2 * Copyright (c) 2001-2007, by Cisco Systems, Inc. All rights reserved.
4 * Redistribution and use in source and binary forms, with or without
5 * modification, are permitted provided that the following conditions are met:
7 * a) Redistributions of source code must retain the above copyright notice,
8 * this list of conditions and the following disclaimer.
10 * b) Redistributions in binary form must reproduce the above copyright
11 * notice, this list of conditions and the following disclaimer in
12 * the documentation and/or other materials provided with the distribution.
14 * c) Neither the name of Cisco Systems, Inc. nor the names of its
15 * contributors may be used to endorse or promote products derived
16 * from this software without specific prior written permission.
18 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
19 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
20 * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
21 * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
22 * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
23 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
24 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
25 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
26 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
27 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
28 * THE POSSIBILITY OF SUCH DAMAGE.
31 /* $KAME: sctp_pcb.h,v 1.21 2005/07/16 01:18:47 suz Exp $ */
33 #include <sys/cdefs.h>
34 __FBSDID("$FreeBSD$");
36 #ifndef __sctp_pcb_h__
37 #define __sctp_pcb_h__
39 #include <netinet/sctp_os.h>
40 #include <netinet/sctp.h>
41 #include <netinet/sctp_constants.h>
42 #include <netinet/sctp_sysctl.h>
44 LIST_HEAD(sctppcbhead, sctp_inpcb);
45 LIST_HEAD(sctpasochead, sctp_tcb);
46 LIST_HEAD(sctpladdr, sctp_laddr);
47 LIST_HEAD(sctpvtaghead, sctp_tagblock);
48 LIST_HEAD(sctp_vrflist, sctp_vrf);
49 LIST_HEAD(sctp_ifnlist, sctp_ifn);
50 LIST_HEAD(sctp_ifalist, sctp_ifa);
51 TAILQ_HEAD(sctp_readhead, sctp_queued_to_read);
52 TAILQ_HEAD(sctp_streamhead, sctp_stream_queue_pending);
54 #include <netinet/sctp_structs.h>
55 #include <netinet/sctp_auth.h>
57 #define SCTP_PCBHASH_ALLADDR(port, mask) (port & mask)
58 #define SCTP_PCBHASH_ASOC(tag, mask) (tag & mask)
61 LIST_ENTRY(sctp_vrf) next_vrf;
62 struct sctp_ifalist *vrf_addr_hash;
63 struct sctp_ifnlist ifnlist;
65 uint32_t tbl_id_v4; /* default v4 table id */
66 uint32_t tbl_id_v6; /* default v6 table id */
67 uint32_t total_ifa_count;
68 u_long vrf_addr_hashmark;
73 struct sctp_ifalist ifalist;
75 LIST_ENTRY(sctp_ifn) next_ifn;
76 LIST_ENTRY(sctp_ifn) next_bucket;
77 void *ifn_p; /* never access without appropriate lock */
80 uint32_t ifn_index; /* shorthand way to look at ifn for reference */
81 uint32_t refcount; /* number of reference held should be >=
83 uint32_t ifa_count; /* IFA's we hold (in our list - ifalist) */
84 uint32_t num_v6; /* number of v6 addresses */
85 uint32_t num_v4; /* number of v4 addresses */
86 uint32_t registered_af; /* registered address family for i/f events */
87 char ifn_name[SCTP_IFNAMSIZ];
90 /* SCTP local IFA flags */
91 #define SCTP_ADDR_VALID 0x00000001 /* its up and active */
92 #define SCTP_BEING_DELETED 0x00000002 /* being deleted, when
93 * refcount = 0. Note that it
94 * is pulled from the ifn list
95 * and ifa_p is nulled right
96 * away but it cannot be freed
98 * pointing to it is deleted. */
99 #define SCTP_ADDR_DEFER_USE 0x00000004 /* Hold off using this one */
100 #define SCTP_ADDR_IFA_UNUSEABLE 0x00000008
103 LIST_ENTRY(sctp_ifa) next_ifa;
104 LIST_ENTRY(sctp_ifa) next_bucket;
105 struct sctp_ifn *ifn_p; /* back pointer to parent ifn */
106 void *ifa; /* pointer to ifa, needed for flag update for
107 * that we MUST lock appropriate locks. This
109 union sctp_sockstore address;
110 uint32_t refcount; /* number of folks refering to this */
112 uint32_t localifa_flags;
113 uint32_t vrf_id; /* vrf_id of this addr (for deleting) */
121 LIST_ENTRY(sctp_laddr) sctp_nxt_addr; /* next in list */
122 struct sctp_ifa *ifa;
123 uint32_t action; /* Used during asconf and adding if no-zero
124 * src-addr selection will not consider this
126 struct timeval start_time; /* time when this address was created */
129 struct sctp_block_entry {
133 struct sctp_timewait {
134 uint32_t tv_sec_at_expire; /* the seconds from boot to expire */
135 uint32_t v_tag; /* the vtag that can not be reused */
136 uint16_t lport; /* the local port used in vtag */
137 uint16_t rport; /* the remote port used in vtag */
140 struct sctp_tagblock {
141 LIST_ENTRY(sctp_tagblock) sctp_nxt_tagblock;
142 struct sctp_timewait vtag_block[SCTP_NUMBER_IN_VTAG_BLOCK];
147 struct sctpasochead *sctp_asochash;
150 struct sctppcbhead *sctp_ephash;
154 * The TCP model represents a substantial overhead in that we get an
155 * additional hash table to keep explicit connections in. The
156 * listening TCP endpoint will exist in the usual ephash above and
157 * accept only INIT's. It will be incapable of sending off an INIT.
158 * When a dg arrives we must look in the normal ephash. If we find a
159 * TCP endpoint that will tell us to go to the specific endpoint
160 * hash and re-hash to find the right assoc/socket. If we find a UDP
161 * model socket we then must complete the lookup. If this fails,
162 * i.e. no association can be found then we must continue to see if
163 * a sctp_peeloff()'d socket is in the tcpephash (a spun off socket
164 * acts like a TCP model connected socket).
166 struct sctppcbhead *sctp_tcpephash;
168 uint32_t hashtblsize;
170 struct sctp_vrflist *sctp_vrfhash;
173 struct sctp_ifnlist *vrf_ifn_hash;
174 u_long vrf_ifn_hashmark;
176 struct sctppcbhead listhead;
177 struct sctpladdr addr_wq;
179 struct sctpiterators iteratorhead;
180 int threads_must_exit;
182 sctp_zone_t ipi_zone_ep;
183 sctp_zone_t ipi_zone_asoc;
184 sctp_zone_t ipi_zone_laddr;
185 sctp_zone_t ipi_zone_net;
186 sctp_zone_t ipi_zone_chunk;
187 sctp_zone_t ipi_zone_readq;
188 sctp_zone_t ipi_zone_strmoq;
189 sctp_zone_t ipi_zone_asconf;
190 sctp_zone_t ipi_zone_asconf_ack;
192 struct rwlock ipi_ep_mtx;
194 struct mtx ipi_iterator_wq_mtx;
195 struct rwlock ipi_addr_mtx;
196 struct mtx ipi_pktlog_mtx;
197 uint32_t ipi_count_ep;
199 /* assoc/tcb zone info */
200 uint32_t ipi_count_asoc;
202 /* local addrlist zone info */
203 uint32_t ipi_count_laddr;
205 /* remote addrlist zone info */
206 uint32_t ipi_count_raddr;
208 /* chunk structure list for output */
209 uint32_t ipi_count_chunk;
211 /* socket queue zone info */
212 uint32_t ipi_count_readq;
214 /* socket queue zone info */
215 uint32_t ipi_count_strmoq;
218 uint32_t ipi_count_vrfs;
221 uint32_t ipi_count_ifns;
224 uint32_t ipi_count_ifas;
226 /* system wide number of free chunks hanging around */
227 uint32_t ipi_free_chunks;
228 uint32_t ipi_free_strmoq;
231 struct sctpvtaghead vtag_timewait[SCTP_STACK_VTAG_HASH_SIZE];
233 /* address work queue handling */
234 #if defined(SCTP_USE_THREAD_BASED_ITERATOR)
235 uint32_t iterator_running;
236 SCTP_PROCESS_STRUCT thread_proc;
238 struct sctp_timer addr_wq_timer;
243 struct sctp_base_info {
245 * All static structures that anchor the system must be here.
247 struct sctp_epinfo sctppcbinfo;
248 struct sctpstat sctpstat;
249 struct sctp_sysctl sctpsysctl;
251 char sctp_pcb_initialized;
252 #if defined(SCTP_PACKET_LOGGING)
253 int packet_log_writers;
255 uint8_t packet_log_buffer[SCTP_PACKET_LOG_SIZE];
260 * Here we have all the relevant information for each SCTP entity created. We
261 * will need to modify this as approprate. We also need to figure out how to
262 * access /dev/random.
265 unsigned int time_of_secret_change; /* number of seconds from
267 uint32_t secret_key[SCTP_HOW_MANY_SECRETS][SCTP_NUMBER_OF_SECRETS];
268 unsigned int size_of_a_cookie;
270 unsigned int sctp_timeoutticks[SCTP_NUM_TMRS];
271 unsigned int sctp_minrto;
272 unsigned int sctp_maxrto;
273 unsigned int initial_rto;
274 int initial_init_rto_max;
276 unsigned int sctp_sack_freq;
277 uint32_t sctp_sws_sender;
278 uint32_t sctp_sws_receiver;
280 uint32_t sctp_default_cc_module;
281 /* authentication related fields */
282 struct sctp_keyhead shared_keys;
283 sctp_auth_chklist_t *local_auth_chunks;
284 sctp_hmaclist_t *local_hmacs;
285 uint16_t default_keyid;
287 /* various thresholds */
288 /* Max times I will init at a guy */
289 uint16_t max_init_times;
291 /* Max times I will send before we consider someone dead */
292 uint16_t max_send_times;
294 uint16_t def_net_failure;
296 /* number of streams to pre-open on a association */
297 uint16_t pre_open_stream_count;
298 uint16_t max_open_streams_intome;
300 /* random number generator */
301 uint32_t random_counter;
302 uint8_t random_numbers[SCTP_SIGNATURE_ALOC_SIZE];
303 uint8_t random_store[SCTP_SIGNATURE_ALOC_SIZE];
306 * This timer is kept running per endpoint. When it fires it will
307 * change the secret key. The default is once a hour
309 struct sctp_timer signature_change;
311 /* Zero copy full buffer timer */
312 struct sctp_timer zero_copy_timer;
313 /* Zero copy app to transport (sendq) read repulse timer */
314 struct sctp_timer zero_copy_sendq_timer;
315 uint32_t def_cookie_life;
318 uint32_t initial_sequence_debug;
319 uint32_t adaptation_layer_indicator;
322 char current_secret_number;
323 char last_secret_number;
326 #ifndef SCTP_ALIGNMENT
327 #define SCTP_ALIGNMENT 32
331 #define SCTP_ALIGNM1 (SCTP_ALIGNMENT-1)
334 #define sctp_lport ip_inp.inp.inp_lport
336 struct sctp_pcbtsn_rlog {
344 #define SCTP_READ_LOG_SIZE 135 /* we choose the number to make a pcb a page */
349 * put an inpcb in front of it all, kind of a waste but we need to
350 * for compatability with all the other stuff.
354 char align[(sizeof(struct in6pcb) + SCTP_ALIGNM1) &
359 /* Socket buffer lock protects read_queue and of course sb_cc */
360 struct sctp_readhead read_queue;
362 LIST_ENTRY(sctp_inpcb) sctp_list; /* lists all endpoints */
363 /* hash of all endpoints for model */
364 LIST_ENTRY(sctp_inpcb) sctp_hash;
365 /* count of local addresses bound, 0 if bound all */
368 /* list of addrs in use by the EP, NULL if bound-all */
369 struct sctpladdr sctp_addr_list;
371 * used for source address selection rotation when we are subset
374 struct sctp_laddr *next_addr_touse;
376 /* back pointer to our socket */
377 struct socket *sctp_socket;
378 uint32_t sctp_flags; /* INP state flag set */
379 uint32_t sctp_features; /* Feature flags */
380 uint32_t sctp_mobility_features; /* Mobility Feature flags */
381 struct sctp_pcb sctp_ep;/* SCTP ep data */
382 /* head of the hash of all associations */
383 struct sctpasochead *sctp_tcbhash;
384 u_long sctp_hashmark;
385 /* head of the list of all associations */
386 struct sctpasochead sctp_asoc_list;
387 #ifdef SCTP_TRACK_FREED_ASOCS
388 struct sctpasochead sctp_asoc_free_list;
390 struct sctp_iterator *inp_starting_point_for_iterator;
391 uint32_t sctp_frag_point;
392 uint32_t partial_delivery_point;
393 uint32_t sctp_context;
394 struct sctp_nonpad_sndrcvinfo def_send;
396 * These three are here for the sosend_dgram
397 * (pkt, pkt_last and control).
398 * routine. However, I don't think anyone in
399 * the current FreeBSD kernel calls this. So
400 * they are candidates with sctp_sendm for
403 struct mbuf *pkt, *pkt_last;
404 struct mbuf *control;
406 struct mtx inp_create_mtx;
407 struct mtx inp_rdata_mtx;
410 uint32_t total_sends;
411 uint32_t total_recvs;
412 uint32_t last_abort_code;
413 uint32_t total_nospaces;
414 struct sctpasochead *sctp_asocidhash;
415 u_long hashasocidmark;
416 uint32_t sctp_associd_counter;
418 #ifdef SCTP_ASOCLOG_OF_TSNS
419 struct sctp_pcbtsn_rlog readlog[SCTP_READ_LOG_SIZE];
420 uint32_t readlog_index;
425 struct socket *sctp_socket; /* back pointer to socket */
426 struct sctp_inpcb *sctp_ep; /* back pointer to ep */
427 LIST_ENTRY(sctp_tcb) sctp_tcbhash; /* next link in hash
429 LIST_ENTRY(sctp_tcb) sctp_tcblist; /* list of all of the
431 LIST_ENTRY(sctp_tcb) sctp_tcbasocidhash; /* next link in asocid
433 LIST_ENTRY(sctp_tcb) sctp_asocs; /* vtag hash list */
434 struct sctp_block_entry *block_entry; /* pointer locked by socket
436 struct sctp_association asoc;
438 * freed_by_sorcv_sincelast is protected by the sockbuf_lock NOT the
439 * tcb_lock. Its special in this way to help avoid extra mutex calls
440 * in the reading of data.
442 uint32_t freed_by_sorcv_sincelast;
443 uint32_t total_sends;
444 uint32_t total_recvs;
445 int freed_from_where;
446 uint16_t rport; /* remote port in network format */
449 struct mtx tcb_send_mtx;
454 #include <netinet/sctp_lock_bsd.h>
457 /* TODO where to put non-_KERNEL things for __Userspace__? */
458 #if defined(_KERNEL) || defined(__Userspace__)
460 /* Attention Julian, this is the extern that
461 * goes with the base info. sctp_pcb.c has
462 * the real definition.
464 extern struct sctp_base_info system_base_info;
467 int SCTP6_ARE_ADDR_EQUAL(struct sockaddr_in6 *a, struct sockaddr_in6 *b);
471 void sctp_fill_pcbinfo(struct sctp_pcbinfo *);
474 sctp_find_ifn(void *ifn, uint32_t ifn_index);
476 struct sctp_vrf *sctp_allocate_vrf(int vrfid);
477 struct sctp_vrf *sctp_find_vrf(uint32_t vrfid);
478 void sctp_free_vrf(struct sctp_vrf *vrf);
481 * Change address state, can be used if
482 * O/S supports telling transports about
483 * changes to IFA/IFN's (link layer triggers).
484 * If a ifn goes down, we will do src-addr-selection
485 * and NOT use that, as a source address. This does
486 * not stop the routing system from routing out
487 * that interface, but we won't put it as a source.
489 void sctp_mark_ifa_addr_down(uint32_t vrf_id, struct sockaddr *addr, const char *if_name, uint32_t ifn_index);
490 void sctp_mark_ifa_addr_up(uint32_t vrf_id, struct sockaddr *addr, const char *if_name, uint32_t ifn_index);
493 sctp_add_addr_to_vrf(uint32_t vrfid,
494 void *ifn, uint32_t ifn_index, uint32_t ifn_type,
496 void *ifa, struct sockaddr *addr, uint32_t ifa_flags,
499 void sctp_update_ifn_mtu(uint32_t ifn_index, uint32_t mtu);
501 void sctp_free_ifn(struct sctp_ifn *sctp_ifnp);
502 void sctp_free_ifa(struct sctp_ifa *sctp_ifap);
506 sctp_del_addr_from_vrf(uint32_t vrfid, struct sockaddr *addr,
507 uint32_t ifn_index, const char *if_name);
511 struct sctp_nets *sctp_findnet(struct sctp_tcb *, struct sockaddr *);
513 struct sctp_inpcb *sctp_pcb_findep(struct sockaddr *, int, int, uint32_t);
516 sctp_inpcb_bind(struct socket *, struct sockaddr *,
517 struct sctp_ifa *, struct thread *);
520 sctp_findassociation_addr(struct mbuf *, int, int,
521 struct sctphdr *, struct sctp_chunkhdr *, struct sctp_inpcb **,
522 struct sctp_nets **, uint32_t vrf_id);
525 sctp_findassociation_addr_sa(struct sockaddr *,
526 struct sockaddr *, struct sctp_inpcb **, struct sctp_nets **, int, uint32_t);
529 sctp_move_pcb_and_assoc(struct sctp_inpcb *, struct sctp_inpcb *,
533 * For this call ep_addr, the to is the destination endpoint address of the
534 * peer (relative to outbound). The from field is only used if the TCP model
535 * is enabled and helps distingush amongst the subset bound (non-boundall).
536 * The TCP model MAY change the actual ep field, this is why it is passed.
539 sctp_findassociation_ep_addr(struct sctp_inpcb **,
540 struct sockaddr *, struct sctp_nets **, struct sockaddr *,
544 sctp_findasoc_ep_asocid_locked(struct sctp_inpcb *inp, sctp_assoc_t asoc_id, int want_lock);
547 sctp_findassociation_ep_asocid(struct sctp_inpcb *,
551 sctp_findassociation_ep_asconf(struct mbuf *, int, int,
552 struct sctphdr *, struct sctp_inpcb **, struct sctp_nets **, uint32_t vrf_id);
554 int sctp_inpcb_alloc(struct socket *so, uint32_t vrf_id);
556 int sctp_is_address_on_local_host(struct sockaddr *addr, uint32_t vrf_id);
558 void sctp_inpcb_free(struct sctp_inpcb *, int, int);
561 sctp_aloc_assoc(struct sctp_inpcb *, struct sockaddr *,
562 int, int *, uint32_t, uint32_t, struct thread *);
564 int sctp_free_assoc(struct sctp_inpcb *, struct sctp_tcb *, int, int);
567 void sctp_delete_from_timewait(uint32_t, uint16_t, uint16_t);
569 int sctp_is_in_timewait(uint32_t tag, uint16_t lport, uint16_t rport);
572 sctp_add_vtag_to_timewait(uint32_t tag, uint32_t time, uint16_t lport, uint16_t rport);
574 void sctp_add_local_addr_ep(struct sctp_inpcb *, struct sctp_ifa *, uint32_t);
576 int sctp_insert_laddr(struct sctpladdr *, struct sctp_ifa *, uint32_t);
578 void sctp_remove_laddr(struct sctp_laddr *);
580 void sctp_del_local_addr_ep(struct sctp_inpcb *, struct sctp_ifa *);
582 int sctp_add_remote_addr(struct sctp_tcb *, struct sockaddr *, int, int);
584 void sctp_remove_net(struct sctp_tcb *, struct sctp_nets *);
586 int sctp_del_remote_addr(struct sctp_tcb *, struct sockaddr *);
588 void sctp_pcb_init(void);
590 void sctp_pcb_finish(void);
592 void sctp_add_local_addr_restricted(struct sctp_tcb *, struct sctp_ifa *);
593 void sctp_del_local_addr_restricted(struct sctp_tcb *, struct sctp_ifa *);
596 sctp_load_addresses_from_init(struct sctp_tcb *, struct mbuf *, int, int,
597 int, struct sctphdr *, struct sockaddr *);
600 sctp_set_primary_addr(struct sctp_tcb *, struct sockaddr *,
603 int sctp_is_vtag_good(struct sctp_inpcb *, uint32_t, uint16_t lport, uint16_t rport, struct timeval *, int);
605 /* void sctp_drain(void); */
607 int sctp_destination_is_reachable(struct sctp_tcb *, struct sockaddr *);
609 int sctp_swap_inpcb_for_listen(struct sctp_inpcb *inp);
612 * Null in last arg inpcb indicate run on ALL ep's. Specific inp in last arg
613 * indicates run on ONLY assoc's of the specified endpoint.
616 sctp_initiate_iterator(inp_func inpf,
627 #endif /* !__sctp_pcb_h__ */