uipc_domain.c revision 1.80 1 /* $NetBSD: uipc_domain.c,v 1.80 2009/04/28 20:54:50 dyoung Exp $ */
2
3 /*
4 * Copyright (c) 1982, 1986, 1993
5 * The Regents of the University of California. All rights reserved.
6 *
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
9 * are met:
10 * 1. Redistributions of source code must retain the above copyright
11 * notice, this list of conditions and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 * notice, this list of conditions and the following disclaimer in the
14 * documentation and/or other materials provided with the distribution.
15 * 3. Neither the name of the University nor the names of its contributors
16 * may be used to endorse or promote products derived from this software
17 * without specific prior written permission.
18 *
19 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29 * SUCH DAMAGE.
30 *
31 * @(#)uipc_domain.c 8.3 (Berkeley) 2/14/95
32 */
33
34 #include <sys/cdefs.h>
35 __KERNEL_RCSID(0, "$NetBSD: uipc_domain.c,v 1.80 2009/04/28 20:54:50 dyoung Exp $");
36
37 #include <sys/param.h>
38 #include <sys/socket.h>
39 #include <sys/socketvar.h>
40 #include <sys/protosw.h>
41 #include <sys/domain.h>
42 #include <sys/mbuf.h>
43 #include <sys/time.h>
44 #include <sys/kernel.h>
45 #include <sys/systm.h>
46 #include <sys/callout.h>
47 #include <sys/queue.h>
48 #include <sys/proc.h>
49 #include <sys/sysctl.h>
50 #include <sys/un.h>
51 #include <sys/unpcb.h>
52 #include <sys/file.h>
53 #include <sys/filedesc.h>
54 #include <sys/kauth.h>
55
56 MALLOC_DECLARE(M_SOCKADDR);
57
58 MALLOC_DEFINE(M_SOCKADDR, "sockaddr", "socket endpoints");
59
60 void pffasttimo(void *);
61 void pfslowtimo(void *);
62
63 struct domainhead domains = STAILQ_HEAD_INITIALIZER(domains);
64 static struct domain *domain_array[AF_MAX];
65
66 callout_t pffasttimo_ch, pfslowtimo_ch;
67
68 /*
69 * Current time values for fast and slow timeouts. We can use u_int
70 * relatively safely. The fast timer will roll over in 27 years and
71 * the slow timer in 68 years.
72 */
73 u_int pfslowtimo_now;
74 u_int pffasttimo_now;
75
76 static struct sysctllog *domain_sysctllog;
77 static void sysctl_net_setup(void);
78
79 void
80 domaininit(void)
81 {
82 __link_set_decl(domains, struct domain);
83 struct domain * const * dpp;
84 struct domain *rt_domain = NULL;
85
86 sysctl_net_setup();
87
88 /*
89 * Add all of the domains. Make sure the PF_ROUTE
90 * domain is added last.
91 */
92 __link_set_foreach(dpp, domains) {
93 if ((*dpp)->dom_family == PF_ROUTE)
94 rt_domain = *dpp;
95 else
96 domain_attach(*dpp);
97 }
98 if (rt_domain)
99 domain_attach(rt_domain);
100
101 callout_init(&pffasttimo_ch, CALLOUT_MPSAFE);
102 callout_init(&pfslowtimo_ch, CALLOUT_MPSAFE);
103
104 callout_reset(&pffasttimo_ch, 1, pffasttimo, NULL);
105 callout_reset(&pfslowtimo_ch, 1, pfslowtimo, NULL);
106 }
107
108 void
109 domain_attach(struct domain *dp)
110 {
111 const struct protosw *pr;
112
113 STAILQ_INSERT_TAIL(&domains, dp, dom_link);
114 if (dp->dom_family < __arraycount(domain_array))
115 domain_array[dp->dom_family] = dp;
116
117 if (dp->dom_init)
118 (*dp->dom_init)();
119
120 #ifdef MBUFTRACE
121 if (dp->dom_mowner.mo_name[0] == '\0') {
122 strncpy(dp->dom_mowner.mo_name, dp->dom_name,
123 sizeof(dp->dom_mowner.mo_name));
124 MOWNER_ATTACH(&dp->dom_mowner);
125 }
126 #endif
127 for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++) {
128 if (pr->pr_init)
129 (*pr->pr_init)();
130 }
131
132 if (max_linkhdr < 16) /* XXX */
133 max_linkhdr = 16;
134 max_hdr = max_linkhdr + max_protohdr;
135 max_datalen = MHLEN - max_hdr;
136 }
137
138 struct domain *
139 pffinddomain(int family)
140 {
141 struct domain *dp;
142
143 if (family < __arraycount(domain_array) && domain_array[family] != NULL)
144 return domain_array[family];
145
146 DOMAIN_FOREACH(dp)
147 if (dp->dom_family == family)
148 return (dp);
149 return (NULL);
150 }
151
152 const struct protosw *
153 pffindtype(int family, int type)
154 {
155 struct domain *dp;
156 const struct protosw *pr;
157
158 dp = pffinddomain(family);
159 if (dp == NULL)
160 return (NULL);
161
162 for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++)
163 if (pr->pr_type && pr->pr_type == type)
164 return (pr);
165
166 return (NULL);
167 }
168
169 const struct protosw *
170 pffindproto(int family, int protocol, int type)
171 {
172 struct domain *dp;
173 const struct protosw *pr;
174 const struct protosw *maybe = NULL;
175
176 if (family == 0)
177 return (NULL);
178
179 dp = pffinddomain(family);
180 if (dp == NULL)
181 return (NULL);
182
183 for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++) {
184 if ((pr->pr_protocol == protocol) && (pr->pr_type == type))
185 return (pr);
186
187 if (type == SOCK_RAW && pr->pr_type == SOCK_RAW &&
188 pr->pr_protocol == 0 && maybe == NULL)
189 maybe = pr;
190 }
191 return (maybe);
192 }
193
194 void *
195 sockaddr_addr(struct sockaddr *sa, socklen_t *slenp)
196 {
197 const struct domain *dom;
198
199 if ((dom = pffinddomain(sa->sa_family)) == NULL ||
200 dom->dom_sockaddr_addr == NULL)
201 return NULL;
202
203 return (*dom->dom_sockaddr_addr)(sa, slenp);
204 }
205
206 const void *
207 sockaddr_const_addr(const struct sockaddr *sa, socklen_t *slenp)
208 {
209 const struct domain *dom;
210
211 if ((dom = pffinddomain(sa->sa_family)) == NULL ||
212 dom->dom_sockaddr_const_addr == NULL)
213 return NULL;
214
215 return (*dom->dom_sockaddr_const_addr)(sa, slenp);
216 }
217
218 const struct sockaddr *
219 sockaddr_any_by_family(int family)
220 {
221 const struct domain *dom;
222
223 if ((dom = pffinddomain(family)) == NULL)
224 return NULL;
225
226 return dom->dom_sa_any;
227 }
228
229 const struct sockaddr *
230 sockaddr_any(const struct sockaddr *sa)
231 {
232 return sockaddr_any_by_family(sa->sa_family);
233 }
234
235 const void *
236 sockaddr_anyaddr(const struct sockaddr *sa, socklen_t *slenp)
237 {
238 const struct sockaddr *any;
239
240 if ((any = sockaddr_any(sa)) == NULL)
241 return NULL;
242
243 return sockaddr_const_addr(any, slenp);
244 }
245
246 struct sockaddr *
247 sockaddr_alloc(sa_family_t af, socklen_t socklen, int flags)
248 {
249 struct sockaddr *sa;
250 socklen_t reallen = MAX(socklen, offsetof(struct sockaddr, sa_data[0]));
251
252 if ((sa = malloc(reallen, M_SOCKADDR, flags)) == NULL)
253 return NULL;
254
255 sa->sa_family = af;
256 sa->sa_len = reallen;
257 return sa;
258 }
259
260 struct sockaddr *
261 sockaddr_copy(struct sockaddr *dst, socklen_t socklen,
262 const struct sockaddr *src)
263 {
264 if (__predict_false(socklen < src->sa_len)) {
265 panic("%s: source too long, %d < %d bytes", __func__, socklen,
266 src->sa_len);
267 }
268 return memcpy(dst, src, src->sa_len);
269 }
270
271 int
272 sockaddr_cmp(const struct sockaddr *sa1, const struct sockaddr *sa2)
273 {
274 int len, rc;
275 struct domain *dom;
276
277 if (sa1->sa_family != sa2->sa_family)
278 return sa1->sa_family - sa2->sa_family;
279
280 dom = pffinddomain(sa1->sa_family);
281
282 if (dom != NULL && dom->dom_sockaddr_cmp != NULL)
283 return (*dom->dom_sockaddr_cmp)(sa1, sa2);
284
285 len = MIN(sa1->sa_len, sa2->sa_len);
286
287 if (dom == NULL || dom->dom_sa_cmplen == 0) {
288 if ((rc = memcmp(sa1, sa2, len)) != 0)
289 return rc;
290 return sa1->sa_len - sa2->sa_len;
291 }
292
293 if ((rc = memcmp((const char *)sa1 + dom->dom_sa_cmpofs,
294 (const char *)sa2 + dom->dom_sa_cmpofs,
295 MIN(dom->dom_sa_cmplen,
296 len - MIN(len, dom->dom_sa_cmpofs)))) != 0)
297 return rc;
298
299 return MIN(dom->dom_sa_cmplen + dom->dom_sa_cmpofs, sa1->sa_len) -
300 MIN(dom->dom_sa_cmplen + dom->dom_sa_cmpofs, sa2->sa_len);
301 }
302
303 struct sockaddr *
304 sockaddr_dup(const struct sockaddr *src, int flags)
305 {
306 struct sockaddr *dst;
307
308 if ((dst = sockaddr_alloc(src->sa_family, src->sa_len, flags)) == NULL)
309 return NULL;
310
311 return sockaddr_copy(dst, dst->sa_len, src);
312 }
313
314 void
315 sockaddr_free(struct sockaddr *sa)
316 {
317 free(sa, M_SOCKADDR);
318 }
319
320 /*
321 * sysctl helper to stuff PF_LOCAL pcbs into sysctl structures
322 */
323 static void
324 sysctl_dounpcb(struct kinfo_pcb *pcb, const struct socket *so)
325 {
326 struct unpcb *unp = sotounpcb(so);
327 struct sockaddr_un *un = unp->unp_addr;
328
329 memset(pcb, 0, sizeof(*pcb));
330
331 pcb->ki_family = so->so_proto->pr_domain->dom_family;
332 pcb->ki_type = so->so_proto->pr_type;
333 pcb->ki_protocol = so->so_proto->pr_protocol;
334 pcb->ki_pflags = unp->unp_flags;
335
336 pcb->ki_pcbaddr = PTRTOUINT64(unp);
337 /* pcb->ki_ppcbaddr = unp has no ppcb... */
338 pcb->ki_sockaddr = PTRTOUINT64(so);
339
340 pcb->ki_sostate = so->so_state;
341 /* pcb->ki_prstate = unp has no state... */
342
343 pcb->ki_rcvq = so->so_rcv.sb_cc;
344 pcb->ki_sndq = so->so_snd.sb_cc;
345
346 un = (struct sockaddr_un *)&pcb->ki_src;
347 /*
348 * local domain sockets may bind without having a local
349 * endpoint. bleah!
350 */
351 if (unp->unp_addr != NULL) {
352 un->sun_len = unp->unp_addr->sun_len;
353 un->sun_family = unp->unp_addr->sun_family;
354 strlcpy(un->sun_path, unp->unp_addr->sun_path,
355 sizeof(pcb->ki_s));
356 }
357 else {
358 un->sun_len = offsetof(struct sockaddr_un, sun_path);
359 un->sun_family = pcb->ki_family;
360 }
361 if (unp->unp_conn != NULL) {
362 un = (struct sockaddr_un *)&pcb->ki_dst;
363 if (unp->unp_conn->unp_addr != NULL) {
364 un->sun_len = unp->unp_conn->unp_addr->sun_len;
365 un->sun_family = unp->unp_conn->unp_addr->sun_family;
366 un->sun_family = unp->unp_conn->unp_addr->sun_family;
367 strlcpy(un->sun_path, unp->unp_conn->unp_addr->sun_path,
368 sizeof(pcb->ki_d));
369 }
370 else {
371 un->sun_len = offsetof(struct sockaddr_un, sun_path);
372 un->sun_family = pcb->ki_family;
373 }
374 }
375
376 pcb->ki_inode = unp->unp_ino;
377 pcb->ki_vnode = PTRTOUINT64(unp->unp_vnode);
378 pcb->ki_conn = PTRTOUINT64(unp->unp_conn);
379 pcb->ki_refs = PTRTOUINT64(unp->unp_refs);
380 pcb->ki_nextref = PTRTOUINT64(unp->unp_nextref);
381 }
382
383 static int
384 sysctl_unpcblist(SYSCTLFN_ARGS)
385 {
386 struct file *fp, *dfp, *np;
387 struct socket *so;
388 struct kinfo_pcb pcb;
389 char *dp;
390 u_int op, arg;
391 size_t len, needed, elem_size, out_size;
392 int error, elem_count, pf, type, pf2;
393
394 if (namelen == 1 && name[0] == CTL_QUERY)
395 return (sysctl_query(SYSCTLFN_CALL(rnode)));
396
397 if (namelen != 4)
398 return (EINVAL);
399
400 if (oldp != NULL) {
401 len = *oldlenp;
402 elem_size = name[2];
403 elem_count = name[3];
404 if (elem_size != sizeof(pcb))
405 return EINVAL;
406 } else {
407 len = 0;
408 elem_size = sizeof(pcb);
409 elem_count = INT_MAX;
410 }
411 error = 0;
412 dp = oldp;
413 op = name[0];
414 arg = name[1];
415 out_size = elem_size;
416 needed = 0;
417
418 if (name - oname != 4)
419 return (EINVAL);
420
421 pf = oname[1];
422 type = oname[2];
423 pf2 = (oldp == NULL) ? 0 : pf;
424
425 /*
426 * allocate dummy file descriptor to make position in list.
427 */
428 sysctl_unlock();
429 if ((dfp = fgetdummy()) == NULL) {
430 sysctl_relock();
431 return ENOMEM;
432 }
433
434 /*
435 * there's no "list" of local domain sockets, so we have
436 * to walk the file list looking for them. :-/
437 */
438 mutex_enter(&filelist_lock);
439 LIST_FOREACH(fp, &filehead, f_list) {
440 np = LIST_NEXT(fp, f_list);
441 if (fp->f_count == 0 || fp->f_type != DTYPE_SOCKET ||
442 fp->f_data == NULL)
443 continue;
444 if (kauth_authorize_generic(l->l_cred,
445 KAUTH_GENERIC_CANSEE, fp->f_cred) != 0)
446 continue;
447 so = (struct socket *)fp->f_data;
448 if (so->so_type != type)
449 continue;
450 if (so->so_proto->pr_domain->dom_family != pf)
451 continue;
452 if (len >= elem_size && elem_count > 0) {
453 mutex_enter(&fp->f_lock);
454 fp->f_count++;
455 mutex_exit(&fp->f_lock);
456 LIST_INSERT_AFTER(fp, dfp, f_list);
457 mutex_exit(&filelist_lock);
458 sysctl_dounpcb(&pcb, so);
459 error = copyout(&pcb, dp, out_size);
460 closef(fp);
461 mutex_enter(&filelist_lock);
462 np = LIST_NEXT(dfp, f_list);
463 LIST_REMOVE(dfp, f_list);
464 if (error)
465 break;
466 dp += elem_size;
467 len -= elem_size;
468 }
469 needed += elem_size;
470 if (elem_count > 0 && elem_count != INT_MAX)
471 elem_count--;
472 }
473 mutex_exit(&filelist_lock);
474 fputdummy(dfp);
475 *oldlenp = needed;
476 if (oldp == NULL)
477 *oldlenp += PCB_SLOP * sizeof(struct kinfo_pcb);
478 sysctl_relock();
479
480 return (error);
481 }
482
483 static void
484 sysctl_net_setup(void)
485 {
486
487 KASSERT(domain_sysctllog == NULL);
488 sysctl_createv(&domain_sysctllog, 0, NULL, NULL,
489 CTLFLAG_PERMANENT,
490 CTLTYPE_NODE, "net", NULL,
491 NULL, 0, NULL, 0,
492 CTL_NET, CTL_EOL);
493 sysctl_createv(&domain_sysctllog, 0, NULL, NULL,
494 CTLFLAG_PERMANENT,
495 CTLTYPE_NODE, "local",
496 SYSCTL_DESCR("PF_LOCAL related settings"),
497 NULL, 0, NULL, 0,
498 CTL_NET, PF_LOCAL, CTL_EOL);
499 sysctl_createv(&domain_sysctllog, 0, NULL, NULL,
500 CTLFLAG_PERMANENT,
501 CTLTYPE_NODE, "stream",
502 SYSCTL_DESCR("SOCK_STREAM settings"),
503 NULL, 0, NULL, 0,
504 CTL_NET, PF_LOCAL, SOCK_STREAM, CTL_EOL);
505 sysctl_createv(&domain_sysctllog, 0, NULL, NULL,
506 CTLFLAG_PERMANENT,
507 CTLTYPE_NODE, "dgram",
508 SYSCTL_DESCR("SOCK_DGRAM settings"),
509 NULL, 0, NULL, 0,
510 CTL_NET, PF_LOCAL, SOCK_DGRAM, CTL_EOL);
511
512 sysctl_createv(&domain_sysctllog, 0, NULL, NULL,
513 CTLFLAG_PERMANENT,
514 CTLTYPE_STRUCT, "pcblist",
515 SYSCTL_DESCR("SOCK_STREAM protocol control block list"),
516 sysctl_unpcblist, 0, NULL, 0,
517 CTL_NET, PF_LOCAL, SOCK_STREAM, CTL_CREATE, CTL_EOL);
518 sysctl_createv(&domain_sysctllog, 0, NULL, NULL,
519 CTLFLAG_PERMANENT,
520 CTLTYPE_STRUCT, "pcblist",
521 SYSCTL_DESCR("SOCK_DGRAM protocol control block list"),
522 sysctl_unpcblist, 0, NULL, 0,
523 CTL_NET, PF_LOCAL, SOCK_DGRAM, CTL_CREATE, CTL_EOL);
524 }
525
526 void
527 pfctlinput(int cmd, const struct sockaddr *sa)
528 {
529 struct domain *dp;
530 const struct protosw *pr;
531
532 DOMAIN_FOREACH(dp) {
533 for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++) {
534 if (pr->pr_ctlinput != NULL)
535 (*pr->pr_ctlinput)(cmd, sa, NULL);
536 }
537 }
538 }
539
540 void
541 pfctlinput2(int cmd, const struct sockaddr *sa, void *ctlparam)
542 {
543 struct domain *dp;
544 const struct protosw *pr;
545
546 if (sa == NULL)
547 return;
548
549 DOMAIN_FOREACH(dp) {
550 /*
551 * the check must be made by xx_ctlinput() anyways, to
552 * make sure we use data item pointed to by ctlparam in
553 * correct way. the following check is made just for safety.
554 */
555 if (dp->dom_family != sa->sa_family)
556 continue;
557
558 for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++) {
559 if (pr->pr_ctlinput != NULL)
560 (*pr->pr_ctlinput)(cmd, sa, ctlparam);
561 }
562 }
563 }
564
565 void
566 pfslowtimo(void *arg)
567 {
568 struct domain *dp;
569 const struct protosw *pr;
570
571 pfslowtimo_now++;
572
573 DOMAIN_FOREACH(dp) {
574 for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++)
575 if (pr->pr_slowtimo)
576 (*pr->pr_slowtimo)();
577 }
578 callout_schedule(&pfslowtimo_ch, hz / 2);
579 }
580
581 void
582 pffasttimo(void *arg)
583 {
584 struct domain *dp;
585 const struct protosw *pr;
586
587 pffasttimo_now++;
588
589 DOMAIN_FOREACH(dp) {
590 for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++)
591 if (pr->pr_fasttimo)
592 (*pr->pr_fasttimo)();
593 }
594 callout_schedule(&pffasttimo_ch, hz / 5);
595 }
596