ip_mroute.c revision 1.51 1 1.51 ad /* $NetBSD: ip_mroute.c,v 1.51 2000/11/08 14:28:15 ad Exp $ */
2 1.13 cgd
3 1.1 hpeyerl /*
4 1.15 mycroft * IP multicast forwarding procedures
5 1.1 hpeyerl *
6 1.1 hpeyerl * Written by David Waitzman, BBN Labs, August 1988.
7 1.1 hpeyerl * Modified by Steve Deering, Stanford, February 1989.
8 1.15 mycroft * Modified by Mark J. Steiglitz, Stanford, May, 1991
9 1.15 mycroft * Modified by Van Jacobson, LBL, January 1993
10 1.15 mycroft * Modified by Ajit Thyagarajan, PARC, August 1993
11 1.15 mycroft * Modified by Bill Fenner, PARC, April 1994
12 1.15 mycroft * Modified by Charles M. Hannum, NetBSD, May 1995.
13 1.1 hpeyerl *
14 1.15 mycroft * MROUTING Revision: 1.2
15 1.1 hpeyerl */
16 1.44 thorpej
17 1.44 thorpej #include "opt_ipsec.h"
18 1.1 hpeyerl
19 1.1 hpeyerl #include <sys/param.h>
20 1.15 mycroft #include <sys/systm.h>
21 1.47 thorpej #include <sys/callout.h>
22 1.1 hpeyerl #include <sys/mbuf.h>
23 1.1 hpeyerl #include <sys/socket.h>
24 1.1 hpeyerl #include <sys/socketvar.h>
25 1.15 mycroft #include <sys/protosw.h>
26 1.15 mycroft #include <sys/errno.h>
27 1.1 hpeyerl #include <sys/time.h>
28 1.15 mycroft #include <sys/kernel.h>
29 1.15 mycroft #include <sys/ioctl.h>
30 1.15 mycroft #include <sys/syslog.h>
31 1.1 hpeyerl #include <net/if.h>
32 1.1 hpeyerl #include <net/route.h>
33 1.1 hpeyerl #include <net/raw_cb.h>
34 1.1 hpeyerl #include <netinet/in.h>
35 1.15 mycroft #include <netinet/in_var.h>
36 1.1 hpeyerl #include <netinet/in_systm.h>
37 1.1 hpeyerl #include <netinet/ip.h>
38 1.15 mycroft #include <netinet/ip_var.h>
39 1.1 hpeyerl #include <netinet/in_pcb.h>
40 1.15 mycroft #include <netinet/udp.h>
41 1.1 hpeyerl #include <netinet/igmp.h>
42 1.1 hpeyerl #include <netinet/igmp_var.h>
43 1.1 hpeyerl #include <netinet/ip_mroute.h>
44 1.37 hwr
45 1.25 christos #include <machine/stdarg.h>
46 1.25 christos
47 1.15 mycroft #define IP_MULTICASTOPTS 0
48 1.15 mycroft #define M_PULLUP(m, len) \
49 1.15 mycroft do { \
50 1.15 mycroft if ((m) && ((m)->m_flags & M_EXT || (m)->m_len < (len))) \
51 1.15 mycroft (m) = m_pullup((m), (len)); \
52 1.15 mycroft } while (0)
53 1.1 hpeyerl
54 1.1 hpeyerl /*
55 1.1 hpeyerl * Globals. All but ip_mrouter and ip_mrtproto could be static,
56 1.1 hpeyerl * except for netstat or debugging purposes.
57 1.1 hpeyerl */
58 1.30 mycroft struct socket *ip_mrouter = 0;
59 1.15 mycroft int ip_mrtproto = IGMP_DVMRP; /* for netstat only */
60 1.15 mycroft
61 1.15 mycroft #define NO_RTE_FOUND 0x1
62 1.15 mycroft #define RTE_FOUND 0x2
63 1.1 hpeyerl
64 1.15 mycroft #define MFCHASH(a, g) \
65 1.29 mycroft ((((a).s_addr >> 20) ^ ((a).s_addr >> 10) ^ (a).s_addr ^ \
66 1.29 mycroft ((g).s_addr >> 20) ^ ((g).s_addr >> 10) ^ (g).s_addr) & mfchash)
67 1.15 mycroft LIST_HEAD(mfchashhdr, mfc) *mfchashtbl;
68 1.15 mycroft u_long mfchash;
69 1.15 mycroft
70 1.15 mycroft u_char nexpire[MFCTBLSIZ];
71 1.15 mycroft struct vif viftable[MAXVIFS];
72 1.15 mycroft struct mrtstat mrtstat;
73 1.15 mycroft u_int mrtdebug = 0; /* debug level */
74 1.15 mycroft #define DEBUG_MFC 0x02
75 1.15 mycroft #define DEBUG_FORWARD 0x04
76 1.15 mycroft #define DEBUG_EXPIRE 0x08
77 1.15 mycroft #define DEBUG_XMIT 0x10
78 1.15 mycroft u_int tbfdebug = 0; /* tbf debug level */
79 1.15 mycroft #ifdef RSVP_ISI
80 1.15 mycroft u_int rsvpdebug = 0; /* rsvp debug level */
81 1.15 mycroft extern struct socket *ip_rsvpd;
82 1.15 mycroft extern int rsvp_on;
83 1.15 mycroft #endif /* RSVP_ISI */
84 1.15 mycroft
85 1.15 mycroft #define EXPIRE_TIMEOUT (hz / 4) /* 4x / second */
86 1.15 mycroft #define UPCALL_EXPIRE 6 /* number of timeouts */
87 1.15 mycroft
88 1.15 mycroft /*
89 1.15 mycroft * Define the token bucket filter structures
90 1.15 mycroft */
91 1.15 mycroft
92 1.31 mycroft #define TBF_REPROCESS (hz / 100) /* 100x / second */
93 1.15 mycroft
94 1.25 christos static int get_sg_cnt __P((struct sioc_sg_req *));
95 1.25 christos static int get_vif_cnt __P((struct sioc_vif_req *));
96 1.25 christos static int ip_mrouter_init __P((struct socket *, struct mbuf *));
97 1.25 christos static int get_version __P((struct mbuf *));
98 1.25 christos static int set_assert __P((struct mbuf *));
99 1.25 christos static int get_assert __P((struct mbuf *));
100 1.25 christos static int add_vif __P((struct mbuf *));
101 1.25 christos static int del_vif __P((struct mbuf *));
102 1.25 christos static void update_mfc __P((struct mfcctl *, struct mfc *));
103 1.25 christos static void expire_mfc __P((struct mfc *));
104 1.25 christos static int add_mfc __P((struct mbuf *));
105 1.25 christos #ifdef UPCALL_TIMING
106 1.25 christos static void collate __P((struct timeval *));
107 1.25 christos #endif
108 1.25 christos static int del_mfc __P((struct mbuf *));
109 1.25 christos static int socket_send __P((struct socket *, struct mbuf *,
110 1.25 christos struct sockaddr_in *));
111 1.25 christos static void expire_upcalls __P((void *));
112 1.25 christos #ifdef RSVP_ISI
113 1.25 christos static int ip_mdq __P((struct mbuf *, struct ifnet *, struct mfc *, vifi_t));
114 1.25 christos #else
115 1.25 christos static int ip_mdq __P((struct mbuf *, struct ifnet *, struct mfc *));
116 1.25 christos #endif
117 1.25 christos static void phyint_send __P((struct ip *, struct vif *, struct mbuf *));
118 1.25 christos static void encap_send __P((struct ip *, struct vif *, struct mbuf *));
119 1.25 christos static void tbf_control __P((struct vif *, struct mbuf *, struct ip *,
120 1.25 christos u_int32_t));
121 1.31 mycroft static void tbf_queue __P((struct vif *, struct mbuf *));
122 1.25 christos static void tbf_process_q __P((struct vif *));
123 1.25 christos static void tbf_reprocess_q __P((void *));
124 1.25 christos static int tbf_dq_sel __P((struct vif *, struct ip *));
125 1.25 christos static void tbf_send_packet __P((struct vif *, struct mbuf *));
126 1.25 christos static void tbf_update_tokens __P((struct vif *));
127 1.25 christos static int priority __P((struct vif *, struct ip *));
128 1.1 hpeyerl
129 1.1 hpeyerl /*
130 1.12 brezak * 'Interfaces' associated with decapsulator (so we can tell
131 1.12 brezak * packets that went through it from ones that get reflected
132 1.12 brezak * by a broken gateway). These interfaces are never linked into
133 1.12 brezak * the system ifnet list & no routes point to them. I.e., packets
134 1.12 brezak * can't be sent this way. They only exist as a placeholder for
135 1.12 brezak * multicast source verification.
136 1.12 brezak */
137 1.17 mycroft #if 0
138 1.12 brezak struct ifnet multicast_decap_if[MAXVIFS];
139 1.17 mycroft #endif
140 1.12 brezak
141 1.17 mycroft #define ENCAP_TTL 64
142 1.17 mycroft #define ENCAP_PROTO IPPROTO_IPIP /* 4 */
143 1.12 brezak
144 1.12 brezak /* prototype IP hdr for encapsulated packets */
145 1.12 brezak struct ip multicast_encap_iphdr = {
146 1.15 mycroft #if BYTE_ORDER == LITTLE_ENDIAN
147 1.12 brezak sizeof(struct ip) >> 2, IPVERSION,
148 1.12 brezak #else
149 1.12 brezak IPVERSION, sizeof(struct ip) >> 2,
150 1.12 brezak #endif
151 1.12 brezak 0, /* tos */
152 1.12 brezak sizeof(struct ip), /* total length */
153 1.12 brezak 0, /* id */
154 1.12 brezak 0, /* frag offset */
155 1.15 mycroft ENCAP_TTL, ENCAP_PROTO,
156 1.12 brezak 0, /* checksum */
157 1.12 brezak };
158 1.12 brezak
159 1.12 brezak /*
160 1.1 hpeyerl * Private variables.
161 1.1 hpeyerl */
162 1.15 mycroft static vifi_t numvifs = 0;
163 1.15 mycroft static int have_encap_tunnel = 0;
164 1.12 brezak
165 1.47 thorpej static struct callout expire_upcalls_ch;
166 1.47 thorpej
167 1.12 brezak /*
168 1.38 thorpej * one-back cache used by mrt_ipip_input to locate a tunnel's vif
169 1.12 brezak * given a datagram's src ip address.
170 1.12 brezak */
171 1.29 mycroft static struct in_addr last_encap_src;
172 1.12 brezak static struct vif *last_encap_vif;
173 1.12 brezak
174 1.12 brezak /*
175 1.15 mycroft * whether or not special PIM assert processing is enabled.
176 1.15 mycroft */
177 1.15 mycroft static int pim_assert;
178 1.15 mycroft /*
179 1.15 mycroft * Rate limit for assert notification messages, in usec
180 1.12 brezak */
181 1.15 mycroft #define ASSERT_MSG_TIME 3000000
182 1.12 brezak
183 1.15 mycroft /*
184 1.15 mycroft * Find a route for a given origin IP address and Multicast group address
185 1.15 mycroft * Type of service parameter to be added in the future!!!
186 1.15 mycroft */
187 1.15 mycroft
188 1.15 mycroft #define MFCFIND(o, g, rt) { \
189 1.48 augustss struct mfc *_rt; \
190 1.30 mycroft (rt) = 0; \
191 1.15 mycroft ++mrtstat.mrts_mfc_lookups; \
192 1.15 mycroft for (_rt = mfchashtbl[MFCHASH(o, g)].lh_first; \
193 1.15 mycroft _rt; _rt = _rt->mfc_hash.le_next) { \
194 1.29 mycroft if (in_hosteq(_rt->mfc_origin, (o)) && \
195 1.29 mycroft in_hosteq(_rt->mfc_mcastgrp, (g)) && \
196 1.30 mycroft _rt->mfc_stall == 0) { \
197 1.15 mycroft (rt) = _rt; \
198 1.15 mycroft break; \
199 1.15 mycroft } \
200 1.15 mycroft } \
201 1.30 mycroft if ((rt) == 0) \
202 1.15 mycroft ++mrtstat.mrts_mfc_misses; \
203 1.12 brezak }
204 1.12 brezak
205 1.12 brezak /*
206 1.15 mycroft * Macros to compute elapsed time efficiently
207 1.15 mycroft * Borrowed from Van Jacobson's scheduling code
208 1.12 brezak */
209 1.15 mycroft #define TV_DELTA(a, b, delta) { \
210 1.48 augustss int xxs; \
211 1.15 mycroft delta = (a).tv_usec - (b).tv_usec; \
212 1.15 mycroft xxs = (a).tv_sec - (b).tv_sec; \
213 1.15 mycroft switch (xxs) { \
214 1.15 mycroft case 2: \
215 1.15 mycroft delta += 1000000; \
216 1.15 mycroft /* fall through */ \
217 1.15 mycroft case 1: \
218 1.15 mycroft delta += 1000000; \
219 1.15 mycroft /* fall through */ \
220 1.15 mycroft case 0: \
221 1.15 mycroft break; \
222 1.15 mycroft default: \
223 1.15 mycroft delta += (1000000 * xxs); \
224 1.15 mycroft break; \
225 1.15 mycroft } \
226 1.15 mycroft }
227 1.15 mycroft
228 1.15 mycroft #ifdef UPCALL_TIMING
229 1.15 mycroft u_int32_t upcall_data[51];
230 1.15 mycroft #endif /* UPCALL_TIMING */
231 1.15 mycroft
232 1.12 brezak /*
233 1.15 mycroft * Handle MRT setsockopt commands to modify the multicast routing tables.
234 1.12 brezak */
235 1.15 mycroft int
236 1.28 mycroft ip_mrouter_set(so, optname, m)
237 1.15 mycroft struct socket *so;
238 1.28 mycroft int optname;
239 1.15 mycroft struct mbuf **m;
240 1.15 mycroft {
241 1.15 mycroft int error;
242 1.15 mycroft
243 1.28 mycroft if (optname != MRT_INIT && so != ip_mrouter)
244 1.28 mycroft error = ENOPROTOOPT;
245 1.15 mycroft else
246 1.28 mycroft switch (optname) {
247 1.15 mycroft case MRT_INIT:
248 1.15 mycroft error = ip_mrouter_init(so, *m);
249 1.15 mycroft break;
250 1.15 mycroft case MRT_DONE:
251 1.15 mycroft error = ip_mrouter_done();
252 1.15 mycroft break;
253 1.15 mycroft case MRT_ADD_VIF:
254 1.15 mycroft error = add_vif(*m);
255 1.15 mycroft break;
256 1.15 mycroft case MRT_DEL_VIF:
257 1.15 mycroft error = del_vif(*m);
258 1.15 mycroft break;
259 1.15 mycroft case MRT_ADD_MFC:
260 1.15 mycroft error = add_mfc(*m);
261 1.15 mycroft break;
262 1.15 mycroft case MRT_DEL_MFC:
263 1.15 mycroft error = del_mfc(*m);
264 1.15 mycroft break;
265 1.15 mycroft case MRT_ASSERT:
266 1.15 mycroft error = set_assert(*m);
267 1.15 mycroft break;
268 1.15 mycroft default:
269 1.28 mycroft error = ENOPROTOOPT;
270 1.15 mycroft break;
271 1.15 mycroft }
272 1.15 mycroft
273 1.15 mycroft if (*m)
274 1.15 mycroft m_free(*m);
275 1.15 mycroft return (error);
276 1.12 brezak }
277 1.12 brezak
278 1.15 mycroft /*
279 1.15 mycroft * Handle MRT getsockopt commands
280 1.15 mycroft */
281 1.15 mycroft int
282 1.28 mycroft ip_mrouter_get(so, optname, m)
283 1.15 mycroft struct socket *so;
284 1.28 mycroft int optname;
285 1.15 mycroft struct mbuf **m;
286 1.12 brezak {
287 1.15 mycroft int error;
288 1.12 brezak
289 1.15 mycroft if (so != ip_mrouter)
290 1.28 mycroft error = ENOPROTOOPT;
291 1.15 mycroft else {
292 1.28 mycroft *m = m_get(M_WAIT, MT_SOOPTS);
293 1.12 brezak
294 1.28 mycroft switch (optname) {
295 1.15 mycroft case MRT_VERSION:
296 1.28 mycroft error = get_version(*m);
297 1.15 mycroft break;
298 1.15 mycroft case MRT_ASSERT:
299 1.28 mycroft error = get_assert(*m);
300 1.15 mycroft break;
301 1.15 mycroft default:
302 1.28 mycroft error = ENOPROTOOPT;
303 1.15 mycroft break;
304 1.15 mycroft }
305 1.15 mycroft
306 1.15 mycroft if (error)
307 1.28 mycroft m_free(*m);
308 1.12 brezak }
309 1.15 mycroft
310 1.15 mycroft return (error);
311 1.12 brezak }
312 1.12 brezak
313 1.1 hpeyerl /*
314 1.15 mycroft * Handle ioctl commands to obtain information from the cache
315 1.1 hpeyerl */
316 1.1 hpeyerl int
317 1.28 mycroft mrt_ioctl(so, cmd, data)
318 1.28 mycroft struct socket *so;
319 1.15 mycroft u_long cmd;
320 1.15 mycroft caddr_t data;
321 1.1 hpeyerl {
322 1.15 mycroft int error;
323 1.1 hpeyerl
324 1.28 mycroft if (so != ip_mrouter)
325 1.15 mycroft error = EINVAL;
326 1.28 mycroft else
327 1.28 mycroft switch (cmd) {
328 1.28 mycroft case SIOCGETVIFCNT:
329 1.28 mycroft error = get_vif_cnt((struct sioc_vif_req *)data);
330 1.28 mycroft break;
331 1.28 mycroft case SIOCGETSGCNT:
332 1.28 mycroft error = get_sg_cnt((struct sioc_sg_req *)data);
333 1.28 mycroft break;
334 1.28 mycroft default:
335 1.28 mycroft error = EINVAL;
336 1.28 mycroft break;
337 1.28 mycroft }
338 1.1 hpeyerl
339 1.15 mycroft return (error);
340 1.15 mycroft }
341 1.1 hpeyerl
342 1.15 mycroft /*
343 1.15 mycroft * returns the packet, byte, rpf-failure count for the source group provided
344 1.15 mycroft */
345 1.15 mycroft static int
346 1.15 mycroft get_sg_cnt(req)
347 1.48 augustss struct sioc_sg_req *req;
348 1.15 mycroft {
349 1.48 augustss struct mfc *rt;
350 1.15 mycroft int s;
351 1.1 hpeyerl
352 1.24 mycroft s = splsoftnet();
353 1.29 mycroft MFCFIND(req->src, req->grp, rt);
354 1.15 mycroft splx(s);
355 1.30 mycroft if (rt != 0) {
356 1.15 mycroft req->pktcnt = rt->mfc_pkt_cnt;
357 1.15 mycroft req->bytecnt = rt->mfc_byte_cnt;
358 1.15 mycroft req->wrong_if = rt->mfc_wrong_if;
359 1.15 mycroft } else
360 1.15 mycroft req->pktcnt = req->bytecnt = req->wrong_if = 0xffffffff;
361 1.1 hpeyerl
362 1.15 mycroft return (0);
363 1.15 mycroft }
364 1.1 hpeyerl
365 1.15 mycroft /*
366 1.15 mycroft * returns the input and output packet and byte counts on the vif provided
367 1.15 mycroft */
368 1.15 mycroft static int
369 1.15 mycroft get_vif_cnt(req)
370 1.48 augustss struct sioc_vif_req *req;
371 1.15 mycroft {
372 1.48 augustss vifi_t vifi = req->vifi;
373 1.1 hpeyerl
374 1.15 mycroft if (vifi >= numvifs)
375 1.15 mycroft return (EINVAL);
376 1.1 hpeyerl
377 1.15 mycroft req->icount = viftable[vifi].v_pkt_in;
378 1.15 mycroft req->ocount = viftable[vifi].v_pkt_out;
379 1.15 mycroft req->ibytes = viftable[vifi].v_bytes_in;
380 1.15 mycroft req->obytes = viftable[vifi].v_bytes_out;
381 1.1 hpeyerl
382 1.15 mycroft return (0);
383 1.1 hpeyerl }
384 1.1 hpeyerl
385 1.1 hpeyerl /*
386 1.1 hpeyerl * Enable multicast routing
387 1.1 hpeyerl */
388 1.1 hpeyerl static int
389 1.15 mycroft ip_mrouter_init(so, m)
390 1.15 mycroft struct socket *so;
391 1.15 mycroft struct mbuf *m;
392 1.1 hpeyerl {
393 1.15 mycroft int *v;
394 1.15 mycroft
395 1.15 mycroft if (mrtdebug)
396 1.15 mycroft log(LOG_DEBUG,
397 1.30 mycroft "ip_mrouter_init: so_type = %d, pr_protocol = %d\n",
398 1.15 mycroft so->so_type, so->so_proto->pr_protocol);
399 1.15 mycroft
400 1.1 hpeyerl if (so->so_type != SOCK_RAW ||
401 1.1 hpeyerl so->so_proto->pr_protocol != IPPROTO_IGMP)
402 1.1 hpeyerl return (EOPNOTSUPP);
403 1.1 hpeyerl
404 1.15 mycroft if (m == 0 || m->m_len < sizeof(int))
405 1.15 mycroft return (EINVAL);
406 1.15 mycroft
407 1.15 mycroft v = mtod(m, int *);
408 1.15 mycroft if (*v != 1)
409 1.15 mycroft return (EINVAL);
410 1.15 mycroft
411 1.30 mycroft if (ip_mrouter != 0)
412 1.1 hpeyerl return (EADDRINUSE);
413 1.1 hpeyerl
414 1.1 hpeyerl ip_mrouter = so;
415 1.1 hpeyerl
416 1.51 ad mfchashtbl =
417 1.51 ad hashinit(MFCTBLSIZ, HASH_LIST, M_MRTABLE, M_WAITOK, &mfchash);
418 1.15 mycroft bzero((caddr_t)nexpire, sizeof(nexpire));
419 1.15 mycroft
420 1.15 mycroft pim_assert = 0;
421 1.15 mycroft
422 1.47 thorpej callout_init(&expire_upcalls_ch);
423 1.47 thorpej callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
424 1.47 thorpej expire_upcalls, NULL);
425 1.15 mycroft
426 1.15 mycroft if (mrtdebug)
427 1.30 mycroft log(LOG_DEBUG, "ip_mrouter_init\n");
428 1.15 mycroft
429 1.1 hpeyerl return (0);
430 1.1 hpeyerl }
431 1.1 hpeyerl
432 1.1 hpeyerl /*
433 1.1 hpeyerl * Disable multicast routing
434 1.1 hpeyerl */
435 1.1 hpeyerl int
436 1.1 hpeyerl ip_mrouter_done()
437 1.1 hpeyerl {
438 1.15 mycroft vifi_t vifi;
439 1.48 augustss struct vif *vifp;
440 1.15 mycroft int i;
441 1.15 mycroft int s;
442 1.15 mycroft
443 1.24 mycroft s = splsoftnet();
444 1.1 hpeyerl
445 1.17 mycroft /* Clear out all the vifs currently in use. */
446 1.1 hpeyerl for (vifi = 0; vifi < numvifs; vifi++) {
447 1.15 mycroft vifp = &viftable[vifi];
448 1.29 mycroft if (!in_nullhost(vifp->v_lcl_addr))
449 1.17 mycroft reset_vif(vifp);
450 1.1 hpeyerl }
451 1.17 mycroft
452 1.1 hpeyerl numvifs = 0;
453 1.15 mycroft pim_assert = 0;
454 1.15 mycroft
455 1.47 thorpej callout_stop(&expire_upcalls_ch);
456 1.15 mycroft
457 1.15 mycroft /*
458 1.15 mycroft * Free all multicast forwarding cache entries.
459 1.15 mycroft */
460 1.15 mycroft for (i = 0; i < MFCTBLSIZ; i++) {
461 1.48 augustss struct mfc *rt, *nrt;
462 1.1 hpeyerl
463 1.15 mycroft for (rt = mfchashtbl[i].lh_first; rt; rt = nrt) {
464 1.15 mycroft nrt = rt->mfc_hash.le_next;
465 1.15 mycroft
466 1.15 mycroft expire_mfc(rt);
467 1.15 mycroft }
468 1.15 mycroft }
469 1.40 mycroft
470 1.15 mycroft free(mfchashtbl, M_MRTABLE);
471 1.40 mycroft mfchashtbl = 0;
472 1.15 mycroft
473 1.17 mycroft /* Reset de-encapsulation cache. */
474 1.15 mycroft have_encap_tunnel = 0;
475 1.15 mycroft
476 1.30 mycroft ip_mrouter = 0;
477 1.15 mycroft
478 1.15 mycroft splx(s);
479 1.15 mycroft
480 1.15 mycroft if (mrtdebug)
481 1.30 mycroft log(LOG_DEBUG, "ip_mrouter_done\n");
482 1.15 mycroft
483 1.15 mycroft return (0);
484 1.15 mycroft }
485 1.15 mycroft
486 1.15 mycroft static int
487 1.15 mycroft get_version(m)
488 1.15 mycroft struct mbuf *m;
489 1.15 mycroft {
490 1.15 mycroft int *v = mtod(m, int *);
491 1.15 mycroft
492 1.15 mycroft *v = 0x0305; /* XXX !!!! */
493 1.15 mycroft m->m_len = sizeof(int);
494 1.15 mycroft return (0);
495 1.15 mycroft }
496 1.15 mycroft
497 1.15 mycroft /*
498 1.15 mycroft * Set PIM assert processing global
499 1.15 mycroft */
500 1.15 mycroft static int
501 1.15 mycroft set_assert(m)
502 1.15 mycroft struct mbuf *m;
503 1.15 mycroft {
504 1.15 mycroft int *i;
505 1.15 mycroft
506 1.15 mycroft if (m == 0 || m->m_len < sizeof(int))
507 1.15 mycroft return (EINVAL);
508 1.1 hpeyerl
509 1.15 mycroft i = mtod(m, int *);
510 1.15 mycroft pim_assert = !!*i;
511 1.15 mycroft return (0);
512 1.15 mycroft }
513 1.15 mycroft
514 1.15 mycroft /*
515 1.15 mycroft * Get PIM assert processing global
516 1.15 mycroft */
517 1.15 mycroft static int
518 1.15 mycroft get_assert(m)
519 1.15 mycroft struct mbuf *m;
520 1.15 mycroft {
521 1.15 mycroft int *i = mtod(m, int *);
522 1.1 hpeyerl
523 1.15 mycroft *i = pim_assert;
524 1.15 mycroft m->m_len = sizeof(int);
525 1.1 hpeyerl return (0);
526 1.1 hpeyerl }
527 1.1 hpeyerl
528 1.15 mycroft static struct sockaddr_in sin = { sizeof(sin), AF_INET };
529 1.15 mycroft
530 1.1 hpeyerl /*
531 1.1 hpeyerl * Add a vif to the vif table
532 1.1 hpeyerl */
533 1.1 hpeyerl static int
534 1.15 mycroft add_vif(m)
535 1.15 mycroft struct mbuf *m;
536 1.15 mycroft {
537 1.48 augustss struct vifctl *vifcp;
538 1.48 augustss struct vif *vifp;
539 1.15 mycroft struct ifaddr *ifa;
540 1.15 mycroft struct ifnet *ifp;
541 1.1 hpeyerl struct ifreq ifr;
542 1.15 mycroft int error, s;
543 1.15 mycroft
544 1.15 mycroft if (m == 0 || m->m_len < sizeof(struct vifctl))
545 1.15 mycroft return (EINVAL);
546 1.1 hpeyerl
547 1.15 mycroft vifcp = mtod(m, struct vifctl *);
548 1.1 hpeyerl if (vifcp->vifc_vifi >= MAXVIFS)
549 1.1 hpeyerl return (EINVAL);
550 1.15 mycroft
551 1.15 mycroft vifp = &viftable[vifcp->vifc_vifi];
552 1.29 mycroft if (!in_nullhost(vifp->v_lcl_addr))
553 1.1 hpeyerl return (EADDRINUSE);
554 1.15 mycroft
555 1.15 mycroft /* Find the interface with an address in AF_INET family. */
556 1.1 hpeyerl sin.sin_addr = vifcp->vifc_lcl_addr;
557 1.18 mycroft ifa = ifa_ifwithaddr(sintosa(&sin));
558 1.1 hpeyerl if (ifa == 0)
559 1.1 hpeyerl return (EADDRNOTAVAIL);
560 1.15 mycroft
561 1.12 brezak if (vifcp->vifc_flags & VIFF_TUNNEL) {
562 1.17 mycroft if (vifcp->vifc_flags & VIFF_SRCRT) {
563 1.30 mycroft log(LOG_ERR, "Source routed tunnels not supported\n");
564 1.15 mycroft return (EOPNOTSUPP);
565 1.12 brezak }
566 1.17 mycroft
567 1.17 mycroft /* Create a fake encapsulation interface. */
568 1.17 mycroft ifp = (struct ifnet *)malloc(sizeof(*ifp), M_MRTABLE, M_WAITOK);
569 1.17 mycroft bzero(ifp, sizeof(*ifp));
570 1.34 christos sprintf(ifp->if_xname, "mdecap%d", vifcp->vifc_vifi);
571 1.17 mycroft
572 1.17 mycroft /* Prepare cached route entry. */
573 1.17 mycroft bzero(&vifp->v_route, sizeof(vifp->v_route));
574 1.17 mycroft
575 1.38 thorpej /* Tell mrt_ipip_input() to start looking at encapsulated packets. */
576 1.17 mycroft have_encap_tunnel = 1;
577 1.12 brezak } else {
578 1.17 mycroft /* Use the physical interface associated with the address. */
579 1.17 mycroft ifp = ifa->ifa_ifp;
580 1.17 mycroft
581 1.15 mycroft /* Make sure the interface supports multicast. */
582 1.12 brezak if ((ifp->if_flags & IFF_MULTICAST) == 0)
583 1.15 mycroft return (EOPNOTSUPP);
584 1.45 thorpej
585 1.15 mycroft /* Enable promiscuous reception of all IP multicasts. */
586 1.19 mycroft satosin(&ifr.ifr_addr)->sin_len = sizeof(struct sockaddr_in);
587 1.15 mycroft satosin(&ifr.ifr_addr)->sin_family = AF_INET;
588 1.29 mycroft satosin(&ifr.ifr_addr)->sin_addr = zeroin_addr;
589 1.1 hpeyerl error = (*ifp->if_ioctl)(ifp, SIOCADDMULTI, (caddr_t)&ifr);
590 1.12 brezak if (error)
591 1.15 mycroft return (error);
592 1.1 hpeyerl }
593 1.45 thorpej
594 1.24 mycroft s = splsoftnet();
595 1.31 mycroft
596 1.15 mycroft /* Define parameters for the tbf structure. */
597 1.31 mycroft vifp->tbf_q = 0;
598 1.31 mycroft vifp->tbf_t = &vifp->tbf_q;
599 1.31 mycroft microtime(&vifp->tbf_last_pkt_t);
600 1.31 mycroft vifp->tbf_n_tok = 0;
601 1.31 mycroft vifp->tbf_q_len = 0;
602 1.31 mycroft vifp->tbf_max_q_len = MAXQSIZE;
603 1.15 mycroft
604 1.1 hpeyerl vifp->v_flags = vifcp->vifc_flags;
605 1.1 hpeyerl vifp->v_threshold = vifcp->vifc_threshold;
606 1.31 mycroft /* scaling up here allows division by 1024 in critical code */
607 1.31 mycroft vifp->v_rate_limit = vifcp->vifc_rate_limit * 1024 / 1000;
608 1.1 hpeyerl vifp->v_lcl_addr = vifcp->vifc_lcl_addr;
609 1.15 mycroft vifp->v_rmt_addr = vifcp->vifc_rmt_addr;
610 1.12 brezak vifp->v_ifp = ifp;
611 1.15 mycroft /* Initialize per vif pkt counters. */
612 1.15 mycroft vifp->v_pkt_in = 0;
613 1.15 mycroft vifp->v_pkt_out = 0;
614 1.15 mycroft vifp->v_bytes_in = 0;
615 1.15 mycroft vifp->v_bytes_out = 0;
616 1.47 thorpej
617 1.47 thorpej callout_init(&vifp->v_repq_ch);
618 1.47 thorpej
619 1.31 mycroft #ifdef RSVP_ISI
620 1.31 mycroft vifp->v_rsvp_on = 0;
621 1.31 mycroft vifp->v_rsvpd = 0;
622 1.31 mycroft #endif /* RSVP_ISI */
623 1.31 mycroft
624 1.12 brezak splx(s);
625 1.15 mycroft
626 1.15 mycroft /* Adjust numvifs up if the vifi is higher than numvifs. */
627 1.1 hpeyerl if (numvifs <= vifcp->vifc_vifi)
628 1.1 hpeyerl numvifs = vifcp->vifc_vifi + 1;
629 1.15 mycroft
630 1.15 mycroft if (mrtdebug)
631 1.30 mycroft log(LOG_DEBUG, "add_vif #%d, lcladdr %x, %s %x, thresh %x, rate %d\n",
632 1.15 mycroft vifcp->vifc_vifi,
633 1.15 mycroft ntohl(vifcp->vifc_lcl_addr.s_addr),
634 1.15 mycroft (vifcp->vifc_flags & VIFF_TUNNEL) ? "rmtaddr" : "mask",
635 1.15 mycroft ntohl(vifcp->vifc_rmt_addr.s_addr),
636 1.15 mycroft vifcp->vifc_threshold,
637 1.15 mycroft vifcp->vifc_rate_limit);
638 1.15 mycroft
639 1.1 hpeyerl return (0);
640 1.1 hpeyerl }
641 1.1 hpeyerl
642 1.17 mycroft void
643 1.17 mycroft reset_vif(vifp)
644 1.48 augustss struct vif *vifp;
645 1.17 mycroft {
646 1.48 augustss struct mbuf *m, *n;
647 1.17 mycroft struct ifnet *ifp;
648 1.17 mycroft struct ifreq ifr;
649 1.17 mycroft
650 1.47 thorpej callout_stop(&vifp->v_repq_ch);
651 1.47 thorpej
652 1.31 mycroft for (m = vifp->tbf_q; m != 0; m = n) {
653 1.31 mycroft n = m->m_nextpkt;
654 1.31 mycroft m_freem(m);
655 1.31 mycroft }
656 1.31 mycroft
657 1.17 mycroft if (vifp->v_flags & VIFF_TUNNEL) {
658 1.17 mycroft free(vifp->v_ifp, M_MRTABLE);
659 1.17 mycroft if (vifp == last_encap_vif) {
660 1.17 mycroft last_encap_vif = 0;
661 1.29 mycroft last_encap_src = zeroin_addr;
662 1.17 mycroft }
663 1.17 mycroft } else {
664 1.19 mycroft satosin(&ifr.ifr_addr)->sin_len = sizeof(struct sockaddr_in);
665 1.17 mycroft satosin(&ifr.ifr_addr)->sin_family = AF_INET;
666 1.29 mycroft satosin(&ifr.ifr_addr)->sin_addr = zeroin_addr;
667 1.17 mycroft ifp = vifp->v_ifp;
668 1.17 mycroft (*ifp->if_ioctl)(ifp, SIOCDELMULTI, (caddr_t)&ifr);
669 1.17 mycroft }
670 1.17 mycroft bzero((caddr_t)vifp, sizeof(*vifp));
671 1.17 mycroft }
672 1.17 mycroft
673 1.1 hpeyerl /*
674 1.1 hpeyerl * Delete a vif from the vif table
675 1.1 hpeyerl */
676 1.1 hpeyerl static int
677 1.15 mycroft del_vif(m)
678 1.15 mycroft struct mbuf *m;
679 1.1 hpeyerl {
680 1.15 mycroft vifi_t *vifip;
681 1.48 augustss struct vif *vifp;
682 1.48 augustss vifi_t vifi;
683 1.15 mycroft int s;
684 1.15 mycroft
685 1.15 mycroft if (m == 0 || m->m_len < sizeof(vifi_t))
686 1.15 mycroft return (EINVAL);
687 1.1 hpeyerl
688 1.15 mycroft vifip = mtod(m, vifi_t *);
689 1.1 hpeyerl if (*vifip >= numvifs)
690 1.1 hpeyerl return (EINVAL);
691 1.15 mycroft
692 1.15 mycroft vifp = &viftable[*vifip];
693 1.29 mycroft if (in_nullhost(vifp->v_lcl_addr))
694 1.1 hpeyerl return (EADDRNOTAVAIL);
695 1.15 mycroft
696 1.24 mycroft s = splsoftnet();
697 1.15 mycroft
698 1.17 mycroft reset_vif(vifp);
699 1.15 mycroft
700 1.1 hpeyerl /* Adjust numvifs down */
701 1.15 mycroft for (vifi = numvifs; vifi > 0; vifi--)
702 1.29 mycroft if (!in_nullhost(viftable[vifi-1].v_lcl_addr))
703 1.1 hpeyerl break;
704 1.15 mycroft numvifs = vifi;
705 1.15 mycroft
706 1.1 hpeyerl splx(s);
707 1.15 mycroft
708 1.15 mycroft if (mrtdebug)
709 1.30 mycroft log(LOG_DEBUG, "del_vif %d, numvifs %d\n", *vifip, numvifs);
710 1.15 mycroft
711 1.1 hpeyerl return (0);
712 1.1 hpeyerl }
713 1.1 hpeyerl
714 1.15 mycroft static void
715 1.15 mycroft update_mfc(mfccp, rt)
716 1.15 mycroft struct mfcctl *mfccp;
717 1.15 mycroft struct mfc *rt;
718 1.1 hpeyerl {
719 1.15 mycroft vifi_t vifi;
720 1.1 hpeyerl
721 1.15 mycroft rt->mfc_parent = mfccp->mfcc_parent;
722 1.15 mycroft for (vifi = 0; vifi < numvifs; vifi++)
723 1.15 mycroft rt->mfc_ttls[vifi] = mfccp->mfcc_ttls[vifi];
724 1.15 mycroft rt->mfc_expire = 0;
725 1.15 mycroft rt->mfc_stall = 0;
726 1.15 mycroft }
727 1.1 hpeyerl
728 1.15 mycroft static void
729 1.15 mycroft expire_mfc(rt)
730 1.15 mycroft struct mfc *rt;
731 1.15 mycroft {
732 1.15 mycroft struct rtdetq *rte, *nrte;
733 1.1 hpeyerl
734 1.30 mycroft for (rte = rt->mfc_stall; rte != 0; rte = nrte) {
735 1.15 mycroft nrte = rte->next;
736 1.15 mycroft m_freem(rte->m);
737 1.15 mycroft free(rte, M_MRTABLE);
738 1.1 hpeyerl }
739 1.1 hpeyerl
740 1.15 mycroft LIST_REMOVE(rt, mfc_hash);
741 1.15 mycroft free(rt, M_MRTABLE);
742 1.1 hpeyerl }
743 1.1 hpeyerl
744 1.1 hpeyerl /*
745 1.15 mycroft * Add an mfc entry
746 1.1 hpeyerl */
747 1.1 hpeyerl static int
748 1.15 mycroft add_mfc(m)
749 1.15 mycroft struct mbuf *m;
750 1.1 hpeyerl {
751 1.15 mycroft struct mfcctl *mfccp;
752 1.25 christos struct mfc *rt;
753 1.25 christos u_int32_t hash = 0;
754 1.15 mycroft struct rtdetq *rte, *nrte;
755 1.48 augustss u_short nstl;
756 1.15 mycroft int s;
757 1.1 hpeyerl
758 1.15 mycroft if (m == 0 || m->m_len < sizeof(struct mfcctl))
759 1.1 hpeyerl return (EINVAL);
760 1.15 mycroft
761 1.15 mycroft mfccp = mtod(m, struct mfcctl *);
762 1.1 hpeyerl
763 1.24 mycroft s = splsoftnet();
764 1.29 mycroft MFCFIND(mfccp->mfcc_origin, mfccp->mfcc_mcastgrp, rt);
765 1.1 hpeyerl
766 1.15 mycroft /* If an entry already exists, just update the fields */
767 1.15 mycroft if (rt) {
768 1.15 mycroft if (mrtdebug & DEBUG_MFC)
769 1.30 mycroft log(LOG_DEBUG,"add_mfc update o %x g %x p %x\n",
770 1.15 mycroft ntohl(mfccp->mfcc_origin.s_addr),
771 1.15 mycroft ntohl(mfccp->mfcc_mcastgrp.s_addr),
772 1.15 mycroft mfccp->mfcc_parent);
773 1.1 hpeyerl
774 1.15 mycroft if (rt->mfc_expire)
775 1.15 mycroft nexpire[hash]--;
776 1.1 hpeyerl
777 1.15 mycroft update_mfc(mfccp, rt);
778 1.1 hpeyerl
779 1.15 mycroft splx(s);
780 1.15 mycroft return (0);
781 1.15 mycroft }
782 1.1 hpeyerl
783 1.15 mycroft /*
784 1.15 mycroft * Find the entry for which the upcall was made and update
785 1.15 mycroft */
786 1.15 mycroft nstl = 0;
787 1.29 mycroft hash = MFCHASH(mfccp->mfcc_origin, mfccp->mfcc_mcastgrp);
788 1.15 mycroft for (rt = mfchashtbl[hash].lh_first; rt; rt = rt->mfc_hash.le_next) {
789 1.29 mycroft if (in_hosteq(rt->mfc_origin, mfccp->mfcc_origin) &&
790 1.29 mycroft in_hosteq(rt->mfc_mcastgrp, mfccp->mfcc_mcastgrp) &&
791 1.30 mycroft rt->mfc_stall != 0) {
792 1.15 mycroft if (nstl++)
793 1.30 mycroft log(LOG_ERR, "add_mfc %s o %x g %x p %x dbx %p\n",
794 1.15 mycroft "multiple kernel entries",
795 1.15 mycroft ntohl(mfccp->mfcc_origin.s_addr),
796 1.15 mycroft ntohl(mfccp->mfcc_mcastgrp.s_addr),
797 1.15 mycroft mfccp->mfcc_parent, rt->mfc_stall);
798 1.15 mycroft
799 1.15 mycroft if (mrtdebug & DEBUG_MFC)
800 1.30 mycroft log(LOG_DEBUG,"add_mfc o %x g %x p %x dbg %p\n",
801 1.15 mycroft ntohl(mfccp->mfcc_origin.s_addr),
802 1.15 mycroft ntohl(mfccp->mfcc_mcastgrp.s_addr),
803 1.15 mycroft mfccp->mfcc_parent, rt->mfc_stall);
804 1.15 mycroft
805 1.15 mycroft if (rt->mfc_expire)
806 1.15 mycroft nexpire[hash]--;
807 1.15 mycroft
808 1.35 mycroft rte = rt->mfc_stall;
809 1.35 mycroft update_mfc(mfccp, rt);
810 1.35 mycroft
811 1.15 mycroft /* free packets Qed at the end of this entry */
812 1.35 mycroft for (; rte != 0; rte = nrte) {
813 1.15 mycroft nrte = rte->next;
814 1.15 mycroft #ifdef RSVP_ISI
815 1.15 mycroft ip_mdq(rte->m, rte->ifp, rt, -1);
816 1.15 mycroft #else
817 1.15 mycroft ip_mdq(rte->m, rte->ifp, rt);
818 1.15 mycroft #endif /* RSVP_ISI */
819 1.15 mycroft m_freem(rte->m);
820 1.15 mycroft #ifdef UPCALL_TIMING
821 1.15 mycroft collate(&rte->t);
822 1.15 mycroft #endif /* UPCALL_TIMING */
823 1.15 mycroft free(rte, M_MRTABLE);
824 1.15 mycroft }
825 1.15 mycroft }
826 1.15 mycroft }
827 1.1 hpeyerl
828 1.15 mycroft if (nstl == 0) {
829 1.15 mycroft /*
830 1.15 mycroft * No mfc; make a new one
831 1.15 mycroft */
832 1.15 mycroft if (mrtdebug & DEBUG_MFC)
833 1.30 mycroft log(LOG_DEBUG,"add_mfc no upcall o %x g %x p %x\n",
834 1.15 mycroft ntohl(mfccp->mfcc_origin.s_addr),
835 1.15 mycroft ntohl(mfccp->mfcc_mcastgrp.s_addr),
836 1.15 mycroft mfccp->mfcc_parent);
837 1.15 mycroft
838 1.15 mycroft rt = (struct mfc *)malloc(sizeof(*rt), M_MRTABLE, M_NOWAIT);
839 1.30 mycroft if (rt == 0) {
840 1.1 hpeyerl splx(s);
841 1.15 mycroft return (ENOBUFS);
842 1.1 hpeyerl }
843 1.15 mycroft
844 1.15 mycroft rt->mfc_origin = mfccp->mfcc_origin;
845 1.15 mycroft rt->mfc_mcastgrp = mfccp->mfcc_mcastgrp;
846 1.15 mycroft /* initialize pkt counters per src-grp */
847 1.15 mycroft rt->mfc_pkt_cnt = 0;
848 1.15 mycroft rt->mfc_byte_cnt = 0;
849 1.15 mycroft rt->mfc_wrong_if = 0;
850 1.15 mycroft timerclear(&rt->mfc_last_assert);
851 1.15 mycroft update_mfc(mfccp, rt);
852 1.15 mycroft
853 1.15 mycroft /* insert new entry at head of hash chain */
854 1.15 mycroft LIST_INSERT_HEAD(&mfchashtbl[hash], rt, mfc_hash);
855 1.15 mycroft }
856 1.15 mycroft
857 1.1 hpeyerl splx(s);
858 1.1 hpeyerl return (0);
859 1.1 hpeyerl }
860 1.1 hpeyerl
861 1.15 mycroft #ifdef UPCALL_TIMING
862 1.15 mycroft /*
863 1.15 mycroft * collect delay statistics on the upcalls
864 1.15 mycroft */
865 1.15 mycroft static void collate(t)
866 1.48 augustss struct timeval *t;
867 1.15 mycroft {
868 1.48 augustss u_int32_t d;
869 1.48 augustss struct timeval tp;
870 1.48 augustss u_int32_t delta;
871 1.15 mycroft
872 1.15 mycroft microtime(&tp);
873 1.15 mycroft
874 1.15 mycroft if (timercmp(t, &tp, <)) {
875 1.15 mycroft TV_DELTA(tp, *t, delta);
876 1.15 mycroft
877 1.15 mycroft d = delta >> 10;
878 1.15 mycroft if (d > 50)
879 1.15 mycroft d = 50;
880 1.15 mycroft
881 1.15 mycroft ++upcall_data[d];
882 1.15 mycroft }
883 1.15 mycroft }
884 1.15 mycroft #endif /* UPCALL_TIMING */
885 1.15 mycroft
886 1.1 hpeyerl /*
887 1.15 mycroft * Delete an mfc entry
888 1.1 hpeyerl */
889 1.1 hpeyerl static int
890 1.15 mycroft del_mfc(m)
891 1.15 mycroft struct mbuf *m;
892 1.1 hpeyerl {
893 1.15 mycroft struct mfcctl *mfccp;
894 1.15 mycroft struct mfc *rt;
895 1.1 hpeyerl int s;
896 1.1 hpeyerl
897 1.15 mycroft if (m == 0 || m->m_len < sizeof(struct mfcctl))
898 1.15 mycroft return (EINVAL);
899 1.15 mycroft
900 1.15 mycroft mfccp = mtod(m, struct mfcctl *);
901 1.15 mycroft
902 1.15 mycroft if (mrtdebug & DEBUG_MFC)
903 1.30 mycroft log(LOG_DEBUG, "del_mfc origin %x mcastgrp %x\n",
904 1.29 mycroft ntohl(mfccp->mfcc_origin.s_addr),
905 1.29 mycroft ntohl(mfccp->mfcc_mcastgrp.s_addr));
906 1.1 hpeyerl
907 1.24 mycroft s = splsoftnet();
908 1.1 hpeyerl
909 1.29 mycroft MFCFIND(mfccp->mfcc_origin, mfccp->mfcc_mcastgrp, rt);
910 1.30 mycroft if (rt == 0) {
911 1.1 hpeyerl splx(s);
912 1.15 mycroft return (EADDRNOTAVAIL);
913 1.1 hpeyerl }
914 1.1 hpeyerl
915 1.15 mycroft LIST_REMOVE(rt, mfc_hash);
916 1.15 mycroft free(rt, M_MRTABLE);
917 1.1 hpeyerl
918 1.1 hpeyerl splx(s);
919 1.1 hpeyerl return (0);
920 1.1 hpeyerl }
921 1.1 hpeyerl
922 1.1 hpeyerl static int
923 1.15 mycroft socket_send(s, mm, src)
924 1.15 mycroft struct socket *s;
925 1.15 mycroft struct mbuf *mm;
926 1.15 mycroft struct sockaddr_in *src;
927 1.1 hpeyerl {
928 1.15 mycroft if (s) {
929 1.18 mycroft if (sbappendaddr(&s->so_rcv, sintosa(src), mm, (struct mbuf *)0) != 0) {
930 1.15 mycroft sorwakeup(s);
931 1.15 mycroft return (0);
932 1.15 mycroft }
933 1.15 mycroft }
934 1.15 mycroft m_freem(mm);
935 1.15 mycroft return (-1);
936 1.1 hpeyerl }
937 1.1 hpeyerl
938 1.1 hpeyerl /*
939 1.1 hpeyerl * IP multicast forwarding function. This function assumes that the packet
940 1.1 hpeyerl * pointed to by "ip" has arrived on (or is about to be sent to) the interface
941 1.1 hpeyerl * pointed to by "ifp", and the packet is to be relayed to other networks
942 1.1 hpeyerl * that have members of the packet's destination IP multicast group.
943 1.1 hpeyerl *
944 1.15 mycroft * The packet is returned unscathed to the caller, unless it is
945 1.15 mycroft * erroneous, in which case a non-zero return value tells the caller to
946 1.1 hpeyerl * discard it.
947 1.1 hpeyerl */
948 1.1 hpeyerl
949 1.15 mycroft #define IP_HDR_LEN 20 /* # bytes of fixed IP header (excluding options) */
950 1.15 mycroft #define TUNNEL_LEN 12 /* # bytes of IP option for tunnel encapsulation */
951 1.1 hpeyerl
952 1.1 hpeyerl int
953 1.15 mycroft #ifdef RSVP_ISI
954 1.15 mycroft ip_mforward(m, ifp, imo)
955 1.15 mycroft #else
956 1.10 mycroft ip_mforward(m, ifp)
957 1.15 mycroft #endif /* RSVP_ISI */
958 1.15 mycroft struct mbuf *m;
959 1.15 mycroft struct ifnet *ifp;
960 1.15 mycroft #ifdef RSVP_ISI
961 1.15 mycroft struct ip_moptions *imo;
962 1.15 mycroft #endif /* RSVP_ISI */
963 1.1 hpeyerl {
964 1.48 augustss struct ip *ip = mtod(m, struct ip *);
965 1.48 augustss struct mfc *rt;
966 1.48 augustss u_char *ipoptions;
967 1.15 mycroft static int srctun = 0;
968 1.48 augustss struct mbuf *mm;
969 1.15 mycroft int s;
970 1.15 mycroft #ifdef RSVP_ISI
971 1.48 augustss struct vif *vifp;
972 1.15 mycroft vifi_t vifi;
973 1.15 mycroft #endif /* RSVP_ISI */
974 1.15 mycroft
975 1.15 mycroft if (mrtdebug & DEBUG_FORWARD)
976 1.30 mycroft log(LOG_DEBUG, "ip_mforward: src %x, dst %x, ifp %p\n",
977 1.15 mycroft ntohl(ip->ip_src.s_addr), ntohl(ip->ip_dst.s_addr), ifp);
978 1.1 hpeyerl
979 1.15 mycroft if (ip->ip_hl < (IP_HDR_LEN + TUNNEL_LEN) >> 2 ||
980 1.15 mycroft (ipoptions = (u_char *)(ip + 1))[1] != IPOPT_LSRR) {
981 1.1 hpeyerl /*
982 1.15 mycroft * Packet arrived via a physical interface or
983 1.15 mycroft * an encapuslated tunnel.
984 1.1 hpeyerl */
985 1.15 mycroft } else {
986 1.1 hpeyerl /*
987 1.15 mycroft * Packet arrived through a source-route tunnel.
988 1.15 mycroft * Source-route tunnels are no longer supported.
989 1.1 hpeyerl */
990 1.15 mycroft if ((srctun++ % 1000) == 0)
991 1.30 mycroft log(LOG_ERR, "ip_mforward: received source-routed packet from %x\n",
992 1.15 mycroft ntohl(ip->ip_src.s_addr));
993 1.15 mycroft
994 1.15 mycroft return (1);
995 1.15 mycroft }
996 1.15 mycroft
997 1.15 mycroft #ifdef RSVP_ISI
998 1.15 mycroft if (imo && ((vifi = imo->imo_multicast_vif) < numvifs)) {
999 1.15 mycroft if (ip->ip_ttl < 255)
1000 1.15 mycroft ip->ip_ttl++; /* compensate for -1 in *_send routines */
1001 1.15 mycroft if (rsvpdebug && ip->ip_p == IPPROTO_RSVP) {
1002 1.15 mycroft vifp = viftable + vifi;
1003 1.34 christos printf("Sending IPPROTO_RSVP from %x to %x on vif %d (%s%s)\n",
1004 1.15 mycroft ntohl(ip->ip_src), ntohl(ip->ip_dst), vifi,
1005 1.15 mycroft (vifp->v_flags & VIFF_TUNNEL) ? "tunnel on " : "",
1006 1.27 thorpej vifp->v_ifp->if_xname);
1007 1.15 mycroft }
1008 1.31 mycroft return (ip_mdq(m, ifp, (struct mfc *)0, vifi));
1009 1.15 mycroft }
1010 1.15 mycroft if (rsvpdebug && ip->ip_p == IPPROTO_RSVP) {
1011 1.34 christos printf("Warning: IPPROTO_RSVP from %x to %x without vif option\n",
1012 1.15 mycroft ntohl(ip->ip_src), ntohl(ip->ip_dst));
1013 1.15 mycroft }
1014 1.15 mycroft #endif /* RSVP_ISI */
1015 1.15 mycroft
1016 1.15 mycroft /*
1017 1.15 mycroft * Don't forward a packet with time-to-live of zero or one,
1018 1.15 mycroft * or a packet destined to a local-only group.
1019 1.15 mycroft */
1020 1.15 mycroft if (ip->ip_ttl <= 1 ||
1021 1.16 mycroft IN_LOCAL_GROUP(ip->ip_dst.s_addr))
1022 1.15 mycroft return (0);
1023 1.15 mycroft
1024 1.15 mycroft /*
1025 1.15 mycroft * Determine forwarding vifs from the forwarding cache table
1026 1.15 mycroft */
1027 1.24 mycroft s = splsoftnet();
1028 1.29 mycroft MFCFIND(ip->ip_src, ip->ip_dst, rt);
1029 1.1 hpeyerl
1030 1.15 mycroft /* Entry exists, so forward if necessary */
1031 1.30 mycroft if (rt != 0) {
1032 1.15 mycroft splx(s);
1033 1.15 mycroft #ifdef RSVP_ISI
1034 1.15 mycroft return (ip_mdq(m, ifp, rt, -1));
1035 1.15 mycroft #else
1036 1.15 mycroft return (ip_mdq(m, ifp, rt));
1037 1.15 mycroft #endif /* RSVP_ISI */
1038 1.15 mycroft } else {
1039 1.1 hpeyerl /*
1040 1.15 mycroft * If we don't have a route for packet's origin,
1041 1.15 mycroft * Make a copy of the packet &
1042 1.15 mycroft * send message to routing daemon
1043 1.1 hpeyerl */
1044 1.15 mycroft
1045 1.48 augustss struct mbuf *mb0;
1046 1.48 augustss struct rtdetq *rte;
1047 1.48 augustss u_int32_t hash;
1048 1.31 mycroft int hlen = ip->ip_hl << 2;
1049 1.15 mycroft #ifdef UPCALL_TIMING
1050 1.15 mycroft struct timeval tp;
1051 1.15 mycroft
1052 1.15 mycroft microtime(&tp);
1053 1.15 mycroft #endif /* UPCALL_TIMING */
1054 1.15 mycroft
1055 1.15 mycroft mrtstat.mrts_no_route++;
1056 1.15 mycroft if (mrtdebug & (DEBUG_FORWARD | DEBUG_MFC))
1057 1.30 mycroft log(LOG_DEBUG, "ip_mforward: no rte s %x g %x\n",
1058 1.15 mycroft ntohl(ip->ip_src.s_addr),
1059 1.15 mycroft ntohl(ip->ip_dst.s_addr));
1060 1.1 hpeyerl
1061 1.1 hpeyerl /*
1062 1.15 mycroft * Allocate mbufs early so that we don't do extra work if we are
1063 1.31 mycroft * just going to fail anyway. Make sure to pullup the header so
1064 1.31 mycroft * that other people can't step on it.
1065 1.1 hpeyerl */
1066 1.15 mycroft rte = (struct rtdetq *)malloc(sizeof(*rte), M_MRTABLE, M_NOWAIT);
1067 1.30 mycroft if (rte == 0) {
1068 1.15 mycroft splx(s);
1069 1.15 mycroft return (ENOBUFS);
1070 1.15 mycroft }
1071 1.15 mycroft mb0 = m_copy(m, 0, M_COPYALL);
1072 1.31 mycroft M_PULLUP(mb0, hlen);
1073 1.30 mycroft if (mb0 == 0) {
1074 1.15 mycroft free(rte, M_MRTABLE);
1075 1.15 mycroft splx(s);
1076 1.15 mycroft return (ENOBUFS);
1077 1.15 mycroft }
1078 1.15 mycroft
1079 1.15 mycroft /* is there an upcall waiting for this packet? */
1080 1.29 mycroft hash = MFCHASH(ip->ip_src, ip->ip_dst);
1081 1.15 mycroft for (rt = mfchashtbl[hash].lh_first; rt; rt = rt->mfc_hash.le_next) {
1082 1.29 mycroft if (in_hosteq(ip->ip_src, rt->mfc_origin) &&
1083 1.29 mycroft in_hosteq(ip->ip_dst, rt->mfc_mcastgrp) &&
1084 1.30 mycroft rt->mfc_stall != 0)
1085 1.15 mycroft break;
1086 1.1 hpeyerl }
1087 1.15 mycroft
1088 1.30 mycroft if (rt == 0) {
1089 1.15 mycroft int i;
1090 1.15 mycroft struct igmpmsg *im;
1091 1.15 mycroft
1092 1.15 mycroft /* no upcall, so make a new entry */
1093 1.15 mycroft rt = (struct mfc *)malloc(sizeof(*rt), M_MRTABLE, M_NOWAIT);
1094 1.30 mycroft if (rt == 0) {
1095 1.15 mycroft free(rte, M_MRTABLE);
1096 1.31 mycroft m_freem(mb0);
1097 1.15 mycroft splx(s);
1098 1.15 mycroft return (ENOBUFS);
1099 1.15 mycroft }
1100 1.15 mycroft /* Make a copy of the header to send to the user level process */
1101 1.15 mycroft mm = m_copy(m, 0, hlen);
1102 1.15 mycroft M_PULLUP(mm, hlen);
1103 1.30 mycroft if (mm == 0) {
1104 1.15 mycroft free(rte, M_MRTABLE);
1105 1.31 mycroft m_freem(mb0);
1106 1.15 mycroft free(rt, M_MRTABLE);
1107 1.15 mycroft splx(s);
1108 1.15 mycroft return (ENOBUFS);
1109 1.15 mycroft }
1110 1.15 mycroft
1111 1.15 mycroft /*
1112 1.15 mycroft * Send message to routing daemon to install
1113 1.15 mycroft * a route into the kernel table
1114 1.15 mycroft */
1115 1.15 mycroft sin.sin_addr = ip->ip_src;
1116 1.15 mycroft
1117 1.15 mycroft im = mtod(mm, struct igmpmsg *);
1118 1.15 mycroft im->im_msgtype = IGMPMSG_NOCACHE;
1119 1.15 mycroft im->im_mbz = 0;
1120 1.15 mycroft
1121 1.15 mycroft mrtstat.mrts_upcalls++;
1122 1.15 mycroft
1123 1.15 mycroft if (socket_send(ip_mrouter, mm, &sin) < 0) {
1124 1.30 mycroft log(LOG_WARNING, "ip_mforward: ip_mrouter socket queue full\n");
1125 1.15 mycroft ++mrtstat.mrts_upq_sockfull;
1126 1.15 mycroft free(rte, M_MRTABLE);
1127 1.31 mycroft m_freem(mb0);
1128 1.15 mycroft free(rt, M_MRTABLE);
1129 1.15 mycroft splx(s);
1130 1.15 mycroft return (ENOBUFS);
1131 1.15 mycroft }
1132 1.15 mycroft
1133 1.15 mycroft /* insert new entry at head of hash chain */
1134 1.23 mycroft rt->mfc_origin = ip->ip_src;
1135 1.23 mycroft rt->mfc_mcastgrp = ip->ip_dst;
1136 1.23 mycroft rt->mfc_pkt_cnt = 0;
1137 1.23 mycroft rt->mfc_byte_cnt = 0;
1138 1.23 mycroft rt->mfc_wrong_if = 0;
1139 1.23 mycroft rt->mfc_expire = UPCALL_EXPIRE;
1140 1.15 mycroft nexpire[hash]++;
1141 1.15 mycroft for (i = 0; i < numvifs; i++)
1142 1.15 mycroft rt->mfc_ttls[i] = 0;
1143 1.15 mycroft rt->mfc_parent = -1;
1144 1.15 mycroft
1145 1.15 mycroft /* link into table */
1146 1.15 mycroft LIST_INSERT_HEAD(&mfchashtbl[hash], rt, mfc_hash);
1147 1.15 mycroft /* Add this entry to the end of the queue */
1148 1.15 mycroft rt->mfc_stall = rte;
1149 1.15 mycroft } else {
1150 1.15 mycroft /* determine if q has overflowed */
1151 1.15 mycroft struct rtdetq **p;
1152 1.48 augustss int npkts = 0;
1153 1.15 mycroft
1154 1.30 mycroft for (p = &rt->mfc_stall; *p != 0; p = &(*p)->next)
1155 1.15 mycroft if (++npkts > MAX_UPQ) {
1156 1.15 mycroft mrtstat.mrts_upq_ovflw++;
1157 1.15 mycroft free(rte, M_MRTABLE);
1158 1.31 mycroft m_freem(mb0);
1159 1.15 mycroft splx(s);
1160 1.15 mycroft return (0);
1161 1.15 mycroft }
1162 1.15 mycroft
1163 1.15 mycroft /* Add this entry to the end of the queue */
1164 1.15 mycroft *p = rte;
1165 1.15 mycroft }
1166 1.15 mycroft
1167 1.30 mycroft rte->next = 0;
1168 1.15 mycroft rte->m = mb0;
1169 1.15 mycroft rte->ifp = ifp;
1170 1.15 mycroft #ifdef UPCALL_TIMING
1171 1.15 mycroft rte->t = tp;
1172 1.15 mycroft #endif /* UPCALL_TIMING */
1173 1.15 mycroft
1174 1.15 mycroft
1175 1.15 mycroft splx(s);
1176 1.15 mycroft
1177 1.15 mycroft return (0);
1178 1.15 mycroft }
1179 1.1 hpeyerl }
1180 1.1 hpeyerl
1181 1.15 mycroft
1182 1.25 christos /*ARGSUSED*/
1183 1.1 hpeyerl static void
1184 1.25 christos expire_upcalls(v)
1185 1.25 christos void *v;
1186 1.1 hpeyerl {
1187 1.15 mycroft int i;
1188 1.15 mycroft int s;
1189 1.15 mycroft
1190 1.24 mycroft s = splsoftnet();
1191 1.15 mycroft
1192 1.15 mycroft for (i = 0; i < MFCTBLSIZ; i++) {
1193 1.48 augustss struct mfc *rt, *nrt;
1194 1.15 mycroft
1195 1.15 mycroft if (nexpire[i] == 0)
1196 1.15 mycroft continue;
1197 1.15 mycroft
1198 1.15 mycroft for (rt = mfchashtbl[i].lh_first; rt; rt = nrt) {
1199 1.15 mycroft nrt = rt->mfc_hash.le_next;
1200 1.1 hpeyerl
1201 1.15 mycroft if (rt->mfc_expire == 0 ||
1202 1.15 mycroft --rt->mfc_expire > 0)
1203 1.15 mycroft continue;
1204 1.15 mycroft nexpire[i]--;
1205 1.15 mycroft
1206 1.15 mycroft ++mrtstat.mrts_cache_cleanups;
1207 1.15 mycroft if (mrtdebug & DEBUG_EXPIRE)
1208 1.15 mycroft log(LOG_DEBUG,
1209 1.30 mycroft "expire_upcalls: expiring (%x %x)\n",
1210 1.15 mycroft ntohl(rt->mfc_origin.s_addr),
1211 1.15 mycroft ntohl(rt->mfc_mcastgrp.s_addr));
1212 1.1 hpeyerl
1213 1.15 mycroft expire_mfc(rt);
1214 1.15 mycroft }
1215 1.15 mycroft }
1216 1.1 hpeyerl
1217 1.15 mycroft splx(s);
1218 1.47 thorpej callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
1219 1.47 thorpej expire_upcalls, NULL);
1220 1.1 hpeyerl }
1221 1.1 hpeyerl
1222 1.15 mycroft /*
1223 1.15 mycroft * Packet forwarding routine once entry in the cache is made
1224 1.15 mycroft */
1225 1.15 mycroft static int
1226 1.15 mycroft #ifdef RSVP_ISI
1227 1.15 mycroft ip_mdq(m, ifp, rt, xmt_vif)
1228 1.15 mycroft #else
1229 1.15 mycroft ip_mdq(m, ifp, rt)
1230 1.15 mycroft #endif /* RSVP_ISI */
1231 1.48 augustss struct mbuf *m;
1232 1.48 augustss struct ifnet *ifp;
1233 1.48 augustss struct mfc *rt;
1234 1.15 mycroft #ifdef RSVP_ISI
1235 1.48 augustss vifi_t xmt_vif;
1236 1.15 mycroft #endif /* RSVP_ISI */
1237 1.1 hpeyerl {
1238 1.48 augustss struct ip *ip = mtod(m, struct ip *);
1239 1.48 augustss vifi_t vifi;
1240 1.48 augustss struct vif *vifp;
1241 1.48 augustss int plen = ntohs(ip->ip_len);
1242 1.15 mycroft
1243 1.15 mycroft /*
1244 1.15 mycroft * Macro to send packet on vif. Since RSVP packets don't get counted on
1245 1.15 mycroft * input, they shouldn't get counted on output, so statistics keeping is
1246 1.15 mycroft * seperate.
1247 1.15 mycroft */
1248 1.15 mycroft #define MC_SEND(ip,vifp,m) { \
1249 1.15 mycroft if ((vifp)->v_flags & VIFF_TUNNEL) \
1250 1.15 mycroft encap_send((ip), (vifp), (m)); \
1251 1.15 mycroft else \
1252 1.15 mycroft phyint_send((ip), (vifp), (m)); \
1253 1.15 mycroft }
1254 1.1 hpeyerl
1255 1.15 mycroft #ifdef RSVP_ISI
1256 1.15 mycroft /*
1257 1.15 mycroft * If xmt_vif is not -1, send on only the requested vif.
1258 1.15 mycroft *
1259 1.15 mycroft * (since vifi_t is u_short, -1 becomes MAXUSHORT, which > numvifs.
1260 1.15 mycroft */
1261 1.15 mycroft if (xmt_vif < numvifs) {
1262 1.15 mycroft MC_SEND(ip, viftable + xmt_vif, m);
1263 1.15 mycroft return (1);
1264 1.15 mycroft }
1265 1.15 mycroft #endif /* RSVP_ISI */
1266 1.15 mycroft
1267 1.15 mycroft /*
1268 1.15 mycroft * Don't forward if it didn't arrive from the parent vif for its origin.
1269 1.15 mycroft */
1270 1.15 mycroft vifi = rt->mfc_parent;
1271 1.15 mycroft if ((vifi >= numvifs) || (viftable[vifi].v_ifp != ifp)) {
1272 1.15 mycroft /* came in the wrong interface */
1273 1.15 mycroft if (mrtdebug & DEBUG_FORWARD)
1274 1.30 mycroft log(LOG_DEBUG, "wrong if: ifp %p vifi %d vififp %p\n",
1275 1.15 mycroft ifp, vifi, viftable[vifi].v_ifp);
1276 1.15 mycroft ++mrtstat.mrts_wrong_if;
1277 1.15 mycroft ++rt->mfc_wrong_if;
1278 1.1 hpeyerl /*
1279 1.15 mycroft * If we are doing PIM assert processing, and we are forwarding
1280 1.15 mycroft * packets on this interface, and it is a broadcast medium
1281 1.15 mycroft * interface (and not a tunnel), send a message to the routing daemon.
1282 1.1 hpeyerl */
1283 1.15 mycroft if (pim_assert && rt->mfc_ttls[vifi] &&
1284 1.15 mycroft (ifp->if_flags & IFF_BROADCAST) &&
1285 1.15 mycroft !(viftable[vifi].v_flags & VIFF_TUNNEL)) {
1286 1.15 mycroft struct mbuf *mm;
1287 1.15 mycroft struct igmpmsg *im;
1288 1.15 mycroft int hlen = ip->ip_hl << 2;
1289 1.15 mycroft struct timeval now;
1290 1.48 augustss u_int32_t delta;
1291 1.15 mycroft
1292 1.15 mycroft microtime(&now);
1293 1.15 mycroft
1294 1.15 mycroft TV_DELTA(rt->mfc_last_assert, now, delta);
1295 1.15 mycroft
1296 1.15 mycroft if (delta > ASSERT_MSG_TIME) {
1297 1.15 mycroft mm = m_copy(m, 0, hlen);
1298 1.15 mycroft M_PULLUP(mm, hlen);
1299 1.30 mycroft if (mm == 0) {
1300 1.15 mycroft return (ENOBUFS);
1301 1.15 mycroft }
1302 1.15 mycroft
1303 1.15 mycroft rt->mfc_last_assert = now;
1304 1.15 mycroft
1305 1.15 mycroft im = mtod(mm, struct igmpmsg *);
1306 1.15 mycroft im->im_msgtype = IGMPMSG_WRONGVIF;
1307 1.15 mycroft im->im_mbz = 0;
1308 1.15 mycroft im->im_vif = vifi;
1309 1.15 mycroft
1310 1.15 mycroft sin.sin_addr = im->im_src;
1311 1.15 mycroft
1312 1.31 mycroft socket_send(ip_mrouter, mm, &sin);
1313 1.15 mycroft }
1314 1.15 mycroft }
1315 1.15 mycroft return (0);
1316 1.15 mycroft }
1317 1.15 mycroft
1318 1.15 mycroft /* If I sourced this packet, it counts as output, else it was input. */
1319 1.29 mycroft if (in_hosteq(ip->ip_src, viftable[vifi].v_lcl_addr)) {
1320 1.15 mycroft viftable[vifi].v_pkt_out++;
1321 1.15 mycroft viftable[vifi].v_bytes_out += plen;
1322 1.15 mycroft } else {
1323 1.15 mycroft viftable[vifi].v_pkt_in++;
1324 1.15 mycroft viftable[vifi].v_bytes_in += plen;
1325 1.15 mycroft }
1326 1.15 mycroft rt->mfc_pkt_cnt++;
1327 1.15 mycroft rt->mfc_byte_cnt += plen;
1328 1.15 mycroft
1329 1.15 mycroft /*
1330 1.15 mycroft * For each vif, decide if a copy of the packet should be forwarded.
1331 1.15 mycroft * Forward if:
1332 1.15 mycroft * - the ttl exceeds the vif's threshold
1333 1.15 mycroft * - there are group members downstream on interface
1334 1.15 mycroft */
1335 1.15 mycroft for (vifp = viftable, vifi = 0; vifi < numvifs; vifp++, vifi++)
1336 1.15 mycroft if ((rt->mfc_ttls[vifi] > 0) &&
1337 1.15 mycroft (ip->ip_ttl > rt->mfc_ttls[vifi])) {
1338 1.15 mycroft vifp->v_pkt_out++;
1339 1.15 mycroft vifp->v_bytes_out += plen;
1340 1.15 mycroft MC_SEND(ip, vifp, m);
1341 1.1 hpeyerl }
1342 1.1 hpeyerl
1343 1.15 mycroft return (0);
1344 1.15 mycroft }
1345 1.15 mycroft
1346 1.15 mycroft #ifdef RSVP_ISI
1347 1.15 mycroft /*
1348 1.15 mycroft * check if a vif number is legal/ok. This is used by ip_output, to export
1349 1.15 mycroft * numvifs there,
1350 1.15 mycroft */
1351 1.15 mycroft int
1352 1.15 mycroft legal_vif_num(vif)
1353 1.15 mycroft int vif;
1354 1.15 mycroft {
1355 1.15 mycroft if (vif >= 0 && vif < numvifs)
1356 1.15 mycroft return (1);
1357 1.15 mycroft else
1358 1.15 mycroft return (0);
1359 1.15 mycroft }
1360 1.15 mycroft #endif /* RSVP_ISI */
1361 1.15 mycroft
1362 1.15 mycroft static void
1363 1.15 mycroft phyint_send(ip, vifp, m)
1364 1.15 mycroft struct ip *ip;
1365 1.15 mycroft struct vif *vifp;
1366 1.15 mycroft struct mbuf *m;
1367 1.15 mycroft {
1368 1.48 augustss struct mbuf *mb_copy;
1369 1.48 augustss int hlen = ip->ip_hl << 2;
1370 1.15 mycroft
1371 1.15 mycroft /*
1372 1.15 mycroft * Make a new reference to the packet; make sure that
1373 1.15 mycroft * the IP header is actually copied, not just referenced,
1374 1.15 mycroft * so that ip_output() only scribbles on the copy.
1375 1.15 mycroft */
1376 1.12 brezak mb_copy = m_copy(m, 0, M_COPYALL);
1377 1.15 mycroft M_PULLUP(mb_copy, hlen);
1378 1.30 mycroft if (mb_copy == 0)
1379 1.1 hpeyerl return;
1380 1.1 hpeyerl
1381 1.15 mycroft if (vifp->v_rate_limit <= 0)
1382 1.15 mycroft tbf_send_packet(vifp, mb_copy);
1383 1.15 mycroft else
1384 1.15 mycroft tbf_control(vifp, mb_copy, mtod(mb_copy, struct ip *), ip->ip_len);
1385 1.12 brezak }
1386 1.12 brezak
1387 1.12 brezak static void
1388 1.12 brezak encap_send(ip, vifp, m)
1389 1.48 augustss struct ip *ip;
1390 1.48 augustss struct vif *vifp;
1391 1.48 augustss struct mbuf *m;
1392 1.48 augustss {
1393 1.48 augustss struct mbuf *mb_copy;
1394 1.48 augustss struct ip *ip_copy;
1395 1.48 augustss int i, len = ip->ip_len + sizeof(multicast_encap_iphdr);
1396 1.12 brezak
1397 1.12 brezak /*
1398 1.12 brezak * copy the old packet & pullup it's IP header into the
1399 1.12 brezak * new mbuf so we can modify it. Try to fill the new
1400 1.12 brezak * mbuf since if we don't the ethernet driver will.
1401 1.12 brezak */
1402 1.15 mycroft MGETHDR(mb_copy, M_DONTWAIT, MT_DATA);
1403 1.30 mycroft if (mb_copy == 0)
1404 1.12 brezak return;
1405 1.15 mycroft mb_copy->m_data += max_linkhdr;
1406 1.15 mycroft mb_copy->m_pkthdr.len = len;
1407 1.12 brezak mb_copy->m_len = sizeof(multicast_encap_iphdr);
1408 1.15 mycroft
1409 1.30 mycroft if ((mb_copy->m_next = m_copy(m, 0, M_COPYALL)) == 0) {
1410 1.12 brezak m_freem(mb_copy);
1411 1.12 brezak return;
1412 1.12 brezak }
1413 1.15 mycroft i = MHLEN - max_linkhdr;
1414 1.12 brezak if (i > len)
1415 1.12 brezak i = len;
1416 1.12 brezak mb_copy = m_pullup(mb_copy, i);
1417 1.30 mycroft if (mb_copy == 0)
1418 1.12 brezak return;
1419 1.15 mycroft
1420 1.12 brezak /*
1421 1.12 brezak * fill in the encapsulating IP header.
1422 1.12 brezak */
1423 1.12 brezak ip_copy = mtod(mb_copy, struct ip *);
1424 1.12 brezak *ip_copy = multicast_encap_iphdr;
1425 1.12 brezak ip_copy->ip_id = htons(ip_id++);
1426 1.15 mycroft ip_copy->ip_len = len;
1427 1.12 brezak ip_copy->ip_src = vifp->v_lcl_addr;
1428 1.12 brezak ip_copy->ip_dst = vifp->v_rmt_addr;
1429 1.15 mycroft
1430 1.12 brezak /*
1431 1.12 brezak * turn the encapsulated IP header back into a valid one.
1432 1.12 brezak */
1433 1.12 brezak ip = (struct ip *)((caddr_t)ip_copy + sizeof(multicast_encap_iphdr));
1434 1.12 brezak --ip->ip_ttl;
1435 1.12 brezak HTONS(ip->ip_len);
1436 1.12 brezak HTONS(ip->ip_off);
1437 1.12 brezak ip->ip_sum = 0;
1438 1.12 brezak mb_copy->m_data += sizeof(multicast_encap_iphdr);
1439 1.12 brezak ip->ip_sum = in_cksum(mb_copy, ip->ip_hl << 2);
1440 1.12 brezak mb_copy->m_data -= sizeof(multicast_encap_iphdr);
1441 1.15 mycroft
1442 1.15 mycroft if (vifp->v_rate_limit <= 0)
1443 1.15 mycroft tbf_send_packet(vifp, mb_copy);
1444 1.15 mycroft else
1445 1.15 mycroft tbf_control(vifp, mb_copy, ip, ip_copy->ip_len);
1446 1.12 brezak }
1447 1.12 brezak
1448 1.12 brezak /*
1449 1.12 brezak * De-encapsulate a packet and feed it back through ip input (this
1450 1.12 brezak * routine is called whenever IP gets a packet with proto type
1451 1.12 brezak * ENCAP_PROTO and a local destination address).
1452 1.39 thorpej *
1453 1.39 thorpej * Return 1 if we handled the packet, 0 if we did not.
1454 1.39 thorpej *
1455 1.50 itojun * Called from encap4_input() in sys/netinet/ip_encap.c.
1456 1.12 brezak */
1457 1.39 thorpej int
1458 1.39 thorpej mrt_ipip_input(m, hlen)
1459 1.25 christos struct mbuf *m;
1460 1.39 thorpej int hlen;
1461 1.25 christos {
1462 1.48 augustss struct ip *ip = mtod(m, struct ip *);
1463 1.48 augustss int s;
1464 1.48 augustss struct ifqueue *ifq;
1465 1.48 augustss struct vif *vifp;
1466 1.25 christos
1467 1.39 thorpej if (!have_encap_tunnel)
1468 1.39 thorpej return (0);
1469 1.22 mycroft
1470 1.12 brezak /*
1471 1.12 brezak * dump the packet if it's not to a multicast destination or if
1472 1.12 brezak * we don't have an encapsulating tunnel with the source.
1473 1.12 brezak * Note: This code assumes that the remote site IP address
1474 1.12 brezak * uniquely identifies the tunnel (i.e., that this site has
1475 1.12 brezak * at most one tunnel with the remote site).
1476 1.12 brezak */
1477 1.16 mycroft if (!IN_MULTICAST(((struct ip *)((char *)ip + hlen))->ip_dst.s_addr)) {
1478 1.12 brezak ++mrtstat.mrts_bad_tunnel;
1479 1.39 thorpej return (0);
1480 1.12 brezak }
1481 1.22 mycroft
1482 1.29 mycroft if (!in_hosteq(ip->ip_src, last_encap_src)) {
1483 1.48 augustss struct vif *vife;
1484 1.15 mycroft
1485 1.12 brezak vifp = viftable;
1486 1.12 brezak vife = vifp + numvifs;
1487 1.22 mycroft for (; vifp < vife; vifp++)
1488 1.22 mycroft if (vifp->v_flags & VIFF_TUNNEL &&
1489 1.29 mycroft in_hosteq(vifp->v_rmt_addr, ip->ip_src))
1490 1.22 mycroft break;
1491 1.22 mycroft if (vifp == vife) {
1492 1.22 mycroft mrtstat.mrts_cant_tunnel++; /*XXX*/
1493 1.22 mycroft if (mrtdebug)
1494 1.39 thorpej log(LOG_DEBUG,
1495 1.39 thorpej "ip_mforward: no tunnel with %x\n",
1496 1.22 mycroft ntohl(ip->ip_src.s_addr));
1497 1.39 thorpej return (0);
1498 1.22 mycroft }
1499 1.22 mycroft last_encap_vif = vifp;
1500 1.29 mycroft last_encap_src = ip->ip_src;
1501 1.22 mycroft } else
1502 1.22 mycroft vifp = last_encap_vif;
1503 1.22 mycroft
1504 1.12 brezak m->m_data += hlen;
1505 1.12 brezak m->m_len -= hlen;
1506 1.12 brezak m->m_pkthdr.len -= hlen;
1507 1.15 mycroft m->m_pkthdr.rcvif = vifp->v_ifp;
1508 1.12 brezak ifq = &ipintrq;
1509 1.12 brezak s = splimp();
1510 1.12 brezak if (IF_QFULL(ifq)) {
1511 1.12 brezak IF_DROP(ifq);
1512 1.12 brezak m_freem(m);
1513 1.12 brezak } else {
1514 1.12 brezak IF_ENQUEUE(ifq, m);
1515 1.12 brezak /*
1516 1.12 brezak * normally we would need a "schednetisr(NETISR_IP)"
1517 1.12 brezak * here but we were called by ip_input and it is going
1518 1.12 brezak * to loop back & try to dequeue the packet we just
1519 1.12 brezak * queued as soon as we return so we avoid the
1520 1.12 brezak * unnecessary software interrrupt.
1521 1.12 brezak */
1522 1.12 brezak }
1523 1.12 brezak splx(s);
1524 1.39 thorpej return (1);
1525 1.1 hpeyerl }
1526 1.15 mycroft
1527 1.15 mycroft /*
1528 1.15 mycroft * Token bucket filter module
1529 1.15 mycroft */
1530 1.15 mycroft static void
1531 1.31 mycroft tbf_control(vifp, m, ip, len)
1532 1.48 augustss struct vif *vifp;
1533 1.48 augustss struct mbuf *m;
1534 1.48 augustss struct ip *ip;
1535 1.48 augustss u_int32_t len;
1536 1.15 mycroft {
1537 1.15 mycroft
1538 1.31 mycroft if (len > MAX_BKT_SIZE) {
1539 1.31 mycroft /* drop if packet is too large */
1540 1.31 mycroft mrtstat.mrts_pkt2large++;
1541 1.31 mycroft m_freem(m);
1542 1.31 mycroft return;
1543 1.31 mycroft }
1544 1.31 mycroft
1545 1.21 mycroft tbf_update_tokens(vifp);
1546 1.15 mycroft
1547 1.21 mycroft /*
1548 1.21 mycroft * If there are enough tokens, and the queue is empty, send this packet
1549 1.21 mycroft * out immediately. Otherwise, try to insert it on this vif's queue.
1550 1.21 mycroft */
1551 1.31 mycroft if (vifp->tbf_q_len == 0) {
1552 1.31 mycroft if (len <= vifp->tbf_n_tok) {
1553 1.31 mycroft vifp->tbf_n_tok -= len;
1554 1.21 mycroft tbf_send_packet(vifp, m);
1555 1.21 mycroft } else {
1556 1.21 mycroft /* queue packet and timeout till later */
1557 1.31 mycroft tbf_queue(vifp, m);
1558 1.47 thorpej callout_reset(&vifp->v_repq_ch, TBF_REPROCESS,
1559 1.47 thorpej tbf_reprocess_q, vifp);
1560 1.21 mycroft }
1561 1.15 mycroft } else {
1562 1.31 mycroft if (vifp->tbf_q_len >= vifp->tbf_max_q_len &&
1563 1.21 mycroft !tbf_dq_sel(vifp, ip)) {
1564 1.21 mycroft /* queue length too much, and couldn't make room */
1565 1.21 mycroft mrtstat.mrts_q_overflow++;
1566 1.21 mycroft m_freem(m);
1567 1.21 mycroft } else {
1568 1.21 mycroft /* queue length low enough, or made room */
1569 1.31 mycroft tbf_queue(vifp, m);
1570 1.21 mycroft tbf_process_q(vifp);
1571 1.21 mycroft }
1572 1.15 mycroft }
1573 1.15 mycroft }
1574 1.15 mycroft
1575 1.15 mycroft /*
1576 1.15 mycroft * adds a packet to the queue at the interface
1577 1.15 mycroft */
1578 1.15 mycroft static void
1579 1.31 mycroft tbf_queue(vifp, m)
1580 1.48 augustss struct vif *vifp;
1581 1.48 augustss struct mbuf *m;
1582 1.15 mycroft {
1583 1.48 augustss int s = splsoftnet();
1584 1.15 mycroft
1585 1.31 mycroft /* insert at tail */
1586 1.31 mycroft *vifp->tbf_t = m;
1587 1.31 mycroft vifp->tbf_t = &m->m_nextpkt;
1588 1.31 mycroft vifp->tbf_q_len++;
1589 1.15 mycroft
1590 1.31 mycroft splx(s);
1591 1.15 mycroft }
1592 1.15 mycroft
1593 1.15 mycroft
1594 1.15 mycroft /*
1595 1.15 mycroft * processes the queue at the interface
1596 1.15 mycroft */
1597 1.15 mycroft static void
1598 1.15 mycroft tbf_process_q(vifp)
1599 1.48 augustss struct vif *vifp;
1600 1.15 mycroft {
1601 1.48 augustss struct mbuf *m;
1602 1.48 augustss int len;
1603 1.48 augustss int s = splsoftnet();
1604 1.15 mycroft
1605 1.31 mycroft /*
1606 1.31 mycroft * Loop through the queue at the interface and send as many packets
1607 1.31 mycroft * as possible.
1608 1.31 mycroft */
1609 1.31 mycroft for (m = vifp->tbf_q;
1610 1.31 mycroft m != 0;
1611 1.31 mycroft m = vifp->tbf_q) {
1612 1.31 mycroft len = mtod(m, struct ip *)->ip_len;
1613 1.31 mycroft
1614 1.31 mycroft /* determine if the packet can be sent */
1615 1.31 mycroft if (len <= vifp->tbf_n_tok) {
1616 1.31 mycroft /* if so,
1617 1.31 mycroft * reduce no of tokens, dequeue the packet,
1618 1.31 mycroft * send the packet.
1619 1.31 mycroft */
1620 1.31 mycroft if ((vifp->tbf_q = m->m_nextpkt) == 0)
1621 1.31 mycroft vifp->tbf_t = &vifp->tbf_q;
1622 1.31 mycroft --vifp->tbf_q_len;
1623 1.15 mycroft
1624 1.31 mycroft m->m_nextpkt = 0;
1625 1.31 mycroft vifp->tbf_n_tok -= len;
1626 1.31 mycroft tbf_send_packet(vifp, m);
1627 1.31 mycroft } else
1628 1.31 mycroft break;
1629 1.31 mycroft }
1630 1.31 mycroft splx(s);
1631 1.15 mycroft }
1632 1.15 mycroft
1633 1.15 mycroft static void
1634 1.20 mycroft tbf_reprocess_q(arg)
1635 1.20 mycroft void *arg;
1636 1.15 mycroft {
1637 1.48 augustss struct vif *vifp = arg;
1638 1.15 mycroft
1639 1.30 mycroft if (ip_mrouter == 0)
1640 1.20 mycroft return;
1641 1.15 mycroft
1642 1.20 mycroft tbf_update_tokens(vifp);
1643 1.20 mycroft tbf_process_q(vifp);
1644 1.15 mycroft
1645 1.31 mycroft if (vifp->tbf_q_len != 0)
1646 1.47 thorpej callout_reset(&vifp->v_repq_ch, TBF_REPROCESS,
1647 1.47 thorpej tbf_reprocess_q, vifp);
1648 1.15 mycroft }
1649 1.15 mycroft
1650 1.15 mycroft /* function that will selectively discard a member of the queue
1651 1.31 mycroft * based on the precedence value and the priority
1652 1.15 mycroft */
1653 1.15 mycroft static int
1654 1.15 mycroft tbf_dq_sel(vifp, ip)
1655 1.48 augustss struct vif *vifp;
1656 1.48 augustss struct ip *ip;
1657 1.15 mycroft {
1658 1.48 augustss u_int p;
1659 1.48 augustss struct mbuf **mp, *m;
1660 1.48 augustss int s = splsoftnet();
1661 1.31 mycroft
1662 1.31 mycroft p = priority(vifp, ip);
1663 1.31 mycroft
1664 1.31 mycroft for (mp = &vifp->tbf_q, m = *mp;
1665 1.31 mycroft m != 0;
1666 1.31 mycroft mp = &m->m_nextpkt, m = *mp) {
1667 1.31 mycroft if (p > priority(vifp, mtod(m, struct ip *))) {
1668 1.31 mycroft if ((*mp = m->m_nextpkt) == 0)
1669 1.31 mycroft vifp->tbf_t = mp;
1670 1.31 mycroft --vifp->tbf_q_len;
1671 1.31 mycroft
1672 1.31 mycroft m_freem(m);
1673 1.31 mycroft mrtstat.mrts_drop_sel++;
1674 1.31 mycroft splx(s);
1675 1.31 mycroft return (1);
1676 1.31 mycroft }
1677 1.15 mycroft }
1678 1.31 mycroft splx(s);
1679 1.31 mycroft return (0);
1680 1.15 mycroft }
1681 1.15 mycroft
1682 1.15 mycroft static void
1683 1.31 mycroft tbf_send_packet(vifp, m)
1684 1.48 augustss struct vif *vifp;
1685 1.48 augustss struct mbuf *m;
1686 1.15 mycroft {
1687 1.31 mycroft int error;
1688 1.31 mycroft int s = splsoftnet();
1689 1.31 mycroft
1690 1.31 mycroft if (vifp->v_flags & VIFF_TUNNEL) {
1691 1.31 mycroft /* If tunnel options */
1692 1.43 itojun #ifdef IPSEC
1693 1.46 itojun /* Don't lookup socket in forwading case */
1694 1.46 itojun ipsec_setsocket(m, NULL);
1695 1.43 itojun #endif
1696 1.31 mycroft ip_output(m, (struct mbuf *)0, &vifp->v_route,
1697 1.31 mycroft IP_FORWARDING, (struct ip_moptions *)0);
1698 1.31 mycroft } else {
1699 1.31 mycroft /* if physical interface option, extract the options and then send */
1700 1.31 mycroft struct ip_moptions imo;
1701 1.15 mycroft
1702 1.31 mycroft imo.imo_multicast_ifp = vifp->v_ifp;
1703 1.31 mycroft imo.imo_multicast_ttl = mtod(m, struct ip *)->ip_ttl - 1;
1704 1.31 mycroft imo.imo_multicast_loop = 1;
1705 1.15 mycroft #ifdef RSVP_ISI
1706 1.31 mycroft imo.imo_multicast_vif = -1;
1707 1.1 hpeyerl #endif
1708 1.15 mycroft
1709 1.43 itojun #ifdef IPSEC
1710 1.46 itojun /* Don't lookup socket in forwading case */
1711 1.46 itojun ipsec_setsocket(m, NULL);
1712 1.43 itojun #endif
1713 1.31 mycroft error = ip_output(m, (struct mbuf *)0, (struct route *)0,
1714 1.31 mycroft IP_FORWARDING|IP_MULTICASTOPTS, &imo);
1715 1.31 mycroft
1716 1.31 mycroft if (mrtdebug & DEBUG_XMIT)
1717 1.42 nathanw log(LOG_DEBUG, "phyint_send on vif %ld err %d\n",
1718 1.42 nathanw (long)(vifp-viftable), error);
1719 1.31 mycroft }
1720 1.31 mycroft splx(s);
1721 1.15 mycroft }
1722 1.15 mycroft
1723 1.15 mycroft /* determine the current time and then
1724 1.15 mycroft * the elapsed time (between the last time and time now)
1725 1.15 mycroft * in milliseconds & update the no. of tokens in the bucket
1726 1.15 mycroft */
1727 1.15 mycroft static void
1728 1.15 mycroft tbf_update_tokens(vifp)
1729 1.48 augustss struct vif *vifp;
1730 1.15 mycroft {
1731 1.31 mycroft struct timeval tp;
1732 1.48 augustss u_int32_t tm;
1733 1.48 augustss int s = splsoftnet();
1734 1.15 mycroft
1735 1.31 mycroft microtime(&tp);
1736 1.15 mycroft
1737 1.31 mycroft TV_DELTA(tp, vifp->tbf_last_pkt_t, tm);
1738 1.15 mycroft
1739 1.31 mycroft /*
1740 1.31 mycroft * This formula is actually
1741 1.31 mycroft * "time in seconds" * "bytes/second".
1742 1.31 mycroft *
1743 1.31 mycroft * (tm / 1000000) * (v_rate_limit * 1000 * (1000/1024) / 8)
1744 1.31 mycroft *
1745 1.31 mycroft * The (1000/1024) was introduced in add_vif to optimize
1746 1.31 mycroft * this divide into a shift.
1747 1.31 mycroft */
1748 1.31 mycroft vifp->tbf_n_tok += tm * vifp->v_rate_limit / 8192;
1749 1.31 mycroft vifp->tbf_last_pkt_t = tp;
1750 1.15 mycroft
1751 1.31 mycroft if (vifp->tbf_n_tok > MAX_BKT_SIZE)
1752 1.31 mycroft vifp->tbf_n_tok = MAX_BKT_SIZE;
1753 1.15 mycroft
1754 1.31 mycroft splx(s);
1755 1.15 mycroft }
1756 1.15 mycroft
1757 1.15 mycroft static int
1758 1.15 mycroft priority(vifp, ip)
1759 1.48 augustss struct vif *vifp;
1760 1.48 augustss struct ip *ip;
1761 1.15 mycroft {
1762 1.48 augustss int prio;
1763 1.15 mycroft
1764 1.15 mycroft /* temporary hack; may add general packet classifier some day */
1765 1.15 mycroft
1766 1.15 mycroft /*
1767 1.15 mycroft * The UDP port space is divided up into four priority ranges:
1768 1.15 mycroft * [0, 16384) : unclassified - lowest priority
1769 1.15 mycroft * [16384, 32768) : audio - highest priority
1770 1.15 mycroft * [32768, 49152) : whiteboard - medium priority
1771 1.15 mycroft * [49152, 65536) : video - low priority
1772 1.15 mycroft */
1773 1.15 mycroft if (ip->ip_p == IPPROTO_UDP) {
1774 1.15 mycroft struct udphdr *udp = (struct udphdr *)(((char *)ip) + (ip->ip_hl << 2));
1775 1.15 mycroft
1776 1.15 mycroft switch (ntohs(udp->uh_dport) & 0xc000) {
1777 1.15 mycroft case 0x4000:
1778 1.15 mycroft prio = 70;
1779 1.15 mycroft break;
1780 1.15 mycroft case 0x8000:
1781 1.15 mycroft prio = 60;
1782 1.15 mycroft break;
1783 1.15 mycroft case 0xc000:
1784 1.15 mycroft prio = 55;
1785 1.15 mycroft break;
1786 1.15 mycroft default:
1787 1.15 mycroft prio = 50;
1788 1.15 mycroft break;
1789 1.15 mycroft }
1790 1.15 mycroft
1791 1.30 mycroft if (tbfdebug > 1)
1792 1.30 mycroft log(LOG_DEBUG, "port %x prio %d\n", ntohs(udp->uh_dport), prio);
1793 1.15 mycroft } else
1794 1.15 mycroft prio = 50;
1795 1.15 mycroft
1796 1.15 mycroft
1797 1.15 mycroft return (prio);
1798 1.15 mycroft }
1799 1.15 mycroft
1800 1.15 mycroft /*
1801 1.15 mycroft * End of token bucket filter modifications
1802 1.15 mycroft */
1803 1.15 mycroft
1804 1.15 mycroft #ifdef RSVP_ISI
1805 1.15 mycroft
1806 1.15 mycroft int
1807 1.15 mycroft ip_rsvp_vif_init(so, m)
1808 1.15 mycroft struct socket *so;
1809 1.15 mycroft struct mbuf *m;
1810 1.15 mycroft {
1811 1.15 mycroft int i;
1812 1.48 augustss int s;
1813 1.15 mycroft
1814 1.15 mycroft if (rsvpdebug)
1815 1.34 christos printf("ip_rsvp_vif_init: so_type = %d, pr_protocol = %d\n",
1816 1.33 christos so->so_type, so->so_proto->pr_protocol);
1817 1.15 mycroft
1818 1.15 mycroft if (so->so_type != SOCK_RAW || so->so_proto->pr_protocol != IPPROTO_RSVP)
1819 1.15 mycroft return (EOPNOTSUPP);
1820 1.15 mycroft
1821 1.15 mycroft /* Check mbuf. */
1822 1.30 mycroft if (m == 0 || m->m_len != sizeof(int)) {
1823 1.15 mycroft return (EINVAL);
1824 1.15 mycroft }
1825 1.15 mycroft i = *(mtod(m, int *));
1826 1.15 mycroft
1827 1.15 mycroft if (rsvpdebug)
1828 1.34 christos printf("ip_rsvp_vif_init: vif = %d rsvp_on = %d\n",i,rsvp_on);
1829 1.15 mycroft
1830 1.24 mycroft s = splsoftnet();
1831 1.15 mycroft
1832 1.15 mycroft /* Check vif. */
1833 1.15 mycroft if (!legal_vif_num(i)) {
1834 1.15 mycroft splx(s);
1835 1.15 mycroft return (EADDRNOTAVAIL);
1836 1.15 mycroft }
1837 1.15 mycroft
1838 1.15 mycroft /* Check if socket is available. */
1839 1.30 mycroft if (viftable[i].v_rsvpd != 0) {
1840 1.15 mycroft splx(s);
1841 1.15 mycroft return (EADDRINUSE);
1842 1.15 mycroft }
1843 1.15 mycroft
1844 1.15 mycroft viftable[i].v_rsvpd = so;
1845 1.15 mycroft /* This may seem silly, but we need to be sure we don't over-increment
1846 1.15 mycroft * the RSVP counter, in case something slips up.
1847 1.15 mycroft */
1848 1.15 mycroft if (!viftable[i].v_rsvp_on) {
1849 1.15 mycroft viftable[i].v_rsvp_on = 1;
1850 1.15 mycroft rsvp_on++;
1851 1.15 mycroft }
1852 1.15 mycroft
1853 1.15 mycroft splx(s);
1854 1.15 mycroft return (0);
1855 1.15 mycroft }
1856 1.15 mycroft
1857 1.15 mycroft int
1858 1.15 mycroft ip_rsvp_vif_done(so, m)
1859 1.15 mycroft struct socket *so;
1860 1.15 mycroft struct mbuf *m;
1861 1.15 mycroft {
1862 1.15 mycroft int i;
1863 1.48 augustss int s;
1864 1.15 mycroft
1865 1.15 mycroft if (rsvpdebug)
1866 1.34 christos printf("ip_rsvp_vif_done: so_type = %d, pr_protocol = %d\n",
1867 1.15 mycroft so->so_type, so->so_proto->pr_protocol);
1868 1.15 mycroft
1869 1.15 mycroft if (so->so_type != SOCK_RAW || so->so_proto->pr_protocol != IPPROTO_RSVP)
1870 1.15 mycroft return (EOPNOTSUPP);
1871 1.15 mycroft
1872 1.15 mycroft /* Check mbuf. */
1873 1.30 mycroft if (m == 0 || m->m_len != sizeof(int)) {
1874 1.15 mycroft return (EINVAL);
1875 1.15 mycroft }
1876 1.15 mycroft i = *(mtod(m, int *));
1877 1.15 mycroft
1878 1.24 mycroft s = splsoftnet();
1879 1.15 mycroft
1880 1.15 mycroft /* Check vif. */
1881 1.15 mycroft if (!legal_vif_num(i)) {
1882 1.15 mycroft splx(s);
1883 1.15 mycroft return (EADDRNOTAVAIL);
1884 1.15 mycroft }
1885 1.15 mycroft
1886 1.15 mycroft if (rsvpdebug)
1887 1.34 christos printf("ip_rsvp_vif_done: v_rsvpd = %x so = %x\n",
1888 1.33 christos viftable[i].v_rsvpd, so);
1889 1.15 mycroft
1890 1.30 mycroft viftable[i].v_rsvpd = 0;
1891 1.15 mycroft /* This may seem silly, but we need to be sure we don't over-decrement
1892 1.15 mycroft * the RSVP counter, in case something slips up.
1893 1.15 mycroft */
1894 1.15 mycroft if (viftable[i].v_rsvp_on) {
1895 1.15 mycroft viftable[i].v_rsvp_on = 0;
1896 1.15 mycroft rsvp_on--;
1897 1.15 mycroft }
1898 1.15 mycroft
1899 1.15 mycroft splx(s);
1900 1.15 mycroft return (0);
1901 1.15 mycroft }
1902 1.15 mycroft
1903 1.25 christos void
1904 1.15 mycroft ip_rsvp_force_done(so)
1905 1.15 mycroft struct socket *so;
1906 1.15 mycroft {
1907 1.15 mycroft int vifi;
1908 1.48 augustss int s;
1909 1.15 mycroft
1910 1.15 mycroft /* Don't bother if it is not the right type of socket. */
1911 1.15 mycroft if (so->so_type != SOCK_RAW || so->so_proto->pr_protocol != IPPROTO_RSVP)
1912 1.15 mycroft return;
1913 1.15 mycroft
1914 1.24 mycroft s = splsoftnet();
1915 1.15 mycroft
1916 1.15 mycroft /* The socket may be attached to more than one vif...this
1917 1.15 mycroft * is perfectly legal.
1918 1.15 mycroft */
1919 1.15 mycroft for (vifi = 0; vifi < numvifs; vifi++) {
1920 1.15 mycroft if (viftable[vifi].v_rsvpd == so) {
1921 1.30 mycroft viftable[vifi].v_rsvpd = 0;
1922 1.15 mycroft /* This may seem silly, but we need to be sure we don't
1923 1.15 mycroft * over-decrement the RSVP counter, in case something slips up.
1924 1.15 mycroft */
1925 1.15 mycroft if (viftable[vifi].v_rsvp_on) {
1926 1.15 mycroft viftable[vifi].v_rsvp_on = 0;
1927 1.15 mycroft rsvp_on--;
1928 1.15 mycroft }
1929 1.15 mycroft }
1930 1.15 mycroft }
1931 1.15 mycroft
1932 1.15 mycroft splx(s);
1933 1.15 mycroft return;
1934 1.15 mycroft }
1935 1.15 mycroft
1936 1.25 christos void
1937 1.15 mycroft rsvp_input(m, ifp)
1938 1.15 mycroft struct mbuf *m;
1939 1.15 mycroft struct ifnet *ifp;
1940 1.15 mycroft {
1941 1.15 mycroft int vifi;
1942 1.48 augustss struct ip *ip = mtod(m, struct ip *);
1943 1.15 mycroft static struct sockaddr_in rsvp_src = { sizeof(sin), AF_INET };
1944 1.48 augustss int s;
1945 1.15 mycroft
1946 1.15 mycroft if (rsvpdebug)
1947 1.34 christos printf("rsvp_input: rsvp_on %d\n",rsvp_on);
1948 1.15 mycroft
1949 1.15 mycroft /* Can still get packets with rsvp_on = 0 if there is a local member
1950 1.15 mycroft * of the group to which the RSVP packet is addressed. But in this
1951 1.15 mycroft * case we want to throw the packet away.
1952 1.15 mycroft */
1953 1.15 mycroft if (!rsvp_on) {
1954 1.15 mycroft m_freem(m);
1955 1.15 mycroft return;
1956 1.15 mycroft }
1957 1.15 mycroft
1958 1.15 mycroft /* If the old-style non-vif-associated socket is set, then use
1959 1.15 mycroft * it and ignore the new ones.
1960 1.15 mycroft */
1961 1.30 mycroft if (ip_rsvpd != 0) {
1962 1.15 mycroft if (rsvpdebug)
1963 1.34 christos printf("rsvp_input: Sending packet up old-style socket\n");
1964 1.43 itojun rip_input(m); /*XXX*/
1965 1.15 mycroft return;
1966 1.15 mycroft }
1967 1.15 mycroft
1968 1.24 mycroft s = splsoftnet();
1969 1.15 mycroft
1970 1.15 mycroft if (rsvpdebug)
1971 1.34 christos printf("rsvp_input: check vifs\n");
1972 1.15 mycroft
1973 1.15 mycroft /* Find which vif the packet arrived on. */
1974 1.15 mycroft for (vifi = 0; vifi < numvifs; vifi++) {
1975 1.15 mycroft if (viftable[vifi].v_ifp == ifp)
1976 1.15 mycroft break;
1977 1.15 mycroft }
1978 1.15 mycroft
1979 1.15 mycroft if (vifi == numvifs) {
1980 1.15 mycroft /* Can't find vif packet arrived on. Drop packet. */
1981 1.15 mycroft if (rsvpdebug)
1982 1.34 christos printf("rsvp_input: Can't find vif for packet...dropping it.\n");
1983 1.15 mycroft m_freem(m);
1984 1.15 mycroft splx(s);
1985 1.15 mycroft return;
1986 1.15 mycroft }
1987 1.15 mycroft
1988 1.15 mycroft if (rsvpdebug)
1989 1.34 christos printf("rsvp_input: check socket\n");
1990 1.15 mycroft
1991 1.30 mycroft if (viftable[vifi].v_rsvpd == 0) {
1992 1.15 mycroft /* drop packet, since there is no specific socket for this
1993 1.15 mycroft * interface */
1994 1.15 mycroft if (rsvpdebug)
1995 1.34 christos printf("rsvp_input: No socket defined for vif %d\n",vifi);
1996 1.15 mycroft m_freem(m);
1997 1.15 mycroft splx(s);
1998 1.15 mycroft return;
1999 1.15 mycroft }
2000 1.15 mycroft
2001 1.15 mycroft rsvp_src.sin_addr = ip->ip_src;
2002 1.15 mycroft
2003 1.15 mycroft if (rsvpdebug && m)
2004 1.34 christos printf("rsvp_input: m->m_len = %d, sbspace() = %d\n",
2005 1.15 mycroft m->m_len,sbspace(&viftable[vifi].v_rsvpd->so_rcv));
2006 1.15 mycroft
2007 1.15 mycroft if (socket_send(viftable[vifi].v_rsvpd, m, &rsvp_src) < 0)
2008 1.15 mycroft if (rsvpdebug)
2009 1.34 christos printf("rsvp_input: Failed to append to socket\n");
2010 1.15 mycroft else
2011 1.15 mycroft if (rsvpdebug)
2012 1.34 christos printf("rsvp_input: send packet up\n");
2013 1.15 mycroft
2014 1.15 mycroft splx(s);
2015 1.15 mycroft }
2016 1.15 mycroft #endif /* RSVP_ISI */
2017