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