static int mly_pci_attach(struct mly_softc *sc) { int i, error; debug_called(1); /* assume failure is 'not configured' */ error = ENXIO; /* * Verify that the adapter is correctly set up in PCI space. */ pci_enable_busmaster(sc->mly_dev); /* * Allocate the PCI register window. */ sc->mly_regs_rid = PCIR_BAR(0); /* first base address register */ if ((sc->mly_regs_resource = bus_alloc_resource_any(sc->mly_dev, SYS_RES_MEMORY, &sc->mly_regs_rid, RF_ACTIVE)) == NULL) { mly_printf(sc, "can't allocate register window\n"); goto fail; } /* * Allocate and connect our interrupt. */ sc->mly_irq_rid = 0; if ((sc->mly_irq = bus_alloc_resource_any(sc->mly_dev, SYS_RES_IRQ, &sc->mly_irq_rid, RF_SHAREABLE | RF_ACTIVE)) == NULL) { mly_printf(sc, "can't allocate interrupt\n"); goto fail; } if (bus_setup_intr(sc->mly_dev, sc->mly_irq, INTR_TYPE_CAM | INTR_ENTROPY | INTR_MPSAFE, NULL, mly_intr, sc, &sc->mly_intr)) { mly_printf(sc, "can't set up interrupt\n"); goto fail; } /* assume failure is 'out of memory' */ error = ENOMEM; /* * Allocate the parent bus DMA tag appropriate for our PCI interface. * * Note that all of these controllers are 64-bit capable. */ if (bus_dma_tag_create(bus_get_dma_tag(sc->mly_dev),/* PCI parent */ 1, 0, /* alignment, boundary */ BUS_SPACE_MAXADDR_32BIT, /* lowaddr */ BUS_SPACE_MAXADDR, /* highaddr */ NULL, NULL, /* filter, filterarg */ BUS_SPACE_MAXSIZE_32BIT, /* maxsize */ BUS_SPACE_UNRESTRICTED, /* nsegments */ BUS_SPACE_MAXSIZE_32BIT, /* maxsegsize */ BUS_DMA_ALLOCNOW, /* flags */ NULL, /* lockfunc */ NULL, /* lockarg */ &sc->mly_parent_dmat)) { mly_printf(sc, "can't allocate parent DMA tag\n"); goto fail; } /* * Create DMA tag for mapping buffers into controller-addressable space. */ if (bus_dma_tag_create(sc->mly_parent_dmat, /* parent */ 1, 0, /* alignment, boundary */ BUS_SPACE_MAXADDR, /* lowaddr */ BUS_SPACE_MAXADDR, /* highaddr */ NULL, NULL, /* filter, filterarg */ DFLTPHYS, /* maxsize */ MLY_MAX_SGENTRIES, /* nsegments */ BUS_SPACE_MAXSIZE_32BIT, /* maxsegsize */ 0, /* flags */ busdma_lock_mutex, /* lockfunc */ &sc->mly_lock, /* lockarg */ &sc->mly_buffer_dmat)) { mly_printf(sc, "can't allocate buffer DMA tag\n"); goto fail; } /* * Initialise the DMA tag for command packets. */ if (bus_dma_tag_create(sc->mly_parent_dmat, /* parent */ 1, 0, /* alignment, boundary */ BUS_SPACE_MAXADDR, /* lowaddr */ BUS_SPACE_MAXADDR, /* highaddr */ NULL, NULL, /* filter, filterarg */ sizeof(union mly_command_packet) * MLY_MAX_COMMANDS, 1, /* maxsize, nsegments */ BUS_SPACE_MAXSIZE_32BIT, /* maxsegsize */ BUS_DMA_ALLOCNOW, /* flags */ NULL, NULL, /* lockfunc, lockarg */ &sc->mly_packet_dmat)) { mly_printf(sc, "can't allocate command packet DMA tag\n"); goto fail; } /* * Detect the hardware interface version */ for (i = 0; mly_identifiers[i].vendor != 0; i++) { if ((mly_identifiers[i].vendor == pci_get_vendor(sc->mly_dev)) && (mly_identifiers[i].device == pci_get_device(sc->mly_dev))) { sc->mly_hwif = mly_identifiers[i].hwif; switch(sc->mly_hwif) { case MLY_HWIF_I960RX: debug(1, "set hardware up for i960RX"); sc->mly_doorbell_true = 0x00; sc->mly_command_mailbox = MLY_I960RX_COMMAND_MAILBOX; sc->mly_status_mailbox = MLY_I960RX_STATUS_MAILBOX; sc->mly_idbr = MLY_I960RX_IDBR; sc->mly_odbr = MLY_I960RX_ODBR; sc->mly_error_status = MLY_I960RX_ERROR_STATUS; sc->mly_interrupt_status = MLY_I960RX_INTERRUPT_STATUS; sc->mly_interrupt_mask = MLY_I960RX_INTERRUPT_MASK; break; case MLY_HWIF_STRONGARM: debug(1, "set hardware up for StrongARM"); sc->mly_doorbell_true = 0xff; /* doorbell 'true' is 0 */ sc->mly_command_mailbox = MLY_STRONGARM_COMMAND_MAILBOX; sc->mly_status_mailbox = MLY_STRONGARM_STATUS_MAILBOX; sc->mly_idbr = MLY_STRONGARM_IDBR; sc->mly_odbr = MLY_STRONGARM_ODBR; sc->mly_error_status = MLY_STRONGARM_ERROR_STATUS; sc->mly_interrupt_status = MLY_STRONGARM_INTERRUPT_STATUS; sc->mly_interrupt_mask = MLY_STRONGARM_INTERRUPT_MASK; break; } break; } } /* * Create the scatter/gather mappings. */ if ((error = mly_sg_map(sc))) goto fail; /* * Allocate and map the memory mailbox */ if ((error = mly_mmbox_map(sc))) goto fail; error = 0; fail: return(error); } /******************************************************************************** * Shut the controller down and detach all our resources. */ static int mly_detach(device_t dev) { int error; if ((error = mly_shutdown(dev)) != 0) return(error); mly_free(device_get_softc(dev)); return(0); } /******************************************************************************** * Bring the controller to a state where it can be safely left alone. * * Note that it should not be necessary to wait for any outstanding commands, * as they should be completed prior to calling here. * * XXX this applies for I/O, but not status polls; we should beware of * the case where a status command is running while we detach. */ static int mly_shutdown(device_t dev) { struct mly_softc *sc = device_get_softc(dev); debug_called(1); MLY_LOCK(sc); if (sc->mly_state & MLY_STATE_OPEN) { MLY_UNLOCK(sc); return(EBUSY); } /* kill the periodic event */ callout_stop(&sc->mly_periodic); #ifdef MLY_DEBUG callout_stop(&sc->mly_timeout); #endif /* flush controller */ mly_printf(sc, "flushing cache..."); printf("%s\n", mly_flush(sc) ? "failed" : "done"); MLY_MASK_INTERRUPTS(sc); MLY_UNLOCK(sc); return(0); } /******************************************************************************* * Take an interrupt, or be poked by other code to look for interrupt-worthy * status. */ static void mly_intr(void *arg) { struct mly_softc *sc = (struct mly_softc *)arg; debug_called(2); MLY_LOCK(sc); mly_done(sc); MLY_UNLOCK(sc); }; /******************************************************************************** ******************************************************************************** Bus-dependant Resource Management ******************************************************************************** ********************************************************************************/ /******************************************************************************** * Allocate memory for the scatter/gather tables */ static int mly_sg_map(struct mly_softc *sc) { size_t segsize; debug_called(1); /* * Create a single tag describing a region large enough to hold all of * the s/g lists we will need. */ segsize = sizeof(struct mly_sg_entry) * MLY_MAX_COMMANDS *MLY_MAX_SGENTRIES; if (bus_dma_tag_create(sc->mly_parent_dmat, /* parent */ 1, 0, /* alignment,boundary */ BUS_SPACE_MAXADDR, /* lowaddr */ BUS_SPACE_MAXADDR, /* highaddr */ NULL, NULL, /* filter, filterarg */ segsize, 1, /* maxsize, nsegments */ BUS_SPACE_MAXSIZE_32BIT, /* maxsegsize */ BUS_DMA_ALLOCNOW, /* flags */ NULL, NULL, /* lockfunc, lockarg */ &sc->mly_sg_dmat)) { mly_printf(sc, "can't allocate scatter/gather DMA tag\n"); return(ENOMEM); } /* * Allocate enough s/g maps for all commands and permanently map them into * controller-visible space. * * XXX this assumes we can get enough space for all the s/g maps in one * contiguous slab. */ if (bus_dmamem_alloc(sc->mly_sg_dmat, (void **)&sc->mly_sg_table, BUS_DMA_NOWAIT, &sc->mly_sg_dmamap)) { mly_printf(sc, "can't allocate s/g table\n"); return(ENOMEM); } if (bus_dmamap_load(sc->mly_sg_dmat, sc->mly_sg_dmamap, sc->mly_sg_table, segsize, mly_sg_map_helper, sc, BUS_DMA_NOWAIT) != 0) return (ENOMEM); return(0); } /******************************************************************************** * Save the physical address of the base of the s/g table. */ static void mly_sg_map_helper(void *arg, bus_dma_segment_t *segs, int nseg, int error) { struct mly_softc *sc = (struct mly_softc *)arg; debug_called(1); /* save base of s/g table's address in bus space */ sc->mly_sg_busaddr = segs->ds_addr; } /******************************************************************************** * Allocate memory for the memory-mailbox interface */ static int mly_mmbox_map(struct mly_softc *sc) { /* * Create a DMA tag for a single contiguous region large enough for the * memory mailbox structure. */ if (bus_dma_tag_create(sc->mly_parent_dmat, /* parent */ 1, 0, /* alignment,boundary */ BUS_SPACE_MAXADDR, /* lowaddr */ BUS_SPACE_MAXADDR, /* highaddr */ NULL, NULL, /* filter, filterarg */ sizeof(struct mly_mmbox), 1, /* maxsize, nsegments */ BUS_SPACE_MAXSIZE_32BIT, /* maxsegsize */ BUS_DMA_ALLOCNOW, /* flags */ NULL, NULL, /* lockfunc, lockarg */ &sc->mly_mmbox_dmat)) { mly_printf(sc, "can't allocate memory mailbox DMA tag\n"); return(ENOMEM); } /* * Allocate the buffer */ if (bus_dmamem_alloc(sc->mly_mmbox_dmat, (void **)&sc->mly_mmbox, BUS_DMA_NOWAIT, &sc->mly_mmbox_dmamap)) { mly_printf(sc, "can't allocate memory mailbox\n"); return(ENOMEM); } if (bus_dmamap_load(sc->mly_mmbox_dmat, sc->mly_mmbox_dmamap, sc->mly_mmbox, sizeof(struct mly_mmbox), mly_mmbox_map_helper, sc, BUS_DMA_NOWAIT) != 0) return (ENOMEM); bzero(sc->mly_mmbox, sizeof(*sc->mly_mmbox)); return(0); } /******************************************************************************** * Save the physical address of the memory mailbox */ static void mly_mmbox_map_helper(void *arg, bus_dma_segment_t *segs, int nseg, int error) { struct mly_softc *sc = (struct mly_softc *)arg; debug_called(1); sc->mly_mmbox_busaddr = segs->ds_addr; } /******************************************************************************** * Free all of the resources associated with (sc) * * Should not be called if the controller is active. */ static void mly_free(struct mly_softc *sc) { debug_called(1); /* Remove the management device */ destroy_dev(sc->mly_dev_t); if (sc->mly_intr) bus_teardown_intr(sc->mly_dev, sc->mly_irq, sc->mly_intr); callout_drain(&sc->mly_periodic); #ifdef MLY_DEBUG callout_drain(&sc->mly_timeout); #endif /* detach from CAM */ mly_cam_detach(sc); /* release command memory */ mly_release_commands(sc); /* throw away the controllerinfo structure */ if (sc->mly_controllerinfo != NULL) free(sc->mly_controllerinfo, M_DEVBUF); /* throw away the controllerparam structure */ if (sc->mly_controllerparam != NULL) free(sc->mly_controllerparam, M_DEVBUF); /* destroy data-transfer DMA tag */ if (sc->mly_buffer_dmat) bus_dma_tag_destroy(sc->mly_buffer_dmat); /* free and destroy DMA memory and tag for s/g lists */ if (sc->mly_sg_table) { bus_dmamap_unload(sc->mly_sg_dmat, sc->mly_sg_dmamap); bus_dmamem_free(sc->mly_sg_dmat, sc->mly_sg_table, sc->mly_sg_dmamap); } if (sc->mly_sg_dmat) bus_dma_tag_destroy(sc->mly_sg_dmat); /* free and destroy DMA memory and tag for memory mailbox */ if (sc->mly_mmbox) { bus_dmamap_unload(sc->mly_mmbox_dmat, sc->mly_mmbox_dmamap); bus_dmamem_free(sc->mly_mmbox_dmat, sc->mly_mmbox, sc->mly_mmbox_dmamap); } if (sc->mly_mmbox_dmat) bus_dma_tag_destroy(sc->mly_mmbox_dmat); /* disconnect the interrupt handler */ if (sc->mly_irq != NULL) bus_release_resource(sc->mly_dev, SYS_RES_IRQ, sc->mly_irq_rid, sc->mly_irq); /* destroy the parent DMA tag */ if (sc->mly_parent_dmat) bus_dma_tag_destroy(sc->mly_parent_dmat); /* release the register window mapping */ if (sc->mly_regs_resource != NULL) bus_release_resource(sc->mly_dev, SYS_RES_MEMORY, sc->mly_regs_rid, sc->mly_regs_resource); mtx_destroy(&sc->mly_lock); } /******************************************************************************** ******************************************************************************** Command Wrappers ******************************************************************************** ********************************************************************************/ /******************************************************************************** * Fill in the mly_controllerinfo and mly_controllerparam fields in the softc. */ static int mly_get_controllerinfo(struct mly_softc *sc) { struct mly_command_ioctl mci; u_int8_t status; int error; debug_called(1); if (sc->mly_controllerinfo != NULL) free(sc->mly_controllerinfo, M_DEVBUF); /* build the getcontrollerinfo ioctl and send it */ bzero(&mci, sizeof(mci)); sc->mly_controllerinfo = NULL; mci.sub_ioctl = MDACIOCTL_GETCONTROLLERINFO; if ((error = mly_ioctl(sc, &mci, (void **)&sc->mly_controllerinfo, sizeof(*sc->mly_controllerinfo), &status, NULL, NULL))) return(error); if (status != 0) return(EIO); if (sc->mly_controllerparam != NULL) free(sc->mly_controllerparam, M_DEVBUF); /* build the getcontrollerparameter ioctl and send it */ bzero(&mci, sizeof(mci)); sc->mly_controllerparam = NULL; mci.sub_ioctl = MDACIOCTL_GETCONTROLLERPARAMETER; if ((error = mly_ioctl(sc, &mci, (void **)&sc->mly_controllerparam, sizeof(*sc->mly_controllerparam), &status, NULL, NULL))) return(error); if (status != 0) return(EIO); return(0); } /******************************************************************************** * Schedule all possible devices for a rescan. * */ static void mly_scan_devices(struct mly_softc *sc) { int bus, target; debug_called(1); /* * Clear any previous BTL information. */ bzero(&sc->mly_btl, sizeof(sc->mly_btl)); /* * Mark all devices as requiring a rescan, and let the next * periodic scan collect them. */ for (bus = 0; bus < sc->mly_cam_channels; bus++) if (MLY_BUS_IS_VALID(sc, bus)) for (target = 0; target < MLY_MAX_TARGETS; target++) sc->mly_btl[bus][target].mb_flags = MLY_BTL_RESCAN; } /******************************************************************************** * Rescan a device, possibly as a consequence of getting an event which suggests * that it may have changed. * * If we suffer resource starvation, we can abandon the rescan as we'll be * retried. */ static void mly_rescan_btl(struct mly_softc *sc, int bus, int target) { struct mly_command *mc; struct mly_command_ioctl *mci; debug_called(1); /* check that this bus is valid */ if (!MLY_BUS_IS_VALID(sc, bus)) return; /* get a command */ if (mly_alloc_command(sc, &mc)) return; /* set up the data buffer */ if ((mc->mc_data = malloc(sizeof(union mly_devinfo), M_DEVBUF, M_NOWAIT | M_ZERO)) == NULL) { mly_release_command(mc); return; } mc->mc_flags |= MLY_CMD_DATAIN; mc->mc_complete = mly_complete_rescan; /* * Build the ioctl. */ mci = (struct mly_command_ioctl *)&mc->mc_packet->ioctl; mci->opcode = MDACMD_IOCTL; mci->addr.phys.controller = 0; mci->timeout.value = 30; mci->timeout.scale = MLY_TIMEOUT_SECONDS; if (MLY_BUS_IS_VIRTUAL(sc, bus)) { mc->mc_length = mci->data_size = sizeof(struct mly_ioctl_getlogdevinfovalid); mci->sub_ioctl = MDACIOCTL_GETLOGDEVINFOVALID; mci->addr.log.logdev = MLY_LOGDEV_ID(sc, bus, target); debug(1, "logical device %d", mci->addr.log.logdev); } else { mc->mc_length = mci->data_size = sizeof(struct mly_ioctl_getphysdevinfovalid); mci->sub_ioctl = MDACIOCTL_GETPHYSDEVINFOVALID; mci->addr.phys.lun = 0; mci->addr.phys.target = target; mci->addr.phys.channel = bus; debug(1, "physical device %d:%d", mci->addr.phys.channel, mci->addr.phys.target); } /* * Dispatch the command. If we successfully send the command, clear the rescan * bit. */ if (mly_start(mc) != 0) { mly_release_command(mc); } else { sc->mly_btl[bus][target].mb_flags &= ~MLY_BTL_RESCAN; /* success */ } } /******************************************************************************** * Handle the completion of a rescan operation */ static void mly_complete_rescan(struct mly_command *mc) { struct mly_softc *sc = mc->mc_sc; struct mly_ioctl_getlogdevinfovalid *ldi; struct mly_ioctl_getphysdevinfovalid *pdi; struct mly_command_ioctl *mci; struct mly_btl btl, *btlp; int bus, target, rescan; debug_called(1); /* * Recover the bus and target from the command. We need these even in * the case where we don't have a useful response. */ mci = (struct mly_command_ioctl *)&mc->mc_packet->ioctl; if (mci->sub_ioctl == MDACIOCTL_GETLOGDEVINFOVALID) { bus = MLY_LOGDEV_BUS(sc, mci->addr.log.logdev); target = MLY_LOGDEV_TARGET(sc, mci->addr.log.logdev); } else { bus = mci->addr.phys.channel; target = mci->addr.phys.target; } /* XXX validate bus/target? */ /* the default result is 'no device' */ bzero(&btl, sizeof(btl)); /* if the rescan completed OK, we have possibly-new BTL data */ if (mc->mc_status == 0) { if (mc->mc_length == sizeof(*ldi)) { ldi = (struct mly_ioctl_getlogdevinfovalid *)mc->mc_data; if ((MLY_LOGDEV_BUS(sc, ldi->logical_device_number) != bus) || (MLY_LOGDEV_TARGET(sc, ldi->logical_device_number) != target)) { mly_printf(sc, "WARNING: BTL rescan for %d:%d returned data for %d:%d instead\n", bus, target, MLY_LOGDEV_BUS(sc, ldi->logical_device_number), MLY_LOGDEV_TARGET(sc, ldi->logical_device_number)); /* XXX what can we do about this? */ } btl.mb_flags = MLY_BTL_LOGICAL; btl.mb_type = ldi->raid_level; btl.mb_state = ldi->state; debug(1, "BTL rescan for %d returns %s, %s", ldi->logical_device_number, mly_describe_code(mly_table_device_type, ldi->raid_level), mly_describe_code(mly_table_device_state, ldi->state)); } else if (mc->mc_length == sizeof(*pdi)) { pdi = (struct mly_ioctl_getphysdevinfovalid *)mc->mc_data; if ((pdi->channel != bus) || (pdi->target != target)) { mly_printf(sc, "WARNING: BTL rescan for %d:%d returned data for %d:%d instead\n", bus, target, pdi->channel, pdi->target); /* XXX what can we do about this? */ } btl.mb_flags = MLY_BTL_PHYSICAL; btl.mb_type = MLY_DEVICE_TYPE_PHYSICAL; btl.mb_state = pdi->state; btl.mb_speed = pdi->speed; btl.mb_width = pdi->width; if (pdi->state != MLY_DEVICE_STATE_UNCONFIGURED) sc->mly_btl[bus][target].mb_flags |= MLY_BTL_PROTECTED; debug(1, "BTL rescan for %d:%d returns %s", bus, target, mly_describe_code(mly_table_device_state, pdi->state)); } else { mly_printf(sc, "BTL rescan result invalid\n"); } } free(mc->mc_data, M_DEVBUF); mly_release_command(mc); /* * Decide whether we need to rescan the device. */ rescan = 0; /* device type changes (usually between 'nothing' and 'something') */ btlp = &sc->mly_btl[bus][target]; if (btl.mb_flags != btlp->mb_flags) { debug(1, "flags changed, rescanning"); rescan = 1; } /* XXX other reasons? */ /* * Update BTL information. */ *btlp = btl; /* * Perform CAM rescan if required. */ if (rescan) mly_cam_rescan_btl(sc, bus, target); } /******************************************************************************** * Get the current health status and set the 'next event' counter to suit. */ static int mly_get_eventstatus(struct mly_softc *sc) { struct mly_command_ioctl mci; struct mly_health_status *mh; u_int8_t status; int error; /* build the gethealthstatus ioctl and send it */ bzero(&mci, sizeof(mci)); mh = NULL; mci.sub_ioctl = MDACIOCTL_GETHEALTHSTATUS; if ((error = mly_ioctl(sc, &mci, (void **)&mh, sizeof(*mh), &status, NULL, NULL))) return(error); if (status != 0) return(EIO); /* get the event counter */ sc->mly_event_change = mh->change_counter; sc->mly_event_waiting = mh->next_event; sc->mly_event_counter = mh->next_event; /* save the health status into the memory mailbox */ bcopy(mh, &sc->mly_mmbox->mmm_health.status, sizeof(*mh)); debug(1, "initial change counter %d, event counter %d", mh->change_counter, mh->next_event); free(mh, M_DEVBUF); return(0); } /******************************************************************************** * Enable the memory mailbox mode. */ static int mly_enable_mmbox(struct mly_softc *sc) { struct mly_command_ioctl mci; u_int8_t *sp, status; int error; debug_called(1); /* build the ioctl and send it */ bzero(&mci, sizeof(mci)); mci.sub_ioctl = MDACIOCTL_SETMEMORYMAILBOX; /* set buffer addresses */ mci.param.setmemorymailbox.command_mailbox_physaddr = sc->mly_mmbox_busaddr + offsetof(struct mly_mmbox, mmm_command); mci.param.setmemorymailbox.status_mailbox_physaddr = sc->mly_mmbox_busaddr + offsetof(struct mly_mmbox, mmm_status); mci.param.setmemorymailbox.health_buffer_physaddr = sc->mly_mmbox_busaddr + offsetof(struct mly_mmbox, mmm_health); /* set buffer sizes - abuse of data_size field is revolting */ sp = (u_int8_t *)&mci.data_size; sp[0] = ((sizeof(union mly_command_packet) * MLY_MMBOX_COMMANDS) / 1024); sp[1] = (sizeof(union mly_status_packet) * MLY_MMBOX_STATUS) / 1024; mci.param.setmemorymailbox.health_buffer_size = sizeof(union mly_health_region) / 1024; debug(1, "memory mailbox at %p (0x%llx/%d 0x%llx/%d 0x%llx/%d", sc->mly_mmbox, mci.param.setmemorymailbox.command_mailbox_physaddr, sp[0], mci.param.setmemorymailbox.status_mailbox_physaddr, sp[1], mci.param.setmemorymailbox.health_buffer_physaddr, mci.param.setmemorymailbox.health_buffer_size); if ((error = mly_ioctl(sc, &mci, NULL, 0, &status, NULL, NULL))) return(error); if (status != 0) return(EIO); sc->mly_state |= MLY_STATE_MMBOX_ACTIVE; debug(1, "memory mailbox active"); return(0); } /******************************************************************************** * Flush all pending I/O from the controller. */ static int mly_flush(struct mly_softc *sc) { struct mly_command_ioctl mci; u_int8_t status; int error; debug_called(1); /* build the ioctl */ bzero(&mci, sizeof(mci)); mci.sub_ioctl = MDACIOCTL_FLUSHDEVICEDATA; mci.param.deviceoperation.operation_device = MLY_OPDEVICE_PHYSICAL_CONTROLLER; /* pass it off to the controller */ if ((error = mly_ioctl(sc, &mci, NULL, 0, &status, NULL, NULL))) return(error); return((status == 0) ? 0 : EIO); } /******************************************************************************** * Perform an ioctl command. * * If (data) is not NULL, the command requires data transfer. If (*data) is NULL * the command requires data transfer from the controller, and we will allocate * a buffer for it. If (*data) is not NULL, the command requires data transfer * to the controller. * * XXX passing in the whole ioctl structure is ugly. Better ideas? * * XXX we don't even try to handle the case where datasize > 4k. We should. */ static int mly_ioctl(struct mly_softc *sc, struct mly_command_ioctl *ioctl, void **data, size_t datasize, u_int8_t *status, void *sense_buffer, size_t *sense_length) { struct mly_command *mc; struct mly_command_ioctl *mci; int error; debug_called(1); MLY_ASSERT_LOCKED(sc); mc = NULL; if (mly_alloc_command(sc, &mc)) { error = ENOMEM; goto out; } /* copy the ioctl structure, but save some important fields and then fixup */ mci = &mc->mc_packet->ioctl; ioctl->sense_buffer_address = mci->sense_buffer_address; ioctl->maximum_sense_size = mci->maximum_sense_size; *mci = *ioctl; mci->opcode = MDACMD_IOCTL; mci->timeout.value = 30; mci->timeout.scale = MLY_TIMEOUT_SECONDS; /* handle the data buffer */ if (data != NULL) { if (*data == NULL) { /* allocate data buffer */ if ((mc->mc_data = malloc(datasize, M_DEVBUF, M_NOWAIT)) == NULL) { error = ENOMEM; goto out; } mc->mc_flags |= MLY_CMD_DATAIN; } else { mc->mc_data = *data; mc->mc_flags |= MLY_CMD_DATAOUT; } mc->mc_length = datasize; mc->mc_packet->generic.data_size = datasize; } /* run the command */ if ((error = mly_immediate_command(mc))) goto out; /* clean up and return any data */ *status = mc->mc_status; if ((mc->mc_sense > 0) && (sense_buffer != NULL)) { bcopy(mc->mc_packet, sense_buffer, mc->mc_sense); *sense_length = mc->mc_sense; goto out; } /* should we return a data pointer? */ if ((data != NULL) && (*data == NULL)) *data = mc->mc_data; /* command completed OK */ error = 0; out: if (mc != NULL) { /* do we need to free a data buffer we allocated? */ if (error && (mc->mc_data != NULL) && (*data == NULL)) free(mc->mc_data, M_DEVBUF); mly_release_command(mc); } return(error); } /******************************************************************************** * Check for event(s) outstanding in the controller. */ static void mly_check_event(struct mly_softc *sc) { /* * The controller may have updated the health status information, * so check for it here. Note that the counters are all in host memory, * so this check is very cheap. Also note that we depend on checking on * completion */ if (sc->mly_mmbox->mmm_health.status.change_counter != sc->mly_event_change) { sc->mly_event_change = sc->mly_mmbox->mmm_health.status.change_counter; debug(1, "event change %d, event status update, %d -> %d", sc->mly_event_change, sc->mly_event_waiting, sc->mly_mmbox->mmm_health.status.next_event); sc->mly_event_waiting = sc->mly_mmbox->mmm_health.status.next_event; /* wake up anyone that might be interested in this */ wakeup(&sc->mly_event_change); } if (sc->mly_event_counter != sc->mly_event_waiting) mly_fetch_event(sc); } /******************************************************************************** * Fetch one event from the controller. * * If we fail due to resource starvation, we'll be retried the next time a * command completes. */ static void mly_fetch_event(struct mly_softc *sc) { struct mly_command *mc; struct mly_command_ioctl *mci; int s; u_int32_t event; debug_called(1); /* get a command */ if (mly_alloc_command(sc, &mc)) return; /* set up the data buffer */ if ((mc->mc_data = malloc(sizeof(struct mly_event), M_DEVBUF, M_NOWAIT | M_ZERO)) == NULL) { mly_release_command(mc); return; } mc->mc_length = sizeof(struct mly_event); mc->mc_flags |= MLY_CMD_DATAIN; mc->mc_complete = mly_complete_event; /* * Get an event number to fetch. It's possible that we've raced with another * context for the last event, in which case there will be no more events. */ s = splcam(); if (sc->mly_event_counter == sc->mly_event_waiting) { mly_release_command(mc); splx(s); return; } event = sc->mly_event_counter++; splx(s); /* * Build the ioctl. * * At this point we are committed to sending this request, as it * will be the only one constructed for this particular event number. */ mci = (struct mly_command_ioctl *)&mc->mc_packet->ioctl; mci->opcode = MDACMD_IOCTL; mci->data_size = sizeof(struct mly_event); mci->addr.phys.lun = (event >> 16) & 0xff; mci->addr.phys.target = (event >> 24) & 0xff; mci->addr.phys.channel = 0; mci->addr.phys.controller = 0; mci->timeout.value = 30; mci->timeout.scale = MLY_TIMEOUT_SECONDS; mci->sub_ioctl = MDACIOCTL_GETEVENT; mci->param.getevent.sequence_number_low = event & 0xffff; debug(1, "fetch event %u", event); /* * Submit the command. * * Note that failure of mly_start() will result in this event never being * fetched. */ if (mly_start(mc) != 0) { mly_printf(sc, "couldn't fetch event %u\n", event); mly_release_command(mc); } } /******************************************************************************** * Handle the completion of an event poll. */ static void mly_complete_event(struct mly_command *mc) { struct mly_softc *sc = mc->mc_sc; struct mly_event *me = (struct mly_event *)mc->mc_data; debug_called(1); /* * If the event was successfully fetched, process it. */ if (mc->mc_status == SCSI_STATUS_OK) { mly_process_event(sc, me); free(me, M_DEVBUF); } mly_release_command(mc); /* * Check for another event. */ mly_check_event(sc); } /******************************************************************************** * Process a controller event. */ static void mly_process_event(struct mly_softc *sc, struct mly_event *me) { struct scsi_sense_data_fixed *ssd; char *fp, *tp; int bus, target, event, class, action; ssd = (struct scsi_sense_data_fixed *)&me->sense[0]; /* * Errors can be reported using vendor-unique sense data. In this case, the * event code will be 0x1c (Request sense data present), the sense key will * be 0x09 (vendor specific), the MSB of the ASC will be set, and the * actual event code will be a 16-bit value comprised of the ASCQ (low byte) * and low seven bits of the ASC (low seven bits of the high byte). */ if ((me->code == 0x1c) && ((ssd->flags & SSD_KEY) == SSD_KEY_Vendor_Specific) && (ssd->add_sense_code & 0x80)) { event = ((int)(ssd->add_sense_code & ~0x80) << 8) + ssd->add_sense_code_qual; } else { event = me->code; } /* look up event, get codes */ fp = mly_describe_code(mly_table_event, event); debug(1, "Event %d code 0x%x", me->sequence_number, me->code); /* quiet event? */ class = fp[0]; if (isupper(class) && bootverbose) class = tolower(class); /* get action code, text string */ action = fp[1]; tp = &fp[2]; /* * Print some information about the event. * * This code uses a table derived from the corresponding portion of the Linux * driver, and thus the parser is very similar. */ switch(class) { case 'p': /* error on physical device */ mly_printf(sc, "physical device %d:%d %s\n", me->channel, me->target, tp); if (action == 'r') sc->mly_btl[me->channel][me->target].mb_flags |= MLY_BTL_RESCAN; break; case 'l': /* error on logical unit */ case 'm': /* message about logical unit */ bus = MLY_LOGDEV_BUS(sc, me->lun); target = MLY_LOGDEV_TARGET(sc, me->lun); mly_name_device(sc, bus, target); mly_printf(sc, "logical device %d (%s) %s\n", me->lun, sc->mly_btl[bus][target].mb_name, tp); if (action == 'r') sc->mly_btl[bus][target].mb_flags |= MLY_BTL_RESCAN; break; case 's': /* report of sense data */ if (((ssd->flags & SSD_KEY) == SSD_KEY_NO_SENSE) || (((ssd->flags & SSD_KEY) == SSD_KEY_NOT_READY) && (ssd->add_sense_code == 0x04) && ((ssd->add_sense_code_qual == 0x01) || (ssd->add_sense_code_qual == 0x02)))) break; /* ignore NO_SENSE or NOT_READY in one case */ mly_printf(sc, "physical device %d:%d %s\n", me->channel, me->target, tp); mly_printf(sc, " sense key %d asc %02x ascq %02x\n", ssd->flags & SSD_KEY, ssd->add_sense_code, ssd->add_sense_code_qual); mly_printf(sc, " info %4D csi %4D\n", ssd->info, "", ssd->cmd_spec_info, ""); if (action == 'r') sc->mly_btl[me->channel][me->target].mb_flags |= MLY_BTL_RESCAN; break; case 'e': mly_printf(sc, tp, me->target, me->lun); printf("\n"); break; case 'c': mly_printf(sc, "controller %s\n", tp); break; case '?': mly_printf(sc, "%s - %d\n", tp, me->code); break; default: /* probably a 'noisy' event being ignored */ break; } } /******************************************************************************** * Perform periodic activities. */ static void mly_periodic(void *data) { struct mly_softc *sc = (struct mly_softc *)data; int bus, target; debug_called(2); MLY_ASSERT_LOCKED(sc); /* * Scan devices. */ for (bus = 0; bus < sc->mly_cam_channels; bus++) { if (MLY_BUS_IS_VALID(sc, bus)) { for (target = 0; target < MLY_MAX_TARGETS; target++) { /* ignore the controller in this scan */ if (target == sc->mly_controllerparam->initiator_id) continue; /* perform device rescan? */ if (sc->mly_btl[bus][target].mb_flags & MLY_BTL_RESCAN) mly_rescan_btl(sc, bus, target); } } } /* check for controller events */ mly_check_event(sc); /* reschedule ourselves */ callout_schedule(&sc->mly_periodic, MLY_PERIODIC_INTERVAL * hz); } /******************************************************************************** ******************************************************************************** Command Processing ******************************************************************************** ********************************************************************************/ /******************************************************************************** * Run a command and wait for it to complete. * */ static int mly_immediate_command(struct mly_command *mc) { struct mly_softc *sc = mc->mc_sc; int error; debug_called(1); MLY_ASSERT_LOCKED(sc); if ((error = mly_start(mc))) { return(error); } if (sc->mly_state & MLY_STATE_INTERRUPTS_ON) { /* sleep on the command */ while(!(mc->mc_flags & MLY_CMD_COMPLETE)) { mtx_sleep(mc, &sc->mly_lock, PRIBIO, "mlywait", 0); } } else { /* spin and collect status while we do */ while(!(mc->mc_flags & MLY_CMD_COMPLETE)) { mly_done(mc->mc_sc); } } return(0); } /******************************************************************************** * Deliver a command to the controller. * * XXX it would be good to just queue commands that we can't submit immediately * and send them later, but we probably want a wrapper for that so that * we don't hang on a failed submission for an immediate command. */ static int mly_start(struct mly_command *mc) { struct mly_softc *sc = mc->mc_sc; union mly_command_packet *pkt; debug_called(2); MLY_ASSERT_LOCKED(sc); /* * Set the command up for delivery to the controller. */ mly_map_command(mc); mc->mc_packet->generic.command_id = mc->mc_slot; #ifdef MLY_DEBUG mc->mc_timestamp = time_second; #endif /* * Do we have to use the hardware mailbox? */ if (!(sc->mly_state & MLY_STATE_MMBOX_ACTIVE)) { /* * Check to see if the controller is ready for us. */ if (MLY_IDBR_TRUE(sc, MLY_HM_CMDSENT)) { return(EBUSY); } mc->mc_flags |= MLY_CMD_BUSY; /* * It's ready, send the command. */ MLY_SET_MBOX(sc, sc->mly_command_mailbox, &mc->mc_packetphys); MLY_SET_REG(sc, sc->mly_idbr, MLY_HM_CMDSENT); } else { /* use memory-mailbox mode */ pkt = &sc->mly_mmbox->mmm_command[sc->mly_mmbox_command_index]; /* check to see if the next index is free yet */ if (pkt->mmbox.flag != 0) { return(EBUSY); } mc->mc_flags |= MLY_CMD_BUSY; /* copy in new command */ bcopy(mc->mc_packet->mmbox.data, pkt->mmbox.data, sizeof(pkt->mmbox.data)); /* barrier to ensure completion of previous write before we write the flag */ bus_barrier(sc->mly_regs_resource, 0, 0, BUS_SPACE_BARRIER_WRITE); /* copy flag last */ pkt->mmbox.flag = mc->mc_packet->mmbox.flag; /* barrier to ensure completion of previous write before we notify the controller */ bus_barrier(sc->mly_regs_resource, 0, 0, BUS_SPACE_BARRIER_WRITE); /* signal controller, update index */ MLY_SET_REG(sc, sc->mly_idbr, MLY_AM_CMDSENT); sc->mly_mmbox_command_index = (sc->mly_mmbox_command_index + 1) % MLY_MMBOX_COMMANDS; } mly_enqueue_busy(mc); return(0); } /******************************************************************************** * Pick up command status from the controller, schedule a completion event */ static void mly_done(struct mly_softc *sc) { struct mly_command *mc; union mly_status_packet *sp; u_int16_t slot; int worked; MLY_ASSERT_LOCKED(sc); worked = 0; /* pick up hardware-mailbox commands */ if (MLY_ODBR_TRUE(sc, MLY_HM_STSREADY)) { slot = MLY_GET_REG2(sc, sc->mly_status_mailbox); if (slot < MLY_SLOT_MAX) { mc = &sc->mly_command[slot - MLY_SLOT_START]; mc->mc_status = MLY_GET_REG(sc, sc->mly_status_mailbox + 2); mc->mc_sense = MLY_GET_REG(sc, sc->mly_status_mailbox + 3); mc->mc_resid = MLY_GET_REG4(sc, sc->mly_status_mailbox + 4); mly_remove_busy(mc); mc->mc_flags &= ~MLY_CMD_BUSY; mly_enqueue_complete(mc); worked = 1; } else { /* slot 0xffff may mean "extremely bogus command" */ mly_printf(sc, "got HM completion for illegal slot %u\n", slot); } /* unconditionally acknowledge status */ MLY_SET_REG(sc, sc->mly_odbr, MLY_HM_STSREADY); MLY_SET_REG(sc, sc->mly_idbr, MLY_HM_STSACK); } /* pick up memory-mailbox commands */ if (MLY_ODBR_TRUE(sc, MLY_AM_STSREADY)) { for (;;) { sp = &sc->mly_mmbox->mmm_status[sc->mly_mmbox_status_index]; /* check for more status */ if (sp->mmbox.flag == 0) break; /* get slot number */ slot = sp->status.command_id; if (slot < MLY_SLOT_MAX) { mc = &sc->mly_command[slot - MLY_SLOT_START]; mc->mc_status = sp->status.status; mc->mc_sense = sp->status.sense_length; mc->mc_resid = sp->status.residue; mly_remove_busy(mc); mc->mc_flags &= ~MLY_CMD_BUSY; mly_enqueue_complete(mc); worked = 1; } else { /* slot 0xffff may mean "extremely bogus command" */ mly_printf(sc, "got AM completion for illegal slot %u at %d\n", slot, sc->mly_mmbox_status_index); } /* clear and move to next index */ sp->mmbox.flag = 0; sc->mly_mmbox_status_index = (sc->mly_mmbox_status_index + 1) % MLY_MMBOX_STATUS; } /* acknowledge that we have collected status value(s) */ MLY_SET_REG(sc, sc->mly_odbr, MLY_AM_STSREADY); } if (worked) { if (sc->mly_state & MLY_STATE_INTERRUPTS_ON) taskqueue_enqueue(taskqueue_thread, &sc->mly_task_complete); else mly_complete(sc); } } /******************************************************************************** * Process completed commands */ static void mly_complete_handler(void *context, int pending) { struct mly_softc *sc = (struct mly_softc *)context; MLY_LOCK(sc); mly_complete(sc); MLY_UNLOCK(sc); } static void mly_complete(struct mly_softc *sc) { struct mly_command *mc; void (* mc_complete)(struct mly_command *mc); debug_called(2); /* * Spin pulling commands off the completed queue and processing them. */ while ((mc = mly_dequeue_complete(sc)) != NULL) { /* * Free controller resources, mark command complete. * * Note that as soon as we mark the command complete, it may be freed * out from under us, so we need to save the mc_complete field in * order to later avoid dereferencing mc. (We would not expect to * have a polling/sleeping consumer with mc_complete != NULL). */ mly_unmap_command(mc); mc_complete = mc->mc_complete; mc->mc_flags |= MLY_CMD_COMPLETE; /* * Call completion handler or wake up sleeping consumer. */ if (mc_complete != NULL) { mc_complete(mc); } else { wakeup(mc); } } /* * XXX if we are deferring commands due to controller-busy status, we should * retry submitting them here. */ } /******************************************************************************** ******************************************************************************** Command Buffer Management ******************************************************************************** ********************************************************************************/ /******************************************************************************** * Allocate a command. */ static int mly_alloc_command(struct mly_softc *sc, struct mly_command **mcp) { struct mly_command *mc; debug_called(3); if ((mc = mly_dequeue_free(sc)) == NULL) return(ENOMEM); *mcp = mc; return(0); } /******************************************************************************** * Release a command back to the freelist. */ static void mly_release_command(struct mly_command *mc) { debug_called(3); /* * Fill in parts of the command that may cause confusion if * a consumer doesn't when we are later allocated. */ mc->mc_data = NULL; mc->mc_flags = 0; mc->mc_complete = NULL; mc->mc_private = NULL; /* * By default, we set up to overwrite the command packet with * sense information. */ mc->mc_packet->generic.sense_buffer_address = mc->mc_packetphys; mc->mc_packet->generic.maximum_sense_size = sizeof(union mly_command_packet); mly_enqueue_free(mc); } /******************************************************************************** * Map helper for command allocation. */ static void mly_alloc_commands_map(void *arg, bus_dma_segment_t *segs, int nseg, int error) { struct mly_softc *sc = (struct mly_softc *)arg; debug_called(1); sc->mly_packetphys = segs[0].ds_addr; } /******************************************************************************** * Allocate and initialise command and packet structures. * * If the controller supports fewer than MLY_MAX_COMMANDS commands, limit our * allocation to that number. If we don't yet know how many commands the * controller supports, allocate a very small set (suitable for initialisation * purposes only). */ static int mly_alloc_commands(struct mly_softc *sc) { struct mly_command *mc; int i, ncmd; if (sc->mly_controllerinfo == NULL) { ncmd = 4; } else { ncmd = min(MLY_MAX_COMMANDS, sc->mly_controllerinfo->maximum_parallel_commands); } /* * Allocate enough space for all the command packets in one chunk and * map them permanently into controller-visible space. */ if (bus_dmamem_alloc(sc->mly_packet_dmat, (void **)&sc->mly_packet, BUS_DMA_NOWAIT, &sc->mly_packetmap)) { return(ENOMEM); } if (bus_dmamap_load(sc->mly_packet_dmat, sc->mly_packetmap, sc->mly_packet, ncmd * sizeof(union mly_command_packet), mly_alloc_commands_map, sc, BUS_DMA_NOWAIT) != 0) return (ENOMEM); for (i = 0; i < ncmd; i++) { mc = &sc->mly_command[i]; bzero(mc, sizeof(*mc)); mc->mc_sc = sc; mc->mc_slot = MLY_SLOT_START + i; mc->mc_packet = sc->mly_packet + i; mc->mc_packetphys = sc->mly_packetphys + (i * sizeof(union mly_command_packet)); if (!bus_dmamap_create(sc->mly_buffer_dmat, 0, &mc->mc_datamap)) mly_release_command(mc); } return(0); } /******************************************************************************** * Free all the storage held by commands. * * Must be called with all commands on the free list. */ static void mly_release_commands(struct mly_softc *sc) { struct mly_command *mc; /* throw away command buffer DMA maps */ while (mly_alloc_command(sc, &mc) == 0) bus_dmamap_destroy(sc->mly_buffer_dmat, mc->mc_datamap); /* release the packet storage */ if (sc->mly_packet != NULL) { bus_dmamap_unload(sc->mly_packet_dmat, sc->mly_packetmap); bus_dmamem_free(sc->mly_packet_dmat, sc->mly_packet, sc->mly_packetmap); sc->mly_packet = NULL; } } /******************************************************************************** * Command-mapping helper function - populate this command's s/g table * with the s/g entries for its data. */ static void mly_map_command_sg(void *arg, bus_dma_segment_t *segs, int nseg, int error) { struct mly_command *mc = (struct mly_command *)arg; struct mly_softc *sc = mc->mc_sc; struct mly_command_generic *gen = &(mc->mc_packet->generic); struct mly_sg_entry *sg; int i, tabofs; debug_called(2); /* can we use the transfer structure directly? */ if (nseg <= 2) { sg = &gen->transfer.direct.sg[0]; gen->command_control.extended_sg_table = 0; } else { tabofs = ((mc->mc_slot - MLY_SLOT_START) * MLY_MAX_SGENTRIES); sg = sc->mly_sg_table + tabofs; gen->transfer.indirect.entries[0] = nseg; gen->transfer.indirect.table_physaddr[0] = sc->mly_sg_busaddr + (tabofs * sizeof(struct mly_sg_entry)); gen->command_control.extended_sg_table = 1; } /* copy the s/g table */ for (i = 0; i < nseg; i++) { sg[i].physaddr = segs[i].ds_addr; sg[i].length = segs[i].ds_len; } } #if 0 /******************************************************************************** * Command-mapping helper function - save the cdb's physical address. * * We don't support 'large' SCSI commands at this time, so this is unused. */ static void mly_map_command_cdb(void *arg, bus_dma_segment_t *segs, int nseg, int error) { struct mly_command *mc = (struct mly_command *)arg; debug_called(2); /* XXX can we safely assume that a CDB will never cross a page boundary? */ if ((segs[0].ds_addr % PAGE_SIZE) > ((segs[0].ds_addr + mc->mc_packet->scsi_large.cdb_length) % PAGE_SIZE)) panic("cdb crosses page boundary"); /* fix up fields in the command packet */ mc->mc_packet->scsi_large.cdb_physaddr = segs[0].ds_addr; } #endif /******************************************************************************** * Map a command into controller-visible space */ static void mly_map_command(struct mly_command *mc) { struct mly_softc *sc = mc->mc_sc; debug_called(2); /* don't map more than once */ if (mc->mc_flags & MLY_CMD_MAPPED) return; /* does the command have a data buffer? */ if (mc->mc_data != NULL) { if (mc->mc_flags & MLY_CMD_CCB) bus_dmamap_load_ccb(sc->mly_buffer_dmat, mc->mc_datamap, mc->mc_data, mly_map_command_sg, mc, 0); else bus_dmamap_load(sc->mly_buffer_dmat, mc->mc_datamap, mc->mc_data, mc->mc_length, mly_map_command_sg, mc, 0); if (mc->mc_flags & MLY_CMD_DATAIN) bus_dmamap_sync(sc->mly_buffer_dmat, mc->mc_datamap, BUS_DMASYNC_PREREAD); if (mc->mc_flags & MLY_CMD_DATAOUT) bus_dmamap_sync(sc->mly_buffer_dmat, mc->mc_datamap, BUS_DMASYNC_PREWRITE); } mc->mc_flags |= MLY_CMD_MAPPED; } /******************************************************************************** * Unmap a command from controller-visible space */ static void mly_unmap_command(struct mly_command *mc) { struct mly_softc *sc = mc->mc_sc; debug_called(2); if (!(mc->mc_flags & MLY_CMD_MAPPED)) return; /* does the command have a data buffer? */ if (mc->mc_data != NULL) { if (mc->mc_flags & MLY_CMD_DATAIN) bus_dmamap_sync(sc->mly_buffer_dmat, mc->mc_datamap, BUS_DMASYNC_POSTREAD); if (mc->mc_flags & MLY_CMD_DATAOUT) bus_dmamap_sync(sc->mly_buffer_dmat, mc->mc_datamap, BUS_DMASYNC_POSTWRITE); bus_dmamap_unload(sc->mly_buffer_dmat, mc->mc_datamap); } mc->mc_flags &= ~MLY_CMD_MAPPED; } /******************************************************************************** ******************************************************************************** CAM interface ******************************************************************************** ********************************************************************************/ /******************************************************************************** * Attach the physical and virtual SCSI busses to CAM. * * Physical bus numbering starts from 0, virtual bus numbering from one greater * than the highest physical bus. Physical busses are only registered if * the kernel environment variable "hw.mly.register_physical_channels" is set. * * When we refer to a "bus", we are referring to the bus number registered with * the SIM, whereas a "channel" is a channel number given to the adapter. In order * to keep things simple, we map these 1:1, so "bus" and "channel" may be used * interchangeably. */ static int mly_cam_attach(struct mly_softc *sc) { struct cam_devq *devq; int chn, i; debug_called(1); /* * Allocate a devq for all our channels combined. */ if ((devq = cam_simq_alloc(sc->mly_controllerinfo->maximum_parallel_commands)) == NULL) { mly_printf(sc, "can't allocate CAM SIM queue\n"); return(ENOMEM); } /* * If physical channel registration has been requested, register these first. * Note that we enable tagged command queueing for physical channels. */ if (testenv("hw.mly.register_physical_channels")) { chn = 0; for (i = 0; i < sc->mly_controllerinfo->physical_channels_present; i++, chn++) { if ((sc->mly_cam_sim[chn] = cam_sim_alloc(mly_cam_action, mly_cam_poll, "mly", sc, device_get_unit(sc->mly_dev), &sc->mly_lock, sc->mly_controllerinfo->maximum_parallel_commands, 1, devq)) == NULL) { return(ENOMEM); } MLY_LOCK(sc); if (xpt_bus_register(sc->mly_cam_sim[chn], sc->mly_dev, chn)) { MLY_UNLOCK(sc); mly_printf(sc, "CAM XPT phsyical channel registration failed\n"); return(ENXIO); } MLY_UNLOCK(sc); debug(1, "registered physical channel %d", chn); } } /* * Register our virtual channels, with bus numbers matching channel numbers. */ chn = sc->mly_controllerinfo->physical_channels_present; for (i = 0; i < sc->mly_controllerinfo->virtual_channels_present; i++, chn++) { if ((sc->mly_cam_sim[chn] = cam_sim_alloc(mly_cam_action, mly_cam_poll, "mly", sc, device_get_unit(sc->mly_dev), &sc->mly_lock, sc->mly_controllerinfo->maximum_parallel_commands, 0, devq)) == NULL) { return(ENOMEM); } MLY_LOCK(sc); if (xpt_bus_register(sc->mly_cam_sim[chn], sc->mly_dev, chn)) { MLY_UNLOCK(sc); mly_printf(sc, "CAM XPT virtual channel registration failed\n"); return(ENXIO); } MLY_UNLOCK(sc); debug(1, "registered virtual channel %d", chn); } /* * This is the total number of channels that (might have been) registered with * CAM. Some may not have been; check the mly_cam_sim array to be certain. */ sc->mly_cam_channels = sc->mly_controllerinfo->physical_channels_present + sc->mly_controllerinfo->virtual_channels_present; return(0); } /******************************************************************************** * Detach from CAM */ static void mly_cam_detach(struct mly_softc *sc) { int i; debug_called(1); MLY_LOCK(sc); for (i = 0; i < sc->mly_cam_channels; i++) { if (sc->mly_cam_sim[i] != NULL) { xpt_bus_deregister(cam_sim_path(sc->mly_cam_sim[i])); cam_sim_free(sc->mly_cam_sim[i], 0); } } MLY_UNLOCK(sc); if (sc->mly_cam_devq != NULL) cam_simq_free(sc->mly_cam_devq); } /************************************************************************ * Rescan a device. */ static void mly_cam_rescan_btl(struct mly_softc *sc, int bus, int target) { union ccb *ccb; debug_called(1); if ((ccb = xpt_alloc_ccb()) == NULL) { mly_printf(sc, "rescan failed (can't allocate CCB)\n"); return; } if (xpt_create_path(&ccb->ccb_h.path, NULL, cam_sim_path(sc->mly_cam_sim[bus]), target, 0) != CAM_REQ_CMP) { mly_printf(sc, "rescan failed (can't create path)\n"); xpt_free_ccb(ccb); return; } debug(1, "rescan target %d:%d", bus, target); xpt_rescan(ccb); } /******************************************************************************** * Handle an action requested by CAM */ static void mly_cam_action(struct cam_sim *sim, union ccb *ccb) { struct mly_softc *sc = cam_sim_softc(sim); debug_called(2); MLY_ASSERT_LOCKED(sc); switch (ccb->ccb_h.func_code) { /* perform SCSI I/O */ case XPT_SCSI_IO: if (!mly_cam_action_io(sim, (struct ccb_scsiio *)&ccb->csio)) return; break; /* perform geometry calculations */ case XPT_CALC_GEOMETRY: { struct ccb_calc_geometry *ccg = &ccb->ccg; u_int32_t secs_per_cylinder; debug(2, "XPT_CALC_GEOMETRY %d:%d:%d", cam_sim_bus(sim), ccb->ccb_h.target_id, ccb->ccb_h.target_lun); if (sc->mly_controllerparam->bios_geometry == MLY_BIOSGEOM_8G) { ccg->heads = 255; ccg->secs_per_track = 63; } else { /* MLY_BIOSGEOM_2G */ ccg->heads = 128; ccg->secs_per_track = 32; } secs_per_cylinder = ccg->heads * ccg->secs_per_track; ccg->cylinders = ccg->volume_size / secs_per_cylinder; ccb->ccb_h.status = CAM_REQ_CMP; break; } /* handle path attribute inquiry */ case XPT_PATH_INQ: { struct ccb_pathinq *cpi = &ccb->cpi; debug(2, "XPT_PATH_INQ %d:%d:%d", cam_sim_bus(sim), ccb->ccb_h.target_id, ccb->ccb_h.target_lun); cpi->version_num = 1; cpi->hba_inquiry = PI_TAG_ABLE; /* XXX extra flags for physical channels? */ cpi->target_sprt = 0; cpi->hba_misc = 0; cpi->max_target = MLY_MAX_TARGETS - 1; cpi->max_lun = MLY_MAX_LUNS - 1; cpi->initiator_id = sc->mly_controllerparam->initiator_id; strlcpy(cpi->sim_vid, "FreeBSD", SIM_IDLEN); - strlcpy(cpi->hba_vid, "Mylex", HBA_IDLEN); - strlcpy(cpi->dev_name, cam_sim_name(sim), DEV_IDLEN); - cpi->unit_number = cam_sim_unit(sim); - cpi->bus_id = cam_sim_bus(sim); + strlcpy(cpi->hba_vid, "Mylex", HBA_IDLEN); + strlcpy(cpi->dev_name, cam_sim_name(sim), DEV_IDLEN); + cpi->unit_number = cam_sim_unit(sim); + cpi->bus_id = cam_sim_bus(sim); cpi->base_transfer_speed = 132 * 1024; /* XXX what to set this to? */ cpi->transport = XPORT_SPI; cpi->transport_version = 2; cpi->protocol = PROTO_SCSI; cpi->protocol_version = SCSI_REV_2; ccb->ccb_h.status = CAM_REQ_CMP; break; } case XPT_GET_TRAN_SETTINGS: { struct ccb_trans_settings *cts = &ccb->cts; int bus, target; struct ccb_trans_settings_scsi *scsi = &cts->proto_specific.scsi; struct ccb_trans_settings_spi *spi = &cts->xport_specific.spi; cts->protocol = PROTO_SCSI; cts->protocol_version = SCSI_REV_2; cts->transport = XPORT_SPI; cts->transport_version = 2; scsi->flags = 0; scsi->valid = 0; spi->flags = 0; spi->valid = 0; bus = cam_sim_bus(sim); target = cts->ccb_h.target_id; debug(2, "XPT_GET_TRAN_SETTINGS %d:%d", bus, target); /* logical device? */ if (sc->mly_btl[bus][target].mb_flags & MLY_BTL_LOGICAL) { /* nothing special for these */ /* physical device? */ } else if (sc->mly_btl[bus][target].mb_flags & MLY_BTL_PHYSICAL) { /* allow CAM to try tagged transactions */ scsi->flags |= CTS_SCSI_FLAGS_TAG_ENB; scsi->valid |= CTS_SCSI_VALID_TQ; /* convert speed (MHz) to usec */ if (sc->mly_btl[bus][target].mb_speed == 0) { spi->sync_period = 1000000 / 5; } else { spi->sync_period = 1000000 / sc->mly_btl[bus][target].mb_speed; } /* convert bus width to CAM internal encoding */ switch (sc->mly_btl[bus][target].mb_width) { case 32: spi->bus_width = MSG_EXT_WDTR_BUS_32_BIT; break; case 16: spi->bus_width = MSG_EXT_WDTR_BUS_16_BIT; break; case 8: default: spi->bus_width = MSG_EXT_WDTR_BUS_8_BIT; break; } spi->valid |= CTS_SPI_VALID_SYNC_RATE | CTS_SPI_VALID_BUS_WIDTH; /* not a device, bail out */ } else { cts->ccb_h.status = CAM_REQ_CMP_ERR; break; } /* disconnect always OK */ spi->flags |= CTS_SPI_FLAGS_DISC_ENB; spi->valid |= CTS_SPI_VALID_DISC; cts->ccb_h.status = CAM_REQ_CMP; break; } default: /* we can't do this */ debug(2, "unspported func_code = 0x%x", ccb->ccb_h.func_code); ccb->ccb_h.status = CAM_REQ_INVALID; break; } xpt_done(ccb); } /******************************************************************************** * Handle an I/O operation requested by CAM */ static int mly_cam_action_io(struct cam_sim *sim, struct ccb_scsiio *csio) { struct mly_softc *sc = cam_sim_softc(sim); struct mly_command *mc; struct mly_command_scsi_small *ss; int bus, target; int error; bus = cam_sim_bus(sim); target = csio->ccb_h.target_id; debug(2, "XPT_SCSI_IO %d:%d:%d", bus, target, csio->ccb_h.target_lun); /* validate bus number */ if (!MLY_BUS_IS_VALID(sc, bus)) { debug(0, " invalid bus %d", bus); csio->ccb_h.status = CAM_REQ_CMP_ERR; } /* check for I/O attempt to a protected device */ if (sc->mly_btl[bus][target].mb_flags & MLY_BTL_PROTECTED) { debug(2, " device protected"); csio->ccb_h.status = CAM_REQ_CMP_ERR; } /* check for I/O attempt to nonexistent device */ if (!(sc->mly_btl[bus][target].mb_flags & (MLY_BTL_LOGICAL | MLY_BTL_PHYSICAL))) { debug(2, " device %d:%d does not exist", bus, target); csio->ccb_h.status = CAM_REQ_CMP_ERR; } /* XXX increase if/when we support large SCSI commands */ if (csio->cdb_len > MLY_CMD_SCSI_SMALL_CDB) { debug(0, " command too large (%d > %d)", csio->cdb_len, MLY_CMD_SCSI_SMALL_CDB); csio->ccb_h.status = CAM_REQ_CMP_ERR; } /* check that the CDB pointer is not to a physical address */ if ((csio->ccb_h.flags & CAM_CDB_POINTER) && (csio->ccb_h.flags & CAM_CDB_PHYS)) { debug(0, " CDB pointer is to physical address"); csio->ccb_h.status = CAM_REQ_CMP_ERR; } /* abandon aborted ccbs or those that have failed validation */ if ((csio->ccb_h.status & CAM_STATUS_MASK) != CAM_REQ_INPROG) { debug(2, "abandoning CCB due to abort/validation failure"); return(EINVAL); } /* * Get a command, or push the ccb back to CAM and freeze the queue. */ if ((error = mly_alloc_command(sc, &mc))) { xpt_freeze_simq(sim, 1); csio->ccb_h.status |= CAM_REQUEUE_REQ; sc->mly_qfrzn_cnt++; return(error); } /* build the command */ mc->mc_data = csio; mc->mc_length = csio->dxfer_len; mc->mc_complete = mly_cam_complete; mc->mc_private = csio; mc->mc_flags |= MLY_CMD_CCB; /* XXX This code doesn't set the data direction in mc_flags. */ /* save the bus number in the ccb for later recovery XXX should be a better way */ csio->ccb_h.sim_priv.entries[0].field = bus; /* build the packet for the controller */ ss = &mc->mc_packet->scsi_small; ss->opcode = MDACMD_SCSI; if (csio->ccb_h.flags & CAM_DIS_DISCONNECT) ss->command_control.disable_disconnect = 1; if ((csio->ccb_h.flags & CAM_DIR_MASK) == CAM_DIR_OUT) ss->command_control.data_direction = MLY_CCB_WRITE; ss->data_size = csio->dxfer_len; ss->addr.phys.lun = csio->ccb_h.target_lun; ss->addr.phys.target = csio->ccb_h.target_id; ss->addr.phys.channel = bus; if (csio->ccb_h.timeout < (60 * 1000)) { ss->timeout.value = csio->ccb_h.timeout / 1000; ss->timeout.scale = MLY_TIMEOUT_SECONDS; } else if (csio->ccb_h.timeout < (60 * 60 * 1000)) { ss->timeout.value = csio->ccb_h.timeout / (60 * 1000); ss->timeout.scale = MLY_TIMEOUT_MINUTES; } else { ss->timeout.value = csio->ccb_h.timeout / (60 * 60 * 1000); /* overflow? */ ss->timeout.scale = MLY_TIMEOUT_HOURS; } ss->maximum_sense_size = csio->sense_len; ss->cdb_length = csio->cdb_len; if (csio->ccb_h.flags & CAM_CDB_POINTER) { bcopy(csio->cdb_io.cdb_ptr, ss->cdb, csio->cdb_len); } else { bcopy(csio->cdb_io.cdb_bytes, ss->cdb, csio->cdb_len); } /* give the command to the controller */ if ((error = mly_start(mc))) { xpt_freeze_simq(sim, 1); csio->ccb_h.status |= CAM_REQUEUE_REQ; sc->mly_qfrzn_cnt++; return(error); } return(0); } /******************************************************************************** * Check for possibly-completed commands. */ static void mly_cam_poll(struct cam_sim *sim) { struct mly_softc *sc = cam_sim_softc(sim); debug_called(2); mly_done(sc); } /******************************************************************************** * Handle completion of a command - pass results back through the CCB */ static void mly_cam_complete(struct mly_command *mc) { struct mly_softc *sc = mc->mc_sc; struct ccb_scsiio *csio = (struct ccb_scsiio *)mc->mc_private; struct scsi_inquiry_data *inq = (struct scsi_inquiry_data *)csio->data_ptr; struct mly_btl *btl; u_int8_t cmd; int bus, target; debug_called(2); csio->scsi_status = mc->mc_status; switch(mc->mc_status) { case SCSI_STATUS_OK: /* * In order to report logical device type and status, we overwrite * the result of the INQUIRY command to logical devices. */ bus = csio->ccb_h.sim_priv.entries[0].field; target = csio->ccb_h.target_id; /* XXX validate bus/target? */ if (sc->mly_btl[bus][target].mb_flags & MLY_BTL_LOGICAL) { if (csio->ccb_h.flags & CAM_CDB_POINTER) { cmd = *csio->cdb_io.cdb_ptr; } else { cmd = csio->cdb_io.cdb_bytes[0]; } if (cmd == INQUIRY) { btl = &sc->mly_btl[bus][target]; padstr(inq->vendor, mly_describe_code(mly_table_device_type, btl->mb_type), 8); padstr(inq->product, mly_describe_code(mly_table_device_state, btl->mb_state), 16); padstr(inq->revision, "", 4); } } debug(2, "SCSI_STATUS_OK"); csio->ccb_h.status = CAM_REQ_CMP; break; case SCSI_STATUS_CHECK_COND: debug(1, "SCSI_STATUS_CHECK_COND sense %d resid %d", mc->mc_sense, mc->mc_resid); csio->ccb_h.status = CAM_SCSI_STATUS_ERROR; bzero(&csio->sense_data, SSD_FULL_SIZE); bcopy(mc->mc_packet, &csio->sense_data, mc->mc_sense); csio->sense_len = mc->mc_sense; csio->ccb_h.status |= CAM_AUTOSNS_VALID; csio->resid = mc->mc_resid; /* XXX this is a signed value... */ break; case SCSI_STATUS_BUSY: debug(1, "SCSI_STATUS_BUSY"); csio->ccb_h.status = CAM_SCSI_BUSY; break; default: debug(1, "unknown status 0x%x", csio->scsi_status); csio->ccb_h.status = CAM_REQ_CMP_ERR; break; } if (sc->mly_qfrzn_cnt) { csio->ccb_h.status |= CAM_RELEASE_SIMQ; sc->mly_qfrzn_cnt--; } xpt_done((union ccb *)csio); mly_release_command(mc); } /******************************************************************************** * Find a peripheral attahed at (bus),(target) */ static struct cam_periph * mly_find_periph(struct mly_softc *sc, int bus, int target) { struct cam_periph *periph; struct cam_path *path; int status; status = xpt_create_path(&path, NULL, cam_sim_path(sc->mly_cam_sim[bus]), target, 0); if (status == CAM_REQ_CMP) { periph = cam_periph_find(path, NULL); xpt_free_path(path); } else { periph = NULL; } return(periph); } /******************************************************************************** * Name the device at (bus)(target) */ static int mly_name_device(struct mly_softc *sc, int bus, int target) { struct cam_periph *periph; if ((periph = mly_find_periph(sc, bus, target)) != NULL) { sprintf(sc->mly_btl[bus][target].mb_name, "%s%d", periph->periph_name, periph->unit_number); return(0); } sc->mly_btl[bus][target].mb_name[0] = 0; return(ENOENT); } /******************************************************************************** ******************************************************************************** Hardware Control ******************************************************************************** ********************************************************************************/ /******************************************************************************** * Handshake with the firmware while the card is being initialised. */ static int mly_fwhandshake(struct mly_softc *sc) { u_int8_t error, param0, param1; int spinup = 0; debug_called(1); /* set HM_STSACK and let the firmware initialise */ MLY_SET_REG(sc, sc->mly_idbr, MLY_HM_STSACK); DELAY(1000); /* too short? */ /* if HM_STSACK is still true, the controller is initialising */ if (!MLY_IDBR_TRUE(sc, MLY_HM_STSACK)) return(0); mly_printf(sc, "controller initialisation started\n"); /* spin waiting for initialisation to finish, or for a message to be delivered */ while (MLY_IDBR_TRUE(sc, MLY_HM_STSACK)) { /* check for a message */ if (MLY_ERROR_VALID(sc)) { error = MLY_GET_REG(sc, sc->mly_error_status) & ~MLY_MSG_EMPTY; param0 = MLY_GET_REG(sc, sc->mly_command_mailbox); param1 = MLY_GET_REG(sc, sc->mly_command_mailbox + 1); switch(error) { case MLY_MSG_SPINUP: if (!spinup) { mly_printf(sc, "drive spinup in progress\n"); spinup = 1; /* only print this once (should print drive being spun?) */ } break; case MLY_MSG_RACE_RECOVERY_FAIL: mly_printf(sc, "mirror race recovery failed, one or more drives offline\n"); break; case MLY_MSG_RACE_IN_PROGRESS: mly_printf(sc, "mirror race recovery in progress\n"); break; case MLY_MSG_RACE_ON_CRITICAL: mly_printf(sc, "mirror race recovery on a critical drive\n"); break; case MLY_MSG_PARITY_ERROR: mly_printf(sc, "FATAL MEMORY PARITY ERROR\n"); return(ENXIO); default: mly_printf(sc, "unknown initialisation code 0x%x\n", error); } } } return(0); } /******************************************************************************** ******************************************************************************** Debugging and Diagnostics ******************************************************************************** ********************************************************************************/ /******************************************************************************** * Print some information about the controller. */ static void mly_describe_controller(struct mly_softc *sc) { struct mly_ioctl_getcontrollerinfo *mi = sc->mly_controllerinfo; mly_printf(sc, "%16s, %d channel%s, firmware %d.%02d-%d-%02d (%02d%02d%02d%02d), %dMB RAM\n", mi->controller_name, mi->physical_channels_present, (mi->physical_channels_present) > 1 ? "s" : "", mi->fw_major, mi->fw_minor, mi->fw_turn, mi->fw_build, /* XXX turn encoding? */ mi->fw_century, mi->fw_year, mi->fw_month, mi->fw_day, mi->memory_size); if (bootverbose) { mly_printf(sc, "%s %s (%x), %dMHz %d-bit %.16s\n", mly_describe_code(mly_table_oemname, mi->oem_information), mly_describe_code(mly_table_controllertype, mi->controller_type), mi->controller_type, mi->interface_speed, mi->interface_width, mi->interface_name); mly_printf(sc, "%dMB %dMHz %d-bit %s%s%s, cache %dMB\n", mi->memory_size, mi->memory_speed, mi->memory_width, mly_describe_code(mly_table_memorytype, mi->memory_type), mi->memory_parity ? "+parity": "",mi->memory_ecc ? "+ECC": "", mi->cache_size); mly_printf(sc, "CPU: %s @ %dMHz\n", mly_describe_code(mly_table_cputype, mi->cpu[0].type), mi->cpu[0].speed); if (mi->l2cache_size != 0) mly_printf(sc, "%dKB L2 cache\n", mi->l2cache_size); if (mi->exmemory_size != 0) mly_printf(sc, "%dMB %dMHz %d-bit private %s%s%s\n", mi->exmemory_size, mi->exmemory_speed, mi->exmemory_width, mly_describe_code(mly_table_memorytype, mi->exmemory_type), mi->exmemory_parity ? "+parity": "",mi->exmemory_ecc ? "+ECC": ""); mly_printf(sc, "battery backup %s\n", mi->bbu_present ? "present" : "not installed"); mly_printf(sc, "maximum data transfer %d blocks, maximum sg entries/command %d\n", mi->maximum_block_count, mi->maximum_sg_entries); mly_printf(sc, "logical devices present/critical/offline %d/%d/%d\n", mi->logical_devices_present, mi->logical_devices_critical, mi->logical_devices_offline); mly_printf(sc, "physical devices present %d\n", mi->physical_devices_present); mly_printf(sc, "physical disks present/offline %d/%d\n", mi->physical_disks_present, mi->physical_disks_offline); mly_printf(sc, "%d physical channel%s, %d virtual channel%s of %d possible\n", mi->physical_channels_present, mi->physical_channels_present == 1 ? "" : "s", mi->virtual_channels_present, mi->virtual_channels_present == 1 ? "" : "s", mi->virtual_channels_possible); mly_printf(sc, "%d parallel commands supported\n", mi->maximum_parallel_commands); mly_printf(sc, "%dMB flash ROM, %d of %d maximum cycles\n", mi->flash_size, mi->flash_age, mi->flash_maximum_age); } } #ifdef MLY_DEBUG /******************************************************************************** * Print some controller state */ static void mly_printstate(struct mly_softc *sc) { mly_printf(sc, "IDBR %02x ODBR %02x ERROR %02x (%x %x %x)\n", MLY_GET_REG(sc, sc->mly_idbr), MLY_GET_REG(sc, sc->mly_odbr), MLY_GET_REG(sc, sc->mly_error_status), sc->mly_idbr, sc->mly_odbr, sc->mly_error_status); mly_printf(sc, "IMASK %02x ISTATUS %02x\n", MLY_GET_REG(sc, sc->mly_interrupt_mask), MLY_GET_REG(sc, sc->mly_interrupt_status)); mly_printf(sc, "COMMAND %02x %02x %02x %02x %02x %02x %02x %02x\n", MLY_GET_REG(sc, sc->mly_command_mailbox), MLY_GET_REG(sc, sc->mly_command_mailbox + 1), MLY_GET_REG(sc, sc->mly_command_mailbox + 2), MLY_GET_REG(sc, sc->mly_command_mailbox + 3), MLY_GET_REG(sc, sc->mly_command_mailbox + 4), MLY_GET_REG(sc, sc->mly_command_mailbox + 5), MLY_GET_REG(sc, sc->mly_command_mailbox + 6), MLY_GET_REG(sc, sc->mly_command_mailbox + 7)); mly_printf(sc, "STATUS %02x %02x %02x %02x %02x %02x %02x %02x\n", MLY_GET_REG(sc, sc->mly_status_mailbox), MLY_GET_REG(sc, sc->mly_status_mailbox + 1), MLY_GET_REG(sc, sc->mly_status_mailbox + 2), MLY_GET_REG(sc, sc->mly_status_mailbox + 3), MLY_GET_REG(sc, sc->mly_status_mailbox + 4), MLY_GET_REG(sc, sc->mly_status_mailbox + 5), MLY_GET_REG(sc, sc->mly_status_mailbox + 6), MLY_GET_REG(sc, sc->mly_status_mailbox + 7)); mly_printf(sc, " %04x %08x\n", MLY_GET_REG2(sc, sc->mly_status_mailbox), MLY_GET_REG4(sc, sc->mly_status_mailbox + 4)); } struct mly_softc *mly_softc0 = NULL; void mly_printstate0(void) { if (mly_softc0 != NULL) mly_printstate(mly_softc0); } /******************************************************************************** * Print a command */ static void mly_print_command(struct mly_command *mc) { struct mly_softc *sc = mc->mc_sc; mly_printf(sc, "COMMAND @ %p\n", mc); mly_printf(sc, " slot %d\n", mc->mc_slot); mly_printf(sc, " status 0x%x\n", mc->mc_status); mly_printf(sc, " sense len %d\n", mc->mc_sense); mly_printf(sc, " resid %d\n", mc->mc_resid); mly_printf(sc, " packet %p/0x%llx\n", mc->mc_packet, mc->mc_packetphys); if (mc->mc_packet != NULL) mly_print_packet(mc); mly_printf(sc, " data %p/%d\n", mc->mc_data, mc->mc_length); mly_printf(sc, " flags %b\n", mc->mc_flags, "\20\1busy\2complete\3slotted\4mapped\5datain\6dataout\n"); mly_printf(sc, " complete %p\n", mc->mc_complete); mly_printf(sc, " private %p\n", mc->mc_private); } /******************************************************************************** * Print a command packet */ static void mly_print_packet(struct mly_command *mc) { struct mly_softc *sc = mc->mc_sc; struct mly_command_generic *ge = (struct mly_command_generic *)mc->mc_packet; struct mly_command_scsi_small *ss = (struct mly_command_scsi_small *)mc->mc_packet; struct mly_command_scsi_large *sl = (struct mly_command_scsi_large *)mc->mc_packet; struct mly_command_ioctl *io = (struct mly_command_ioctl *)mc->mc_packet; int transfer; mly_printf(sc, " command_id %d\n", ge->command_id); mly_printf(sc, " opcode %d\n", ge->opcode); mly_printf(sc, " command_control fua %d dpo %d est %d dd %s nas %d ddis %d\n", ge->command_control.force_unit_access, ge->command_control.disable_page_out, ge->command_control.extended_sg_table, (ge->command_control.data_direction == MLY_CCB_WRITE) ? "WRITE" : "READ", ge->command_control.no_auto_sense, ge->command_control.disable_disconnect); mly_printf(sc, " data_size %d\n", ge->data_size); mly_printf(sc, " sense_buffer_address 0x%llx\n", ge->sense_buffer_address); mly_printf(sc, " lun %d\n", ge->addr.phys.lun); mly_printf(sc, " target %d\n", ge->addr.phys.target); mly_printf(sc, " channel %d\n", ge->addr.phys.channel); mly_printf(sc, " logical device %d\n", ge->addr.log.logdev); mly_printf(sc, " controller %d\n", ge->addr.phys.controller); mly_printf(sc, " timeout %d %s\n", ge->timeout.value, (ge->timeout.scale == MLY_TIMEOUT_SECONDS) ? "seconds" : ((ge->timeout.scale == MLY_TIMEOUT_MINUTES) ? "minutes" : "hours")); mly_printf(sc, " maximum_sense_size %d\n", ge->maximum_sense_size); switch(ge->opcode) { case MDACMD_SCSIPT: case MDACMD_SCSI: mly_printf(sc, " cdb length %d\n", ss->cdb_length); mly_printf(sc, " cdb %*D\n", ss->cdb_length, ss->cdb, " "); transfer = 1; break; case MDACMD_SCSILC: case MDACMD_SCSILCPT: mly_printf(sc, " cdb length %d\n", sl->cdb_length); mly_printf(sc, " cdb 0x%llx\n", sl->cdb_physaddr); transfer = 1; break; case MDACMD_IOCTL: mly_printf(sc, " sub_ioctl 0x%x\n", io->sub_ioctl); switch(io->sub_ioctl) { case MDACIOCTL_SETMEMORYMAILBOX: mly_printf(sc, " health_buffer_size %d\n", io->param.setmemorymailbox.health_buffer_size); mly_printf(sc, " health_buffer_phys 0x%llx\n", io->param.setmemorymailbox.health_buffer_physaddr); mly_printf(sc, " command_mailbox 0x%llx\n", io->param.setmemorymailbox.command_mailbox_physaddr); mly_printf(sc, " status_mailbox 0x%llx\n", io->param.setmemorymailbox.status_mailbox_physaddr); transfer = 0; break; case MDACIOCTL_SETREALTIMECLOCK: case MDACIOCTL_GETHEALTHSTATUS: case MDACIOCTL_GETCONTROLLERINFO: case MDACIOCTL_GETLOGDEVINFOVALID: case MDACIOCTL_GETPHYSDEVINFOVALID: case MDACIOCTL_GETPHYSDEVSTATISTICS: case MDACIOCTL_GETLOGDEVSTATISTICS: case MDACIOCTL_GETCONTROLLERSTATISTICS: case MDACIOCTL_GETBDT_FOR_SYSDRIVE: case MDACIOCTL_CREATENEWCONF: case MDACIOCTL_ADDNEWCONF: case MDACIOCTL_GETDEVCONFINFO: case MDACIOCTL_GETFREESPACELIST: case MDACIOCTL_MORE: case MDACIOCTL_SETPHYSDEVPARAMETER: case MDACIOCTL_GETPHYSDEVPARAMETER: case MDACIOCTL_GETLOGDEVPARAMETER: case MDACIOCTL_SETLOGDEVPARAMETER: mly_printf(sc, " param %10D\n", io->param.data.param, " "); transfer = 1; break; case MDACIOCTL_GETEVENT: mly_printf(sc, " event %d\n", io->param.getevent.sequence_number_low + ((u_int32_t)io->addr.log.logdev << 16)); transfer = 1; break; case MDACIOCTL_SETRAIDDEVSTATE: mly_printf(sc, " state %d\n", io->param.setraiddevstate.state); transfer = 0; break; case MDACIOCTL_XLATEPHYSDEVTORAIDDEV: mly_printf(sc, " raid_device %d\n", io->param.xlatephysdevtoraiddev.raid_device); mly_printf(sc, " controller %d\n", io->param.xlatephysdevtoraiddev.controller); mly_printf(sc, " channel %d\n", io->param.xlatephysdevtoraiddev.channel); mly_printf(sc, " target %d\n", io->param.xlatephysdevtoraiddev.target); mly_printf(sc, " lun %d\n", io->param.xlatephysdevtoraiddev.lun); transfer = 0; break; case MDACIOCTL_GETGROUPCONFINFO: mly_printf(sc, " group %d\n", io->param.getgroupconfinfo.group); transfer = 1; break; case MDACIOCTL_GET_SUBSYSTEM_DATA: case MDACIOCTL_SET_SUBSYSTEM_DATA: case MDACIOCTL_STARTDISOCVERY: case MDACIOCTL_INITPHYSDEVSTART: case MDACIOCTL_INITPHYSDEVSTOP: case MDACIOCTL_INITRAIDDEVSTART: case MDACIOCTL_INITRAIDDEVSTOP: case MDACIOCTL_REBUILDRAIDDEVSTART: case MDACIOCTL_REBUILDRAIDDEVSTOP: case MDACIOCTL_MAKECONSISTENTDATASTART: case MDACIOCTL_MAKECONSISTENTDATASTOP: case MDACIOCTL_CONSISTENCYCHECKSTART: case MDACIOCTL_CONSISTENCYCHECKSTOP: case MDACIOCTL_RESETDEVICE: case MDACIOCTL_FLUSHDEVICEDATA: case MDACIOCTL_PAUSEDEVICE: case MDACIOCTL_UNPAUSEDEVICE: case MDACIOCTL_LOCATEDEVICE: case MDACIOCTL_SETMASTERSLAVEMODE: case MDACIOCTL_DELETERAIDDEV: case MDACIOCTL_REPLACEINTERNALDEV: case MDACIOCTL_CLEARCONF: case MDACIOCTL_GETCONTROLLERPARAMETER: case MDACIOCTL_SETCONTRLLERPARAMETER: case MDACIOCTL_CLEARCONFSUSPMODE: case MDACIOCTL_STOREIMAGE: case MDACIOCTL_READIMAGE: case MDACIOCTL_FLASHIMAGES: case MDACIOCTL_RENAMERAIDDEV: default: /* no idea what to print */ transfer = 0; break; } break; case MDACMD_IOCTLCHECK: case MDACMD_MEMCOPY: default: transfer = 0; break; /* print nothing */ } if (transfer) { if (ge->command_control.extended_sg_table) { mly_printf(sc, " sg table 0x%llx/%d\n", ge->transfer.indirect.table_physaddr[0], ge->transfer.indirect.entries[0]); } else { mly_printf(sc, " 0000 0x%llx/%lld\n", ge->transfer.direct.sg[0].physaddr, ge->transfer.direct.sg[0].length); mly_printf(sc, " 0001 0x%llx/%lld\n", ge->transfer.direct.sg[1].physaddr, ge->transfer.direct.sg[1].length); } } } /******************************************************************************** * Panic in a slightly informative fashion */ static void mly_panic(struct mly_softc *sc, char *reason) { mly_printstate(sc); panic(reason); } /******************************************************************************** * Print queue statistics, callable from DDB. */ void mly_print_controller(int controller) { struct mly_softc *sc; if ((sc = devclass_get_softc(devclass_find("mly"), controller)) == NULL) { printf("mly: controller %d invalid\n", controller); } else { device_printf(sc->mly_dev, "queue curr max\n"); device_printf(sc->mly_dev, "free %04d/%04d\n", sc->mly_qstat[MLYQ_FREE].q_length, sc->mly_qstat[MLYQ_FREE].q_max); device_printf(sc->mly_dev, "busy %04d/%04d\n", sc->mly_qstat[MLYQ_BUSY].q_length, sc->mly_qstat[MLYQ_BUSY].q_max); device_printf(sc->mly_dev, "complete %04d/%04d\n", sc->mly_qstat[MLYQ_COMPLETE].q_length, sc->mly_qstat[MLYQ_COMPLETE].q_max); } } #endif /******************************************************************************** ******************************************************************************** Control device interface ******************************************************************************** ********************************************************************************/ /******************************************************************************** * Accept an open operation on the control device. */ static int mly_user_open(struct cdev *dev, int flags, int fmt, struct thread *td) { struct mly_softc *sc = dev->si_drv1; MLY_LOCK(sc); sc->mly_state |= MLY_STATE_OPEN; MLY_UNLOCK(sc); return(0); } /******************************************************************************** * Accept the last close on the control device. */ static int mly_user_close(struct cdev *dev, int flags, int fmt, struct thread *td) { struct mly_softc *sc = dev->si_drv1; MLY_LOCK(sc); sc->mly_state &= ~MLY_STATE_OPEN; MLY_UNLOCK(sc); return (0); } /******************************************************************************** * Handle controller-specific control operations. */ static int mly_user_ioctl(struct cdev *dev, u_long cmd, caddr_t addr, int32_t flag, struct thread *td) { struct mly_softc *sc = (struct mly_softc *)dev->si_drv1; struct mly_user_command *uc = (struct mly_user_command *)addr; struct mly_user_health *uh = (struct mly_user_health *)addr; switch(cmd) { case MLYIO_COMMAND: return(mly_user_command(sc, uc)); case MLYIO_HEALTH: return(mly_user_health(sc, uh)); default: return(ENOIOCTL); } } /******************************************************************************** * Execute a command passed in from userspace. * * The control structure contains the actual command for the controller, as well * as the user-space data pointer and data size, and an optional sense buffer * size/pointer. On completion, the data size is adjusted to the command * residual, and the sense buffer size to the size of the returned sense data. * */ static int mly_user_command(struct mly_softc *sc, struct mly_user_command *uc) { struct mly_command *mc; int error; /* allocate a command */ MLY_LOCK(sc); if (mly_alloc_command(sc, &mc)) { MLY_UNLOCK(sc); error = ENOMEM; goto out; /* XXX Linux version will wait for a command */ } MLY_UNLOCK(sc); /* handle data size/direction */ mc->mc_length = (uc->DataTransferLength >= 0) ? uc->DataTransferLength : -uc->DataTransferLength; if (mc->mc_length > 0) { if ((mc->mc_data = malloc(mc->mc_length, M_DEVBUF, M_NOWAIT)) == NULL) { error = ENOMEM; goto out; } } if (uc->DataTransferLength > 0) { mc->mc_flags |= MLY_CMD_DATAIN; bzero(mc->mc_data, mc->mc_length); } if (uc->DataTransferLength < 0) { mc->mc_flags |= MLY_CMD_DATAOUT; if ((error = copyin(uc->DataTransferBuffer, mc->mc_data, mc->mc_length)) != 0) goto out; } /* copy the controller command */ bcopy(&uc->CommandMailbox, mc->mc_packet, sizeof(uc->CommandMailbox)); /* clear command completion handler so that we get woken up */ mc->mc_complete = NULL; /* execute the command */ MLY_LOCK(sc); if ((error = mly_start(mc)) != 0) { MLY_UNLOCK(sc); goto out; } while (!(mc->mc_flags & MLY_CMD_COMPLETE)) mtx_sleep(mc, &sc->mly_lock, PRIBIO, "mlyioctl", 0); MLY_UNLOCK(sc); /* return the data to userspace */ if (uc->DataTransferLength > 0) if ((error = copyout(mc->mc_data, uc->DataTransferBuffer, mc->mc_length)) != 0) goto out; /* return the sense buffer to userspace */ if ((uc->RequestSenseLength > 0) && (mc->mc_sense > 0)) { if ((error = copyout(mc->mc_packet, uc->RequestSenseBuffer, min(uc->RequestSenseLength, mc->mc_sense))) != 0) goto out; } /* return command results to userspace (caller will copy out) */ uc->DataTransferLength = mc->mc_resid; uc->RequestSenseLength = min(uc->RequestSenseLength, mc->mc_sense); uc->CommandStatus = mc->mc_status; error = 0; out: if (mc->mc_data != NULL) free(mc->mc_data, M_DEVBUF); if (mc != NULL) { MLY_LOCK(sc); mly_release_command(mc); MLY_UNLOCK(sc); } return(error); } /******************************************************************************** * Return health status to userspace. If the health change index in the user * structure does not match that currently exported by the controller, we * return the current status immediately. Otherwise, we block until either * interrupted or new status is delivered. */ static int mly_user_health(struct mly_softc *sc, struct mly_user_health *uh) { struct mly_health_status mh; int error; /* fetch the current health status from userspace */ if ((error = copyin(uh->HealthStatusBuffer, &mh, sizeof(mh))) != 0) return(error); /* spin waiting for a status update */ MLY_LOCK(sc); error = EWOULDBLOCK; while ((error != 0) && (sc->mly_event_change == mh.change_counter)) error = mtx_sleep(&sc->mly_event_change, &sc->mly_lock, PRIBIO | PCATCH, "mlyhealth", 0); mh = sc->mly_mmbox->mmm_health.status; MLY_UNLOCK(sc); /* copy the controller's health status buffer out */ error = copyout(&mh, uh->HealthStatusBuffer, sizeof(mh)); return(error); } #ifdef MLY_DEBUG static void mly_timeout(void *arg) { struct mly_softc *sc; struct mly_command *mc; int deadline; sc = arg; MLY_ASSERT_LOCKED(sc); deadline = time_second - MLY_CMD_TIMEOUT; TAILQ_FOREACH(mc, &sc->mly_busy, mc_link) { if ((mc->mc_timestamp < deadline)) { device_printf(sc->mly_dev, "COMMAND %p TIMEOUT AFTER %d SECONDS\n", mc, (int)(time_second - mc->mc_timestamp)); } } callout_reset(&sc->mly_timeout, MLY_CMD_TIMEOUT * hz, mly_timeout, sc); } #endif Index: projects/clang400-import/sys/dev/ncr/ncr.c =================================================================== --- projects/clang400-import/sys/dev/ncr/ncr.c (revision 311459) +++ projects/clang400-import/sys/dev/ncr/ncr.c (revision 311460) @@ -1,7119 +1,7119 @@ /************************************************************************** ** ** ** Device driver for the NCR 53C8XX PCI-SCSI-Controller Family. ** **------------------------------------------------------------------------- ** ** Written for 386bsd and FreeBSD by ** Wolfgang Stanglmeier ** Stefan Esser ** **------------------------------------------------------------------------- */ /*- ** Copyright (c) 1994 Wolfgang Stanglmeier. All rights reserved. ** ** Redistribution and use in source and binary forms, with or without ** modification, are permitted provided that the following conditions ** are met: ** 1. Redistributions of source code must retain the above copyright ** notice, this list of conditions and the following disclaimer. ** 2. Redistributions in binary form must reproduce the above copyright ** notice, this list of conditions and the following disclaimer in the ** documentation and/or other materials provided with the distribution. ** 3. The name of the author may not be used to endorse or promote products ** derived from this software without specific prior written permission. ** ** THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR ** IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES ** OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. ** IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, ** INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT ** NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, ** DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY ** THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT ** (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF ** THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. ** *************************************************************************** */ #include __FBSDID("$FreeBSD$"); #define NCR_GETCC_WITHMSG #if defined (__FreeBSD__) && defined(_KERNEL) #include "opt_ncr.h" #endif /*========================================================== ** ** Configuration and Debugging ** ** May be overwritten in ** **========================================================== */ /* ** SCSI address of this device. ** The boot routines should have set it. ** If not, use this. */ #ifndef SCSI_NCR_MYADDR #define SCSI_NCR_MYADDR (7) #endif /* SCSI_NCR_MYADDR */ /* ** The default synchronous period factor ** (0=asynchronous) ** If maximum synchronous frequency is defined, use it instead. */ #ifndef SCSI_NCR_MAX_SYNC #ifndef SCSI_NCR_DFLT_SYNC #define SCSI_NCR_DFLT_SYNC (12) #endif /* SCSI_NCR_DFLT_SYNC */ #else #if SCSI_NCR_MAX_SYNC == 0 #define SCSI_NCR_DFLT_SYNC 0 #else #define SCSI_NCR_DFLT_SYNC (250000 / SCSI_NCR_MAX_SYNC) #endif #endif /* ** The minimal asynchronous pre-scaler period (ns) ** Shall be 40. */ #ifndef SCSI_NCR_MIN_ASYNC #define SCSI_NCR_MIN_ASYNC (40) #endif /* SCSI_NCR_MIN_ASYNC */ /* ** The maximal bus with (in log2 byte) ** (0=8 bit, 1=16 bit) */ #ifndef SCSI_NCR_MAX_WIDE #define SCSI_NCR_MAX_WIDE (1) #endif /* SCSI_NCR_MAX_WIDE */ /*========================================================== ** ** Configuration and Debugging ** **========================================================== */ /* ** Number of targets supported by the driver. ** n permits target numbers 0..n-1. ** Default is 7, meaning targets #0..#6. ** #7 .. is myself. */ #define MAX_TARGET (16) /* ** Number of logic units supported by the driver. ** n enables logic unit numbers 0..n-1. ** The common SCSI devices require only ** one lun, so take 1 as the default. */ #ifndef MAX_LUN #define MAX_LUN (8) #endif /* MAX_LUN */ /* ** The maximum number of jobs scheduled for starting. ** There should be one slot per target, and one slot ** for each tag of each target in use. */ #define MAX_START (256) /* ** The maximum number of segments a transfer is split into. */ #define MAX_SCATTER (33) /* ** The maximum transfer length (should be >= 64k). ** MUST NOT be greater than (MAX_SCATTER-1) * PAGE_SIZE. */ #define MAX_SIZE ((MAX_SCATTER-1) * (long) PAGE_SIZE) /* ** other */ #define NCR_SNOOP_TIMEOUT (1000000) /*========================================================== ** ** Include files ** **========================================================== */ #include #include #ifdef _KERNEL #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #endif #include #include #include #include #include #include #include #include #include #include /*========================================================== ** ** Debugging tags ** **========================================================== */ #define DEBUG_ALLOC (0x0001) #define DEBUG_PHASE (0x0002) #define DEBUG_POLL (0x0004) #define DEBUG_QUEUE (0x0008) #define DEBUG_RESULT (0x0010) #define DEBUG_SCATTER (0x0020) #define DEBUG_SCRIPT (0x0040) #define DEBUG_TINY (0x0080) #define DEBUG_TIMING (0x0100) #define DEBUG_NEGO (0x0200) #define DEBUG_TAGS (0x0400) #define DEBUG_FREEZE (0x0800) #define DEBUG_RESTART (0x1000) /* ** Enable/Disable debug messages. ** Can be changed at runtime too. */ #ifdef SCSI_NCR_DEBUG #define DEBUG_FLAGS ncr_debug #else /* SCSI_NCR_DEBUG */ #define SCSI_NCR_DEBUG 0 #define DEBUG_FLAGS 0 #endif /* SCSI_NCR_DEBUG */ /*========================================================== ** ** assert () ** **========================================================== ** ** modified copy from 386bsd:/usr/include/sys/assert.h ** **---------------------------------------------------------- */ #ifdef DIAGNOSTIC #define assert(expression) { \ KASSERT(expression, ("%s", #expression)); \ } #else #define assert(expression) { \ if (!(expression)) { \ (void)printf("assertion \"%s\" failed: " \ "file \"%s\", line %d\n", \ #expression, __FILE__, __LINE__); \ } \ } #endif /*========================================================== ** ** Access to the controller chip. ** **========================================================== */ #define INB(r) bus_read_1(np->reg_res, offsetof(struct ncr_reg, r)) #define INW(r) bus_read_2(np->reg_res, offsetof(struct ncr_reg, r)) #define INL(r) bus_read_4(np->reg_res, offsetof(struct ncr_reg, r)) #define OUTB(r, val) bus_write_1(np->reg_res, offsetof(struct ncr_reg, r), val) #define OUTW(r, val) bus_write_2(np->reg_res, offsetof(struct ncr_reg, r), val) #define OUTL(r, val) bus_write_4(np->reg_res, offsetof(struct ncr_reg, r), val) #define OUTL_OFF(o, val) bus_write_4(np->reg_res, o, val) #define INB_OFF(o) bus_read_1(np->reg_res, o) #define INW_OFF(o) bus_read_2(np->reg_res, o) #define INL_OFF(o) bus_read_4(np->reg_res, o) #define READSCRIPT_OFF(base, off) \ (base ? *((volatile u_int32_t *)((volatile char *)base + (off))) : \ bus_read_4(np->sram_res, off)) #define WRITESCRIPT_OFF(base, off, val) \ do { \ if (base) \ *((volatile u_int32_t *) \ ((volatile char *)base + (off))) = (val); \ else \ bus_write_4(np->sram_res, off, val); \ } while (0) #define READSCRIPT(r) \ READSCRIPT_OFF(np->script, offsetof(struct script, r)) #define WRITESCRIPT(r, val) \ WRITESCRIPT_OFF(np->script, offsetof(struct script, r), val) /* ** Set bit field ON, OFF */ #define OUTONB(r, m) OUTB(r, INB(r) | (m)) #define OUTOFFB(r, m) OUTB(r, INB(r) & ~(m)) #define OUTONW(r, m) OUTW(r, INW(r) | (m)) #define OUTOFFW(r, m) OUTW(r, INW(r) & ~(m)) #define OUTONL(r, m) OUTL(r, INL(r) | (m)) #define OUTOFFL(r, m) OUTL(r, INL(r) & ~(m)) /*========================================================== ** ** Command control block states. ** **========================================================== */ #define HS_IDLE (0) #define HS_BUSY (1) #define HS_NEGOTIATE (2) /* sync/wide data transfer*/ #define HS_DISCONNECT (3) /* Disconnected by target */ #define HS_COMPLETE (4) #define HS_SEL_TIMEOUT (5) /* Selection timeout */ #define HS_RESET (6) /* SCSI reset */ #define HS_ABORTED (7) /* Transfer aborted */ #define HS_TIMEOUT (8) /* Software timeout */ #define HS_FAIL (9) /* SCSI or PCI bus errors */ #define HS_UNEXPECTED (10) /* Unexpected disconnect */ #define HS_STALL (11) /* QUEUE FULL or BUSY */ #define HS_DONEMASK (0xfc) /*========================================================== ** ** Software Interrupt Codes ** **========================================================== */ #define SIR_SENSE_RESTART (1) #define SIR_SENSE_FAILED (2) #define SIR_STALL_RESTART (3) #define SIR_STALL_QUEUE (4) #define SIR_NEGO_SYNC (5) #define SIR_NEGO_WIDE (6) #define SIR_NEGO_FAILED (7) #define SIR_NEGO_PROTO (8) #define SIR_REJECT_RECEIVED (9) #define SIR_REJECT_SENT (10) #define SIR_IGN_RESIDUE (11) #define SIR_MISSING_SAVE (12) #define SIR_MAX (12) /*========================================================== ** ** Extended error codes. ** xerr_status field of struct nccb. ** **========================================================== */ #define XE_OK (0) #define XE_EXTRA_DATA (1) /* unexpected data phase */ #define XE_BAD_PHASE (2) /* illegal phase (4/5) */ /*========================================================== ** ** Negotiation status. ** nego_status field of struct nccb. ** **========================================================== */ #define NS_SYNC (1) #define NS_WIDE (2) /*========================================================== ** ** XXX These are no longer used. Remove once the ** script is updated. ** "Special features" of targets. ** quirks field of struct tcb. ** actualquirks field of struct nccb. ** **========================================================== */ #define QUIRK_AUTOSAVE (0x01) #define QUIRK_NOMSG (0x02) #define QUIRK_NOSYNC (0x10) #define QUIRK_NOWIDE16 (0x20) #define QUIRK_NOTAGS (0x40) #define QUIRK_UPDATE (0x80) /*========================================================== ** ** Misc. ** **========================================================== */ #define CCB_MAGIC (0xf2691ad2) #define MAX_TAGS (32) /* hard limit */ /*========================================================== ** ** OS dependencies. ** **========================================================== */ #define PRINT_ADDR(ccb) xpt_print_path((ccb)->ccb_h.path) /*========================================================== ** ** Declaration of structs. ** **========================================================== */ struct tcb; struct lcb; struct nccb; struct ncb; struct script; typedef struct ncb * ncb_p; typedef struct tcb * tcb_p; typedef struct lcb * lcb_p; typedef struct nccb * nccb_p; struct link { ncrcmd l_cmd; ncrcmd l_paddr; }; struct usrcmd { u_long target; u_long lun; u_long data; u_long cmd; }; #define UC_SETSYNC 10 #define UC_SETTAGS 11 #define UC_SETDEBUG 12 #define UC_SETORDER 13 #define UC_SETWIDE 14 #define UC_SETFLAG 15 #define UF_TRACE (0x01) /*--------------------------------------- ** ** Timestamps for profiling ** **--------------------------------------- */ /* Type of the kernel variable `ticks'. XXX should be declared with the var. */ typedef int ticks_t; struct tstamp { ticks_t start; ticks_t end; ticks_t select; ticks_t command; ticks_t data; ticks_t status; ticks_t disconnect; }; /* ** profiling data (per device) */ struct profile { u_long num_trans; u_long num_bytes; u_long num_disc; u_long num_break; u_long num_int; u_long num_fly; u_long ms_setup; u_long ms_data; u_long ms_disc; u_long ms_post; }; /*========================================================== ** ** Declaration of structs: target control block ** **========================================================== */ #define NCR_TRANS_CUR 0x01 /* Modify current neogtiation status */ #define NCR_TRANS_ACTIVE 0x03 /* Assume this is the active target */ #define NCR_TRANS_GOAL 0x04 /* Modify negotiation goal */ #define NCR_TRANS_USER 0x08 /* Modify user negotiation settings */ struct ncr_transinfo { u_int8_t width; u_int8_t period; u_int8_t offset; }; struct ncr_target_tinfo { /* Hardware version of our sync settings */ u_int8_t disc_tag; #define NCR_CUR_DISCENB 0x01 #define NCR_CUR_TAGENB 0x02 #define NCR_USR_DISCENB 0x04 #define NCR_USR_TAGENB 0x08 u_int8_t sval; struct ncr_transinfo current; struct ncr_transinfo goal; struct ncr_transinfo user; /* Hardware version of our wide settings */ u_int8_t wval; }; struct tcb { /* ** during reselection the ncr jumps to this point ** with SFBR set to the encoded target number ** with bit 7 set. ** if it's not this target, jump to the next. ** ** JUMP IF (SFBR != #target#) ** @(next tcb) */ struct link jump_tcb; /* ** load the actual values for the sxfer and the scntl3 ** register (sync/wide mode). ** ** SCR_COPY (1); ** @(sval field of this tcb) ** @(sxfer register) ** SCR_COPY (1); ** @(wval field of this tcb) ** @(scntl3 register) */ ncrcmd getscr[6]; /* ** if next message is "identify" ** then load the message to SFBR, ** else load 0 to SFBR. ** ** CALL ** */ struct link call_lun; /* ** now look for the right lun. ** ** JUMP ** @(first nccb of this lun) */ struct link jump_lcb; /* ** pointer to interrupted getcc nccb */ nccb_p hold_cp; /* ** pointer to nccb used for negotiating. ** Avoid to start a nego for all queued commands ** when tagged command queuing is enabled. */ nccb_p nego_cp; /* ** statistical data */ u_long transfers; u_long bytes; /* ** user settable limits for sync transfer ** and tagged commands. */ struct ncr_target_tinfo tinfo; /* ** the lcb's of this tcb */ lcb_p lp[MAX_LUN]; }; /*========================================================== ** ** Declaration of structs: lun control block ** **========================================================== */ struct lcb { /* ** during reselection the ncr jumps to this point ** with SFBR set to the "Identify" message. ** if it's not this lun, jump to the next. ** ** JUMP IF (SFBR != #lun#) ** @(next lcb of this target) */ struct link jump_lcb; /* ** if next message is "simple tag", ** then load the tag to SFBR, ** else load 0 to SFBR. ** ** CALL ** */ struct link call_tag; /* ** now look for the right nccb. ** ** JUMP ** @(first nccb of this lun) */ struct link jump_nccb; /* ** start of the nccb chain */ nccb_p next_nccb; /* ** Control of tagged queueing */ u_char reqnccbs; u_char reqlink; u_char actlink; u_char usetags; u_char lasttag; }; /*========================================================== ** ** Declaration of structs: COMMAND control block ** **========================================================== ** ** This substructure is copied from the nccb to a ** global address after selection (or reselection) ** and copied back before disconnect. ** ** These fields are accessible to the script processor. ** **---------------------------------------------------------- */ struct head { /* ** Execution of a nccb starts at this point. ** It's a jump to the "SELECT" label ** of the script. ** ** After successful selection the script ** processor overwrites it with a jump to ** the IDLE label of the script. */ struct link launch; /* ** Saved data pointer. ** Points to the position in the script ** responsible for the actual transfer ** of data. ** It's written after reception of a ** "SAVE_DATA_POINTER" message. ** The goalpointer points after ** the last transfer command. */ u_int32_t savep; u_int32_t lastp; u_int32_t goalp; /* ** The virtual address of the nccb ** containing this header. */ nccb_p cp; /* ** space for some timestamps to gather ** profiling data about devices and this driver. */ struct tstamp stamp; /* ** status fields. */ u_char status[8]; }; /* ** The status bytes are used by the host and the script processor. ** ** The first four byte are copied to the scratchb register ** (declared as scr0..scr3 in ncr_reg.h) just after the select/reselect, ** and copied back just after disconnecting. ** Inside the script the XX_REG are used. ** ** The last four bytes are used inside the script by "COPY" commands. ** Because source and destination must have the same alignment ** in a longword, the fields HAVE to be at the chosen offsets. ** xerr_st (4) 0 (0x34) scratcha ** sync_st (5) 1 (0x05) sxfer ** wide_st (7) 3 (0x03) scntl3 */ /* ** First four bytes (script) */ #define QU_REG scr0 #define HS_REG scr1 #define HS_PRT nc_scr1 #define SS_REG scr2 #define PS_REG scr3 /* ** First four bytes (host) */ #define actualquirks phys.header.status[0] #define host_status phys.header.status[1] #define s_status phys.header.status[2] #define parity_status phys.header.status[3] /* ** Last four bytes (script) */ #define xerr_st header.status[4] /* MUST be ==0 mod 4 */ #define sync_st header.status[5] /* MUST be ==1 mod 4 */ #define nego_st header.status[6] #define wide_st header.status[7] /* MUST be ==3 mod 4 */ /* ** Last four bytes (host) */ #define xerr_status phys.xerr_st #define sync_status phys.sync_st #define nego_status phys.nego_st #define wide_status phys.wide_st /*========================================================== ** ** Declaration of structs: Data structure block ** **========================================================== ** ** During execution of a nccb by the script processor, ** the DSA (data structure address) register points ** to this substructure of the nccb. ** This substructure contains the header with ** the script-processor-changable data and ** data blocks for the indirect move commands. ** **---------------------------------------------------------- */ struct dsb { /* ** Header. ** Has to be the first entry, ** because it's jumped to by the ** script processor */ struct head header; /* ** Table data for Script */ struct scr_tblsel select; struct scr_tblmove smsg ; struct scr_tblmove smsg2 ; struct scr_tblmove cmd ; struct scr_tblmove scmd ; struct scr_tblmove sense ; struct scr_tblmove data [MAX_SCATTER]; }; /*========================================================== ** ** Declaration of structs: Command control block. ** **========================================================== ** ** During execution of a nccb by the script processor, ** the DSA (data structure address) register points ** to this substructure of the nccb. ** This substructure contains the header with ** the script-processor-changable data and then ** data blocks for the indirect move commands. ** **---------------------------------------------------------- */ struct nccb { /* ** This filler ensures that the global header is ** cache line size aligned. */ ncrcmd filler[4]; /* ** during reselection the ncr jumps to this point. ** If a "SIMPLE_TAG" message was received, ** then SFBR is set to the tag. ** else SFBR is set to 0 ** If looking for another tag, jump to the next nccb. ** ** JUMP IF (SFBR != #TAG#) ** @(next nccb of this lun) */ struct link jump_nccb; /* ** After execution of this call, the return address ** (in the TEMP register) points to the following ** data structure block. ** So copy it to the DSA register, and start ** processing of this data structure. ** ** CALL ** */ struct link call_tmp; /* ** This is the data structure which is ** to be executed by the script processor. */ struct dsb phys; /* ** If a data transfer phase is terminated too early ** (after reception of a message (i.e. DISCONNECT)), ** we have to prepare a mini script to transfer ** the rest of the data. */ ncrcmd patch[8]; /* ** The general SCSI driver provides a ** pointer to a control block. */ union ccb *ccb; /* ** We prepare a message to be sent after selection, ** and a second one to be sent after getcc selection. ** Contents are IDENTIFY and SIMPLE_TAG. ** While negotiating sync or wide transfer, ** a SDTM or WDTM message is appended. */ u_char scsi_smsg [8]; u_char scsi_smsg2[8]; /* ** Lock this nccb. ** Flag is used while looking for a free nccb. */ u_long magic; /* ** Physical address of this instance of nccb */ u_long p_nccb; /* ** Completion time out for this job. ** It's set to time of start + allowed number of seconds. */ time_t tlimit; /* ** All nccbs of one hostadapter are chained. */ nccb_p link_nccb; /* ** All nccbs of one target/lun are chained. */ nccb_p next_nccb; /* ** Sense command */ u_char sensecmd[6]; /* ** Tag for this transfer. ** It's patched into jump_nccb. ** If it's not zero, a SIMPLE_TAG ** message is included in smsg. */ u_char tag; }; #define CCB_PHYS(cp,lbl) (cp->p_nccb + offsetof(struct nccb, lbl)) /*========================================================== ** ** Declaration of structs: NCR device descriptor ** **========================================================== */ struct ncb { /* ** The global header. ** Accessible to both the host and the ** script-processor. ** We assume it is cache line size aligned. */ struct head header; device_t dev; /*----------------------------------------------- ** Scripts .. **----------------------------------------------- ** ** During reselection the ncr jumps to this point. ** The SFBR register is loaded with the encoded target id. ** ** Jump to the first target. ** ** JUMP ** @(next tcb) */ struct link jump_tcb; /*----------------------------------------------- ** Configuration .. **----------------------------------------------- ** ** virtual and physical addresses ** of the 53c810 chip. */ int reg_rid; struct resource *reg_res; int sram_rid; struct resource *sram_res; struct resource *irq_res; void *irq_handle; /* ** Scripts instance virtual address. */ struct script *script; struct scripth *scripth; /* ** Scripts instance physical address. */ u_long p_script; u_long p_scripth; /* ** The SCSI address of the host adapter. */ u_char myaddr; /* ** timing parameters */ u_char minsync; /* Minimum sync period factor */ u_char maxsync; /* Maximum sync period factor */ u_char maxoffs; /* Max scsi offset */ u_char clock_divn; /* Number of clock divisors */ u_long clock_khz; /* SCSI clock frequency in KHz */ u_long features; /* Chip features map */ u_char multiplier; /* Clock multiplier (1,2,4) */ u_char maxburst; /* log base 2 of dwords burst */ /* ** BIOS supplied PCI bus options */ u_char rv_scntl3; u_char rv_dcntl; u_char rv_dmode; u_char rv_ctest3; u_char rv_ctest4; u_char rv_ctest5; u_char rv_gpcntl; u_char rv_stest2; /*----------------------------------------------- ** CAM SIM information for this instance **----------------------------------------------- */ struct cam_sim *sim; struct cam_path *path; /*----------------------------------------------- ** Job control **----------------------------------------------- ** ** Commands from user */ struct usrcmd user; /* ** Target data */ struct tcb target[MAX_TARGET]; /* ** Start queue. */ u_int32_t squeue [MAX_START]; u_short squeueput; /* ** Timeout handler */ time_t heartbeat; u_short ticks; u_short latetime; time_t lasttime; struct callout timer; /*----------------------------------------------- ** Debug and profiling **----------------------------------------------- ** ** register dump */ struct ncr_reg regdump; time_t regtime; /* ** Profiling data */ struct profile profile; u_long disc_phys; u_long disc_ref; /* ** Head of list of all nccbs for this controller. */ nccb_p link_nccb; /* ** message buffers. ** Should be longword aligned, ** because they're written with a ** COPY script command. */ u_char msgout[8]; u_char msgin [8]; u_int32_t lastmsg; /* ** Buffer for STATUS_IN phase. */ u_char scratch; /* ** controller chip dependent maximal transfer width. */ u_char maxwide; struct mtx lock; #ifdef NCR_IOMAPPED /* ** address of the ncr control registers in io space */ pci_port_t port; #endif }; #define NCB_SCRIPT_PHYS(np,lbl) (np->p_script + offsetof (struct script, lbl)) #define NCB_SCRIPTH_PHYS(np,lbl) (np->p_scripth + offsetof (struct scripth,lbl)) /*========================================================== ** ** ** Script for NCR-Processor. ** ** Use ncr_script_fill() to create the variable parts. ** Use ncr_script_copy_and_bind() to make a copy and ** bind to physical addresses. ** ** **========================================================== ** ** We have to know the offsets of all labels before ** we reach them (for forward jumps). ** Therefore we declare a struct here. ** If you make changes inside the script, ** DONT FORGET TO CHANGE THE LENGTHS HERE! ** **---------------------------------------------------------- */ /* ** Script fragments which are loaded into the on-board RAM ** of 825A, 875 and 895 chips. */ struct script { ncrcmd start [ 7]; ncrcmd start0 [ 2]; ncrcmd start1 [ 3]; ncrcmd startpos [ 1]; ncrcmd trysel [ 8]; ncrcmd skip [ 8]; ncrcmd skip2 [ 3]; ncrcmd idle [ 2]; ncrcmd select [ 18]; ncrcmd prepare [ 4]; ncrcmd loadpos [ 14]; ncrcmd prepare2 [ 24]; ncrcmd setmsg [ 5]; ncrcmd clrack [ 2]; ncrcmd dispatch [ 33]; ncrcmd no_data [ 17]; ncrcmd checkatn [ 10]; ncrcmd command [ 15]; ncrcmd status [ 27]; ncrcmd msg_in [ 26]; ncrcmd msg_bad [ 6]; ncrcmd complete [ 13]; ncrcmd cleanup [ 12]; ncrcmd cleanup0 [ 9]; ncrcmd signal [ 12]; ncrcmd save_dp [ 5]; ncrcmd restore_dp [ 5]; ncrcmd disconnect [ 12]; ncrcmd disconnect0 [ 5]; ncrcmd disconnect1 [ 23]; ncrcmd msg_out [ 9]; ncrcmd msg_out_done [ 7]; ncrcmd badgetcc [ 6]; ncrcmd reselect [ 8]; ncrcmd reselect1 [ 8]; ncrcmd reselect2 [ 8]; ncrcmd resel_tmp [ 5]; ncrcmd resel_lun [ 18]; ncrcmd resel_tag [ 24]; ncrcmd data_in [MAX_SCATTER * 4 + 7]; ncrcmd data_out [MAX_SCATTER * 4 + 7]; }; /* ** Script fragments which stay in main memory for all chips. */ struct scripth { ncrcmd tryloop [MAX_START*5+2]; ncrcmd msg_parity [ 6]; ncrcmd msg_reject [ 8]; ncrcmd msg_ign_residue [ 32]; ncrcmd msg_extended [ 18]; ncrcmd msg_ext_2 [ 18]; ncrcmd msg_wdtr [ 27]; ncrcmd msg_ext_3 [ 18]; ncrcmd msg_sdtr [ 27]; ncrcmd msg_out_abort [ 10]; ncrcmd getcc [ 4]; ncrcmd getcc1 [ 5]; #ifdef NCR_GETCC_WITHMSG ncrcmd getcc2 [ 29]; #else ncrcmd getcc2 [ 14]; #endif ncrcmd getcc3 [ 6]; ncrcmd aborttag [ 4]; ncrcmd abort [ 22]; ncrcmd snooptest [ 9]; ncrcmd snoopend [ 2]; }; /*========================================================== ** ** ** Function headers. ** ** **========================================================== */ #ifdef _KERNEL static nccb_p ncr_alloc_nccb(ncb_p np, u_long target, u_long lun); static void ncr_complete(ncb_p np, nccb_p cp); static int ncr_delta(int * from, int * to); static void ncr_exception(ncb_p np); static void ncr_free_nccb(ncb_p np, nccb_p cp); static void ncr_freeze_devq(ncb_p np, struct cam_path *path); static void ncr_selectclock(ncb_p np, u_char scntl3); static void ncr_getclock(ncb_p np, u_char multiplier); static nccb_p ncr_get_nccb(ncb_p np, u_long t,u_long l); #if 0 static u_int32_t ncr_info(int unit); #endif static void ncr_init(ncb_p np, char * msg, u_long code); static void ncr_intr(void *vnp); static void ncr_intr_locked(ncb_p np); static void ncr_int_ma(ncb_p np, u_char dstat); static void ncr_int_sir(ncb_p np); static void ncr_int_sto(ncb_p np); #if 0 static void ncr_min_phys(struct buf *bp); #endif static void ncr_poll(struct cam_sim *sim); static void ncb_profile(ncb_p np, nccb_p cp); static void ncr_script_copy_and_bind(ncb_p np, ncrcmd *src, ncrcmd *dst, int len); static void ncr_script_fill(struct script * scr, struct scripth *scrh); static int ncr_scatter(struct dsb* phys, vm_offset_t vaddr, vm_size_t datalen); static void ncr_getsync(ncb_p np, u_char sfac, u_char *fakp, u_char *scntl3p); static void ncr_setsync(ncb_p np, nccb_p cp,u_char scntl3,u_char sxfer, u_char period); static void ncr_setwide(ncb_p np, nccb_p cp, u_char wide, u_char ack); static int ncr_show_msg(u_char * msg); static int ncr_snooptest(ncb_p np); static void ncr_action(struct cam_sim *sim, union ccb *ccb); static void ncr_timeout(void *arg); static void ncr_wakeup(ncb_p np, u_long code); static int ncr_probe(device_t dev); static int ncr_attach(device_t dev); #endif /* _KERNEL */ /*========================================================== ** ** ** Global static data. ** ** **========================================================== */ #ifdef _KERNEL static int ncr_debug = SCSI_NCR_DEBUG; SYSCTL_INT(_debug, OID_AUTO, ncr_debug, CTLFLAG_RW, &ncr_debug, 0, ""); static int ncr_cache; /* to be aligned _NOT_ static */ /*========================================================== ** ** ** Global static data: auto configure ** ** **========================================================== */ #define NCR_810_ID (0x00011000ul) #define NCR_815_ID (0x00041000ul) #define NCR_820_ID (0x00021000ul) #define NCR_825_ID (0x00031000ul) #define NCR_860_ID (0x00061000ul) #define NCR_875_ID (0x000f1000ul) #define NCR_875_ID2 (0x008f1000ul) #define NCR_885_ID (0x000d1000ul) #define NCR_895_ID (0x000c1000ul) #define NCR_896_ID (0x000b1000ul) #define NCR_895A_ID (0x00121000ul) #define NCR_1510D_ID (0x000a1000ul) /*========================================================== ** ** ** Scripts for NCR-Processor. ** ** Use ncr_script_bind for binding to physical addresses. ** ** **========================================================== ** ** NADDR generates a reference to a field of the controller data. ** PADDR generates a reference to another part of the script. ** RADDR generates a reference to a script processor register. ** FADDR generates a reference to a script processor register ** with offset. ** **---------------------------------------------------------- */ #define RELOC_SOFTC 0x40000000 #define RELOC_LABEL 0x50000000 #define RELOC_REGISTER 0x60000000 #define RELOC_KVAR 0x70000000 #define RELOC_LABELH 0x80000000 #define RELOC_MASK 0xf0000000 #define NADDR(label) (RELOC_SOFTC | offsetof(struct ncb, label)) #define PADDR(label) (RELOC_LABEL | offsetof(struct script, label)) #define PADDRH(label) (RELOC_LABELH | offsetof(struct scripth, label)) #define RADDR(label) (RELOC_REGISTER | REG(label)) #define FADDR(label,ofs)(RELOC_REGISTER | ((REG(label))+(ofs))) #define KVAR(which) (RELOC_KVAR | (which)) #define KVAR_SECOND (0) #define KVAR_TICKS (1) #define KVAR_NCR_CACHE (2) #define SCRIPT_KVAR_FIRST (0) #define SCRIPT_KVAR_LAST (3) /* * Kernel variables referenced in the scripts. * THESE MUST ALL BE ALIGNED TO A 4-BYTE BOUNDARY. */ static volatile void *script_kvars[] = { &time_second, &ticks, &ncr_cache }; static struct script script0 = { /*--------------------------< START >-----------------------*/ { /* ** Claim to be still alive ... */ SCR_COPY (sizeof (((struct ncb *)0)->heartbeat)), KVAR (KVAR_SECOND), NADDR (heartbeat), /* ** Make data structure address invalid. ** clear SIGP. */ SCR_LOAD_REG (dsa, 0xff), 0, SCR_FROM_REG (ctest2), 0, }/*-------------------------< START0 >----------------------*/,{ /* ** Hook for interrupted GetConditionCode. ** Will be patched to ... IFTRUE by ** the interrupt handler. */ SCR_INT ^ IFFALSE (0), SIR_SENSE_RESTART, }/*-------------------------< START1 >----------------------*/,{ /* ** Hook for stalled start queue. ** Will be patched to IFTRUE by the interrupt handler. */ SCR_INT ^ IFFALSE (0), SIR_STALL_RESTART, /* ** Then jump to a certain point in tryloop. ** Due to the lack of indirect addressing the code ** is self modifying here. */ SCR_JUMP, }/*-------------------------< STARTPOS >--------------------*/,{ PADDRH(tryloop), }/*-------------------------< TRYSEL >----------------------*/,{ /* ** Now: ** DSA: Address of a Data Structure ** or Address of the IDLE-Label. ** ** TEMP: Address of a script, which tries to ** start the NEXT entry. ** ** Save the TEMP register into the SCRATCHA register. ** Then copy the DSA to TEMP and RETURN. ** This is kind of an indirect jump. ** (The script processor has NO stack, so the ** CALL is actually a jump and link, and the ** RETURN is an indirect jump.) ** ** If the slot was empty, DSA contains the address ** of the IDLE part of this script. The processor ** jumps to IDLE and waits for a reselect. ** It will wake up and try the same slot again ** after the SIGP bit becomes set by the host. ** ** If the slot was not empty, DSA contains ** the address of the phys-part of a nccb. ** The processor jumps to this address. ** phys starts with head, ** head starts with launch, ** so actually the processor jumps to ** the lauch part. ** If the entry is scheduled for execution, ** then launch contains a jump to SELECT. ** If it's not scheduled, it contains a jump to IDLE. */ SCR_COPY (4), RADDR (temp), RADDR (scratcha), SCR_COPY (4), RADDR (dsa), RADDR (temp), SCR_RETURN, 0 }/*-------------------------< SKIP >------------------------*/,{ /* ** This entry has been canceled. ** Next time use the next slot. */ SCR_COPY (4), RADDR (scratcha), PADDR (startpos), /* ** patch the launch field. ** should look like an idle process. */ SCR_COPY_F (4), RADDR (dsa), PADDR (skip2), SCR_COPY (8), PADDR (idle), }/*-------------------------< SKIP2 >-----------------------*/,{ 0, SCR_JUMP, PADDR(start), }/*-------------------------< IDLE >------------------------*/,{ /* ** Nothing to do? ** Wait for reselect. */ SCR_JUMP, PADDR(reselect), }/*-------------------------< SELECT >----------------------*/,{ /* ** DSA contains the address of a scheduled ** data structure. ** ** SCRATCHA contains the address of the script, ** which starts the next entry. ** ** Set Initiator mode. ** ** (Target mode is left as an exercise for the reader) */ SCR_CLR (SCR_TRG), 0, SCR_LOAD_REG (HS_REG, 0xff), 0, /* ** And try to select this target. */ SCR_SEL_TBL_ATN ^ offsetof (struct dsb, select), PADDR (reselect), /* ** Now there are 4 possibilities: ** ** (1) The ncr loses arbitration. ** This is ok, because it will try again, ** when the bus becomes idle. ** (But beware of the timeout function!) ** ** (2) The ncr is reselected. ** Then the script processor takes the jump ** to the RESELECT label. ** ** (3) The ncr completes the selection. ** Then it will execute the next statement. ** ** (4) There is a selection timeout. ** Then the ncr should interrupt the host and stop. ** Unfortunately, it seems to continue execution ** of the script. But it will fail with an ** IID-interrupt on the next WHEN. */ SCR_JUMPR ^ IFTRUE (WHEN (SCR_MSG_IN)), 0, /* ** Send the IDENTIFY and SIMPLE_TAG messages ** (and the MSG_EXT_SDTR message) */ SCR_MOVE_TBL ^ SCR_MSG_OUT, offsetof (struct dsb, smsg), #ifdef undef /* XXX better fail than try to deal with this ... */ SCR_JUMPR ^ IFTRUE (WHEN (SCR_MSG_OUT)), -16, #endif SCR_CLR (SCR_ATN), 0, SCR_COPY (1), RADDR (sfbr), NADDR (lastmsg), /* ** Selection complete. ** Next time use the next slot. */ SCR_COPY (4), RADDR (scratcha), PADDR (startpos), }/*-------------------------< PREPARE >----------------------*/,{ /* ** The ncr doesn't have an indirect load ** or store command. So we have to ** copy part of the control block to a ** fixed place, where we can access it. ** ** We patch the address part of a ** COPY command with the DSA-register. */ SCR_COPY_F (4), RADDR (dsa), PADDR (loadpos), /* ** then we do the actual copy. */ SCR_COPY (sizeof (struct head)), /* ** continued after the next label ... */ }/*-------------------------< LOADPOS >---------------------*/,{ 0, NADDR (header), /* ** Mark this nccb as not scheduled. */ SCR_COPY (8), PADDR (idle), NADDR (header.launch), /* ** Set a time stamp for this selection */ SCR_COPY (sizeof (ticks)), KVAR (KVAR_TICKS), NADDR (header.stamp.select), /* ** load the savep (saved pointer) into ** the TEMP register (actual pointer) */ SCR_COPY (4), NADDR (header.savep), RADDR (temp), /* ** Initialize the status registers */ SCR_COPY (4), NADDR (header.status), RADDR (scr0), }/*-------------------------< PREPARE2 >---------------------*/,{ /* ** Load the synchronous mode register */ SCR_COPY (1), NADDR (sync_st), RADDR (sxfer), /* ** Load the wide mode and timing register */ SCR_COPY (1), NADDR (wide_st), RADDR (scntl3), /* ** Initialize the msgout buffer with a NOOP message. */ SCR_LOAD_REG (scratcha, MSG_NOOP), 0, SCR_COPY (1), RADDR (scratcha), NADDR (msgout), SCR_COPY (1), RADDR (scratcha), NADDR (msgin), /* ** Message in phase ? */ SCR_JUMP ^ IFFALSE (WHEN (SCR_MSG_IN)), PADDR (dispatch), /* ** Extended or reject message ? */ SCR_FROM_REG (sbdl), 0, SCR_JUMP ^ IFTRUE (DATA (MSG_EXTENDED)), PADDR (msg_in), SCR_JUMP ^ IFTRUE (DATA (MSG_MESSAGE_REJECT)), PADDRH (msg_reject), /* ** normal processing */ SCR_JUMP, PADDR (dispatch), }/*-------------------------< SETMSG >----------------------*/,{ SCR_COPY (1), RADDR (scratcha), NADDR (msgout), SCR_SET (SCR_ATN), 0, }/*-------------------------< CLRACK >----------------------*/,{ /* ** Terminate possible pending message phase. */ SCR_CLR (SCR_ACK), 0, }/*-----------------------< DISPATCH >----------------------*/,{ SCR_FROM_REG (HS_REG), 0, SCR_INT ^ IFTRUE (DATA (HS_NEGOTIATE)), SIR_NEGO_FAILED, /* ** remove bogus output signals */ SCR_REG_REG (socl, SCR_AND, CACK|CATN), 0, SCR_RETURN ^ IFTRUE (WHEN (SCR_DATA_OUT)), 0, SCR_RETURN ^ IFTRUE (IF (SCR_DATA_IN)), 0, SCR_JUMP ^ IFTRUE (IF (SCR_MSG_OUT)), PADDR (msg_out), SCR_JUMP ^ IFTRUE (IF (SCR_MSG_IN)), PADDR (msg_in), SCR_JUMP ^ IFTRUE (IF (SCR_COMMAND)), PADDR (command), SCR_JUMP ^ IFTRUE (IF (SCR_STATUS)), PADDR (status), /* ** Discard one illegal phase byte, if required. */ SCR_LOAD_REG (scratcha, XE_BAD_PHASE), 0, SCR_COPY (1), RADDR (scratcha), NADDR (xerr_st), SCR_JUMPR ^ IFFALSE (IF (SCR_ILG_OUT)), 8, SCR_MOVE_ABS (1) ^ SCR_ILG_OUT, NADDR (scratch), SCR_JUMPR ^ IFFALSE (IF (SCR_ILG_IN)), 8, SCR_MOVE_ABS (1) ^ SCR_ILG_IN, NADDR (scratch), SCR_JUMP, PADDR (dispatch), }/*-------------------------< NO_DATA >--------------------*/,{ /* ** The target wants to transfer too much data ** or in the wrong direction. ** Remember that in extended error. */ SCR_LOAD_REG (scratcha, XE_EXTRA_DATA), 0, SCR_COPY (1), RADDR (scratcha), NADDR (xerr_st), /* ** Discard one data byte, if required. */ SCR_JUMPR ^ IFFALSE (WHEN (SCR_DATA_OUT)), 8, SCR_MOVE_ABS (1) ^ SCR_DATA_OUT, NADDR (scratch), SCR_JUMPR ^ IFFALSE (IF (SCR_DATA_IN)), 8, SCR_MOVE_ABS (1) ^ SCR_DATA_IN, NADDR (scratch), /* ** .. and repeat as required. */ SCR_CALL, PADDR (dispatch), SCR_JUMP, PADDR (no_data), }/*-------------------------< CHECKATN >--------------------*/,{ /* ** If AAP (bit 1 of scntl0 register) is set ** and a parity error is detected, ** the script processor asserts ATN. ** ** The target should switch to a MSG_OUT phase ** to get the message. */ SCR_FROM_REG (socl), 0, SCR_JUMP ^ IFFALSE (MASK (CATN, CATN)), PADDR (dispatch), /* ** count it */ SCR_REG_REG (PS_REG, SCR_ADD, 1), 0, /* ** Prepare a MSG_INITIATOR_DET_ERR message ** (initiator detected error). ** The target should retry the transfer. */ SCR_LOAD_REG (scratcha, MSG_INITIATOR_DET_ERR), 0, SCR_JUMP, PADDR (setmsg), }/*-------------------------< COMMAND >--------------------*/,{ /* ** If this is not a GETCC transfer ... */ SCR_FROM_REG (SS_REG), 0, /*<<<*/ SCR_JUMPR ^ IFTRUE (DATA (SCSI_STATUS_CHECK_COND)), 28, /* ** ... set a timestamp ... */ SCR_COPY (sizeof (ticks)), KVAR (KVAR_TICKS), NADDR (header.stamp.command), /* ** ... and send the command */ SCR_MOVE_TBL ^ SCR_COMMAND, offsetof (struct dsb, cmd), SCR_JUMP, PADDR (dispatch), /* ** Send the GETCC command */ /*>>>*/ SCR_MOVE_TBL ^ SCR_COMMAND, offsetof (struct dsb, scmd), SCR_JUMP, PADDR (dispatch), }/*-------------------------< STATUS >--------------------*/,{ /* ** set the timestamp. */ SCR_COPY (sizeof (ticks)), KVAR (KVAR_TICKS), NADDR (header.stamp.status), /* ** If this is a GETCC transfer, */ SCR_FROM_REG (SS_REG), 0, /*<<<*/ SCR_JUMPR ^ IFFALSE (DATA (SCSI_STATUS_CHECK_COND)), 40, /* ** get the status */ SCR_MOVE_ABS (1) ^ SCR_STATUS, NADDR (scratch), /* ** Save status to scsi_status. ** Mark as complete. ** And wait for disconnect. */ SCR_TO_REG (SS_REG), 0, SCR_REG_REG (SS_REG, SCR_OR, SCSI_STATUS_SENSE), 0, SCR_LOAD_REG (HS_REG, HS_COMPLETE), 0, SCR_JUMP, PADDR (checkatn), /* ** If it was no GETCC transfer, ** save the status to scsi_status. */ /*>>>*/ SCR_MOVE_ABS (1) ^ SCR_STATUS, NADDR (scratch), SCR_TO_REG (SS_REG), 0, /* ** if it was no check condition ... */ SCR_JUMP ^ IFTRUE (DATA (SCSI_STATUS_CHECK_COND)), PADDR (checkatn), /* ** ... mark as complete. */ SCR_LOAD_REG (HS_REG, HS_COMPLETE), 0, SCR_JUMP, PADDR (checkatn), }/*-------------------------< MSG_IN >--------------------*/,{ /* ** Get the first byte of the message ** and save it to SCRATCHA. ** ** The script processor doesn't negate the ** ACK signal after this transfer. */ SCR_MOVE_ABS (1) ^ SCR_MSG_IN, NADDR (msgin[0]), /* ** Check for message parity error. */ SCR_TO_REG (scratcha), 0, SCR_FROM_REG (socl), 0, SCR_JUMP ^ IFTRUE (MASK (CATN, CATN)), PADDRH (msg_parity), SCR_FROM_REG (scratcha), 0, /* ** Parity was ok, handle this message. */ SCR_JUMP ^ IFTRUE (DATA (MSG_CMDCOMPLETE)), PADDR (complete), SCR_JUMP ^ IFTRUE (DATA (MSG_SAVEDATAPOINTER)), PADDR (save_dp), SCR_JUMP ^ IFTRUE (DATA (MSG_RESTOREPOINTERS)), PADDR (restore_dp), SCR_JUMP ^ IFTRUE (DATA (MSG_DISCONNECT)), PADDR (disconnect), SCR_JUMP ^ IFTRUE (DATA (MSG_EXTENDED)), PADDRH (msg_extended), SCR_JUMP ^ IFTRUE (DATA (MSG_NOOP)), PADDR (clrack), SCR_JUMP ^ IFTRUE (DATA (MSG_MESSAGE_REJECT)), PADDRH (msg_reject), SCR_JUMP ^ IFTRUE (DATA (MSG_IGN_WIDE_RESIDUE)), PADDRH (msg_ign_residue), /* ** Rest of the messages left as ** an exercise ... ** ** Unimplemented messages: ** fall through to MSG_BAD. */ }/*-------------------------< MSG_BAD >------------------*/,{ /* ** unimplemented message - reject it. */ SCR_INT, SIR_REJECT_SENT, SCR_LOAD_REG (scratcha, MSG_MESSAGE_REJECT), 0, SCR_JUMP, PADDR (setmsg), }/*-------------------------< COMPLETE >-----------------*/,{ /* ** Complete message. ** ** If it's not the get condition code, ** copy TEMP register to LASTP in header. */ SCR_FROM_REG (SS_REG), 0, /*<<<*/ SCR_JUMPR ^ IFTRUE (MASK (SCSI_STATUS_SENSE, SCSI_STATUS_SENSE)), 12, SCR_COPY (4), RADDR (temp), NADDR (header.lastp), /*>>>*/ /* ** When we terminate the cycle by clearing ACK, ** the target may disconnect immediately. ** ** We don't want to be told of an ** "unexpected disconnect", ** so we disable this feature. */ SCR_REG_REG (scntl2, SCR_AND, 0x7f), 0, /* ** Terminate cycle ... */ SCR_CLR (SCR_ACK|SCR_ATN), 0, /* ** ... and wait for the disconnect. */ SCR_WAIT_DISC, 0, }/*-------------------------< CLEANUP >-------------------*/,{ /* ** dsa: Pointer to nccb ** or xxxxxxFF (no nccb) ** ** HS_REG: Host-Status (<>0!) */ SCR_FROM_REG (dsa), 0, SCR_JUMP ^ IFTRUE (DATA (0xff)), PADDR (signal), /* ** dsa is valid. ** save the status registers */ SCR_COPY (4), RADDR (scr0), NADDR (header.status), /* ** and copy back the header to the nccb. */ SCR_COPY_F (4), RADDR (dsa), PADDR (cleanup0), SCR_COPY (sizeof (struct head)), NADDR (header), }/*-------------------------< CLEANUP0 >--------------------*/,{ 0, /* ** If command resulted in "check condition" ** status and is not yet completed, ** try to get the condition code. */ SCR_FROM_REG (HS_REG), 0, /*<<<*/ SCR_JUMPR ^ IFFALSE (MASK (0, HS_DONEMASK)), 16, SCR_FROM_REG (SS_REG), 0, SCR_JUMP ^ IFTRUE (DATA (SCSI_STATUS_CHECK_COND)), PADDRH(getcc2), }/*-------------------------< SIGNAL >----------------------*/,{ /* ** if status = queue full, ** reinsert in startqueue and stall queue. */ /*>>>*/ SCR_FROM_REG (SS_REG), 0, SCR_INT ^ IFTRUE (DATA (SCSI_STATUS_QUEUE_FULL)), SIR_STALL_QUEUE, /* ** And make the DSA register invalid. */ SCR_LOAD_REG (dsa, 0xff), /* invalid */ 0, /* ** if job completed ... */ SCR_FROM_REG (HS_REG), 0, /* ** ... signal completion to the host */ SCR_INT_FLY ^ IFFALSE (MASK (0, HS_DONEMASK)), 0, /* ** Auf zu neuen Schandtaten! */ SCR_JUMP, PADDR(start), }/*-------------------------< SAVE_DP >------------------*/,{ /* ** SAVE_DP message: ** Copy TEMP register to SAVEP in header. */ SCR_COPY (4), RADDR (temp), NADDR (header.savep), SCR_JUMP, PADDR (clrack), }/*-------------------------< RESTORE_DP >---------------*/,{ /* ** RESTORE_DP message: ** Copy SAVEP in header to TEMP register. */ SCR_COPY (4), NADDR (header.savep), RADDR (temp), SCR_JUMP, PADDR (clrack), }/*-------------------------< DISCONNECT >---------------*/,{ /* ** If QUIRK_AUTOSAVE is set, ** do a "save pointer" operation. */ SCR_FROM_REG (QU_REG), 0, /*<<<*/ SCR_JUMPR ^ IFFALSE (MASK (QUIRK_AUTOSAVE, QUIRK_AUTOSAVE)), 12, /* ** like SAVE_DP message: ** Copy TEMP register to SAVEP in header. */ SCR_COPY (4), RADDR (temp), NADDR (header.savep), /*>>>*/ /* ** Check if temp==savep or temp==goalp: ** if not, log a missing save pointer message. ** In fact, it's a comparison mod 256. ** ** Hmmm, I hadn't thought that I would be urged to ** write this kind of ugly self modifying code. ** ** It's unbelievable, but the ncr53c8xx isn't able ** to subtract one register from another. */ SCR_FROM_REG (temp), 0, /* ** You are not expected to understand this .. ** ** CAUTION: only little endian architectures supported! XXX */ SCR_COPY_F (1), NADDR (header.savep), PADDR (disconnect0), }/*-------------------------< DISCONNECT0 >--------------*/,{ /*<<<*/ SCR_JUMPR ^ IFTRUE (DATA (1)), 20, /* ** neither this */ SCR_COPY_F (1), NADDR (header.goalp), PADDR (disconnect1), }/*-------------------------< DISCONNECT1 >--------------*/,{ SCR_INT ^ IFFALSE (DATA (1)), SIR_MISSING_SAVE, /*>>>*/ /* ** DISCONNECTing ... ** ** disable the "unexpected disconnect" feature, ** and remove the ACK signal. */ SCR_REG_REG (scntl2, SCR_AND, 0x7f), 0, SCR_CLR (SCR_ACK|SCR_ATN), 0, /* ** Wait for the disconnect. */ SCR_WAIT_DISC, 0, /* ** Profiling: ** Set a time stamp, ** and count the disconnects. */ SCR_COPY (sizeof (ticks)), KVAR (KVAR_TICKS), NADDR (header.stamp.disconnect), SCR_COPY (4), NADDR (disc_phys), RADDR (temp), SCR_REG_REG (temp, SCR_ADD, 0x01), 0, SCR_COPY (4), RADDR (temp), NADDR (disc_phys), /* ** Status is: DISCONNECTED. */ SCR_LOAD_REG (HS_REG, HS_DISCONNECT), 0, SCR_JUMP, PADDR (cleanup), }/*-------------------------< MSG_OUT >-------------------*/,{ /* ** The target requests a message. */ SCR_MOVE_ABS (1) ^ SCR_MSG_OUT, NADDR (msgout), SCR_COPY (1), RADDR (sfbr), NADDR (lastmsg), /* ** If it was no ABORT message ... */ SCR_JUMP ^ IFTRUE (DATA (MSG_ABORT)), PADDRH (msg_out_abort), /* ** ... wait for the next phase ** if it's a message out, send it again, ... */ SCR_JUMP ^ IFTRUE (WHEN (SCR_MSG_OUT)), PADDR (msg_out), }/*-------------------------< MSG_OUT_DONE >--------------*/,{ /* ** ... else clear the message ... */ SCR_LOAD_REG (scratcha, MSG_NOOP), 0, SCR_COPY (4), RADDR (scratcha), NADDR (msgout), /* ** ... and process the next phase */ SCR_JUMP, PADDR (dispatch), }/*------------------------< BADGETCC >---------------------*/,{ /* ** If SIGP was set, clear it and try again. */ SCR_FROM_REG (ctest2), 0, SCR_JUMP ^ IFTRUE (MASK (CSIGP,CSIGP)), PADDRH (getcc2), SCR_INT, SIR_SENSE_FAILED, }/*-------------------------< RESELECT >--------------------*/,{ /* ** This NOP will be patched with LED OFF ** SCR_REG_REG (gpreg, SCR_OR, 0x01) */ SCR_NO_OP, 0, /* ** make the DSA invalid. */ SCR_LOAD_REG (dsa, 0xff), 0, SCR_CLR (SCR_TRG), 0, /* ** Sleep waiting for a reselection. ** If SIGP is set, special treatment. ** ** Zu allem bereit .. */ SCR_WAIT_RESEL, PADDR(reselect2), }/*-------------------------< RESELECT1 >--------------------*/,{ /* ** This NOP will be patched with LED ON ** SCR_REG_REG (gpreg, SCR_AND, 0xfe) */ SCR_NO_OP, 0, /* ** ... zu nichts zu gebrauchen ? ** ** load the target id into the SFBR ** and jump to the control block. ** ** Look at the declarations of ** - struct ncb ** - struct tcb ** - struct lcb ** - struct nccb ** to understand what's going on. */ SCR_REG_SFBR (ssid, SCR_AND, 0x8F), 0, SCR_TO_REG (sdid), 0, SCR_JUMP, NADDR (jump_tcb), }/*-------------------------< RESELECT2 >-------------------*/,{ /* ** This NOP will be patched with LED ON ** SCR_REG_REG (gpreg, SCR_AND, 0xfe) */ SCR_NO_OP, 0, /* ** If it's not connected :( ** -> interrupted by SIGP bit. ** Jump to start. */ SCR_FROM_REG (ctest2), 0, SCR_JUMP ^ IFTRUE (MASK (CSIGP,CSIGP)), PADDR (start), SCR_JUMP, PADDR (reselect), }/*-------------------------< RESEL_TMP >-------------------*/,{ /* ** The return address in TEMP ** is in fact the data structure address, ** so copy it to the DSA register. */ SCR_COPY (4), RADDR (temp), RADDR (dsa), SCR_JUMP, PADDR (prepare), }/*-------------------------< RESEL_LUN >-------------------*/,{ /* ** come back to this point ** to get an IDENTIFY message ** Wait for a msg_in phase. */ /*<<<*/ SCR_JUMPR ^ IFFALSE (WHEN (SCR_MSG_IN)), 48, /* ** message phase ** It's not a sony, it's a trick: ** read the data without acknowledging it. */ SCR_FROM_REG (sbdl), 0, /*<<<*/ SCR_JUMPR ^ IFFALSE (MASK (MSG_IDENTIFYFLAG, 0x98)), 32, /* ** It WAS an Identify message. ** get it and ack it! */ SCR_MOVE_ABS (1) ^ SCR_MSG_IN, NADDR (msgin), SCR_CLR (SCR_ACK), 0, /* ** Mask out the lun. */ SCR_REG_REG (sfbr, SCR_AND, 0x07), 0, SCR_RETURN, 0, /* ** No message phase or no IDENTIFY message: ** return 0. */ /*>>>*/ SCR_LOAD_SFBR (0), 0, SCR_RETURN, 0, }/*-------------------------< RESEL_TAG >-------------------*/,{ /* ** come back to this point ** to get a SIMPLE_TAG message ** Wait for a MSG_IN phase. */ /*<<<*/ SCR_JUMPR ^ IFFALSE (WHEN (SCR_MSG_IN)), 64, /* ** message phase ** It's a trick - read the data ** without acknowledging it. */ SCR_FROM_REG (sbdl), 0, /*<<<*/ SCR_JUMPR ^ IFFALSE (DATA (MSG_SIMPLE_Q_TAG)), 48, /* ** It WAS a SIMPLE_TAG message. ** get it and ack it! */ SCR_MOVE_ABS (1) ^ SCR_MSG_IN, NADDR (msgin), SCR_CLR (SCR_ACK), 0, /* ** Wait for the second byte (the tag) */ /*<<<*/ SCR_JUMPR ^ IFFALSE (WHEN (SCR_MSG_IN)), 24, /* ** Get it and ack it! */ SCR_MOVE_ABS (1) ^ SCR_MSG_IN, NADDR (msgin), SCR_CLR (SCR_ACK|SCR_CARRY), 0, SCR_RETURN, 0, /* ** No message phase or no SIMPLE_TAG message ** or no second byte: return 0. */ /*>>>*/ SCR_LOAD_SFBR (0), 0, SCR_SET (SCR_CARRY), 0, SCR_RETURN, 0, }/*-------------------------< DATA_IN >--------------------*/,{ /* ** Because the size depends on the ** #define MAX_SCATTER parameter, ** it is filled in at runtime. ** ** SCR_JUMP ^ IFFALSE (WHEN (SCR_DATA_IN)), ** PADDR (no_data), ** SCR_COPY (sizeof (ticks)), ** KVAR (KVAR_TICKS), ** NADDR (header.stamp.data), ** SCR_MOVE_TBL ^ SCR_DATA_IN, ** offsetof (struct dsb, data[ 0]), ** ** ##===========< i=1; i========= ** || SCR_CALL ^ IFFALSE (WHEN (SCR_DATA_IN)), ** || PADDR (checkatn), ** || SCR_MOVE_TBL ^ SCR_DATA_IN, ** || offsetof (struct dsb, data[ i]), ** ##========================================== ** ** SCR_CALL, ** PADDR (checkatn), ** SCR_JUMP, ** PADDR (no_data), */ 0 }/*-------------------------< DATA_OUT >-------------------*/,{ /* ** Because the size depends on the ** #define MAX_SCATTER parameter, ** it is filled in at runtime. ** ** SCR_JUMP ^ IFFALSE (WHEN (SCR_DATA_OUT)), ** PADDR (no_data), ** SCR_COPY (sizeof (ticks)), ** KVAR (KVAR_TICKS), ** NADDR (header.stamp.data), ** SCR_MOVE_TBL ^ SCR_DATA_OUT, ** offsetof (struct dsb, data[ 0]), ** ** ##===========< i=1; i========= ** || SCR_CALL ^ IFFALSE (WHEN (SCR_DATA_OUT)), ** || PADDR (dispatch), ** || SCR_MOVE_TBL ^ SCR_DATA_OUT, ** || offsetof (struct dsb, data[ i]), ** ##========================================== ** ** SCR_CALL, ** PADDR (dispatch), ** SCR_JUMP, ** PADDR (no_data), ** **--------------------------------------------------------- */ (u_long)0 }/*--------------------------------------------------------*/ }; static struct scripth scripth0 = { /*-------------------------< TRYLOOP >---------------------*/{ /* ** Load an entry of the start queue into dsa ** and try to start it by jumping to TRYSEL. ** ** Because the size depends on the ** #define MAX_START parameter, it is filled ** in at runtime. ** **----------------------------------------------------------- ** ** ##===========< I=0; i=========== ** || SCR_COPY (4), ** || NADDR (squeue[i]), ** || RADDR (dsa), ** || SCR_CALL, ** || PADDR (trysel), ** ##========================================== ** ** SCR_JUMP, ** PADDRH(tryloop), ** **----------------------------------------------------------- */ 0 }/*-------------------------< MSG_PARITY >---------------*/,{ /* ** count it */ SCR_REG_REG (PS_REG, SCR_ADD, 0x01), 0, /* ** send a "message parity error" message. */ SCR_LOAD_REG (scratcha, MSG_PARITY_ERROR), 0, SCR_JUMP, PADDR (setmsg), }/*-------------------------< MSG_MESSAGE_REJECT >---------------*/,{ /* ** If a negotiation was in progress, ** negotiation failed. */ SCR_FROM_REG (HS_REG), 0, SCR_INT ^ IFTRUE (DATA (HS_NEGOTIATE)), SIR_NEGO_FAILED, /* ** else make host log this message */ SCR_INT ^ IFFALSE (DATA (HS_NEGOTIATE)), SIR_REJECT_RECEIVED, SCR_JUMP, PADDR (clrack), }/*-------------------------< MSG_IGN_RESIDUE >----------*/,{ /* ** Terminate cycle */ SCR_CLR (SCR_ACK), 0, SCR_JUMP ^ IFFALSE (WHEN (SCR_MSG_IN)), PADDR (dispatch), /* ** get residue size. */ SCR_MOVE_ABS (1) ^ SCR_MSG_IN, NADDR (msgin[1]), /* ** Check for message parity error. */ SCR_TO_REG (scratcha), 0, SCR_FROM_REG (socl), 0, SCR_JUMP ^ IFTRUE (MASK (CATN, CATN)), PADDRH (msg_parity), SCR_FROM_REG (scratcha), 0, /* ** Size is 0 .. ignore message. */ SCR_JUMP ^ IFTRUE (DATA (0)), PADDR (clrack), /* ** Size is not 1 .. have to interrupt. */ /*<<<*/ SCR_JUMPR ^ IFFALSE (DATA (1)), 40, /* ** Check for residue byte in swide register */ SCR_FROM_REG (scntl2), 0, /*<<<*/ SCR_JUMPR ^ IFFALSE (MASK (WSR, WSR)), 16, /* ** There IS data in the swide register. ** Discard it. */ SCR_REG_REG (scntl2, SCR_OR, WSR), 0, SCR_JUMP, PADDR (clrack), /* ** Load again the size to the sfbr register. */ /*>>>*/ SCR_FROM_REG (scratcha), 0, /*>>>*/ SCR_INT, SIR_IGN_RESIDUE, SCR_JUMP, PADDR (clrack), }/*-------------------------< MSG_EXTENDED >-------------*/,{ /* ** Terminate cycle */ SCR_CLR (SCR_ACK), 0, SCR_JUMP ^ IFFALSE (WHEN (SCR_MSG_IN)), PADDR (dispatch), /* ** get length. */ SCR_MOVE_ABS (1) ^ SCR_MSG_IN, NADDR (msgin[1]), /* ** Check for message parity error. */ SCR_TO_REG (scratcha), 0, SCR_FROM_REG (socl), 0, SCR_JUMP ^ IFTRUE (MASK (CATN, CATN)), PADDRH (msg_parity), SCR_FROM_REG (scratcha), 0, /* */ SCR_JUMP ^ IFTRUE (DATA (3)), PADDRH (msg_ext_3), SCR_JUMP ^ IFFALSE (DATA (2)), PADDR (msg_bad), }/*-------------------------< MSG_EXT_2 >----------------*/,{ SCR_CLR (SCR_ACK), 0, SCR_JUMP ^ IFFALSE (WHEN (SCR_MSG_IN)), PADDR (dispatch), /* ** get extended message code. */ SCR_MOVE_ABS (1) ^ SCR_MSG_IN, NADDR (msgin[2]), /* ** Check for message parity error. */ SCR_TO_REG (scratcha), 0, SCR_FROM_REG (socl), 0, SCR_JUMP ^ IFTRUE (MASK (CATN, CATN)), PADDRH (msg_parity), SCR_FROM_REG (scratcha), 0, SCR_JUMP ^ IFTRUE (DATA (MSG_EXT_WDTR)), PADDRH (msg_wdtr), /* ** unknown extended message */ SCR_JUMP, PADDR (msg_bad) }/*-------------------------< MSG_WDTR >-----------------*/,{ SCR_CLR (SCR_ACK), 0, SCR_JUMP ^ IFFALSE (WHEN (SCR_MSG_IN)), PADDR (dispatch), /* ** get data bus width */ SCR_MOVE_ABS (1) ^ SCR_MSG_IN, NADDR (msgin[3]), SCR_FROM_REG (socl), 0, SCR_JUMP ^ IFTRUE (MASK (CATN, CATN)), PADDRH (msg_parity), /* ** let the host do the real work. */ SCR_INT, SIR_NEGO_WIDE, /* ** let the target fetch our answer. */ SCR_SET (SCR_ATN), 0, SCR_CLR (SCR_ACK), 0, SCR_INT ^ IFFALSE (WHEN (SCR_MSG_OUT)), SIR_NEGO_PROTO, /* ** Send the MSG_EXT_WDTR */ SCR_MOVE_ABS (4) ^ SCR_MSG_OUT, NADDR (msgout), SCR_CLR (SCR_ATN), 0, SCR_COPY (1), RADDR (sfbr), NADDR (lastmsg), SCR_JUMP, PADDR (msg_out_done), }/*-------------------------< MSG_EXT_3 >----------------*/,{ SCR_CLR (SCR_ACK), 0, SCR_JUMP ^ IFFALSE (WHEN (SCR_MSG_IN)), PADDR (dispatch), /* ** get extended message code. */ SCR_MOVE_ABS (1) ^ SCR_MSG_IN, NADDR (msgin[2]), /* ** Check for message parity error. */ SCR_TO_REG (scratcha), 0, SCR_FROM_REG (socl), 0, SCR_JUMP ^ IFTRUE (MASK (CATN, CATN)), PADDRH (msg_parity), SCR_FROM_REG (scratcha), 0, SCR_JUMP ^ IFTRUE (DATA (MSG_EXT_SDTR)), PADDRH (msg_sdtr), /* ** unknown extended message */ SCR_JUMP, PADDR (msg_bad) }/*-------------------------< MSG_SDTR >-----------------*/,{ SCR_CLR (SCR_ACK), 0, SCR_JUMP ^ IFFALSE (WHEN (SCR_MSG_IN)), PADDR (dispatch), /* ** get period and offset */ SCR_MOVE_ABS (2) ^ SCR_MSG_IN, NADDR (msgin[3]), SCR_FROM_REG (socl), 0, SCR_JUMP ^ IFTRUE (MASK (CATN, CATN)), PADDRH (msg_parity), /* ** let the host do the real work. */ SCR_INT, SIR_NEGO_SYNC, /* ** let the target fetch our answer. */ SCR_SET (SCR_ATN), 0, SCR_CLR (SCR_ACK), 0, SCR_INT ^ IFFALSE (WHEN (SCR_MSG_OUT)), SIR_NEGO_PROTO, /* ** Send the MSG_EXT_SDTR */ SCR_MOVE_ABS (5) ^ SCR_MSG_OUT, NADDR (msgout), SCR_CLR (SCR_ATN), 0, SCR_COPY (1), RADDR (sfbr), NADDR (lastmsg), SCR_JUMP, PADDR (msg_out_done), }/*-------------------------< MSG_OUT_ABORT >-------------*/,{ /* ** After ABORT message, ** ** expect an immediate disconnect, ... */ SCR_REG_REG (scntl2, SCR_AND, 0x7f), 0, SCR_CLR (SCR_ACK|SCR_ATN), 0, SCR_WAIT_DISC, 0, /* ** ... and set the status to "ABORTED" */ SCR_LOAD_REG (HS_REG, HS_ABORTED), 0, SCR_JUMP, PADDR (cleanup), }/*-------------------------< GETCC >-----------------------*/,{ /* ** The ncr doesn't have an indirect load ** or store command. So we have to ** copy part of the control block to a ** fixed place, where we can modify it. ** ** We patch the address part of a COPY command ** with the address of the dsa register ... */ SCR_COPY_F (4), RADDR (dsa), PADDRH (getcc1), /* ** ... then we do the actual copy. */ SCR_COPY (sizeof (struct head)), }/*-------------------------< GETCC1 >----------------------*/,{ 0, NADDR (header), /* ** Initialize the status registers */ SCR_COPY (4), NADDR (header.status), RADDR (scr0), }/*-------------------------< GETCC2 >----------------------*/,{ /* ** Get the condition code from a target. ** ** DSA points to a data structure. ** Set TEMP to the script location ** that receives the condition code. ** ** Because there is no script command ** to load a longword into a register, ** we use a CALL command. */ /*<<<*/ SCR_CALLR, 24, /* ** Get the condition code. */ SCR_MOVE_TBL ^ SCR_DATA_IN, offsetof (struct dsb, sense), /* ** No data phase may follow! */ SCR_CALL, PADDR (checkatn), SCR_JUMP, PADDR (no_data), /*>>>*/ /* ** The CALL jumps to this point. ** Prepare for a RESTORE_POINTER message. ** Save the TEMP register into the saved pointer. */ SCR_COPY (4), RADDR (temp), NADDR (header.savep), /* ** Load scratcha, because in case of a selection timeout, ** the host will expect a new value for startpos in ** the scratcha register. */ SCR_COPY (4), PADDR (startpos), RADDR (scratcha), #ifdef NCR_GETCC_WITHMSG /* ** If QUIRK_NOMSG is set, select without ATN. ** and don't send a message. */ SCR_FROM_REG (QU_REG), 0, SCR_JUMP ^ IFTRUE (MASK (QUIRK_NOMSG, QUIRK_NOMSG)), PADDRH(getcc3), /* ** Then try to connect to the target. ** If we are reselected, special treatment ** of the current job is required before ** accepting the reselection. */ SCR_SEL_TBL_ATN ^ offsetof (struct dsb, select), PADDR(badgetcc), /* ** Send the IDENTIFY message. ** In case of short transfer, remove ATN. */ SCR_MOVE_TBL ^ SCR_MSG_OUT, offsetof (struct dsb, smsg2), SCR_CLR (SCR_ATN), 0, /* ** save the first byte of the message. */ SCR_COPY (1), RADDR (sfbr), NADDR (lastmsg), SCR_JUMP, PADDR (prepare2), #endif }/*-------------------------< GETCC3 >----------------------*/,{ /* ** Try to connect to the target. ** If we are reselected, special treatment ** of the current job is required before ** accepting the reselection. ** ** Silly target won't accept a message. ** Select without ATN. */ SCR_SEL_TBL ^ offsetof (struct dsb, select), PADDR(badgetcc), /* ** Force error if selection timeout */ SCR_JUMPR ^ IFTRUE (WHEN (SCR_MSG_IN)), 0, /* ** don't negotiate. */ SCR_JUMP, PADDR (prepare2), }/*-------------------------< ABORTTAG >-------------------*/,{ /* ** Abort a bad reselection. ** Set the message to ABORT vs. ABORT_TAG */ SCR_LOAD_REG (scratcha, MSG_ABORT_TAG), 0, SCR_JUMPR ^ IFFALSE (CARRYSET), 8, }/*-------------------------< ABORT >----------------------*/,{ SCR_LOAD_REG (scratcha, MSG_ABORT), 0, SCR_COPY (1), RADDR (scratcha), NADDR (msgout), SCR_SET (SCR_ATN), 0, SCR_CLR (SCR_ACK), 0, /* ** and send it. ** we expect an immediate disconnect */ SCR_REG_REG (scntl2, SCR_AND, 0x7f), 0, SCR_MOVE_ABS (1) ^ SCR_MSG_OUT, NADDR (msgout), SCR_COPY (1), RADDR (sfbr), NADDR (lastmsg), SCR_CLR (SCR_ACK|SCR_ATN), 0, SCR_WAIT_DISC, 0, SCR_JUMP, PADDR (start), }/*-------------------------< SNOOPTEST >-------------------*/,{ /* ** Read the variable. */ SCR_COPY (4), KVAR (KVAR_NCR_CACHE), RADDR (scratcha), /* ** Write the variable. */ SCR_COPY (4), RADDR (temp), KVAR (KVAR_NCR_CACHE), /* ** Read back the variable. */ SCR_COPY (4), KVAR (KVAR_NCR_CACHE), RADDR (temp), }/*-------------------------< SNOOPEND >-------------------*/,{ /* ** And stop. */ SCR_INT, 99, }/*--------------------------------------------------------*/ }; /*========================================================== ** ** ** Fill in #define dependent parts of the script ** ** **========================================================== */ static void ncr_script_fill (struct script * scr, struct scripth * scrh) { int i; ncrcmd *p; p = scrh->tryloop; for (i=0; itryloop + sizeof (scrh->tryloop)); p = scr->data_in; *p++ =SCR_JUMP ^ IFFALSE (WHEN (SCR_DATA_IN)); *p++ =PADDR (no_data); *p++ =SCR_COPY (sizeof (ticks)); *p++ =(ncrcmd) KVAR (KVAR_TICKS); *p++ =NADDR (header.stamp.data); *p++ =SCR_MOVE_TBL ^ SCR_DATA_IN; *p++ =offsetof (struct dsb, data[ 0]); for (i=1; idata_in + sizeof (scr->data_in)); p = scr->data_out; *p++ =SCR_JUMP ^ IFFALSE (WHEN (SCR_DATA_OUT)); *p++ =PADDR (no_data); *p++ =SCR_COPY (sizeof (ticks)); *p++ =(ncrcmd) KVAR (KVAR_TICKS); *p++ =NADDR (header.stamp.data); *p++ =SCR_MOVE_TBL ^ SCR_DATA_OUT; *p++ =offsetof (struct dsb, data[ 0]); for (i=1; idata_out + sizeof (scr->data_out)); } /*========================================================== ** ** ** Copy and rebind a script. ** ** **========================================================== */ static void ncr_script_copy_and_bind (ncb_p np, ncrcmd *src, ncrcmd *dst, int len) { ncrcmd opcode, new, old, tmp1, tmp2; ncrcmd *start, *end; int relocs, offset; start = src; end = src + len/4; offset = 0; while (src < end) { opcode = *src++; WRITESCRIPT_OFF(dst, offset, opcode); offset += 4; /* ** If we forget to change the length ** in struct script, a field will be ** padded with 0. This is an illegal ** command. */ if (opcode == 0) { device_printf(np->dev, "ERROR0 IN SCRIPT at %d.\n", (int)(src - start - 1)); DELAY (1000000); } if (DEBUG_FLAGS & DEBUG_SCRIPT) printf ("%p: <%x>\n", (src-1), (unsigned)opcode); /* ** We don't have to decode ALL commands */ switch (opcode >> 28) { case 0xc: /* ** COPY has TWO arguments. */ relocs = 2; tmp1 = src[0]; if ((tmp1 & RELOC_MASK) == RELOC_KVAR) tmp1 = 0; tmp2 = src[1]; if ((tmp2 & RELOC_MASK) == RELOC_KVAR) tmp2 = 0; if ((tmp1 ^ tmp2) & 3) { device_printf(np->dev, "ERROR1 IN SCRIPT at %d.\n", (int)(src - start - 1)); DELAY (1000000); } /* ** If PREFETCH feature not enabled, remove ** the NO FLUSH bit if present. */ if ((opcode & SCR_NO_FLUSH) && !(np->features&FE_PFEN)) WRITESCRIPT_OFF(dst, offset - 4, (opcode & ~SCR_NO_FLUSH)); break; case 0x0: /* ** MOVE (absolute address) */ relocs = 1; break; case 0x8: /* ** JUMP / CALL ** dont't relocate if relative :-) */ if (opcode & 0x00800000) relocs = 0; else relocs = 1; break; case 0x4: case 0x5: case 0x6: case 0x7: relocs = 1; break; default: relocs = 0; break; } if (relocs) { while (relocs--) { old = *src++; switch (old & RELOC_MASK) { case RELOC_REGISTER: new = (old & ~RELOC_MASK) + rman_get_start(np->reg_res); break; case RELOC_LABEL: new = (old & ~RELOC_MASK) + np->p_script; break; case RELOC_LABELH: new = (old & ~RELOC_MASK) + np->p_scripth; break; case RELOC_SOFTC: new = (old & ~RELOC_MASK) + vtophys(np); break; case RELOC_KVAR: if (((old & ~RELOC_MASK) < SCRIPT_KVAR_FIRST) || ((old & ~RELOC_MASK) > SCRIPT_KVAR_LAST)) panic("ncr KVAR out of range"); new = vtophys(script_kvars[old & ~RELOC_MASK]); break; case 0: /* Don't relocate a 0 address. */ if (old == 0) { new = old; break; } /* FALLTHROUGH */ default: panic("ncr_script_copy_and_bind: weird relocation %x @ %d\n", old, (int)(src - start)); break; } WRITESCRIPT_OFF(dst, offset, new); offset += 4; } } else { WRITESCRIPT_OFF(dst, offset, *src++); offset += 4; } } } /*========================================================== ** ** ** Auto configuration. ** ** **========================================================== */ #if 0 /*---------------------------------------------------------- ** ** Reduce the transfer length to the max value ** we can transfer safely. ** ** Reading a block greater then MAX_SIZE from the ** raw (character) device exercises a memory leak ** in the vm subsystem. This is common to ALL devices. ** We have submitted a description of this bug to ** . ** It should be fixed in the current release. ** **---------------------------------------------------------- */ void ncr_min_phys (struct buf *bp) { if ((unsigned long)bp->b_bcount > MAX_SIZE) bp->b_bcount = MAX_SIZE; } #endif #if 0 /*---------------------------------------------------------- ** ** Maximal number of outstanding requests per target. ** **---------------------------------------------------------- */ u_int32_t ncr_info (int unit) { return (1); /* may be changed later */ } #endif /*---------------------------------------------------------- ** ** NCR chip devices table and chip look up function. ** Features bit are defined in ncrreg.h. Is it the ** right place? ** **---------------------------------------------------------- */ typedef struct { unsigned long device_id; unsigned short minrevid; char *name; unsigned char maxburst; unsigned char maxoffs; unsigned char clock_divn; unsigned int features; } ncr_chip; static ncr_chip ncr_chip_table[] = { {NCR_810_ID, 0x00, "ncr 53c810 fast10 scsi", 4, 8, 4, FE_ERL} , {NCR_810_ID, 0x10, "ncr 53c810a fast10 scsi", 4, 8, 4, FE_ERL|FE_LDSTR|FE_PFEN|FE_BOF} , {NCR_815_ID, 0x00, "ncr 53c815 fast10 scsi", 4, 8, 4, FE_ERL|FE_BOF} , {NCR_820_ID, 0x00, "ncr 53c820 fast10 wide scsi", 4, 8, 4, FE_WIDE|FE_ERL} , {NCR_825_ID, 0x00, "ncr 53c825 fast10 wide scsi", 4, 8, 4, FE_WIDE|FE_ERL|FE_BOF} , {NCR_825_ID, 0x10, "ncr 53c825a fast10 wide scsi", 7, 8, 4, FE_WIDE|FE_CACHE_SET|FE_DFS|FE_LDSTR|FE_PFEN|FE_RAM} , {NCR_860_ID, 0x00, "ncr 53c860 fast20 scsi", 4, 8, 5, FE_ULTRA|FE_CLK80|FE_CACHE_SET|FE_LDSTR|FE_PFEN} , {NCR_875_ID, 0x00, "ncr 53c875 fast20 wide scsi", 7, 16, 5, FE_WIDE|FE_ULTRA|FE_CLK80|FE_CACHE_SET|FE_DFS|FE_LDSTR|FE_PFEN|FE_RAM} , {NCR_875_ID, 0x02, "ncr 53c875 fast20 wide scsi", 7, 16, 5, FE_WIDE|FE_ULTRA|FE_DBLR|FE_CACHE_SET|FE_DFS|FE_LDSTR|FE_PFEN|FE_RAM} , {NCR_875_ID2, 0x00, "ncr 53c875j fast20 wide scsi", 7, 16, 5, FE_WIDE|FE_ULTRA|FE_DBLR|FE_CACHE_SET|FE_DFS|FE_LDSTR|FE_PFEN|FE_RAM} , {NCR_885_ID, 0x00, "ncr 53c885 fast20 wide scsi", 7, 16, 5, FE_WIDE|FE_ULTRA|FE_DBLR|FE_CACHE_SET|FE_DFS|FE_LDSTR|FE_PFEN|FE_RAM} , {NCR_895_ID, 0x00, "ncr 53c895 fast40 wide scsi", 7, 31, 7, FE_WIDE|FE_ULTRA2|FE_QUAD|FE_CACHE_SET|FE_DFS|FE_LDSTR|FE_PFEN|FE_RAM} , {NCR_896_ID, 0x00, "ncr 53c896 fast40 wide scsi", 7, 31, 7, FE_WIDE|FE_ULTRA2|FE_QUAD|FE_CACHE_SET|FE_DFS|FE_LDSTR|FE_PFEN|FE_RAM} , {NCR_895A_ID, 0x00, "ncr 53c895a fast40 wide scsi", 7, 31, 7, FE_WIDE|FE_ULTRA2|FE_QUAD|FE_CACHE_SET|FE_DFS|FE_LDSTR|FE_PFEN|FE_RAM} , {NCR_1510D_ID, 0x00, "ncr 53c1510d fast40 wide scsi", 7, 31, 7, FE_WIDE|FE_ULTRA2|FE_QUAD|FE_CACHE_SET|FE_DFS|FE_LDSTR|FE_PFEN|FE_RAM} }; static int ncr_chip_lookup(u_long device_id, u_char revision_id) { int i, found; found = -1; for (i = 0; i < nitems(ncr_chip_table); i++) { if (device_id == ncr_chip_table[i].device_id && ncr_chip_table[i].minrevid <= revision_id) { if (found < 0 || ncr_chip_table[found].minrevid < ncr_chip_table[i].minrevid) { found = i; } } } return found; } /*---------------------------------------------------------- ** ** Probe the hostadapter. ** **---------------------------------------------------------- */ static int ncr_probe (device_t dev) { int i; i = ncr_chip_lookup(pci_get_devid(dev), pci_get_revid(dev)); if (i >= 0) { device_set_desc(dev, ncr_chip_table[i].name); return (BUS_PROBE_DEFAULT); } return (ENXIO); } /*========================================================== ** ** NCR chip clock divisor table. ** Divisors are multiplied by 10,000,000 in order to make ** calculations more simple. ** **========================================================== */ #define _5M 5000000 static u_long div_10M[] = {2*_5M, 3*_5M, 4*_5M, 6*_5M, 8*_5M, 12*_5M, 16*_5M}; /*=============================================================== ** ** NCR chips allow burst lengths of 2, 4, 8, 16, 32, 64, 128 ** transfers. 32,64,128 are only supported by 875 and 895 chips. ** We use log base 2 (burst length) as internal code, with ** value 0 meaning "burst disabled". ** **=============================================================== */ /* * Burst length from burst code. */ #define burst_length(bc) (!(bc))? 0 : 1 << (bc) /* * Burst code from io register bits. */ #define burst_code(dmode, ctest4, ctest5) \ (ctest4) & 0x80? 0 : (((dmode) & 0xc0) >> 6) + ((ctest5) & 0x04) + 1 /* * Set initial io register bits from burst code. */ static void ncr_init_burst(ncb_p np, u_char bc) { np->rv_ctest4 &= ~0x80; np->rv_dmode &= ~(0x3 << 6); np->rv_ctest5 &= ~0x4; if (!bc) { np->rv_ctest4 |= 0x80; } else { --bc; np->rv_dmode |= ((bc & 0x3) << 6); np->rv_ctest5 |= (bc & 0x4); } } /*========================================================== ** ** ** Auto configuration: attach and init a host adapter. ** ** **========================================================== */ static int ncr_attach (device_t dev) { ncb_p np = (struct ncb*) device_get_softc(dev); u_char rev = 0; u_long period; int i, rid; u_int8_t usrsync; u_int8_t usrwide; struct cam_devq *devq; /* ** allocate and initialize structures. */ np->dev = dev; mtx_init(&np->lock, "ncr", NULL, MTX_DEF); callout_init_mtx(&np->timer, &np->lock, 0); /* ** Try to map the controller chip to ** virtual and physical memory. */ np->reg_rid = PCIR_BAR(1); np->reg_res = bus_alloc_resource_any(dev, SYS_RES_MEMORY, &np->reg_rid, RF_ACTIVE); if (!np->reg_res) { device_printf(dev, "could not map memory\n"); return ENXIO; } /* ** Now the INB INW INL OUTB OUTW OUTL macros ** can be used safely. */ #ifdef NCR_IOMAPPED /* ** Try to map the controller chip into iospace. */ if (!pci_map_port (config_id, 0x10, &np->port)) return; #endif /* ** Save some controller register default values */ np->rv_scntl3 = INB(nc_scntl3) & 0x77; np->rv_dmode = INB(nc_dmode) & 0xce; np->rv_dcntl = INB(nc_dcntl) & 0xa9; np->rv_ctest3 = INB(nc_ctest3) & 0x01; np->rv_ctest4 = INB(nc_ctest4) & 0x88; np->rv_ctest5 = INB(nc_ctest5) & 0x24; np->rv_gpcntl = INB(nc_gpcntl); np->rv_stest2 = INB(nc_stest2) & 0x20; if (bootverbose >= 2) { printf ("\tBIOS values: SCNTL3:%02x DMODE:%02x DCNTL:%02x\n", np->rv_scntl3, np->rv_dmode, np->rv_dcntl); printf ("\t CTEST3:%02x CTEST4:%02x CTEST5:%02x\n", np->rv_ctest3, np->rv_ctest4, np->rv_ctest5); } np->rv_dcntl |= NOCOM; /* ** Do chip dependent initialization. */ rev = pci_get_revid(dev); /* ** Get chip features from chips table. */ i = ncr_chip_lookup(pci_get_devid(dev), rev); if (i >= 0) { np->maxburst = ncr_chip_table[i].maxburst; np->maxoffs = ncr_chip_table[i].maxoffs; np->clock_divn = ncr_chip_table[i].clock_divn; np->features = ncr_chip_table[i].features; } else { /* Should'nt happen if probe() is ok */ np->maxburst = 4; np->maxoffs = 8; np->clock_divn = 4; np->features = FE_ERL; } np->maxwide = np->features & FE_WIDE ? 1 : 0; np->clock_khz = np->features & FE_CLK80 ? 80000 : 40000; if (np->features & FE_QUAD) np->multiplier = 4; else if (np->features & FE_DBLR) np->multiplier = 2; else np->multiplier = 1; /* ** Get the frequency of the chip's clock. ** Find the right value for scntl3. */ if (np->features & (FE_ULTRA|FE_ULTRA2)) ncr_getclock(np, np->multiplier); #ifdef NCR_TEKRAM_EEPROM if (bootverbose) { device_printf(dev, "Tekram EEPROM read %s\n", read_tekram_eeprom (np, NULL) ? "succeeded" : "failed"); } #endif /* NCR_TEKRAM_EEPROM */ /* * If scntl3 != 0, we assume BIOS is present. */ if (np->rv_scntl3) np->features |= FE_BIOS; /* * Divisor to be used for async (timer pre-scaler). */ i = np->clock_divn - 1; while (i >= 0) { --i; if (10ul * SCSI_NCR_MIN_ASYNC * np->clock_khz > div_10M[i]) { ++i; break; } } np->rv_scntl3 = i+1; /* * Minimum synchronous period factor supported by the chip. * Btw, 'period' is in tenths of nanoseconds. */ period = howmany(4 * div_10M[0], np->clock_khz); if (period <= 250) np->minsync = 10; else if (period <= 303) np->minsync = 11; else if (period <= 500) np->minsync = 12; else np->minsync = howmany(period, 40); /* * Check against chip SCSI standard support (SCSI-2,ULTRA,ULTRA2). */ if (np->minsync < 25 && !(np->features & (FE_ULTRA|FE_ULTRA2))) np->minsync = 25; else if (np->minsync < 12 && !(np->features & FE_ULTRA2)) np->minsync = 12; /* * Maximum synchronous period factor supported by the chip. */ period = (11 * div_10M[np->clock_divn - 1]) / (4 * np->clock_khz); np->maxsync = period > 2540 ? 254 : period / 10; /* * Now, some features available with Symbios compatible boards. * LED support through GPIO0 and DIFF support. */ #ifdef SCSI_NCR_SYMBIOS_COMPAT if (!(np->rv_gpcntl & 0x01)) np->features |= FE_LED0; #if 0 /* Not safe enough without NVRAM support or user settable option */ if (!(INB(nc_gpreg) & 0x08)) np->features |= FE_DIFF; #endif #endif /* SCSI_NCR_SYMBIOS_COMPAT */ /* * Prepare initial IO registers settings. * Trust BIOS only if we believe we have one and if we want to. */ #ifdef SCSI_NCR_TRUST_BIOS if (!(np->features & FE_BIOS)) { #else if (1) { #endif np->rv_dmode = 0; np->rv_dcntl = NOCOM; np->rv_ctest3 = 0; np->rv_ctest4 = MPEE; np->rv_ctest5 = 0; np->rv_stest2 = 0; if (np->features & FE_ERL) np->rv_dmode |= ERL; /* Enable Read Line */ if (np->features & FE_BOF) np->rv_dmode |= BOF; /* Burst Opcode Fetch */ if (np->features & FE_ERMP) np->rv_dmode |= ERMP; /* Enable Read Multiple */ if (np->features & FE_CLSE) np->rv_dcntl |= CLSE; /* Cache Line Size Enable */ if (np->features & FE_WRIE) np->rv_ctest3 |= WRIE; /* Write and Invalidate */ if (np->features & FE_PFEN) np->rv_dcntl |= PFEN; /* Prefetch Enable */ if (np->features & FE_DFS) np->rv_ctest5 |= DFS; /* Dma Fifo Size */ if (np->features & FE_DIFF) np->rv_stest2 |= 0x20; /* Differential mode */ ncr_init_burst(np, np->maxburst); /* Max dwords burst length */ } else { np->maxburst = burst_code(np->rv_dmode, np->rv_ctest4, np->rv_ctest5); } /* ** Get on-chip SRAM address, if supported */ if ((np->features & FE_RAM) && sizeof(struct script) <= 4096) { np->sram_rid = PCIR_BAR(2); np->sram_res = bus_alloc_resource_any(dev, SYS_RES_MEMORY, &np->sram_rid, RF_ACTIVE); } /* ** Allocate structure for script relocation. */ if (np->sram_res != NULL) { np->script = NULL; np->p_script = rman_get_start(np->sram_res); } else if (sizeof (struct script) > PAGE_SIZE) { np->script = (struct script*) contigmalloc (round_page(sizeof (struct script)), M_DEVBUF, M_WAITOK, 0, 0xffffffff, PAGE_SIZE, 0); } else { np->script = (struct script *) malloc (sizeof (struct script), M_DEVBUF, M_WAITOK); } if (sizeof (struct scripth) > PAGE_SIZE) { np->scripth = (struct scripth*) contigmalloc (round_page(sizeof (struct scripth)), M_DEVBUF, M_WAITOK, 0, 0xffffffff, PAGE_SIZE, 0); } else { np->scripth = (struct scripth *) malloc (sizeof (struct scripth), M_DEVBUF, M_WAITOK); } #ifdef SCSI_NCR_PCI_CONFIG_FIXUP /* ** If cache line size is enabled, check PCI config space and ** try to fix it up if necessary. */ #ifdef PCIR_CACHELNSZ /* To be sure that new PCI stuff is present */ { u_char cachelnsz = pci_read_config(dev, PCIR_CACHELNSZ, 1); u_short command = pci_read_config(dev, PCIR_COMMAND, 2); if (!cachelnsz) { cachelnsz = 8; device_printf(dev, "setting PCI cache line size register to %d.\n", (int)cachelnsz); pci_write_config(dev, PCIR_CACHELNSZ, cachelnsz, 1); } if (!(command & PCIM_CMD_MWRICEN)) { command |= PCIM_CMD_MWRICEN; device_printf(dev, "setting PCI command write and invalidate.\n"); pci_write_config(dev, PCIR_COMMAND, command, 2); } } #endif /* PCIR_CACHELNSZ */ #endif /* SCSI_NCR_PCI_CONFIG_FIXUP */ /* Initialize per-target user settings */ usrsync = 0; if (SCSI_NCR_DFLT_SYNC) { usrsync = SCSI_NCR_DFLT_SYNC; if (usrsync > np->maxsync) usrsync = np->maxsync; if (usrsync < np->minsync) usrsync = np->minsync; } usrwide = (SCSI_NCR_MAX_WIDE); if (usrwide > np->maxwide) usrwide=np->maxwide; for (i=0;itarget[i]; tp->tinfo.user.period = usrsync; tp->tinfo.user.offset = usrsync != 0 ? np->maxoffs : 0; tp->tinfo.user.width = usrwide; tp->tinfo.disc_tag = NCR_CUR_DISCENB | NCR_CUR_TAGENB | NCR_USR_DISCENB | NCR_USR_TAGENB; } /* ** Bells and whistles ;-) */ if (bootverbose) device_printf(dev, "minsync=%d, maxsync=%d, maxoffs=%d, %d dwords burst, %s dma fifo\n", np->minsync, np->maxsync, np->maxoffs, burst_length(np->maxburst), (np->rv_ctest5 & DFS) ? "large" : "normal"); /* ** Print some complementary information that can be helpful. */ if (bootverbose) device_printf(dev, "%s, %s IRQ driver%s\n", np->rv_stest2 & 0x20 ? "differential" : "single-ended", np->rv_dcntl & IRQM ? "totem pole" : "open drain", np->sram_res ? ", using on-chip SRAM" : ""); /* ** Patch scripts to physical addresses */ ncr_script_fill (&script0, &scripth0); if (np->script) np->p_script = vtophys(np->script); np->p_scripth = vtophys(np->scripth); ncr_script_copy_and_bind (np, (ncrcmd *) &script0, (ncrcmd *) np->script, sizeof(struct script)); ncr_script_copy_and_bind (np, (ncrcmd *) &scripth0, (ncrcmd *) np->scripth, sizeof(struct scripth)); /* ** Patch the script for LED support. */ if (np->features & FE_LED0) { WRITESCRIPT(reselect[0], SCR_REG_REG(gpreg, SCR_OR, 0x01)); WRITESCRIPT(reselect1[0], SCR_REG_REG(gpreg, SCR_AND, 0xfe)); WRITESCRIPT(reselect2[0], SCR_REG_REG(gpreg, SCR_AND, 0xfe)); } /* ** init data structure */ np->jump_tcb.l_cmd = SCR_JUMP; np->jump_tcb.l_paddr = NCB_SCRIPTH_PHYS (np, abort); /* ** Get SCSI addr of host adapter (set by bios?). */ np->myaddr = INB(nc_scid) & 0x07; if (!np->myaddr) np->myaddr = SCSI_NCR_MYADDR; #ifdef NCR_DUMP_REG /* ** Log the initial register contents */ { int reg; for (reg=0; reg<256; reg+=4) { if (reg%16==0) printf ("reg[%2x]", reg); printf (" %08x", (int)pci_conf_read (config_id, reg)); if (reg%16==12) printf ("\n"); } } #endif /* NCR_DUMP_REG */ /* ** Reset chip. */ OUTB (nc_istat, SRST); DELAY (1000); OUTB (nc_istat, 0 ); /* ** Now check the cache handling of the pci chipset. */ if (ncr_snooptest (np)) { printf ("CACHE INCORRECTLY CONFIGURED.\n"); return EINVAL; } /* ** Install the interrupt handler. */ rid = 0; np->irq_res = bus_alloc_resource_any(dev, SYS_RES_IRQ, &rid, RF_SHAREABLE | RF_ACTIVE); if (np->irq_res == NULL) { device_printf(dev, "interruptless mode: reduced performance.\n"); } else { bus_setup_intr(dev, np->irq_res, INTR_TYPE_CAM | INTR_ENTROPY | INTR_MPSAFE, NULL, ncr_intr, np, &np->irq_handle); } /* ** Create the device queue. We only allow MAX_START-1 concurrent ** transactions so we can be sure to have one element free in our ** start queue to reset to the idle loop. */ devq = cam_simq_alloc(MAX_START - 1); if (devq == NULL) return ENOMEM; /* ** Now tell the generic SCSI layer ** about our bus. */ np->sim = cam_sim_alloc(ncr_action, ncr_poll, "ncr", np, device_get_unit(dev), &np->lock, 1, MAX_TAGS, devq); if (np->sim == NULL) { cam_simq_free(devq); return ENOMEM; } mtx_lock(&np->lock); if (xpt_bus_register(np->sim, dev, 0) != CAM_SUCCESS) { cam_sim_free(np->sim, /*free_devq*/ TRUE); mtx_unlock(&np->lock); return ENOMEM; } if (xpt_create_path(&np->path, /*periph*/NULL, cam_sim_path(np->sim), CAM_TARGET_WILDCARD, CAM_LUN_WILDCARD) != CAM_REQ_CMP) { xpt_bus_deregister(cam_sim_path(np->sim)); cam_sim_free(np->sim, /*free_devq*/TRUE); mtx_unlock(&np->lock); return ENOMEM; } /* ** start the timeout daemon */ ncr_timeout (np); np->lasttime=0; mtx_unlock(&np->lock); return 0; } /*========================================================== ** ** ** Process pending device interrupts. ** ** **========================================================== */ static void ncr_intr(vnp) void *vnp; { ncb_p np = vnp; mtx_lock(&np->lock); ncr_intr_locked(np); mtx_unlock(&np->lock); } static void ncr_intr_locked(ncb_p np) { if (DEBUG_FLAGS & DEBUG_TINY) printf ("["); if (INB(nc_istat) & (INTF|SIP|DIP)) { /* ** Repeat until no outstanding ints */ do { ncr_exception (np); } while (INB(nc_istat) & (INTF|SIP|DIP)); np->ticks = 100; } if (DEBUG_FLAGS & DEBUG_TINY) printf ("]\n"); } /*========================================================== ** ** ** Start execution of a SCSI command. ** This is called from the generic SCSI driver. ** ** **========================================================== */ static void ncr_action (struct cam_sim *sim, union ccb *ccb) { ncb_p np; np = (ncb_p) cam_sim_softc(sim); mtx_assert(&np->lock, MA_OWNED); switch (ccb->ccb_h.func_code) { /* Common cases first */ case XPT_SCSI_IO: /* Execute the requested I/O operation */ { nccb_p cp; lcb_p lp; tcb_p tp; struct ccb_scsiio *csio; u_int8_t *msgptr; u_int msglen; u_int msglen2; int segments; u_int8_t nego; u_int8_t idmsg; int qidx; tp = &np->target[ccb->ccb_h.target_id]; csio = &ccb->csio; /* * Make sure we support this request. We can't do * PHYS pointers. */ if (ccb->ccb_h.flags & CAM_CDB_PHYS) { ccb->ccb_h.status = CAM_REQ_INVALID; xpt_done(ccb); return; } /* * Last time we need to check if this CCB needs to * be aborted. */ if ((ccb->ccb_h.status & CAM_STATUS_MASK) != CAM_REQ_INPROG) { xpt_done(ccb); return; } ccb->ccb_h.status |= CAM_SIM_QUEUED; /*--------------------------------------------------- ** ** Assign an nccb / bind ccb ** **---------------------------------------------------- */ cp = ncr_get_nccb (np, ccb->ccb_h.target_id, ccb->ccb_h.target_lun); if (cp == NULL) { /* XXX JGibbs - Freeze SIMQ */ ccb->ccb_h.status = CAM_RESRC_UNAVAIL; xpt_done(ccb); return; } cp->ccb = ccb; /*--------------------------------------------------- ** ** timestamp ** **---------------------------------------------------- */ /* ** XXX JGibbs - Isn't this expensive ** enough to be conditionalized?? */ bzero (&cp->phys.header.stamp, sizeof (struct tstamp)); cp->phys.header.stamp.start = ticks; nego = 0; if (tp->nego_cp == NULL) { if (tp->tinfo.current.width != tp->tinfo.goal.width) { tp->nego_cp = cp; nego = NS_WIDE; } else if ((tp->tinfo.current.period != tp->tinfo.goal.period) || (tp->tinfo.current.offset != tp->tinfo.goal.offset)) { tp->nego_cp = cp; nego = NS_SYNC; } } /*--------------------------------------------------- ** ** choose a new tag ... ** **---------------------------------------------------- */ lp = tp->lp[ccb->ccb_h.target_lun]; if ((ccb->ccb_h.flags & CAM_TAG_ACTION_VALID) != 0 && (ccb->csio.tag_action != CAM_TAG_ACTION_NONE) && (nego == 0)) { /* ** assign a tag to this nccb */ while (!cp->tag) { nccb_p cp2 = lp->next_nccb; lp->lasttag = lp->lasttag % 255 + 1; while (cp2 && cp2->tag != lp->lasttag) cp2 = cp2->next_nccb; if (cp2) continue; cp->tag=lp->lasttag; if (DEBUG_FLAGS & DEBUG_TAGS) { PRINT_ADDR(ccb); printf ("using tag #%d.\n", cp->tag); } } } else { cp->tag=0; } /*---------------------------------------------------- ** ** Build the identify / tag / sdtr message ** **---------------------------------------------------- */ idmsg = MSG_IDENTIFYFLAG | ccb->ccb_h.target_lun; if (tp->tinfo.disc_tag & NCR_CUR_DISCENB) idmsg |= MSG_IDENTIFY_DISCFLAG; msgptr = cp->scsi_smsg; msglen = 0; msgptr[msglen++] = idmsg; if (cp->tag) { msgptr[msglen++] = ccb->csio.tag_action; msgptr[msglen++] = cp->tag; } switch (nego) { case NS_SYNC: msgptr[msglen++] = MSG_EXTENDED; msgptr[msglen++] = MSG_EXT_SDTR_LEN; msgptr[msglen++] = MSG_EXT_SDTR; msgptr[msglen++] = tp->tinfo.goal.period; msgptr[msglen++] = tp->tinfo.goal.offset; if (DEBUG_FLAGS & DEBUG_NEGO) { PRINT_ADDR(ccb); printf ("sync msgout: "); ncr_show_msg (&cp->scsi_smsg [msglen-5]); printf (".\n"); }; break; case NS_WIDE: msgptr[msglen++] = MSG_EXTENDED; msgptr[msglen++] = MSG_EXT_WDTR_LEN; msgptr[msglen++] = MSG_EXT_WDTR; msgptr[msglen++] = tp->tinfo.goal.width; if (DEBUG_FLAGS & DEBUG_NEGO) { PRINT_ADDR(ccb); printf ("wide msgout: "); ncr_show_msg (&cp->scsi_smsg [msglen-4]); printf (".\n"); }; break; } /*---------------------------------------------------- ** ** Build the identify message for getcc. ** **---------------------------------------------------- */ cp->scsi_smsg2 [0] = idmsg; msglen2 = 1; /*---------------------------------------------------- ** ** Build the data descriptors ** **---------------------------------------------------- */ /* XXX JGibbs - Handle other types of I/O */ if ((ccb->ccb_h.flags & CAM_DIR_MASK) != CAM_DIR_NONE) { segments = ncr_scatter(&cp->phys, (vm_offset_t)csio->data_ptr, (vm_size_t)csio->dxfer_len); if (segments < 0) { ccb->ccb_h.status = CAM_REQ_TOO_BIG; ncr_free_nccb(np, cp); xpt_done(ccb); return; } if ((ccb->ccb_h.flags & CAM_DIR_MASK) == CAM_DIR_IN) { cp->phys.header.savep = NCB_SCRIPT_PHYS (np, data_in); cp->phys.header.goalp = cp->phys.header.savep +20 +segments*16; } else { /* CAM_DIR_OUT */ cp->phys.header.savep = NCB_SCRIPT_PHYS (np, data_out); cp->phys.header.goalp = cp->phys.header.savep +20 +segments*16; } } else { cp->phys.header.savep = NCB_SCRIPT_PHYS (np, no_data); cp->phys.header.goalp = cp->phys.header.savep; } cp->phys.header.lastp = cp->phys.header.savep; /*---------------------------------------------------- ** ** fill in nccb ** **---------------------------------------------------- ** ** ** physical -> virtual backlink ** Generic SCSI command */ cp->phys.header.cp = cp; /* ** Startqueue */ cp->phys.header.launch.l_paddr = NCB_SCRIPT_PHYS (np, select); cp->phys.header.launch.l_cmd = SCR_JUMP; /* ** select */ cp->phys.select.sel_id = ccb->ccb_h.target_id; cp->phys.select.sel_scntl3 = tp->tinfo.wval; cp->phys.select.sel_sxfer = tp->tinfo.sval; /* ** message */ cp->phys.smsg.addr = CCB_PHYS (cp, scsi_smsg); cp->phys.smsg.size = msglen; cp->phys.smsg2.addr = CCB_PHYS (cp, scsi_smsg2); cp->phys.smsg2.size = msglen2; /* ** command */ cp->phys.cmd.addr = vtophys (scsiio_cdb_ptr(csio)); cp->phys.cmd.size = csio->cdb_len; /* ** sense command */ cp->phys.scmd.addr = CCB_PHYS (cp, sensecmd); cp->phys.scmd.size = 6; /* ** patch requested size into sense command */ cp->sensecmd[0] = 0x03; cp->sensecmd[1] = ccb->ccb_h.target_lun << 5; cp->sensecmd[4] = csio->sense_len; /* ** sense data */ cp->phys.sense.addr = vtophys (&csio->sense_data); cp->phys.sense.size = csio->sense_len; /* ** status */ cp->actualquirks = QUIRK_NOMSG; cp->host_status = nego ? HS_NEGOTIATE : HS_BUSY; cp->s_status = SCSI_STATUS_ILLEGAL; cp->parity_status = 0; cp->xerr_status = XE_OK; cp->sync_status = tp->tinfo.sval; cp->nego_status = nego; cp->wide_status = tp->tinfo.wval; /*---------------------------------------------------- ** ** Critical region: start this job. ** **---------------------------------------------------- */ /* ** reselect pattern and activate this job. */ cp->jump_nccb.l_cmd = (SCR_JUMP ^ IFFALSE (DATA (cp->tag))); cp->tlimit = time_second + ccb->ccb_h.timeout / 1000 + 2; cp->magic = CCB_MAGIC; /* ** insert into start queue. */ qidx = np->squeueput + 1; if (qidx >= MAX_START) qidx = 0; np->squeue [qidx ] = NCB_SCRIPT_PHYS (np, idle); np->squeue [np->squeueput] = CCB_PHYS (cp, phys); np->squeueput = qidx; if(DEBUG_FLAGS & DEBUG_QUEUE) device_printf(np->dev, "queuepos=%d tryoffset=%d.\n", np->squeueput, (unsigned)(READSCRIPT(startpos[0]) - (NCB_SCRIPTH_PHYS (np, tryloop)))); /* ** Script processor may be waiting for reselect. ** Wake it up. */ OUTB (nc_istat, SIGP); break; } case XPT_RESET_DEV: /* Bus Device Reset the specified SCSI device */ case XPT_EN_LUN: /* Enable LUN as a target */ case XPT_TARGET_IO: /* Execute target I/O request */ case XPT_ACCEPT_TARGET_IO: /* Accept Host Target Mode CDB */ case XPT_CONT_TARGET_IO: /* Continue Host Target I/O Connection*/ case XPT_ABORT: /* Abort the specified CCB */ /* XXX Implement */ ccb->ccb_h.status = CAM_REQ_INVALID; xpt_done(ccb); break; case XPT_SET_TRAN_SETTINGS: { struct ccb_trans_settings *cts = &ccb->cts; tcb_p tp; u_int update_type; struct ccb_trans_settings_scsi *scsi = &cts->proto_specific.scsi; struct ccb_trans_settings_spi *spi = &cts->xport_specific.spi; update_type = 0; if (cts->type == CTS_TYPE_CURRENT_SETTINGS) update_type |= NCR_TRANS_GOAL; if (cts->type == CTS_TYPE_USER_SETTINGS) update_type |= NCR_TRANS_USER; tp = &np->target[ccb->ccb_h.target_id]; /* Tag and disc enables */ if ((spi->valid & CTS_SPI_VALID_DISC) != 0) { if (update_type & NCR_TRANS_GOAL) { if ((spi->flags & CTS_SPI_FLAGS_DISC_ENB) != 0) tp->tinfo.disc_tag |= NCR_CUR_DISCENB; else tp->tinfo.disc_tag &= ~NCR_CUR_DISCENB; } if (update_type & NCR_TRANS_USER) { if ((spi->flags & CTS_SPI_FLAGS_DISC_ENB) != 0) tp->tinfo.disc_tag |= NCR_USR_DISCENB; else tp->tinfo.disc_tag &= ~NCR_USR_DISCENB; } } if ((scsi->valid & CTS_SCSI_VALID_TQ) != 0) { if (update_type & NCR_TRANS_GOAL) { if ((scsi->flags & CTS_SCSI_FLAGS_TAG_ENB) != 0) tp->tinfo.disc_tag |= NCR_CUR_TAGENB; else tp->tinfo.disc_tag &= ~NCR_CUR_TAGENB; } if (update_type & NCR_TRANS_USER) { if ((scsi->flags & CTS_SCSI_FLAGS_TAG_ENB) != 0) tp->tinfo.disc_tag |= NCR_USR_TAGENB; else tp->tinfo.disc_tag &= ~NCR_USR_TAGENB; } } /* Filter bus width and sync negotiation settings */ if ((spi->valid & CTS_SPI_VALID_BUS_WIDTH) != 0) { if (spi->bus_width > np->maxwide) spi->bus_width = np->maxwide; } if (((spi->valid & CTS_SPI_VALID_SYNC_RATE) != 0) || ((spi->valid & CTS_SPI_VALID_SYNC_OFFSET) != 0)) { if ((spi->valid & CTS_SPI_VALID_SYNC_RATE) != 0) { if (spi->sync_period != 0 && (spi->sync_period < np->minsync)) spi->sync_period = np->minsync; } if ((spi->valid & CTS_SPI_VALID_SYNC_OFFSET) != 0) { if (spi->sync_offset == 0) spi->sync_period = 0; if (spi->sync_offset > np->maxoffs) spi->sync_offset = np->maxoffs; } } if ((update_type & NCR_TRANS_USER) != 0) { if ((spi->valid & CTS_SPI_VALID_SYNC_RATE) != 0) tp->tinfo.user.period = spi->sync_period; if ((spi->valid & CTS_SPI_VALID_SYNC_OFFSET) != 0) tp->tinfo.user.offset = spi->sync_offset; if ((spi->valid & CTS_SPI_VALID_BUS_WIDTH) != 0) tp->tinfo.user.width = spi->bus_width; } if ((update_type & NCR_TRANS_GOAL) != 0) { if ((spi->valid & CTS_SPI_VALID_SYNC_RATE) != 0) tp->tinfo.goal.period = spi->sync_period; if ((spi->valid & CTS_SPI_VALID_SYNC_OFFSET) != 0) tp->tinfo.goal.offset = spi->sync_offset; if ((spi->valid & CTS_SPI_VALID_BUS_WIDTH) != 0) tp->tinfo.goal.width = spi->bus_width; } ccb->ccb_h.status = CAM_REQ_CMP; xpt_done(ccb); break; } case XPT_GET_TRAN_SETTINGS: /* Get default/user set transfer settings for the target */ { struct ccb_trans_settings *cts = &ccb->cts; struct ncr_transinfo *tinfo; tcb_p tp = &np->target[ccb->ccb_h.target_id]; struct ccb_trans_settings_scsi *scsi = &cts->proto_specific.scsi; struct ccb_trans_settings_spi *spi = &cts->xport_specific.spi; cts->protocol = PROTO_SCSI; cts->protocol_version = SCSI_REV_2; cts->transport = XPORT_SPI; cts->transport_version = 2; if (cts->type == CTS_TYPE_CURRENT_SETTINGS) { tinfo = &tp->tinfo.current; if (tp->tinfo.disc_tag & NCR_CUR_DISCENB) spi->flags |= CTS_SPI_FLAGS_DISC_ENB; else spi->flags &= ~CTS_SPI_FLAGS_DISC_ENB; if (tp->tinfo.disc_tag & NCR_CUR_TAGENB) scsi->flags |= CTS_SCSI_FLAGS_TAG_ENB; else scsi->flags &= ~CTS_SCSI_FLAGS_TAG_ENB; } else { tinfo = &tp->tinfo.user; if (tp->tinfo.disc_tag & NCR_USR_DISCENB) spi->flags |= CTS_SPI_FLAGS_DISC_ENB; else spi->flags &= ~CTS_SPI_FLAGS_DISC_ENB; if (tp->tinfo.disc_tag & NCR_USR_TAGENB) scsi->flags |= CTS_SCSI_FLAGS_TAG_ENB; else scsi->flags &= ~CTS_SCSI_FLAGS_TAG_ENB; } spi->sync_period = tinfo->period; spi->sync_offset = tinfo->offset; spi->bus_width = tinfo->width; spi->valid = CTS_SPI_VALID_SYNC_RATE | CTS_SPI_VALID_SYNC_OFFSET | CTS_SPI_VALID_BUS_WIDTH | CTS_SPI_VALID_DISC; scsi->valid = CTS_SCSI_VALID_TQ; ccb->ccb_h.status = CAM_REQ_CMP; xpt_done(ccb); break; } case XPT_CALC_GEOMETRY: { /* XXX JGibbs - I'm sure the NCR uses a different strategy, * but it should be able to deal with Adaptec * geometry too. */ cam_calc_geometry(&ccb->ccg, /*extended*/1); xpt_done(ccb); break; } case XPT_RESET_BUS: /* Reset the specified SCSI bus */ { OUTB (nc_scntl1, CRST); ccb->ccb_h.status = CAM_REQ_CMP; DELAY(10000); /* Wait until our interrupt handler sees it */ xpt_done(ccb); break; } case XPT_TERM_IO: /* Terminate the I/O process */ /* XXX Implement */ ccb->ccb_h.status = CAM_REQ_INVALID; xpt_done(ccb); break; case XPT_PATH_INQ: /* Path routing inquiry */ { struct ccb_pathinq *cpi = &ccb->cpi; cpi->version_num = 1; /* XXX??? */ cpi->hba_inquiry = PI_SDTR_ABLE|PI_TAG_ABLE; if ((np->features & FE_WIDE) != 0) cpi->hba_inquiry |= PI_WIDE_16; cpi->target_sprt = 0; cpi->hba_misc = 0; cpi->hba_eng_cnt = 0; cpi->max_target = (np->features & FE_WIDE) ? 15 : 7; cpi->max_lun = MAX_LUN - 1; cpi->initiator_id = np->myaddr; cpi->bus_id = cam_sim_bus(sim); cpi->base_transfer_speed = 3300; strlcpy(cpi->sim_vid, "FreeBSD", SIM_IDLEN); strlcpy(cpi->hba_vid, "Symbios", HBA_IDLEN); strlcpy(cpi->dev_name, cam_sim_name(sim), DEV_IDLEN); cpi->unit_number = cam_sim_unit(sim); - cpi->transport = XPORT_SPI; - cpi->transport_version = 2; - cpi->protocol = PROTO_SCSI; - cpi->protocol_version = SCSI_REV_2; + cpi->transport = XPORT_SPI; + cpi->transport_version = 2; + cpi->protocol = PROTO_SCSI; + cpi->protocol_version = SCSI_REV_2; cpi->ccb_h.status = CAM_REQ_CMP; xpt_done(ccb); break; } default: ccb->ccb_h.status = CAM_REQ_INVALID; xpt_done(ccb); break; } } /*========================================================== ** ** ** Complete execution of a SCSI command. ** Signal completion to the generic SCSI driver. ** ** **========================================================== */ static void ncr_complete (ncb_p np, nccb_p cp) { union ccb *ccb; tcb_p tp; /* ** Sanity check */ if (!cp || (cp->magic!=CCB_MAGIC) || !cp->ccb) return; cp->magic = 1; cp->tlimit= 0; /* ** No Reselect anymore. */ cp->jump_nccb.l_cmd = (SCR_JUMP); /* ** No starting. */ cp->phys.header.launch.l_paddr= NCB_SCRIPT_PHYS (np, idle); /* ** timestamp */ ncb_profile (np, cp); if (DEBUG_FLAGS & DEBUG_TINY) printf ("CCB=%x STAT=%x/%x\n", (int)(intptr_t)cp & 0xfff, cp->host_status,cp->s_status); ccb = cp->ccb; cp->ccb = NULL; tp = &np->target[ccb->ccb_h.target_id]; /* ** We do not queue more than 1 nccb per target ** with negotiation at any time. If this nccb was ** used for negotiation, clear this info in the tcb. */ if (cp == tp->nego_cp) tp->nego_cp = NULL; /* ** Check for parity errors. */ /* XXX JGibbs - What about reporting them??? */ if (cp->parity_status) { PRINT_ADDR(ccb); printf ("%d parity error(s), fallback.\n", cp->parity_status); /* ** fallback to asynch transfer. */ tp->tinfo.goal.period = 0; tp->tinfo.goal.offset = 0; } /* ** Check for extended errors. */ if (cp->xerr_status != XE_OK) { PRINT_ADDR(ccb); switch (cp->xerr_status) { case XE_EXTRA_DATA: printf ("extraneous data discarded.\n"); break; case XE_BAD_PHASE: printf ("illegal scsi phase (4/5).\n"); break; default: printf ("extended error %d.\n", cp->xerr_status); break; } if (cp->host_status==HS_COMPLETE) cp->host_status = HS_FAIL; } /* ** Check the status. */ if (cp->host_status == HS_COMPLETE) { if (cp->s_status == SCSI_STATUS_OK) { /* ** All went well. */ /* XXX JGibbs - Properly calculate residual */ tp->bytes += ccb->csio.dxfer_len; tp->transfers ++; ccb->ccb_h.status = CAM_REQ_CMP; } else if ((cp->s_status & SCSI_STATUS_SENSE) != 0) { /* * XXX Could be TERMIO too. Should record * original status. */ ccb->csio.scsi_status = SCSI_STATUS_CHECK_COND; cp->s_status &= ~SCSI_STATUS_SENSE; if (cp->s_status == SCSI_STATUS_OK) { ccb->ccb_h.status = CAM_AUTOSNS_VALID|CAM_SCSI_STATUS_ERROR; } else { ccb->ccb_h.status = CAM_AUTOSENSE_FAIL; } } else { ccb->ccb_h.status = CAM_SCSI_STATUS_ERROR; ccb->csio.scsi_status = cp->s_status; } } else if (cp->host_status == HS_SEL_TIMEOUT) { /* ** Device failed selection */ ccb->ccb_h.status = CAM_SEL_TIMEOUT; } else if (cp->host_status == HS_TIMEOUT) { /* ** No response */ ccb->ccb_h.status = CAM_CMD_TIMEOUT; } else if (cp->host_status == HS_STALL) { ccb->ccb_h.status = CAM_REQUEUE_REQ; } else { /* ** Other protocol messes */ PRINT_ADDR(ccb); printf ("COMMAND FAILED (%x %x) @%p.\n", cp->host_status, cp->s_status, cp); ccb->ccb_h.status = CAM_CMD_TIMEOUT; } if ((ccb->ccb_h.status & CAM_STATUS_MASK) != CAM_REQ_CMP) { xpt_freeze_devq(ccb->ccb_h.path, /*count*/1); ccb->ccb_h.status |= CAM_DEV_QFRZN; } /* ** Free this nccb */ ncr_free_nccb (np, cp); /* ** signal completion to generic driver. */ xpt_done (ccb); } /*========================================================== ** ** ** Signal all (or one) control block done. ** ** **========================================================== */ static void ncr_wakeup (ncb_p np, u_long code) { /* ** Starting at the default nccb and following ** the links, complete all jobs with a ** host_status greater than "disconnect". ** ** If the "code" parameter is not zero, ** complete all jobs that are not IDLE. */ nccb_p cp = np->link_nccb; while (cp) { switch (cp->host_status) { case HS_IDLE: break; case HS_DISCONNECT: if(DEBUG_FLAGS & DEBUG_TINY) printf ("D"); /* FALLTHROUGH */ case HS_BUSY: case HS_NEGOTIATE: if (!code) break; cp->host_status = code; /* FALLTHROUGH */ default: ncr_complete (np, cp); break; } cp = cp -> link_nccb; } } static void ncr_freeze_devq (ncb_p np, struct cam_path *path) { nccb_p cp; int i; int count; int firstskip; /* ** Starting at the first nccb and following ** the links, complete all jobs that match ** the passed in path and are in the start queue. */ cp = np->link_nccb; count = 0; firstskip = 0; while (cp) { switch (cp->host_status) { case HS_BUSY: case HS_NEGOTIATE: if ((cp->phys.header.launch.l_paddr == NCB_SCRIPT_PHYS (np, select)) && (xpt_path_comp(path, cp->ccb->ccb_h.path) >= 0)) { /* Mark for removal from the start queue */ for (i = 1; i < MAX_START; i++) { int idx; idx = np->squeueput - i; if (idx < 0) idx = MAX_START + idx; if (np->squeue[idx] == CCB_PHYS(cp, phys)) { np->squeue[idx] = NCB_SCRIPT_PHYS (np, skip); if (i > firstskip) firstskip = i; break; } } cp->host_status=HS_STALL; ncr_complete (np, cp); count++; } break; default: break; } cp = cp->link_nccb; } if (count > 0) { int j; int bidx; /* Compress the start queue */ j = 0; bidx = np->squeueput; i = np->squeueput - firstskip; if (i < 0) i = MAX_START + i; for (;;) { bidx = i - j; if (bidx < 0) bidx = MAX_START + bidx; if (np->squeue[i] == NCB_SCRIPT_PHYS (np, skip)) { j++; } else if (j != 0) { np->squeue[bidx] = np->squeue[i]; if (np->squeue[bidx] == NCB_SCRIPT_PHYS(np, idle)) break; } i = (i + 1) % MAX_START; } np->squeueput = bidx; } } /*========================================================== ** ** ** Start NCR chip. ** ** **========================================================== */ static void ncr_init(ncb_p np, char * msg, u_long code) { int i; /* ** Reset chip. */ OUTB (nc_istat, SRST); DELAY (1000); OUTB (nc_istat, 0); /* ** Message. */ if (msg) device_printf(np->dev, "restart (%s).\n", msg); /* ** Clear Start Queue */ for (i=0;i squeue [i] = NCB_SCRIPT_PHYS (np, idle); /* ** Start at first entry. */ np->squeueput = 0; WRITESCRIPT(startpos[0], NCB_SCRIPTH_PHYS (np, tryloop)); WRITESCRIPT(start0 [0], SCR_INT ^ IFFALSE (0)); /* ** Wakeup all pending jobs. */ ncr_wakeup (np, code); /* ** Init chip. */ OUTB (nc_istat, 0x00 ); /* Remove Reset, abort ... */ OUTB (nc_scntl0, 0xca ); /* full arb., ena parity, par->ATN */ OUTB (nc_scntl1, 0x00 ); /* odd parity, and remove CRST!! */ ncr_selectclock(np, np->rv_scntl3); /* Select SCSI clock */ OUTB (nc_scid , RRE|np->myaddr);/* host adapter SCSI address */ OUTW (nc_respid, 1ul<myaddr);/* id to respond to */ OUTB (nc_istat , SIGP ); /* Signal Process */ OUTB (nc_dmode , np->rv_dmode); /* XXX modify burstlen ??? */ OUTB (nc_dcntl , np->rv_dcntl); OUTB (nc_ctest3, np->rv_ctest3); OUTB (nc_ctest5, np->rv_ctest5); OUTB (nc_ctest4, np->rv_ctest4);/* enable master parity checking */ OUTB (nc_stest2, np->rv_stest2|EXT); /* Extended Sreq/Sack filtering */ OUTB (nc_stest3, TE ); /* TolerANT enable */ OUTB (nc_stime0, 0x0b ); /* HTH = disabled, STO = 0.1 sec. */ if (bootverbose >= 2) { printf ("\tACTUAL values:SCNTL3:%02x DMODE:%02x DCNTL:%02x\n", np->rv_scntl3, np->rv_dmode, np->rv_dcntl); printf ("\t CTEST3:%02x CTEST4:%02x CTEST5:%02x\n", np->rv_ctest3, np->rv_ctest4, np->rv_ctest5); } /* ** Enable GPIO0 pin for writing if LED support. */ if (np->features & FE_LED0) { OUTOFFB (nc_gpcntl, 0x01); } /* ** Fill in target structure. */ for (i=0;itarget[i]; tp->tinfo.sval = 0; tp->tinfo.wval = np->rv_scntl3; tp->tinfo.current.period = 0; tp->tinfo.current.offset = 0; tp->tinfo.current.width = MSG_EXT_WDTR_BUS_8_BIT; } /* ** enable ints */ OUTW (nc_sien , STO|HTH|MA|SGE|UDC|RST); OUTB (nc_dien , MDPE|BF|ABRT|SSI|SIR|IID); /* ** Start script processor. */ OUTL (nc_dsp, NCB_SCRIPT_PHYS (np, start)); /* * Notify the XPT of the event */ if (code == HS_RESET) xpt_async(AC_BUS_RESET, np->path, NULL); } static void ncr_poll(struct cam_sim *sim) { ncr_intr_locked(cam_sim_softc(sim)); } /*========================================================== ** ** Get clock factor and sync divisor for a given ** synchronous factor period. ** Returns the clock factor (in sxfer) and scntl3 ** synchronous divisor field. ** **========================================================== */ static void ncr_getsync(ncb_p np, u_char sfac, u_char *fakp, u_char *scntl3p) { u_long clk = np->clock_khz; /* SCSI clock frequency in kHz */ int div = np->clock_divn; /* Number of divisors supported */ u_long fak; /* Sync factor in sxfer */ u_long per; /* Period in tenths of ns */ u_long kpc; /* (per * clk) */ /* ** Compute the synchronous period in tenths of nano-seconds */ if (sfac <= 10) per = 250; else if (sfac == 11) per = 303; else if (sfac == 12) per = 500; else per = 40 * sfac; /* ** Look for the greatest clock divisor that allows an ** input speed faster than the period. */ kpc = per * clk; while (--div >= 0) if (kpc >= (div_10M[div] * 4)) break; /* ** Calculate the lowest clock factor that allows an output ** speed not faster than the period. */ fak = (kpc - 1) / div_10M[div] + 1; #if 0 /* You can #if 1 if you think this optimization is useful */ per = (fak * div_10M[div]) / clk; /* ** Why not to try the immediate lower divisor and to choose ** the one that allows the fastest output speed ? ** We dont want input speed too much greater than output speed. */ if (div >= 1 && fak < 6) { u_long fak2, per2; fak2 = (kpc - 1) / div_10M[div-1] + 1; per2 = (fak2 * div_10M[div-1]) / clk; if (per2 < per && fak2 <= 6) { fak = fak2; per = per2; --div; } } #endif if (fak < 4) fak = 4; /* Should never happen, too bad ... */ /* ** Compute and return sync parameters for the ncr */ *fakp = fak - 4; *scntl3p = ((div+1) << 4) + (sfac < 25 ? 0x80 : 0); } /*========================================================== ** ** Switch sync mode for current job and its target ** **========================================================== */ static void ncr_setsync(ncb_p np, nccb_p cp, u_char scntl3, u_char sxfer, u_char period) { union ccb *ccb; struct ccb_trans_settings neg; tcb_p tp; int div; u_int target = INB (nc_sdid) & 0x0f; u_int period_10ns; assert (cp); if (!cp) return; ccb = cp->ccb; assert (ccb); if (!ccb) return; assert (target == ccb->ccb_h.target_id); tp = &np->target[target]; if (!scntl3 || !(sxfer & 0x1f)) scntl3 = np->rv_scntl3; scntl3 = (scntl3 & 0xf0) | (tp->tinfo.wval & EWS) | (np->rv_scntl3 & 0x07); /* ** Deduce the value of controller sync period from scntl3. ** period is in tenths of nano-seconds. */ div = ((scntl3 >> 4) & 0x7); if ((sxfer & 0x1f) && div) period_10ns = (((sxfer>>5)+4)*div_10M[div-1])/np->clock_khz; else period_10ns = 0; tp->tinfo.goal.period = period; tp->tinfo.goal.offset = sxfer & 0x1f; tp->tinfo.current.period = period; tp->tinfo.current.offset = sxfer & 0x1f; /* ** Stop there if sync parameters are unchanged */ if (tp->tinfo.sval == sxfer && tp->tinfo.wval == scntl3) return; tp->tinfo.sval = sxfer; tp->tinfo.wval = scntl3; if (sxfer & 0x1f) { /* ** Disable extended Sreq/Sack filtering */ if (period_10ns <= 2000) OUTOFFB (nc_stest2, EXT); } /* ** Tell the SCSI layer about the ** new transfer parameters. */ memset(&neg, 0, sizeof (neg)); neg.protocol = PROTO_SCSI; neg.protocol_version = SCSI_REV_2; neg.transport = XPORT_SPI; neg.transport_version = 2; neg.xport_specific.spi.sync_period = period; neg.xport_specific.spi.sync_offset = sxfer & 0x1f; neg.xport_specific.spi.valid = CTS_SPI_VALID_SYNC_RATE | CTS_SPI_VALID_SYNC_OFFSET; xpt_setup_ccb(&neg.ccb_h, ccb->ccb_h.path, /*priority*/1); xpt_async(AC_TRANSFER_NEG, ccb->ccb_h.path, &neg); /* ** set actual value and sync_status */ OUTB (nc_sxfer, sxfer); np->sync_st = sxfer; OUTB (nc_scntl3, scntl3); np->wide_st = scntl3; /* ** patch ALL nccbs of this target. */ for (cp = np->link_nccb; cp; cp = cp->link_nccb) { if (!cp->ccb) continue; if (cp->ccb->ccb_h.target_id != target) continue; cp->sync_status = sxfer; cp->wide_status = scntl3; } } /*========================================================== ** ** Switch wide mode for current job and its target ** SCSI specs say: a SCSI device that accepts a WDTR ** message shall reset the synchronous agreement to ** asynchronous mode. ** **========================================================== */ static void ncr_setwide (ncb_p np, nccb_p cp, u_char wide, u_char ack) { union ccb *ccb; struct ccb_trans_settings neg; u_int target = INB (nc_sdid) & 0x0f; tcb_p tp; u_char scntl3; u_char sxfer; assert (cp); if (!cp) return; ccb = cp->ccb; assert (ccb); if (!ccb) return; assert (target == ccb->ccb_h.target_id); tp = &np->target[target]; tp->tinfo.current.width = wide; tp->tinfo.goal.width = wide; tp->tinfo.current.period = 0; tp->tinfo.current.offset = 0; scntl3 = (tp->tinfo.wval & (~EWS)) | (wide ? EWS : 0); sxfer = ack ? 0 : tp->tinfo.sval; /* ** Stop there if sync/wide parameters are unchanged */ if (tp->tinfo.sval == sxfer && tp->tinfo.wval == scntl3) return; tp->tinfo.sval = sxfer; tp->tinfo.wval = scntl3; /* Tell the SCSI layer about the new transfer params */ memset(&neg, 0, sizeof (neg)); neg.protocol = PROTO_SCSI; neg.protocol_version = SCSI_REV_2; neg.transport = XPORT_SPI; neg.transport_version = 2; neg.xport_specific.spi.bus_width = (scntl3 & EWS) ? MSG_EXT_WDTR_BUS_16_BIT : MSG_EXT_WDTR_BUS_8_BIT; neg.xport_specific.spi.sync_period = 0; neg.xport_specific.spi.sync_offset = 0; neg.xport_specific.spi.valid = CTS_SPI_VALID_SYNC_RATE | CTS_SPI_VALID_SYNC_OFFSET | CTS_SPI_VALID_BUS_WIDTH; xpt_setup_ccb(&neg.ccb_h, ccb->ccb_h.path, /*priority*/1); xpt_async(AC_TRANSFER_NEG, ccb->ccb_h.path, &neg); /* ** set actual value and sync_status */ OUTB (nc_sxfer, sxfer); np->sync_st = sxfer; OUTB (nc_scntl3, scntl3); np->wide_st = scntl3; /* ** patch ALL nccbs of this target. */ for (cp = np->link_nccb; cp; cp = cp->link_nccb) { if (!cp->ccb) continue; if (cp->ccb->ccb_h.target_id != target) continue; cp->sync_status = sxfer; cp->wide_status = scntl3; } } /*========================================================== ** ** ** ncr timeout handler. ** ** **========================================================== ** ** Misused to keep the driver running when ** interrupts are not configured correctly. ** **---------------------------------------------------------- */ static void ncr_timeout (void *arg) { ncb_p np = arg; time_t thistime = time_second; ticks_t step = np->ticks; u_long count = 0; long signed t; nccb_p cp; mtx_assert(&np->lock, MA_OWNED); if (np->lasttime != thistime) { np->lasttime = thistime; /*---------------------------------------------------- ** ** handle ncr chip timeouts ** ** Assumption: ** We have a chance to arbitrate for the ** SCSI bus at least every 10 seconds. ** **---------------------------------------------------- */ t = thistime - np->heartbeat; if (t<2) np->latetime=0; else np->latetime++; if (np->latetime>2) { /* ** If there are no requests, the script ** processor will sleep on SEL_WAIT_RESEL. ** But we have to check whether it died. ** Let's try to wake it up. */ OUTB (nc_istat, SIGP); } /*---------------------------------------------------- ** ** handle nccb timeouts ** **---------------------------------------------------- */ for (cp=np->link_nccb; cp; cp=cp->link_nccb) { /* ** look for timed out nccbs. */ if (!cp->host_status) continue; count++; if (cp->tlimit > thistime) continue; /* ** Disable reselect. ** Remove it from startqueue. */ cp->jump_nccb.l_cmd = (SCR_JUMP); if (cp->phys.header.launch.l_paddr == NCB_SCRIPT_PHYS (np, select)) { device_printf(np->dev, "timeout nccb=%p (skip)\n", cp); cp->phys.header.launch.l_paddr = NCB_SCRIPT_PHYS (np, skip); } switch (cp->host_status) { case HS_BUSY: case HS_NEGOTIATE: /* FALLTHROUGH */ case HS_DISCONNECT: cp->host_status=HS_TIMEOUT; } cp->tag = 0; /* ** wakeup this nccb. */ ncr_complete (np, cp); } } callout_reset(&np->timer, step ? step : 1, ncr_timeout, np); if (INB(nc_istat) & (INTF|SIP|DIP)) { /* ** Process pending interrupts. */ if (DEBUG_FLAGS & DEBUG_TINY) printf ("{"); ncr_exception (np); if (DEBUG_FLAGS & DEBUG_TINY) printf ("}"); } } /*========================================================== ** ** log message for real hard errors ** ** "ncr0 targ 0?: ERROR (ds:si) (so-si-sd) (sxfer/scntl3) @ name (dsp:dbc)." ** " reg: r0 r1 r2 r3 r4 r5 r6 ..... rf." ** ** exception register: ** ds: dstat ** si: sist ** ** SCSI bus lines: ** so: control lines as driver by NCR. ** si: control lines as seen by NCR. ** sd: scsi data lines as seen by NCR. ** ** wide/fastmode: ** sxfer: (see the manual) ** scntl3: (see the manual) ** ** current script command: ** dsp: script address (relative to start of script). ** dbc: first word of script command. ** ** First 16 register of the chip: ** r0..rf ** **========================================================== */ static void ncr_log_hard_error(ncb_p np, u_short sist, u_char dstat) { u_int32_t dsp; int script_ofs; int script_size; char *script_name; u_char *script_base; int i; dsp = INL (nc_dsp); if (np->p_script < dsp && dsp <= np->p_script + sizeof(struct script)) { script_ofs = dsp - np->p_script; script_size = sizeof(struct script); script_base = (u_char *) np->script; script_name = "script"; } else if (np->p_scripth < dsp && dsp <= np->p_scripth + sizeof(struct scripth)) { script_ofs = dsp - np->p_scripth; script_size = sizeof(struct scripth); script_base = (u_char *) np->scripth; script_name = "scripth"; } else { script_ofs = dsp; script_size = 0; script_base = 0; script_name = "mem"; } device_printf(np->dev, "%d: ERROR (%x:%x) (%x-%x-%x) (%x/%x) @ (%s %x:%08x).\n", (unsigned)INB (nc_sdid)&0x0f, dstat, sist, (unsigned)INB (nc_socl), (unsigned)INB (nc_sbcl), (unsigned)INB (nc_sbdl), (unsigned)INB (nc_sxfer),(unsigned)INB (nc_scntl3), script_name, script_ofs, (unsigned)INL (nc_dbc)); if (((script_ofs & 3) == 0) && (unsigned)script_ofs < script_size) { device_printf(np->dev, "script cmd = %08x\n", (int)READSCRIPT_OFF(script_base, script_ofs)); } device_printf(np->dev, "regdump:"); for (i=0; i<16;i++) printf (" %02x", (unsigned)INB_OFF(i)); printf (".\n"); } /*========================================================== ** ** ** ncr chip exception handler. ** ** **========================================================== */ static void ncr_exception (ncb_p np) { u_char istat, dstat; u_short sist; /* ** interrupt on the fly ? */ while ((istat = INB (nc_istat)) & INTF) { if (DEBUG_FLAGS & DEBUG_TINY) printf ("F "); OUTB (nc_istat, INTF); np->profile.num_fly++; ncr_wakeup (np, 0); } if (!(istat & (SIP|DIP))) { return; } /* ** Steinbach's Guideline for Systems Programming: ** Never test for an error condition you don't know how to handle. */ sist = (istat & SIP) ? INW (nc_sist) : 0; dstat = (istat & DIP) ? INB (nc_dstat) : 0; np->profile.num_int++; if (DEBUG_FLAGS & DEBUG_TINY) printf ("<%d|%x:%x|%x:%x>", INB(nc_scr0), dstat,sist, (unsigned)INL(nc_dsp), (unsigned)INL(nc_dbc)); if ((dstat==DFE) && (sist==PAR)) return; /*========================================================== ** ** First the normal cases. ** **========================================================== */ /*------------------------------------------- ** SCSI reset **------------------------------------------- */ if (sist & RST) { ncr_init (np, bootverbose ? "scsi reset" : NULL, HS_RESET); return; } /*------------------------------------------- ** selection timeout ** ** IID excluded from dstat mask! ** (chip bug) **------------------------------------------- */ if ((sist & STO) && !(sist & (GEN|HTH|MA|SGE|UDC|RST|PAR)) && !(dstat & (MDPE|BF|ABRT|SIR))) { ncr_int_sto (np); return; } /*------------------------------------------- ** Phase mismatch. **------------------------------------------- */ if ((sist & MA) && !(sist & (STO|GEN|HTH|SGE|UDC|RST|PAR)) && !(dstat & (MDPE|BF|ABRT|SIR|IID))) { ncr_int_ma (np, dstat); return; } /*---------------------------------------- ** move command with length 0 **---------------------------------------- */ if ((dstat & IID) && !(sist & (STO|GEN|HTH|MA|SGE|UDC|RST|PAR)) && !(dstat & (MDPE|BF|ABRT|SIR)) && ((INL(nc_dbc) & 0xf8000000) == SCR_MOVE_TBL)) { /* ** Target wants more data than available. ** The "no_data" script will do it. */ OUTL (nc_dsp, NCB_SCRIPT_PHYS (np, no_data)); return; } /*------------------------------------------- ** Programmed interrupt **------------------------------------------- */ if ((dstat & SIR) && !(sist & (STO|GEN|HTH|MA|SGE|UDC|RST|PAR)) && !(dstat & (MDPE|BF|ABRT|IID)) && (INB(nc_dsps) <= SIR_MAX)) { ncr_int_sir (np); return; } /*======================================== ** log message for real hard errors **======================================== */ ncr_log_hard_error(np, sist, dstat); /*======================================== ** do the register dump **======================================== */ if (time_second - np->regtime > 10) { int i; np->regtime = time_second; for (i=0; iregdump); i++) ((volatile char*)&np->regdump)[i] = INB_OFF(i); np->regdump.nc_dstat = dstat; np->regdump.nc_sist = sist; } /*---------------------------------------- ** clean up the dma fifo **---------------------------------------- */ if ( (INB(nc_sstat0) & (ILF|ORF|OLF) ) || (INB(nc_sstat1) & (FF3210) ) || (INB(nc_sstat2) & (ILF1|ORF1|OLF1)) || /* wide .. */ !(dstat & DFE)) { device_printf(np->dev, "have to clear fifos.\n"); OUTB (nc_stest3, TE|CSF); /* clear scsi fifo */ OUTB (nc_ctest3, np->rv_ctest3 | CLF); /* clear dma fifo */ } /*---------------------------------------- ** handshake timeout **---------------------------------------- */ if (sist & HTH) { device_printf(np->dev, "handshake timeout\n"); OUTB (nc_scntl1, CRST); DELAY (1000); OUTB (nc_scntl1, 0x00); OUTB (nc_scr0, HS_FAIL); OUTL (nc_dsp, NCB_SCRIPT_PHYS (np, cleanup)); return; } /*---------------------------------------- ** unexpected disconnect **---------------------------------------- */ if ((sist & UDC) && !(sist & (STO|GEN|HTH|MA|SGE|RST|PAR)) && !(dstat & (MDPE|BF|ABRT|SIR|IID))) { OUTB (nc_scr0, HS_UNEXPECTED); OUTL (nc_dsp, NCB_SCRIPT_PHYS (np, cleanup)); return; } /*---------------------------------------- ** cannot disconnect **---------------------------------------- */ if ((dstat & IID) && !(sist & (STO|GEN|HTH|MA|SGE|UDC|RST|PAR)) && !(dstat & (MDPE|BF|ABRT|SIR)) && ((INL(nc_dbc) & 0xf8000000) == SCR_WAIT_DISC)) { /* ** Unexpected data cycle while waiting for disconnect. */ if (INB(nc_sstat2) & LDSC) { /* ** It's an early reconnect. ** Let's continue ... */ OUTB (nc_dcntl, np->rv_dcntl | STD); /* ** info message */ device_printf(np->dev, "INFO: LDSC while IID.\n"); return; } device_printf(np->dev, "target %d doesn't release the bus.\n", INB (nc_sdid)&0x0f); /* ** return without restarting the NCR. ** timeout will do the real work. */ return; } /*---------------------------------------- ** single step **---------------------------------------- */ if ((dstat & SSI) && !(sist & (STO|GEN|HTH|MA|SGE|UDC|RST|PAR)) && !(dstat & (MDPE|BF|ABRT|SIR|IID))) { OUTB (nc_dcntl, np->rv_dcntl | STD); return; } /* ** @RECOVER@ HTH, SGE, ABRT. ** ** We should try to recover from these interrupts. ** They may occur if there are problems with synch transfers, or ** if targets are switched on or off while the driver is running. */ if (sist & SGE) { /* clear scsi offsets */ OUTB (nc_ctest3, np->rv_ctest3 | CLF); } /* ** Freeze controller to be able to read the messages. */ if (DEBUG_FLAGS & DEBUG_FREEZE) { int i; unsigned char val; for (i=0; i<0x60; i++) { switch (i%16) { case 0: device_printf(np->dev, "reg[%d0]: ", i / 16); break; case 4: case 8: case 12: printf (" "); break; } val = bus_read_1(np->reg_res, i); printf (" %x%x", val/16, val%16); if (i%16==15) printf (".\n"); } callout_stop(&np->timer); device_printf(np->dev, "halted!\n"); /* ** don't restart controller ... */ OUTB (nc_istat, SRST); return; } #ifdef NCR_FREEZE /* ** Freeze system to be able to read the messages. */ printf ("ncr: fatal error: system halted - press reset to reboot ..."); for (;;); #endif /* ** sorry, have to kill ALL jobs ... */ ncr_init (np, "fatal error", HS_FAIL); } /*========================================================== ** ** ncr chip exception handler for selection timeout ** **========================================================== ** ** There seems to be a bug in the 53c810. ** Although a STO-Interrupt is pending, ** it continues executing script commands. ** But it will fail and interrupt (IID) on ** the next instruction where it's looking ** for a valid phase. ** **---------------------------------------------------------- */ static void ncr_int_sto (ncb_p np) { u_long dsa, scratcha, diff; nccb_p cp; if (DEBUG_FLAGS & DEBUG_TINY) printf ("T"); /* ** look for nccb and set the status. */ dsa = INL (nc_dsa); cp = np->link_nccb; while (cp && (CCB_PHYS (cp, phys) != dsa)) cp = cp->link_nccb; if (cp) { cp-> host_status = HS_SEL_TIMEOUT; ncr_complete (np, cp); } /* ** repair start queue */ scratcha = INL (nc_scratcha); diff = scratcha - NCB_SCRIPTH_PHYS (np, tryloop); /* assert ((diff <= MAX_START * 20) && !(diff % 20));*/ if ((diff <= MAX_START * 20) && !(diff % 20)) { WRITESCRIPT(startpos[0], scratcha); OUTL (nc_dsp, NCB_SCRIPT_PHYS (np, start)); return; } ncr_init (np, "selection timeout", HS_FAIL); } /*========================================================== ** ** ** ncr chip exception handler for phase errors. ** ** **========================================================== ** ** We have to construct a new transfer descriptor, ** to transfer the rest of the current block. ** **---------------------------------------------------------- */ static void ncr_int_ma (ncb_p np, u_char dstat) { u_int32_t dbc; u_int32_t rest; u_int32_t dsa; u_int32_t dsp; u_int32_t nxtdsp; volatile void *vdsp_base; size_t vdsp_off; u_int32_t oadr, olen; u_int32_t *tblp, *newcmd; u_char cmd, sbcl, ss0, ss2, ctest5; u_short delta; nccb_p cp; dsp = INL (nc_dsp); dsa = INL (nc_dsa); dbc = INL (nc_dbc); ss0 = INB (nc_sstat0); ss2 = INB (nc_sstat2); sbcl= INB (nc_sbcl); cmd = dbc >> 24; rest= dbc & 0xffffff; ctest5 = (np->rv_ctest5 & DFS) ? INB (nc_ctest5) : 0; if (ctest5 & DFS) delta=(((ctest5<<8) | (INB (nc_dfifo) & 0xff)) - rest) & 0x3ff; else delta=(INB (nc_dfifo) - rest) & 0x7f; /* ** The data in the dma fifo has not been transferred to ** the target -> add the amount to the rest ** and clear the data. ** Check the sstat2 register in case of wide transfer. */ if (!(dstat & DFE)) rest += delta; if (ss0 & OLF) rest++; if (ss0 & ORF) rest++; if (INB(nc_scntl3) & EWS) { if (ss2 & OLF1) rest++; if (ss2 & ORF1) rest++; } OUTB (nc_ctest3, np->rv_ctest3 | CLF); /* clear dma fifo */ OUTB (nc_stest3, TE|CSF); /* clear scsi fifo */ /* ** locate matching cp */ cp = np->link_nccb; while (cp && (CCB_PHYS (cp, phys) != dsa)) cp = cp->link_nccb; if (!cp) { device_printf(np->dev, "SCSI phase error fixup: CCB already dequeued (%p)\n", (void *)np->header.cp); return; } if (cp != np->header.cp) { device_printf(np->dev, "SCSI phase error fixup: CCB address mismatch " "(%p != %p) np->nccb = %p\n", (void *)cp, (void *)np->header.cp, (void *)np->link_nccb); /* return;*/ } /* ** find the interrupted script command, ** and the address at which to continue. */ if (dsp == vtophys (&cp->patch[2])) { vdsp_base = cp; vdsp_off = offsetof(struct nccb, patch[0]); nxtdsp = READSCRIPT_OFF(vdsp_base, vdsp_off + 3*4); } else if (dsp == vtophys (&cp->patch[6])) { vdsp_base = cp; vdsp_off = offsetof(struct nccb, patch[4]); nxtdsp = READSCRIPT_OFF(vdsp_base, vdsp_off + 3*4); } else if (dsp > np->p_script && dsp <= np->p_script + sizeof(struct script)) { vdsp_base = np->script; vdsp_off = dsp - np->p_script - 8; nxtdsp = dsp; } else { vdsp_base = np->scripth; vdsp_off = dsp - np->p_scripth - 8; nxtdsp = dsp; } /* ** log the information */ if (DEBUG_FLAGS & (DEBUG_TINY|DEBUG_PHASE)) { printf ("P%x%x ",cmd&7, sbcl&7); printf ("RL=%d D=%d SS0=%x ", (unsigned) rest, (unsigned) delta, ss0); } if (DEBUG_FLAGS & DEBUG_PHASE) { printf ("\nCP=%p CP2=%p DSP=%x NXT=%x VDSP=%p CMD=%x ", cp, np->header.cp, dsp, nxtdsp, (volatile char*)vdsp_base+vdsp_off, cmd); } /* ** get old startaddress and old length. */ oadr = READSCRIPT_OFF(vdsp_base, vdsp_off + 1*4); if (cmd & 0x10) { /* Table indirect */ tblp = (u_int32_t *) ((char*) &cp->phys + oadr); olen = tblp[0]; oadr = tblp[1]; } else { tblp = (u_int32_t *) 0; olen = READSCRIPT_OFF(vdsp_base, vdsp_off) & 0xffffff; } if (DEBUG_FLAGS & DEBUG_PHASE) { printf ("OCMD=%x\nTBLP=%p OLEN=%lx OADR=%lx\n", (unsigned) (READSCRIPT_OFF(vdsp_base, vdsp_off) >> 24), (void *) tblp, (u_long) olen, (u_long) oadr); } /* ** if old phase not dataphase, leave here. */ if (cmd != (READSCRIPT_OFF(vdsp_base, vdsp_off) >> 24)) { PRINT_ADDR(cp->ccb); printf ("internal error: cmd=%02x != %02x=(vdsp[0] >> 24)\n", (unsigned)cmd, (unsigned)READSCRIPT_OFF(vdsp_base, vdsp_off) >> 24); return; } if (cmd & 0x06) { PRINT_ADDR(cp->ccb); printf ("phase change %x-%x %d@%08x resid=%d.\n", cmd&7, sbcl&7, (unsigned)olen, (unsigned)oadr, (unsigned)rest); OUTB (nc_dcntl, np->rv_dcntl | STD); return; } /* ** choose the correct patch area. ** if savep points to one, choose the other. */ newcmd = cp->patch; if (cp->phys.header.savep == vtophys (newcmd)) newcmd+=4; /* ** fillin the commands */ newcmd[0] = ((cmd & 0x0f) << 24) | rest; newcmd[1] = oadr + olen - rest; newcmd[2] = SCR_JUMP; newcmd[3] = nxtdsp; if (DEBUG_FLAGS & DEBUG_PHASE) { PRINT_ADDR(cp->ccb); printf ("newcmd[%d] %x %x %x %x.\n", (int)(newcmd - cp->patch), (unsigned)newcmd[0], (unsigned)newcmd[1], (unsigned)newcmd[2], (unsigned)newcmd[3]); } /* ** fake the return address (to the patch). ** and restart script processor at dispatcher. */ np->profile.num_break++; OUTL (nc_temp, vtophys (newcmd)); if ((cmd & 7) == 0) OUTL (nc_dsp, NCB_SCRIPT_PHYS (np, dispatch)); else OUTL (nc_dsp, NCB_SCRIPT_PHYS (np, checkatn)); } /*========================================================== ** ** ** ncr chip exception handler for programmed interrupts. ** ** **========================================================== */ static int ncr_show_msg (u_char * msg) { u_char i; printf ("%x",*msg); if (*msg==MSG_EXTENDED) { for (i=1;i<8;i++) { if (i-1>msg[1]) break; printf ("-%x",msg[i]); } return (i+1); } else if ((*msg & 0xf0) == 0x20) { printf ("-%x",msg[1]); return (2); } return (1); } static void ncr_int_sir (ncb_p np) { u_char scntl3; u_char chg, ofs, per, fak, wide; u_char num = INB (nc_dsps); nccb_p cp=0; u_long dsa; u_int target = INB (nc_sdid) & 0x0f; tcb_p tp = &np->target[target]; int i; if (DEBUG_FLAGS & DEBUG_TINY) printf ("I#%d", num); switch (num) { case SIR_SENSE_RESTART: case SIR_STALL_RESTART: break; default: /* ** lookup the nccb */ dsa = INL (nc_dsa); cp = np->link_nccb; while (cp && (CCB_PHYS (cp, phys) != dsa)) cp = cp->link_nccb; assert (cp); if (!cp) goto out; assert (cp == np->header.cp); if (cp != np->header.cp) goto out; } switch (num) { /*-------------------------------------------------------------------- ** ** Processing of interrupted getcc selects ** **-------------------------------------------------------------------- */ case SIR_SENSE_RESTART: /*------------------------------------------ ** Script processor is idle. ** Look for interrupted "check cond" **------------------------------------------ */ if (DEBUG_FLAGS & DEBUG_RESTART) device_printf(np->dev, "int#%d", num); cp = (nccb_p) 0; for (i=0; itarget[i]; if (DEBUG_FLAGS & DEBUG_RESTART) printf ("+"); cp = tp->hold_cp; if (!cp) continue; if (DEBUG_FLAGS & DEBUG_RESTART) printf ("+"); if ((cp->host_status==HS_BUSY) && (cp->s_status==SCSI_STATUS_CHECK_COND)) break; if (DEBUG_FLAGS & DEBUG_RESTART) printf ("- (remove)"); tp->hold_cp = cp = (nccb_p) 0; } if (cp) { if (DEBUG_FLAGS & DEBUG_RESTART) printf ("+ restart job ..\n"); OUTL (nc_dsa, CCB_PHYS (cp, phys)); OUTL (nc_dsp, NCB_SCRIPTH_PHYS (np, getcc)); return; } /* ** no job, resume normal processing */ if (DEBUG_FLAGS & DEBUG_RESTART) printf (" -- remove trap\n"); WRITESCRIPT(start0[0], SCR_INT ^ IFFALSE (0)); break; case SIR_SENSE_FAILED: /*------------------------------------------- ** While trying to select for ** getting the condition code, ** a target reselected us. **------------------------------------------- */ if (DEBUG_FLAGS & DEBUG_RESTART) { PRINT_ADDR(cp->ccb); printf ("in getcc reselect by t%d.\n", INB(nc_ssid) & 0x0f); } /* ** Mark this job */ cp->host_status = HS_BUSY; cp->s_status = SCSI_STATUS_CHECK_COND; np->target[cp->ccb->ccb_h.target_id].hold_cp = cp; /* ** And patch code to restart it. */ WRITESCRIPT(start0[0], SCR_INT); break; /*----------------------------------------------------------------------------- ** ** Was Sie schon immer ueber transfermode negotiation wissen wollten ... ** ** We try to negotiate sync and wide transfer only after ** a successful inquire command. We look at byte 7 of the ** inquire data to determine the capabilities if the target. ** ** When we try to negotiate, we append the negotiation message ** to the identify and (maybe) simple tag message. ** The host status field is set to HS_NEGOTIATE to mark this ** situation. ** ** If the target doesn't answer this message immediately ** (as required by the standard), the SIR_NEGO_FAIL interrupt ** will be raised eventually. ** The handler removes the HS_NEGOTIATE status, and sets the ** negotiated value to the default (async / nowide). ** ** If we receive a matching answer immediately, we check it ** for validity, and set the values. ** ** If we receive a Reject message immediately, we assume the ** negotiation has failed, and fall back to standard values. ** ** If we receive a negotiation message while not in HS_NEGOTIATE ** state, it's a target initiated negotiation. We prepare a ** (hopefully) valid answer, set our parameters, and send back ** this answer to the target. ** ** If the target doesn't fetch the answer (no message out phase), ** we assume the negotiation has failed, and fall back to default ** settings. ** ** When we set the values, we adjust them in all nccbs belonging ** to this target, in the controller's register, and in the "phys" ** field of the controller's struct ncb. ** ** Possible cases: hs sir msg_in value send goto ** We try try to negotiate: ** -> target doesnt't msgin NEG FAIL noop defa. - dispatch ** -> target rejected our msg NEG FAIL reject defa. - dispatch ** -> target answered (ok) NEG SYNC sdtr set - clrack ** -> target answered (!ok) NEG SYNC sdtr defa. REJ--->msg_bad ** -> target answered (ok) NEG WIDE wdtr set - clrack ** -> target answered (!ok) NEG WIDE wdtr defa. REJ--->msg_bad ** -> any other msgin NEG FAIL noop defa. - dispatch ** ** Target tries to negotiate: ** -> incoming message --- SYNC sdtr set SDTR - ** -> incoming message --- WIDE wdtr set WDTR - ** We sent our answer: ** -> target doesn't msgout --- PROTO ? defa. - dispatch ** **----------------------------------------------------------------------------- */ case SIR_NEGO_FAILED: /*------------------------------------------------------- ** ** Negotiation failed. ** Target doesn't send an answer message, ** or target rejected our message. ** ** Remove negotiation request. ** **------------------------------------------------------- */ OUTB (HS_PRT, HS_BUSY); /* FALLTHROUGH */ case SIR_NEGO_PROTO: /*------------------------------------------------------- ** ** Negotiation failed. ** Target doesn't fetch the answer message. ** **------------------------------------------------------- */ if (DEBUG_FLAGS & DEBUG_NEGO) { PRINT_ADDR(cp->ccb); printf ("negotiation failed sir=%x status=%x.\n", num, cp->nego_status); } /* ** any error in negotiation: ** fall back to default mode. */ switch (cp->nego_status) { case NS_SYNC: ncr_setsync (np, cp, 0, 0xe0, 0); break; case NS_WIDE: ncr_setwide (np, cp, 0, 0); break; } np->msgin [0] = MSG_NOOP; np->msgout[0] = MSG_NOOP; cp->nego_status = 0; OUTL (nc_dsp, NCB_SCRIPT_PHYS (np, dispatch)); break; case SIR_NEGO_SYNC: /* ** Synchronous request message received. */ if (DEBUG_FLAGS & DEBUG_NEGO) { PRINT_ADDR(cp->ccb); printf ("sync msgin: "); (void) ncr_show_msg (np->msgin); printf (".\n"); } /* ** get requested values. */ chg = 0; per = np->msgin[3]; ofs = np->msgin[4]; if (ofs==0) per=255; /* ** check values against driver limits. */ if (per < np->minsync) {chg = 1; per = np->minsync;} if (per < tp->tinfo.user.period) {chg = 1; per = tp->tinfo.user.period;} if (ofs > tp->tinfo.user.offset) {chg = 1; ofs = tp->tinfo.user.offset;} /* ** Check against controller limits. */ fak = 7; scntl3 = 0; if (ofs != 0) { ncr_getsync(np, per, &fak, &scntl3); if (fak > 7) { chg = 1; ofs = 0; } } if (ofs == 0) { fak = 7; per = 0; scntl3 = 0; } if (DEBUG_FLAGS & DEBUG_NEGO) { PRINT_ADDR(cp->ccb); printf ("sync: per=%d scntl3=0x%x ofs=%d fak=%d chg=%d.\n", per, scntl3, ofs, fak, chg); } if (INB (HS_PRT) == HS_NEGOTIATE) { OUTB (HS_PRT, HS_BUSY); switch (cp->nego_status) { case NS_SYNC: /* ** This was an answer message */ if (chg) { /* ** Answer wasn't acceptable. */ ncr_setsync (np, cp, 0, 0xe0, 0); OUTL (nc_dsp, NCB_SCRIPT_PHYS (np, msg_bad)); } else { /* ** Answer is ok. */ ncr_setsync (np,cp,scntl3,(fak<<5)|ofs, per); OUTL (nc_dsp, NCB_SCRIPT_PHYS (np, clrack)); } return; case NS_WIDE: ncr_setwide (np, cp, 0, 0); break; } } /* ** It was a request. Set value and ** prepare an answer message */ ncr_setsync (np, cp, scntl3, (fak<<5)|ofs, per); np->msgout[0] = MSG_EXTENDED; np->msgout[1] = 3; np->msgout[2] = MSG_EXT_SDTR; np->msgout[3] = per; np->msgout[4] = ofs; cp->nego_status = NS_SYNC; if (DEBUG_FLAGS & DEBUG_NEGO) { PRINT_ADDR(cp->ccb); printf ("sync msgout: "); (void) ncr_show_msg (np->msgout); printf (".\n"); } if (!ofs) { OUTL (nc_dsp, NCB_SCRIPT_PHYS (np, msg_bad)); return; } np->msgin [0] = MSG_NOOP; break; case SIR_NEGO_WIDE: /* ** Wide request message received. */ if (DEBUG_FLAGS & DEBUG_NEGO) { PRINT_ADDR(cp->ccb); printf ("wide msgin: "); (void) ncr_show_msg (np->msgin); printf (".\n"); } /* ** get requested values. */ chg = 0; wide = np->msgin[3]; /* ** check values against driver limits. */ if (wide > tp->tinfo.user.width) {chg = 1; wide = tp->tinfo.user.width;} if (DEBUG_FLAGS & DEBUG_NEGO) { PRINT_ADDR(cp->ccb); printf ("wide: wide=%d chg=%d.\n", wide, chg); } if (INB (HS_PRT) == HS_NEGOTIATE) { OUTB (HS_PRT, HS_BUSY); switch (cp->nego_status) { case NS_WIDE: /* ** This was an answer message */ if (chg) { /* ** Answer wasn't acceptable. */ ncr_setwide (np, cp, 0, 1); OUTL (nc_dsp, NCB_SCRIPT_PHYS (np, msg_bad)); } else { /* ** Answer is ok. */ ncr_setwide (np, cp, wide, 1); OUTL (nc_dsp, NCB_SCRIPT_PHYS (np, clrack)); } return; case NS_SYNC: ncr_setsync (np, cp, 0, 0xe0, 0); break; } } /* ** It was a request, set value and ** prepare an answer message */ ncr_setwide (np, cp, wide, 1); np->msgout[0] = MSG_EXTENDED; np->msgout[1] = 2; np->msgout[2] = MSG_EXT_WDTR; np->msgout[3] = wide; np->msgin [0] = MSG_NOOP; cp->nego_status = NS_WIDE; if (DEBUG_FLAGS & DEBUG_NEGO) { PRINT_ADDR(cp->ccb); printf ("wide msgout: "); (void) ncr_show_msg (np->msgout); printf (".\n"); } break; /*-------------------------------------------------------------------- ** ** Processing of special messages ** **-------------------------------------------------------------------- */ case SIR_REJECT_RECEIVED: /*----------------------------------------------- ** ** We received a MSG_MESSAGE_REJECT message. ** **----------------------------------------------- */ PRINT_ADDR(cp->ccb); printf ("MSG_MESSAGE_REJECT received (%x:%x).\n", (unsigned)np->lastmsg, np->msgout[0]); break; case SIR_REJECT_SENT: /*----------------------------------------------- ** ** We received an unknown message ** **----------------------------------------------- */ PRINT_ADDR(cp->ccb); printf ("MSG_MESSAGE_REJECT sent for "); (void) ncr_show_msg (np->msgin); printf (".\n"); break; /*-------------------------------------------------------------------- ** ** Processing of special messages ** **-------------------------------------------------------------------- */ case SIR_IGN_RESIDUE: /*----------------------------------------------- ** ** We received an IGNORE RESIDUE message, ** which couldn't be handled by the script. ** **----------------------------------------------- */ PRINT_ADDR(cp->ccb); printf ("MSG_IGN_WIDE_RESIDUE received, but not yet implemented.\n"); break; case SIR_MISSING_SAVE: /*----------------------------------------------- ** ** We received an DISCONNECT message, ** but the datapointer wasn't saved before. ** **----------------------------------------------- */ PRINT_ADDR(cp->ccb); printf ("MSG_DISCONNECT received, but datapointer not saved:\n" "\tdata=%x save=%x goal=%x.\n", (unsigned) INL (nc_temp), (unsigned) np->header.savep, (unsigned) np->header.goalp); break; /*-------------------------------------------------------------------- ** ** Processing of a "SCSI_STATUS_QUEUE_FULL" status. ** ** XXX JGibbs - We should do the same thing for BUSY status. ** ** The current command has been rejected, ** because there are too many in the command queue. ** We have started too many commands for that target. ** **-------------------------------------------------------------------- */ case SIR_STALL_QUEUE: cp->xerr_status = XE_OK; cp->host_status = HS_COMPLETE; cp->s_status = SCSI_STATUS_QUEUE_FULL; ncr_freeze_devq(np, cp->ccb->ccb_h.path); ncr_complete(np, cp); /* FALLTHROUGH */ case SIR_STALL_RESTART: /*----------------------------------------------- ** ** Enable selecting again, ** if NO disconnected jobs. ** **----------------------------------------------- */ /* ** Look for a disconnected job. */ cp = np->link_nccb; while (cp && cp->host_status != HS_DISCONNECT) cp = cp->link_nccb; /* ** if there is one, ... */ if (cp) { /* ** wait for reselection */ OUTL (nc_dsp, NCB_SCRIPT_PHYS (np, reselect)); return; } /* ** else remove the interrupt. */ device_printf(np->dev, "queue empty.\n"); WRITESCRIPT(start1[0], SCR_INT ^ IFFALSE (0)); break; } out: OUTB (nc_dcntl, np->rv_dcntl | STD); } /*========================================================== ** ** ** Acquire a control block ** ** **========================================================== */ static nccb_p ncr_get_nccb (ncb_p np, u_long target, u_long lun) { lcb_p lp; nccb_p cp = NULL; /* ** Lun structure available ? */ lp = np->target[target].lp[lun]; if (lp) { cp = lp->next_nccb; /* ** Look for free CCB */ while (cp && cp->magic) { cp = cp->next_nccb; } } /* ** if nothing available, create one. */ if (cp == NULL) cp = ncr_alloc_nccb(np, target, lun); if (cp != NULL) { if (cp->magic) { device_printf(np->dev, "Bogus free cp found\n"); return (NULL); } cp->magic = 1; } return (cp); } /*========================================================== ** ** ** Release one control block ** ** **========================================================== */ static void ncr_free_nccb (ncb_p np, nccb_p cp) { /* ** sanity */ assert (cp != NULL); cp -> host_status = HS_IDLE; cp -> magic = 0; } /*========================================================== ** ** ** Allocation of resources for Targets/Luns/Tags. ** ** **========================================================== */ static nccb_p ncr_alloc_nccb (ncb_p np, u_long target, u_long lun) { tcb_p tp; lcb_p lp; nccb_p cp; assert (np != NULL); if (target>=MAX_TARGET) return(NULL); if (lun >=MAX_LUN ) return(NULL); tp=&np->target[target]; if (!tp->jump_tcb.l_cmd) { /* ** initialize it. */ tp->jump_tcb.l_cmd = (SCR_JUMP^IFFALSE (DATA (0x80 + target))); tp->jump_tcb.l_paddr = np->jump_tcb.l_paddr; tp->getscr[0] = (np->features & FE_PFEN)? SCR_COPY(1) : SCR_COPY_F(1); tp->getscr[1] = vtophys (&tp->tinfo.sval); tp->getscr[2] = rman_get_start(np->reg_res) + offsetof (struct ncr_reg, nc_sxfer); tp->getscr[3] = (np->features & FE_PFEN)? SCR_COPY(1) : SCR_COPY_F(1); tp->getscr[4] = vtophys (&tp->tinfo.wval); tp->getscr[5] = rman_get_start(np->reg_res) + offsetof (struct ncr_reg, nc_scntl3); assert (((offsetof(struct ncr_reg, nc_sxfer) ^ (offsetof(struct tcb ,tinfo) + offsetof(struct ncr_target_tinfo, sval))) & 3) == 0); assert (((offsetof(struct ncr_reg, nc_scntl3) ^ (offsetof(struct tcb, tinfo) + offsetof(struct ncr_target_tinfo, wval))) &3) == 0); tp->call_lun.l_cmd = (SCR_CALL); tp->call_lun.l_paddr = NCB_SCRIPT_PHYS (np, resel_lun); tp->jump_lcb.l_cmd = (SCR_JUMP); tp->jump_lcb.l_paddr = NCB_SCRIPTH_PHYS (np, abort); np->jump_tcb.l_paddr = vtophys (&tp->jump_tcb); } /* ** Logic unit control block */ lp = tp->lp[lun]; if (!lp) { /* ** Allocate a lcb */ lp = (lcb_p) malloc (sizeof (struct lcb), M_DEVBUF, M_NOWAIT | M_ZERO); if (!lp) return(NULL); /* ** Initialize it */ lp->jump_lcb.l_cmd = (SCR_JUMP ^ IFFALSE (DATA (lun))); lp->jump_lcb.l_paddr = tp->jump_lcb.l_paddr; lp->call_tag.l_cmd = (SCR_CALL); lp->call_tag.l_paddr = NCB_SCRIPT_PHYS (np, resel_tag); lp->jump_nccb.l_cmd = (SCR_JUMP); lp->jump_nccb.l_paddr = NCB_SCRIPTH_PHYS (np, aborttag); lp->actlink = 1; /* ** Chain into LUN list */ tp->jump_lcb.l_paddr = vtophys (&lp->jump_lcb); tp->lp[lun] = lp; } /* ** Allocate a nccb */ cp = (nccb_p) malloc (sizeof (struct nccb), M_DEVBUF, M_NOWAIT|M_ZERO); if (!cp) return (NULL); if (DEBUG_FLAGS & DEBUG_ALLOC) { printf ("new nccb @%p.\n", cp); } /* ** Fill in physical addresses */ cp->p_nccb = vtophys (cp); /* ** Chain into reselect list */ cp->jump_nccb.l_cmd = SCR_JUMP; cp->jump_nccb.l_paddr = lp->jump_nccb.l_paddr; lp->jump_nccb.l_paddr = CCB_PHYS (cp, jump_nccb); cp->call_tmp.l_cmd = SCR_CALL; cp->call_tmp.l_paddr = NCB_SCRIPT_PHYS (np, resel_tmp); /* ** Chain into wakeup list */ cp->link_nccb = np->link_nccb; np->link_nccb = cp; /* ** Chain into CCB list */ cp->next_nccb = lp->next_nccb; lp->next_nccb = cp; return (cp); } /*========================================================== ** ** ** Build Scatter Gather Block ** ** **========================================================== ** ** The transfer area may be scattered among ** several non adjacent physical pages. ** ** We may use MAX_SCATTER blocks. ** **---------------------------------------------------------- */ static int ncr_scatter (struct dsb* phys, vm_offset_t vaddr, vm_size_t datalen) { u_long paddr, pnext; u_short segment = 0; u_long segsize, segaddr; u_long size, csize = 0; u_long chunk = MAX_SIZE; int free; bzero (&phys->data, sizeof (phys->data)); if (!datalen) return (0); paddr = vtophys (vaddr); /* ** insert extra break points at a distance of chunk. ** We try to reduce the number of interrupts caused ** by unexpected phase changes due to disconnects. ** A typical harddisk may disconnect before ANY block. ** If we wanted to avoid unexpected phase changes at all ** we had to use a break point every 512 bytes. ** Of course the number of scatter/gather blocks is ** limited. */ free = MAX_SCATTER - 1; if (vaddr & PAGE_MASK) free -= datalen / PAGE_SIZE; if (free>1) while ((chunk * free >= 2 * datalen) && (chunk>=1024)) chunk /= 2; if(DEBUG_FLAGS & DEBUG_SCATTER) printf("ncr?:\tscattering virtual=%p size=%d chunk=%d.\n", (void *) vaddr, (unsigned) datalen, (unsigned) chunk); /* ** Build data descriptors. */ while (datalen && (segment < MAX_SCATTER)) { /* ** this segment is empty */ segsize = 0; segaddr = paddr; pnext = paddr; if (!csize) csize = chunk; while ((datalen) && (paddr == pnext) && (csize)) { /* ** continue this segment */ pnext = (paddr & (~PAGE_MASK)) + PAGE_SIZE; /* ** Compute max size */ size = pnext - paddr; /* page size */ if (size > datalen) size = datalen; /* data size */ if (size > csize ) size = csize ; /* chunksize */ segsize += size; vaddr += size; csize -= size; datalen -= size; paddr = vtophys (vaddr); } if(DEBUG_FLAGS & DEBUG_SCATTER) printf ("\tseg #%d addr=%x size=%d (rest=%d).\n", segment, (unsigned) segaddr, (unsigned) segsize, (unsigned) datalen); phys->data[segment].addr = segaddr; phys->data[segment].size = segsize; segment++; } if (datalen) { printf("ncr?: scatter/gather failed (residue=%d).\n", (unsigned) datalen); return (-1); } return (segment); } /*========================================================== ** ** ** Test the pci bus snoop logic :-( ** ** Has to be called with interrupts disabled. ** ** **========================================================== */ #ifndef NCR_IOMAPPED static int ncr_regtest (struct ncb* np) { register volatile u_int32_t data; /* ** ncr registers may NOT be cached. ** write 0xffffffff to a read only register area, ** and try to read it back. */ data = 0xffffffff; OUTL_OFF(offsetof(struct ncr_reg, nc_dstat), data); data = INL_OFF(offsetof(struct ncr_reg, nc_dstat)); #if 1 if (data == 0xffffffff) { #else if ((data & 0xe2f0fffd) != 0x02000080) { #endif printf ("CACHE TEST FAILED: reg dstat-sstat2 readback %x.\n", (unsigned) data); return (0x10); } return (0); } #endif static int ncr_snooptest (struct ncb* np) { u_int32_t ncr_rd, ncr_wr, ncr_bk, host_rd, host_wr, pc; int i, err=0; #ifndef NCR_IOMAPPED err |= ncr_regtest (np); if (err) return (err); #endif /* ** init */ pc = NCB_SCRIPTH_PHYS (np, snooptest); host_wr = 1; ncr_wr = 2; /* ** Set memory and register. */ ncr_cache = host_wr; OUTL (nc_temp, ncr_wr); /* ** Start script (exchange values) */ OUTL (nc_dsp, pc); /* ** Wait 'til done (with timeout) */ for (i=0; i=NCR_SNOOP_TIMEOUT) { printf ("CACHE TEST FAILED: timeout.\n"); return (0x20); } /* ** Check termination position. */ if (pc != NCB_SCRIPTH_PHYS (np, snoopend)+8) { printf ("CACHE TEST FAILED: script execution failed.\n"); printf ("start=%08lx, pc=%08lx, end=%08lx\n", (u_long) NCB_SCRIPTH_PHYS (np, snooptest), (u_long) pc, (u_long) NCB_SCRIPTH_PHYS (np, snoopend) +8); return (0x40); } /* ** Show results. */ if (host_wr != ncr_rd) { printf ("CACHE TEST FAILED: host wrote %d, ncr read %d.\n", (int) host_wr, (int) ncr_rd); err |= 1; } if (host_rd != ncr_wr) { printf ("CACHE TEST FAILED: ncr wrote %d, host read %d.\n", (int) ncr_wr, (int) host_rd); err |= 2; } if (ncr_bk != ncr_wr) { printf ("CACHE TEST FAILED: ncr wrote %d, read back %d.\n", (int) ncr_wr, (int) ncr_bk); err |= 4; } return (err); } /*========================================================== ** ** ** Profiling the drivers and targets performance. ** ** **========================================================== */ /* ** Compute the difference in milliseconds. **/ static int ncr_delta (int *from, int *to) { if (!from) return (-1); if (!to) return (-2); return ((to - from) * 1000 / hz); } #define PROFILE cp->phys.header.stamp static void ncb_profile (ncb_p np, nccb_p cp) { int co, da, st, en, di, se, post,work,disc; u_long diff; PROFILE.end = ticks; st = ncr_delta (&PROFILE.start,&PROFILE.status); if (st<0) return; /* status not reached */ da = ncr_delta (&PROFILE.start,&PROFILE.data); if (da<0) return; /* No data transfer phase */ co = ncr_delta (&PROFILE.start,&PROFILE.command); if (co<0) return; /* command not executed */ en = ncr_delta (&PROFILE.start,&PROFILE.end), di = ncr_delta (&PROFILE.start,&PROFILE.disconnect), se = ncr_delta (&PROFILE.start,&PROFILE.select); post = en - st; /* ** @PROFILE@ Disconnect time invalid if multiple disconnects */ if (di>=0) disc = se-di; else disc = 0; work = (st - co) - disc; diff = (np->disc_phys - np->disc_ref) & 0xff; np->disc_ref += diff; np->profile.num_trans += 1; if (cp->ccb) np->profile.num_bytes += cp->ccb->csio.dxfer_len; np->profile.num_disc += diff; np->profile.ms_setup += co; np->profile.ms_data += work; np->profile.ms_disc += disc; np->profile.ms_post += post; } #undef PROFILE /*========================================================== ** ** Determine the ncr's clock frequency. ** This is essential for the negotiation ** of the synchronous transfer rate. ** **========================================================== ** ** Note: we have to return the correct value. ** THERE IS NO SAVE DEFAULT VALUE. ** ** Most NCR/SYMBIOS boards are delivered with a 40 Mhz clock. ** 53C860 and 53C875 rev. 1 support fast20 transfers but ** do not have a clock doubler and so are provided with a ** 80 MHz clock. All other fast20 boards incorporate a doubler ** and so should be delivered with a 40 MHz clock. ** The future fast40 chips (895/895) use a 40 Mhz base clock ** and provide a clock quadrupler (160 Mhz). The code below ** tries to deal as cleverly as possible with all this stuff. ** **---------------------------------------------------------- */ /* * Select NCR SCSI clock frequency */ static void ncr_selectclock(ncb_p np, u_char scntl3) { if (np->multiplier < 2) { OUTB(nc_scntl3, scntl3); return; } if (bootverbose >= 2) device_printf(np->dev, "enabling clock multiplier\n"); OUTB(nc_stest1, DBLEN); /* Enable clock multiplier */ if (np->multiplier > 2) { /* Poll bit 5 of stest4 for quadrupler */ int i = 20; while (!(INB(nc_stest4) & LCKFRQ) && --i > 0) DELAY(20); if (!i) device_printf(np->dev, "the chip cannot lock the frequency\n"); } else /* Wait 20 micro-seconds for doubler */ DELAY(20); OUTB(nc_stest3, HSC); /* Halt the scsi clock */ OUTB(nc_scntl3, scntl3); OUTB(nc_stest1, (DBLEN|DBLSEL));/* Select clock multiplier */ OUTB(nc_stest3, 0x00); /* Restart scsi clock */ } /* * calculate NCR SCSI clock frequency (in KHz) */ static unsigned ncrgetfreq (ncb_p np, int gen) { int ms = 0; /* * Measure GEN timer delay in order * to calculate SCSI clock frequency * * This code will never execute too * many loop iterations (if DELAY is * reasonably correct). It could get * too low a delay (too high a freq.) * if the CPU is slow executing the * loop for some reason (an NMI, for * example). For this reason we will * if multiple measurements are to be * performed trust the higher delay * (lower frequency returned). */ OUTB (nc_stest1, 0); /* make sure clock doubler is OFF */ OUTW (nc_sien , 0); /* mask all scsi interrupts */ (void) INW (nc_sist); /* clear pending scsi interrupt */ OUTB (nc_dien , 0); /* mask all dma interrupts */ (void) INW (nc_sist); /* another one, just to be sure :) */ OUTB (nc_scntl3, 4); /* set pre-scaler to divide by 3 */ OUTB (nc_stime1, 0); /* disable general purpose timer */ OUTB (nc_stime1, gen); /* set to nominal delay of (1<= 2) printf ("\tDelay (GEN=%d): %u msec\n", gen, ms); /* * adjust for prescaler, and convert into KHz */ return ms ? ((1 << gen) * 4440) / ms : 0; } static void ncr_getclock (ncb_p np, u_char multiplier) { unsigned char scntl3; unsigned char stest1; scntl3 = INB(nc_scntl3); stest1 = INB(nc_stest1); np->multiplier = 1; if (multiplier > 1) { np->multiplier = multiplier; np->clock_khz = 40000 * multiplier; } else { if ((scntl3 & 7) == 0) { unsigned f1, f2; /* throw away first result */ (void) ncrgetfreq (np, 11); f1 = ncrgetfreq (np, 11); f2 = ncrgetfreq (np, 11); if (bootverbose >= 2) printf ("\tNCR clock is %uKHz, %uKHz\n", f1, f2); if (f1 > f2) f1 = f2; /* trust lower result */ if (f1 > 45000) { scntl3 = 5; /* >45Mhz: assume 80MHz */ } else { scntl3 = 3; /* <45Mhz: assume 40MHz */ } } else if ((scntl3 & 7) == 5) np->clock_khz = 80000; /* Probably a 875 rev. 1 ? */ } } /*=========================================================================*/ #ifdef NCR_TEKRAM_EEPROM struct tekram_eeprom_dev { u_char devmode; #define TKR_PARCHK 0x01 #define TKR_TRYSYNC 0x02 #define TKR_ENDISC 0x04 #define TKR_STARTUNIT 0x08 #define TKR_USETAGS 0x10 #define TKR_TRYWIDE 0x20 u_char syncparam; /* max. sync transfer rate (table ?) */ u_char filler1; u_char filler2; }; struct tekram_eeprom { struct tekram_eeprom_dev dev[16]; u_char adaptid; u_char adaptmode; #define TKR_ADPT_GT2DRV 0x01 #define TKR_ADPT_GT1GB 0x02 #define TKR_ADPT_RSTBUS 0x04 #define TKR_ADPT_ACTNEG 0x08 #define TKR_ADPT_NOSEEK 0x10 #define TKR_ADPT_MORLUN 0x20 u_char delay; /* unit ? ( table ??? ) */ u_char tags; /* use 4 times as many ... */ u_char filler[60]; }; static void tekram_write_bit (ncb_p np, int bit) { u_char val = 0x10 + ((bit & 1) << 1); DELAY(10); OUTB (nc_gpreg, val); DELAY(10); OUTB (nc_gpreg, val | 0x04); DELAY(10); OUTB (nc_gpreg, val); DELAY(10); } static int tekram_read_bit (ncb_p np) { OUTB (nc_gpreg, 0x10); DELAY(10); OUTB (nc_gpreg, 0x14); DELAY(10); return INB (nc_gpreg) & 1; } static u_short read_tekram_eeprom_reg (ncb_p np, int reg) { int bit; u_short result = 0; int cmd = 0x80 | reg; OUTB (nc_gpreg, 0x10); tekram_write_bit (np, 1); for (bit = 7; bit >= 0; bit--) { tekram_write_bit (np, cmd >> bit); } for (bit = 0; bit < 16; bit++) { result <<= 1; result |= tekram_read_bit (np); } OUTB (nc_gpreg, 0x00); return result; } static int read_tekram_eeprom(ncb_p np, struct tekram_eeprom *buffer) { u_short *p = (u_short *) buffer; u_short sum = 0; int i; if (INB (nc_gpcntl) != 0x09) { return 0; } for (i = 0; i < 64; i++) { u_short val; if((i&0x0f) == 0) printf ("%02x:", i*2); val = read_tekram_eeprom_reg (np, i); if (p) *p++ = val; sum += val; if((i&0x01) == 0x00) printf (" "); printf ("%02x%02x", val & 0xff, (val >> 8) & 0xff); if((i&0x0f) == 0x0f) printf ("\n"); } printf ("Sum = %04x\n", sum); return sum == 0x1234; } #endif /* NCR_TEKRAM_EEPROM */ static device_method_t ncr_methods[] = { /* Device interface */ DEVMETHOD(device_probe, ncr_probe), DEVMETHOD(device_attach, ncr_attach), { 0, 0 } }; static driver_t ncr_driver = { "ncr", ncr_methods, sizeof(struct ncb), }; static devclass_t ncr_devclass; DRIVER_MODULE(ncr, pci, ncr_driver, ncr_devclass, 0, 0); MODULE_DEPEND(ncr, cam, 1, 1, 1); MODULE_DEPEND(ncr, pci, 1, 1, 1); /*=========================================================================*/ #endif /* _KERNEL */ Index: projects/clang400-import/sys/dev/nvme/nvme_sim.c =================================================================== --- projects/clang400-import/sys/dev/nvme/nvme_sim.c (revision 311459) +++ projects/clang400-import/sys/dev/nvme/nvme_sim.c (revision 311460) @@ -1,395 +1,395 @@ /*- * Copyright (c) 2016 Netflix, Inc * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer, * without modification, immediately at the beginning of the file. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include // Yes, this is wrong. #include #include "nvme_private.h" #define ccb_accb_ptr spriv_ptr0 #define ccb_ctrlr_ptr spriv_ptr1 static void nvme_sim_action(struct cam_sim *sim, union ccb *ccb); static void nvme_sim_poll(struct cam_sim *sim); #define sim2softc(sim) ((struct nvme_sim_softc *)cam_sim_softc(sim)) #define sim2ns(sim) (sim2softc(sim)->s_ns) #define sim2ctrlr(sim) (sim2softc(sim)->s_ctrlr) struct nvme_sim_softc { struct nvme_controller *s_ctrlr; struct nvme_namespace *s_ns; struct cam_sim *s_sim; struct cam_path *s_path; }; static void nvme_sim_nvmeio_done(void *ccb_arg, const struct nvme_completion *cpl) { union ccb *ccb = (union ccb *)ccb_arg; /* * Let the periph know the completion, and let it sort out what * it means. Make our best guess, though for the status code. */ memcpy(&ccb->nvmeio.cpl, cpl, sizeof(*cpl)); if (nvme_completion_is_error(cpl)) ccb->ccb_h.status = CAM_REQ_CMP_ERR; else ccb->ccb_h.status = CAM_REQ_CMP; xpt_done(ccb); } static void nvme_sim_nvmeio(struct cam_sim *sim, union ccb *ccb) { struct ccb_nvmeio *nvmeio = &ccb->nvmeio; struct nvme_request *req; void *payload; uint32_t size; struct nvme_controller *ctrlr; ctrlr = sim2ctrlr(sim); payload = nvmeio->data_ptr; size = nvmeio->dxfer_len; /* SG LIST ??? */ if ((nvmeio->ccb_h.flags & CAM_DATA_MASK) == CAM_DATA_BIO) req = nvme_allocate_request_bio((struct bio *)payload, nvme_sim_nvmeio_done, ccb); else if (payload == NULL) req = nvme_allocate_request_null(nvme_sim_nvmeio_done, ccb); else req = nvme_allocate_request_vaddr(payload, size, nvme_sim_nvmeio_done, ccb); if (req == NULL) { nvmeio->ccb_h.status = CAM_RESRC_UNAVAIL; xpt_done(ccb); return; } memcpy(&req->cmd, &ccb->nvmeio.cmd, sizeof(ccb->nvmeio.cmd)); nvme_ctrlr_submit_io_request(ctrlr, req); ccb->ccb_h.status |= CAM_SIM_QUEUED; } static void nvme_sim_action(struct cam_sim *sim, union ccb *ccb) { struct nvme_controller *ctrlr; struct nvme_namespace *ns; CAM_DEBUG(ccb->ccb_h.path, CAM_DEBUG_TRACE, ("nvme_sim_action: func= %#x\n", ccb->ccb_h.func_code)); /* * XXX when we support multiple namespaces in the base driver we'll need * to revisit how all this gets stored and saved in the periph driver's * reserved areas. Right now we store all three in the softc of the sim. */ ns = sim2ns(sim); ctrlr = sim2ctrlr(sim); mtx_assert(&ctrlr->lock, MA_OWNED); switch (ccb->ccb_h.func_code) { case XPT_CALC_GEOMETRY: /* Calculate Geometry Totally nuts ? XXX */ /* * Only meaningful for old-school SCSI disks since only the SCSI * da driver generates them. Reject all these that slip through. */ /*FALLTHROUGH*/ case XPT_ABORT: /* Abort the specified CCB */ case XPT_EN_LUN: /* Enable LUN as a target */ case XPT_TARGET_IO: /* Execute target I/O request */ case XPT_ACCEPT_TARGET_IO: /* Accept Host Target Mode CDB */ case XPT_CONT_TARGET_IO: /* Continue Host Target I/O Connection*/ /* * Only target mode generates these, and only for SCSI. They are * all invalid/unsupported for NVMe. */ ccb->ccb_h.status = CAM_REQ_INVALID; break; case XPT_SET_TRAN_SETTINGS: /* * NVMe doesn't really have different transfer settings, but * other parts of CAM think failure here is a big deal. */ ccb->ccb_h.status = CAM_REQ_CMP; break; case XPT_PATH_INQ: /* Path routing inquiry */ { struct ccb_pathinq *cpi = &ccb->cpi; /* * NVMe may have multiple LUNs on the same path. Current generation * of NVMe devives support only a single name space. Multiple name * space drives are coming, but it's unclear how we should report * them up the stack. */ cpi->version_num = 1; cpi->hba_inquiry = 0; cpi->target_sprt = 0; cpi->hba_misc = PIM_UNMAPPED /* | PIM_NOSCAN */; cpi->hba_eng_cnt = 0; cpi->max_target = 0; cpi->max_lun = ctrlr->cdata.nn; cpi->maxio = nvme_ns_get_max_io_xfer_size(ns); cpi->initiator_id = 0; cpi->bus_id = cam_sim_bus(sim); cpi->base_transfer_speed = 4000000; /* 4 GB/s 4 lanes pcie 3 */ strlcpy(cpi->sim_vid, "FreeBSD", SIM_IDLEN); strlcpy(cpi->hba_vid, "NVMe", HBA_IDLEN); strlcpy(cpi->dev_name, cam_sim_name(sim), DEV_IDLEN); cpi->unit_number = cam_sim_unit(sim); - cpi->transport = XPORT_NVME; /* XXX XPORT_PCIE ? */ - cpi->transport_version = 1; /* XXX Get PCIe spec ? */ - cpi->protocol = PROTO_NVME; - cpi->protocol_version = NVME_REV_1; /* Groks all 1.x NVMe cards */ + cpi->transport = XPORT_NVME; /* XXX XPORT_PCIE ? */ + cpi->transport_version = 1; /* XXX Get PCIe spec ? */ + cpi->protocol = PROTO_NVME; + cpi->protocol_version = NVME_REV_1; /* Groks all 1.x NVMe cards */ cpi->xport_specific.nvme.nsid = ns->id; cpi->ccb_h.status = CAM_REQ_CMP; break; } case XPT_GET_TRAN_SETTINGS: /* Get transport settings */ { struct ccb_trans_settings *cts; struct ccb_trans_settings_nvme *nvmep; struct ccb_trans_settings_nvme *nvmex; cts = &ccb->cts; nvmex = &cts->xport_specific.nvme; nvmep = &cts->proto_specific.nvme; nvmex->valid = CTS_NVME_VALID_SPEC; nvmex->spec_major = 1; /* XXX read from card */ nvmex->spec_minor = 2; nvmex->spec_tiny = 0; nvmep->valid = CTS_NVME_VALID_SPEC; nvmep->spec_major = 1; /* XXX read from card */ nvmep->spec_minor = 2; nvmep->spec_tiny = 0; cts->transport = XPORT_NVME; cts->protocol = PROTO_NVME; cts->ccb_h.status = CAM_REQ_CMP; break; } case XPT_TERM_IO: /* Terminate the I/O process */ /* * every driver handles this, but nothing generates it. Assume * it's OK to just say 'that worked'. */ /*FALLTHROUGH*/ case XPT_RESET_DEV: /* Bus Device Reset the specified device */ case XPT_RESET_BUS: /* Reset the specified bus */ /* * NVMe doesn't really support physically resetting the bus. It's part * of the bus scanning dance, so return sucess to tell the process to * proceed. */ ccb->ccb_h.status = CAM_REQ_CMP; break; case XPT_NVME_IO: /* Execute the requested I/O operation */ nvme_sim_nvmeio(sim, ccb); return; /* no done */ default: ccb->ccb_h.status = CAM_REQ_INVALID; break; } xpt_done(ccb); } static void nvme_sim_poll(struct cam_sim *sim) { nvme_ctrlr_intx_handler(sim2ctrlr(sim)); } static void * nvme_sim_new_controller(struct nvme_controller *ctrlr) { struct cam_devq *devq; int max_trans; int unit; struct nvme_sim_softc *sc = NULL; max_trans = 256;/* XXX not so simple -- must match queues */ unit = device_get_unit(ctrlr->dev); devq = cam_simq_alloc(max_trans); if (devq == NULL) return NULL; sc = malloc(sizeof(*sc), M_NVME, M_ZERO | M_WAITOK); sc->s_ctrlr = ctrlr; sc->s_sim = cam_sim_alloc(nvme_sim_action, nvme_sim_poll, "nvme", sc, unit, &ctrlr->lock, max_trans, max_trans, devq); if (sc->s_sim == NULL) { printf("Failed to allocate a sim\n"); cam_simq_free(devq); free(sc, M_NVME); return NULL; } return sc; } static void nvme_sim_rescan_target(struct nvme_controller *ctrlr, struct cam_path *path) { union ccb *ccb; ccb = xpt_alloc_ccb_nowait(); if (ccb == NULL) { printf("unable to alloc CCB for rescan\n"); return; } if (xpt_clone_path(&ccb->ccb_h.path, path) != CAM_REQ_CMP) { printf("unable to copy path for rescan\n"); xpt_free_ccb(ccb); return; } xpt_rescan(ccb); } static void * nvme_sim_new_ns(struct nvme_namespace *ns, void *sc_arg) { struct nvme_sim_softc *sc = sc_arg; struct nvme_controller *ctrlr = sc->s_ctrlr; int i; sc->s_ns = ns; /* * XXX this is creating one bus per ns, but it should be one * XXX target per controller, and one LUN per namespace. * XXX Current drives only support one NS, so there's time * XXX to fix it later when new drives arrive. * * XXX I'm pretty sure the xpt_bus_register() call below is * XXX like super lame and it really belongs in the sim_new_ctrlr * XXX callback. Then the create_path below would be pretty close * XXX to being right. Except we should be per-ns not per-ctrlr * XXX data. */ mtx_lock(&ctrlr->lock); /* Create bus */ /* * XXX do I need to lock ctrlr->lock ? * XXX do I need to lock the path? * ata and scsi seem to in their code, but their discovery is * somewhat more asynchronous. We're only every called one at a * time, and nothing is in parallel. */ i = 0; if (xpt_bus_register(sc->s_sim, ctrlr->dev, 0) != CAM_SUCCESS) goto error; i++; if (xpt_create_path(&sc->s_path, /*periph*/NULL, cam_sim_path(sc->s_sim), 1, ns->id) != CAM_REQ_CMP) goto error; i++; sc->s_path->device->nvme_data = nvme_ns_get_data(ns); sc->s_path->device->nvme_cdata = nvme_ctrlr_get_data(ns->ctrlr); /* Scan bus */ nvme_sim_rescan_target(ctrlr, sc->s_path); mtx_unlock(&ctrlr->lock); return ns; error: switch (i) { case 2: xpt_free_path(sc->s_path); case 1: xpt_bus_deregister(cam_sim_path(sc->s_sim)); case 0: cam_sim_free(sc->s_sim, /*free_devq*/TRUE); } mtx_unlock(&ctrlr->lock); return NULL; } static void nvme_sim_controller_fail(void *ctrlr_arg) { /* XXX cleanup XXX */ } struct nvme_consumer *consumer_cookie; static void nvme_sim_init(void) { consumer_cookie = nvme_register_consumer(nvme_sim_new_ns, nvme_sim_new_controller, NULL, nvme_sim_controller_fail); } SYSINIT(nvme_sim_register, SI_SUB_DRIVERS, SI_ORDER_ANY, nvme_sim_init, NULL); static void nvme_sim_uninit(void) { /* XXX Cleanup */ nvme_unregister_consumer(consumer_cookie); } SYSUNINIT(nvme_sim_unregister, SI_SUB_DRIVERS, SI_ORDER_ANY, nvme_sim_uninit, NULL); Index: projects/clang400-import/sys/dev/rtwn/rtl8192c/pci/r92ce_init.c =================================================================== --- projects/clang400-import/sys/dev/rtwn/rtl8192c/pci/r92ce_init.c (revision 311459) +++ projects/clang400-import/sys/dev/rtwn/rtl8192c/pci/r92ce_init.c (revision 311460) @@ -1,327 +1,323 @@ /* $OpenBSD: if_rtwn.c,v 1.6 2015/08/28 00:03:53 deraadt Exp $ */ /*- * Copyright (c) 2010 Damien Bergamini * Copyright (c) 2015 Stefan Sperling * Copyright (c) 2016 Andriy Voskoboinyk * * 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 "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 void r92ce_init_intr(struct rtwn_softc *sc) { /* Disable interrupts. */ rtwn_write_4(sc, R92C_HISR, 0x00000000); rtwn_write_4(sc, R92C_HIMR, 0x00000000); } void r92ce_init_edca(struct rtwn_softc *sc) { /* SIFS */ rtwn_write_2(sc, R92C_SPEC_SIFS, 0x1010); rtwn_write_2(sc, R92C_MAC_SPEC_SIFS, 0x1010); rtwn_write_2(sc, R92C_SIFS_CCK, 0x1010); rtwn_write_2(sc, R92C_SIFS_OFDM, 0x0e0e); /* TXOP */ rtwn_write_4(sc, R92C_EDCA_BE_PARAM, 0x005ea42b); rtwn_write_4(sc, R92C_EDCA_BK_PARAM, 0x0000a44f); rtwn_write_4(sc, R92C_EDCA_VI_PARAM, 0x005e4322); rtwn_write_4(sc, R92C_EDCA_VO_PARAM, 0x002f3222); } void r92ce_init_bb(struct rtwn_softc *sc) { /* Enable BB and RF. */ rtwn_setbits_2(sc, R92C_SYS_FUNC_EN, 0, R92C_SYS_FUNC_EN_BBRSTB | R92C_SYS_FUNC_EN_BB_GLB_RST | R92C_SYS_FUNC_EN_DIO_RF); rtwn_write_2(sc, R92C_AFE_PLL_CTRL, 0xdb83); rtwn_write_1(sc, R92C_RF_CTRL, R92C_RF_CTRL_EN | R92C_RF_CTRL_RSTB | R92C_RF_CTRL_SDMRSTB); rtwn_write_1(sc, R92C_SYS_FUNC_EN, R92C_SYS_FUNC_EN_DIO_PCIE | R92C_SYS_FUNC_EN_PCIEA | R92C_SYS_FUNC_EN_PPLL | R92C_SYS_FUNC_EN_BB_GLB_RST | R92C_SYS_FUNC_EN_BBRSTB); rtwn_write_1(sc, R92C_AFE_XTAL_CTRL + 1, 0x80); rtwn_setbits_4(sc, R92C_LEDCFG0, 0, 0x00800000); r92c_init_bb_common(sc); - - if (rtwn_bb_read(sc, R92C_HSSI_PARAM2(0)) & - R92C_HSSI_PARAM2_CCK_HIPWR) - sc->sc_flags |= RTWN_FLAG_CCK_HIPWR; } int r92ce_power_on(struct rtwn_softc *sc) { struct r92c_softc *rs = sc->sc_priv; uint32_t reg; int ntries; /* Wait for autoload done bit. */ for (ntries = 0; ntries < 1000; ntries++) { if (rtwn_read_1(sc, R92C_APS_FSMCO) & R92C_APS_FSMCO_PFM_ALDN) break; DELAY(5); } if (ntries == 1000) { device_printf(sc->sc_dev, "timeout waiting for chip autoload\n"); return (ETIMEDOUT); } /* Unlock ISO/CLK/Power control register. */ rtwn_write_1(sc, R92C_RSV_CTRL, 0); if (rs->board_type != R92C_BOARD_TYPE_DONGLE) { /* bt coex */ rtwn_setbits_4(sc, R92C_APS_FSMCO, 0, R92C_APS_FSMCO_SOP_ABG | R92C_APS_FSMCO_SOP_AMB | R92C_APS_FSMCO_XOP_BTCK); } /* Move SPS into PWM mode. */ rtwn_write_1(sc, R92C_SPS0_CTRL, 0x2b); /* Set low byte to 0x0f, leave others unchanged. */ rtwn_write_1(sc, R92C_AFE_XTAL_CTRL, 0x0f); /* TODO: check if we need this for 8188CE */ if (rs->board_type != R92C_BOARD_TYPE_DONGLE) { /* bt coex */ /* XXX magic from linux */ rtwn_setbits_4(sc, R92C_AFE_XTAL_CTRL, 0x024800, 0); } rtwn_setbits_2(sc, R92C_SYS_ISO_CTRL, 0xff00, R92C_SYS_ISO_CTRL_PWC_EV12V | R92C_SYS_ISO_CTRL_DIOR); DELAY(200); /* TODO: linux does additional btcoex stuff here */ /* Auto enable WLAN. */ rtwn_setbits_2(sc, R92C_APS_FSMCO, 0, R92C_APS_FSMCO_APFM_ONMAC); for (ntries = 0; ntries < 1000; ntries++) { if (!(rtwn_read_2(sc, R92C_APS_FSMCO) & R92C_APS_FSMCO_APFM_ONMAC)) break; DELAY(5); } if (ntries == 1000) { device_printf(sc->sc_dev, "timeout waiting for MAC auto ON\n"); return (ETIMEDOUT); } /* Enable radio, GPIO and LED functions. */ rtwn_write_2(sc, R92C_APS_FSMCO, R92C_APS_FSMCO_AFSM_PCIE | R92C_APS_FSMCO_PDN_EN | R92C_APS_FSMCO_PFM_ALDN); /* Release RF digital isolation. */ rtwn_setbits_2(sc, R92C_SYS_ISO_CTRL, R92C_SYS_ISO_CTRL_DIOR, 0); if (rs->chip & R92C_CHIP_92C) rtwn_write_1(sc, R92C_PCIE_CTRL_REG + 3, 0x77); else rtwn_write_1(sc, R92C_PCIE_CTRL_REG + 3, 0x22); rtwn_write_4(sc, R92C_INT_MIG, 0); if (rs->board_type != R92C_BOARD_TYPE_DONGLE) { /* bt coex */ /* XXX magic from linux */ rtwn_setbits_1(sc, R92C_AFE_XTAL_CTRL + 2, 0x02, 0); } rtwn_setbits_1(sc, R92C_GPIO_MUXCFG, R92C_GPIO_MUXCFG_RFKILL, 0); reg = rtwn_read_1(sc, R92C_GPIO_IO_SEL); if (!(reg & R92C_GPIO_IO_SEL_RFKILL)) { device_printf(sc->sc_dev, "radio is disabled by hardware switch\n"); /* XXX how driver will know when radio will be enabled? */ return (EPERM); } /* Initialize MAC. */ rtwn_setbits_1(sc, R92C_APSD_CTRL, R92C_APSD_CTRL_OFF, 0); for (ntries = 0; ntries < 200; ntries++) { if (!(rtwn_read_1(sc, R92C_APSD_CTRL) & R92C_APSD_CTRL_OFF_STATUS)) break; DELAY(500); } if (ntries == 200) { device_printf(sc->sc_dev, "timeout waiting for MAC initialization\n"); return (ETIMEDOUT); } /* Enable MAC DMA/WMAC/SCHEDULE/SEC blocks. */ rtwn_setbits_2(sc, R92C_CR, 0, R92C_CR_HCI_TXDMA_EN | R92C_CR_HCI_RXDMA_EN | R92C_CR_TXDMA_EN | R92C_CR_RXDMA_EN | R92C_CR_PROTOCOL_EN | R92C_CR_SCHEDULE_EN | R92C_CR_MACTXEN | R92C_CR_MACRXEN | ((sc->sc_hwcrypto != RTWN_CRYPTO_SW) ? R92C_CR_ENSEC : 0)); rtwn_write_4(sc, R92C_MCUTST_1, 0x0); return (0); } void r92ce_power_off(struct rtwn_softc *sc) { #ifndef RTWN_WITHOUT_UCODE struct r92c_softc *rs = sc->sc_priv; /* Deinit C2H event handler. */ callout_stop(&rs->rs_c2h_report); rs->rs_c2h_paused = 0; rs->rs_c2h_pending = 0; rs->rs_c2h_timeout = hz; #endif /* Stop hardware. */ /* Disable interrupts. */ rtwn_write_4(sc, R92C_HISR, 0); rtwn_write_4(sc, R92C_HIMR, 0); /* Stop hardware. */ rtwn_write_1(sc, R92C_TXPAUSE, R92C_TX_QUEUE_ALL); /* Turn off RF. */ rtwn_write_1(sc, R92C_RF_CTRL, 0); /* Reset BB state machine */ rtwn_setbits_1(sc, R92C_SYS_FUNC_EN, 0, R92C_SYS_FUNC_EN_BB_GLB_RST); rtwn_setbits_1(sc, R92C_SYS_FUNC_EN, R92C_SYS_FUNC_EN_BB_GLB_RST, 0); /* Disable MAC DMA/WMAC/SCHEDULE/SEC blocks. */ rtwn_setbits_2(sc, R92C_CR, R92C_CR_HCI_TXDMA_EN | R92C_CR_HCI_RXDMA_EN | R92C_CR_TXDMA_EN | R92C_CR_RXDMA_EN | R92C_CR_PROTOCOL_EN | R92C_CR_SCHEDULE_EN | R92C_CR_MACTXEN | R92C_CR_MACRXEN | R92C_CR_ENSEC, 0); /* If firmware in ram code, do reset. */ #ifndef RTWN_WITHOUT_UCODE if (rtwn_read_1(sc, R92C_MCUFWDL) & R92C_MCUFWDL_RAM_DL_SEL) r92ce_fw_reset(sc, RTWN_FW_RESET_SHUTDOWN); #endif /* TODO: linux does additional btcoex stuff here */ rtwn_write_2(sc, R92C_AFE_PLL_CTRL, 0x80); /* linux magic number */ rtwn_write_1(sc, R92C_SPS0_CTRL, 0x23); /* ditto */ rtwn_write_1(sc, R92C_AFE_XTAL_CTRL, 0x0e); /* different with btcoex */ rtwn_write_1(sc, R92C_RSV_CTRL, 0x0e); rtwn_write_1(sc, R92C_APS_FSMCO, R92C_APS_FSMCO_PDN_EN); } void r92ce_init_ampdu(struct rtwn_softc *sc) { /* Setup AMPDU aggregation. */ rtwn_write_4(sc, R92C_AGGLEN_LMT, 0x99997631); /* MCS7~0 */ rtwn_write_1(sc, R92C_AGGR_BREAK_TIME, 0x16); } void r92ce_post_init(struct rtwn_softc *sc) { rtwn_write_2(sc, R92C_FWHW_TXQ_CTRL, 0x1f00 | R92C_FWHW_TXQ_CTRL_AMPDU_RTY_NEW); rtwn_write_1(sc, R92C_BCN_MAX_ERR, 0xff); /* Perform LO and IQ calibrations. */ r92ce_iq_calib(sc); /* Perform LC calibration. */ r92c_lc_calib(sc); r92c_pa_bias_init(sc); /* Fix for lower temperature. */ rtwn_write_1(sc, 0x15, 0xe9); #ifndef RTWN_WITHOUT_UCODE if (sc->sc_flags & RTWN_FW_LOADED) { struct r92c_softc *rs = sc->sc_priv; if (sc->sc_ratectl_sysctl == RTWN_RATECTL_FW) { /* XXX TODO: fix (see comment in r92cu_init.c) */ sc->sc_ratectl = RTWN_RATECTL_NET80211; } else sc->sc_ratectl = sc->sc_ratectl_sysctl; /* Start C2H event handling. */ callout_reset(&rs->rs_c2h_report, rs->rs_c2h_timeout, r92c_handle_c2h_report, sc); } else #endif sc->sc_ratectl = RTWN_RATECTL_NONE; } Index: projects/clang400-import/sys/dev/rtwn/rtl8192c/r92c_init.c =================================================================== --- projects/clang400-import/sys/dev/rtwn/rtl8192c/r92c_init.c (revision 311459) +++ projects/clang400-import/sys/dev/rtwn/rtl8192c/r92c_init.c (revision 311460) @@ -1,314 +1,317 @@ /* $OpenBSD: if_urtwn.c,v 1.16 2011/02/10 17:26:40 jakemsr Exp $ */ /*- * Copyright (c) 2010 Damien Bergamini * Copyright (c) 2014 Kevin Lo * Copyright (c) 2015-2016 Andriy Voskoboinyk * * 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 "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 int r92c_check_condition(struct rtwn_softc *sc, const uint8_t cond[]) { struct r92c_softc *rs = sc->sc_priv; uint8_t mask; int i; if (cond[0] == 0) return (1); RTWN_DPRINTF(sc, RTWN_DEBUG_RESET, "%s: condition byte 0: %02X; chip %02X, board %02X\n", __func__, cond[0], rs->chip, rs->board_type); if (!(rs->chip & R92C_CHIP_92C)) { if (rs->board_type == R92C_BOARD_TYPE_HIGHPA) mask = R92C_COND_RTL8188RU; else if (rs->board_type == R92C_BOARD_TYPE_MINICARD) mask = R92C_COND_RTL8188CE; else mask = R92C_COND_RTL8188CU; } else { if (rs->board_type == R92C_BOARD_TYPE_MINICARD) mask = R92C_COND_RTL8192CE; else mask = R92C_COND_RTL8192CU; } for (i = 0; i < RTWN_MAX_CONDITIONS && cond[i] != 0; i++) if ((cond[i] & mask) == mask) return (1); return (0); } int r92c_set_page_size(struct rtwn_softc *sc) { return (rtwn_write_1(sc, R92C_PBP, SM(R92C_PBP_PSRX, R92C_PBP_128) | SM(R92C_PBP_PSTX, R92C_PBP_128)) == 0); } void r92c_init_bb_common(struct rtwn_softc *sc) { struct r92c_softc *rs = sc->sc_priv; int i, j; /* Write BB initialization values. */ for (i = 0; i < sc->bb_size; i++) { const struct rtwn_bb_prog *bb_prog = &sc->bb_prog[i]; while (!rtwn_check_condition(sc, bb_prog->cond)) { KASSERT(bb_prog->next != NULL, ("%s: wrong condition value (i %d)\n", __func__, i)); bb_prog = bb_prog->next; } for (j = 0; j < bb_prog->count; j++) { RTWN_DPRINTF(sc, RTWN_DEBUG_RESET, "BB: reg 0x%03x, val 0x%08x\n", bb_prog->reg[j], bb_prog->val[j]); rtwn_bb_write(sc, bb_prog->reg[j], bb_prog->val[j]); rtwn_delay(sc, 1); } } if (rs->chip & R92C_CHIP_92C_1T2R) { /* 8192C 1T only configuration. */ rtwn_bb_setbits(sc, R92C_FPGA0_TXINFO, 0x03, 0x02); rtwn_bb_setbits(sc, R92C_FPGA1_TXINFO, 0x300033, 0x200022); rtwn_bb_setbits(sc, R92C_CCK0_AFESETTING, 0xff000000, 0x45000000); rtwn_bb_setbits(sc, R92C_OFDM0_TRXPATHENA, 0xff, 0x23); rtwn_bb_setbits(sc, R92C_OFDM0_AGCPARAM1, 0x30, 0x10); rtwn_bb_setbits(sc, 0xe74, 0x0c000000, 0x08000000); rtwn_bb_setbits(sc, 0xe78, 0x0c000000, 0x08000000); rtwn_bb_setbits(sc, 0xe7c, 0x0c000000, 0x08000000); rtwn_bb_setbits(sc, 0xe80, 0x0c000000, 0x08000000); rtwn_bb_setbits(sc, 0xe88, 0x0c000000, 0x08000000); } /* Write AGC values. */ for (i = 0; i < sc->agc_size; i++) { const struct rtwn_agc_prog *agc_prog = &sc->agc_prog[i]; while (!rtwn_check_condition(sc, agc_prog->cond)) { KASSERT(agc_prog->next != NULL, ("%s: wrong condition value (2) (i %d)\n", __func__, i)); agc_prog = agc_prog->next; } for (j = 0; j < agc_prog->count; j++) { RTWN_DPRINTF(sc, RTWN_DEBUG_RESET, "AGC: val 0x%08x\n", agc_prog->val[j]); rtwn_bb_write(sc, R92C_OFDM0_AGCRSSITABLE, agc_prog->val[j]); rtwn_delay(sc, 1); } } + + if (rtwn_bb_read(sc, R92C_HSSI_PARAM2(0)) & R92C_HSSI_PARAM2_CCK_HIPWR) + sc->sc_flags |= RTWN_FLAG_CCK_HIPWR; } int r92c_init_rf_chain(struct rtwn_softc *sc, const struct rtwn_rf_prog *rf_prog, int chain) { int i, j; RTWN_DPRINTF(sc, RTWN_DEBUG_RESET, "%s: chain %d\n", __func__, chain); for (i = 0; rf_prog[i].reg != NULL; i++) { const struct rtwn_rf_prog *prog = &rf_prog[i]; while (!rtwn_check_condition(sc, prog->cond)) { KASSERT(prog->next != NULL, ("%s: wrong condition value (i %d)\n", __func__, i)); prog = prog->next; } for (j = 0; j < prog->count; j++) { RTWN_DPRINTF(sc, RTWN_DEBUG_RESET, "RF: reg 0x%02x, val 0x%05x\n", prog->reg[j], prog->val[j]); /* * These are fake RF registers offsets that * indicate a delay is required. */ /* NB: we are using 'value' to store required delay. */ if (prog->reg[j] > 0xf8) { rtwn_delay(sc, prog->val[j]); continue; } rtwn_rf_write(sc, chain, prog->reg[j], prog->val[j]); rtwn_delay(sc, 1); } } return (i); } void r92c_init_rf(struct rtwn_softc *sc) { struct r92c_softc *rs = sc->sc_priv; uint32_t reg, type; int i, chain, idx, off; for (chain = 0, i = 0; chain < sc->nrxchains; chain++, i++) { /* Save RF_ENV control type. */ idx = chain / 2; off = (chain % 2) * 16; reg = rtwn_bb_read(sc, R92C_FPGA0_RFIFACESW(idx)); type = (reg >> off) & 0x10; /* Set RF_ENV enable. */ rtwn_bb_setbits(sc, R92C_FPGA0_RFIFACEOE(chain), 0, 0x100000); rtwn_delay(sc, 1); /* Set RF_ENV output high. */ rtwn_bb_setbits(sc, R92C_FPGA0_RFIFACEOE(chain), 0, 0x10); rtwn_delay(sc, 1); /* Set address and data lengths of RF registers. */ rtwn_bb_setbits(sc, R92C_HSSI_PARAM2(chain), R92C_HSSI_PARAM2_ADDR_LENGTH, 0); rtwn_delay(sc, 1); rtwn_bb_setbits(sc, R92C_HSSI_PARAM2(chain), R92C_HSSI_PARAM2_DATA_LENGTH, 0); rtwn_delay(sc, 1); /* Write RF initialization values for this chain. */ i += r92c_init_rf_chain(sc, &sc->rf_prog[i], chain); /* Restore RF_ENV control type. */ rtwn_bb_setbits(sc, R92C_FPGA0_RFIFACESW(idx), 0x10 << off, type << off); /* Cache RF register CHNLBW. */ rs->rf_chnlbw[chain] = rtwn_rf_read(sc, chain, R92C_RF_CHNLBW); } if ((rs->chip & (R92C_CHIP_UMC_A_CUT | R92C_CHIP_92C)) == R92C_CHIP_UMC_A_CUT) { rtwn_rf_write(sc, 0, R92C_RF_RX_G1, 0x30255); rtwn_rf_write(sc, 0, R92C_RF_RX_G2, 0x50a00); } /* Turn CCK and OFDM blocks on. */ rtwn_bb_setbits(sc, R92C_FPGA0_RFMOD, 0, R92C_RFMOD_CCK_EN); rtwn_bb_setbits(sc, R92C_FPGA0_RFMOD, 0, R92C_RFMOD_OFDM_EN); } void r92c_init_edca(struct rtwn_softc *sc) { /* SIFS */ rtwn_write_2(sc, R92C_SPEC_SIFS, 0x100a); rtwn_write_2(sc, R92C_MAC_SPEC_SIFS, 0x100a); rtwn_write_2(sc, R92C_SIFS_CCK, 0x100a); rtwn_write_2(sc, R92C_SIFS_OFDM, 0x100a); /* TXOP */ rtwn_write_4(sc, R92C_EDCA_BE_PARAM, 0x005ea42b); rtwn_write_4(sc, R92C_EDCA_BK_PARAM, 0x0000a44f); rtwn_write_4(sc, R92C_EDCA_VI_PARAM, 0x005ea324); rtwn_write_4(sc, R92C_EDCA_VO_PARAM, 0x002fa226); } void r92c_init_ampdu(struct rtwn_softc *sc) { /* Setup AMPDU aggregation. */ rtwn_write_4(sc, R92C_AGGLEN_LMT, 0x99997631); /* MCS7~0 */ rtwn_write_1(sc, R92C_AGGR_BREAK_TIME, 0x16); rtwn_write_2(sc, R92C_MAX_AGGR_NUM, 0x0708); } void r92c_init_antsel(struct rtwn_softc *sc) { uint32_t reg; if (sc->ntxchains != 1 || sc->nrxchains != 1) return; rtwn_setbits_1(sc, R92C_LEDCFG2, 0, 0x80); rtwn_bb_setbits(sc, R92C_FPGA0_RFPARAM(0), 0, 0x2000); reg = rtwn_bb_read(sc, R92C_FPGA0_RFIFACEOE(0)); sc->sc_ant = MS(reg, R92C_FPGA0_RFIFACEOE0_ANT); /* XXX */ } void r92c_pa_bias_init(struct rtwn_softc *sc) { struct r92c_softc *rs = sc->sc_priv; int i; for (i = 0; i < sc->nrxchains; i++) { if (rs->pa_setting & (1 << i)) continue; r92c_rf_write(sc, i, R92C_RF_IPA, 0x0f406); r92c_rf_write(sc, i, R92C_RF_IPA, 0x4f406); r92c_rf_write(sc, i, R92C_RF_IPA, 0x8f406); r92c_rf_write(sc, i, R92C_RF_IPA, 0xcf406); } if (!(rs->pa_setting & 0x10)) rtwn_setbits_1(sc, 0x16, 0xf0, 0x90); } Index: projects/clang400-import/sys/dev/rtwn/rtl8192c/usb/r92cu_init.c =================================================================== --- projects/clang400-import/sys/dev/rtwn/rtl8192c/usb/r92cu_init.c (revision 311459) +++ projects/clang400-import/sys/dev/rtwn/rtl8192c/usb/r92cu_init.c (revision 311460) @@ -1,395 +1,391 @@ /* $OpenBSD: if_urtwn.c,v 1.16 2011/02/10 17:26:40 jakemsr Exp $ */ /*- * Copyright (c) 2010 Damien Bergamini * Copyright (c) 2014 Kevin Lo * Copyright (c) 2015-2016 Andriy Voskoboinyk * * 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 "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 void r92cu_init_bb(struct rtwn_softc *sc) { /* Enable BB and RF. */ rtwn_setbits_2(sc, R92C_SYS_FUNC_EN, 0, R92C_SYS_FUNC_EN_BBRSTB | R92C_SYS_FUNC_EN_BB_GLB_RST | R92C_SYS_FUNC_EN_DIO_RF); rtwn_write_2(sc, R92C_AFE_PLL_CTRL, 0xdb83); rtwn_write_1(sc, R92C_RF_CTRL, R92C_RF_CTRL_EN | R92C_RF_CTRL_RSTB | R92C_RF_CTRL_SDMRSTB); rtwn_write_1(sc, R92C_SYS_FUNC_EN, R92C_SYS_FUNC_EN_USBA | R92C_SYS_FUNC_EN_USBD | R92C_SYS_FUNC_EN_BB_GLB_RST | R92C_SYS_FUNC_EN_BBRSTB); rtwn_write_1(sc, R92C_LDOHCI12_CTRL, 0x0f); rtwn_write_1(sc, 0x15, 0xe9); rtwn_write_1(sc, R92C_AFE_XTAL_CTRL + 1, 0x80); r92c_init_bb_common(sc); - - if (rtwn_bb_read(sc, R92C_HSSI_PARAM2(0)) & - R92C_HSSI_PARAM2_CCK_HIPWR) - sc->sc_flags |= RTWN_FLAG_CCK_HIPWR; } int r92cu_power_on(struct rtwn_softc *sc) { #define RTWN_CHK(res) do { \ if (res != 0) \ return (EIO); \ } while(0) uint32_t reg; int ntries; /* Wait for autoload done bit. */ for (ntries = 0; ntries < 5000; ntries++) { if (rtwn_read_1(sc, R92C_APS_FSMCO) & R92C_APS_FSMCO_PFM_ALDN) break; rtwn_delay(sc, 10); } if (ntries == 5000) { device_printf(sc->sc_dev, "timeout waiting for chip autoload\n"); return (ETIMEDOUT); } /* Unlock ISO/CLK/Power control register. */ RTWN_CHK(rtwn_write_1(sc, R92C_RSV_CTRL, 0)); /* Move SPS into PWM mode. */ RTWN_CHK(rtwn_write_1(sc, R92C_SPS0_CTRL, 0x2b)); /* just in case if power_off() was not properly executed. */ rtwn_delay(sc, 100); reg = rtwn_read_1(sc, R92C_LDOV12D_CTRL); if (!(reg & R92C_LDOV12D_CTRL_LDV12_EN)) { RTWN_CHK(rtwn_write_1(sc, R92C_LDOV12D_CTRL, reg | R92C_LDOV12D_CTRL_LDV12_EN)); rtwn_delay(sc, 100); RTWN_CHK(rtwn_setbits_1(sc, R92C_SYS_ISO_CTRL, R92C_SYS_ISO_CTRL_MD2PP, 0)); } /* Auto enable WLAN. */ RTWN_CHK(rtwn_setbits_1_shift(sc, R92C_APS_FSMCO, 0, R92C_APS_FSMCO_APFM_ONMAC, 1)); for (ntries = 0; ntries < 5000; ntries++) { if (!(rtwn_read_2(sc, R92C_APS_FSMCO) & R92C_APS_FSMCO_APFM_ONMAC)) break; rtwn_delay(sc, 10); } if (ntries == 5000) { device_printf(sc->sc_dev, "timeout waiting for MAC auto ON\n"); return (ETIMEDOUT); } /* Enable radio, GPIO and LED functions. */ RTWN_CHK(rtwn_write_2(sc, R92C_APS_FSMCO, R92C_APS_FSMCO_AFSM_HSUS | R92C_APS_FSMCO_PDN_EN | R92C_APS_FSMCO_PFM_ALDN)); /* Release RF digital isolation. */ RTWN_CHK(rtwn_setbits_1_shift(sc, R92C_SYS_ISO_CTRL, R92C_SYS_ISO_CTRL_DIOR, 0, 1)); /* Initialize MAC. */ RTWN_CHK(rtwn_setbits_1(sc, R92C_APSD_CTRL, R92C_APSD_CTRL_OFF, 0)); for (ntries = 0; ntries < 1000; ntries++) { if (!(rtwn_read_1(sc, R92C_APSD_CTRL) & R92C_APSD_CTRL_OFF_STATUS)) break; rtwn_delay(sc, 50); } if (ntries == 1000) { device_printf(sc->sc_dev, "timeout waiting for MAC initialization\n"); return (ETIMEDOUT); } /* Enable MAC DMA/WMAC/SCHEDULE/SEC blocks. */ RTWN_CHK(rtwn_setbits_2(sc, R92C_CR, 0, R92C_CR_HCI_TXDMA_EN | R92C_CR_TXDMA_EN | R92C_CR_HCI_RXDMA_EN | R92C_CR_RXDMA_EN | R92C_CR_PROTOCOL_EN | R92C_CR_SCHEDULE_EN | ((sc->sc_hwcrypto != RTWN_CRYPTO_SW) ? R92C_CR_ENSEC : 0) | R92C_CR_CALTMR_EN)); RTWN_CHK(rtwn_write_1(sc, 0xfe10, 0x19)); return (0); #undef RTWN_CHK } void r92cu_power_off(struct rtwn_softc *sc) { #ifndef RTWN_WITHOUT_UCODE struct r92c_softc *rs = sc->sc_priv; #endif uint32_t reg; int error; /* Deinit C2H event handler. */ #ifndef RTWN_WITHOUT_UCODE callout_stop(&rs->rs_c2h_report); rs->rs_c2h_paused = 0; rs->rs_c2h_pending = 0; rs->rs_c2h_timeout = hz; #endif /* Block all Tx queues. */ error = rtwn_write_1(sc, R92C_TXPAUSE, R92C_TX_QUEUE_ALL); if (error == ENXIO) /* hardware gone */ return; /* Disable RF */ rtwn_rf_write(sc, 0, 0, 0); rtwn_write_1(sc, R92C_APSD_CTRL, R92C_APSD_CTRL_OFF); /* Reset BB state machine */ rtwn_write_1(sc, R92C_SYS_FUNC_EN, R92C_SYS_FUNC_EN_USBD | R92C_SYS_FUNC_EN_USBA | R92C_SYS_FUNC_EN_BB_GLB_RST); rtwn_write_1(sc, R92C_SYS_FUNC_EN, R92C_SYS_FUNC_EN_USBD | R92C_SYS_FUNC_EN_USBA); /* * Reset digital sequence */ #ifndef RTWN_WITHOUT_UCODE if (rtwn_read_1(sc, R92C_MCUFWDL) & R92C_MCUFWDL_RDY) { /* Reset MCU ready status */ rtwn_write_1(sc, R92C_MCUFWDL, 0); /* If firmware in ram code, do reset */ r92c_fw_reset(sc, RTWN_FW_RESET_SHUTDOWN); } #endif /* Reset MAC and Enable 8051 */ rtwn_write_1(sc, R92C_SYS_FUNC_EN + 1, (R92C_SYS_FUNC_EN_CPUEN | R92C_SYS_FUNC_EN_ELDR | R92C_SYS_FUNC_EN_HWPDN) >> 8); /* Reset MCU ready status */ rtwn_write_1(sc, R92C_MCUFWDL, 0); /* Disable MAC clock */ rtwn_write_2(sc, R92C_SYS_CLKR, R92C_SYS_CLKR_ANAD16V_EN | R92C_SYS_CLKR_ANA8M | R92C_SYS_CLKR_LOADER_EN | R92C_SYS_CLKR_80M_SSC_DIS | R92C_SYS_CLKR_SYS_EN | R92C_SYS_CLKR_RING_EN | 0x4000); /* Disable AFE PLL */ rtwn_write_1(sc, R92C_AFE_PLL_CTRL, 0x80); /* Gated AFE DIG_CLOCK */ rtwn_write_2(sc, R92C_AFE_XTAL_CTRL, 0x880F); /* Isolated digital to PON */ rtwn_write_1(sc, R92C_SYS_ISO_CTRL, R92C_SYS_ISO_CTRL_MD2PP | R92C_SYS_ISO_CTRL_PA2PCIE | R92C_SYS_ISO_CTRL_PD2CORE | R92C_SYS_ISO_CTRL_IP2MAC | R92C_SYS_ISO_CTRL_DIOP | R92C_SYS_ISO_CTRL_DIOE); /* * Pull GPIO PIN to balance level and LED control */ /* 1. Disable GPIO[7:0] */ rtwn_write_2(sc, R92C_GPIO_IOSEL, 0x0000); reg = rtwn_read_4(sc, R92C_GPIO_PIN_CTRL) & ~0x0000ff00; reg |= ((reg << 8) & 0x0000ff00) | 0x00ff0000; rtwn_write_4(sc, R92C_GPIO_PIN_CTRL, reg); /* Disable GPIO[10:8] */ rtwn_write_1(sc, R92C_MAC_PINMUX_CFG, 0x00); reg = rtwn_read_2(sc, R92C_GPIO_IO_SEL) & ~0x00f0; reg |= (((reg & 0x000f) << 4) | 0x0780); rtwn_write_2(sc, R92C_GPIO_IO_SEL, reg); /* Disable LED0 & 1 */ rtwn_write_2(sc, R92C_LEDCFG0, 0x8080); /* * Reset digital sequence */ /* Disable ELDR clock */ rtwn_write_2(sc, R92C_SYS_CLKR, R92C_SYS_CLKR_ANAD16V_EN | R92C_SYS_CLKR_ANA8M | R92C_SYS_CLKR_LOADER_EN | R92C_SYS_CLKR_80M_SSC_DIS | R92C_SYS_CLKR_SYS_EN | R92C_SYS_CLKR_RING_EN | 0x4000); /* Isolated ELDR to PON */ rtwn_write_1(sc, R92C_SYS_ISO_CTRL + 1, (R92C_SYS_ISO_CTRL_DIOR | R92C_SYS_ISO_CTRL_PWC_EV12V) >> 8); /* * Disable analog sequence */ /* Disable A15 power */ rtwn_write_1(sc, R92C_LDOA15_CTRL, R92C_LDOA15_CTRL_OBUF); /* Disable digital core power */ rtwn_setbits_1(sc, R92C_LDOV12D_CTRL, R92C_LDOV12D_CTRL_LDV12_EN, 0); /* Enter PFM mode */ rtwn_write_1(sc, R92C_SPS0_CTRL, 0x23); /* Set USB suspend */ rtwn_write_2(sc, R92C_APS_FSMCO, R92C_APS_FSMCO_APDM_HOST | R92C_APS_FSMCO_AFSM_HSUS | R92C_APS_FSMCO_PFM_ALDN); /* Lock ISO/CLK/Power control register. */ rtwn_write_1(sc, R92C_RSV_CTRL, 0x0E); } void r92cu_init_intr(struct rtwn_softc *sc) { rtwn_write_4(sc, R92C_HISR, 0xffffffff); rtwn_write_4(sc, R92C_HIMR, 0xffffffff); } void r92cu_init_tx_agg(struct rtwn_softc *sc) { struct rtwn_usb_softc *uc = RTWN_USB_SOFTC(sc); uint32_t reg; reg = rtwn_read_4(sc, R92C_TDECTRL); reg = RW(reg, R92C_TDECTRL_BLK_DESC_NUM, uc->tx_agg_desc_num); rtwn_write_4(sc, R92C_TDECTRL, reg); } void r92cu_init_rx_agg(struct rtwn_softc *sc) { /* Rx aggregation (DMA & USB). */ rtwn_setbits_1(sc, R92C_TRXDMA_CTRL, 0, R92C_TRXDMA_CTRL_RXDMA_AGG_EN); rtwn_setbits_1(sc, R92C_USB_SPECIAL_OPTION, 0, R92C_USB_SPECIAL_OPTION_AGG_EN); /* XXX dehardcode */ rtwn_write_1(sc, R92C_RXDMA_AGG_PG_TH, 48); rtwn_write_1(sc, R92C_USB_DMA_AGG_TO, 4); rtwn_write_1(sc, R92C_USB_AGG_TH, 8); rtwn_write_1(sc, R92C_USB_AGG_TO, 6); } void r92cu_post_init(struct rtwn_softc *sc) { /* Perform LO and IQ calibrations. */ r92c_iq_calib(sc); /* Perform LC calibration. */ r92c_lc_calib(sc); /* Fix USB interference issue. */ rtwn_write_1(sc, 0xfe40, 0xe0); rtwn_write_1(sc, 0xfe41, 0x8d); rtwn_write_1(sc, 0xfe42, 0x80); r92c_pa_bias_init(sc); /* Fix for lower temperature. */ rtwn_write_1(sc, 0x15, 0xe9); #ifndef RTWN_WITHOUT_UCODE if (sc->sc_flags & RTWN_FW_LOADED) { struct r92c_softc *rs = sc->sc_priv; if (sc->sc_ratectl_sysctl == RTWN_RATECTL_FW) { /* XXX firmware RA does not work yet */ sc->sc_ratectl = RTWN_RATECTL_NET80211; } else sc->sc_ratectl = sc->sc_ratectl_sysctl; /* Start C2H event handling. */ callout_reset(&rs->rs_c2h_report, rs->rs_c2h_timeout, r92c_handle_c2h_report, sc); } else #endif sc->sc_ratectl = RTWN_RATECTL_NONE; } Index: projects/clang400-import/sys/dev/twa/tw_osl_cam.c =================================================================== --- projects/clang400-import/sys/dev/twa/tw_osl_cam.c (revision 311459) +++ projects/clang400-import/sys/dev/twa/tw_osl_cam.c (revision 311460) @@ -1,686 +1,686 @@ /* * Copyright (c) 2004-07 Applied Micro Circuits Corporation. * Copyright (c) 2004-05 Vinod Kashyap. * 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. * * $FreeBSD$ */ /* * AMCC'S 3ware driver for 9000 series storage controllers. * * Author: Vinod Kashyap * Modifications by: Adam Radford * Modifications by: Manjunath Ranganathaiah */ /* * FreeBSD CAM related functions. */ #include #include #include #include #include #include #include #include #include static TW_VOID twa_action(struct cam_sim *sim, union ccb *ccb); static TW_VOID twa_poll(struct cam_sim *sim); static TW_INT32 tw_osli_execute_scsi(struct tw_osli_req_context *req, union ccb *ccb); /* * Function name: tw_osli_cam_attach * Description: Attaches the driver to CAM. * * Input: sc -- ptr to OSL internal ctlr context * Output: None * Return value: 0 -- success * non-zero-- failure */ TW_INT32 tw_osli_cam_attach(struct twa_softc *sc) { struct cam_devq *devq; tw_osli_dbg_dprintf(3, sc, "entered"); /* * Create the device queue for our SIM. */ if ((devq = cam_simq_alloc(TW_OSLI_MAX_NUM_IOS)) == NULL) { tw_osli_printf(sc, "error = %d", TW_CL_SEVERITY_ERROR_STRING, TW_CL_MESSAGE_SOURCE_FREEBSD_DRIVER, 0x2100, "Failed to create SIM device queue", ENOMEM); return(ENOMEM); } /* * Create a SIM entry. Though we can support TW_OSLI_MAX_NUM_REQUESTS * simultaneous requests, we claim to be able to handle only * TW_OSLI_MAX_NUM_IOS (two less), so that we always have a request * packet available to service ioctls and AENs. */ tw_osli_dbg_dprintf(3, sc, "Calling cam_sim_alloc"); sc->sim = cam_sim_alloc(twa_action, twa_poll, "twa", sc, device_get_unit(sc->bus_dev), sc->sim_lock, TW_OSLI_MAX_NUM_IOS, 1, devq); if (sc->sim == NULL) { cam_simq_free(devq); tw_osli_printf(sc, "error = %d", TW_CL_SEVERITY_ERROR_STRING, TW_CL_MESSAGE_SOURCE_FREEBSD_DRIVER, 0x2101, "Failed to create a SIM entry", ENOMEM); return(ENOMEM); } /* * Register the bus. */ tw_osli_dbg_dprintf(3, sc, "Calling xpt_bus_register"); mtx_lock(sc->sim_lock); if (xpt_bus_register(sc->sim, sc->bus_dev, 0) != CAM_SUCCESS) { cam_sim_free(sc->sim, TRUE); sc->sim = NULL; /* so cam_detach will not try to free it */ tw_osli_printf(sc, "error = %d", TW_CL_SEVERITY_ERROR_STRING, TW_CL_MESSAGE_SOURCE_FREEBSD_DRIVER, 0x2102, "Failed to register the bus", ENXIO); mtx_unlock(sc->sim_lock); return(ENXIO); } tw_osli_dbg_dprintf(3, sc, "Calling xpt_create_path"); if (xpt_create_path(&sc->path, NULL, cam_sim_path(sc->sim), CAM_TARGET_WILDCARD, CAM_LUN_WILDCARD) != CAM_REQ_CMP) { xpt_bus_deregister(cam_sim_path (sc->sim)); /* Passing TRUE to cam_sim_free will free the devq as well. */ cam_sim_free(sc->sim, TRUE); tw_osli_printf(sc, "error = %d", TW_CL_SEVERITY_ERROR_STRING, TW_CL_MESSAGE_SOURCE_FREEBSD_DRIVER, 0x2103, "Failed to create path", ENXIO); mtx_unlock(sc->sim_lock); return(ENXIO); } mtx_unlock(sc->sim_lock); tw_osli_dbg_dprintf(3, sc, "exiting"); return(0); } /* * Function name: tw_osli_cam_detach * Description: Detaches the driver from CAM. * * Input: sc -- ptr to OSL internal ctlr context * Output: None * Return value: None */ TW_VOID tw_osli_cam_detach(struct twa_softc *sc) { tw_osli_dbg_dprintf(3, sc, "entered"); mtx_lock(sc->sim_lock); if (sc->path) xpt_free_path(sc->path); if (sc->sim) { xpt_bus_deregister(cam_sim_path(sc->sim)); /* Passing TRUE to cam_sim_free will free the devq as well. */ cam_sim_free(sc->sim, TRUE); } /* It's ok have 1 hold count while destroying the mutex */ mtx_destroy(sc->sim_lock); } /* * Function name: tw_osli_execute_scsi * Description: Build a fw cmd, based on a CAM style ccb, and * send it down. * * Input: req -- ptr to OSL internal request context * ccb -- ptr to CAM style ccb * Output: None * Return value: 0 -- success * non-zero-- failure */ TW_INT32 tw_osli_execute_scsi(struct tw_osli_req_context *req, union ccb *ccb) { struct twa_softc *sc = req->ctlr; struct tw_cl_req_packet *req_pkt; struct tw_cl_scsi_req_packet *scsi_req; struct ccb_hdr *ccb_h = &(ccb->ccb_h); struct ccb_scsiio *csio = &(ccb->csio); TW_INT32 error; tw_osli_dbg_dprintf(10, sc, "SCSI I/O request 0x%x", csio->cdb_io.cdb_bytes[0]); if (ccb_h->target_id >= TW_CL_MAX_NUM_UNITS) { tw_osli_dbg_dprintf(3, sc, "Invalid target. PTL = %x %x %jx", ccb_h->path_id, ccb_h->target_id, (uintmax_t)ccb_h->target_lun); ccb_h->status |= CAM_TID_INVALID; xpt_done(ccb); return(1); } if (ccb_h->target_lun >= TW_CL_MAX_NUM_LUNS) { tw_osli_dbg_dprintf(3, sc, "Invalid lun. PTL = %x %x %jx", ccb_h->path_id, ccb_h->target_id, (uintmax_t)ccb_h->target_lun); ccb_h->status |= CAM_LUN_INVALID; xpt_done(ccb); return(1); } if(ccb_h->flags & CAM_CDB_PHYS) { tw_osli_printf(sc, "", TW_CL_SEVERITY_ERROR_STRING, TW_CL_MESSAGE_SOURCE_FREEBSD_DRIVER, 0x2105, "Physical CDB address!"); ccb_h->status = CAM_REQ_INVALID; xpt_done(ccb); return(1); } /* * We are going to work on this request. Mark it as enqueued (though * we don't actually queue it...) */ ccb_h->status |= CAM_SIM_QUEUED; if((ccb_h->flags & CAM_DIR_MASK) != CAM_DIR_NONE) { if(ccb_h->flags & CAM_DIR_IN) req->flags |= TW_OSLI_REQ_FLAGS_DATA_IN; else req->flags |= TW_OSLI_REQ_FLAGS_DATA_OUT; } /* Build the CL understood request packet for SCSI cmds. */ req_pkt = &req->req_pkt; req_pkt->status = 0; req_pkt->tw_osl_callback = tw_osl_complete_io; scsi_req = &(req_pkt->gen_req_pkt.scsi_req); scsi_req->unit = ccb_h->target_id; scsi_req->lun = ccb_h->target_lun; scsi_req->sense_len = 0; scsi_req->sense_data = (TW_UINT8 *)(&csio->sense_data); scsi_req->scsi_status = 0; if(ccb_h->flags & CAM_CDB_POINTER) scsi_req->cdb = csio->cdb_io.cdb_ptr; else scsi_req->cdb = csio->cdb_io.cdb_bytes; scsi_req->cdb_len = csio->cdb_len; if (csio->dxfer_len > TW_CL_MAX_IO_SIZE) { tw_osli_printf(sc, "size = %d", TW_CL_SEVERITY_ERROR_STRING, TW_CL_MESSAGE_SOURCE_FREEBSD_DRIVER, 0x2106, "I/O size too big", csio->dxfer_len); ccb_h->status = CAM_REQ_TOO_BIG; ccb_h->status &= ~CAM_SIM_QUEUED; xpt_done(ccb); return(1); } if ((ccb_h->flags & CAM_DATA_MASK) == CAM_DATA_VADDR) { if ((req->length = csio->dxfer_len) != 0) { req->data = csio->data_ptr; scsi_req->sgl_entries = 1; } } else req->flags |= TW_OSLI_REQ_FLAGS_CCB; req->deadline = tw_osl_get_local_time() + (ccb_h->timeout / 1000); /* * twa_map_load_data_callback will fill in the SGL, * and submit the I/O. */ error = tw_osli_map_request(req); if ((error) && (req->flags & TW_OSLI_REQ_FLAGS_FAILED)) { req->deadline = 0; ccb_h->status = CAM_REQ_CMP_ERR; ccb_h->status &= ~CAM_SIM_QUEUED; xpt_done(ccb); } return(error); } /* * Function name: twa_action * Description: Driver entry point for CAM's use. * * Input: sim -- sim corresponding to the ctlr * ccb -- ptr to CAM request * Output: None * Return value: None */ TW_VOID twa_action(struct cam_sim *sim, union ccb *ccb) { struct twa_softc *sc = (struct twa_softc *)cam_sim_softc(sim); struct ccb_hdr *ccb_h = &(ccb->ccb_h); switch (ccb_h->func_code) { case XPT_SCSI_IO: /* SCSI I/O */ { struct tw_osli_req_context *req; req = tw_osli_get_request(sc); if (req == NULL) { tw_osli_dbg_dprintf(2, sc, "Cannot get request pkt."); /* * Freeze the simq to maintain ccb ordering. The next * ccb that gets completed will unfreeze the simq. */ ccb_h->status &= ~CAM_SIM_QUEUED; ccb_h->status |= CAM_REQUEUE_REQ; xpt_done(ccb); break; } if ((tw_cl_is_reset_needed(&(req->ctlr->ctlr_handle)))) { ccb_h->status &= ~CAM_SIM_QUEUED; ccb_h->status |= CAM_REQUEUE_REQ; xpt_done(ccb); tw_osli_req_q_insert_tail(req, TW_OSLI_FREE_Q); break; } req->req_handle.osl_req_ctxt = req; req->req_handle.is_io = TW_CL_TRUE; req->orig_req = ccb; if (tw_osli_execute_scsi(req, ccb)) tw_osli_req_q_insert_tail(req, TW_OSLI_FREE_Q); break; } case XPT_ABORT: tw_osli_dbg_dprintf(2, sc, "Abort request."); ccb_h->status = CAM_UA_ABORT; xpt_done(ccb); break; case XPT_RESET_BUS: tw_cl_create_event(&(sc->ctlr_handle), TW_CL_FALSE, TW_CL_MESSAGE_SOURCE_FREEBSD_DRIVER, 0x2108, 0x3, TW_CL_SEVERITY_INFO_STRING, "Received Reset Bus request from CAM", " "); tw_cl_set_reset_needed(&(sc->ctlr_handle)); ccb_h->status = CAM_REQ_CMP; xpt_done(ccb); break; case XPT_SET_TRAN_SETTINGS: tw_osli_dbg_dprintf(3, sc, "XPT_SET_TRAN_SETTINGS"); /* * This command is not supported, since it's very specific * to SCSI, and we are doing ATA. */ ccb_h->status = CAM_FUNC_NOTAVAIL; xpt_done(ccb); break; case XPT_GET_TRAN_SETTINGS: { struct ccb_trans_settings *cts = &ccb->cts; struct ccb_trans_settings_scsi *scsi = &cts->proto_specific.scsi; struct ccb_trans_settings_spi *spi = &cts->xport_specific.spi; cts->protocol = PROTO_SCSI; cts->protocol_version = SCSI_REV_2; cts->transport = XPORT_SPI; cts->transport_version = 2; spi->valid = CTS_SPI_VALID_DISC; spi->flags = CTS_SPI_FLAGS_DISC_ENB; scsi->valid = CTS_SCSI_VALID_TQ; scsi->flags = CTS_SCSI_FLAGS_TAG_ENB; tw_osli_dbg_dprintf(3, sc, "XPT_GET_TRAN_SETTINGS"); ccb_h->status = CAM_REQ_CMP; xpt_done(ccb); break; } case XPT_CALC_GEOMETRY: tw_osli_dbg_dprintf(3, sc, "XPT_CALC_GEOMETRY"); cam_calc_geometry(&ccb->ccg, 1/* extended */); xpt_done(ccb); break; case XPT_PATH_INQ: /* Path inquiry -- get twa properties */ { struct ccb_pathinq *path_inq = &ccb->cpi; tw_osli_dbg_dprintf(3, sc, "XPT_PATH_INQ request"); path_inq->version_num = 1; path_inq->hba_inquiry = 0; path_inq->target_sprt = 0; path_inq->hba_misc = 0; path_inq->hba_eng_cnt = 0; path_inq->max_target = TW_CL_MAX_NUM_UNITS; path_inq->max_lun = TW_CL_MAX_NUM_LUNS - 1; path_inq->unit_number = cam_sim_unit(sim); path_inq->bus_id = cam_sim_bus(sim); path_inq->initiator_id = TW_CL_MAX_NUM_UNITS; path_inq->base_transfer_speed = 100000; strlcpy(path_inq->sim_vid, "FreeBSD", SIM_IDLEN); strlcpy(path_inq->hba_vid, "3ware", HBA_IDLEN); strlcpy(path_inq->dev_name, cam_sim_name(sim), DEV_IDLEN); - path_inq->transport = XPORT_SPI; - path_inq->transport_version = 2; - path_inq->protocol = PROTO_SCSI; - path_inq->protocol_version = SCSI_REV_2; - path_inq->maxio = TW_CL_MAX_IO_SIZE; + path_inq->transport = XPORT_SPI; + path_inq->transport_version = 2; + path_inq->protocol = PROTO_SCSI; + path_inq->protocol_version = SCSI_REV_2; + path_inq->maxio = TW_CL_MAX_IO_SIZE; ccb_h->status = CAM_REQ_CMP; xpt_done(ccb); break; } default: tw_osli_dbg_dprintf(3, sc, "func_code = %x", ccb_h->func_code); ccb_h->status = CAM_REQ_INVALID; xpt_done(ccb); break; } } /* * Function name: twa_poll * Description: Driver entry point called when interrupts are not * available. * * Input: sim -- sim corresponding to the controller * Output: None * Return value: None */ TW_VOID twa_poll(struct cam_sim *sim) { struct twa_softc *sc = (struct twa_softc *)(cam_sim_softc(sim)); tw_osli_dbg_dprintf(3, sc, "entering; sc = %p", sc); tw_cl_interrupt(&(sc->ctlr_handle)); tw_osli_dbg_dprintf(3, sc, "exiting; sc = %p", sc); } /* * Function name: tw_osli_request_bus_scan * Description: Requests CAM for a scan of the bus. * * Input: sc -- ptr to per ctlr structure * Output: None * Return value: 0 -- success * non-zero-- failure */ TW_INT32 tw_osli_request_bus_scan(struct twa_softc *sc) { union ccb *ccb; tw_osli_dbg_dprintf(3, sc, "entering"); /* If we get here before sc->sim is initialized, return an error. */ if (!(sc->sim)) return(ENXIO); if ((ccb = xpt_alloc_ccb()) == NULL) return(ENOMEM); mtx_lock(sc->sim_lock); if (xpt_create_path(&ccb->ccb_h.path, NULL, cam_sim_path(sc->sim), CAM_TARGET_WILDCARD, CAM_LUN_WILDCARD) != CAM_REQ_CMP) { xpt_free_ccb(ccb); mtx_unlock(sc->sim_lock); return(EIO); } xpt_rescan(ccb); mtx_unlock(sc->sim_lock); return(0); } /* * Function name: tw_osli_disallow_new_requests * Description: Calls the appropriate CAM function, so as to freeze * the flow of new requests from CAM to this controller. * * Input: sc -- ptr to OSL internal ctlr context * req_handle -- ptr to request handle sent by OSL. * Output: None * Return value: None */ TW_VOID tw_osli_disallow_new_requests(struct twa_softc *sc, struct tw_cl_req_handle *req_handle) { /* Only freeze/release the simq for IOs */ if (req_handle->is_io) { struct tw_osli_req_context *req = req_handle->osl_req_ctxt; union ccb *ccb = (union ccb *)(req->orig_req); xpt_freeze_simq(sc->sim, 1); ccb->ccb_h.status |= CAM_RELEASE_SIMQ; } } /* * Function name: tw_osl_timeout * Description: Call to timeout(). * * Input: req_handle -- ptr to request handle sent by OSL. * Output: None * Return value: None */ TW_VOID tw_osl_timeout(struct tw_cl_req_handle *req_handle) { struct tw_osli_req_context *req = req_handle->osl_req_ctxt; union ccb *ccb = (union ccb *)(req->orig_req); struct ccb_hdr *ccb_h = &(ccb->ccb_h); req->deadline = tw_osl_get_local_time() + (ccb_h->timeout / 1000); } /* * Function name: tw_osl_untimeout * Description: Inverse of call to timeout(). * * Input: req_handle -- ptr to request handle sent by OSL. * Output: None * Return value: None */ TW_VOID tw_osl_untimeout(struct tw_cl_req_handle *req_handle) { struct tw_osli_req_context *req = req_handle->osl_req_ctxt; req->deadline = 0; } /* * Function name: tw_osl_scan_bus * Description: CL calls this function to request for a bus scan. * * Input: ctlr_handle -- ptr to controller handle * Output: None * Return value: None */ TW_VOID tw_osl_scan_bus(struct tw_cl_ctlr_handle *ctlr_handle) { struct twa_softc *sc = ctlr_handle->osl_ctlr_ctxt; TW_INT32 error; if ((error = tw_osli_request_bus_scan(sc))) tw_osli_printf(sc, "error = %d", TW_CL_SEVERITY_ERROR_STRING, TW_CL_MESSAGE_SOURCE_FREEBSD_DRIVER, 0x2109, "Bus scan request to CAM failed", error); } /* * Function name: tw_osl_complete_io * Description: Called to complete CAM scsi requests. * * Input: req_handle -- ptr to request handle * Output: None * Return value: None */ TW_VOID tw_osl_complete_io(struct tw_cl_req_handle *req_handle) { struct tw_osli_req_context *req = req_handle->osl_req_ctxt; struct tw_cl_req_packet *req_pkt = (struct tw_cl_req_packet *)(&req->req_pkt); struct tw_cl_scsi_req_packet *scsi_req; struct twa_softc *sc = req->ctlr; union ccb *ccb = (union ccb *)(req->orig_req); tw_osli_dbg_dprintf(10, sc, "entering"); if (req->state != TW_OSLI_REQ_STATE_BUSY) tw_osli_printf(sc, "request = %p, status = %d", TW_CL_SEVERITY_ERROR_STRING, TW_CL_MESSAGE_SOURCE_FREEBSD_DRIVER, 0x210A, "Unposted command completed!!", req, req->state); /* * Remove request from the busy queue. Just mark it complete. * There's no need to move it into the complete queue as we are * going to be done with it right now. */ req->state = TW_OSLI_REQ_STATE_COMPLETE; tw_osli_req_q_remove_item(req, TW_OSLI_BUSY_Q); tw_osli_unmap_request(req); req->deadline = 0; if (req->error_code) { /* This request never got submitted to the firmware. */ if (req->error_code == EBUSY) { /* * Cmd queue is full, or the Common Layer is out of * resources. The simq will already have been frozen. * When this ccb gets completed will unfreeze the simq. */ ccb->ccb_h.status |= CAM_REQUEUE_REQ; } else if (req->error_code == EFBIG) ccb->ccb_h.status = CAM_REQ_TOO_BIG; else ccb->ccb_h.status = CAM_REQ_CMP_ERR; } else { scsi_req = &(req_pkt->gen_req_pkt.scsi_req); if (req_pkt->status == TW_CL_ERR_REQ_SUCCESS) ccb->ccb_h.status = CAM_REQ_CMP; else { if (req_pkt->status & TW_CL_ERR_REQ_INVALID_TARGET) ccb->ccb_h.status |= CAM_SEL_TIMEOUT; else if (req_pkt->status & TW_CL_ERR_REQ_INVALID_LUN) ccb->ccb_h.status |= CAM_DEV_NOT_THERE; else if (req_pkt->status & TW_CL_ERR_REQ_SCSI_ERROR) ccb->ccb_h.status |= CAM_SCSI_STATUS_ERROR; else if (req_pkt->status & TW_CL_ERR_REQ_BUS_RESET) ccb->ccb_h.status |= (CAM_REQUEUE_REQ | CAM_SCSI_BUS_RESET); /* * If none of the above errors occurred, simply * mark completion error. */ if (ccb->ccb_h.status == 0) ccb->ccb_h.status = CAM_REQ_CMP_ERR; if (req_pkt->status & TW_CL_ERR_REQ_AUTO_SENSE_VALID) { ccb->csio.sense_len = scsi_req->sense_len; ccb->ccb_h.status |= CAM_AUTOSNS_VALID; } } ccb->csio.scsi_status = scsi_req->scsi_status; } ccb->ccb_h.status &= ~CAM_SIM_QUEUED; mtx_lock(sc->sim_lock); xpt_done(ccb); mtx_unlock(sc->sim_lock); if (! req->error_code) /* twa_action will free the request otherwise */ tw_osli_req_q_insert_tail(req, TW_OSLI_FREE_Q); } Index: projects/clang400-import/sys/fs/nfs/nfs_commonsubs.c =================================================================== --- projects/clang400-import/sys/fs/nfs/nfs_commonsubs.c (revision 311459) +++ projects/clang400-import/sys/fs/nfs/nfs_commonsubs.c (revision 311460) @@ -1,4244 +1,4248 @@ /*- * 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("$FreeBSD$"); /* * These functions support the macros and help fiddle mbuf chains for * the nfs op functions. They do things like create the rpc header and * copy data between mbuf chains and uio lists. */ #ifndef APPLEKEXT #include "opt_inet6.h" #include #include /* * Data items converted to xdr at startup, since they are constant * This is kinda hokey, but may save a little time doing byte swaps */ u_int32_t newnfs_true, newnfs_false, newnfs_xdrneg1; /* And other global data */ nfstype nfsv34_type[9] = { NFNON, NFREG, NFDIR, NFBLK, NFCHR, NFLNK, NFSOCK, NFFIFO, NFNON }; enum vtype newnv2tov_type[8] = { VNON, VREG, VDIR, VBLK, VCHR, VLNK, VNON, VNON }; enum vtype nv34tov_type[8]={ VNON, VREG, VDIR, VBLK, VCHR, VLNK, VSOCK, VFIFO }; struct timeval nfsboottime; /* Copy boottime once, so it never changes */ int nfscl_ticks; int nfsrv_useacl = 1; struct nfssockreq nfsrv_nfsuserdsock; int nfsrv_nfsuserd = 0; struct nfsreqhead nfsd_reqq; uid_t nfsrv_defaultuid; gid_t nfsrv_defaultgid; int nfsrv_lease = NFSRV_LEASE; int ncl_mbuf_mlen = MLEN; int nfsd_enable_stringtouid = 0; NFSNAMEIDMUTEX; NFSSOCKMUTEX; extern int nfsrv_lughashsize; /* * This array of structures indicates, for V4: * retfh - which of 3 types of calling args are used * 0 - doesn't change cfh or use a sfh * 1 - replaces cfh with a new one (unless it returns an error status) * 2 - uses cfh and sfh * needscfh - if the op wants a cfh and premtime * 0 - doesn't use a cfh * 1 - uses a cfh, but doesn't want pre-op attributes * 2 - uses a cfh and wants pre-op attributes * savereply - indicates a non-idempotent Op * 0 - not non-idempotent * 1 - non-idempotent * Ops that are ordered via seqid# are handled separately from these * non-idempotent Ops. * Define it here, since it is used by both the client and server. */ struct nfsv4_opflag nfsv4_opflag[NFSV41_NOPS] = { { 0, 0, 0, 0, LK_EXCLUSIVE, 1, 1 }, /* undef */ { 0, 0, 0, 0, LK_EXCLUSIVE, 1, 1 }, /* undef */ { 0, 0, 0, 0, LK_EXCLUSIVE, 1, 1 }, /* undef */ { 0, 1, 0, 0, LK_SHARED, 1, 1 }, /* Access */ { 0, 1, 0, 0, LK_EXCLUSIVE, 1, 0 }, /* Close */ { 0, 2, 0, 1, LK_EXCLUSIVE, 1, 1 }, /* Commit */ { 1, 2, 1, 1, LK_EXCLUSIVE, 1, 1 }, /* Create */ { 0, 0, 0, 0, LK_EXCLUSIVE, 1, 0 }, /* Delegpurge */ { 0, 1, 0, 0, LK_EXCLUSIVE, 1, 0 }, /* Delegreturn */ { 0, 1, 0, 0, LK_SHARED, 1, 1 }, /* Getattr */ { 0, 1, 0, 0, LK_EXCLUSIVE, 1, 1 }, /* GetFH */ { 2, 1, 1, 1, LK_EXCLUSIVE, 1, 1 }, /* Link */ { 0, 1, 0, 0, LK_EXCLUSIVE, 1, 0 }, /* Lock */ { 0, 1, 0, 0, LK_EXCLUSIVE, 1, 0 }, /* LockT */ { 0, 1, 0, 0, LK_EXCLUSIVE, 1, 0 }, /* LockU */ { 1, 2, 0, 0, LK_EXCLUSIVE, 1, 1 }, /* Lookup */ { 1, 2, 0, 0, LK_EXCLUSIVE, 1, 1 }, /* Lookupp */ { 0, 1, 0, 0, LK_EXCLUSIVE, 1, 1 }, /* NVerify */ { 1, 1, 0, 1, LK_EXCLUSIVE, 1, 0 }, /* Open */ { 1, 1, 0, 0, LK_EXCLUSIVE, 1, 0 }, /* OpenAttr */ { 0, 1, 0, 0, LK_EXCLUSIVE, 1, 0 }, /* OpenConfirm */ { 0, 1, 0, 0, LK_EXCLUSIVE, 1, 0 }, /* OpenDowngrade */ { 1, 0, 0, 0, LK_EXCLUSIVE, 1, 1 }, /* PutFH */ { 1, 0, 0, 0, LK_EXCLUSIVE, 1, 1 }, /* PutPubFH */ { 1, 0, 0, 0, LK_EXCLUSIVE, 1, 1 }, /* PutRootFH */ { 0, 1, 0, 0, LK_SHARED, 1, 0 }, /* Read */ { 0, 1, 0, 0, LK_SHARED, 1, 1 }, /* Readdir */ { 0, 1, 0, 0, LK_SHARED, 1, 1 }, /* ReadLink */ { 0, 2, 1, 1, LK_EXCLUSIVE, 1, 1 }, /* Remove */ { 2, 1, 1, 1, LK_EXCLUSIVE, 1, 1 }, /* Rename */ { 0, 0, 0, 0, LK_EXCLUSIVE, 1, 0 }, /* Renew */ { 0, 0, 0, 0, LK_EXCLUSIVE, 1, 1 }, /* RestoreFH */ { 0, 1, 0, 0, LK_EXCLUSIVE, 1, 1 }, /* SaveFH */ { 0, 1, 0, 0, LK_EXCLUSIVE, 1, 1 }, /* SecInfo */ { 0, 2, 1, 1, LK_EXCLUSIVE, 1, 0 }, /* Setattr */ { 0, 0, 0, 0, LK_EXCLUSIVE, 1, 1 }, /* SetClientID */ { 0, 0, 0, 0, LK_EXCLUSIVE, 1, 1 }, /* SetClientIDConfirm */ { 0, 1, 0, 0, LK_EXCLUSIVE, 1, 1 }, /* Verify */ { 0, 2, 1, 1, LK_EXCLUSIVE, 1, 0 }, /* Write */ { 0, 0, 0, 0, LK_EXCLUSIVE, 1, 0 }, /* ReleaseLockOwner */ { 0, 0, 0, 0, LK_EXCLUSIVE, 1, 1 }, /* Backchannel Ctrl */ { 0, 0, 0, 0, LK_EXCLUSIVE, 1, 1 }, /* Bind Conn to Sess */ { 0, 0, 0, 0, LK_EXCLUSIVE, 0, 0 }, /* Exchange ID */ { 0, 0, 0, 0, LK_EXCLUSIVE, 0, 0 }, /* Create Session */ { 0, 0, 0, 0, LK_EXCLUSIVE, 0, 0 }, /* Destroy Session */ { 0, 0, 0, 0, LK_EXCLUSIVE, 1, 0 }, /* Free StateID */ { 0, 0, 0, 0, LK_EXCLUSIVE, 1, 1 }, /* Get Dir Deleg */ { 0, 0, 0, 0, LK_EXCLUSIVE, 1, 1 }, /* Get Device Info */ { 0, 0, 0, 0, LK_EXCLUSIVE, 1, 1 }, /* Get Device List */ { 0, 1, 0, 0, LK_EXCLUSIVE, 1, 1 }, /* Layout Commit */ { 0, 1, 0, 0, LK_EXCLUSIVE, 1, 1 }, /* Layout Get */ { 0, 1, 0, 0, LK_EXCLUSIVE, 1, 0 }, /* Layout Return */ { 0, 0, 0, 0, LK_EXCLUSIVE, 1, 1 }, /* Secinfo No name */ { 0, 0, 0, 0, LK_EXCLUSIVE, 1, 0 }, /* Sequence */ { 0, 0, 0, 0, LK_EXCLUSIVE, 1, 1 }, /* Set SSV */ { 0, 0, 0, 0, LK_EXCLUSIVE, 1, 1 }, /* Test StateID */ { 0, 0, 0, 0, LK_EXCLUSIVE, 1, 1 }, /* Want Delegation */ { 0, 0, 0, 0, LK_EXCLUSIVE, 0, 0 }, /* Destroy ClientID */ { 0, 0, 0, 0, LK_EXCLUSIVE, 1, 1 }, /* Reclaim Complete */ }; #endif /* !APPLEKEXT */ static int ncl_mbuf_mhlen = MHLEN; static int nfsrv_usercnt = 0; static int nfsrv_dnsnamelen; static u_char *nfsrv_dnsname = NULL; static int nfsrv_usermax = 999999999; struct nfsrv_lughash { struct mtx mtx; struct nfsuserhashhead lughead; }; static struct nfsrv_lughash *nfsuserhash; static struct nfsrv_lughash *nfsusernamehash; static struct nfsrv_lughash *nfsgrouphash; static struct nfsrv_lughash *nfsgroupnamehash; /* * This static array indicates whether or not the RPC generates a large * reply. This is used by nfs_reply() to decide whether or not an mbuf * cluster should be allocated. (If a cluster is required by an RPC * marked 0 in this array, the code will still work, just not quite as * efficiently.) */ int nfs_bigreply[NFSV41_NPROCS] = { 0, 0, 0, 1, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0 }; /* local functions */ static int nfsrv_skipace(struct nfsrv_descript *nd, int *acesizep); static void nfsv4_wanted(struct nfsv4lock *lp); static int nfsrv_cmpmixedcase(u_char *cp, u_char *cp2, int len); static int nfsrv_getuser(int procnum, uid_t uid, gid_t gid, char *name, NFSPROC_T *p); static void nfsrv_removeuser(struct nfsusrgrp *usrp, int isuser); static int nfsrv_getrefstr(struct nfsrv_descript *, u_char **, u_char **, int *, int *); static void nfsrv_refstrbigenough(int, u_char **, u_char **, int *); #ifndef APPLE /* * copies mbuf chain to the uio scatter/gather list */ int nfsm_mbufuio(struct nfsrv_descript *nd, struct uio *uiop, int siz) { char *mbufcp, *uiocp; int xfer, left, len; mbuf_t mp; long uiosiz, rem; int error = 0; mp = nd->nd_md; mbufcp = nd->nd_dpos; len = NFSMTOD(mp, caddr_t) + mbuf_len(mp) - mbufcp; rem = NFSM_RNDUP(siz) - siz; while (siz > 0) { if (uiop->uio_iovcnt <= 0 || uiop->uio_iov == NULL) { error = EBADRPC; goto out; } left = uiop->uio_iov->iov_len; uiocp = uiop->uio_iov->iov_base; if (left > siz) left = siz; uiosiz = left; while (left > 0) { while (len == 0) { mp = mbuf_next(mp); if (mp == NULL) { error = EBADRPC; goto out; } mbufcp = NFSMTOD(mp, caddr_t); len = mbuf_len(mp); KASSERT(len >= 0, ("len %d, corrupted mbuf?", len)); } xfer = (left > len) ? len : left; #ifdef notdef /* Not Yet.. */ if (uiop->uio_iov->iov_op != NULL) (*(uiop->uio_iov->iov_op)) (mbufcp, uiocp, xfer); else #endif if (uiop->uio_segflg == UIO_SYSSPACE) NFSBCOPY(mbufcp, uiocp, xfer); else copyout(mbufcp, CAST_USER_ADDR_T(uiocp), xfer); left -= xfer; len -= xfer; mbufcp += xfer; uiocp += xfer; uiop->uio_offset += xfer; uiop->uio_resid -= xfer; } if (uiop->uio_iov->iov_len <= siz) { uiop->uio_iovcnt--; uiop->uio_iov++; } else { uiop->uio_iov->iov_base = (void *) ((char *)uiop->uio_iov->iov_base + uiosiz); uiop->uio_iov->iov_len -= uiosiz; } siz -= uiosiz; } nd->nd_dpos = mbufcp; nd->nd_md = mp; if (rem > 0) { if (len < rem) error = nfsm_advance(nd, rem, len); else nd->nd_dpos += rem; } out: NFSEXITCODE2(error, nd); return (error); } #endif /* !APPLE */ /* * Help break down an mbuf chain by setting the first siz bytes contiguous * pointed to by returned val. * This is used by the macro NFSM_DISSECT for tough * cases. */ APPLESTATIC void * nfsm_dissct(struct nfsrv_descript *nd, int siz, int how) { mbuf_t mp2; int siz2, xfer; caddr_t p; int left; caddr_t retp; retp = NULL; left = NFSMTOD(nd->nd_md, caddr_t) + mbuf_len(nd->nd_md) - nd->nd_dpos; while (left == 0) { nd->nd_md = mbuf_next(nd->nd_md); if (nd->nd_md == NULL) return (retp); left = mbuf_len(nd->nd_md); nd->nd_dpos = NFSMTOD(nd->nd_md, caddr_t); } if (left >= siz) { retp = nd->nd_dpos; nd->nd_dpos += siz; } else if (mbuf_next(nd->nd_md) == NULL) { return (retp); } else if (siz > ncl_mbuf_mhlen) { panic("nfs S too big"); } else { MGET(mp2, MT_DATA, how); if (mp2 == NULL) return (NULL); mbuf_setnext(mp2, mbuf_next(nd->nd_md)); mbuf_setnext(nd->nd_md, mp2); mbuf_setlen(nd->nd_md, mbuf_len(nd->nd_md) - left); nd->nd_md = mp2; retp = p = NFSMTOD(mp2, caddr_t); NFSBCOPY(nd->nd_dpos, p, left); /* Copy what was left */ siz2 = siz - left; p += left; mp2 = mbuf_next(mp2); /* Loop around copying up the siz2 bytes */ while (siz2 > 0) { if (mp2 == NULL) return (NULL); xfer = (siz2 > mbuf_len(mp2)) ? mbuf_len(mp2) : siz2; if (xfer > 0) { NFSBCOPY(NFSMTOD(mp2, caddr_t), p, xfer); NFSM_DATAP(mp2, xfer); mbuf_setlen(mp2, mbuf_len(mp2) - xfer); p += xfer; siz2 -= xfer; } if (siz2 > 0) mp2 = mbuf_next(mp2); } mbuf_setlen(nd->nd_md, siz); nd->nd_md = mp2; nd->nd_dpos = NFSMTOD(mp2, caddr_t); } return (retp); } /* * Advance the position in the mbuf chain. * If offs == 0, this is a no-op, but it is simpler to just return from * here than check for offs > 0 for all calls to nfsm_advance. * If left == -1, it should be calculated here. */ APPLESTATIC int nfsm_advance(struct nfsrv_descript *nd, int offs, int left) { int error = 0; if (offs == 0) goto out; /* * A negative offs should be considered a serious problem. */ if (offs < 0) panic("nfsrv_advance"); /* * If left == -1, calculate it here. */ if (left == -1) left = NFSMTOD(nd->nd_md, caddr_t) + mbuf_len(nd->nd_md) - nd->nd_dpos; /* * Loop around, advancing over the mbuf data. */ while (offs > left) { offs -= left; nd->nd_md = mbuf_next(nd->nd_md); if (nd->nd_md == NULL) { error = EBADRPC; goto out; } left = mbuf_len(nd->nd_md); nd->nd_dpos = NFSMTOD(nd->nd_md, caddr_t); } nd->nd_dpos += offs; out: NFSEXITCODE(error); return (error); } /* * Copy a string into mbuf(s). * Return the number of bytes output, including XDR overheads. */ APPLESTATIC int nfsm_strtom(struct nfsrv_descript *nd, const char *cp, int siz) { mbuf_t m2; int xfer, left; mbuf_t m1; int rem, bytesize; u_int32_t *tl; char *cp2; NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(siz); rem = NFSM_RNDUP(siz) - siz; bytesize = NFSX_UNSIGNED + siz + rem; m2 = nd->nd_mb; cp2 = nd->nd_bpos; left = M_TRAILINGSPACE(m2); /* * Loop around copying the string to mbuf(s). */ while (siz > 0) { if (left == 0) { if (siz > ncl_mbuf_mlen) NFSMCLGET(m1, M_WAITOK); else NFSMGET(m1); mbuf_setlen(m1, 0); mbuf_setnext(m2, m1); m2 = m1; cp2 = NFSMTOD(m2, caddr_t); left = M_TRAILINGSPACE(m2); } if (left >= siz) xfer = siz; else xfer = left; NFSBCOPY(cp, cp2, xfer); cp += xfer; mbuf_setlen(m2, mbuf_len(m2) + xfer); siz -= xfer; left -= xfer; if (siz == 0 && rem) { if (left < rem) panic("nfsm_strtom"); NFSBZERO(cp2 + xfer, rem); mbuf_setlen(m2, mbuf_len(m2) + rem); } } nd->nd_mb = m2; nd->nd_bpos = NFSMTOD(m2, caddr_t) + mbuf_len(m2); return (bytesize); } /* * Called once to initialize data structures... */ APPLESTATIC void newnfs_init(void) { static int nfs_inited = 0; if (nfs_inited) return; nfs_inited = 1; newnfs_true = txdr_unsigned(TRUE); newnfs_false = txdr_unsigned(FALSE); newnfs_xdrneg1 = txdr_unsigned(-1); nfscl_ticks = (hz * NFS_TICKINTVL + 500) / 1000; if (nfscl_ticks < 1) nfscl_ticks = 1; NFSSETBOOTTIME(nfsboottime); /* * Initialize reply list and start timer */ TAILQ_INIT(&nfsd_reqq); NFS_TIMERINIT; } /* * Put a file handle in an mbuf list. * If the size argument == 0, just use the default size. * set_true == 1 if there should be an newnfs_true prepended on the file handle. * Return the number of bytes output, including XDR overhead. */ APPLESTATIC int nfsm_fhtom(struct nfsrv_descript *nd, u_int8_t *fhp, int size, int set_true) { u_int32_t *tl; u_int8_t *cp; int fullsiz, rem, bytesize = 0; if (size == 0) size = NFSX_MYFH; switch (nd->nd_flag & (ND_NFSV2 | ND_NFSV3 | ND_NFSV4)) { case ND_NFSV2: if (size > NFSX_V2FH) panic("fh size > NFSX_V2FH for NFSv2"); NFSM_BUILD(cp, u_int8_t *, NFSX_V2FH); NFSBCOPY(fhp, cp, size); if (size < NFSX_V2FH) NFSBZERO(cp + size, NFSX_V2FH - size); bytesize = NFSX_V2FH; break; case ND_NFSV3: case ND_NFSV4: fullsiz = NFSM_RNDUP(size); rem = fullsiz - size; if (set_true) { bytesize = 2 * NFSX_UNSIGNED + fullsiz; NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = newnfs_true; } else { bytesize = NFSX_UNSIGNED + fullsiz; } (void) nfsm_strtom(nd, fhp, size); break; } return (bytesize); } /* * This function compares two net addresses by family and returns TRUE * if they are the same host. * If there is any doubt, return FALSE. * The AF_INET family is handled as a special case so that address mbufs * don't need to be saved to store "struct in_addr", which is only 4 bytes. */ APPLESTATIC int nfsaddr_match(int family, union nethostaddr *haddr, NFSSOCKADDR_T nam) { struct sockaddr_in *inetaddr; switch (family) { case AF_INET: inetaddr = NFSSOCKADDR(nam, struct sockaddr_in *); if (inetaddr->sin_family == AF_INET && inetaddr->sin_addr.s_addr == haddr->had_inet.s_addr) return (1); break; #ifdef INET6 case AF_INET6: { struct sockaddr_in6 *inetaddr6; inetaddr6 = NFSSOCKADDR(nam, struct sockaddr_in6 *); /* XXX - should test sin6_scope_id ? */ if (inetaddr6->sin6_family == AF_INET6 && IN6_ARE_ADDR_EQUAL(&inetaddr6->sin6_addr, &haddr->had_inet6)) return (1); } break; #endif } return (0); } /* * Similar to the above, but takes to NFSSOCKADDR_T args. */ APPLESTATIC int nfsaddr2_match(NFSSOCKADDR_T nam1, NFSSOCKADDR_T nam2) { struct sockaddr_in *addr1, *addr2; struct sockaddr *inaddr; inaddr = NFSSOCKADDR(nam1, struct sockaddr *); switch (inaddr->sa_family) { case AF_INET: addr1 = NFSSOCKADDR(nam1, struct sockaddr_in *); addr2 = NFSSOCKADDR(nam2, struct sockaddr_in *); if (addr2->sin_family == AF_INET && addr1->sin_addr.s_addr == addr2->sin_addr.s_addr) return (1); break; #ifdef INET6 case AF_INET6: { struct sockaddr_in6 *inet6addr1, *inet6addr2; inet6addr1 = NFSSOCKADDR(nam1, struct sockaddr_in6 *); inet6addr2 = NFSSOCKADDR(nam2, struct sockaddr_in6 *); /* XXX - should test sin6_scope_id ? */ if (inet6addr2->sin6_family == AF_INET6 && IN6_ARE_ADDR_EQUAL(&inet6addr1->sin6_addr, &inet6addr2->sin6_addr)) return (1); } break; #endif } return (0); } /* * Trim the stuff already dissected off the mbuf list. */ APPLESTATIC void newnfs_trimleading(nd) struct nfsrv_descript *nd; { mbuf_t m, n; int offs; /* * First, free up leading mbufs. */ if (nd->nd_mrep != nd->nd_md) { m = nd->nd_mrep; while (mbuf_next(m) != nd->nd_md) { if (mbuf_next(m) == NULL) panic("nfsm trim leading"); m = mbuf_next(m); } mbuf_setnext(m, NULL); mbuf_freem(nd->nd_mrep); } m = nd->nd_md; /* * Now, adjust this mbuf, based on nd_dpos. */ offs = nd->nd_dpos - NFSMTOD(m, caddr_t); if (offs == mbuf_len(m)) { n = m; m = mbuf_next(m); if (m == NULL) panic("nfsm trim leading2"); mbuf_setnext(n, NULL); mbuf_freem(n); } else if (offs > 0) { mbuf_setlen(m, mbuf_len(m) - offs); NFSM_DATAP(m, offs); } else if (offs < 0) panic("nfsm trimleading offs"); nd->nd_mrep = m; nd->nd_md = m; nd->nd_dpos = NFSMTOD(m, caddr_t); } /* * Trim trailing data off the mbuf list being built. */ APPLESTATIC void newnfs_trimtrailing(nd, mb, bpos) struct nfsrv_descript *nd; mbuf_t mb; caddr_t bpos; { if (mbuf_next(mb)) { mbuf_freem(mbuf_next(mb)); mbuf_setnext(mb, NULL); } mbuf_setlen(mb, bpos - NFSMTOD(mb, caddr_t)); nd->nd_mb = mb; nd->nd_bpos = bpos; } /* * Dissect a file handle on the client. */ APPLESTATIC int nfsm_getfh(struct nfsrv_descript *nd, struct nfsfh **nfhpp) { u_int32_t *tl; struct nfsfh *nfhp; int error, len; *nfhpp = NULL; if (nd->nd_flag & (ND_NFSV3 | ND_NFSV4)) { NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); if ((len = fxdr_unsigned(int, *tl)) <= 0 || len > NFSX_FHMAX) { error = EBADRPC; goto nfsmout; } } else len = NFSX_V2FH; MALLOC(nfhp, struct nfsfh *, sizeof (struct nfsfh) + len, M_NFSFH, M_WAITOK); error = nfsrv_mtostr(nd, nfhp->nfh_fh, len); if (error) { FREE((caddr_t)nfhp, M_NFSFH); goto nfsmout; } nfhp->nfh_len = len; *nfhpp = nfhp; nfsmout: NFSEXITCODE2(error, nd); return (error); } /* * Break down the nfsv4 acl. * If the aclp == NULL or won't fit in an acl, just discard the acl info. */ APPLESTATIC int nfsrv_dissectacl(struct nfsrv_descript *nd, NFSACL_T *aclp, int *aclerrp, int *aclsizep, __unused NFSPROC_T *p) { u_int32_t *tl; int i, aclsize; int acecnt, error = 0, aceerr = 0, acesize; *aclerrp = 0; if (aclp) aclp->acl_cnt = 0; /* * Parse out the ace entries and expect them to conform to * what can be supported by R/W/X bits. */ NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); aclsize = NFSX_UNSIGNED; acecnt = fxdr_unsigned(int, *tl); if (acecnt > ACL_MAX_ENTRIES) aceerr = NFSERR_ATTRNOTSUPP; if (nfsrv_useacl == 0) aceerr = NFSERR_ATTRNOTSUPP; for (i = 0; i < acecnt; i++) { if (aclp && !aceerr) error = nfsrv_dissectace(nd, &aclp->acl_entry[i], &aceerr, &acesize, p); else error = nfsrv_skipace(nd, &acesize); if (error) goto nfsmout; aclsize += acesize; } if (aclp && !aceerr) aclp->acl_cnt = acecnt; if (aceerr) *aclerrp = aceerr; if (aclsizep) *aclsizep = aclsize; nfsmout: NFSEXITCODE2(error, nd); return (error); } /* * Skip over an NFSv4 ace entry. Just dissect the xdr and discard it. */ static int nfsrv_skipace(struct nfsrv_descript *nd, int *acesizep) { u_int32_t *tl; int error, len = 0; NFSM_DISSECT(tl, u_int32_t *, 4 * NFSX_UNSIGNED); len = fxdr_unsigned(int, *(tl + 3)); error = nfsm_advance(nd, NFSM_RNDUP(len), -1); nfsmout: *acesizep = NFSM_RNDUP(len) + (4 * NFSX_UNSIGNED); NFSEXITCODE2(error, nd); return (error); } /* * Get attribute bits from an mbuf list. * Returns EBADRPC for a parsing error, 0 otherwise. * If the clearinvalid flag is set, clear the bits not supported. */ APPLESTATIC int nfsrv_getattrbits(struct nfsrv_descript *nd, nfsattrbit_t *attrbitp, int *cntp, int *retnotsupp) { u_int32_t *tl; int cnt, i, outcnt; int error = 0; NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); cnt = fxdr_unsigned(int, *tl); if (cnt < 0) { error = NFSERR_BADXDR; goto nfsmout; } if (cnt > NFSATTRBIT_MAXWORDS) outcnt = NFSATTRBIT_MAXWORDS; else outcnt = cnt; NFSZERO_ATTRBIT(attrbitp); if (outcnt > 0) { NFSM_DISSECT(tl, u_int32_t *, outcnt * NFSX_UNSIGNED); for (i = 0; i < outcnt; i++) attrbitp->bits[i] = fxdr_unsigned(u_int32_t, *tl++); } for (i = 0; i < (cnt - outcnt); i++) { NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); if (retnotsupp != NULL && *tl != 0) *retnotsupp = NFSERR_ATTRNOTSUPP; } if (cntp) *cntp = NFSX_UNSIGNED + (cnt * NFSX_UNSIGNED); nfsmout: NFSEXITCODE2(error, nd); return (error); } /* * Get the attributes for V4. * If the compare flag is true, test for any attribute changes, * otherwise return the attribute values. * These attributes cover fields in "struct vattr", "struct statfs", * "struct nfsfsinfo", the file handle and the lease duration. * The value of retcmpp is set to 1 if all attributes are the same, * and 0 otherwise. * Returns EBADRPC if it can't be parsed, 0 otherwise. */ APPLESTATIC int nfsv4_loadattr(struct nfsrv_descript *nd, vnode_t vp, struct nfsvattr *nap, struct nfsfh **nfhpp, fhandle_t *fhp, int fhsize, struct nfsv3_pathconf *pc, struct statfs *sbp, struct nfsstatfs *sfp, struct nfsfsinfo *fsp, NFSACL_T *aclp, int compare, int *retcmpp, u_int32_t *leasep, u_int32_t *rderrp, NFSPROC_T *p, struct ucred *cred) { u_int32_t *tl; int i = 0, j, k, l = 0, m, bitpos, attrsum = 0; int error, tfhsize, aceerr, attrsize, cnt, retnotsup; u_char *cp, *cp2, namestr[NFSV4_SMALLSTR + 1]; nfsattrbit_t attrbits, retattrbits, checkattrbits; struct nfsfh *tnfhp; struct nfsreferral *refp; u_quad_t tquad; nfsquad_t tnfsquad; struct timespec temptime; uid_t uid; gid_t gid; long fid; u_int32_t freenum = 0, tuint; u_int64_t uquad = 0, thyp, thyp2; #ifdef QUOTA struct dqblk dqb; uid_t savuid; #endif static struct timeval last64fileid; static size_t count64fileid; static struct timeval last64mountfileid; static size_t count64mountfileid; static struct timeval warninterval = { 60, 0 }; if (compare) { retnotsup = 0; error = nfsrv_getattrbits(nd, &attrbits, NULL, &retnotsup); } else { error = nfsrv_getattrbits(nd, &attrbits, NULL, NULL); } if (error) goto nfsmout; if (compare) { *retcmpp = retnotsup; } else { /* * Just set default values to some of the important ones. */ if (nap != NULL) { nap->na_type = VREG; nap->na_mode = 0; nap->na_rdev = (NFSDEV_T)0; nap->na_mtime.tv_sec = 0; nap->na_mtime.tv_nsec = 0; nap->na_gen = 0; nap->na_flags = 0; nap->na_blocksize = NFS_FABLKSIZE; } if (sbp != NULL) { sbp->f_bsize = NFS_FABLKSIZE; sbp->f_blocks = 0; sbp->f_bfree = 0; sbp->f_bavail = 0; sbp->f_files = 0; sbp->f_ffree = 0; } if (fsp != NULL) { fsp->fs_rtmax = 8192; fsp->fs_rtpref = 8192; fsp->fs_maxname = NFS_MAXNAMLEN; fsp->fs_wtmax = 8192; fsp->fs_wtpref = 8192; fsp->fs_wtmult = NFS_FABLKSIZE; fsp->fs_dtpref = 8192; fsp->fs_maxfilesize = 0xffffffffffffffffull; fsp->fs_timedelta.tv_sec = 0; fsp->fs_timedelta.tv_nsec = 1; fsp->fs_properties = (NFSV3_FSFLINK | NFSV3_FSFSYMLINK | NFSV3_FSFHOMOGENEOUS | NFSV3_FSFCANSETTIME); } if (pc != NULL) { pc->pc_linkmax = LINK_MAX; pc->pc_namemax = NAME_MAX; pc->pc_notrunc = 0; pc->pc_chownrestricted = 0; pc->pc_caseinsensitive = 0; pc->pc_casepreserving = 1; } } /* * Loop around getting the attributes. */ NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); attrsize = fxdr_unsigned(int, *tl); for (bitpos = 0; bitpos < NFSATTRBIT_MAX; bitpos++) { if (attrsum > attrsize) { error = NFSERR_BADXDR; goto nfsmout; } if (NFSISSET_ATTRBIT(&attrbits, bitpos)) switch (bitpos) { case NFSATTRBIT_SUPPORTEDATTRS: retnotsup = 0; if (compare || nap == NULL) error = nfsrv_getattrbits(nd, &retattrbits, &cnt, &retnotsup); else error = nfsrv_getattrbits(nd, &nap->na_suppattr, &cnt, &retnotsup); if (error) goto nfsmout; if (compare && !(*retcmpp)) { NFSSETSUPP_ATTRBIT(&checkattrbits); /* Some filesystem do not support NFSv4ACL */ if (nfsrv_useacl == 0 || nfs_supportsnfsv4acls(vp) == 0) { NFSCLRBIT_ATTRBIT(&checkattrbits, NFSATTRBIT_ACL); NFSCLRBIT_ATTRBIT(&checkattrbits, NFSATTRBIT_ACLSUPPORT); } if (!NFSEQUAL_ATTRBIT(&retattrbits, &checkattrbits) || retnotsup) *retcmpp = NFSERR_NOTSAME; } attrsum += cnt; break; case NFSATTRBIT_TYPE: NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); if (compare) { if (!(*retcmpp)) { if (nap->na_type != nfsv34tov_type(*tl)) *retcmpp = NFSERR_NOTSAME; } } else if (nap != NULL) { nap->na_type = nfsv34tov_type(*tl); } attrsum += NFSX_UNSIGNED; break; case NFSATTRBIT_FHEXPIRETYPE: NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); if (compare && !(*retcmpp)) { if (fxdr_unsigned(int, *tl) != NFSV4FHTYPE_PERSISTENT) *retcmpp = NFSERR_NOTSAME; } attrsum += NFSX_UNSIGNED; break; case NFSATTRBIT_CHANGE: NFSM_DISSECT(tl, u_int32_t *, NFSX_HYPER); if (compare) { if (!(*retcmpp)) { if (nap->na_filerev != fxdr_hyper(tl)) *retcmpp = NFSERR_NOTSAME; } } else if (nap != NULL) { nap->na_filerev = fxdr_hyper(tl); } attrsum += NFSX_HYPER; break; case NFSATTRBIT_SIZE: NFSM_DISSECT(tl, u_int32_t *, NFSX_HYPER); if (compare) { if (!(*retcmpp)) { if (nap->na_size != fxdr_hyper(tl)) *retcmpp = NFSERR_NOTSAME; } } else if (nap != NULL) { nap->na_size = fxdr_hyper(tl); } attrsum += NFSX_HYPER; break; case NFSATTRBIT_LINKSUPPORT: NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); if (compare) { if (!(*retcmpp)) { if (fsp->fs_properties & NFSV3_FSFLINK) { if (*tl == newnfs_false) *retcmpp = NFSERR_NOTSAME; } else { if (*tl == newnfs_true) *retcmpp = NFSERR_NOTSAME; } } } else if (fsp != NULL) { if (*tl == newnfs_true) fsp->fs_properties |= NFSV3_FSFLINK; else fsp->fs_properties &= ~NFSV3_FSFLINK; } attrsum += NFSX_UNSIGNED; break; case NFSATTRBIT_SYMLINKSUPPORT: NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); if (compare) { if (!(*retcmpp)) { if (fsp->fs_properties & NFSV3_FSFSYMLINK) { if (*tl == newnfs_false) *retcmpp = NFSERR_NOTSAME; } else { if (*tl == newnfs_true) *retcmpp = NFSERR_NOTSAME; } } } else if (fsp != NULL) { if (*tl == newnfs_true) fsp->fs_properties |= NFSV3_FSFSYMLINK; else fsp->fs_properties &= ~NFSV3_FSFSYMLINK; } attrsum += NFSX_UNSIGNED; break; case NFSATTRBIT_NAMEDATTR: NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); if (compare && !(*retcmpp)) { if (*tl != newnfs_false) *retcmpp = NFSERR_NOTSAME; } attrsum += NFSX_UNSIGNED; break; case NFSATTRBIT_FSID: NFSM_DISSECT(tl, u_int32_t *, 4 * NFSX_UNSIGNED); thyp = fxdr_hyper(tl); tl += 2; thyp2 = fxdr_hyper(tl); if (compare) { if (*retcmpp == 0) { if (thyp != (u_int64_t) vfs_statfs(vnode_mount(vp))->f_fsid.val[0] || thyp2 != (u_int64_t) vfs_statfs(vnode_mount(vp))->f_fsid.val[1]) *retcmpp = NFSERR_NOTSAME; } } else if (nap != NULL) { nap->na_filesid[0] = thyp; nap->na_filesid[1] = thyp2; } attrsum += (4 * NFSX_UNSIGNED); break; case NFSATTRBIT_UNIQUEHANDLES: NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); if (compare && !(*retcmpp)) { if (*tl != newnfs_true) *retcmpp = NFSERR_NOTSAME; } attrsum += NFSX_UNSIGNED; break; case NFSATTRBIT_LEASETIME: NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); if (compare) { if (fxdr_unsigned(int, *tl) != nfsrv_lease && !(*retcmpp)) *retcmpp = NFSERR_NOTSAME; } else if (leasep != NULL) { *leasep = fxdr_unsigned(u_int32_t, *tl); } attrsum += NFSX_UNSIGNED; break; case NFSATTRBIT_RDATTRERROR: NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); if (compare) { if (!(*retcmpp)) *retcmpp = NFSERR_INVAL; } else if (rderrp != NULL) { *rderrp = fxdr_unsigned(u_int32_t, *tl); } attrsum += NFSX_UNSIGNED; break; case NFSATTRBIT_ACL: if (compare) { if (!(*retcmpp)) { if (nfsrv_useacl && nfs_supportsnfsv4acls(vp)) { NFSACL_T *naclp; naclp = acl_alloc(M_WAITOK); error = nfsrv_dissectacl(nd, naclp, &aceerr, &cnt, p); if (error) { acl_free(naclp); goto nfsmout; } if (aceerr || aclp == NULL || nfsrv_compareacl(aclp, naclp)) *retcmpp = NFSERR_NOTSAME; acl_free(naclp); } else { error = nfsrv_dissectacl(nd, NULL, &aceerr, &cnt, p); *retcmpp = NFSERR_ATTRNOTSUPP; } } } else { if (vp != NULL && aclp != NULL) error = nfsrv_dissectacl(nd, aclp, &aceerr, &cnt, p); else error = nfsrv_dissectacl(nd, NULL, &aceerr, &cnt, p); if (error) goto nfsmout; } attrsum += cnt; break; case NFSATTRBIT_ACLSUPPORT: NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); if (compare && !(*retcmpp)) { if (nfsrv_useacl && nfs_supportsnfsv4acls(vp)) { if (fxdr_unsigned(u_int32_t, *tl) != NFSV4ACE_SUPTYPES) *retcmpp = NFSERR_NOTSAME; } else { *retcmpp = NFSERR_ATTRNOTSUPP; } } attrsum += NFSX_UNSIGNED; break; case NFSATTRBIT_ARCHIVE: NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); if (compare && !(*retcmpp)) *retcmpp = NFSERR_ATTRNOTSUPP; attrsum += NFSX_UNSIGNED; break; case NFSATTRBIT_CANSETTIME: NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); if (compare) { if (!(*retcmpp)) { if (fsp->fs_properties & NFSV3_FSFCANSETTIME) { if (*tl == newnfs_false) *retcmpp = NFSERR_NOTSAME; } else { if (*tl == newnfs_true) *retcmpp = NFSERR_NOTSAME; } } } else if (fsp != NULL) { if (*tl == newnfs_true) fsp->fs_properties |= NFSV3_FSFCANSETTIME; else fsp->fs_properties &= ~NFSV3_FSFCANSETTIME; } attrsum += NFSX_UNSIGNED; break; case NFSATTRBIT_CASEINSENSITIVE: NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); if (compare) { if (!(*retcmpp)) { if (*tl != newnfs_false) *retcmpp = NFSERR_NOTSAME; } } else if (pc != NULL) { pc->pc_caseinsensitive = fxdr_unsigned(u_int32_t, *tl); } attrsum += NFSX_UNSIGNED; break; case NFSATTRBIT_CASEPRESERVING: NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); if (compare) { if (!(*retcmpp)) { if (*tl != newnfs_true) *retcmpp = NFSERR_NOTSAME; } } else if (pc != NULL) { pc->pc_casepreserving = fxdr_unsigned(u_int32_t, *tl); } attrsum += NFSX_UNSIGNED; break; case NFSATTRBIT_CHOWNRESTRICTED: NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); if (compare) { if (!(*retcmpp)) { if (*tl != newnfs_true) *retcmpp = NFSERR_NOTSAME; } } else if (pc != NULL) { pc->pc_chownrestricted = fxdr_unsigned(u_int32_t, *tl); } attrsum += NFSX_UNSIGNED; break; case NFSATTRBIT_FILEHANDLE: error = nfsm_getfh(nd, &tnfhp); if (error) goto nfsmout; tfhsize = tnfhp->nfh_len; if (compare) { if (!(*retcmpp) && !NFSRV_CMPFH(tnfhp->nfh_fh, tfhsize, fhp, fhsize)) *retcmpp = NFSERR_NOTSAME; FREE((caddr_t)tnfhp, M_NFSFH); } else if (nfhpp != NULL) { *nfhpp = tnfhp; } else { FREE((caddr_t)tnfhp, M_NFSFH); } attrsum += (NFSX_UNSIGNED + NFSM_RNDUP(tfhsize)); break; case NFSATTRBIT_FILEID: NFSM_DISSECT(tl, u_int32_t *, NFSX_HYPER); thyp = fxdr_hyper(tl); if (compare) { if (!(*retcmpp)) { if ((u_int64_t)nap->na_fileid != thyp) *retcmpp = NFSERR_NOTSAME; } } else if (nap != NULL) { if (*tl++) { count64fileid++; if (ratecheck(&last64fileid, &warninterval)) { printf("NFSv4 fileid > 32bits (%zu occurrences)\n", count64fileid); count64fileid = 0; } } nap->na_fileid = thyp; } attrsum += NFSX_HYPER; break; case NFSATTRBIT_FILESAVAIL: NFSM_DISSECT(tl, u_int32_t *, NFSX_HYPER); if (compare) { if (!(*retcmpp) && sfp->sf_afiles != fxdr_hyper(tl)) *retcmpp = NFSERR_NOTSAME; } else if (sfp != NULL) { sfp->sf_afiles = fxdr_hyper(tl); } attrsum += NFSX_HYPER; break; case NFSATTRBIT_FILESFREE: NFSM_DISSECT(tl, u_int32_t *, NFSX_HYPER); if (compare) { if (!(*retcmpp) && sfp->sf_ffiles != fxdr_hyper(tl)) *retcmpp = NFSERR_NOTSAME; } else if (sfp != NULL) { sfp->sf_ffiles = fxdr_hyper(tl); } attrsum += NFSX_HYPER; break; case NFSATTRBIT_FILESTOTAL: NFSM_DISSECT(tl, u_int32_t *, NFSX_HYPER); if (compare) { if (!(*retcmpp) && sfp->sf_tfiles != fxdr_hyper(tl)) *retcmpp = NFSERR_NOTSAME; } else if (sfp != NULL) { sfp->sf_tfiles = fxdr_hyper(tl); } attrsum += NFSX_HYPER; break; case NFSATTRBIT_FSLOCATIONS: error = nfsrv_getrefstr(nd, &cp, &cp2, &l, &m); if (error) goto nfsmout; attrsum += l; if (compare && !(*retcmpp)) { refp = nfsv4root_getreferral(vp, NULL, 0); if (refp != NULL) { if (cp == NULL || cp2 == NULL || strcmp(cp, "/") || strcmp(cp2, refp->nfr_srvlist)) *retcmpp = NFSERR_NOTSAME; } else if (m == 0) { *retcmpp = NFSERR_NOTSAME; } } if (cp != NULL) free(cp, M_NFSSTRING); if (cp2 != NULL) free(cp2, M_NFSSTRING); break; case NFSATTRBIT_HIDDEN: NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); if (compare && !(*retcmpp)) *retcmpp = NFSERR_ATTRNOTSUPP; attrsum += NFSX_UNSIGNED; break; case NFSATTRBIT_HOMOGENEOUS: NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); if (compare) { if (!(*retcmpp)) { if (fsp->fs_properties & NFSV3_FSFHOMOGENEOUS) { if (*tl == newnfs_false) *retcmpp = NFSERR_NOTSAME; } else { if (*tl == newnfs_true) *retcmpp = NFSERR_NOTSAME; } } } else if (fsp != NULL) { if (*tl == newnfs_true) fsp->fs_properties |= NFSV3_FSFHOMOGENEOUS; else fsp->fs_properties &= ~NFSV3_FSFHOMOGENEOUS; } attrsum += NFSX_UNSIGNED; break; case NFSATTRBIT_MAXFILESIZE: NFSM_DISSECT(tl, u_int32_t *, NFSX_HYPER); tnfsquad.qval = fxdr_hyper(tl); if (compare) { if (!(*retcmpp)) { tquad = NFSRV_MAXFILESIZE; if (tquad != tnfsquad.qval) *retcmpp = NFSERR_NOTSAME; } } else if (fsp != NULL) { fsp->fs_maxfilesize = tnfsquad.qval; } attrsum += NFSX_HYPER; break; case NFSATTRBIT_MAXLINK: NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); if (compare) { if (!(*retcmpp)) { if (fxdr_unsigned(int, *tl) != LINK_MAX) *retcmpp = NFSERR_NOTSAME; } } else if (pc != NULL) { pc->pc_linkmax = fxdr_unsigned(u_int32_t, *tl); } attrsum += NFSX_UNSIGNED; break; case NFSATTRBIT_MAXNAME: NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); if (compare) { if (!(*retcmpp)) { if (fsp->fs_maxname != fxdr_unsigned(u_int32_t, *tl)) *retcmpp = NFSERR_NOTSAME; } } else { tuint = fxdr_unsigned(u_int32_t, *tl); /* * Some Linux NFSv4 servers report this * as 0 or 4billion, so I'll set it to * NFS_MAXNAMLEN. If a server actually creates * a name longer than NFS_MAXNAMLEN, it will * get an error back. */ if (tuint == 0 || tuint > NFS_MAXNAMLEN) tuint = NFS_MAXNAMLEN; if (fsp != NULL) fsp->fs_maxname = tuint; if (pc != NULL) pc->pc_namemax = tuint; } attrsum += NFSX_UNSIGNED; break; case NFSATTRBIT_MAXREAD: NFSM_DISSECT(tl, u_int32_t *, NFSX_HYPER); if (compare) { if (!(*retcmpp)) { if (fsp->fs_rtmax != fxdr_unsigned(u_int32_t, *(tl + 1)) || *tl != 0) *retcmpp = NFSERR_NOTSAME; } } else if (fsp != NULL) { fsp->fs_rtmax = fxdr_unsigned(u_int32_t, *++tl); fsp->fs_rtpref = fsp->fs_rtmax; fsp->fs_dtpref = fsp->fs_rtpref; } attrsum += NFSX_HYPER; break; case NFSATTRBIT_MAXWRITE: NFSM_DISSECT(tl, u_int32_t *, NFSX_HYPER); if (compare) { if (!(*retcmpp)) { if (fsp->fs_wtmax != fxdr_unsigned(u_int32_t, *(tl + 1)) || *tl != 0) *retcmpp = NFSERR_NOTSAME; } } else if (fsp != NULL) { fsp->fs_wtmax = fxdr_unsigned(int, *++tl); fsp->fs_wtpref = fsp->fs_wtmax; } attrsum += NFSX_HYPER; break; case NFSATTRBIT_MIMETYPE: NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); i = fxdr_unsigned(int, *tl); attrsum += (NFSX_UNSIGNED + NFSM_RNDUP(i)); error = nfsm_advance(nd, NFSM_RNDUP(i), -1); if (error) goto nfsmout; if (compare && !(*retcmpp)) *retcmpp = NFSERR_ATTRNOTSUPP; break; case NFSATTRBIT_MODE: NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); if (compare) { if (!(*retcmpp)) { if (nap->na_mode != nfstov_mode(*tl)) *retcmpp = NFSERR_NOTSAME; } } else if (nap != NULL) { nap->na_mode = nfstov_mode(*tl); } attrsum += NFSX_UNSIGNED; break; case NFSATTRBIT_NOTRUNC: NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); if (compare) { if (!(*retcmpp)) { if (*tl != newnfs_true) *retcmpp = NFSERR_NOTSAME; } } else if (pc != NULL) { pc->pc_notrunc = fxdr_unsigned(u_int32_t, *tl); } attrsum += NFSX_UNSIGNED; break; case NFSATTRBIT_NUMLINKS: NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); tuint = fxdr_unsigned(u_int32_t, *tl); if (compare) { if (!(*retcmpp)) { if ((u_int32_t)nap->na_nlink != tuint) *retcmpp = NFSERR_NOTSAME; } } else if (nap != NULL) { nap->na_nlink = tuint; } attrsum += NFSX_UNSIGNED; break; case NFSATTRBIT_OWNER: NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); j = fxdr_unsigned(int, *tl); if (j < 0) { error = NFSERR_BADXDR; goto nfsmout; } attrsum += (NFSX_UNSIGNED + NFSM_RNDUP(j)); if (j > NFSV4_SMALLSTR) cp = malloc(j + 1, M_NFSSTRING, M_WAITOK); else cp = namestr; error = nfsrv_mtostr(nd, cp, j); if (error) { if (j > NFSV4_SMALLSTR) free(cp, M_NFSSTRING); goto nfsmout; } if (compare) { if (!(*retcmpp)) { if (nfsv4_strtouid(nd, cp, j, &uid, p) || nap->na_uid != uid) *retcmpp = NFSERR_NOTSAME; } } else if (nap != NULL) { if (nfsv4_strtouid(nd, cp, j, &uid, p)) nap->na_uid = nfsrv_defaultuid; else nap->na_uid = uid; } if (j > NFSV4_SMALLSTR) free(cp, M_NFSSTRING); break; case NFSATTRBIT_OWNERGROUP: NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); j = fxdr_unsigned(int, *tl); if (j < 0) { error = NFSERR_BADXDR; goto nfsmout; } attrsum += (NFSX_UNSIGNED + NFSM_RNDUP(j)); if (j > NFSV4_SMALLSTR) cp = malloc(j + 1, M_NFSSTRING, M_WAITOK); else cp = namestr; error = nfsrv_mtostr(nd, cp, j); if (error) { if (j > NFSV4_SMALLSTR) free(cp, M_NFSSTRING); goto nfsmout; } if (compare) { if (!(*retcmpp)) { if (nfsv4_strtogid(nd, cp, j, &gid, p) || nap->na_gid != gid) *retcmpp = NFSERR_NOTSAME; } } else if (nap != NULL) { if (nfsv4_strtogid(nd, cp, j, &gid, p)) nap->na_gid = nfsrv_defaultgid; else nap->na_gid = gid; } if (j > NFSV4_SMALLSTR) free(cp, M_NFSSTRING); break; case NFSATTRBIT_QUOTAHARD: NFSM_DISSECT(tl, u_int32_t *, NFSX_HYPER); if (sbp != NULL) { if (priv_check_cred(cred, PRIV_VFS_EXCEEDQUOTA, 0)) freenum = sbp->f_bfree; else freenum = sbp->f_bavail; #ifdef QUOTA /* * ufs_quotactl() insists that the uid argument * equal p_ruid for non-root quota access, so * we'll just make sure that's the case. */ savuid = p->p_cred->p_ruid; p->p_cred->p_ruid = cred->cr_uid; if (!VFS_QUOTACTL(vnode_mount(vp),QCMD(Q_GETQUOTA, USRQUOTA), cred->cr_uid, (caddr_t)&dqb)) freenum = min(dqb.dqb_bhardlimit, freenum); p->p_cred->p_ruid = savuid; #endif /* QUOTA */ uquad = (u_int64_t)freenum; NFSQUOTABLKTOBYTE(uquad, sbp->f_bsize); } if (compare && !(*retcmpp)) { if (uquad != fxdr_hyper(tl)) *retcmpp = NFSERR_NOTSAME; } attrsum += NFSX_HYPER; break; case NFSATTRBIT_QUOTASOFT: NFSM_DISSECT(tl, u_int32_t *, NFSX_HYPER); if (sbp != NULL) { if (priv_check_cred(cred, PRIV_VFS_EXCEEDQUOTA, 0)) freenum = sbp->f_bfree; else freenum = sbp->f_bavail; #ifdef QUOTA /* * ufs_quotactl() insists that the uid argument * equal p_ruid for non-root quota access, so * we'll just make sure that's the case. */ savuid = p->p_cred->p_ruid; p->p_cred->p_ruid = cred->cr_uid; if (!VFS_QUOTACTL(vnode_mount(vp),QCMD(Q_GETQUOTA, USRQUOTA), cred->cr_uid, (caddr_t)&dqb)) freenum = min(dqb.dqb_bsoftlimit, freenum); p->p_cred->p_ruid = savuid; #endif /* QUOTA */ uquad = (u_int64_t)freenum; NFSQUOTABLKTOBYTE(uquad, sbp->f_bsize); } if (compare && !(*retcmpp)) { if (uquad != fxdr_hyper(tl)) *retcmpp = NFSERR_NOTSAME; } attrsum += NFSX_HYPER; break; case NFSATTRBIT_QUOTAUSED: NFSM_DISSECT(tl, u_int32_t *, NFSX_HYPER); if (sbp != NULL) { freenum = 0; #ifdef QUOTA /* * ufs_quotactl() insists that the uid argument * equal p_ruid for non-root quota access, so * we'll just make sure that's the case. */ savuid = p->p_cred->p_ruid; p->p_cred->p_ruid = cred->cr_uid; if (!VFS_QUOTACTL(vnode_mount(vp),QCMD(Q_GETQUOTA, USRQUOTA), cred->cr_uid, (caddr_t)&dqb)) freenum = dqb.dqb_curblocks; p->p_cred->p_ruid = savuid; #endif /* QUOTA */ uquad = (u_int64_t)freenum; NFSQUOTABLKTOBYTE(uquad, sbp->f_bsize); } if (compare && !(*retcmpp)) { if (uquad != fxdr_hyper(tl)) *retcmpp = NFSERR_NOTSAME; } attrsum += NFSX_HYPER; break; case NFSATTRBIT_RAWDEV: NFSM_DISSECT(tl, u_int32_t *, NFSX_V4SPECDATA); j = fxdr_unsigned(int, *tl++); k = fxdr_unsigned(int, *tl); if (compare) { if (!(*retcmpp)) { if (nap->na_rdev != NFSMAKEDEV(j, k)) *retcmpp = NFSERR_NOTSAME; } } else if (nap != NULL) { nap->na_rdev = NFSMAKEDEV(j, k); } attrsum += NFSX_V4SPECDATA; break; case NFSATTRBIT_SPACEAVAIL: NFSM_DISSECT(tl, u_int32_t *, NFSX_HYPER); if (compare) { if (!(*retcmpp) && sfp->sf_abytes != fxdr_hyper(tl)) *retcmpp = NFSERR_NOTSAME; } else if (sfp != NULL) { sfp->sf_abytes = fxdr_hyper(tl); } attrsum += NFSX_HYPER; break; case NFSATTRBIT_SPACEFREE: NFSM_DISSECT(tl, u_int32_t *, NFSX_HYPER); if (compare) { if (!(*retcmpp) && sfp->sf_fbytes != fxdr_hyper(tl)) *retcmpp = NFSERR_NOTSAME; } else if (sfp != NULL) { sfp->sf_fbytes = fxdr_hyper(tl); } attrsum += NFSX_HYPER; break; case NFSATTRBIT_SPACETOTAL: NFSM_DISSECT(tl, u_int32_t *, NFSX_HYPER); if (compare) { if (!(*retcmpp) && sfp->sf_tbytes != fxdr_hyper(tl)) *retcmpp = NFSERR_NOTSAME; } else if (sfp != NULL) { sfp->sf_tbytes = fxdr_hyper(tl); } attrsum += NFSX_HYPER; break; case NFSATTRBIT_SPACEUSED: NFSM_DISSECT(tl, u_int32_t *, NFSX_HYPER); thyp = fxdr_hyper(tl); if (compare) { if (!(*retcmpp)) { if ((u_int64_t)nap->na_bytes != thyp) *retcmpp = NFSERR_NOTSAME; } } else if (nap != NULL) { nap->na_bytes = thyp; } attrsum += NFSX_HYPER; break; case NFSATTRBIT_SYSTEM: NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); if (compare && !(*retcmpp)) *retcmpp = NFSERR_ATTRNOTSUPP; attrsum += NFSX_UNSIGNED; break; case NFSATTRBIT_TIMEACCESS: NFSM_DISSECT(tl, u_int32_t *, NFSX_V4TIME); fxdr_nfsv4time(tl, &temptime); if (compare) { if (!(*retcmpp)) { if (!NFS_CMPTIME(temptime, nap->na_atime)) *retcmpp = NFSERR_NOTSAME; } } else if (nap != NULL) { nap->na_atime = temptime; } attrsum += NFSX_V4TIME; break; case NFSATTRBIT_TIMEACCESSSET: NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); attrsum += NFSX_UNSIGNED; i = fxdr_unsigned(int, *tl); if (i == NFSV4SATTRTIME_TOCLIENT) { NFSM_DISSECT(tl, u_int32_t *, NFSX_V4TIME); attrsum += NFSX_V4TIME; } if (compare && !(*retcmpp)) *retcmpp = NFSERR_INVAL; break; case NFSATTRBIT_TIMEBACKUP: NFSM_DISSECT(tl, u_int32_t *, NFSX_V4TIME); if (compare && !(*retcmpp)) *retcmpp = NFSERR_ATTRNOTSUPP; attrsum += NFSX_V4TIME; break; case NFSATTRBIT_TIMECREATE: NFSM_DISSECT(tl, u_int32_t *, NFSX_V4TIME); if (compare && !(*retcmpp)) *retcmpp = NFSERR_ATTRNOTSUPP; attrsum += NFSX_V4TIME; break; case NFSATTRBIT_TIMEDELTA: NFSM_DISSECT(tl, u_int32_t *, NFSX_V4TIME); if (fsp != NULL) { if (compare) { if (!(*retcmpp)) { if ((u_int32_t)fsp->fs_timedelta.tv_sec != fxdr_unsigned(u_int32_t, *(tl + 1)) || (u_int32_t)fsp->fs_timedelta.tv_nsec != (fxdr_unsigned(u_int32_t, *(tl + 2)) % 1000000000) || *tl != 0) *retcmpp = NFSERR_NOTSAME; } } else { fxdr_nfsv4time(tl, &fsp->fs_timedelta); } } attrsum += NFSX_V4TIME; break; case NFSATTRBIT_TIMEMETADATA: NFSM_DISSECT(tl, u_int32_t *, NFSX_V4TIME); fxdr_nfsv4time(tl, &temptime); if (compare) { if (!(*retcmpp)) { if (!NFS_CMPTIME(temptime, nap->na_ctime)) *retcmpp = NFSERR_NOTSAME; } } else if (nap != NULL) { nap->na_ctime = temptime; } attrsum += NFSX_V4TIME; break; case NFSATTRBIT_TIMEMODIFY: NFSM_DISSECT(tl, u_int32_t *, NFSX_V4TIME); fxdr_nfsv4time(tl, &temptime); if (compare) { if (!(*retcmpp)) { if (!NFS_CMPTIME(temptime, nap->na_mtime)) *retcmpp = NFSERR_NOTSAME; } } else if (nap != NULL) { nap->na_mtime = temptime; } attrsum += NFSX_V4TIME; break; case NFSATTRBIT_TIMEMODIFYSET: NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); attrsum += NFSX_UNSIGNED; i = fxdr_unsigned(int, *tl); if (i == NFSV4SATTRTIME_TOCLIENT) { NFSM_DISSECT(tl, u_int32_t *, NFSX_V4TIME); attrsum += NFSX_V4TIME; } if (compare && !(*retcmpp)) *retcmpp = NFSERR_INVAL; break; case NFSATTRBIT_MOUNTEDONFILEID: NFSM_DISSECT(tl, u_int32_t *, NFSX_HYPER); thyp = fxdr_hyper(tl); if (compare) { if (!(*retcmpp)) { if (*tl++) { *retcmpp = NFSERR_NOTSAME; } else { if (!vp || !nfsrv_atroot(vp, &fid)) fid = nap->na_fileid; if ((u_int64_t)fid != thyp) *retcmpp = NFSERR_NOTSAME; } } } else if (nap != NULL) { if (*tl++) { count64mountfileid++; if (ratecheck(&last64mountfileid, &warninterval)) { printf("NFSv4 mounted on fileid > 32bits (%zu occurrences)\n", count64mountfileid); count64mountfileid = 0; } } nap->na_mntonfileno = thyp; } attrsum += NFSX_HYPER; break; case NFSATTRBIT_SUPPATTREXCLCREAT: retnotsup = 0; error = nfsrv_getattrbits(nd, &retattrbits, &cnt, &retnotsup); if (error) goto nfsmout; if (compare && !(*retcmpp)) { NFSSETSUPP_ATTRBIT(&checkattrbits); NFSCLRNOTSETABLE_ATTRBIT(&checkattrbits); NFSCLRBIT_ATTRBIT(&checkattrbits, NFSATTRBIT_TIMEACCESSSET); if (!NFSEQUAL_ATTRBIT(&retattrbits, &checkattrbits) || retnotsup) *retcmpp = NFSERR_NOTSAME; } attrsum += cnt; break; default: printf("EEK! nfsv4_loadattr unknown attr=%d\n", bitpos); if (compare && !(*retcmpp)) *retcmpp = NFSERR_ATTRNOTSUPP; /* * and get out of the loop, since we can't parse * the unknown attrbute data. */ bitpos = NFSATTRBIT_MAX; break; } } /* * some clients pad the attrlist, so we need to skip over the * padding. */ if (attrsum > attrsize) { error = NFSERR_BADXDR; } else { attrsize = NFSM_RNDUP(attrsize); if (attrsum < attrsize) error = nfsm_advance(nd, attrsize - attrsum, -1); } nfsmout: NFSEXITCODE2(error, nd); return (error); } /* * Implement sleep locks for newnfs. The nfslock_usecnt allows for a * shared lock and the NFSXXX_LOCK flag permits an exclusive lock. * The first argument is a pointer to an nfsv4lock structure. * The second argument is 1 iff a blocking lock is wanted. * If this argument is 0, the call waits until no thread either wants nor * holds an exclusive lock. * It returns 1 if the lock was acquired, 0 otherwise. * If several processes call this function concurrently wanting the exclusive * lock, one will get the lock and the rest will return without getting the * lock. (If the caller must have the lock, it simply calls this function in a * loop until the function returns 1 to indicate the lock was acquired.) * Any usecnt must be decremented by calling nfsv4_relref() before * calling nfsv4_lock(). It was done this way, so nfsv4_lock() could * be called in a loop. * The isleptp argument is set to indicate if the call slept, iff not NULL * and the mp argument indicates to check for a forced dismount, iff not * NULL. */ APPLESTATIC int nfsv4_lock(struct nfsv4lock *lp, int iwantlock, int *isleptp, void *mutex, struct mount *mp) { if (isleptp) *isleptp = 0; /* * If a lock is wanted, loop around until the lock is acquired by * someone and then released. If I want the lock, try to acquire it. * For a lock to be issued, no lock must be in force and the usecnt * must be zero. */ if (iwantlock) { if (!(lp->nfslock_lock & NFSV4LOCK_LOCK) && lp->nfslock_usecnt == 0) { lp->nfslock_lock &= ~NFSV4LOCK_LOCKWANTED; lp->nfslock_lock |= NFSV4LOCK_LOCK; return (1); } lp->nfslock_lock |= NFSV4LOCK_LOCKWANTED; } while (lp->nfslock_lock & (NFSV4LOCK_LOCK | NFSV4LOCK_LOCKWANTED)) { if (mp != NULL && (mp->mnt_kern_flag & MNTK_UNMOUNTF) != 0) { lp->nfslock_lock &= ~NFSV4LOCK_LOCKWANTED; return (0); } lp->nfslock_lock |= NFSV4LOCK_WANTED; if (isleptp) *isleptp = 1; (void) nfsmsleep(&lp->nfslock_lock, mutex, PZERO - 1, "nfsv4lck", NULL); if (iwantlock && !(lp->nfslock_lock & NFSV4LOCK_LOCK) && lp->nfslock_usecnt == 0) { lp->nfslock_lock &= ~NFSV4LOCK_LOCKWANTED; lp->nfslock_lock |= NFSV4LOCK_LOCK; return (1); } } return (0); } /* * Release the lock acquired by nfsv4_lock(). * The second argument is set to 1 to indicate the nfslock_usecnt should be * incremented, as well. */ APPLESTATIC void nfsv4_unlock(struct nfsv4lock *lp, int incref) { lp->nfslock_lock &= ~NFSV4LOCK_LOCK; if (incref) lp->nfslock_usecnt++; nfsv4_wanted(lp); } /* * Release a reference cnt. */ APPLESTATIC void nfsv4_relref(struct nfsv4lock *lp) { if (lp->nfslock_usecnt <= 0) panic("nfsv4root ref cnt"); lp->nfslock_usecnt--; if (lp->nfslock_usecnt == 0) nfsv4_wanted(lp); } /* * Get a reference cnt. * This function will wait for any exclusive lock to be released, but will * not wait for threads that want the exclusive lock. If priority needs * to be given to threads that need the exclusive lock, a call to nfsv4_lock() * with the 2nd argument == 0 should be done before calling nfsv4_getref(). * If the mp argument is not NULL, check for MNTK_UNMOUNTF being set and * return without getting a refcnt for that case. */ APPLESTATIC void nfsv4_getref(struct nfsv4lock *lp, int *isleptp, void *mutex, struct mount *mp) { if (isleptp) *isleptp = 0; /* * Wait for a lock held. */ while (lp->nfslock_lock & NFSV4LOCK_LOCK) { if (mp != NULL && (mp->mnt_kern_flag & MNTK_UNMOUNTF) != 0) return; lp->nfslock_lock |= NFSV4LOCK_WANTED; if (isleptp) *isleptp = 1; (void) nfsmsleep(&lp->nfslock_lock, mutex, PZERO - 1, "nfsv4gr", NULL); } if (mp != NULL && (mp->mnt_kern_flag & MNTK_UNMOUNTF) != 0) return; lp->nfslock_usecnt++; } /* * Get a reference as above, but return failure instead of sleeping if * an exclusive lock is held. */ APPLESTATIC int nfsv4_getref_nonblock(struct nfsv4lock *lp) { if ((lp->nfslock_lock & NFSV4LOCK_LOCK) != 0) return (0); lp->nfslock_usecnt++; return (1); } /* * Test for a lock. Return 1 if locked, 0 otherwise. */ APPLESTATIC int nfsv4_testlock(struct nfsv4lock *lp) { if ((lp->nfslock_lock & NFSV4LOCK_LOCK) == 0 && lp->nfslock_usecnt == 0) return (0); return (1); } /* * Wake up anyone sleeping, waiting for this lock. */ static void nfsv4_wanted(struct nfsv4lock *lp) { if (lp->nfslock_lock & NFSV4LOCK_WANTED) { lp->nfslock_lock &= ~NFSV4LOCK_WANTED; wakeup((caddr_t)&lp->nfslock_lock); } } /* * Copy a string from an mbuf list into a character array. * Return EBADRPC if there is an mbuf error, * 0 otherwise. */ APPLESTATIC int nfsrv_mtostr(struct nfsrv_descript *nd, char *str, int siz) { char *cp; int xfer, len; mbuf_t mp; int rem, error = 0; mp = nd->nd_md; cp = nd->nd_dpos; len = NFSMTOD(mp, caddr_t) + mbuf_len(mp) - cp; rem = NFSM_RNDUP(siz) - siz; while (siz > 0) { if (len > siz) xfer = siz; else xfer = len; NFSBCOPY(cp, str, xfer); str += xfer; siz -= xfer; if (siz > 0) { mp = mbuf_next(mp); if (mp == NULL) { error = EBADRPC; goto out; } cp = NFSMTOD(mp, caddr_t); len = mbuf_len(mp); } else { cp += xfer; len -= xfer; } } *str = '\0'; nd->nd_dpos = cp; nd->nd_md = mp; if (rem > 0) { if (len < rem) error = nfsm_advance(nd, rem, len); else nd->nd_dpos += rem; } out: NFSEXITCODE2(error, nd); return (error); } /* * Fill in the attributes as marked by the bitmap (V4). */ APPLESTATIC int nfsv4_fillattr(struct nfsrv_descript *nd, struct mount *mp, vnode_t vp, NFSACL_T *saclp, struct vattr *vap, fhandle_t *fhp, int rderror, nfsattrbit_t *attrbitp, struct ucred *cred, NFSPROC_T *p, int isdgram, int reterr, int supports_nfsv4acls, int at_root, uint64_t mounted_on_fileno) { int bitpos, retnum = 0; u_int32_t *tl; int siz, prefixnum, error; u_char *cp, namestr[NFSV4_SMALLSTR]; nfsattrbit_t attrbits, retbits; nfsattrbit_t *retbitp = &retbits; u_int32_t freenum, *retnump; u_int64_t uquad; - struct statfs fs; + struct statfs *fs; struct nfsfsinfo fsinf; struct timespec temptime; NFSACL_T *aclp, *naclp = NULL; #ifdef QUOTA struct dqblk dqb; uid_t savuid; #endif /* * First, set the bits that can be filled and get fsinfo. */ NFSSET_ATTRBIT(retbitp, attrbitp); /* * If both p and cred are NULL, it is a client side setattr call. * If both p and cred are not NULL, it is a server side reply call. * If p is not NULL and cred is NULL, it is a client side callback * reply call. */ if (p == NULL && cred == NULL) { NFSCLRNOTSETABLE_ATTRBIT(retbitp); aclp = saclp; } else { NFSCLRNOTFILLABLE_ATTRBIT(retbitp); naclp = acl_alloc(M_WAITOK); aclp = naclp; } nfsvno_getfs(&fsinf, isdgram); #ifndef APPLE /* * Get the VFS_STATFS(), since some attributes need them. */ + fs = malloc(sizeof(struct statfs), M_STATFS, M_WAITOK); if (NFSISSETSTATFS_ATTRBIT(retbitp)) { - error = VFS_STATFS(mp, &fs); + error = VFS_STATFS(mp, fs); if (error != 0) { if (reterr) { nd->nd_repstat = NFSERR_ACCES; + free(fs, M_STATFS); return (0); } NFSCLRSTATFS_ATTRBIT(retbitp); } } #endif /* * And the NFSv4 ACL... */ if (NFSISSET_ATTRBIT(retbitp, NFSATTRBIT_ACLSUPPORT) && (nfsrv_useacl == 0 || ((cred != NULL || p != NULL) && supports_nfsv4acls == 0))) { NFSCLRBIT_ATTRBIT(retbitp, NFSATTRBIT_ACLSUPPORT); } if (NFSISSET_ATTRBIT(retbitp, NFSATTRBIT_ACL)) { if (nfsrv_useacl == 0 || ((cred != NULL || p != NULL) && supports_nfsv4acls == 0)) { NFSCLRBIT_ATTRBIT(retbitp, NFSATTRBIT_ACL); } else if (naclp != NULL) { if (NFSVOPLOCK(vp, LK_SHARED) == 0) { error = VOP_ACCESSX(vp, VREAD_ACL, cred, p); if (error == 0) error = VOP_GETACL(vp, ACL_TYPE_NFS4, naclp, cred, p); NFSVOPUNLOCK(vp, 0); } else error = NFSERR_PERM; if (error != 0) { if (reterr) { nd->nd_repstat = NFSERR_ACCES; + free(fs, M_STATFS); return (0); } NFSCLRBIT_ATTRBIT(retbitp, NFSATTRBIT_ACL); } } } /* * Put out the attribute bitmap for the ones being filled in * and get the field for the number of attributes returned. */ prefixnum = nfsrv_putattrbit(nd, retbitp); NFSM_BUILD(retnump, u_int32_t *, NFSX_UNSIGNED); prefixnum += NFSX_UNSIGNED; /* * Now, loop around filling in the attributes for each bit set. */ for (bitpos = 0; bitpos < NFSATTRBIT_MAX; bitpos++) { if (NFSISSET_ATTRBIT(retbitp, bitpos)) { switch (bitpos) { case NFSATTRBIT_SUPPORTEDATTRS: NFSSETSUPP_ATTRBIT(&attrbits); if (nfsrv_useacl == 0 || ((cred != NULL || p != NULL) && supports_nfsv4acls == 0)) { NFSCLRBIT_ATTRBIT(&attrbits,NFSATTRBIT_ACLSUPPORT); NFSCLRBIT_ATTRBIT(&attrbits,NFSATTRBIT_ACL); } retnum += nfsrv_putattrbit(nd, &attrbits); break; case NFSATTRBIT_TYPE: NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = vtonfsv34_type(vap->va_type); retnum += NFSX_UNSIGNED; break; case NFSATTRBIT_FHEXPIRETYPE: NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(NFSV4FHTYPE_PERSISTENT); retnum += NFSX_UNSIGNED; break; case NFSATTRBIT_CHANGE: NFSM_BUILD(tl, u_int32_t *, NFSX_HYPER); txdr_hyper(vap->va_filerev, tl); retnum += NFSX_HYPER; break; case NFSATTRBIT_SIZE: NFSM_BUILD(tl, u_int32_t *, NFSX_HYPER); txdr_hyper(vap->va_size, tl); retnum += NFSX_HYPER; break; case NFSATTRBIT_LINKSUPPORT: NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); if (fsinf.fs_properties & NFSV3FSINFO_LINK) *tl = newnfs_true; else *tl = newnfs_false; retnum += NFSX_UNSIGNED; break; case NFSATTRBIT_SYMLINKSUPPORT: NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); if (fsinf.fs_properties & NFSV3FSINFO_SYMLINK) *tl = newnfs_true; else *tl = newnfs_false; retnum += NFSX_UNSIGNED; break; case NFSATTRBIT_NAMEDATTR: NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = newnfs_false; retnum += NFSX_UNSIGNED; break; case NFSATTRBIT_FSID: NFSM_BUILD(tl, u_int32_t *, NFSX_V4FSID); *tl++ = 0; *tl++ = txdr_unsigned(mp->mnt_stat.f_fsid.val[0]); *tl++ = 0; *tl = txdr_unsigned(mp->mnt_stat.f_fsid.val[1]); retnum += NFSX_V4FSID; break; case NFSATTRBIT_UNIQUEHANDLES: NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = newnfs_true; retnum += NFSX_UNSIGNED; break; case NFSATTRBIT_LEASETIME: NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(nfsrv_lease); retnum += NFSX_UNSIGNED; break; case NFSATTRBIT_RDATTRERROR: NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(rderror); retnum += NFSX_UNSIGNED; break; /* * Recommended Attributes. (Only the supported ones.) */ case NFSATTRBIT_ACL: retnum += nfsrv_buildacl(nd, aclp, vnode_vtype(vp), p); break; case NFSATTRBIT_ACLSUPPORT: NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(NFSV4ACE_SUPTYPES); retnum += NFSX_UNSIGNED; break; case NFSATTRBIT_CANSETTIME: NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); if (fsinf.fs_properties & NFSV3FSINFO_CANSETTIME) *tl = newnfs_true; else *tl = newnfs_false; retnum += NFSX_UNSIGNED; break; case NFSATTRBIT_CASEINSENSITIVE: NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = newnfs_false; retnum += NFSX_UNSIGNED; break; case NFSATTRBIT_CASEPRESERVING: NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = newnfs_true; retnum += NFSX_UNSIGNED; break; case NFSATTRBIT_CHOWNRESTRICTED: NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = newnfs_true; retnum += NFSX_UNSIGNED; break; case NFSATTRBIT_FILEHANDLE: retnum += nfsm_fhtom(nd, (u_int8_t *)fhp, 0, 0); break; case NFSATTRBIT_FILEID: NFSM_BUILD(tl, u_int32_t *, NFSX_HYPER); *tl++ = 0; *tl = txdr_unsigned(vap->va_fileid); retnum += NFSX_HYPER; break; case NFSATTRBIT_FILESAVAIL: /* * Check quota and use min(quota, f_ffree). */ - freenum = fs.f_ffree; + freenum = fs->f_ffree; #ifdef QUOTA /* * ufs_quotactl() insists that the uid argument * equal p_ruid for non-root quota access, so * we'll just make sure that's the case. */ savuid = p->p_cred->p_ruid; p->p_cred->p_ruid = cred->cr_uid; if (!VFS_QUOTACTL(mp, QCMD(Q_GETQUOTA,USRQUOTA), cred->cr_uid, (caddr_t)&dqb)) freenum = min(dqb.dqb_isoftlimit-dqb.dqb_curinodes, freenum); p->p_cred->p_ruid = savuid; #endif /* QUOTA */ NFSM_BUILD(tl, u_int32_t *, NFSX_HYPER); *tl++ = 0; *tl = txdr_unsigned(freenum); retnum += NFSX_HYPER; break; case NFSATTRBIT_FILESFREE: NFSM_BUILD(tl, u_int32_t *, NFSX_HYPER); *tl++ = 0; - *tl = txdr_unsigned(fs.f_ffree); + *tl = txdr_unsigned(fs->f_ffree); retnum += NFSX_HYPER; break; case NFSATTRBIT_FILESTOTAL: NFSM_BUILD(tl, u_int32_t *, NFSX_HYPER); *tl++ = 0; - *tl = txdr_unsigned(fs.f_files); + *tl = txdr_unsigned(fs->f_files); retnum += NFSX_HYPER; break; case NFSATTRBIT_FSLOCATIONS: NFSM_BUILD(tl, u_int32_t *, 2 * NFSX_UNSIGNED); *tl++ = 0; *tl = 0; retnum += 2 * NFSX_UNSIGNED; break; case NFSATTRBIT_HOMOGENEOUS: NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); if (fsinf.fs_properties & NFSV3FSINFO_HOMOGENEOUS) *tl = newnfs_true; else *tl = newnfs_false; retnum += NFSX_UNSIGNED; break; case NFSATTRBIT_MAXFILESIZE: NFSM_BUILD(tl, u_int32_t *, NFSX_HYPER); uquad = NFSRV_MAXFILESIZE; txdr_hyper(uquad, tl); retnum += NFSX_HYPER; break; case NFSATTRBIT_MAXLINK: NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(LINK_MAX); retnum += NFSX_UNSIGNED; break; case NFSATTRBIT_MAXNAME: NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(NFS_MAXNAMLEN); retnum += NFSX_UNSIGNED; break; case NFSATTRBIT_MAXREAD: NFSM_BUILD(tl, u_int32_t *, NFSX_HYPER); *tl++ = 0; *tl = txdr_unsigned(fsinf.fs_rtmax); retnum += NFSX_HYPER; break; case NFSATTRBIT_MAXWRITE: NFSM_BUILD(tl, u_int32_t *, NFSX_HYPER); *tl++ = 0; *tl = txdr_unsigned(fsinf.fs_wtmax); retnum += NFSX_HYPER; break; case NFSATTRBIT_MODE: NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = vtonfsv34_mode(vap->va_mode); retnum += NFSX_UNSIGNED; break; case NFSATTRBIT_NOTRUNC: NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = newnfs_true; retnum += NFSX_UNSIGNED; break; case NFSATTRBIT_NUMLINKS: NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(vap->va_nlink); retnum += NFSX_UNSIGNED; break; case NFSATTRBIT_OWNER: cp = namestr; nfsv4_uidtostr(vap->va_uid, &cp, &siz, p); retnum += nfsm_strtom(nd, cp, siz); if (cp != namestr) free(cp, M_NFSSTRING); break; case NFSATTRBIT_OWNERGROUP: cp = namestr; nfsv4_gidtostr(vap->va_gid, &cp, &siz, p); retnum += nfsm_strtom(nd, cp, siz); if (cp != namestr) free(cp, M_NFSSTRING); break; case NFSATTRBIT_QUOTAHARD: if (priv_check_cred(cred, PRIV_VFS_EXCEEDQUOTA, 0)) - freenum = fs.f_bfree; + freenum = fs->f_bfree; else - freenum = fs.f_bavail; + freenum = fs->f_bavail; #ifdef QUOTA /* * ufs_quotactl() insists that the uid argument * equal p_ruid for non-root quota access, so * we'll just make sure that's the case. */ savuid = p->p_cred->p_ruid; p->p_cred->p_ruid = cred->cr_uid; if (!VFS_QUOTACTL(mp, QCMD(Q_GETQUOTA,USRQUOTA), cred->cr_uid, (caddr_t)&dqb)) freenum = min(dqb.dqb_bhardlimit, freenum); p->p_cred->p_ruid = savuid; #endif /* QUOTA */ NFSM_BUILD(tl, u_int32_t *, NFSX_HYPER); uquad = (u_int64_t)freenum; - NFSQUOTABLKTOBYTE(uquad, fs.f_bsize); + NFSQUOTABLKTOBYTE(uquad, fs->f_bsize); txdr_hyper(uquad, tl); retnum += NFSX_HYPER; break; case NFSATTRBIT_QUOTASOFT: if (priv_check_cred(cred, PRIV_VFS_EXCEEDQUOTA, 0)) - freenum = fs.f_bfree; + freenum = fs->f_bfree; else - freenum = fs.f_bavail; + freenum = fs->f_bavail; #ifdef QUOTA /* * ufs_quotactl() insists that the uid argument * equal p_ruid for non-root quota access, so * we'll just make sure that's the case. */ savuid = p->p_cred->p_ruid; p->p_cred->p_ruid = cred->cr_uid; if (!VFS_QUOTACTL(mp, QCMD(Q_GETQUOTA,USRQUOTA), cred->cr_uid, (caddr_t)&dqb)) freenum = min(dqb.dqb_bsoftlimit, freenum); p->p_cred->p_ruid = savuid; #endif /* QUOTA */ NFSM_BUILD(tl, u_int32_t *, NFSX_HYPER); uquad = (u_int64_t)freenum; - NFSQUOTABLKTOBYTE(uquad, fs.f_bsize); + NFSQUOTABLKTOBYTE(uquad, fs->f_bsize); txdr_hyper(uquad, tl); retnum += NFSX_HYPER; break; case NFSATTRBIT_QUOTAUSED: freenum = 0; #ifdef QUOTA /* * ufs_quotactl() insists that the uid argument * equal p_ruid for non-root quota access, so * we'll just make sure that's the case. */ savuid = p->p_cred->p_ruid; p->p_cred->p_ruid = cred->cr_uid; if (!VFS_QUOTACTL(mp, QCMD(Q_GETQUOTA,USRQUOTA), cred->cr_uid, (caddr_t)&dqb)) freenum = dqb.dqb_curblocks; p->p_cred->p_ruid = savuid; #endif /* QUOTA */ NFSM_BUILD(tl, u_int32_t *, NFSX_HYPER); uquad = (u_int64_t)freenum; - NFSQUOTABLKTOBYTE(uquad, fs.f_bsize); + NFSQUOTABLKTOBYTE(uquad, fs->f_bsize); txdr_hyper(uquad, tl); retnum += NFSX_HYPER; break; case NFSATTRBIT_RAWDEV: NFSM_BUILD(tl, u_int32_t *, NFSX_V4SPECDATA); *tl++ = txdr_unsigned(NFSMAJOR(vap->va_rdev)); *tl = txdr_unsigned(NFSMINOR(vap->va_rdev)); retnum += NFSX_V4SPECDATA; break; case NFSATTRBIT_SPACEAVAIL: NFSM_BUILD(tl, u_int32_t *, NFSX_HYPER); if (priv_check_cred(cred, PRIV_VFS_BLOCKRESERVE, 0)) - uquad = (u_int64_t)fs.f_bfree; + uquad = (u_int64_t)fs->f_bfree; else - uquad = (u_int64_t)fs.f_bavail; - uquad *= fs.f_bsize; + uquad = (u_int64_t)fs->f_bavail; + uquad *= fs->f_bsize; txdr_hyper(uquad, tl); retnum += NFSX_HYPER; break; case NFSATTRBIT_SPACEFREE: NFSM_BUILD(tl, u_int32_t *, NFSX_HYPER); - uquad = (u_int64_t)fs.f_bfree; - uquad *= fs.f_bsize; + uquad = (u_int64_t)fs->f_bfree; + uquad *= fs->f_bsize; txdr_hyper(uquad, tl); retnum += NFSX_HYPER; break; case NFSATTRBIT_SPACETOTAL: NFSM_BUILD(tl, u_int32_t *, NFSX_HYPER); - uquad = (u_int64_t)fs.f_blocks; - uquad *= fs.f_bsize; + uquad = (u_int64_t)fs->f_blocks; + uquad *= fs->f_bsize; txdr_hyper(uquad, tl); retnum += NFSX_HYPER; break; case NFSATTRBIT_SPACEUSED: NFSM_BUILD(tl, u_int32_t *, NFSX_HYPER); txdr_hyper(vap->va_bytes, tl); retnum += NFSX_HYPER; break; case NFSATTRBIT_TIMEACCESS: NFSM_BUILD(tl, u_int32_t *, NFSX_V4TIME); txdr_nfsv4time(&vap->va_atime, tl); retnum += NFSX_V4TIME; break; case NFSATTRBIT_TIMEACCESSSET: if ((vap->va_vaflags & VA_UTIMES_NULL) == 0) { NFSM_BUILD(tl, u_int32_t *, NFSX_V4SETTIME); *tl++ = txdr_unsigned(NFSV4SATTRTIME_TOCLIENT); txdr_nfsv4time(&vap->va_atime, tl); retnum += NFSX_V4SETTIME; } else { NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(NFSV4SATTRTIME_TOSERVER); retnum += NFSX_UNSIGNED; } break; case NFSATTRBIT_TIMEDELTA: NFSM_BUILD(tl, u_int32_t *, NFSX_V4TIME); temptime.tv_sec = 0; temptime.tv_nsec = 1000000000 / hz; txdr_nfsv4time(&temptime, tl); retnum += NFSX_V4TIME; break; case NFSATTRBIT_TIMEMETADATA: NFSM_BUILD(tl, u_int32_t *, NFSX_V4TIME); txdr_nfsv4time(&vap->va_ctime, tl); retnum += NFSX_V4TIME; break; case NFSATTRBIT_TIMEMODIFY: NFSM_BUILD(tl, u_int32_t *, NFSX_V4TIME); txdr_nfsv4time(&vap->va_mtime, tl); retnum += NFSX_V4TIME; break; case NFSATTRBIT_TIMEMODIFYSET: if ((vap->va_vaflags & VA_UTIMES_NULL) == 0) { NFSM_BUILD(tl, u_int32_t *, NFSX_V4SETTIME); *tl++ = txdr_unsigned(NFSV4SATTRTIME_TOCLIENT); txdr_nfsv4time(&vap->va_mtime, tl); retnum += NFSX_V4SETTIME; } else { NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(NFSV4SATTRTIME_TOSERVER); retnum += NFSX_UNSIGNED; } break; case NFSATTRBIT_MOUNTEDONFILEID: NFSM_BUILD(tl, u_int32_t *, NFSX_HYPER); if (at_root != 0) uquad = mounted_on_fileno; else uquad = (u_int64_t)vap->va_fileid; txdr_hyper(uquad, tl); retnum += NFSX_HYPER; break; case NFSATTRBIT_SUPPATTREXCLCREAT: NFSSETSUPP_ATTRBIT(&attrbits); NFSCLRNOTSETABLE_ATTRBIT(&attrbits); NFSCLRBIT_ATTRBIT(&attrbits, NFSATTRBIT_TIMEACCESSSET); retnum += nfsrv_putattrbit(nd, &attrbits); break; default: printf("EEK! Bad V4 attribute bitpos=%d\n", bitpos); } } } if (naclp != NULL) acl_free(naclp); + free(fs, M_STATFS); *retnump = txdr_unsigned(retnum); return (retnum + prefixnum); } /* * Put the attribute bits onto an mbuf list. * Return the number of bytes of output generated. */ APPLESTATIC int nfsrv_putattrbit(struct nfsrv_descript *nd, nfsattrbit_t *attrbitp) { u_int32_t *tl; int cnt, i, bytesize; for (cnt = NFSATTRBIT_MAXWORDS; cnt > 0; cnt--) if (attrbitp->bits[cnt - 1]) break; bytesize = (cnt + 1) * NFSX_UNSIGNED; NFSM_BUILD(tl, u_int32_t *, bytesize); *tl++ = txdr_unsigned(cnt); for (i = 0; i < cnt; i++) *tl++ = txdr_unsigned(attrbitp->bits[i]); return (bytesize); } /* * Convert a uid to a string. * If the lookup fails, just output the digits. * uid - the user id * cpp - points to a buffer of size NFSV4_SMALLSTR * (malloc a larger one, as required) * retlenp - pointer to length to be returned */ APPLESTATIC void nfsv4_uidtostr(uid_t uid, u_char **cpp, int *retlenp, NFSPROC_T *p) { int i; struct nfsusrgrp *usrp; u_char *cp = *cpp; uid_t tmp; int cnt, hasampersand, len = NFSV4_SMALLSTR, ret; struct nfsrv_lughash *hp; cnt = 0; tryagain: if (nfsrv_dnsnamelen > 0) { /* * Always map nfsrv_defaultuid to "nobody". */ if (uid == nfsrv_defaultuid) { i = nfsrv_dnsnamelen + 7; if (i > len) { if (len > NFSV4_SMALLSTR) free(cp, M_NFSSTRING); cp = malloc(i, M_NFSSTRING, M_WAITOK); *cpp = cp; len = i; goto tryagain; } *retlenp = i; NFSBCOPY("nobody@", cp, 7); cp += 7; NFSBCOPY(nfsrv_dnsname, cp, nfsrv_dnsnamelen); return; } hasampersand = 0; hp = NFSUSERHASH(uid); mtx_lock(&hp->mtx); TAILQ_FOREACH(usrp, &hp->lughead, lug_numhash) { if (usrp->lug_uid == uid) { if (usrp->lug_expiry < NFSD_MONOSEC) break; /* * If the name doesn't already have an '@' * in it, append @domainname to it. */ for (i = 0; i < usrp->lug_namelen; i++) { if (usrp->lug_name[i] == '@') { hasampersand = 1; break; } } if (hasampersand) i = usrp->lug_namelen; else i = usrp->lug_namelen + nfsrv_dnsnamelen + 1; if (i > len) { mtx_unlock(&hp->mtx); if (len > NFSV4_SMALLSTR) free(cp, M_NFSSTRING); cp = malloc(i, M_NFSSTRING, M_WAITOK); *cpp = cp; len = i; goto tryagain; } *retlenp = i; NFSBCOPY(usrp->lug_name, cp, usrp->lug_namelen); if (!hasampersand) { cp += usrp->lug_namelen; *cp++ = '@'; NFSBCOPY(nfsrv_dnsname, cp, nfsrv_dnsnamelen); } TAILQ_REMOVE(&hp->lughead, usrp, lug_numhash); TAILQ_INSERT_TAIL(&hp->lughead, usrp, lug_numhash); mtx_unlock(&hp->mtx); return; } } mtx_unlock(&hp->mtx); cnt++; ret = nfsrv_getuser(RPCNFSUSERD_GETUID, uid, (gid_t)0, NULL, p); if (ret == 0 && cnt < 2) goto tryagain; } /* * No match, just return a string of digits. */ tmp = uid; i = 0; while (tmp || i == 0) { tmp /= 10; i++; } len = (i > len) ? len : i; *retlenp = len; cp += (len - 1); tmp = uid; for (i = 0; i < len; i++) { *cp-- = '0' + (tmp % 10); tmp /= 10; } return; } /* * Get a credential for the uid with the server's group list. * If none is found, just return the credential passed in after * logging a warning message. */ struct ucred * nfsrv_getgrpscred(struct ucred *oldcred) { struct nfsusrgrp *usrp; struct ucred *newcred; int cnt, ret; uid_t uid; struct nfsrv_lughash *hp; cnt = 0; uid = oldcred->cr_uid; tryagain: if (nfsrv_dnsnamelen > 0) { hp = NFSUSERHASH(uid); mtx_lock(&hp->mtx); TAILQ_FOREACH(usrp, &hp->lughead, lug_numhash) { if (usrp->lug_uid == uid) { if (usrp->lug_expiry < NFSD_MONOSEC) break; if (usrp->lug_cred != NULL) { newcred = crhold(usrp->lug_cred); crfree(oldcred); } else newcred = oldcred; TAILQ_REMOVE(&hp->lughead, usrp, lug_numhash); TAILQ_INSERT_TAIL(&hp->lughead, usrp, lug_numhash); mtx_unlock(&hp->mtx); return (newcred); } } mtx_unlock(&hp->mtx); cnt++; ret = nfsrv_getuser(RPCNFSUSERD_GETUID, uid, (gid_t)0, NULL, curthread); if (ret == 0 && cnt < 2) goto tryagain; } return (oldcred); } /* * Convert a string to a uid. * If no conversion is possible return NFSERR_BADOWNER, otherwise * return 0. * If this is called from a client side mount using AUTH_SYS and the * string is made up entirely of digits, just convert the string to * a number. */ APPLESTATIC int nfsv4_strtouid(struct nfsrv_descript *nd, u_char *str, int len, uid_t *uidp, NFSPROC_T *p) { int i; char *cp, *endstr, *str0; struct nfsusrgrp *usrp; int cnt, ret; int error = 0; uid_t tuid; struct nfsrv_lughash *hp, *hp2; if (len == 0) { error = NFSERR_BADOWNER; goto out; } /* If a string of digits and an AUTH_SYS mount, just convert it. */ str0 = str; tuid = (uid_t)strtoul(str0, &endstr, 10); if ((endstr - str0) == len) { /* A numeric string. */ if ((nd->nd_flag & ND_KERBV) == 0 && ((nd->nd_flag & ND_NFSCL) != 0 || nfsd_enable_stringtouid != 0)) *uidp = tuid; else error = NFSERR_BADOWNER; goto out; } /* * Look for an '@'. */ cp = strchr(str0, '@'); if (cp != NULL) i = (int)(cp++ - str0); else i = len; cnt = 0; tryagain: if (nfsrv_dnsnamelen > 0) { /* * If an '@' is found and the domain name matches, search for * the name with dns stripped off. * Mixed case alpahbetics will match for the domain name, but * all upper case will not. */ if (cnt == 0 && i < len && i > 0 && (len - 1 - i) == nfsrv_dnsnamelen && !nfsrv_cmpmixedcase(cp, nfsrv_dnsname, nfsrv_dnsnamelen)) { len -= (nfsrv_dnsnamelen + 1); *(cp - 1) = '\0'; } /* * Check for the special case of "nobody". */ if (len == 6 && !NFSBCMP(str, "nobody", 6)) { *uidp = nfsrv_defaultuid; error = 0; goto out; } hp = NFSUSERNAMEHASH(str, len); mtx_lock(&hp->mtx); TAILQ_FOREACH(usrp, &hp->lughead, lug_namehash) { if (usrp->lug_namelen == len && !NFSBCMP(usrp->lug_name, str, len)) { if (usrp->lug_expiry < NFSD_MONOSEC) break; hp2 = NFSUSERHASH(usrp->lug_uid); mtx_lock(&hp2->mtx); TAILQ_REMOVE(&hp2->lughead, usrp, lug_numhash); TAILQ_INSERT_TAIL(&hp2->lughead, usrp, lug_numhash); *uidp = usrp->lug_uid; mtx_unlock(&hp2->mtx); mtx_unlock(&hp->mtx); error = 0; goto out; } } mtx_unlock(&hp->mtx); cnt++; ret = nfsrv_getuser(RPCNFSUSERD_GETUSER, (uid_t)0, (gid_t)0, str, p); if (ret == 0 && cnt < 2) goto tryagain; } error = NFSERR_BADOWNER; out: NFSEXITCODE(error); return (error); } /* * Convert a gid to a string. * gid - the group id * cpp - points to a buffer of size NFSV4_SMALLSTR * (malloc a larger one, as required) * retlenp - pointer to length to be returned */ APPLESTATIC void nfsv4_gidtostr(gid_t gid, u_char **cpp, int *retlenp, NFSPROC_T *p) { int i; struct nfsusrgrp *usrp; u_char *cp = *cpp; gid_t tmp; int cnt, hasampersand, len = NFSV4_SMALLSTR, ret; struct nfsrv_lughash *hp; cnt = 0; tryagain: if (nfsrv_dnsnamelen > 0) { /* * Always map nfsrv_defaultgid to "nogroup". */ if (gid == nfsrv_defaultgid) { i = nfsrv_dnsnamelen + 8; if (i > len) { if (len > NFSV4_SMALLSTR) free(cp, M_NFSSTRING); cp = malloc(i, M_NFSSTRING, M_WAITOK); *cpp = cp; len = i; goto tryagain; } *retlenp = i; NFSBCOPY("nogroup@", cp, 8); cp += 8; NFSBCOPY(nfsrv_dnsname, cp, nfsrv_dnsnamelen); return; } hasampersand = 0; hp = NFSGROUPHASH(gid); mtx_lock(&hp->mtx); TAILQ_FOREACH(usrp, &hp->lughead, lug_numhash) { if (usrp->lug_gid == gid) { if (usrp->lug_expiry < NFSD_MONOSEC) break; /* * If the name doesn't already have an '@' * in it, append @domainname to it. */ for (i = 0; i < usrp->lug_namelen; i++) { if (usrp->lug_name[i] == '@') { hasampersand = 1; break; } } if (hasampersand) i = usrp->lug_namelen; else i = usrp->lug_namelen + nfsrv_dnsnamelen + 1; if (i > len) { mtx_unlock(&hp->mtx); if (len > NFSV4_SMALLSTR) free(cp, M_NFSSTRING); cp = malloc(i, M_NFSSTRING, M_WAITOK); *cpp = cp; len = i; goto tryagain; } *retlenp = i; NFSBCOPY(usrp->lug_name, cp, usrp->lug_namelen); if (!hasampersand) { cp += usrp->lug_namelen; *cp++ = '@'; NFSBCOPY(nfsrv_dnsname, cp, nfsrv_dnsnamelen); } TAILQ_REMOVE(&hp->lughead, usrp, lug_numhash); TAILQ_INSERT_TAIL(&hp->lughead, usrp, lug_numhash); mtx_unlock(&hp->mtx); return; } } mtx_unlock(&hp->mtx); cnt++; ret = nfsrv_getuser(RPCNFSUSERD_GETGID, (uid_t)0, gid, NULL, p); if (ret == 0 && cnt < 2) goto tryagain; } /* * No match, just return a string of digits. */ tmp = gid; i = 0; while (tmp || i == 0) { tmp /= 10; i++; } len = (i > len) ? len : i; *retlenp = len; cp += (len - 1); tmp = gid; for (i = 0; i < len; i++) { *cp-- = '0' + (tmp % 10); tmp /= 10; } return; } /* * Convert a string to a gid. * If no conversion is possible return NFSERR_BADOWNER, otherwise * return 0. * If this is called from a client side mount using AUTH_SYS and the * string is made up entirely of digits, just convert the string to * a number. */ APPLESTATIC int nfsv4_strtogid(struct nfsrv_descript *nd, u_char *str, int len, gid_t *gidp, NFSPROC_T *p) { int i; char *cp, *endstr, *str0; struct nfsusrgrp *usrp; int cnt, ret; int error = 0; gid_t tgid; struct nfsrv_lughash *hp, *hp2; if (len == 0) { error = NFSERR_BADOWNER; goto out; } /* If a string of digits and an AUTH_SYS mount, just convert it. */ str0 = str; tgid = (gid_t)strtoul(str0, &endstr, 10); if ((endstr - str0) == len) { /* A numeric string. */ if ((nd->nd_flag & ND_KERBV) == 0 && ((nd->nd_flag & ND_NFSCL) != 0 || nfsd_enable_stringtouid != 0)) *gidp = tgid; else error = NFSERR_BADOWNER; goto out; } /* * Look for an '@'. */ cp = strchr(str0, '@'); if (cp != NULL) i = (int)(cp++ - str0); else i = len; cnt = 0; tryagain: if (nfsrv_dnsnamelen > 0) { /* * If an '@' is found and the dns name matches, search for the * name with the dns stripped off. */ if (cnt == 0 && i < len && i > 0 && (len - 1 - i) == nfsrv_dnsnamelen && !nfsrv_cmpmixedcase(cp, nfsrv_dnsname, nfsrv_dnsnamelen)) { len -= (nfsrv_dnsnamelen + 1); *(cp - 1) = '\0'; } /* * Check for the special case of "nogroup". */ if (len == 7 && !NFSBCMP(str, "nogroup", 7)) { *gidp = nfsrv_defaultgid; error = 0; goto out; } hp = NFSGROUPNAMEHASH(str, len); mtx_lock(&hp->mtx); TAILQ_FOREACH(usrp, &hp->lughead, lug_namehash) { if (usrp->lug_namelen == len && !NFSBCMP(usrp->lug_name, str, len)) { if (usrp->lug_expiry < NFSD_MONOSEC) break; hp2 = NFSGROUPHASH(usrp->lug_gid); mtx_lock(&hp2->mtx); TAILQ_REMOVE(&hp2->lughead, usrp, lug_numhash); TAILQ_INSERT_TAIL(&hp2->lughead, usrp, lug_numhash); *gidp = usrp->lug_gid; mtx_unlock(&hp2->mtx); mtx_unlock(&hp->mtx); error = 0; goto out; } } mtx_unlock(&hp->mtx); cnt++; ret = nfsrv_getuser(RPCNFSUSERD_GETGROUP, (uid_t)0, (gid_t)0, str, p); if (ret == 0 && cnt < 2) goto tryagain; } error = NFSERR_BADOWNER; out: NFSEXITCODE(error); return (error); } /* * Cmp len chars, allowing mixed case in the first argument to match lower * case in the second, but not if the first argument is all upper case. * Return 0 for a match, 1 otherwise. */ static int nfsrv_cmpmixedcase(u_char *cp, u_char *cp2, int len) { int i; u_char tmp; int fndlower = 0; for (i = 0; i < len; i++) { if (*cp >= 'A' && *cp <= 'Z') { tmp = *cp++ + ('a' - 'A'); } else { tmp = *cp++; if (tmp >= 'a' && tmp <= 'z') fndlower = 1; } if (tmp != *cp2++) return (1); } if (fndlower) return (0); else return (1); } /* * Set the port for the nfsuserd. */ APPLESTATIC int nfsrv_nfsuserdport(u_short port, NFSPROC_T *p) { struct nfssockreq *rp; struct sockaddr_in *ad; int error; NFSLOCKNAMEID(); if (nfsrv_nfsuserd) { NFSUNLOCKNAMEID(); error = EPERM; goto out; } nfsrv_nfsuserd = 1; NFSUNLOCKNAMEID(); /* * Set up the socket record and connect. */ rp = &nfsrv_nfsuserdsock; rp->nr_client = NULL; rp->nr_sotype = SOCK_DGRAM; rp->nr_soproto = IPPROTO_UDP; rp->nr_lock = (NFSR_RESERVEDPORT | NFSR_LOCALHOST); rp->nr_cred = NULL; NFSSOCKADDRALLOC(rp->nr_nam); NFSSOCKADDRSIZE(rp->nr_nam, sizeof (struct sockaddr_in)); ad = NFSSOCKADDR(rp->nr_nam, struct sockaddr_in *); ad->sin_family = AF_INET; ad->sin_addr.s_addr = htonl((u_int32_t)0x7f000001); /* 127.0.0.1 */ ad->sin_port = port; rp->nr_prog = RPCPROG_NFSUSERD; rp->nr_vers = RPCNFSUSERD_VERS; error = newnfs_connect(NULL, rp, NFSPROCCRED(p), p, 0); if (error) { NFSSOCKADDRFREE(rp->nr_nam); nfsrv_nfsuserd = 0; } out: NFSEXITCODE(error); return (error); } /* * Delete the nfsuserd port. */ APPLESTATIC void nfsrv_nfsuserddelport(void) { NFSLOCKNAMEID(); if (nfsrv_nfsuserd == 0) { NFSUNLOCKNAMEID(); return; } nfsrv_nfsuserd = 0; NFSUNLOCKNAMEID(); newnfs_disconnect(&nfsrv_nfsuserdsock); NFSSOCKADDRFREE(nfsrv_nfsuserdsock.nr_nam); } /* * Do upcalls to the nfsuserd, for cache misses of the owner/ownergroup * name<-->id cache. * Returns 0 upon success, non-zero otherwise. */ static int nfsrv_getuser(int procnum, uid_t uid, gid_t gid, char *name, NFSPROC_T *p) { u_int32_t *tl; struct nfsrv_descript *nd; int len; struct nfsrv_descript nfsd; struct ucred *cred; int error; NFSLOCKNAMEID(); if (nfsrv_nfsuserd == 0) { NFSUNLOCKNAMEID(); error = EPERM; goto out; } NFSUNLOCKNAMEID(); nd = &nfsd; cred = newnfs_getcred(); nd->nd_flag = ND_GSSINITREPLY; nfsrvd_rephead(nd); nd->nd_procnum = procnum; if (procnum == RPCNFSUSERD_GETUID || procnum == RPCNFSUSERD_GETGID) { NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); if (procnum == RPCNFSUSERD_GETUID) *tl = txdr_unsigned(uid); else *tl = txdr_unsigned(gid); } else { len = strlen(name); (void) nfsm_strtom(nd, name, len); } error = newnfs_request(nd, NULL, NULL, &nfsrv_nfsuserdsock, NULL, NULL, cred, RPCPROG_NFSUSERD, RPCNFSUSERD_VERS, NULL, 0, NULL, NULL); NFSFREECRED(cred); if (!error) { mbuf_freem(nd->nd_mrep); error = nd->nd_repstat; } out: NFSEXITCODE(error); return (error); } /* * This function is called from the nfssvc(2) system call, to update the * kernel user/group name list(s) for the V4 owner and ownergroup attributes. */ APPLESTATIC int nfssvc_idname(struct nfsd_idargs *nidp) { struct nfsusrgrp *nusrp, *usrp, *newusrp; struct nfsrv_lughash *hp_name, *hp_idnum, *thp; int i, group_locked, groupname_locked, user_locked, username_locked; int error = 0; u_char *cp; gid_t *grps; struct ucred *cr; static int onethread = 0; static time_t lasttime = 0; if (nidp->nid_namelen <= 0 || nidp->nid_namelen > MAXHOSTNAMELEN) { error = EINVAL; goto out; } if (nidp->nid_flag & NFSID_INITIALIZE) { cp = malloc(nidp->nid_namelen + 1, M_NFSSTRING, M_WAITOK); error = copyin(CAST_USER_ADDR_T(nidp->nid_name), cp, nidp->nid_namelen); if (error != 0) { free(cp, M_NFSSTRING); goto out; } if (atomic_cmpset_acq_int(&nfsrv_dnsnamelen, 0, 0) == 0) { /* * Free up all the old stuff and reinitialize hash * lists. All mutexes for both lists must be locked, * with the user/group name ones before the uid/gid * ones, to avoid a LOR. */ for (i = 0; i < nfsrv_lughashsize; i++) mtx_lock(&nfsusernamehash[i].mtx); for (i = 0; i < nfsrv_lughashsize; i++) mtx_lock(&nfsuserhash[i].mtx); for (i = 0; i < nfsrv_lughashsize; i++) TAILQ_FOREACH_SAFE(usrp, &nfsuserhash[i].lughead, lug_numhash, nusrp) nfsrv_removeuser(usrp, 1); for (i = 0; i < nfsrv_lughashsize; i++) mtx_unlock(&nfsuserhash[i].mtx); for (i = 0; i < nfsrv_lughashsize; i++) mtx_unlock(&nfsusernamehash[i].mtx); for (i = 0; i < nfsrv_lughashsize; i++) mtx_lock(&nfsgroupnamehash[i].mtx); for (i = 0; i < nfsrv_lughashsize; i++) mtx_lock(&nfsgrouphash[i].mtx); for (i = 0; i < nfsrv_lughashsize; i++) TAILQ_FOREACH_SAFE(usrp, &nfsgrouphash[i].lughead, lug_numhash, nusrp) nfsrv_removeuser(usrp, 0); for (i = 0; i < nfsrv_lughashsize; i++) mtx_unlock(&nfsgrouphash[i].mtx); for (i = 0; i < nfsrv_lughashsize; i++) mtx_unlock(&nfsgroupnamehash[i].mtx); free(nfsrv_dnsname, M_NFSSTRING); nfsrv_dnsname = NULL; } if (nfsuserhash == NULL) { /* Allocate the hash tables. */ nfsuserhash = malloc(sizeof(struct nfsrv_lughash) * nfsrv_lughashsize, M_NFSUSERGROUP, M_WAITOK | M_ZERO); for (i = 0; i < nfsrv_lughashsize; i++) mtx_init(&nfsuserhash[i].mtx, "nfsuidhash", NULL, MTX_DEF | MTX_DUPOK); nfsusernamehash = malloc(sizeof(struct nfsrv_lughash) * nfsrv_lughashsize, M_NFSUSERGROUP, M_WAITOK | M_ZERO); for (i = 0; i < nfsrv_lughashsize; i++) mtx_init(&nfsusernamehash[i].mtx, "nfsusrhash", NULL, MTX_DEF | MTX_DUPOK); nfsgrouphash = malloc(sizeof(struct nfsrv_lughash) * nfsrv_lughashsize, M_NFSUSERGROUP, M_WAITOK | M_ZERO); for (i = 0; i < nfsrv_lughashsize; i++) mtx_init(&nfsgrouphash[i].mtx, "nfsgidhash", NULL, MTX_DEF | MTX_DUPOK); nfsgroupnamehash = malloc(sizeof(struct nfsrv_lughash) * nfsrv_lughashsize, M_NFSUSERGROUP, M_WAITOK | M_ZERO); for (i = 0; i < nfsrv_lughashsize; i++) mtx_init(&nfsgroupnamehash[i].mtx, "nfsgrphash", NULL, MTX_DEF | MTX_DUPOK); } /* (Re)initialize the list heads. */ for (i = 0; i < nfsrv_lughashsize; i++) TAILQ_INIT(&nfsuserhash[i].lughead); for (i = 0; i < nfsrv_lughashsize; i++) TAILQ_INIT(&nfsusernamehash[i].lughead); for (i = 0; i < nfsrv_lughashsize; i++) TAILQ_INIT(&nfsgrouphash[i].lughead); for (i = 0; i < nfsrv_lughashsize; i++) TAILQ_INIT(&nfsgroupnamehash[i].lughead); /* * Put name in "DNS" string. */ nfsrv_dnsname = cp; nfsrv_defaultuid = nidp->nid_uid; nfsrv_defaultgid = nidp->nid_gid; nfsrv_usercnt = 0; nfsrv_usermax = nidp->nid_usermax; atomic_store_rel_int(&nfsrv_dnsnamelen, nidp->nid_namelen); goto out; } /* * malloc the new one now, so any potential sleep occurs before * manipulation of the lists. */ newusrp = malloc(sizeof(struct nfsusrgrp) + nidp->nid_namelen, M_NFSUSERGROUP, M_WAITOK | M_ZERO); error = copyin(CAST_USER_ADDR_T(nidp->nid_name), newusrp->lug_name, nidp->nid_namelen); if (error == 0 && nidp->nid_ngroup > 0 && (nidp->nid_flag & NFSID_ADDUID) != 0) { grps = malloc(sizeof(gid_t) * nidp->nid_ngroup, M_TEMP, M_WAITOK); error = copyin(CAST_USER_ADDR_T(nidp->nid_grps), grps, sizeof(gid_t) * nidp->nid_ngroup); if (error == 0) { /* * Create a credential just like svc_getcred(), * but using the group list provided. */ cr = crget(); cr->cr_uid = cr->cr_ruid = cr->cr_svuid = nidp->nid_uid; crsetgroups(cr, nidp->nid_ngroup, grps); cr->cr_rgid = cr->cr_svgid = cr->cr_groups[0]; cr->cr_prison = &prison0; prison_hold(cr->cr_prison); #ifdef MAC mac_cred_associate_nfsd(cr); #endif newusrp->lug_cred = cr; } free(grps, M_TEMP); } if (error) { free(newusrp, M_NFSUSERGROUP); goto out; } newusrp->lug_namelen = nidp->nid_namelen; /* * The lock order is username[0]->[nfsrv_lughashsize - 1] followed * by uid[0]->[nfsrv_lughashsize - 1], with the same for group. * The flags user_locked, username_locked, group_locked and * groupname_locked are set to indicate all of those hash lists are * locked. hp_name != NULL and hp_idnum != NULL indicates that * the respective one mutex is locked. */ user_locked = username_locked = group_locked = groupname_locked = 0; hp_name = hp_idnum = NULL; /* * Delete old entries, as required. */ if (nidp->nid_flag & (NFSID_DELUID | NFSID_ADDUID)) { /* Must lock all username hash lists first, to avoid a LOR. */ for (i = 0; i < nfsrv_lughashsize; i++) mtx_lock(&nfsusernamehash[i].mtx); username_locked = 1; hp_idnum = NFSUSERHASH(nidp->nid_uid); mtx_lock(&hp_idnum->mtx); TAILQ_FOREACH_SAFE(usrp, &hp_idnum->lughead, lug_numhash, nusrp) { if (usrp->lug_uid == nidp->nid_uid) nfsrv_removeuser(usrp, 1); } } else if (nidp->nid_flag & (NFSID_DELUSERNAME | NFSID_ADDUSERNAME)) { hp_name = NFSUSERNAMEHASH(newusrp->lug_name, newusrp->lug_namelen); mtx_lock(&hp_name->mtx); TAILQ_FOREACH_SAFE(usrp, &hp_name->lughead, lug_namehash, nusrp) { if (usrp->lug_namelen == newusrp->lug_namelen && !NFSBCMP(usrp->lug_name, newusrp->lug_name, usrp->lug_namelen)) { thp = NFSUSERHASH(usrp->lug_uid); mtx_lock(&thp->mtx); nfsrv_removeuser(usrp, 1); mtx_unlock(&thp->mtx); } } hp_idnum = NFSUSERHASH(nidp->nid_uid); mtx_lock(&hp_idnum->mtx); } else if (nidp->nid_flag & (NFSID_DELGID | NFSID_ADDGID)) { /* Must lock all groupname hash lists first, to avoid a LOR. */ for (i = 0; i < nfsrv_lughashsize; i++) mtx_lock(&nfsgroupnamehash[i].mtx); groupname_locked = 1; hp_idnum = NFSGROUPHASH(nidp->nid_gid); mtx_lock(&hp_idnum->mtx); TAILQ_FOREACH_SAFE(usrp, &hp_idnum->lughead, lug_numhash, nusrp) { if (usrp->lug_gid == nidp->nid_gid) nfsrv_removeuser(usrp, 0); } } else if (nidp->nid_flag & (NFSID_DELGROUPNAME | NFSID_ADDGROUPNAME)) { hp_name = NFSGROUPNAMEHASH(newusrp->lug_name, newusrp->lug_namelen); mtx_lock(&hp_name->mtx); TAILQ_FOREACH_SAFE(usrp, &hp_name->lughead, lug_namehash, nusrp) { if (usrp->lug_namelen == newusrp->lug_namelen && !NFSBCMP(usrp->lug_name, newusrp->lug_name, usrp->lug_namelen)) { thp = NFSGROUPHASH(usrp->lug_gid); mtx_lock(&thp->mtx); nfsrv_removeuser(usrp, 0); mtx_unlock(&thp->mtx); } } hp_idnum = NFSGROUPHASH(nidp->nid_gid); mtx_lock(&hp_idnum->mtx); } /* * Now, we can add the new one. */ if (nidp->nid_usertimeout) newusrp->lug_expiry = NFSD_MONOSEC + nidp->nid_usertimeout; else newusrp->lug_expiry = NFSD_MONOSEC + 5; if (nidp->nid_flag & (NFSID_ADDUID | NFSID_ADDUSERNAME)) { newusrp->lug_uid = nidp->nid_uid; thp = NFSUSERHASH(newusrp->lug_uid); mtx_assert(&thp->mtx, MA_OWNED); TAILQ_INSERT_TAIL(&thp->lughead, newusrp, lug_numhash); thp = NFSUSERNAMEHASH(newusrp->lug_name, newusrp->lug_namelen); mtx_assert(&thp->mtx, MA_OWNED); TAILQ_INSERT_TAIL(&thp->lughead, newusrp, lug_namehash); atomic_add_int(&nfsrv_usercnt, 1); } else if (nidp->nid_flag & (NFSID_ADDGID | NFSID_ADDGROUPNAME)) { newusrp->lug_gid = nidp->nid_gid; thp = NFSGROUPHASH(newusrp->lug_gid); mtx_assert(&thp->mtx, MA_OWNED); TAILQ_INSERT_TAIL(&thp->lughead, newusrp, lug_numhash); thp = NFSGROUPNAMEHASH(newusrp->lug_name, newusrp->lug_namelen); mtx_assert(&thp->mtx, MA_OWNED); TAILQ_INSERT_TAIL(&thp->lughead, newusrp, lug_namehash); atomic_add_int(&nfsrv_usercnt, 1); } else { if (newusrp->lug_cred != NULL) crfree(newusrp->lug_cred); free(newusrp, M_NFSUSERGROUP); } /* * Once per second, allow one thread to trim the cache. */ if (lasttime < NFSD_MONOSEC && atomic_cmpset_acq_int(&onethread, 0, 1) != 0) { /* * First, unlock the single mutexes, so that all entries * can be locked and any LOR is avoided. */ if (hp_name != NULL) { mtx_unlock(&hp_name->mtx); hp_name = NULL; } if (hp_idnum != NULL) { mtx_unlock(&hp_idnum->mtx); hp_idnum = NULL; } if ((nidp->nid_flag & (NFSID_DELUID | NFSID_ADDUID | NFSID_DELUSERNAME | NFSID_ADDUSERNAME)) != 0) { if (username_locked == 0) { for (i = 0; i < nfsrv_lughashsize; i++) mtx_lock(&nfsusernamehash[i].mtx); username_locked = 1; } KASSERT(user_locked == 0, ("nfssvc_idname: user_locked")); for (i = 0; i < nfsrv_lughashsize; i++) mtx_lock(&nfsuserhash[i].mtx); user_locked = 1; for (i = 0; i < nfsrv_lughashsize; i++) { TAILQ_FOREACH_SAFE(usrp, &nfsuserhash[i].lughead, lug_numhash, nusrp) if (usrp->lug_expiry < NFSD_MONOSEC) nfsrv_removeuser(usrp, 1); } for (i = 0; i < nfsrv_lughashsize; i++) { /* * Trim the cache using an approximate LRU * algorithm. This code deletes the least * recently used entry on each hash list. */ if (nfsrv_usercnt <= nfsrv_usermax) break; usrp = TAILQ_FIRST(&nfsuserhash[i].lughead); if (usrp != NULL) nfsrv_removeuser(usrp, 1); } } else { if (groupname_locked == 0) { for (i = 0; i < nfsrv_lughashsize; i++) mtx_lock(&nfsgroupnamehash[i].mtx); groupname_locked = 1; } KASSERT(group_locked == 0, ("nfssvc_idname: group_locked")); for (i = 0; i < nfsrv_lughashsize; i++) mtx_lock(&nfsgrouphash[i].mtx); group_locked = 1; for (i = 0; i < nfsrv_lughashsize; i++) { TAILQ_FOREACH_SAFE(usrp, &nfsgrouphash[i].lughead, lug_numhash, nusrp) if (usrp->lug_expiry < NFSD_MONOSEC) nfsrv_removeuser(usrp, 0); } for (i = 0; i < nfsrv_lughashsize; i++) { /* * Trim the cache using an approximate LRU * algorithm. This code deletes the least * recently user entry on each hash list. */ if (nfsrv_usercnt <= nfsrv_usermax) break; usrp = TAILQ_FIRST(&nfsgrouphash[i].lughead); if (usrp != NULL) nfsrv_removeuser(usrp, 0); } } lasttime = NFSD_MONOSEC; atomic_store_rel_int(&onethread, 0); } /* Now, unlock all locked mutexes. */ if (hp_idnum != NULL) mtx_unlock(&hp_idnum->mtx); if (hp_name != NULL) mtx_unlock(&hp_name->mtx); if (user_locked != 0) for (i = 0; i < nfsrv_lughashsize; i++) mtx_unlock(&nfsuserhash[i].mtx); if (username_locked != 0) for (i = 0; i < nfsrv_lughashsize; i++) mtx_unlock(&nfsusernamehash[i].mtx); if (group_locked != 0) for (i = 0; i < nfsrv_lughashsize; i++) mtx_unlock(&nfsgrouphash[i].mtx); if (groupname_locked != 0) for (i = 0; i < nfsrv_lughashsize; i++) mtx_unlock(&nfsgroupnamehash[i].mtx); out: NFSEXITCODE(error); return (error); } /* * Remove a user/group name element. */ static void nfsrv_removeuser(struct nfsusrgrp *usrp, int isuser) { struct nfsrv_lughash *hp; if (isuser != 0) { hp = NFSUSERHASH(usrp->lug_uid); mtx_assert(&hp->mtx, MA_OWNED); TAILQ_REMOVE(&hp->lughead, usrp, lug_numhash); hp = NFSUSERNAMEHASH(usrp->lug_name, usrp->lug_namelen); mtx_assert(&hp->mtx, MA_OWNED); TAILQ_REMOVE(&hp->lughead, usrp, lug_namehash); } else { hp = NFSGROUPHASH(usrp->lug_gid); mtx_assert(&hp->mtx, MA_OWNED); TAILQ_REMOVE(&hp->lughead, usrp, lug_numhash); hp = NFSGROUPNAMEHASH(usrp->lug_name, usrp->lug_namelen); mtx_assert(&hp->mtx, MA_OWNED); TAILQ_REMOVE(&hp->lughead, usrp, lug_namehash); } atomic_add_int(&nfsrv_usercnt, -1); if (usrp->lug_cred != NULL) crfree(usrp->lug_cred); free(usrp, M_NFSUSERGROUP); } /* * Free up all the allocations related to the name<-->id cache. * This function should only be called when the nfsuserd daemon isn't * running, since it doesn't do any locking. * This function is meant to be used when the nfscommon module is unloaded. */ APPLESTATIC void nfsrv_cleanusergroup(void) { struct nfsrv_lughash *hp, *hp2; struct nfsusrgrp *nusrp, *usrp; int i; if (nfsuserhash == NULL) return; for (i = 0; i < nfsrv_lughashsize; i++) { hp = &nfsuserhash[i]; TAILQ_FOREACH_SAFE(usrp, &hp->lughead, lug_numhash, nusrp) { TAILQ_REMOVE(&hp->lughead, usrp, lug_numhash); hp2 = NFSUSERNAMEHASH(usrp->lug_name, usrp->lug_namelen); TAILQ_REMOVE(&hp2->lughead, usrp, lug_namehash); if (usrp->lug_cred != NULL) crfree(usrp->lug_cred); free(usrp, M_NFSUSERGROUP); } hp = &nfsgrouphash[i]; TAILQ_FOREACH_SAFE(usrp, &hp->lughead, lug_numhash, nusrp) { TAILQ_REMOVE(&hp->lughead, usrp, lug_numhash); hp2 = NFSGROUPNAMEHASH(usrp->lug_name, usrp->lug_namelen); TAILQ_REMOVE(&hp2->lughead, usrp, lug_namehash); if (usrp->lug_cred != NULL) crfree(usrp->lug_cred); free(usrp, M_NFSUSERGROUP); } mtx_destroy(&nfsuserhash[i].mtx); mtx_destroy(&nfsusernamehash[i].mtx); mtx_destroy(&nfsgroupnamehash[i].mtx); mtx_destroy(&nfsgrouphash[i].mtx); } free(nfsuserhash, M_NFSUSERGROUP); free(nfsusernamehash, M_NFSUSERGROUP); free(nfsgrouphash, M_NFSUSERGROUP); free(nfsgroupnamehash, M_NFSUSERGROUP); free(nfsrv_dnsname, M_NFSSTRING); } /* * This function scans a byte string and checks for UTF-8 compliance. * It returns 0 if it conforms and NFSERR_INVAL if not. */ APPLESTATIC int nfsrv_checkutf8(u_int8_t *cp, int len) { u_int32_t val = 0x0; int cnt = 0, gotd = 0, shift = 0; u_int8_t byte; static int utf8_shift[5] = { 7, 11, 16, 21, 26 }; int error = 0; /* * Here are what the variables are used for: * val - the calculated value of a multibyte char, used to check * that it was coded with the correct range * cnt - the number of 10xxxxxx bytes to follow * gotd - set for a char of Dxxx, so D800<->DFFF can be checked for * shift - lower order bits of range (ie. "val >> shift" should * not be 0, in other words, dividing by the lower bound * of the range should get a non-zero value) * byte - used to calculate cnt */ while (len > 0) { if (cnt > 0) { /* This handles the 10xxxxxx bytes */ if ((*cp & 0xc0) != 0x80 || (gotd && (*cp & 0x20))) { error = NFSERR_INVAL; goto out; } gotd = 0; val <<= 6; val |= (*cp & 0x3f); cnt--; if (cnt == 0 && (val >> shift) == 0x0) { error = NFSERR_INVAL; goto out; } } else if (*cp & 0x80) { /* first byte of multi byte char */ byte = *cp; while ((byte & 0x40) && cnt < 6) { cnt++; byte <<= 1; } if (cnt == 0 || cnt == 6) { error = NFSERR_INVAL; goto out; } val = (*cp & (0x3f >> cnt)); shift = utf8_shift[cnt - 1]; if (cnt == 2 && val == 0xd) /* Check for the 0xd800-0xdfff case */ gotd = 1; } cp++; len--; } if (cnt > 0) error = NFSERR_INVAL; out: NFSEXITCODE(error); return (error); } /* * Parse the xdr for an NFSv4 FsLocations attribute. Return two malloc'd * strings, one with the root path in it and the other with the list of * locations. The list is in the same format as is found in nfr_refs. * It is a "," separated list of entries, where each of them is of the * form :. For example * "nfsv4-test:/sub2,nfsv4-test2:/user/mnt,nfsv4-test2:/user/mnt2" * The nilp argument is set to 1 for the special case of a null fs_root * and an empty server list. * It returns NFSERR_BADXDR, if the xdr can't be parsed and returns the * number of xdr bytes parsed in sump. */ static int nfsrv_getrefstr(struct nfsrv_descript *nd, u_char **fsrootp, u_char **srvp, int *sump, int *nilp) { u_int32_t *tl; u_char *cp = NULL, *cp2 = NULL, *cp3, *str; int i, j, len, stringlen, cnt, slen, siz, xdrsum, error = 0, nsrv; struct list { SLIST_ENTRY(list) next; int len; u_char host[1]; } *lsp, *nlsp; SLIST_HEAD(, list) head; *fsrootp = NULL; *srvp = NULL; *nilp = 0; /* * Get the fs_root path and check for the special case of null path * and 0 length server list. */ NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); len = fxdr_unsigned(int, *tl); if (len < 0 || len > 10240) { error = NFSERR_BADXDR; goto nfsmout; } if (len == 0) { NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); if (*tl != 0) { error = NFSERR_BADXDR; goto nfsmout; } *nilp = 1; *sump = 2 * NFSX_UNSIGNED; error = 0; goto nfsmout; } cp = malloc(len + 1, M_NFSSTRING, M_WAITOK); error = nfsrv_mtostr(nd, cp, len); if (!error) { NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); cnt = fxdr_unsigned(int, *tl); if (cnt <= 0) error = NFSERR_BADXDR; } if (error) goto nfsmout; /* * Now, loop through the location list and make up the srvlist. */ xdrsum = (2 * NFSX_UNSIGNED) + NFSM_RNDUP(len); cp2 = cp3 = malloc(1024, M_NFSSTRING, M_WAITOK); slen = 1024; siz = 0; for (i = 0; i < cnt; i++) { SLIST_INIT(&head); NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); nsrv = fxdr_unsigned(int, *tl); if (nsrv <= 0) { error = NFSERR_BADXDR; goto nfsmout; } /* * Handle the first server by putting it in the srvstr. */ NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); len = fxdr_unsigned(int, *tl); if (len <= 0 || len > 1024) { error = NFSERR_BADXDR; goto nfsmout; } nfsrv_refstrbigenough(siz + len + 3, &cp2, &cp3, &slen); if (cp3 != cp2) { *cp3++ = ','; siz++; } error = nfsrv_mtostr(nd, cp3, len); if (error) goto nfsmout; cp3 += len; *cp3++ = ':'; siz += (len + 1); xdrsum += (2 * NFSX_UNSIGNED) + NFSM_RNDUP(len); for (j = 1; j < nsrv; j++) { /* * Yuck, put them in an slist and process them later. */ NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); len = fxdr_unsigned(int, *tl); if (len <= 0 || len > 1024) { error = NFSERR_BADXDR; goto nfsmout; } lsp = (struct list *)malloc(sizeof (struct list) + len, M_TEMP, M_WAITOK); error = nfsrv_mtostr(nd, lsp->host, len); if (error) goto nfsmout; xdrsum += NFSX_UNSIGNED + NFSM_RNDUP(len); lsp->len = len; SLIST_INSERT_HEAD(&head, lsp, next); } /* * Finally, we can get the path. */ NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); len = fxdr_unsigned(int, *tl); if (len <= 0 || len > 1024) { error = NFSERR_BADXDR; goto nfsmout; } nfsrv_refstrbigenough(siz + len + 1, &cp2, &cp3, &slen); error = nfsrv_mtostr(nd, cp3, len); if (error) goto nfsmout; xdrsum += NFSX_UNSIGNED + NFSM_RNDUP(len); str = cp3; stringlen = len; cp3 += len; siz += len; SLIST_FOREACH_SAFE(lsp, &head, next, nlsp) { nfsrv_refstrbigenough(siz + lsp->len + stringlen + 3, &cp2, &cp3, &slen); *cp3++ = ','; NFSBCOPY(lsp->host, cp3, lsp->len); cp3 += lsp->len; *cp3++ = ':'; NFSBCOPY(str, cp3, stringlen); cp3 += stringlen; *cp3 = '\0'; siz += (lsp->len + stringlen + 2); free((caddr_t)lsp, M_TEMP); } } *fsrootp = cp; *srvp = cp2; *sump = xdrsum; NFSEXITCODE2(0, nd); return (0); nfsmout: if (cp != NULL) free(cp, M_NFSSTRING); if (cp2 != NULL) free(cp2, M_NFSSTRING); NFSEXITCODE2(error, nd); return (error); } /* * Make the malloc'd space large enough. This is a pain, but the xdr * doesn't set an upper bound on the side, so... */ static void nfsrv_refstrbigenough(int siz, u_char **cpp, u_char **cpp2, int *slenp) { u_char *cp; int i; if (siz <= *slenp) return; cp = malloc(siz + 1024, M_NFSSTRING, M_WAITOK); NFSBCOPY(*cpp, cp, *slenp); free(*cpp, M_NFSSTRING); i = *cpp2 - *cpp; *cpp = cp; *cpp2 = cp + i; *slenp = siz + 1024; } /* * Initialize the reply header data structures. */ APPLESTATIC void nfsrvd_rephead(struct nfsrv_descript *nd) { mbuf_t mreq; /* * If this is a big reply, use a cluster. */ if ((nd->nd_flag & ND_GSSINITREPLY) == 0 && nfs_bigreply[nd->nd_procnum]) { NFSMCLGET(mreq, M_WAITOK); nd->nd_mreq = mreq; nd->nd_mb = mreq; } else { NFSMGET(mreq); nd->nd_mreq = mreq; nd->nd_mb = mreq; } nd->nd_bpos = NFSMTOD(mreq, caddr_t); mbuf_setlen(mreq, 0); if ((nd->nd_flag & ND_GSSINITREPLY) == 0) NFSM_BUILD(nd->nd_errp, int *, NFSX_UNSIGNED); } /* * Lock a socket against others. * Currently used to serialize connect/disconnect attempts. */ int newnfs_sndlock(int *flagp) { struct timespec ts; NFSLOCKSOCK(); while (*flagp & NFSR_SNDLOCK) { *flagp |= NFSR_WANTSND; ts.tv_sec = 0; ts.tv_nsec = 0; (void) nfsmsleep((caddr_t)flagp, NFSSOCKMUTEXPTR, PZERO - 1, "nfsndlck", &ts); } *flagp |= NFSR_SNDLOCK; NFSUNLOCKSOCK(); return (0); } /* * Unlock the stream socket for others. */ void newnfs_sndunlock(int *flagp) { NFSLOCKSOCK(); if ((*flagp & NFSR_SNDLOCK) == 0) panic("nfs sndunlock"); *flagp &= ~NFSR_SNDLOCK; if (*flagp & NFSR_WANTSND) { *flagp &= ~NFSR_WANTSND; wakeup((caddr_t)flagp); } NFSUNLOCKSOCK(); } APPLESTATIC int nfsv4_getipaddr(struct nfsrv_descript *nd, struct sockaddr_storage *sa, int *isudp) { struct sockaddr_in *sad; struct sockaddr_in6 *sad6; struct in_addr saddr; uint32_t portnum, *tl; int af = 0, i, j, k; char addr[64], protocol[5], *cp; int cantparse = 0, error = 0; uint16_t portv; NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); i = fxdr_unsigned(int, *tl); if (i >= 3 && i <= 4) { error = nfsrv_mtostr(nd, protocol, i); if (error) goto nfsmout; if (strcmp(protocol, "tcp") == 0) { af = AF_INET; *isudp = 0; } else if (strcmp(protocol, "udp") == 0) { af = AF_INET; *isudp = 1; } else if (strcmp(protocol, "tcp6") == 0) { af = AF_INET6; *isudp = 0; } else if (strcmp(protocol, "udp6") == 0) { af = AF_INET6; *isudp = 1; } else cantparse = 1; } else { cantparse = 1; if (i > 0) { error = nfsm_advance(nd, NFSM_RNDUP(i), -1); if (error) goto nfsmout; } } NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); i = fxdr_unsigned(int, *tl); if (i < 0) { error = NFSERR_BADXDR; goto nfsmout; } else if (cantparse == 0 && i >= 11 && i < 64) { /* * The shortest address is 11chars and the longest is < 64. */ error = nfsrv_mtostr(nd, addr, i); if (error) goto nfsmout; /* Find the port# at the end and extract that. */ i = strlen(addr); k = 0; cp = &addr[i - 1]; /* Count back two '.'s from end to get port# field. */ for (j = 0; j < i; j++) { if (*cp == '.') { k++; if (k == 2) break; } cp--; } if (k == 2) { /* * The NFSv4 port# is appended as .N.N, where N is * a decimal # in the range 0-255, just like an inet4 * address. Cheat and use inet_aton(), which will * return a Class A address and then shift the high * order 8bits over to convert it to the port#. */ *cp++ = '\0'; if (inet_aton(cp, &saddr) == 1) { portnum = ntohl(saddr.s_addr); portv = (uint16_t)((portnum >> 16) | (portnum & 0xff)); } else cantparse = 1; } else cantparse = 1; if (cantparse == 0) { if (af == AF_INET) { sad = (struct sockaddr_in *)sa; if (inet_pton(af, addr, &sad->sin_addr) == 1) { sad->sin_len = sizeof(*sad); sad->sin_family = AF_INET; sad->sin_port = htons(portv); return (0); } } else { sad6 = (struct sockaddr_in6 *)sa; if (inet_pton(af, addr, &sad6->sin6_addr) == 1) { sad6->sin6_len = sizeof(*sad6); sad6->sin6_family = AF_INET6; sad6->sin6_port = htons(portv); return (0); } } } } else { if (i > 0) { error = nfsm_advance(nd, NFSM_RNDUP(i), -1); if (error) goto nfsmout; } } error = EPERM; nfsmout: return (error); } /* * Handle an NFSv4.1 Sequence request for the session. * If reply != NULL, use it to return the cached reply, as required. * The client gets a cached reply via this call for callbacks, however the * server gets a cached reply via the nfsv4_seqsess_cachereply() call. */ int nfsv4_seqsession(uint32_t seqid, uint32_t slotid, uint32_t highslot, struct nfsslot *slots, struct mbuf **reply, uint16_t maxslot) { int error; error = 0; if (reply != NULL) *reply = NULL; if (slotid > maxslot) return (NFSERR_BADSLOT); if (seqid == slots[slotid].nfssl_seq) { /* A retry. */ if (slots[slotid].nfssl_inprog != 0) error = NFSERR_DELAY; else if (slots[slotid].nfssl_reply != NULL) { if (reply != NULL) { *reply = slots[slotid].nfssl_reply; slots[slotid].nfssl_reply = NULL; } slots[slotid].nfssl_inprog = 1; error = NFSERR_REPLYFROMCACHE; } else /* No reply cached, so just do it. */ slots[slotid].nfssl_inprog = 1; } else if ((slots[slotid].nfssl_seq + 1) == seqid) { if (slots[slotid].nfssl_reply != NULL) m_freem(slots[slotid].nfssl_reply); slots[slotid].nfssl_reply = NULL; slots[slotid].nfssl_inprog = 1; slots[slotid].nfssl_seq++; } else error = NFSERR_SEQMISORDERED; return (error); } /* * Cache this reply for the slot. * Use the "rep" argument to return the cached reply if repstat is set to * NFSERR_REPLYFROMCACHE. The client never sets repstat to this value. */ void nfsv4_seqsess_cacherep(uint32_t slotid, struct nfsslot *slots, int repstat, struct mbuf **rep) { if (repstat == NFSERR_REPLYFROMCACHE) { *rep = slots[slotid].nfssl_reply; slots[slotid].nfssl_reply = NULL; } else { if (slots[slotid].nfssl_reply != NULL) m_freem(slots[slotid].nfssl_reply); slots[slotid].nfssl_reply = *rep; } slots[slotid].nfssl_inprog = 0; } /* * Generate the xdr for an NFSv4.1 Sequence Operation. */ APPLESTATIC void nfsv4_setsequence(struct nfsmount *nmp, struct nfsrv_descript *nd, struct nfsclsession *sep, int dont_replycache) { uint32_t *tl, slotseq = 0; int error, maxslot, slotpos; uint8_t sessionid[NFSX_V4SESSIONID]; error = nfsv4_sequencelookup(nmp, sep, &slotpos, &maxslot, &slotseq, sessionid); /* Build the Sequence arguments. */ NFSM_BUILD(tl, uint32_t *, NFSX_V4SESSIONID + 4 * NFSX_UNSIGNED); nd->nd_sequence = tl; bcopy(sessionid, tl, NFSX_V4SESSIONID); tl += NFSX_V4SESSIONID / NFSX_UNSIGNED; nd->nd_slotseq = tl; if (error == 0) { *tl++ = txdr_unsigned(slotseq); *tl++ = txdr_unsigned(slotpos); *tl++ = txdr_unsigned(maxslot); if (dont_replycache == 0) *tl = newnfs_true; else *tl = newnfs_false; } else { /* * There are two errors and the rest of the session can * just be zeros. * NFSERR_BADSESSION: This bad session should just generate * the same error again when the RPC is retried. * ESTALE: A forced dismount is in progress and will cause the * RPC to fail later. */ *tl++ = 0; *tl++ = 0; *tl++ = 0; *tl = 0; } nd->nd_flag |= ND_HASSEQUENCE; } int nfsv4_sequencelookup(struct nfsmount *nmp, struct nfsclsession *sep, int *slotposp, int *maxslotp, uint32_t *slotseqp, uint8_t *sessionid) { int i, maxslot, slotpos; uint64_t bitval; /* Find an unused slot. */ slotpos = -1; maxslot = -1; mtx_lock(&sep->nfsess_mtx); do { if (nmp != NULL && sep->nfsess_defunct != 0) { /* Just return the bad session. */ bcopy(sep->nfsess_sessionid, sessionid, NFSX_V4SESSIONID); mtx_unlock(&sep->nfsess_mtx); return (NFSERR_BADSESSION); } bitval = 1; for (i = 0; i < sep->nfsess_foreslots; i++) { if ((bitval & sep->nfsess_slots) == 0) { slotpos = i; sep->nfsess_slots |= bitval; sep->nfsess_slotseq[i]++; *slotseqp = sep->nfsess_slotseq[i]; break; } bitval <<= 1; } if (slotpos == -1) { /* * If a forced dismount is in progress, just return. * This RPC attempt will fail when it calls * newnfs_request(). */ if (nmp != NULL && (nmp->nm_mountp->mnt_kern_flag & MNTK_UNMOUNTF) != 0) { mtx_unlock(&sep->nfsess_mtx); return (ESTALE); } /* Wake up once/sec, to check for a forced dismount. */ (void)mtx_sleep(&sep->nfsess_slots, &sep->nfsess_mtx, PZERO, "nfsclseq", hz); } } while (slotpos == -1); /* Now, find the highest slot in use. (nfsc_slots is 64bits) */ bitval = 1; for (i = 0; i < 64; i++) { if ((bitval & sep->nfsess_slots) != 0) maxslot = i; bitval <<= 1; } bcopy(sep->nfsess_sessionid, sessionid, NFSX_V4SESSIONID); mtx_unlock(&sep->nfsess_mtx); *slotposp = slotpos; *maxslotp = maxslot; return (0); } /* * Free a session slot. */ APPLESTATIC void nfsv4_freeslot(struct nfsclsession *sep, int slot) { uint64_t bitval; bitval = 1; if (slot > 0) bitval <<= slot; mtx_lock(&sep->nfsess_mtx); if ((bitval & sep->nfsess_slots) == 0) printf("freeing free slot!!\n"); sep->nfsess_slots &= ~bitval; wakeup(&sep->nfsess_slots); mtx_unlock(&sep->nfsess_mtx); } Index: projects/clang400-import/sys/fs/nfsserver/nfs_nfsdserv.c =================================================================== --- projects/clang400-import/sys/fs/nfsserver/nfs_nfsdserv.c (revision 311459) +++ projects/clang400-import/sys/fs/nfsserver/nfs_nfsdserv.c (revision 311460) @@ -1,4098 +1,4101 @@ /*- * 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("$FreeBSD$"); /* * nfs version 2, 3 and 4 server calls to vnode ops * - these routines generally have 3 phases * 1 - break down and validate rpc request in mbuf list * 2 - do the vnode ops for the request, usually by calling a nfsvno_XXX() * function in nfsd_port.c * 3 - build the rpc reply in an mbuf list * For nfsv4, these functions are called for each Op within the Compound RPC. */ #ifndef APPLEKEXT #include /* Global vars */ extern u_int32_t newnfs_false, newnfs_true; extern enum vtype nv34tov_type[8]; extern struct timeval nfsboottime; extern int nfs_rootfhset; extern int nfsrv_enable_crossmntpt; extern int nfsrv_statehashsize; #endif /* !APPLEKEXT */ static int nfs_async = 0; SYSCTL_DECL(_vfs_nfsd); SYSCTL_INT(_vfs_nfsd, OID_AUTO, async, CTLFLAG_RW, &nfs_async, 0, "Tell client that writes were synced even though they were not"); /* * This list defines the GSS mechanisms supported. * (Don't ask me how you get these strings from the RFC stuff like * iso(1), org(3)... but someone did it, so I don't need to know.) */ static struct nfsgss_mechlist nfsgss_mechlist[] = { { 9, "\052\206\110\206\367\022\001\002\002", 11 }, { 0, "", 0 }, }; /* local functions */ static void nfsrvd_symlinksub(struct nfsrv_descript *nd, struct nameidata *ndp, struct nfsvattr *nvap, fhandle_t *fhp, vnode_t *vpp, vnode_t dirp, struct nfsvattr *dirforp, struct nfsvattr *diraftp, int *diraft_retp, nfsattrbit_t *attrbitp, NFSACL_T *aclp, NFSPROC_T *p, struct nfsexstuff *exp, char *pathcp, int pathlen); static void nfsrvd_mkdirsub(struct nfsrv_descript *nd, struct nameidata *ndp, struct nfsvattr *nvap, fhandle_t *fhp, vnode_t *vpp, vnode_t dirp, struct nfsvattr *dirforp, struct nfsvattr *diraftp, int *diraft_retp, nfsattrbit_t *attrbitp, NFSACL_T *aclp, NFSPROC_T *p, struct nfsexstuff *exp); /* * nfs access service (not a part of NFS V2) */ APPLESTATIC int nfsrvd_access(struct nfsrv_descript *nd, __unused int isdgram, vnode_t vp, NFSPROC_T *p, struct nfsexstuff *exp) { u_int32_t *tl; int getret, error = 0; struct nfsvattr nva; u_int32_t testmode, nfsmode, supported = 0; accmode_t deletebit; if (nd->nd_repstat) { nfsrv_postopattr(nd, 1, &nva); goto out; } NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); nfsmode = fxdr_unsigned(u_int32_t, *tl); if ((nd->nd_flag & ND_NFSV4) && (nfsmode & ~(NFSACCESS_READ | NFSACCESS_LOOKUP | NFSACCESS_MODIFY | NFSACCESS_EXTEND | NFSACCESS_DELETE | NFSACCESS_EXECUTE))) { nd->nd_repstat = NFSERR_INVAL; vput(vp); goto out; } if (nfsmode & NFSACCESS_READ) { supported |= NFSACCESS_READ; if (nfsvno_accchk(vp, VREAD, nd->nd_cred, exp, p, NFSACCCHK_NOOVERRIDE, NFSACCCHK_VPISLOCKED, &supported)) nfsmode &= ~NFSACCESS_READ; } if (nfsmode & NFSACCESS_MODIFY) { supported |= NFSACCESS_MODIFY; if (nfsvno_accchk(vp, VWRITE, nd->nd_cred, exp, p, NFSACCCHK_NOOVERRIDE, NFSACCCHK_VPISLOCKED, &supported)) nfsmode &= ~NFSACCESS_MODIFY; } if (nfsmode & NFSACCESS_EXTEND) { supported |= NFSACCESS_EXTEND; if (nfsvno_accchk(vp, VWRITE | VAPPEND, nd->nd_cred, exp, p, NFSACCCHK_NOOVERRIDE, NFSACCCHK_VPISLOCKED, &supported)) nfsmode &= ~NFSACCESS_EXTEND; } if (nfsmode & NFSACCESS_DELETE) { supported |= NFSACCESS_DELETE; if (vp->v_type == VDIR) deletebit = VDELETE_CHILD; else deletebit = VDELETE; if (nfsvno_accchk(vp, deletebit, nd->nd_cred, exp, p, NFSACCCHK_NOOVERRIDE, NFSACCCHK_VPISLOCKED, &supported)) nfsmode &= ~NFSACCESS_DELETE; } if (vnode_vtype(vp) == VDIR) testmode = NFSACCESS_LOOKUP; else testmode = NFSACCESS_EXECUTE; if (nfsmode & testmode) { supported |= (nfsmode & testmode); if (nfsvno_accchk(vp, VEXEC, nd->nd_cred, exp, p, NFSACCCHK_NOOVERRIDE, NFSACCCHK_VPISLOCKED, &supported)) nfsmode &= ~testmode; } nfsmode &= supported; if (nd->nd_flag & ND_NFSV3) { getret = nfsvno_getattr(vp, &nva, nd->nd_cred, p, 1); nfsrv_postopattr(nd, getret, &nva); } vput(vp); if (nd->nd_flag & ND_NFSV4) { NFSM_BUILD(tl, u_int32_t *, 2 * NFSX_UNSIGNED); *tl++ = txdr_unsigned(supported); } else NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(nfsmode); out: NFSEXITCODE2(0, nd); return (0); nfsmout: vput(vp); NFSEXITCODE2(error, nd); return (error); } /* * nfs getattr service */ APPLESTATIC int nfsrvd_getattr(struct nfsrv_descript *nd, int isdgram, vnode_t vp, NFSPROC_T *p, __unused struct nfsexstuff *exp) { struct nfsvattr nva; fhandle_t fh; int at_root = 0, error = 0, supports_nfsv4acls; struct nfsreferral *refp; nfsattrbit_t attrbits, tmpbits; struct mount *mp; struct vnode *tvp = NULL; struct vattr va; uint64_t mounted_on_fileno = 0; accmode_t accmode; if (nd->nd_repstat) goto out; if (nd->nd_flag & ND_NFSV4) { error = nfsrv_getattrbits(nd, &attrbits, NULL, NULL); if (error) { vput(vp); goto out; } /* * Check for a referral. */ refp = nfsv4root_getreferral(vp, NULL, 0); if (refp != NULL) { (void) nfsrv_putreferralattr(nd, &attrbits, refp, 1, &nd->nd_repstat); vput(vp); goto out; } if (nd->nd_repstat == 0) { accmode = 0; NFSSET_ATTRBIT(&tmpbits, &attrbits); /* * GETATTR with write-only attr time_access_set and time_modify_set * should return NFS4ERR_INVAL. */ if (NFSISSET_ATTRBIT(&tmpbits, NFSATTRBIT_TIMEACCESSSET) || NFSISSET_ATTRBIT(&tmpbits, NFSATTRBIT_TIMEMODIFYSET)){ error = NFSERR_INVAL; vput(vp); goto out; } if (NFSISSET_ATTRBIT(&tmpbits, NFSATTRBIT_ACL)) { NFSCLRBIT_ATTRBIT(&tmpbits, NFSATTRBIT_ACL); accmode |= VREAD_ACL; } if (NFSNONZERO_ATTRBIT(&tmpbits)) accmode |= VREAD_ATTRIBUTES; if (accmode != 0) nd->nd_repstat = nfsvno_accchk(vp, accmode, nd->nd_cred, exp, p, NFSACCCHK_NOOVERRIDE, NFSACCCHK_VPISLOCKED, NULL); } } if (!nd->nd_repstat) nd->nd_repstat = nfsvno_getattr(vp, &nva, nd->nd_cred, p, 1); if (!nd->nd_repstat) { if (nd->nd_flag & ND_NFSV4) { if (NFSISSET_ATTRBIT(&attrbits, NFSATTRBIT_FILEHANDLE)) nd->nd_repstat = nfsvno_getfh(vp, &fh, p); if (!nd->nd_repstat) nd->nd_repstat = nfsrv_checkgetattr(nd, vp, &nva, &attrbits, nd->nd_cred, p); if (nd->nd_repstat == 0) { supports_nfsv4acls = nfs_supportsnfsv4acls(vp); mp = vp->v_mount; if (nfsrv_enable_crossmntpt != 0 && vp->v_type == VDIR && (vp->v_vflag & VV_ROOT) != 0 && vp != rootvnode) { tvp = mp->mnt_vnodecovered; VREF(tvp); at_root = 1; } else at_root = 0; vfs_ref(mp); NFSVOPUNLOCK(vp, 0); if (at_root != 0) { if ((nd->nd_repstat = NFSVOPLOCK(tvp, LK_SHARED)) == 0) { nd->nd_repstat = VOP_GETATTR( tvp, &va, nd->nd_cred); vput(tvp); } else vrele(tvp); if (nd->nd_repstat == 0) mounted_on_fileno = (uint64_t) va.va_fileid; else at_root = 0; } if (nd->nd_repstat == 0) nd->nd_repstat = vfs_busy(mp, 0); vfs_rel(mp); if (nd->nd_repstat == 0) { (void)nfsvno_fillattr(nd, mp, vp, &nva, &fh, 0, &attrbits, nd->nd_cred, p, isdgram, 1, supports_nfsv4acls, at_root, mounted_on_fileno); vfs_unbusy(mp); } vrele(vp); } else vput(vp); } else { nfsrv_fillattr(nd, &nva); vput(vp); } } else { vput(vp); } out: NFSEXITCODE2(error, nd); return (error); } /* * nfs setattr service */ APPLESTATIC int nfsrvd_setattr(struct nfsrv_descript *nd, __unused int isdgram, vnode_t vp, NFSPROC_T *p, struct nfsexstuff *exp) { struct nfsvattr nva, nva2; u_int32_t *tl; int preat_ret = 1, postat_ret = 1, gcheck = 0, error = 0; struct timespec guard = { 0, 0 }; nfsattrbit_t attrbits, retbits; nfsv4stateid_t stateid; NFSACL_T *aclp = NULL; if (nd->nd_repstat) { nfsrv_wcc(nd, preat_ret, &nva2, postat_ret, &nva); goto out; } #ifdef NFS4_ACL_EXTATTR_NAME aclp = acl_alloc(M_WAITOK); aclp->acl_cnt = 0; #endif NFSVNO_ATTRINIT(&nva); NFSZERO_ATTRBIT(&retbits); if (nd->nd_flag & ND_NFSV4) { NFSM_DISSECT(tl, u_int32_t *, NFSX_STATEID); stateid.seqid = fxdr_unsigned(u_int32_t, *tl++); NFSBCOPY((caddr_t)tl,(caddr_t)stateid.other,NFSX_STATEIDOTHER); } error = nfsrv_sattr(nd, vp, &nva, &attrbits, aclp, p); if (error) goto nfsmout; preat_ret = nfsvno_getattr(vp, &nva2, nd->nd_cred, p, 1); if (!nd->nd_repstat) nd->nd_repstat = preat_ret; if (nd->nd_flag & ND_NFSV3) { NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); gcheck = fxdr_unsigned(int, *tl); if (gcheck) { NFSM_DISSECT(tl, u_int32_t *, 2 * NFSX_UNSIGNED); fxdr_nfsv3time(tl, &guard); } if (!nd->nd_repstat && gcheck && (nva2.na_ctime.tv_sec != guard.tv_sec || nva2.na_ctime.tv_nsec != guard.tv_nsec)) nd->nd_repstat = NFSERR_NOT_SYNC; if (nd->nd_repstat) { vput(vp); #ifdef NFS4_ACL_EXTATTR_NAME acl_free(aclp); #endif nfsrv_wcc(nd, preat_ret, &nva2, postat_ret, &nva); goto out; } } else if (!nd->nd_repstat && (nd->nd_flag & ND_NFSV4)) nd->nd_repstat = nfsrv_checkuidgid(nd, &nva); /* * Now that we have all the fields, lets do it. * If the size is being changed write access is required, otherwise * just check for a read only file system. */ if (!nd->nd_repstat) { if (NFSVNO_NOTSETSIZE(&nva)) { if (NFSVNO_EXRDONLY(exp) || (vfs_flags(vnode_mount(vp)) & MNT_RDONLY)) nd->nd_repstat = EROFS; } else { if (vnode_vtype(vp) != VREG) nd->nd_repstat = EINVAL; else if (nva2.na_uid != nd->nd_cred->cr_uid || NFSVNO_EXSTRICTACCESS(exp)) nd->nd_repstat = nfsvno_accchk(vp, VWRITE, nd->nd_cred, exp, p, NFSACCCHK_NOOVERRIDE, NFSACCCHK_VPISLOCKED, NULL); } } if (!nd->nd_repstat && (nd->nd_flag & ND_NFSV4)) nd->nd_repstat = nfsrv_checksetattr(vp, nd, &stateid, &nva, &attrbits, exp, p); if (!nd->nd_repstat && (nd->nd_flag & ND_NFSV4)) { /* * For V4, try setting the attrbutes in sets, so that the * reply bitmap will be correct for an error case. */ if (NFSISSET_ATTRBIT(&attrbits, NFSATTRBIT_OWNER) || NFSISSET_ATTRBIT(&attrbits, NFSATTRBIT_OWNERGROUP)) { NFSVNO_ATTRINIT(&nva2); NFSVNO_SETATTRVAL(&nva2, uid, nva.na_uid); NFSVNO_SETATTRVAL(&nva2, gid, nva.na_gid); nd->nd_repstat = nfsvno_setattr(vp, &nva2, nd->nd_cred, p, exp); if (!nd->nd_repstat) { if (NFSISSET_ATTRBIT(&attrbits, NFSATTRBIT_OWNER)) NFSSETBIT_ATTRBIT(&retbits, NFSATTRBIT_OWNER); if (NFSISSET_ATTRBIT(&attrbits, NFSATTRBIT_OWNERGROUP)) NFSSETBIT_ATTRBIT(&retbits, NFSATTRBIT_OWNERGROUP); } } if (!nd->nd_repstat && NFSISSET_ATTRBIT(&attrbits, NFSATTRBIT_SIZE)) { NFSVNO_ATTRINIT(&nva2); NFSVNO_SETATTRVAL(&nva2, size, nva.na_size); nd->nd_repstat = nfsvno_setattr(vp, &nva2, nd->nd_cred, p, exp); if (!nd->nd_repstat) NFSSETBIT_ATTRBIT(&retbits, NFSATTRBIT_SIZE); } if (!nd->nd_repstat && (NFSISSET_ATTRBIT(&attrbits, NFSATTRBIT_TIMEACCESSSET) || NFSISSET_ATTRBIT(&attrbits, NFSATTRBIT_TIMEMODIFYSET))) { NFSVNO_ATTRINIT(&nva2); NFSVNO_SETATTRVAL(&nva2, atime, nva.na_atime); NFSVNO_SETATTRVAL(&nva2, mtime, nva.na_mtime); if (nva.na_vaflags & VA_UTIMES_NULL) { nva2.na_vaflags |= VA_UTIMES_NULL; NFSVNO_SETACTIVE(&nva2, vaflags); } nd->nd_repstat = nfsvno_setattr(vp, &nva2, nd->nd_cred, p, exp); if (!nd->nd_repstat) { if (NFSISSET_ATTRBIT(&attrbits, NFSATTRBIT_TIMEACCESSSET)) NFSSETBIT_ATTRBIT(&retbits, NFSATTRBIT_TIMEACCESSSET); if (NFSISSET_ATTRBIT(&attrbits, NFSATTRBIT_TIMEMODIFYSET)) NFSSETBIT_ATTRBIT(&retbits, NFSATTRBIT_TIMEMODIFYSET); } } if (!nd->nd_repstat && NFSISSET_ATTRBIT(&attrbits, NFSATTRBIT_MODE)) { NFSVNO_ATTRINIT(&nva2); NFSVNO_SETATTRVAL(&nva2, mode, nva.na_mode); nd->nd_repstat = nfsvno_setattr(vp, &nva2, nd->nd_cred, p, exp); if (!nd->nd_repstat) NFSSETBIT_ATTRBIT(&retbits, NFSATTRBIT_MODE); } #ifdef NFS4_ACL_EXTATTR_NAME if (!nd->nd_repstat && aclp->acl_cnt > 0 && NFSISSET_ATTRBIT(&attrbits, NFSATTRBIT_ACL)) { nd->nd_repstat = nfsrv_setacl(vp, aclp, nd->nd_cred, p); if (!nd->nd_repstat) NFSSETBIT_ATTRBIT(&retbits, NFSATTRBIT_ACL); } #endif } else if (!nd->nd_repstat) { nd->nd_repstat = nfsvno_setattr(vp, &nva, nd->nd_cred, p, exp); } if (nd->nd_flag & (ND_NFSV2 | ND_NFSV3)) { postat_ret = nfsvno_getattr(vp, &nva, nd->nd_cred, p, 1); if (!nd->nd_repstat) nd->nd_repstat = postat_ret; } vput(vp); #ifdef NFS4_ACL_EXTATTR_NAME acl_free(aclp); #endif if (nd->nd_flag & ND_NFSV3) nfsrv_wcc(nd, preat_ret, &nva2, postat_ret, &nva); else if (nd->nd_flag & ND_NFSV4) (void) nfsrv_putattrbit(nd, &retbits); else if (!nd->nd_repstat) nfsrv_fillattr(nd, &nva); out: NFSEXITCODE2(0, nd); return (0); nfsmout: vput(vp); #ifdef NFS4_ACL_EXTATTR_NAME acl_free(aclp); #endif if (nd->nd_flag & ND_NFSV4) { /* * For all nd_repstat, the V4 reply includes a bitmap, * even NFSERR_BADXDR, which is what this will end up * returning. */ (void) nfsrv_putattrbit(nd, &retbits); } NFSEXITCODE2(error, nd); return (error); } /* * nfs lookup rpc * (Also performs lookup parent for v4) */ APPLESTATIC int nfsrvd_lookup(struct nfsrv_descript *nd, __unused int isdgram, vnode_t dp, vnode_t *vpp, fhandle_t *fhp, NFSPROC_T *p, struct nfsexstuff *exp) { struct nameidata named; vnode_t vp, dirp = NULL; int error = 0, dattr_ret = 1; struct nfsvattr nva, dattr; char *bufp; u_long *hashp; if (nd->nd_repstat) { nfsrv_postopattr(nd, dattr_ret, &dattr); goto out; } /* * For some reason, if dp is a symlink, the error * returned is supposed to be NFSERR_SYMLINK and not NFSERR_NOTDIR. */ if (dp->v_type == VLNK && (nd->nd_flag & ND_NFSV4)) { nd->nd_repstat = NFSERR_SYMLINK; vrele(dp); goto out; } NFSNAMEICNDSET(&named.ni_cnd, nd->nd_cred, LOOKUP, LOCKLEAF | SAVESTART); nfsvno_setpathbuf(&named, &bufp, &hashp); error = nfsrv_parsename(nd, bufp, hashp, &named.ni_pathlen); if (error) { vrele(dp); nfsvno_relpathbuf(&named); goto out; } if (!nd->nd_repstat) { nd->nd_repstat = nfsvno_namei(nd, &named, dp, 0, exp, p, &dirp); } else { vrele(dp); nfsvno_relpathbuf(&named); } if (nd->nd_repstat) { if (dirp) { if (nd->nd_flag & ND_NFSV3) dattr_ret = nfsvno_getattr(dirp, &dattr, nd->nd_cred, p, 0); vrele(dirp); } if (nd->nd_flag & ND_NFSV3) nfsrv_postopattr(nd, dattr_ret, &dattr); goto out; } if (named.ni_startdir) vrele(named.ni_startdir); nfsvno_relpathbuf(&named); vp = named.ni_vp; if ((nd->nd_flag & ND_NFSV4) != 0 && !NFSVNO_EXPORTED(exp) && vp->v_type != VDIR && vp->v_type != VLNK) /* * Only allow lookup of VDIR and VLNK for traversal of * non-exported volumes during NFSv4 mounting. */ nd->nd_repstat = ENOENT; if (nd->nd_repstat == 0) nd->nd_repstat = nfsvno_getfh(vp, fhp, p); if (!(nd->nd_flag & ND_NFSV4) && !nd->nd_repstat) nd->nd_repstat = nfsvno_getattr(vp, &nva, nd->nd_cred, p, 1); if (vpp != NULL && nd->nd_repstat == 0) *vpp = vp; else vput(vp); if (dirp) { if (nd->nd_flag & ND_NFSV3) dattr_ret = nfsvno_getattr(dirp, &dattr, nd->nd_cred, p, 0); vrele(dirp); } if (nd->nd_repstat) { if (nd->nd_flag & ND_NFSV3) nfsrv_postopattr(nd, dattr_ret, &dattr); goto out; } if (nd->nd_flag & ND_NFSV2) { (void) nfsm_fhtom(nd, (u_int8_t *)fhp, 0, 0); nfsrv_fillattr(nd, &nva); } else if (nd->nd_flag & ND_NFSV3) { (void) nfsm_fhtom(nd, (u_int8_t *)fhp, 0, 0); nfsrv_postopattr(nd, 0, &nva); nfsrv_postopattr(nd, dattr_ret, &dattr); } out: NFSEXITCODE2(error, nd); return (error); } /* * nfs readlink service */ APPLESTATIC int nfsrvd_readlink(struct nfsrv_descript *nd, __unused int isdgram, vnode_t vp, NFSPROC_T *p, __unused struct nfsexstuff *exp) { u_int32_t *tl; mbuf_t mp = NULL, mpend = NULL; int getret = 1, len; struct nfsvattr nva; if (nd->nd_repstat) { nfsrv_postopattr(nd, getret, &nva); goto out; } if (vnode_vtype(vp) != VLNK) { if (nd->nd_flag & ND_NFSV2) nd->nd_repstat = ENXIO; else nd->nd_repstat = EINVAL; } if (!nd->nd_repstat) nd->nd_repstat = nfsvno_readlink(vp, nd->nd_cred, p, &mp, &mpend, &len); if (nd->nd_flag & ND_NFSV3) getret = nfsvno_getattr(vp, &nva, nd->nd_cred, p, 1); vput(vp); if (nd->nd_flag & ND_NFSV3) nfsrv_postopattr(nd, getret, &nva); if (nd->nd_repstat) goto out; NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(len); mbuf_setnext(nd->nd_mb, mp); nd->nd_mb = mpend; nd->nd_bpos = NFSMTOD(mpend, caddr_t) + mbuf_len(mpend); out: NFSEXITCODE2(0, nd); return (0); } /* * nfs read service */ APPLESTATIC int nfsrvd_read(struct nfsrv_descript *nd, __unused int isdgram, vnode_t vp, NFSPROC_T *p, struct nfsexstuff *exp) { u_int32_t *tl; int error = 0, cnt, getret = 1, reqlen, eof = 0; mbuf_t m2, m3; struct nfsvattr nva; off_t off = 0x0; struct nfsstate st, *stp = &st; struct nfslock lo, *lop = &lo; nfsv4stateid_t stateid; nfsquad_t clientid; if (nd->nd_repstat) { nfsrv_postopattr(nd, getret, &nva); goto out; } if (nd->nd_flag & ND_NFSV2) { NFSM_DISSECT(tl, u_int32_t *, 2 * NFSX_UNSIGNED); off = (off_t)fxdr_unsigned(u_int32_t, *tl++); reqlen = fxdr_unsigned(int, *tl); } else if (nd->nd_flag & ND_NFSV3) { NFSM_DISSECT(tl, u_int32_t *, 3 * NFSX_UNSIGNED); off = fxdr_hyper(tl); tl += 2; reqlen = fxdr_unsigned(int, *tl); } else { NFSM_DISSECT(tl, u_int32_t *, NFSX_STATEID + 3*NFSX_UNSIGNED); reqlen = fxdr_unsigned(int, *(tl + 6)); } if (reqlen > NFS_SRVMAXDATA(nd)) { reqlen = NFS_SRVMAXDATA(nd); } else if (reqlen < 0) { error = EBADRPC; goto nfsmout; } if (nd->nd_flag & ND_NFSV4) { stp->ls_flags = (NFSLCK_CHECK | NFSLCK_READACCESS); lop->lo_flags = NFSLCK_READ; stp->ls_ownerlen = 0; stp->ls_op = NULL; stp->ls_uid = nd->nd_cred->cr_uid; stp->ls_stateid.seqid = fxdr_unsigned(u_int32_t, *tl++); clientid.lval[0] = stp->ls_stateid.other[0] = *tl++; clientid.lval[1] = stp->ls_stateid.other[1] = *tl++; if ((nd->nd_flag & ND_IMPLIEDCLID) != 0) { if ((nd->nd_flag & ND_NFSV41) != 0) clientid.qval = nd->nd_clientid.qval; else if (nd->nd_clientid.qval != clientid.qval) printf("EEK1 multiple clids\n"); } else { if ((nd->nd_flag & ND_NFSV41) != 0) printf("EEK! no clientid from session\n"); nd->nd_flag |= ND_IMPLIEDCLID; nd->nd_clientid.qval = clientid.qval; } stp->ls_stateid.other[2] = *tl++; off = fxdr_hyper(tl); lop->lo_first = off; tl += 2; lop->lo_end = off + reqlen; /* * Paranoia, just in case it wraps around. */ if (lop->lo_end < off) lop->lo_end = NFS64BITSSET; } if (vnode_vtype(vp) != VREG) { if (nd->nd_flag & ND_NFSV3) nd->nd_repstat = EINVAL; else nd->nd_repstat = (vnode_vtype(vp) == VDIR) ? EISDIR : EINVAL; } getret = nfsvno_getattr(vp, &nva, nd->nd_cred, p, 1); if (!nd->nd_repstat) nd->nd_repstat = getret; if (!nd->nd_repstat && (nva.na_uid != nd->nd_cred->cr_uid || NFSVNO_EXSTRICTACCESS(exp))) { nd->nd_repstat = nfsvno_accchk(vp, VREAD, nd->nd_cred, exp, p, NFSACCCHK_ALLOWOWNER, NFSACCCHK_VPISLOCKED, NULL); if (nd->nd_repstat) nd->nd_repstat = nfsvno_accchk(vp, VEXEC, nd->nd_cred, exp, p, NFSACCCHK_ALLOWOWNER, NFSACCCHK_VPISLOCKED, NULL); } if ((nd->nd_flag & ND_NFSV4) && !nd->nd_repstat) nd->nd_repstat = nfsrv_lockctrl(vp, &stp, &lop, NULL, clientid, &stateid, exp, nd, p); if (nd->nd_repstat) { vput(vp); if (nd->nd_flag & ND_NFSV3) nfsrv_postopattr(nd, getret, &nva); goto out; } if (off >= nva.na_size) { cnt = 0; eof = 1; } else if (reqlen == 0) cnt = 0; else if ((off + reqlen) >= nva.na_size) { cnt = nva.na_size - off; eof = 1; } else cnt = reqlen; m3 = NULL; if (cnt > 0) { nd->nd_repstat = nfsvno_read(vp, off, cnt, nd->nd_cred, p, &m3, &m2); if (!(nd->nd_flag & ND_NFSV4)) { getret = nfsvno_getattr(vp, &nva, nd->nd_cred, p, 1); if (!nd->nd_repstat) nd->nd_repstat = getret; } if (nd->nd_repstat) { vput(vp); if (m3) mbuf_freem(m3); if (nd->nd_flag & ND_NFSV3) nfsrv_postopattr(nd, getret, &nva); goto out; } } vput(vp); if (nd->nd_flag & ND_NFSV2) { nfsrv_fillattr(nd, &nva); NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); } else { if (nd->nd_flag & ND_NFSV3) { nfsrv_postopattr(nd, getret, &nva); NFSM_BUILD(tl, u_int32_t *, 3 * NFSX_UNSIGNED); *tl++ = txdr_unsigned(cnt); } else NFSM_BUILD(tl, u_int32_t *, 2 * NFSX_UNSIGNED); if (eof) *tl++ = newnfs_true; else *tl++ = newnfs_false; } *tl = txdr_unsigned(cnt); if (m3) { mbuf_setnext(nd->nd_mb, m3); nd->nd_mb = m2; nd->nd_bpos = NFSMTOD(m2, caddr_t) + mbuf_len(m2); } out: NFSEXITCODE2(0, nd); return (0); nfsmout: vput(vp); NFSEXITCODE2(error, nd); return (error); } /* * nfs write service */ APPLESTATIC int nfsrvd_write(struct nfsrv_descript *nd, __unused int isdgram, vnode_t vp, NFSPROC_T *p, struct nfsexstuff *exp) { int i, cnt; u_int32_t *tl; mbuf_t mp; struct nfsvattr nva, forat; int aftat_ret = 1, retlen, len, error = 0, forat_ret = 1; int stable = NFSWRITE_FILESYNC; off_t off; struct nfsstate st, *stp = &st; struct nfslock lo, *lop = &lo; nfsv4stateid_t stateid; nfsquad_t clientid; if (nd->nd_repstat) { nfsrv_wcc(nd, forat_ret, &forat, aftat_ret, &nva); goto out; } if (nd->nd_flag & ND_NFSV2) { NFSM_DISSECT(tl, u_int32_t *, 4 * NFSX_UNSIGNED); off = (off_t)fxdr_unsigned(u_int32_t, *++tl); tl += 2; retlen = len = fxdr_unsigned(int32_t, *tl); } else if (nd->nd_flag & ND_NFSV3) { NFSM_DISSECT(tl, u_int32_t *, 5 * NFSX_UNSIGNED); off = fxdr_hyper(tl); tl += 3; stable = fxdr_unsigned(int, *tl++); retlen = len = fxdr_unsigned(int32_t, *tl); } else { NFSM_DISSECT(tl, u_int32_t *, NFSX_STATEID + 4 * NFSX_UNSIGNED); stp->ls_flags = (NFSLCK_CHECK | NFSLCK_WRITEACCESS); lop->lo_flags = NFSLCK_WRITE; stp->ls_ownerlen = 0; stp->ls_op = NULL; stp->ls_uid = nd->nd_cred->cr_uid; stp->ls_stateid.seqid = fxdr_unsigned(u_int32_t, *tl++); clientid.lval[0] = stp->ls_stateid.other[0] = *tl++; clientid.lval[1] = stp->ls_stateid.other[1] = *tl++; if ((nd->nd_flag & ND_IMPLIEDCLID) != 0) { if ((nd->nd_flag & ND_NFSV41) != 0) clientid.qval = nd->nd_clientid.qval; else if (nd->nd_clientid.qval != clientid.qval) printf("EEK2 multiple clids\n"); } else { if ((nd->nd_flag & ND_NFSV41) != 0) printf("EEK! no clientid from session\n"); nd->nd_flag |= ND_IMPLIEDCLID; nd->nd_clientid.qval = clientid.qval; } stp->ls_stateid.other[2] = *tl++; off = fxdr_hyper(tl); lop->lo_first = off; tl += 2; stable = fxdr_unsigned(int, *tl++); retlen = len = fxdr_unsigned(int32_t, *tl); lop->lo_end = off + len; /* * Paranoia, just in case it wraps around, which shouldn't * ever happen anyhow. */ if (lop->lo_end < lop->lo_first) lop->lo_end = NFS64BITSSET; } /* * Loop through the mbuf chain, counting how many mbufs are a * part of this write operation, so the iovec size is known. */ cnt = 0; mp = nd->nd_md; i = NFSMTOD(mp, caddr_t) + mbuf_len(mp) - nd->nd_dpos; while (len > 0) { if (i > 0) { len -= i; cnt++; } mp = mbuf_next(mp); if (!mp) { if (len > 0) { error = EBADRPC; goto nfsmout; } } else i = mbuf_len(mp); } if (retlen > NFS_SRVMAXIO || retlen < 0) nd->nd_repstat = EIO; if (vnode_vtype(vp) != VREG && !nd->nd_repstat) { if (nd->nd_flag & ND_NFSV3) nd->nd_repstat = EINVAL; else nd->nd_repstat = (vnode_vtype(vp) == VDIR) ? EISDIR : EINVAL; } forat_ret = nfsvno_getattr(vp, &forat, nd->nd_cred, p, 1); if (!nd->nd_repstat) nd->nd_repstat = forat_ret; if (!nd->nd_repstat && (forat.na_uid != nd->nd_cred->cr_uid || NFSVNO_EXSTRICTACCESS(exp))) nd->nd_repstat = nfsvno_accchk(vp, VWRITE, nd->nd_cred, exp, p, NFSACCCHK_ALLOWOWNER, NFSACCCHK_VPISLOCKED, NULL); if ((nd->nd_flag & ND_NFSV4) && !nd->nd_repstat) { nd->nd_repstat = nfsrv_lockctrl(vp, &stp, &lop, NULL, clientid, &stateid, exp, nd, p); } if (nd->nd_repstat) { vput(vp); if (nd->nd_flag & ND_NFSV3) nfsrv_wcc(nd, forat_ret, &forat, aftat_ret, &nva); goto out; } /* * For NFS Version 2, it is not obvious what a write of zero length * should do, but I might as well be consistent with Version 3, * which is to return ok so long as there are no permission problems. */ if (retlen > 0) { nd->nd_repstat = nfsvno_write(vp, off, retlen, cnt, stable, nd->nd_md, nd->nd_dpos, nd->nd_cred, p); error = nfsm_advance(nd, NFSM_RNDUP(retlen), -1); if (error) panic("nfsrv_write mbuf"); } if (nd->nd_flag & ND_NFSV4) aftat_ret = 0; else aftat_ret = nfsvno_getattr(vp, &nva, nd->nd_cred, p, 1); vput(vp); if (!nd->nd_repstat) nd->nd_repstat = aftat_ret; if (nd->nd_flag & (ND_NFSV3 | ND_NFSV4)) { if (nd->nd_flag & ND_NFSV3) nfsrv_wcc(nd, forat_ret, &forat, aftat_ret, &nva); if (nd->nd_repstat) goto out; NFSM_BUILD(tl, u_int32_t *, 4 * NFSX_UNSIGNED); *tl++ = txdr_unsigned(retlen); /* * If nfs_async is set, then pretend the write was FILESYNC. * Warning: Doing this violates RFC1813 and runs a risk * of data written by a client being lost when the server * crashes/reboots. */ if (stable == NFSWRITE_UNSTABLE && nfs_async == 0) *tl++ = txdr_unsigned(stable); else *tl++ = txdr_unsigned(NFSWRITE_FILESYNC); /* * Actually, there is no need to txdr these fields, * but it may make the values more human readable, * for debugging purposes. */ *tl++ = txdr_unsigned(nfsboottime.tv_sec); *tl = txdr_unsigned(nfsboottime.tv_usec); } else if (!nd->nd_repstat) nfsrv_fillattr(nd, &nva); out: NFSEXITCODE2(0, nd); return (0); nfsmout: vput(vp); NFSEXITCODE2(error, nd); return (error); } /* * nfs create service (creates regular files for V2 and V3. Spec. files for V2.) * now does a truncate to 0 length via. setattr if it already exists * The core creation routine has been extracted out into nfsrv_creatsub(), * so it can also be used by nfsrv_open() for V4. */ APPLESTATIC int nfsrvd_create(struct nfsrv_descript *nd, __unused int isdgram, vnode_t dp, NFSPROC_T *p, struct nfsexstuff *exp) { struct nfsvattr nva, dirfor, diraft; struct nfsv2_sattr *sp; struct nameidata named; u_int32_t *tl; int error = 0, tsize, dirfor_ret = 1, diraft_ret = 1; int how = NFSCREATE_UNCHECKED, exclusive_flag = 0; NFSDEV_T rdev = 0; vnode_t vp = NULL, dirp = NULL; fhandle_t fh; char *bufp; u_long *hashp; enum vtype vtyp; int32_t cverf[2], tverf[2] = { 0, 0 }; if (nd->nd_repstat) { nfsrv_wcc(nd, dirfor_ret, &dirfor, diraft_ret, &diraft); goto out; } NFSNAMEICNDSET(&named.ni_cnd, nd->nd_cred, CREATE, LOCKPARENT | LOCKLEAF | SAVESTART | NOCACHE); nfsvno_setpathbuf(&named, &bufp, &hashp); error = nfsrv_parsename(nd, bufp, hashp, &named.ni_pathlen); if (error) goto nfsmout; if (!nd->nd_repstat) { NFSVNO_ATTRINIT(&nva); if (nd->nd_flag & ND_NFSV2) { NFSM_DISSECT(sp, struct nfsv2_sattr *, NFSX_V2SATTR); vtyp = IFTOVT(fxdr_unsigned(u_int32_t, sp->sa_mode)); if (vtyp == VNON) vtyp = VREG; NFSVNO_SETATTRVAL(&nva, type, vtyp); NFSVNO_SETATTRVAL(&nva, mode, nfstov_mode(sp->sa_mode)); switch (nva.na_type) { case VREG: tsize = fxdr_unsigned(int32_t, sp->sa_size); if (tsize != -1) NFSVNO_SETATTRVAL(&nva, size, (u_quad_t)tsize); break; case VCHR: case VBLK: case VFIFO: rdev = fxdr_unsigned(NFSDEV_T, sp->sa_size); break; default: break; } } else { NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); how = fxdr_unsigned(int, *tl); switch (how) { case NFSCREATE_GUARDED: case NFSCREATE_UNCHECKED: error = nfsrv_sattr(nd, NULL, &nva, NULL, NULL, p); if (error) goto nfsmout; break; case NFSCREATE_EXCLUSIVE: NFSM_DISSECT(tl, u_int32_t *, NFSX_VERF); cverf[0] = *tl++; cverf[1] = *tl; exclusive_flag = 1; break; } NFSVNO_SETATTRVAL(&nva, type, VREG); } } if (nd->nd_repstat) { nfsvno_relpathbuf(&named); if (nd->nd_flag & ND_NFSV3) { dirfor_ret = nfsvno_getattr(dp, &dirfor, nd->nd_cred, p, 1); nfsrv_wcc(nd, dirfor_ret, &dirfor, diraft_ret, &diraft); } vput(dp); goto out; } nd->nd_repstat = nfsvno_namei(nd, &named, dp, 1, exp, p, &dirp); if (dirp) { if (nd->nd_flag & ND_NFSV2) { vrele(dirp); dirp = NULL; } else { dirfor_ret = nfsvno_getattr(dirp, &dirfor, nd->nd_cred, p, 0); } } if (nd->nd_repstat) { if (nd->nd_flag & ND_NFSV3) nfsrv_wcc(nd, dirfor_ret, &dirfor, diraft_ret, &diraft); if (dirp) vrele(dirp); goto out; } if (!(nd->nd_flag & ND_NFSV2)) { switch (how) { case NFSCREATE_GUARDED: if (named.ni_vp) nd->nd_repstat = EEXIST; break; case NFSCREATE_UNCHECKED: break; case NFSCREATE_EXCLUSIVE: if (named.ni_vp == NULL) NFSVNO_SETATTRVAL(&nva, mode, 0); break; } } /* * Iff doesn't exist, create it * otherwise just truncate to 0 length * should I set the mode too ? */ nd->nd_repstat = nfsvno_createsub(nd, &named, &vp, &nva, &exclusive_flag, cverf, rdev, p, exp); if (!nd->nd_repstat) { nd->nd_repstat = nfsvno_getfh(vp, &fh, p); if (!nd->nd_repstat) nd->nd_repstat = nfsvno_getattr(vp, &nva, nd->nd_cred, p, 1); vput(vp); if (!nd->nd_repstat) { tverf[0] = nva.na_atime.tv_sec; tverf[1] = nva.na_atime.tv_nsec; } } if (nd->nd_flag & ND_NFSV2) { if (!nd->nd_repstat) { (void) nfsm_fhtom(nd, (u_int8_t *)&fh, 0, 0); nfsrv_fillattr(nd, &nva); } } else { if (exclusive_flag && !nd->nd_repstat && (cverf[0] != tverf[0] || cverf[1] != tverf[1])) nd->nd_repstat = EEXIST; diraft_ret = nfsvno_getattr(dirp, &diraft, nd->nd_cred, p, 0); vrele(dirp); if (!nd->nd_repstat) { (void) nfsm_fhtom(nd, (u_int8_t *)&fh, 0, 1); nfsrv_postopattr(nd, 0, &nva); } nfsrv_wcc(nd, dirfor_ret, &dirfor, diraft_ret, &diraft); } out: NFSEXITCODE2(0, nd); return (0); nfsmout: vput(dp); nfsvno_relpathbuf(&named); NFSEXITCODE2(error, nd); return (error); } /* * nfs v3 mknod service (and v4 create) */ APPLESTATIC int nfsrvd_mknod(struct nfsrv_descript *nd, __unused int isdgram, vnode_t dp, vnode_t *vpp, fhandle_t *fhp, NFSPROC_T *p, struct nfsexstuff *exp) { struct nfsvattr nva, dirfor, diraft; u_int32_t *tl; struct nameidata named; int error = 0, dirfor_ret = 1, diraft_ret = 1, pathlen; u_int32_t major, minor; enum vtype vtyp = VNON; nfstype nfs4type = NFNON; vnode_t vp, dirp = NULL; nfsattrbit_t attrbits; char *bufp = NULL, *pathcp = NULL; u_long *hashp, cnflags; NFSACL_T *aclp = NULL; NFSVNO_ATTRINIT(&nva); cnflags = (LOCKPARENT | SAVESTART); if (nd->nd_repstat) { nfsrv_wcc(nd, dirfor_ret, &dirfor, diraft_ret, &diraft); goto out; } #ifdef NFS4_ACL_EXTATTR_NAME aclp = acl_alloc(M_WAITOK); aclp->acl_cnt = 0; #endif /* * For V4, the creation stuff is here, Yuck! */ if (nd->nd_flag & ND_NFSV4) { NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); vtyp = nfsv34tov_type(*tl); nfs4type = fxdr_unsigned(nfstype, *tl); switch (nfs4type) { case NFLNK: error = nfsvno_getsymlink(nd, &nva, p, &pathcp, &pathlen); if (error) goto nfsmout; break; case NFCHR: case NFBLK: NFSM_DISSECT(tl, u_int32_t *, 2 * NFSX_UNSIGNED); major = fxdr_unsigned(u_int32_t, *tl++); minor = fxdr_unsigned(u_int32_t, *tl); nva.na_rdev = NFSMAKEDEV(major, minor); break; case NFSOCK: case NFFIFO: break; case NFDIR: cnflags = (LOCKPARENT | SAVENAME); break; default: nd->nd_repstat = NFSERR_BADTYPE; vrele(dp); #ifdef NFS4_ACL_EXTATTR_NAME acl_free(aclp); #endif goto out; } } NFSNAMEICNDSET(&named.ni_cnd, nd->nd_cred, CREATE, cnflags | NOCACHE); nfsvno_setpathbuf(&named, &bufp, &hashp); error = nfsrv_parsename(nd, bufp, hashp, &named.ni_pathlen); if (error) goto nfsmout; if (!nd->nd_repstat) { if (nd->nd_flag & ND_NFSV3) { NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); vtyp = nfsv34tov_type(*tl); } error = nfsrv_sattr(nd, NULL, &nva, &attrbits, aclp, p); if (error) goto nfsmout; nva.na_type = vtyp; if (!nd->nd_repstat && (nd->nd_flag & ND_NFSV3) && (vtyp == VCHR || vtyp == VBLK)) { NFSM_DISSECT(tl, u_int32_t *, 2 * NFSX_UNSIGNED); major = fxdr_unsigned(u_int32_t, *tl++); minor = fxdr_unsigned(u_int32_t, *tl); nva.na_rdev = NFSMAKEDEV(major, minor); } } dirfor_ret = nfsvno_getattr(dp, &dirfor, nd->nd_cred, p, 0); if (!nd->nd_repstat && (nd->nd_flag & ND_NFSV4)) { if (!dirfor_ret && NFSVNO_ISSETGID(&nva) && dirfor.na_gid == nva.na_gid) NFSVNO_UNSET(&nva, gid); nd->nd_repstat = nfsrv_checkuidgid(nd, &nva); } if (nd->nd_repstat) { vrele(dp); #ifdef NFS4_ACL_EXTATTR_NAME acl_free(aclp); #endif nfsvno_relpathbuf(&named); if (pathcp) FREE(pathcp, M_TEMP); if (nd->nd_flag & ND_NFSV3) nfsrv_wcc(nd, dirfor_ret, &dirfor, diraft_ret, &diraft); goto out; } /* * Yuck! For V4, mkdir and link are here and some V4 clients don't fill * in va_mode, so we'll have to set a default here. */ if (NFSVNO_NOTSETMODE(&nva)) { if (vtyp == VLNK) nva.na_mode = 0755; else nva.na_mode = 0400; } if (vtyp == VDIR) named.ni_cnd.cn_flags |= WILLBEDIR; nd->nd_repstat = nfsvno_namei(nd, &named, dp, 0, exp, p, &dirp); if (nd->nd_repstat) { if (dirp) { if (nd->nd_flag & ND_NFSV3) dirfor_ret = nfsvno_getattr(dirp, &dirfor, nd->nd_cred, p, 0); vrele(dirp); } #ifdef NFS4_ACL_EXTATTR_NAME acl_free(aclp); #endif if (nd->nd_flag & ND_NFSV3) nfsrv_wcc(nd, dirfor_ret, &dirfor, diraft_ret, &diraft); goto out; } if (dirp) dirfor_ret = nfsvno_getattr(dirp, &dirfor, nd->nd_cred, p, 0); if ((nd->nd_flag & ND_NFSV4) && (vtyp == VDIR || vtyp == VLNK)) { if (vtyp == VDIR) { nfsrvd_mkdirsub(nd, &named, &nva, fhp, vpp, dirp, &dirfor, &diraft, &diraft_ret, &attrbits, aclp, p, exp); #ifdef NFS4_ACL_EXTATTR_NAME acl_free(aclp); #endif goto out; } else if (vtyp == VLNK) { nfsrvd_symlinksub(nd, &named, &nva, fhp, vpp, dirp, &dirfor, &diraft, &diraft_ret, &attrbits, aclp, p, exp, pathcp, pathlen); #ifdef NFS4_ACL_EXTATTR_NAME acl_free(aclp); #endif FREE(pathcp, M_TEMP); goto out; } } nd->nd_repstat = nfsvno_mknod(&named, &nva, nd->nd_cred, p); if (!nd->nd_repstat) { vp = named.ni_vp; nfsrv_fixattr(nd, vp, &nva, aclp, p, &attrbits, exp); nd->nd_repstat = nfsvno_getfh(vp, fhp, p); if ((nd->nd_flag & ND_NFSV3) && !nd->nd_repstat) nd->nd_repstat = nfsvno_getattr(vp, &nva, nd->nd_cred, p, 1); if (vpp != NULL && nd->nd_repstat == 0) { NFSVOPUNLOCK(vp, 0); *vpp = vp; } else vput(vp); } diraft_ret = nfsvno_getattr(dirp, &diraft, nd->nd_cred, p, 0); vrele(dirp); if (!nd->nd_repstat) { if (nd->nd_flag & ND_NFSV3) { (void) nfsm_fhtom(nd, (u_int8_t *)fhp, 0, 1); nfsrv_postopattr(nd, 0, &nva); } else { NFSM_BUILD(tl, u_int32_t *, 5 * NFSX_UNSIGNED); *tl++ = newnfs_false; txdr_hyper(dirfor.na_filerev, tl); tl += 2; txdr_hyper(diraft.na_filerev, tl); (void) nfsrv_putattrbit(nd, &attrbits); } } if (nd->nd_flag & ND_NFSV3) nfsrv_wcc(nd, dirfor_ret, &dirfor, diraft_ret, &diraft); #ifdef NFS4_ACL_EXTATTR_NAME acl_free(aclp); #endif out: NFSEXITCODE2(0, nd); return (0); nfsmout: vrele(dp); #ifdef NFS4_ACL_EXTATTR_NAME acl_free(aclp); #endif if (bufp) nfsvno_relpathbuf(&named); if (pathcp) FREE(pathcp, M_TEMP); NFSEXITCODE2(error, nd); return (error); } /* * nfs remove service */ APPLESTATIC int nfsrvd_remove(struct nfsrv_descript *nd, __unused int isdgram, vnode_t dp, NFSPROC_T *p, struct nfsexstuff *exp) { struct nameidata named; u_int32_t *tl; int error = 0, dirfor_ret = 1, diraft_ret = 1; vnode_t dirp = NULL; struct nfsvattr dirfor, diraft; char *bufp; u_long *hashp; if (nd->nd_repstat) { nfsrv_wcc(nd, dirfor_ret, &dirfor, diraft_ret, &diraft); goto out; } NFSNAMEICNDSET(&named.ni_cnd, nd->nd_cred, DELETE, LOCKPARENT | LOCKLEAF); nfsvno_setpathbuf(&named, &bufp, &hashp); error = nfsrv_parsename(nd, bufp, hashp, &named.ni_pathlen); if (error) { vput(dp); nfsvno_relpathbuf(&named); goto out; } if (!nd->nd_repstat) { nd->nd_repstat = nfsvno_namei(nd, &named, dp, 1, exp, p, &dirp); } else { vput(dp); nfsvno_relpathbuf(&named); } if (dirp) { if (!(nd->nd_flag & ND_NFSV2)) { dirfor_ret = nfsvno_getattr(dirp, &dirfor, nd->nd_cred, p, 0); } else { vrele(dirp); dirp = NULL; } } if (!nd->nd_repstat) { if (nd->nd_flag & ND_NFSV4) { if (vnode_vtype(named.ni_vp) == VDIR) nd->nd_repstat = nfsvno_rmdirsub(&named, 1, nd->nd_cred, p, exp); else nd->nd_repstat = nfsvno_removesub(&named, 1, nd->nd_cred, p, exp); } else if (nd->nd_procnum == NFSPROC_RMDIR) { nd->nd_repstat = nfsvno_rmdirsub(&named, 0, nd->nd_cred, p, exp); } else { nd->nd_repstat = nfsvno_removesub(&named, 0, nd->nd_cred, p, exp); } } if (!(nd->nd_flag & ND_NFSV2)) { if (dirp) { diraft_ret = nfsvno_getattr(dirp, &diraft, nd->nd_cred, p, 0); vrele(dirp); } if (nd->nd_flag & ND_NFSV3) { nfsrv_wcc(nd, dirfor_ret, &dirfor, diraft_ret, &diraft); } else if (!nd->nd_repstat) { NFSM_BUILD(tl, u_int32_t *, 5 * NFSX_UNSIGNED); *tl++ = newnfs_false; txdr_hyper(dirfor.na_filerev, tl); tl += 2; txdr_hyper(diraft.na_filerev, tl); } } out: NFSEXITCODE2(error, nd); return (error); } /* * nfs rename service */ APPLESTATIC int nfsrvd_rename(struct nfsrv_descript *nd, int isdgram, vnode_t dp, vnode_t todp, NFSPROC_T *p, struct nfsexstuff *exp, struct nfsexstuff *toexp) { u_int32_t *tl; int error = 0, fdirfor_ret = 1, fdiraft_ret = 1; int tdirfor_ret = 1, tdiraft_ret = 1; struct nameidata fromnd, tond; vnode_t fdirp = NULL, tdirp = NULL, tdp = NULL; struct nfsvattr fdirfor, fdiraft, tdirfor, tdiraft; struct nfsexstuff tnes; struct nfsrvfh tfh; char *bufp, *tbufp = NULL; u_long *hashp; fhandle_t fh; if (nd->nd_repstat) { nfsrv_wcc(nd, fdirfor_ret, &fdirfor, fdiraft_ret, &fdiraft); nfsrv_wcc(nd, tdirfor_ret, &tdirfor, tdiraft_ret, &tdiraft); goto out; } if (!(nd->nd_flag & ND_NFSV2)) fdirfor_ret = nfsvno_getattr(dp, &fdirfor, nd->nd_cred, p, 1); tond.ni_cnd.cn_nameiop = 0; tond.ni_startdir = NULL; NFSNAMEICNDSET(&fromnd.ni_cnd, nd->nd_cred, DELETE, WANTPARENT | SAVESTART); nfsvno_setpathbuf(&fromnd, &bufp, &hashp); error = nfsrv_parsename(nd, bufp, hashp, &fromnd.ni_pathlen); if (error) { vput(dp); if (todp) vrele(todp); nfsvno_relpathbuf(&fromnd); goto out; } /* * Unlock dp in this code section, so it is unlocked before * tdp gets locked. This avoids a potential LOR if tdp is the * parent directory of dp. */ if (nd->nd_flag & ND_NFSV4) { tdp = todp; tnes = *toexp; if (dp != tdp) { NFSVOPUNLOCK(dp, 0); tdirfor_ret = nfsvno_getattr(tdp, &tdirfor, nd->nd_cred, p, 0); /* Might lock tdp. */ } else { tdirfor_ret = nfsvno_getattr(tdp, &tdirfor, nd->nd_cred, p, 1); NFSVOPUNLOCK(dp, 0); } } else { tfh.nfsrvfh_len = 0; error = nfsrv_mtofh(nd, &tfh); if (error == 0) error = nfsvno_getfh(dp, &fh, p); if (error) { vput(dp); /* todp is always NULL except NFSv4 */ nfsvno_relpathbuf(&fromnd); goto out; } /* If this is the same file handle, just VREF() the vnode. */ if (tfh.nfsrvfh_len == NFSX_MYFH && !NFSBCMP(tfh.nfsrvfh_data, &fh, NFSX_MYFH)) { VREF(dp); tdp = dp; tnes = *exp; tdirfor_ret = nfsvno_getattr(tdp, &tdirfor, nd->nd_cred, p, 1); NFSVOPUNLOCK(dp, 0); } else { NFSVOPUNLOCK(dp, 0); nd->nd_cred->cr_uid = nd->nd_saveduid; nfsd_fhtovp(nd, &tfh, LK_EXCLUSIVE, &tdp, &tnes, NULL, 0, p); /* Locks tdp. */ if (tdp) { tdirfor_ret = nfsvno_getattr(tdp, &tdirfor, nd->nd_cred, p, 1); NFSVOPUNLOCK(tdp, 0); } } } NFSNAMEICNDSET(&tond.ni_cnd, nd->nd_cred, RENAME, LOCKPARENT | LOCKLEAF | NOCACHE | SAVESTART); nfsvno_setpathbuf(&tond, &tbufp, &hashp); if (!nd->nd_repstat) { error = nfsrv_parsename(nd, tbufp, hashp, &tond.ni_pathlen); if (error) { if (tdp) vrele(tdp); vrele(dp); nfsvno_relpathbuf(&fromnd); nfsvno_relpathbuf(&tond); goto out; } } if (nd->nd_repstat) { if (nd->nd_flag & ND_NFSV3) { nfsrv_wcc(nd, fdirfor_ret, &fdirfor, fdiraft_ret, &fdiraft); nfsrv_wcc(nd, tdirfor_ret, &tdirfor, tdiraft_ret, &tdiraft); } if (tdp) vrele(tdp); vrele(dp); nfsvno_relpathbuf(&fromnd); nfsvno_relpathbuf(&tond); goto out; } /* * Done parsing, now down to business. */ nd->nd_repstat = nfsvno_namei(nd, &fromnd, dp, 0, exp, p, &fdirp); if (nd->nd_repstat) { if (nd->nd_flag & ND_NFSV3) { nfsrv_wcc(nd, fdirfor_ret, &fdirfor, fdiraft_ret, &fdiraft); nfsrv_wcc(nd, tdirfor_ret, &tdirfor, tdiraft_ret, &tdiraft); } if (fdirp) vrele(fdirp); if (tdp) vrele(tdp); nfsvno_relpathbuf(&tond); goto out; } if (vnode_vtype(fromnd.ni_vp) == VDIR) tond.ni_cnd.cn_flags |= WILLBEDIR; nd->nd_repstat = nfsvno_namei(nd, &tond, tdp, 0, &tnes, p, &tdirp); nd->nd_repstat = nfsvno_rename(&fromnd, &tond, nd->nd_repstat, nd->nd_flag, nd->nd_cred, p); if (fdirp) fdiraft_ret = nfsvno_getattr(fdirp, &fdiraft, nd->nd_cred, p, 0); if (tdirp) tdiraft_ret = nfsvno_getattr(tdirp, &tdiraft, nd->nd_cred, p, 0); if (fdirp) vrele(fdirp); if (tdirp) vrele(tdirp); if (nd->nd_flag & ND_NFSV3) { nfsrv_wcc(nd, fdirfor_ret, &fdirfor, fdiraft_ret, &fdiraft); nfsrv_wcc(nd, tdirfor_ret, &tdirfor, tdiraft_ret, &tdiraft); } else if ((nd->nd_flag & ND_NFSV4) && !nd->nd_repstat) { NFSM_BUILD(tl, u_int32_t *, 10 * NFSX_UNSIGNED); *tl++ = newnfs_false; txdr_hyper(fdirfor.na_filerev, tl); tl += 2; txdr_hyper(fdiraft.na_filerev, tl); tl += 2; *tl++ = newnfs_false; txdr_hyper(tdirfor.na_filerev, tl); tl += 2; txdr_hyper(tdiraft.na_filerev, tl); } out: NFSEXITCODE2(error, nd); return (error); } /* * nfs link service */ APPLESTATIC int nfsrvd_link(struct nfsrv_descript *nd, int isdgram, vnode_t vp, vnode_t tovp, NFSPROC_T *p, struct nfsexstuff *exp, struct nfsexstuff *toexp) { struct nameidata named; u_int32_t *tl; int error = 0, dirfor_ret = 1, diraft_ret = 1, getret = 1; vnode_t dirp = NULL, dp = NULL; struct nfsvattr dirfor, diraft, at; struct nfsexstuff tnes; struct nfsrvfh dfh; char *bufp; u_long *hashp; if (nd->nd_repstat) { nfsrv_postopattr(nd, getret, &at); nfsrv_wcc(nd, dirfor_ret, &dirfor, diraft_ret, &diraft); goto out; } NFSVOPUNLOCK(vp, 0); if (vnode_vtype(vp) == VDIR) { if (nd->nd_flag & ND_NFSV4) nd->nd_repstat = NFSERR_ISDIR; else nd->nd_repstat = NFSERR_INVAL; if (tovp) vrele(tovp); } if (!nd->nd_repstat) { if (nd->nd_flag & ND_NFSV4) { dp = tovp; tnes = *toexp; } else { error = nfsrv_mtofh(nd, &dfh); if (error) { vrele(vp); /* tovp is always NULL unless NFSv4 */ goto out; } nfsd_fhtovp(nd, &dfh, LK_EXCLUSIVE, &dp, &tnes, NULL, 0, p); if (dp) NFSVOPUNLOCK(dp, 0); } } NFSNAMEICNDSET(&named.ni_cnd, nd->nd_cred, CREATE, LOCKPARENT | SAVENAME | NOCACHE); if (!nd->nd_repstat) { nfsvno_setpathbuf(&named, &bufp, &hashp); error = nfsrv_parsename(nd, bufp, hashp, &named.ni_pathlen); if (error) { vrele(vp); if (dp) vrele(dp); nfsvno_relpathbuf(&named); goto out; } if (!nd->nd_repstat) { nd->nd_repstat = nfsvno_namei(nd, &named, dp, 0, &tnes, p, &dirp); } else { if (dp) vrele(dp); nfsvno_relpathbuf(&named); } } if (dirp) { if (nd->nd_flag & ND_NFSV2) { vrele(dirp); dirp = NULL; } else { dirfor_ret = nfsvno_getattr(dirp, &dirfor, nd->nd_cred, p, 0); } } if (!nd->nd_repstat) nd->nd_repstat = nfsvno_link(&named, vp, nd->nd_cred, p, exp); if (nd->nd_flag & ND_NFSV3) getret = nfsvno_getattr(vp, &at, nd->nd_cred, p, 0); if (dirp) { diraft_ret = nfsvno_getattr(dirp, &diraft, nd->nd_cred, p, 0); vrele(dirp); } vrele(vp); if (nd->nd_flag & ND_NFSV3) { nfsrv_postopattr(nd, getret, &at); nfsrv_wcc(nd, dirfor_ret, &dirfor, diraft_ret, &diraft); } else if ((nd->nd_flag & ND_NFSV4) && !nd->nd_repstat) { NFSM_BUILD(tl, u_int32_t *, 5 * NFSX_UNSIGNED); *tl++ = newnfs_false; txdr_hyper(dirfor.na_filerev, tl); tl += 2; txdr_hyper(diraft.na_filerev, tl); } out: NFSEXITCODE2(error, nd); return (error); } /* * nfs symbolic link service */ APPLESTATIC int nfsrvd_symlink(struct nfsrv_descript *nd, __unused int isdgram, vnode_t dp, vnode_t *vpp, fhandle_t *fhp, NFSPROC_T *p, struct nfsexstuff *exp) { struct nfsvattr nva, dirfor, diraft; struct nameidata named; int error = 0, dirfor_ret = 1, diraft_ret = 1, pathlen; vnode_t dirp = NULL; char *bufp, *pathcp = NULL; u_long *hashp; if (nd->nd_repstat) { nfsrv_wcc(nd, dirfor_ret, &dirfor, diraft_ret, &diraft); goto out; } if (vpp) *vpp = NULL; NFSVNO_ATTRINIT(&nva); NFSNAMEICNDSET(&named.ni_cnd, nd->nd_cred, CREATE, LOCKPARENT | SAVESTART | NOCACHE); nfsvno_setpathbuf(&named, &bufp, &hashp); error = nfsrv_parsename(nd, bufp, hashp, &named.ni_pathlen); if (!error && !nd->nd_repstat) error = nfsvno_getsymlink(nd, &nva, p, &pathcp, &pathlen); if (error) { vrele(dp); nfsvno_relpathbuf(&named); goto out; } if (!nd->nd_repstat) { nd->nd_repstat = nfsvno_namei(nd, &named, dp, 0, exp, p, &dirp); } else { vrele(dp); nfsvno_relpathbuf(&named); } if (dirp != NULL && !(nd->nd_flag & ND_NFSV3)) { vrele(dirp); dirp = NULL; } /* * And call nfsrvd_symlinksub() to do the common code. It will * return EBADRPC upon a parsing error, 0 otherwise. */ if (!nd->nd_repstat) { if (dirp != NULL) dirfor_ret = nfsvno_getattr(dirp, &dirfor, nd->nd_cred, p, 0); nfsrvd_symlinksub(nd, &named, &nva, fhp, vpp, dirp, &dirfor, &diraft, &diraft_ret, NULL, NULL, p, exp, pathcp, pathlen); } else if (dirp != NULL) { dirfor_ret = nfsvno_getattr(dirp, &dirfor, nd->nd_cred, p, 0); vrele(dirp); } if (pathcp) FREE(pathcp, M_TEMP); if (nd->nd_flag & ND_NFSV3) { if (!nd->nd_repstat) { (void) nfsm_fhtom(nd, (u_int8_t *)fhp, 0, 1); nfsrv_postopattr(nd, 0, &nva); } nfsrv_wcc(nd, dirfor_ret, &dirfor, diraft_ret, &diraft); } out: NFSEXITCODE2(error, nd); return (error); } /* * Common code for creating a symbolic link. */ static void nfsrvd_symlinksub(struct nfsrv_descript *nd, struct nameidata *ndp, struct nfsvattr *nvap, fhandle_t *fhp, vnode_t *vpp, vnode_t dirp, struct nfsvattr *dirforp, struct nfsvattr *diraftp, int *diraft_retp, nfsattrbit_t *attrbitp, NFSACL_T *aclp, NFSPROC_T *p, struct nfsexstuff *exp, char *pathcp, int pathlen) { u_int32_t *tl; nd->nd_repstat = nfsvno_symlink(ndp, nvap, pathcp, pathlen, !(nd->nd_flag & ND_NFSV2), nd->nd_saveduid, nd->nd_cred, p, exp); if (!nd->nd_repstat && !(nd->nd_flag & ND_NFSV2)) { nfsrv_fixattr(nd, ndp->ni_vp, nvap, aclp, p, attrbitp, exp); if (nd->nd_flag & ND_NFSV3) { nd->nd_repstat = nfsvno_getfh(ndp->ni_vp, fhp, p); if (!nd->nd_repstat) nd->nd_repstat = nfsvno_getattr(ndp->ni_vp, nvap, nd->nd_cred, p, 1); } if (vpp != NULL && nd->nd_repstat == 0) { NFSVOPUNLOCK(ndp->ni_vp, 0); *vpp = ndp->ni_vp; } else vput(ndp->ni_vp); } if (dirp) { *diraft_retp = nfsvno_getattr(dirp, diraftp, nd->nd_cred, p, 0); vrele(dirp); } if ((nd->nd_flag & ND_NFSV4) && !nd->nd_repstat) { NFSM_BUILD(tl, u_int32_t *, 5 * NFSX_UNSIGNED); *tl++ = newnfs_false; txdr_hyper(dirforp->na_filerev, tl); tl += 2; txdr_hyper(diraftp->na_filerev, tl); (void) nfsrv_putattrbit(nd, attrbitp); } NFSEXITCODE2(0, nd); } /* * nfs mkdir service */ APPLESTATIC int nfsrvd_mkdir(struct nfsrv_descript *nd, __unused int isdgram, vnode_t dp, vnode_t *vpp, fhandle_t *fhp, NFSPROC_T *p, struct nfsexstuff *exp) { struct nfsvattr nva, dirfor, diraft; struct nameidata named; u_int32_t *tl; int error = 0, dirfor_ret = 1, diraft_ret = 1; vnode_t dirp = NULL; char *bufp; u_long *hashp; if (nd->nd_repstat) { nfsrv_wcc(nd, dirfor_ret, &dirfor, diraft_ret, &diraft); goto out; } NFSNAMEICNDSET(&named.ni_cnd, nd->nd_cred, CREATE, LOCKPARENT | SAVENAME | NOCACHE); nfsvno_setpathbuf(&named, &bufp, &hashp); error = nfsrv_parsename(nd, bufp, hashp, &named.ni_pathlen); if (error) goto nfsmout; if (!nd->nd_repstat) { NFSVNO_ATTRINIT(&nva); if (nd->nd_flag & ND_NFSV3) { error = nfsrv_sattr(nd, NULL, &nva, NULL, NULL, p); if (error) goto nfsmout; } else { NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); nva.na_mode = nfstov_mode(*tl++); } } if (!nd->nd_repstat) { nd->nd_repstat = nfsvno_namei(nd, &named, dp, 0, exp, p, &dirp); } else { vrele(dp); nfsvno_relpathbuf(&named); } if (dirp != NULL && !(nd->nd_flag & ND_NFSV3)) { vrele(dirp); dirp = NULL; } if (nd->nd_repstat) { if (dirp != NULL) { dirfor_ret = nfsvno_getattr(dirp, &dirfor, nd->nd_cred, p, 0); vrele(dirp); } if (nd->nd_flag & ND_NFSV3) nfsrv_wcc(nd, dirfor_ret, &dirfor, diraft_ret, &diraft); goto out; } if (dirp != NULL) dirfor_ret = nfsvno_getattr(dirp, &dirfor, nd->nd_cred, p, 0); /* * Call nfsrvd_mkdirsub() for the code common to V4 as well. */ nfsrvd_mkdirsub(nd, &named, &nva, fhp, vpp, dirp, &dirfor, &diraft, &diraft_ret, NULL, NULL, p, exp); if (nd->nd_flag & ND_NFSV3) { if (!nd->nd_repstat) { (void) nfsm_fhtom(nd, (u_int8_t *)fhp, 0, 1); nfsrv_postopattr(nd, 0, &nva); } nfsrv_wcc(nd, dirfor_ret, &dirfor, diraft_ret, &diraft); } else if (!nd->nd_repstat) { (void) nfsm_fhtom(nd, (u_int8_t *)fhp, 0, 0); nfsrv_fillattr(nd, &nva); } out: NFSEXITCODE2(0, nd); return (0); nfsmout: vrele(dp); nfsvno_relpathbuf(&named); NFSEXITCODE2(error, nd); return (error); } /* * Code common to mkdir for V2,3 and 4. */ static void nfsrvd_mkdirsub(struct nfsrv_descript *nd, struct nameidata *ndp, struct nfsvattr *nvap, fhandle_t *fhp, vnode_t *vpp, vnode_t dirp, struct nfsvattr *dirforp, struct nfsvattr *diraftp, int *diraft_retp, nfsattrbit_t *attrbitp, NFSACL_T *aclp, NFSPROC_T *p, struct nfsexstuff *exp) { vnode_t vp; u_int32_t *tl; NFSVNO_SETATTRVAL(nvap, type, VDIR); nd->nd_repstat = nfsvno_mkdir(ndp, nvap, nd->nd_saveduid, nd->nd_cred, p, exp); if (!nd->nd_repstat) { vp = ndp->ni_vp; nfsrv_fixattr(nd, vp, nvap, aclp, p, attrbitp, exp); nd->nd_repstat = nfsvno_getfh(vp, fhp, p); if (!(nd->nd_flag & ND_NFSV4) && !nd->nd_repstat) nd->nd_repstat = nfsvno_getattr(vp, nvap, nd->nd_cred, p, 1); if (vpp && !nd->nd_repstat) { NFSVOPUNLOCK(vp, 0); *vpp = vp; } else { vput(vp); } } if (dirp) { *diraft_retp = nfsvno_getattr(dirp, diraftp, nd->nd_cred, p, 0); vrele(dirp); } if ((nd->nd_flag & ND_NFSV4) && !nd->nd_repstat) { NFSM_BUILD(tl, u_int32_t *, 5 * NFSX_UNSIGNED); *tl++ = newnfs_false; txdr_hyper(dirforp->na_filerev, tl); tl += 2; txdr_hyper(diraftp->na_filerev, tl); (void) nfsrv_putattrbit(nd, attrbitp); } NFSEXITCODE2(0, nd); } /* * nfs commit service */ APPLESTATIC int nfsrvd_commit(struct nfsrv_descript *nd, __unused int isdgram, vnode_t vp, NFSPROC_T *p, __unused struct nfsexstuff *exp) { struct nfsvattr bfor, aft; u_int32_t *tl; int error = 0, for_ret = 1, aft_ret = 1, cnt; u_int64_t off; if (nd->nd_repstat) { nfsrv_wcc(nd, for_ret, &bfor, aft_ret, &aft); goto out; } /* Return NFSERR_ISDIR in NFSv4 when commit on a directory. */ if (vp->v_type != VREG) { if (nd->nd_flag & ND_NFSV3) error = NFSERR_NOTSUPP; else error = (vp->v_type == VDIR) ? NFSERR_ISDIR : NFSERR_INVAL; goto nfsmout; } NFSM_DISSECT(tl, u_int32_t *, 3 * NFSX_UNSIGNED); /* * XXX At this time VOP_FSYNC() does not accept offset and byte * count parameters, so these arguments are useless (someday maybe). */ off = fxdr_hyper(tl); tl += 2; cnt = fxdr_unsigned(int, *tl); if (nd->nd_flag & ND_NFSV3) for_ret = nfsvno_getattr(vp, &bfor, nd->nd_cred, p, 1); nd->nd_repstat = nfsvno_fsync(vp, off, cnt, nd->nd_cred, p); if (nd->nd_flag & ND_NFSV3) { aft_ret = nfsvno_getattr(vp, &aft, nd->nd_cred, p, 1); nfsrv_wcc(nd, for_ret, &bfor, aft_ret, &aft); } vput(vp); if (!nd->nd_repstat) { NFSM_BUILD(tl, u_int32_t *, NFSX_VERF); *tl++ = txdr_unsigned(nfsboottime.tv_sec); *tl = txdr_unsigned(nfsboottime.tv_usec); } out: NFSEXITCODE2(0, nd); return (0); nfsmout: vput(vp); NFSEXITCODE2(error, nd); return (error); } /* * nfs statfs service */ APPLESTATIC int nfsrvd_statfs(struct nfsrv_descript *nd, __unused int isdgram, vnode_t vp, NFSPROC_T *p, __unused struct nfsexstuff *exp) { struct statfs *sf; u_int32_t *tl; int getret = 1; struct nfsvattr at; - struct statfs sfs; u_quad_t tval; + sf = NULL; if (nd->nd_repstat) { nfsrv_postopattr(nd, getret, &at); goto out; } - sf = &sfs; + sf = malloc(sizeof(struct statfs), M_STATFS, M_WAITOK); nd->nd_repstat = nfsvno_statfs(vp, sf); getret = nfsvno_getattr(vp, &at, nd->nd_cred, p, 1); vput(vp); if (nd->nd_flag & ND_NFSV3) nfsrv_postopattr(nd, getret, &at); if (nd->nd_repstat) goto out; if (nd->nd_flag & ND_NFSV2) { NFSM_BUILD(tl, u_int32_t *, NFSX_V2STATFS); *tl++ = txdr_unsigned(NFS_V2MAXDATA); *tl++ = txdr_unsigned(sf->f_bsize); *tl++ = txdr_unsigned(sf->f_blocks); *tl++ = txdr_unsigned(sf->f_bfree); *tl = txdr_unsigned(sf->f_bavail); } else { NFSM_BUILD(tl, u_int32_t *, NFSX_V3STATFS); tval = (u_quad_t)sf->f_blocks; tval *= (u_quad_t)sf->f_bsize; txdr_hyper(tval, tl); tl += 2; tval = (u_quad_t)sf->f_bfree; tval *= (u_quad_t)sf->f_bsize; txdr_hyper(tval, tl); tl += 2; tval = (u_quad_t)sf->f_bavail; tval *= (u_quad_t)sf->f_bsize; txdr_hyper(tval, tl); tl += 2; tval = (u_quad_t)sf->f_files; txdr_hyper(tval, tl); tl += 2; tval = (u_quad_t)sf->f_ffree; txdr_hyper(tval, tl); tl += 2; tval = (u_quad_t)sf->f_ffree; txdr_hyper(tval, tl); tl += 2; *tl = 0; } out: + free(sf, M_STATFS); NFSEXITCODE2(0, nd); return (0); } /* * nfs fsinfo service */ APPLESTATIC int nfsrvd_fsinfo(struct nfsrv_descript *nd, int isdgram, vnode_t vp, NFSPROC_T *p, __unused struct nfsexstuff *exp) { u_int32_t *tl; struct nfsfsinfo fs; int getret = 1; struct nfsvattr at; if (nd->nd_repstat) { nfsrv_postopattr(nd, getret, &at); goto out; } getret = nfsvno_getattr(vp, &at, nd->nd_cred, p, 1); nfsvno_getfs(&fs, isdgram); vput(vp); nfsrv_postopattr(nd, getret, &at); NFSM_BUILD(tl, u_int32_t *, NFSX_V3FSINFO); *tl++ = txdr_unsigned(fs.fs_rtmax); *tl++ = txdr_unsigned(fs.fs_rtpref); *tl++ = txdr_unsigned(fs.fs_rtmult); *tl++ = txdr_unsigned(fs.fs_wtmax); *tl++ = txdr_unsigned(fs.fs_wtpref); *tl++ = txdr_unsigned(fs.fs_wtmult); *tl++ = txdr_unsigned(fs.fs_dtpref); txdr_hyper(fs.fs_maxfilesize, tl); tl += 2; txdr_nfsv3time(&fs.fs_timedelta, tl); tl += 2; *tl = txdr_unsigned(fs.fs_properties); out: NFSEXITCODE2(0, nd); return (0); } /* * nfs pathconf service */ APPLESTATIC int nfsrvd_pathconf(struct nfsrv_descript *nd, __unused int isdgram, vnode_t vp, NFSPROC_T *p, __unused struct nfsexstuff *exp) { struct nfsv3_pathconf *pc; int getret = 1; register_t linkmax, namemax, chownres, notrunc; struct nfsvattr at; if (nd->nd_repstat) { nfsrv_postopattr(nd, getret, &at); goto out; } nd->nd_repstat = nfsvno_pathconf(vp, _PC_LINK_MAX, &linkmax, nd->nd_cred, p); if (!nd->nd_repstat) nd->nd_repstat = nfsvno_pathconf(vp, _PC_NAME_MAX, &namemax, nd->nd_cred, p); if (!nd->nd_repstat) nd->nd_repstat=nfsvno_pathconf(vp, _PC_CHOWN_RESTRICTED, &chownres, nd->nd_cred, p); if (!nd->nd_repstat) nd->nd_repstat = nfsvno_pathconf(vp, _PC_NO_TRUNC, ¬runc, nd->nd_cred, p); getret = nfsvno_getattr(vp, &at, nd->nd_cred, p, 1); vput(vp); nfsrv_postopattr(nd, getret, &at); if (!nd->nd_repstat) { NFSM_BUILD(pc, struct nfsv3_pathconf *, NFSX_V3PATHCONF); pc->pc_linkmax = txdr_unsigned(linkmax); pc->pc_namemax = txdr_unsigned(namemax); pc->pc_notrunc = txdr_unsigned(notrunc); pc->pc_chownrestricted = txdr_unsigned(chownres); /* * These should probably be supported by VOP_PATHCONF(), but * until msdosfs is exportable (why would you want to?), the * Unix defaults should be ok. */ pc->pc_caseinsensitive = newnfs_false; pc->pc_casepreserving = newnfs_true; } out: NFSEXITCODE2(0, nd); return (0); } /* * nfsv4 lock service */ APPLESTATIC int nfsrvd_lock(struct nfsrv_descript *nd, __unused int isdgram, vnode_t vp, NFSPROC_T *p, struct nfsexstuff *exp) { u_int32_t *tl; int i; struct nfsstate *stp = NULL; struct nfslock *lop; struct nfslockconflict cf; int error = 0; u_short flags = NFSLCK_LOCK, lflags; u_int64_t offset, len; nfsv4stateid_t stateid; nfsquad_t clientid; NFSM_DISSECT(tl, u_int32_t *, 7 * NFSX_UNSIGNED); i = fxdr_unsigned(int, *tl++); switch (i) { case NFSV4LOCKT_READW: flags |= NFSLCK_BLOCKING; case NFSV4LOCKT_READ: lflags = NFSLCK_READ; break; case NFSV4LOCKT_WRITEW: flags |= NFSLCK_BLOCKING; case NFSV4LOCKT_WRITE: lflags = NFSLCK_WRITE; break; default: nd->nd_repstat = NFSERR_BADXDR; goto nfsmout; } if (*tl++ == newnfs_true) flags |= NFSLCK_RECLAIM; offset = fxdr_hyper(tl); tl += 2; len = fxdr_hyper(tl); tl += 2; if (*tl == newnfs_true) flags |= NFSLCK_OPENTOLOCK; if (flags & NFSLCK_OPENTOLOCK) { NFSM_DISSECT(tl, u_int32_t *, 5 * NFSX_UNSIGNED + NFSX_STATEID); i = fxdr_unsigned(int, *(tl+4+(NFSX_STATEID / NFSX_UNSIGNED))); if (i <= 0 || i > NFSV4_OPAQUELIMIT) { nd->nd_repstat = NFSERR_BADXDR; goto nfsmout; } MALLOC(stp, struct nfsstate *, sizeof (struct nfsstate) + i, M_NFSDSTATE, M_WAITOK); stp->ls_ownerlen = i; stp->ls_op = nd->nd_rp; stp->ls_seq = fxdr_unsigned(int, *tl++); stp->ls_stateid.seqid = fxdr_unsigned(u_int32_t, *tl++); NFSBCOPY((caddr_t)tl, (caddr_t)stp->ls_stateid.other, NFSX_STATEIDOTHER); tl += (NFSX_STATEIDOTHER / NFSX_UNSIGNED); stp->ls_opentolockseq = fxdr_unsigned(int, *tl++); clientid.lval[0] = *tl++; clientid.lval[1] = *tl++; if ((nd->nd_flag & ND_IMPLIEDCLID) != 0) { if ((nd->nd_flag & ND_NFSV41) != 0) clientid.qval = nd->nd_clientid.qval; else if (nd->nd_clientid.qval != clientid.qval) printf("EEK3 multiple clids\n"); } else { if ((nd->nd_flag & ND_NFSV41) != 0) printf("EEK! no clientid from session\n"); nd->nd_flag |= ND_IMPLIEDCLID; nd->nd_clientid.qval = clientid.qval; } error = nfsrv_mtostr(nd, stp->ls_owner, stp->ls_ownerlen); if (error) goto nfsmout; } else { NFSM_DISSECT(tl, u_int32_t *, NFSX_STATEID + NFSX_UNSIGNED); MALLOC(stp, struct nfsstate *, sizeof (struct nfsstate), M_NFSDSTATE, M_WAITOK); stp->ls_ownerlen = 0; stp->ls_op = nd->nd_rp; stp->ls_stateid.seqid = fxdr_unsigned(u_int32_t, *tl++); NFSBCOPY((caddr_t)tl, (caddr_t)stp->ls_stateid.other, NFSX_STATEIDOTHER); tl += (NFSX_STATEIDOTHER / NFSX_UNSIGNED); stp->ls_seq = fxdr_unsigned(int, *tl); clientid.lval[0] = stp->ls_stateid.other[0]; clientid.lval[1] = stp->ls_stateid.other[1]; if ((nd->nd_flag & ND_IMPLIEDCLID) != 0) { if ((nd->nd_flag & ND_NFSV41) != 0) clientid.qval = nd->nd_clientid.qval; else if (nd->nd_clientid.qval != clientid.qval) printf("EEK4 multiple clids\n"); } else { if ((nd->nd_flag & ND_NFSV41) != 0) printf("EEK! no clientid from session\n"); nd->nd_flag |= ND_IMPLIEDCLID; nd->nd_clientid.qval = clientid.qval; } } MALLOC(lop, struct nfslock *, sizeof (struct nfslock), M_NFSDLOCK, M_WAITOK); lop->lo_first = offset; if (len == NFS64BITSSET) { lop->lo_end = NFS64BITSSET; } else { lop->lo_end = offset + len; if (lop->lo_end <= lop->lo_first) nd->nd_repstat = NFSERR_INVAL; } lop->lo_flags = lflags; stp->ls_flags = flags; stp->ls_uid = nd->nd_cred->cr_uid; /* * Do basic access checking. */ if (!nd->nd_repstat && vnode_vtype(vp) != VREG) { if (vnode_vtype(vp) == VDIR) nd->nd_repstat = NFSERR_ISDIR; else nd->nd_repstat = NFSERR_INVAL; } if (!nd->nd_repstat) { if (lflags & NFSLCK_WRITE) { nd->nd_repstat = nfsvno_accchk(vp, VWRITE, nd->nd_cred, exp, p, NFSACCCHK_ALLOWOWNER, NFSACCCHK_VPISLOCKED, NULL); } else { nd->nd_repstat = nfsvno_accchk(vp, VREAD, nd->nd_cred, exp, p, NFSACCCHK_ALLOWOWNER, NFSACCCHK_VPISLOCKED, NULL); if (nd->nd_repstat) nd->nd_repstat = nfsvno_accchk(vp, VEXEC, nd->nd_cred, exp, p, NFSACCCHK_ALLOWOWNER, NFSACCCHK_VPISLOCKED, NULL); } } /* * We call nfsrv_lockctrl() even if nd_repstat set, so that the * seqid# gets updated. nfsrv_lockctrl() will return the value * of nd_repstat, if it gets that far. */ nd->nd_repstat = nfsrv_lockctrl(vp, &stp, &lop, &cf, clientid, &stateid, exp, nd, p); if (lop) FREE((caddr_t)lop, M_NFSDLOCK); if (stp) FREE((caddr_t)stp, M_NFSDSTATE); if (!nd->nd_repstat) { NFSM_BUILD(tl, u_int32_t *, NFSX_STATEID); *tl++ = txdr_unsigned(stateid.seqid); NFSBCOPY((caddr_t)stateid.other,(caddr_t)tl,NFSX_STATEIDOTHER); } else if (nd->nd_repstat == NFSERR_DENIED) { NFSM_BUILD(tl, u_int32_t *, 7 * NFSX_UNSIGNED); txdr_hyper(cf.cl_first, tl); tl += 2; if (cf.cl_end == NFS64BITSSET) len = NFS64BITSSET; else len = cf.cl_end - cf.cl_first; txdr_hyper(len, tl); tl += 2; if (cf.cl_flags == NFSLCK_WRITE) *tl++ = txdr_unsigned(NFSV4LOCKT_WRITE); else *tl++ = txdr_unsigned(NFSV4LOCKT_READ); *tl++ = stateid.other[0]; *tl = stateid.other[1]; (void) nfsm_strtom(nd, cf.cl_owner, cf.cl_ownerlen); } vput(vp); NFSEXITCODE2(0, nd); return (0); nfsmout: vput(vp); if (stp) free((caddr_t)stp, M_NFSDSTATE); NFSEXITCODE2(error, nd); return (error); } /* * nfsv4 lock test service */ APPLESTATIC int nfsrvd_lockt(struct nfsrv_descript *nd, __unused int isdgram, vnode_t vp, NFSPROC_T *p, struct nfsexstuff *exp) { u_int32_t *tl; int i; struct nfsstate *stp = NULL; struct nfslock lo, *lop = &lo; struct nfslockconflict cf; int error = 0; nfsv4stateid_t stateid; nfsquad_t clientid; u_int64_t len; NFSM_DISSECT(tl, u_int32_t *, 8 * NFSX_UNSIGNED); i = fxdr_unsigned(int, *(tl + 7)); if (i <= 0 || i > NFSV4_OPAQUELIMIT) { nd->nd_repstat = NFSERR_BADXDR; goto nfsmout; } MALLOC(stp, struct nfsstate *, sizeof (struct nfsstate) + i, M_NFSDSTATE, M_WAITOK); stp->ls_ownerlen = i; stp->ls_op = NULL; stp->ls_flags = NFSLCK_TEST; stp->ls_uid = nd->nd_cred->cr_uid; i = fxdr_unsigned(int, *tl++); switch (i) { case NFSV4LOCKT_READW: stp->ls_flags |= NFSLCK_BLOCKING; case NFSV4LOCKT_READ: lo.lo_flags = NFSLCK_READ; break; case NFSV4LOCKT_WRITEW: stp->ls_flags |= NFSLCK_BLOCKING; case NFSV4LOCKT_WRITE: lo.lo_flags = NFSLCK_WRITE; break; default: nd->nd_repstat = NFSERR_BADXDR; goto nfsmout; } lo.lo_first = fxdr_hyper(tl); tl += 2; len = fxdr_hyper(tl); if (len == NFS64BITSSET) { lo.lo_end = NFS64BITSSET; } else { lo.lo_end = lo.lo_first + len; if (lo.lo_end <= lo.lo_first) nd->nd_repstat = NFSERR_INVAL; } tl += 2; clientid.lval[0] = *tl++; clientid.lval[1] = *tl; if ((nd->nd_flag & ND_IMPLIEDCLID) != 0) { if ((nd->nd_flag & ND_NFSV41) != 0) clientid.qval = nd->nd_clientid.qval; else if (nd->nd_clientid.qval != clientid.qval) printf("EEK5 multiple clids\n"); } else { if ((nd->nd_flag & ND_NFSV41) != 0) printf("EEK! no clientid from session\n"); nd->nd_flag |= ND_IMPLIEDCLID; nd->nd_clientid.qval = clientid.qval; } error = nfsrv_mtostr(nd, stp->ls_owner, stp->ls_ownerlen); if (error) goto nfsmout; if (!nd->nd_repstat && vnode_vtype(vp) != VREG) { if (vnode_vtype(vp) == VDIR) nd->nd_repstat = NFSERR_ISDIR; else nd->nd_repstat = NFSERR_INVAL; } if (!nd->nd_repstat) nd->nd_repstat = nfsrv_lockctrl(vp, &stp, &lop, &cf, clientid, &stateid, exp, nd, p); if (nd->nd_repstat) { if (nd->nd_repstat == NFSERR_DENIED) { NFSM_BUILD(tl, u_int32_t *, 7 * NFSX_UNSIGNED); txdr_hyper(cf.cl_first, tl); tl += 2; if (cf.cl_end == NFS64BITSSET) len = NFS64BITSSET; else len = cf.cl_end - cf.cl_first; txdr_hyper(len, tl); tl += 2; if (cf.cl_flags == NFSLCK_WRITE) *tl++ = txdr_unsigned(NFSV4LOCKT_WRITE); else *tl++ = txdr_unsigned(NFSV4LOCKT_READ); *tl++ = stp->ls_stateid.other[0]; *tl = stp->ls_stateid.other[1]; (void) nfsm_strtom(nd, cf.cl_owner, cf.cl_ownerlen); } } vput(vp); if (stp) FREE((caddr_t)stp, M_NFSDSTATE); NFSEXITCODE2(0, nd); return (0); nfsmout: vput(vp); if (stp) free((caddr_t)stp, M_NFSDSTATE); NFSEXITCODE2(error, nd); return (error); } /* * nfsv4 unlock service */ APPLESTATIC int nfsrvd_locku(struct nfsrv_descript *nd, __unused int isdgram, vnode_t vp, NFSPROC_T *p, struct nfsexstuff *exp) { u_int32_t *tl; int i; struct nfsstate *stp; struct nfslock *lop; int error = 0; nfsv4stateid_t stateid; nfsquad_t clientid; u_int64_t len; NFSM_DISSECT(tl, u_int32_t *, 6 * NFSX_UNSIGNED + NFSX_STATEID); MALLOC(stp, struct nfsstate *, sizeof (struct nfsstate), M_NFSDSTATE, M_WAITOK); MALLOC(lop, struct nfslock *, sizeof (struct nfslock), M_NFSDLOCK, M_WAITOK); stp->ls_flags = NFSLCK_UNLOCK; lop->lo_flags = NFSLCK_UNLOCK; stp->ls_op = nd->nd_rp; i = fxdr_unsigned(int, *tl++); switch (i) { case NFSV4LOCKT_READW: stp->ls_flags |= NFSLCK_BLOCKING; case NFSV4LOCKT_READ: break; case NFSV4LOCKT_WRITEW: stp->ls_flags |= NFSLCK_BLOCKING; case NFSV4LOCKT_WRITE: break; default: nd->nd_repstat = NFSERR_BADXDR; free(stp, M_NFSDSTATE); free(lop, M_NFSDLOCK); goto nfsmout; } stp->ls_ownerlen = 0; stp->ls_uid = nd->nd_cred->cr_uid; stp->ls_seq = fxdr_unsigned(int, *tl++); stp->ls_stateid.seqid = fxdr_unsigned(u_int32_t, *tl++); NFSBCOPY((caddr_t)tl, (caddr_t)stp->ls_stateid.other, NFSX_STATEIDOTHER); tl += (NFSX_STATEIDOTHER / NFSX_UNSIGNED); lop->lo_first = fxdr_hyper(tl); tl += 2; len = fxdr_hyper(tl); if (len == NFS64BITSSET) { lop->lo_end = NFS64BITSSET; } else { lop->lo_end = lop->lo_first + len; if (lop->lo_end <= lop->lo_first) nd->nd_repstat = NFSERR_INVAL; } clientid.lval[0] = stp->ls_stateid.other[0]; clientid.lval[1] = stp->ls_stateid.other[1]; if ((nd->nd_flag & ND_IMPLIEDCLID) != 0) { if ((nd->nd_flag & ND_NFSV41) != 0) clientid.qval = nd->nd_clientid.qval; else if (nd->nd_clientid.qval != clientid.qval) printf("EEK6 multiple clids\n"); } else { if ((nd->nd_flag & ND_NFSV41) != 0) printf("EEK! no clientid from session\n"); nd->nd_flag |= ND_IMPLIEDCLID; nd->nd_clientid.qval = clientid.qval; } if (!nd->nd_repstat && vnode_vtype(vp) != VREG) { if (vnode_vtype(vp) == VDIR) nd->nd_repstat = NFSERR_ISDIR; else nd->nd_repstat = NFSERR_INVAL; } /* * Call nfsrv_lockctrl() even if nd_repstat is set, so that the * seqid# gets incremented. nfsrv_lockctrl() will return the * value of nd_repstat, if it gets that far. */ nd->nd_repstat = nfsrv_lockctrl(vp, &stp, &lop, NULL, clientid, &stateid, exp, nd, p); if (stp) FREE((caddr_t)stp, M_NFSDSTATE); if (lop) free((caddr_t)lop, M_NFSDLOCK); if (!nd->nd_repstat) { NFSM_BUILD(tl, u_int32_t *, NFSX_STATEID); *tl++ = txdr_unsigned(stateid.seqid); NFSBCOPY((caddr_t)stateid.other,(caddr_t)tl,NFSX_STATEIDOTHER); } nfsmout: vput(vp); NFSEXITCODE2(error, nd); return (error); } /* * nfsv4 open service */ APPLESTATIC int nfsrvd_open(struct nfsrv_descript *nd, __unused int isdgram, vnode_t dp, vnode_t *vpp, __unused fhandle_t *fhp, NFSPROC_T *p, struct nfsexstuff *exp) { u_int32_t *tl; int i, retext; struct nfsstate *stp = NULL; int error = 0, create, claim, exclusive_flag = 0; u_int32_t rflags = NFSV4OPEN_LOCKTYPEPOSIX, acemask; int how = NFSCREATE_UNCHECKED; int32_t cverf[2], tverf[2] = { 0, 0 }; vnode_t vp = NULL, dirp = NULL; struct nfsvattr nva, dirfor, diraft; struct nameidata named; nfsv4stateid_t stateid, delegstateid; nfsattrbit_t attrbits; nfsquad_t clientid; char *bufp = NULL; u_long *hashp; NFSACL_T *aclp = NULL; #ifdef NFS4_ACL_EXTATTR_NAME aclp = acl_alloc(M_WAITOK); aclp->acl_cnt = 0; #endif NFSZERO_ATTRBIT(&attrbits); named.ni_startdir = NULL; named.ni_cnd.cn_nameiop = 0; NFSM_DISSECT(tl, u_int32_t *, 6 * NFSX_UNSIGNED); i = fxdr_unsigned(int, *(tl + 5)); if (i <= 0 || i > NFSV4_OPAQUELIMIT) { nd->nd_repstat = NFSERR_BADXDR; goto nfsmout; } MALLOC(stp, struct nfsstate *, sizeof (struct nfsstate) + i, M_NFSDSTATE, M_WAITOK); stp->ls_ownerlen = i; stp->ls_op = nd->nd_rp; stp->ls_flags = NFSLCK_OPEN; stp->ls_uid = nd->nd_cred->cr_uid; stp->ls_seq = fxdr_unsigned(u_int32_t, *tl++); i = fxdr_unsigned(int, *tl++); retext = 0; if ((i & (NFSV4OPEN_WANTDELEGMASK | NFSV4OPEN_WANTSIGNALDELEG | NFSV4OPEN_WANTPUSHDELEG)) != 0 && (nd->nd_flag & ND_NFSV41) != 0) { retext = 1; /* For now, ignore these. */ i &= ~(NFSV4OPEN_WANTPUSHDELEG | NFSV4OPEN_WANTSIGNALDELEG); switch (i & NFSV4OPEN_WANTDELEGMASK) { case NFSV4OPEN_WANTANYDELEG: stp->ls_flags |= (NFSLCK_WANTRDELEG | NFSLCK_WANTWDELEG); i &= ~NFSV4OPEN_WANTDELEGMASK; break; case NFSV4OPEN_WANTREADDELEG: stp->ls_flags |= NFSLCK_WANTRDELEG; i &= ~NFSV4OPEN_WANTDELEGMASK; break; case NFSV4OPEN_WANTWRITEDELEG: stp->ls_flags |= NFSLCK_WANTWDELEG; i &= ~NFSV4OPEN_WANTDELEGMASK; break; case NFSV4OPEN_WANTNODELEG: stp->ls_flags |= NFSLCK_WANTNODELEG; i &= ~NFSV4OPEN_WANTDELEGMASK; break; case NFSV4OPEN_WANTCANCEL: printf("NFSv4: ignore Open WantCancel\n"); i &= ~NFSV4OPEN_WANTDELEGMASK; break; default: /* nd_repstat will be set to NFSERR_INVAL below. */ break; } } switch (i) { case NFSV4OPEN_ACCESSREAD: stp->ls_flags |= NFSLCK_READACCESS; break; case NFSV4OPEN_ACCESSWRITE: stp->ls_flags |= NFSLCK_WRITEACCESS; break; case NFSV4OPEN_ACCESSBOTH: stp->ls_flags |= (NFSLCK_READACCESS | NFSLCK_WRITEACCESS); break; default: nd->nd_repstat = NFSERR_INVAL; } i = fxdr_unsigned(int, *tl++); switch (i) { case NFSV4OPEN_DENYNONE: break; case NFSV4OPEN_DENYREAD: stp->ls_flags |= NFSLCK_READDENY; break; case NFSV4OPEN_DENYWRITE: stp->ls_flags |= NFSLCK_WRITEDENY; break; case NFSV4OPEN_DENYBOTH: stp->ls_flags |= (NFSLCK_READDENY | NFSLCK_WRITEDENY); break; default: nd->nd_repstat = NFSERR_INVAL; } clientid.lval[0] = *tl++; clientid.lval[1] = *tl; if ((nd->nd_flag & ND_IMPLIEDCLID) != 0) { if ((nd->nd_flag & ND_NFSV41) != 0) clientid.qval = nd->nd_clientid.qval; else if (nd->nd_clientid.qval != clientid.qval) printf("EEK7 multiple clids\n"); } else { if ((nd->nd_flag & ND_NFSV41) != 0) printf("EEK! no clientid from session\n"); nd->nd_flag |= ND_IMPLIEDCLID; nd->nd_clientid.qval = clientid.qval; } error = nfsrv_mtostr(nd, stp->ls_owner, stp->ls_ownerlen); if (error) goto nfsmout; NFSVNO_ATTRINIT(&nva); NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); create = fxdr_unsigned(int, *tl); if (!nd->nd_repstat) nd->nd_repstat = nfsvno_getattr(dp, &dirfor, nd->nd_cred, p, 0); if (create == NFSV4OPEN_CREATE) { nva.na_type = VREG; nva.na_mode = 0; NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); how = fxdr_unsigned(int, *tl); switch (how) { case NFSCREATE_UNCHECKED: case NFSCREATE_GUARDED: error = nfsv4_sattr(nd, NULL, &nva, &attrbits, aclp, p); if (error) goto nfsmout; /* * If the na_gid being set is the same as that of * the directory it is going in, clear it, since * that is what will be set by default. This allows * a user that isn't in that group to do the create. */ if (!nd->nd_repstat && NFSVNO_ISSETGID(&nva) && nva.na_gid == dirfor.na_gid) NFSVNO_UNSET(&nva, gid); if (!nd->nd_repstat) nd->nd_repstat = nfsrv_checkuidgid(nd, &nva); break; case NFSCREATE_EXCLUSIVE: NFSM_DISSECT(tl, u_int32_t *, NFSX_VERF); cverf[0] = *tl++; cverf[1] = *tl; break; case NFSCREATE_EXCLUSIVE41: NFSM_DISSECT(tl, u_int32_t *, NFSX_VERF); cverf[0] = *tl++; cverf[1] = *tl; error = nfsv4_sattr(nd, vp, &nva, &attrbits, aclp, p); if (error != 0) goto nfsmout; if (NFSISSET_ATTRBIT(&attrbits, NFSATTRBIT_TIMEACCESSSET)) nd->nd_repstat = NFSERR_INVAL; /* * If the na_gid being set is the same as that of * the directory it is going in, clear it, since * that is what will be set by default. This allows * a user that isn't in that group to do the create. */ if (nd->nd_repstat == 0 && NFSVNO_ISSETGID(&nva) && nva.na_gid == dirfor.na_gid) NFSVNO_UNSET(&nva, gid); if (nd->nd_repstat == 0) nd->nd_repstat = nfsrv_checkuidgid(nd, &nva); break; default: nd->nd_repstat = NFSERR_BADXDR; goto nfsmout; } } else if (create != NFSV4OPEN_NOCREATE) { nd->nd_repstat = NFSERR_BADXDR; goto nfsmout; } /* * Now, handle the claim, which usually includes looking up a * name in the directory referenced by dp. The exception is * NFSV4OPEN_CLAIMPREVIOUS. */ NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); claim = fxdr_unsigned(int, *tl); if (claim == NFSV4OPEN_CLAIMDELEGATECUR) { NFSM_DISSECT(tl, u_int32_t *, NFSX_STATEID); stateid.seqid = fxdr_unsigned(u_int32_t, *tl++); NFSBCOPY((caddr_t)tl,(caddr_t)stateid.other,NFSX_STATEIDOTHER); stp->ls_flags |= NFSLCK_DELEGCUR; } else if (claim == NFSV4OPEN_CLAIMDELEGATEPREV) { stp->ls_flags |= NFSLCK_DELEGPREV; } if (claim == NFSV4OPEN_CLAIMNULL || claim == NFSV4OPEN_CLAIMDELEGATECUR || claim == NFSV4OPEN_CLAIMDELEGATEPREV) { if (!nd->nd_repstat && create == NFSV4OPEN_CREATE && claim != NFSV4OPEN_CLAIMNULL) nd->nd_repstat = NFSERR_INVAL; if (nd->nd_repstat) { nd->nd_repstat = nfsrv_opencheck(clientid, &stateid, stp, NULL, nd, p, nd->nd_repstat); goto nfsmout; } if (create == NFSV4OPEN_CREATE) NFSNAMEICNDSET(&named.ni_cnd, nd->nd_cred, CREATE, LOCKPARENT | LOCKLEAF | SAVESTART | NOCACHE); else NFSNAMEICNDSET(&named.ni_cnd, nd->nd_cred, LOOKUP, LOCKLEAF | SAVESTART); nfsvno_setpathbuf(&named, &bufp, &hashp); error = nfsrv_parsename(nd, bufp, hashp, &named.ni_pathlen); if (error) { vrele(dp); #ifdef NFS4_ACL_EXTATTR_NAME acl_free(aclp); #endif FREE((caddr_t)stp, M_NFSDSTATE); nfsvno_relpathbuf(&named); NFSEXITCODE2(error, nd); return (error); } if (!nd->nd_repstat) { nd->nd_repstat = nfsvno_namei(nd, &named, dp, 0, exp, p, &dirp); } else { vrele(dp); nfsvno_relpathbuf(&named); } if (create == NFSV4OPEN_CREATE) { switch (how) { case NFSCREATE_UNCHECKED: if (named.ni_vp) { /* * Clear the setable attribute bits, except * for Size, if it is being truncated. */ NFSZERO_ATTRBIT(&attrbits); if (NFSVNO_ISSETSIZE(&nva)) NFSSETBIT_ATTRBIT(&attrbits, NFSATTRBIT_SIZE); } break; case NFSCREATE_GUARDED: if (named.ni_vp && !nd->nd_repstat) nd->nd_repstat = EEXIST; break; case NFSCREATE_EXCLUSIVE: exclusive_flag = 1; if (!named.ni_vp) nva.na_mode = 0; break; case NFSCREATE_EXCLUSIVE41: exclusive_flag = 1; break; } } nfsvno_open(nd, &named, clientid, &stateid, stp, &exclusive_flag, &nva, cverf, create, aclp, &attrbits, nd->nd_cred, p, exp, &vp); } else if (claim == NFSV4OPEN_CLAIMPREVIOUS || claim == NFSV4OPEN_CLAIMFH) { if (claim == NFSV4OPEN_CLAIMPREVIOUS) { NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); i = fxdr_unsigned(int, *tl); switch (i) { case NFSV4OPEN_DELEGATEREAD: stp->ls_flags |= NFSLCK_DELEGREAD; break; case NFSV4OPEN_DELEGATEWRITE: stp->ls_flags |= NFSLCK_DELEGWRITE; case NFSV4OPEN_DELEGATENONE: break; default: nd->nd_repstat = NFSERR_BADXDR; goto nfsmout; } stp->ls_flags |= NFSLCK_RECLAIM; } else { /* CLAIM_NULL_FH */ if (nd->nd_repstat == 0 && create == NFSV4OPEN_CREATE) nd->nd_repstat = NFSERR_INVAL; } vp = dp; NFSVOPLOCK(vp, LK_EXCLUSIVE | LK_RETRY); if ((vp->v_iflag & VI_DOOMED) == 0) nd->nd_repstat = nfsrv_opencheck(clientid, &stateid, stp, vp, nd, p, nd->nd_repstat); else nd->nd_repstat = NFSERR_PERM; } else { nd->nd_repstat = NFSERR_BADXDR; goto nfsmout; } /* * Do basic access checking. */ if (!nd->nd_repstat && vnode_vtype(vp) != VREG) { /* * The IETF working group decided that this is the correct * error return for all non-regular files. */ nd->nd_repstat = (vp->v_type == VDIR) ? NFSERR_ISDIR : NFSERR_SYMLINK; } if (!nd->nd_repstat && (stp->ls_flags & NFSLCK_WRITEACCESS)) nd->nd_repstat = nfsvno_accchk(vp, VWRITE, nd->nd_cred, exp, p, NFSACCCHK_ALLOWOWNER, NFSACCCHK_VPISLOCKED, NULL); if (!nd->nd_repstat && (stp->ls_flags & NFSLCK_READACCESS)) { nd->nd_repstat = nfsvno_accchk(vp, VREAD, nd->nd_cred, exp, p, NFSACCCHK_ALLOWOWNER, NFSACCCHK_VPISLOCKED, NULL); if (nd->nd_repstat) nd->nd_repstat = nfsvno_accchk(vp, VEXEC, nd->nd_cred, exp, p, NFSACCCHK_ALLOWOWNER, NFSACCCHK_VPISLOCKED, NULL); } if (!nd->nd_repstat) { nd->nd_repstat = nfsvno_getattr(vp, &nva, nd->nd_cred, p, 1); if (!nd->nd_repstat) { tverf[0] = nva.na_atime.tv_sec; tverf[1] = nva.na_atime.tv_nsec; } } if (!nd->nd_repstat && exclusive_flag && (cverf[0] != tverf[0] || cverf[1] != tverf[1])) nd->nd_repstat = EEXIST; /* * Do the open locking/delegation stuff. */ if (!nd->nd_repstat) nd->nd_repstat = nfsrv_openctrl(nd, vp, &stp, clientid, &stateid, &delegstateid, &rflags, exp, p, nva.na_filerev); /* * vp must be unlocked before the call to nfsvno_getattr(dirp,...) * below, to avoid a deadlock with the lookup in nfsvno_namei() above. * (ie: Leave the NFSVOPUNLOCK() about here.) */ if (vp) NFSVOPUNLOCK(vp, 0); if (stp) FREE((caddr_t)stp, M_NFSDSTATE); if (!nd->nd_repstat && dirp) nd->nd_repstat = nfsvno_getattr(dirp, &diraft, nd->nd_cred, p, 0); if (!nd->nd_repstat) { NFSM_BUILD(tl, u_int32_t *, NFSX_STATEID + 6 * NFSX_UNSIGNED); *tl++ = txdr_unsigned(stateid.seqid); NFSBCOPY((caddr_t)stateid.other,(caddr_t)tl,NFSX_STATEIDOTHER); tl += (NFSX_STATEIDOTHER / NFSX_UNSIGNED); if (claim == NFSV4OPEN_CLAIMPREVIOUS) { *tl++ = newnfs_true; *tl++ = 0; *tl++ = 0; *tl++ = 0; *tl++ = 0; } else { *tl++ = newnfs_false; /* Since dirp is not locked */ txdr_hyper(dirfor.na_filerev, tl); tl += 2; txdr_hyper(diraft.na_filerev, tl); tl += 2; } *tl = txdr_unsigned(rflags & NFSV4OPEN_RFLAGS); (void) nfsrv_putattrbit(nd, &attrbits); NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); if (rflags & NFSV4OPEN_READDELEGATE) *tl = txdr_unsigned(NFSV4OPEN_DELEGATEREAD); else if (rflags & NFSV4OPEN_WRITEDELEGATE) *tl = txdr_unsigned(NFSV4OPEN_DELEGATEWRITE); else if (retext != 0) { *tl = txdr_unsigned(NFSV4OPEN_DELEGATENONEEXT); if ((rflags & NFSV4OPEN_WDCONTENTION) != 0) { NFSM_BUILD(tl, u_int32_t *, 2 * NFSX_UNSIGNED); *tl++ = txdr_unsigned(NFSV4OPEN_CONTENTION); *tl = newnfs_false; } else if ((rflags & NFSV4OPEN_WDRESOURCE) != 0) { NFSM_BUILD(tl, u_int32_t *, 2 * NFSX_UNSIGNED); *tl++ = txdr_unsigned(NFSV4OPEN_RESOURCE); *tl = newnfs_false; } else { NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(NFSV4OPEN_NOTWANTED); } } else *tl = txdr_unsigned(NFSV4OPEN_DELEGATENONE); if (rflags & (NFSV4OPEN_READDELEGATE|NFSV4OPEN_WRITEDELEGATE)) { NFSM_BUILD(tl, u_int32_t *, NFSX_STATEID+NFSX_UNSIGNED); *tl++ = txdr_unsigned(delegstateid.seqid); NFSBCOPY((caddr_t)delegstateid.other, (caddr_t)tl, NFSX_STATEIDOTHER); tl += (NFSX_STATEIDOTHER / NFSX_UNSIGNED); if (rflags & NFSV4OPEN_RECALL) *tl = newnfs_true; else *tl = newnfs_false; if (rflags & NFSV4OPEN_WRITEDELEGATE) { NFSM_BUILD(tl, u_int32_t *, 3 * NFSX_UNSIGNED); *tl++ = txdr_unsigned(NFSV4OPEN_LIMITSIZE); txdr_hyper(nva.na_size, tl); } NFSM_BUILD(tl, u_int32_t *, 3 * NFSX_UNSIGNED); *tl++ = txdr_unsigned(NFSV4ACE_ALLOWEDTYPE); *tl++ = txdr_unsigned(0x0); acemask = NFSV4ACE_ALLFILESMASK; if (nva.na_mode & S_IRUSR) acemask |= NFSV4ACE_READMASK; if (nva.na_mode & S_IWUSR) acemask |= NFSV4ACE_WRITEMASK; if (nva.na_mode & S_IXUSR) acemask |= NFSV4ACE_EXECUTEMASK; *tl = txdr_unsigned(acemask); (void) nfsm_strtom(nd, "OWNER@", 6); } *vpp = vp; } else if (vp) { vrele(vp); } if (dirp) vrele(dirp); #ifdef NFS4_ACL_EXTATTR_NAME acl_free(aclp); #endif NFSEXITCODE2(0, nd); return (0); nfsmout: vrele(dp); #ifdef NFS4_ACL_EXTATTR_NAME acl_free(aclp); #endif if (stp) FREE((caddr_t)stp, M_NFSDSTATE); NFSEXITCODE2(error, nd); return (error); } /* * nfsv4 close service */ APPLESTATIC int nfsrvd_close(struct nfsrv_descript *nd, __unused int isdgram, vnode_t vp, NFSPROC_T *p, __unused struct nfsexstuff *exp) { u_int32_t *tl; struct nfsstate st, *stp = &st; int error = 0; nfsv4stateid_t stateid; nfsquad_t clientid; NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED + NFSX_STATEID); stp->ls_seq = fxdr_unsigned(u_int32_t, *tl++); stp->ls_ownerlen = 0; stp->ls_op = nd->nd_rp; stp->ls_uid = nd->nd_cred->cr_uid; stp->ls_stateid.seqid = fxdr_unsigned(u_int32_t, *tl++); NFSBCOPY((caddr_t)tl, (caddr_t)stp->ls_stateid.other, NFSX_STATEIDOTHER); stp->ls_flags = NFSLCK_CLOSE; clientid.lval[0] = stp->ls_stateid.other[0]; clientid.lval[1] = stp->ls_stateid.other[1]; if ((nd->nd_flag & ND_IMPLIEDCLID) != 0) { if ((nd->nd_flag & ND_NFSV41) != 0) clientid.qval = nd->nd_clientid.qval; else if (nd->nd_clientid.qval != clientid.qval) printf("EEK8 multiple clids\n"); } else { if ((nd->nd_flag & ND_NFSV41) != 0) printf("EEK! no clientid from session\n"); nd->nd_flag |= ND_IMPLIEDCLID; nd->nd_clientid.qval = clientid.qval; } nd->nd_repstat = nfsrv_openupdate(vp, stp, clientid, &stateid, nd, p); vput(vp); if (!nd->nd_repstat) { NFSM_BUILD(tl, u_int32_t *, NFSX_STATEID); *tl++ = txdr_unsigned(stateid.seqid); NFSBCOPY((caddr_t)stateid.other,(caddr_t)tl,NFSX_STATEIDOTHER); } NFSEXITCODE2(0, nd); return (0); nfsmout: vput(vp); NFSEXITCODE2(error, nd); return (error); } /* * nfsv4 delegpurge service */ APPLESTATIC int nfsrvd_delegpurge(struct nfsrv_descript *nd, __unused int isdgram, __unused vnode_t vp, __unused NFSPROC_T *p, __unused struct nfsexstuff *exp) { u_int32_t *tl; int error = 0; nfsquad_t clientid; if (nfs_rootfhset == 0 || nfsd_checkrootexp(nd) != 0) { nd->nd_repstat = NFSERR_WRONGSEC; goto nfsmout; } NFSM_DISSECT(tl, u_int32_t *, 2 * NFSX_UNSIGNED); clientid.lval[0] = *tl++; clientid.lval[1] = *tl; if ((nd->nd_flag & ND_IMPLIEDCLID) != 0) { if ((nd->nd_flag & ND_NFSV41) != 0) clientid.qval = nd->nd_clientid.qval; else if (nd->nd_clientid.qval != clientid.qval) printf("EEK9 multiple clids\n"); } else { if ((nd->nd_flag & ND_NFSV41) != 0) printf("EEK! no clientid from session\n"); nd->nd_flag |= ND_IMPLIEDCLID; nd->nd_clientid.qval = clientid.qval; } nd->nd_repstat = nfsrv_delegupdate(nd, clientid, NULL, NULL, NFSV4OP_DELEGPURGE, nd->nd_cred, p); nfsmout: NFSEXITCODE2(error, nd); return (error); } /* * nfsv4 delegreturn service */ APPLESTATIC int nfsrvd_delegreturn(struct nfsrv_descript *nd, __unused int isdgram, vnode_t vp, NFSPROC_T *p, __unused struct nfsexstuff *exp) { u_int32_t *tl; int error = 0; nfsv4stateid_t stateid; nfsquad_t clientid; NFSM_DISSECT(tl, u_int32_t *, NFSX_STATEID); stateid.seqid = fxdr_unsigned(u_int32_t, *tl++); NFSBCOPY((caddr_t)tl, (caddr_t)stateid.other, NFSX_STATEIDOTHER); clientid.lval[0] = stateid.other[0]; clientid.lval[1] = stateid.other[1]; if ((nd->nd_flag & ND_IMPLIEDCLID) != 0) { if ((nd->nd_flag & ND_NFSV41) != 0) clientid.qval = nd->nd_clientid.qval; else if (nd->nd_clientid.qval != clientid.qval) printf("EEK10 multiple clids\n"); } else { if ((nd->nd_flag & ND_NFSV41) != 0) printf("EEK! no clientid from session\n"); nd->nd_flag |= ND_IMPLIEDCLID; nd->nd_clientid.qval = clientid.qval; } nd->nd_repstat = nfsrv_delegupdate(nd, clientid, &stateid, vp, NFSV4OP_DELEGRETURN, nd->nd_cred, p); nfsmout: vput(vp); NFSEXITCODE2(error, nd); return (error); } /* * nfsv4 get file handle service */ APPLESTATIC int nfsrvd_getfh(struct nfsrv_descript *nd, __unused int isdgram, vnode_t vp, NFSPROC_T *p, __unused struct nfsexstuff *exp) { fhandle_t fh; nd->nd_repstat = nfsvno_getfh(vp, &fh, p); vput(vp); if (!nd->nd_repstat) (void) nfsm_fhtom(nd, (u_int8_t *)&fh, 0, 0); NFSEXITCODE2(0, nd); return (0); } /* * nfsv4 open confirm service */ APPLESTATIC int nfsrvd_openconfirm(struct nfsrv_descript *nd, __unused int isdgram, vnode_t vp, NFSPROC_T *p, __unused struct nfsexstuff *exp) { u_int32_t *tl; struct nfsstate st, *stp = &st; int error = 0; nfsv4stateid_t stateid; nfsquad_t clientid; if ((nd->nd_flag & ND_NFSV41) != 0) { nd->nd_repstat = NFSERR_NOTSUPP; goto nfsmout; } NFSM_DISSECT(tl, u_int32_t *, NFSX_STATEID + NFSX_UNSIGNED); stp->ls_ownerlen = 0; stp->ls_op = nd->nd_rp; stp->ls_uid = nd->nd_cred->cr_uid; stp->ls_stateid.seqid = fxdr_unsigned(u_int32_t, *tl++); NFSBCOPY((caddr_t)tl, (caddr_t)stp->ls_stateid.other, NFSX_STATEIDOTHER); tl += (NFSX_STATEIDOTHER / NFSX_UNSIGNED); stp->ls_seq = fxdr_unsigned(u_int32_t, *tl); stp->ls_flags = NFSLCK_CONFIRM; clientid.lval[0] = stp->ls_stateid.other[0]; clientid.lval[1] = stp->ls_stateid.other[1]; if ((nd->nd_flag & ND_IMPLIEDCLID) != 0) { if ((nd->nd_flag & ND_NFSV41) != 0) clientid.qval = nd->nd_clientid.qval; else if (nd->nd_clientid.qval != clientid.qval) printf("EEK11 multiple clids\n"); } else { if ((nd->nd_flag & ND_NFSV41) != 0) printf("EEK! no clientid from session\n"); nd->nd_flag |= ND_IMPLIEDCLID; nd->nd_clientid.qval = clientid.qval; } nd->nd_repstat = nfsrv_openupdate(vp, stp, clientid, &stateid, nd, p); if (!nd->nd_repstat) { NFSM_BUILD(tl, u_int32_t *, NFSX_STATEID); *tl++ = txdr_unsigned(stateid.seqid); NFSBCOPY((caddr_t)stateid.other,(caddr_t)tl,NFSX_STATEIDOTHER); } nfsmout: vput(vp); NFSEXITCODE2(error, nd); return (error); } /* * nfsv4 open downgrade service */ APPLESTATIC int nfsrvd_opendowngrade(struct nfsrv_descript *nd, __unused int isdgram, vnode_t vp, NFSPROC_T *p, __unused struct nfsexstuff *exp) { u_int32_t *tl; int i; struct nfsstate st, *stp = &st; int error = 0; nfsv4stateid_t stateid; nfsquad_t clientid; /* opendowngrade can only work on a file object.*/ if (vp->v_type != VREG) { error = NFSERR_INVAL; goto nfsmout; } NFSM_DISSECT(tl, u_int32_t *, NFSX_STATEID + 3 * NFSX_UNSIGNED); stp->ls_ownerlen = 0; stp->ls_op = nd->nd_rp; stp->ls_uid = nd->nd_cred->cr_uid; stp->ls_stateid.seqid = fxdr_unsigned(u_int32_t, *tl++); NFSBCOPY((caddr_t)tl, (caddr_t)stp->ls_stateid.other, NFSX_STATEIDOTHER); tl += (NFSX_STATEIDOTHER / NFSX_UNSIGNED); stp->ls_seq = fxdr_unsigned(u_int32_t, *tl++); i = fxdr_unsigned(int, *tl++); switch (i) { case NFSV4OPEN_ACCESSREAD: stp->ls_flags = (NFSLCK_READACCESS | NFSLCK_DOWNGRADE); break; case NFSV4OPEN_ACCESSWRITE: stp->ls_flags = (NFSLCK_WRITEACCESS | NFSLCK_DOWNGRADE); break; case NFSV4OPEN_ACCESSBOTH: stp->ls_flags = (NFSLCK_READACCESS | NFSLCK_WRITEACCESS | NFSLCK_DOWNGRADE); break; default: nd->nd_repstat = NFSERR_BADXDR; } i = fxdr_unsigned(int, *tl); switch (i) { case NFSV4OPEN_DENYNONE: break; case NFSV4OPEN_DENYREAD: stp->ls_flags |= NFSLCK_READDENY; break; case NFSV4OPEN_DENYWRITE: stp->ls_flags |= NFSLCK_WRITEDENY; break; case NFSV4OPEN_DENYBOTH: stp->ls_flags |= (NFSLCK_READDENY | NFSLCK_WRITEDENY); break; default: nd->nd_repstat = NFSERR_BADXDR; } clientid.lval[0] = stp->ls_stateid.other[0]; clientid.lval[1] = stp->ls_stateid.other[1]; if ((nd->nd_flag & ND_IMPLIEDCLID) != 0) { if ((nd->nd_flag & ND_NFSV41) != 0) clientid.qval = nd->nd_clientid.qval; else if (nd->nd_clientid.qval != clientid.qval) printf("EEK12 multiple clids\n"); } else { if ((nd->nd_flag & ND_NFSV41) != 0) printf("EEK! no clientid from session\n"); nd->nd_flag |= ND_IMPLIEDCLID; nd->nd_clientid.qval = clientid.qval; } if (!nd->nd_repstat) nd->nd_repstat = nfsrv_openupdate(vp, stp, clientid, &stateid, nd, p); if (!nd->nd_repstat) { NFSM_BUILD(tl, u_int32_t *, NFSX_STATEID); *tl++ = txdr_unsigned(stateid.seqid); NFSBCOPY((caddr_t)stateid.other,(caddr_t)tl,NFSX_STATEIDOTHER); } nfsmout: vput(vp); NFSEXITCODE2(error, nd); return (error); } /* * nfsv4 renew lease service */ APPLESTATIC int nfsrvd_renew(struct nfsrv_descript *nd, __unused int isdgram, __unused vnode_t vp, NFSPROC_T *p, __unused struct nfsexstuff *exp) { u_int32_t *tl; int error = 0; nfsquad_t clientid; if ((nd->nd_flag & ND_NFSV41) != 0) { nd->nd_repstat = NFSERR_NOTSUPP; goto nfsmout; } if (nfs_rootfhset == 0 || nfsd_checkrootexp(nd) != 0) { nd->nd_repstat = NFSERR_WRONGSEC; goto nfsmout; } NFSM_DISSECT(tl, u_int32_t *, NFSX_HYPER); clientid.lval[0] = *tl++; clientid.lval[1] = *tl; if ((nd->nd_flag & ND_IMPLIEDCLID) != 0) { if ((nd->nd_flag & ND_NFSV41) != 0) clientid.qval = nd->nd_clientid.qval; else if (nd->nd_clientid.qval != clientid.qval) printf("EEK13 multiple clids\n"); } else { if ((nd->nd_flag & ND_NFSV41) != 0) printf("EEK! no clientid from session\n"); nd->nd_flag |= ND_IMPLIEDCLID; nd->nd_clientid.qval = clientid.qval; } nd->nd_repstat = nfsrv_getclient(clientid, (CLOPS_RENEWOP|CLOPS_RENEW), NULL, NULL, (nfsquad_t)((u_quad_t)0), 0, nd, p); nfsmout: NFSEXITCODE2(error, nd); return (error); } /* * nfsv4 security info service */ APPLESTATIC int nfsrvd_secinfo(struct nfsrv_descript *nd, int isdgram, vnode_t dp, NFSPROC_T *p, struct nfsexstuff *exp) { u_int32_t *tl; int len; struct nameidata named; vnode_t dirp = NULL, vp; struct nfsrvfh fh; struct nfsexstuff retnes; u_int32_t *sizp; int error = 0, savflag, i; char *bufp; u_long *hashp; /* * All this just to get the export flags for the name. */ NFSNAMEICNDSET(&named.ni_cnd, nd->nd_cred, LOOKUP, LOCKLEAF | SAVESTART); nfsvno_setpathbuf(&named, &bufp, &hashp); error = nfsrv_parsename(nd, bufp, hashp, &named.ni_pathlen); if (error) { vput(dp); nfsvno_relpathbuf(&named); goto out; } if (!nd->nd_repstat) { nd->nd_repstat = nfsvno_namei(nd, &named, dp, 1, exp, p, &dirp); } else { vput(dp); nfsvno_relpathbuf(&named); } if (dirp) vrele(dirp); if (nd->nd_repstat) goto out; vrele(named.ni_startdir); nfsvno_relpathbuf(&named); fh.nfsrvfh_len = NFSX_MYFH; vp = named.ni_vp; nd->nd_repstat = nfsvno_getfh(vp, (fhandle_t *)fh.nfsrvfh_data, p); vput(vp); savflag = nd->nd_flag; if (!nd->nd_repstat) { nfsd_fhtovp(nd, &fh, LK_SHARED, &vp, &retnes, NULL, 0, p); if (vp) vput(vp); } nd->nd_flag = savflag; if (nd->nd_repstat) goto out; /* * Finally have the export flags for name, so we can create * the security info. */ len = 0; NFSM_BUILD(sizp, u_int32_t *, NFSX_UNSIGNED); for (i = 0; i < retnes.nes_numsecflavor; i++) { if (retnes.nes_secflavors[i] == AUTH_SYS) { NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl = txdr_unsigned(RPCAUTH_UNIX); len++; } else if (retnes.nes_secflavors[i] == RPCSEC_GSS_KRB5) { NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl++ = txdr_unsigned(RPCAUTH_GSS); (void) nfsm_strtom(nd, nfsgss_mechlist[KERBV_MECH].str, nfsgss_mechlist[KERBV_MECH].len); NFSM_BUILD(tl, u_int32_t *, 2 * NFSX_UNSIGNED); *tl++ = txdr_unsigned(GSS_KERBV_QOP); *tl = txdr_unsigned(RPCAUTHGSS_SVCNONE); len++; } else if (retnes.nes_secflavors[i] == RPCSEC_GSS_KRB5I) { NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl++ = txdr_unsigned(RPCAUTH_GSS); (void) nfsm_strtom(nd, nfsgss_mechlist[KERBV_MECH].str, nfsgss_mechlist[KERBV_MECH].len); NFSM_BUILD(tl, u_int32_t *, 2 * NFSX_UNSIGNED); *tl++ = txdr_unsigned(GSS_KERBV_QOP); *tl = txdr_unsigned(RPCAUTHGSS_SVCINTEGRITY); len++; } else if (retnes.nes_secflavors[i] == RPCSEC_GSS_KRB5P) { NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED); *tl++ = txdr_unsigned(RPCAUTH_GSS); (void) nfsm_strtom(nd, nfsgss_mechlist[KERBV_MECH].str, nfsgss_mechlist[KERBV_MECH].len); NFSM_BUILD(tl, u_int32_t *, 2 * NFSX_UNSIGNED); *tl++ = txdr_unsigned(GSS_KERBV_QOP); *tl = txdr_unsigned(RPCAUTHGSS_SVCPRIVACY); len++; } } *sizp = txdr_unsigned(len); out: NFSEXITCODE2(error, nd); return (error); } /* * nfsv4 set client id service */ APPLESTATIC int nfsrvd_setclientid(struct nfsrv_descript *nd, __unused int isdgram, __unused vnode_t vp, NFSPROC_T *p, __unused struct nfsexstuff *exp) { u_int32_t *tl; int i; int error = 0, idlen; struct nfsclient *clp = NULL; struct sockaddr_in *rad; u_char *verf, *ucp, *ucp2, addrbuf[24]; nfsquad_t clientid, confirm; if ((nd->nd_flag & ND_NFSV41) != 0) { nd->nd_repstat = NFSERR_NOTSUPP; goto nfsmout; } if (nfs_rootfhset == 0 || nfsd_checkrootexp(nd) != 0) { nd->nd_repstat = NFSERR_WRONGSEC; goto out; } NFSM_DISSECT(tl, u_int32_t *, NFSX_VERF + NFSX_UNSIGNED); verf = (u_char *)tl; tl += (NFSX_VERF / NFSX_UNSIGNED); i = fxdr_unsigned(int, *tl); if (i > NFSV4_OPAQUELIMIT || i <= 0) { nd->nd_repstat = NFSERR_BADXDR; goto nfsmout; } idlen = i; if (nd->nd_flag & ND_GSS) i += nd->nd_princlen; clp = malloc(sizeof(struct nfsclient) + i, M_NFSDCLIENT, M_WAITOK | M_ZERO); clp->lc_stateid = malloc(sizeof(struct nfsstatehead) * nfsrv_statehashsize, M_NFSDCLIENT, M_WAITOK); NFSINITSOCKMUTEX(&clp->lc_req.nr_mtx); NFSSOCKADDRALLOC(clp->lc_req.nr_nam); NFSSOCKADDRSIZE(clp->lc_req.nr_nam, sizeof (struct sockaddr_in)); clp->lc_req.nr_cred = NULL; NFSBCOPY(verf, clp->lc_verf, NFSX_VERF); clp->lc_idlen = idlen; error = nfsrv_mtostr(nd, clp->lc_id, idlen); if (error) goto nfsmout; if (nd->nd_flag & ND_GSS) { clp->lc_flags = LCL_GSS; if (nd->nd_flag & ND_GSSINTEGRITY) clp->lc_flags |= LCL_GSSINTEGRITY; else if (nd->nd_flag & ND_GSSPRIVACY) clp->lc_flags |= LCL_GSSPRIVACY; } else { clp->lc_flags = 0; } if ((nd->nd_flag & ND_GSS) && nd->nd_princlen > 0) { clp->lc_flags |= LCL_NAME; clp->lc_namelen = nd->nd_princlen; clp->lc_name = &clp->lc_id[idlen]; NFSBCOPY(nd->nd_principal, clp->lc_name, clp->lc_namelen); } else { clp->lc_uid = nd->nd_cred->cr_uid; clp->lc_gid = nd->nd_cred->cr_gid; } NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); clp->lc_program = fxdr_unsigned(u_int32_t, *tl); error = nfsrv_getclientipaddr(nd, clp); if (error) goto nfsmout; NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); clp->lc_callback = fxdr_unsigned(u_int32_t, *tl); /* * nfsrv_setclient() does the actual work of adding it to the * client list. If there is no error, the structure has been * linked into the client list and clp should no longer be used * here. When an error is returned, it has not been linked in, * so it should be free'd. */ nd->nd_repstat = nfsrv_setclient(nd, &clp, &clientid, &confirm, p); if (nd->nd_repstat == NFSERR_CLIDINUSE) { if (clp->lc_flags & LCL_TCPCALLBACK) (void) nfsm_strtom(nd, "tcp", 3); else (void) nfsm_strtom(nd, "udp", 3); rad = NFSSOCKADDR(clp->lc_req.nr_nam, struct sockaddr_in *); ucp = (u_char *)&rad->sin_addr.s_addr; ucp2 = (u_char *)&rad->sin_port; sprintf(addrbuf, "%d.%d.%d.%d.%d.%d", ucp[0] & 0xff, ucp[1] & 0xff, ucp[2] & 0xff, ucp[3] & 0xff, ucp2[0] & 0xff, ucp2[1] & 0xff); (void) nfsm_strtom(nd, addrbuf, strlen(addrbuf)); } if (clp) { NFSSOCKADDRFREE(clp->lc_req.nr_nam); NFSFREEMUTEX(&clp->lc_req.nr_mtx); free(clp->lc_stateid, M_NFSDCLIENT); free(clp, M_NFSDCLIENT); } if (!nd->nd_repstat) { NFSM_BUILD(tl, u_int32_t *, 2 * NFSX_HYPER); *tl++ = clientid.lval[0]; *tl++ = clientid.lval[1]; *tl++ = confirm.lval[0]; *tl = confirm.lval[1]; } out: NFSEXITCODE2(0, nd); return (0); nfsmout: if (clp) { NFSSOCKADDRFREE(clp->lc_req.nr_nam); NFSFREEMUTEX(&clp->lc_req.nr_mtx); free(clp->lc_stateid, M_NFSDCLIENT); free(clp, M_NFSDCLIENT); } NFSEXITCODE2(error, nd); return (error); } /* * nfsv4 set client id confirm service */ APPLESTATIC int nfsrvd_setclientidcfrm(struct nfsrv_descript *nd, __unused int isdgram, __unused vnode_t vp, NFSPROC_T *p, __unused struct nfsexstuff *exp) { u_int32_t *tl; int error = 0; nfsquad_t clientid, confirm; if ((nd->nd_flag & ND_NFSV41) != 0) { nd->nd_repstat = NFSERR_NOTSUPP; goto nfsmout; } if (nfs_rootfhset == 0 || nfsd_checkrootexp(nd) != 0) { nd->nd_repstat = NFSERR_WRONGSEC; goto nfsmout; } NFSM_DISSECT(tl, u_int32_t *, 2 * NFSX_HYPER); clientid.lval[0] = *tl++; clientid.lval[1] = *tl++; confirm.lval[0] = *tl++; confirm.lval[1] = *tl; /* * nfsrv_getclient() searches the client list for a match and * returns the appropriate NFSERR status. */ nd->nd_repstat = nfsrv_getclient(clientid, (CLOPS_CONFIRM|CLOPS_RENEW), NULL, NULL, confirm, 0, nd, p); nfsmout: NFSEXITCODE2(error, nd); return (error); } /* * nfsv4 verify service */ APPLESTATIC int nfsrvd_verify(struct nfsrv_descript *nd, int isdgram, vnode_t vp, NFSPROC_T *p, __unused struct nfsexstuff *exp) { int error = 0, ret, fhsize = NFSX_MYFH; struct nfsvattr nva; - struct statfs sf; + struct statfs *sf; struct nfsfsinfo fs; fhandle_t fh; + sf = malloc(sizeof(struct statfs), M_STATFS, M_WAITOK); nd->nd_repstat = nfsvno_getattr(vp, &nva, nd->nd_cred, p, 1); if (!nd->nd_repstat) - nd->nd_repstat = nfsvno_statfs(vp, &sf); + nd->nd_repstat = nfsvno_statfs(vp, sf); if (!nd->nd_repstat) nd->nd_repstat = nfsvno_getfh(vp, &fh, p); if (!nd->nd_repstat) { nfsvno_getfs(&fs, isdgram); error = nfsv4_loadattr(nd, vp, &nva, NULL, &fh, fhsize, NULL, - &sf, NULL, &fs, NULL, 1, &ret, NULL, NULL, p, nd->nd_cred); + sf, NULL, &fs, NULL, 1, &ret, NULL, NULL, p, nd->nd_cred); if (!error) { if (nd->nd_procnum == NFSV4OP_NVERIFY) { if (ret == 0) nd->nd_repstat = NFSERR_SAME; else if (ret != NFSERR_NOTSAME) nd->nd_repstat = ret; } else if (ret) nd->nd_repstat = ret; } } vput(vp); + free(sf, M_STATFS); NFSEXITCODE2(error, nd); return (error); } /* * nfs openattr rpc */ APPLESTATIC int nfsrvd_openattr(struct nfsrv_descript *nd, __unused int isdgram, vnode_t dp, __unused vnode_t *vpp, __unused fhandle_t *fhp, __unused NFSPROC_T *p, __unused struct nfsexstuff *exp) { u_int32_t *tl; int error = 0, createdir; NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); createdir = fxdr_unsigned(int, *tl); nd->nd_repstat = NFSERR_NOTSUPP; nfsmout: vrele(dp); NFSEXITCODE2(error, nd); return (error); } /* * nfsv4 release lock owner service */ APPLESTATIC int nfsrvd_releaselckown(struct nfsrv_descript *nd, __unused int isdgram, __unused vnode_t vp, NFSPROC_T *p, __unused struct nfsexstuff *exp) { u_int32_t *tl; struct nfsstate *stp = NULL; int error = 0, len; nfsquad_t clientid; if ((nd->nd_flag & ND_NFSV41) != 0) { nd->nd_repstat = NFSERR_NOTSUPP; goto nfsmout; } if (nfs_rootfhset == 0 || nfsd_checkrootexp(nd) != 0) { nd->nd_repstat = NFSERR_WRONGSEC; goto nfsmout; } NFSM_DISSECT(tl, u_int32_t *, 3 * NFSX_UNSIGNED); len = fxdr_unsigned(int, *(tl + 2)); if (len <= 0 || len > NFSV4_OPAQUELIMIT) { nd->nd_repstat = NFSERR_BADXDR; goto nfsmout; } MALLOC(stp, struct nfsstate *, sizeof (struct nfsstate) + len, M_NFSDSTATE, M_WAITOK); stp->ls_ownerlen = len; stp->ls_op = NULL; stp->ls_flags = NFSLCK_RELEASE; stp->ls_uid = nd->nd_cred->cr_uid; clientid.lval[0] = *tl++; clientid.lval[1] = *tl; if ((nd->nd_flag & ND_IMPLIEDCLID) != 0) { if ((nd->nd_flag & ND_NFSV41) != 0) clientid.qval = nd->nd_clientid.qval; else if (nd->nd_clientid.qval != clientid.qval) printf("EEK14 multiple clids\n"); } else { if ((nd->nd_flag & ND_NFSV41) != 0) printf("EEK! no clientid from session\n"); nd->nd_flag |= ND_IMPLIEDCLID; nd->nd_clientid.qval = clientid.qval; } error = nfsrv_mtostr(nd, stp->ls_owner, len); if (error) goto nfsmout; nd->nd_repstat = nfsrv_releaselckown(stp, clientid, p); FREE((caddr_t)stp, M_NFSDSTATE); NFSEXITCODE2(0, nd); return (0); nfsmout: if (stp) free((caddr_t)stp, M_NFSDSTATE); NFSEXITCODE2(error, nd); return (error); } /* * nfsv4 exchange_id service */ APPLESTATIC int nfsrvd_exchangeid(struct nfsrv_descript *nd, __unused int isdgram, __unused vnode_t vp, NFSPROC_T *p, __unused struct nfsexstuff *exp) { uint32_t *tl; int error = 0, i, idlen; struct nfsclient *clp = NULL; nfsquad_t clientid, confirm; uint8_t *verf; uint32_t sp4type, v41flags; uint64_t owner_minor; struct timespec verstime; if (nfs_rootfhset == 0 || nfsd_checkrootexp(nd) != 0) { nd->nd_repstat = NFSERR_WRONGSEC; goto nfsmout; } NFSM_DISSECT(tl, u_int32_t *, NFSX_VERF + NFSX_UNSIGNED); verf = (uint8_t *)tl; tl += (NFSX_VERF / NFSX_UNSIGNED); i = fxdr_unsigned(int, *tl); if (i > NFSV4_OPAQUELIMIT || i <= 0) { nd->nd_repstat = NFSERR_BADXDR; goto nfsmout; } idlen = i; if (nd->nd_flag & ND_GSS) i += nd->nd_princlen; clp = malloc(sizeof(struct nfsclient) + i, M_NFSDCLIENT, M_WAITOK | M_ZERO); clp->lc_stateid = malloc(sizeof(struct nfsstatehead) * nfsrv_statehashsize, M_NFSDCLIENT, M_WAITOK); NFSINITSOCKMUTEX(&clp->lc_req.nr_mtx); NFSSOCKADDRALLOC(clp->lc_req.nr_nam); NFSSOCKADDRSIZE(clp->lc_req.nr_nam, sizeof (struct sockaddr_in)); clp->lc_req.nr_cred = NULL; NFSBCOPY(verf, clp->lc_verf, NFSX_VERF); clp->lc_idlen = idlen; error = nfsrv_mtostr(nd, clp->lc_id, idlen); if (error != 0) goto nfsmout; if ((nd->nd_flag & ND_GSS) != 0) { clp->lc_flags = LCL_GSS | LCL_NFSV41; if ((nd->nd_flag & ND_GSSINTEGRITY) != 0) clp->lc_flags |= LCL_GSSINTEGRITY; else if ((nd->nd_flag & ND_GSSPRIVACY) != 0) clp->lc_flags |= LCL_GSSPRIVACY; } else clp->lc_flags = LCL_NFSV41; if ((nd->nd_flag & ND_GSS) != 0 && nd->nd_princlen > 0) { clp->lc_flags |= LCL_NAME; clp->lc_namelen = nd->nd_princlen; clp->lc_name = &clp->lc_id[idlen]; NFSBCOPY(nd->nd_principal, clp->lc_name, clp->lc_namelen); } else { clp->lc_uid = nd->nd_cred->cr_uid; clp->lc_gid = nd->nd_cred->cr_gid; } NFSM_DISSECT(tl, u_int32_t *, 2 * NFSX_UNSIGNED); v41flags = fxdr_unsigned(uint32_t, *tl++); if ((v41flags & ~(NFSV4EXCH_SUPPMOVEDREFER | NFSV4EXCH_SUPPMOVEDMIGR | NFSV4EXCH_BINDPRINCSTATEID | NFSV4EXCH_MASKPNFS | NFSV4EXCH_UPDCONFIRMEDRECA)) != 0) { nd->nd_repstat = NFSERR_INVAL; goto nfsmout; } if ((v41flags & NFSV4EXCH_UPDCONFIRMEDRECA) != 0) confirm.lval[1] = 1; else confirm.lval[1] = 0; v41flags = NFSV4EXCH_USENONPNFS; sp4type = fxdr_unsigned(uint32_t, *tl); if (sp4type != NFSV4EXCH_SP4NONE) { nd->nd_repstat = NFSERR_NOTSUPP; goto nfsmout; } /* * nfsrv_setclient() does the actual work of adding it to the * client list. If there is no error, the structure has been * linked into the client list and clp should no longer be used * here. When an error is returned, it has not been linked in, * so it should be free'd. */ nd->nd_repstat = nfsrv_setclient(nd, &clp, &clientid, &confirm, p); if (clp != NULL) { NFSSOCKADDRFREE(clp->lc_req.nr_nam); NFSFREEMUTEX(&clp->lc_req.nr_mtx); free(clp->lc_stateid, M_NFSDCLIENT); free(clp, M_NFSDCLIENT); } if (nd->nd_repstat == 0) { if (confirm.lval[1] != 0) v41flags |= NFSV4EXCH_CONFIRMEDR; NFSM_BUILD(tl, uint32_t *, 2 * NFSX_HYPER + 3 * NFSX_UNSIGNED); *tl++ = clientid.lval[0]; /* ClientID */ *tl++ = clientid.lval[1]; *tl++ = txdr_unsigned(confirm.lval[0]); /* SequenceID */ *tl++ = txdr_unsigned(v41flags); /* Exch flags */ *tl++ = txdr_unsigned(NFSV4EXCH_SP4NONE); /* No SSV */ owner_minor = 0; /* Owner */ txdr_hyper(owner_minor, tl); /* Minor */ (void)nfsm_strtom(nd, nd->nd_cred->cr_prison->pr_hostuuid, strlen(nd->nd_cred->cr_prison->pr_hostuuid)); /* Major */ NFSM_BUILD(tl, uint32_t *, 3 * NFSX_UNSIGNED); *tl++ = txdr_unsigned(NFSX_UNSIGNED); *tl++ = time_uptime; /* Make scope a unique value. */ *tl = txdr_unsigned(1); (void)nfsm_strtom(nd, "freebsd.org", strlen("freebsd.org")); (void)nfsm_strtom(nd, version, strlen(version)); NFSM_BUILD(tl, uint32_t *, NFSX_V4TIME); verstime.tv_sec = 1293840000; /* Jan 1, 2011 */ verstime.tv_nsec = 0; txdr_nfsv4time(&verstime, tl); } NFSEXITCODE2(0, nd); return (0); nfsmout: if (clp != NULL) { NFSSOCKADDRFREE(clp->lc_req.nr_nam); NFSFREEMUTEX(&clp->lc_req.nr_mtx); free(clp->lc_stateid, M_NFSDCLIENT); free(clp, M_NFSDCLIENT); } NFSEXITCODE2(error, nd); return (error); } /* * nfsv4 create session service */ APPLESTATIC int nfsrvd_createsession(struct nfsrv_descript *nd, __unused int isdgram, __unused vnode_t vp, NFSPROC_T *p, __unused struct nfsexstuff *exp) { uint32_t *tl; int error = 0; nfsquad_t clientid, confirm; struct nfsdsession *sep = NULL; uint32_t rdmacnt; if (nfs_rootfhset == 0 || nfsd_checkrootexp(nd) != 0) { nd->nd_repstat = NFSERR_WRONGSEC; goto nfsmout; } sep = (struct nfsdsession *)malloc(sizeof(struct nfsdsession), M_NFSDSESSION, M_WAITOK | M_ZERO); sep->sess_refcnt = 1; mtx_init(&sep->sess_cbsess.nfsess_mtx, "nfscbsession", NULL, MTX_DEF); NFSM_DISSECT(tl, uint32_t *, NFSX_HYPER + 2 * NFSX_UNSIGNED); clientid.lval[0] = *tl++; clientid.lval[1] = *tl++; confirm.lval[0] = fxdr_unsigned(uint32_t, *tl++); sep->sess_crflags = fxdr_unsigned(uint32_t, *tl); /* Persistent sessions and RDMA are not supported. */ sep->sess_crflags &= NFSV4CRSESS_CONNBACKCHAN; /* Fore channel attributes. */ NFSM_DISSECT(tl, uint32_t *, 7 * NFSX_UNSIGNED); tl++; /* Header pad always 0. */ sep->sess_maxreq = fxdr_unsigned(uint32_t, *tl++); sep->sess_maxresp = fxdr_unsigned(uint32_t, *tl++); sep->sess_maxrespcached = fxdr_unsigned(uint32_t, *tl++); sep->sess_maxops = fxdr_unsigned(uint32_t, *tl++); sep->sess_maxslots = fxdr_unsigned(uint32_t, *tl++); if (sep->sess_maxslots > NFSV4_SLOTS) sep->sess_maxslots = NFSV4_SLOTS; rdmacnt = fxdr_unsigned(uint32_t, *tl); if (rdmacnt > 1) { nd->nd_repstat = NFSERR_BADXDR; goto nfsmout; } else if (rdmacnt == 1) NFSM_DISSECT(tl, uint32_t *, NFSX_UNSIGNED); /* Back channel attributes. */ NFSM_DISSECT(tl, uint32_t *, 7 * NFSX_UNSIGNED); tl++; /* Header pad always 0. */ sep->sess_cbmaxreq = fxdr_unsigned(uint32_t, *tl++); sep->sess_cbmaxresp = fxdr_unsigned(uint32_t, *tl++); sep->sess_cbmaxrespcached = fxdr_unsigned(uint32_t, *tl++); sep->sess_cbmaxops = fxdr_unsigned(uint32_t, *tl++); sep->sess_cbsess.nfsess_foreslots = fxdr_unsigned(uint32_t, *tl++); rdmacnt = fxdr_unsigned(uint32_t, *tl); if (rdmacnt > 1) { nd->nd_repstat = NFSERR_BADXDR; goto nfsmout; } else if (rdmacnt == 1) NFSM_DISSECT(tl, uint32_t *, NFSX_UNSIGNED); NFSM_DISSECT(tl, uint32_t *, NFSX_UNSIGNED); sep->sess_cbprogram = fxdr_unsigned(uint32_t, *tl); /* * nfsrv_getclient() searches the client list for a match and * returns the appropriate NFSERR status. */ nd->nd_repstat = nfsrv_getclient(clientid, CLOPS_CONFIRM | CLOPS_RENEW, NULL, sep, confirm, sep->sess_cbprogram, nd, p); if (nd->nd_repstat == 0) { NFSM_BUILD(tl, uint32_t *, NFSX_V4SESSIONID); NFSBCOPY(sep->sess_sessionid, tl, NFSX_V4SESSIONID); NFSM_BUILD(tl, uint32_t *, 18 * NFSX_UNSIGNED); *tl++ = txdr_unsigned(confirm.lval[0]); /* sequenceid */ *tl++ = txdr_unsigned(sep->sess_crflags); /* Fore channel attributes. */ *tl++ = 0; *tl++ = txdr_unsigned(sep->sess_maxreq); *tl++ = txdr_unsigned(sep->sess_maxresp); *tl++ = txdr_unsigned(sep->sess_maxrespcached); *tl++ = txdr_unsigned(sep->sess_maxops); *tl++ = txdr_unsigned(sep->sess_maxslots); *tl++ = txdr_unsigned(1); *tl++ = txdr_unsigned(0); /* No RDMA. */ /* Back channel attributes. */ *tl++ = 0; *tl++ = txdr_unsigned(sep->sess_cbmaxreq); *tl++ = txdr_unsigned(sep->sess_cbmaxresp); *tl++ = txdr_unsigned(sep->sess_cbmaxrespcached); *tl++ = txdr_unsigned(sep->sess_cbmaxops); *tl++ = txdr_unsigned(sep->sess_cbsess.nfsess_foreslots); *tl++ = txdr_unsigned(1); *tl = txdr_unsigned(0); /* No RDMA. */ } nfsmout: if (nd->nd_repstat != 0 && sep != NULL) free(sep, M_NFSDSESSION); NFSEXITCODE2(error, nd); return (error); } /* * nfsv4 sequence service */ APPLESTATIC int nfsrvd_sequence(struct nfsrv_descript *nd, __unused int isdgram, __unused vnode_t vp, __unused NFSPROC_T *p, __unused struct nfsexstuff *exp) { uint32_t *tl; uint32_t highest_slotid, sequenceid, sflags, target_highest_slotid; int cache_this, error = 0; if (nfs_rootfhset == 0 || nfsd_checkrootexp(nd) != 0) { nd->nd_repstat = NFSERR_WRONGSEC; goto nfsmout; } NFSM_DISSECT(tl, uint32_t *, NFSX_V4SESSIONID); NFSBCOPY(tl, nd->nd_sessionid, NFSX_V4SESSIONID); NFSM_DISSECT(tl, uint32_t *, 4 * NFSX_UNSIGNED); sequenceid = fxdr_unsigned(uint32_t, *tl++); nd->nd_slotid = fxdr_unsigned(uint32_t, *tl++); highest_slotid = fxdr_unsigned(uint32_t, *tl++); if (*tl == newnfs_true) cache_this = 1; else cache_this = 0; nd->nd_flag |= ND_HASSEQUENCE; nd->nd_repstat = nfsrv_checksequence(nd, sequenceid, &highest_slotid, &target_highest_slotid, cache_this, &sflags, p); if (nd->nd_repstat == 0) { NFSM_BUILD(tl, uint32_t *, NFSX_V4SESSIONID); NFSBCOPY(nd->nd_sessionid, tl, NFSX_V4SESSIONID); NFSM_BUILD(tl, uint32_t *, 5 * NFSX_UNSIGNED); *tl++ = txdr_unsigned(sequenceid); *tl++ = txdr_unsigned(nd->nd_slotid); *tl++ = txdr_unsigned(highest_slotid); *tl++ = txdr_unsigned(target_highest_slotid); *tl = txdr_unsigned(sflags); } nfsmout: NFSEXITCODE2(error, nd); return (error); } /* * nfsv4 reclaim complete service */ APPLESTATIC int nfsrvd_reclaimcomplete(struct nfsrv_descript *nd, __unused int isdgram, __unused vnode_t vp, __unused NFSPROC_T *p, __unused struct nfsexstuff *exp) { uint32_t *tl; int error = 0; if (nfs_rootfhset == 0 || nfsd_checkrootexp(nd) != 0) { nd->nd_repstat = NFSERR_WRONGSEC; goto nfsmout; } NFSM_DISSECT(tl, uint32_t *, NFSX_UNSIGNED); if (*tl == newnfs_true) nd->nd_repstat = NFSERR_NOTSUPP; else nd->nd_repstat = nfsrv_checkreclaimcomplete(nd); nfsmout: NFSEXITCODE2(error, nd); return (error); } /* * nfsv4 destroy clientid service */ APPLESTATIC int nfsrvd_destroyclientid(struct nfsrv_descript *nd, __unused int isdgram, __unused vnode_t vp, NFSPROC_T *p, __unused struct nfsexstuff *exp) { uint32_t *tl; nfsquad_t clientid; int error = 0; if (nfs_rootfhset == 0 || nfsd_checkrootexp(nd) != 0) { nd->nd_repstat = NFSERR_WRONGSEC; goto nfsmout; } NFSM_DISSECT(tl, uint32_t *, 2 * NFSX_UNSIGNED); clientid.lval[0] = *tl++; clientid.lval[1] = *tl; nd->nd_repstat = nfsrv_destroyclient(clientid, p); nfsmout: NFSEXITCODE2(error, nd); return (error); } /* * nfsv4 destroy session service */ APPLESTATIC int nfsrvd_destroysession(struct nfsrv_descript *nd, __unused int isdgram, __unused vnode_t vp, __unused NFSPROC_T *p, __unused struct nfsexstuff *exp) { uint8_t *cp, sessid[NFSX_V4SESSIONID]; int error = 0; if (nfs_rootfhset == 0 || nfsd_checkrootexp(nd) != 0) { nd->nd_repstat = NFSERR_WRONGSEC; goto nfsmout; } NFSM_DISSECT(cp, uint8_t *, NFSX_V4SESSIONID); NFSBCOPY(cp, sessid, NFSX_V4SESSIONID); nd->nd_repstat = nfsrv_destroysession(nd, sessid); nfsmout: NFSEXITCODE2(error, nd); return (error); } /* * nfsv4 free stateid service */ APPLESTATIC int nfsrvd_freestateid(struct nfsrv_descript *nd, __unused int isdgram, __unused vnode_t vp, NFSPROC_T *p, __unused struct nfsexstuff *exp) { uint32_t *tl; nfsv4stateid_t stateid; int error = 0; if (nfs_rootfhset == 0 || nfsd_checkrootexp(nd) != 0) { nd->nd_repstat = NFSERR_WRONGSEC; goto nfsmout; } NFSM_DISSECT(tl, uint32_t *, NFSX_STATEID); stateid.seqid = fxdr_unsigned(uint32_t, *tl++); NFSBCOPY(tl, stateid.other, NFSX_STATEIDOTHER); nd->nd_repstat = nfsrv_freestateid(nd, &stateid, p); nfsmout: NFSEXITCODE2(error, nd); return (error); } /* * nfsv4 service not supported */ APPLESTATIC int nfsrvd_notsupp(struct nfsrv_descript *nd, __unused int isdgram, __unused vnode_t vp, __unused NFSPROC_T *p, __unused struct nfsexstuff *exp) { nd->nd_repstat = NFSERR_NOTSUPP; NFSEXITCODE2(0, nd); return (0); } Index: projects/clang400-import/sys/fs/nullfs/null_vfsops.c =================================================================== --- projects/clang400-import/sys/fs/nullfs/null_vfsops.c (revision 311459) +++ projects/clang400-import/sys/fs/nullfs/null_vfsops.c (revision 311460) @@ -1,457 +1,461 @@ /*- * Copyright (c) 1992, 1993, 1995 * The Regents of the University of California. All rights reserved. * * This code is derived from software donated to Berkeley by * Jan-Simon Pendry. * * 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. * * @(#)null_vfsops.c 8.2 (Berkeley) 1/21/94 * * @(#)lofs_vfsops.c 1.2 (Berkeley) 6/18/92 * $FreeBSD$ */ /* * Null Layer * (See null_vnops.c for a description of what this does.) */ #include #include #include #include #include #include #include #include #include #include #include #include static MALLOC_DEFINE(M_NULLFSMNT, "nullfs_mount", "NULLFS mount structure"); static vfs_fhtovp_t nullfs_fhtovp; static vfs_mount_t nullfs_mount; static vfs_quotactl_t nullfs_quotactl; static vfs_root_t nullfs_root; static vfs_sync_t nullfs_sync; static vfs_statfs_t nullfs_statfs; static vfs_unmount_t nullfs_unmount; static vfs_vget_t nullfs_vget; static vfs_extattrctl_t nullfs_extattrctl; /* * Mount null layer */ static int nullfs_mount(struct mount *mp) { int error = 0; struct vnode *lowerrootvp, *vp; struct vnode *nullm_rootvp; struct null_mount *xmp; struct thread *td = curthread; char *target; int isvnunlocked = 0, len; struct nameidata nd, *ndp = &nd; NULLFSDEBUG("nullfs_mount(mp = %p)\n", (void *)mp); if (!prison_allow(td->td_ucred, PR_ALLOW_MOUNT_NULLFS)) return (EPERM); if (mp->mnt_flag & MNT_ROOTFS) return (EOPNOTSUPP); /* * Update is a no-op */ if (mp->mnt_flag & MNT_UPDATE) { /* * Only support update mounts for NFS export. */ if (vfs_flagopt(mp->mnt_optnew, "export", NULL, 0)) return (0); else return (EOPNOTSUPP); } /* * Get argument */ error = vfs_getopt(mp->mnt_optnew, "target", (void **)&target, &len); if (error || target[len - 1] != '\0') return (EINVAL); /* * Unlock lower node to avoid possible deadlock. */ if ((mp->mnt_vnodecovered->v_op == &null_vnodeops) && VOP_ISLOCKED(mp->mnt_vnodecovered) == LK_EXCLUSIVE) { VOP_UNLOCK(mp->mnt_vnodecovered, 0); isvnunlocked = 1; } /* * Find lower node */ NDINIT(ndp, LOOKUP, FOLLOW|LOCKLEAF, UIO_SYSSPACE, target, curthread); error = namei(ndp); /* * Re-lock vnode. * XXXKIB This is deadlock-prone as well. */ if (isvnunlocked) vn_lock(mp->mnt_vnodecovered, LK_EXCLUSIVE | LK_RETRY); if (error) return (error); NDFREE(ndp, NDF_ONLY_PNBUF); /* * Sanity check on lower vnode */ lowerrootvp = ndp->ni_vp; /* * Check multi null mount to avoid `lock against myself' panic. */ if (lowerrootvp == VTONULL(mp->mnt_vnodecovered)->null_lowervp) { NULLFSDEBUG("nullfs_mount: multi null mount?\n"); vput(lowerrootvp); return (EDEADLK); } xmp = (struct null_mount *) malloc(sizeof(struct null_mount), M_NULLFSMNT, M_WAITOK | M_ZERO); /* * Save reference to underlying FS */ xmp->nullm_vfs = lowerrootvp->v_mount; /* * Save reference. Each mount also holds * a reference on the root vnode. */ error = null_nodeget(mp, lowerrootvp, &vp); /* * Make sure the node alias worked */ if (error) { free(xmp, M_NULLFSMNT); return (error); } /* * Keep a held reference to the root vnode. * It is vrele'd in nullfs_unmount. */ nullm_rootvp = vp; nullm_rootvp->v_vflag |= VV_ROOT; xmp->nullm_rootvp = nullm_rootvp; /* * Unlock the node (either the lower or the alias) */ VOP_UNLOCK(vp, 0); if (NULLVPTOLOWERVP(nullm_rootvp)->v_mount->mnt_flag & MNT_LOCAL) { MNT_ILOCK(mp); mp->mnt_flag |= MNT_LOCAL; MNT_IUNLOCK(mp); } xmp->nullm_flags |= NULLM_CACHE; if (vfs_getopt(mp->mnt_optnew, "nocache", NULL, NULL) == 0 || (xmp->nullm_vfs->mnt_kern_flag & MNTK_NULL_NOCACHE) != 0) xmp->nullm_flags &= ~NULLM_CACHE; MNT_ILOCK(mp); if ((xmp->nullm_flags & NULLM_CACHE) != 0) { mp->mnt_kern_flag |= lowerrootvp->v_mount->mnt_kern_flag & (MNTK_SHARED_WRITES | MNTK_LOOKUP_SHARED | MNTK_EXTENDED_SHARED); } mp->mnt_kern_flag |= MNTK_LOOKUP_EXCL_DOTDOT; mp->mnt_kern_flag |= lowerrootvp->v_mount->mnt_kern_flag & (MNTK_USES_BCACHE | MNTK_NO_IOPF | MNTK_UNMAPPED_BUFS); MNT_IUNLOCK(mp); mp->mnt_data = xmp; vfs_getnewfsid(mp); if ((xmp->nullm_flags & NULLM_CACHE) != 0) { MNT_ILOCK(xmp->nullm_vfs); TAILQ_INSERT_TAIL(&xmp->nullm_vfs->mnt_uppers, mp, mnt_upper_link); MNT_IUNLOCK(xmp->nullm_vfs); } vfs_mountedfrom(mp, target); NULLFSDEBUG("nullfs_mount: lower %s, alias at %s\n", mp->mnt_stat.f_mntfromname, mp->mnt_stat.f_mntonname); return (0); } /* * Free reference to null layer */ static int nullfs_unmount(mp, mntflags) struct mount *mp; int mntflags; { struct null_mount *mntdata; struct mount *ump; int error, flags; NULLFSDEBUG("nullfs_unmount: mp = %p\n", (void *)mp); if (mntflags & MNT_FORCE) flags = FORCECLOSE; else flags = 0; /* There is 1 extra root vnode reference (nullm_rootvp). */ error = vflush(mp, 1, flags, curthread); if (error) return (error); /* * Finally, throw away the null_mount structure */ mntdata = mp->mnt_data; ump = mntdata->nullm_vfs; if ((mntdata->nullm_flags & NULLM_CACHE) != 0) { MNT_ILOCK(ump); while ((ump->mnt_kern_flag & MNTK_VGONE_UPPER) != 0) { ump->mnt_kern_flag |= MNTK_VGONE_WAITER; msleep(&ump->mnt_uppers, &ump->mnt_mtx, 0, "vgnupw", 0); } TAILQ_REMOVE(&ump->mnt_uppers, mp, mnt_upper_link); MNT_IUNLOCK(ump); } mp->mnt_data = NULL; free(mntdata, M_NULLFSMNT); return (0); } static int nullfs_root(mp, flags, vpp) struct mount *mp; int flags; struct vnode **vpp; { struct vnode *vp; NULLFSDEBUG("nullfs_root(mp = %p, vp = %p->%p)\n", (void *)mp, (void *)MOUNTTONULLMOUNT(mp)->nullm_rootvp, (void *)NULLVPTOLOWERVP(MOUNTTONULLMOUNT(mp)->nullm_rootvp)); /* * Return locked reference to root. */ vp = MOUNTTONULLMOUNT(mp)->nullm_rootvp; VREF(vp); ASSERT_VOP_UNLOCKED(vp, "root vnode is locked"); vn_lock(vp, flags | LK_RETRY); *vpp = vp; return 0; } static int nullfs_quotactl(mp, cmd, uid, arg) struct mount *mp; int cmd; uid_t uid; void *arg; { return VFS_QUOTACTL(MOUNTTONULLMOUNT(mp)->nullm_vfs, cmd, uid, arg); } static int nullfs_statfs(mp, sbp) struct mount *mp; struct statfs *sbp; { int error; - struct statfs mstat; + struct statfs *mstat; NULLFSDEBUG("nullfs_statfs(mp = %p, vp = %p->%p)\n", (void *)mp, (void *)MOUNTTONULLMOUNT(mp)->nullm_rootvp, (void *)NULLVPTOLOWERVP(MOUNTTONULLMOUNT(mp)->nullm_rootvp)); - bzero(&mstat, sizeof(mstat)); + mstat = malloc(sizeof(struct statfs), M_STATFS, M_WAITOK | M_ZERO); - error = VFS_STATFS(MOUNTTONULLMOUNT(mp)->nullm_vfs, &mstat); - if (error) + error = VFS_STATFS(MOUNTTONULLMOUNT(mp)->nullm_vfs, mstat); + if (error) { + free(mstat, M_STATFS); return (error); + } /* now copy across the "interesting" information and fake the rest */ - sbp->f_type = mstat.f_type; + sbp->f_type = mstat->f_type; sbp->f_flags = (sbp->f_flags & (MNT_RDONLY | MNT_NOEXEC | MNT_NOSUID | - MNT_UNION | MNT_NOSYMFOLLOW)) | (mstat.f_flags & ~MNT_ROOTFS); - sbp->f_bsize = mstat.f_bsize; - sbp->f_iosize = mstat.f_iosize; - sbp->f_blocks = mstat.f_blocks; - sbp->f_bfree = mstat.f_bfree; - sbp->f_bavail = mstat.f_bavail; - sbp->f_files = mstat.f_files; - sbp->f_ffree = mstat.f_ffree; + MNT_UNION | MNT_NOSYMFOLLOW)) | (mstat->f_flags & ~MNT_ROOTFS); + sbp->f_bsize = mstat->f_bsize; + sbp->f_iosize = mstat->f_iosize; + sbp->f_blocks = mstat->f_blocks; + sbp->f_bfree = mstat->f_bfree; + sbp->f_bavail = mstat->f_bavail; + sbp->f_files = mstat->f_files; + sbp->f_ffree = mstat->f_ffree; + + free(mstat, M_STATFS); return (0); } static int nullfs_sync(mp, waitfor) struct mount *mp; int waitfor; { /* * XXX - Assumes no data cached at null layer. */ return (0); } static int nullfs_vget(mp, ino, flags, vpp) struct mount *mp; ino_t ino; int flags; struct vnode **vpp; { int error; KASSERT((flags & LK_TYPE_MASK) != 0, ("nullfs_vget: no lock requested")); error = VFS_VGET(MOUNTTONULLMOUNT(mp)->nullm_vfs, ino, flags, vpp); if (error != 0) return (error); return (null_nodeget(mp, *vpp, vpp)); } static int nullfs_fhtovp(mp, fidp, flags, vpp) struct mount *mp; struct fid *fidp; int flags; struct vnode **vpp; { int error; error = VFS_FHTOVP(MOUNTTONULLMOUNT(mp)->nullm_vfs, fidp, flags, vpp); if (error != 0) return (error); return (null_nodeget(mp, *vpp, vpp)); } static int nullfs_extattrctl(mp, cmd, filename_vp, namespace, attrname) struct mount *mp; int cmd; struct vnode *filename_vp; int namespace; const char *attrname; { return (VFS_EXTATTRCTL(MOUNTTONULLMOUNT(mp)->nullm_vfs, cmd, filename_vp, namespace, attrname)); } static void nullfs_reclaim_lowervp(struct mount *mp, struct vnode *lowervp) { struct vnode *vp; vp = null_hashget(mp, lowervp); if (vp == NULL) return; VTONULL(vp)->null_flags |= NULLV_NOUNLOCK; vgone(vp); vput(vp); } static void nullfs_unlink_lowervp(struct mount *mp, struct vnode *lowervp) { struct vnode *vp; struct null_node *xp; vp = null_hashget(mp, lowervp); if (vp == NULL) return; xp = VTONULL(vp); xp->null_flags |= NULLV_DROP | NULLV_NOUNLOCK; vhold(vp); vunref(vp); if (vp->v_usecount == 0) { /* * If vunref() dropped the last use reference on the * nullfs vnode, it must be reclaimed, and its lock * was split from the lower vnode lock. Need to do * extra unlock before allowing the final vdrop() to * free the vnode. */ KASSERT((vp->v_iflag & VI_DOOMED) != 0, ("not reclaimed nullfs vnode %p", vp)); VOP_UNLOCK(vp, 0); } else { /* * Otherwise, the nullfs vnode still shares the lock * with the lower vnode, and must not be unlocked. * Also clear the NULLV_NOUNLOCK, the flag is not * relevant for future reclamations. */ ASSERT_VOP_ELOCKED(vp, "unlink_lowervp"); KASSERT((vp->v_iflag & VI_DOOMED) == 0, ("reclaimed nullfs vnode %p", vp)); xp->null_flags &= ~NULLV_NOUNLOCK; } vdrop(vp); } static struct vfsops null_vfsops = { .vfs_extattrctl = nullfs_extattrctl, .vfs_fhtovp = nullfs_fhtovp, .vfs_init = nullfs_init, .vfs_mount = nullfs_mount, .vfs_quotactl = nullfs_quotactl, .vfs_root = nullfs_root, .vfs_statfs = nullfs_statfs, .vfs_sync = nullfs_sync, .vfs_uninit = nullfs_uninit, .vfs_unmount = nullfs_unmount, .vfs_vget = nullfs_vget, .vfs_reclaim_lowervp = nullfs_reclaim_lowervp, .vfs_unlink_lowervp = nullfs_unlink_lowervp, }; VFS_SET(null_vfsops, nullfs, VFCF_LOOPBACK | VFCF_JAIL); Index: projects/clang400-import/sys/fs/unionfs/union_vfsops.c =================================================================== --- projects/clang400-import/sys/fs/unionfs/union_vfsops.c (revision 311459) +++ projects/clang400-import/sys/fs/unionfs/union_vfsops.c (revision 311460) @@ -1,498 +1,506 @@ /*- * Copyright (c) 1994, 1995 The Regents of the University of California. * Copyright (c) 1994, 1995 Jan-Simon Pendry. * Copyright (c) 2005, 2006, 2012 Masanori Ozawa , ONGS Inc. * Copyright (c) 2006, 2012 Daichi Goto * All rights reserved. * * This code is derived from software donated to Berkeley by * Jan-Simon Pendry. * * 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. * * @(#)union_vfsops.c 8.20 (Berkeley) 5/20/95 * $FreeBSD$ */ #include #include #include #include #include #include #include #include #include #include #include #include #include static MALLOC_DEFINE(M_UNIONFSMNT, "UNIONFS mount", "UNIONFS mount structure"); static vfs_fhtovp_t unionfs_fhtovp; static vfs_checkexp_t unionfs_checkexp; static vfs_mount_t unionfs_domount; static vfs_quotactl_t unionfs_quotactl; static vfs_root_t unionfs_root; static vfs_sync_t unionfs_sync; static vfs_statfs_t unionfs_statfs; static vfs_unmount_t unionfs_unmount; static vfs_vget_t unionfs_vget; static vfs_extattrctl_t unionfs_extattrctl; static struct vfsops unionfs_vfsops; /* * Mount unionfs layer. */ static int unionfs_domount(struct mount *mp) { int error; struct vnode *lowerrootvp; struct vnode *upperrootvp; struct unionfs_mount *ump; struct thread *td; char *target; char *tmp; char *ep; int len; size_t done; int below; uid_t uid; gid_t gid; u_short udir; u_short ufile; unionfs_copymode copymode; unionfs_whitemode whitemode; struct nameidata nd, *ndp; struct vattr va; UNIONFSDEBUG("unionfs_mount(mp = %p)\n", (void *)mp); error = 0; below = 0; uid = 0; gid = 0; udir = 0; ufile = 0; copymode = UNIONFS_TRANSPARENT; /* default */ whitemode = UNIONFS_WHITE_ALWAYS; ndp = &nd; td = curthread; if (mp->mnt_flag & MNT_ROOTFS) { vfs_mount_error(mp, "Cannot union mount root filesystem"); return (EOPNOTSUPP); } /* * Update is a no operation. */ if (mp->mnt_flag & MNT_UPDATE) { vfs_mount_error(mp, "unionfs does not support mount update"); return (EOPNOTSUPP); } /* * Get argument */ error = vfs_getopt(mp->mnt_optnew, "target", (void **)&target, &len); if (error) error = vfs_getopt(mp->mnt_optnew, "from", (void **)&target, &len); if (error || target[len - 1] != '\0') { vfs_mount_error(mp, "Invalid target"); return (EINVAL); } if (vfs_getopt(mp->mnt_optnew, "below", NULL, NULL) == 0) below = 1; if (vfs_getopt(mp->mnt_optnew, "udir", (void **)&tmp, NULL) == 0) { if (tmp != NULL) udir = (mode_t)strtol(tmp, &ep, 8); if (tmp == NULL || *ep) { vfs_mount_error(mp, "Invalid udir"); return (EINVAL); } udir &= S_IRWXU | S_IRWXG | S_IRWXO; } if (vfs_getopt(mp->mnt_optnew, "ufile", (void **)&tmp, NULL) == 0) { if (tmp != NULL) ufile = (mode_t)strtol(tmp, &ep, 8); if (tmp == NULL || *ep) { vfs_mount_error(mp, "Invalid ufile"); return (EINVAL); } ufile &= S_IRWXU | S_IRWXG | S_IRWXO; } /* check umask, uid and gid */ if (udir == 0 && ufile != 0) udir = ufile; if (ufile == 0 && udir != 0) ufile = udir; vn_lock(mp->mnt_vnodecovered, LK_SHARED | LK_RETRY); error = VOP_GETATTR(mp->mnt_vnodecovered, &va, mp->mnt_cred); if (!error) { if (udir == 0) udir = va.va_mode; if (ufile == 0) ufile = va.va_mode; uid = va.va_uid; gid = va.va_gid; } VOP_UNLOCK(mp->mnt_vnodecovered, LK_RELEASE); if (error) return (error); if (mp->mnt_cred->cr_ruid == 0) { /* root only */ if (vfs_getopt(mp->mnt_optnew, "uid", (void **)&tmp, NULL) == 0) { if (tmp != NULL) uid = (uid_t)strtol(tmp, &ep, 10); if (tmp == NULL || *ep) { vfs_mount_error(mp, "Invalid uid"); return (EINVAL); } } if (vfs_getopt(mp->mnt_optnew, "gid", (void **)&tmp, NULL) == 0) { if (tmp != NULL) gid = (gid_t)strtol(tmp, &ep, 10); if (tmp == NULL || *ep) { vfs_mount_error(mp, "Invalid gid"); return (EINVAL); } } if (vfs_getopt(mp->mnt_optnew, "copymode", (void **)&tmp, NULL) == 0) { if (tmp == NULL) { vfs_mount_error(mp, "Invalid copymode"); return (EINVAL); } else if (strcasecmp(tmp, "traditional") == 0) copymode = UNIONFS_TRADITIONAL; else if (strcasecmp(tmp, "transparent") == 0) copymode = UNIONFS_TRANSPARENT; else if (strcasecmp(tmp, "masquerade") == 0) copymode = UNIONFS_MASQUERADE; else { vfs_mount_error(mp, "Invalid copymode"); return (EINVAL); } } if (vfs_getopt(mp->mnt_optnew, "whiteout", (void **)&tmp, NULL) == 0) { if (tmp == NULL) { vfs_mount_error(mp, "Invalid whiteout mode"); return (EINVAL); } else if (strcasecmp(tmp, "always") == 0) whitemode = UNIONFS_WHITE_ALWAYS; else if (strcasecmp(tmp, "whenneeded") == 0) whitemode = UNIONFS_WHITE_WHENNEEDED; else { vfs_mount_error(mp, "Invalid whiteout mode"); return (EINVAL); } } } /* If copymode is UNIONFS_TRADITIONAL, uid/gid is mounted user. */ if (copymode == UNIONFS_TRADITIONAL) { uid = mp->mnt_cred->cr_ruid; gid = mp->mnt_cred->cr_rgid; } UNIONFSDEBUG("unionfs_mount: uid=%d, gid=%d\n", uid, gid); UNIONFSDEBUG("unionfs_mount: udir=0%03o, ufile=0%03o\n", udir, ufile); UNIONFSDEBUG("unionfs_mount: copymode=%d\n", copymode); /* * Find upper node */ NDINIT(ndp, LOOKUP, FOLLOW | LOCKLEAF, UIO_SYSSPACE, target, td); if ((error = namei(ndp))) return (error); NDFREE(ndp, NDF_ONLY_PNBUF); /* get root vnodes */ lowerrootvp = mp->mnt_vnodecovered; upperrootvp = ndp->ni_vp; /* create unionfs_mount */ ump = (struct unionfs_mount *)malloc(sizeof(struct unionfs_mount), M_UNIONFSMNT, M_WAITOK | M_ZERO); /* * Save reference */ if (below) { VOP_UNLOCK(upperrootvp, LK_RELEASE); vn_lock(lowerrootvp, LK_EXCLUSIVE | LK_RETRY); ump->um_lowervp = upperrootvp; ump->um_uppervp = lowerrootvp; } else { ump->um_lowervp = lowerrootvp; ump->um_uppervp = upperrootvp; } ump->um_rootvp = NULLVP; ump->um_uid = uid; ump->um_gid = gid; ump->um_udir = udir; ump->um_ufile = ufile; ump->um_copymode = copymode; ump->um_whitemode = whitemode; mp->mnt_data = ump; /* * Copy upper layer's RDONLY flag. */ mp->mnt_flag |= ump->um_uppervp->v_mount->mnt_flag & MNT_RDONLY; /* * Unlock the node */ VOP_UNLOCK(ump->um_uppervp, LK_RELEASE); /* * Get the unionfs root vnode. */ error = unionfs_nodeget(mp, ump->um_uppervp, ump->um_lowervp, NULLVP, &(ump->um_rootvp), NULL, td); vrele(upperrootvp); if (error) { free(ump, M_UNIONFSMNT); mp->mnt_data = NULL; return (error); } MNT_ILOCK(mp); if ((ump->um_lowervp->v_mount->mnt_flag & MNT_LOCAL) && (ump->um_uppervp->v_mount->mnt_flag & MNT_LOCAL)) mp->mnt_flag |= MNT_LOCAL; MNT_IUNLOCK(mp); /* * Get new fsid */ vfs_getnewfsid(mp); len = MNAMELEN - 1; tmp = mp->mnt_stat.f_mntfromname; copystr((below ? ":" : ":"), tmp, len, &done); len -= done - 1; tmp += done - 1; copystr(target, tmp, len, NULL); UNIONFSDEBUG("unionfs_mount: from %s, on %s\n", mp->mnt_stat.f_mntfromname, mp->mnt_stat.f_mntonname); return (0); } /* * Free reference to unionfs layer */ static int unionfs_unmount(struct mount *mp, int mntflags) { struct unionfs_mount *ump; int error; int num; int freeing; int flags; UNIONFSDEBUG("unionfs_unmount: mp = %p\n", (void *)mp); ump = MOUNTTOUNIONFSMOUNT(mp); flags = 0; if (mntflags & MNT_FORCE) flags |= FORCECLOSE; /* vflush (no need to call vrele) */ for (freeing = 0; (error = vflush(mp, 1, flags, curthread)) != 0;) { num = mp->mnt_nvnodelistsize; if (num == freeing) break; freeing = num; } if (error) return (error); free(ump, M_UNIONFSMNT); mp->mnt_data = NULL; return (0); } static int unionfs_root(struct mount *mp, int flags, struct vnode **vpp) { struct unionfs_mount *ump; struct vnode *vp; ump = MOUNTTOUNIONFSMOUNT(mp); vp = ump->um_rootvp; UNIONFSDEBUG("unionfs_root: rootvp=%p locked=%x\n", vp, VOP_ISLOCKED(vp)); vref(vp); if (flags & LK_TYPE_MASK) vn_lock(vp, flags); *vpp = vp; return (0); } static int unionfs_quotactl(struct mount *mp, int cmd, uid_t uid, void *arg) { struct unionfs_mount *ump; ump = MOUNTTOUNIONFSMOUNT(mp); /* * Writing is always performed to upper vnode. */ return (VFS_QUOTACTL(ump->um_uppervp->v_mount, cmd, uid, arg)); } static int unionfs_statfs(struct mount *mp, struct statfs *sbp) { struct unionfs_mount *ump; int error; - struct statfs mstat; + struct statfs *mstat; uint64_t lbsize; ump = MOUNTTOUNIONFSMOUNT(mp); UNIONFSDEBUG("unionfs_statfs(mp = %p, lvp = %p, uvp = %p)\n", (void *)mp, (void *)ump->um_lowervp, (void *)ump->um_uppervp); - bzero(&mstat, sizeof(mstat)); + mstat = malloc(sizeof(struct statfs), M_STATFS, M_WAITOK | M_ZERO); - error = VFS_STATFS(ump->um_lowervp->v_mount, &mstat); - if (error) + error = VFS_STATFS(ump->um_lowervp->v_mount, mstat); + if (error) { + free(mstat, M_STATFS); return (error); + } /* now copy across the "interesting" information and fake the rest */ - sbp->f_blocks = mstat.f_blocks; - sbp->f_files = mstat.f_files; + sbp->f_blocks = mstat->f_blocks; + sbp->f_files = mstat->f_files; - lbsize = mstat.f_bsize; + lbsize = mstat->f_bsize; - error = VFS_STATFS(ump->um_uppervp->v_mount, &mstat); - if (error) + error = VFS_STATFS(ump->um_uppervp->v_mount, mstat); + if (error) { + free(mstat, M_STATFS); return (error); + } + /* * The FS type etc is copy from upper vfs. * (write able vfs have priority) */ - sbp->f_type = mstat.f_type; - sbp->f_flags = mstat.f_flags; - sbp->f_bsize = mstat.f_bsize; - sbp->f_iosize = mstat.f_iosize; + sbp->f_type = mstat->f_type; + sbp->f_flags = mstat->f_flags; + sbp->f_bsize = mstat->f_bsize; + sbp->f_iosize = mstat->f_iosize; - if (mstat.f_bsize != lbsize) - sbp->f_blocks = ((off_t)sbp->f_blocks * lbsize) / mstat.f_bsize; + if (mstat->f_bsize != lbsize) + sbp->f_blocks = ((off_t)sbp->f_blocks * lbsize) / + mstat->f_bsize; - sbp->f_blocks += mstat.f_blocks; - sbp->f_bfree = mstat.f_bfree; - sbp->f_bavail = mstat.f_bavail; - sbp->f_files += mstat.f_files; - sbp->f_ffree = mstat.f_ffree; + sbp->f_blocks += mstat->f_blocks; + sbp->f_bfree = mstat->f_bfree; + sbp->f_bavail = mstat->f_bavail; + sbp->f_files += mstat->f_files; + sbp->f_ffree = mstat->f_ffree; + + free(mstat, M_STATFS); return (0); } static int unionfs_sync(struct mount *mp, int waitfor) { /* nothing to do */ return (0); } static int unionfs_vget(struct mount *mp, ino_t ino, int flags, struct vnode **vpp) { return (EOPNOTSUPP); } static int unionfs_fhtovp(struct mount *mp, struct fid *fidp, int flags, struct vnode **vpp) { return (EOPNOTSUPP); } static int unionfs_checkexp(struct mount *mp, struct sockaddr *nam, int *extflagsp, struct ucred **credanonp, int *numsecflavors, int **secflavors) { return (EOPNOTSUPP); } static int unionfs_extattrctl(struct mount *mp, int cmd, struct vnode *filename_vp, int namespace, const char *attrname) { struct unionfs_mount *ump; struct unionfs_node *unp; ump = MOUNTTOUNIONFSMOUNT(mp); unp = VTOUNIONFS(filename_vp); if (unp->un_uppervp != NULLVP) { return (VFS_EXTATTRCTL(ump->um_uppervp->v_mount, cmd, unp->un_uppervp, namespace, attrname)); } else { return (VFS_EXTATTRCTL(ump->um_lowervp->v_mount, cmd, unp->un_lowervp, namespace, attrname)); } } static struct vfsops unionfs_vfsops = { .vfs_checkexp = unionfs_checkexp, .vfs_extattrctl = unionfs_extattrctl, .vfs_fhtovp = unionfs_fhtovp, .vfs_init = unionfs_init, .vfs_mount = unionfs_domount, .vfs_quotactl = unionfs_quotactl, .vfs_root = unionfs_root, .vfs_statfs = unionfs_statfs, .vfs_sync = unionfs_sync, .vfs_uninit = unionfs_uninit, .vfs_unmount = unionfs_unmount, .vfs_vget = unionfs_vget, }; VFS_SET(unionfs_vfsops, unionfs, VFCF_LOOPBACK); Index: projects/clang400-import/sys/i386/ibcs2/ibcs2_stat.c =================================================================== --- projects/clang400-import/sys/i386/ibcs2/ibcs2_stat.c (revision 311459) +++ projects/clang400-import/sys/i386/ibcs2/ibcs2_stat.c (revision 311460) @@ -1,238 +1,243 @@ /*- * Copyright (c) 1995 Scott Bartram * Copyright (c) 1995 Steven Wallace * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. The name of the author may not be used to endorse or promote products * derived from this software without specific prior written permission * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include +#include #include #include #include #include #include #include #include #include #include #include #include #include static void bsd_stat2ibcs_stat(struct stat *, struct ibcs2_stat *); static int cvt_statfs(struct statfs *, caddr_t, int); static void bsd_stat2ibcs_stat(st, st4) struct stat *st; struct ibcs2_stat *st4; { bzero(st4, sizeof(*st4)); st4->st_dev = (ibcs2_dev_t)st->st_dev; st4->st_ino = (ibcs2_ino_t)st->st_ino; st4->st_mode = (ibcs2_mode_t)st->st_mode; st4->st_nlink= (ibcs2_nlink_t)st->st_nlink; st4->st_uid = (ibcs2_uid_t)st->st_uid; st4->st_gid = (ibcs2_gid_t)st->st_gid; st4->st_rdev = (ibcs2_dev_t)st->st_rdev; if (st->st_size < (quad_t)1 << 32) st4->st_size = (ibcs2_off_t)st->st_size; else st4->st_size = -2; st4->st_atim = (ibcs2_time_t)st->st_atim.tv_sec; st4->st_mtim = (ibcs2_time_t)st->st_mtim.tv_sec; st4->st_ctim = (ibcs2_time_t)st->st_ctim.tv_sec; } static int cvt_statfs(sp, buf, len) struct statfs *sp; caddr_t buf; int len; { struct ibcs2_statfs ssfs; if (len < 0) return (EINVAL); else if (len > sizeof(ssfs)) len = sizeof(ssfs); bzero(&ssfs, sizeof ssfs); ssfs.f_fstyp = 0; ssfs.f_bsize = sp->f_bsize; ssfs.f_frsize = 0; ssfs.f_blocks = sp->f_blocks; ssfs.f_bfree = sp->f_bfree; ssfs.f_files = sp->f_files; ssfs.f_ffree = sp->f_ffree; ssfs.f_fname[0] = 0; ssfs.f_fpack[0] = 0; return copyout((caddr_t)&ssfs, buf, len); } int ibcs2_statfs(td, uap) struct thread *td; struct ibcs2_statfs_args *uap; { - struct statfs sf; + struct statfs *sf; char *path; int error; CHECKALTEXIST(td, uap->path, &path); - error = kern_statfs(td, path, UIO_SYSSPACE, &sf); + sf = malloc(sizeof(struct statfs), M_STATFS, M_WAITOK); + error = kern_statfs(td, path, UIO_SYSSPACE, sf); free(path, M_TEMP); - if (error) - return (error); - return cvt_statfs(&sf, (caddr_t)uap->buf, uap->len); + if (error == 0) + error = cvt_statfs(sf, (caddr_t)uap->buf, uap->len); + free(sf, M_STATFS); + return (error); } int ibcs2_fstatfs(td, uap) struct thread *td; struct ibcs2_fstatfs_args *uap; { - struct statfs sf; + struct statfs *sf; int error; - error = kern_fstatfs(td, uap->fd, &sf); - if (error) - return (error); - return cvt_statfs(&sf, (caddr_t)uap->buf, uap->len); + sf = malloc(sizeof(struct statfs), M_STATFS, M_WAITOK); + error = kern_fstatfs(td, uap->fd, sf); + if (error == 0) + error = cvt_statfs(sf, (caddr_t)uap->buf, uap->len); + free(sf, M_STATFS); + return (error); } int ibcs2_stat(td, uap) struct thread *td; struct ibcs2_stat_args *uap; { struct ibcs2_stat ibcs2_st; struct stat st; char *path; int error; CHECKALTEXIST(td, uap->path, &path); error = kern_statat(td, 0, AT_FDCWD, path, UIO_SYSSPACE, &st, NULL); free(path, M_TEMP); if (error) return (error); bsd_stat2ibcs_stat(&st, &ibcs2_st); return copyout((caddr_t)&ibcs2_st, (caddr_t)uap->st, ibcs2_stat_len); } int ibcs2_lstat(td, uap) struct thread *td; struct ibcs2_lstat_args *uap; { struct ibcs2_stat ibcs2_st; struct stat st; char *path; int error; CHECKALTEXIST(td, uap->path, &path); error = kern_statat(td, AT_SYMLINK_NOFOLLOW, AT_FDCWD, path, UIO_SYSSPACE, &st, NULL); free(path, M_TEMP); if (error) return (error); bsd_stat2ibcs_stat(&st, &ibcs2_st); return copyout((caddr_t)&ibcs2_st, (caddr_t)uap->st, ibcs2_stat_len); } int ibcs2_fstat(td, uap) struct thread *td; struct ibcs2_fstat_args *uap; { struct ibcs2_stat ibcs2_st; struct stat st; int error; error = kern_fstat(td, uap->fd, &st); if (error) return (error); bsd_stat2ibcs_stat(&st, &ibcs2_st); return copyout((caddr_t)&ibcs2_st, (caddr_t)uap->st, ibcs2_stat_len); } int ibcs2_utssys(td, uap) struct thread *td; struct ibcs2_utssys_args *uap; { switch (uap->flag) { case 0: /* uname(2) */ { char machine_name[9], *p; struct ibcs2_utsname sut; bzero(&sut, ibcs2_utsname_len); strncpy(sut.sysname, IBCS2_UNAME_SYSNAME, sizeof(sut.sysname) - 1); strncpy(sut.release, IBCS2_UNAME_RELEASE, sizeof(sut.release) - 1); strncpy(sut.version, IBCS2_UNAME_VERSION, sizeof(sut.version) - 1); getcredhostname(td->td_ucred, machine_name, sizeof(machine_name) - 1); p = strchr(machine_name, '.'); if ( p ) *p = '\0'; strncpy(sut.nodename, machine_name, sizeof(sut.nodename) - 1); strncpy(sut.machine, machine, sizeof(sut.machine) - 1); DPRINTF(("IBCS2 uname: sys=%s rel=%s ver=%s node=%s mach=%s\n", sut.sysname, sut.release, sut.version, sut.nodename, sut.machine)); return copyout((caddr_t)&sut, (caddr_t)uap->a1, ibcs2_utsname_len); } case 2: /* ustat(2) */ { return ENOSYS; /* XXX - TODO */ } default: return ENOSYS; } } Index: projects/clang400-import/sys/kern/kern_acct.c =================================================================== --- projects/clang400-import/sys/kern/kern_acct.c (revision 311459) +++ projects/clang400-import/sys/kern/kern_acct.c (revision 311460) @@ -1,647 +1,652 @@ /*- * Copyright (c) 1982, 1986, 1989, 1993 * The Regents of the University of California. All rights reserved. * (c) UNIX System Laboratories, Inc. * Copyright (c) 2005 Robert N. M. Watson * All rights reserved. * * All or some portions of this file are derived from material licensed * to the University of California by American Telephone and Telegraph * Co. or Unix System Laboratories, Inc. and are reproduced herein with * the permission of UNIX System Laboratories, Inc. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 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. * * Copyright (c) 1994 Christopher G. Demetriou * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. All advertising materials mentioning features or use of this software * must display the following acknowledgement: * This product includes software developed by the University of * California, Berkeley and its contributors. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * @(#)kern_acct.c 8.1 (Berkeley) 6/14/93 */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include +#include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include /* * The routines implemented in this file are described in: * Leffler, et al.: The Design and Implementation of the 4.3BSD * UNIX Operating System (Addison Welley, 1989) * on pages 62-63. * On May 2007 the historic 3 bits base 8 exponent, 13 bit fraction * compt_t representation described in the above reference was replaced * with that of IEEE-754 floats. * * Arguably, to simplify accounting operations, this mechanism should * be replaced by one in which an accounting log file (similar to /dev/klog) * is read by a user process, etc. However, that has its own problems. */ /* Floating point definitions from . */ #define FLT_MANT_DIG 24 /* p */ #define FLT_MAX_EXP 128 /* emax */ /* * Internal accounting functions. * The former's operation is described in Leffler, et al., and the latter * was provided by UCB with the 4.4BSD-Lite release */ static uint32_t encode_timeval(struct timeval); static uint32_t encode_long(long); static void acctwatch(void); static void acct_thread(void *); static int acct_disable(struct thread *, int); /* * Accounting vnode pointer, saved vnode pointer, and flags for each. * acct_sx protects against changes to the active vnode and credentials * while accounting records are being committed to disk. */ static int acct_configured; static int acct_suspended; static struct vnode *acct_vp; static struct ucred *acct_cred; static struct plimit *acct_limit; static int acct_flags; static struct sx acct_sx; SX_SYSINIT(acct, &acct_sx, "acct_sx"); /* * State of the accounting kthread. */ static int acct_state; #define ACCT_RUNNING 1 /* Accounting kthread is running. */ #define ACCT_EXITREQ 2 /* Accounting kthread should exit. */ /* * Values associated with enabling and disabling accounting */ static int acctsuspend = 2; /* stop accounting when < 2% free space left */ SYSCTL_INT(_kern, OID_AUTO, acct_suspend, CTLFLAG_RW, &acctsuspend, 0, "percentage of free disk space below which accounting stops"); static int acctresume = 4; /* resume when free space risen to > 4% */ SYSCTL_INT(_kern, OID_AUTO, acct_resume, CTLFLAG_RW, &acctresume, 0, "percentage of free disk space above which accounting resumes"); static int acctchkfreq = 15; /* frequency (in seconds) to check space */ static int sysctl_acct_chkfreq(SYSCTL_HANDLER_ARGS) { int error, value; /* Write out the old value. */ error = SYSCTL_OUT(req, &acctchkfreq, sizeof(int)); if (error || req->newptr == NULL) return (error); /* Read in and verify the new value. */ error = SYSCTL_IN(req, &value, sizeof(int)); if (error) return (error); if (value <= 0) return (EINVAL); acctchkfreq = value; return (0); } SYSCTL_PROC(_kern, OID_AUTO, acct_chkfreq, CTLTYPE_INT|CTLFLAG_RW, &acctchkfreq, 0, sysctl_acct_chkfreq, "I", "frequency for checking the free space"); SYSCTL_INT(_kern, OID_AUTO, acct_configured, CTLFLAG_RD, &acct_configured, 0, "Accounting configured or not"); SYSCTL_INT(_kern, OID_AUTO, acct_suspended, CTLFLAG_RD, &acct_suspended, 0, "Accounting suspended or not"); /* * Accounting system call. Written based on the specification and previous * implementation done by Mark Tinguely. */ int sys_acct(struct thread *td, struct acct_args *uap) { struct nameidata nd; int error, flags, i, replacing; error = priv_check(td, PRIV_ACCT); if (error) return (error); /* * If accounting is to be started to a file, open that file for * appending and make sure it's a 'normal'. */ if (uap->path != NULL) { NDINIT(&nd, LOOKUP, NOFOLLOW | AUDITVNODE1, UIO_USERSPACE, uap->path, td); flags = FWRITE | O_APPEND; error = vn_open(&nd, &flags, 0, NULL); if (error) return (error); NDFREE(&nd, NDF_ONLY_PNBUF); #ifdef MAC error = mac_system_check_acct(td->td_ucred, nd.ni_vp); if (error) { VOP_UNLOCK(nd.ni_vp, 0); vn_close(nd.ni_vp, flags, td->td_ucred, td); return (error); } #endif VOP_UNLOCK(nd.ni_vp, 0); if (nd.ni_vp->v_type != VREG) { vn_close(nd.ni_vp, flags, td->td_ucred, td); return (EACCES); } #ifdef MAC } else { error = mac_system_check_acct(td->td_ucred, NULL); if (error) return (error); #endif } /* * Disallow concurrent access to the accounting vnode while we swap * it out, in order to prevent access after close. */ sx_xlock(&acct_sx); /* * Don't log spurious disable/enable messages if we are * switching from one accounting file to another due to log * rotation. */ replacing = (acct_vp != NULL && uap->path != NULL); /* * If accounting was previously enabled, kill the old space-watcher, * close the file, and (if no new file was specified, leave). Reset * the suspended state regardless of whether accounting remains * enabled. */ acct_suspended = 0; if (acct_vp != NULL) error = acct_disable(td, !replacing); if (uap->path == NULL) { if (acct_state & ACCT_RUNNING) { acct_state |= ACCT_EXITREQ; wakeup(&acct_state); } sx_xunlock(&acct_sx); return (error); } /* * Create our own plimit object without limits. It will be assigned * to exiting processes. */ acct_limit = lim_alloc(); for (i = 0; i < RLIM_NLIMITS; i++) acct_limit->pl_rlimit[i].rlim_cur = acct_limit->pl_rlimit[i].rlim_max = RLIM_INFINITY; /* * Save the new accounting file vnode, and schedule the new * free space watcher. */ acct_vp = nd.ni_vp; acct_cred = crhold(td->td_ucred); acct_flags = flags; if (acct_state & ACCT_RUNNING) acct_state &= ~ACCT_EXITREQ; else { /* * Try to start up an accounting kthread. We may start more * than one, but if so the extras will commit suicide as * soon as they start up. */ error = kproc_create(acct_thread, NULL, NULL, 0, 0, "accounting"); if (error) { (void) acct_disable(td, 0); sx_xunlock(&acct_sx); log(LOG_NOTICE, "Unable to start accounting thread\n"); return (error); } } acct_configured = 1; sx_xunlock(&acct_sx); if (!replacing) log(LOG_NOTICE, "Accounting enabled\n"); return (error); } /* * Disable currently in-progress accounting by closing the vnode, dropping * our reference to the credential, and clearing the vnode's flags. */ static int acct_disable(struct thread *td, int logging) { int error; sx_assert(&acct_sx, SX_XLOCKED); error = vn_close(acct_vp, acct_flags, acct_cred, td); crfree(acct_cred); lim_free(acct_limit); acct_configured = 0; acct_vp = NULL; acct_cred = NULL; acct_flags = 0; if (logging) log(LOG_NOTICE, "Accounting disabled\n"); return (error); } /* * Write out process accounting information, on process exit. * Data to be written out is specified in Leffler, et al. * and are enumerated below. (They're also noted in the system * "acct.h" header file.) */ int acct_process(struct thread *td) { struct acctv2 acct; struct timeval ut, st, tmp; struct plimit *oldlim; struct proc *p; struct rusage ru; int t, ret; /* * Lockless check of accounting condition before doing the hard * work. */ if (acct_vp == NULL || acct_suspended) return (0); sx_slock(&acct_sx); /* * If accounting isn't enabled, don't bother. Have to check again * once we own the lock in case we raced with disabling of accounting * by another thread. */ if (acct_vp == NULL || acct_suspended) { sx_sunlock(&acct_sx); return (0); } p = td->td_proc; /* * Get process accounting information. */ sx_slock(&proctree_lock); PROC_LOCK(p); /* (1) The terminal from which the process was started */ if ((p->p_flag & P_CONTROLT) && p->p_pgrp->pg_session->s_ttyp) acct.ac_tty = tty_udev(p->p_pgrp->pg_session->s_ttyp); else acct.ac_tty = NODEV; sx_sunlock(&proctree_lock); /* (2) The name of the command that ran */ bcopy(p->p_comm, acct.ac_comm, sizeof acct.ac_comm); /* (3) The amount of user and system time that was used */ rufetchcalc(p, &ru, &ut, &st); acct.ac_utime = encode_timeval(ut); acct.ac_stime = encode_timeval(st); /* (4) The elapsed time the command ran (and its starting time) */ getboottime(&tmp); timevaladd(&tmp, &p->p_stats->p_start); acct.ac_btime = tmp.tv_sec; microuptime(&tmp); timevalsub(&tmp, &p->p_stats->p_start); acct.ac_etime = encode_timeval(tmp); /* (5) The average amount of memory used */ tmp = ut; timevaladd(&tmp, &st); /* Convert tmp (i.e. u + s) into hz units to match ru_i*. */ t = tmp.tv_sec * hz + tmp.tv_usec / tick; if (t) acct.ac_mem = encode_long((ru.ru_ixrss + ru.ru_idrss + + ru.ru_isrss) / t); else acct.ac_mem = 0; /* (6) The number of disk I/O operations done */ acct.ac_io = encode_long(ru.ru_inblock + ru.ru_oublock); /* (7) The UID and GID of the process */ acct.ac_uid = p->p_ucred->cr_ruid; acct.ac_gid = p->p_ucred->cr_rgid; /* (8) The boolean flags that tell how the process terminated, etc. */ acct.ac_flagx = p->p_acflag; /* Setup ancillary structure fields. */ acct.ac_flagx |= ANVER; acct.ac_zero = 0; acct.ac_version = 2; acct.ac_len = acct.ac_len2 = sizeof(acct); /* * Eliminate rlimits (file size limit in particular). */ oldlim = p->p_limit; p->p_limit = lim_hold(acct_limit); PROC_UNLOCK(p); lim_free(oldlim); /* * Write the accounting information to the file. */ ret = vn_rdwr(UIO_WRITE, acct_vp, (caddr_t)&acct, sizeof (acct), (off_t)0, UIO_SYSSPACE, IO_APPEND|IO_UNIT, acct_cred, NOCRED, NULL, td); sx_sunlock(&acct_sx); return (ret); } /* FLOAT_CONVERSION_START (Regression testing; don't remove this line.) */ /* Convert timevals and longs into IEEE-754 bit patterns. */ /* Mantissa mask (MSB is implied, so subtract 1). */ #define MANT_MASK ((1 << (FLT_MANT_DIG - 1)) - 1) /* * We calculate integer values to a precision of approximately * 28 bits. * This is high-enough precision to fill the 24 float bits * and low-enough to avoid overflowing the 32 int bits. */ #define CALC_BITS 28 /* log_2(1000000). */ #define LOG2_1M 20 /* * Convert the elements of a timeval into a 32-bit word holding * the bits of a IEEE-754 float. * The float value represents the timeval's value in microsecond units. */ static uint32_t encode_timeval(struct timeval tv) { int log2_s; int val, exp; /* Unnormalized value and exponent */ int norm_exp; /* Normalized exponent */ int shift; /* * First calculate value and exponent to about CALC_BITS precision. * Note that the following conditionals have been ordered so that * the most common cases appear first. */ if (tv.tv_sec == 0) { if (tv.tv_usec == 0) return (0); exp = 0; val = tv.tv_usec; } else { /* * Calculate the value to a precision of approximately * CALC_BITS. */ log2_s = fls(tv.tv_sec) - 1; if (log2_s + LOG2_1M < CALC_BITS) { exp = 0; val = 1000000 * tv.tv_sec + tv.tv_usec; } else { exp = log2_s + LOG2_1M - CALC_BITS; val = (unsigned int)(((uint64_t)1000000 * tv.tv_sec + tv.tv_usec) >> exp); } } /* Now normalize and pack the value into an IEEE-754 float. */ norm_exp = fls(val) - 1; shift = FLT_MANT_DIG - norm_exp - 1; #ifdef ACCT_DEBUG printf("val=%d exp=%d shift=%d log2(val)=%d\n", val, exp, shift, norm_exp); printf("exp=%x mant=%x\n", FLT_MAX_EXP - 1 + exp + norm_exp, ((shift > 0 ? (val << shift) : (val >> -shift)) & MANT_MASK)); #endif return (((FLT_MAX_EXP - 1 + exp + norm_exp) << (FLT_MANT_DIG - 1)) | ((shift > 0 ? val << shift : val >> -shift) & MANT_MASK)); } /* * Convert a non-negative long value into the bit pattern of * an IEEE-754 float value. */ static uint32_t encode_long(long val) { int norm_exp; /* Normalized exponent */ int shift; if (val == 0) return (0); if (val < 0) { log(LOG_NOTICE, "encode_long: negative value %ld in accounting record\n", val); val = LONG_MAX; } norm_exp = fls(val) - 1; shift = FLT_MANT_DIG - norm_exp - 1; #ifdef ACCT_DEBUG printf("val=%d shift=%d log2(val)=%d\n", val, shift, norm_exp); printf("exp=%x mant=%x\n", FLT_MAX_EXP - 1 + exp + norm_exp, ((shift > 0 ? (val << shift) : (val >> -shift)) & MANT_MASK)); #endif return (((FLT_MAX_EXP - 1 + norm_exp) << (FLT_MANT_DIG - 1)) | ((shift > 0 ? val << shift : val >> -shift) & MANT_MASK)); } /* FLOAT_CONVERSION_END (Regression testing; don't remove this line.) */ /* * Periodically check the filesystem to see if accounting * should be turned on or off. Beware the case where the vnode * has been vgone()'d out from underneath us, e.g. when the file * system containing the accounting file has been forcibly unmounted. */ /* ARGSUSED */ static void acctwatch(void) { - struct statfs sb; + struct statfs *sp; sx_assert(&acct_sx, SX_XLOCKED); /* * If accounting was disabled before our kthread was scheduled, * then acct_vp might be NULL. If so, just ask our kthread to * exit and return. */ if (acct_vp == NULL) { acct_state |= ACCT_EXITREQ; return; } /* * If our vnode is no longer valid, tear it down and signal the * accounting thread to die. */ if (acct_vp->v_type == VBAD) { (void) acct_disable(NULL, 1); acct_state |= ACCT_EXITREQ; return; } /* * Stopping here is better than continuing, maybe it will be VBAD * next time around. */ - if (VFS_STATFS(acct_vp->v_mount, &sb) < 0) + sp = malloc(sizeof(struct statfs), M_STATFS, M_WAITOK); + if (VFS_STATFS(acct_vp->v_mount, sp) < 0) { + free(sp, M_STATFS); return; + } if (acct_suspended) { - if (sb.f_bavail > (int64_t)(acctresume * sb.f_blocks / + if (sp->f_bavail > (int64_t)(acctresume * sp->f_blocks / 100)) { acct_suspended = 0; log(LOG_NOTICE, "Accounting resumed\n"); } } else { - if (sb.f_bavail <= (int64_t)(acctsuspend * sb.f_blocks / + if (sp->f_bavail <= (int64_t)(acctsuspend * sp->f_blocks / 100)) { acct_suspended = 1; log(LOG_NOTICE, "Accounting suspended\n"); } } + free(sp, M_STATFS); } /* * The main loop for the dedicated kernel thread that periodically calls * acctwatch(). */ static void acct_thread(void *dummy) { u_char pri; /* This is a low-priority kernel thread. */ pri = PRI_MAX_KERN; thread_lock(curthread); sched_prio(curthread, pri); thread_unlock(curthread); /* If another accounting kthread is already running, just die. */ sx_xlock(&acct_sx); if (acct_state & ACCT_RUNNING) { sx_xunlock(&acct_sx); kproc_exit(0); } acct_state |= ACCT_RUNNING; /* Loop until we are asked to exit. */ while (!(acct_state & ACCT_EXITREQ)) { /* Perform our periodic checks. */ acctwatch(); /* * We check this flag again before sleeping since the * acctwatch() might have shut down accounting and asked us * to exit. */ if (!(acct_state & ACCT_EXITREQ)) { sx_sleep(&acct_state, &acct_sx, 0, "-", acctchkfreq * hz); } } /* * Acknowledge the exit request and shutdown. We clear both the * exit request and running flags. */ acct_state = 0; sx_xunlock(&acct_sx); kproc_exit(0); } Index: projects/clang400-import/sys/kern/kern_exec.c =================================================================== --- projects/clang400-import/sys/kern/kern_exec.c (revision 311459) +++ projects/clang400-import/sys/kern/kern_exec.c (revision 311460) @@ -1,1633 +1,1697 @@ /*- * Copyright (c) 1993, David Greenman * 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_capsicum.h" #include "opt_hwpmc_hooks.h" #include "opt_ktrace.h" #include "opt_vm.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 #ifdef KTRACE #include #endif #include #include #include #include #include #include #include #include #include #ifdef HWPMC_HOOKS #include #endif #include #include #include #ifdef KDTRACE_HOOKS #include dtrace_execexit_func_t dtrace_fasttrap_exec; #endif SDT_PROVIDER_DECLARE(proc); SDT_PROBE_DEFINE1(proc, , , exec, "char *"); SDT_PROBE_DEFINE1(proc, , , exec__failure, "int"); SDT_PROBE_DEFINE1(proc, , , exec__success, "char *"); MALLOC_DEFINE(M_PARGS, "proc-args", "Process arguments"); int coredump_pack_fileinfo = 1; SYSCTL_INT(_kern, OID_AUTO, coredump_pack_fileinfo, CTLFLAG_RWTUN, &coredump_pack_fileinfo, 0, "Enable file path packing in 'procstat -f' coredump notes"); int coredump_pack_vmmapinfo = 1; SYSCTL_INT(_kern, OID_AUTO, coredump_pack_vmmapinfo, CTLFLAG_RWTUN, &coredump_pack_vmmapinfo, 0, "Enable file path packing in 'procstat -v' coredump notes"); static int sysctl_kern_ps_strings(SYSCTL_HANDLER_ARGS); static int sysctl_kern_usrstack(SYSCTL_HANDLER_ARGS); static int sysctl_kern_stackprot(SYSCTL_HANDLER_ARGS); static int do_execve(struct thread *td, struct image_args *args, struct mac *mac_p); /* XXX This should be vm_size_t. */ SYSCTL_PROC(_kern, KERN_PS_STRINGS, ps_strings, CTLTYPE_ULONG|CTLFLAG_RD| CTLFLAG_MPSAFE, NULL, 0, sysctl_kern_ps_strings, "LU", ""); /* XXX This should be vm_size_t. */ SYSCTL_PROC(_kern, KERN_USRSTACK, usrstack, CTLTYPE_ULONG|CTLFLAG_RD| CTLFLAG_CAPRD|CTLFLAG_MPSAFE, NULL, 0, sysctl_kern_usrstack, "LU", ""); SYSCTL_PROC(_kern, OID_AUTO, stackprot, CTLTYPE_INT|CTLFLAG_RD|CTLFLAG_MPSAFE, NULL, 0, sysctl_kern_stackprot, "I", ""); u_long ps_arg_cache_limit = PAGE_SIZE / 16; SYSCTL_ULONG(_kern, OID_AUTO, ps_arg_cache_limit, CTLFLAG_RW, &ps_arg_cache_limit, 0, ""); static int disallow_high_osrel; SYSCTL_INT(_kern, OID_AUTO, disallow_high_osrel, CTLFLAG_RW, &disallow_high_osrel, 0, "Disallow execution of binaries built for higher version of the world"); static int map_at_zero = 0; SYSCTL_INT(_security_bsd, OID_AUTO, map_at_zero, CTLFLAG_RWTUN, &map_at_zero, 0, "Permit processes to map an object at virtual address 0."); static int sysctl_kern_ps_strings(SYSCTL_HANDLER_ARGS) { struct proc *p; int error; p = curproc; #ifdef SCTL_MASK32 if (req->flags & SCTL_MASK32) { unsigned int val; val = (unsigned int)p->p_sysent->sv_psstrings; error = SYSCTL_OUT(req, &val, sizeof(val)); } else #endif error = SYSCTL_OUT(req, &p->p_sysent->sv_psstrings, sizeof(p->p_sysent->sv_psstrings)); return error; } static int sysctl_kern_usrstack(SYSCTL_HANDLER_ARGS) { struct proc *p; int error; p = curproc; #ifdef SCTL_MASK32 if (req->flags & SCTL_MASK32) { unsigned int val; val = (unsigned int)p->p_sysent->sv_usrstack; error = SYSCTL_OUT(req, &val, sizeof(val)); } else #endif error = SYSCTL_OUT(req, &p->p_sysent->sv_usrstack, sizeof(p->p_sysent->sv_usrstack)); return error; } static int sysctl_kern_stackprot(SYSCTL_HANDLER_ARGS) { struct proc *p; p = curproc; return (SYSCTL_OUT(req, &p->p_sysent->sv_stackprot, sizeof(p->p_sysent->sv_stackprot))); } /* * Each of the items is a pointer to a `const struct execsw', hence the * double pointer here. */ static const struct execsw **execsw; #ifndef _SYS_SYSPROTO_H_ struct execve_args { char *fname; char **argv; char **envv; }; #endif int sys_execve(struct thread *td, struct execve_args *uap) { struct image_args args; struct vmspace *oldvmspace; int error; error = pre_execve(td, &oldvmspace); if (error != 0) return (error); error = exec_copyin_args(&args, uap->fname, UIO_USERSPACE, uap->argv, uap->envv); if (error == 0) error = kern_execve(td, &args, NULL); post_execve(td, error, oldvmspace); return (error); } #ifndef _SYS_SYSPROTO_H_ struct fexecve_args { int fd; char **argv; char **envv; } #endif int sys_fexecve(struct thread *td, struct fexecve_args *uap) { struct image_args args; struct vmspace *oldvmspace; int error; error = pre_execve(td, &oldvmspace); if (error != 0) return (error); error = exec_copyin_args(&args, NULL, UIO_SYSSPACE, uap->argv, uap->envv); if (error == 0) { args.fd = uap->fd; error = kern_execve(td, &args, NULL); } post_execve(td, error, oldvmspace); return (error); } #ifndef _SYS_SYSPROTO_H_ struct __mac_execve_args { char *fname; char **argv; char **envv; struct mac *mac_p; }; #endif int sys___mac_execve(struct thread *td, struct __mac_execve_args *uap) { #ifdef MAC struct image_args args; struct vmspace *oldvmspace; int error; error = pre_execve(td, &oldvmspace); if (error != 0) return (error); error = exec_copyin_args(&args, uap->fname, UIO_USERSPACE, uap->argv, uap->envv); if (error == 0) error = kern_execve(td, &args, uap->mac_p); post_execve(td, error, oldvmspace); return (error); #else return (ENOSYS); #endif } int pre_execve(struct thread *td, struct vmspace **oldvmspace) { struct proc *p; int error; KASSERT(td == curthread, ("non-current thread %p", td)); error = 0; p = td->td_proc; if ((p->p_flag & P_HADTHREADS) != 0) { PROC_LOCK(p); if (thread_single(p, SINGLE_BOUNDARY) != 0) error = ERESTART; PROC_UNLOCK(p); } KASSERT(error != 0 || (td->td_pflags & TDP_EXECVMSPC) == 0, ("nested execve")); *oldvmspace = p->p_vmspace; return (error); } void post_execve(struct thread *td, int error, struct vmspace *oldvmspace) { struct proc *p; KASSERT(td == curthread, ("non-current thread %p", td)); p = td->td_proc; if ((p->p_flag & P_HADTHREADS) != 0) { PROC_LOCK(p); /* * If success, we upgrade to SINGLE_EXIT state to * force other threads to suicide. */ if (error == 0) thread_single(p, SINGLE_EXIT); else thread_single_end(p, SINGLE_BOUNDARY); PROC_UNLOCK(p); } if ((td->td_pflags & TDP_EXECVMSPC) != 0) { KASSERT(p->p_vmspace != oldvmspace, ("oldvmspace still used")); vmspace_free(oldvmspace); td->td_pflags &= ~TDP_EXECVMSPC; } } /* * XXX: kern_execve has the astonishing property of not always returning to * the caller. If sufficiently bad things happen during the call to * do_execve(), it can end up calling exit1(); as a result, callers must * avoid doing anything which they might need to undo (e.g., allocating * memory). */ int kern_execve(struct thread *td, struct image_args *args, struct mac *mac_p) { AUDIT_ARG_ARGV(args->begin_argv, args->argc, args->begin_envv - args->begin_argv); AUDIT_ARG_ENVV(args->begin_envv, args->envc, args->endp - args->begin_envv); return (do_execve(td, args, mac_p)); } /* * In-kernel implementation of execve(). All arguments are assumed to be * userspace pointers from the passed thread. */ static int do_execve(td, args, mac_p) struct thread *td; struct image_args *args; struct mac *mac_p; { struct proc *p = td->td_proc; struct nameidata nd; struct ucred *oldcred; struct uidinfo *euip = NULL; register_t *stack_base; int error, i; struct image_params image_params, *imgp; struct vattr attr; int (*img_first)(struct image_params *); struct pargs *oldargs = NULL, *newargs = NULL; struct sigacts *oldsigacts = NULL, *newsigacts = NULL; #ifdef KTRACE struct vnode *tracevp = NULL; struct ucred *tracecred = NULL; #endif struct vnode *oldtextvp = NULL, *newtextvp; cap_rights_t rights; int credential_changing; int textset; #ifdef MAC struct label *interpvplabel = NULL; int will_transition; #endif #ifdef HWPMC_HOOKS struct pmckern_procexec pe; #endif static const char fexecv_proc_title[] = "(fexecv)"; imgp = &image_params; /* * Lock the process and set the P_INEXEC flag to indicate that * it should be left alone until we're done here. This is * necessary to avoid race conditions - e.g. in ptrace() - * that might allow a local user to illicitly obtain elevated * privileges. */ PROC_LOCK(p); KASSERT((p->p_flag & P_INEXEC) == 0, ("%s(): process already has P_INEXEC flag", __func__)); p->p_flag |= P_INEXEC; PROC_UNLOCK(p); /* * Initialize part of the common data */ bzero(imgp, sizeof(*imgp)); imgp->proc = p; imgp->attr = &attr; imgp->args = args; oldcred = p->p_ucred; #ifdef MAC error = mac_execve_enter(imgp, mac_p); if (error) goto exec_fail; #endif /* * Translate the file name. namei() returns a vnode pointer * in ni_vp among other things. * * XXXAUDIT: It would be desirable to also audit the name of the * interpreter if this is an interpreted binary. */ if (args->fname != NULL) { NDINIT(&nd, LOOKUP, ISOPEN | LOCKLEAF | FOLLOW | SAVENAME | AUDITVNODE1, UIO_SYSSPACE, args->fname, td); } SDT_PROBE1(proc, , , exec, args->fname); interpret: if (args->fname != NULL) { #ifdef CAPABILITY_MODE /* * While capability mode can't reach this point via direct * path arguments to execve(), we also don't allow * interpreters to be used in capability mode (for now). * Catch indirect lookups and return a permissions error. */ if (IN_CAPABILITY_MODE(td)) { error = ECAPMODE; goto exec_fail; } #endif error = namei(&nd); if (error) goto exec_fail; newtextvp = nd.ni_vp; imgp->vp = newtextvp; } else { AUDIT_ARG_FD(args->fd); /* * Descriptors opened only with O_EXEC or O_RDONLY are allowed. */ error = fgetvp_exec(td, args->fd, cap_rights_init(&rights, CAP_FEXECVE), &newtextvp); if (error) goto exec_fail; vn_lock(newtextvp, LK_EXCLUSIVE | LK_RETRY); AUDIT_ARG_VNODE1(newtextvp); imgp->vp = newtextvp; } /* * Check file permissions (also 'opens' file) */ error = exec_check_permissions(imgp); if (error) goto exec_fail_dealloc; imgp->object = imgp->vp->v_object; if (imgp->object != NULL) vm_object_reference(imgp->object); /* * Set VV_TEXT now so no one can write to the executable while we're * activating it. * * Remember if this was set before and unset it in case this is not * actually an executable image. */ textset = VOP_IS_TEXT(imgp->vp); VOP_SET_TEXT(imgp->vp); error = exec_map_first_page(imgp); if (error) goto exec_fail_dealloc; imgp->proc->p_osrel = 0; /* * Implement image setuid/setgid. * * Determine new credentials before attempting image activators * so that it can be used by process_exec handlers to determine * credential/setid changes. * * Don't honor setuid/setgid if the filesystem prohibits it or if * the process is being traced. * * We disable setuid/setgid/etc in capability mode on the basis * that most setugid applications are not written with that * environment in mind, and will therefore almost certainly operate * incorrectly. In principle there's no reason that setugid * applications might not be useful in capability mode, so we may want * to reconsider this conservative design choice in the future. * * XXXMAC: For the time being, use NOSUID to also prohibit * transitions on the file system. */ credential_changing = 0; credential_changing |= (attr.va_mode & S_ISUID) && oldcred->cr_uid != attr.va_uid; credential_changing |= (attr.va_mode & S_ISGID) && oldcred->cr_gid != attr.va_gid; #ifdef MAC will_transition = mac_vnode_execve_will_transition(oldcred, imgp->vp, interpvplabel, imgp); credential_changing |= will_transition; #endif if (credential_changing && #ifdef CAPABILITY_MODE ((oldcred->cr_flags & CRED_FLAG_CAPMODE) == 0) && #endif (imgp->vp->v_mount->mnt_flag & MNT_NOSUID) == 0 && (p->p_flag & P_TRACED) == 0) { imgp->credential_setid = true; VOP_UNLOCK(imgp->vp, 0); imgp->newcred = crdup(oldcred); if (attr.va_mode & S_ISUID) { euip = uifind(attr.va_uid); change_euid(imgp->newcred, euip); } vn_lock(imgp->vp, LK_EXCLUSIVE | LK_RETRY); if (attr.va_mode & S_ISGID) change_egid(imgp->newcred, attr.va_gid); /* * Implement correct POSIX saved-id behavior. * * XXXMAC: Note that the current logic will save the * uid and gid if a MAC domain transition occurs, even * though maybe it shouldn't. */ change_svuid(imgp->newcred, imgp->newcred->cr_uid); change_svgid(imgp->newcred, imgp->newcred->cr_gid); } else { /* * Implement correct POSIX saved-id behavior. * * XXX: It's not clear that the existing behavior is * POSIX-compliant. A number of sources indicate that the * saved uid/gid should only be updated if the new ruid is * not equal to the old ruid, or the new euid is not equal * to the old euid and the new euid is not equal to the old * ruid. The FreeBSD code always updates the saved uid/gid. * Also, this code uses the new (replaced) euid and egid as * the source, which may or may not be the right ones to use. */ if (oldcred->cr_svuid != oldcred->cr_uid || oldcred->cr_svgid != oldcred->cr_gid) { VOP_UNLOCK(imgp->vp, 0); imgp->newcred = crdup(oldcred); vn_lock(imgp->vp, LK_EXCLUSIVE | LK_RETRY); change_svuid(imgp->newcred, imgp->newcred->cr_uid); change_svgid(imgp->newcred, imgp->newcred->cr_gid); } } /* The new credentials are installed into the process later. */ /* * Do the best to calculate the full path to the image file. */ if (args->fname != NULL && args->fname[0] == '/') imgp->execpath = args->fname; else { VOP_UNLOCK(imgp->vp, 0); if (vn_fullpath(td, imgp->vp, &imgp->execpath, &imgp->freepath) != 0) imgp->execpath = args->fname; vn_lock(imgp->vp, LK_EXCLUSIVE | LK_RETRY); } /* * If the current process has a special image activator it * wants to try first, call it. For example, emulating shell * scripts differently. */ error = -1; if ((img_first = imgp->proc->p_sysent->sv_imgact_try) != NULL) error = img_first(imgp); /* * Loop through the list of image activators, calling each one. * An activator returns -1 if there is no match, 0 on success, * and an error otherwise. */ for (i = 0; error == -1 && execsw[i]; ++i) { if (execsw[i]->ex_imgact == NULL || execsw[i]->ex_imgact == img_first) { continue; } error = (*execsw[i]->ex_imgact)(imgp); } if (error) { if (error == -1) { if (textset == 0) VOP_UNSET_TEXT(imgp->vp); error = ENOEXEC; } goto exec_fail_dealloc; } /* * Special interpreter operation, cleanup and loop up to try to * activate the interpreter. */ if (imgp->interpreted) { exec_unmap_first_page(imgp); /* * VV_TEXT needs to be unset for scripts. There is a short * period before we determine that something is a script where * VV_TEXT will be set. The vnode lock is held over this * entire period so nothing should illegitimately be blocked. */ VOP_UNSET_TEXT(imgp->vp); /* free name buffer and old vnode */ if (args->fname != NULL) NDFREE(&nd, NDF_ONLY_PNBUF); #ifdef MAC mac_execve_interpreter_enter(newtextvp, &interpvplabel); #endif if (imgp->opened) { VOP_CLOSE(newtextvp, FREAD, td->td_ucred, td); imgp->opened = 0; } vput(newtextvp); vm_object_deallocate(imgp->object); imgp->object = NULL; imgp->credential_setid = false; if (imgp->newcred != NULL) { crfree(imgp->newcred); imgp->newcred = NULL; } imgp->execpath = NULL; free(imgp->freepath, M_TEMP); imgp->freepath = NULL; /* set new name to that of the interpreter */ NDINIT(&nd, LOOKUP, LOCKLEAF | FOLLOW | SAVENAME, UIO_SYSSPACE, imgp->interpreter_name, td); args->fname = imgp->interpreter_name; goto interpret; } /* * NB: We unlock the vnode here because it is believed that none * of the sv_copyout_strings/sv_fixup operations require the vnode. */ VOP_UNLOCK(imgp->vp, 0); if (disallow_high_osrel && P_OSREL_MAJOR(p->p_osrel) > P_OSREL_MAJOR(__FreeBSD_version)) { error = ENOEXEC; uprintf("Osrel %d for image %s too high\n", p->p_osrel, imgp->execpath != NULL ? imgp->execpath : ""); vn_lock(imgp->vp, LK_SHARED | LK_RETRY); goto exec_fail_dealloc; } /* ABI enforces the use of Capsicum. Switch into capabilities mode. */ if (SV_PROC_FLAG(p, SV_CAPSICUM)) sys_cap_enter(td, NULL); /* * Copy out strings (args and env) and initialize stack base */ if (p->p_sysent->sv_copyout_strings) stack_base = (*p->p_sysent->sv_copyout_strings)(imgp); else stack_base = exec_copyout_strings(imgp); /* * If custom stack fixup routine present for this process * let it do the stack setup. * Else stuff argument count as first item on stack */ if (p->p_sysent->sv_fixup != NULL) (*p->p_sysent->sv_fixup)(&stack_base, imgp); else suword(--stack_base, imgp->args->argc); if (args->fdp != NULL) { /* Install a brand new file descriptor table. */ fdinstall_remapped(td, args->fdp); args->fdp = NULL; } else { /* * Keep on using the existing file descriptor table. For * security and other reasons, the file descriptor table * cannot be shared after an exec. */ fdunshare(td); /* close files on exec */ fdcloseexec(td); } /* * Malloc things before we need locks. */ i = imgp->args->begin_envv - imgp->args->begin_argv; /* Cache arguments if they fit inside our allowance */ if (ps_arg_cache_limit >= i + sizeof(struct pargs)) { newargs = pargs_alloc(i); bcopy(imgp->args->begin_argv, newargs->ar_args, i); } /* * For security and other reasons, signal handlers cannot * be shared after an exec. The new process gets a copy of the old * handlers. In execsigs(), the new process will have its signals * reset. */ if (sigacts_shared(p->p_sigacts)) { oldsigacts = p->p_sigacts; newsigacts = sigacts_alloc(); sigacts_copy(newsigacts, oldsigacts); } vn_lock(imgp->vp, LK_SHARED | LK_RETRY); PROC_LOCK(p); if (oldsigacts) p->p_sigacts = newsigacts; /* Stop profiling */ stopprofclock(p); /* reset caught signals */ execsigs(p); /* name this process - nameiexec(p, ndp) */ bzero(p->p_comm, sizeof(p->p_comm)); if (args->fname) bcopy(nd.ni_cnd.cn_nameptr, p->p_comm, min(nd.ni_cnd.cn_namelen, MAXCOMLEN)); else if (vn_commname(newtextvp, p->p_comm, sizeof(p->p_comm)) != 0) bcopy(fexecv_proc_title, p->p_comm, sizeof(fexecv_proc_title)); bcopy(p->p_comm, td->td_name, sizeof(td->td_name)); #ifdef KTR sched_clear_tdname(td); #endif /* * mark as execed, wakeup the process that vforked (if any) and tell * it that it now has its own resources back */ p->p_flag |= P_EXEC; if ((p->p_flag2 & P2_NOTRACE_EXEC) == 0) p->p_flag2 &= ~P2_NOTRACE; if (p->p_flag & P_PPWAIT) { p->p_flag &= ~(P_PPWAIT | P_PPTRACE); cv_broadcast(&p->p_pwait); /* STOPs are no longer ignored, arrange for AST */ signotify(td); } /* * Implement image setuid/setgid installation. */ if (imgp->credential_setid) { /* * Turn off syscall tracing for set-id programs, except for * root. Record any set-id flags first to make sure that * we do not regain any tracing during a possible block. */ setsugid(p); #ifdef KTRACE if (p->p_tracecred != NULL && priv_check_cred(p->p_tracecred, PRIV_DEBUG_DIFFCRED, 0)) ktrprocexec(p, &tracecred, &tracevp); #endif /* * Close any file descriptors 0..2 that reference procfs, * then make sure file descriptors 0..2 are in use. * * Both fdsetugidsafety() and fdcheckstd() may call functions * taking sleepable locks, so temporarily drop our locks. */ PROC_UNLOCK(p); VOP_UNLOCK(imgp->vp, 0); fdsetugidsafety(td); error = fdcheckstd(td); vn_lock(imgp->vp, LK_SHARED | LK_RETRY); if (error != 0) goto exec_fail_dealloc; PROC_LOCK(p); #ifdef MAC if (will_transition) { mac_vnode_execve_transition(oldcred, imgp->newcred, imgp->vp, interpvplabel, imgp); } #endif } else { if (oldcred->cr_uid == oldcred->cr_ruid && oldcred->cr_gid == oldcred->cr_rgid) p->p_flag &= ~P_SUGID; } /* * Set the new credentials. */ if (imgp->newcred != NULL) { proc_set_cred(p, imgp->newcred); crfree(oldcred); oldcred = NULL; } /* * Store the vp for use in procfs. This vnode was referenced by namei * or fgetvp_exec. */ oldtextvp = p->p_textvp; p->p_textvp = newtextvp; #ifdef KDTRACE_HOOKS /* * Tell the DTrace fasttrap provider about the exec if it * has declared an interest. */ if (dtrace_fasttrap_exec) dtrace_fasttrap_exec(p); #endif /* * Notify others that we exec'd, and clear the P_INEXEC flag * as we're now a bona fide freshly-execed process. */ KNOTE_LOCKED(p->p_klist, NOTE_EXEC); p->p_flag &= ~P_INEXEC; /* clear "fork but no exec" flag, as we _are_ execing */ p->p_acflag &= ~AFORK; /* * Free any previous argument cache and replace it with * the new argument cache, if any. */ oldargs = p->p_args; p->p_args = newargs; newargs = NULL; #ifdef HWPMC_HOOKS /* * Check if system-wide sampling is in effect or if the * current process is using PMCs. If so, do exec() time * processing. This processing needs to happen AFTER the * P_INEXEC flag is cleared. * * The proc lock needs to be released before taking the PMC * SX. */ if (PMC_SYSTEM_SAMPLING_ACTIVE() || PMC_PROC_IS_USING_PMCS(p)) { PROC_UNLOCK(p); VOP_UNLOCK(imgp->vp, 0); pe.pm_credentialschanged = credential_changing; pe.pm_entryaddr = imgp->entry_addr; PMC_CALL_HOOK_X(td, PMC_FN_PROCESS_EXEC, (void *) &pe); vn_lock(imgp->vp, LK_SHARED | LK_RETRY); } else PROC_UNLOCK(p); #else /* !HWPMC_HOOKS */ PROC_UNLOCK(p); #endif /* Set values passed into the program in registers. */ if (p->p_sysent->sv_setregs) (*p->p_sysent->sv_setregs)(td, imgp, (u_long)(uintptr_t)stack_base); else exec_setregs(td, imgp, (u_long)(uintptr_t)stack_base); vfs_mark_atime(imgp->vp, td->td_ucred); SDT_PROBE1(proc, , , exec__success, args->fname); exec_fail_dealloc: if (imgp->firstpage != NULL) exec_unmap_first_page(imgp); if (imgp->vp != NULL) { if (args->fname) NDFREE(&nd, NDF_ONLY_PNBUF); if (imgp->opened) VOP_CLOSE(imgp->vp, FREAD, td->td_ucred, td); if (error != 0) vput(imgp->vp); else VOP_UNLOCK(imgp->vp, 0); } if (imgp->object != NULL) vm_object_deallocate(imgp->object); free(imgp->freepath, M_TEMP); if (error == 0) { PROC_LOCK(p); if (p->p_ptevents & PTRACE_EXEC) td->td_dbgflags |= TDB_EXEC; PROC_UNLOCK(p); /* * Stop the process here if its stop event mask has * the S_EXEC bit set. */ STOPEVENT(p, S_EXEC, 0); } else { exec_fail: /* we're done here, clear P_INEXEC */ PROC_LOCK(p); p->p_flag &= ~P_INEXEC; PROC_UNLOCK(p); SDT_PROBE1(proc, , , exec__failure, error); } if (imgp->newcred != NULL && oldcred != NULL) crfree(imgp->newcred); #ifdef MAC mac_execve_exit(imgp); mac_execve_interpreter_exit(interpvplabel); #endif exec_free_args(args); /* * Handle deferred decrement of ref counts. */ if (oldtextvp != NULL) vrele(oldtextvp); #ifdef KTRACE if (tracevp != NULL) vrele(tracevp); if (tracecred != NULL) crfree(tracecred); #endif pargs_drop(oldargs); pargs_drop(newargs); if (oldsigacts != NULL) sigacts_free(oldsigacts); if (euip != NULL) uifree(euip); if (error && imgp->vmspace_destroyed) { /* sorry, no more process anymore. exit gracefully */ exit1(td, 0, SIGABRT); /* NOT REACHED */ } #ifdef KTRACE if (error == 0) ktrprocctor(p); #endif return (error); } int exec_map_first_page(imgp) struct image_params *imgp; { int rv, i, after, initial_pagein; vm_page_t ma[VM_INITIAL_PAGEIN]; vm_object_t object; if (imgp->firstpage != NULL) exec_unmap_first_page(imgp); object = imgp->vp->v_object; if (object == NULL) return (EACCES); VM_OBJECT_WLOCK(object); #if VM_NRESERVLEVEL > 0 vm_object_color(object, 0); #endif ma[0] = vm_page_grab(object, 0, VM_ALLOC_NORMAL | VM_ALLOC_NOBUSY); if (ma[0]->valid != VM_PAGE_BITS_ALL) { vm_page_xbusy(ma[0]); if (!vm_pager_has_page(object, 0, NULL, &after)) { vm_page_lock(ma[0]); vm_page_free(ma[0]); vm_page_unlock(ma[0]); VM_OBJECT_WUNLOCK(object); return (EIO); } initial_pagein = min(after, VM_INITIAL_PAGEIN); KASSERT(initial_pagein <= object->size, ("%s: initial_pagein %d object->size %ju", __func__, initial_pagein, (uintmax_t )object->size)); for (i = 1; i < initial_pagein; i++) { if ((ma[i] = vm_page_next(ma[i - 1])) != NULL) { if (ma[i]->valid) break; if (vm_page_tryxbusy(ma[i])) break; } else { ma[i] = vm_page_alloc(object, i, VM_ALLOC_NORMAL); if (ma[i] == NULL) break; } } initial_pagein = i; rv = vm_pager_get_pages(object, ma, initial_pagein, NULL, NULL); if (rv != VM_PAGER_OK) { for (i = 0; i < initial_pagein; i++) { vm_page_lock(ma[i]); vm_page_free(ma[i]); vm_page_unlock(ma[i]); } VM_OBJECT_WUNLOCK(object); return (EIO); } vm_page_xunbusy(ma[0]); for (i = 1; i < initial_pagein; i++) vm_page_readahead_finish(ma[i]); } vm_page_lock(ma[0]); vm_page_hold(ma[0]); vm_page_activate(ma[0]); vm_page_unlock(ma[0]); VM_OBJECT_WUNLOCK(object); imgp->firstpage = sf_buf_alloc(ma[0], 0); imgp->image_header = (char *)sf_buf_kva(imgp->firstpage); return (0); } void exec_unmap_first_page(imgp) struct image_params *imgp; { vm_page_t m; if (imgp->firstpage != NULL) { m = sf_buf_page(imgp->firstpage); sf_buf_free(imgp->firstpage); imgp->firstpage = NULL; vm_page_lock(m); vm_page_unhold(m); vm_page_unlock(m); } } /* * Destroy old address space, and allocate a new stack * The new stack is only SGROWSIZ large because it is grown * automatically in trap.c. */ int exec_new_vmspace(imgp, sv) struct image_params *imgp; struct sysentvec *sv; { int error; struct proc *p = imgp->proc; struct vmspace *vmspace = p->p_vmspace; vm_object_t obj; struct rlimit rlim_stack; vm_offset_t sv_minuser, stack_addr; vm_map_t map; u_long ssiz; imgp->vmspace_destroyed = 1; imgp->sysent = sv; /* May be called with Giant held */ EVENTHANDLER_INVOKE(process_exec, p, imgp); /* * Blow away entire process VM, if address space not shared, * otherwise, create a new VM space so that other threads are * not disrupted */ map = &vmspace->vm_map; if (map_at_zero) sv_minuser = sv->sv_minuser; else sv_minuser = MAX(sv->sv_minuser, PAGE_SIZE); if (vmspace->vm_refcnt == 1 && vm_map_min(map) == sv_minuser && vm_map_max(map) == sv->sv_maxuser) { shmexit(vmspace); pmap_remove_pages(vmspace_pmap(vmspace)); vm_map_remove(map, vm_map_min(map), vm_map_max(map)); } else { error = vmspace_exec(p, sv_minuser, sv->sv_maxuser); if (error) return (error); vmspace = p->p_vmspace; map = &vmspace->vm_map; } /* Map a shared page */ obj = sv->sv_shared_page_obj; if (obj != NULL) { vm_object_reference(obj); error = vm_map_fixed(map, obj, 0, sv->sv_shared_page_base, sv->sv_shared_page_len, VM_PROT_READ | VM_PROT_EXECUTE, VM_PROT_READ | VM_PROT_EXECUTE, MAP_INHERIT_SHARE | MAP_ACC_NO_CHARGE); if (error) { vm_object_deallocate(obj); return (error); } } /* Allocate a new stack */ if (imgp->stack_sz != 0) { ssiz = trunc_page(imgp->stack_sz); PROC_LOCK(p); lim_rlimit_proc(p, RLIMIT_STACK, &rlim_stack); PROC_UNLOCK(p); if (ssiz > rlim_stack.rlim_max) ssiz = rlim_stack.rlim_max; if (ssiz > rlim_stack.rlim_cur) { rlim_stack.rlim_cur = ssiz; kern_setrlimit(curthread, RLIMIT_STACK, &rlim_stack); } } else if (sv->sv_maxssiz != NULL) { ssiz = *sv->sv_maxssiz; } else { ssiz = maxssiz; } stack_addr = sv->sv_usrstack - ssiz; error = vm_map_stack(map, stack_addr, (vm_size_t)ssiz, obj != NULL && imgp->stack_prot != 0 ? imgp->stack_prot : sv->sv_stackprot, VM_PROT_ALL, MAP_STACK_GROWS_DOWN); if (error) return (error); /* * vm_ssize and vm_maxsaddr are somewhat antiquated concepts, but they * are still used to enforce the stack rlimit on the process stack. */ vmspace->vm_ssize = sgrowsiz >> PAGE_SHIFT; vmspace->vm_maxsaddr = (char *)stack_addr; return (0); } /* * Copy out argument and environment strings from the old process address * space into the temporary string buffer. */ int exec_copyin_args(struct image_args *args, char *fname, enum uio_seg segflg, char **argv, char **envv) { u_long argp, envp; int error; size_t length; bzero(args, sizeof(*args)); if (argv == NULL) return (EFAULT); /* * Allocate demand-paged memory for the file name, argument, and * environment strings. */ error = exec_alloc_args(args); if (error != 0) return (error); /* * Copy the file name. */ if (fname != NULL) { args->fname = args->buf; error = (segflg == UIO_SYSSPACE) ? copystr(fname, args->fname, PATH_MAX, &length) : copyinstr(fname, args->fname, PATH_MAX, &length); if (error != 0) goto err_exit; } else length = 0; args->begin_argv = args->buf + length; args->endp = args->begin_argv; args->stringspace = ARG_MAX; /* * extract arguments first */ for (;;) { error = fueword(argv++, &argp); if (error == -1) { error = EFAULT; goto err_exit; } if (argp == 0) break; error = copyinstr((void *)(uintptr_t)argp, args->endp, args->stringspace, &length); if (error != 0) { if (error == ENAMETOOLONG) error = E2BIG; goto err_exit; } args->stringspace -= length; args->endp += length; args->argc++; } args->begin_envv = args->endp; /* * extract environment strings */ if (envv) { for (;;) { error = fueword(envv++, &envp); if (error == -1) { error = EFAULT; goto err_exit; } if (envp == 0) break; error = copyinstr((void *)(uintptr_t)envp, args->endp, args->stringspace, &length); if (error != 0) { if (error == ENAMETOOLONG) error = E2BIG; goto err_exit; } args->stringspace -= length; args->endp += length; args->envc++; } } return (0); err_exit: exec_free_args(args); return (error); } int exec_copyin_data_fds(struct thread *td, struct image_args *args, const void *data, size_t datalen, const int *fds, size_t fdslen) { struct filedesc *ofdp; const char *p; int *kfds; int error; memset(args, '\0', sizeof(*args)); ofdp = td->td_proc->p_fd; if (datalen >= ARG_MAX || fdslen > ofdp->fd_lastfile + 1) return (E2BIG); error = exec_alloc_args(args); if (error != 0) return (error); args->begin_argv = args->buf; args->stringspace = ARG_MAX; if (datalen > 0) { /* * Argument buffer has been provided. Copy it into the * kernel as a single string and add a terminating null * byte. */ error = copyin(data, args->begin_argv, datalen); if (error != 0) goto err_exit; args->begin_argv[datalen] = '\0'; args->endp = args->begin_argv + datalen + 1; args->stringspace -= datalen + 1; /* * Traditional argument counting. Count the number of * null bytes. */ for (p = args->begin_argv; p < args->endp; ++p) if (*p == '\0') ++args->argc; } else { /* No argument buffer provided. */ args->endp = args->begin_argv; } /* There are no environment variables. */ args->begin_envv = args->endp; /* Create new file descriptor table. */ kfds = malloc(fdslen * sizeof(int), M_TEMP, M_WAITOK); error = copyin(fds, kfds, fdslen * sizeof(int)); if (error != 0) { free(kfds, M_TEMP); goto err_exit; } error = fdcopy_remapped(ofdp, kfds, fdslen, &args->fdp); free(kfds, M_TEMP); if (error != 0) goto err_exit; return (0); err_exit: exec_free_args(args); return (error); } +struct exec_args_kva { + vm_offset_t addr; + SLIST_ENTRY(exec_args_kva) next; +}; + +static DPCPU_DEFINE(struct exec_args_kva *, exec_args_kva); + +static SLIST_HEAD(, exec_args_kva) exec_args_kva_freelist; +static struct mtx exec_args_kva_mtx; + +static void +exec_prealloc_args_kva(void *arg __unused) +{ + struct exec_args_kva *argkva; + u_int i; + + SLIST_INIT(&exec_args_kva_freelist); + mtx_init(&exec_args_kva_mtx, "exec args kva", NULL, MTX_DEF); + for (i = 0; i < exec_map_entries; i++) { + argkva = malloc(sizeof(*argkva), M_PARGS, M_WAITOK); + argkva->addr = kmap_alloc_wait(exec_map, exec_map_entry_size); + SLIST_INSERT_HEAD(&exec_args_kva_freelist, argkva, next); + } +} +SYSINIT(exec_args_kva, SI_SUB_EXEC, SI_ORDER_ANY, exec_prealloc_args_kva, NULL); + +static vm_offset_t +exec_alloc_args_kva(void **cookie) +{ + struct exec_args_kva *argkva; + + argkva = (void *)atomic_readandclear_ptr( + (uintptr_t *)DPCPU_PTR(exec_args_kva)); + if (argkva == NULL) { + mtx_lock(&exec_args_kva_mtx); + while ((argkva = SLIST_FIRST(&exec_args_kva_freelist)) == NULL) + (void)mtx_sleep(&exec_args_kva_freelist, + &exec_args_kva_mtx, 0, "execkva", 0); + SLIST_REMOVE_HEAD(&exec_args_kva_freelist, next); + mtx_unlock(&exec_args_kva_mtx); + } + *(struct exec_args_kva **)cookie = argkva; + return (argkva->addr); +} + +static void +exec_free_args_kva(void *cookie) +{ + struct exec_args_kva *argkva; + vm_offset_t base; + + argkva = cookie; + base = argkva->addr; + + vm_map_madvise(exec_map, base, base + exec_map_entry_size, MADV_FREE); + if (!atomic_cmpset_ptr((uintptr_t *)DPCPU_PTR(exec_args_kva), + (uintptr_t)NULL, (uintptr_t)argkva)) { + mtx_lock(&exec_args_kva_mtx); + SLIST_INSERT_HEAD(&exec_args_kva_freelist, argkva, next); + wakeup_one(&exec_args_kva_freelist); + mtx_unlock(&exec_args_kva_mtx); + } +} + /* * Allocate temporary demand-paged, zero-filled memory for the file name, - * argument, and environment strings. Returns zero if the allocation succeeds - * and ENOMEM otherwise. + * argument, and environment strings. */ int exec_alloc_args(struct image_args *args) { - args->buf = (char *)kmap_alloc_wait(exec_map, PATH_MAX + ARG_MAX); - return (args->buf != NULL ? 0 : ENOMEM); + args->buf = (char *)exec_alloc_args_kva(&args->bufkva); + return (0); } void exec_free_args(struct image_args *args) { if (args->buf != NULL) { - kmap_free_wakeup(exec_map, (vm_offset_t)args->buf, - PATH_MAX + ARG_MAX); + exec_free_args_kva(args->bufkva); args->buf = NULL; } if (args->fname_buf != NULL) { free(args->fname_buf, M_TEMP); args->fname_buf = NULL; } if (args->fdp != NULL) fdescfree_remapped(args->fdp); } /* * Copy strings out to the new process address space, constructing new arg * and env vector tables. Return a pointer to the base so that it can be used * as the initial stack pointer. */ register_t * exec_copyout_strings(imgp) struct image_params *imgp; { int argc, envc; char **vectp; char *stringp; uintptr_t destp; register_t *stack_base; struct ps_strings *arginfo; struct proc *p; size_t execpath_len; int szsigcode, szps; char canary[sizeof(long) * 8]; szps = sizeof(pagesizes[0]) * MAXPAGESIZES; /* * Calculate string base and vector table pointers. * Also deal with signal trampoline code for this exec type. */ if (imgp->execpath != NULL && imgp->auxargs != NULL) execpath_len = strlen(imgp->execpath) + 1; else execpath_len = 0; p = imgp->proc; szsigcode = 0; arginfo = (struct ps_strings *)p->p_sysent->sv_psstrings; if (p->p_sysent->sv_sigcode_base == 0) { if (p->p_sysent->sv_szsigcode != NULL) szsigcode = *(p->p_sysent->sv_szsigcode); } destp = (uintptr_t)arginfo; /* * install sigcode */ if (szsigcode != 0) { destp -= szsigcode; destp = rounddown2(destp, sizeof(void *)); copyout(p->p_sysent->sv_sigcode, (void *)destp, szsigcode); } /* * Copy the image path for the rtld. */ if (execpath_len != 0) { destp -= execpath_len; imgp->execpathp = destp; copyout(imgp->execpath, (void *)destp, execpath_len); } /* * Prepare the canary for SSP. */ arc4rand(canary, sizeof(canary), 0); destp -= sizeof(canary); imgp->canary = destp; copyout(canary, (void *)destp, sizeof(canary)); imgp->canarylen = sizeof(canary); /* * Prepare the pagesizes array. */ destp -= szps; destp = rounddown2(destp, sizeof(void *)); imgp->pagesizes = destp; copyout(pagesizes, (void *)destp, szps); imgp->pagesizeslen = szps; destp -= ARG_MAX - imgp->args->stringspace; destp = rounddown2(destp, sizeof(void *)); /* * If we have a valid auxargs ptr, prepare some room * on the stack. */ if (imgp->auxargs) { /* * 'AT_COUNT*2' is size for the ELF Auxargs data. This is for * lower compatibility. */ imgp->auxarg_size = (imgp->auxarg_size) ? imgp->auxarg_size : (AT_COUNT * 2); /* * The '+ 2' is for the null pointers at the end of each of * the arg and env vector sets,and imgp->auxarg_size is room * for argument of Runtime loader. */ vectp = (char **)(destp - (imgp->args->argc + imgp->args->envc + 2 + imgp->auxarg_size) * sizeof(char *)); } else { /* * The '+ 2' is for the null pointers at the end of each of * the arg and env vector sets */ vectp = (char **)(destp - (imgp->args->argc + imgp->args->envc + 2) * sizeof(char *)); } /* * vectp also becomes our initial stack base */ stack_base = (register_t *)vectp; stringp = imgp->args->begin_argv; argc = imgp->args->argc; envc = imgp->args->envc; /* * Copy out strings - arguments and environment. */ copyout(stringp, (void *)destp, ARG_MAX - imgp->args->stringspace); /* * Fill in "ps_strings" struct for ps, w, etc. */ suword(&arginfo->ps_argvstr, (long)(intptr_t)vectp); suword32(&arginfo->ps_nargvstr, argc); /* * Fill in argument portion of vector table. */ for (; argc > 0; --argc) { suword(vectp++, (long)(intptr_t)destp); while (*stringp++ != 0) destp++; destp++; } /* a null vector table pointer separates the argp's from the envp's */ suword(vectp++, 0); suword(&arginfo->ps_envstr, (long)(intptr_t)vectp); suword32(&arginfo->ps_nenvstr, envc); /* * Fill in environment portion of vector table. */ for (; envc > 0; --envc) { suword(vectp++, (long)(intptr_t)destp); while (*stringp++ != 0) destp++; destp++; } /* end of vector table is a null pointer */ suword(vectp, 0); return (stack_base); } /* * Check permissions of file to execute. * Called with imgp->vp locked. * Return 0 for success or error code on failure. */ int exec_check_permissions(imgp) struct image_params *imgp; { struct vnode *vp = imgp->vp; struct vattr *attr = imgp->attr; struct thread *td; int error, writecount; td = curthread; /* Get file attributes */ error = VOP_GETATTR(vp, attr, td->td_ucred); if (error) return (error); #ifdef MAC error = mac_vnode_check_exec(td->td_ucred, imgp->vp, imgp); if (error) return (error); #endif /* * 1) Check if file execution is disabled for the filesystem that * this file resides on. * 2) Ensure that at least one execute bit is on. Otherwise, a * privileged user will always succeed, and we don't want this * to happen unless the file really is executable. * 3) Ensure that the file is a regular file. */ if ((vp->v_mount->mnt_flag & MNT_NOEXEC) || (attr->va_mode & (S_IXUSR | S_IXGRP | S_IXOTH)) == 0 || (attr->va_type != VREG)) return (EACCES); /* * Zero length files can't be exec'd */ if (attr->va_size == 0) return (ENOEXEC); /* * Check for execute permission to file based on current credentials. */ error = VOP_ACCESS(vp, VEXEC, td->td_ucred, td); if (error) return (error); /* * Check number of open-for-writes on the file and deny execution * if there are any. */ error = VOP_GET_WRITECOUNT(vp, &writecount); if (error != 0) return (error); if (writecount != 0) return (ETXTBSY); /* * Call filesystem specific open routine (which does nothing in the * general case). */ error = VOP_OPEN(vp, FREAD, td->td_ucred, td, NULL); if (error == 0) imgp->opened = 1; return (error); } /* * Exec handler registration */ int exec_register(execsw_arg) const struct execsw *execsw_arg; { const struct execsw **es, **xs, **newexecsw; int count = 2; /* New slot and trailing NULL */ if (execsw) for (es = execsw; *es; es++) count++; newexecsw = malloc(count * sizeof(*es), M_TEMP, M_WAITOK); xs = newexecsw; if (execsw) for (es = execsw; *es; es++) *xs++ = *es; *xs++ = execsw_arg; *xs = NULL; if (execsw) free(execsw, M_TEMP); execsw = newexecsw; return (0); } int exec_unregister(execsw_arg) const struct execsw *execsw_arg; { const struct execsw **es, **xs, **newexecsw; int count = 1; if (execsw == NULL) panic("unregister with no handlers left?\n"); for (es = execsw; *es; es++) { if (*es == execsw_arg) break; } if (*es == NULL) return (ENOENT); for (es = execsw; *es; es++) if (*es != execsw_arg) count++; newexecsw = malloc(count * sizeof(*es), M_TEMP, M_WAITOK); xs = newexecsw; for (es = execsw; *es; es++) if (*es != execsw_arg) *xs++ = *es; *xs = NULL; if (execsw) free(execsw, M_TEMP); execsw = newexecsw; return (0); } Index: projects/clang400-import/sys/kern/kern_sendfile.c =================================================================== --- projects/clang400-import/sys/kern/kern_sendfile.c (revision 311459) +++ projects/clang400-import/sys/kern/kern_sendfile.c (revision 311460) @@ -1,1042 +1,1040 @@ /*- * Copyright (c) 2013-2015 Gleb Smirnoff * Copyright (c) 1998, David Greenman. 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. */ #include __FBSDID("$FreeBSD$"); #include "opt_compat.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 -extern vm_page_t bogus_page; - /* * Structure describing a single sendfile(2) I/O, which may consist of * several underlying pager I/Os. * * The syscall context allocates the structure and initializes 'nios' * to 1. As sendfile_swapin() runs through pages and starts asynchronous * paging operations, it increments 'nios'. * * Every I/O completion calls sendfile_iodone(), which decrements the 'nios', * and the syscall also calls sendfile_iodone() after allocating all mbufs, * linking them and sending to socket. Whoever reaches zero 'nios' is * responsible to * call pru_ready on the socket, to notify it of readyness * of the data. */ struct sf_io { volatile u_int nios; u_int error; int npages; struct file *sock_fp; struct mbuf *m; vm_page_t pa[]; }; /* * Structure used to track requests with SF_SYNC flag. */ struct sendfile_sync { struct mtx mtx; struct cv cv; unsigned count; }; counter_u64_t sfstat[sizeof(struct sfstat) / sizeof(uint64_t)]; static void sfstat_init(const void *unused) { COUNTER_ARRAY_ALLOC(sfstat, sizeof(struct sfstat) / sizeof(uint64_t), M_WAITOK); } SYSINIT(sfstat, SI_SUB_MBUF, SI_ORDER_FIRST, sfstat_init, NULL); static int sfstat_sysctl(SYSCTL_HANDLER_ARGS) { struct sfstat s; COUNTER_ARRAY_COPY(sfstat, &s, sizeof(s) / sizeof(uint64_t)); if (req->newptr) COUNTER_ARRAY_ZERO(sfstat, sizeof(s) / sizeof(uint64_t)); return (SYSCTL_OUT(req, &s, sizeof(s))); } SYSCTL_PROC(_kern_ipc, OID_AUTO, sfstat, CTLTYPE_OPAQUE | CTLFLAG_RW, NULL, 0, sfstat_sysctl, "I", "sendfile statistics"); /* * Detach mapped page and release resources back to the system. Called * by mbuf(9) code when last reference to a page is freed. */ void sf_ext_free(void *arg1, void *arg2) { struct sf_buf *sf = arg1; struct sendfile_sync *sfs = arg2; vm_page_t pg = sf_buf_page(sf); sf_buf_free(sf); vm_page_lock(pg); /* * Check for the object going away on us. This can * happen since we don't hold a reference to it. * If so, we're responsible for freeing the page. */ if (vm_page_unwire(pg, PQ_INACTIVE) && pg->object == NULL) vm_page_free(pg); vm_page_unlock(pg); if (sfs != NULL) { mtx_lock(&sfs->mtx); KASSERT(sfs->count > 0, ("Sendfile sync botchup count == 0")); if (--sfs->count == 0) cv_signal(&sfs->cv); mtx_unlock(&sfs->mtx); } } /* * Same as above, but forces the page to be detached from the object * and go into free pool. */ void sf_ext_free_nocache(void *arg1, void *arg2) { struct sf_buf *sf = arg1; struct sendfile_sync *sfs = arg2; vm_page_t pg = sf_buf_page(sf); sf_buf_free(sf); vm_page_lock(pg); if (vm_page_unwire(pg, PQ_NONE)) { vm_object_t obj; /* Try to free the page, but only if it is cheap to. */ if ((obj = pg->object) == NULL) vm_page_free(pg); else if (!vm_page_xbusied(pg) && VM_OBJECT_TRYWLOCK(obj)) { vm_page_free(pg); VM_OBJECT_WUNLOCK(obj); } else vm_page_deactivate(pg); } vm_page_unlock(pg); if (sfs != NULL) { mtx_lock(&sfs->mtx); KASSERT(sfs->count > 0, ("Sendfile sync botchup count == 0")); if (--sfs->count == 0) cv_signal(&sfs->cv); mtx_unlock(&sfs->mtx); } } /* * Helper function to calculate how much data to put into page i of n. * Only first and last pages are special. */ static inline off_t xfsize(int i, int n, off_t off, off_t len) { if (i == 0) return (omin(PAGE_SIZE - (off & PAGE_MASK), len)); if (i == n - 1 && ((off + len) & PAGE_MASK) > 0) return ((off + len) & PAGE_MASK); return (PAGE_SIZE); } /* * Helper function to get offset within object for i page. */ static inline vm_offset_t vmoff(int i, off_t off) { if (i == 0) return ((vm_offset_t)off); return (trunc_page(off + i * PAGE_SIZE)); } /* * Helper function used when allocation of a page or sf_buf failed. * Pretend as if we don't have enough space, subtract xfsize() of * all pages that failed. */ static inline void fixspace(int old, int new, off_t off, int *space) { KASSERT(old > new, ("%s: old %d new %d", __func__, old, new)); /* Subtract last one. */ *space -= xfsize(old - 1, old, off, *space); old--; if (new == old) /* There was only one page. */ return; /* Subtract first one. */ if (new == 0) { *space -= xfsize(0, old, off, *space); new++; } /* Rest of pages are full sized. */ *space -= (old - new) * PAGE_SIZE; KASSERT(*space >= 0, ("%s: space went backwards", __func__)); } /* * I/O completion callback. */ static void sendfile_iodone(void *arg, vm_page_t *pg, int count, int error) { struct sf_io *sfio = arg; struct socket *so; for (int i = 0; i < count; i++) if (pg[i] != bogus_page) vm_page_xunbusy(pg[i]); if (error) sfio->error = error; if (!refcount_release(&sfio->nios)) return; so = sfio->sock_fp->f_data; if (sfio->error) { struct mbuf *m; /* * I/O operation failed. The state of data in the socket * is now inconsistent, and all what we can do is to tear * it down. Protocol abort method would tear down protocol * state, free all ready mbufs and detach not ready ones. * We will free the mbufs corresponding to this I/O manually. * * The socket would be marked with EIO and made available * for read, so that application receives EIO on next * syscall and eventually closes the socket. */ so->so_proto->pr_usrreqs->pru_abort(so); so->so_error = EIO; m = sfio->m; for (int i = 0; i < sfio->npages; i++) m = m_free(m); } else { CURVNET_SET(so->so_vnet); (void )(so->so_proto->pr_usrreqs->pru_ready)(so, sfio->m, sfio->npages); CURVNET_RESTORE(); } /* XXXGL: curthread */ fdrop(sfio->sock_fp, curthread); free(sfio, M_TEMP); } /* * Iterate through pages vector and request paging for non-valid pages. */ static int sendfile_swapin(vm_object_t obj, struct sf_io *sfio, off_t off, off_t len, int npages, int rhpages, int flags) { vm_page_t *pa = sfio->pa; int nios; nios = 0; flags = (flags & SF_NODISKIO) ? VM_ALLOC_NOWAIT : 0; /* * First grab all the pages and wire them. Note that we grab * only required pages. Readahead pages are dealt with later. */ VM_OBJECT_WLOCK(obj); for (int i = 0; i < npages; i++) { pa[i] = vm_page_grab(obj, OFF_TO_IDX(vmoff(i, off)), VM_ALLOC_WIRED | VM_ALLOC_NORMAL | flags); if (pa[i] == NULL) { npages = i; rhpages = 0; break; } } for (int i = 0; i < npages;) { int j, a, count, rv; /* Skip valid pages. */ if (vm_page_is_valid(pa[i], vmoff(i, off) & PAGE_MASK, xfsize(i, npages, off, len))) { vm_page_xunbusy(pa[i]); SFSTAT_INC(sf_pages_valid); i++; continue; } /* * Next page is invalid. Check if it belongs to pager. It * may not be there, which is a regular situation for shmem * pager. For vnode pager this happens only in case of * a sparse file. * * Important feature of vm_pager_has_page() is the hint * stored in 'a', about how many pages we can pagein after * this page in a single I/O. */ if (!vm_pager_has_page(obj, OFF_TO_IDX(vmoff(i, off)), NULL, &a)) { pmap_zero_page(pa[i]); pa[i]->valid = VM_PAGE_BITS_ALL; pa[i]->dirty = 0; vm_page_xunbusy(pa[i]); i++; continue; } /* * We want to pagein as many pages as possible, limited only * by the 'a' hint and actual request. */ count = min(a + 1, npages - i); /* * We should not pagein into a valid page, thus we first trim * any valid pages off the end of request, and substitute * to bogus_page those, that are in the middle. */ for (j = i + count - 1; j > i; j--) { if (vm_page_is_valid(pa[j], vmoff(j, off) & PAGE_MASK, xfsize(j, npages, off, len))) { count--; rhpages = 0; } else break; } for (j = i + 1; j < i + count - 1; j++) if (vm_page_is_valid(pa[j], vmoff(j, off) & PAGE_MASK, xfsize(j, npages, off, len))) { vm_page_xunbusy(pa[j]); SFSTAT_INC(sf_pages_valid); SFSTAT_INC(sf_pages_bogus); pa[j] = bogus_page; } refcount_acquire(&sfio->nios); rv = vm_pager_get_pages_async(obj, pa + i, count, NULL, i + count == npages ? &rhpages : NULL, &sendfile_iodone, sfio); KASSERT(rv == VM_PAGER_OK, ("%s: pager fail obj %p page %p", __func__, obj, pa[i])); SFSTAT_INC(sf_iocnt); SFSTAT_ADD(sf_pages_read, count); if (i + count == npages) SFSTAT_ADD(sf_rhpages_read, rhpages); /* * Restore the valid page pointers. They are already * unbusied, but still wired. */ for (j = i; j < i + count; j++) if (pa[j] == bogus_page) { pa[j] = vm_page_lookup(obj, OFF_TO_IDX(vmoff(j, off))); KASSERT(pa[j], ("%s: page %p[%d] disappeared", __func__, pa, j)); } i += count; nios++; } VM_OBJECT_WUNLOCK(obj); if (nios == 0 && npages != 0) SFSTAT_INC(sf_noiocnt); return (nios); } static int sendfile_getobj(struct thread *td, struct file *fp, vm_object_t *obj_res, struct vnode **vp_res, struct shmfd **shmfd_res, off_t *obj_size, int *bsize) { struct vattr va; vm_object_t obj; struct vnode *vp; struct shmfd *shmfd; int error; vp = *vp_res = NULL; obj = NULL; shmfd = *shmfd_res = NULL; *bsize = 0; /* * The file descriptor must be a regular file and have a * backing VM object. */ if (fp->f_type == DTYPE_VNODE) { vp = fp->f_vnode; vn_lock(vp, LK_SHARED | LK_RETRY); if (vp->v_type != VREG) { error = EINVAL; goto out; } *bsize = vp->v_mount->mnt_stat.f_iosize; error = VOP_GETATTR(vp, &va, td->td_ucred); if (error != 0) goto out; *obj_size = va.va_size; obj = vp->v_object; if (obj == NULL) { error = EINVAL; goto out; } } else if (fp->f_type == DTYPE_SHM) { error = 0; shmfd = fp->f_data; obj = shmfd->shm_object; *obj_size = shmfd->shm_size; } else { error = EINVAL; goto out; } VM_OBJECT_WLOCK(obj); if ((obj->flags & OBJ_DEAD) != 0) { VM_OBJECT_WUNLOCK(obj); error = EBADF; goto out; } /* * Temporarily increase the backing VM object's reference * count so that a forced reclamation of its vnode does not * immediately destroy it. */ vm_object_reference_locked(obj); VM_OBJECT_WUNLOCK(obj); *obj_res = obj; *vp_res = vp; *shmfd_res = shmfd; out: if (vp != NULL) VOP_UNLOCK(vp, 0); return (error); } static int sendfile_getsock(struct thread *td, int s, struct file **sock_fp, struct socket **so) { cap_rights_t rights; int error; *sock_fp = NULL; *so = NULL; /* * The socket must be a stream socket and connected. */ error = getsock_cap(td, s, cap_rights_init(&rights, CAP_SEND), sock_fp, NULL, NULL); if (error != 0) return (error); *so = (*sock_fp)->f_data; if ((*so)->so_type != SOCK_STREAM) return (EINVAL); if (((*so)->so_state & SS_ISCONNECTED) == 0) return (ENOTCONN); return (0); } int vn_sendfile(struct file *fp, int sockfd, struct uio *hdr_uio, struct uio *trl_uio, off_t offset, size_t nbytes, off_t *sent, int flags, struct thread *td) { struct file *sock_fp; struct vnode *vp; struct vm_object *obj; struct socket *so; struct mbuf *m, *mh, *mhtail; struct sf_buf *sf; struct shmfd *shmfd; struct sendfile_sync *sfs; struct vattr va; off_t off, sbytes, rem, obj_size; int error, softerr, bsize, hdrlen; obj = NULL; so = NULL; m = mh = NULL; sfs = NULL; hdrlen = sbytes = 0; softerr = 0; error = sendfile_getobj(td, fp, &obj, &vp, &shmfd, &obj_size, &bsize); if (error != 0) return (error); error = sendfile_getsock(td, sockfd, &sock_fp, &so); if (error != 0) goto out; #ifdef MAC error = mac_socket_check_send(td->td_ucred, so); if (error != 0) goto out; #endif SFSTAT_INC(sf_syscalls); SFSTAT_ADD(sf_rhpages_requested, SF_READAHEAD(flags)); if (flags & SF_SYNC) { sfs = malloc(sizeof *sfs, M_TEMP, M_WAITOK | M_ZERO); mtx_init(&sfs->mtx, "sendfile", NULL, MTX_DEF); cv_init(&sfs->cv, "sendfile"); } rem = nbytes ? omin(nbytes, obj_size - offset) : obj_size - offset; /* * Protect against multiple writers to the socket. * * XXXRW: Historically this has assumed non-interruptibility, so now * we implement that, but possibly shouldn't. */ (void)sblock(&so->so_snd, SBL_WAIT | SBL_NOINTR); /* * Loop through the pages of the file, starting with the requested * offset. Get a file page (do I/O if necessary), map the file page * into an sf_buf, attach an mbuf header to the sf_buf, and queue * it on the socket. * This is done in two loops. The inner loop turns as many pages * as it can, up to available socket buffer space, without blocking * into mbufs to have it bulk delivered into the socket send buffer. * The outer loop checks the state and available space of the socket * and takes care of the overall progress. */ for (off = offset; rem > 0; ) { struct sf_io *sfio; vm_page_t *pa; struct mbuf *mtail; int nios, space, npages, rhpages; mtail = NULL; /* * Check the socket state for ongoing connection, * no errors and space in socket buffer. * If space is low allow for the remainder of the * file to be processed if it fits the socket buffer. * Otherwise block in waiting for sufficient space * to proceed, or if the socket is nonblocking, return * to userland with EAGAIN while reporting how far * we've come. * We wait until the socket buffer has significant free * space to do bulk sends. This makes good use of file * system read ahead and allows packet segmentation * offloading hardware to take over lots of work. If * we were not careful here we would send off only one * sfbuf at a time. */ SOCKBUF_LOCK(&so->so_snd); if (so->so_snd.sb_lowat < so->so_snd.sb_hiwat / 2) so->so_snd.sb_lowat = so->so_snd.sb_hiwat / 2; retry_space: if (so->so_snd.sb_state & SBS_CANTSENDMORE) { error = EPIPE; SOCKBUF_UNLOCK(&so->so_snd); goto done; } else if (so->so_error) { error = so->so_error; so->so_error = 0; SOCKBUF_UNLOCK(&so->so_snd); goto done; } space = sbspace(&so->so_snd); if (space < rem && (space <= 0 || space < so->so_snd.sb_lowat)) { if (so->so_state & SS_NBIO) { SOCKBUF_UNLOCK(&so->so_snd); error = EAGAIN; goto done; } /* * sbwait drops the lock while sleeping. * When we loop back to retry_space the * state may have changed and we retest * for it. */ error = sbwait(&so->so_snd); /* * An error from sbwait usually indicates that we've * been interrupted by a signal. If we've sent anything * then return bytes sent, otherwise return the error. */ if (error != 0) { SOCKBUF_UNLOCK(&so->so_snd); goto done; } goto retry_space; } SOCKBUF_UNLOCK(&so->so_snd); /* * At the beginning of the first loop check if any headers * are specified and copy them into mbufs. Reduce space in * the socket buffer by the size of the header mbuf chain. * Clear hdr_uio here and hdrlen at the end of the first loop. */ if (hdr_uio != NULL && hdr_uio->uio_resid > 0) { hdr_uio->uio_td = td; hdr_uio->uio_rw = UIO_WRITE; mh = m_uiotombuf(hdr_uio, M_WAITOK, space, 0, 0); hdrlen = m_length(mh, &mhtail); space -= hdrlen; /* * If header consumed all the socket buffer space, * don't waste CPU cycles and jump to the end. */ if (space == 0) { sfio = NULL; nios = 0; goto prepend_header; } hdr_uio = NULL; } if (vp != NULL) { error = vn_lock(vp, LK_SHARED); if (error != 0) goto done; error = VOP_GETATTR(vp, &va, td->td_ucred); if (error != 0 || off >= va.va_size) { VOP_UNLOCK(vp, 0); goto done; } if (va.va_size != obj_size) { if (nbytes == 0) rem += va.va_size - obj_size; else if (offset + nbytes > va.va_size) rem -= (offset + nbytes - va.va_size); obj_size = va.va_size; } } if (space > rem) space = rem; npages = howmany(space + (off & PAGE_MASK), PAGE_SIZE); /* * Calculate maximum allowed number of pages for readahead * at this iteration. If SF_USER_READAHEAD was set, we don't * do any heuristics and use exactly the value supplied by * application. Otherwise, we allow readahead up to "rem". * If application wants more, let it be, but there is no * reason to go above MAXPHYS. Also check against "obj_size", * since vm_pager_has_page() can hint beyond EOF. */ if (flags & SF_USER_READAHEAD) { rhpages = SF_READAHEAD(flags); } else { rhpages = howmany(rem + (off & PAGE_MASK), PAGE_SIZE) - npages; rhpages += SF_READAHEAD(flags); } rhpages = min(howmany(MAXPHYS, PAGE_SIZE), rhpages); rhpages = min(howmany(obj_size - trunc_page(off), PAGE_SIZE) - npages, rhpages); sfio = malloc(sizeof(struct sf_io) + npages * sizeof(vm_page_t), M_TEMP, M_WAITOK); refcount_init(&sfio->nios, 1); sfio->error = 0; nios = sendfile_swapin(obj, sfio, off, space, npages, rhpages, flags); /* * Loop and construct maximum sized mbuf chain to be bulk * dumped into socket buffer. */ pa = sfio->pa; for (int i = 0; i < npages; i++) { struct mbuf *m0; /* * If a page wasn't grabbed successfully, then * trim the array. Can happen only with SF_NODISKIO. */ if (pa[i] == NULL) { SFSTAT_INC(sf_busy); fixspace(npages, i, off, &space); npages = i; softerr = EBUSY; break; } /* * Get a sendfile buf. When allocating the * first buffer for mbuf chain, we usually * wait as long as necessary, but this wait * can be interrupted. For consequent * buffers, do not sleep, since several * threads might exhaust the buffers and then * deadlock. */ sf = sf_buf_alloc(pa[i], m != NULL ? SFB_NOWAIT : SFB_CATCH); if (sf == NULL) { SFSTAT_INC(sf_allocfail); for (int j = i; j < npages; j++) { vm_page_lock(pa[j]); vm_page_unwire(pa[j], PQ_INACTIVE); vm_page_unlock(pa[j]); } if (m == NULL) softerr = ENOBUFS; fixspace(npages, i, off, &space); npages = i; break; } m0 = m_get(M_WAITOK, MT_DATA); m0->m_ext.ext_buf = (char *)sf_buf_kva(sf); m0->m_ext.ext_size = PAGE_SIZE; m0->m_ext.ext_arg1 = sf; m0->m_ext.ext_arg2 = sfs; /* * SF_NOCACHE sets the page as being freed upon send. * However, we ignore it for the last page in 'space', * if the page is truncated, and we got more data to * send (rem > space), or if we have readahead * configured (rhpages > 0). */ if ((flags & SF_NOCACHE) == 0 || (i == npages - 1 && ((off + space) & PAGE_MASK) && (rem > space || rhpages > 0))) m0->m_ext.ext_type = EXT_SFBUF; else m0->m_ext.ext_type = EXT_SFBUF_NOCACHE; m0->m_ext.ext_flags = EXT_FLAG_EMBREF; m0->m_ext.ext_count = 1; m0->m_flags |= (M_EXT | M_RDONLY); if (nios) m0->m_flags |= M_NOTREADY; m0->m_data = (char *)sf_buf_kva(sf) + (vmoff(i, off) & PAGE_MASK); m0->m_len = xfsize(i, npages, off, space); if (i == 0) sfio->m = m0; /* Append to mbuf chain. */ if (mtail != NULL) mtail->m_next = m0; else m = m0; mtail = m0; if (sfs != NULL) { mtx_lock(&sfs->mtx); sfs->count++; mtx_unlock(&sfs->mtx); } } if (vp != NULL) VOP_UNLOCK(vp, 0); /* Keep track of bytes processed. */ off += space; rem -= space; /* Prepend header, if any. */ if (hdrlen) { prepend_header: mhtail->m_next = m; m = mh; mh = NULL; } if (m == NULL) { KASSERT(softerr, ("%s: m NULL, no error", __func__)); error = softerr; free(sfio, M_TEMP); goto done; } /* Add the buffer chain to the socket buffer. */ KASSERT(m_length(m, NULL) == space + hdrlen, ("%s: mlen %u space %d hdrlen %d", __func__, m_length(m, NULL), space, hdrlen)); CURVNET_SET(so->so_vnet); if (nios == 0) { /* * If sendfile_swapin() didn't initiate any I/Os, * which happens if all data is cached in VM, then * we can send data right now without the * PRUS_NOTREADY flag. */ free(sfio, M_TEMP); error = (*so->so_proto->pr_usrreqs->pru_send) (so, 0, m, NULL, NULL, td); } else { sfio->sock_fp = sock_fp; sfio->npages = npages; fhold(sock_fp); error = (*so->so_proto->pr_usrreqs->pru_send) (so, PRUS_NOTREADY, m, NULL, NULL, td); sendfile_iodone(sfio, NULL, 0, 0); } CURVNET_RESTORE(); m = NULL; /* pru_send always consumes */ if (error) goto done; sbytes += space + hdrlen; if (hdrlen) hdrlen = 0; if (softerr) { error = softerr; goto done; } } /* * Send trailers. Wimp out and use writev(2). */ if (trl_uio != NULL) { sbunlock(&so->so_snd); error = kern_writev(td, sockfd, trl_uio); if (error == 0) sbytes += td->td_retval[0]; goto out; } done: sbunlock(&so->so_snd); out: /* * If there was no error we have to clear td->td_retval[0] * because it may have been set by writev. */ if (error == 0) { td->td_retval[0] = 0; } if (sent != NULL) { (*sent) = sbytes; } if (obj != NULL) vm_object_deallocate(obj); if (so) fdrop(sock_fp, td); if (m) m_freem(m); if (mh) m_freem(mh); if (sfs != NULL) { mtx_lock(&sfs->mtx); if (sfs->count != 0) cv_wait(&sfs->cv, &sfs->mtx); KASSERT(sfs->count == 0, ("sendfile sync still busy")); cv_destroy(&sfs->cv); mtx_destroy(&sfs->mtx); free(sfs, M_TEMP); } if (error == ERESTART) error = EINTR; return (error); } static int sendfile(struct thread *td, struct sendfile_args *uap, int compat) { struct sf_hdtr hdtr; struct uio *hdr_uio, *trl_uio; struct file *fp; cap_rights_t rights; off_t sbytes; int error; /* * File offset must be positive. If it goes beyond EOF * we send only the header/trailer and no payload data. */ if (uap->offset < 0) return (EINVAL); hdr_uio = trl_uio = NULL; if (uap->hdtr != NULL) { error = copyin(uap->hdtr, &hdtr, sizeof(hdtr)); if (error != 0) goto out; if (hdtr.headers != NULL) { error = copyinuio(hdtr.headers, hdtr.hdr_cnt, &hdr_uio); if (error != 0) goto out; #ifdef COMPAT_FREEBSD4 /* * In FreeBSD < 5.0 the nbytes to send also included * the header. If compat is specified subtract the * header size from nbytes. */ if (compat) { if (uap->nbytes > hdr_uio->uio_resid) uap->nbytes -= hdr_uio->uio_resid; else uap->nbytes = 0; } #endif } if (hdtr.trailers != NULL) { error = copyinuio(hdtr.trailers, hdtr.trl_cnt, &trl_uio); if (error != 0) goto out; } } AUDIT_ARG_FD(uap->fd); /* * sendfile(2) can start at any offset within a file so we require * CAP_READ+CAP_SEEK = CAP_PREAD. */ if ((error = fget_read(td, uap->fd, cap_rights_init(&rights, CAP_PREAD), &fp)) != 0) { goto out; } error = fo_sendfile(fp, uap->s, hdr_uio, trl_uio, uap->offset, uap->nbytes, &sbytes, uap->flags, td); fdrop(fp, td); if (uap->sbytes != NULL) copyout(&sbytes, uap->sbytes, sizeof(off_t)); out: free(hdr_uio, M_IOV); free(trl_uio, M_IOV); return (error); } /* * sendfile(2) * * int sendfile(int fd, int s, off_t offset, size_t nbytes, * struct sf_hdtr *hdtr, off_t *sbytes, int flags) * * Send a file specified by 'fd' and starting at 'offset' to a socket * specified by 's'. Send only 'nbytes' of the file or until EOF if nbytes == * 0. Optionally add a header and/or trailer to the socket output. If * specified, write the total number of bytes sent into *sbytes. */ int sys_sendfile(struct thread *td, struct sendfile_args *uap) { return (sendfile(td, uap, 0)); } #ifdef COMPAT_FREEBSD4 int freebsd4_sendfile(struct thread *td, struct freebsd4_sendfile_args *uap) { struct sendfile_args args; args.fd = uap->fd; args.s = uap->s; args.offset = uap->offset; args.nbytes = uap->nbytes; args.hdtr = uap->hdtr; args.sbytes = uap->sbytes; args.flags = uap->flags; return (sendfile(td, &args, 1)); } #endif /* COMPAT_FREEBSD4 */ Index: projects/clang400-import/sys/kern/vfs_bio.c =================================================================== --- projects/clang400-import/sys/kern/vfs_bio.c (revision 311459) +++ projects/clang400-import/sys/kern/vfs_bio.c (revision 311460) @@ -1,5002 +1,4990 @@ /*- * Copyright (c) 2004 Poul-Henning Kamp * Copyright (c) 1994,1997 John S. Dyson * Copyright (c) 2013 The FreeBSD Foundation * All rights reserved. * * Portions of this software were developed by Konstantin Belousov * 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. */ /* * this file contains a new buffer I/O scheme implementing a coherent * VM object and buffer cache scheme. Pains have been taken to make * sure that the performance degradation associated with schemes such * as this is not realized. * * Author: John S. Dyson * Significant help during the development and debugging phases * had been provided by David Greenman, also of the FreeBSD core team. * * see man buf(9) for more info. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "opt_compat.h" #include "opt_swap.h" static MALLOC_DEFINE(M_BIOBUF, "biobuf", "BIO buffer"); struct bio_ops bioops; /* I/O operation notification */ struct buf_ops buf_ops_bio = { .bop_name = "buf_ops_bio", .bop_write = bufwrite, .bop_strategy = bufstrategy, .bop_sync = bufsync, .bop_bdflush = bufbdflush, }; static struct buf *buf; /* buffer header pool */ extern struct buf *swbuf; /* Swap buffer header pool. */ caddr_t unmapped_buf; /* Used below and for softdep flushing threads in ufs/ffs/ffs_softdep.c */ struct proc *bufdaemonproc; struct proc *bufspacedaemonproc; static int inmem(struct vnode *vp, daddr_t blkno); static void vm_hold_free_pages(struct buf *bp, int newbsize); static void vm_hold_load_pages(struct buf *bp, vm_offset_t from, vm_offset_t to); static void vfs_page_set_valid(struct buf *bp, vm_ooffset_t off, vm_page_t m); static void vfs_page_set_validclean(struct buf *bp, vm_ooffset_t off, vm_page_t m); static void vfs_clean_pages_dirty_buf(struct buf *bp); static void vfs_setdirty_locked_object(struct buf *bp); static void vfs_vmio_invalidate(struct buf *bp); static void vfs_vmio_truncate(struct buf *bp, int npages); static void vfs_vmio_extend(struct buf *bp, int npages, int size); static int vfs_bio_clcheck(struct vnode *vp, int size, daddr_t lblkno, daddr_t blkno); static int buf_flush(struct vnode *vp, int); static int buf_recycle(bool); static int buf_scan(bool); static int flushbufqueues(struct vnode *, int, int); static void buf_daemon(void); static void bremfreel(struct buf *bp); static __inline void bd_wakeup(void); static int sysctl_runningspace(SYSCTL_HANDLER_ARGS); static void bufkva_reclaim(vmem_t *, int); static void bufkva_free(struct buf *); static int buf_import(void *, void **, int, int); static void buf_release(void *, void **, int); #if defined(COMPAT_FREEBSD4) || defined(COMPAT_FREEBSD5) || \ defined(COMPAT_FREEBSD6) || defined(COMPAT_FREEBSD7) static int sysctl_bufspace(SYSCTL_HANDLER_ARGS); #endif int vmiodirenable = TRUE; SYSCTL_INT(_vfs, OID_AUTO, vmiodirenable, CTLFLAG_RW, &vmiodirenable, 0, "Use the VM system for directory writes"); long runningbufspace; SYSCTL_LONG(_vfs, OID_AUTO, runningbufspace, CTLFLAG_RD, &runningbufspace, 0, "Amount of presently outstanding async buffer io"); static long bufspace; #if defined(COMPAT_FREEBSD4) || defined(COMPAT_FREEBSD5) || \ defined(COMPAT_FREEBSD6) || defined(COMPAT_FREEBSD7) SYSCTL_PROC(_vfs, OID_AUTO, bufspace, CTLTYPE_LONG|CTLFLAG_MPSAFE|CTLFLAG_RD, &bufspace, 0, sysctl_bufspace, "L", "Virtual memory used for buffers"); #else SYSCTL_LONG(_vfs, OID_AUTO, bufspace, CTLFLAG_RD, &bufspace, 0, "Physical memory used for buffers"); #endif static long bufkvaspace; SYSCTL_LONG(_vfs, OID_AUTO, bufkvaspace, CTLFLAG_RD, &bufkvaspace, 0, "Kernel virtual memory used for buffers"); static long maxbufspace; SYSCTL_LONG(_vfs, OID_AUTO, maxbufspace, CTLFLAG_RW, &maxbufspace, 0, "Maximum allowed value of bufspace (including metadata)"); static long bufmallocspace; SYSCTL_LONG(_vfs, OID_AUTO, bufmallocspace, CTLFLAG_RD, &bufmallocspace, 0, "Amount of malloced memory for buffers"); static long maxbufmallocspace; SYSCTL_LONG(_vfs, OID_AUTO, maxmallocbufspace, CTLFLAG_RW, &maxbufmallocspace, 0, "Maximum amount of malloced memory for buffers"); static long lobufspace; SYSCTL_LONG(_vfs, OID_AUTO, lobufspace, CTLFLAG_RW, &lobufspace, 0, "Minimum amount of buffers we want to have"); long hibufspace; SYSCTL_LONG(_vfs, OID_AUTO, hibufspace, CTLFLAG_RW, &hibufspace, 0, "Maximum allowed value of bufspace (excluding metadata)"); long bufspacethresh; SYSCTL_LONG(_vfs, OID_AUTO, bufspacethresh, CTLFLAG_RW, &bufspacethresh, 0, "Bufspace consumed before waking the daemon to free some"); static int buffreekvacnt; SYSCTL_INT(_vfs, OID_AUTO, buffreekvacnt, CTLFLAG_RW, &buffreekvacnt, 0, "Number of times we have freed the KVA space from some buffer"); static int bufdefragcnt; SYSCTL_INT(_vfs, OID_AUTO, bufdefragcnt, CTLFLAG_RW, &bufdefragcnt, 0, "Number of times we have had to repeat buffer allocation to defragment"); static long lorunningspace; SYSCTL_PROC(_vfs, OID_AUTO, lorunningspace, CTLTYPE_LONG | CTLFLAG_MPSAFE | CTLFLAG_RW, &lorunningspace, 0, sysctl_runningspace, "L", "Minimum preferred space used for in-progress I/O"); static long hirunningspace; SYSCTL_PROC(_vfs, OID_AUTO, hirunningspace, CTLTYPE_LONG | CTLFLAG_MPSAFE | CTLFLAG_RW, &hirunningspace, 0, sysctl_runningspace, "L", "Maximum amount of space to use for in-progress I/O"); int dirtybufferflushes; SYSCTL_INT(_vfs, OID_AUTO, dirtybufferflushes, CTLFLAG_RW, &dirtybufferflushes, 0, "Number of bdwrite to bawrite conversions to limit dirty buffers"); int bdwriteskip; SYSCTL_INT(_vfs, OID_AUTO, bdwriteskip, CTLFLAG_RW, &bdwriteskip, 0, "Number of buffers supplied to bdwrite with snapshot deadlock risk"); int altbufferflushes; SYSCTL_INT(_vfs, OID_AUTO, altbufferflushes, CTLFLAG_RW, &altbufferflushes, 0, "Number of fsync flushes to limit dirty buffers"); static int recursiveflushes; SYSCTL_INT(_vfs, OID_AUTO, recursiveflushes, CTLFLAG_RW, &recursiveflushes, 0, "Number of flushes skipped due to being recursive"); static int numdirtybuffers; SYSCTL_INT(_vfs, OID_AUTO, numdirtybuffers, CTLFLAG_RD, &numdirtybuffers, 0, "Number of buffers that are dirty (has unwritten changes) at the moment"); static int lodirtybuffers; SYSCTL_INT(_vfs, OID_AUTO, lodirtybuffers, CTLFLAG_RW, &lodirtybuffers, 0, "How many buffers we want to have free before bufdaemon can sleep"); static int hidirtybuffers; SYSCTL_INT(_vfs, OID_AUTO, hidirtybuffers, CTLFLAG_RW, &hidirtybuffers, 0, "When the number of dirty buffers is considered severe"); int dirtybufthresh; SYSCTL_INT(_vfs, OID_AUTO, dirtybufthresh, CTLFLAG_RW, &dirtybufthresh, 0, "Number of bdwrite to bawrite conversions to clear dirty buffers"); static int numfreebuffers; SYSCTL_INT(_vfs, OID_AUTO, numfreebuffers, CTLFLAG_RD, &numfreebuffers, 0, "Number of free buffers"); static int lofreebuffers; SYSCTL_INT(_vfs, OID_AUTO, lofreebuffers, CTLFLAG_RW, &lofreebuffers, 0, "Target number of free buffers"); static int hifreebuffers; SYSCTL_INT(_vfs, OID_AUTO, hifreebuffers, CTLFLAG_RW, &hifreebuffers, 0, "Threshold for clean buffer recycling"); static int getnewbufcalls; SYSCTL_INT(_vfs, OID_AUTO, getnewbufcalls, CTLFLAG_RW, &getnewbufcalls, 0, "Number of calls to getnewbuf"); static int getnewbufrestarts; SYSCTL_INT(_vfs, OID_AUTO, getnewbufrestarts, CTLFLAG_RW, &getnewbufrestarts, 0, "Number of times getnewbuf has had to restart a buffer acquisition"); static int mappingrestarts; SYSCTL_INT(_vfs, OID_AUTO, mappingrestarts, CTLFLAG_RW, &mappingrestarts, 0, "Number of times getblk has had to restart a buffer mapping for " "unmapped buffer"); static int numbufallocfails; SYSCTL_INT(_vfs, OID_AUTO, numbufallocfails, CTLFLAG_RW, &numbufallocfails, 0, "Number of times buffer allocations failed"); static int flushbufqtarget = 100; SYSCTL_INT(_vfs, OID_AUTO, flushbufqtarget, CTLFLAG_RW, &flushbufqtarget, 0, "Amount of work to do in flushbufqueues when helping bufdaemon"); static long notbufdflushes; SYSCTL_LONG(_vfs, OID_AUTO, notbufdflushes, CTLFLAG_RD, ¬bufdflushes, 0, "Number of dirty buffer flushes done by the bufdaemon helpers"); static long barrierwrites; SYSCTL_LONG(_vfs, OID_AUTO, barrierwrites, CTLFLAG_RW, &barrierwrites, 0, "Number of barrier writes"); SYSCTL_INT(_vfs, OID_AUTO, unmapped_buf_allowed, CTLFLAG_RD, &unmapped_buf_allowed, 0, "Permit the use of the unmapped i/o"); /* * This lock synchronizes access to bd_request. */ static struct mtx_padalign bdlock; /* * This lock protects the runningbufreq and synchronizes runningbufwakeup and * waitrunningbufspace(). */ static struct mtx_padalign rbreqlock; /* * Lock that protects needsbuffer and the sleeps/wakeups surrounding it. */ static struct rwlock_padalign nblock; /* * Lock that protects bdirtywait. */ static struct mtx_padalign bdirtylock; /* * Wakeup point for bufdaemon, as well as indicator of whether it is already * active. Set to 1 when the bufdaemon is already "on" the queue, 0 when it * is idling. */ static int bd_request; /* * Request/wakeup point for the bufspace daemon. */ static int bufspace_request; /* * Request for the buf daemon to write more buffers than is indicated by * lodirtybuf. This may be necessary to push out excess dependencies or * defragment the address space where a simple count of the number of dirty * buffers is insufficient to characterize the demand for flushing them. */ static int bd_speedupreq; /* - * bogus page -- for I/O to/from partially complete buffers - * this is a temporary solution to the problem, but it is not - * really that bad. it would be better to split the buffer - * for input in the case of buffers partially already in memory, - * but the code is intricate enough already. - */ -vm_page_t bogus_page; - -/* * Synchronization (sleep/wakeup) variable for active buffer space requests. * Set when wait starts, cleared prior to wakeup(). * Used in runningbufwakeup() and waitrunningbufspace(). */ static int runningbufreq; /* * Synchronization (sleep/wakeup) variable for buffer requests. * Can contain the VFS_BIO_NEED flags defined below; setting/clearing is done * by and/or. * Used in numdirtywakeup(), bufspace_wakeup(), bwillwrite(), * getnewbuf(), and getblk(). */ static volatile int needsbuffer; /* * Synchronization for bwillwrite() waiters. */ static int bdirtywait; /* * Definitions for the buffer free lists. */ #define QUEUE_NONE 0 /* on no queue */ #define QUEUE_EMPTY 1 /* empty buffer headers */ #define QUEUE_DIRTY 2 /* B_DELWRI buffers */ #define QUEUE_CLEAN 3 /* non-B_DELWRI buffers */ #define QUEUE_SENTINEL 1024 /* not an queue index, but mark for sentinel */ /* Maximum number of clean buffer queues. */ #define CLEAN_QUEUES 16 /* Configured number of clean queues. */ static int clean_queues; /* Maximum number of buffer queues. */ #define BUFFER_QUEUES (QUEUE_CLEAN + CLEAN_QUEUES) /* Queues for free buffers with various properties */ static TAILQ_HEAD(bqueues, buf) bufqueues[BUFFER_QUEUES] = { { 0 } }; #ifdef INVARIANTS static int bq_len[BUFFER_QUEUES]; #endif /* * Lock for each bufqueue */ static struct mtx_padalign bqlocks[BUFFER_QUEUES]; /* * per-cpu empty buffer cache. */ uma_zone_t buf_zone; /* * Single global constant for BUF_WMESG, to avoid getting multiple references. * buf_wmesg is referred from macros. */ const char *buf_wmesg = BUF_WMESG; static int sysctl_runningspace(SYSCTL_HANDLER_ARGS) { long value; int error; value = *(long *)arg1; error = sysctl_handle_long(oidp, &value, 0, req); if (error != 0 || req->newptr == NULL) return (error); mtx_lock(&rbreqlock); if (arg1 == &hirunningspace) { if (value < lorunningspace) error = EINVAL; else hirunningspace = value; } else { KASSERT(arg1 == &lorunningspace, ("%s: unknown arg1", __func__)); if (value > hirunningspace) error = EINVAL; else lorunningspace = value; } mtx_unlock(&rbreqlock); return (error); } #if defined(COMPAT_FREEBSD4) || defined(COMPAT_FREEBSD5) || \ defined(COMPAT_FREEBSD6) || defined(COMPAT_FREEBSD7) static int sysctl_bufspace(SYSCTL_HANDLER_ARGS) { long lvalue; int ivalue; if (sizeof(int) == sizeof(long) || req->oldlen >= sizeof(long)) return (sysctl_handle_long(oidp, arg1, arg2, req)); lvalue = *(long *)arg1; if (lvalue > INT_MAX) /* On overflow, still write out a long to trigger ENOMEM. */ return (sysctl_handle_long(oidp, &lvalue, 0, req)); ivalue = lvalue; return (sysctl_handle_int(oidp, &ivalue, 0, req)); } #endif static int bqcleanq(void) { static int nextq; return ((atomic_fetchadd_int(&nextq, 1) % clean_queues) + QUEUE_CLEAN); } static int bqisclean(int qindex) { return (qindex >= QUEUE_CLEAN && qindex < QUEUE_CLEAN + CLEAN_QUEUES); } /* * bqlock: * * Return the appropriate queue lock based on the index. */ static inline struct mtx * bqlock(int qindex) { return (struct mtx *)&bqlocks[qindex]; } /* * bdirtywakeup: * * Wakeup any bwillwrite() waiters. */ static void bdirtywakeup(void) { mtx_lock(&bdirtylock); if (bdirtywait) { bdirtywait = 0; wakeup(&bdirtywait); } mtx_unlock(&bdirtylock); } /* * bdirtysub: * * Decrement the numdirtybuffers count by one and wakeup any * threads blocked in bwillwrite(). */ static void bdirtysub(void) { if (atomic_fetchadd_int(&numdirtybuffers, -1) == (lodirtybuffers + hidirtybuffers) / 2) bdirtywakeup(); } /* * bdirtyadd: * * Increment the numdirtybuffers count by one and wakeup the buf * daemon if needed. */ static void bdirtyadd(void) { /* * Only do the wakeup once as we cross the boundary. The * buf daemon will keep running until the condition clears. */ if (atomic_fetchadd_int(&numdirtybuffers, 1) == (lodirtybuffers + hidirtybuffers) / 2) bd_wakeup(); } /* * bufspace_wakeup: * * Called when buffer space is potentially available for recovery. * getnewbuf() will block on this flag when it is unable to free * sufficient buffer space. Buffer space becomes recoverable when * bp's get placed back in the queues. */ static void bufspace_wakeup(void) { /* * If someone is waiting for bufspace, wake them up. * * Since needsbuffer is set prior to doing an additional queue * scan it is safe to check for the flag prior to acquiring the * lock. The thread that is preparing to scan again before * blocking would discover the buf we released. */ if (needsbuffer) { rw_rlock(&nblock); if (atomic_cmpset_int(&needsbuffer, 1, 0) == 1) wakeup(__DEVOLATILE(void *, &needsbuffer)); rw_runlock(&nblock); } } /* * bufspace_daemonwakeup: * * Wakeup the daemon responsible for freeing clean bufs. */ static void bufspace_daemonwakeup(void) { rw_rlock(&nblock); if (bufspace_request == 0) { bufspace_request = 1; wakeup(&bufspace_request); } rw_runlock(&nblock); } /* * bufspace_adjust: * * Adjust the reported bufspace for a KVA managed buffer, possibly * waking any waiters. */ static void bufspace_adjust(struct buf *bp, int bufsize) { long space; int diff; KASSERT((bp->b_flags & B_MALLOC) == 0, ("bufspace_adjust: malloc buf %p", bp)); diff = bufsize - bp->b_bufsize; if (diff < 0) { atomic_subtract_long(&bufspace, -diff); bufspace_wakeup(); } else { space = atomic_fetchadd_long(&bufspace, diff); /* Wake up the daemon on the transition. */ if (space < bufspacethresh && space + diff >= bufspacethresh) bufspace_daemonwakeup(); } bp->b_bufsize = bufsize; } /* * bufspace_reserve: * * Reserve bufspace before calling allocbuf(). metadata has a * different space limit than data. */ static int bufspace_reserve(int size, bool metadata) { long limit; long space; if (metadata) limit = maxbufspace; else limit = hibufspace; do { space = bufspace; if (space + size > limit) return (ENOSPC); } while (atomic_cmpset_long(&bufspace, space, space + size) == 0); /* Wake up the daemon on the transition. */ if (space < bufspacethresh && space + size >= bufspacethresh) bufspace_daemonwakeup(); return (0); } /* * bufspace_release: * * Release reserved bufspace after bufspace_adjust() has consumed it. */ static void bufspace_release(int size) { atomic_subtract_long(&bufspace, size); bufspace_wakeup(); } /* * bufspace_wait: * * Wait for bufspace, acting as the buf daemon if a locked vnode is * supplied. needsbuffer must be set in a safe fashion prior to * polling for space. The operation must be re-tried on return. */ static void bufspace_wait(struct vnode *vp, int gbflags, int slpflag, int slptimeo) { struct thread *td; int error, fl, norunbuf; if ((gbflags & GB_NOWAIT_BD) != 0) return; td = curthread; rw_wlock(&nblock); while (needsbuffer != 0) { if (vp != NULL && vp->v_type != VCHR && (td->td_pflags & TDP_BUFNEED) == 0) { rw_wunlock(&nblock); /* * getblk() is called with a vnode locked, and * some majority of the dirty buffers may as * well belong to the vnode. Flushing the * buffers there would make a progress that * cannot be achieved by the buf_daemon, that * cannot lock the vnode. */ norunbuf = ~(TDP_BUFNEED | TDP_NORUNNINGBUF) | (td->td_pflags & TDP_NORUNNINGBUF); /* * Play bufdaemon. The getnewbuf() function * may be called while the thread owns lock * for another dirty buffer for the same * vnode, which makes it impossible to use * VOP_FSYNC() there, due to the buffer lock * recursion. */ td->td_pflags |= TDP_BUFNEED | TDP_NORUNNINGBUF; fl = buf_flush(vp, flushbufqtarget); td->td_pflags &= norunbuf; rw_wlock(&nblock); if (fl != 0) continue; if (needsbuffer == 0) break; } error = rw_sleep(__DEVOLATILE(void *, &needsbuffer), &nblock, (PRIBIO + 4) | slpflag, "newbuf", slptimeo); if (error != 0) break; } rw_wunlock(&nblock); } /* * bufspace_daemon: * * buffer space management daemon. Tries to maintain some marginal * amount of free buffer space so that requesting processes neither * block nor work to reclaim buffers. */ static void bufspace_daemon(void) { for (;;) { kproc_suspend_check(bufspacedaemonproc); /* * Free buffers from the clean queue until we meet our * targets. * * Theory of operation: The buffer cache is most efficient * when some free buffer headers and space are always * available to getnewbuf(). This daemon attempts to prevent * the excessive blocking and synchronization associated * with shortfall. It goes through three phases according * demand: * * 1) The daemon wakes up voluntarily once per-second * during idle periods when the counters are below * the wakeup thresholds (bufspacethresh, lofreebuffers). * * 2) The daemon wakes up as we cross the thresholds * ahead of any potential blocking. This may bounce * slightly according to the rate of consumption and * release. * * 3) The daemon and consumers are starved for working * clean buffers. This is the 'bufspace' sleep below * which will inefficiently trade bufs with bqrelse * until we return to condition 2. */ while (bufspace > lobufspace || numfreebuffers < hifreebuffers) { if (buf_recycle(false) != 0) { atomic_set_int(&needsbuffer, 1); if (buf_recycle(false) != 0) { rw_wlock(&nblock); if (needsbuffer) rw_sleep(__DEVOLATILE(void *, &needsbuffer), &nblock, PRIBIO|PDROP, "bufspace", hz/10); else rw_wunlock(&nblock); } } maybe_yield(); } /* * Re-check our limits under the exclusive nblock. */ rw_wlock(&nblock); if (bufspace < bufspacethresh && numfreebuffers > lofreebuffers) { bufspace_request = 0; rw_sleep(&bufspace_request, &nblock, PRIBIO|PDROP, "-", hz); } else rw_wunlock(&nblock); } } static struct kproc_desc bufspace_kp = { "bufspacedaemon", bufspace_daemon, &bufspacedaemonproc }; SYSINIT(bufspacedaemon, SI_SUB_KTHREAD_BUF, SI_ORDER_FIRST, kproc_start, &bufspace_kp); /* * bufmallocadjust: * * Adjust the reported bufspace for a malloc managed buffer, possibly * waking any waiters. */ static void bufmallocadjust(struct buf *bp, int bufsize) { int diff; KASSERT((bp->b_flags & B_MALLOC) != 0, ("bufmallocadjust: non-malloc buf %p", bp)); diff = bufsize - bp->b_bufsize; if (diff < 0) atomic_subtract_long(&bufmallocspace, -diff); else atomic_add_long(&bufmallocspace, diff); bp->b_bufsize = bufsize; } /* * runningwakeup: * * Wake up processes that are waiting on asynchronous writes to fall * below lorunningspace. */ static void runningwakeup(void) { mtx_lock(&rbreqlock); if (runningbufreq) { runningbufreq = 0; wakeup(&runningbufreq); } mtx_unlock(&rbreqlock); } /* * runningbufwakeup: * * Decrement the outstanding write count according. */ void runningbufwakeup(struct buf *bp) { long space, bspace; bspace = bp->b_runningbufspace; if (bspace == 0) return; space = atomic_fetchadd_long(&runningbufspace, -bspace); KASSERT(space >= bspace, ("runningbufspace underflow %ld %ld", space, bspace)); bp->b_runningbufspace = 0; /* * Only acquire the lock and wakeup on the transition from exceeding * the threshold to falling below it. */ if (space < lorunningspace) return; if (space - bspace > lorunningspace) return; runningwakeup(); } /* * waitrunningbufspace() * * runningbufspace is a measure of the amount of I/O currently * running. This routine is used in async-write situations to * prevent creating huge backups of pending writes to a device. * Only asynchronous writes are governed by this function. * * This does NOT turn an async write into a sync write. It waits * for earlier writes to complete and generally returns before the * caller's write has reached the device. */ void waitrunningbufspace(void) { mtx_lock(&rbreqlock); while (runningbufspace > hirunningspace) { runningbufreq = 1; msleep(&runningbufreq, &rbreqlock, PVM, "wdrain", 0); } mtx_unlock(&rbreqlock); } /* * vfs_buf_test_cache: * * Called when a buffer is extended. This function clears the B_CACHE * bit if the newly extended portion of the buffer does not contain * valid data. */ static __inline void vfs_buf_test_cache(struct buf *bp, vm_ooffset_t foff, vm_offset_t off, vm_offset_t size, vm_page_t m) { VM_OBJECT_ASSERT_LOCKED(m->object); if (bp->b_flags & B_CACHE) { int base = (foff + off) & PAGE_MASK; if (vm_page_is_valid(m, base, size) == 0) bp->b_flags &= ~B_CACHE; } } /* Wake up the buffer daemon if necessary */ static __inline void bd_wakeup(void) { mtx_lock(&bdlock); if (bd_request == 0) { bd_request = 1; wakeup(&bd_request); } mtx_unlock(&bdlock); } /* * bd_speedup - speedup the buffer cache flushing code */ void bd_speedup(void) { int needwake; mtx_lock(&bdlock); needwake = 0; if (bd_speedupreq == 0 || bd_request == 0) needwake = 1; bd_speedupreq = 1; bd_request = 1; if (needwake) wakeup(&bd_request); mtx_unlock(&bdlock); } #ifndef NSWBUF_MIN #define NSWBUF_MIN 16 #endif #ifdef __i386__ #define TRANSIENT_DENOM 5 #else #define TRANSIENT_DENOM 10 #endif /* * Calculating buffer cache scaling values and reserve space for buffer * headers. This is called during low level kernel initialization and * may be called more then once. We CANNOT write to the memory area * being reserved at this time. */ caddr_t kern_vfs_bio_buffer_alloc(caddr_t v, long physmem_est) { int tuned_nbuf; long maxbuf, maxbuf_sz, buf_sz, biotmap_sz; /* * physmem_est is in pages. Convert it to kilobytes (assumes * PAGE_SIZE is >= 1K) */ physmem_est = physmem_est * (PAGE_SIZE / 1024); /* * The nominal buffer size (and minimum KVA allocation) is BKVASIZE. * For the first 64MB of ram nominally allocate sufficient buffers to * cover 1/4 of our ram. Beyond the first 64MB allocate additional * buffers to cover 1/10 of our ram over 64MB. When auto-sizing * the buffer cache we limit the eventual kva reservation to * maxbcache bytes. * * factor represents the 1/4 x ram conversion. */ if (nbuf == 0) { int factor = 4 * BKVASIZE / 1024; nbuf = 50; if (physmem_est > 4096) nbuf += min((physmem_est - 4096) / factor, 65536 / factor); if (physmem_est > 65536) nbuf += min((physmem_est - 65536) * 2 / (factor * 5), 32 * 1024 * 1024 / (factor * 5)); if (maxbcache && nbuf > maxbcache / BKVASIZE) nbuf = maxbcache / BKVASIZE; tuned_nbuf = 1; } else tuned_nbuf = 0; /* XXX Avoid unsigned long overflows later on with maxbufspace. */ maxbuf = (LONG_MAX / 3) / BKVASIZE; if (nbuf > maxbuf) { if (!tuned_nbuf) printf("Warning: nbufs lowered from %d to %ld\n", nbuf, maxbuf); nbuf = maxbuf; } /* * Ideal allocation size for the transient bio submap is 10% * of the maximal space buffer map. This roughly corresponds * to the amount of the buffer mapped for typical UFS load. * * Clip the buffer map to reserve space for the transient * BIOs, if its extent is bigger than 90% (80% on i386) of the * maximum buffer map extent on the platform. * * The fall-back to the maxbuf in case of maxbcache unset, * allows to not trim the buffer KVA for the architectures * with ample KVA space. */ if (bio_transient_maxcnt == 0 && unmapped_buf_allowed) { maxbuf_sz = maxbcache != 0 ? maxbcache : maxbuf * BKVASIZE; buf_sz = (long)nbuf * BKVASIZE; if (buf_sz < maxbuf_sz / TRANSIENT_DENOM * (TRANSIENT_DENOM - 1)) { /* * There is more KVA than memory. Do not * adjust buffer map size, and assign the rest * of maxbuf to transient map. */ biotmap_sz = maxbuf_sz - buf_sz; } else { /* * Buffer map spans all KVA we could afford on * this platform. Give 10% (20% on i386) of * the buffer map to the transient bio map. */ biotmap_sz = buf_sz / TRANSIENT_DENOM; buf_sz -= biotmap_sz; } if (biotmap_sz / INT_MAX > MAXPHYS) bio_transient_maxcnt = INT_MAX; else bio_transient_maxcnt = biotmap_sz / MAXPHYS; /* * Artificially limit to 1024 simultaneous in-flight I/Os * using the transient mapping. */ if (bio_transient_maxcnt > 1024) bio_transient_maxcnt = 1024; if (tuned_nbuf) nbuf = buf_sz / BKVASIZE; } /* * swbufs are used as temporary holders for I/O, such as paging I/O. * We have no less then 16 and no more then 256. */ nswbuf = min(nbuf / 4, 256); TUNABLE_INT_FETCH("kern.nswbuf", &nswbuf); if (nswbuf < NSWBUF_MIN) nswbuf = NSWBUF_MIN; /* * Reserve space for the buffer cache buffers */ swbuf = (void *)v; v = (caddr_t)(swbuf + nswbuf); buf = (void *)v; v = (caddr_t)(buf + nbuf); return(v); } /* Initialize the buffer subsystem. Called before use of any buffers. */ void bufinit(void) { struct buf *bp; int i; CTASSERT(MAXBCACHEBUF >= MAXBSIZE); mtx_init(&bqlocks[QUEUE_DIRTY], "bufq dirty lock", NULL, MTX_DEF); mtx_init(&bqlocks[QUEUE_EMPTY], "bufq empty lock", NULL, MTX_DEF); for (i = QUEUE_CLEAN; i < QUEUE_CLEAN + CLEAN_QUEUES; i++) mtx_init(&bqlocks[i], "bufq clean lock", NULL, MTX_DEF); mtx_init(&rbreqlock, "runningbufspace lock", NULL, MTX_DEF); rw_init(&nblock, "needsbuffer lock"); mtx_init(&bdlock, "buffer daemon lock", NULL, MTX_DEF); mtx_init(&bdirtylock, "dirty buf lock", NULL, MTX_DEF); /* next, make a null set of free lists */ for (i = 0; i < BUFFER_QUEUES; i++) TAILQ_INIT(&bufqueues[i]); unmapped_buf = (caddr_t)kva_alloc(MAXPHYS); /* finally, initialize each buffer header and stick on empty q */ for (i = 0; i < nbuf; i++) { bp = &buf[i]; bzero(bp, sizeof *bp); bp->b_flags = B_INVAL; bp->b_rcred = NOCRED; bp->b_wcred = NOCRED; bp->b_qindex = QUEUE_EMPTY; bp->b_xflags = 0; bp->b_data = bp->b_kvabase = unmapped_buf; LIST_INIT(&bp->b_dep); BUF_LOCKINIT(bp); TAILQ_INSERT_TAIL(&bufqueues[QUEUE_EMPTY], bp, b_freelist); #ifdef INVARIANTS bq_len[QUEUE_EMPTY]++; #endif } /* * maxbufspace is the absolute maximum amount of buffer space we are * allowed to reserve in KVM and in real terms. The absolute maximum * is nominally used by metadata. hibufspace is the nominal maximum * used by most other requests. The differential is required to * ensure that metadata deadlocks don't occur. * * maxbufspace is based on BKVASIZE. Allocating buffers larger then * this may result in KVM fragmentation which is not handled optimally * by the system. XXX This is less true with vmem. We could use * PAGE_SIZE. */ maxbufspace = (long)nbuf * BKVASIZE; hibufspace = lmax(3 * maxbufspace / 4, maxbufspace - MAXBCACHEBUF * 10); lobufspace = (hibufspace / 20) * 19; /* 95% */ bufspacethresh = lobufspace + (hibufspace - lobufspace) / 2; /* * Note: The 16 MiB upper limit for hirunningspace was chosen * arbitrarily and may need further tuning. It corresponds to * 128 outstanding write IO requests (if IO size is 128 KiB), * which fits with many RAID controllers' tagged queuing limits. * The lower 1 MiB limit is the historical upper limit for * hirunningspace. */ hirunningspace = lmax(lmin(roundup(hibufspace / 64, MAXBCACHEBUF), 16 * 1024 * 1024), 1024 * 1024); lorunningspace = roundup((hirunningspace * 2) / 3, MAXBCACHEBUF); /* * Limit the amount of malloc memory since it is wired permanently into * the kernel space. Even though this is accounted for in the buffer * allocation, we don't want the malloced region to grow uncontrolled. * The malloc scheme improves memory utilization significantly on * average (small) directories. */ maxbufmallocspace = hibufspace / 20; /* * Reduce the chance of a deadlock occurring by limiting the number * of delayed-write dirty buffers we allow to stack up. */ hidirtybuffers = nbuf / 4 + 20; dirtybufthresh = hidirtybuffers * 9 / 10; numdirtybuffers = 0; /* * To support extreme low-memory systems, make sure hidirtybuffers * cannot eat up all available buffer space. This occurs when our * minimum cannot be met. We try to size hidirtybuffers to 3/4 our * buffer space assuming BKVASIZE'd buffers. */ while ((long)hidirtybuffers * BKVASIZE > 3 * hibufspace / 4) { hidirtybuffers >>= 1; } lodirtybuffers = hidirtybuffers / 2; /* * lofreebuffers should be sufficient to avoid stalling waiting on * buf headers under heavy utilization. The bufs in per-cpu caches * are counted as free but will be unavailable to threads executing * on other cpus. * * hifreebuffers is the free target for the bufspace daemon. This * should be set appropriately to limit work per-iteration. */ lofreebuffers = MIN((nbuf / 25) + (20 * mp_ncpus), 128 * mp_ncpus); hifreebuffers = (3 * lofreebuffers) / 2; numfreebuffers = nbuf; - - bogus_page = vm_page_alloc(NULL, 0, VM_ALLOC_NOOBJ | - VM_ALLOC_NORMAL | VM_ALLOC_WIRED); /* Setup the kva and free list allocators. */ vmem_set_reclaim(buffer_arena, bufkva_reclaim); buf_zone = uma_zcache_create("buf free cache", sizeof(struct buf), NULL, NULL, NULL, NULL, buf_import, buf_release, NULL, 0); /* * Size the clean queue according to the amount of buffer space. * One queue per-256mb up to the max. More queues gives better * concurrency but less accurate LRU. */ clean_queues = MIN(howmany(maxbufspace, 256*1024*1024), CLEAN_QUEUES); } #ifdef INVARIANTS static inline void vfs_buf_check_mapped(struct buf *bp) { KASSERT(bp->b_kvabase != unmapped_buf, ("mapped buf: b_kvabase was not updated %p", bp)); KASSERT(bp->b_data != unmapped_buf, ("mapped buf: b_data was not updated %p", bp)); KASSERT(bp->b_data < unmapped_buf || bp->b_data >= unmapped_buf + MAXPHYS, ("b_data + b_offset unmapped %p", bp)); } static inline void vfs_buf_check_unmapped(struct buf *bp) { KASSERT(bp->b_data == unmapped_buf, ("unmapped buf: corrupted b_data %p", bp)); } #define BUF_CHECK_MAPPED(bp) vfs_buf_check_mapped(bp) #define BUF_CHECK_UNMAPPED(bp) vfs_buf_check_unmapped(bp) #else #define BUF_CHECK_MAPPED(bp) do {} while (0) #define BUF_CHECK_UNMAPPED(bp) do {} while (0) #endif static int isbufbusy(struct buf *bp) { if (((bp->b_flags & B_INVAL) == 0 && BUF_ISLOCKED(bp)) || ((bp->b_flags & (B_DELWRI | B_INVAL)) == B_DELWRI)) return (1); return (0); } /* * Shutdown the system cleanly to prepare for reboot, halt, or power off. */ void bufshutdown(int show_busybufs) { static int first_buf_printf = 1; struct buf *bp; int iter, nbusy, pbusy; #ifndef PREEMPTION int subiter; #endif /* * Sync filesystems for shutdown */ wdog_kern_pat(WD_LASTVAL); sys_sync(curthread, NULL); /* * With soft updates, some buffers that are * written will be remarked as dirty until other * buffers are written. */ for (iter = pbusy = 0; iter < 20; iter++) { nbusy = 0; for (bp = &buf[nbuf]; --bp >= buf; ) if (isbufbusy(bp)) nbusy++; if (nbusy == 0) { if (first_buf_printf) printf("All buffers synced."); break; } if (first_buf_printf) { printf("Syncing disks, buffers remaining... "); first_buf_printf = 0; } printf("%d ", nbusy); if (nbusy < pbusy) iter = 0; pbusy = nbusy; wdog_kern_pat(WD_LASTVAL); sys_sync(curthread, NULL); #ifdef PREEMPTION /* * Drop Giant and spin for a while to allow * interrupt threads to run. */ DROP_GIANT(); DELAY(50000 * iter); PICKUP_GIANT(); #else /* * Drop Giant and context switch several times to * allow interrupt threads to run. */ DROP_GIANT(); for (subiter = 0; subiter < 50 * iter; subiter++) { thread_lock(curthread); mi_switch(SW_VOL, NULL); thread_unlock(curthread); DELAY(1000); } PICKUP_GIANT(); #endif } printf("\n"); /* * Count only busy local buffers to prevent forcing * a fsck if we're just a client of a wedged NFS server */ nbusy = 0; for (bp = &buf[nbuf]; --bp >= buf; ) { if (isbufbusy(bp)) { #if 0 /* XXX: This is bogus. We should probably have a BO_REMOTE flag instead */ if (bp->b_dev == NULL) { TAILQ_REMOVE(&mountlist, bp->b_vp->v_mount, mnt_list); continue; } #endif nbusy++; if (show_busybufs > 0) { printf( "%d: buf:%p, vnode:%p, flags:%0x, blkno:%jd, lblkno:%jd, buflock:", nbusy, bp, bp->b_vp, bp->b_flags, (intmax_t)bp->b_blkno, (intmax_t)bp->b_lblkno); BUF_LOCKPRINTINFO(bp); if (show_busybufs > 1) vn_printf(bp->b_vp, "vnode content: "); } } } if (nbusy) { /* * Failed to sync all blocks. Indicate this and don't * unmount filesystems (thus forcing an fsck on reboot). */ printf("Giving up on %d buffers\n", nbusy); DELAY(5000000); /* 5 seconds */ } else { if (!first_buf_printf) printf("Final sync complete\n"); /* * Unmount filesystems */ if (panicstr == NULL) vfs_unmountall(); } swapoff_all(); DELAY(100000); /* wait for console output to finish */ } static void bpmap_qenter(struct buf *bp) { BUF_CHECK_MAPPED(bp); /* * bp->b_data is relative to bp->b_offset, but * bp->b_offset may be offset into the first page. */ bp->b_data = (caddr_t)trunc_page((vm_offset_t)bp->b_data); pmap_qenter((vm_offset_t)bp->b_data, bp->b_pages, bp->b_npages); bp->b_data = (caddr_t)((vm_offset_t)bp->b_data | (vm_offset_t)(bp->b_offset & PAGE_MASK)); } /* * binsfree: * * Insert the buffer into the appropriate free list. */ static void binsfree(struct buf *bp, int qindex) { struct mtx *olock, *nlock; if (qindex != QUEUE_EMPTY) { BUF_ASSERT_XLOCKED(bp); } /* * Stick to the same clean queue for the lifetime of the buf to * limit locking below. Otherwise pick ont sequentially. */ if (qindex == QUEUE_CLEAN) { if (bqisclean(bp->b_qindex)) qindex = bp->b_qindex; else qindex = bqcleanq(); } /* * Handle delayed bremfree() processing. */ nlock = bqlock(qindex); if (bp->b_flags & B_REMFREE) { olock = bqlock(bp->b_qindex); mtx_lock(olock); bremfreel(bp); if (olock != nlock) { mtx_unlock(olock); mtx_lock(nlock); } } else mtx_lock(nlock); if (bp->b_qindex != QUEUE_NONE) panic("binsfree: free buffer onto another queue???"); bp->b_qindex = qindex; if (bp->b_flags & B_AGE) TAILQ_INSERT_HEAD(&bufqueues[bp->b_qindex], bp, b_freelist); else TAILQ_INSERT_TAIL(&bufqueues[bp->b_qindex], bp, b_freelist); #ifdef INVARIANTS bq_len[bp->b_qindex]++; #endif mtx_unlock(nlock); } /* * buf_free: * * Free a buffer to the buf zone once it no longer has valid contents. */ static void buf_free(struct buf *bp) { if (bp->b_flags & B_REMFREE) bremfreef(bp); if (bp->b_vflags & BV_BKGRDINPROG) panic("losing buffer 1"); if (bp->b_rcred != NOCRED) { crfree(bp->b_rcred); bp->b_rcred = NOCRED; } if (bp->b_wcred != NOCRED) { crfree(bp->b_wcred); bp->b_wcred = NOCRED; } if (!LIST_EMPTY(&bp->b_dep)) buf_deallocate(bp); bufkva_free(bp); BUF_UNLOCK(bp); uma_zfree(buf_zone, bp); atomic_add_int(&numfreebuffers, 1); bufspace_wakeup(); } /* * buf_import: * * Import bufs into the uma cache from the buf list. The system still * expects a static array of bufs and much of the synchronization * around bufs assumes type stable storage. As a result, UMA is used * only as a per-cpu cache of bufs still maintained on a global list. */ static int buf_import(void *arg, void **store, int cnt, int flags) { struct buf *bp; int i; mtx_lock(&bqlocks[QUEUE_EMPTY]); for (i = 0; i < cnt; i++) { bp = TAILQ_FIRST(&bufqueues[QUEUE_EMPTY]); if (bp == NULL) break; bremfreel(bp); store[i] = bp; } mtx_unlock(&bqlocks[QUEUE_EMPTY]); return (i); } /* * buf_release: * * Release bufs from the uma cache back to the buffer queues. */ static void buf_release(void *arg, void **store, int cnt) { int i; for (i = 0; i < cnt; i++) binsfree(store[i], QUEUE_EMPTY); } /* * buf_alloc: * * Allocate an empty buffer header. */ static struct buf * buf_alloc(void) { struct buf *bp; bp = uma_zalloc(buf_zone, M_NOWAIT); if (bp == NULL) { bufspace_daemonwakeup(); atomic_add_int(&numbufallocfails, 1); return (NULL); } /* * Wake-up the bufspace daemon on transition. */ if (atomic_fetchadd_int(&numfreebuffers, -1) == lofreebuffers) bufspace_daemonwakeup(); if (BUF_LOCK(bp, LK_EXCLUSIVE | LK_NOWAIT, NULL) != 0) panic("getnewbuf_empty: Locked buf %p on free queue.", bp); KASSERT(bp->b_vp == NULL, ("bp: %p still has vnode %p.", bp, bp->b_vp)); KASSERT((bp->b_flags & (B_DELWRI | B_NOREUSE)) == 0, ("invalid buffer %p flags %#x", bp, bp->b_flags)); KASSERT((bp->b_xflags & (BX_VNCLEAN|BX_VNDIRTY)) == 0, ("bp: %p still on a buffer list. xflags %X", bp, bp->b_xflags)); KASSERT(bp->b_npages == 0, ("bp: %p still has %d vm pages\n", bp, bp->b_npages)); KASSERT(bp->b_kvasize == 0, ("bp: %p still has kva\n", bp)); KASSERT(bp->b_bufsize == 0, ("bp: %p still has bufspace\n", bp)); bp->b_flags = 0; bp->b_ioflags = 0; bp->b_xflags = 0; bp->b_vflags = 0; bp->b_vp = NULL; bp->b_blkno = bp->b_lblkno = 0; bp->b_offset = NOOFFSET; bp->b_iodone = 0; bp->b_error = 0; bp->b_resid = 0; bp->b_bcount = 0; bp->b_npages = 0; bp->b_dirtyoff = bp->b_dirtyend = 0; bp->b_bufobj = NULL; bp->b_data = bp->b_kvabase = unmapped_buf; bp->b_fsprivate1 = NULL; bp->b_fsprivate2 = NULL; bp->b_fsprivate3 = NULL; LIST_INIT(&bp->b_dep); return (bp); } /* * buf_qrecycle: * * Free a buffer from the given bufqueue. kva controls whether the * freed buf must own some kva resources. This is used for * defragmenting. */ static int buf_qrecycle(int qindex, bool kva) { struct buf *bp, *nbp; if (kva) atomic_add_int(&bufdefragcnt, 1); nbp = NULL; mtx_lock(&bqlocks[qindex]); nbp = TAILQ_FIRST(&bufqueues[qindex]); /* * Run scan, possibly freeing data and/or kva mappings on the fly * depending. */ while ((bp = nbp) != NULL) { /* * Calculate next bp (we can only use it if we do not * release the bqlock). */ nbp = TAILQ_NEXT(bp, b_freelist); /* * If we are defragging then we need a buffer with * some kva to reclaim. */ if (kva && bp->b_kvasize == 0) continue; if (BUF_LOCK(bp, LK_EXCLUSIVE | LK_NOWAIT, NULL) != 0) continue; /* * Skip buffers with background writes in progress. */ if ((bp->b_vflags & BV_BKGRDINPROG) != 0) { BUF_UNLOCK(bp); continue; } KASSERT(bp->b_qindex == qindex, ("getnewbuf: inconsistent queue %d bp %p", qindex, bp)); /* * NOTE: nbp is now entirely invalid. We can only restart * the scan from this point on. */ bremfreel(bp); mtx_unlock(&bqlocks[qindex]); /* * Requeue the background write buffer with error and * restart the scan. */ if ((bp->b_vflags & BV_BKGRDERR) != 0) { bqrelse(bp); mtx_lock(&bqlocks[qindex]); nbp = TAILQ_FIRST(&bufqueues[qindex]); continue; } bp->b_flags |= B_INVAL; brelse(bp); return (0); } mtx_unlock(&bqlocks[qindex]); return (ENOBUFS); } /* * buf_recycle: * * Iterate through all clean queues until we find a buf to recycle or * exhaust the search. */ static int buf_recycle(bool kva) { int qindex, first_qindex; qindex = first_qindex = bqcleanq(); do { if (buf_qrecycle(qindex, kva) == 0) return (0); if (++qindex == QUEUE_CLEAN + clean_queues) qindex = QUEUE_CLEAN; } while (qindex != first_qindex); return (ENOBUFS); } /* * buf_scan: * * Scan the clean queues looking for a buffer to recycle. needsbuffer * is set on failure so that the caller may optionally bufspace_wait() * in a race-free fashion. */ static int buf_scan(bool defrag) { int error; /* * To avoid heavy synchronization and wakeup races we set * needsbuffer and re-poll before failing. This ensures that * no frees can be missed between an unsuccessful poll and * going to sleep in a synchronized fashion. */ if ((error = buf_recycle(defrag)) != 0) { atomic_set_int(&needsbuffer, 1); bufspace_daemonwakeup(); error = buf_recycle(defrag); } if (error == 0) atomic_add_int(&getnewbufrestarts, 1); return (error); } /* * bremfree: * * Mark the buffer for removal from the appropriate free list. * */ void bremfree(struct buf *bp) { CTR3(KTR_BUF, "bremfree(%p) vp %p flags %X", bp, bp->b_vp, bp->b_flags); KASSERT((bp->b_flags & B_REMFREE) == 0, ("bremfree: buffer %p already marked for delayed removal.", bp)); KASSERT(bp->b_qindex != QUEUE_NONE, ("bremfree: buffer %p not on a queue.", bp)); BUF_ASSERT_XLOCKED(bp); bp->b_flags |= B_REMFREE; } /* * bremfreef: * * Force an immediate removal from a free list. Used only in nfs when * it abuses the b_freelist pointer. */ void bremfreef(struct buf *bp) { struct mtx *qlock; qlock = bqlock(bp->b_qindex); mtx_lock(qlock); bremfreel(bp); mtx_unlock(qlock); } /* * bremfreel: * * Removes a buffer from the free list, must be called with the * correct qlock held. */ static void bremfreel(struct buf *bp) { CTR3(KTR_BUF, "bremfreel(%p) vp %p flags %X", bp, bp->b_vp, bp->b_flags); KASSERT(bp->b_qindex != QUEUE_NONE, ("bremfreel: buffer %p not on a queue.", bp)); if (bp->b_qindex != QUEUE_EMPTY) { BUF_ASSERT_XLOCKED(bp); } mtx_assert(bqlock(bp->b_qindex), MA_OWNED); TAILQ_REMOVE(&bufqueues[bp->b_qindex], bp, b_freelist); #ifdef INVARIANTS KASSERT(bq_len[bp->b_qindex] >= 1, ("queue %d underflow", bp->b_qindex)); bq_len[bp->b_qindex]--; #endif bp->b_qindex = QUEUE_NONE; bp->b_flags &= ~B_REMFREE; } /* * bufkva_free: * * Free the kva allocation for a buffer. * */ static void bufkva_free(struct buf *bp) { #ifdef INVARIANTS if (bp->b_kvasize == 0) { KASSERT(bp->b_kvabase == unmapped_buf && bp->b_data == unmapped_buf, ("Leaked KVA space on %p", bp)); } else if (buf_mapped(bp)) BUF_CHECK_MAPPED(bp); else BUF_CHECK_UNMAPPED(bp); #endif if (bp->b_kvasize == 0) return; vmem_free(buffer_arena, (vm_offset_t)bp->b_kvabase, bp->b_kvasize); atomic_subtract_long(&bufkvaspace, bp->b_kvasize); atomic_add_int(&buffreekvacnt, 1); bp->b_data = bp->b_kvabase = unmapped_buf; bp->b_kvasize = 0; } /* * bufkva_alloc: * * Allocate the buffer KVA and set b_kvasize and b_kvabase. */ static int bufkva_alloc(struct buf *bp, int maxsize, int gbflags) { vm_offset_t addr; int error; KASSERT((gbflags & GB_UNMAPPED) == 0 || (gbflags & GB_KVAALLOC) != 0, ("Invalid gbflags 0x%x in %s", gbflags, __func__)); bufkva_free(bp); addr = 0; error = vmem_alloc(buffer_arena, maxsize, M_BESTFIT | M_NOWAIT, &addr); if (error != 0) { /* * Buffer map is too fragmented. Request the caller * to defragment the map. */ return (error); } bp->b_kvabase = (caddr_t)addr; bp->b_kvasize = maxsize; atomic_add_long(&bufkvaspace, bp->b_kvasize); if ((gbflags & GB_UNMAPPED) != 0) { bp->b_data = unmapped_buf; BUF_CHECK_UNMAPPED(bp); } else { bp->b_data = bp->b_kvabase; BUF_CHECK_MAPPED(bp); } return (0); } /* * bufkva_reclaim: * * Reclaim buffer kva by freeing buffers holding kva. This is a vmem * callback that fires to avoid returning failure. */ static void bufkva_reclaim(vmem_t *vmem, int flags) { int i; for (i = 0; i < 5; i++) if (buf_scan(true) != 0) break; return; } /* * Attempt to initiate asynchronous I/O on read-ahead blocks. We must * clear BIO_ERROR and B_INVAL prior to initiating I/O . If B_CACHE is set, * the buffer is valid and we do not have to do anything. */ void breada(struct vnode * vp, daddr_t * rablkno, int * rabsize, int cnt, struct ucred * cred) { struct buf *rabp; int i; for (i = 0; i < cnt; i++, rablkno++, rabsize++) { if (inmem(vp, *rablkno)) continue; rabp = getblk(vp, *rablkno, *rabsize, 0, 0, 0); if ((rabp->b_flags & B_CACHE) == 0) { if (!TD_IS_IDLETHREAD(curthread)) { #ifdef RACCT if (racct_enable) { PROC_LOCK(curproc); racct_add_buf(curproc, rabp, 0); PROC_UNLOCK(curproc); } #endif /* RACCT */ curthread->td_ru.ru_inblock++; } rabp->b_flags |= B_ASYNC; rabp->b_flags &= ~B_INVAL; rabp->b_ioflags &= ~BIO_ERROR; rabp->b_iocmd = BIO_READ; if (rabp->b_rcred == NOCRED && cred != NOCRED) rabp->b_rcred = crhold(cred); vfs_busy_pages(rabp, 0); BUF_KERNPROC(rabp); rabp->b_iooffset = dbtob(rabp->b_blkno); bstrategy(rabp); } else { brelse(rabp); } } } /* * Entry point for bread() and breadn() via #defines in sys/buf.h. * * Get a buffer with the specified data. Look in the cache first. We * must clear BIO_ERROR and B_INVAL prior to initiating I/O. If B_CACHE * is set, the buffer is valid and we do not have to do anything, see * getblk(). Also starts asynchronous I/O on read-ahead blocks. * * Always return a NULL buffer pointer (in bpp) when returning an error. */ int breadn_flags(struct vnode *vp, daddr_t blkno, int size, daddr_t *rablkno, int *rabsize, int cnt, struct ucred *cred, int flags, struct buf **bpp) { struct buf *bp; int rv = 0, readwait = 0; CTR3(KTR_BUF, "breadn(%p, %jd, %d)", vp, blkno, size); /* * Can only return NULL if GB_LOCK_NOWAIT flag is specified. */ *bpp = bp = getblk(vp, blkno, size, 0, 0, flags); if (bp == NULL) return (EBUSY); /* if not found in cache, do some I/O */ if ((bp->b_flags & B_CACHE) == 0) { if (!TD_IS_IDLETHREAD(curthread)) { #ifdef RACCT if (racct_enable) { PROC_LOCK(curproc); racct_add_buf(curproc, bp, 0); PROC_UNLOCK(curproc); } #endif /* RACCT */ curthread->td_ru.ru_inblock++; } bp->b_iocmd = BIO_READ; bp->b_flags &= ~B_INVAL; bp->b_ioflags &= ~BIO_ERROR; if (bp->b_rcred == NOCRED && cred != NOCRED) bp->b_rcred = crhold(cred); vfs_busy_pages(bp, 0); bp->b_iooffset = dbtob(bp->b_blkno); bstrategy(bp); ++readwait; } breada(vp, rablkno, rabsize, cnt, cred); if (readwait) { rv = bufwait(bp); if (rv != 0) { brelse(bp); *bpp = NULL; } } return (rv); } /* * Write, release buffer on completion. (Done by iodone * if async). Do not bother writing anything if the buffer * is invalid. * * Note that we set B_CACHE here, indicating that buffer is * fully valid and thus cacheable. This is true even of NFS * now so we set it generally. This could be set either here * or in biodone() since the I/O is synchronous. We put it * here. */ int bufwrite(struct buf *bp) { int oldflags; struct vnode *vp; long space; int vp_md; CTR3(KTR_BUF, "bufwrite(%p) vp %p flags %X", bp, bp->b_vp, bp->b_flags); if ((bp->b_bufobj->bo_flag & BO_DEAD) != 0) { bp->b_flags |= B_INVAL | B_RELBUF; bp->b_flags &= ~B_CACHE; brelse(bp); return (ENXIO); } if (bp->b_flags & B_INVAL) { brelse(bp); return (0); } if (bp->b_flags & B_BARRIER) barrierwrites++; oldflags = bp->b_flags; BUF_ASSERT_HELD(bp); KASSERT(!(bp->b_vflags & BV_BKGRDINPROG), ("FFS background buffer should not get here %p", bp)); vp = bp->b_vp; if (vp) vp_md = vp->v_vflag & VV_MD; else vp_md = 0; /* * Mark the buffer clean. Increment the bufobj write count * before bundirty() call, to prevent other thread from seeing * empty dirty list and zero counter for writes in progress, * falsely indicating that the bufobj is clean. */ bufobj_wref(bp->b_bufobj); bundirty(bp); bp->b_flags &= ~B_DONE; bp->b_ioflags &= ~BIO_ERROR; bp->b_flags |= B_CACHE; bp->b_iocmd = BIO_WRITE; vfs_busy_pages(bp, 1); /* * Normal bwrites pipeline writes */ bp->b_runningbufspace = bp->b_bufsize; space = atomic_fetchadd_long(&runningbufspace, bp->b_runningbufspace); if (!TD_IS_IDLETHREAD(curthread)) { #ifdef RACCT if (racct_enable) { PROC_LOCK(curproc); racct_add_buf(curproc, bp, 1); PROC_UNLOCK(curproc); } #endif /* RACCT */ curthread->td_ru.ru_oublock++; } if (oldflags & B_ASYNC) BUF_KERNPROC(bp); bp->b_iooffset = dbtob(bp->b_blkno); buf_track(bp, __func__); bstrategy(bp); if ((oldflags & B_ASYNC) == 0) { int rtval = bufwait(bp); brelse(bp); return (rtval); } else if (space > hirunningspace) { /* * don't allow the async write to saturate the I/O * system. We will not deadlock here because * we are blocking waiting for I/O that is already in-progress * to complete. We do not block here if it is the update * or syncer daemon trying to clean up as that can lead * to deadlock. */ if ((curthread->td_pflags & TDP_NORUNNINGBUF) == 0 && !vp_md) waitrunningbufspace(); } return (0); } void bufbdflush(struct bufobj *bo, struct buf *bp) { struct buf *nbp; if (bo->bo_dirty.bv_cnt > dirtybufthresh + 10) { (void) VOP_FSYNC(bp->b_vp, MNT_NOWAIT, curthread); altbufferflushes++; } else if (bo->bo_dirty.bv_cnt > dirtybufthresh) { BO_LOCK(bo); /* * Try to find a buffer to flush. */ TAILQ_FOREACH(nbp, &bo->bo_dirty.bv_hd, b_bobufs) { if ((nbp->b_vflags & BV_BKGRDINPROG) || BUF_LOCK(nbp, LK_EXCLUSIVE | LK_NOWAIT, NULL)) continue; if (bp == nbp) panic("bdwrite: found ourselves"); BO_UNLOCK(bo); /* Don't countdeps with the bo lock held. */ if (buf_countdeps(nbp, 0)) { BO_LOCK(bo); BUF_UNLOCK(nbp); continue; } if (nbp->b_flags & B_CLUSTEROK) { vfs_bio_awrite(nbp); } else { bremfree(nbp); bawrite(nbp); } dirtybufferflushes++; break; } if (nbp == NULL) BO_UNLOCK(bo); } } /* * Delayed write. (Buffer is marked dirty). Do not bother writing * anything if the buffer is marked invalid. * * Note that since the buffer must be completely valid, we can safely * set B_CACHE. In fact, we have to set B_CACHE here rather then in * biodone() in order to prevent getblk from writing the buffer * out synchronously. */ void bdwrite(struct buf *bp) { struct thread *td = curthread; struct vnode *vp; struct bufobj *bo; CTR3(KTR_BUF, "bdwrite(%p) vp %p flags %X", bp, bp->b_vp, bp->b_flags); KASSERT(bp->b_bufobj != NULL, ("No b_bufobj %p", bp)); KASSERT((bp->b_flags & B_BARRIER) == 0, ("Barrier request in delayed write %p", bp)); BUF_ASSERT_HELD(bp); if (bp->b_flags & B_INVAL) { brelse(bp); return; } /* * If we have too many dirty buffers, don't create any more. * If we are wildly over our limit, then force a complete * cleanup. Otherwise, just keep the situation from getting * out of control. Note that we have to avoid a recursive * disaster and not try to clean up after our own cleanup! */ vp = bp->b_vp; bo = bp->b_bufobj; if ((td->td_pflags & (TDP_COWINPROGRESS|TDP_INBDFLUSH)) == 0) { td->td_pflags |= TDP_INBDFLUSH; BO_BDFLUSH(bo, bp); td->td_pflags &= ~TDP_INBDFLUSH; } else recursiveflushes++; bdirty(bp); /* * Set B_CACHE, indicating that the buffer is fully valid. This is * true even of NFS now. */ bp->b_flags |= B_CACHE; /* * This bmap keeps the system from needing to do the bmap later, * perhaps when the system is attempting to do a sync. Since it * is likely that the indirect block -- or whatever other datastructure * that the filesystem needs is still in memory now, it is a good * thing to do this. Note also, that if the pageout daemon is * requesting a sync -- there might not be enough memory to do * the bmap then... So, this is important to do. */ if (vp->v_type != VCHR && bp->b_lblkno == bp->b_blkno) { VOP_BMAP(vp, bp->b_lblkno, NULL, &bp->b_blkno, NULL, NULL); } buf_track(bp, __func__); /* * Set the *dirty* buffer range based upon the VM system dirty * pages. * * Mark the buffer pages as clean. We need to do this here to * satisfy the vnode_pager and the pageout daemon, so that it * thinks that the pages have been "cleaned". Note that since * the pages are in a delayed write buffer -- the VFS layer * "will" see that the pages get written out on the next sync, * or perhaps the cluster will be completed. */ vfs_clean_pages_dirty_buf(bp); bqrelse(bp); /* * note: we cannot initiate I/O from a bdwrite even if we wanted to, * due to the softdep code. */ } /* * bdirty: * * Turn buffer into delayed write request. We must clear BIO_READ and * B_RELBUF, and we must set B_DELWRI. We reassign the buffer to * itself to properly update it in the dirty/clean lists. We mark it * B_DONE to ensure that any asynchronization of the buffer properly * clears B_DONE ( else a panic will occur later ). * * bdirty() is kinda like bdwrite() - we have to clear B_INVAL which * might have been set pre-getblk(). Unlike bwrite/bdwrite, bdirty() * should only be called if the buffer is known-good. * * Since the buffer is not on a queue, we do not update the numfreebuffers * count. * * The buffer must be on QUEUE_NONE. */ void bdirty(struct buf *bp) { CTR3(KTR_BUF, "bdirty(%p) vp %p flags %X", bp, bp->b_vp, bp->b_flags); KASSERT(bp->b_bufobj != NULL, ("No b_bufobj %p", bp)); KASSERT(bp->b_flags & B_REMFREE || bp->b_qindex == QUEUE_NONE, ("bdirty: buffer %p still on queue %d", bp, bp->b_qindex)); BUF_ASSERT_HELD(bp); bp->b_flags &= ~(B_RELBUF); bp->b_iocmd = BIO_WRITE; if ((bp->b_flags & B_DELWRI) == 0) { bp->b_flags |= /* XXX B_DONE | */ B_DELWRI; reassignbuf(bp); bdirtyadd(); } } /* * bundirty: * * Clear B_DELWRI for buffer. * * Since the buffer is not on a queue, we do not update the numfreebuffers * count. * * The buffer must be on QUEUE_NONE. */ void bundirty(struct buf *bp) { CTR3(KTR_BUF, "bundirty(%p) vp %p flags %X", bp, bp->b_vp, bp->b_flags); KASSERT(bp->b_bufobj != NULL, ("No b_bufobj %p", bp)); KASSERT(bp->b_flags & B_REMFREE || bp->b_qindex == QUEUE_NONE, ("bundirty: buffer %p still on queue %d", bp, bp->b_qindex)); BUF_ASSERT_HELD(bp); if (bp->b_flags & B_DELWRI) { bp->b_flags &= ~B_DELWRI; reassignbuf(bp); bdirtysub(); } /* * Since it is now being written, we can clear its deferred write flag. */ bp->b_flags &= ~B_DEFERRED; } /* * bawrite: * * Asynchronous write. Start output on a buffer, but do not wait for * it to complete. The buffer is released when the output completes. * * bwrite() ( or the VOP routine anyway ) is responsible for handling * B_INVAL buffers. Not us. */ void bawrite(struct buf *bp) { bp->b_flags |= B_ASYNC; (void) bwrite(bp); } /* * babarrierwrite: * * Asynchronous barrier write. Start output on a buffer, but do not * wait for it to complete. Place a write barrier after this write so * that this buffer and all buffers written before it are committed to * the disk before any buffers written after this write are committed * to the disk. The buffer is released when the output completes. */ void babarrierwrite(struct buf *bp) { bp->b_flags |= B_ASYNC | B_BARRIER; (void) bwrite(bp); } /* * bbarrierwrite: * * Synchronous barrier write. Start output on a buffer and wait for * it to complete. Place a write barrier after this write so that * this buffer and all buffers written before it are committed to * the disk before any buffers written after this write are committed * to the disk. The buffer is released when the output completes. */ int bbarrierwrite(struct buf *bp) { bp->b_flags |= B_BARRIER; return (bwrite(bp)); } /* * bwillwrite: * * Called prior to the locking of any vnodes when we are expecting to * write. We do not want to starve the buffer cache with too many * dirty buffers so we block here. By blocking prior to the locking * of any vnodes we attempt to avoid the situation where a locked vnode * prevents the various system daemons from flushing related buffers. */ void bwillwrite(void) { if (numdirtybuffers >= hidirtybuffers) { mtx_lock(&bdirtylock); while (numdirtybuffers >= hidirtybuffers) { bdirtywait = 1; msleep(&bdirtywait, &bdirtylock, (PRIBIO + 4), "flswai", 0); } mtx_unlock(&bdirtylock); } } /* * Return true if we have too many dirty buffers. */ int buf_dirty_count_severe(void) { return(numdirtybuffers >= hidirtybuffers); } /* * brelse: * * Release a busy buffer and, if requested, free its resources. The * buffer will be stashed in the appropriate bufqueue[] allowing it * to be accessed later as a cache entity or reused for other purposes. */ void brelse(struct buf *bp) { int qindex; /* * Many functions erroneously call brelse with a NULL bp under rare * error conditions. Simply return when called with a NULL bp. */ if (bp == NULL) return; CTR3(KTR_BUF, "brelse(%p) vp %p flags %X", bp, bp->b_vp, bp->b_flags); KASSERT(!(bp->b_flags & (B_CLUSTER|B_PAGING)), ("brelse: inappropriate B_PAGING or B_CLUSTER bp %p", bp)); KASSERT((bp->b_flags & B_VMIO) != 0 || (bp->b_flags & B_NOREUSE) == 0, ("brelse: non-VMIO buffer marked NOREUSE")); if (BUF_LOCKRECURSED(bp)) { /* * Do not process, in particular, do not handle the * B_INVAL/B_RELBUF and do not release to free list. */ BUF_UNLOCK(bp); return; } if (bp->b_flags & B_MANAGED) { bqrelse(bp); return; } if ((bp->b_vflags & (BV_BKGRDINPROG | BV_BKGRDERR)) == BV_BKGRDERR) { BO_LOCK(bp->b_bufobj); bp->b_vflags &= ~BV_BKGRDERR; BO_UNLOCK(bp->b_bufobj); bdirty(bp); } if (bp->b_iocmd == BIO_WRITE && (bp->b_ioflags & BIO_ERROR) && !(bp->b_flags & B_INVAL)) { /* * Failed write, redirty. Must clear BIO_ERROR to prevent * pages from being scrapped. */ bp->b_ioflags &= ~BIO_ERROR; bdirty(bp); } else if ((bp->b_flags & (B_NOCACHE | B_INVAL)) || (bp->b_ioflags & BIO_ERROR) || (bp->b_bufsize <= 0)) { /* * Either a failed read I/O or we were asked to free or not * cache the buffer. */ bp->b_flags |= B_INVAL; if (!LIST_EMPTY(&bp->b_dep)) buf_deallocate(bp); if (bp->b_flags & B_DELWRI) bdirtysub(); bp->b_flags &= ~(B_DELWRI | B_CACHE); if ((bp->b_flags & B_VMIO) == 0) { allocbuf(bp, 0); if (bp->b_vp) brelvp(bp); } } /* * We must clear B_RELBUF if B_DELWRI is set. If vfs_vmio_truncate() * is called with B_DELWRI set, the underlying pages may wind up * getting freed causing a previous write (bdwrite()) to get 'lost' * because pages associated with a B_DELWRI bp are marked clean. * * We still allow the B_INVAL case to call vfs_vmio_truncate(), even * if B_DELWRI is set. */ if (bp->b_flags & B_DELWRI) bp->b_flags &= ~B_RELBUF; /* * VMIO buffer rundown. It is not very necessary to keep a VMIO buffer * constituted, not even NFS buffers now. Two flags effect this. If * B_INVAL, the struct buf is invalidated but the VM object is kept * around ( i.e. so it is trivial to reconstitute the buffer later ). * * If BIO_ERROR or B_NOCACHE is set, pages in the VM object will be * invalidated. BIO_ERROR cannot be set for a failed write unless the * buffer is also B_INVAL because it hits the re-dirtying code above. * * Normally we can do this whether a buffer is B_DELWRI or not. If * the buffer is an NFS buffer, it is tracking piecemeal writes or * the commit state and we cannot afford to lose the buffer. If the * buffer has a background write in progress, we need to keep it * around to prevent it from being reconstituted and starting a second * background write. */ if ((bp->b_flags & B_VMIO) && (bp->b_flags & B_NOCACHE || (bp->b_ioflags & BIO_ERROR && bp->b_iocmd == BIO_READ)) && !(bp->b_vp->v_mount != NULL && (bp->b_vp->v_mount->mnt_vfc->vfc_flags & VFCF_NETWORK) != 0 && !vn_isdisk(bp->b_vp, NULL) && (bp->b_flags & B_DELWRI))) { vfs_vmio_invalidate(bp); allocbuf(bp, 0); } if ((bp->b_flags & (B_INVAL | B_RELBUF)) != 0 || (bp->b_flags & (B_DELWRI | B_NOREUSE)) == B_NOREUSE) { allocbuf(bp, 0); bp->b_flags &= ~B_NOREUSE; if (bp->b_vp != NULL) brelvp(bp); } /* * If the buffer has junk contents signal it and eventually * clean up B_DELWRI and diassociate the vnode so that gbincore() * doesn't find it. */ if (bp->b_bufsize == 0 || (bp->b_ioflags & BIO_ERROR) != 0 || (bp->b_flags & (B_INVAL | B_NOCACHE | B_RELBUF)) != 0) bp->b_flags |= B_INVAL; if (bp->b_flags & B_INVAL) { if (bp->b_flags & B_DELWRI) bundirty(bp); if (bp->b_vp) brelvp(bp); } buf_track(bp, __func__); /* buffers with no memory */ if (bp->b_bufsize == 0) { buf_free(bp); return; } /* buffers with junk contents */ if (bp->b_flags & (B_INVAL | B_NOCACHE | B_RELBUF) || (bp->b_ioflags & BIO_ERROR)) { bp->b_xflags &= ~(BX_BKGRDWRITE | BX_ALTDATA); if (bp->b_vflags & BV_BKGRDINPROG) panic("losing buffer 2"); qindex = QUEUE_CLEAN; bp->b_flags |= B_AGE; /* remaining buffers */ } else if (bp->b_flags & B_DELWRI) qindex = QUEUE_DIRTY; else qindex = QUEUE_CLEAN; binsfree(bp, qindex); bp->b_flags &= ~(B_ASYNC | B_NOCACHE | B_AGE | B_RELBUF | B_DIRECT); if ((bp->b_flags & B_DELWRI) == 0 && (bp->b_xflags & BX_VNDIRTY)) panic("brelse: not dirty"); /* unlock */ BUF_UNLOCK(bp); if (qindex == QUEUE_CLEAN) bufspace_wakeup(); } /* * Release a buffer back to the appropriate queue but do not try to free * it. The buffer is expected to be used again soon. * * bqrelse() is used by bdwrite() to requeue a delayed write, and used by * biodone() to requeue an async I/O on completion. It is also used when * known good buffers need to be requeued but we think we may need the data * again soon. * * XXX we should be able to leave the B_RELBUF hint set on completion. */ void bqrelse(struct buf *bp) { int qindex; CTR3(KTR_BUF, "bqrelse(%p) vp %p flags %X", bp, bp->b_vp, bp->b_flags); KASSERT(!(bp->b_flags & (B_CLUSTER|B_PAGING)), ("bqrelse: inappropriate B_PAGING or B_CLUSTER bp %p", bp)); qindex = QUEUE_NONE; if (BUF_LOCKRECURSED(bp)) { /* do not release to free list */ BUF_UNLOCK(bp); return; } bp->b_flags &= ~(B_ASYNC | B_NOCACHE | B_AGE | B_RELBUF); if (bp->b_flags & B_MANAGED) { if (bp->b_flags & B_REMFREE) bremfreef(bp); goto out; } /* buffers with stale but valid contents */ if ((bp->b_flags & B_DELWRI) != 0 || (bp->b_vflags & (BV_BKGRDINPROG | BV_BKGRDERR)) == BV_BKGRDERR) { BO_LOCK(bp->b_bufobj); bp->b_vflags &= ~BV_BKGRDERR; BO_UNLOCK(bp->b_bufobj); qindex = QUEUE_DIRTY; } else { if ((bp->b_flags & B_DELWRI) == 0 && (bp->b_xflags & BX_VNDIRTY)) panic("bqrelse: not dirty"); if ((bp->b_flags & B_NOREUSE) != 0) { brelse(bp); return; } qindex = QUEUE_CLEAN; } binsfree(bp, qindex); out: buf_track(bp, __func__); /* unlock */ BUF_UNLOCK(bp); if (qindex == QUEUE_CLEAN) bufspace_wakeup(); } /* * Complete I/O to a VMIO backed page. Validate the pages as appropriate, * restore bogus pages. */ static void vfs_vmio_iodone(struct buf *bp) { vm_ooffset_t foff; vm_page_t m; vm_object_t obj; struct vnode *vp; int bogus, i, iosize; obj = bp->b_bufobj->bo_object; KASSERT(obj->paging_in_progress >= bp->b_npages, ("vfs_vmio_iodone: paging in progress(%d) < b_npages(%d)", obj->paging_in_progress, bp->b_npages)); vp = bp->b_vp; KASSERT(vp->v_holdcnt > 0, ("vfs_vmio_iodone: vnode %p has zero hold count", vp)); KASSERT(vp->v_object != NULL, ("vfs_vmio_iodone: vnode %p has no vm_object", vp)); foff = bp->b_offset; KASSERT(bp->b_offset != NOOFFSET, ("vfs_vmio_iodone: bp %p has no buffer offset", bp)); bogus = 0; iosize = bp->b_bcount - bp->b_resid; VM_OBJECT_WLOCK(obj); for (i = 0; i < bp->b_npages; i++) { int resid; resid = ((foff + PAGE_SIZE) & ~(off_t)PAGE_MASK) - foff; if (resid > iosize) resid = iosize; /* * cleanup bogus pages, restoring the originals */ m = bp->b_pages[i]; if (m == bogus_page) { bogus = 1; m = vm_page_lookup(obj, OFF_TO_IDX(foff)); if (m == NULL) panic("biodone: page disappeared!"); bp->b_pages[i] = m; } else if ((bp->b_iocmd == BIO_READ) && resid > 0) { /* * In the write case, the valid and clean bits are * already changed correctly ( see bdwrite() ), so we * only need to do this here in the read case. */ KASSERT((m->dirty & vm_page_bits(foff & PAGE_MASK, resid)) == 0, ("vfs_vmio_iodone: page %p " "has unexpected dirty bits", m)); vfs_page_set_valid(bp, foff, m); } KASSERT(OFF_TO_IDX(foff) == m->pindex, ("vfs_vmio_iodone: foff(%jd)/pindex(%ju) mismatch", (intmax_t)foff, (uintmax_t)m->pindex)); vm_page_sunbusy(m); foff = (foff + PAGE_SIZE) & ~(off_t)PAGE_MASK; iosize -= resid; } vm_object_pip_wakeupn(obj, bp->b_npages); VM_OBJECT_WUNLOCK(obj); if (bogus && buf_mapped(bp)) { BUF_CHECK_MAPPED(bp); pmap_qenter(trunc_page((vm_offset_t)bp->b_data), bp->b_pages, bp->b_npages); } } /* * Unwire a page held by a buf and place it on the appropriate vm queue. */ static void vfs_vmio_unwire(struct buf *bp, vm_page_t m) { bool freed; vm_page_lock(m); if (vm_page_unwire(m, PQ_NONE)) { /* * Determine if the page should be freed before adding * it to the inactive queue. */ if (m->valid == 0) { freed = !vm_page_busied(m); if (freed) vm_page_free(m); } else if ((bp->b_flags & B_DIRECT) != 0) freed = vm_page_try_to_free(m); else freed = false; if (!freed) { /* * If the page is unlikely to be reused, let the * VM know. Otherwise, maintain LRU page * ordering and put the page at the tail of the * inactive queue. */ if ((bp->b_flags & B_NOREUSE) != 0) vm_page_deactivate_noreuse(m); else vm_page_deactivate(m); } } vm_page_unlock(m); } /* * Perform page invalidation when a buffer is released. The fully invalid * pages will be reclaimed later in vfs_vmio_truncate(). */ static void vfs_vmio_invalidate(struct buf *bp) { vm_object_t obj; vm_page_t m; int i, resid, poffset, presid; if (buf_mapped(bp)) { BUF_CHECK_MAPPED(bp); pmap_qremove(trunc_page((vm_offset_t)bp->b_data), bp->b_npages); } else BUF_CHECK_UNMAPPED(bp); /* * Get the base offset and length of the buffer. Note that * in the VMIO case if the buffer block size is not * page-aligned then b_data pointer may not be page-aligned. * But our b_pages[] array *IS* page aligned. * * block sizes less then DEV_BSIZE (usually 512) are not * supported due to the page granularity bits (m->valid, * m->dirty, etc...). * * See man buf(9) for more information */ obj = bp->b_bufobj->bo_object; resid = bp->b_bufsize; poffset = bp->b_offset & PAGE_MASK; VM_OBJECT_WLOCK(obj); for (i = 0; i < bp->b_npages; i++) { m = bp->b_pages[i]; if (m == bogus_page) panic("vfs_vmio_invalidate: Unexpected bogus page."); bp->b_pages[i] = NULL; presid = resid > (PAGE_SIZE - poffset) ? (PAGE_SIZE - poffset) : resid; KASSERT(presid >= 0, ("brelse: extra page")); while (vm_page_xbusied(m)) { vm_page_lock(m); VM_OBJECT_WUNLOCK(obj); vm_page_busy_sleep(m, "mbncsh", true); VM_OBJECT_WLOCK(obj); } if (pmap_page_wired_mappings(m) == 0) vm_page_set_invalid(m, poffset, presid); vfs_vmio_unwire(bp, m); resid -= presid; poffset = 0; } VM_OBJECT_WUNLOCK(obj); bp->b_npages = 0; } /* * Page-granular truncation of an existing VMIO buffer. */ static void vfs_vmio_truncate(struct buf *bp, int desiredpages) { vm_object_t obj; vm_page_t m; int i; if (bp->b_npages == desiredpages) return; if (buf_mapped(bp)) { BUF_CHECK_MAPPED(bp); pmap_qremove((vm_offset_t)trunc_page((vm_offset_t)bp->b_data) + (desiredpages << PAGE_SHIFT), bp->b_npages - desiredpages); } else BUF_CHECK_UNMAPPED(bp); obj = bp->b_bufobj->bo_object; if (obj != NULL) VM_OBJECT_WLOCK(obj); for (i = desiredpages; i < bp->b_npages; i++) { m = bp->b_pages[i]; KASSERT(m != bogus_page, ("allocbuf: bogus page found")); bp->b_pages[i] = NULL; vfs_vmio_unwire(bp, m); } if (obj != NULL) VM_OBJECT_WUNLOCK(obj); bp->b_npages = desiredpages; } /* * Byte granular extension of VMIO buffers. */ static void vfs_vmio_extend(struct buf *bp, int desiredpages, int size) { /* * We are growing the buffer, possibly in a * byte-granular fashion. */ vm_object_t obj; vm_offset_t toff; vm_offset_t tinc; vm_page_t m; /* * Step 1, bring in the VM pages from the object, allocating * them if necessary. We must clear B_CACHE if these pages * are not valid for the range covered by the buffer. */ obj = bp->b_bufobj->bo_object; VM_OBJECT_WLOCK(obj); while (bp->b_npages < desiredpages) { /* * We must allocate system pages since blocking * here could interfere with paging I/O, no * matter which process we are. * * Only exclusive busy can be tested here. * Blocking on shared busy might lead to * deadlocks once allocbuf() is called after * pages are vfs_busy_pages(). */ m = vm_page_grab(obj, OFF_TO_IDX(bp->b_offset) + bp->b_npages, VM_ALLOC_NOBUSY | VM_ALLOC_SYSTEM | VM_ALLOC_WIRED | VM_ALLOC_IGN_SBUSY | VM_ALLOC_COUNT(desiredpages - bp->b_npages)); if (m->valid == 0) bp->b_flags &= ~B_CACHE; bp->b_pages[bp->b_npages] = m; ++bp->b_npages; } /* * Step 2. We've loaded the pages into the buffer, * we have to figure out if we can still have B_CACHE * set. Note that B_CACHE is set according to the * byte-granular range ( bcount and size ), not the * aligned range ( newbsize ). * * The VM test is against m->valid, which is DEV_BSIZE * aligned. Needless to say, the validity of the data * needs to also be DEV_BSIZE aligned. Note that this * fails with NFS if the server or some other client * extends the file's EOF. If our buffer is resized, * B_CACHE may remain set! XXX */ toff = bp->b_bcount; tinc = PAGE_SIZE - ((bp->b_offset + toff) & PAGE_MASK); while ((bp->b_flags & B_CACHE) && toff < size) { vm_pindex_t pi; if (tinc > (size - toff)) tinc = size - toff; pi = ((bp->b_offset & PAGE_MASK) + toff) >> PAGE_SHIFT; m = bp->b_pages[pi]; vfs_buf_test_cache(bp, bp->b_offset, toff, tinc, m); toff += tinc; tinc = PAGE_SIZE; } VM_OBJECT_WUNLOCK(obj); /* * Step 3, fixup the KVA pmap. */ if (buf_mapped(bp)) bpmap_qenter(bp); else BUF_CHECK_UNMAPPED(bp); } /* * Check to see if a block at a particular lbn is available for a clustered * write. */ static int vfs_bio_clcheck(struct vnode *vp, int size, daddr_t lblkno, daddr_t blkno) { struct buf *bpa; int match; match = 0; /* If the buf isn't in core skip it */ if ((bpa = gbincore(&vp->v_bufobj, lblkno)) == NULL) return (0); /* If the buf is busy we don't want to wait for it */ if (BUF_LOCK(bpa, LK_EXCLUSIVE | LK_NOWAIT, NULL) != 0) return (0); /* Only cluster with valid clusterable delayed write buffers */ if ((bpa->b_flags & (B_DELWRI | B_CLUSTEROK | B_INVAL)) != (B_DELWRI | B_CLUSTEROK)) goto done; if (bpa->b_bufsize != size) goto done; /* * Check to see if it is in the expected place on disk and that the * block has been mapped. */ if ((bpa->b_blkno != bpa->b_lblkno) && (bpa->b_blkno == blkno)) match = 1; done: BUF_UNLOCK(bpa); return (match); } /* * vfs_bio_awrite: * * Implement clustered async writes for clearing out B_DELWRI buffers. * This is much better then the old way of writing only one buffer at * a time. Note that we may not be presented with the buffers in the * correct order, so we search for the cluster in both directions. */ int vfs_bio_awrite(struct buf *bp) { struct bufobj *bo; int i; int j; daddr_t lblkno = bp->b_lblkno; struct vnode *vp = bp->b_vp; int ncl; int nwritten; int size; int maxcl; int gbflags; bo = &vp->v_bufobj; gbflags = (bp->b_data == unmapped_buf) ? GB_UNMAPPED : 0; /* * right now we support clustered writing only to regular files. If * we find a clusterable block we could be in the middle of a cluster * rather then at the beginning. */ if ((vp->v_type == VREG) && (vp->v_mount != 0) && /* Only on nodes that have the size info */ (bp->b_flags & (B_CLUSTEROK | B_INVAL)) == B_CLUSTEROK) { size = vp->v_mount->mnt_stat.f_iosize; maxcl = MAXPHYS / size; BO_RLOCK(bo); for (i = 1; i < maxcl; i++) if (vfs_bio_clcheck(vp, size, lblkno + i, bp->b_blkno + ((i * size) >> DEV_BSHIFT)) == 0) break; for (j = 1; i + j <= maxcl && j <= lblkno; j++) if (vfs_bio_clcheck(vp, size, lblkno - j, bp->b_blkno - ((j * size) >> DEV_BSHIFT)) == 0) break; BO_RUNLOCK(bo); --j; ncl = i + j; /* * this is a possible cluster write */ if (ncl != 1) { BUF_UNLOCK(bp); nwritten = cluster_wbuild(vp, size, lblkno - j, ncl, gbflags); return (nwritten); } } bremfree(bp); bp->b_flags |= B_ASYNC; /* * default (old) behavior, writing out only one block * * XXX returns b_bufsize instead of b_bcount for nwritten? */ nwritten = bp->b_bufsize; (void) bwrite(bp); return (nwritten); } /* * getnewbuf_kva: * * Allocate KVA for an empty buf header according to gbflags. */ static int getnewbuf_kva(struct buf *bp, int gbflags, int maxsize) { if ((gbflags & (GB_UNMAPPED | GB_KVAALLOC)) != GB_UNMAPPED) { /* * In order to keep fragmentation sane we only allocate kva * in BKVASIZE chunks. XXX with vmem we can do page size. */ maxsize = (maxsize + BKVAMASK) & ~BKVAMASK; if (maxsize != bp->b_kvasize && bufkva_alloc(bp, maxsize, gbflags)) return (ENOSPC); } return (0); } /* * getnewbuf: * * Find and initialize a new buffer header, freeing up existing buffers * in the bufqueues as necessary. The new buffer is returned locked. * * We block if: * We have insufficient buffer headers * We have insufficient buffer space * buffer_arena is too fragmented ( space reservation fails ) * If we have to flush dirty buffers ( but we try to avoid this ) * * The caller is responsible for releasing the reserved bufspace after * allocbuf() is called. */ static struct buf * getnewbuf(struct vnode *vp, int slpflag, int slptimeo, int maxsize, int gbflags) { struct buf *bp; bool metadata, reserved; bp = NULL; KASSERT((gbflags & (GB_UNMAPPED | GB_KVAALLOC)) != GB_KVAALLOC, ("GB_KVAALLOC only makes sense with GB_UNMAPPED")); if (!unmapped_buf_allowed) gbflags &= ~(GB_UNMAPPED | GB_KVAALLOC); if (vp == NULL || (vp->v_vflag & (VV_MD | VV_SYSTEM)) != 0 || vp->v_type == VCHR) metadata = true; else metadata = false; atomic_add_int(&getnewbufcalls, 1); reserved = false; do { if (reserved == false && bufspace_reserve(maxsize, metadata) != 0) continue; reserved = true; if ((bp = buf_alloc()) == NULL) continue; if (getnewbuf_kva(bp, gbflags, maxsize) == 0) return (bp); break; } while(buf_scan(false) == 0); if (reserved) atomic_subtract_long(&bufspace, maxsize); if (bp != NULL) { bp->b_flags |= B_INVAL; brelse(bp); } bufspace_wait(vp, gbflags, slpflag, slptimeo); return (NULL); } /* * buf_daemon: * * buffer flushing daemon. Buffers are normally flushed by the * update daemon but if it cannot keep up this process starts to * take the load in an attempt to prevent getnewbuf() from blocking. */ static struct kproc_desc buf_kp = { "bufdaemon", buf_daemon, &bufdaemonproc }; SYSINIT(bufdaemon, SI_SUB_KTHREAD_BUF, SI_ORDER_FIRST, kproc_start, &buf_kp); static int buf_flush(struct vnode *vp, int target) { int flushed; flushed = flushbufqueues(vp, target, 0); if (flushed == 0) { /* * Could not find any buffers without rollback * dependencies, so just write the first one * in the hopes of eventually making progress. */ if (vp != NULL && target > 2) target /= 2; flushbufqueues(vp, target, 1); } return (flushed); } static void buf_daemon() { int lodirty; /* * This process needs to be suspended prior to shutdown sync. */ EVENTHANDLER_REGISTER(shutdown_pre_sync, kproc_shutdown, bufdaemonproc, SHUTDOWN_PRI_LAST); /* * This process is allowed to take the buffer cache to the limit */ curthread->td_pflags |= TDP_NORUNNINGBUF | TDP_BUFNEED; mtx_lock(&bdlock); for (;;) { bd_request = 0; mtx_unlock(&bdlock); kproc_suspend_check(bufdaemonproc); lodirty = lodirtybuffers; if (bd_speedupreq) { lodirty = numdirtybuffers / 2; bd_speedupreq = 0; } /* * Do the flush. Limit the amount of in-transit I/O we * allow to build up, otherwise we would completely saturate * the I/O system. */ while (numdirtybuffers > lodirty) { if (buf_flush(NULL, numdirtybuffers - lodirty) == 0) break; kern_yield(PRI_USER); } /* * Only clear bd_request if we have reached our low water * mark. The buf_daemon normally waits 1 second and * then incrementally flushes any dirty buffers that have * built up, within reason. * * If we were unable to hit our low water mark and couldn't * find any flushable buffers, we sleep for a short period * to avoid endless loops on unlockable buffers. */ mtx_lock(&bdlock); if (numdirtybuffers <= lodirtybuffers) { /* * We reached our low water mark, reset the * request and sleep until we are needed again. * The sleep is just so the suspend code works. */ bd_request = 0; /* * Do an extra wakeup in case dirty threshold * changed via sysctl and the explicit transition * out of shortfall was missed. */ bdirtywakeup(); if (runningbufspace <= lorunningspace) runningwakeup(); msleep(&bd_request, &bdlock, PVM, "psleep", hz); } else { /* * We couldn't find any flushable dirty buffers but * still have too many dirty buffers, we * have to sleep and try again. (rare) */ msleep(&bd_request, &bdlock, PVM, "qsleep", hz / 10); } } } /* * flushbufqueues: * * Try to flush a buffer in the dirty queue. We must be careful to * free up B_INVAL buffers instead of write them, which NFS is * particularly sensitive to. */ static int flushwithdeps = 0; SYSCTL_INT(_vfs, OID_AUTO, flushwithdeps, CTLFLAG_RW, &flushwithdeps, 0, "Number of buffers flushed with dependecies that require rollbacks"); static int flushbufqueues(struct vnode *lvp, int target, int flushdeps) { struct buf *sentinel; struct vnode *vp; struct mount *mp; struct buf *bp; int hasdeps; int flushed; int queue; int error; bool unlock; flushed = 0; queue = QUEUE_DIRTY; bp = NULL; sentinel = malloc(sizeof(struct buf), M_TEMP, M_WAITOK | M_ZERO); sentinel->b_qindex = QUEUE_SENTINEL; mtx_lock(&bqlocks[queue]); TAILQ_INSERT_HEAD(&bufqueues[queue], sentinel, b_freelist); mtx_unlock(&bqlocks[queue]); while (flushed != target) { maybe_yield(); mtx_lock(&bqlocks[queue]); bp = TAILQ_NEXT(sentinel, b_freelist); if (bp != NULL) { TAILQ_REMOVE(&bufqueues[queue], sentinel, b_freelist); TAILQ_INSERT_AFTER(&bufqueues[queue], bp, sentinel, b_freelist); } else { mtx_unlock(&bqlocks[queue]); break; } /* * Skip sentinels inserted by other invocations of the * flushbufqueues(), taking care to not reorder them. * * Only flush the buffers that belong to the * vnode locked by the curthread. */ if (bp->b_qindex == QUEUE_SENTINEL || (lvp != NULL && bp->b_vp != lvp)) { mtx_unlock(&bqlocks[queue]); continue; } error = BUF_LOCK(bp, LK_EXCLUSIVE | LK_NOWAIT, NULL); mtx_unlock(&bqlocks[queue]); if (error != 0) continue; /* * BKGRDINPROG can only be set with the buf and bufobj * locks both held. We tolerate a race to clear it here. */ if ((bp->b_vflags & BV_BKGRDINPROG) != 0 || (bp->b_flags & B_DELWRI) == 0) { BUF_UNLOCK(bp); continue; } if (bp->b_flags & B_INVAL) { bremfreef(bp); brelse(bp); flushed++; continue; } if (!LIST_EMPTY(&bp->b_dep) && buf_countdeps(bp, 0)) { if (flushdeps == 0) { BUF_UNLOCK(bp); continue; } hasdeps = 1; } else hasdeps = 0; /* * We must hold the lock on a vnode before writing * one of its buffers. Otherwise we may confuse, or * in the case of a snapshot vnode, deadlock the * system. * * The lock order here is the reverse of the normal * of vnode followed by buf lock. This is ok because * the NOWAIT will prevent deadlock. */ vp = bp->b_vp; if (vn_start_write(vp, &mp, V_NOWAIT) != 0) { BUF_UNLOCK(bp); continue; } if (lvp == NULL) { unlock = true; error = vn_lock(vp, LK_EXCLUSIVE | LK_NOWAIT); } else { ASSERT_VOP_LOCKED(vp, "getbuf"); unlock = false; error = VOP_ISLOCKED(vp) == LK_EXCLUSIVE ? 0 : vn_lock(vp, LK_TRYUPGRADE); } if (error == 0) { CTR3(KTR_BUF, "flushbufqueue(%p) vp %p flags %X", bp, bp->b_vp, bp->b_flags); if (curproc == bufdaemonproc) { vfs_bio_awrite(bp); } else { bremfree(bp); bwrite(bp); notbufdflushes++; } vn_finished_write(mp); if (unlock) VOP_UNLOCK(vp, 0); flushwithdeps += hasdeps; flushed++; /* * Sleeping on runningbufspace while holding * vnode lock leads to deadlock. */ if (curproc == bufdaemonproc && runningbufspace > hirunningspace) waitrunningbufspace(); continue; } vn_finished_write(mp); BUF_UNLOCK(bp); } mtx_lock(&bqlocks[queue]); TAILQ_REMOVE(&bufqueues[queue], sentinel, b_freelist); mtx_unlock(&bqlocks[queue]); free(sentinel, M_TEMP); return (flushed); } /* * Check to see if a block is currently memory resident. */ struct buf * incore(struct bufobj *bo, daddr_t blkno) { struct buf *bp; BO_RLOCK(bo); bp = gbincore(bo, blkno); BO_RUNLOCK(bo); return (bp); } /* * Returns true if no I/O is needed to access the * associated VM object. This is like incore except * it also hunts around in the VM system for the data. */ static int inmem(struct vnode * vp, daddr_t blkno) { vm_object_t obj; vm_offset_t toff, tinc, size; vm_page_t m; vm_ooffset_t off; ASSERT_VOP_LOCKED(vp, "inmem"); if (incore(&vp->v_bufobj, blkno)) return 1; if (vp->v_mount == NULL) return 0; obj = vp->v_object; if (obj == NULL) return (0); size = PAGE_SIZE; if (size > vp->v_mount->mnt_stat.f_iosize) size = vp->v_mount->mnt_stat.f_iosize; off = (vm_ooffset_t)blkno * (vm_ooffset_t)vp->v_mount->mnt_stat.f_iosize; VM_OBJECT_RLOCK(obj); for (toff = 0; toff < vp->v_mount->mnt_stat.f_iosize; toff += tinc) { m = vm_page_lookup(obj, OFF_TO_IDX(off + toff)); if (!m) goto notinmem; tinc = size; if (tinc > PAGE_SIZE - ((toff + off) & PAGE_MASK)) tinc = PAGE_SIZE - ((toff + off) & PAGE_MASK); if (vm_page_is_valid(m, (vm_offset_t) ((toff + off) & PAGE_MASK), tinc) == 0) goto notinmem; } VM_OBJECT_RUNLOCK(obj); return 1; notinmem: VM_OBJECT_RUNLOCK(obj); return (0); } /* * Set the dirty range for a buffer based on the status of the dirty * bits in the pages comprising the buffer. The range is limited * to the size of the buffer. * * Tell the VM system that the pages associated with this buffer * are clean. This is used for delayed writes where the data is * going to go to disk eventually without additional VM intevention. * * Note that while we only really need to clean through to b_bcount, we * just go ahead and clean through to b_bufsize. */ static void vfs_clean_pages_dirty_buf(struct buf *bp) { vm_ooffset_t foff, noff, eoff; vm_page_t m; int i; if ((bp->b_flags & B_VMIO) == 0 || bp->b_bufsize == 0) return; foff = bp->b_offset; KASSERT(bp->b_offset != NOOFFSET, ("vfs_clean_pages_dirty_buf: no buffer offset")); VM_OBJECT_WLOCK(bp->b_bufobj->bo_object); vfs_drain_busy_pages(bp); vfs_setdirty_locked_object(bp); for (i = 0; i < bp->b_npages; i++) { noff = (foff + PAGE_SIZE) & ~(off_t)PAGE_MASK; eoff = noff; if (eoff > bp->b_offset + bp->b_bufsize) eoff = bp->b_offset + bp->b_bufsize; m = bp->b_pages[i]; vfs_page_set_validclean(bp, foff, m); /* vm_page_clear_dirty(m, foff & PAGE_MASK, eoff - foff); */ foff = noff; } VM_OBJECT_WUNLOCK(bp->b_bufobj->bo_object); } static void vfs_setdirty_locked_object(struct buf *bp) { vm_object_t object; int i; object = bp->b_bufobj->bo_object; VM_OBJECT_ASSERT_WLOCKED(object); /* * We qualify the scan for modified pages on whether the * object has been flushed yet. */ if ((object->flags & OBJ_MIGHTBEDIRTY) != 0) { vm_offset_t boffset; vm_offset_t eoffset; /* * test the pages to see if they have been modified directly * by users through the VM system. */ for (i = 0; i < bp->b_npages; i++) vm_page_test_dirty(bp->b_pages[i]); /* * Calculate the encompassing dirty range, boffset and eoffset, * (eoffset - boffset) bytes. */ for (i = 0; i < bp->b_npages; i++) { if (bp->b_pages[i]->dirty) break; } boffset = (i << PAGE_SHIFT) - (bp->b_offset & PAGE_MASK); for (i = bp->b_npages - 1; i >= 0; --i) { if (bp->b_pages[i]->dirty) { break; } } eoffset = ((i + 1) << PAGE_SHIFT) - (bp->b_offset & PAGE_MASK); /* * Fit it to the buffer. */ if (eoffset > bp->b_bcount) eoffset = bp->b_bcount; /* * If we have a good dirty range, merge with the existing * dirty range. */ if (boffset < eoffset) { if (bp->b_dirtyoff > boffset) bp->b_dirtyoff = boffset; if (bp->b_dirtyend < eoffset) bp->b_dirtyend = eoffset; } } } /* * Allocate the KVA mapping for an existing buffer. * If an unmapped buffer is provided but a mapped buffer is requested, take * also care to properly setup mappings between pages and KVA. */ static void bp_unmapped_get_kva(struct buf *bp, daddr_t blkno, int size, int gbflags) { int bsize, maxsize, need_mapping, need_kva; off_t offset; need_mapping = bp->b_data == unmapped_buf && (gbflags & GB_UNMAPPED) == 0; need_kva = bp->b_kvabase == unmapped_buf && bp->b_data == unmapped_buf && (gbflags & GB_KVAALLOC) != 0; if (!need_mapping && !need_kva) return; BUF_CHECK_UNMAPPED(bp); if (need_mapping && bp->b_kvabase != unmapped_buf) { /* * Buffer is not mapped, but the KVA was already * reserved at the time of the instantiation. Use the * allocated space. */ goto has_addr; } /* * Calculate the amount of the address space we would reserve * if the buffer was mapped. */ bsize = vn_isdisk(bp->b_vp, NULL) ? DEV_BSIZE : bp->b_bufobj->bo_bsize; KASSERT(bsize != 0, ("bsize == 0, check bo->bo_bsize")); offset = blkno * bsize; maxsize = size + (offset & PAGE_MASK); maxsize = imax(maxsize, bsize); while (bufkva_alloc(bp, maxsize, gbflags) != 0) { if ((gbflags & GB_NOWAIT_BD) != 0) { /* * XXXKIB: defragmentation cannot * succeed, not sure what else to do. */ panic("GB_NOWAIT_BD and GB_UNMAPPED %p", bp); } atomic_add_int(&mappingrestarts, 1); bufspace_wait(bp->b_vp, gbflags, 0, 0); } has_addr: if (need_mapping) { /* b_offset is handled by bpmap_qenter. */ bp->b_data = bp->b_kvabase; BUF_CHECK_MAPPED(bp); bpmap_qenter(bp); } } /* * getblk: * * Get a block given a specified block and offset into a file/device. * The buffers B_DONE bit will be cleared on return, making it almost * ready for an I/O initiation. B_INVAL may or may not be set on * return. The caller should clear B_INVAL prior to initiating a * READ. * * For a non-VMIO buffer, B_CACHE is set to the opposite of B_INVAL for * an existing buffer. * * For a VMIO buffer, B_CACHE is modified according to the backing VM. * If getblk()ing a previously 0-sized invalid buffer, B_CACHE is set * and then cleared based on the backing VM. If the previous buffer is * non-0-sized but invalid, B_CACHE will be cleared. * * If getblk() must create a new buffer, the new buffer is returned with * both B_INVAL and B_CACHE clear unless it is a VMIO buffer, in which * case it is returned with B_INVAL clear and B_CACHE set based on the * backing VM. * * getblk() also forces a bwrite() for any B_DELWRI buffer whos * B_CACHE bit is clear. * * What this means, basically, is that the caller should use B_CACHE to * determine whether the buffer is fully valid or not and should clear * B_INVAL prior to issuing a read. If the caller intends to validate * the buffer by loading its data area with something, the caller needs * to clear B_INVAL. If the caller does this without issuing an I/O, * the caller should set B_CACHE ( as an optimization ), else the caller * should issue the I/O and biodone() will set B_CACHE if the I/O was * a write attempt or if it was a successful read. If the caller * intends to issue a READ, the caller must clear B_INVAL and BIO_ERROR * prior to issuing the READ. biodone() will *not* clear B_INVAL. */ struct buf * getblk(struct vnode *vp, daddr_t blkno, int size, int slpflag, int slptimeo, int flags) { struct buf *bp; struct bufobj *bo; int bsize, error, maxsize, vmio; off_t offset; CTR3(KTR_BUF, "getblk(%p, %ld, %d)", vp, (long)blkno, size); KASSERT((flags & (GB_UNMAPPED | GB_KVAALLOC)) != GB_KVAALLOC, ("GB_KVAALLOC only makes sense with GB_UNMAPPED")); ASSERT_VOP_LOCKED(vp, "getblk"); if (size > MAXBCACHEBUF) panic("getblk: size(%d) > MAXBCACHEBUF(%d)\n", size, MAXBCACHEBUF); if (!unmapped_buf_allowed) flags &= ~(GB_UNMAPPED | GB_KVAALLOC); bo = &vp->v_bufobj; loop: BO_RLOCK(bo); bp = gbincore(bo, blkno); if (bp != NULL) { int lockflags; /* * Buffer is in-core. If the buffer is not busy nor managed, * it must be on a queue. */ lockflags = LK_EXCLUSIVE | LK_SLEEPFAIL | LK_INTERLOCK; if (flags & GB_LOCK_NOWAIT) lockflags |= LK_NOWAIT; error = BUF_TIMELOCK(bp, lockflags, BO_LOCKPTR(bo), "getblk", slpflag, slptimeo); /* * If we slept and got the lock we have to restart in case * the buffer changed identities. */ if (error == ENOLCK) goto loop; /* We timed out or were interrupted. */ else if (error) return (NULL); /* If recursed, assume caller knows the rules. */ else if (BUF_LOCKRECURSED(bp)) goto end; /* * The buffer is locked. B_CACHE is cleared if the buffer is * invalid. Otherwise, for a non-VMIO buffer, B_CACHE is set * and for a VMIO buffer B_CACHE is adjusted according to the * backing VM cache. */ if (bp->b_flags & B_INVAL) bp->b_flags &= ~B_CACHE; else if ((bp->b_flags & (B_VMIO | B_INVAL)) == 0) bp->b_flags |= B_CACHE; if (bp->b_flags & B_MANAGED) MPASS(bp->b_qindex == QUEUE_NONE); else bremfree(bp); /* * check for size inconsistencies for non-VMIO case. */ if (bp->b_bcount != size) { if ((bp->b_flags & B_VMIO) == 0 || (size > bp->b_kvasize)) { if (bp->b_flags & B_DELWRI) { bp->b_flags |= B_NOCACHE; bwrite(bp); } else { if (LIST_EMPTY(&bp->b_dep)) { bp->b_flags |= B_RELBUF; brelse(bp); } else { bp->b_flags |= B_NOCACHE; bwrite(bp); } } goto loop; } } /* * Handle the case of unmapped buffer which should * become mapped, or the buffer for which KVA * reservation is requested. */ bp_unmapped_get_kva(bp, blkno, size, flags); /* * If the size is inconsistent in the VMIO case, we can resize * the buffer. This might lead to B_CACHE getting set or * cleared. If the size has not changed, B_CACHE remains * unchanged from its previous state. */ allocbuf(bp, size); KASSERT(bp->b_offset != NOOFFSET, ("getblk: no buffer offset")); /* * A buffer with B_DELWRI set and B_CACHE clear must * be committed before we can return the buffer in * order to prevent the caller from issuing a read * ( due to B_CACHE not being set ) and overwriting * it. * * Most callers, including NFS and FFS, need this to * operate properly either because they assume they * can issue a read if B_CACHE is not set, or because * ( for example ) an uncached B_DELWRI might loop due * to softupdates re-dirtying the buffer. In the latter * case, B_CACHE is set after the first write completes, * preventing further loops. * NOTE! b*write() sets B_CACHE. If we cleared B_CACHE * above while extending the buffer, we cannot allow the * buffer to remain with B_CACHE set after the write * completes or it will represent a corrupt state. To * deal with this we set B_NOCACHE to scrap the buffer * after the write. * * We might be able to do something fancy, like setting * B_CACHE in bwrite() except if B_DELWRI is already set, * so the below call doesn't set B_CACHE, but that gets real * confusing. This is much easier. */ if ((bp->b_flags & (B_CACHE|B_DELWRI)) == B_DELWRI) { bp->b_flags |= B_NOCACHE; bwrite(bp); goto loop; } bp->b_flags &= ~B_DONE; } else { /* * Buffer is not in-core, create new buffer. The buffer * returned by getnewbuf() is locked. Note that the returned * buffer is also considered valid (not marked B_INVAL). */ BO_RUNLOCK(bo); /* * If the user does not want us to create the buffer, bail out * here. */ if (flags & GB_NOCREAT) return NULL; if (numfreebuffers == 0 && TD_IS_IDLETHREAD(curthread)) return NULL; bsize = vn_isdisk(vp, NULL) ? DEV_BSIZE : bo->bo_bsize; KASSERT(bsize != 0, ("bsize == 0, check bo->bo_bsize")); offset = blkno * bsize; vmio = vp->v_object != NULL; if (vmio) { maxsize = size + (offset & PAGE_MASK); } else { maxsize = size; /* Do not allow non-VMIO notmapped buffers. */ flags &= ~(GB_UNMAPPED | GB_KVAALLOC); } maxsize = imax(maxsize, bsize); bp = getnewbuf(vp, slpflag, slptimeo, maxsize, flags); if (bp == NULL) { if (slpflag || slptimeo) return NULL; /* * XXX This is here until the sleep path is diagnosed * enough to work under very low memory conditions. * * There's an issue on low memory, 4BSD+non-preempt * systems (eg MIPS routers with 32MB RAM) where buffer * exhaustion occurs without sleeping for buffer * reclaimation. This just sticks in a loop and * constantly attempts to allocate a buffer, which * hits exhaustion and tries to wakeup bufdaemon. * This never happens because we never yield. * * The real solution is to identify and fix these cases * so we aren't effectively busy-waiting in a loop * until the reclaimation path has cycles to run. */ kern_yield(PRI_USER); goto loop; } /* * This code is used to make sure that a buffer is not * created while the getnewbuf routine is blocked. * This can be a problem whether the vnode is locked or not. * If the buffer is created out from under us, we have to * throw away the one we just created. * * Note: this must occur before we associate the buffer * with the vp especially considering limitations in * the splay tree implementation when dealing with duplicate * lblkno's. */ BO_LOCK(bo); if (gbincore(bo, blkno)) { BO_UNLOCK(bo); bp->b_flags |= B_INVAL; brelse(bp); bufspace_release(maxsize); goto loop; } /* * Insert the buffer into the hash, so that it can * be found by incore. */ bp->b_blkno = bp->b_lblkno = blkno; bp->b_offset = offset; bgetvp(vp, bp); BO_UNLOCK(bo); /* * set B_VMIO bit. allocbuf() the buffer bigger. Since the * buffer size starts out as 0, B_CACHE will be set by * allocbuf() for the VMIO case prior to it testing the * backing store for validity. */ if (vmio) { bp->b_flags |= B_VMIO; KASSERT(vp->v_object == bp->b_bufobj->bo_object, ("ARGH! different b_bufobj->bo_object %p %p %p\n", bp, vp->v_object, bp->b_bufobj->bo_object)); } else { bp->b_flags &= ~B_VMIO; KASSERT(bp->b_bufobj->bo_object == NULL, ("ARGH! has b_bufobj->bo_object %p %p\n", bp, bp->b_bufobj->bo_object)); BUF_CHECK_MAPPED(bp); } allocbuf(bp, size); bufspace_release(maxsize); bp->b_flags &= ~B_DONE; } CTR4(KTR_BUF, "getblk(%p, %ld, %d) = %p", vp, (long)blkno, size, bp); BUF_ASSERT_HELD(bp); end: buf_track(bp, __func__); KASSERT(bp->b_bufobj == bo, ("bp %p wrong b_bufobj %p should be %p", bp, bp->b_bufobj, bo)); return (bp); } /* * Get an empty, disassociated buffer of given size. The buffer is initially * set to B_INVAL. */ struct buf * geteblk(int size, int flags) { struct buf *bp; int maxsize; maxsize = (size + BKVAMASK) & ~BKVAMASK; while ((bp = getnewbuf(NULL, 0, 0, maxsize, flags)) == NULL) { if ((flags & GB_NOWAIT_BD) && (curthread->td_pflags & TDP_BUFNEED) != 0) return (NULL); } allocbuf(bp, size); bufspace_release(maxsize); bp->b_flags |= B_INVAL; /* b_dep cleared by getnewbuf() */ BUF_ASSERT_HELD(bp); return (bp); } /* * Truncate the backing store for a non-vmio buffer. */ static void vfs_nonvmio_truncate(struct buf *bp, int newbsize) { if (bp->b_flags & B_MALLOC) { /* * malloced buffers are not shrunk */ if (newbsize == 0) { bufmallocadjust(bp, 0); free(bp->b_data, M_BIOBUF); bp->b_data = bp->b_kvabase; bp->b_flags &= ~B_MALLOC; } return; } vm_hold_free_pages(bp, newbsize); bufspace_adjust(bp, newbsize); } /* * Extend the backing for a non-VMIO buffer. */ static void vfs_nonvmio_extend(struct buf *bp, int newbsize) { caddr_t origbuf; int origbufsize; /* * We only use malloced memory on the first allocation. * and revert to page-allocated memory when the buffer * grows. * * There is a potential smp race here that could lead * to bufmallocspace slightly passing the max. It * is probably extremely rare and not worth worrying * over. */ if (bp->b_bufsize == 0 && newbsize <= PAGE_SIZE/2 && bufmallocspace < maxbufmallocspace) { bp->b_data = malloc(newbsize, M_BIOBUF, M_WAITOK); bp->b_flags |= B_MALLOC; bufmallocadjust(bp, newbsize); return; } /* * If the buffer is growing on its other-than-first * allocation then we revert to the page-allocation * scheme. */ origbuf = NULL; origbufsize = 0; if (bp->b_flags & B_MALLOC) { origbuf = bp->b_data; origbufsize = bp->b_bufsize; bp->b_data = bp->b_kvabase; bufmallocadjust(bp, 0); bp->b_flags &= ~B_MALLOC; newbsize = round_page(newbsize); } vm_hold_load_pages(bp, (vm_offset_t) bp->b_data + bp->b_bufsize, (vm_offset_t) bp->b_data + newbsize); if (origbuf != NULL) { bcopy(origbuf, bp->b_data, origbufsize); free(origbuf, M_BIOBUF); } bufspace_adjust(bp, newbsize); } /* * This code constitutes the buffer memory from either anonymous system * memory (in the case of non-VMIO operations) or from an associated * VM object (in the case of VMIO operations). This code is able to * resize a buffer up or down. * * Note that this code is tricky, and has many complications to resolve * deadlock or inconsistent data situations. Tread lightly!!! * There are B_CACHE and B_DELWRI interactions that must be dealt with by * the caller. Calling this code willy nilly can result in the loss of data. * * allocbuf() only adjusts B_CACHE for VMIO buffers. getblk() deals with * B_CACHE for the non-VMIO case. */ int allocbuf(struct buf *bp, int size) { int newbsize; BUF_ASSERT_HELD(bp); if (bp->b_bcount == size) return (1); if (bp->b_kvasize != 0 && bp->b_kvasize < size) panic("allocbuf: buffer too small"); newbsize = roundup2(size, DEV_BSIZE); if ((bp->b_flags & B_VMIO) == 0) { if ((bp->b_flags & B_MALLOC) == 0) newbsize = round_page(newbsize); /* * Just get anonymous memory from the kernel. Don't * mess with B_CACHE. */ if (newbsize < bp->b_bufsize) vfs_nonvmio_truncate(bp, newbsize); else if (newbsize > bp->b_bufsize) vfs_nonvmio_extend(bp, newbsize); } else { int desiredpages; desiredpages = (size == 0) ? 0 : num_pages((bp->b_offset & PAGE_MASK) + newbsize); if (bp->b_flags & B_MALLOC) panic("allocbuf: VMIO buffer can't be malloced"); /* * Set B_CACHE initially if buffer is 0 length or will become * 0-length. */ if (size == 0 || bp->b_bufsize == 0) bp->b_flags |= B_CACHE; if (newbsize < bp->b_bufsize) vfs_vmio_truncate(bp, desiredpages); /* XXX This looks as if it should be newbsize > b_bufsize */ else if (size > bp->b_bcount) vfs_vmio_extend(bp, desiredpages, size); bufspace_adjust(bp, newbsize); } bp->b_bcount = size; /* requested buffer size. */ return (1); } extern int inflight_transient_maps; void biodone(struct bio *bp) { struct mtx *mtxp; void (*done)(struct bio *); vm_offset_t start, end; biotrack(bp, __func__); if ((bp->bio_flags & BIO_TRANSIENT_MAPPING) != 0) { bp->bio_flags &= ~BIO_TRANSIENT_MAPPING; bp->bio_flags |= BIO_UNMAPPED; start = trunc_page((vm_offset_t)bp->bio_data); end = round_page((vm_offset_t)bp->bio_data + bp->bio_length); bp->bio_data = unmapped_buf; pmap_qremove(start, OFF_TO_IDX(end - start)); vmem_free(transient_arena, start, end - start); atomic_add_int(&inflight_transient_maps, -1); } done = bp->bio_done; if (done == NULL) { mtxp = mtx_pool_find(mtxpool_sleep, bp); mtx_lock(mtxp); bp->bio_flags |= BIO_DONE; wakeup(bp); mtx_unlock(mtxp); } else { bp->bio_flags |= BIO_DONE; done(bp); } } /* * Wait for a BIO to finish. */ int biowait(struct bio *bp, const char *wchan) { struct mtx *mtxp; mtxp = mtx_pool_find(mtxpool_sleep, bp); mtx_lock(mtxp); while ((bp->bio_flags & BIO_DONE) == 0) msleep(bp, mtxp, PRIBIO, wchan, 0); mtx_unlock(mtxp); if (bp->bio_error != 0) return (bp->bio_error); if (!(bp->bio_flags & BIO_ERROR)) return (0); return (EIO); } void biofinish(struct bio *bp, struct devstat *stat, int error) { if (error) { bp->bio_error = error; bp->bio_flags |= BIO_ERROR; } if (stat != NULL) devstat_end_transaction_bio(stat, bp); biodone(bp); } #if defined(BUF_TRACKING) || defined(FULL_BUF_TRACKING) void biotrack_buf(struct bio *bp, const char *location) { buf_track(bp->bio_track_bp, location); } #endif /* * bufwait: * * Wait for buffer I/O completion, returning error status. The buffer * is left locked and B_DONE on return. B_EINTR is converted into an EINTR * error and cleared. */ int bufwait(struct buf *bp) { if (bp->b_iocmd == BIO_READ) bwait(bp, PRIBIO, "biord"); else bwait(bp, PRIBIO, "biowr"); if (bp->b_flags & B_EINTR) { bp->b_flags &= ~B_EINTR; return (EINTR); } if (bp->b_ioflags & BIO_ERROR) { return (bp->b_error ? bp->b_error : EIO); } else { return (0); } } /* * bufdone: * * Finish I/O on a buffer, optionally calling a completion function. * This is usually called from an interrupt so process blocking is * not allowed. * * biodone is also responsible for setting B_CACHE in a B_VMIO bp. * In a non-VMIO bp, B_CACHE will be set on the next getblk() * assuming B_INVAL is clear. * * For the VMIO case, we set B_CACHE if the op was a read and no * read error occurred, or if the op was a write. B_CACHE is never * set if the buffer is invalid or otherwise uncacheable. * * biodone does not mess with B_INVAL, allowing the I/O routine or the * initiator to leave B_INVAL set to brelse the buffer out of existence * in the biodone routine. */ void bufdone(struct buf *bp) { struct bufobj *dropobj; void (*biodone)(struct buf *); buf_track(bp, __func__); CTR3(KTR_BUF, "bufdone(%p) vp %p flags %X", bp, bp->b_vp, bp->b_flags); dropobj = NULL; KASSERT(!(bp->b_flags & B_DONE), ("biodone: bp %p already done", bp)); BUF_ASSERT_HELD(bp); runningbufwakeup(bp); if (bp->b_iocmd == BIO_WRITE) dropobj = bp->b_bufobj; /* call optional completion function if requested */ if (bp->b_iodone != NULL) { biodone = bp->b_iodone; bp->b_iodone = NULL; (*biodone) (bp); if (dropobj) bufobj_wdrop(dropobj); return; } bufdone_finish(bp); if (dropobj) bufobj_wdrop(dropobj); } void bufdone_finish(struct buf *bp) { BUF_ASSERT_HELD(bp); if (!LIST_EMPTY(&bp->b_dep)) buf_complete(bp); if (bp->b_flags & B_VMIO) { /* * Set B_CACHE if the op was a normal read and no error * occurred. B_CACHE is set for writes in the b*write() * routines. */ if (bp->b_iocmd == BIO_READ && !(bp->b_flags & (B_INVAL|B_NOCACHE)) && !(bp->b_ioflags & BIO_ERROR)) bp->b_flags |= B_CACHE; vfs_vmio_iodone(bp); } /* * For asynchronous completions, release the buffer now. The brelse * will do a wakeup there if necessary - so no need to do a wakeup * here in the async case. The sync case always needs to do a wakeup. */ if (bp->b_flags & B_ASYNC) { if ((bp->b_flags & (B_NOCACHE | B_INVAL | B_RELBUF)) || (bp->b_ioflags & BIO_ERROR)) brelse(bp); else bqrelse(bp); } else bdone(bp); } /* * This routine is called in lieu of iodone in the case of * incomplete I/O. This keeps the busy status for pages * consistent. */ void vfs_unbusy_pages(struct buf *bp) { int i; vm_object_t obj; vm_page_t m; runningbufwakeup(bp); if (!(bp->b_flags & B_VMIO)) return; obj = bp->b_bufobj->bo_object; VM_OBJECT_WLOCK(obj); for (i = 0; i < bp->b_npages; i++) { m = bp->b_pages[i]; if (m == bogus_page) { m = vm_page_lookup(obj, OFF_TO_IDX(bp->b_offset) + i); if (!m) panic("vfs_unbusy_pages: page missing\n"); bp->b_pages[i] = m; if (buf_mapped(bp)) { BUF_CHECK_MAPPED(bp); pmap_qenter(trunc_page((vm_offset_t)bp->b_data), bp->b_pages, bp->b_npages); } else BUF_CHECK_UNMAPPED(bp); } vm_page_sunbusy(m); } vm_object_pip_wakeupn(obj, bp->b_npages); VM_OBJECT_WUNLOCK(obj); } /* * vfs_page_set_valid: * * Set the valid bits in a page based on the supplied offset. The * range is restricted to the buffer's size. * * This routine is typically called after a read completes. */ static void vfs_page_set_valid(struct buf *bp, vm_ooffset_t off, vm_page_t m) { vm_ooffset_t eoff; /* * Compute the end offset, eoff, such that [off, eoff) does not span a * page boundary and eoff is not greater than the end of the buffer. * The end of the buffer, in this case, is our file EOF, not the * allocation size of the buffer. */ eoff = (off + PAGE_SIZE) & ~(vm_ooffset_t)PAGE_MASK; if (eoff > bp->b_offset + bp->b_bcount) eoff = bp->b_offset + bp->b_bcount; /* * Set valid range. This is typically the entire buffer and thus the * entire page. */ if (eoff > off) vm_page_set_valid_range(m, off & PAGE_MASK, eoff - off); } /* * vfs_page_set_validclean: * * Set the valid bits and clear the dirty bits in a page based on the * supplied offset. The range is restricted to the buffer's size. */ static void vfs_page_set_validclean(struct buf *bp, vm_ooffset_t off, vm_page_t m) { vm_ooffset_t soff, eoff; /* * Start and end offsets in buffer. eoff - soff may not cross a * page boundary or cross the end of the buffer. The end of the * buffer, in this case, is our file EOF, not the allocation size * of the buffer. */ soff = off; eoff = (off + PAGE_SIZE) & ~(off_t)PAGE_MASK; if (eoff > bp->b_offset + bp->b_bcount) eoff = bp->b_offset + bp->b_bcount; /* * Set valid range. This is typically the entire buffer and thus the * entire page. */ if (eoff > soff) { vm_page_set_validclean( m, (vm_offset_t) (soff & PAGE_MASK), (vm_offset_t) (eoff - soff) ); } } /* * Ensure that all buffer pages are not exclusive busied. If any page is * exclusive busy, drain it. */ void vfs_drain_busy_pages(struct buf *bp) { vm_page_t m; int i, last_busied; VM_OBJECT_ASSERT_WLOCKED(bp->b_bufobj->bo_object); last_busied = 0; for (i = 0; i < bp->b_npages; i++) { m = bp->b_pages[i]; if (vm_page_xbusied(m)) { for (; last_busied < i; last_busied++) vm_page_sbusy(bp->b_pages[last_busied]); while (vm_page_xbusied(m)) { vm_page_lock(m); VM_OBJECT_WUNLOCK(bp->b_bufobj->bo_object); vm_page_busy_sleep(m, "vbpage", true); VM_OBJECT_WLOCK(bp->b_bufobj->bo_object); } } } for (i = 0; i < last_busied; i++) vm_page_sunbusy(bp->b_pages[i]); } /* * This routine is called before a device strategy routine. * It is used to tell the VM system that paging I/O is in * progress, and treat the pages associated with the buffer * almost as being exclusive busy. Also the object paging_in_progress * flag is handled to make sure that the object doesn't become * inconsistent. * * Since I/O has not been initiated yet, certain buffer flags * such as BIO_ERROR or B_INVAL may be in an inconsistent state * and should be ignored. */ void vfs_busy_pages(struct buf *bp, int clear_modify) { int i, bogus; vm_object_t obj; vm_ooffset_t foff; vm_page_t m; if (!(bp->b_flags & B_VMIO)) return; obj = bp->b_bufobj->bo_object; foff = bp->b_offset; KASSERT(bp->b_offset != NOOFFSET, ("vfs_busy_pages: no buffer offset")); VM_OBJECT_WLOCK(obj); vfs_drain_busy_pages(bp); if (bp->b_bufsize != 0) vfs_setdirty_locked_object(bp); bogus = 0; for (i = 0; i < bp->b_npages; i++) { m = bp->b_pages[i]; if ((bp->b_flags & B_CLUSTER) == 0) { vm_object_pip_add(obj, 1); vm_page_sbusy(m); } /* * When readying a buffer for a read ( i.e * clear_modify == 0 ), it is important to do * bogus_page replacement for valid pages in * partially instantiated buffers. Partially * instantiated buffers can, in turn, occur when * reconstituting a buffer from its VM backing store * base. We only have to do this if B_CACHE is * clear ( which causes the I/O to occur in the * first place ). The replacement prevents the read * I/O from overwriting potentially dirty VM-backed * pages. XXX bogus page replacement is, uh, bogus. * It may not work properly with small-block devices. * We need to find a better way. */ if (clear_modify) { pmap_remove_write(m); vfs_page_set_validclean(bp, foff, m); } else if (m->valid == VM_PAGE_BITS_ALL && (bp->b_flags & B_CACHE) == 0) { bp->b_pages[i] = bogus_page; bogus++; } foff = (foff + PAGE_SIZE) & ~(off_t)PAGE_MASK; } VM_OBJECT_WUNLOCK(obj); if (bogus && buf_mapped(bp)) { BUF_CHECK_MAPPED(bp); pmap_qenter(trunc_page((vm_offset_t)bp->b_data), bp->b_pages, bp->b_npages); } } /* * vfs_bio_set_valid: * * Set the range within the buffer to valid. The range is * relative to the beginning of the buffer, b_offset. Note that * b_offset itself may be offset from the beginning of the first * page. */ void vfs_bio_set_valid(struct buf *bp, int base, int size) { int i, n; vm_page_t m; if (!(bp->b_flags & B_VMIO)) return; /* * Fixup base to be relative to beginning of first page. * Set initial n to be the maximum number of bytes in the * first page that can be validated. */ base += (bp->b_offset & PAGE_MASK); n = PAGE_SIZE - (base & PAGE_MASK); VM_OBJECT_WLOCK(bp->b_bufobj->bo_object); for (i = base / PAGE_SIZE; size > 0 && i < bp->b_npages; ++i) { m = bp->b_pages[i]; if (n > size) n = size; vm_page_set_valid_range(m, base & PAGE_MASK, n); base += n; size -= n; n = PAGE_SIZE; } VM_OBJECT_WUNLOCK(bp->b_bufobj->bo_object); } /* * vfs_bio_clrbuf: * * If the specified buffer is a non-VMIO buffer, clear the entire * buffer. If the specified buffer is a VMIO buffer, clear and * validate only the previously invalid portions of the buffer. * This routine essentially fakes an I/O, so we need to clear * BIO_ERROR and B_INVAL. * * Note that while we only theoretically need to clear through b_bcount, * we go ahead and clear through b_bufsize. */ void vfs_bio_clrbuf(struct buf *bp) { int i, j, mask, sa, ea, slide; if ((bp->b_flags & (B_VMIO | B_MALLOC)) != B_VMIO) { clrbuf(bp); return; } bp->b_flags &= ~B_INVAL; bp->b_ioflags &= ~BIO_ERROR; VM_OBJECT_WLOCK(bp->b_bufobj->bo_object); if ((bp->b_npages == 1) && (bp->b_bufsize < PAGE_SIZE) && (bp->b_offset & PAGE_MASK) == 0) { if (bp->b_pages[0] == bogus_page) goto unlock; mask = (1 << (bp->b_bufsize / DEV_BSIZE)) - 1; VM_OBJECT_ASSERT_WLOCKED(bp->b_pages[0]->object); if ((bp->b_pages[0]->valid & mask) == mask) goto unlock; if ((bp->b_pages[0]->valid & mask) == 0) { pmap_zero_page_area(bp->b_pages[0], 0, bp->b_bufsize); bp->b_pages[0]->valid |= mask; goto unlock; } } sa = bp->b_offset & PAGE_MASK; slide = 0; for (i = 0; i < bp->b_npages; i++, sa = 0) { slide = imin(slide + PAGE_SIZE, bp->b_offset + bp->b_bufsize); ea = slide & PAGE_MASK; if (ea == 0) ea = PAGE_SIZE; if (bp->b_pages[i] == bogus_page) continue; j = sa / DEV_BSIZE; mask = ((1 << ((ea - sa) / DEV_BSIZE)) - 1) << j; VM_OBJECT_ASSERT_WLOCKED(bp->b_pages[i]->object); if ((bp->b_pages[i]->valid & mask) == mask) continue; if ((bp->b_pages[i]->valid & mask) == 0) pmap_zero_page_area(bp->b_pages[i], sa, ea - sa); else { for (; sa < ea; sa += DEV_BSIZE, j++) { if ((bp->b_pages[i]->valid & (1 << j)) == 0) { pmap_zero_page_area(bp->b_pages[i], sa, DEV_BSIZE); } } } bp->b_pages[i]->valid |= mask; } unlock: VM_OBJECT_WUNLOCK(bp->b_bufobj->bo_object); bp->b_resid = 0; } void vfs_bio_bzero_buf(struct buf *bp, int base, int size) { vm_page_t m; int i, n; if (buf_mapped(bp)) { BUF_CHECK_MAPPED(bp); bzero(bp->b_data + base, size); } else { BUF_CHECK_UNMAPPED(bp); n = PAGE_SIZE - (base & PAGE_MASK); for (i = base / PAGE_SIZE; size > 0 && i < bp->b_npages; ++i) { m = bp->b_pages[i]; if (n > size) n = size; pmap_zero_page_area(m, base & PAGE_MASK, n); base += n; size -= n; n = PAGE_SIZE; } } } /* * Update buffer flags based on I/O request parameters, optionally releasing the * buffer. If it's VMIO or direct I/O, the buffer pages are released to the VM, * where they may be placed on a page queue (VMIO) or freed immediately (direct * I/O). Otherwise the buffer is released to the cache. */ static void b_io_dismiss(struct buf *bp, int ioflag, bool release) { KASSERT((ioflag & IO_NOREUSE) == 0 || (ioflag & IO_VMIO) != 0, ("buf %p non-VMIO noreuse", bp)); if ((ioflag & IO_DIRECT) != 0) bp->b_flags |= B_DIRECT; if ((ioflag & (IO_VMIO | IO_DIRECT)) != 0 && LIST_EMPTY(&bp->b_dep)) { bp->b_flags |= B_RELBUF; if ((ioflag & IO_NOREUSE) != 0) bp->b_flags |= B_NOREUSE; if (release) brelse(bp); } else if (release) bqrelse(bp); } void vfs_bio_brelse(struct buf *bp, int ioflag) { b_io_dismiss(bp, ioflag, true); } void vfs_bio_set_flags(struct buf *bp, int ioflag) { b_io_dismiss(bp, ioflag, false); } /* * vm_hold_load_pages and vm_hold_free_pages get pages into * a buffers address space. The pages are anonymous and are * not associated with a file object. */ static void vm_hold_load_pages(struct buf *bp, vm_offset_t from, vm_offset_t to) { vm_offset_t pg; vm_page_t p; int index; BUF_CHECK_MAPPED(bp); to = round_page(to); from = round_page(from); index = (from - trunc_page((vm_offset_t)bp->b_data)) >> PAGE_SHIFT; for (pg = from; pg < to; pg += PAGE_SIZE, index++) { tryagain: /* * note: must allocate system pages since blocking here * could interfere with paging I/O, no matter which * process we are. */ p = vm_page_alloc(NULL, 0, VM_ALLOC_SYSTEM | VM_ALLOC_NOOBJ | VM_ALLOC_WIRED | VM_ALLOC_COUNT((to - pg) >> PAGE_SHIFT)); if (p == NULL) { VM_WAIT; goto tryagain; } pmap_qenter(pg, &p, 1); bp->b_pages[index] = p; } bp->b_npages = index; } /* Return pages associated with this buf to the vm system */ static void vm_hold_free_pages(struct buf *bp, int newbsize) { vm_offset_t from; vm_page_t p; int index, newnpages; BUF_CHECK_MAPPED(bp); from = round_page((vm_offset_t)bp->b_data + newbsize); newnpages = (from - trunc_page((vm_offset_t)bp->b_data)) >> PAGE_SHIFT; if (bp->b_npages > newnpages) pmap_qremove(from, bp->b_npages - newnpages); for (index = newnpages; index < bp->b_npages; index++) { p = bp->b_pages[index]; bp->b_pages[index] = NULL; if (vm_page_sbusied(p)) printf("vm_hold_free_pages: blkno: %jd, lblkno: %jd\n", (intmax_t)bp->b_blkno, (intmax_t)bp->b_lblkno); p->wire_count--; vm_page_free(p); atomic_subtract_int(&vm_cnt.v_wire_count, 1); } bp->b_npages = newnpages; } /* * Map an IO request into kernel virtual address space. * * All requests are (re)mapped into kernel VA space. * Notice that we use b_bufsize for the size of the buffer * to be mapped. b_bcount might be modified by the driver. * * Note that even if the caller determines that the address space should * be valid, a race or a smaller-file mapped into a larger space may * actually cause vmapbuf() to fail, so all callers of vmapbuf() MUST * check the return value. * * This function only works with pager buffers. */ int vmapbuf(struct buf *bp, int mapbuf) { vm_prot_t prot; int pidx; if (bp->b_bufsize < 0) return (-1); prot = VM_PROT_READ; if (bp->b_iocmd == BIO_READ) prot |= VM_PROT_WRITE; /* Less backwards than it looks */ if ((pidx = vm_fault_quick_hold_pages(&curproc->p_vmspace->vm_map, (vm_offset_t)bp->b_data, bp->b_bufsize, prot, bp->b_pages, btoc(MAXPHYS))) < 0) return (-1); bp->b_npages = pidx; bp->b_offset = ((vm_offset_t)bp->b_data) & PAGE_MASK; if (mapbuf || !unmapped_buf_allowed) { pmap_qenter((vm_offset_t)bp->b_kvabase, bp->b_pages, pidx); bp->b_data = bp->b_kvabase + bp->b_offset; } else bp->b_data = unmapped_buf; return(0); } /* * Free the io map PTEs associated with this IO operation. * We also invalidate the TLB entries and restore the original b_addr. * * This function only works with pager buffers. */ void vunmapbuf(struct buf *bp) { int npages; npages = bp->b_npages; if (buf_mapped(bp)) pmap_qremove(trunc_page((vm_offset_t)bp->b_data), npages); vm_page_unhold_pages(bp->b_pages, npages); bp->b_data = unmapped_buf; } void bdone(struct buf *bp) { struct mtx *mtxp; mtxp = mtx_pool_find(mtxpool_sleep, bp); mtx_lock(mtxp); bp->b_flags |= B_DONE; wakeup(bp); mtx_unlock(mtxp); } void bwait(struct buf *bp, u_char pri, const char *wchan) { struct mtx *mtxp; mtxp = mtx_pool_find(mtxpool_sleep, bp); mtx_lock(mtxp); while ((bp->b_flags & B_DONE) == 0) msleep(bp, mtxp, pri, wchan, 0); mtx_unlock(mtxp); } int bufsync(struct bufobj *bo, int waitfor) { return (VOP_FSYNC(bo2vnode(bo), waitfor, curthread)); } void bufstrategy(struct bufobj *bo, struct buf *bp) { int i = 0; struct vnode *vp; vp = bp->b_vp; KASSERT(vp == bo->bo_private, ("Inconsistent vnode bufstrategy")); KASSERT(vp->v_type != VCHR && vp->v_type != VBLK, ("Wrong vnode in bufstrategy(bp=%p, vp=%p)", bp, vp)); i = VOP_STRATEGY(vp, bp); KASSERT(i == 0, ("VOP_STRATEGY failed bp=%p vp=%p", bp, bp->b_vp)); } void bufobj_wrefl(struct bufobj *bo) { KASSERT(bo != NULL, ("NULL bo in bufobj_wref")); ASSERT_BO_WLOCKED(bo); bo->bo_numoutput++; } void bufobj_wref(struct bufobj *bo) { KASSERT(bo != NULL, ("NULL bo in bufobj_wref")); BO_LOCK(bo); bo->bo_numoutput++; BO_UNLOCK(bo); } void bufobj_wdrop(struct bufobj *bo) { KASSERT(bo != NULL, ("NULL bo in bufobj_wdrop")); BO_LOCK(bo); KASSERT(bo->bo_numoutput > 0, ("bufobj_wdrop non-positive count")); if ((--bo->bo_numoutput == 0) && (bo->bo_flag & BO_WWAIT)) { bo->bo_flag &= ~BO_WWAIT; wakeup(&bo->bo_numoutput); } BO_UNLOCK(bo); } int bufobj_wwait(struct bufobj *bo, int slpflag, int timeo) { int error; KASSERT(bo != NULL, ("NULL bo in bufobj_wwait")); ASSERT_BO_WLOCKED(bo); error = 0; while (bo->bo_numoutput) { bo->bo_flag |= BO_WWAIT; error = msleep(&bo->bo_numoutput, BO_LOCKPTR(bo), slpflag | (PRIBIO + 1), "bo_wwait", timeo); if (error) break; } return (error); } /* * Set bio_data or bio_ma for struct bio from the struct buf. */ void bdata2bio(struct buf *bp, struct bio *bip) { if (!buf_mapped(bp)) { KASSERT(unmapped_buf_allowed, ("unmapped")); bip->bio_ma = bp->b_pages; bip->bio_ma_n = bp->b_npages; bip->bio_data = unmapped_buf; bip->bio_ma_offset = (vm_offset_t)bp->b_offset & PAGE_MASK; bip->bio_flags |= BIO_UNMAPPED; KASSERT(round_page(bip->bio_ma_offset + bip->bio_length) / PAGE_SIZE == bp->b_npages, ("Buffer %p too short: %d %lld %d", bp, bip->bio_ma_offset, (long long)bip->bio_length, bip->bio_ma_n)); } else { bip->bio_data = bp->b_data; bip->bio_ma = NULL; } } /* * The MIPS pmap code currently doesn't handle aliased pages. * The VIPT caches may not handle page aliasing themselves, leading * to data corruption. * * As such, this code makes a system extremely unhappy if said * system doesn't support unaliasing the above situation in hardware. * Some "recent" systems (eg some mips24k/mips74k cores) don't enable * this feature at build time, so it has to be handled in software. * * Once the MIPS pmap/cache code grows to support this function on * earlier chips, it should be flipped back off. */ #ifdef __mips__ static int buf_pager_relbuf = 1; #else static int buf_pager_relbuf = 0; #endif SYSCTL_INT(_vfs, OID_AUTO, buf_pager_relbuf, CTLFLAG_RWTUN, &buf_pager_relbuf, 0, "Make buffer pager release buffers after reading"); /* * The buffer pager. It uses buffer reads to validate pages. * * In contrast to the generic local pager from vm/vnode_pager.c, this * pager correctly and easily handles volumes where the underlying * device block size is greater than the machine page size. The * buffer cache transparently extends the requested page run to be * aligned at the block boundary, and does the necessary bogus page * replacements in the addends to avoid obliterating already valid * pages. * * The only non-trivial issue is that the exclusive busy state for * pages, which is assumed by the vm_pager_getpages() interface, is * incompatible with the VMIO buffer cache's desire to share-busy the * pages. This function performs a trivial downgrade of the pages' * state before reading buffers, and a less trivial upgrade from the * shared-busy to excl-busy state after the read. */ int vfs_bio_getpages(struct vnode *vp, vm_page_t *ma, int count, int *rbehind, int *rahead, vbg_get_lblkno_t get_lblkno, vbg_get_blksize_t get_blksize) { vm_page_t m; vm_object_t object; struct buf *bp; struct mount *mp; daddr_t lbn, lbnp; vm_ooffset_t la, lb, poff, poffe; long bsize; int bo_bs, br_flags, error, i, pgsin, pgsin_a, pgsin_b; bool redo, lpart; object = vp->v_object; mp = vp->v_mount; la = IDX_TO_OFF(ma[count - 1]->pindex); if (la >= object->un_pager.vnp.vnp_size) return (VM_PAGER_BAD); lpart = la + PAGE_SIZE > object->un_pager.vnp.vnp_size; bo_bs = get_blksize(vp, get_lblkno(vp, IDX_TO_OFF(ma[0]->pindex))); /* * Calculate read-ahead, behind and total pages. */ pgsin = count; lb = IDX_TO_OFF(ma[0]->pindex); pgsin_b = OFF_TO_IDX(lb - rounddown2(lb, bo_bs)); pgsin += pgsin_b; if (rbehind != NULL) *rbehind = pgsin_b; pgsin_a = OFF_TO_IDX(roundup2(la, bo_bs) - la); if (la + IDX_TO_OFF(pgsin_a) >= object->un_pager.vnp.vnp_size) pgsin_a = OFF_TO_IDX(roundup2(object->un_pager.vnp.vnp_size, PAGE_SIZE) - la); pgsin += pgsin_a; if (rahead != NULL) *rahead = pgsin_a; PCPU_INC(cnt.v_vnodein); PCPU_ADD(cnt.v_vnodepgsin, pgsin); br_flags = (mp != NULL && (mp->mnt_kern_flag & MNTK_UNMAPPED_BUFS) != 0) ? GB_UNMAPPED : 0; VM_OBJECT_WLOCK(object); again: for (i = 0; i < count; i++) vm_page_busy_downgrade(ma[i]); VM_OBJECT_WUNLOCK(object); lbnp = -1; for (i = 0; i < count; i++) { m = ma[i]; /* * Pages are shared busy and the object lock is not * owned, which together allow for the pages' * invalidation. The racy test for validity avoids * useless creation of the buffer for the most typical * case when invalidation is not used in redo or for * parallel read. The shared->excl upgrade loop at * the end of the function catches the race in a * reliable way (protected by the object lock). */ if (m->valid == VM_PAGE_BITS_ALL) continue; poff = IDX_TO_OFF(m->pindex); poffe = MIN(poff + PAGE_SIZE, object->un_pager.vnp.vnp_size); for (; poff < poffe; poff += bsize) { lbn = get_lblkno(vp, poff); if (lbn == lbnp) goto next_page; lbnp = lbn; bsize = get_blksize(vp, lbn); error = bread_gb(vp, lbn, bsize, curthread->td_ucred, br_flags, &bp); if (error != 0) goto end_pages; if (LIST_EMPTY(&bp->b_dep)) { /* * Invalidation clears m->valid, but * may leave B_CACHE flag if the * buffer existed at the invalidation * time. In this case, recycle the * buffer to do real read on next * bread() after redo. * * Otherwise B_RELBUF is not strictly * necessary, enable to reduce buf * cache pressure. */ if (buf_pager_relbuf || m->valid != VM_PAGE_BITS_ALL) bp->b_flags |= B_RELBUF; bp->b_flags &= ~B_NOCACHE; brelse(bp); } else { bqrelse(bp); } } KASSERT(1 /* racy, enable for debugging */ || m->valid == VM_PAGE_BITS_ALL || i == count - 1, ("buf %d %p invalid", i, m)); if (i == count - 1 && lpart) { VM_OBJECT_WLOCK(object); if (m->valid != 0 && m->valid != VM_PAGE_BITS_ALL) vm_page_zero_invalid(m, TRUE); VM_OBJECT_WUNLOCK(object); } next_page:; } end_pages: VM_OBJECT_WLOCK(object); redo = false; for (i = 0; i < count; i++) { vm_page_sunbusy(ma[i]); ma[i] = vm_page_grab(object, ma[i]->pindex, VM_ALLOC_NORMAL); /* * Since the pages were only sbusy while neither the * buffer nor the object lock was held by us, or * reallocated while vm_page_grab() slept for busy * relinguish, they could have been invalidated. * Recheck the valid bits and re-read as needed. * * Note that the last page is made fully valid in the * read loop, and partial validity for the page at * index count - 1 could mean that the page was * invalidated or removed, so we must restart for * safety as well. */ if (ma[i]->valid != VM_PAGE_BITS_ALL) redo = true; } if (redo && error == 0) goto again; VM_OBJECT_WUNLOCK(object); return (error != 0 ? VM_PAGER_ERROR : VM_PAGER_OK); } #include "opt_ddb.h" #ifdef DDB #include /* DDB command to show buffer data */ DB_SHOW_COMMAND(buffer, db_show_buffer) { /* get args */ struct buf *bp = (struct buf *)addr; #ifdef FULL_BUF_TRACKING uint32_t i, j; #endif if (!have_addr) { db_printf("usage: show buffer \n"); return; } db_printf("buf at %p\n", bp); db_printf("b_flags = 0x%b, b_xflags=0x%b, b_vflags=0x%b\n", (u_int)bp->b_flags, PRINT_BUF_FLAGS, (u_int)bp->b_xflags, PRINT_BUF_XFLAGS, (u_int)bp->b_vflags, PRINT_BUF_VFLAGS); db_printf( "b_error = %d, b_bufsize = %ld, b_bcount = %ld, b_resid = %ld\n" "b_bufobj = (%p), b_data = %p, b_blkno = %jd, b_lblkno = %jd, " "b_dep = %p\n", bp->b_error, bp->b_bufsize, bp->b_bcount, bp->b_resid, bp->b_bufobj, bp->b_data, (intmax_t)bp->b_blkno, (intmax_t)bp->b_lblkno, bp->b_dep.lh_first); db_printf("b_kvabase = %p, b_kvasize = %d\n", bp->b_kvabase, bp->b_kvasize); if (bp->b_npages) { int i; db_printf("b_npages = %d, pages(OBJ, IDX, PA): ", bp->b_npages); for (i = 0; i < bp->b_npages; i++) { vm_page_t m; m = bp->b_pages[i]; if (m != NULL) db_printf("(%p, 0x%lx, 0x%lx)", m->object, (u_long)m->pindex, (u_long)VM_PAGE_TO_PHYS(m)); else db_printf("( ??? )"); if ((i + 1) < bp->b_npages) db_printf(","); } db_printf("\n"); } #if defined(FULL_BUF_TRACKING) db_printf("b_io_tracking: b_io_tcnt = %u\n", bp->b_io_tcnt); i = bp->b_io_tcnt % BUF_TRACKING_SIZE; for (j = 1; j <= BUF_TRACKING_SIZE; j++) db_printf(" %2u: %s\n", j, bp->b_io_tracking[BUF_TRACKING_ENTRY(i - j)]); #elif defined(BUF_TRACKING) db_printf("b_io_tracking: %s\n", bp->b_io_tracking); #endif db_printf(" "); BUF_LOCKPRINTINFO(bp); } DB_SHOW_COMMAND(lockedbufs, lockedbufs) { struct buf *bp; int i; for (i = 0; i < nbuf; i++) { bp = &buf[i]; if (BUF_ISLOCKED(bp)) { db_show_buffer((uintptr_t)bp, 1, 0, NULL); db_printf("\n"); } } } DB_SHOW_COMMAND(vnodebufs, db_show_vnodebufs) { struct vnode *vp; struct buf *bp; if (!have_addr) { db_printf("usage: show vnodebufs \n"); return; } vp = (struct vnode *)addr; db_printf("Clean buffers:\n"); TAILQ_FOREACH(bp, &vp->v_bufobj.bo_clean.bv_hd, b_bobufs) { db_show_buffer((uintptr_t)bp, 1, 0, NULL); db_printf("\n"); } db_printf("Dirty buffers:\n"); TAILQ_FOREACH(bp, &vp->v_bufobj.bo_dirty.bv_hd, b_bobufs) { db_show_buffer((uintptr_t)bp, 1, 0, NULL); db_printf("\n"); } } DB_COMMAND(countfreebufs, db_coundfreebufs) { struct buf *bp; int i, used = 0, nfree = 0; if (have_addr) { db_printf("usage: countfreebufs\n"); return; } for (i = 0; i < nbuf; i++) { bp = &buf[i]; if (bp->b_qindex == QUEUE_EMPTY) nfree++; else used++; } db_printf("Counted %d free, %d used (%d tot)\n", nfree, used, nfree + used); db_printf("numfreebuffers is %d\n", numfreebuffers); } #endif /* DDB */ Index: projects/clang400-import/sys/kern/vfs_cluster.c =================================================================== --- projects/clang400-import/sys/kern/vfs_cluster.c (revision 311459) +++ projects/clang400-import/sys/kern/vfs_cluster.c (revision 311460) @@ -1,1064 +1,1061 @@ /*- * Copyright (c) 1993 * The Regents of the University of California. All rights reserved. * Modifications/enhancements: * Copyright (c) 1995 John S. Dyson. 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. * * @(#)vfs_cluster.c 8.7 (Berkeley) 2/13/94 */ #include __FBSDID("$FreeBSD$"); #include "opt_debug_cluster.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #if defined(CLUSTERDEBUG) static int rcluster= 0; SYSCTL_INT(_debug, OID_AUTO, rcluster, CTLFLAG_RW, &rcluster, 0, "Debug VFS clustering code"); #endif static MALLOC_DEFINE(M_SEGMENT, "cl_savebuf", "cluster_save buffer"); static struct cluster_save *cluster_collectbufs(struct vnode *vp, struct buf *last_bp, int gbflags); static struct buf *cluster_rbuild(struct vnode *vp, u_quad_t filesize, daddr_t lbn, daddr_t blkno, long size, int run, int gbflags, struct buf *fbp); static void cluster_callback(struct buf *); static int write_behind = 1; SYSCTL_INT(_vfs, OID_AUTO, write_behind, CTLFLAG_RW, &write_behind, 0, "Cluster write-behind; 0: disable, 1: enable, 2: backed off"); static int read_max = 64; SYSCTL_INT(_vfs, OID_AUTO, read_max, CTLFLAG_RW, &read_max, 0, "Cluster read-ahead max block count"); static int read_min = 1; SYSCTL_INT(_vfs, OID_AUTO, read_min, CTLFLAG_RW, &read_min, 0, "Cluster read min block count"); -/* Page expended to mark partially backed buffers */ -extern vm_page_t bogus_page; - /* * Read data to a buf, including read-ahead if we find this to be beneficial. * cluster_read replaces bread. */ int cluster_read(struct vnode *vp, u_quad_t filesize, daddr_t lblkno, long size, struct ucred *cred, long totread, int seqcount, int gbflags, struct buf **bpp) { struct buf *bp, *rbp, *reqbp; struct bufobj *bo; daddr_t blkno, origblkno; int maxra, racluster; int error, ncontig; int i; error = 0; bo = &vp->v_bufobj; if (!unmapped_buf_allowed) gbflags &= ~GB_UNMAPPED; /* * Try to limit the amount of read-ahead by a few * ad-hoc parameters. This needs work!!! */ racluster = vp->v_mount->mnt_iosize_max / size; maxra = seqcount; maxra = min(read_max, maxra); maxra = min(nbuf/8, maxra); if (((u_quad_t)(lblkno + maxra + 1) * size) > filesize) maxra = (filesize / size) - lblkno; /* * get the requested block */ *bpp = reqbp = bp = getblk(vp, lblkno, size, 0, 0, gbflags); if (bp == NULL) return (EBUSY); origblkno = lblkno; /* * if it is in the cache, then check to see if the reads have been * sequential. If they have, then try some read-ahead, otherwise * back-off on prospective read-aheads. */ if (bp->b_flags & B_CACHE) { if (!seqcount) { return 0; } else if ((bp->b_flags & B_RAM) == 0) { return 0; } else { bp->b_flags &= ~B_RAM; BO_RLOCK(bo); for (i = 1; i < maxra; i++) { /* * Stop if the buffer does not exist or it * is invalid (about to go away?) */ rbp = gbincore(&vp->v_bufobj, lblkno+i); if (rbp == NULL || (rbp->b_flags & B_INVAL)) break; /* * Set another read-ahead mark so we know * to check again. (If we can lock the * buffer without waiting) */ if ((((i % racluster) == (racluster - 1)) || (i == (maxra - 1))) && (0 == BUF_LOCK(rbp, LK_EXCLUSIVE | LK_NOWAIT, NULL))) { rbp->b_flags |= B_RAM; BUF_UNLOCK(rbp); } } BO_RUNLOCK(bo); if (i >= maxra) { return 0; } lblkno += i; } reqbp = bp = NULL; /* * If it isn't in the cache, then get a chunk from * disk if sequential, otherwise just get the block. */ } else { off_t firstread = bp->b_offset; int nblks; long minread; KASSERT(bp->b_offset != NOOFFSET, ("cluster_read: no buffer offset")); ncontig = 0; /* * Adjust totread if needed */ minread = read_min * size; if (minread > totread) totread = minread; /* * Compute the total number of blocks that we should read * synchronously. */ if (firstread + totread > filesize) totread = filesize - firstread; nblks = howmany(totread, size); if (nblks > racluster) nblks = racluster; /* * Now compute the number of contiguous blocks. */ if (nblks > 1) { error = VOP_BMAP(vp, lblkno, NULL, &blkno, &ncontig, NULL); /* * If this failed to map just do the original block. */ if (error || blkno == -1) ncontig = 0; } /* * If we have contiguous data available do a cluster * otherwise just read the requested block. */ if (ncontig) { /* Account for our first block. */ ncontig = min(ncontig + 1, nblks); if (ncontig < nblks) nblks = ncontig; bp = cluster_rbuild(vp, filesize, lblkno, blkno, size, nblks, gbflags, bp); lblkno += (bp->b_bufsize / size); } else { bp->b_flags |= B_RAM; bp->b_iocmd = BIO_READ; lblkno += 1; } } /* * handle the synchronous read so that it is available ASAP. */ if (bp) { if ((bp->b_flags & B_CLUSTER) == 0) { vfs_busy_pages(bp, 0); } bp->b_flags &= ~B_INVAL; bp->b_ioflags &= ~BIO_ERROR; if ((bp->b_flags & B_ASYNC) || bp->b_iodone != NULL) BUF_KERNPROC(bp); bp->b_iooffset = dbtob(bp->b_blkno); bstrategy(bp); #ifdef RACCT if (racct_enable) { PROC_LOCK(curproc); racct_add_buf(curproc, bp, 0); PROC_UNLOCK(curproc); } #endif /* RACCT */ curthread->td_ru.ru_inblock++; } /* * If we have been doing sequential I/O, then do some read-ahead. */ while (lblkno < (origblkno + maxra)) { error = VOP_BMAP(vp, lblkno, NULL, &blkno, &ncontig, NULL); if (error) break; if (blkno == -1) break; /* * We could throttle ncontig here by maxra but we might as * well read the data if it is contiguous. We're throttled * by racluster anyway. */ if (ncontig) { ncontig = min(ncontig + 1, racluster); rbp = cluster_rbuild(vp, filesize, lblkno, blkno, size, ncontig, gbflags, NULL); lblkno += (rbp->b_bufsize / size); if (rbp->b_flags & B_DELWRI) { bqrelse(rbp); continue; } } else { rbp = getblk(vp, lblkno, size, 0, 0, gbflags); lblkno += 1; if (rbp->b_flags & B_DELWRI) { bqrelse(rbp); continue; } rbp->b_flags |= B_ASYNC | B_RAM; rbp->b_iocmd = BIO_READ; rbp->b_blkno = blkno; } if (rbp->b_flags & B_CACHE) { rbp->b_flags &= ~B_ASYNC; bqrelse(rbp); continue; } if ((rbp->b_flags & B_CLUSTER) == 0) { vfs_busy_pages(rbp, 0); } rbp->b_flags &= ~B_INVAL; rbp->b_ioflags &= ~BIO_ERROR; if ((rbp->b_flags & B_ASYNC) || rbp->b_iodone != NULL) BUF_KERNPROC(rbp); rbp->b_iooffset = dbtob(rbp->b_blkno); bstrategy(rbp); #ifdef RACCT if (racct_enable) { PROC_LOCK(curproc); racct_add_buf(curproc, rbp, 0); PROC_UNLOCK(curproc); } #endif /* RACCT */ curthread->td_ru.ru_inblock++; } if (reqbp) { /* * Like bread, always brelse() the buffer when * returning an error. */ error = bufwait(reqbp); if (error != 0) { brelse(reqbp); *bpp = NULL; } } return (error); } /* * If blocks are contiguous on disk, use this to provide clustered * read ahead. We will read as many blocks as possible sequentially * and then parcel them up into logical blocks in the buffer hash table. */ static struct buf * cluster_rbuild(struct vnode *vp, u_quad_t filesize, daddr_t lbn, daddr_t blkno, long size, int run, int gbflags, struct buf *fbp) { struct buf *bp, *tbp; daddr_t bn; off_t off; long tinc, tsize; int i, inc, j, k, toff; KASSERT(size == vp->v_mount->mnt_stat.f_iosize, ("cluster_rbuild: size %ld != f_iosize %jd\n", size, (intmax_t)vp->v_mount->mnt_stat.f_iosize)); /* * avoid a division */ while ((u_quad_t) size * (lbn + run) > filesize) { --run; } if (fbp) { tbp = fbp; tbp->b_iocmd = BIO_READ; } else { tbp = getblk(vp, lbn, size, 0, 0, gbflags); if (tbp->b_flags & B_CACHE) return tbp; tbp->b_flags |= B_ASYNC | B_RAM; tbp->b_iocmd = BIO_READ; } tbp->b_blkno = blkno; if( (tbp->b_flags & B_MALLOC) || ((tbp->b_flags & B_VMIO) == 0) || (run <= 1) ) return tbp; bp = trypbuf(&cluster_pbuf_freecnt); if (bp == NULL) return tbp; /* * We are synthesizing a buffer out of vm_page_t's, but * if the block size is not page aligned then the starting * address may not be either. Inherit the b_data offset * from the original buffer. */ bp->b_flags = B_ASYNC | B_CLUSTER | B_VMIO; if ((gbflags & GB_UNMAPPED) != 0) { bp->b_data = unmapped_buf; } else { bp->b_data = (char *)((vm_offset_t)bp->b_data | ((vm_offset_t)tbp->b_data & PAGE_MASK)); } bp->b_iocmd = BIO_READ; bp->b_iodone = cluster_callback; bp->b_blkno = blkno; bp->b_lblkno = lbn; bp->b_offset = tbp->b_offset; KASSERT(bp->b_offset != NOOFFSET, ("cluster_rbuild: no buffer offset")); pbgetvp(vp, bp); TAILQ_INIT(&bp->b_cluster.cluster_head); bp->b_bcount = 0; bp->b_bufsize = 0; bp->b_npages = 0; inc = btodb(size); for (bn = blkno, i = 0; i < run; ++i, bn += inc) { if (i == 0) { VM_OBJECT_WLOCK(tbp->b_bufobj->bo_object); vfs_drain_busy_pages(tbp); vm_object_pip_add(tbp->b_bufobj->bo_object, tbp->b_npages); for (k = 0; k < tbp->b_npages; k++) vm_page_sbusy(tbp->b_pages[k]); VM_OBJECT_WUNLOCK(tbp->b_bufobj->bo_object); } else { if ((bp->b_npages * PAGE_SIZE) + round_page(size) > vp->v_mount->mnt_iosize_max) { break; } tbp = getblk(vp, lbn + i, size, 0, 0, GB_LOCK_NOWAIT | (gbflags & GB_UNMAPPED)); /* Don't wait around for locked bufs. */ if (tbp == NULL) break; /* * Stop scanning if the buffer is fully valid * (marked B_CACHE), or locked (may be doing a * background write), or if the buffer is not * VMIO backed. The clustering code can only deal * with VMIO-backed buffers. The bo lock is not * required for the BKGRDINPROG check since it * can not be set without the buf lock. */ if ((tbp->b_vflags & BV_BKGRDINPROG) || (tbp->b_flags & B_CACHE) || (tbp->b_flags & B_VMIO) == 0) { bqrelse(tbp); break; } /* * The buffer must be completely invalid in order to * take part in the cluster. If it is partially valid * then we stop. */ off = tbp->b_offset; tsize = size; VM_OBJECT_WLOCK(tbp->b_bufobj->bo_object); for (j = 0; tsize > 0; j++) { toff = off & PAGE_MASK; tinc = tsize; if (toff + tinc > PAGE_SIZE) tinc = PAGE_SIZE - toff; VM_OBJECT_ASSERT_WLOCKED(tbp->b_pages[j]->object); if ((tbp->b_pages[j]->valid & vm_page_bits(toff, tinc)) != 0) break; if (vm_page_xbusied(tbp->b_pages[j])) break; vm_object_pip_add(tbp->b_bufobj->bo_object, 1); vm_page_sbusy(tbp->b_pages[j]); off += tinc; tsize -= tinc; } if (tsize > 0) { clean_sbusy: vm_object_pip_add(tbp->b_bufobj->bo_object, -j); for (k = 0; k < j; k++) vm_page_sunbusy(tbp->b_pages[k]); VM_OBJECT_WUNLOCK(tbp->b_bufobj->bo_object); bqrelse(tbp); break; } VM_OBJECT_WUNLOCK(tbp->b_bufobj->bo_object); /* * Set a read-ahead mark as appropriate */ if ((fbp && (i == 1)) || (i == (run - 1))) tbp->b_flags |= B_RAM; /* * Set the buffer up for an async read (XXX should * we do this only if we do not wind up brelse()ing?). * Set the block number if it isn't set, otherwise * if it is make sure it matches the block number we * expect. */ tbp->b_flags |= B_ASYNC; tbp->b_iocmd = BIO_READ; if (tbp->b_blkno == tbp->b_lblkno) { tbp->b_blkno = bn; } else if (tbp->b_blkno != bn) { VM_OBJECT_WLOCK(tbp->b_bufobj->bo_object); goto clean_sbusy; } } /* * XXX fbp from caller may not be B_ASYNC, but we are going * to biodone() it in cluster_callback() anyway */ BUF_KERNPROC(tbp); TAILQ_INSERT_TAIL(&bp->b_cluster.cluster_head, tbp, b_cluster.cluster_entry); VM_OBJECT_WLOCK(tbp->b_bufobj->bo_object); for (j = 0; j < tbp->b_npages; j += 1) { vm_page_t m; m = tbp->b_pages[j]; if ((bp->b_npages == 0) || (bp->b_pages[bp->b_npages-1] != m)) { bp->b_pages[bp->b_npages] = m; bp->b_npages++; } if (m->valid == VM_PAGE_BITS_ALL) tbp->b_pages[j] = bogus_page; } VM_OBJECT_WUNLOCK(tbp->b_bufobj->bo_object); /* * Don't inherit tbp->b_bufsize as it may be larger due to * a non-page-aligned size. Instead just aggregate using * 'size'. */ if (tbp->b_bcount != size) printf("warning: tbp->b_bcount wrong %ld vs %ld\n", tbp->b_bcount, size); if (tbp->b_bufsize != size) printf("warning: tbp->b_bufsize wrong %ld vs %ld\n", tbp->b_bufsize, size); bp->b_bcount += size; bp->b_bufsize += size; } /* * Fully valid pages in the cluster are already good and do not need * to be re-read from disk. Replace the page with bogus_page */ VM_OBJECT_WLOCK(bp->b_bufobj->bo_object); for (j = 0; j < bp->b_npages; j++) { VM_OBJECT_ASSERT_WLOCKED(bp->b_pages[j]->object); if (bp->b_pages[j]->valid == VM_PAGE_BITS_ALL) bp->b_pages[j] = bogus_page; } VM_OBJECT_WUNLOCK(bp->b_bufobj->bo_object); if (bp->b_bufsize > bp->b_kvasize) panic("cluster_rbuild: b_bufsize(%ld) > b_kvasize(%d)\n", bp->b_bufsize, bp->b_kvasize); if (buf_mapped(bp)) { pmap_qenter(trunc_page((vm_offset_t) bp->b_data), (vm_page_t *)bp->b_pages, bp->b_npages); } return (bp); } /* * Cleanup after a clustered read or write. * This is complicated by the fact that any of the buffers might have * extra memory (if there were no empty buffer headers at allocbuf time) * that we will need to shift around. */ static void cluster_callback(bp) struct buf *bp; { struct buf *nbp, *tbp; int error = 0; /* * Must propagate errors to all the components. */ if (bp->b_ioflags & BIO_ERROR) error = bp->b_error; if (buf_mapped(bp)) { pmap_qremove(trunc_page((vm_offset_t) bp->b_data), bp->b_npages); } /* * Move memory from the large cluster buffer into the component * buffers and mark IO as done on these. */ for (tbp = TAILQ_FIRST(&bp->b_cluster.cluster_head); tbp; tbp = nbp) { nbp = TAILQ_NEXT(&tbp->b_cluster, cluster_entry); if (error) { tbp->b_ioflags |= BIO_ERROR; tbp->b_error = error; } else { tbp->b_dirtyoff = tbp->b_dirtyend = 0; tbp->b_flags &= ~B_INVAL; tbp->b_ioflags &= ~BIO_ERROR; /* * XXX the bdwrite()/bqrelse() issued during * cluster building clears B_RELBUF (see bqrelse() * comment). If direct I/O was specified, we have * to restore it here to allow the buffer and VM * to be freed. */ if (tbp->b_flags & B_DIRECT) tbp->b_flags |= B_RELBUF; } bufdone(tbp); } pbrelvp(bp); relpbuf(bp, &cluster_pbuf_freecnt); } /* * cluster_wbuild_wb: * * Implement modified write build for cluster. * * write_behind = 0 write behind disabled * write_behind = 1 write behind normal (default) * write_behind = 2 write behind backed-off */ static __inline int cluster_wbuild_wb(struct vnode *vp, long size, daddr_t start_lbn, int len, int gbflags) { int r = 0; switch (write_behind) { case 2: if (start_lbn < len) break; start_lbn -= len; /* FALLTHROUGH */ case 1: r = cluster_wbuild(vp, size, start_lbn, len, gbflags); /* FALLTHROUGH */ default: /* FALLTHROUGH */ break; } return(r); } /* * Do clustered write for FFS. * * Three cases: * 1. Write is not sequential (write asynchronously) * Write is sequential: * 2. beginning of cluster - begin cluster * 3. middle of a cluster - add to cluster * 4. end of a cluster - asynchronously write cluster */ void cluster_write(struct vnode *vp, struct buf *bp, u_quad_t filesize, int seqcount, int gbflags) { daddr_t lbn; int maxclen, cursize; int lblocksize; int async; if (!unmapped_buf_allowed) gbflags &= ~GB_UNMAPPED; if (vp->v_type == VREG) { async = DOINGASYNC(vp); lblocksize = vp->v_mount->mnt_stat.f_iosize; } else { async = 0; lblocksize = bp->b_bufsize; } lbn = bp->b_lblkno; KASSERT(bp->b_offset != NOOFFSET, ("cluster_write: no buffer offset")); /* Initialize vnode to beginning of file. */ if (lbn == 0) vp->v_lasta = vp->v_clen = vp->v_cstart = vp->v_lastw = 0; if (vp->v_clen == 0 || lbn != vp->v_lastw + 1 || (bp->b_blkno != vp->v_lasta + btodb(lblocksize))) { maxclen = vp->v_mount->mnt_iosize_max / lblocksize - 1; if (vp->v_clen != 0) { /* * Next block is not sequential. * * If we are not writing at end of file, the process * seeked to another point in the file since its last * write, or we have reached our maximum cluster size, * then push the previous cluster. Otherwise try * reallocating to make it sequential. * * Change to algorithm: only push previous cluster if * it was sequential from the point of view of the * seqcount heuristic, otherwise leave the buffer * intact so we can potentially optimize the I/O * later on in the buf_daemon or update daemon * flush. */ cursize = vp->v_lastw - vp->v_cstart + 1; if (((u_quad_t) bp->b_offset + lblocksize) != filesize || lbn != vp->v_lastw + 1 || vp->v_clen <= cursize) { if (!async && seqcount > 0) { cluster_wbuild_wb(vp, lblocksize, vp->v_cstart, cursize, gbflags); } } else { struct buf **bpp, **endbp; struct cluster_save *buflist; buflist = cluster_collectbufs(vp, bp, gbflags); endbp = &buflist->bs_children [buflist->bs_nchildren - 1]; if (VOP_REALLOCBLKS(vp, buflist)) { /* * Failed, push the previous cluster * if *really* writing sequentially * in the logical file (seqcount > 1), * otherwise delay it in the hopes that * the low level disk driver can * optimize the write ordering. */ for (bpp = buflist->bs_children; bpp < endbp; bpp++) brelse(*bpp); free(buflist, M_SEGMENT); if (seqcount > 1) { cluster_wbuild_wb(vp, lblocksize, vp->v_cstart, cursize, gbflags); } } else { /* * Succeeded, keep building cluster. */ for (bpp = buflist->bs_children; bpp <= endbp; bpp++) bdwrite(*bpp); free(buflist, M_SEGMENT); vp->v_lastw = lbn; vp->v_lasta = bp->b_blkno; return; } } } /* * Consider beginning a cluster. If at end of file, make * cluster as large as possible, otherwise find size of * existing cluster. */ if ((vp->v_type == VREG) && ((u_quad_t) bp->b_offset + lblocksize) != filesize && (bp->b_blkno == bp->b_lblkno) && (VOP_BMAP(vp, lbn, NULL, &bp->b_blkno, &maxclen, NULL) || bp->b_blkno == -1)) { bawrite(bp); vp->v_clen = 0; vp->v_lasta = bp->b_blkno; vp->v_cstart = lbn + 1; vp->v_lastw = lbn; return; } vp->v_clen = maxclen; if (!async && maxclen == 0) { /* I/O not contiguous */ vp->v_cstart = lbn + 1; bawrite(bp); } else { /* Wait for rest of cluster */ vp->v_cstart = lbn; bdwrite(bp); } } else if (lbn == vp->v_cstart + vp->v_clen) { /* * At end of cluster, write it out if seqcount tells us we * are operating sequentially, otherwise let the buf or * update daemon handle it. */ bdwrite(bp); if (seqcount > 1) { cluster_wbuild_wb(vp, lblocksize, vp->v_cstart, vp->v_clen + 1, gbflags); } vp->v_clen = 0; vp->v_cstart = lbn + 1; } else if (vm_page_count_severe()) { /* * We are low on memory, get it going NOW */ bawrite(bp); } else { /* * In the middle of a cluster, so just delay the I/O for now. */ bdwrite(bp); } vp->v_lastw = lbn; vp->v_lasta = bp->b_blkno; } /* * This is an awful lot like cluster_rbuild...wish they could be combined. * The last lbn argument is the current block on which I/O is being * performed. Check to see that it doesn't fall in the middle of * the current block (if last_bp == NULL). */ int cluster_wbuild(struct vnode *vp, long size, daddr_t start_lbn, int len, int gbflags) { struct buf *bp, *tbp; struct bufobj *bo; int i, j; int totalwritten = 0; int dbsize = btodb(size); if (!unmapped_buf_allowed) gbflags &= ~GB_UNMAPPED; bo = &vp->v_bufobj; while (len > 0) { /* * If the buffer is not delayed-write (i.e. dirty), or it * is delayed-write but either locked or inval, it cannot * partake in the clustered write. */ BO_LOCK(bo); if ((tbp = gbincore(&vp->v_bufobj, start_lbn)) == NULL || (tbp->b_vflags & BV_BKGRDINPROG)) { BO_UNLOCK(bo); ++start_lbn; --len; continue; } if (BUF_LOCK(tbp, LK_EXCLUSIVE | LK_NOWAIT | LK_INTERLOCK, BO_LOCKPTR(bo))) { ++start_lbn; --len; continue; } if ((tbp->b_flags & (B_INVAL | B_DELWRI)) != B_DELWRI) { BUF_UNLOCK(tbp); ++start_lbn; --len; continue; } bremfree(tbp); tbp->b_flags &= ~B_DONE; /* * Extra memory in the buffer, punt on this buffer. * XXX we could handle this in most cases, but we would * have to push the extra memory down to after our max * possible cluster size and then potentially pull it back * up if the cluster was terminated prematurely--too much * hassle. */ if (((tbp->b_flags & (B_CLUSTEROK | B_MALLOC | B_VMIO)) != (B_CLUSTEROK | B_VMIO)) || (tbp->b_bcount != tbp->b_bufsize) || (tbp->b_bcount != size) || (len == 1) || ((bp = (vp->v_vflag & VV_MD) != 0 ? trypbuf(&cluster_pbuf_freecnt) : getpbuf(&cluster_pbuf_freecnt)) == NULL)) { totalwritten += tbp->b_bufsize; bawrite(tbp); ++start_lbn; --len; continue; } /* * We got a pbuf to make the cluster in. * so initialise it. */ TAILQ_INIT(&bp->b_cluster.cluster_head); bp->b_bcount = 0; bp->b_bufsize = 0; bp->b_npages = 0; if (tbp->b_wcred != NOCRED) bp->b_wcred = crhold(tbp->b_wcred); bp->b_blkno = tbp->b_blkno; bp->b_lblkno = tbp->b_lblkno; bp->b_offset = tbp->b_offset; /* * We are synthesizing a buffer out of vm_page_t's, but * if the block size is not page aligned then the starting * address may not be either. Inherit the b_data offset * from the original buffer. */ if ((gbflags & GB_UNMAPPED) == 0 || (tbp->b_flags & B_VMIO) == 0) { bp->b_data = (char *)((vm_offset_t)bp->b_data | ((vm_offset_t)tbp->b_data & PAGE_MASK)); } else { bp->b_data = unmapped_buf; } bp->b_flags |= B_CLUSTER | (tbp->b_flags & (B_VMIO | B_NEEDCOMMIT)); bp->b_iodone = cluster_callback; pbgetvp(vp, bp); /* * From this location in the file, scan forward to see * if there are buffers with adjacent data that need to * be written as well. */ for (i = 0; i < len; ++i, ++start_lbn) { if (i != 0) { /* If not the first buffer */ /* * If the adjacent data is not even in core it * can't need to be written. */ BO_LOCK(bo); if ((tbp = gbincore(bo, start_lbn)) == NULL || (tbp->b_vflags & BV_BKGRDINPROG)) { BO_UNLOCK(bo); break; } /* * If it IS in core, but has different * characteristics, or is locked (which * means it could be undergoing a background * I/O or be in a weird state), then don't * cluster with it. */ if (BUF_LOCK(tbp, LK_EXCLUSIVE | LK_NOWAIT | LK_INTERLOCK, BO_LOCKPTR(bo))) break; if ((tbp->b_flags & (B_VMIO | B_CLUSTEROK | B_INVAL | B_DELWRI | B_NEEDCOMMIT)) != (B_DELWRI | B_CLUSTEROK | (bp->b_flags & (B_VMIO | B_NEEDCOMMIT))) || tbp->b_wcred != bp->b_wcred) { BUF_UNLOCK(tbp); break; } /* * Check that the combined cluster * would make sense with regard to pages * and would not be too large */ if ((tbp->b_bcount != size) || ((bp->b_blkno + (dbsize * i)) != tbp->b_blkno) || ((tbp->b_npages + bp->b_npages) > (vp->v_mount->mnt_iosize_max / PAGE_SIZE))) { BUF_UNLOCK(tbp); break; } /* * Ok, it's passed all the tests, * so remove it from the free list * and mark it busy. We will use it. */ bremfree(tbp); tbp->b_flags &= ~B_DONE; } /* end of code for non-first buffers only */ /* * If the IO is via the VM then we do some * special VM hackery (yuck). Since the buffer's * block size may not be page-aligned it is possible * for a page to be shared between two buffers. We * have to get rid of the duplication when building * the cluster. */ if (tbp->b_flags & B_VMIO) { vm_page_t m; VM_OBJECT_WLOCK(tbp->b_bufobj->bo_object); if (i == 0) { vfs_drain_busy_pages(tbp); } else { /* if not first buffer */ for (j = 0; j < tbp->b_npages; j += 1) { m = tbp->b_pages[j]; if (vm_page_xbusied(m)) { VM_OBJECT_WUNLOCK( tbp->b_object); bqrelse(tbp); goto finishcluster; } } } for (j = 0; j < tbp->b_npages; j += 1) { m = tbp->b_pages[j]; vm_page_sbusy(m); vm_object_pip_add(m->object, 1); if ((bp->b_npages == 0) || (bp->b_pages[bp->b_npages - 1] != m)) { bp->b_pages[bp->b_npages] = m; bp->b_npages++; } } VM_OBJECT_WUNLOCK(tbp->b_bufobj->bo_object); } bp->b_bcount += size; bp->b_bufsize += size; /* * If any of the clustered buffers have their * B_BARRIER flag set, transfer that request to * the cluster. */ bp->b_flags |= (tbp->b_flags & B_BARRIER); tbp->b_flags &= ~(B_DONE | B_BARRIER); tbp->b_flags |= B_ASYNC; tbp->b_ioflags &= ~BIO_ERROR; tbp->b_iocmd = BIO_WRITE; bundirty(tbp); reassignbuf(tbp); /* put on clean list */ bufobj_wref(tbp->b_bufobj); BUF_KERNPROC(tbp); buf_track(tbp, __func__); TAILQ_INSERT_TAIL(&bp->b_cluster.cluster_head, tbp, b_cluster.cluster_entry); } finishcluster: if (buf_mapped(bp)) { pmap_qenter(trunc_page((vm_offset_t) bp->b_data), (vm_page_t *)bp->b_pages, bp->b_npages); } if (bp->b_bufsize > bp->b_kvasize) panic( "cluster_wbuild: b_bufsize(%ld) > b_kvasize(%d)\n", bp->b_bufsize, bp->b_kvasize); totalwritten += bp->b_bufsize; bp->b_dirtyoff = 0; bp->b_dirtyend = bp->b_bufsize; bawrite(bp); len -= i; } return totalwritten; } /* * Collect together all the buffers in a cluster. * Plus add one additional buffer. */ static struct cluster_save * cluster_collectbufs(struct vnode *vp, struct buf *last_bp, int gbflags) { struct cluster_save *buflist; struct buf *bp; daddr_t lbn; int i, len; len = vp->v_lastw - vp->v_cstart + 1; buflist = malloc(sizeof(struct buf *) * (len + 1) + sizeof(*buflist), M_SEGMENT, M_WAITOK); buflist->bs_nchildren = 0; buflist->bs_children = (struct buf **) (buflist + 1); for (lbn = vp->v_cstart, i = 0; i < len; lbn++, i++) { (void)bread_gb(vp, lbn, last_bp->b_bcount, NOCRED, gbflags, &bp); buflist->bs_children[i] = bp; if (bp->b_blkno == bp->b_lblkno) VOP_BMAP(vp, bp->b_lblkno, NULL, &bp->b_blkno, NULL, NULL); } buflist->bs_children[i] = bp = last_bp; if (bp->b_blkno == bp->b_lblkno) VOP_BMAP(vp, bp->b_lblkno, NULL, &bp->b_blkno, NULL, NULL); buflist->bs_nchildren = i + 1; return (buflist); } Index: projects/clang400-import/sys/kern/vfs_default.c =================================================================== --- projects/clang400-import/sys/kern/vfs_default.c (revision 311459) +++ projects/clang400-import/sys/kern/vfs_default.c (revision 311460) @@ -1,1323 +1,1328 @@ /*- * Copyright (c) 1989, 1993 * The Regents of the University of California. All rights reserved. * * This code is derived from software contributed * to Berkeley by John Heidemann of the UCLA Ficus project. * * Source: * @(#)i405_init.c 2.10 92/04/27 UCLA Ficus project * * 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. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include static int vop_nolookup(struct vop_lookup_args *); static int vop_norename(struct vop_rename_args *); static int vop_nostrategy(struct vop_strategy_args *); static int get_next_dirent(struct vnode *vp, struct dirent **dpp, char *dirbuf, int dirbuflen, off_t *off, char **cpos, int *len, int *eofflag, struct thread *td); static int dirent_exists(struct vnode *vp, const char *dirname, struct thread *td); #define DIRENT_MINSIZE (sizeof(struct dirent) - (MAXNAMLEN+1) + 4) static int vop_stdis_text(struct vop_is_text_args *ap); static int vop_stdset_text(struct vop_set_text_args *ap); static int vop_stdunset_text(struct vop_unset_text_args *ap); static int vop_stdget_writecount(struct vop_get_writecount_args *ap); static int vop_stdadd_writecount(struct vop_add_writecount_args *ap); static int vop_stdfdatasync(struct vop_fdatasync_args *ap); static int vop_stdgetpages_async(struct vop_getpages_async_args *ap); /* * This vnode table stores what we want to do if the filesystem doesn't * implement a particular VOP. * * If there is no specific entry here, we will return EOPNOTSUPP. * * Note that every filesystem has to implement either vop_access * or vop_accessx; failing to do so will result in immediate crash * due to stack overflow, as vop_stdaccess() calls vop_stdaccessx(), * which calls vop_stdaccess() etc. */ struct vop_vector default_vnodeops = { .vop_default = NULL, .vop_bypass = VOP_EOPNOTSUPP, .vop_access = vop_stdaccess, .vop_accessx = vop_stdaccessx, .vop_advise = vop_stdadvise, .vop_advlock = vop_stdadvlock, .vop_advlockasync = vop_stdadvlockasync, .vop_advlockpurge = vop_stdadvlockpurge, .vop_allocate = vop_stdallocate, .vop_bmap = vop_stdbmap, .vop_close = VOP_NULL, .vop_fsync = VOP_NULL, .vop_fdatasync = vop_stdfdatasync, .vop_getpages = vop_stdgetpages, .vop_getpages_async = vop_stdgetpages_async, .vop_getwritemount = vop_stdgetwritemount, .vop_inactive = VOP_NULL, .vop_ioctl = VOP_ENOTTY, .vop_kqfilter = vop_stdkqfilter, .vop_islocked = vop_stdislocked, .vop_lock1 = vop_stdlock, .vop_lookup = vop_nolookup, .vop_open = VOP_NULL, .vop_pathconf = VOP_EINVAL, .vop_poll = vop_nopoll, .vop_putpages = vop_stdputpages, .vop_readlink = VOP_EINVAL, .vop_rename = vop_norename, .vop_revoke = VOP_PANIC, .vop_strategy = vop_nostrategy, .vop_unlock = vop_stdunlock, .vop_vptocnp = vop_stdvptocnp, .vop_vptofh = vop_stdvptofh, .vop_unp_bind = vop_stdunp_bind, .vop_unp_connect = vop_stdunp_connect, .vop_unp_detach = vop_stdunp_detach, .vop_is_text = vop_stdis_text, .vop_set_text = vop_stdset_text, .vop_unset_text = vop_stdunset_text, .vop_get_writecount = vop_stdget_writecount, .vop_add_writecount = vop_stdadd_writecount, }; /* * Series of placeholder functions for various error returns for * VOPs. */ int vop_eopnotsupp(struct vop_generic_args *ap) { /* printf("vop_notsupp[%s]\n", ap->a_desc->vdesc_name); */ return (EOPNOTSUPP); } int vop_ebadf(struct vop_generic_args *ap) { return (EBADF); } int vop_enotty(struct vop_generic_args *ap) { return (ENOTTY); } int vop_einval(struct vop_generic_args *ap) { return (EINVAL); } int vop_enoent(struct vop_generic_args *ap) { return (ENOENT); } int vop_null(struct vop_generic_args *ap) { return (0); } /* * Helper function to panic on some bad VOPs in some filesystems. */ int vop_panic(struct vop_generic_args *ap) { panic("filesystem goof: vop_panic[%s]", ap->a_desc->vdesc_name); } /* * vop_std and vop_no are default functions for use by * filesystems that need the "default reasonable" implementation for a * particular operation. * * The documentation for the operations they implement exists (if it exists) * in the VOP_(9) manpage (all uppercase). */ /* * Default vop for filesystems that do not support name lookup */ static int vop_nolookup(ap) struct vop_lookup_args /* { struct vnode *a_dvp; struct vnode **a_vpp; struct componentname *a_cnp; } */ *ap; { *ap->a_vpp = NULL; return (ENOTDIR); } /* * vop_norename: * * Handle unlock and reference counting for arguments of vop_rename * for filesystems that do not implement rename operation. */ static int vop_norename(struct vop_rename_args *ap) { vop_rename_fail(ap); return (EOPNOTSUPP); } /* * vop_nostrategy: * * Strategy routine for VFS devices that have none. * * BIO_ERROR and B_INVAL must be cleared prior to calling any strategy * routine. Typically this is done for a BIO_READ strategy call. * Typically B_INVAL is assumed to already be clear prior to a write * and should not be cleared manually unless you just made the buffer * invalid. BIO_ERROR should be cleared either way. */ static int vop_nostrategy (struct vop_strategy_args *ap) { printf("No strategy for buffer at %p\n", ap->a_bp); vn_printf(ap->a_vp, "vnode "); ap->a_bp->b_ioflags |= BIO_ERROR; ap->a_bp->b_error = EOPNOTSUPP; bufdone(ap->a_bp); return (EOPNOTSUPP); } static int get_next_dirent(struct vnode *vp, struct dirent **dpp, char *dirbuf, int dirbuflen, off_t *off, char **cpos, int *len, int *eofflag, struct thread *td) { int error, reclen; struct uio uio; struct iovec iov; struct dirent *dp; KASSERT(VOP_ISLOCKED(vp), ("vp %p is not locked", vp)); KASSERT(vp->v_type == VDIR, ("vp %p is not a directory", vp)); if (*len == 0) { iov.iov_base = dirbuf; iov.iov_len = dirbuflen; uio.uio_iov = &iov; uio.uio_iovcnt = 1; uio.uio_offset = *off; uio.uio_resid = dirbuflen; uio.uio_segflg = UIO_SYSSPACE; uio.uio_rw = UIO_READ; uio.uio_td = td; *eofflag = 0; #ifdef MAC error = mac_vnode_check_readdir(td->td_ucred, vp); if (error == 0) #endif error = VOP_READDIR(vp, &uio, td->td_ucred, eofflag, NULL, NULL); if (error) return (error); *off = uio.uio_offset; *cpos = dirbuf; *len = (dirbuflen - uio.uio_resid); if (*len == 0) return (ENOENT); } dp = (struct dirent *)(*cpos); reclen = dp->d_reclen; *dpp = dp; /* check for malformed directory.. */ if (reclen < DIRENT_MINSIZE) return (EINVAL); *cpos += reclen; *len -= reclen; return (0); } /* * Check if a named file exists in a given directory vnode. */ static int dirent_exists(struct vnode *vp, const char *dirname, struct thread *td) { char *dirbuf, *cpos; int error, eofflag, dirbuflen, len, found; off_t off; struct dirent *dp; struct vattr va; KASSERT(VOP_ISLOCKED(vp), ("vp %p is not locked", vp)); KASSERT(vp->v_type == VDIR, ("vp %p is not a directory", vp)); found = 0; error = VOP_GETATTR(vp, &va, td->td_ucred); if (error) return (found); dirbuflen = DEV_BSIZE; if (dirbuflen < va.va_blocksize) dirbuflen = va.va_blocksize; dirbuf = (char *)malloc(dirbuflen, M_TEMP, M_WAITOK); off = 0; len = 0; do { error = get_next_dirent(vp, &dp, dirbuf, dirbuflen, &off, &cpos, &len, &eofflag, td); if (error) goto out; if (dp->d_type != DT_WHT && dp->d_fileno != 0 && strcmp(dp->d_name, dirname) == 0) { found = 1; goto out; } } while (len > 0 || !eofflag); out: free(dirbuf, M_TEMP); return (found); } int vop_stdaccess(struct vop_access_args *ap) { KASSERT((ap->a_accmode & ~(VEXEC | VWRITE | VREAD | VADMIN | VAPPEND)) == 0, ("invalid bit in accmode")); return (VOP_ACCESSX(ap->a_vp, ap->a_accmode, ap->a_cred, ap->a_td)); } int vop_stdaccessx(struct vop_accessx_args *ap) { int error; accmode_t accmode = ap->a_accmode; error = vfs_unixify_accmode(&accmode); if (error != 0) return (error); if (accmode == 0) return (0); return (VOP_ACCESS(ap->a_vp, accmode, ap->a_cred, ap->a_td)); } /* * Advisory record locking support */ int vop_stdadvlock(struct vop_advlock_args *ap) { struct vnode *vp; struct vattr vattr; int error; vp = ap->a_vp; if (ap->a_fl->l_whence == SEEK_END) { /* * The NFSv4 server must avoid doing a vn_lock() here, since it * can deadlock the nfsd threads, due to a LOR. Fortunately * the NFSv4 server always uses SEEK_SET and this code is * only required for the SEEK_END case. */ vn_lock(vp, LK_SHARED | LK_RETRY); error = VOP_GETATTR(vp, &vattr, curthread->td_ucred); VOP_UNLOCK(vp, 0); if (error) return (error); } else vattr.va_size = 0; return (lf_advlock(ap, &(vp->v_lockf), vattr.va_size)); } int vop_stdadvlockasync(struct vop_advlockasync_args *ap) { struct vnode *vp; struct vattr vattr; int error; vp = ap->a_vp; if (ap->a_fl->l_whence == SEEK_END) { /* The size argument is only needed for SEEK_END. */ vn_lock(vp, LK_SHARED | LK_RETRY); error = VOP_GETATTR(vp, &vattr, curthread->td_ucred); VOP_UNLOCK(vp, 0); if (error) return (error); } else vattr.va_size = 0; return (lf_advlockasync(ap, &(vp->v_lockf), vattr.va_size)); } int vop_stdadvlockpurge(struct vop_advlockpurge_args *ap) { struct vnode *vp; vp = ap->a_vp; lf_purgelocks(vp, &vp->v_lockf); return (0); } /* * vop_stdpathconf: * * Standard implementation of POSIX pathconf, to get information about limits * for a filesystem. * Override per filesystem for the case where the filesystem has smaller * limits. */ int vop_stdpathconf(ap) struct vop_pathconf_args /* { struct vnode *a_vp; int a_name; int *a_retval; } */ *ap; { switch (ap->a_name) { case _PC_ASYNC_IO: *ap->a_retval = _POSIX_ASYNCHRONOUS_IO; return (0); case _PC_NAME_MAX: *ap->a_retval = NAME_MAX; return (0); case _PC_PATH_MAX: *ap->a_retval = PATH_MAX; return (0); case _PC_LINK_MAX: *ap->a_retval = LINK_MAX; return (0); case _PC_MAX_CANON: *ap->a_retval = MAX_CANON; return (0); case _PC_MAX_INPUT: *ap->a_retval = MAX_INPUT; return (0); case _PC_PIPE_BUF: *ap->a_retval = PIPE_BUF; return (0); case _PC_CHOWN_RESTRICTED: *ap->a_retval = 1; return (0); case _PC_VDISABLE: *ap->a_retval = _POSIX_VDISABLE; return (0); default: return (EINVAL); } /* NOTREACHED */ } /* * Standard lock, unlock and islocked functions. */ int vop_stdlock(ap) struct vop_lock1_args /* { struct vnode *a_vp; int a_flags; char *file; int line; } */ *ap; { struct vnode *vp = ap->a_vp; 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)); } /* See above. */ int vop_stdunlock(ap) struct vop_unlock_args /* { struct vnode *a_vp; int a_flags; } */ *ap; { struct vnode *vp = ap->a_vp; return (lockmgr(vp->v_vnlock, ap->a_flags | LK_RELEASE, VI_MTX(vp))); } /* See above. */ int vop_stdislocked(ap) struct vop_islocked_args /* { struct vnode *a_vp; } */ *ap; { return (lockstatus(ap->a_vp->v_vnlock)); } /* * Return true for select/poll. */ int vop_nopoll(ap) struct vop_poll_args /* { struct vnode *a_vp; int a_events; struct ucred *a_cred; struct thread *a_td; } */ *ap; { return (poll_no_poll(ap->a_events)); } /* * Implement poll for local filesystems that support it. */ int vop_stdpoll(ap) struct vop_poll_args /* { struct vnode *a_vp; int a_events; struct ucred *a_cred; struct thread *a_td; } */ *ap; { if (ap->a_events & ~POLLSTANDARD) return (vn_pollrecord(ap->a_vp, ap->a_td, ap->a_events)); return (ap->a_events & (POLLIN | POLLOUT | POLLRDNORM | POLLWRNORM)); } /* * Return our mount point, as we will take charge of the writes. */ int vop_stdgetwritemount(ap) struct vop_getwritemount_args /* { struct vnode *a_vp; struct mount **a_mpp; } */ *ap; { struct mount *mp; /* * XXX Since this is called unlocked we may be recycled while * attempting to ref the mount. If this is the case or mountpoint * will be set to NULL. We only have to prevent this call from * returning with a ref to an incorrect mountpoint. It is not * harmful to return with a ref to our previous mountpoint. */ mp = ap->a_vp->v_mount; if (mp != NULL) { vfs_ref(mp); if (mp != ap->a_vp->v_mount) { vfs_rel(mp); mp = NULL; } } *(ap->a_mpp) = mp; return (0); } /* XXX Needs good comment and VOP_BMAP(9) manpage */ int vop_stdbmap(ap) struct vop_bmap_args /* { struct vnode *a_vp; daddr_t a_bn; struct bufobj **a_bop; daddr_t *a_bnp; int *a_runp; int *a_runb; } */ *ap; { if (ap->a_bop != NULL) *ap->a_bop = &ap->a_vp->v_bufobj; if (ap->a_bnp != NULL) *ap->a_bnp = ap->a_bn * btodb(ap->a_vp->v_mount->mnt_stat.f_iosize); if (ap->a_runp != NULL) *ap->a_runp = 0; if (ap->a_runb != NULL) *ap->a_runb = 0; return (0); } int vop_stdfsync(ap) struct vop_fsync_args /* { struct vnode *a_vp; int a_waitfor; struct thread *a_td; } */ *ap; { struct vnode *vp = ap->a_vp; struct buf *bp; struct bufobj *bo; struct buf *nbp; int error = 0; int maxretry = 1000; /* large, arbitrarily chosen */ bo = &vp->v_bufobj; BO_LOCK(bo); loop1: /* * MARK/SCAN initialization to avoid infinite loops. */ TAILQ_FOREACH(bp, &bo->bo_dirty.bv_hd, b_bobufs) { bp->b_vflags &= ~BV_SCANNED; bp->b_error = 0; } /* * Flush all dirty buffers associated with a vnode. */ loop2: TAILQ_FOREACH_SAFE(bp, &bo->bo_dirty.bv_hd, b_bobufs, nbp) { if ((bp->b_vflags & BV_SCANNED) != 0) continue; bp->b_vflags |= BV_SCANNED; if (BUF_LOCK(bp, LK_EXCLUSIVE | LK_NOWAIT, NULL)) { if (ap->a_waitfor != MNT_WAIT) continue; if (BUF_LOCK(bp, LK_EXCLUSIVE | LK_INTERLOCK | LK_SLEEPFAIL, BO_LOCKPTR(bo)) != 0) { BO_LOCK(bo); goto loop1; } BO_LOCK(bo); } BO_UNLOCK(bo); KASSERT(bp->b_bufobj == bo, ("bp %p wrong b_bufobj %p should be %p", bp, bp->b_bufobj, bo)); if ((bp->b_flags & B_DELWRI) == 0) panic("fsync: not dirty"); if ((vp->v_object != NULL) && (bp->b_flags & B_CLUSTEROK)) { vfs_bio_awrite(bp); } else { bremfree(bp); bawrite(bp); } BO_LOCK(bo); goto loop2; } /* * If synchronous the caller expects us to completely resolve all * dirty buffers in the system. Wait for in-progress I/O to * complete (which could include background bitmap writes), then * retry if dirty blocks still exist. */ if (ap->a_waitfor == MNT_WAIT) { bufobj_wwait(bo, 0, 0); if (bo->bo_dirty.bv_cnt > 0) { /* * If we are unable to write any of these buffers * then we fail now rather than trying endlessly * to write them out. */ TAILQ_FOREACH(bp, &bo->bo_dirty.bv_hd, b_bobufs) if ((error = bp->b_error) == 0) continue; if (error == 0 && --maxretry >= 0) goto loop1; error = EAGAIN; } } BO_UNLOCK(bo); if (error == EAGAIN) vn_printf(vp, "fsync: giving up on dirty "); return (error); } static int vop_stdfdatasync(struct vop_fdatasync_args *ap) { return (VOP_FSYNC(ap->a_vp, MNT_WAIT, ap->a_td)); } int vop_stdfdatasync_buf(struct vop_fdatasync_args *ap) { struct vop_fsync_args apf; apf.a_vp = ap->a_vp; apf.a_waitfor = MNT_WAIT; apf.a_td = ap->a_td; return (vop_stdfsync(&apf)); } /* XXX Needs good comment and more info in the manpage (VOP_GETPAGES(9)). */ int vop_stdgetpages(ap) struct vop_getpages_args /* { struct vnode *a_vp; vm_page_t *a_m; int a_count; int *a_rbehind; int *a_rahead; } */ *ap; { return vnode_pager_generic_getpages(ap->a_vp, ap->a_m, ap->a_count, ap->a_rbehind, ap->a_rahead, NULL, NULL); } static int vop_stdgetpages_async(struct vop_getpages_async_args *ap) { int error; error = VOP_GETPAGES(ap->a_vp, ap->a_m, ap->a_count, ap->a_rbehind, ap->a_rahead); ap->a_iodone(ap->a_arg, ap->a_m, ap->a_count, error); return (error); } int vop_stdkqfilter(struct vop_kqfilter_args *ap) { return vfs_kqfilter(ap); } /* XXX Needs good comment and more info in the manpage (VOP_PUTPAGES(9)). */ int vop_stdputpages(ap) struct vop_putpages_args /* { struct vnode *a_vp; vm_page_t *a_m; int a_count; int a_sync; int *a_rtvals; } */ *ap; { return vnode_pager_generic_putpages(ap->a_vp, ap->a_m, ap->a_count, ap->a_sync, ap->a_rtvals); } int vop_stdvptofh(struct vop_vptofh_args *ap) { return (EOPNOTSUPP); } int vop_stdvptocnp(struct vop_vptocnp_args *ap) { struct vnode *vp = ap->a_vp; struct vnode **dvp = ap->a_vpp; struct ucred *cred = ap->a_cred; char *buf = ap->a_buf; int *buflen = ap->a_buflen; char *dirbuf, *cpos; int i, error, eofflag, dirbuflen, flags, locked, len, covered; off_t off; ino_t fileno; struct vattr va; struct nameidata nd; struct thread *td; struct dirent *dp; struct vnode *mvp; i = *buflen; error = 0; covered = 0; td = curthread; if (vp->v_type != VDIR) return (ENOENT); error = VOP_GETATTR(vp, &va, cred); if (error) return (error); VREF(vp); locked = VOP_ISLOCKED(vp); VOP_UNLOCK(vp, 0); NDINIT_ATVP(&nd, LOOKUP, FOLLOW | LOCKSHARED | LOCKLEAF, UIO_SYSSPACE, "..", vp, td); flags = FREAD; error = vn_open_cred(&nd, &flags, 0, VN_OPEN_NOAUDIT, cred, NULL); if (error) { vn_lock(vp, locked | LK_RETRY); return (error); } NDFREE(&nd, NDF_ONLY_PNBUF); mvp = *dvp = nd.ni_vp; if (vp->v_mount != (*dvp)->v_mount && ((*dvp)->v_vflag & VV_ROOT) && ((*dvp)->v_mount->mnt_flag & MNT_UNION)) { *dvp = (*dvp)->v_mount->mnt_vnodecovered; VREF(mvp); VOP_UNLOCK(mvp, 0); vn_close(mvp, FREAD, cred, td); VREF(*dvp); vn_lock(*dvp, LK_SHARED | LK_RETRY); covered = 1; } fileno = va.va_fileid; dirbuflen = DEV_BSIZE; if (dirbuflen < va.va_blocksize) dirbuflen = va.va_blocksize; dirbuf = (char *)malloc(dirbuflen, M_TEMP, M_WAITOK); if ((*dvp)->v_type != VDIR) { error = ENOENT; goto out; } off = 0; len = 0; do { /* call VOP_READDIR of parent */ error = get_next_dirent(*dvp, &dp, dirbuf, dirbuflen, &off, &cpos, &len, &eofflag, td); if (error) goto out; if ((dp->d_type != DT_WHT) && (dp->d_fileno == fileno)) { if (covered) { VOP_UNLOCK(*dvp, 0); vn_lock(mvp, LK_SHARED | LK_RETRY); if (dirent_exists(mvp, dp->d_name, td)) { error = ENOENT; VOP_UNLOCK(mvp, 0); vn_lock(*dvp, LK_SHARED | LK_RETRY); goto out; } VOP_UNLOCK(mvp, 0); vn_lock(*dvp, LK_SHARED | LK_RETRY); } i -= dp->d_namlen; if (i < 0) { error = ENOMEM; goto out; } if (dp->d_namlen == 1 && dp->d_name[0] == '.') { error = ENOENT; } else { bcopy(dp->d_name, buf + i, dp->d_namlen); error = 0; } goto out; } } while (len > 0 || !eofflag); error = ENOENT; out: free(dirbuf, M_TEMP); if (!error) { *buflen = i; vref(*dvp); } if (covered) { vput(*dvp); vrele(mvp); } else { VOP_UNLOCK(mvp, 0); vn_close(mvp, FREAD, cred, td); } vn_lock(vp, locked | LK_RETRY); return (error); } int vop_stdallocate(struct vop_allocate_args *ap) { #ifdef __notyet__ - struct statfs sfs; + struct statfs *sfs; + off_t maxfilesize = 0; #endif struct iovec aiov; struct vattr vattr, *vap; struct uio auio; off_t fsize, len, cur, offset; uint8_t *buf; struct thread *td; struct vnode *vp; size_t iosize; int error; buf = NULL; error = 0; td = curthread; vap = &vattr; vp = ap->a_vp; len = *ap->a_len; offset = *ap->a_offset; error = VOP_GETATTR(vp, vap, td->td_ucred); if (error != 0) goto out; fsize = vap->va_size; iosize = vap->va_blocksize; if (iosize == 0) iosize = BLKDEV_IOSIZE; if (iosize > MAXPHYS) iosize = MAXPHYS; buf = malloc(iosize, M_TEMP, M_WAITOK); #ifdef __notyet__ /* * Check if the filesystem sets f_maxfilesize; if not use * VOP_SETATTR to perform the check. */ - error = VFS_STATFS(vp->v_mount, &sfs, td); + sfs = malloc(sizeof(struct statfs), M_STATFS, M_WAITOK); + error = VFS_STATFS(vp->v_mount, sfs, td); + if (error == 0) + maxfilesize = sfs->f_maxfilesize; + free(sfs, M_STATFS); if (error != 0) goto out; - if (sfs.f_maxfilesize) { - if (offset > sfs.f_maxfilesize || len > sfs.f_maxfilesize || - offset + len > sfs.f_maxfilesize) { + if (maxfilesize) { + if (offset > maxfilesize || len > maxfilesize || + offset + len > maxfilesize) { error = EFBIG; goto out; } } else #endif if (offset + len > vap->va_size) { /* * Test offset + len against the filesystem's maxfilesize. */ VATTR_NULL(vap); vap->va_size = offset + len; error = VOP_SETATTR(vp, vap, td->td_ucred); if (error != 0) goto out; VATTR_NULL(vap); vap->va_size = fsize; error = VOP_SETATTR(vp, vap, td->td_ucred); if (error != 0) goto out; } for (;;) { /* * Read and write back anything below the nominal file * size. There's currently no way outside the filesystem * to know whether this area is sparse or not. */ cur = iosize; if ((offset % iosize) != 0) cur -= (offset % iosize); if (cur > len) cur = len; if (offset < fsize) { aiov.iov_base = buf; aiov.iov_len = cur; auio.uio_iov = &aiov; auio.uio_iovcnt = 1; auio.uio_offset = offset; auio.uio_resid = cur; auio.uio_segflg = UIO_SYSSPACE; auio.uio_rw = UIO_READ; auio.uio_td = td; error = VOP_READ(vp, &auio, 0, td->td_ucred); if (error != 0) break; if (auio.uio_resid > 0) { bzero(buf + cur - auio.uio_resid, auio.uio_resid); } } else { bzero(buf, cur); } aiov.iov_base = buf; aiov.iov_len = cur; auio.uio_iov = &aiov; auio.uio_iovcnt = 1; auio.uio_offset = offset; auio.uio_resid = cur; auio.uio_segflg = UIO_SYSSPACE; auio.uio_rw = UIO_WRITE; auio.uio_td = td; error = VOP_WRITE(vp, &auio, 0, td->td_ucred); if (error != 0) break; len -= cur; offset += cur; if (len == 0) break; if (should_yield()) break; } out: *ap->a_len = len; *ap->a_offset = offset; free(buf, M_TEMP); return (error); } int vop_stdadvise(struct vop_advise_args *ap) { struct vnode *vp; struct bufobj *bo; daddr_t startn, endn; off_t start, end; int bsize, error; vp = ap->a_vp; switch (ap->a_advice) { case POSIX_FADV_WILLNEED: /* * Do nothing for now. Filesystems should provide a * custom method which starts an asynchronous read of * the requested region. */ error = 0; break; case POSIX_FADV_DONTNEED: error = 0; vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); if (vp->v_iflag & VI_DOOMED) { VOP_UNLOCK(vp, 0); break; } /* * Deactivate pages in the specified range from the backing VM * object. Pages that are resident in the buffer cache will * remain wired until their corresponding buffers are released * below. */ if (vp->v_object != NULL) { start = trunc_page(ap->a_start); end = round_page(ap->a_end); VM_OBJECT_WLOCK(vp->v_object); vm_object_page_noreuse(vp->v_object, OFF_TO_IDX(start), OFF_TO_IDX(end)); VM_OBJECT_WUNLOCK(vp->v_object); } bo = &vp->v_bufobj; BO_RLOCK(bo); bsize = vp->v_bufobj.bo_bsize; startn = ap->a_start / bsize; endn = ap->a_end / bsize; error = bnoreuselist(&bo->bo_clean, bo, startn, endn); if (error == 0) error = bnoreuselist(&bo->bo_dirty, bo, startn, endn); BO_RUNLOCK(bo); VOP_UNLOCK(vp, 0); break; default: error = EINVAL; break; } return (error); } int vop_stdunp_bind(struct vop_unp_bind_args *ap) { ap->a_vp->v_socket = ap->a_socket; return (0); } int vop_stdunp_connect(struct vop_unp_connect_args *ap) { *ap->a_socket = ap->a_vp->v_socket; return (0); } int vop_stdunp_detach(struct vop_unp_detach_args *ap) { ap->a_vp->v_socket = NULL; return (0); } static int vop_stdis_text(struct vop_is_text_args *ap) { return ((ap->a_vp->v_vflag & VV_TEXT) != 0); } static int vop_stdset_text(struct vop_set_text_args *ap) { ap->a_vp->v_vflag |= VV_TEXT; return (0); } static int vop_stdunset_text(struct vop_unset_text_args *ap) { ap->a_vp->v_vflag &= ~VV_TEXT; return (0); } static int vop_stdget_writecount(struct vop_get_writecount_args *ap) { *ap->a_writecount = ap->a_vp->v_writecount; return (0); } static int vop_stdadd_writecount(struct vop_add_writecount_args *ap) { ap->a_vp->v_writecount += ap->a_inc; return (0); } /* * vfs default ops * used to fill the vfs function table to get reasonable default return values. */ int vfs_stdroot (mp, flags, vpp) struct mount *mp; int flags; struct vnode **vpp; { return (EOPNOTSUPP); } int vfs_stdstatfs (mp, sbp) struct mount *mp; struct statfs *sbp; { return (EOPNOTSUPP); } int vfs_stdquotactl (mp, cmds, uid, arg) struct mount *mp; int cmds; uid_t uid; void *arg; { return (EOPNOTSUPP); } int vfs_stdsync(mp, waitfor) struct mount *mp; int waitfor; { struct vnode *vp, *mvp; struct thread *td; int error, lockreq, allerror = 0; td = curthread; lockreq = LK_EXCLUSIVE | LK_INTERLOCK; if (waitfor != MNT_WAIT) lockreq |= LK_NOWAIT; /* * Force stale buffer cache information to be flushed. */ loop: MNT_VNODE_FOREACH_ALL(vp, mp, mvp) { if (vp->v_bufobj.bo_dirty.bv_cnt == 0) { VI_UNLOCK(vp); continue; } if ((error = vget(vp, lockreq, td)) != 0) { if (error == ENOENT) { MNT_VNODE_FOREACH_ALL_ABORT(mp, mvp); goto loop; } continue; } error = VOP_FSYNC(vp, waitfor, td); if (error) allerror = error; vput(vp); } return (allerror); } int vfs_stdnosync (mp, waitfor) struct mount *mp; int waitfor; { return (0); } int vfs_stdvget (mp, ino, flags, vpp) struct mount *mp; ino_t ino; int flags; struct vnode **vpp; { return (EOPNOTSUPP); } int vfs_stdfhtovp (mp, fhp, flags, vpp) struct mount *mp; struct fid *fhp; int flags; struct vnode **vpp; { return (EOPNOTSUPP); } int vfs_stdinit (vfsp) struct vfsconf *vfsp; { return (0); } int vfs_stduninit (vfsp) struct vfsconf *vfsp; { return(0); } int vfs_stdextattrctl(mp, cmd, filename_vp, attrnamespace, attrname) struct mount *mp; int cmd; struct vnode *filename_vp; int attrnamespace; const char *attrname; { if (filename_vp != NULL) VOP_UNLOCK(filename_vp, 0); return (EOPNOTSUPP); } int vfs_stdsysctl(mp, op, req) struct mount *mp; fsctlop_t op; struct sysctl_req *req; { return (EOPNOTSUPP); } /* end of vfs default ops */ Index: projects/clang400-import/sys/kern/vfs_mount.c =================================================================== --- projects/clang400-import/sys/kern/vfs_mount.c (revision 311459) +++ projects/clang400-import/sys/kern/vfs_mount.c (revision 311460) @@ -1,2015 +1,2016 @@ /*- * Copyright (c) 1999-2004 Poul-Henning Kamp * Copyright (c) 1999 Michael Smith * Copyright (c) 1989, 1993 * The Regents of the University of California. All rights reserved. * (c) UNIX System Laboratories, Inc. * All or some portions of this file are derived from material licensed * to the University of California by American Telephone and Telegraph * Co. or Unix System Laboratories, Inc. and are reproduced herein with * the permission of UNIX System Laboratories, Inc. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 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 AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #define VFS_MOUNTARG_SIZE_MAX (1024 * 64) static int vfs_domount(struct thread *td, const char *fstype, char *fspath, uint64_t fsflags, struct vfsoptlist **optlist); static void free_mntarg(struct mntarg *ma); static int usermount = 0; SYSCTL_INT(_vfs, OID_AUTO, usermount, CTLFLAG_RW, &usermount, 0, "Unprivileged users may mount and unmount file systems"); MALLOC_DEFINE(M_MOUNT, "mount", "vfs mount structure"); +MALLOC_DEFINE(M_STATFS, "statfs", "statfs structure"); static uma_zone_t mount_zone; /* List of mounted filesystems. */ struct mntlist mountlist = TAILQ_HEAD_INITIALIZER(mountlist); /* For any iteration/modification of mountlist */ struct mtx mountlist_mtx; MTX_SYSINIT(mountlist, &mountlist_mtx, "mountlist", MTX_DEF); /* * Global opts, taken by all filesystems */ static const char *global_opts[] = { "errmsg", "fstype", "fspath", "ro", "rw", "nosuid", "noexec", NULL }; static int mount_init(void *mem, int size, int flags) { struct mount *mp; mp = (struct mount *)mem; mtx_init(&mp->mnt_mtx, "struct mount mtx", NULL, MTX_DEF); mtx_init(&mp->mnt_listmtx, "struct mount vlist mtx", NULL, MTX_DEF); lockinit(&mp->mnt_explock, PVFS, "explock", 0, 0); return (0); } static void mount_fini(void *mem, int size) { struct mount *mp; mp = (struct mount *)mem; lockdestroy(&mp->mnt_explock); mtx_destroy(&mp->mnt_listmtx); mtx_destroy(&mp->mnt_mtx); } static void vfs_mount_init(void *dummy __unused) { mount_zone = uma_zcreate("Mountpoints", sizeof(struct mount), NULL, NULL, mount_init, mount_fini, UMA_ALIGN_PTR, UMA_ZONE_NOFREE); } SYSINIT(vfs_mount, SI_SUB_VFS, SI_ORDER_ANY, vfs_mount_init, NULL); /* * --------------------------------------------------------------------- * Functions for building and sanitizing the mount options */ /* Remove one mount option. */ static void vfs_freeopt(struct vfsoptlist *opts, struct vfsopt *opt) { TAILQ_REMOVE(opts, opt, link); free(opt->name, M_MOUNT); if (opt->value != NULL) free(opt->value, M_MOUNT); free(opt, M_MOUNT); } /* Release all resources related to the mount options. */ void vfs_freeopts(struct vfsoptlist *opts) { struct vfsopt *opt; while (!TAILQ_EMPTY(opts)) { opt = TAILQ_FIRST(opts); vfs_freeopt(opts, opt); } free(opts, M_MOUNT); } void vfs_deleteopt(struct vfsoptlist *opts, const char *name) { struct vfsopt *opt, *temp; if (opts == NULL) return; TAILQ_FOREACH_SAFE(opt, opts, link, temp) { if (strcmp(opt->name, name) == 0) vfs_freeopt(opts, opt); } } static int vfs_isopt_ro(const char *opt) { if (strcmp(opt, "ro") == 0 || strcmp(opt, "rdonly") == 0 || strcmp(opt, "norw") == 0) return (1); return (0); } static int vfs_isopt_rw(const char *opt) { if (strcmp(opt, "rw") == 0 || strcmp(opt, "noro") == 0) return (1); return (0); } /* * Check if options are equal (with or without the "no" prefix). */ static int vfs_equalopts(const char *opt1, const char *opt2) { char *p; /* "opt" vs. "opt" or "noopt" vs. "noopt" */ if (strcmp(opt1, opt2) == 0) return (1); /* "noopt" vs. "opt" */ if (strncmp(opt1, "no", 2) == 0 && strcmp(opt1 + 2, opt2) == 0) return (1); /* "opt" vs. "noopt" */ if (strncmp(opt2, "no", 2) == 0 && strcmp(opt1, opt2 + 2) == 0) return (1); while ((p = strchr(opt1, '.')) != NULL && !strncmp(opt1, opt2, ++p - opt1)) { opt2 += p - opt1; opt1 = p; /* "foo.noopt" vs. "foo.opt" */ if (strncmp(opt1, "no", 2) == 0 && strcmp(opt1 + 2, opt2) == 0) return (1); /* "foo.opt" vs. "foo.noopt" */ if (strncmp(opt2, "no", 2) == 0 && strcmp(opt1, opt2 + 2) == 0) return (1); } /* "ro" / "rdonly" / "norw" / "rw" / "noro" */ if ((vfs_isopt_ro(opt1) || vfs_isopt_rw(opt1)) && (vfs_isopt_ro(opt2) || vfs_isopt_rw(opt2))) return (1); return (0); } /* * If a mount option is specified several times, * (with or without the "no" prefix) only keep * the last occurrence of it. */ static void vfs_sanitizeopts(struct vfsoptlist *opts) { struct vfsopt *opt, *opt2, *tmp; TAILQ_FOREACH_REVERSE(opt, opts, vfsoptlist, link) { opt2 = TAILQ_PREV(opt, vfsoptlist, link); while (opt2 != NULL) { if (vfs_equalopts(opt->name, opt2->name)) { tmp = TAILQ_PREV(opt2, vfsoptlist, link); vfs_freeopt(opts, opt2); opt2 = tmp; } else { opt2 = TAILQ_PREV(opt2, vfsoptlist, link); } } } } /* * Build a linked list of mount options from a struct uio. */ int vfs_buildopts(struct uio *auio, struct vfsoptlist **options) { struct vfsoptlist *opts; struct vfsopt *opt; size_t memused, namelen, optlen; unsigned int i, iovcnt; int error; opts = malloc(sizeof(struct vfsoptlist), M_MOUNT, M_WAITOK); TAILQ_INIT(opts); memused = 0; iovcnt = auio->uio_iovcnt; for (i = 0; i < iovcnt; i += 2) { namelen = auio->uio_iov[i].iov_len; optlen = auio->uio_iov[i + 1].iov_len; memused += sizeof(struct vfsopt) + optlen + namelen; /* * Avoid consuming too much memory, and attempts to overflow * memused. */ if (memused > VFS_MOUNTARG_SIZE_MAX || optlen > VFS_MOUNTARG_SIZE_MAX || namelen > VFS_MOUNTARG_SIZE_MAX) { error = EINVAL; goto bad; } opt = malloc(sizeof(struct vfsopt), M_MOUNT, M_WAITOK); opt->name = malloc(namelen, M_MOUNT, M_WAITOK); opt->value = NULL; opt->len = 0; opt->pos = i / 2; opt->seen = 0; /* * Do this early, so jumps to "bad" will free the current * option. */ TAILQ_INSERT_TAIL(opts, opt, link); if (auio->uio_segflg == UIO_SYSSPACE) { bcopy(auio->uio_iov[i].iov_base, opt->name, namelen); } else { error = copyin(auio->uio_iov[i].iov_base, opt->name, namelen); if (error) goto bad; } /* Ensure names are null-terminated strings. */ if (namelen == 0 || opt->name[namelen - 1] != '\0') { error = EINVAL; goto bad; } if (optlen != 0) { opt->len = optlen; opt->value = malloc(optlen, M_MOUNT, M_WAITOK); if (auio->uio_segflg == UIO_SYSSPACE) { bcopy(auio->uio_iov[i + 1].iov_base, opt->value, optlen); } else { error = copyin(auio->uio_iov[i + 1].iov_base, opt->value, optlen); if (error) goto bad; } } } vfs_sanitizeopts(opts); *options = opts; return (0); bad: vfs_freeopts(opts); return (error); } /* * Merge the old mount options with the new ones passed * in the MNT_UPDATE case. * * XXX: This function will keep a "nofoo" option in the new * options. E.g, if the option's canonical name is "foo", * "nofoo" ends up in the mount point's active options. */ static void vfs_mergeopts(struct vfsoptlist *toopts, struct vfsoptlist *oldopts) { struct vfsopt *opt, *new; TAILQ_FOREACH(opt, oldopts, link) { new = malloc(sizeof(struct vfsopt), M_MOUNT, M_WAITOK); new->name = strdup(opt->name, M_MOUNT); if (opt->len != 0) { new->value = malloc(opt->len, M_MOUNT, M_WAITOK); bcopy(opt->value, new->value, opt->len); } else new->value = NULL; new->len = opt->len; new->seen = opt->seen; TAILQ_INSERT_HEAD(toopts, new, link); } vfs_sanitizeopts(toopts); } /* * Mount a filesystem. */ int sys_nmount(td, uap) struct thread *td; struct nmount_args /* { struct iovec *iovp; unsigned int iovcnt; int flags; } */ *uap; { struct uio *auio; int error; u_int iovcnt; uint64_t flags; /* * Mount flags are now 64-bits. On 32-bit archtectures only * 32-bits are passed in, but from here on everything handles * 64-bit flags correctly. */ flags = uap->flags; AUDIT_ARG_FFLAGS(flags); CTR4(KTR_VFS, "%s: iovp %p with iovcnt %d and flags %d", __func__, uap->iovp, uap->iovcnt, flags); /* * Filter out MNT_ROOTFS. We do not want clients of nmount() in * userspace to set this flag, but we must filter it out if we want * MNT_UPDATE on the root file system to work. * MNT_ROOTFS should only be set by the kernel when mounting its * root file system. */ flags &= ~MNT_ROOTFS; iovcnt = uap->iovcnt; /* * Check that we have an even number of iovec's * and that we have at least two options. */ if ((iovcnt & 1) || (iovcnt < 4)) { CTR2(KTR_VFS, "%s: failed for invalid iovcnt %d", __func__, uap->iovcnt); return (EINVAL); } error = copyinuio(uap->iovp, iovcnt, &auio); if (error) { CTR2(KTR_VFS, "%s: failed for invalid uio op with %d errno", __func__, error); return (error); } error = vfs_donmount(td, flags, auio); free(auio, M_IOV); return (error); } /* * --------------------------------------------------------------------- * Various utility functions */ void vfs_ref(struct mount *mp) { CTR2(KTR_VFS, "%s: mp %p", __func__, mp); MNT_ILOCK(mp); MNT_REF(mp); MNT_IUNLOCK(mp); } void vfs_rel(struct mount *mp) { CTR2(KTR_VFS, "%s: mp %p", __func__, mp); MNT_ILOCK(mp); MNT_REL(mp); MNT_IUNLOCK(mp); } /* * Allocate and initialize the mount point struct. */ struct mount * vfs_mount_alloc(struct vnode *vp, struct vfsconf *vfsp, const char *fspath, struct ucred *cred) { struct mount *mp; mp = uma_zalloc(mount_zone, M_WAITOK); bzero(&mp->mnt_startzero, __rangeof(struct mount, mnt_startzero, mnt_endzero)); TAILQ_INIT(&mp->mnt_nvnodelist); mp->mnt_nvnodelistsize = 0; TAILQ_INIT(&mp->mnt_activevnodelist); mp->mnt_activevnodelistsize = 0; TAILQ_INIT(&mp->mnt_tmpfreevnodelist); mp->mnt_tmpfreevnodelistsize = 0; mp->mnt_ref = 0; (void) vfs_busy(mp, MBF_NOWAIT); atomic_add_acq_int(&vfsp->vfc_refcount, 1); mp->mnt_op = vfsp->vfc_vfsops; mp->mnt_vfc = vfsp; mp->mnt_stat.f_type = vfsp->vfc_typenum; mp->mnt_gen++; strlcpy(mp->mnt_stat.f_fstypename, vfsp->vfc_name, MFSNAMELEN); mp->mnt_vnodecovered = vp; mp->mnt_cred = crdup(cred); mp->mnt_stat.f_owner = cred->cr_uid; strlcpy(mp->mnt_stat.f_mntonname, fspath, MNAMELEN); mp->mnt_iosize_max = DFLTPHYS; #ifdef MAC mac_mount_init(mp); mac_mount_create(cred, mp); #endif arc4rand(&mp->mnt_hashseed, sizeof mp->mnt_hashseed, 0); TAILQ_INIT(&mp->mnt_uppers); return (mp); } /* * Destroy the mount struct previously allocated by vfs_mount_alloc(). */ void vfs_mount_destroy(struct mount *mp) { MNT_ILOCK(mp); mp->mnt_kern_flag |= MNTK_REFEXPIRE; if (mp->mnt_kern_flag & MNTK_MWAIT) { mp->mnt_kern_flag &= ~MNTK_MWAIT; wakeup(mp); } while (mp->mnt_ref) msleep(mp, MNT_MTX(mp), PVFS, "mntref", 0); KASSERT(mp->mnt_ref == 0, ("%s: invalid refcount in the drain path @ %s:%d", __func__, __FILE__, __LINE__)); if (mp->mnt_writeopcount != 0) panic("vfs_mount_destroy: nonzero writeopcount"); if (mp->mnt_secondary_writes != 0) panic("vfs_mount_destroy: nonzero secondary_writes"); atomic_subtract_rel_int(&mp->mnt_vfc->vfc_refcount, 1); if (!TAILQ_EMPTY(&mp->mnt_nvnodelist)) { struct vnode *vp; TAILQ_FOREACH(vp, &mp->mnt_nvnodelist, v_nmntvnodes) vn_printf(vp, "dangling vnode "); panic("unmount: dangling vnode"); } KASSERT(TAILQ_EMPTY(&mp->mnt_uppers), ("mnt_uppers")); if (mp->mnt_nvnodelistsize != 0) panic("vfs_mount_destroy: nonzero nvnodelistsize"); if (mp->mnt_activevnodelistsize != 0) panic("vfs_mount_destroy: nonzero activevnodelistsize"); if (mp->mnt_tmpfreevnodelistsize != 0) panic("vfs_mount_destroy: nonzero tmpfreevnodelistsize"); if (mp->mnt_lockref != 0) panic("vfs_mount_destroy: nonzero lock refcount"); MNT_IUNLOCK(mp); #ifdef MAC mac_mount_destroy(mp); #endif if (mp->mnt_opt != NULL) vfs_freeopts(mp->mnt_opt); crfree(mp->mnt_cred); uma_zfree(mount_zone, mp); } int vfs_donmount(struct thread *td, uint64_t fsflags, struct uio *fsoptions) { struct vfsoptlist *optlist; struct vfsopt *opt, *tmp_opt; char *fstype, *fspath, *errmsg; int error, fstypelen, fspathlen, errmsg_len, errmsg_pos; errmsg = fspath = NULL; errmsg_len = fspathlen = 0; errmsg_pos = -1; error = vfs_buildopts(fsoptions, &optlist); if (error) return (error); if (vfs_getopt(optlist, "errmsg", (void **)&errmsg, &errmsg_len) == 0) errmsg_pos = vfs_getopt_pos(optlist, "errmsg"); /* * We need these two options before the others, * and they are mandatory for any filesystem. * Ensure they are NUL terminated as well. */ fstypelen = 0; error = vfs_getopt(optlist, "fstype", (void **)&fstype, &fstypelen); if (error || fstype[fstypelen - 1] != '\0') { error = EINVAL; if (errmsg != NULL) strncpy(errmsg, "Invalid fstype", errmsg_len); goto bail; } fspathlen = 0; error = vfs_getopt(optlist, "fspath", (void **)&fspath, &fspathlen); if (error || fspath[fspathlen - 1] != '\0') { error = EINVAL; if (errmsg != NULL) strncpy(errmsg, "Invalid fspath", errmsg_len); goto bail; } /* * We need to see if we have the "update" option * before we call vfs_domount(), since vfs_domount() has special * logic based on MNT_UPDATE. This is very important * when we want to update the root filesystem. */ TAILQ_FOREACH_SAFE(opt, optlist, link, tmp_opt) { if (strcmp(opt->name, "update") == 0) { fsflags |= MNT_UPDATE; vfs_freeopt(optlist, opt); } else if (strcmp(opt->name, "async") == 0) fsflags |= MNT_ASYNC; else if (strcmp(opt->name, "force") == 0) { fsflags |= MNT_FORCE; vfs_freeopt(optlist, opt); } else if (strcmp(opt->name, "reload") == 0) { fsflags |= MNT_RELOAD; vfs_freeopt(optlist, opt); } else if (strcmp(opt->name, "multilabel") == 0) fsflags |= MNT_MULTILABEL; else if (strcmp(opt->name, "noasync") == 0) fsflags &= ~MNT_ASYNC; else if (strcmp(opt->name, "noatime") == 0) fsflags |= MNT_NOATIME; else if (strcmp(opt->name, "atime") == 0) { free(opt->name, M_MOUNT); opt->name = strdup("nonoatime", M_MOUNT); } else if (strcmp(opt->name, "noclusterr") == 0) fsflags |= MNT_NOCLUSTERR; else if (strcmp(opt->name, "clusterr") == 0) { free(opt->name, M_MOUNT); opt->name = strdup("nonoclusterr", M_MOUNT); } else if (strcmp(opt->name, "noclusterw") == 0) fsflags |= MNT_NOCLUSTERW; else if (strcmp(opt->name, "clusterw") == 0) { free(opt->name, M_MOUNT); opt->name = strdup("nonoclusterw", M_MOUNT); } else if (strcmp(opt->name, "noexec") == 0) fsflags |= MNT_NOEXEC; else if (strcmp(opt->name, "exec") == 0) { free(opt->name, M_MOUNT); opt->name = strdup("nonoexec", M_MOUNT); } else if (strcmp(opt->name, "nosuid") == 0) fsflags |= MNT_NOSUID; else if (strcmp(opt->name, "suid") == 0) { free(opt->name, M_MOUNT); opt->name = strdup("nonosuid", M_MOUNT); } else if (strcmp(opt->name, "nosymfollow") == 0) fsflags |= MNT_NOSYMFOLLOW; else if (strcmp(opt->name, "symfollow") == 0) { free(opt->name, M_MOUNT); opt->name = strdup("nonosymfollow", M_MOUNT); } else if (strcmp(opt->name, "noro") == 0) fsflags &= ~MNT_RDONLY; else if (strcmp(opt->name, "rw") == 0) fsflags &= ~MNT_RDONLY; else if (strcmp(opt->name, "ro") == 0) fsflags |= MNT_RDONLY; else if (strcmp(opt->name, "rdonly") == 0) { free(opt->name, M_MOUNT); opt->name = strdup("ro", M_MOUNT); fsflags |= MNT_RDONLY; } else if (strcmp(opt->name, "suiddir") == 0) fsflags |= MNT_SUIDDIR; else if (strcmp(opt->name, "sync") == 0) fsflags |= MNT_SYNCHRONOUS; else if (strcmp(opt->name, "union") == 0) fsflags |= MNT_UNION; else if (strcmp(opt->name, "automounted") == 0) { fsflags |= MNT_AUTOMOUNTED; vfs_freeopt(optlist, opt); } } /* * Be ultra-paranoid about making sure the type and fspath * variables will fit in our mp buffers, including the * terminating NUL. */ if (fstypelen > MFSNAMELEN || fspathlen > MNAMELEN) { error = ENAMETOOLONG; goto bail; } error = vfs_domount(td, fstype, fspath, fsflags, &optlist); bail: /* copyout the errmsg */ if (errmsg_pos != -1 && ((2 * errmsg_pos + 1) < fsoptions->uio_iovcnt) && errmsg_len > 0 && errmsg != NULL) { if (fsoptions->uio_segflg == UIO_SYSSPACE) { bcopy(errmsg, fsoptions->uio_iov[2 * errmsg_pos + 1].iov_base, fsoptions->uio_iov[2 * errmsg_pos + 1].iov_len); } else { copyout(errmsg, fsoptions->uio_iov[2 * errmsg_pos + 1].iov_base, fsoptions->uio_iov[2 * errmsg_pos + 1].iov_len); } } if (optlist != NULL) vfs_freeopts(optlist); return (error); } /* * Old mount API. */ #ifndef _SYS_SYSPROTO_H_ struct mount_args { char *type; char *path; int flags; caddr_t data; }; #endif /* ARGSUSED */ int sys_mount(td, uap) struct thread *td; struct mount_args /* { char *type; char *path; int flags; caddr_t data; } */ *uap; { char *fstype; struct vfsconf *vfsp = NULL; struct mntarg *ma = NULL; uint64_t flags; int error; /* * Mount flags are now 64-bits. On 32-bit architectures only * 32-bits are passed in, but from here on everything handles * 64-bit flags correctly. */ flags = uap->flags; AUDIT_ARG_FFLAGS(flags); /* * Filter out MNT_ROOTFS. We do not want clients of mount() in * userspace to set this flag, but we must filter it out if we want * MNT_UPDATE on the root file system to work. * MNT_ROOTFS should only be set by the kernel when mounting its * root file system. */ flags &= ~MNT_ROOTFS; fstype = malloc(MFSNAMELEN, M_TEMP, M_WAITOK); error = copyinstr(uap->type, fstype, MFSNAMELEN, NULL); if (error) { free(fstype, M_TEMP); return (error); } AUDIT_ARG_TEXT(fstype); vfsp = vfs_byname_kld(fstype, td, &error); free(fstype, M_TEMP); if (vfsp == NULL) return (ENOENT); if (vfsp->vfc_vfsops->vfs_cmount == NULL) return (EOPNOTSUPP); ma = mount_argsu(ma, "fstype", uap->type, MFSNAMELEN); ma = mount_argsu(ma, "fspath", uap->path, MNAMELEN); ma = mount_argb(ma, flags & MNT_RDONLY, "noro"); ma = mount_argb(ma, !(flags & MNT_NOSUID), "nosuid"); ma = mount_argb(ma, !(flags & MNT_NOEXEC), "noexec"); error = vfsp->vfc_vfsops->vfs_cmount(ma, uap->data, flags); return (error); } /* * vfs_domount_first(): first file system mount (not update) */ static int vfs_domount_first( struct thread *td, /* Calling thread. */ struct vfsconf *vfsp, /* File system type. */ char *fspath, /* Mount path. */ struct vnode *vp, /* Vnode to be covered. */ uint64_t fsflags, /* Flags common to all filesystems. */ struct vfsoptlist **optlist /* Options local to the filesystem. */ ) { struct vattr va; struct mount *mp; struct vnode *newdp; int error; ASSERT_VOP_ELOCKED(vp, __func__); KASSERT((fsflags & MNT_UPDATE) == 0, ("MNT_UPDATE shouldn't be here")); /* * If the user is not root, ensure that they own the directory * onto which we are attempting to mount. */ error = VOP_GETATTR(vp, &va, td->td_ucred); if (error == 0 && va.va_uid != td->td_ucred->cr_uid) error = priv_check_cred(td->td_ucred, PRIV_VFS_ADMIN, 0); if (error == 0) error = vinvalbuf(vp, V_SAVE, 0, 0); if (error == 0 && vp->v_type != VDIR) error = ENOTDIR; if (error == 0) { VI_LOCK(vp); if ((vp->v_iflag & VI_MOUNT) == 0 && vp->v_mountedhere == NULL) vp->v_iflag |= VI_MOUNT; else error = EBUSY; VI_UNLOCK(vp); } if (error != 0) { vput(vp); return (error); } VOP_UNLOCK(vp, 0); /* Allocate and initialize the filesystem. */ mp = vfs_mount_alloc(vp, vfsp, fspath, td->td_ucred); /* XXXMAC: pass to vfs_mount_alloc? */ mp->mnt_optnew = *optlist; /* Set the mount level flags. */ mp->mnt_flag = (fsflags & (MNT_UPDATEMASK | MNT_ROOTFS | MNT_RDONLY)); /* * Mount the filesystem. * XXX The final recipients of VFS_MOUNT just overwrite the ndp they * get. No freeing of cn_pnbuf. */ error = VFS_MOUNT(mp); if (error != 0) { vfs_unbusy(mp); vfs_mount_destroy(mp); VI_LOCK(vp); vp->v_iflag &= ~VI_MOUNT; VI_UNLOCK(vp); vrele(vp); return (error); } if (mp->mnt_opt != NULL) vfs_freeopts(mp->mnt_opt); mp->mnt_opt = mp->mnt_optnew; *optlist = NULL; (void)VFS_STATFS(mp, &mp->mnt_stat); /* * Prevent external consumers of mount options from reading mnt_optnew. */ mp->mnt_optnew = NULL; MNT_ILOCK(mp); if ((mp->mnt_flag & MNT_ASYNC) != 0 && (mp->mnt_kern_flag & MNTK_NOASYNC) == 0) mp->mnt_kern_flag |= MNTK_ASYNC; else mp->mnt_kern_flag &= ~MNTK_ASYNC; MNT_IUNLOCK(mp); vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); cache_purge(vp); VI_LOCK(vp); vp->v_iflag &= ~VI_MOUNT; VI_UNLOCK(vp); vp->v_mountedhere = mp; /* Place the new filesystem at the end of the mount list. */ mtx_lock(&mountlist_mtx); TAILQ_INSERT_TAIL(&mountlist, mp, mnt_list); mtx_unlock(&mountlist_mtx); vfs_event_signal(NULL, VQ_MOUNT, 0); if (VFS_ROOT(mp, LK_EXCLUSIVE, &newdp)) panic("mount: lost mount"); VOP_UNLOCK(vp, 0); EVENTHANDLER_INVOKE(vfs_mounted, mp, newdp, td); VOP_UNLOCK(newdp, 0); mountcheckdirs(vp, newdp); vrele(newdp); if ((mp->mnt_flag & MNT_RDONLY) == 0) vfs_allocate_syncvnode(mp); vfs_unbusy(mp); return (0); } /* * vfs_domount_update(): update of mounted file system */ static int vfs_domount_update( struct thread *td, /* Calling thread. */ struct vnode *vp, /* Mount point vnode. */ uint64_t fsflags, /* Flags common to all filesystems. */ struct vfsoptlist **optlist /* Options local to the filesystem. */ ) { struct export_args export; void *bufp; struct mount *mp; int error, export_error, len; uint64_t flag; ASSERT_VOP_ELOCKED(vp, __func__); KASSERT((fsflags & MNT_UPDATE) != 0, ("MNT_UPDATE should be here")); mp = vp->v_mount; if ((vp->v_vflag & VV_ROOT) == 0) { if (vfs_copyopt(*optlist, "export", &export, sizeof(export)) == 0) error = EXDEV; else error = EINVAL; vput(vp); return (error); } /* * We only allow the filesystem to be reloaded if it * is currently mounted read-only. */ flag = mp->mnt_flag; if ((fsflags & MNT_RELOAD) != 0 && (flag & MNT_RDONLY) == 0) { vput(vp); return (EOPNOTSUPP); /* Needs translation */ } /* * Only privileged root, or (if MNT_USER is set) the user that * did the original mount is permitted to update it. */ error = vfs_suser(mp, td); if (error != 0) { vput(vp); return (error); } if (vfs_busy(mp, MBF_NOWAIT)) { vput(vp); return (EBUSY); } VI_LOCK(vp); if ((vp->v_iflag & VI_MOUNT) != 0 || vp->v_mountedhere != NULL) { VI_UNLOCK(vp); vfs_unbusy(mp); vput(vp); return (EBUSY); } vp->v_iflag |= VI_MOUNT; VI_UNLOCK(vp); VOP_UNLOCK(vp, 0); MNT_ILOCK(mp); if ((mp->mnt_kern_flag & MNTK_UNMOUNT) != 0) { MNT_IUNLOCK(mp); error = EBUSY; goto end; } mp->mnt_flag &= ~MNT_UPDATEMASK; mp->mnt_flag |= fsflags & (MNT_RELOAD | MNT_FORCE | MNT_UPDATE | MNT_SNAPSHOT | MNT_ROOTFS | MNT_UPDATEMASK | MNT_RDONLY); if ((mp->mnt_flag & MNT_ASYNC) == 0) mp->mnt_kern_flag &= ~MNTK_ASYNC; MNT_IUNLOCK(mp); mp->mnt_optnew = *optlist; vfs_mergeopts(mp->mnt_optnew, mp->mnt_opt); /* * Mount the filesystem. * XXX The final recipients of VFS_MOUNT just overwrite the ndp they * get. No freeing of cn_pnbuf. */ error = VFS_MOUNT(mp); export_error = 0; /* Process the export option. */ if (error == 0 && vfs_getopt(mp->mnt_optnew, "export", &bufp, &len) == 0) { /* Assume that there is only 1 ABI for each length. */ switch (len) { case (sizeof(struct oexport_args)): bzero(&export, sizeof(export)); /* FALLTHROUGH */ case (sizeof(export)): bcopy(bufp, &export, len); export_error = vfs_export(mp, &export); break; default: export_error = EINVAL; break; } } MNT_ILOCK(mp); if (error == 0) { mp->mnt_flag &= ~(MNT_UPDATE | MNT_RELOAD | MNT_FORCE | MNT_SNAPSHOT); } else { /* * If we fail, restore old mount flags. MNT_QUOTA is special, * because it is not part of MNT_UPDATEMASK, but it could have * changed in the meantime if quotactl(2) was called. * All in all we want current value of MNT_QUOTA, not the old * one. */ mp->mnt_flag = (mp->mnt_flag & MNT_QUOTA) | (flag & ~MNT_QUOTA); } if ((mp->mnt_flag & MNT_ASYNC) != 0 && (mp->mnt_kern_flag & MNTK_NOASYNC) == 0) mp->mnt_kern_flag |= MNTK_ASYNC; else mp->mnt_kern_flag &= ~MNTK_ASYNC; MNT_IUNLOCK(mp); if (error != 0) goto end; if (mp->mnt_opt != NULL) vfs_freeopts(mp->mnt_opt); mp->mnt_opt = mp->mnt_optnew; *optlist = NULL; (void)VFS_STATFS(mp, &mp->mnt_stat); /* * Prevent external consumers of mount options from reading * mnt_optnew. */ mp->mnt_optnew = NULL; if ((mp->mnt_flag & MNT_RDONLY) == 0) vfs_allocate_syncvnode(mp); else vfs_deallocate_syncvnode(mp); end: vfs_unbusy(mp); VI_LOCK(vp); vp->v_iflag &= ~VI_MOUNT; VI_UNLOCK(vp); vrele(vp); return (error != 0 ? error : export_error); } /* * vfs_domount(): actually attempt a filesystem mount. */ static int vfs_domount( struct thread *td, /* Calling thread. */ const char *fstype, /* Filesystem type. */ char *fspath, /* Mount path. */ uint64_t fsflags, /* Flags common to all filesystems. */ struct vfsoptlist **optlist /* Options local to the filesystem. */ ) { struct vfsconf *vfsp; struct nameidata nd; struct vnode *vp; char *pathbuf; int error; /* * Be ultra-paranoid about making sure the type and fspath * variables will fit in our mp buffers, including the * terminating NUL. */ if (strlen(fstype) >= MFSNAMELEN || strlen(fspath) >= MNAMELEN) return (ENAMETOOLONG); if (jailed(td->td_ucred) || usermount == 0) { if ((error = priv_check(td, PRIV_VFS_MOUNT)) != 0) return (error); } /* * Do not allow NFS export or MNT_SUIDDIR by unprivileged users. */ if (fsflags & MNT_EXPORTED) { error = priv_check(td, PRIV_VFS_MOUNT_EXPORTED); if (error) return (error); } if (fsflags & MNT_SUIDDIR) { error = priv_check(td, PRIV_VFS_MOUNT_SUIDDIR); if (error) return (error); } /* * Silently enforce MNT_NOSUID and MNT_USER for unprivileged users. */ if ((fsflags & (MNT_NOSUID | MNT_USER)) != (MNT_NOSUID | MNT_USER)) { if (priv_check(td, PRIV_VFS_MOUNT_NONUSER) != 0) fsflags |= MNT_NOSUID | MNT_USER; } /* Load KLDs before we lock the covered vnode to avoid reversals. */ vfsp = NULL; if ((fsflags & MNT_UPDATE) == 0) { /* Don't try to load KLDs if we're mounting the root. */ if (fsflags & MNT_ROOTFS) vfsp = vfs_byname(fstype); else vfsp = vfs_byname_kld(fstype, td, &error); if (vfsp == NULL) return (ENODEV); if (jailed(td->td_ucred) && !(vfsp->vfc_flags & VFCF_JAIL)) return (EPERM); } /* * Get vnode to be covered or mount point's vnode in case of MNT_UPDATE. */ NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF | AUDITVNODE1, UIO_SYSSPACE, fspath, td); error = namei(&nd); if (error != 0) return (error); NDFREE(&nd, NDF_ONLY_PNBUF); vp = nd.ni_vp; if ((fsflags & MNT_UPDATE) == 0) { pathbuf = malloc(MNAMELEN, M_TEMP, M_WAITOK); strcpy(pathbuf, fspath); error = vn_path_to_global_path(td, vp, pathbuf, MNAMELEN); /* debug.disablefullpath == 1 results in ENODEV */ if (error == 0 || error == ENODEV) { error = vfs_domount_first(td, vfsp, pathbuf, vp, fsflags, optlist); } free(pathbuf, M_TEMP); } else error = vfs_domount_update(td, vp, fsflags, optlist); return (error); } /* * Unmount a filesystem. * * Note: unmount takes a path to the vnode mounted on as argument, not * special file (as before). */ #ifndef _SYS_SYSPROTO_H_ struct unmount_args { char *path; int flags; }; #endif /* ARGSUSED */ int sys_unmount(struct thread *td, struct unmount_args *uap) { struct nameidata nd; struct mount *mp; char *pathbuf; int error, id0, id1; AUDIT_ARG_VALUE(uap->flags); if (jailed(td->td_ucred) || usermount == 0) { error = priv_check(td, PRIV_VFS_UNMOUNT); if (error) return (error); } pathbuf = malloc(MNAMELEN, M_TEMP, M_WAITOK); error = copyinstr(uap->path, pathbuf, MNAMELEN, NULL); if (error) { free(pathbuf, M_TEMP); return (error); } if (uap->flags & MNT_BYFSID) { AUDIT_ARG_TEXT(pathbuf); /* Decode the filesystem ID. */ if (sscanf(pathbuf, "FSID:%d:%d", &id0, &id1) != 2) { free(pathbuf, M_TEMP); return (EINVAL); } mtx_lock(&mountlist_mtx); TAILQ_FOREACH_REVERSE(mp, &mountlist, mntlist, mnt_list) { if (mp->mnt_stat.f_fsid.val[0] == id0 && mp->mnt_stat.f_fsid.val[1] == id1) { vfs_ref(mp); break; } } mtx_unlock(&mountlist_mtx); } else { /* * Try to find global path for path argument. */ NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF | AUDITVNODE1, UIO_SYSSPACE, pathbuf, td); if (namei(&nd) == 0) { NDFREE(&nd, NDF_ONLY_PNBUF); error = vn_path_to_global_path(td, nd.ni_vp, pathbuf, MNAMELEN); if (error == 0 || error == ENODEV) vput(nd.ni_vp); } mtx_lock(&mountlist_mtx); TAILQ_FOREACH_REVERSE(mp, &mountlist, mntlist, mnt_list) { if (strcmp(mp->mnt_stat.f_mntonname, pathbuf) == 0) { vfs_ref(mp); break; } } mtx_unlock(&mountlist_mtx); } free(pathbuf, M_TEMP); if (mp == NULL) { /* * Previously we returned ENOENT for a nonexistent path and * EINVAL for a non-mountpoint. We cannot tell these apart * now, so in the !MNT_BYFSID case return the more likely * EINVAL for compatibility. */ return ((uap->flags & MNT_BYFSID) ? ENOENT : EINVAL); } /* * Don't allow unmounting the root filesystem. */ if (mp->mnt_flag & MNT_ROOTFS) { vfs_rel(mp); return (EINVAL); } error = dounmount(mp, uap->flags, td); return (error); } /* * Return error if any of the vnodes, ignoring the root vnode * and the syncer vnode, have non-zero usecount. * * This function is purely advisory - it can return false positives * and negatives. */ static int vfs_check_usecounts(struct mount *mp) { struct vnode *vp, *mvp; MNT_VNODE_FOREACH_ALL(vp, mp, mvp) { if ((vp->v_vflag & VV_ROOT) == 0 && vp->v_type != VNON && vp->v_usecount != 0) { VI_UNLOCK(vp); MNT_VNODE_FOREACH_ALL_ABORT(mp, mvp); return (EBUSY); } VI_UNLOCK(vp); } return (0); } static void dounmount_cleanup(struct mount *mp, struct vnode *coveredvp, int mntkflags) { mtx_assert(MNT_MTX(mp), MA_OWNED); mp->mnt_kern_flag &= ~mntkflags; if ((mp->mnt_kern_flag & MNTK_MWAIT) != 0) { mp->mnt_kern_flag &= ~MNTK_MWAIT; wakeup(mp); } MNT_IUNLOCK(mp); if (coveredvp != NULL) { VOP_UNLOCK(coveredvp, 0); vdrop(coveredvp); } vn_finished_write(mp); } /* * Do the actual filesystem unmount. */ int dounmount(struct mount *mp, int flags, struct thread *td) { struct vnode *coveredvp, *fsrootvp; int error; uint64_t async_flag; int mnt_gen_r; if ((coveredvp = mp->mnt_vnodecovered) != NULL) { mnt_gen_r = mp->mnt_gen; VI_LOCK(coveredvp); vholdl(coveredvp); vn_lock(coveredvp, LK_EXCLUSIVE | LK_INTERLOCK | LK_RETRY); /* * Check for mp being unmounted while waiting for the * covered vnode lock. */ if (coveredvp->v_mountedhere != mp || coveredvp->v_mountedhere->mnt_gen != mnt_gen_r) { VOP_UNLOCK(coveredvp, 0); vdrop(coveredvp); vfs_rel(mp); return (EBUSY); } } /* * Only privileged root, or (if MNT_USER is set) the user that did the * original mount is permitted to unmount this filesystem. */ error = vfs_suser(mp, td); if (error != 0) { if (coveredvp != NULL) { VOP_UNLOCK(coveredvp, 0); vdrop(coveredvp); } vfs_rel(mp); return (error); } vn_start_write(NULL, &mp, V_WAIT | V_MNTREF); MNT_ILOCK(mp); if ((mp->mnt_kern_flag & MNTK_UNMOUNT) != 0 || (mp->mnt_flag & MNT_UPDATE) != 0 || !TAILQ_EMPTY(&mp->mnt_uppers)) { dounmount_cleanup(mp, coveredvp, 0); return (EBUSY); } mp->mnt_kern_flag |= MNTK_UNMOUNT | MNTK_NOINSMNTQ; if (flags & MNT_NONBUSY) { MNT_IUNLOCK(mp); error = vfs_check_usecounts(mp); MNT_ILOCK(mp); if (error != 0) { dounmount_cleanup(mp, coveredvp, MNTK_UNMOUNT | MNTK_NOINSMNTQ); return (error); } } /* Allow filesystems to detect that a forced unmount is in progress. */ if (flags & MNT_FORCE) { mp->mnt_kern_flag |= MNTK_UNMOUNTF; MNT_IUNLOCK(mp); /* * Must be done after setting MNTK_UNMOUNTF and before * waiting for mnt_lockref to become 0. */ VFS_PURGE(mp); MNT_ILOCK(mp); } error = 0; if (mp->mnt_lockref) { mp->mnt_kern_flag |= MNTK_DRAINING; error = msleep(&mp->mnt_lockref, MNT_MTX(mp), PVFS, "mount drain", 0); } MNT_IUNLOCK(mp); KASSERT(mp->mnt_lockref == 0, ("%s: invalid lock refcount in the drain path @ %s:%d", __func__, __FILE__, __LINE__)); KASSERT(error == 0, ("%s: invalid return value for msleep in the drain path @ %s:%d", __func__, __FILE__, __LINE__)); if (mp->mnt_flag & MNT_EXPUBLIC) vfs_setpublicfs(NULL, NULL, NULL); /* * From now, we can claim that the use reference on the * coveredvp is ours, and the ref can be released only by * successfull unmount by us, or left for later unmount * attempt. The previously acquired hold reference is no * longer needed to protect the vnode from reuse. */ if (coveredvp != NULL) vdrop(coveredvp); vfs_msync(mp, MNT_WAIT); MNT_ILOCK(mp); async_flag = mp->mnt_flag & MNT_ASYNC; mp->mnt_flag &= ~MNT_ASYNC; mp->mnt_kern_flag &= ~MNTK_ASYNC; MNT_IUNLOCK(mp); cache_purgevfs(mp, false); /* remove cache entries for this file sys */ vfs_deallocate_syncvnode(mp); /* * For forced unmounts, move process cdir/rdir refs on the fs root * vnode to the covered vnode. For non-forced unmounts we want * such references to cause an EBUSY error. */ if ((flags & MNT_FORCE) && VFS_ROOT(mp, LK_EXCLUSIVE, &fsrootvp) == 0) { if (mp->mnt_vnodecovered != NULL && (mp->mnt_flag & MNT_IGNORE) == 0) mountcheckdirs(fsrootvp, mp->mnt_vnodecovered); if (fsrootvp == rootvnode) { vrele(rootvnode); rootvnode = NULL; } vput(fsrootvp); } if ((mp->mnt_flag & MNT_RDONLY) != 0 || (flags & MNT_FORCE) != 0 || (error = VFS_SYNC(mp, MNT_WAIT)) == 0) error = VFS_UNMOUNT(mp, flags); vn_finished_write(mp); /* * If we failed to flush the dirty blocks for this mount point, * undo all the cdir/rdir and rootvnode changes we made above. * Unless we failed to do so because the device is reporting that * it doesn't exist anymore. */ if (error && error != ENXIO) { if ((flags & MNT_FORCE) && VFS_ROOT(mp, LK_EXCLUSIVE, &fsrootvp) == 0) { if (mp->mnt_vnodecovered != NULL && (mp->mnt_flag & MNT_IGNORE) == 0) mountcheckdirs(mp->mnt_vnodecovered, fsrootvp); if (rootvnode == NULL) { rootvnode = fsrootvp; vref(rootvnode); } vput(fsrootvp); } MNT_ILOCK(mp); mp->mnt_kern_flag &= ~MNTK_NOINSMNTQ; if ((mp->mnt_flag & MNT_RDONLY) == 0) { MNT_IUNLOCK(mp); vfs_allocate_syncvnode(mp); MNT_ILOCK(mp); } mp->mnt_kern_flag &= ~(MNTK_UNMOUNT | MNTK_UNMOUNTF); mp->mnt_flag |= async_flag; if ((mp->mnt_flag & MNT_ASYNC) != 0 && (mp->mnt_kern_flag & MNTK_NOASYNC) == 0) mp->mnt_kern_flag |= MNTK_ASYNC; if (mp->mnt_kern_flag & MNTK_MWAIT) { mp->mnt_kern_flag &= ~MNTK_MWAIT; wakeup(mp); } MNT_IUNLOCK(mp); if (coveredvp) VOP_UNLOCK(coveredvp, 0); return (error); } mtx_lock(&mountlist_mtx); TAILQ_REMOVE(&mountlist, mp, mnt_list); mtx_unlock(&mountlist_mtx); EVENTHANDLER_INVOKE(vfs_unmounted, mp, td); if (coveredvp != NULL) { coveredvp->v_mountedhere = NULL; vput(coveredvp); } vfs_event_signal(NULL, VQ_UNMOUNT, 0); if (mp == rootdevmp) rootdevmp = NULL; vfs_mount_destroy(mp); return (0); } /* * Report errors during filesystem mounting. */ void vfs_mount_error(struct mount *mp, const char *fmt, ...) { struct vfsoptlist *moptlist = mp->mnt_optnew; va_list ap; int error, len; char *errmsg; error = vfs_getopt(moptlist, "errmsg", (void **)&errmsg, &len); if (error || errmsg == NULL || len <= 0) return; va_start(ap, fmt); vsnprintf(errmsg, (size_t)len, fmt, ap); va_end(ap); } void vfs_opterror(struct vfsoptlist *opts, const char *fmt, ...) { va_list ap; int error, len; char *errmsg; error = vfs_getopt(opts, "errmsg", (void **)&errmsg, &len); if (error || errmsg == NULL || len <= 0) return; va_start(ap, fmt); vsnprintf(errmsg, (size_t)len, fmt, ap); va_end(ap); } /* * --------------------------------------------------------------------- * Functions for querying mount options/arguments from filesystems. */ /* * Check that no unknown options are given */ int vfs_filteropt(struct vfsoptlist *opts, const char **legal) { struct vfsopt *opt; char errmsg[255]; const char **t, *p, *q; int ret = 0; TAILQ_FOREACH(opt, opts, link) { p = opt->name; q = NULL; if (p[0] == 'n' && p[1] == 'o') q = p + 2; for(t = global_opts; *t != NULL; t++) { if (strcmp(*t, p) == 0) break; if (q != NULL) { if (strcmp(*t, q) == 0) break; } } if (*t != NULL) continue; for(t = legal; *t != NULL; t++) { if (strcmp(*t, p) == 0) break; if (q != NULL) { if (strcmp(*t, q) == 0) break; } } if (*t != NULL) continue; snprintf(errmsg, sizeof(errmsg), "mount option <%s> is unknown", p); ret = EINVAL; } if (ret != 0) { TAILQ_FOREACH(opt, opts, link) { if (strcmp(opt->name, "errmsg") == 0) { strncpy((char *)opt->value, errmsg, opt->len); break; } } if (opt == NULL) printf("%s\n", errmsg); } return (ret); } /* * Get a mount option by its name. * * Return 0 if the option was found, ENOENT otherwise. * If len is non-NULL it will be filled with the length * of the option. If buf is non-NULL, it will be filled * with the address of the option. */ int vfs_getopt(opts, name, buf, len) struct vfsoptlist *opts; const char *name; void **buf; int *len; { struct vfsopt *opt; KASSERT(opts != NULL, ("vfs_getopt: caller passed 'opts' as NULL")); TAILQ_FOREACH(opt, opts, link) { if (strcmp(name, opt->name) == 0) { opt->seen = 1; if (len != NULL) *len = opt->len; if (buf != NULL) *buf = opt->value; return (0); } } return (ENOENT); } int vfs_getopt_pos(struct vfsoptlist *opts, const char *name) { struct vfsopt *opt; if (opts == NULL) return (-1); TAILQ_FOREACH(opt, opts, link) { if (strcmp(name, opt->name) == 0) { opt->seen = 1; return (opt->pos); } } return (-1); } int vfs_getopt_size(struct vfsoptlist *opts, const char *name, off_t *value) { char *opt_value, *vtp; quad_t iv; int error, opt_len; error = vfs_getopt(opts, name, (void **)&opt_value, &opt_len); if (error != 0) return (error); if (opt_len == 0 || opt_value == NULL) return (EINVAL); if (opt_value[0] == '\0' || opt_value[opt_len - 1] != '\0') return (EINVAL); iv = strtoq(opt_value, &vtp, 0); if (vtp == opt_value || (vtp[0] != '\0' && vtp[1] != '\0')) return (EINVAL); if (iv < 0) return (EINVAL); switch (vtp[0]) { case 't': case 'T': iv *= 1024; case 'g': case 'G': iv *= 1024; case 'm': case 'M': iv *= 1024; case 'k': case 'K': iv *= 1024; case '\0': break; default: return (EINVAL); } *value = iv; return (0); } char * vfs_getopts(struct vfsoptlist *opts, const char *name, int *error) { struct vfsopt *opt; *error = 0; TAILQ_FOREACH(opt, opts, link) { if (strcmp(name, opt->name) != 0) continue; opt->seen = 1; if (opt->len == 0 || ((char *)opt->value)[opt->len - 1] != '\0') { *error = EINVAL; return (NULL); } return (opt->value); } *error = ENOENT; return (NULL); } int vfs_flagopt(struct vfsoptlist *opts, const char *name, uint64_t *w, uint64_t val) { struct vfsopt *opt; TAILQ_FOREACH(opt, opts, link) { if (strcmp(name, opt->name) == 0) { opt->seen = 1; if (w != NULL) *w |= val; return (1); } } if (w != NULL) *w &= ~val; return (0); } int vfs_scanopt(struct vfsoptlist *opts, const char *name, const char *fmt, ...) { va_list ap; struct vfsopt *opt; int ret; KASSERT(opts != NULL, ("vfs_getopt: caller passed 'opts' as NULL")); TAILQ_FOREACH(opt, opts, link) { if (strcmp(name, opt->name) != 0) continue; opt->seen = 1; if (opt->len == 0 || opt->value == NULL) return (0); if (((char *)opt->value)[opt->len - 1] != '\0') return (0); va_start(ap, fmt); ret = vsscanf(opt->value, fmt, ap); va_end(ap); return (ret); } return (0); } int vfs_setopt(struct vfsoptlist *opts, const char *name, void *value, int len) { struct vfsopt *opt; TAILQ_FOREACH(opt, opts, link) { if (strcmp(name, opt->name) != 0) continue; opt->seen = 1; if (opt->value == NULL) opt->len = len; else { if (opt->len != len) return (EINVAL); bcopy(value, opt->value, len); } return (0); } return (ENOENT); } int vfs_setopt_part(struct vfsoptlist *opts, const char *name, void *value, int len) { struct vfsopt *opt; TAILQ_FOREACH(opt, opts, link) { if (strcmp(name, opt->name) != 0) continue; opt->seen = 1; if (opt->value == NULL) opt->len = len; else { if (opt->len < len) return (EINVAL); opt->len = len; bcopy(value, opt->value, len); } return (0); } return (ENOENT); } int vfs_setopts(struct vfsoptlist *opts, const char *name, const char *value) { struct vfsopt *opt; TAILQ_FOREACH(opt, opts, link) { if (strcmp(name, opt->name) != 0) continue; opt->seen = 1; if (opt->value == NULL) opt->len = strlen(value) + 1; else if (strlcpy(opt->value, value, opt->len) >= opt->len) return (EINVAL); return (0); } return (ENOENT); } /* * Find and copy a mount option. * * The size of the buffer has to be specified * in len, if it is not the same length as the * mount option, EINVAL is returned. * Returns ENOENT if the option is not found. */ int vfs_copyopt(opts, name, dest, len) struct vfsoptlist *opts; const char *name; void *dest; int len; { struct vfsopt *opt; KASSERT(opts != NULL, ("vfs_copyopt: caller passed 'opts' as NULL")); TAILQ_FOREACH(opt, opts, link) { if (strcmp(name, opt->name) == 0) { opt->seen = 1; if (len != opt->len) return (EINVAL); bcopy(opt->value, dest, opt->len); return (0); } } return (ENOENT); } int __vfs_statfs(struct mount *mp, struct statfs *sbp) { int error; error = mp->mnt_op->vfs_statfs(mp, &mp->mnt_stat); if (sbp != &mp->mnt_stat) *sbp = mp->mnt_stat; return (error); } void vfs_mountedfrom(struct mount *mp, const char *from) { bzero(mp->mnt_stat.f_mntfromname, sizeof mp->mnt_stat.f_mntfromname); strlcpy(mp->mnt_stat.f_mntfromname, from, sizeof mp->mnt_stat.f_mntfromname); } /* * --------------------------------------------------------------------- * This is the api for building mount args and mounting filesystems from * inside the kernel. * * The API works by accumulation of individual args. First error is * latched. * * XXX: should be documented in new manpage kernel_mount(9) */ /* A memory allocation which must be freed when we are done */ struct mntaarg { SLIST_ENTRY(mntaarg) next; }; /* The header for the mount arguments */ struct mntarg { struct iovec *v; int len; int error; SLIST_HEAD(, mntaarg) list; }; /* * Add a boolean argument. * * flag is the boolean value. * name must start with "no". */ struct mntarg * mount_argb(struct mntarg *ma, int flag, const char *name) { KASSERT(name[0] == 'n' && name[1] == 'o', ("mount_argb(...,%s): name must start with 'no'", name)); return (mount_arg(ma, name + (flag ? 2 : 0), NULL, 0)); } /* * Add an argument printf style */ struct mntarg * mount_argf(struct mntarg *ma, const char *name, const char *fmt, ...) { va_list ap; struct mntaarg *maa; struct sbuf *sb; int len; if (ma == NULL) { ma = malloc(sizeof *ma, M_MOUNT, M_WAITOK | M_ZERO); SLIST_INIT(&ma->list); } if (ma->error) return (ma); ma->v = realloc(ma->v, sizeof *ma->v * (ma->len + 2), M_MOUNT, M_WAITOK); ma->v[ma->len].iov_base = (void *)(uintptr_t)name; ma->v[ma->len].iov_len = strlen(name) + 1; ma->len++; sb = sbuf_new_auto(); va_start(ap, fmt); sbuf_vprintf(sb, fmt, ap); va_end(ap); sbuf_finish(sb); len = sbuf_len(sb) + 1; maa = malloc(sizeof *maa + len, M_MOUNT, M_WAITOK | M_ZERO); SLIST_INSERT_HEAD(&ma->list, maa, next); bcopy(sbuf_data(sb), maa + 1, len); sbuf_delete(sb); ma->v[ma->len].iov_base = maa + 1; ma->v[ma->len].iov_len = len; ma->len++; return (ma); } /* * Add an argument which is a userland string. */ struct mntarg * mount_argsu(struct mntarg *ma, const char *name, const void *val, int len) { struct mntaarg *maa; char *tbuf; if (val == NULL) return (ma); if (ma == NULL) { ma = malloc(sizeof *ma, M_MOUNT, M_WAITOK | M_ZERO); SLIST_INIT(&ma->list); } if (ma->error) return (ma); maa = malloc(sizeof *maa + len, M_MOUNT, M_WAITOK | M_ZERO); SLIST_INSERT_HEAD(&ma->list, maa, next); tbuf = (void *)(maa + 1); ma->error = copyinstr(val, tbuf, len, NULL); return (mount_arg(ma, name, tbuf, -1)); } /* * Plain argument. * * If length is -1, treat value as a C string. */ struct mntarg * mount_arg(struct mntarg *ma, const char *name, const void *val, int len) { if (ma == NULL) { ma = malloc(sizeof *ma, M_MOUNT, M_WAITOK | M_ZERO); SLIST_INIT(&ma->list); } if (ma->error) return (ma); ma->v = realloc(ma->v, sizeof *ma->v * (ma->len + 2), M_MOUNT, M_WAITOK); ma->v[ma->len].iov_base = (void *)(uintptr_t)name; ma->v[ma->len].iov_len = strlen(name) + 1; ma->len++; ma->v[ma->len].iov_base = (void *)(uintptr_t)val; if (len < 0) ma->v[ma->len].iov_len = strlen(val) + 1; else ma->v[ma->len].iov_len = len; ma->len++; return (ma); } /* * Free a mntarg structure */ static void free_mntarg(struct mntarg *ma) { struct mntaarg *maa; while (!SLIST_EMPTY(&ma->list)) { maa = SLIST_FIRST(&ma->list); SLIST_REMOVE_HEAD(&ma->list, next); free(maa, M_MOUNT); } free(ma->v, M_MOUNT); free(ma, M_MOUNT); } /* * Mount a filesystem */ int kernel_mount(struct mntarg *ma, uint64_t flags) { struct uio auio; int error; KASSERT(ma != NULL, ("kernel_mount NULL ma")); KASSERT(ma->v != NULL, ("kernel_mount NULL ma->v")); KASSERT(!(ma->len & 1), ("kernel_mount odd ma->len (%d)", ma->len)); auio.uio_iov = ma->v; auio.uio_iovcnt = ma->len; auio.uio_segflg = UIO_SYSSPACE; error = ma->error; if (!error) error = vfs_donmount(curthread, flags, &auio); free_mntarg(ma); return (error); } /* * A printflike function to mount a filesystem. */ int kernel_vmount(int flags, ...) { struct mntarg *ma = NULL; va_list ap; const char *cp; const void *vp; int error; va_start(ap, flags); for (;;) { cp = va_arg(ap, const char *); if (cp == NULL) break; vp = va_arg(ap, const void *); ma = mount_arg(ma, cp, vp, (vp != NULL ? -1 : 0)); } va_end(ap); error = kernel_mount(ma, flags); return (error); } void vfs_oexport_conv(const struct oexport_args *oexp, struct export_args *exp) { bcopy(oexp, exp, sizeof(*oexp)); exp->ex_numsecflavors = 0; } Index: projects/clang400-import/sys/kern/vfs_syscalls.c =================================================================== --- projects/clang400-import/sys/kern/vfs_syscalls.c (revision 311459) +++ projects/clang400-import/sys/kern/vfs_syscalls.c (revision 311460) @@ -1,4681 +1,4701 @@ /*- * Copyright (c) 1989, 1993 * The Regents of the University of California. All rights reserved. * (c) UNIX System Laboratories, Inc. * All or some portions of this file are derived from material licensed * to the University of California by American Telephone and Telegraph * Co. or Unix System Laboratories, Inc. and are reproduced herein with * the permission of UNIX System Laboratories, Inc. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 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. * * @(#)vfs_syscalls.c 8.13 (Berkeley) 4/15/94 */ #include __FBSDID("$FreeBSD$"); #include "opt_capsicum.h" #include "opt_compat.h" #include "opt_ktrace.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef KTRACE #include #endif #include #include #include #include #include #include #include #include MALLOC_DEFINE(M_FADVISE, "fadvise", "posix_fadvise(2) information"); SDT_PROVIDER_DEFINE(vfs); SDT_PROBE_DEFINE2(vfs, , stat, mode, "char *", "int"); SDT_PROBE_DEFINE2(vfs, , stat, reg, "char *", "int"); static int kern_chflagsat(struct thread *td, int fd, const char *path, enum uio_seg pathseg, u_long flags, int atflag); static int setfflags(struct thread *td, struct vnode *, u_long); static int getutimes(const struct timeval *, enum uio_seg, struct timespec *); static int getutimens(const struct timespec *, enum uio_seg, struct timespec *, int *); static int setutimes(struct thread *td, struct vnode *, const struct timespec *, int, int); static int vn_access(struct vnode *vp, int user_flags, struct ucred *cred, struct thread *td); /* * Sync each mounted filesystem. */ #ifndef _SYS_SYSPROTO_H_ struct sync_args { int dummy; }; #endif /* ARGSUSED */ int sys_sync(td, uap) struct thread *td; struct sync_args *uap; { struct mount *mp, *nmp; int save; mtx_lock(&mountlist_mtx); for (mp = TAILQ_FIRST(&mountlist); mp != NULL; mp = nmp) { if (vfs_busy(mp, MBF_NOWAIT | MBF_MNTLSTLOCK)) { nmp = TAILQ_NEXT(mp, mnt_list); continue; } if ((mp->mnt_flag & MNT_RDONLY) == 0 && vn_start_write(NULL, &mp, V_NOWAIT) == 0) { save = curthread_pflags_set(TDP_SYNCIO); vfs_msync(mp, MNT_NOWAIT); VFS_SYNC(mp, MNT_NOWAIT); curthread_pflags_restore(save); vn_finished_write(mp); } mtx_lock(&mountlist_mtx); nmp = TAILQ_NEXT(mp, mnt_list); vfs_unbusy(mp); } mtx_unlock(&mountlist_mtx); return (0); } /* * Change filesystem quotas. */ #ifndef _SYS_SYSPROTO_H_ struct quotactl_args { char *path; int cmd; int uid; caddr_t arg; }; #endif int sys_quotactl(td, uap) struct thread *td; register struct quotactl_args /* { char *path; int cmd; int uid; caddr_t arg; } */ *uap; { struct mount *mp; struct nameidata nd; int error; AUDIT_ARG_CMD(uap->cmd); AUDIT_ARG_UID(uap->uid); if (!prison_allow(td->td_ucred, PR_ALLOW_QUOTAS)) return (EPERM); NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF | AUDITVNODE1, UIO_USERSPACE, uap->path, td); if ((error = namei(&nd)) != 0) return (error); NDFREE(&nd, NDF_ONLY_PNBUF); mp = nd.ni_vp->v_mount; vfs_ref(mp); vput(nd.ni_vp); error = vfs_busy(mp, 0); vfs_rel(mp); if (error != 0) return (error); error = VFS_QUOTACTL(mp, uap->cmd, uap->uid, uap->arg); /* * Since quota on operation typically needs to open quota * file, the Q_QUOTAON handler needs to unbusy the mount point * before calling into namei. Otherwise, unmount might be * started between two vfs_busy() invocations (first is our, * second is from mount point cross-walk code in lookup()), * causing deadlock. * * Require that Q_QUOTAON handles the vfs_busy() reference on * its own, always returning with ubusied mount point. */ if ((uap->cmd >> SUBCMDSHIFT) != Q_QUOTAON) vfs_unbusy(mp); return (error); } /* * Used by statfs conversion routines to scale the block size up if * necessary so that all of the block counts are <= 'max_size'. Note * that 'max_size' should be a bitmask, i.e. 2^n - 1 for some non-zero * value of 'n'. */ void statfs_scale_blocks(struct statfs *sf, long max_size) { uint64_t count; int shift; KASSERT(powerof2(max_size + 1), ("%s: invalid max_size", __func__)); /* * Attempt to scale the block counts to give a more accurate * overview to userland of the ratio of free space to used * space. To do this, find the largest block count and compute * a divisor that lets it fit into a signed integer <= max_size. */ if (sf->f_bavail < 0) count = -sf->f_bavail; else count = sf->f_bavail; count = MAX(sf->f_blocks, MAX(sf->f_bfree, count)); if (count <= max_size) return; count >>= flsl(max_size); shift = 0; while (count > 0) { shift++; count >>=1; } sf->f_bsize <<= shift; sf->f_blocks >>= shift; sf->f_bfree >>= shift; sf->f_bavail >>= shift; } static int kern_do_statfs(struct thread *td, struct mount *mp, struct statfs *buf) { struct statfs *sp; int error; if (mp == NULL) return (EBADF); error = vfs_busy(mp, 0); vfs_rel(mp); if (error != 0) return (error); #ifdef MAC error = mac_mount_check_stat(td->td_ucred, mp); if (error != 0) goto out; #endif /* * Set these in case the underlying filesystem fails to do so. */ sp = &mp->mnt_stat; sp->f_version = STATFS_VERSION; sp->f_namemax = NAME_MAX; sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK; error = VFS_STATFS(mp, sp); if (error != 0) goto out; *buf = *sp; if (priv_check(td, PRIV_VFS_GENERATION)) { buf->f_fsid.val[0] = buf->f_fsid.val[1] = 0; prison_enforce_statfs(td->td_ucred, mp, buf); } out: vfs_unbusy(mp); return (error); } /* * Get filesystem statistics. */ #ifndef _SYS_SYSPROTO_H_ struct statfs_args { char *path; struct statfs *buf; }; #endif int sys_statfs(td, uap) struct thread *td; register struct statfs_args /* { char *path; struct statfs *buf; } */ *uap; { - struct statfs sf; + struct statfs *sfp; int error; - error = kern_statfs(td, uap->path, UIO_USERSPACE, &sf); + sfp = malloc(sizeof(struct statfs), M_STATFS, M_WAITOK); + error = kern_statfs(td, uap->path, UIO_USERSPACE, sfp); if (error == 0) - error = copyout(&sf, uap->buf, sizeof(sf)); + error = copyout(sfp, uap->buf, sizeof(struct statfs)); + free(sfp, M_STATFS); return (error); } int kern_statfs(struct thread *td, char *path, enum uio_seg pathseg, struct statfs *buf) { struct mount *mp; struct nameidata nd; int error; NDINIT(&nd, LOOKUP, FOLLOW | LOCKSHARED | LOCKLEAF | AUDITVNODE1, pathseg, path, td); error = namei(&nd); if (error != 0) return (error); mp = nd.ni_vp->v_mount; vfs_ref(mp); NDFREE(&nd, NDF_ONLY_PNBUF); vput(nd.ni_vp); return (kern_do_statfs(td, mp, buf)); } /* * Get filesystem statistics. */ #ifndef _SYS_SYSPROTO_H_ struct fstatfs_args { int fd; struct statfs *buf; }; #endif int sys_fstatfs(td, uap) struct thread *td; register struct fstatfs_args /* { int fd; struct statfs *buf; } */ *uap; { - struct statfs sf; + struct statfs *sfp; int error; - error = kern_fstatfs(td, uap->fd, &sf); + sfp = malloc(sizeof(struct statfs), M_STATFS, M_WAITOK); + error = kern_fstatfs(td, uap->fd, sfp); if (error == 0) - error = copyout(&sf, uap->buf, sizeof(sf)); + error = copyout(sfp, uap->buf, sizeof(struct statfs)); + free(sfp, M_STATFS); return (error); } int kern_fstatfs(struct thread *td, int fd, struct statfs *buf) { struct file *fp; struct mount *mp; struct vnode *vp; cap_rights_t rights; int error; AUDIT_ARG_FD(fd); error = getvnode(td, fd, cap_rights_init(&rights, CAP_FSTATFS), &fp); if (error != 0) return (error); vp = fp->f_vnode; vn_lock(vp, LK_SHARED | LK_RETRY); #ifdef AUDIT AUDIT_ARG_VNODE1(vp); #endif mp = vp->v_mount; if (mp != NULL) vfs_ref(mp); VOP_UNLOCK(vp, 0); fdrop(fp, td); return (kern_do_statfs(td, mp, buf)); } /* * Get statistics on all filesystems. */ #ifndef _SYS_SYSPROTO_H_ struct getfsstat_args { struct statfs *buf; long bufsize; int mode; }; #endif int sys_getfsstat(td, uap) struct thread *td; register struct getfsstat_args /* { struct statfs *buf; long bufsize; int mode; } */ *uap; { size_t count; int error; if (uap->bufsize < 0 || uap->bufsize > SIZE_MAX) return (EINVAL); error = kern_getfsstat(td, &uap->buf, uap->bufsize, &count, UIO_USERSPACE, uap->mode); if (error == 0) td->td_retval[0] = count; return (error); } /* * If (bufsize > 0 && bufseg == UIO_SYSSPACE) * The caller is responsible for freeing memory which will be allocated * in '*buf'. */ int kern_getfsstat(struct thread *td, struct statfs **buf, size_t bufsize, size_t *countp, enum uio_seg bufseg, int mode) { struct mount *mp, *nmp; - struct statfs *sfsp, *sp, sb, *tofree; + struct statfs *sfsp, *sp, *sptmp, *tofree; size_t count, maxcount; int error; switch (mode) { case MNT_WAIT: case MNT_NOWAIT: break; default: if (bufseg == UIO_SYSSPACE) *buf = NULL; return (EINVAL); } restart: maxcount = bufsize / sizeof(struct statfs); if (bufsize == 0) { sfsp = NULL; tofree = NULL; } else if (bufseg == UIO_USERSPACE) { sfsp = *buf; tofree = NULL; } else /* if (bufseg == UIO_SYSSPACE) */ { count = 0; mtx_lock(&mountlist_mtx); TAILQ_FOREACH(mp, &mountlist, mnt_list) { count++; } mtx_unlock(&mountlist_mtx); if (maxcount > count) maxcount = count; tofree = sfsp = *buf = malloc(maxcount * sizeof(struct statfs), - M_TEMP, M_WAITOK); + M_STATFS, M_WAITOK); } count = 0; mtx_lock(&mountlist_mtx); for (mp = TAILQ_FIRST(&mountlist); mp != NULL; mp = nmp) { if (prison_canseemount(td->td_ucred, mp) != 0) { nmp = TAILQ_NEXT(mp, mnt_list); continue; } #ifdef MAC if (mac_mount_check_stat(td->td_ucred, mp) != 0) { nmp = TAILQ_NEXT(mp, mnt_list); continue; } #endif if (mode == MNT_WAIT) { if (vfs_busy(mp, MBF_MNTLSTLOCK) != 0) { /* * If vfs_busy() failed, and MBF_NOWAIT * wasn't passed, then the mp is gone. * Furthermore, because of MBF_MNTLSTLOCK, * the mountlist_mtx was dropped. We have * no other choice than to start over. */ mtx_unlock(&mountlist_mtx); - free(tofree, M_TEMP); + free(tofree, M_STATFS); goto restart; } } else { if (vfs_busy(mp, MBF_NOWAIT | MBF_MNTLSTLOCK) != 0) { nmp = TAILQ_NEXT(mp, mnt_list); continue; } } - if (sfsp && count < maxcount) { + if (sfsp != NULL && count < maxcount) { sp = &mp->mnt_stat; /* * Set these in case the underlying filesystem * fails to do so. */ sp->f_version = STATFS_VERSION; sp->f_namemax = NAME_MAX; sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK; /* * If MNT_NOWAIT is specified, do not refresh * the fsstat cache. */ if (mode != MNT_NOWAIT) { error = VFS_STATFS(mp, sp); if (error != 0) { mtx_lock(&mountlist_mtx); nmp = TAILQ_NEXT(mp, mnt_list); vfs_unbusy(mp); continue; } } if (priv_check(td, PRIV_VFS_GENERATION)) { - bcopy(sp, &sb, sizeof(sb)); - sb.f_fsid.val[0] = sb.f_fsid.val[1] = 0; - prison_enforce_statfs(td->td_ucred, mp, &sb); - sp = &sb; - } - if (bufseg == UIO_SYSSPACE) + sptmp = malloc(sizeof(struct statfs), M_STATFS, + M_WAITOK); + *sptmp = *sp; + sptmp->f_fsid.val[0] = sptmp->f_fsid.val[1] = 0; + prison_enforce_statfs(td->td_ucred, mp, sptmp); + sp = sptmp; + } else + sptmp = NULL; + if (bufseg == UIO_SYSSPACE) { bcopy(sp, sfsp, sizeof(*sp)); - else /* if (bufseg == UIO_USERSPACE) */ { + free(sptmp, M_STATFS); + } else /* if (bufseg == UIO_USERSPACE) */ { error = copyout(sp, sfsp, sizeof(*sp)); + free(sptmp, M_STATFS); if (error != 0) { vfs_unbusy(mp); return (error); } } sfsp++; } count++; mtx_lock(&mountlist_mtx); nmp = TAILQ_NEXT(mp, mnt_list); vfs_unbusy(mp); } mtx_unlock(&mountlist_mtx); - if (sfsp && count > maxcount) + if (sfsp != NULL && count > maxcount) *countp = maxcount; else *countp = count; return (0); } #ifdef COMPAT_FREEBSD4 /* * Get old format filesystem statistics. */ static void cvtstatfs(struct statfs *, struct ostatfs *); #ifndef _SYS_SYSPROTO_H_ struct freebsd4_statfs_args { char *path; struct ostatfs *buf; }; #endif int freebsd4_statfs(td, uap) struct thread *td; struct freebsd4_statfs_args /* { char *path; struct ostatfs *buf; } */ *uap; { struct ostatfs osb; - struct statfs sf; + struct statfs *sfp; int error; - error = kern_statfs(td, uap->path, UIO_USERSPACE, &sf); - if (error != 0) - return (error); - cvtstatfs(&sf, &osb); - return (copyout(&osb, uap->buf, sizeof(osb))); + sfp = malloc(sizeof(struct statfs), M_STATFS, M_WAITOK); + error = kern_statfs(td, uap->path, UIO_USERSPACE, sfp); + if (error == 0) { + cvtstatfs(sfp, &osb); + error = copyout(&osb, uap->buf, sizeof(osb)); + } + free(sfp, M_STATFS); + return (error); } /* * Get filesystem statistics. */ #ifndef _SYS_SYSPROTO_H_ struct freebsd4_fstatfs_args { int fd; struct ostatfs *buf; }; #endif int freebsd4_fstatfs(td, uap) struct thread *td; struct freebsd4_fstatfs_args /* { int fd; struct ostatfs *buf; } */ *uap; { struct ostatfs osb; - struct statfs sf; + struct statfs *sfp; int error; - error = kern_fstatfs(td, uap->fd, &sf); - if (error != 0) - return (error); - cvtstatfs(&sf, &osb); - return (copyout(&osb, uap->buf, sizeof(osb))); + sfp = malloc(sizeof(struct statfs), M_STATFS, M_WAITOK); + error = kern_fstatfs(td, uap->fd, sfp); + if (error == 0) { + cvtstatfs(sfp, &osb); + error = copyout(&osb, uap->buf, sizeof(osb)); + } + free(sfp, M_STATFS); + return (error); } /* * Get statistics on all filesystems. */ #ifndef _SYS_SYSPROTO_H_ struct freebsd4_getfsstat_args { struct ostatfs *buf; long bufsize; int mode; }; #endif int freebsd4_getfsstat(td, uap) struct thread *td; register struct freebsd4_getfsstat_args /* { struct ostatfs *buf; long bufsize; int mode; } */ *uap; { struct statfs *buf, *sp; struct ostatfs osb; size_t count, size; int error; if (uap->bufsize < 0) return (EINVAL); count = uap->bufsize / sizeof(struct ostatfs); if (count > SIZE_MAX / sizeof(struct statfs)) return (EINVAL); size = count * sizeof(struct statfs); error = kern_getfsstat(td, &buf, size, &count, UIO_SYSSPACE, uap->mode); td->td_retval[0] = count; if (size != 0) { sp = buf; while (count != 0 && error == 0) { cvtstatfs(sp, &osb); error = copyout(&osb, uap->buf, sizeof(osb)); sp++; uap->buf++; count--; } - free(buf, M_TEMP); + free(buf, M_STATFS); } return (error); } /* * Implement fstatfs() for (NFS) file handles. */ #ifndef _SYS_SYSPROTO_H_ struct freebsd4_fhstatfs_args { struct fhandle *u_fhp; struct ostatfs *buf; }; #endif int freebsd4_fhstatfs(td, uap) struct thread *td; struct freebsd4_fhstatfs_args /* { struct fhandle *u_fhp; struct ostatfs *buf; } */ *uap; { struct ostatfs osb; - struct statfs sf; + struct statfs *sfp; fhandle_t fh; int error; error = copyin(uap->u_fhp, &fh, sizeof(fhandle_t)); if (error != 0) return (error); - error = kern_fhstatfs(td, fh, &sf); - if (error != 0) - return (error); - cvtstatfs(&sf, &osb); - return (copyout(&osb, uap->buf, sizeof(osb))); + sfp = malloc(sizeof(struct statfs), M_STATFS, M_WAITOK); + error = kern_fhstatfs(td, fh, sfp); + if (error == 0) { + cvtstatfs(sfp, &osb); + error = copyout(&osb, uap->buf, sizeof(osb)); + } + free(sfp, M_STATFS); + return (error); } /* * Convert a new format statfs structure to an old format statfs structure. */ static void cvtstatfs(nsp, osp) struct statfs *nsp; struct ostatfs *osp; { statfs_scale_blocks(nsp, LONG_MAX); bzero(osp, sizeof(*osp)); osp->f_bsize = nsp->f_bsize; osp->f_iosize = MIN(nsp->f_iosize, LONG_MAX); osp->f_blocks = nsp->f_blocks; osp->f_bfree = nsp->f_bfree; osp->f_bavail = nsp->f_bavail; osp->f_files = MIN(nsp->f_files, LONG_MAX); osp->f_ffree = MIN(nsp->f_ffree, LONG_MAX); osp->f_owner = nsp->f_owner; osp->f_type = nsp->f_type; osp->f_flags = nsp->f_flags; osp->f_syncwrites = MIN(nsp->f_syncwrites, LONG_MAX); osp->f_asyncwrites = MIN(nsp->f_asyncwrites, LONG_MAX); osp->f_syncreads = MIN(nsp->f_syncreads, LONG_MAX); osp->f_asyncreads = MIN(nsp->f_asyncreads, LONG_MAX); strlcpy(osp->f_fstypename, nsp->f_fstypename, MIN(MFSNAMELEN, OMFSNAMELEN)); strlcpy(osp->f_mntonname, nsp->f_mntonname, MIN(MNAMELEN, OMNAMELEN)); strlcpy(osp->f_mntfromname, nsp->f_mntfromname, MIN(MNAMELEN, OMNAMELEN)); osp->f_fsid = nsp->f_fsid; } #endif /* COMPAT_FREEBSD4 */ /* * Change current working directory to a given file descriptor. */ #ifndef _SYS_SYSPROTO_H_ struct fchdir_args { int fd; }; #endif int sys_fchdir(td, uap) struct thread *td; struct fchdir_args /* { int fd; } */ *uap; { struct vnode *vp, *tdp; struct mount *mp; struct file *fp; cap_rights_t rights; int error; AUDIT_ARG_FD(uap->fd); error = getvnode(td, uap->fd, cap_rights_init(&rights, CAP_FCHDIR), &fp); if (error != 0) return (error); vp = fp->f_vnode; vrefact(vp); fdrop(fp, td); vn_lock(vp, LK_SHARED | LK_RETRY); AUDIT_ARG_VNODE1(vp); error = change_dir(vp, td); while (!error && (mp = vp->v_mountedhere) != NULL) { if (vfs_busy(mp, 0)) continue; error = VFS_ROOT(mp, LK_SHARED, &tdp); vfs_unbusy(mp); if (error != 0) break; vput(vp); vp = tdp; } if (error != 0) { vput(vp); return (error); } VOP_UNLOCK(vp, 0); pwd_chdir(td, vp); return (0); } /* * Change current working directory (``.''). */ #ifndef _SYS_SYSPROTO_H_ struct chdir_args { char *path; }; #endif int sys_chdir(td, uap) struct thread *td; struct chdir_args /* { char *path; } */ *uap; { return (kern_chdir(td, uap->path, UIO_USERSPACE)); } int kern_chdir(struct thread *td, char *path, enum uio_seg pathseg) { struct nameidata nd; int error; NDINIT(&nd, LOOKUP, FOLLOW | LOCKSHARED | LOCKLEAF | AUDITVNODE1, pathseg, path, td); if ((error = namei(&nd)) != 0) return (error); if ((error = change_dir(nd.ni_vp, td)) != 0) { vput(nd.ni_vp); NDFREE(&nd, NDF_ONLY_PNBUF); return (error); } VOP_UNLOCK(nd.ni_vp, 0); NDFREE(&nd, NDF_ONLY_PNBUF); pwd_chdir(td, nd.ni_vp); return (0); } /* * Change notion of root (``/'') directory. */ #ifndef _SYS_SYSPROTO_H_ struct chroot_args { char *path; }; #endif int sys_chroot(td, uap) struct thread *td; struct chroot_args /* { char *path; } */ *uap; { struct nameidata nd; int error; error = priv_check(td, PRIV_VFS_CHROOT); if (error != 0) return (error); NDINIT(&nd, LOOKUP, FOLLOW | LOCKSHARED | LOCKLEAF | AUDITVNODE1, UIO_USERSPACE, uap->path, td); error = namei(&nd); if (error != 0) goto error; error = change_dir(nd.ni_vp, td); if (error != 0) goto e_vunlock; #ifdef MAC error = mac_vnode_check_chroot(td->td_ucred, nd.ni_vp); if (error != 0) goto e_vunlock; #endif VOP_UNLOCK(nd.ni_vp, 0); error = pwd_chroot(td, nd.ni_vp); vrele(nd.ni_vp); NDFREE(&nd, NDF_ONLY_PNBUF); return (error); e_vunlock: vput(nd.ni_vp); error: NDFREE(&nd, NDF_ONLY_PNBUF); return (error); } /* * Common routine for chroot and chdir. Callers must provide a locked vnode * instance. */ int change_dir(vp, td) struct vnode *vp; struct thread *td; { #ifdef MAC int error; #endif ASSERT_VOP_LOCKED(vp, "change_dir(): vp not locked"); if (vp->v_type != VDIR) return (ENOTDIR); #ifdef MAC error = mac_vnode_check_chdir(td->td_ucred, vp); if (error != 0) return (error); #endif return (VOP_ACCESS(vp, VEXEC, td->td_ucred, td)); } static __inline void flags_to_rights(int flags, cap_rights_t *rightsp) { if (flags & O_EXEC) { cap_rights_set(rightsp, CAP_FEXECVE); } else { switch ((flags & O_ACCMODE)) { case O_RDONLY: cap_rights_set(rightsp, CAP_READ); break; case O_RDWR: cap_rights_set(rightsp, CAP_READ); /* FALLTHROUGH */ case O_WRONLY: cap_rights_set(rightsp, CAP_WRITE); if (!(flags & (O_APPEND | O_TRUNC))) cap_rights_set(rightsp, CAP_SEEK); break; } } if (flags & O_CREAT) cap_rights_set(rightsp, CAP_CREATE); if (flags & O_TRUNC) cap_rights_set(rightsp, CAP_FTRUNCATE); if (flags & (O_SYNC | O_FSYNC)) cap_rights_set(rightsp, CAP_FSYNC); if (flags & (O_EXLOCK | O_SHLOCK)) cap_rights_set(rightsp, CAP_FLOCK); } /* * Check permissions, allocate an open file structure, and call the device * open routine if any. */ #ifndef _SYS_SYSPROTO_H_ struct open_args { char *path; int flags; int mode; }; #endif int sys_open(td, uap) struct thread *td; register struct open_args /* { char *path; int flags; int mode; } */ *uap; { return (kern_openat(td, AT_FDCWD, uap->path, UIO_USERSPACE, uap->flags, uap->mode)); } #ifndef _SYS_SYSPROTO_H_ struct openat_args { int fd; char *path; int flag; int mode; }; #endif int sys_openat(struct thread *td, struct openat_args *uap) { AUDIT_ARG_FD(uap->fd); return (kern_openat(td, uap->fd, uap->path, UIO_USERSPACE, uap->flag, uap->mode)); } int kern_openat(struct thread *td, int fd, char *path, enum uio_seg pathseg, int flags, int mode) { struct proc *p = td->td_proc; struct filedesc *fdp = p->p_fd; struct file *fp; struct vnode *vp; struct nameidata nd; cap_rights_t rights; int cmode, error, indx; indx = -1; AUDIT_ARG_FFLAGS(flags); AUDIT_ARG_MODE(mode); cap_rights_init(&rights, CAP_LOOKUP); flags_to_rights(flags, &rights); /* * Only one of the O_EXEC, O_RDONLY, O_WRONLY and O_RDWR flags * may be specified. */ if (flags & O_EXEC) { if (flags & O_ACCMODE) return (EINVAL); } else if ((flags & O_ACCMODE) == O_ACCMODE) { return (EINVAL); } else { flags = FFLAGS(flags); } /* * Allocate a file structure. The descriptor to reference it * is allocated and set by finstall() below. */ error = falloc_noinstall(td, &fp); if (error != 0) return (error); /* * An extra reference on `fp' has been held for us by * falloc_noinstall(). */ /* Set the flags early so the finit in devfs can pick them up. */ fp->f_flag = flags & FMASK; cmode = ((mode & ~fdp->fd_cmask) & ALLPERMS) & ~S_ISTXT; NDINIT_ATRIGHTS(&nd, LOOKUP, FOLLOW | AUDITVNODE1, pathseg, path, fd, &rights, td); td->td_dupfd = -1; /* XXX check for fdopen */ error = vn_open(&nd, &flags, cmode, fp); if (error != 0) { /* * If the vn_open replaced the method vector, something * wonderous happened deep below and we just pass it up * pretending we know what we do. */ if (error == ENXIO && fp->f_ops != &badfileops) goto success; /* * Handle special fdopen() case. bleh. * * Don't do this for relative (capability) lookups; we don't * understand exactly what would happen, and we don't think * that it ever should. */ if ((nd.ni_lcf & NI_LCF_STRICTRELATIVE) == 0 && (error == ENODEV || error == ENXIO) && td->td_dupfd >= 0) { error = dupfdopen(td, fdp, td->td_dupfd, flags, error, &indx); if (error == 0) goto success; } goto bad; } td->td_dupfd = 0; NDFREE(&nd, NDF_ONLY_PNBUF); vp = nd.ni_vp; /* * Store the vnode, for any f_type. Typically, the vnode use * count is decremented by direct call to vn_closefile() for * files that switched type in the cdevsw fdopen() method. */ fp->f_vnode = vp; /* * If the file wasn't claimed by devfs bind it to the normal * vnode operations here. */ if (fp->f_ops == &badfileops) { KASSERT(vp->v_type != VFIFO, ("Unexpected fifo.")); fp->f_seqcount = 1; finit(fp, (flags & FMASK) | (fp->f_flag & FHASLOCK), DTYPE_VNODE, vp, &vnops); } VOP_UNLOCK(vp, 0); if (flags & O_TRUNC) { error = fo_truncate(fp, 0, td->td_ucred, td); if (error != 0) goto bad; } success: /* * If we haven't already installed the FD (for dupfdopen), do so now. */ if (indx == -1) { struct filecaps *fcaps; #ifdef CAPABILITIES if ((nd.ni_lcf & NI_LCF_STRICTRELATIVE) != 0) fcaps = &nd.ni_filecaps; else #endif fcaps = NULL; error = finstall(td, fp, &indx, flags, fcaps); /* On success finstall() consumes fcaps. */ if (error != 0) { filecaps_free(&nd.ni_filecaps); goto bad; } } else { filecaps_free(&nd.ni_filecaps); } /* * Release our private reference, leaving the one associated with * the descriptor table intact. */ fdrop(fp, td); td->td_retval[0] = indx; return (0); bad: KASSERT(indx == -1, ("indx=%d, should be -1", indx)); fdrop(fp, td); return (error); } #ifdef COMPAT_43 /* * Create a file. */ #ifndef _SYS_SYSPROTO_H_ struct ocreat_args { char *path; int mode; }; #endif int ocreat(td, uap) struct thread *td; register struct ocreat_args /* { char *path; int mode; } */ *uap; { return (kern_openat(td, AT_FDCWD, uap->path, UIO_USERSPACE, O_WRONLY | O_CREAT | O_TRUNC, uap->mode)); } #endif /* COMPAT_43 */ /* * Create a special file. */ #ifndef _SYS_SYSPROTO_H_ struct mknod_args { char *path; int mode; int dev; }; #endif int sys_mknod(td, uap) struct thread *td; register struct mknod_args /* { char *path; int mode; int dev; } */ *uap; { return (kern_mknodat(td, AT_FDCWD, uap->path, UIO_USERSPACE, uap->mode, uap->dev)); } #ifndef _SYS_SYSPROTO_H_ struct mknodat_args { int fd; char *path; mode_t mode; dev_t dev; }; #endif int sys_mknodat(struct thread *td, struct mknodat_args *uap) { return (kern_mknodat(td, uap->fd, uap->path, UIO_USERSPACE, uap->mode, uap->dev)); } int kern_mknodat(struct thread *td, int fd, char *path, enum uio_seg pathseg, int mode, int dev) { struct vnode *vp; struct mount *mp; struct vattr vattr; struct nameidata nd; cap_rights_t rights; int error, whiteout = 0; AUDIT_ARG_MODE(mode); AUDIT_ARG_DEV(dev); switch (mode & S_IFMT) { case S_IFCHR: case S_IFBLK: error = priv_check(td, PRIV_VFS_MKNOD_DEV); if (error == 0 && dev == VNOVAL) error = EINVAL; break; case S_IFMT: error = priv_check(td, PRIV_VFS_MKNOD_BAD); break; case S_IFWHT: error = priv_check(td, PRIV_VFS_MKNOD_WHT); break; case S_IFIFO: if (dev == 0) return (kern_mkfifoat(td, fd, path, pathseg, mode)); /* FALLTHROUGH */ default: error = EINVAL; break; } if (error != 0) return (error); restart: bwillwrite(); NDINIT_ATRIGHTS(&nd, CREATE, LOCKPARENT | SAVENAME | AUDITVNODE1 | NOCACHE, pathseg, path, fd, cap_rights_init(&rights, CAP_MKNODAT), td); if ((error = namei(&nd)) != 0) return (error); vp = nd.ni_vp; if (vp != NULL) { NDFREE(&nd, NDF_ONLY_PNBUF); if (vp == nd.ni_dvp) vrele(nd.ni_dvp); else vput(nd.ni_dvp); vrele(vp); return (EEXIST); } else { VATTR_NULL(&vattr); vattr.va_mode = (mode & ALLPERMS) & ~td->td_proc->p_fd->fd_cmask; vattr.va_rdev = dev; whiteout = 0; switch (mode & S_IFMT) { case S_IFMT: /* used by badsect to flag bad sectors */ vattr.va_type = VBAD; break; case S_IFCHR: vattr.va_type = VCHR; break; case S_IFBLK: vattr.va_type = VBLK; break; case S_IFWHT: whiteout = 1; break; default: panic("kern_mknod: invalid mode"); } } if (vn_start_write(nd.ni_dvp, &mp, V_NOWAIT) != 0) { NDFREE(&nd, NDF_ONLY_PNBUF); vput(nd.ni_dvp); if ((error = vn_start_write(NULL, &mp, V_XSLEEP | PCATCH)) != 0) return (error); goto restart; } #ifdef MAC if (error == 0 && !whiteout) error = mac_vnode_check_create(td->td_ucred, nd.ni_dvp, &nd.ni_cnd, &vattr); #endif if (error == 0) { if (whiteout) error = VOP_WHITEOUT(nd.ni_dvp, &nd.ni_cnd, CREATE); else { error = VOP_MKNOD(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr); if (error == 0) vput(nd.ni_vp); } } NDFREE(&nd, NDF_ONLY_PNBUF); vput(nd.ni_dvp); vn_finished_write(mp); return (error); } /* * Create a named pipe. */ #ifndef _SYS_SYSPROTO_H_ struct mkfifo_args { char *path; int mode; }; #endif int sys_mkfifo(td, uap) struct thread *td; register struct mkfifo_args /* { char *path; int mode; } */ *uap; { return (kern_mkfifoat(td, AT_FDCWD, uap->path, UIO_USERSPACE, uap->mode)); } #ifndef _SYS_SYSPROTO_H_ struct mkfifoat_args { int fd; char *path; mode_t mode; }; #endif int sys_mkfifoat(struct thread *td, struct mkfifoat_args *uap) { return (kern_mkfifoat(td, uap->fd, uap->path, UIO_USERSPACE, uap->mode)); } int kern_mkfifoat(struct thread *td, int fd, char *path, enum uio_seg pathseg, int mode) { struct mount *mp; struct vattr vattr; struct nameidata nd; cap_rights_t rights; int error; AUDIT_ARG_MODE(mode); restart: bwillwrite(); NDINIT_ATRIGHTS(&nd, CREATE, LOCKPARENT | SAVENAME | AUDITVNODE1 | NOCACHE, pathseg, path, fd, cap_rights_init(&rights, CAP_MKFIFOAT), td); if ((error = namei(&nd)) != 0) return (error); if (nd.ni_vp != NULL) { NDFREE(&nd, NDF_ONLY_PNBUF); if (nd.ni_vp == nd.ni_dvp) vrele(nd.ni_dvp); else vput(nd.ni_dvp); vrele(nd.ni_vp); return (EEXIST); } if (vn_start_write(nd.ni_dvp, &mp, V_NOWAIT) != 0) { NDFREE(&nd, NDF_ONLY_PNBUF); vput(nd.ni_dvp); if ((error = vn_start_write(NULL, &mp, V_XSLEEP | PCATCH)) != 0) return (error); goto restart; } VATTR_NULL(&vattr); vattr.va_type = VFIFO; vattr.va_mode = (mode & ALLPERMS) & ~td->td_proc->p_fd->fd_cmask; #ifdef MAC error = mac_vnode_check_create(td->td_ucred, nd.ni_dvp, &nd.ni_cnd, &vattr); if (error != 0) goto out; #endif error = VOP_MKNOD(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr); if (error == 0) vput(nd.ni_vp); #ifdef MAC out: #endif vput(nd.ni_dvp); vn_finished_write(mp); NDFREE(&nd, NDF_ONLY_PNBUF); return (error); } /* * Make a hard file link. */ #ifndef _SYS_SYSPROTO_H_ struct link_args { char *path; char *link; }; #endif int sys_link(td, uap) struct thread *td; register struct link_args /* { char *path; char *link; } */ *uap; { return (kern_linkat(td, AT_FDCWD, AT_FDCWD, uap->path, uap->link, UIO_USERSPACE, FOLLOW)); } #ifndef _SYS_SYSPROTO_H_ struct linkat_args { int fd1; char *path1; int fd2; char *path2; int flag; }; #endif int sys_linkat(struct thread *td, struct linkat_args *uap) { int flag; flag = uap->flag; if (flag & ~AT_SYMLINK_FOLLOW) return (EINVAL); return (kern_linkat(td, uap->fd1, uap->fd2, uap->path1, uap->path2, UIO_USERSPACE, (flag & AT_SYMLINK_FOLLOW) ? FOLLOW : NOFOLLOW)); } int hardlink_check_uid = 0; SYSCTL_INT(_security_bsd, OID_AUTO, hardlink_check_uid, CTLFLAG_RW, &hardlink_check_uid, 0, "Unprivileged processes cannot create hard links to files owned by other " "users"); static int hardlink_check_gid = 0; SYSCTL_INT(_security_bsd, OID_AUTO, hardlink_check_gid, CTLFLAG_RW, &hardlink_check_gid, 0, "Unprivileged processes cannot create hard links to files owned by other " "groups"); static int can_hardlink(struct vnode *vp, struct ucred *cred) { struct vattr va; int error; if (!hardlink_check_uid && !hardlink_check_gid) return (0); error = VOP_GETATTR(vp, &va, cred); if (error != 0) return (error); if (hardlink_check_uid && cred->cr_uid != va.va_uid) { error = priv_check_cred(cred, PRIV_VFS_LINK, 0); if (error != 0) return (error); } if (hardlink_check_gid && !groupmember(va.va_gid, cred)) { error = priv_check_cred(cred, PRIV_VFS_LINK, 0); if (error != 0) return (error); } return (0); } int kern_linkat(struct thread *td, int fd1, int fd2, char *path1, char *path2, enum uio_seg segflg, int follow) { struct vnode *vp; struct mount *mp; struct nameidata nd; cap_rights_t rights; int error; again: bwillwrite(); NDINIT_ATRIGHTS(&nd, LOOKUP, follow | AUDITVNODE1, segflg, path1, fd1, cap_rights_init(&rights, CAP_LINKAT_SOURCE), td); if ((error = namei(&nd)) != 0) return (error); NDFREE(&nd, NDF_ONLY_PNBUF); vp = nd.ni_vp; if (vp->v_type == VDIR) { vrele(vp); return (EPERM); /* POSIX */ } NDINIT_ATRIGHTS(&nd, CREATE, LOCKPARENT | SAVENAME | AUDITVNODE2 | NOCACHE, segflg, path2, fd2, cap_rights_init(&rights, CAP_LINKAT_TARGET), td); if ((error = namei(&nd)) == 0) { if (nd.ni_vp != NULL) { NDFREE(&nd, NDF_ONLY_PNBUF); if (nd.ni_dvp == nd.ni_vp) vrele(nd.ni_dvp); else vput(nd.ni_dvp); vrele(nd.ni_vp); vrele(vp); return (EEXIST); } else if (nd.ni_dvp->v_mount != vp->v_mount) { /* * Cross-device link. No need to recheck * vp->v_type, since it cannot change, except * to VBAD. */ NDFREE(&nd, NDF_ONLY_PNBUF); vput(nd.ni_dvp); vrele(vp); return (EXDEV); } else if ((error = vn_lock(vp, LK_EXCLUSIVE)) == 0) { error = can_hardlink(vp, td->td_ucred); #ifdef MAC if (error == 0) error = mac_vnode_check_link(td->td_ucred, nd.ni_dvp, vp, &nd.ni_cnd); #endif if (error != 0) { vput(vp); vput(nd.ni_dvp); NDFREE(&nd, NDF_ONLY_PNBUF); return (error); } error = vn_start_write(vp, &mp, V_NOWAIT); if (error != 0) { vput(vp); vput(nd.ni_dvp); NDFREE(&nd, NDF_ONLY_PNBUF); error = vn_start_write(NULL, &mp, V_XSLEEP | PCATCH); if (error != 0) return (error); goto again; } error = VOP_LINK(nd.ni_dvp, vp, &nd.ni_cnd); VOP_UNLOCK(vp, 0); vput(nd.ni_dvp); vn_finished_write(mp); NDFREE(&nd, NDF_ONLY_PNBUF); } else { vput(nd.ni_dvp); NDFREE(&nd, NDF_ONLY_PNBUF); vrele(vp); goto again; } } vrele(vp); return (error); } /* * Make a symbolic link. */ #ifndef _SYS_SYSPROTO_H_ struct symlink_args { char *path; char *link; }; #endif int sys_symlink(td, uap) struct thread *td; register struct symlink_args /* { char *path; char *link; } */ *uap; { return (kern_symlinkat(td, uap->path, AT_FDCWD, uap->link, UIO_USERSPACE)); } #ifndef _SYS_SYSPROTO_H_ struct symlinkat_args { char *path; int fd; char *path2; }; #endif int sys_symlinkat(struct thread *td, struct symlinkat_args *uap) { return (kern_symlinkat(td, uap->path1, uap->fd, uap->path2, UIO_USERSPACE)); } int kern_symlinkat(struct thread *td, char *path1, int fd, char *path2, enum uio_seg segflg) { struct mount *mp; struct vattr vattr; char *syspath; struct nameidata nd; int error; cap_rights_t rights; if (segflg == UIO_SYSSPACE) { syspath = path1; } else { syspath = uma_zalloc(namei_zone, M_WAITOK); if ((error = copyinstr(path1, syspath, MAXPATHLEN, NULL)) != 0) goto out; } AUDIT_ARG_TEXT(syspath); restart: bwillwrite(); NDINIT_ATRIGHTS(&nd, CREATE, LOCKPARENT | SAVENAME | AUDITVNODE1 | NOCACHE, segflg, path2, fd, cap_rights_init(&rights, CAP_SYMLINKAT), td); if ((error = namei(&nd)) != 0) goto out; if (nd.ni_vp) { NDFREE(&nd, NDF_ONLY_PNBUF); if (nd.ni_vp == nd.ni_dvp) vrele(nd.ni_dvp); else vput(nd.ni_dvp); vrele(nd.ni_vp); error = EEXIST; goto out; } if (vn_start_write(nd.ni_dvp, &mp, V_NOWAIT) != 0) { NDFREE(&nd, NDF_ONLY_PNBUF); vput(nd.ni_dvp); if ((error = vn_start_write(NULL, &mp, V_XSLEEP | PCATCH)) != 0) goto out; goto restart; } VATTR_NULL(&vattr); vattr.va_mode = ACCESSPERMS &~ td->td_proc->p_fd->fd_cmask; #ifdef MAC vattr.va_type = VLNK; error = mac_vnode_check_create(td->td_ucred, nd.ni_dvp, &nd.ni_cnd, &vattr); if (error != 0) goto out2; #endif error = VOP_SYMLINK(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr, syspath); if (error == 0) vput(nd.ni_vp); #ifdef MAC out2: #endif NDFREE(&nd, NDF_ONLY_PNBUF); vput(nd.ni_dvp); vn_finished_write(mp); out: if (segflg != UIO_SYSSPACE) uma_zfree(namei_zone, syspath); return (error); } /* * Delete a whiteout from the filesystem. */ int sys_undelete(td, uap) struct thread *td; register struct undelete_args /* { char *path; } */ *uap; { struct mount *mp; struct nameidata nd; int error; restart: bwillwrite(); NDINIT(&nd, DELETE, LOCKPARENT | DOWHITEOUT | AUDITVNODE1, UIO_USERSPACE, uap->path, td); error = namei(&nd); if (error != 0) return (error); if (nd.ni_vp != NULLVP || !(nd.ni_cnd.cn_flags & ISWHITEOUT)) { NDFREE(&nd, NDF_ONLY_PNBUF); if (nd.ni_vp == nd.ni_dvp) vrele(nd.ni_dvp); else vput(nd.ni_dvp); if (nd.ni_vp) vrele(nd.ni_vp); return (EEXIST); } if (vn_start_write(nd.ni_dvp, &mp, V_NOWAIT) != 0) { NDFREE(&nd, NDF_ONLY_PNBUF); vput(nd.ni_dvp); if ((error = vn_start_write(NULL, &mp, V_XSLEEP | PCATCH)) != 0) return (error); goto restart; } error = VOP_WHITEOUT(nd.ni_dvp, &nd.ni_cnd, DELETE); NDFREE(&nd, NDF_ONLY_PNBUF); vput(nd.ni_dvp); vn_finished_write(mp); return (error); } /* * Delete a name from the filesystem. */ #ifndef _SYS_SYSPROTO_H_ struct unlink_args { char *path; }; #endif int sys_unlink(td, uap) struct thread *td; struct unlink_args /* { char *path; } */ *uap; { return (kern_unlinkat(td, AT_FDCWD, uap->path, UIO_USERSPACE, 0)); } #ifndef _SYS_SYSPROTO_H_ struct unlinkat_args { int fd; char *path; int flag; }; #endif int sys_unlinkat(struct thread *td, struct unlinkat_args *uap) { int flag = uap->flag; int fd = uap->fd; char *path = uap->path; if (flag & ~AT_REMOVEDIR) return (EINVAL); if (flag & AT_REMOVEDIR) return (kern_rmdirat(td, fd, path, UIO_USERSPACE)); else return (kern_unlinkat(td, fd, path, UIO_USERSPACE, 0)); } int kern_unlinkat(struct thread *td, int fd, char *path, enum uio_seg pathseg, ino_t oldinum) { struct mount *mp; struct vnode *vp; struct nameidata nd; struct stat sb; cap_rights_t rights; int error; restart: bwillwrite(); NDINIT_ATRIGHTS(&nd, DELETE, LOCKPARENT | LOCKLEAF | AUDITVNODE1, pathseg, path, fd, cap_rights_init(&rights, CAP_UNLINKAT), td); if ((error = namei(&nd)) != 0) return (error == EINVAL ? EPERM : error); vp = nd.ni_vp; if (vp->v_type == VDIR && oldinum == 0) { error = EPERM; /* POSIX */ } else if (oldinum != 0 && ((error = vn_stat(vp, &sb, td->td_ucred, NOCRED, td)) == 0) && sb.st_ino != oldinum) { error = EIDRM; /* Identifier removed */ } else { /* * The root of a mounted filesystem cannot be deleted. * * XXX: can this only be a VDIR case? */ if (vp->v_vflag & VV_ROOT) error = EBUSY; } if (error == 0) { if (vn_start_write(nd.ni_dvp, &mp, V_NOWAIT) != 0) { NDFREE(&nd, NDF_ONLY_PNBUF); vput(nd.ni_dvp); if (vp == nd.ni_dvp) vrele(vp); else vput(vp); if ((error = vn_start_write(NULL, &mp, V_XSLEEP | PCATCH)) != 0) return (error); goto restart; } #ifdef MAC error = mac_vnode_check_unlink(td->td_ucred, nd.ni_dvp, vp, &nd.ni_cnd); if (error != 0) goto out; #endif vfs_notify_upper(vp, VFS_NOTIFY_UPPER_UNLINK); error = VOP_REMOVE(nd.ni_dvp, vp, &nd.ni_cnd); #ifdef MAC out: #endif vn_finished_write(mp); } NDFREE(&nd, NDF_ONLY_PNBUF); vput(nd.ni_dvp); if (vp == nd.ni_dvp) vrele(vp); else vput(vp); return (error); } /* * Reposition read/write file offset. */ #ifndef _SYS_SYSPROTO_H_ struct lseek_args { int fd; int pad; off_t offset; int whence; }; #endif int sys_lseek(td, uap) struct thread *td; register struct lseek_args /* { int fd; int pad; off_t offset; int whence; } */ *uap; { struct file *fp; cap_rights_t rights; int error; AUDIT_ARG_FD(uap->fd); error = fget(td, uap->fd, cap_rights_init(&rights, CAP_SEEK), &fp); if (error != 0) return (error); error = (fp->f_ops->fo_flags & DFLAG_SEEKABLE) != 0 ? fo_seek(fp, uap->offset, uap->whence, td) : ESPIPE; fdrop(fp, td); return (error); } #if defined(COMPAT_43) /* * Reposition read/write file offset. */ #ifndef _SYS_SYSPROTO_H_ struct olseek_args { int fd; long offset; int whence; }; #endif int olseek(td, uap) struct thread *td; register struct olseek_args /* { int fd; long offset; int whence; } */ *uap; { struct lseek_args /* { int fd; int pad; off_t offset; int whence; } */ nuap; nuap.fd = uap->fd; nuap.offset = uap->offset; nuap.whence = uap->whence; return (sys_lseek(td, &nuap)); } #endif /* COMPAT_43 */ #if defined(COMPAT_FREEBSD6) /* Version with the 'pad' argument */ int freebsd6_lseek(td, uap) struct thread *td; register struct freebsd6_lseek_args *uap; { struct lseek_args ouap; ouap.fd = uap->fd; ouap.offset = uap->offset; ouap.whence = uap->whence; return (sys_lseek(td, &ouap)); } #endif /* * Check access permissions using passed credentials. */ static int vn_access(vp, user_flags, cred, td) struct vnode *vp; int user_flags; struct ucred *cred; struct thread *td; { accmode_t accmode; int error; /* Flags == 0 means only check for existence. */ if (user_flags == 0) return (0); accmode = 0; if (user_flags & R_OK) accmode |= VREAD; if (user_flags & W_OK) accmode |= VWRITE; if (user_flags & X_OK) accmode |= VEXEC; #ifdef MAC error = mac_vnode_check_access(cred, vp, accmode); if (error != 0) return (error); #endif if ((accmode & VWRITE) == 0 || (error = vn_writechk(vp)) == 0) error = VOP_ACCESS(vp, accmode, cred, td); return (error); } /* * Check access permissions using "real" credentials. */ #ifndef _SYS_SYSPROTO_H_ struct access_args { char *path; int amode; }; #endif int sys_access(td, uap) struct thread *td; register struct access_args /* { char *path; int amode; } */ *uap; { return (kern_accessat(td, AT_FDCWD, uap->path, UIO_USERSPACE, 0, uap->amode)); } #ifndef _SYS_SYSPROTO_H_ struct faccessat_args { int dirfd; char *path; int amode; int flag; } #endif int sys_faccessat(struct thread *td, struct faccessat_args *uap) { return (kern_accessat(td, uap->fd, uap->path, UIO_USERSPACE, uap->flag, uap->amode)); } int kern_accessat(struct thread *td, int fd, char *path, enum uio_seg pathseg, int flag, int amode) { struct ucred *cred, *usecred; struct vnode *vp; struct nameidata nd; cap_rights_t rights; int error; if (flag & ~AT_EACCESS) return (EINVAL); if (amode != F_OK && (amode & ~(R_OK | W_OK | X_OK)) != 0) return (EINVAL); /* * Create and modify a temporary credential instead of one that * is potentially shared (if we need one). */ cred = td->td_ucred; if ((flag & AT_EACCESS) == 0 && ((cred->cr_uid != cred->cr_ruid || cred->cr_rgid != cred->cr_groups[0]))) { usecred = crdup(cred); usecred->cr_uid = cred->cr_ruid; usecred->cr_groups[0] = cred->cr_rgid; td->td_ucred = usecred; } else usecred = cred; AUDIT_ARG_VALUE(amode); NDINIT_ATRIGHTS(&nd, LOOKUP, FOLLOW | LOCKSHARED | LOCKLEAF | AUDITVNODE1, pathseg, path, fd, cap_rights_init(&rights, CAP_FSTAT), td); if ((error = namei(&nd)) != 0) goto out; vp = nd.ni_vp; error = vn_access(vp, amode, usecred, td); NDFREE(&nd, NDF_ONLY_PNBUF); vput(vp); out: if (usecred != cred) { td->td_ucred = cred; crfree(usecred); } return (error); } /* * Check access permissions using "effective" credentials. */ #ifndef _SYS_SYSPROTO_H_ struct eaccess_args { char *path; int amode; }; #endif int sys_eaccess(td, uap) struct thread *td; register struct eaccess_args /* { char *path; int amode; } */ *uap; { return (kern_accessat(td, AT_FDCWD, uap->path, UIO_USERSPACE, AT_EACCESS, uap->amode)); } #if defined(COMPAT_43) /* * Get file status; this version follows links. */ #ifndef _SYS_SYSPROTO_H_ struct ostat_args { char *path; struct ostat *ub; }; #endif int ostat(td, uap) struct thread *td; register struct ostat_args /* { char *path; struct ostat *ub; } */ *uap; { struct stat sb; struct ostat osb; int error; error = kern_statat(td, 0, AT_FDCWD, uap->path, UIO_USERSPACE, &sb, NULL); if (error != 0) return (error); cvtstat(&sb, &osb); return (copyout(&osb, uap->ub, sizeof (osb))); } /* * Get file status; this version does not follow links. */ #ifndef _SYS_SYSPROTO_H_ struct olstat_args { char *path; struct ostat *ub; }; #endif int olstat(td, uap) struct thread *td; register struct olstat_args /* { char *path; struct ostat *ub; } */ *uap; { struct stat sb; struct ostat osb; int error; error = kern_statat(td, AT_SYMLINK_NOFOLLOW, AT_FDCWD, uap->path, UIO_USERSPACE, &sb, NULL); if (error != 0) return (error); cvtstat(&sb, &osb); return (copyout(&osb, uap->ub, sizeof (osb))); } /* * Convert from an old to a new stat structure. */ void cvtstat(st, ost) struct stat *st; struct ostat *ost; { bzero(ost, sizeof(*ost)); ost->st_dev = st->st_dev; ost->st_ino = st->st_ino; ost->st_mode = st->st_mode; ost->st_nlink = st->st_nlink; ost->st_uid = st->st_uid; ost->st_gid = st->st_gid; ost->st_rdev = st->st_rdev; if (st->st_size < (quad_t)1 << 32) ost->st_size = st->st_size; else ost->st_size = -2; ost->st_atim = st->st_atim; ost->st_mtim = st->st_mtim; ost->st_ctim = st->st_ctim; ost->st_blksize = st->st_blksize; ost->st_blocks = st->st_blocks; ost->st_flags = st->st_flags; ost->st_gen = st->st_gen; } #endif /* COMPAT_43 */ /* * Get file status; this version follows links. */ #ifndef _SYS_SYSPROTO_H_ struct stat_args { char *path; struct stat *ub; }; #endif int sys_stat(td, uap) struct thread *td; register struct stat_args /* { char *path; struct stat *ub; } */ *uap; { struct stat sb; int error; error = kern_statat(td, 0, AT_FDCWD, uap->path, UIO_USERSPACE, &sb, NULL); if (error == 0) error = copyout(&sb, uap->ub, sizeof (sb)); return (error); } #ifndef _SYS_SYSPROTO_H_ struct fstatat_args { int fd; char *path; struct stat *buf; int flag; } #endif int sys_fstatat(struct thread *td, struct fstatat_args *uap) { struct stat sb; int error; error = kern_statat(td, uap->flag, uap->fd, uap->path, UIO_USERSPACE, &sb, NULL); if (error == 0) error = copyout(&sb, uap->buf, sizeof (sb)); return (error); } int kern_statat(struct thread *td, int flag, int fd, char *path, enum uio_seg pathseg, struct stat *sbp, void (*hook)(struct vnode *vp, struct stat *sbp)) { struct nameidata nd; struct stat sb; cap_rights_t rights; int error; if (flag & ~AT_SYMLINK_NOFOLLOW) return (EINVAL); NDINIT_ATRIGHTS(&nd, LOOKUP, ((flag & AT_SYMLINK_NOFOLLOW) ? NOFOLLOW : FOLLOW) | LOCKSHARED | LOCKLEAF | AUDITVNODE1, pathseg, path, fd, cap_rights_init(&rights, CAP_FSTAT), td); if ((error = namei(&nd)) != 0) return (error); error = vn_stat(nd.ni_vp, &sb, td->td_ucred, NOCRED, td); if (error == 0) { SDT_PROBE2(vfs, , stat, mode, path, sb.st_mode); if (S_ISREG(sb.st_mode)) SDT_PROBE2(vfs, , stat, reg, path, pathseg); if (__predict_false(hook != NULL)) hook(nd.ni_vp, &sb); } NDFREE(&nd, NDF_ONLY_PNBUF); vput(nd.ni_vp); if (error != 0) return (error); *sbp = sb; #ifdef KTRACE if (KTRPOINT(td, KTR_STRUCT)) ktrstat(&sb); #endif return (0); } /* * Get file status; this version does not follow links. */ #ifndef _SYS_SYSPROTO_H_ struct lstat_args { char *path; struct stat *ub; }; #endif int sys_lstat(td, uap) struct thread *td; register struct lstat_args /* { char *path; struct stat *ub; } */ *uap; { struct stat sb; int error; error = kern_statat(td, AT_SYMLINK_NOFOLLOW, AT_FDCWD, uap->path, UIO_USERSPACE, &sb, NULL); if (error == 0) error = copyout(&sb, uap->ub, sizeof (sb)); return (error); } /* * Implementation of the NetBSD [l]stat() functions. */ void cvtnstat(sb, nsb) struct stat *sb; struct nstat *nsb; { bzero(nsb, sizeof *nsb); nsb->st_dev = sb->st_dev; nsb->st_ino = sb->st_ino; nsb->st_mode = sb->st_mode; nsb->st_nlink = sb->st_nlink; nsb->st_uid = sb->st_uid; nsb->st_gid = sb->st_gid; nsb->st_rdev = sb->st_rdev; nsb->st_atim = sb->st_atim; nsb->st_mtim = sb->st_mtim; nsb->st_ctim = sb->st_ctim; nsb->st_size = sb->st_size; nsb->st_blocks = sb->st_blocks; nsb->st_blksize = sb->st_blksize; nsb->st_flags = sb->st_flags; nsb->st_gen = sb->st_gen; nsb->st_birthtim = sb->st_birthtim; } #ifndef _SYS_SYSPROTO_H_ struct nstat_args { char *path; struct nstat *ub; }; #endif int sys_nstat(td, uap) struct thread *td; register struct nstat_args /* { char *path; struct nstat *ub; } */ *uap; { struct stat sb; struct nstat nsb; int error; error = kern_statat(td, 0, AT_FDCWD, uap->path, UIO_USERSPACE, &sb, NULL); if (error != 0) return (error); cvtnstat(&sb, &nsb); return (copyout(&nsb, uap->ub, sizeof (nsb))); } /* * NetBSD lstat. Get file status; this version does not follow links. */ #ifndef _SYS_SYSPROTO_H_ struct lstat_args { char *path; struct stat *ub; }; #endif int sys_nlstat(td, uap) struct thread *td; register struct nlstat_args /* { char *path; struct nstat *ub; } */ *uap; { struct stat sb; struct nstat nsb; int error; error = kern_statat(td, AT_SYMLINK_NOFOLLOW, AT_FDCWD, uap->path, UIO_USERSPACE, &sb, NULL); if (error != 0) return (error); cvtnstat(&sb, &nsb); return (copyout(&nsb, uap->ub, sizeof (nsb))); } /* * Get configurable pathname variables. */ #ifndef _SYS_SYSPROTO_H_ struct pathconf_args { char *path; int name; }; #endif int sys_pathconf(td, uap) struct thread *td; register struct pathconf_args /* { char *path; int name; } */ *uap; { return (kern_pathconf(td, uap->path, UIO_USERSPACE, uap->name, FOLLOW)); } #ifndef _SYS_SYSPROTO_H_ struct lpathconf_args { char *path; int name; }; #endif int sys_lpathconf(td, uap) struct thread *td; register struct lpathconf_args /* { char *path; int name; } */ *uap; { return (kern_pathconf(td, uap->path, UIO_USERSPACE, uap->name, NOFOLLOW)); } int kern_pathconf(struct thread *td, char *path, enum uio_seg pathseg, int name, u_long flags) { struct nameidata nd; int error; NDINIT(&nd, LOOKUP, LOCKSHARED | LOCKLEAF | AUDITVNODE1 | flags, pathseg, path, td); if ((error = namei(&nd)) != 0) return (error); NDFREE(&nd, NDF_ONLY_PNBUF); error = VOP_PATHCONF(nd.ni_vp, name, td->td_retval); vput(nd.ni_vp); return (error); } /* * Return target name of a symbolic link. */ #ifndef _SYS_SYSPROTO_H_ struct readlink_args { char *path; char *buf; size_t count; }; #endif int sys_readlink(td, uap) struct thread *td; register struct readlink_args /* { char *path; char *buf; size_t count; } */ *uap; { return (kern_readlinkat(td, AT_FDCWD, uap->path, UIO_USERSPACE, uap->buf, UIO_USERSPACE, uap->count)); } #ifndef _SYS_SYSPROTO_H_ struct readlinkat_args { int fd; char *path; char *buf; size_t bufsize; }; #endif int sys_readlinkat(struct thread *td, struct readlinkat_args *uap) { return (kern_readlinkat(td, uap->fd, uap->path, UIO_USERSPACE, uap->buf, UIO_USERSPACE, uap->bufsize)); } int kern_readlinkat(struct thread *td, int fd, char *path, enum uio_seg pathseg, char *buf, enum uio_seg bufseg, size_t count) { struct vnode *vp; struct iovec aiov; struct uio auio; struct nameidata nd; int error; if (count > IOSIZE_MAX) return (EINVAL); NDINIT_AT(&nd, LOOKUP, NOFOLLOW | LOCKSHARED | LOCKLEAF | AUDITVNODE1, pathseg, path, fd, td); if ((error = namei(&nd)) != 0) return (error); NDFREE(&nd, NDF_ONLY_PNBUF); vp = nd.ni_vp; #ifdef MAC error = mac_vnode_check_readlink(td->td_ucred, vp); if (error != 0) { vput(vp); return (error); } #endif if (vp->v_type != VLNK) error = EINVAL; else { aiov.iov_base = buf; aiov.iov_len = count; auio.uio_iov = &aiov; auio.uio_iovcnt = 1; auio.uio_offset = 0; auio.uio_rw = UIO_READ; auio.uio_segflg = bufseg; auio.uio_td = td; auio.uio_resid = count; error = VOP_READLINK(vp, &auio, td->td_ucred); td->td_retval[0] = count - auio.uio_resid; } vput(vp); return (error); } /* * Common implementation code for chflags() and fchflags(). */ static int setfflags(td, vp, flags) struct thread *td; struct vnode *vp; u_long flags; { struct mount *mp; struct vattr vattr; int error; /* We can't support the value matching VNOVAL. */ if (flags == VNOVAL) return (EOPNOTSUPP); /* * Prevent non-root users from setting flags on devices. When * a device is reused, users can retain ownership of the device * if they are allowed to set flags and programs assume that * chown can't fail when done as root. */ if (vp->v_type == VCHR || vp->v_type == VBLK) { error = priv_check(td, PRIV_VFS_CHFLAGS_DEV); if (error != 0) return (error); } if ((error = vn_start_write(vp, &mp, V_WAIT | PCATCH)) != 0) return (error); VATTR_NULL(&vattr); vattr.va_flags = flags; vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); #ifdef MAC error = mac_vnode_check_setflags(td->td_ucred, vp, vattr.va_flags); if (error == 0) #endif error = VOP_SETATTR(vp, &vattr, td->td_ucred); VOP_UNLOCK(vp, 0); vn_finished_write(mp); return (error); } /* * Change flags of a file given a path name. */ #ifndef _SYS_SYSPROTO_H_ struct chflags_args { const char *path; u_long flags; }; #endif int sys_chflags(td, uap) struct thread *td; register struct chflags_args /* { const char *path; u_long flags; } */ *uap; { return (kern_chflagsat(td, AT_FDCWD, uap->path, UIO_USERSPACE, uap->flags, 0)); } #ifndef _SYS_SYSPROTO_H_ struct chflagsat_args { int fd; const char *path; u_long flags; int atflag; } #endif int sys_chflagsat(struct thread *td, struct chflagsat_args *uap) { int fd = uap->fd; const char *path = uap->path; u_long flags = uap->flags; int atflag = uap->atflag; if (atflag & ~AT_SYMLINK_NOFOLLOW) return (EINVAL); return (kern_chflagsat(td, fd, path, UIO_USERSPACE, flags, atflag)); } /* * Same as chflags() but doesn't follow symlinks. */ int sys_lchflags(td, uap) struct thread *td; register struct lchflags_args /* { const char *path; u_long flags; } */ *uap; { return (kern_chflagsat(td, AT_FDCWD, uap->path, UIO_USERSPACE, uap->flags, AT_SYMLINK_NOFOLLOW)); } static int kern_chflagsat(struct thread *td, int fd, const char *path, enum uio_seg pathseg, u_long flags, int atflag) { struct nameidata nd; cap_rights_t rights; int error, follow; AUDIT_ARG_FFLAGS(flags); follow = (atflag & AT_SYMLINK_NOFOLLOW) ? NOFOLLOW : FOLLOW; NDINIT_ATRIGHTS(&nd, LOOKUP, follow | AUDITVNODE1, pathseg, path, fd, cap_rights_init(&rights, CAP_FCHFLAGS), td); if ((error = namei(&nd)) != 0) return (error); NDFREE(&nd, NDF_ONLY_PNBUF); error = setfflags(td, nd.ni_vp, flags); vrele(nd.ni_vp); return (error); } /* * Change flags of a file given a file descriptor. */ #ifndef _SYS_SYSPROTO_H_ struct fchflags_args { int fd; u_long flags; }; #endif int sys_fchflags(td, uap) struct thread *td; register struct fchflags_args /* { int fd; u_long flags; } */ *uap; { struct file *fp; cap_rights_t rights; int error; AUDIT_ARG_FD(uap->fd); AUDIT_ARG_FFLAGS(uap->flags); error = getvnode(td, uap->fd, cap_rights_init(&rights, CAP_FCHFLAGS), &fp); if (error != 0) return (error); #ifdef AUDIT vn_lock(fp->f_vnode, LK_SHARED | LK_RETRY); AUDIT_ARG_VNODE1(fp->f_vnode); VOP_UNLOCK(fp->f_vnode, 0); #endif error = setfflags(td, fp->f_vnode, uap->flags); fdrop(fp, td); return (error); } /* * Common implementation code for chmod(), lchmod() and fchmod(). */ int setfmode(td, cred, vp, mode) struct thread *td; struct ucred *cred; struct vnode *vp; int mode; { struct mount *mp; struct vattr vattr; int error; if ((error = vn_start_write(vp, &mp, V_WAIT | PCATCH)) != 0) return (error); vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); VATTR_NULL(&vattr); vattr.va_mode = mode & ALLPERMS; #ifdef MAC error = mac_vnode_check_setmode(cred, vp, vattr.va_mode); if (error == 0) #endif error = VOP_SETATTR(vp, &vattr, cred); VOP_UNLOCK(vp, 0); vn_finished_write(mp); return (error); } /* * Change mode of a file given path name. */ #ifndef _SYS_SYSPROTO_H_ struct chmod_args { char *path; int mode; }; #endif int sys_chmod(td, uap) struct thread *td; register struct chmod_args /* { char *path; int mode; } */ *uap; { return (kern_fchmodat(td, AT_FDCWD, uap->path, UIO_USERSPACE, uap->mode, 0)); } #ifndef _SYS_SYSPROTO_H_ struct fchmodat_args { int dirfd; char *path; mode_t mode; int flag; } #endif int sys_fchmodat(struct thread *td, struct fchmodat_args *uap) { int flag = uap->flag; int fd = uap->fd; char *path = uap->path; mode_t mode = uap->mode; if (flag & ~AT_SYMLINK_NOFOLLOW) return (EINVAL); return (kern_fchmodat(td, fd, path, UIO_USERSPACE, mode, flag)); } /* * Change mode of a file given path name (don't follow links.) */ #ifndef _SYS_SYSPROTO_H_ struct lchmod_args { char *path; int mode; }; #endif int sys_lchmod(td, uap) struct thread *td; register struct lchmod_args /* { char *path; int mode; } */ *uap; { return (kern_fchmodat(td, AT_FDCWD, uap->path, UIO_USERSPACE, uap->mode, AT_SYMLINK_NOFOLLOW)); } int kern_fchmodat(struct thread *td, int fd, char *path, enum uio_seg pathseg, mode_t mode, int flag) { struct nameidata nd; cap_rights_t rights; int error, follow; AUDIT_ARG_MODE(mode); follow = (flag & AT_SYMLINK_NOFOLLOW) ? NOFOLLOW : FOLLOW; NDINIT_ATRIGHTS(&nd, LOOKUP, follow | AUDITVNODE1, pathseg, path, fd, cap_rights_init(&rights, CAP_FCHMOD), td); if ((error = namei(&nd)) != 0) return (error); NDFREE(&nd, NDF_ONLY_PNBUF); error = setfmode(td, td->td_ucred, nd.ni_vp, mode); vrele(nd.ni_vp); return (error); } /* * Change mode of a file given a file descriptor. */ #ifndef _SYS_SYSPROTO_H_ struct fchmod_args { int fd; int mode; }; #endif int sys_fchmod(struct thread *td, struct fchmod_args *uap) { struct file *fp; cap_rights_t rights; int error; AUDIT_ARG_FD(uap->fd); AUDIT_ARG_MODE(uap->mode); error = fget(td, uap->fd, cap_rights_init(&rights, CAP_FCHMOD), &fp); if (error != 0) return (error); error = fo_chmod(fp, uap->mode, td->td_ucred, td); fdrop(fp, td); return (error); } /* * Common implementation for chown(), lchown(), and fchown() */ int setfown(td, cred, vp, uid, gid) struct thread *td; struct ucred *cred; struct vnode *vp; uid_t uid; gid_t gid; { struct mount *mp; struct vattr vattr; int error; if ((error = vn_start_write(vp, &mp, V_WAIT | PCATCH)) != 0) return (error); vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); VATTR_NULL(&vattr); vattr.va_uid = uid; vattr.va_gid = gid; #ifdef MAC error = mac_vnode_check_setowner(cred, vp, vattr.va_uid, vattr.va_gid); if (error == 0) #endif error = VOP_SETATTR(vp, &vattr, cred); VOP_UNLOCK(vp, 0); vn_finished_write(mp); return (error); } /* * Set ownership given a path name. */ #ifndef _SYS_SYSPROTO_H_ struct chown_args { char *path; int uid; int gid; }; #endif int sys_chown(td, uap) struct thread *td; register struct chown_args /* { char *path; int uid; int gid; } */ *uap; { return (kern_fchownat(td, AT_FDCWD, uap->path, UIO_USERSPACE, uap->uid, uap->gid, 0)); } #ifndef _SYS_SYSPROTO_H_ struct fchownat_args { int fd; const char * path; uid_t uid; gid_t gid; int flag; }; #endif int sys_fchownat(struct thread *td, struct fchownat_args *uap) { int flag; flag = uap->flag; if (flag & ~AT_SYMLINK_NOFOLLOW) return (EINVAL); return (kern_fchownat(td, uap->fd, uap->path, UIO_USERSPACE, uap->uid, uap->gid, uap->flag)); } int kern_fchownat(struct thread *td, int fd, char *path, enum uio_seg pathseg, int uid, int gid, int flag) { struct nameidata nd; cap_rights_t rights; int error, follow; AUDIT_ARG_OWNER(uid, gid); follow = (flag & AT_SYMLINK_NOFOLLOW) ? NOFOLLOW : FOLLOW; NDINIT_ATRIGHTS(&nd, LOOKUP, follow | AUDITVNODE1, pathseg, path, fd, cap_rights_init(&rights, CAP_FCHOWN), td); if ((error = namei(&nd)) != 0) return (error); NDFREE(&nd, NDF_ONLY_PNBUF); error = setfown(td, td->td_ucred, nd.ni_vp, uid, gid); vrele(nd.ni_vp); return (error); } /* * Set ownership given a path name, do not cross symlinks. */ #ifndef _SYS_SYSPROTO_H_ struct lchown_args { char *path; int uid; int gid; }; #endif int sys_lchown(td, uap) struct thread *td; register struct lchown_args /* { char *path; int uid; int gid; } */ *uap; { return (kern_fchownat(td, AT_FDCWD, uap->path, UIO_USERSPACE, uap->uid, uap->gid, AT_SYMLINK_NOFOLLOW)); } /* * Set ownership given a file descriptor. */ #ifndef _SYS_SYSPROTO_H_ struct fchown_args { int fd; int uid; int gid; }; #endif int sys_fchown(td, uap) struct thread *td; register struct fchown_args /* { int fd; int uid; int gid; } */ *uap; { struct file *fp; cap_rights_t rights; int error; AUDIT_ARG_FD(uap->fd); AUDIT_ARG_OWNER(uap->uid, uap->gid); error = fget(td, uap->fd, cap_rights_init(&rights, CAP_FCHOWN), &fp); if (error != 0) return (error); error = fo_chown(fp, uap->uid, uap->gid, td->td_ucred, td); fdrop(fp, td); return (error); } /* * Common implementation code for utimes(), lutimes(), and futimes(). */ static int getutimes(usrtvp, tvpseg, tsp) const struct timeval *usrtvp; enum uio_seg tvpseg; struct timespec *tsp; { struct timeval tv[2]; const struct timeval *tvp; int error; if (usrtvp == NULL) { vfs_timestamp(&tsp[0]); tsp[1] = tsp[0]; } else { if (tvpseg == UIO_SYSSPACE) { tvp = usrtvp; } else { if ((error = copyin(usrtvp, tv, sizeof(tv))) != 0) return (error); tvp = tv; } if (tvp[0].tv_usec < 0 || tvp[0].tv_usec >= 1000000 || tvp[1].tv_usec < 0 || tvp[1].tv_usec >= 1000000) return (EINVAL); TIMEVAL_TO_TIMESPEC(&tvp[0], &tsp[0]); TIMEVAL_TO_TIMESPEC(&tvp[1], &tsp[1]); } return (0); } /* * Common implementation code for futimens(), utimensat(). */ #define UTIMENS_NULL 0x1 #define UTIMENS_EXIT 0x2 static int getutimens(const struct timespec *usrtsp, enum uio_seg tspseg, struct timespec *tsp, int *retflags) { struct timespec tsnow; int error; vfs_timestamp(&tsnow); *retflags = 0; if (usrtsp == NULL) { tsp[0] = tsnow; tsp[1] = tsnow; *retflags |= UTIMENS_NULL; return (0); } if (tspseg == UIO_SYSSPACE) { tsp[0] = usrtsp[0]; tsp[1] = usrtsp[1]; } else if ((error = copyin(usrtsp, tsp, sizeof(*tsp) * 2)) != 0) return (error); if (tsp[0].tv_nsec == UTIME_OMIT && tsp[1].tv_nsec == UTIME_OMIT) *retflags |= UTIMENS_EXIT; if (tsp[0].tv_nsec == UTIME_NOW && tsp[1].tv_nsec == UTIME_NOW) *retflags |= UTIMENS_NULL; if (tsp[0].tv_nsec == UTIME_OMIT) tsp[0].tv_sec = VNOVAL; else if (tsp[0].tv_nsec == UTIME_NOW) tsp[0] = tsnow; else if (tsp[0].tv_nsec < 0 || tsp[0].tv_nsec >= 1000000000L) return (EINVAL); if (tsp[1].tv_nsec == UTIME_OMIT) tsp[1].tv_sec = VNOVAL; else if (tsp[1].tv_nsec == UTIME_NOW) tsp[1] = tsnow; else if (tsp[1].tv_nsec < 0 || tsp[1].tv_nsec >= 1000000000L) return (EINVAL); return (0); } /* * Common implementation code for utimes(), lutimes(), futimes(), futimens(), * and utimensat(). */ static int setutimes(td, vp, ts, numtimes, nullflag) struct thread *td; struct vnode *vp; const struct timespec *ts; int numtimes; int nullflag; { struct mount *mp; struct vattr vattr; int error, setbirthtime; if ((error = vn_start_write(vp, &mp, V_WAIT | PCATCH)) != 0) return (error); vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); setbirthtime = 0; if (numtimes < 3 && !VOP_GETATTR(vp, &vattr, td->td_ucred) && timespeccmp(&ts[1], &vattr.va_birthtime, < )) setbirthtime = 1; VATTR_NULL(&vattr); vattr.va_atime = ts[0]; vattr.va_mtime = ts[1]; if (setbirthtime) vattr.va_birthtime = ts[1]; if (numtimes > 2) vattr.va_birthtime = ts[2]; if (nullflag) vattr.va_vaflags |= VA_UTIMES_NULL; #ifdef MAC error = mac_vnode_check_setutimes(td->td_ucred, vp, vattr.va_atime, vattr.va_mtime); #endif if (error == 0) error = VOP_SETATTR(vp, &vattr, td->td_ucred); VOP_UNLOCK(vp, 0); vn_finished_write(mp); return (error); } /* * Set the access and modification times of a file. */ #ifndef _SYS_SYSPROTO_H_ struct utimes_args { char *path; struct timeval *tptr; }; #endif int sys_utimes(td, uap) struct thread *td; register struct utimes_args /* { char *path; struct timeval *tptr; } */ *uap; { return (kern_utimesat(td, AT_FDCWD, uap->path, UIO_USERSPACE, uap->tptr, UIO_USERSPACE)); } #ifndef _SYS_SYSPROTO_H_ struct futimesat_args { int fd; const char * path; const struct timeval * times; }; #endif int sys_futimesat(struct thread *td, struct futimesat_args *uap) { return (kern_utimesat(td, uap->fd, uap->path, UIO_USERSPACE, uap->times, UIO_USERSPACE)); } int kern_utimesat(struct thread *td, int fd, char *path, enum uio_seg pathseg, struct timeval *tptr, enum uio_seg tptrseg) { struct nameidata nd; struct timespec ts[2]; cap_rights_t rights; int error; if ((error = getutimes(tptr, tptrseg, ts)) != 0) return (error); NDINIT_ATRIGHTS(&nd, LOOKUP, FOLLOW | AUDITVNODE1, pathseg, path, fd, cap_rights_init(&rights, CAP_FUTIMES), td); if ((error = namei(&nd)) != 0) return (error); NDFREE(&nd, NDF_ONLY_PNBUF); error = setutimes(td, nd.ni_vp, ts, 2, tptr == NULL); vrele(nd.ni_vp); return (error); } /* * Set the access and modification times of a file. */ #ifndef _SYS_SYSPROTO_H_ struct lutimes_args { char *path; struct timeval *tptr; }; #endif int sys_lutimes(td, uap) struct thread *td; register struct lutimes_args /* { char *path; struct timeval *tptr; } */ *uap; { return (kern_lutimes(td, uap->path, UIO_USERSPACE, uap->tptr, UIO_USERSPACE)); } int kern_lutimes(struct thread *td, char *path, enum uio_seg pathseg, struct timeval *tptr, enum uio_seg tptrseg) { struct timespec ts[2]; struct nameidata nd; int error; if ((error = getutimes(tptr, tptrseg, ts)) != 0) return (error); NDINIT(&nd, LOOKUP, NOFOLLOW | AUDITVNODE1, pathseg, path, td); if ((error = namei(&nd)) != 0) return (error); NDFREE(&nd, NDF_ONLY_PNBUF); error = setutimes(td, nd.ni_vp, ts, 2, tptr == NULL); vrele(nd.ni_vp); return (error); } /* * Set the access and modification times of a file. */ #ifndef _SYS_SYSPROTO_H_ struct futimes_args { int fd; struct timeval *tptr; }; #endif int sys_futimes(td, uap) struct thread *td; register struct futimes_args /* { int fd; struct timeval *tptr; } */ *uap; { return (kern_futimes(td, uap->fd, uap->tptr, UIO_USERSPACE)); } int kern_futimes(struct thread *td, int fd, struct timeval *tptr, enum uio_seg tptrseg) { struct timespec ts[2]; struct file *fp; cap_rights_t rights; int error; AUDIT_ARG_FD(fd); error = getutimes(tptr, tptrseg, ts); if (error != 0) return (error); error = getvnode(td, fd, cap_rights_init(&rights, CAP_FUTIMES), &fp); if (error != 0) return (error); #ifdef AUDIT vn_lock(fp->f_vnode, LK_SHARED | LK_RETRY); AUDIT_ARG_VNODE1(fp->f_vnode); VOP_UNLOCK(fp->f_vnode, 0); #endif error = setutimes(td, fp->f_vnode, ts, 2, tptr == NULL); fdrop(fp, td); return (error); } int sys_futimens(struct thread *td, struct futimens_args *uap) { return (kern_futimens(td, uap->fd, uap->times, UIO_USERSPACE)); } int kern_futimens(struct thread *td, int fd, struct timespec *tptr, enum uio_seg tptrseg) { struct timespec ts[2]; struct file *fp; cap_rights_t rights; int error, flags; AUDIT_ARG_FD(fd); error = getutimens(tptr, tptrseg, ts, &flags); if (error != 0) return (error); if (flags & UTIMENS_EXIT) return (0); error = getvnode(td, fd, cap_rights_init(&rights, CAP_FUTIMES), &fp); if (error != 0) return (error); #ifdef AUDIT vn_lock(fp->f_vnode, LK_SHARED | LK_RETRY); AUDIT_ARG_VNODE1(fp->f_vnode); VOP_UNLOCK(fp->f_vnode, 0); #endif error = setutimes(td, fp->f_vnode, ts, 2, flags & UTIMENS_NULL); fdrop(fp, td); return (error); } int sys_utimensat(struct thread *td, struct utimensat_args *uap) { return (kern_utimensat(td, uap->fd, uap->path, UIO_USERSPACE, uap->times, UIO_USERSPACE, uap->flag)); } int kern_utimensat(struct thread *td, int fd, char *path, enum uio_seg pathseg, struct timespec *tptr, enum uio_seg tptrseg, int flag) { struct nameidata nd; struct timespec ts[2]; cap_rights_t rights; int error, flags; if (flag & ~AT_SYMLINK_NOFOLLOW) return (EINVAL); if ((error = getutimens(tptr, tptrseg, ts, &flags)) != 0) return (error); NDINIT_ATRIGHTS(&nd, LOOKUP, ((flag & AT_SYMLINK_NOFOLLOW) ? NOFOLLOW : FOLLOW) | AUDITVNODE1, pathseg, path, fd, cap_rights_init(&rights, CAP_FUTIMES), td); if ((error = namei(&nd)) != 0) return (error); /* * We are allowed to call namei() regardless of 2xUTIME_OMIT. * POSIX states: * "If both tv_nsec fields are UTIME_OMIT... EACCESS may be detected." * "Search permission is denied by a component of the path prefix." */ NDFREE(&nd, NDF_ONLY_PNBUF); if ((flags & UTIMENS_EXIT) == 0) error = setutimes(td, nd.ni_vp, ts, 2, flags & UTIMENS_NULL); vrele(nd.ni_vp); return (error); } /* * Truncate a file given its path name. */ #ifndef _SYS_SYSPROTO_H_ struct truncate_args { char *path; int pad; off_t length; }; #endif int sys_truncate(td, uap) struct thread *td; register struct truncate_args /* { char *path; int pad; off_t length; } */ *uap; { return (kern_truncate(td, uap->path, UIO_USERSPACE, uap->length)); } int kern_truncate(struct thread *td, char *path, enum uio_seg pathseg, off_t length) { struct mount *mp; struct vnode *vp; void *rl_cookie; struct vattr vattr; struct nameidata nd; int error; if (length < 0) return(EINVAL); NDINIT(&nd, LOOKUP, FOLLOW | AUDITVNODE1, pathseg, path, td); if ((error = namei(&nd)) != 0) return (error); vp = nd.ni_vp; rl_cookie = vn_rangelock_wlock(vp, 0, OFF_MAX); if ((error = vn_start_write(vp, &mp, V_WAIT | PCATCH)) != 0) { vn_rangelock_unlock(vp, rl_cookie); vrele(vp); return (error); } NDFREE(&nd, NDF_ONLY_PNBUF); vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); if (vp->v_type == VDIR) error = EISDIR; #ifdef MAC else if ((error = mac_vnode_check_write(td->td_ucred, NOCRED, vp))) { } #endif else if ((error = vn_writechk(vp)) == 0 && (error = VOP_ACCESS(vp, VWRITE, td->td_ucred, td)) == 0) { VATTR_NULL(&vattr); vattr.va_size = length; error = VOP_SETATTR(vp, &vattr, td->td_ucred); } VOP_UNLOCK(vp, 0); vn_finished_write(mp); vn_rangelock_unlock(vp, rl_cookie); vrele(vp); return (error); } #if defined(COMPAT_43) /* * Truncate a file given its path name. */ #ifndef _SYS_SYSPROTO_H_ struct otruncate_args { char *path; long length; }; #endif int otruncate(td, uap) struct thread *td; register struct otruncate_args /* { char *path; long length; } */ *uap; { struct truncate_args /* { char *path; int pad; off_t length; } */ nuap; nuap.path = uap->path; nuap.length = uap->length; return (sys_truncate(td, &nuap)); } #endif /* COMPAT_43 */ #if defined(COMPAT_FREEBSD6) /* Versions with the pad argument */ int freebsd6_truncate(struct thread *td, struct freebsd6_truncate_args *uap) { struct truncate_args ouap; ouap.path = uap->path; ouap.length = uap->length; return (sys_truncate(td, &ouap)); } int freebsd6_ftruncate(struct thread *td, struct freebsd6_ftruncate_args *uap) { struct ftruncate_args ouap; ouap.fd = uap->fd; ouap.length = uap->length; return (sys_ftruncate(td, &ouap)); } #endif int kern_fsync(struct thread *td, int fd, bool fullsync) { struct vnode *vp; struct mount *mp; struct file *fp; cap_rights_t rights; int error, lock_flags; AUDIT_ARG_FD(fd); error = getvnode(td, fd, cap_rights_init(&rights, CAP_FSYNC), &fp); if (error != 0) return (error); vp = fp->f_vnode; #if 0 if (!fullsync) /* XXXKIB: compete outstanding aio writes */; #endif error = vn_start_write(vp, &mp, V_WAIT | PCATCH); if (error != 0) goto drop; if (MNT_SHARED_WRITES(mp) || ((mp == NULL) && MNT_SHARED_WRITES(vp->v_mount))) { lock_flags = LK_SHARED; } else { lock_flags = LK_EXCLUSIVE; } vn_lock(vp, lock_flags | LK_RETRY); AUDIT_ARG_VNODE1(vp); if (vp->v_object != NULL) { VM_OBJECT_WLOCK(vp->v_object); vm_object_page_clean(vp->v_object, 0, 0, 0); VM_OBJECT_WUNLOCK(vp->v_object); } error = fullsync ? VOP_FSYNC(vp, MNT_WAIT, td) : VOP_FDATASYNC(vp, td); VOP_UNLOCK(vp, 0); vn_finished_write(mp); drop: fdrop(fp, td); return (error); } /* * Sync an open file. */ #ifndef _SYS_SYSPROTO_H_ struct fsync_args { int fd; }; #endif int sys_fsync(struct thread *td, struct fsync_args *uap) { return (kern_fsync(td, uap->fd, true)); } int sys_fdatasync(struct thread *td, struct fdatasync_args *uap) { return (kern_fsync(td, uap->fd, false)); } /* * Rename files. Source and destination must either both be directories, or * both not be directories. If target is a directory, it must be empty. */ #ifndef _SYS_SYSPROTO_H_ struct rename_args { char *from; char *to; }; #endif int sys_rename(td, uap) struct thread *td; register struct rename_args /* { char *from; char *to; } */ *uap; { return (kern_renameat(td, AT_FDCWD, uap->from, AT_FDCWD, uap->to, UIO_USERSPACE)); } #ifndef _SYS_SYSPROTO_H_ struct renameat_args { int oldfd; char *old; int newfd; char *new; }; #endif int sys_renameat(struct thread *td, struct renameat_args *uap) { return (kern_renameat(td, uap->oldfd, uap->old, uap->newfd, uap->new, UIO_USERSPACE)); } int kern_renameat(struct thread *td, int oldfd, char *old, int newfd, char *new, enum uio_seg pathseg) { struct mount *mp = NULL; struct vnode *tvp, *fvp, *tdvp; struct nameidata fromnd, tond; cap_rights_t rights; int error; again: bwillwrite(); #ifdef MAC NDINIT_ATRIGHTS(&fromnd, DELETE, LOCKPARENT | LOCKLEAF | SAVESTART | AUDITVNODE1, pathseg, old, oldfd, cap_rights_init(&rights, CAP_RENAMEAT_SOURCE), td); #else NDINIT_ATRIGHTS(&fromnd, DELETE, WANTPARENT | SAVESTART | AUDITVNODE1, pathseg, old, oldfd, cap_rights_init(&rights, CAP_RENAMEAT_SOURCE), td); #endif if ((error = namei(&fromnd)) != 0) return (error); #ifdef MAC error = mac_vnode_check_rename_from(td->td_ucred, fromnd.ni_dvp, fromnd.ni_vp, &fromnd.ni_cnd); VOP_UNLOCK(fromnd.ni_dvp, 0); if (fromnd.ni_dvp != fromnd.ni_vp) VOP_UNLOCK(fromnd.ni_vp, 0); #endif fvp = fromnd.ni_vp; NDINIT_ATRIGHTS(&tond, RENAME, LOCKPARENT | LOCKLEAF | NOCACHE | SAVESTART | AUDITVNODE2, pathseg, new, newfd, cap_rights_init(&rights, CAP_RENAMEAT_TARGET), td); if (fromnd.ni_vp->v_type == VDIR) tond.ni_cnd.cn_flags |= WILLBEDIR; if ((error = namei(&tond)) != 0) { /* Translate error code for rename("dir1", "dir2/."). */ if (error == EISDIR && fvp->v_type == VDIR) error = EINVAL; NDFREE(&fromnd, NDF_ONLY_PNBUF); vrele(fromnd.ni_dvp); vrele(fvp); goto out1; } tdvp = tond.ni_dvp; tvp = tond.ni_vp; error = vn_start_write(fvp, &mp, V_NOWAIT); if (error != 0) { NDFREE(&fromnd, NDF_ONLY_PNBUF); NDFREE(&tond, NDF_ONLY_PNBUF); if (tvp != NULL) vput(tvp); if (tdvp == tvp) vrele(tdvp); else vput(tdvp); vrele(fromnd.ni_dvp); vrele(fvp); vrele(tond.ni_startdir); if (fromnd.ni_startdir != NULL) vrele(fromnd.ni_startdir); error = vn_start_write(NULL, &mp, V_XSLEEP | PCATCH); if (error != 0) return (error); goto again; } if (tvp != NULL) { if (fvp->v_type == VDIR && tvp->v_type != VDIR) { error = ENOTDIR; goto out; } else if (fvp->v_type != VDIR && tvp->v_type == VDIR) { error = EISDIR; goto out; } #ifdef CAPABILITIES if (newfd != AT_FDCWD) { /* * If the target already exists we require CAP_UNLINKAT * from 'newfd'. */ error = cap_check(&tond.ni_filecaps.fc_rights, cap_rights_init(&rights, CAP_UNLINKAT)); if (error != 0) goto out; } #endif } if (fvp == tdvp) { error = EINVAL; goto out; } /* * If the source is the same as the destination (that is, if they * are links to the same vnode), then there is nothing to do. */ if (fvp == tvp) error = -1; #ifdef MAC else error = mac_vnode_check_rename_to(td->td_ucred, tdvp, tond.ni_vp, fromnd.ni_dvp == tdvp, &tond.ni_cnd); #endif out: if (error == 0) { error = VOP_RENAME(fromnd.ni_dvp, fromnd.ni_vp, &fromnd.ni_cnd, tond.ni_dvp, tond.ni_vp, &tond.ni_cnd); NDFREE(&fromnd, NDF_ONLY_PNBUF); NDFREE(&tond, NDF_ONLY_PNBUF); } else { NDFREE(&fromnd, NDF_ONLY_PNBUF); NDFREE(&tond, NDF_ONLY_PNBUF); if (tvp != NULL) vput(tvp); if (tdvp == tvp) vrele(tdvp); else vput(tdvp); vrele(fromnd.ni_dvp); vrele(fvp); } vrele(tond.ni_startdir); vn_finished_write(mp); out1: if (fromnd.ni_startdir) vrele(fromnd.ni_startdir); if (error == -1) return (0); return (error); } /* * Make a directory file. */ #ifndef _SYS_SYSPROTO_H_ struct mkdir_args { char *path; int mode; }; #endif int sys_mkdir(td, uap) struct thread *td; register struct mkdir_args /* { char *path; int mode; } */ *uap; { return (kern_mkdirat(td, AT_FDCWD, uap->path, UIO_USERSPACE, uap->mode)); } #ifndef _SYS_SYSPROTO_H_ struct mkdirat_args { int fd; char *path; mode_t mode; }; #endif int sys_mkdirat(struct thread *td, struct mkdirat_args *uap) { return (kern_mkdirat(td, uap->fd, uap->path, UIO_USERSPACE, uap->mode)); } int kern_mkdirat(struct thread *td, int fd, char *path, enum uio_seg segflg, int mode) { struct mount *mp; struct vnode *vp; struct vattr vattr; struct nameidata nd; cap_rights_t rights; int error; AUDIT_ARG_MODE(mode); restart: bwillwrite(); NDINIT_ATRIGHTS(&nd, CREATE, LOCKPARENT | SAVENAME | AUDITVNODE1 | NOCACHE, segflg, path, fd, cap_rights_init(&rights, CAP_MKDIRAT), td); nd.ni_cnd.cn_flags |= WILLBEDIR; if ((error = namei(&nd)) != 0) return (error); vp = nd.ni_vp; if (vp != NULL) { NDFREE(&nd, NDF_ONLY_PNBUF); /* * XXX namei called with LOCKPARENT but not LOCKLEAF has * the strange behaviour of leaving the vnode unlocked * if the target is the same vnode as the parent. */ if (vp == nd.ni_dvp) vrele(nd.ni_dvp); else vput(nd.ni_dvp); vrele(vp); return (EEXIST); } if (vn_start_write(nd.ni_dvp, &mp, V_NOWAIT) != 0) { NDFREE(&nd, NDF_ONLY_PNBUF); vput(nd.ni_dvp); if ((error = vn_start_write(NULL, &mp, V_XSLEEP | PCATCH)) != 0) return (error); goto restart; } VATTR_NULL(&vattr); vattr.va_type = VDIR; vattr.va_mode = (mode & ACCESSPERMS) &~ td->td_proc->p_fd->fd_cmask; #ifdef MAC error = mac_vnode_check_create(td->td_ucred, nd.ni_dvp, &nd.ni_cnd, &vattr); if (error != 0) goto out; #endif error = VOP_MKDIR(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, &vattr); #ifdef MAC out: #endif NDFREE(&nd, NDF_ONLY_PNBUF); vput(nd.ni_dvp); if (error == 0) vput(nd.ni_vp); vn_finished_write(mp); return (error); } /* * Remove a directory file. */ #ifndef _SYS_SYSPROTO_H_ struct rmdir_args { char *path; }; #endif int sys_rmdir(td, uap) struct thread *td; struct rmdir_args /* { char *path; } */ *uap; { return (kern_rmdirat(td, AT_FDCWD, uap->path, UIO_USERSPACE)); } int kern_rmdirat(struct thread *td, int fd, char *path, enum uio_seg pathseg) { struct mount *mp; struct vnode *vp; struct nameidata nd; cap_rights_t rights; int error; restart: bwillwrite(); NDINIT_ATRIGHTS(&nd, DELETE, LOCKPARENT | LOCKLEAF | AUDITVNODE1, pathseg, path, fd, cap_rights_init(&rights, CAP_UNLINKAT), td); if ((error = namei(&nd)) != 0) return (error); vp = nd.ni_vp; if (vp->v_type != VDIR) { error = ENOTDIR; goto out; } /* * No rmdir "." please. */ if (nd.ni_dvp == vp) { error = EINVAL; goto out; } /* * The root of a mounted filesystem cannot be deleted. */ if (vp->v_vflag & VV_ROOT) { error = EBUSY; goto out; } #ifdef MAC error = mac_vnode_check_unlink(td->td_ucred, nd.ni_dvp, vp, &nd.ni_cnd); if (error != 0) goto out; #endif if (vn_start_write(nd.ni_dvp, &mp, V_NOWAIT) != 0) { NDFREE(&nd, NDF_ONLY_PNBUF); vput(vp); if (nd.ni_dvp == vp) vrele(nd.ni_dvp); else vput(nd.ni_dvp); if ((error = vn_start_write(NULL, &mp, V_XSLEEP | PCATCH)) != 0) return (error); goto restart; } vfs_notify_upper(vp, VFS_NOTIFY_UPPER_UNLINK); error = VOP_RMDIR(nd.ni_dvp, nd.ni_vp, &nd.ni_cnd); vn_finished_write(mp); out: NDFREE(&nd, NDF_ONLY_PNBUF); vput(vp); if (nd.ni_dvp == vp) vrele(nd.ni_dvp); else vput(nd.ni_dvp); return (error); } #ifdef COMPAT_43 /* * Read a block of directory entries in a filesystem independent format. */ #ifndef _SYS_SYSPROTO_H_ struct ogetdirentries_args { int fd; char *buf; u_int count; long *basep; }; #endif int ogetdirentries(struct thread *td, struct ogetdirentries_args *uap) { long loff; int error; error = kern_ogetdirentries(td, uap, &loff); if (error == 0) error = copyout(&loff, uap->basep, sizeof(long)); return (error); } int kern_ogetdirentries(struct thread *td, struct ogetdirentries_args *uap, long *ploff) { struct vnode *vp; struct file *fp; struct uio auio, kuio; struct iovec aiov, kiov; struct dirent *dp, *edp; cap_rights_t rights; caddr_t dirbuf; int error, eofflag, readcnt; long loff; off_t foffset; /* XXX arbitrary sanity limit on `count'. */ if (uap->count > 64 * 1024) return (EINVAL); error = getvnode(td, uap->fd, cap_rights_init(&rights, CAP_READ), &fp); if (error != 0) return (error); if ((fp->f_flag & FREAD) == 0) { fdrop(fp, td); return (EBADF); } vp = fp->f_vnode; foffset = foffset_lock(fp, 0); unionread: if (vp->v_type != VDIR) { foffset_unlock(fp, foffset, 0); fdrop(fp, td); return (EINVAL); } aiov.iov_base = uap->buf; aiov.iov_len = uap->count; auio.uio_iov = &aiov; auio.uio_iovcnt = 1; auio.uio_rw = UIO_READ; auio.uio_segflg = UIO_USERSPACE; auio.uio_td = td; auio.uio_resid = uap->count; vn_lock(vp, LK_SHARED | LK_RETRY); loff = auio.uio_offset = foffset; #ifdef MAC error = mac_vnode_check_readdir(td->td_ucred, vp); if (error != 0) { VOP_UNLOCK(vp, 0); foffset_unlock(fp, foffset, FOF_NOUPDATE); fdrop(fp, td); return (error); } #endif # if (BYTE_ORDER != LITTLE_ENDIAN) if (vp->v_mount->mnt_maxsymlinklen <= 0) { error = VOP_READDIR(vp, &auio, fp->f_cred, &eofflag, NULL, NULL); foffset = auio.uio_offset; } else # endif { kuio = auio; kuio.uio_iov = &kiov; kuio.uio_segflg = UIO_SYSSPACE; kiov.iov_len = uap->count; dirbuf = malloc(uap->count, M_TEMP, M_WAITOK); kiov.iov_base = dirbuf; error = VOP_READDIR(vp, &kuio, fp->f_cred, &eofflag, NULL, NULL); foffset = kuio.uio_offset; if (error == 0) { readcnt = uap->count - kuio.uio_resid; edp = (struct dirent *)&dirbuf[readcnt]; for (dp = (struct dirent *)dirbuf; dp < edp; ) { # if (BYTE_ORDER == LITTLE_ENDIAN) /* * The expected low byte of * dp->d_namlen is our dp->d_type. * The high MBZ byte of dp->d_namlen * is our dp->d_namlen. */ dp->d_type = dp->d_namlen; dp->d_namlen = 0; # else /* * The dp->d_type is the high byte * of the expected dp->d_namlen, * so must be zero'ed. */ dp->d_type = 0; # endif if (dp->d_reclen > 0) { dp = (struct dirent *) ((char *)dp + dp->d_reclen); } else { error = EIO; break; } } if (dp >= edp) error = uiomove(dirbuf, readcnt, &auio); } free(dirbuf, M_TEMP); } if (error != 0) { VOP_UNLOCK(vp, 0); foffset_unlock(fp, foffset, 0); fdrop(fp, td); return (error); } if (uap->count == auio.uio_resid && (vp->v_vflag & VV_ROOT) && (vp->v_mount->mnt_flag & MNT_UNION)) { struct vnode *tvp = vp; vp = vp->v_mount->mnt_vnodecovered; VREF(vp); fp->f_vnode = vp; fp->f_data = vp; foffset = 0; vput(tvp); goto unionread; } VOP_UNLOCK(vp, 0); foffset_unlock(fp, foffset, 0); fdrop(fp, td); td->td_retval[0] = uap->count - auio.uio_resid; if (error == 0) *ploff = loff; return (error); } #endif /* COMPAT_43 */ /* * Read a block of directory entries in a filesystem independent format. */ #ifndef _SYS_SYSPROTO_H_ struct getdirentries_args { int fd; char *buf; u_int count; long *basep; }; #endif int sys_getdirentries(td, uap) struct thread *td; register struct getdirentries_args /* { int fd; char *buf; u_int count; long *basep; } */ *uap; { long base; int error; error = kern_getdirentries(td, uap->fd, uap->buf, uap->count, &base, NULL, UIO_USERSPACE); if (error != 0) return (error); if (uap->basep != NULL) error = copyout(&base, uap->basep, sizeof(long)); return (error); } int kern_getdirentries(struct thread *td, int fd, char *buf, u_int count, long *basep, ssize_t *residp, enum uio_seg bufseg) { struct vnode *vp; struct file *fp; struct uio auio; struct iovec aiov; cap_rights_t rights; long loff; int error, eofflag; off_t foffset; AUDIT_ARG_FD(fd); if (count > IOSIZE_MAX) return (EINVAL); auio.uio_resid = count; error = getvnode(td, fd, cap_rights_init(&rights, CAP_READ), &fp); if (error != 0) return (error); if ((fp->f_flag & FREAD) == 0) { fdrop(fp, td); return (EBADF); } vp = fp->f_vnode; foffset = foffset_lock(fp, 0); unionread: if (vp->v_type != VDIR) { error = EINVAL; goto fail; } aiov.iov_base = buf; aiov.iov_len = count; auio.uio_iov = &aiov; auio.uio_iovcnt = 1; auio.uio_rw = UIO_READ; auio.uio_segflg = bufseg; auio.uio_td = td; vn_lock(vp, LK_SHARED | LK_RETRY); AUDIT_ARG_VNODE1(vp); loff = auio.uio_offset = foffset; #ifdef MAC error = mac_vnode_check_readdir(td->td_ucred, vp); if (error == 0) #endif error = VOP_READDIR(vp, &auio, fp->f_cred, &eofflag, NULL, NULL); foffset = auio.uio_offset; if (error != 0) { VOP_UNLOCK(vp, 0); goto fail; } if (count == auio.uio_resid && (vp->v_vflag & VV_ROOT) && (vp->v_mount->mnt_flag & MNT_UNION)) { struct vnode *tvp = vp; vp = vp->v_mount->mnt_vnodecovered; VREF(vp); fp->f_vnode = vp; fp->f_data = vp; foffset = 0; vput(tvp); goto unionread; } VOP_UNLOCK(vp, 0); *basep = loff; if (residp != NULL) *residp = auio.uio_resid; td->td_retval[0] = count - auio.uio_resid; fail: foffset_unlock(fp, foffset, 0); fdrop(fp, td); return (error); } #ifndef _SYS_SYSPROTO_H_ struct getdents_args { int fd; char *buf; size_t count; }; #endif int sys_getdents(td, uap) struct thread *td; register struct getdents_args /* { int fd; char *buf; u_int count; } */ *uap; { struct getdirentries_args ap; ap.fd = uap->fd; ap.buf = uap->buf; ap.count = uap->count; ap.basep = NULL; return (sys_getdirentries(td, &ap)); } /* * Set the mode mask for creation of filesystem nodes. */ #ifndef _SYS_SYSPROTO_H_ struct umask_args { int newmask; }; #endif int sys_umask(td, uap) struct thread *td; struct umask_args /* { int newmask; } */ *uap; { struct filedesc *fdp; fdp = td->td_proc->p_fd; FILEDESC_XLOCK(fdp); td->td_retval[0] = fdp->fd_cmask; fdp->fd_cmask = uap->newmask & ALLPERMS; FILEDESC_XUNLOCK(fdp); return (0); } /* * Void all references to file by ripping underlying filesystem away from * vnode. */ #ifndef _SYS_SYSPROTO_H_ struct revoke_args { char *path; }; #endif int sys_revoke(td, uap) struct thread *td; register struct revoke_args /* { char *path; } */ *uap; { struct vnode *vp; struct vattr vattr; struct nameidata nd; int error; NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF | AUDITVNODE1, UIO_USERSPACE, uap->path, td); if ((error = namei(&nd)) != 0) return (error); vp = nd.ni_vp; NDFREE(&nd, NDF_ONLY_PNBUF); if (vp->v_type != VCHR || vp->v_rdev == NULL) { error = EINVAL; goto out; } #ifdef MAC error = mac_vnode_check_revoke(td->td_ucred, vp); if (error != 0) goto out; #endif error = VOP_GETATTR(vp, &vattr, td->td_ucred); if (error != 0) goto out; if (td->td_ucred->cr_uid != vattr.va_uid) { error = priv_check(td, PRIV_VFS_ADMIN); if (error != 0) goto out; } if (vcount(vp) > 1) VOP_REVOKE(vp, REVOKEALL); out: vput(vp); return (error); } /* * Convert a user file descriptor to a kernel file entry and check that, if it * is a capability, the correct rights are present. A reference on the file * entry is held upon returning. */ int getvnode(struct thread *td, int fd, cap_rights_t *rightsp, struct file **fpp) { struct file *fp; int error; error = fget_unlocked(td->td_proc->p_fd, fd, rightsp, &fp, NULL); if (error != 0) return (error); /* * The file could be not of the vnode type, or it may be not * yet fully initialized, in which case the f_vnode pointer * may be set, but f_ops is still badfileops. E.g., * devfs_open() transiently create such situation to * facilitate csw d_fdopen(). * * Dupfdopen() handling in kern_openat() installs the * half-baked file into the process descriptor table, allowing * other thread to dereference it. Guard against the race by * checking f_ops. */ if (fp->f_vnode == NULL || fp->f_ops == &badfileops) { fdrop(fp, td); return (EINVAL); } *fpp = fp; return (0); } /* * Get an (NFS) file handle. */ #ifndef _SYS_SYSPROTO_H_ struct lgetfh_args { char *fname; fhandle_t *fhp; }; #endif int sys_lgetfh(td, uap) struct thread *td; register struct lgetfh_args *uap; { struct nameidata nd; fhandle_t fh; register struct vnode *vp; int error; error = priv_check(td, PRIV_VFS_GETFH); if (error != 0) return (error); NDINIT(&nd, LOOKUP, NOFOLLOW | LOCKLEAF | AUDITVNODE1, UIO_USERSPACE, uap->fname, td); error = namei(&nd); if (error != 0) return (error); NDFREE(&nd, NDF_ONLY_PNBUF); vp = nd.ni_vp; bzero(&fh, sizeof(fh)); fh.fh_fsid = vp->v_mount->mnt_stat.f_fsid; error = VOP_VPTOFH(vp, &fh.fh_fid); vput(vp); if (error == 0) error = copyout(&fh, uap->fhp, sizeof (fh)); return (error); } #ifndef _SYS_SYSPROTO_H_ struct getfh_args { char *fname; fhandle_t *fhp; }; #endif int sys_getfh(td, uap) struct thread *td; register struct getfh_args *uap; { struct nameidata nd; fhandle_t fh; register struct vnode *vp; int error; error = priv_check(td, PRIV_VFS_GETFH); if (error != 0) return (error); NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF | AUDITVNODE1, UIO_USERSPACE, uap->fname, td); error = namei(&nd); if (error != 0) return (error); NDFREE(&nd, NDF_ONLY_PNBUF); vp = nd.ni_vp; bzero(&fh, sizeof(fh)); fh.fh_fsid = vp->v_mount->mnt_stat.f_fsid; error = VOP_VPTOFH(vp, &fh.fh_fid); vput(vp); if (error == 0) error = copyout(&fh, uap->fhp, sizeof (fh)); return (error); } /* * syscall for the rpc.lockd to use to translate a NFS file handle into an * open descriptor. * * warning: do not remove the priv_check() call or this becomes one giant * security hole. */ #ifndef _SYS_SYSPROTO_H_ struct fhopen_args { const struct fhandle *u_fhp; int flags; }; #endif int sys_fhopen(td, uap) struct thread *td; struct fhopen_args /* { const struct fhandle *u_fhp; int flags; } */ *uap; { struct mount *mp; struct vnode *vp; struct fhandle fhp; struct file *fp; int fmode, error; int indx; error = priv_check(td, PRIV_VFS_FHOPEN); if (error != 0) return (error); indx = -1; fmode = FFLAGS(uap->flags); /* why not allow a non-read/write open for our lockd? */ if (((fmode & (FREAD | FWRITE)) == 0) || (fmode & O_CREAT)) return (EINVAL); error = copyin(uap->u_fhp, &fhp, sizeof(fhp)); if (error != 0) return(error); /* find the mount point */ mp = vfs_busyfs(&fhp.fh_fsid); if (mp == NULL) return (ESTALE); /* now give me my vnode, it gets returned to me locked */ error = VFS_FHTOVP(mp, &fhp.fh_fid, LK_EXCLUSIVE, &vp); vfs_unbusy(mp); if (error != 0) return (error); error = falloc_noinstall(td, &fp); if (error != 0) { vput(vp); return (error); } /* * An extra reference on `fp' has been held for us by * falloc_noinstall(). */ #ifdef INVARIANTS td->td_dupfd = -1; #endif error = vn_open_vnode(vp, fmode, td->td_ucred, td, fp); if (error != 0) { KASSERT(fp->f_ops == &badfileops, ("VOP_OPEN in fhopen() set f_ops")); KASSERT(td->td_dupfd < 0, ("fhopen() encountered fdopen()")); vput(vp); goto bad; } #ifdef INVARIANTS td->td_dupfd = 0; #endif fp->f_vnode = vp; fp->f_seqcount = 1; finit(fp, (fmode & FMASK) | (fp->f_flag & FHASLOCK), DTYPE_VNODE, vp, &vnops); VOP_UNLOCK(vp, 0); if ((fmode & O_TRUNC) != 0) { error = fo_truncate(fp, 0, td->td_ucred, td); if (error != 0) goto bad; } error = finstall(td, fp, &indx, fmode, NULL); bad: fdrop(fp, td); td->td_retval[0] = indx; return (error); } /* * Stat an (NFS) file handle. */ #ifndef _SYS_SYSPROTO_H_ struct fhstat_args { struct fhandle *u_fhp; struct stat *sb; }; #endif int sys_fhstat(td, uap) struct thread *td; register struct fhstat_args /* { struct fhandle *u_fhp; struct stat *sb; } */ *uap; { struct stat sb; struct fhandle fh; int error; error = copyin(uap->u_fhp, &fh, sizeof(fh)); if (error != 0) return (error); error = kern_fhstat(td, fh, &sb); if (error == 0) error = copyout(&sb, uap->sb, sizeof(sb)); return (error); } int kern_fhstat(struct thread *td, struct fhandle fh, struct stat *sb) { struct mount *mp; struct vnode *vp; int error; error = priv_check(td, PRIV_VFS_FHSTAT); if (error != 0) return (error); if ((mp = vfs_busyfs(&fh.fh_fsid)) == NULL) return (ESTALE); error = VFS_FHTOVP(mp, &fh.fh_fid, LK_EXCLUSIVE, &vp); vfs_unbusy(mp); if (error != 0) return (error); error = vn_stat(vp, sb, td->td_ucred, NOCRED, td); vput(vp); return (error); } /* * Implement fstatfs() for (NFS) file handles. */ #ifndef _SYS_SYSPROTO_H_ struct fhstatfs_args { struct fhandle *u_fhp; struct statfs *buf; }; #endif int sys_fhstatfs(td, uap) struct thread *td; struct fhstatfs_args /* { struct fhandle *u_fhp; struct statfs *buf; } */ *uap; { - struct statfs sf; + struct statfs *sfp; fhandle_t fh; int error; error = copyin(uap->u_fhp, &fh, sizeof(fhandle_t)); if (error != 0) return (error); - error = kern_fhstatfs(td, fh, &sf); - if (error != 0) - return (error); - return (copyout(&sf, uap->buf, sizeof(sf))); + sfp = malloc(sizeof(struct statfs), M_STATFS, M_WAITOK); + error = kern_fhstatfs(td, fh, sfp); + if (error == 0) + error = copyout(sfp, uap->buf, sizeof(*sfp)); + free(sfp, M_STATFS); + return (error); } int kern_fhstatfs(struct thread *td, fhandle_t fh, struct statfs *buf) { struct statfs *sp; struct mount *mp; struct vnode *vp; int error; error = priv_check(td, PRIV_VFS_FHSTATFS); if (error != 0) return (error); if ((mp = vfs_busyfs(&fh.fh_fsid)) == NULL) return (ESTALE); error = VFS_FHTOVP(mp, &fh.fh_fid, LK_EXCLUSIVE, &vp); if (error != 0) { vfs_unbusy(mp); return (error); } vput(vp); error = prison_canseemount(td->td_ucred, mp); if (error != 0) goto out; #ifdef MAC error = mac_mount_check_stat(td->td_ucred, mp); if (error != 0) goto out; #endif /* * Set these in case the underlying filesystem fails to do so. */ sp = &mp->mnt_stat; sp->f_version = STATFS_VERSION; sp->f_namemax = NAME_MAX; sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK; error = VFS_STATFS(mp, sp); if (error == 0) *buf = *sp; out: vfs_unbusy(mp); return (error); } int kern_posix_fallocate(struct thread *td, int fd, off_t offset, off_t len) { struct file *fp; struct mount *mp; struct vnode *vp; cap_rights_t rights; off_t olen, ooffset; int error; if (offset < 0 || len <= 0) return (EINVAL); /* Check for wrap. */ if (offset > OFF_MAX - len) return (EFBIG); error = fget(td, fd, cap_rights_init(&rights, CAP_WRITE), &fp); if (error != 0) return (error); if ((fp->f_ops->fo_flags & DFLAG_SEEKABLE) == 0) { error = ESPIPE; goto out; } if ((fp->f_flag & FWRITE) == 0) { error = EBADF; goto out; } if (fp->f_type != DTYPE_VNODE) { error = ENODEV; goto out; } vp = fp->f_vnode; if (vp->v_type != VREG) { error = ENODEV; goto out; } /* Allocating blocks may take a long time, so iterate. */ for (;;) { olen = len; ooffset = offset; bwillwrite(); mp = NULL; error = vn_start_write(vp, &mp, V_WAIT | PCATCH); if (error != 0) break; error = vn_lock(vp, LK_EXCLUSIVE); if (error != 0) { vn_finished_write(mp); break; } #ifdef MAC error = mac_vnode_check_write(td->td_ucred, fp->f_cred, vp); if (error == 0) #endif error = VOP_ALLOCATE(vp, &offset, &len); VOP_UNLOCK(vp, 0); vn_finished_write(mp); if (olen + ooffset != offset + len) { panic("offset + len changed from %jx/%jx to %jx/%jx", ooffset, olen, offset, len); } if (error != 0 || len == 0) break; KASSERT(olen > len, ("Iteration did not make progress?")); maybe_yield(); } out: fdrop(fp, td); return (error); } int sys_posix_fallocate(struct thread *td, struct posix_fallocate_args *uap) { int error; error = kern_posix_fallocate(td, uap->fd, uap->offset, uap->len); return (kern_posix_error(td, error)); } /* * Unlike madvise(2), we do not make a best effort to remember every * possible caching hint. Instead, we remember the last setting with * the exception that we will allow POSIX_FADV_NORMAL to adjust the * region of any current setting. */ int kern_posix_fadvise(struct thread *td, int fd, off_t offset, off_t len, int advice) { struct fadvise_info *fa, *new; struct file *fp; struct vnode *vp; cap_rights_t rights; off_t end; int error; if (offset < 0 || len < 0 || offset > OFF_MAX - len) return (EINVAL); switch (advice) { case POSIX_FADV_SEQUENTIAL: case POSIX_FADV_RANDOM: case POSIX_FADV_NOREUSE: new = malloc(sizeof(*fa), M_FADVISE, M_WAITOK); break; case POSIX_FADV_NORMAL: case POSIX_FADV_WILLNEED: case POSIX_FADV_DONTNEED: new = NULL; break; default: return (EINVAL); } /* XXX: CAP_POSIX_FADVISE? */ error = fget(td, fd, cap_rights_init(&rights), &fp); if (error != 0) goto out; if ((fp->f_ops->fo_flags & DFLAG_SEEKABLE) == 0) { error = ESPIPE; goto out; } if (fp->f_type != DTYPE_VNODE) { error = ENODEV; goto out; } vp = fp->f_vnode; if (vp->v_type != VREG) { error = ENODEV; goto out; } if (len == 0) end = OFF_MAX; else end = offset + len - 1; switch (advice) { case POSIX_FADV_SEQUENTIAL: case POSIX_FADV_RANDOM: case POSIX_FADV_NOREUSE: /* * Try to merge any existing non-standard region with * this new region if possible, otherwise create a new * non-standard region for this request. */ mtx_pool_lock(mtxpool_sleep, fp); fa = fp->f_advice; if (fa != NULL && fa->fa_advice == advice && ((fa->fa_start <= end && fa->fa_end >= offset) || (end != OFF_MAX && fa->fa_start == end + 1) || (fa->fa_end != OFF_MAX && fa->fa_end + 1 == offset))) { if (offset < fa->fa_start) fa->fa_start = offset; if (end > fa->fa_end) fa->fa_end = end; } else { new->fa_advice = advice; new->fa_start = offset; new->fa_end = end; fp->f_advice = new; new = fa; } mtx_pool_unlock(mtxpool_sleep, fp); break; case POSIX_FADV_NORMAL: /* * If a the "normal" region overlaps with an existing * non-standard region, trim or remove the * non-standard region. */ mtx_pool_lock(mtxpool_sleep, fp); fa = fp->f_advice; if (fa != NULL) { if (offset <= fa->fa_start && end >= fa->fa_end) { new = fa; fp->f_advice = NULL; } else if (offset <= fa->fa_start && end >= fa->fa_start) fa->fa_start = end + 1; else if (offset <= fa->fa_end && end >= fa->fa_end) fa->fa_end = offset - 1; else if (offset >= fa->fa_start && end <= fa->fa_end) { /* * If the "normal" region is a middle * portion of the existing * non-standard region, just remove * the whole thing rather than picking * one side or the other to * preserve. */ new = fa; fp->f_advice = NULL; } } mtx_pool_unlock(mtxpool_sleep, fp); break; case POSIX_FADV_WILLNEED: case POSIX_FADV_DONTNEED: error = VOP_ADVISE(vp, offset, end, advice); break; } out: if (fp != NULL) fdrop(fp, td); free(new, M_FADVISE); return (error); } int sys_posix_fadvise(struct thread *td, struct posix_fadvise_args *uap) { int error; error = kern_posix_fadvise(td, uap->fd, uap->offset, uap->len, uap->advice); return (kern_posix_error(td, error)); } Index: projects/clang400-import/sys/mips/include/db_machdep.h =================================================================== --- projects/clang400-import/sys/mips/include/db_machdep.h (revision 311459) +++ projects/clang400-import/sys/mips/include/db_machdep.h (revision 311460) @@ -1,99 +1,98 @@ /* $OpenBSD: db_machdep.h,v 1.2 1998/09/15 10:50:12 pefo Exp $ */ /* * Copyright (c) 1998 Per Fogelstrom, Opsycon AB * * 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 under OpenBSD by * Per Fogelstrom, Opsycon AB, Sweden. * 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. * * JNPR: db_machdep.h,v 1.7 2006/10/16 12:30:34 katta * $FreeBSD$ */ #ifndef _MIPS_DB_MACHDEP_H_ #define _MIPS_DB_MACHDEP_H_ #include #include #include typedef struct trapframe db_regs_t; extern db_regs_t ddb_regs; /* register state */ typedef vm_offset_t db_addr_t; /* address - unsigned */ typedef register_t db_expr_t; /* expression - signed */ #if BYTE_ORDER == _BIG_ENDIAN #define BYTE_MSF (1) #endif #define SOFTWARE_SSTEP /* Need software single step */ #define SOFTWARE_SSTEP_EMUL /* next_instr_address() emulates 100% */ db_addr_t next_instr_address(db_addr_t, boolean_t); #define BKPT_SIZE (4) #define BKPT_SET(ins) (MIPS_BREAK_DDB) #define DB_VALID_BREAKPOINT(addr) (((addr) & 3) == 0) #define IS_BREAKPOINT_TRAP(type, code) ((type) == T_BREAK) #define IS_WATCHPOINT_TRAP(type, code) (0) /* XXX mips3 watchpoint */ #define PC_REGS() ((db_addr_t)kdb_thrctx->pcb_regs.pc) #define BKPT_SKIP \ do { \ if((db_get_value(kdb_frame->pc, sizeof(int), FALSE) & \ ~MIPS_BREAK_VAL_MASK) == MIPS_BREAK_INSTR) { \ kdb_frame->pc += BKPT_SIZE; \ kdb_thrctx->pcb_regs.pc += BKPT_SIZE; \ } \ } while (0); /* * Test of instructions to see class. */ #define IT_CALL 0x01 #define IT_BRANCH 0x02 #define IT_LOAD 0x03 #define IT_STORE 0x04 #define inst_branch(i) (db_inst_type(i) == IT_BRANCH) #define inst_trap_return(i) ((i) & 0) #define inst_call(i) (db_inst_type(i) == IT_CALL) #define inst_return(i) ((i) == 0x03e00008) #define inst_load(i) (db_inst_type(i) == IT_LOAD) #define inst_store(i) (db_inst_type(i) == IT_STORE) #define DB_SMALL_VALUE_MAX 0x7fffffff #define DB_SMALL_VALUE_MIN (-0x400001) int db_inst_type(int); db_addr_t branch_taken(int inst, db_addr_t pc); -void stacktrace_subr(register_t pc, register_t sp, register_t ra, int (*)(const char *, ...)); int32_t kdbpeek(int *); int64_t kdbpeekd(int *); #endif /* !_MIPS_DB_MACHDEP_H_ */ Index: projects/clang400-import/sys/mips/mips/db_trace.c =================================================================== --- projects/clang400-import/sys/mips/mips/db_trace.c (revision 311459) +++ projects/clang400-import/sys/mips/mips/db_trace.c (revision 311460) @@ -1,509 +1,451 @@ /*- * Copyright (c) 2004-2005, Juniper Networks, Inc. * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY 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. * * JNPR: db_trace.c,v 1.8 2007/08/09 11:23:32 katta */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include extern char _locore[]; extern char _locoreEnd[]; extern char edata[]; /* * A function using a stack frame has the following instruction as the first * one: [d]addiu sp,sp,- * * We make use of this to detect starting address of a function. This works * better than using 'j ra' instruction to signify end of the previous * function (for e.g. functions like boot() or panic() do not actually * emit a 'j ra' instruction). * * XXX the abi does not require that the addiu instruction be the first one. */ #define MIPS_START_OF_FUNCTION(ins) ((((ins) & 0xffff8000) == 0x27bd8000) \ || (((ins) & 0xffff8000) == 0x67bd8000)) /* * MIPS ABI 3.0 requires that all functions return using the 'j ra' instruction * * XXX gcc doesn't do this for functions with __noreturn__ attribute. */ #define MIPS_END_OF_FUNCTION(ins) ((ins) == 0x03e00008) #if defined(__mips_n64) # define MIPS_IS_VALID_KERNELADDR(reg) ((((reg) & 3) == 0) && \ ((vm_offset_t)(reg) >= MIPS_XKPHYS_START)) #else # define MIPS_IS_VALID_KERNELADDR(reg) ((((reg) & 3) == 0) && \ ((vm_offset_t)(reg) >= MIPS_KSEG0_START)) #endif -/* - * Functions ``special'' enough to print by name - */ -#ifdef __STDC__ -#define Name(_fn) { (void*)_fn, # _fn } -#else -#define Name(_fn) { _fn, "_fn"} -#endif -static struct { - void *addr; - char *name; -} names[] = { - - Name(trap), - Name(MipsKernGenException), - Name(MipsUserGenException), - Name(MipsKernIntr), - Name(MipsUserIntr), - Name(cpu_switch), - { - 0, 0 - } -}; - -/* - * Map a function address to a string name, if known; or a hex string. - */ -static const char * -fn_name(uintptr_t addr) +static void +stacktrace_subr(register_t pc, register_t sp, register_t ra) { - static char buf[17]; - int i = 0; - - db_expr_t diff; - c_db_sym_t sym; - const char *symname; - - diff = 0; - symname = NULL; - sym = db_search_symbol((db_addr_t)addr, DB_STGY_ANY, &diff); - db_symbol_values(sym, &symname, NULL); - if (symname && diff == 0) - return (symname); - - for (i = 0; names[i].name; i++) - if (names[i].addr == (void *)addr) - return (names[i].name); - sprintf(buf, "%jx", (uintmax_t)addr); - return (buf); -} - -void -stacktrace_subr(register_t pc, register_t sp, register_t ra, - int (*printfn) (const char *,...)) -{ InstFmt i; /* * Arrays for a0..a3 registers and flags if content * of these registers is valid, e.g. obtained from the stack */ int valid_args[4]; register_t args[4]; register_t va, subr, cause, badvaddr; unsigned instr, mask; unsigned int frames = 0; int more, stksize, j; register_t next_ra; bool trapframe; /* Jump here when done with a frame, to start a new one */ loop: /* * Invalidate arguments values */ valid_args[0] = 0; valid_args[1] = 0; valid_args[2] = 0; valid_args[3] = 0; next_ra = 0; stksize = 0; subr = 0; trapframe = false; if (frames++ > 100) { - (*printfn) ("\nstackframe count exceeded\n"); - /* return breaks stackframe-size heuristics with gcc -O2 */ - goto finish; /* XXX */ + db_printf("\nstackframe count exceeded\n"); + return; } - /* check for bad SP: could foul up next frame */ - /*XXX MIPS64 bad: this hard-coded SP is lame */ + + /* Check for bad SP: could foul up next frame. */ if (!MIPS_IS_VALID_KERNELADDR(sp)) { - (*printfn) ("SP 0x%jx: not in kernel\n", sp); + db_printf("SP 0x%jx: not in kernel\n", sp); ra = 0; subr = 0; goto done; } #define Between(x, y, z) \ ( ((x) <= (y)) && ((y) < (z)) ) #define pcBetween(a,b) \ Between((uintptr_t)a, pc, (uintptr_t)b) /* * Check for current PC in exception handler code that don't have a * preceding "j ra" at the tail of the preceding function. Depends * on relative ordering of functions in exception.S, swtch.S. */ if (pcBetween(MipsKernGenException, MipsUserGenException)) { subr = (uintptr_t)MipsKernGenException; trapframe = true; } else if (pcBetween(MipsUserGenException, MipsKernIntr)) subr = (uintptr_t)MipsUserGenException; else if (pcBetween(MipsKernIntr, MipsUserIntr)) { subr = (uintptr_t)MipsKernIntr; trapframe = true; } else if (pcBetween(MipsUserIntr, MipsTLBInvalidException)) subr = (uintptr_t)MipsUserIntr; else if (pcBetween(MipsTLBInvalidException, MipsTLBMissException)) { subr = (uintptr_t)MipsTLBInvalidException; if (pc == (uintptr_t)MipsKStackOverflow) trapframe = true; } else if (pcBetween(fork_trampoline, savectx)) subr = (uintptr_t)fork_trampoline; else if (pcBetween(savectx, cpu_throw)) subr = (uintptr_t)savectx; else if (pcBetween(cpu_throw, cpu_switch)) subr = (uintptr_t)cpu_throw; else if (pcBetween(cpu_switch, MipsSwitchFPState)) subr = (uintptr_t)cpu_switch; else if (pcBetween(_locore, _locoreEnd)) { subr = (uintptr_t)_locore; ra = 0; goto done; } - /* check for bad PC */ - /*XXX MIPS64 bad: These hard coded constants are lame */ + + /* Check for bad PC. */ if (!MIPS_IS_VALID_KERNELADDR(pc)) { - (*printfn) ("PC 0x%jx: not in kernel\n", pc); + db_printf("PC 0x%jx: not in kernel\n", pc); ra = 0; goto done; } /* * For a trapframe, skip to the output and afterwards pull the * previous registers out of the trapframe instead of decoding * the function prologue. */ if (trapframe) goto done; /* * Find the beginning of the current subroutine by scanning * backwards from the current PC for the end of the previous * subroutine. */ if (!subr) { va = pc - sizeof(int); while (1) { instr = kdbpeek((int *)va); if (MIPS_START_OF_FUNCTION(instr)) break; if (MIPS_END_OF_FUNCTION(instr)) { /* skip over branch-delay slot instruction */ va += 2 * sizeof(int); break; } va -= sizeof(int); } /* skip over nulls which might separate .o files */ while ((instr = kdbpeek((int *)va)) == 0) va += sizeof(int); subr = va; } /* scan forwards to find stack size and any saved registers */ stksize = 0; more = 3; mask = 0; for (va = subr; more; va += sizeof(int), more = (more == 3) ? 3 : more - 1) { /* stop if hit our current position */ if (va >= pc) break; instr = kdbpeek((int *)va); i.word = instr; switch (i.JType.op) { case OP_SPECIAL: switch (i.RType.func) { case OP_JR: case OP_JALR: more = 2; /* stop after next instruction */ break; case OP_SYSCALL: case OP_BREAK: more = 1; /* stop now */ } break; case OP_BCOND: case OP_J: case OP_JAL: case OP_BEQ: case OP_BNE: case OP_BLEZ: case OP_BGTZ: more = 2; /* stop after next instruction */ break; case OP_COP0: case OP_COP1: case OP_COP2: case OP_COP3: switch (i.RType.rs) { case OP_BCx: case OP_BCy: more = 2; /* stop after next instruction */ } break; case OP_SW: /* look for saved registers on the stack */ if (i.IType.rs != 29) break; /* * only restore the first one except RA for * MipsKernGenException case */ if (mask & (1 << i.IType.rt)) { if (subr == (uintptr_t)MipsKernGenException && i.IType.rt == 31) next_ra = kdbpeek((int *)(sp + (short)i.IType.imm)); break; } mask |= (1 << i.IType.rt); switch (i.IType.rt) { case 4:/* a0 */ args[0] = kdbpeek((int *)(sp + (short)i.IType.imm)); valid_args[0] = 1; break; case 5:/* a1 */ args[1] = kdbpeek((int *)(sp + (short)i.IType.imm)); valid_args[1] = 1; break; case 6:/* a2 */ args[2] = kdbpeek((int *)(sp + (short)i.IType.imm)); valid_args[2] = 1; break; case 7:/* a3 */ args[3] = kdbpeek((int *)(sp + (short)i.IType.imm)); valid_args[3] = 1; break; case 31: /* ra */ ra = kdbpeek((int *)(sp + (short)i.IType.imm)); } break; case OP_SD: /* look for saved registers on the stack */ if (i.IType.rs != 29) break; /* only restore the first one */ if (mask & (1 << i.IType.rt)) break; mask |= (1 << i.IType.rt); switch (i.IType.rt) { case 4:/* a0 */ args[0] = kdbpeekd((int *)(sp + (short)i.IType.imm)); valid_args[0] = 1; break; case 5:/* a1 */ args[1] = kdbpeekd((int *)(sp + (short)i.IType.imm)); valid_args[1] = 1; break; case 6:/* a2 */ args[2] = kdbpeekd((int *)(sp + (short)i.IType.imm)); valid_args[2] = 1; break; case 7:/* a3 */ args[3] = kdbpeekd((int *)(sp + (short)i.IType.imm)); valid_args[3] = 1; break; case 31: /* ra */ ra = kdbpeekd((int *)(sp + (short)i.IType.imm)); } break; case OP_ADDI: case OP_ADDIU: case OP_DADDI: case OP_DADDIU: /* look for stack pointer adjustment */ if (i.IType.rs != 29 || i.IType.rt != 29) break; stksize = -((short)i.IType.imm); } } done: - (*printfn) ("%s+%x (", fn_name(subr), pc - subr); + db_printsym(pc, DB_STGY_PROC); + db_printf(" ("); for (j = 0; j < 4; j ++) { if (j > 0) - (*printfn)(","); + db_printf(","); if (valid_args[j]) - (*printfn)("%jx", (uintmax_t)(u_register_t)args[j]); + db_printf("%jx", (uintmax_t)(u_register_t)args[j]); else - (*printfn)("?"); + db_printf("?"); } - (*printfn) (") ra %jx sp %jx sz %d\n", + db_printf(") ra %jx sp %jx sz %d\n", (uintmax_t)(u_register_t) ra, (uintmax_t)(u_register_t) sp, stksize); if (trapframe) { #define TF_REG(base, reg) ((base) + CALLFRAME_SIZ + ((reg) * SZREG)) #if defined(__mips_n64) || defined(__mips_n32) pc = kdbpeekd((int *)TF_REG(sp, PC)); ra = kdbpeekd((int *)TF_REG(sp, RA)); sp = kdbpeekd((int *)TF_REG(sp, SP)); cause = kdbpeekd((int *)TF_REG(sp, CAUSE)); badvaddr = kdbpeekd((int *)TF_REG(sp, BADVADDR)); #else pc = kdbpeek((int *)TF_REG(sp, PC)); ra = kdbpeek((int *)TF_REG(sp, RA)); sp = kdbpeek((int *)TF_REG(sp, SP)); cause = kdbpeek((int *)TF_REG(sp, CAUSE)); badvaddr = kdbpeek((int *)TF_REG(sp, BADVADDR)); #endif #undef TF_REG - (*printfn) ("--- exception, cause %jx badvaddr %jx ---\n", + db_printf("--- exception, cause %jx badvaddr %jx ---\n", (uintmax_t)cause, (uintmax_t)badvaddr); goto loop; } else if (ra) { if (pc == ra && stksize == 0) - (*printfn) ("stacktrace: loop!\n"); + db_printf("stacktrace: loop!\n"); else { pc = ra; sp += stksize; ra = next_ra; goto loop; } - } else { -finish: - if (curproc) - (*printfn) ("pid %d\n", curproc->p_pid); - else - (*printfn) ("curproc NULL\n"); } } int db_md_set_watchpoint(db_expr_t addr, db_expr_t size) { return(0); } int db_md_clr_watchpoint(db_expr_t addr, db_expr_t size) { return(0); } void db_md_list_watchpoints() { } void db_trace_self(void) { register_t pc, ra, sp; sp = (register_t)(intptr_t)__builtin_frame_address(0); ra = (register_t)(intptr_t)__builtin_return_address(0); __asm __volatile( "jal 99f\n" "nop\n" "99:\n" "move %0, $31\n" /* get ra */ "move $31, %1\n" /* restore ra */ : "=r" (pc) : "r" (ra)); - stacktrace_subr(pc, sp, ra, db_printf); + stacktrace_subr(pc, sp, ra); return; } int db_trace_thread(struct thread *thr, int count) { register_t pc, ra, sp; struct pcb *ctx; ctx = kdb_thr_ctx(thr); sp = (register_t)ctx->pcb_context[PCB_REG_SP]; pc = (register_t)ctx->pcb_context[PCB_REG_PC]; ra = (register_t)ctx->pcb_context[PCB_REG_RA]; - stacktrace_subr(pc, sp, ra, db_printf); + stacktrace_subr(pc, sp, ra); return (0); } void db_show_mdpcpu(struct pcpu *pc) { db_printf("ipis = 0x%x\n", pc->pc_pending_ipis); db_printf("next ASID = %d\n", pc->pc_next_asid); db_printf("GENID = %d\n", pc->pc_asid_generation); return; } Index: projects/clang400-import/sys/mips/mips/trap.c =================================================================== --- projects/clang400-import/sys/mips/mips/trap.c (revision 311459) +++ projects/clang400-import/sys/mips/mips/trap.c (revision 311460) @@ -1,1737 +1,1719 @@ /* $OpenBSD: trap.c,v 1.19 1998/09/30 12:40:41 pefo Exp $ */ /* tracked to 1.23 */ /*- * Copyright (c) 1988 University of Utah. * Copyright (c) 1992, 1993 * The Regents of the University of California. All rights reserved. * * This code is derived from software contributed to Berkeley by * the Systems Programming Group of the University of Utah Computer * Science Department and Ralph Campbell. * * 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: Utah Hdr: trap.c 1.32 91/04/06 * * from: @(#)trap.c 8.5 (Berkeley) 1/11/94 * JNPR: trap.c,v 1.13.2.2 2007/08/29 10:03:49 girish */ #include __FBSDID("$FreeBSD$"); #include "opt_compat.h" #include "opt_ddb.h" #include "opt_ktrace.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef KTRACE #include #endif #include #include #include #include #include #include #include #include #include #include #ifdef DDB #include #include #include #include #endif #ifdef KDTRACE_HOOKS #include #endif #ifdef TRAP_DEBUG int trap_debug = 0; SYSCTL_INT(_machdep, OID_AUTO, trap_debug, CTLFLAG_RW, &trap_debug, 0, "Debug information on all traps"); #endif #define lbu_macro(data, addr) \ __asm __volatile ("lbu %0, 0x0(%1)" \ : "=r" (data) /* outputs */ \ : "r" (addr)); /* inputs */ #define lb_macro(data, addr) \ __asm __volatile ("lb %0, 0x0(%1)" \ : "=r" (data) /* outputs */ \ : "r" (addr)); /* inputs */ #define lwl_macro(data, addr) \ __asm __volatile ("lwl %0, 0x0(%1)" \ : "=r" (data) /* outputs */ \ : "r" (addr)); /* inputs */ #define lwr_macro(data, addr) \ __asm __volatile ("lwr %0, 0x0(%1)" \ : "=r" (data) /* outputs */ \ : "r" (addr)); /* inputs */ #define ldl_macro(data, addr) \ __asm __volatile ("ldl %0, 0x0(%1)" \ : "=r" (data) /* outputs */ \ : "r" (addr)); /* inputs */ #define ldr_macro(data, addr) \ __asm __volatile ("ldr %0, 0x0(%1)" \ : "=r" (data) /* outputs */ \ : "r" (addr)); /* inputs */ #define sb_macro(data, addr) \ __asm __volatile ("sb %0, 0x0(%1)" \ : /* outputs */ \ : "r" (data), "r" (addr)); /* inputs */ #define swl_macro(data, addr) \ __asm __volatile ("swl %0, 0x0(%1)" \ : /* outputs */ \ : "r" (data), "r" (addr)); /* inputs */ #define swr_macro(data, addr) \ __asm __volatile ("swr %0, 0x0(%1)" \ : /* outputs */ \ : "r" (data), "r" (addr)); /* inputs */ #define sdl_macro(data, addr) \ __asm __volatile ("sdl %0, 0x0(%1)" \ : /* outputs */ \ : "r" (data), "r" (addr)); /* inputs */ #define sdr_macro(data, addr) \ __asm __volatile ("sdr %0, 0x0(%1)" \ : /* outputs */ \ : "r" (data), "r" (addr)); /* inputs */ static void log_illegal_instruction(const char *, struct trapframe *); static void log_bad_page_fault(char *, struct trapframe *, int); static void log_frame_dump(struct trapframe *frame); static void get_mapping_info(vm_offset_t, pd_entry_t **, pt_entry_t **); int (*dtrace_invop_jump_addr)(struct trapframe *); #ifdef TRAP_DEBUG static void trap_frame_dump(struct trapframe *frame); #endif void (*machExceptionTable[]) (void)= { /* * The kernel exception handlers. */ MipsKernIntr, /* external interrupt */ MipsKernGenException, /* TLB modification */ MipsTLBInvalidException,/* TLB miss (load or instr. fetch) */ MipsTLBInvalidException,/* TLB miss (store) */ MipsKernGenException, /* address error (load or I-fetch) */ MipsKernGenException, /* address error (store) */ MipsKernGenException, /* bus error (I-fetch) */ MipsKernGenException, /* bus error (load or store) */ MipsKernGenException, /* system call */ MipsKernGenException, /* breakpoint */ MipsKernGenException, /* reserved instruction */ MipsKernGenException, /* coprocessor unusable */ MipsKernGenException, /* arithmetic overflow */ MipsKernGenException, /* trap exception */ MipsKernGenException, /* virtual coherence exception inst */ MipsKernGenException, /* floating point exception */ MipsKernGenException, /* reserved */ MipsKernGenException, /* reserved */ MipsKernGenException, /* reserved */ MipsKernGenException, /* reserved */ MipsKernGenException, /* reserved */ MipsKernGenException, /* reserved */ MipsKernGenException, /* reserved */ MipsKernGenException, /* watch exception */ MipsKernGenException, /* reserved */ MipsKernGenException, /* reserved */ MipsKernGenException, /* reserved */ MipsKernGenException, /* reserved */ MipsKernGenException, /* reserved */ MipsKernGenException, /* reserved */ MipsKernGenException, /* reserved */ MipsKernGenException, /* virtual coherence exception data */ /* * The user exception handlers. */ MipsUserIntr, /* 0 */ MipsUserGenException, /* 1 */ MipsTLBInvalidException,/* 2 */ MipsTLBInvalidException,/* 3 */ MipsUserGenException, /* 4 */ MipsUserGenException, /* 5 */ MipsUserGenException, /* 6 */ MipsUserGenException, /* 7 */ MipsUserGenException, /* 8 */ MipsUserGenException, /* 9 */ MipsUserGenException, /* 10 */ MipsUserGenException, /* 11 */ MipsUserGenException, /* 12 */ MipsUserGenException, /* 13 */ MipsUserGenException, /* 14 */ MipsUserGenException, /* 15 */ MipsUserGenException, /* 16 */ MipsUserGenException, /* 17 */ MipsUserGenException, /* 18 */ MipsUserGenException, /* 19 */ MipsUserGenException, /* 20 */ MipsUserGenException, /* 21 */ MipsUserGenException, /* 22 */ MipsUserGenException, /* 23 */ MipsUserGenException, /* 24 */ MipsUserGenException, /* 25 */ MipsUserGenException, /* 26 */ MipsUserGenException, /* 27 */ MipsUserGenException, /* 28 */ MipsUserGenException, /* 29 */ MipsUserGenException, /* 20 */ MipsUserGenException, /* 31 */ }; char *trap_type[] = { "external interrupt", "TLB modification", "TLB miss (load or instr. fetch)", "TLB miss (store)", "address error (load or I-fetch)", "address error (store)", "bus error (I-fetch)", "bus error (load or store)", "system call", "breakpoint", "reserved instruction", "coprocessor unusable", "arithmetic overflow", "trap", "virtual coherency instruction", "floating point", "reserved 16", "reserved 17", "reserved 18", "reserved 19", "reserved 20", "reserved 21", "reserved 22", "watch", "reserved 24", "reserved 25", "reserved 26", "reserved 27", "reserved 28", "reserved 29", "reserved 30", "virtual coherency data", }; #if !defined(SMP) && (defined(DDB) || defined(DEBUG)) struct trapdebug trapdebug[TRAPSIZE], *trp = trapdebug; #endif -#if defined(DDB) || defined(DEBUG) -void stacktrace(struct trapframe *); -void logstacktrace(struct trapframe *); -#endif - #define KERNLAND(x) ((vm_offset_t)(x) >= VM_MIN_KERNEL_ADDRESS && (vm_offset_t)(x) < VM_MAX_KERNEL_ADDRESS) #define DELAYBRANCH(x) ((int)(x) < 0) /* * MIPS load/store access type */ enum { MIPS_LHU_ACCESS = 1, MIPS_LH_ACCESS, MIPS_LWU_ACCESS, MIPS_LW_ACCESS, MIPS_LD_ACCESS, MIPS_SH_ACCESS, MIPS_SW_ACCESS, MIPS_SD_ACCESS }; char *access_name[] = { "Load Halfword Unsigned", "Load Halfword", "Load Word Unsigned", "Load Word", "Load Doubleword", "Store Halfword", "Store Word", "Store Doubleword" }; #ifdef CPU_CNMIPS #include #endif static int allow_unaligned_acc = 1; SYSCTL_INT(_vm, OID_AUTO, allow_unaligned_acc, CTLFLAG_RW, &allow_unaligned_acc, 0, "Allow unaligned accesses"); /* * FP emulation is assumed to work on O32, but the code is outdated and crufty * enough that it's a more sensible default to have it disabled when using * other ABIs. At the very least, it needs a lot of help in using * type-semantic ABI-oblivious macros for everything it does. */ #if defined(__mips_o32) static int emulate_fp = 1; #else static int emulate_fp = 0; #endif SYSCTL_INT(_machdep, OID_AUTO, emulate_fp, CTLFLAG_RW, &emulate_fp, 0, "Emulate unimplemented FPU instructions"); static int emulate_unaligned_access(struct trapframe *frame, int mode); extern void fswintrberr(void); /* XXX */ int cpu_fetch_syscall_args(struct thread *td, struct syscall_args *sa) { struct trapframe *locr0 = td->td_frame; struct sysentvec *se; int error, nsaved; bzero(sa->args, sizeof(sa->args)); /* compute next PC after syscall instruction */ td->td_pcb->pcb_tpc = sa->trapframe->pc; /* Remember if restart */ if (DELAYBRANCH(sa->trapframe->cause)) /* Check BD bit */ locr0->pc = MipsEmulateBranch(locr0, sa->trapframe->pc, 0, 0); else locr0->pc += sizeof(int); sa->code = locr0->v0; switch (sa->code) { case SYS___syscall: case SYS_syscall: /* * This is an indirect syscall, in which the code is the first argument. */ #if (!defined(__mips_n32) && !defined(__mips_n64)) || defined(COMPAT_FREEBSD32) if (sa->code == SYS___syscall && SV_PROC_FLAG(td->td_proc, SV_ILP32)) { /* * Like syscall, but code is a quad, so as to maintain alignment * for the rest of the arguments. */ if (_QUAD_LOWWORD == 0) sa->code = locr0->a0; else sa->code = locr0->a1; sa->args[0] = locr0->a2; sa->args[1] = locr0->a3; nsaved = 2; break; } #endif /* * This is either not a quad syscall, or is a quad syscall with a * new ABI in which quads fit in a single register. */ sa->code = locr0->a0; sa->args[0] = locr0->a1; sa->args[1] = locr0->a2; sa->args[2] = locr0->a3; nsaved = 3; #if defined(__mips_n32) || defined(__mips_n64) #ifdef COMPAT_FREEBSD32 if (!SV_PROC_FLAG(td->td_proc, SV_ILP32)) { #endif /* * Non-o32 ABIs support more arguments in registers. */ sa->args[3] = locr0->a4; sa->args[4] = locr0->a5; sa->args[5] = locr0->a6; sa->args[6] = locr0->a7; nsaved += 4; #ifdef COMPAT_FREEBSD32 } #endif #endif break; default: /* * A direct syscall, arguments are just parameters to the syscall. */ sa->args[0] = locr0->a0; sa->args[1] = locr0->a1; sa->args[2] = locr0->a2; sa->args[3] = locr0->a3; nsaved = 4; #if defined (__mips_n32) || defined(__mips_n64) #ifdef COMPAT_FREEBSD32 if (!SV_PROC_FLAG(td->td_proc, SV_ILP32)) { #endif /* * Non-o32 ABIs support more arguments in registers. */ sa->args[4] = locr0->a4; sa->args[5] = locr0->a5; sa->args[6] = locr0->a6; sa->args[7] = locr0->a7; nsaved += 4; #ifdef COMPAT_FREEBSD32 } #endif #endif break; } #ifdef TRAP_DEBUG if (trap_debug) printf("SYSCALL #%d pid:%u\n", sa->code, td->td_proc->p_pid); #endif se = td->td_proc->p_sysent; /* * XXX * Shouldn't this go before switching on the code? */ if (se->sv_mask) sa->code &= se->sv_mask; if (sa->code >= se->sv_size) sa->callp = &se->sv_table[0]; else sa->callp = &se->sv_table[sa->code]; sa->narg = sa->callp->sy_narg; if (sa->narg > nsaved) { #if defined(__mips_n32) || defined(__mips_n64) /* * XXX * Is this right for new ABIs? I think the 4 there * should be 8, size there are 8 registers to skip, * not 4, but I'm not certain. */ #ifdef COMPAT_FREEBSD32 if (!SV_PROC_FLAG(td->td_proc, SV_ILP32)) #endif printf("SYSCALL #%u pid:%u, narg (%u) > nsaved (%u).\n", sa->code, td->td_proc->p_pid, sa->narg, nsaved); #endif #if (defined(__mips_n32) || defined(__mips_n64)) && defined(COMPAT_FREEBSD32) if (SV_PROC_FLAG(td->td_proc, SV_ILP32)) { unsigned i; int32_t arg; error = 0; /* XXX GCC is awful. */ for (i = nsaved; i < sa->narg; i++) { error = copyin((caddr_t)(intptr_t)(locr0->sp + (4 + (i - nsaved)) * sizeof(int32_t)), (caddr_t)&arg, sizeof arg); if (error != 0) break; sa->args[i] = arg; } } else #endif error = copyin((caddr_t)(intptr_t)(locr0->sp + 4 * sizeof(register_t)), (caddr_t)&sa->args[nsaved], (u_int)(sa->narg - nsaved) * sizeof(register_t)); if (error != 0) { locr0->v0 = error; locr0->a3 = 1; } } else error = 0; if (error == 0) { td->td_retval[0] = 0; td->td_retval[1] = locr0->v1; } return (error); } #undef __FBSDID #define __FBSDID(x) #include "../../kern/subr_syscall.c" /* * Handle an exception. * Called from MipsKernGenException() or MipsUserGenException() * when a processor trap occurs. * In the case of a kernel trap, we return the pc where to resume if * p->p_addr->u_pcb.pcb_onfault is set, otherwise, return old pc. */ register_t trap(struct trapframe *trapframe) { int type, usermode; int i = 0; unsigned ucode = 0; struct thread *td = curthread; struct proc *p = curproc; vm_prot_t ftype; pmap_t pmap; int access_type; ksiginfo_t ksi; char *msg = NULL; intptr_t addr = 0; register_t pc; int cop; register_t *frame_regs; trapdebug_enter(trapframe, 0); type = (trapframe->cause & MIPS_CR_EXC_CODE) >> MIPS_CR_EXC_CODE_SHIFT; if (TRAPF_USERMODE(trapframe)) { type |= T_USER; usermode = 1; } else { usermode = 0; } /* * Enable hardware interrupts if they were on before the trap. If it * was off disable all so we don't accidently enable it when doing a * return to userland. */ if (trapframe->sr & MIPS_SR_INT_IE) { set_intr_mask(trapframe->sr & MIPS_SR_INT_MASK); intr_enable(); } else { intr_disable(); } #ifdef TRAP_DEBUG if (trap_debug) { static vm_offset_t last_badvaddr = 0; static vm_offset_t this_badvaddr = 0; static int count = 0; u_int32_t pid; printf("trap type %x (%s - ", type, trap_type[type & (~T_USER)]); if (type & T_USER) printf("user mode)\n"); else printf("kernel mode)\n"); #ifdef SMP printf("cpuid = %d\n", PCPU_GET(cpuid)); #endif pid = mips_rd_entryhi() & TLBHI_ASID_MASK; printf("badaddr = %#jx, pc = %#jx, ra = %#jx, sp = %#jx, sr = %jx, pid = %d, ASID = %u\n", (intmax_t)trapframe->badvaddr, (intmax_t)trapframe->pc, (intmax_t)trapframe->ra, (intmax_t)trapframe->sp, (intmax_t)trapframe->sr, (curproc ? curproc->p_pid : -1), pid); switch (type & ~T_USER) { case T_TLB_MOD: case T_TLB_LD_MISS: case T_TLB_ST_MISS: case T_ADDR_ERR_LD: case T_ADDR_ERR_ST: this_badvaddr = trapframe->badvaddr; break; case T_SYSCALL: this_badvaddr = trapframe->ra; break; default: this_badvaddr = trapframe->pc; break; } if ((last_badvaddr == this_badvaddr) && ((type & ~T_USER) != T_SYSCALL) && ((type & ~T_USER) != T_COP_UNUSABLE)) { if (++count == 3) { trap_frame_dump(trapframe); panic("too many faults at %p\n", (void *)last_badvaddr); } } else { last_badvaddr = this_badvaddr; count = 0; } } #endif #ifdef KDTRACE_HOOKS /* * A trap can occur while DTrace executes a probe. Before * executing the probe, DTrace blocks re-scheduling and sets * a flag in its per-cpu flags to indicate that it doesn't * want to fault. On returning from the probe, the no-fault * flag is cleared and finally re-scheduling is enabled. * * If the DTrace kernel module has registered a trap handler, * call it and if it returns non-zero, assume that it has * handled the trap and modified the trap frame so that this * function can return normally. */ /* * XXXDTRACE: add pid probe handler here (if ever) */ if (!usermode) { if (dtrace_trap_func != NULL && (*dtrace_trap_func)(trapframe, type) != 0) return (trapframe->pc); } #endif switch (type) { case T_MCHECK: #ifdef DDB kdb_trap(type, 0, trapframe); #endif panic("MCHECK\n"); break; case T_TLB_MOD: /* check for kernel address */ if (KERNLAND(trapframe->badvaddr)) { if (pmap_emulate_modified(kernel_pmap, trapframe->badvaddr) != 0) { ftype = VM_PROT_WRITE; goto kernel_fault; } return (trapframe->pc); } /* FALLTHROUGH */ case T_TLB_MOD + T_USER: pmap = &p->p_vmspace->vm_pmap; if (pmap_emulate_modified(pmap, trapframe->badvaddr) != 0) { ftype = VM_PROT_WRITE; goto dofault; } if (!usermode) return (trapframe->pc); goto out; case T_TLB_LD_MISS: case T_TLB_ST_MISS: ftype = (type == T_TLB_ST_MISS) ? VM_PROT_WRITE : VM_PROT_READ; /* check for kernel address */ if (KERNLAND(trapframe->badvaddr)) { vm_offset_t va; int rv; kernel_fault: va = trunc_page((vm_offset_t)trapframe->badvaddr); rv = vm_fault(kernel_map, va, ftype, VM_FAULT_NORMAL); if (rv == KERN_SUCCESS) return (trapframe->pc); if (td->td_pcb->pcb_onfault != NULL) { pc = (register_t)(intptr_t)td->td_pcb->pcb_onfault; td->td_pcb->pcb_onfault = NULL; return (pc); } goto err; } /* * It is an error for the kernel to access user space except * through the copyin/copyout routines. */ if (td->td_pcb->pcb_onfault == NULL) goto err; /* check for fuswintr() or suswintr() getting a page fault */ /* XXX There must be a nicer way to do this. */ if (td->td_pcb->pcb_onfault == fswintrberr) { pc = (register_t)(intptr_t)td->td_pcb->pcb_onfault; td->td_pcb->pcb_onfault = NULL; return (pc); } goto dofault; case T_TLB_LD_MISS + T_USER: ftype = VM_PROT_READ; goto dofault; case T_TLB_ST_MISS + T_USER: ftype = VM_PROT_WRITE; dofault: { vm_offset_t va; struct vmspace *vm; vm_map_t map; int rv = 0; vm = p->p_vmspace; map = &vm->vm_map; va = trunc_page((vm_offset_t)trapframe->badvaddr); if (KERNLAND(trapframe->badvaddr)) { /* * Don't allow user-mode faults in kernel * address space. */ goto nogo; } rv = vm_fault(map, va, ftype, VM_FAULT_NORMAL); /* * XXXDTRACE: add dtrace_doubletrap_func here? */ #ifdef VMFAULT_TRACE printf("vm_fault(%p (pmap %p), %p (%p), %x, %d) -> %x at pc %p\n", map, &vm->vm_pmap, (void *)va, (void *)(intptr_t)trapframe->badvaddr, ftype, VM_FAULT_NORMAL, rv, (void *)(intptr_t)trapframe->pc); #endif if (rv == KERN_SUCCESS) { if (!usermode) { return (trapframe->pc); } goto out; } nogo: if (!usermode) { if (td->td_pcb->pcb_onfault != NULL) { pc = (register_t)(intptr_t)td->td_pcb->pcb_onfault; td->td_pcb->pcb_onfault = NULL; return (pc); } goto err; } i = SIGSEGV; if (rv == KERN_PROTECTION_FAILURE) ucode = SEGV_ACCERR; else ucode = SEGV_MAPERR; addr = trapframe->pc; msg = "BAD_PAGE_FAULT"; log_bad_page_fault(msg, trapframe, type); break; } case T_ADDR_ERR_LD + T_USER: /* misaligned or kseg access */ case T_ADDR_ERR_ST + T_USER: /* misaligned or kseg access */ if (trapframe->badvaddr < 0 || trapframe->badvaddr >= VM_MAXUSER_ADDRESS) { msg = "ADDRESS_SPACE_ERR"; } else if (allow_unaligned_acc) { int mode; if (type == (T_ADDR_ERR_LD + T_USER)) mode = VM_PROT_READ; else mode = VM_PROT_WRITE; access_type = emulate_unaligned_access(trapframe, mode); if (access_type != 0) goto out; msg = "ALIGNMENT_FIX_ERR"; } else { msg = "ADDRESS_ERR"; } /* FALL THROUGH */ case T_BUS_ERR_IFETCH + T_USER: /* BERR asserted to cpu */ case T_BUS_ERR_LD_ST + T_USER: /* BERR asserted to cpu */ ucode = 0; /* XXX should be VM_PROT_something */ i = SIGBUS; addr = trapframe->pc; if (!msg) msg = "BUS_ERR"; log_bad_page_fault(msg, trapframe, type); break; case T_SYSCALL + T_USER: { struct syscall_args sa; int error; sa.trapframe = trapframe; error = syscallenter(td, &sa); #if !defined(SMP) && (defined(DDB) || defined(DEBUG)) if (trp == trapdebug) trapdebug[TRAPSIZE - 1].code = sa.code; else trp[-1].code = sa.code; #endif trapdebug_enter(td->td_frame, -sa.code); /* * The sync'ing of I & D caches for SYS_ptrace() is * done by procfs_domem() through procfs_rwmem() * instead of being done here under a special check * for SYS_ptrace(). */ syscallret(td, error, &sa); return (trapframe->pc); } #if defined(KDTRACE_HOOKS) || defined(DDB) case T_BREAK: #ifdef KDTRACE_HOOKS if (!usermode && dtrace_invop_jump_addr != 0) { dtrace_invop_jump_addr(trapframe); return (trapframe->pc); } #endif #ifdef DDB kdb_trap(type, 0, trapframe); return (trapframe->pc); #endif #endif case T_BREAK + T_USER: { intptr_t va; uint32_t instr; /* compute address of break instruction */ va = trapframe->pc; if (DELAYBRANCH(trapframe->cause)) va += sizeof(int); /* read break instruction */ instr = fuword32((caddr_t)va); #if 0 printf("trap: %s (%d) breakpoint %x at %x: (adr %x ins %x)\n", p->p_comm, p->p_pid, instr, trapframe->pc, p->p_md.md_ss_addr, p->p_md.md_ss_instr); /* XXX */ #endif if (td->td_md.md_ss_addr != va || instr != MIPS_BREAK_SSTEP) { i = SIGTRAP; addr = trapframe->pc; break; } /* * The restoration of the original instruction and * the clearing of the breakpoint will be done later * by the call to ptrace_clear_single_step() in * issignal() when SIGTRAP is processed. */ addr = trapframe->pc; i = SIGTRAP; break; } case T_IWATCH + T_USER: case T_DWATCH + T_USER: { intptr_t va; /* compute address of trapped instruction */ va = trapframe->pc; if (DELAYBRANCH(trapframe->cause)) va += sizeof(int); printf("watch exception @ %p\n", (void *)va); i = SIGTRAP; addr = va; break; } case T_TRAP + T_USER: { intptr_t va; uint32_t instr; struct trapframe *locr0 = td->td_frame; /* compute address of trap instruction */ va = trapframe->pc; if (DELAYBRANCH(trapframe->cause)) va += sizeof(int); /* read break instruction */ instr = fuword32((caddr_t)va); if (DELAYBRANCH(trapframe->cause)) { /* Check BD bit */ locr0->pc = MipsEmulateBranch(locr0, trapframe->pc, 0, 0); } else { locr0->pc += sizeof(int); } addr = va; i = SIGEMT; /* Stuff it with something for now */ break; } case T_RES_INST + T_USER: { InstFmt inst; inst = *(InstFmt *)(intptr_t)trapframe->pc; switch (inst.RType.op) { case OP_SPECIAL3: switch (inst.RType.func) { case OP_RDHWR: /* Register 29 used for TLS */ if (inst.RType.rd == 29) { frame_regs = &(trapframe->zero); frame_regs[inst.RType.rt] = (register_t)(intptr_t)td->td_md.md_tls; frame_regs[inst.RType.rt] += td->td_md.md_tls_tcb_offset; trapframe->pc += sizeof(int); goto out; } break; } break; } log_illegal_instruction("RES_INST", trapframe); i = SIGILL; addr = trapframe->pc; } break; case T_C2E: case T_C2E + T_USER: goto err; break; case T_COP_UNUSABLE: #ifdef CPU_CNMIPS cop = (trapframe->cause & MIPS_CR_COP_ERR) >> MIPS_CR_COP_ERR_SHIFT; /* Handle only COP2 exception */ if (cop != 2) goto err; addr = trapframe->pc; /* save userland cop2 context if it has been touched */ if ((td->td_md.md_flags & MDTD_COP2USED) && (td->td_md.md_cop2owner == COP2_OWNER_USERLAND)) { if (td->td_md.md_ucop2) octeon_cop2_save(td->td_md.md_ucop2); else panic("COP2 was used in user mode but md_ucop2 is NULL"); } if (td->td_md.md_cop2 == NULL) { td->td_md.md_cop2 = octeon_cop2_alloc_ctx(); if (td->td_md.md_cop2 == NULL) panic("Failed to allocate COP2 context"); memset(td->td_md.md_cop2, 0, sizeof(*td->td_md.md_cop2)); } octeon_cop2_restore(td->td_md.md_cop2); /* Make userland re-request its context */ td->td_frame->sr &= ~MIPS_SR_COP_2_BIT; td->td_md.md_flags |= MDTD_COP2USED; td->td_md.md_cop2owner = COP2_OWNER_KERNEL; /* Enable COP2, it will be disabled in cpu_switch */ mips_wr_status(mips_rd_status() | MIPS_SR_COP_2_BIT); return (trapframe->pc); #else goto err; break; #endif case T_COP_UNUSABLE + T_USER: cop = (trapframe->cause & MIPS_CR_COP_ERR) >> MIPS_CR_COP_ERR_SHIFT; if (cop == 1) { #if !defined(CPU_HAVEFPU) /* FP (COP1) instruction */ log_illegal_instruction("COP1_UNUSABLE", trapframe); i = SIGILL; break; #else addr = trapframe->pc; MipsSwitchFPState(PCPU_GET(fpcurthread), td->td_frame); PCPU_SET(fpcurthread, td); #if defined(__mips_n64) td->td_frame->sr |= MIPS_SR_COP_1_BIT | MIPS_SR_FR; #else td->td_frame->sr |= MIPS_SR_COP_1_BIT; #endif td->td_md.md_flags |= MDTD_FPUSED; goto out; #endif } #ifdef CPU_CNMIPS else if (cop == 2) { addr = trapframe->pc; if ((td->td_md.md_flags & MDTD_COP2USED) && (td->td_md.md_cop2owner == COP2_OWNER_KERNEL)) { if (td->td_md.md_cop2) octeon_cop2_save(td->td_md.md_cop2); else panic("COP2 was used in kernel mode but md_cop2 is NULL"); } if (td->td_md.md_ucop2 == NULL) { td->td_md.md_ucop2 = octeon_cop2_alloc_ctx(); if (td->td_md.md_ucop2 == NULL) panic("Failed to allocate userland COP2 context"); memset(td->td_md.md_ucop2, 0, sizeof(*td->td_md.md_ucop2)); } octeon_cop2_restore(td->td_md.md_ucop2); td->td_frame->sr |= MIPS_SR_COP_2_BIT; td->td_md.md_flags |= MDTD_COP2USED; td->td_md.md_cop2owner = COP2_OWNER_USERLAND; goto out; } #endif else { log_illegal_instruction("COPn_UNUSABLE", trapframe); i = SIGILL; /* only FPU instructions allowed */ break; } case T_FPE: #if !defined(SMP) && (defined(DDB) || defined(DEBUG)) trapDump("fpintr"); #else printf("FPU Trap: PC %#jx CR %x SR %x\n", (intmax_t)trapframe->pc, (unsigned)trapframe->cause, (unsigned)trapframe->sr); goto err; #endif case T_FPE + T_USER: if (!emulate_fp) { i = SIGFPE; addr = trapframe->pc; break; } MipsFPTrap(trapframe->sr, trapframe->cause, trapframe->pc); goto out; case T_OVFLOW + T_USER: i = SIGFPE; addr = trapframe->pc; break; case T_ADDR_ERR_LD: /* misaligned access */ case T_ADDR_ERR_ST: /* misaligned access */ #ifdef TRAP_DEBUG if (trap_debug) { printf("+++ ADDR_ERR: type = %d, badvaddr = %#jx\n", type, (intmax_t)trapframe->badvaddr); } #endif /* Only allow emulation on a user address */ if (allow_unaligned_acc && ((vm_offset_t)trapframe->badvaddr < VM_MAXUSER_ADDRESS)) { int mode; if (type == T_ADDR_ERR_LD) mode = VM_PROT_READ; else mode = VM_PROT_WRITE; access_type = emulate_unaligned_access(trapframe, mode); if (access_type != 0) return (trapframe->pc); } /* FALLTHROUGH */ case T_BUS_ERR_LD_ST: /* BERR asserted to cpu */ if (td->td_pcb->pcb_onfault != NULL) { pc = (register_t)(intptr_t)td->td_pcb->pcb_onfault; td->td_pcb->pcb_onfault = NULL; return (pc); } /* FALLTHROUGH */ default: err: #if !defined(SMP) && defined(DEBUG) - stacktrace(!usermode ? trapframe : td->td_frame); trapDump("trap"); #endif #ifdef SMP printf("cpu:%d-", PCPU_GET(cpuid)); #endif printf("Trap cause = %d (%s - ", type, trap_type[type & (~T_USER)]); if (type & T_USER) printf("user mode)\n"); else printf("kernel mode)\n"); #ifdef TRAP_DEBUG if (trap_debug) printf("badvaddr = %#jx, pc = %#jx, ra = %#jx, sr = %#jxx\n", (intmax_t)trapframe->badvaddr, (intmax_t)trapframe->pc, (intmax_t)trapframe->ra, (intmax_t)trapframe->sr); #endif #ifdef KDB if (debugger_on_panic || kdb_active) { kdb_trap(type, 0, trapframe); } #endif panic("trap"); } td->td_frame->pc = trapframe->pc; td->td_frame->cause = trapframe->cause; td->td_frame->badvaddr = trapframe->badvaddr; ksiginfo_init_trap(&ksi); ksi.ksi_signo = i; ksi.ksi_code = ucode; ksi.ksi_addr = (void *)addr; ksi.ksi_trapno = type; trapsignal(td, &ksi); out: /* * Note: we should only get here if returning to user mode. */ userret(td, trapframe); return (trapframe->pc); } #if !defined(SMP) && (defined(DDB) || defined(DEBUG)) void trapDump(char *msg) { register_t s; int i; s = intr_disable(); printf("trapDump(%s)\n", msg); for (i = 0; i < TRAPSIZE; i++) { if (trp == trapdebug) { trp = &trapdebug[TRAPSIZE - 1]; } else { trp--; } if (trp->cause == 0) break; printf("%s: ADR %jx PC %jx CR %jx SR %jx\n", trap_type[(trp->cause & MIPS_CR_EXC_CODE) >> MIPS_CR_EXC_CODE_SHIFT], (intmax_t)trp->vadr, (intmax_t)trp->pc, (intmax_t)trp->cause, (intmax_t)trp->status); printf(" RA %jx SP %jx code %d\n", (intmax_t)trp->ra, (intmax_t)trp->sp, (int)trp->code); } intr_restore(s); } #endif /* * Return the resulting PC as if the branch was executed. */ uintptr_t MipsEmulateBranch(struct trapframe *framePtr, uintptr_t instPC, int fpcCSR, uintptr_t instptr) { InstFmt inst; register_t *regsPtr = (register_t *) framePtr; uintptr_t retAddr = 0; int condition; #define GetBranchDest(InstPtr, inst) \ (InstPtr + 4 + ((short)inst.IType.imm << 2)) if (instptr) { if (instptr < MIPS_KSEG0_START) inst.word = fuword32((void *)instptr); else inst = *(InstFmt *) instptr; } else { if ((vm_offset_t)instPC < MIPS_KSEG0_START) inst.word = fuword32((void *)instPC); else inst = *(InstFmt *) instPC; } switch ((int)inst.JType.op) { case OP_SPECIAL: switch ((int)inst.RType.func) { case OP_JR: case OP_JALR: retAddr = regsPtr[inst.RType.rs]; break; default: retAddr = instPC + 4; break; } break; case OP_BCOND: switch ((int)inst.IType.rt) { case OP_BLTZ: case OP_BLTZL: case OP_BLTZAL: case OP_BLTZALL: if ((int)(regsPtr[inst.RType.rs]) < 0) retAddr = GetBranchDest(instPC, inst); else retAddr = instPC + 8; break; case OP_BGEZ: case OP_BGEZL: case OP_BGEZAL: case OP_BGEZALL: if ((int)(regsPtr[inst.RType.rs]) >= 0) retAddr = GetBranchDest(instPC, inst); else retAddr = instPC + 8; break; case OP_TGEI: case OP_TGEIU: case OP_TLTI: case OP_TLTIU: case OP_TEQI: case OP_TNEI: retAddr = instPC + 4; /* Like syscall... */ break; default: panic("MipsEmulateBranch: Bad branch cond"); } break; case OP_J: case OP_JAL: retAddr = (inst.JType.target << 2) | ((unsigned)(instPC + 4) & 0xF0000000); break; case OP_BEQ: case OP_BEQL: if (regsPtr[inst.RType.rs] == regsPtr[inst.RType.rt]) retAddr = GetBranchDest(instPC, inst); else retAddr = instPC + 8; break; case OP_BNE: case OP_BNEL: if (regsPtr[inst.RType.rs] != regsPtr[inst.RType.rt]) retAddr = GetBranchDest(instPC, inst); else retAddr = instPC + 8; break; case OP_BLEZ: case OP_BLEZL: if ((int)(regsPtr[inst.RType.rs]) <= 0) retAddr = GetBranchDest(instPC, inst); else retAddr = instPC + 8; break; case OP_BGTZ: case OP_BGTZL: if ((int)(regsPtr[inst.RType.rs]) > 0) retAddr = GetBranchDest(instPC, inst); else retAddr = instPC + 8; break; case OP_COP1: switch (inst.RType.rs) { case OP_BCx: case OP_BCy: if ((inst.RType.rt & COPz_BC_TF_MASK) == COPz_BC_TRUE) condition = fpcCSR & MIPS_FPU_COND_BIT; else condition = !(fpcCSR & MIPS_FPU_COND_BIT); if (condition) retAddr = GetBranchDest(instPC, inst); else retAddr = instPC + 8; break; default: retAddr = instPC + 4; } break; default: retAddr = instPC + 4; } return (retAddr); } - - -#if defined(DDB) || defined(DEBUG) -/* - * Print a stack backtrace. - */ -void -stacktrace(struct trapframe *regs) -{ - stacktrace_subr(regs->pc, regs->sp, regs->ra, printf); -} -#endif static void log_frame_dump(struct trapframe *frame) { log(LOG_ERR, "Trapframe Register Dump:\n"); log(LOG_ERR, "\tzero: %#jx\tat: %#jx\tv0: %#jx\tv1: %#jx\n", (intmax_t)0, (intmax_t)frame->ast, (intmax_t)frame->v0, (intmax_t)frame->v1); log(LOG_ERR, "\ta0: %#jx\ta1: %#jx\ta2: %#jx\ta3: %#jx\n", (intmax_t)frame->a0, (intmax_t)frame->a1, (intmax_t)frame->a2, (intmax_t)frame->a3); #if defined(__mips_n32) || defined(__mips_n64) log(LOG_ERR, "\ta4: %#jx\ta5: %#jx\ta6: %#jx\ta6: %#jx\n", (intmax_t)frame->a4, (intmax_t)frame->a5, (intmax_t)frame->a6, (intmax_t)frame->a7); log(LOG_ERR, "\tt0: %#jx\tt1: %#jx\tt2: %#jx\tt3: %#jx\n", (intmax_t)frame->t0, (intmax_t)frame->t1, (intmax_t)frame->t2, (intmax_t)frame->t3); #else log(LOG_ERR, "\tt0: %#jx\tt1: %#jx\tt2: %#jx\tt3: %#jx\n", (intmax_t)frame->t0, (intmax_t)frame->t1, (intmax_t)frame->t2, (intmax_t)frame->t3); log(LOG_ERR, "\tt4: %#jx\tt5: %#jx\tt6: %#jx\tt7: %#jx\n", (intmax_t)frame->t4, (intmax_t)frame->t5, (intmax_t)frame->t6, (intmax_t)frame->t7); #endif log(LOG_ERR, "\tt8: %#jx\tt9: %#jx\ts0: %#jx\ts1: %#jx\n", (intmax_t)frame->t8, (intmax_t)frame->t9, (intmax_t)frame->s0, (intmax_t)frame->s1); log(LOG_ERR, "\ts2: %#jx\ts3: %#jx\ts4: %#jx\ts5: %#jx\n", (intmax_t)frame->s2, (intmax_t)frame->s3, (intmax_t)frame->s4, (intmax_t)frame->s5); log(LOG_ERR, "\ts6: %#jx\ts7: %#jx\tk0: %#jx\tk1: %#jx\n", (intmax_t)frame->s6, (intmax_t)frame->s7, (intmax_t)frame->k0, (intmax_t)frame->k1); log(LOG_ERR, "\tgp: %#jx\tsp: %#jx\ts8: %#jx\tra: %#jx\n", (intmax_t)frame->gp, (intmax_t)frame->sp, (intmax_t)frame->s8, (intmax_t)frame->ra); log(LOG_ERR, "\tsr: %#jx\tmullo: %#jx\tmulhi: %#jx\tbadvaddr: %#jx\n", (intmax_t)frame->sr, (intmax_t)frame->mullo, (intmax_t)frame->mulhi, (intmax_t)frame->badvaddr); #ifdef IC_REG log(LOG_ERR, "\tcause: %#jx\tpc: %#jx\tic: %#jx\n", (intmax_t)frame->cause, (intmax_t)frame->pc, (intmax_t)frame->ic); #else log(LOG_ERR, "\tcause: %#jx\tpc: %#jx\n", (intmax_t)frame->cause, (intmax_t)frame->pc); #endif } #ifdef TRAP_DEBUG static void trap_frame_dump(struct trapframe *frame) { printf("Trapframe Register Dump:\n"); printf("\tzero: %#jx\tat: %#jx\tv0: %#jx\tv1: %#jx\n", (intmax_t)0, (intmax_t)frame->ast, (intmax_t)frame->v0, (intmax_t)frame->v1); printf("\ta0: %#jx\ta1: %#jx\ta2: %#jx\ta3: %#jx\n", (intmax_t)frame->a0, (intmax_t)frame->a1, (intmax_t)frame->a2, (intmax_t)frame->a3); #if defined(__mips_n32) || defined(__mips_n64) printf("\ta4: %#jx\ta5: %#jx\ta6: %#jx\ta7: %#jx\n", (intmax_t)frame->a4, (intmax_t)frame->a5, (intmax_t)frame->a6, (intmax_t)frame->a7); printf("\tt0: %#jx\tt1: %#jx\tt2: %#jx\tt3: %#jx\n", (intmax_t)frame->t0, (intmax_t)frame->t1, (intmax_t)frame->t2, (intmax_t)frame->t3); #else printf("\tt0: %#jx\tt1: %#jx\tt2: %#jx\tt3: %#jx\n", (intmax_t)frame->t0, (intmax_t)frame->t1, (intmax_t)frame->t2, (intmax_t)frame->t3); printf("\tt4: %#jx\tt5: %#jx\tt6: %#jx\tt7: %#jx\n", (intmax_t)frame->t4, (intmax_t)frame->t5, (intmax_t)frame->t6, (intmax_t)frame->t7); #endif printf("\tt8: %#jx\tt9: %#jx\ts0: %#jx\ts1: %#jx\n", (intmax_t)frame->t8, (intmax_t)frame->t9, (intmax_t)frame->s0, (intmax_t)frame->s1); printf("\ts2: %#jx\ts3: %#jx\ts4: %#jx\ts5: %#jx\n", (intmax_t)frame->s2, (intmax_t)frame->s3, (intmax_t)frame->s4, (intmax_t)frame->s5); printf("\ts6: %#jx\ts7: %#jx\tk0: %#jx\tk1: %#jx\n", (intmax_t)frame->s6, (intmax_t)frame->s7, (intmax_t)frame->k0, (intmax_t)frame->k1); printf("\tgp: %#jx\tsp: %#jx\ts8: %#jx\tra: %#jx\n", (intmax_t)frame->gp, (intmax_t)frame->sp, (intmax_t)frame->s8, (intmax_t)frame->ra); printf("\tsr: %#jx\tmullo: %#jx\tmulhi: %#jx\tbadvaddr: %#jx\n", (intmax_t)frame->sr, (intmax_t)frame->mullo, (intmax_t)frame->mulhi, (intmax_t)frame->badvaddr); #ifdef IC_REG printf("\tcause: %#jx\tpc: %#jx\tic: %#jx\n", (intmax_t)frame->cause, (intmax_t)frame->pc, (intmax_t)frame->ic); #else printf("\tcause: %#jx\tpc: %#jx\n", (intmax_t)frame->cause, (intmax_t)frame->pc); #endif } #endif static void get_mapping_info(vm_offset_t va, pd_entry_t **pdepp, pt_entry_t **ptepp) { pt_entry_t *ptep; pd_entry_t *pdep; struct proc *p = curproc; pdep = (&(p->p_vmspace->vm_pmap.pm_segtab[(va >> SEGSHIFT) & (NPDEPG - 1)])); if (*pdep) ptep = pmap_pte(&p->p_vmspace->vm_pmap, va); else ptep = (pt_entry_t *)0; *pdepp = pdep; *ptepp = ptep; } static void log_illegal_instruction(const char *msg, struct trapframe *frame) { pt_entry_t *ptep; pd_entry_t *pdep; unsigned int *addr; struct thread *td; struct proc *p; register_t pc; td = curthread; p = td->td_proc; #ifdef SMP printf("cpuid = %d\n", PCPU_GET(cpuid)); #endif pc = frame->pc + (DELAYBRANCH(frame->cause) ? 4 : 0); log(LOG_ERR, "%s: pid %d tid %ld (%s), uid %d: pc %#jx ra %#jx\n", msg, p->p_pid, (long)td->td_tid, p->p_comm, p->p_ucred ? p->p_ucred->cr_uid : -1, (intmax_t)pc, (intmax_t)frame->ra); /* log registers in trap frame */ log_frame_dump(frame); get_mapping_info((vm_offset_t)pc, &pdep, &ptep); /* * Dump a few words around faulting instruction, if the addres is * valid. */ if (!(pc & 3) && useracc((caddr_t)(intptr_t)pc, sizeof(int) * 4, VM_PROT_READ)) { /* dump page table entry for faulting instruction */ log(LOG_ERR, "Page table info for pc address %#jx: pde = %p, pte = %#jx\n", (intmax_t)pc, (void *)(intptr_t)*pdep, (uintmax_t)(ptep ? *ptep : 0)); addr = (unsigned int *)(intptr_t)pc; log(LOG_ERR, "Dumping 4 words starting at pc address %p: \n", addr); log(LOG_ERR, "%08x %08x %08x %08x\n", addr[0], addr[1], addr[2], addr[3]); } else { log(LOG_ERR, "pc address %#jx is inaccessible, pde = %p, pte = %#jx\n", (intmax_t)pc, (void *)(intptr_t)*pdep, (uintmax_t)(ptep ? *ptep : 0)); } } static void log_bad_page_fault(char *msg, struct trapframe *frame, int trap_type) { pt_entry_t *ptep; pd_entry_t *pdep; unsigned int *addr; struct thread *td; struct proc *p; char *read_or_write; register_t pc; trap_type &= ~T_USER; td = curthread; p = td->td_proc; #ifdef SMP printf("cpuid = %d\n", PCPU_GET(cpuid)); #endif switch (trap_type) { case T_TLB_MOD: case T_TLB_ST_MISS: case T_ADDR_ERR_ST: read_or_write = "write"; break; case T_TLB_LD_MISS: case T_ADDR_ERR_LD: case T_BUS_ERR_IFETCH: read_or_write = "read"; break; default: read_or_write = "unknown"; } pc = frame->pc + (DELAYBRANCH(frame->cause) ? 4 : 0); log(LOG_ERR, "%s: pid %d tid %ld (%s), uid %d: pc %#jx got a %s fault " "(type %#x) at %#jx\n", msg, p->p_pid, (long)td->td_tid, p->p_comm, p->p_ucred ? p->p_ucred->cr_uid : -1, (intmax_t)pc, read_or_write, trap_type, (intmax_t)frame->badvaddr); /* log registers in trap frame */ log_frame_dump(frame); get_mapping_info((vm_offset_t)pc, &pdep, &ptep); /* * Dump a few words around faulting instruction, if the addres is * valid. */ if (!(pc & 3) && (pc != frame->badvaddr) && (trap_type != T_BUS_ERR_IFETCH) && useracc((caddr_t)(intptr_t)pc, sizeof(int) * 4, VM_PROT_READ)) { /* dump page table entry for faulting instruction */ log(LOG_ERR, "Page table info for pc address %#jx: pde = %p, pte = %#jx\n", (intmax_t)pc, (void *)(intptr_t)*pdep, (uintmax_t)(ptep ? *ptep : 0)); addr = (unsigned int *)(intptr_t)pc; log(LOG_ERR, "Dumping 4 words starting at pc address %p: \n", addr); log(LOG_ERR, "%08x %08x %08x %08x\n", addr[0], addr[1], addr[2], addr[3]); } else { log(LOG_ERR, "pc address %#jx is inaccessible, pde = %p, pte = %#jx\n", (intmax_t)pc, (void *)(intptr_t)*pdep, (uintmax_t)(ptep ? *ptep : 0)); } get_mapping_info((vm_offset_t)frame->badvaddr, &pdep, &ptep); log(LOG_ERR, "Page table info for bad address %#jx: pde = %p, pte = %#jx\n", (intmax_t)frame->badvaddr, (void *)(intptr_t)*pdep, (uintmax_t)(ptep ? *ptep : 0)); } /* * Unaligned load/store emulation */ static int mips_unaligned_load_store(struct trapframe *frame, int mode, register_t addr, register_t pc) { register_t *reg = (register_t *) frame; u_int32_t inst = *((u_int32_t *)(intptr_t)pc); register_t value_msb, value; unsigned size; /* * ADDR_ERR faults have higher priority than TLB * Miss faults. Therefore, it is necessary to * verify that the faulting address is a valid * virtual address within the process' address space * before trying to emulate the unaligned access. */ switch (MIPS_INST_OPCODE(inst)) { case OP_LHU: case OP_LH: case OP_SH: size = 2; break; case OP_LWU: case OP_LW: case OP_SW: size = 4; break; case OP_LD: case OP_SD: size = 8; break; default: printf("%s: unhandled opcode in address error: %#x\n", __func__, MIPS_INST_OPCODE(inst)); return (0); } if (!useracc((void *)rounddown2((vm_offset_t)addr, size), size * 2, mode)) return (0); /* * XXX * Handle LL/SC LLD/SCD. */ switch (MIPS_INST_OPCODE(inst)) { case OP_LHU: KASSERT(mode == VM_PROT_READ, ("access mode must be read for load instruction.")); lbu_macro(value_msb, addr); addr += 1; lbu_macro(value, addr); value |= value_msb << 8; reg[MIPS_INST_RT(inst)] = value; return (MIPS_LHU_ACCESS); case OP_LH: KASSERT(mode == VM_PROT_READ, ("access mode must be read for load instruction.")); lb_macro(value_msb, addr); addr += 1; lbu_macro(value, addr); value |= value_msb << 8; reg[MIPS_INST_RT(inst)] = value; return (MIPS_LH_ACCESS); case OP_LWU: KASSERT(mode == VM_PROT_READ, ("access mode must be read for load instruction.")); lwl_macro(value, addr); addr += 3; lwr_macro(value, addr); value &= 0xffffffff; reg[MIPS_INST_RT(inst)] = value; return (MIPS_LWU_ACCESS); case OP_LW: KASSERT(mode == VM_PROT_READ, ("access mode must be read for load instruction.")); lwl_macro(value, addr); addr += 3; lwr_macro(value, addr); reg[MIPS_INST_RT(inst)] = value; return (MIPS_LW_ACCESS); #if defined(__mips_n32) || defined(__mips_n64) case OP_LD: KASSERT(mode == VM_PROT_READ, ("access mode must be read for load instruction.")); ldl_macro(value, addr); addr += 7; ldr_macro(value, addr); reg[MIPS_INST_RT(inst)] = value; return (MIPS_LD_ACCESS); #endif case OP_SH: KASSERT(mode == VM_PROT_WRITE, ("access mode must be write for store instruction.")); value = reg[MIPS_INST_RT(inst)]; value_msb = value >> 8; sb_macro(value_msb, addr); addr += 1; sb_macro(value, addr); return (MIPS_SH_ACCESS); case OP_SW: KASSERT(mode == VM_PROT_WRITE, ("access mode must be write for store instruction.")); value = reg[MIPS_INST_RT(inst)]; swl_macro(value, addr); addr += 3; swr_macro(value, addr); return (MIPS_SW_ACCESS); #if defined(__mips_n32) || defined(__mips_n64) case OP_SD: KASSERT(mode == VM_PROT_WRITE, ("access mode must be write for store instruction.")); value = reg[MIPS_INST_RT(inst)]; sdl_macro(value, addr); addr += 7; sdr_macro(value, addr); return (MIPS_SD_ACCESS); #endif } panic("%s: should not be reached.", __func__); } /* * XXX TODO: SMP? */ static struct timeval unaligned_lasterr; static int unaligned_curerr; static int unaligned_pps_log_limit = 4; SYSCTL_INT(_machdep, OID_AUTO, unaligned_log_pps_limit, CTLFLAG_RWTUN, &unaligned_pps_log_limit, 0, "limit number of userland unaligned log messages per second"); static int emulate_unaligned_access(struct trapframe *frame, int mode) { register_t pc; int access_type = 0; struct thread *td = curthread; struct proc *p = curproc; pc = frame->pc + (DELAYBRANCH(frame->cause) ? 4 : 0); /* * Fall through if it's instruction fetch exception */ if (!((pc & 3) || (pc == frame->badvaddr))) { /* * Handle unaligned load and store */ /* * Return access type if the instruction was emulated. * Otherwise restore pc and fall through. */ access_type = mips_unaligned_load_store(frame, mode, frame->badvaddr, pc); if (access_type) { if (DELAYBRANCH(frame->cause)) frame->pc = MipsEmulateBranch(frame, frame->pc, 0, 0); else frame->pc += 4; if (ppsratecheck(&unaligned_lasterr, &unaligned_curerr, unaligned_pps_log_limit)) { /* XXX TODO: keep global/tid/pid counters? */ log(LOG_INFO, "Unaligned %s: pid=%ld (%s), tid=%ld, " "pc=%#jx, badvaddr=%#jx\n", access_name[access_type - 1], (long) p->p_pid, p->p_comm, (long) td->td_tid, (intmax_t)pc, (intmax_t)frame->badvaddr); } } } return access_type; } Index: projects/clang400-import/sys/net/if_media.h =================================================================== --- projects/clang400-import/sys/net/if_media.h (revision 311459) +++ projects/clang400-import/sys/net/if_media.h (revision 311460) @@ -1,838 +1,842 @@ /* $NetBSD: if_media.h,v 1.3 1997/03/26 01:19:27 thorpej Exp $ */ /* $FreeBSD$ */ /*- * Copyright (c) 1997 * Jonathan Stone and Jason R. Thorpe. All rights reserved. * * This software is derived from information provided by Matt Thomas. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. All advertising materials mentioning features or use of this software * must display the following acknowledgement: * This product includes software developed by Jonathan Stone * and Jason R. Thorpe for the NetBSD Project. * 4. The names of the authors may not be used to endorse or promote products * derived from this software without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE AUTHORS ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #ifndef _NET_IF_MEDIA_H_ #define _NET_IF_MEDIA_H_ /* * Prototypes and definitions for BSD/OS-compatible network interface * media selection. * * Where it is safe to do so, this code strays slightly from the BSD/OS * design. Software which uses the API (device drivers, basically) * shouldn't notice any difference. * * Many thanks to Matt Thomas for providing the information necessary * to implement this interface. */ #ifdef _KERNEL #include struct ifnet; /* * Driver callbacks for media status and change requests. */ typedef int (*ifm_change_cb_t)(struct ifnet *); typedef void (*ifm_stat_cb_t)(struct ifnet *, struct ifmediareq *req); /* * In-kernel representation of a single supported media type. */ struct ifmedia_entry { LIST_ENTRY(ifmedia_entry) ifm_list; int ifm_media; /* description of this media attachment */ int ifm_data; /* for driver-specific use */ void *ifm_aux; /* for driver-specific use */ }; /* * One of these goes into a network interface's softc structure. * It is used to keep general media state. */ struct ifmedia { int ifm_mask; /* mask of changes we don't care about */ int ifm_media; /* current user-set media word */ struct ifmedia_entry *ifm_cur; /* currently selected media */ LIST_HEAD(, ifmedia_entry) ifm_list; /* list of all supported media */ ifm_change_cb_t ifm_change; /* media change driver callback */ ifm_stat_cb_t ifm_status; /* media status driver callback */ }; /* Initialize an interface's struct if_media field. */ void ifmedia_init(struct ifmedia *ifm, int dontcare_mask, ifm_change_cb_t change_callback, ifm_stat_cb_t status_callback); /* Remove all mediums from a struct ifmedia. */ void ifmedia_removeall( struct ifmedia *ifm); /* Add one supported medium to a struct ifmedia. */ void ifmedia_add(struct ifmedia *ifm, int mword, int data, void *aux); /* Add an array (of ifmedia_entry) media to a struct ifmedia. */ void ifmedia_list_add(struct ifmedia *mp, struct ifmedia_entry *lp, int count); /* Set default media type on initialization. */ void ifmedia_set(struct ifmedia *ifm, int mword); /* Common ioctl function for getting/setting media, called by driver. */ int ifmedia_ioctl(struct ifnet *ifp, struct ifreq *ifr, struct ifmedia *ifm, u_long cmd); /* Compute baudrate for a given media. */ uint64_t ifmedia_baudrate(int); #endif /*_KERNEL */ /* * if_media Options word: * Bits Use * ---- ------- * 0-4 Media variant * 5-7 Media type * 8-15 Type specific options (includes added variant bits on Ethernet) * 16-18 Mode (for multi-mode devices) * 19 RFU * 20-27 Shared (global) options * 28-31 Instance */ /* * Ethernet * In order to use more than 31 subtypes, Ethernet uses some of the option * bits as part of the subtype field. See the options section below for * relevant definitions */ #define IFM_ETHER 0x00000020 #define IFM_ETHER_SUBTYPE(x) (((x) & IFM_TMASK) | \ (((x) & (IFM_ETH_XTYPE >> IFM_ETH_XSHIFT)) << IFM_ETH_XSHIFT)) #define IFM_X(x) IFM_ETHER_SUBTYPE(x) /* internal shorthand */ #define IFM_ETHER_SUBTYPE_SET(x) (IFM_ETHER_SUBTYPE(x) | IFM_ETHER) #define IFM_ETHER_SUBTYPE_GET(x) ((x) & (IFM_TMASK|IFM_ETH_XTYPE)) #define IFM_ETHER_IS_EXTENDED(x) ((x) & IFM_ETH_XTYPE) #define IFM_10_T 3 /* 10BaseT - RJ45 */ #define IFM_10_2 4 /* 10Base2 - Thinnet */ #define IFM_10_5 5 /* 10Base5 - AUI */ #define IFM_100_TX 6 /* 100BaseTX - RJ45 */ #define IFM_100_FX 7 /* 100BaseFX - Fiber */ #define IFM_100_T4 8 /* 100BaseT4 - 4 pair cat 3 */ #define IFM_100_VG 9 /* 100VG-AnyLAN */ #define IFM_100_T2 10 /* 100BaseT2 */ #define IFM_1000_SX 11 /* 1000BaseSX - multi-mode fiber */ #define IFM_10_STP 12 /* 10BaseT over shielded TP */ #define IFM_10_FL 13 /* 10BaseFL - Fiber */ #define IFM_1000_LX 14 /* 1000baseLX - single-mode fiber */ #define IFM_1000_CX 15 /* 1000baseCX - 150ohm STP */ #define IFM_1000_T 16 /* 1000baseT - 4 pair cat 5 */ #define IFM_HPNA_1 17 /* HomePNA 1.0 (1Mb/s) */ #define IFM_10G_LR 18 /* 10GBase-LR 1310nm Single-mode */ #define IFM_10G_SR 19 /* 10GBase-SR 850nm Multi-mode */ #define IFM_10G_CX4 20 /* 10GBase CX4 copper */ #define IFM_2500_SX 21 /* 2500BaseSX - multi-mode fiber */ #define IFM_10G_TWINAX 22 /* 10GBase Twinax copper */ #define IFM_10G_TWINAX_LONG 23 /* 10GBase Twinax Long copper */ #define IFM_10G_LRM 24 /* 10GBase-LRM 850nm Multi-mode */ #define IFM_UNKNOWN 25 /* media types not defined yet */ #define IFM_10G_T 26 /* 10GBase-T - RJ45 */ #define IFM_40G_CR4 27 /* 40GBase-CR4 */ #define IFM_40G_SR4 28 /* 40GBase-SR4 */ #define IFM_40G_LR4 29 /* 40GBase-LR4 */ #define IFM_1000_KX 30 /* 1000Base-KX backplane */ #define IFM_OTHER 31 /* Other: one of the following */ /* following types are not visible to old binaries using only IFM_TMASK */ #define IFM_10G_KX4 IFM_X(32) /* 10GBase-KX4 backplane */ #define IFM_10G_KR IFM_X(33) /* 10GBase-KR backplane */ #define IFM_10G_CR1 IFM_X(34) /* 10GBase-CR1 Twinax splitter */ #define IFM_20G_KR2 IFM_X(35) /* 20GBase-KR2 backplane */ #define IFM_2500_KX IFM_X(36) /* 2500Base-KX backplane */ #define IFM_2500_T IFM_X(37) /* 2500Base-T - RJ45 (NBaseT) */ #define IFM_5000_T IFM_X(38) /* 5000Base-T - RJ45 (NBaseT) */ #define IFM_50G_PCIE IFM_X(39) /* 50G Ethernet over PCIE */ #define IFM_25G_PCIE IFM_X(40) /* 25G Ethernet over PCIE */ #define IFM_1000_SGMII IFM_X(41) /* 1G media interface */ #define IFM_10G_SFI IFM_X(42) /* 10G media interface */ #define IFM_40G_XLPPI IFM_X(43) /* 40G media interface */ #define IFM_1000_CX_SGMII IFM_X(44) /* 1000Base-CX-SGMII */ #define IFM_40G_KR4 IFM_X(45) /* 40GBase-KR4 */ #define IFM_10G_ER IFM_X(46) /* 10GBase-ER */ #define IFM_100G_CR4 IFM_X(47) /* 100GBase-CR4 */ #define IFM_100G_SR4 IFM_X(48) /* 100GBase-SR4 */ #define IFM_100G_KR4 IFM_X(49) /* 100GBase-KR4 */ #define IFM_100G_LR4 IFM_X(50) /* 100GBase-LR4 */ #define IFM_56G_R4 IFM_X(51) /* 56GBase-R4 */ #define IFM_100_T IFM_X(52) /* 100BaseT - RJ45 */ #define IFM_25G_CR IFM_X(53) /* 25GBase-CR */ #define IFM_25G_KR IFM_X(54) /* 25GBase-KR */ #define IFM_25G_SR IFM_X(55) /* 25GBase-SR */ #define IFM_50G_CR2 IFM_X(56) /* 50GBase-CR2 */ #define IFM_50G_KR2 IFM_X(57) /* 50GBase-KR2 */ /* * Please update ieee8023ad_lacp.c:lacp_compose_key() * after adding new Ethernet media types. */ /* Note IFM_X(511) is the max! */ /* Ethernet option values; includes bits used for extended variant field */ #define IFM_ETH_MASTER 0x00000100 /* master mode (1000baseT) */ #define IFM_ETH_RXPAUSE 0x00000200 /* receive PAUSE frames */ #define IFM_ETH_TXPAUSE 0x00000400 /* transmit PAUSE frames */ #define IFM_ETH_XTYPE 0x00007800 /* extended media variants */ #define IFM_ETH_XSHIFT 6 /* shift XTYPE next to TMASK */ /* * Token ring */ #define IFM_TOKEN 0x00000040 #define IFM_TOK_STP4 3 /* Shielded twisted pair 4m - DB9 */ #define IFM_TOK_STP16 4 /* Shielded twisted pair 16m - DB9 */ #define IFM_TOK_UTP4 5 /* Unshielded twisted pair 4m - RJ45 */ #define IFM_TOK_UTP16 6 /* Unshielded twisted pair 16m - RJ45 */ #define IFM_TOK_STP100 7 /* Shielded twisted pair 100m - DB9 */ #define IFM_TOK_UTP100 8 /* Unshielded twisted pair 100m - RJ45 */ #define IFM_TOK_ETR 0x00000200 /* Early token release */ #define IFM_TOK_SRCRT 0x00000400 /* Enable source routing features */ #define IFM_TOK_ALLR 0x00000800 /* All routes / Single route bcast */ #define IFM_TOK_DTR 0x00002000 /* Dedicated token ring */ #define IFM_TOK_CLASSIC 0x00004000 /* Classic token ring */ #define IFM_TOK_AUTO 0x00008000 /* Automatic Dedicate/Classic token ring */ /* * FDDI */ #define IFM_FDDI 0x00000060 #define IFM_FDDI_SMF 3 /* Single-mode fiber */ #define IFM_FDDI_MMF 4 /* Multi-mode fiber */ #define IFM_FDDI_UTP 5 /* CDDI / UTP */ #define IFM_FDDI_DA 0x00000100 /* Dual attach / single attach */ /* * IEEE 802.11 Wireless */ #define IFM_IEEE80211 0x00000080 /* NB: 0,1,2 are auto, manual, none defined below */ #define IFM_IEEE80211_FH1 3 /* Frequency Hopping 1Mbps */ #define IFM_IEEE80211_FH2 4 /* Frequency Hopping 2Mbps */ #define IFM_IEEE80211_DS1 5 /* Direct Sequence 1Mbps */ #define IFM_IEEE80211_DS2 6 /* Direct Sequence 2Mbps */ #define IFM_IEEE80211_DS5 7 /* Direct Sequence 5.5Mbps */ #define IFM_IEEE80211_DS11 8 /* Direct Sequence 11Mbps */ #define IFM_IEEE80211_DS22 9 /* Direct Sequence 22Mbps */ #define IFM_IEEE80211_OFDM6 10 /* OFDM 6Mbps */ #define IFM_IEEE80211_OFDM9 11 /* OFDM 9Mbps */ #define IFM_IEEE80211_OFDM12 12 /* OFDM 12Mbps */ #define IFM_IEEE80211_OFDM18 13 /* OFDM 18Mbps */ #define IFM_IEEE80211_OFDM24 14 /* OFDM 24Mbps */ #define IFM_IEEE80211_OFDM36 15 /* OFDM 36Mbps */ #define IFM_IEEE80211_OFDM48 16 /* OFDM 48Mbps */ #define IFM_IEEE80211_OFDM54 17 /* OFDM 54Mbps */ #define IFM_IEEE80211_OFDM72 18 /* OFDM 72Mbps */ #define IFM_IEEE80211_DS354k 19 /* Direct Sequence 354Kbps */ #define IFM_IEEE80211_DS512k 20 /* Direct Sequence 512Kbps */ #define IFM_IEEE80211_OFDM3 21 /* OFDM 3Mbps */ #define IFM_IEEE80211_OFDM4 22 /* OFDM 4.5Mbps */ #define IFM_IEEE80211_OFDM27 23 /* OFDM 27Mbps */ /* NB: not enough bits to express MCS fully */ #define IFM_IEEE80211_MCS 24 /* HT MCS rate */ #define IFM_IEEE80211_VHT 25 /* HT MCS rate */ #define IFM_IEEE80211_ADHOC 0x00000100 /* Operate in Adhoc mode */ #define IFM_IEEE80211_HOSTAP 0x00000200 /* Operate in Host AP mode */ #define IFM_IEEE80211_IBSS 0x00000400 /* Operate in IBSS mode */ #define IFM_IEEE80211_WDS 0x00000800 /* Operate in WDS mode */ #define IFM_IEEE80211_TURBO 0x00001000 /* Operate in turbo mode */ #define IFM_IEEE80211_MONITOR 0x00002000 /* Operate in monitor mode */ #define IFM_IEEE80211_MBSS 0x00004000 /* Operate in MBSS mode */ /* operating mode for multi-mode devices */ #define IFM_IEEE80211_11A 0x00010000 /* 5Ghz, OFDM mode */ #define IFM_IEEE80211_11B 0x00020000 /* Direct Sequence mode */ #define IFM_IEEE80211_11G 0x00030000 /* 2Ghz, CCK mode */ #define IFM_IEEE80211_FH 0x00040000 /* 2Ghz, GFSK mode */ #define IFM_IEEE80211_11NA 0x00050000 /* 5Ghz, HT mode */ #define IFM_IEEE80211_11NG 0x00060000 /* 2Ghz, HT mode */ #define IFM_IEEE80211_VHT5G 0x00070000 /* 5Ghz, VHT mode */ #define IFM_IEEE80211_VHT2G 0x00080000 /* 2Ghz, VHT mode */ /* * ATM */ #define IFM_ATM 0x000000a0 #define IFM_ATM_UNKNOWN 3 #define IFM_ATM_UTP_25 4 #define IFM_ATM_TAXI_100 5 #define IFM_ATM_TAXI_140 6 #define IFM_ATM_MM_155 7 #define IFM_ATM_SM_155 8 #define IFM_ATM_UTP_155 9 #define IFM_ATM_MM_622 10 #define IFM_ATM_SM_622 11 #define IFM_ATM_VIRTUAL 12 #define IFM_ATM_SDH 0x00000100 /* SDH instead of SONET */ #define IFM_ATM_NOSCRAMB 0x00000200 /* no scrambling */ #define IFM_ATM_UNASSIGNED 0x00000400 /* unassigned cells */ /* * Shared media sub-types */ #define IFM_AUTO 0 /* Autoselect best media */ #define IFM_MANUAL 1 /* Jumper/dipswitch selects media */ #define IFM_NONE 2 /* Deselect all media */ /* * Shared options */ #define IFM_FDX 0x00100000 /* Force full duplex */ #define IFM_HDX 0x00200000 /* Force half duplex */ #define IFM_FLOW 0x00400000 /* enable hardware flow control */ #define IFM_FLAG0 0x01000000 /* Driver defined flag */ #define IFM_FLAG1 0x02000000 /* Driver defined flag */ #define IFM_FLAG2 0x04000000 /* Driver defined flag */ #define IFM_LOOP 0x08000000 /* Put hardware in loopback */ /* * Masks */ #define IFM_NMASK 0x000000e0 /* Network type */ #define IFM_TMASK 0x0000001f /* Media sub-type */ #define IFM_IMASK 0xf0000000 /* Instance */ #define IFM_ISHIFT 28 /* Instance shift */ #define IFM_OMASK 0x0000ff00 /* Type specific options */ #define IFM_MMASK 0x00070000 /* Mode */ #define IFM_MSHIFT 16 /* Mode shift */ #define IFM_GMASK 0x0ff00000 /* Global options */ /* Ethernet flow control mask */ #define IFM_ETH_FMASK (IFM_FLOW | IFM_ETH_RXPAUSE | IFM_ETH_TXPAUSE) /* * Status bits */ #define IFM_AVALID 0x00000001 /* Active bit valid */ #define IFM_ACTIVE 0x00000002 /* Interface attached to working net */ /* Mask of "status valid" bits, for ifconfig(8). */ #define IFM_STATUS_VALID IFM_AVALID /* List of "status valid" bits, for ifconfig(8). */ #define IFM_STATUS_VALID_LIST { \ IFM_AVALID, \ 0 \ } /* * Macros to extract various bits of information from the media word. */ #define IFM_TYPE(x) ((x) & IFM_NMASK) #define IFM_SUBTYPE(x) \ (IFM_TYPE(x) == IFM_ETHER ? IFM_ETHER_SUBTYPE_GET(x) : ((x) & IFM_TMASK)) #define IFM_TYPE_MATCH(x,y) \ (IFM_TYPE(x) == IFM_TYPE(y) && IFM_SUBTYPE(x) == IFM_SUBTYPE(y)) #define IFM_TYPE_OPTIONS(x) ((x) & IFM_OMASK) #define IFM_INST(x) (((x) & IFM_IMASK) >> IFM_ISHIFT) #define IFM_OPTIONS(x) ((x) & (IFM_OMASK | IFM_GMASK)) #define IFM_MODE(x) ((x) & IFM_MMASK) #define IFM_INST_MAX IFM_INST(IFM_IMASK) /* * Macro to create a media word. */ #define IFM_MAKEWORD(type, subtype, options, instance) \ ((type) | (subtype) | (options) | ((instance) << IFM_ISHIFT)) #define IFM_MAKEMODE(mode) \ (((mode) << IFM_MSHIFT) & IFM_MMASK) /* * NetBSD extension not defined in the BSDI API. This is used in various * places to get the canonical description for a given type/subtype. * * NOTE: all but the top-level type descriptions must contain NO whitespace! * Otherwise, parsing these in ifconfig(8) would be a nightmare. */ struct ifmedia_description { int ifmt_word; /* word value; may be masked */ const char *ifmt_string; /* description */ }; #define IFM_TYPE_DESCRIPTIONS { \ { IFM_ETHER, "Ethernet" }, \ { IFM_TOKEN, "Token ring" }, \ { IFM_FDDI, "FDDI" }, \ { IFM_IEEE80211, "IEEE 802.11 Wireless Ethernet" }, \ { IFM_ATM, "ATM" }, \ { 0, NULL }, \ } #define IFM_SUBTYPE_ETHERNET_DESCRIPTIONS { \ { IFM_10_T, "10baseT/UTP" }, \ { IFM_10_2, "10base2/BNC" }, \ { IFM_10_5, "10base5/AUI" }, \ { IFM_100_TX, "100baseTX" }, \ { IFM_100_FX, "100baseFX" }, \ { IFM_100_T4, "100baseT4" }, \ { IFM_100_VG, "100baseVG" }, \ { IFM_100_T2, "100baseT2" }, \ { IFM_10_STP, "10baseSTP" }, \ { IFM_10_FL, "10baseFL" }, \ { IFM_1000_SX, "1000baseSX" }, \ { IFM_1000_LX, "1000baseLX" }, \ { IFM_1000_CX, "1000baseCX" }, \ { IFM_1000_T, "1000baseT" }, \ { IFM_HPNA_1, "homePNA" }, \ { IFM_10G_LR, "10Gbase-LR" }, \ { IFM_10G_SR, "10Gbase-SR" }, \ { IFM_10G_CX4, "10Gbase-CX4" }, \ { IFM_2500_SX, "2500BaseSX" }, \ { IFM_10G_LRM, "10Gbase-LRM" }, \ { IFM_10G_TWINAX, "10Gbase-Twinax" }, \ { IFM_10G_TWINAX_LONG, "10Gbase-Twinax-Long" }, \ { IFM_UNKNOWN, "Unknown" }, \ { IFM_10G_T, "10Gbase-T" }, \ { IFM_40G_CR4, "40Gbase-CR4" }, \ { IFM_40G_SR4, "40Gbase-SR4" }, \ { IFM_40G_LR4, "40Gbase-LR4" }, \ { IFM_1000_KX, "1000Base-KX" }, \ { IFM_OTHER, "Other" }, \ { IFM_10G_KX4, "10GBase-KX4" }, \ { IFM_10G_KR, "10GBase-KR" }, \ { IFM_10G_CR1, "10GBase-CR1" }, \ { IFM_20G_KR2, "20GBase-KR2" }, \ { IFM_2500_KX, "2500Base-KX" }, \ { IFM_2500_T, "2500Base-T" }, \ { IFM_5000_T, "5000Base-T" }, \ { IFM_50G_PCIE, "PCIExpress-50G" }, \ { IFM_25G_PCIE, "PCIExpress-25G" }, \ { IFM_1000_SGMII, "1000Base-SGMII" }, \ { IFM_10G_SFI, "10GBase-SFI" }, \ { IFM_40G_XLPPI, "40GBase-XLPPI" }, \ { IFM_1000_CX_SGMII, "1000Base-CX-SGMII" }, \ { IFM_40G_KR4, "40GBase-KR4" }, \ { IFM_10G_ER, "10GBase-ER" }, \ { IFM_100G_CR4, "100GBase-CR4" }, \ { IFM_100G_SR4, "100GBase-SR4" }, \ { IFM_100G_KR4, "100GBase-KR4" }, \ { IFM_100G_LR4, "100GBase-LR4" }, \ { IFM_56G_R4, "56GBase-R4" }, \ { IFM_100_T, "100BaseT" }, \ { IFM_25G_CR, "25GBase-CR" }, \ { IFM_25G_KR, "25GBase-KR" }, \ { IFM_25G_SR, "25GBase-SR" }, \ { IFM_50G_CR2, "50GBase-CR2" }, \ { IFM_50G_KR2, "50GBase-KR2" }, \ { 0, NULL }, \ } #define IFM_SUBTYPE_ETHERNET_ALIASES { \ { IFM_10_T, "10baseT" }, \ { IFM_10_T, "UTP" }, \ { IFM_10_T, "10UTP" }, \ { IFM_10_2, "BNC" }, \ { IFM_10_2, "10BNC" }, \ { IFM_10_5, "AUI" }, \ { IFM_10_5, "10AUI" }, \ { IFM_100_TX, "100TX" }, \ { IFM_100_T4, "100T4" }, \ { IFM_100_VG, "100VG" }, \ { IFM_100_T2, "100T2" }, \ { IFM_10_STP, "10STP" }, \ { IFM_10_FL, "10FL" }, \ { IFM_1000_SX, "1000SX" }, \ { IFM_1000_LX, "1000LX" }, \ { IFM_1000_CX, "1000CX" }, \ { IFM_1000_T, "1000baseTX" }, \ { IFM_1000_T, "1000TX" }, \ { IFM_1000_T, "1000T" }, \ { IFM_2500_SX, "2500SX" }, \ \ /* \ * Shorthands for common media+option combinations as announced \ * by miibus(4) \ */ \ { IFM_10_T | IFM_FDX, "10baseT-FDX" }, \ { IFM_10_T | IFM_FDX | IFM_FLOW, "10baseT-FDX-flow" }, \ { IFM_100_TX | IFM_FDX, "100baseTX-FDX" }, \ { IFM_100_TX | IFM_FDX | IFM_FLOW, "100baseTX-FDX-flow" }, \ { IFM_1000_T | IFM_FDX, "1000baseT-FDX" }, \ { IFM_1000_T | IFM_FDX | IFM_FLOW, "1000baseT-FDX-flow" }, \ { IFM_1000_T | IFM_FDX | IFM_FLOW | IFM_ETH_MASTER, \ "1000baseT-FDX-flow-master" }, \ { IFM_1000_T | IFM_FDX | IFM_ETH_MASTER, \ "1000baseT-FDX-master" }, \ { IFM_1000_T | IFM_ETH_MASTER, "1000baseT-master" }, \ \ { 0, NULL }, \ } #define IFM_SUBTYPE_ETHERNET_OPTION_DESCRIPTIONS { \ { IFM_ETH_MASTER, "master" }, \ { IFM_ETH_RXPAUSE, "rxpause" }, \ { IFM_ETH_TXPAUSE, "txpause" }, \ { 0, NULL }, \ } #define IFM_SUBTYPE_TOKENRING_DESCRIPTIONS { \ { IFM_TOK_STP4, "DB9/4Mbit" }, \ { IFM_TOK_STP16, "DB9/16Mbit" }, \ { IFM_TOK_UTP4, "UTP/4Mbit" }, \ { IFM_TOK_UTP16, "UTP/16Mbit" }, \ { IFM_TOK_STP100, "STP/100Mbit" }, \ { IFM_TOK_UTP100, "UTP/100Mbit" }, \ { 0, NULL }, \ } #define IFM_SUBTYPE_TOKENRING_ALIASES { \ { IFM_TOK_STP4, "4STP" }, \ { IFM_TOK_STP16, "16STP" }, \ { IFM_TOK_UTP4, "4UTP" }, \ { IFM_TOK_UTP16, "16UTP" }, \ { IFM_TOK_STP100, "100STP" }, \ { IFM_TOK_UTP100, "100UTP" }, \ { 0, NULL }, \ } #define IFM_SUBTYPE_TOKENRING_OPTION_DESCRIPTIONS { \ { IFM_TOK_ETR, "EarlyTokenRelease" }, \ { IFM_TOK_SRCRT, "SourceRouting" }, \ { IFM_TOK_ALLR, "AllRoutes" }, \ { IFM_TOK_DTR, "Dedicated" }, \ { IFM_TOK_CLASSIC,"Classic" }, \ { IFM_TOK_AUTO, " " }, \ { 0, NULL }, \ } #define IFM_SUBTYPE_FDDI_DESCRIPTIONS { \ { IFM_FDDI_SMF, "Single-mode" }, \ { IFM_FDDI_MMF, "Multi-mode" }, \ { IFM_FDDI_UTP, "UTP" }, \ { 0, NULL }, \ } #define IFM_SUBTYPE_FDDI_ALIASES { \ { IFM_FDDI_SMF, "SMF" }, \ { IFM_FDDI_MMF, "MMF" }, \ { IFM_FDDI_UTP, "CDDI" }, \ { 0, NULL }, \ } #define IFM_SUBTYPE_FDDI_OPTION_DESCRIPTIONS { \ { IFM_FDDI_DA, "Dual-attach" }, \ { 0, NULL }, \ } #define IFM_SUBTYPE_IEEE80211_DESCRIPTIONS { \ { IFM_IEEE80211_FH1, "FH/1Mbps" }, \ { IFM_IEEE80211_FH2, "FH/2Mbps" }, \ { IFM_IEEE80211_DS1, "DS/1Mbps" }, \ { IFM_IEEE80211_DS2, "DS/2Mbps" }, \ { IFM_IEEE80211_DS5, "DS/5.5Mbps" }, \ { IFM_IEEE80211_DS11, "DS/11Mbps" }, \ { IFM_IEEE80211_DS22, "DS/22Mbps" }, \ { IFM_IEEE80211_OFDM6, "OFDM/6Mbps" }, \ { IFM_IEEE80211_OFDM9, "OFDM/9Mbps" }, \ { IFM_IEEE80211_OFDM12, "OFDM/12Mbps" }, \ { IFM_IEEE80211_OFDM18, "OFDM/18Mbps" }, \ { IFM_IEEE80211_OFDM24, "OFDM/24Mbps" }, \ { IFM_IEEE80211_OFDM36, "OFDM/36Mbps" }, \ { IFM_IEEE80211_OFDM48, "OFDM/48Mbps" }, \ { IFM_IEEE80211_OFDM54, "OFDM/54Mbps" }, \ { IFM_IEEE80211_OFDM72, "OFDM/72Mbps" }, \ { IFM_IEEE80211_DS354k, "DS/354Kbps" }, \ { IFM_IEEE80211_DS512k, "DS/512Kbps" }, \ { IFM_IEEE80211_OFDM3, "OFDM/3Mbps" }, \ { IFM_IEEE80211_OFDM4, "OFDM/4.5Mbps" }, \ { IFM_IEEE80211_OFDM27, "OFDM/27Mbps" }, \ { IFM_IEEE80211_MCS, "MCS" }, \ + { IFM_IEEE80211_VHT, "VHT" }, \ { 0, NULL }, \ } #define IFM_SUBTYPE_IEEE80211_ALIASES { \ { IFM_IEEE80211_FH1, "FH1" }, \ { IFM_IEEE80211_FH2, "FH2" }, \ { IFM_IEEE80211_FH1, "FrequencyHopping/1Mbps" }, \ { IFM_IEEE80211_FH2, "FrequencyHopping/2Mbps" }, \ { IFM_IEEE80211_DS1, "DS1" }, \ { IFM_IEEE80211_DS2, "DS2" }, \ { IFM_IEEE80211_DS5, "DS5.5" }, \ { IFM_IEEE80211_DS11, "DS11" }, \ { IFM_IEEE80211_DS22, "DS22" }, \ { IFM_IEEE80211_DS1, "DirectSequence/1Mbps" }, \ { IFM_IEEE80211_DS2, "DirectSequence/2Mbps" }, \ { IFM_IEEE80211_DS5, "DirectSequence/5.5Mbps" }, \ { IFM_IEEE80211_DS11, "DirectSequence/11Mbps" }, \ { IFM_IEEE80211_DS22, "DirectSequence/22Mbps" }, \ { IFM_IEEE80211_OFDM6, "OFDM6" }, \ { IFM_IEEE80211_OFDM9, "OFDM9" }, \ { IFM_IEEE80211_OFDM12, "OFDM12" }, \ { IFM_IEEE80211_OFDM18, "OFDM18" }, \ { IFM_IEEE80211_OFDM24, "OFDM24" }, \ { IFM_IEEE80211_OFDM36, "OFDM36" }, \ { IFM_IEEE80211_OFDM48, "OFDM48" }, \ { IFM_IEEE80211_OFDM54, "OFDM54" }, \ { IFM_IEEE80211_OFDM72, "OFDM72" }, \ { IFM_IEEE80211_DS1, "CCK1" }, \ { IFM_IEEE80211_DS2, "CCK2" }, \ { IFM_IEEE80211_DS5, "CCK5.5" }, \ { IFM_IEEE80211_DS11, "CCK11" }, \ { IFM_IEEE80211_DS354k, "DS354K" }, \ { IFM_IEEE80211_DS354k, "DirectSequence/354Kbps" }, \ { IFM_IEEE80211_DS512k, "DS512K" }, \ { IFM_IEEE80211_DS512k, "DirectSequence/512Kbps" }, \ { IFM_IEEE80211_OFDM3, "OFDM3" }, \ { IFM_IEEE80211_OFDM4, "OFDM4.5" }, \ { IFM_IEEE80211_OFDM27, "OFDM27" }, \ { IFM_IEEE80211_MCS, "MCS" }, \ + { IFM_IEEE80211_VHT, "VHT" }, \ { 0, NULL }, \ } #define IFM_SUBTYPE_IEEE80211_OPTION_DESCRIPTIONS { \ { IFM_IEEE80211_ADHOC, "adhoc" }, \ { IFM_IEEE80211_HOSTAP, "hostap" }, \ { IFM_IEEE80211_IBSS, "ibss" }, \ { IFM_IEEE80211_WDS, "wds" }, \ { IFM_IEEE80211_TURBO, "turbo" }, \ { IFM_IEEE80211_MONITOR, "monitor" }, \ { IFM_IEEE80211_MBSS, "mesh" }, \ { 0, NULL }, \ } #define IFM_SUBTYPE_IEEE80211_MODE_DESCRIPTIONS { \ { IFM_AUTO, "autoselect" }, \ { IFM_IEEE80211_11A, "11a" }, \ { IFM_IEEE80211_11B, "11b" }, \ { IFM_IEEE80211_11G, "11g" }, \ { IFM_IEEE80211_FH, "fh" }, \ { IFM_IEEE80211_11NA, "11na" }, \ { IFM_IEEE80211_11NG, "11ng" }, \ + { IFM_IEEE80211_VHT5G, "11ac" }, \ + { IFM_IEEE80211_VHT2G, "11ac2" }, \ { 0, NULL }, \ } #define IFM_SUBTYPE_IEEE80211_MODE_ALIASES { \ { IFM_AUTO, "auto" }, \ { 0, NULL }, \ } #define IFM_SUBTYPE_ATM_DESCRIPTIONS { \ { IFM_ATM_UNKNOWN, "Unknown" }, \ { IFM_ATM_UTP_25, "UTP/25.6MBit" }, \ { IFM_ATM_TAXI_100, "Taxi/100MBit" }, \ { IFM_ATM_TAXI_140, "Taxi/140MBit" }, \ { IFM_ATM_MM_155, "Multi-mode/155MBit" }, \ { IFM_ATM_SM_155, "Single-mode/155MBit" }, \ { IFM_ATM_UTP_155, "UTP/155MBit" }, \ { IFM_ATM_MM_622, "Multi-mode/622MBit" }, \ { IFM_ATM_SM_622, "Single-mode/622MBit" }, \ { IFM_ATM_VIRTUAL, "Virtual" }, \ { 0, NULL }, \ } #define IFM_SUBTYPE_ATM_ALIASES { \ { IFM_ATM_UNKNOWN, "UNKNOWN" }, \ { IFM_ATM_UTP_25, "UTP-25" }, \ { IFM_ATM_TAXI_100, "TAXI-100" }, \ { IFM_ATM_TAXI_140, "TAXI-140" }, \ { IFM_ATM_MM_155, "MM-155" }, \ { IFM_ATM_SM_155, "SM-155" }, \ { IFM_ATM_UTP_155, "UTP-155" }, \ { IFM_ATM_MM_622, "MM-622" }, \ { IFM_ATM_SM_622, "SM-622" }, \ { IFM_ATM_VIRTUAL, "VIRTUAL" }, \ { 0, NULL }, \ } #define IFM_SUBTYPE_ATM_OPTION_DESCRIPTIONS { \ { IFM_ATM_SDH, "SDH" }, \ { IFM_ATM_NOSCRAMB, "Noscramb" }, \ { IFM_ATM_UNASSIGNED, "Unassigned" }, \ { 0, NULL }, \ } #define IFM_SUBTYPE_SHARED_DESCRIPTIONS { \ { IFM_AUTO, "autoselect" }, \ { IFM_MANUAL, "manual" }, \ { IFM_NONE, "none" }, \ { 0, NULL }, \ } #define IFM_SUBTYPE_SHARED_ALIASES { \ { IFM_AUTO, "auto" }, \ \ /* \ * Shorthands for common media+option combinations as announced \ * by miibus(4) \ */ \ { IFM_AUTO | IFM_FLOW, "auto-flow" }, \ \ { 0, NULL }, \ } #define IFM_SHARED_OPTION_DESCRIPTIONS { \ { IFM_FDX, "full-duplex" }, \ { IFM_HDX, "half-duplex" }, \ { IFM_FLOW, "flowcontrol" }, \ { IFM_FLAG0, "flag0" }, \ { IFM_FLAG1, "flag1" }, \ { IFM_FLAG2, "flag2" }, \ { IFM_LOOP, "hw-loopback" }, \ { 0, NULL }, \ } #define IFM_SHARED_OPTION_ALIASES { \ { IFM_FDX, "fdx" }, \ { IFM_HDX, "hdx" }, \ { IFM_FLOW, "flow" }, \ { IFM_LOOP, "loop" }, \ { IFM_LOOP, "loopback" }, \ { 0, NULL }, \ } /* * Baudrate descriptions for the various media types. */ struct ifmedia_baudrate { int ifmb_word; /* media word */ uint64_t ifmb_baudrate; /* corresponding baudrate */ }; #define IFM_BAUDRATE_DESCRIPTIONS { \ { IFM_ETHER | IFM_10_T, IF_Mbps(10) }, \ { IFM_ETHER | IFM_10_2, IF_Mbps(10) }, \ { IFM_ETHER | IFM_10_5, IF_Mbps(10) }, \ { IFM_ETHER | IFM_100_TX, IF_Mbps(100) }, \ { IFM_ETHER | IFM_100_FX, IF_Mbps(100) }, \ { IFM_ETHER | IFM_100_T4, IF_Mbps(100) }, \ { IFM_ETHER | IFM_100_VG, IF_Mbps(100) }, \ { IFM_ETHER | IFM_100_T2, IF_Mbps(100) }, \ { IFM_ETHER | IFM_1000_SX, IF_Mbps(1000) }, \ { IFM_ETHER | IFM_10_STP, IF_Mbps(10) }, \ { IFM_ETHER | IFM_10_FL, IF_Mbps(10) }, \ { IFM_ETHER | IFM_1000_LX, IF_Mbps(1000) }, \ { IFM_ETHER | IFM_1000_CX, IF_Mbps(1000) }, \ { IFM_ETHER | IFM_1000_T, IF_Mbps(1000) }, \ { IFM_ETHER | IFM_HPNA_1, IF_Mbps(1) }, \ { IFM_ETHER | IFM_10G_LR, IF_Gbps(10ULL) }, \ { IFM_ETHER | IFM_10G_SR, IF_Gbps(10ULL) }, \ { IFM_ETHER | IFM_10G_CX4, IF_Gbps(10ULL) }, \ { IFM_ETHER | IFM_2500_SX, IF_Mbps(2500ULL) }, \ { IFM_ETHER | IFM_10G_TWINAX, IF_Gbps(10ULL) }, \ { IFM_ETHER | IFM_10G_TWINAX_LONG, IF_Gbps(10ULL) }, \ { IFM_ETHER | IFM_10G_LRM, IF_Gbps(10ULL) }, \ { IFM_ETHER | IFM_10G_T, IF_Gbps(10ULL) }, \ { IFM_ETHER | IFM_40G_CR4, IF_Gbps(40ULL) }, \ { IFM_ETHER | IFM_40G_SR4, IF_Gbps(40ULL) }, \ { IFM_ETHER | IFM_40G_LR4, IF_Gbps(40ULL) }, \ { IFM_ETHER | IFM_1000_KX, IF_Mbps(1000) }, \ { IFM_ETHER | IFM_10G_KX4, IF_Gbps(10ULL) }, \ { IFM_ETHER | IFM_10G_KR, IF_Gbps(10ULL) }, \ { IFM_ETHER | IFM_10G_CR1, IF_Gbps(10ULL) }, \ { IFM_ETHER | IFM_20G_KR2, IF_Gbps(20ULL) }, \ { IFM_ETHER | IFM_2500_KX, IF_Mbps(2500) }, \ { IFM_ETHER | IFM_2500_T, IF_Mbps(2500) }, \ { IFM_ETHER | IFM_5000_T, IF_Mbps(5000) }, \ { IFM_ETHER | IFM_50G_PCIE, IF_Gbps(50ULL) }, \ { IFM_ETHER | IFM_25G_PCIE, IF_Gbps(25ULL) }, \ { IFM_ETHER | IFM_1000_SGMII, IF_Mbps(1000) }, \ { IFM_ETHER | IFM_10G_SFI, IF_Gbps(10ULL) }, \ { IFM_ETHER | IFM_40G_XLPPI, IF_Gbps(40ULL) }, \ { IFM_ETHER | IFM_1000_CX_SGMII, IF_Mbps(1000) }, \ { IFM_ETHER | IFM_40G_KR4, IF_Gbps(40ULL) }, \ { IFM_ETHER | IFM_10G_ER, IF_Gbps(10ULL) }, \ { IFM_ETHER | IFM_100G_CR4, IF_Gbps(100ULL) }, \ { IFM_ETHER | IFM_100G_SR4, IF_Gbps(100ULL) }, \ { IFM_ETHER | IFM_100G_KR4, IF_Gbps(100ULL) }, \ { IFM_ETHER | IFM_100G_LR4, IF_Gbps(100ULL) }, \ { IFM_ETHER | IFM_56G_R4, IF_Gbps(56ULL) }, \ { IFM_ETHER | IFM_100_T, IF_Mbps(100ULL) }, \ { IFM_ETHER | IFM_25G_CR, IF_Gbps(25ULL) }, \ { IFM_ETHER | IFM_25G_KR, IF_Gbps(25ULL) }, \ { IFM_ETHER | IFM_25G_SR, IF_Gbps(25ULL) }, \ { IFM_ETHER | IFM_50G_CR2, IF_Gbps(50ULL) }, \ { IFM_ETHER | IFM_50G_KR2, IF_Gbps(50ULL) }, \ \ { IFM_TOKEN | IFM_TOK_STP4, IF_Mbps(4) }, \ { IFM_TOKEN | IFM_TOK_STP16, IF_Mbps(16) }, \ { IFM_TOKEN | IFM_TOK_UTP4, IF_Mbps(4) }, \ { IFM_TOKEN | IFM_TOK_UTP16, IF_Mbps(16) }, \ \ { IFM_FDDI | IFM_FDDI_SMF, IF_Mbps(100) }, \ { IFM_FDDI | IFM_FDDI_MMF, IF_Mbps(100) }, \ { IFM_FDDI | IFM_FDDI_UTP, IF_Mbps(100) }, \ \ { IFM_IEEE80211 | IFM_IEEE80211_FH1, IF_Mbps(1) }, \ { IFM_IEEE80211 | IFM_IEEE80211_FH2, IF_Mbps(2) }, \ { IFM_IEEE80211 | IFM_IEEE80211_DS2, IF_Mbps(2) }, \ { IFM_IEEE80211 | IFM_IEEE80211_DS5, IF_Kbps(5500) }, \ { IFM_IEEE80211 | IFM_IEEE80211_DS11, IF_Mbps(11) }, \ { IFM_IEEE80211 | IFM_IEEE80211_DS1, IF_Mbps(1) }, \ { IFM_IEEE80211 | IFM_IEEE80211_DS22, IF_Mbps(22) }, \ { IFM_IEEE80211 | IFM_IEEE80211_OFDM6, IF_Mbps(6) }, \ { IFM_IEEE80211 | IFM_IEEE80211_OFDM9, IF_Mbps(9) }, \ { IFM_IEEE80211 | IFM_IEEE80211_OFDM12, IF_Mbps(12) }, \ { IFM_IEEE80211 | IFM_IEEE80211_OFDM18, IF_Mbps(18) }, \ { IFM_IEEE80211 | IFM_IEEE80211_OFDM24, IF_Mbps(24) }, \ { IFM_IEEE80211 | IFM_IEEE80211_OFDM36, IF_Mbps(36) }, \ { IFM_IEEE80211 | IFM_IEEE80211_OFDM48, IF_Mbps(48) }, \ { IFM_IEEE80211 | IFM_IEEE80211_OFDM54, IF_Mbps(54) }, \ { IFM_IEEE80211 | IFM_IEEE80211_OFDM72, IF_Mbps(72) }, \ \ { 0, 0 }, \ } /* * Status descriptions for the various media types. */ struct ifmedia_status_description { int ifms_type; int ifms_valid; int ifms_bit; const char *ifms_string[2]; }; #define IFM_STATUS_DESC(ifms, bit) \ (ifms)->ifms_string[((ifms)->ifms_bit & (bit)) ? 1 : 0] #define IFM_STATUS_DESCRIPTIONS { \ { IFM_ETHER, IFM_AVALID, IFM_ACTIVE, \ { "no carrier", "active" } }, \ { IFM_FDDI, IFM_AVALID, IFM_ACTIVE, \ { "no ring", "inserted" } }, \ { IFM_TOKEN, IFM_AVALID, IFM_ACTIVE, \ { "no ring", "inserted" } }, \ { IFM_IEEE80211, IFM_AVALID, IFM_ACTIVE, \ { "no network", "active" } }, \ { IFM_ATM, IFM_AVALID, IFM_ACTIVE, \ { "no network", "active" } }, \ { 0, 0, 0, \ { NULL, NULL } } \ } #endif /* _NET_IF_MEDIA_H_ */ Index: projects/clang400-import/sys/net80211/ieee80211_var.h =================================================================== --- projects/clang400-import/sys/net80211/ieee80211_var.h (revision 311459) +++ projects/clang400-import/sys/net80211/ieee80211_var.h (revision 311460) @@ -1,978 +1,983 @@ /*- * Copyright (c) 2001 Atsushi Onoe * 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. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. * * $FreeBSD$ */ #ifndef _NET80211_IEEE80211_VAR_H_ #define _NET80211_IEEE80211_VAR_H_ /* * Definitions for IEEE 802.11 drivers. */ /* NB: portability glue must go first */ #if defined(__NetBSD__) #include #elif defined(__FreeBSD__) #include #elif defined(__linux__) #include #else #error "No support for your operating system!" #endif #include #include #include #include #include #include /* for ieee80211_stats */ #include #include #include #include #include #include #define IEEE80211_TXPOWER_MAX 100 /* .5 dBm (XXX units?) */ #define IEEE80211_TXPOWER_MIN 0 /* kill radio */ #define IEEE80211_DTIM_DEFAULT 1 /* default DTIM period */ #define IEEE80211_BINTVAL_DEFAULT 100 /* default beacon interval (TU's) */ #define IEEE80211_BMISS_MAX 2 /* maximum consecutive bmiss allowed */ #define IEEE80211_HWBMISS_DEFAULT 7 /* h/w bmiss threshold (beacons) */ #define IEEE80211_BGSCAN_INTVAL_MIN 15 /* min bg scan intvl (secs) */ #define IEEE80211_BGSCAN_INTVAL_DEFAULT (5*60) /* default bg scan intvl */ #define IEEE80211_BGSCAN_IDLE_MIN 100 /* min idle time (ms) */ #define IEEE80211_BGSCAN_IDLE_DEFAULT 250 /* default idle time (ms) */ #define IEEE80211_SCAN_VALID_MIN 10 /* min scan valid time (secs) */ #define IEEE80211_SCAN_VALID_DEFAULT 60 /* default scan valid time */ #define IEEE80211_PS_SLEEP 0x1 /* STA is in power saving mode */ #define IEEE80211_PS_MAX_QUEUE 50 /* maximum saved packets */ #define IEEE80211_FIXED_RATE_NONE 0xff #define IEEE80211_TXMAX_DEFAULT 6 /* default ucast max retries */ #define IEEE80211_RTS_DEFAULT IEEE80211_RTS_MAX #define IEEE80211_FRAG_DEFAULT IEEE80211_FRAG_MAX #define IEEE80211_MS_TO_TU(x) (((x) * 1000) / 1024) #define IEEE80211_TU_TO_MS(x) (((x) * 1024) / 1000) /* XXX TODO: cap this at 1, in case hz is not 1000 */ #define IEEE80211_TU_TO_TICKS(x)(((uint64_t)(x) * 1024 * hz) / (1000 * 1000)) /* * 802.11 control state is split into a common portion that maps * 1-1 to a physical device and one or more "Virtual AP's" (VAP) * that are bound to an ieee80211com instance and share a single * underlying device. Each VAP has a corresponding OS device * entity through which traffic flows and that applications use * for issuing ioctls, etc. */ /* * Data common to one or more virtual AP's. State shared by * the underlying device and the net80211 layer is exposed here; * e.g. device-specific callbacks. */ struct ieee80211vap; typedef void (*ieee80211vap_attach)(struct ieee80211vap *); struct ieee80211_appie { uint16_t ie_len; /* size of ie_data */ uint8_t ie_data[]; /* user-specified IE's */ }; struct ieee80211_tdma_param; struct ieee80211_rate_table; struct ieee80211_tx_ampdu; struct ieee80211_rx_ampdu; struct ieee80211_superg; struct ieee80211_frame; struct ieee80211com { void *ic_softc; /* driver softc */ const char *ic_name; /* usually device name */ ieee80211_com_lock_t ic_comlock; /* state update lock */ ieee80211_tx_lock_t ic_txlock; /* ic/vap TX lock */ ieee80211_ff_lock_t ic_fflock; /* stageq/ni_tx_superg lock */ LIST_ENTRY(ieee80211com) ic_next; /* on global list */ TAILQ_HEAD(, ieee80211vap) ic_vaps; /* list of vap instances */ int ic_headroom; /* driver tx headroom needs */ enum ieee80211_phytype ic_phytype; /* XXX wrong for multi-mode */ enum ieee80211_opmode ic_opmode; /* operation mode */ struct callout ic_inact; /* inactivity processing */ struct taskqueue *ic_tq; /* deferred state thread */ struct task ic_parent_task; /* deferred parent processing */ struct task ic_promisc_task;/* deferred promisc update */ struct task ic_mcast_task; /* deferred mcast update */ struct task ic_chan_task; /* deferred channel change */ struct task ic_bmiss_task; /* deferred beacon miss hndlr */ struct task ic_chw_task; /* deferred HT CHW update */ struct task ic_wme_task; /* deferred WME update */ struct task ic_restart_task; /* deferred device restart */ counter_u64_t ic_ierrors; /* input errors */ counter_u64_t ic_oerrors; /* output errors */ uint32_t ic_flags; /* state flags */ uint32_t ic_flags_ext; /* extended state flags */ uint32_t ic_flags_ht; /* HT state flags */ uint32_t ic_flags_ven; /* vendor state flags */ uint32_t ic_caps; /* capabilities */ uint32_t ic_htcaps; /* HT capabilities */ uint32_t ic_htextcaps; /* HT extended capabilities */ uint32_t ic_cryptocaps; /* crypto capabilities */ /* set of mode capabilities */ uint8_t ic_modecaps[IEEE80211_MODE_BYTES]; uint8_t ic_promisc; /* vap's needing promisc mode */ uint8_t ic_allmulti; /* vap's needing all multicast*/ uint8_t ic_nrunning; /* vap's marked running */ uint8_t ic_curmode; /* current mode */ uint8_t ic_macaddr[IEEE80211_ADDR_LEN]; uint16_t ic_bintval; /* beacon interval */ uint16_t ic_lintval; /* listen interval */ uint16_t ic_holdover; /* PM hold over duration */ uint16_t ic_txpowlimit; /* global tx power limit */ struct ieee80211_rateset ic_sup_rates[IEEE80211_MODE_MAX]; /* * Channel state: * * ic_channels is the set of available channels for the device; * it is setup by the driver * ic_nchans is the number of valid entries in ic_channels * ic_chan_avail is a bit vector of these channels used to check * whether a channel is available w/o searching the channel table. * ic_chan_active is a (potentially) constrained subset of * ic_chan_avail that reflects any mode setting or user-specified * limit on the set of channels to use/scan * ic_curchan is the current channel the device is set to; it may * be different from ic_bsschan when we are off-channel scanning * or otherwise doing background work * ic_bsschan is the channel selected for operation; it may * be undefined (IEEE80211_CHAN_ANYC) * ic_prevchan is a cached ``previous channel'' used to optimize * lookups when switching back+forth between two channels * (e.g. for dynamic turbo) */ int ic_nchans; /* # entries in ic_channels */ struct ieee80211_channel ic_channels[IEEE80211_CHAN_MAX]; uint8_t ic_chan_avail[IEEE80211_CHAN_BYTES]; uint8_t ic_chan_active[IEEE80211_CHAN_BYTES]; uint8_t ic_chan_scan[IEEE80211_CHAN_BYTES]; struct ieee80211_channel *ic_curchan; /* current channel */ const struct ieee80211_rate_table *ic_rt; /* table for ic_curchan */ struct ieee80211_channel *ic_bsschan; /* bss channel */ struct ieee80211_channel *ic_prevchan; /* previous channel */ struct ieee80211_regdomain ic_regdomain;/* regulatory data */ struct ieee80211_appie *ic_countryie; /* calculated country ie */ struct ieee80211_channel *ic_countryie_chan; /* 802.11h/DFS state */ struct ieee80211_channel *ic_csa_newchan;/* channel for doing CSA */ short ic_csa_mode; /* mode for doing CSA */ short ic_csa_count; /* count for doing CSA */ struct ieee80211_dfs_state ic_dfs; /* DFS state */ struct ieee80211_scan_state *ic_scan; /* scan state */ struct ieee80211_scan_methods *ic_scan_methods; /* scan methods */ int ic_lastdata; /* time of last data frame */ int ic_lastscan; /* time last scan completed */ /* NB: this is the union of all vap stations/neighbors */ int ic_max_keyix; /* max h/w key index */ struct ieee80211_node_table ic_sta; /* stations/neighbors */ struct ieee80211_ageq ic_stageq; /* frame staging queue */ uint32_t ic_hash_key; /* random key for mac hash */ /* XXX multi-bss: split out common/vap parts */ struct ieee80211_wme_state ic_wme; /* WME/WMM state */ /* XXX multi-bss: can per-vap be done/make sense? */ enum ieee80211_protmode ic_protmode; /* 802.11g protection mode */ uint16_t ic_nonerpsta; /* # non-ERP stations */ uint16_t ic_longslotsta; /* # long slot time stations */ uint16_t ic_sta_assoc; /* stations associated */ uint16_t ic_ht_sta_assoc;/* HT stations associated */ uint16_t ic_ht40_sta_assoc;/* HT40 stations associated */ uint8_t ic_curhtprotmode;/* HTINFO bss state */ enum ieee80211_protmode ic_htprotmode; /* HT protection mode */ int ic_lastnonerp; /* last time non-ERP sta noted*/ int ic_lastnonht; /* last time non-HT sta noted */ uint8_t ic_rxstream; /* # RX streams */ uint8_t ic_txstream; /* # TX streams */ /* VHT information */ uint32_t ic_vhtcaps; /* VHT capabilities */ uint32_t ic_vhtextcaps; /* VHT extended capabilities (TODO) */ - struct ieee80211_vht_mcs_info iv_vht_mcsinfo; /* Support TX/RX VHT MCS */ - uint32_t ic_vht_spare[4]; + struct ieee80211_vht_mcs_info ic_vht_mcsinfo; /* Support TX/RX VHT MCS */ + uint32_t ic_flags_vht; /* VHT state flags */ + uint32_t ic_vht_spare[3]; /* optional state for Atheros SuperG protocol extensions */ struct ieee80211_superg *ic_superg; /* radiotap handling */ struct ieee80211_radiotap_header *ic_th;/* tx radiotap headers */ void *ic_txchan; /* channel state in ic_th */ struct ieee80211_radiotap_header *ic_rh;/* rx radiotap headers */ void *ic_rxchan; /* channel state in ic_rh */ int ic_montaps; /* active monitor mode taps */ /* virtual ap create/delete */ struct ieee80211vap* (*ic_vap_create)(struct ieee80211com *, const char [IFNAMSIZ], int, enum ieee80211_opmode, int, const uint8_t [IEEE80211_ADDR_LEN], const uint8_t [IEEE80211_ADDR_LEN]); void (*ic_vap_delete)(struct ieee80211vap *); /* device specific ioctls */ int (*ic_ioctl)(struct ieee80211com *, u_long, void *); /* start/stop device */ void (*ic_parent)(struct ieee80211com *); /* operating mode attachment */ ieee80211vap_attach ic_vattach[IEEE80211_OPMODE_MAX]; /* return hardware/radio capabilities */ void (*ic_getradiocaps)(struct ieee80211com *, int, int *, struct ieee80211_channel []); /* check and/or prepare regdomain state change */ int (*ic_setregdomain)(struct ieee80211com *, struct ieee80211_regdomain *, int, struct ieee80211_channel []); int (*ic_set_quiet)(struct ieee80211_node *, u_int8_t *quiet_elm); /* regular transmit */ int (*ic_transmit)(struct ieee80211com *, struct mbuf *); /* send/recv 802.11 management frame */ int (*ic_send_mgmt)(struct ieee80211_node *, int, int); /* send raw 802.11 frame */ int (*ic_raw_xmit)(struct ieee80211_node *, struct mbuf *, const struct ieee80211_bpf_params *); /* update device state for 802.11 slot time change */ void (*ic_updateslot)(struct ieee80211com *); /* handle multicast state changes */ void (*ic_update_mcast)(struct ieee80211com *); /* handle promiscuous mode changes */ void (*ic_update_promisc)(struct ieee80211com *); /* new station association callback/notification */ void (*ic_newassoc)(struct ieee80211_node *, int); /* TDMA update notification */ void (*ic_tdma_update)(struct ieee80211_node *, const struct ieee80211_tdma_param *, int); /* node state management */ struct ieee80211_node* (*ic_node_alloc)(struct ieee80211vap *, const uint8_t [IEEE80211_ADDR_LEN]); void (*ic_node_free)(struct ieee80211_node *); void (*ic_node_cleanup)(struct ieee80211_node *); void (*ic_node_age)(struct ieee80211_node *); void (*ic_node_drain)(struct ieee80211_node *); int8_t (*ic_node_getrssi)(const struct ieee80211_node*); void (*ic_node_getsignal)(const struct ieee80211_node*, int8_t *, int8_t *); void (*ic_node_getmimoinfo)( const struct ieee80211_node*, struct ieee80211_mimo_info *); /* scanning support */ void (*ic_scan_start)(struct ieee80211com *); void (*ic_scan_end)(struct ieee80211com *); void (*ic_set_channel)(struct ieee80211com *); void (*ic_scan_curchan)(struct ieee80211_scan_state *, unsigned long); void (*ic_scan_mindwell)(struct ieee80211_scan_state *); /* * 802.11n ADDBA support. A simple/generic implementation * of A-MPDU tx aggregation is provided; the driver may * override these methods to provide their own support. * A-MPDU rx re-ordering happens automatically if the * driver passes out-of-order frames to ieee80211_input * from an assocated HT station. */ int (*ic_recv_action)(struct ieee80211_node *, const struct ieee80211_frame *, const uint8_t *frm, const uint8_t *efrm); int (*ic_send_action)(struct ieee80211_node *, int category, int action, void *); /* check if A-MPDU should be enabled this station+ac */ int (*ic_ampdu_enable)(struct ieee80211_node *, struct ieee80211_tx_ampdu *); /* start/stop doing A-MPDU tx aggregation for a station */ int (*ic_addba_request)(struct ieee80211_node *, struct ieee80211_tx_ampdu *, int dialogtoken, int baparamset, int batimeout); int (*ic_addba_response)(struct ieee80211_node *, struct ieee80211_tx_ampdu *, int status, int baparamset, int batimeout); void (*ic_addba_stop)(struct ieee80211_node *, struct ieee80211_tx_ampdu *); void (*ic_addba_response_timeout)(struct ieee80211_node *, struct ieee80211_tx_ampdu *); /* BAR response received */ void (*ic_bar_response)(struct ieee80211_node *, struct ieee80211_tx_ampdu *, int status); /* start/stop doing A-MPDU rx processing for a station */ int (*ic_ampdu_rx_start)(struct ieee80211_node *, struct ieee80211_rx_ampdu *, int baparamset, int batimeout, int baseqctl); void (*ic_ampdu_rx_stop)(struct ieee80211_node *, struct ieee80211_rx_ampdu *); /* The channel width has changed (20<->2040) */ void (*ic_update_chw)(struct ieee80211com *); uint64_t ic_spare[7]; }; struct ieee80211_aclator; struct ieee80211_tdma_state; struct ieee80211_mesh_state; struct ieee80211_hwmp_state; struct ieee80211vap { struct ifmedia iv_media; /* interface media config */ struct ifnet *iv_ifp; /* associated device */ struct bpf_if *iv_rawbpf; /* packet filter structure */ struct sysctl_ctx_list *iv_sysctl; /* dynamic sysctl context */ struct sysctl_oid *iv_oid; /* net.wlan.X sysctl oid */ TAILQ_ENTRY(ieee80211vap) iv_next; /* list of vap instances */ struct ieee80211com *iv_ic; /* back ptr to common state */ /* MAC address: ifp or ic */ uint8_t iv_myaddr[IEEE80211_ADDR_LEN]; uint32_t iv_debug; /* debug msg flags */ struct ieee80211_stats iv_stats; /* statistics */ uint32_t iv_flags; /* state flags */ uint32_t iv_flags_ext; /* extended state flags */ uint32_t iv_flags_ht; /* HT state flags */ uint32_t iv_flags_ven; /* vendor state flags */ uint32_t iv_ifflags; /* ifnet flags */ uint32_t iv_caps; /* capabilities */ uint32_t iv_htcaps; /* HT capabilities */ uint32_t iv_htextcaps; /* HT extended capabilities */ enum ieee80211_opmode iv_opmode; /* operation mode */ enum ieee80211_state iv_state; /* state machine state */ enum ieee80211_state iv_nstate; /* pending state */ int iv_nstate_arg; /* pending state arg */ struct task iv_nstate_task; /* deferred state processing */ struct task iv_swbmiss_task;/* deferred iv_bmiss call */ struct callout iv_mgtsend; /* mgmt frame response timer */ /* inactivity timer settings */ int iv_inact_init; /* setting for new station */ int iv_inact_auth; /* auth but not assoc setting */ int iv_inact_run; /* authorized setting */ int iv_inact_probe; /* inactive probe time */ /* VHT flags */ uint32_t iv_flags_vht; /* VHT state flags */ uint32_t iv_vhtcaps; /* VHT capabilities */ uint32_t iv_vhtextcaps; /* VHT extended capabilities (TODO) */ struct ieee80211_vht_mcs_info iv_vht_mcsinfo; uint32_t iv_vht_spare[4]; int iv_des_nssid; /* # desired ssids */ struct ieee80211_scan_ssid iv_des_ssid[1];/* desired ssid table */ uint8_t iv_des_bssid[IEEE80211_ADDR_LEN]; struct ieee80211_channel *iv_des_chan; /* desired channel */ uint16_t iv_des_mode; /* desired mode */ int iv_nicknamelen; /* XXX junk */ uint8_t iv_nickname[IEEE80211_NWID_LEN]; u_int iv_bgscanidle; /* bg scan idle threshold */ u_int iv_bgscanintvl; /* bg scan min interval */ u_int iv_scanvalid; /* scan cache valid threshold */ u_int iv_scanreq_duration; u_int iv_scanreq_mindwell; u_int iv_scanreq_maxdwell; uint16_t iv_scanreq_flags;/* held scan request params */ uint8_t iv_scanreq_nssid; struct ieee80211_scan_ssid iv_scanreq_ssid[IEEE80211_SCAN_MAX_SSID]; /* sta-mode roaming state */ enum ieee80211_roamingmode iv_roaming; /* roaming mode */ struct ieee80211_roamparam iv_roamparms[IEEE80211_MODE_MAX]; uint8_t iv_bmissthreshold; uint8_t iv_bmiss_count; /* current beacon miss count */ int iv_bmiss_max; /* max bmiss before scan */ uint16_t iv_swbmiss_count;/* beacons in last period */ uint16_t iv_swbmiss_period;/* s/w bmiss period */ struct callout iv_swbmiss; /* s/w beacon miss timer */ int iv_ampdu_rxmax; /* A-MPDU rx limit (bytes) */ int iv_ampdu_density;/* A-MPDU density */ int iv_ampdu_limit; /* A-MPDU tx limit (bytes) */ int iv_amsdu_limit; /* A-MSDU tx limit (bytes) */ u_int iv_ampdu_mintraffic[WME_NUM_AC]; struct ieee80211_beacon_offsets iv_bcn_off; uint32_t *iv_aid_bitmap; /* association id map */ uint16_t iv_max_aid; uint16_t iv_sta_assoc; /* stations associated */ uint16_t iv_ps_sta; /* stations in power save */ uint16_t iv_ps_pending; /* ps sta's w/ pending frames */ uint16_t iv_txseq; /* mcast xmit seq# space */ uint16_t iv_tim_len; /* ic_tim_bitmap size (bytes) */ uint8_t *iv_tim_bitmap; /* power-save stations w/ data*/ uint8_t iv_dtim_period; /* DTIM period */ uint8_t iv_dtim_count; /* DTIM count from last bcn */ /* set/unset aid pwrsav state */ uint8_t iv_quiet; /* Quiet Element */ uint8_t iv_quiet_count; /* constant count for Quiet Element */ uint8_t iv_quiet_count_value; /* variable count for Quiet Element */ uint8_t iv_quiet_period; /* period for Quiet Element */ uint16_t iv_quiet_duration; /* duration for Quiet Element */ uint16_t iv_quiet_offset; /* offset for Quiet Element */ int iv_csa_count; /* count for doing CSA */ struct ieee80211_node *iv_bss; /* information for this node */ struct ieee80211_txparam iv_txparms[IEEE80211_MODE_MAX]; uint16_t iv_rtsthreshold; uint16_t iv_fragthreshold; int iv_inact_timer; /* inactivity timer wait */ /* application-specified IE's to attach to mgt frames */ struct ieee80211_appie *iv_appie_beacon; struct ieee80211_appie *iv_appie_probereq; struct ieee80211_appie *iv_appie_proberesp; struct ieee80211_appie *iv_appie_assocreq; struct ieee80211_appie *iv_appie_assocresp; struct ieee80211_appie *iv_appie_wpa; uint8_t *iv_wpa_ie; uint8_t *iv_rsn_ie; /* Key management */ uint16_t iv_max_keyix; /* max h/w key index */ ieee80211_keyix iv_def_txkey; /* default/group tx key index */ struct ieee80211_key iv_nw_keys[IEEE80211_WEP_NKID]; int (*iv_key_alloc)(struct ieee80211vap *, struct ieee80211_key *, ieee80211_keyix *, ieee80211_keyix *); int (*iv_key_delete)(struct ieee80211vap *, const struct ieee80211_key *); int (*iv_key_set)(struct ieee80211vap *, const struct ieee80211_key *); void (*iv_key_update_begin)(struct ieee80211vap *); void (*iv_key_update_end)(struct ieee80211vap *); void (*iv_update_deftxkey)(struct ieee80211vap *, ieee80211_keyix deftxkey); const struct ieee80211_authenticator *iv_auth; /* authenticator glue */ void *iv_ec; /* private auth state */ const struct ieee80211_aclator *iv_acl; /* acl glue */ void *iv_as; /* private aclator state */ const struct ieee80211_ratectl *iv_rate; void *iv_rs; /* private ratectl state */ struct ieee80211_tdma_state *iv_tdma; /* tdma state */ struct ieee80211_mesh_state *iv_mesh; /* MBSS state */ struct ieee80211_hwmp_state *iv_hwmp; /* HWMP state */ /* operate-mode detach hook */ void (*iv_opdetach)(struct ieee80211vap *); /* receive processing */ int (*iv_input)(struct ieee80211_node *, struct mbuf *, const struct ieee80211_rx_stats *, int, int); void (*iv_recv_mgmt)(struct ieee80211_node *, struct mbuf *, int, const struct ieee80211_rx_stats *, int, int); void (*iv_recv_ctl)(struct ieee80211_node *, struct mbuf *, int); void (*iv_deliver_data)(struct ieee80211vap *, struct ieee80211_node *, struct mbuf *); #if 0 /* send processing */ int (*iv_send_mgmt)(struct ieee80211_node *, int, int); #endif /* beacon miss processing */ void (*iv_bmiss)(struct ieee80211vap *); /* reset device state after 802.11 parameter/state change */ int (*iv_reset)(struct ieee80211vap *, u_long); /* [schedule] beacon frame update */ void (*iv_update_beacon)(struct ieee80211vap *, int); /* power save handling */ void (*iv_update_ps)(struct ieee80211vap *, int); int (*iv_set_tim)(struct ieee80211_node *, int); void (*iv_node_ps)(struct ieee80211_node *, int); void (*iv_sta_ps)(struct ieee80211vap *, int); void (*iv_recv_pspoll)(struct ieee80211_node *, struct mbuf *); /* state machine processing */ int (*iv_newstate)(struct ieee80211vap *, enum ieee80211_state, int); /* 802.3 output method for raw frame xmit */ int (*iv_output)(struct ifnet *, struct mbuf *, const struct sockaddr *, struct route *); uint64_t iv_spare[6]; }; MALLOC_DECLARE(M_80211_VAP); #define IEEE80211_ADDR_EQ(a1,a2) (memcmp(a1,a2,IEEE80211_ADDR_LEN) == 0) #define IEEE80211_ADDR_COPY(dst,src) memcpy(dst,src,IEEE80211_ADDR_LEN) /* ic_flags/iv_flags */ #define IEEE80211_F_TURBOP 0x00000001 /* CONF: ATH Turbo enabled*/ #define IEEE80211_F_COMP 0x00000002 /* CONF: ATH comp enabled */ #define IEEE80211_F_FF 0x00000004 /* CONF: ATH FF enabled */ #define IEEE80211_F_BURST 0x00000008 /* CONF: bursting enabled */ /* NB: this is intentionally setup to be IEEE80211_CAPINFO_PRIVACY */ #define IEEE80211_F_PRIVACY 0x00000010 /* CONF: privacy enabled */ #define IEEE80211_F_PUREG 0x00000020 /* CONF: 11g w/o 11b sta's */ #define IEEE80211_F_SCAN 0x00000080 /* STATUS: scanning */ #define IEEE80211_F_ASCAN 0x00000100 /* STATUS: active scan */ #define IEEE80211_F_SIBSS 0x00000200 /* STATUS: start IBSS */ /* NB: this is intentionally setup to be IEEE80211_CAPINFO_SHORT_SLOTTIME */ #define IEEE80211_F_SHSLOT 0x00000400 /* STATUS: use short slot time*/ #define IEEE80211_F_PMGTON 0x00000800 /* CONF: Power mgmt enable */ #define IEEE80211_F_DESBSSID 0x00001000 /* CONF: des_bssid is set */ #define IEEE80211_F_WME 0x00002000 /* CONF: enable WME use */ #define IEEE80211_F_BGSCAN 0x00004000 /* CONF: bg scan enabled (???)*/ #define IEEE80211_F_SWRETRY 0x00008000 /* CONF: sw tx retry enabled */ #define IEEE80211_F_TXPOW_FIXED 0x00010000 /* TX Power: fixed rate */ #define IEEE80211_F_IBSSON 0x00020000 /* CONF: IBSS creation enable */ #define IEEE80211_F_SHPREAMBLE 0x00040000 /* STATUS: use short preamble */ #define IEEE80211_F_DATAPAD 0x00080000 /* CONF: do alignment pad */ #define IEEE80211_F_USEPROT 0x00100000 /* STATUS: protection enabled */ #define IEEE80211_F_USEBARKER 0x00200000 /* STATUS: use barker preamble*/ #define IEEE80211_F_CSAPENDING 0x00400000 /* STATUS: chan switch pending*/ #define IEEE80211_F_WPA1 0x00800000 /* CONF: WPA enabled */ #define IEEE80211_F_WPA2 0x01000000 /* CONF: WPA2 enabled */ #define IEEE80211_F_WPA 0x01800000 /* CONF: WPA/WPA2 enabled */ #define IEEE80211_F_DROPUNENC 0x02000000 /* CONF: drop unencrypted */ #define IEEE80211_F_COUNTERM 0x04000000 /* CONF: TKIP countermeasures */ #define IEEE80211_F_HIDESSID 0x08000000 /* CONF: hide SSID in beacon */ #define IEEE80211_F_NOBRIDGE 0x10000000 /* CONF: dis. internal bridge */ #define IEEE80211_F_PCF 0x20000000 /* CONF: PCF enabled */ #define IEEE80211_F_DOTH 0x40000000 /* CONF: 11h enabled */ #define IEEE80211_F_DWDS 0x80000000 /* CONF: Dynamic WDS enabled */ #define IEEE80211_F_BITS \ "\20\1TURBOP\2COMP\3FF\4BURST\5PRIVACY\6PUREG\10SCAN\11ASCAN\12SIBSS" \ "\13SHSLOT\14PMGTON\15DESBSSID\16WME\17BGSCAN\20SWRETRY\21TXPOW_FIXED" \ "\22IBSSON\23SHPREAMBLE\24DATAPAD\25USEPROT\26USERBARKER\27CSAPENDING" \ "\30WPA1\31WPA2\32DROPUNENC\33COUNTERM\34HIDESSID\35NOBRIDG\36PCF" \ "\37DOTH\40DWDS" /* Atheros protocol-specific flags */ #define IEEE80211_F_ATHEROS \ (IEEE80211_F_FF | IEEE80211_F_COMP | IEEE80211_F_TURBOP) /* Check if an Atheros capability was negotiated for use */ #define IEEE80211_ATH_CAP(vap, ni, bit) \ ((vap)->iv_flags & (ni)->ni_ath_flags & (bit)) /* ic_flags_ext/iv_flags_ext */ #define IEEE80211_FEXT_INACT 0x00000002 /* CONF: sta inact handling */ #define IEEE80211_FEXT_SCANWAIT 0x00000004 /* STATUS: awaiting scan */ /* 0x00000006 reserved */ #define IEEE80211_FEXT_BGSCAN 0x00000008 /* STATUS: complete bgscan */ #define IEEE80211_FEXT_WPS 0x00000010 /* CONF: WPS enabled */ #define IEEE80211_FEXT_TSN 0x00000020 /* CONF: TSN enabled */ #define IEEE80211_FEXT_SCANREQ 0x00000040 /* STATUS: scan req params */ #define IEEE80211_FEXT_RESUME 0x00000080 /* STATUS: start on resume */ #define IEEE80211_FEXT_4ADDR 0x00000100 /* CONF: apply 4-addr encap */ #define IEEE80211_FEXT_NONERP_PR 0x00000200 /* STATUS: non-ERP sta present*/ #define IEEE80211_FEXT_SWBMISS 0x00000400 /* CONF: do bmiss in s/w */ #define IEEE80211_FEXT_DFS 0x00000800 /* CONF: DFS enabled */ #define IEEE80211_FEXT_DOTD 0x00001000 /* CONF: 11d enabled */ #define IEEE80211_FEXT_STATEWAIT 0x00002000 /* STATUS: awaiting state chg */ #define IEEE80211_FEXT_REINIT 0x00004000 /* STATUS: INIT state first */ #define IEEE80211_FEXT_BPF 0x00008000 /* STATUS: BPF tap present */ /* NB: immutable: should be set only when creating a vap */ #define IEEE80211_FEXT_WDSLEGACY 0x00010000 /* CONF: legacy WDS operation */ #define IEEE80211_FEXT_PROBECHAN 0x00020000 /* CONF: probe passive channel*/ #define IEEE80211_FEXT_UNIQMAC 0x00040000 /* CONF: user or computed mac */ #define IEEE80211_FEXT_SCAN_OFFLOAD 0x00080000 /* CONF: scan is fully offloaded */ #define IEEE80211_FEXT_BITS \ "\20\2INACT\3SCANWAIT\4BGSCAN\5WPS\6TSN\7SCANREQ\10RESUME" \ "\0114ADDR\12NONEPR_PR\13SWBMISS\14DFS\15DOTD\16STATEWAIT\17REINIT" \ "\20BPF\21WDSLEGACY\22PROBECHAN\23UNIQMAC\24SCAN_OFFLOAD" /* ic_flags_ht/iv_flags_ht */ #define IEEE80211_FHT_NONHT_PR 0x00000001 /* STATUS: non-HT sta present */ #define IEEE80211_FHT_GF 0x00040000 /* CONF: Greenfield enabled */ #define IEEE80211_FHT_HT 0x00080000 /* CONF: HT supported */ #define IEEE80211_FHT_AMPDU_TX 0x00100000 /* CONF: A-MPDU tx supported */ #define IEEE80211_FHT_AMPDU_RX 0x00200000 /* CONF: A-MPDU rx supported */ #define IEEE80211_FHT_AMSDU_TX 0x00400000 /* CONF: A-MSDU tx supported */ #define IEEE80211_FHT_AMSDU_RX 0x00800000 /* CONF: A-MSDU rx supported */ #define IEEE80211_FHT_USEHT40 0x01000000 /* CONF: 20/40 use enabled */ #define IEEE80211_FHT_PUREN 0x02000000 /* CONF: 11n w/o legacy sta's */ #define IEEE80211_FHT_SHORTGI20 0x04000000 /* CONF: short GI in HT20 */ #define IEEE80211_FHT_SHORTGI40 0x08000000 /* CONF: short GI in HT40 */ #define IEEE80211_FHT_HTCOMPAT 0x10000000 /* CONF: HT vendor OUI's */ #define IEEE80211_FHT_RIFS 0x20000000 /* CONF: RIFS enabled */ #define IEEE80211_FHT_STBC_TX 0x40000000 /* CONF: STBC tx enabled */ #define IEEE80211_FHT_STBC_RX 0x80000000 /* CONF: STBC rx enabled */ #define IEEE80211_FHT_BITS \ "\20\1NONHT_PR" \ "\23GF\24HT\25AMPDU_TX\26AMPDU_TX" \ "\27AMSDU_TX\30AMSDU_RX\31USEHT40\32PUREN\33SHORTGI20\34SHORTGI40" \ "\35HTCOMPAT\36RIFS\37STBC_TX\40STBC_RX" #define IEEE80211_FVEN_BITS "\20" + +#define IEEE80211_FVHT_VHT 0x000000001 /* CONF: VHT supported */ +#define IEEE80211_VFHT_BITS \ + "\20\1VHT" int ic_printf(struct ieee80211com *, const char *, ...) __printflike(2, 3); void ieee80211_ifattach(struct ieee80211com *); void ieee80211_ifdetach(struct ieee80211com *); int ieee80211_vap_setup(struct ieee80211com *, struct ieee80211vap *, const char name[IFNAMSIZ], int unit, enum ieee80211_opmode opmode, int flags, const uint8_t bssid[IEEE80211_ADDR_LEN]); int ieee80211_vap_attach(struct ieee80211vap *, ifm_change_cb_t, ifm_stat_cb_t, const uint8_t macaddr[IEEE80211_ADDR_LEN]); void ieee80211_vap_detach(struct ieee80211vap *); const struct ieee80211_rateset *ieee80211_get_suprates(struct ieee80211com *ic, const struct ieee80211_channel *); void ieee80211_announce(struct ieee80211com *); void ieee80211_announce_channels(struct ieee80211com *); void ieee80211_drain(struct ieee80211com *); void ieee80211_chan_init(struct ieee80211com *); struct ieee80211com *ieee80211_find_vap(const uint8_t mac[IEEE80211_ADDR_LEN]); struct ieee80211com *ieee80211_find_com(const char *name); typedef void ieee80211_com_iter_func(void *, struct ieee80211com *); void ieee80211_iterate_coms(ieee80211_com_iter_func *, void *); int ieee80211_media_change(struct ifnet *); void ieee80211_media_status(struct ifnet *, struct ifmediareq *); int ieee80211_ioctl(struct ifnet *, u_long, caddr_t); int ieee80211_rate2media(struct ieee80211com *, int, enum ieee80211_phymode); int ieee80211_media2rate(int); int ieee80211_mhz2ieee(u_int, u_int); int ieee80211_chan2ieee(struct ieee80211com *, const struct ieee80211_channel *); u_int ieee80211_ieee2mhz(u_int, u_int); int ieee80211_add_channel(struct ieee80211_channel[], int, int *, uint8_t, uint16_t, int8_t, uint32_t, const uint8_t[]); int ieee80211_add_channel_ht40(struct ieee80211_channel[], int, int *, uint8_t, int8_t, uint32_t); uint32_t ieee80211_get_channel_center_freq(const struct ieee80211_channel *); uint32_t ieee80211_get_channel_center_freq1(const struct ieee80211_channel *); uint32_t ieee80211_get_channel_center_freq2(const struct ieee80211_channel *); int ieee80211_add_channel_list_2ghz(struct ieee80211_channel[], int, int *, const uint8_t[], int, const uint8_t[], int); int ieee80211_add_channel_list_5ghz(struct ieee80211_channel[], int, int *, const uint8_t[], int, const uint8_t[], int); struct ieee80211_channel *ieee80211_find_channel(struct ieee80211com *, int freq, int flags); struct ieee80211_channel *ieee80211_find_channel_byieee(struct ieee80211com *, int ieee, int flags); struct ieee80211_channel *ieee80211_lookup_channel_rxstatus(struct ieee80211vap *, const struct ieee80211_rx_stats *); int ieee80211_setmode(struct ieee80211com *, enum ieee80211_phymode); enum ieee80211_phymode ieee80211_chan2mode(const struct ieee80211_channel *); uint32_t ieee80211_mac_hash(const struct ieee80211com *, const uint8_t addr[IEEE80211_ADDR_LEN]); char ieee80211_channel_type_char(const struct ieee80211_channel *c); #define ieee80211_get_current_channel(_ic) ((_ic)->ic_curchan) #define ieee80211_get_home_channel(_ic) ((_ic)->ic_bsschan) #define ieee80211_get_vap_desired_channel(_iv) ((_iv)->iv_des_chan) void ieee80211_radiotap_attach(struct ieee80211com *, struct ieee80211_radiotap_header *th, int tlen, uint32_t tx_radiotap, struct ieee80211_radiotap_header *rh, int rlen, uint32_t rx_radiotap); void ieee80211_radiotap_attachv(struct ieee80211com *, struct ieee80211_radiotap_header *th, int tlen, int n_tx_v, uint32_t tx_radiotap, struct ieee80211_radiotap_header *rh, int rlen, int n_rx_v, uint32_t rx_radiotap); void ieee80211_radiotap_detach(struct ieee80211com *); void ieee80211_radiotap_vattach(struct ieee80211vap *); void ieee80211_radiotap_vdetach(struct ieee80211vap *); void ieee80211_radiotap_chan_change(struct ieee80211com *); void ieee80211_radiotap_tx(struct ieee80211vap *, struct mbuf *); void ieee80211_radiotap_rx(struct ieee80211vap *, struct mbuf *); void ieee80211_radiotap_rx_all(struct ieee80211com *, struct mbuf *); static __inline int ieee80211_radiotap_active(const struct ieee80211com *ic) { return (ic->ic_flags_ext & IEEE80211_FEXT_BPF) != 0; } static __inline int ieee80211_radiotap_active_vap(const struct ieee80211vap *vap) { return (vap->iv_flags_ext & IEEE80211_FEXT_BPF) || vap->iv_ic->ic_montaps != 0; } /* * Enqueue a task on the state thread. */ static __inline void ieee80211_runtask(struct ieee80211com *ic, struct task *task) { taskqueue_enqueue(ic->ic_tq, task); } /* * Wait for a queued task to complete. */ static __inline void ieee80211_draintask(struct ieee80211com *ic, struct task *task) { taskqueue_drain(ic->ic_tq, task); } /* * Key update synchronization methods. XXX should not be visible. */ static __inline void ieee80211_key_update_begin(struct ieee80211vap *vap) { vap->iv_key_update_begin(vap); } static __inline void ieee80211_key_update_end(struct ieee80211vap *vap) { vap->iv_key_update_end(vap); } /* * XXX these need to be here for IEEE80211_F_DATAPAD */ /* * Return the space occupied by the 802.11 header and any * padding required by the driver. This works for a * management or data frame. */ static __inline int ieee80211_hdrspace(struct ieee80211com *ic, const void *data) { int size = ieee80211_hdrsize(data); if (ic->ic_flags & IEEE80211_F_DATAPAD) size = roundup(size, sizeof(uint32_t)); return size; } /* * Like ieee80211_hdrspace, but handles any type of frame. */ static __inline int ieee80211_anyhdrspace(struct ieee80211com *ic, const void *data) { int size = ieee80211_anyhdrsize(data); if (ic->ic_flags & IEEE80211_F_DATAPAD) size = roundup(size, sizeof(uint32_t)); return size; } /* * Notify a vap that beacon state has been updated. */ static __inline void ieee80211_beacon_notify(struct ieee80211vap *vap, int what) { if (vap->iv_state == IEEE80211_S_RUN) vap->iv_update_beacon(vap, what); } /* * Calculate HT channel promotion flags for a channel. * XXX belongs in ieee80211_ht.h but needs IEEE80211_FHT_* */ static __inline int ieee80211_htchanflags(const struct ieee80211_channel *c) { return IEEE80211_IS_CHAN_HT40(c) ? IEEE80211_FHT_HT | IEEE80211_FHT_USEHT40 : IEEE80211_IS_CHAN_HT(c) ? IEEE80211_FHT_HT : 0; } /* * Fetch the current TX power (cap) for the given node. * * This includes the node and ic/vap TX power limit as needed, * but it doesn't take into account any per-rate limit. */ static __inline uint16_t ieee80211_get_node_txpower(struct ieee80211_node *ni) { struct ieee80211com *ic = ni->ni_ic; uint16_t txpower; txpower = ni->ni_txpower; txpower = MIN(txpower, ic->ic_txpowlimit); if (ic->ic_curchan != NULL) { txpower = MIN(txpower, 2 * ic->ic_curchan->ic_maxregpower); txpower = MIN(txpower, ic->ic_curchan->ic_maxpower); } return (txpower); } /* * Debugging facilities compiled in when IEEE80211_DEBUG is defined. * * The intent is that any problem in the net80211 layer can be * diagnosed by inspecting the statistics (dumped by the wlanstats * program) and/or the msgs generated by net80211. Messages are * broken into functional classes and can be controlled with the * wlandebug program. Certain of these msg groups are for facilities * that are no longer part of net80211 (e.g. IEEE80211_MSG_DOT1XSM). */ #define IEEE80211_MSG_11N 0x80000000 /* 11n mode debug */ #define IEEE80211_MSG_DEBUG 0x40000000 /* IFF_DEBUG equivalent */ #define IEEE80211_MSG_DUMPPKTS 0x20000000 /* IFF_LINK2 equivalant */ #define IEEE80211_MSG_CRYPTO 0x10000000 /* crypto work */ #define IEEE80211_MSG_INPUT 0x08000000 /* input handling */ #define IEEE80211_MSG_XRATE 0x04000000 /* rate set handling */ #define IEEE80211_MSG_ELEMID 0x02000000 /* element id parsing */ #define IEEE80211_MSG_NODE 0x01000000 /* node handling */ #define IEEE80211_MSG_ASSOC 0x00800000 /* association handling */ #define IEEE80211_MSG_AUTH 0x00400000 /* authentication handling */ #define IEEE80211_MSG_SCAN 0x00200000 /* scanning */ #define IEEE80211_MSG_OUTPUT 0x00100000 /* output handling */ #define IEEE80211_MSG_STATE 0x00080000 /* state machine */ #define IEEE80211_MSG_POWER 0x00040000 /* power save handling */ #define IEEE80211_MSG_HWMP 0x00020000 /* hybrid mesh protocol */ #define IEEE80211_MSG_DOT1XSM 0x00010000 /* 802.1x state machine */ #define IEEE80211_MSG_RADIUS 0x00008000 /* 802.1x radius client */ #define IEEE80211_MSG_RADDUMP 0x00004000 /* dump 802.1x radius packets */ #define IEEE80211_MSG_MESH 0x00002000 /* mesh networking */ #define IEEE80211_MSG_WPA 0x00001000 /* WPA/RSN protocol */ #define IEEE80211_MSG_ACL 0x00000800 /* ACL handling */ #define IEEE80211_MSG_WME 0x00000400 /* WME protocol */ #define IEEE80211_MSG_SUPERG 0x00000200 /* Atheros SuperG protocol */ #define IEEE80211_MSG_DOTH 0x00000100 /* 802.11h support */ #define IEEE80211_MSG_INACT 0x00000080 /* inactivity handling */ #define IEEE80211_MSG_ROAM 0x00000040 /* sta-mode roaming */ #define IEEE80211_MSG_RATECTL 0x00000020 /* tx rate control */ #define IEEE80211_MSG_ACTION 0x00000010 /* action frame handling */ #define IEEE80211_MSG_WDS 0x00000008 /* WDS handling */ #define IEEE80211_MSG_IOCTL 0x00000004 /* ioctl handling */ #define IEEE80211_MSG_TDMA 0x00000002 /* TDMA handling */ #define IEEE80211_MSG_ANY 0xffffffff /* anything */ #define IEEE80211_MSG_BITS \ "\20\2TDMA\3IOCTL\4WDS\5ACTION\6RATECTL\7ROAM\10INACT\11DOTH\12SUPERG" \ "\13WME\14ACL\15WPA\16RADKEYS\17RADDUMP\20RADIUS\21DOT1XSM\22HWMP" \ "\23POWER\24STATE\25OUTPUT\26SCAN\27AUTH\30ASSOC\31NODE\32ELEMID" \ "\33XRATE\34INPUT\35CRYPTO\36DUPMPKTS\37DEBUG\04011N" #ifdef IEEE80211_DEBUG #define ieee80211_msg(_vap, _m) ((_vap)->iv_debug & (_m)) #define IEEE80211_DPRINTF(_vap, _m, _fmt, ...) do { \ if (ieee80211_msg(_vap, _m)) \ ieee80211_note(_vap, _fmt, __VA_ARGS__); \ } while (0) #define IEEE80211_NOTE(_vap, _m, _ni, _fmt, ...) do { \ if (ieee80211_msg(_vap, _m)) \ ieee80211_note_mac(_vap, (_ni)->ni_macaddr, _fmt, __VA_ARGS__);\ } while (0) #define IEEE80211_NOTE_MAC(_vap, _m, _mac, _fmt, ...) do { \ if (ieee80211_msg(_vap, _m)) \ ieee80211_note_mac(_vap, _mac, _fmt, __VA_ARGS__); \ } while (0) #define IEEE80211_NOTE_FRAME(_vap, _m, _wh, _fmt, ...) do { \ if (ieee80211_msg(_vap, _m)) \ ieee80211_note_frame(_vap, _wh, _fmt, __VA_ARGS__); \ } while (0) void ieee80211_note(const struct ieee80211vap *, const char *, ...); void ieee80211_note_mac(const struct ieee80211vap *, const uint8_t mac[IEEE80211_ADDR_LEN], const char *, ...); void ieee80211_note_frame(const struct ieee80211vap *, const struct ieee80211_frame *, const char *, ...); #define ieee80211_msg_debug(_vap) \ ((_vap)->iv_debug & IEEE80211_MSG_DEBUG) #define ieee80211_msg_dumppkts(_vap) \ ((_vap)->iv_debug & IEEE80211_MSG_DUMPPKTS) #define ieee80211_msg_input(_vap) \ ((_vap)->iv_debug & IEEE80211_MSG_INPUT) #define ieee80211_msg_radius(_vap) \ ((_vap)->iv_debug & IEEE80211_MSG_RADIUS) #define ieee80211_msg_dumpradius(_vap) \ ((_vap)->iv_debug & IEEE80211_MSG_RADDUMP) #define ieee80211_msg_dumpradkeys(_vap) \ ((_vap)->iv_debug & IEEE80211_MSG_RADKEYS) #define ieee80211_msg_scan(_vap) \ ((_vap)->iv_debug & IEEE80211_MSG_SCAN) #define ieee80211_msg_assoc(_vap) \ ((_vap)->iv_debug & IEEE80211_MSG_ASSOC) /* * Emit a debug message about discarding a frame or information * element. One format is for extracting the mac address from * the frame header; the other is for when a header is not * available or otherwise appropriate. */ #define IEEE80211_DISCARD(_vap, _m, _wh, _type, _fmt, ...) do { \ if ((_vap)->iv_debug & (_m)) \ ieee80211_discard_frame(_vap, _wh, _type, _fmt, __VA_ARGS__);\ } while (0) #define IEEE80211_DISCARD_IE(_vap, _m, _wh, _type, _fmt, ...) do { \ if ((_vap)->iv_debug & (_m)) \ ieee80211_discard_ie(_vap, _wh, _type, _fmt, __VA_ARGS__);\ } while (0) #define IEEE80211_DISCARD_MAC(_vap, _m, _mac, _type, _fmt, ...) do { \ if ((_vap)->iv_debug & (_m)) \ ieee80211_discard_mac(_vap, _mac, _type, _fmt, __VA_ARGS__);\ } while (0) void ieee80211_discard_frame(const struct ieee80211vap *, const struct ieee80211_frame *, const char *type, const char *fmt, ...); void ieee80211_discard_ie(const struct ieee80211vap *, const struct ieee80211_frame *, const char *type, const char *fmt, ...); void ieee80211_discard_mac(const struct ieee80211vap *, const uint8_t mac[IEEE80211_ADDR_LEN], const char *type, const char *fmt, ...); #else #define IEEE80211_DPRINTF(_vap, _m, _fmt, ...) #define IEEE80211_NOTE(_vap, _m, _ni, _fmt, ...) #define IEEE80211_NOTE_FRAME(_vap, _m, _wh, _fmt, ...) #define IEEE80211_NOTE_MAC(_vap, _m, _mac, _fmt, ...) #define ieee80211_msg_dumppkts(_vap) 0 #define ieee80211_msg(_vap, _m) 0 #define IEEE80211_DISCARD(_vap, _m, _wh, _type, _fmt, ...) #define IEEE80211_DISCARD_IE(_vap, _m, _wh, _type, _fmt, ...) #define IEEE80211_DISCARD_MAC(_vap, _m, _mac, _type, _fmt, ...) #endif #endif /* _NET80211_IEEE80211_VAR_H_ */ Index: projects/clang400-import/sys/netinet/tcp_hostcache.c =================================================================== --- projects/clang400-import/sys/netinet/tcp_hostcache.c (revision 311459) +++ projects/clang400-import/sys/netinet/tcp_hostcache.c (revision 311460) @@ -1,739 +1,744 @@ /*- * Copyright (c) 2002 Andre Oppermann, Internet Business Solutions AG * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. The name of the author may not be used to endorse or promote * products derived from this software without specific prior written * permission. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR 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. */ /* * The tcp_hostcache moves the tcp-specific cached metrics from the routing * table to a dedicated structure indexed by the remote IP address. It keeps * information on the measured TCP parameters of past TCP sessions to allow * better initial start values to be used with later connections to/from the * same source. Depending on the network parameters (delay, max MTU, * congestion window) between local and remote sites, this can lead to * significant speed-ups for new TCP connections after the first one. * * Due to the tcp_hostcache, all TCP-specific metrics information in the * routing table have been removed. The inpcb no longer keeps a pointer to * the routing entry, and protocol-initiated route cloning has been removed * as well. With these changes, the routing table has gone back to being * more lightwight and only carries information related to packet forwarding. * * tcp_hostcache is designed for multiple concurrent access in SMP * environments and high contention. All bucket rows have their own lock and * thus multiple lookups and modifies can be done at the same time as long as * they are in different bucket rows. If a request for insertion of a new * record can't be satisfied, it simply returns an empty structure. Nobody * and nothing outside of tcp_hostcache.c will ever point directly to any * entry in the tcp_hostcache. All communication is done in an * object-oriented way and only functions of tcp_hostcache will manipulate * hostcache entries. Otherwise, we are unable to achieve good behaviour in * concurrent access situations. Since tcp_hostcache is only caching * information, there are no fatal consequences if we either can't satisfy * any particular request or have to drop/overwrite an existing entry because * of bucket limit memory constrains. */ /* * Many thanks to jlemon for basic structure of tcp_syncache which is being * followed here. */ #include __FBSDID("$FreeBSD$"); #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 #ifdef INET6 #include #include #endif #include #include #include #ifdef INET6 #include #endif #include /* Arbitrary values */ #define TCP_HOSTCACHE_HASHSIZE 512 #define TCP_HOSTCACHE_BUCKETLIMIT 30 #define TCP_HOSTCACHE_EXPIRE 60*60 /* one hour */ #define TCP_HOSTCACHE_PRUNE 5*60 /* every 5 minutes */ static VNET_DEFINE(struct tcp_hostcache, tcp_hostcache); #define V_tcp_hostcache VNET(tcp_hostcache) static VNET_DEFINE(struct callout, tcp_hc_callout); #define V_tcp_hc_callout VNET(tcp_hc_callout) static struct hc_metrics *tcp_hc_lookup(struct in_conninfo *); static struct hc_metrics *tcp_hc_insert(struct in_conninfo *); static int sysctl_tcp_hc_list(SYSCTL_HANDLER_ARGS); static int sysctl_tcp_hc_purgenow(SYSCTL_HANDLER_ARGS); static void tcp_hc_purge_internal(int); static void tcp_hc_purge(void *); static SYSCTL_NODE(_net_inet_tcp, OID_AUTO, hostcache, CTLFLAG_RW, 0, "TCP Host cache"); VNET_DEFINE(int, tcp_use_hostcache) = 1; #define V_tcp_use_hostcache VNET(tcp_use_hostcache) SYSCTL_INT(_net_inet_tcp_hostcache, OID_AUTO, enable, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(tcp_use_hostcache), 0, "Enable the TCP hostcache"); SYSCTL_UINT(_net_inet_tcp_hostcache, OID_AUTO, cachelimit, CTLFLAG_VNET | CTLFLAG_RDTUN, &VNET_NAME(tcp_hostcache.cache_limit), 0, "Overall entry limit for hostcache"); SYSCTL_UINT(_net_inet_tcp_hostcache, OID_AUTO, hashsize, CTLFLAG_VNET | CTLFLAG_RDTUN, &VNET_NAME(tcp_hostcache.hashsize), 0, "Size of TCP hostcache hashtable"); SYSCTL_UINT(_net_inet_tcp_hostcache, OID_AUTO, bucketlimit, CTLFLAG_VNET | CTLFLAG_RDTUN, &VNET_NAME(tcp_hostcache.bucket_limit), 0, "Per-bucket hash limit for hostcache"); SYSCTL_UINT(_net_inet_tcp_hostcache, OID_AUTO, count, CTLFLAG_VNET | CTLFLAG_RD, &VNET_NAME(tcp_hostcache.cache_count), 0, "Current number of entries in hostcache"); SYSCTL_INT(_net_inet_tcp_hostcache, OID_AUTO, expire, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(tcp_hostcache.expire), 0, "Expire time of TCP hostcache entries"); SYSCTL_INT(_net_inet_tcp_hostcache, OID_AUTO, prune, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(tcp_hostcache.prune), 0, "Time between purge runs"); SYSCTL_INT(_net_inet_tcp_hostcache, OID_AUTO, purge, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(tcp_hostcache.purgeall), 0, "Expire all entires on next purge run"); SYSCTL_PROC(_net_inet_tcp_hostcache, OID_AUTO, list, CTLTYPE_STRING | CTLFLAG_RD | CTLFLAG_SKIP, 0, 0, sysctl_tcp_hc_list, "A", "List of all hostcache entries"); SYSCTL_PROC(_net_inet_tcp_hostcache, OID_AUTO, purgenow, CTLTYPE_INT | CTLFLAG_RW, NULL, 0, sysctl_tcp_hc_purgenow, "I", "Immediately purge all entries"); static MALLOC_DEFINE(M_HOSTCACHE, "hostcache", "TCP hostcache"); #define HOSTCACHE_HASH(ip) \ (((ip)->s_addr ^ ((ip)->s_addr >> 7) ^ ((ip)->s_addr >> 17)) & \ V_tcp_hostcache.hashmask) /* XXX: What is the recommended hash to get good entropy for IPv6 addresses? */ #define HOSTCACHE_HASH6(ip6) \ (((ip6)->s6_addr32[0] ^ \ (ip6)->s6_addr32[1] ^ \ (ip6)->s6_addr32[2] ^ \ (ip6)->s6_addr32[3]) & \ V_tcp_hostcache.hashmask) #define THC_LOCK(lp) mtx_lock(lp) #define THC_UNLOCK(lp) mtx_unlock(lp) void tcp_hc_init(void) { u_int cache_limit; int i; /* * Initialize hostcache structures. */ V_tcp_hostcache.cache_count = 0; V_tcp_hostcache.hashsize = TCP_HOSTCACHE_HASHSIZE; V_tcp_hostcache.bucket_limit = TCP_HOSTCACHE_BUCKETLIMIT; V_tcp_hostcache.expire = TCP_HOSTCACHE_EXPIRE; V_tcp_hostcache.prune = TCP_HOSTCACHE_PRUNE; TUNABLE_INT_FETCH("net.inet.tcp.hostcache.hashsize", &V_tcp_hostcache.hashsize); if (!powerof2(V_tcp_hostcache.hashsize)) { printf("WARNING: hostcache hash size is not a power of 2.\n"); V_tcp_hostcache.hashsize = TCP_HOSTCACHE_HASHSIZE; /* default */ } V_tcp_hostcache.hashmask = V_tcp_hostcache.hashsize - 1; TUNABLE_INT_FETCH("net.inet.tcp.hostcache.bucketlimit", &V_tcp_hostcache.bucket_limit); cache_limit = V_tcp_hostcache.hashsize * V_tcp_hostcache.bucket_limit; V_tcp_hostcache.cache_limit = cache_limit; TUNABLE_INT_FETCH("net.inet.tcp.hostcache.cachelimit", &V_tcp_hostcache.cache_limit); if (V_tcp_hostcache.cache_limit > cache_limit) V_tcp_hostcache.cache_limit = cache_limit; /* * Allocate the hash table. */ V_tcp_hostcache.hashbase = (struct hc_head *) malloc(V_tcp_hostcache.hashsize * sizeof(struct hc_head), M_HOSTCACHE, M_WAITOK | M_ZERO); /* * Initialize the hash buckets. */ for (i = 0; i < V_tcp_hostcache.hashsize; i++) { TAILQ_INIT(&V_tcp_hostcache.hashbase[i].hch_bucket); V_tcp_hostcache.hashbase[i].hch_length = 0; mtx_init(&V_tcp_hostcache.hashbase[i].hch_mtx, "tcp_hc_entry", NULL, MTX_DEF); } /* * Allocate the hostcache entries. */ V_tcp_hostcache.zone = uma_zcreate("hostcache", sizeof(struct hc_metrics), NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, 0); uma_zone_set_max(V_tcp_hostcache.zone, V_tcp_hostcache.cache_limit); /* * Set up periodic cache cleanup. */ callout_init(&V_tcp_hc_callout, 1); callout_reset(&V_tcp_hc_callout, V_tcp_hostcache.prune * hz, tcp_hc_purge, curvnet); } #ifdef VIMAGE void tcp_hc_destroy(void) { int i; callout_drain(&V_tcp_hc_callout); /* Purge all hc entries. */ tcp_hc_purge_internal(1); /* Free the uma zone and the allocated hash table. */ uma_zdestroy(V_tcp_hostcache.zone); for (i = 0; i < V_tcp_hostcache.hashsize; i++) mtx_destroy(&V_tcp_hostcache.hashbase[i].hch_mtx); free(V_tcp_hostcache.hashbase, M_HOSTCACHE); } #endif /* * Internal function: look up an entry in the hostcache or return NULL. * * If an entry has been returned, the caller becomes responsible for * unlocking the bucket row after he is done reading/modifying the entry. */ static struct hc_metrics * tcp_hc_lookup(struct in_conninfo *inc) { int hash; struct hc_head *hc_head; struct hc_metrics *hc_entry; if (!V_tcp_use_hostcache) return NULL; KASSERT(inc != NULL, ("tcp_hc_lookup with NULL in_conninfo pointer")); /* * Hash the foreign ip address. */ if (inc->inc_flags & INC_ISIPV6) hash = HOSTCACHE_HASH6(&inc->inc6_faddr); else hash = HOSTCACHE_HASH(&inc->inc_faddr); hc_head = &V_tcp_hostcache.hashbase[hash]; /* * Acquire lock for this bucket row; we release the lock if we don't * find an entry, otherwise the caller has to unlock after he is * done. */ THC_LOCK(&hc_head->hch_mtx); /* * Iterate through entries in bucket row looking for a match. */ TAILQ_FOREACH(hc_entry, &hc_head->hch_bucket, rmx_q) { if (inc->inc_flags & INC_ISIPV6) { /* XXX: check ip6_zoneid */ if (memcmp(&inc->inc6_faddr, &hc_entry->ip6, sizeof(inc->inc6_faddr)) == 0) return hc_entry; } else { if (memcmp(&inc->inc_faddr, &hc_entry->ip4, sizeof(inc->inc_faddr)) == 0) return hc_entry; } } /* * We were unsuccessful and didn't find anything. */ THC_UNLOCK(&hc_head->hch_mtx); return NULL; } /* * Internal function: insert an entry into the hostcache or return NULL if * unable to allocate a new one. * * If an entry has been returned, the caller becomes responsible for * unlocking the bucket row after he is done reading/modifying the entry. */ static struct hc_metrics * tcp_hc_insert(struct in_conninfo *inc) { int hash; struct hc_head *hc_head; struct hc_metrics *hc_entry; if (!V_tcp_use_hostcache) return NULL; KASSERT(inc != NULL, ("tcp_hc_insert with NULL in_conninfo pointer")); /* * Hash the foreign ip address. */ if (inc->inc_flags & INC_ISIPV6) hash = HOSTCACHE_HASH6(&inc->inc6_faddr); else hash = HOSTCACHE_HASH(&inc->inc_faddr); hc_head = &V_tcp_hostcache.hashbase[hash]; /* * Acquire lock for this bucket row; we release the lock if we don't * find an entry, otherwise the caller has to unlock after he is * done. */ THC_LOCK(&hc_head->hch_mtx); /* * If the bucket limit is reached, reuse the least-used element. */ if (hc_head->hch_length >= V_tcp_hostcache.bucket_limit || V_tcp_hostcache.cache_count >= V_tcp_hostcache.cache_limit) { hc_entry = TAILQ_LAST(&hc_head->hch_bucket, hc_qhead); /* * At first we were dropping the last element, just to * reacquire it in the next two lines again, which isn't very * efficient. Instead just reuse the least used element. * We may drop something that is still "in-use" but we can be * "lossy". * Just give up if this bucket row is empty and we don't have * anything to replace. */ if (hc_entry == NULL) { THC_UNLOCK(&hc_head->hch_mtx); return NULL; } TAILQ_REMOVE(&hc_head->hch_bucket, hc_entry, rmx_q); V_tcp_hostcache.hashbase[hash].hch_length--; V_tcp_hostcache.cache_count--; TCPSTAT_INC(tcps_hc_bucketoverflow); #if 0 uma_zfree(V_tcp_hostcache.zone, hc_entry); #endif } else { /* * Allocate a new entry, or balk if not possible. */ hc_entry = uma_zalloc(V_tcp_hostcache.zone, M_NOWAIT); if (hc_entry == NULL) { THC_UNLOCK(&hc_head->hch_mtx); return NULL; } } /* * Initialize basic information of hostcache entry. */ bzero(hc_entry, sizeof(*hc_entry)); if (inc->inc_flags & INC_ISIPV6) { hc_entry->ip6 = inc->inc6_faddr; hc_entry->ip6_zoneid = inc->inc6_zoneid; } else hc_entry->ip4 = inc->inc_faddr; hc_entry->rmx_head = hc_head; hc_entry->rmx_expire = V_tcp_hostcache.expire; /* * Put it upfront. */ TAILQ_INSERT_HEAD(&hc_head->hch_bucket, hc_entry, rmx_q); V_tcp_hostcache.hashbase[hash].hch_length++; V_tcp_hostcache.cache_count++; TCPSTAT_INC(tcps_hc_added); return hc_entry; } /* * External function: look up an entry in the hostcache and fill out the * supplied TCP metrics structure. Fills in NULL when no entry was found or * a value is not set. */ void tcp_hc_get(struct in_conninfo *inc, struct hc_metrics_lite *hc_metrics_lite) { struct hc_metrics *hc_entry; if (!V_tcp_use_hostcache) return; /* * Find the right bucket. */ hc_entry = tcp_hc_lookup(inc); /* * If we don't have an existing object. */ if (hc_entry == NULL) { bzero(hc_metrics_lite, sizeof(*hc_metrics_lite)); return; } hc_entry->rmx_hits++; hc_entry->rmx_expire = V_tcp_hostcache.expire; /* start over again */ hc_metrics_lite->rmx_mtu = hc_entry->rmx_mtu; hc_metrics_lite->rmx_ssthresh = hc_entry->rmx_ssthresh; hc_metrics_lite->rmx_rtt = hc_entry->rmx_rtt; hc_metrics_lite->rmx_rttvar = hc_entry->rmx_rttvar; hc_metrics_lite->rmx_cwnd = hc_entry->rmx_cwnd; hc_metrics_lite->rmx_sendpipe = hc_entry->rmx_sendpipe; hc_metrics_lite->rmx_recvpipe = hc_entry->rmx_recvpipe; /* * Unlock bucket row. */ THC_UNLOCK(&hc_entry->rmx_head->hch_mtx); } /* * External function: look up an entry in the hostcache and return the * discovered path MTU. Returns 0 if no entry is found or value is not * set. */ uint32_t tcp_hc_getmtu(struct in_conninfo *inc) { struct hc_metrics *hc_entry; uint32_t mtu; if (!V_tcp_use_hostcache) return 0; hc_entry = tcp_hc_lookup(inc); if (hc_entry == NULL) { return 0; } hc_entry->rmx_hits++; hc_entry->rmx_expire = V_tcp_hostcache.expire; /* start over again */ mtu = hc_entry->rmx_mtu; THC_UNLOCK(&hc_entry->rmx_head->hch_mtx); return mtu; } /* * External function: update the MTU value of an entry in the hostcache. * Creates a new entry if none was found. */ void tcp_hc_updatemtu(struct in_conninfo *inc, uint32_t mtu) { struct hc_metrics *hc_entry; if (!V_tcp_use_hostcache) return; /* * Find the right bucket. */ hc_entry = tcp_hc_lookup(inc); /* * If we don't have an existing object, try to insert a new one. */ if (hc_entry == NULL) { hc_entry = tcp_hc_insert(inc); if (hc_entry == NULL) return; } hc_entry->rmx_updates++; hc_entry->rmx_expire = V_tcp_hostcache.expire; /* start over again */ hc_entry->rmx_mtu = mtu; /* * Put it upfront so we find it faster next time. */ TAILQ_REMOVE(&hc_entry->rmx_head->hch_bucket, hc_entry, rmx_q); TAILQ_INSERT_HEAD(&hc_entry->rmx_head->hch_bucket, hc_entry, rmx_q); /* * Unlock bucket row. */ THC_UNLOCK(&hc_entry->rmx_head->hch_mtx); } /* * External function: update the TCP metrics of an entry in the hostcache. * Creates a new entry if none was found. */ void tcp_hc_update(struct in_conninfo *inc, struct hc_metrics_lite *hcml) { struct hc_metrics *hc_entry; if (!V_tcp_use_hostcache) return; hc_entry = tcp_hc_lookup(inc); if (hc_entry == NULL) { hc_entry = tcp_hc_insert(inc); if (hc_entry == NULL) return; } hc_entry->rmx_updates++; hc_entry->rmx_expire = V_tcp_hostcache.expire; /* start over again */ if (hcml->rmx_rtt != 0) { if (hc_entry->rmx_rtt == 0) hc_entry->rmx_rtt = hcml->rmx_rtt; else hc_entry->rmx_rtt = ((uint64_t)hc_entry->rmx_rtt + (uint64_t)hcml->rmx_rtt) / 2; TCPSTAT_INC(tcps_cachedrtt); } if (hcml->rmx_rttvar != 0) { if (hc_entry->rmx_rttvar == 0) hc_entry->rmx_rttvar = hcml->rmx_rttvar; else hc_entry->rmx_rttvar = ((uint64_t)hc_entry->rmx_rttvar + (uint64_t)hcml->rmx_rttvar) / 2; TCPSTAT_INC(tcps_cachedrttvar); } if (hcml->rmx_ssthresh != 0) { if (hc_entry->rmx_ssthresh == 0) hc_entry->rmx_ssthresh = hcml->rmx_ssthresh; else hc_entry->rmx_ssthresh = (hc_entry->rmx_ssthresh + hcml->rmx_ssthresh) / 2; TCPSTAT_INC(tcps_cachedssthresh); } if (hcml->rmx_cwnd != 0) { if (hc_entry->rmx_cwnd == 0) hc_entry->rmx_cwnd = hcml->rmx_cwnd; else hc_entry->rmx_cwnd = ((uint64_t)hc_entry->rmx_cwnd + (uint64_t)hcml->rmx_cwnd) / 2; /* TCPSTAT_INC(tcps_cachedcwnd); */ } if (hcml->rmx_sendpipe != 0) { if (hc_entry->rmx_sendpipe == 0) hc_entry->rmx_sendpipe = hcml->rmx_sendpipe; else hc_entry->rmx_sendpipe = ((uint64_t)hc_entry->rmx_sendpipe + (uint64_t)hcml->rmx_sendpipe) /2; /* TCPSTAT_INC(tcps_cachedsendpipe); */ } if (hcml->rmx_recvpipe != 0) { if (hc_entry->rmx_recvpipe == 0) hc_entry->rmx_recvpipe = hcml->rmx_recvpipe; else hc_entry->rmx_recvpipe = ((uint64_t)hc_entry->rmx_recvpipe + (uint64_t)hcml->rmx_recvpipe) /2; /* TCPSTAT_INC(tcps_cachedrecvpipe); */ } TAILQ_REMOVE(&hc_entry->rmx_head->hch_bucket, hc_entry, rmx_q); TAILQ_INSERT_HEAD(&hc_entry->rmx_head->hch_bucket, hc_entry, rmx_q); THC_UNLOCK(&hc_entry->rmx_head->hch_mtx); } /* * Sysctl function: prints the list and values of all hostcache entries in * unsorted order. */ static int sysctl_tcp_hc_list(SYSCTL_HANDLER_ARGS) { const int linesize = 128; struct sbuf sb; int i, error; struct hc_metrics *hc_entry; #ifdef INET6 char ip6buf[INET6_ADDRSTRLEN]; #endif + + if (jailed_without_vnet(curthread->td_ucred) != 0) + return (EPERM); sbuf_new(&sb, NULL, linesize * (V_tcp_hostcache.cache_count + 1), SBUF_INCLUDENUL); sbuf_printf(&sb, "\nIP address MTU SSTRESH RTT RTTVAR " " CWND SENDPIPE RECVPIPE HITS UPD EXP\n"); #define msec(u) (((u) + 500) / 1000) for (i = 0; i < V_tcp_hostcache.hashsize; i++) { THC_LOCK(&V_tcp_hostcache.hashbase[i].hch_mtx); TAILQ_FOREACH(hc_entry, &V_tcp_hostcache.hashbase[i].hch_bucket, rmx_q) { sbuf_printf(&sb, "%-15s %5u %8u %6lums %6lums %8u %8u %8u %4lu " "%4lu %4i\n", hc_entry->ip4.s_addr ? inet_ntoa(hc_entry->ip4) : #ifdef INET6 ip6_sprintf(ip6buf, &hc_entry->ip6), #else "IPv6?", #endif hc_entry->rmx_mtu, hc_entry->rmx_ssthresh, msec((u_long)hc_entry->rmx_rtt * (RTM_RTTUNIT / (hz * TCP_RTT_SCALE))), msec((u_long)hc_entry->rmx_rttvar * (RTM_RTTUNIT / (hz * TCP_RTTVAR_SCALE))), hc_entry->rmx_cwnd, hc_entry->rmx_sendpipe, hc_entry->rmx_recvpipe, hc_entry->rmx_hits, hc_entry->rmx_updates, hc_entry->rmx_expire); } THC_UNLOCK(&V_tcp_hostcache.hashbase[i].hch_mtx); } #undef msec error = sbuf_finish(&sb); if (error == 0) error = SYSCTL_OUT(req, sbuf_data(&sb), sbuf_len(&sb)); sbuf_delete(&sb); return(error); } /* * Caller has to make sure the curvnet is set properly. */ static void tcp_hc_purge_internal(int all) { struct hc_metrics *hc_entry, *hc_next; int i; for (i = 0; i < V_tcp_hostcache.hashsize; i++) { THC_LOCK(&V_tcp_hostcache.hashbase[i].hch_mtx); TAILQ_FOREACH_SAFE(hc_entry, &V_tcp_hostcache.hashbase[i].hch_bucket, rmx_q, hc_next) { if (all || hc_entry->rmx_expire <= 0) { TAILQ_REMOVE(&V_tcp_hostcache.hashbase[i].hch_bucket, hc_entry, rmx_q); uma_zfree(V_tcp_hostcache.zone, hc_entry); V_tcp_hostcache.hashbase[i].hch_length--; V_tcp_hostcache.cache_count--; } else hc_entry->rmx_expire -= V_tcp_hostcache.prune; } THC_UNLOCK(&V_tcp_hostcache.hashbase[i].hch_mtx); } } /* * Expire and purge (old|all) entries in the tcp_hostcache. Runs * periodically from the callout. */ static void tcp_hc_purge(void *arg) { CURVNET_SET((struct vnet *) arg); int all = 0; if (V_tcp_hostcache.purgeall) { all = 1; V_tcp_hostcache.purgeall = 0; } tcp_hc_purge_internal(all); callout_reset(&V_tcp_hc_callout, V_tcp_hostcache.prune * hz, tcp_hc_purge, arg); CURVNET_RESTORE(); } /* * Expire and purge all entries in hostcache immediately. */ static int sysctl_tcp_hc_purgenow(SYSCTL_HANDLER_ARGS) { int error, val; val = 0; error = sysctl_handle_int(oidp, &val, 0, req); if (error || !req->newptr) return (error); tcp_hc_purge_internal(1); callout_reset(&V_tcp_hc_callout, V_tcp_hostcache.prune * hz, tcp_hc_purge, curvnet); return (0); } Index: projects/clang400-import/sys/sys/imgact.h =================================================================== --- projects/clang400-import/sys/sys/imgact.h (revision 311459) +++ projects/clang400-import/sys/sys/imgact.h (revision 311460) @@ -1,113 +1,114 @@ /*- * Copyright (c) 1993, David Greenman * 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 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 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. * * $FreeBSD$ */ #ifndef _SYS_IMGACT_H_ #define _SYS_IMGACT_H_ #include #include #define MAXSHELLCMDLEN PAGE_SIZE struct ucred; struct image_args { char *buf; /* pointer to string buffer */ + void *bufkva; /* cookie for string buffer KVA */ char *begin_argv; /* beginning of argv in buf */ char *begin_envv; /* beginning of envv in buf */ char *endp; /* current `end' pointer of arg & env strings */ char *fname; /* pointer to filename of executable (system space) */ char *fname_buf; /* pointer to optional malloc(M_TEMP) buffer */ int stringspace; /* space left in arg & env buffer */ int argc; /* count of argument strings */ int envc; /* count of environment strings */ int fd; /* file descriptor of the executable */ struct filedesc *fdp; /* new file descriptor table */ }; struct image_params { struct proc *proc; /* our process struct */ struct label *execlabel; /* optional exec label */ struct vnode *vp; /* pointer to vnode of file to exec */ struct vm_object *object; /* The vm object for this vp */ struct vattr *attr; /* attributes of file */ const char *image_header; /* head of file to exec */ unsigned long entry_addr; /* entry address of target executable */ unsigned long reloc_base; /* load address of image */ char vmspace_destroyed; /* flag - we've blown away original vm space */ #define IMGACT_SHELL 0x1 #define IMGACT_BINMISC 0x2 unsigned char interpreted; /* mask of interpreters that have run */ char opened; /* flag - we have opened executable vnode */ char *interpreter_name; /* name of the interpreter */ void *auxargs; /* ELF Auxinfo structure pointer */ struct sf_buf *firstpage; /* first page that we mapped */ unsigned long ps_strings; /* PS_STRINGS for BSD/OS binaries */ size_t auxarg_size; struct image_args *args; /* system call arguments */ struct sysentvec *sysent; /* system entry vector */ char *execpath; unsigned long execpathp; char *freepath; unsigned long canary; int canarylen; unsigned long pagesizes; int pagesizeslen; vm_prot_t stack_prot; u_long stack_sz; struct ucred *newcred; /* new credentials if changing */ bool credential_setid; /* true if becoming setid */ }; #ifdef _KERNEL struct sysentvec; struct thread; struct vmspace; #define IMGACT_CORE_COMPRESS 0x01 int exec_alloc_args(struct image_args *); int exec_check_permissions(struct image_params *); register_t *exec_copyout_strings(struct image_params *); void exec_free_args(struct image_args *); int exec_new_vmspace(struct image_params *, struct sysentvec *); void exec_setregs(struct thread *, struct image_params *, u_long); int exec_shell_imgact(struct image_params *); int exec_copyin_args(struct image_args *, char *, enum uio_seg, char **, char **); int exec_copyin_data_fds(struct thread *, struct image_args *, const void *, size_t, const int *, size_t); int pre_execve(struct thread *td, struct vmspace **oldvmspace); void post_execve(struct thread *td, int error, struct vmspace *oldvmspace); #endif #endif /* !_SYS_IMGACT_H_ */ Index: projects/clang400-import/sys/sys/mount.h =================================================================== --- projects/clang400-import/sys/sys/mount.h (revision 311459) +++ projects/clang400-import/sys/sys/mount.h (revision 311460) @@ -1,958 +1,959 @@ /*- * Copyright (c) 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.h 8.21 (Berkeley) 5/20/95 * $FreeBSD$ */ #ifndef _SYS_MOUNT_H_ #define _SYS_MOUNT_H_ #include #include #ifdef _KERNEL #include #include #include #include #endif /* * NOTE: When changing statfs structure, mount structure, MNT_* flags or * MNTK_* flags also update DDB show mount command in vfs_subr.c. */ typedef struct fsid { int32_t val[2]; } fsid_t; /* filesystem id type */ /* * File identifier. * These are unique per filesystem on a single machine. */ #define MAXFIDSZ 16 struct fid { u_short fid_len; /* length of data in bytes */ u_short fid_data0; /* force longword alignment */ char fid_data[MAXFIDSZ]; /* data (variable length) */ }; /* * filesystem statistics */ #define MFSNAMELEN 16 /* length of type name including null */ #define MNAMELEN 88 /* size of on/from name bufs */ #define STATFS_VERSION 0x20030518 /* current version number */ struct statfs { uint32_t f_version; /* structure version number */ uint32_t f_type; /* type of filesystem */ uint64_t f_flags; /* copy of mount exported flags */ uint64_t f_bsize; /* filesystem fragment size */ uint64_t f_iosize; /* optimal transfer block size */ uint64_t f_blocks; /* total data blocks in filesystem */ uint64_t f_bfree; /* free blocks in filesystem */ int64_t f_bavail; /* free blocks avail to non-superuser */ uint64_t f_files; /* total file nodes in filesystem */ int64_t f_ffree; /* free nodes avail to non-superuser */ uint64_t f_syncwrites; /* count of sync writes since mount */ uint64_t f_asyncwrites; /* count of async writes since mount */ uint64_t f_syncreads; /* count of sync reads since mount */ uint64_t f_asyncreads; /* count of async reads since mount */ uint64_t f_spare[10]; /* unused spare */ uint32_t f_namemax; /* maximum filename length */ uid_t f_owner; /* user that mounted the filesystem */ fsid_t f_fsid; /* filesystem id */ char f_charspare[80]; /* spare string space */ char f_fstypename[MFSNAMELEN]; /* filesystem type name */ char f_mntfromname[MNAMELEN]; /* mounted filesystem */ char f_mntonname[MNAMELEN]; /* directory on which mounted */ }; #ifdef _KERNEL #define OMFSNAMELEN 16 /* length of fs type name, including null */ #define OMNAMELEN (88 - 2 * sizeof(long)) /* size of on/from name bufs */ /* XXX getfsstat.2 is out of date with write and read counter changes here. */ /* XXX statfs.2 is out of date with read counter changes here. */ struct ostatfs { long f_spare2; /* placeholder */ long f_bsize; /* fundamental filesystem block size */ long f_iosize; /* optimal transfer block size */ long f_blocks; /* total data blocks in filesystem */ long f_bfree; /* free blocks in fs */ long f_bavail; /* free blocks avail to non-superuser */ long f_files; /* total file nodes in filesystem */ long f_ffree; /* free file nodes in fs */ fsid_t f_fsid; /* filesystem id */ uid_t f_owner; /* user that mounted the filesystem */ int f_type; /* type of filesystem */ int f_flags; /* copy of mount exported flags */ long f_syncwrites; /* count of sync writes since mount */ long f_asyncwrites; /* count of async writes since mount */ char f_fstypename[OMFSNAMELEN]; /* fs type name */ char f_mntonname[OMNAMELEN]; /* directory on which mounted */ long f_syncreads; /* count of sync reads since mount */ long f_asyncreads; /* count of async reads since mount */ short f_spares1; /* unused spare */ char f_mntfromname[OMNAMELEN];/* mounted filesystem */ short f_spares2; /* unused spare */ /* * XXX on machines where longs are aligned to 8-byte boundaries, there * is an unnamed int32_t here. This spare was after the apparent end * of the struct until we bit off the read counters from f_mntonname. */ long f_spare[2]; /* unused spare */ }; TAILQ_HEAD(vnodelst, vnode); /* Mount options list */ TAILQ_HEAD(vfsoptlist, vfsopt); struct vfsopt { TAILQ_ENTRY(vfsopt) link; char *name; void *value; int len; int pos; int seen; }; /* * Structure per mounted filesystem. Each mounted filesystem has an * array of operations and an instance record. The filesystems are * put on a doubly linked list. * * Lock reference: * l - mnt_listmtx * m - mountlist_mtx * i - interlock * v - vnode freelist mutex * * Unmarked fields are considered stable as long as a ref is held. * */ struct mount { struct mtx mnt_mtx; /* mount structure interlock */ int mnt_gen; /* struct mount generation */ #define mnt_startzero mnt_list TAILQ_ENTRY(mount) mnt_list; /* (m) mount list */ struct vfsops *mnt_op; /* operations on fs */ struct vfsconf *mnt_vfc; /* configuration info */ struct vnode *mnt_vnodecovered; /* vnode we mounted on */ struct vnode *mnt_syncer; /* syncer vnode */ int mnt_ref; /* (i) Reference count */ struct vnodelst mnt_nvnodelist; /* (i) list of vnodes */ int mnt_nvnodelistsize; /* (i) # of vnodes */ int mnt_writeopcount; /* (i) write syscalls pending */ int mnt_kern_flag; /* (i) kernel only flags */ uint64_t mnt_flag; /* (i) flags shared with user */ struct vfsoptlist *mnt_opt; /* current mount options */ struct vfsoptlist *mnt_optnew; /* new options passed to fs */ int mnt_maxsymlinklen; /* max size of short symlink */ struct statfs mnt_stat; /* cache of filesystem stats */ struct ucred *mnt_cred; /* credentials of mounter */ void * mnt_data; /* private data */ time_t mnt_time; /* last time written*/ int mnt_iosize_max; /* max size for clusters, etc */ struct netexport *mnt_export; /* export list */ struct label *mnt_label; /* MAC label for the fs */ u_int mnt_hashseed; /* Random seed for vfs_hash */ int mnt_lockref; /* (i) Lock reference count */ int mnt_secondary_writes; /* (i) # of secondary writes */ int mnt_secondary_accwrites;/* (i) secondary wr. starts */ struct thread *mnt_susp_owner; /* (i) thread owning suspension */ #define mnt_endzero mnt_gjprovider char *mnt_gjprovider; /* gjournal provider name */ struct mtx mnt_listmtx; struct vnodelst mnt_activevnodelist; /* (l) list of active vnodes */ int mnt_activevnodelistsize;/* (l) # of active vnodes */ struct vnodelst mnt_tmpfreevnodelist; /* (l) list of free vnodes */ int mnt_tmpfreevnodelistsize;/* (l) # of free vnodes */ struct lock mnt_explock; /* vfs_export walkers lock */ TAILQ_ENTRY(mount) mnt_upper_link; /* (m) we in the all uppers */ TAILQ_HEAD(, mount) mnt_uppers; /* (m) upper mounts over us*/ }; /* * Definitions for MNT_VNODE_FOREACH_ALL. */ struct vnode *__mnt_vnode_next_all(struct vnode **mvp, struct mount *mp); struct vnode *__mnt_vnode_first_all(struct vnode **mvp, struct mount *mp); void __mnt_vnode_markerfree_all(struct vnode **mvp, struct mount *mp); #define MNT_VNODE_FOREACH_ALL(vp, mp, mvp) \ for (vp = __mnt_vnode_first_all(&(mvp), (mp)); \ (vp) != NULL; vp = __mnt_vnode_next_all(&(mvp), (mp))) #define MNT_VNODE_FOREACH_ALL_ABORT(mp, mvp) \ do { \ MNT_ILOCK(mp); \ __mnt_vnode_markerfree_all(&(mvp), (mp)); \ /* MNT_IUNLOCK(mp); -- done in above function */ \ mtx_assert(MNT_MTX(mp), MA_NOTOWNED); \ } while (0) /* * Definitions for MNT_VNODE_FOREACH_ACTIVE. */ struct vnode *__mnt_vnode_next_active(struct vnode **mvp, struct mount *mp); struct vnode *__mnt_vnode_first_active(struct vnode **mvp, struct mount *mp); void __mnt_vnode_markerfree_active(struct vnode **mvp, struct mount *); #define MNT_VNODE_FOREACH_ACTIVE(vp, mp, mvp) \ for (vp = __mnt_vnode_first_active(&(mvp), (mp)); \ (vp) != NULL; vp = __mnt_vnode_next_active(&(mvp), (mp))) #define MNT_VNODE_FOREACH_ACTIVE_ABORT(mp, mvp) \ __mnt_vnode_markerfree_active(&(mvp), (mp)) #define MNT_ILOCK(mp) mtx_lock(&(mp)->mnt_mtx) #define MNT_ITRYLOCK(mp) mtx_trylock(&(mp)->mnt_mtx) #define MNT_IUNLOCK(mp) mtx_unlock(&(mp)->mnt_mtx) #define MNT_MTX(mp) (&(mp)->mnt_mtx) #define MNT_REF(mp) (mp)->mnt_ref++ #define MNT_REL(mp) do { \ KASSERT((mp)->mnt_ref > 0, ("negative mnt_ref")); \ (mp)->mnt_ref--; \ if ((mp)->mnt_ref == 0) \ wakeup((mp)); \ } while (0) #endif /* _KERNEL */ /* * User specifiable flags, stored in mnt_flag. */ #define MNT_RDONLY 0x0000000000000001ULL /* read only filesystem */ #define MNT_SYNCHRONOUS 0x0000000000000002ULL /* fs written synchronously */ #define MNT_NOEXEC 0x0000000000000004ULL /* can't exec from filesystem */ #define MNT_NOSUID 0x0000000000000008ULL /* don't honor setuid fs bits */ #define MNT_NFS4ACLS 0x0000000000000010ULL /* enable NFS version 4 ACLs */ #define MNT_UNION 0x0000000000000020ULL /* union with underlying fs */ #define MNT_ASYNC 0x0000000000000040ULL /* fs written asynchronously */ #define MNT_SUIDDIR 0x0000000000100000ULL /* special SUID dir handling */ #define MNT_SOFTDEP 0x0000000000200000ULL /* using soft updates */ #define MNT_NOSYMFOLLOW 0x0000000000400000ULL /* do not follow symlinks */ #define MNT_GJOURNAL 0x0000000002000000ULL /* GEOM journal support enabled */ #define MNT_MULTILABEL 0x0000000004000000ULL /* MAC support for objects */ #define MNT_ACLS 0x0000000008000000ULL /* ACL support enabled */ #define MNT_NOATIME 0x0000000010000000ULL /* dont update file access time */ #define MNT_NOCLUSTERR 0x0000000040000000ULL /* disable cluster read */ #define MNT_NOCLUSTERW 0x0000000080000000ULL /* disable cluster write */ #define MNT_SUJ 0x0000000100000000ULL /* using journaled soft updates */ #define MNT_AUTOMOUNTED 0x0000000200000000ULL /* mounted by automountd(8) */ /* * NFS export related mount flags. */ #define MNT_EXRDONLY 0x0000000000000080ULL /* exported read only */ #define MNT_EXPORTED 0x0000000000000100ULL /* filesystem is exported */ #define MNT_DEFEXPORTED 0x0000000000000200ULL /* exported to the world */ #define MNT_EXPORTANON 0x0000000000000400ULL /* anon uid mapping for all */ #define MNT_EXKERB 0x0000000000000800ULL /* exported with Kerberos */ #define MNT_EXPUBLIC 0x0000000020000000ULL /* public export (WebNFS) */ /* * Flags set by internal operations, * but visible to the user. * XXX some of these are not quite right.. (I've never seen the root flag set) */ #define MNT_LOCAL 0x0000000000001000ULL /* filesystem is stored locally */ #define MNT_QUOTA 0x0000000000002000ULL /* quotas are enabled on fs */ #define MNT_ROOTFS 0x0000000000004000ULL /* identifies the root fs */ #define MNT_USER 0x0000000000008000ULL /* mounted by a user */ #define MNT_IGNORE 0x0000000000800000ULL /* do not show entry in df */ /* * Mask of flags that are visible to statfs(). * XXX I think that this could now become (~(MNT_CMDFLAGS)) * but the 'mount' program may need changing to handle this. */ #define MNT_VISFLAGMASK (MNT_RDONLY | MNT_SYNCHRONOUS | MNT_NOEXEC | \ MNT_NOSUID | MNT_UNION | MNT_SUJ | \ MNT_ASYNC | MNT_EXRDONLY | MNT_EXPORTED | \ MNT_DEFEXPORTED | MNT_EXPORTANON| MNT_EXKERB | \ MNT_LOCAL | MNT_USER | MNT_QUOTA | \ MNT_ROOTFS | MNT_NOATIME | MNT_NOCLUSTERR| \ MNT_NOCLUSTERW | MNT_SUIDDIR | MNT_SOFTDEP | \ MNT_IGNORE | MNT_EXPUBLIC | MNT_NOSYMFOLLOW | \ MNT_GJOURNAL | MNT_MULTILABEL | MNT_ACLS | \ MNT_NFS4ACLS | MNT_AUTOMOUNTED) /* Mask of flags that can be updated. */ #define MNT_UPDATEMASK (MNT_NOSUID | MNT_NOEXEC | \ MNT_SYNCHRONOUS | MNT_UNION | MNT_ASYNC | \ MNT_NOATIME | \ MNT_NOSYMFOLLOW | MNT_IGNORE | \ MNT_NOCLUSTERR | MNT_NOCLUSTERW | MNT_SUIDDIR | \ MNT_ACLS | MNT_USER | MNT_NFS4ACLS | \ MNT_AUTOMOUNTED) /* * External filesystem command modifier flags. * Unmount can use the MNT_FORCE flag. * XXX: These are not STATES and really should be somewhere else. * XXX: MNT_BYFSID and MNT_NONBUSY collide with MNT_ACLS and MNT_MULTILABEL, * but because MNT_ACLS and MNT_MULTILABEL are only used for mount(2), * and MNT_BYFSID and MNT_NONBUSY are only used for unmount(2), * it's harmless. */ #define MNT_UPDATE 0x0000000000010000ULL /* not real mount, just update */ #define MNT_DELEXPORT 0x0000000000020000ULL /* delete export host lists */ #define MNT_RELOAD 0x0000000000040000ULL /* reload filesystem data */ #define MNT_FORCE 0x0000000000080000ULL /* force unmount or readonly */ #define MNT_SNAPSHOT 0x0000000001000000ULL /* snapshot the filesystem */ #define MNT_NONBUSY 0x0000000004000000ULL /* check vnode use counts. */ #define MNT_BYFSID 0x0000000008000000ULL /* specify filesystem by ID. */ #define MNT_CMDFLAGS (MNT_UPDATE | MNT_DELEXPORT | MNT_RELOAD | \ MNT_FORCE | MNT_SNAPSHOT | MNT_NONBUSY | \ MNT_BYFSID) /* * Internal filesystem control flags stored in mnt_kern_flag. * * MNTK_UNMOUNT locks the mount entry so that name lookup cannot proceed * past the mount point. This keeps the subtree stable during mounts * and unmounts. * * MNTK_UNMOUNTF permits filesystems to detect a forced unmount while * dounmount() is still waiting to lock the mountpoint. This allows * the filesystem to cancel operations that might otherwise deadlock * with the unmount attempt (used by NFS). * * MNTK_NOINSMNTQ is strict subset of MNTK_UNMOUNT. They are separated * to allow for failed unmount attempt to restore the syncer vnode for * the mount. */ #define MNTK_UNMOUNTF 0x00000001 /* forced unmount in progress */ #define MNTK_ASYNC 0x00000002 /* filtered async flag */ #define MNTK_SOFTDEP 0x00000004 /* async disabled by softdep */ #define MNTK_NOINSMNTQ 0x00000008 /* insmntque is not allowed */ #define MNTK_DRAINING 0x00000010 /* lock draining is happening */ #define MNTK_REFEXPIRE 0x00000020 /* refcount expiring is happening */ #define MNTK_EXTENDED_SHARED 0x00000040 /* Allow shared locking for more ops */ #define MNTK_SHARED_WRITES 0x00000080 /* Allow shared locking for writes */ #define MNTK_NO_IOPF 0x00000100 /* Disallow page faults during reads and writes. Filesystem shall properly handle i/o state on EFAULT. */ #define MNTK_VGONE_UPPER 0x00000200 #define MNTK_VGONE_WAITER 0x00000400 #define MNTK_LOOKUP_EXCL_DOTDOT 0x00000800 #define MNTK_MARKER 0x00001000 #define MNTK_UNMAPPED_BUFS 0x00002000 #define MNTK_USES_BCACHE 0x00004000 /* FS uses the buffer cache. */ #define MNTK_NOASYNC 0x00800000 /* disable async */ #define MNTK_UNMOUNT 0x01000000 /* unmount in progress */ #define MNTK_MWAIT 0x02000000 /* waiting for unmount to finish */ #define MNTK_SUSPEND 0x08000000 /* request write suspension */ #define MNTK_SUSPEND2 0x04000000 /* block secondary writes */ #define MNTK_SUSPENDED 0x10000000 /* write operations are suspended */ #define MNTK_NULL_NOCACHE 0x20000000 /* auto disable cache for nullfs mounts over this fs */ #define MNTK_LOOKUP_SHARED 0x40000000 /* FS supports shared lock lookups */ #define MNTK_NOKNOTE 0x80000000 /* Don't send KNOTEs from VOP hooks */ #ifdef _KERNEL static inline int MNT_SHARED_WRITES(struct mount *mp) { return (mp != NULL && (mp->mnt_kern_flag & MNTK_SHARED_WRITES) != 0); } static inline int MNT_EXTENDED_SHARED(struct mount *mp) { return (mp != NULL && (mp->mnt_kern_flag & MNTK_EXTENDED_SHARED) != 0); } #endif /* * Sysctl CTL_VFS definitions. * * Second level identifier specifies which filesystem. Second level * identifier VFS_VFSCONF returns information about all filesystems. * Second level identifier VFS_GENERIC is non-terminal. */ #define VFS_VFSCONF 0 /* get configured filesystems */ #define VFS_GENERIC 0 /* generic filesystem information */ /* * Third level identifiers for VFS_GENERIC are given below; third * level identifiers for specific filesystems are given in their * mount specific header files. */ #define VFS_MAXTYPENUM 1 /* int: highest defined filesystem type */ #define VFS_CONF 2 /* struct: vfsconf for filesystem given as next argument */ /* * Flags for various system call interfaces. * * waitfor flags to vfs_sync() and getfsstat() */ #define MNT_WAIT 1 /* synchronously wait for I/O to complete */ #define MNT_NOWAIT 2 /* start all I/O, but do not wait for it */ #define MNT_LAZY 3 /* push data not written by filesystem syncer */ #define MNT_SUSPEND 4 /* Suspend file system after sync */ /* * Generic file handle */ struct fhandle { fsid_t fh_fsid; /* Filesystem id of mount point */ struct fid fh_fid; /* Filesys specific id */ }; typedef struct fhandle fhandle_t; /* * Old export arguments without security flavor list */ struct oexport_args { int ex_flags; /* export related flags */ uid_t ex_root; /* mapping for root uid */ struct xucred ex_anon; /* mapping for anonymous user */ struct sockaddr *ex_addr; /* net address to which exported */ u_char ex_addrlen; /* and the net address length */ struct sockaddr *ex_mask; /* mask of valid bits in saddr */ u_char ex_masklen; /* and the smask length */ char *ex_indexfile; /* index file for WebNFS URLs */ }; /* * Export arguments for local filesystem mount calls. */ #define MAXSECFLAVORS 5 struct export_args { int ex_flags; /* export related flags */ uid_t ex_root; /* mapping for root uid */ struct xucred ex_anon; /* mapping for anonymous user */ struct sockaddr *ex_addr; /* net address to which exported */ u_char ex_addrlen; /* and the net address length */ struct sockaddr *ex_mask; /* mask of valid bits in saddr */ u_char ex_masklen; /* and the smask length */ char *ex_indexfile; /* index file for WebNFS URLs */ int ex_numsecflavors; /* security flavor count */ int ex_secflavors[MAXSECFLAVORS]; /* list of security flavors */ }; /* * Structure holding information for a publicly exported filesystem * (WebNFS). Currently the specs allow just for one such filesystem. */ struct nfs_public { int np_valid; /* Do we hold valid information */ fhandle_t np_handle; /* Filehandle for pub fs (internal) */ struct mount *np_mount; /* Mountpoint of exported fs */ char *np_index; /* Index file */ }; /* * Filesystem configuration information. One of these exists for each * type of filesystem supported by the kernel. These are searched at * mount time to identify the requested filesystem. * * XXX: Never change the first two arguments! */ struct vfsconf { u_int vfc_version; /* ABI version number */ char vfc_name[MFSNAMELEN]; /* filesystem type name */ struct vfsops *vfc_vfsops; /* filesystem operations vector */ int vfc_typenum; /* historic filesystem type number */ int vfc_refcount; /* number mounted of this type */ int vfc_flags; /* permanent flags */ struct vfsoptdecl *vfc_opts; /* mount options */ TAILQ_ENTRY(vfsconf) vfc_list; /* list of vfscons */ }; /* Userland version of the struct vfsconf. */ struct xvfsconf { struct vfsops *vfc_vfsops; /* filesystem operations vector */ char vfc_name[MFSNAMELEN]; /* filesystem type name */ int vfc_typenum; /* historic filesystem type number */ int vfc_refcount; /* number mounted of this type */ int vfc_flags; /* permanent flags */ struct vfsconf *vfc_next; /* next in list */ }; #ifndef BURN_BRIDGES struct ovfsconf { void *vfc_vfsops; char vfc_name[32]; int vfc_index; int vfc_refcount; int vfc_flags; }; #endif /* * NB: these flags refer to IMPLEMENTATION properties, not properties of * any actual mounts; i.e., it does not make sense to change the flags. */ #define VFCF_STATIC 0x00010000 /* statically compiled into kernel */ #define VFCF_NETWORK 0x00020000 /* may get data over the network */ #define VFCF_READONLY 0x00040000 /* writes are not implemented */ #define VFCF_SYNTHETIC 0x00080000 /* data does not represent real files */ #define VFCF_LOOPBACK 0x00100000 /* aliases some other mounted FS */ #define VFCF_UNICODE 0x00200000 /* stores file names as Unicode */ #define VFCF_JAIL 0x00400000 /* can be mounted from within a jail */ #define VFCF_DELEGADMIN 0x00800000 /* supports delegated administration */ #define VFCF_SBDRY 0x01000000 /* defer stop requests */ typedef uint32_t fsctlop_t; struct vfsidctl { int vc_vers; /* should be VFSIDCTL_VERS1 (below) */ fsid_t vc_fsid; /* fsid to operate on */ char vc_fstypename[MFSNAMELEN]; /* type of fs 'nfs' or '*' */ fsctlop_t vc_op; /* operation VFS_CTL_* (below) */ void *vc_ptr; /* pointer to data structure */ size_t vc_len; /* sizeof said structure */ u_int32_t vc_spare[12]; /* spare (must be zero) */ }; /* vfsidctl API version. */ #define VFS_CTL_VERS1 0x01 /* * New style VFS sysctls, do not reuse/conflict with the namespace for * private sysctls. * All "global" sysctl ops have the 33rd bit set: * 0x...1.... * Private sysctl ops should have the 33rd bit unset. */ #define VFS_CTL_QUERY 0x00010001 /* anything wrong? (vfsquery) */ #define VFS_CTL_TIMEO 0x00010002 /* set timeout for vfs notification */ #define VFS_CTL_NOLOCKS 0x00010003 /* disable file locking */ struct vfsquery { u_int32_t vq_flags; u_int32_t vq_spare[31]; }; /* vfsquery flags */ #define VQ_NOTRESP 0x0001 /* server down */ #define VQ_NEEDAUTH 0x0002 /* server bad auth */ #define VQ_LOWDISK 0x0004 /* we're low on space */ #define VQ_MOUNT 0x0008 /* new filesystem arrived */ #define VQ_UNMOUNT 0x0010 /* filesystem has left */ #define VQ_DEAD 0x0020 /* filesystem is dead, needs force unmount */ #define VQ_ASSIST 0x0040 /* filesystem needs assistance from external program */ #define VQ_NOTRESPLOCK 0x0080 /* server lockd down */ #define VQ_FLAG0100 0x0100 /* placeholder */ #define VQ_FLAG0200 0x0200 /* placeholder */ #define VQ_FLAG0400 0x0400 /* placeholder */ #define VQ_FLAG0800 0x0800 /* placeholder */ #define VQ_FLAG1000 0x1000 /* placeholder */ #define VQ_FLAG2000 0x2000 /* placeholder */ #define VQ_FLAG4000 0x4000 /* placeholder */ #define VQ_FLAG8000 0x8000 /* placeholder */ #ifdef _KERNEL /* Point a sysctl request at a vfsidctl's data. */ #define VCTLTOREQ(vc, req) \ do { \ (req)->newptr = (vc)->vc_ptr; \ (req)->newlen = (vc)->vc_len; \ (req)->newidx = 0; \ } while (0) #endif struct iovec; struct uio; #ifdef _KERNEL /* * vfs_busy specific flags and mask. */ #define MBF_NOWAIT 0x01 #define MBF_MNTLSTLOCK 0x02 #define MBF_MASK (MBF_NOWAIT | MBF_MNTLSTLOCK) #ifdef MALLOC_DECLARE MALLOC_DECLARE(M_MOUNT); +MALLOC_DECLARE(M_STATFS); #endif extern int maxvfsconf; /* highest defined filesystem type */ TAILQ_HEAD(vfsconfhead, vfsconf); extern struct vfsconfhead vfsconf; /* * Operations supported on mounted filesystem. */ struct mount_args; struct nameidata; struct sysctl_req; struct mntarg; typedef int vfs_cmount_t(struct mntarg *ma, void *data, uint64_t flags); typedef int vfs_unmount_t(struct mount *mp, int mntflags); typedef int vfs_root_t(struct mount *mp, int flags, struct vnode **vpp); typedef int vfs_quotactl_t(struct mount *mp, int cmds, uid_t uid, void *arg); typedef int vfs_statfs_t(struct mount *mp, struct statfs *sbp); typedef int vfs_sync_t(struct mount *mp, int waitfor); typedef int vfs_vget_t(struct mount *mp, ino_t ino, int flags, struct vnode **vpp); typedef int vfs_fhtovp_t(struct mount *mp, struct fid *fhp, int flags, struct vnode **vpp); typedef int vfs_checkexp_t(struct mount *mp, struct sockaddr *nam, int *extflagsp, struct ucred **credanonp, int *numsecflavors, int **secflavors); typedef int vfs_init_t(struct vfsconf *); typedef int vfs_uninit_t(struct vfsconf *); typedef int vfs_extattrctl_t(struct mount *mp, int cmd, struct vnode *filename_vp, int attrnamespace, const char *attrname); typedef int vfs_mount_t(struct mount *mp); typedef int vfs_sysctl_t(struct mount *mp, fsctlop_t op, struct sysctl_req *req); typedef void vfs_susp_clean_t(struct mount *mp); typedef void vfs_notify_lowervp_t(struct mount *mp, struct vnode *lowervp); typedef void vfs_purge_t(struct mount *mp); struct vfsops { vfs_mount_t *vfs_mount; vfs_cmount_t *vfs_cmount; vfs_unmount_t *vfs_unmount; vfs_root_t *vfs_root; vfs_quotactl_t *vfs_quotactl; vfs_statfs_t *vfs_statfs; vfs_sync_t *vfs_sync; vfs_vget_t *vfs_vget; vfs_fhtovp_t *vfs_fhtovp; vfs_checkexp_t *vfs_checkexp; vfs_init_t *vfs_init; vfs_uninit_t *vfs_uninit; vfs_extattrctl_t *vfs_extattrctl; vfs_sysctl_t *vfs_sysctl; vfs_susp_clean_t *vfs_susp_clean; vfs_notify_lowervp_t *vfs_reclaim_lowervp; vfs_notify_lowervp_t *vfs_unlink_lowervp; vfs_purge_t *vfs_purge; vfs_mount_t *vfs_spare[6]; /* spares for ABI compat */ }; vfs_statfs_t __vfs_statfs; #define VFS_PROLOGUE(MP) do { \ struct mount *mp__; \ int _prev_stops; \ \ mp__ = (MP); \ _prev_stops = sigdeferstop((mp__ != NULL && \ (mp__->mnt_vfc->vfc_flags & VFCF_SBDRY) != 0) ? \ SIGDEFERSTOP_SILENT : SIGDEFERSTOP_NOP); #define VFS_EPILOGUE(MP) \ sigallowstop(_prev_stops); \ } while (0) #define VFS_MOUNT(MP) ({ \ int _rc; \ \ VFS_PROLOGUE(MP); \ _rc = (*(MP)->mnt_op->vfs_mount)(MP); \ VFS_EPILOGUE(MP); \ _rc; }) #define VFS_UNMOUNT(MP, FORCE) ({ \ int _rc; \ \ VFS_PROLOGUE(MP); \ _rc = (*(MP)->mnt_op->vfs_unmount)(MP, FORCE); \ VFS_EPILOGUE(MP); \ _rc; }) #define VFS_ROOT(MP, FLAGS, VPP) ({ \ int _rc; \ \ VFS_PROLOGUE(MP); \ _rc = (*(MP)->mnt_op->vfs_root)(MP, FLAGS, VPP); \ VFS_EPILOGUE(MP); \ _rc; }) #define VFS_QUOTACTL(MP, C, U, A) ({ \ int _rc; \ \ VFS_PROLOGUE(MP); \ _rc = (*(MP)->mnt_op->vfs_quotactl)(MP, C, U, A); \ VFS_EPILOGUE(MP); \ _rc; }) #define VFS_STATFS(MP, SBP) ({ \ int _rc; \ \ VFS_PROLOGUE(MP); \ _rc = __vfs_statfs((MP), (SBP)); \ VFS_EPILOGUE(MP); \ _rc; }) #define VFS_SYNC(MP, WAIT) ({ \ int _rc; \ \ VFS_PROLOGUE(MP); \ _rc = (*(MP)->mnt_op->vfs_sync)(MP, WAIT); \ VFS_EPILOGUE(MP); \ _rc; }) #define VFS_VGET(MP, INO, FLAGS, VPP) ({ \ int _rc; \ \ VFS_PROLOGUE(MP); \ _rc = (*(MP)->mnt_op->vfs_vget)(MP, INO, FLAGS, VPP); \ VFS_EPILOGUE(MP); \ _rc; }) #define VFS_FHTOVP(MP, FIDP, FLAGS, VPP) ({ \ int _rc; \ \ VFS_PROLOGUE(MP); \ _rc = (*(MP)->mnt_op->vfs_fhtovp)(MP, FIDP, FLAGS, VPP); \ VFS_EPILOGUE(MP); \ _rc; }) #define VFS_CHECKEXP(MP, NAM, EXFLG, CRED, NUMSEC, SEC) ({ \ int _rc; \ \ VFS_PROLOGUE(MP); \ _rc = (*(MP)->mnt_op->vfs_checkexp)(MP, NAM, EXFLG, CRED, NUMSEC,\ SEC); \ VFS_EPILOGUE(MP); \ _rc; }) #define VFS_EXTATTRCTL(MP, C, FN, NS, N) ({ \ int _rc; \ \ VFS_PROLOGUE(MP); \ _rc = (*(MP)->mnt_op->vfs_extattrctl)(MP, C, FN, NS, N); \ VFS_EPILOGUE(MP); \ _rc; }) #define VFS_SYSCTL(MP, OP, REQ) ({ \ int _rc; \ \ VFS_PROLOGUE(MP); \ _rc = (*(MP)->mnt_op->vfs_sysctl)(MP, OP, REQ); \ VFS_EPILOGUE(MP); \ _rc; }) #define VFS_SUSP_CLEAN(MP) do { \ if (*(MP)->mnt_op->vfs_susp_clean != NULL) { \ VFS_PROLOGUE(MP); \ (*(MP)->mnt_op->vfs_susp_clean)(MP); \ VFS_EPILOGUE(MP); \ } \ } while (0) #define VFS_RECLAIM_LOWERVP(MP, VP) do { \ if (*(MP)->mnt_op->vfs_reclaim_lowervp != NULL) { \ VFS_PROLOGUE(MP); \ (*(MP)->mnt_op->vfs_reclaim_lowervp)((MP), (VP)); \ VFS_EPILOGUE(MP); \ } \ } while (0) #define VFS_UNLINK_LOWERVP(MP, VP) do { \ if (*(MP)->mnt_op->vfs_unlink_lowervp != NULL) { \ VFS_PROLOGUE(MP); \ (*(MP)->mnt_op->vfs_unlink_lowervp)((MP), (VP)); \ VFS_EPILOGUE(MP); \ } \ } while (0) #define VFS_PURGE(MP) do { \ if (*(MP)->mnt_op->vfs_purge != NULL) { \ VFS_PROLOGUE(MP); \ (*(MP)->mnt_op->vfs_purge)(MP); \ VFS_EPILOGUE(MP); \ } \ } while (0) #define VFS_KNOTE_LOCKED(vp, hint) do \ { \ if (((vp)->v_vflag & VV_NOKNOTE) == 0) \ VN_KNOTE((vp), (hint), KNF_LISTLOCKED); \ } while (0) #define VFS_KNOTE_UNLOCKED(vp, hint) do \ { \ if (((vp)->v_vflag & VV_NOKNOTE) == 0) \ VN_KNOTE((vp), (hint), 0); \ } while (0) #define VFS_NOTIFY_UPPER_RECLAIM 1 #define VFS_NOTIFY_UPPER_UNLINK 2 #include /* * Version numbers. */ #define VFS_VERSION_00 0x19660120 #define VFS_VERSION_01 0x20121030 #define VFS_VERSION VFS_VERSION_01 #define VFS_SET(vfsops, fsname, flags) \ static struct vfsconf fsname ## _vfsconf = { \ .vfc_version = VFS_VERSION, \ .vfc_name = #fsname, \ .vfc_vfsops = &vfsops, \ .vfc_typenum = -1, \ .vfc_flags = flags, \ }; \ static moduledata_t fsname ## _mod = { \ #fsname, \ vfs_modevent, \ & fsname ## _vfsconf \ }; \ DECLARE_MODULE(fsname, fsname ## _mod, SI_SUB_VFS, SI_ORDER_MIDDLE) /* * exported vnode operations */ int dounmount(struct mount *, int, struct thread *); int kernel_mount(struct mntarg *ma, uint64_t flags); int kernel_vmount(int flags, ...); struct mntarg *mount_arg(struct mntarg *ma, const char *name, const void *val, int len); struct mntarg *mount_argb(struct mntarg *ma, int flag, const char *name); struct mntarg *mount_argf(struct mntarg *ma, const char *name, const char *fmt, ...); struct mntarg *mount_argsu(struct mntarg *ma, const char *name, const void *val, int len); void statfs_scale_blocks(struct statfs *sf, long max_size); struct vfsconf *vfs_byname(const char *); struct vfsconf *vfs_byname_kld(const char *, struct thread *td, int *); void vfs_mount_destroy(struct mount *); void vfs_event_signal(fsid_t *, u_int32_t, intptr_t); void vfs_freeopts(struct vfsoptlist *opts); void vfs_deleteopt(struct vfsoptlist *opts, const char *name); int vfs_buildopts(struct uio *auio, struct vfsoptlist **options); int vfs_flagopt(struct vfsoptlist *opts, const char *name, uint64_t *w, uint64_t val); int vfs_getopt(struct vfsoptlist *, const char *, void **, int *); int vfs_getopt_pos(struct vfsoptlist *opts, const char *name); int vfs_getopt_size(struct vfsoptlist *opts, const char *name, off_t *value); char *vfs_getopts(struct vfsoptlist *, const char *, int *error); int vfs_copyopt(struct vfsoptlist *, const char *, void *, int); int vfs_filteropt(struct vfsoptlist *, const char **legal); void vfs_opterror(struct vfsoptlist *opts, const char *fmt, ...); int vfs_scanopt(struct vfsoptlist *opts, const char *name, const char *fmt, ...); int vfs_setopt(struct vfsoptlist *opts, const char *name, void *value, int len); int vfs_setopt_part(struct vfsoptlist *opts, const char *name, void *value, int len); int vfs_setopts(struct vfsoptlist *opts, const char *name, const char *value); int vfs_setpublicfs /* set publicly exported fs */ (struct mount *, struct netexport *, struct export_args *); void vfs_msync(struct mount *, int); int vfs_busy(struct mount *, int); int vfs_export /* process mount export info */ (struct mount *, struct export_args *); void vfs_allocate_syncvnode(struct mount *); void vfs_deallocate_syncvnode(struct mount *); int vfs_donmount(struct thread *td, uint64_t fsflags, struct uio *fsoptions); void vfs_getnewfsid(struct mount *); struct cdev *vfs_getrootfsid(struct mount *); struct mount *vfs_getvfs(fsid_t *); /* return vfs given fsid */ struct mount *vfs_busyfs(fsid_t *); int vfs_modevent(module_t, int, void *); void vfs_mount_error(struct mount *, const char *, ...); void vfs_mountroot(void); /* mount our root filesystem */ void vfs_mountedfrom(struct mount *, const char *from); void vfs_notify_upper(struct vnode *, int); void vfs_oexport_conv(const struct oexport_args *oexp, struct export_args *exp); void vfs_ref(struct mount *); void vfs_rel(struct mount *); struct mount *vfs_mount_alloc(struct vnode *, struct vfsconf *, const char *, struct ucred *); int vfs_suser(struct mount *, struct thread *); void vfs_unbusy(struct mount *); void vfs_unmountall(void); extern TAILQ_HEAD(mntlist, mount) mountlist; /* mounted filesystem list */ extern struct mtx mountlist_mtx; extern struct nfs_public nfs_pub; extern struct sx vfsconf_sx; #define vfsconf_lock() sx_xlock(&vfsconf_sx) #define vfsconf_unlock() sx_xunlock(&vfsconf_sx) #define vfsconf_slock() sx_slock(&vfsconf_sx) #define vfsconf_sunlock() sx_sunlock(&vfsconf_sx) /* * Declarations for these vfs default operations are located in * kern/vfs_default.c. They will be automatically used to replace * null entries in VFS ops tables when registering a new filesystem * type in the global table. */ vfs_root_t vfs_stdroot; vfs_quotactl_t vfs_stdquotactl; vfs_statfs_t vfs_stdstatfs; vfs_sync_t vfs_stdsync; vfs_sync_t vfs_stdnosync; vfs_vget_t vfs_stdvget; vfs_fhtovp_t vfs_stdfhtovp; vfs_checkexp_t vfs_stdcheckexp; vfs_init_t vfs_stdinit; vfs_uninit_t vfs_stduninit; vfs_extattrctl_t vfs_stdextattrctl; vfs_sysctl_t vfs_stdsysctl; void syncer_suspend(void); void syncer_resume(void); #else /* !_KERNEL */ #include struct stat; __BEGIN_DECLS int fhopen(const struct fhandle *, int); int fhstat(const struct fhandle *, struct stat *); int fhstatfs(const struct fhandle *, struct statfs *); int fstatfs(int, struct statfs *); int getfh(const char *, fhandle_t *); int getfsstat(struct statfs *, long, int); int getmntinfo(struct statfs **, int); int lgetfh(const char *, fhandle_t *); int mount(const char *, const char *, int, void *); int nmount(struct iovec *, unsigned int, int); int statfs(const char *, struct statfs *); int unmount(const char *, int); /* C library stuff */ int getvfsbyname(const char *, struct xvfsconf *); __END_DECLS #endif /* _KERNEL */ #endif /* !_SYS_MOUNT_H_ */ Index: projects/clang400-import/sys/vm/vm_init.c =================================================================== --- projects/clang400-import/sys/vm/vm_init.c (revision 311459) +++ projects/clang400-import/sys/vm/vm_init.c (revision 311460) @@ -1,280 +1,285 @@ /*- * Copyright (c) 1991, 1993 * The Regents of the University of California. All rights reserved. * * This code is derived from software contributed to Berkeley by * The Mach Operating System project at Carnegie-Mellon University. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * from: @(#)vm_init.c 8.1 (Berkeley) 6/11/93 * * * Copyright (c) 1987, 1990 Carnegie-Mellon University. * All rights reserved. * * Authors: Avadis Tevanian, Jr., Michael Wayne Young * * Permission to use, copy, modify and distribute this software and * its documentation is hereby granted, provided that both the copyright * notice and this permission notice appear in all copies of the * software, derivative works or modified versions, and any portions * thereof, and that both notices appear in supporting documentation. * * CARNEGIE MELLON ALLOWS FREE USE OF THIS SOFTWARE IN ITS "AS IS" * CONDITION. CARNEGIE MELLON DISCLAIMS ANY LIABILITY OF ANY KIND * FOR ANY DAMAGES WHATSOEVER RESULTING FROM THE USE OF THIS SOFTWARE. * * Carnegie Mellon requests users of this software to return to * * Software Distribution Coordinator or Software.Distribution@CS.CMU.EDU * School of Computer Science * Carnegie Mellon University * Pittsburgh PA 15213-3890 * * any improvements or extensions that they make and grant Carnegie the * rights to redistribute these changes. */ /* * Initialize the Virtual Memory subsystem. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include long physmem; -static int exec_map_entries = 16; -SYSCTL_INT(_vm, OID_AUTO, exec_map_entries, CTLFLAG_RDTUN, &exec_map_entries, 0, - "Maximum number of simultaneous execs"); - /* * System initialization */ static void vm_mem_init(void *); SYSINIT(vm_mem, SI_SUB_VM, SI_ORDER_FIRST, vm_mem_init, NULL); /* * Import kva into the kernel arena. */ static int kva_import(void *unused, vmem_size_t size, int flags, vmem_addr_t *addrp) { vm_offset_t addr; int result; addr = vm_map_min(kernel_map); result = vm_map_find(kernel_map, NULL, 0, &addr, size, 0, VMFS_SUPER_SPACE, VM_PROT_ALL, VM_PROT_ALL, MAP_NOFAULT); if (result != KERN_SUCCESS) return (ENOMEM); *addrp = addr; return (0); } /* * vm_init initializes the virtual memory system. * This is done only by the first cpu up. * * The start and end address of physical memory is passed in. */ /* ARGSUSED*/ static void vm_mem_init(dummy) void *dummy; { /* * Initializes resident memory structures. From here on, all physical * memory is accounted for, and we use only virtual addresses. */ vm_set_page_size(); virtual_avail = vm_page_startup(virtual_avail); /* * Initialize other VM packages */ vmem_startup(); vm_object_init(); vm_map_startup(); kmem_init(virtual_avail, virtual_end); /* * Initialize the kernel_arena. This can grow on demand. */ vmem_init(kernel_arena, "kernel arena", 0, 0, PAGE_SIZE, 0, 0); vmem_set_import(kernel_arena, kva_import, NULL, NULL, #if VM_NRESERVLEVEL > 0 1 << (VM_LEVEL_0_ORDER + PAGE_SHIFT)); #else /* On non-superpage architectures want large import sizes. */ PAGE_SIZE * 1024); #endif kmem_init_zero_region(); pmap_init(); vm_pager_init(); } void vm_ksubmap_init(struct kva_md_info *kmi) { vm_offset_t firstaddr; caddr_t v; vm_size_t size = 0; long physmem_est; vm_offset_t minaddr; vm_offset_t maxaddr; /* * Allocate space for system data structures. * The first available kernel virtual address is in "v". * As pages of kernel virtual memory are allocated, "v" is incremented. * As pages of memory are allocated and cleared, * "firstaddr" is incremented. */ /* * Make two passes. The first pass calculates how much memory is * needed and allocates it. The second pass assigns virtual * addresses to the various data structures. */ firstaddr = 0; again: v = (caddr_t)firstaddr; /* * Discount the physical memory larger than the size of kernel_map * to avoid eating up all of KVA space. */ physmem_est = lmin(physmem, btoc(kernel_map->max_offset - kernel_map->min_offset)); v = kern_vfs_bio_buffer_alloc(v, physmem_est); /* * End of first pass, size has been calculated so allocate memory */ if (firstaddr == 0) { size = (vm_size_t)v; #ifdef VM_FREELIST_DMA32 /* * Try to protect 32-bit DMAable memory from the largest * early alloc of wired mem. */ firstaddr = kmem_alloc_attr(kernel_arena, size, M_ZERO | M_NOWAIT, (vm_paddr_t)1 << 32, ~(vm_paddr_t)0, VM_MEMATTR_DEFAULT); if (firstaddr == 0) #endif firstaddr = kmem_malloc(kernel_arena, size, M_ZERO | M_WAITOK); if (firstaddr == 0) panic("startup: no room for tables"); goto again; } /* * End of second pass, addresses have been assigned */ if ((vm_size_t)((char *)v - firstaddr) != size) panic("startup: table size inconsistency"); /* * Allocate the clean map to hold all of the paging and I/O virtual * memory. */ size = (long)nbuf * BKVASIZE + (long)nswbuf * MAXPHYS + (long)bio_transient_maxcnt * MAXPHYS; kmi->clean_sva = firstaddr = kva_alloc(size); kmi->clean_eva = firstaddr + size; /* * Allocate the buffer arena. * * Enable the quantum cache if we have more than 4 cpus. This * avoids lock contention at the expense of some fragmentation. */ size = (long)nbuf * BKVASIZE; kmi->buffer_sva = firstaddr; kmi->buffer_eva = kmi->buffer_sva + size; vmem_init(buffer_arena, "buffer arena", kmi->buffer_sva, size, PAGE_SIZE, (mp_ncpus > 4) ? BKVASIZE * 8 : 0, 0); firstaddr += size; /* * Now swap kva. */ swapbkva = firstaddr; size = (long)nswbuf * MAXPHYS; firstaddr += size; /* * And optionally transient bio space. */ if (bio_transient_maxcnt != 0) { size = (long)bio_transient_maxcnt * MAXPHYS; vmem_init(transient_arena, "transient arena", firstaddr, size, PAGE_SIZE, 0, 0); firstaddr += size; } if (firstaddr != kmi->clean_eva) panic("Clean map calculation incorrect"); /* - * Allocate the pageable submaps. + * Allocate the pageable submaps. We may cache an exec map entry per + * CPU, so we therefore need to reserve space for at least ncpu+1 + * entries to avoid deadlock. The exec map is also used by some image + * activators, so we leave a fixed number of pages for their use. */ +#ifdef __LP64__ + exec_map_entries = 8 * mp_ncpus; +#else + exec_map_entries = 2 * mp_ncpus + 4; +#endif + exec_map_entry_size = round_page(PATH_MAX + ARG_MAX); exec_map = kmem_suballoc(kernel_map, &minaddr, &maxaddr, - exec_map_entries * round_page(PATH_MAX + ARG_MAX), FALSE); + exec_map_entries * exec_map_entry_size + 64 * PAGE_SIZE, FALSE); pipe_map = kmem_suballoc(kernel_map, &minaddr, &maxaddr, maxpipekva, FALSE); } Index: projects/clang400-import/sys/vm/vm_kern.c =================================================================== --- projects/clang400-import/sys/vm/vm_kern.c (revision 311459) +++ projects/clang400-import/sys/vm/vm_kern.c (revision 311460) @@ -1,559 +1,562 @@ /*- * Copyright (c) 1991, 1993 * The Regents of the University of California. All rights reserved. * * This code is derived from software contributed to Berkeley by * The Mach Operating System project at Carnegie-Mellon University. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * from: @(#)vm_kern.c 8.3 (Berkeley) 1/12/94 * * * Copyright (c) 1987, 1990 Carnegie-Mellon University. * All rights reserved. * * Authors: Avadis Tevanian, Jr., Michael Wayne Young * * Permission to use, copy, modify and distribute this software and * its documentation is hereby granted, provided that both the copyright * notice and this permission notice appear in all copies of the * software, derivative works or modified versions, and any portions * thereof, and that both notices appear in supporting documentation. * * CARNEGIE MELLON ALLOWS FREE USE OF THIS SOFTWARE IN ITS "AS IS" * CONDITION. CARNEGIE MELLON DISCLAIMS ANY LIABILITY OF ANY KIND * FOR ANY DAMAGES WHATSOEVER RESULTING FROM THE USE OF THIS SOFTWARE. * * Carnegie Mellon requests users of this software to return to * * Software Distribution Coordinator or Software.Distribution@CS.CMU.EDU * School of Computer Science * Carnegie Mellon University * Pittsburgh PA 15213-3890 * * any improvements or extensions that they make and grant Carnegie the * rights to redistribute these changes. */ /* * Kernel memory management. */ #include __FBSDID("$FreeBSD$"); #include #include #include /* for ticks and hz */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include vm_map_t kernel_map; vm_map_t exec_map; vm_map_t pipe_map; const void *zero_region; CTASSERT((ZERO_REGION_SIZE & PAGE_MASK) == 0); /* NB: Used by kernel debuggers. */ const u_long vm_maxuser_address = VM_MAXUSER_ADDRESS; +u_int exec_map_entry_size; +u_int exec_map_entries; + SYSCTL_ULONG(_vm, OID_AUTO, min_kernel_address, CTLFLAG_RD, SYSCTL_NULL_ULONG_PTR, VM_MIN_KERNEL_ADDRESS, "Min kernel address"); SYSCTL_ULONG(_vm, OID_AUTO, max_kernel_address, CTLFLAG_RD, #if defined(__arm__) || defined(__sparc64__) &vm_max_kernel_address, 0, #else SYSCTL_NULL_ULONG_PTR, VM_MAX_KERNEL_ADDRESS, #endif "Max kernel address"); /* * kva_alloc: * * Allocate a virtual address range with no underlying object and * no initial mapping to physical memory. Any mapping from this * range to physical memory must be explicitly created prior to * its use, typically with pmap_qenter(). Any attempt to create * a mapping on demand through vm_fault() will result in a panic. */ vm_offset_t kva_alloc(size) vm_size_t size; { vm_offset_t addr; size = round_page(size); if (vmem_alloc(kernel_arena, size, M_BESTFIT | M_NOWAIT, &addr)) return (0); return (addr); } /* * kva_free: * * Release a region of kernel virtual memory allocated * with kva_alloc, and return the physical pages * associated with that region. * * This routine may not block on kernel maps. */ void kva_free(addr, size) vm_offset_t addr; vm_size_t size; { size = round_page(size); vmem_free(kernel_arena, addr, size); } /* * Allocates a region from the kernel address map and physical pages * within the specified address range to the kernel object. Creates a * wired mapping from this region to these pages, and returns the * region's starting virtual address. The allocated pages are not * necessarily physically contiguous. If M_ZERO is specified through the * given flags, then the pages are zeroed before they are mapped. */ vm_offset_t kmem_alloc_attr(vmem_t *vmem, vm_size_t size, int flags, vm_paddr_t low, vm_paddr_t high, vm_memattr_t memattr) { vm_object_t object = vmem == kmem_arena ? kmem_object : kernel_object; vm_offset_t addr, i; vm_ooffset_t offset; vm_page_t m; int pflags, tries; size = round_page(size); if (vmem_alloc(vmem, size, M_BESTFIT | flags, &addr)) return (0); offset = addr - VM_MIN_KERNEL_ADDRESS; pflags = malloc2vm_flags(flags) | VM_ALLOC_NOBUSY | VM_ALLOC_WIRED; VM_OBJECT_WLOCK(object); for (i = 0; i < size; i += PAGE_SIZE) { tries = 0; retry: m = vm_page_alloc_contig(object, OFF_TO_IDX(offset + i), pflags, 1, low, high, PAGE_SIZE, 0, memattr); if (m == NULL) { VM_OBJECT_WUNLOCK(object); if (tries < ((flags & M_NOWAIT) != 0 ? 1 : 3)) { if (!vm_page_reclaim_contig(pflags, 1, low, high, PAGE_SIZE, 0) && (flags & M_WAITOK) != 0) VM_WAIT; VM_OBJECT_WLOCK(object); tries++; goto retry; } kmem_unback(object, addr, i); vmem_free(vmem, addr, size); return (0); } if ((flags & M_ZERO) && (m->flags & PG_ZERO) == 0) pmap_zero_page(m); m->valid = VM_PAGE_BITS_ALL; pmap_enter(kernel_pmap, addr + i, m, VM_PROT_ALL, VM_PROT_ALL | PMAP_ENTER_WIRED, 0); } VM_OBJECT_WUNLOCK(object); return (addr); } /* * Allocates a region from the kernel address map and physically * contiguous pages within the specified address range to the kernel * object. Creates a wired mapping from this region to these pages, and * returns the region's starting virtual address. If M_ZERO is specified * through the given flags, then the pages are zeroed before they are * mapped. */ vm_offset_t kmem_alloc_contig(struct vmem *vmem, vm_size_t size, int flags, vm_paddr_t low, vm_paddr_t high, u_long alignment, vm_paddr_t boundary, vm_memattr_t memattr) { vm_object_t object = vmem == kmem_arena ? kmem_object : kernel_object; vm_offset_t addr, tmp; vm_ooffset_t offset; vm_page_t end_m, m; u_long npages; int pflags, tries; size = round_page(size); if (vmem_alloc(vmem, size, flags | M_BESTFIT, &addr)) return (0); offset = addr - VM_MIN_KERNEL_ADDRESS; pflags = malloc2vm_flags(flags) | VM_ALLOC_NOBUSY | VM_ALLOC_WIRED; npages = atop(size); VM_OBJECT_WLOCK(object); tries = 0; retry: m = vm_page_alloc_contig(object, OFF_TO_IDX(offset), pflags, npages, low, high, alignment, boundary, memattr); if (m == NULL) { VM_OBJECT_WUNLOCK(object); if (tries < ((flags & M_NOWAIT) != 0 ? 1 : 3)) { if (!vm_page_reclaim_contig(pflags, npages, low, high, alignment, boundary) && (flags & M_WAITOK) != 0) VM_WAIT; VM_OBJECT_WLOCK(object); tries++; goto retry; } vmem_free(vmem, addr, size); return (0); } end_m = m + npages; tmp = addr; for (; m < end_m; m++) { if ((flags & M_ZERO) && (m->flags & PG_ZERO) == 0) pmap_zero_page(m); m->valid = VM_PAGE_BITS_ALL; pmap_enter(kernel_pmap, tmp, m, VM_PROT_ALL, VM_PROT_ALL | PMAP_ENTER_WIRED, 0); tmp += PAGE_SIZE; } VM_OBJECT_WUNLOCK(object); return (addr); } /* * kmem_suballoc: * * Allocates a map to manage a subrange * of the kernel virtual address space. * * Arguments are as follows: * * parent Map to take range from * min, max Returned endpoints of map * size Size of range to find * superpage_align Request that min is superpage aligned */ vm_map_t kmem_suballoc(vm_map_t parent, vm_offset_t *min, vm_offset_t *max, vm_size_t size, boolean_t superpage_align) { int ret; vm_map_t result; size = round_page(size); *min = vm_map_min(parent); ret = vm_map_find(parent, NULL, 0, min, size, 0, superpage_align ? VMFS_SUPER_SPACE : VMFS_ANY_SPACE, VM_PROT_ALL, VM_PROT_ALL, MAP_ACC_NO_CHARGE); if (ret != KERN_SUCCESS) panic("kmem_suballoc: bad status return of %d", ret); *max = *min + size; result = vm_map_create(vm_map_pmap(parent), *min, *max); if (result == NULL) panic("kmem_suballoc: cannot create submap"); if (vm_map_submap(parent, *min, *max, result) != KERN_SUCCESS) panic("kmem_suballoc: unable to change range to submap"); return (result); } /* * kmem_malloc: * * Allocate wired-down pages in the kernel's address space. */ vm_offset_t kmem_malloc(struct vmem *vmem, vm_size_t size, int flags) { vm_offset_t addr; int rv; size = round_page(size); if (vmem_alloc(vmem, size, flags | M_BESTFIT, &addr)) return (0); rv = kmem_back((vmem == kmem_arena) ? kmem_object : kernel_object, addr, size, flags); if (rv != KERN_SUCCESS) { vmem_free(vmem, addr, size); return (0); } return (addr); } /* * kmem_back: * * Allocate physical pages for the specified virtual address range. */ int kmem_back(vm_object_t object, vm_offset_t addr, vm_size_t size, int flags) { vm_offset_t offset, i; vm_page_t m; int pflags; KASSERT(object == kmem_object || object == kernel_object, ("kmem_back: only supports kernel objects.")); offset = addr - VM_MIN_KERNEL_ADDRESS; pflags = malloc2vm_flags(flags) | VM_ALLOC_NOBUSY | VM_ALLOC_WIRED; VM_OBJECT_WLOCK(object); for (i = 0; i < size; i += PAGE_SIZE) { retry: m = vm_page_alloc(object, OFF_TO_IDX(offset + i), pflags); /* * Ran out of space, free everything up and return. Don't need * to lock page queues here as we know that the pages we got * aren't on any queues. */ if (m == NULL) { VM_OBJECT_WUNLOCK(object); if ((flags & M_NOWAIT) == 0) { VM_WAIT; VM_OBJECT_WLOCK(object); goto retry; } kmem_unback(object, addr, i); return (KERN_NO_SPACE); } if (flags & M_ZERO && (m->flags & PG_ZERO) == 0) pmap_zero_page(m); KASSERT((m->oflags & VPO_UNMANAGED) != 0, ("kmem_malloc: page %p is managed", m)); m->valid = VM_PAGE_BITS_ALL; pmap_enter(kernel_pmap, addr + i, m, VM_PROT_ALL, VM_PROT_ALL | PMAP_ENTER_WIRED, 0); } VM_OBJECT_WUNLOCK(object); return (KERN_SUCCESS); } /* * kmem_unback: * * Unmap and free the physical pages underlying the specified virtual * address range. * * A physical page must exist within the specified object at each index * that is being unmapped. */ void kmem_unback(vm_object_t object, vm_offset_t addr, vm_size_t size) { vm_page_t m; vm_offset_t i, offset; KASSERT(object == kmem_object || object == kernel_object, ("kmem_unback: only supports kernel objects.")); pmap_remove(kernel_pmap, addr, addr + size); offset = addr - VM_MIN_KERNEL_ADDRESS; VM_OBJECT_WLOCK(object); for (i = 0; i < size; i += PAGE_SIZE) { m = vm_page_lookup(object, OFF_TO_IDX(offset + i)); vm_page_unwire(m, PQ_NONE); vm_page_free(m); } VM_OBJECT_WUNLOCK(object); } /* * kmem_free: * * Free memory allocated with kmem_malloc. The size must match the * original allocation. */ void kmem_free(struct vmem *vmem, vm_offset_t addr, vm_size_t size) { size = round_page(size); kmem_unback((vmem == kmem_arena) ? kmem_object : kernel_object, addr, size); vmem_free(vmem, addr, size); } /* * kmap_alloc_wait: * * Allocates pageable memory from a sub-map of the kernel. If the submap * has no room, the caller sleeps waiting for more memory in the submap. * * This routine may block. */ vm_offset_t kmap_alloc_wait(map, size) vm_map_t map; vm_size_t size; { vm_offset_t addr; size = round_page(size); if (!swap_reserve(size)) return (0); for (;;) { /* * To make this work for more than one map, use the map's lock * to lock out sleepers/wakers. */ vm_map_lock(map); if (vm_map_findspace(map, vm_map_min(map), size, &addr) == 0) break; /* no space now; see if we can ever get space */ if (vm_map_max(map) - vm_map_min(map) < size) { vm_map_unlock(map); swap_release(size); return (0); } map->needs_wakeup = TRUE; vm_map_unlock_and_wait(map, 0); } vm_map_insert(map, NULL, 0, addr, addr + size, VM_PROT_ALL, VM_PROT_ALL, MAP_ACC_CHARGED); vm_map_unlock(map); return (addr); } /* * kmap_free_wakeup: * * Returns memory to a submap of the kernel, and wakes up any processes * waiting for memory in that map. */ void kmap_free_wakeup(map, addr, size) vm_map_t map; vm_offset_t addr; vm_size_t size; { vm_map_lock(map); (void) vm_map_delete(map, trunc_page(addr), round_page(addr + size)); if (map->needs_wakeup) { map->needs_wakeup = FALSE; vm_map_wakeup(map); } vm_map_unlock(map); } void kmem_init_zero_region(void) { vm_offset_t addr, i; vm_page_t m; /* * Map a single physical page of zeros to a larger virtual range. * This requires less looping in places that want large amounts of * zeros, while not using much more physical resources. */ addr = kva_alloc(ZERO_REGION_SIZE); m = vm_page_alloc(NULL, 0, VM_ALLOC_NORMAL | VM_ALLOC_NOOBJ | VM_ALLOC_WIRED | VM_ALLOC_ZERO); if ((m->flags & PG_ZERO) == 0) pmap_zero_page(m); for (i = 0; i < ZERO_REGION_SIZE; i += PAGE_SIZE) pmap_qenter(addr + i, &m, 1); pmap_protect(kernel_pmap, addr, addr + ZERO_REGION_SIZE, VM_PROT_READ); zero_region = (const void *)addr; } /* * kmem_init: * * Create the kernel map; insert a mapping covering kernel text, * data, bss, and all space allocated thus far (`boostrap' data). The * new map will thus map the range between VM_MIN_KERNEL_ADDRESS and * `start' as allocated, and the range between `start' and `end' as free. */ void kmem_init(start, end) vm_offset_t start, end; { vm_map_t m; m = vm_map_create(kernel_pmap, VM_MIN_KERNEL_ADDRESS, end); m->system_map = 1; vm_map_lock(m); /* N.B.: cannot use kgdb to debug, starting with this assignment ... */ kernel_map = m; (void) vm_map_insert(m, NULL, (vm_ooffset_t) 0, #ifdef __amd64__ KERNBASE, #else VM_MIN_KERNEL_ADDRESS, #endif start, VM_PROT_ALL, VM_PROT_ALL, MAP_NOFAULT); /* ... and ending with the completion of the above `insert' */ vm_map_unlock(m); } #ifdef DIAGNOSTIC /* * Allow userspace to directly trigger the VM drain routine for testing * purposes. */ static int debug_vm_lowmem(SYSCTL_HANDLER_ARGS) { int error, i; i = 0; error = sysctl_handle_int(oidp, &i, 0, req); if (error) return (error); if (i) EVENTHANDLER_INVOKE(vm_lowmem, 0); return (0); } SYSCTL_PROC(_debug, OID_AUTO, vm_lowmem, CTLTYPE_INT | CTLFLAG_RW, 0, 0, debug_vm_lowmem, "I", "set to trigger vm_lowmem event"); #endif Index: projects/clang400-import/sys/vm/vm_kern.h =================================================================== --- projects/clang400-import/sys/vm/vm_kern.h (revision 311459) +++ projects/clang400-import/sys/vm/vm_kern.h (revision 311460) @@ -1,78 +1,80 @@ /*- * Copyright (c) 1991, 1993 * The Regents of the University of California. All rights reserved. * * This code is derived from software contributed to Berkeley by * The Mach Operating System project at Carnegie-Mellon University. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * from: @(#)vm_kern.h 8.1 (Berkeley) 6/11/93 * * * Copyright (c) 1987, 1990 Carnegie-Mellon University. * All rights reserved. * * Authors: Avadis Tevanian, Jr., Michael Wayne Young * * Permission to use, copy, modify and distribute this software and * its documentation is hereby granted, provided that both the copyright * notice and this permission notice appear in all copies of the * software, derivative works or modified versions, and any portions * thereof, and that both notices appear in supporting documentation. * * CARNEGIE MELLON ALLOWS FREE USE OF THIS SOFTWARE IN ITS "AS IS" * CONDITION. CARNEGIE MELLON DISCLAIMS ANY LIABILITY OF ANY KIND * FOR ANY DAMAGES WHATSOEVER RESULTING FROM THE USE OF THIS SOFTWARE. * * Carnegie Mellon requests users of this software to return to * * Software Distribution Coordinator or Software.Distribution@CS.CMU.EDU * School of Computer Science * Carnegie Mellon University * Pittsburgh PA 15213-3890 * * any improvements or extensions that they make and grant Carnegie the * rights to redistribute these changes. * * $FreeBSD$ */ #ifndef _VM_VM_KERN_H_ -#define _VM_VM_KERN_H_ 1 +#define _VM_VM_KERN_H_ /* Kernel memory management definitions. */ extern vm_map_t kernel_map; extern vm_map_t exec_map; extern vm_map_t pipe_map; extern struct vmem *kernel_arena; extern struct vmem *kmem_arena; extern struct vmem *buffer_arena; extern struct vmem *transient_arena; extern struct vmem *memguard_arena; extern vm_offset_t swapbkva; extern u_long vm_kmem_size; +extern u_int exec_map_entries; +extern u_int exec_map_entry_size; -#endif /* _VM_VM_KERN_H_ */ +#endif /* _VM_VM_KERN_H_ */ Index: projects/clang400-import/sys/vm/vm_page.c =================================================================== --- projects/clang400-import/sys/vm/vm_page.c (revision 311459) +++ projects/clang400-import/sys/vm/vm_page.c (revision 311460) @@ -1,3592 +1,3600 @@ /*- * Copyright (c) 1991 Regents of the University of California. * All rights reserved. * Copyright (c) 1998 Matthew Dillon. All Rights Reserved. * * This code is derived from software contributed to Berkeley by * The Mach Operating System project at Carnegie-Mellon University. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * from: @(#)vm_page.c 7.4 (Berkeley) 5/7/91 */ /*- * Copyright (c) 1987, 1990 Carnegie-Mellon University. * All rights reserved. * * Authors: Avadis Tevanian, Jr., Michael Wayne Young * * Permission to use, copy, modify and distribute this software and * its documentation is hereby granted, provided that both the copyright * notice and this permission notice appear in all copies of the * software, derivative works or modified versions, and any portions * thereof, and that both notices appear in supporting documentation. * * CARNEGIE MELLON ALLOWS FREE USE OF THIS SOFTWARE IN ITS "AS IS" * CONDITION. CARNEGIE MELLON DISCLAIMS ANY LIABILITY OF ANY KIND * FOR ANY DAMAGES WHATSOEVER RESULTING FROM THE USE OF THIS SOFTWARE. * * Carnegie Mellon requests users of this software to return to * * Software Distribution Coordinator or Software.Distribution@CS.CMU.EDU * School of Computer Science * Carnegie Mellon University * Pittsburgh PA 15213-3890 * * any improvements or extensions that they make and grant Carnegie the * rights to redistribute these changes. */ /* * GENERAL RULES ON VM_PAGE MANIPULATION * * - A page queue lock is required when adding or removing a page from a * page queue regardless of other locks or the busy state of a page. * * * In general, no thread besides the page daemon can acquire or * hold more than one page queue lock at a time. * * * The page daemon can acquire and hold any pair of page queue * locks in any order. * * - The object lock is required when inserting or removing * pages from an object (vm_page_insert() or vm_page_remove()). * */ /* * Resident memory management module. */ #include __FBSDID("$FreeBSD$"); #include "opt_vm.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 /* * Associated with page of user-allocatable memory is a * page structure. */ struct vm_domain vm_dom[MAXMEMDOM]; struct mtx_padalign vm_page_queue_free_mtx; struct mtx_padalign pa_lock[PA_LOCK_COUNT]; +/* + * bogus page -- for I/O to/from partially complete buffers, + * or for paging into sparsely invalid regions. + */ +vm_page_t bogus_page; + vm_page_t vm_page_array; long vm_page_array_size; long first_page; static int boot_pages = UMA_BOOT_PAGES; SYSCTL_INT(_vm, OID_AUTO, boot_pages, CTLFLAG_RDTUN | CTLFLAG_NOFETCH, &boot_pages, 0, "number of pages allocated for bootstrapping the VM system"); static int pa_tryrelock_restart; SYSCTL_INT(_vm, OID_AUTO, tryrelock_restart, CTLFLAG_RD, &pa_tryrelock_restart, 0, "Number of tryrelock restarts"); static TAILQ_HEAD(, vm_page) blacklist_head; static int sysctl_vm_page_blacklist(SYSCTL_HANDLER_ARGS); SYSCTL_PROC(_vm, OID_AUTO, page_blacklist, CTLTYPE_STRING | CTLFLAG_RD | CTLFLAG_MPSAFE, NULL, 0, sysctl_vm_page_blacklist, "A", "Blacklist pages"); /* Is the page daemon waiting for free pages? */ static int vm_pageout_pages_needed; static uma_zone_t fakepg_zone; static void vm_page_alloc_check(vm_page_t m); static void vm_page_clear_dirty_mask(vm_page_t m, vm_page_bits_t pagebits); static void vm_page_enqueue(uint8_t queue, vm_page_t m); static void vm_page_free_wakeup(void); -static void vm_page_init_fakepg(void *dummy); +static void vm_page_init(void *dummy); static int vm_page_insert_after(vm_page_t m, vm_object_t object, vm_pindex_t pindex, vm_page_t mpred); static void vm_page_insert_radixdone(vm_page_t m, vm_object_t object, vm_page_t mpred); static int vm_page_reclaim_run(int req_class, u_long npages, vm_page_t m_run, vm_paddr_t high); -SYSINIT(vm_page, SI_SUB_VM, SI_ORDER_SECOND, vm_page_init_fakepg, NULL); +SYSINIT(vm_page, SI_SUB_VM, SI_ORDER_SECOND, vm_page_init, NULL); static void -vm_page_init_fakepg(void *dummy) +vm_page_init(void *dummy) { fakepg_zone = uma_zcreate("fakepg", sizeof(struct vm_page), NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, UMA_ZONE_NOFREE | UMA_ZONE_VM); + bogus_page = vm_page_alloc(NULL, 0, VM_ALLOC_NOOBJ | + VM_ALLOC_NORMAL | VM_ALLOC_WIRED); } /* Make sure that u_long is at least 64 bits when PAGE_SIZE is 32K. */ #if PAGE_SIZE == 32768 #ifdef CTASSERT CTASSERT(sizeof(u_long) >= 8); #endif #endif /* * Try to acquire a physical address lock while a pmap is locked. If we * fail to trylock we unlock and lock the pmap directly and cache the * locked pa in *locked. The caller should then restart their loop in case * the virtual to physical mapping has changed. */ int vm_page_pa_tryrelock(pmap_t pmap, vm_paddr_t pa, vm_paddr_t *locked) { vm_paddr_t lockpa; lockpa = *locked; *locked = pa; if (lockpa) { PA_LOCK_ASSERT(lockpa, MA_OWNED); if (PA_LOCKPTR(pa) == PA_LOCKPTR(lockpa)) return (0); PA_UNLOCK(lockpa); } if (PA_TRYLOCK(pa)) return (0); PMAP_UNLOCK(pmap); atomic_add_int(&pa_tryrelock_restart, 1); PA_LOCK(pa); PMAP_LOCK(pmap); return (EAGAIN); } /* * vm_set_page_size: * * Sets the page size, perhaps based upon the memory * size. Must be called before any use of page-size * dependent functions. */ void vm_set_page_size(void) { if (vm_cnt.v_page_size == 0) vm_cnt.v_page_size = PAGE_SIZE; if (((vm_cnt.v_page_size - 1) & vm_cnt.v_page_size) != 0) panic("vm_set_page_size: page size not a power of two"); } /* * vm_page_blacklist_next: * * Find the next entry in the provided string of blacklist * addresses. Entries are separated by space, comma, or newline. * If an invalid integer is encountered then the rest of the * string is skipped. Updates the list pointer to the next * character, or NULL if the string is exhausted or invalid. */ static vm_paddr_t vm_page_blacklist_next(char **list, char *end) { vm_paddr_t bad; char *cp, *pos; if (list == NULL || *list == NULL) return (0); if (**list =='\0') { *list = NULL; return (0); } /* * If there's no end pointer then the buffer is coming from * the kenv and we know it's null-terminated. */ if (end == NULL) end = *list + strlen(*list); /* Ensure that strtoq() won't walk off the end */ if (*end != '\0') { if (*end == '\n' || *end == ' ' || *end == ',') *end = '\0'; else { printf("Blacklist not terminated, skipping\n"); *list = NULL; return (0); } } for (pos = *list; *pos != '\0'; pos = cp) { bad = strtoq(pos, &cp, 0); if (*cp == '\0' || *cp == ' ' || *cp == ',' || *cp == '\n') { if (bad == 0) { if (++cp < end) continue; else break; } } else break; if (*cp == '\0' || ++cp >= end) *list = NULL; else *list = cp; return (trunc_page(bad)); } printf("Garbage in RAM blacklist, skipping\n"); *list = NULL; return (0); } /* * vm_page_blacklist_check: * * Iterate through the provided string of blacklist addresses, pulling * each entry out of the physical allocator free list and putting it * onto a list for reporting via the vm.page_blacklist sysctl. */ static void vm_page_blacklist_check(char *list, char *end) { vm_paddr_t pa; vm_page_t m; char *next; int ret; next = list; while (next != NULL) { if ((pa = vm_page_blacklist_next(&next, end)) == 0) continue; m = vm_phys_paddr_to_vm_page(pa); if (m == NULL) continue; mtx_lock(&vm_page_queue_free_mtx); ret = vm_phys_unfree_page(m); mtx_unlock(&vm_page_queue_free_mtx); if (ret == TRUE) { TAILQ_INSERT_TAIL(&blacklist_head, m, listq); if (bootverbose) printf("Skipping page with pa 0x%jx\n", (uintmax_t)pa); } } } /* * vm_page_blacklist_load: * * Search for a special module named "ram_blacklist". It'll be a * plain text file provided by the user via the loader directive * of the same name. */ static void vm_page_blacklist_load(char **list, char **end) { void *mod; u_char *ptr; u_int len; mod = NULL; ptr = NULL; mod = preload_search_by_type("ram_blacklist"); if (mod != NULL) { ptr = preload_fetch_addr(mod); len = preload_fetch_size(mod); } *list = ptr; if (ptr != NULL) *end = ptr + len; else *end = NULL; return; } static int sysctl_vm_page_blacklist(SYSCTL_HANDLER_ARGS) { vm_page_t m; struct sbuf sbuf; int error, first; first = 1; error = sysctl_wire_old_buffer(req, 0); if (error != 0) return (error); sbuf_new_for_sysctl(&sbuf, NULL, 128, req); TAILQ_FOREACH(m, &blacklist_head, listq) { sbuf_printf(&sbuf, "%s%#jx", first ? "" : ",", (uintmax_t)m->phys_addr); first = 0; } error = sbuf_finish(&sbuf); sbuf_delete(&sbuf); return (error); } static void vm_page_domain_init(struct vm_domain *vmd) { struct vm_pagequeue *pq; int i; *__DECONST(char **, &vmd->vmd_pagequeues[PQ_INACTIVE].pq_name) = "vm inactive pagequeue"; *__DECONST(u_int **, &vmd->vmd_pagequeues[PQ_INACTIVE].pq_vcnt) = &vm_cnt.v_inactive_count; *__DECONST(char **, &vmd->vmd_pagequeues[PQ_ACTIVE].pq_name) = "vm active pagequeue"; *__DECONST(u_int **, &vmd->vmd_pagequeues[PQ_ACTIVE].pq_vcnt) = &vm_cnt.v_active_count; *__DECONST(char **, &vmd->vmd_pagequeues[PQ_LAUNDRY].pq_name) = "vm laundry pagequeue"; *__DECONST(int **, &vmd->vmd_pagequeues[PQ_LAUNDRY].pq_vcnt) = &vm_cnt.v_laundry_count; *__DECONST(char **, &vmd->vmd_pagequeues[PQ_UNSWAPPABLE].pq_name) = "vm unswappable pagequeue"; /* Unswappable dirty pages are counted as being in the laundry. */ *__DECONST(int **, &vmd->vmd_pagequeues[PQ_UNSWAPPABLE].pq_vcnt) = &vm_cnt.v_laundry_count; vmd->vmd_page_count = 0; vmd->vmd_free_count = 0; vmd->vmd_segs = 0; vmd->vmd_oom = FALSE; for (i = 0; i < PQ_COUNT; i++) { pq = &vmd->vmd_pagequeues[i]; TAILQ_INIT(&pq->pq_pl); mtx_init(&pq->pq_mutex, pq->pq_name, "vm pagequeue", MTX_DEF | MTX_DUPOK); } } /* * vm_page_startup: * * Initializes the resident memory module. * * Allocates memory for the page cells, and * for the object/offset-to-page hash table headers. * Each page cell is initialized and placed on the free list. */ vm_offset_t vm_page_startup(vm_offset_t vaddr) { vm_offset_t mapped; vm_paddr_t page_range; vm_paddr_t new_end; int i; vm_paddr_t pa; vm_paddr_t last_pa; char *list, *listend; vm_paddr_t end; vm_paddr_t biggestsize; vm_paddr_t low_water, high_water; int biggestone; int pages_per_zone; biggestsize = 0; biggestone = 0; vaddr = round_page(vaddr); for (i = 0; phys_avail[i + 1]; i += 2) { phys_avail[i] = round_page(phys_avail[i]); phys_avail[i + 1] = trunc_page(phys_avail[i + 1]); } low_water = phys_avail[0]; high_water = phys_avail[1]; for (i = 0; i < vm_phys_nsegs; i++) { if (vm_phys_segs[i].start < low_water) low_water = vm_phys_segs[i].start; if (vm_phys_segs[i].end > high_water) high_water = vm_phys_segs[i].end; } for (i = 0; phys_avail[i + 1]; i += 2) { vm_paddr_t size = phys_avail[i + 1] - phys_avail[i]; if (size > biggestsize) { biggestone = i; biggestsize = size; } if (phys_avail[i] < low_water) low_water = phys_avail[i]; if (phys_avail[i + 1] > high_water) high_water = phys_avail[i + 1]; } end = phys_avail[biggestone+1]; /* * Initialize the page and queue locks. */ mtx_init(&vm_page_queue_free_mtx, "vm page free queue", NULL, MTX_DEF); for (i = 0; i < PA_LOCK_COUNT; i++) mtx_init(&pa_lock[i], "vm page", NULL, MTX_DEF); for (i = 0; i < vm_ndomains; i++) vm_page_domain_init(&vm_dom[i]); /* * Almost all of the pages needed for boot strapping UMA are used * for zone structures, so if the number of CPUs results in those * structures taking more than one page each, we set aside more pages * in proportion to the zone structure size. */ pages_per_zone = howmany(sizeof(struct uma_zone) + sizeof(struct uma_cache) * (mp_maxid + 1), UMA_SLAB_SIZE); if (pages_per_zone > 1) { /* Reserve more pages so that we don't run out. */ boot_pages = UMA_BOOT_PAGES_ZONES * pages_per_zone; } /* * Allocate memory for use when boot strapping the kernel memory * allocator. * * CTFLAG_RDTUN doesn't work during the early boot process, so we must * manually fetch the value. */ TUNABLE_INT_FETCH("vm.boot_pages", &boot_pages); new_end = end - (boot_pages * UMA_SLAB_SIZE); new_end = trunc_page(new_end); mapped = pmap_map(&vaddr, new_end, end, VM_PROT_READ | VM_PROT_WRITE); bzero((void *)mapped, end - new_end); uma_startup((void *)mapped, boot_pages); #if defined(__aarch64__) || defined(__amd64__) || defined(__arm__) || \ defined(__i386__) || defined(__mips__) /* * Allocate a bitmap to indicate that a random physical page * needs to be included in a minidump. * * The amd64 port needs this to indicate which direct map pages * need to be dumped, via calls to dump_add_page()/dump_drop_page(). * * However, i386 still needs this workspace internally within the * minidump code. In theory, they are not needed on i386, but are * included should the sf_buf code decide to use them. */ last_pa = 0; for (i = 0; dump_avail[i + 1] != 0; i += 2) if (dump_avail[i + 1] > last_pa) last_pa = dump_avail[i + 1]; page_range = last_pa / PAGE_SIZE; vm_page_dump_size = round_page(roundup2(page_range, NBBY) / NBBY); new_end -= vm_page_dump_size; vm_page_dump = (void *)(uintptr_t)pmap_map(&vaddr, new_end, new_end + vm_page_dump_size, VM_PROT_READ | VM_PROT_WRITE); bzero((void *)vm_page_dump, vm_page_dump_size); #endif #ifdef __amd64__ /* * Request that the physical pages underlying the message buffer be * included in a crash dump. Since the message buffer is accessed * through the direct map, they are not automatically included. */ pa = DMAP_TO_PHYS((vm_offset_t)msgbufp->msg_ptr); last_pa = pa + round_page(msgbufsize); while (pa < last_pa) { dump_add_page(pa); pa += PAGE_SIZE; } #endif /* * Compute the number of pages of memory that will be available for * use (taking into account the overhead of a page structure per * page). */ first_page = low_water / PAGE_SIZE; #ifdef VM_PHYSSEG_SPARSE page_range = 0; for (i = 0; i < vm_phys_nsegs; i++) { page_range += atop(vm_phys_segs[i].end - vm_phys_segs[i].start); } for (i = 0; phys_avail[i + 1] != 0; i += 2) page_range += atop(phys_avail[i + 1] - phys_avail[i]); #elif defined(VM_PHYSSEG_DENSE) page_range = high_water / PAGE_SIZE - first_page; #else #error "Either VM_PHYSSEG_DENSE or VM_PHYSSEG_SPARSE must be defined." #endif end = new_end; /* * Reserve an unmapped guard page to trap access to vm_page_array[-1]. */ vaddr += PAGE_SIZE; /* * Initialize the mem entry structures now, and put them in the free * queue. */ new_end = trunc_page(end - page_range * sizeof(struct vm_page)); mapped = pmap_map(&vaddr, new_end, end, VM_PROT_READ | VM_PROT_WRITE); vm_page_array = (vm_page_t) mapped; #if VM_NRESERVLEVEL > 0 /* * Allocate memory for the reservation management system's data * structures. */ new_end = vm_reserv_startup(&vaddr, new_end, high_water); #endif #if defined(__aarch64__) || defined(__amd64__) || defined(__mips__) /* * pmap_map on arm64, amd64, and mips can come out of the direct-map, * not kvm like i386, so the pages must be tracked for a crashdump to * include this data. This includes the vm_page_array and the early * UMA bootstrap pages. */ for (pa = new_end; pa < phys_avail[biggestone + 1]; pa += PAGE_SIZE) dump_add_page(pa); #endif phys_avail[biggestone + 1] = new_end; /* * Add physical memory segments corresponding to the available * physical pages. */ for (i = 0; phys_avail[i + 1] != 0; i += 2) vm_phys_add_seg(phys_avail[i], phys_avail[i + 1]); /* * Clear all of the page structures */ bzero((caddr_t) vm_page_array, page_range * sizeof(struct vm_page)); for (i = 0; i < page_range; i++) vm_page_array[i].order = VM_NFREEORDER; vm_page_array_size = page_range; /* * Initialize the physical memory allocator. */ vm_phys_init(); /* * Add every available physical page that is not blacklisted to * the free lists. */ vm_cnt.v_page_count = 0; vm_cnt.v_free_count = 0; for (i = 0; phys_avail[i + 1] != 0; i += 2) { pa = phys_avail[i]; last_pa = phys_avail[i + 1]; while (pa < last_pa) { vm_phys_add_page(pa); pa += PAGE_SIZE; } } TAILQ_INIT(&blacklist_head); vm_page_blacklist_load(&list, &listend); vm_page_blacklist_check(list, listend); list = kern_getenv("vm.blacklist"); vm_page_blacklist_check(list, NULL); freeenv(list); #if VM_NRESERVLEVEL > 0 /* * Initialize the reservation management system. */ vm_reserv_init(); #endif return (vaddr); } void vm_page_reference(vm_page_t m) { vm_page_aflag_set(m, PGA_REFERENCED); } /* * vm_page_busy_downgrade: * * Downgrade an exclusive busy page into a single shared busy page. */ void vm_page_busy_downgrade(vm_page_t m) { u_int x; bool locked; vm_page_assert_xbusied(m); locked = mtx_owned(vm_page_lockptr(m)); for (;;) { x = m->busy_lock; x &= VPB_BIT_WAITERS; if (x != 0 && !locked) vm_page_lock(m); if (atomic_cmpset_rel_int(&m->busy_lock, VPB_SINGLE_EXCLUSIVER | x, VPB_SHARERS_WORD(1))) break; if (x != 0 && !locked) vm_page_unlock(m); } if (x != 0) { wakeup(m); if (!locked) vm_page_unlock(m); } } /* * vm_page_sbusied: * * Return a positive value if the page is shared busied, 0 otherwise. */ int vm_page_sbusied(vm_page_t m) { u_int x; x = m->busy_lock; return ((x & VPB_BIT_SHARED) != 0 && x != VPB_UNBUSIED); } /* * vm_page_sunbusy: * * Shared unbusy a page. */ void vm_page_sunbusy(vm_page_t m) { u_int x; vm_page_assert_sbusied(m); for (;;) { x = m->busy_lock; if (VPB_SHARERS(x) > 1) { if (atomic_cmpset_int(&m->busy_lock, x, x - VPB_ONE_SHARER)) break; continue; } if ((x & VPB_BIT_WAITERS) == 0) { KASSERT(x == VPB_SHARERS_WORD(1), ("vm_page_sunbusy: invalid lock state")); if (atomic_cmpset_int(&m->busy_lock, VPB_SHARERS_WORD(1), VPB_UNBUSIED)) break; continue; } KASSERT(x == (VPB_SHARERS_WORD(1) | VPB_BIT_WAITERS), ("vm_page_sunbusy: invalid lock state for waiters")); vm_page_lock(m); if (!atomic_cmpset_int(&m->busy_lock, x, VPB_UNBUSIED)) { vm_page_unlock(m); continue; } wakeup(m); vm_page_unlock(m); break; } } /* * vm_page_busy_sleep: * * Sleep and release the page lock, using the page pointer as wchan. * This is used to implement the hard-path of busying mechanism. * * The given page must be locked. * * If nonshared is true, sleep only if the page is xbusy. */ void vm_page_busy_sleep(vm_page_t m, const char *wmesg, bool nonshared) { u_int x; vm_page_assert_locked(m); x = m->busy_lock; if (x == VPB_UNBUSIED || (nonshared && (x & VPB_BIT_SHARED) != 0) || ((x & VPB_BIT_WAITERS) == 0 && !atomic_cmpset_int(&m->busy_lock, x, x | VPB_BIT_WAITERS))) { vm_page_unlock(m); return; } msleep(m, vm_page_lockptr(m), PVM | PDROP, wmesg, 0); } /* * vm_page_trysbusy: * * Try to shared busy a page. * If the operation succeeds 1 is returned otherwise 0. * The operation never sleeps. */ int vm_page_trysbusy(vm_page_t m) { u_int x; for (;;) { x = m->busy_lock; if ((x & VPB_BIT_SHARED) == 0) return (0); if (atomic_cmpset_acq_int(&m->busy_lock, x, x + VPB_ONE_SHARER)) return (1); } } static void vm_page_xunbusy_locked(vm_page_t m) { vm_page_assert_xbusied(m); vm_page_assert_locked(m); atomic_store_rel_int(&m->busy_lock, VPB_UNBUSIED); /* There is a waiter, do wakeup() instead of vm_page_flash(). */ wakeup(m); } void vm_page_xunbusy_maybelocked(vm_page_t m) { bool lockacq; vm_page_assert_xbusied(m); /* * Fast path for unbusy. If it succeeds, we know that there * are no waiters, so we do not need a wakeup. */ if (atomic_cmpset_rel_int(&m->busy_lock, VPB_SINGLE_EXCLUSIVER, VPB_UNBUSIED)) return; lockacq = !mtx_owned(vm_page_lockptr(m)); if (lockacq) vm_page_lock(m); vm_page_xunbusy_locked(m); if (lockacq) vm_page_unlock(m); } /* * vm_page_xunbusy_hard: * * Called after the first try the exclusive unbusy of a page failed. * It is assumed that the waiters bit is on. */ void vm_page_xunbusy_hard(vm_page_t m) { vm_page_assert_xbusied(m); vm_page_lock(m); vm_page_xunbusy_locked(m); vm_page_unlock(m); } /* * vm_page_flash: * * Wakeup anyone waiting for the page. * The ownership bits do not change. * * The given page must be locked. */ void vm_page_flash(vm_page_t m) { u_int x; vm_page_lock_assert(m, MA_OWNED); for (;;) { x = m->busy_lock; if ((x & VPB_BIT_WAITERS) == 0) return; if (atomic_cmpset_int(&m->busy_lock, x, x & (~VPB_BIT_WAITERS))) break; } wakeup(m); } /* * Keep page from being freed by the page daemon * much of the same effect as wiring, except much lower * overhead and should be used only for *very* temporary * holding ("wiring"). */ void vm_page_hold(vm_page_t mem) { vm_page_lock_assert(mem, MA_OWNED); mem->hold_count++; } void vm_page_unhold(vm_page_t mem) { vm_page_lock_assert(mem, MA_OWNED); KASSERT(mem->hold_count >= 1, ("vm_page_unhold: hold count < 0!!!")); --mem->hold_count; if (mem->hold_count == 0 && (mem->flags & PG_UNHOLDFREE) != 0) vm_page_free_toq(mem); } /* * vm_page_unhold_pages: * * Unhold each of the pages that is referenced by the given array. */ void vm_page_unhold_pages(vm_page_t *ma, int count) { struct mtx *mtx, *new_mtx; mtx = NULL; for (; count != 0; count--) { /* * Avoid releasing and reacquiring the same page lock. */ new_mtx = vm_page_lockptr(*ma); if (mtx != new_mtx) { if (mtx != NULL) mtx_unlock(mtx); mtx = new_mtx; mtx_lock(mtx); } vm_page_unhold(*ma); ma++; } if (mtx != NULL) mtx_unlock(mtx); } vm_page_t PHYS_TO_VM_PAGE(vm_paddr_t pa) { vm_page_t m; #ifdef VM_PHYSSEG_SPARSE m = vm_phys_paddr_to_vm_page(pa); if (m == NULL) m = vm_phys_fictitious_to_vm_page(pa); return (m); #elif defined(VM_PHYSSEG_DENSE) long pi; pi = atop(pa); if (pi >= first_page && (pi - first_page) < vm_page_array_size) { m = &vm_page_array[pi - first_page]; return (m); } return (vm_phys_fictitious_to_vm_page(pa)); #else #error "Either VM_PHYSSEG_DENSE or VM_PHYSSEG_SPARSE must be defined." #endif } /* * vm_page_getfake: * * Create a fictitious page with the specified physical address and * memory attribute. The memory attribute is the only the machine- * dependent aspect of a fictitious page that must be initialized. */ vm_page_t vm_page_getfake(vm_paddr_t paddr, vm_memattr_t memattr) { vm_page_t m; m = uma_zalloc(fakepg_zone, M_WAITOK | M_ZERO); vm_page_initfake(m, paddr, memattr); return (m); } void vm_page_initfake(vm_page_t m, vm_paddr_t paddr, vm_memattr_t memattr) { if ((m->flags & PG_FICTITIOUS) != 0) { /* * The page's memattr might have changed since the * previous initialization. Update the pmap to the * new memattr. */ goto memattr; } m->phys_addr = paddr; m->queue = PQ_NONE; /* Fictitious pages don't use "segind". */ m->flags = PG_FICTITIOUS; /* Fictitious pages don't use "order" or "pool". */ m->oflags = VPO_UNMANAGED; m->busy_lock = VPB_SINGLE_EXCLUSIVER; m->wire_count = 1; pmap_page_init(m); memattr: pmap_page_set_memattr(m, memattr); } /* * vm_page_putfake: * * Release a fictitious page. */ void vm_page_putfake(vm_page_t m) { KASSERT((m->oflags & VPO_UNMANAGED) != 0, ("managed %p", m)); KASSERT((m->flags & PG_FICTITIOUS) != 0, ("vm_page_putfake: bad page %p", m)); uma_zfree(fakepg_zone, m); } /* * vm_page_updatefake: * * Update the given fictitious page to the specified physical address and * memory attribute. */ void vm_page_updatefake(vm_page_t m, vm_paddr_t paddr, vm_memattr_t memattr) { KASSERT((m->flags & PG_FICTITIOUS) != 0, ("vm_page_updatefake: bad page %p", m)); m->phys_addr = paddr; pmap_page_set_memattr(m, memattr); } /* * vm_page_free: * * Free a page. */ void vm_page_free(vm_page_t m) { m->flags &= ~PG_ZERO; vm_page_free_toq(m); } /* * vm_page_free_zero: * * Free a page to the zerod-pages queue */ void vm_page_free_zero(vm_page_t m) { m->flags |= PG_ZERO; vm_page_free_toq(m); } /* * Unbusy and handle the page queueing for a page from a getpages request that * was optionally read ahead or behind. */ void vm_page_readahead_finish(vm_page_t m) { /* We shouldn't put invalid pages on queues. */ KASSERT(m->valid != 0, ("%s: %p is invalid", __func__, m)); /* * Since the page is not the actually needed one, whether it should * be activated or deactivated is not obvious. Empirical results * have shown that deactivating the page is usually the best choice, * unless the page is wanted by another thread. */ vm_page_lock(m); if ((m->busy_lock & VPB_BIT_WAITERS) != 0) vm_page_activate(m); else vm_page_deactivate(m); vm_page_unlock(m); vm_page_xunbusy(m); } /* * vm_page_sleep_if_busy: * * Sleep and release the page queues lock if the page is busied. * Returns TRUE if the thread slept. * * The given page must be unlocked and object containing it must * be locked. */ int vm_page_sleep_if_busy(vm_page_t m, const char *msg) { vm_object_t obj; vm_page_lock_assert(m, MA_NOTOWNED); VM_OBJECT_ASSERT_WLOCKED(m->object); if (vm_page_busied(m)) { /* * The page-specific object must be cached because page * identity can change during the sleep, causing the * re-lock of a different object. * It is assumed that a reference to the object is already * held by the callers. */ obj = m->object; vm_page_lock(m); VM_OBJECT_WUNLOCK(obj); vm_page_busy_sleep(m, msg, false); VM_OBJECT_WLOCK(obj); return (TRUE); } return (FALSE); } /* * vm_page_dirty_KBI: [ internal use only ] * * Set all bits in the page's dirty field. * * The object containing the specified page must be locked if the * call is made from the machine-independent layer. * * See vm_page_clear_dirty_mask(). * * This function should only be called by vm_page_dirty(). */ void vm_page_dirty_KBI(vm_page_t m) { /* Refer to this operation by its public name. */ KASSERT(m->valid == VM_PAGE_BITS_ALL, ("vm_page_dirty: page is invalid!")); m->dirty = VM_PAGE_BITS_ALL; } /* * vm_page_insert: [ internal use only ] * * Inserts the given mem entry into the object and object list. * * The object must be locked. */ int vm_page_insert(vm_page_t m, vm_object_t object, vm_pindex_t pindex) { vm_page_t mpred; VM_OBJECT_ASSERT_WLOCKED(object); mpred = vm_radix_lookup_le(&object->rtree, pindex); return (vm_page_insert_after(m, object, pindex, mpred)); } /* * vm_page_insert_after: * * Inserts the page "m" into the specified object at offset "pindex". * * The page "mpred" must immediately precede the offset "pindex" within * the specified object. * * The object must be locked. */ static int vm_page_insert_after(vm_page_t m, vm_object_t object, vm_pindex_t pindex, vm_page_t mpred) { vm_page_t msucc; VM_OBJECT_ASSERT_WLOCKED(object); KASSERT(m->object == NULL, ("vm_page_insert_after: page already inserted")); if (mpred != NULL) { KASSERT(mpred->object == object, ("vm_page_insert_after: object doesn't contain mpred")); KASSERT(mpred->pindex < pindex, ("vm_page_insert_after: mpred doesn't precede pindex")); msucc = TAILQ_NEXT(mpred, listq); } else msucc = TAILQ_FIRST(&object->memq); if (msucc != NULL) KASSERT(msucc->pindex > pindex, ("vm_page_insert_after: msucc doesn't succeed pindex")); /* * Record the object/offset pair in this page */ m->object = object; m->pindex = pindex; /* * Now link into the object's ordered list of backed pages. */ if (vm_radix_insert(&object->rtree, m)) { m->object = NULL; m->pindex = 0; return (1); } vm_page_insert_radixdone(m, object, mpred); return (0); } /* * vm_page_insert_radixdone: * * Complete page "m" insertion into the specified object after the * radix trie hooking. * * The page "mpred" must precede the offset "m->pindex" within the * specified object. * * The object must be locked. */ static void vm_page_insert_radixdone(vm_page_t m, vm_object_t object, vm_page_t mpred) { VM_OBJECT_ASSERT_WLOCKED(object); KASSERT(object != NULL && m->object == object, ("vm_page_insert_radixdone: page %p has inconsistent object", m)); if (mpred != NULL) { KASSERT(mpred->object == object, ("vm_page_insert_after: object doesn't contain mpred")); KASSERT(mpred->pindex < m->pindex, ("vm_page_insert_after: mpred doesn't precede pindex")); } if (mpred != NULL) TAILQ_INSERT_AFTER(&object->memq, mpred, m, listq); else TAILQ_INSERT_HEAD(&object->memq, m, listq); /* * Show that the object has one more resident page. */ object->resident_page_count++; /* * Hold the vnode until the last page is released. */ if (object->resident_page_count == 1 && object->type == OBJT_VNODE) vhold(object->handle); /* * Since we are inserting a new and possibly dirty page, * update the object's OBJ_MIGHTBEDIRTY flag. */ if (pmap_page_is_write_mapped(m)) vm_object_set_writeable_dirty(object); } /* * vm_page_remove: * * Removes the specified page from its containing object, but does not * invalidate any backing storage. * * The object must be locked. The page must be locked if it is managed. */ void vm_page_remove(vm_page_t m) { vm_object_t object; vm_page_t mrem; if ((m->oflags & VPO_UNMANAGED) == 0) vm_page_assert_locked(m); if ((object = m->object) == NULL) return; VM_OBJECT_ASSERT_WLOCKED(object); if (vm_page_xbusied(m)) vm_page_xunbusy_maybelocked(m); mrem = vm_radix_remove(&object->rtree, m->pindex); KASSERT(mrem == m, ("removed page %p, expected page %p", mrem, m)); /* * Now remove from the object's list of backed pages. */ TAILQ_REMOVE(&object->memq, m, listq); /* * And show that the object has one fewer resident page. */ object->resident_page_count--; /* * The vnode may now be recycled. */ if (object->resident_page_count == 0 && object->type == OBJT_VNODE) vdrop(object->handle); m->object = NULL; } /* * vm_page_lookup: * * Returns the page associated with the object/offset * pair specified; if none is found, NULL is returned. * * The object must be locked. */ vm_page_t vm_page_lookup(vm_object_t object, vm_pindex_t pindex) { VM_OBJECT_ASSERT_LOCKED(object); return (vm_radix_lookup(&object->rtree, pindex)); } /* * vm_page_find_least: * * Returns the page associated with the object with least pindex * greater than or equal to the parameter pindex, or NULL. * * The object must be locked. */ vm_page_t vm_page_find_least(vm_object_t object, vm_pindex_t pindex) { vm_page_t m; VM_OBJECT_ASSERT_LOCKED(object); if ((m = TAILQ_FIRST(&object->memq)) != NULL && m->pindex < pindex) m = vm_radix_lookup_ge(&object->rtree, pindex); return (m); } /* * Returns the given page's successor (by pindex) within the object if it is * resident; if none is found, NULL is returned. * * The object must be locked. */ vm_page_t vm_page_next(vm_page_t m) { vm_page_t next; VM_OBJECT_ASSERT_LOCKED(m->object); if ((next = TAILQ_NEXT(m, listq)) != NULL) { MPASS(next->object == m->object); if (next->pindex != m->pindex + 1) next = NULL; } return (next); } /* * Returns the given page's predecessor (by pindex) within the object if it is * resident; if none is found, NULL is returned. * * The object must be locked. */ vm_page_t vm_page_prev(vm_page_t m) { vm_page_t prev; VM_OBJECT_ASSERT_LOCKED(m->object); if ((prev = TAILQ_PREV(m, pglist, listq)) != NULL) { MPASS(prev->object == m->object); if (prev->pindex != m->pindex - 1) prev = NULL; } return (prev); } /* * Uses the page mnew as a replacement for an existing page at index * pindex which must be already present in the object. * * The existing page must not be on a paging queue. */ vm_page_t vm_page_replace(vm_page_t mnew, vm_object_t object, vm_pindex_t pindex) { vm_page_t mold; VM_OBJECT_ASSERT_WLOCKED(object); KASSERT(mnew->object == NULL, ("vm_page_replace: page already in object")); /* * This function mostly follows vm_page_insert() and * vm_page_remove() without the radix, object count and vnode * dance. Double check such functions for more comments. */ mnew->object = object; mnew->pindex = pindex; mold = vm_radix_replace(&object->rtree, mnew); KASSERT(mold->queue == PQ_NONE, ("vm_page_replace: mold is on a paging queue")); /* Keep the resident page list in sorted order. */ TAILQ_INSERT_AFTER(&object->memq, mold, mnew, listq); TAILQ_REMOVE(&object->memq, mold, listq); mold->object = NULL; vm_page_xunbusy_maybelocked(mold); /* * The object's resident_page_count does not change because we have * swapped one page for another, but OBJ_MIGHTBEDIRTY. */ if (pmap_page_is_write_mapped(mnew)) vm_object_set_writeable_dirty(object); return (mold); } /* * vm_page_rename: * * Move the given memory entry from its * current object to the specified target object/offset. * * Note: swap associated with the page must be invalidated by the move. We * have to do this for several reasons: (1) we aren't freeing the * page, (2) we are dirtying the page, (3) the VM system is probably * moving the page from object A to B, and will then later move * the backing store from A to B and we can't have a conflict. * * Note: we *always* dirty the page. It is necessary both for the * fact that we moved it, and because we may be invalidating * swap. * * The objects must be locked. */ int vm_page_rename(vm_page_t m, vm_object_t new_object, vm_pindex_t new_pindex) { vm_page_t mpred; vm_pindex_t opidx; VM_OBJECT_ASSERT_WLOCKED(new_object); mpred = vm_radix_lookup_le(&new_object->rtree, new_pindex); KASSERT(mpred == NULL || mpred->pindex != new_pindex, ("vm_page_rename: pindex already renamed")); /* * Create a custom version of vm_page_insert() which does not depend * by m_prev and can cheat on the implementation aspects of the * function. */ opidx = m->pindex; m->pindex = new_pindex; if (vm_radix_insert(&new_object->rtree, m)) { m->pindex = opidx; return (1); } /* * The operation cannot fail anymore. The removal must happen before * the listq iterator is tainted. */ m->pindex = opidx; vm_page_lock(m); vm_page_remove(m); /* Return back to the new pindex to complete vm_page_insert(). */ m->pindex = new_pindex; m->object = new_object; vm_page_unlock(m); vm_page_insert_radixdone(m, new_object, mpred); vm_page_dirty(m); return (0); } /* * vm_page_alloc: * * Allocate and return a page that is associated with the specified * object and offset pair. By default, this page is exclusive busied. * * The caller must always specify an allocation class. * * allocation classes: * VM_ALLOC_NORMAL normal process request * VM_ALLOC_SYSTEM system *really* needs a page * VM_ALLOC_INTERRUPT interrupt time request * * optional allocation flags: * VM_ALLOC_COUNT(number) the number of additional pages that the caller * intends to allocate * VM_ALLOC_NOBUSY do not exclusive busy the page * VM_ALLOC_NODUMP do not include the page in a kernel core dump * VM_ALLOC_NOOBJ page is not associated with an object and * should not be exclusive busy * VM_ALLOC_SBUSY shared busy the allocated page * VM_ALLOC_WIRED wire the allocated page * VM_ALLOC_ZERO prefer a zeroed page * * This routine may not sleep. */ vm_page_t vm_page_alloc(vm_object_t object, vm_pindex_t pindex, int req) { vm_page_t m, mpred; int flags, req_class; mpred = NULL; /* XXX: pacify gcc */ KASSERT((object != NULL) == ((req & VM_ALLOC_NOOBJ) == 0) && (object != NULL || (req & VM_ALLOC_SBUSY) == 0) && ((req & (VM_ALLOC_NOBUSY | VM_ALLOC_SBUSY)) != (VM_ALLOC_NOBUSY | VM_ALLOC_SBUSY)), ("vm_page_alloc: inconsistent object(%p)/req(%x)", object, req)); if (object != NULL) VM_OBJECT_ASSERT_WLOCKED(object); req_class = req & VM_ALLOC_CLASS_MASK; /* * The page daemon is allowed to dig deeper into the free page list. */ if (curproc == pageproc && req_class != VM_ALLOC_INTERRUPT) req_class = VM_ALLOC_SYSTEM; if (object != NULL) { mpred = vm_radix_lookup_le(&object->rtree, pindex); KASSERT(mpred == NULL || mpred->pindex != pindex, ("vm_page_alloc: pindex already allocated")); } /* * Allocate a page if the number of free pages exceeds the minimum * for the request class. */ mtx_lock(&vm_page_queue_free_mtx); if (vm_cnt.v_free_count > vm_cnt.v_free_reserved || (req_class == VM_ALLOC_SYSTEM && vm_cnt.v_free_count > vm_cnt.v_interrupt_free_min) || (req_class == VM_ALLOC_INTERRUPT && vm_cnt.v_free_count > 0)) { /* * Can we allocate the page from a reservation? */ #if VM_NRESERVLEVEL > 0 if (object == NULL || (object->flags & (OBJ_COLORED | OBJ_FICTITIOUS)) != OBJ_COLORED || (m = vm_reserv_alloc_page(object, pindex, mpred)) == NULL) #endif { /* * If not, allocate it from the free page queues. */ m = vm_phys_alloc_pages(object != NULL ? VM_FREEPOOL_DEFAULT : VM_FREEPOOL_DIRECT, 0); #if VM_NRESERVLEVEL > 0 if (m == NULL && vm_reserv_reclaim_inactive()) { m = vm_phys_alloc_pages(object != NULL ? VM_FREEPOOL_DEFAULT : VM_FREEPOOL_DIRECT, 0); } #endif } } else { /* * Not allocatable, give up. */ mtx_unlock(&vm_page_queue_free_mtx); atomic_add_int(&vm_pageout_deficit, max((u_int)req >> VM_ALLOC_COUNT_SHIFT, 1)); pagedaemon_wakeup(); return (NULL); } /* * At this point we had better have found a good page. */ KASSERT(m != NULL, ("vm_page_alloc: missing page")); vm_phys_freecnt_adj(m, -1); mtx_unlock(&vm_page_queue_free_mtx); vm_page_alloc_check(m); /* * Initialize the page. Only the PG_ZERO flag is inherited. */ flags = 0; if ((req & VM_ALLOC_ZERO) != 0) flags = PG_ZERO; flags &= m->flags; if ((req & VM_ALLOC_NODUMP) != 0) flags |= PG_NODUMP; m->flags = flags; m->aflags = 0; m->oflags = object == NULL || (object->flags & OBJ_UNMANAGED) != 0 ? VPO_UNMANAGED : 0; m->busy_lock = VPB_UNBUSIED; if ((req & (VM_ALLOC_NOBUSY | VM_ALLOC_NOOBJ | VM_ALLOC_SBUSY)) == 0) m->busy_lock = VPB_SINGLE_EXCLUSIVER; if ((req & VM_ALLOC_SBUSY) != 0) m->busy_lock = VPB_SHARERS_WORD(1); if (req & VM_ALLOC_WIRED) { /* * The page lock is not required for wiring a page until that * page is inserted into the object. */ atomic_add_int(&vm_cnt.v_wire_count, 1); m->wire_count = 1; } m->act_count = 0; if (object != NULL) { if (vm_page_insert_after(m, object, pindex, mpred)) { pagedaemon_wakeup(); if (req & VM_ALLOC_WIRED) { atomic_subtract_int(&vm_cnt.v_wire_count, 1); m->wire_count = 0; } KASSERT(m->object == NULL, ("page %p has object", m)); m->oflags = VPO_UNMANAGED; m->busy_lock = VPB_UNBUSIED; /* Don't change PG_ZERO. */ vm_page_free_toq(m); return (NULL); } /* Ignore device objects; the pager sets "memattr" for them. */ if (object->memattr != VM_MEMATTR_DEFAULT && (object->flags & OBJ_FICTITIOUS) == 0) pmap_page_set_memattr(m, object->memattr); } else m->pindex = pindex; /* * Don't wakeup too often - wakeup the pageout daemon when * we would be nearly out of memory. */ if (vm_paging_needed()) pagedaemon_wakeup(); return (m); } /* * vm_page_alloc_contig: * * Allocate a contiguous set of physical pages of the given size "npages" * from the free lists. All of the physical pages must be at or above * the given physical address "low" and below the given physical address * "high". The given value "alignment" determines the alignment of the * first physical page in the set. If the given value "boundary" is * non-zero, then the set of physical pages cannot cross any physical * address boundary that is a multiple of that value. Both "alignment" * and "boundary" must be a power of two. * * If the specified memory attribute, "memattr", is VM_MEMATTR_DEFAULT, * then the memory attribute setting for the physical pages is configured * to the object's memory attribute setting. Otherwise, the memory * attribute setting for the physical pages is configured to "memattr", * overriding the object's memory attribute setting. However, if the * object's memory attribute setting is not VM_MEMATTR_DEFAULT, then the * memory attribute setting for the physical pages cannot be configured * to VM_MEMATTR_DEFAULT. * * The specified object may not contain fictitious pages. * * The caller must always specify an allocation class. * * allocation classes: * VM_ALLOC_NORMAL normal process request * VM_ALLOC_SYSTEM system *really* needs a page * VM_ALLOC_INTERRUPT interrupt time request * * optional allocation flags: * VM_ALLOC_NOBUSY do not exclusive busy the page * VM_ALLOC_NODUMP do not include the page in a kernel core dump * VM_ALLOC_NOOBJ page is not associated with an object and * should not be exclusive busy * VM_ALLOC_SBUSY shared busy the allocated page * VM_ALLOC_WIRED wire the allocated page * VM_ALLOC_ZERO prefer a zeroed page * * This routine may not sleep. */ vm_page_t vm_page_alloc_contig(vm_object_t object, vm_pindex_t pindex, int req, u_long npages, vm_paddr_t low, vm_paddr_t high, u_long alignment, vm_paddr_t boundary, vm_memattr_t memattr) { vm_page_t m, m_ret, mpred; u_int busy_lock, flags, oflags; int req_class; mpred = NULL; /* XXX: pacify gcc */ KASSERT((object != NULL) == ((req & VM_ALLOC_NOOBJ) == 0) && (object != NULL || (req & VM_ALLOC_SBUSY) == 0) && ((req & (VM_ALLOC_NOBUSY | VM_ALLOC_SBUSY)) != (VM_ALLOC_NOBUSY | VM_ALLOC_SBUSY)), ("vm_page_alloc_contig: inconsistent object(%p)/req(%x)", object, req)); if (object != NULL) { VM_OBJECT_ASSERT_WLOCKED(object); KASSERT((object->flags & OBJ_FICTITIOUS) == 0, ("vm_page_alloc_contig: object %p has fictitious pages", object)); } KASSERT(npages > 0, ("vm_page_alloc_contig: npages is zero")); req_class = req & VM_ALLOC_CLASS_MASK; /* * The page daemon is allowed to dig deeper into the free page list. */ if (curproc == pageproc && req_class != VM_ALLOC_INTERRUPT) req_class = VM_ALLOC_SYSTEM; if (object != NULL) { mpred = vm_radix_lookup_le(&object->rtree, pindex); KASSERT(mpred == NULL || mpred->pindex != pindex, ("vm_page_alloc_contig: pindex already allocated")); } /* * Can we allocate the pages without the number of free pages falling * below the lower bound for the allocation class? */ mtx_lock(&vm_page_queue_free_mtx); if (vm_cnt.v_free_count >= npages + vm_cnt.v_free_reserved || (req_class == VM_ALLOC_SYSTEM && vm_cnt.v_free_count >= npages + vm_cnt.v_interrupt_free_min) || (req_class == VM_ALLOC_INTERRUPT && vm_cnt.v_free_count >= npages)) { /* * Can we allocate the pages from a reservation? */ #if VM_NRESERVLEVEL > 0 retry: if (object == NULL || (object->flags & OBJ_COLORED) == 0 || (m_ret = vm_reserv_alloc_contig(object, pindex, npages, low, high, alignment, boundary, mpred)) == NULL) #endif /* * If not, allocate them from the free page queues. */ m_ret = vm_phys_alloc_contig(npages, low, high, alignment, boundary); } else { mtx_unlock(&vm_page_queue_free_mtx); atomic_add_int(&vm_pageout_deficit, npages); pagedaemon_wakeup(); return (NULL); } if (m_ret != NULL) vm_phys_freecnt_adj(m_ret, -npages); else { #if VM_NRESERVLEVEL > 0 if (vm_reserv_reclaim_contig(npages, low, high, alignment, boundary)) goto retry; #endif } mtx_unlock(&vm_page_queue_free_mtx); if (m_ret == NULL) return (NULL); for (m = m_ret; m < &m_ret[npages]; m++) vm_page_alloc_check(m); /* * Initialize the pages. Only the PG_ZERO flag is inherited. */ flags = 0; if ((req & VM_ALLOC_ZERO) != 0) flags = PG_ZERO; if ((req & VM_ALLOC_NODUMP) != 0) flags |= PG_NODUMP; oflags = object == NULL || (object->flags & OBJ_UNMANAGED) != 0 ? VPO_UNMANAGED : 0; busy_lock = VPB_UNBUSIED; if ((req & (VM_ALLOC_NOBUSY | VM_ALLOC_NOOBJ | VM_ALLOC_SBUSY)) == 0) busy_lock = VPB_SINGLE_EXCLUSIVER; if ((req & VM_ALLOC_SBUSY) != 0) busy_lock = VPB_SHARERS_WORD(1); if ((req & VM_ALLOC_WIRED) != 0) atomic_add_int(&vm_cnt.v_wire_count, npages); if (object != NULL) { if (object->memattr != VM_MEMATTR_DEFAULT && memattr == VM_MEMATTR_DEFAULT) memattr = object->memattr; } for (m = m_ret; m < &m_ret[npages]; m++) { m->aflags = 0; m->flags = (m->flags | PG_NODUMP) & flags; m->busy_lock = busy_lock; if ((req & VM_ALLOC_WIRED) != 0) m->wire_count = 1; m->act_count = 0; m->oflags = oflags; if (object != NULL) { if (vm_page_insert_after(m, object, pindex, mpred)) { pagedaemon_wakeup(); if ((req & VM_ALLOC_WIRED) != 0) atomic_subtract_int( &vm_cnt.v_wire_count, npages); KASSERT(m->object == NULL, ("page %p has object", m)); mpred = m; for (m = m_ret; m < &m_ret[npages]; m++) { if (m <= mpred && (req & VM_ALLOC_WIRED) != 0) m->wire_count = 0; m->oflags = VPO_UNMANAGED; m->busy_lock = VPB_UNBUSIED; /* Don't change PG_ZERO. */ vm_page_free_toq(m); } return (NULL); } mpred = m; } else m->pindex = pindex; if (memattr != VM_MEMATTR_DEFAULT) pmap_page_set_memattr(m, memattr); pindex++; } if (vm_paging_needed()) pagedaemon_wakeup(); return (m_ret); } /* * Check a page that has been freshly dequeued from a freelist. */ static void vm_page_alloc_check(vm_page_t m) { KASSERT(m->object == NULL, ("page %p has object", m)); KASSERT(m->queue == PQ_NONE, ("page %p has unexpected queue %d", m, m->queue)); KASSERT(m->wire_count == 0, ("page %p is wired", m)); KASSERT(m->hold_count == 0, ("page %p is held", m)); KASSERT(!vm_page_busied(m), ("page %p is busy", m)); KASSERT(m->dirty == 0, ("page %p is dirty", m)); KASSERT(pmap_page_get_memattr(m) == VM_MEMATTR_DEFAULT, ("page %p has unexpected memattr %d", m, pmap_page_get_memattr(m))); KASSERT(m->valid == 0, ("free page %p is valid", m)); } /* * vm_page_alloc_freelist: * * Allocate a physical page from the specified free page list. * * The caller must always specify an allocation class. * * allocation classes: * VM_ALLOC_NORMAL normal process request * VM_ALLOC_SYSTEM system *really* needs a page * VM_ALLOC_INTERRUPT interrupt time request * * optional allocation flags: * VM_ALLOC_COUNT(number) the number of additional pages that the caller * intends to allocate * VM_ALLOC_WIRED wire the allocated page * VM_ALLOC_ZERO prefer a zeroed page * * This routine may not sleep. */ vm_page_t vm_page_alloc_freelist(int flind, int req) { vm_page_t m; u_int flags; int req_class; req_class = req & VM_ALLOC_CLASS_MASK; /* * The page daemon is allowed to dig deeper into the free page list. */ if (curproc == pageproc && req_class != VM_ALLOC_INTERRUPT) req_class = VM_ALLOC_SYSTEM; /* * Do not allocate reserved pages unless the req has asked for it. */ mtx_lock(&vm_page_queue_free_mtx); if (vm_cnt.v_free_count > vm_cnt.v_free_reserved || (req_class == VM_ALLOC_SYSTEM && vm_cnt.v_free_count > vm_cnt.v_interrupt_free_min) || (req_class == VM_ALLOC_INTERRUPT && vm_cnt.v_free_count > 0)) m = vm_phys_alloc_freelist_pages(flind, VM_FREEPOOL_DIRECT, 0); else { mtx_unlock(&vm_page_queue_free_mtx); atomic_add_int(&vm_pageout_deficit, max((u_int)req >> VM_ALLOC_COUNT_SHIFT, 1)); pagedaemon_wakeup(); return (NULL); } if (m == NULL) { mtx_unlock(&vm_page_queue_free_mtx); return (NULL); } vm_phys_freecnt_adj(m, -1); mtx_unlock(&vm_page_queue_free_mtx); vm_page_alloc_check(m); /* * Initialize the page. Only the PG_ZERO flag is inherited. */ m->aflags = 0; flags = 0; if ((req & VM_ALLOC_ZERO) != 0) flags = PG_ZERO; m->flags &= flags; if ((req & VM_ALLOC_WIRED) != 0) { /* * The page lock is not required for wiring a page that does * not belong to an object. */ atomic_add_int(&vm_cnt.v_wire_count, 1); m->wire_count = 1; } /* Unmanaged pages don't use "act_count". */ m->oflags = VPO_UNMANAGED; if (vm_paging_needed()) pagedaemon_wakeup(); return (m); } #define VPSC_ANY 0 /* No restrictions. */ #define VPSC_NORESERV 1 /* Skip reservations; implies VPSC_NOSUPER. */ #define VPSC_NOSUPER 2 /* Skip superpages. */ /* * vm_page_scan_contig: * * Scan vm_page_array[] between the specified entries "m_start" and * "m_end" for a run of contiguous physical pages that satisfy the * specified conditions, and return the lowest page in the run. The * specified "alignment" determines the alignment of the lowest physical * page in the run. If the specified "boundary" is non-zero, then the * run of physical pages cannot span a physical address that is a * multiple of "boundary". * * "m_end" is never dereferenced, so it need not point to a vm_page * structure within vm_page_array[]. * * "npages" must be greater than zero. "m_start" and "m_end" must not * span a hole (or discontiguity) in the physical address space. Both * "alignment" and "boundary" must be a power of two. */ vm_page_t vm_page_scan_contig(u_long npages, vm_page_t m_start, vm_page_t m_end, u_long alignment, vm_paddr_t boundary, int options) { struct mtx *m_mtx, *new_mtx; vm_object_t object; vm_paddr_t pa; vm_page_t m, m_run; #if VM_NRESERVLEVEL > 0 int level; #endif int m_inc, order, run_ext, run_len; KASSERT(npages > 0, ("npages is 0")); KASSERT(powerof2(alignment), ("alignment is not a power of 2")); KASSERT(powerof2(boundary), ("boundary is not a power of 2")); m_run = NULL; run_len = 0; m_mtx = NULL; for (m = m_start; m < m_end && run_len < npages; m += m_inc) { KASSERT((m->flags & (PG_FICTITIOUS | PG_MARKER)) == 0, ("page %p is PG_FICTITIOUS or PG_MARKER", m)); /* * If the current page would be the start of a run, check its * physical address against the end, alignment, and boundary * conditions. If it doesn't satisfy these conditions, either * terminate the scan or advance to the next page that * satisfies the failed condition. */ if (run_len == 0) { KASSERT(m_run == NULL, ("m_run != NULL")); if (m + npages > m_end) break; pa = VM_PAGE_TO_PHYS(m); if ((pa & (alignment - 1)) != 0) { m_inc = atop(roundup2(pa, alignment) - pa); continue; } if (rounddown2(pa ^ (pa + ptoa(npages) - 1), boundary) != 0) { m_inc = atop(roundup2(pa, boundary) - pa); continue; } } else KASSERT(m_run != NULL, ("m_run == NULL")); /* * Avoid releasing and reacquiring the same page lock. */ new_mtx = vm_page_lockptr(m); if (m_mtx != new_mtx) { if (m_mtx != NULL) mtx_unlock(m_mtx); m_mtx = new_mtx; mtx_lock(m_mtx); } m_inc = 1; retry: if (m->wire_count != 0 || m->hold_count != 0) run_ext = 0; #if VM_NRESERVLEVEL > 0 else if ((level = vm_reserv_level(m)) >= 0 && (options & VPSC_NORESERV) != 0) { run_ext = 0; /* Advance to the end of the reservation. */ pa = VM_PAGE_TO_PHYS(m); m_inc = atop(roundup2(pa + 1, vm_reserv_size(level)) - pa); } #endif else if ((object = m->object) != NULL) { /* * The page is considered eligible for relocation if * and only if it could be laundered or reclaimed by * the page daemon. */ if (!VM_OBJECT_TRYRLOCK(object)) { mtx_unlock(m_mtx); VM_OBJECT_RLOCK(object); mtx_lock(m_mtx); if (m->object != object) { /* * The page may have been freed. */ VM_OBJECT_RUNLOCK(object); goto retry; } else if (m->wire_count != 0 || m->hold_count != 0) { run_ext = 0; goto unlock; } } KASSERT((m->flags & PG_UNHOLDFREE) == 0, ("page %p is PG_UNHOLDFREE", m)); /* Don't care: PG_NODUMP, PG_ZERO. */ if (object->type != OBJT_DEFAULT && object->type != OBJT_SWAP && object->type != OBJT_VNODE) { run_ext = 0; #if VM_NRESERVLEVEL > 0 } else if ((options & VPSC_NOSUPER) != 0 && (level = vm_reserv_level_iffullpop(m)) >= 0) { run_ext = 0; /* Advance to the end of the superpage. */ pa = VM_PAGE_TO_PHYS(m); m_inc = atop(roundup2(pa + 1, vm_reserv_size(level)) - pa); #endif } else if (object->memattr == VM_MEMATTR_DEFAULT && m->queue != PQ_NONE && !vm_page_busied(m)) { /* * The page is allocated but eligible for * relocation. Extend the current run by one * page. */ KASSERT(pmap_page_get_memattr(m) == VM_MEMATTR_DEFAULT, ("page %p has an unexpected memattr", m)); KASSERT((m->oflags & (VPO_SWAPINPROG | VPO_SWAPSLEEP | VPO_UNMANAGED)) == 0, ("page %p has unexpected oflags", m)); /* Don't care: VPO_NOSYNC. */ run_ext = 1; } else run_ext = 0; unlock: VM_OBJECT_RUNLOCK(object); #if VM_NRESERVLEVEL > 0 } else if (level >= 0) { /* * The page is reserved but not yet allocated. In * other words, it is still free. Extend the current * run by one page. */ run_ext = 1; #endif } else if ((order = m->order) < VM_NFREEORDER) { /* * The page is enqueued in the physical memory * allocator's free page queues. Moreover, it is the * first page in a power-of-two-sized run of * contiguous free pages. Add these pages to the end * of the current run, and jump ahead. */ run_ext = 1 << order; m_inc = 1 << order; } else { /* * Skip the page for one of the following reasons: (1) * It is enqueued in the physical memory allocator's * free page queues. However, it is not the first * page in a run of contiguous free pages. (This case * rarely occurs because the scan is performed in * ascending order.) (2) It is not reserved, and it is * transitioning from free to allocated. (Conversely, * the transition from allocated to free for managed * pages is blocked by the page lock.) (3) It is * allocated but not contained by an object and not * wired, e.g., allocated by Xen's balloon driver. */ run_ext = 0; } /* * Extend or reset the current run of pages. */ if (run_ext > 0) { if (run_len == 0) m_run = m; run_len += run_ext; } else { if (run_len > 0) { m_run = NULL; run_len = 0; } } } if (m_mtx != NULL) mtx_unlock(m_mtx); if (run_len >= npages) return (m_run); return (NULL); } /* * vm_page_reclaim_run: * * Try to relocate each of the allocated virtual pages within the * specified run of physical pages to a new physical address. Free the * physical pages underlying the relocated virtual pages. A virtual page * is relocatable if and only if it could be laundered or reclaimed by * the page daemon. Whenever possible, a virtual page is relocated to a * physical address above "high". * * Returns 0 if every physical page within the run was already free or * just freed by a successful relocation. Otherwise, returns a non-zero * value indicating why the last attempt to relocate a virtual page was * unsuccessful. * * "req_class" must be an allocation class. */ static int vm_page_reclaim_run(int req_class, u_long npages, vm_page_t m_run, vm_paddr_t high) { struct mtx *m_mtx, *new_mtx; struct spglist free; vm_object_t object; vm_paddr_t pa; vm_page_t m, m_end, m_new; int error, order, req; KASSERT((req_class & VM_ALLOC_CLASS_MASK) == req_class, ("req_class is not an allocation class")); SLIST_INIT(&free); error = 0; m = m_run; m_end = m_run + npages; m_mtx = NULL; for (; error == 0 && m < m_end; m++) { KASSERT((m->flags & (PG_FICTITIOUS | PG_MARKER)) == 0, ("page %p is PG_FICTITIOUS or PG_MARKER", m)); /* * Avoid releasing and reacquiring the same page lock. */ new_mtx = vm_page_lockptr(m); if (m_mtx != new_mtx) { if (m_mtx != NULL) mtx_unlock(m_mtx); m_mtx = new_mtx; mtx_lock(m_mtx); } retry: if (m->wire_count != 0 || m->hold_count != 0) error = EBUSY; else if ((object = m->object) != NULL) { /* * The page is relocated if and only if it could be * laundered or reclaimed by the page daemon. */ if (!VM_OBJECT_TRYWLOCK(object)) { mtx_unlock(m_mtx); VM_OBJECT_WLOCK(object); mtx_lock(m_mtx); if (m->object != object) { /* * The page may have been freed. */ VM_OBJECT_WUNLOCK(object); goto retry; } else if (m->wire_count != 0 || m->hold_count != 0) { error = EBUSY; goto unlock; } } KASSERT((m->flags & PG_UNHOLDFREE) == 0, ("page %p is PG_UNHOLDFREE", m)); /* Don't care: PG_NODUMP, PG_ZERO. */ if (object->type != OBJT_DEFAULT && object->type != OBJT_SWAP && object->type != OBJT_VNODE) error = EINVAL; else if (object->memattr != VM_MEMATTR_DEFAULT) error = EINVAL; else if (m->queue != PQ_NONE && !vm_page_busied(m)) { KASSERT(pmap_page_get_memattr(m) == VM_MEMATTR_DEFAULT, ("page %p has an unexpected memattr", m)); KASSERT((m->oflags & (VPO_SWAPINPROG | VPO_SWAPSLEEP | VPO_UNMANAGED)) == 0, ("page %p has unexpected oflags", m)); /* Don't care: VPO_NOSYNC. */ if (m->valid != 0) { /* * First, try to allocate a new page * that is above "high". Failing * that, try to allocate a new page * that is below "m_run". Allocate * the new page between the end of * "m_run" and "high" only as a last * resort. */ req = req_class | VM_ALLOC_NOOBJ; if ((m->flags & PG_NODUMP) != 0) req |= VM_ALLOC_NODUMP; if (trunc_page(high) != ~(vm_paddr_t)PAGE_MASK) { m_new = vm_page_alloc_contig( NULL, 0, req, 1, round_page(high), ~(vm_paddr_t)0, PAGE_SIZE, 0, VM_MEMATTR_DEFAULT); } else m_new = NULL; if (m_new == NULL) { pa = VM_PAGE_TO_PHYS(m_run); m_new = vm_page_alloc_contig( NULL, 0, req, 1, 0, pa - 1, PAGE_SIZE, 0, VM_MEMATTR_DEFAULT); } if (m_new == NULL) { pa += ptoa(npages); m_new = vm_page_alloc_contig( NULL, 0, req, 1, pa, high, PAGE_SIZE, 0, VM_MEMATTR_DEFAULT); } if (m_new == NULL) { error = ENOMEM; goto unlock; } KASSERT(m_new->wire_count == 0, ("page %p is wired", m)); /* * Replace "m" with the new page. For * vm_page_replace(), "m" must be busy * and dequeued. Finally, change "m" * as if vm_page_free() was called. */ if (object->ref_count != 0) pmap_remove_all(m); m_new->aflags = m->aflags; KASSERT(m_new->oflags == VPO_UNMANAGED, ("page %p is managed", m)); m_new->oflags = m->oflags & VPO_NOSYNC; pmap_copy_page(m, m_new); m_new->valid = m->valid; m_new->dirty = m->dirty; m->flags &= ~PG_ZERO; vm_page_xbusy(m); vm_page_remque(m); vm_page_replace_checked(m_new, object, m->pindex, m); m->valid = 0; vm_page_undirty(m); /* * The new page must be deactivated * before the object is unlocked. */ new_mtx = vm_page_lockptr(m_new); if (m_mtx != new_mtx) { mtx_unlock(m_mtx); m_mtx = new_mtx; mtx_lock(m_mtx); } vm_page_deactivate(m_new); } else { m->flags &= ~PG_ZERO; vm_page_remque(m); vm_page_remove(m); KASSERT(m->dirty == 0, ("page %p is dirty", m)); } SLIST_INSERT_HEAD(&free, m, plinks.s.ss); } else error = EBUSY; unlock: VM_OBJECT_WUNLOCK(object); } else { mtx_lock(&vm_page_queue_free_mtx); order = m->order; if (order < VM_NFREEORDER) { /* * The page is enqueued in the physical memory * allocator's free page queues. Moreover, it * is the first page in a power-of-two-sized * run of contiguous free pages. Jump ahead * to the last page within that run, and * continue from there. */ m += (1 << order) - 1; } #if VM_NRESERVLEVEL > 0 else if (vm_reserv_is_page_free(m)) order = 0; #endif mtx_unlock(&vm_page_queue_free_mtx); if (order == VM_NFREEORDER) error = EINVAL; } } if (m_mtx != NULL) mtx_unlock(m_mtx); if ((m = SLIST_FIRST(&free)) != NULL) { mtx_lock(&vm_page_queue_free_mtx); do { SLIST_REMOVE_HEAD(&free, plinks.s.ss); vm_phys_freecnt_adj(m, 1); #if VM_NRESERVLEVEL > 0 if (!vm_reserv_free_page(m)) #else if (true) #endif vm_phys_free_pages(m, 0); } while ((m = SLIST_FIRST(&free)) != NULL); vm_page_free_wakeup(); mtx_unlock(&vm_page_queue_free_mtx); } return (error); } #define NRUNS 16 CTASSERT(powerof2(NRUNS)); #define RUN_INDEX(count) ((count) & (NRUNS - 1)) #define MIN_RECLAIM 8 /* * vm_page_reclaim_contig: * * Reclaim allocated, contiguous physical memory satisfying the specified * conditions by relocating the virtual pages using that physical memory. * Returns true if reclamation is successful and false otherwise. Since * relocation requires the allocation of physical pages, reclamation may * fail due to a shortage of free pages. When reclamation fails, callers * are expected to perform VM_WAIT before retrying a failed allocation * operation, e.g., vm_page_alloc_contig(). * * The caller must always specify an allocation class through "req". * * allocation classes: * VM_ALLOC_NORMAL normal process request * VM_ALLOC_SYSTEM system *really* needs a page * VM_ALLOC_INTERRUPT interrupt time request * * The optional allocation flags are ignored. * * "npages" must be greater than zero. Both "alignment" and "boundary" * must be a power of two. */ bool vm_page_reclaim_contig(int req, u_long npages, vm_paddr_t low, vm_paddr_t high, u_long alignment, vm_paddr_t boundary) { vm_paddr_t curr_low; vm_page_t m_run, m_runs[NRUNS]; u_long count, reclaimed; int error, i, options, req_class; KASSERT(npages > 0, ("npages is 0")); KASSERT(powerof2(alignment), ("alignment is not a power of 2")); KASSERT(powerof2(boundary), ("boundary is not a power of 2")); req_class = req & VM_ALLOC_CLASS_MASK; /* * The page daemon is allowed to dig deeper into the free page list. */ if (curproc == pageproc && req_class != VM_ALLOC_INTERRUPT) req_class = VM_ALLOC_SYSTEM; /* * Return if the number of free pages cannot satisfy the requested * allocation. */ count = vm_cnt.v_free_count; if (count < npages + vm_cnt.v_free_reserved || (count < npages + vm_cnt.v_interrupt_free_min && req_class == VM_ALLOC_SYSTEM) || (count < npages && req_class == VM_ALLOC_INTERRUPT)) return (false); /* * Scan up to three times, relaxing the restrictions ("options") on * the reclamation of reservations and superpages each time. */ for (options = VPSC_NORESERV;;) { /* * Find the highest runs that satisfy the given constraints * and restrictions, and record them in "m_runs". */ curr_low = low; count = 0; for (;;) { m_run = vm_phys_scan_contig(npages, curr_low, high, alignment, boundary, options); if (m_run == NULL) break; curr_low = VM_PAGE_TO_PHYS(m_run) + ptoa(npages); m_runs[RUN_INDEX(count)] = m_run; count++; } /* * Reclaim the highest runs in LIFO (descending) order until * the number of reclaimed pages, "reclaimed", is at least * MIN_RECLAIM. Reset "reclaimed" each time because each * reclamation is idempotent, and runs will (likely) recur * from one scan to the next as restrictions are relaxed. */ reclaimed = 0; for (i = 0; count > 0 && i < NRUNS; i++) { count--; m_run = m_runs[RUN_INDEX(count)]; error = vm_page_reclaim_run(req_class, npages, m_run, high); if (error == 0) { reclaimed += npages; if (reclaimed >= MIN_RECLAIM) return (true); } } /* * Either relax the restrictions on the next scan or return if * the last scan had no restrictions. */ if (options == VPSC_NORESERV) options = VPSC_NOSUPER; else if (options == VPSC_NOSUPER) options = VPSC_ANY; else if (options == VPSC_ANY) return (reclaimed != 0); } } /* * vm_wait: (also see VM_WAIT macro) * * Sleep until free pages are available for allocation. * - Called in various places before memory allocations. */ void vm_wait(void) { mtx_lock(&vm_page_queue_free_mtx); if (curproc == pageproc) { vm_pageout_pages_needed = 1; msleep(&vm_pageout_pages_needed, &vm_page_queue_free_mtx, PDROP | PSWP, "VMWait", 0); } else { if (__predict_false(pageproc == NULL)) panic("vm_wait in early boot"); if (!vm_pageout_wanted) { vm_pageout_wanted = true; wakeup(&vm_pageout_wanted); } vm_pages_needed = true; msleep(&vm_cnt.v_free_count, &vm_page_queue_free_mtx, PDROP | PVM, "vmwait", 0); } } /* * vm_waitpfault: (also see VM_WAITPFAULT macro) * * Sleep until free pages are available for allocation. * - Called only in vm_fault so that processes page faulting * can be easily tracked. * - Sleeps at a lower priority than vm_wait() so that vm_wait()ing * processes will be able to grab memory first. Do not change * this balance without careful testing first. */ void vm_waitpfault(void) { mtx_lock(&vm_page_queue_free_mtx); if (!vm_pageout_wanted) { vm_pageout_wanted = true; wakeup(&vm_pageout_wanted); } vm_pages_needed = true; msleep(&vm_cnt.v_free_count, &vm_page_queue_free_mtx, PDROP | PUSER, "pfault", 0); } struct vm_pagequeue * vm_page_pagequeue(vm_page_t m) { if (vm_page_in_laundry(m)) return (&vm_dom[0].vmd_pagequeues[m->queue]); else return (&vm_phys_domain(m)->vmd_pagequeues[m->queue]); } /* * vm_page_dequeue: * * Remove the given page from its current page queue. * * The page must be locked. */ void vm_page_dequeue(vm_page_t m) { struct vm_pagequeue *pq; vm_page_assert_locked(m); KASSERT(m->queue < PQ_COUNT, ("vm_page_dequeue: page %p is not queued", m)); pq = vm_page_pagequeue(m); vm_pagequeue_lock(pq); m->queue = PQ_NONE; TAILQ_REMOVE(&pq->pq_pl, m, plinks.q); vm_pagequeue_cnt_dec(pq); vm_pagequeue_unlock(pq); } /* * vm_page_dequeue_locked: * * Remove the given page from its current page queue. * * The page and page queue must be locked. */ void vm_page_dequeue_locked(vm_page_t m) { struct vm_pagequeue *pq; vm_page_lock_assert(m, MA_OWNED); pq = vm_page_pagequeue(m); vm_pagequeue_assert_locked(pq); m->queue = PQ_NONE; TAILQ_REMOVE(&pq->pq_pl, m, plinks.q); vm_pagequeue_cnt_dec(pq); } /* * vm_page_enqueue: * * Add the given page to the specified page queue. * * The page must be locked. */ static void vm_page_enqueue(uint8_t queue, vm_page_t m) { struct vm_pagequeue *pq; vm_page_lock_assert(m, MA_OWNED); KASSERT(queue < PQ_COUNT, ("vm_page_enqueue: invalid queue %u request for page %p", queue, m)); if (queue == PQ_LAUNDRY || queue == PQ_UNSWAPPABLE) pq = &vm_dom[0].vmd_pagequeues[queue]; else pq = &vm_phys_domain(m)->vmd_pagequeues[queue]; vm_pagequeue_lock(pq); m->queue = queue; TAILQ_INSERT_TAIL(&pq->pq_pl, m, plinks.q); vm_pagequeue_cnt_inc(pq); vm_pagequeue_unlock(pq); } /* * vm_page_requeue: * * Move the given page to the tail of its current page queue. * * The page must be locked. */ void vm_page_requeue(vm_page_t m) { struct vm_pagequeue *pq; vm_page_lock_assert(m, MA_OWNED); KASSERT(m->queue != PQ_NONE, ("vm_page_requeue: page %p is not queued", m)); pq = vm_page_pagequeue(m); vm_pagequeue_lock(pq); TAILQ_REMOVE(&pq->pq_pl, m, plinks.q); TAILQ_INSERT_TAIL(&pq->pq_pl, m, plinks.q); vm_pagequeue_unlock(pq); } /* * vm_page_requeue_locked: * * Move the given page to the tail of its current page queue. * * The page queue must be locked. */ void vm_page_requeue_locked(vm_page_t m) { struct vm_pagequeue *pq; KASSERT(m->queue != PQ_NONE, ("vm_page_requeue_locked: page %p is not queued", m)); pq = vm_page_pagequeue(m); vm_pagequeue_assert_locked(pq); TAILQ_REMOVE(&pq->pq_pl, m, plinks.q); TAILQ_INSERT_TAIL(&pq->pq_pl, m, plinks.q); } /* * vm_page_activate: * * Put the specified page on the active list (if appropriate). * Ensure that act_count is at least ACT_INIT but do not otherwise * mess with it. * * The page must be locked. */ void vm_page_activate(vm_page_t m) { int queue; vm_page_lock_assert(m, MA_OWNED); if ((queue = m->queue) != PQ_ACTIVE) { if (m->wire_count == 0 && (m->oflags & VPO_UNMANAGED) == 0) { if (m->act_count < ACT_INIT) m->act_count = ACT_INIT; if (queue != PQ_NONE) vm_page_dequeue(m); vm_page_enqueue(PQ_ACTIVE, m); } else KASSERT(queue == PQ_NONE, ("vm_page_activate: wired page %p is queued", m)); } else { if (m->act_count < ACT_INIT) m->act_count = ACT_INIT; } } /* * vm_page_free_wakeup: * * Helper routine for vm_page_free_toq(). This routine is called * when a page is added to the free queues. * * The page queues must be locked. */ static inline void vm_page_free_wakeup(void) { mtx_assert(&vm_page_queue_free_mtx, MA_OWNED); /* * if pageout daemon needs pages, then tell it that there are * some free. */ if (vm_pageout_pages_needed && vm_cnt.v_free_count >= vm_cnt.v_pageout_free_min) { wakeup(&vm_pageout_pages_needed); vm_pageout_pages_needed = 0; } /* * wakeup processes that are waiting on memory if we hit a * high water mark. And wakeup scheduler process if we have * lots of memory. this process will swapin processes. */ if (vm_pages_needed && !vm_page_count_min()) { vm_pages_needed = false; wakeup(&vm_cnt.v_free_count); } } /* * vm_page_free_toq: * * Returns the given page to the free list, * disassociating it with any VM object. * * The object must be locked. The page must be locked if it is managed. */ void vm_page_free_toq(vm_page_t m) { if ((m->oflags & VPO_UNMANAGED) == 0) { vm_page_lock_assert(m, MA_OWNED); KASSERT(!pmap_page_is_mapped(m), ("vm_page_free_toq: freeing mapped page %p", m)); } else KASSERT(m->queue == PQ_NONE, ("vm_page_free_toq: unmanaged page %p is queued", m)); PCPU_INC(cnt.v_tfree); if (vm_page_sbusied(m)) panic("vm_page_free: freeing busy page %p", m); /* * Unqueue, then remove page. Note that we cannot destroy * the page here because we do not want to call the pager's * callback routine until after we've put the page on the * appropriate free queue. */ vm_page_remque(m); vm_page_remove(m); /* * If fictitious remove object association and * return, otherwise delay object association removal. */ if ((m->flags & PG_FICTITIOUS) != 0) { return; } m->valid = 0; vm_page_undirty(m); if (m->wire_count != 0) panic("vm_page_free: freeing wired page %p", m); if (m->hold_count != 0) { m->flags &= ~PG_ZERO; KASSERT((m->flags & PG_UNHOLDFREE) == 0, ("vm_page_free: freeing PG_UNHOLDFREE page %p", m)); m->flags |= PG_UNHOLDFREE; } else { /* * Restore the default memory attribute to the page. */ if (pmap_page_get_memattr(m) != VM_MEMATTR_DEFAULT) pmap_page_set_memattr(m, VM_MEMATTR_DEFAULT); /* * Insert the page into the physical memory allocator's free * page queues. */ mtx_lock(&vm_page_queue_free_mtx); vm_phys_freecnt_adj(m, 1); #if VM_NRESERVLEVEL > 0 if (!vm_reserv_free_page(m)) #else if (TRUE) #endif vm_phys_free_pages(m, 0); vm_page_free_wakeup(); mtx_unlock(&vm_page_queue_free_mtx); } } /* * vm_page_wire: * * Mark this page as wired down by yet * another map, removing it from paging queues * as necessary. * * If the page is fictitious, then its wire count must remain one. * * The page must be locked. */ void vm_page_wire(vm_page_t m) { /* * Only bump the wire statistics if the page is not already wired, * and only unqueue the page if it is on some queue (if it is unmanaged * it is already off the queues). */ vm_page_lock_assert(m, MA_OWNED); if ((m->flags & PG_FICTITIOUS) != 0) { KASSERT(m->wire_count == 1, ("vm_page_wire: fictitious page %p's wire count isn't one", m)); return; } if (m->wire_count == 0) { KASSERT((m->oflags & VPO_UNMANAGED) == 0 || m->queue == PQ_NONE, ("vm_page_wire: unmanaged page %p is queued", m)); vm_page_remque(m); atomic_add_int(&vm_cnt.v_wire_count, 1); } m->wire_count++; KASSERT(m->wire_count != 0, ("vm_page_wire: wire_count overflow m=%p", m)); } /* * vm_page_unwire: * * Release one wiring of the specified page, potentially allowing it to be * paged out. Returns TRUE if the number of wirings transitions to zero and * FALSE otherwise. * * Only managed pages belonging to an object can be paged out. If the number * of wirings transitions to zero and the page is eligible for page out, then * the page is added to the specified paging queue (unless PQ_NONE is * specified). * * If a page is fictitious, then its wire count must always be one. * * A managed page must be locked. */ boolean_t vm_page_unwire(vm_page_t m, uint8_t queue) { KASSERT(queue < PQ_COUNT || queue == PQ_NONE, ("vm_page_unwire: invalid queue %u request for page %p", queue, m)); if ((m->oflags & VPO_UNMANAGED) == 0) vm_page_assert_locked(m); if ((m->flags & PG_FICTITIOUS) != 0) { KASSERT(m->wire_count == 1, ("vm_page_unwire: fictitious page %p's wire count isn't one", m)); return (FALSE); } if (m->wire_count > 0) { m->wire_count--; if (m->wire_count == 0) { atomic_subtract_int(&vm_cnt.v_wire_count, 1); if ((m->oflags & VPO_UNMANAGED) == 0 && m->object != NULL && queue != PQ_NONE) vm_page_enqueue(queue, m); return (TRUE); } else return (FALSE); } else panic("vm_page_unwire: page %p's wire count is zero", m); } /* * Move the specified page to the inactive queue. * * Normally, "noreuse" is FALSE, resulting in LRU ordering of the inactive * queue. However, setting "noreuse" to TRUE will accelerate the specified * page's reclamation, but it will not unmap the page from any address space. * This is implemented by inserting the page near the head of the inactive * queue, using a marker page to guide FIFO insertion ordering. * * The page must be locked. */ static inline void _vm_page_deactivate(vm_page_t m, boolean_t noreuse) { struct vm_pagequeue *pq; int queue; vm_page_assert_locked(m); /* * Ignore if the page is already inactive, unless it is unlikely to be * reactivated. */ if ((queue = m->queue) == PQ_INACTIVE && !noreuse) return; if (m->wire_count == 0 && (m->oflags & VPO_UNMANAGED) == 0) { pq = &vm_phys_domain(m)->vmd_pagequeues[PQ_INACTIVE]; /* Avoid multiple acquisitions of the inactive queue lock. */ if (queue == PQ_INACTIVE) { vm_pagequeue_lock(pq); vm_page_dequeue_locked(m); } else { if (queue != PQ_NONE) vm_page_dequeue(m); vm_pagequeue_lock(pq); } m->queue = PQ_INACTIVE; if (noreuse) TAILQ_INSERT_BEFORE(&vm_phys_domain(m)->vmd_inacthead, m, plinks.q); else TAILQ_INSERT_TAIL(&pq->pq_pl, m, plinks.q); vm_pagequeue_cnt_inc(pq); vm_pagequeue_unlock(pq); } } /* * Move the specified page to the inactive queue. * * The page must be locked. */ void vm_page_deactivate(vm_page_t m) { _vm_page_deactivate(m, FALSE); } /* * Move the specified page to the inactive queue with the expectation * that it is unlikely to be reused. * * The page must be locked. */ void vm_page_deactivate_noreuse(vm_page_t m) { _vm_page_deactivate(m, TRUE); } /* * vm_page_launder * * Put a page in the laundry. */ void vm_page_launder(vm_page_t m) { int queue; vm_page_assert_locked(m); if ((queue = m->queue) != PQ_LAUNDRY) { if (m->wire_count == 0 && (m->oflags & VPO_UNMANAGED) == 0) { if (queue != PQ_NONE) vm_page_dequeue(m); vm_page_enqueue(PQ_LAUNDRY, m); } else KASSERT(queue == PQ_NONE, ("wired page %p is queued", m)); } } /* * vm_page_unswappable * * Put a page in the PQ_UNSWAPPABLE holding queue. */ void vm_page_unswappable(vm_page_t m) { vm_page_assert_locked(m); KASSERT(m->wire_count == 0 && (m->oflags & VPO_UNMANAGED) == 0, ("page %p already unswappable", m)); if (m->queue != PQ_NONE) vm_page_dequeue(m); vm_page_enqueue(PQ_UNSWAPPABLE, m); } /* * vm_page_try_to_free() * * Attempt to free the page. If we cannot free it, we do nothing. * 1 is returned on success, 0 on failure. */ int vm_page_try_to_free(vm_page_t m) { vm_page_lock_assert(m, MA_OWNED); if (m->object != NULL) VM_OBJECT_ASSERT_WLOCKED(m->object); if (m->dirty || m->hold_count || m->wire_count || (m->oflags & VPO_UNMANAGED) != 0 || vm_page_busied(m)) return (0); pmap_remove_all(m); if (m->dirty) return (0); vm_page_free(m); return (1); } /* * vm_page_advise * * Deactivate or do nothing, as appropriate. * * The object and page must be locked. */ void vm_page_advise(vm_page_t m, int advice) { vm_page_assert_locked(m); VM_OBJECT_ASSERT_WLOCKED(m->object); if (advice == MADV_FREE) /* * Mark the page clean. This will allow the page to be freed * without first paging it out. MADV_FREE pages are often * quickly reused by malloc(3), so we do not do anything that * would result in a page fault on a later access. */ vm_page_undirty(m); else if (advice != MADV_DONTNEED) return; /* * Clear any references to the page. Otherwise, the page daemon will * immediately reactivate the page. */ vm_page_aflag_clear(m, PGA_REFERENCED); if (advice != MADV_FREE && m->dirty == 0 && pmap_is_modified(m)) vm_page_dirty(m); /* * Place clean pages near the head of the inactive queue rather than * the tail, thus defeating the queue's LRU operation and ensuring that * the page will be reused quickly. Dirty pages not already in the * laundry are moved there. */ if (m->dirty == 0) vm_page_deactivate_noreuse(m); else vm_page_launder(m); } /* * Grab a page, waiting until we are waken up due to the page * changing state. We keep on waiting, if the page continues * to be in the object. If the page doesn't exist, first allocate it * and then conditionally zero it. * * This routine may sleep. * * The object must be locked on entry. The lock will, however, be released * and reacquired if the routine sleeps. */ vm_page_t vm_page_grab(vm_object_t object, vm_pindex_t pindex, int allocflags) { vm_page_t m; int sleep; VM_OBJECT_ASSERT_WLOCKED(object); KASSERT((allocflags & VM_ALLOC_SBUSY) == 0 || (allocflags & VM_ALLOC_IGN_SBUSY) != 0, ("vm_page_grab: VM_ALLOC_SBUSY/VM_ALLOC_IGN_SBUSY mismatch")); retrylookup: if ((m = vm_page_lookup(object, pindex)) != NULL) { sleep = (allocflags & VM_ALLOC_IGN_SBUSY) != 0 ? vm_page_xbusied(m) : vm_page_busied(m); if (sleep) { if ((allocflags & VM_ALLOC_NOWAIT) != 0) return (NULL); /* * Reference the page before unlocking and * sleeping so that the page daemon is less * likely to reclaim it. */ vm_page_aflag_set(m, PGA_REFERENCED); vm_page_lock(m); VM_OBJECT_WUNLOCK(object); vm_page_busy_sleep(m, "pgrbwt", (allocflags & VM_ALLOC_IGN_SBUSY) != 0); VM_OBJECT_WLOCK(object); goto retrylookup; } else { if ((allocflags & VM_ALLOC_WIRED) != 0) { vm_page_lock(m); vm_page_wire(m); vm_page_unlock(m); } if ((allocflags & (VM_ALLOC_NOBUSY | VM_ALLOC_SBUSY)) == 0) vm_page_xbusy(m); if ((allocflags & VM_ALLOC_SBUSY) != 0) vm_page_sbusy(m); return (m); } } m = vm_page_alloc(object, pindex, allocflags); if (m == NULL) { if ((allocflags & VM_ALLOC_NOWAIT) != 0) return (NULL); VM_OBJECT_WUNLOCK(object); VM_WAIT; VM_OBJECT_WLOCK(object); goto retrylookup; } if (allocflags & VM_ALLOC_ZERO && (m->flags & PG_ZERO) == 0) pmap_zero_page(m); return (m); } /* * Mapping function for valid or dirty bits in a page. * * Inputs are required to range within a page. */ vm_page_bits_t vm_page_bits(int base, int size) { int first_bit; int last_bit; KASSERT( base + size <= PAGE_SIZE, ("vm_page_bits: illegal base/size %d/%d", base, size) ); if (size == 0) /* handle degenerate case */ return (0); first_bit = base >> DEV_BSHIFT; last_bit = (base + size - 1) >> DEV_BSHIFT; return (((vm_page_bits_t)2 << last_bit) - ((vm_page_bits_t)1 << first_bit)); } /* * vm_page_set_valid_range: * * Sets portions of a page valid. The arguments are expected * to be DEV_BSIZE aligned but if they aren't the bitmap is inclusive * of any partial chunks touched by the range. The invalid portion of * such chunks will be zeroed. * * (base + size) must be less then or equal to PAGE_SIZE. */ void vm_page_set_valid_range(vm_page_t m, int base, int size) { int endoff, frag; VM_OBJECT_ASSERT_WLOCKED(m->object); if (size == 0) /* handle degenerate case */ return; /* * If the base is not DEV_BSIZE aligned and the valid * bit is clear, we have to zero out a portion of the * first block. */ if ((frag = rounddown2(base, DEV_BSIZE)) != base && (m->valid & (1 << (base >> DEV_BSHIFT))) == 0) pmap_zero_page_area(m, frag, base - frag); /* * If the ending offset is not DEV_BSIZE aligned and the * valid bit is clear, we have to zero out a portion of * the last block. */ endoff = base + size; if ((frag = rounddown2(endoff, DEV_BSIZE)) != endoff && (m->valid & (1 << (endoff >> DEV_BSHIFT))) == 0) pmap_zero_page_area(m, endoff, DEV_BSIZE - (endoff & (DEV_BSIZE - 1))); /* * Assert that no previously invalid block that is now being validated * is already dirty. */ KASSERT((~m->valid & vm_page_bits(base, size) & m->dirty) == 0, ("vm_page_set_valid_range: page %p is dirty", m)); /* * Set valid bits inclusive of any overlap. */ m->valid |= vm_page_bits(base, size); } /* * Clear the given bits from the specified page's dirty field. */ static __inline void vm_page_clear_dirty_mask(vm_page_t m, vm_page_bits_t pagebits) { uintptr_t addr; #if PAGE_SIZE < 16384 int shift; #endif /* * If the object is locked and the page is neither exclusive busy nor * write mapped, then the page's dirty field cannot possibly be * set by a concurrent pmap operation. */ VM_OBJECT_ASSERT_WLOCKED(m->object); if (!vm_page_xbusied(m) && !pmap_page_is_write_mapped(m)) m->dirty &= ~pagebits; else { /* * The pmap layer can call vm_page_dirty() without * holding a distinguished lock. The combination of * the object's lock and an atomic operation suffice * to guarantee consistency of the page dirty field. * * For PAGE_SIZE == 32768 case, compiler already * properly aligns the dirty field, so no forcible * alignment is needed. Only require existence of * atomic_clear_64 when page size is 32768. */ addr = (uintptr_t)&m->dirty; #if PAGE_SIZE == 32768 atomic_clear_64((uint64_t *)addr, pagebits); #elif PAGE_SIZE == 16384 atomic_clear_32((uint32_t *)addr, pagebits); #else /* PAGE_SIZE <= 8192 */ /* * Use a trick to perform a 32-bit atomic on the * containing aligned word, to not depend on the existence * of atomic_clear_{8, 16}. */ shift = addr & (sizeof(uint32_t) - 1); #if BYTE_ORDER == BIG_ENDIAN shift = (sizeof(uint32_t) - sizeof(m->dirty) - shift) * NBBY; #else shift *= NBBY; #endif addr &= ~(sizeof(uint32_t) - 1); atomic_clear_32((uint32_t *)addr, pagebits << shift); #endif /* PAGE_SIZE */ } } /* * vm_page_set_validclean: * * Sets portions of a page valid and clean. The arguments are expected * to be DEV_BSIZE aligned but if they aren't the bitmap is inclusive * of any partial chunks touched by the range. The invalid portion of * such chunks will be zero'd. * * (base + size) must be less then or equal to PAGE_SIZE. */ void vm_page_set_validclean(vm_page_t m, int base, int size) { vm_page_bits_t oldvalid, pagebits; int endoff, frag; VM_OBJECT_ASSERT_WLOCKED(m->object); if (size == 0) /* handle degenerate case */ return; /* * If the base is not DEV_BSIZE aligned and the valid * bit is clear, we have to zero out a portion of the * first block. */ if ((frag = rounddown2(base, DEV_BSIZE)) != base && (m->valid & ((vm_page_bits_t)1 << (base >> DEV_BSHIFT))) == 0) pmap_zero_page_area(m, frag, base - frag); /* * If the ending offset is not DEV_BSIZE aligned and the * valid bit is clear, we have to zero out a portion of * the last block. */ endoff = base + size; if ((frag = rounddown2(endoff, DEV_BSIZE)) != endoff && (m->valid & ((vm_page_bits_t)1 << (endoff >> DEV_BSHIFT))) == 0) pmap_zero_page_area(m, endoff, DEV_BSIZE - (endoff & (DEV_BSIZE - 1))); /* * Set valid, clear dirty bits. If validating the entire * page we can safely clear the pmap modify bit. We also * use this opportunity to clear the VPO_NOSYNC flag. If a process * takes a write fault on a MAP_NOSYNC memory area the flag will * be set again. * * We set valid bits inclusive of any overlap, but we can only * clear dirty bits for DEV_BSIZE chunks that are fully within * the range. */ oldvalid = m->valid; pagebits = vm_page_bits(base, size); m->valid |= pagebits; #if 0 /* NOT YET */ if ((frag = base & (DEV_BSIZE - 1)) != 0) { frag = DEV_BSIZE - frag; base += frag; size -= frag; if (size < 0) size = 0; } pagebits = vm_page_bits(base, size & (DEV_BSIZE - 1)); #endif if (base == 0 && size == PAGE_SIZE) { /* * The page can only be modified within the pmap if it is * mapped, and it can only be mapped if it was previously * fully valid. */ if (oldvalid == VM_PAGE_BITS_ALL) /* * Perform the pmap_clear_modify() first. Otherwise, * a concurrent pmap operation, such as * pmap_protect(), could clear a modification in the * pmap and set the dirty field on the page before * pmap_clear_modify() had begun and after the dirty * field was cleared here. */ pmap_clear_modify(m); m->dirty = 0; m->oflags &= ~VPO_NOSYNC; } else if (oldvalid != VM_PAGE_BITS_ALL) m->dirty &= ~pagebits; else vm_page_clear_dirty_mask(m, pagebits); } void vm_page_clear_dirty(vm_page_t m, int base, int size) { vm_page_clear_dirty_mask(m, vm_page_bits(base, size)); } /* * vm_page_set_invalid: * * Invalidates DEV_BSIZE'd chunks within a page. Both the * valid and dirty bits for the effected areas are cleared. */ void vm_page_set_invalid(vm_page_t m, int base, int size) { vm_page_bits_t bits; vm_object_t object; object = m->object; VM_OBJECT_ASSERT_WLOCKED(object); if (object->type == OBJT_VNODE && base == 0 && IDX_TO_OFF(m->pindex) + size >= object->un_pager.vnp.vnp_size) bits = VM_PAGE_BITS_ALL; else bits = vm_page_bits(base, size); if (object->ref_count != 0 && m->valid == VM_PAGE_BITS_ALL && bits != 0) pmap_remove_all(m); KASSERT((bits == 0 && m->valid == VM_PAGE_BITS_ALL) || !pmap_page_is_mapped(m), ("vm_page_set_invalid: page %p is mapped", m)); m->valid &= ~bits; m->dirty &= ~bits; } /* * vm_page_zero_invalid() * * The kernel assumes that the invalid portions of a page contain * garbage, but such pages can be mapped into memory by user code. * When this occurs, we must zero out the non-valid portions of the * page so user code sees what it expects. * * Pages are most often semi-valid when the end of a file is mapped * into memory and the file's size is not page aligned. */ void vm_page_zero_invalid(vm_page_t m, boolean_t setvalid) { int b; int i; VM_OBJECT_ASSERT_WLOCKED(m->object); /* * Scan the valid bits looking for invalid sections that * must be zeroed. Invalid sub-DEV_BSIZE'd areas ( where the * valid bit may be set ) have already been zeroed by * vm_page_set_validclean(). */ for (b = i = 0; i <= PAGE_SIZE / DEV_BSIZE; ++i) { if (i == (PAGE_SIZE / DEV_BSIZE) || (m->valid & ((vm_page_bits_t)1 << i))) { if (i > b) { pmap_zero_page_area(m, b << DEV_BSHIFT, (i - b) << DEV_BSHIFT); } b = i + 1; } } /* * setvalid is TRUE when we can safely set the zero'd areas * as being valid. We can do this if there are no cache consistancy * issues. e.g. it is ok to do with UFS, but not ok to do with NFS. */ if (setvalid) m->valid = VM_PAGE_BITS_ALL; } /* * vm_page_is_valid: * * Is (partial) page valid? Note that the case where size == 0 * will return FALSE in the degenerate case where the page is * entirely invalid, and TRUE otherwise. */ int vm_page_is_valid(vm_page_t m, int base, int size) { vm_page_bits_t bits; VM_OBJECT_ASSERT_LOCKED(m->object); bits = vm_page_bits(base, size); return (m->valid != 0 && (m->valid & bits) == bits); } /* * vm_page_ps_is_valid: * * Returns TRUE if the entire (super)page is valid and FALSE otherwise. */ boolean_t vm_page_ps_is_valid(vm_page_t m) { int i, npages; VM_OBJECT_ASSERT_LOCKED(m->object); npages = atop(pagesizes[m->psind]); /* * The physically contiguous pages that make up a superpage, i.e., a * page with a page size index ("psind") greater than zero, will * occupy adjacent entries in vm_page_array[]. */ for (i = 0; i < npages; i++) { if (m[i].valid != VM_PAGE_BITS_ALL) return (FALSE); } return (TRUE); } /* * Set the page's dirty bits if the page is modified. */ void vm_page_test_dirty(vm_page_t m) { VM_OBJECT_ASSERT_WLOCKED(m->object); if (m->dirty != VM_PAGE_BITS_ALL && pmap_is_modified(m)) vm_page_dirty(m); } void vm_page_lock_KBI(vm_page_t m, const char *file, int line) { mtx_lock_flags_(vm_page_lockptr(m), 0, file, line); } void vm_page_unlock_KBI(vm_page_t m, const char *file, int line) { mtx_unlock_flags_(vm_page_lockptr(m), 0, file, line); } int vm_page_trylock_KBI(vm_page_t m, const char *file, int line) { return (mtx_trylock_flags_(vm_page_lockptr(m), 0, file, line)); } #if defined(INVARIANTS) || defined(INVARIANT_SUPPORT) void vm_page_assert_locked_KBI(vm_page_t m, const char *file, int line) { vm_page_lock_assert_KBI(m, MA_OWNED, file, line); } void vm_page_lock_assert_KBI(vm_page_t m, int a, const char *file, int line) { mtx_assert_(vm_page_lockptr(m), a, file, line); } #endif #ifdef INVARIANTS void vm_page_object_lock_assert(vm_page_t m) { /* * Certain of the page's fields may only be modified by the * holder of the containing object's lock or the exclusive busy. * holder. Unfortunately, the holder of the write busy is * not recorded, and thus cannot be checked here. */ if (m->object != NULL && !vm_page_xbusied(m)) VM_OBJECT_ASSERT_WLOCKED(m->object); } void vm_page_assert_pga_writeable(vm_page_t m, uint8_t bits) { if ((bits & PGA_WRITEABLE) == 0) return; /* * The PGA_WRITEABLE flag can only be set if the page is * managed, is exclusively busied or the object is locked. * Currently, this flag is only set by pmap_enter(). */ KASSERT((m->oflags & VPO_UNMANAGED) == 0, ("PGA_WRITEABLE on unmanaged page")); if (!vm_page_xbusied(m)) VM_OBJECT_ASSERT_LOCKED(m->object); } #endif #include "opt_ddb.h" #ifdef DDB #include #include DB_SHOW_COMMAND(page, vm_page_print_page_info) { db_printf("vm_cnt.v_free_count: %d\n", vm_cnt.v_free_count); db_printf("vm_cnt.v_inactive_count: %d\n", vm_cnt.v_inactive_count); db_printf("vm_cnt.v_active_count: %d\n", vm_cnt.v_active_count); db_printf("vm_cnt.v_laundry_count: %d\n", vm_cnt.v_laundry_count); db_printf("vm_cnt.v_wire_count: %d\n", vm_cnt.v_wire_count); db_printf("vm_cnt.v_free_reserved: %d\n", vm_cnt.v_free_reserved); db_printf("vm_cnt.v_free_min: %d\n", vm_cnt.v_free_min); db_printf("vm_cnt.v_free_target: %d\n", vm_cnt.v_free_target); db_printf("vm_cnt.v_inactive_target: %d\n", vm_cnt.v_inactive_target); } DB_SHOW_COMMAND(pageq, vm_page_print_pageq_info) { int dom; db_printf("pq_free %d\n", vm_cnt.v_free_count); for (dom = 0; dom < vm_ndomains; dom++) { db_printf( "dom %d page_cnt %d free %d pq_act %d pq_inact %d pq_laund %d pq_unsw %d\n", dom, vm_dom[dom].vmd_page_count, vm_dom[dom].vmd_free_count, vm_dom[dom].vmd_pagequeues[PQ_ACTIVE].pq_cnt, vm_dom[dom].vmd_pagequeues[PQ_INACTIVE].pq_cnt, vm_dom[dom].vmd_pagequeues[PQ_LAUNDRY].pq_cnt, vm_dom[dom].vmd_pagequeues[PQ_UNSWAPPABLE].pq_cnt); } } DB_SHOW_COMMAND(pginfo, vm_page_print_pginfo) { vm_page_t m; boolean_t phys; if (!have_addr) { db_printf("show pginfo addr\n"); return; } phys = strchr(modif, 'p') != NULL; if (phys) m = PHYS_TO_VM_PAGE(addr); else m = (vm_page_t)addr; db_printf( "page %p obj %p pidx 0x%jx phys 0x%jx q %d hold %d wire %d\n" " af 0x%x of 0x%x f 0x%x act %d busy %x valid 0x%x dirty 0x%x\n", m, m->object, (uintmax_t)m->pindex, (uintmax_t)m->phys_addr, m->queue, m->hold_count, m->wire_count, m->aflags, m->oflags, m->flags, m->act_count, m->busy_lock, m->valid, m->dirty); } #endif /* DDB */ Index: projects/clang400-import/sys/vm/vm_page.h =================================================================== --- projects/clang400-import/sys/vm/vm_page.h (revision 311459) +++ projects/clang400-import/sys/vm/vm_page.h (revision 311460) @@ -1,720 +1,722 @@ /*- * Copyright (c) 1991, 1993 * The Regents of the University of California. All rights reserved. * * This code is derived from software contributed to Berkeley by * The Mach Operating System project at Carnegie-Mellon University. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * from: @(#)vm_page.h 8.2 (Berkeley) 12/13/93 * * * Copyright (c) 1987, 1990 Carnegie-Mellon University. * All rights reserved. * * Authors: Avadis Tevanian, Jr., Michael Wayne Young * * Permission to use, copy, modify and distribute this software and * its documentation is hereby granted, provided that both the copyright * notice and this permission notice appear in all copies of the * software, derivative works or modified versions, and any portions * thereof, and that both notices appear in supporting documentation. * * CARNEGIE MELLON ALLOWS FREE USE OF THIS SOFTWARE IN ITS "AS IS" * CONDITION. CARNEGIE MELLON DISCLAIMS ANY LIABILITY OF ANY KIND * FOR ANY DAMAGES WHATSOEVER RESULTING FROM THE USE OF THIS SOFTWARE. * * Carnegie Mellon requests users of this software to return to * * Software Distribution Coordinator or Software.Distribution@CS.CMU.EDU * School of Computer Science * Carnegie Mellon University * Pittsburgh PA 15213-3890 * * any improvements or extensions that they make and grant Carnegie the * rights to redistribute these changes. * * $FreeBSD$ */ /* * Resident memory system definitions. */ #ifndef _VM_PAGE_ #define _VM_PAGE_ #include /* * Management of resident (logical) pages. * * A small structure is kept for each resident * page, indexed by page number. Each structure * is an element of several collections: * * A radix tree used to quickly * perform object/offset lookups * * A list of all pages for a given object, * so they can be quickly deactivated at * time of deallocation. * * An ordered list of pages due for pageout. * * In addition, the structure contains the object * and offset to which this page belongs (for pageout), * and sundry status bits. * * In general, operations on this structure's mutable fields are * synchronized using either one of or a combination of the lock on the * object that the page belongs to (O), the pool lock for the page (P), * or the lock for either the free or paging queue (Q). If a field is * annotated below with two of these locks, then holding either lock is * sufficient for read access, but both locks are required for write * access. * * In contrast, the synchronization of accesses to the page's * dirty field is machine dependent (M). In the * machine-independent layer, the lock on the object that the * page belongs to must be held in order to operate on the field. * However, the pmap layer is permitted to set all bits within * the field without holding that lock. If the underlying * architecture does not support atomic read-modify-write * operations on the field's type, then the machine-independent * layer uses a 32-bit atomic on the aligned 32-bit word that * contains the dirty field. In the machine-independent layer, * the implementation of read-modify-write operations on the * field is encapsulated in vm_page_clear_dirty_mask(). */ #if PAGE_SIZE == 4096 #define VM_PAGE_BITS_ALL 0xffu typedef uint8_t vm_page_bits_t; #elif PAGE_SIZE == 8192 #define VM_PAGE_BITS_ALL 0xffffu typedef uint16_t vm_page_bits_t; #elif PAGE_SIZE == 16384 #define VM_PAGE_BITS_ALL 0xffffffffu typedef uint32_t vm_page_bits_t; #elif PAGE_SIZE == 32768 #define VM_PAGE_BITS_ALL 0xfffffffffffffffflu typedef uint64_t vm_page_bits_t; #endif struct vm_page { union { TAILQ_ENTRY(vm_page) q; /* page queue or free list (Q) */ struct { SLIST_ENTRY(vm_page) ss; /* private slists */ void *pv; } s; struct { u_long p; u_long v; } memguard; } plinks; TAILQ_ENTRY(vm_page) listq; /* pages in same object (O) */ vm_object_t object; /* which object am I in (O,P) */ vm_pindex_t pindex; /* offset into object (O,P) */ vm_paddr_t phys_addr; /* physical address of page */ struct md_page md; /* machine dependent stuff */ u_int wire_count; /* wired down maps refs (P) */ volatile u_int busy_lock; /* busy owners lock */ uint16_t hold_count; /* page hold count (P) */ uint16_t flags; /* page PG_* flags (P) */ uint8_t aflags; /* access is atomic */ uint8_t oflags; /* page VPO_* flags (O) */ uint8_t queue; /* page queue index (P,Q) */ int8_t psind; /* pagesizes[] index (O) */ int8_t segind; uint8_t order; /* index of the buddy queue */ uint8_t pool; u_char act_count; /* page usage count (P) */ /* NOTE that these must support one bit per DEV_BSIZE in a page */ /* so, on normal X86 kernels, they must be at least 8 bits wide */ vm_page_bits_t valid; /* map of valid DEV_BSIZE chunks (O) */ vm_page_bits_t dirty; /* map of dirty DEV_BSIZE chunks (M) */ }; /* * Page flags stored in oflags: * * Access to these page flags is synchronized by the lock on the object * containing the page (O). * * Note: VPO_UNMANAGED (used by OBJT_DEVICE, OBJT_PHYS and OBJT_SG) * indicates that the page is not under PV management but * otherwise should be treated as a normal page. Pages not * under PV management cannot be paged out via the * object/vm_page_t because there is no knowledge of their pte * mappings, and such pages are also not on any PQ queue. * */ #define VPO_UNUSED01 0x01 /* --available-- */ #define VPO_SWAPSLEEP 0x02 /* waiting for swap to finish */ #define VPO_UNMANAGED 0x04 /* no PV management for page */ #define VPO_SWAPINPROG 0x08 /* swap I/O in progress on page */ #define VPO_NOSYNC 0x10 /* do not collect for syncer */ /* * Busy page implementation details. * The algorithm is taken mostly by rwlock(9) and sx(9) locks implementation, * even if the support for owner identity is removed because of size * constraints. Checks on lock recursion are then not possible, while the * lock assertions effectiveness is someway reduced. */ #define VPB_BIT_SHARED 0x01 #define VPB_BIT_EXCLUSIVE 0x02 #define VPB_BIT_WAITERS 0x04 #define VPB_BIT_FLAGMASK \ (VPB_BIT_SHARED | VPB_BIT_EXCLUSIVE | VPB_BIT_WAITERS) #define VPB_SHARERS_SHIFT 3 #define VPB_SHARERS(x) \ (((x) & ~VPB_BIT_FLAGMASK) >> VPB_SHARERS_SHIFT) #define VPB_SHARERS_WORD(x) ((x) << VPB_SHARERS_SHIFT | VPB_BIT_SHARED) #define VPB_ONE_SHARER (1 << VPB_SHARERS_SHIFT) #define VPB_SINGLE_EXCLUSIVER VPB_BIT_EXCLUSIVE #define VPB_UNBUSIED VPB_SHARERS_WORD(0) #define PQ_NONE 255 #define PQ_INACTIVE 0 #define PQ_ACTIVE 1 #define PQ_LAUNDRY 2 #define PQ_UNSWAPPABLE 3 #define PQ_COUNT 4 TAILQ_HEAD(pglist, vm_page); SLIST_HEAD(spglist, vm_page); struct vm_pagequeue { struct mtx pq_mutex; struct pglist pq_pl; int pq_cnt; u_int * const pq_vcnt; const char * const pq_name; } __aligned(CACHE_LINE_SIZE); struct vm_domain { struct vm_pagequeue vmd_pagequeues[PQ_COUNT]; u_int vmd_page_count; u_int vmd_free_count; long vmd_segs; /* bitmask of the segments */ boolean_t vmd_oom; int vmd_oom_seq; int vmd_last_active_scan; struct vm_page vmd_laundry_marker; struct vm_page vmd_marker; /* marker for pagedaemon private use */ struct vm_page vmd_inacthead; /* marker for LRU-defeating insertions */ }; extern struct vm_domain vm_dom[MAXMEMDOM]; #define vm_pagequeue_assert_locked(pq) mtx_assert(&(pq)->pq_mutex, MA_OWNED) #define vm_pagequeue_lock(pq) mtx_lock(&(pq)->pq_mutex) #define vm_pagequeue_lockptr(pq) (&(pq)->pq_mutex) #define vm_pagequeue_unlock(pq) mtx_unlock(&(pq)->pq_mutex) #ifdef _KERNEL +extern vm_page_t bogus_page; + static __inline void vm_pagequeue_cnt_add(struct vm_pagequeue *pq, int addend) { #ifdef notyet vm_pagequeue_assert_locked(pq); #endif pq->pq_cnt += addend; atomic_add_int(pq->pq_vcnt, addend); } #define vm_pagequeue_cnt_inc(pq) vm_pagequeue_cnt_add((pq), 1) #define vm_pagequeue_cnt_dec(pq) vm_pagequeue_cnt_add((pq), -1) #endif /* _KERNEL */ extern struct mtx_padalign vm_page_queue_free_mtx; extern struct mtx_padalign pa_lock[]; #if defined(__arm__) #define PDRSHIFT PDR_SHIFT #elif !defined(PDRSHIFT) #define PDRSHIFT 21 #endif #define pa_index(pa) ((pa) >> PDRSHIFT) #define PA_LOCKPTR(pa) ((struct mtx *)(&pa_lock[pa_index(pa) % PA_LOCK_COUNT])) #define PA_LOCKOBJPTR(pa) ((struct lock_object *)PA_LOCKPTR((pa))) #define PA_LOCK(pa) mtx_lock(PA_LOCKPTR(pa)) #define PA_TRYLOCK(pa) mtx_trylock(PA_LOCKPTR(pa)) #define PA_UNLOCK(pa) mtx_unlock(PA_LOCKPTR(pa)) #define PA_UNLOCK_COND(pa) \ do { \ if ((pa) != 0) { \ PA_UNLOCK((pa)); \ (pa) = 0; \ } \ } while (0) #define PA_LOCK_ASSERT(pa, a) mtx_assert(PA_LOCKPTR(pa), (a)) #ifdef KLD_MODULE #define vm_page_lock(m) vm_page_lock_KBI((m), LOCK_FILE, LOCK_LINE) #define vm_page_unlock(m) vm_page_unlock_KBI((m), LOCK_FILE, LOCK_LINE) #define vm_page_trylock(m) vm_page_trylock_KBI((m), LOCK_FILE, LOCK_LINE) #else /* !KLD_MODULE */ #define vm_page_lockptr(m) (PA_LOCKPTR(VM_PAGE_TO_PHYS((m)))) #define vm_page_lock(m) mtx_lock(vm_page_lockptr((m))) #define vm_page_unlock(m) mtx_unlock(vm_page_lockptr((m))) #define vm_page_trylock(m) mtx_trylock(vm_page_lockptr((m))) #endif #if defined(INVARIANTS) #define vm_page_assert_locked(m) \ vm_page_assert_locked_KBI((m), __FILE__, __LINE__) #define vm_page_lock_assert(m, a) \ vm_page_lock_assert_KBI((m), (a), __FILE__, __LINE__) #else #define vm_page_assert_locked(m) #define vm_page_lock_assert(m, a) #endif /* * The vm_page's aflags are updated using atomic operations. To set or clear * these flags, the functions vm_page_aflag_set() and vm_page_aflag_clear() * must be used. Neither these flags nor these functions are part of the KBI. * * PGA_REFERENCED may be cleared only if the page is locked. It is set by * both the MI and MD VM layers. However, kernel loadable modules should not * directly set this flag. They should call vm_page_reference() instead. * * PGA_WRITEABLE is set exclusively on managed pages by pmap_enter(). * When it does so, the object must be locked, or the page must be * exclusive busied. The MI VM layer must never access this flag * directly. Instead, it should call pmap_page_is_write_mapped(). * * PGA_EXECUTABLE may be set by pmap routines, and indicates that a page has * at least one executable mapping. It is not consumed by the MI VM layer. */ #define PGA_WRITEABLE 0x01 /* page may be mapped writeable */ #define PGA_REFERENCED 0x02 /* page has been referenced */ #define PGA_EXECUTABLE 0x04 /* page may be mapped executable */ /* * Page flags. If changed at any other time than page allocation or * freeing, the modification must be protected by the vm_page lock. */ #define PG_FICTITIOUS 0x0004 /* physical page doesn't exist */ #define PG_ZERO 0x0008 /* page is zeroed */ #define PG_MARKER 0x0010 /* special queue marker page */ #define PG_NODUMP 0x0080 /* don't include this page in a dump */ #define PG_UNHOLDFREE 0x0100 /* delayed free of a held page */ /* * Misc constants. */ #define ACT_DECLINE 1 #define ACT_ADVANCE 3 #define ACT_INIT 5 #define ACT_MAX 64 #ifdef _KERNEL #include #include /* * Each pageable resident page falls into one of five lists: * * free * Available for allocation now. * * inactive * Low activity, candidates for reclamation. * This list is approximately LRU ordered. * * laundry * This is the list of pages that should be * paged out next. * * unswappable * Dirty anonymous pages that cannot be paged * out because no swap device is configured. * * active * Pages that are "active", i.e., they have been * recently referenced. * */ extern int vm_page_zero_count; extern vm_page_t vm_page_array; /* First resident page in table */ extern long vm_page_array_size; /* number of vm_page_t's */ extern long first_page; /* first physical page number */ #define VM_PAGE_TO_PHYS(entry) ((entry)->phys_addr) /* * PHYS_TO_VM_PAGE() returns the vm_page_t object that represents a memory * page to which the given physical address belongs. The correct vm_page_t * object is returned for addresses that are not page-aligned. */ vm_page_t PHYS_TO_VM_PAGE(vm_paddr_t pa); /* * Page allocation parameters for vm_page for the functions * vm_page_alloc(), vm_page_grab(), vm_page_alloc_contig() and * vm_page_alloc_freelist(). Some functions support only a subset * of the flags, and ignore others, see the flags legend. * * Bits 0 - 1 define class. * Bits 2 - 15 dedicated for flags. * Legend: * (a) - vm_page_alloc() supports the flag. * (c) - vm_page_alloc_contig() supports the flag. * (f) - vm_page_alloc_freelist() supports the flag. * (g) - vm_page_grab() supports the flag. * Bits above 15 define the count of additional pages that the caller * intends to allocate. */ #define VM_ALLOC_NORMAL 0 #define VM_ALLOC_INTERRUPT 1 #define VM_ALLOC_SYSTEM 2 #define VM_ALLOC_CLASS_MASK 3 #define VM_ALLOC_WIRED 0x0020 /* (acfg) Allocate non pageable page */ #define VM_ALLOC_ZERO 0x0040 /* (acfg) Try to obtain a zeroed page */ #define VM_ALLOC_NOOBJ 0x0100 /* (acg) No associated object */ #define VM_ALLOC_NOBUSY 0x0200 /* (acg) Do not busy the page */ #define VM_ALLOC_IGN_SBUSY 0x1000 /* (g) Ignore shared busy flag */ #define VM_ALLOC_NODUMP 0x2000 /* (ag) don't include in dump */ #define VM_ALLOC_SBUSY 0x4000 /* (acg) Shared busy the page */ #define VM_ALLOC_NOWAIT 0x8000 /* (g) Do not sleep, return NULL */ #define VM_ALLOC_COUNT_SHIFT 16 #define VM_ALLOC_COUNT(count) ((count) << VM_ALLOC_COUNT_SHIFT) #ifdef M_NOWAIT static inline int malloc2vm_flags(int malloc_flags) { int pflags; KASSERT((malloc_flags & M_USE_RESERVE) == 0 || (malloc_flags & M_NOWAIT) != 0, ("M_USE_RESERVE requires M_NOWAIT")); pflags = (malloc_flags & M_USE_RESERVE) != 0 ? VM_ALLOC_INTERRUPT : VM_ALLOC_SYSTEM; if ((malloc_flags & M_ZERO) != 0) pflags |= VM_ALLOC_ZERO; if ((malloc_flags & M_NODUMP) != 0) pflags |= VM_ALLOC_NODUMP; return (pflags); } #endif void vm_page_busy_downgrade(vm_page_t m); void vm_page_busy_sleep(vm_page_t m, const char *msg, bool nonshared); void vm_page_flash(vm_page_t m); void vm_page_hold(vm_page_t mem); void vm_page_unhold(vm_page_t mem); void vm_page_free(vm_page_t m); void vm_page_free_zero(vm_page_t m); void vm_page_activate (vm_page_t); void vm_page_advise(vm_page_t m, int advice); vm_page_t vm_page_alloc (vm_object_t, vm_pindex_t, int); vm_page_t vm_page_alloc_contig(vm_object_t object, vm_pindex_t pindex, int req, u_long npages, vm_paddr_t low, vm_paddr_t high, u_long alignment, vm_paddr_t boundary, vm_memattr_t memattr); vm_page_t vm_page_alloc_freelist(int, int); vm_page_t vm_page_grab (vm_object_t, vm_pindex_t, int); int vm_page_try_to_free (vm_page_t); void vm_page_deactivate (vm_page_t); void vm_page_deactivate_noreuse(vm_page_t); void vm_page_dequeue(vm_page_t m); void vm_page_dequeue_locked(vm_page_t m); vm_page_t vm_page_find_least(vm_object_t, vm_pindex_t); vm_page_t vm_page_getfake(vm_paddr_t paddr, vm_memattr_t memattr); void vm_page_initfake(vm_page_t m, vm_paddr_t paddr, vm_memattr_t memattr); int vm_page_insert (vm_page_t, vm_object_t, vm_pindex_t); void vm_page_launder(vm_page_t m); vm_page_t vm_page_lookup (vm_object_t, vm_pindex_t); vm_page_t vm_page_next(vm_page_t m); int vm_page_pa_tryrelock(pmap_t, vm_paddr_t, vm_paddr_t *); struct vm_pagequeue *vm_page_pagequeue(vm_page_t m); vm_page_t vm_page_prev(vm_page_t m); boolean_t vm_page_ps_is_valid(vm_page_t m); void vm_page_putfake(vm_page_t m); void vm_page_readahead_finish(vm_page_t m); bool vm_page_reclaim_contig(int req, u_long npages, vm_paddr_t low, vm_paddr_t high, u_long alignment, vm_paddr_t boundary); void vm_page_reference(vm_page_t m); void vm_page_remove (vm_page_t); int vm_page_rename (vm_page_t, vm_object_t, vm_pindex_t); vm_page_t vm_page_replace(vm_page_t mnew, vm_object_t object, vm_pindex_t pindex); void vm_page_requeue(vm_page_t m); void vm_page_requeue_locked(vm_page_t m); int vm_page_sbusied(vm_page_t m); vm_page_t vm_page_scan_contig(u_long npages, vm_page_t m_start, vm_page_t m_end, u_long alignment, vm_paddr_t boundary, int options); void vm_page_set_valid_range(vm_page_t m, int base, int size); int vm_page_sleep_if_busy(vm_page_t m, const char *msg); vm_offset_t vm_page_startup(vm_offset_t vaddr); void vm_page_sunbusy(vm_page_t m); int vm_page_trysbusy(vm_page_t m); void vm_page_unhold_pages(vm_page_t *ma, int count); void vm_page_unswappable(vm_page_t m); boolean_t vm_page_unwire(vm_page_t m, uint8_t queue); void vm_page_updatefake(vm_page_t m, vm_paddr_t paddr, vm_memattr_t memattr); void vm_page_wire (vm_page_t); void vm_page_xunbusy_hard(vm_page_t m); void vm_page_xunbusy_maybelocked(vm_page_t m); void vm_page_set_validclean (vm_page_t, int, int); void vm_page_clear_dirty (vm_page_t, int, int); void vm_page_set_invalid (vm_page_t, int, int); int vm_page_is_valid (vm_page_t, int, int); void vm_page_test_dirty (vm_page_t); vm_page_bits_t vm_page_bits(int base, int size); void vm_page_zero_invalid(vm_page_t m, boolean_t setvalid); void vm_page_free_toq(vm_page_t m); void vm_page_dirty_KBI(vm_page_t m); void vm_page_lock_KBI(vm_page_t m, const char *file, int line); void vm_page_unlock_KBI(vm_page_t m, const char *file, int line); int vm_page_trylock_KBI(vm_page_t m, const char *file, int line); #if defined(INVARIANTS) || defined(INVARIANT_SUPPORT) void vm_page_assert_locked_KBI(vm_page_t m, const char *file, int line); void vm_page_lock_assert_KBI(vm_page_t m, int a, const char *file, int line); #endif #define vm_page_assert_sbusied(m) \ KASSERT(vm_page_sbusied(m), \ ("vm_page_assert_sbusied: page %p not shared busy @ %s:%d", \ (m), __FILE__, __LINE__)) #define vm_page_assert_unbusied(m) \ KASSERT(!vm_page_busied(m), \ ("vm_page_assert_unbusied: page %p busy @ %s:%d", \ (m), __FILE__, __LINE__)) #define vm_page_assert_xbusied(m) \ KASSERT(vm_page_xbusied(m), \ ("vm_page_assert_xbusied: page %p not exclusive busy @ %s:%d", \ (m), __FILE__, __LINE__)) #define vm_page_busied(m) \ ((m)->busy_lock != VPB_UNBUSIED) #define vm_page_sbusy(m) do { \ if (!vm_page_trysbusy(m)) \ panic("%s: page %p failed shared busying", __func__, \ (m)); \ } while (0) #define vm_page_tryxbusy(m) \ (atomic_cmpset_acq_int(&(m)->busy_lock, VPB_UNBUSIED, \ VPB_SINGLE_EXCLUSIVER)) #define vm_page_xbusied(m) \ (((m)->busy_lock & VPB_SINGLE_EXCLUSIVER) != 0) #define vm_page_xbusy(m) do { \ if (!vm_page_tryxbusy(m)) \ panic("%s: page %p failed exclusive busying", __func__, \ (m)); \ } while (0) /* Note: page m's lock must not be owned by the caller. */ #define vm_page_xunbusy(m) do { \ if (!atomic_cmpset_rel_int(&(m)->busy_lock, \ VPB_SINGLE_EXCLUSIVER, VPB_UNBUSIED)) \ vm_page_xunbusy_hard(m); \ } while (0) #ifdef INVARIANTS void vm_page_object_lock_assert(vm_page_t m); #define VM_PAGE_OBJECT_LOCK_ASSERT(m) vm_page_object_lock_assert(m) void vm_page_assert_pga_writeable(vm_page_t m, uint8_t bits); #define VM_PAGE_ASSERT_PGA_WRITEABLE(m, bits) \ vm_page_assert_pga_writeable(m, bits) #else #define VM_PAGE_OBJECT_LOCK_ASSERT(m) (void)0 #define VM_PAGE_ASSERT_PGA_WRITEABLE(m, bits) (void)0 #endif /* * We want to use atomic updates for the aflags field, which is 8 bits wide. * However, not all architectures support atomic operations on 8-bit * destinations. In order that we can easily use a 32-bit operation, we * require that the aflags field be 32-bit aligned. */ CTASSERT(offsetof(struct vm_page, aflags) % sizeof(uint32_t) == 0); /* * Clear the given bits in the specified page. */ static inline void vm_page_aflag_clear(vm_page_t m, uint8_t bits) { uint32_t *addr, val; /* * The PGA_REFERENCED flag can only be cleared if the page is locked. */ if ((bits & PGA_REFERENCED) != 0) vm_page_assert_locked(m); /* * Access the whole 32-bit word containing the aflags field with an * atomic update. Parallel non-atomic updates to the other fields * within this word are handled properly by the atomic update. */ addr = (void *)&m->aflags; KASSERT(((uintptr_t)addr & (sizeof(uint32_t) - 1)) == 0, ("vm_page_aflag_clear: aflags is misaligned")); val = bits; #if BYTE_ORDER == BIG_ENDIAN val <<= 24; #endif atomic_clear_32(addr, val); } /* * Set the given bits in the specified page. */ static inline void vm_page_aflag_set(vm_page_t m, uint8_t bits) { uint32_t *addr, val; VM_PAGE_ASSERT_PGA_WRITEABLE(m, bits); /* * Access the whole 32-bit word containing the aflags field with an * atomic update. Parallel non-atomic updates to the other fields * within this word are handled properly by the atomic update. */ addr = (void *)&m->aflags; KASSERT(((uintptr_t)addr & (sizeof(uint32_t) - 1)) == 0, ("vm_page_aflag_set: aflags is misaligned")); val = bits; #if BYTE_ORDER == BIG_ENDIAN val <<= 24; #endif atomic_set_32(addr, val); } /* * vm_page_dirty: * * Set all bits in the page's dirty field. * * The object containing the specified page must be locked if the * call is made from the machine-independent layer. * * See vm_page_clear_dirty_mask(). */ static __inline void vm_page_dirty(vm_page_t m) { /* Use vm_page_dirty_KBI() under INVARIANTS to save memory. */ #if defined(KLD_MODULE) || defined(INVARIANTS) vm_page_dirty_KBI(m); #else m->dirty = VM_PAGE_BITS_ALL; #endif } /* * vm_page_remque: * * If the given page is in a page queue, then remove it from that page * queue. * * The page must be locked. */ static inline void vm_page_remque(vm_page_t m) { if (m->queue != PQ_NONE) vm_page_dequeue(m); } /* * vm_page_undirty: * * Set page to not be dirty. Note: does not clear pmap modify bits */ static __inline void vm_page_undirty(vm_page_t m) { VM_PAGE_OBJECT_LOCK_ASSERT(m); m->dirty = 0; } static inline void vm_page_replace_checked(vm_page_t mnew, vm_object_t object, vm_pindex_t pindex, vm_page_t mold) { vm_page_t mret; mret = vm_page_replace(mnew, object, pindex); KASSERT(mret == mold, ("invalid page replacement, mold=%p, mret=%p", mold, mret)); /* Unused if !INVARIANTS. */ (void)mold; (void)mret; } static inline bool vm_page_active(vm_page_t m) { return (m->queue == PQ_ACTIVE); } static inline bool vm_page_inactive(vm_page_t m) { return (m->queue == PQ_INACTIVE); } static inline bool vm_page_in_laundry(vm_page_t m) { return (m->queue == PQ_LAUNDRY || m->queue == PQ_UNSWAPPABLE); } #endif /* _KERNEL */ #endif /* !_VM_PAGE_ */ Index: projects/clang400-import/sys/vm/vm_pager.c =================================================================== --- projects/clang400-import/sys/vm/vm_pager.c (revision 311459) +++ projects/clang400-import/sys/vm/vm_pager.c (revision 311460) @@ -1,552 +1,550 @@ /*- * Copyright (c) 1991, 1993 * The Regents of the University of California. All rights reserved. * * This code is derived from software contributed to Berkeley by * The Mach Operating System project at Carnegie-Mellon University. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * from: @(#)vm_pager.c 8.6 (Berkeley) 1/12/94 * * * Copyright (c) 1987, 1990 Carnegie-Mellon University. * All rights reserved. * * Authors: Avadis Tevanian, Jr., Michael Wayne Young * * Permission to use, copy, modify and distribute this software and * its documentation is hereby granted, provided that both the copyright * notice and this permission notice appear in all copies of the * software, derivative works or modified versions, and any portions * thereof, and that both notices appear in supporting documentation. * * CARNEGIE MELLON ALLOWS FREE USE OF THIS SOFTWARE IN ITS "AS IS" * CONDITION. CARNEGIE MELLON DISCLAIMS ANY LIABILITY OF ANY KIND * FOR ANY DAMAGES WHATSOEVER RESULTING FROM THE USE OF THIS SOFTWARE. * * Carnegie Mellon requests users of this software to return to * * Software Distribution Coordinator or Software.Distribution@CS.CMU.EDU * School of Computer Science * Carnegie Mellon University * Pittsburgh PA 15213-3890 * * any improvements or extensions that they make and grant Carnegie the * rights to redistribute these changes. */ /* * Paging space routine stubs. Emulates a matchmaker-like interface * for builtin pagers. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include -extern vm_page_t bogus_page; - int cluster_pbuf_freecnt = -1; /* unlimited to begin with */ struct buf *swbuf; static int dead_pager_getpages(vm_object_t, vm_page_t *, int, int *, int *); static vm_object_t dead_pager_alloc(void *, vm_ooffset_t, vm_prot_t, vm_ooffset_t, struct ucred *); static void dead_pager_putpages(vm_object_t, vm_page_t *, int, int, int *); static boolean_t dead_pager_haspage(vm_object_t, vm_pindex_t, int *, int *); static void dead_pager_dealloc(vm_object_t); static int dead_pager_getpages(vm_object_t obj, vm_page_t *ma, int count, int *rbehind, int *rahead) { return (VM_PAGER_FAIL); } static vm_object_t dead_pager_alloc(void *handle, vm_ooffset_t size, vm_prot_t prot, vm_ooffset_t off, struct ucred *cred) { return (NULL); } static void dead_pager_putpages(vm_object_t object, vm_page_t *m, int count, int flags, int *rtvals) { int i; for (i = 0; i < count; i++) rtvals[i] = VM_PAGER_AGAIN; } static int dead_pager_haspage(vm_object_t object, vm_pindex_t pindex, int *prev, int *next) { if (prev != NULL) *prev = 0; if (next != NULL) *next = 0; return (FALSE); } static void dead_pager_dealloc(vm_object_t object) { } static struct pagerops deadpagerops = { .pgo_alloc = dead_pager_alloc, .pgo_dealloc = dead_pager_dealloc, .pgo_getpages = dead_pager_getpages, .pgo_putpages = dead_pager_putpages, .pgo_haspage = dead_pager_haspage, }; struct pagerops *pagertab[] = { &defaultpagerops, /* OBJT_DEFAULT */ &swappagerops, /* OBJT_SWAP */ &vnodepagerops, /* OBJT_VNODE */ &devicepagerops, /* OBJT_DEVICE */ &physpagerops, /* OBJT_PHYS */ &deadpagerops, /* OBJT_DEAD */ &sgpagerops, /* OBJT_SG */ &mgtdevicepagerops, /* OBJT_MGTDEVICE */ }; /* * Kernel address space for mapping pages. * Used by pagers where KVAs are needed for IO. * * XXX needs to be large enough to support the number of pending async * cleaning requests (NPENDINGIO == 64) * the maximum swap cluster size * (MAXPHYS == 64k) if you want to get the most efficiency. */ struct mtx_padalign pbuf_mtx; static TAILQ_HEAD(swqueue, buf) bswlist; static int bswneeded; vm_offset_t swapbkva; /* swap buffers kva */ void vm_pager_init(void) { struct pagerops **pgops; TAILQ_INIT(&bswlist); /* * Initialize known pagers */ for (pgops = pagertab; pgops < &pagertab[nitems(pagertab)]; pgops++) if ((*pgops)->pgo_init != NULL) (*(*pgops)->pgo_init)(); } void vm_pager_bufferinit(void) { struct buf *bp; int i; mtx_init(&pbuf_mtx, "pbuf mutex", NULL, MTX_DEF); bp = swbuf; /* * Now set up swap and physical I/O buffer headers. */ for (i = 0; i < nswbuf; i++, bp++) { TAILQ_INSERT_HEAD(&bswlist, bp, b_freelist); BUF_LOCKINIT(bp); LIST_INIT(&bp->b_dep); bp->b_rcred = bp->b_wcred = NOCRED; bp->b_xflags = 0; } cluster_pbuf_freecnt = nswbuf / 2; vnode_pbuf_freecnt = nswbuf / 2 + 1; vnode_async_pbuf_freecnt = nswbuf / 2; } /* * Allocate an instance of a pager of the given type. * Size, protection and offset parameters are passed in for pagers that * need to perform page-level validation (e.g. the device pager). */ vm_object_t vm_pager_allocate(objtype_t type, void *handle, vm_ooffset_t size, vm_prot_t prot, vm_ooffset_t off, struct ucred *cred) { vm_object_t ret; struct pagerops *ops; ops = pagertab[type]; if (ops) ret = (*ops->pgo_alloc)(handle, size, prot, off, cred); else ret = NULL; return (ret); } /* * The object must be locked. */ void vm_pager_deallocate(vm_object_t object) { VM_OBJECT_ASSERT_WLOCKED(object); (*pagertab[object->type]->pgo_dealloc) (object); } static void vm_pager_assert_in(vm_object_t object, vm_page_t *m, int count) { #ifdef INVARIANTS VM_OBJECT_ASSERT_WLOCKED(object); KASSERT(count > 0, ("%s: 0 count", __func__)); /* * All pages must be busied, not mapped, not fully valid, * not dirty and belong to the proper object. */ for (int i = 0 ; i < count; i++) { if (m[i] == bogus_page) continue; vm_page_assert_xbusied(m[i]); KASSERT(!pmap_page_is_mapped(m[i]), ("%s: page %p is mapped", __func__, m[i])); KASSERT(m[i]->valid != VM_PAGE_BITS_ALL, ("%s: request for a valid page %p", __func__, m[i])); KASSERT(m[i]->dirty == 0, ("%s: page %p is dirty", __func__, m[i])); KASSERT(m[i]->object == object, ("%s: wrong object %p/%p", __func__, object, m[i]->object)); } #endif } /* * Page in the pages for the object using its associated pager. * The requested page must be fully valid on successful return. */ int vm_pager_get_pages(vm_object_t object, vm_page_t *m, int count, int *rbehind, int *rahead) { #ifdef INVARIANTS vm_pindex_t pindex = m[0]->pindex; #endif int r; vm_pager_assert_in(object, m, count); r = (*pagertab[object->type]->pgo_getpages)(object, m, count, rbehind, rahead); if (r != VM_PAGER_OK) return (r); for (int i = 0; i < count; i++) { /* * If pager has replaced a page, assert that it had * updated the array. */ KASSERT(m[i] == vm_page_lookup(object, pindex++), ("%s: mismatch page %p pindex %ju", __func__, m[i], (uintmax_t )pindex - 1)); /* * Zero out partially filled data. */ if (m[i]->valid != VM_PAGE_BITS_ALL) vm_page_zero_invalid(m[i], TRUE); } return (VM_PAGER_OK); } int vm_pager_get_pages_async(vm_object_t object, vm_page_t *m, int count, int *rbehind, int *rahead, pgo_getpages_iodone_t iodone, void *arg) { vm_pager_assert_in(object, m, count); return ((*pagertab[object->type]->pgo_getpages_async)(object, m, count, rbehind, rahead, iodone, arg)); } /* * vm_pager_put_pages() - inline, see vm/vm_pager.h * vm_pager_has_page() - inline, see vm/vm_pager.h */ /* * Search the specified pager object list for an object with the * specified handle. If an object with the specified handle is found, * increase its reference count and return it. Otherwise, return NULL. * * The pager object list must be locked. */ vm_object_t vm_pager_object_lookup(struct pagerlst *pg_list, void *handle) { vm_object_t object; TAILQ_FOREACH(object, pg_list, pager_object_list) { if (object->handle == handle) { VM_OBJECT_WLOCK(object); if ((object->flags & OBJ_DEAD) == 0) { vm_object_reference_locked(object); VM_OBJECT_WUNLOCK(object); break; } VM_OBJECT_WUNLOCK(object); } } return (object); } /* * initialize a physical buffer */ /* * XXX This probably belongs in vfs_bio.c */ static void initpbuf(struct buf *bp) { KASSERT(bp->b_bufobj == NULL, ("initpbuf with bufobj")); KASSERT(bp->b_vp == NULL, ("initpbuf with vp")); bp->b_rcred = NOCRED; bp->b_wcred = NOCRED; bp->b_qindex = 0; /* On no queue (QUEUE_NONE) */ bp->b_kvabase = (caddr_t)(MAXPHYS * (bp - swbuf)) + swapbkva; bp->b_data = bp->b_kvabase; bp->b_kvasize = MAXPHYS; bp->b_flags = 0; bp->b_xflags = 0; bp->b_ioflags = 0; bp->b_iodone = NULL; bp->b_error = 0; BUF_LOCK(bp, LK_EXCLUSIVE, NULL); buf_track(bp, __func__); } /* * allocate a physical buffer * * There are a limited number (nswbuf) of physical buffers. We need * to make sure that no single subsystem is able to hog all of them, * so each subsystem implements a counter which is typically initialized * to 1/2 nswbuf. getpbuf() decrements this counter in allocation and * increments it on release, and blocks if the counter hits zero. A * subsystem may initialize the counter to -1 to disable the feature, * but it must still be sure to match up all uses of getpbuf() with * relpbuf() using the same variable. * * NOTE: pfreecnt can be NULL, but this 'feature' will be removed * relatively soon when the rest of the subsystems get smart about it. XXX */ struct buf * getpbuf(int *pfreecnt) { struct buf *bp; mtx_lock(&pbuf_mtx); for (;;) { if (pfreecnt != NULL) { while (*pfreecnt == 0) { msleep(pfreecnt, &pbuf_mtx, PVM, "wswbuf0", 0); } } /* get a bp from the swap buffer header pool */ if ((bp = TAILQ_FIRST(&bswlist)) != NULL) break; bswneeded = 1; msleep(&bswneeded, &pbuf_mtx, PVM, "wswbuf1", 0); /* loop in case someone else grabbed one */ } TAILQ_REMOVE(&bswlist, bp, b_freelist); if (pfreecnt) --*pfreecnt; mtx_unlock(&pbuf_mtx); initpbuf(bp); return (bp); } /* * allocate a physical buffer, if one is available. * * Note that there is no NULL hack here - all subsystems using this * call understand how to use pfreecnt. */ struct buf * trypbuf(int *pfreecnt) { struct buf *bp; mtx_lock(&pbuf_mtx); if (*pfreecnt == 0 || (bp = TAILQ_FIRST(&bswlist)) == NULL) { mtx_unlock(&pbuf_mtx); return NULL; } TAILQ_REMOVE(&bswlist, bp, b_freelist); --*pfreecnt; mtx_unlock(&pbuf_mtx); initpbuf(bp); return (bp); } /* * release a physical buffer * * NOTE: pfreecnt can be NULL, but this 'feature' will be removed * relatively soon when the rest of the subsystems get smart about it. XXX */ void relpbuf(struct buf *bp, int *pfreecnt) { if (bp->b_rcred != NOCRED) { crfree(bp->b_rcred); bp->b_rcred = NOCRED; } if (bp->b_wcred != NOCRED) { crfree(bp->b_wcred); bp->b_wcred = NOCRED; } KASSERT(bp->b_vp == NULL, ("relpbuf with vp")); KASSERT(bp->b_bufobj == NULL, ("relpbuf with bufobj")); buf_track(bp, __func__); BUF_UNLOCK(bp); mtx_lock(&pbuf_mtx); TAILQ_INSERT_HEAD(&bswlist, bp, b_freelist); if (bswneeded) { bswneeded = 0; wakeup(&bswneeded); } if (pfreecnt) { if (++*pfreecnt == 1) wakeup(pfreecnt); } mtx_unlock(&pbuf_mtx); } /* * Associate a p-buffer with a vnode. * * Also sets B_PAGING flag to indicate that vnode is not fully associated * with the buffer. i.e. the bp has not been linked into the vnode or * ref-counted. */ void pbgetvp(struct vnode *vp, struct buf *bp) { KASSERT(bp->b_vp == NULL, ("pbgetvp: not free")); KASSERT(bp->b_bufobj == NULL, ("pbgetvp: not free (bufobj)")); bp->b_vp = vp; bp->b_flags |= B_PAGING; bp->b_bufobj = &vp->v_bufobj; } /* * Associate a p-buffer with a vnode. * * Also sets B_PAGING flag to indicate that vnode is not fully associated * with the buffer. i.e. the bp has not been linked into the vnode or * ref-counted. */ void pbgetbo(struct bufobj *bo, struct buf *bp) { KASSERT(bp->b_vp == NULL, ("pbgetbo: not free (vnode)")); KASSERT(bp->b_bufobj == NULL, ("pbgetbo: not free (bufobj)")); bp->b_flags |= B_PAGING; bp->b_bufobj = bo; } /* * Disassociate a p-buffer from a vnode. */ void pbrelvp(struct buf *bp) { KASSERT(bp->b_vp != NULL, ("pbrelvp: NULL")); KASSERT(bp->b_bufobj != NULL, ("pbrelvp: NULL bufobj")); KASSERT((bp->b_xflags & (BX_VNDIRTY | BX_VNCLEAN)) == 0, ("pbrelvp: pager buf on vnode list.")); bp->b_vp = NULL; bp->b_bufobj = NULL; bp->b_flags &= ~B_PAGING; } /* * Disassociate a p-buffer from a bufobj. */ void pbrelbo(struct buf *bp) { KASSERT(bp->b_vp == NULL, ("pbrelbo: vnode")); KASSERT(bp->b_bufobj != NULL, ("pbrelbo: NULL bufobj")); KASSERT((bp->b_xflags & (BX_VNDIRTY | BX_VNCLEAN)) == 0, ("pbrelbo: pager buf on vnode list.")); bp->b_bufobj = NULL; bp->b_flags &= ~B_PAGING; } Index: projects/clang400-import/sys/vm/vnode_pager.c =================================================================== --- projects/clang400-import/sys/vm/vnode_pager.c (revision 311459) +++ projects/clang400-import/sys/vm/vnode_pager.c (revision 311460) @@ -1,1414 +1,1418 @@ /*- * Copyright (c) 1990 University of Utah. * Copyright (c) 1991 The Regents of the University of California. * All rights reserved. * Copyright (c) 1993, 1994 John S. Dyson * Copyright (c) 1995, David Greenman * * This code is derived from software contributed to Berkeley by * the Systems Programming Group of the University of Utah Computer * Science Department. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. All advertising materials mentioning features or use of this software * must display the following acknowledgement: * This product includes software developed by the University of * California, Berkeley and its contributors. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * from: @(#)vnode_pager.c 7.5 (Berkeley) 4/20/91 */ /* * Page to/from files (vnodes). */ /* * TODO: * Implement VOP_GETPAGES/PUTPAGES interface for filesystems. Will * greatly re-simplify the vnode_pager. */ #include __FBSDID("$FreeBSD$"); #include "opt_vm.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include static int vnode_pager_addr(struct vnode *vp, vm_ooffset_t address, daddr_t *rtaddress, int *run); static int vnode_pager_input_smlfs(vm_object_t object, vm_page_t m); static int vnode_pager_input_old(vm_object_t object, vm_page_t m); static void vnode_pager_dealloc(vm_object_t); static int vnode_pager_getpages(vm_object_t, vm_page_t *, int, int *, int *); static int vnode_pager_getpages_async(vm_object_t, vm_page_t *, int, int *, int *, vop_getpages_iodone_t, void *); static void vnode_pager_putpages(vm_object_t, vm_page_t *, int, int, int *); static boolean_t vnode_pager_haspage(vm_object_t, vm_pindex_t, int *, int *); static vm_object_t vnode_pager_alloc(void *, vm_ooffset_t, vm_prot_t, vm_ooffset_t, struct ucred *cred); static int vnode_pager_generic_getpages_done(struct buf *); static void vnode_pager_generic_getpages_done_async(struct buf *); struct pagerops vnodepagerops = { .pgo_alloc = vnode_pager_alloc, .pgo_dealloc = vnode_pager_dealloc, .pgo_getpages = vnode_pager_getpages, .pgo_getpages_async = vnode_pager_getpages_async, .pgo_putpages = vnode_pager_putpages, .pgo_haspage = vnode_pager_haspage, }; int vnode_pbuf_freecnt; int vnode_async_pbuf_freecnt; /* Create the VM system backing object for this vnode */ int vnode_create_vobject(struct vnode *vp, off_t isize, struct thread *td) { vm_object_t object; vm_ooffset_t size = isize; struct vattr va; if (!vn_isdisk(vp, NULL) && vn_canvmio(vp) == FALSE) return (0); while ((object = vp->v_object) != NULL) { VM_OBJECT_WLOCK(object); if (!(object->flags & OBJ_DEAD)) { VM_OBJECT_WUNLOCK(object); return (0); } VOP_UNLOCK(vp, 0); vm_object_set_flag(object, OBJ_DISCONNECTWNT); VM_OBJECT_SLEEP(object, object, PDROP | PVM, "vodead", 0); vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); } if (size == 0) { if (vn_isdisk(vp, NULL)) { size = IDX_TO_OFF(INT_MAX); } else { if (VOP_GETATTR(vp, &va, td->td_ucred)) return (0); size = va.va_size; } } object = vnode_pager_alloc(vp, size, 0, 0, td->td_ucred); /* * Dereference the reference we just created. This assumes * that the object is associated with the vp. */ VM_OBJECT_WLOCK(object); object->ref_count--; VM_OBJECT_WUNLOCK(object); vrele(vp); KASSERT(vp->v_object != NULL, ("vnode_create_vobject: NULL object")); return (0); } void vnode_destroy_vobject(struct vnode *vp) { struct vm_object *obj; obj = vp->v_object; if (obj == NULL) return; ASSERT_VOP_ELOCKED(vp, "vnode_destroy_vobject"); VM_OBJECT_WLOCK(obj); umtx_shm_object_terminated(obj); if (obj->ref_count == 0) { /* * don't double-terminate the object */ if ((obj->flags & OBJ_DEAD) == 0) { vm_object_terminate(obj); } else { /* * Waiters were already handled during object * termination. The exclusive vnode lock hopefully * prevented new waiters from referencing the dying * object. */ KASSERT((obj->flags & OBJ_DISCONNECTWNT) == 0, ("OBJ_DISCONNECTWNT set obj %p flags %x", obj, obj->flags)); vp->v_object = NULL; VM_OBJECT_WUNLOCK(obj); } } else { /* * Woe to the process that tries to page now :-). */ vm_pager_deallocate(obj); VM_OBJECT_WUNLOCK(obj); } KASSERT(vp->v_object == NULL, ("vp %p obj %p", vp, vp->v_object)); } /* * Allocate (or lookup) pager for a vnode. * Handle is a vnode pointer. * * MPSAFE */ vm_object_t vnode_pager_alloc(void *handle, vm_ooffset_t size, vm_prot_t prot, vm_ooffset_t offset, struct ucred *cred) { vm_object_t object; struct vnode *vp; /* * Pageout to vnode, no can do yet. */ if (handle == NULL) return (NULL); vp = (struct vnode *) handle; /* * If the object is being terminated, wait for it to * go away. */ retry: while ((object = vp->v_object) != NULL) { VM_OBJECT_WLOCK(object); if ((object->flags & OBJ_DEAD) == 0) break; vm_object_set_flag(object, OBJ_DISCONNECTWNT); VM_OBJECT_SLEEP(object, object, PDROP | PVM, "vadead", 0); } KASSERT(vp->v_usecount != 0, ("vnode_pager_alloc: no vnode reference")); if (object == NULL) { /* * Add an object of the appropriate size */ object = vm_object_allocate(OBJT_VNODE, OFF_TO_IDX(round_page(size))); object->un_pager.vnp.vnp_size = size; object->un_pager.vnp.writemappings = 0; object->handle = handle; VI_LOCK(vp); if (vp->v_object != NULL) { /* * Object has been created while we were sleeping */ VI_UNLOCK(vp); VM_OBJECT_WLOCK(object); KASSERT(object->ref_count == 1, ("leaked ref %p %d", object, object->ref_count)); object->type = OBJT_DEAD; object->ref_count = 0; VM_OBJECT_WUNLOCK(object); vm_object_destroy(object); goto retry; } vp->v_object = object; VI_UNLOCK(vp); } else { object->ref_count++; #if VM_NRESERVLEVEL > 0 vm_object_color(object, 0); #endif VM_OBJECT_WUNLOCK(object); } vrefact(vp); return (object); } /* * The object must be locked. */ static void vnode_pager_dealloc(vm_object_t object) { struct vnode *vp; int refs; vp = object->handle; if (vp == NULL) panic("vnode_pager_dealloc: pager already dealloced"); VM_OBJECT_ASSERT_WLOCKED(object); vm_object_pip_wait(object, "vnpdea"); refs = object->ref_count; object->handle = NULL; object->type = OBJT_DEAD; if (object->flags & OBJ_DISCONNECTWNT) { vm_object_clear_flag(object, OBJ_DISCONNECTWNT); wakeup(object); } ASSERT_VOP_ELOCKED(vp, "vnode_pager_dealloc"); if (object->un_pager.vnp.writemappings > 0) { object->un_pager.vnp.writemappings = 0; VOP_ADD_WRITECOUNT(vp, -1); CTR3(KTR_VFS, "%s: vp %p v_writecount decreased to %d", __func__, vp, vp->v_writecount); } vp->v_object = NULL; VOP_UNSET_TEXT(vp); VM_OBJECT_WUNLOCK(object); while (refs-- > 0) vunref(vp); VM_OBJECT_WLOCK(object); } static boolean_t vnode_pager_haspage(vm_object_t object, vm_pindex_t pindex, int *before, int *after) { struct vnode *vp = object->handle; daddr_t bn; int err; daddr_t reqblock; int poff; int bsize; int pagesperblock, blocksperpage; VM_OBJECT_ASSERT_WLOCKED(object); /* * If no vp or vp is doomed or marked transparent to VM, we do not * have the page. */ if (vp == NULL || vp->v_iflag & VI_DOOMED) return FALSE; /* * If the offset is beyond end of file we do * not have the page. */ if (IDX_TO_OFF(pindex) >= object->un_pager.vnp.vnp_size) return FALSE; bsize = vp->v_mount->mnt_stat.f_iosize; pagesperblock = bsize / PAGE_SIZE; blocksperpage = 0; if (pagesperblock > 0) { reqblock = pindex / pagesperblock; } else { blocksperpage = (PAGE_SIZE / bsize); reqblock = pindex * blocksperpage; } VM_OBJECT_WUNLOCK(object); err = VOP_BMAP(vp, reqblock, NULL, &bn, after, before); VM_OBJECT_WLOCK(object); if (err) return TRUE; if (bn == -1) return FALSE; if (pagesperblock > 0) { poff = pindex - (reqblock * pagesperblock); if (before) { *before *= pagesperblock; *before += poff; } if (after) { /* * The BMAP vop can report a partial block in the * 'after', but must not report blocks after EOF. * Assert the latter, and truncate 'after' in case * of the former. */ KASSERT((reqblock + *after) * pagesperblock < roundup2(object->size, pagesperblock), ("%s: reqblock %jd after %d size %ju", __func__, (intmax_t )reqblock, *after, (uintmax_t )object->size)); *after *= pagesperblock; *after += pagesperblock - (poff + 1); if (pindex + *after >= object->size) *after = object->size - 1 - pindex; } } else { if (before) { *before /= blocksperpage; } if (after) { *after /= blocksperpage; } } return TRUE; } /* * Lets the VM system know about a change in size for a file. * We adjust our own internal size and flush any cached pages in * the associated object that are affected by the size change. * * Note: this routine may be invoked as a result of a pager put * operation (possibly at object termination time), so we must be careful. */ void vnode_pager_setsize(struct vnode *vp, vm_ooffset_t nsize) { vm_object_t object; vm_page_t m; vm_pindex_t nobjsize; if ((object = vp->v_object) == NULL) return; /* ASSERT_VOP_ELOCKED(vp, "vnode_pager_setsize and not locked vnode"); */ VM_OBJECT_WLOCK(object); if (object->type == OBJT_DEAD) { VM_OBJECT_WUNLOCK(object); return; } KASSERT(object->type == OBJT_VNODE, ("not vnode-backed object %p", object)); if (nsize == object->un_pager.vnp.vnp_size) { /* * Hasn't changed size */ VM_OBJECT_WUNLOCK(object); return; } nobjsize = OFF_TO_IDX(nsize + PAGE_MASK); if (nsize < object->un_pager.vnp.vnp_size) { /* * File has shrunk. Toss any cached pages beyond the new EOF. */ if (nobjsize < object->size) vm_object_page_remove(object, nobjsize, object->size, 0); /* * this gets rid of garbage at the end of a page that is now * only partially backed by the vnode. * * XXX for some reason (I don't know yet), if we take a * completely invalid page and mark it partially valid * it can screw up NFS reads, so we don't allow the case. */ if ((nsize & PAGE_MASK) && (m = vm_page_lookup(object, OFF_TO_IDX(nsize))) != NULL && m->valid != 0) { int base = (int)nsize & PAGE_MASK; int size = PAGE_SIZE - base; /* * Clear out partial-page garbage in case * the page has been mapped. */ pmap_zero_page_area(m, base, size); /* * Update the valid bits to reflect the blocks that * have been zeroed. Some of these valid bits may * have already been set. */ vm_page_set_valid_range(m, base, size); /* * Round "base" to the next block boundary so that the * dirty bit for a partially zeroed block is not * cleared. */ base = roundup2(base, DEV_BSIZE); /* * Clear out partial-page dirty bits. * * note that we do not clear out the valid * bits. This would prevent bogus_page * replacement from working properly. */ vm_page_clear_dirty(m, base, PAGE_SIZE - base); } } object->un_pager.vnp.vnp_size = nsize; object->size = nobjsize; VM_OBJECT_WUNLOCK(object); } /* * calculate the linear (byte) disk address of specified virtual * file address */ static int vnode_pager_addr(struct vnode *vp, vm_ooffset_t address, daddr_t *rtaddress, int *run) { int bsize; int err; daddr_t vblock; daddr_t voffset; if (address < 0) return -1; if (vp->v_iflag & VI_DOOMED) return -1; bsize = vp->v_mount->mnt_stat.f_iosize; vblock = address / bsize; voffset = address % bsize; err = VOP_BMAP(vp, vblock, NULL, rtaddress, run, NULL); if (err == 0) { if (*rtaddress != -1) *rtaddress += voffset / DEV_BSIZE; if (run) { *run += 1; *run *= bsize/PAGE_SIZE; *run -= voffset/PAGE_SIZE; } } return (err); } /* * small block filesystem vnode pager input */ static int vnode_pager_input_smlfs(vm_object_t object, vm_page_t m) { struct vnode *vp; struct bufobj *bo; struct buf *bp; struct sf_buf *sf; daddr_t fileaddr; vm_offset_t bsize; vm_page_bits_t bits; int error, i; error = 0; vp = object->handle; if (vp->v_iflag & VI_DOOMED) return VM_PAGER_BAD; bsize = vp->v_mount->mnt_stat.f_iosize; VOP_BMAP(vp, 0, &bo, 0, NULL, NULL); sf = sf_buf_alloc(m, 0); for (i = 0; i < PAGE_SIZE / bsize; i++) { vm_ooffset_t address; bits = vm_page_bits(i * bsize, bsize); if (m->valid & bits) continue; address = IDX_TO_OFF(m->pindex) + i * bsize; if (address >= object->un_pager.vnp.vnp_size) { fileaddr = -1; } else { error = vnode_pager_addr(vp, address, &fileaddr, NULL); if (error) break; } if (fileaddr != -1) { bp = getpbuf(&vnode_pbuf_freecnt); /* build a minimal buffer header */ bp->b_iocmd = BIO_READ; bp->b_iodone = bdone; KASSERT(bp->b_rcred == NOCRED, ("leaking read ucred")); KASSERT(bp->b_wcred == NOCRED, ("leaking write ucred")); bp->b_rcred = crhold(curthread->td_ucred); bp->b_wcred = crhold(curthread->td_ucred); bp->b_data = (caddr_t)sf_buf_kva(sf) + i * bsize; bp->b_blkno = fileaddr; pbgetbo(bo, bp); bp->b_vp = vp; bp->b_bcount = bsize; bp->b_bufsize = bsize; bp->b_runningbufspace = bp->b_bufsize; atomic_add_long(&runningbufspace, bp->b_runningbufspace); /* do the input */ bp->b_iooffset = dbtob(bp->b_blkno); bstrategy(bp); bwait(bp, PVM, "vnsrd"); if ((bp->b_ioflags & BIO_ERROR) != 0) error = EIO; /* * free the buffer header back to the swap buffer pool */ bp->b_vp = NULL; pbrelbo(bp); relpbuf(bp, &vnode_pbuf_freecnt); if (error) break; } else bzero((caddr_t)sf_buf_kva(sf) + i * bsize, bsize); KASSERT((m->dirty & bits) == 0, ("vnode_pager_input_smlfs: page %p is dirty", m)); VM_OBJECT_WLOCK(object); m->valid |= bits; VM_OBJECT_WUNLOCK(object); } sf_buf_free(sf); if (error) { return VM_PAGER_ERROR; } return VM_PAGER_OK; } /* * old style vnode pager input routine */ static int vnode_pager_input_old(vm_object_t object, vm_page_t m) { struct uio auio; struct iovec aiov; int error; int size; struct sf_buf *sf; struct vnode *vp; VM_OBJECT_ASSERT_WLOCKED(object); error = 0; /* * Return failure if beyond current EOF */ if (IDX_TO_OFF(m->pindex) >= object->un_pager.vnp.vnp_size) { return VM_PAGER_BAD; } else { size = PAGE_SIZE; if (IDX_TO_OFF(m->pindex) + size > object->un_pager.vnp.vnp_size) size = object->un_pager.vnp.vnp_size - IDX_TO_OFF(m->pindex); vp = object->handle; VM_OBJECT_WUNLOCK(object); /* * Allocate a kernel virtual address and initialize so that * we can use VOP_READ/WRITE routines. */ sf = sf_buf_alloc(m, 0); aiov.iov_base = (caddr_t)sf_buf_kva(sf); aiov.iov_len = size; auio.uio_iov = &aiov; auio.uio_iovcnt = 1; auio.uio_offset = IDX_TO_OFF(m->pindex); auio.uio_segflg = UIO_SYSSPACE; auio.uio_rw = UIO_READ; auio.uio_resid = size; auio.uio_td = curthread; error = VOP_READ(vp, &auio, 0, curthread->td_ucred); if (!error) { int count = size - auio.uio_resid; if (count == 0) error = EINVAL; else if (count != PAGE_SIZE) bzero((caddr_t)sf_buf_kva(sf) + count, PAGE_SIZE - count); } sf_buf_free(sf); VM_OBJECT_WLOCK(object); } KASSERT(m->dirty == 0, ("vnode_pager_input_old: page %p is dirty", m)); if (!error) m->valid = VM_PAGE_BITS_ALL; return error ? VM_PAGER_ERROR : VM_PAGER_OK; } /* * generic vnode pager input routine */ /* * Local media VFS's that do not implement their own VOP_GETPAGES * should have their VOP_GETPAGES call to vnode_pager_generic_getpages() * to implement the previous behaviour. * * All other FS's should use the bypass to get to the local media * backing vp's VOP_GETPAGES. */ static int vnode_pager_getpages(vm_object_t object, vm_page_t *m, int count, int *rbehind, int *rahead) { struct vnode *vp; int rtval; vp = object->handle; VM_OBJECT_WUNLOCK(object); rtval = VOP_GETPAGES(vp, m, count, rbehind, rahead); KASSERT(rtval != EOPNOTSUPP, ("vnode_pager: FS getpages not implemented\n")); VM_OBJECT_WLOCK(object); return rtval; } static int vnode_pager_getpages_async(vm_object_t object, vm_page_t *m, int count, int *rbehind, int *rahead, vop_getpages_iodone_t iodone, void *arg) { struct vnode *vp; int rtval; vp = object->handle; VM_OBJECT_WUNLOCK(object); rtval = VOP_GETPAGES_ASYNC(vp, m, count, rbehind, rahead, iodone, arg); KASSERT(rtval != EOPNOTSUPP, ("vnode_pager: FS getpages_async not implemented\n")); VM_OBJECT_WLOCK(object); return (rtval); } /* * The implementation of VOP_GETPAGES() and VOP_GETPAGES_ASYNC() for * local filesystems, where partially valid pages can only occur at * the end of file. */ int vnode_pager_local_getpages(struct vop_getpages_args *ap) { return (vnode_pager_generic_getpages(ap->a_vp, ap->a_m, ap->a_count, ap->a_rbehind, ap->a_rahead, NULL, NULL)); } int vnode_pager_local_getpages_async(struct vop_getpages_async_args *ap) { return (vnode_pager_generic_getpages(ap->a_vp, ap->a_m, ap->a_count, ap->a_rbehind, ap->a_rahead, ap->a_iodone, ap->a_arg)); } /* * This is now called from local media FS's to operate against their * own vnodes if they fail to implement VOP_GETPAGES. */ int vnode_pager_generic_getpages(struct vnode *vp, vm_page_t *m, int count, int *a_rbehind, int *a_rahead, vop_getpages_iodone_t iodone, void *arg) { vm_object_t object; struct bufobj *bo; struct buf *bp; off_t foff; #ifdef INVARIANTS off_t blkno0; #endif int bsize, pagesperblock, *freecnt; int error, before, after, rbehind, rahead, poff, i; int bytecount, secmask; KASSERT(vp->v_type != VCHR && vp->v_type != VBLK, ("%s does not support devices", __func__)); if (vp->v_iflag & VI_DOOMED) return (VM_PAGER_BAD); object = vp->v_object; foff = IDX_TO_OFF(m[0]->pindex); bsize = vp->v_mount->mnt_stat.f_iosize; pagesperblock = bsize / PAGE_SIZE; KASSERT(foff < object->un_pager.vnp.vnp_size, ("%s: page %p offset beyond vp %p size", __func__, m[0], vp)); KASSERT(count <= sizeof(bp->b_pages), ("%s: requested %d pages", __func__, count)); /* * The last page has valid blocks. Invalid part can only * exist at the end of file, and the page is made fully valid * by zeroing in vm_pager_get_pages(). */ if (m[count - 1]->valid != 0 && --count == 0) { if (iodone != NULL) iodone(arg, m, 1, 0); return (VM_PAGER_OK); } /* * Synchronous and asynchronous paging operations use different * free pbuf counters. This is done to avoid asynchronous requests * to consume all pbufs. * Allocate the pbuf at the very beginning of the function, so that * if we are low on certain kind of pbufs don't even proceed to BMAP, * but sleep. */ freecnt = iodone != NULL ? &vnode_async_pbuf_freecnt : &vnode_pbuf_freecnt; bp = getpbuf(freecnt); /* * Get the underlying device blocks for the file with VOP_BMAP(). * If the file system doesn't support VOP_BMAP, use old way of * getting pages via VOP_READ. */ error = VOP_BMAP(vp, foff / bsize, &bo, &bp->b_blkno, &after, &before); if (error == EOPNOTSUPP) { relpbuf(bp, freecnt); VM_OBJECT_WLOCK(object); for (i = 0; i < count; i++) { PCPU_INC(cnt.v_vnodein); PCPU_INC(cnt.v_vnodepgsin); error = vnode_pager_input_old(object, m[i]); if (error) break; } VM_OBJECT_WUNLOCK(object); return (error); } else if (error != 0) { relpbuf(bp, freecnt); return (VM_PAGER_ERROR); } /* * If the file system supports BMAP, but blocksize is smaller * than a page size, then use special small filesystem code. */ if (pagesperblock == 0) { relpbuf(bp, freecnt); for (i = 0; i < count; i++) { PCPU_INC(cnt.v_vnodein); PCPU_INC(cnt.v_vnodepgsin); error = vnode_pager_input_smlfs(object, m[i]); if (error) break; } return (error); } /* * A sparse file can be encountered only for a single page request, * which may not be preceded by call to vm_pager_haspage(). */ if (bp->b_blkno == -1) { KASSERT(count == 1, ("%s: array[%d] request to a sparse file %p", __func__, count, vp)); relpbuf(bp, freecnt); pmap_zero_page(m[0]); KASSERT(m[0]->dirty == 0, ("%s: page %p is dirty", __func__, m[0])); VM_OBJECT_WLOCK(object); m[0]->valid = VM_PAGE_BITS_ALL; VM_OBJECT_WUNLOCK(object); return (VM_PAGER_OK); } #ifdef INVARIANTS blkno0 = bp->b_blkno; #endif bp->b_blkno += (foff % bsize) / DEV_BSIZE; /* Recalculate blocks available after/before to pages. */ poff = (foff % bsize) / PAGE_SIZE; before *= pagesperblock; before += poff; after *= pagesperblock; after += pagesperblock - (poff + 1); if (m[0]->pindex + after >= object->size) after = object->size - 1 - m[0]->pindex; KASSERT(count <= after + 1, ("%s: %d pages asked, can do only %d", __func__, count, after + 1)); after -= count - 1; /* Trim requested rbehind/rahead to possible values. */ rbehind = a_rbehind ? *a_rbehind : 0; rahead = a_rahead ? *a_rahead : 0; rbehind = min(rbehind, before); rbehind = min(rbehind, m[0]->pindex); rahead = min(rahead, after); rahead = min(rahead, object->size - m[count - 1]->pindex); /* * Check that total amount of pages fit into buf. Trim rbehind and * rahead evenly if not. */ if (rbehind + rahead + count > nitems(bp->b_pages)) { int trim, sum; trim = rbehind + rahead + count - nitems(bp->b_pages) + 1; sum = rbehind + rahead; if (rbehind == before) { /* Roundup rbehind trim to block size. */ rbehind -= roundup(trim * rbehind / sum, pagesperblock); if (rbehind < 0) rbehind = 0; } else rbehind -= trim * rbehind / sum; rahead -= trim * rahead / sum; } KASSERT(rbehind + rahead + count <= nitems(bp->b_pages), ("%s: behind %d ahead %d count %d", __func__, rbehind, rahead, count)); /* * Fill in the bp->b_pages[] array with requested and optional * read behind or read ahead pages. Read behind pages are looked * up in a backward direction, down to a first cached page. Same * for read ahead pages, but there is no need to shift the array * in case of encountering a cached page. */ i = bp->b_npages = 0; if (rbehind) { vm_pindex_t startpindex, tpindex; vm_page_t p; VM_OBJECT_WLOCK(object); startpindex = m[0]->pindex - rbehind; if ((p = TAILQ_PREV(m[0], pglist, listq)) != NULL && p->pindex >= startpindex) startpindex = p->pindex + 1; /* tpindex is unsigned; beware of numeric underflow. */ for (tpindex = m[0]->pindex - 1; tpindex >= startpindex && tpindex < m[0]->pindex; tpindex--, i++) { p = vm_page_alloc(object, tpindex, VM_ALLOC_NORMAL); if (p == NULL) { /* Shift the array. */ for (int j = 0; j < i; j++) bp->b_pages[j] = bp->b_pages[j + tpindex + 1 - startpindex]; break; } bp->b_pages[tpindex - startpindex] = p; } bp->b_pgbefore = i; bp->b_npages += i; bp->b_blkno -= IDX_TO_OFF(i) / DEV_BSIZE; } else bp->b_pgbefore = 0; /* Requested pages. */ for (int j = 0; j < count; j++, i++) bp->b_pages[i] = m[j]; bp->b_npages += count; if (rahead) { vm_pindex_t endpindex, tpindex; vm_page_t p; if (!VM_OBJECT_WOWNED(object)) VM_OBJECT_WLOCK(object); endpindex = m[count - 1]->pindex + rahead + 1; if ((p = TAILQ_NEXT(m[count - 1], listq)) != NULL && p->pindex < endpindex) endpindex = p->pindex; if (endpindex > object->size) endpindex = object->size; for (tpindex = m[count - 1]->pindex + 1; tpindex < endpindex; i++, tpindex++) { p = vm_page_alloc(object, tpindex, VM_ALLOC_NORMAL); if (p == NULL) break; bp->b_pages[i] = p; } bp->b_pgafter = i - bp->b_npages; bp->b_npages = i; } else bp->b_pgafter = 0; if (VM_OBJECT_WOWNED(object)) VM_OBJECT_WUNLOCK(object); /* Report back actual behind/ahead read. */ if (a_rbehind) *a_rbehind = bp->b_pgbefore; if (a_rahead) *a_rahead = bp->b_pgafter; #ifdef INVARIANTS KASSERT(bp->b_npages <= nitems(bp->b_pages), ("%s: buf %p overflowed", __func__, bp)); - for (int j = 1; j < bp->b_npages; j++) - KASSERT(bp->b_pages[j]->pindex - 1 == - bp->b_pages[j - 1]->pindex, - ("%s: pages array not consecutive, bp %p", __func__, bp)); + for (int j = 1, prev = 1; j < bp->b_npages; j++) { + if (bp->b_pages[j] == bogus_page) + continue; + KASSERT(bp->b_pages[j]->pindex - bp->b_pages[prev]->pindex == + j - prev, ("%s: pages array not consecutive, bp %p", + __func__, bp)); + prev = j; + } #endif /* * Recalculate first offset and bytecount with regards to read behind. * Truncate bytecount to vnode real size and round up physical size * for real devices. */ foff = IDX_TO_OFF(bp->b_pages[0]->pindex); bytecount = bp->b_npages << PAGE_SHIFT; if ((foff + bytecount) > object->un_pager.vnp.vnp_size) bytecount = object->un_pager.vnp.vnp_size - foff; secmask = bo->bo_bsize - 1; KASSERT(secmask < PAGE_SIZE && secmask > 0, ("%s: sector size %d too large", __func__, secmask + 1)); bytecount = (bytecount + secmask) & ~secmask; /* * And map the pages to be read into the kva, if the filesystem * requires mapped buffers. */ if ((vp->v_mount->mnt_kern_flag & MNTK_UNMAPPED_BUFS) != 0 && unmapped_buf_allowed) { bp->b_data = unmapped_buf; bp->b_offset = 0; } else { bp->b_data = bp->b_kvabase; pmap_qenter((vm_offset_t)bp->b_data, bp->b_pages, bp->b_npages); } /* Build a minimal buffer header. */ bp->b_iocmd = BIO_READ; KASSERT(bp->b_rcred == NOCRED, ("leaking read ucred")); KASSERT(bp->b_wcred == NOCRED, ("leaking write ucred")); bp->b_rcred = crhold(curthread->td_ucred); bp->b_wcred = crhold(curthread->td_ucred); pbgetbo(bo, bp); bp->b_vp = vp; bp->b_bcount = bp->b_bufsize = bp->b_runningbufspace = bytecount; bp->b_iooffset = dbtob(bp->b_blkno); KASSERT(IDX_TO_OFF(m[0]->pindex - bp->b_pages[0]->pindex) == (blkno0 - bp->b_blkno) * DEV_BSIZE + IDX_TO_OFF(m[0]->pindex) % bsize, ("wrong offsets bsize %d m[0] %ju b_pages[0] %ju " "blkno0 %ju b_blkno %ju", bsize, (uintmax_t)m[0]->pindex, (uintmax_t)bp->b_pages[0]->pindex, (uintmax_t)blkno0, (uintmax_t)bp->b_blkno)); atomic_add_long(&runningbufspace, bp->b_runningbufspace); PCPU_INC(cnt.v_vnodein); PCPU_ADD(cnt.v_vnodepgsin, bp->b_npages); if (iodone != NULL) { /* async */ bp->b_pgiodone = iodone; bp->b_caller1 = arg; bp->b_iodone = vnode_pager_generic_getpages_done_async; bp->b_flags |= B_ASYNC; BUF_KERNPROC(bp); bstrategy(bp); return (VM_PAGER_OK); } else { bp->b_iodone = bdone; bstrategy(bp); bwait(bp, PVM, "vnread"); error = vnode_pager_generic_getpages_done(bp); for (i = 0; i < bp->b_npages; i++) bp->b_pages[i] = NULL; bp->b_vp = NULL; pbrelbo(bp); relpbuf(bp, &vnode_pbuf_freecnt); return (error != 0 ? VM_PAGER_ERROR : VM_PAGER_OK); } } static void vnode_pager_generic_getpages_done_async(struct buf *bp) { int error; error = vnode_pager_generic_getpages_done(bp); /* Run the iodone upon the requested range. */ bp->b_pgiodone(bp->b_caller1, bp->b_pages + bp->b_pgbefore, bp->b_npages - bp->b_pgbefore - bp->b_pgafter, error); for (int i = 0; i < bp->b_npages; i++) bp->b_pages[i] = NULL; bp->b_vp = NULL; pbrelbo(bp); relpbuf(bp, &vnode_async_pbuf_freecnt); } static int vnode_pager_generic_getpages_done(struct buf *bp) { vm_object_t object; off_t tfoff, nextoff; int i, error; error = (bp->b_ioflags & BIO_ERROR) != 0 ? EIO : 0; object = bp->b_vp->v_object; if (error == 0 && bp->b_bcount != bp->b_npages * PAGE_SIZE) { if (!buf_mapped(bp)) { bp->b_data = bp->b_kvabase; pmap_qenter((vm_offset_t)bp->b_data, bp->b_pages, bp->b_npages); } bzero(bp->b_data + bp->b_bcount, PAGE_SIZE * bp->b_npages - bp->b_bcount); } if (buf_mapped(bp)) { pmap_qremove((vm_offset_t)bp->b_data, bp->b_npages); bp->b_data = unmapped_buf; } VM_OBJECT_WLOCK(object); for (i = 0, tfoff = IDX_TO_OFF(bp->b_pages[0]->pindex); i < bp->b_npages; i++, tfoff = nextoff) { vm_page_t mt; nextoff = tfoff + PAGE_SIZE; mt = bp->b_pages[i]; if (nextoff <= object->un_pager.vnp.vnp_size) { /* * Read filled up entire page. */ mt->valid = VM_PAGE_BITS_ALL; KASSERT(mt->dirty == 0, ("%s: page %p is dirty", __func__, mt)); KASSERT(!pmap_page_is_mapped(mt), ("%s: page %p is mapped", __func__, mt)); } else { /* * Read did not fill up entire page. * * Currently we do not set the entire page valid, * we just try to clear the piece that we couldn't * read. */ vm_page_set_valid_range(mt, 0, object->un_pager.vnp.vnp_size - tfoff); KASSERT((mt->dirty & vm_page_bits(0, object->un_pager.vnp.vnp_size - tfoff)) == 0, ("%s: page %p is dirty", __func__, mt)); } if (i < bp->b_pgbefore || i >= bp->b_npages - bp->b_pgafter) vm_page_readahead_finish(mt); } VM_OBJECT_WUNLOCK(object); if (error != 0) printf("%s: I/O read error %d\n", __func__, error); return (error); } /* * EOPNOTSUPP is no longer legal. For local media VFS's that do not * implement their own VOP_PUTPAGES, their VOP_PUTPAGES should call to * vnode_pager_generic_putpages() to implement the previous behaviour. * * All other FS's should use the bypass to get to the local media * backing vp's VOP_PUTPAGES. */ static void vnode_pager_putpages(vm_object_t object, vm_page_t *m, int count, int flags, int *rtvals) { int rtval; struct vnode *vp; int bytes = count * PAGE_SIZE; /* * Force synchronous operation if we are extremely low on memory * to prevent a low-memory deadlock. VOP operations often need to * allocate more memory to initiate the I/O ( i.e. do a BMAP * operation ). The swapper handles the case by limiting the amount * of asynchronous I/O, but that sort of solution doesn't scale well * for the vnode pager without a lot of work. * * Also, the backing vnode's iodone routine may not wake the pageout * daemon up. This should be probably be addressed XXX. */ if (vm_cnt.v_free_count < vm_cnt.v_pageout_free_min) flags |= VM_PAGER_PUT_SYNC; /* * Call device-specific putpages function */ vp = object->handle; VM_OBJECT_WUNLOCK(object); rtval = VOP_PUTPAGES(vp, m, bytes, flags, rtvals); KASSERT(rtval != EOPNOTSUPP, ("vnode_pager: stale FS putpages\n")); VM_OBJECT_WLOCK(object); } /* * This is now called from local media FS's to operate against their * own vnodes if they fail to implement VOP_PUTPAGES. * * This is typically called indirectly via the pageout daemon and * clustering has already typically occurred, so in general we ask the * underlying filesystem to write the data out asynchronously rather * then delayed. */ int vnode_pager_generic_putpages(struct vnode *vp, vm_page_t *ma, int bytecount, int flags, int *rtvals) { int i; vm_object_t object; vm_page_t m; int count; int maxsize, ncount; vm_ooffset_t poffset; struct uio auio; struct iovec aiov; int error; int ioflags; int ppscheck = 0; static struct timeval lastfail; static int curfail; object = vp->v_object; count = bytecount / PAGE_SIZE; for (i = 0; i < count; i++) rtvals[i] = VM_PAGER_ERROR; if ((int64_t)ma[0]->pindex < 0) { printf("vnode_pager_putpages: attempt to write meta-data!!! -- 0x%lx(%lx)\n", (long)ma[0]->pindex, (u_long)ma[0]->dirty); rtvals[0] = VM_PAGER_BAD; return VM_PAGER_BAD; } maxsize = count * PAGE_SIZE; ncount = count; poffset = IDX_TO_OFF(ma[0]->pindex); /* * If the page-aligned write is larger then the actual file we * have to invalidate pages occurring beyond the file EOF. However, * there is an edge case where a file may not be page-aligned where * the last page is partially invalid. In this case the filesystem * may not properly clear the dirty bits for the entire page (which * could be VM_PAGE_BITS_ALL due to the page having been mmap()d). * With the page locked we are free to fix-up the dirty bits here. * * We do not under any circumstances truncate the valid bits, as * this will screw up bogus page replacement. */ VM_OBJECT_WLOCK(object); if (maxsize + poffset > object->un_pager.vnp.vnp_size) { if (object->un_pager.vnp.vnp_size > poffset) { int pgoff; maxsize = object->un_pager.vnp.vnp_size - poffset; ncount = btoc(maxsize); if ((pgoff = (int)maxsize & PAGE_MASK) != 0) { /* * If the object is locked and the following * conditions hold, then the page's dirty * field cannot be concurrently changed by a * pmap operation. */ m = ma[ncount - 1]; vm_page_assert_sbusied(m); KASSERT(!pmap_page_is_write_mapped(m), ("vnode_pager_generic_putpages: page %p is not read-only", m)); vm_page_clear_dirty(m, pgoff, PAGE_SIZE - pgoff); } } else { maxsize = 0; ncount = 0; } if (ncount < count) { for (i = ncount; i < count; i++) { rtvals[i] = VM_PAGER_BAD; } } } VM_OBJECT_WUNLOCK(object); /* * pageouts are already clustered, use IO_ASYNC to force a bawrite() * rather then a bdwrite() to prevent paging I/O from saturating * the buffer cache. Dummy-up the sequential heuristic to cause * large ranges to cluster. If neither IO_SYNC or IO_ASYNC is set, * the system decides how to cluster. */ ioflags = IO_VMIO; if (flags & (VM_PAGER_PUT_SYNC | VM_PAGER_PUT_INVAL)) ioflags |= IO_SYNC; else if ((flags & VM_PAGER_CLUSTER_OK) == 0) ioflags |= IO_ASYNC; ioflags |= (flags & VM_PAGER_PUT_INVAL) ? IO_INVAL: 0; ioflags |= (flags & VM_PAGER_PUT_NOREUSE) ? IO_NOREUSE : 0; ioflags |= IO_SEQMAX << IO_SEQSHIFT; aiov.iov_base = (caddr_t) 0; aiov.iov_len = maxsize; auio.uio_iov = &aiov; auio.uio_iovcnt = 1; auio.uio_offset = poffset; auio.uio_segflg = UIO_NOCOPY; auio.uio_rw = UIO_WRITE; auio.uio_resid = maxsize; auio.uio_td = (struct thread *) 0; error = VOP_WRITE(vp, &auio, ioflags, curthread->td_ucred); PCPU_INC(cnt.v_vnodeout); PCPU_ADD(cnt.v_vnodepgsout, ncount); if (error) { if ((ppscheck = ppsratecheck(&lastfail, &curfail, 1))) printf("vnode_pager_putpages: I/O error %d\n", error); } if (auio.uio_resid) { if (ppscheck || ppsratecheck(&lastfail, &curfail, 1)) printf("vnode_pager_putpages: residual I/O %zd at %lu\n", auio.uio_resid, (u_long)ma[0]->pindex); } for (i = 0; i < ncount; i++) { rtvals[i] = VM_PAGER_OK; } return rtvals[0]; } void vnode_pager_undirty_pages(vm_page_t *ma, int *rtvals, int written) { vm_object_t obj; int i, pos; if (written == 0) return; obj = ma[0]->object; VM_OBJECT_WLOCK(obj); for (i = 0, pos = 0; pos < written; i++, pos += PAGE_SIZE) { if (pos < trunc_page(written)) { rtvals[i] = VM_PAGER_OK; vm_page_undirty(ma[i]); } else { /* Partially written page. */ rtvals[i] = VM_PAGER_AGAIN; vm_page_clear_dirty(ma[i], 0, written & PAGE_MASK); } } VM_OBJECT_WUNLOCK(obj); } void vnode_pager_update_writecount(vm_object_t object, vm_offset_t start, vm_offset_t end) { struct vnode *vp; vm_ooffset_t old_wm; VM_OBJECT_WLOCK(object); if (object->type != OBJT_VNODE) { VM_OBJECT_WUNLOCK(object); return; } old_wm = object->un_pager.vnp.writemappings; object->un_pager.vnp.writemappings += (vm_ooffset_t)end - start; vp = object->handle; if (old_wm == 0 && object->un_pager.vnp.writemappings != 0) { ASSERT_VOP_ELOCKED(vp, "v_writecount inc"); VOP_ADD_WRITECOUNT(vp, 1); CTR3(KTR_VFS, "%s: vp %p v_writecount increased to %d", __func__, vp, vp->v_writecount); } else if (old_wm != 0 && object->un_pager.vnp.writemappings == 0) { ASSERT_VOP_ELOCKED(vp, "v_writecount dec"); VOP_ADD_WRITECOUNT(vp, -1); CTR3(KTR_VFS, "%s: vp %p v_writecount decreased to %d", __func__, vp, vp->v_writecount); } VM_OBJECT_WUNLOCK(object); } void vnode_pager_release_writecount(vm_object_t object, vm_offset_t start, vm_offset_t end) { struct vnode *vp; struct mount *mp; vm_offset_t inc; VM_OBJECT_WLOCK(object); /* * First, recheck the object type to account for the race when * the vnode is reclaimed. */ if (object->type != OBJT_VNODE) { VM_OBJECT_WUNLOCK(object); return; } /* * Optimize for the case when writemappings is not going to * zero. */ inc = end - start; if (object->un_pager.vnp.writemappings != inc) { object->un_pager.vnp.writemappings -= inc; VM_OBJECT_WUNLOCK(object); return; } vp = object->handle; vhold(vp); VM_OBJECT_WUNLOCK(object); mp = NULL; vn_start_write(vp, &mp, V_WAIT); vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); /* * Decrement the object's writemappings, by swapping the start * and end arguments for vnode_pager_update_writecount(). If * there was not a race with vnode reclaimation, then the * vnode's v_writecount is decremented. */ vnode_pager_update_writecount(object, end, start); VOP_UNLOCK(vp, 0); vdrop(vp); if (mp != NULL) vn_finished_write(mp); } Index: projects/clang400-import/usr.bin/netstat/if.c =================================================================== --- projects/clang400-import/usr.bin/netstat/if.c (revision 311459) +++ projects/clang400-import/usr.bin/netstat/if.c (revision 311460) @@ -1,655 +1,655 @@ /*- * Copyright (c) 2013 Gleb Smirnoff * Copyright (c) 1983, 1988, 1993 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #if 0 #ifndef lint static char sccsid[] = "@(#)if.c 8.3 (Berkeley) 4/28/95"; #endif /* not lint */ #endif #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #ifdef PF #include #include #endif #include #include #include #include #ifdef INET6 #include #endif #include #include #include #include #include #include #include #include #include #include "netstat.h" static void sidewaysintpr(void); #ifdef PF static const char* pfsyncacts[] = { /* PFSYNC_ACT_CLR */ "clear all request", /* PFSYNC_ACT_INS */ "state insert", /* PFSYNC_ACT_INS_ACK */ "state inserted ack", /* PFSYNC_ACT_UPD */ "state update", /* PFSYNC_ACT_UPD_C */ "compressed state update", /* PFSYNC_ACT_UPD_REQ */ "uncompressed state request", /* PFSYNC_ACT_DEL */ "state delete", /* PFSYNC_ACT_DEL_C */ "compressed state delete", /* PFSYNC_ACT_INS_F */ "fragment insert", /* PFSYNC_ACT_DEL_F */ "fragment delete", /* PFSYNC_ACT_BUS */ "bulk update mark", /* PFSYNC_ACT_TDB */ "TDB replay counter update", /* PFSYNC_ACT_EOF */ "end of frame mark", }; static const char* pfsyncacts_name[] = { /* PFSYNC_ACT_CLR */ "clear-all-request", /* PFSYNC_ACT_INS */ "state-insert", /* PFSYNC_ACT_INS_ACK */ "state-inserted-ack", /* PFSYNC_ACT_UPD */ "state-update", /* PFSYNC_ACT_UPD_C */ "compressed-state-update", /* PFSYNC_ACT_UPD_REQ */ "uncompressed-state-request", /* PFSYNC_ACT_DEL */ "state-delete", /* PFSYNC_ACT_DEL_C */ "compressed-state-delete", /* PFSYNC_ACT_INS_F */ "fragment-insert", /* PFSYNC_ACT_DEL_F */ "fragment-delete", /* PFSYNC_ACT_BUS */ "bulk-update-mark", /* PFSYNC_ACT_TDB */ "TDB-replay-counter-update", /* PFSYNC_ACT_EOF */ "end-of-frame-mark", }; static void pfsync_acts_stats(const char *list, const char *desc, uint64_t *a) { int i; xo_open_list(list); for (i = 0; i < PFSYNC_ACT_MAX; i++, a++) { if (*a || sflag <= 1) { xo_open_instance(list); xo_emit("\t\t{e:name}{:count/%ju} {N:/%s%s %s}\n", pfsyncacts_name[i], (uintmax_t)(*a), pfsyncacts[i], plural(*a), desc); xo_close_instance(list); } } xo_close_list(list); } /* * Dump pfsync statistics structure. */ void pfsync_stats(u_long off, const char *name, int af1 __unused, int proto __unused) { struct pfsyncstats pfsyncstat; if (fetch_stats("net.pfsync.stats", off, &pfsyncstat, sizeof(pfsyncstat), kread) != 0) return; xo_emit("{T:/%s}:\n", name); xo_open_container(name); #define p(f, m) if (pfsyncstat.f || sflag <= 1) \ xo_emit(m, (uintmax_t)pfsyncstat.f, plural(pfsyncstat.f)) p(pfsyncs_ipackets, "\t{:received-inet-packets/%ju} " "{N:/packet%s received (IPv4)}\n"); p(pfsyncs_ipackets6, "\t{:received-inet6-packets/%ju} " "{N:/packet%s received (IPv6)}\n"); pfsync_acts_stats("input-histogram", "received", &pfsyncstat.pfsyncs_iacts[0]); p(pfsyncs_badif, "\t\t/{:dropped-bad-interface/%ju} " "{N:/packet%s discarded for bad interface}\n"); p(pfsyncs_badttl, "\t\t{:dropped-bad-ttl/%ju} " "{N:/packet%s discarded for bad ttl}\n"); p(pfsyncs_hdrops, "\t\t{:dropped-short-header/%ju} " "{N:/packet%s shorter than header}\n"); p(pfsyncs_badver, "\t\t{:dropped-bad-version/%ju} " "{N:/packet%s discarded for bad version}\n"); p(pfsyncs_badauth, "\t\t{:dropped-bad-auth/%ju} " "{N:/packet%s discarded for bad HMAC}\n"); p(pfsyncs_badact,"\t\t{:dropped-bad-action/%ju} " "{N:/packet%s discarded for bad action}\n"); p(pfsyncs_badlen, "\t\t{:dropped-short/%ju} " "{N:/packet%s discarded for short packet}\n"); p(pfsyncs_badval, "\t\t{:dropped-bad-values/%ju} " "{N:/state%s discarded for bad values}\n"); p(pfsyncs_stale, "\t\t{:dropped-stale-state/%ju} " "{N:/stale state%s}\n"); p(pfsyncs_badstate, "\t\t{:dropped-failed-lookup/%ju} " "{N:/failed state lookup\\/insert%s}\n"); p(pfsyncs_opackets, "\t{:sent-inet-packets/%ju} " "{N:/packet%s sent (IPv4})\n"); p(pfsyncs_opackets6, "\t{:send-inet6-packets/%ju} " "{N:/packet%s sent (IPv6})\n"); pfsync_acts_stats("output-histogram", "sent", &pfsyncstat.pfsyncs_oacts[0]); p(pfsyncs_onomem, "\t\t{:discarded-no-memory/%ju} " "{N:/failure%s due to mbuf memory error}\n"); p(pfsyncs_oerrors, "\t\t{:send-errors/%ju} " "{N:/send error%s}\n"); #undef p xo_close_container(name); } #endif /* PF */ /* * Display a formatted value, or a '-' in the same space. */ static void show_stat(const char *fmt, int width, const char *name, u_long value, short showvalue, int div1000) { const char *lsep, *rsep; char newfmt[64]; lsep = ""; if (strncmp(fmt, "LS", 2) == 0) { lsep = " "; fmt += 2; } rsep = " "; if (strncmp(fmt, "NRS", 3) == 0) { rsep = ""; fmt += 3; } if (showvalue == 0) { /* Print just dash. */ xo_emit("{P:/%s}{D:/%*s}{P:/%s}", lsep, width, "-", rsep); return; } /* * XXX: workaround {P:} modifier can't be empty and doesn't seem to * take args... so we need to conditionally include it in the format. */ #define maybe_pad(pad) do { \ if (strlen(pad)) { \ snprintf(newfmt, sizeof(newfmt), "{P:%s}", pad); \ xo_emit(newfmt); \ } \ } while (0) if (hflag) { char buf[5]; /* Format in human readable form. */ humanize_number(buf, sizeof(buf), (int64_t)value, "", HN_AUTOSCALE, HN_NOSPACE | HN_DECIMAL | \ ((div1000) ? HN_DIVISOR_1000 : 0)); maybe_pad(lsep); snprintf(newfmt, sizeof(newfmt), "{:%s/%%%ds}", name, width); xo_emit(newfmt, buf); maybe_pad(rsep); } else { /* Construct the format string. */ maybe_pad(lsep); snprintf(newfmt, sizeof(newfmt), "{:%s/%%%d%s}", name, width, fmt); xo_emit(newfmt, value); maybe_pad(rsep); } } /* * Find next multiaddr for a given interface name. */ static struct ifmaddrs * next_ifma(struct ifmaddrs *ifma, const char *name, const sa_family_t family) { for(; ifma != NULL; ifma = ifma->ifma_next) { struct sockaddr_dl *sdl; sdl = (struct sockaddr_dl *)ifma->ifma_name; if (ifma->ifma_addr->sa_family == family && strcmp(sdl->sdl_data, name) == 0) break; } return (ifma); } /* * Print a description of the network interfaces. */ void intpr(void (*pfunc)(char *), int af) { struct ifaddrs *ifap, *ifa; struct ifmaddrs *ifmap, *ifma; u_int ifn_len_max = 5, ifn_len; u_int has_ipv6 = 0, net_len = 13, addr_len = 17; if (interval) return sidewaysintpr(); if (getifaddrs(&ifap) != 0) err(EX_OSERR, "getifaddrs"); if (aflag && getifmaddrs(&ifmap) != 0) err(EX_OSERR, "getifmaddrs"); if (Wflag) { for (ifa = ifap; ifa; ifa = ifa->ifa_next) { if (interface != NULL && strcmp(ifa->ifa_name, interface) != 0) continue; if (af != AF_UNSPEC && ifa->ifa_addr->sa_family != af) continue; ifn_len = strlen(ifa->ifa_name); if ((ifa->ifa_flags & IFF_UP) == 0) ++ifn_len; ifn_len_max = MAX(ifn_len_max, ifn_len); if (ifa->ifa_addr->sa_family == AF_INET6) has_ipv6 = 1; } if (has_ipv6) { net_len = 24; addr_len = 39; } else net_len = 18; } xo_open_list("interface"); if (!pfunc) { xo_emit("{T:/%-*.*s}", ifn_len_max, ifn_len_max, "Name"); xo_emit(" {T:/%5.5s} {T:/%-*.*s} {T:/%-*.*s} {T:/%8.8s} " "{T:/%5.5s} {T:/%5.5s}", "Mtu", net_len, net_len, "Network", addr_len, addr_len, "Address", "Ipkts", "Ierrs", "Idrop"); if (bflag) xo_emit(" {T:/%10.10s}","Ibytes"); xo_emit(" {T:/%8.8s} {T:/%5.5s}", "Opkts", "Oerrs"); if (bflag) xo_emit(" {T:/%10.10s}","Obytes"); xo_emit(" {T:/%5s}", "Coll"); if (dflag) xo_emit(" {T:/%5.5s}", "Drop"); xo_emit("\n"); } for (ifa = ifap; ifa; ifa = ifa->ifa_next) { bool network = false, link = false; char *name, *xname, buf[IFNAMSIZ+1]; const char *nn, *rn; if (interface != NULL && strcmp(ifa->ifa_name, interface) != 0) continue; name = ifa->ifa_name; if (pfunc) { (*pfunc)(name); /* * Skip all ifaddrs belonging to same interface. */ while(ifa->ifa_next != NULL && (strcmp(ifa->ifa_next->ifa_name, name) == 0)) { ifa = ifa->ifa_next; } continue; } if (af != AF_UNSPEC && ifa->ifa_addr->sa_family != af) continue; xo_open_instance("interface"); if ((ifa->ifa_flags & IFF_UP) == 0) { xname = stpcpy(buf, name); *xname++ = '*'; *xname = '\0'; xname = buf; } else xname = name; xo_emit("{d:/%-*.*s}{etk:name}{eq:flags/0x%x}", ifn_len_max, ifn_len_max, xname, name, ifa->ifa_flags); #define IFA_MTU(ifa) (((struct if_data *)(ifa)->ifa_data)->ifi_mtu) show_stat("lu", 6, "mtu", IFA_MTU(ifa), IFA_MTU(ifa), 0); #undef IFA_MTU switch (ifa->ifa_addr->sa_family) { case AF_UNSPEC: xo_emit("{:network/%-*.*s} ", net_len, net_len, "none"); xo_emit("{:address/%-*.*s} ", addr_len, addr_len, "none"); break; case AF_INET: #ifdef INET6 case AF_INET6: #endif /* INET6 */ nn = netname(ifa->ifa_addr, ifa->ifa_netmask); rn = routename(ifa->ifa_addr, numeric_addr); if (Wflag) { xo_emit("{t:network/%-*s} ", net_len, nn); xo_emit("{t:address/%-*s} ", addr_len, rn); } else { xo_emit("{d:network/%-*.*s}{et:network} ", net_len, net_len, nn, nn); xo_emit("{d:address/%-*.*s}{et:address} ", addr_len, addr_len, rn, rn); } network = true; break; case AF_LINK: { struct sockaddr_dl *sdl; - char linknum[10]; + char linknum[sizeof("")]; sdl = (struct sockaddr_dl *)ifa->ifa_addr; - sprintf(linknum, "", sdl->sdl_index); + snprintf(linknum, sizeof(linknum), "", sdl->sdl_index); xo_emit("{t:network/%-*.*s} ", net_len, net_len, linknum); if (sdl->sdl_nlen == 0 && sdl->sdl_alen == 0 && sdl->sdl_slen == 0) xo_emit("{P:/%*s} ", addr_len, ""); else xo_emit("{t:address/%-*.*s} ", addr_len, addr_len, routename(ifa->ifa_addr, 1)); link = true; break; } } #define IFA_STAT(s) (((struct if_data *)ifa->ifa_data)->ifi_ ## s) show_stat("lu", 8, "received-packets", IFA_STAT(ipackets), link|network, 1); show_stat("lu", 5, "received-errors", IFA_STAT(ierrors), link, 1); show_stat("lu", 5, "dropped-packets", IFA_STAT(iqdrops), link, 1); if (bflag) show_stat("lu", 10, "received-bytes", IFA_STAT(ibytes), link|network, 0); show_stat("lu", 8, "sent-packets", IFA_STAT(opackets), link|network, 1); show_stat("lu", 5, "send-errors", IFA_STAT(oerrors), link, 1); if (bflag) show_stat("lu", 10, "sent-bytes", IFA_STAT(obytes), link|network, 0); show_stat("NRSlu", 5, "collisions", IFA_STAT(collisions), link, 1); if (dflag) show_stat("LSlu", 5, "dropped-packets", IFA_STAT(oqdrops), link, 1); xo_emit("\n"); if (!aflag) { xo_close_instance("interface"); continue; } /* * Print family's multicast addresses. */ xo_open_list("multicast-address"); for (ifma = next_ifma(ifmap, ifa->ifa_name, ifa->ifa_addr->sa_family); ifma != NULL; ifma = next_ifma(ifma, ifa->ifa_name, ifa->ifa_addr->sa_family)) { const char *fmt = NULL; xo_open_instance("multicast-address"); switch (ifma->ifma_addr->sa_family) { case AF_LINK: { struct sockaddr_dl *sdl; sdl = (struct sockaddr_dl *)ifma->ifma_addr; if (sdl->sdl_type != IFT_ETHER && sdl->sdl_type != IFT_FDDI) break; } /* FALLTHROUGH */ case AF_INET: #ifdef INET6 case AF_INET6: #endif /* INET6 */ fmt = routename(ifma->ifma_addr, numeric_addr); break; } if (fmt) { if (Wflag) xo_emit("{P:/%27s }" "{t:address/%-17s/}", "", fmt); else xo_emit("{P:/%25s }" "{t:address/%-17.17s/}", "", fmt); if (ifma->ifma_addr->sa_family == AF_LINK) { xo_emit(" {:received-packets/%8lu}", IFA_STAT(imcasts)); xo_emit("{P:/%*s}", bflag? 17 : 6, ""); xo_emit(" {:sent-packets/%8lu}", IFA_STAT(omcasts)); } xo_emit("\n"); } xo_close_instance("multicast-address"); ifma = ifma->ifma_next; } xo_close_list("multicast-address"); xo_close_instance("interface"); } xo_close_list("interface"); freeifaddrs(ifap); if (aflag) freeifmaddrs(ifmap); } struct iftot { u_long ift_ip; /* input packets */ u_long ift_ie; /* input errors */ u_long ift_id; /* input drops */ u_long ift_op; /* output packets */ u_long ift_oe; /* output errors */ u_long ift_od; /* output drops */ u_long ift_co; /* collisions */ u_long ift_ib; /* input bytes */ u_long ift_ob; /* output bytes */ }; /* * Obtain stats for interface(s). */ static void fill_iftot(struct iftot *st) { struct ifaddrs *ifap, *ifa; bool found = false; if (getifaddrs(&ifap) != 0) xo_err(EX_OSERR, "getifaddrs"); bzero(st, sizeof(*st)); for (ifa = ifap; ifa; ifa = ifa->ifa_next) { if (ifa->ifa_addr->sa_family != AF_LINK) continue; if (interface) { if (strcmp(ifa->ifa_name, interface) == 0) found = true; else continue; } st->ift_ip += IFA_STAT(ipackets); st->ift_ie += IFA_STAT(ierrors); st->ift_id += IFA_STAT(iqdrops); st->ift_ib += IFA_STAT(ibytes); st->ift_op += IFA_STAT(opackets); st->ift_oe += IFA_STAT(oerrors); st->ift_od += IFA_STAT(oqdrops); st->ift_ob += IFA_STAT(obytes); st->ift_co += IFA_STAT(collisions); } if (interface && found == false) xo_err(EX_DATAERR, "interface %s not found", interface); freeifaddrs(ifap); } /* * Set a flag to indicate that a signal from the periodic itimer has been * caught. */ static sig_atomic_t signalled; static void catchalarm(int signo __unused) { signalled = true; } /* * Print a running summary of interface statistics. * Repeat display every interval seconds, showing statistics * collected over that interval. Assumes that interval is non-zero. * First line printed at top of screen is always cumulative. */ static void sidewaysintpr(void) { struct iftot ift[2], *new, *old; struct itimerval interval_it; int oldmask, line; new = &ift[0]; old = &ift[1]; fill_iftot(old); (void)signal(SIGALRM, catchalarm); signalled = false; interval_it.it_interval.tv_sec = interval; interval_it.it_interval.tv_usec = 0; interval_it.it_value = interval_it.it_interval; setitimer(ITIMER_REAL, &interval_it, NULL); xo_open_list("interface-statistics"); banner: xo_emit("{T:/%17s} {T:/%14s} {T:/%16s}\n", "input", interface != NULL ? interface : "(Total)", "output"); xo_emit("{T:/%10s} {T:/%5s} {T:/%5s} {T:/%10s} {T:/%10s} {T:/%5s} " "{T:/%10s} {T:/%5s}", "packets", "errs", "idrops", "bytes", "packets", "errs", "bytes", "colls"); if (dflag) xo_emit(" {T:/%5.5s}", "drops"); xo_emit("\n"); xo_flush(); line = 0; loop: if ((noutputs != 0) && (--noutputs == 0)) { xo_close_list("interface-statistics"); return; } oldmask = sigblock(sigmask(SIGALRM)); while (!signalled) sigpause(0); signalled = false; sigsetmask(oldmask); line++; fill_iftot(new); xo_open_instance("stats"); show_stat("lu", 10, "received-packets", new->ift_ip - old->ift_ip, 1, 1); show_stat("lu", 5, "received-errors", new->ift_ie - old->ift_ie, 1, 1); show_stat("lu", 5, "dropped-packets", new->ift_id - old->ift_id, 1, 1); show_stat("lu", 10, "received-bytes", new->ift_ib - old->ift_ib, 1, 0); show_stat("lu", 10, "sent-packets", new->ift_op - old->ift_op, 1, 1); show_stat("lu", 5, "send-errors", new->ift_oe - old->ift_oe, 1, 1); show_stat("lu", 10, "sent-bytes", new->ift_ob - old->ift_ob, 1, 0); show_stat("NRSlu", 5, "collisions", new->ift_co - old->ift_co, 1, 1); if (dflag) show_stat("LSlu", 5, "dropped-packets", new->ift_od - old->ift_od, 1, 1); xo_close_instance("stats"); xo_emit("\n"); xo_flush(); if (new == &ift[0]) { new = &ift[1]; old = &ift[0]; } else { new = &ift[0]; old = &ift[1]; } if (line == 21) goto banner; else goto loop; /* NOTREACHED */ } Index: projects/clang400-import/usr.bin/netstat/inet.c =================================================================== --- projects/clang400-import/usr.bin/netstat/inet.c (revision 311459) +++ projects/clang400-import/usr.bin/netstat/inet.c (revision 311460) @@ -1,1488 +1,1494 @@ /*- * Copyright (c) 1983, 1988, 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. */ #if 0 #ifndef lint static char sccsid[] = "@(#)inet.c 8.5 (Berkeley) 5/24/95"; #endif /* not lint */ #endif #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef INET6 #include #endif /* INET6 */ #include #include #include #include #include #include #include #include #include #define TCPSTATES #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "netstat.h" #include "nl_defs.h" -char *inetname(struct in_addr *); void inetprint(const char *, struct in_addr *, int, const char *, int, const int); #ifdef INET6 static int udp_done, tcp_done, sdp_done; #endif /* INET6 */ static int pcblist_sysctl(int proto, const char *name, char **bufp, int istcp __unused) { const char *mibvar; char *buf; size_t len; switch (proto) { case IPPROTO_TCP: mibvar = "net.inet.tcp.pcblist"; break; case IPPROTO_UDP: mibvar = "net.inet.udp.pcblist"; break; case IPPROTO_DIVERT: mibvar = "net.inet.divert.pcblist"; break; default: mibvar = "net.inet.raw.pcblist"; break; } if (strncmp(name, "sdp", 3) == 0) mibvar = "net.inet.sdp.pcblist"; len = 0; if (sysctlbyname(mibvar, 0, &len, 0, 0) < 0) { if (errno != ENOENT) xo_warn("sysctl: %s", mibvar); return (0); } if ((buf = malloc(len)) == NULL) { xo_warnx("malloc %lu bytes", (u_long)len); return (0); } if (sysctlbyname(mibvar, buf, &len, 0, 0) < 0) { xo_warn("sysctl: %s", mibvar); free(buf); return (0); } *bufp = buf; return (1); } /* * Copied directly from uipc_socket2.c. We leave out some fields that are in * nested structures that aren't used to avoid extra work. */ static void sbtoxsockbuf(struct sockbuf *sb, struct xsockbuf *xsb) { xsb->sb_cc = sb->sb_ccc; xsb->sb_hiwat = sb->sb_hiwat; xsb->sb_mbcnt = sb->sb_mbcnt; xsb->sb_mcnt = sb->sb_mcnt; xsb->sb_ccnt = sb->sb_ccnt; xsb->sb_mbmax = sb->sb_mbmax; xsb->sb_lowat = sb->sb_lowat; xsb->sb_flags = sb->sb_flags; xsb->sb_timeo = sb->sb_timeo; } int sotoxsocket(struct socket *so, struct xsocket *xso) { struct protosw proto; struct domain domain; bzero(xso, sizeof *xso); xso->xso_len = sizeof *xso; xso->xso_so = so; xso->so_type = so->so_type; xso->so_options = so->so_options; xso->so_linger = so->so_linger; xso->so_state = so->so_state; xso->so_pcb = so->so_pcb; if (kread((uintptr_t)so->so_proto, &proto, sizeof(proto)) != 0) return (-1); xso->xso_protocol = proto.pr_protocol; if (kread((uintptr_t)proto.pr_domain, &domain, sizeof(domain)) != 0) return (-1); xso->xso_family = domain.dom_family; xso->so_qlen = so->so_qlen; xso->so_incqlen = so->so_incqlen; xso->so_qlimit = so->so_qlimit; xso->so_timeo = so->so_timeo; xso->so_error = so->so_error; xso->so_oobmark = so->so_oobmark; sbtoxsockbuf(&so->so_snd, &xso->so_snd); sbtoxsockbuf(&so->so_rcv, &xso->so_rcv); return (0); } static int pcblist_kvm(u_long off, char **bufp, int istcp) { struct inpcbinfo pcbinfo; struct inpcbhead listhead; struct inpcb *inp; struct xinpcb xi; struct xinpgen xig; struct xtcpcb xt; struct socket so; struct xsocket *xso; char *buf, *p; size_t len; if (off == 0) return (0); kread(off, &pcbinfo, sizeof(pcbinfo)); if (istcp) len = 2 * sizeof(xig) + (pcbinfo.ipi_count + pcbinfo.ipi_count / 8) * sizeof(struct xtcpcb); else len = 2 * sizeof(xig) + (pcbinfo.ipi_count + pcbinfo.ipi_count / 8) * sizeof(struct xinpcb); if ((buf = malloc(len)) == NULL) { xo_warnx("malloc %lu bytes", (u_long)len); return (0); } p = buf; #define COPYOUT(obj, size) do { \ if (len < (size)) { \ xo_warnx("buffer size exceeded"); \ goto fail; \ } \ bcopy((obj), p, (size)); \ len -= (size); \ p += (size); \ } while (0) #define KREAD(off, buf, len) do { \ if (kread((uintptr_t)(off), (buf), (len)) != 0) \ goto fail; \ } while (0) /* Write out header. */ xig.xig_len = sizeof xig; xig.xig_count = pcbinfo.ipi_count; xig.xig_gen = pcbinfo.ipi_gencnt; xig.xig_sogen = 0; COPYOUT(&xig, sizeof xig); /* Walk the PCB list. */ xt.xt_len = sizeof xt; xi.xi_len = sizeof xi; if (istcp) xso = &xt.xt_socket; else xso = &xi.xi_socket; KREAD(pcbinfo.ipi_listhead, &listhead, sizeof(listhead)); LIST_FOREACH(inp, &listhead, inp_list) { if (istcp) { KREAD(inp, &xt.xt_inp, sizeof(*inp)); inp = &xt.xt_inp; } else { KREAD(inp, &xi.xi_inp, sizeof(*inp)); inp = &xi.xi_inp; } if (inp->inp_gencnt > pcbinfo.ipi_gencnt) continue; if (istcp) { if (inp->inp_ppcb == NULL) bzero(&xt.xt_tp, sizeof xt.xt_tp); else if (inp->inp_flags & INP_TIMEWAIT) { bzero(&xt.xt_tp, sizeof xt.xt_tp); xt.xt_tp.t_state = TCPS_TIME_WAIT; } else KREAD(inp->inp_ppcb, &xt.xt_tp, sizeof xt.xt_tp); } if (inp->inp_socket) { KREAD(inp->inp_socket, &so, sizeof(so)); if (sotoxsocket(&so, xso) != 0) goto fail; } else { bzero(xso, sizeof(*xso)); if (istcp) xso->xso_protocol = IPPROTO_TCP; } if (istcp) COPYOUT(&xt, sizeof xt); else COPYOUT(&xi, sizeof xi); } /* Reread the pcbinfo and write out the footer. */ kread(off, &pcbinfo, sizeof(pcbinfo)); xig.xig_count = pcbinfo.ipi_count; xig.xig_gen = pcbinfo.ipi_gencnt; COPYOUT(&xig, sizeof xig); *bufp = buf; return (1); fail: free(buf); return (0); #undef COPYOUT #undef KREAD } /* * Print a summary of connections related to an Internet * protocol. For TCP, also give state of connection. * Listening processes (aflag) are suppressed unless the * -a (all) flag is specified. */ void protopr(u_long off, const char *name, int af1, int proto) { int istcp; static int first = 1; char *buf; const char *vchar; struct tcpcb *tp = NULL; struct inpcb *inp; struct xinpgen *xig, *oxig; struct xsocket *so; struct xtcp_timer *timer; istcp = 0; switch (proto) { case IPPROTO_TCP: #ifdef INET6 if (strncmp(name, "sdp", 3) != 0) { if (tcp_done != 0) return; else tcp_done = 1; } else { if (sdp_done != 0) return; else sdp_done = 1; } #endif istcp = 1; break; case IPPROTO_UDP: #ifdef INET6 if (udp_done != 0) return; else udp_done = 1; #endif break; } if (live) { if (!pcblist_sysctl(proto, name, &buf, istcp)) return; } else { if (!pcblist_kvm(off, &buf, istcp)) return; } oxig = xig = (struct xinpgen *)buf; for (xig = (struct xinpgen *)((char *)xig + xig->xig_len); xig->xig_len > sizeof(struct xinpgen); xig = (struct xinpgen *)((char *)xig + xig->xig_len)) { if (istcp) { timer = &((struct xtcpcb *)xig)->xt_timer; tp = &((struct xtcpcb *)xig)->xt_tp; inp = &((struct xtcpcb *)xig)->xt_inp; so = &((struct xtcpcb *)xig)->xt_socket; } else { inp = &((struct xinpcb *)xig)->xi_inp; so = &((struct xinpcb *)xig)->xi_socket; timer = NULL; } /* Ignore sockets for protocols other than the desired one. */ if (so->xso_protocol != proto) continue; /* Ignore PCBs which were freed during copyout. */ if (inp->inp_gencnt > oxig->xig_gen) continue; if ((af1 == AF_INET && (inp->inp_vflag & INP_IPV4) == 0) #ifdef INET6 || (af1 == AF_INET6 && (inp->inp_vflag & INP_IPV6) == 0) #endif /* INET6 */ || (af1 == AF_UNSPEC && ((inp->inp_vflag & INP_IPV4) == 0 #ifdef INET6 && (inp->inp_vflag & INP_IPV6) == 0 #endif /* INET6 */ )) ) continue; if (!aflag && ( (istcp && tp->t_state == TCPS_LISTEN) || (af1 == AF_INET && inet_lnaof(inp->inp_laddr) == INADDR_ANY) #ifdef INET6 || (af1 == AF_INET6 && IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) #endif /* INET6 */ || (af1 == AF_UNSPEC && (((inp->inp_vflag & INP_IPV4) != 0 && inet_lnaof(inp->inp_laddr) == INADDR_ANY) #ifdef INET6 || ((inp->inp_vflag & INP_IPV6) != 0 && IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) #endif )) )) continue; if (first) { if (!Lflag) { xo_emit("Active Internet connections"); if (aflag) xo_emit(" (including servers)"); } else xo_emit( "Current listen queue sizes (qlen/incqlen/maxqlen)"); xo_emit("\n"); if (Aflag) xo_emit("{T:/%-*s} ", 2 * (int)sizeof(void *), "Tcpcb"); if (Lflag) xo_emit((Aflag && !Wflag) ? "{T:/%-5.5s} {T:/%-32.32s} {T:/%-18.18s}" : ((!Wflag || af1 == AF_INET) ? "{T:/%-5.5s} {T:/%-32.32s} {T:/%-22.22s}" : "{T:/%-5.5s} {T:/%-32.32s} {T:/%-45.45s}"), "Proto", "Listen", "Local Address"); else if (Tflag) xo_emit((Aflag && !Wflag) ? "{T:/%-5.5s} {T:/%-6.6s} {T:/%-6.6s} {T:/%-6.6s} {T:/%-18.18s} {T:/%s}" : ((!Wflag || af1 == AF_INET) ? "{T:/%-5.5s} {T:/%-6.6s} {T:/%-6.6s} {T:/%-6.6s} {T:/%-22.22s} {T:/%s}" : "{T:/%-5.5s} {T:/%-6.6s} {T:/%-6.6s} {T:/%-6.6s} {T:/%-45.45s} {T:/%s}"), "Proto", "Rexmit", "OOORcv", "0-win", "Local Address", "Foreign Address"); else { xo_emit((Aflag && !Wflag) ? "{T:/%-5.5s} {T:/%-6.6s} {T:/%-6.6s} {T:/%-18.18s} {T:/%-18.18s}" : ((!Wflag || af1 == AF_INET) ? "{T:/%-5.5s} {T:/%-6.6s} {T:/%-6.6s} {T:/%-22.22s} {T:/%-22.22s}" : "{T:/%-5.5s} {T:/%-6.6s} {T:/%-6.6s} {T:/%-45.45s} {T:/%-45.45s}"), "Proto", "Recv-Q", "Send-Q", "Local Address", "Foreign Address"); if (!xflag && !Rflag) xo_emit(" (state)"); } if (xflag) { xo_emit(" {T:/%-6.6s} {T:/%-6.6s} {T:/%-6.6s} " "{T:/%-6.6s} {T:/%-6.6s} {T:/%-6.6s} " "{T:/%-6.6s} {T:/%-6.6s} {T:/%-6.6s} " "{T:/%-6.6s} {T:/%-6.6s} {T:/%-6.6s}", "R-MBUF", "S-MBUF", "R-CLUS", "S-CLUS", "R-HIWA", "S-HIWA", "R-LOWA", "S-LOWA", "R-BCNT", "S-BCNT", "R-BMAX", "S-BMAX"); xo_emit(" {T:/%7.7s} {T:/%7.7s} {T:/%7.7s} " "{T:/%7.7s} {T:/%7.7s} {T:/%7.7s}", "rexmt", "persist", "keep", "2msl", "delack", "rcvtime"); } else if (Rflag) { xo_emit(" {T:/%8.8s} {T:/%5.5s}", "flowid", "ftype"); } xo_emit("\n"); first = 0; } if (Lflag && so->so_qlimit == 0) continue; xo_open_instance("socket"); if (Aflag) { if (istcp) xo_emit("{q:address/%*lx} ", 2 * (int)sizeof(void *), (u_long)inp->inp_ppcb); else xo_emit("{q:adddress/%*lx} ", 2 * (int)sizeof(void *), (u_long)so->so_pcb); } #ifdef INET6 if ((inp->inp_vflag & INP_IPV6) != 0) vchar = ((inp->inp_vflag & INP_IPV4) != 0) ? "46" : "6"; else #endif vchar = ((inp->inp_vflag & INP_IPV4) != 0) ? "4" : ""; if (istcp && (tp->t_flags & TF_TOE) != 0) xo_emit("{:protocol/%-3.3s%-2.2s/%s%s} ", "toe", vchar); else xo_emit("{:protocol/%-3.3s%-2.2s/%s%s} ", name, vchar); if (Lflag) { char buf1[33]; snprintf(buf1, sizeof buf1, "%u/%u/%u", so->so_qlen, so->so_incqlen, so->so_qlimit); xo_emit("{:listen-queue-sizes/%-32.32s} ", buf1); } else if (Tflag) { if (istcp) xo_emit("{:sent-retransmit-packets/%6u} " "{:received-out-of-order-packets/%6u} " "{:sent-zero-window/%6u} ", tp->t_sndrexmitpack, tp->t_rcvoopack, tp->t_sndzerowin); else xo_emit("{P:/%21s}", ""); } else { xo_emit("{:receive-bytes-waiting/%6u} " "{:send-bytes-waiting/%6u} ", so->so_rcv.sb_cc, so->so_snd.sb_cc); } if (numeric_port) { if (inp->inp_vflag & INP_IPV4) { inetprint("local", &inp->inp_laddr, (int)inp->inp_lport, name, 1, af1); if (!Lflag) inetprint("remote", &inp->inp_faddr, (int)inp->inp_fport, name, 1, af1); } #ifdef INET6 else if (inp->inp_vflag & INP_IPV6) { inet6print("local", &inp->in6p_laddr, (int)inp->inp_lport, name, 1); if (!Lflag) inet6print("remote", &inp->in6p_faddr, (int)inp->inp_fport, name, 1); } /* else nothing printed now */ #endif /* INET6 */ } else if (inp->inp_flags & INP_ANONPORT) { if (inp->inp_vflag & INP_IPV4) { inetprint("local", &inp->inp_laddr, (int)inp->inp_lport, name, 1, af1); if (!Lflag) inetprint("remote", &inp->inp_faddr, (int)inp->inp_fport, name, 0, af1); } #ifdef INET6 else if (inp->inp_vflag & INP_IPV6) { inet6print("local", &inp->in6p_laddr, (int)inp->inp_lport, name, 1); if (!Lflag) inet6print("remote", &inp->in6p_faddr, (int)inp->inp_fport, name, 0); } /* else nothing printed now */ #endif /* INET6 */ } else { if (inp->inp_vflag & INP_IPV4) { inetprint("local", &inp->inp_laddr, (int)inp->inp_lport, name, 0, af1); if (!Lflag) inetprint("remote", &inp->inp_faddr, (int)inp->inp_fport, name, inp->inp_lport != inp->inp_fport, af1); } #ifdef INET6 else if (inp->inp_vflag & INP_IPV6) { inet6print("local", &inp->in6p_laddr, (int)inp->inp_lport, name, 0); if (!Lflag) inet6print("remote", &inp->in6p_faddr, (int)inp->inp_fport, name, inp->inp_lport != inp->inp_fport); } /* else nothing printed now */ #endif /* INET6 */ } if (xflag) { xo_emit("{:receive-mbufs/%6u} {:send-mbufs/%6u} " "{:receive-clusters/%6u} {:send-clusters/%6u} " "{:receive-high-water/%6u} {:send-high-water/%6u} " "{:receive-low-water/%6u} {:send-low-water/%6u} " "{:receive-mbuf-bytes/%6u} {:send-mbuf-bytes/%6u} " "{:receive-mbuf-bytes-max/%6u} " "{:send-mbuf-bytes-max/%6u}", so->so_rcv.sb_mcnt, so->so_snd.sb_mcnt, so->so_rcv.sb_ccnt, so->so_snd.sb_ccnt, so->so_rcv.sb_hiwat, so->so_snd.sb_hiwat, so->so_rcv.sb_lowat, so->so_snd.sb_lowat, so->so_rcv.sb_mbcnt, so->so_snd.sb_mbcnt, so->so_rcv.sb_mbmax, so->so_snd.sb_mbmax); if (timer != NULL) xo_emit(" {:retransmit-timer/%4d.%02d} " "{:persist-timer/%4d.%02d} " "{:keepalive-timer/%4d.%02d} " "{:msl2-timer/%4d.%02d} " "{:delay-ack-timer/%4d.%02d} " "{:inactivity-timer/%4d.%02d}", timer->tt_rexmt / 1000, (timer->tt_rexmt % 1000) / 10, timer->tt_persist / 1000, (timer->tt_persist % 1000) / 10, timer->tt_keep / 1000, (timer->tt_keep % 1000) / 10, timer->tt_2msl / 1000, (timer->tt_2msl % 1000) / 10, timer->tt_delack / 1000, (timer->tt_delack % 1000) / 10, timer->t_rcvtime / 1000, (timer->t_rcvtime % 1000) / 10); } if (istcp && !Lflag && !xflag && !Tflag && !Rflag) { if (tp->t_state < 0 || tp->t_state >= TCP_NSTATES) xo_emit("{:tcp-state/%d}", tp->t_state); else { xo_emit("{:tcp-state/%s}", tcpstates[tp->t_state]); #if defined(TF_NEEDSYN) && defined(TF_NEEDFIN) /* Show T/TCP `hidden state' */ if (tp->t_flags & (TF_NEEDSYN|TF_NEEDFIN)) xo_emit("{:need-syn-or-fin/*}"); #endif /* defined(TF_NEEDSYN) && defined(TF_NEEDFIN) */ } } if (Rflag) { /* XXX: is this right Alfred */ xo_emit(" {:flow-id/%08x} {:flow-type/%5d}", inp->inp_flowid, inp->inp_flowtype); } xo_emit("\n"); xo_close_instance("socket"); } if (xig != oxig && xig->xig_gen != oxig->xig_gen) { if (oxig->xig_count > xig->xig_count) { xo_emit("Some {d:lost/%s} sockets may have been " "deleted.\n", name); } else if (oxig->xig_count < xig->xig_count) { xo_emit("Some {d:created/%s} sockets may have been " "created.\n", name); } else { xo_emit("Some {d:changed/%s} sockets may have been " "created or deleted.\n", name); } } free(buf); } /* * Dump TCP statistics structure. */ void tcp_stats(u_long off, const char *name, int af1 __unused, int proto __unused) { struct tcpstat tcpstat; uint64_t tcps_states[TCP_NSTATES]; #ifdef INET6 if (tcp_done != 0) return; else tcp_done = 1; #endif if (fetch_stats("net.inet.tcp.stats", off, &tcpstat, sizeof(tcpstat), kread_counters) != 0) return; if (fetch_stats_ro("net.inet.tcp.states", nl[N_TCPS_STATES].n_value, &tcps_states, sizeof(tcps_states), kread_counters) != 0) return; xo_open_container("tcp"); xo_emit("{T:/%s}:\n", name); #define p(f, m) if (tcpstat.f || sflag <= 1) \ xo_emit(m, (uintmax_t )tcpstat.f, plural(tcpstat.f)) #define p1a(f, m) if (tcpstat.f || sflag <= 1) \ xo_emit(m, (uintmax_t )tcpstat.f) #define p2(f1, f2, m) if (tcpstat.f1 || tcpstat.f2 || sflag <= 1) \ xo_emit(m, (uintmax_t )tcpstat.f1, plural(tcpstat.f1), \ (uintmax_t )tcpstat.f2, plural(tcpstat.f2)) #define p2a(f1, f2, m) if (tcpstat.f1 || tcpstat.f2 || sflag <= 1) \ xo_emit(m, (uintmax_t )tcpstat.f1, plural(tcpstat.f1), \ (uintmax_t )tcpstat.f2) #define p3(f, m) if (tcpstat.f || sflag <= 1) \ xo_emit(m, (uintmax_t )tcpstat.f, pluralies(tcpstat.f)) p(tcps_sndtotal, "\t{:sent-packets/%ju} {N:/packet%s sent}\n"); p2(tcps_sndpack,tcps_sndbyte, "\t\t{:sent-data-packets/%ju} " "{N:/data packet%s} ({:sent-data-bytes/%ju} {N:/byte%s})\n"); p2(tcps_sndrexmitpack, tcps_sndrexmitbyte, "\t\t" "{:sent-retransmitted-packets/%ju} {N:/data packet%s} " "({:sent-retransmitted-bytes/%ju} {N:/byte%s}) " "{N:retransmitted}\n"); p(tcps_sndrexmitbad, "\t\t" "{:sent-unnecessary-retransmitted-packets/%ju} " "{N:/data packet%s unnecessarily retransmitted}\n"); p(tcps_mturesent, "\t\t{:sent-resends-by-mtu-discovery/%ju} " "{N:/resend%s initiated by MTU discovery}\n"); p2a(tcps_sndacks, tcps_delack, "\t\t{:sent-ack-only-packets/%ju} " "{N:/ack-only packet%s/} ({:sent-packets-delayed/%ju} " "{N:delayed})\n"); p(tcps_sndurg, "\t\t{:sent-urg-only-packets/%ju} " "{N:/URG only packet%s}\n"); p(tcps_sndprobe, "\t\t{:sent-window-probe-packets/%ju} " "{N:/window probe packet%s}\n"); p(tcps_sndwinup, "\t\t{:sent-window-update-packets/%ju} " "{N:/window update packet%s}\n"); p(tcps_sndctrl, "\t\t{:sent-control-packets/%ju} " "{N:/control packet%s}\n"); p(tcps_rcvtotal, "\t{:received-packets/%ju} " "{N:/packet%s received}\n"); p2(tcps_rcvackpack, tcps_rcvackbyte, "\t\t" "{:received-ack-packets/%ju} {N:/ack%s} " "{N:(for} {:received-ack-bytes/%ju} {N:/byte%s})\n"); p(tcps_rcvdupack, "\t\t{:received-duplicate-acks/%ju} " "{N:/duplicate ack%s}\n"); p(tcps_rcvacktoomuch, "\t\t{:received-acks-for-unsent-data/%ju} " "{N:/ack%s for unsent data}\n"); p2(tcps_rcvpack, tcps_rcvbyte, "\t\t" "{:received-in-sequence-packets/%ju} {N:/packet%s} " "({:received-in-sequence-bytes/%ju} {N:/byte%s}) " "{N:received in-sequence}\n"); p2(tcps_rcvduppack, tcps_rcvdupbyte, "\t\t" "{:received-completely-duplicate-packets/%ju} " "{N:/completely duplicate packet%s} " "({:received-completely-duplicate-bytes/%ju} {N:/byte%s})\n"); p(tcps_pawsdrop, "\t\t{:received-old-duplicate-packets/%ju} " "{N:/old duplicate packet%s}\n"); p2(tcps_rcvpartduppack, tcps_rcvpartdupbyte, "\t\t" "{:received-some-duplicate-packets/%ju} " "{N:/packet%s with some dup. data} " "({:received-some-duplicate-bytes/%ju} {N:/byte%s duped/})\n"); p2(tcps_rcvoopack, tcps_rcvoobyte, "\t\t{:received-out-of-order/%ju} " "{N:/out-of-order packet%s} " "({:received-out-of-order-bytes/%ju} {N:/byte%s})\n"); p2(tcps_rcvpackafterwin, tcps_rcvbyteafterwin, "\t\t" "{:received-after-window-packets/%ju} {N:/packet%s} " "({:received-after-window-bytes/%ju} {N:/byte%s}) " "{N:of data after window}\n"); p(tcps_rcvwinprobe, "\t\t{:received-window-probes/%ju} " "{N:/window probe%s}\n"); p(tcps_rcvwinupd, "\t\t{:receive-window-update-packets/%ju} " "{N:/window update packet%s}\n"); p(tcps_rcvafterclose, "\t\t{:received-after-close-packets/%ju} " "{N:/packet%s received after close}\n"); p(tcps_rcvbadsum, "\t\t{:discard-bad-checksum/%ju} " "{N:/discarded for bad checksum%s}\n"); p(tcps_rcvbadoff, "\t\t{:discard-bad-header-offset/%ju} " "{N:/discarded for bad header offset field%s}\n"); p1a(tcps_rcvshort, "\t\t{:discard-too-short/%ju} " "{N:discarded because packet too short}\n"); p1a(tcps_rcvmemdrop, "\t\t{:discard-memory-problems/%ju} " "{N:discarded due to memory problems}\n"); p(tcps_connattempt, "\t{:connection-requests/%ju} " "{N:/connection request%s}\n"); p(tcps_accepts, "\t{:connections-accepts/%ju} " "{N:/connection accept%s}\n"); p(tcps_badsyn, "\t{:bad-connection-attempts/%ju} " "{N:/bad connection attempt%s}\n"); p(tcps_listendrop, "\t{:listen-queue-overflows/%ju} " "{N:/listen queue overflow%s}\n"); p(tcps_badrst, "\t{:ignored-in-window-resets/%ju} " "{N:/ignored RSTs in the window%s}\n"); p(tcps_connects, "\t{:connections-established/%ju} " "{N:/connection%s established (including accepts)}\n"); p(tcps_usedrtt, "\t\t{:connections-hostcache-rtt/%ju} " "{N:/time%s used RTT from hostcache}\n"); p(tcps_usedrttvar, "\t\t{:connections-hostcache-rttvar/%ju} " "{N:/time%s used RTT variance from hostcache}\n"); p(tcps_usedssthresh, "\t\t{:connections-hostcache-ssthresh/%ju} " "{N:/time%s used slow-start threshold from hostcache}\n"); p2(tcps_closed, tcps_drops, "\t{:connections-closed/%ju} " "{N:/connection%s closed (including} " "{:connection-drops/%ju} {N:/drop%s})\n"); p(tcps_cachedrtt, "\t\t{:connections-updated-rtt-on-close/%ju} " "{N:/connection%s updated cached RTT on close}\n"); p(tcps_cachedrttvar, "\t\t" "{:connections-updated-variance-on-close/%ju} " "{N:/connection%s updated cached RTT variance on close}\n"); p(tcps_cachedssthresh, "\t\t" "{:connections-updated-ssthresh-on-close/%ju} " "{N:/connection%s updated cached ssthresh on close}\n"); p(tcps_conndrops, "\t{:embryonic-connections-dropped/%ju} " "{N:/embryonic connection%s dropped}\n"); p2(tcps_rttupdated, tcps_segstimed, "\t{:segments-updated-rtt/%ju} " "{N:/segment%s updated rtt (of} " "{:segment-update-attempts/%ju} {N:/attempt%s})\n"); p(tcps_rexmttimeo, "\t{:retransmit-timeouts/%ju} " "{N:/retransmit timeout%s}\n"); p(tcps_timeoutdrop, "\t\t" "{:connections-dropped-by-retransmit-timeout/%ju} " "{N:/connection%s dropped by rexmit timeout}\n"); p(tcps_persisttimeo, "\t{:persist-timeout/%ju} " "{N:/persist timeout%s}\n"); p(tcps_persistdrop, "\t\t" "{:connections-dropped-by-persist-timeout/%ju} " "{N:/connection%s dropped by persist timeout}\n"); p(tcps_finwait2_drops, "\t" "{:connections-dropped-by-finwait2-timeout/%ju} " "{N:/Connection%s (fin_wait_2) dropped because of timeout}\n"); p(tcps_keeptimeo, "\t{:keepalive-timeout/%ju} " "{N:/keepalive timeout%s}\n"); p(tcps_keepprobe, "\t\t{:keepalive-probes/%ju} " "{N:/keepalive probe%s sent}\n"); p(tcps_keepdrops, "\t\t{:connections-dropped-by-keepalives/%ju} " "{N:/connection%s dropped by keepalive}\n"); p(tcps_predack, "\t{:ack-header-predictions/%ju} " "{N:/correct ACK header prediction%s}\n"); p(tcps_preddat, "\t{:data-packet-header-predictions/%ju} " "{N:/correct data packet header prediction%s}\n"); xo_open_container("syncache"); p3(tcps_sc_added, "\t{:entries-added/%ju} " "{N:/syncache entr%s added}\n"); p1a(tcps_sc_retransmitted, "\t\t{:retransmitted/%ju} " "{N:/retransmitted}\n"); p1a(tcps_sc_dupsyn, "\t\t{:duplicates/%ju} {N:/dupsyn}\n"); p1a(tcps_sc_dropped, "\t\t{:dropped/%ju} {N:/dropped}\n"); p1a(tcps_sc_completed, "\t\t{:completed/%ju} {N:/completed}\n"); p1a(tcps_sc_bucketoverflow, "\t\t{:bucket-overflow/%ju} " "{N:/bucket overflow}\n"); p1a(tcps_sc_cacheoverflow, "\t\t{:cache-overflow/%ju} " "{N:/cache overflow}\n"); p1a(tcps_sc_reset, "\t\t{:reset/%ju} {N:/reset}\n"); p1a(tcps_sc_stale, "\t\t{:stale/%ju} {N:/stale}\n"); p1a(tcps_sc_aborted, "\t\t{:aborted/%ju} {N:/aborted}\n"); p1a(tcps_sc_badack, "\t\t{:bad-ack/%ju} {N:/badack}\n"); p1a(tcps_sc_unreach, "\t\t{:unreachable/%ju} {N:/unreach}\n"); p(tcps_sc_zonefail, "\t\t{:zone-failures/%ju} {N:/zone failure%s}\n"); p(tcps_sc_sendcookie, "\t{:sent-cookies/%ju} {N:/cookie%s sent}\n"); p(tcps_sc_recvcookie, "\t{:receivd-cookies/%ju} " "{N:/cookie%s received}\n"); xo_close_container("syncache"); xo_open_container("hostcache"); p3(tcps_hc_added, "\t{:entries-added/%ju} " "{N:/hostcache entr%s added}\n"); p1a(tcps_hc_bucketoverflow, "\t\t{:buffer-overflows/%ju} " "{N:/bucket overflow}\n"); xo_close_container("hostcache"); xo_open_container("sack"); p(tcps_sack_recovery_episode, "\t{:recovery-episodes/%ju} " "{N:/SACK recovery episode%s}\n"); p(tcps_sack_rexmits, "\t{:segment-retransmits/%ju} " "{N:/segment rexmit%s in SACK recovery episodes}\n"); p(tcps_sack_rexmit_bytes, "\t{:byte-retransmits/%ju} " "{N:/byte rexmit%s in SACK recovery episodes}\n"); p(tcps_sack_rcv_blocks, "\t{:received-blocks/%ju} " "{N:/SACK option%s (SACK blocks) received}\n"); p(tcps_sack_send_blocks, "\t{:sent-option-blocks/%ju} " "{N:/SACK option%s (SACK blocks) sent}\n"); p1a(tcps_sack_sboverflow, "\t{:scoreboard-overflows/%ju} " "{N:/SACK scoreboard overflow}\n"); xo_close_container("sack"); xo_open_container("ecn"); p(tcps_ecn_ce, "\t{:ce-packets/%ju} " "{N:/packet%s with ECN CE bit set}\n"); p(tcps_ecn_ect0, "\t{:ect0-packets/%ju} " "{N:/packet%s with ECN ECT(0) bit set}\n"); p(tcps_ecn_ect1, "\t{:ect1-packets/%ju} " "{N:/packet%s with ECN ECT(1) bit set}\n"); p(tcps_ecn_shs, "\t{:handshakes/%ju} " "{N:/successful ECN handshake%s}\n"); p(tcps_ecn_rcwnd, "\t{:congestion-reductions/%ju} " "{N:/time%s ECN reduced the congestion window}\n"); #undef p #undef p1a #undef p2 #undef p2a #undef p3 xo_close_container("ecn"); xo_open_container("TCP connection count by state"); xo_emit("{T:/TCP connection count by state}:\n"); for (int i = 0; i < TCP_NSTATES; i++) { /* * XXXGL: is there a way in libxo to use %s * in the "content string" of a format * string? I failed to do that, that's why * a temporary buffer is used to construct * format string for xo_emit(). */ char fmtbuf[80]; if (sflag > 1 && tcps_states[i] == 0) continue; snprintf(fmtbuf, sizeof(fmtbuf), "\t{:%s/%%ju} " "{Np:/connection ,connections} in %s state\n", tcpstates[i], tcpstates[i]); xo_emit(fmtbuf, (uintmax_t )tcps_states[i]); } xo_close_container("TCP connection count by state"); xo_close_container("tcp"); } /* * Dump UDP statistics structure. */ void udp_stats(u_long off, const char *name, int af1 __unused, int proto __unused) { struct udpstat udpstat; uint64_t delivered; #ifdef INET6 if (udp_done != 0) return; else udp_done = 1; #endif if (fetch_stats("net.inet.udp.stats", off, &udpstat, sizeof(udpstat), kread_counters) != 0) return; xo_open_container("udp"); xo_emit("{T:/%s}:\n", name); #define p(f, m) if (udpstat.f || sflag <= 1) \ xo_emit("\t" m, (uintmax_t)udpstat.f, plural(udpstat.f)) #define p1a(f, m) if (udpstat.f || sflag <= 1) \ xo_emit("\t" m, (uintmax_t)udpstat.f) p(udps_ipackets, "{:received-datagrams/%ju} " "{N:/datagram%s received}\n"); p1a(udps_hdrops, "{:dropped-incomplete-headers/%ju} " "{N:/with incomplete header}\n"); p1a(udps_badlen, "{:dropped-bad-data-length/%ju} " "{N:/with bad data length field}\n"); p1a(udps_badsum, "{:dropped-bad-checksum/%ju} " "{N:/with bad checksum}\n"); p1a(udps_nosum, "{:dropped-no-checksum/%ju} " "{N:/with no checksum}\n"); p1a(udps_noport, "{:dropped-no-socket/%ju} " "{N:/dropped due to no socket}\n"); p(udps_noportbcast, "{:dropped-broadcast-multicast/%ju} " "{N:/broadcast\\/multicast datagram%s undelivered}\n"); p1a(udps_fullsock, "{:dropped-full-socket-buffer/%ju} " "{N:/dropped due to full socket buffers}\n"); p1a(udpps_pcbhashmiss, "{:not-for-hashed-pcb/%ju} " "{N:/not for hashed pcb}\n"); delivered = udpstat.udps_ipackets - udpstat.udps_hdrops - udpstat.udps_badlen - udpstat.udps_badsum - udpstat.udps_noport - udpstat.udps_noportbcast - udpstat.udps_fullsock; if (delivered || sflag <= 1) xo_emit("\t{:delivered-packets/%ju} {N:/delivered}\n", (uint64_t)delivered); p(udps_opackets, "{:output-packets/%ju} {N:/datagram%s output}\n"); /* the next statistic is cumulative in udps_noportbcast */ p(udps_filtermcast, "{:multicast-source-filter-matches/%ju} " "{N:/time%s multicast source filter matched}\n"); #undef p #undef p1a xo_close_container("udp"); } /* * Dump CARP statistics structure. */ void carp_stats(u_long off, const char *name, int af1 __unused, int proto __unused) { struct carpstats carpstat; if (fetch_stats("net.inet.carp.stats", off, &carpstat, sizeof(carpstat), kread_counters) != 0) return; xo_open_container(name); xo_emit("{T:/%s}:\n", name); #define p(f, m) if (carpstat.f || sflag <= 1) \ xo_emit(m, (uintmax_t)carpstat.f, plural(carpstat.f)) #define p2(f, m) if (carpstat.f || sflag <= 1) \ xo_emit(m, (uintmax_t)carpstat.f) p(carps_ipackets, "\t{:received-inet-packets/%ju} " "{N:/packet%s received (IPv4)}\n"); p(carps_ipackets6, "\t{:received-inet6-packets/%ju} " "{N:/packet%s received (IPv6)}\n"); p(carps_badttl, "\t\t{:dropped-wrong-ttl/%ju} " "{N:/packet%s discarded for wrong TTL}\n"); p(carps_hdrops, "\t\t{:dropped-short-header/%ju} " "{N:/packet%s shorter than header}\n"); p(carps_badsum, "\t\t{:dropped-bad-checksum/%ju} " "{N:/discarded for bad checksum%s}\n"); p(carps_badver, "\t\t{:dropped-bad-version/%ju} " "{N:/discarded packet%s with a bad version}\n"); p2(carps_badlen, "\t\t{:dropped-short-packet/%ju} " "{N:/discarded because packet too short}\n"); p2(carps_badauth, "\t\t{:dropped-bad-authentication/%ju} " "{N:/discarded for bad authentication}\n"); p2(carps_badvhid, "\t\t{:dropped-bad-vhid/%ju} " "{N:/discarded for bad vhid}\n"); p2(carps_badaddrs, "\t\t{:dropped-bad-address-list/%ju} " "{N:/discarded because of a bad address list}\n"); p(carps_opackets, "\t{:sent-inet-packets/%ju} " "{N:/packet%s sent (IPv4)}\n"); p(carps_opackets6, "\t{:sent-inet6-packets/%ju} " "{N:/packet%s sent (IPv6)}\n"); p2(carps_onomem, "\t\t{:send-failed-memory-error/%ju} " "{N:/send failed due to mbuf memory error}\n"); #if notyet p(carps_ostates, "\t\t{:send-state-updates/%s} " "{N:/state update%s sent}\n"); #endif #undef p #undef p2 xo_close_container(name); } /* * Dump IP statistics structure. */ void ip_stats(u_long off, const char *name, int af1 __unused, int proto __unused) { struct ipstat ipstat; if (fetch_stats("net.inet.ip.stats", off, &ipstat, sizeof(ipstat), kread_counters) != 0) return; xo_open_container(name); xo_emit("{T:/%s}:\n", name); #define p(f, m) if (ipstat.f || sflag <= 1) \ xo_emit(m, (uintmax_t )ipstat.f, plural(ipstat.f)) #define p1a(f, m) if (ipstat.f || sflag <= 1) \ xo_emit(m, (uintmax_t )ipstat.f) p(ips_total, "\t{:received-packets/%ju} " "{N:/total packet%s received}\n"); p(ips_badsum, "\t{:dropped-bad-checksum/%ju} " "{N:/bad header checksum%s}\n"); p1a(ips_toosmall, "\t{:dropped-below-minimum-size/%ju} " "{N:/with size smaller than minimum}\n"); p1a(ips_tooshort, "\t{:dropped-short-packets/%ju} " "{N:/with data size < data length}\n"); p1a(ips_toolong, "\t{:dropped-too-long/%ju} " "{N:/with ip length > max ip packet size}\n"); p1a(ips_badhlen, "\t{:dropped-short-header-length/%ju} " "{N:/with header length < data size}\n"); p1a(ips_badlen, "\t{:dropped-short-data/%ju} " "{N:/with data length < header length}\n"); p1a(ips_badoptions, "\t{:dropped-bad-options/%ju} " "{N:/with bad options}\n"); p1a(ips_badvers, "\t{:dropped-bad-version/%ju} " "{N:/with incorrect version number}\n"); p(ips_fragments, "\t{:received-fragments/%ju} " "{N:/fragment%s received}\n"); p(ips_fragdropped, "\t{:dropped-fragments/%ju} " "{N:/fragment%s dropped (dup or out of space)}\n"); p(ips_fragtimeout, "\t{:dropped-fragments-after-timeout/%ju} " "{N:/fragment%s dropped after timeout}\n"); p(ips_reassembled, "\t{:reassembled-packets/%ju} " "{N:/packet%s reassembled ok}\n"); p(ips_delivered, "\t{:received-local-packets/%ju} " "{N:/packet%s for this host}\n"); p(ips_noproto, "\t{:dropped-unknown-protocol/%ju} " "{N:/packet%s for unknown\\/unsupported protocol}\n"); p(ips_forward, "\t{:forwarded-packets/%ju} " "{N:/packet%s forwarded}"); p(ips_fastforward, " ({:fast-forwarded-packets/%ju} " "{N:/packet%s fast forwarded})"); if (ipstat.ips_forward || sflag <= 1) xo_emit("\n"); p(ips_cantforward, "\t{:packets-cannot-forward/%ju} " "{N:/packet%s not forwardable}\n"); p(ips_notmember, "\t{:received-unknown-multicast-group/%ju} " "{N:/packet%s received for unknown multicast group}\n"); p(ips_redirectsent, "\t{:redirects-sent/%ju} " "{N:/redirect%s sent}\n"); p(ips_localout, "\t{:sent-packets/%ju} " "{N:/packet%s sent from this host}\n"); p(ips_rawout, "\t{:send-packets-fabricated-header/%ju} " "{N:/packet%s sent with fabricated ip header}\n"); p(ips_odropped, "\t{:discard-no-mbufs/%ju} " "{N:/output packet%s dropped due to no bufs, etc.}\n"); p(ips_noroute, "\t{:discard-no-route/%ju} " "{N:/output packet%s discarded due to no route}\n"); p(ips_fragmented, "\t{:sent-fragments/%ju} " "{N:/output datagram%s fragmented}\n"); p(ips_ofragments, "\t{:fragments-created/%ju} " "{N:/fragment%s created}\n"); p(ips_cantfrag, "\t{:discard-cannot-fragment/%ju} " "{N:/datagram%s that can't be fragmented}\n"); p(ips_nogif, "\t{:discard-tunnel-no-gif/%ju} " "{N:/tunneling packet%s that can't find gif}\n"); p(ips_badaddr, "\t{:discard-bad-address/%ju} " "{N:/datagram%s with bad address in header}\n"); #undef p #undef p1a xo_close_container(name); } /* * Dump ARP statistics structure. */ void arp_stats(u_long off, const char *name, int af1 __unused, int proto __unused) { struct arpstat arpstat; if (fetch_stats("net.link.ether.arp.stats", off, &arpstat, sizeof(arpstat), kread_counters) != 0) return; xo_open_container(name); xo_emit("{T:/%s}:\n", name); #define p(f, m) if (arpstat.f || sflag <= 1) \ xo_emit("\t" m, (uintmax_t)arpstat.f, plural(arpstat.f)) #define p2(f, m) if (arpstat.f || sflag <= 1) \ xo_emit("\t" m, (uintmax_t)arpstat.f, pluralies(arpstat.f)) p(txrequests, "{:sent-requests/%ju} {N:/ARP request%s sent}\n"); p2(txreplies, "{:sent-replies/%ju} {N:/ARP repl%s sent}\n"); p(rxrequests, "{:received-requests/%ju} " "{N:/ARP request%s received}\n"); p2(rxreplies, "{:received-replies/%ju} " "{N:/ARP repl%s received}\n"); p(received, "{:received-packers/%ju} " "{N:/ARP packet%s received}\n"); p(dropped, "{:dropped-no-entry/%ju} " "{N:/total packet%s dropped due to no ARP entry}\n"); p(timeouts, "{:entries-timeout/%ju} " "{N:/ARP entry%s timed out}\n"); p(dupips, "{:dropped-duplicate-address/%ju} " "{N:/Duplicate IP%s seen}\n"); #undef p #undef p2 xo_close_container(name); } static const char *icmpnames[ICMP_MAXTYPE + 1] = { "echo reply", /* RFC 792 */ "#1", "#2", "destination unreachable", /* RFC 792 */ "source quench", /* RFC 792 */ "routing redirect", /* RFC 792 */ "#6", "#7", "echo", /* RFC 792 */ "router advertisement", /* RFC 1256 */ "router solicitation", /* RFC 1256 */ "time exceeded", /* RFC 792 */ "parameter problem", /* RFC 792 */ "time stamp", /* RFC 792 */ "time stamp reply", /* RFC 792 */ "information request", /* RFC 792 */ "information request reply", /* RFC 792 */ "address mask request", /* RFC 950 */ "address mask reply", /* RFC 950 */ "#19", "#20", "#21", "#22", "#23", "#24", "#25", "#26", "#27", "#28", "#29", "icmp traceroute", /* RFC 1393 */ "datagram conversion error", /* RFC 1475 */ "mobile host redirect", "IPv6 where-are-you", "IPv6 i-am-here", "mobile registration req", "mobile registration reply", "domain name request", /* RFC 1788 */ "domain name reply", /* RFC 1788 */ "icmp SKIP", "icmp photuris", /* RFC 2521 */ }; /* * Dump ICMP statistics. */ void icmp_stats(u_long off, const char *name, int af1 __unused, int proto __unused) { struct icmpstat icmpstat; size_t len; int i, first; if (fetch_stats("net.inet.icmp.stats", off, &icmpstat, sizeof(icmpstat), kread_counters) != 0) return; xo_open_container(name); xo_emit("{T:/%s}:\n", name); #define p(f, m) if (icmpstat.f || sflag <= 1) \ xo_emit(m, icmpstat.f, plural(icmpstat.f)) #define p1a(f, m) if (icmpstat.f || sflag <= 1) \ xo_emit(m, icmpstat.f) #define p2(f, m) if (icmpstat.f || sflag <= 1) \ xo_emit(m, icmpstat.f, plurales(icmpstat.f)) p(icps_error, "\t{:icmp-calls/%lu} " "{N:/call%s to icmp_error}\n"); p(icps_oldicmp, "\t{:errors-not-from-message/%lu} " "{N:/error%s not generated in response to an icmp message}\n"); for (first = 1, i = 0; i < ICMP_MAXTYPE + 1; i++) { if (icmpstat.icps_outhist[i] != 0) { if (first) { xo_open_list("output-histogram"); xo_emit("\tOutput histogram:\n"); first = 0; } xo_open_instance("output-histogram"); if (icmpnames[i] != NULL) xo_emit("\t\t{k:name/%s}: {:count/%lu}\n", icmpnames[i], icmpstat.icps_outhist[i]); else xo_emit("\t\tunknown ICMP #{k:name/%d}: " "{:count/%lu}\n", i, icmpstat.icps_outhist[i]); xo_close_instance("output-histogram"); } } if (!first) xo_close_list("output-histogram"); p(icps_badcode, "\t{:dropped-bad-code/%lu} " "{N:/message%s with bad code fields}\n"); p(icps_tooshort, "\t{:dropped-too-short/%lu} " "{N:/message%s less than the minimum length}\n"); p(icps_checksum, "\t{:dropped-bad-checksum/%lu} " "{N:/message%s with bad checksum}\n"); p(icps_badlen, "\t{:dropped-bad-length/%lu} " "{N:/message%s with bad length}\n"); p1a(icps_bmcastecho, "\t{:dropped-multicast-echo/%lu} " "{N:/multicast echo requests ignored}\n"); p1a(icps_bmcasttstamp, "\t{:dropped-multicast-timestamp/%lu} " "{N:/multicast timestamp requests ignored}\n"); for (first = 1, i = 0; i < ICMP_MAXTYPE + 1; i++) { if (icmpstat.icps_inhist[i] != 0) { if (first) { xo_open_list("input-histogram"); xo_emit("\tInput histogram:\n"); first = 0; } xo_open_instance("input-histogram"); if (icmpnames[i] != NULL) xo_emit("\t\t{k:name/%s}: {:count/%lu}\n", icmpnames[i], icmpstat.icps_inhist[i]); else xo_emit( "\t\tunknown ICMP #{k:name/%d}: {:count/%lu}\n", i, icmpstat.icps_inhist[i]); xo_close_instance("input-histogram"); } } if (!first) xo_close_list("input-histogram"); p(icps_reflect, "\t{:sent-packets/%lu} " "{N:/message response%s generated}\n"); p2(icps_badaddr, "\t{:discard-invalid-return-address/%lu} " "{N:/invalid return address%s}\n"); p(icps_noroute, "\t{:discard-no-route/%lu} " "{N:/no return route%s}\n"); #undef p #undef p1a #undef p2 if (live) { len = sizeof i; if (sysctlbyname("net.inet.icmp.maskrepl", &i, &len, NULL, 0) < 0) return; xo_emit("\tICMP address mask responses are " "{q:icmp-address-responses/%sabled}\n", i ? "en" : "dis"); } xo_close_container(name); } /* * Dump IGMP statistics structure. */ void igmp_stats(u_long off, const char *name, int af1 __unused, int proto __unused) { struct igmpstat igmpstat; if (fetch_stats("net.inet.igmp.stats", 0, &igmpstat, sizeof(igmpstat), kread) != 0) return; if (igmpstat.igps_version != IGPS_VERSION_3) { xo_warnx("%s: version mismatch (%d != %d)", __func__, igmpstat.igps_version, IGPS_VERSION_3); } if (igmpstat.igps_len != IGPS_VERSION3_LEN) { xo_warnx("%s: size mismatch (%d != %d)", __func__, igmpstat.igps_len, IGPS_VERSION3_LEN); } xo_open_container(name); xo_emit("{T:/%s}:\n", name); #define p64(f, m) if (igmpstat.f || sflag <= 1) \ xo_emit(m, (uintmax_t) igmpstat.f, plural(igmpstat.f)) #define py64(f, m) if (igmpstat.f || sflag <= 1) \ xo_emit(m, (uintmax_t) igmpstat.f, pluralies(igmpstat.f)) p64(igps_rcv_total, "\t{:received-messages/%ju} " "{N:/message%s received}\n"); p64(igps_rcv_tooshort, "\t{:dropped-too-short/%ju} " "{N:/message%s received with too few bytes}\n"); p64(igps_rcv_badttl, "\t{:dropped-wrong-ttl/%ju} " "{N:/message%s received with wrong TTL}\n"); p64(igps_rcv_badsum, "\t{:dropped-bad-checksum/%ju} " "{N:/message%s received with bad checksum}\n"); py64(igps_rcv_v1v2_queries, "\t{:received-membership-queries/%ju} " "{N:/V1\\/V2 membership quer%s received}\n"); py64(igps_rcv_v3_queries, "\t{:received-v3-membership-queries/%ju} " "{N:/V3 membership quer%s received}\n"); py64(igps_rcv_badqueries, "\t{:dropped-membership-queries/%ju} " "{N:/membership quer%s received with invalid field(s)}\n"); py64(igps_rcv_gen_queries, "\t{:received-general-queries/%ju} " "{N:/general quer%s received}\n"); py64(igps_rcv_group_queries, "\t{:received-group-queries/%ju} " "{N:/group quer%s received}\n"); py64(igps_rcv_gsr_queries, "\t{:received-group-source-queries/%ju} " "{N:/group-source quer%s received}\n"); py64(igps_drop_gsr_queries, "\t{:dropped-group-source-queries/%ju} " "{N:/group-source quer%s dropped}\n"); p64(igps_rcv_reports, "\t{:received-membership-requests/%ju} " "{N:/membership report%s received}\n"); p64(igps_rcv_badreports, "\t{:dropped-membership-reports/%ju} " "{N:/membership report%s received with invalid field(s)}\n"); p64(igps_rcv_ourreports, "\t" "{:received-membership-reports-matching/%ju} " "{N:/membership report%s received for groups to which we belong}" "\n"); p64(igps_rcv_nora, "\t{:received-v3-reports-no-router-alert/%ju} " "{N:/V3 report%s received without Router Alert}\n"); p64(igps_snd_reports, "\t{:sent-membership-reports/%ju} " "{N:/membership report%s sent}\n"); #undef p64 #undef py64 xo_close_container(name); } /* * Dump PIM statistics structure. */ void pim_stats(u_long off __unused, const char *name, int af1 __unused, int proto __unused) { struct pimstat pimstat; if (fetch_stats("net.inet.pim.stats", off, &pimstat, sizeof(pimstat), kread_counters) != 0) return; xo_open_container(name); xo_emit("{T:/%s}:\n", name); #define p(f, m) if (pimstat.f || sflag <= 1) \ xo_emit(m, (uintmax_t)pimstat.f, plural(pimstat.f)) #define py(f, m) if (pimstat.f || sflag <= 1) \ xo_emit(m, (uintmax_t)pimstat.f, pimstat.f != 1 ? "ies" : "y") p(pims_rcv_total_msgs, "\t{:received-messages/%ju} " "{N:/message%s received}\n"); p(pims_rcv_total_bytes, "\t{:received-bytes/%ju} " "{N:/byte%s received}\n"); p(pims_rcv_tooshort, "\t{:dropped-too-short/%ju} " "{N:/message%s received with too few bytes}\n"); p(pims_rcv_badsum, "\t{:dropped-bad-checksum/%ju} " "{N:/message%s received with bad checksum}\n"); p(pims_rcv_badversion, "\t{:dropped-bad-version/%ju} " "{N:/message%s received with bad version}\n"); p(pims_rcv_registers_msgs, "\t{:received-data-register-messages/%ju} " "{N:/data register message%s received}\n"); p(pims_rcv_registers_bytes, "\t{:received-data-register-bytes/%ju} " "{N:/data register byte%s received}\n"); p(pims_rcv_registers_wrongiif, "\t" "{:received-data-register-wrong-interface/%ju} " "{N:/data register message%s received on wrong iif}\n"); p(pims_rcv_badregisters, "\t{:received-bad-registers/%ju} " "{N:/bad register%s received}\n"); p(pims_snd_registers_msgs, "\t{:sent-data-register-messages/%ju} " "{N:/data register message%s sent}\n"); p(pims_snd_registers_bytes, "\t{:sent-data-register-bytes/%ju} " "{N:/data register byte%s sent}\n"); #undef p #undef py xo_close_container(name); } /* * Pretty print an Internet address (net address + port). */ void inetprint(const char *container, struct in_addr *in, int port, const char *proto, int num_port, const int af1) { struct servent *sp = 0; char line[80], *cp; int width; + size_t alen, plen; if (container) xo_open_container(container); if (Wflag) - sprintf(line, "%s.", inetname(in)); + snprintf(line, sizeof(line), "%s.", inetname(in)); else - sprintf(line, "%.*s.", (Aflag && !num_port) ? 12 : 16, inetname(in)); - cp = strchr(line, '\0'); + snprintf(line, sizeof(line), "%.*s.", + (Aflag && !num_port) ? 12 : 16, inetname(in)); + alen = strlen(line); + cp = line + alen; if (!num_port && port) sp = getservbyport((int)port, proto); if (sp || port == 0) - sprintf(cp, "%.15s ", sp ? sp->s_name : "*"); + snprintf(cp, sizeof(line) - alen, + "%.15s ", sp ? sp->s_name : "*"); else - sprintf(cp, "%d ", ntohs((u_short)port)); + snprintf(cp, sizeof(line) - alen, + "%d ", ntohs((u_short)port)); width = (Aflag && !Wflag) ? 18 : ((!Wflag || af1 == AF_INET) ? 22 : 45); if (Wflag) xo_emit("{d:target/%-*s} ", width, line); else xo_emit("{d:target/%-*.*s} ", width, width, line); - int alen = cp - line - 1, plen = strlen(cp) - 1; + plen = strlen(cp) - 1; + alen--; xo_emit("{e:address/%*.*s}{e:port/%*.*s}", alen, alen, line, plen, plen, cp); if (container) xo_close_container(container); } /* * Construct an Internet address representation. * If numeric_addr has been supplied, give * numeric value, otherwise try for symbolic name. */ char * inetname(struct in_addr *inp) { char *cp; static char line[MAXHOSTNAMELEN]; struct hostent *hp; struct netent *np; cp = 0; if (!numeric_addr && inp->s_addr != INADDR_ANY) { int net = inet_netof(*inp); int lna = inet_lnaof(*inp); if (lna == INADDR_ANY) { np = getnetbyaddr(net, AF_INET); if (np) cp = np->n_name; } if (cp == NULL) { hp = gethostbyaddr((char *)inp, sizeof (*inp), AF_INET); if (hp) { cp = hp->h_name; trimdomain(cp, strlen(cp)); } } } if (inp->s_addr == INADDR_ANY) strcpy(line, "*"); else if (cp) { strlcpy(line, cp, sizeof(line)); } else { inp->s_addr = ntohl(inp->s_addr); #define C(x) ((u_int)((x) & 0xff)) - sprintf(line, "%u.%u.%u.%u", C(inp->s_addr >> 24), - C(inp->s_addr >> 16), C(inp->s_addr >> 8), C(inp->s_addr)); + snprintf(line, sizeof(line), "%u.%u.%u.%u", + C(inp->s_addr >> 24), C(inp->s_addr >> 16), + C(inp->s_addr >> 8), C(inp->s_addr)); } return (line); } Index: projects/clang400-import/usr.bin/netstat/inet6.c =================================================================== --- projects/clang400-import/usr.bin/netstat/inet6.c (revision 311459) +++ projects/clang400-import/usr.bin/netstat/inet6.c (revision 311460) @@ -1,1348 +1,1352 @@ /* BSDI inet.c,v 2.3 1995/10/24 02:19:29 prb Exp */ /*- * Copyright (c) 1983, 1988, 1993 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #if 0 #ifndef lint static char sccsid[] = "@(#)inet6.c 8.4 (Berkeley) 4/20/94"; #endif /* not lint */ #endif #include __FBSDID("$FreeBSD$"); #ifdef INET6 #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "netstat.h" -char *inet6name(struct in6_addr *); - static char ntop_buf[INET6_ADDRSTRLEN]; static const char *ip6nh[] = { "hop by hop", "ICMP", "IGMP", "#3", "IP", "#5", "TCP", "#7", "#8", "#9", "#10", "#11", "#12", "#13", "#14", "#15", "#16", "UDP", "#18", "#19", "#20", "#21", "IDP", "#23", "#24", "#25", "#26", "#27", "#28", "TP", "#30", "#31", "#32", "#33", "#34", "#35", "#36", "#37", "#38", "#39", "#40", "IP6", "#42", "routing", "fragment", "#45", "#46", "#47", "#48", "#49", "ESP", "AH", "#52", "#53", "#54", "#55", "#56", "#57", "ICMP6", "no next header", "destination option", "#61", "mobility", "#63", "#64", "#65", "#66", "#67", "#68", "#69", "#70", "#71", "#72", "#73", "#74", "#75", "#76", "#77", "#78", "#79", "ISOIP", "#81", "#82", "#83", "#84", "#85", "#86", "#87", "#88", "OSPF", "#80", "#91", "#92", "#93", "#94", "#95", "#96", "Ethernet", "#98", "#99", "#100", "#101", "#102", "PIM", "#104", "#105", "#106", "#107", "#108", "#109", "#110", "#111", "#112", "#113", "#114", "#115", "#116", "#117", "#118", "#119", "#120", "#121", "#122", "#123", "#124", "#125", "#126", "#127", "#128", "#129", "#130", "#131", "SCTP", "#133", "#134", "#135", "UDPLite", "#137", "#138", "#139", "#140", "#141", "#142", "#143", "#144", "#145", "#146", "#147", "#148", "#149", "#150", "#151", "#152", "#153", "#154", "#155", "#156", "#157", "#158", "#159", "#160", "#161", "#162", "#163", "#164", "#165", "#166", "#167", "#168", "#169", "#170", "#171", "#172", "#173", "#174", "#175", "#176", "#177", "#178", "#179", "#180", "#181", "#182", "#183", "#184", "#185", "#186", "#187", "#188", "#189", "#180", "#191", "#192", "#193", "#194", "#195", "#196", "#197", "#198", "#199", "#200", "#201", "#202", "#203", "#204", "#205", "#206", "#207", "#208", "#209", "#210", "#211", "#212", "#213", "#214", "#215", "#216", "#217", "#218", "#219", "#220", "#221", "#222", "#223", "#224", "#225", "#226", "#227", "#228", "#229", "#230", "#231", "#232", "#233", "#234", "#235", "#236", "#237", "#238", "#239", "#240", "#241", "#242", "#243", "#244", "#245", "#246", "#247", "#248", "#249", "#250", "#251", "#252", "#253", "#254", "#255", }; static const char *srcrule_str[] = { "first candidate", "same address", "appropriate scope", "deprecated address", "home address", "outgoing interface", "matching label", "public/temporary address", "alive interface", "better virtual status", "preferred source", "rule #11", "rule #12", "rule #13", "longest match", "rule #15", }; /* * Dump IP6 statistics structure. */ void ip6_stats(u_long off, const char *name, int af1 __unused, int proto __unused) { struct ip6stat ip6stat; int first, i; if (fetch_stats("net.inet6.ip6.stats", off, &ip6stat, sizeof(ip6stat), kread_counters) != 0) return; xo_open_container(name); xo_emit("{T:/%s}:\n", name); #define p(f, m) if (ip6stat.f || sflag <= 1) \ xo_emit(m, (uintmax_t)ip6stat.f, plural(ip6stat.f)) #define p1a(f, m) if (ip6stat.f || sflag <= 1) \ xo_emit(m, (uintmax_t)ip6stat.f) p(ip6s_total, "\t{:received-packets/%ju} " "{N:/total packet%s received}\n"); p1a(ip6s_toosmall, "\t{:dropped-below-minimum-size/%ju} " "{N:/with size smaller than minimum}\n"); p1a(ip6s_tooshort, "\t{:dropped-short-packets/%ju} " "{N:/with data size < data length}\n"); p1a(ip6s_badoptions, "\t{:dropped-bad-options/%ju} " "{N:/with bad options}\n"); p1a(ip6s_badvers, "\t{:dropped-bad-version/%ju} " "{N:/with incorrect version number}\n"); p(ip6s_fragments, "\t{:received-fragments/%ju} " "{N:/fragment%s received}\n"); p(ip6s_fragdropped, "\t{:dropped-fragment/%ju} " "{N:/fragment%s dropped (dup or out of space)}\n"); p(ip6s_fragtimeout, "\t{:dropped-fragment-after-timeout/%ju} " "{N:/fragment%s dropped after timeout}\n"); p(ip6s_fragoverflow, "\t{:dropped-fragments-overflow/%ju} " "{N:/fragment%s that exceeded limit}\n"); p(ip6s_reassembled, "\t{:reassembled-packets/%ju} " "{N:/packet%s reassembled ok}\n"); p(ip6s_delivered, "\t{:received-local-packets/%ju} " "{N:/packet%s for this host}\n"); p(ip6s_forward, "\t{:forwarded-packets/%ju} " "{N:/packet%s forwarded}\n"); p(ip6s_cantforward, "\t{:packets-not-forwardable/%ju} " "{N:/packet%s not forwardable}\n"); p(ip6s_redirectsent, "\t{:sent-redirects/%ju} " "{N:/redirect%s sent}\n"); p(ip6s_localout, "\t{:sent-packets/%ju} " "{N:/packet%s sent from this host}\n"); p(ip6s_rawout, "\t{:send-packets-fabricated-header/%ju} " "{N:/packet%s sent with fabricated ip header}\n"); p(ip6s_odropped, "\t{:discard-no-mbufs/%ju} " "{N:/output packet%s dropped due to no bufs, etc.}\n"); p(ip6s_noroute, "\t{:discard-no-route/%ju} " "{N:/output packet%s discarded due to no route}\n"); p(ip6s_fragmented, "\t{:sent-fragments/%ju} " "{N:/output datagram%s fragmented}\n"); p(ip6s_ofragments, "\t{:fragments-created/%ju} " "{N:/fragment%s created}\n"); p(ip6s_cantfrag, "\t{:discard-cannot-fragment/%ju} " "{N:/datagram%s that can't be fragmented}\n"); p(ip6s_badscope, "\t{:discard-scope-violations/%ju} " "{N:/packet%s that violated scope rules}\n"); p(ip6s_notmember, "\t{:multicast-no-join-packets/%ju} " "{N:/multicast packet%s which we don't join}\n"); for (first = 1, i = 0; i < IP6S_HDRCNT; i++) if (ip6stat.ip6s_nxthist[i] != 0) { if (first) { xo_emit("\t{T:Input histogram}:\n"); xo_open_list("input-histogram"); first = 0; } xo_open_instance("input-histogram"); xo_emit("\t\t{k:name/%s}: {:count/%ju}\n", ip6nh[i], (uintmax_t)ip6stat.ip6s_nxthist[i]); xo_close_instance("input-histogram"); } if (!first) xo_close_list("input-histogram"); xo_open_container("mbuf-statistics"); xo_emit("\t{T:Mbuf statistics}:\n"); xo_emit("\t\t{:one-mbuf/%ju} {N:/one mbuf}\n", (uintmax_t)ip6stat.ip6s_m1); for (first = 1, i = 0; i < IP6S_M2MMAX; i++) { char ifbuf[IFNAMSIZ]; if (ip6stat.ip6s_m2m[i] != 0) { if (first) { xo_emit("\t\t{N:two or more mbuf}:\n"); xo_open_list("mbuf-data"); first = 0; } xo_open_instance("mbuf-data"); xo_emit("\t\t\t{k:name/%s}= {:count/%ju}\n", if_indextoname(i, ifbuf), (uintmax_t)ip6stat.ip6s_m2m[i]); xo_close_instance("mbuf-data"); } } if (!first) xo_close_list("mbuf-data"); xo_emit("\t\t{:one-extra-mbuf/%ju} {N:one ext mbuf}\n", (uintmax_t)ip6stat.ip6s_mext1); xo_emit("\t\t{:two-or-more-extra-mbufs/%ju} " "{N:/two or more ext mbuf}\n", (uintmax_t)ip6stat.ip6s_mext2m); xo_close_container("mbuf-statistics"); p(ip6s_exthdrtoolong, "\t{:dropped-header-too-long/%ju} " "{N:/packet%s whose headers are not contiguous}\n"); p(ip6s_nogif, "\t{:discard-tunnel-no-gif/%ju} " "{N:/tunneling packet%s that can't find gif}\n"); p(ip6s_toomanyhdr, "\t{:dropped-too-many-headers/%ju} " "{N:/packet%s discarded because of too many headers}\n"); /* for debugging source address selection */ #define PRINT_SCOPESTAT(s,i) do {\ switch(i) { /* XXX hardcoding in each case */\ case 1:\ p(s, "\t\t{ke:name/interface-locals}{:count/%ju} " \ "{N:/interface-local%s}\n"); \ break;\ case 2:\ p(s,"\t\t{ke:name/link-locals}{:count/%ju} " \ "{N:/link-local%s}\n"); \ break;\ case 5:\ p(s,"\t\t{ke:name/site-locals}{:count/%ju} " \ "{N:/site-local%s}\n");\ break;\ case 14:\ p(s,"\t\t{ke:name/globals}{:count/%ju} " \ "{N:/global%s}\n");\ break;\ default:\ xo_emit("\t\t{qke:name/%#x}{:count/%ju} " \ "{N:/addresses scope=%#x}\n",\ i, (uintmax_t)ip6stat.s, i); \ }\ } while (0); xo_open_container("source-address-selection"); p(ip6s_sources_none, "\t{:address-selection-failures/%ju} " "{N:/failure%s of source address selection}\n"); for (first = 1, i = 0; i < IP6S_SCOPECNT; i++) { if (ip6stat.ip6s_sources_sameif[i]) { if (first) { xo_open_list("outgoing-interface"); xo_emit("\tsource addresses on an outgoing " "I/F\n"); first = 0; } xo_open_instance("outgoing-interface"); PRINT_SCOPESTAT(ip6s_sources_sameif[i], i); xo_close_instance("outgoing-interface"); } } if (!first) xo_close_list("outgoing-interface"); for (first = 1, i = 0; i < IP6S_SCOPECNT; i++) { if (ip6stat.ip6s_sources_otherif[i]) { if (first) { xo_open_list("non-outgoing-interface"); xo_emit("\tsource addresses on a non-outgoing " "I/F\n"); first = 0; } xo_open_instance("non-outgoing-interface"); PRINT_SCOPESTAT(ip6s_sources_otherif[i], i); xo_close_instance("non-outgoing-interface"); } } if (!first) xo_close_list("non-outgoing-interface"); for (first = 1, i = 0; i < IP6S_SCOPECNT; i++) { if (ip6stat.ip6s_sources_samescope[i]) { if (first) { xo_open_list("same-source"); xo_emit("\tsource addresses of same scope\n"); first = 0; } xo_open_instance("same-source"); PRINT_SCOPESTAT(ip6s_sources_samescope[i], i); xo_close_instance("same-source"); } } if (!first) xo_close_list("same-source"); for (first = 1, i = 0; i < IP6S_SCOPECNT; i++) { if (ip6stat.ip6s_sources_otherscope[i]) { if (first) { xo_open_list("different-scope"); xo_emit("\tsource addresses of a different " "scope\n"); first = 0; } xo_open_instance("different-scope"); PRINT_SCOPESTAT(ip6s_sources_otherscope[i], i); xo_close_instance("different-scope"); } } if (!first) xo_close_list("different-scope"); for (first = 1, i = 0; i < IP6S_SCOPECNT; i++) { if (ip6stat.ip6s_sources_deprecated[i]) { if (first) { xo_open_list("deprecated-source"); xo_emit("\tdeprecated source addresses\n"); first = 0; } xo_open_instance("deprecated-source"); PRINT_SCOPESTAT(ip6s_sources_deprecated[i], i); xo_close_instance("deprecated-source"); } } if (!first) xo_close_list("deprecated-source"); for (first = 1, i = 0; i < IP6S_RULESMAX; i++) { if (ip6stat.ip6s_sources_rule[i]) { if (first) { xo_open_list("rules-applied"); xo_emit("\t{T:Source addresses selection " "rule applied}:\n"); first = 0; } xo_open_instance("rules-applied"); xo_emit("\t\t{ke:name/%s}{:count/%ju} {d:name/%s}\n", srcrule_str[i], (uintmax_t)ip6stat.ip6s_sources_rule[i], srcrule_str[i]); xo_close_instance("rules-applied"); } } if (!first) xo_close_list("rules-applied"); xo_close_container("source-address-selection"); #undef p #undef p1a xo_close_container(name); } /* * Dump IPv6 per-interface statistics based on RFC 2465. */ void ip6_ifstats(char *ifname) { struct in6_ifreq ifr; int s; #define p(f, m) if (ifr.ifr_ifru.ifru_stat.f || sflag <= 1) \ xo_emit(m, (uintmax_t)ifr.ifr_ifru.ifru_stat.f, \ plural(ifr.ifr_ifru.ifru_stat.f)) if ((s = socket(AF_INET6, SOCK_DGRAM, 0)) < 0) { xo_warn("Warning: socket(AF_INET6)"); return; } strcpy(ifr.ifr_name, ifname); if (ioctl(s, SIOCGIFSTAT_IN6, (char *)&ifr) < 0) { if (errno != EPFNOSUPPORT) xo_warn("Warning: ioctl(SIOCGIFSTAT_IN6)"); goto end; } xo_emit("{T:/ip6 on %s}:\n", ifr.ifr_name); xo_open_instance("ip6-interface-statistics"); xo_emit("{ke:name/%s}", ifr.ifr_name); p(ifs6_in_receive, "\t{:received-packets/%ju} " "{N:/total input datagram%s}\n"); p(ifs6_in_hdrerr, "\t{:dropped-invalid-header/%ju} " "{N:/datagram%s with invalid header received}\n"); p(ifs6_in_toobig, "\t{:dropped-mtu-exceeded/%ju} " "{N:/datagram%s exceeded MTU received}\n"); p(ifs6_in_noroute, "\t{:dropped-no-route/%ju} " "{N:/datagram%s with no route received}\n"); p(ifs6_in_addrerr, "\t{:dropped-invalid-destination/%ju} " "{N:/datagram%s with invalid dst received}\n"); p(ifs6_in_protounknown, "\t{:dropped-unknown-protocol/%ju} " "{N:/datagram%s with unknown proto received}\n"); p(ifs6_in_truncated, "\t{:dropped-truncated/%ju} " "{N:/truncated datagram%s received}\n"); p(ifs6_in_discard, "\t{:dropped-discarded/%ju} " "{N:/input datagram%s discarded}\n"); p(ifs6_in_deliver, "\t{:received-valid-packets/%ju} " "{N:/datagram%s delivered to an upper layer protocol}\n"); p(ifs6_out_forward, "\t{:sent-forwarded/%ju} " "{N:/datagram%s forwarded to this interface}\n"); p(ifs6_out_request, "\t{:sent-packets/%ju} " "{N:/datagram%s sent from an upper layer protocol}\n"); p(ifs6_out_discard, "\t{:discard-packets/%ju} " "{N:/total discarded output datagram%s}\n"); p(ifs6_out_fragok, "\t{:discard-fragments/%ju} " "{N:/output datagram%s fragmented}\n"); p(ifs6_out_fragfail, "\t{:fragments-failed/%ju} " "{N:/output datagram%s failed on fragment}\n"); p(ifs6_out_fragcreat, "\t{:fragments-created/%ju} " "{N:/output datagram%s succeeded on fragment}\n"); p(ifs6_reass_reqd, "\t{:reassembly-required/%ju} " "{N:/incoming datagram%s fragmented}\n"); p(ifs6_reass_ok, "\t{:reassembled-packets/%ju} " "{N:/datagram%s reassembled}\n"); p(ifs6_reass_fail, "\t{:reassembly-failed/%ju} " "{N:/datagram%s failed on reassembly}\n"); p(ifs6_in_mcast, "\t{:received-multicast/%ju} " "{N:/multicast datagram%s received}\n"); p(ifs6_out_mcast, "\t{:sent-multicast/%ju} " "{N:/multicast datagram%s sent}\n"); end: xo_close_instance("ip6-interface-statistics"); close(s); #undef p } static const char *icmp6names[] = { "#0", "unreach", "packet too big", "time exceed", "parameter problem", "#5", "#6", "#7", "#8", "#9", "#10", "#11", "#12", "#13", "#14", "#15", "#16", "#17", "#18", "#19", "#20", "#21", "#22", "#23", "#24", "#25", "#26", "#27", "#28", "#29", "#30", "#31", "#32", "#33", "#34", "#35", "#36", "#37", "#38", "#39", "#40", "#41", "#42", "#43", "#44", "#45", "#46", "#47", "#48", "#49", "#50", "#51", "#52", "#53", "#54", "#55", "#56", "#57", "#58", "#59", "#60", "#61", "#62", "#63", "#64", "#65", "#66", "#67", "#68", "#69", "#70", "#71", "#72", "#73", "#74", "#75", "#76", "#77", "#78", "#79", "#80", "#81", "#82", "#83", "#84", "#85", "#86", "#87", "#88", "#89", "#80", "#91", "#92", "#93", "#94", "#95", "#96", "#97", "#98", "#99", "#100", "#101", "#102", "#103", "#104", "#105", "#106", "#107", "#108", "#109", "#110", "#111", "#112", "#113", "#114", "#115", "#116", "#117", "#118", "#119", "#120", "#121", "#122", "#123", "#124", "#125", "#126", "#127", "echo", "echo reply", "multicast listener query", "MLDv1 listener report", "MLDv1 listener done", "router solicitation", "router advertisement", "neighbor solicitation", "neighbor advertisement", "redirect", "router renumbering", "node information request", "node information reply", "inverse neighbor solicitation", "inverse neighbor advertisement", "MLDv2 listener report", "#144", "#145", "#146", "#147", "#148", "#149", "#150", "#151", "#152", "#153", "#154", "#155", "#156", "#157", "#158", "#159", "#160", "#161", "#162", "#163", "#164", "#165", "#166", "#167", "#168", "#169", "#170", "#171", "#172", "#173", "#174", "#175", "#176", "#177", "#178", "#179", "#180", "#181", "#182", "#183", "#184", "#185", "#186", "#187", "#188", "#189", "#180", "#191", "#192", "#193", "#194", "#195", "#196", "#197", "#198", "#199", "#200", "#201", "#202", "#203", "#204", "#205", "#206", "#207", "#208", "#209", "#210", "#211", "#212", "#213", "#214", "#215", "#216", "#217", "#218", "#219", "#220", "#221", "#222", "#223", "#224", "#225", "#226", "#227", "#228", "#229", "#230", "#231", "#232", "#233", "#234", "#235", "#236", "#237", "#238", "#239", "#240", "#241", "#242", "#243", "#244", "#245", "#246", "#247", "#248", "#249", "#250", "#251", "#252", "#253", "#254", "#255", }; /* * Dump ICMP6 statistics. */ void icmp6_stats(u_long off, const char *name, int af1 __unused, int proto __unused) { struct icmp6stat icmp6stat; int i, first; if (fetch_stats("net.inet6.icmp6.stats", off, &icmp6stat, sizeof(icmp6stat), kread_counters) != 0) return; xo_emit("{T:/%s}:\n", name); xo_open_container(name); #define p(f, m) if (icmp6stat.f || sflag <= 1) \ xo_emit(m, (uintmax_t)icmp6stat.f, plural(icmp6stat.f)) #define p_5(f, m) if (icmp6stat.f || sflag <= 1) \ xo_emit(m, (uintmax_t)icmp6stat.f) p(icp6s_error, "\t{:icmp6-calls/%ju} " "{N:/call%s to icmp6_error}\n"); p(icp6s_canterror, "\t{:errors-not-generated-from-message/%ju} " "{N:/error%s not generated in response to an icmp6 message}\n"); p(icp6s_toofreq, "\t{:errors-discarded-by-rate-limitation/%ju} " "{N:/error%s not generated because of rate limitation}\n"); #define NELEM (int)(sizeof(icmp6stat.icp6s_outhist)/sizeof(icmp6stat.icp6s_outhist[0])) for (first = 1, i = 0; i < NELEM; i++) if (icmp6stat.icp6s_outhist[i] != 0) { if (first) { xo_open_list("output-histogram"); xo_emit("\t{T:Output histogram}:\n"); first = 0; } xo_open_instance("output-histogram"); xo_emit("\t\t{k:name/%s}: {:count/%ju}\n", icmp6names[i], (uintmax_t)icmp6stat.icp6s_outhist[i]); xo_close_instance("output-histogram"); } if (!first) xo_close_list("output-histogram"); #undef NELEM p(icp6s_badcode, "\t{:dropped-bad-code/%ju} " "{N:/message%s with bad code fields}\n"); p(icp6s_tooshort, "\t{:dropped-too-short/%ju} " "{N:/message%s < minimum length}\n"); p(icp6s_checksum, "\t{:dropped-bad-checksum/%ju} " "{N:/bad checksum%s}\n"); p(icp6s_badlen, "\t{:dropped-bad-length/%ju} " "{N:/message%s with bad length}\n"); #define NELEM (int)(sizeof(icmp6stat.icp6s_inhist)/sizeof(icmp6stat.icp6s_inhist[0])) for (first = 1, i = 0; i < NELEM; i++) if (icmp6stat.icp6s_inhist[i] != 0) { if (first) { xo_open_list("input-histogram"); xo_emit("\t{T:Input histogram}:\n"); first = 0; } xo_open_instance("input-histogram"); xo_emit("\t\t{k:name/%s}: {:count/%ju}\n", icmp6names[i], (uintmax_t)icmp6stat.icp6s_inhist[i]); xo_close_instance("input-histogram"); } if (!first) xo_close_list("input-histogram"); #undef NELEM xo_emit("\t{T:Histogram of error messages to be generated}:\n"); xo_open_container("errors"); p_5(icp6s_odst_unreach_noroute, "\t\t{:no-route/%ju} " "{N:/no route}\n"); p_5(icp6s_odst_unreach_admin, "\t\t{:admin-prohibited/%ju} " "{N:/administratively prohibited}\n"); p_5(icp6s_odst_unreach_beyondscope, "\t\t{:beyond-scope/%ju} " "{N:/beyond scope}\n"); p_5(icp6s_odst_unreach_addr, "\t\t{:address-unreachable/%ju} " "{N:/address unreachable}\n"); p_5(icp6s_odst_unreach_noport, "\t\t{:port-unreachable/%ju} " "{N:/port unreachable}\n"); p_5(icp6s_opacket_too_big, "\t\t{:packet-too-big/%ju} " "{N:/packet too big}\n"); p_5(icp6s_otime_exceed_transit, "\t\t{:time-exceed-transmit/%ju} " "{N:/time exceed transit}\n"); p_5(icp6s_otime_exceed_reassembly, "\t\t{:time-exceed-reassembly/%ju} " "{N:/time exceed reassembly}\n"); p_5(icp6s_oparamprob_header, "\t\t{:bad-header/%ju} " "{N:/erroneous header field}\n"); p_5(icp6s_oparamprob_nextheader, "\t\t{:bad-next-header/%ju} " "{N:/unrecognized next header}\n"); p_5(icp6s_oparamprob_option, "\t\t{:bad-option/%ju} " "{N:/unrecognized option}\n"); p_5(icp6s_oredirect, "\t\t{:redirects/%ju} " "{N:/redirect}\n"); p_5(icp6s_ounknown, "\t\t{:unknown/%ju} {N:unknown}\n"); p(icp6s_reflect, "\t{:reflect/%ju} " "{N:/message response%s generated}\n"); p(icp6s_nd_toomanyopt, "\t{:too-many-nd-options/%ju} " "{N:/message%s with too many ND options}\n"); p(icp6s_nd_badopt, "\t{:bad-nd-options/%ju} " "{N:/message%s with bad ND options}\n"); p(icp6s_badns, "\t{:bad-neighbor-solicitation/%ju} " "{N:/bad neighbor solicitation message%s}\n"); p(icp6s_badna, "\t{:bad-neighbor-advertisement/%ju} " "{N:/bad neighbor advertisement message%s}\n"); p(icp6s_badrs, "\t{:bad-router-solicitation/%ju} " "{N:/bad router solicitation message%s}\n"); p(icp6s_badra, "\t{:bad-router-advertisement/%ju} " "{N:/bad router advertisement message%s}\n"); p(icp6s_badredirect, "\t{:bad-redirect/%ju} " "{N:/bad redirect message%s}\n"); xo_close_container("errors"); p(icp6s_pmtuchg, "\t{:path-mtu-changes/%ju} {N:/path MTU change%s}\n"); #undef p #undef p_5 xo_close_container(name); } /* * Dump ICMPv6 per-interface statistics based on RFC 2466. */ void icmp6_ifstats(char *ifname) { struct in6_ifreq ifr; int s; #define p(f, m) if (ifr.ifr_ifru.ifru_icmp6stat.f || sflag <= 1) \ xo_emit(m, (uintmax_t)ifr.ifr_ifru.ifru_icmp6stat.f, \ plural(ifr.ifr_ifru.ifru_icmp6stat.f)) #define p2(f, m) if (ifr.ifr_ifru.ifru_icmp6stat.f || sflag <= 1) \ xo_emit(m, (uintmax_t)ifr.ifr_ifru.ifru_icmp6stat.f, \ pluralies(ifr.ifr_ifru.ifru_icmp6stat.f)) if ((s = socket(AF_INET6, SOCK_DGRAM, 0)) < 0) { xo_warn("Warning: socket(AF_INET6)"); return; } strcpy(ifr.ifr_name, ifname); if (ioctl(s, SIOCGIFSTAT_ICMP6, (char *)&ifr) < 0) { if (errno != EPFNOSUPPORT) xo_warn("Warning: ioctl(SIOCGIFSTAT_ICMP6)"); goto end; } xo_emit("{T:/icmp6 on %s}:\n", ifr.ifr_name); xo_open_instance("icmp6-interface-statistics"); xo_emit("{ke:name/%s}", ifr.ifr_name); p(ifs6_in_msg, "\t{:received-packets/%ju} " "{N:/total input message%s}\n"); p(ifs6_in_error, "\t{:received-errors/%ju} " "{N:/total input error message%s}\n"); p(ifs6_in_dstunreach, "\t{:received-destination-unreachable/%ju} " "{N:/input destination unreachable error%s}\n"); p(ifs6_in_adminprohib, "\t{:received-admin-prohibited/%ju} " "{N:/input administratively prohibited error%s}\n"); p(ifs6_in_timeexceed, "\t{:received-time-exceeded/%ju} " "{N:/input time exceeded error%s}\n"); p(ifs6_in_paramprob, "\t{:received-bad-parameter/%ju} " "{N:/input parameter problem error%s}\n"); p(ifs6_in_pkttoobig, "\t{:received-packet-too-big/%ju} " "{N:/input packet too big error%s}\n"); p(ifs6_in_echo, "\t{:received-echo-requests/%ju} " "{N:/input echo request%s}\n"); p2(ifs6_in_echoreply, "\t{:received-echo-replies/%ju} " "{N:/input echo repl%s}\n"); p(ifs6_in_routersolicit, "\t{:received-router-solicitation/%ju} " "{N:/input router solicitation%s}\n"); p(ifs6_in_routeradvert, "\t{:received-router-advertisement/%ju} " "{N:/input router advertisement%s}\n"); p(ifs6_in_neighborsolicit, "\t{:received-neighbor-solicitation/%ju} " "{N:/input neighbor solicitation%s}\n"); p(ifs6_in_neighboradvert, "\t{:received-neighbor-advertisement/%ju} " "{N:/input neighbor advertisement%s}\n"); p(ifs6_in_redirect, "\t{received-redirects/%ju} " "{N:/input redirect%s}\n"); p2(ifs6_in_mldquery, "\t{:received-mld-queries/%ju} " "{N:/input MLD quer%s}\n"); p(ifs6_in_mldreport, "\t{:received-mld-reports/%ju} " "{N:/input MLD report%s}\n"); p(ifs6_in_mlddone, "\t{:received-mld-done/%ju} " "{N:/input MLD done%s}\n"); p(ifs6_out_msg, "\t{:sent-packets/%ju} " "{N:/total output message%s}\n"); p(ifs6_out_error, "\t{:sent-errors/%ju} " "{N:/total output error message%s}\n"); p(ifs6_out_dstunreach, "\t{:sent-destination-unreachable/%ju} " "{N:/output destination unreachable error%s}\n"); p(ifs6_out_adminprohib, "\t{:sent-admin-prohibited/%ju} " "{N:/output administratively prohibited error%s}\n"); p(ifs6_out_timeexceed, "\t{:sent-time-exceeded/%ju} " "{N:/output time exceeded error%s}\n"); p(ifs6_out_paramprob, "\t{:sent-bad-parameter/%ju} " "{N:/output parameter problem error%s}\n"); p(ifs6_out_pkttoobig, "\t{:sent-packet-too-big/%ju} " "{N:/output packet too big error%s}\n"); p(ifs6_out_echo, "\t{:sent-echo-requests/%ju} " "{N:/output echo request%s}\n"); p2(ifs6_out_echoreply, "\t{:sent-echo-replies/%ju} " "{N:/output echo repl%s}\n"); p(ifs6_out_routersolicit, "\t{:sent-router-solicitation/%ju} " "{N:/output router solicitation%s}\n"); p(ifs6_out_routeradvert, "\t{:sent-router-advertisement/%ju} " "{N:/output router advertisement%s}\n"); p(ifs6_out_neighborsolicit, "\t{:sent-neighbor-solicitation/%ju} " "{N:/output neighbor solicitation%s}\n"); p(ifs6_out_neighboradvert, "\t{:sent-neighbor-advertisement/%ju} " "{N:/output neighbor advertisement%s}\n"); p(ifs6_out_redirect, "\t{:sent-redirects/%ju} " "{N:/output redirect%s}\n"); p2(ifs6_out_mldquery, "\t{:sent-mld-queries/%ju} " "{N:/output MLD quer%s}\n"); p(ifs6_out_mldreport, "\t{:sent-mld-reports/%ju} " "{N:/output MLD report%s}\n"); p(ifs6_out_mlddone, "\t{:sent-mld-dones/%ju} " "{N:/output MLD done%s}\n"); end: xo_close_instance("icmp6-interface-statistics"); close(s); #undef p } /* * Dump PIM statistics structure. */ void pim6_stats(u_long off, const char *name, int af1 __unused, int proto __unused) { struct pim6stat pim6stat; if (fetch_stats("net.inet6.pim.stats", off, &pim6stat, sizeof(pim6stat), kread) != 0) return; xo_emit("{T:/%s}:\n", name); xo_open_container(name); #define p(f, m) if (pim6stat.f || sflag <= 1) \ xo_emit(m, (uintmax_t)pim6stat.f, plural(pim6stat.f)) p(pim6s_rcv_total, "\t{:received-packets/%ju} " "{N:/message%s received}\n"); p(pim6s_rcv_tooshort, "\t{:dropped-too-short/%ju} " "{N:/message%s received with too few bytes}\n"); p(pim6s_rcv_badsum, "\t{:dropped-bad-checksum/%ju} " "{N:/message%s received with bad checksum}\n"); p(pim6s_rcv_badversion, "\t{:dropped-bad-version/%ju} " "{N:/message%s received with bad version}\n"); p(pim6s_rcv_registers, "\t{:received-registers/%ju} " "{N:/register%s received}\n"); p(pim6s_rcv_badregisters, "\t{:received-bad-registers/%ju} " "{N:/bad register%s received}\n"); p(pim6s_snd_registers, "\t{:sent-registers/%ju} " "{N:/register%s sent}\n"); #undef p xo_close_container(name); } /* * Dump raw ip6 statistics structure. */ void rip6_stats(u_long off, const char *name, int af1 __unused, int proto __unused) { struct rip6stat rip6stat; u_quad_t delivered; if (fetch_stats("net.inet6.ip6.rip6stats", off, &rip6stat, sizeof(rip6stat), kread_counters) != 0) return; xo_emit("{T:/%s}:\n", name); xo_open_container(name); #define p(f, m) if (rip6stat.f || sflag <= 1) \ xo_emit(m, (uintmax_t)rip6stat.f, plural(rip6stat.f)) p(rip6s_ipackets, "\t{:received-packets/%ju} " "{N:/message%s received}\n"); p(rip6s_isum, "\t{:input-checksum-computation/%ju} " "{N:/checksum calculation%s on inbound}\n"); p(rip6s_badsum, "\t{:received-bad-checksum/%ju} " "{N:/message%s with bad checksum}\n"); p(rip6s_nosock, "\t{:dropped-no-socket/%ju} " "{N:/message%s dropped due to no socket}\n"); p(rip6s_nosockmcast, "\t{:dropped-multicast-no-socket/%ju} " "{N:/multicast message%s dropped due to no socket}\n"); p(rip6s_fullsock, "\t{:dropped-full-socket-buffer/%ju} " "{N:/message%s dropped due to full socket buffers}\n"); delivered = rip6stat.rip6s_ipackets - rip6stat.rip6s_badsum - rip6stat.rip6s_nosock - rip6stat.rip6s_nosockmcast - rip6stat.rip6s_fullsock; if (delivered || sflag <= 1) xo_emit("\t{:delivered-packets/%ju} {N:/delivered}\n", (uintmax_t)delivered); p(rip6s_opackets, "\t{:sent-packets/%ju} " "{N:/datagram%s output}\n"); #undef p xo_close_container(name); } /* * Pretty print an Internet address (net address + port). * Take numeric_addr and numeric_port into consideration. */ #define GETSERVBYPORT6(port, proto, ret)\ {\ if (strcmp((proto), "tcp6") == 0)\ (ret) = getservbyport((int)(port), "tcp");\ else if (strcmp((proto), "udp6") == 0)\ (ret) = getservbyport((int)(port), "udp");\ else\ (ret) = getservbyport((int)(port), (proto));\ }; void inet6print(const char *container, struct in6_addr *in6, int port, const char *proto, int numeric) { struct servent *sp = 0; char line[80], *cp; int width; + size_t alen, plen; if (container) xo_open_container(container); - sprintf(line, "%.*s.", Wflag ? 39 : (Aflag && !numeric) ? 12 : 16, + snprintf(line, sizeof(line), "%.*s.", + Wflag ? 39 : (Aflag && !numeric) ? 12 : 16, inet6name(in6)); - cp = strchr(line, '\0'); + alen = strlen(line); + cp = line + alen; if (!numeric && port) GETSERVBYPORT6(port, proto, sp); if (sp || port == 0) - sprintf(cp, "%.15s", sp ? sp->s_name : "*"); + snprintf(cp, sizeof(line) - alen, + "%.15s", sp ? sp->s_name : "*"); else - sprintf(cp, "%d", ntohs((u_short)port)); + snprintf(cp, sizeof(line) - alen, + "%d", ntohs((u_short)port)); width = Wflag ? 45 : Aflag ? 18 : 22; xo_emit("{d:target/%-*.*s} ", width, width, line); - int alen = cp - line - 1, plen = strlen(cp) - 1; + plen = strlen(cp); + alen--; xo_emit("{e:address/%*.*s}{e:port/%*.*s}", alen, alen, line, plen, plen, cp); if (container) xo_close_container(container); } /* * Construct an Internet address representation. * If the numeric_addr has been supplied, give * numeric value, otherwise try for symbolic name. */ char * inet6name(struct in6_addr *in6p) { struct sockaddr_in6 sin6; char hbuf[NI_MAXHOST], *cp; - static char line[50]; + static char line[NI_MAXHOST]; static char domain[MAXHOSTNAMELEN]; static int first = 1; int flags, error; if (IN6_IS_ADDR_UNSPECIFIED(in6p)) { strcpy(line, "*"); return (line); } if (first && !numeric_addr) { first = 0; - if (gethostname(domain, MAXHOSTNAMELEN) == 0 && + if (gethostname(domain, sizeof(domain)) == 0 && (cp = strchr(domain, '.'))) - (void) strcpy(domain, cp + 1); + strlcpy(domain, cp + 1, sizeof(domain)); else domain[0] = 0; } memset(&sin6, 0, sizeof(sin6)); memcpy(&sin6.sin6_addr, in6p, sizeof(*in6p)); sin6.sin6_family = AF_INET6; /* XXX: in6p.s6_addr[2] can contain scopeid. */ in6_fillscopeid(&sin6); flags = (numeric_addr) ? NI_NUMERICHOST : 0; error = getnameinfo((struct sockaddr *)&sin6, sizeof(sin6), hbuf, sizeof(hbuf), NULL, 0, flags); if (error == 0) { if ((flags & NI_NUMERICHOST) == 0 && (cp = strchr(hbuf, '.')) && !strcmp(cp + 1, domain)) *cp = 0; - strcpy(line, hbuf); + strlcpy(line, hbuf, sizeof(line)); } else { /* XXX: this should not happen. */ - sprintf(line, "%s", + snprintf(line, sizeof(line), "%s", inet_ntop(AF_INET6, (void *)&sin6.sin6_addr, ntop_buf, sizeof(ntop_buf))); } return (line); } #endif /*INET6*/ Index: projects/clang400-import/usr.bin/netstat/mroute.c =================================================================== --- projects/clang400-import/usr.bin/netstat/mroute.c (revision 311459) +++ projects/clang400-import/usr.bin/netstat/mroute.c (revision 311460) @@ -1,454 +1,458 @@ /*- * Copyright (c) 1989 Stephen Deering * Copyright (c) 1992, 1993 * The Regents of the University of California. All rights reserved. * * This code is derived from software contributed to Berkeley by * Stephen Deering of Stanford University. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. All advertising materials mentioning features or use of this software * must display the following acknowledgement: * This product includes software developed by the University of * California, Berkeley and its contributors. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * @(#)mroute.c 8.2 (Berkeley) 4/28/95 */ #include __FBSDID("$FreeBSD$"); /* * Print multicast routing structures and statistics. * * MROUTING 1.0 */ #include #include #include #include #include #include #include #include #include #include #include #include #define _KERNEL 1 #include #undef _KERNEL #include #include #include #include #include #include #include #include "netstat.h" #include "nl_defs.h" static void print_bw_meter(struct bw_meter *, int *); static void print_mfc(struct mfc *, int, int *); static void print_bw_meter(struct bw_meter *bw_meter, int *banner_printed) { char s1[256], s2[256], s3[256]; struct timeval now, end, delta; gettimeofday(&now, NULL); if (! *banner_printed) { xo_open_list("bandwidth-meter"); xo_emit(" {T:Bandwidth Meters}\n"); xo_emit(" {T:/%-30s}", "Measured(Start|Packets|Bytes)"); xo_emit(" {T:/%s}", "Type"); xo_emit(" {T:/%-30s}", "Thresh(Interval|Packets|Bytes)"); xo_emit(" {T:Remain}"); xo_emit("\n"); *banner_printed = 1; } xo_open_instance("bandwidth-meter"); /* The measured values */ if (bw_meter->bm_flags & BW_METER_UNIT_PACKETS) { - sprintf(s1, "%ju", (uintmax_t)bw_meter->bm_measured.b_packets); + snprintf(s1, sizeof(s1), "%ju", + (uintmax_t)bw_meter->bm_measured.b_packets); xo_emit("{e:measured-packets/%ju}", (uintmax_t)bw_meter->bm_measured.b_packets); } else - sprintf(s1, "?"); + strcpy(s1, "?"); if (bw_meter->bm_flags & BW_METER_UNIT_BYTES) { - sprintf(s2, "%ju", (uintmax_t)bw_meter->bm_measured.b_bytes); + snprintf(s2, sizeof(s2), "%ju", + (uintmax_t)bw_meter->bm_measured.b_bytes); xo_emit("{e:measured-bytes/%ju}", (uintmax_t)bw_meter->bm_measured.b_bytes); } else - sprintf(s2, "?"); + strcpy(s2, "?"); xo_emit(" {[:-30}{:start-time/%lu.%06lu}|{q:measured-packets/%s}" "|{q:measured-bytes%s}{]:}", (u_long)bw_meter->bm_start_time.tv_sec, (u_long)bw_meter->bm_start_time.tv_usec, s1, s2); /* The type of entry */ xo_emit(" {t:type/%-3s}", (bw_meter->bm_flags & BW_METER_GEQ) ? ">=" : (bw_meter->bm_flags & BW_METER_LEQ) ? "<=" : "?"); /* The threshold values */ if (bw_meter->bm_flags & BW_METER_UNIT_PACKETS) { - sprintf(s1, "%ju", (uintmax_t)bw_meter->bm_threshold.b_packets); + snprintf(s1, sizeof(s1), "%ju", + (uintmax_t)bw_meter->bm_threshold.b_packets); xo_emit("{e:threshold-packets/%ju}", (uintmax_t)bw_meter->bm_threshold.b_packets); } else - sprintf(s1, "?"); + strcpy(s1, "?"); if (bw_meter->bm_flags & BW_METER_UNIT_BYTES) { - sprintf(s2, "%ju", (uintmax_t)bw_meter->bm_threshold.b_bytes); + snprintf(s2, sizeof(s2), "%ju", + (uintmax_t)bw_meter->bm_threshold.b_bytes); xo_emit("{e:threshold-bytes/%ju}", (uintmax_t)bw_meter->bm_threshold.b_bytes); } else - sprintf(s2, "?"); + strcpy(s2, "?"); xo_emit(" {[:-30}{:threshold-time/%lu.%06lu}|{q:threshold-packets/%s}" "|{q:threshold-bytes%s}{]:}", (u_long)bw_meter->bm_threshold.b_time.tv_sec, (u_long)bw_meter->bm_threshold.b_time.tv_usec, s1, s2); /* Remaining time */ timeradd(&bw_meter->bm_start_time, &bw_meter->bm_threshold.b_time, &end); if (timercmp(&now, &end, <=)) { timersub(&end, &now, &delta); - sprintf(s3, "%lu.%06lu", + snprintf(s3, sizeof(s3), "%lu.%06lu", (u_long)delta.tv_sec, (u_long)delta.tv_usec); } else { /* Negative time */ timersub(&now, &end, &delta); - sprintf(s3, "-%lu.06%lu", + snprintf(s3, sizeof(s3), "-%lu.06%lu", (u_long)delta.tv_sec, (u_long)delta.tv_usec); } xo_emit(" {:remaining-time/%s}", s3); xo_open_instance("bandwidth-meter"); xo_emit("\n"); } static void print_mfc(struct mfc *m, int maxvif, int *banner_printed) { struct sockaddr_in sin; struct sockaddr *sa = (struct sockaddr *)&sin; struct bw_meter bw_meter, *bwm; int bw_banner_printed; int error; vifi_t vifi; bw_banner_printed = 0; memset(&sin, 0, sizeof(sin)); sin.sin_len = sizeof(sin); sin.sin_family = AF_INET; if (! *banner_printed) { xo_open_list("multicast-forwarding-entry"); xo_emit("\n{T:IPv4 Multicast Forwarding Table}\n" " {T:Origin} {T:Group} " " {T:Packets In-Vif} {T:Out-Vifs:Ttls}\n"); *banner_printed = 1; } memcpy(&sin.sin_addr, &m->mfc_origin, sizeof(sin.sin_addr)); xo_emit(" {:origin-address/%-15.15s}", routename(sa, numeric_addr)); memcpy(&sin.sin_addr, &m->mfc_mcastgrp, sizeof(sin.sin_addr)); xo_emit(" {:group-address/%-15.15s}", routename(sa, numeric_addr)); xo_emit(" {:sent-packets/%9lu}", m->mfc_pkt_cnt); xo_emit(" {:parent/%3d} ", m->mfc_parent); xo_open_list("vif-ttl"); for (vifi = 0; vifi <= maxvif; vifi++) { if (m->mfc_ttls[vifi] > 0) { xo_open_instance("vif-ttl"); xo_emit(" {k:vif/%u}:{:ttl/%u}", vifi, m->mfc_ttls[vifi]); xo_close_instance("vif-ttl"); } } xo_close_list("vif-ttl"); xo_emit("\n"); /* * XXX We break the rules and try to use KVM to read the * bandwidth meters, they are not retrievable via sysctl yet. */ bwm = m->mfc_bw_meter; while (bwm != NULL) { error = kread((u_long)bwm, (char *)&bw_meter, sizeof(bw_meter)); if (error) break; print_bw_meter(&bw_meter, &bw_banner_printed); bwm = bw_meter.bm_mfc_next; } if (banner_printed) xo_close_list("bandwidth-meter"); } void mroutepr() { struct sockaddr_in sin; struct sockaddr *sa = (struct sockaddr *)&sin; struct vif viftable[MAXVIFS]; struct vif *v; struct mfc *m; u_long pmfchashtbl, pmfctablesize, pviftbl; int banner_printed; int saved_numeric_addr; size_t len; vifi_t vifi, maxvif; saved_numeric_addr = numeric_addr; numeric_addr = 1; memset(&sin, 0, sizeof(sin)); sin.sin_len = sizeof(sin); sin.sin_family = AF_INET; /* * TODO: * The VIF table will move to hanging off the struct if_info for * each IPv4 configured interface. Currently it is statically * allocated, and retrieved either using KVM or an opaque SYSCTL. * * This can't happen until the API documented in multicast(4) * is itself refactored. The historical reason why VIFs use * a separate ifindex space is entirely due to the legacy * capability of the MROUTING code to create IPIP tunnels on * the fly to support DVMRP. When gif(4) became available, this * functionality was deprecated, as PIM does not use it. */ maxvif = 0; pmfchashtbl = pmfctablesize = pviftbl = 0; len = sizeof(viftable); if (live) { if (sysctlbyname("net.inet.ip.viftable", viftable, &len, NULL, 0) < 0) { xo_warn("sysctl: net.inet.ip.viftable"); return; } } else { pmfchashtbl = nl[N_MFCHASHTBL].n_value; pmfctablesize = nl[N_MFCTABLESIZE].n_value; pviftbl = nl[N_VIFTABLE].n_value; if (pmfchashtbl == 0 || pmfctablesize == 0 || pviftbl == 0) { xo_warnx("No IPv4 MROUTING kernel support."); return; } kread(pviftbl, (char *)viftable, sizeof(viftable)); } banner_printed = 0; for (vifi = 0, v = viftable; vifi < MAXVIFS; ++vifi, ++v) { if (v->v_lcl_addr.s_addr == 0) continue; maxvif = vifi; if (!banner_printed) { xo_emit("\n{T:IPv4 Virtual Interface Table\n" " Vif Thresh Local-Address " "Remote-Address Pkts-In Pkts-Out}\n"); banner_printed = 1; xo_open_list("vif"); } xo_open_instance("vif"); memcpy(&sin.sin_addr, &v->v_lcl_addr, sizeof(sin.sin_addr)); xo_emit(" {:vif/%2u} {:threshold/%6u} {:route/%-15.15s}", /* opposite math of add_vif() */ vifi, v->v_threshold, routename(sa, numeric_addr)); memcpy(&sin.sin_addr, &v->v_rmt_addr, sizeof(sin.sin_addr)); xo_emit(" {:source/%-15.15s}", (v->v_flags & VIFF_TUNNEL) ? routename(sa, numeric_addr) : ""); xo_emit(" {:received-packets/%9lu} {:sent-packets/%9lu}\n", v->v_pkt_in, v->v_pkt_out); xo_close_instance("vif"); } if (banner_printed) xo_close_list("vif"); else xo_emit("\n{T:IPv4 Virtual Interface Table is empty}\n"); banner_printed = 0; /* * TODO: * The MFC table will move into the AF_INET radix trie in future. * In 8.x, it becomes a dynamically allocated structure referenced * by a hashed LIST, allowing more than 256 entries w/o kernel tuning. * * If retrieved via opaque SYSCTL, the kernel will coalesce it into * a static table for us. * If retrieved via KVM, the hash list pointers must be followed. */ if (live) { struct mfc *mfctable; len = 0; if (sysctlbyname("net.inet.ip.mfctable", NULL, &len, NULL, 0) < 0) { xo_warn("sysctl: net.inet.ip.mfctable"); return; } mfctable = malloc(len); if (mfctable == NULL) { xo_warnx("malloc %lu bytes", (u_long)len); return; } if (sysctlbyname("net.inet.ip.mfctable", mfctable, &len, NULL, 0) < 0) { free(mfctable); xo_warn("sysctl: net.inet.ip.mfctable"); return; } m = mfctable; while (len >= sizeof(*m)) { print_mfc(m++, maxvif, &banner_printed); len -= sizeof(*m); } if (banner_printed) xo_close_list("multicast-forwarding-entry"); if (len != 0) xo_warnx("print_mfc: %lu trailing bytes", (u_long)len); free(mfctable); } else { LIST_HEAD(, mfc) *mfchashtbl; u_long i, mfctablesize; struct mfc mfc; int error; error = kread(pmfctablesize, (char *)&mfctablesize, sizeof(u_long)); if (error) { xo_warn("kread: mfctablesize"); return; } len = sizeof(*mfchashtbl) * mfctablesize; mfchashtbl = malloc(len); if (mfchashtbl == NULL) { xo_warnx("malloc %lu bytes", (u_long)len); return; } kread(pmfchashtbl, (char *)&mfchashtbl, len); for (i = 0; i < mfctablesize; i++) { LIST_FOREACH(m, &mfchashtbl[i], mfc_hash) { kread((u_long)m, (char *)&mfc, sizeof(mfc)); print_mfc(m, maxvif, &banner_printed); } } if (banner_printed) xo_close_list("multicast-forwarding-entry"); free(mfchashtbl); } if (!banner_printed) xo_emit("\n{T:IPv4 Multicast Forwarding Table is empty}\n"); xo_emit("\n"); numeric_addr = saved_numeric_addr; } void mrt_stats() { struct mrtstat mrtstat; u_long mstaddr; mstaddr = nl[N_MRTSTAT].n_value; if (mstaddr == 0) { fprintf(stderr, "No IPv4 MROUTING kernel support.\n"); return; } if (fetch_stats("net.inet.ip.mrtstat", mstaddr, &mrtstat, sizeof(mrtstat), kread_counters) != 0) return; xo_emit("{T:IPv4 multicast forwarding}:\n"); #define p(f, m) if (mrtstat.f || sflag <= 1) \ xo_emit(m, (uintmax_t)mrtstat.f, plural(mrtstat.f)) #define p2(f, m) if (mrtstat.f || sflag <= 1) \ xo_emit(m, (uintmax_t)mrtstat.f, plurales(mrtstat.f)) xo_open_container("multicast-statistics"); p(mrts_mfc_lookups, "\t{:cache-lookups/%ju} " "{N:/multicast forwarding cache lookup%s}\n"); p2(mrts_mfc_misses, "\t{:cache-misses/%ju} " "{N:/multicast forwarding cache miss%s}\n"); p(mrts_upcalls, "\t{:upcalls-total/%ju} " "{N:/upcall%s to multicast routing daemon}\n"); p(mrts_upq_ovflw, "\t{:upcall-overflows/%ju} " "{N:/upcall queue overflow%s}\n"); p(mrts_upq_sockfull, "\t{:upcalls-dropped-full-buffer/%ju} " "{N:/upcall%s dropped due to full socket buffer}\n"); p(mrts_cache_cleanups, "\t{:cache-cleanups/%ju} " "{N:/cache cleanup%s}\n"); p(mrts_no_route, "\t{:dropped-no-origin/%ju} " "{N:/datagram%s with no route for origin}\n"); p(mrts_bad_tunnel, "\t{:dropped-bad-tunnel/%ju} " "{N:/datagram%s arrived with bad tunneling}\n"); p(mrts_cant_tunnel, "\t{:dropped-could-not-tunnel/%ju} " "{N:/datagram%s could not be tunneled}\n"); p(mrts_wrong_if, "\t{:dropped-wrong-incoming-interface/%ju} " "{N:/datagram%s arrived on wrong interface}\n"); p(mrts_drop_sel, "\t{:dropped-selectively/%ju} " "{N:/datagram%s selectively dropped}\n"); p(mrts_q_overflow, "\t{:dropped-queue-overflow/%ju} " "{N:/datagram%s dropped due to queue overflow}\n"); p(mrts_pkt2large, "\t{:dropped-too-large/%ju} " "{N:/datagram%s dropped for being too large}\n"); #undef p2 #undef p } Index: projects/clang400-import/usr.bin/netstat/netstat.h =================================================================== --- projects/clang400-import/usr.bin/netstat/netstat.h (revision 311459) +++ projects/clang400-import/usr.bin/netstat/netstat.h (revision 311460) @@ -1,148 +1,157 @@ /*- * Copyright (c) 1992, 1993 * 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. * * @(#)netstat.h 8.2 (Berkeley) 1/4/94 * $FreeBSD$ */ #include #define satosin(sa) ((struct sockaddr_in *)(sa)) #define satosin6(sa) ((struct sockaddr_in6 *)(sa)) #define sin6tosa(sin6) ((struct sockaddr *)(sin6)) extern int Aflag; /* show addresses of protocol control block */ extern int aflag; /* show all sockets (including servers) */ extern int bflag; /* show i/f total bytes in/out */ extern int dflag; /* show i/f dropped packets */ extern int gflag; /* show group (multicast) routing or stats */ extern int hflag; /* show counters in human readable format */ extern int iflag; /* show interfaces */ extern int Lflag; /* show size of listen queues */ extern int mflag; /* show memory stats */ extern int noutputs; /* how much outputs before we exit */ extern int numeric_addr; /* show addresses numerically */ extern int numeric_port; /* show ports numerically */ extern int rflag; /* show routing tables (or routing stats) */ extern int Rflag; /* show flowid / RSS information */ extern int sflag; /* show protocol statistics */ extern int Tflag; /* show TCP control block info */ extern int Wflag; /* wide display */ extern int xflag; /* extended display, includes all socket buffer info */ extern int zflag; /* zero stats */ extern int interval; /* repeat interval for i/f stats */ extern char *interface; /* desired i/f for stats, or NULL for all i/fs */ extern int unit; /* unit number for above */ extern int live; /* true if we are examining a live system */ typedef int kreadfn_t(u_long, void *, size_t); int fetch_stats(const char *, u_long, void *, size_t, kreadfn_t); int fetch_stats_ro(const char *, u_long, void *, size_t, kreadfn_t); int kread(u_long addr, void *buf, size_t size); uint64_t kread_counter(u_long addr); int kread_counters(u_long addr, void *buf, size_t size); void kset_dpcpu(u_int); const char *plural(uintmax_t); const char *plurales(uintmax_t); const char *pluralies(uintmax_t); struct sockaddr; struct socket; struct xsocket; int sotoxsocket(struct socket *, struct xsocket *); void protopr(u_long, const char *, int, int); void tcp_stats(u_long, const char *, int, int); void udp_stats(u_long, const char *, int, int); #ifdef SCTP void sctp_protopr(u_long, const char *, int, int); void sctp_stats(u_long, const char *, int, int); #endif void arp_stats(u_long, const char *, int, int); void ip_stats(u_long, const char *, int, int); void icmp_stats(u_long, const char *, int, int); void igmp_stats(u_long, const char *, int, int); void pim_stats(u_long, const char *, int, int); void carp_stats(u_long, const char *, int, int); void pfsync_stats(u_long, const char *, int, int); #ifdef IPSEC void ipsec_stats(u_long, const char *, int, int); void esp_stats(u_long, const char *, int, int); void ah_stats(u_long, const char *, int, int); void ipcomp_stats(u_long, const char *, int, int); #endif +#ifdef INET +struct in_addr; + +char *inetname(struct in_addr *); +#endif + #ifdef INET6 +struct in6_addr; + +char *inet6name(struct in6_addr *); void ip6_stats(u_long, const char *, int, int); void ip6_ifstats(char *); void icmp6_stats(u_long, const char *, int, int); void icmp6_ifstats(char *); void pim6_stats(u_long, const char *, int, int); void rip6_stats(u_long, const char *, int, int); void mroute6pr(void); void mrt6_stats(void); struct sockaddr_in6; struct in6_addr; void in6_fillscopeid(struct sockaddr_in6 *); void inet6print(const char *, struct in6_addr *, int, const char *, int); #endif /*INET6*/ #ifdef IPSEC void pfkey_stats(u_long, const char *, int, int); #endif void mbpr(void *, u_long); void netisr_stats(void); void hostpr(u_long, u_long); void impstats(u_long, u_long); void intpr(void (*)(char *), int); void pr_family(int); void rt_stats(void); void flowtable_stats(void); char *routename(struct sockaddr *, int); const char *netname(struct sockaddr *, struct sockaddr *); void routepr(int, int); #ifdef NETGRAPH void netgraphprotopr(u_long, const char *, int, int); #endif void unixpr(u_long, u_long, u_long, u_long, u_long, bool *); void mroutepr(void); void mrt_stats(void); void bpf_stats(char *); Index: projects/clang400-import/usr.bin/netstat/route.c =================================================================== --- projects/clang400-import/usr.bin/netstat/route.c (revision 311459) +++ projects/clang400-import/usr.bin/netstat/route.c (revision 311460) @@ -1,789 +1,799 @@ /*- * Copyright (c) 1983, 1988, 1993 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #if 0 #ifndef lint static char sccsid[] = "From: @(#)route.c 8.6 (Berkeley) 4/28/95"; #endif /* not lint */ #endif #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include +#include #include #include #include #include #include #include #include #include #include #include "netstat.h" #include "nl_defs.h" /* * Definitions for showing gateway flags. */ static struct bits { u_long b_mask; char b_val; const char *b_name; } bits[] = { { RTF_UP, 'U', "up" }, { RTF_GATEWAY, 'G', "gateway" }, { RTF_HOST, 'H', "host" }, { RTF_REJECT, 'R', "reject" }, { RTF_DYNAMIC, 'D', "dynamic" }, { RTF_MODIFIED, 'M', "modified" }, { RTF_DONE, 'd', "done" }, /* Completed -- for routing msgs only */ { RTF_XRESOLVE, 'X', "xresolve" }, { RTF_STATIC, 'S', "static" }, { RTF_PROTO1, '1', "proto1" }, { RTF_PROTO2, '2', "proto2" }, { RTF_PROTO3, '3', "proto3" }, { RTF_BLACKHOLE,'B', "blackhole" }, { RTF_BROADCAST,'b', "broadcast" }, #ifdef RTF_LLINFO { RTF_LLINFO, 'L', "llinfo" }, #endif { 0 , 0, NULL } }; struct ifmap_entry { char ifname[IFNAMSIZ]; }; static struct ifmap_entry *ifmap; static int ifmap_size; static struct timespec uptime; static const char *netname4(in_addr_t, in_addr_t); #ifdef INET6 static const char *netname6(struct sockaddr_in6 *, struct sockaddr_in6 *); #endif static void p_rtable_sysctl(int, int); static void p_rtentry_sysctl(const char *name, struct rt_msghdr *); static int p_sockaddr(const char *name, struct sockaddr *, struct sockaddr *, int, int); static const char *fmt_sockaddr(struct sockaddr *sa, struct sockaddr *mask, int flags); static void p_flags(int, const char *); static const char *fmt_flags(int f); -static void domask(char *, in_addr_t, u_long); +static void domask(char *, size_t, u_long); /* * Print routing tables. */ void routepr(int fibnum, int af) { size_t intsize; int numfibs; if (live == 0) return; intsize = sizeof(int); if (fibnum == -1 && sysctlbyname("net.my_fibnum", &fibnum, &intsize, NULL, 0) == -1) fibnum = 0; if (sysctlbyname("net.fibs", &numfibs, &intsize, NULL, 0) == -1) numfibs = 1; if (fibnum < 0 || fibnum > numfibs - 1) errx(EX_USAGE, "%d: invalid fib", fibnum); /* * Since kernel & userland use different timebase * (time_uptime vs time_second) and we are reading kernel memory * directly we should do rt_expire --> expire_time conversion. */ if (clock_gettime(CLOCK_UPTIME, &uptime) < 0) err(EX_OSERR, "clock_gettime() failed"); xo_open_container("route-information"); xo_emit("{T:Routing tables}"); if (fibnum) xo_emit(" ({L:fib}: {:fib/%d})", fibnum); xo_emit("\n"); p_rtable_sysctl(fibnum, af); xo_close_container("route-information"); } /* * Print address family header before a section of the routing table. */ void pr_family(int af1) { const char *afname; switch (af1) { case AF_INET: afname = "Internet"; break; #ifdef INET6 case AF_INET6: afname = "Internet6"; break; #endif /*INET6*/ case AF_ISO: afname = "ISO"; break; case AF_CCITT: afname = "X.25"; break; case AF_NETGRAPH: afname = "Netgraph"; break; default: afname = NULL; break; } if (afname) xo_emit("\n{k:address-family/%s}:\n", afname); else xo_emit("\n{L:Protocol Family} {k:address-family/%d}:\n", af1); } /* column widths; each followed by one space */ #ifndef INET6 #define WID_DST_DEFAULT(af) 18 /* width of destination column */ #define WID_GW_DEFAULT(af) 18 /* width of gateway column */ #define WID_IF_DEFAULT(af) (Wflag ? 10 : 8) /* width of netif column */ #else #define WID_DST_DEFAULT(af) \ ((af) == AF_INET6 ? (numeric_addr ? 33: 18) : 18) #define WID_GW_DEFAULT(af) \ ((af) == AF_INET6 ? (numeric_addr ? 29 : 18) : 18) #define WID_IF_DEFAULT(af) ((af) == AF_INET6 ? 8 : (Wflag ? 10 : 8)) #endif /*INET6*/ static int wid_dst; static int wid_gw; static int wid_flags; static int wid_pksent; static int wid_mtu; static int wid_if; static int wid_expire; /* * Print header for routing table columns. */ static void pr_rthdr(int af1 __unused) { if (Wflag) { xo_emit("{T:/%-*.*s} {T:/%-*.*s} {T:/%-*.*s} {T:/%*.*s} " "{T:/%*.*s} {T:/%*.*s} {T:/%*s}\n", wid_dst, wid_dst, "Destination", wid_gw, wid_gw, "Gateway", wid_flags, wid_flags, "Flags", wid_pksent, wid_pksent, "Use", wid_mtu, wid_mtu, "Mtu", wid_if, wid_if, "Netif", wid_expire, "Expire"); } else { xo_emit("{T:/%-*.*s} {T:/%-*.*s} {T:/%-*.*s} {T:/%*.*s} " "{T:/%*s}\n", wid_dst, wid_dst, "Destination", wid_gw, wid_gw, "Gateway", wid_flags, wid_flags, "Flags", wid_if, wid_if, "Netif", wid_expire, "Expire"); } } static void p_rtable_sysctl(int fibnum, int af) { size_t needed; int mib[7]; char *buf, *next, *lim; struct rt_msghdr *rtm; struct sockaddr *sa; int fam = AF_UNSPEC, ifindex = 0, size; int need_table_close = false; struct ifaddrs *ifap, *ifa; struct sockaddr_dl *sdl; /* * Retrieve interface list at first * since we need #ifindex -> if_xname match */ if (getifaddrs(&ifap) != 0) err(EX_OSERR, "getifaddrs"); for (ifa = ifap; ifa; ifa = ifa->ifa_next) { if (ifa->ifa_addr->sa_family != AF_LINK) continue; sdl = (struct sockaddr_dl *)ifa->ifa_addr; ifindex = sdl->sdl_index; if (ifindex >= ifmap_size) { size = roundup(ifindex + 1, 32) * sizeof(struct ifmap_entry); if ((ifmap = realloc(ifmap, size)) == NULL) errx(2, "realloc(%d) failed", size); memset(&ifmap[ifmap_size], 0, size - ifmap_size * sizeof(struct ifmap_entry)); ifmap_size = roundup(ifindex + 1, 32); } if (*ifmap[ifindex].ifname != '\0') continue; strlcpy(ifmap[ifindex].ifname, ifa->ifa_name, IFNAMSIZ); } freeifaddrs(ifap); mib[0] = CTL_NET; mib[1] = PF_ROUTE; mib[2] = 0; mib[3] = af; mib[4] = NET_RT_DUMP; mib[5] = 0; mib[6] = fibnum; if (sysctl(mib, nitems(mib), NULL, &needed, NULL, 0) < 0) err(EX_OSERR, "sysctl: net.route.0.%d.dump.%d estimate", af, fibnum); if ((buf = malloc(needed)) == NULL) errx(2, "malloc(%lu)", (unsigned long)needed); if (sysctl(mib, nitems(mib), buf, &needed, NULL, 0) < 0) err(1, "sysctl: net.route.0.%d.dump.%d", af, fibnum); lim = buf + needed; xo_open_container("route-table"); xo_open_list("rt-family"); for (next = buf; next < lim; next += rtm->rtm_msglen) { rtm = (struct rt_msghdr *)next; if (rtm->rtm_version != RTM_VERSION) continue; /* * Peek inside header to determine AF */ sa = (struct sockaddr *)(rtm + 1); /* Only print family first time. */ if (fam != sa->sa_family) { if (need_table_close) { xo_close_list("rt-entry"); xo_close_instance("rt-family"); } need_table_close = true; fam = sa->sa_family; wid_dst = WID_DST_DEFAULT(fam); wid_gw = WID_GW_DEFAULT(fam); wid_flags = 6; wid_pksent = 8; wid_mtu = 6; wid_if = WID_IF_DEFAULT(fam); wid_expire = 6; xo_open_instance("rt-family"); pr_family(fam); xo_open_list("rt-entry"); pr_rthdr(fam); } p_rtentry_sysctl("rt-entry", rtm); } if (need_table_close) { xo_close_list("rt-entry"); xo_close_instance("rt-family"); } xo_close_list("rt-family"); xo_close_container("route-table"); free(buf); } static void p_rtentry_sysctl(const char *name, struct rt_msghdr *rtm) { struct sockaddr *sa, *addr[RTAX_MAX]; char buffer[128]; char prettyname[128]; int i, protrusion; xo_open_instance(name); sa = (struct sockaddr *)(rtm + 1); for (i = 0; i < RTAX_MAX; i++) { if (rtm->rtm_addrs & (1 << i)) addr[i] = sa; sa = (struct sockaddr *)((char *)sa + SA_SIZE(sa)); } protrusion = p_sockaddr("destination", addr[RTAX_DST], addr[RTAX_NETMASK], rtm->rtm_flags, wid_dst); protrusion = p_sockaddr("gateway", addr[RTAX_GATEWAY], NULL, RTF_HOST, wid_gw - protrusion); snprintf(buffer, sizeof(buffer), "{[:-%d}{:flags/%%s}{]:} ", wid_flags - protrusion); p_flags(rtm->rtm_flags, buffer); if (Wflag) { xo_emit("{t:use/%*lu} ", wid_pksent, rtm->rtm_rmx.rmx_pksent); if (rtm->rtm_rmx.rmx_mtu != 0) xo_emit("{t:mtu/%*lu} ", wid_mtu, rtm->rtm_rmx.rmx_mtu); else xo_emit("{P:/%*s} ", wid_mtu, ""); } memset(prettyname, 0, sizeof(prettyname)); if (rtm->rtm_index < ifmap_size) { strlcpy(prettyname, ifmap[rtm->rtm_index].ifname, sizeof(prettyname)); if (*prettyname == '\0') strlcpy(prettyname, "---", sizeof(prettyname)); } if (Wflag) xo_emit("{t:interface-name/%*s}", wid_if, prettyname); else xo_emit("{t:interface-name/%*.*s}", wid_if, wid_if, prettyname); if (rtm->rtm_rmx.rmx_expire) { time_t expire_time; if ((expire_time = rtm->rtm_rmx.rmx_expire - uptime.tv_sec) > 0) xo_emit(" {:expire-time/%*d}", wid_expire, (int)expire_time); } xo_emit("\n"); xo_close_instance(name); } static int p_sockaddr(const char *name, struct sockaddr *sa, struct sockaddr *mask, int flags, int width) { const char *cp; char buf[128]; int protrusion; cp = fmt_sockaddr(sa, mask, flags); if (width < 0) { snprintf(buf, sizeof(buf), "{:%s/%%s} ", name); xo_emit(buf, cp); protrusion = 0; } else { if (Wflag != 0 || numeric_addr) { snprintf(buf, sizeof(buf), "{[:%d}{:%s/%%s}{]:} ", -width, name); xo_emit(buf, cp); protrusion = strlen(cp) - width; if (protrusion < 0) protrusion = 0; } else { snprintf(buf, sizeof(buf), "{[:%d}{:%s/%%-.*s}{]:} ", -width, name); xo_emit(buf, width, cp); protrusion = 0; } } return (protrusion); } static const char * fmt_sockaddr(struct sockaddr *sa, struct sockaddr *mask, int flags) { static char buf[128]; const char *cp; if (sa == NULL) return ("null"); switch(sa->sa_family) { #ifdef INET6 case AF_INET6: /* * The sa6->sin6_scope_id must be filled here because * this sockaddr is extracted from kmem(4) directly * and has KAME-specific embedded scope id in * sa6->sin6_addr.s6_addr[2]. */ in6_fillscopeid(satosin6(sa)); /* FALLTHROUGH */ #endif /*INET6*/ case AF_INET: if (flags & RTF_HOST) cp = routename(sa, numeric_addr); else if (mask) cp = netname(sa, mask); else cp = netname(sa, NULL); break; case AF_NETGRAPH: { strlcpy(buf, ((struct sockaddr_ng *)sa)->sg_data, sizeof(buf)); cp = buf; break; } case AF_LINK: { #if 0 struct sockaddr_dl *sdl = (struct sockaddr_dl *)sa; /* Interface route. */ if (sdl->sdl_nlen) cp = sdl->sdl_data; else #endif cp = routename(sa, 1); break; } default: { u_char *s = (u_char *)sa->sa_data, *slim; char *cq, *cqlim; cq = buf; slim = sa->sa_len + (u_char *) sa; - cqlim = cq + sizeof(buf) - 6; - cq += sprintf(cq, "(%d)", sa->sa_family); + cqlim = cq + sizeof(buf) - sizeof(" ffff"); + snprintf(cq, sizeof(cq), "(%d)", sa->sa_family); + cq += strlen(cq); while (s < slim && cq < cqlim) { - cq += sprintf(cq, " %02x", *s++); - if (s < slim) - cq += sprintf(cq, "%02x", *s++); + snprintf(cq, sizeof(" ff"), " %02x", *s++); + cq += strlen(cq); + if (s < slim) { + snprintf(cq, sizeof("ff"), "%02x", *s++); + cq += strlen(cq); + } } cp = buf; } } return (cp); } static void p_flags(int f, const char *format) { struct bits *p; xo_emit(format, fmt_flags(f)); xo_open_list("flags_pretty"); for (p = bits; p->b_mask; p++) if (p->b_mask & f) xo_emit("{le:flags_pretty/%s}", p->b_name); xo_close_list("flags_pretty"); } static const char * fmt_flags(int f) { static char name[33]; char *flags; struct bits *p = bits; for (flags = name; p->b_mask; p++) if (p->b_mask & f) *flags++ = p->b_val; *flags = '\0'; return (name); } char * routename(struct sockaddr *sa, int flags) { static char line[NI_MAXHOST]; int error, f; f = (flags) ? NI_NUMERICHOST : 0; error = getnameinfo(sa, sa->sa_len, line, sizeof(line), NULL, 0, f); if (error) { const void *src; switch (sa->sa_family) { #ifdef INET case AF_INET: src = &satosin(sa)->sin_addr; break; #endif /* INET */ #ifdef INET6 case AF_INET6: src = &satosin6(sa)->sin6_addr; break; #endif /* INET6 */ default: return(line); } inet_ntop(sa->sa_family, src, line, sizeof(line) - 1); return (line); } trimdomain(line, strlen(line)); return (line); } #define NSHIFT(m) ( \ (m) == IN_CLASSA_NET ? IN_CLASSA_NSHIFT : \ (m) == IN_CLASSB_NET ? IN_CLASSB_NSHIFT : \ (m) == IN_CLASSC_NET ? IN_CLASSC_NSHIFT : \ 0) static void -domask(char *dst, in_addr_t addr __unused, u_long mask) +domask(char *dst, size_t buflen, u_long mask) { int b, i; if (mask == 0) { *dst = '\0'; return; } i = 0; for (b = 0; b < 32; b++) if (mask & (1 << b)) { int bb; i = b; for (bb = b+1; bb < 32; bb++) if (!(mask & (1 << bb))) { i = -1; /* noncontig */ break; } break; } if (i == -1) - sprintf(dst, "&0x%lx", mask); + snprintf(dst, buflen, "&0x%lx", mask); else - sprintf(dst, "/%d", 32-i); + snprintf(dst, buflen, "/%d", 32-i); } /* * Return the name of the network whose address is given. */ const char * netname(struct sockaddr *sa, struct sockaddr *mask) { switch (sa->sa_family) { case AF_INET: if (mask != NULL) return (netname4(satosin(sa)->sin_addr.s_addr, satosin(mask)->sin_addr.s_addr)); else return (netname4(satosin(sa)->sin_addr.s_addr, INADDR_ANY)); break; #ifdef INET6 case AF_INET6: return (netname6(satosin6(sa), satosin6(mask))); #endif /* INET6 */ default: return (NULL); } } static const char * netname4(in_addr_t in, in_addr_t mask) { char *cp = 0; - static char line[MAXHOSTNAMELEN + sizeof("/xx")]; + static char line[MAXHOSTNAMELEN + sizeof("&0xffffffff")]; char nline[INET_ADDRSTRLEN]; struct netent *np = 0; in_addr_t i; if (in == INADDR_ANY && mask == 0) { strlcpy(line, "default", sizeof(line)); return (line); } /* It is ok to supply host address. */ in &= mask; i = ntohl(in); if (!numeric_addr && i) { np = getnetbyaddr(i >> NSHIFT(ntohl(mask)), AF_INET); if (np != NULL) { cp = np->n_name; trimdomain(cp, strlen(cp)); } } if (cp != NULL) strlcpy(line, cp, sizeof(line)); else { inet_ntop(AF_INET, &in, nline, sizeof(nline)); strlcpy(line, nline, sizeof(line)); - domask(line + strlen(line), i, ntohl(mask)); + domask(line + strlen(line), sizeof(line) - strlen(line), ntohl(mask)); } return (line); } #undef NSHIFT #ifdef INET6 void in6_fillscopeid(struct sockaddr_in6 *sa6) { #if defined(__KAME__) /* * XXX: This is a special workaround for KAME kernels. * sin6_scope_id field of SA should be set in the future. */ if (IN6_IS_ADDR_LINKLOCAL(&sa6->sin6_addr) || IN6_IS_ADDR_MC_NODELOCAL(&sa6->sin6_addr) || IN6_IS_ADDR_MC_LINKLOCAL(&sa6->sin6_addr)) { if (sa6->sin6_scope_id == 0) sa6->sin6_scope_id = ntohs(*(u_int16_t *)&sa6->sin6_addr.s6_addr[2]); sa6->sin6_addr.s6_addr[2] = sa6->sin6_addr.s6_addr[3] = 0; } #endif } /* Mask to length table. To check an invalid value, (length + 1) is used. */ -static int masktolen[256] = { +static const u_char masktolen[256] = { [0xff] = 8 + 1, [0xfe] = 7 + 1, [0xfc] = 6 + 1, [0xf8] = 5 + 1, [0xf0] = 4 + 1, [0xe0] = 3 + 1, [0xc0] = 2 + 1, [0x80] = 1 + 1, [0x00] = 0 + 1, }; static const char * netname6(struct sockaddr_in6 *sa6, struct sockaddr_in6 *mask) { static char line[NI_MAXHOST + sizeof("/xxx") - 1]; struct sockaddr_in6 addr; char nline[NI_MAXHOST]; + char maskbuf[sizeof("/xxx")]; u_char *p, *lim; - int masklen, illegal = 0, i; + u_char masklen; + int i; + bool illegal = false; if (mask) { p = (u_char *)&mask->sin6_addr; for (masklen = 0, lim = p + 16; p < lim; p++) { - if (masktolen[*p] > 0) + if (masktolen[*p] > 0) { /* -1 is required. */ - masklen += masktolen[*p] - 1; - else - illegal++; + masklen += (masktolen[*p] - 1); + } else + illegal = true; } if (illegal) xo_error("illegal prefixlen\n"); memcpy(&addr, sa6, sizeof(addr)); for (i = 0; i < 16; ++i) addr.sin6_addr.s6_addr[i] &= mask->sin6_addr.s6_addr[i]; sa6 = &addr; } else masklen = 128; if (masklen == 0 && IN6_IS_ADDR_UNSPECIFIED(&sa6->sin6_addr)) return("default"); getnameinfo((struct sockaddr *)sa6, sa6->sin6_len, nline, sizeof(nline), NULL, 0, NI_NUMERICHOST); if (numeric_addr) strlcpy(line, nline, sizeof(line)); else getnameinfo((struct sockaddr *)sa6, sa6->sin6_len, line, sizeof(line), NULL, 0, 0); - if (numeric_addr || strcmp(line, nline) == 0) - sprintf(&line[strlen(line)], "/%d", masklen); + if (numeric_addr || strcmp(line, nline) == 0) { + snprintf(maskbuf, sizeof(maskbuf), "/%d", masklen); + strlcat(line, maskbuf, sizeof(line)); + } return (line); } #endif /*INET6*/ /* * Print routing statistics */ void rt_stats(void) { struct rtstat rtstat; u_long rtsaddr, rttaddr; int rttrash; if ((rtsaddr = nl[N_RTSTAT].n_value) == 0) { xo_emit("{W:rtstat: symbol not in namelist}\n"); return; } if ((rttaddr = nl[N_RTTRASH].n_value) == 0) { xo_emit("{W:rttrash: symbol not in namelist}\n"); return; } kread(rtsaddr, (char *)&rtstat, sizeof (rtstat)); kread(rttaddr, (char *)&rttrash, sizeof (rttrash)); xo_emit("{T:routing}:\n"); #define p(f, m) if (rtstat.f || sflag <= 1) \ xo_emit(m, rtstat.f, plural(rtstat.f)) p(rts_badredirect, "\t{:bad-redirects/%hu} " "{N:/bad routing redirect%s}\n"); p(rts_dynamic, "\t{:dynamically-created/%hu} " "{N:/dynamically created route%s}\n"); p(rts_newgateway, "\t{:new-gateways/%hu} " "{N:/new gateway%s due to redirects}\n"); p(rts_unreach, "\t{:unreachable-destination/%hu} " "{N:/destination%s found unreachable}\n"); p(rts_wildcard, "\t{:wildcard-uses/%hu} " "{N:/use%s of a wildcard route}\n"); #undef p if (rttrash || sflag <= 1) xo_emit("\t{:unused-but-not-freed/%u} " "{N:/route%s not in table but not freed}\n", rttrash, plural(rttrash)); } Index: projects/clang400-import/usr.bin/netstat/sctp.c =================================================================== --- projects/clang400-import/usr.bin/netstat/sctp.c (revision 311459) +++ projects/clang400-import/usr.bin/netstat/sctp.c (revision 311460) @@ -1,843 +1,841 @@ /*- * Copyright (c) 2001-2007, by Weongyo Jeong. All rights reserved. * Copyright (c) 2011, by Michael Tuexen. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions are met: * * a) Redistributions of source code must retain the above copyright notice, * this list of conditions and the following disclaimer. * * b) 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. * * c) Neither the name of Cisco Systems, Inc. nor the names of its * contributors may be used to endorse or promote products derived * from this software without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF * THE POSSIBILITY OF SUCH DAMAGE. */ #if 0 #ifndef lint static char sccsid[] = "@(#)sctp.c 0.1 (Berkeley) 4/18/2007"; #endif /* not lint */ #endif #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "netstat.h" #include #ifdef SCTP static void sctp_statesprint(uint32_t state); #define NETSTAT_SCTP_STATES_CLOSED 0x0 #define NETSTAT_SCTP_STATES_BOUND 0x1 #define NETSTAT_SCTP_STATES_LISTEN 0x2 #define NETSTAT_SCTP_STATES_COOKIE_WAIT 0x3 #define NETSTAT_SCTP_STATES_COOKIE_ECHOED 0x4 #define NETSTAT_SCTP_STATES_ESTABLISHED 0x5 #define NETSTAT_SCTP_STATES_SHUTDOWN_SENT 0x6 #define NETSTAT_SCTP_STATES_SHUTDOWN_RECEIVED 0x7 #define NETSTAT_SCTP_STATES_SHUTDOWN_ACK_SENT 0x8 #define NETSTAT_SCTP_STATES_SHUTDOWN_PENDING 0x9 static const char *sctpstates[] = { "CLOSED", "BOUND", "LISTEN", "COOKIE_WAIT", "COOKIE_ECHOED", "ESTABLISHED", "SHUTDOWN_SENT", "SHUTDOWN_RECEIVED", "SHUTDOWN_ACK_SENT", "SHUTDOWN_PENDING" }; static LIST_HEAD(xladdr_list, xladdr_entry) xladdr_head; struct xladdr_entry { struct xsctp_laddr *xladdr; LIST_ENTRY(xladdr_entry) xladdr_entries; }; static LIST_HEAD(xraddr_list, xraddr_entry) xraddr_head; struct xraddr_entry { struct xsctp_raddr *xraddr; LIST_ENTRY(xraddr_entry) xraddr_entries; }; -#ifdef INET -char * -inetname(struct in_addr *inp); -#endif - -#ifdef INET6 -char * -inet6name(struct in6_addr *in6p); -#endif - static void sctp_print_address(const char *container, union sctp_sockstore *address, int port, int num_port) { struct servent *sp = 0; char line[80], *cp; int width; + size_t alen, plen; if (container) xo_open_container(container); switch (address->sa.sa_family) { #ifdef INET case AF_INET: - sprintf(line, "%.*s.", Wflag ? 39 : 16, inetname(&address->sin.sin_addr)); + snprintf(line, sizeof(line), "%.*s.", + Wflag ? 39 : 16, inetname(&address->sin.sin_addr)); break; #endif #ifdef INET6 case AF_INET6: - sprintf(line, "%.*s.", Wflag ? 39 : 16, inet6name(&address->sin6.sin6_addr)); + snprintf(line, sizeof(line), "%.*s.", + Wflag ? 39 : 16, inet6name(&address->sin6.sin6_addr)); break; #endif default: - sprintf(line, "%.*s.", Wflag ? 39 : 16, ""); + snprintf(line, sizeof(line), "%.*s.", + Wflag ? 39 : 16, ""); break; } - cp = strchr(line, '\0'); + alen = strlen(line); + cp = line + alen; if (!num_port && port) sp = getservbyport((int)port, "sctp"); if (sp || port == 0) - sprintf(cp, "%.15s ", sp ? sp->s_name : "*"); + snprintf(cp, sizeof(line) - alen, + "%.15s ", sp ? sp->s_name : "*"); else - sprintf(cp, "%d ", ntohs((u_short)port)); + snprintf(cp, sizeof(line) - alen, + "%d ", ntohs((u_short)port)); width = Wflag ? 45 : 22; xo_emit("{d:target/%-*.*s} ", width, width, line); - int alen = cp - line - 1, plen = strlen(cp) - 1; + plen = strlen(cp) - 1; + alen--; xo_emit("{e:address/%*.*s}{e:port/%*.*s}", alen, alen, line, plen, plen, cp); if (container) xo_close_container(container); } static int sctp_skip_xinpcb_ifneed(char *buf, const size_t buflen, size_t *offset) { int exist_tcb = 0; struct xsctp_tcb *xstcb; struct xsctp_raddr *xraddr; struct xsctp_laddr *xladdr; while (*offset < buflen) { xladdr = (struct xsctp_laddr *)(buf + *offset); *offset += sizeof(struct xsctp_laddr); if (xladdr->last == 1) break; } while (*offset < buflen) { xstcb = (struct xsctp_tcb *)(buf + *offset); *offset += sizeof(struct xsctp_tcb); if (xstcb->last == 1) break; exist_tcb = 1; while (*offset < buflen) { xladdr = (struct xsctp_laddr *)(buf + *offset); *offset += sizeof(struct xsctp_laddr); if (xladdr->last == 1) break; } while (*offset < buflen) { xraddr = (struct xsctp_raddr *)(buf + *offset); *offset += sizeof(struct xsctp_raddr); if (xraddr->last == 1) break; } } /* * If Lflag is set, we don't care about the return value. */ if (Lflag) return 0; return exist_tcb; } static void sctp_process_tcb(struct xsctp_tcb *xstcb, char *buf, const size_t buflen, size_t *offset, int *indent) { int i, xl_total = 0, xr_total = 0, x_max; struct xsctp_raddr *xraddr; struct xsctp_laddr *xladdr; struct xladdr_entry *prev_xl = NULL, *xl = NULL, *xl_tmp; struct xraddr_entry *prev_xr = NULL, *xr = NULL, *xr_tmp; LIST_INIT(&xladdr_head); LIST_INIT(&xraddr_head); /* * Make `struct xladdr_list' list and `struct xraddr_list' list * to handle the address flexibly. */ while (*offset < buflen) { xladdr = (struct xsctp_laddr *)(buf + *offset); *offset += sizeof(struct xsctp_laddr); if (xladdr->last == 1) break; prev_xl = xl; xl = malloc(sizeof(struct xladdr_entry)); if (xl == NULL) { xo_warnx("malloc %lu bytes", (u_long)sizeof(struct xladdr_entry)); goto out; } xl->xladdr = xladdr; if (prev_xl == NULL) LIST_INSERT_HEAD(&xladdr_head, xl, xladdr_entries); else LIST_INSERT_AFTER(prev_xl, xl, xladdr_entries); xl_total++; } while (*offset < buflen) { xraddr = (struct xsctp_raddr *)(buf + *offset); *offset += sizeof(struct xsctp_raddr); if (xraddr->last == 1) break; prev_xr = xr; xr = malloc(sizeof(struct xraddr_entry)); if (xr == NULL) { xo_warnx("malloc %lu bytes", (u_long)sizeof(struct xraddr_entry)); goto out; } xr->xraddr = xraddr; if (prev_xr == NULL) LIST_INSERT_HEAD(&xraddr_head, xr, xraddr_entries); else LIST_INSERT_AFTER(prev_xr, xr, xraddr_entries); xr_total++; } /* * Let's print the address infos. */ xo_open_list("address"); xl = LIST_FIRST(&xladdr_head); xr = LIST_FIRST(&xraddr_head); x_max = MAX(xl_total, xr_total); for (i = 0; i < x_max; i++) { xo_open_instance("address"); if (((*indent == 0) && i > 0) || *indent > 0) xo_emit("{P:/%-12s} ", " "); if (xl != NULL) { sctp_print_address("local", &(xl->xladdr->address), htons(xstcb->local_port), numeric_port); } else { if (Wflag) { xo_emit("{P:/%-45s} ", " "); } else { xo_emit("{P:/%-22s} ", " "); } } if (xr != NULL && !Lflag) { sctp_print_address("remote", &(xr->xraddr->address), htons(xstcb->remote_port), numeric_port); } if (xl != NULL) xl = LIST_NEXT(xl, xladdr_entries); if (xr != NULL) xr = LIST_NEXT(xr, xraddr_entries); if (i == 0 && !Lflag) sctp_statesprint(xstcb->state); if (i < x_max) xo_emit("\n"); xo_close_instance("address"); } out: /* * Free the list which be used to handle the address. */ xl = LIST_FIRST(&xladdr_head); while (xl != NULL) { xl_tmp = LIST_NEXT(xl, xladdr_entries); free(xl); xl = xl_tmp; } xr = LIST_FIRST(&xraddr_head); while (xr != NULL) { xr_tmp = LIST_NEXT(xr, xraddr_entries); free(xr); xr = xr_tmp; } } static void sctp_process_inpcb(struct xsctp_inpcb *xinpcb, char *buf, const size_t buflen, size_t *offset) { int indent = 0, xladdr_total = 0, is_listening = 0; static int first = 1; const char *tname, *pname; struct xsctp_tcb *xstcb; struct xsctp_laddr *xladdr; size_t offset_laddr; int process_closed; if (xinpcb->maxqlen > 0) is_listening = 1; if (first) { if (!Lflag) { xo_emit("Active SCTP associations"); if (aflag) xo_emit(" (including servers)"); } else xo_emit("Current listen queue sizes (qlen/maxqlen)"); xo_emit("\n"); if (Lflag) xo_emit("{T:/%-6.6s} {T:/%-5.5s} {T:/%-8.8s} " "{T:/%-22.22s}\n", "Proto", "Type", "Listen", "Local Address"); else if (Wflag) xo_emit("{T:/%-6.6s} {T:/%-5.5s} {T:/%-45.45s} " "{T:/%-45.45s} {T:/%s}\n", "Proto", "Type", "Local Address", "Foreign Address", "(state)"); else xo_emit("{T:/%-6.6s} {T:/%-5.5s} {T:/%-22.22s} " "{T:/%-22.22s} {T:/%s}\n", "Proto", "Type", "Local Address", "Foreign Address", "(state)"); first = 0; } xladdr = (struct xsctp_laddr *)(buf + *offset); if ((!aflag && is_listening) || (Lflag && !is_listening)) { sctp_skip_xinpcb_ifneed(buf, buflen, offset); return; } if (xinpcb->flags & SCTP_PCB_FLAGS_BOUND_V6) { /* Can't distinguish between sctp46 and sctp6 */ pname = "sctp46"; } else { pname = "sctp4"; } if (xinpcb->flags & SCTP_PCB_FLAGS_TCPTYPE) tname = "1to1"; else if (xinpcb->flags & SCTP_PCB_FLAGS_UDPTYPE) tname = "1toN"; else tname = "????"; if (Lflag) { char buf1[22]; snprintf(buf1, sizeof buf1, "%u/%u", xinpcb->qlen, xinpcb->maxqlen); xo_emit("{:protocol/%-6.6s/%s} {:type/%-5.5s/%s} ", pname, tname); xo_emit("{d:queues/%-8.8s}{e:queue-len/%hu}" "{e:max-queue-len/%hu} ", buf1, xinpcb->qlen, xinpcb->maxqlen); } offset_laddr = *offset; process_closed = 0; xo_open_list("local-address"); retry: while (*offset < buflen) { xladdr = (struct xsctp_laddr *)(buf + *offset); *offset += sizeof(struct xsctp_laddr); if (xladdr->last) { if (aflag && !Lflag && (xladdr_total == 0) && process_closed) { xo_open_instance("local-address"); xo_emit("{:protocol/%-6.6s/%s} " "{:type/%-5.5s/%s} ", pname, tname); if (Wflag) { xo_emit("{P:/%-91.91s/%s} " "{:state/CLOSED}", " "); } else { xo_emit("{P:/%-45.45s/%s} " "{:state/CLOSED}", " "); } xo_close_instance("local-address"); } if (process_closed || is_listening) { xo_emit("\n"); } break; } if (!Lflag && !is_listening && !process_closed) continue; xo_open_instance("local-address"); if (xladdr_total == 0) { if (!Lflag) { xo_emit("{:protocol/%-6.6s/%s} " "{:type/%-5.5s/%s} ", pname, tname); } } else { xo_emit("\n"); xo_emit(Lflag ? "{P:/%-21.21s} " : "{P:/%-12.12s} ", " "); } sctp_print_address("local", &(xladdr->address), htons(xinpcb->local_port), numeric_port); if (aflag && !Lflag && xladdr_total == 0) { if (Wflag) { if (process_closed) { xo_emit("{P:/%-45.45s} " "{:state/CLOSED}", " "); } else { xo_emit("{P:/%-45.45s} " "{:state/LISTEN}", " "); } } else { if (process_closed) { xo_emit("{P:/%-22.22s} " "{:state/CLOSED}", " "); } else { xo_emit("{P:/%-22.22s} " "{:state/LISTEN}", " "); } } } xladdr_total++; xo_close_instance("local-address"); } xstcb = (struct xsctp_tcb *)(buf + *offset); *offset += sizeof(struct xsctp_tcb); if (aflag && (xladdr_total == 0) && xstcb->last && !process_closed) { process_closed = 1; *offset = offset_laddr; goto retry; } while (xstcb->last == 0 && *offset < buflen) { xo_emit("{:protocol/%-6.6s/%s} {:type/%-5.5s/%s} ", pname, tname); sctp_process_tcb(xstcb, buf, buflen, offset, &indent); indent++; xstcb = (struct xsctp_tcb *)(buf + *offset); *offset += sizeof(struct xsctp_tcb); } xo_close_list("local-address"); } /* * Print a summary of SCTP connections related to an Internet * protocol. */ void sctp_protopr(u_long off __unused, const char *name __unused, int af1 __unused, int proto) { char *buf; const char *mibvar = "net.inet.sctp.assoclist"; size_t offset = 0; size_t len = 0; struct xsctp_inpcb *xinpcb; if (proto != IPPROTO_SCTP) return; if (sysctlbyname(mibvar, 0, &len, 0, 0) < 0) { if (errno != ENOENT) xo_warn("sysctl: %s", mibvar); return; } if ((buf = malloc(len)) == NULL) { xo_warnx("malloc %lu bytes", (u_long)len); return; } if (sysctlbyname(mibvar, buf, &len, 0, 0) < 0) { xo_warn("sysctl: %s", mibvar); free(buf); return; } xinpcb = (struct xsctp_inpcb *)(buf + offset); offset += sizeof(struct xsctp_inpcb); while (xinpcb->last == 0 && offset < len) { sctp_process_inpcb(xinpcb, buf, (const size_t)len, &offset); xinpcb = (struct xsctp_inpcb *)(buf + offset); offset += sizeof(struct xsctp_inpcb); } free(buf); } static void sctp_statesprint(uint32_t state) { int idx; switch (state) { case SCTP_CLOSED: idx = NETSTAT_SCTP_STATES_CLOSED; break; case SCTP_BOUND: idx = NETSTAT_SCTP_STATES_BOUND; break; case SCTP_LISTEN: idx = NETSTAT_SCTP_STATES_LISTEN; break; case SCTP_COOKIE_WAIT: idx = NETSTAT_SCTP_STATES_COOKIE_WAIT; break; case SCTP_COOKIE_ECHOED: idx = NETSTAT_SCTP_STATES_COOKIE_ECHOED; break; case SCTP_ESTABLISHED: idx = NETSTAT_SCTP_STATES_ESTABLISHED; break; case SCTP_SHUTDOWN_SENT: idx = NETSTAT_SCTP_STATES_SHUTDOWN_SENT; break; case SCTP_SHUTDOWN_RECEIVED: idx = NETSTAT_SCTP_STATES_SHUTDOWN_RECEIVED; break; case SCTP_SHUTDOWN_ACK_SENT: idx = NETSTAT_SCTP_STATES_SHUTDOWN_ACK_SENT; break; case SCTP_SHUTDOWN_PENDING: idx = NETSTAT_SCTP_STATES_SHUTDOWN_PENDING; break; default: xo_emit("UNKNOWN {:state/0x%08x}", state); return; } xo_emit("{:state/%s}", sctpstates[idx]); } /* * Dump SCTP statistics structure. */ void sctp_stats(u_long off, const char *name, int af1 __unused, int proto __unused) { struct sctpstat sctpstat; if (fetch_stats("net.inet.sctp.stats", off, &sctpstat, sizeof(sctpstat), kread) != 0) return; xo_open_container(name); xo_emit("{T:/%s}:\n", name); #define p(f, m) if (sctpstat.f || sflag <= 1) \ xo_emit(m, (uintmax_t)sctpstat.f, plural(sctpstat.f)) #define p1a(f, m) if (sctpstat.f || sflag <= 1) \ xo_emit(m, (uintmax_t)sctpstat.f) /* * input statistics */ p(sctps_recvpackets, "\t{:received-packets/%ju} " "{N:/input packet%s}\n"); p(sctps_recvdatagrams, "\t\t{:received-datagrams/%ju} " "{N:/datagram%s}\n"); p(sctps_recvpktwithdata, "\t\t{:received-with-data/%ju} " "{N:/packet%s that had data}\n"); p(sctps_recvsacks, "\t\t{:received-sack-chunks/%ju} " "{N:/input SACK chunk%s}\n"); p(sctps_recvdata, "\t\t{:received-data-chunks/%ju} " "{N:/input DATA chunk%s}\n"); p(sctps_recvdupdata, "\t\t{:received-duplicate-data-chunks/%ju} " "{N:/duplicate DATA chunk%s}\n"); p(sctps_recvheartbeat, "\t\t{:received-hb-chunks/%ju} " "{N:/input HB chunk%s}\n"); p(sctps_recvheartbeatack, "\t\t{:received-hb-ack-chunks/%ju} " "{N:/HB-ACK chunk%s}\n"); p(sctps_recvecne, "\t\t{:received-ecne-chunks/%ju} " "{N:/input ECNE chunk%s}\n"); p(sctps_recvauth, "\t\t{:received-auth-chunks/%ju} " "{N:/input AUTH chunk%s}\n"); p(sctps_recvauthmissing, "\t\t{:dropped-missing-auth/%ju} " "{N:/chunk%s missing AUTH}\n"); p(sctps_recvivalhmacid, "\t\t{:dropped-invalid-hmac/%ju} " "{N:/invalid HMAC id%s received}\n"); p(sctps_recvivalkeyid, "\t\t{:dropped-invalid-secret/%ju} " "{N:/invalid secret id%s received}\n"); p1a(sctps_recvauthfailed, "\t\t{:dropped-auth-failed/%ju} " "{N:/auth failed}\n"); p1a(sctps_recvexpress, "\t\t{:received-fast-path/%ju} " "{N:/fast path receives all one chunk}\n"); p1a(sctps_recvexpressm, "\t\t{:receives-fast-path-multipart/%ju} " "{N:/fast path multi-part data}\n"); /* * output statistics */ p(sctps_sendpackets, "\t{:sent-packets/%ju} " "{N:/output packet%s}\n"); p(sctps_sendsacks, "\t\t{:sent-sacks/%ju} " "{N:/output SACK%s}\n"); p(sctps_senddata, "\t\t{:sent-data-chunks/%ju} " "{N:/output DATA chunk%s}\n"); p(sctps_sendretransdata, "\t\t{:sent-retransmitted-data-chunks/%ju} " "{N:/retransmitted DATA chunk%s}\n"); p(sctps_sendfastretrans, "\t\t" "{:sent-fast-retransmitted-data-chunks/%ju} " "{N:/fast retransmitted DATA chunk%s}\n"); p(sctps_sendmultfastretrans, "\t\t" "{:sent-fast-retransmitted-data-chunk-multiple-times/%ju} " "{N:/FR'%s that happened more than once to same chunk}\n"); p(sctps_sendheartbeat, "\t\t{:sent-hb-chunks/%ju} " "{N:/output HB chunk%s}\n"); p(sctps_sendecne, "\t\t{:sent-ecne-chunks/%ju} " "{N:/output ECNE chunk%s}\n"); p(sctps_sendauth, "\t\t{:sent-auth-chunks/%ju} " "{N:/output AUTH chunk%s}\n"); p1a(sctps_senderrors, "\t\t{:send-errors/%ju} " "{N:/ip_output error counter}\n"); /* * PCKDROPREP statistics */ xo_emit("\t{T:Packet drop statistics}:\n"); xo_open_container("drop-statistics"); p1a(sctps_pdrpfmbox, "\t\t{:middle-box/%ju} " "{N:/from middle box}\n"); p1a(sctps_pdrpfehos, "\t\t{:end-host/%ju} " "{N:/from end host}\n"); p1a(sctps_pdrpmbda, "\t\t{:with-data/%ju} " "{N:/with data}\n"); p1a(sctps_pdrpmbct, "\t\t{:non-data/%ju} " "{N:/non-data, non-endhost}\n"); p1a(sctps_pdrpbwrpt, "\t\t{:non-endhost/%ju} " "{N:/non-endhost, bandwidth rep only}\n"); p1a(sctps_pdrpcrupt, "\t\t{:short-header/%ju} " "{N:/not enough for chunk header}\n"); p1a(sctps_pdrpnedat, "\t\t{:short-data/%ju} " "{N:/not enough data to confirm}\n"); p1a(sctps_pdrppdbrk, "\t\t{:chunk-break/%ju} " "{N:/where process_chunk_drop said break}\n"); p1a(sctps_pdrptsnnf, "\t\t{:tsn-not-found/%ju} " "{N:/failed to find TSN}\n"); p1a(sctps_pdrpdnfnd, "\t\t{:reverse-tsn/%ju} " "{N:/attempt reverse TSN lookup}\n"); p1a(sctps_pdrpdiwnp, "\t\t{:confirmed-zero-window/%ju} " "{N:/e-host confirms zero-rwnd}\n"); p1a(sctps_pdrpdizrw, "\t\t{:middle-box-no-space/%ju} " "{N:/midbox confirms no space}\n"); p1a(sctps_pdrpbadd, "\t\t{:bad-data/%ju} " "{N:/data did not match TSN}\n"); p(sctps_pdrpmark, "\t\t{:tsn-marked-fast-retransmission/%ju} " "{N:/TSN'%s marked for Fast Retran}\n"); xo_close_container("drop-statistics"); /* * Timeouts */ xo_emit("\t{T:Timeouts}:\n"); xo_open_container("timeouts"); p(sctps_timoiterator, "\t\t{:iterator/%ju} " "{N:/iterator timer%s fired}\n"); p(sctps_timodata, "\t\t{:t3-data/%ju} " "{N:/T3 data time out%s}\n"); p(sctps_timowindowprobe, "\t\t{:window-probe/%ju} " "{N:/window probe (T3) timer%s fired}\n"); p(sctps_timoinit, "\t\t{:init-timer/%ju} " "{N:/INIT timer%s fired}\n"); p(sctps_timosack, "\t\t{:sack-timer/%ju} " "{N:/sack timer%s fired}\n"); p(sctps_timoshutdown, "\t\t{:shutdown-timer/%ju} " "{N:/shutdown timer%s fired}\n"); p(sctps_timoheartbeat, "\t\t{:heartbeat-timer/%ju} " "{N:/heartbeat timer%s fired}\n"); p1a(sctps_timocookie, "\t\t{:cookie-timer/%ju} " "{N:/a cookie timeout fired}\n"); p1a(sctps_timosecret, "\t\t{:endpoint-changed-cookie/%ju} " "{N:/an endpoint changed its cook}ie" "secret\n"); p(sctps_timopathmtu, "\t\t{:pmtu-timer/%ju} " "{N:/PMTU timer%s fired}\n"); p(sctps_timoshutdownack, "\t\t{:shutdown-timer/%ju} " "{N:/shutdown ack timer%s fired}\n"); p(sctps_timoshutdownguard, "\t\t{:shutdown-guard-timer/%ju} " "{N:/shutdown guard timer%s fired}\n"); p(sctps_timostrmrst, "\t\t{:stream-reset-timer/%ju} " "{N:/stream reset timer%s fired}\n"); p(sctps_timoearlyfr, "\t\t{:early-fast-retransmission-timer/%ju} " "{N:/early FR timer%s fired}\n"); p1a(sctps_timoasconf, "\t\t{:asconf-timer/%ju} " "{N:/an asconf timer fired}\n"); p1a(sctps_timoautoclose, "\t\t{:auto-close-timer/%ju} " "{N:/auto close timer fired}\n"); p(sctps_timoassockill, "\t\t{:asoc-free-timer/%ju} " "{N:/asoc free timer%s expired}\n"); p(sctps_timoinpkill, "\t\t{:input-free-timer/%ju} " "{N:/inp free timer%s expired}\n"); xo_close_container("timeouts"); #if 0 /* * Early fast retransmission counters */ p(sctps_earlyfrstart, "\t%ju TODO:sctps_earlyfrstart\n"); p(sctps_earlyfrstop, "\t%ju TODO:sctps_earlyfrstop\n"); p(sctps_earlyfrmrkretrans, "\t%ju TODO:sctps_earlyfrmrkretrans\n"); p(sctps_earlyfrstpout, "\t%ju TODO:sctps_earlyfrstpout\n"); p(sctps_earlyfrstpidsck1, "\t%ju TODO:sctps_earlyfrstpidsck1\n"); p(sctps_earlyfrstpidsck2, "\t%ju TODO:sctps_earlyfrstpidsck2\n"); p(sctps_earlyfrstpidsck3, "\t%ju TODO:sctps_earlyfrstpidsck3\n"); p(sctps_earlyfrstpidsck4, "\t%ju TODO:sctps_earlyfrstpidsck4\n"); p(sctps_earlyfrstrid, "\t%ju TODO:sctps_earlyfrstrid\n"); p(sctps_earlyfrstrout, "\t%ju TODO:sctps_earlyfrstrout\n"); p(sctps_earlyfrstrtmr, "\t%ju TODO:sctps_earlyfrstrtmr\n"); #endif /* * Others */ p1a(sctps_hdrops, "\t{:dropped-too-short/%ju} " "{N:/packet shorter than header}\n"); p1a(sctps_badsum, "\t{:dropped-bad-checksum/%ju} " "{N:/checksum error}\n"); p1a(sctps_noport, "\t{:dropped-no-endpoint/%ju} " "{N:/no endpoint for port}\n"); p1a(sctps_badvtag, "\t{:dropped-bad-v-tag/%ju} " "{N:/bad v-tag}\n"); p1a(sctps_badsid, "\t{:dropped-bad-sid/%ju} " "{N:/bad SID}\n"); p1a(sctps_nomem, "\t{:dropped-no-memory/%ju} " "{N:/no memory}\n"); p1a(sctps_fastretransinrtt, "\t{:multiple-fast-retransmits-in-rtt/%ju} " "{N:/number of multiple FR in a RT}T window\n"); #if 0 p(sctps_markedretrans, "\t%ju TODO:sctps_markedretrans\n"); #endif p1a(sctps_naglesent, "\t{:rfc813-sent/%ju} " "{N:/RFC813 allowed sending}\n"); p1a(sctps_naglequeued, "\t{:rfc813-queued/%ju} " "{N:/RFC813 does not allow sending}\n"); p1a(sctps_maxburstqueued, "\t{:max-burst-queued/%ju} " "{N:/times max burst prohibited sending}\n"); p1a(sctps_ifnomemqueued, "\t{:no-memory-in-interface/%ju} " "{N:/look ahead tells us no memory in interface}\n"); p(sctps_windowprobed, "\t{:sent-window-probes/%ju} " "{N:/number%s of window probes sent}\n"); p(sctps_lowlevelerr, "\t{:low-level-err/%ju} " "{N:/time%s an output error to clamp down on next user send}\n"); p(sctps_lowlevelerrusr, "\t{:low-level-user-error/%ju} " "{N:/time%s sctp_senderrors were caused from a user}\n"); p(sctps_datadropchklmt, "\t{:dropped-chunk-limit/%ju} " "{N:/number of in data drop%s due to chunk limit reached}\n"); p(sctps_datadroprwnd, "\t{:dropped-rwnd-limit/%ju} " "{N:/number of in data drop%s due to rwnd limit reached}\n"); p(sctps_ecnereducedcwnd, "\t{:ecn-reduced-cwnd/%ju} " "{N:/time%s a ECN reduced the cwnd}\n"); p1a(sctps_vtagexpress, "\t{:v-tag-express-lookup/%ju} " "{N:/used express lookup via vtag}\n"); p1a(sctps_vtagbogus, "\t{:v-tag-collision/%ju} " "{N:/collision in express lookup}\n"); p(sctps_primary_randry, "\t{:sender-ran-dry/%ju} " "{N:/time%s the sender ran dry of user data on primary}\n"); p1a(sctps_cmt_randry, "\t{:cmt-ran-dry/%ju} " "{N:/same for above}\n"); p(sctps_slowpath_sack, "\t{:slow-path-sack/%ju} " "{N:/sack%s the slow way}\n"); p(sctps_wu_sacks_sent, "\t{:sent-window-update-only-sack/%ju} " "{N:/window update only sack%s sent}\n"); p(sctps_sends_with_flags, "\t{:sent-with-sinfo/%ju} " "{N:/send%s with sinfo_flags !=0}\n"); p(sctps_sends_with_unord, "\t{:sent-with-unordered/%ju} " "{N:/unordered send%s}\n"); p(sctps_sends_with_eof, "\t{:sent-with-eof/%ju} " "{N:/send%s with EOF flag set}\n"); p(sctps_sends_with_abort, "\t{:sent-with-abort/%ju} " "{N:/send%s with ABORT flag set}\n"); p(sctps_protocol_drain_calls, "\t{:protocol-drain-called/%ju} " "{N:/time%s protocol drain called}\n"); p(sctps_protocol_drains_done, "\t{:protocol-drain/%ju} " "{N:/time%s we did a protocol drain}\n"); p(sctps_read_peeks, "\t{:read-with-peek/%ju} " "{N:/time%s recv was called with peek}\n"); p(sctps_cached_chk, "\t{:cached-chunks/%ju} " "{N:/cached chunk%s used}\n"); p1a(sctps_cached_strmoq, "\t{:cached-output-queue-used/%ju} " "{N:/cached stream oq's used}\n"); p(sctps_left_abandon, "\t{:messages-abandoned/%ju} " "{N:/unread message%s abandonded by close}\n"); p1a(sctps_send_burst_avoid, "\t{:send-burst-avoidance/%ju} " "{N:/send burst avoidance, already max burst inflight to net}\n"); p1a(sctps_send_cwnd_avoid, "\t{:send-cwnd-avoidance/%ju} " "{N:/send cwnd full avoidance, already max burst inflight " "to net}\n"); p(sctps_fwdtsn_map_over, "\t{:tsn-map-overruns/%ju} " "{N:/number of map array over-run%s via fwd-tsn's}\n"); #undef p #undef p1a xo_close_container(name); } #endif /* SCTP */ Index: projects/clang400-import/usr.bin/netstat/unix.c =================================================================== --- projects/clang400-import/usr.bin/netstat/unix.c (revision 311459) +++ projects/clang400-import/usr.bin/netstat/unix.c (revision 311460) @@ -1,331 +1,331 @@ /*- * Copyright (c) 1983, 1988, 1993 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #if 0 #ifndef lint static char sccsid[] = "@(#)unix.c 8.1 (Berkeley) 6/6/93"; #endif /* not lint */ #endif #include __FBSDID("$FreeBSD$"); /* * Display protocol blocks in the unix domain. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "netstat.h" static void unixdomainpr(struct xunpcb *, struct xsocket *); static const char *const socktype[] = { "#0", "stream", "dgram", "raw", "rdm", "seqpacket" }; static int pcblist_sysctl(int type, char **bufp) { char *buf; size_t len; char mibvar[sizeof "net.local.seqpacket.pcblist"]; - sprintf(mibvar, "net.local.%s.pcblist", socktype[type]); + snprintf(mibvar, sizeof(mibvar), "net.local.%s.pcblist", socktype[type]); len = 0; if (sysctlbyname(mibvar, 0, &len, 0, 0) < 0) { if (errno != ENOENT) xo_warn("sysctl: %s", mibvar); return (-1); } if ((buf = malloc(len)) == NULL) { xo_warnx("malloc %lu bytes", (u_long)len); return (-2); } if (sysctlbyname(mibvar, buf, &len, 0, 0) < 0) { xo_warn("sysctl: %s", mibvar); free(buf); return (-2); } *bufp = buf; return (0); } static int pcblist_kvm(u_long count_off, u_long gencnt_off, u_long head_off, char **bufp) { struct unp_head head; struct unpcb *unp, unp_conn; u_char sun_len; struct socket so; struct xunpgen xug; struct xunpcb xu; unp_gen_t unp_gencnt; u_int unp_count; char *buf, *p; size_t len; if (count_off == 0 || gencnt_off == 0) return (-2); if (head_off == 0) return (-1); kread(count_off, &unp_count, sizeof(unp_count)); len = 2 * sizeof(xug) + (unp_count + unp_count / 8) * sizeof(xu); if ((buf = malloc(len)) == NULL) { xo_warnx("malloc %lu bytes", (u_long)len); return (-2); } p = buf; #define COPYOUT(obj, size) do { \ if (len < (size)) { \ xo_warnx("buffer size exceeded"); \ goto fail; \ } \ bcopy((obj), p, (size)); \ len -= (size); \ p += (size); \ } while (0) #define KREAD(off, buf, len) do { \ if (kread((uintptr_t)(off), (buf), (len)) != 0) \ goto fail; \ } while (0) /* Write out header. */ kread(gencnt_off, &unp_gencnt, sizeof(unp_gencnt)); xug.xug_len = sizeof xug; xug.xug_count = unp_count; xug.xug_gen = unp_gencnt; xug.xug_sogen = 0; COPYOUT(&xug, sizeof xug); /* Walk the PCB list. */ xu.xu_len = sizeof xu; KREAD(head_off, &head, sizeof(head)); LIST_FOREACH(unp, &head, unp_link) { xu.xu_unpp = unp; KREAD(unp, &xu.xu_unp, sizeof (*unp)); unp = &xu.xu_unp; if (unp->unp_gencnt > unp_gencnt) continue; if (unp->unp_addr != NULL) { KREAD(unp->unp_addr, &sun_len, sizeof(sun_len)); KREAD(unp->unp_addr, &xu.xu_addr, sun_len); } if (unp->unp_conn != NULL) { KREAD(unp->unp_conn, &unp_conn, sizeof(unp_conn)); if (unp_conn.unp_addr != NULL) { KREAD(unp_conn.unp_addr, &sun_len, sizeof(sun_len)); KREAD(unp_conn.unp_addr, &xu.xu_caddr, sun_len); } } KREAD(unp->unp_socket, &so, sizeof(so)); if (sotoxsocket(&so, &xu.xu_socket) != 0) goto fail; COPYOUT(&xu, sizeof(xu)); } /* Reread the counts and write the footer. */ kread(count_off, &unp_count, sizeof(unp_count)); kread(gencnt_off, &unp_gencnt, sizeof(unp_gencnt)); xug.xug_count = unp_count; xug.xug_gen = unp_gencnt; COPYOUT(&xug, sizeof xug); *bufp = buf; return (0); fail: free(buf); return (-1); #undef COPYOUT #undef KREAD } void unixpr(u_long count_off, u_long gencnt_off, u_long dhead_off, u_long shead_off, u_long sphead_off, bool *first) { char *buf; int ret, type; struct xsocket *so; struct xunpgen *xug, *oxug; struct xunpcb *xunp; u_long head_off; buf = NULL; for (type = SOCK_STREAM; type <= SOCK_SEQPACKET; type++) { if (live) ret = pcblist_sysctl(type, &buf); else { head_off = 0; switch (type) { case SOCK_STREAM: head_off = shead_off; break; case SOCK_DGRAM: head_off = dhead_off; break; case SOCK_SEQPACKET: head_off = sphead_off; break; } ret = pcblist_kvm(count_off, gencnt_off, head_off, &buf); } if (ret == -1) continue; if (ret < 0) return; oxug = xug = (struct xunpgen *)buf; for (xug = (struct xunpgen *)((char *)xug + xug->xug_len); xug->xug_len > sizeof(struct xunpgen); xug = (struct xunpgen *)((char *)xug + xug->xug_len)) { xunp = (struct xunpcb *)xug; so = &xunp->xu_socket; /* Ignore PCBs which were freed during copyout. */ if (xunp->xu_unp.unp_gencnt > oxug->xug_gen) continue; if (*first) { xo_open_list("socket"); *first = false; } xo_open_instance("socket"); unixdomainpr(xunp, so); xo_close_instance("socket"); } if (xug != oxug && xug->xug_gen != oxug->xug_gen) { if (oxug->xug_count > xug->xug_count) { xo_emit("Some {:type/%s} sockets may have " "been {:action/deleted}.\n", socktype[type]); } else if (oxug->xug_count < xug->xug_count) { xo_emit("Some {:type/%s} sockets may have " "been {:action/created}.\n", socktype[type]); } else { xo_emit("Some {:type/%s} sockets may have " "been {:action/created or deleted}", socktype[type]); } } free(buf); } } static void unixdomainpr(struct xunpcb *xunp, struct xsocket *so) { struct unpcb *unp; struct sockaddr_un *sa; static int first = 1; char buf1[33]; static const char *titles[2] = { "{T:/%-8.8s} {T:/%-6.6s} {T:/%-6.6s} {T:/%-6.6s} {T:/%8.8s} " "{T:/%8.8s} {T:/%8.8s} {T:/%8.8s} {T:Addr}\n", "{T:/%-16.16s} {T:/%-6.6s} {T:/%-6.6s} {T:/%-6.6s} {T:/%16.16s} " "{T:/%16.16s} {T:/%16.16s} {T:/%16.16s} {T:Addr}\n" }; static const char *format[2] = { "{q:address/%8lx} {t:type/%-6.6s} " "{:receive-bytes-waiting/%6u} " "{:send-bytes-waiting/%6u} " "{q:vnode/%8lx} {q:connection/%8lx} " "{q:first-reference/%8lx} {q:next-reference/%8lx}", "{q:address/%16lx} {t:type/%-6.6s} " "{:receive-bytes-waiting/%6u} " "{:send-bytes-waiting/%6u} " "{q:vnode/%16lx} {q:connection/%16lx} " "{q:first-reference/%16lx} {q:next-reference/%16lx}" }; int fmt = (sizeof(void *) == 8) ? 1 : 0; unp = &xunp->xu_unp; if (unp->unp_addr) sa = &xunp->xu_addr; else sa = (struct sockaddr_un *)0; if (first && !Lflag) { xo_emit("{T:Active UNIX domain sockets}\n"); xo_emit(titles[fmt], "Address", "Type", "Recv-Q", "Send-Q", "Inode", "Conn", "Refs", "Nextref"); first = 0; } if (Lflag && so->so_qlimit == 0) return; if (Lflag) { snprintf(buf1, sizeof buf1, "%u/%u/%u", so->so_qlen, so->so_incqlen, so->so_qlimit); xo_emit("unix {d:socket/%-32.32s}{e:queue-length/%u}" "{e:incomplete-queue-length/%u}{e:queue-limit/%u}", buf1, so->so_qlen, so->so_incqlen, so->so_qlimit); } else { xo_emit(format[fmt], (long)so->so_pcb, socktype[so->so_type], so->so_rcv.sb_cc, so->so_snd.sb_cc, (long)unp->unp_vnode, (long)unp->unp_conn, (long)LIST_FIRST(&unp->unp_refs), (long)LIST_NEXT(unp, unp_reflink)); } if (sa) xo_emit(" {:path/%.*s}", (int)(sa->sun_len - offsetof(struct sockaddr_un, sun_path)), sa->sun_path); xo_emit("\n"); } Index: projects/clang400-import/usr.bin/tail/reverse.c =================================================================== --- projects/clang400-import/usr.bin/tail/reverse.c (revision 311459) +++ projects/clang400-import/usr.bin/tail/reverse.c (revision 311460) @@ -1,284 +1,291 @@ /*- * Copyright (c) 1991, 1993 * The Regents of the University of California. All rights reserved. * * This code is derived from software contributed to Berkeley by * Edward Sze-Tyan Wang. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #if 0 #ifndef lint static char sccsid[] = "@(#)reverse.c 8.1 (Berkeley) 6/6/93"; #endif /* not lint */ #endif #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include "extern.h" static void r_buf(FILE *, const char *); static void r_reg(FILE *, const char *, enum STYLE, off_t, struct stat *); /* * reverse -- display input in reverse order by line. * * There are six separate cases for this -- regular and non-regular * files by bytes, lines or the whole file. * * BYTES display N bytes * REG mmap the file and display the lines * NOREG cyclically read characters into a wrap-around buffer * * LINES display N lines * REG mmap the file and display the lines * NOREG cyclically read lines into a wrap-around array of buffers * * FILE display the entire file * REG mmap the file and display the lines * NOREG cyclically read input into a linked list of buffers */ void reverse(FILE *fp, const char *fn, enum STYLE style, off_t off, struct stat *sbp) { if (style != REVERSE && off == 0) return; if (S_ISREG(sbp->st_mode)) r_reg(fp, fn, style, off, sbp); else switch(style) { case FBYTES: case RBYTES: bytes(fp, fn, off); break; case FLINES: case RLINES: lines(fp, fn, off); break; case REVERSE: r_buf(fp, fn); break; default: break; } } /* * r_reg -- display a regular file in reverse order by line. */ static void r_reg(FILE *fp, const char *fn, enum STYLE style, off_t off, struct stat *sbp) { struct mapinfo map; off_t curoff, size, lineend; int i; if (!(size = sbp->st_size)) return; map.start = NULL; map.mapoff = map.maxoff = size; map.fd = fileno(fp); + map.maplen = 0; /* * Last char is special, ignore whether newline or not. Note that * size == 0 is dealt with above, and size == 1 sets curoff to -1. */ curoff = size - 2; lineend = size; while (curoff >= 0) { if (curoff < map.mapoff || curoff >= map.mapoff + (off_t)map.maplen) { if (maparound(&map, curoff) != 0) { ierr(fn); return; } } for (i = curoff - map.mapoff; i >= 0; i--) { if (style == RBYTES && --off == 0) break; if (map.start[i] == '\n') break; } /* `i' is either the map offset of a '\n', or -1. */ curoff = map.mapoff + i; if (i < 0) continue; /* Print the line and update offsets. */ if (mapprint(&map, curoff + 1, lineend - curoff - 1) != 0) { ierr(fn); return; } lineend = curoff + 1; curoff--; if (style == RLINES) off--; if (off == 0 && style != REVERSE) { /* Avoid printing anything below. */ curoff = 0; break; } } if (curoff < 0 && mapprint(&map, 0, lineend) != 0) { ierr(fn); return; } if (map.start != NULL && munmap(map.start, map.maplen)) ierr(fn); } typedef struct bf { struct bf *next; struct bf *prev; int len; char *l; } BF; /* * r_buf -- display a non-regular file in reverse order by line. * * This is the function that saves the entire input, storing the data in a * doubly linked list of buffers and then displays them in reverse order. * It has the usual nastiness of trying to find the newlines, as there's no * guarantee that a newline occurs anywhere in the file, let alone in any * particular buffer. If we run out of memory, input is discarded (and the * user warned). */ static void r_buf(FILE *fp, const char *fn) { BF *mark, *tl, *tr; int ch, len, llen; char *p; off_t enomem; tl = NULL; #define BSZ (128 * 1024) for (mark = NULL, enomem = 0;;) { /* * Allocate a new block and link it into place in a doubly * linked list. If out of memory, toss the LRU block and * keep going. */ if (enomem || (tl = malloc(sizeof(BF))) == NULL || (tl->l = malloc(BSZ)) == NULL) { if (!mark) err(1, "malloc"); - tl = enomem ? tl->next : mark; + if (enomem) + tl = tl->next; + else { + if (tl) + free(tl); + tl = mark; + } enomem += tl->len; } else if (mark) { tl->next = mark; tl->prev = mark->prev; mark->prev->next = tl; mark->prev = tl; } else { mark = tl; mark->next = mark->prev = mark; } /* Fill the block with input data. */ for (p = tl->l, len = 0; len < BSZ && (ch = getc(fp)) != EOF; ++len) *p++ = ch; if (ferror(fp)) { ierr(fn); return; } /* * If no input data for this block and we tossed some data, * recover it. */ if (!len && enomem) { enomem -= tl->len; tl = tl->prev; break; } tl->len = len; if (ch == EOF) break; } if (enomem) { warnx("warning: %jd bytes discarded", (intmax_t)enomem); rval = 1; } /* * Step through the blocks in the reverse order read. The last char * is special, ignore whether newline or not. */ for (mark = tl;;) { for (p = tl->l + (len = tl->len) - 1, llen = 0; len--; --p, ++llen) if (*p == '\n') { if (llen) { WR(p + 1, llen); llen = 0; } if (tl == mark) continue; for (tr = tl->next; tr->len; tr = tr->next) { WR(tr->l, tr->len); tr->len = 0; if (tr == mark) break; } } tl->len = llen; if ((tl = tl->prev) == mark) break; } tl = tl->next; if (tl->len) { WR(tl->l, tl->len); tl->len = 0; } while ((tl = tl->next)->len) { WR(tl->l, tl->len); tl->len = 0; } } Index: projects/clang400-import/usr.sbin/bsnmpd/modules/snmp_hostres/hostres_scalars.c =================================================================== --- projects/clang400-import/usr.sbin/bsnmpd/modules/snmp_hostres/hostres_scalars.c (revision 311459) +++ projects/clang400-import/usr.sbin/bsnmpd/modules/snmp_hostres/hostres_scalars.c (revision 311460) @@ -1,492 +1,492 @@ /*- * Copyright (c) 2005-2006 The FreeBSD Project * All rights reserved. * * Author: Victor Cruceru * * Redistribution of this software and documentation 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 or documentation 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. * * $FreeBSD$ */ /* * Host Resources MIB scalars implementation for SNMPd. */ -#include +#include #include #include #include #include #include #include #include #include "hostres_snmp.h" #include "hostres_oid.h" #include "hostres_tree.h" /* boot timestamp in centi-seconds */ static uint64_t kernel_boot; /* physical memory size in Kb */ static uint64_t phys_mem_size; /* boot line (malloced) */ static u_char *boot_line; /* maximum number of processes */ static uint32_t max_proc; /** * Free all static data */ void fini_scalars(void) { free(boot_line); } /** * Get system uptime in hundredths of seconds since the epoch * Returns 0 in case of an error */ static int OS_getSystemUptime(uint32_t *ut) { struct timeval right_now; uint64_t now; if (kernel_boot == 0) { /* first time, do the sysctl */ struct timeval kernel_boot_timestamp; int mib[2] = { CTL_KERN, KERN_BOOTTIME }; size_t len = sizeof(kernel_boot_timestamp); - if (sysctl(mib, 2, &kernel_boot_timestamp, + if (sysctl(mib, nitems(mib), &kernel_boot_timestamp, &len, NULL, 0) == -1) { syslog(LOG_ERR, "sysctl KERN_BOOTTIME failed: %m"); return (SNMP_ERR_GENERR); } HRDBG("boot timestamp from kernel: {%lld, %ld}", (long long)kernel_boot_timestamp.tv_sec, (long)kernel_boot_timestamp.tv_usec); kernel_boot = ((uint64_t)kernel_boot_timestamp.tv_sec * 100) + (kernel_boot_timestamp.tv_usec / 10000); } if (gettimeofday(&right_now, NULL) < 0) { syslog(LOG_ERR, "gettimeofday failed: %m"); return (SNMP_ERR_GENERR); } now = ((uint64_t)right_now.tv_sec * 100) + (right_now.tv_usec / 10000); if (now - kernel_boot > UINT32_MAX) *ut = UINT32_MAX; else *ut = now - kernel_boot; return (SNMP_ERR_NOERROR); } /** * Get system local date and time in a foramt suitable for DateAndTime TC: * field octets contents range * ----- ------ -------- ----- * 1 1-2 year* 0..65536 * 2 3 month 1..12 * 3 4 day 1..31 * 4 5 hour 0..23 * 5 6 minutes 0..59 * 6 7 seconds 0..60 * (use 60 for leap-second) * 7 8 deci-seconds 0..9 * 8 9 direction from UTC '+' / '-' * 9 10 hours from UTC* 0..13 * 10 11 minutes from UTC 0..59 * * * Notes: * - the value of year is in network-byte order * - daylight saving time in New Zealand is +13 * * For example, Tuesday May 26, 1992 at 1:30:15 PM EDT would be * displayed as: * * 1992-5-26,13:30:15.0,-4:0 * * Returns -1 in case of an error or the length of the string (8 or 11) * Actually returns always 11 on freebsd */ static int OS_getSystemDate(struct snmp_value *value) { u_char s_date_time[11]; struct tm tloc_tm; time_t tloc_time_t; struct timeval right_now; int string_len; if (gettimeofday(&right_now, NULL) < 0) { syslog(LOG_ERR, "gettimeofday failed: %m"); return (SNMP_ERR_GENERR); } tloc_time_t = right_now.tv_sec; if (localtime_r(&tloc_time_t, &tloc_tm) == NULL) { syslog(LOG_ERR, "localtime_r() failed: %m "); return (SNMP_ERR_GENERR); } string_len = make_date_time(s_date_time, &tloc_tm, right_now.tv_usec / 100000); return (string_get(value, s_date_time, string_len)); } /** * Get kernel boot path. For FreeBSD it seems that no arguments are * present. Returns NULL if an error occurred. The returned data is a * pointer to a global storage. */ int OS_getSystemInitialLoadParameters(u_char **params) { if (boot_line == NULL) { int mib[2] = { CTL_KERN, KERN_BOOTFILE }; char *buf; size_t buf_len = 0; /* get the needed buffer len */ if (sysctl(mib, 2, NULL, &buf_len, NULL, 0) != 0) { syslog(LOG_ERR, "sysctl({CTL_KERN,KERN_BOOTFILE}) failed: %m"); return (SNMP_ERR_GENERR); } if ((buf = malloc(buf_len)) == NULL) { syslog(LOG_ERR, "malloc failed"); return (SNMP_ERR_GENERR); } if (sysctl(mib, 2, buf, &buf_len, NULL, 0)) { syslog(LOG_ERR, "sysctl({CTL_KERN,KERN_BOOTFILE}) failed: %m"); free(buf); return (SNMP_ERR_GENERR); } boot_line = buf; HRDBG("kernel boot file: %s", boot_line); } *params = boot_line; return (SNMP_ERR_NOERROR); } /** * Get number of current users which are logged in */ static int OS_getSystemNumUsers(uint32_t *nu) { struct utmpx *utmp; setutxent(); *nu = 0; while ((utmp = getutxent()) != NULL) { if (utmp->ut_type == USER_PROCESS) (*nu)++; } endutxent(); return (SNMP_ERR_NOERROR); } /** * Get number of current processes existing into the system */ static int OS_getSystemProcesses(uint32_t *proc_count) { int pc; if (hr_kd == NULL) return (SNMP_ERR_GENERR); if (kvm_getprocs(hr_kd, KERN_PROC_PROC, 0, &pc) == NULL) { syslog(LOG_ERR, "kvm_getprocs failed: %m"); return (SNMP_ERR_GENERR); } *proc_count = pc; return (SNMP_ERR_NOERROR); } /** * Get maximum number of processes allowed on this system */ static int OS_getSystemMaxProcesses(uint32_t *mproc) { if (max_proc == 0) { int mib[2] = { CTL_KERN, KERN_MAXPROC }; int mp; size_t len = sizeof(mp); if (sysctl(mib, 2, &mp, &len, NULL, 0) == -1) { syslog(LOG_ERR, "sysctl KERN_MAXPROC failed: %m"); return (SNMP_ERR_GENERR); } max_proc = mp; } *mproc = max_proc; return (SNMP_ERR_NOERROR); } /* * Get the physical memeory size in Kbytes. * Returns SNMP error code. */ static int OS_getMemorySize(uint32_t *ms) { if (phys_mem_size == 0) { int mib[2] = { CTL_HW, HW_PHYSMEM }; u_long physmem; size_t len = sizeof(physmem); if (sysctl(mib, 2, &physmem, &len, NULL, 0) == -1) { syslog(LOG_ERR, "sysctl({ CTL_HW, HW_PHYSMEM }) failed: %m"); return (SNMP_ERR_GENERR); } phys_mem_size = physmem / 1024; } if (phys_mem_size > UINT32_MAX) *ms = UINT32_MAX; else *ms = phys_mem_size; return (SNMP_ERR_NOERROR); } /* * Try to use the s_date_time parameter as a DateAndTime TC to fill in * the second parameter. * Returns 0 on succes and -1 for an error. * Bug: time zone info is not used */ static struct timeval * OS_checkSystemDateInput(const u_char *str, u_int len) { struct tm tm_to_set; time_t t; struct timeval *tv; if (len != 8 && len != 11) return (NULL); if (str[2] == 0 || str[2] > 12 || str[3] == 0 || str[3] > 31 || str[4] > 23 || str[5] > 59 || str[6] > 60 || str[7] > 9) return (NULL); tm_to_set.tm_year = ((str[0] << 8) + str[1]) - 1900; tm_to_set.tm_mon = str[2] - 1; tm_to_set.tm_mday = str[3]; tm_to_set.tm_hour = str[4]; tm_to_set.tm_min = str[5]; tm_to_set.tm_sec = str[6]; tm_to_set.tm_isdst = 0; /* now make UTC from it */ if ((t = timegm(&tm_to_set)) == (time_t)-1) return (NULL); /* now apply timezone if specified */ if (len == 11) { if (str[9] > 13 || str[10] > 59) return (NULL); if (str[8] == '+') t += 3600 * str[9] + 60 * str[10]; else t -= 3600 * str[9] + 60 * str[10]; } if ((tv = malloc(sizeof(*tv))) == NULL) return (NULL); tv->tv_sec = t; tv->tv_usec = (int32_t)str[7] * 100000; return (tv); } /* * Set system date and time. Timezone is not changed */ static int OS_setSystemDate(const struct timeval *timeval_to_set) { if (settimeofday(timeval_to_set, NULL) == -1) { syslog(LOG_ERR, "settimeofday failed: %m"); return (SNMP_ERR_GENERR); } return (SNMP_ERR_NOERROR); } /* * prototype of this function was genrated by gensnmptree tool in header file * hostres_tree.h * Returns SNMP_ERR_NOERROR on success */ int op_hrSystem(struct snmp_context *ctx, struct snmp_value *value, u_int sub, u_int iidx __unused, enum snmp_op curr_op) { int err; u_char *str; switch (curr_op) { case SNMP_OP_GET: switch (value->var.subs[sub - 1]) { case LEAF_hrSystemUptime: return (OS_getSystemUptime(&value->v.uint32)); case LEAF_hrSystemDate: return (OS_getSystemDate(value)); case LEAF_hrSystemInitialLoadDevice: value->v.uint32 = 0; /* FIXME */ return (SNMP_ERR_NOERROR); case LEAF_hrSystemInitialLoadParameters: if ((err = OS_getSystemInitialLoadParameters(&str)) != SNMP_ERR_NOERROR) return (err); return (string_get(value, str, -1)); case LEAF_hrSystemNumUsers: return (OS_getSystemNumUsers(&value->v.uint32)); case LEAF_hrSystemProcesses: return (OS_getSystemProcesses(&value->v.uint32)); case LEAF_hrSystemMaxProcesses: return (OS_getSystemMaxProcesses(&value->v.uint32)); } abort(); case SNMP_OP_SET: switch (value->var.subs[sub - 1]) { case LEAF_hrSystemDate: if ((ctx->scratch->ptr1 = OS_checkSystemDateInput(value->v.octetstring.octets, value->v.octetstring.len)) == NULL) return (SNMP_ERR_WRONG_VALUE); return (SNMP_ERR_NOERROR); case LEAF_hrSystemInitialLoadDevice: case LEAF_hrSystemInitialLoadParameters: return (SNMP_ERR_NOT_WRITEABLE); } abort(); case SNMP_OP_ROLLBACK: switch (value->var.subs[sub - 1]) { case LEAF_hrSystemDate: free(ctx->scratch->ptr1); return (SNMP_ERR_NOERROR); case LEAF_hrSystemInitialLoadDevice: case LEAF_hrSystemInitialLoadParameters: abort(); } abort(); case SNMP_OP_COMMIT: switch (value->var.subs[sub - 1]) { case LEAF_hrSystemDate: (void)OS_setSystemDate(ctx->scratch->ptr1); free(ctx->scratch->ptr1); return (SNMP_ERR_NOERROR); case LEAF_hrSystemInitialLoadDevice: case LEAF_hrSystemInitialLoadParameters: abort(); } abort(); case SNMP_OP_GETNEXT: abort(); } abort(); } /* * prototype of this function was genrated by gensnmptree tool * in the header file hostres_tree.h * Returns SNMP_ERR_NOERROR on success */ int op_hrStorage(struct snmp_context *ctx __unused, struct snmp_value *value, u_int sub, u_int iidx __unused, enum snmp_op curr_op) { /* only GET is possible */ switch (curr_op) { case SNMP_OP_GET: switch (value->var.subs[sub - 1]) { case LEAF_hrMemorySize: return (OS_getMemorySize(&value->v.uint32)); } abort(); case SNMP_OP_SET: return (SNMP_ERR_NOT_WRITEABLE); case SNMP_OP_ROLLBACK: case SNMP_OP_COMMIT: case SNMP_OP_GETNEXT: abort(); } abort(); } Index: projects/clang400-import/usr.sbin/inetd/inetd.c =================================================================== --- projects/clang400-import/usr.sbin/inetd/inetd.c (revision 311459) +++ projects/clang400-import/usr.sbin/inetd/inetd.c (revision 311460) @@ -1,2565 +1,2573 @@ /* * Copyright (c) 1983, 1991, 1993, 1994 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #ifndef lint static const char copyright[] = "@(#) Copyright (c) 1983, 1991, 1993, 1994\n\ The Regents of the University of California. All rights reserved.\n"; #endif /* not lint */ #ifndef lint #if 0 static char sccsid[] = "@(#)from: inetd.c 8.4 (Berkeley) 4/13/94"; #endif #endif /* not lint */ #include __FBSDID("$FreeBSD$"); /* * Inetd - Internet super-server * * This program invokes all internet services as needed. Connection-oriented * services are invoked each time a connection is made, by creating a process. * This process is passed the connection as file descriptor 0 and is expected * to do a getpeername to find out the source host and port. * * Datagram oriented services are invoked when a datagram * arrives; a process is created and passed a pending message * on file descriptor 0. Datagram servers may either connect * to their peer, freeing up the original socket for inetd * to receive further messages on, or ``take over the socket'', * processing all arriving datagrams and, eventually, timing * out. The first type of server is said to be ``multi-threaded''; * the second type of server ``single-threaded''. * * Inetd uses a configuration file which is read at startup * and, possibly, at some later time in response to a hangup signal. * The configuration file is ``free format'' with fields given in the * order shown below. Continuation lines for an entry must begin with * a space or tab. All fields must be present in each entry. * * service name must be in /etc/services * or name a tcpmux service * or specify a unix domain socket * socket type stream/dgram/raw/rdm/seqpacket * protocol tcp[4][6], udp[4][6], unix * wait/nowait single-threaded/multi-threaded * user[:group][/login-class] user/group/login-class to run daemon as * server program full path name * server program arguments maximum of MAXARGS (20) * * TCP services without official port numbers are handled with the * RFC1078-based tcpmux internal service. Tcpmux listens on port 1 for * requests. When a connection is made from a foreign host, the service * requested is passed to tcpmux, which looks it up in the servtab list * and returns the proper entry for the service. Tcpmux returns a * negative reply if the service doesn't exist, otherwise the invoked * server is expected to return the positive reply if the service type in * inetd.conf file has the prefix "tcpmux/". If the service type has the * prefix "tcpmux/+", tcpmux will return the positive reply for the * process; this is for compatibility with older server code, and also * allows you to invoke programs that use stdin/stdout without putting any * special server code in them. Services that use tcpmux are "nowait" * because they do not have a well-known port and hence cannot listen * for new requests. * * For RPC services * service name/version must be in /etc/rpc * socket type stream/dgram/raw/rdm/seqpacket * protocol rpc/tcp[4][6], rpc/udp[4][6] * wait/nowait single-threaded/multi-threaded * user[:group][/login-class] user/group/login-class to run daemon as * server program full path name * server program arguments maximum of MAXARGS * * Comment lines are indicated by a `#' in column 1. * * #ifdef IPSEC * Comment lines that start with "#@" denote IPsec policy string, as described * in ipsec_set_policy(3). This will affect all the following items in * inetd.conf(8). To reset the policy, just use "#@" line. By default, * there's no IPsec policy. * #endif */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "inetd.h" #include "pathnames.h" #ifdef IPSEC #include #ifndef IPSEC_POLICY_IPSEC /* no ipsec support on old ipsec */ #undef IPSEC #endif #endif #ifndef LIBWRAP_ALLOW_FACILITY # define LIBWRAP_ALLOW_FACILITY LOG_AUTH #endif #ifndef LIBWRAP_ALLOW_SEVERITY # define LIBWRAP_ALLOW_SEVERITY LOG_INFO #endif #ifndef LIBWRAP_DENY_FACILITY # define LIBWRAP_DENY_FACILITY LOG_AUTH #endif #ifndef LIBWRAP_DENY_SEVERITY # define LIBWRAP_DENY_SEVERITY LOG_WARNING #endif #define ISWRAP(sep) \ ( ((wrap_ex && !(sep)->se_bi) || (wrap_bi && (sep)->se_bi)) \ && (sep->se_family == AF_INET || sep->se_family == AF_INET6) \ && ( ((sep)->se_accept && (sep)->se_socktype == SOCK_STREAM) \ || (sep)->se_socktype == SOCK_DGRAM)) #ifdef LOGIN_CAP #include /* see init.c */ #define RESOURCE_RC "daemon" #endif #ifndef MAXCHILD #define MAXCHILD -1 /* maximum number of this service < 0 = no limit */ #endif #ifndef MAXCPM #define MAXCPM -1 /* rate limit invocations from a single remote address, < 0 = no limit */ #endif #ifndef MAXPERIP #define MAXPERIP -1 /* maximum number of this service from a single remote address, < 0 = no limit */ #endif #ifndef TOOMANY #define TOOMANY 256 /* don't start more than TOOMANY */ #endif #define CNT_INTVL 60 /* servers in CNT_INTVL sec. */ #define RETRYTIME (60*10) /* retry after bind or server fail */ #define MAX_MAXCHLD 32767 /* max allowable max children */ #define SIGBLOCK (sigmask(SIGCHLD)|sigmask(SIGHUP)|sigmask(SIGALRM)) #define satosin(sa) ((struct sockaddr_in *)(void *)sa) #define csatosin(sa) ((const struct sockaddr_in *)(const void *)sa) #ifdef INET6 #define satosin6(sa) ((struct sockaddr_in6 *)(void *)sa) #define csatosin6(sa) ((const struct sockaddr_in6 *)(const void *)sa) #endif static void close_sep(struct servtab *); static void flag_signal(int); static void flag_config(int); static void config(void); static int cpmip(const struct servtab *, int); static void endconfig(void); static struct servtab *enter(struct servtab *); static void freeconfig(struct servtab *); static struct servtab *getconfigent(void); static int matchservent(const char *, const char *, const char *); static char *nextline(FILE *); static void addchild(struct servtab *, int); static void flag_reapchild(int); static void reapchild(void); static void enable(struct servtab *); static void disable(struct servtab *); static void flag_retry(int); static void retry(void); static int setconfig(void); static void setup(struct servtab *); #ifdef IPSEC static void ipsecsetup(struct servtab *); #endif static void unregisterrpc(register struct servtab *sep); static struct conninfo *search_conn(struct servtab *sep, int ctrl); static int room_conn(struct servtab *sep, struct conninfo *conn); static void addchild_conn(struct conninfo *conn, pid_t pid); static void reapchild_conn(pid_t pid); static void free_conn(struct conninfo *conn); static void resize_conn(struct servtab *sep, int maxperip); static void free_connlist(struct servtab *sep); static void free_proc(struct procinfo *); static struct procinfo *search_proc(pid_t pid, int add); static int hashval(char *p, int len); static char *skip(char **); static char *sskip(char **); static char *newstr(const char *); static void print_service(const char *, const struct servtab *); /* tcpd.h */ int allow_severity; int deny_severity; static int wrap_ex = 0; static int wrap_bi = 0; int debug = 0; static int dolog = 0; static int maxsock; /* highest-numbered descriptor */ static fd_set allsock; static int options; static int timingout; static int toomany = TOOMANY; static int maxchild = MAXCHILD; static int maxcpm = MAXCPM; static int maxperip = MAXPERIP; static struct servent *sp; static struct rpcent *rpc; static char *hostname = NULL; static struct sockaddr_in *bind_sa4; static int v4bind_ok = 0; #ifdef INET6 static struct sockaddr_in6 *bind_sa6; static int v6bind_ok = 0; #endif static int signalpipe[2]; #ifdef SANITY_CHECK static int nsock; #endif static uid_t euid; static gid_t egid; static mode_t mask; struct servtab *servtab; static const char *CONFIG = _PATH_INETDCONF; static const char *pid_file = _PATH_INETDPID; static struct pidfh *pfh = NULL; static struct netconfig *udpconf, *tcpconf, *udp6conf, *tcp6conf; static LIST_HEAD(, procinfo) proctable[PERIPSIZE]; static int getvalue(const char *arg, int *value, const char *whine) { int tmp; char *p; tmp = strtol(arg, &p, 0); if (tmp < 0 || *p) { syslog(LOG_ERR, whine, arg); return 1; /* failure */ } *value = tmp; return 0; /* success */ } static sa_family_t whichaf(struct request_info *req) { struct sockaddr *sa; sa = (struct sockaddr *)req->client->sin; if (sa == NULL) return AF_UNSPEC; +#ifdef INET6 if (sa->sa_family == AF_INET6 && IN6_IS_ADDR_V4MAPPED(&satosin6(sa)->sin6_addr)) return AF_INET; +#endif return sa->sa_family; } int main(int argc, char **argv) { struct servtab *sep; struct passwd *pwd; struct group *grp; struct sigaction sa, saalrm, sachld, sahup, sapipe; int ch, dofork; pid_t pid; char buf[50]; #ifdef LOGIN_CAP login_cap_t *lc = NULL; #endif struct request_info req; int denied; char *service = NULL; struct sockaddr_storage peer; int i; struct addrinfo hints, *res; const char *servname; int error; struct conninfo *conn; openlog("inetd", LOG_PID | LOG_NOWAIT | LOG_PERROR, LOG_DAEMON); while ((ch = getopt(argc, argv, "dlwWR:a:c:C:p:s:")) != -1) switch(ch) { case 'd': debug = 1; options |= SO_DEBUG; break; case 'l': dolog = 1; break; case 'R': getvalue(optarg, &toomany, "-R %s: bad value for service invocation rate"); break; case 'c': getvalue(optarg, &maxchild, "-c %s: bad value for maximum children"); break; case 'C': getvalue(optarg, &maxcpm, "-C %s: bad value for maximum children/minute"); break; case 'a': hostname = optarg; break; case 'p': pid_file = optarg; break; case 's': getvalue(optarg, &maxperip, "-s %s: bad value for maximum children per source address"); break; case 'w': wrap_ex++; break; case 'W': wrap_bi++; break; case '?': default: syslog(LOG_ERR, "usage: inetd [-dlwW] [-a address] [-R rate]" " [-c maximum] [-C rate]" " [-p pidfile] [conf-file]"); exit(EX_USAGE); } /* * Initialize Bind Addrs. * When hostname is NULL, wild card bind addrs are obtained from * getaddrinfo(). But getaddrinfo() requires at least one of * hostname or servname is non NULL. * So when hostname is NULL, set dummy value to servname. * Since getaddrinfo() doesn't accept numeric servname, and * we doesn't use ai_socktype of struct addrinfo returned * from getaddrinfo(), we set dummy value to ai_socktype. */ servname = (hostname == NULL) ? "0" /* dummy */ : NULL; bzero(&hints, sizeof(struct addrinfo)); hints.ai_flags = AI_PASSIVE; hints.ai_family = AF_UNSPEC; hints.ai_socktype = SOCK_STREAM; /* dummy */ error = getaddrinfo(hostname, servname, &hints, &res); if (error != 0) { syslog(LOG_ERR, "-a %s: %s", hostname, gai_strerror(error)); if (error == EAI_SYSTEM) syslog(LOG_ERR, "%s", strerror(errno)); exit(EX_USAGE); } do { if (res->ai_addr == NULL) { syslog(LOG_ERR, "-a %s: getaddrinfo failed", hostname); exit(EX_USAGE); } switch (res->ai_addr->sa_family) { case AF_INET: if (v4bind_ok) continue; bind_sa4 = satosin(res->ai_addr); /* init port num in case servname is dummy */ bind_sa4->sin_port = 0; v4bind_ok = 1; continue; #ifdef INET6 case AF_INET6: if (v6bind_ok) continue; bind_sa6 = satosin6(res->ai_addr); /* init port num in case servname is dummy */ bind_sa6->sin6_port = 0; v6bind_ok = 1; continue; #endif } if (v4bind_ok #ifdef INET6 && v6bind_ok #endif ) break; } while ((res = res->ai_next) != NULL); if (!v4bind_ok #ifdef INET6 && !v6bind_ok #endif ) { syslog(LOG_ERR, "-a %s: unknown address family", hostname); exit(EX_USAGE); } euid = geteuid(); egid = getegid(); umask(mask = umask(0777)); argc -= optind; argv += optind; if (argc > 0) CONFIG = argv[0]; if (access(CONFIG, R_OK) < 0) syslog(LOG_ERR, "Accessing %s: %m, continuing anyway.", CONFIG); if (debug == 0) { pid_t otherpid; pfh = pidfile_open(pid_file, 0600, &otherpid); if (pfh == NULL) { if (errno == EEXIST) { syslog(LOG_ERR, "%s already running, pid: %d", getprogname(), otherpid); exit(EX_OSERR); } syslog(LOG_WARNING, "pidfile_open() failed: %m"); } if (daemon(0, 0) < 0) { syslog(LOG_WARNING, "daemon(0,0) failed: %m"); } /* From now on we don't want syslog messages going to stderr. */ closelog(); openlog("inetd", LOG_PID | LOG_NOWAIT, LOG_DAEMON); /* * In case somebody has started inetd manually, we need to * clear the logname, so that old servers run as root do not * get the user's logname.. */ if (setlogin("") < 0) { syslog(LOG_WARNING, "cannot clear logname: %m"); /* no big deal if it fails.. */ } if (pfh != NULL && pidfile_write(pfh) == -1) { syslog(LOG_WARNING, "pidfile_write(): %m"); } } if (madvise(NULL, 0, MADV_PROTECT) != 0) syslog(LOG_WARNING, "madvise() failed: %s", strerror(errno)); for (i = 0; i < PERIPSIZE; ++i) LIST_INIT(&proctable[i]); if (v4bind_ok) { udpconf = getnetconfigent("udp"); tcpconf = getnetconfigent("tcp"); if (udpconf == NULL || tcpconf == NULL) { syslog(LOG_ERR, "unknown rpc/udp or rpc/tcp"); exit(EX_USAGE); } } #ifdef INET6 if (v6bind_ok) { udp6conf = getnetconfigent("udp6"); tcp6conf = getnetconfigent("tcp6"); if (udp6conf == NULL || tcp6conf == NULL) { syslog(LOG_ERR, "unknown rpc/udp6 or rpc/tcp6"); exit(EX_USAGE); } } #endif sa.sa_flags = 0; sigemptyset(&sa.sa_mask); sigaddset(&sa.sa_mask, SIGALRM); sigaddset(&sa.sa_mask, SIGCHLD); sigaddset(&sa.sa_mask, SIGHUP); sa.sa_handler = flag_retry; sigaction(SIGALRM, &sa, &saalrm); config(); sa.sa_handler = flag_config; sigaction(SIGHUP, &sa, &sahup); sa.sa_handler = flag_reapchild; sigaction(SIGCHLD, &sa, &sachld); sa.sa_handler = SIG_IGN; sigaction(SIGPIPE, &sa, &sapipe); { /* space for daemons to overwrite environment for ps */ #define DUMMYSIZE 100 char dummy[DUMMYSIZE]; (void)memset(dummy, 'x', DUMMYSIZE - 1); dummy[DUMMYSIZE - 1] = '\0'; (void)setenv("inetd_dummy", dummy, 1); } if (pipe2(signalpipe, O_CLOEXEC) != 0) { syslog(LOG_ERR, "pipe: %m"); exit(EX_OSERR); } FD_SET(signalpipe[0], &allsock); #ifdef SANITY_CHECK nsock++; #endif if (signalpipe[0] > maxsock) maxsock = signalpipe[0]; if (signalpipe[1] > maxsock) maxsock = signalpipe[1]; for (;;) { int n, ctrl; fd_set readable; #ifdef SANITY_CHECK if (nsock == 0) { syslog(LOG_ERR, "%s: nsock=0", __func__); exit(EX_SOFTWARE); } #endif readable = allsock; if ((n = select(maxsock + 1, &readable, (fd_set *)0, (fd_set *)0, (struct timeval *)0)) <= 0) { if (n < 0 && errno != EINTR) { syslog(LOG_WARNING, "select: %m"); sleep(1); } continue; } /* handle any queued signal flags */ if (FD_ISSET(signalpipe[0], &readable)) { int nsig; if (ioctl(signalpipe[0], FIONREAD, &nsig) != 0) { syslog(LOG_ERR, "ioctl: %m"); exit(EX_OSERR); } while (--nsig >= 0) { char c; if (read(signalpipe[0], &c, 1) != 1) { syslog(LOG_ERR, "read: %m"); exit(EX_OSERR); } if (debug) warnx("handling signal flag %c", c); switch(c) { case 'A': /* sigalrm */ retry(); break; case 'C': /* sigchld */ reapchild(); break; case 'H': /* sighup */ config(); break; } } } for (sep = servtab; n && sep; sep = sep->se_next) if (sep->se_fd != -1 && FD_ISSET(sep->se_fd, &readable)) { n--; if (debug) warnx("someone wants %s", sep->se_service); dofork = !sep->se_bi || sep->se_bi->bi_fork || ISWRAP(sep); conn = NULL; if (sep->se_accept && sep->se_socktype == SOCK_STREAM) { i = 1; if (ioctl(sep->se_fd, FIONBIO, &i) < 0) syslog(LOG_ERR, "ioctl (FIONBIO, 1): %m"); ctrl = accept(sep->se_fd, (struct sockaddr *)0, (socklen_t *)0); if (debug) warnx("accept, ctrl %d", ctrl); if (ctrl < 0) { if (errno != EINTR) syslog(LOG_WARNING, "accept (for %s): %m", sep->se_service); if (sep->se_accept && sep->se_socktype == SOCK_STREAM) close(ctrl); continue; } i = 0; if (ioctl(sep->se_fd, FIONBIO, &i) < 0) syslog(LOG_ERR, "ioctl1(FIONBIO, 0): %m"); if (ioctl(ctrl, FIONBIO, &i) < 0) syslog(LOG_ERR, "ioctl2(FIONBIO, 0): %m"); if (cpmip(sep, ctrl) < 0) { close(ctrl); continue; } if (dofork && (conn = search_conn(sep, ctrl)) != NULL && !room_conn(sep, conn)) { close(ctrl); continue; } } else ctrl = sep->se_fd; if (dolog && !ISWRAP(sep)) { char pname[NI_MAXHOST] = "unknown"; socklen_t sl; sl = sizeof(peer); if (getpeername(ctrl, (struct sockaddr *) &peer, &sl)) { sl = sizeof(peer); if (recvfrom(ctrl, buf, sizeof(buf), MSG_PEEK, (struct sockaddr *)&peer, &sl) >= 0) { getnameinfo((struct sockaddr *)&peer, peer.ss_len, pname, sizeof(pname), NULL, 0, NI_NUMERICHOST); } } else { getnameinfo((struct sockaddr *)&peer, peer.ss_len, pname, sizeof(pname), NULL, 0, NI_NUMERICHOST); } syslog(LOG_INFO,"%s from %s", sep->se_service, pname); } (void) sigblock(SIGBLOCK); pid = 0; /* * Fork for all external services, builtins which need to * fork and anything we're wrapping (as wrapping might * block or use hosts_options(5) twist). */ if (dofork) { if (sep->se_count++ == 0) (void)clock_gettime(CLOCK_MONOTONIC_FAST, &sep->se_time); else if (toomany > 0 && sep->se_count >= toomany) { struct timespec now; (void)clock_gettime(CLOCK_MONOTONIC_FAST, &now); if (now.tv_sec - sep->se_time.tv_sec > CNT_INTVL) { sep->se_time = now; sep->se_count = 1; } else { syslog(LOG_ERR, "%s/%s server failing (looping), service terminated", sep->se_service, sep->se_proto); if (sep->se_accept && sep->se_socktype == SOCK_STREAM) close(ctrl); close_sep(sep); free_conn(conn); sigsetmask(0L); if (!timingout) { timingout = 1; alarm(RETRYTIME); } continue; } } pid = fork(); } if (pid < 0) { syslog(LOG_ERR, "fork: %m"); if (sep->se_accept && sep->se_socktype == SOCK_STREAM) close(ctrl); free_conn(conn); sigsetmask(0L); sleep(1); continue; } if (pid) { addchild_conn(conn, pid); addchild(sep, pid); } sigsetmask(0L); if (pid == 0) { pidfile_close(pfh); if (dofork) { sigaction(SIGALRM, &saalrm, (struct sigaction *)0); sigaction(SIGCHLD, &sachld, (struct sigaction *)0); sigaction(SIGHUP, &sahup, (struct sigaction *)0); /* SIGPIPE reset before exec */ } /* * Call tcpmux to find the real service to exec. */ if (sep->se_bi && sep->se_bi->bi_fn == (bi_fn_t *) tcpmux) { sep = tcpmux(ctrl); if (sep == NULL) { close(ctrl); _exit(0); } } if (ISWRAP(sep)) { inetd_setproctitle("wrapping", ctrl); service = sep->se_server_name ? sep->se_server_name : sep->se_service; request_init(&req, RQ_DAEMON, service, RQ_FILE, ctrl, 0); fromhost(&req); deny_severity = LIBWRAP_DENY_FACILITY|LIBWRAP_DENY_SEVERITY; allow_severity = LIBWRAP_ALLOW_FACILITY|LIBWRAP_ALLOW_SEVERITY; denied = !hosts_access(&req); if (denied) { syslog(deny_severity, "refused connection from %.500s, service %s (%s%s)", eval_client(&req), service, sep->se_proto, (whichaf(&req) == AF_INET6) ? "6" : ""); if (sep->se_socktype != SOCK_STREAM) recv(ctrl, buf, sizeof (buf), 0); if (dofork) { sleep(1); _exit(0); } } if (dolog) { syslog(allow_severity, "connection from %.500s, service %s (%s%s)", eval_client(&req), service, sep->se_proto, (whichaf(&req) == AF_INET6) ? "6" : ""); } } if (sep->se_bi) { (*sep->se_bi->bi_fn)(ctrl, sep); } else { if (debug) warnx("%d execl %s", getpid(), sep->se_server); /* Clear close-on-exec. */ if (fcntl(ctrl, F_SETFD, 0) < 0) { syslog(LOG_ERR, "%s/%s: fcntl (F_SETFD, 0): %m", sep->se_service, sep->se_proto); _exit(EX_OSERR); } if (ctrl != 0) { dup2(ctrl, 0); close(ctrl); } dup2(0, 1); dup2(0, 2); if ((pwd = getpwnam(sep->se_user)) == NULL) { syslog(LOG_ERR, "%s/%s: %s: no such user", sep->se_service, sep->se_proto, sep->se_user); if (sep->se_socktype != SOCK_STREAM) recv(0, buf, sizeof (buf), 0); _exit(EX_NOUSER); } grp = NULL; if ( sep->se_group != NULL && (grp = getgrnam(sep->se_group)) == NULL ) { syslog(LOG_ERR, "%s/%s: %s: no such group", sep->se_service, sep->se_proto, sep->se_group); if (sep->se_socktype != SOCK_STREAM) recv(0, buf, sizeof (buf), 0); _exit(EX_NOUSER); } if (grp != NULL) pwd->pw_gid = grp->gr_gid; #ifdef LOGIN_CAP if ((lc = login_getclass(sep->se_class)) == NULL) { /* error syslogged by getclass */ syslog(LOG_ERR, "%s/%s: %s: login class error", sep->se_service, sep->se_proto, sep->se_class); if (sep->se_socktype != SOCK_STREAM) recv(0, buf, sizeof (buf), 0); _exit(EX_NOUSER); } #endif if (setsid() < 0) { syslog(LOG_ERR, "%s: can't setsid(): %m", sep->se_service); /* _exit(EX_OSERR); not fatal yet */ } #ifdef LOGIN_CAP if (setusercontext(lc, pwd, pwd->pw_uid, LOGIN_SETALL & ~LOGIN_SETMAC) != 0) { syslog(LOG_ERR, "%s: can't setusercontext(..%s..): %m", sep->se_service, sep->se_user); _exit(EX_OSERR); } login_close(lc); #else if (pwd->pw_uid) { if (setlogin(sep->se_user) < 0) { syslog(LOG_ERR, "%s: can't setlogin(%s): %m", sep->se_service, sep->se_user); /* _exit(EX_OSERR); not yet */ } if (setgid(pwd->pw_gid) < 0) { syslog(LOG_ERR, "%s: can't set gid %d: %m", sep->se_service, pwd->pw_gid); _exit(EX_OSERR); } (void) initgroups(pwd->pw_name, pwd->pw_gid); if (setuid(pwd->pw_uid) < 0) { syslog(LOG_ERR, "%s: can't set uid %d: %m", sep->se_service, pwd->pw_uid); _exit(EX_OSERR); } } #endif sigaction(SIGPIPE, &sapipe, (struct sigaction *)0); execv(sep->se_server, sep->se_argv); syslog(LOG_ERR, "cannot execute %s: %m", sep->se_server); if (sep->se_socktype != SOCK_STREAM) recv(0, buf, sizeof (buf), 0); } if (dofork) _exit(0); } if (sep->se_accept && sep->se_socktype == SOCK_STREAM) close(ctrl); } } } /* * Add a signal flag to the signal flag queue for later handling */ static void flag_signal(int c) { char ch = c; if (write(signalpipe[1], &ch, 1) != 1) { syslog(LOG_ERR, "write: %m"); _exit(EX_OSERR); } } /* * Record a new child pid for this service. If we've reached the * limit on children, then stop accepting incoming requests. */ static void addchild(struct servtab *sep, pid_t pid) { if (sep->se_maxchild <= 0) return; #ifdef SANITY_CHECK if (sep->se_numchild >= sep->se_maxchild) { syslog(LOG_ERR, "%s: %d >= %d", __func__, sep->se_numchild, sep->se_maxchild); exit(EX_SOFTWARE); } #endif sep->se_pids[sep->se_numchild++] = pid; if (sep->se_numchild == sep->se_maxchild) disable(sep); } /* * Some child process has exited. See if it's on somebody's list. */ static void flag_reapchild(int signo __unused) { flag_signal('C'); } static void reapchild(void) { int k, status; pid_t pid; struct servtab *sep; for (;;) { pid = wait3(&status, WNOHANG, (struct rusage *)0); if (pid <= 0) break; if (debug) warnx("%d reaped, %s %u", pid, WIFEXITED(status) ? "status" : "signal", WIFEXITED(status) ? WEXITSTATUS(status) : WTERMSIG(status)); for (sep = servtab; sep; sep = sep->se_next) { for (k = 0; k < sep->se_numchild; k++) if (sep->se_pids[k] == pid) break; if (k == sep->se_numchild) continue; if (sep->se_numchild == sep->se_maxchild) enable(sep); sep->se_pids[k] = sep->se_pids[--sep->se_numchild]; if (WIFSIGNALED(status) || WEXITSTATUS(status)) syslog(LOG_WARNING, "%s[%d]: exited, %s %u", sep->se_server, pid, WIFEXITED(status) ? "status" : "signal", WIFEXITED(status) ? WEXITSTATUS(status) : WTERMSIG(status)); break; } reapchild_conn(pid); } } static void flag_config(int signo __unused) { flag_signal('H'); } static void config(void) { struct servtab *sep, *new, **sepp; long omask; int new_nomapped; #ifdef LOGIN_CAP login_cap_t *lc = NULL; #endif if (!setconfig()) { syslog(LOG_ERR, "%s: %m", CONFIG); return; } for (sep = servtab; sep; sep = sep->se_next) sep->se_checked = 0; while ((new = getconfigent())) { if (getpwnam(new->se_user) == NULL) { syslog(LOG_ERR, "%s/%s: no such user '%s', service ignored", new->se_service, new->se_proto, new->se_user); continue; } if (new->se_group && getgrnam(new->se_group) == NULL) { syslog(LOG_ERR, "%s/%s: no such group '%s', service ignored", new->se_service, new->se_proto, new->se_group); continue; } #ifdef LOGIN_CAP if ((lc = login_getclass(new->se_class)) == NULL) { /* error syslogged by getclass */ syslog(LOG_ERR, "%s/%s: %s: login class error, service ignored", new->se_service, new->se_proto, new->se_class); continue; } login_close(lc); #endif new_nomapped = new->se_nomapped; for (sep = servtab; sep; sep = sep->se_next) if (strcmp(sep->se_service, new->se_service) == 0 && strcmp(sep->se_proto, new->se_proto) == 0 && sep->se_rpc == new->se_rpc && sep->se_socktype == new->se_socktype && sep->se_family == new->se_family) break; if (sep != 0) { int i; #define SWAP(t,a, b) { t c = a; a = b; b = c; } omask = sigblock(SIGBLOCK); if (sep->se_nomapped != new->se_nomapped) { /* for rpc keep old nommaped till unregister */ if (!sep->se_rpc) sep->se_nomapped = new->se_nomapped; sep->se_reset = 1; } /* copy over outstanding child pids */ if (sep->se_maxchild > 0 && new->se_maxchild > 0) { new->se_numchild = sep->se_numchild; if (new->se_numchild > new->se_maxchild) new->se_numchild = new->se_maxchild; memcpy(new->se_pids, sep->se_pids, new->se_numchild * sizeof(*new->se_pids)); } SWAP(pid_t *, sep->se_pids, new->se_pids); sep->se_maxchild = new->se_maxchild; sep->se_numchild = new->se_numchild; sep->se_maxcpm = new->se_maxcpm; resize_conn(sep, new->se_maxperip); sep->se_maxperip = new->se_maxperip; sep->se_bi = new->se_bi; /* might need to turn on or off service now */ if (sep->se_fd >= 0) { if (sep->se_maxchild > 0 && sep->se_numchild == sep->se_maxchild) { if (FD_ISSET(sep->se_fd, &allsock)) disable(sep); } else { if (!FD_ISSET(sep->se_fd, &allsock)) enable(sep); } } sep->se_accept = new->se_accept; SWAP(char *, sep->se_user, new->se_user); SWAP(char *, sep->se_group, new->se_group); #ifdef LOGIN_CAP SWAP(char *, sep->se_class, new->se_class); #endif SWAP(char *, sep->se_server, new->se_server); SWAP(char *, sep->se_server_name, new->se_server_name); for (i = 0; i < MAXARGV; i++) SWAP(char *, sep->se_argv[i], new->se_argv[i]); #ifdef IPSEC SWAP(char *, sep->se_policy, new->se_policy); ipsecsetup(sep); #endif sigsetmask(omask); freeconfig(new); if (debug) print_service("REDO", sep); } else { sep = enter(new); if (debug) print_service("ADD ", sep); } sep->se_checked = 1; if (ISMUX(sep)) { sep->se_fd = -1; continue; } switch (sep->se_family) { case AF_INET: if (!v4bind_ok) { sep->se_fd = -1; continue; } break; #ifdef INET6 case AF_INET6: if (!v6bind_ok) { sep->se_fd = -1; continue; } break; #endif } if (!sep->se_rpc) { if (sep->se_family != AF_UNIX) { sp = getservbyname(sep->se_service, sep->se_proto); if (sp == 0) { syslog(LOG_ERR, "%s/%s: unknown service", sep->se_service, sep->se_proto); sep->se_checked = 0; continue; } } switch (sep->se_family) { case AF_INET: if (sp->s_port != sep->se_ctrladdr4.sin_port) { sep->se_ctrladdr4.sin_port = sp->s_port; sep->se_reset = 1; } break; #ifdef INET6 case AF_INET6: if (sp->s_port != sep->se_ctrladdr6.sin6_port) { sep->se_ctrladdr6.sin6_port = sp->s_port; sep->se_reset = 1; } break; #endif } if (sep->se_reset != 0 && sep->se_fd >= 0) close_sep(sep); } else { rpc = getrpcbyname(sep->se_service); if (rpc == 0) { syslog(LOG_ERR, "%s/%s unknown RPC service", sep->se_service, sep->se_proto); if (sep->se_fd != -1) (void) close(sep->se_fd); sep->se_fd = -1; continue; } if (sep->se_reset != 0 || rpc->r_number != sep->se_rpc_prog) { if (sep->se_rpc_prog) unregisterrpc(sep); sep->se_rpc_prog = rpc->r_number; if (sep->se_fd != -1) (void) close(sep->se_fd); sep->se_fd = -1; } sep->se_nomapped = new_nomapped; } sep->se_reset = 0; if (sep->se_fd == -1) setup(sep); } endconfig(); /* * Purge anything not looked at above. */ omask = sigblock(SIGBLOCK); sepp = &servtab; while ((sep = *sepp)) { if (sep->se_checked) { sepp = &sep->se_next; continue; } *sepp = sep->se_next; if (sep->se_fd >= 0) close_sep(sep); if (debug) print_service("FREE", sep); if (sep->se_rpc && sep->se_rpc_prog > 0) unregisterrpc(sep); freeconfig(sep); free(sep); } (void) sigsetmask(omask); } static void unregisterrpc(struct servtab *sep) { u_int i; struct servtab *sepp; long omask; struct netconfig *netid4, *netid6; omask = sigblock(SIGBLOCK); netid4 = sep->se_socktype == SOCK_DGRAM ? udpconf : tcpconf; netid6 = sep->se_socktype == SOCK_DGRAM ? udp6conf : tcp6conf; if (sep->se_family == AF_INET) netid6 = NULL; else if (sep->se_nomapped) netid4 = NULL; /* * Conflict if same prog and protocol - In that case one should look * to versions, but it is not interesting: having separate servers for * different versions does not work well. * Therefore one do not unregister if there is a conflict. * There is also transport conflict if destroying INET when INET46 * exists, or destroying INET46 when INET exists */ for (sepp = servtab; sepp; sepp = sepp->se_next) { if (sepp == sep) continue; if (sepp->se_checked == 0 || !sepp->se_rpc || strcmp(sep->se_proto, sepp->se_proto) != 0 || sep->se_rpc_prog != sepp->se_rpc_prog) continue; if (sepp->se_family == AF_INET) netid4 = NULL; if (sepp->se_family == AF_INET6) { netid6 = NULL; if (!sep->se_nomapped) netid4 = NULL; } if (netid4 == NULL && netid6 == NULL) return; } if (debug) print_service("UNREG", sep); for (i = sep->se_rpc_lowvers; i <= sep->se_rpc_highvers; i++) { if (netid4) rpcb_unset(sep->se_rpc_prog, i, netid4); if (netid6) rpcb_unset(sep->se_rpc_prog, i, netid6); } if (sep->se_fd != -1) (void) close(sep->se_fd); sep->se_fd = -1; (void) sigsetmask(omask); } static void flag_retry(int signo __unused) { flag_signal('A'); } static void retry(void) { struct servtab *sep; timingout = 0; for (sep = servtab; sep; sep = sep->se_next) if (sep->se_fd == -1 && !ISMUX(sep)) setup(sep); } static void setup(struct servtab *sep) { int on = 1; /* Set all listening sockets to close-on-exec. */ if ((sep->se_fd = socket(sep->se_family, sep->se_socktype | SOCK_CLOEXEC, 0)) < 0) { if (debug) warn("socket failed on %s/%s", sep->se_service, sep->se_proto); syslog(LOG_ERR, "%s/%s: socket: %m", sep->se_service, sep->se_proto); return; } #define turnon(fd, opt) \ setsockopt(fd, SOL_SOCKET, opt, (char *)&on, sizeof (on)) if (strcmp(sep->se_proto, "tcp") == 0 && (options & SO_DEBUG) && turnon(sep->se_fd, SO_DEBUG) < 0) syslog(LOG_ERR, "setsockopt (SO_DEBUG): %m"); if (turnon(sep->se_fd, SO_REUSEADDR) < 0) syslog(LOG_ERR, "setsockopt (SO_REUSEADDR): %m"); #ifdef SO_PRIVSTATE if (turnon(sep->se_fd, SO_PRIVSTATE) < 0) syslog(LOG_ERR, "setsockopt (SO_PRIVSTATE): %m"); #endif /* tftpd opens a new connection then needs more infos */ +#ifdef INET6 if ((sep->se_family == AF_INET6) && (strcmp(sep->se_proto, "udp") == 0) && (sep->se_accept == 0) && (setsockopt(sep->se_fd, IPPROTO_IPV6, IPV6_RECVPKTINFO, (char *)&on, sizeof (on)) < 0)) syslog(LOG_ERR, "setsockopt (IPV6_RECVPKTINFO): %m"); if (sep->se_family == AF_INET6) { int flag = sep->se_nomapped ? 1 : 0; if (setsockopt(sep->se_fd, IPPROTO_IPV6, IPV6_V6ONLY, (char *)&flag, sizeof (flag)) < 0) syslog(LOG_ERR, "setsockopt (IPV6_V6ONLY): %m"); } +#endif #undef turnon #ifdef IPSEC ipsecsetup(sep); #endif if (sep->se_family == AF_UNIX) { (void) unlink(sep->se_ctrladdr_un.sun_path); umask(0777); /* Make socket with conservative permissions */ } if (bind(sep->se_fd, (struct sockaddr *)&sep->se_ctrladdr, sep->se_ctrladdr_size) < 0) { if (debug) warn("bind failed on %s/%s", sep->se_service, sep->se_proto); syslog(LOG_ERR, "%s/%s: bind: %m", sep->se_service, sep->se_proto); (void) close(sep->se_fd); sep->se_fd = -1; if (!timingout) { timingout = 1; alarm(RETRYTIME); } if (sep->se_family == AF_UNIX) umask(mask); return; } if (sep->se_family == AF_UNIX) { /* Ick - fch{own,mod} don't work on Unix domain sockets */ if (chown(sep->se_service, sep->se_sockuid, sep->se_sockgid) < 0) syslog(LOG_ERR, "chown socket: %m"); if (chmod(sep->se_service, sep->se_sockmode) < 0) syslog(LOG_ERR, "chmod socket: %m"); umask(mask); } if (sep->se_rpc) { u_int i; socklen_t len = sep->se_ctrladdr_size; struct netconfig *netid, *netid2 = NULL; +#ifdef INET6 struct sockaddr_in sock; +#endif struct netbuf nbuf, nbuf2; if (getsockname(sep->se_fd, (struct sockaddr*)&sep->se_ctrladdr, &len) < 0){ syslog(LOG_ERR, "%s/%s: getsockname: %m", sep->se_service, sep->se_proto); (void) close(sep->se_fd); sep->se_fd = -1; return; } nbuf.buf = &sep->se_ctrladdr; nbuf.len = sep->se_ctrladdr.sa_len; if (sep->se_family == AF_INET) netid = sep->se_socktype==SOCK_DGRAM? udpconf:tcpconf; +#ifdef INET6 else { netid = sep->se_socktype==SOCK_DGRAM? udp6conf:tcp6conf; if (!sep->se_nomapped) { /* INET and INET6 */ netid2 = netid==udp6conf? udpconf:tcpconf; memset(&sock, 0, sizeof sock); /* ADDR_ANY */ nbuf2.buf = &sock; nbuf2.len = sock.sin_len = sizeof sock; sock.sin_family = AF_INET; sock.sin_port = sep->se_ctrladdr6.sin6_port; } } +#endif if (debug) print_service("REG ", sep); for (i = sep->se_rpc_lowvers; i <= sep->se_rpc_highvers; i++) { rpcb_unset(sep->se_rpc_prog, i, netid); rpcb_set(sep->se_rpc_prog, i, netid, &nbuf); if (netid2) { rpcb_unset(sep->se_rpc_prog, i, netid2); rpcb_set(sep->se_rpc_prog, i, netid2, &nbuf2); } } } if (sep->se_socktype == SOCK_STREAM) listen(sep->se_fd, -1); enable(sep); if (debug) { warnx("registered %s on %d", sep->se_server, sep->se_fd); } } #ifdef IPSEC static void ipsecsetup(struct servtab *sep) { char *buf; char *policy_in = NULL; char *policy_out = NULL; int level; int opt; switch (sep->se_family) { case AF_INET: level = IPPROTO_IP; opt = IP_IPSEC_POLICY; break; #ifdef INET6 case AF_INET6: level = IPPROTO_IPV6; opt = IPV6_IPSEC_POLICY; break; #endif default: return; } if (!sep->se_policy || sep->se_policy[0] == '\0') { static char def_in[] = "in entrust", def_out[] = "out entrust"; policy_in = def_in; policy_out = def_out; } else { if (!strncmp("in", sep->se_policy, 2)) policy_in = sep->se_policy; else if (!strncmp("out", sep->se_policy, 3)) policy_out = sep->se_policy; else { syslog(LOG_ERR, "invalid security policy \"%s\"", sep->se_policy); return; } } if (policy_in != NULL) { buf = ipsec_set_policy(policy_in, strlen(policy_in)); if (buf != NULL) { if (setsockopt(sep->se_fd, level, opt, buf, ipsec_get_policylen(buf)) < 0 && debug != 0) warnx("%s/%s: ipsec initialization failed; %s", sep->se_service, sep->se_proto, policy_in); free(buf); } else syslog(LOG_ERR, "invalid security policy \"%s\"", policy_in); } if (policy_out != NULL) { buf = ipsec_set_policy(policy_out, strlen(policy_out)); if (buf != NULL) { if (setsockopt(sep->se_fd, level, opt, buf, ipsec_get_policylen(buf)) < 0 && debug != 0) warnx("%s/%s: ipsec initialization failed; %s", sep->se_service, sep->se_proto, policy_out); free(buf); } else syslog(LOG_ERR, "invalid security policy \"%s\"", policy_out); } } #endif /* * Finish with a service and its socket. */ static void close_sep(struct servtab *sep) { if (sep->se_fd >= 0) { if (FD_ISSET(sep->se_fd, &allsock)) disable(sep); (void) close(sep->se_fd); sep->se_fd = -1; } sep->se_count = 0; sep->se_numchild = 0; /* forget about any existing children */ } static int matchservent(const char *name1, const char *name2, const char *proto) { char **alias, *p; struct servent *se; if (strcmp(proto, "unix") == 0) { if ((p = strrchr(name1, '/')) != NULL) name1 = p + 1; if ((p = strrchr(name2, '/')) != NULL) name2 = p + 1; } if (strcmp(name1, name2) == 0) return(1); if ((se = getservbyname(name1, proto)) != NULL) { if (strcmp(name2, se->s_name) == 0) return(1); for (alias = se->s_aliases; *alias; alias++) if (strcmp(name2, *alias) == 0) return(1); } return(0); } static struct servtab * enter(struct servtab *cp) { struct servtab *sep; long omask; sep = (struct servtab *)malloc(sizeof (*sep)); if (sep == (struct servtab *)0) { syslog(LOG_ERR, "malloc: %m"); exit(EX_OSERR); } *sep = *cp; sep->se_fd = -1; omask = sigblock(SIGBLOCK); sep->se_next = servtab; servtab = sep; sigsetmask(omask); return (sep); } static void enable(struct servtab *sep) { if (debug) warnx( "enabling %s, fd %d", sep->se_service, sep->se_fd); #ifdef SANITY_CHECK if (sep->se_fd < 0) { syslog(LOG_ERR, "%s: %s: bad fd", __func__, sep->se_service); exit(EX_SOFTWARE); } if (ISMUX(sep)) { syslog(LOG_ERR, "%s: %s: is mux", __func__, sep->se_service); exit(EX_SOFTWARE); } if (FD_ISSET(sep->se_fd, &allsock)) { syslog(LOG_ERR, "%s: %s: not off", __func__, sep->se_service); exit(EX_SOFTWARE); } nsock++; #endif FD_SET(sep->se_fd, &allsock); if (sep->se_fd > maxsock) maxsock = sep->se_fd; } static void disable(struct servtab *sep) { if (debug) warnx( "disabling %s, fd %d", sep->se_service, sep->se_fd); #ifdef SANITY_CHECK if (sep->se_fd < 0) { syslog(LOG_ERR, "%s: %s: bad fd", __func__, sep->se_service); exit(EX_SOFTWARE); } if (ISMUX(sep)) { syslog(LOG_ERR, "%s: %s: is mux", __func__, sep->se_service); exit(EX_SOFTWARE); } if (!FD_ISSET(sep->se_fd, &allsock)) { syslog(LOG_ERR, "%s: %s: not on", __func__, sep->se_service); exit(EX_SOFTWARE); } if (nsock == 0) { syslog(LOG_ERR, "%s: nsock=0", __func__); exit(EX_SOFTWARE); } nsock--; #endif FD_CLR(sep->se_fd, &allsock); if (sep->se_fd == maxsock) maxsock--; } static FILE *fconfig = NULL; static struct servtab serv; static char line[LINE_MAX]; static int setconfig(void) { if (fconfig != NULL) { fseek(fconfig, 0L, SEEK_SET); return (1); } fconfig = fopen(CONFIG, "r"); return (fconfig != NULL); } static void endconfig(void) { if (fconfig) { (void) fclose(fconfig); fconfig = NULL; } } static struct servtab * getconfigent(void) { struct servtab *sep = &serv; int argc; char *cp, *arg, *s; char *versp; static char TCPMUX_TOKEN[] = "tcpmux/"; #define MUX_LEN (sizeof(TCPMUX_TOKEN)-1) #ifdef IPSEC char *policy; #endif int v4bind; #ifdef INET6 int v6bind; #endif int i; #ifdef IPSEC policy = NULL; #endif more: v4bind = 0; #ifdef INET6 v6bind = 0; #endif while ((cp = nextline(fconfig)) != NULL) { #ifdef IPSEC /* lines starting with #@ is not a comment, but the policy */ if (cp[0] == '#' && cp[1] == '@') { char *p; for (p = cp + 2; p && *p && isspace(*p); p++) ; if (*p == '\0') { if (policy) free(policy); policy = NULL; } else if (ipsec_get_policylen(p) >= 0) { if (policy) free(policy); policy = newstr(p); } else { syslog(LOG_ERR, "%s: invalid ipsec policy \"%s\"", CONFIG, p); exit(EX_CONFIG); } } #endif if (*cp == '#' || *cp == '\0') continue; break; } if (cp == NULL) return ((struct servtab *)0); /* * clear the static buffer, since some fields (se_ctrladdr, * for example) don't get initialized here. */ memset(sep, 0, sizeof *sep); arg = skip(&cp); if (cp == NULL) { /* got an empty line containing just blanks/tabs. */ goto more; } if (arg[0] == ':') { /* :user:group:perm: */ char *user, *group, *perm; struct passwd *pw; struct group *gr; user = arg+1; if ((group = strchr(user, ':')) == NULL) { syslog(LOG_ERR, "no group after user '%s'", user); goto more; } *group++ = '\0'; if ((perm = strchr(group, ':')) == NULL) { syslog(LOG_ERR, "no mode after group '%s'", group); goto more; } *perm++ = '\0'; if ((pw = getpwnam(user)) == NULL) { syslog(LOG_ERR, "no such user '%s'", user); goto more; } sep->se_sockuid = pw->pw_uid; if ((gr = getgrnam(group)) == NULL) { syslog(LOG_ERR, "no such user '%s'", group); goto more; } sep->se_sockgid = gr->gr_gid; sep->se_sockmode = strtol(perm, &arg, 8); if (*arg != ':') { syslog(LOG_ERR, "bad mode '%s'", perm); goto more; } *arg++ = '\0'; } else { sep->se_sockuid = euid; sep->se_sockgid = egid; sep->se_sockmode = 0200; } if (strncmp(arg, TCPMUX_TOKEN, MUX_LEN) == 0) { char *c = arg + MUX_LEN; if (*c == '+') { sep->se_type = MUXPLUS_TYPE; c++; } else sep->se_type = MUX_TYPE; sep->se_service = newstr(c); } else { sep->se_service = newstr(arg); sep->se_type = NORM_TYPE; } arg = sskip(&cp); if (strcmp(arg, "stream") == 0) sep->se_socktype = SOCK_STREAM; else if (strcmp(arg, "dgram") == 0) sep->se_socktype = SOCK_DGRAM; else if (strcmp(arg, "rdm") == 0) sep->se_socktype = SOCK_RDM; else if (strcmp(arg, "seqpacket") == 0) sep->se_socktype = SOCK_SEQPACKET; else if (strcmp(arg, "raw") == 0) sep->se_socktype = SOCK_RAW; else sep->se_socktype = -1; arg = sskip(&cp); if (strncmp(arg, "tcp", 3) == 0) { sep->se_proto = newstr(strsep(&arg, "/")); if (arg != NULL && (strcmp(arg, "faith") == 0)) { syslog(LOG_ERR, "faith has been deprecated"); goto more; } } else { if (sep->se_type == NORM_TYPE && strncmp(arg, "faith/", 6) == 0) { syslog(LOG_ERR, "faith has been deprecated"); goto more; } sep->se_proto = newstr(arg); } if (strncmp(sep->se_proto, "rpc/", 4) == 0) { memmove(sep->se_proto, sep->se_proto + 4, strlen(sep->se_proto) + 1 - 4); sep->se_rpc = 1; sep->se_rpc_prog = sep->se_rpc_lowvers = sep->se_rpc_highvers = 0; if ((versp = strrchr(sep->se_service, '/'))) { *versp++ = '\0'; switch (sscanf(versp, "%u-%u", &sep->se_rpc_lowvers, &sep->se_rpc_highvers)) { case 2: break; case 1: sep->se_rpc_highvers = sep->se_rpc_lowvers; break; default: syslog(LOG_ERR, "bad RPC version specifier; %s", sep->se_service); freeconfig(sep); goto more; } } else { sep->se_rpc_lowvers = sep->se_rpc_highvers = 1; } } sep->se_nomapped = 0; if (strcmp(sep->se_proto, "unix") == 0) { sep->se_family = AF_UNIX; } else { while (isdigit(sep->se_proto[strlen(sep->se_proto) - 1])) { #ifdef INET6 if (sep->se_proto[strlen(sep->se_proto) - 1] == '6') { sep->se_proto[strlen(sep->se_proto) - 1] = '\0'; v6bind = 1; continue; } #endif if (sep->se_proto[strlen(sep->se_proto) - 1] == '4') { sep->se_proto[strlen(sep->se_proto) - 1] = '\0'; v4bind = 1; continue; } /* illegal version num */ syslog(LOG_ERR, "bad IP version for %s", sep->se_proto); freeconfig(sep); goto more; } #ifdef INET6 if (v6bind && !v6bind_ok) { syslog(LOG_INFO, "IPv6 bind is ignored for %s", sep->se_service); if (v4bind && v4bind_ok) v6bind = 0; else { freeconfig(sep); goto more; } } if (v6bind) { sep->se_family = AF_INET6; if (!v4bind || !v4bind_ok) sep->se_nomapped = 1; } else #endif { /* default to v4 bind if not v6 bind */ if (!v4bind_ok) { syslog(LOG_NOTICE, "IPv4 bind is ignored for %s", sep->se_service); freeconfig(sep); goto more; } sep->se_family = AF_INET; } } /* init ctladdr */ switch(sep->se_family) { case AF_INET: memcpy(&sep->se_ctrladdr4, bind_sa4, sizeof(sep->se_ctrladdr4)); sep->se_ctrladdr_size = sizeof(sep->se_ctrladdr4); break; #ifdef INET6 case AF_INET6: memcpy(&sep->se_ctrladdr6, bind_sa6, sizeof(sep->se_ctrladdr6)); sep->se_ctrladdr_size = sizeof(sep->se_ctrladdr6); break; #endif case AF_UNIX: if (strlen(sep->se_service) >= sizeof(sep->se_ctrladdr_un.sun_path)) { syslog(LOG_ERR, "domain socket pathname too long for service %s", sep->se_service); goto more; } memset(&sep->se_ctrladdr, 0, sizeof(sep->se_ctrladdr)); sep->se_ctrladdr_un.sun_family = sep->se_family; sep->se_ctrladdr_un.sun_len = strlen(sep->se_service); strcpy(sep->se_ctrladdr_un.sun_path, sep->se_service); sep->se_ctrladdr_size = SUN_LEN(&sep->se_ctrladdr_un); } arg = sskip(&cp); if (!strncmp(arg, "wait", 4)) sep->se_accept = 0; else if (!strncmp(arg, "nowait", 6)) sep->se_accept = 1; else { syslog(LOG_ERR, "%s: bad wait/nowait for service %s", CONFIG, sep->se_service); goto more; } sep->se_maxchild = -1; sep->se_maxcpm = -1; sep->se_maxperip = -1; if ((s = strchr(arg, '/')) != NULL) { char *eptr; u_long val; val = strtoul(s + 1, &eptr, 10); if (eptr == s + 1 || val > MAX_MAXCHLD) { syslog(LOG_ERR, "%s: bad max-child for service %s", CONFIG, sep->se_service); goto more; } if (debug) if (!sep->se_accept && val != 1) warnx("maxchild=%lu for wait service %s" " not recommended", val, sep->se_service); sep->se_maxchild = val; if (*eptr == '/') sep->se_maxcpm = strtol(eptr + 1, &eptr, 10); if (*eptr == '/') sep->se_maxperip = strtol(eptr + 1, &eptr, 10); /* * explicitly do not check for \0 for future expansion / * backwards compatibility */ } if (ISMUX(sep)) { /* * Silently enforce "nowait" mode for TCPMUX services * since they don't have an assigned port to listen on. */ sep->se_accept = 1; if (strcmp(sep->se_proto, "tcp")) { syslog(LOG_ERR, "%s: bad protocol for tcpmux service %s", CONFIG, sep->se_service); goto more; } if (sep->se_socktype != SOCK_STREAM) { syslog(LOG_ERR, "%s: bad socket type for tcpmux service %s", CONFIG, sep->se_service); goto more; } } sep->se_user = newstr(sskip(&cp)); #ifdef LOGIN_CAP if ((s = strrchr(sep->se_user, '/')) != NULL) { *s = '\0'; sep->se_class = newstr(s + 1); } else sep->se_class = newstr(RESOURCE_RC); #endif if ((s = strrchr(sep->se_user, ':')) != NULL) { *s = '\0'; sep->se_group = newstr(s + 1); } else sep->se_group = NULL; sep->se_server = newstr(sskip(&cp)); if ((sep->se_server_name = strrchr(sep->se_server, '/'))) sep->se_server_name++; if (strcmp(sep->se_server, "internal") == 0) { struct biltin *bi; for (bi = biltins; bi->bi_service; bi++) if (bi->bi_socktype == sep->se_socktype && matchservent(bi->bi_service, sep->se_service, sep->se_proto)) break; if (bi->bi_service == 0) { syslog(LOG_ERR, "internal service %s unknown", sep->se_service); goto more; } sep->se_accept = 1; /* force accept mode for built-ins */ sep->se_bi = bi; } else sep->se_bi = NULL; if (sep->se_maxperip < 0) sep->se_maxperip = maxperip; if (sep->se_maxcpm < 0) sep->se_maxcpm = maxcpm; if (sep->se_maxchild < 0) { /* apply default max-children */ if (sep->se_bi && sep->se_bi->bi_maxchild >= 0) sep->se_maxchild = sep->se_bi->bi_maxchild; else if (sep->se_accept) sep->se_maxchild = MAX(maxchild, 0); else sep->se_maxchild = 1; } if (sep->se_maxchild > 0) { sep->se_pids = malloc(sep->se_maxchild * sizeof(*sep->se_pids)); if (sep->se_pids == NULL) { syslog(LOG_ERR, "malloc: %m"); exit(EX_OSERR); } } argc = 0; for (arg = skip(&cp); cp; arg = skip(&cp)) if (argc < MAXARGV) { sep->se_argv[argc++] = newstr(arg); } else { syslog(LOG_ERR, "%s: too many arguments for service %s", CONFIG, sep->se_service); goto more; } while (argc <= MAXARGV) sep->se_argv[argc++] = NULL; for (i = 0; i < PERIPSIZE; ++i) LIST_INIT(&sep->se_conn[i]); #ifdef IPSEC sep->se_policy = policy ? newstr(policy) : NULL; #endif return (sep); } static void freeconfig(struct servtab *cp) { int i; if (cp->se_service) free(cp->se_service); if (cp->se_proto) free(cp->se_proto); if (cp->se_user) free(cp->se_user); if (cp->se_group) free(cp->se_group); #ifdef LOGIN_CAP if (cp->se_class) free(cp->se_class); #endif if (cp->se_server) free(cp->se_server); if (cp->se_pids) free(cp->se_pids); for (i = 0; i < MAXARGV; i++) if (cp->se_argv[i]) free(cp->se_argv[i]); free_connlist(cp); #ifdef IPSEC if (cp->se_policy) free(cp->se_policy); #endif } /* * Safe skip - if skip returns null, log a syntax error in the * configuration file and exit. */ static char * sskip(char **cpp) { char *cp; cp = skip(cpp); if (cp == NULL) { syslog(LOG_ERR, "%s: syntax error", CONFIG); exit(EX_DATAERR); } return (cp); } static char * skip(char **cpp) { char *cp = *cpp; char *start; char quote = '\0'; again: while (*cp == ' ' || *cp == '\t') cp++; if (*cp == '\0') { int c; c = getc(fconfig); (void) ungetc(c, fconfig); if (c == ' ' || c == '\t') if ((cp = nextline(fconfig))) goto again; *cpp = (char *)0; return ((char *)0); } if (*cp == '"' || *cp == '\'') quote = *cp++; start = cp; if (quote) while (*cp && *cp != quote) cp++; else while (*cp && *cp != ' ' && *cp != '\t') cp++; if (*cp != '\0') *cp++ = '\0'; *cpp = cp; return (start); } static char * nextline(FILE *fd) { char *cp; if (fgets(line, sizeof (line), fd) == NULL) return ((char *)0); cp = strchr(line, '\n'); if (cp) *cp = '\0'; return (line); } static char * newstr(const char *cp) { char *cr; if ((cr = strdup(cp != NULL ? cp : ""))) return (cr); syslog(LOG_ERR, "strdup: %m"); exit(EX_OSERR); } void inetd_setproctitle(const char *a, int s) { socklen_t size; struct sockaddr_storage ss; char buf[80], pbuf[NI_MAXHOST]; size = sizeof(ss); if (getpeername(s, (struct sockaddr *)&ss, &size) == 0) { getnameinfo((struct sockaddr *)&ss, size, pbuf, sizeof(pbuf), NULL, 0, NI_NUMERICHOST); (void) sprintf(buf, "%s [%s]", a, pbuf); } else (void) sprintf(buf, "%s", a); setproctitle("%s", buf); } int check_loop(const struct sockaddr *sa, const struct servtab *sep) { struct servtab *se2; char pname[NI_MAXHOST]; for (se2 = servtab; se2; se2 = se2->se_next) { if (!se2->se_bi || se2->se_socktype != SOCK_DGRAM) continue; switch (se2->se_family) { case AF_INET: if (csatosin(sa)->sin_port == se2->se_ctrladdr4.sin_port) goto isloop; continue; #ifdef INET6 case AF_INET6: if (csatosin6(sa)->sin6_port == se2->se_ctrladdr6.sin6_port) goto isloop; continue; #endif default: continue; } isloop: getnameinfo(sa, sa->sa_len, pname, sizeof(pname), NULL, 0, NI_NUMERICHOST); syslog(LOG_WARNING, "%s/%s:%s/%s loop request REFUSED from %s", sep->se_service, sep->se_proto, se2->se_service, se2->se_proto, pname); return 1; } return 0; } /* * print_service: * Dump relevant information to stderr */ static void print_service(const char *action, const struct servtab *sep) { fprintf(stderr, "%s: %s proto=%s accept=%d max=%d user=%s group=%s" #ifdef LOGIN_CAP "class=%s" #endif " builtin=%p server=%s" #ifdef IPSEC " policy=\"%s\"" #endif "\n", action, sep->se_service, sep->se_proto, sep->se_accept, sep->se_maxchild, sep->se_user, sep->se_group, #ifdef LOGIN_CAP sep->se_class, #endif (void *) sep->se_bi, sep->se_server #ifdef IPSEC , (sep->se_policy ? sep->se_policy : "") #endif ); } #define CPMHSIZE 256 #define CPMHMASK (CPMHSIZE-1) #define CHTGRAN 10 #define CHTSIZE 6 typedef struct CTime { unsigned long ct_Ticks; int ct_Count; } CTime; typedef struct CHash { union { struct in_addr c4_Addr; struct in6_addr c6_Addr; } cu_Addr; #define ch_Addr4 cu_Addr.c4_Addr #define ch_Addr6 cu_Addr.c6_Addr int ch_Family; time_t ch_LTime; char *ch_Service; CTime ch_Times[CHTSIZE]; } CHash; static CHash CHashAry[CPMHSIZE]; static int cpmip(const struct servtab *sep, int ctrl) { struct sockaddr_storage rss; socklen_t rssLen = sizeof(rss); int r = 0; /* * If getpeername() fails, just let it through (if logging is * enabled the condition is caught elsewhere) */ if (sep->se_maxcpm > 0 && (sep->se_family == AF_INET || sep->se_family == AF_INET6) && getpeername(ctrl, (struct sockaddr *)&rss, &rssLen) == 0 ) { time_t t = time(NULL); int hv = 0xABC3D20F; int i; int cnt = 0; CHash *chBest = NULL; unsigned int ticks = t / CHTGRAN; struct sockaddr_in *sin4; #ifdef INET6 struct sockaddr_in6 *sin6; #endif sin4 = (struct sockaddr_in *)&rss; #ifdef INET6 sin6 = (struct sockaddr_in6 *)&rss; #endif { char *p; int addrlen; switch (rss.ss_family) { case AF_INET: p = (char *)&sin4->sin_addr; addrlen = sizeof(struct in_addr); break; #ifdef INET6 case AF_INET6: p = (char *)&sin6->sin6_addr; addrlen = sizeof(struct in6_addr); break; #endif default: /* should not happen */ return -1; } for (i = 0; i < addrlen; ++i, ++p) { hv = (hv << 5) ^ (hv >> 23) ^ *p; } hv = (hv ^ (hv >> 16)); } for (i = 0; i < 5; ++i) { CHash *ch = &CHashAry[(hv + i) & CPMHMASK]; if (rss.ss_family == AF_INET && ch->ch_Family == AF_INET && sin4->sin_addr.s_addr == ch->ch_Addr4.s_addr && ch->ch_Service && strcmp(sep->se_service, ch->ch_Service) == 0) { chBest = ch; break; } #ifdef INET6 if (rss.ss_family == AF_INET6 && ch->ch_Family == AF_INET6 && IN6_ARE_ADDR_EQUAL(&sin6->sin6_addr, &ch->ch_Addr6) != 0 && ch->ch_Service && strcmp(sep->se_service, ch->ch_Service) == 0) { chBest = ch; break; } #endif if (chBest == NULL || ch->ch_LTime == 0 || ch->ch_LTime < chBest->ch_LTime) { chBest = ch; } } if ((rss.ss_family == AF_INET && (chBest->ch_Family != AF_INET || sin4->sin_addr.s_addr != chBest->ch_Addr4.s_addr)) || chBest->ch_Service == NULL || strcmp(sep->se_service, chBest->ch_Service) != 0) { chBest->ch_Family = sin4->sin_family; chBest->ch_Addr4 = sin4->sin_addr; if (chBest->ch_Service) free(chBest->ch_Service); chBest->ch_Service = strdup(sep->se_service); bzero(chBest->ch_Times, sizeof(chBest->ch_Times)); } #ifdef INET6 if ((rss.ss_family == AF_INET6 && (chBest->ch_Family != AF_INET6 || IN6_ARE_ADDR_EQUAL(&sin6->sin6_addr, &chBest->ch_Addr6) == 0)) || chBest->ch_Service == NULL || strcmp(sep->se_service, chBest->ch_Service) != 0) { chBest->ch_Family = sin6->sin6_family; chBest->ch_Addr6 = sin6->sin6_addr; if (chBest->ch_Service) free(chBest->ch_Service); chBest->ch_Service = strdup(sep->se_service); bzero(chBest->ch_Times, sizeof(chBest->ch_Times)); } #endif chBest->ch_LTime = t; { CTime *ct = &chBest->ch_Times[ticks % CHTSIZE]; if (ct->ct_Ticks != ticks) { ct->ct_Ticks = ticks; ct->ct_Count = 0; } ++ct->ct_Count; } for (i = 0; i < CHTSIZE; ++i) { CTime *ct = &chBest->ch_Times[i]; if (ct->ct_Ticks <= ticks && ct->ct_Ticks >= ticks - CHTSIZE) { cnt += ct->ct_Count; } } if ((cnt * 60) / (CHTSIZE * CHTGRAN) > sep->se_maxcpm) { char pname[NI_MAXHOST]; getnameinfo((struct sockaddr *)&rss, ((struct sockaddr *)&rss)->sa_len, pname, sizeof(pname), NULL, 0, NI_NUMERICHOST); r = -1; syslog(LOG_ERR, "%s from %s exceeded counts/min (limit %d/min)", sep->se_service, pname, sep->se_maxcpm); } } return(r); } static struct conninfo * search_conn(struct servtab *sep, int ctrl) { struct sockaddr_storage ss; socklen_t sslen = sizeof(ss); struct conninfo *conn; int hv; char pname[NI_MAXHOST], pname2[NI_MAXHOST]; if (sep->se_maxperip <= 0) return NULL; /* * If getpeername() fails, just let it through (if logging is * enabled the condition is caught elsewhere) */ if (getpeername(ctrl, (struct sockaddr *)&ss, &sslen) != 0) return NULL; switch (ss.ss_family) { case AF_INET: hv = hashval((char *)&((struct sockaddr_in *)&ss)->sin_addr, sizeof(struct in_addr)); break; #ifdef INET6 case AF_INET6: hv = hashval((char *)&((struct sockaddr_in6 *)&ss)->sin6_addr, sizeof(struct in6_addr)); break; #endif default: /* * Since we only support AF_INET and AF_INET6, just * let other than AF_INET and AF_INET6 through. */ return NULL; } if (getnameinfo((struct sockaddr *)&ss, sslen, pname, sizeof(pname), NULL, 0, NI_NUMERICHOST) != 0) return NULL; LIST_FOREACH(conn, &sep->se_conn[hv], co_link) { if (getnameinfo((struct sockaddr *)&conn->co_addr, conn->co_addr.ss_len, pname2, sizeof(pname2), NULL, 0, NI_NUMERICHOST) == 0 && strcmp(pname, pname2) == 0) break; } if (conn == NULL) { if ((conn = malloc(sizeof(struct conninfo))) == NULL) { syslog(LOG_ERR, "malloc: %m"); exit(EX_OSERR); } conn->co_proc = malloc(sep->se_maxperip * sizeof(*conn->co_proc)); if (conn->co_proc == NULL) { syslog(LOG_ERR, "malloc: %m"); exit(EX_OSERR); } memcpy(&conn->co_addr, (struct sockaddr *)&ss, sslen); conn->co_numchild = 0; LIST_INSERT_HEAD(&sep->se_conn[hv], conn, co_link); } /* * Since a child process is not invoked yet, we cannot * determine a pid of a child. So, co_proc and co_numchild * should be filled leter. */ return conn; } static int room_conn(struct servtab *sep, struct conninfo *conn) { char pname[NI_MAXHOST]; if (conn->co_numchild >= sep->se_maxperip) { getnameinfo((struct sockaddr *)&conn->co_addr, conn->co_addr.ss_len, pname, sizeof(pname), NULL, 0, NI_NUMERICHOST); syslog(LOG_ERR, "%s from %s exceeded counts (limit %d)", sep->se_service, pname, sep->se_maxperip); return 0; } return 1; } static void addchild_conn(struct conninfo *conn, pid_t pid) { struct procinfo *proc; if (conn == NULL) return; if ((proc = search_proc(pid, 1)) != NULL) { if (proc->pr_conn != NULL) { syslog(LOG_ERR, "addchild_conn: child already on process list"); exit(EX_OSERR); } proc->pr_conn = conn; } conn->co_proc[conn->co_numchild++] = proc; } static void reapchild_conn(pid_t pid) { struct procinfo *proc; struct conninfo *conn; int i; if ((proc = search_proc(pid, 0)) == NULL) return; if ((conn = proc->pr_conn) == NULL) return; for (i = 0; i < conn->co_numchild; ++i) if (conn->co_proc[i] == proc) { conn->co_proc[i] = conn->co_proc[--conn->co_numchild]; break; } free_proc(proc); free_conn(conn); } static void resize_conn(struct servtab *sep, int maxpip) { struct conninfo *conn; int i, j; if (sep->se_maxperip <= 0) return; if (maxpip <= 0) { free_connlist(sep); return; } for (i = 0; i < PERIPSIZE; ++i) { LIST_FOREACH(conn, &sep->se_conn[i], co_link) { for (j = maxpip; j < conn->co_numchild; ++j) free_proc(conn->co_proc[j]); conn->co_proc = realloc(conn->co_proc, maxpip * sizeof(*conn->co_proc)); if (conn->co_proc == NULL) { syslog(LOG_ERR, "realloc: %m"); exit(EX_OSERR); } if (conn->co_numchild > maxpip) conn->co_numchild = maxpip; } } } static void free_connlist(struct servtab *sep) { struct conninfo *conn; int i, j; for (i = 0; i < PERIPSIZE; ++i) { while ((conn = LIST_FIRST(&sep->se_conn[i])) != NULL) { for (j = 0; j < conn->co_numchild; ++j) free_proc(conn->co_proc[j]); conn->co_numchild = 0; free_conn(conn); } } } static void free_conn(struct conninfo *conn) { if (conn == NULL) return; if (conn->co_numchild <= 0) { LIST_REMOVE(conn, co_link); free(conn->co_proc); free(conn); } } static struct procinfo * search_proc(pid_t pid, int add) { struct procinfo *proc; int hv; hv = hashval((char *)&pid, sizeof(pid)); LIST_FOREACH(proc, &proctable[hv], pr_link) { if (proc->pr_pid == pid) break; } if (proc == NULL && add) { if ((proc = malloc(sizeof(struct procinfo))) == NULL) { syslog(LOG_ERR, "malloc: %m"); exit(EX_OSERR); } proc->pr_pid = pid; proc->pr_conn = NULL; LIST_INSERT_HEAD(&proctable[hv], proc, pr_link); } return proc; } static void free_proc(struct procinfo *proc) { if (proc == NULL) return; LIST_REMOVE(proc, pr_link); free(proc); } static int hashval(char *p, int len) { int i, hv = 0xABC3D20F; for (i = 0; i < len; ++i, ++p) hv = (hv << 5) ^ (hv >> 23) ^ *p; hv = (hv ^ (hv >> 16)) & (PERIPSIZE - 1); return hv; } Index: projects/clang400-import =================================================================== --- projects/clang400-import (revision 311459) +++ projects/clang400-import (revision 311460) Property changes on: projects/clang400-import ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head:r311314-311459