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