usched: Allow process to change self cpu affinity
[dragonfly.git] / sys / netinet / igmp.c
blob19879844fb600ed20004787206ee169834d0e0c1
1 /*
2 * Copyright (c) 1988 Stephen Deering.
3 * Copyright (c) 1992, 1993
4 * The Regents of the University of California. All rights reserved.
6 * This code is derived from software contributed to Berkeley by
7 * Stephen Deering of Stanford University.
9 * Redistribution and use in source and binary forms, with or without
10 * modification, are permitted provided that the following conditions
11 * are met:
12 * 1. Redistributions of source code must retain the above copyright
13 * notice, this list of conditions and the following disclaimer.
14 * 2. Redistributions in binary form must reproduce the above copyright
15 * notice, this list of conditions and the following disclaimer in the
16 * documentation and/or other materials provided with the distribution.
17 * 3. Neither the name of the University nor the names of its contributors
18 * may be used to endorse or promote products derived from this software
19 * without specific prior written permission.
21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31 * SUCH DAMAGE.
33 * @(#)igmp.c 8.1 (Berkeley) 7/19/93
34 * $FreeBSD: src/sys/netinet/igmp.c,v 1.29.2.2 2003/01/23 21:06:44 sam Exp $
38 * Internet Group Management Protocol (IGMP) routines.
40 * Written by Steve Deering, Stanford, May 1988.
41 * Modified by Rosen Sharma, Stanford, Aug 1994.
42 * Modified by Bill Fenner, Xerox PARC, Feb 1995.
43 * Modified to fully comply to IGMPv2 by Bill Fenner, Oct 1995.
45 * MULTICAST Revision: 3.5.1.4
48 #include <sys/param.h>
49 #include <sys/systm.h>
50 #include <sys/malloc.h>
51 #include <sys/mbuf.h>
52 #include <sys/socket.h>
53 #include <sys/protosw.h>
54 #include <sys/kernel.h>
55 #include <sys/sysctl.h>
56 #include <sys/in_cksum.h>
57 #include <sys/thread2.h>
59 #include <machine/stdarg.h>
61 #include <net/if.h>
62 #include <net/route.h>
63 #include <net/netmsg2.h>
64 #include <net/netisr2.h>
66 #include <netinet/in.h>
67 #include <netinet/in_var.h>
68 #include <netinet/in_systm.h>
69 #include <netinet/ip.h>
70 #include <netinet/ip_var.h>
71 #include <netinet/igmp.h>
72 #include <netinet/igmp_var.h>
74 static MALLOC_DEFINE(M_IGMP, "igmp", "igmp state");
76 static struct router_info *
77 find_rti (struct ifnet *ifp);
79 static struct igmpstat igmpstat;
81 SYSCTL_STRUCT(_net_inet_igmp, IGMPCTL_STATS, stats, CTLFLAG_RW,
82 &igmpstat, igmpstat, "IGMP statistics");
84 static int igmp_timers_are_running;
85 static u_long igmp_all_hosts_group;
86 static u_long igmp_all_rtrs_group;
87 static struct mbuf *router_alert;
88 static struct router_info *Head;
90 static struct netmsg_base igmp_slowtimo_netmsg;
91 static struct netmsg_base igmp_fasttimo_netmsg;
93 static void igmp_sendpkt (struct in_multi *, int, unsigned long);
94 static void igmp_slowtimo_dispatch(netmsg_t);
95 static void igmp_fasttimo_dispatch(netmsg_t);
97 void
98 igmp_init(void)
100 struct ipoption *ra;
103 * To avoid byte-swapping the same value over and over again.
105 igmp_all_hosts_group = htonl(INADDR_ALLHOSTS_GROUP);
106 igmp_all_rtrs_group = htonl(INADDR_ALLRTRS_GROUP);
108 igmp_timers_are_running = 0;
111 * Construct a Router Alert option to use in outgoing packets
113 MGET(router_alert, M_NOWAIT, MT_DATA);
114 ra = mtod(router_alert, struct ipoption *);
115 ra->ipopt_dst.s_addr = 0;
116 ra->ipopt_list[0] = IPOPT_RA; /* Router Alert Option */
117 ra->ipopt_list[1] = 0x04; /* 4 bytes long */
118 ra->ipopt_list[2] = 0x00;
119 ra->ipopt_list[3] = 0x00;
120 router_alert->m_len = sizeof(ra->ipopt_dst) + ra->ipopt_list[1];
122 Head = NULL;
124 netmsg_init(&igmp_slowtimo_netmsg, NULL, &netisr_adone_rport,
125 MSGF_PRIORITY, igmp_slowtimo_dispatch);
126 netmsg_init(&igmp_fasttimo_netmsg, NULL, &netisr_adone_rport,
127 MSGF_PRIORITY, igmp_fasttimo_dispatch);
130 static struct router_info *
131 find_rti(struct ifnet *ifp)
133 struct router_info *rti = Head;
135 #ifdef IGMP_DEBUG
136 kprintf("[igmp.c, _find_rti] --> entering \n");
137 #endif
138 while (rti) {
139 if (rti->rti_ifp == ifp) {
140 #ifdef IGMP_DEBUG
141 kprintf("[igmp.c, _find_rti] --> found old entry \n");
142 #endif
143 return rti;
145 rti = rti->rti_next;
147 rti = kmalloc(sizeof *rti, M_IGMP, M_INTWAIT);
148 rti->rti_ifp = ifp;
149 rti->rti_type = IGMP_V2_ROUTER;
150 rti->rti_time = 0;
151 rti->rti_next = Head;
152 Head = rti;
153 #ifdef IGMP_DEBUG
154 kprintf("[igmp.c, _find_rti] --> created an entry \n");
155 #endif
156 return rti;
160 igmp_input(struct mbuf **mp, int *offp, int proto)
162 struct mbuf *m = *mp;
163 int iphlen;
164 struct igmp *igmp;
165 struct ip *ip;
166 int igmplen;
167 struct ifnet *ifp = m->m_pkthdr.rcvif;
168 int minlen;
169 struct in_multi *inm;
170 struct in_ifaddr *ia;
171 struct in_multistep step;
172 struct router_info *rti;
173 int timer; /** timer value in the igmp query header **/
175 iphlen = *offp;
176 *mp = NULL;
178 ++igmpstat.igps_rcv_total;
180 ip = mtod(m, struct ip *);
181 igmplen = ip->ip_len;
184 * Validate lengths
186 if (igmplen < IGMP_MINLEN) {
187 ++igmpstat.igps_rcv_tooshort;
188 m_freem(m);
189 return(IPPROTO_DONE);
191 minlen = iphlen + IGMP_MINLEN;
192 if ((m->m_flags & M_EXT || m->m_len < minlen) &&
193 (m = m_pullup(m, minlen)) == NULL) {
194 ++igmpstat.igps_rcv_tooshort;
195 return(IPPROTO_DONE);
199 * Validate checksum
201 m->m_data += iphlen;
202 m->m_len -= iphlen;
203 igmp = mtod(m, struct igmp *);
204 if (in_cksum(m, igmplen)) {
205 ++igmpstat.igps_rcv_badsum;
206 m_freem(m);
207 return(IPPROTO_DONE);
209 m->m_data -= iphlen;
210 m->m_len += iphlen;
212 ip = mtod(m, struct ip *);
213 timer = igmp->igmp_code * PR_FASTHZ / IGMP_TIMER_SCALE;
214 if (timer == 0)
215 timer = 1;
216 rti = find_rti(ifp);
219 * In the IGMPv2 specification, there are 3 states and a flag.
221 * In Non-Member state, we simply don't have a membership record.
222 * In Delaying Member state, our timer is running (inm->inm_timer)
223 * In Idle Member state, our timer is not running (inm->inm_timer==0)
225 * The flag is inm->inm_state, it is set to IGMP_OTHERMEMBER if
226 * we have heard a report from another member, or IGMP_IREPORTEDLAST
227 * if I sent the last report.
229 switch (igmp->igmp_type) {
231 case IGMP_MEMBERSHIP_QUERY:
232 ++igmpstat.igps_rcv_queries;
234 if (ifp->if_flags & IFF_LOOPBACK)
235 break;
237 if (igmp->igmp_code == 0) {
239 * Old router. Remember that the querier on this
240 * interface is old, and set the timer to the
241 * value in RFC 1112.
244 rti->rti_type = IGMP_V1_ROUTER;
245 rti->rti_time = 0;
247 timer = IGMP_MAX_HOST_REPORT_DELAY * PR_FASTHZ;
249 if (ip->ip_dst.s_addr != igmp_all_hosts_group ||
250 igmp->igmp_group.s_addr != 0) {
251 ++igmpstat.igps_rcv_badqueries;
252 m_freem(m);
253 return(IPPROTO_DONE);
255 } else {
257 * New router. Simply do the new validity check.
260 if (igmp->igmp_group.s_addr != 0 &&
261 !IN_MULTICAST(ntohl(igmp->igmp_group.s_addr))) {
262 ++igmpstat.igps_rcv_badqueries;
263 m_freem(m);
264 return(IPPROTO_DONE);
269 * - Start the timers in all of our membership records
270 * that the query applies to for the interface on
271 * which the query arrived excl. those that belong
272 * to the "all-hosts" group (224.0.0.1).
273 * - Restart any timer that is already running but has
274 * a value longer than the requested timeout.
275 * - Use the value specified in the query message as
276 * the maximum timeout.
278 IN_FIRST_MULTI(step, inm);
279 while (inm != NULL) {
280 if (inm->inm_ifp == ifp &&
281 inm->inm_addr.s_addr != igmp_all_hosts_group &&
282 (igmp->igmp_group.s_addr == 0 ||
283 igmp->igmp_group.s_addr == inm->inm_addr.s_addr)) {
284 if (inm->inm_timer == 0 ||
285 inm->inm_timer > timer) {
286 inm->inm_timer =
287 IGMP_RANDOM_DELAY(timer);
288 igmp_timers_are_running = 1;
291 IN_NEXT_MULTI(step, inm);
294 break;
296 case IGMP_V1_MEMBERSHIP_REPORT:
297 case IGMP_V2_MEMBERSHIP_REPORT:
299 * For fast leave to work, we have to know that we are the
300 * last person to send a report for this group. Reports
301 * can potentially get looped back if we are a multicast
302 * router, so discard reports sourced by me.
304 ia = IFP_TO_IA(ifp);
305 if (ia && ip->ip_src.s_addr == IA_SIN(ia)->sin_addr.s_addr)
306 break;
308 ++igmpstat.igps_rcv_reports;
310 if (ifp->if_flags & IFF_LOOPBACK)
311 break;
313 if (!IN_MULTICAST(ntohl(igmp->igmp_group.s_addr))) {
314 ++igmpstat.igps_rcv_badreports;
315 m_freem(m);
316 return(IPPROTO_DONE);
320 * KLUDGE: if the IP source address of the report has an
321 * unspecified (i.e., zero) subnet number, as is allowed for
322 * a booting host, replace it with the correct subnet number
323 * so that a process-level multicast routing demon can
324 * determine which subnet it arrived from. This is necessary
325 * to compensate for the lack of any way for a process to
326 * determine the arrival interface of an incoming packet.
328 if ((ntohl(ip->ip_src.s_addr) & IN_CLASSA_NET) == 0)
329 if (ia) ip->ip_src.s_addr = htonl(ia->ia_subnet);
332 * If we belong to the group being reported, stop
333 * our timer for that group.
335 inm = IN_LOOKUP_MULTI(&igmp->igmp_group, ifp);
337 if (inm != NULL) {
338 inm->inm_timer = 0;
339 ++igmpstat.igps_rcv_ourreports;
341 inm->inm_state = IGMP_OTHERMEMBER;
344 break;
348 * Pass all valid IGMP packets up to any process(es) listening
349 * on a raw IGMP socket.
351 *mp = m;
352 rip_input(mp, offp, proto);
353 return(IPPROTO_DONE);
356 void
357 igmp_joingroup(struct in_multi *inm)
359 crit_enter();
360 if (inm->inm_addr.s_addr == igmp_all_hosts_group
361 || inm->inm_ifp->if_flags & IFF_LOOPBACK) {
362 inm->inm_timer = 0;
363 inm->inm_state = IGMP_OTHERMEMBER;
364 } else {
365 inm->inm_rti = find_rti(inm->inm_ifp);
366 igmp_sendpkt(inm, inm->inm_rti->rti_type, 0);
367 inm->inm_timer = IGMP_RANDOM_DELAY(
368 IGMP_MAX_HOST_REPORT_DELAY*PR_FASTHZ);
369 inm->inm_state = IGMP_IREPORTEDLAST;
370 igmp_timers_are_running = 1;
372 crit_exit();
375 void
376 igmp_leavegroup(struct in_multi *inm)
378 if (inm->inm_state == IGMP_IREPORTEDLAST &&
379 inm->inm_addr.s_addr != igmp_all_hosts_group &&
380 !(inm->inm_ifp->if_flags & IFF_LOOPBACK) &&
381 inm->inm_rti->rti_type != IGMP_V1_ROUTER)
382 igmp_sendpkt(inm, IGMP_V2_LEAVE_GROUP, igmp_all_rtrs_group);
385 static void
386 igmp_fasttimo_ipi(void *arg __unused)
388 struct lwkt_msg *msg = &igmp_fasttimo_netmsg.lmsg;
390 crit_enter();
391 if (msg->ms_flags & MSGF_DONE)
392 lwkt_sendmsg_oncpu(netisr_cpuport(0), msg);
393 crit_exit();
396 void
397 igmp_fasttimo(void)
399 lwkt_send_ipiq_bycpu(0, igmp_fasttimo_ipi, NULL);
402 static void
403 igmp_fasttimo_dispatch(netmsg_t nmsg)
405 struct in_multi *inm;
406 struct in_multistep step;
408 crit_enter();
409 lwkt_replymsg(&nmsg->lmsg, 0); /* reply ASAP */
410 crit_exit();
413 * Quick check to see if any work needs to be done, in order
414 * to minimize the overhead of fasttimo processing.
417 if (!igmp_timers_are_running)
418 return;
420 igmp_timers_are_running = 0;
421 IN_FIRST_MULTI(step, inm);
422 while (inm != NULL) {
423 if (inm->inm_timer == 0) {
424 /* do nothing */
425 } else if (--inm->inm_timer == 0) {
426 igmp_sendpkt(inm, inm->inm_rti->rti_type, 0);
427 inm->inm_state = IGMP_IREPORTEDLAST;
428 } else {
429 igmp_timers_are_running = 1;
431 IN_NEXT_MULTI(step, inm);
435 static void
436 igmp_slowtimo_ipi(void *arg __unused)
438 struct lwkt_msg *msg = &igmp_slowtimo_netmsg.lmsg;
440 crit_enter();
441 if (msg->ms_flags & MSGF_DONE)
442 lwkt_sendmsg_oncpu(netisr_cpuport(0), msg);
443 crit_exit();
446 void
447 igmp_slowtimo(void)
449 lwkt_send_ipiq_bycpu(0, igmp_slowtimo_ipi, NULL);
452 static void
453 igmp_slowtimo_dispatch(netmsg_t nmsg)
455 struct router_info *rti = Head;
457 crit_enter();
458 lwkt_replymsg(&nmsg->lmsg, 0); /* reply ASAP */
459 crit_exit();
461 #ifdef IGMP_DEBUG
462 kprintf("[igmp.c,_slowtimo] -- > entering \n");
463 #endif
464 while (rti) {
465 if (rti->rti_type == IGMP_V1_ROUTER) {
466 rti->rti_time++;
467 if (rti->rti_time >= IGMP_AGE_THRESHOLD) {
468 rti->rti_type = IGMP_V2_ROUTER;
471 rti = rti->rti_next;
473 #ifdef IGMP_DEBUG
474 kprintf("[igmp.c,_slowtimo] -- > exiting \n");
475 #endif
478 static struct route igmprt;
480 static void
481 igmp_sendpkt(struct in_multi *inm, int type, unsigned long addr)
483 struct mbuf *m;
484 struct igmp *igmp;
485 struct ip *ip;
486 struct ip_moptions imo;
488 MGETHDR(m, M_NOWAIT, MT_HEADER);
489 if (m == NULL)
490 return;
492 m->m_pkthdr.rcvif = loif;
493 m->m_pkthdr.len = sizeof(struct ip) + IGMP_MINLEN;
494 MH_ALIGN(m, IGMP_MINLEN + sizeof(struct ip));
495 m->m_data += sizeof(struct ip);
496 m->m_len = IGMP_MINLEN;
497 igmp = mtod(m, struct igmp *);
498 igmp->igmp_type = type;
499 igmp->igmp_code = 0;
500 igmp->igmp_group = inm->inm_addr;
501 igmp->igmp_cksum = 0;
502 igmp->igmp_cksum = in_cksum(m, IGMP_MINLEN);
504 m->m_data -= sizeof(struct ip);
505 m->m_len += sizeof(struct ip);
506 ip = mtod(m, struct ip *);
507 ip->ip_tos = 0;
508 ip->ip_len = sizeof(struct ip) + IGMP_MINLEN;
509 ip->ip_off = 0;
510 ip->ip_p = IPPROTO_IGMP;
511 ip->ip_src.s_addr = INADDR_ANY;
512 ip->ip_dst.s_addr = addr ? addr : igmp->igmp_group.s_addr;
514 imo.imo_multicast_ifp = inm->inm_ifp;
515 imo.imo_multicast_ttl = 1;
516 imo.imo_multicast_vif = -1;
518 * Request loopback of the report if we are acting as a multicast
519 * router, so that the process-level routing demon can hear it.
521 imo.imo_multicast_loop = (ip_mrouter != NULL);
524 * XXX
525 * Do we have to worry about reentrancy here? Don't think so.
527 ip_output(m, router_alert, &igmprt, 0, &imo, NULL);
529 ++igmpstat.igps_snd_reports;