1 /*-
2 * SPDX-License-Identifier: BSD-3-Clause
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
32 #include <sys/param.h>
33 #include <sys/socket.h>
34 #include <sys/protosw.h>
35 #include <sys/domain.h>
36 #include <sys/eventhandler.h>
37 #include <sys/epoch.h>
38 #include <sys/mbuf.h>
39 #include <sys/kernel.h>
40 #include <sys/lock.h>
41 #include <sys/mutex.h>
42 #include <sys/rmlock.h>
43 #include <sys/socketvar.h>
44 #include <sys/systm.h>
45
46 #include <machine/atomic.h>
47
48 #include <net/vnet.h>
49
50 struct domainhead domains = SLIST_HEAD_INITIALIZER(&domains);
51 int domain_init_status = 1;
52 static struct mtx dom_mtx; /* domain list lock */
53 MTX_SYSINIT(domain, &dom_mtx, "domain list", MTX_DEF);
54
55 static int
pr_accept_notsupp(struct socket * so,struct sockaddr * sa)56 pr_accept_notsupp(struct socket *so, struct sockaddr *sa)
57 {
58 return (EOPNOTSUPP);
59 }
60
61 static int
pr_bind_notsupp(struct socket * so,struct sockaddr * nam,struct thread * td)62 pr_bind_notsupp(struct socket *so, struct sockaddr *nam, struct thread *td)
63 {
64 return (EOPNOTSUPP);
65 }
66
67 static int
pr_bindat_notsupp(int fd,struct socket * so,struct sockaddr * nam,struct thread * td)68 pr_bindat_notsupp(int fd, struct socket *so, struct sockaddr *nam,
69 struct thread *td)
70 {
71 return (EOPNOTSUPP);
72 }
73
74 static int
pr_connect_notsupp(struct socket * so,struct sockaddr * nam,struct thread * td)75 pr_connect_notsupp(struct socket *so, struct sockaddr *nam, struct thread *td)
76 {
77 return (EOPNOTSUPP);
78 }
79
80 static int
pr_connectat_notsupp(int fd,struct socket * so,struct sockaddr * nam,struct thread * td)81 pr_connectat_notsupp(int fd, struct socket *so, struct sockaddr *nam,
82 struct thread *td)
83 {
84 return (EOPNOTSUPP);
85 }
86
87 static int
pr_connect2_notsupp(struct socket * so1,struct socket * so2)88 pr_connect2_notsupp(struct socket *so1, struct socket *so2)
89 {
90 return (EOPNOTSUPP);
91 }
92
93 static int
pr_control_notsupp(struct socket * so,u_long cmd,void * data,struct ifnet * ifp,struct thread * td)94 pr_control_notsupp(struct socket *so, u_long cmd, void *data,
95 struct ifnet *ifp, struct thread *td)
96 {
97 return (EOPNOTSUPP);
98 }
99
100 static int
pr_ctloutput_notsupp(struct socket * so,struct sockopt * sopt)101 pr_ctloutput_notsupp(struct socket *so, struct sockopt *sopt)
102 {
103 return (ENOPROTOOPT);
104 }
105
106 static int
pr_disconnect_notsupp(struct socket * so)107 pr_disconnect_notsupp(struct socket *so)
108 {
109 return (EOPNOTSUPP);
110 }
111
112 int
pr_listen_notsupp(struct socket * so,int backlog,struct thread * td)113 pr_listen_notsupp(struct socket *so, int backlog, struct thread *td)
114 {
115 return (EOPNOTSUPP);
116 }
117
118 static int
pr_peeraddr_notsupp(struct socket * so,struct sockaddr * nam)119 pr_peeraddr_notsupp(struct socket *so, struct sockaddr *nam)
120 {
121 return (EOPNOTSUPP);
122 }
123
124 static int
pr_rcvd_notsupp(struct socket * so,int flags)125 pr_rcvd_notsupp(struct socket *so, int flags)
126 {
127 return (EOPNOTSUPP);
128 }
129
130 static int
pr_rcvoob_notsupp(struct socket * so,struct mbuf * m,int flags)131 pr_rcvoob_notsupp(struct socket *so, struct mbuf *m, int flags)
132 {
133 return (EOPNOTSUPP);
134 }
135
136 static int
pr_send_notsupp(struct socket * so,int flags,struct mbuf * m,struct sockaddr * addr,struct mbuf * control,struct thread * td)137 pr_send_notsupp(struct socket *so, int flags, struct mbuf *m,
138 struct sockaddr *addr, struct mbuf *control, struct thread *td)
139 {
140 if (control != NULL)
141 m_freem(control);
142 if ((flags & PRUS_NOTREADY) == 0)
143 m_freem(m);
144 return (EOPNOTSUPP);
145 }
146
147 static int
pr_ready_notsupp(struct socket * so,struct mbuf * m,int count)148 pr_ready_notsupp(struct socket *so, struct mbuf *m, int count)
149 {
150 return (EOPNOTSUPP);
151 }
152
153 static int
pr_shutdown_notsupp(struct socket * so,enum shutdown_how how)154 pr_shutdown_notsupp(struct socket *so, enum shutdown_how how)
155 {
156 return (EOPNOTSUPP);
157 }
158
159 static int
pr_sockaddr_notsupp(struct socket * so,struct sockaddr * nam)160 pr_sockaddr_notsupp(struct socket *so, struct sockaddr *nam)
161 {
162 return (EOPNOTSUPP);
163 }
164
165 static int
pr_sosend_notsupp(struct socket * so,struct sockaddr * addr,struct uio * uio,struct mbuf * top,struct mbuf * control,int flags,struct thread * td)166 pr_sosend_notsupp(struct socket *so, struct sockaddr *addr, struct uio *uio,
167 struct mbuf *top, struct mbuf *control, int flags, struct thread *td)
168 {
169 return (EOPNOTSUPP);
170 }
171
172 static int
pr_soreceive_notsupp(struct socket * so,struct sockaddr ** paddr,struct uio * uio,struct mbuf ** mp0,struct mbuf ** controlp,int * flagsp)173 pr_soreceive_notsupp(struct socket *so, struct sockaddr **paddr,
174 struct uio *uio, struct mbuf **mp0, struct mbuf **controlp, int *flagsp)
175 {
176 return (EOPNOTSUPP);
177 }
178
179 static void
pr_init(struct domain * dom,struct protosw * pr)180 pr_init(struct domain *dom, struct protosw *pr)
181 {
182
183 KASSERT(pr->pr_attach != NULL,
184 ("%s: protocol doesn't have pr_attach", __func__));
185
186 pr->pr_domain = dom;
187
188 #define DEFAULT(foo, bar) if (pr->foo == NULL) pr->foo = bar
189 DEFAULT(pr_sosend, sosend_generic);
190 DEFAULT(pr_soreceive, soreceive_generic);
191 DEFAULT(pr_sopoll, sopoll_generic);
192 DEFAULT(pr_setsbopt, sbsetopt);
193 DEFAULT(pr_aio_queue, soaio_queue_generic);
194
195 #define NOTSUPP(foo) if (pr->foo == NULL) pr->foo = foo ## _notsupp
196 NOTSUPP(pr_accept);
197 NOTSUPP(pr_bind);
198 NOTSUPP(pr_bindat);
199 NOTSUPP(pr_connect);
200 NOTSUPP(pr_connect2);
201 NOTSUPP(pr_connectat);
202 NOTSUPP(pr_control);
203 NOTSUPP(pr_ctloutput);
204 NOTSUPP(pr_disconnect);
205 NOTSUPP(pr_listen);
206 NOTSUPP(pr_peeraddr);
207 NOTSUPP(pr_rcvd);
208 NOTSUPP(pr_rcvoob);
209 NOTSUPP(pr_send);
210 NOTSUPP(pr_shutdown);
211 NOTSUPP(pr_sockaddr);
212 NOTSUPP(pr_sosend);
213 NOTSUPP(pr_soreceive);
214 NOTSUPP(pr_ready);
215 }
216
217 /*
218 * Add a new protocol domain to the list of supported domains
219 * Note: you can't unload it again because a socket may be using it.
220 * XXX can't fail at this time.
221 */
222 void
domain_add(struct domain * dp)223 domain_add(struct domain *dp)
224 {
225 struct protosw *pr;
226
227 MPASS(IS_DEFAULT_VNET(curvnet));
228
229 if (dp->dom_probe != NULL && (*dp->dom_probe)() != 0)
230 return;
231
232 for (int i = 0; i < dp->dom_nprotosw; i++)
233 if ((pr = dp->dom_protosw[i]) != NULL)
234 pr_init(dp, pr);
235
236 mtx_lock(&dom_mtx);
237 #ifdef INVARIANTS
238 struct domain *tmp;
239 SLIST_FOREACH(tmp, &domains, dom_next)
240 MPASS(tmp->dom_family != dp->dom_family);
241 #endif
242 SLIST_INSERT_HEAD(&domains, dp, dom_next);
243 mtx_unlock(&dom_mtx);
244 }
245
246 void
domain_remove(struct domain * dp)247 domain_remove(struct domain *dp)
248 {
249
250 if ((dp->dom_flags & DOMF_UNLOADABLE) == 0)
251 return;
252
253 mtx_lock(&dom_mtx);
254 SLIST_REMOVE(&domains, dp, domain, dom_next);
255 mtx_unlock(&dom_mtx);
256 }
257
258 static void
domainfinalize(void * dummy)259 domainfinalize(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 SYSINIT(domainfin, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_FIRST, domainfinalize,
268 NULL);
269
270 struct domain *
pffinddomain(int family)271 pffinddomain(int family)
272 {
273 struct domain *dp;
274
275 SLIST_FOREACH(dp, &domains, dom_next)
276 if (dp->dom_family == family)
277 return (dp);
278 return (NULL);
279 }
280
281 struct protosw *
pffindproto(int family,int type,int proto)282 pffindproto(int family, int type, int proto)
283 {
284 struct domain *dp;
285 struct protosw *pr;
286
287 dp = pffinddomain(family);
288 if (dp == NULL)
289 return (NULL);
290
291 for (int i = 0; i < dp->dom_nprotosw; i++)
292 if ((pr = dp->dom_protosw[i]) != NULL && pr->pr_type == type &&
293 (pr->pr_protocol == 0 || proto == 0 ||
294 pr->pr_protocol == proto))
295 return (pr);
296
297 return (NULL);
298 }
299
300 /*
301 * The caller must make sure that the new protocol is fully set up and ready to
302 * accept requests before it is registered.
303 */
304 int
protosw_register(struct domain * dp,struct protosw * npr)305 protosw_register(struct domain *dp, struct protosw *npr)
306 {
307 struct protosw **prp;
308
309 MPASS(dp);
310 MPASS(npr && npr->pr_type > 0 && npr->pr_protocol > 0);
311
312 prp = NULL;
313 /*
314 * Protect us against races when two protocol registrations for
315 * the same protocol happen at the same time.
316 */
317 mtx_lock(&dom_mtx);
318 for (int i = 0; i < dp->dom_nprotosw; i++) {
319 if (dp->dom_protosw[i] == NULL) {
320 /* Remember the first free spacer. */
321 if (prp == NULL)
322 prp = &dp->dom_protosw[i];
323 } else {
324 /*
325 * The new protocol must not yet exist.
326 * XXXAO: Check only protocol?
327 * XXXGL: Maybe assert that it doesn't exist?
328 */
329 if ((dp->dom_protosw[i]->pr_type == npr->pr_type) &&
330 (dp->dom_protosw[i]->pr_protocol ==
331 npr->pr_protocol)) {
332 mtx_unlock(&dom_mtx);
333 return (EEXIST);
334 }
335
336 }
337 }
338
339 /* If no free spacer is found we can't add the new protocol. */
340 if (prp == NULL) {
341 mtx_unlock(&dom_mtx);
342 return (ENOMEM);
343 }
344
345 pr_init(dp, npr);
346 *prp = npr;
347 mtx_unlock(&dom_mtx);
348
349 return (0);
350 }
351
352 /*
353 * The caller must make sure the protocol and its functions correctly shut down
354 * all sockets and release all locks and memory references.
355 */
356 int
protosw_unregister(struct protosw * pr)357 protosw_unregister(struct protosw *pr)
358 {
359 struct domain *dp;
360 struct protosw **prp;
361
362 dp = pr->pr_domain;
363 prp = NULL;
364
365 mtx_lock(&dom_mtx);
366 /* The protocol must exist and only once. */
367 for (int i = 0; i < dp->dom_nprotosw; i++) {
368 if (dp->dom_protosw[i] == pr) {
369 KASSERT(prp == NULL,
370 ("%s: domain %p protocol %p registered twice\n",
371 __func__, dp, pr));
372 prp = &dp->dom_protosw[i];
373 }
374 }
375
376 /* Protocol does not exist. XXXGL: assert that it does? */
377 if (prp == NULL) {
378 mtx_unlock(&dom_mtx);
379 return (EPROTONOSUPPORT);
380 }
381
382 /* De-orbit the protocol and make the slot available again. */
383 *prp = NULL;
384 mtx_unlock(&dom_mtx);
385
386 return (0);
387 }
388