slirp/slirp.c

   1 /*
   2  * libslirp glue
   3  *
   4  * Copyright (c) 2004-2008 Fabrice Bellard
   5  *
   6  * Permission is hereby granted, free of charge, to any person obtaining a copy
   7  * of this software and associated documentation files (the "Software"), to deal
   8  * in the Software without restriction, including without limitation the rights
   9  * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
  10  * copies of the Software, and to permit persons to whom the Software is
  11  * furnished to do so, subject to the following conditions:
  12  *
  13  * The above copyright notice and this permission notice shall be included in
  14  * all copies or substantial portions of the Software.
  15  *
  16  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
  17  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
  18  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
  19  * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
  20  * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
  21  * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
  22  * THE SOFTWARE.
  23  */
  24 #include "qemu-common.h"
  25 #include "qemu/timer.h"
  26 #include "char/char.h"
  27 #include "slirp.h"
  28 #include "hw/hw.h"
  29
  30 /* host loopback address */
  31 struct in_addr loopback_addr;
  32 /* host loopback network mask */
  33 unsigned long loopback_mask;
  34
  35 /* emulated hosts use the MAC addr 52:55:IP:IP:IP:IP */
  36 static const uint8_t special_ethaddr[ETH_ALEN] = {
  37     0x52, 0x55, 0x00, 0x00, 0x00, 0x00
  38 };
  39
  40 static const uint8_t zero_ethaddr[ETH_ALEN] = { 0, 0, 0, 0, 0, 0 };
  41
  42 /* XXX: suppress those select globals */
  43 fd_set *global_readfds, *global_writefds, *global_xfds;
  44
  45 u_int curtime;
  46 static u_int time_fasttimo, last_slowtimo;
  47 static int do_slowtimo;
  48
  49 static QTAILQ_HEAD(slirp_instances, Slirp) slirp_instances =
  50     QTAILQ_HEAD_INITIALIZER(slirp_instances);
  51
  52 static struct in_addr dns_addr;
  53 static u_int dns_addr_time;
  54
  55 #ifdef _WIN32
  56
  57 int get_dns_addr(struct in_addr *pdns_addr)
  58 {
  59     FIXED_INFO *FixedInfo=NULL;
  60     ULONG    BufLen;
  61     DWORD    ret;
  62     IP_ADDR_STRING *pIPAddr;
  63     struct in_addr tmp_addr;
  64
  65     if (dns_addr.s_addr != 0 && (curtime - dns_addr_time) < 1000) {
  66         *pdns_addr = dns_addr;
  67         return 0;
  68     }
  69
  70     FixedInfo = (FIXED_INFO *)GlobalAlloc(GPTR, sizeof(FIXED_INFO));
  71     BufLen = sizeof(FIXED_INFO);
  72
  73     if (ERROR_BUFFER_OVERFLOW == GetNetworkParams(FixedInfo, &BufLen)) {
  74         if (FixedInfo) {
  75             GlobalFree(FixedInfo);
  76             FixedInfo = NULL;
  77         }
  78         FixedInfo = GlobalAlloc(GPTR, BufLen);
  79     }
  80
  81     if ((ret = GetNetworkParams(FixedInfo, &BufLen)) != ERROR_SUCCESS) {
  82         printf("GetNetworkParams failed. ret = %08x\n", (u_int)ret );
  83         if (FixedInfo) {
  84             GlobalFree(FixedInfo);
  85             FixedInfo = NULL;
  86         }
  87         return -1;
  88     }
  89
  90     pIPAddr = &(FixedInfo->DnsServerList);
  91     inet_aton(pIPAddr->IpAddress.String, &tmp_addr);
  92     *pdns_addr = tmp_addr;
  93     dns_addr = tmp_addr;
  94     dns_addr_time = curtime;
  95     if (FixedInfo) {
  96         GlobalFree(FixedInfo);
  97         FixedInfo = NULL;
  98     }
  99     return 0;
 100 }
 101
 102 static void winsock_cleanup(void)
 103 {
 104     WSACleanup();
 105 }
 106
 107 #else
 108
 109 static struct stat dns_addr_stat;
 110
 111 int get_dns_addr(struct in_addr *pdns_addr)
 112 {
 113     char buff[512];
 114     char buff2[257];
 115     FILE *f;
 116     int found = 0;
 117     struct in_addr tmp_addr;
 118
 119     if (dns_addr.s_addr != 0) {
 120         struct stat old_stat;
 121         if ((curtime - dns_addr_time) < 1000) {
 122             *pdns_addr = dns_addr;
 123             return 0;
 124         }
 125         old_stat = dns_addr_stat;
 126         if (stat("/etc/resolv.conf", &dns_addr_stat) != 0)
 127             return -1;
 128         if ((dns_addr_stat.st_dev == old_stat.st_dev)
 129             && (dns_addr_stat.st_ino == old_stat.st_ino)
 130             && (dns_addr_stat.st_size == old_stat.st_size)
 131             && (dns_addr_stat.st_mtime == old_stat.st_mtime)) {
 132             *pdns_addr = dns_addr;
 133             return 0;
 134         }
 135     }
 136
 137     f = fopen("/etc/resolv.conf", "r");
 138     if (!f)
 139         return -1;
 140
 141 #ifdef DEBUG
 142     lprint("IP address of your DNS(s): ");
 143 #endif
 144     while (fgets(buff, 512, f) != NULL) {
 145         if (sscanf(buff, "nameserver%*[ \t]%256s", buff2) == 1) {
 146             if (!inet_aton(buff2, &tmp_addr))
 147                 continue;
 148             /* If it's the first one, set it to dns_addr */
 149             if (!found) {
 150                 *pdns_addr = tmp_addr;
 151                 dns_addr = tmp_addr;
 152                 dns_addr_time = curtime;
 153             }
 154 #ifdef DEBUG
 155             else
 156                 lprint(", ");
 157 #endif
 158             if (++found > 3) {
 159 #ifdef DEBUG
 160                 lprint("(more)");
 161 #endif
 162                 break;
 163             }
 164 #ifdef DEBUG
 165             else
 166                 lprint("%s", inet_ntoa(tmp_addr));
 167 #endif
 168         }
 169     }
 170     fclose(f);
 171     if (!found)
 172         return -1;
 173     return 0;
 174 }
 175
 176 #endif
 177
 178 static void slirp_init_once(void)
 179 {
 180     static int initialized;
 181 #ifdef _WIN32
 182     WSADATA Data;
 183 #endif
 184
 185     if (initialized) {
 186         return;
 187     }
 188     initialized = 1;
 189
 190 #ifdef _WIN32
 191     WSAStartup(MAKEWORD(2,0), &Data);
 192     atexit(winsock_cleanup);
 193 #endif
 194
 195     loopback_addr.s_addr = htonl(INADDR_LOOPBACK);
 196     loopback_mask = htonl(IN_CLASSA_NET);
 197 }
 198
 199 static void slirp_state_save(QEMUFile *f, void *opaque);
 200 static int slirp_state_load(QEMUFile *f, void *opaque, int version_id);
 201
 202 Slirp *slirp_init(int restricted, struct in_addr vnetwork,
 203                   struct in_addr vnetmask, struct in_addr vhost,
 204                   const char *vhostname, const char *tftp_path,
 205                   const char *bootfile, struct in_addr vdhcp_start,
 206                   struct in_addr vnameserver, const char **vdnssearch,
 207                   void *opaque)
 208 {
 209     Slirp *slirp = g_malloc0(sizeof(Slirp));
 210
 211     slirp_init_once();
 212
 213     slirp->restricted = restricted;
 214
 215     if_init(slirp);
 216     ip_init(slirp);
 217
 218     /* Initialise mbufs *after* setting the MTU */
 219     m_init(slirp);
 220
 221     slirp->vnetwork_addr = vnetwork;
 222     slirp->vnetwork_mask = vnetmask;
 223     slirp->vhost_addr = vhost;
 224     if (vhostname) {
 225         pstrcpy(slirp->client_hostname, sizeof(slirp->client_hostname),
 226                 vhostname);
 227     }
 228     if (tftp_path) {
 229         slirp->tftp_prefix = g_strdup(tftp_path);
 230     }
 231     if (bootfile) {
 232         slirp->bootp_filename = g_strdup(bootfile);
 233     }
 234     slirp->vdhcp_startaddr = vdhcp_start;
 235     slirp->vnameserver_addr = vnameserver;
 236
 237     if (vdnssearch) {
 238         translate_dnssearch(slirp, vdnssearch);
 239     }
 240
 241     slirp->opaque = opaque;
 242
 243     register_savevm(NULL, "slirp", 0, 3,
 244                     slirp_state_save, slirp_state_load, slirp);
 245
 246     QTAILQ_INSERT_TAIL(&slirp_instances, slirp, entry);
 247
 248     return slirp;
 249 }
 250
 251 void slirp_cleanup(Slirp *slirp)
 252 {
 253     QTAILQ_REMOVE(&slirp_instances, slirp, entry);
 254
 255     unregister_savevm(NULL, "slirp", slirp);
 256
 257     ip_cleanup(slirp);
 258     m_cleanup(slirp);
 259
 260     g_free(slirp->vdnssearch);
 261     g_free(slirp->tftp_prefix);
 262     g_free(slirp->bootp_filename);
 263     g_free(slirp);
 264 }
 265
 266 #define CONN_CANFSEND(so) (((so)->so_state & (SS_FCANTSENDMORE|SS_ISFCONNECTED)) == SS_ISFCONNECTED)
 267 #define CONN_CANFRCV(so) (((so)->so_state & (SS_FCANTRCVMORE|SS_ISFCONNECTED)) == SS_ISFCONNECTED)
 268 #define UPD_NFDS(x) if (nfds < (x)) nfds = (x)
 269
 270 void slirp_update_timeout(uint32_t *timeout)
 271 {
 272     if (!QTAILQ_EMPTY(&slirp_instances)) {
 273         *timeout = MIN(1000, *timeout);
 274     }
 275 }
 276
 277 void slirp_select_fill(int *pnfds,
 278                        fd_set *readfds, fd_set *writefds, fd_set *xfds)
 279 {
 280     Slirp *slirp;
 281     struct socket *so, *so_next;
 282     int nfds;
 283
 284     if (QTAILQ_EMPTY(&slirp_instances)) {
 285         return;
 286     }
 287
 288     /* fail safe */
 289     global_readfds = NULL;
 290     global_writefds = NULL;
 291     global_xfds = NULL;
 292
 293     nfds = *pnfds;
 294         /*
 295          * First, TCP sockets
 296          */
 297         do_slowtimo = 0;
 298
 299         QTAILQ_FOREACH(slirp, &slirp_instances, entry) {
 300                 /*
 301                  * *_slowtimo needs calling if there are IP fragments
 302                  * in the fragment queue, or there are TCP connections active
 303                  */
 304                 do_slowtimo |= ((slirp->tcb.so_next != &slirp->tcb) ||
 305                     (&slirp->ipq.ip_link != slirp->ipq.ip_link.next));
 306
 307                 for (so = slirp->tcb.so_next; so != &slirp->tcb;
 308                      so = so_next) {
 309                         so_next = so->so_next;
 310
 311                         /*
 312                          * See if we need a tcp_fasttimo
 313                          */
 314                         if (time_fasttimo == 0 && so->so_tcpcb->t_flags & TF_DELACK)
 315                            time_fasttimo = curtime; /* Flag when we want a fasttimo */
 316
 317                         /*
 318                          * NOFDREF can include still connecting to local-host,
 319                          * newly socreated() sockets etc. Don't want to select these.
 320                          */
 321                         if (so->so_state & SS_NOFDREF || so->s == -1)
 322                            continue;
 323
 324                         /*
 325                          * Set for reading sockets which are accepting
 326                          */
 327                         if (so->so_state & SS_FACCEPTCONN) {
 328                                 FD_SET(so->s, readfds);
 329                                 UPD_NFDS(so->s);
 330                                 continue;
 331                         }
 332
 333                         /*
 334                          * Set for writing sockets which are connecting
 335                          */
 336                         if (so->so_state & SS_ISFCONNECTING) {
 337                                 FD_SET(so->s, writefds);
 338                                 UPD_NFDS(so->s);
 339                                 continue;
 340                         }
 341
 342                         /*
 343                          * Set for writing if we are connected, can send more, and
 344                          * we have something to send
 345                          */
 346                         if (CONN_CANFSEND(so) && so->so_rcv.sb_cc) {
 347                                 FD_SET(so->s, writefds);
 348                                 UPD_NFDS(so->s);
 349                         }
 350
 351                         /*
 352                          * Set for reading (and urgent data) if we are connected, can
 353                          * receive more, and we have room for it XXX /2 ?
 354                          */
 355                         if (CONN_CANFRCV(so) && (so->so_snd.sb_cc < (so->so_snd.sb_datalen/2))) {
 356                                 FD_SET(so->s, readfds);
 357                                 FD_SET(so->s, xfds);
 358                                 UPD_NFDS(so->s);
 359                         }
 360                 }
 361
 362                 /*
 363                  * UDP sockets
 364                  */
 365                 for (so = slirp->udb.so_next; so != &slirp->udb;
 366                      so = so_next) {
 367                         so_next = so->so_next;
 368
 369                         /*
 370                          * See if it's timed out
 371                          */
 372                         if (so->so_expire) {
 373                                 if (so->so_expire <= curtime) {
 374                                         udp_detach(so);
 375                                         continue;
 376                                 } else
 377                                         do_slowtimo = 1; /* Let socket expire */
 378                         }
 379
 380                         /*
 381                          * When UDP packets are received from over the
 382                          * link, they're sendto()'d straight away, so
 383                          * no need for setting for writing
 384                          * Limit the number of packets queued by this session
 385                          * to 4.  Note that even though we try and limit this
 386                          * to 4 packets, the session could have more queued
 387                          * if the packets needed to be fragmented
 388                          * (XXX <= 4 ?)
 389                          */
 390                         if ((so->so_state & SS_ISFCONNECTED) && so->so_queued <= 4) {
 391                                 FD_SET(so->s, readfds);
 392                                 UPD_NFDS(so->s);
 393                         }
 394                 }
 395
 396                 /*
 397                  * ICMP sockets
 398                  */
 399                 for (so = slirp->icmp.so_next; so != &slirp->icmp;
 400                      so = so_next) {
 401                     so_next = so->so_next;
 402
 403                     /*
 404                      * See if it's timed out
 405                      */
 406                     if (so->so_expire) {
 407                         if (so->so_expire <= curtime) {
 408                             icmp_detach(so);
 409                             continue;
 410                         } else {
 411                             do_slowtimo = 1; /* Let socket expire */
 412                         }
 413                     }
 414
 415                     if (so->so_state & SS_ISFCONNECTED) {
 416                         FD_SET(so->s, readfds);
 417                         UPD_NFDS(so->s);
 418                     }
 419                 }
 420         }
 421
 422         *pnfds = nfds;
 423 }
 424
 425 void slirp_select_poll(fd_set *readfds, fd_set *writefds, fd_set *xfds,
 426                        int select_error)
 427 {
 428     Slirp *slirp;
 429     struct socket *so, *so_next;
 430     int ret;
 431
 432     if (QTAILQ_EMPTY(&slirp_instances)) {
 433         return;
 434     }
 435
 436     global_readfds = readfds;
 437     global_writefds = writefds;
 438     global_xfds = xfds;
 439
 440     curtime = qemu_get_clock_ms(rt_clock);
 441
 442     QTAILQ_FOREACH(slirp, &slirp_instances, entry) {
 443         /*
 444          * See if anything has timed out
 445          */
 446                 if (time_fasttimo && ((curtime - time_fasttimo) >= 2)) {
 447                         tcp_fasttimo(slirp);
 448                         time_fasttimo = 0;
 449                 }
 450                 if (do_slowtimo && ((curtime - last_slowtimo) >= 499)) {
 451                         ip_slowtimo(slirp);
 452                         tcp_slowtimo(slirp);
 453                         last_slowtimo = curtime;
 454                 }
 455
 456         /*
 457          * Check sockets
 458          */
 459         if (!select_error) {
 460                 /*
 461                  * Check TCP sockets
 462                  */
 463                 for (so = slirp->tcb.so_next; so != &slirp->tcb;
 464                      so = so_next) {
 465                         so_next = so->so_next;
 466
 467                         /*
 468                          * FD_ISSET is meaningless on these sockets
 469                          * (and they can crash the program)
 470                          */
 471                         if (so->so_state & SS_NOFDREF || so->s == -1)
 472                            continue;
 473
 474                         /*
 475                          * Check for URG data
 476                          * This will soread as well, so no need to
 477                          * test for readfds below if this succeeds
 478                          */
 479                         if (FD_ISSET(so->s, xfds))
 480                            sorecvoob(so);
 481                         /*
 482                          * Check sockets for reading
 483                          */
 484                         else if (FD_ISSET(so->s, readfds)) {
 485                                 /*
 486                                  * Check for incoming connections
 487                                  */
 488                                 if (so->so_state & SS_FACCEPTCONN) {
 489                                         tcp_connect(so);
 490                                         continue;
 491                                 } /* else */
 492                                 ret = soread(so);
 493
 494                                 /* Output it if we read something */
 495                                 if (ret > 0)
 496                                    tcp_output(sototcpcb(so));
 497                         }
 498
 499                         /*
 500                          * Check sockets for writing
 501                          */
 502                         if (FD_ISSET(so->s, writefds)) {
 503                           /*
 504                            * Check for non-blocking, still-connecting sockets
 505                            */
 506                           if (so->so_state & SS_ISFCONNECTING) {
 507                             /* Connected */
 508                             so->so_state &= ~SS_ISFCONNECTING;
 509
 510                             ret = send(so->s, (const void *) &ret, 0, 0);
 511                             if (ret < 0) {
 512                               /* XXXXX Must fix, zero bytes is a NOP */
 513                               if (errno == EAGAIN || errno == EWOULDBLOCK ||
 514                                   errno == EINPROGRESS || errno == ENOTCONN)
 515                                 continue;
 516
 517                               /* else failed */
 518                               so->so_state &= SS_PERSISTENT_MASK;
 519                               so->so_state |= SS_NOFDREF;
 520                             }
 521                             /* else so->so_state &= ~SS_ISFCONNECTING; */
 522
 523                             /*
 524                              * Continue tcp_input
 525                              */
 526                             tcp_input((struct mbuf *)NULL, sizeof(struct ip), so);
 527                             /* continue; */
 528                           } else
 529                             ret = sowrite(so);
 530                           /*
 531                            * XXXXX If we wrote something (a lot), there
 532                            * could be a need for a window update.
 533                            * In the worst case, the remote will send
 534                            * a window probe to get things going again
 535                            */
 536                         }
 537
 538                         /*
 539                          * Probe a still-connecting, non-blocking socket
 540                          * to check if it's still alive
 541                          */
 542 #ifdef PROBE_CONN
 543                         if (so->so_state & SS_ISFCONNECTING) {
 544                           ret = qemu_recv(so->s, &ret, 0,0);
 545
 546                           if (ret < 0) {
 547                             /* XXX */
 548                             if (errno == EAGAIN || errno == EWOULDBLOCK ||
 549                                 errno == EINPROGRESS || errno == ENOTCONN)
 550                               continue; /* Still connecting, continue */
 551
 552                             /* else failed */
 553                             so->so_state &= SS_PERSISTENT_MASK;
 554                             so->so_state |= SS_NOFDREF;
 555
 556                             /* tcp_input will take care of it */
 557                           } else {
 558                             ret = send(so->s, &ret, 0,0);
 559                             if (ret < 0) {
 560                               /* XXX */
 561                               if (errno == EAGAIN || errno == EWOULDBLOCK ||
 562                                   errno == EINPROGRESS || errno == ENOTCONN)
 563                                 continue;
 564                               /* else failed */
 565                               so->so_state &= SS_PERSISTENT_MASK;
 566                               so->so_state |= SS_NOFDREF;
 567                             } else
 568                               so->so_state &= ~SS_ISFCONNECTING;
 569
 570                           }
 571                           tcp_input((struct mbuf *)NULL, sizeof(struct ip),so);
 572                         } /* SS_ISFCONNECTING */
 573 #endif
 574                 }
 575
 576                 /*
 577                  * Now UDP sockets.
 578                  * Incoming packets are sent straight away, they're not buffered.
 579                  * Incoming UDP data isn't buffered either.
 580                  */
 581                 for (so = slirp->udb.so_next; so != &slirp->udb;
 582                      so = so_next) {
 583                         so_next = so->so_next;
 584
 585                         if (so->s != -1 && FD_ISSET(so->s, readfds)) {
 586                             sorecvfrom(so);
 587                         }
 588                 }
 589
 590                 /*
 591                  * Check incoming ICMP relies.
 592                  */
 593                 for (so = slirp->icmp.so_next; so != &slirp->icmp;
 594                      so = so_next) {
 595                      so_next = so->so_next;
 596
 597                     if (so->s != -1 && FD_ISSET(so->s, readfds)) {
 598                         icmp_receive(so);
 599                     }
 600                 }
 601         }
 602
 603         if_start(slirp);
 604     }
 605
 606         /* clear global file descriptor sets.
 607          * these reside on the stack in vl.c
 608          * so they're unusable if we're not in
 609          * slirp_select_fill or slirp_select_poll.
 610          */
 611          global_readfds = NULL;
 612          global_writefds = NULL;
 613          global_xfds = NULL;
 614 }
 615
 616 static void arp_input(Slirp *slirp, const uint8_t *pkt, int pkt_len)
 617 {
 618     struct arphdr *ah = (struct arphdr *)(pkt + ETH_HLEN);
 619     uint8_t arp_reply[max(ETH_HLEN + sizeof(struct arphdr), 64)];
 620     struct ethhdr *reh = (struct ethhdr *)arp_reply;
 621     struct arphdr *rah = (struct arphdr *)(arp_reply + ETH_HLEN);
 622     int ar_op;
 623     struct ex_list *ex_ptr;
 624
 625     ar_op = ntohs(ah->ar_op);
 626     switch(ar_op) {
 627     case ARPOP_REQUEST:
 628         if (ah->ar_tip == ah->ar_sip) {
 629             /* Gratuitous ARP */
 630             arp_table_add(slirp, ah->ar_sip, ah->ar_sha);
 631             return;
 632         }
 633
 634         if ((ah->ar_tip & slirp->vnetwork_mask.s_addr) ==
 635             slirp->vnetwork_addr.s_addr) {
 636             if (ah->ar_tip == slirp->vnameserver_addr.s_addr ||
 637                 ah->ar_tip == slirp->vhost_addr.s_addr)
 638                 goto arp_ok;
 639             for (ex_ptr = slirp->exec_list; ex_ptr; ex_ptr = ex_ptr->ex_next) {
 640                 if (ex_ptr->ex_addr.s_addr == ah->ar_tip)
 641                     goto arp_ok;
 642             }
 643             return;
 644         arp_ok:
 645             memset(arp_reply, 0, sizeof(arp_reply));
 646
 647             arp_table_add(slirp, ah->ar_sip, ah->ar_sha);
 648
 649             /* ARP request for alias/dns mac address */
 650             memcpy(reh->h_dest, pkt + ETH_ALEN, ETH_ALEN);
 651             memcpy(reh->h_source, special_ethaddr, ETH_ALEN - 4);
 652             memcpy(&reh->h_source[2], &ah->ar_tip, 4);
 653             reh->h_proto = htons(ETH_P_ARP);
 654
 655             rah->ar_hrd = htons(1);
 656             rah->ar_pro = htons(ETH_P_IP);
 657             rah->ar_hln = ETH_ALEN;
 658             rah->ar_pln = 4;
 659             rah->ar_op = htons(ARPOP_REPLY);
 660             memcpy(rah->ar_sha, reh->h_source, ETH_ALEN);
 661             rah->ar_sip = ah->ar_tip;
 662             memcpy(rah->ar_tha, ah->ar_sha, ETH_ALEN);
 663             rah->ar_tip = ah->ar_sip;
 664             slirp_output(slirp->opaque, arp_reply, sizeof(arp_reply));
 665         }
 666         break;
 667     case ARPOP_REPLY:
 668         arp_table_add(slirp, ah->ar_sip, ah->ar_sha);
 669         break;
 670     default:
 671         break;
 672     }
 673 }
 674
 675 void slirp_input(Slirp *slirp, const uint8_t *pkt, int pkt_len)
 676 {
 677     struct mbuf *m;
 678     int proto;
 679
 680     if (pkt_len < ETH_HLEN)
 681         return;
 682
 683     proto = ntohs(*(uint16_t *)(pkt + 12));
 684     switch(proto) {
 685     case ETH_P_ARP:
 686         arp_input(slirp, pkt, pkt_len);
 687         break;
 688     case ETH_P_IP:
 689         m = m_get(slirp);
 690         if (!m)
 691             return;
 692         /* Note: we add to align the IP header */
 693         if (M_FREEROOM(m) < pkt_len + 2) {
 694             m_inc(m, pkt_len + 2);
 695         }
 696         m->m_len = pkt_len + 2;
 697         memcpy(m->m_data + 2, pkt, pkt_len);
 698
 699         m->m_data += 2 + ETH_HLEN;
 700         m->m_len -= 2 + ETH_HLEN;
 701
 702         ip_input(m);
 703         break;
 704     default:
 705         break;
 706     }
 707 }
 708
 709 /* Output the IP packet to the ethernet device. Returns 0 if the packet must be
 710  * re-queued.
 711  */
 712 int if_encap(Slirp *slirp, struct mbuf *ifm)
 713 {
 714     uint8_t buf[1600];
 715     struct ethhdr *eh = (struct ethhdr *)buf;
 716     uint8_t ethaddr[ETH_ALEN];
 717     const struct ip *iph = (const struct ip *)ifm->m_data;
 718
 719     if (ifm->m_len + ETH_HLEN > sizeof(buf)) {
 720         return 1;
 721     }
 722
 723     if (!arp_table_search(slirp, iph->ip_dst.s_addr, ethaddr)) {
 724         uint8_t arp_req[ETH_HLEN + sizeof(struct arphdr)];
 725         struct ethhdr *reh = (struct ethhdr *)arp_req;
 726         struct arphdr *rah = (struct arphdr *)(arp_req + ETH_HLEN);
 727
 728         if (!ifm->arp_requested) {
 729             /* If the client addr is not known, send an ARP request */
 730             memset(reh->h_dest, 0xff, ETH_ALEN);
 731             memcpy(reh->h_source, special_ethaddr, ETH_ALEN - 4);
 732             memcpy(&reh->h_source[2], &slirp->vhost_addr, 4);
 733             reh->h_proto = htons(ETH_P_ARP);
 734             rah->ar_hrd = htons(1);
 735             rah->ar_pro = htons(ETH_P_IP);
 736             rah->ar_hln = ETH_ALEN;
 737             rah->ar_pln = 4;
 738             rah->ar_op = htons(ARPOP_REQUEST);
 739
 740             /* source hw addr */
 741             memcpy(rah->ar_sha, special_ethaddr, ETH_ALEN - 4);
 742             memcpy(&rah->ar_sha[2], &slirp->vhost_addr, 4);
 743
 744             /* source IP */
 745             rah->ar_sip = slirp->vhost_addr.s_addr;
 746
 747             /* target hw addr (none) */
 748             memset(rah->ar_tha, 0, ETH_ALEN);
 749
 750             /* target IP */
 751             rah->ar_tip = iph->ip_dst.s_addr;
 752             slirp->client_ipaddr = iph->ip_dst;
 753             slirp_output(slirp->opaque, arp_req, sizeof(arp_req));
 754             ifm->arp_requested = true;
 755
 756             /* Expire request and drop outgoing packet after 1 second */
 757             ifm->expiration_date = qemu_get_clock_ns(rt_clock) + 1000000000ULL;
 758         }
 759         return 0;
 760     } else {
 761         memcpy(eh->h_dest, ethaddr, ETH_ALEN);
 762         memcpy(eh->h_source, special_ethaddr, ETH_ALEN - 4);
 763         /* XXX: not correct */
 764         memcpy(&eh->h_source[2], &slirp->vhost_addr, 4);
 765         eh->h_proto = htons(ETH_P_IP);
 766         memcpy(buf + sizeof(struct ethhdr), ifm->m_data, ifm->m_len);
 767         slirp_output(slirp->opaque, buf, ifm->m_len + ETH_HLEN);
 768         return 1;
 769     }
 770 }
 771
 772 /* Drop host forwarding rule, return 0 if found. */
 773 int slirp_remove_hostfwd(Slirp *slirp, int is_udp, struct in_addr host_addr,
 774                          int host_port)
 775 {
 776     struct socket *so;
 777     struct socket *head = (is_udp ? &slirp->udb : &slirp->tcb);
 778     struct sockaddr_in addr;
 779     int port = htons(host_port);
 780     socklen_t addr_len;
 781
 782     for (so = head->so_next; so != head; so = so->so_next) {
 783         addr_len = sizeof(addr);
 784         if ((so->so_state & SS_HOSTFWD) &&
 785             getsockname(so->s, (struct sockaddr *)&addr, &addr_len) == 0 &&
 786             addr.sin_addr.s_addr == host_addr.s_addr &&
 787             addr.sin_port == port) {
 788             close(so->s);
 789             sofree(so);
 790             return 0;
 791         }
 792     }
 793
 794     return -1;
 795 }
 796
 797 int slirp_add_hostfwd(Slirp *slirp, int is_udp, struct in_addr host_addr,
 798                       int host_port, struct in_addr guest_addr, int guest_port)
 799 {
 800     if (!guest_addr.s_addr) {
 801         guest_addr = slirp->vdhcp_startaddr;
 802     }
 803     if (is_udp) {
 804         if (!udp_listen(slirp, host_addr.s_addr, htons(host_port),
 805                         guest_addr.s_addr, htons(guest_port), SS_HOSTFWD))
 806             return -1;
 807     } else {
 808         if (!tcp_listen(slirp, host_addr.s_addr, htons(host_port),
 809                         guest_addr.s_addr, htons(guest_port), SS_HOSTFWD))
 810             return -1;
 811     }
 812     return 0;
 813 }
 814
 815 int slirp_add_exec(Slirp *slirp, int do_pty, const void *args,
 816                    struct in_addr *guest_addr, int guest_port)
 817 {
 818     if (!guest_addr->s_addr) {
 819         guest_addr->s_addr = slirp->vnetwork_addr.s_addr |
 820             (htonl(0x0204) & ~slirp->vnetwork_mask.s_addr);
 821     }
 822     if ((guest_addr->s_addr & slirp->vnetwork_mask.s_addr) !=
 823         slirp->vnetwork_addr.s_addr ||
 824         guest_addr->s_addr == slirp->vhost_addr.s_addr ||
 825         guest_addr->s_addr == slirp->vnameserver_addr.s_addr) {
 826         return -1;
 827     }
 828     return add_exec(&slirp->exec_list, do_pty, (char *)args, *guest_addr,
 829                     htons(guest_port));
 830 }
 831
 832 ssize_t slirp_send(struct socket *so, const void *buf, size_t len, int flags)
 833 {
 834         if (so->s == -1 && so->extra) {
 835                 qemu_chr_fe_write(so->extra, buf, len);
 836                 return len;
 837         }
 838
 839         return send(so->s, buf, len, flags);
 840 }
 841
 842 static struct socket *
 843 slirp_find_ctl_socket(Slirp *slirp, struct in_addr guest_addr, int guest_port)
 844 {
 845     struct socket *so;
 846
 847     for (so = slirp->tcb.so_next; so != &slirp->tcb; so = so->so_next) {
 848         if (so->so_faddr.s_addr == guest_addr.s_addr &&
 849             htons(so->so_fport) == guest_port) {
 850             return so;
 851         }
 852     }
 853     return NULL;
 854 }
 855
 856 size_t slirp_socket_can_recv(Slirp *slirp, struct in_addr guest_addr,
 857                              int guest_port)
 858 {
 859         struct iovec iov[2];
 860         struct socket *so;
 861
 862         so = slirp_find_ctl_socket(slirp, guest_addr, guest_port);
 863
 864         if (!so || so->so_state & SS_NOFDREF)
 865                 return 0;
 866
 867         if (!CONN_CANFRCV(so) || so->so_snd.sb_cc >= (so->so_snd.sb_datalen/2))
 868                 return 0;
 869
 870         return sopreprbuf(so, iov, NULL);
 871 }
 872
 873 void slirp_socket_recv(Slirp *slirp, struct in_addr guest_addr, int guest_port,
 874                        const uint8_t *buf, int size)
 875 {
 876     int ret;
 877     struct socket *so = slirp_find_ctl_socket(slirp, guest_addr, guest_port);
 878
 879     if (!so)
 880         return;
 881
 882     ret = soreadbuf(so, (const char *)buf, size);
 883
 884     if (ret > 0)
 885         tcp_output(sototcpcb(so));
 886 }
 887
 888 static void slirp_tcp_save(QEMUFile *f, struct tcpcb *tp)
 889 {
 890     int i;
 891
 892     qemu_put_sbe16(f, tp->t_state);
 893     for (i = 0; i < TCPT_NTIMERS; i++)
 894         qemu_put_sbe16(f, tp->t_timer[i]);
 895     qemu_put_sbe16(f, tp->t_rxtshift);
 896     qemu_put_sbe16(f, tp->t_rxtcur);
 897     qemu_put_sbe16(f, tp->t_dupacks);
 898     qemu_put_be16(f, tp->t_maxseg);
 899     qemu_put_sbyte(f, tp->t_force);
 900     qemu_put_be16(f, tp->t_flags);
 901     qemu_put_be32(f, tp->snd_una);
 902     qemu_put_be32(f, tp->snd_nxt);
 903     qemu_put_be32(f, tp->snd_up);
 904     qemu_put_be32(f, tp->snd_wl1);
 905     qemu_put_be32(f, tp->snd_wl2);
 906     qemu_put_be32(f, tp->iss);
 907     qemu_put_be32(f, tp->snd_wnd);
 908     qemu_put_be32(f, tp->rcv_wnd);
 909     qemu_put_be32(f, tp->rcv_nxt);
 910     qemu_put_be32(f, tp->rcv_up);
 911     qemu_put_be32(f, tp->irs);
 912     qemu_put_be32(f, tp->rcv_adv);
 913     qemu_put_be32(f, tp->snd_max);
 914     qemu_put_be32(f, tp->snd_cwnd);
 915     qemu_put_be32(f, tp->snd_ssthresh);
 916     qemu_put_sbe16(f, tp->t_idle);
 917     qemu_put_sbe16(f, tp->t_rtt);
 918     qemu_put_be32(f, tp->t_rtseq);
 919     qemu_put_sbe16(f, tp->t_srtt);
 920     qemu_put_sbe16(f, tp->t_rttvar);
 921     qemu_put_be16(f, tp->t_rttmin);
 922     qemu_put_be32(f, tp->max_sndwnd);
 923     qemu_put_byte(f, tp->t_oobflags);
 924     qemu_put_byte(f, tp->t_iobc);
 925     qemu_put_sbe16(f, tp->t_softerror);
 926     qemu_put_byte(f, tp->snd_scale);
 927     qemu_put_byte(f, tp->rcv_scale);
 928     qemu_put_byte(f, tp->request_r_scale);
 929     qemu_put_byte(f, tp->requested_s_scale);
 930     qemu_put_be32(f, tp->ts_recent);
 931     qemu_put_be32(f, tp->ts_recent_age);
 932     qemu_put_be32(f, tp->last_ack_sent);
 933 }
 934
 935 static void slirp_sbuf_save(QEMUFile *f, struct sbuf *sbuf)
 936 {
 937     uint32_t off;
 938
 939     qemu_put_be32(f, sbuf->sb_cc);
 940     qemu_put_be32(f, sbuf->sb_datalen);
 941     off = (uint32_t)(sbuf->sb_wptr - sbuf->sb_data);
 942     qemu_put_sbe32(f, off);
 943     off = (uint32_t)(sbuf->sb_rptr - sbuf->sb_data);
 944     qemu_put_sbe32(f, off);
 945     qemu_put_buffer(f, (unsigned char*)sbuf->sb_data, sbuf->sb_datalen);
 946 }
 947
 948 static void slirp_socket_save(QEMUFile *f, struct socket *so)
 949 {
 950     qemu_put_be32(f, so->so_urgc);
 951     qemu_put_be32(f, so->so_faddr.s_addr);
 952     qemu_put_be32(f, so->so_laddr.s_addr);
 953     qemu_put_be16(f, so->so_fport);
 954     qemu_put_be16(f, so->so_lport);
 955     qemu_put_byte(f, so->so_iptos);
 956     qemu_put_byte(f, so->so_emu);
 957     qemu_put_byte(f, so->so_type);
 958     qemu_put_be32(f, so->so_state);
 959     slirp_sbuf_save(f, &so->so_rcv);
 960     slirp_sbuf_save(f, &so->so_snd);
 961     slirp_tcp_save(f, so->so_tcpcb);
 962 }
 963
 964 static void slirp_bootp_save(QEMUFile *f, Slirp *slirp)
 965 {
 966     int i;
 967
 968     for (i = 0; i < NB_BOOTP_CLIENTS; i++) {
 969         qemu_put_be16(f, slirp->bootp_clients[i].allocated);
 970         qemu_put_buffer(f, slirp->bootp_clients[i].macaddr, 6);
 971     }
 972 }
 973
 974 static void slirp_state_save(QEMUFile *f, void *opaque)
 975 {
 976     Slirp *slirp = opaque;
 977     struct ex_list *ex_ptr;
 978
 979     for (ex_ptr = slirp->exec_list; ex_ptr; ex_ptr = ex_ptr->ex_next)
 980         if (ex_ptr->ex_pty == 3) {
 981             struct socket *so;
 982             so = slirp_find_ctl_socket(slirp, ex_ptr->ex_addr,
 983                                        ntohs(ex_ptr->ex_fport));
 984             if (!so)
 985                 continue;
 986
 987             qemu_put_byte(f, 42);
 988             slirp_socket_save(f, so);
 989         }
 990     qemu_put_byte(f, 0);
 991
 992     qemu_put_be16(f, slirp->ip_id);
 993
 994     slirp_bootp_save(f, slirp);
 995 }
 996
 997 static void slirp_tcp_load(QEMUFile *f, struct tcpcb *tp)
 998 {
 999     int i;
1000
1001     tp->t_state = qemu_get_sbe16(f);
1002     for (i = 0; i < TCPT_NTIMERS; i++)
1003         tp->t_timer[i] = qemu_get_sbe16(f);
1004     tp->t_rxtshift = qemu_get_sbe16(f);
1005     tp->t_rxtcur = qemu_get_sbe16(f);
1006     tp->t_dupacks = qemu_get_sbe16(f);
1007     tp->t_maxseg = qemu_get_be16(f);
1008     tp->t_force = qemu_get_sbyte(f);
1009     tp->t_flags = qemu_get_be16(f);
1010     tp->snd_una = qemu_get_be32(f);
1011     tp->snd_nxt = qemu_get_be32(f);
1012     tp->snd_up = qemu_get_be32(f);
1013     tp->snd_wl1 = qemu_get_be32(f);
1014     tp->snd_wl2 = qemu_get_be32(f);
1015     tp->iss = qemu_get_be32(f);
1016     tp->snd_wnd = qemu_get_be32(f);
1017     tp->rcv_wnd = qemu_get_be32(f);
1018     tp->rcv_nxt = qemu_get_be32(f);
1019     tp->rcv_up = qemu_get_be32(f);
1020     tp->irs = qemu_get_be32(f);
1021     tp->rcv_adv = qemu_get_be32(f);
1022     tp->snd_max = qemu_get_be32(f);
1023     tp->snd_cwnd = qemu_get_be32(f);
1024     tp->snd_ssthresh = qemu_get_be32(f);
1025     tp->t_idle = qemu_get_sbe16(f);
1026     tp->t_rtt = qemu_get_sbe16(f);
1027     tp->t_rtseq = qemu_get_be32(f);
1028     tp->t_srtt = qemu_get_sbe16(f);
1029     tp->t_rttvar = qemu_get_sbe16(f);
1030     tp->t_rttmin = qemu_get_be16(f);
1031     tp->max_sndwnd = qemu_get_be32(f);
1032     tp->t_oobflags = qemu_get_byte(f);
1033     tp->t_iobc = qemu_get_byte(f);
1034     tp->t_softerror = qemu_get_sbe16(f);
1035     tp->snd_scale = qemu_get_byte(f);
1036     tp->rcv_scale = qemu_get_byte(f);
1037     tp->request_r_scale = qemu_get_byte(f);
1038     tp->requested_s_scale = qemu_get_byte(f);
1039     tp->ts_recent = qemu_get_be32(f);
1040     tp->ts_recent_age = qemu_get_be32(f);
1041     tp->last_ack_sent = qemu_get_be32(f);
1042     tcp_template(tp);
1043 }
1044
1045 static int slirp_sbuf_load(QEMUFile *f, struct sbuf *sbuf)
1046 {
1047     uint32_t off, sb_cc, sb_datalen;
1048
1049     sb_cc = qemu_get_be32(f);
1050     sb_datalen = qemu_get_be32(f);
1051
1052     sbreserve(sbuf, sb_datalen);
1053
1054     if (sbuf->sb_datalen != sb_datalen)
1055         return -ENOMEM;
1056
1057     sbuf->sb_cc = sb_cc;
1058
1059     off = qemu_get_sbe32(f);
1060     sbuf->sb_wptr = sbuf->sb_data + off;
1061     off = qemu_get_sbe32(f);
1062     sbuf->sb_rptr = sbuf->sb_data + off;
1063     qemu_get_buffer(f, (unsigned char*)sbuf->sb_data, sbuf->sb_datalen);
1064
1065     return 0;
1066 }
1067
1068 static int slirp_socket_load(QEMUFile *f, struct socket *so)
1069 {
1070     if (tcp_attach(so) < 0)
1071         return -ENOMEM;
1072
1073     so->so_urgc = qemu_get_be32(f);
1074     so->so_faddr.s_addr = qemu_get_be32(f);
1075     so->so_laddr.s_addr = qemu_get_be32(f);
1076     so->so_fport = qemu_get_be16(f);
1077     so->so_lport = qemu_get_be16(f);
1078     so->so_iptos = qemu_get_byte(f);
1079     so->so_emu = qemu_get_byte(f);
1080     so->so_type = qemu_get_byte(f);
1081     so->so_state = qemu_get_be32(f);
1082     if (slirp_sbuf_load(f, &so->so_rcv) < 0)
1083         return -ENOMEM;
1084     if (slirp_sbuf_load(f, &so->so_snd) < 0)
1085         return -ENOMEM;
1086     slirp_tcp_load(f, so->so_tcpcb);
1087
1088     return 0;
1089 }
1090
1091 static void slirp_bootp_load(QEMUFile *f, Slirp *slirp)
1092 {
1093     int i;
1094
1095     for (i = 0; i < NB_BOOTP_CLIENTS; i++) {
1096         slirp->bootp_clients[i].allocated = qemu_get_be16(f);
1097         qemu_get_buffer(f, slirp->bootp_clients[i].macaddr, 6);
1098     }
1099 }
1100
1101 static int slirp_state_load(QEMUFile *f, void *opaque, int version_id)
1102 {
1103     Slirp *slirp = opaque;
1104     struct ex_list *ex_ptr;
1105
1106     while (qemu_get_byte(f)) {
1107         int ret;
1108         struct socket *so = socreate(slirp);
1109
1110         if (!so)
1111             return -ENOMEM;
1112
1113         ret = slirp_socket_load(f, so);
1114
1115         if (ret < 0)
1116             return ret;
1117
1118         if ((so->so_faddr.s_addr & slirp->vnetwork_mask.s_addr) !=
1119             slirp->vnetwork_addr.s_addr) {
1120             return -EINVAL;
1121         }
1122         for (ex_ptr = slirp->exec_list; ex_ptr; ex_ptr = ex_ptr->ex_next) {
1123             if (ex_ptr->ex_pty == 3 &&
1124                 so->so_faddr.s_addr == ex_ptr->ex_addr.s_addr &&
1125                 so->so_fport == ex_ptr->ex_fport) {
1126                 break;
1127             }
1128         }
1129         if (!ex_ptr)
1130             return -EINVAL;
1131
1132         so->extra = (void *)ex_ptr->ex_exec;
1133     }
1134
1135     if (version_id >= 2) {
1136         slirp->ip_id = qemu_get_be16(f);
1137     }
1138
1139     if (version_id >= 3) {
1140         slirp_bootp_load(f, slirp);
1141     }
1142
1143     return 0;
1144 }