xref: /freebsd/sys/netinet/in_pcb.h (revision 5b9c547c)
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  */
83 struct in_endpoints {
84 	u_int16_t	ie_fport;		/* foreign port */
85 	u_int16_t	ie_lport;		/* local port */
86 	/* protocol dependent part, local and foreign addr */
87 	union {
88 		/* foreign host table entry */
89 		struct	in_addr_4in6 ie46_foreign;
90 		struct	in6_addr ie6_foreign;
91 	} ie_dependfaddr;
92 	union {
93 		/* local host table entry */
94 		struct	in_addr_4in6 ie46_local;
95 		struct	in6_addr ie6_local;
96 	} ie_dependladdr;
97 	u_int32_t	ie6_zoneid;		/* scope zone id */
98 };
99 #define	ie_faddr	ie_dependfaddr.ie46_foreign.ia46_addr4
100 #define	ie_laddr	ie_dependladdr.ie46_local.ia46_addr4
101 #define	ie6_faddr	ie_dependfaddr.ie6_foreign
102 #define	ie6_laddr	ie_dependladdr.ie6_local
103 
104 /*
105  * XXX The defines for inc_* are hacks and should be changed to direct
106  * references.
107  */
108 struct in_conninfo {
109 	u_int8_t	inc_flags;
110 	u_int8_t	inc_len;
111 	u_int16_t	inc_fibnum;	/* XXX was pad, 16 bits is plenty */
112 	/* protocol dependent part */
113 	struct	in_endpoints inc_ie;
114 };
115 
116 /*
117  * Flags for inc_flags.
118  */
119 #define	INC_ISIPV6	0x01
120 
121 #define inc_isipv6	inc_flags	/* temp compatability */
122 #define	inc_fport	inc_ie.ie_fport
123 #define	inc_lport	inc_ie.ie_lport
124 #define	inc_faddr	inc_ie.ie_faddr
125 #define	inc_laddr	inc_ie.ie_laddr
126 #define	inc6_faddr	inc_ie.ie6_faddr
127 #define	inc6_laddr	inc_ie.ie6_laddr
128 #define	inc6_zoneid	inc_ie.ie6_zoneid
129 
130 struct	icmp6_filter;
131 
132 /*-
133  * struct inpcb captures the network layer state for TCP, UDP, and raw IPv4
134  * and IPv6 sockets.  In the case of TCP, further per-connection state is
135  * hung off of inp_ppcb most of the time.  Almost all fields of struct inpcb
136  * are static after creation or protected by a per-inpcb rwlock, inp_lock.  A
137  * few fields also require the global pcbinfo lock for the inpcb to be held,
138  * when modified, such as the global connection lists and hashes, as well as
139  * binding information (which affects which hash a connection is on).  This
140  * model means that connections can be looked up without holding the
141  * per-connection lock, which is important for performance when attempting to
142  * find the connection for a packet given its IP and port tuple.  Writing to
143  * these fields that write locks be held on both the inpcb and global locks.
144  *
145  * Key:
146  * (c) - Constant after initialization
147  * (g) - Protected by the pcbgroup lock
148  * (i) - Protected by the inpcb lock
149  * (p) - Protected by the pcbinfo lock for the inpcb
150  * (s) - Protected by another subsystem's locks
151  * (x) - Undefined locking
152  *
153  * A few other notes:
154  *
155  * When a read lock is held, stability of the field is guaranteed; to write
156  * to a field, a write lock must generally be held.
157  *
158  * netinet/netinet6-layer code should not assume that the inp_socket pointer
159  * is safe to dereference without inp_lock being held, even for protocols
160  * other than TCP (where the inpcb persists during TIMEWAIT even after the
161  * socket has been freed), or there may be close(2)-related races.
162  *
163  * The inp_vflag field is overloaded, and would otherwise ideally be (c).
164  */
165 struct inpcb {
166 	LIST_ENTRY(inpcb) inp_hash;	/* (i/p) hash list */
167 	LIST_ENTRY(inpcb) inp_pcbgrouphash;	/* (g/i) hash list */
168 	LIST_ENTRY(inpcb) inp_list;	/* (i/p) list for all PCBs for proto */
169 	void	*inp_ppcb;		/* (i) pointer to per-protocol pcb */
170 	struct	inpcbinfo *inp_pcbinfo;	/* (c) PCB list info */
171 	struct	inpcbgroup *inp_pcbgroup; /* (g/i) PCB group list */
172 	LIST_ENTRY(inpcb) inp_pcbgroup_wild; /* (g/i/p) group wildcard entry */
173 	struct	socket *inp_socket;	/* (i) back pointer to socket */
174 	struct	ucred	*inp_cred;	/* (c) cache of socket cred */
175 	u_int32_t inp_flow;		/* (i) IPv6 flow information */
176 	int	inp_flags;		/* (i) generic IP/datagram flags */
177 	int	inp_flags2;		/* (i) generic IP/datagram flags #2*/
178 	u_char	inp_vflag;		/* (i) IP version flag (v4/v6) */
179 	u_char	inp_ip_ttl;		/* (i) time to live proto */
180 	u_char	inp_ip_p;		/* (c) protocol proto */
181 	u_char	inp_ip_minttl;		/* (i) minimum TTL or drop */
182 	uint32_t inp_flowid;		/* (x) flow id / queue id */
183 	u_int	inp_refcount;		/* (i) refcount */
184 	void	*inp_pspare[5];		/* (x) route caching / general use */
185 	uint32_t inp_flowtype;		/* (x) M_HASHTYPE value */
186 	uint32_t inp_rss_listen_bucket;	/* (x) overridden RSS listen bucket */
187 	u_int	inp_ispare[4];		/* (x) route caching / user cookie /
188 					 *     general use */
189 
190 	/* Local and foreign ports, local and foreign addr. */
191 	struct	in_conninfo inp_inc;	/* (i/p) list for PCB's local port */
192 
193 	/* MAC and IPSEC policy information. */
194 	struct	label *inp_label;	/* (i) MAC label */
195 	struct	inpcbpolicy *inp_sp;    /* (s) for IPSEC */
196 
197 	/* Protocol-dependent part; options. */
198 	struct {
199 		u_char	inp4_ip_tos;		/* (i) type of service proto */
200 		struct	mbuf *inp4_options;	/* (i) IP options */
201 		struct	ip_moptions *inp4_moptions; /* (i) IP mcast options */
202 	} inp_depend4;
203 	struct {
204 		/* (i) IP options */
205 		struct	mbuf *inp6_options;
206 		/* (i) IP6 options for outgoing packets */
207 		struct	ip6_pktopts *inp6_outputopts;
208 		/* (i) IP multicast options */
209 		struct	ip6_moptions *inp6_moptions;
210 		/* (i) ICMPv6 code type filter */
211 		struct	icmp6_filter *inp6_icmp6filt;
212 		/* (i) IPV6_CHECKSUM setsockopt */
213 		int	inp6_cksum;
214 		short	inp6_hops;
215 	} inp_depend6;
216 	LIST_ENTRY(inpcb) inp_portlist;	/* (i/p) */
217 	struct	inpcbport *inp_phd;	/* (i/p) head of this list */
218 #define inp_zero_size offsetof(struct inpcb, inp_gencnt)
219 	inp_gen_t	inp_gencnt;	/* (c) generation count */
220 	struct llentry	*inp_lle;	/* cached L2 information */
221 	struct rtentry	*inp_rt;	/* cached L3 information */
222 	struct rwlock	inp_lock;
223 };
224 #define	inp_fport	inp_inc.inc_fport
225 #define	inp_lport	inp_inc.inc_lport
226 #define	inp_faddr	inp_inc.inc_faddr
227 #define	inp_laddr	inp_inc.inc_laddr
228 #define	inp_ip_tos	inp_depend4.inp4_ip_tos
229 #define	inp_options	inp_depend4.inp4_options
230 #define	inp_moptions	inp_depend4.inp4_moptions
231 
232 #define	in6p_faddr	inp_inc.inc6_faddr
233 #define	in6p_laddr	inp_inc.inc6_laddr
234 #define	in6p_zoneid	inp_inc.inc6_zoneid
235 #define	in6p_hops	inp_depend6.inp6_hops	/* default hop limit */
236 #define	in6p_flowinfo	inp_flow
237 #define	in6p_options	inp_depend6.inp6_options
238 #define	in6p_outputopts	inp_depend6.inp6_outputopts
239 #define	in6p_moptions	inp_depend6.inp6_moptions
240 #define	in6p_icmp6filt	inp_depend6.inp6_icmp6filt
241 #define	in6p_cksum	inp_depend6.inp6_cksum
242 
243 #define	inp_vnet	inp_pcbinfo->ipi_vnet
244 
245 /*
246  * The range of the generation count, as used in this implementation, is 9e19.
247  * We would have to create 300 billion connections per second for this number
248  * to roll over in a year.  This seems sufficiently unlikely that we simply
249  * don't concern ourselves with that possibility.
250  */
251 
252 /*
253  * Interface exported to userland by various protocols which use inpcbs.  Hack
254  * alert -- only define if struct xsocket is in scope.
255  */
256 #ifdef _SYS_SOCKETVAR_H_
257 struct	xinpcb {
258 	size_t	xi_len;		/* length of this structure */
259 	struct	inpcb xi_inp;
260 	struct	xsocket xi_socket;
261 	u_quad_t	xi_alignment_hack;
262 };
263 
264 struct	xinpgen {
265 	size_t	xig_len;	/* length of this structure */
266 	u_int	xig_count;	/* number of PCBs at this time */
267 	inp_gen_t xig_gen;	/* generation count at this time */
268 	so_gen_t xig_sogen;	/* socket generation count at this time */
269 };
270 #endif /* _SYS_SOCKETVAR_H_ */
271 
272 struct inpcbport {
273 	LIST_ENTRY(inpcbport) phd_hash;
274 	struct inpcbhead phd_pcblist;
275 	u_short phd_port;
276 };
277 
278 /*-
279  * Global data structure for each high-level protocol (UDP, TCP, ...) in both
280  * IPv4 and IPv6.  Holds inpcb lists and information for managing them.
281  *
282  * Each pcbinfo is protected by two locks: ipi_lock and ipi_hash_lock,
283  * the former covering mutable global fields (such as the global pcb list),
284  * and the latter covering the hashed lookup tables.  The lock order is:
285  *
286  *    ipi_lock (before) inpcb locks (before) {ipi_hash_lock, pcbgroup locks}
287  *
288  * Locking key:
289  *
290  * (c) Constant or nearly constant after initialisation
291  * (g) Locked by ipi_lock
292  * (h) Read using either ipi_hash_lock or inpcb lock; write requires both
293  * (p) Protected by one or more pcbgroup locks
294  * (x) Synchronisation properties poorly defined
295  */
296 struct inpcbinfo {
297 	/*
298 	 * Global lock protecting global inpcb list, inpcb count, etc.
299 	 */
300 	struct rwlock		 ipi_lock;
301 
302 	/*
303 	 * Global list of inpcbs on the protocol.
304 	 */
305 	struct inpcbhead	*ipi_listhead;		/* (g) */
306 	u_int			 ipi_count;		/* (g) */
307 
308 	/*
309 	 * Generation count -- incremented each time a connection is allocated
310 	 * or freed.
311 	 */
312 	u_quad_t		 ipi_gencnt;		/* (g) */
313 
314 	/*
315 	 * Fields associated with port lookup and allocation.
316 	 */
317 	u_short			 ipi_lastport;		/* (x) */
318 	u_short			 ipi_lastlow;		/* (x) */
319 	u_short			 ipi_lasthi;		/* (x) */
320 
321 	/*
322 	 * UMA zone from which inpcbs are allocated for this protocol.
323 	 */
324 	struct	uma_zone	*ipi_zone;		/* (c) */
325 
326 	/*
327 	 * Connection groups associated with this protocol.  These fields are
328 	 * constant, but pcbgroup structures themselves are protected by
329 	 * per-pcbgroup locks.
330 	 */
331 	struct inpcbgroup	*ipi_pcbgroups;		/* (c) */
332 	u_int			 ipi_npcbgroups;	/* (c) */
333 	u_int			 ipi_hashfields;	/* (c) */
334 
335 	/*
336 	 * Global lock protecting non-pcbgroup hash lookup tables.
337 	 */
338 	struct rwlock		 ipi_hash_lock;
339 
340 	/*
341 	 * Global hash of inpcbs, hashed by local and foreign addresses and
342 	 * port numbers.
343 	 */
344 	struct inpcbhead	*ipi_hashbase;		/* (h) */
345 	u_long			 ipi_hashmask;		/* (h) */
346 
347 	/*
348 	 * Global hash of inpcbs, hashed by only local port number.
349 	 */
350 	struct inpcbporthead	*ipi_porthashbase;	/* (h) */
351 	u_long			 ipi_porthashmask;	/* (h) */
352 
353 	/*
354 	 * List of wildcard inpcbs for use with pcbgroups.  In the past, was
355 	 * per-pcbgroup but is now global.  All pcbgroup locks must be held
356 	 * to modify the list, so any is sufficient to read it.
357 	 */
358 	struct inpcbhead	*ipi_wildbase;		/* (p) */
359 	u_long			 ipi_wildmask;		/* (p) */
360 
361 	/*
362 	 * Pointer to network stack instance
363 	 */
364 	struct vnet		*ipi_vnet;		/* (c) */
365 
366 	/*
367 	 * general use 2
368 	 */
369 	void 			*ipi_pspare[2];
370 };
371 
372 #ifdef _KERNEL
373 /*
374  * Connection groups hold sets of connections that have similar CPU/thread
375  * affinity.  Each connection belongs to exactly one connection group.
376  */
377 struct inpcbgroup {
378 	/*
379 	 * Per-connection group hash of inpcbs, hashed by local and foreign
380 	 * addresses and port numbers.
381 	 */
382 	struct inpcbhead	*ipg_hashbase;		/* (c) */
383 	u_long			 ipg_hashmask;		/* (c) */
384 
385 	/*
386 	 * Notional affinity of this pcbgroup.
387 	 */
388 	u_int			 ipg_cpu;		/* (p) */
389 
390 	/*
391 	 * Per-connection group lock, not to be confused with ipi_lock.
392 	 * Protects the hash table hung off the group, but also the global
393 	 * wildcard list in inpcbinfo.
394 	 */
395 	struct mtx		 ipg_lock;
396 } __aligned(CACHE_LINE_SIZE);
397 
398 #define INP_LOCK_INIT(inp, d, t) \
399 	rw_init_flags(&(inp)->inp_lock, (t), RW_RECURSE |  RW_DUPOK)
400 #define INP_LOCK_DESTROY(inp)	rw_destroy(&(inp)->inp_lock)
401 #define INP_RLOCK(inp)		rw_rlock(&(inp)->inp_lock)
402 #define INP_WLOCK(inp)		rw_wlock(&(inp)->inp_lock)
403 #define INP_TRY_RLOCK(inp)	rw_try_rlock(&(inp)->inp_lock)
404 #define INP_TRY_WLOCK(inp)	rw_try_wlock(&(inp)->inp_lock)
405 #define INP_RUNLOCK(inp)	rw_runlock(&(inp)->inp_lock)
406 #define INP_WUNLOCK(inp)	rw_wunlock(&(inp)->inp_lock)
407 #define	INP_TRY_UPGRADE(inp)	rw_try_upgrade(&(inp)->inp_lock)
408 #define	INP_DOWNGRADE(inp)	rw_downgrade(&(inp)->inp_lock)
409 #define	INP_WLOCKED(inp)	rw_wowned(&(inp)->inp_lock)
410 #define	INP_LOCK_ASSERT(inp)	rw_assert(&(inp)->inp_lock, RA_LOCKED)
411 #define	INP_RLOCK_ASSERT(inp)	rw_assert(&(inp)->inp_lock, RA_RLOCKED)
412 #define	INP_WLOCK_ASSERT(inp)	rw_assert(&(inp)->inp_lock, RA_WLOCKED)
413 #define	INP_UNLOCK_ASSERT(inp)	rw_assert(&(inp)->inp_lock, RA_UNLOCKED)
414 
415 /*
416  * These locking functions are for inpcb consumers outside of sys/netinet,
417  * more specifically, they were added for the benefit of TOE drivers. The
418  * macros are reserved for use by the stack.
419  */
420 void inp_wlock(struct inpcb *);
421 void inp_wunlock(struct inpcb *);
422 void inp_rlock(struct inpcb *);
423 void inp_runlock(struct inpcb *);
424 
425 #ifdef INVARIANTS
426 void inp_lock_assert(struct inpcb *);
427 void inp_unlock_assert(struct inpcb *);
428 #else
429 static __inline void
430 inp_lock_assert(struct inpcb *inp __unused)
431 {
432 }
433 
434 static __inline void
435 inp_unlock_assert(struct inpcb *inp __unused)
436 {
437 }
438 
439 #endif
440 
441 void	inp_apply_all(void (*func)(struct inpcb *, void *), void *arg);
442 int 	inp_ip_tos_get(const struct inpcb *inp);
443 void 	inp_ip_tos_set(struct inpcb *inp, int val);
444 struct socket *
445 	inp_inpcbtosocket(struct inpcb *inp);
446 struct tcpcb *
447 	inp_inpcbtotcpcb(struct inpcb *inp);
448 void 	inp_4tuple_get(struct inpcb *inp, uint32_t *laddr, uint16_t *lp,
449 		uint32_t *faddr, uint16_t *fp);
450 short	inp_so_options(const struct inpcb *inp);
451 
452 #endif /* _KERNEL */
453 
454 #define INP_INFO_LOCK_INIT(ipi, d) \
455 	rw_init_flags(&(ipi)->ipi_lock, (d), RW_RECURSE)
456 #define INP_INFO_LOCK_DESTROY(ipi)  rw_destroy(&(ipi)->ipi_lock)
457 #define INP_INFO_RLOCK(ipi)	rw_rlock(&(ipi)->ipi_lock)
458 #define INP_INFO_WLOCK(ipi)	rw_wlock(&(ipi)->ipi_lock)
459 #define INP_INFO_TRY_RLOCK(ipi)	rw_try_rlock(&(ipi)->ipi_lock)
460 #define INP_INFO_TRY_WLOCK(ipi)	rw_try_wlock(&(ipi)->ipi_lock)
461 #define INP_INFO_TRY_UPGRADE(ipi)	rw_try_upgrade(&(ipi)->ipi_lock)
462 #define INP_INFO_RUNLOCK(ipi)	rw_runlock(&(ipi)->ipi_lock)
463 #define INP_INFO_WUNLOCK(ipi)	rw_wunlock(&(ipi)->ipi_lock)
464 #define	INP_INFO_LOCK_ASSERT(ipi)	rw_assert(&(ipi)->ipi_lock, RA_LOCKED)
465 #define INP_INFO_RLOCK_ASSERT(ipi)	rw_assert(&(ipi)->ipi_lock, RA_RLOCKED)
466 #define INP_INFO_WLOCK_ASSERT(ipi)	rw_assert(&(ipi)->ipi_lock, RA_WLOCKED)
467 #define INP_INFO_UNLOCK_ASSERT(ipi)	rw_assert(&(ipi)->ipi_lock, RA_UNLOCKED)
468 
469 #define	INP_HASH_LOCK_INIT(ipi, d) \
470 	rw_init_flags(&(ipi)->ipi_hash_lock, (d), 0)
471 #define	INP_HASH_LOCK_DESTROY(ipi)	rw_destroy(&(ipi)->ipi_hash_lock)
472 #define	INP_HASH_RLOCK(ipi)		rw_rlock(&(ipi)->ipi_hash_lock)
473 #define	INP_HASH_WLOCK(ipi)		rw_wlock(&(ipi)->ipi_hash_lock)
474 #define	INP_HASH_RUNLOCK(ipi)		rw_runlock(&(ipi)->ipi_hash_lock)
475 #define	INP_HASH_WUNLOCK(ipi)		rw_wunlock(&(ipi)->ipi_hash_lock)
476 #define	INP_HASH_LOCK_ASSERT(ipi)	rw_assert(&(ipi)->ipi_hash_lock, \
477 					    RA_LOCKED)
478 #define	INP_HASH_WLOCK_ASSERT(ipi)	rw_assert(&(ipi)->ipi_hash_lock, \
479 					    RA_WLOCKED)
480 
481 #define	INP_GROUP_LOCK_INIT(ipg, d)	mtx_init(&(ipg)->ipg_lock, (d), NULL, \
482 					    MTX_DEF | MTX_DUPOK)
483 #define	INP_GROUP_LOCK_DESTROY(ipg)	mtx_destroy(&(ipg)->ipg_lock)
484 
485 #define	INP_GROUP_LOCK(ipg)		mtx_lock(&(ipg)->ipg_lock)
486 #define	INP_GROUP_LOCK_ASSERT(ipg)	mtx_assert(&(ipg)->ipg_lock, MA_OWNED)
487 #define	INP_GROUP_UNLOCK(ipg)		mtx_unlock(&(ipg)->ipg_lock)
488 
489 #define INP_PCBHASH(faddr, lport, fport, mask) \
490 	(((faddr) ^ ((faddr) >> 16) ^ ntohs((lport) ^ (fport))) & (mask))
491 #define INP_PCBPORTHASH(lport, mask) \
492 	(ntohs((lport)) & (mask))
493 #define	INP6_PCBHASHKEY(faddr)	((faddr)->s6_addr32[3])
494 
495 /*
496  * Flags for inp_vflags -- historically version flags only
497  */
498 #define	INP_IPV4	0x1
499 #define	INP_IPV6	0x2
500 #define	INP_IPV6PROTO	0x4		/* opened under IPv6 protocol */
501 
502 /*
503  * Flags for inp_flags.
504  */
505 #define	INP_RECVOPTS		0x00000001 /* receive incoming IP options */
506 #define	INP_RECVRETOPTS		0x00000002 /* receive IP options for reply */
507 #define	INP_RECVDSTADDR		0x00000004 /* receive IP dst address */
508 #define	INP_HDRINCL		0x00000008 /* user supplies entire IP header */
509 #define	INP_HIGHPORT		0x00000010 /* user wants "high" port binding */
510 #define	INP_LOWPORT		0x00000020 /* user wants "low" port binding */
511 #define	INP_ANONPORT		0x00000040 /* port chosen for user */
512 #define	INP_RECVIF		0x00000080 /* receive incoming interface */
513 #define	INP_MTUDISC		0x00000100 /* user can do MTU discovery */
514 				   	   /* 0x000200 unused: was INP_FAITH */
515 #define	INP_RECVTTL		0x00000400 /* receive incoming IP TTL */
516 #define	INP_DONTFRAG		0x00000800 /* don't fragment packet */
517 #define	INP_BINDANY		0x00001000 /* allow bind to any address */
518 #define	INP_INHASHLIST		0x00002000 /* in_pcbinshash() has been called */
519 #define	INP_RECVTOS		0x00004000 /* receive incoming IP TOS */
520 #define	IN6P_IPV6_V6ONLY	0x00008000 /* restrict AF_INET6 socket for v6 */
521 #define	IN6P_PKTINFO		0x00010000 /* receive IP6 dst and I/F */
522 #define	IN6P_HOPLIMIT		0x00020000 /* receive hoplimit */
523 #define	IN6P_HOPOPTS		0x00040000 /* receive hop-by-hop options */
524 #define	IN6P_DSTOPTS		0x00080000 /* receive dst options after rthdr */
525 #define	IN6P_RTHDR		0x00100000 /* receive routing header */
526 #define	IN6P_RTHDRDSTOPTS	0x00200000 /* receive dstoptions before rthdr */
527 #define	IN6P_TCLASS		0x00400000 /* receive traffic class value */
528 #define	IN6P_AUTOFLOWLABEL	0x00800000 /* attach flowlabel automatically */
529 #define	INP_TIMEWAIT		0x01000000 /* in TIMEWAIT, ppcb is tcptw */
530 #define	INP_ONESBCAST		0x02000000 /* send all-ones broadcast */
531 #define	INP_DROPPED		0x04000000 /* protocol drop flag */
532 #define	INP_SOCKREF		0x08000000 /* strong socket reference */
533 #define	INP_RESERVED_0          0x10000000 /* reserved field */
534 #define	INP_RESERVED_1          0x20000000 /* reserved field */
535 #define	IN6P_RFC2292		0x40000000 /* used RFC2292 API on the socket */
536 #define	IN6P_MTU		0x80000000 /* receive path MTU */
537 
538 #define	INP_CONTROLOPTS		(INP_RECVOPTS|INP_RECVRETOPTS|INP_RECVDSTADDR|\
539 				 INP_RECVIF|INP_RECVTTL|INP_RECVTOS|\
540 				 IN6P_PKTINFO|IN6P_HOPLIMIT|IN6P_HOPOPTS|\
541 				 IN6P_DSTOPTS|IN6P_RTHDR|IN6P_RTHDRDSTOPTS|\
542 				 IN6P_TCLASS|IN6P_AUTOFLOWLABEL|IN6P_RFC2292|\
543 				 IN6P_MTU)
544 
545 /*
546  * Flags for inp_flags2.
547  */
548 #define	INP_LLE_VALID		0x00000001 /* cached lle is valid */
549 #define	INP_RT_VALID		0x00000002 /* cached rtentry is valid */
550 #define	INP_PCBGROUPWILD	0x00000004 /* in pcbgroup wildcard list */
551 #define	INP_REUSEPORT		0x00000008 /* SO_REUSEPORT option is set */
552 #define	INP_FREED		0x00000010 /* inp itself is not valid */
553 #define	INP_REUSEADDR		0x00000020 /* SO_REUSEADDR option is set */
554 #define	INP_BINDMULTI		0x00000040 /* IP_BINDMULTI option is set */
555 #define	INP_RSS_BUCKET_SET	0x00000080 /* IP_RSS_LISTEN_BUCKET is set */
556 #define	INP_RECVFLOWID		0x00000100 /* populate recv datagram with flow info */
557 #define	INP_RECVRSSBUCKETID	0x00000200 /* populate recv datagram with bucket id */
558 
559 /*
560  * Flags passed to in_pcblookup*() functions.
561  */
562 #define	INPLOOKUP_WILDCARD	0x00000001	/* Allow wildcard sockets. */
563 #define	INPLOOKUP_RLOCKPCB	0x00000002	/* Return inpcb read-locked. */
564 #define	INPLOOKUP_WLOCKPCB	0x00000004	/* Return inpcb write-locked. */
565 
566 #define	INPLOOKUP_MASK	(INPLOOKUP_WILDCARD | INPLOOKUP_RLOCKPCB | \
567 			    INPLOOKUP_WLOCKPCB)
568 
569 #define	sotoinpcb(so)	((struct inpcb *)(so)->so_pcb)
570 #define	sotoin6pcb(so)	sotoinpcb(so) /* for KAME src sync over BSD*'s */
571 
572 #define	INP_SOCKAF(so) so->so_proto->pr_domain->dom_family
573 
574 #define	INP_CHECK_SOCKAF(so, af)	(INP_SOCKAF(so) == af)
575 
576 /*
577  * Constants for pcbinfo.ipi_hashfields.
578  */
579 #define	IPI_HASHFIELDS_NONE	0
580 #define	IPI_HASHFIELDS_2TUPLE	1
581 #define	IPI_HASHFIELDS_4TUPLE	2
582 
583 #ifdef _KERNEL
584 VNET_DECLARE(int, ipport_reservedhigh);
585 VNET_DECLARE(int, ipport_reservedlow);
586 VNET_DECLARE(int, ipport_lowfirstauto);
587 VNET_DECLARE(int, ipport_lowlastauto);
588 VNET_DECLARE(int, ipport_firstauto);
589 VNET_DECLARE(int, ipport_lastauto);
590 VNET_DECLARE(int, ipport_hifirstauto);
591 VNET_DECLARE(int, ipport_hilastauto);
592 VNET_DECLARE(int, ipport_randomized);
593 VNET_DECLARE(int, ipport_randomcps);
594 VNET_DECLARE(int, ipport_randomtime);
595 VNET_DECLARE(int, ipport_stoprandom);
596 VNET_DECLARE(int, ipport_tcpallocs);
597 
598 #define	V_ipport_reservedhigh	VNET(ipport_reservedhigh)
599 #define	V_ipport_reservedlow	VNET(ipport_reservedlow)
600 #define	V_ipport_lowfirstauto	VNET(ipport_lowfirstauto)
601 #define	V_ipport_lowlastauto	VNET(ipport_lowlastauto)
602 #define	V_ipport_firstauto	VNET(ipport_firstauto)
603 #define	V_ipport_lastauto	VNET(ipport_lastauto)
604 #define	V_ipport_hifirstauto	VNET(ipport_hifirstauto)
605 #define	V_ipport_hilastauto	VNET(ipport_hilastauto)
606 #define	V_ipport_randomized	VNET(ipport_randomized)
607 #define	V_ipport_randomcps	VNET(ipport_randomcps)
608 #define	V_ipport_randomtime	VNET(ipport_randomtime)
609 #define	V_ipport_stoprandom	VNET(ipport_stoprandom)
610 #define	V_ipport_tcpallocs	VNET(ipport_tcpallocs)
611 
612 void	in_pcbinfo_destroy(struct inpcbinfo *);
613 void	in_pcbinfo_init(struct inpcbinfo *, const char *, struct inpcbhead *,
614 	    int, int, char *, uma_init, uma_fini, uint32_t, u_int);
615 
616 int	in_pcbbind_check_bindmulti(const struct inpcb *ni,
617 	    const struct inpcb *oi);
618 
619 struct inpcbgroup *
620 	in_pcbgroup_byhash(struct inpcbinfo *, u_int, uint32_t);
621 struct inpcbgroup *
622 	in_pcbgroup_byinpcb(struct inpcb *);
623 struct inpcbgroup *
624 	in_pcbgroup_bytuple(struct inpcbinfo *, struct in_addr, u_short,
625 	    struct in_addr, u_short);
626 void	in_pcbgroup_destroy(struct inpcbinfo *);
627 int	in_pcbgroup_enabled(struct inpcbinfo *);
628 void	in_pcbgroup_init(struct inpcbinfo *, u_int, int);
629 void	in_pcbgroup_remove(struct inpcb *);
630 void	in_pcbgroup_update(struct inpcb *);
631 void	in_pcbgroup_update_mbuf(struct inpcb *, struct mbuf *);
632 
633 void	in_pcbpurgeif0(struct inpcbinfo *, struct ifnet *);
634 int	in_pcballoc(struct socket *, struct inpcbinfo *);
635 int	in_pcbbind(struct inpcb *, struct sockaddr *, struct ucred *);
636 int	in_pcb_lport(struct inpcb *, struct in_addr *, u_short *,
637 	    struct ucred *, int);
638 int	in_pcbbind_setup(struct inpcb *, struct sockaddr *, in_addr_t *,
639 	    u_short *, struct ucred *);
640 int	in_pcbconnect(struct inpcb *, struct sockaddr *, struct ucred *);
641 int	in_pcbconnect_mbuf(struct inpcb *, struct sockaddr *, struct ucred *,
642 	    struct mbuf *);
643 int	in_pcbconnect_setup(struct inpcb *, struct sockaddr *, in_addr_t *,
644 	    u_short *, in_addr_t *, u_short *, struct inpcb **,
645 	    struct ucred *);
646 void	in_pcbdetach(struct inpcb *);
647 void	in_pcbdisconnect(struct inpcb *);
648 void	in_pcbdrop(struct inpcb *);
649 void	in_pcbfree(struct inpcb *);
650 int	in_pcbinshash(struct inpcb *);
651 int	in_pcbinshash_nopcbgroup(struct inpcb *);
652 int	in_pcbladdr(struct inpcb *, struct in_addr *, struct in_addr *,
653 	    struct ucred *);
654 struct inpcb *
655 	in_pcblookup_local(struct inpcbinfo *,
656 	    struct in_addr, u_short, int, struct ucred *);
657 struct inpcb *
658 	in_pcblookup(struct inpcbinfo *, struct in_addr, u_int,
659 	    struct in_addr, u_int, int, struct ifnet *);
660 struct inpcb *
661 	in_pcblookup_mbuf(struct inpcbinfo *, struct in_addr, u_int,
662 	    struct in_addr, u_int, int, struct ifnet *, struct mbuf *);
663 void	in_pcbnotifyall(struct inpcbinfo *pcbinfo, struct in_addr,
664 	    int, struct inpcb *(*)(struct inpcb *, int));
665 void	in_pcbref(struct inpcb *);
666 void	in_pcbrehash(struct inpcb *);
667 void	in_pcbrehash_mbuf(struct inpcb *, struct mbuf *);
668 int	in_pcbrele(struct inpcb *);
669 int	in_pcbrele_rlocked(struct inpcb *);
670 int	in_pcbrele_wlocked(struct inpcb *);
671 void	in_pcbsetsolabel(struct socket *so);
672 int	in_getpeeraddr(struct socket *so, struct sockaddr **nam);
673 int	in_getsockaddr(struct socket *so, struct sockaddr **nam);
674 struct sockaddr *
675 	in_sockaddr(in_port_t port, struct in_addr *addr);
676 void	in_pcbsosetlabel(struct socket *so);
677 #endif /* _KERNEL */
678 
679 #endif /* !_NETINET_IN_PCB_H_ */
680