1 /*- 2 * SPDX-License-Identifier: BSD-3-Clause 3 * 4 * Copyright (c) 1982, 1986, 1993 5 * The Regents of the University of California. All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 3. Neither the name of the University nor the names of its contributors 16 * may be used to endorse or promote products derived from this software 17 * without specific prior written permission. 18 * 19 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 22 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 29 * SUCH DAMAGE. 30 * 31 * @(#)if_loop.c 8.2 (Berkeley) 1/9/95 32 * $FreeBSD$ 33 */ 34 35 /* 36 * Loopback interface driver for protocol testing and timing. 37 */ 38 39 #include "opt_inet.h" 40 #include "opt_inet6.h" 41 #include "opt_rss.h" 42 43 #include <sys/param.h> 44 #include <sys/systm.h> 45 #include <sys/kernel.h> 46 #include <sys/mbuf.h> 47 #include <sys/module.h> 48 #include <machine/bus.h> 49 #include <sys/rman.h> 50 #include <sys/socket.h> 51 #include <sys/sockio.h> 52 #include <sys/sysctl.h> 53 54 #include <net/if.h> 55 #include <net/if_var.h> 56 #include <net/if_private.h> 57 #include <net/if_clone.h> 58 #include <net/if_types.h> 59 #include <net/netisr.h> 60 #include <net/route.h> 61 #include <net/bpf.h> 62 #include <net/vnet.h> 63 64 #ifdef INET 65 #include <netinet/in.h> 66 #include <netinet/in_var.h> 67 #endif 68 69 #ifdef INET6 70 #ifndef INET 71 #include <netinet/in.h> 72 #endif 73 #include <netinet6/in6_var.h> 74 #include <netinet/ip6.h> 75 #endif 76 77 #include <security/mac/mac_framework.h> 78 79 #ifdef TINY_LOMTU 80 #define LOMTU (1024+512) 81 #elif defined(LARGE_LOMTU) 82 #define LOMTU 131072 83 #else 84 #define LOMTU 16384 85 #endif 86 87 #define LO_CSUM_FEATURES (CSUM_IP | CSUM_TCP | CSUM_UDP | CSUM_SCTP) 88 #define LO_CSUM_FEATURES6 (CSUM_TCP_IPV6 | CSUM_UDP_IPV6 | CSUM_SCTP_IPV6) 89 #define LO_CSUM_SET (CSUM_DATA_VALID | CSUM_DATA_VALID_IPV6 | \ 90 CSUM_PSEUDO_HDR | \ 91 CSUM_IP_CHECKED | CSUM_IP_VALID | \ 92 CSUM_SCTP_VALID) 93 94 static int loioctl(struct ifnet *, u_long, caddr_t); 95 static int looutput(struct ifnet *ifp, struct mbuf *m, 96 const struct sockaddr *dst, struct route *ro); 97 98 VNET_DEFINE(struct ifnet *, loif); /* Used externally */ 99 100 #ifdef VIMAGE 101 VNET_DEFINE_STATIC(struct if_clone *, lo_cloner); 102 #define V_lo_cloner VNET(lo_cloner) 103 #endif 104 105 static struct if_clone *lo_cloner; 106 static const char loname[] = "lo"; 107 108 static int 109 lo_clone_destroy(struct if_clone *ifc, struct ifnet *ifp, uint32_t flags) 110 { 111 if (ifp->if_dunit == 0 && (flags & IFC_F_FORCE) == 0) 112 return (EINVAL); 113 114 #ifndef VIMAGE 115 /* XXX: destroying lo0 will lead to panics. */ 116 KASSERT(V_loif != ifp, ("%s: destroying lo0", __func__)); 117 #endif 118 119 bpfdetach(ifp); 120 if_detach(ifp); 121 if_free(ifp); 122 123 return (0); 124 } 125 126 static int 127 lo_clone_create(struct if_clone *ifc, char *name, size_t len, 128 struct ifc_data *ifd, struct ifnet **ifpp) 129 { 130 struct ifnet *ifp; 131 132 ifp = if_alloc(IFT_LOOP); 133 if (ifp == NULL) 134 return (ENOSPC); 135 136 if_initname(ifp, loname, ifd->unit); 137 ifp->if_mtu = LOMTU; 138 ifp->if_flags = IFF_LOOPBACK | IFF_MULTICAST; 139 ifp->if_ioctl = loioctl; 140 ifp->if_output = looutput; 141 ifp->if_snd.ifq_maxlen = ifqmaxlen; 142 ifp->if_capabilities = ifp->if_capenable = 143 IFCAP_HWCSUM | IFCAP_HWCSUM_IPV6 | IFCAP_LINKSTATE; 144 ifp->if_hwassist = LO_CSUM_FEATURES | LO_CSUM_FEATURES6; 145 if_attach(ifp); 146 bpfattach(ifp, DLT_NULL, sizeof(u_int32_t)); 147 if (V_loif == NULL) 148 V_loif = ifp; 149 *ifpp = ifp; 150 151 return (0); 152 } 153 154 static void 155 vnet_loif_init(const void *unused __unused) 156 { 157 struct if_clone_addreq req = { 158 .create_f = lo_clone_create, 159 .destroy_f = lo_clone_destroy, 160 .flags = IFC_F_AUTOUNIT, 161 }; 162 lo_cloner = ifc_attach_cloner(loname, &req); 163 #ifdef VIMAGE 164 V_lo_cloner = lo_cloner; 165 #endif 166 struct ifc_data ifd = { .unit = 0 }; 167 ifc_create_ifp(loname, &ifd, NULL); 168 } 169 VNET_SYSINIT(vnet_loif_init, SI_SUB_PSEUDO, SI_ORDER_ANY, 170 vnet_loif_init, NULL); 171 172 #ifdef VIMAGE 173 static void 174 vnet_loif_uninit(const void *unused __unused) 175 { 176 177 ifc_detach_cloner(V_lo_cloner); 178 V_loif = NULL; 179 } 180 VNET_SYSUNINIT(vnet_loif_uninit, SI_SUB_INIT_IF, SI_ORDER_SECOND, 181 vnet_loif_uninit, NULL); 182 #endif 183 184 static int 185 loop_modevent(module_t mod, int type, void *data) 186 { 187 188 switch (type) { 189 case MOD_LOAD: 190 break; 191 192 case MOD_UNLOAD: 193 printf("loop module unload - not possible for this module type\n"); 194 return (EINVAL); 195 196 default: 197 return (EOPNOTSUPP); 198 } 199 return (0); 200 } 201 202 static moduledata_t loop_mod = { 203 "if_lo", 204 loop_modevent, 205 0 206 }; 207 208 DECLARE_MODULE(if_lo, loop_mod, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY); 209 210 static int 211 looutput(struct ifnet *ifp, struct mbuf *m, const struct sockaddr *dst, 212 struct route *ro) 213 { 214 u_int32_t af; 215 #ifdef MAC 216 int error; 217 #endif 218 219 M_ASSERTPKTHDR(m); /* check if we have the packet header */ 220 221 #ifdef MAC 222 error = mac_ifnet_check_transmit(ifp, m); 223 if (error) { 224 m_freem(m); 225 return (error); 226 } 227 #endif 228 229 if (ro != NULL && ro->ro_flags & (RT_REJECT|RT_BLACKHOLE)) { 230 m_freem(m); 231 return (ro->ro_flags & RT_BLACKHOLE ? 0 : EHOSTUNREACH); 232 } 233 234 if_inc_counter(ifp, IFCOUNTER_OPACKETS, 1); 235 if_inc_counter(ifp, IFCOUNTER_OBYTES, m->m_pkthdr.len); 236 237 #ifdef RSS 238 M_HASHTYPE_CLEAR(m); 239 #endif 240 241 /* BPF writes need to be handled specially. */ 242 if (dst->sa_family == AF_UNSPEC || dst->sa_family == pseudo_AF_HDRCMPLT) 243 bcopy(dst->sa_data, &af, sizeof(af)); 244 else 245 af = RO_GET_FAMILY(ro, dst); 246 247 #if 1 /* XXX */ 248 switch (af) { 249 case AF_INET: 250 if (ifp->if_capenable & IFCAP_RXCSUM) { 251 m->m_pkthdr.csum_data = 0xffff; 252 m->m_pkthdr.csum_flags = LO_CSUM_SET; 253 } 254 m->m_pkthdr.csum_flags &= ~LO_CSUM_FEATURES; 255 break; 256 case AF_INET6: 257 #if 0 258 /* 259 * XXX-BZ for now always claim the checksum is good despite 260 * any interface flags. This is a workaround for 9.1-R and 261 * a proper solution ought to be sought later. 262 */ 263 if (ifp->if_capenable & IFCAP_RXCSUM_IPV6) { 264 m->m_pkthdr.csum_data = 0xffff; 265 m->m_pkthdr.csum_flags = LO_CSUM_SET; 266 } 267 #else 268 m->m_pkthdr.csum_data = 0xffff; 269 m->m_pkthdr.csum_flags = LO_CSUM_SET; 270 #endif 271 m->m_pkthdr.csum_flags &= ~LO_CSUM_FEATURES6; 272 break; 273 default: 274 printf("looutput: af=%d unexpected\n", af); 275 m_freem(m); 276 return (EAFNOSUPPORT); 277 } 278 #endif 279 return (if_simloop(ifp, m, af, 0)); 280 } 281 282 /* 283 * if_simloop() 284 * 285 * This function is to support software emulation of hardware loopback, 286 * i.e., for interfaces with the IFF_SIMPLEX attribute. Since they can't 287 * hear their own broadcasts, we create a copy of the packet that we 288 * would normally receive via a hardware loopback. 289 * 290 * This function expects the packet to include the media header of length hlen. 291 */ 292 int 293 if_simloop(struct ifnet *ifp, struct mbuf *m, int af, int hlen) 294 { 295 int isr; 296 297 M_ASSERTPKTHDR(m); 298 m_tag_delete_nonpersistent(m); 299 m->m_pkthdr.rcvif = ifp; 300 301 #ifdef MAC 302 mac_ifnet_create_mbuf(ifp, m); 303 #endif 304 305 /* 306 * Let BPF see incoming packet in the following manner: 307 * - Emulated packet loopback for a simplex interface 308 * (net/if_ethersubr.c) 309 * -> passes it to ifp's BPF 310 * - IPv4/v6 multicast packet loopback (netinet(6)/ip(6)_output.c) 311 * -> not passes it to any BPF 312 * - Normal packet loopback from myself to myself (net/if_loop.c) 313 * -> passes to lo0's BPF (even in case of IPv6, where ifp!=lo0) 314 */ 315 if (hlen > 0) { 316 if (bpf_peers_present(ifp->if_bpf)) { 317 bpf_mtap(ifp->if_bpf, m); 318 } 319 } else { 320 if (bpf_peers_present(V_loif->if_bpf)) { 321 if ((m->m_flags & M_MCAST) == 0 || V_loif == ifp) { 322 /* XXX beware sizeof(af) != 4 */ 323 u_int32_t af1 = af; 324 325 /* 326 * We need to prepend the address family. 327 */ 328 bpf_mtap2(V_loif->if_bpf, &af1, sizeof(af1), m); 329 } 330 } 331 } 332 333 /* Strip away media header */ 334 if (hlen > 0) { 335 m_adj(m, hlen); 336 #ifndef __NO_STRICT_ALIGNMENT 337 /* 338 * Some archs do not like unaligned data, so 339 * we move data down in the first mbuf. 340 */ 341 if (mtod(m, vm_offset_t) & 3) { 342 KASSERT(hlen >= 3, ("if_simloop: hlen too small")); 343 bcopy(m->m_data, 344 (char *)(mtod(m, vm_offset_t) 345 - (mtod(m, vm_offset_t) & 3)), 346 m->m_len); 347 m->m_data -= (mtod(m,vm_offset_t) & 3); 348 } 349 #endif 350 } 351 352 /* Deliver to upper layer protocol */ 353 switch (af) { 354 #ifdef INET 355 case AF_INET: 356 isr = NETISR_IP; 357 break; 358 #endif 359 #ifdef INET6 360 case AF_INET6: 361 m->m_flags |= M_LOOP; 362 isr = NETISR_IPV6; 363 break; 364 #endif 365 default: 366 printf("if_simloop: can't handle af=%d\n", af); 367 m_freem(m); 368 return (EAFNOSUPPORT); 369 } 370 if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1); 371 if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len); 372 netisr_queue(isr, m); /* mbuf is free'd on failure. */ 373 return (0); 374 } 375 376 /* 377 * Process an ioctl request. 378 */ 379 static int 380 loioctl(struct ifnet *ifp, u_long cmd, caddr_t data) 381 { 382 struct ifreq *ifr = (struct ifreq *)data; 383 int error = 0, mask; 384 385 switch (cmd) { 386 case SIOCSIFADDR: 387 ifp->if_flags |= IFF_UP; 388 ifp->if_drv_flags |= IFF_DRV_RUNNING; 389 if_link_state_change(ifp, LINK_STATE_UP); 390 /* 391 * Everything else is done at a higher level. 392 */ 393 break; 394 395 case SIOCADDMULTI: 396 case SIOCDELMULTI: 397 if (ifr == NULL) { 398 error = EAFNOSUPPORT; /* XXX */ 399 break; 400 } 401 switch (ifr->ifr_addr.sa_family) { 402 #ifdef INET 403 case AF_INET: 404 break; 405 #endif 406 #ifdef INET6 407 case AF_INET6: 408 break; 409 #endif 410 411 default: 412 error = EAFNOSUPPORT; 413 break; 414 } 415 break; 416 417 case SIOCSIFMTU: 418 ifp->if_mtu = ifr->ifr_mtu; 419 break; 420 421 case SIOCSIFFLAGS: 422 if_link_state_change(ifp, (ifp->if_flags & IFF_UP) ? 423 LINK_STATE_UP: LINK_STATE_DOWN); 424 break; 425 426 case SIOCSIFCAP: 427 mask = ifp->if_capenable ^ ifr->ifr_reqcap; 428 if ((mask & IFCAP_RXCSUM) != 0) 429 ifp->if_capenable ^= IFCAP_RXCSUM; 430 if ((mask & IFCAP_TXCSUM) != 0) 431 ifp->if_capenable ^= IFCAP_TXCSUM; 432 if ((mask & IFCAP_RXCSUM_IPV6) != 0) { 433 #if 0 434 ifp->if_capenable ^= IFCAP_RXCSUM_IPV6; 435 #else 436 error = EOPNOTSUPP; 437 break; 438 #endif 439 } 440 if ((mask & IFCAP_TXCSUM_IPV6) != 0) { 441 #if 0 442 ifp->if_capenable ^= IFCAP_TXCSUM_IPV6; 443 #else 444 error = EOPNOTSUPP; 445 break; 446 #endif 447 } 448 ifp->if_hwassist = 0; 449 if (ifp->if_capenable & IFCAP_TXCSUM) 450 ifp->if_hwassist = LO_CSUM_FEATURES; 451 #if 0 452 if (ifp->if_capenable & IFCAP_TXCSUM_IPV6) 453 ifp->if_hwassist |= LO_CSUM_FEATURES6; 454 #endif 455 break; 456 457 default: 458 error = EINVAL; 459 } 460 return (error); 461 } 462