| 1 | /* |
| 2 | * Copyright (c) 1982, 1989, 1993 |
| 3 | * The Regents of the University of California. All rights reserved. |
| 4 | * |
| 5 | * Redistribution and use in source and binary forms, with or without |
| 6 | * modification, are permitted provided that the following conditions |
| 7 | * are met: |
| 8 | * 1. Redistributions of source code must retain the above copyright |
| 9 | * notice, this list of conditions and the following disclaimer. |
| 10 | * 2. Redistributions in binary form must reproduce the above copyright |
| 11 | * notice, this list of conditions and the following disclaimer in the |
| 12 | * documentation and/or other materials provided with the distribution. |
| 13 | * 3. All advertising materials mentioning features or use of this software |
| 14 | * must display the following acknowledgement: |
| 15 | * This product includes software developed by the University of |
| 16 | * California, Berkeley and its contributors. |
| 17 | * 4. Neither the name of the University nor the names of its contributors |
| 18 | * may be used to endorse or promote products derived from this software |
| 19 | * without specific prior written permission. |
| 20 | * |
| 21 | * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND |
| 22 | * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE |
| 23 | * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE |
| 24 | * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE |
| 25 | * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL |
| 26 | * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS |
| 27 | * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) |
| 28 | * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT |
| 29 | * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY |
| 30 | * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF |
| 31 | * SUCH DAMAGE. |
| 32 | * |
| 33 | * @(#)if_ethersubr.c 8.1 (Berkeley) 6/10/93 |
| 34 | * $FreeBSD: src/sys/net/if_ethersubr.c,v 1.70.2.33 2003/04/28 15:45:53 archie Exp $ |
| 35 | * $DragonFly: src/sys/net/if_ethersubr.c,v 1.67 2008/06/21 03:58:09 sephe Exp $ |
| 36 | */ |
| 37 | |
| 38 | #include "opt_atalk.h" |
| 39 | #include "opt_inet.h" |
| 40 | #include "opt_inet6.h" |
| 41 | #include "opt_ipx.h" |
| 42 | #include "opt_netgraph.h" |
| 43 | #include "opt_carp.h" |
| 44 | #include "opt_ethernet.h" |
| 45 | |
| 46 | #include <sys/param.h> |
| 47 | #include <sys/systm.h> |
| 48 | #include <sys/globaldata.h> |
| 49 | #include <sys/kernel.h> |
| 50 | #include <sys/malloc.h> |
| 51 | #include <sys/mbuf.h> |
| 52 | #include <sys/msgport.h> |
| 53 | #include <sys/socket.h> |
| 54 | #include <sys/sockio.h> |
| 55 | #include <sys/sysctl.h> |
| 56 | #include <sys/thread.h> |
| 57 | #include <sys/thread2.h> |
| 58 | |
| 59 | #include <net/if.h> |
| 60 | #include <net/netisr.h> |
| 61 | #include <net/route.h> |
| 62 | #include <net/if_llc.h> |
| 63 | #include <net/if_dl.h> |
| 64 | #include <net/if_types.h> |
| 65 | #include <net/ifq_var.h> |
| 66 | #include <net/bpf.h> |
| 67 | #include <net/ethernet.h> |
| 68 | #include <net/vlan/if_vlan_ether.h> |
| 69 | |
| 70 | #if defined(INET) || defined(INET6) |
| 71 | #include <netinet/in.h> |
| 72 | #include <netinet/in_var.h> |
| 73 | #include <netinet/if_ether.h> |
| 74 | #include <net/ipfw/ip_fw.h> |
| 75 | #include <net/dummynet/ip_dummynet.h> |
| 76 | #endif |
| 77 | #ifdef INET6 |
| 78 | #include <netinet6/nd6.h> |
| 79 | #endif |
| 80 | |
| 81 | #ifdef CARP |
| 82 | #include <netinet/ip_carp.h> |
| 83 | #endif |
| 84 | |
| 85 | #ifdef IPX |
| 86 | #include <netproto/ipx/ipx.h> |
| 87 | #include <netproto/ipx/ipx_if.h> |
| 88 | int (*ef_inputp)(struct ifnet*, const struct ether_header *eh, struct mbuf *m); |
| 89 | int (*ef_outputp)(struct ifnet *ifp, struct mbuf **mp, struct sockaddr *dst, |
| 90 | short *tp, int *hlen); |
| 91 | #endif |
| 92 | |
| 93 | #ifdef NS |
| 94 | #include <netns/ns.h> |
| 95 | #include <netns/ns_if.h> |
| 96 | ushort ns_nettype; |
| 97 | int ether_outputdebug = 0; |
| 98 | int ether_inputdebug = 0; |
| 99 | #endif |
| 100 | |
| 101 | #ifdef NETATALK |
| 102 | #include <netproto/atalk/at.h> |
| 103 | #include <netproto/atalk/at_var.h> |
| 104 | #include <netproto/atalk/at_extern.h> |
| 105 | |
| 106 | #define llc_snap_org_code llc_un.type_snap.org_code |
| 107 | #define llc_snap_ether_type llc_un.type_snap.ether_type |
| 108 | |
| 109 | extern u_char at_org_code[3]; |
| 110 | extern u_char aarp_org_code[3]; |
| 111 | #endif /* NETATALK */ |
| 112 | |
| 113 | /* netgraph node hooks for ng_ether(4) */ |
| 114 | void (*ng_ether_input_p)(struct ifnet *ifp, struct mbuf **mp); |
| 115 | void (*ng_ether_input_orphan_p)(struct ifnet *ifp, |
| 116 | struct mbuf *m, const struct ether_header *eh); |
| 117 | int (*ng_ether_output_p)(struct ifnet *ifp, struct mbuf **mp); |
| 118 | void (*ng_ether_attach_p)(struct ifnet *ifp); |
| 119 | void (*ng_ether_detach_p)(struct ifnet *ifp); |
| 120 | |
| 121 | int (*vlan_input_p)(struct mbuf *, struct mbuf_chain *); |
| 122 | |
| 123 | static int ether_output(struct ifnet *, struct mbuf *, struct sockaddr *, |
| 124 | struct rtentry *); |
| 125 | static void ether_restore_header(struct mbuf **, const struct ether_header *, |
| 126 | const struct ether_header *); |
| 127 | static void ether_demux_chain(struct ifnet *, struct mbuf *, |
| 128 | struct mbuf_chain *); |
| 129 | |
| 130 | /* |
| 131 | * if_bridge support |
| 132 | */ |
| 133 | struct mbuf *(*bridge_input_p)(struct ifnet *, struct mbuf *); |
| 134 | int (*bridge_output_p)(struct ifnet *, struct mbuf *); |
| 135 | void (*bridge_dn_p)(struct mbuf *, struct ifnet *); |
| 136 | |
| 137 | static int ether_resolvemulti(struct ifnet *, struct sockaddr **, |
| 138 | struct sockaddr *); |
| 139 | |
| 140 | const uint8_t etherbroadcastaddr[ETHER_ADDR_LEN] = { |
| 141 | 0xff, 0xff, 0xff, 0xff, 0xff, 0xff |
| 142 | }; |
| 143 | |
| 144 | #define gotoerr(e) do { error = (e); goto bad; } while (0) |
| 145 | #define IFP2AC(ifp) ((struct arpcom *)(ifp)) |
| 146 | |
| 147 | static boolean_t ether_ipfw_chk(struct mbuf **m0, struct ifnet *dst, |
| 148 | struct ip_fw **rule, |
| 149 | const struct ether_header *eh); |
| 150 | |
| 151 | static int ether_ipfw; |
| 152 | static u_int ether_restore_hdr; |
| 153 | static u_int ether_prepend_hdr; |
| 154 | |
| 155 | SYSCTL_DECL(_net_link); |
| 156 | SYSCTL_NODE(_net_link, IFT_ETHER, ether, CTLFLAG_RW, 0, "Ethernet"); |
| 157 | SYSCTL_INT(_net_link_ether, OID_AUTO, ipfw, CTLFLAG_RW, |
| 158 | ðer_ipfw, 0, "Pass ether pkts through firewall"); |
| 159 | SYSCTL_UINT(_net_link_ether, OID_AUTO, restore_hdr, CTLFLAG_RW, |
| 160 | ðer_restore_hdr, 0, "# of ether header restoration"); |
| 161 | SYSCTL_UINT(_net_link_ether, OID_AUTO, prepend_hdr, CTLFLAG_RW, |
| 162 | ðer_prepend_hdr, 0, |
| 163 | "# of ether header restoration which prepends mbuf"); |
| 164 | |
| 165 | /* |
| 166 | * Ethernet output routine. |
| 167 | * Encapsulate a packet of type family for the local net. |
| 168 | * Use trailer local net encapsulation if enough data in first |
| 169 | * packet leaves a multiple of 512 bytes of data in remainder. |
| 170 | * Assumes that ifp is actually pointer to arpcom structure. |
| 171 | */ |
| 172 | static int |
| 173 | ether_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *dst, |
| 174 | struct rtentry *rt) |
| 175 | { |
| 176 | struct ether_header *eh, *deh; |
| 177 | u_char *edst; |
| 178 | int loop_copy = 0; |
| 179 | int hlen = ETHER_HDR_LEN; /* link layer header length */ |
| 180 | struct arpcom *ac = IFP2AC(ifp); |
| 181 | int error; |
| 182 | |
| 183 | ASSERT_NOT_SERIALIZED(ifp->if_serializer); |
| 184 | |
| 185 | if (ifp->if_flags & IFF_MONITOR) |
| 186 | gotoerr(ENETDOWN); |
| 187 | if ((ifp->if_flags & (IFF_UP | IFF_RUNNING)) != (IFF_UP | IFF_RUNNING)) |
| 188 | gotoerr(ENETDOWN); |
| 189 | |
| 190 | M_PREPEND(m, sizeof(struct ether_header), MB_DONTWAIT); |
| 191 | if (m == NULL) |
| 192 | return (ENOBUFS); |
| 193 | eh = mtod(m, struct ether_header *); |
| 194 | edst = eh->ether_dhost; |
| 195 | |
| 196 | /* |
| 197 | * Fill in the destination ethernet address and frame type. |
| 198 | */ |
| 199 | switch (dst->sa_family) { |
| 200 | #ifdef INET |
| 201 | case AF_INET: |
| 202 | if (!arpresolve(ifp, rt, m, dst, edst)) |
| 203 | return (0); /* if not yet resolved */ |
| 204 | eh->ether_type = htons(ETHERTYPE_IP); |
| 205 | break; |
| 206 | #endif |
| 207 | #ifdef INET6 |
| 208 | case AF_INET6: |
| 209 | if (!nd6_storelladdr(&ac->ac_if, rt, m, dst, edst)) |
| 210 | return (0); /* Something bad happenned. */ |
| 211 | eh->ether_type = htons(ETHERTYPE_IPV6); |
| 212 | break; |
| 213 | #endif |
| 214 | #ifdef IPX |
| 215 | case AF_IPX: |
| 216 | if (ef_outputp != NULL) { |
| 217 | error = ef_outputp(ifp, &m, dst, &eh->ether_type, |
| 218 | &hlen); |
| 219 | if (error) |
| 220 | goto bad; |
| 221 | } else { |
| 222 | eh->ether_type = htons(ETHERTYPE_IPX); |
| 223 | bcopy(&(((struct sockaddr_ipx *)dst)->sipx_addr.x_host), |
| 224 | edst, ETHER_ADDR_LEN); |
| 225 | } |
| 226 | break; |
| 227 | #endif |
| 228 | #ifdef NETATALK |
| 229 | case AF_APPLETALK: { |
| 230 | struct at_ifaddr *aa; |
| 231 | |
| 232 | if ((aa = at_ifawithnet((struct sockaddr_at *)dst)) == NULL) { |
| 233 | error = 0; /* XXX */ |
| 234 | goto bad; |
| 235 | } |
| 236 | /* |
| 237 | * In the phase 2 case, need to prepend an mbuf for |
| 238 | * the llc header. Since we must preserve the value |
| 239 | * of m, which is passed to us by value, we m_copy() |
| 240 | * the first mbuf, and use it for our llc header. |
| 241 | */ |
| 242 | if (aa->aa_flags & AFA_PHASE2) { |
| 243 | struct llc llc; |
| 244 | |
| 245 | M_PREPEND(m, sizeof(struct llc), MB_DONTWAIT); |
| 246 | eh = mtod(m, struct ether_header *); |
| 247 | edst = eh->ether_dhost; |
| 248 | llc.llc_dsap = llc.llc_ssap = LLC_SNAP_LSAP; |
| 249 | llc.llc_control = LLC_UI; |
| 250 | bcopy(at_org_code, llc.llc_snap_org_code, |
| 251 | sizeof at_org_code); |
| 252 | llc.llc_snap_ether_type = htons(ETHERTYPE_AT); |
| 253 | bcopy(&llc, |
| 254 | mtod(m, caddr_t) + sizeof(struct ether_header), |
| 255 | sizeof(struct llc)); |
| 256 | eh->ether_type = htons(m->m_pkthdr.len); |
| 257 | hlen = sizeof(struct llc) + ETHER_HDR_LEN; |
| 258 | } else { |
| 259 | eh->ether_type = htons(ETHERTYPE_AT); |
| 260 | } |
| 261 | if (!aarpresolve(ac, m, (struct sockaddr_at *)dst, edst)) |
| 262 | return (0); |
| 263 | break; |
| 264 | } |
| 265 | #endif |
| 266 | #ifdef NS |
| 267 | case AF_NS: |
| 268 | switch(ns_nettype) { |
| 269 | default: |
| 270 | case 0x8137: /* Novell Ethernet_II Ethernet TYPE II */ |
| 271 | eh->ether_type = 0x8137; |
| 272 | break; |
| 273 | case 0x0: /* Novell 802.3 */ |
| 274 | eh->ether_type = htons(m->m_pkthdr.len); |
| 275 | break; |
| 276 | case 0xe0e0: /* Novell 802.2 and Token-Ring */ |
| 277 | M_PREPEND(m, 3, MB_DONTWAIT); |
| 278 | eh = mtod(m, struct ether_header *); |
| 279 | edst = eh->ether_dhost; |
| 280 | eh->ether_type = htons(m->m_pkthdr.len); |
| 281 | cp = mtod(m, u_char *) + sizeof(struct ether_header); |
| 282 | *cp++ = 0xE0; |
| 283 | *cp++ = 0xE0; |
| 284 | *cp++ = 0x03; |
| 285 | break; |
| 286 | } |
| 287 | bcopy(&(((struct sockaddr_ns *)dst)->sns_addr.x_host), edst, |
| 288 | ETHER_ADDR_LEN); |
| 289 | /* |
| 290 | * XXX if ns_thishost is the same as the node's ethernet |
| 291 | * address then just the default code will catch this anyhow. |
| 292 | * So I'm not sure if this next clause should be here at all? |
| 293 | * [JRE] |
| 294 | */ |
| 295 | if (bcmp(edst, &ns_thishost, ETHER_ADDR_LEN) == 0) { |
| 296 | m->m_pkthdr.rcvif = ifp; |
| 297 | netisr_dispatch(NETISR_NS, m); |
| 298 | return (error); |
| 299 | } |
| 300 | if (bcmp(edst, &ns_broadhost, ETHER_ADDR_LEN) == 0) |
| 301 | m->m_flags |= M_BCAST; |
| 302 | break; |
| 303 | #endif |
| 304 | case pseudo_AF_HDRCMPLT: |
| 305 | case AF_UNSPEC: |
| 306 | loop_copy = -1; /* if this is for us, don't do it */ |
| 307 | deh = (struct ether_header *)dst->sa_data; |
| 308 | memcpy(edst, deh->ether_dhost, ETHER_ADDR_LEN); |
| 309 | eh->ether_type = deh->ether_type; |
| 310 | break; |
| 311 | |
| 312 | default: |
| 313 | if_printf(ifp, "can't handle af%d\n", dst->sa_family); |
| 314 | gotoerr(EAFNOSUPPORT); |
| 315 | } |
| 316 | |
| 317 | if (dst->sa_family == pseudo_AF_HDRCMPLT) /* unlikely */ |
| 318 | memcpy(eh->ether_shost, |
| 319 | ((struct ether_header *)dst->sa_data)->ether_shost, |
| 320 | ETHER_ADDR_LEN); |
| 321 | else |
| 322 | memcpy(eh->ether_shost, ac->ac_enaddr, ETHER_ADDR_LEN); |
| 323 | |
| 324 | /* |
| 325 | * Bridges require special output handling. |
| 326 | */ |
| 327 | if (ifp->if_bridge) { |
| 328 | KASSERT(bridge_output_p != NULL, |
| 329 | ("%s: if_bridge not loaded!", __func__)); |
| 330 | return bridge_output_p(ifp, m); |
| 331 | } |
| 332 | |
| 333 | /* |
| 334 | * If a simplex interface, and the packet is being sent to our |
| 335 | * Ethernet address or a broadcast address, loopback a copy. |
| 336 | * XXX To make a simplex device behave exactly like a duplex |
| 337 | * device, we should copy in the case of sending to our own |
| 338 | * ethernet address (thus letting the original actually appear |
| 339 | * on the wire). However, we don't do that here for security |
| 340 | * reasons and compatibility with the original behavior. |
| 341 | */ |
| 342 | if ((ifp->if_flags & IFF_SIMPLEX) && (loop_copy != -1)) { |
| 343 | int csum_flags = 0; |
| 344 | |
| 345 | if (m->m_pkthdr.csum_flags & CSUM_IP) |
| 346 | csum_flags |= (CSUM_IP_CHECKED | CSUM_IP_VALID); |
| 347 | if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA) |
| 348 | csum_flags |= (CSUM_DATA_VALID | CSUM_PSEUDO_HDR); |
| 349 | if ((m->m_flags & M_BCAST) || (loop_copy > 0)) { |
| 350 | struct mbuf *n; |
| 351 | |
| 352 | if ((n = m_copypacket(m, MB_DONTWAIT)) != NULL) { |
| 353 | n->m_pkthdr.csum_flags |= csum_flags; |
| 354 | if (csum_flags & CSUM_DATA_VALID) |
| 355 | n->m_pkthdr.csum_data = 0xffff; |
| 356 | if_simloop(ifp, n, dst->sa_family, hlen); |
| 357 | } else |
| 358 | ifp->if_iqdrops++; |
| 359 | } else if (bcmp(eh->ether_dhost, eh->ether_shost, |
| 360 | ETHER_ADDR_LEN) == 0) { |
| 361 | m->m_pkthdr.csum_flags |= csum_flags; |
| 362 | if (csum_flags & CSUM_DATA_VALID) |
| 363 | m->m_pkthdr.csum_data = 0xffff; |
| 364 | if_simloop(ifp, m, dst->sa_family, hlen); |
| 365 | return (0); /* XXX */ |
| 366 | } |
| 367 | } |
| 368 | |
| 369 | #ifdef CARP |
| 370 | if (ifp->if_carp && (error = carp_output(ifp, m, dst, NULL))) |
| 371 | goto bad; |
| 372 | #endif |
| 373 | |
| 374 | |
| 375 | /* Handle ng_ether(4) processing, if any */ |
| 376 | if (ng_ether_output_p != NULL) { |
| 377 | if ((error = (*ng_ether_output_p)(ifp, &m)) != 0) |
| 378 | goto bad; |
| 379 | if (m == NULL) |
| 380 | return (0); |
| 381 | } |
| 382 | |
| 383 | /* Continue with link-layer output */ |
| 384 | return ether_output_frame(ifp, m); |
| 385 | |
| 386 | bad: |
| 387 | m_freem(m); |
| 388 | return (error); |
| 389 | } |
| 390 | |
| 391 | /* |
| 392 | * Ethernet link layer output routine to send a raw frame to the device. |
| 393 | * |
| 394 | * This assumes that the 14 byte Ethernet header is present and contiguous |
| 395 | * in the first mbuf. |
| 396 | */ |
| 397 | int |
| 398 | ether_output_frame(struct ifnet *ifp, struct mbuf *m) |
| 399 | { |
| 400 | struct ip_fw *rule = NULL; |
| 401 | int error = 0; |
| 402 | struct altq_pktattr pktattr; |
| 403 | struct m_tag *mtag; |
| 404 | |
| 405 | ASSERT_NOT_SERIALIZED(ifp->if_serializer); |
| 406 | |
| 407 | /* Extract info from dummynet tag */ |
| 408 | mtag = m_tag_find(m, PACKET_TAG_DUMMYNET, NULL); |
| 409 | if (mtag != NULL) { |
| 410 | rule = ((struct dn_pkt *)m_tag_data(mtag))->dn_priv; |
| 411 | |
| 412 | m_tag_delete(m, mtag); |
| 413 | mtag = NULL; |
| 414 | } |
| 415 | |
| 416 | if (ifq_is_enabled(&ifp->if_snd)) |
| 417 | altq_etherclassify(&ifp->if_snd, m, &pktattr); |
| 418 | crit_enter(); |
| 419 | if (IPFW_LOADED && ether_ipfw != 0) { |
| 420 | struct ether_header save_eh, *eh; |
| 421 | |
| 422 | eh = mtod(m, struct ether_header *); |
| 423 | save_eh = *eh; |
| 424 | m_adj(m, ETHER_HDR_LEN); |
| 425 | if (!ether_ipfw_chk(&m, ifp, &rule, eh)) { |
| 426 | crit_exit(); |
| 427 | if (m != NULL) { |
| 428 | m_freem(m); |
| 429 | return ENOBUFS; /* pkt dropped */ |
| 430 | } else |
| 431 | return 0; /* consumed e.g. in a pipe */ |
| 432 | } |
| 433 | |
| 434 | /* packet was ok, restore the ethernet header */ |
| 435 | ether_restore_header(&m, eh, &save_eh); |
| 436 | if (m == NULL) { |
| 437 | crit_exit(); |
| 438 | return ENOBUFS; |
| 439 | } |
| 440 | } |
| 441 | crit_exit(); |
| 442 | |
| 443 | /* |
| 444 | * Queue message on interface, update output statistics if |
| 445 | * successful, and start output if interface not yet active. |
| 446 | */ |
| 447 | error = ifq_dispatch(ifp, m, &pktattr); |
| 448 | return (error); |
| 449 | } |
| 450 | |
| 451 | /* |
| 452 | * ipfw processing for ethernet packets (in and out). |
| 453 | * The second parameter is NULL from ether_demux(), and ifp from |
| 454 | * ether_output_frame(). |
| 455 | */ |
| 456 | static boolean_t |
| 457 | ether_ipfw_chk(struct mbuf **m0, struct ifnet *dst, struct ip_fw **rule, |
| 458 | const struct ether_header *eh) |
| 459 | { |
| 460 | struct ether_header save_eh = *eh; /* might be a ptr in m */ |
| 461 | struct ip_fw_args args; |
| 462 | struct m_tag *mtag; |
| 463 | int i; |
| 464 | |
| 465 | if (*rule != NULL && fw_one_pass) |
| 466 | return TRUE; /* dummynet packet, already partially processed */ |
| 467 | |
| 468 | /* |
| 469 | * I need some amount of data to be contiguous. |
| 470 | */ |
| 471 | i = min((*m0)->m_pkthdr.len, max_protohdr); |
| 472 | if ((*m0)->m_len < i) { |
| 473 | *m0 = m_pullup(*m0, i); |
| 474 | if (*m0 == NULL) |
| 475 | return FALSE; |
| 476 | } |
| 477 | |
| 478 | args.m = *m0; /* the packet we are looking at */ |
| 479 | args.oif = dst; /* destination, if any */ |
| 480 | if ((mtag = m_tag_find(*m0, PACKET_TAG_IPFW_DIVERT, NULL)) != NULL) |
| 481 | m_tag_delete(*m0, mtag); |
| 482 | args.rule = *rule; /* matching rule to restart */ |
| 483 | args.next_hop = NULL; /* we do not support forward yet */ |
| 484 | args.eh = &save_eh; /* MAC header for bridged/MAC packets */ |
| 485 | i = ip_fw_chk_ptr(&args); |
| 486 | *m0 = args.m; |
| 487 | *rule = args.rule; |
| 488 | |
| 489 | if ((i & IP_FW_PORT_DENY_FLAG) || *m0 == NULL) /* drop */ |
| 490 | return FALSE; |
| 491 | |
| 492 | if (i == 0) /* a PASS rule. */ |
| 493 | return TRUE; |
| 494 | |
| 495 | if (i & IP_FW_PORT_DYNT_FLAG) { |
| 496 | /* |
| 497 | * Pass the pkt to dummynet, which consumes it. |
| 498 | */ |
| 499 | struct mbuf *m; |
| 500 | |
| 501 | m = *m0; /* pass the original to dummynet */ |
| 502 | *m0 = NULL; /* and nothing back to the caller */ |
| 503 | |
| 504 | ether_restore_header(&m, eh, &save_eh); |
| 505 | if (m == NULL) |
| 506 | return FALSE; |
| 507 | |
| 508 | ip_fw_dn_io_ptr(m, (i & 0xffff), |
| 509 | dst ? DN_TO_ETH_OUT: DN_TO_ETH_DEMUX, &args); |
| 510 | return FALSE; |
| 511 | } |
| 512 | /* |
| 513 | * XXX at some point add support for divert/forward actions. |
| 514 | * If none of the above matches, we have to drop the pkt. |
| 515 | */ |
| 516 | return FALSE; |
| 517 | } |
| 518 | |
| 519 | /* |
| 520 | * Process a received Ethernet packet. |
| 521 | * |
| 522 | * The ethernet header is assumed to be in the mbuf so the caller |
| 523 | * MUST MAKE SURE that there are at least sizeof(struct ether_header) |
| 524 | * bytes in the first mbuf. |
| 525 | * |
| 526 | * This allows us to concentrate in one place a bunch of code which |
| 527 | * is replicated in all device drivers. Also, many functions called |
| 528 | * from ether_input() try to put the eh back into the mbuf, so we |
| 529 | * can later propagate the 'contiguous packet' interface to them. |
| 530 | * |
| 531 | * NOTA BENE: for all drivers "eh" is a pointer into the first mbuf or |
| 532 | * cluster, right before m_data. So be very careful when working on m, |
| 533 | * as you could destroy *eh !! |
| 534 | * |
| 535 | * First we perform any link layer operations, then continue to the |
| 536 | * upper layers with ether_demux(). |
| 537 | */ |
| 538 | void |
| 539 | ether_input_chain(struct ifnet *ifp, struct mbuf *m, struct mbuf_chain *chain) |
| 540 | { |
| 541 | struct ether_header *eh; |
| 542 | |
| 543 | ASSERT_SERIALIZED(ifp->if_serializer); |
| 544 | M_ASSERTPKTHDR(m); |
| 545 | |
| 546 | /* Discard packet if interface is not up */ |
| 547 | if (!(ifp->if_flags & IFF_UP)) { |
| 548 | m_freem(m); |
| 549 | return; |
| 550 | } |
| 551 | |
| 552 | if (m->m_len < sizeof(struct ether_header)) { |
| 553 | /* XXX error in the caller. */ |
| 554 | m_freem(m); |
| 555 | return; |
| 556 | } |
| 557 | eh = mtod(m, struct ether_header *); |
| 558 | |
| 559 | if (ntohs(eh->ether_type) == ETHERTYPE_VLAN && |
| 560 | (m->m_flags & M_VLANTAG) == 0) { |
| 561 | /* |
| 562 | * Extract vlan tag if hardware does not do it for us |
| 563 | */ |
| 564 | vlan_ether_decap(&m); |
| 565 | if (m == NULL) |
| 566 | return; |
| 567 | eh = mtod(m, struct ether_header *); |
| 568 | } |
| 569 | |
| 570 | m->m_pkthdr.rcvif = ifp; |
| 571 | |
| 572 | if (ETHER_IS_MULTICAST(eh->ether_dhost)) { |
| 573 | if (bcmp(ifp->if_broadcastaddr, eh->ether_dhost, |
| 574 | ifp->if_addrlen) == 0) |
| 575 | m->m_flags |= M_BCAST; |
| 576 | else |
| 577 | m->m_flags |= M_MCAST; |
| 578 | ifp->if_imcasts++; |
| 579 | } |
| 580 | |
| 581 | ETHER_BPF_MTAP(ifp, m); |
| 582 | |
| 583 | ifp->if_ibytes += m->m_pkthdr.len; |
| 584 | |
| 585 | if (ifp->if_flags & IFF_MONITOR) { |
| 586 | /* |
| 587 | * Interface marked for monitoring; discard packet. |
| 588 | */ |
| 589 | m_freem(m); |
| 590 | return; |
| 591 | } |
| 592 | |
| 593 | /* |
| 594 | * Tap the packet off here for a bridge. bridge_input() |
| 595 | * will return NULL if it has consumed the packet, otherwise |
| 596 | * it gets processed as normal. Note that bridge_input() |
| 597 | * will always return the original packet if we need to |
| 598 | * process it locally. |
| 599 | */ |
| 600 | if (ifp->if_bridge) { |
| 601 | KASSERT(bridge_input_p != NULL, |
| 602 | ("%s: if_bridge not loaded!", __func__)); |
| 603 | |
| 604 | if(m->m_flags & M_PROTO1) { |
| 605 | m->m_flags &= ~M_PROTO1; |
| 606 | } else { |
| 607 | /* clear M_PROMISC, in case the packets comes from a vlan */ |
| 608 | /* m->m_flags &= ~M_PROMISC; */ |
| 609 | lwkt_serialize_exit(ifp->if_serializer); |
| 610 | m = bridge_input_p(ifp, m); |
| 611 | lwkt_serialize_enter(ifp->if_serializer); |
| 612 | if (m == NULL) |
| 613 | return; |
| 614 | |
| 615 | KASSERT(ifp == m->m_pkthdr.rcvif, |
| 616 | ("bridge_input_p changed rcvif\n")); |
| 617 | |
| 618 | /* 'm' may be changed by bridge_input_p() */ |
| 619 | eh = mtod(m, struct ether_header *); |
| 620 | } |
| 621 | } |
| 622 | |
| 623 | /* Handle ng_ether(4) processing, if any */ |
| 624 | if (ng_ether_input_p != NULL) { |
| 625 | ng_ether_input_p(ifp, &m); |
| 626 | if (m == NULL) |
| 627 | return; |
| 628 | |
| 629 | /* 'm' may be changed by ng_ether_input_p() */ |
| 630 | eh = mtod(m, struct ether_header *); |
| 631 | } |
| 632 | |
| 633 | /* Continue with upper layer processing */ |
| 634 | ether_demux_chain(ifp, m, chain); |
| 635 | } |
| 636 | |
| 637 | void |
| 638 | ether_input(struct ifnet *ifp, struct mbuf *m) |
| 639 | { |
| 640 | ether_input_chain(ifp, m, NULL); |
| 641 | } |
| 642 | |
| 643 | /* |
| 644 | * Upper layer processing for a received Ethernet packet. |
| 645 | */ |
| 646 | static void |
| 647 | ether_demux_chain(struct ifnet *ifp, struct mbuf *m, struct mbuf_chain *chain) |
| 648 | { |
| 649 | struct ether_header save_eh, *eh; |
| 650 | int isr; |
| 651 | u_short ether_type; |
| 652 | struct ip_fw *rule = NULL; |
| 653 | struct m_tag *mtag; |
| 654 | #ifdef NETATALK |
| 655 | struct llc *l; |
| 656 | #endif |
| 657 | |
| 658 | M_ASSERTPKTHDR(m); |
| 659 | KASSERT(m->m_len >= ETHER_HDR_LEN, |
| 660 | ("ether header is no contiguous!\n")); |
| 661 | |
| 662 | eh = mtod(m, struct ether_header *); |
| 663 | save_eh = *eh; |
| 664 | |
| 665 | /* XXX old crufty stuff, needs to be removed */ |
| 666 | m_adj(m, sizeof(struct ether_header)); |
| 667 | |
| 668 | /* Extract info from dummynet tag */ |
| 669 | mtag = m_tag_find(m, PACKET_TAG_DUMMYNET, NULL); |
| 670 | if (mtag != NULL) { |
| 671 | rule = ((struct dn_pkt *)m_tag_data(mtag))->dn_priv; |
| 672 | KKASSERT(ifp == NULL); |
| 673 | ifp = m->m_pkthdr.rcvif; |
| 674 | |
| 675 | m_tag_delete(m, mtag); |
| 676 | mtag = NULL; |
| 677 | } |
| 678 | if (rule) /* packet is passing the second time */ |
| 679 | goto post_stats; |
| 680 | |
| 681 | #ifdef CARP |
| 682 | /* |
| 683 | * XXX: Okay, we need to call carp_forus() and - if it is for |
| 684 | * us jump over code that does the normal check |
| 685 | * "ac_enaddr == ether_dhost". The check sequence is a bit |
| 686 | * different from OpenBSD, so we jump over as few code as |
| 687 | * possible, to catch _all_ sanity checks. This needs |
| 688 | * evaluation, to see if the carp ether_dhost values break any |
| 689 | * of these checks! |
| 690 | */ |
| 691 | if (ifp->if_carp && carp_forus(ifp->if_carp, eh->ether_dhost)) |
| 692 | goto post_stats; |
| 693 | #endif |
| 694 | |
| 695 | /* |
| 696 | * Discard packet if upper layers shouldn't see it because |
| 697 | * it was unicast to a different Ethernet address. If the |
| 698 | * driver is working properly, then this situation can only |
| 699 | * happen when the interface is in promiscuous mode. |
| 700 | */ |
| 701 | if (((ifp->if_flags & (IFF_PROMISC | IFF_PPROMISC)) == IFF_PROMISC) && |
| 702 | (eh->ether_dhost[0] & 1) == 0 && |
| 703 | bcmp(eh->ether_dhost, IFP2AC(ifp)->ac_enaddr, ETHER_ADDR_LEN)) { |
| 704 | m_freem(m); |
| 705 | return; |
| 706 | } |
| 707 | |
| 708 | post_stats: |
| 709 | if (IPFW_LOADED && ether_ipfw != 0) { |
| 710 | if (!ether_ipfw_chk(&m, NULL, &rule, eh)) { |
| 711 | m_freem(m); |
| 712 | return; |
| 713 | } |
| 714 | } |
| 715 | |
| 716 | ether_type = ntohs(save_eh.ether_type); |
| 717 | |
| 718 | if (m->m_flags & M_VLANTAG) { |
| 719 | if (ether_type == ETHERTYPE_VLAN) { |
| 720 | /* |
| 721 | * To prevent possible dangerous recursion, |
| 722 | * we don't do vlan-in-vlan |
| 723 | */ |
| 724 | m->m_pkthdr.rcvif->if_noproto++; |
| 725 | m_freem(m); |
| 726 | return; |
| 727 | } |
| 728 | |
| 729 | if (vlan_input_p != NULL) { |
| 730 | ether_restore_header(&m, eh, &save_eh); |
| 731 | if (m != NULL) |
| 732 | vlan_input_p(m, chain); |
| 733 | } else { |
| 734 | m->m_pkthdr.rcvif->if_noproto++; |
| 735 | m_freem(m); |
| 736 | } |
| 737 | return; |
| 738 | } |
| 739 | KKASSERT(ether_type != ETHERTYPE_VLAN); |
| 740 | |
| 741 | switch (ether_type) { |
| 742 | #ifdef INET |
| 743 | case ETHERTYPE_IP: |
| 744 | if (ipflow_fastforward(m, ifp->if_serializer)) |
| 745 | return; |
| 746 | isr = NETISR_IP; |
| 747 | break; |
| 748 | |
| 749 | case ETHERTYPE_ARP: |
| 750 | if (ifp->if_flags & IFF_NOARP) { |
| 751 | /* Discard packet if ARP is disabled on interface */ |
| 752 | m_freem(m); |
| 753 | return; |
| 754 | } |
| 755 | isr = NETISR_ARP; |
| 756 | break; |
| 757 | #endif |
| 758 | |
| 759 | #ifdef INET6 |
| 760 | case ETHERTYPE_IPV6: |
| 761 | isr = NETISR_IPV6; |
| 762 | break; |
| 763 | #endif |
| 764 | |
| 765 | #ifdef IPX |
| 766 | case ETHERTYPE_IPX: |
| 767 | if (ef_inputp && ef_inputp(ifp, &save_eh, m) == 0) |
| 768 | return; |
| 769 | isr = NETISR_IPX; |
| 770 | break; |
| 771 | #endif |
| 772 | |
| 773 | #ifdef NS |
| 774 | case 0x8137: /* Novell Ethernet_II Ethernet TYPE II */ |
| 775 | isr = NETISR_NS; |
| 776 | break; |
| 777 | |
| 778 | #endif |
| 779 | |
| 780 | #ifdef NETATALK |
| 781 | case ETHERTYPE_AT: |
| 782 | isr = NETISR_ATALK1; |
| 783 | break; |
| 784 | case ETHERTYPE_AARP: |
| 785 | isr = NETISR_AARP; |
| 786 | break; |
| 787 | #endif |
| 788 | |
| 789 | default: |
| 790 | #ifdef IPX |
| 791 | if (ef_inputp && ef_inputp(ifp, &save_eh, m) == 0) |
| 792 | return; |
| 793 | #endif |
| 794 | #ifdef NS |
| 795 | checksum = mtod(m, ushort *); |
| 796 | /* Novell 802.3 */ |
| 797 | if ((ether_type <= ETHERMTU) && |
| 798 | ((*checksum == 0xffff) || (*checksum == 0xE0E0))) { |
| 799 | if (*checksum == 0xE0E0) { |
| 800 | m->m_pkthdr.len -= 3; |
| 801 | m->m_len -= 3; |
| 802 | m->m_data += 3; |
| 803 | } |
| 804 | isr = NETISR_NS; |
| 805 | break; |
| 806 | } |
| 807 | #endif |
| 808 | #ifdef NETATALK |
| 809 | if (ether_type > ETHERMTU) |
| 810 | goto dropanyway; |
| 811 | l = mtod(m, struct llc *); |
| 812 | if (l->llc_dsap == LLC_SNAP_LSAP && |
| 813 | l->llc_ssap == LLC_SNAP_LSAP && |
| 814 | l->llc_control == LLC_UI) { |
| 815 | if (bcmp(&(l->llc_snap_org_code)[0], at_org_code, |
| 816 | sizeof at_org_code) == 0 && |
| 817 | ntohs(l->llc_snap_ether_type) == ETHERTYPE_AT) { |
| 818 | m_adj(m, sizeof(struct llc)); |
| 819 | isr = NETISR_ATALK2; |
| 820 | break; |
| 821 | } |
| 822 | if (bcmp(&(l->llc_snap_org_code)[0], aarp_org_code, |
| 823 | sizeof aarp_org_code) == 0 && |
| 824 | ntohs(l->llc_snap_ether_type) == ETHERTYPE_AARP) { |
| 825 | m_adj(m, sizeof(struct llc)); |
| 826 | isr = NETISR_AARP; |
| 827 | break; |
| 828 | } |
| 829 | } |
| 830 | dropanyway: |
| 831 | #endif |
| 832 | if (ng_ether_input_orphan_p != NULL) |
| 833 | (*ng_ether_input_orphan_p)(ifp, m, &save_eh); |
| 834 | else |
| 835 | m_freem(m); |
| 836 | return; |
| 837 | } |
| 838 | |
| 839 | #ifdef ETHER_INPUT_CHAIN |
| 840 | if (chain != NULL) { |
| 841 | struct mbuf_chain *c; |
| 842 | lwkt_port_t port; |
| 843 | int cpuid; |
| 844 | |
| 845 | port = netisr_mport(isr, &m); |
| 846 | if (port == NULL) |
| 847 | return; |
| 848 | |
| 849 | m->m_pkthdr.header = port; /* XXX */ |
| 850 | cpuid = port->mpu_td->td_gd->gd_cpuid; |
| 851 | |
| 852 | c = &chain[cpuid]; |
| 853 | if (c->mc_head == NULL) { |
| 854 | c->mc_head = c->mc_tail = m; |
| 855 | } else { |
| 856 | c->mc_tail->m_nextpkt = m; |
| 857 | c->mc_tail = m; |
| 858 | } |
| 859 | m->m_nextpkt = NULL; |
| 860 | } else |
| 861 | #endif /* ETHER_INPUT_CHAIN */ |
| 862 | netisr_dispatch(isr, m); |
| 863 | } |
| 864 | |
| 865 | void |
| 866 | ether_demux(struct ifnet *ifp, struct mbuf *m) |
| 867 | { |
| 868 | ether_demux_chain(ifp, m, NULL); |
| 869 | } |
| 870 | |
| 871 | /* |
| 872 | * Perform common duties while attaching to interface list |
| 873 | */ |
| 874 | |
| 875 | void |
| 876 | ether_ifattach(struct ifnet *ifp, uint8_t *lla, lwkt_serialize_t serializer) |
| 877 | { |
| 878 | ether_ifattach_bpf(ifp, lla, DLT_EN10MB, sizeof(struct ether_header), |
| 879 | serializer); |
| 880 | } |
| 881 | |
| 882 | void |
| 883 | ether_ifattach_bpf(struct ifnet *ifp, uint8_t *lla, u_int dlt, u_int hdrlen, |
| 884 | lwkt_serialize_t serializer) |
| 885 | { |
| 886 | struct sockaddr_dl *sdl; |
| 887 | |
| 888 | ifp->if_type = IFT_ETHER; |
| 889 | ifp->if_addrlen = ETHER_ADDR_LEN; |
| 890 | ifp->if_hdrlen = ETHER_HDR_LEN; |
| 891 | if_attach(ifp, serializer); |
| 892 | ifp->if_mtu = ETHERMTU; |
| 893 | if (ifp->if_baudrate == 0) |
| 894 | ifp->if_baudrate = 10000000; |
| 895 | ifp->if_output = ether_output; |
| 896 | ifp->if_input = ether_input; |
| 897 | ifp->if_resolvemulti = ether_resolvemulti; |
| 898 | ifp->if_broadcastaddr = etherbroadcastaddr; |
| 899 | sdl = IF_LLSOCKADDR(ifp); |
| 900 | sdl->sdl_type = IFT_ETHER; |
| 901 | sdl->sdl_alen = ifp->if_addrlen; |
| 902 | bcopy(lla, LLADDR(sdl), ifp->if_addrlen); |
| 903 | /* |
| 904 | * XXX Keep the current drivers happy. |
| 905 | * XXX Remove once all drivers have been cleaned up |
| 906 | */ |
| 907 | if (lla != IFP2AC(ifp)->ac_enaddr) |
| 908 | bcopy(lla, IFP2AC(ifp)->ac_enaddr, ifp->if_addrlen); |
| 909 | bpfattach(ifp, dlt, hdrlen); |
| 910 | if (ng_ether_attach_p != NULL) |
| 911 | (*ng_ether_attach_p)(ifp); |
| 912 | |
| 913 | if_printf(ifp, "MAC address: %6D\n", lla, ":"); |
| 914 | } |
| 915 | |
| 916 | /* |
| 917 | * Perform common duties while detaching an Ethernet interface |
| 918 | */ |
| 919 | void |
| 920 | ether_ifdetach(struct ifnet *ifp) |
| 921 | { |
| 922 | if_down(ifp); |
| 923 | |
| 924 | if (ng_ether_detach_p != NULL) |
| 925 | (*ng_ether_detach_p)(ifp); |
| 926 | bpfdetach(ifp); |
| 927 | if_detach(ifp); |
| 928 | } |
| 929 | |
| 930 | int |
| 931 | ether_ioctl(struct ifnet *ifp, int command, caddr_t data) |
| 932 | { |
| 933 | struct ifaddr *ifa = (struct ifaddr *) data; |
| 934 | struct ifreq *ifr = (struct ifreq *) data; |
| 935 | int error = 0; |
| 936 | |
| 937 | #define IF_INIT(ifp) \ |
| 938 | do { \ |
| 939 | if (((ifp)->if_flags & IFF_UP) == 0) { \ |
| 940 | (ifp)->if_flags |= IFF_UP; \ |
| 941 | (ifp)->if_init((ifp)->if_softc); \ |
| 942 | } \ |
| 943 | } while (0) |
| 944 | |
| 945 | ASSERT_SERIALIZED(ifp->if_serializer); |
| 946 | |
| 947 | switch (command) { |
| 948 | case SIOCSIFADDR: |
| 949 | switch (ifa->ifa_addr->sa_family) { |
| 950 | #ifdef INET |
| 951 | case AF_INET: |
| 952 | IF_INIT(ifp); /* before arpwhohas */ |
| 953 | arp_ifinit(ifp, ifa); |
| 954 | break; |
| 955 | #endif |
| 956 | #ifdef IPX |
| 957 | /* |
| 958 | * XXX - This code is probably wrong |
| 959 | */ |
| 960 | case AF_IPX: |
| 961 | { |
| 962 | struct ipx_addr *ina = &IA_SIPX(ifa)->sipx_addr; |
| 963 | struct arpcom *ac = IFP2AC(ifp); |
| 964 | |
| 965 | if (ipx_nullhost(*ina)) |
| 966 | ina->x_host = *(union ipx_host *) ac->ac_enaddr; |
| 967 | else |
| 968 | bcopy(ina->x_host.c_host, ac->ac_enaddr, |
| 969 | sizeof ac->ac_enaddr); |
| 970 | |
| 971 | IF_INIT(ifp); /* Set new address. */ |
| 972 | break; |
| 973 | } |
| 974 | #endif |
| 975 | #ifdef NS |
| 976 | /* |
| 977 | * XXX - This code is probably wrong |
| 978 | */ |
| 979 | case AF_NS: |
| 980 | { |
| 981 | struct ns_addr *ina = &(IA_SNS(ifa)->sns_addr); |
| 982 | struct arpcom *ac = IFP2AC(ifp); |
| 983 | |
| 984 | if (ns_nullhost(*ina)) |
| 985 | ina->x_host = *(union ns_host *)(ac->ac_enaddr); |
| 986 | else |
| 987 | bcopy(ina->x_host.c_host, ac->ac_enaddr, |
| 988 | sizeof ac->ac_enaddr); |
| 989 | |
| 990 | /* |
| 991 | * Set new address |
| 992 | */ |
| 993 | IF_INIT(ifp); |
| 994 | break; |
| 995 | } |
| 996 | #endif |
| 997 | default: |
| 998 | IF_INIT(ifp); |
| 999 | break; |
| 1000 | } |
| 1001 | break; |
| 1002 | |
| 1003 | case SIOCGIFADDR: |
| 1004 | bcopy(IFP2AC(ifp)->ac_enaddr, |
| 1005 | ((struct sockaddr *)ifr->ifr_data)->sa_data, |
| 1006 | ETHER_ADDR_LEN); |
| 1007 | break; |
| 1008 | |
| 1009 | case SIOCSIFMTU: |
| 1010 | /* |
| 1011 | * Set the interface MTU. |
| 1012 | */ |
| 1013 | if (ifr->ifr_mtu > ETHERMTU) { |
| 1014 | error = EINVAL; |
| 1015 | } else { |
| 1016 | ifp->if_mtu = ifr->ifr_mtu; |
| 1017 | } |
| 1018 | break; |
| 1019 | default: |
| 1020 | error = EINVAL; |
| 1021 | break; |
| 1022 | } |
| 1023 | return (error); |
| 1024 | |
| 1025 | #undef IF_INIT |
| 1026 | } |
| 1027 | |
| 1028 | int |
| 1029 | ether_resolvemulti( |
| 1030 | struct ifnet *ifp, |
| 1031 | struct sockaddr **llsa, |
| 1032 | struct sockaddr *sa) |
| 1033 | { |
| 1034 | struct sockaddr_dl *sdl; |
| 1035 | struct sockaddr_in *sin; |
| 1036 | #ifdef INET6 |
| 1037 | struct sockaddr_in6 *sin6; |
| 1038 | #endif |
| 1039 | u_char *e_addr; |
| 1040 | |
| 1041 | switch(sa->sa_family) { |
| 1042 | case AF_LINK: |
| 1043 | /* |
| 1044 | * No mapping needed. Just check that it's a valid MC address. |
| 1045 | */ |
| 1046 | sdl = (struct sockaddr_dl *)sa; |
| 1047 | e_addr = LLADDR(sdl); |
| 1048 | if ((e_addr[0] & 1) != 1) |
| 1049 | return EADDRNOTAVAIL; |
| 1050 | *llsa = 0; |
| 1051 | return 0; |
| 1052 | |
| 1053 | #ifdef INET |
| 1054 | case AF_INET: |
| 1055 | sin = (struct sockaddr_in *)sa; |
| 1056 | if (!IN_MULTICAST(ntohl(sin->sin_addr.s_addr))) |
| 1057 | return EADDRNOTAVAIL; |
| 1058 | MALLOC(sdl, struct sockaddr_dl *, sizeof *sdl, M_IFMADDR, |
| 1059 | M_WAITOK | M_ZERO); |
| 1060 | sdl->sdl_len = sizeof *sdl; |
| 1061 | sdl->sdl_family = AF_LINK; |
| 1062 | sdl->sdl_index = ifp->if_index; |
| 1063 | sdl->sdl_type = IFT_ETHER; |
| 1064 | sdl->sdl_alen = ETHER_ADDR_LEN; |
| 1065 | e_addr = LLADDR(sdl); |
| 1066 | ETHER_MAP_IP_MULTICAST(&sin->sin_addr, e_addr); |
| 1067 | *llsa = (struct sockaddr *)sdl; |
| 1068 | return 0; |
| 1069 | #endif |
| 1070 | #ifdef INET6 |
| 1071 | case AF_INET6: |
| 1072 | sin6 = (struct sockaddr_in6 *)sa; |
| 1073 | if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) { |
| 1074 | /* |
| 1075 | * An IP6 address of 0 means listen to all |
| 1076 | * of the Ethernet multicast address used for IP6. |
| 1077 | * (This is used for multicast routers.) |
| 1078 | */ |
| 1079 | ifp->if_flags |= IFF_ALLMULTI; |
| 1080 | *llsa = 0; |
| 1081 | return 0; |
| 1082 | } |
| 1083 | if (!IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) |
| 1084 | return EADDRNOTAVAIL; |
| 1085 | MALLOC(sdl, struct sockaddr_dl *, sizeof *sdl, M_IFMADDR, |
| 1086 | M_WAITOK | M_ZERO); |
| 1087 | sdl->sdl_len = sizeof *sdl; |
| 1088 | sdl->sdl_family = AF_LINK; |
| 1089 | sdl->sdl_index = ifp->if_index; |
| 1090 | sdl->sdl_type = IFT_ETHER; |
| 1091 | sdl->sdl_alen = ETHER_ADDR_LEN; |
| 1092 | e_addr = LLADDR(sdl); |
| 1093 | ETHER_MAP_IPV6_MULTICAST(&sin6->sin6_addr, e_addr); |
| 1094 | *llsa = (struct sockaddr *)sdl; |
| 1095 | return 0; |
| 1096 | #endif |
| 1097 | |
| 1098 | default: |
| 1099 | /* |
| 1100 | * Well, the text isn't quite right, but it's the name |
| 1101 | * that counts... |
| 1102 | */ |
| 1103 | return EAFNOSUPPORT; |
| 1104 | } |
| 1105 | } |
| 1106 | |
| 1107 | #if 0 |
| 1108 | /* |
| 1109 | * This is for reference. We have a table-driven version |
| 1110 | * of the little-endian crc32 generator, which is faster |
| 1111 | * than the double-loop. |
| 1112 | */ |
| 1113 | uint32_t |
| 1114 | ether_crc32_le(const uint8_t *buf, size_t len) |
| 1115 | { |
| 1116 | uint32_t c, crc, carry; |
| 1117 | size_t i, j; |
| 1118 | |
| 1119 | crc = 0xffffffffU; /* initial value */ |
| 1120 | |
| 1121 | for (i = 0; i < len; i++) { |
| 1122 | c = buf[i]; |
| 1123 | for (j = 0; j < 8; j++) { |
| 1124 | carry = ((crc & 0x01) ? 1 : 0) ^ (c & 0x01); |
| 1125 | crc >>= 1; |
| 1126 | c >>= 1; |
| 1127 | if (carry) |
| 1128 | crc = (crc ^ ETHER_CRC_POLY_LE); |
| 1129 | } |
| 1130 | } |
| 1131 | |
| 1132 | return (crc); |
| 1133 | } |
| 1134 | #else |
| 1135 | uint32_t |
| 1136 | ether_crc32_le(const uint8_t *buf, size_t len) |
| 1137 | { |
| 1138 | static const uint32_t crctab[] = { |
| 1139 | 0x00000000, 0x1db71064, 0x3b6e20c8, 0x26d930ac, |
| 1140 | 0x76dc4190, 0x6b6b51f4, 0x4db26158, 0x5005713c, |
| 1141 | 0xedb88320, 0xf00f9344, 0xd6d6a3e8, 0xcb61b38c, |
| 1142 | 0x9b64c2b0, 0x86d3d2d4, 0xa00ae278, 0xbdbdf21c |
| 1143 | }; |
| 1144 | uint32_t crc; |
| 1145 | size_t i; |
| 1146 | |
| 1147 | crc = 0xffffffffU; /* initial value */ |
| 1148 | |
| 1149 | for (i = 0; i < len; i++) { |
| 1150 | crc ^= buf[i]; |
| 1151 | crc = (crc >> 4) ^ crctab[crc & 0xf]; |
| 1152 | crc = (crc >> 4) ^ crctab[crc & 0xf]; |
| 1153 | } |
| 1154 | |
| 1155 | return (crc); |
| 1156 | } |
| 1157 | #endif |
| 1158 | |
| 1159 | uint32_t |
| 1160 | ether_crc32_be(const uint8_t *buf, size_t len) |
| 1161 | { |
| 1162 | uint32_t c, crc, carry; |
| 1163 | size_t i, j; |
| 1164 | |
| 1165 | crc = 0xffffffffU; /* initial value */ |
| 1166 | |
| 1167 | for (i = 0; i < len; i++) { |
| 1168 | c = buf[i]; |
| 1169 | for (j = 0; j < 8; j++) { |
| 1170 | carry = ((crc & 0x80000000U) ? 1 : 0) ^ (c & 0x01); |
| 1171 | crc <<= 1; |
| 1172 | c >>= 1; |
| 1173 | if (carry) |
| 1174 | crc = (crc ^ ETHER_CRC_POLY_BE) | carry; |
| 1175 | } |
| 1176 | } |
| 1177 | |
| 1178 | return (crc); |
| 1179 | } |
| 1180 | |
| 1181 | /* |
| 1182 | * find the size of ethernet header, and call classifier |
| 1183 | */ |
| 1184 | void |
| 1185 | altq_etherclassify(struct ifaltq *ifq, struct mbuf *m, |
| 1186 | struct altq_pktattr *pktattr) |
| 1187 | { |
| 1188 | struct ether_header *eh; |
| 1189 | uint16_t ether_type; |
| 1190 | int hlen, af, hdrsize; |
| 1191 | caddr_t hdr; |
| 1192 | |
| 1193 | hlen = sizeof(struct ether_header); |
| 1194 | eh = mtod(m, struct ether_header *); |
| 1195 | |
| 1196 | ether_type = ntohs(eh->ether_type); |
| 1197 | if (ether_type < ETHERMTU) { |
| 1198 | /* ick! LLC/SNAP */ |
| 1199 | struct llc *llc = (struct llc *)(eh + 1); |
| 1200 | hlen += 8; |
| 1201 | |
| 1202 | if (m->m_len < hlen || |
| 1203 | llc->llc_dsap != LLC_SNAP_LSAP || |
| 1204 | llc->llc_ssap != LLC_SNAP_LSAP || |
| 1205 | llc->llc_control != LLC_UI) |
| 1206 | goto bad; /* not snap! */ |
| 1207 | |
| 1208 | ether_type = ntohs(llc->llc_un.type_snap.ether_type); |
| 1209 | } |
| 1210 | |
| 1211 | if (ether_type == ETHERTYPE_IP) { |
| 1212 | af = AF_INET; |
| 1213 | hdrsize = 20; /* sizeof(struct ip) */ |
| 1214 | #ifdef INET6 |
| 1215 | } else if (ether_type == ETHERTYPE_IPV6) { |
| 1216 | af = AF_INET6; |
| 1217 | hdrsize = 40; /* sizeof(struct ip6_hdr) */ |
| 1218 | #endif |
| 1219 | } else |
| 1220 | goto bad; |
| 1221 | |
| 1222 | while (m->m_len <= hlen) { |
| 1223 | hlen -= m->m_len; |
| 1224 | m = m->m_next; |
| 1225 | } |
| 1226 | hdr = m->m_data + hlen; |
| 1227 | if (m->m_len < hlen + hdrsize) { |
| 1228 | /* |
| 1229 | * ip header is not in a single mbuf. this should not |
| 1230 | * happen in the current code. |
| 1231 | * (todo: use m_pulldown in the future) |
| 1232 | */ |
| 1233 | goto bad; |
| 1234 | } |
| 1235 | m->m_data += hlen; |
| 1236 | m->m_len -= hlen; |
| 1237 | ifq_classify(ifq, m, af, pktattr); |
| 1238 | m->m_data -= hlen; |
| 1239 | m->m_len += hlen; |
| 1240 | |
| 1241 | return; |
| 1242 | |
| 1243 | bad: |
| 1244 | pktattr->pattr_class = NULL; |
| 1245 | pktattr->pattr_hdr = NULL; |
| 1246 | pktattr->pattr_af = AF_UNSPEC; |
| 1247 | } |
| 1248 | |
| 1249 | static void |
| 1250 | ether_restore_header(struct mbuf **m0, const struct ether_header *eh, |
| 1251 | const struct ether_header *save_eh) |
| 1252 | { |
| 1253 | struct mbuf *m = *m0; |
| 1254 | |
| 1255 | ether_restore_hdr++; |
| 1256 | |
| 1257 | /* |
| 1258 | * Prepend the header, optimize for the common case of |
| 1259 | * eh pointing into the mbuf. |
| 1260 | */ |
| 1261 | if ((const void *)(eh + 1) == (void *)m->m_data) { |
| 1262 | m->m_data -= ETHER_HDR_LEN; |
| 1263 | m->m_len += ETHER_HDR_LEN; |
| 1264 | m->m_pkthdr.len += ETHER_HDR_LEN; |
| 1265 | } else { |
| 1266 | ether_prepend_hdr++; |
| 1267 | |
| 1268 | M_PREPEND(m, ETHER_HDR_LEN, MB_DONTWAIT); |
| 1269 | if (m != NULL) { |
| 1270 | bcopy(save_eh, mtod(m, struct ether_header *), |
| 1271 | ETHER_HDR_LEN); |
| 1272 | } |
| 1273 | } |
| 1274 | *m0 = m; |
| 1275 | } |
| 1276 | |
| 1277 | #ifdef ETHER_INPUT_CHAIN |
| 1278 | |
| 1279 | static void |
| 1280 | ether_input_ipifunc(void *arg) |
| 1281 | { |
| 1282 | struct mbuf *m, *next; |
| 1283 | lwkt_port_t port; |
| 1284 | |
| 1285 | m = arg; |
| 1286 | do { |
| 1287 | next = m->m_nextpkt; |
| 1288 | m->m_nextpkt = NULL; |
| 1289 | |
| 1290 | port = m->m_pkthdr.header; |
| 1291 | m->m_pkthdr.header = NULL; |
| 1292 | |
| 1293 | lwkt_sendmsg(port, |
| 1294 | &m->m_hdr.mh_netmsg.nm_netmsg.nm_lmsg); |
| 1295 | |
| 1296 | m = next; |
| 1297 | } while (m != NULL); |
| 1298 | } |
| 1299 | |
| 1300 | void |
| 1301 | ether_input_dispatch(struct mbuf_chain *chain) |
| 1302 | { |
| 1303 | #ifdef SMP |
| 1304 | int i; |
| 1305 | |
| 1306 | for (i = 0; i < ncpus; ++i) { |
| 1307 | if (chain[i].mc_head != NULL) { |
| 1308 | lwkt_send_ipiq(globaldata_find(i), |
| 1309 | ether_input_ipifunc, chain[i].mc_head); |
| 1310 | } |
| 1311 | } |
| 1312 | #else |
| 1313 | ether_input_ipifunc(chain->mc_head); |
| 1314 | #endif |
| 1315 | } |
| 1316 | |
| 1317 | #endif /* ETHER_INPUT_CHAIN */ |