1 /*- 2 * Copyright (c) 1982, 1986, 1990, 1993 3 * The Regents of the University of California. 4 * Copyright (c) 2010-2011 Juniper Networks, Inc. 5 * All rights reserved. 6 * 7 * Portions of this software were developed by Robert N. M. Watson under 8 * contract to Juniper Networks, Inc. 9 * 10 * Redistribution and use in source and binary forms, with or without 11 * modification, are permitted provided that the following conditions 12 * are met: 13 * 1. Redistributions of source code must retain the above copyright 14 * notice, this list of conditions and the following disclaimer. 15 * 2. Redistributions in binary form must reproduce the above copyright 16 * notice, this list of conditions and the following disclaimer in the 17 * documentation and/or other materials provided with the distribution. 18 * 4. Neither the name of the University nor the names of its contributors 19 * may be used to endorse or promote products derived from this software 20 * without specific prior written permission. 21 * 22 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 23 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 24 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 25 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 26 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 27 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 28 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 29 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 30 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 31 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 32 * SUCH DAMAGE. 33 * 34 * @(#)in_pcb.h 8.1 (Berkeley) 6/10/93 35 * $FreeBSD$ 36 */ 37 38 #ifndef _NETINET_IN_PCB_H_ 39 #define _NETINET_IN_PCB_H_ 40 41 #include <sys/queue.h> 42 #include <sys/_lock.h> 43 #include <sys/_mutex.h> 44 #include <sys/_rwlock.h> 45 #include <net/route.h> 46 47 #ifdef _KERNEL 48 #include <sys/lock.h> 49 #include <sys/rwlock.h> 50 #include <net/vnet.h> 51 #include <vm/uma.h> 52 #endif 53 54 #define in6pcb inpcb /* for KAME src sync over BSD*'s */ 55 #define in6p_sp inp_sp /* for KAME src sync over BSD*'s */ 56 struct inpcbpolicy; 57 58 /* 59 * struct inpcb is the common protocol control block structure used in most 60 * IP transport protocols. 61 * 62 * Pointers to local and foreign host table entries, local and foreign socket 63 * numbers, and pointers up (to a socket structure) and down (to a 64 * protocol-specific control block) are stored here. 65 */ 66 LIST_HEAD(inpcbhead, inpcb); 67 LIST_HEAD(inpcbporthead, inpcbport); 68 typedef u_quad_t inp_gen_t; 69 70 /* 71 * PCB with AF_INET6 null bind'ed laddr can receive AF_INET input packet. 72 * So, AF_INET6 null laddr is also used as AF_INET null laddr, by utilizing 73 * the following structure. 74 */ 75 struct in_addr_4in6 { 76 u_int32_t ia46_pad32[3]; 77 struct in_addr ia46_addr4; 78 }; 79 80 /* 81 * NOTE: ipv6 addrs should be 64-bit aligned, per RFC 2553. in_conninfo has 82 * some extra padding to accomplish this. 83 * NOTE 2: tcp_syncache.c uses first 5 32-bit words, which identify fport, 84 * lport, faddr to generate hash, so these fields shouldn't be moved. 85 */ 86 struct in_endpoints { 87 u_int16_t ie_fport; /* foreign port */ 88 u_int16_t ie_lport; /* local port */ 89 /* protocol dependent part, local and foreign addr */ 90 union { 91 /* foreign host table entry */ 92 struct in_addr_4in6 ie46_foreign; 93 struct in6_addr ie6_foreign; 94 } ie_dependfaddr; 95 union { 96 /* local host table entry */ 97 struct in_addr_4in6 ie46_local; 98 struct in6_addr ie6_local; 99 } ie_dependladdr; 100 u_int32_t ie6_zoneid; /* scope zone id */ 101 }; 102 #define ie_faddr ie_dependfaddr.ie46_foreign.ia46_addr4 103 #define ie_laddr ie_dependladdr.ie46_local.ia46_addr4 104 #define ie6_faddr ie_dependfaddr.ie6_foreign 105 #define ie6_laddr ie_dependladdr.ie6_local 106 107 /* 108 * XXX The defines for inc_* are hacks and should be changed to direct 109 * references. 110 */ 111 struct in_conninfo { 112 u_int8_t inc_flags; 113 u_int8_t inc_len; 114 u_int16_t inc_fibnum; /* XXX was pad, 16 bits is plenty */ 115 /* protocol dependent part */ 116 struct in_endpoints inc_ie; 117 }; 118 119 /* 120 * Flags for inc_flags. 121 */ 122 #define INC_ISIPV6 0x01 123 124 #define inc_isipv6 inc_flags /* temp compatibility */ 125 #define inc_fport inc_ie.ie_fport 126 #define inc_lport inc_ie.ie_lport 127 #define inc_faddr inc_ie.ie_faddr 128 #define inc_laddr inc_ie.ie_laddr 129 #define inc6_faddr inc_ie.ie6_faddr 130 #define inc6_laddr inc_ie.ie6_laddr 131 #define inc6_zoneid inc_ie.ie6_zoneid 132 133 struct icmp6_filter; 134 135 /*- 136 * struct inpcb captures the network layer state for TCP, UDP, and raw IPv4 and 137 * IPv6 sockets. In the case of TCP and UDP, further per-connection state is 138 * hung off of inp_ppcb most of the time. Almost all fields of struct inpcb 139 * are static after creation or protected by a per-inpcb rwlock, inp_lock. A 140 * few fields are protected by multiple locks as indicated in the locking notes 141 * below. For these fields, all of the listed locks must be write-locked for 142 * any modifications. However, these fields can be safely read while any one of 143 * the listed locks are read-locked. This model can permit greater concurrency 144 * for read operations. For example, connections can be looked up while only 145 * holding a read lock on the global pcblist lock. This is important for 146 * performance when attempting to find the connection for a packet given its IP 147 * and port tuple. 148 * 149 * One noteworthy exception is that the global pcbinfo lock follows a different 150 * set of rules in relation to the inp_list field. Rather than being 151 * write-locked for modifications and read-locked for list iterations, it must 152 * be read-locked during modifications and write-locked during list iterations. 153 * This ensures that the relatively rare global list iterations safely walk a 154 * stable snapshot of connections while allowing more common list modifications 155 * to safely grab the pcblist lock just while adding or removing a connection 156 * from the global list. 157 * 158 * Key: 159 * (c) - Constant after initialization 160 * (g) - Protected by the pcbgroup lock 161 * (i) - Protected by the inpcb lock 162 * (p) - Protected by the pcbinfo lock for the inpcb 163 * (l) - Protected by the pcblist lock for the inpcb 164 * (h) - Protected by the pcbhash lock for the inpcb 165 * (s) - Protected by another subsystem's locks 166 * (x) - Undefined locking 167 * 168 * A few other notes: 169 * 170 * When a read lock is held, stability of the field is guaranteed; to write 171 * to a field, a write lock must generally be held. 172 * 173 * netinet/netinet6-layer code should not assume that the inp_socket pointer 174 * is safe to dereference without inp_lock being held, even for protocols 175 * other than TCP (where the inpcb persists during TIMEWAIT even after the 176 * socket has been freed), or there may be close(2)-related races. 177 * 178 * The inp_vflag field is overloaded, and would otherwise ideally be (c). 179 * 180 * TODO: Currently only the TCP stack is leveraging the global pcbinfo lock 181 * read-lock usage during modification, this model can be applied to other 182 * protocols (especially SCTP). 183 */ 184 struct inpcb { 185 LIST_ENTRY(inpcb) inp_hash; /* (h/i) hash list */ 186 LIST_ENTRY(inpcb) inp_pcbgrouphash; /* (g/i) hash list */ 187 LIST_ENTRY(inpcb) inp_list; /* (p/l) list for all PCBs for proto */ 188 /* (p[w]) for list iteration */ 189 /* (p[r]/l) for addition/removal */ 190 void *inp_ppcb; /* (i) pointer to per-protocol pcb */ 191 struct inpcbinfo *inp_pcbinfo; /* (c) PCB list info */ 192 struct inpcbgroup *inp_pcbgroup; /* (g/i) PCB group list */ 193 LIST_ENTRY(inpcb) inp_pcbgroup_wild; /* (g/i/h) group wildcard entry */ 194 struct socket *inp_socket; /* (i) back pointer to socket */ 195 struct ucred *inp_cred; /* (c) cache of socket cred */ 196 u_int32_t inp_flow; /* (i) IPv6 flow information */ 197 int inp_flags; /* (i) generic IP/datagram flags */ 198 int inp_flags2; /* (i) generic IP/datagram flags #2*/ 199 u_char inp_vflag; /* (i) IP version flag (v4/v6) */ 200 u_char inp_ip_ttl; /* (i) time to live proto */ 201 u_char inp_ip_p; /* (c) protocol proto */ 202 u_char inp_ip_minttl; /* (i) minimum TTL or drop */ 203 uint32_t inp_flowid; /* (x) flow id / queue id */ 204 u_int inp_refcount; /* (i) refcount */ 205 void *inp_pspare[5]; /* (x) packet pacing / general use */ 206 uint32_t inp_flowtype; /* (x) M_HASHTYPE value */ 207 uint32_t inp_rss_listen_bucket; /* (x) overridden RSS listen bucket */ 208 u_int inp_ispare[4]; /* (x) packet pacing / user cookie / 209 * general use */ 210 211 /* Local and foreign ports, local and foreign addr. */ 212 struct in_conninfo inp_inc; /* (i) list for PCB's local port */ 213 214 /* MAC and IPSEC policy information. */ 215 struct label *inp_label; /* (i) MAC label */ 216 struct inpcbpolicy *inp_sp; /* (s) for IPSEC */ 217 218 /* Protocol-dependent part; options. */ 219 struct { 220 u_char inp4_ip_tos; /* (i) type of service proto */ 221 struct mbuf *inp4_options; /* (i) IP options */ 222 struct ip_moptions *inp4_moptions; /* (i) IP mcast options */ 223 } inp_depend4; 224 struct { 225 /* (i) IP options */ 226 struct mbuf *inp6_options; 227 /* (i) IP6 options for outgoing packets */ 228 struct ip6_pktopts *inp6_outputopts; 229 /* (i) IP multicast options */ 230 struct ip6_moptions *inp6_moptions; 231 /* (i) ICMPv6 code type filter */ 232 struct icmp6_filter *inp6_icmp6filt; 233 /* (i) IPV6_CHECKSUM setsockopt */ 234 int inp6_cksum; 235 short inp6_hops; 236 } inp_depend6; 237 LIST_ENTRY(inpcb) inp_portlist; /* (i/h) */ 238 struct inpcbport *inp_phd; /* (i/h) head of this list */ 239 #define inp_zero_size offsetof(struct inpcb, inp_gencnt) 240 inp_gen_t inp_gencnt; /* (c) generation count */ 241 struct llentry *inp_lle; /* cached L2 information */ 242 struct rwlock inp_lock; 243 rt_gen_t inp_rt_cookie; /* generation for route entry */ 244 union { /* cached L3 information */ 245 struct route inpu_route; 246 struct route_in6 inpu_route6; 247 } inp_rtu; 248 #define inp_route inp_rtu.inpu_route 249 #define inp_route6 inp_rtu.inpu_route6 250 }; 251 #define inp_fport inp_inc.inc_fport 252 #define inp_lport inp_inc.inc_lport 253 #define inp_faddr inp_inc.inc_faddr 254 #define inp_laddr inp_inc.inc_laddr 255 #define inp_ip_tos inp_depend4.inp4_ip_tos 256 #define inp_options inp_depend4.inp4_options 257 #define inp_moptions inp_depend4.inp4_moptions 258 259 #define in6p_faddr inp_inc.inc6_faddr 260 #define in6p_laddr inp_inc.inc6_laddr 261 #define in6p_zoneid inp_inc.inc6_zoneid 262 #define in6p_hops inp_depend6.inp6_hops /* default hop limit */ 263 #define in6p_flowinfo inp_flow 264 #define in6p_options inp_depend6.inp6_options 265 #define in6p_outputopts inp_depend6.inp6_outputopts 266 #define in6p_moptions inp_depend6.inp6_moptions 267 #define in6p_icmp6filt inp_depend6.inp6_icmp6filt 268 #define in6p_cksum inp_depend6.inp6_cksum 269 270 #define inp_vnet inp_pcbinfo->ipi_vnet 271 272 /* 273 * The range of the generation count, as used in this implementation, is 9e19. 274 * We would have to create 300 billion connections per second for this number 275 * to roll over in a year. This seems sufficiently unlikely that we simply 276 * don't concern ourselves with that possibility. 277 */ 278 279 /* 280 * Interface exported to userland by various protocols which use inpcbs. Hack 281 * alert -- only define if struct xsocket is in scope. 282 */ 283 #ifdef _SYS_SOCKETVAR_H_ 284 struct xinpcb { 285 size_t xi_len; /* length of this structure */ 286 struct inpcb xi_inp; 287 struct xsocket xi_socket; 288 u_quad_t xi_alignment_hack; 289 }; 290 291 struct xinpgen { 292 size_t xig_len; /* length of this structure */ 293 u_int xig_count; /* number of PCBs at this time */ 294 inp_gen_t xig_gen; /* generation count at this time */ 295 so_gen_t xig_sogen; /* socket generation count at this time */ 296 }; 297 #endif /* _SYS_SOCKETVAR_H_ */ 298 299 struct inpcbport { 300 LIST_ENTRY(inpcbport) phd_hash; 301 struct inpcbhead phd_pcblist; 302 u_short phd_port; 303 }; 304 305 /*- 306 * Global data structure for each high-level protocol (UDP, TCP, ...) in both 307 * IPv4 and IPv6. Holds inpcb lists and information for managing them. 308 * 309 * Each pcbinfo is protected by three locks: ipi_lock, ipi_hash_lock and 310 * ipi_list_lock: 311 * - ipi_lock covering the global pcb list stability during loop iteration, 312 * - ipi_hash_lock covering the hashed lookup tables, 313 * - ipi_list_lock covering mutable global fields (such as the global 314 * pcb list) 315 * 316 * The lock order is: 317 * 318 * ipi_lock (before) 319 * inpcb locks (before) 320 * ipi_list locks (before) 321 * {ipi_hash_lock, pcbgroup locks} 322 * 323 * Locking key: 324 * 325 * (c) Constant or nearly constant after initialisation 326 * (g) Locked by ipi_lock 327 * (l) Locked by ipi_list_lock 328 * (h) Read using either ipi_hash_lock or inpcb lock; write requires both 329 * (p) Protected by one or more pcbgroup locks 330 * (x) Synchronisation properties poorly defined 331 */ 332 struct inpcbinfo { 333 /* 334 * Global lock protecting full inpcb list traversal 335 */ 336 struct rwlock ipi_lock; 337 338 /* 339 * Global list of inpcbs on the protocol. 340 */ 341 struct inpcbhead *ipi_listhead; /* (g/l) */ 342 u_int ipi_count; /* (l) */ 343 344 /* 345 * Generation count -- incremented each time a connection is allocated 346 * or freed. 347 */ 348 u_quad_t ipi_gencnt; /* (l) */ 349 350 /* 351 * Fields associated with port lookup and allocation. 352 */ 353 u_short ipi_lastport; /* (x) */ 354 u_short ipi_lastlow; /* (x) */ 355 u_short ipi_lasthi; /* (x) */ 356 357 /* 358 * UMA zone from which inpcbs are allocated for this protocol. 359 */ 360 struct uma_zone *ipi_zone; /* (c) */ 361 362 /* 363 * Connection groups associated with this protocol. These fields are 364 * constant, but pcbgroup structures themselves are protected by 365 * per-pcbgroup locks. 366 */ 367 struct inpcbgroup *ipi_pcbgroups; /* (c) */ 368 u_int ipi_npcbgroups; /* (c) */ 369 u_int ipi_hashfields; /* (c) */ 370 371 /* 372 * Global lock protecting non-pcbgroup hash lookup tables. 373 */ 374 struct rwlock ipi_hash_lock; 375 376 /* 377 * Global hash of inpcbs, hashed by local and foreign addresses and 378 * port numbers. 379 */ 380 struct inpcbhead *ipi_hashbase; /* (h) */ 381 u_long ipi_hashmask; /* (h) */ 382 383 /* 384 * Global hash of inpcbs, hashed by only local port number. 385 */ 386 struct inpcbporthead *ipi_porthashbase; /* (h) */ 387 u_long ipi_porthashmask; /* (h) */ 388 389 /* 390 * List of wildcard inpcbs for use with pcbgroups. In the past, was 391 * per-pcbgroup but is now global. All pcbgroup locks must be held 392 * to modify the list, so any is sufficient to read it. 393 */ 394 struct inpcbhead *ipi_wildbase; /* (p) */ 395 u_long ipi_wildmask; /* (p) */ 396 397 /* 398 * Pointer to network stack instance 399 */ 400 struct vnet *ipi_vnet; /* (c) */ 401 402 /* 403 * general use 2 404 */ 405 void *ipi_pspare[2]; 406 407 /* 408 * Global lock protecting global inpcb list, inpcb count, etc. 409 */ 410 struct rwlock ipi_list_lock; 411 }; 412 413 #ifdef _KERNEL 414 /* 415 * Connection groups hold sets of connections that have similar CPU/thread 416 * affinity. Each connection belongs to exactly one connection group. 417 */ 418 struct inpcbgroup { 419 /* 420 * Per-connection group hash of inpcbs, hashed by local and foreign 421 * addresses and port numbers. 422 */ 423 struct inpcbhead *ipg_hashbase; /* (c) */ 424 u_long ipg_hashmask; /* (c) */ 425 426 /* 427 * Notional affinity of this pcbgroup. 428 */ 429 u_int ipg_cpu; /* (p) */ 430 431 /* 432 * Per-connection group lock, not to be confused with ipi_lock. 433 * Protects the hash table hung off the group, but also the global 434 * wildcard list in inpcbinfo. 435 */ 436 struct mtx ipg_lock; 437 } __aligned(CACHE_LINE_SIZE); 438 439 #define INP_LOCK_INIT(inp, d, t) \ 440 rw_init_flags(&(inp)->inp_lock, (t), RW_RECURSE | RW_DUPOK) 441 #define INP_LOCK_DESTROY(inp) rw_destroy(&(inp)->inp_lock) 442 #define INP_RLOCK(inp) rw_rlock(&(inp)->inp_lock) 443 #define INP_WLOCK(inp) rw_wlock(&(inp)->inp_lock) 444 #define INP_TRY_RLOCK(inp) rw_try_rlock(&(inp)->inp_lock) 445 #define INP_TRY_WLOCK(inp) rw_try_wlock(&(inp)->inp_lock) 446 #define INP_RUNLOCK(inp) rw_runlock(&(inp)->inp_lock) 447 #define INP_WUNLOCK(inp) rw_wunlock(&(inp)->inp_lock) 448 #define INP_TRY_UPGRADE(inp) rw_try_upgrade(&(inp)->inp_lock) 449 #define INP_DOWNGRADE(inp) rw_downgrade(&(inp)->inp_lock) 450 #define INP_WLOCKED(inp) rw_wowned(&(inp)->inp_lock) 451 #define INP_LOCK_ASSERT(inp) rw_assert(&(inp)->inp_lock, RA_LOCKED) 452 #define INP_RLOCK_ASSERT(inp) rw_assert(&(inp)->inp_lock, RA_RLOCKED) 453 #define INP_WLOCK_ASSERT(inp) rw_assert(&(inp)->inp_lock, RA_WLOCKED) 454 #define INP_UNLOCK_ASSERT(inp) rw_assert(&(inp)->inp_lock, RA_UNLOCKED) 455 456 /* 457 * These locking functions are for inpcb consumers outside of sys/netinet, 458 * more specifically, they were added for the benefit of TOE drivers. The 459 * macros are reserved for use by the stack. 460 */ 461 void inp_wlock(struct inpcb *); 462 void inp_wunlock(struct inpcb *); 463 void inp_rlock(struct inpcb *); 464 void inp_runlock(struct inpcb *); 465 466 #ifdef INVARIANTS 467 void inp_lock_assert(struct inpcb *); 468 void inp_unlock_assert(struct inpcb *); 469 #else 470 static __inline void 471 inp_lock_assert(struct inpcb *inp __unused) 472 { 473 } 474 475 static __inline void 476 inp_unlock_assert(struct inpcb *inp __unused) 477 { 478 } 479 480 #endif 481 482 void inp_apply_all(void (*func)(struct inpcb *, void *), void *arg); 483 int inp_ip_tos_get(const struct inpcb *inp); 484 void inp_ip_tos_set(struct inpcb *inp, int val); 485 struct socket * 486 inp_inpcbtosocket(struct inpcb *inp); 487 struct tcpcb * 488 inp_inpcbtotcpcb(struct inpcb *inp); 489 void inp_4tuple_get(struct inpcb *inp, uint32_t *laddr, uint16_t *lp, 490 uint32_t *faddr, uint16_t *fp); 491 short inp_so_options(const struct inpcb *inp); 492 493 #endif /* _KERNEL */ 494 495 #define INP_INFO_LOCK_INIT(ipi, d) \ 496 rw_init_flags(&(ipi)->ipi_lock, (d), RW_RECURSE) 497 #define INP_INFO_LOCK_DESTROY(ipi) rw_destroy(&(ipi)->ipi_lock) 498 #define INP_INFO_RLOCK(ipi) rw_rlock(&(ipi)->ipi_lock) 499 #define INP_INFO_WLOCK(ipi) rw_wlock(&(ipi)->ipi_lock) 500 #define INP_INFO_TRY_RLOCK(ipi) rw_try_rlock(&(ipi)->ipi_lock) 501 #define INP_INFO_TRY_WLOCK(ipi) rw_try_wlock(&(ipi)->ipi_lock) 502 #define INP_INFO_TRY_UPGRADE(ipi) rw_try_upgrade(&(ipi)->ipi_lock) 503 #define INP_INFO_WLOCKED(ipi) rw_wowned(&(ipi)->ipi_lock) 504 #define INP_INFO_RUNLOCK(ipi) rw_runlock(&(ipi)->ipi_lock) 505 #define INP_INFO_WUNLOCK(ipi) rw_wunlock(&(ipi)->ipi_lock) 506 #define INP_INFO_LOCK_ASSERT(ipi) rw_assert(&(ipi)->ipi_lock, RA_LOCKED) 507 #define INP_INFO_RLOCK_ASSERT(ipi) rw_assert(&(ipi)->ipi_lock, RA_RLOCKED) 508 #define INP_INFO_WLOCK_ASSERT(ipi) rw_assert(&(ipi)->ipi_lock, RA_WLOCKED) 509 #define INP_INFO_UNLOCK_ASSERT(ipi) rw_assert(&(ipi)->ipi_lock, RA_UNLOCKED) 510 511 #define INP_LIST_LOCK_INIT(ipi, d) \ 512 rw_init_flags(&(ipi)->ipi_list_lock, (d), 0) 513 #define INP_LIST_LOCK_DESTROY(ipi) rw_destroy(&(ipi)->ipi_list_lock) 514 #define INP_LIST_RLOCK(ipi) rw_rlock(&(ipi)->ipi_list_lock) 515 #define INP_LIST_WLOCK(ipi) rw_wlock(&(ipi)->ipi_list_lock) 516 #define INP_LIST_TRY_RLOCK(ipi) rw_try_rlock(&(ipi)->ipi_list_lock) 517 #define INP_LIST_TRY_WLOCK(ipi) rw_try_wlock(&(ipi)->ipi_list_lock) 518 #define INP_LIST_TRY_UPGRADE(ipi) rw_try_upgrade(&(ipi)->ipi_list_lock) 519 #define INP_LIST_RUNLOCK(ipi) rw_runlock(&(ipi)->ipi_list_lock) 520 #define INP_LIST_WUNLOCK(ipi) rw_wunlock(&(ipi)->ipi_list_lock) 521 #define INP_LIST_LOCK_ASSERT(ipi) \ 522 rw_assert(&(ipi)->ipi_list_lock, RA_LOCKED) 523 #define INP_LIST_RLOCK_ASSERT(ipi) \ 524 rw_assert(&(ipi)->ipi_list_lock, RA_RLOCKED) 525 #define INP_LIST_WLOCK_ASSERT(ipi) \ 526 rw_assert(&(ipi)->ipi_list_lock, RA_WLOCKED) 527 #define INP_LIST_UNLOCK_ASSERT(ipi) \ 528 rw_assert(&(ipi)->ipi_list_lock, RA_UNLOCKED) 529 530 #define INP_HASH_LOCK_INIT(ipi, d) \ 531 rw_init_flags(&(ipi)->ipi_hash_lock, (d), 0) 532 #define INP_HASH_LOCK_DESTROY(ipi) rw_destroy(&(ipi)->ipi_hash_lock) 533 #define INP_HASH_RLOCK(ipi) rw_rlock(&(ipi)->ipi_hash_lock) 534 #define INP_HASH_WLOCK(ipi) rw_wlock(&(ipi)->ipi_hash_lock) 535 #define INP_HASH_RUNLOCK(ipi) rw_runlock(&(ipi)->ipi_hash_lock) 536 #define INP_HASH_WUNLOCK(ipi) rw_wunlock(&(ipi)->ipi_hash_lock) 537 #define INP_HASH_LOCK_ASSERT(ipi) rw_assert(&(ipi)->ipi_hash_lock, \ 538 RA_LOCKED) 539 #define INP_HASH_WLOCK_ASSERT(ipi) rw_assert(&(ipi)->ipi_hash_lock, \ 540 RA_WLOCKED) 541 542 #define INP_GROUP_LOCK_INIT(ipg, d) mtx_init(&(ipg)->ipg_lock, (d), NULL, \ 543 MTX_DEF | MTX_DUPOK) 544 #define INP_GROUP_LOCK_DESTROY(ipg) mtx_destroy(&(ipg)->ipg_lock) 545 546 #define INP_GROUP_LOCK(ipg) mtx_lock(&(ipg)->ipg_lock) 547 #define INP_GROUP_LOCK_ASSERT(ipg) mtx_assert(&(ipg)->ipg_lock, MA_OWNED) 548 #define INP_GROUP_UNLOCK(ipg) mtx_unlock(&(ipg)->ipg_lock) 549 550 #define INP_PCBHASH(faddr, lport, fport, mask) \ 551 (((faddr) ^ ((faddr) >> 16) ^ ntohs((lport) ^ (fport))) & (mask)) 552 #define INP_PCBPORTHASH(lport, mask) \ 553 (ntohs((lport)) & (mask)) 554 #define INP6_PCBHASHKEY(faddr) ((faddr)->s6_addr32[3]) 555 556 /* 557 * Flags for inp_vflags -- historically version flags only 558 */ 559 #define INP_IPV4 0x1 560 #define INP_IPV6 0x2 561 #define INP_IPV6PROTO 0x4 /* opened under IPv6 protocol */ 562 563 /* 564 * Flags for inp_flags. 565 */ 566 #define INP_RECVOPTS 0x00000001 /* receive incoming IP options */ 567 #define INP_RECVRETOPTS 0x00000002 /* receive IP options for reply */ 568 #define INP_RECVDSTADDR 0x00000004 /* receive IP dst address */ 569 #define INP_HDRINCL 0x00000008 /* user supplies entire IP header */ 570 #define INP_HIGHPORT 0x00000010 /* user wants "high" port binding */ 571 #define INP_LOWPORT 0x00000020 /* user wants "low" port binding */ 572 #define INP_ANONPORT 0x00000040 /* port chosen for user */ 573 #define INP_RECVIF 0x00000080 /* receive incoming interface */ 574 #define INP_MTUDISC 0x00000100 /* user can do MTU discovery */ 575 /* 0x000200 unused: was INP_FAITH */ 576 #define INP_RECVTTL 0x00000400 /* receive incoming IP TTL */ 577 #define INP_DONTFRAG 0x00000800 /* don't fragment packet */ 578 #define INP_BINDANY 0x00001000 /* allow bind to any address */ 579 #define INP_INHASHLIST 0x00002000 /* in_pcbinshash() has been called */ 580 #define INP_RECVTOS 0x00004000 /* receive incoming IP TOS */ 581 #define IN6P_IPV6_V6ONLY 0x00008000 /* restrict AF_INET6 socket for v6 */ 582 #define IN6P_PKTINFO 0x00010000 /* receive IP6 dst and I/F */ 583 #define IN6P_HOPLIMIT 0x00020000 /* receive hoplimit */ 584 #define IN6P_HOPOPTS 0x00040000 /* receive hop-by-hop options */ 585 #define IN6P_DSTOPTS 0x00080000 /* receive dst options after rthdr */ 586 #define IN6P_RTHDR 0x00100000 /* receive routing header */ 587 #define IN6P_RTHDRDSTOPTS 0x00200000 /* receive dstoptions before rthdr */ 588 #define IN6P_TCLASS 0x00400000 /* receive traffic class value */ 589 #define IN6P_AUTOFLOWLABEL 0x00800000 /* attach flowlabel automatically */ 590 #define INP_TIMEWAIT 0x01000000 /* in TIMEWAIT, ppcb is tcptw */ 591 #define INP_ONESBCAST 0x02000000 /* send all-ones broadcast */ 592 #define INP_DROPPED 0x04000000 /* protocol drop flag */ 593 #define INP_SOCKREF 0x08000000 /* strong socket reference */ 594 #define INP_RESERVED_0 0x10000000 /* reserved field */ 595 #define INP_RESERVED_1 0x20000000 /* reserved field */ 596 #define IN6P_RFC2292 0x40000000 /* used RFC2292 API on the socket */ 597 #define IN6P_MTU 0x80000000 /* receive path MTU */ 598 599 #define INP_CONTROLOPTS (INP_RECVOPTS|INP_RECVRETOPTS|INP_RECVDSTADDR|\ 600 INP_RECVIF|INP_RECVTTL|INP_RECVTOS|\ 601 IN6P_PKTINFO|IN6P_HOPLIMIT|IN6P_HOPOPTS|\ 602 IN6P_DSTOPTS|IN6P_RTHDR|IN6P_RTHDRDSTOPTS|\ 603 IN6P_TCLASS|IN6P_AUTOFLOWLABEL|IN6P_RFC2292|\ 604 IN6P_MTU) 605 606 /* 607 * Flags for inp_flags2. 608 */ 609 #define INP_LLE_VALID 0x00000001 /* cached lle is valid */ 610 #define INP_RT_VALID 0x00000002 /* cached rtentry is valid */ 611 #define INP_PCBGROUPWILD 0x00000004 /* in pcbgroup wildcard list */ 612 #define INP_REUSEPORT 0x00000008 /* SO_REUSEPORT option is set */ 613 #define INP_FREED 0x00000010 /* inp itself is not valid */ 614 #define INP_REUSEADDR 0x00000020 /* SO_REUSEADDR option is set */ 615 #define INP_BINDMULTI 0x00000040 /* IP_BINDMULTI option is set */ 616 #define INP_RSS_BUCKET_SET 0x00000080 /* IP_RSS_LISTEN_BUCKET is set */ 617 #define INP_RECVFLOWID 0x00000100 /* populate recv datagram with flow info */ 618 #define INP_RECVRSSBUCKETID 0x00000200 /* populate recv datagram with bucket id */ 619 620 /* 621 * Flags passed to in_pcblookup*() functions. 622 */ 623 #define INPLOOKUP_WILDCARD 0x00000001 /* Allow wildcard sockets. */ 624 #define INPLOOKUP_RLOCKPCB 0x00000002 /* Return inpcb read-locked. */ 625 #define INPLOOKUP_WLOCKPCB 0x00000004 /* Return inpcb write-locked. */ 626 627 #define INPLOOKUP_MASK (INPLOOKUP_WILDCARD | INPLOOKUP_RLOCKPCB | \ 628 INPLOOKUP_WLOCKPCB) 629 630 #define sotoinpcb(so) ((struct inpcb *)(so)->so_pcb) 631 #define sotoin6pcb(so) sotoinpcb(so) /* for KAME src sync over BSD*'s */ 632 633 #define INP_SOCKAF(so) so->so_proto->pr_domain->dom_family 634 635 #define INP_CHECK_SOCKAF(so, af) (INP_SOCKAF(so) == af) 636 637 /* 638 * Constants for pcbinfo.ipi_hashfields. 639 */ 640 #define IPI_HASHFIELDS_NONE 0 641 #define IPI_HASHFIELDS_2TUPLE 1 642 #define IPI_HASHFIELDS_4TUPLE 2 643 644 #ifdef _KERNEL 645 VNET_DECLARE(int, ipport_reservedhigh); 646 VNET_DECLARE(int, ipport_reservedlow); 647 VNET_DECLARE(int, ipport_lowfirstauto); 648 VNET_DECLARE(int, ipport_lowlastauto); 649 VNET_DECLARE(int, ipport_firstauto); 650 VNET_DECLARE(int, ipport_lastauto); 651 VNET_DECLARE(int, ipport_hifirstauto); 652 VNET_DECLARE(int, ipport_hilastauto); 653 VNET_DECLARE(int, ipport_randomized); 654 VNET_DECLARE(int, ipport_randomcps); 655 VNET_DECLARE(int, ipport_randomtime); 656 VNET_DECLARE(int, ipport_stoprandom); 657 VNET_DECLARE(int, ipport_tcpallocs); 658 659 #define V_ipport_reservedhigh VNET(ipport_reservedhigh) 660 #define V_ipport_reservedlow VNET(ipport_reservedlow) 661 #define V_ipport_lowfirstauto VNET(ipport_lowfirstauto) 662 #define V_ipport_lowlastauto VNET(ipport_lowlastauto) 663 #define V_ipport_firstauto VNET(ipport_firstauto) 664 #define V_ipport_lastauto VNET(ipport_lastauto) 665 #define V_ipport_hifirstauto VNET(ipport_hifirstauto) 666 #define V_ipport_hilastauto VNET(ipport_hilastauto) 667 #define V_ipport_randomized VNET(ipport_randomized) 668 #define V_ipport_randomcps VNET(ipport_randomcps) 669 #define V_ipport_randomtime VNET(ipport_randomtime) 670 #define V_ipport_stoprandom VNET(ipport_stoprandom) 671 #define V_ipport_tcpallocs VNET(ipport_tcpallocs) 672 673 void in_pcbinfo_destroy(struct inpcbinfo *); 674 void in_pcbinfo_init(struct inpcbinfo *, const char *, struct inpcbhead *, 675 int, int, char *, uma_init, uma_fini, uint32_t, u_int); 676 677 int in_pcbbind_check_bindmulti(const struct inpcb *ni, 678 const struct inpcb *oi); 679 680 struct inpcbgroup * 681 in_pcbgroup_byhash(struct inpcbinfo *, u_int, uint32_t); 682 struct inpcbgroup * 683 in_pcbgroup_byinpcb(struct inpcb *); 684 struct inpcbgroup * 685 in_pcbgroup_bytuple(struct inpcbinfo *, struct in_addr, u_short, 686 struct in_addr, u_short); 687 void in_pcbgroup_destroy(struct inpcbinfo *); 688 int in_pcbgroup_enabled(struct inpcbinfo *); 689 void in_pcbgroup_init(struct inpcbinfo *, u_int, int); 690 void in_pcbgroup_remove(struct inpcb *); 691 void in_pcbgroup_update(struct inpcb *); 692 void in_pcbgroup_update_mbuf(struct inpcb *, struct mbuf *); 693 694 void in_pcbpurgeif0(struct inpcbinfo *, struct ifnet *); 695 int in_pcballoc(struct socket *, struct inpcbinfo *); 696 int in_pcbbind(struct inpcb *, struct sockaddr *, struct ucred *); 697 int in_pcb_lport(struct inpcb *, struct in_addr *, u_short *, 698 struct ucred *, int); 699 int in_pcbbind_setup(struct inpcb *, struct sockaddr *, in_addr_t *, 700 u_short *, struct ucred *); 701 int in_pcbconnect(struct inpcb *, struct sockaddr *, struct ucred *); 702 int in_pcbconnect_mbuf(struct inpcb *, struct sockaddr *, struct ucred *, 703 struct mbuf *); 704 int in_pcbconnect_setup(struct inpcb *, struct sockaddr *, in_addr_t *, 705 u_short *, in_addr_t *, u_short *, struct inpcb **, 706 struct ucred *); 707 void in_pcbdetach(struct inpcb *); 708 void in_pcbdisconnect(struct inpcb *); 709 void in_pcbdrop(struct inpcb *); 710 void in_pcbfree(struct inpcb *); 711 int in_pcbinshash(struct inpcb *); 712 int in_pcbinshash_nopcbgroup(struct inpcb *); 713 int in_pcbladdr(struct inpcb *, struct in_addr *, struct in_addr *, 714 struct ucred *); 715 struct inpcb * 716 in_pcblookup_local(struct inpcbinfo *, 717 struct in_addr, u_short, int, struct ucred *); 718 struct inpcb * 719 in_pcblookup(struct inpcbinfo *, struct in_addr, u_int, 720 struct in_addr, u_int, int, struct ifnet *); 721 struct inpcb * 722 in_pcblookup_mbuf(struct inpcbinfo *, struct in_addr, u_int, 723 struct in_addr, u_int, int, struct ifnet *, struct mbuf *); 724 void in_pcbnotifyall(struct inpcbinfo *pcbinfo, struct in_addr, 725 int, struct inpcb *(*)(struct inpcb *, int)); 726 void in_pcbref(struct inpcb *); 727 void in_pcbrehash(struct inpcb *); 728 void in_pcbrehash_mbuf(struct inpcb *, struct mbuf *); 729 int in_pcbrele(struct inpcb *); 730 int in_pcbrele_rlocked(struct inpcb *); 731 int in_pcbrele_wlocked(struct inpcb *); 732 void in_losing(struct inpcb *); 733 void in_pcbsetsolabel(struct socket *so); 734 int in_getpeeraddr(struct socket *so, struct sockaddr **nam); 735 int in_getsockaddr(struct socket *so, struct sockaddr **nam); 736 struct sockaddr * 737 in_sockaddr(in_port_t port, struct in_addr *addr); 738 void in_pcbsosetlabel(struct socket *so); 739 #endif /* _KERNEL */ 740 741 #endif /* !_NETINET_IN_PCB_H_ */ 742