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 __FBSDID("$FreeBSD$");
37 #include <sys/param.h>
38 #include <sys/socket.h>
39 #include <sys/protosw.h>
40 #include <sys/domain.h>
41 #include <sys/eventhandler.h>
42 #include <sys/epoch.h>
44 #include <sys/kernel.h>
46 #include <sys/mutex.h>
47 #include <sys/rmlock.h>
48 #include <sys/socketvar.h>
49 #include <sys/systm.h>
51 #include <machine/atomic.h>
56 * System initialization
58 * Note: domain initialization takes place on a per domain basis
59 * as a result of traversing a SYSINIT linker set. Most likely,
60 * each domain would want to call DOMAIN_SET(9) itself, which
61 * would cause the domain to be added just after domaininit()
62 * is called during startup.
64 * See DOMAIN_SET(9) for details on its use.
67 static void domaininit(void *);
68 SYSINIT(domain, SI_SUB_PROTO_DOMAININIT, SI_ORDER_ANY, domaininit, NULL);
70 static void domainfinalize(void *);
71 SYSINIT(domainfin, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_FIRST, domainfinalize,
74 static struct callout pffast_callout;
75 static struct callout pfslow_callout;
77 static void pffasttimo(void *);
78 static void pfslowtimo(void *);
80 static struct rmlock pftimo_lock;
81 RM_SYSINIT(pftimo_lock, &pftimo_lock, "pftimo");
83 static LIST_HEAD(, protosw) pffast_list =
84 LIST_HEAD_INITIALIZER(pffast_list);
85 static LIST_HEAD(, protosw) pfslow_list =
86 LIST_HEAD_INITIALIZER(pfslow_list);
88 struct domain *domains; /* registered protocol domains */
89 int domain_init_status = 0;
90 static struct mtx dom_mtx; /* domain list lock */
91 MTX_SYSINIT(domain, &dom_mtx, "domain list", MTX_DEF);
94 * Dummy protocol specific user requests function pointer array.
95 * All functions return EOPNOTSUPP.
97 struct pr_usrreqs nousrreqs = {
98 .pru_accept = pru_accept_notsupp,
99 .pru_attach = pru_attach_notsupp,
100 .pru_bind = pru_bind_notsupp,
101 .pru_connect = pru_connect_notsupp,
102 .pru_connect2 = pru_connect2_notsupp,
103 .pru_control = pru_control_notsupp,
104 .pru_disconnect = pru_disconnect_notsupp,
105 .pru_listen = pru_listen_notsupp,
106 .pru_peeraddr = pru_peeraddr_notsupp,
107 .pru_rcvd = pru_rcvd_notsupp,
108 .pru_rcvoob = pru_rcvoob_notsupp,
109 .pru_send = pru_send_notsupp,
110 .pru_sense = pru_sense_null,
111 .pru_shutdown = pru_shutdown_notsupp,
112 .pru_sockaddr = pru_sockaddr_notsupp,
113 .pru_sosend = pru_sosend_notsupp,
114 .pru_soreceive = pru_soreceive_notsupp,
115 .pru_sopoll = pru_sopoll_notsupp,
119 protosw_init(struct protosw *pr)
121 struct pr_usrreqs *pu;
124 KASSERT(pu != NULL, ("protosw_init: %ssw[%d] has no usrreqs!",
125 pr->pr_domain->dom_name,
126 (int)(pr - pr->pr_domain->dom_protosw)));
129 * Protocol switch methods fall into three categories: mandatory,
130 * mandatory but protosw_init() provides a default, and optional.
132 * For true protocols (i.e., pru_attach != NULL), KASSERT truly
133 * mandatory methods with no defaults, and initialize defaults for
134 * other mandatory methods if the protocol hasn't defined an
135 * implementation (NULL function pointer).
138 if (pu->pru_attach != NULL) {
139 KASSERT(pu->pru_abort != NULL,
140 ("protosw_init: %ssw[%d] pru_abort NULL",
141 pr->pr_domain->dom_name,
142 (int)(pr - pr->pr_domain->dom_protosw)));
143 KASSERT(pu->pru_send != NULL,
144 ("protosw_init: %ssw[%d] pru_send NULL",
145 pr->pr_domain->dom_name,
146 (int)(pr - pr->pr_domain->dom_protosw)));
150 #define DEFAULT(foo, bar) if ((foo) == NULL) (foo) = (bar)
151 DEFAULT(pu->pru_accept, pru_accept_notsupp);
152 DEFAULT(pu->pru_aio_queue, pru_aio_queue_notsupp);
153 DEFAULT(pu->pru_bind, pru_bind_notsupp);
154 DEFAULT(pu->pru_bindat, pru_bindat_notsupp);
155 DEFAULT(pu->pru_connect, pru_connect_notsupp);
156 DEFAULT(pu->pru_connect2, pru_connect2_notsupp);
157 DEFAULT(pu->pru_connectat, pru_connectat_notsupp);
158 DEFAULT(pu->pru_control, pru_control_notsupp);
159 DEFAULT(pu->pru_disconnect, pru_disconnect_notsupp);
160 DEFAULT(pu->pru_listen, pru_listen_notsupp);
161 DEFAULT(pu->pru_peeraddr, pru_peeraddr_notsupp);
162 DEFAULT(pu->pru_rcvd, pru_rcvd_notsupp);
163 DEFAULT(pu->pru_rcvoob, pru_rcvoob_notsupp);
164 DEFAULT(pu->pru_sense, pru_sense_null);
165 DEFAULT(pu->pru_shutdown, pru_shutdown_notsupp);
166 DEFAULT(pu->pru_sockaddr, pru_sockaddr_notsupp);
167 DEFAULT(pu->pru_sosend, sosend_generic);
168 DEFAULT(pu->pru_soreceive, soreceive_generic);
169 DEFAULT(pu->pru_sopoll, sopoll_generic);
170 DEFAULT(pu->pru_ready, pru_ready_notsupp);
177 * Add a new protocol domain to the list of supported domains
178 * Note: you cant unload it again because a socket may be using it.
179 * XXX can't fail at this time.
182 domain_init(void *arg)
184 struct domain *dp = arg;
188 flags = atomic_load_acq_int(&dp->dom_flags);
189 if ((flags & DOMF_SUPPORTED) == 0)
191 KASSERT((flags & DOMF_INITED) == 0 || !IS_DEFAULT_VNET(curvnet),
192 ("Premature initialization of domain in non-default vnet"));
195 for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++) {
199 * Note that with VIMAGE enabled, domain_init() will be
200 * re-invoked for each new vnet that's created. The below lists
201 * are intended to be system-wide, so avoid altering global
202 * state for non-default vnets.
204 if (IS_DEFAULT_VNET(curvnet)) {
205 rm_wlock(&pftimo_lock);
206 if (pr->pr_fasttimo != NULL)
207 LIST_INSERT_HEAD(&pffast_list, pr,
209 if (pr->pr_slowtimo != NULL)
210 LIST_INSERT_HEAD(&pfslow_list, pr,
212 rm_wunlock(&pftimo_lock);
217 * update global information about maximums
219 max_hdr = max_linkhdr + max_protohdr;
220 max_datalen = MHLEN - max_hdr;
222 panic("%s: max_datalen < 1", __func__);
223 if (IS_DEFAULT_VNET(curvnet))
224 atomic_set_rel_int(&dp->dom_flags, DOMF_INITED);
229 vnet_domain_init(void *arg)
232 /* Virtualized case is no different -- call init functions. */
237 vnet_domain_uninit(void *arg)
239 struct domain *dp = arg;
241 if ((atomic_load_acq_int(&dp->dom_flags) & DOMF_SUPPORTED) == 0)
244 (*dp->dom_destroy)();
249 * Add a new protocol domain to the list of supported domains
250 * Note: you cant unload it again because a socket may be using it.
251 * XXX can't fail at this time.
254 domain_add(void *data)
258 dp = (struct domain *)data;
259 if (dp->dom_probe != NULL && (*dp->dom_probe)() != 0)
261 atomic_set_rel_int(&dp->dom_flags, DOMF_SUPPORTED);
263 dp->dom_next = domains;
266 KASSERT(domain_init_status >= 1,
267 ("attempt to domain_add(%s) before domaininit()",
270 if (domain_init_status < 1)
271 printf("WARNING: attempt to domain_add(%s) before "
272 "domaininit()\n", dp->dom_name);
274 mtx_unlock(&dom_mtx);
279 domaininit(void *dummy)
282 if (max_linkhdr < 16) /* XXX */
285 callout_init(&pffast_callout, 1);
286 callout_init(&pfslow_callout, 1);
289 KASSERT(domain_init_status == 0, ("domaininit called too late!"));
290 domain_init_status = 1;
291 mtx_unlock(&dom_mtx);
296 domainfinalize(void *dummy)
300 KASSERT(domain_init_status == 1, ("domainfinalize called too late!"));
301 domain_init_status = 2;
302 mtx_unlock(&dom_mtx);
304 callout_reset(&pffast_callout, 1, pffasttimo, NULL);
305 callout_reset(&pfslow_callout, 1, pfslowtimo, NULL);
309 pffinddomain(int family)
313 for (dp = domains; dp != NULL; dp = dp->dom_next)
314 if (dp->dom_family == family)
320 pffindtype(int family, int type)
325 dp = pffinddomain(family);
329 for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++)
330 if (pr->pr_type && pr->pr_type == type)
336 pffindproto(int family, int protocol, int type)
340 struct protosw *maybe;
346 dp = pffinddomain(family);
350 for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++) {
351 if ((pr->pr_protocol == protocol) && (pr->pr_type == type))
354 if (type == SOCK_RAW && pr->pr_type == SOCK_RAW &&
355 pr->pr_protocol == 0 && maybe == NULL)
362 * The caller must make sure that the new protocol is fully set up and ready to
363 * accept requests before it is registered.
366 pf_proto_register(int family, struct protosw *npr)
368 VNET_ITERATOR_DECL(vnet_iter);
370 struct protosw *pr, *fpr;
374 return (EPFNOSUPPORT);
375 if (npr->pr_type == 0)
377 if (npr->pr_protocol == 0)
378 return (EPROTONOSUPPORT);
379 if (npr->pr_usrreqs == NULL)
382 /* Try to find the specified domain based on the family. */
383 dp = pffinddomain(family);
385 return (EPFNOSUPPORT);
387 /* Initialize backpointer to struct domain. */
392 * Protect us against races when two protocol registrations for
393 * the same protocol happen at the same time.
397 /* The new protocol must not yet exist. */
398 for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++) {
399 if ((pr->pr_type == npr->pr_type) &&
400 (pr->pr_protocol == npr->pr_protocol)) {
401 mtx_unlock(&dom_mtx);
402 return (EEXIST); /* XXX: Check only protocol? */
404 /* While here, remember the first free spacer. */
405 if ((fpr == NULL) && (pr->pr_protocol == PROTO_SPACER))
409 /* If no free spacer is found we can't add the new protocol. */
411 mtx_unlock(&dom_mtx);
415 /* Copy the new struct protosw over the spacer. */
416 bcopy(npr, fpr, sizeof(*fpr));
418 rm_wlock(&pftimo_lock);
419 if (fpr->pr_fasttimo != NULL)
420 LIST_INSERT_HEAD(&pffast_list, fpr, pr_fasttimos);
421 if (fpr->pr_slowtimo != NULL)
422 LIST_INSERT_HEAD(&pfslow_list, fpr, pr_slowtimos);
423 rm_wunlock(&pftimo_lock);
425 /* Job is done, no more protection required. */
426 mtx_unlock(&dom_mtx);
428 /* Initialize and activate the protocol. */
430 VNET_FOREACH(vnet_iter) {
431 CURVNET_SET_QUIET(vnet_iter);
441 * The caller must make sure the protocol and its functions correctly shut down
442 * all sockets and release all locks and memory references.
445 pf_proto_unregister(int family, int protocol, int type)
448 struct protosw *pr, *dpr;
452 return (EPFNOSUPPORT);
454 return (EPROTONOSUPPORT);
458 /* Try to find the specified domain based on the family type. */
459 dp = pffinddomain(family);
461 return (EPFNOSUPPORT);
465 /* Lock out everyone else while we are manipulating the protosw. */
468 /* The protocol must exist and only once. */
469 for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++) {
470 if ((pr->pr_type == type) && (pr->pr_protocol == protocol)) {
472 mtx_unlock(&dom_mtx);
473 return (EMLINK); /* Should not happen! */
479 /* Protocol does not exist. */
481 mtx_unlock(&dom_mtx);
482 return (EPROTONOSUPPORT);
485 rm_wlock(&pftimo_lock);
486 if (dpr->pr_fasttimo != NULL)
487 LIST_REMOVE(dpr, pr_fasttimos);
488 if (dpr->pr_slowtimo != NULL)
489 LIST_REMOVE(dpr, pr_slowtimos);
490 rm_wunlock(&pftimo_lock);
492 /* De-orbit the protocol and make the slot available again. */
495 dpr->pr_protocol = PROTO_SPACER;
497 dpr->pr_input = NULL;
498 dpr->pr_output = NULL;
499 dpr->pr_ctlinput = NULL;
500 dpr->pr_ctloutput = NULL;
502 dpr->pr_fasttimo = NULL;
503 dpr->pr_slowtimo = NULL;
504 dpr->pr_drain = NULL;
505 dpr->pr_usrreqs = &nousrreqs;
507 /* Job is done, not more protection required. */
508 mtx_unlock(&dom_mtx);
514 pfctlinput(int cmd, struct sockaddr *sa)
519 for (dp = domains; dp; dp = dp->dom_next)
520 for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++)
522 (*pr->pr_ctlinput)(cmd, sa, (void *)0);
526 pfslowtimo(void *arg)
528 struct rm_priotracker tracker;
529 struct epoch_tracker et;
532 rm_rlock(&pftimo_lock, &tracker);
534 LIST_FOREACH(pr, &pfslow_list, pr_slowtimos) {
535 (*pr->pr_slowtimo)();
538 rm_runlock(&pftimo_lock, &tracker);
539 callout_reset(&pfslow_callout, hz/2, pfslowtimo, NULL);
543 pffasttimo(void *arg)
545 struct rm_priotracker tracker;
546 struct epoch_tracker et;
549 rm_rlock(&pftimo_lock, &tracker);
551 LIST_FOREACH(pr, &pffast_list, pr_fasttimos) {
552 (*pr->pr_fasttimo)();
555 rm_runlock(&pftimo_lock, &tracker);
556 callout_reset(&pffast_callout, hz/5, pffasttimo, NULL);