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