in_pcb.h revision 189657
1/*-
2 * Copyright (c) 1982, 1986, 1990, 1993
3 *	The Regents of the University of California.
4 * All rights reserved.
5 *
6 * Redistribution and use in source and binary forms, with or without
7 * modification, are permitted provided that the following conditions
8 * are met:
9 * 1. Redistributions of source code must retain the above copyright
10 *    notice, this list of conditions and the following disclaimer.
11 * 2. Redistributions in binary form must reproduce the above copyright
12 *    notice, this list of conditions and the following disclaimer in the
13 *    documentation and/or other materials provided with the distribution.
14 * 4. Neither the name of the University nor the names of its contributors
15 *    may be used to endorse or promote products derived from this software
16 *    without specific prior written permission.
17 *
18 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
19 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
20 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
21 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
22 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
23 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
24 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
25 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
26 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
27 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
28 * SUCH DAMAGE.
29 *
30 *	@(#)in_pcb.h	8.1 (Berkeley) 6/10/93
31 * $FreeBSD: head/sys/netinet/in_pcb.h 189657 2009-03-11 00:29:22Z rwatson $
32 */
33
34#ifndef _NETINET_IN_PCB_H_
35#define _NETINET_IN_PCB_H_
36
37#include <sys/queue.h>
38#include <sys/_lock.h>
39#include <sys/_mutex.h>
40#include <sys/_rwlock.h>
41
42#include <net/route.h>
43
44#ifdef _KERNEL
45#include <sys/rwlock.h>
46#endif
47
48#define	in6pcb		inpcb	/* for KAME src sync over BSD*'s */
49#define	in6p_sp		inp_sp	/* for KAME src sync over BSD*'s */
50struct inpcbpolicy;
51
52/*
53 * struct inpcb is the common protocol control block structure used in most
54 * IP transport protocols.
55 *
56 * Pointers to local and foreign host table entries, local and foreign socket
57 * numbers, and pointers up (to a socket structure) and down (to a
58 * protocol-specific control block) are stored here.
59 */
60LIST_HEAD(inpcbhead, inpcb);
61LIST_HEAD(inpcbporthead, inpcbport);
62typedef	u_quad_t	inp_gen_t;
63
64/*
65 * PCB with AF_INET6 null bind'ed laddr can receive AF_INET input packet.
66 * So, AF_INET6 null laddr is also used as AF_INET null laddr, by utilizing
67 * the following structure.
68 */
69struct in_addr_4in6 {
70	u_int32_t	ia46_pad32[3];
71	struct	in_addr	ia46_addr4;
72};
73
74/*
75 * NOTE: ipv6 addrs should be 64-bit aligned, per RFC 2553.  in_conninfo has
76 * some extra padding to accomplish this.
77 */
78struct in_endpoints {
79	u_int16_t	ie_fport;		/* foreign port */
80	u_int16_t	ie_lport;		/* local port */
81	/* protocol dependent part, local and foreign addr */
82	union {
83		/* foreign host table entry */
84		struct	in_addr_4in6 ie46_foreign;
85		struct	in6_addr ie6_foreign;
86	} ie_dependfaddr;
87	union {
88		/* local host table entry */
89		struct	in_addr_4in6 ie46_local;
90		struct	in6_addr ie6_local;
91	} ie_dependladdr;
92};
93#define	ie_faddr	ie_dependfaddr.ie46_foreign.ia46_addr4
94#define	ie_laddr	ie_dependladdr.ie46_local.ia46_addr4
95#define	ie6_faddr	ie_dependfaddr.ie6_foreign
96#define	ie6_laddr	ie_dependladdr.ie6_local
97
98/*
99 * XXX The defines for inc_* are hacks and should be changed to direct
100 * references.
101 */
102struct in_conninfo {
103	u_int8_t	inc_flags;
104	u_int8_t	inc_len;
105	u_int16_t	inc_fibnum;	/* XXX was pad, 16 bits is plenty */
106	/* protocol dependent part */
107	struct	in_endpoints inc_ie;
108};
109
110/*
111 * Flags for inc_flags.
112 */
113#define	INC_ISIPV6	0x01
114
115#define inc_isipv6	inc_flags	/* temp compatability */
116#define	inc_fport	inc_ie.ie_fport
117#define	inc_lport	inc_ie.ie_lport
118#define	inc_faddr	inc_ie.ie_faddr
119#define	inc_laddr	inc_ie.ie_laddr
120#define	inc6_faddr	inc_ie.ie6_faddr
121#define	inc6_laddr	inc_ie.ie6_laddr
122
123struct	icmp6_filter;
124
125/*-
126 * struct inpcb captures the network layer state for TCP, UDP, and raw IPv4
127 * and IPv6 sockets.  In the case of TCP, further per-connection state is
128 * hung off of inp_ppcb most of the time.  Almost all fields of struct inpcb
129 * are static after creation or protected by a per-inpcb rwlock, inp_lock.  A
130 * few fields also require the global pcbinfo lock for the inpcb to be held,
131 * when modified, such as the global connection lists and hashes, as well as
132 * binding information (which affects which hash a connection is on).  This
133 * model means that connections can be looked up without holding the
134 * per-connection lock, which is important for performance when attempting to
135 * find the connection for a packet given its IP and port tuple.  Writing to
136 * these fields that write locks be held on both the inpcb and global locks.
137 *
138 * Key:
139 * (c) - Constant after initialization
140 * (i) - Protected by the inpcb lock
141 * (p) - Protected by the pcbinfo lock for the inpcb
142 * (s) - Protected by another subsystem's locks
143 * (x) - Undefined locking
144 *
145 * A few other notes:
146 *
147 * When a read lock is held, stability of the field is guaranteed; to write
148 * to a field, a write lock must generally be held.
149 *
150 * netinet/netinet6-layer code should not assume that the inp_socket pointer
151 * is safe to dereference without inp_lock being held, even for protocols
152 * other than TCP (where the inpcb persists during TIMEWAIT even after the
153 * socket has been freed), or there may be close(2)-related races.
154 *
155 * The inp_vflag field is overloaded, and would otherwise ideally be (c).
156 */
157struct inpcb {
158	LIST_ENTRY(inpcb) inp_hash;	/* (i/p) hash list */
159	LIST_ENTRY(inpcb) inp_list;	/* (i/p) list for all PCBs for proto */
160	void	*inp_ppcb;		/* (i) pointer to per-protocol pcb */
161	struct	inpcbinfo *inp_pcbinfo;	/* (c) PCB list info */
162	struct	socket *inp_socket;	/* (i) back pointer to socket */
163	struct	ucred	*inp_cred;	/* (c) cache of socket cred */
164	u_int32_t inp_flow;		/* (i) IPv6 flow information */
165	int	inp_flags;		/* (i) generic IP/datagram flags */
166	u_char	inp_vflag;		/* (i) IP version flag (v4/v6) */
167	u_char	inp_ip_ttl;		/* (i) time to live proto */
168	u_char	inp_ip_p;		/* (c) protocol proto */
169	u_char	inp_ip_minttl;		/* (i) minimum TTL or drop */
170	uint32_t inp_ispare1;		/* (x) connection id / queue id */
171	u_int	inp_refcount;		/* (i) refcount */
172	void	*inp_pspare[2];		/* (x) rtentry / general use */
173
174	/* Local and foreign ports, local and foreign addr. */
175	struct	in_conninfo inp_inc;	/* (i/p) list for PCB's local port */
176
177	/* MAC and IPSEC policy information. */
178	struct	label *inp_label;	/* (i) MAC label */
179	struct	inpcbpolicy *inp_sp;    /* (s) for IPSEC */
180
181	/* Protocol-dependent part; options. */
182	struct {
183		u_char	inp4_ip_tos;		/* (i) type of service proto */
184		struct	mbuf *inp4_options;	/* (i) IP options */
185		struct	ip_moptions *inp4_moptions; /* (i) IP mcast options */
186	} inp_depend4;
187	struct {
188		/* (i) IP options */
189		struct	mbuf *inp6_options;
190		/* (i) IP6 options for outgoing packets */
191		struct	ip6_pktopts *inp6_outputopts;
192		/* (i) IP multicast options */
193		struct	ip6_moptions *inp6_moptions;
194		/* (i) ICMPv6 code type filter */
195		struct	icmp6_filter *inp6_icmp6filt;
196		/* (i) IPV6_CHECKSUM setsockopt */
197		int	inp6_cksum;
198		short	inp6_hops;
199	} inp_depend6;
200	LIST_ENTRY(inpcb) inp_portlist;	/* (i/p) */
201	struct	inpcbport *inp_phd;	/* (i/p) head of this list */
202#define inp_zero_size offsetof(struct inpcb, inp_gencnt)
203	inp_gen_t	inp_gencnt;	/* (c) generation count */
204	struct rwlock	inp_lock;
205};
206#define	inp_fport	inp_inc.inc_fport
207#define	inp_lport	inp_inc.inc_lport
208#define	inp_faddr	inp_inc.inc_faddr
209#define	inp_laddr	inp_inc.inc_laddr
210#define	inp_ip_tos	inp_depend4.inp4_ip_tos
211#define	inp_options	inp_depend4.inp4_options
212#define	inp_moptions	inp_depend4.inp4_moptions
213
214#define	in6p_faddr	inp_inc.inc6_faddr
215#define	in6p_laddr	inp_inc.inc6_laddr
216#define	in6p_hops	inp_depend6.inp6_hops	/* default hop limit */
217#define	in6p_flowinfo	inp_flow
218#define	in6p_options	inp_depend6.inp6_options
219#define	in6p_outputopts	inp_depend6.inp6_outputopts
220#define	in6p_moptions	inp_depend6.inp6_moptions
221#define	in6p_icmp6filt	inp_depend6.inp6_icmp6filt
222#define	in6p_cksum	inp_depend6.inp6_cksum
223
224/*
225 * The range of the generation count, as used in this implementation, is 9e19.
226 * We would have to create 300 billion connections per second for this number
227 * to roll over in a year.  This seems sufficiently unlikely that we simply
228 * don't concern ourselves with that possibility.
229 */
230
231/*
232 * Interface exported to userland by various protocols which use inpcbs.  Hack
233 * alert -- only define if struct xsocket is in scope.
234 */
235#ifdef _SYS_SOCKETVAR_H_
236struct	xinpcb {
237	size_t	xi_len;		/* length of this structure */
238	struct	inpcb xi_inp;
239	struct	xsocket xi_socket;
240	u_quad_t	xi_alignment_hack;
241};
242
243struct	xinpgen {
244	size_t	xig_len;	/* length of this structure */
245	u_int	xig_count;	/* number of PCBs at this time */
246	inp_gen_t xig_gen;	/* generation count at this time */
247	so_gen_t xig_sogen;	/* socket generation count at this time */
248};
249#endif /* _SYS_SOCKETVAR_H_ */
250
251struct inpcbport {
252	LIST_ENTRY(inpcbport) phd_hash;
253	struct inpcbhead phd_pcblist;
254	u_short phd_port;
255};
256
257/*
258 * Global data structure for each high-level protocol (UDP, TCP, ...) in both
259 * IPv4 and IPv6.  Holds inpcb lists and information for managing them.
260 */
261struct inpcbinfo {
262	/*
263	 * Global list of inpcbs on the protocol.
264	 */
265	struct inpcbhead	*ipi_listhead;
266	u_int			 ipi_count;
267
268	/*
269	 * Global hash of inpcbs, hashed by local and foreign addresses and
270	 * port numbers.
271	 */
272	struct inpcbhead	*ipi_hashbase;
273	u_long			 ipi_hashmask;
274
275	/*
276	 * Global hash of inpcbs, hashed by only local port number.
277	 */
278	struct inpcbporthead	*ipi_porthashbase;
279	u_long			 ipi_porthashmask;
280
281	/*
282	 * Fields associated with port lookup and allocation.
283	 */
284	u_short			 ipi_lastport;
285	u_short			 ipi_lastlow;
286	u_short			 ipi_lasthi;
287
288	/*
289	 * UMA zone from which inpcbs are allocated for this protocol.
290	 */
291	struct	uma_zone	*ipi_zone;
292
293	/*
294	 * Generation count--incremented each time a connection is allocated
295	 * or freed.
296	 */
297	u_quad_t		 ipi_gencnt;
298	struct rwlock		 ipi_lock;
299
300	/*
301	 * vimage 1
302	 * general use 1
303	 */
304	void 			*ipi_pspare[2];
305};
306
307#define INP_LOCK_INIT(inp, d, t) \
308	rw_init_flags(&(inp)->inp_lock, (t), RW_RECURSE |  RW_DUPOK)
309#define INP_LOCK_DESTROY(inp)	rw_destroy(&(inp)->inp_lock)
310#define INP_RLOCK(inp)		rw_rlock(&(inp)->inp_lock)
311#define INP_WLOCK(inp)		rw_wlock(&(inp)->inp_lock)
312#define INP_TRY_RLOCK(inp)	rw_try_rlock(&(inp)->inp_lock)
313#define INP_TRY_WLOCK(inp)	rw_try_wlock(&(inp)->inp_lock)
314#define INP_RUNLOCK(inp)	rw_runlock(&(inp)->inp_lock)
315#define INP_WUNLOCK(inp)	rw_wunlock(&(inp)->inp_lock)
316#define INP_LOCK_ASSERT(inp)	rw_assert(&(inp)->inp_lock, RA_LOCKED)
317#define	INP_RLOCK_ASSERT(inp)	rw_assert(&(inp)->inp_lock, RA_RLOCKED)
318#define	INP_WLOCK_ASSERT(inp)	rw_assert(&(inp)->inp_lock, RA_WLOCKED)
319#define	INP_UNLOCK_ASSERT(inp)	rw_assert(&(inp)->inp_lock, RA_UNLOCKED)
320
321#ifdef _KERNEL
322/*
323 * These locking functions are for inpcb consumers outside of sys/netinet,
324 * more specifically, they were added for the benefit of TOE drivers. The
325 * macros are reserved for use by the stack.
326 */
327void inp_wlock(struct inpcb *);
328void inp_wunlock(struct inpcb *);
329void inp_rlock(struct inpcb *);
330void inp_runlock(struct inpcb *);
331
332#ifdef INVARIANTS
333void inp_lock_assert(struct inpcb *);
334void inp_unlock_assert(struct inpcb *);
335#else
336static __inline void
337inp_lock_assert(struct inpcb *inp __unused)
338{
339}
340
341static __inline void
342inp_unlock_assert(struct inpcb *inp __unused)
343{
344}
345
346#endif
347
348void	inp_apply_all(void (*func)(struct inpcb *, void *), void *arg);
349int 	inp_ip_tos_get(const struct inpcb *inp);
350void 	inp_ip_tos_set(struct inpcb *inp, int val);
351struct socket *
352	inp_inpcbtosocket(struct inpcb *inp);
353struct tcpcb *
354	inp_inpcbtotcpcb(struct inpcb *inp);
355void 	inp_4tuple_get(struct inpcb *inp, uint32_t *laddr, uint16_t *lp,
356		uint32_t *faddr, uint16_t *fp);
357
358#endif /* _KERNEL */
359
360#define INP_INFO_LOCK_INIT(ipi, d) \
361	rw_init_flags(&(ipi)->ipi_lock, (d), RW_RECURSE)
362#define INP_INFO_LOCK_DESTROY(ipi)  rw_destroy(&(ipi)->ipi_lock)
363#define INP_INFO_RLOCK(ipi)	rw_rlock(&(ipi)->ipi_lock)
364#define INP_INFO_WLOCK(ipi)	rw_wlock(&(ipi)->ipi_lock)
365#define INP_INFO_TRY_RLOCK(ipi)	rw_try_rlock(&(ipi)->ipi_lock)
366#define INP_INFO_TRY_WLOCK(ipi)	rw_try_wlock(&(ipi)->ipi_lock)
367#define INP_INFO_RUNLOCK(ipi)	rw_runlock(&(ipi)->ipi_lock)
368#define INP_INFO_WUNLOCK(ipi)	rw_wunlock(&(ipi)->ipi_lock)
369#define	INP_INFO_LOCK_ASSERT(ipi)	rw_assert(&(ipi)->ipi_lock, RA_LOCKED)
370#define INP_INFO_RLOCK_ASSERT(ipi)	rw_assert(&(ipi)->ipi_lock, RA_RLOCKED)
371#define INP_INFO_WLOCK_ASSERT(ipi)	rw_assert(&(ipi)->ipi_lock, RA_WLOCKED)
372#define INP_INFO_UNLOCK_ASSERT(ipi)	rw_assert(&(ipi)->ipi_lock, RA_UNLOCKED)
373
374#define INP_PCBHASH(faddr, lport, fport, mask) \
375	(((faddr) ^ ((faddr) >> 16) ^ ntohs((lport) ^ (fport))) & (mask))
376#define INP_PCBPORTHASH(lport, mask) \
377	(ntohs((lport)) & (mask))
378
379/*
380 * Flags for inp_vflags -- historically version flags only, but now quite a
381 * bit more due to an overflow of inp_flag, leading to some locking ambiguity
382 * as some bits are stable from initial allocation, and others may change.
383 */
384#define	INP_IPV4	0x1
385#define	INP_IPV6	0x2
386#define	INP_IPV6PROTO	0x4		/* opened under IPv6 protocol */
387#define	INP_TIMEWAIT	0x8		/* inpcb in TIMEWAIT, ppcb is tcptw */
388#define	INP_ONESBCAST	0x10		/* send all-ones broadcast */
389#define	INP_DROPPED	0x20		/* protocol drop flag */
390#define	INP_SOCKREF	0x40		/* strong socket reference */
391
392/*
393 * Flags for inp_flag.
394 */
395#define	INP_RECVOPTS		0x01	/* receive incoming IP options */
396#define	INP_RECVRETOPTS		0x02	/* receive IP options for reply */
397#define	INP_RECVDSTADDR		0x04	/* receive IP dst address */
398#define	INP_HDRINCL		0x08	/* user supplies entire IP header */
399#define	INP_HIGHPORT		0x10	/* user wants "high" port binding */
400#define	INP_LOWPORT		0x20	/* user wants "low" port binding */
401#define	INP_ANONPORT		0x40	/* port chosen for user */
402#define	INP_RECVIF		0x80	/* receive incoming interface */
403#define	INP_MTUDISC		0x100	/* user can do MTU discovery */
404#define	INP_FAITH		0x200	/* accept FAITH'ed connections */
405#define	INP_RECVTTL		0x400	/* receive incoming IP TTL */
406#define	INP_DONTFRAG		0x800	/* don't fragment packet */
407#define	INP_NONLOCALOK		0x1000	/* Allow bind to spoof any address */
408					/* - requires options IP_NONLOCALBIND */
409#define	INP_INHASHLIST		0x2000	/* in_pcbinshash() has been called */
410
411#define IN6P_IPV6_V6ONLY	0x008000 /* restrict AF_INET6 socket for v6 */
412
413#define	IN6P_PKTINFO		0x010000 /* receive IP6 dst and I/F */
414#define	IN6P_HOPLIMIT		0x020000 /* receive hoplimit */
415#define	IN6P_HOPOPTS		0x040000 /* receive hop-by-hop options */
416#define	IN6P_DSTOPTS		0x080000 /* receive dst options after rthdr */
417#define	IN6P_RTHDR		0x100000 /* receive routing header */
418#define	IN6P_RTHDRDSTOPTS	0x200000 /* receive dstoptions before rthdr */
419#define	IN6P_TCLASS		0x400000 /* receive traffic class value */
420#define	IN6P_AUTOFLOWLABEL	0x800000 /* attach flowlabel automatically */
421#define	IN6P_RFC2292		0x40000000 /* used RFC2292 API on the socket */
422#define	IN6P_MTU		0x80000000 /* receive path MTU */
423
424#define	INP_CONTROLOPTS		(INP_RECVOPTS|INP_RECVRETOPTS|INP_RECVDSTADDR|\
425				 INP_RECVIF|INP_RECVTTL|\
426				 IN6P_PKTINFO|IN6P_HOPLIMIT|IN6P_HOPOPTS|\
427				 IN6P_DSTOPTS|IN6P_RTHDR|IN6P_RTHDRDSTOPTS|\
428				 IN6P_TCLASS|IN6P_AUTOFLOWLABEL|IN6P_RFC2292|\
429				 IN6P_MTU)
430
431#define	INPLOOKUP_WILDCARD	1
432#define	sotoinpcb(so)	((struct inpcb *)(so)->so_pcb)
433#define	sotoin6pcb(so)	sotoinpcb(so) /* for KAME src sync over BSD*'s */
434
435#define	INP_SOCKAF(so) so->so_proto->pr_domain->dom_family
436
437#define	INP_CHECK_SOCKAF(so, af)	(INP_SOCKAF(so) == af)
438
439#ifdef _KERNEL
440#ifdef VIMAGE_GLOBALS
441extern int	ipport_reservedhigh;
442extern int	ipport_reservedlow;
443extern int	ipport_lowfirstauto;
444extern int	ipport_lowlastauto;
445extern int	ipport_firstauto;
446extern int	ipport_lastauto;
447extern int	ipport_hifirstauto;
448extern int	ipport_hilastauto;
449extern int	ipport_randomized;
450extern int	ipport_randomcps;
451extern int	ipport_randomtime;
452extern int	ipport_stoprandom;
453extern int	ipport_tcpallocs;
454#endif
455extern struct callout ipport_tick_callout;
456
457void	in_pcbpurgeif0(struct inpcbinfo *, struct ifnet *);
458int	in_pcballoc(struct socket *, struct inpcbinfo *);
459int	in_pcbbind(struct inpcb *, struct sockaddr *, struct ucred *);
460int	in_pcbbind_setup(struct inpcb *, struct sockaddr *, in_addr_t *,
461	    u_short *, struct ucred *);
462int	in_pcbconnect(struct inpcb *, struct sockaddr *, struct ucred *);
463int	in_pcbconnect_setup(struct inpcb *, struct sockaddr *, in_addr_t *,
464	    u_short *, in_addr_t *, u_short *, struct inpcb **,
465	    struct ucred *);
466void	in_pcbdetach(struct inpcb *);
467void	in_pcbdisconnect(struct inpcb *);
468void	in_pcbdrop(struct inpcb *);
469void	in_pcbfree(struct inpcb *);
470int	in_pcbinshash(struct inpcb *);
471struct inpcb *
472	in_pcblookup_local(struct inpcbinfo *,
473	    struct in_addr, u_short, int, struct ucred *);
474struct inpcb *
475	in_pcblookup_hash(struct inpcbinfo *, struct in_addr, u_int,
476	    struct in_addr, u_int, int, struct ifnet *);
477void	in_pcbnotifyall(struct inpcbinfo *pcbinfo, struct in_addr,
478	    int, struct inpcb *(*)(struct inpcb *, int));
479void	in_pcbref(struct inpcb *);
480void	in_pcbrehash(struct inpcb *);
481int	in_pcbrele(struct inpcb *);
482void	in_pcbsetsolabel(struct socket *so);
483int	in_getpeeraddr(struct socket *so, struct sockaddr **nam);
484int	in_getsockaddr(struct socket *so, struct sockaddr **nam);
485struct sockaddr *
486	in_sockaddr(in_port_t port, struct in_addr *addr);
487void	in_pcbsosetlabel(struct socket *so);
488void	in_pcbremlists(struct inpcb *inp);
489void	ipport_tick(void *xtp);
490
491/*
492 * Debugging routines compiled in when DDB is present.
493 */
494void	db_print_inpcb(struct inpcb *inp, const char *name, int indent);
495
496#endif /* _KERNEL */
497
498#endif /* !_NETINET_IN_PCB_H_ */
499