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 * 3. 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 m_snd_tag; 185 struct inpcb { 186 LIST_ENTRY(inpcb) inp_hash; /* (h/i) hash list */ 187 LIST_ENTRY(inpcb) inp_pcbgrouphash; /* (g/i) hash list */ 188 LIST_ENTRY(inpcb) inp_list; /* (p/l) list for all PCBs for proto */ 189 /* (p[w]) for list iteration */ 190 /* (p[r]/l) for addition/removal */ 191 void *inp_ppcb; /* (i) pointer to per-protocol pcb */ 192 struct inpcbinfo *inp_pcbinfo; /* (c) PCB list info */ 193 struct inpcbgroup *inp_pcbgroup; /* (g/i) PCB group list */ 194 LIST_ENTRY(inpcb) inp_pcbgroup_wild; /* (g/i/h) group wildcard entry */ 195 struct socket *inp_socket; /* (i) back pointer to socket */ 196 struct ucred *inp_cred; /* (c) cache of socket cred */ 197 u_int32_t inp_flow; /* (i) IPv6 flow information */ 198 int inp_flags; /* (i) generic IP/datagram flags */ 199 int inp_flags2; /* (i) generic IP/datagram flags #2*/ 200 u_char inp_vflag; /* (i) IP version flag (v4/v6) */ 201 u_char inp_ip_ttl; /* (i) time to live proto */ 202 u_char inp_ip_p; /* (c) protocol proto */ 203 u_char inp_ip_minttl; /* (i) minimum TTL or drop */ 204 uint32_t inp_flowid; /* (x) flow id / queue id */ 205 u_int inp_refcount; /* (i) refcount */ 206 struct m_snd_tag *inp_snd_tag; /* (i) send tag for outgoing mbufs */ 207 void *inp_pspare[4]; /* (x) general use */ 208 uint32_t inp_flowtype; /* (x) M_HASHTYPE value */ 209 uint32_t inp_rss_listen_bucket; /* (x) overridden RSS listen bucket */ 210 u_int inp_ispare[4]; /* (x) user cookie / general use */ 211 212 /* Local and foreign ports, local and foreign addr. */ 213 struct in_conninfo inp_inc; /* (i) list for PCB's local port */ 214 215 /* MAC and IPSEC policy information. */ 216 struct label *inp_label; /* (i) MAC label */ 217 struct inpcbpolicy *inp_sp; /* (s) for IPSEC */ 218 219 /* Protocol-dependent part; options. */ 220 struct { 221 u_char inp4_ip_tos; /* (i) type of service proto */ 222 struct mbuf *inp4_options; /* (i) IP options */ 223 struct ip_moptions *inp4_moptions; /* (i) IP mcast options */ 224 } inp_depend4; 225 struct { 226 /* (i) IP options */ 227 struct mbuf *inp6_options; 228 /* (i) IP6 options for outgoing packets */ 229 struct ip6_pktopts *inp6_outputopts; 230 /* (i) IP multicast options */ 231 struct ip6_moptions *inp6_moptions; 232 /* (i) ICMPv6 code type filter */ 233 struct icmp6_filter *inp6_icmp6filt; 234 /* (i) IPV6_CHECKSUM setsockopt */ 235 int inp6_cksum; 236 short inp6_hops; 237 } inp_depend6; 238 LIST_ENTRY(inpcb) inp_portlist; /* (i/h) */ 239 struct inpcbport *inp_phd; /* (i/h) head of this list */ 240 #define inp_zero_size offsetof(struct inpcb, inp_gencnt) 241 inp_gen_t inp_gencnt; /* (c) generation count */ 242 struct llentry *inp_lle; /* cached L2 information */ 243 struct rwlock inp_lock; 244 rt_gen_t inp_rt_cookie; /* generation for route entry */ 245 union { /* cached L3 information */ 246 struct route inpu_route; 247 struct route_in6 inpu_route6; 248 } inp_rtu; 249 #define inp_route inp_rtu.inpu_route 250 #define inp_route6 inp_rtu.inpu_route6 251 }; 252 #define inp_fport inp_inc.inc_fport 253 #define inp_lport inp_inc.inc_lport 254 #define inp_faddr inp_inc.inc_faddr 255 #define inp_laddr inp_inc.inc_laddr 256 #define inp_ip_tos inp_depend4.inp4_ip_tos 257 #define inp_options inp_depend4.inp4_options 258 #define inp_moptions inp_depend4.inp4_moptions 259 260 #define in6p_faddr inp_inc.inc6_faddr 261 #define in6p_laddr inp_inc.inc6_laddr 262 #define in6p_zoneid inp_inc.inc6_zoneid 263 #define in6p_hops inp_depend6.inp6_hops /* default hop limit */ 264 #define in6p_flowinfo inp_flow 265 #define in6p_options inp_depend6.inp6_options 266 #define in6p_outputopts inp_depend6.inp6_outputopts 267 #define in6p_moptions inp_depend6.inp6_moptions 268 #define in6p_icmp6filt inp_depend6.inp6_icmp6filt 269 #define in6p_cksum inp_depend6.inp6_cksum 270 271 #define inp_vnet inp_pcbinfo->ipi_vnet 272 273 /* 274 * The range of the generation count, as used in this implementation, is 9e19. 275 * We would have to create 300 billion connections per second for this number 276 * to roll over in a year. This seems sufficiently unlikely that we simply 277 * don't concern ourselves with that possibility. 278 */ 279 280 /* 281 * Interface exported to userland by various protocols which use inpcbs. Hack 282 * alert -- only define if struct xsocket is in scope. 283 */ 284 #ifdef _SYS_SOCKETVAR_H_ 285 struct xinpcb { 286 size_t xi_len; /* length of this structure */ 287 struct inpcb xi_inp; 288 struct xsocket xi_socket; 289 u_quad_t xi_alignment_hack; 290 }; 291 292 struct xinpgen { 293 size_t xig_len; /* length of this structure */ 294 u_int xig_count; /* number of PCBs at this time */ 295 inp_gen_t xig_gen; /* generation count at this time */ 296 so_gen_t xig_sogen; /* socket generation count at this time */ 297 }; 298 #endif /* _SYS_SOCKETVAR_H_ */ 299 300 struct inpcbport { 301 LIST_ENTRY(inpcbport) phd_hash; 302 struct inpcbhead phd_pcblist; 303 u_short phd_port; 304 }; 305 306 /*- 307 * Global data structure for each high-level protocol (UDP, TCP, ...) in both 308 * IPv4 and IPv6. Holds inpcb lists and information for managing them. 309 * 310 * Each pcbinfo is protected by three locks: ipi_lock, ipi_hash_lock and 311 * ipi_list_lock: 312 * - ipi_lock covering the global pcb list stability during loop iteration, 313 * - ipi_hash_lock covering the hashed lookup tables, 314 * - ipi_list_lock covering mutable global fields (such as the global 315 * pcb list) 316 * 317 * The lock order is: 318 * 319 * ipi_lock (before) 320 * inpcb locks (before) 321 * ipi_list locks (before) 322 * {ipi_hash_lock, pcbgroup locks} 323 * 324 * Locking key: 325 * 326 * (c) Constant or nearly constant after initialisation 327 * (g) Locked by ipi_lock 328 * (l) Locked by ipi_list_lock 329 * (h) Read using either ipi_hash_lock or inpcb lock; write requires both 330 * (p) Protected by one or more pcbgroup locks 331 * (x) Synchronisation properties poorly defined 332 */ 333 struct inpcbinfo { 334 /* 335 * Global lock protecting full inpcb list traversal 336 */ 337 struct rwlock ipi_lock; 338 339 /* 340 * Global list of inpcbs on the protocol. 341 */ 342 struct inpcbhead *ipi_listhead; /* (g/l) */ 343 u_int ipi_count; /* (l) */ 344 345 /* 346 * Generation count -- incremented each time a connection is allocated 347 * or freed. 348 */ 349 u_quad_t ipi_gencnt; /* (l) */ 350 351 /* 352 * Fields associated with port lookup and allocation. 353 */ 354 u_short ipi_lastport; /* (x) */ 355 u_short ipi_lastlow; /* (x) */ 356 u_short ipi_lasthi; /* (x) */ 357 358 /* 359 * UMA zone from which inpcbs are allocated for this protocol. 360 */ 361 struct uma_zone *ipi_zone; /* (c) */ 362 363 /* 364 * Connection groups associated with this protocol. These fields are 365 * constant, but pcbgroup structures themselves are protected by 366 * per-pcbgroup locks. 367 */ 368 struct inpcbgroup *ipi_pcbgroups; /* (c) */ 369 u_int ipi_npcbgroups; /* (c) */ 370 u_int ipi_hashfields; /* (c) */ 371 372 /* 373 * Global lock protecting non-pcbgroup hash lookup tables. 374 */ 375 struct rwlock ipi_hash_lock; 376 377 /* 378 * Global hash of inpcbs, hashed by local and foreign addresses and 379 * port numbers. 380 */ 381 struct inpcbhead *ipi_hashbase; /* (h) */ 382 u_long ipi_hashmask; /* (h) */ 383 384 /* 385 * Global hash of inpcbs, hashed by only local port number. 386 */ 387 struct inpcbporthead *ipi_porthashbase; /* (h) */ 388 u_long ipi_porthashmask; /* (h) */ 389 390 /* 391 * List of wildcard inpcbs for use with pcbgroups. In the past, was 392 * per-pcbgroup but is now global. All pcbgroup locks must be held 393 * to modify the list, so any is sufficient to read it. 394 */ 395 struct inpcbhead *ipi_wildbase; /* (p) */ 396 u_long ipi_wildmask; /* (p) */ 397 398 /* 399 * Pointer to network stack instance 400 */ 401 struct vnet *ipi_vnet; /* (c) */ 402 403 /* 404 * general use 2 405 */ 406 void *ipi_pspare[2]; 407 408 /* 409 * Global lock protecting global inpcb list, inpcb count, etc. 410 */ 411 struct rwlock ipi_list_lock; 412 }; 413 414 #ifdef _KERNEL 415 /* 416 * Connection groups hold sets of connections that have similar CPU/thread 417 * affinity. Each connection belongs to exactly one connection group. 418 */ 419 struct inpcbgroup { 420 /* 421 * Per-connection group hash of inpcbs, hashed by local and foreign 422 * addresses and port numbers. 423 */ 424 struct inpcbhead *ipg_hashbase; /* (c) */ 425 u_long ipg_hashmask; /* (c) */ 426 427 /* 428 * Notional affinity of this pcbgroup. 429 */ 430 u_int ipg_cpu; /* (p) */ 431 432 /* 433 * Per-connection group lock, not to be confused with ipi_lock. 434 * Protects the hash table hung off the group, but also the global 435 * wildcard list in inpcbinfo. 436 */ 437 struct mtx ipg_lock; 438 } __aligned(CACHE_LINE_SIZE); 439 440 #define INP_LOCK_INIT(inp, d, t) \ 441 rw_init_flags(&(inp)->inp_lock, (t), RW_RECURSE | RW_DUPOK) 442 #define INP_LOCK_DESTROY(inp) rw_destroy(&(inp)->inp_lock) 443 #define INP_RLOCK(inp) rw_rlock(&(inp)->inp_lock) 444 #define INP_WLOCK(inp) rw_wlock(&(inp)->inp_lock) 445 #define INP_TRY_RLOCK(inp) rw_try_rlock(&(inp)->inp_lock) 446 #define INP_TRY_WLOCK(inp) rw_try_wlock(&(inp)->inp_lock) 447 #define INP_RUNLOCK(inp) rw_runlock(&(inp)->inp_lock) 448 #define INP_WUNLOCK(inp) rw_wunlock(&(inp)->inp_lock) 449 #define INP_TRY_UPGRADE(inp) rw_try_upgrade(&(inp)->inp_lock) 450 #define INP_DOWNGRADE(inp) rw_downgrade(&(inp)->inp_lock) 451 #define INP_WLOCKED(inp) rw_wowned(&(inp)->inp_lock) 452 #define INP_LOCK_ASSERT(inp) rw_assert(&(inp)->inp_lock, RA_LOCKED) 453 #define INP_RLOCK_ASSERT(inp) rw_assert(&(inp)->inp_lock, RA_RLOCKED) 454 #define INP_WLOCK_ASSERT(inp) rw_assert(&(inp)->inp_lock, RA_WLOCKED) 455 #define INP_UNLOCK_ASSERT(inp) rw_assert(&(inp)->inp_lock, RA_UNLOCKED) 456 457 /* 458 * These locking functions are for inpcb consumers outside of sys/netinet, 459 * more specifically, they were added for the benefit of TOE drivers. The 460 * macros are reserved for use by the stack. 461 */ 462 void inp_wlock(struct inpcb *); 463 void inp_wunlock(struct inpcb *); 464 void inp_rlock(struct inpcb *); 465 void inp_runlock(struct inpcb *); 466 467 #ifdef INVARIANT_SUPPORT 468 void inp_lock_assert(struct inpcb *); 469 void inp_unlock_assert(struct inpcb *); 470 #else 471 #define inp_lock_assert(inp) do {} while (0) 472 #define inp_unlock_assert(inp) do {} while (0) 473 #endif 474 475 void inp_apply_all(void (*func)(struct inpcb *, void *), void *arg); 476 int inp_ip_tos_get(const struct inpcb *inp); 477 void inp_ip_tos_set(struct inpcb *inp, int val); 478 struct socket * 479 inp_inpcbtosocket(struct inpcb *inp); 480 struct tcpcb * 481 inp_inpcbtotcpcb(struct inpcb *inp); 482 void inp_4tuple_get(struct inpcb *inp, uint32_t *laddr, uint16_t *lp, 483 uint32_t *faddr, uint16_t *fp); 484 short inp_so_options(const struct inpcb *inp); 485 486 #endif /* _KERNEL */ 487 488 #define INP_INFO_LOCK_INIT(ipi, d) \ 489 rw_init_flags(&(ipi)->ipi_lock, (d), RW_RECURSE) 490 #define INP_INFO_LOCK_DESTROY(ipi) rw_destroy(&(ipi)->ipi_lock) 491 #define INP_INFO_RLOCK(ipi) rw_rlock(&(ipi)->ipi_lock) 492 #define INP_INFO_WLOCK(ipi) rw_wlock(&(ipi)->ipi_lock) 493 #define INP_INFO_TRY_RLOCK(ipi) rw_try_rlock(&(ipi)->ipi_lock) 494 #define INP_INFO_TRY_WLOCK(ipi) rw_try_wlock(&(ipi)->ipi_lock) 495 #define INP_INFO_TRY_UPGRADE(ipi) rw_try_upgrade(&(ipi)->ipi_lock) 496 #define INP_INFO_WLOCKED(ipi) rw_wowned(&(ipi)->ipi_lock) 497 #define INP_INFO_RUNLOCK(ipi) rw_runlock(&(ipi)->ipi_lock) 498 #define INP_INFO_WUNLOCK(ipi) rw_wunlock(&(ipi)->ipi_lock) 499 #define INP_INFO_LOCK_ASSERT(ipi) rw_assert(&(ipi)->ipi_lock, RA_LOCKED) 500 #define INP_INFO_RLOCK_ASSERT(ipi) rw_assert(&(ipi)->ipi_lock, RA_RLOCKED) 501 #define INP_INFO_WLOCK_ASSERT(ipi) rw_assert(&(ipi)->ipi_lock, RA_WLOCKED) 502 #define INP_INFO_UNLOCK_ASSERT(ipi) rw_assert(&(ipi)->ipi_lock, RA_UNLOCKED) 503 504 #define INP_LIST_LOCK_INIT(ipi, d) \ 505 rw_init_flags(&(ipi)->ipi_list_lock, (d), 0) 506 #define INP_LIST_LOCK_DESTROY(ipi) rw_destroy(&(ipi)->ipi_list_lock) 507 #define INP_LIST_RLOCK(ipi) rw_rlock(&(ipi)->ipi_list_lock) 508 #define INP_LIST_WLOCK(ipi) rw_wlock(&(ipi)->ipi_list_lock) 509 #define INP_LIST_TRY_RLOCK(ipi) rw_try_rlock(&(ipi)->ipi_list_lock) 510 #define INP_LIST_TRY_WLOCK(ipi) rw_try_wlock(&(ipi)->ipi_list_lock) 511 #define INP_LIST_TRY_UPGRADE(ipi) rw_try_upgrade(&(ipi)->ipi_list_lock) 512 #define INP_LIST_RUNLOCK(ipi) rw_runlock(&(ipi)->ipi_list_lock) 513 #define INP_LIST_WUNLOCK(ipi) rw_wunlock(&(ipi)->ipi_list_lock) 514 #define INP_LIST_LOCK_ASSERT(ipi) \ 515 rw_assert(&(ipi)->ipi_list_lock, RA_LOCKED) 516 #define INP_LIST_RLOCK_ASSERT(ipi) \ 517 rw_assert(&(ipi)->ipi_list_lock, RA_RLOCKED) 518 #define INP_LIST_WLOCK_ASSERT(ipi) \ 519 rw_assert(&(ipi)->ipi_list_lock, RA_WLOCKED) 520 #define INP_LIST_UNLOCK_ASSERT(ipi) \ 521 rw_assert(&(ipi)->ipi_list_lock, RA_UNLOCKED) 522 523 #define INP_HASH_LOCK_INIT(ipi, d) \ 524 rw_init_flags(&(ipi)->ipi_hash_lock, (d), 0) 525 #define INP_HASH_LOCK_DESTROY(ipi) rw_destroy(&(ipi)->ipi_hash_lock) 526 #define INP_HASH_RLOCK(ipi) rw_rlock(&(ipi)->ipi_hash_lock) 527 #define INP_HASH_WLOCK(ipi) rw_wlock(&(ipi)->ipi_hash_lock) 528 #define INP_HASH_RUNLOCK(ipi) rw_runlock(&(ipi)->ipi_hash_lock) 529 #define INP_HASH_WUNLOCK(ipi) rw_wunlock(&(ipi)->ipi_hash_lock) 530 #define INP_HASH_LOCK_ASSERT(ipi) rw_assert(&(ipi)->ipi_hash_lock, \ 531 RA_LOCKED) 532 #define INP_HASH_WLOCK_ASSERT(ipi) rw_assert(&(ipi)->ipi_hash_lock, \ 533 RA_WLOCKED) 534 535 #define INP_GROUP_LOCK_INIT(ipg, d) mtx_init(&(ipg)->ipg_lock, (d), NULL, \ 536 MTX_DEF | MTX_DUPOK) 537 #define INP_GROUP_LOCK_DESTROY(ipg) mtx_destroy(&(ipg)->ipg_lock) 538 539 #define INP_GROUP_LOCK(ipg) mtx_lock(&(ipg)->ipg_lock) 540 #define INP_GROUP_LOCK_ASSERT(ipg) mtx_assert(&(ipg)->ipg_lock, MA_OWNED) 541 #define INP_GROUP_UNLOCK(ipg) mtx_unlock(&(ipg)->ipg_lock) 542 543 #define INP_PCBHASH(faddr, lport, fport, mask) \ 544 (((faddr) ^ ((faddr) >> 16) ^ ntohs((lport) ^ (fport))) & (mask)) 545 #define INP_PCBPORTHASH(lport, mask) \ 546 (ntohs((lport)) & (mask)) 547 #define INP6_PCBHASHKEY(faddr) ((faddr)->s6_addr32[3]) 548 549 /* 550 * Flags for inp_vflags -- historically version flags only 551 */ 552 #define INP_IPV4 0x1 553 #define INP_IPV6 0x2 554 #define INP_IPV6PROTO 0x4 /* opened under IPv6 protocol */ 555 556 /* 557 * Flags for inp_flags. 558 */ 559 #define INP_RECVOPTS 0x00000001 /* receive incoming IP options */ 560 #define INP_RECVRETOPTS 0x00000002 /* receive IP options for reply */ 561 #define INP_RECVDSTADDR 0x00000004 /* receive IP dst address */ 562 #define INP_HDRINCL 0x00000008 /* user supplies entire IP header */ 563 #define INP_HIGHPORT 0x00000010 /* user wants "high" port binding */ 564 #define INP_LOWPORT 0x00000020 /* user wants "low" port binding */ 565 #define INP_ANONPORT 0x00000040 /* port chosen for user */ 566 #define INP_RECVIF 0x00000080 /* receive incoming interface */ 567 #define INP_MTUDISC 0x00000100 /* user can do MTU discovery */ 568 /* 0x000200 unused: was INP_FAITH */ 569 #define INP_RECVTTL 0x00000400 /* receive incoming IP TTL */ 570 #define INP_DONTFRAG 0x00000800 /* don't fragment packet */ 571 #define INP_BINDANY 0x00001000 /* allow bind to any address */ 572 #define INP_INHASHLIST 0x00002000 /* in_pcbinshash() has been called */ 573 #define INP_RECVTOS 0x00004000 /* receive incoming IP TOS */ 574 #define IN6P_IPV6_V6ONLY 0x00008000 /* restrict AF_INET6 socket for v6 */ 575 #define IN6P_PKTINFO 0x00010000 /* receive IP6 dst and I/F */ 576 #define IN6P_HOPLIMIT 0x00020000 /* receive hoplimit */ 577 #define IN6P_HOPOPTS 0x00040000 /* receive hop-by-hop options */ 578 #define IN6P_DSTOPTS 0x00080000 /* receive dst options after rthdr */ 579 #define IN6P_RTHDR 0x00100000 /* receive routing header */ 580 #define IN6P_RTHDRDSTOPTS 0x00200000 /* receive dstoptions before rthdr */ 581 #define IN6P_TCLASS 0x00400000 /* receive traffic class value */ 582 #define IN6P_AUTOFLOWLABEL 0x00800000 /* attach flowlabel automatically */ 583 #define INP_TIMEWAIT 0x01000000 /* in TIMEWAIT, ppcb is tcptw */ 584 #define INP_ONESBCAST 0x02000000 /* send all-ones broadcast */ 585 #define INP_DROPPED 0x04000000 /* protocol drop flag */ 586 #define INP_SOCKREF 0x08000000 /* strong socket reference */ 587 #define INP_RESERVED_0 0x10000000 /* reserved field */ 588 #define INP_RESERVED_1 0x20000000 /* reserved field */ 589 #define IN6P_RFC2292 0x40000000 /* used RFC2292 API on the socket */ 590 #define IN6P_MTU 0x80000000 /* receive path MTU */ 591 592 #define INP_CONTROLOPTS (INP_RECVOPTS|INP_RECVRETOPTS|INP_RECVDSTADDR|\ 593 INP_RECVIF|INP_RECVTTL|INP_RECVTOS|\ 594 IN6P_PKTINFO|IN6P_HOPLIMIT|IN6P_HOPOPTS|\ 595 IN6P_DSTOPTS|IN6P_RTHDR|IN6P_RTHDRDSTOPTS|\ 596 IN6P_TCLASS|IN6P_AUTOFLOWLABEL|IN6P_RFC2292|\ 597 IN6P_MTU) 598 599 /* 600 * Flags for inp_flags2. 601 */ 602 #define INP_LLE_VALID 0x00000001 /* cached lle is valid */ 603 #define INP_RT_VALID 0x00000002 /* cached rtentry is valid */ 604 #define INP_PCBGROUPWILD 0x00000004 /* in pcbgroup wildcard list */ 605 #define INP_REUSEPORT 0x00000008 /* SO_REUSEPORT option is set */ 606 #define INP_FREED 0x00000010 /* inp itself is not valid */ 607 #define INP_REUSEADDR 0x00000020 /* SO_REUSEADDR option is set */ 608 #define INP_BINDMULTI 0x00000040 /* IP_BINDMULTI option is set */ 609 #define INP_RSS_BUCKET_SET 0x00000080 /* IP_RSS_LISTEN_BUCKET is set */ 610 #define INP_RECVFLOWID 0x00000100 /* populate recv datagram with flow info */ 611 #define INP_RECVRSSBUCKETID 0x00000200 /* populate recv datagram with bucket id */ 612 #define INP_RATE_LIMIT_CHANGED 0x00000400 /* rate limit needs attention */ 613 #define INP_ORIGDSTADDR 0x00000800 /* receive IP dst address/port */ 614 615 /* 616 * Flags passed to in_pcblookup*() functions. 617 */ 618 #define INPLOOKUP_WILDCARD 0x00000001 /* Allow wildcard sockets. */ 619 #define INPLOOKUP_RLOCKPCB 0x00000002 /* Return inpcb read-locked. */ 620 #define INPLOOKUP_WLOCKPCB 0x00000004 /* Return inpcb write-locked. */ 621 622 #define INPLOOKUP_MASK (INPLOOKUP_WILDCARD | INPLOOKUP_RLOCKPCB | \ 623 INPLOOKUP_WLOCKPCB) 624 625 #define sotoinpcb(so) ((struct inpcb *)(so)->so_pcb) 626 #define sotoin6pcb(so) sotoinpcb(so) /* for KAME src sync over BSD*'s */ 627 628 #define INP_SOCKAF(so) so->so_proto->pr_domain->dom_family 629 630 #define INP_CHECK_SOCKAF(so, af) (INP_SOCKAF(so) == af) 631 632 /* 633 * Constants for pcbinfo.ipi_hashfields. 634 */ 635 #define IPI_HASHFIELDS_NONE 0 636 #define IPI_HASHFIELDS_2TUPLE 1 637 #define IPI_HASHFIELDS_4TUPLE 2 638 639 #ifdef _KERNEL 640 VNET_DECLARE(int, ipport_reservedhigh); 641 VNET_DECLARE(int, ipport_reservedlow); 642 VNET_DECLARE(int, ipport_lowfirstauto); 643 VNET_DECLARE(int, ipport_lowlastauto); 644 VNET_DECLARE(int, ipport_firstauto); 645 VNET_DECLARE(int, ipport_lastauto); 646 VNET_DECLARE(int, ipport_hifirstauto); 647 VNET_DECLARE(int, ipport_hilastauto); 648 VNET_DECLARE(int, ipport_randomized); 649 VNET_DECLARE(int, ipport_randomcps); 650 VNET_DECLARE(int, ipport_randomtime); 651 VNET_DECLARE(int, ipport_stoprandom); 652 VNET_DECLARE(int, ipport_tcpallocs); 653 654 #define V_ipport_reservedhigh VNET(ipport_reservedhigh) 655 #define V_ipport_reservedlow VNET(ipport_reservedlow) 656 #define V_ipport_lowfirstauto VNET(ipport_lowfirstauto) 657 #define V_ipport_lowlastauto VNET(ipport_lowlastauto) 658 #define V_ipport_firstauto VNET(ipport_firstauto) 659 #define V_ipport_lastauto VNET(ipport_lastauto) 660 #define V_ipport_hifirstauto VNET(ipport_hifirstauto) 661 #define V_ipport_hilastauto VNET(ipport_hilastauto) 662 #define V_ipport_randomized VNET(ipport_randomized) 663 #define V_ipport_randomcps VNET(ipport_randomcps) 664 #define V_ipport_randomtime VNET(ipport_randomtime) 665 #define V_ipport_stoprandom VNET(ipport_stoprandom) 666 #define V_ipport_tcpallocs VNET(ipport_tcpallocs) 667 668 void in_pcbinfo_destroy(struct inpcbinfo *); 669 void in_pcbinfo_init(struct inpcbinfo *, const char *, struct inpcbhead *, 670 int, int, char *, uma_init, uma_fini, uint32_t, u_int); 671 672 int in_pcbbind_check_bindmulti(const struct inpcb *ni, 673 const struct inpcb *oi); 674 675 struct inpcbgroup * 676 in_pcbgroup_byhash(struct inpcbinfo *, u_int, uint32_t); 677 struct inpcbgroup * 678 in_pcbgroup_byinpcb(struct inpcb *); 679 struct inpcbgroup * 680 in_pcbgroup_bytuple(struct inpcbinfo *, struct in_addr, u_short, 681 struct in_addr, u_short); 682 void in_pcbgroup_destroy(struct inpcbinfo *); 683 int in_pcbgroup_enabled(struct inpcbinfo *); 684 void in_pcbgroup_init(struct inpcbinfo *, u_int, int); 685 void in_pcbgroup_remove(struct inpcb *); 686 void in_pcbgroup_update(struct inpcb *); 687 void in_pcbgroup_update_mbuf(struct inpcb *, struct mbuf *); 688 689 void in_pcbpurgeif0(struct inpcbinfo *, struct ifnet *); 690 int in_pcballoc(struct socket *, struct inpcbinfo *); 691 int in_pcbbind(struct inpcb *, struct sockaddr *, struct ucred *); 692 int in_pcb_lport(struct inpcb *, struct in_addr *, u_short *, 693 struct ucred *, int); 694 int in_pcbbind_setup(struct inpcb *, struct sockaddr *, in_addr_t *, 695 u_short *, struct ucred *); 696 int in_pcbconnect(struct inpcb *, struct sockaddr *, struct ucred *); 697 int in_pcbconnect_mbuf(struct inpcb *, struct sockaddr *, struct ucred *, 698 struct mbuf *); 699 int in_pcbconnect_setup(struct inpcb *, struct sockaddr *, in_addr_t *, 700 u_short *, in_addr_t *, u_short *, struct inpcb **, 701 struct ucred *); 702 void in_pcbdetach(struct inpcb *); 703 void in_pcbdisconnect(struct inpcb *); 704 void in_pcbdrop(struct inpcb *); 705 void in_pcbfree(struct inpcb *); 706 int in_pcbinshash(struct inpcb *); 707 int in_pcbinshash_nopcbgroup(struct inpcb *); 708 int in_pcbladdr(struct inpcb *, struct in_addr *, struct in_addr *, 709 struct ucred *); 710 struct inpcb * 711 in_pcblookup_local(struct inpcbinfo *, 712 struct in_addr, u_short, int, struct ucred *); 713 struct inpcb * 714 in_pcblookup(struct inpcbinfo *, struct in_addr, u_int, 715 struct in_addr, u_int, int, struct ifnet *); 716 struct inpcb * 717 in_pcblookup_mbuf(struct inpcbinfo *, struct in_addr, u_int, 718 struct in_addr, u_int, int, struct ifnet *, struct mbuf *); 719 void in_pcbnotifyall(struct inpcbinfo *pcbinfo, struct in_addr, 720 int, struct inpcb *(*)(struct inpcb *, int)); 721 void in_pcbref(struct inpcb *); 722 void in_pcbrehash(struct inpcb *); 723 void in_pcbrehash_mbuf(struct inpcb *, struct mbuf *); 724 int in_pcbrele(struct inpcb *); 725 int in_pcbrele_rlocked(struct inpcb *); 726 int in_pcbrele_wlocked(struct inpcb *); 727 void in_losing(struct inpcb *); 728 void in_pcbsetsolabel(struct socket *so); 729 int in_getpeeraddr(struct socket *so, struct sockaddr **nam); 730 int in_getsockaddr(struct socket *so, struct sockaddr **nam); 731 struct sockaddr * 732 in_sockaddr(in_port_t port, struct in_addr *addr); 733 void in_pcbsosetlabel(struct socket *so); 734 #ifdef RATELIMIT 735 int in_pcbattach_txrtlmt(struct inpcb *, struct ifnet *, uint32_t, uint32_t, uint32_t); 736 void in_pcbdetach_txrtlmt(struct inpcb *); 737 int in_pcbmodify_txrtlmt(struct inpcb *, uint32_t); 738 int in_pcbquery_txrtlmt(struct inpcb *, uint32_t *); 739 void in_pcboutput_txrtlmt(struct inpcb *, struct ifnet *, struct mbuf *); 740 void in_pcboutput_eagain(struct inpcb *); 741 #endif 742 #endif /* _KERNEL */ 743 744 #endif /* !_NETINET_IN_PCB_H_ */ 745