1 | /* |
---|
2 | * IP multicast forwarding procedures |
---|
3 | * |
---|
4 | * Written by David Waitzman, BBN Labs, August 1988. |
---|
5 | * Modified by Steve Deering, Stanford, February 1989. |
---|
6 | * Modified by Mark J. Steiglitz, Stanford, May, 1991 |
---|
7 | * Modified by Van Jacobson, LBL, January 1993 |
---|
8 | * Modified by Ajit Thyagarajan, PARC, August 1993 |
---|
9 | * Modified by Bill Fenner, PARC, April 1995 |
---|
10 | * |
---|
11 | * MROUTING Revision: 3.5 |
---|
12 | * $Id$ |
---|
13 | */ |
---|
14 | |
---|
15 | #include "opt_mrouting.h" |
---|
16 | |
---|
17 | #include <sys/param.h> |
---|
18 | #include <rtems/bsd/sys/queue.h> |
---|
19 | #include <sys/systm.h> |
---|
20 | #include <sys/sysctl.h> |
---|
21 | #include <sys/mbuf.h> |
---|
22 | #include <sys/socket.h> |
---|
23 | #include <sys/socketvar.h> |
---|
24 | #include <sys/protosw.h> |
---|
25 | #include <sys/errno.h> |
---|
26 | #include <sys/time.h> |
---|
27 | #include <sys/kernel.h> |
---|
28 | #include <sys/ioctl.h> |
---|
29 | #include <sys/syslog.h> |
---|
30 | #include <net/if.h> |
---|
31 | #include <net/route.h> |
---|
32 | #include <netinet/in.h> |
---|
33 | #include <netinet/in_systm.h> |
---|
34 | #include <netinet/ip.h> |
---|
35 | #include <netinet/ip_var.h> |
---|
36 | #include <netinet/in_pcb.h> |
---|
37 | #include <netinet/in_var.h> |
---|
38 | #include <netinet/igmp.h> |
---|
39 | #include <netinet/igmp_var.h> |
---|
40 | #include <netinet/ip_mroute.h> |
---|
41 | #include <netinet/udp.h> |
---|
42 | |
---|
43 | #ifndef NTOHL |
---|
44 | #if BYTE_ORDER != BIG_ENDIAN |
---|
45 | #define NTOHL(d) ((d) = ntohl((d))) |
---|
46 | #define NTOHS(d) ((d) = ntohs((u_short)(d))) |
---|
47 | #define HTONL(d) ((d) = htonl((d))) |
---|
48 | #define HTONS(d) ((d) = htons((u_short)(d))) |
---|
49 | #else |
---|
50 | #define NTOHL(d) |
---|
51 | #define NTOHS(d) |
---|
52 | #define HTONL(d) |
---|
53 | #define HTONS(d) |
---|
54 | #endif |
---|
55 | #endif |
---|
56 | |
---|
57 | #ifndef MROUTING |
---|
58 | extern u_long _ip_mcast_src(int vifi); |
---|
59 | extern int _ip_mforward(struct ip *ip, struct ifnet *ifp, |
---|
60 | struct mbuf *m, struct ip_moptions *imo); |
---|
61 | extern int _ip_mrouter_done(void); |
---|
62 | extern int _ip_mrouter_get(int cmd, struct socket *so, |
---|
63 | struct mbuf **m); |
---|
64 | extern int _ip_mrouter_set(int cmd, struct socket *so, |
---|
65 | struct mbuf *m); |
---|
66 | extern int _mrt_ioctl(int req, caddr_t data, struct proc *p); |
---|
67 | |
---|
68 | /* |
---|
69 | * Dummy routines and globals used when multicast routing is not compiled in. |
---|
70 | */ |
---|
71 | |
---|
72 | struct socket *ip_mrouter = NULL; |
---|
73 | /* static u_int ip_mrtproto = 0; */ |
---|
74 | /* static struct mrtstat mrtstat; */ |
---|
75 | u_int rsvpdebug = 0; |
---|
76 | |
---|
77 | int |
---|
78 | _ip_mrouter_set(int cmd, struct socket *so, struct mbuf *m) |
---|
79 | { |
---|
80 | return(EOPNOTSUPP); |
---|
81 | } |
---|
82 | |
---|
83 | int (*ip_mrouter_set)(int, struct socket *, struct mbuf *) = _ip_mrouter_set; |
---|
84 | |
---|
85 | |
---|
86 | int |
---|
87 | _ip_mrouter_get(int cmd, struct socket *so, struct mbuf **m) |
---|
88 | { |
---|
89 | return(EOPNOTSUPP); |
---|
90 | } |
---|
91 | |
---|
92 | int (*ip_mrouter_get)(int, struct socket *, struct mbuf **) = _ip_mrouter_get; |
---|
93 | |
---|
94 | int |
---|
95 | _ip_mrouter_done(void) |
---|
96 | { |
---|
97 | return(0); |
---|
98 | } |
---|
99 | |
---|
100 | int (*ip_mrouter_done)(void) = _ip_mrouter_done; |
---|
101 | |
---|
102 | int |
---|
103 | _ip_mforward(struct ip *ip, struct ifnet *ifp, struct mbuf *m, |
---|
104 | struct ip_moptions *imo) |
---|
105 | { |
---|
106 | return(0); |
---|
107 | } |
---|
108 | |
---|
109 | int (*ip_mforward)(struct ip *, struct ifnet *, struct mbuf *, |
---|
110 | struct ip_moptions *) = _ip_mforward; |
---|
111 | |
---|
112 | int |
---|
113 | _mrt_ioctl(int req, caddr_t data, struct proc *p) |
---|
114 | { |
---|
115 | return EOPNOTSUPP; |
---|
116 | } |
---|
117 | |
---|
118 | int (*mrt_ioctl)(int, caddr_t, struct proc *) = _mrt_ioctl; |
---|
119 | |
---|
120 | void |
---|
121 | rsvp_input(struct mbuf *m, int iphlen) |
---|
122 | { |
---|
123 | /* Can still get packets with rsvp_on = 0 if there is a local member |
---|
124 | * of the group to which the RSVP packet is addressed. But in this |
---|
125 | * case we want to throw the packet away. |
---|
126 | */ |
---|
127 | if (!rsvp_on) { |
---|
128 | m_freem(m); |
---|
129 | return; |
---|
130 | } |
---|
131 | |
---|
132 | if (ip_rsvpd != NULL) { |
---|
133 | if (rsvpdebug) |
---|
134 | printf("rsvp_input: Sending packet up old-style socket\n"); |
---|
135 | rip_input(m, iphlen); |
---|
136 | return; |
---|
137 | } |
---|
138 | /* Drop the packet */ |
---|
139 | m_freem(m); |
---|
140 | } |
---|
141 | |
---|
142 | void ipip_input(struct mbuf *m, int iphlen) |
---|
143 | { /* XXX must fixup manually */ |
---|
144 | rip_input(m, iphlen); |
---|
145 | } |
---|
146 | |
---|
147 | int (*legal_vif_num)(int) = 0; |
---|
148 | |
---|
149 | /* |
---|
150 | * This should never be called, since IP_MULTICAST_VIF should fail, but |
---|
151 | * just in case it does get called, the code a little lower in ip_output |
---|
152 | * will assign the packet a local address. |
---|
153 | */ |
---|
154 | u_long |
---|
155 | _ip_mcast_src(int vifi) { return INADDR_ANY; } |
---|
156 | u_long (*ip_mcast_src)(int) = _ip_mcast_src; |
---|
157 | |
---|
158 | int |
---|
159 | ip_rsvp_vif_init(struct socket *so, struct mbuf *m) |
---|
160 | { |
---|
161 | return(EINVAL); |
---|
162 | } |
---|
163 | |
---|
164 | int |
---|
165 | ip_rsvp_vif_done(struct socket *so, struct mbuf *m) |
---|
166 | { |
---|
167 | return(EINVAL); |
---|
168 | } |
---|
169 | |
---|
170 | void |
---|
171 | ip_rsvp_force_done(struct socket *so) |
---|
172 | { |
---|
173 | return; |
---|
174 | } |
---|
175 | |
---|
176 | #else /* MROUTING */ |
---|
177 | |
---|
178 | #define M_HASCL(m) ((m)->m_flags & M_EXT) |
---|
179 | |
---|
180 | #define INSIZ sizeof(struct in_addr) |
---|
181 | #define same(a1, a2) \ |
---|
182 | (bcmp((caddr_t)(a1), (caddr_t)(a2), INSIZ) == 0) |
---|
183 | |
---|
184 | #define MT_MRTABLE MT_RTABLE /* since nothing else uses it */ |
---|
185 | |
---|
186 | /* |
---|
187 | * Globals. All but ip_mrouter and ip_mrtproto could be static, |
---|
188 | * except for netstat or debugging purposes. |
---|
189 | */ |
---|
190 | #ifndef MROUTE_LKM |
---|
191 | struct socket *ip_mrouter = NULL; |
---|
192 | struct mrtstat mrtstat; |
---|
193 | |
---|
194 | int ip_mrtproto = IGMP_DVMRP; /* for netstat only */ |
---|
195 | #else /* MROUTE_LKM */ |
---|
196 | extern void X_ipip_input(struct mbuf *m, int iphlen); |
---|
197 | extern struct mrtstat mrtstat; |
---|
198 | static int ip_mrtproto; |
---|
199 | #endif |
---|
200 | |
---|
201 | #define NO_RTE_FOUND 0x1 |
---|
202 | #define RTE_FOUND 0x2 |
---|
203 | |
---|
204 | static struct mbuf *mfctable[MFCTBLSIZ]; |
---|
205 | static u_char nexpire[MFCTBLSIZ]; |
---|
206 | static struct vif viftable[MAXVIFS]; |
---|
207 | static u_int mrtdebug = 0; /* debug level */ |
---|
208 | #define DEBUG_MFC 0x02 |
---|
209 | #define DEBUG_FORWARD 0x04 |
---|
210 | #define DEBUG_EXPIRE 0x08 |
---|
211 | #define DEBUG_XMIT 0x10 |
---|
212 | static u_int tbfdebug = 0; /* tbf debug level */ |
---|
213 | static u_int rsvpdebug = 0; /* rsvp debug level */ |
---|
214 | |
---|
215 | #define EXPIRE_TIMEOUT (hz / 4) /* 4x / second */ |
---|
216 | #define UPCALL_EXPIRE 6 /* number of timeouts */ |
---|
217 | |
---|
218 | /* |
---|
219 | * Define the token bucket filter structures |
---|
220 | * tbftable -> each vif has one of these for storing info |
---|
221 | */ |
---|
222 | |
---|
223 | static struct tbf tbftable[MAXVIFS]; |
---|
224 | #define TBF_REPROCESS (hz / 100) /* 100x / second */ |
---|
225 | |
---|
226 | /* |
---|
227 | * 'Interfaces' associated with decapsulator (so we can tell |
---|
228 | * packets that went through it from ones that get reflected |
---|
229 | * by a broken gateway). These interfaces are never linked into |
---|
230 | * the system ifnet list & no routes point to them. I.e., packets |
---|
231 | * can't be sent this way. They only exist as a placeholder for |
---|
232 | * multicast source verification. |
---|
233 | */ |
---|
234 | static struct ifnet multicast_decap_if[MAXVIFS]; |
---|
235 | |
---|
236 | #define ENCAP_TTL 64 |
---|
237 | #define ENCAP_PROTO IPPROTO_IPIP /* 4 */ |
---|
238 | |
---|
239 | /* prototype IP hdr for encapsulated packets */ |
---|
240 | static struct ip multicast_encap_iphdr = { |
---|
241 | #if BYTE_ORDER == LITTLE_ENDIAN |
---|
242 | sizeof(struct ip) >> 2, IPVERSION, |
---|
243 | #else |
---|
244 | IPVERSION, sizeof(struct ip) >> 2, |
---|
245 | #endif |
---|
246 | 0, /* tos */ |
---|
247 | sizeof(struct ip), /* total length */ |
---|
248 | 0, /* id */ |
---|
249 | 0, /* frag offset */ |
---|
250 | ENCAP_TTL, ENCAP_PROTO, |
---|
251 | 0, /* checksum */ |
---|
252 | }; |
---|
253 | |
---|
254 | /* |
---|
255 | * Private variables. |
---|
256 | */ |
---|
257 | static vifi_t numvifs = 0; |
---|
258 | static int have_encap_tunnel = 0; |
---|
259 | |
---|
260 | /* |
---|
261 | * one-back cache used by ipip_input to locate a tunnel's vif |
---|
262 | * given a datagram's src ip address. |
---|
263 | */ |
---|
264 | static u_long last_encap_src; |
---|
265 | static struct vif *last_encap_vif; |
---|
266 | |
---|
267 | static u_long X_ip_mcast_src(int vifi); |
---|
268 | static int X_ip_mforward(struct ip *ip, struct ifnet *ifp, struct mbuf *m, struct ip_moptions *imo); |
---|
269 | static int X_ip_mrouter_done(void); |
---|
270 | static int X_ip_mrouter_get(int cmd, struct socket *so, struct mbuf **m); |
---|
271 | static int X_ip_mrouter_set(int cmd, struct socket *so, struct mbuf *m); |
---|
272 | static int X_legal_vif_num(int vif); |
---|
273 | static int X_mrt_ioctl(int cmd, caddr_t data); |
---|
274 | |
---|
275 | static int get_sg_cnt(struct sioc_sg_req *); |
---|
276 | static int get_vif_cnt(struct sioc_vif_req *); |
---|
277 | static int ip_mrouter_init(struct socket *, struct mbuf *); |
---|
278 | static int add_vif(struct vifctl *); |
---|
279 | static int del_vif(vifi_t *); |
---|
280 | static int add_mfc(struct mfcctl *); |
---|
281 | static int del_mfc(struct mfcctl *); |
---|
282 | static int socket_send(struct socket *, struct mbuf *, struct sockaddr_in *); |
---|
283 | static int get_version(struct mbuf *); |
---|
284 | static int get_assert(struct mbuf *); |
---|
285 | static int set_assert(int *); |
---|
286 | static void expire_upcalls(void *); |
---|
287 | static int ip_mdq(struct mbuf *, struct ifnet *, struct mfc *, |
---|
288 | vifi_t); |
---|
289 | static void phyint_send(struct ip *, struct vif *, struct mbuf *); |
---|
290 | static void encap_send(struct ip *, struct vif *, struct mbuf *); |
---|
291 | static void tbf_control(struct vif *, struct mbuf *, struct ip *, u_long); |
---|
292 | static void tbf_queue(struct vif *, struct mbuf *); |
---|
293 | static void tbf_process_q(struct vif *); |
---|
294 | static void tbf_reprocess_q(void *); |
---|
295 | static int tbf_dq_sel(struct vif *, struct ip *); |
---|
296 | static void tbf_send_packet(struct vif *, struct mbuf *); |
---|
297 | static void tbf_update_tokens(struct vif *); |
---|
298 | static int priority(struct vif *, struct ip *); |
---|
299 | void multiencap_decap(struct mbuf *); |
---|
300 | |
---|
301 | /* |
---|
302 | * whether or not special PIM assert processing is enabled. |
---|
303 | */ |
---|
304 | static int pim_assert; |
---|
305 | /* |
---|
306 | * Rate limit for assert notification messages, in usec |
---|
307 | */ |
---|
308 | #define ASSERT_MSG_TIME 3000000 |
---|
309 | |
---|
310 | /* |
---|
311 | * Hash function for a source, group entry |
---|
312 | */ |
---|
313 | #define MFCHASH(a, g) MFCHASHMOD(((a) >> 20) ^ ((a) >> 10) ^ (a) ^ \ |
---|
314 | ((g) >> 20) ^ ((g) >> 10) ^ (g)) |
---|
315 | |
---|
316 | /* |
---|
317 | * Find a route for a given origin IP address and Multicast group address |
---|
318 | * Type of service parameter to be added in the future!!! |
---|
319 | */ |
---|
320 | |
---|
321 | #define MFCFIND(o, g, rt) { \ |
---|
322 | register struct mbuf *_mb_rt = mfctable[MFCHASH(o,g)]; \ |
---|
323 | register struct mfc *_rt = NULL; \ |
---|
324 | rt = NULL; \ |
---|
325 | ++mrtstat.mrts_mfc_lookups; \ |
---|
326 | while (_mb_rt) { \ |
---|
327 | _rt = mtod(_mb_rt, struct mfc *); \ |
---|
328 | if ((_rt->mfc_origin.s_addr == o) && \ |
---|
329 | (_rt->mfc_mcastgrp.s_addr == g) && \ |
---|
330 | (_mb_rt->m_act == NULL)) { \ |
---|
331 | rt = _rt; \ |
---|
332 | break; \ |
---|
333 | } \ |
---|
334 | _mb_rt = _mb_rt->m_next; \ |
---|
335 | } \ |
---|
336 | if (rt == NULL) { \ |
---|
337 | ++mrtstat.mrts_mfc_misses; \ |
---|
338 | } \ |
---|
339 | } |
---|
340 | |
---|
341 | |
---|
342 | /* |
---|
343 | * Macros to compute elapsed time efficiently |
---|
344 | * Borrowed from Van Jacobson's scheduling code |
---|
345 | */ |
---|
346 | #define TV_DELTA(a, b, delta) { \ |
---|
347 | register int xxs; \ |
---|
348 | \ |
---|
349 | delta = (a).tv_usec - (b).tv_usec; \ |
---|
350 | if ((xxs = (a).tv_sec - (b).tv_sec)) { \ |
---|
351 | switch (xxs) { \ |
---|
352 | case 2: \ |
---|
353 | delta += 1000000; \ |
---|
354 | /* fall through */ \ |
---|
355 | case 1: \ |
---|
356 | delta += 1000000; \ |
---|
357 | break; \ |
---|
358 | default: \ |
---|
359 | delta += (1000000 * xxs); \ |
---|
360 | } \ |
---|
361 | } \ |
---|
362 | } |
---|
363 | |
---|
364 | #define TV_LT(a, b) (((a).tv_usec < (b).tv_usec && \ |
---|
365 | (a).tv_sec <= (b).tv_sec) || (a).tv_sec < (b).tv_sec) |
---|
366 | |
---|
367 | #ifdef UPCALL_TIMING |
---|
368 | u_long upcall_data[51]; |
---|
369 | static void collate(struct timeval *); |
---|
370 | #endif /* UPCALL_TIMING */ |
---|
371 | |
---|
372 | |
---|
373 | /* |
---|
374 | * Handle MRT setsockopt commands to modify the multicast routing tables. |
---|
375 | */ |
---|
376 | static int |
---|
377 | X_ip_mrouter_set(int cmd, struct socket *so, struct mbuf *m) |
---|
378 | { |
---|
379 | if (cmd != MRT_INIT && so != ip_mrouter) return EACCES; |
---|
380 | |
---|
381 | switch (cmd) { |
---|
382 | case MRT_INIT: return ip_mrouter_init(so, m); |
---|
383 | case MRT_DONE: return ip_mrouter_done(); |
---|
384 | case MRT_ADD_VIF: return add_vif (mtod(m, struct vifctl *)); |
---|
385 | case MRT_DEL_VIF: return del_vif (mtod(m, vifi_t *)); |
---|
386 | case MRT_ADD_MFC: return add_mfc (mtod(m, struct mfcctl *)); |
---|
387 | case MRT_DEL_MFC: return del_mfc (mtod(m, struct mfcctl *)); |
---|
388 | case MRT_ASSERT: return set_assert(mtod(m, int *)); |
---|
389 | default: return EOPNOTSUPP; |
---|
390 | } |
---|
391 | } |
---|
392 | |
---|
393 | #ifndef MROUTE_LKM |
---|
394 | int (*ip_mrouter_set)(int, struct socket *, struct mbuf *) = X_ip_mrouter_set; |
---|
395 | #endif |
---|
396 | |
---|
397 | /* |
---|
398 | * Handle MRT getsockopt commands |
---|
399 | */ |
---|
400 | static int |
---|
401 | X_ip_mrouter_get(int cmd, struct socket *so, struct mbuf **m) |
---|
402 | { |
---|
403 | struct mbuf *mb; |
---|
404 | |
---|
405 | if (so != ip_mrouter) return EACCES; |
---|
406 | |
---|
407 | *m = mb = m_get(M_WAIT, MT_SOOPTS); |
---|
408 | |
---|
409 | switch (cmd) { |
---|
410 | case MRT_VERSION: return get_version(mb); |
---|
411 | case MRT_ASSERT: return get_assert(mb); |
---|
412 | default: return EOPNOTSUPP; |
---|
413 | } |
---|
414 | } |
---|
415 | |
---|
416 | #ifndef MROUTE_LKM |
---|
417 | int (*ip_mrouter_get)(int, struct socket *, struct mbuf **) = X_ip_mrouter_get; |
---|
418 | #endif |
---|
419 | |
---|
420 | /* |
---|
421 | * Handle ioctl commands to obtain information from the cache |
---|
422 | */ |
---|
423 | static int |
---|
424 | X_mrt_ioctl(int cmd, caddr_t data) |
---|
425 | { |
---|
426 | int error = 0; |
---|
427 | |
---|
428 | switch (cmd) { |
---|
429 | case (SIOCGETVIFCNT): |
---|
430 | return (get_vif_cnt((struct sioc_vif_req *)data)); |
---|
431 | break; |
---|
432 | case (SIOCGETSGCNT): |
---|
433 | return (get_sg_cnt((struct sioc_sg_req *)data)); |
---|
434 | break; |
---|
435 | default: |
---|
436 | return (EINVAL); |
---|
437 | break; |
---|
438 | } |
---|
439 | return error; |
---|
440 | } |
---|
441 | |
---|
442 | #ifndef MROUTE_LKM |
---|
443 | int (*mrt_ioctl)(int, caddr_t) = X_mrt_ioctl; |
---|
444 | #endif |
---|
445 | |
---|
446 | /* |
---|
447 | * returns the packet, byte, rpf-failure count for the source group provided |
---|
448 | */ |
---|
449 | static int |
---|
450 | get_sg_cnt(struct sioc_sg_req *req) |
---|
451 | { |
---|
452 | register struct mfc *rt; |
---|
453 | int s; |
---|
454 | |
---|
455 | s = splnet(); |
---|
456 | MFCFIND(req->src.s_addr, req->grp.s_addr, rt); |
---|
457 | splx(s); |
---|
458 | if (rt != NULL) { |
---|
459 | req->pktcnt = rt->mfc_pkt_cnt; |
---|
460 | req->bytecnt = rt->mfc_byte_cnt; |
---|
461 | req->wrong_if = rt->mfc_wrong_if; |
---|
462 | } else |
---|
463 | req->pktcnt = req->bytecnt = req->wrong_if = 0xffffffff; |
---|
464 | |
---|
465 | return 0; |
---|
466 | } |
---|
467 | |
---|
468 | /* |
---|
469 | * returns the input and output packet and byte counts on the vif provided |
---|
470 | */ |
---|
471 | static int |
---|
472 | get_vif_cnt(struct sioc_vif_req *req) |
---|
473 | { |
---|
474 | register vifi_t vifi = req->vifi; |
---|
475 | |
---|
476 | if (vifi >= numvifs) return EINVAL; |
---|
477 | |
---|
478 | req->icount = viftable[vifi].v_pkt_in; |
---|
479 | req->ocount = viftable[vifi].v_pkt_out; |
---|
480 | req->ibytes = viftable[vifi].v_bytes_in; |
---|
481 | req->obytes = viftable[vifi].v_bytes_out; |
---|
482 | |
---|
483 | return 0; |
---|
484 | } |
---|
485 | |
---|
486 | /* |
---|
487 | * Enable multicast routing |
---|
488 | */ |
---|
489 | static int |
---|
490 | ip_mrouter_init(struct socket *so, struct mbuf *m) |
---|
491 | { |
---|
492 | int *v; |
---|
493 | |
---|
494 | if (mrtdebug) |
---|
495 | log(LOG_DEBUG,"ip_mrouter_init: so_type = %d, pr_protocol = %d\n", |
---|
496 | so->so_type, so->so_proto->pr_protocol); |
---|
497 | |
---|
498 | if (so->so_type != SOCK_RAW || |
---|
499 | so->so_proto->pr_protocol != IPPROTO_IGMP) return EOPNOTSUPP; |
---|
500 | |
---|
501 | if (!m || (m->m_len != sizeof(int *))) |
---|
502 | return ENOPROTOOPT; |
---|
503 | |
---|
504 | v = mtod(m, int *); |
---|
505 | if (*v != 1) |
---|
506 | return ENOPROTOOPT; |
---|
507 | |
---|
508 | if (ip_mrouter != NULL) return EADDRINUSE; |
---|
509 | |
---|
510 | ip_mrouter = so; |
---|
511 | |
---|
512 | bzero((caddr_t)mfctable, sizeof(mfctable)); |
---|
513 | bzero((caddr_t)nexpire, sizeof(nexpire)); |
---|
514 | |
---|
515 | pim_assert = 0; |
---|
516 | |
---|
517 | timeout(expire_upcalls, (caddr_t)NULL, EXPIRE_TIMEOUT); |
---|
518 | |
---|
519 | if (mrtdebug) |
---|
520 | log(LOG_DEBUG, "ip_mrouter_init\n"); |
---|
521 | |
---|
522 | return 0; |
---|
523 | } |
---|
524 | |
---|
525 | /* |
---|
526 | * Disable multicast routing |
---|
527 | */ |
---|
528 | static int |
---|
529 | X_ip_mrouter_done(void) |
---|
530 | { |
---|
531 | vifi_t vifi; |
---|
532 | int i; |
---|
533 | struct ifnet *ifp; |
---|
534 | struct ifreq ifr; |
---|
535 | struct mbuf *mb_rt; |
---|
536 | struct mbuf *m; |
---|
537 | struct rtdetq *rte; |
---|
538 | int s; |
---|
539 | |
---|
540 | s = splnet(); |
---|
541 | |
---|
542 | /* |
---|
543 | * For each phyint in use, disable promiscuous reception of all IP |
---|
544 | * multicasts. |
---|
545 | */ |
---|
546 | for (vifi = 0; vifi < numvifs; vifi++) { |
---|
547 | if (viftable[vifi].v_lcl_addr.s_addr != 0 && |
---|
548 | !(viftable[vifi].v_flags & VIFF_TUNNEL)) { |
---|
549 | ((struct sockaddr_in *)&(ifr.ifr_addr))->sin_family = AF_INET; |
---|
550 | ((struct sockaddr_in *)&(ifr.ifr_addr))->sin_addr.s_addr |
---|
551 | = INADDR_ANY; |
---|
552 | ifp = viftable[vifi].v_ifp; |
---|
553 | (*ifp->if_ioctl)(ifp, SIOCDELMULTI, (caddr_t)&ifr); |
---|
554 | } |
---|
555 | } |
---|
556 | bzero((caddr_t)tbftable, sizeof(tbftable)); |
---|
557 | bzero((caddr_t)viftable, sizeof(viftable)); |
---|
558 | numvifs = 0; |
---|
559 | pim_assert = 0; |
---|
560 | |
---|
561 | untimeout(expire_upcalls, (caddr_t)NULL); |
---|
562 | |
---|
563 | /* |
---|
564 | * Free all multicast forwarding cache entries. |
---|
565 | */ |
---|
566 | for (i = 0; i < MFCTBLSIZ; i++) { |
---|
567 | mb_rt = mfctable[i]; |
---|
568 | while (mb_rt) { |
---|
569 | if (mb_rt->m_act != NULL) { |
---|
570 | while (mb_rt->m_act) { |
---|
571 | m = mb_rt->m_act; |
---|
572 | mb_rt->m_act = m->m_act; |
---|
573 | rte = mtod(m, struct rtdetq *); |
---|
574 | m_freem(rte->m); |
---|
575 | m_free(m); |
---|
576 | } |
---|
577 | } |
---|
578 | mb_rt = m_free(mb_rt); |
---|
579 | } |
---|
580 | } |
---|
581 | |
---|
582 | bzero((caddr_t)mfctable, sizeof(mfctable)); |
---|
583 | |
---|
584 | /* |
---|
585 | * Reset de-encapsulation cache |
---|
586 | */ |
---|
587 | last_encap_src = 0; |
---|
588 | last_encap_vif = NULL; |
---|
589 | have_encap_tunnel = 0; |
---|
590 | |
---|
591 | ip_mrouter = NULL; |
---|
592 | |
---|
593 | splx(s); |
---|
594 | |
---|
595 | if (mrtdebug) |
---|
596 | log(LOG_DEBUG, "ip_mrouter_done\n"); |
---|
597 | |
---|
598 | return 0; |
---|
599 | } |
---|
600 | |
---|
601 | #ifndef MROUTE_LKM |
---|
602 | int (*ip_mrouter_done)(void) = X_ip_mrouter_done; |
---|
603 | #endif |
---|
604 | |
---|
605 | static int |
---|
606 | get_version(struct mbuf *mb) |
---|
607 | { |
---|
608 | int *v; |
---|
609 | |
---|
610 | v = mtod(mb, int *); |
---|
611 | |
---|
612 | *v = 0x0305; /* XXX !!!! */ |
---|
613 | mb->m_len = sizeof(int); |
---|
614 | |
---|
615 | return 0; |
---|
616 | } |
---|
617 | |
---|
618 | /* |
---|
619 | * Set PIM assert processing global |
---|
620 | */ |
---|
621 | static int |
---|
622 | set_assert(int *i) |
---|
623 | { |
---|
624 | if ((*i != 1) && (*i != 0)) |
---|
625 | return EINVAL; |
---|
626 | |
---|
627 | pim_assert = *i; |
---|
628 | |
---|
629 | return 0; |
---|
630 | } |
---|
631 | |
---|
632 | /* |
---|
633 | * Get PIM assert processing global |
---|
634 | */ |
---|
635 | static int |
---|
636 | get_assert(struct mbuf *m) |
---|
637 | { |
---|
638 | int *i; |
---|
639 | |
---|
640 | i = mtod(m, int *); |
---|
641 | |
---|
642 | *i = pim_assert; |
---|
643 | |
---|
644 | return 0; |
---|
645 | } |
---|
646 | |
---|
647 | /* |
---|
648 | * Add a vif to the vif table |
---|
649 | */ |
---|
650 | static int |
---|
651 | add_vif(struct vifctl *vifcp) |
---|
652 | { |
---|
653 | register struct vif *vifp = viftable + vifcp->vifc_vifi; |
---|
654 | static struct sockaddr_in sin = {sizeof sin, AF_INET}; |
---|
655 | struct ifaddr *ifa; |
---|
656 | struct ifnet *ifp; |
---|
657 | struct ifreq ifr; |
---|
658 | int error, s; |
---|
659 | struct tbf *v_tbf = tbftable + vifcp->vifc_vifi; |
---|
660 | |
---|
661 | if (vifcp->vifc_vifi >= MAXVIFS) return EINVAL; |
---|
662 | if (vifp->v_lcl_addr.s_addr != 0) return EADDRINUSE; |
---|
663 | |
---|
664 | /* Find the interface with an address in AF_INET family */ |
---|
665 | sin.sin_addr = vifcp->vifc_lcl_addr; |
---|
666 | ifa = ifa_ifwithaddr((struct sockaddr *)&sin); |
---|
667 | if (ifa == 0) return EADDRNOTAVAIL; |
---|
668 | ifp = ifa->ifa_ifp; |
---|
669 | |
---|
670 | if (vifcp->vifc_flags & VIFF_TUNNEL) { |
---|
671 | if ((vifcp->vifc_flags & VIFF_SRCRT) == 0) { |
---|
672 | /* |
---|
673 | * An encapsulating tunnel is wanted. Tell ipip_input() to |
---|
674 | * start paying attention to encapsulated packets. |
---|
675 | */ |
---|
676 | if (have_encap_tunnel == 0) { |
---|
677 | have_encap_tunnel = 1; |
---|
678 | for (s = 0; s < MAXVIFS; ++s) { |
---|
679 | multicast_decap_if[s].if_name = "mdecap"; |
---|
680 | multicast_decap_if[s].if_unit = s; |
---|
681 | } |
---|
682 | } |
---|
683 | /* |
---|
684 | * Set interface to fake encapsulator interface |
---|
685 | */ |
---|
686 | ifp = &multicast_decap_if[vifcp->vifc_vifi]; |
---|
687 | /* |
---|
688 | * Prepare cached route entry |
---|
689 | */ |
---|
690 | bzero(&vifp->v_route, sizeof(vifp->v_route)); |
---|
691 | } else { |
---|
692 | log(LOG_ERR, "source routed tunnels not supported\n"); |
---|
693 | return EOPNOTSUPP; |
---|
694 | } |
---|
695 | } else { |
---|
696 | /* Make sure the interface supports multicast */ |
---|
697 | if ((ifp->if_flags & IFF_MULTICAST) == 0) |
---|
698 | return EOPNOTSUPP; |
---|
699 | |
---|
700 | /* Enable promiscuous reception of all IP multicasts from the if */ |
---|
701 | ((struct sockaddr_in *)&(ifr.ifr_addr))->sin_family = AF_INET; |
---|
702 | ((struct sockaddr_in *)&(ifr.ifr_addr))->sin_addr.s_addr = INADDR_ANY; |
---|
703 | s = splnet(); |
---|
704 | error = (*ifp->if_ioctl)(ifp, SIOCADDMULTI, (caddr_t)&ifr); |
---|
705 | splx(s); |
---|
706 | if (error) |
---|
707 | return error; |
---|
708 | } |
---|
709 | |
---|
710 | s = splnet(); |
---|
711 | /* define parameters for the tbf structure */ |
---|
712 | vifp->v_tbf = v_tbf; |
---|
713 | GET_TIME(vifp->v_tbf->tbf_last_pkt_t); |
---|
714 | vifp->v_tbf->tbf_n_tok = 0; |
---|
715 | vifp->v_tbf->tbf_q_len = 0; |
---|
716 | vifp->v_tbf->tbf_max_q_len = MAXQSIZE; |
---|
717 | vifp->v_tbf->tbf_q = vifp->v_tbf->tbf_t = NULL; |
---|
718 | |
---|
719 | vifp->v_flags = vifcp->vifc_flags; |
---|
720 | vifp->v_threshold = vifcp->vifc_threshold; |
---|
721 | vifp->v_lcl_addr = vifcp->vifc_lcl_addr; |
---|
722 | vifp->v_rmt_addr = vifcp->vifc_rmt_addr; |
---|
723 | vifp->v_ifp = ifp; |
---|
724 | /* scaling up here allows division by 1024 in critical code */ |
---|
725 | vifp->v_rate_limit= vifcp->vifc_rate_limit * 1024 / 1000; |
---|
726 | vifp->v_rsvp_on = 0; |
---|
727 | vifp->v_rsvpd = NULL; |
---|
728 | /* initialize per vif pkt counters */ |
---|
729 | vifp->v_pkt_in = 0; |
---|
730 | vifp->v_pkt_out = 0; |
---|
731 | vifp->v_bytes_in = 0; |
---|
732 | vifp->v_bytes_out = 0; |
---|
733 | splx(s); |
---|
734 | |
---|
735 | /* Adjust numvifs up if the vifi is higher than numvifs */ |
---|
736 | if (numvifs <= vifcp->vifc_vifi) numvifs = vifcp->vifc_vifi + 1; |
---|
737 | |
---|
738 | if (mrtdebug) |
---|
739 | log(LOG_DEBUG, "add_vif #%d, lcladdr %x, %s %x, thresh %x, rate %d\n", |
---|
740 | vifcp->vifc_vifi, |
---|
741 | ntohl(vifcp->vifc_lcl_addr.s_addr), |
---|
742 | (vifcp->vifc_flags & VIFF_TUNNEL) ? "rmtaddr" : "mask", |
---|
743 | ntohl(vifcp->vifc_rmt_addr.s_addr), |
---|
744 | vifcp->vifc_threshold, |
---|
745 | vifcp->vifc_rate_limit); |
---|
746 | |
---|
747 | return 0; |
---|
748 | } |
---|
749 | |
---|
750 | /* |
---|
751 | * Delete a vif from the vif table |
---|
752 | */ |
---|
753 | static int |
---|
754 | del_vif(vifi_t *vifip) |
---|
755 | { |
---|
756 | register struct vif *vifp = viftable + *vifip; |
---|
757 | register vifi_t vifi; |
---|
758 | register struct mbuf *m; |
---|
759 | struct ifnet *ifp; |
---|
760 | struct ifreq ifr; |
---|
761 | int s; |
---|
762 | |
---|
763 | if (*vifip >= numvifs) return EINVAL; |
---|
764 | if (vifp->v_lcl_addr.s_addr == 0) return EADDRNOTAVAIL; |
---|
765 | |
---|
766 | s = splnet(); |
---|
767 | |
---|
768 | if (!(vifp->v_flags & VIFF_TUNNEL)) { |
---|
769 | ((struct sockaddr_in *)&(ifr.ifr_addr))->sin_family = AF_INET; |
---|
770 | ((struct sockaddr_in *)&(ifr.ifr_addr))->sin_addr.s_addr = INADDR_ANY; |
---|
771 | ifp = vifp->v_ifp; |
---|
772 | (*ifp->if_ioctl)(ifp, SIOCDELMULTI, (caddr_t)&ifr); |
---|
773 | } |
---|
774 | |
---|
775 | if (vifp == last_encap_vif) { |
---|
776 | last_encap_vif = 0; |
---|
777 | last_encap_src = 0; |
---|
778 | } |
---|
779 | |
---|
780 | /* |
---|
781 | * Free packets queued at the interface |
---|
782 | */ |
---|
783 | while (vifp->v_tbf->tbf_q) { |
---|
784 | m = vifp->v_tbf->tbf_q; |
---|
785 | vifp->v_tbf->tbf_q = m->m_act; |
---|
786 | m_freem(m); |
---|
787 | } |
---|
788 | |
---|
789 | bzero((caddr_t)vifp->v_tbf, sizeof(*(vifp->v_tbf))); |
---|
790 | bzero((caddr_t)vifp, sizeof (*vifp)); |
---|
791 | |
---|
792 | /* Adjust numvifs down */ |
---|
793 | for (vifi = numvifs; vifi > 0; vifi--) |
---|
794 | if (viftable[vifi-1].v_lcl_addr.s_addr != 0) break; |
---|
795 | numvifs = vifi; |
---|
796 | |
---|
797 | splx(s); |
---|
798 | |
---|
799 | if (mrtdebug) |
---|
800 | log(LOG_DEBUG, "del_vif %d, numvifs %d\n", *vifip, numvifs); |
---|
801 | |
---|
802 | return 0; |
---|
803 | } |
---|
804 | |
---|
805 | /* |
---|
806 | * Add an mfc entry |
---|
807 | */ |
---|
808 | static int |
---|
809 | add_mfc(struct mfcctl *mfccp) |
---|
810 | { |
---|
811 | struct mfc *rt; |
---|
812 | register struct mbuf *mb_rt; |
---|
813 | u_long hash; |
---|
814 | struct mbuf *mb_ntry; |
---|
815 | struct rtdetq *rte; |
---|
816 | register u_short nstl; |
---|
817 | int s; |
---|
818 | int i; |
---|
819 | |
---|
820 | MFCFIND(mfccp->mfcc_origin.s_addr, mfccp->mfcc_mcastgrp.s_addr, rt); |
---|
821 | |
---|
822 | /* If an entry already exists, just update the fields */ |
---|
823 | if (rt) { |
---|
824 | if (mrtdebug & DEBUG_MFC) |
---|
825 | log(LOG_DEBUG,"add_mfc update o %x g %x p %x\n", |
---|
826 | ntohl(mfccp->mfcc_origin.s_addr), |
---|
827 | ntohl(mfccp->mfcc_mcastgrp.s_addr), |
---|
828 | mfccp->mfcc_parent); |
---|
829 | |
---|
830 | s = splnet(); |
---|
831 | rt->mfc_parent = mfccp->mfcc_parent; |
---|
832 | for (i = 0; i < numvifs; i++) |
---|
833 | rt->mfc_ttls[i] = mfccp->mfcc_ttls[i]; |
---|
834 | splx(s); |
---|
835 | return 0; |
---|
836 | } |
---|
837 | |
---|
838 | /* |
---|
839 | * Find the entry for which the upcall was made and update |
---|
840 | */ |
---|
841 | s = splnet(); |
---|
842 | hash = MFCHASH(mfccp->mfcc_origin.s_addr, mfccp->mfcc_mcastgrp.s_addr); |
---|
843 | for (mb_rt = mfctable[hash], nstl = 0; mb_rt; mb_rt = mb_rt->m_next) { |
---|
844 | |
---|
845 | rt = mtod(mb_rt, struct mfc *); |
---|
846 | if ((rt->mfc_origin.s_addr == mfccp->mfcc_origin.s_addr) && |
---|
847 | (rt->mfc_mcastgrp.s_addr == mfccp->mfcc_mcastgrp.s_addr) && |
---|
848 | (mb_rt->m_act != NULL)) { |
---|
849 | |
---|
850 | if (nstl++) |
---|
851 | log(LOG_ERR, "add_mfc %s o %x g %x p %x dbx %x\n", |
---|
852 | "multiple kernel entries", |
---|
853 | ntohl(mfccp->mfcc_origin.s_addr), |
---|
854 | ntohl(mfccp->mfcc_mcastgrp.s_addr), |
---|
855 | mfccp->mfcc_parent, mb_rt->m_act); |
---|
856 | |
---|
857 | if (mrtdebug & DEBUG_MFC) |
---|
858 | log(LOG_DEBUG,"add_mfc o %x g %x p %x dbg %x\n", |
---|
859 | ntohl(mfccp->mfcc_origin.s_addr), |
---|
860 | ntohl(mfccp->mfcc_mcastgrp.s_addr), |
---|
861 | mfccp->mfcc_parent, mb_rt->m_act); |
---|
862 | |
---|
863 | rt->mfc_origin = mfccp->mfcc_origin; |
---|
864 | rt->mfc_mcastgrp = mfccp->mfcc_mcastgrp; |
---|
865 | rt->mfc_parent = mfccp->mfcc_parent; |
---|
866 | for (i = 0; i < numvifs; i++) |
---|
867 | rt->mfc_ttls[i] = mfccp->mfcc_ttls[i]; |
---|
868 | /* initialize pkt counters per src-grp */ |
---|
869 | rt->mfc_pkt_cnt = 0; |
---|
870 | rt->mfc_byte_cnt = 0; |
---|
871 | rt->mfc_wrong_if = 0; |
---|
872 | rt->mfc_last_assert.tv_sec = rt->mfc_last_assert.tv_usec = 0; |
---|
873 | |
---|
874 | rt->mfc_expire = 0; /* Don't clean this guy up */ |
---|
875 | nexpire[hash]--; |
---|
876 | |
---|
877 | /* free packets Qed at the end of this entry */ |
---|
878 | while (mb_rt->m_act) { |
---|
879 | mb_ntry = mb_rt->m_act; |
---|
880 | rte = mtod(mb_ntry, struct rtdetq *); |
---|
881 | /* #ifdef RSVP_ISI */ |
---|
882 | ip_mdq(rte->m, rte->ifp, rt, -1); |
---|
883 | /* #endif */ |
---|
884 | mb_rt->m_act = mb_ntry->m_act; |
---|
885 | m_freem(rte->m); |
---|
886 | #ifdef UPCALL_TIMING |
---|
887 | collate(&(rte->t)); |
---|
888 | #endif /* UPCALL_TIMING */ |
---|
889 | m_free(mb_ntry); |
---|
890 | } |
---|
891 | } |
---|
892 | } |
---|
893 | |
---|
894 | /* |
---|
895 | * It is possible that an entry is being inserted without an upcall |
---|
896 | */ |
---|
897 | if (nstl == 0) { |
---|
898 | if (mrtdebug & DEBUG_MFC) |
---|
899 | log(LOG_DEBUG,"add_mfc no upcall h %d o %x g %x p %x\n", |
---|
900 | hash, ntohl(mfccp->mfcc_origin.s_addr), |
---|
901 | ntohl(mfccp->mfcc_mcastgrp.s_addr), |
---|
902 | mfccp->mfcc_parent); |
---|
903 | |
---|
904 | for (mb_rt = mfctable[hash]; mb_rt; mb_rt = mb_rt->m_next) { |
---|
905 | |
---|
906 | rt = mtod(mb_rt, struct mfc *); |
---|
907 | if ((rt->mfc_origin.s_addr == mfccp->mfcc_origin.s_addr) && |
---|
908 | (rt->mfc_mcastgrp.s_addr == mfccp->mfcc_mcastgrp.s_addr)) { |
---|
909 | |
---|
910 | rt->mfc_origin = mfccp->mfcc_origin; |
---|
911 | rt->mfc_mcastgrp = mfccp->mfcc_mcastgrp; |
---|
912 | rt->mfc_parent = mfccp->mfcc_parent; |
---|
913 | for (i = 0; i < numvifs; i++) |
---|
914 | rt->mfc_ttls[i] = mfccp->mfcc_ttls[i]; |
---|
915 | /* initialize pkt counters per src-grp */ |
---|
916 | rt->mfc_pkt_cnt = 0; |
---|
917 | rt->mfc_byte_cnt = 0; |
---|
918 | rt->mfc_wrong_if = 0; |
---|
919 | rt->mfc_last_assert.tv_sec = rt->mfc_last_assert.tv_usec = 0; |
---|
920 | if (rt->mfc_expire) |
---|
921 | nexpire[hash]--; |
---|
922 | rt->mfc_expire = 0; |
---|
923 | } |
---|
924 | } |
---|
925 | if (mb_rt == NULL) { |
---|
926 | /* no upcall, so make a new entry */ |
---|
927 | MGET(mb_rt, M_DONTWAIT, MT_MRTABLE); |
---|
928 | if (mb_rt == NULL) { |
---|
929 | splx(s); |
---|
930 | return ENOBUFS; |
---|
931 | } |
---|
932 | |
---|
933 | rt = mtod(mb_rt, struct mfc *); |
---|
934 | |
---|
935 | /* insert new entry at head of hash chain */ |
---|
936 | rt->mfc_origin = mfccp->mfcc_origin; |
---|
937 | rt->mfc_mcastgrp = mfccp->mfcc_mcastgrp; |
---|
938 | rt->mfc_parent = mfccp->mfcc_parent; |
---|
939 | for (i = 0; i < numvifs; i++) |
---|
940 | rt->mfc_ttls[i] = mfccp->mfcc_ttls[i]; |
---|
941 | /* initialize pkt counters per src-grp */ |
---|
942 | rt->mfc_pkt_cnt = 0; |
---|
943 | rt->mfc_byte_cnt = 0; |
---|
944 | rt->mfc_wrong_if = 0; |
---|
945 | rt->mfc_last_assert.tv_sec = rt->mfc_last_assert.tv_usec = 0; |
---|
946 | rt->mfc_expire = 0; |
---|
947 | |
---|
948 | /* link into table */ |
---|
949 | mb_rt->m_next = mfctable[hash]; |
---|
950 | mfctable[hash] = mb_rt; |
---|
951 | mb_rt->m_act = NULL; |
---|
952 | } |
---|
953 | } |
---|
954 | splx(s); |
---|
955 | return 0; |
---|
956 | } |
---|
957 | |
---|
958 | #ifdef UPCALL_TIMING |
---|
959 | /* |
---|
960 | * collect delay statistics on the upcalls |
---|
961 | */ |
---|
962 | static void collate(struct timeval *t) |
---|
963 | { |
---|
964 | register u_long d; |
---|
965 | register struct timeval tp; |
---|
966 | register u_long delta; |
---|
967 | |
---|
968 | GET_TIME(tp); |
---|
969 | |
---|
970 | if (TV_LT(*t, tp)) |
---|
971 | { |
---|
972 | TV_DELTA(tp, *t, delta); |
---|
973 | |
---|
974 | d = delta >> 10; |
---|
975 | if (d > 50) |
---|
976 | d = 50; |
---|
977 | |
---|
978 | ++upcall_data[d]; |
---|
979 | } |
---|
980 | } |
---|
981 | #endif /* UPCALL_TIMING */ |
---|
982 | |
---|
983 | /* |
---|
984 | * Delete an mfc entry |
---|
985 | */ |
---|
986 | static int |
---|
987 | del_mfc(struct mfcctl *mfccp) |
---|
988 | { |
---|
989 | struct in_addr origin; |
---|
990 | struct in_addr mcastgrp; |
---|
991 | struct mfc *rt; |
---|
992 | struct mbuf *mb_rt; |
---|
993 | struct mbuf **nptr; |
---|
994 | u_long hash; |
---|
995 | int s; |
---|
996 | |
---|
997 | origin = mfccp->mfcc_origin; |
---|
998 | mcastgrp = mfccp->mfcc_mcastgrp; |
---|
999 | hash = MFCHASH(origin.s_addr, mcastgrp.s_addr); |
---|
1000 | |
---|
1001 | if (mrtdebug & DEBUG_MFC) |
---|
1002 | log(LOG_DEBUG,"del_mfc orig %x mcastgrp %x\n", |
---|
1003 | ntohl(origin.s_addr), ntohl(mcastgrp.s_addr)); |
---|
1004 | |
---|
1005 | s = splnet(); |
---|
1006 | |
---|
1007 | nptr = &mfctable[hash]; |
---|
1008 | while ((mb_rt = *nptr) != NULL) { |
---|
1009 | rt = mtod(mb_rt, struct mfc *); |
---|
1010 | if (origin.s_addr == rt->mfc_origin.s_addr && |
---|
1011 | mcastgrp.s_addr == rt->mfc_mcastgrp.s_addr && |
---|
1012 | mb_rt->m_act == NULL) |
---|
1013 | break; |
---|
1014 | |
---|
1015 | nptr = &mb_rt->m_next; |
---|
1016 | } |
---|
1017 | if (mb_rt == NULL) { |
---|
1018 | splx(s); |
---|
1019 | return EADDRNOTAVAIL; |
---|
1020 | } |
---|
1021 | |
---|
1022 | MFREE(mb_rt, *nptr); |
---|
1023 | |
---|
1024 | splx(s); |
---|
1025 | |
---|
1026 | return 0; |
---|
1027 | } |
---|
1028 | |
---|
1029 | /* |
---|
1030 | * Send a message to mrouted on the multicast routing socket |
---|
1031 | */ |
---|
1032 | static int |
---|
1033 | socket_send(struct socket *s, struct mbuf *mm, struct sockaddr_in *src) |
---|
1034 | { |
---|
1035 | if (s) { |
---|
1036 | if (sbappendaddr(&s->so_rcv, |
---|
1037 | (struct sockaddr *)src, |
---|
1038 | mm, (struct mbuf *)0) != 0) { |
---|
1039 | sorwakeup(s); |
---|
1040 | return 0; |
---|
1041 | } |
---|
1042 | } |
---|
1043 | m_freem(mm); |
---|
1044 | return -1; |
---|
1045 | } |
---|
1046 | |
---|
1047 | /* |
---|
1048 | * IP multicast forwarding function. This function assumes that the packet |
---|
1049 | * pointed to by "ip" has arrived on (or is about to be sent to) the interface |
---|
1050 | * pointed to by "ifp", and the packet is to be relayed to other networks |
---|
1051 | * that have members of the packet's destination IP multicast group. |
---|
1052 | * |
---|
1053 | * The packet is returned unscathed to the caller, unless it is |
---|
1054 | * erroneous, in which case a non-zero return value tells the caller to |
---|
1055 | * discard it. |
---|
1056 | */ |
---|
1057 | |
---|
1058 | #define IP_HDR_LEN 20 /* # bytes of fixed IP header (excluding options) */ |
---|
1059 | #define TUNNEL_LEN 12 /* # bytes of IP option for tunnel encapsulation */ |
---|
1060 | |
---|
1061 | static int |
---|
1062 | X_ip_mforward(struct ip *ip, struct ifnet *ifp, struct mbuf *m, |
---|
1063 | struct ip_moptions *imo) |
---|
1064 | { |
---|
1065 | register struct mfc *rt; |
---|
1066 | register u_char *ipoptions; |
---|
1067 | static struct sockaddr_in k_igmpsrc = { sizeof k_igmpsrc, AF_INET }; |
---|
1068 | static int srctun = 0; |
---|
1069 | register struct mbuf *mm; |
---|
1070 | int s; |
---|
1071 | vifi_t vifi; |
---|
1072 | struct vif *vifp; |
---|
1073 | |
---|
1074 | if (mrtdebug & DEBUG_FORWARD) |
---|
1075 | log(LOG_DEBUG, "ip_mforward: src %x, dst %x, ifp %x\n", |
---|
1076 | ntohl(ip->ip_src.s_addr), ntohl(ip->ip_dst.s_addr), ifp); |
---|
1077 | |
---|
1078 | if (ip->ip_hl < (IP_HDR_LEN + TUNNEL_LEN) >> 2 || |
---|
1079 | (ipoptions = (u_char *)(ip + 1))[1] != IPOPT_LSRR ) { |
---|
1080 | /* |
---|
1081 | * Packet arrived via a physical interface or |
---|
1082 | * an encapsulated tunnel. |
---|
1083 | */ |
---|
1084 | } else { |
---|
1085 | /* |
---|
1086 | * Packet arrived through a source-route tunnel. |
---|
1087 | * Source-route tunnels are no longer supported. |
---|
1088 | */ |
---|
1089 | if ((srctun++ % 1000) == 0) |
---|
1090 | log(LOG_ERR, "ip_mforward: received source-routed packet from %x\n", |
---|
1091 | ntohl(ip->ip_src.s_addr)); |
---|
1092 | |
---|
1093 | return 1; |
---|
1094 | } |
---|
1095 | |
---|
1096 | if ((imo) && ((vifi = imo->imo_multicast_vif) < numvifs)) { |
---|
1097 | if (ip->ip_ttl < 255) |
---|
1098 | ip->ip_ttl++; /* compensate for -1 in *_send routines */ |
---|
1099 | if (rsvpdebug && ip->ip_p == IPPROTO_RSVP) { |
---|
1100 | vifp = viftable + vifi; |
---|
1101 | printf("Sending IPPROTO_RSVP from %lx to %lx on vif %d (%s%s%d)\n", |
---|
1102 | ntohl(ip->ip_src.s_addr), ntohl(ip->ip_dst.s_addr), vifi, |
---|
1103 | (vifp->v_flags & VIFF_TUNNEL) ? "tunnel on " : "", |
---|
1104 | vifp->v_ifp->if_name, vifp->v_ifp->if_unit); |
---|
1105 | } |
---|
1106 | return (ip_mdq(m, ifp, NULL, vifi)); |
---|
1107 | } |
---|
1108 | if (rsvpdebug && ip->ip_p == IPPROTO_RSVP) { |
---|
1109 | printf("Warning: IPPROTO_RSVP from %lx to %lx without vif option\n", |
---|
1110 | ntohl(ip->ip_src.s_addr), ntohl(ip->ip_dst.s_addr)); |
---|
1111 | if(!imo) |
---|
1112 | printf("In fact, no options were specified at all\n"); |
---|
1113 | } |
---|
1114 | |
---|
1115 | /* |
---|
1116 | * Don't forward a packet with time-to-live of zero or one, |
---|
1117 | * or a packet destined to a local-only group. |
---|
1118 | */ |
---|
1119 | if (ip->ip_ttl <= 1 || |
---|
1120 | ntohl(ip->ip_dst.s_addr) <= INADDR_MAX_LOCAL_GROUP) |
---|
1121 | return 0; |
---|
1122 | |
---|
1123 | /* |
---|
1124 | * Determine forwarding vifs from the forwarding cache table |
---|
1125 | */ |
---|
1126 | s = splnet(); |
---|
1127 | MFCFIND(ip->ip_src.s_addr, ip->ip_dst.s_addr, rt); |
---|
1128 | |
---|
1129 | /* Entry exists, so forward if necessary */ |
---|
1130 | if (rt != NULL) { |
---|
1131 | splx(s); |
---|
1132 | return (ip_mdq(m, ifp, rt, -1)); |
---|
1133 | } else { |
---|
1134 | /* |
---|
1135 | * If we don't have a route for packet's origin, |
---|
1136 | * Make a copy of the packet & |
---|
1137 | * send message to routing daemon |
---|
1138 | */ |
---|
1139 | |
---|
1140 | register struct mbuf *mb_rt; |
---|
1141 | register struct mbuf *mb_ntry; |
---|
1142 | register struct mbuf *mb0; |
---|
1143 | register struct rtdetq *rte; |
---|
1144 | register struct mbuf *rte_m; |
---|
1145 | register u_long hash; |
---|
1146 | register int npkts; |
---|
1147 | int hlen = ip->ip_hl << 2; |
---|
1148 | #ifdef UPCALL_TIMING |
---|
1149 | struct timeval tp; |
---|
1150 | |
---|
1151 | GET_TIME(tp); |
---|
1152 | #endif |
---|
1153 | |
---|
1154 | mrtstat.mrts_no_route++; |
---|
1155 | if (mrtdebug & (DEBUG_FORWARD | DEBUG_MFC)) |
---|
1156 | log(LOG_DEBUG, "ip_mforward: no rte s %x g %x\n", |
---|
1157 | ntohl(ip->ip_src.s_addr), |
---|
1158 | ntohl(ip->ip_dst.s_addr)); |
---|
1159 | |
---|
1160 | /* |
---|
1161 | * Allocate mbufs early so that we don't do extra work if we are |
---|
1162 | * just going to fail anyway. Make sure to pullup the header so |
---|
1163 | * that other people can't step on it. |
---|
1164 | */ |
---|
1165 | MGET(mb_ntry, M_DONTWAIT, MT_DATA); |
---|
1166 | if (mb_ntry == NULL) { |
---|
1167 | splx(s); |
---|
1168 | return ENOBUFS; |
---|
1169 | } |
---|
1170 | mb0 = m_copy(m, 0, M_COPYALL); |
---|
1171 | if (mb0 && (M_HASCL(mb0) || mb0->m_len < hlen)) |
---|
1172 | mb0 = m_pullup(mb0, hlen); |
---|
1173 | if (mb0 == NULL) { |
---|
1174 | m_free(mb_ntry); |
---|
1175 | splx(s); |
---|
1176 | return ENOBUFS; |
---|
1177 | } |
---|
1178 | |
---|
1179 | /* is there an upcall waiting for this packet? */ |
---|
1180 | hash = MFCHASH(ip->ip_src.s_addr, ip->ip_dst.s_addr); |
---|
1181 | for (mb_rt = mfctable[hash]; mb_rt; mb_rt = mb_rt->m_next) { |
---|
1182 | rt = mtod(mb_rt, struct mfc *); |
---|
1183 | if ((ip->ip_src.s_addr == rt->mfc_origin.s_addr) && |
---|
1184 | (ip->ip_dst.s_addr == rt->mfc_mcastgrp.s_addr) && |
---|
1185 | (mb_rt->m_act != NULL)) |
---|
1186 | break; |
---|
1187 | } |
---|
1188 | |
---|
1189 | if (mb_rt == NULL) { |
---|
1190 | int i; |
---|
1191 | struct igmpmsg *im; |
---|
1192 | |
---|
1193 | /* no upcall, so make a new entry */ |
---|
1194 | MGET(mb_rt, M_DONTWAIT, MT_MRTABLE); |
---|
1195 | if (mb_rt == NULL) { |
---|
1196 | m_free(mb_ntry); |
---|
1197 | m_freem(mb0); |
---|
1198 | splx(s); |
---|
1199 | return ENOBUFS; |
---|
1200 | } |
---|
1201 | /* Make a copy of the header to send to the user level process */ |
---|
1202 | mm = m_copy(mb0, 0, hlen); |
---|
1203 | if (mm == NULL) { |
---|
1204 | m_free(mb_ntry); |
---|
1205 | m_freem(mb0); |
---|
1206 | m_free(mb_rt); |
---|
1207 | splx(s); |
---|
1208 | return ENOBUFS; |
---|
1209 | } |
---|
1210 | |
---|
1211 | /* |
---|
1212 | * Send message to routing daemon to install |
---|
1213 | * a route into the kernel table |
---|
1214 | */ |
---|
1215 | k_igmpsrc.sin_addr = ip->ip_src; |
---|
1216 | |
---|
1217 | im = mtod(mm, struct igmpmsg *); |
---|
1218 | im->im_msgtype = IGMPMSG_NOCACHE; |
---|
1219 | im->im_mbz = 0; |
---|
1220 | |
---|
1221 | mrtstat.mrts_upcalls++; |
---|
1222 | |
---|
1223 | if (socket_send(ip_mrouter, mm, &k_igmpsrc) < 0) { |
---|
1224 | log(LOG_WARNING, "ip_mforward: ip_mrouter socket queue full\n"); |
---|
1225 | ++mrtstat.mrts_upq_sockfull; |
---|
1226 | m_free(mb_ntry); |
---|
1227 | m_freem(mb0); |
---|
1228 | m_free(mb_rt); |
---|
1229 | splx(s); |
---|
1230 | return ENOBUFS; |
---|
1231 | } |
---|
1232 | |
---|
1233 | rt = mtod(mb_rt, struct mfc *); |
---|
1234 | |
---|
1235 | /* insert new entry at head of hash chain */ |
---|
1236 | rt->mfc_origin.s_addr = ip->ip_src.s_addr; |
---|
1237 | rt->mfc_mcastgrp.s_addr = ip->ip_dst.s_addr; |
---|
1238 | rt->mfc_expire = UPCALL_EXPIRE; |
---|
1239 | nexpire[hash]++; |
---|
1240 | for (i = 0; i < numvifs; i++) |
---|
1241 | rt->mfc_ttls[i] = 0; |
---|
1242 | rt->mfc_parent = -1; |
---|
1243 | |
---|
1244 | /* link into table */ |
---|
1245 | mb_rt->m_next = mfctable[hash]; |
---|
1246 | mfctable[hash] = mb_rt; |
---|
1247 | mb_rt->m_act = NULL; |
---|
1248 | |
---|
1249 | rte_m = mb_rt; |
---|
1250 | } else { |
---|
1251 | /* determine if q has overflowed */ |
---|
1252 | for (rte_m = mb_rt, npkts = 0; rte_m->m_act; rte_m = rte_m->m_act) |
---|
1253 | npkts++; |
---|
1254 | |
---|
1255 | if (npkts > MAX_UPQ) { |
---|
1256 | mrtstat.mrts_upq_ovflw++; |
---|
1257 | m_free(mb_ntry); |
---|
1258 | m_freem(mb0); |
---|
1259 | splx(s); |
---|
1260 | return 0; |
---|
1261 | } |
---|
1262 | } |
---|
1263 | |
---|
1264 | mb_ntry->m_act = NULL; |
---|
1265 | rte = mtod(mb_ntry, struct rtdetq *); |
---|
1266 | |
---|
1267 | rte->m = mb0; |
---|
1268 | rte->ifp = ifp; |
---|
1269 | #ifdef UPCALL_TIMING |
---|
1270 | rte->t = tp; |
---|
1271 | #endif |
---|
1272 | |
---|
1273 | /* Add this entry to the end of the queue */ |
---|
1274 | rte_m->m_act = mb_ntry; |
---|
1275 | |
---|
1276 | splx(s); |
---|
1277 | |
---|
1278 | return 0; |
---|
1279 | } |
---|
1280 | } |
---|
1281 | |
---|
1282 | #ifndef MROUTE_LKM |
---|
1283 | int (*ip_mforward)(struct ip *, struct ifnet *, struct mbuf *, |
---|
1284 | struct ip_moptions *) = X_ip_mforward; |
---|
1285 | #endif |
---|
1286 | |
---|
1287 | /* |
---|
1288 | * Clean up the cache entry if upcall is not serviced |
---|
1289 | */ |
---|
1290 | static void |
---|
1291 | expire_upcalls(void *unused) |
---|
1292 | { |
---|
1293 | struct mbuf *mb_rt, *m, **nptr; |
---|
1294 | struct rtdetq *rte; |
---|
1295 | struct mfc *mfc; |
---|
1296 | int i; |
---|
1297 | int s; |
---|
1298 | |
---|
1299 | s = splnet(); |
---|
1300 | for (i = 0; i < MFCTBLSIZ; i++) { |
---|
1301 | if (nexpire[i] == 0) |
---|
1302 | continue; |
---|
1303 | nptr = &mfctable[i]; |
---|
1304 | for (mb_rt = *nptr; mb_rt != NULL; mb_rt = *nptr) { |
---|
1305 | mfc = mtod(mb_rt, struct mfc *); |
---|
1306 | |
---|
1307 | /* |
---|
1308 | * Skip real cache entries |
---|
1309 | * Make sure it wasn't marked to not expire (shouldn't happen) |
---|
1310 | * If it expires now |
---|
1311 | */ |
---|
1312 | if (mb_rt->m_act != NULL && |
---|
1313 | mfc->mfc_expire != 0 && |
---|
1314 | --mfc->mfc_expire == 0) { |
---|
1315 | if (mrtdebug & DEBUG_EXPIRE) |
---|
1316 | log(LOG_DEBUG, "expire_upcalls: expiring (%x %x)\n", |
---|
1317 | ntohl(mfc->mfc_origin.s_addr), |
---|
1318 | ntohl(mfc->mfc_mcastgrp.s_addr)); |
---|
1319 | /* |
---|
1320 | * drop all the packets |
---|
1321 | * free the mbuf with the pkt, if, timing info |
---|
1322 | */ |
---|
1323 | while (mb_rt->m_act) { |
---|
1324 | m = mb_rt->m_act; |
---|
1325 | mb_rt->m_act = m->m_act; |
---|
1326 | |
---|
1327 | rte = mtod(m, struct rtdetq *); |
---|
1328 | m_freem(rte->m); |
---|
1329 | m_free(m); |
---|
1330 | } |
---|
1331 | ++mrtstat.mrts_cache_cleanups; |
---|
1332 | nexpire[i]--; |
---|
1333 | |
---|
1334 | MFREE(mb_rt, *nptr); |
---|
1335 | } else { |
---|
1336 | nptr = &mb_rt->m_next; |
---|
1337 | } |
---|
1338 | } |
---|
1339 | } |
---|
1340 | splx(s); |
---|
1341 | timeout(expire_upcalls, (caddr_t)NULL, EXPIRE_TIMEOUT); |
---|
1342 | } |
---|
1343 | |
---|
1344 | /* |
---|
1345 | * Packet forwarding routine once entry in the cache is made |
---|
1346 | */ |
---|
1347 | static int |
---|
1348 | ip_mdq(struct mbuf *m, struct ifnet *ifp, struct mfc *rt, |
---|
1349 | vifi_t xmt_vif) |
---|
1350 | { |
---|
1351 | register struct ip *ip = mtod(m, struct ip *); |
---|
1352 | register vifi_t vifi; |
---|
1353 | register struct vif *vifp; |
---|
1354 | register int plen = ntohs(ip->ip_len); |
---|
1355 | |
---|
1356 | /* |
---|
1357 | * Macro to send packet on vif. Since RSVP packets don't get counted on |
---|
1358 | * input, they shouldn't get counted on output, so statistics keeping is |
---|
1359 | * seperate. |
---|
1360 | */ |
---|
1361 | #define MC_SEND(ip,vifp,m) { \ |
---|
1362 | if ((vifp)->v_flags & VIFF_TUNNEL) \ |
---|
1363 | encap_send((ip), (vifp), (m)); \ |
---|
1364 | else \ |
---|
1365 | phyint_send((ip), (vifp), (m)); \ |
---|
1366 | } |
---|
1367 | |
---|
1368 | /* |
---|
1369 | * If xmt_vif is not -1, send on only the requested vif. |
---|
1370 | * |
---|
1371 | * (since vifi_t is u_short, -1 becomes MAXUSHORT, which > numvifs.) |
---|
1372 | */ |
---|
1373 | if (xmt_vif < numvifs) { |
---|
1374 | MC_SEND(ip, viftable + xmt_vif, m); |
---|
1375 | return 1; |
---|
1376 | } |
---|
1377 | |
---|
1378 | /* |
---|
1379 | * Don't forward if it didn't arrive from the parent vif for its origin. |
---|
1380 | */ |
---|
1381 | vifi = rt->mfc_parent; |
---|
1382 | if ((vifi >= numvifs) || (viftable[vifi].v_ifp != ifp)) { |
---|
1383 | /* came in the wrong interface */ |
---|
1384 | if (mrtdebug & DEBUG_FORWARD) |
---|
1385 | log(LOG_DEBUG, "wrong if: ifp %x vifi %d vififp %x\n", |
---|
1386 | ifp, vifi, viftable[vifi].v_ifp); |
---|
1387 | ++mrtstat.mrts_wrong_if; |
---|
1388 | ++rt->mfc_wrong_if; |
---|
1389 | /* |
---|
1390 | * If we are doing PIM assert processing, and we are forwarding |
---|
1391 | * packets on this interface, and it is a broadcast medium |
---|
1392 | * interface (and not a tunnel), send a message to the routing daemon. |
---|
1393 | */ |
---|
1394 | if (pim_assert && rt->mfc_ttls[vifi] && |
---|
1395 | (ifp->if_flags & IFF_BROADCAST) && |
---|
1396 | !(viftable[vifi].v_flags & VIFF_TUNNEL)) { |
---|
1397 | struct sockaddr_in k_igmpsrc; |
---|
1398 | struct mbuf *mm; |
---|
1399 | struct igmpmsg *im; |
---|
1400 | int hlen = ip->ip_hl << 2; |
---|
1401 | struct timeval now; |
---|
1402 | register u_long delta; |
---|
1403 | |
---|
1404 | GET_TIME(now); |
---|
1405 | |
---|
1406 | TV_DELTA(rt->mfc_last_assert, now, delta); |
---|
1407 | |
---|
1408 | if (delta > ASSERT_MSG_TIME) { |
---|
1409 | mm = m_copy(m, 0, hlen); |
---|
1410 | if (mm && (M_HASCL(mm) || mm->m_len < hlen)) |
---|
1411 | mm = m_pullup(mm, hlen); |
---|
1412 | if (mm == NULL) { |
---|
1413 | return ENOBUFS; |
---|
1414 | } |
---|
1415 | |
---|
1416 | rt->mfc_last_assert = now; |
---|
1417 | |
---|
1418 | im = mtod(mm, struct igmpmsg *); |
---|
1419 | im->im_msgtype = IGMPMSG_WRONGVIF; |
---|
1420 | im->im_mbz = 0; |
---|
1421 | im->im_vif = vifi; |
---|
1422 | |
---|
1423 | k_igmpsrc.sin_addr = im->im_src; |
---|
1424 | |
---|
1425 | socket_send(ip_mrouter, mm, &k_igmpsrc); |
---|
1426 | } |
---|
1427 | } |
---|
1428 | return 0; |
---|
1429 | } |
---|
1430 | |
---|
1431 | /* If I sourced this packet, it counts as output, else it was input. */ |
---|
1432 | if (ip->ip_src.s_addr == viftable[vifi].v_lcl_addr.s_addr) { |
---|
1433 | viftable[vifi].v_pkt_out++; |
---|
1434 | viftable[vifi].v_bytes_out += plen; |
---|
1435 | } else { |
---|
1436 | viftable[vifi].v_pkt_in++; |
---|
1437 | viftable[vifi].v_bytes_in += plen; |
---|
1438 | } |
---|
1439 | rt->mfc_pkt_cnt++; |
---|
1440 | rt->mfc_byte_cnt += plen; |
---|
1441 | |
---|
1442 | /* |
---|
1443 | * For each vif, decide if a copy of the packet should be forwarded. |
---|
1444 | * Forward if: |
---|
1445 | * - the ttl exceeds the vif's threshold |
---|
1446 | * - there are group members downstream on interface |
---|
1447 | */ |
---|
1448 | for (vifp = viftable, vifi = 0; vifi < numvifs; vifp++, vifi++) |
---|
1449 | if ((rt->mfc_ttls[vifi] > 0) && |
---|
1450 | (ip->ip_ttl > rt->mfc_ttls[vifi])) { |
---|
1451 | vifp->v_pkt_out++; |
---|
1452 | vifp->v_bytes_out += plen; |
---|
1453 | MC_SEND(ip, vifp, m); |
---|
1454 | } |
---|
1455 | |
---|
1456 | return 0; |
---|
1457 | } |
---|
1458 | |
---|
1459 | /* |
---|
1460 | * check if a vif number is legal/ok. This is used by ip_output, to export |
---|
1461 | * numvifs there, |
---|
1462 | */ |
---|
1463 | static int |
---|
1464 | X_legal_vif_num(int vif) |
---|
1465 | { |
---|
1466 | if (vif >= 0 && vif < numvifs) |
---|
1467 | return(1); |
---|
1468 | else |
---|
1469 | return(0); |
---|
1470 | } |
---|
1471 | |
---|
1472 | #ifndef MROUTE_LKM |
---|
1473 | int (*legal_vif_num)(int) = X_legal_vif_num; |
---|
1474 | #endif |
---|
1475 | |
---|
1476 | /* |
---|
1477 | * Return the local address used by this vif |
---|
1478 | */ |
---|
1479 | static u_long |
---|
1480 | X_ip_mcast_src(int vifi) |
---|
1481 | { |
---|
1482 | if (vifi >= 0 && vifi < numvifs) |
---|
1483 | return viftable[vifi].v_lcl_addr.s_addr; |
---|
1484 | else |
---|
1485 | return INADDR_ANY; |
---|
1486 | } |
---|
1487 | |
---|
1488 | #ifndef MROUTE_LKM |
---|
1489 | u_long (*ip_mcast_src)(int) = X_ip_mcast_src; |
---|
1490 | #endif |
---|
1491 | |
---|
1492 | static void |
---|
1493 | phyint_send(struct ip *ip, struct vif *vifp, struct mbuf *m) |
---|
1494 | { |
---|
1495 | register struct mbuf *mb_copy; |
---|
1496 | register int hlen = ip->ip_hl << 2; |
---|
1497 | |
---|
1498 | /* |
---|
1499 | * Make a new reference to the packet; make sure that |
---|
1500 | * the IP header is actually copied, not just referenced, |
---|
1501 | * so that ip_output() only scribbles on the copy. |
---|
1502 | */ |
---|
1503 | mb_copy = m_copy(m, 0, M_COPYALL); |
---|
1504 | if (mb_copy && (M_HASCL(mb_copy) || mb_copy->m_len < hlen)) |
---|
1505 | mb_copy = m_pullup(mb_copy, hlen); |
---|
1506 | if (mb_copy == NULL) |
---|
1507 | return; |
---|
1508 | |
---|
1509 | if (vifp->v_rate_limit <= 0) |
---|
1510 | tbf_send_packet(vifp, mb_copy); |
---|
1511 | else |
---|
1512 | tbf_control(vifp, mb_copy, mtod(mb_copy, struct ip *), ip->ip_len); |
---|
1513 | } |
---|
1514 | |
---|
1515 | static void |
---|
1516 | encap_send(struct ip *ip, struct vif *vifp, struct mbuf *m) |
---|
1517 | { |
---|
1518 | register struct mbuf *mb_copy; |
---|
1519 | register struct ip *ip_copy; |
---|
1520 | register int i, len = ip->ip_len; |
---|
1521 | |
---|
1522 | /* |
---|
1523 | * copy the old packet & pullup it's IP header into the |
---|
1524 | * new mbuf so we can modify it. Try to fill the new |
---|
1525 | * mbuf since if we don't the ethernet driver will. |
---|
1526 | */ |
---|
1527 | MGETHDR(mb_copy, M_DONTWAIT, MT_HEADER); |
---|
1528 | if (mb_copy == NULL) |
---|
1529 | return; |
---|
1530 | mb_copy->m_data += max_linkhdr; |
---|
1531 | mb_copy->m_len = sizeof(multicast_encap_iphdr); |
---|
1532 | |
---|
1533 | if ((mb_copy->m_next = m_copy(m, 0, M_COPYALL)) == NULL) { |
---|
1534 | m_freem(mb_copy); |
---|
1535 | return; |
---|
1536 | } |
---|
1537 | i = MHLEN - M_LEADINGSPACE(mb_copy); |
---|
1538 | if (i > len) |
---|
1539 | i = len; |
---|
1540 | mb_copy = m_pullup(mb_copy, i); |
---|
1541 | if (mb_copy == NULL) |
---|
1542 | return; |
---|
1543 | mb_copy->m_pkthdr.len = len + sizeof(multicast_encap_iphdr); |
---|
1544 | |
---|
1545 | /* |
---|
1546 | * fill in the encapsulating IP header. |
---|
1547 | */ |
---|
1548 | ip_copy = mtod(mb_copy, struct ip *); |
---|
1549 | *ip_copy = multicast_encap_iphdr; |
---|
1550 | ip_copy->ip_id = htons(ip_id++); |
---|
1551 | ip_copy->ip_len += len; |
---|
1552 | ip_copy->ip_src = vifp->v_lcl_addr; |
---|
1553 | ip_copy->ip_dst = vifp->v_rmt_addr; |
---|
1554 | |
---|
1555 | /* |
---|
1556 | * turn the encapsulated IP header back into a valid one. |
---|
1557 | */ |
---|
1558 | ip = (struct ip *)((caddr_t)ip_copy + sizeof(multicast_encap_iphdr)); |
---|
1559 | --ip->ip_ttl; |
---|
1560 | HTONS(ip->ip_len); |
---|
1561 | HTONS(ip->ip_off); |
---|
1562 | ip->ip_sum = 0; |
---|
1563 | mb_copy->m_data += sizeof(multicast_encap_iphdr); |
---|
1564 | ip->ip_sum = in_cksum(mb_copy, ip->ip_hl << 2); |
---|
1565 | mb_copy->m_data -= sizeof(multicast_encap_iphdr); |
---|
1566 | |
---|
1567 | if (vifp->v_rate_limit <= 0) |
---|
1568 | tbf_send_packet(vifp, mb_copy); |
---|
1569 | else |
---|
1570 | tbf_control(vifp, mb_copy, ip, ip_copy->ip_len); |
---|
1571 | } |
---|
1572 | |
---|
1573 | /* |
---|
1574 | * De-encapsulate a packet and feed it back through ip input (this |
---|
1575 | * routine is called whenever IP gets a packet with proto type |
---|
1576 | * ENCAP_PROTO and a local destination address). |
---|
1577 | */ |
---|
1578 | void |
---|
1579 | #ifdef MROUTE_LKM |
---|
1580 | X_ipip_input(struct mbuf *m, int iphlen) |
---|
1581 | #else |
---|
1582 | ipip_input(struct mbuf *m, int iphlen) |
---|
1583 | #endif |
---|
1584 | { |
---|
1585 | struct ifnet *ifp = m->m_pkthdr.rcvif; |
---|
1586 | register struct ip *ip = mtod(m, struct ip *); |
---|
1587 | register int hlen = ip->ip_hl << 2; |
---|
1588 | register int s; |
---|
1589 | register struct ifqueue *ifq; |
---|
1590 | register struct vif *vifp; |
---|
1591 | |
---|
1592 | if (!have_encap_tunnel) { |
---|
1593 | rip_input(m, iphlen); |
---|
1594 | return; |
---|
1595 | } |
---|
1596 | /* |
---|
1597 | * dump the packet if it's not to a multicast destination or if |
---|
1598 | * we don't have an encapsulating tunnel with the source. |
---|
1599 | * Note: This code assumes that the remote site IP address |
---|
1600 | * uniquely identifies the tunnel (i.e., that this site has |
---|
1601 | * at most one tunnel with the remote site). |
---|
1602 | */ |
---|
1603 | if (! IN_MULTICAST(ntohl(((struct ip *)((char *)ip + hlen))->ip_dst.s_addr))) { |
---|
1604 | ++mrtstat.mrts_bad_tunnel; |
---|
1605 | m_freem(m); |
---|
1606 | return; |
---|
1607 | } |
---|
1608 | if (ip->ip_src.s_addr != last_encap_src) { |
---|
1609 | register struct vif *vife; |
---|
1610 | |
---|
1611 | vifp = viftable; |
---|
1612 | vife = vifp + numvifs; |
---|
1613 | last_encap_src = ip->ip_src.s_addr; |
---|
1614 | last_encap_vif = 0; |
---|
1615 | for ( ; vifp < vife; ++vifp) |
---|
1616 | if (vifp->v_rmt_addr.s_addr == ip->ip_src.s_addr) { |
---|
1617 | if ((vifp->v_flags & (VIFF_TUNNEL|VIFF_SRCRT)) |
---|
1618 | == VIFF_TUNNEL) |
---|
1619 | last_encap_vif = vifp; |
---|
1620 | break; |
---|
1621 | } |
---|
1622 | } |
---|
1623 | if ((vifp = last_encap_vif) == 0) { |
---|
1624 | last_encap_src = 0; |
---|
1625 | mrtstat.mrts_cant_tunnel++; /*XXX*/ |
---|
1626 | m_freem(m); |
---|
1627 | if (mrtdebug) |
---|
1628 | log(LOG_DEBUG, "ip_mforward: no tunnel with %x\n", |
---|
1629 | ntohl(ip->ip_src.s_addr)); |
---|
1630 | return; |
---|
1631 | } |
---|
1632 | ifp = vifp->v_ifp; |
---|
1633 | |
---|
1634 | if (hlen > IP_HDR_LEN) |
---|
1635 | ip_stripoptions(m, (struct mbuf *) 0); |
---|
1636 | m->m_data += IP_HDR_LEN; |
---|
1637 | m->m_len -= IP_HDR_LEN; |
---|
1638 | m->m_pkthdr.len -= IP_HDR_LEN; |
---|
1639 | m->m_pkthdr.rcvif = ifp; |
---|
1640 | |
---|
1641 | ifq = &ipintrq; |
---|
1642 | s = splimp(); |
---|
1643 | if (IF_QFULL(ifq)) { |
---|
1644 | IF_DROP(ifq); |
---|
1645 | m_freem(m); |
---|
1646 | } else { |
---|
1647 | IF_ENQUEUE(ifq, m); |
---|
1648 | /* |
---|
1649 | * normally we would need a "schednetisr(NETISR_IP)" |
---|
1650 | * here but we were called by ip_input and it is going |
---|
1651 | * to loop back & try to dequeue the packet we just |
---|
1652 | * queued as soon as we return so we avoid the |
---|
1653 | * unnecessary software interrrupt. |
---|
1654 | */ |
---|
1655 | } |
---|
1656 | splx(s); |
---|
1657 | } |
---|
1658 | |
---|
1659 | /* |
---|
1660 | * Token bucket filter module |
---|
1661 | */ |
---|
1662 | |
---|
1663 | static void |
---|
1664 | tbf_control(struct vif *vifp, struct mbuf *m, struct ip *ip, u_long p_len) |
---|
1665 | { |
---|
1666 | register struct tbf *t = vifp->v_tbf; |
---|
1667 | |
---|
1668 | if (p_len > MAX_BKT_SIZE) { |
---|
1669 | /* drop if packet is too large */ |
---|
1670 | mrtstat.mrts_pkt2large++; |
---|
1671 | m_freem(m); |
---|
1672 | return; |
---|
1673 | } |
---|
1674 | |
---|
1675 | tbf_update_tokens(vifp); |
---|
1676 | |
---|
1677 | /* if there are enough tokens, |
---|
1678 | * and the queue is empty, |
---|
1679 | * send this packet out |
---|
1680 | */ |
---|
1681 | |
---|
1682 | if (t->tbf_q_len == 0) { |
---|
1683 | /* queue empty, send packet if enough tokens */ |
---|
1684 | if (p_len <= t->tbf_n_tok) { |
---|
1685 | t->tbf_n_tok -= p_len; |
---|
1686 | tbf_send_packet(vifp, m); |
---|
1687 | } else { |
---|
1688 | /* queue packet and timeout till later */ |
---|
1689 | tbf_queue(vifp, m); |
---|
1690 | timeout(tbf_reprocess_q, (caddr_t)vifp, TBF_REPROCESS); |
---|
1691 | } |
---|
1692 | } else if (t->tbf_q_len < t->tbf_max_q_len) { |
---|
1693 | /* finite queue length, so queue pkts and process queue */ |
---|
1694 | tbf_queue(vifp, m); |
---|
1695 | tbf_process_q(vifp); |
---|
1696 | } else { |
---|
1697 | /* queue length too much, try to dq and queue and process */ |
---|
1698 | if (!tbf_dq_sel(vifp, ip)) { |
---|
1699 | mrtstat.mrts_q_overflow++; |
---|
1700 | m_freem(m); |
---|
1701 | return; |
---|
1702 | } else { |
---|
1703 | tbf_queue(vifp, m); |
---|
1704 | tbf_process_q(vifp); |
---|
1705 | } |
---|
1706 | } |
---|
1707 | return; |
---|
1708 | } |
---|
1709 | |
---|
1710 | /* |
---|
1711 | * adds a packet to the queue at the interface |
---|
1712 | */ |
---|
1713 | static void |
---|
1714 | tbf_queue(struct vif *vifp, struct mbuf *m) |
---|
1715 | { |
---|
1716 | register int s = splnet(); |
---|
1717 | register struct tbf *t = vifp->v_tbf; |
---|
1718 | |
---|
1719 | if (t->tbf_t == NULL) { |
---|
1720 | /* Queue was empty */ |
---|
1721 | t->tbf_q = m; |
---|
1722 | } else { |
---|
1723 | /* Insert at tail */ |
---|
1724 | t->tbf_t->m_act = m; |
---|
1725 | } |
---|
1726 | |
---|
1727 | /* Set new tail pointer */ |
---|
1728 | t->tbf_t = m; |
---|
1729 | |
---|
1730 | #ifdef DIAGNOSTIC |
---|
1731 | /* Make sure we didn't get fed a bogus mbuf */ |
---|
1732 | if (m->m_act) |
---|
1733 | panic("tbf_queue: m_act"); |
---|
1734 | #endif |
---|
1735 | m->m_act = NULL; |
---|
1736 | |
---|
1737 | t->tbf_q_len++; |
---|
1738 | |
---|
1739 | splx(s); |
---|
1740 | } |
---|
1741 | |
---|
1742 | |
---|
1743 | /* |
---|
1744 | * processes the queue at the interface |
---|
1745 | */ |
---|
1746 | static void |
---|
1747 | tbf_process_q(struct vif *vifp) |
---|
1748 | { |
---|
1749 | register struct mbuf *m; |
---|
1750 | register int len; |
---|
1751 | register int s = splnet(); |
---|
1752 | register struct tbf *t = vifp->v_tbf; |
---|
1753 | |
---|
1754 | /* loop through the queue at the interface and send as many packets |
---|
1755 | * as possible |
---|
1756 | */ |
---|
1757 | while (t->tbf_q_len > 0) { |
---|
1758 | m = t->tbf_q; |
---|
1759 | |
---|
1760 | len = mtod(m, struct ip *)->ip_len; |
---|
1761 | |
---|
1762 | /* determine if the packet can be sent */ |
---|
1763 | if (len <= t->tbf_n_tok) { |
---|
1764 | /* if so, |
---|
1765 | * reduce no of tokens, dequeue the packet, |
---|
1766 | * send the packet. |
---|
1767 | */ |
---|
1768 | t->tbf_n_tok -= len; |
---|
1769 | |
---|
1770 | t->tbf_q = m->m_act; |
---|
1771 | if (--t->tbf_q_len == 0) |
---|
1772 | t->tbf_t = NULL; |
---|
1773 | |
---|
1774 | m->m_act = NULL; |
---|
1775 | tbf_send_packet(vifp, m); |
---|
1776 | |
---|
1777 | } else break; |
---|
1778 | } |
---|
1779 | splx(s); |
---|
1780 | } |
---|
1781 | |
---|
1782 | static void |
---|
1783 | tbf_reprocess_q(void *xvifp) |
---|
1784 | { |
---|
1785 | register struct vif *vifp = xvifp; |
---|
1786 | if (ip_mrouter == NULL) |
---|
1787 | return; |
---|
1788 | |
---|
1789 | tbf_update_tokens(vifp); |
---|
1790 | |
---|
1791 | tbf_process_q(vifp); |
---|
1792 | |
---|
1793 | if (vifp->v_tbf->tbf_q_len) |
---|
1794 | timeout(tbf_reprocess_q, (caddr_t)vifp, TBF_REPROCESS); |
---|
1795 | } |
---|
1796 | |
---|
1797 | /* function that will selectively discard a member of the queue |
---|
1798 | * based on the precedence value and the priority |
---|
1799 | */ |
---|
1800 | static int |
---|
1801 | tbf_dq_sel(struct vif *vifp, struct ip *ip) |
---|
1802 | { |
---|
1803 | register int s = splnet(); |
---|
1804 | register u_int p; |
---|
1805 | register struct mbuf *m, *last; |
---|
1806 | register struct mbuf **np; |
---|
1807 | register struct tbf *t = vifp->v_tbf; |
---|
1808 | |
---|
1809 | p = priority(vifp, ip); |
---|
1810 | |
---|
1811 | np = &t->tbf_q; |
---|
1812 | last = NULL; |
---|
1813 | while ((m = *np) != NULL) { |
---|
1814 | if (p > priority(vifp, mtod(m, struct ip *))) { |
---|
1815 | *np = m->m_act; |
---|
1816 | /* If we're removing the last packet, fix the tail pointer */ |
---|
1817 | if (m == t->tbf_t) |
---|
1818 | t->tbf_t = last; |
---|
1819 | m_freem(m); |
---|
1820 | /* it's impossible for the queue to be empty, but |
---|
1821 | * we check anyway. */ |
---|
1822 | if (--t->tbf_q_len == 0) |
---|
1823 | t->tbf_t = NULL; |
---|
1824 | splx(s); |
---|
1825 | mrtstat.mrts_drop_sel++; |
---|
1826 | return(1); |
---|
1827 | } |
---|
1828 | np = &m->m_act; |
---|
1829 | last = m; |
---|
1830 | } |
---|
1831 | splx(s); |
---|
1832 | return(0); |
---|
1833 | } |
---|
1834 | |
---|
1835 | static void |
---|
1836 | tbf_send_packet(struct vif *vifp, struct mbuf *m) |
---|
1837 | { |
---|
1838 | struct ip_moptions imo; |
---|
1839 | int error; |
---|
1840 | static struct route ro; |
---|
1841 | int s = splnet(); |
---|
1842 | |
---|
1843 | if (vifp->v_flags & VIFF_TUNNEL) { |
---|
1844 | /* If tunnel options */ |
---|
1845 | ip_output(m, (struct mbuf *)0, &vifp->v_route, |
---|
1846 | IP_FORWARDING, (struct ip_moptions *)0); |
---|
1847 | } else { |
---|
1848 | imo.imo_multicast_ifp = vifp->v_ifp; |
---|
1849 | imo.imo_multicast_ttl = mtod(m, struct ip *)->ip_ttl - 1; |
---|
1850 | imo.imo_multicast_loop = 1; |
---|
1851 | imo.imo_multicast_vif = -1; |
---|
1852 | |
---|
1853 | /* |
---|
1854 | * Re-entrancy should not be a problem here, because |
---|
1855 | * the packets that we send out and are looped back at us |
---|
1856 | * should get rejected because they appear to come from |
---|
1857 | * the loopback interface, thus preventing looping. |
---|
1858 | */ |
---|
1859 | error = ip_output(m, (struct mbuf *)0, &ro, |
---|
1860 | IP_FORWARDING, &imo); |
---|
1861 | |
---|
1862 | if (mrtdebug & DEBUG_XMIT) |
---|
1863 | log(LOG_DEBUG, "phyint_send on vif %d err %d\n", |
---|
1864 | vifp - viftable, error); |
---|
1865 | } |
---|
1866 | splx(s); |
---|
1867 | } |
---|
1868 | |
---|
1869 | /* determine the current time and then |
---|
1870 | * the elapsed time (between the last time and time now) |
---|
1871 | * in milliseconds & update the no. of tokens in the bucket |
---|
1872 | */ |
---|
1873 | static void |
---|
1874 | tbf_update_tokens(struct vif *vifp) |
---|
1875 | { |
---|
1876 | struct timeval tp; |
---|
1877 | register u_long tm; |
---|
1878 | register int s = splnet(); |
---|
1879 | register struct tbf *t = vifp->v_tbf; |
---|
1880 | |
---|
1881 | GET_TIME(tp); |
---|
1882 | |
---|
1883 | TV_DELTA(tp, t->tbf_last_pkt_t, tm); |
---|
1884 | |
---|
1885 | /* |
---|
1886 | * This formula is actually |
---|
1887 | * "time in seconds" * "bytes/second". |
---|
1888 | * |
---|
1889 | * (tm / 1000000) * (v_rate_limit * 1000 * (1000/1024) / 8) |
---|
1890 | * |
---|
1891 | * The (1000/1024) was introduced in add_vif to optimize |
---|
1892 | * this divide into a shift. |
---|
1893 | */ |
---|
1894 | t->tbf_n_tok += tm * vifp->v_rate_limit / 1024 / 8; |
---|
1895 | t->tbf_last_pkt_t = tp; |
---|
1896 | |
---|
1897 | if (t->tbf_n_tok > MAX_BKT_SIZE) |
---|
1898 | t->tbf_n_tok = MAX_BKT_SIZE; |
---|
1899 | |
---|
1900 | splx(s); |
---|
1901 | } |
---|
1902 | |
---|
1903 | static int |
---|
1904 | priority(struct vif *vifp, struct ip *ip) |
---|
1905 | { |
---|
1906 | register int prio; |
---|
1907 | |
---|
1908 | /* temporary hack; may add general packet classifier some day */ |
---|
1909 | |
---|
1910 | /* |
---|
1911 | * The UDP port space is divided up into four priority ranges: |
---|
1912 | * [0, 16384) : unclassified - lowest priority |
---|
1913 | * [16384, 32768) : audio - highest priority |
---|
1914 | * [32768, 49152) : whiteboard - medium priority |
---|
1915 | * [49152, 65536) : video - low priority |
---|
1916 | */ |
---|
1917 | if (ip->ip_p == IPPROTO_UDP) { |
---|
1918 | struct udphdr *udp = (struct udphdr *)(((char *)ip) + (ip->ip_hl << 2)); |
---|
1919 | switch (ntohs(udp->uh_dport) & 0xc000) { |
---|
1920 | case 0x4000: |
---|
1921 | prio = 70; |
---|
1922 | break; |
---|
1923 | case 0x8000: |
---|
1924 | prio = 60; |
---|
1925 | break; |
---|
1926 | case 0xc000: |
---|
1927 | prio = 55; |
---|
1928 | break; |
---|
1929 | default: |
---|
1930 | prio = 50; |
---|
1931 | break; |
---|
1932 | } |
---|
1933 | if (tbfdebug > 1) |
---|
1934 | log(LOG_DEBUG, "port %x prio%d\n", ntohs(udp->uh_dport), prio); |
---|
1935 | } else { |
---|
1936 | prio = 50; |
---|
1937 | } |
---|
1938 | return prio; |
---|
1939 | } |
---|
1940 | |
---|
1941 | /* |
---|
1942 | * End of token bucket filter modifications |
---|
1943 | */ |
---|
1944 | |
---|
1945 | int |
---|
1946 | ip_rsvp_vif_init(struct socket *so, struct mbuf *m) |
---|
1947 | { |
---|
1948 | int i; |
---|
1949 | register int s; |
---|
1950 | |
---|
1951 | if (rsvpdebug) |
---|
1952 | printf("ip_rsvp_vif_init: so_type = %d, pr_protocol = %d\n", |
---|
1953 | so->so_type, so->so_proto->pr_protocol); |
---|
1954 | |
---|
1955 | if (so->so_type != SOCK_RAW || so->so_proto->pr_protocol != IPPROTO_RSVP) |
---|
1956 | return EOPNOTSUPP; |
---|
1957 | |
---|
1958 | /* Check mbuf. */ |
---|
1959 | if (m == NULL || m->m_len != sizeof(int)) { |
---|
1960 | return EINVAL; |
---|
1961 | } |
---|
1962 | i = *(mtod(m, int *)); |
---|
1963 | |
---|
1964 | if (rsvpdebug) |
---|
1965 | printf("ip_rsvp_vif_init: vif = %d rsvp_on = %d\n",i,rsvp_on); |
---|
1966 | |
---|
1967 | s = splnet(); |
---|
1968 | |
---|
1969 | /* Check vif. */ |
---|
1970 | if (!legal_vif_num(i)) { |
---|
1971 | splx(s); |
---|
1972 | return EADDRNOTAVAIL; |
---|
1973 | } |
---|
1974 | |
---|
1975 | /* Check if socket is available. */ |
---|
1976 | if (viftable[i].v_rsvpd != NULL) { |
---|
1977 | splx(s); |
---|
1978 | return EADDRINUSE; |
---|
1979 | } |
---|
1980 | |
---|
1981 | viftable[i].v_rsvpd = so; |
---|
1982 | /* This may seem silly, but we need to be sure we don't over-increment |
---|
1983 | * the RSVP counter, in case something slips up. |
---|
1984 | */ |
---|
1985 | if (!viftable[i].v_rsvp_on) { |
---|
1986 | viftable[i].v_rsvp_on = 1; |
---|
1987 | rsvp_on++; |
---|
1988 | } |
---|
1989 | |
---|
1990 | splx(s); |
---|
1991 | return 0; |
---|
1992 | } |
---|
1993 | |
---|
1994 | int |
---|
1995 | ip_rsvp_vif_done(struct socket *so, struct mbuf *m) |
---|
1996 | { |
---|
1997 | int i; |
---|
1998 | register int s; |
---|
1999 | |
---|
2000 | if (rsvpdebug) |
---|
2001 | printf("ip_rsvp_vif_done: so_type = %d, pr_protocol = %d\n", |
---|
2002 | so->so_type, so->so_proto->pr_protocol); |
---|
2003 | |
---|
2004 | if (so->so_type != SOCK_RAW || so->so_proto->pr_protocol != IPPROTO_RSVP) |
---|
2005 | return EOPNOTSUPP; |
---|
2006 | |
---|
2007 | /* Check mbuf. */ |
---|
2008 | if (m == NULL || m->m_len != sizeof(int)) { |
---|
2009 | return EINVAL; |
---|
2010 | } |
---|
2011 | i = *(mtod(m, int *)); |
---|
2012 | |
---|
2013 | s = splnet(); |
---|
2014 | |
---|
2015 | /* Check vif. */ |
---|
2016 | if (!legal_vif_num(i)) { |
---|
2017 | splx(s); |
---|
2018 | return EADDRNOTAVAIL; |
---|
2019 | } |
---|
2020 | |
---|
2021 | if (rsvpdebug) |
---|
2022 | printf("ip_rsvp_vif_done: v_rsvpd = %p so = %p\n", |
---|
2023 | viftable[i].v_rsvpd, so); |
---|
2024 | |
---|
2025 | viftable[i].v_rsvpd = NULL; |
---|
2026 | /* This may seem silly, but we need to be sure we don't over-decrement |
---|
2027 | * the RSVP counter, in case something slips up. |
---|
2028 | */ |
---|
2029 | if (viftable[i].v_rsvp_on) { |
---|
2030 | viftable[i].v_rsvp_on = 0; |
---|
2031 | rsvp_on--; |
---|
2032 | } |
---|
2033 | |
---|
2034 | splx(s); |
---|
2035 | return 0; |
---|
2036 | } |
---|
2037 | |
---|
2038 | void |
---|
2039 | ip_rsvp_force_done(struct socket *so) |
---|
2040 | { |
---|
2041 | int vifi; |
---|
2042 | register int s; |
---|
2043 | |
---|
2044 | /* Don't bother if it is not the right type of socket. */ |
---|
2045 | if (so->so_type != SOCK_RAW || so->so_proto->pr_protocol != IPPROTO_RSVP) |
---|
2046 | return; |
---|
2047 | |
---|
2048 | s = splnet(); |
---|
2049 | |
---|
2050 | /* The socket may be attached to more than one vif...this |
---|
2051 | * is perfectly legal. |
---|
2052 | */ |
---|
2053 | for (vifi = 0; vifi < numvifs; vifi++) { |
---|
2054 | if (viftable[vifi].v_rsvpd == so) { |
---|
2055 | viftable[vifi].v_rsvpd = NULL; |
---|
2056 | /* This may seem silly, but we need to be sure we don't |
---|
2057 | * over-decrement the RSVP counter, in case something slips up. |
---|
2058 | */ |
---|
2059 | if (viftable[vifi].v_rsvp_on) { |
---|
2060 | viftable[vifi].v_rsvp_on = 0; |
---|
2061 | rsvp_on--; |
---|
2062 | } |
---|
2063 | } |
---|
2064 | } |
---|
2065 | |
---|
2066 | splx(s); |
---|
2067 | return; |
---|
2068 | } |
---|
2069 | |
---|
2070 | void |
---|
2071 | rsvp_input(struct mbuf *m, int iphlen) |
---|
2072 | { |
---|
2073 | int vifi; |
---|
2074 | register struct ip *ip = mtod(m, struct ip *); |
---|
2075 | static struct sockaddr_in rsvp_src = { sizeof rsvp_src, AF_INET }; |
---|
2076 | register int s; |
---|
2077 | struct ifnet *ifp; |
---|
2078 | |
---|
2079 | if (rsvpdebug) |
---|
2080 | printf("rsvp_input: rsvp_on %d\n",rsvp_on); |
---|
2081 | |
---|
2082 | /* Can still get packets with rsvp_on = 0 if there is a local member |
---|
2083 | * of the group to which the RSVP packet is addressed. But in this |
---|
2084 | * case we want to throw the packet away. |
---|
2085 | */ |
---|
2086 | if (!rsvp_on) { |
---|
2087 | m_freem(m); |
---|
2088 | return; |
---|
2089 | } |
---|
2090 | |
---|
2091 | /* If the old-style non-vif-associated socket is set, then use |
---|
2092 | * it and ignore the new ones. |
---|
2093 | */ |
---|
2094 | if (ip_rsvpd != NULL) { |
---|
2095 | if (rsvpdebug) |
---|
2096 | printf("rsvp_input: Sending packet up old-style socket\n"); |
---|
2097 | rip_input(m, iphlen); |
---|
2098 | return; |
---|
2099 | } |
---|
2100 | |
---|
2101 | s = splnet(); |
---|
2102 | |
---|
2103 | if (rsvpdebug) |
---|
2104 | printf("rsvp_input: check vifs\n"); |
---|
2105 | |
---|
2106 | #ifdef DIAGNOSTIC |
---|
2107 | if (!(m->m_flags & M_PKTHDR)) |
---|
2108 | panic("rsvp_input no hdr"); |
---|
2109 | #endif |
---|
2110 | |
---|
2111 | ifp = m->m_pkthdr.rcvif; |
---|
2112 | /* Find which vif the packet arrived on. */ |
---|
2113 | for (vifi = 0; vifi < numvifs; vifi++) { |
---|
2114 | if (viftable[vifi].v_ifp == ifp) |
---|
2115 | break; |
---|
2116 | } |
---|
2117 | |
---|
2118 | if (vifi == numvifs) { |
---|
2119 | /* Can't find vif packet arrived on. Drop packet. */ |
---|
2120 | if (rsvpdebug) |
---|
2121 | printf("rsvp_input: Can't find vif for packet...dropping it.\n"); |
---|
2122 | m_freem(m); |
---|
2123 | splx(s); |
---|
2124 | return; |
---|
2125 | } |
---|
2126 | |
---|
2127 | if (rsvpdebug) |
---|
2128 | printf("rsvp_input: check socket\n"); |
---|
2129 | |
---|
2130 | if (viftable[vifi].v_rsvpd == NULL) { |
---|
2131 | /* drop packet, since there is no specific socket for this |
---|
2132 | * interface */ |
---|
2133 | if (rsvpdebug) |
---|
2134 | printf("rsvp_input: No socket defined for vif %d\n",vifi); |
---|
2135 | m_freem(m); |
---|
2136 | splx(s); |
---|
2137 | return; |
---|
2138 | } |
---|
2139 | rsvp_src.sin_addr = ip->ip_src; |
---|
2140 | |
---|
2141 | if (rsvpdebug && m) |
---|
2142 | printf("rsvp_input: m->m_len = %d, sbspace() = %ld\n", |
---|
2143 | m->m_len,sbspace(&(viftable[vifi].v_rsvpd->so_rcv))); |
---|
2144 | |
---|
2145 | if (socket_send(viftable[vifi].v_rsvpd, m, &rsvp_src) < 0) |
---|
2146 | if (rsvpdebug) |
---|
2147 | printf("rsvp_input: Failed to append to socket\n"); |
---|
2148 | else |
---|
2149 | if (rsvpdebug) |
---|
2150 | printf("rsvp_input: send packet up\n"); |
---|
2151 | |
---|
2152 | splx(s); |
---|
2153 | } |
---|
2154 | |
---|
2155 | #ifdef MROUTE_LKM |
---|
2156 | #include <sys/conf.h> |
---|
2157 | #include <sys/exec.h> |
---|
2158 | #include <sys/sysent.h> |
---|
2159 | #include <sys/lkm.h> |
---|
2160 | |
---|
2161 | MOD_MISC("ip_mroute_mod") |
---|
2162 | |
---|
2163 | static int |
---|
2164 | ip_mroute_mod_handle(struct lkm_table *lkmtp, int cmd) |
---|
2165 | { |
---|
2166 | int i; |
---|
2167 | struct lkm_misc *args = lkmtp->private.lkm_misc; |
---|
2168 | int err = 0; |
---|
2169 | |
---|
2170 | switch(cmd) { |
---|
2171 | static int (*old_ip_mrouter_cmd)(); |
---|
2172 | static int (*old_ip_mrouter_done)(); |
---|
2173 | static int (*old_ip_mforward)(); |
---|
2174 | static int (*old_mrt_ioctl)(); |
---|
2175 | static void (*old_proto4_input)(); |
---|
2176 | static int (*old_legal_vif_num)(); |
---|
2177 | extern struct protosw inetsw[]; |
---|
2178 | |
---|
2179 | case LKM_E_LOAD: |
---|
2180 | if(lkmexists(lkmtp) || ip_mrtproto) |
---|
2181 | return(EEXIST); |
---|
2182 | old_ip_mrouter_cmd = ip_mrouter_cmd; |
---|
2183 | ip_mrouter_cmd = X_ip_mrouter_cmd; |
---|
2184 | old_ip_mrouter_done = ip_mrouter_done; |
---|
2185 | ip_mrouter_done = X_ip_mrouter_done; |
---|
2186 | old_ip_mforward = ip_mforward; |
---|
2187 | ip_mforward = X_ip_mforward; |
---|
2188 | old_mrt_ioctl = mrt_ioctl; |
---|
2189 | mrt_ioctl = X_mrt_ioctl; |
---|
2190 | old_proto4_input = inetsw[ip_protox[ENCAP_PROTO]].pr_input; |
---|
2191 | inetsw[ip_protox[ENCAP_PROTO]].pr_input = X_ipip_input; |
---|
2192 | old_legal_vif_num = legal_vif_num; |
---|
2193 | legal_vif_num = X_legal_vif_num; |
---|
2194 | ip_mrtproto = IGMP_DVMRP; |
---|
2195 | |
---|
2196 | printf("\nIP multicast routing loaded\n"); |
---|
2197 | break; |
---|
2198 | |
---|
2199 | case LKM_E_UNLOAD: |
---|
2200 | if (ip_mrouter) |
---|
2201 | return EINVAL; |
---|
2202 | |
---|
2203 | ip_mrouter_cmd = old_ip_mrouter_cmd; |
---|
2204 | ip_mrouter_done = old_ip_mrouter_done; |
---|
2205 | ip_mforward = old_ip_mforward; |
---|
2206 | mrt_ioctl = old_mrt_ioctl; |
---|
2207 | inetsw[ip_protox[ENCAP_PROTO]].pr_input = old_proto4_input; |
---|
2208 | legal_vif_num = old_legal_vif_num; |
---|
2209 | ip_mrtproto = 0; |
---|
2210 | break; |
---|
2211 | |
---|
2212 | default: |
---|
2213 | err = EINVAL; |
---|
2214 | break; |
---|
2215 | } |
---|
2216 | |
---|
2217 | return(err); |
---|
2218 | } |
---|
2219 | |
---|
2220 | int |
---|
2221 | ip_mroute_mod(struct lkm_table *lkmtp, int cmd, int ver) { |
---|
2222 | DISPATCH(lkmtp, cmd, ver, ip_mroute_mod_handle, ip_mroute_mod_handle, |
---|
2223 | nosys); |
---|
2224 | } |
---|
2225 | |
---|
2226 | #endif /* MROUTE_LKM */ |
---|
2227 | #endif /* MROUTING */ |
---|