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
46 #ifdef _KERNEL
47 #include <sys/lock.h>
48 #include <sys/rwlock.h>
49 #include <net/vnet.h>
50 #include <vm/uma.h>
51 #endif
52
53 #define in6pcb inpcb /* for KAME src sync over BSD*'s */
54 #define in6p_sp inp_sp /* for KAME src sync over BSD*'s */
55 struct inpcbpolicy;
56
57 /*
58 * struct inpcb is the common protocol control block structure used in most
59 * IP transport protocols.
60 *
61 * Pointers to local and foreign host table entries, local and foreign socket
62 * numbers, and pointers up (to a socket structure) and down (to a
63 * protocol-specific control block) are stored here.
64 */
65 LIST_HEAD(inpcbhead, inpcb);
66 LIST_HEAD(inpcbporthead, inpcbport);
67 typedef u_quad_t inp_gen_t;
68
69 /*
70 * PCB with AF_INET6 null bind'ed laddr can receive AF_INET input packet.
71 * So, AF_INET6 null laddr is also used as AF_INET null laddr, by utilizing
72 * the following structure.
73 */
74 struct in_addr_4in6 {
75 u_int32_t ia46_pad32[3];
76 struct in_addr ia46_addr4;
77 };
78
79 /*
80 * NOTE: ipv6 addrs should be 64-bit aligned, per RFC 2553. in_conninfo has
81 * some extra padding to accomplish this.
82 * NOTE 2: tcp_syncache.c uses first 5 32-bit words, which identify fport,
83 * lport, faddr to generate hash, so these fields shouldn't be moved.
84 */
85 struct in_endpoints {
86 u_int16_t ie_fport; /* foreign port */
87 u_int16_t ie_lport; /* local port */
88 /* protocol dependent part, local and foreign addr */
89 union {
90 /* foreign host table entry */
91 struct in_addr_4in6 ie46_foreign;
92 struct in6_addr ie6_foreign;
93 } ie_dependfaddr;
94 union {
95 /* local host table entry */
96 struct in_addr_4in6 ie46_local;
97 struct in6_addr ie6_local;
98 } ie_dependladdr;
99 u_int32_t ie6_zoneid; /* scope zone id */
100 };
101 #define ie_faddr ie_dependfaddr.ie46_foreign.ia46_addr4
102 #define ie_laddr ie_dependladdr.ie46_local.ia46_addr4
103 #define ie6_faddr ie_dependfaddr.ie6_foreign
104 #define ie6_laddr ie_dependladdr.ie6_local
105
106 /*
107 * XXX The defines for inc_* are hacks and should be changed to direct
108 * references.
109 */
110 struct in_conninfo {
111 u_int8_t inc_flags;
112 u_int8_t inc_len;
113 u_int16_t inc_fibnum; /* XXX was pad, 16 bits is plenty */
114 /* protocol dependent part */
115 struct in_endpoints inc_ie;
116 };
117
118 /*
119 * Flags for inc_flags.
120 */
121 #define INC_ISIPV6 0x01
122
123 #define inc_isipv6 inc_flags /* temp compatability */
124 #define inc_fport inc_ie.ie_fport
125 #define inc_lport inc_ie.ie_lport
126 #define inc_faddr inc_ie.ie_faddr
127 #define inc_laddr inc_ie.ie_laddr
128 #define inc6_faddr inc_ie.ie6_faddr
129 #define inc6_laddr inc_ie.ie6_laddr
130 #define inc6_zoneid inc_ie.ie6_zoneid
131
132 struct icmp6_filter;
133
134 /*-
135 * struct inpcb captures the network layer state for TCP, UDP, and raw IPv4 and
136 * IPv6 sockets. In the case of TCP and UDP, further per-connection state is
137 * hung off of inp_ppcb most of the time. Almost all fields of struct inpcb
138 * are static after creation or protected by a per-inpcb rwlock, inp_lock. A
139 * few fields are protected by multiple locks as indicated in the locking notes
140 * below. For these fields, all of the listed locks must be write-locked for
141 * any modifications. However, these fields can be safely read while any one of
142 * the listed locks are read-locked. This model can permit greater concurrency
143 * for read operations. For example, connections can be looked up while only
144 * holding a read lock on the global pcblist lock. This is important for
145 * performance when attempting to find the connection for a packet given its IP
146 * and port tuple.
147 *
148 * One noteworthy exception is that the global pcbinfo lock follows a different
149 * set of rules in relation to the inp_list field. Rather than being
150 * write-locked for modifications and read-locked for list iterations, it must
151 * be read-locked during modifications and write-locked during list iterations.
152 * This ensures that the relatively rare global list iterations safely walk a
153 * stable snapshot of connections while allowing more common list modifications
154 * to safely grab the pcblist lock just while adding or removing a connection
155 * from the global list.
156 *
157 * Key:
158 * (a) - Atomically incremented
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 struct in_ifaddr *inp_ifaddr; /* (i) reference to the local ifaddr */
206 u_int inp_rt_gen; /* (a) generation count of routing entry */
207 void *inp_pspare[5]; /* (x) route caching / 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) route caching / user cookie /
211 * general use */
212
213 /* Local and foreign ports, local and foreign addr. */
214 struct in_conninfo inp_inc; /* (i) list for PCB's local port */
215
216 /* MAC and IPSEC policy information. */
217 struct label *inp_label; /* (i) MAC label */
218 struct inpcbpolicy *inp_sp; /* (s) for IPSEC */
219
220 /* Protocol-dependent part; options. */
221 struct {
222 u_char inp4_ip_tos; /* (i) type of service proto */
223 struct mbuf *inp4_options; /* (i) IP options */
224 struct ip_moptions *inp4_moptions; /* (i) IP mcast options */
225 } inp_depend4;
226 struct {
227 /* (i) IP options */
228 struct mbuf *inp6_options;
229 /* (i) IP6 options for outgoing packets */
230 struct ip6_pktopts *inp6_outputopts;
231 /* (i) IP multicast options */
232 struct ip6_moptions *inp6_moptions;
233 /* (i) ICMPv6 code type filter */
234 struct icmp6_filter *inp6_icmp6filt;
235 /* (i) IPV6_CHECKSUM setsockopt */
236 int inp6_cksum;
237 short inp6_hops;
238 } inp_depend6;
239 LIST_ENTRY(inpcb) inp_portlist; /* (i/h) */
240 struct inpcbport *inp_phd; /* (i/h) head of this list */
241 #define inp_zero_size offsetof(struct inpcb, inp_gencnt)
242 inp_gen_t inp_gencnt; /* (c) generation count */
243 struct rtentry *inp_rt; /* cached L3 information */
244 char *inp_prepend; /* cached L2 information */
245 uint16_t inp_plen;
246 struct rwlock inp_lock;
247 };
248 #define inp_fport inp_inc.inc_fport
249 #define inp_lport inp_inc.inc_lport
250 #define inp_faddr inp_inc.inc_faddr
251 #define inp_laddr inp_inc.inc_laddr
252 #define inp_ip_tos inp_depend4.inp4_ip_tos
253 #define inp_options inp_depend4.inp4_options
254 #define inp_moptions inp_depend4.inp4_moptions
255
256 #define in6p_faddr inp_inc.inc6_faddr
257 #define in6p_laddr inp_inc.inc6_laddr
258 #define in6p_zoneid inp_inc.inc6_zoneid
259 #define in6p_hops inp_depend6.inp6_hops /* default hop limit */
260 #define in6p_flowinfo inp_flow
261 #define in6p_options inp_depend6.inp6_options
262 #define in6p_outputopts inp_depend6.inp6_outputopts
263 #define in6p_moptions inp_depend6.inp6_moptions
264 #define in6p_icmp6filt inp_depend6.inp6_icmp6filt
265 #define in6p_cksum inp_depend6.inp6_cksum
266
267 #define inp_vnet inp_pcbinfo->ipi_vnet
268
269 /*
270 * The range of the generation count, as used in this implementation, is 9e19.
271 * We would have to create 300 billion connections per second for this number
272 * to roll over in a year. This seems sufficiently unlikely that we simply
273 * don't concern ourselves with that possibility.
274 */
275
276 /*
277 * Interface exported to userland by various protocols which use inpcbs. Hack
278 * alert -- only define if struct xsocket is in scope.
279 */
280 #ifdef _SYS_SOCKETVAR_H_
281 struct xinpcb {
282 size_t xi_len; /* length of this structure */
283 struct inpcb xi_inp;
284 struct xsocket xi_socket;
285 u_quad_t xi_alignment_hack;
286 };
287
288 struct xinpgen {
289 size_t xig_len; /* length of this structure */
290 u_int xig_count; /* number of PCBs at this time */
291 inp_gen_t xig_gen; /* generation count at this time */
292 so_gen_t xig_sogen; /* socket generation count at this time */
293 };
294 #endif /* _SYS_SOCKETVAR_H_ */
295
296 struct inpcbport {
297 LIST_ENTRY(inpcbport) phd_hash;
298 struct inpcbhead phd_pcblist;
299 u_short phd_port;
300 };
301
302 /*-
303 * Global data structure for each high-level protocol (UDP, TCP, ...) in both
304 * IPv4 and IPv6. Holds inpcb lists and information for managing them.
305 *
306 * Each pcbinfo is protected by three locks: ipi_lock, ipi_hash_lock and
307 * ipi_list_lock:
308 * - ipi_lock covering the global pcb list stability during loop iteration,
309 * - ipi_hash_lock covering the hashed lookup tables,
310 * - ipi_list_lock covering mutable global fields (such as the global
311 * pcb list)
312 *
313 * The lock order is:
314 *
315 * ipi_lock (before)
316 * inpcb locks (before)
317 * ipi_list locks (before)
318 * {ipi_hash_lock, pcbgroup locks}
319 *
320 * Locking key:
321 *
322 * (c) Constant or nearly constant after initialisation
323 * (g) Locked by ipi_lock
324 * (l) Locked by ipi_list_lock
325 * (h) Read using either ipi_hash_lock or inpcb lock; write requires both
326 * (p) Protected by one or more pcbgroup locks
327 * (x) Synchronisation properties poorly defined
328 */
329 struct inpcbinfo {
330 /*
331 * Global lock protecting full inpcb list traversal
332 */
333 struct rwlock ipi_lock;
334
335 /*
336 * Global list of inpcbs on the protocol.
337 */
338 struct inpcbhead *ipi_listhead; /* (g/l) */
339 u_int ipi_count; /* (l) */
340
341 /*
342 * Generation count -- incremented each time a connection is allocated
343 * or freed.
344 */
345 u_quad_t ipi_gencnt; /* (l) */
346
347 /*
348 * Fields associated with port lookup and allocation.
349 */
350 u_short ipi_lastport; /* (x) */
351 u_short ipi_lastlow; /* (x) */
352 u_short ipi_lasthi; /* (x) */
353
354 /*
355 * UMA zone from which inpcbs are allocated for this protocol.
356 */
357 struct uma_zone *ipi_zone; /* (c) */
358
359 /*
360 * Connection groups associated with this protocol. These fields are
361 * constant, but pcbgroup structures themselves are protected by
362 * per-pcbgroup locks.
363 */
364 struct inpcbgroup *ipi_pcbgroups; /* (c) */
365 u_int ipi_npcbgroups; /* (c) */
366 u_int ipi_hashfields; /* (c) */
367
368 /*
369 * Global lock protecting non-pcbgroup hash lookup tables.
370 */
371 struct rwlock ipi_hash_lock;
372
373 /*
374 * Global hash of inpcbs, hashed by local and foreign addresses and
375 * port numbers.
376 */
377 struct inpcbhead *ipi_hashbase; /* (h) */
378 u_long ipi_hashmask; /* (h) */
379
380 /*
381 * Global hash of inpcbs, hashed by only local port number.
382 */
383 struct inpcbporthead *ipi_porthashbase; /* (h) */
384 u_long ipi_porthashmask; /* (h) */
385
386 /*
387 * List of wildcard inpcbs for use with pcbgroups. In the past, was
388 * per-pcbgroup but is now global. All pcbgroup locks must be held
389 * to modify the list, so any is sufficient to read it.
390 */
391 struct inpcbhead *ipi_wildbase; /* (p) */
392 u_long ipi_wildmask; /* (p) */
393
394 /*
395 * Pointer to network stack instance
396 */
397 struct vnet *ipi_vnet; /* (c) */
398
399 /*
400 * general use 2
401 */
402 void *ipi_pspare[2];
403
404 /*
405 * Global lock protecting global inpcb list, inpcb count, etc.
406 */
407 struct rwlock ipi_list_lock;
408 };
409
410 #ifdef _KERNEL
411 /*
412 * Connection groups hold sets of connections that have similar CPU/thread
413 * affinity. Each connection belongs to exactly one connection group.
414 */
415 struct inpcbgroup {
416 /*
417 * Per-connection group hash of inpcbs, hashed by local and foreign
418 * addresses and port numbers.
419 */
420 struct inpcbhead *ipg_hashbase; /* (c) */
421 u_long ipg_hashmask; /* (c) */
422
423 /*
424 * Notional affinity of this pcbgroup.
425 */
426 u_int ipg_cpu; /* (p) */
427
428 /*
429 * Per-connection group lock, not to be confused with ipi_lock.
430 * Protects the hash table hung off the group, but also the global
431 * wildcard list in inpcbinfo.
432 */
433 struct mtx ipg_lock;
434 };
435
436 #define INP_LOCK_INIT(inp, d, t) \
437 rw_init_flags(&(inp)->inp_lock, (t), RW_RECURSE | RW_DUPOK)
438 #define INP_LOCK_DESTROY(inp) rw_destroy(&(inp)->inp_lock)
439 #define INP_RLOCK(inp) rw_rlock(&(inp)->inp_lock)
440 #define INP_WLOCK(inp) rw_wlock(&(inp)->inp_lock)
441 #define INP_TRY_RLOCK(inp) rw_try_rlock(&(inp)->inp_lock)
442 #define INP_TRY_WLOCK(inp) rw_try_wlock(&(inp)->inp_lock)
443 #define INP_RUNLOCK(inp) rw_runlock(&(inp)->inp_lock)
444 #define INP_WUNLOCK(inp) rw_wunlock(&(inp)->inp_lock)
445 #define INP_TRY_UPGRADE(inp) rw_try_upgrade(&(inp)->inp_lock)
446 #define INP_DOWNGRADE(inp) rw_downgrade(&(inp)->inp_lock)
447 #define INP_WLOCKED(inp) rw_wowned(&(inp)->inp_lock)
448 #define INP_LOCK_ASSERT(inp) rw_assert(&(inp)->inp_lock, RA_LOCKED)
449 #define INP_RLOCK_ASSERT(inp) rw_assert(&(inp)->inp_lock, RA_RLOCKED)
450 #define INP_WLOCK_ASSERT(inp) rw_assert(&(inp)->inp_lock, RA_WLOCKED)
451 #define INP_UNLOCK_ASSERT(inp) rw_assert(&(inp)->inp_lock, RA_UNLOCKED)
452
453 /*
454 * These locking functions are for inpcb consumers outside of sys/netinet,
455 * more specifically, they were added for the benefit of TOE drivers. The
456 * macros are reserved for use by the stack.
457 */
458 void inp_wlock(struct inpcb *);
459 void inp_wunlock(struct inpcb *);
460 void inp_rlock(struct inpcb *);
461 void inp_runlock(struct inpcb *);
462
463 #ifdef INVARIANTS
464 void inp_lock_assert(struct inpcb *);
465 void inp_unlock_assert(struct inpcb *);
466 #else
467 static __inline void
inp_lock_assert(struct inpcb * inp __unused)468 inp_lock_assert(struct inpcb *inp __unused)
469 {
470 }
471
472 static __inline void
inp_unlock_assert(struct inpcb * inp __unused)473 inp_unlock_assert(struct inpcb *inp __unused)
474 {
475 }
476
477 #endif
478
479 void inp_apply_all(void (*func)(struct inpcb *, void *), void *arg);
480 int inp_ip_tos_get(const struct inpcb *inp);
481 void inp_ip_tos_set(struct inpcb *inp, int val);
482 struct socket *
483 inp_inpcbtosocket(struct inpcb *inp);
484 struct tcpcb *
485 inp_inpcbtotcpcb(struct inpcb *inp);
486 void inp_4tuple_get(struct inpcb *inp, uint32_t *laddr, uint16_t *lp,
487 uint32_t *faddr, uint16_t *fp);
488 short inp_so_options(const struct inpcb *inp);
489
490 #endif /* _KERNEL */
491
492 #define INP_INFO_LOCK_INIT(ipi, d) \
493 rw_init_flags(&(ipi)->ipi_lock, (d), RW_RECURSE)
494 #define INP_INFO_LOCK_DESTROY(ipi) rw_destroy(&(ipi)->ipi_lock)
495 #define INP_INFO_RLOCK(ipi) rw_rlock(&(ipi)->ipi_lock)
496 #define INP_INFO_WLOCK(ipi) rw_wlock(&(ipi)->ipi_lock)
497 #define INP_INFO_TRY_RLOCK(ipi) rw_try_rlock(&(ipi)->ipi_lock)
498 #define INP_INFO_TRY_WLOCK(ipi) rw_try_wlock(&(ipi)->ipi_lock)
499 #define INP_INFO_TRY_UPGRADE(ipi) rw_try_upgrade(&(ipi)->ipi_lock)
500 #define INP_INFO_WLOCKED(ipi) rw_wowned(&(ipi)->ipi_lock)
501 #define INP_INFO_RUNLOCK(ipi) rw_runlock(&(ipi)->ipi_lock)
502 #define INP_INFO_WUNLOCK(ipi) rw_wunlock(&(ipi)->ipi_lock)
503 #define INP_INFO_LOCK_ASSERT(ipi) rw_assert(&(ipi)->ipi_lock, RA_LOCKED)
504 #define INP_INFO_RLOCK_ASSERT(ipi) rw_assert(&(ipi)->ipi_lock, RA_RLOCKED)
505 #define INP_INFO_WLOCK_ASSERT(ipi) rw_assert(&(ipi)->ipi_lock, RA_WLOCKED)
506 #define INP_INFO_UNLOCK_ASSERT(ipi) rw_assert(&(ipi)->ipi_lock, RA_UNLOCKED)
507
508 #define INP_LIST_LOCK_INIT(ipi, d) \
509 rw_init_flags(&(ipi)->ipi_list_lock, (d), 0)
510 #define INP_LIST_LOCK_DESTROY(ipi) rw_destroy(&(ipi)->ipi_list_lock)
511 #define INP_LIST_RLOCK(ipi) rw_rlock(&(ipi)->ipi_list_lock)
512 #define INP_LIST_WLOCK(ipi) rw_wlock(&(ipi)->ipi_list_lock)
513 #define INP_LIST_TRY_RLOCK(ipi) rw_try_rlock(&(ipi)->ipi_list_lock)
514 #define INP_LIST_TRY_WLOCK(ipi) rw_try_wlock(&(ipi)->ipi_list_lock)
515 #define INP_LIST_TRY_UPGRADE(ipi) rw_try_upgrade(&(ipi)->ipi_list_lock)
516 #define INP_LIST_RUNLOCK(ipi) rw_runlock(&(ipi)->ipi_list_lock)
517 #define INP_LIST_WUNLOCK(ipi) rw_wunlock(&(ipi)->ipi_list_lock)
518 #define INP_LIST_LOCK_ASSERT(ipi) \
519 rw_assert(&(ipi)->ipi_list_lock, RA_LOCKED)
520 #define INP_LIST_RLOCK_ASSERT(ipi) \
521 rw_assert(&(ipi)->ipi_list_lock, RA_RLOCKED)
522 #define INP_LIST_WLOCK_ASSERT(ipi) \
523 rw_assert(&(ipi)->ipi_list_lock, RA_WLOCKED)
524 #define INP_LIST_UNLOCK_ASSERT(ipi) \
525 rw_assert(&(ipi)->ipi_list_lock, RA_UNLOCKED)
526
527 #define INP_HASH_LOCK_INIT(ipi, d) \
528 rw_init_flags(&(ipi)->ipi_hash_lock, (d), 0)
529 #define INP_HASH_LOCK_DESTROY(ipi) rw_destroy(&(ipi)->ipi_hash_lock)
530 #define INP_HASH_RLOCK(ipi) rw_rlock(&(ipi)->ipi_hash_lock)
531 #define INP_HASH_WLOCK(ipi) rw_wlock(&(ipi)->ipi_hash_lock)
532 #define INP_HASH_RUNLOCK(ipi) rw_runlock(&(ipi)->ipi_hash_lock)
533 #define INP_HASH_WUNLOCK(ipi) rw_wunlock(&(ipi)->ipi_hash_lock)
534 #define INP_HASH_LOCK_ASSERT(ipi) rw_assert(&(ipi)->ipi_hash_lock, \
535 RA_LOCKED)
536 #define INP_HASH_WLOCK_ASSERT(ipi) rw_assert(&(ipi)->ipi_hash_lock, \
537 RA_WLOCKED)
538
539 #define INP_GROUP_LOCK_INIT(ipg, d) mtx_init(&(ipg)->ipg_lock, (d), NULL, \
540 MTX_DEF | MTX_DUPOK)
541 #define INP_GROUP_LOCK_DESTROY(ipg) mtx_destroy(&(ipg)->ipg_lock)
542
543 #define INP_GROUP_LOCK(ipg) mtx_lock(&(ipg)->ipg_lock)
544 #define INP_GROUP_LOCK_ASSERT(ipg) mtx_assert(&(ipg)->ipg_lock, MA_OWNED)
545 #define INP_GROUP_UNLOCK(ipg) mtx_unlock(&(ipg)->ipg_lock)
546
547 #define INP_PCBHASH(faddr, lport, fport, mask) \
548 (((faddr) ^ ((faddr) >> 16) ^ ntohs((lport) ^ (fport))) & (mask))
549 #define INP_PCBPORTHASH(lport, mask) \
550 (ntohs((lport)) & (mask))
551 #define INP6_PCBHASHKEY(faddr) ((faddr)->s6_addr32[3])
552
553 /*
554 * Flags for inp_vflags -- historically version flags only
555 */
556 #define INP_IPV4 0x1
557 #define INP_IPV6 0x2
558 #define INP_IPV6PROTO 0x4 /* opened under IPv6 protocol */
559
560 /*
561 * Flags for inp_flags.
562 */
563 #define INP_RECVOPTS 0x00000001 /* receive incoming IP options */
564 #define INP_RECVRETOPTS 0x00000002 /* receive IP options for reply */
565 #define INP_RECVDSTADDR 0x00000004 /* receive IP dst address */
566 #define INP_HDRINCL 0x00000008 /* user supplies entire IP header */
567 #define INP_HIGHPORT 0x00000010 /* user wants "high" port binding */
568 #define INP_LOWPORT 0x00000020 /* user wants "low" port binding */
569 #define INP_ANONPORT 0x00000040 /* port chosen for user */
570 #define INP_RECVIF 0x00000080 /* receive incoming interface */
571 #define INP_MTUDISC 0x00000100 /* user can do MTU discovery */
572 /* 0x000200 unused: was INP_FAITH */
573 #define INP_RECVTTL 0x00000400 /* receive incoming IP TTL */
574 #define INP_DONTFRAG 0x00000800 /* don't fragment packet */
575 #define INP_BINDANY 0x00001000 /* allow bind to any address */
576 #define INP_INHASHLIST 0x00002000 /* in_pcbinshash() has been called */
577 #define INP_RECVTOS 0x00004000 /* receive incoming IP TOS */
578 #define IN6P_IPV6_V6ONLY 0x00008000 /* restrict AF_INET6 socket for v6 */
579 #define IN6P_PKTINFO 0x00010000 /* receive IP6 dst and I/F */
580 #define IN6P_HOPLIMIT 0x00020000 /* receive hoplimit */
581 #define IN6P_HOPOPTS 0x00040000 /* receive hop-by-hop options */
582 #define IN6P_DSTOPTS 0x00080000 /* receive dst options after rthdr */
583 #define IN6P_RTHDR 0x00100000 /* receive routing header */
584 #define IN6P_RTHDRDSTOPTS 0x00200000 /* receive dstoptions before rthdr */
585 #define IN6P_TCLASS 0x00400000 /* receive traffic class value */
586 #define IN6P_AUTOFLOWLABEL 0x00800000 /* attach flowlabel automatically */
587 #define INP_TIMEWAIT 0x01000000 /* in TIMEWAIT, ppcb is tcptw */
588 #define INP_ONESBCAST 0x02000000 /* send all-ones broadcast */
589 #define INP_DROPPED 0x04000000 /* protocol drop flag */
590 #define INP_SOCKREF 0x08000000 /* strong socket reference */
591 #define INP_RESERVED_0 0x10000000 /* reserved field */
592 #define INP_RESERVED_1 0x20000000 /* reserved field */
593 #define IN6P_RFC2292 0x40000000 /* used RFC2292 API on the socket */
594 #define IN6P_MTU 0x80000000 /* receive path MTU */
595
596 #define INP_CONTROLOPTS (INP_RECVOPTS|INP_RECVRETOPTS|INP_RECVDSTADDR|\
597 INP_RECVIF|INP_RECVTTL|INP_RECVTOS|\
598 IN6P_PKTINFO|IN6P_HOPLIMIT|IN6P_HOPOPTS|\
599 IN6P_DSTOPTS|IN6P_RTHDR|IN6P_RTHDRDSTOPTS|\
600 IN6P_TCLASS|IN6P_AUTOFLOWLABEL|IN6P_RFC2292|\
601 IN6P_MTU)
602
603 /*
604 * Flags for inp_flags2.
605 */
606 #define INP_PCBGROUPWILD 0x00000004 /* in pcbgroup wildcard list */
607 #define INP_REUSEPORT 0x00000008 /* SO_REUSEPORT option is set */
608 #define INP_FREED 0x00000010 /* inp itself is not valid */
609 #define INP_REUSEADDR 0x00000020 /* SO_REUSEADDR option is set */
610 #define INP_BINDMULTI 0x00000040 /* IP_BINDMULTI option is set */
611 #define INP_RSS_BUCKET_SET 0x00000080 /* IP_RSS_LISTEN_BUCKET is set */
612 #define INP_RECVFLOWID 0x00000100 /* populate recv datagram with flow info */
613 #define INP_RECVRSSBUCKETID 0x00000200 /* populate recv datagram with bucket id */
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_pcbrtalloc(struct inpcb *inp);
728 int in_rt_valid(struct inpcb *inp);
729 void in_pcbsetsolabel(struct socket *so);
730 int in_getpeeraddr(struct socket *so, struct sockaddr **nam);
731 int in_getsockaddr(struct socket *so, struct sockaddr **nam);
732 struct sockaddr *
733 in_sockaddr(in_port_t port, struct in_addr *addr);
734 void in_pcbsosetlabel(struct socket *so);
735 #endif /* _KERNEL */
736
737 #endif /* !_NETINET_IN_PCB_H_ */
738