[39e6e65a] | 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 <sys/queue.h> |
---|
| 19 | #include <sys/systm.h> |
---|
[78219ad1] | 20 | #include <sys/sysctl.h> |
---|
[39e6e65a] | 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 __P((int vifi)); |
---|
| 59 | extern int _ip_mforward __P((struct ip *ip, struct ifnet *ifp, |
---|
| 60 | struct mbuf *m, struct ip_moptions *imo)); |
---|
| 61 | extern int _ip_mrouter_done __P((void)); |
---|
| 62 | extern int _ip_mrouter_get __P((int cmd, struct socket *so, |
---|
| 63 | struct mbuf **m)); |
---|
| 64 | extern int _ip_mrouter_set __P((int cmd, struct socket *so, |
---|
| 65 | struct mbuf *m)); |
---|
| 66 | extern int _mrt_ioctl __P((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; |
---|
[33679ec4] | 73 | /* static u_int ip_mrtproto = 0; */ |
---|
| 74 | /* static struct mrtstat mrtstat; */ |
---|
[39e6e65a] | 75 | u_int rsvpdebug = 0; |
---|
| 76 | |
---|
| 77 | int |
---|
| 78 | _ip_mrouter_set(cmd, so, m) |
---|
| 79 | int cmd; |
---|
| 80 | struct socket *so; |
---|
| 81 | struct mbuf *m; |
---|
| 82 | { |
---|
| 83 | return(EOPNOTSUPP); |
---|
| 84 | } |
---|
| 85 | |
---|
| 86 | int (*ip_mrouter_set)(int, struct socket *, struct mbuf *) = _ip_mrouter_set; |
---|
| 87 | |
---|
| 88 | |
---|
| 89 | int |
---|
| 90 | _ip_mrouter_get(cmd, so, m) |
---|
| 91 | int cmd; |
---|
| 92 | struct socket *so; |
---|
| 93 | struct mbuf **m; |
---|
| 94 | { |
---|
| 95 | return(EOPNOTSUPP); |
---|
| 96 | } |
---|
| 97 | |
---|
| 98 | int (*ip_mrouter_get)(int, struct socket *, struct mbuf **) = _ip_mrouter_get; |
---|
| 99 | |
---|
| 100 | int |
---|
| 101 | _ip_mrouter_done() |
---|
| 102 | { |
---|
| 103 | return(0); |
---|
| 104 | } |
---|
| 105 | |
---|
| 106 | int (*ip_mrouter_done)(void) = _ip_mrouter_done; |
---|
| 107 | |
---|
| 108 | int |
---|
| 109 | _ip_mforward(ip, ifp, m, imo) |
---|
| 110 | struct ip *ip; |
---|
| 111 | struct ifnet *ifp; |
---|
| 112 | struct mbuf *m; |
---|
| 113 | struct ip_moptions *imo; |
---|
| 114 | { |
---|
| 115 | return(0); |
---|
| 116 | } |
---|
| 117 | |
---|
| 118 | int (*ip_mforward)(struct ip *, struct ifnet *, struct mbuf *, |
---|
| 119 | struct ip_moptions *) = _ip_mforward; |
---|
| 120 | |
---|
| 121 | int |
---|
| 122 | _mrt_ioctl(int req, caddr_t data, struct proc *p) |
---|
| 123 | { |
---|
| 124 | return EOPNOTSUPP; |
---|
| 125 | } |
---|
| 126 | |
---|
| 127 | int (*mrt_ioctl)(int, caddr_t, struct proc *) = _mrt_ioctl; |
---|
| 128 | |
---|
| 129 | void |
---|
| 130 | rsvp_input(m, iphlen) /* XXX must fixup manually */ |
---|
| 131 | struct mbuf *m; |
---|
| 132 | int iphlen; |
---|
| 133 | { |
---|
| 134 | /* Can still get packets with rsvp_on = 0 if there is a local member |
---|
| 135 | * of the group to which the RSVP packet is addressed. But in this |
---|
| 136 | * case we want to throw the packet away. |
---|
| 137 | */ |
---|
| 138 | if (!rsvp_on) { |
---|
| 139 | m_freem(m); |
---|
| 140 | return; |
---|
| 141 | } |
---|
| 142 | |
---|
| 143 | if (ip_rsvpd != NULL) { |
---|
| 144 | if (rsvpdebug) |
---|
| 145 | printf("rsvp_input: Sending packet up old-style socket\n"); |
---|
| 146 | rip_input(m, iphlen); |
---|
| 147 | return; |
---|
| 148 | } |
---|
| 149 | /* Drop the packet */ |
---|
| 150 | m_freem(m); |
---|
| 151 | } |
---|
| 152 | |
---|
| 153 | void ipip_input(struct mbuf *m, int iphlen) { /* XXX must fixup manually */ |
---|
| 154 | rip_input(m, iphlen); |
---|
| 155 | } |
---|
| 156 | |
---|
| 157 | int (*legal_vif_num)(int) = 0; |
---|
| 158 | |
---|
| 159 | /* |
---|
| 160 | * This should never be called, since IP_MULTICAST_VIF should fail, but |
---|
| 161 | * just in case it does get called, the code a little lower in ip_output |
---|
| 162 | * will assign the packet a local address. |
---|
| 163 | */ |
---|
| 164 | u_long |
---|
| 165 | _ip_mcast_src(int vifi) { return INADDR_ANY; } |
---|
| 166 | u_long (*ip_mcast_src)(int) = _ip_mcast_src; |
---|
| 167 | |
---|
| 168 | int |
---|
| 169 | ip_rsvp_vif_init(so, m) |
---|
| 170 | struct socket *so; |
---|
| 171 | struct mbuf *m; |
---|
| 172 | { |
---|
| 173 | return(EINVAL); |
---|
| 174 | } |
---|
| 175 | |
---|
| 176 | int |
---|
| 177 | ip_rsvp_vif_done(so, m) |
---|
| 178 | struct socket *so; |
---|
| 179 | struct mbuf *m; |
---|
| 180 | { |
---|
| 181 | return(EINVAL); |
---|
| 182 | } |
---|
| 183 | |
---|
| 184 | void |
---|
| 185 | ip_rsvp_force_done(so) |
---|
| 186 | struct socket *so; |
---|
| 187 | { |
---|
| 188 | return; |
---|
| 189 | } |
---|
| 190 | |
---|
| 191 | #else /* MROUTING */ |
---|
| 192 | |
---|
| 193 | #define M_HASCL(m) ((m)->m_flags & M_EXT) |
---|
| 194 | |
---|
| 195 | #define INSIZ sizeof(struct in_addr) |
---|
| 196 | #define same(a1, a2) \ |
---|
| 197 | (bcmp((caddr_t)(a1), (caddr_t)(a2), INSIZ) == 0) |
---|
| 198 | |
---|
| 199 | #define MT_MRTABLE MT_RTABLE /* since nothing else uses it */ |
---|
| 200 | |
---|
| 201 | /* |
---|
| 202 | * Globals. All but ip_mrouter and ip_mrtproto could be static, |
---|
| 203 | * except for netstat or debugging purposes. |
---|
| 204 | */ |
---|
| 205 | #ifndef MROUTE_LKM |
---|
| 206 | struct socket *ip_mrouter = NULL; |
---|
| 207 | struct mrtstat mrtstat; |
---|
| 208 | |
---|
| 209 | int ip_mrtproto = IGMP_DVMRP; /* for netstat only */ |
---|
| 210 | #else /* MROUTE_LKM */ |
---|
| 211 | extern void X_ipip_input __P((struct mbuf *m, int iphlen)); |
---|
| 212 | extern struct mrtstat mrtstat; |
---|
| 213 | static int ip_mrtproto; |
---|
| 214 | #endif |
---|
| 215 | |
---|
| 216 | #define NO_RTE_FOUND 0x1 |
---|
| 217 | #define RTE_FOUND 0x2 |
---|
| 218 | |
---|
| 219 | static struct mbuf *mfctable[MFCTBLSIZ]; |
---|
| 220 | static u_char nexpire[MFCTBLSIZ]; |
---|
| 221 | static struct vif viftable[MAXVIFS]; |
---|
| 222 | static u_int mrtdebug = 0; /* debug level */ |
---|
| 223 | #define DEBUG_MFC 0x02 |
---|
| 224 | #define DEBUG_FORWARD 0x04 |
---|
| 225 | #define DEBUG_EXPIRE 0x08 |
---|
| 226 | #define DEBUG_XMIT 0x10 |
---|
| 227 | static u_int tbfdebug = 0; /* tbf debug level */ |
---|
| 228 | static u_int rsvpdebug = 0; /* rsvp debug level */ |
---|
| 229 | |
---|
| 230 | #define EXPIRE_TIMEOUT (hz / 4) /* 4x / second */ |
---|
| 231 | #define UPCALL_EXPIRE 6 /* number of timeouts */ |
---|
| 232 | |
---|
| 233 | /* |
---|
| 234 | * Define the token bucket filter structures |
---|
| 235 | * tbftable -> each vif has one of these for storing info |
---|
| 236 | */ |
---|
| 237 | |
---|
| 238 | static struct tbf tbftable[MAXVIFS]; |
---|
| 239 | #define TBF_REPROCESS (hz / 100) /* 100x / second */ |
---|
| 240 | |
---|
| 241 | /* |
---|
| 242 | * 'Interfaces' associated with decapsulator (so we can tell |
---|
| 243 | * packets that went through it from ones that get reflected |
---|
| 244 | * by a broken gateway). These interfaces are never linked into |
---|
| 245 | * the system ifnet list & no routes point to them. I.e., packets |
---|
| 246 | * can't be sent this way. They only exist as a placeholder for |
---|
| 247 | * multicast source verification. |
---|
| 248 | */ |
---|
| 249 | static struct ifnet multicast_decap_if[MAXVIFS]; |
---|
| 250 | |
---|
| 251 | #define ENCAP_TTL 64 |
---|
| 252 | #define ENCAP_PROTO IPPROTO_IPIP /* 4 */ |
---|
| 253 | |
---|
| 254 | /* prototype IP hdr for encapsulated packets */ |
---|
| 255 | static struct ip multicast_encap_iphdr = { |
---|
| 256 | #if BYTE_ORDER == LITTLE_ENDIAN |
---|
| 257 | sizeof(struct ip) >> 2, IPVERSION, |
---|
| 258 | #else |
---|
| 259 | IPVERSION, sizeof(struct ip) >> 2, |
---|
| 260 | #endif |
---|
| 261 | 0, /* tos */ |
---|
| 262 | sizeof(struct ip), /* total length */ |
---|
| 263 | 0, /* id */ |
---|
| 264 | 0, /* frag offset */ |
---|
| 265 | ENCAP_TTL, ENCAP_PROTO, |
---|
| 266 | 0, /* checksum */ |
---|
| 267 | }; |
---|
| 268 | |
---|
| 269 | /* |
---|
| 270 | * Private variables. |
---|
| 271 | */ |
---|
| 272 | static vifi_t numvifs = 0; |
---|
| 273 | static int have_encap_tunnel = 0; |
---|
| 274 | |
---|
| 275 | /* |
---|
| 276 | * one-back cache used by ipip_input to locate a tunnel's vif |
---|
| 277 | * given a datagram's src ip address. |
---|
| 278 | */ |
---|
| 279 | static u_long last_encap_src; |
---|
| 280 | static struct vif *last_encap_vif; |
---|
| 281 | |
---|
| 282 | static u_long X_ip_mcast_src __P((int vifi)); |
---|
| 283 | static int X_ip_mforward __P((struct ip *ip, struct ifnet *ifp, struct mbuf *m, struct ip_moptions *imo)); |
---|
| 284 | static int X_ip_mrouter_done __P((void)); |
---|
| 285 | static int X_ip_mrouter_get __P((int cmd, struct socket *so, struct mbuf **m)); |
---|
| 286 | static int X_ip_mrouter_set __P((int cmd, struct socket *so, struct mbuf *m)); |
---|
| 287 | static int X_legal_vif_num __P((int vif)); |
---|
| 288 | static int X_mrt_ioctl __P((int cmd, caddr_t data)); |
---|
| 289 | |
---|
| 290 | static int get_sg_cnt(struct sioc_sg_req *); |
---|
| 291 | static int get_vif_cnt(struct sioc_vif_req *); |
---|
| 292 | static int ip_mrouter_init(struct socket *, struct mbuf *); |
---|
| 293 | static int add_vif(struct vifctl *); |
---|
| 294 | static int del_vif(vifi_t *); |
---|
| 295 | static int add_mfc(struct mfcctl *); |
---|
| 296 | static int del_mfc(struct mfcctl *); |
---|
| 297 | static int socket_send(struct socket *, struct mbuf *, struct sockaddr_in *); |
---|
| 298 | static int get_version(struct mbuf *); |
---|
| 299 | static int get_assert(struct mbuf *); |
---|
| 300 | static int set_assert(int *); |
---|
| 301 | static void expire_upcalls(void *); |
---|
| 302 | static int ip_mdq(struct mbuf *, struct ifnet *, struct mfc *, |
---|
| 303 | vifi_t); |
---|
| 304 | static void phyint_send(struct ip *, struct vif *, struct mbuf *); |
---|
| 305 | static void encap_send(struct ip *, struct vif *, struct mbuf *); |
---|
| 306 | static void tbf_control(struct vif *, struct mbuf *, struct ip *, u_long); |
---|
| 307 | static void tbf_queue(struct vif *, struct mbuf *); |
---|
| 308 | static void tbf_process_q(struct vif *); |
---|
| 309 | static void tbf_reprocess_q(void *); |
---|
| 310 | static int tbf_dq_sel(struct vif *, struct ip *); |
---|
| 311 | static void tbf_send_packet(struct vif *, struct mbuf *); |
---|
| 312 | static void tbf_update_tokens(struct vif *); |
---|
| 313 | static int priority(struct vif *, struct ip *); |
---|
| 314 | void multiencap_decap(struct mbuf *); |
---|
| 315 | |
---|
| 316 | /* |
---|
| 317 | * whether or not special PIM assert processing is enabled. |
---|
| 318 | */ |
---|
| 319 | static int pim_assert; |
---|
| 320 | /* |
---|
| 321 | * Rate limit for assert notification messages, in usec |
---|
| 322 | */ |
---|
| 323 | #define ASSERT_MSG_TIME 3000000 |
---|
| 324 | |
---|
| 325 | /* |
---|
| 326 | * Hash function for a source, group entry |
---|
| 327 | */ |
---|
| 328 | #define MFCHASH(a, g) MFCHASHMOD(((a) >> 20) ^ ((a) >> 10) ^ (a) ^ \ |
---|
| 329 | ((g) >> 20) ^ ((g) >> 10) ^ (g)) |
---|
| 330 | |
---|
| 331 | /* |
---|
| 332 | * Find a route for a given origin IP address and Multicast group address |
---|
| 333 | * Type of service parameter to be added in the future!!! |
---|
| 334 | */ |
---|
| 335 | |
---|
| 336 | #define MFCFIND(o, g, rt) { \ |
---|
| 337 | register struct mbuf *_mb_rt = mfctable[MFCHASH(o,g)]; \ |
---|
| 338 | register struct mfc *_rt = NULL; \ |
---|
| 339 | rt = NULL; \ |
---|
| 340 | ++mrtstat.mrts_mfc_lookups; \ |
---|
| 341 | while (_mb_rt) { \ |
---|
| 342 | _rt = mtod(_mb_rt, struct mfc *); \ |
---|
| 343 | if ((_rt->mfc_origin.s_addr == o) && \ |
---|
| 344 | (_rt->mfc_mcastgrp.s_addr == g) && \ |
---|
| 345 | (_mb_rt->m_act == NULL)) { \ |
---|
| 346 | rt = _rt; \ |
---|
| 347 | break; \ |
---|
| 348 | } \ |
---|
| 349 | _mb_rt = _mb_rt->m_next; \ |
---|
| 350 | } \ |
---|
| 351 | if (rt == NULL) { \ |
---|
| 352 | ++mrtstat.mrts_mfc_misses; \ |
---|
| 353 | } \ |
---|
| 354 | } |
---|
| 355 | |
---|
| 356 | |
---|
| 357 | /* |
---|
| 358 | * Macros to compute elapsed time efficiently |
---|
| 359 | * Borrowed from Van Jacobson's scheduling code |
---|
| 360 | */ |
---|
| 361 | #define TV_DELTA(a, b, delta) { \ |
---|
| 362 | register int xxs; \ |
---|
| 363 | \ |
---|
| 364 | delta = (a).tv_usec - (b).tv_usec; \ |
---|
| 365 | if ((xxs = (a).tv_sec - (b).tv_sec)) { \ |
---|
| 366 | switch (xxs) { \ |
---|
| 367 | case 2: \ |
---|
| 368 | delta += 1000000; \ |
---|
| 369 | /* fall through */ \ |
---|
| 370 | case 1: \ |
---|
| 371 | delta += 1000000; \ |
---|
| 372 | break; \ |
---|
| 373 | default: \ |
---|
| 374 | delta += (1000000 * xxs); \ |
---|
| 375 | } \ |
---|
| 376 | } \ |
---|
| 377 | } |
---|
| 378 | |
---|
| 379 | #define TV_LT(a, b) (((a).tv_usec < (b).tv_usec && \ |
---|
| 380 | (a).tv_sec <= (b).tv_sec) || (a).tv_sec < (b).tv_sec) |
---|
| 381 | |
---|
| 382 | #ifdef UPCALL_TIMING |
---|
| 383 | u_long upcall_data[51]; |
---|
| 384 | static void collate(struct timeval *); |
---|
| 385 | #endif /* UPCALL_TIMING */ |
---|
| 386 | |
---|
| 387 | |
---|
| 388 | /* |
---|
| 389 | * Handle MRT setsockopt commands to modify the multicast routing tables. |
---|
| 390 | */ |
---|
| 391 | static int |
---|
| 392 | X_ip_mrouter_set(cmd, so, m) |
---|
| 393 | int cmd; |
---|
| 394 | struct socket *so; |
---|
| 395 | struct mbuf *m; |
---|
| 396 | { |
---|
| 397 | if (cmd != MRT_INIT && so != ip_mrouter) return EACCES; |
---|
| 398 | |
---|
| 399 | switch (cmd) { |
---|
| 400 | case MRT_INIT: return ip_mrouter_init(so, m); |
---|
| 401 | case MRT_DONE: return ip_mrouter_done(); |
---|
| 402 | case MRT_ADD_VIF: return add_vif (mtod(m, struct vifctl *)); |
---|
| 403 | case MRT_DEL_VIF: return del_vif (mtod(m, vifi_t *)); |
---|
| 404 | case MRT_ADD_MFC: return add_mfc (mtod(m, struct mfcctl *)); |
---|
| 405 | case MRT_DEL_MFC: return del_mfc (mtod(m, struct mfcctl *)); |
---|
| 406 | case MRT_ASSERT: return set_assert(mtod(m, int *)); |
---|
| 407 | default: return EOPNOTSUPP; |
---|
| 408 | } |
---|
| 409 | } |
---|
| 410 | |
---|
| 411 | #ifndef MROUTE_LKM |
---|
| 412 | int (*ip_mrouter_set)(int, struct socket *, struct mbuf *) = X_ip_mrouter_set; |
---|
| 413 | #endif |
---|
| 414 | |
---|
| 415 | /* |
---|
| 416 | * Handle MRT getsockopt commands |
---|
| 417 | */ |
---|
| 418 | static int |
---|
| 419 | X_ip_mrouter_get(cmd, so, m) |
---|
| 420 | int cmd; |
---|
| 421 | struct socket *so; |
---|
| 422 | struct mbuf **m; |
---|
| 423 | { |
---|
| 424 | struct mbuf *mb; |
---|
| 425 | |
---|
| 426 | if (so != ip_mrouter) return EACCES; |
---|
| 427 | |
---|
| 428 | *m = mb = m_get(M_WAIT, MT_SOOPTS); |
---|
| 429 | |
---|
| 430 | switch (cmd) { |
---|
| 431 | case MRT_VERSION: return get_version(mb); |
---|
| 432 | case MRT_ASSERT: return get_assert(mb); |
---|
| 433 | default: return EOPNOTSUPP; |
---|
| 434 | } |
---|
| 435 | } |
---|
| 436 | |
---|
| 437 | #ifndef MROUTE_LKM |
---|
| 438 | int (*ip_mrouter_get)(int, struct socket *, struct mbuf **) = X_ip_mrouter_get; |
---|
| 439 | #endif |
---|
| 440 | |
---|
| 441 | /* |
---|
| 442 | * Handle ioctl commands to obtain information from the cache |
---|
| 443 | */ |
---|
| 444 | static int |
---|
| 445 | X_mrt_ioctl(cmd, data) |
---|
| 446 | int cmd; |
---|
| 447 | caddr_t data; |
---|
| 448 | { |
---|
| 449 | int error = 0; |
---|
| 450 | |
---|
| 451 | switch (cmd) { |
---|
| 452 | case (SIOCGETVIFCNT): |
---|
| 453 | return (get_vif_cnt((struct sioc_vif_req *)data)); |
---|
| 454 | break; |
---|
| 455 | case (SIOCGETSGCNT): |
---|
| 456 | return (get_sg_cnt((struct sioc_sg_req *)data)); |
---|
| 457 | break; |
---|
| 458 | default: |
---|
| 459 | return (EINVAL); |
---|
| 460 | break; |
---|
| 461 | } |
---|
| 462 | return error; |
---|
| 463 | } |
---|
| 464 | |
---|
| 465 | #ifndef MROUTE_LKM |
---|
| 466 | int (*mrt_ioctl)(int, caddr_t) = X_mrt_ioctl; |
---|
| 467 | #endif |
---|
| 468 | |
---|
| 469 | /* |
---|
| 470 | * returns the packet, byte, rpf-failure count for the source group provided |
---|
| 471 | */ |
---|
| 472 | static int |
---|
| 473 | get_sg_cnt(req) |
---|
| 474 | register struct sioc_sg_req *req; |
---|
| 475 | { |
---|
| 476 | register struct mfc *rt; |
---|
| 477 | int s; |
---|
| 478 | |
---|
| 479 | s = splnet(); |
---|
| 480 | MFCFIND(req->src.s_addr, req->grp.s_addr, rt); |
---|
| 481 | splx(s); |
---|
| 482 | if (rt != NULL) { |
---|
| 483 | req->pktcnt = rt->mfc_pkt_cnt; |
---|
| 484 | req->bytecnt = rt->mfc_byte_cnt; |
---|
| 485 | req->wrong_if = rt->mfc_wrong_if; |
---|
| 486 | } else |
---|
| 487 | req->pktcnt = req->bytecnt = req->wrong_if = 0xffffffff; |
---|
| 488 | |
---|
| 489 | return 0; |
---|
| 490 | } |
---|
| 491 | |
---|
| 492 | /* |
---|
| 493 | * returns the input and output packet and byte counts on the vif provided |
---|
| 494 | */ |
---|
| 495 | static int |
---|
| 496 | get_vif_cnt(req) |
---|
| 497 | register struct sioc_vif_req *req; |
---|
| 498 | { |
---|
| 499 | register vifi_t vifi = req->vifi; |
---|
| 500 | |
---|
| 501 | if (vifi >= numvifs) return EINVAL; |
---|
| 502 | |
---|
| 503 | req->icount = viftable[vifi].v_pkt_in; |
---|
| 504 | req->ocount = viftable[vifi].v_pkt_out; |
---|
| 505 | req->ibytes = viftable[vifi].v_bytes_in; |
---|
| 506 | req->obytes = viftable[vifi].v_bytes_out; |
---|
| 507 | |
---|
| 508 | return 0; |
---|
| 509 | } |
---|
| 510 | |
---|
| 511 | /* |
---|
| 512 | * Enable multicast routing |
---|
| 513 | */ |
---|
| 514 | static int |
---|
| 515 | ip_mrouter_init(so, m) |
---|
| 516 | struct socket *so; |
---|
| 517 | struct mbuf *m; |
---|
| 518 | { |
---|
| 519 | int *v; |
---|
| 520 | |
---|
| 521 | if (mrtdebug) |
---|
| 522 | log(LOG_DEBUG,"ip_mrouter_init: so_type = %d, pr_protocol = %d\n", |
---|
| 523 | so->so_type, so->so_proto->pr_protocol); |
---|
| 524 | |
---|
| 525 | if (so->so_type != SOCK_RAW || |
---|
| 526 | so->so_proto->pr_protocol != IPPROTO_IGMP) return EOPNOTSUPP; |
---|
| 527 | |
---|
| 528 | if (!m || (m->m_len != sizeof(int *))) |
---|
| 529 | return ENOPROTOOPT; |
---|
| 530 | |
---|
| 531 | v = mtod(m, int *); |
---|
| 532 | if (*v != 1) |
---|
| 533 | return ENOPROTOOPT; |
---|
| 534 | |
---|
| 535 | if (ip_mrouter != NULL) return EADDRINUSE; |
---|
| 536 | |
---|
| 537 | ip_mrouter = so; |
---|
| 538 | |
---|
| 539 | bzero((caddr_t)mfctable, sizeof(mfctable)); |
---|
| 540 | bzero((caddr_t)nexpire, sizeof(nexpire)); |
---|
| 541 | |
---|
| 542 | pim_assert = 0; |
---|
| 543 | |
---|
| 544 | timeout(expire_upcalls, (caddr_t)NULL, EXPIRE_TIMEOUT); |
---|
| 545 | |
---|
| 546 | if (mrtdebug) |
---|
| 547 | log(LOG_DEBUG, "ip_mrouter_init\n"); |
---|
| 548 | |
---|
| 549 | return 0; |
---|
| 550 | } |
---|
| 551 | |
---|
| 552 | /* |
---|
| 553 | * Disable multicast routing |
---|
| 554 | */ |
---|
| 555 | static int |
---|
| 556 | X_ip_mrouter_done() |
---|
| 557 | { |
---|
| 558 | vifi_t vifi; |
---|
| 559 | int i; |
---|
| 560 | struct ifnet *ifp; |
---|
| 561 | struct ifreq ifr; |
---|
| 562 | struct mbuf *mb_rt; |
---|
| 563 | struct mbuf *m; |
---|
| 564 | struct rtdetq *rte; |
---|
| 565 | int s; |
---|
| 566 | |
---|
| 567 | s = splnet(); |
---|
| 568 | |
---|
| 569 | /* |
---|
| 570 | * For each phyint in use, disable promiscuous reception of all IP |
---|
| 571 | * multicasts. |
---|
| 572 | */ |
---|
| 573 | for (vifi = 0; vifi < numvifs; vifi++) { |
---|
| 574 | if (viftable[vifi].v_lcl_addr.s_addr != 0 && |
---|
| 575 | !(viftable[vifi].v_flags & VIFF_TUNNEL)) { |
---|
| 576 | ((struct sockaddr_in *)&(ifr.ifr_addr))->sin_family = AF_INET; |
---|
| 577 | ((struct sockaddr_in *)&(ifr.ifr_addr))->sin_addr.s_addr |
---|
| 578 | = INADDR_ANY; |
---|
| 579 | ifp = viftable[vifi].v_ifp; |
---|
| 580 | (*ifp->if_ioctl)(ifp, SIOCDELMULTI, (caddr_t)&ifr); |
---|
| 581 | } |
---|
| 582 | } |
---|
| 583 | bzero((caddr_t)tbftable, sizeof(tbftable)); |
---|
| 584 | bzero((caddr_t)viftable, sizeof(viftable)); |
---|
| 585 | numvifs = 0; |
---|
| 586 | pim_assert = 0; |
---|
| 587 | |
---|
| 588 | untimeout(expire_upcalls, (caddr_t)NULL); |
---|
| 589 | |
---|
| 590 | /* |
---|
| 591 | * Free all multicast forwarding cache entries. |
---|
| 592 | */ |
---|
| 593 | for (i = 0; i < MFCTBLSIZ; i++) { |
---|
| 594 | mb_rt = mfctable[i]; |
---|
| 595 | while (mb_rt) { |
---|
| 596 | if (mb_rt->m_act != NULL) { |
---|
| 597 | while (mb_rt->m_act) { |
---|
| 598 | m = mb_rt->m_act; |
---|
| 599 | mb_rt->m_act = m->m_act; |
---|
| 600 | rte = mtod(m, struct rtdetq *); |
---|
| 601 | m_freem(rte->m); |
---|
| 602 | m_free(m); |
---|
| 603 | } |
---|
| 604 | } |
---|
| 605 | mb_rt = m_free(mb_rt); |
---|
| 606 | } |
---|
| 607 | } |
---|
| 608 | |
---|
| 609 | bzero((caddr_t)mfctable, sizeof(mfctable)); |
---|
| 610 | |
---|
| 611 | /* |
---|
| 612 | * Reset de-encapsulation cache |
---|
| 613 | */ |
---|
| 614 | last_encap_src = 0; |
---|
| 615 | last_encap_vif = NULL; |
---|
| 616 | have_encap_tunnel = 0; |
---|
| 617 | |
---|
| 618 | ip_mrouter = NULL; |
---|
| 619 | |
---|
| 620 | splx(s); |
---|
| 621 | |
---|
| 622 | if (mrtdebug) |
---|
| 623 | log(LOG_DEBUG, "ip_mrouter_done\n"); |
---|
| 624 | |
---|
| 625 | return 0; |
---|
| 626 | } |
---|
| 627 | |
---|
| 628 | #ifndef MROUTE_LKM |
---|
| 629 | int (*ip_mrouter_done)(void) = X_ip_mrouter_done; |
---|
| 630 | #endif |
---|
| 631 | |
---|
| 632 | static int |
---|
| 633 | get_version(mb) |
---|
| 634 | struct mbuf *mb; |
---|
| 635 | { |
---|
| 636 | int *v; |
---|
| 637 | |
---|
| 638 | v = mtod(mb, int *); |
---|
| 639 | |
---|
| 640 | *v = 0x0305; /* XXX !!!! */ |
---|
| 641 | mb->m_len = sizeof(int); |
---|
| 642 | |
---|
| 643 | return 0; |
---|
| 644 | } |
---|
| 645 | |
---|
| 646 | /* |
---|
| 647 | * Set PIM assert processing global |
---|
| 648 | */ |
---|
| 649 | static int |
---|
| 650 | set_assert(i) |
---|
| 651 | int *i; |
---|
| 652 | { |
---|
| 653 | if ((*i != 1) && (*i != 0)) |
---|
| 654 | return EINVAL; |
---|
| 655 | |
---|
| 656 | pim_assert = *i; |
---|
| 657 | |
---|
| 658 | return 0; |
---|
| 659 | } |
---|
| 660 | |
---|
| 661 | /* |
---|
| 662 | * Get PIM assert processing global |
---|
| 663 | */ |
---|
| 664 | static int |
---|
| 665 | get_assert(m) |
---|
| 666 | struct mbuf *m; |
---|
| 667 | { |
---|
| 668 | int *i; |
---|
| 669 | |
---|
| 670 | i = mtod(m, int *); |
---|
| 671 | |
---|
| 672 | *i = pim_assert; |
---|
| 673 | |
---|
| 674 | return 0; |
---|
| 675 | } |
---|
| 676 | |
---|
| 677 | /* |
---|
| 678 | * Add a vif to the vif table |
---|
| 679 | */ |
---|
| 680 | static int |
---|
| 681 | add_vif(vifcp) |
---|
| 682 | register struct vifctl *vifcp; |
---|
| 683 | { |
---|
| 684 | register struct vif *vifp = viftable + vifcp->vifc_vifi; |
---|
| 685 | static struct sockaddr_in sin = {sizeof sin, AF_INET}; |
---|
| 686 | struct ifaddr *ifa; |
---|
| 687 | struct ifnet *ifp; |
---|
| 688 | struct ifreq ifr; |
---|
| 689 | int error, s; |
---|
| 690 | struct tbf *v_tbf = tbftable + vifcp->vifc_vifi; |
---|
| 691 | |
---|
| 692 | if (vifcp->vifc_vifi >= MAXVIFS) return EINVAL; |
---|
| 693 | if (vifp->v_lcl_addr.s_addr != 0) return EADDRINUSE; |
---|
| 694 | |
---|
| 695 | /* Find the interface with an address in AF_INET family */ |
---|
| 696 | sin.sin_addr = vifcp->vifc_lcl_addr; |
---|
| 697 | ifa = ifa_ifwithaddr((struct sockaddr *)&sin); |
---|
| 698 | if (ifa == 0) return EADDRNOTAVAIL; |
---|
| 699 | ifp = ifa->ifa_ifp; |
---|
| 700 | |
---|
| 701 | if (vifcp->vifc_flags & VIFF_TUNNEL) { |
---|
| 702 | if ((vifcp->vifc_flags & VIFF_SRCRT) == 0) { |
---|
| 703 | /* |
---|
| 704 | * An encapsulating tunnel is wanted. Tell ipip_input() to |
---|
| 705 | * start paying attention to encapsulated packets. |
---|
| 706 | */ |
---|
| 707 | if (have_encap_tunnel == 0) { |
---|
| 708 | have_encap_tunnel = 1; |
---|
| 709 | for (s = 0; s < MAXVIFS; ++s) { |
---|
| 710 | multicast_decap_if[s].if_name = "mdecap"; |
---|
| 711 | multicast_decap_if[s].if_unit = s; |
---|
| 712 | } |
---|
| 713 | } |
---|
| 714 | /* |
---|
| 715 | * Set interface to fake encapsulator interface |
---|
| 716 | */ |
---|
| 717 | ifp = &multicast_decap_if[vifcp->vifc_vifi]; |
---|
| 718 | /* |
---|
| 719 | * Prepare cached route entry |
---|
| 720 | */ |
---|
| 721 | bzero(&vifp->v_route, sizeof(vifp->v_route)); |
---|
| 722 | } else { |
---|
| 723 | log(LOG_ERR, "source routed tunnels not supported\n"); |
---|
| 724 | return EOPNOTSUPP; |
---|
| 725 | } |
---|
| 726 | } else { |
---|
| 727 | /* Make sure the interface supports multicast */ |
---|
| 728 | if ((ifp->if_flags & IFF_MULTICAST) == 0) |
---|
| 729 | return EOPNOTSUPP; |
---|
| 730 | |
---|
| 731 | /* Enable promiscuous reception of all IP multicasts from the if */ |
---|
| 732 | ((struct sockaddr_in *)&(ifr.ifr_addr))->sin_family = AF_INET; |
---|
| 733 | ((struct sockaddr_in *)&(ifr.ifr_addr))->sin_addr.s_addr = INADDR_ANY; |
---|
| 734 | s = splnet(); |
---|
| 735 | error = (*ifp->if_ioctl)(ifp, SIOCADDMULTI, (caddr_t)&ifr); |
---|
| 736 | splx(s); |
---|
| 737 | if (error) |
---|
| 738 | return error; |
---|
| 739 | } |
---|
| 740 | |
---|
| 741 | s = splnet(); |
---|
| 742 | /* define parameters for the tbf structure */ |
---|
| 743 | vifp->v_tbf = v_tbf; |
---|
| 744 | GET_TIME(vifp->v_tbf->tbf_last_pkt_t); |
---|
| 745 | vifp->v_tbf->tbf_n_tok = 0; |
---|
| 746 | vifp->v_tbf->tbf_q_len = 0; |
---|
| 747 | vifp->v_tbf->tbf_max_q_len = MAXQSIZE; |
---|
| 748 | vifp->v_tbf->tbf_q = vifp->v_tbf->tbf_t = NULL; |
---|
| 749 | |
---|
| 750 | vifp->v_flags = vifcp->vifc_flags; |
---|
| 751 | vifp->v_threshold = vifcp->vifc_threshold; |
---|
| 752 | vifp->v_lcl_addr = vifcp->vifc_lcl_addr; |
---|
| 753 | vifp->v_rmt_addr = vifcp->vifc_rmt_addr; |
---|
| 754 | vifp->v_ifp = ifp; |
---|
| 755 | /* scaling up here allows division by 1024 in critical code */ |
---|
| 756 | vifp->v_rate_limit= vifcp->vifc_rate_limit * 1024 / 1000; |
---|
| 757 | vifp->v_rsvp_on = 0; |
---|
| 758 | vifp->v_rsvpd = NULL; |
---|
| 759 | /* initialize per vif pkt counters */ |
---|
| 760 | vifp->v_pkt_in = 0; |
---|
| 761 | vifp->v_pkt_out = 0; |
---|
| 762 | vifp->v_bytes_in = 0; |
---|
| 763 | vifp->v_bytes_out = 0; |
---|
| 764 | splx(s); |
---|
| 765 | |
---|
| 766 | /* Adjust numvifs up if the vifi is higher than numvifs */ |
---|
| 767 | if (numvifs <= vifcp->vifc_vifi) numvifs = vifcp->vifc_vifi + 1; |
---|
| 768 | |
---|
| 769 | if (mrtdebug) |
---|
| 770 | log(LOG_DEBUG, "add_vif #%d, lcladdr %x, %s %x, thresh %x, rate %d\n", |
---|
| 771 | vifcp->vifc_vifi, |
---|
| 772 | ntohl(vifcp->vifc_lcl_addr.s_addr), |
---|
| 773 | (vifcp->vifc_flags & VIFF_TUNNEL) ? "rmtaddr" : "mask", |
---|
| 774 | ntohl(vifcp->vifc_rmt_addr.s_addr), |
---|
| 775 | vifcp->vifc_threshold, |
---|
| 776 | vifcp->vifc_rate_limit); |
---|
| 777 | |
---|
| 778 | return 0; |
---|
| 779 | } |
---|
| 780 | |
---|
| 781 | /* |
---|
| 782 | * Delete a vif from the vif table |
---|
| 783 | */ |
---|
| 784 | static int |
---|
| 785 | del_vif(vifip) |
---|
| 786 | vifi_t *vifip; |
---|
| 787 | { |
---|
| 788 | register struct vif *vifp = viftable + *vifip; |
---|
| 789 | register vifi_t vifi; |
---|
| 790 | register struct mbuf *m; |
---|
| 791 | struct ifnet *ifp; |
---|
| 792 | struct ifreq ifr; |
---|
| 793 | int s; |
---|
| 794 | |
---|
| 795 | if (*vifip >= numvifs) return EINVAL; |
---|
| 796 | if (vifp->v_lcl_addr.s_addr == 0) return EADDRNOTAVAIL; |
---|
| 797 | |
---|
| 798 | s = splnet(); |
---|
| 799 | |
---|
| 800 | if (!(vifp->v_flags & VIFF_TUNNEL)) { |
---|
| 801 | ((struct sockaddr_in *)&(ifr.ifr_addr))->sin_family = AF_INET; |
---|
| 802 | ((struct sockaddr_in *)&(ifr.ifr_addr))->sin_addr.s_addr = INADDR_ANY; |
---|
| 803 | ifp = vifp->v_ifp; |
---|
| 804 | (*ifp->if_ioctl)(ifp, SIOCDELMULTI, (caddr_t)&ifr); |
---|
| 805 | } |
---|
| 806 | |
---|
| 807 | if (vifp == last_encap_vif) { |
---|
| 808 | last_encap_vif = 0; |
---|
| 809 | last_encap_src = 0; |
---|
| 810 | } |
---|
| 811 | |
---|
| 812 | /* |
---|
| 813 | * Free packets queued at the interface |
---|
| 814 | */ |
---|
| 815 | while (vifp->v_tbf->tbf_q) { |
---|
| 816 | m = vifp->v_tbf->tbf_q; |
---|
| 817 | vifp->v_tbf->tbf_q = m->m_act; |
---|
| 818 | m_freem(m); |
---|
| 819 | } |
---|
| 820 | |
---|
| 821 | bzero((caddr_t)vifp->v_tbf, sizeof(*(vifp->v_tbf))); |
---|
| 822 | bzero((caddr_t)vifp, sizeof (*vifp)); |
---|
| 823 | |
---|
| 824 | /* Adjust numvifs down */ |
---|
| 825 | for (vifi = numvifs; vifi > 0; vifi--) |
---|
| 826 | if (viftable[vifi-1].v_lcl_addr.s_addr != 0) break; |
---|
| 827 | numvifs = vifi; |
---|
| 828 | |
---|
| 829 | splx(s); |
---|
| 830 | |
---|
| 831 | if (mrtdebug) |
---|
| 832 | log(LOG_DEBUG, "del_vif %d, numvifs %d\n", *vifip, numvifs); |
---|
| 833 | |
---|
| 834 | return 0; |
---|
| 835 | } |
---|
| 836 | |
---|
| 837 | /* |
---|
| 838 | * Add an mfc entry |
---|
| 839 | */ |
---|
| 840 | static int |
---|
| 841 | add_mfc(mfccp) |
---|
| 842 | struct mfcctl *mfccp; |
---|
| 843 | { |
---|
| 844 | struct mfc *rt; |
---|
| 845 | register struct mbuf *mb_rt; |
---|
| 846 | u_long hash; |
---|
| 847 | struct mbuf *mb_ntry; |
---|
| 848 | struct rtdetq *rte; |
---|
| 849 | register u_short nstl; |
---|
| 850 | int s; |
---|
| 851 | int i; |
---|
| 852 | |
---|
| 853 | MFCFIND(mfccp->mfcc_origin.s_addr, mfccp->mfcc_mcastgrp.s_addr, rt); |
---|
| 854 | |
---|
| 855 | /* If an entry already exists, just update the fields */ |
---|
| 856 | if (rt) { |
---|
| 857 | if (mrtdebug & DEBUG_MFC) |
---|
| 858 | log(LOG_DEBUG,"add_mfc update o %x g %x p %x\n", |
---|
| 859 | ntohl(mfccp->mfcc_origin.s_addr), |
---|
| 860 | ntohl(mfccp->mfcc_mcastgrp.s_addr), |
---|
| 861 | mfccp->mfcc_parent); |
---|
| 862 | |
---|
| 863 | s = splnet(); |
---|
| 864 | rt->mfc_parent = mfccp->mfcc_parent; |
---|
| 865 | for (i = 0; i < numvifs; i++) |
---|
| 866 | rt->mfc_ttls[i] = mfccp->mfcc_ttls[i]; |
---|
| 867 | splx(s); |
---|
| 868 | return 0; |
---|
| 869 | } |
---|
| 870 | |
---|
| 871 | /* |
---|
| 872 | * Find the entry for which the upcall was made and update |
---|
| 873 | */ |
---|
| 874 | s = splnet(); |
---|
| 875 | hash = MFCHASH(mfccp->mfcc_origin.s_addr, mfccp->mfcc_mcastgrp.s_addr); |
---|
| 876 | for (mb_rt = mfctable[hash], nstl = 0; mb_rt; mb_rt = mb_rt->m_next) { |
---|
| 877 | |
---|
| 878 | rt = mtod(mb_rt, struct mfc *); |
---|
| 879 | if ((rt->mfc_origin.s_addr == mfccp->mfcc_origin.s_addr) && |
---|
| 880 | (rt->mfc_mcastgrp.s_addr == mfccp->mfcc_mcastgrp.s_addr) && |
---|
| 881 | (mb_rt->m_act != NULL)) { |
---|
| 882 | |
---|
| 883 | if (nstl++) |
---|
| 884 | log(LOG_ERR, "add_mfc %s o %x g %x p %x dbx %x\n", |
---|
| 885 | "multiple kernel entries", |
---|
| 886 | ntohl(mfccp->mfcc_origin.s_addr), |
---|
| 887 | ntohl(mfccp->mfcc_mcastgrp.s_addr), |
---|
| 888 | mfccp->mfcc_parent, mb_rt->m_act); |
---|
| 889 | |
---|
| 890 | if (mrtdebug & DEBUG_MFC) |
---|
| 891 | log(LOG_DEBUG,"add_mfc o %x g %x p %x dbg %x\n", |
---|
| 892 | ntohl(mfccp->mfcc_origin.s_addr), |
---|
| 893 | ntohl(mfccp->mfcc_mcastgrp.s_addr), |
---|
| 894 | mfccp->mfcc_parent, mb_rt->m_act); |
---|
| 895 | |
---|
| 896 | rt->mfc_origin = mfccp->mfcc_origin; |
---|
| 897 | rt->mfc_mcastgrp = mfccp->mfcc_mcastgrp; |
---|
| 898 | rt->mfc_parent = mfccp->mfcc_parent; |
---|
| 899 | for (i = 0; i < numvifs; i++) |
---|
| 900 | rt->mfc_ttls[i] = mfccp->mfcc_ttls[i]; |
---|
| 901 | /* initialize pkt counters per src-grp */ |
---|
| 902 | rt->mfc_pkt_cnt = 0; |
---|
| 903 | rt->mfc_byte_cnt = 0; |
---|
| 904 | rt->mfc_wrong_if = 0; |
---|
| 905 | rt->mfc_last_assert.tv_sec = rt->mfc_last_assert.tv_usec = 0; |
---|
| 906 | |
---|
| 907 | rt->mfc_expire = 0; /* Don't clean this guy up */ |
---|
| 908 | nexpire[hash]--; |
---|
| 909 | |
---|
| 910 | /* free packets Qed at the end of this entry */ |
---|
| 911 | while (mb_rt->m_act) { |
---|
| 912 | mb_ntry = mb_rt->m_act; |
---|
| 913 | rte = mtod(mb_ntry, struct rtdetq *); |
---|
| 914 | /* #ifdef RSVP_ISI */ |
---|
| 915 | ip_mdq(rte->m, rte->ifp, rt, -1); |
---|
| 916 | /* #endif */ |
---|
| 917 | mb_rt->m_act = mb_ntry->m_act; |
---|
| 918 | m_freem(rte->m); |
---|
| 919 | #ifdef UPCALL_TIMING |
---|
| 920 | collate(&(rte->t)); |
---|
| 921 | #endif /* UPCALL_TIMING */ |
---|
| 922 | m_free(mb_ntry); |
---|
| 923 | } |
---|
| 924 | } |
---|
| 925 | } |
---|
| 926 | |
---|
| 927 | /* |
---|
| 928 | * It is possible that an entry is being inserted without an upcall |
---|
| 929 | */ |
---|
| 930 | if (nstl == 0) { |
---|
| 931 | if (mrtdebug & DEBUG_MFC) |
---|
| 932 | log(LOG_DEBUG,"add_mfc no upcall h %d o %x g %x p %x\n", |
---|
| 933 | hash, ntohl(mfccp->mfcc_origin.s_addr), |
---|
| 934 | ntohl(mfccp->mfcc_mcastgrp.s_addr), |
---|
| 935 | mfccp->mfcc_parent); |
---|
| 936 | |
---|
| 937 | for (mb_rt = mfctable[hash]; mb_rt; mb_rt = mb_rt->m_next) { |
---|
| 938 | |
---|
| 939 | rt = mtod(mb_rt, struct mfc *); |
---|
| 940 | if ((rt->mfc_origin.s_addr == mfccp->mfcc_origin.s_addr) && |
---|
| 941 | (rt->mfc_mcastgrp.s_addr == mfccp->mfcc_mcastgrp.s_addr)) { |
---|
| 942 | |
---|
| 943 | rt->mfc_origin = mfccp->mfcc_origin; |
---|
| 944 | rt->mfc_mcastgrp = mfccp->mfcc_mcastgrp; |
---|
| 945 | rt->mfc_parent = mfccp->mfcc_parent; |
---|
| 946 | for (i = 0; i < numvifs; i++) |
---|
| 947 | rt->mfc_ttls[i] = mfccp->mfcc_ttls[i]; |
---|
| 948 | /* initialize pkt counters per src-grp */ |
---|
| 949 | rt->mfc_pkt_cnt = 0; |
---|
| 950 | rt->mfc_byte_cnt = 0; |
---|
| 951 | rt->mfc_wrong_if = 0; |
---|
| 952 | rt->mfc_last_assert.tv_sec = rt->mfc_last_assert.tv_usec = 0; |
---|
| 953 | if (rt->mfc_expire) |
---|
| 954 | nexpire[hash]--; |
---|
| 955 | rt->mfc_expire = 0; |
---|
| 956 | } |
---|
| 957 | } |
---|
| 958 | if (mb_rt == NULL) { |
---|
| 959 | /* no upcall, so make a new entry */ |
---|
| 960 | MGET(mb_rt, M_DONTWAIT, MT_MRTABLE); |
---|
| 961 | if (mb_rt == NULL) { |
---|
| 962 | splx(s); |
---|
| 963 | return ENOBUFS; |
---|
| 964 | } |
---|
| 965 | |
---|
| 966 | rt = mtod(mb_rt, struct mfc *); |
---|
| 967 | |
---|
| 968 | /* insert new entry at head of hash chain */ |
---|
| 969 | rt->mfc_origin = mfccp->mfcc_origin; |
---|
| 970 | rt->mfc_mcastgrp = mfccp->mfcc_mcastgrp; |
---|
| 971 | rt->mfc_parent = mfccp->mfcc_parent; |
---|
| 972 | for (i = 0; i < numvifs; i++) |
---|
| 973 | rt->mfc_ttls[i] = mfccp->mfcc_ttls[i]; |
---|
| 974 | /* initialize pkt counters per src-grp */ |
---|
| 975 | rt->mfc_pkt_cnt = 0; |
---|
| 976 | rt->mfc_byte_cnt = 0; |
---|
| 977 | rt->mfc_wrong_if = 0; |
---|
| 978 | rt->mfc_last_assert.tv_sec = rt->mfc_last_assert.tv_usec = 0; |
---|
| 979 | rt->mfc_expire = 0; |
---|
| 980 | |
---|
| 981 | /* link into table */ |
---|
| 982 | mb_rt->m_next = mfctable[hash]; |
---|
| 983 | mfctable[hash] = mb_rt; |
---|
| 984 | mb_rt->m_act = NULL; |
---|
| 985 | } |
---|
| 986 | } |
---|
| 987 | splx(s); |
---|
| 988 | return 0; |
---|
| 989 | } |
---|
| 990 | |
---|
| 991 | #ifdef UPCALL_TIMING |
---|
| 992 | /* |
---|
| 993 | * collect delay statistics on the upcalls |
---|
| 994 | */ |
---|
| 995 | static void collate(t) |
---|
| 996 | register struct timeval *t; |
---|
| 997 | { |
---|
| 998 | register u_long d; |
---|
| 999 | register struct timeval tp; |
---|
| 1000 | register u_long delta; |
---|
| 1001 | |
---|
| 1002 | GET_TIME(tp); |
---|
| 1003 | |
---|
| 1004 | if (TV_LT(*t, tp)) |
---|
| 1005 | { |
---|
| 1006 | TV_DELTA(tp, *t, delta); |
---|
| 1007 | |
---|
| 1008 | d = delta >> 10; |
---|
| 1009 | if (d > 50) |
---|
| 1010 | d = 50; |
---|
| 1011 | |
---|
| 1012 | ++upcall_data[d]; |
---|
| 1013 | } |
---|
| 1014 | } |
---|
| 1015 | #endif /* UPCALL_TIMING */ |
---|
| 1016 | |
---|
| 1017 | /* |
---|
| 1018 | * Delete an mfc entry |
---|
| 1019 | */ |
---|
| 1020 | static int |
---|
| 1021 | del_mfc(mfccp) |
---|
| 1022 | struct mfcctl *mfccp; |
---|
| 1023 | { |
---|
| 1024 | struct in_addr origin; |
---|
| 1025 | struct in_addr mcastgrp; |
---|
| 1026 | struct mfc *rt; |
---|
| 1027 | struct mbuf *mb_rt; |
---|
| 1028 | struct mbuf **nptr; |
---|
| 1029 | u_long hash; |
---|
| 1030 | int s; |
---|
| 1031 | |
---|
| 1032 | origin = mfccp->mfcc_origin; |
---|
| 1033 | mcastgrp = mfccp->mfcc_mcastgrp; |
---|
| 1034 | hash = MFCHASH(origin.s_addr, mcastgrp.s_addr); |
---|
| 1035 | |
---|
| 1036 | if (mrtdebug & DEBUG_MFC) |
---|
| 1037 | log(LOG_DEBUG,"del_mfc orig %x mcastgrp %x\n", |
---|
| 1038 | ntohl(origin.s_addr), ntohl(mcastgrp.s_addr)); |
---|
| 1039 | |
---|
| 1040 | s = splnet(); |
---|
| 1041 | |
---|
| 1042 | nptr = &mfctable[hash]; |
---|
| 1043 | while ((mb_rt = *nptr) != NULL) { |
---|
| 1044 | rt = mtod(mb_rt, struct mfc *); |
---|
| 1045 | if (origin.s_addr == rt->mfc_origin.s_addr && |
---|
| 1046 | mcastgrp.s_addr == rt->mfc_mcastgrp.s_addr && |
---|
| 1047 | mb_rt->m_act == NULL) |
---|
| 1048 | break; |
---|
| 1049 | |
---|
| 1050 | nptr = &mb_rt->m_next; |
---|
| 1051 | } |
---|
| 1052 | if (mb_rt == NULL) { |
---|
| 1053 | splx(s); |
---|
| 1054 | return EADDRNOTAVAIL; |
---|
| 1055 | } |
---|
| 1056 | |
---|
| 1057 | MFREE(mb_rt, *nptr); |
---|
| 1058 | |
---|
| 1059 | splx(s); |
---|
| 1060 | |
---|
| 1061 | return 0; |
---|
| 1062 | } |
---|
| 1063 | |
---|
| 1064 | /* |
---|
| 1065 | * Send a message to mrouted on the multicast routing socket |
---|
| 1066 | */ |
---|
| 1067 | static int |
---|
| 1068 | socket_send(s, mm, src) |
---|
| 1069 | struct socket *s; |
---|
| 1070 | struct mbuf *mm; |
---|
| 1071 | struct sockaddr_in *src; |
---|
| 1072 | { |
---|
| 1073 | if (s) { |
---|
| 1074 | if (sbappendaddr(&s->so_rcv, |
---|
| 1075 | (struct sockaddr *)src, |
---|
| 1076 | mm, (struct mbuf *)0) != 0) { |
---|
| 1077 | sorwakeup(s); |
---|
| 1078 | return 0; |
---|
| 1079 | } |
---|
| 1080 | } |
---|
| 1081 | m_freem(mm); |
---|
| 1082 | return -1; |
---|
| 1083 | } |
---|
| 1084 | |
---|
| 1085 | /* |
---|
| 1086 | * IP multicast forwarding function. This function assumes that the packet |
---|
| 1087 | * pointed to by "ip" has arrived on (or is about to be sent to) the interface |
---|
| 1088 | * pointed to by "ifp", and the packet is to be relayed to other networks |
---|
| 1089 | * that have members of the packet's destination IP multicast group. |
---|
| 1090 | * |
---|
| 1091 | * The packet is returned unscathed to the caller, unless it is |
---|
| 1092 | * erroneous, in which case a non-zero return value tells the caller to |
---|
| 1093 | * discard it. |
---|
| 1094 | */ |
---|
| 1095 | |
---|
| 1096 | #define IP_HDR_LEN 20 /* # bytes of fixed IP header (excluding options) */ |
---|
| 1097 | #define TUNNEL_LEN 12 /* # bytes of IP option for tunnel encapsulation */ |
---|
| 1098 | |
---|
| 1099 | static int |
---|
| 1100 | X_ip_mforward(ip, ifp, m, imo) |
---|
| 1101 | register struct ip *ip; |
---|
| 1102 | struct ifnet *ifp; |
---|
| 1103 | struct mbuf *m; |
---|
| 1104 | struct ip_moptions *imo; |
---|
| 1105 | { |
---|
| 1106 | register struct mfc *rt; |
---|
| 1107 | register u_char *ipoptions; |
---|
| 1108 | static struct sockaddr_in k_igmpsrc = { sizeof k_igmpsrc, AF_INET }; |
---|
| 1109 | static int srctun = 0; |
---|
| 1110 | register struct mbuf *mm; |
---|
| 1111 | int s; |
---|
| 1112 | vifi_t vifi; |
---|
| 1113 | struct vif *vifp; |
---|
| 1114 | |
---|
| 1115 | if (mrtdebug & DEBUG_FORWARD) |
---|
| 1116 | log(LOG_DEBUG, "ip_mforward: src %x, dst %x, ifp %x\n", |
---|
| 1117 | ntohl(ip->ip_src.s_addr), ntohl(ip->ip_dst.s_addr), ifp); |
---|
| 1118 | |
---|
| 1119 | if (ip->ip_hl < (IP_HDR_LEN + TUNNEL_LEN) >> 2 || |
---|
| 1120 | (ipoptions = (u_char *)(ip + 1))[1] != IPOPT_LSRR ) { |
---|
| 1121 | /* |
---|
| 1122 | * Packet arrived via a physical interface or |
---|
| 1123 | * an encapsulated tunnel. |
---|
| 1124 | */ |
---|
| 1125 | } else { |
---|
| 1126 | /* |
---|
| 1127 | * Packet arrived through a source-route tunnel. |
---|
| 1128 | * Source-route tunnels are no longer supported. |
---|
| 1129 | */ |
---|
| 1130 | if ((srctun++ % 1000) == 0) |
---|
| 1131 | log(LOG_ERR, "ip_mforward: received source-routed packet from %x\n", |
---|
| 1132 | ntohl(ip->ip_src.s_addr)); |
---|
| 1133 | |
---|
| 1134 | return 1; |
---|
| 1135 | } |
---|
| 1136 | |
---|
| 1137 | if ((imo) && ((vifi = imo->imo_multicast_vif) < numvifs)) { |
---|
| 1138 | if (ip->ip_ttl < 255) |
---|
| 1139 | ip->ip_ttl++; /* compensate for -1 in *_send routines */ |
---|
| 1140 | if (rsvpdebug && ip->ip_p == IPPROTO_RSVP) { |
---|
| 1141 | vifp = viftable + vifi; |
---|
| 1142 | printf("Sending IPPROTO_RSVP from %lx to %lx on vif %d (%s%s%d)\n", |
---|
| 1143 | ntohl(ip->ip_src.s_addr), ntohl(ip->ip_dst.s_addr), vifi, |
---|
| 1144 | (vifp->v_flags & VIFF_TUNNEL) ? "tunnel on " : "", |
---|
| 1145 | vifp->v_ifp->if_name, vifp->v_ifp->if_unit); |
---|
| 1146 | } |
---|
| 1147 | return (ip_mdq(m, ifp, NULL, vifi)); |
---|
| 1148 | } |
---|
| 1149 | if (rsvpdebug && ip->ip_p == IPPROTO_RSVP) { |
---|
| 1150 | printf("Warning: IPPROTO_RSVP from %lx to %lx without vif option\n", |
---|
| 1151 | ntohl(ip->ip_src.s_addr), ntohl(ip->ip_dst.s_addr)); |
---|
| 1152 | if(!imo) |
---|
| 1153 | printf("In fact, no options were specified at all\n"); |
---|
| 1154 | } |
---|
| 1155 | |
---|
| 1156 | /* |
---|
| 1157 | * Don't forward a packet with time-to-live of zero or one, |
---|
| 1158 | * or a packet destined to a local-only group. |
---|
| 1159 | */ |
---|
| 1160 | if (ip->ip_ttl <= 1 || |
---|
| 1161 | ntohl(ip->ip_dst.s_addr) <= INADDR_MAX_LOCAL_GROUP) |
---|
| 1162 | return 0; |
---|
| 1163 | |
---|
| 1164 | /* |
---|
| 1165 | * Determine forwarding vifs from the forwarding cache table |
---|
| 1166 | */ |
---|
| 1167 | s = splnet(); |
---|
| 1168 | MFCFIND(ip->ip_src.s_addr, ip->ip_dst.s_addr, rt); |
---|
| 1169 | |
---|
| 1170 | /* Entry exists, so forward if necessary */ |
---|
| 1171 | if (rt != NULL) { |
---|
| 1172 | splx(s); |
---|
| 1173 | return (ip_mdq(m, ifp, rt, -1)); |
---|
| 1174 | } else { |
---|
| 1175 | /* |
---|
| 1176 | * If we don't have a route for packet's origin, |
---|
| 1177 | * Make a copy of the packet & |
---|
| 1178 | * send message to routing daemon |
---|
| 1179 | */ |
---|
| 1180 | |
---|
| 1181 | register struct mbuf *mb_rt; |
---|
| 1182 | register struct mbuf *mb_ntry; |
---|
| 1183 | register struct mbuf *mb0; |
---|
| 1184 | register struct rtdetq *rte; |
---|
| 1185 | register struct mbuf *rte_m; |
---|
| 1186 | register u_long hash; |
---|
| 1187 | register int npkts; |
---|
| 1188 | int hlen = ip->ip_hl << 2; |
---|
| 1189 | #ifdef UPCALL_TIMING |
---|
| 1190 | struct timeval tp; |
---|
| 1191 | |
---|
| 1192 | GET_TIME(tp); |
---|
| 1193 | #endif |
---|
| 1194 | |
---|
| 1195 | mrtstat.mrts_no_route++; |
---|
| 1196 | if (mrtdebug & (DEBUG_FORWARD | DEBUG_MFC)) |
---|
| 1197 | log(LOG_DEBUG, "ip_mforward: no rte s %x g %x\n", |
---|
| 1198 | ntohl(ip->ip_src.s_addr), |
---|
| 1199 | ntohl(ip->ip_dst.s_addr)); |
---|
| 1200 | |
---|
| 1201 | /* |
---|
| 1202 | * Allocate mbufs early so that we don't do extra work if we are |
---|
| 1203 | * just going to fail anyway. Make sure to pullup the header so |
---|
| 1204 | * that other people can't step on it. |
---|
| 1205 | */ |
---|
| 1206 | MGET(mb_ntry, M_DONTWAIT, MT_DATA); |
---|
| 1207 | if (mb_ntry == NULL) { |
---|
| 1208 | splx(s); |
---|
| 1209 | return ENOBUFS; |
---|
| 1210 | } |
---|
| 1211 | mb0 = m_copy(m, 0, M_COPYALL); |
---|
| 1212 | if (mb0 && (M_HASCL(mb0) || mb0->m_len < hlen)) |
---|
| 1213 | mb0 = m_pullup(mb0, hlen); |
---|
| 1214 | if (mb0 == NULL) { |
---|
| 1215 | m_free(mb_ntry); |
---|
| 1216 | splx(s); |
---|
| 1217 | return ENOBUFS; |
---|
| 1218 | } |
---|
| 1219 | |
---|
| 1220 | /* is there an upcall waiting for this packet? */ |
---|
| 1221 | hash = MFCHASH(ip->ip_src.s_addr, ip->ip_dst.s_addr); |
---|
| 1222 | for (mb_rt = mfctable[hash]; mb_rt; mb_rt = mb_rt->m_next) { |
---|
| 1223 | rt = mtod(mb_rt, struct mfc *); |
---|
| 1224 | if ((ip->ip_src.s_addr == rt->mfc_origin.s_addr) && |
---|
| 1225 | (ip->ip_dst.s_addr == rt->mfc_mcastgrp.s_addr) && |
---|
| 1226 | (mb_rt->m_act != NULL)) |
---|
| 1227 | break; |
---|
| 1228 | } |
---|
| 1229 | |
---|
| 1230 | if (mb_rt == NULL) { |
---|
| 1231 | int i; |
---|
| 1232 | struct igmpmsg *im; |
---|
| 1233 | |
---|
| 1234 | /* no upcall, so make a new entry */ |
---|
| 1235 | MGET(mb_rt, M_DONTWAIT, MT_MRTABLE); |
---|
| 1236 | if (mb_rt == NULL) { |
---|
| 1237 | m_free(mb_ntry); |
---|
| 1238 | m_freem(mb0); |
---|
| 1239 | splx(s); |
---|
| 1240 | return ENOBUFS; |
---|
| 1241 | } |
---|
| 1242 | /* Make a copy of the header to send to the user level process */ |
---|
| 1243 | mm = m_copy(mb0, 0, hlen); |
---|
| 1244 | if (mm == NULL) { |
---|
| 1245 | m_free(mb_ntry); |
---|
| 1246 | m_freem(mb0); |
---|
| 1247 | m_free(mb_rt); |
---|
| 1248 | splx(s); |
---|
| 1249 | return ENOBUFS; |
---|
| 1250 | } |
---|
| 1251 | |
---|
| 1252 | /* |
---|
| 1253 | * Send message to routing daemon to install |
---|
| 1254 | * a route into the kernel table |
---|
| 1255 | */ |
---|
| 1256 | k_igmpsrc.sin_addr = ip->ip_src; |
---|
| 1257 | |
---|
| 1258 | im = mtod(mm, struct igmpmsg *); |
---|
| 1259 | im->im_msgtype = IGMPMSG_NOCACHE; |
---|
| 1260 | im->im_mbz = 0; |
---|
| 1261 | |
---|
| 1262 | mrtstat.mrts_upcalls++; |
---|
| 1263 | |
---|
| 1264 | if (socket_send(ip_mrouter, mm, &k_igmpsrc) < 0) { |
---|
| 1265 | log(LOG_WARNING, "ip_mforward: ip_mrouter socket queue full\n"); |
---|
| 1266 | ++mrtstat.mrts_upq_sockfull; |
---|
| 1267 | m_free(mb_ntry); |
---|
| 1268 | m_freem(mb0); |
---|
| 1269 | m_free(mb_rt); |
---|
| 1270 | splx(s); |
---|
| 1271 | return ENOBUFS; |
---|
| 1272 | } |
---|
| 1273 | |
---|
| 1274 | rt = mtod(mb_rt, struct mfc *); |
---|
| 1275 | |
---|
| 1276 | /* insert new entry at head of hash chain */ |
---|
| 1277 | rt->mfc_origin.s_addr = ip->ip_src.s_addr; |
---|
| 1278 | rt->mfc_mcastgrp.s_addr = ip->ip_dst.s_addr; |
---|
| 1279 | rt->mfc_expire = UPCALL_EXPIRE; |
---|
| 1280 | nexpire[hash]++; |
---|
| 1281 | for (i = 0; i < numvifs; i++) |
---|
| 1282 | rt->mfc_ttls[i] = 0; |
---|
| 1283 | rt->mfc_parent = -1; |
---|
| 1284 | |
---|
| 1285 | /* link into table */ |
---|
| 1286 | mb_rt->m_next = mfctable[hash]; |
---|
| 1287 | mfctable[hash] = mb_rt; |
---|
| 1288 | mb_rt->m_act = NULL; |
---|
| 1289 | |
---|
| 1290 | rte_m = mb_rt; |
---|
| 1291 | } else { |
---|
| 1292 | /* determine if q has overflowed */ |
---|
| 1293 | for (rte_m = mb_rt, npkts = 0; rte_m->m_act; rte_m = rte_m->m_act) |
---|
| 1294 | npkts++; |
---|
| 1295 | |
---|
| 1296 | if (npkts > MAX_UPQ) { |
---|
| 1297 | mrtstat.mrts_upq_ovflw++; |
---|
| 1298 | m_free(mb_ntry); |
---|
| 1299 | m_freem(mb0); |
---|
| 1300 | splx(s); |
---|
| 1301 | return 0; |
---|
| 1302 | } |
---|
| 1303 | } |
---|
| 1304 | |
---|
| 1305 | mb_ntry->m_act = NULL; |
---|
| 1306 | rte = mtod(mb_ntry, struct rtdetq *); |
---|
| 1307 | |
---|
| 1308 | rte->m = mb0; |
---|
| 1309 | rte->ifp = ifp; |
---|
| 1310 | #ifdef UPCALL_TIMING |
---|
| 1311 | rte->t = tp; |
---|
| 1312 | #endif |
---|
| 1313 | |
---|
| 1314 | /* Add this entry to the end of the queue */ |
---|
| 1315 | rte_m->m_act = mb_ntry; |
---|
| 1316 | |
---|
| 1317 | splx(s); |
---|
| 1318 | |
---|
| 1319 | return 0; |
---|
| 1320 | } |
---|
| 1321 | } |
---|
| 1322 | |
---|
| 1323 | #ifndef MROUTE_LKM |
---|
| 1324 | int (*ip_mforward)(struct ip *, struct ifnet *, struct mbuf *, |
---|
| 1325 | struct ip_moptions *) = X_ip_mforward; |
---|
| 1326 | #endif |
---|
| 1327 | |
---|
| 1328 | /* |
---|
| 1329 | * Clean up the cache entry if upcall is not serviced |
---|
| 1330 | */ |
---|
| 1331 | static void |
---|
| 1332 | expire_upcalls(void *unused) |
---|
| 1333 | { |
---|
| 1334 | struct mbuf *mb_rt, *m, **nptr; |
---|
| 1335 | struct rtdetq *rte; |
---|
| 1336 | struct mfc *mfc; |
---|
| 1337 | int i; |
---|
| 1338 | int s; |
---|
| 1339 | |
---|
| 1340 | s = splnet(); |
---|
| 1341 | for (i = 0; i < MFCTBLSIZ; i++) { |
---|
| 1342 | if (nexpire[i] == 0) |
---|
| 1343 | continue; |
---|
| 1344 | nptr = &mfctable[i]; |
---|
| 1345 | for (mb_rt = *nptr; mb_rt != NULL; mb_rt = *nptr) { |
---|
| 1346 | mfc = mtod(mb_rt, struct mfc *); |
---|
| 1347 | |
---|
| 1348 | /* |
---|
| 1349 | * Skip real cache entries |
---|
| 1350 | * Make sure it wasn't marked to not expire (shouldn't happen) |
---|
| 1351 | * If it expires now |
---|
| 1352 | */ |
---|
| 1353 | if (mb_rt->m_act != NULL && |
---|
| 1354 | mfc->mfc_expire != 0 && |
---|
| 1355 | --mfc->mfc_expire == 0) { |
---|
| 1356 | if (mrtdebug & DEBUG_EXPIRE) |
---|
| 1357 | log(LOG_DEBUG, "expire_upcalls: expiring (%x %x)\n", |
---|
| 1358 | ntohl(mfc->mfc_origin.s_addr), |
---|
| 1359 | ntohl(mfc->mfc_mcastgrp.s_addr)); |
---|
| 1360 | /* |
---|
| 1361 | * drop all the packets |
---|
| 1362 | * free the mbuf with the pkt, if, timing info |
---|
| 1363 | */ |
---|
| 1364 | while (mb_rt->m_act) { |
---|
| 1365 | m = mb_rt->m_act; |
---|
| 1366 | mb_rt->m_act = m->m_act; |
---|
| 1367 | |
---|
| 1368 | rte = mtod(m, struct rtdetq *); |
---|
| 1369 | m_freem(rte->m); |
---|
| 1370 | m_free(m); |
---|
| 1371 | } |
---|
| 1372 | ++mrtstat.mrts_cache_cleanups; |
---|
| 1373 | nexpire[i]--; |
---|
| 1374 | |
---|
| 1375 | MFREE(mb_rt, *nptr); |
---|
| 1376 | } else { |
---|
| 1377 | nptr = &mb_rt->m_next; |
---|
| 1378 | } |
---|
| 1379 | } |
---|
| 1380 | } |
---|
| 1381 | splx(s); |
---|
| 1382 | timeout(expire_upcalls, (caddr_t)NULL, EXPIRE_TIMEOUT); |
---|
| 1383 | } |
---|
| 1384 | |
---|
| 1385 | /* |
---|
| 1386 | * Packet forwarding routine once entry in the cache is made |
---|
| 1387 | */ |
---|
| 1388 | static int |
---|
| 1389 | ip_mdq(m, ifp, rt, xmt_vif) |
---|
| 1390 | register struct mbuf *m; |
---|
| 1391 | register struct ifnet *ifp; |
---|
| 1392 | register struct mfc *rt; |
---|
| 1393 | register vifi_t xmt_vif; |
---|
| 1394 | { |
---|
| 1395 | register struct ip *ip = mtod(m, struct ip *); |
---|
| 1396 | register vifi_t vifi; |
---|
| 1397 | register struct vif *vifp; |
---|
| 1398 | register int plen = ntohs(ip->ip_len); |
---|
| 1399 | |
---|
| 1400 | /* |
---|
| 1401 | * Macro to send packet on vif. Since RSVP packets don't get counted on |
---|
| 1402 | * input, they shouldn't get counted on output, so statistics keeping is |
---|
| 1403 | * seperate. |
---|
| 1404 | */ |
---|
| 1405 | #define MC_SEND(ip,vifp,m) { \ |
---|
| 1406 | if ((vifp)->v_flags & VIFF_TUNNEL) \ |
---|
| 1407 | encap_send((ip), (vifp), (m)); \ |
---|
| 1408 | else \ |
---|
| 1409 | phyint_send((ip), (vifp), (m)); \ |
---|
| 1410 | } |
---|
| 1411 | |
---|
| 1412 | /* |
---|
| 1413 | * If xmt_vif is not -1, send on only the requested vif. |
---|
| 1414 | * |
---|
| 1415 | * (since vifi_t is u_short, -1 becomes MAXUSHORT, which > numvifs.) |
---|
| 1416 | */ |
---|
| 1417 | if (xmt_vif < numvifs) { |
---|
| 1418 | MC_SEND(ip, viftable + xmt_vif, m); |
---|
| 1419 | return 1; |
---|
| 1420 | } |
---|
| 1421 | |
---|
| 1422 | /* |
---|
| 1423 | * Don't forward if it didn't arrive from the parent vif for its origin. |
---|
| 1424 | */ |
---|
| 1425 | vifi = rt->mfc_parent; |
---|
| 1426 | if ((vifi >= numvifs) || (viftable[vifi].v_ifp != ifp)) { |
---|
| 1427 | /* came in the wrong interface */ |
---|
| 1428 | if (mrtdebug & DEBUG_FORWARD) |
---|
| 1429 | log(LOG_DEBUG, "wrong if: ifp %x vifi %d vififp %x\n", |
---|
| 1430 | ifp, vifi, viftable[vifi].v_ifp); |
---|
| 1431 | ++mrtstat.mrts_wrong_if; |
---|
| 1432 | ++rt->mfc_wrong_if; |
---|
| 1433 | /* |
---|
| 1434 | * If we are doing PIM assert processing, and we are forwarding |
---|
| 1435 | * packets on this interface, and it is a broadcast medium |
---|
| 1436 | * interface (and not a tunnel), send a message to the routing daemon. |
---|
| 1437 | */ |
---|
| 1438 | if (pim_assert && rt->mfc_ttls[vifi] && |
---|
| 1439 | (ifp->if_flags & IFF_BROADCAST) && |
---|
| 1440 | !(viftable[vifi].v_flags & VIFF_TUNNEL)) { |
---|
| 1441 | struct sockaddr_in k_igmpsrc; |
---|
| 1442 | struct mbuf *mm; |
---|
| 1443 | struct igmpmsg *im; |
---|
| 1444 | int hlen = ip->ip_hl << 2; |
---|
| 1445 | struct timeval now; |
---|
| 1446 | register u_long delta; |
---|
| 1447 | |
---|
| 1448 | GET_TIME(now); |
---|
| 1449 | |
---|
| 1450 | TV_DELTA(rt->mfc_last_assert, now, delta); |
---|
| 1451 | |
---|
| 1452 | if (delta > ASSERT_MSG_TIME) { |
---|
| 1453 | mm = m_copy(m, 0, hlen); |
---|
| 1454 | if (mm && (M_HASCL(mm) || mm->m_len < hlen)) |
---|
| 1455 | mm = m_pullup(mm, hlen); |
---|
| 1456 | if (mm == NULL) { |
---|
| 1457 | return ENOBUFS; |
---|
| 1458 | } |
---|
| 1459 | |
---|
| 1460 | rt->mfc_last_assert = now; |
---|
| 1461 | |
---|
| 1462 | im = mtod(mm, struct igmpmsg *); |
---|
| 1463 | im->im_msgtype = IGMPMSG_WRONGVIF; |
---|
| 1464 | im->im_mbz = 0; |
---|
| 1465 | im->im_vif = vifi; |
---|
| 1466 | |
---|
| 1467 | k_igmpsrc.sin_addr = im->im_src; |
---|
| 1468 | |
---|
| 1469 | socket_send(ip_mrouter, mm, &k_igmpsrc); |
---|
| 1470 | } |
---|
| 1471 | } |
---|
| 1472 | return 0; |
---|
| 1473 | } |
---|
| 1474 | |
---|
| 1475 | /* If I sourced this packet, it counts as output, else it was input. */ |
---|
| 1476 | if (ip->ip_src.s_addr == viftable[vifi].v_lcl_addr.s_addr) { |
---|
| 1477 | viftable[vifi].v_pkt_out++; |
---|
| 1478 | viftable[vifi].v_bytes_out += plen; |
---|
| 1479 | } else { |
---|
| 1480 | viftable[vifi].v_pkt_in++; |
---|
| 1481 | viftable[vifi].v_bytes_in += plen; |
---|
| 1482 | } |
---|
| 1483 | rt->mfc_pkt_cnt++; |
---|
| 1484 | rt->mfc_byte_cnt += plen; |
---|
| 1485 | |
---|
| 1486 | /* |
---|
| 1487 | * For each vif, decide if a copy of the packet should be forwarded. |
---|
| 1488 | * Forward if: |
---|
| 1489 | * - the ttl exceeds the vif's threshold |
---|
| 1490 | * - there are group members downstream on interface |
---|
| 1491 | */ |
---|
| 1492 | for (vifp = viftable, vifi = 0; vifi < numvifs; vifp++, vifi++) |
---|
| 1493 | if ((rt->mfc_ttls[vifi] > 0) && |
---|
| 1494 | (ip->ip_ttl > rt->mfc_ttls[vifi])) { |
---|
| 1495 | vifp->v_pkt_out++; |
---|
| 1496 | vifp->v_bytes_out += plen; |
---|
| 1497 | MC_SEND(ip, vifp, m); |
---|
| 1498 | } |
---|
| 1499 | |
---|
| 1500 | return 0; |
---|
| 1501 | } |
---|
| 1502 | |
---|
| 1503 | /* |
---|
| 1504 | * check if a vif number is legal/ok. This is used by ip_output, to export |
---|
| 1505 | * numvifs there, |
---|
| 1506 | */ |
---|
| 1507 | static int |
---|
| 1508 | X_legal_vif_num(vif) |
---|
| 1509 | int vif; |
---|
| 1510 | { |
---|
| 1511 | if (vif >= 0 && vif < numvifs) |
---|
| 1512 | return(1); |
---|
| 1513 | else |
---|
| 1514 | return(0); |
---|
| 1515 | } |
---|
| 1516 | |
---|
| 1517 | #ifndef MROUTE_LKM |
---|
| 1518 | int (*legal_vif_num)(int) = X_legal_vif_num; |
---|
| 1519 | #endif |
---|
| 1520 | |
---|
| 1521 | /* |
---|
| 1522 | * Return the local address used by this vif |
---|
| 1523 | */ |
---|
| 1524 | static u_long |
---|
| 1525 | X_ip_mcast_src(vifi) |
---|
| 1526 | int vifi; |
---|
| 1527 | { |
---|
| 1528 | if (vifi >= 0 && vifi < numvifs) |
---|
| 1529 | return viftable[vifi].v_lcl_addr.s_addr; |
---|
| 1530 | else |
---|
| 1531 | return INADDR_ANY; |
---|
| 1532 | } |
---|
| 1533 | |
---|
| 1534 | #ifndef MROUTE_LKM |
---|
| 1535 | u_long (*ip_mcast_src)(int) = X_ip_mcast_src; |
---|
| 1536 | #endif |
---|
| 1537 | |
---|
| 1538 | static void |
---|
| 1539 | phyint_send(ip, vifp, m) |
---|
| 1540 | struct ip *ip; |
---|
| 1541 | struct vif *vifp; |
---|
| 1542 | struct mbuf *m; |
---|
| 1543 | { |
---|
| 1544 | register struct mbuf *mb_copy; |
---|
| 1545 | register int hlen = ip->ip_hl << 2; |
---|
| 1546 | |
---|
| 1547 | /* |
---|
| 1548 | * Make a new reference to the packet; make sure that |
---|
| 1549 | * the IP header is actually copied, not just referenced, |
---|
| 1550 | * so that ip_output() only scribbles on the copy. |
---|
| 1551 | */ |
---|
| 1552 | mb_copy = m_copy(m, 0, M_COPYALL); |
---|
| 1553 | if (mb_copy && (M_HASCL(mb_copy) || mb_copy->m_len < hlen)) |
---|
| 1554 | mb_copy = m_pullup(mb_copy, hlen); |
---|
| 1555 | if (mb_copy == NULL) |
---|
| 1556 | return; |
---|
| 1557 | |
---|
| 1558 | if (vifp->v_rate_limit <= 0) |
---|
| 1559 | tbf_send_packet(vifp, mb_copy); |
---|
| 1560 | else |
---|
| 1561 | tbf_control(vifp, mb_copy, mtod(mb_copy, struct ip *), ip->ip_len); |
---|
| 1562 | } |
---|
| 1563 | |
---|
| 1564 | static void |
---|
| 1565 | encap_send(ip, vifp, m) |
---|
| 1566 | register struct ip *ip; |
---|
| 1567 | register struct vif *vifp; |
---|
| 1568 | register struct mbuf *m; |
---|
| 1569 | { |
---|
| 1570 | register struct mbuf *mb_copy; |
---|
| 1571 | register struct ip *ip_copy; |
---|
| 1572 | register int i, len = ip->ip_len; |
---|
| 1573 | |
---|
| 1574 | /* |
---|
| 1575 | * copy the old packet & pullup it's IP header into the |
---|
| 1576 | * new mbuf so we can modify it. Try to fill the new |
---|
| 1577 | * mbuf since if we don't the ethernet driver will. |
---|
| 1578 | */ |
---|
| 1579 | MGETHDR(mb_copy, M_DONTWAIT, MT_HEADER); |
---|
| 1580 | if (mb_copy == NULL) |
---|
| 1581 | return; |
---|
| 1582 | mb_copy->m_data += max_linkhdr; |
---|
| 1583 | mb_copy->m_len = sizeof(multicast_encap_iphdr); |
---|
| 1584 | |
---|
| 1585 | if ((mb_copy->m_next = m_copy(m, 0, M_COPYALL)) == NULL) { |
---|
| 1586 | m_freem(mb_copy); |
---|
| 1587 | return; |
---|
| 1588 | } |
---|
| 1589 | i = MHLEN - M_LEADINGSPACE(mb_copy); |
---|
| 1590 | if (i > len) |
---|
| 1591 | i = len; |
---|
| 1592 | mb_copy = m_pullup(mb_copy, i); |
---|
| 1593 | if (mb_copy == NULL) |
---|
| 1594 | return; |
---|
| 1595 | mb_copy->m_pkthdr.len = len + sizeof(multicast_encap_iphdr); |
---|
| 1596 | |
---|
| 1597 | /* |
---|
| 1598 | * fill in the encapsulating IP header. |
---|
| 1599 | */ |
---|
| 1600 | ip_copy = mtod(mb_copy, struct ip *); |
---|
| 1601 | *ip_copy = multicast_encap_iphdr; |
---|
| 1602 | ip_copy->ip_id = htons(ip_id++); |
---|
| 1603 | ip_copy->ip_len += len; |
---|
| 1604 | ip_copy->ip_src = vifp->v_lcl_addr; |
---|
| 1605 | ip_copy->ip_dst = vifp->v_rmt_addr; |
---|
| 1606 | |
---|
| 1607 | /* |
---|
| 1608 | * turn the encapsulated IP header back into a valid one. |
---|
| 1609 | */ |
---|
| 1610 | ip = (struct ip *)((caddr_t)ip_copy + sizeof(multicast_encap_iphdr)); |
---|
| 1611 | --ip->ip_ttl; |
---|
| 1612 | HTONS(ip->ip_len); |
---|
| 1613 | HTONS(ip->ip_off); |
---|
| 1614 | ip->ip_sum = 0; |
---|
| 1615 | mb_copy->m_data += sizeof(multicast_encap_iphdr); |
---|
| 1616 | ip->ip_sum = in_cksum(mb_copy, ip->ip_hl << 2); |
---|
| 1617 | mb_copy->m_data -= sizeof(multicast_encap_iphdr); |
---|
| 1618 | |
---|
| 1619 | if (vifp->v_rate_limit <= 0) |
---|
| 1620 | tbf_send_packet(vifp, mb_copy); |
---|
| 1621 | else |
---|
| 1622 | tbf_control(vifp, mb_copy, ip, ip_copy->ip_len); |
---|
| 1623 | } |
---|
| 1624 | |
---|
| 1625 | /* |
---|
| 1626 | * De-encapsulate a packet and feed it back through ip input (this |
---|
| 1627 | * routine is called whenever IP gets a packet with proto type |
---|
| 1628 | * ENCAP_PROTO and a local destination address). |
---|
| 1629 | */ |
---|
| 1630 | void |
---|
| 1631 | #ifdef MROUTE_LKM |
---|
| 1632 | X_ipip_input(m, iphlen) |
---|
| 1633 | #else |
---|
| 1634 | ipip_input(m, iphlen) |
---|
| 1635 | #endif |
---|
| 1636 | register struct mbuf *m; |
---|
| 1637 | int iphlen; |
---|
| 1638 | { |
---|
| 1639 | struct ifnet *ifp = m->m_pkthdr.rcvif; |
---|
| 1640 | register struct ip *ip = mtod(m, struct ip *); |
---|
| 1641 | register int hlen = ip->ip_hl << 2; |
---|
| 1642 | register int s; |
---|
| 1643 | register struct ifqueue *ifq; |
---|
| 1644 | register struct vif *vifp; |
---|
| 1645 | |
---|
| 1646 | if (!have_encap_tunnel) { |
---|
| 1647 | rip_input(m, iphlen); |
---|
| 1648 | return; |
---|
| 1649 | } |
---|
| 1650 | /* |
---|
| 1651 | * dump the packet if it's not to a multicast destination or if |
---|
| 1652 | * we don't have an encapsulating tunnel with the source. |
---|
| 1653 | * Note: This code assumes that the remote site IP address |
---|
| 1654 | * uniquely identifies the tunnel (i.e., that this site has |
---|
| 1655 | * at most one tunnel with the remote site). |
---|
| 1656 | */ |
---|
| 1657 | if (! IN_MULTICAST(ntohl(((struct ip *)((char *)ip + hlen))->ip_dst.s_addr))) { |
---|
| 1658 | ++mrtstat.mrts_bad_tunnel; |
---|
| 1659 | m_freem(m); |
---|
| 1660 | return; |
---|
| 1661 | } |
---|
| 1662 | if (ip->ip_src.s_addr != last_encap_src) { |
---|
| 1663 | register struct vif *vife; |
---|
| 1664 | |
---|
| 1665 | vifp = viftable; |
---|
| 1666 | vife = vifp + numvifs; |
---|
| 1667 | last_encap_src = ip->ip_src.s_addr; |
---|
| 1668 | last_encap_vif = 0; |
---|
| 1669 | for ( ; vifp < vife; ++vifp) |
---|
| 1670 | if (vifp->v_rmt_addr.s_addr == ip->ip_src.s_addr) { |
---|
| 1671 | if ((vifp->v_flags & (VIFF_TUNNEL|VIFF_SRCRT)) |
---|
| 1672 | == VIFF_TUNNEL) |
---|
| 1673 | last_encap_vif = vifp; |
---|
| 1674 | break; |
---|
| 1675 | } |
---|
| 1676 | } |
---|
| 1677 | if ((vifp = last_encap_vif) == 0) { |
---|
| 1678 | last_encap_src = 0; |
---|
| 1679 | mrtstat.mrts_cant_tunnel++; /*XXX*/ |
---|
| 1680 | m_freem(m); |
---|
| 1681 | if (mrtdebug) |
---|
| 1682 | log(LOG_DEBUG, "ip_mforward: no tunnel with %x\n", |
---|
| 1683 | ntohl(ip->ip_src.s_addr)); |
---|
| 1684 | return; |
---|
| 1685 | } |
---|
| 1686 | ifp = vifp->v_ifp; |
---|
| 1687 | |
---|
| 1688 | if (hlen > IP_HDR_LEN) |
---|
| 1689 | ip_stripoptions(m, (struct mbuf *) 0); |
---|
| 1690 | m->m_data += IP_HDR_LEN; |
---|
| 1691 | m->m_len -= IP_HDR_LEN; |
---|
| 1692 | m->m_pkthdr.len -= IP_HDR_LEN; |
---|
| 1693 | m->m_pkthdr.rcvif = ifp; |
---|
| 1694 | |
---|
| 1695 | ifq = &ipintrq; |
---|
| 1696 | s = splimp(); |
---|
| 1697 | if (IF_QFULL(ifq)) { |
---|
| 1698 | IF_DROP(ifq); |
---|
| 1699 | m_freem(m); |
---|
| 1700 | } else { |
---|
| 1701 | IF_ENQUEUE(ifq, m); |
---|
| 1702 | /* |
---|
| 1703 | * normally we would need a "schednetisr(NETISR_IP)" |
---|
| 1704 | * here but we were called by ip_input and it is going |
---|
| 1705 | * to loop back & try to dequeue the packet we just |
---|
| 1706 | * queued as soon as we return so we avoid the |
---|
| 1707 | * unnecessary software interrrupt. |
---|
| 1708 | */ |
---|
| 1709 | } |
---|
| 1710 | splx(s); |
---|
| 1711 | } |
---|
| 1712 | |
---|
| 1713 | /* |
---|
| 1714 | * Token bucket filter module |
---|
| 1715 | */ |
---|
| 1716 | |
---|
| 1717 | static void |
---|
| 1718 | tbf_control(vifp, m, ip, p_len) |
---|
| 1719 | register struct vif *vifp; |
---|
| 1720 | register struct mbuf *m; |
---|
| 1721 | register struct ip *ip; |
---|
| 1722 | register u_long p_len; |
---|
| 1723 | { |
---|
| 1724 | register struct tbf *t = vifp->v_tbf; |
---|
| 1725 | |
---|
| 1726 | if (p_len > MAX_BKT_SIZE) { |
---|
| 1727 | /* drop if packet is too large */ |
---|
| 1728 | mrtstat.mrts_pkt2large++; |
---|
| 1729 | m_freem(m); |
---|
| 1730 | return; |
---|
| 1731 | } |
---|
| 1732 | |
---|
| 1733 | tbf_update_tokens(vifp); |
---|
| 1734 | |
---|
| 1735 | /* if there are enough tokens, |
---|
| 1736 | * and the queue is empty, |
---|
| 1737 | * send this packet out |
---|
| 1738 | */ |
---|
| 1739 | |
---|
| 1740 | if (t->tbf_q_len == 0) { |
---|
| 1741 | /* queue empty, send packet if enough tokens */ |
---|
| 1742 | if (p_len <= t->tbf_n_tok) { |
---|
| 1743 | t->tbf_n_tok -= p_len; |
---|
| 1744 | tbf_send_packet(vifp, m); |
---|
| 1745 | } else { |
---|
| 1746 | /* queue packet and timeout till later */ |
---|
| 1747 | tbf_queue(vifp, m); |
---|
| 1748 | timeout(tbf_reprocess_q, (caddr_t)vifp, TBF_REPROCESS); |
---|
| 1749 | } |
---|
| 1750 | } else if (t->tbf_q_len < t->tbf_max_q_len) { |
---|
| 1751 | /* finite queue length, so queue pkts and process queue */ |
---|
| 1752 | tbf_queue(vifp, m); |
---|
| 1753 | tbf_process_q(vifp); |
---|
| 1754 | } else { |
---|
| 1755 | /* queue length too much, try to dq and queue and process */ |
---|
| 1756 | if (!tbf_dq_sel(vifp, ip)) { |
---|
| 1757 | mrtstat.mrts_q_overflow++; |
---|
| 1758 | m_freem(m); |
---|
| 1759 | return; |
---|
| 1760 | } else { |
---|
| 1761 | tbf_queue(vifp, m); |
---|
| 1762 | tbf_process_q(vifp); |
---|
| 1763 | } |
---|
| 1764 | } |
---|
| 1765 | return; |
---|
| 1766 | } |
---|
| 1767 | |
---|
| 1768 | /* |
---|
| 1769 | * adds a packet to the queue at the interface |
---|
| 1770 | */ |
---|
| 1771 | static void |
---|
| 1772 | tbf_queue(vifp, m) |
---|
| 1773 | register struct vif *vifp; |
---|
| 1774 | register struct mbuf *m; |
---|
| 1775 | { |
---|
| 1776 | register int s = splnet(); |
---|
| 1777 | register struct tbf *t = vifp->v_tbf; |
---|
| 1778 | |
---|
| 1779 | if (t->tbf_t == NULL) { |
---|
| 1780 | /* Queue was empty */ |
---|
| 1781 | t->tbf_q = m; |
---|
| 1782 | } else { |
---|
| 1783 | /* Insert at tail */ |
---|
| 1784 | t->tbf_t->m_act = m; |
---|
| 1785 | } |
---|
| 1786 | |
---|
| 1787 | /* Set new tail pointer */ |
---|
| 1788 | t->tbf_t = m; |
---|
| 1789 | |
---|
| 1790 | #ifdef DIAGNOSTIC |
---|
| 1791 | /* Make sure we didn't get fed a bogus mbuf */ |
---|
| 1792 | if (m->m_act) |
---|
| 1793 | panic("tbf_queue: m_act"); |
---|
| 1794 | #endif |
---|
| 1795 | m->m_act = NULL; |
---|
| 1796 | |
---|
| 1797 | t->tbf_q_len++; |
---|
| 1798 | |
---|
| 1799 | splx(s); |
---|
| 1800 | } |
---|
| 1801 | |
---|
| 1802 | |
---|
| 1803 | /* |
---|
| 1804 | * processes the queue at the interface |
---|
| 1805 | */ |
---|
| 1806 | static void |
---|
| 1807 | tbf_process_q(vifp) |
---|
| 1808 | register struct vif *vifp; |
---|
| 1809 | { |
---|
| 1810 | register struct mbuf *m; |
---|
| 1811 | register int len; |
---|
| 1812 | register int s = splnet(); |
---|
| 1813 | register struct tbf *t = vifp->v_tbf; |
---|
| 1814 | |
---|
| 1815 | /* loop through the queue at the interface and send as many packets |
---|
| 1816 | * as possible |
---|
| 1817 | */ |
---|
| 1818 | while (t->tbf_q_len > 0) { |
---|
| 1819 | m = t->tbf_q; |
---|
| 1820 | |
---|
| 1821 | len = mtod(m, struct ip *)->ip_len; |
---|
| 1822 | |
---|
| 1823 | /* determine if the packet can be sent */ |
---|
| 1824 | if (len <= t->tbf_n_tok) { |
---|
| 1825 | /* if so, |
---|
| 1826 | * reduce no of tokens, dequeue the packet, |
---|
| 1827 | * send the packet. |
---|
| 1828 | */ |
---|
| 1829 | t->tbf_n_tok -= len; |
---|
| 1830 | |
---|
| 1831 | t->tbf_q = m->m_act; |
---|
| 1832 | if (--t->tbf_q_len == 0) |
---|
| 1833 | t->tbf_t = NULL; |
---|
| 1834 | |
---|
| 1835 | m->m_act = NULL; |
---|
| 1836 | tbf_send_packet(vifp, m); |
---|
| 1837 | |
---|
| 1838 | } else break; |
---|
| 1839 | } |
---|
| 1840 | splx(s); |
---|
| 1841 | } |
---|
| 1842 | |
---|
| 1843 | static void |
---|
| 1844 | tbf_reprocess_q(xvifp) |
---|
| 1845 | void *xvifp; |
---|
| 1846 | { |
---|
| 1847 | register struct vif *vifp = xvifp; |
---|
| 1848 | if (ip_mrouter == NULL) |
---|
| 1849 | return; |
---|
| 1850 | |
---|
| 1851 | tbf_update_tokens(vifp); |
---|
| 1852 | |
---|
| 1853 | tbf_process_q(vifp); |
---|
| 1854 | |
---|
| 1855 | if (vifp->v_tbf->tbf_q_len) |
---|
| 1856 | timeout(tbf_reprocess_q, (caddr_t)vifp, TBF_REPROCESS); |
---|
| 1857 | } |
---|
| 1858 | |
---|
| 1859 | /* function that will selectively discard a member of the queue |
---|
| 1860 | * based on the precedence value and the priority |
---|
| 1861 | */ |
---|
| 1862 | static int |
---|
| 1863 | tbf_dq_sel(vifp, ip) |
---|
| 1864 | register struct vif *vifp; |
---|
| 1865 | register struct ip *ip; |
---|
| 1866 | { |
---|
| 1867 | register int s = splnet(); |
---|
| 1868 | register u_int p; |
---|
| 1869 | register struct mbuf *m, *last; |
---|
| 1870 | register struct mbuf **np; |
---|
| 1871 | register struct tbf *t = vifp->v_tbf; |
---|
| 1872 | |
---|
| 1873 | p = priority(vifp, ip); |
---|
| 1874 | |
---|
| 1875 | np = &t->tbf_q; |
---|
| 1876 | last = NULL; |
---|
| 1877 | while ((m = *np) != NULL) { |
---|
| 1878 | if (p > priority(vifp, mtod(m, struct ip *))) { |
---|
| 1879 | *np = m->m_act; |
---|
| 1880 | /* If we're removing the last packet, fix the tail pointer */ |
---|
| 1881 | if (m == t->tbf_t) |
---|
| 1882 | t->tbf_t = last; |
---|
| 1883 | m_freem(m); |
---|
| 1884 | /* it's impossible for the queue to be empty, but |
---|
| 1885 | * we check anyway. */ |
---|
| 1886 | if (--t->tbf_q_len == 0) |
---|
| 1887 | t->tbf_t = NULL; |
---|
| 1888 | splx(s); |
---|
| 1889 | mrtstat.mrts_drop_sel++; |
---|
| 1890 | return(1); |
---|
| 1891 | } |
---|
| 1892 | np = &m->m_act; |
---|
| 1893 | last = m; |
---|
| 1894 | } |
---|
| 1895 | splx(s); |
---|
| 1896 | return(0); |
---|
| 1897 | } |
---|
| 1898 | |
---|
| 1899 | static void |
---|
| 1900 | tbf_send_packet(vifp, m) |
---|
| 1901 | register struct vif *vifp; |
---|
| 1902 | register struct mbuf *m; |
---|
| 1903 | { |
---|
| 1904 | struct ip_moptions imo; |
---|
| 1905 | int error; |
---|
| 1906 | static struct route ro; |
---|
| 1907 | int s = splnet(); |
---|
| 1908 | |
---|
| 1909 | if (vifp->v_flags & VIFF_TUNNEL) { |
---|
| 1910 | /* If tunnel options */ |
---|
| 1911 | ip_output(m, (struct mbuf *)0, &vifp->v_route, |
---|
| 1912 | IP_FORWARDING, (struct ip_moptions *)0); |
---|
| 1913 | } else { |
---|
| 1914 | imo.imo_multicast_ifp = vifp->v_ifp; |
---|
| 1915 | imo.imo_multicast_ttl = mtod(m, struct ip *)->ip_ttl - 1; |
---|
| 1916 | imo.imo_multicast_loop = 1; |
---|
| 1917 | imo.imo_multicast_vif = -1; |
---|
| 1918 | |
---|
| 1919 | /* |
---|
| 1920 | * Re-entrancy should not be a problem here, because |
---|
| 1921 | * the packets that we send out and are looped back at us |
---|
| 1922 | * should get rejected because they appear to come from |
---|
| 1923 | * the loopback interface, thus preventing looping. |
---|
| 1924 | */ |
---|
| 1925 | error = ip_output(m, (struct mbuf *)0, &ro, |
---|
| 1926 | IP_FORWARDING, &imo); |
---|
| 1927 | |
---|
| 1928 | if (mrtdebug & DEBUG_XMIT) |
---|
| 1929 | log(LOG_DEBUG, "phyint_send on vif %d err %d\n", |
---|
| 1930 | vifp - viftable, error); |
---|
| 1931 | } |
---|
| 1932 | splx(s); |
---|
| 1933 | } |
---|
| 1934 | |
---|
| 1935 | /* determine the current time and then |
---|
| 1936 | * the elapsed time (between the last time and time now) |
---|
| 1937 | * in milliseconds & update the no. of tokens in the bucket |
---|
| 1938 | */ |
---|
| 1939 | static void |
---|
| 1940 | tbf_update_tokens(vifp) |
---|
| 1941 | register struct vif *vifp; |
---|
| 1942 | { |
---|
| 1943 | struct timeval tp; |
---|
| 1944 | register u_long tm; |
---|
| 1945 | register int s = splnet(); |
---|
| 1946 | register struct tbf *t = vifp->v_tbf; |
---|
| 1947 | |
---|
| 1948 | GET_TIME(tp); |
---|
| 1949 | |
---|
| 1950 | TV_DELTA(tp, t->tbf_last_pkt_t, tm); |
---|
| 1951 | |
---|
| 1952 | /* |
---|
| 1953 | * This formula is actually |
---|
| 1954 | * "time in seconds" * "bytes/second". |
---|
| 1955 | * |
---|
| 1956 | * (tm / 1000000) * (v_rate_limit * 1000 * (1000/1024) / 8) |
---|
| 1957 | * |
---|
| 1958 | * The (1000/1024) was introduced in add_vif to optimize |
---|
| 1959 | * this divide into a shift. |
---|
| 1960 | */ |
---|
| 1961 | t->tbf_n_tok += tm * vifp->v_rate_limit / 1024 / 8; |
---|
| 1962 | t->tbf_last_pkt_t = tp; |
---|
| 1963 | |
---|
| 1964 | if (t->tbf_n_tok > MAX_BKT_SIZE) |
---|
| 1965 | t->tbf_n_tok = MAX_BKT_SIZE; |
---|
| 1966 | |
---|
| 1967 | splx(s); |
---|
| 1968 | } |
---|
| 1969 | |
---|
| 1970 | static int |
---|
| 1971 | priority(vifp, ip) |
---|
| 1972 | register struct vif *vifp; |
---|
| 1973 | register struct ip *ip; |
---|
| 1974 | { |
---|
| 1975 | register int prio; |
---|
| 1976 | |
---|
| 1977 | /* temporary hack; may add general packet classifier some day */ |
---|
| 1978 | |
---|
| 1979 | /* |
---|
| 1980 | * The UDP port space is divided up into four priority ranges: |
---|
| 1981 | * [0, 16384) : unclassified - lowest priority |
---|
| 1982 | * [16384, 32768) : audio - highest priority |
---|
| 1983 | * [32768, 49152) : whiteboard - medium priority |
---|
| 1984 | * [49152, 65536) : video - low priority |
---|
| 1985 | */ |
---|
| 1986 | if (ip->ip_p == IPPROTO_UDP) { |
---|
| 1987 | struct udphdr *udp = (struct udphdr *)(((char *)ip) + (ip->ip_hl << 2)); |
---|
| 1988 | switch (ntohs(udp->uh_dport) & 0xc000) { |
---|
| 1989 | case 0x4000: |
---|
| 1990 | prio = 70; |
---|
| 1991 | break; |
---|
| 1992 | case 0x8000: |
---|
| 1993 | prio = 60; |
---|
| 1994 | break; |
---|
| 1995 | case 0xc000: |
---|
| 1996 | prio = 55; |
---|
| 1997 | break; |
---|
| 1998 | default: |
---|
| 1999 | prio = 50; |
---|
| 2000 | break; |
---|
| 2001 | } |
---|
| 2002 | if (tbfdebug > 1) |
---|
| 2003 | log(LOG_DEBUG, "port %x prio%d\n", ntohs(udp->uh_dport), prio); |
---|
| 2004 | } else { |
---|
| 2005 | prio = 50; |
---|
| 2006 | } |
---|
| 2007 | return prio; |
---|
| 2008 | } |
---|
| 2009 | |
---|
| 2010 | /* |
---|
| 2011 | * End of token bucket filter modifications |
---|
| 2012 | */ |
---|
| 2013 | |
---|
| 2014 | int |
---|
| 2015 | ip_rsvp_vif_init(so, m) |
---|
| 2016 | struct socket *so; |
---|
| 2017 | struct mbuf *m; |
---|
| 2018 | { |
---|
| 2019 | int i; |
---|
| 2020 | register int s; |
---|
| 2021 | |
---|
| 2022 | if (rsvpdebug) |
---|
| 2023 | printf("ip_rsvp_vif_init: so_type = %d, pr_protocol = %d\n", |
---|
| 2024 | so->so_type, so->so_proto->pr_protocol); |
---|
| 2025 | |
---|
| 2026 | if (so->so_type != SOCK_RAW || so->so_proto->pr_protocol != IPPROTO_RSVP) |
---|
| 2027 | return EOPNOTSUPP; |
---|
| 2028 | |
---|
| 2029 | /* Check mbuf. */ |
---|
| 2030 | if (m == NULL || m->m_len != sizeof(int)) { |
---|
| 2031 | return EINVAL; |
---|
| 2032 | } |
---|
| 2033 | i = *(mtod(m, int *)); |
---|
| 2034 | |
---|
| 2035 | if (rsvpdebug) |
---|
| 2036 | printf("ip_rsvp_vif_init: vif = %d rsvp_on = %d\n",i,rsvp_on); |
---|
| 2037 | |
---|
| 2038 | s = splnet(); |
---|
| 2039 | |
---|
| 2040 | /* Check vif. */ |
---|
| 2041 | if (!legal_vif_num(i)) { |
---|
| 2042 | splx(s); |
---|
| 2043 | return EADDRNOTAVAIL; |
---|
| 2044 | } |
---|
| 2045 | |
---|
| 2046 | /* Check if socket is available. */ |
---|
| 2047 | if (viftable[i].v_rsvpd != NULL) { |
---|
| 2048 | splx(s); |
---|
| 2049 | return EADDRINUSE; |
---|
| 2050 | } |
---|
| 2051 | |
---|
| 2052 | viftable[i].v_rsvpd = so; |
---|
| 2053 | /* This may seem silly, but we need to be sure we don't over-increment |
---|
| 2054 | * the RSVP counter, in case something slips up. |
---|
| 2055 | */ |
---|
| 2056 | if (!viftable[i].v_rsvp_on) { |
---|
| 2057 | viftable[i].v_rsvp_on = 1; |
---|
| 2058 | rsvp_on++; |
---|
| 2059 | } |
---|
| 2060 | |
---|
| 2061 | splx(s); |
---|
| 2062 | return 0; |
---|
| 2063 | } |
---|
| 2064 | |
---|
| 2065 | int |
---|
| 2066 | ip_rsvp_vif_done(so, m) |
---|
| 2067 | struct socket *so; |
---|
| 2068 | struct mbuf *m; |
---|
| 2069 | { |
---|
| 2070 | int i; |
---|
| 2071 | register int s; |
---|
| 2072 | |
---|
| 2073 | if (rsvpdebug) |
---|
| 2074 | printf("ip_rsvp_vif_done: so_type = %d, pr_protocol = %d\n", |
---|
| 2075 | so->so_type, so->so_proto->pr_protocol); |
---|
| 2076 | |
---|
| 2077 | if (so->so_type != SOCK_RAW || so->so_proto->pr_protocol != IPPROTO_RSVP) |
---|
| 2078 | return EOPNOTSUPP; |
---|
| 2079 | |
---|
| 2080 | /* Check mbuf. */ |
---|
| 2081 | if (m == NULL || m->m_len != sizeof(int)) { |
---|
| 2082 | return EINVAL; |
---|
| 2083 | } |
---|
| 2084 | i = *(mtod(m, int *)); |
---|
| 2085 | |
---|
| 2086 | s = splnet(); |
---|
| 2087 | |
---|
| 2088 | /* Check vif. */ |
---|
| 2089 | if (!legal_vif_num(i)) { |
---|
| 2090 | splx(s); |
---|
| 2091 | return EADDRNOTAVAIL; |
---|
| 2092 | } |
---|
| 2093 | |
---|
| 2094 | if (rsvpdebug) |
---|
| 2095 | printf("ip_rsvp_vif_done: v_rsvpd = %p so = %p\n", |
---|
| 2096 | viftable[i].v_rsvpd, so); |
---|
| 2097 | |
---|
| 2098 | viftable[i].v_rsvpd = NULL; |
---|
| 2099 | /* This may seem silly, but we need to be sure we don't over-decrement |
---|
| 2100 | * the RSVP counter, in case something slips up. |
---|
| 2101 | */ |
---|
| 2102 | if (viftable[i].v_rsvp_on) { |
---|
| 2103 | viftable[i].v_rsvp_on = 0; |
---|
| 2104 | rsvp_on--; |
---|
| 2105 | } |
---|
| 2106 | |
---|
| 2107 | splx(s); |
---|
| 2108 | return 0; |
---|
| 2109 | } |
---|
| 2110 | |
---|
| 2111 | void |
---|
| 2112 | ip_rsvp_force_done(so) |
---|
| 2113 | struct socket *so; |
---|
| 2114 | { |
---|
| 2115 | int vifi; |
---|
| 2116 | register int s; |
---|
| 2117 | |
---|
| 2118 | /* Don't bother if it is not the right type of socket. */ |
---|
| 2119 | if (so->so_type != SOCK_RAW || so->so_proto->pr_protocol != IPPROTO_RSVP) |
---|
| 2120 | return; |
---|
| 2121 | |
---|
| 2122 | s = splnet(); |
---|
| 2123 | |
---|
| 2124 | /* The socket may be attached to more than one vif...this |
---|
| 2125 | * is perfectly legal. |
---|
| 2126 | */ |
---|
| 2127 | for (vifi = 0; vifi < numvifs; vifi++) { |
---|
| 2128 | if (viftable[vifi].v_rsvpd == so) { |
---|
| 2129 | viftable[vifi].v_rsvpd = NULL; |
---|
| 2130 | /* This may seem silly, but we need to be sure we don't |
---|
| 2131 | * over-decrement the RSVP counter, in case something slips up. |
---|
| 2132 | */ |
---|
| 2133 | if (viftable[vifi].v_rsvp_on) { |
---|
| 2134 | viftable[vifi].v_rsvp_on = 0; |
---|
| 2135 | rsvp_on--; |
---|
| 2136 | } |
---|
| 2137 | } |
---|
| 2138 | } |
---|
| 2139 | |
---|
| 2140 | splx(s); |
---|
| 2141 | return; |
---|
| 2142 | } |
---|
| 2143 | |
---|
| 2144 | void |
---|
| 2145 | rsvp_input(m, iphlen) |
---|
| 2146 | struct mbuf *m; |
---|
| 2147 | int iphlen; |
---|
| 2148 | { |
---|
| 2149 | int vifi; |
---|
| 2150 | register struct ip *ip = mtod(m, struct ip *); |
---|
| 2151 | static struct sockaddr_in rsvp_src = { sizeof rsvp_src, AF_INET }; |
---|
| 2152 | register int s; |
---|
| 2153 | struct ifnet *ifp; |
---|
| 2154 | |
---|
| 2155 | if (rsvpdebug) |
---|
| 2156 | printf("rsvp_input: rsvp_on %d\n",rsvp_on); |
---|
| 2157 | |
---|
| 2158 | /* Can still get packets with rsvp_on = 0 if there is a local member |
---|
| 2159 | * of the group to which the RSVP packet is addressed. But in this |
---|
| 2160 | * case we want to throw the packet away. |
---|
| 2161 | */ |
---|
| 2162 | if (!rsvp_on) { |
---|
| 2163 | m_freem(m); |
---|
| 2164 | return; |
---|
| 2165 | } |
---|
| 2166 | |
---|
| 2167 | /* If the old-style non-vif-associated socket is set, then use |
---|
| 2168 | * it and ignore the new ones. |
---|
| 2169 | */ |
---|
| 2170 | if (ip_rsvpd != NULL) { |
---|
| 2171 | if (rsvpdebug) |
---|
| 2172 | printf("rsvp_input: Sending packet up old-style socket\n"); |
---|
| 2173 | rip_input(m, iphlen); |
---|
| 2174 | return; |
---|
| 2175 | } |
---|
| 2176 | |
---|
| 2177 | s = splnet(); |
---|
| 2178 | |
---|
| 2179 | if (rsvpdebug) |
---|
| 2180 | printf("rsvp_input: check vifs\n"); |
---|
| 2181 | |
---|
| 2182 | #ifdef DIAGNOSTIC |
---|
| 2183 | if (!(m->m_flags & M_PKTHDR)) |
---|
| 2184 | panic("rsvp_input no hdr"); |
---|
| 2185 | #endif |
---|
| 2186 | |
---|
| 2187 | ifp = m->m_pkthdr.rcvif; |
---|
| 2188 | /* Find which vif the packet arrived on. */ |
---|
| 2189 | for (vifi = 0; vifi < numvifs; vifi++) { |
---|
| 2190 | if (viftable[vifi].v_ifp == ifp) |
---|
| 2191 | break; |
---|
| 2192 | } |
---|
| 2193 | |
---|
| 2194 | if (vifi == numvifs) { |
---|
| 2195 | /* Can't find vif packet arrived on. Drop packet. */ |
---|
| 2196 | if (rsvpdebug) |
---|
| 2197 | printf("rsvp_input: Can't find vif for packet...dropping it.\n"); |
---|
| 2198 | m_freem(m); |
---|
| 2199 | splx(s); |
---|
| 2200 | return; |
---|
| 2201 | } |
---|
| 2202 | |
---|
| 2203 | if (rsvpdebug) |
---|
| 2204 | printf("rsvp_input: check socket\n"); |
---|
| 2205 | |
---|
| 2206 | if (viftable[vifi].v_rsvpd == NULL) { |
---|
| 2207 | /* drop packet, since there is no specific socket for this |
---|
| 2208 | * interface */ |
---|
| 2209 | if (rsvpdebug) |
---|
| 2210 | printf("rsvp_input: No socket defined for vif %d\n",vifi); |
---|
| 2211 | m_freem(m); |
---|
| 2212 | splx(s); |
---|
| 2213 | return; |
---|
| 2214 | } |
---|
| 2215 | rsvp_src.sin_addr = ip->ip_src; |
---|
| 2216 | |
---|
| 2217 | if (rsvpdebug && m) |
---|
| 2218 | printf("rsvp_input: m->m_len = %d, sbspace() = %ld\n", |
---|
| 2219 | m->m_len,sbspace(&(viftable[vifi].v_rsvpd->so_rcv))); |
---|
| 2220 | |
---|
| 2221 | if (socket_send(viftable[vifi].v_rsvpd, m, &rsvp_src) < 0) |
---|
| 2222 | if (rsvpdebug) |
---|
| 2223 | printf("rsvp_input: Failed to append to socket\n"); |
---|
| 2224 | else |
---|
| 2225 | if (rsvpdebug) |
---|
| 2226 | printf("rsvp_input: send packet up\n"); |
---|
| 2227 | |
---|
| 2228 | splx(s); |
---|
| 2229 | } |
---|
| 2230 | |
---|
| 2231 | #ifdef MROUTE_LKM |
---|
| 2232 | #include <sys/conf.h> |
---|
| 2233 | #include <sys/exec.h> |
---|
| 2234 | #include <sys/sysent.h> |
---|
| 2235 | #include <sys/lkm.h> |
---|
| 2236 | |
---|
| 2237 | MOD_MISC("ip_mroute_mod") |
---|
| 2238 | |
---|
| 2239 | static int |
---|
| 2240 | ip_mroute_mod_handle(struct lkm_table *lkmtp, int cmd) |
---|
| 2241 | { |
---|
| 2242 | int i; |
---|
| 2243 | struct lkm_misc *args = lkmtp->private.lkm_misc; |
---|
| 2244 | int err = 0; |
---|
| 2245 | |
---|
| 2246 | switch(cmd) { |
---|
| 2247 | static int (*old_ip_mrouter_cmd)(); |
---|
| 2248 | static int (*old_ip_mrouter_done)(); |
---|
| 2249 | static int (*old_ip_mforward)(); |
---|
| 2250 | static int (*old_mrt_ioctl)(); |
---|
| 2251 | static void (*old_proto4_input)(); |
---|
| 2252 | static int (*old_legal_vif_num)(); |
---|
| 2253 | extern struct protosw inetsw[]; |
---|
| 2254 | |
---|
| 2255 | case LKM_E_LOAD: |
---|
| 2256 | if(lkmexists(lkmtp) || ip_mrtproto) |
---|
| 2257 | return(EEXIST); |
---|
| 2258 | old_ip_mrouter_cmd = ip_mrouter_cmd; |
---|
| 2259 | ip_mrouter_cmd = X_ip_mrouter_cmd; |
---|
| 2260 | old_ip_mrouter_done = ip_mrouter_done; |
---|
| 2261 | ip_mrouter_done = X_ip_mrouter_done; |
---|
| 2262 | old_ip_mforward = ip_mforward; |
---|
| 2263 | ip_mforward = X_ip_mforward; |
---|
| 2264 | old_mrt_ioctl = mrt_ioctl; |
---|
| 2265 | mrt_ioctl = X_mrt_ioctl; |
---|
| 2266 | old_proto4_input = inetsw[ip_protox[ENCAP_PROTO]].pr_input; |
---|
| 2267 | inetsw[ip_protox[ENCAP_PROTO]].pr_input = X_ipip_input; |
---|
| 2268 | old_legal_vif_num = legal_vif_num; |
---|
| 2269 | legal_vif_num = X_legal_vif_num; |
---|
| 2270 | ip_mrtproto = IGMP_DVMRP; |
---|
| 2271 | |
---|
| 2272 | printf("\nIP multicast routing loaded\n"); |
---|
| 2273 | break; |
---|
| 2274 | |
---|
| 2275 | case LKM_E_UNLOAD: |
---|
| 2276 | if (ip_mrouter) |
---|
| 2277 | return EINVAL; |
---|
| 2278 | |
---|
| 2279 | ip_mrouter_cmd = old_ip_mrouter_cmd; |
---|
| 2280 | ip_mrouter_done = old_ip_mrouter_done; |
---|
| 2281 | ip_mforward = old_ip_mforward; |
---|
| 2282 | mrt_ioctl = old_mrt_ioctl; |
---|
| 2283 | inetsw[ip_protox[ENCAP_PROTO]].pr_input = old_proto4_input; |
---|
| 2284 | legal_vif_num = old_legal_vif_num; |
---|
| 2285 | ip_mrtproto = 0; |
---|
| 2286 | break; |
---|
| 2287 | |
---|
| 2288 | default: |
---|
| 2289 | err = EINVAL; |
---|
| 2290 | break; |
---|
| 2291 | } |
---|
| 2292 | |
---|
| 2293 | return(err); |
---|
| 2294 | } |
---|
| 2295 | |
---|
| 2296 | int |
---|
| 2297 | ip_mroute_mod(struct lkm_table *lkmtp, int cmd, int ver) { |
---|
| 2298 | DISPATCH(lkmtp, cmd, ver, ip_mroute_mod_handle, ip_mroute_mod_handle, |
---|
| 2299 | nosys); |
---|
| 2300 | } |
---|
| 2301 | |
---|
| 2302 | #endif /* MROUTE_LKM */ |
---|
| 2303 | #endif /* MROUTING */ |
---|