1/*-
2 * Copyright (c) 1982, 1986, 1993
3 *	The Regents of the University of California.  All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 * 1. Redistributions of source code must retain the above copyright
9 *    notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 *    notice, this list of conditions and the following disclaimer in the
12 *    documentation and/or other materials provided with the distribution.
13 * 4. Neither the name of the University nor the names of its contributors
14 *    may be used to endorse or promote products derived from this software
15 *    without specific prior written permission.
16 *
17 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27 * SUCH DAMAGE.
28 *
29 *	@(#)uipc_domain.c	8.2 (Berkeley) 10/18/93
30 */
31
32#include <sys/cdefs.h>
33__FBSDID("$FreeBSD$");
34
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/mbuf.h>
41#include <sys/kernel.h>
42#include <sys/lock.h>
43#include <sys/mutex.h>
44#include <sys/socketvar.h>
45#include <sys/systm.h>
46
47#include <net/vnet.h>
48
49/*
50 * System initialization
51 *
52 * Note: domain initialization takes place on a per domain basis
53 * as a result of traversing a SYSINIT linker set.  Most likely,
54 * each domain would want to call DOMAIN_SET(9) itself, which
55 * would cause the domain to be added just after domaininit()
56 * is called during startup.
57 *
58 * See DOMAIN_SET(9) for details on its use.
59 */
60
61static void domaininit(void *);
62SYSINIT(domain, SI_SUB_PROTO_DOMAININIT, SI_ORDER_ANY, domaininit, NULL);
63
64static void domainfinalize(void *);
65SYSINIT(domainfin, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_FIRST, domainfinalize,
66    NULL);
67
68static struct callout pffast_callout;
69static struct callout pfslow_callout;
70
71static void	pffasttimo(void *);
72static void	pfslowtimo(void *);
73
74struct domain *domains;		/* registered protocol domains */
75int domain_init_status = 0;
76static struct mtx dom_mtx;		/* domain list lock */
77MTX_SYSINIT(domain, &dom_mtx, "domain list", MTX_DEF);
78
79/*
80 * Dummy protocol specific user requests function pointer array.
81 * All functions return EOPNOTSUPP.
82 */
83struct pr_usrreqs nousrreqs = {
84	.pru_accept =		pru_accept_notsupp,
85	.pru_attach =		pru_attach_notsupp,
86	.pru_bind =		pru_bind_notsupp,
87	.pru_connect =		pru_connect_notsupp,
88	.pru_connect2 =		pru_connect2_notsupp,
89	.pru_control =		pru_control_notsupp,
90	.pru_disconnect	=	pru_disconnect_notsupp,
91	.pru_listen =		pru_listen_notsupp,
92	.pru_peeraddr =		pru_peeraddr_notsupp,
93	.pru_rcvd =		pru_rcvd_notsupp,
94	.pru_rcvoob =		pru_rcvoob_notsupp,
95	.pru_send =		pru_send_notsupp,
96	.pru_sense =		pru_sense_null,
97	.pru_shutdown =		pru_shutdown_notsupp,
98	.pru_sockaddr =		pru_sockaddr_notsupp,
99	.pru_sosend =		pru_sosend_notsupp,
100	.pru_soreceive =	pru_soreceive_notsupp,
101	.pru_sopoll =		pru_sopoll_notsupp,
102};
103
104static void
105protosw_init(struct protosw *pr)
106{
107	struct pr_usrreqs *pu;
108
109	pu = pr->pr_usrreqs;
110	KASSERT(pu != NULL, ("protosw_init: %ssw[%d] has no usrreqs!",
111	    pr->pr_domain->dom_name,
112	    (int)(pr - pr->pr_domain->dom_protosw)));
113
114	/*
115	 * Protocol switch methods fall into three categories: mandatory,
116	 * mandatory but protosw_init() provides a default, and optional.
117	 *
118	 * For true protocols (i.e., pru_attach != NULL), KASSERT truly
119	 * mandatory methods with no defaults, and initialize defaults for
120	 * other mandatory methods if the protocol hasn't defined an
121	 * implementation (NULL function pointer).
122	 */
123#if 0
124	if (pu->pru_attach != NULL) {
125		KASSERT(pu->pru_abort != NULL,
126		    ("protosw_init: %ssw[%d] pru_abort NULL",
127		    pr->pr_domain->dom_name,
128		    (int)(pr - pr->pr_domain->dom_protosw)));
129		KASSERT(pu->pru_send != NULL,
130		    ("protosw_init: %ssw[%d] pru_send NULL",
131		    pr->pr_domain->dom_name,
132		    (int)(pr - pr->pr_domain->dom_protosw)));
133	}
134#endif
135
136#define DEFAULT(foo, bar)	if ((foo) == NULL)  (foo) = (bar)
137	DEFAULT(pu->pru_accept, pru_accept_notsupp);
138	DEFAULT(pu->pru_aio_queue, pru_aio_queue_notsupp);
139	DEFAULT(pu->pru_bind, pru_bind_notsupp);
140	DEFAULT(pu->pru_bindat, pru_bindat_notsupp);
141	DEFAULT(pu->pru_connect, pru_connect_notsupp);
142	DEFAULT(pu->pru_connect2, pru_connect2_notsupp);
143	DEFAULT(pu->pru_connectat, pru_connectat_notsupp);
144	DEFAULT(pu->pru_control, pru_control_notsupp);
145	DEFAULT(pu->pru_disconnect, pru_disconnect_notsupp);
146	DEFAULT(pu->pru_listen, pru_listen_notsupp);
147	DEFAULT(pu->pru_peeraddr, pru_peeraddr_notsupp);
148	DEFAULT(pu->pru_rcvd, pru_rcvd_notsupp);
149	DEFAULT(pu->pru_rcvoob, pru_rcvoob_notsupp);
150	DEFAULT(pu->pru_sense, pru_sense_null);
151	DEFAULT(pu->pru_shutdown, pru_shutdown_notsupp);
152	DEFAULT(pu->pru_sockaddr, pru_sockaddr_notsupp);
153	DEFAULT(pu->pru_sosend, sosend_generic);
154	DEFAULT(pu->pru_soreceive, soreceive_generic);
155	DEFAULT(pu->pru_sopoll, sopoll_generic);
156	DEFAULT(pu->pru_ready, pru_ready_notsupp);
157#undef DEFAULT
158	if (pr->pr_init)
159		(*pr->pr_init)();
160}
161
162/*
163 * Add a new protocol domain to the list of supported domains
164 * Note: you cant unload it again because a socket may be using it.
165 * XXX can't fail at this time.
166 */
167void
168domain_init(void *arg)
169{
170	struct domain *dp = arg;
171	struct protosw *pr;
172
173	if (dp->dom_init)
174		(*dp->dom_init)();
175	for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++)
176		protosw_init(pr);
177	/*
178	 * update global information about maximums
179	 */
180	max_hdr = max_linkhdr + max_protohdr;
181	max_datalen = MHLEN - max_hdr;
182	if (max_datalen < 1)
183		panic("%s: max_datalen < 1", __func__);
184}
185
186#ifdef VIMAGE
187void
188vnet_domain_init(void *arg)
189{
190
191	/* Virtualized case is no different -- call init functions. */
192	domain_init(arg);
193}
194
195void
196vnet_domain_uninit(void *arg)
197{
198	struct domain *dp = arg;
199
200	if (dp->dom_destroy)
201		(*dp->dom_destroy)();
202}
203#endif
204
205/*
206 * Add a new protocol domain to the list of supported domains
207 * Note: you cant unload it again because a socket may be using it.
208 * XXX can't fail at this time.
209 */
210void
211domain_add(void *data)
212{
213	struct domain *dp;
214
215	dp = (struct domain *)data;
216	mtx_lock(&dom_mtx);
217	dp->dom_next = domains;
218	domains = dp;
219
220	KASSERT(domain_init_status >= 1,
221	    ("attempt to domain_add(%s) before domaininit()",
222	    dp->dom_name));
223#ifndef INVARIANTS
224	if (domain_init_status < 1)
225		printf("WARNING: attempt to domain_add(%s) before "
226		    "domaininit()\n", dp->dom_name);
227#endif
228#ifdef notyet
229	KASSERT(domain_init_status < 2,
230	    ("attempt to domain_add(%s) after domainfinalize()",
231	    dp->dom_name));
232#else
233	if (domain_init_status >= 2)
234		printf("WARNING: attempt to domain_add(%s) after "
235		    "domainfinalize()\n", dp->dom_name);
236#endif
237	mtx_unlock(&dom_mtx);
238}
239
240/* ARGSUSED*/
241static void
242domaininit(void *dummy)
243{
244
245	if (max_linkhdr < 16)		/* XXX */
246		max_linkhdr = 16;
247
248	callout_init(&pffast_callout, 1);
249	callout_init(&pfslow_callout, 1);
250
251	mtx_lock(&dom_mtx);
252	KASSERT(domain_init_status == 0, ("domaininit called too late!"));
253	domain_init_status = 1;
254	mtx_unlock(&dom_mtx);
255}
256
257/* ARGSUSED*/
258static void
259domainfinalize(void *dummy)
260{
261
262	mtx_lock(&dom_mtx);
263	KASSERT(domain_init_status == 1, ("domainfinalize called too late!"));
264	domain_init_status = 2;
265	mtx_unlock(&dom_mtx);
266
267	callout_reset(&pffast_callout, 1, pffasttimo, NULL);
268	callout_reset(&pfslow_callout, 1, pfslowtimo, NULL);
269}
270
271struct domain *
272pffinddomain(int family)
273{
274	struct domain *dp;
275
276	for (dp = domains; dp != NULL; dp = dp->dom_next)
277		if (dp->dom_family == family)
278			return (dp);
279	return (NULL);
280}
281
282struct protosw *
283pffindtype(int family, int type)
284{
285	struct domain *dp;
286	struct protosw *pr;
287
288	dp = pffinddomain(family);
289	if (dp == NULL)
290		return (NULL);
291
292	for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++)
293		if (pr->pr_type && pr->pr_type == type)
294			return (pr);
295	return (NULL);
296}
297
298struct protosw *
299pffindproto(int family, int protocol, int type)
300{
301	struct domain *dp;
302	struct protosw *pr;
303	struct protosw *maybe;
304
305	maybe = NULL;
306	if (family == 0)
307		return (NULL);
308
309	dp = pffinddomain(family);
310	if (dp == NULL)
311		return (NULL);
312
313	for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++) {
314		if ((pr->pr_protocol == protocol) && (pr->pr_type == type))
315			return (pr);
316
317		if (type == SOCK_RAW && pr->pr_type == SOCK_RAW &&
318		    pr->pr_protocol == 0 && maybe == NULL)
319			maybe = pr;
320	}
321	return (maybe);
322}
323
324/*
325 * The caller must make sure that the new protocol is fully set up and ready to
326 * accept requests before it is registered.
327 */
328int
329pf_proto_register(int family, struct protosw *npr)
330{
331	VNET_ITERATOR_DECL(vnet_iter);
332	struct domain *dp;
333	struct protosw *pr, *fpr;
334
335	/* Sanity checks. */
336	if (family == 0)
337		return (EPFNOSUPPORT);
338	if (npr->pr_type == 0)
339		return (EPROTOTYPE);
340	if (npr->pr_protocol == 0)
341		return (EPROTONOSUPPORT);
342	if (npr->pr_usrreqs == NULL)
343		return (ENXIO);
344
345	/* Try to find the specified domain based on the family. */
346	dp = pffinddomain(family);
347	if (dp == NULL)
348		return (EPFNOSUPPORT);
349
350	/* Initialize backpointer to struct domain. */
351	npr->pr_domain = dp;
352	fpr = NULL;
353
354	/*
355	 * Protect us against races when two protocol registrations for
356	 * the same protocol happen at the same time.
357	 */
358	mtx_lock(&dom_mtx);
359
360	/* The new protocol must not yet exist. */
361	for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++) {
362		if ((pr->pr_type == npr->pr_type) &&
363		    (pr->pr_protocol == npr->pr_protocol)) {
364			mtx_unlock(&dom_mtx);
365			return (EEXIST);	/* XXX: Check only protocol? */
366		}
367		/* While here, remember the first free spacer. */
368		if ((fpr == NULL) && (pr->pr_protocol == PROTO_SPACER))
369			fpr = pr;
370	}
371
372	/* If no free spacer is found we can't add the new protocol. */
373	if (fpr == NULL) {
374		mtx_unlock(&dom_mtx);
375		return (ENOMEM);
376	}
377
378	/* Copy the new struct protosw over the spacer. */
379	bcopy(npr, fpr, sizeof(*fpr));
380
381	/* Job is done, no more protection required. */
382	mtx_unlock(&dom_mtx);
383
384	/* Initialize and activate the protocol. */
385	VNET_LIST_RLOCK();
386	VNET_FOREACH(vnet_iter) {
387		CURVNET_SET_QUIET(vnet_iter);
388		protosw_init(fpr);
389		CURVNET_RESTORE();
390	}
391	VNET_LIST_RUNLOCK();
392
393	return (0);
394}
395
396/*
397 * The caller must make sure the protocol and its functions correctly shut down
398 * all sockets and release all locks and memory references.
399 */
400int
401pf_proto_unregister(int family, int protocol, int type)
402{
403	struct domain *dp;
404	struct protosw *pr, *dpr;
405
406	/* Sanity checks. */
407	if (family == 0)
408		return (EPFNOSUPPORT);
409	if (protocol == 0)
410		return (EPROTONOSUPPORT);
411	if (type == 0)
412		return (EPROTOTYPE);
413
414	/* Try to find the specified domain based on the family type. */
415	dp = pffinddomain(family);
416	if (dp == NULL)
417		return (EPFNOSUPPORT);
418
419	dpr = NULL;
420
421	/* Lock out everyone else while we are manipulating the protosw. */
422	mtx_lock(&dom_mtx);
423
424	/* The protocol must exist and only once. */
425	for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++) {
426		if ((pr->pr_type == type) && (pr->pr_protocol == protocol)) {
427			if (dpr != NULL) {
428				mtx_unlock(&dom_mtx);
429				return (EMLINK);   /* Should not happen! */
430			} else
431				dpr = pr;
432		}
433	}
434
435	/* Protocol does not exist. */
436	if (dpr == NULL) {
437		mtx_unlock(&dom_mtx);
438		return (EPROTONOSUPPORT);
439	}
440
441	/* De-orbit the protocol and make the slot available again. */
442	dpr->pr_type = 0;
443	dpr->pr_domain = dp;
444	dpr->pr_protocol = PROTO_SPACER;
445	dpr->pr_flags = 0;
446	dpr->pr_input = NULL;
447	dpr->pr_output = NULL;
448	dpr->pr_ctlinput = NULL;
449	dpr->pr_ctloutput = NULL;
450	dpr->pr_init = NULL;
451	dpr->pr_fasttimo = NULL;
452	dpr->pr_slowtimo = NULL;
453	dpr->pr_drain = NULL;
454	dpr->pr_usrreqs = &nousrreqs;
455
456	/* Job is done, not more protection required. */
457	mtx_unlock(&dom_mtx);
458
459	return (0);
460}
461
462void
463pfctlinput(int cmd, struct sockaddr *sa)
464{
465	struct domain *dp;
466	struct protosw *pr;
467
468	for (dp = domains; dp; dp = dp->dom_next)
469		for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++)
470			if (pr->pr_ctlinput)
471				(*pr->pr_ctlinput)(cmd, sa, (void *)0);
472}
473
474void
475pfctlinput2(int cmd, struct sockaddr *sa, void *ctlparam)
476{
477	struct domain *dp;
478	struct protosw *pr;
479
480	if (!sa)
481		return;
482	for (dp = domains; dp; dp = dp->dom_next) {
483		/*
484		 * the check must be made by xx_ctlinput() anyways, to
485		 * make sure we use data item pointed to by ctlparam in
486		 * correct way.  the following check is made just for safety.
487		 */
488		if (dp->dom_family != sa->sa_family)
489			continue;
490
491		for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++)
492			if (pr->pr_ctlinput)
493				(*pr->pr_ctlinput)(cmd, sa, ctlparam);
494	}
495}
496
497static void
498pfslowtimo(void *arg)
499{
500	struct domain *dp;
501	struct protosw *pr;
502
503	for (dp = domains; dp; dp = dp->dom_next)
504		for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++)
505			if (pr->pr_slowtimo)
506				(*pr->pr_slowtimo)();
507	callout_reset(&pfslow_callout, hz/2, pfslowtimo, NULL);
508}
509
510static void
511pffasttimo(void *arg)
512{
513	struct domain *dp;
514	struct protosw *pr;
515
516	for (dp = domains; dp; dp = dp->dom_next)
517		for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++)
518			if (pr->pr_fasttimo)
519				(*pr->pr_fasttimo)();
520	callout_reset(&pffast_callout, hz/5, pffasttimo, NULL);
521}
522