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 61 static void domaininit(void *); 62 SYSINIT(domain, SI_SUB_PROTO_DOMAININIT, SI_ORDER_ANY, domaininit, NULL); 63 64 static void domainfinalize(void *); 65 SYSINIT(domainfin, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_FIRST, domainfinalize, 66 NULL); 67 68 static struct callout pffast_callout; 69 static struct callout pfslow_callout; 70 71 static void pffasttimo(void *); 72 static void pfslowtimo(void *); 73 74 struct domain *domains; /* registered protocol domains */ 75 int domain_init_status = 0; 76 static struct mtx dom_mtx; /* domain list lock */ 77 MTX_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 */ 83 struct 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 }; 103 104 static void 105 protosw_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 */ 167 void 168 domain_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 187 void 188 vnet_domain_init(void *arg) 189 { 190 191 /* Virtualized case is no different -- call init functions. */ 192 domain_init(arg); 193 } 194 195 void 196 vnet_domain_uninit(void *arg) 197 { 198 struct domain *dp = arg; 199 struct protosw *pr; 200 201 for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++) 202 if (pr->pr_destroy) 203 (*pr->pr_destroy)(); 204 if (dp->dom_destroy) 205 (*dp->dom_destroy)(); 206 } 207 #endif 208 209 /* 210 * Add a new protocol domain to the list of supported domains 211 * Note: you cant unload it again because a socket may be using it. 212 * XXX can't fail at this time. 213 */ 214 void 215 domain_add(void *data) 216 { 217 struct domain *dp; 218 219 dp = (struct domain *)data; 220 mtx_lock(&dom_mtx); 221 dp->dom_next = domains; 222 domains = dp; 223 224 KASSERT(domain_init_status >= 1, 225 ("attempt to domain_add(%s) before domaininit()", 226 dp->dom_name)); 227 #ifndef INVARIANTS 228 if (domain_init_status < 1) 229 printf("WARNING: attempt to domain_add(%s) before " 230 "domaininit()\n", dp->dom_name); 231 #endif 232 #ifdef notyet 233 KASSERT(domain_init_status < 2, 234 ("attempt to domain_add(%s) after domainfinalize()", 235 dp->dom_name)); 236 #else 237 if (domain_init_status >= 2) 238 printf("WARNING: attempt to domain_add(%s) after " 239 "domainfinalize()\n", dp->dom_name); 240 #endif 241 mtx_unlock(&dom_mtx); 242 } 243 244 /* ARGSUSED*/ 245 static void 246 domaininit(void *dummy) 247 { 248 249 if (max_linkhdr < 16) /* XXX */ 250 max_linkhdr = 16; 251 252 callout_init(&pffast_callout, 1); 253 callout_init(&pfslow_callout, 1); 254 255 mtx_lock(&dom_mtx); 256 KASSERT(domain_init_status == 0, ("domaininit called too late!")); 257 domain_init_status = 1; 258 mtx_unlock(&dom_mtx); 259 } 260 261 /* ARGSUSED*/ 262 static void 263 domainfinalize(void *dummy) 264 { 265 266 mtx_lock(&dom_mtx); 267 KASSERT(domain_init_status == 1, ("domainfinalize called too late!")); 268 domain_init_status = 2; 269 mtx_unlock(&dom_mtx); 270 271 callout_reset(&pffast_callout, 1, pffasttimo, NULL); 272 callout_reset(&pfslow_callout, 1, pfslowtimo, NULL); 273 } 274 275 struct domain * 276 pffinddomain(int family) 277 { 278 struct domain *dp; 279 280 for (dp = domains; dp != NULL; dp = dp->dom_next) 281 if (dp->dom_family == family) 282 return (dp); 283 return (NULL); 284 } 285 286 struct protosw * 287 pffindtype(int family, int type) 288 { 289 struct domain *dp; 290 struct protosw *pr; 291 292 dp = pffinddomain(family); 293 if (dp == NULL) 294 return (NULL); 295 296 for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++) 297 if (pr->pr_type && pr->pr_type == type) 298 return (pr); 299 return (NULL); 300 } 301 302 struct protosw * 303 pffindproto(int family, int protocol, int type) 304 { 305 struct domain *dp; 306 struct protosw *pr; 307 struct protosw *maybe; 308 309 maybe = NULL; 310 if (family == 0) 311 return (NULL); 312 313 dp = pffinddomain(family); 314 if (dp == NULL) 315 return (NULL); 316 317 for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++) { 318 if ((pr->pr_protocol == protocol) && (pr->pr_type == type)) 319 return (pr); 320 321 if (type == SOCK_RAW && pr->pr_type == SOCK_RAW && 322 pr->pr_protocol == 0 && maybe == NULL) 323 maybe = pr; 324 } 325 return (maybe); 326 } 327 328 /* 329 * The caller must make sure that the new protocol is fully set up and ready to 330 * accept requests before it is registered. 331 */ 332 int 333 pf_proto_register(int family, struct protosw *npr) 334 { 335 VNET_ITERATOR_DECL(vnet_iter); 336 struct domain *dp; 337 struct protosw *pr, *fpr; 338 339 /* Sanity checks. */ 340 if (family == 0) 341 return (EPFNOSUPPORT); 342 if (npr->pr_type == 0) 343 return (EPROTOTYPE); 344 if (npr->pr_protocol == 0) 345 return (EPROTONOSUPPORT); 346 if (npr->pr_usrreqs == NULL) 347 return (ENXIO); 348 349 /* Try to find the specified domain based on the family. */ 350 dp = pffinddomain(family); 351 if (dp == NULL) 352 return (EPFNOSUPPORT); 353 354 /* Initialize backpointer to struct domain. */ 355 npr->pr_domain = dp; 356 fpr = NULL; 357 358 /* 359 * Protect us against races when two protocol registrations for 360 * the same protocol happen at the same time. 361 */ 362 mtx_lock(&dom_mtx); 363 364 /* The new protocol must not yet exist. */ 365 for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++) { 366 if ((pr->pr_type == npr->pr_type) && 367 (pr->pr_protocol == npr->pr_protocol)) { 368 mtx_unlock(&dom_mtx); 369 return (EEXIST); /* XXX: Check only protocol? */ 370 } 371 /* While here, remember the first free spacer. */ 372 if ((fpr == NULL) && (pr->pr_protocol == PROTO_SPACER)) 373 fpr = pr; 374 } 375 376 /* If no free spacer is found we can't add the new protocol. */ 377 if (fpr == NULL) { 378 mtx_unlock(&dom_mtx); 379 return (ENOMEM); 380 } 381 382 /* Copy the new struct protosw over the spacer. */ 383 bcopy(npr, fpr, sizeof(*fpr)); 384 385 /* Job is done, no more protection required. */ 386 mtx_unlock(&dom_mtx); 387 388 /* Initialize and activate the protocol. */ 389 VNET_LIST_RLOCK(); 390 VNET_FOREACH(vnet_iter) { 391 CURVNET_SET_QUIET(vnet_iter); 392 protosw_init(fpr); 393 CURVNET_RESTORE(); 394 } 395 VNET_LIST_RUNLOCK(); 396 397 return (0); 398 } 399 400 /* 401 * The caller must make sure the protocol and its functions correctly shut down 402 * all sockets and release all locks and memory references. 403 */ 404 int 405 pf_proto_unregister(int family, int protocol, int type) 406 { 407 struct domain *dp; 408 struct protosw *pr, *dpr; 409 410 /* Sanity checks. */ 411 if (family == 0) 412 return (EPFNOSUPPORT); 413 if (protocol == 0) 414 return (EPROTONOSUPPORT); 415 if (type == 0) 416 return (EPROTOTYPE); 417 418 /* Try to find the specified domain based on the family type. */ 419 dp = pffinddomain(family); 420 if (dp == NULL) 421 return (EPFNOSUPPORT); 422 423 dpr = NULL; 424 425 /* Lock out everyone else while we are manipulating the protosw. */ 426 mtx_lock(&dom_mtx); 427 428 /* The protocol must exist and only once. */ 429 for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++) { 430 if ((pr->pr_type == type) && (pr->pr_protocol == protocol)) { 431 if (dpr != NULL) { 432 mtx_unlock(&dom_mtx); 433 return (EMLINK); /* Should not happen! */ 434 } else 435 dpr = pr; 436 } 437 } 438 439 /* Protocol does not exist. */ 440 if (dpr == NULL) { 441 mtx_unlock(&dom_mtx); 442 return (EPROTONOSUPPORT); 443 } 444 445 /* De-orbit the protocol and make the slot available again. */ 446 dpr->pr_type = 0; 447 dpr->pr_domain = dp; 448 dpr->pr_protocol = PROTO_SPACER; 449 dpr->pr_flags = 0; 450 dpr->pr_input = NULL; 451 dpr->pr_output = NULL; 452 dpr->pr_ctlinput = NULL; 453 dpr->pr_ctloutput = NULL; 454 dpr->pr_init = NULL; 455 dpr->pr_fasttimo = NULL; 456 dpr->pr_slowtimo = NULL; 457 dpr->pr_drain = NULL; 458 dpr->pr_usrreqs = &nousrreqs; 459 460 /* Job is done, not more protection required. */ 461 mtx_unlock(&dom_mtx); 462 463 return (0); 464 } 465 466 void 467 pfctlinput(int cmd, struct sockaddr *sa) 468 { 469 struct domain *dp; 470 struct protosw *pr; 471 472 for (dp = domains; dp; dp = dp->dom_next) 473 for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++) 474 if (pr->pr_ctlinput) 475 (*pr->pr_ctlinput)(cmd, sa, (void *)0); 476 } 477 478 void 479 pfctlinput2(int cmd, struct sockaddr *sa, void *ctlparam) 480 { 481 struct domain *dp; 482 struct protosw *pr; 483 484 if (!sa) 485 return; 486 for (dp = domains; dp; dp = dp->dom_next) { 487 /* 488 * the check must be made by xx_ctlinput() anyways, to 489 * make sure we use data item pointed to by ctlparam in 490 * correct way. the following check is made just for safety. 491 */ 492 if (dp->dom_family != sa->sa_family) 493 continue; 494 495 for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++) 496 if (pr->pr_ctlinput) 497 (*pr->pr_ctlinput)(cmd, sa, ctlparam); 498 } 499 } 500 501 static void 502 pfslowtimo(void *arg) 503 { 504 struct domain *dp; 505 struct protosw *pr; 506 507 for (dp = domains; dp; dp = dp->dom_next) 508 for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++) 509 if (pr->pr_slowtimo) 510 (*pr->pr_slowtimo)(); 511 callout_reset(&pfslow_callout, hz/2, pfslowtimo, NULL); 512 } 513 514 static void 515 pffasttimo(void *arg) 516 { 517 struct domain *dp; 518 struct protosw *pr; 519 520 for (dp = domains; dp; dp = dp->dom_next) 521 for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++) 522 if (pr->pr_fasttimo) 523 (*pr->pr_fasttimo)(); 524 callout_reset(&pffast_callout, hz/5, pffasttimo, NULL); 525 } 526