2 * SPDX-License-Identifier: BSD-3-Clause
4 * Copyright (c) 1982, 1986, 1993
5 * The Regents of the University of California. All rights reserved.
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
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.
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
31 * @(#)uipc_domain.c 8.2 (Berkeley) 10/18/93
34 #include <sys/cdefs.h>
35 #include <sys/param.h>
36 #include <sys/socket.h>
37 #include <sys/protosw.h>
38 #include <sys/domain.h>
39 #include <sys/eventhandler.h>
40 #include <sys/epoch.h>
42 #include <sys/kernel.h>
44 #include <sys/mutex.h>
45 #include <sys/rmlock.h>
46 #include <sys/socketvar.h>
47 #include <sys/systm.h>
52 * System initialization
54 * Note: domain initialization takes place on a per domain basis
55 * as a result of traversing a SYSINIT linker set. Most likely,
56 * each domain would want to call DOMAIN_SET(9) itself, which
57 * would cause the domain to be added just after domaininit()
58 * is called during startup.
60 * See DOMAIN_SET(9) for details on its use.
63 static void domaininit(void *);
64 SYSINIT(domain, SI_SUB_PROTO_DOMAININIT, SI_ORDER_ANY, domaininit, NULL);
66 static void domainfinalize(void *);
67 SYSINIT(domainfin, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_FIRST, domainfinalize,
70 static struct callout pffast_callout;
71 static struct callout pfslow_callout;
73 static void pffasttimo(void *);
74 static void pfslowtimo(void *);
76 static struct rmlock pftimo_lock;
77 RM_SYSINIT(pftimo_lock, &pftimo_lock, "pftimo");
79 static LIST_HEAD(, protosw) pffast_list =
80 LIST_HEAD_INITIALIZER(pffast_list);
81 static LIST_HEAD(, protosw) pfslow_list =
82 LIST_HEAD_INITIALIZER(pfslow_list);
84 struct domain *domains; /* registered protocol domains */
85 int domain_init_status = 0;
86 static struct mtx dom_mtx; /* domain list lock */
87 MTX_SYSINIT(domain, &dom_mtx, "domain list", MTX_DEF);
90 * Dummy protocol specific user requests function pointer array.
91 * All functions return EOPNOTSUPP.
93 struct pr_usrreqs nousrreqs = {
94 .pru_accept = pru_accept_notsupp,
95 .pru_attach = pru_attach_notsupp,
96 .pru_bind = pru_bind_notsupp,
97 .pru_connect = pru_connect_notsupp,
98 .pru_connect2 = pru_connect2_notsupp,
99 .pru_control = pru_control_notsupp,
100 .pru_disconnect = pru_disconnect_notsupp,
101 .pru_listen = pru_listen_notsupp,
102 .pru_peeraddr = pru_peeraddr_notsupp,
103 .pru_rcvd = pru_rcvd_notsupp,
104 .pru_rcvoob = pru_rcvoob_notsupp,
105 .pru_send = pru_send_notsupp,
106 .pru_sense = pru_sense_null,
107 .pru_shutdown = pru_shutdown_notsupp,
108 .pru_sockaddr = pru_sockaddr_notsupp,
109 .pru_sosend = pru_sosend_notsupp,
110 .pru_soreceive = pru_soreceive_notsupp,
111 .pru_sopoll = pru_sopoll_notsupp,
115 protosw_init(struct protosw *pr)
117 struct pr_usrreqs *pu;
120 KASSERT(pu != NULL, ("protosw_init: %ssw[%d] has no usrreqs!",
121 pr->pr_domain->dom_name,
122 (int)(pr - pr->pr_domain->dom_protosw)));
125 * Protocol switch methods fall into three categories: mandatory,
126 * mandatory but protosw_init() provides a default, and optional.
128 * For true protocols (i.e., pru_attach != NULL), KASSERT truly
129 * mandatory methods with no defaults, and initialize defaults for
130 * other mandatory methods if the protocol hasn't defined an
131 * implementation (NULL function pointer).
134 if (pu->pru_attach != NULL) {
135 KASSERT(pu->pru_abort != NULL,
136 ("protosw_init: %ssw[%d] pru_abort NULL",
137 pr->pr_domain->dom_name,
138 (int)(pr - pr->pr_domain->dom_protosw)));
139 KASSERT(pu->pru_send != NULL,
140 ("protosw_init: %ssw[%d] pru_send NULL",
141 pr->pr_domain->dom_name,
142 (int)(pr - pr->pr_domain->dom_protosw)));
146 #define DEFAULT(foo, bar) if ((foo) == NULL) (foo) = (bar)
147 DEFAULT(pu->pru_accept, pru_accept_notsupp);
148 DEFAULT(pu->pru_aio_queue, pru_aio_queue_notsupp);
149 DEFAULT(pu->pru_bind, pru_bind_notsupp);
150 DEFAULT(pu->pru_bindat, pru_bindat_notsupp);
151 DEFAULT(pu->pru_connect, pru_connect_notsupp);
152 DEFAULT(pu->pru_connect2, pru_connect2_notsupp);
153 DEFAULT(pu->pru_connectat, pru_connectat_notsupp);
154 DEFAULT(pu->pru_control, pru_control_notsupp);
155 DEFAULT(pu->pru_disconnect, pru_disconnect_notsupp);
156 DEFAULT(pu->pru_listen, pru_listen_notsupp);
157 DEFAULT(pu->pru_peeraddr, pru_peeraddr_notsupp);
158 DEFAULT(pu->pru_rcvd, pru_rcvd_notsupp);
159 DEFAULT(pu->pru_rcvoob, pru_rcvoob_notsupp);
160 DEFAULT(pu->pru_sense, pru_sense_null);
161 DEFAULT(pu->pru_shutdown, pru_shutdown_notsupp);
162 DEFAULT(pu->pru_sockaddr, pru_sockaddr_notsupp);
163 DEFAULT(pu->pru_sosend, sosend_generic);
164 DEFAULT(pu->pru_soreceive, soreceive_generic);
165 DEFAULT(pu->pru_sopoll, sopoll_generic);
166 DEFAULT(pu->pru_ready, pru_ready_notsupp);
173 * Add a new protocol domain to the list of supported domains
174 * Note: you cant unload it again because a socket may be using it.
175 * XXX can't fail at this time.
178 domain_init(void *arg)
180 struct domain *dp = arg;
185 for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++) {
189 * Note that with VIMAGE enabled, domain_init() will be
190 * re-invoked for each new vnet that's created. The below lists
191 * are intended to be system-wide, so avoid altering global
192 * state for non-default vnets.
194 if (IS_DEFAULT_VNET(curvnet)) {
195 rm_wlock(&pftimo_lock);
196 if (pr->pr_fasttimo != NULL)
197 LIST_INSERT_HEAD(&pffast_list, pr,
199 if (pr->pr_slowtimo != NULL)
200 LIST_INSERT_HEAD(&pfslow_list, pr,
202 rm_wunlock(&pftimo_lock);
207 * update global information about maximums
209 max_hdr = max_linkhdr + max_protohdr;
210 max_datalen = MHLEN - max_hdr;
212 panic("%s: max_datalen < 1", __func__);
217 vnet_domain_init(void *arg)
220 /* Virtualized case is no different -- call init functions. */
225 vnet_domain_uninit(void *arg)
227 struct domain *dp = arg;
230 (*dp->dom_destroy)();
235 * Add a new protocol domain to the list of supported domains
236 * Note: you cant unload it again because a socket may be using it.
237 * XXX can't fail at this time.
240 domain_add(void *data)
244 dp = (struct domain *)data;
246 dp->dom_next = domains;
249 KASSERT(domain_init_status >= 1,
250 ("attempt to domain_add(%s) before domaininit()",
253 if (domain_init_status < 1)
254 printf("WARNING: attempt to domain_add(%s) before "
255 "domaininit()\n", dp->dom_name);
258 KASSERT(domain_init_status < 2,
259 ("attempt to domain_add(%s) after domainfinalize()",
262 if (domain_init_status >= 2)
263 printf("WARNING: attempt to domain_add(%s) after "
264 "domainfinalize()\n", dp->dom_name);
266 mtx_unlock(&dom_mtx);
270 domain_remove(void *data)
272 struct domain *dp = (struct domain *)data;
274 if (dp->dom_family != PF_NETLINK)
279 domains = dp->dom_next;
282 for (curr = domains; curr != NULL; curr = curr->dom_next) {
283 if (curr->dom_next == dp) {
284 curr->dom_next = dp->dom_next;
289 mtx_unlock(&dom_mtx);
294 domaininit(void *dummy)
297 if (max_linkhdr < 16) /* XXX */
300 callout_init(&pffast_callout, 1);
301 callout_init(&pfslow_callout, 1);
304 KASSERT(domain_init_status == 0, ("domaininit called too late!"));
305 domain_init_status = 1;
306 mtx_unlock(&dom_mtx);
311 domainfinalize(void *dummy)
315 KASSERT(domain_init_status == 1, ("domainfinalize called too late!"));
316 domain_init_status = 2;
317 mtx_unlock(&dom_mtx);
319 callout_reset(&pffast_callout, 1, pffasttimo, NULL);
320 callout_reset(&pfslow_callout, 1, pfslowtimo, NULL);
324 pffinddomain(int family)
328 for (dp = domains; dp != NULL; dp = dp->dom_next)
329 if (dp->dom_family == family)
335 pffindtype(int family, int type)
340 dp = pffinddomain(family);
344 for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++)
345 if (pr->pr_type && pr->pr_type == type)
351 pffindproto(int family, int protocol, int type)
355 struct protosw *maybe;
361 dp = pffinddomain(family);
365 for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++) {
366 if ((pr->pr_protocol == protocol) && (pr->pr_type == type))
369 if (type == SOCK_RAW && pr->pr_type == SOCK_RAW &&
370 pr->pr_protocol == 0 && maybe == NULL)
377 * The caller must make sure that the new protocol is fully set up and ready to
378 * accept requests before it is registered.
381 pf_proto_register(int family, struct protosw *npr)
383 VNET_ITERATOR_DECL(vnet_iter);
385 struct protosw *pr, *fpr;
389 return (EPFNOSUPPORT);
390 if (npr->pr_type == 0)
392 if (npr->pr_protocol == 0)
393 return (EPROTONOSUPPORT);
394 if (npr->pr_usrreqs == NULL)
397 /* Try to find the specified domain based on the family. */
398 dp = pffinddomain(family);
400 return (EPFNOSUPPORT);
402 /* Initialize backpointer to struct domain. */
407 * Protect us against races when two protocol registrations for
408 * the same protocol happen at the same time.
412 /* The new protocol must not yet exist. */
413 for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++) {
414 if ((pr->pr_type == npr->pr_type) &&
415 (pr->pr_protocol == npr->pr_protocol)) {
416 mtx_unlock(&dom_mtx);
417 return (EEXIST); /* XXX: Check only protocol? */
419 /* While here, remember the first free spacer. */
420 if ((fpr == NULL) && (pr->pr_protocol == PROTO_SPACER))
424 /* If no free spacer is found we can't add the new protocol. */
426 mtx_unlock(&dom_mtx);
430 /* Copy the new struct protosw over the spacer. */
431 bcopy(npr, fpr, sizeof(*fpr));
433 rm_wlock(&pftimo_lock);
434 if (fpr->pr_fasttimo != NULL)
435 LIST_INSERT_HEAD(&pffast_list, fpr, pr_fasttimos);
436 if (fpr->pr_slowtimo != NULL)
437 LIST_INSERT_HEAD(&pfslow_list, fpr, pr_slowtimos);
438 rm_wunlock(&pftimo_lock);
440 /* Job is done, no more protection required. */
441 mtx_unlock(&dom_mtx);
443 /* Initialize and activate the protocol. */
445 VNET_FOREACH(vnet_iter) {
446 CURVNET_SET_QUIET(vnet_iter);
456 * The caller must make sure the protocol and its functions correctly shut down
457 * all sockets and release all locks and memory references.
460 pf_proto_unregister(int family, int protocol, int type)
463 struct protosw *pr, *dpr;
467 return (EPFNOSUPPORT);
469 return (EPROTONOSUPPORT);
473 /* Try to find the specified domain based on the family type. */
474 dp = pffinddomain(family);
476 return (EPFNOSUPPORT);
480 /* Lock out everyone else while we are manipulating the protosw. */
483 /* The protocol must exist and only once. */
484 for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++) {
485 if ((pr->pr_type == type) && (pr->pr_protocol == protocol)) {
487 mtx_unlock(&dom_mtx);
488 return (EMLINK); /* Should not happen! */
494 /* Protocol does not exist. */
496 mtx_unlock(&dom_mtx);
497 return (EPROTONOSUPPORT);
500 rm_wlock(&pftimo_lock);
501 if (dpr->pr_fasttimo != NULL)
502 LIST_REMOVE(dpr, pr_fasttimos);
503 if (dpr->pr_slowtimo != NULL)
504 LIST_REMOVE(dpr, pr_slowtimos);
505 rm_wunlock(&pftimo_lock);
507 /* De-orbit the protocol and make the slot available again. */
510 dpr->pr_protocol = PROTO_SPACER;
512 dpr->pr_input = NULL;
513 dpr->pr_output = NULL;
514 dpr->pr_ctlinput = NULL;
515 dpr->pr_ctloutput = NULL;
517 dpr->pr_fasttimo = NULL;
518 dpr->pr_slowtimo = NULL;
519 dpr->pr_drain = NULL;
520 dpr->pr_usrreqs = &nousrreqs;
522 /* Job is done, not more protection required. */
523 mtx_unlock(&dom_mtx);
529 pfctlinput(int cmd, struct sockaddr *sa)
536 for (dp = domains; dp; dp = dp->dom_next)
537 for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++)
539 (*pr->pr_ctlinput)(cmd, sa, (void *)0);
543 pfslowtimo(void *arg)
545 struct rm_priotracker tracker;
546 struct epoch_tracker et;
549 rm_rlock(&pftimo_lock, &tracker);
551 LIST_FOREACH(pr, &pfslow_list, pr_slowtimos) {
552 (*pr->pr_slowtimo)();
555 rm_runlock(&pftimo_lock, &tracker);
556 callout_reset(&pfslow_callout, hz / PR_SLOWHZ, pfslowtimo, NULL);
560 pffasttimo(void *arg)
562 struct rm_priotracker tracker;
563 struct epoch_tracker et;
566 rm_rlock(&pftimo_lock, &tracker);
568 LIST_FOREACH(pr, &pffast_list, pr_fasttimos) {
569 (*pr->pr_fasttimo)();
572 rm_runlock(&pftimo_lock, &tracker);
573 callout_reset(&pffast_callout, hz / PR_FASTHZ, pffasttimo, NULL);