1 /* 2 * Copyright (c) 1988 Regents of the University of California. 3 * All rights reserved. 4 * 5 * Redistribution and use in source and binary forms are permitted 6 * provided that the above copyright notice and this paragraph are 7 * duplicated in all such forms and that any documentation, 8 * advertising materials, and other materials related to such 9 * distribution and use acknowledge that the software was developed 10 * by the University of California, Berkeley. The name of the 11 * University may not be used to endorse or promote products derived 12 * from this software without specific prior written permission. 13 * THIS SOFTWARE IS PROVIDED ``AS IS'' AND WITHOUT ANY EXPRESS OR 14 * IMPLIED WARRANTIES, INCLUDING, WITHOUT LIMITATION, THE IMPLIED 15 * WARRANTIES OF MERCHANTIBILITY AND FITNESS FOR A PARTICULAR PURPOSE. 16 * 17 * @(#)rtsock.c 7.10 (Berkeley) 06/20/90 18 */ 19 20 #include "param.h" 21 #include "mbuf.h" 22 #include "user.h" 23 #include "proc.h" 24 #include "socket.h" 25 #include "socketvar.h" 26 #include "domain.h" 27 #include "protosw.h" 28 #include "errno.h" 29 30 #include "af.h" 31 #include "if.h" 32 #include "route.h" 33 #include "raw_cb.h" 34 35 #include "machine/mtpr.h" 36 37 struct sockaddr route_dst = { 0, PF_ROUTE, }; 38 struct sockaddr route_src = { 0, PF_ROUTE, }; 39 struct sockproto route_proto = { PF_ROUTE, }; 40 41 /*ARGSUSED*/ 42 route_usrreq(so, req, m, nam, control) 43 register struct socket *so; 44 int req; 45 struct mbuf *m, *nam, *control; 46 { 47 register int error = 0; 48 register struct rawcb *rp = sotorawcb(so); 49 if (req == PRU_ATTACH) { 50 MALLOC(rp, struct rawcb *, sizeof(*rp), M_PCB, M_WAITOK); 51 if (so->so_pcb = (caddr_t)rp) 52 bzero(so->so_pcb, sizeof(*rp)); 53 54 } 55 if (req == PRU_DETACH && rp) { 56 int af = rp->rcb_proto.sp_protocol; 57 if (af == AF_INET) 58 route_cb.ip_count--; 59 else if (af == AF_NS) 60 route_cb.ns_count--; 61 else if (af == AF_ISO) 62 route_cb.iso_count--; 63 route_cb.any_count--; 64 } 65 error = raw_usrreq(so, req, m, nam, control); 66 rp = sotorawcb(so); 67 if (req == PRU_ATTACH && rp) { 68 int af = rp->rcb_proto.sp_protocol; 69 if (error) { 70 free((caddr_t)rp, M_PCB); 71 return (error); 72 } 73 if (af == AF_INET) 74 route_cb.ip_count++; 75 else if (af == AF_NS) 76 route_cb.ns_count++; 77 else if (af == AF_ISO) 78 route_cb.iso_count++; 79 rp->rcb_faddr = &route_src; 80 route_cb.any_count++; 81 soisconnected(so); 82 so->so_options |= SO_USELOOPBACK; 83 } 84 return (error); 85 } 86 #define ROUNDUP(a) (1 + (((a) - 1) | (sizeof(long) - 1))) 87 88 /*ARGSUSED*/ 89 route_output(m, so) 90 register struct mbuf *m; 91 struct socket *so; 92 { 93 register struct rt_msghdr *rtm; 94 register struct rtentry *rt = 0; 95 struct rtentry *saved_nrt = 0; 96 struct sockaddr *dst = 0, *gate = 0, *netmask = 0, *genmask = 0; 97 struct sockaddr *ifpaddr = 0; 98 caddr_t cp, lim; 99 int len, error = 0; 100 struct ifnet *ifp = 0; 101 struct ifaddr *ifa; 102 extern struct ifaddr *ifaof_ifpforaddr(), *ifa_ifwithroute(); 103 104 #define senderr(e) { error = e; goto flush;} 105 if (m == 0 || m->m_len < sizeof(long)) 106 return (ENOBUFS); 107 if ((m = m_pullup(m, sizeof(long))) == 0) 108 return (ENOBUFS); 109 if ((m->m_flags & M_PKTHDR) == 0) 110 panic("route_output"); 111 len = m->m_pkthdr.len; 112 rtm = mtod(m, struct rt_msghdr *); 113 if (len < rtm->rtm_msglen) 114 senderr(EINVAL); 115 R_Malloc(rtm, struct rt_msghdr *, len); 116 if (rtm == 0) 117 senderr(ENOBUFS); 118 m_copydata(m, 0, len, (caddr_t)rtm); 119 if (rtm->rtm_version != RTM_VERSION) 120 senderr(EPROTONOSUPPORT); 121 rtm->rtm_pid = u.u_procp->p_pid; 122 lim = len + (caddr_t) rtm; 123 cp = (caddr_t) (rtm + 1); 124 if (rtm->rtm_addrs & RTA_DST) { 125 dst = (struct sockaddr *)cp; 126 cp += ROUNDUP(dst->sa_len); 127 } else 128 senderr(EINVAL); 129 if ((rtm->rtm_addrs & RTA_GATEWAY) && cp < lim) { 130 gate = (struct sockaddr *)cp; 131 cp += ROUNDUP(gate->sa_len); 132 } 133 if ((rtm->rtm_addrs & RTA_NETMASK) && cp < lim) { 134 netmask = (struct sockaddr *)cp; 135 if (*cp) 136 cp += ROUNDUP(netmask->sa_len); 137 else 138 cp += sizeof(long); 139 140 } 141 if ((rtm->rtm_addrs & RTA_GENMASK) && cp < lim) { 142 struct radix_node *t, *rn_addmask(); 143 genmask = (struct sockaddr *)cp; 144 if (*cp) 145 cp += ROUNDUP(netmask->sa_len); 146 else 147 cp += sizeof(long); 148 t = rn_addmask(genmask, 1, 2); 149 if (t && Bcmp(genmask, t->rn_key, *(u_char *)genmask) == 0) 150 genmask = (struct sockaddr *)(t->rn_key); 151 else 152 senderr(ENOBUFS); 153 } 154 if ((rtm->rtm_addrs & RTA_IFP) && cp < lim) { 155 ifpaddr = (struct sockaddr *)cp; 156 } 157 switch (rtm->rtm_type) { 158 case RTM_ADD: 159 if (gate == 0) 160 senderr(EINVAL); 161 error = rtrequest(RTM_ADD, dst, gate, netmask, 162 rtm->rtm_flags, &saved_nrt); 163 if (error == 0 && saved_nrt) { 164 rt_setmetrics(rtm->rtm_inits, 165 &rtm->rtm_rmx, &saved_nrt->rt_rmx); 166 saved_nrt->rt_refcnt--; 167 saved_nrt->rt_genmask = genmask; 168 } 169 break; 170 171 case RTM_DELETE: 172 error = rtrequest(RTM_DELETE, dst, gate, netmask, 173 rtm->rtm_flags, (struct rtentry **)0); 174 break; 175 176 case RTM_GET: 177 case RTM_CHANGE: 178 case RTM_LOCK: 179 rt = rtalloc1(dst, 0); 180 if (rt == 0) 181 senderr(ESRCH); 182 switch(rtm->rtm_type) { 183 struct sockaddr *outmask; 184 185 case RTM_GET: 186 netmask = rt_mask(rt); 187 len = sizeof(*rtm) + ROUNDUP(rt_key(rt)->sa_len); 188 rtm->rtm_addrs = RTA_DST; 189 if (rt->rt_gateway) { 190 len += ROUNDUP(rt->rt_gateway->sa_len); 191 rtm->rtm_addrs |= RTA_GATEWAY; 192 } 193 if (netmask) { 194 len += netmask->sa_len; 195 rtm->rtm_addrs |= RTA_NETMASK; 196 } 197 if (len > rtm->rtm_msglen) { 198 struct rt_msghdr *new_rtm; 199 R_Malloc(new_rtm, struct rt_msghdr *, len); 200 if (new_rtm == 0) 201 senderr(ENOBUFS); 202 Bcopy(rtm, new_rtm, rtm->rtm_msglen); 203 Free(rtm); rtm = new_rtm; 204 gate = (struct sockaddr *) 205 (ROUNDUP(rt->rt_gateway->sa_len) 206 + (char *)dst); 207 Bcopy(&rt->rt_gateway, gate, 208 rt->rt_gateway->sa_len); 209 rtm->rtm_flags = rt->rt_flags; 210 if (netmask) { 211 outmask = (struct sockaddr *) 212 (ROUNDUP(netmask->sa_len)+(char *)gate); 213 Bcopy(netmask, outmask, netmask->sa_len); 214 } 215 } 216 break; 217 218 case RTM_CHANGE: 219 if (gate == 0 || netmask != 0) 220 senderr(EINVAL); 221 if (gate->sa_len > (len = rt->rt_gateway->sa_len)) 222 senderr(EDQUOT); 223 if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest) 224 rt->rt_ifa->ifa_rtrequest(RTM_DELETE, rt, gate); 225 /* new gateway could require new ifaddr, ifp; 226 flags may also be different; ifp may be specified 227 by ll sockaddr when protocol address is ambiguous */ 228 if (ifpaddr && 229 (ifa = ifa_ifwithnet(ifpaddr)) && 230 (ifp = ifa->ifa_ifp) && 231 (ifa = ifaof_ifpforaddr(gate, ifp))) { 232 /* We got it */ 233 } else { 234 ifa = 0; ifp = 0; 235 } 236 Bcopy(gate, rt->rt_gateway, len); 237 rt->rt_gateway->sa_len = len; 238 rt_setmetrics(rtm->rtm_inits, 239 &rtm->rtm_rmx, &rt->rt_rmx); 240 if (ifa == 0) 241 ifa = ifa_ifwithroute(rt->rt_flags, rt_key(rt), 242 gate); 243 if (ifa) { 244 if (rt->rt_ifa != ifa) { 245 rt->rt_ifa = ifa; 246 rt->rt_ifp = ifa->ifa_ifp; 247 } 248 } 249 if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest) 250 rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, gate); 251 if (genmask) 252 rt->rt_genmask = genmask; 253 /* 254 * Fall into 255 */ 256 case RTM_LOCK: 257 rt->rt_rmx.rmx_locks |= 258 (rtm->rtm_inits & rtm->rtm_rmx.rmx_locks); 259 rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits); 260 break; 261 } 262 goto cleanup; 263 264 default: 265 senderr(EOPNOTSUPP); 266 } 267 268 flush: 269 if (rtm) { 270 if (error) 271 rtm->rtm_errno = error; 272 else 273 rtm->rtm_flags |= RTF_DONE; 274 } 275 cleanup: 276 if (rt) 277 rtfree(rt); 278 { 279 register struct rawcb *rp = 0; 280 /* 281 * Check to see if we don't want our own messages. 282 */ 283 if ((so->so_options & SO_USELOOPBACK) == 0) { 284 if (route_cb.any_count <= 1) { 285 if (rtm) 286 Free(rtm); 287 m_freem(m); 288 return (error); 289 } 290 /* There is another listener, so construct message */ 291 rp = sotorawcb(so); 292 } 293 if (cp = (caddr_t)rtm) { 294 m_copyback(m, 0, len, cp); 295 Free(rtm); 296 } 297 if (rp) 298 rp->rcb_proto.sp_family = 0; /* Avoid us */ 299 route_proto.sp_protocol = dst->sa_family; 300 raw_input(m, &route_proto, &route_src, &route_dst); 301 if (rp) 302 rp->rcb_proto.sp_family = PF_ROUTE; 303 } 304 return (error); 305 } 306 307 rt_setmetrics(which, in, out) 308 u_long which; 309 register struct rt_metrics *in, *out; 310 { 311 #define metric(f, e) if (which & (f)) out->e = in->e; 312 metric(RTV_RPIPE, rmx_recvpipe); 313 metric(RTV_SPIPE, rmx_sendpipe); 314 metric(RTV_SSTHRESH, rmx_ssthresh); 315 metric(RTV_RTT, rmx_rtt); 316 metric(RTV_RTTVAR, rmx_rttvar); 317 metric(RTV_HOPCOUNT, rmx_hopcount); 318 metric(RTV_MTU, rmx_mtu); 319 #undef metric 320 } 321 322 /* 323 * Copy data from a buffer back into the indicated mbuf chain, 324 * starting "off" bytes from the beginning, extending the mbuf 325 * chain if necessary. 326 */ 327 m_copyback(m0, off, len, cp) 328 struct mbuf *m0; 329 register int off; 330 register int len; 331 caddr_t cp; 332 333 { 334 register int mlen; 335 register struct mbuf *m = m0, *n; 336 int totlen = 0; 337 338 if (m0 == 0) 339 return; 340 while (off >= (mlen = m->m_len)) { 341 off -= mlen; 342 totlen += mlen; 343 if (m->m_next == 0) { 344 n = m_getclr(M_DONTWAIT, m->m_type); 345 if (n == 0) 346 goto out; 347 n->m_len = min(MLEN, len + off); 348 m->m_next = n; 349 } 350 m = m->m_next; 351 } 352 while (len > 0) { 353 mlen = min (m->m_len - off, len); 354 bcopy(cp, off + mtod(m, caddr_t), (unsigned)mlen); 355 cp += mlen; 356 len -= mlen; 357 mlen += off; 358 off = 0; 359 totlen += mlen; 360 if (len == 0) 361 break; 362 if (m->m_next == 0) { 363 n = m_get(M_DONTWAIT, m->m_type); 364 if (n == 0) 365 break; 366 n->m_len = min(MLEN, len); 367 m->m_next = n; 368 } 369 m = m->m_next; 370 } 371 out: if (((m = m0)->m_flags & M_PKTHDR) && (m->m_pkthdr.len < totlen)) 372 m->m_pkthdr.len = totlen; 373 } 374 375 /* 376 * The miss message and losing message are very similar. 377 */ 378 379 rt_missmsg(type, dst, gate, mask, src, flags, error) 380 register struct sockaddr *dst; 381 struct sockaddr *gate, *mask, *src; 382 { 383 register struct rt_msghdr *rtm; 384 register struct mbuf *m; 385 int dlen = ROUNDUP(dst->sa_len); 386 int len = dlen + sizeof(*rtm); 387 388 if (route_cb.any_count == 0) 389 return; 390 m = m_gethdr(M_DONTWAIT, MT_DATA); 391 if (m == 0) 392 return; 393 m->m_pkthdr.len = m->m_len = min(len, MHLEN); 394 m->m_pkthdr.rcvif = 0; 395 rtm = mtod(m, struct rt_msghdr *); 396 bzero((caddr_t)rtm, sizeof(*rtm)); /*XXX assumes sizeof(*rtm) < MHLEN*/ 397 rtm->rtm_flags = RTF_DONE | flags; 398 rtm->rtm_msglen = len; 399 rtm->rtm_version = RTM_VERSION; 400 rtm->rtm_type = type; 401 rtm->rtm_addrs = RTA_DST; 402 if (type == RTM_OLDADD || type == RTM_OLDDEL) { 403 rtm->rtm_pid = u.u_procp->p_pid; 404 } 405 m_copyback(m, sizeof (*rtm), dlen, (caddr_t)dst); 406 if (gate) { 407 dlen = ROUNDUP(gate->sa_len); 408 m_copyback(m, len , dlen, (caddr_t)gate); 409 len += dlen; 410 rtm->rtm_addrs |= RTA_GATEWAY; 411 } 412 if (mask) { 413 if (mask->sa_len) 414 dlen = ROUNDUP(mask->sa_len); 415 else 416 dlen = sizeof(long); 417 m_copyback(m, len , dlen, (caddr_t)mask); 418 len += dlen; 419 rtm->rtm_addrs |= RTA_NETMASK; 420 } 421 if (src) { 422 dlen = ROUNDUP(src->sa_len); 423 m_copyback(m, len , dlen, (caddr_t)src); 424 len += dlen; 425 rtm->rtm_addrs |= RTA_AUTHOR; 426 } 427 if (m->m_pkthdr.len != len) { 428 m_freem(m); 429 return; 430 } 431 rtm->rtm_errno = error; 432 rtm->rtm_msglen = len; 433 route_proto.sp_protocol = dst->sa_family; 434 raw_input(m, &route_proto, &route_src, &route_dst); 435 } 436 437 #include "kinfo.h" 438 struct walkarg { 439 int w_op, w_arg; 440 int w_given, w_needed; 441 caddr_t w_where; 442 struct { 443 struct rt_msghdr m_rtm; 444 char m_sabuf[128]; 445 } w_m; 446 #define w_rtm w_m.m_rtm 447 }; 448 /* 449 * This is used in dumping the kernel table via getkinfo(). 450 */ 451 rt_dumpentry(rn, w) 452 struct radix_node *rn; 453 register struct walkarg *w; 454 { 455 register struct sockaddr *sa; 456 int n, error; 457 458 for (; rn; rn = rn->rn_dupedkey) { 459 int count = 0, size = sizeof(w->w_rtm); 460 register struct rtentry *rt = (struct rtentry *)rn; 461 462 if (rn->rn_flags & RNF_ROOT) 463 continue; 464 if (w->w_op == KINFO_RT_FLAGS && !(rt->rt_flags & w->w_arg)) 465 continue; 466 #define next(a, l) {size += (l); w->w_rtm.rtm_addrs |= (a); } 467 w->w_rtm.rtm_addrs = 0; 468 if (sa = rt_key(rt)) 469 next(RTA_DST, ROUNDUP(sa->sa_len)); 470 if (sa = rt->rt_gateway) 471 next(RTA_GATEWAY, ROUNDUP(sa->sa_len)); 472 if (sa = rt_mask(rt)) 473 next(RTA_NETMASK, 474 sa->sa_len ? ROUNDUP(sa->sa_len) : sizeof(long)); 475 if (sa = rt->rt_genmask) 476 next(RTA_GENMASK, ROUNDUP(sa->sa_len)); 477 w->w_needed += size; 478 if (w->w_where == NULL || w->w_needed > 0) 479 continue; 480 w->w_rtm.rtm_msglen = size; 481 w->w_rtm.rtm_flags = rt->rt_flags; 482 w->w_rtm.rtm_use = rt->rt_use; 483 w->w_rtm.rtm_rmx = rt->rt_rmx; 484 w->w_rtm.rtm_index = rt->rt_ifp->if_index; 485 #undef next 486 #define next(l) {n = (l); Bcopy(sa, cp, n); cp += n;} 487 if (size <= sizeof(w->w_m)) { 488 register caddr_t cp = (caddr_t)(w->w_m.m_sabuf); 489 if (sa = rt_key(rt)) 490 next(ROUNDUP(sa->sa_len)); 491 if (sa = rt->rt_gateway) 492 next(ROUNDUP(sa->sa_len)); 493 if (sa = rt_mask(rt)) 494 next(sa->sa_len ? ROUNDUP(sa->sa_len) : sizeof(long)); 495 if (sa = rt->rt_genmask) 496 next(ROUNDUP(sa->sa_len)); 497 #undef next 498 #define next(s, l) {n = (l); \ 499 if (error = copyout((caddr_t)(s), w->w_where, n)) return (error); \ 500 w->w_where += n;} 501 502 next(&w->w_m, size); /* Copy rtmsg and sockaddrs back */ 503 continue; 504 } 505 next(&w->w_rtm, sizeof(w->w_rtm)); 506 if (sa = rt_key(rt)) 507 next(sa, ROUNDUP(sa->sa_len)); 508 if (sa = rt->rt_gateway) 509 next(sa, ROUNDUP(sa->sa_len)); 510 if (sa = rt_mask(rt)) 511 next(sa, sa->sa_len ? ROUNDUP(sa->sa_len) : sizeof(long)); 512 if (sa = rt->rt_genmask) 513 next(sa, ROUNDUP(sa->sa_len)); 514 } 515 return (0); 516 #undef next 517 } 518 519 kinfo_rtable(op, where, given, arg, needed) 520 int op, arg; 521 caddr_t where; 522 int *given, *needed; 523 { 524 register struct radix_node_head *rnh; 525 int s, error = 0; 526 u_char af = ki_af(op); 527 struct walkarg w; 528 529 op &= 0xffff; 530 if (op != KINFO_RT_DUMP && op != KINFO_RT_FLAGS) 531 return (EINVAL); 532 533 Bzero(&w, sizeof(w)); 534 if ((w.w_where = where) && given) 535 w.w_given = *given; 536 w.w_needed = 0 - w.w_given; 537 w.w_arg = arg; 538 w.w_op = op; 539 w.w_rtm.rtm_version = RTM_VERSION; 540 w.w_rtm.rtm_type = RTM_GET; 541 542 s = splnet(); 543 for (rnh = radix_node_head; rnh; rnh = rnh->rnh_next) { 544 if (rnh->rnh_af == 0) 545 continue; 546 if (af && af != rnh->rnh_af) 547 continue; 548 error = rt_walk(rnh->rnh_treetop, rt_dumpentry, &w); 549 if (error) 550 break; 551 } 552 w.w_needed += w.w_given; 553 if (where && given) 554 *given = w.w_where - where; 555 else 556 w.w_needed = (11 * w.w_needed) / 10; 557 *needed = w.w_needed; 558 splx(s); 559 return (error); 560 } 561 562 rt_walk(rn, f, w) 563 register struct radix_node *rn; 564 register int (*f)(); 565 struct walkarg *w; 566 { 567 int error; 568 for (;;) { 569 while (rn->rn_b >= 0) 570 rn = rn->rn_l; /* First time through node, go left */ 571 if (error = (*f)(rn, w)) 572 return (error); /* Process Leaf */ 573 while (rn->rn_p->rn_r == rn) { /* if coming back from right */ 574 rn = rn->rn_p; /* go back up */ 575 if (rn->rn_flags & RNF_ROOT) 576 return 0; 577 } 578 rn = rn->rn_p->rn_r; /* otherwise, go right*/ 579 } 580 } 581 582 /* 583 * Definitions of protocols supported in the ROUTE domain. 584 */ 585 586 int raw_init(),raw_usrreq(),raw_input(),raw_ctlinput(); 587 extern struct domain routedomain; /* or at least forward */ 588 589 struct protosw routesw[] = { 590 { SOCK_RAW, &routedomain, 0, PR_ATOMIC|PR_ADDR, 591 raw_input, route_output, raw_ctlinput, 0, 592 route_usrreq, 593 raw_init, 0, 0, 0, 594 }, 595 { 0, 0, 0, 0, 596 raw_input, 0, raw_ctlinput, 0, 597 raw_usrreq, 598 raw_init, 0, 0, 0, 599 } 600 }; 601 602 int unp_externalize(), unp_dispose(); 603 604 struct domain routedomain = 605 { PF_ROUTE, "route", 0, 0, 0, 606 routesw, &routesw[sizeof(routesw)/sizeof(routesw[0])] }; 607