src/src/host.c

   1 /*************************************************
   2 *     Exim - an Internet mail transport agent    *
   3 *************************************************/
   4
   5 /* Copyright (c) University of Cambridge 1995 - 2009 */
   6 /* See the file NOTICE for conditions of use and distribution. */
   7
   8 /* Functions for finding hosts, either by gethostbyname(), gethostbyaddr(), or
   9 directly via the DNS. When IPv6 is supported, getipnodebyname() and
  10 getipnodebyaddr() may be used instead of gethostbyname() and gethostbyaddr(),
  11 if the newer functions are available. This module also contains various other
  12 functions concerned with hosts and addresses, and a random number function,
  13 used for randomizing hosts with equal MXs but available for use in other parts
  14 of Exim. */
  15
  16
  17 #include "exim.h"
  18
  19
  20 /* Static variable for preserving the list of interface addresses in case it is
  21 used more than once. */
  22
  23 static ip_address_item *local_interface_data = NULL;
  24
  25
  26 #ifdef USE_INET_NTOA_FIX
  27 /*************************************************
  28 *         Replacement for broken inet_ntoa()     *
  29 *************************************************/
  30
  31 /* On IRIX systems, gcc uses a different structure passing convention to the
  32 native libraries. This causes inet_ntoa() to always yield 0.0.0.0 or
  33 255.255.255.255. To get round this, we provide a private version of the
  34 function here. It is used only if USE_INET_NTOA_FIX is set, which should happen
  35 only when gcc is in use on an IRIX system. Code send to me by J.T. Breitner,
  36 with these comments:
  37
  38   code by Stuart Levy
  39   as seen in comp.sys.sgi.admin
  40
  41 August 2005: Apparently this is also needed for AIX systems; USE_INET_NTOA_FIX
  42 should now be set for them as well.
  43
  44 Arguments:  sa  an in_addr structure
  45 Returns:        pointer to static text string
  46 */
  47
  48 char *
  49 inet_ntoa(struct in_addr sa)
  50 {
  51 static uschar addr[20];
  52 sprintf(addr, "%d.%d.%d.%d",
  53         (US &sa.s_addr)[0],
  54         (US &sa.s_addr)[1],
  55         (US &sa.s_addr)[2],
  56         (US &sa.s_addr)[3]);
  57   return addr;
  58 }
  59 #endif
  60
  61
  62
  63 /*************************************************
  64 *              Random number generator           *
  65 *************************************************/
  66
  67 /* This is a simple pseudo-random number generator. It does not have to be
  68 very good for the uses to which it is put. When running the regression tests,
  69 start with a fixed seed.
  70
  71 If you need better, see pseudo_random_number() which is potentially stronger,
  72 if a crypto library is available, but might end up just calling this instead.
  73
  74 Arguments:
  75   limit:    one more than the largest number required
  76
  77 Returns:    a pseudo-random number in the range 0 to limit-1
  78 */
  79
  80 int
  81 random_number(int limit)
  82 {
  83 if (limit < 1)
  84   return 0;
  85 if (random_seed == 0)
  86   {
  87   if (running_in_test_harness) random_seed = 42; else
  88     {
  89     int p = (int)getpid();
  90     random_seed = (int)time(NULL) ^ ((p << 16) | p);
  91     }
  92   }
  93 random_seed = 1103515245 * random_seed + 12345;
  94 return (unsigned int)(random_seed >> 16) % limit;
  95 }
  96
  97
  98
  99 /*************************************************
 100 *       Replace gethostbyname() when testing     *
 101 *************************************************/
 102
 103 /* This function is called instead of gethostbyname(), gethostbyname2(), or
 104 getipnodebyname() when running in the test harness. It recognizes the name
 105 "manyhome.test.ex" and generates a humungous number of IP addresses. It also
 106 recognizes an unqualified "localhost" and forces it to the appropriate loopback
 107 address. IP addresses are treated as literals. For other names, it uses the DNS
 108 to find the host name. In the test harness, this means it will access only the
 109 fake DNS resolver.
 110
 111 Arguments:
 112   name          the host name or a textual IP address
 113   af            AF_INET or AF_INET6
 114   error_num     where to put an error code:
 115                 HOST_NOT_FOUND/TRY_AGAIN/NO_RECOVERY/NO_DATA
 116
 117 Returns:        a hostent structure or NULL for an error
 118 */
 119
 120 static struct hostent *
 121 host_fake_gethostbyname(uschar *name, int af, int *error_num)
 122 {
 123 #if HAVE_IPV6
 124 int alen = (af == AF_INET)? sizeof(struct in_addr):sizeof(struct in6_addr);
 125 #else
 126 int alen = sizeof(struct in_addr);
 127 #endif
 128
 129 int ipa;
 130 uschar *lname = name;
 131 uschar *adds;
 132 uschar **alist;
 133 struct hostent *yield;
 134 dns_answer dnsa;
 135 dns_scan dnss;
 136 dns_record *rr;
 137
 138 DEBUG(D_host_lookup)
 139   debug_printf("using host_fake_gethostbyname for %s (%s)\n", name,
 140     (af == AF_INET)? "IPv4" : "IPv6");
 141
 142 /* Handle the name that needs a vast number of IP addresses */
 143
 144 if (Ustrcmp(name, "manyhome.test.ex") == 0 && af == AF_INET)
 145   {
 146   int i, j;
 147   yield = store_get(sizeof(struct hostent));
 148   alist = store_get(2049 * sizeof(char *));
 149   adds  = store_get(2048 * alen);
 150   yield->h_name = CS name;
 151   yield->h_aliases = NULL;
 152   yield->h_addrtype = af;
 153   yield->h_length = alen;
 154   yield->h_addr_list = CSS alist;
 155   for (i = 104; i <= 111; i++)
 156     {
 157     for (j = 0; j <= 255; j++)
 158       {
 159       *alist++ = adds;
 160       *adds++ = 10;
 161       *adds++ = 250;
 162       *adds++ = i;
 163       *adds++ = j;
 164       }
 165     }
 166   *alist = NULL;
 167   return yield;
 168   }
 169
 170 /* Handle unqualified "localhost" */
 171
 172 if (Ustrcmp(name, "localhost") == 0)
 173   lname = (af == AF_INET)? US"127.0.0.1" : US"::1";
 174
 175 /* Handle a literal IP address */
 176
 177 ipa = string_is_ip_address(lname, NULL);
 178 if (ipa != 0)
 179   {
 180   if ((ipa == 4 && af == AF_INET) ||
 181       (ipa == 6 && af == AF_INET6))
 182     {
 183     int i, n;
 184     int x[4];
 185     yield = store_get(sizeof(struct hostent));
 186     alist = store_get(2 * sizeof(char *));
 187     adds  = store_get(alen);
 188     yield->h_name = CS name;
 189     yield->h_aliases = NULL;
 190     yield->h_addrtype = af;
 191     yield->h_length = alen;
 192     yield->h_addr_list = CSS alist;
 193     *alist++ = adds;
 194     n = host_aton(lname, x);
 195     for (i = 0; i < n; i++)
 196       {
 197       int y = x[i];
 198       *adds++ = (y >> 24) & 255;
 199       *adds++ = (y >> 16) & 255;
 200       *adds++ = (y >> 8) & 255;
 201       *adds++ = y & 255;
 202       }
 203     *alist = NULL;
 204     }
 205
 206   /* Wrong kind of literal address */
 207
 208   else
 209     {
 210     *error_num = HOST_NOT_FOUND;
 211     return NULL;
 212     }
 213   }
 214
 215 /* Handle a host name */
 216
 217 else
 218   {
 219   int type = (af == AF_INET)? T_A:T_AAAA;
 220   int rc = dns_lookup(&dnsa, lname, type, NULL);
 221   int count = 0;
 222
 223   switch(rc)
 224     {
 225     case DNS_SUCCEED: break;
 226     case DNS_NOMATCH: *error_num = HOST_NOT_FOUND; return NULL;
 227     case DNS_NODATA:  *error_num = NO_DATA; return NULL;
 228     case DNS_AGAIN:   *error_num = TRY_AGAIN; return NULL;
 229     default:
 230     case DNS_FAIL:    *error_num = NO_RECOVERY; return NULL;
 231     }
 232
 233   for (rr = dns_next_rr(&dnsa, &dnss, RESET_ANSWERS);
 234        rr != NULL;
 235        rr = dns_next_rr(&dnsa, &dnss, RESET_NEXT))
 236     {
 237     if (rr->type == type) count++;
 238     }
 239
 240   yield = store_get(sizeof(struct hostent));
 241   alist = store_get((count + 1) * sizeof(char **));
 242   adds  = store_get(count *alen);
 243
 244   yield->h_name = CS name;
 245   yield->h_aliases = NULL;
 246   yield->h_addrtype = af;
 247   yield->h_length = alen;
 248   yield->h_addr_list = CSS alist;
 249
 250   for (rr = dns_next_rr(&dnsa, &dnss, RESET_ANSWERS);
 251        rr != NULL;
 252        rr = dns_next_rr(&dnsa, &dnss, RESET_NEXT))
 253     {
 254     int i, n;
 255     int x[4];
 256     dns_address *da;
 257     if (rr->type != type) continue;
 258     da = dns_address_from_rr(&dnsa, rr);
 259     *alist++ = adds;
 260     n = host_aton(da->address, x);
 261     for (i = 0; i < n; i++)
 262       {
 263       int y = x[i];
 264       *adds++ = (y >> 24) & 255;
 265       *adds++ = (y >> 16) & 255;
 266       *adds++ = (y >> 8) & 255;
 267       *adds++ = y & 255;
 268       }
 269     }
 270   *alist = NULL;
 271   }
 272
 273 return yield;
 274 }
 275
 276
 277
 278 /*************************************************
 279 *       Build chain of host items from list      *
 280 *************************************************/
 281
 282 /* This function builds a chain of host items from a textual list of host
 283 names. It does not do any lookups. If randomize is true, the chain is build in
 284 a randomized order. There may be multiple groups of independently randomized
 285 hosts; they are delimited by a host name consisting of just "+".
 286
 287 Arguments:
 288   anchor      anchor for the chain
 289   list        text list
 290   randomize   TRUE for randomizing
 291
 292 Returns:      nothing
 293 */
 294
 295 void
 296 host_build_hostlist(host_item **anchor, uschar *list, BOOL randomize)
 297 {
 298 int sep = 0;
 299 int fake_mx = MX_NONE;          /* This value is actually -1 */
 300 uschar *name;
 301 uschar buffer[1024];
 302
 303 if (list == NULL) return;
 304 if (randomize) fake_mx--;       /* Start at -2 for randomizing */
 305
 306 *anchor = NULL;
 307
 308 while ((name = string_nextinlist(&list, &sep, buffer, sizeof(buffer))) != NULL)
 309   {
 310   host_item *h;
 311
 312   if (name[0] == '+' && name[1] == 0)   /* "+" delimits a randomized group */
 313     {                                   /* ignore if not randomizing */
 314     if (randomize) fake_mx--;
 315     continue;
 316     }
 317
 318   h = store_get(sizeof(host_item));
 319   h->name = string_copy(name);
 320   h->address = NULL;
 321   h->port = PORT_NONE;
 322   h->mx = fake_mx;
 323   h->sort_key = randomize? (-fake_mx)*1000 + random_number(1000) : 0;
 324   h->status = hstatus_unknown;
 325   h->why = hwhy_unknown;
 326   h->last_try = 0;
 327
 328   if (*anchor == NULL)
 329     {
 330     h->next = NULL;
 331     *anchor = h;
 332     }
 333   else
 334     {
 335     host_item *hh = *anchor;
 336     if (h->sort_key < hh->sort_key)
 337       {
 338       h->next = hh;
 339       *anchor = h;
 340       }
 341     else
 342       {
 343       while (hh->next != NULL && h->sort_key >= (hh->next)->sort_key)
 344         hh = hh->next;
 345       h->next = hh->next;
 346       hh->next = h;
 347       }
 348     }
 349   }
 350 }
 351
 352
 353
 354
 355
 356 /*************************************************
 357 *        Extract port from address string        *
 358 *************************************************/
 359
 360 /* In the spool file, and in the -oMa and -oMi options, a host plus port is
 361 given as an IP address followed by a dot and a port number. This function
 362 decodes this.
 363
 364 An alternative format for the -oMa and -oMi options is [ip address]:port which
 365 is what Exim 4 uses for output, because it seems to becoming commonly used,
 366 whereas the dot form confuses some programs/people. So we recognize that form
 367 too.
 368
 369 Argument:
 370   address    points to the string; if there is a port, the '.' in the string
 371              is overwritten with zero to terminate the address; if the string
 372              is in the [xxx]:ppp format, the address is shifted left and the
 373              brackets are removed
 374
 375 Returns:     0 if there is no port, else the port number. If there's a syntax
 376              error, leave the incoming address alone, and return 0.
 377 */
 378
 379 int
 380 host_address_extract_port(uschar *address)
 381 {
 382 int port = 0;
 383 uschar *endptr;
 384
 385 /* Handle the "bracketed with colon on the end" format */
 386
 387 if (*address == '[')
 388   {
 389   uschar *rb = address + 1;
 390   while (*rb != 0 && *rb != ']') rb++;
 391   if (*rb++ == 0) return 0;        /* Missing ]; leave invalid address */
 392   if (*rb == ':')
 393     {
 394     port = Ustrtol(rb + 1, &endptr, 10);
 395     if (*endptr != 0) return 0;    /* Invalid port; leave invalid address */
 396     }
 397   else if (*rb != 0) return 0;     /* Bad syntax; leave invalid address */
 398   memmove(address, address + 1, rb - address - 2);
 399   rb[-2] = 0;
 400   }
 401
 402 /* Handle the "dot on the end" format */
 403
 404 else
 405   {
 406   int skip = -3;                   /* Skip 3 dots in IPv4 addresses */
 407   address--;
 408   while (*(++address) != 0)
 409     {
 410     int ch = *address;
 411     if (ch == ':') skip = 0;       /* Skip 0 dots in IPv6 addresses */
 412       else if (ch == '.' && skip++ >= 0) break;
 413     }
 414   if (*address == 0) return 0;
 415   port = Ustrtol(address + 1, &endptr, 10);
 416   if (*endptr != 0) return 0;      /* Invalid port; leave invalid address */
 417   *address = 0;
 418   }
 419
 420 return port;
 421 }
 422
 423
 424 /*************************************************
 425 *         Get port from a host item's name       *
 426 *************************************************/
 427
 428 /* This function is called when finding the IP address for a host that is in a
 429 list of hosts explicitly configured, such as in the manualroute router, or in a
 430 fallback hosts list. We see if there is a port specification at the end of the
 431 host name, and if so, remove it. A minimum length of 3 is required for the
 432 original name; nothing shorter is recognized as having a port.
 433
 434 We test for a name ending with a sequence of digits; if preceded by colon we
 435 have a port if the character before the colon is ] and the name starts with [
 436 or if there are no other colons in the name (i.e. it's not an IPv6 address).
 437
 438 Arguments:  pointer to the host item
 439 Returns:    a port number or PORT_NONE
 440 */
 441
 442 int
 443 host_item_get_port(host_item *h)
 444 {
 445 uschar *p;
 446 int port, x;
 447 int len = Ustrlen(h->name);
 448
 449 if (len < 3 || (p = h->name + len - 1, !isdigit(*p))) return PORT_NONE;
 450
 451 /* Extract potential port number */
 452
 453 port = *p-- - '0';
 454 x = 10;
 455
 456 while (p > h->name + 1 && isdigit(*p))
 457   {
 458   port += (*p-- - '0') * x;
 459   x *= 10;
 460   }
 461
 462 /* The smallest value of p at this point is h->name + 1. */
 463
 464 if (*p != ':') return PORT_NONE;
 465
 466 if (p[-1] == ']' && h->name[0] == '[')
 467   h->name = string_copyn(h->name + 1, p - h->name - 2);
 468 else if (Ustrchr(h->name, ':') == p)
 469   h->name = string_copyn(h->name, p - h->name);
 470 else return PORT_NONE;
 471
 472 DEBUG(D_route|D_host_lookup) debug_printf("host=%s port=%d\n", h->name, port);
 473 return port;
 474 }
 475
 476
 477
 478 #ifndef STAND_ALONE    /* Omit when standalone testing */
 479
 480 /*************************************************
 481 *     Build sender_fullhost and sender_rcvhost   *
 482 *************************************************/
 483
 484 /* This function is called when sender_host_name and/or sender_helo_name
 485 have been set. Or might have been set - for a local message read off the spool
 486 they won't be. In that case, do nothing. Otherwise, set up the fullhost string
 487 as follows:
 488
 489 (a) No sender_host_name or sender_helo_name: "[ip address]"
 490 (b) Just sender_host_name: "host_name [ip address]"
 491 (c) Just sender_helo_name: "(helo_name) [ip address]" unless helo is IP
 492             in which case: "[ip address}"
 493 (d) The two are identical: "host_name [ip address]" includes helo = IP
 494 (e) The two are different: "host_name (helo_name) [ip address]"
 495
 496 If log_incoming_port is set, the sending host's port number is added to the IP
 497 address.
 498
 499 This function also builds sender_rcvhost for use in Received: lines, whose
 500 syntax is a bit different. This value also includes the RFC 1413 identity.
 501 There wouldn't be two different variables if I had got all this right in the
 502 first place.
 503
 504 Because this data may survive over more than one incoming SMTP message, it has
 505 to be in permanent store.
 506
 507 Arguments:  none
 508 Returns:    nothing
 509 */
 510
 511 void
 512 host_build_sender_fullhost(void)
 513 {
 514 BOOL show_helo = TRUE;
 515 uschar *address;
 516 int len;
 517 int old_pool = store_pool;
 518
 519 if (sender_host_address == NULL) return;
 520
 521 store_pool = POOL_PERM;
 522
 523 /* Set up address, with or without the port. After discussion, it seems that
 524 the only format that doesn't cause trouble is [aaaa]:pppp. However, we can't
 525 use this directly as the first item for Received: because it ain't an RFC 2822
 526 domain. Sigh. */
 527
 528 address = string_sprintf("[%s]:%d", sender_host_address, sender_host_port);
 529 if ((log_extra_selector & LX_incoming_port) == 0 || sender_host_port <= 0)
 530   *(Ustrrchr(address, ':')) = 0;
 531
 532 /* If there's no EHLO/HELO data, we can't show it. */
 533
 534 if (sender_helo_name == NULL) show_helo = FALSE;
 535
 536 /* If HELO/EHLO was followed by an IP literal, it's messy because of two
 537 features of IPv6. Firstly, there's the "IPv6:" prefix (Exim is liberal and
 538 doesn't require this, for historical reasons). Secondly, IPv6 addresses may not
 539 be given in canonical form, so we have to canonicize them before comparing. As
 540 it happens, the code works for both IPv4 and IPv6. */
 541
 542 else if (sender_helo_name[0] == '[' &&
 543          sender_helo_name[(len=Ustrlen(sender_helo_name))-1] == ']')
 544   {
 545   int offset = 1;
 546   uschar *helo_ip;
 547
 548   if (strncmpic(sender_helo_name + 1, US"IPv6:", 5) == 0) offset += 5;
 549   if (strncmpic(sender_helo_name + 1, US"IPv4:", 5) == 0) offset += 5;
 550
 551   helo_ip = string_copyn(sender_helo_name + offset, len - offset - 1);
 552
 553   if (string_is_ip_address(helo_ip, NULL) != 0)
 554     {
 555     int x[4], y[4];
 556     int sizex, sizey;
 557     uschar ipx[48], ipy[48];    /* large enough for full IPv6 */
 558
 559     sizex = host_aton(helo_ip, x);
 560     sizey = host_aton(sender_host_address, y);
 561
 562     (void)host_nmtoa(sizex, x, -1, ipx, ':');
 563     (void)host_nmtoa(sizey, y, -1, ipy, ':');
 564
 565     if (strcmpic(ipx, ipy) == 0) show_helo = FALSE;
 566     }
 567   }
 568
 569 /* Host name is not verified */
 570
 571 if (sender_host_name == NULL)
 572   {
 573   uschar *portptr = Ustrstr(address, "]:");
 574   int size = 0;
 575   int ptr = 0;
 576   int adlen;    /* Sun compiler doesn't like ++ in initializers */
 577
 578   adlen = (portptr == NULL)? Ustrlen(address) : (++portptr - address);
 579   sender_fullhost = (sender_helo_name == NULL)? address :
 580     string_sprintf("(%s) %s", sender_helo_name, address);
 581
 582   sender_rcvhost = string_cat(NULL, &size, &ptr, address, adlen);
 583
 584   if (sender_ident != NULL || show_helo || portptr != NULL)
 585     {
 586     int firstptr;
 587     sender_rcvhost = string_cat(sender_rcvhost, &size, &ptr, US" (", 2);
 588     firstptr = ptr;
 589
 590     if (portptr != NULL)
 591       sender_rcvhost = string_append(sender_rcvhost, &size, &ptr, 2, US"port=",
 592         portptr + 1);
 593
 594     if (show_helo)
 595       sender_rcvhost = string_append(sender_rcvhost, &size, &ptr, 2,
 596         (firstptr == ptr)? US"helo=" : US" helo=", sender_helo_name);
 597
 598     if (sender_ident != NULL)
 599       sender_rcvhost = string_append(sender_rcvhost, &size, &ptr, 2,
 600         (firstptr == ptr)? US"ident=" : US" ident=", sender_ident);
 601
 602     sender_rcvhost = string_cat(sender_rcvhost, &size, &ptr, US")", 1);
 603     }
 604
 605   sender_rcvhost[ptr] = 0;   /* string_cat() always leaves room */
 606
 607   /* Release store, because string_cat allocated a minimum of 100 bytes that
 608   are rarely completely used. */
 609
 610   store_reset(sender_rcvhost + ptr + 1);
 611   }
 612
 613 /* Host name is known and verified. Unless we've already found that the HELO
 614 data matches the IP address, compare it with the name. */
 615
 616 else
 617   {
 618   if (show_helo && strcmpic(sender_host_name, sender_helo_name) == 0)
 619     show_helo = FALSE;
 620
 621   if (show_helo)
 622     {
 623     sender_fullhost = string_sprintf("%s (%s) %s", sender_host_name,
 624       sender_helo_name, address);
 625     sender_rcvhost = (sender_ident == NULL)?
 626       string_sprintf("%s (%s helo=%s)", sender_host_name,
 627         address, sender_helo_name) :
 628       string_sprintf("%s\n\t(%s helo=%s ident=%s)", sender_host_name,
 629         address, sender_helo_name, sender_ident);
 630     }
 631   else
 632     {
 633     sender_fullhost = string_sprintf("%s %s", sender_host_name, address);
 634     sender_rcvhost = (sender_ident == NULL)?
 635       string_sprintf("%s (%s)", sender_host_name, address) :
 636       string_sprintf("%s (%s ident=%s)", sender_host_name, address,
 637         sender_ident);
 638     }
 639   }
 640
 641 store_pool = old_pool;
 642
 643 DEBUG(D_host_lookup) debug_printf("sender_fullhost = %s\n", sender_fullhost);
 644 DEBUG(D_host_lookup) debug_printf("sender_rcvhost = %s\n", sender_rcvhost);
 645 }
 646
 647
 648
 649 /*************************************************
 650 *          Build host+ident message              *
 651 *************************************************/
 652
 653 /* Used when logging rejections and various ACL and SMTP incidents. The text
 654 return depends on whether sender_fullhost and sender_ident are set or not:
 655
 656   no ident, no host   => U=unknown
 657   no ident, host set  => H=sender_fullhost
 658   ident set, no host  => U=ident
 659   ident set, host set => H=sender_fullhost U=ident
 660
 661 Arguments:
 662   useflag   TRUE if first item to be flagged (H= or U=); if there are two
 663               items, the second is always flagged
 664
 665 Returns:    pointer to a string in big_buffer
 666 */
 667
 668 uschar *
 669 host_and_ident(BOOL useflag)
 670 {
 671 if (sender_fullhost == NULL)
 672   {
 673   (void)string_format(big_buffer, big_buffer_size, "%s%s", useflag? "U=" : "",
 674      (sender_ident == NULL)? US"unknown" : sender_ident);
 675   }
 676 else
 677   {
 678   uschar *flag = useflag? US"H=" : US"";
 679   uschar *iface = US"";
 680   if ((log_extra_selector & LX_incoming_interface) != 0 &&
 681        interface_address != NULL)
 682     iface = string_sprintf(" I=[%s]:%d", interface_address, interface_port);
 683   if (sender_ident == NULL)
 684     (void)string_format(big_buffer, big_buffer_size, "%s%s%s",
 685       flag, sender_fullhost, iface);
 686   else
 687     (void)string_format(big_buffer, big_buffer_size, "%s%s%s U=%s",
 688       flag, sender_fullhost, iface, sender_ident);
 689   }
 690 return big_buffer;
 691 }
 692
 693 #endif   /* STAND_ALONE */
 694
 695
 696
 697
 698 /*************************************************
 699 *         Build list of local interfaces         *
 700 *************************************************/
 701
 702 /* This function interprets the contents of the local_interfaces or
 703 extra_local_interfaces options, and creates an ip_address_item block for each
 704 item on the list. There is no special interpretation of any IP addresses; in
 705 particular, 0.0.0.0 and ::0 are returned without modification. If any address
 706 includes a port, it is set in the block. Otherwise the port value is set to
 707 zero.
 708
 709 Arguments:
 710   list        the list
 711   name        the name of the option being expanded
 712
 713 Returns:      a chain of ip_address_items, each containing to a textual
 714               version of an IP address, and a port number (host order) or
 715               zero if no port was given with the address
 716 */
 717
 718 ip_address_item *
 719 host_build_ifacelist(uschar *list, uschar *name)
 720 {
 721 int sep = 0;
 722 uschar *s;
 723 uschar buffer[64];
 724 ip_address_item *yield = NULL;
 725 ip_address_item *last = NULL;
 726 ip_address_item *next;
 727
 728 while ((s = string_nextinlist(&list, &sep, buffer, sizeof(buffer))) != NULL)
 729   {
 730   int ipv;
 731   int port = host_address_extract_port(s);            /* Leaves just the IP address */
 732   if ((ipv = string_is_ip_address(s, NULL)) == 0)
 733     log_write(0, LOG_MAIN|LOG_PANIC_DIE, "Malformed IP address \"%s\" in %s",
 734       s, name);
 735
 736   /* Skip IPv6 addresses if IPv6 is disabled. */
 737
 738   if (disable_ipv6 && ipv == 6) continue;
 739
 740   /* This use of strcpy() is OK because we have checked that s is a valid IP
 741   address above. The field in the ip_address_item is large enough to hold an
 742   IPv6 address. */
 743
 744   next = store_get(sizeof(ip_address_item));
 745   next->next = NULL;
 746   Ustrcpy(next->address, s);
 747   next->port = port;
 748   next->v6_include_v4 = FALSE;
 749
 750   if (yield == NULL) yield = last = next; else
 751     {
 752     last->next = next;
 753     last = next;
 754     }
 755   }
 756
 757 return yield;
 758 }
 759
 760
 761
 762
 763
 764 /*************************************************
 765 *         Find addresses on local interfaces     *
 766 *************************************************/
 767
 768 /* This function finds the addresses of local IP interfaces. These are used
 769 when testing for routing to the local host. As the function may be called more
 770 than once, the list is preserved in permanent store, pointed to by a static
 771 variable, to save doing the work more than once per process.
 772
 773 The generic list of interfaces is obtained by calling host_build_ifacelist()
 774 for local_interfaces and extra_local_interfaces. This list scanned to remove
 775 duplicates (which may exist with different ports - not relevant here). If
 776 either of the wildcard IP addresses (0.0.0.0 and ::0) are encountered, they are
 777 replaced by the appropriate (IPv4 or IPv6) list of actual local interfaces,
 778 obtained from os_find_running_interfaces().
 779
 780 Arguments:    none
 781 Returns:      a chain of ip_address_items, each containing to a textual
 782               version of an IP address; the port numbers are not relevant
 783 */
 784
 785
 786 /* First, a local subfunction to add an interface to a list in permanent store,
 787 but only if there isn't a previous copy of that address on the list. */
 788
 789 static ip_address_item *
 790 add_unique_interface(ip_address_item *list, ip_address_item *ipa)
 791 {
 792 ip_address_item *ipa2;
 793 for (ipa2 = list; ipa2 != NULL; ipa2 = ipa2->next)
 794   if (Ustrcmp(ipa2->address, ipa->address) == 0) return list;
 795 ipa2 = store_get_perm(sizeof(ip_address_item));
 796 *ipa2 = *ipa;
 797 ipa2->next = list;
 798 return ipa2;
 799 }
 800
 801
 802 /* This is the globally visible function */
 803
 804 ip_address_item *
 805 host_find_interfaces(void)
 806 {
 807 ip_address_item *running_interfaces = NULL;
 808
 809 if (local_interface_data == NULL)
 810   {
 811   void *reset_item = store_get(0);
 812   ip_address_item *dlist = host_build_ifacelist(local_interfaces,
 813     US"local_interfaces");
 814   ip_address_item *xlist = host_build_ifacelist(extra_local_interfaces,
 815     US"extra_local_interfaces");
 816   ip_address_item *ipa;
 817
 818   if (dlist == NULL) dlist = xlist; else
 819     {
 820     for (ipa = dlist; ipa->next != NULL; ipa = ipa->next);
 821     ipa->next = xlist;
 822     }
 823
 824   for (ipa = dlist; ipa != NULL; ipa = ipa->next)
 825     {
 826     if (Ustrcmp(ipa->address, "0.0.0.0") == 0 ||
 827         Ustrcmp(ipa->address, "::0") == 0)
 828       {
 829       ip_address_item *ipa2;
 830       BOOL ipv6 = ipa->address[0] == ':';
 831       if (running_interfaces == NULL)
 832         running_interfaces = os_find_running_interfaces();
 833       for (ipa2 = running_interfaces; ipa2 != NULL; ipa2 = ipa2->next)
 834         {
 835         if ((Ustrchr(ipa2->address, ':') != NULL) == ipv6)
 836           local_interface_data = add_unique_interface(local_interface_data,
 837           ipa2);
 838         }
 839       }
 840     else
 841       {
 842       local_interface_data = add_unique_interface(local_interface_data, ipa);
 843       DEBUG(D_interface)
 844         {
 845         debug_printf("Configured local interface: address=%s", ipa->address);
 846         if (ipa->port != 0) debug_printf(" port=%d", ipa->port);
 847         debug_printf("\n");
 848         }
 849       }
 850     }
 851   store_reset(reset_item);
 852   }
 853
 854 return local_interface_data;
 855 }
 856
 857
 858
 859
 860
 861 /*************************************************
 862 *        Convert network IP address to text      *
 863 *************************************************/
 864
 865 /* Given an IPv4 or IPv6 address in binary, convert it to a text
 866 string and return the result in a piece of new store. The address can
 867 either be given directly, or passed over in a sockaddr structure. Note
 868 that this isn't the converse of host_aton() because of byte ordering
 869 differences. See host_nmtoa() below.
 870
 871 Arguments:
 872   type       if < 0 then arg points to a sockaddr, else
 873              either AF_INET or AF_INET6
 874   arg        points to a sockaddr if type is < 0, or
 875              points to an IPv4 address (32 bits), or
 876              points to an IPv6 address (128 bits),
 877              in both cases, in network byte order
 878   buffer     if NULL, the result is returned in gotten store;
 879              else points to a buffer to hold the answer
 880   portptr    points to where to put the port number, if non NULL; only
 881              used when type < 0
 882
 883 Returns:     pointer to character string
 884 */
 885
 886 uschar *
 887 host_ntoa(int type, const void *arg, uschar *buffer, int *portptr)
 888 {
 889 uschar *yield;
 890
 891 /* The new world. It is annoying that we have to fish out the address from
 892 different places in the block, depending on what kind of address it is. It
 893 is also a pain that inet_ntop() returns a const uschar *, whereas the IPv4
 894 function inet_ntoa() returns just uschar *, and some picky compilers insist
 895 on warning if one assigns a const uschar * to a uschar *. Hence the casts. */
 896
 897 #if HAVE_IPV6
 898 uschar addr_buffer[46];
 899 if (type < 0)
 900   {
 901   int family = ((struct sockaddr *)arg)->sa_family;
 902   if (family == AF_INET6)
 903     {
 904     struct sockaddr_in6 *sk = (struct sockaddr_in6 *)arg;
 905     yield = (uschar *)inet_ntop(family, &(sk->sin6_addr), CS addr_buffer,
 906       sizeof(addr_buffer));
 907     if (portptr != NULL) *portptr = ntohs(sk->sin6_port);
 908     }
 909   else
 910     {
 911     struct sockaddr_in *sk = (struct sockaddr_in *)arg;
 912     yield = (uschar *)inet_ntop(family, &(sk->sin_addr), CS addr_buffer,
 913       sizeof(addr_buffer));
 914     if (portptr != NULL) *portptr = ntohs(sk->sin_port);
 915     }
 916   }
 917 else
 918   {
 919   yield = (uschar *)inet_ntop(type, arg, CS addr_buffer, sizeof(addr_buffer));
 920   }
 921
 922 /* If the result is a mapped IPv4 address, show it in V4 format. */
 923
 924 if (Ustrncmp(yield, "::ffff:", 7) == 0) yield += 7;
 925
 926 #else  /* HAVE_IPV6 */
 927
 928 /* The old world */
 929
 930 if (type < 0)
 931   {
 932   yield = US inet_ntoa(((struct sockaddr_in *)arg)->sin_addr);
 933   if (portptr != NULL) *portptr = ntohs(((struct sockaddr_in *)arg)->sin_port);
 934   }
 935 else
 936   yield = US inet_ntoa(*((struct in_addr *)arg));
 937 #endif
 938
 939 /* If there is no buffer, put the string into some new store. */
 940
 941 if (buffer == NULL) return string_copy(yield);
 942
 943 /* Callers of this function with a non-NULL buffer must ensure that it is
 944 large enough to hold an IPv6 address, namely, at least 46 bytes. That's what
 945 makes this use of strcpy() OK. */
 946
 947 Ustrcpy(buffer, yield);
 948 return buffer;
 949 }
 950
 951
 952
 953
 954 /*************************************************
 955 *         Convert address text to binary         *
 956 *************************************************/
 957
 958 /* Given the textual form of an IP address, convert it to binary in an
 959 array of ints. IPv4 addresses occupy one int; IPv6 addresses occupy 4 ints.
 960 The result has the first byte in the most significant byte of the first int. In
 961 other words, the result is not in network byte order, but in host byte order.
 962 As a result, this is not the converse of host_ntoa(), which expects network
 963 byte order. See host_nmtoa() below.
 964
 965 Arguments:
 966   address    points to the textual address, checked for syntax
 967   bin        points to an array of 4 ints
 968
 969 Returns:     the number of ints used
 970 */
 971
 972 int
 973 host_aton(uschar *address, int *bin)
 974 {
 975 int x[4];
 976 int v4offset = 0;
 977
 978 /* Handle IPv6 address, which may end with an IPv4 address. It may also end
 979 with a "scope", introduced by a percent sign. This code is NOT enclosed in #if
 980 HAVE_IPV6 in order that IPv6 addresses are recognized even if IPv6 is not
 981 supported. */
 982
 983 if (Ustrchr(address, ':') != NULL)
 984   {
 985   uschar *p = address;
 986   uschar *component[8];
 987   BOOL ipv4_ends = FALSE;
 988   int ci = 0;
 989   int nulloffset = 0;
 990   int v6count = 8;
 991   int i;
 992
 993   /* If the address starts with a colon, it will start with two colons.
 994   Just lose the first one, which will leave a null first component. */
 995
 996   if (*p == ':') p++;
 997
 998   /* Split the address into components separated by colons. The input address
 999   is supposed to be checked for syntax. There was a case where this was
1000   overlooked; to guard against that happening again, check here and crash if
1001   there are too many components. */
1002
1003   while (*p != 0 && *p != '%')
1004     {
1005     int len = Ustrcspn(p, ":%");
1006     if (len == 0) nulloffset = ci;
1007     if (ci > 7) log_write(0, LOG_MAIN|LOG_PANIC_DIE,
1008       "Internal error: invalid IPv6 address \"%s\" passed to host_aton()",
1009       address);
1010     component[ci++] = p;
1011     p += len;
1012     if (*p == ':') p++;
1013     }
1014
1015   /* If the final component contains a dot, it is a trailing v4 address.
1016   As the syntax is known to be checked, just set up for a trailing
1017   v4 address and restrict the v6 part to 6 components. */
1018
1019   if (Ustrchr(component[ci-1], '.') != NULL)
1020     {
1021     address = component[--ci];
1022     ipv4_ends = TRUE;
1023     v4offset = 3;
1024     v6count = 6;
1025     }
1026
1027   /* If there are fewer than 6 or 8 components, we have to insert some
1028   more empty ones in the middle. */
1029
1030   if (ci < v6count)
1031     {
1032     int insert_count = v6count - ci;
1033     for (i = v6count-1; i > nulloffset + insert_count; i--)
1034       component[i] = component[i - insert_count];
1035     while (i > nulloffset) component[i--] = US"";
1036     }
1037
1038   /* Now turn the components into binary in pairs and bung them
1039   into the vector of ints. */
1040
1041   for (i = 0; i < v6count; i += 2)
1042     bin[i/2] = (Ustrtol(component[i], NULL, 16) << 16) +
1043       Ustrtol(component[i+1], NULL, 16);
1044
1045   /* If there was no terminating v4 component, we are done. */
1046
1047   if (!ipv4_ends) return 4;
1048   }
1049
1050 /* Handle IPv4 address */
1051
1052 (void)sscanf(CS address, "%d.%d.%d.%d", x, x+1, x+2, x+3);
1053 bin[v4offset] = (x[0] << 24) + (x[1] << 16) + (x[2] << 8) + x[3];
1054 return v4offset+1;
1055 }
1056
1057
1058 /*************************************************
1059 *           Apply mask to an IP address          *
1060 *************************************************/
1061
1062 /* Mask an address held in 1 or 4 ints, with the ms bit in the ms bit of the
1063 first int, etc.
1064
1065 Arguments:
1066   count        the number of ints
1067   binary       points to the ints to be masked
1068   mask         the count of ms bits to leave, or -1 if no masking
1069
1070 Returns:       nothing
1071 */
1072
1073 void
1074 host_mask(int count, int *binary, int mask)
1075 {
1076 int i;
1077 if (mask < 0) mask = 99999;
1078 for (i = 0; i < count; i++)
1079   {
1080   int wordmask;
1081   if (mask == 0) wordmask = 0;
1082   else if (mask < 32)
1083     {
1084     wordmask = (-1) << (32 - mask);
1085     mask = 0;
1086     }
1087   else
1088     {
1089     wordmask = -1;
1090     mask -= 32;
1091     }
1092   binary[i] &= wordmask;
1093   }
1094 }
1095
1096
1097
1098
1099 /*************************************************
1100 *     Convert masked IP address in ints to text  *
1101 *************************************************/
1102
1103 /* We can't use host_ntoa() because it assumes the binary values are in network
1104 byte order, and these are the result of host_aton(), which puts them in ints in
1105 host byte order. Also, we really want IPv6 addresses to be in a canonical
1106 format, so we output them with no abbreviation. In a number of cases we can't
1107 use the normal colon separator in them because it terminates keys in lsearch
1108 files, so we want to use dot instead. There's an argument that specifies what
1109 to use for IPv6 addresses.
1110
1111 Arguments:
1112   count       1 or 4 (number of ints)
1113   binary      points to the ints
1114   mask        mask value; if < 0 don't add to result
1115   buffer      big enough to hold the result
1116   sep         component separator character for IPv6 addresses
1117
1118 Returns:      the number of characters placed in buffer, not counting
1119               the final nul.
1120 */
1121
1122 int
1123 host_nmtoa(int count, int *binary, int mask, uschar *buffer, int sep)
1124 {
1125 int i, j;
1126 uschar *tt = buffer;
1127
1128 if (count == 1)
1129   {
1130   j = binary[0];
1131   for (i = 24; i >= 0; i -= 8)
1132     {
1133     sprintf(CS tt, "%d.", (j >> i) & 255);
1134     while (*tt) tt++;
1135     }
1136   }
1137 else
1138   {
1139   for (i = 0; i < 4; i++)
1140     {
1141     j = binary[i];
1142     sprintf(CS tt, "%04x%c%04x%c", (j >> 16) & 0xffff, sep, j & 0xffff, sep);
1143     while (*tt) tt++;
1144     }
1145   }
1146
1147 tt--;   /* lose final separator */
1148
1149 if (mask < 0)
1150   *tt = 0;
1151 else
1152   {
1153   sprintf(CS tt, "/%d", mask);
1154   while (*tt) tt++;
1155   }
1156
1157 return tt - buffer;
1158 }
1159
1160
1161
1162 /*************************************************
1163 *        Check port for tls_on_connect           *
1164 *************************************************/
1165
1166 /* This function checks whether a given incoming port is configured for tls-
1167 on-connect. It is called from the daemon and from inetd handling. If the global
1168 option tls_on_connect is already set, all ports operate this way. Otherwise, we
1169 check the tls_on_connect_ports option for a list of ports.
1170
1171 Argument:  a port number
1172 Returns:   TRUE or FALSE
1173 */
1174
1175 BOOL
1176 host_is_tls_on_connect_port(int port)
1177 {
1178 int sep = 0;
1179 uschar buffer[32];
1180 uschar *list = tls_on_connect_ports;
1181 uschar *s;
1182
1183 if (tls_on_connect) return TRUE;
1184
1185 while ((s = string_nextinlist(&list, &sep, buffer, sizeof(buffer))) != NULL)
1186   {
1187   uschar *end;
1188   int lport = Ustrtol(s, &end, 10);
1189   if (*end != 0) log_write(0, LOG_MAIN|LOG_PANIC_DIE, "tls_on_connect_ports "
1190     "contains \"%s\", which is not a port number: exim abandoned", s);
1191   if (lport == port) return TRUE;
1192   }
1193
1194 return FALSE;
1195 }
1196
1197
1198
1199 /*************************************************
1200 *        Check whether host is in a network      *
1201 *************************************************/
1202
1203 /* This function checks whether a given IP address matches a pattern that
1204 represents either a single host, or a network (using CIDR notation). The caller
1205 of this function must check the syntax of the arguments before calling it.
1206
1207 Arguments:
1208   host        string representation of the ip-address to check
1209   net         string representation of the network, with optional CIDR mask
1210   maskoffset  offset to the / that introduces the mask in the key
1211               zero if there is no mask
1212
1213 Returns:
1214   TRUE   the host is inside the network
1215   FALSE  the host is NOT inside the network
1216 */
1217
1218 BOOL
1219 host_is_in_net(uschar *host, uschar *net, int maskoffset)
1220 {
1221 int i;
1222 int address[4];
1223 int incoming[4];
1224 int mlen;
1225 int size = host_aton(net, address);
1226 int insize;
1227
1228 /* No mask => all bits to be checked */
1229
1230 if (maskoffset == 0) mlen = 99999;    /* Big number */
1231   else mlen = Uatoi(net + maskoffset + 1);
1232
1233 /* Convert the incoming address to binary. */
1234
1235 insize = host_aton(host, incoming);
1236
1237 /* Convert IPv4 addresses given in IPv6 compatible mode, which represent
1238    connections from IPv4 hosts to IPv6 hosts, that is, addresses of the form
1239    ::ffff:<v4address>, to IPv4 format. */
1240
1241 if (insize == 4 && incoming[0] == 0 && incoming[1] == 0 &&
1242     incoming[2] == 0xffff)
1243   {
1244   insize = 1;
1245   incoming[0] = incoming[3];
1246   }
1247
1248 /* No match if the sizes don't agree. */
1249
1250 if (insize != size) return FALSE;
1251
1252 /* Else do the masked comparison. */
1253
1254 for (i = 0; i < size; i++)
1255   {
1256   int mask;
1257   if (mlen == 0) mask = 0;
1258   else if (mlen < 32)
1259     {
1260     mask = (-1) << (32 - mlen);
1261     mlen = 0;
1262     }
1263   else
1264     {
1265     mask = -1;
1266     mlen -= 32;
1267     }
1268   if ((incoming[i] & mask) != (address[i] & mask)) return FALSE;
1269   }
1270
1271 return TRUE;
1272 }
1273
1274
1275
1276 /*************************************************
1277 *       Scan host list for local hosts           *
1278 *************************************************/
1279
1280 /* Scan through a chain of addresses and check whether any of them is the
1281 address of an interface on the local machine. If so, remove that address and
1282 any previous ones with the same MX value, and all subsequent ones (which will
1283 have greater or equal MX values) from the chain. Note: marking them as unusable
1284 is NOT the right thing to do because it causes the hosts not to be used for
1285 other domains, for which they may well be correct.
1286
1287 The hosts may be part of a longer chain; we only process those between the
1288 initial pointer and the "last" pointer.
1289
1290 There is also a list of "pseudo-local" host names which are checked against the
1291 host names. Any match causes that host item to be treated the same as one which
1292 matches a local IP address.
1293
1294 If the very first host is a local host, then all MX records had a precedence
1295 greater than or equal to that of the local host. Either there's a problem in
1296 the DNS, or an apparently remote name turned out to be an abbreviation for the
1297 local host. Give a specific return code, and let the caller decide what to do.
1298 Otherwise, give a success code if at least one host address has been found.
1299
1300 Arguments:
1301   host        pointer to the first host in the chain
1302   lastptr     pointer to pointer to the last host in the chain (may be updated)
1303   removed     if not NULL, set TRUE if some local addresses were removed
1304                 from the list
1305
1306 Returns:
1307   HOST_FOUND       if there is at least one host with an IP address on the chain
1308                      and an MX value less than any MX value associated with the
1309                      local host
1310   HOST_FOUND_LOCAL if a local host is among the lowest-numbered MX hosts; when
1311                      the host addresses were obtained from A records or
1312                      gethostbyname(), the MX values are set to -1.
1313   HOST_FIND_FAILED if no valid hosts with set IP addresses were found
1314 */
1315
1316 int
1317 host_scan_for_local_hosts(host_item *host, host_item **lastptr, BOOL *removed)
1318 {
1319 int yield = HOST_FIND_FAILED;
1320 host_item *last = *lastptr;
1321 host_item *prev = NULL;
1322 host_item *h;
1323
1324 if (removed != NULL) *removed = FALSE;
1325
1326 if (local_interface_data == NULL) local_interface_data = host_find_interfaces();
1327
1328 for (h = host; h != last->next; h = h->next)
1329   {
1330   #ifndef STAND_ALONE
1331   if (hosts_treat_as_local != NULL)
1332     {
1333     int rc;
1334     uschar *save = deliver_domain;
1335     deliver_domain = h->name;   /* set $domain */
1336     rc = match_isinlist(string_copylc(h->name), &hosts_treat_as_local, 0,
1337       &domainlist_anchor, NULL, MCL_DOMAIN, TRUE, NULL);
1338     deliver_domain = save;
1339     if (rc == OK) goto FOUND_LOCAL;
1340     }
1341   #endif
1342
1343   /* It seems that on many operating systems, 0.0.0.0 is treated as a synonym
1344   for 127.0.0.1 and refers to the local host. We therefore force it always to
1345   be treated as local. */
1346
1347   if (h->address != NULL)
1348     {
1349     ip_address_item *ip;
1350     if (Ustrcmp(h->address, "0.0.0.0") == 0) goto FOUND_LOCAL;
1351     for (ip = local_interface_data; ip != NULL; ip = ip->next)
1352       if (Ustrcmp(h->address, ip->address) == 0) goto FOUND_LOCAL;
1353     yield = HOST_FOUND;  /* At least one remote address has been found */
1354     }
1355
1356   /* Update prev to point to the last host item before any that have
1357   the same MX value as the one we have just considered. */
1358
1359   if (h->next == NULL || h->next->mx != h->mx) prev = h;
1360   }
1361
1362 return yield;  /* No local hosts found: return HOST_FOUND or HOST_FIND_FAILED */
1363
1364 /* A host whose IP address matches a local IP address, or whose name matches
1365 something in hosts_treat_as_local has been found. */
1366
1367 FOUND_LOCAL:
1368
1369 if (prev == NULL)
1370   {
1371   HDEBUG(D_host_lookup) debug_printf((h->mx >= 0)?
1372     "local host has lowest MX\n" :
1373     "local host found for non-MX address\n");
1374   return HOST_FOUND_LOCAL;
1375   }
1376
1377 HDEBUG(D_host_lookup)
1378   {
1379   debug_printf("local host in host list - removed hosts:\n");
1380   for (h = prev->next; h != last->next; h = h->next)
1381     debug_printf("  %s %s %d\n", h->name, h->address, h->mx);
1382   }
1383
1384 if (removed != NULL) *removed = TRUE;
1385 prev->next = last->next;
1386 *lastptr = prev;
1387 return yield;
1388 }
1389
1390
1391
1392
1393 /*************************************************
1394 *        Remove duplicate IPs in host list       *
1395 *************************************************/
1396
1397 /* You would think that administrators could set up their DNS records so that
1398 one ended up with a list of unique IP addresses after looking up A or MX
1399 records, but apparently duplication is common. So we scan such lists and
1400 remove the later duplicates. Note that we may get lists in which some host
1401 addresses are not set.
1402
1403 Arguments:
1404   host        pointer to the first host in the chain
1405   lastptr     pointer to pointer to the last host in the chain (may be updated)
1406
1407 Returns:      nothing
1408 */
1409
1410 static void
1411 host_remove_duplicates(host_item *host, host_item **lastptr)
1412 {
1413 while (host != *lastptr)
1414   {
1415   if (host->address != NULL)
1416     {
1417     host_item *h = host;
1418     while (h != *lastptr)
1419       {
1420       if (h->next->address != NULL &&
1421           Ustrcmp(h->next->address, host->address) == 0)
1422         {
1423         DEBUG(D_host_lookup) debug_printf("duplicate IP address %s (MX=%d) "
1424           "removed\n", host->address, h->next->mx);
1425         if (h->next == *lastptr) *lastptr = h;
1426         h->next = h->next->next;
1427         }
1428       else h = h->next;
1429       }
1430     }
1431   /* If the last item was removed, host may have become == *lastptr */
1432   if (host != *lastptr) host = host->next;
1433   }
1434 }
1435
1436
1437
1438
1439 /*************************************************
1440 *    Find sender host name by gethostbyaddr()    *
1441 *************************************************/
1442
1443 /* This used to be the only way it was done, but it turns out that not all
1444 systems give aliases for calls to gethostbyaddr() - or one of the modern
1445 equivalents like getipnodebyaddr(). Fortunately, multiple PTR records are rare,
1446 but they can still exist. This function is now used only when a DNS lookup of
1447 the IP address fails, in order to give access to /etc/hosts.
1448
1449 Arguments:   none
1450 Returns:     OK, DEFER, FAIL
1451 */
1452
1453 static int
1454 host_name_lookup_byaddr(void)
1455 {
1456 int len;
1457 uschar *s, *t;
1458 struct hostent *hosts;
1459 struct in_addr addr;
1460
1461 /* Lookup on IPv6 system */
1462
1463 #if HAVE_IPV6
1464 if (Ustrchr(sender_host_address, ':') != NULL)
1465   {
1466   struct in6_addr addr6;
1467   if (inet_pton(AF_INET6, CS sender_host_address, &addr6) != 1)
1468     log_write(0, LOG_MAIN|LOG_PANIC_DIE, "unable to parse \"%s\" as an "
1469       "IPv6 address", sender_host_address);
1470   #if HAVE_GETIPNODEBYADDR
1471   hosts = getipnodebyaddr(CS &addr6, sizeof(addr6), AF_INET6, &h_errno);
1472   #else
1473   hosts = gethostbyaddr(CS &addr6, sizeof(addr6), AF_INET6);
1474   #endif
1475   }
1476 else
1477   {
1478   if (inet_pton(AF_INET, CS sender_host_address, &addr) != 1)
1479     log_write(0, LOG_MAIN|LOG_PANIC_DIE, "unable to parse \"%s\" as an "
1480       "IPv4 address", sender_host_address);
1481   #if HAVE_GETIPNODEBYADDR
1482   hosts = getipnodebyaddr(CS &addr, sizeof(addr), AF_INET, &h_errno);
1483   #else
1484   hosts = gethostbyaddr(CS &addr, sizeof(addr), AF_INET);
1485   #endif
1486   }
1487
1488 /* Do lookup on IPv4 system */
1489
1490 #else
1491 addr.s_addr = (S_ADDR_TYPE)inet_addr(CS sender_host_address);
1492 hosts = gethostbyaddr(CS(&addr), sizeof(addr), AF_INET);
1493 #endif
1494
1495 /* Failed to look up the host. */
1496
1497 if (hosts == NULL)
1498   {
1499   HDEBUG(D_host_lookup) debug_printf("IP address lookup failed: h_errno=%d\n",
1500     h_errno);
1501   return (h_errno == TRY_AGAIN || h_errno == NO_RECOVERY) ? DEFER : FAIL;
1502   }
1503
1504 /* It seems there are some records in the DNS that yield an empty name. We
1505 treat this as non-existent. In some operating systems, this is returned as an
1506 empty string; in others as a single dot. */
1507
1508 if (hosts->h_name == NULL || hosts->h_name[0] == 0 || hosts->h_name[0] == '.')
1509   {
1510   HDEBUG(D_host_lookup) debug_printf("IP address lookup yielded an empty name: "
1511     "treated as non-existent host name\n");
1512   return FAIL;
1513   }
1514
1515 /* Copy and lowercase the name, which is in static storage in many systems.
1516 Put it in permanent memory. */
1517
1518 s = (uschar *)hosts->h_name;
1519 len = Ustrlen(s) + 1;
1520 t = sender_host_name = store_get_perm(len);
1521 while (*s != 0) *t++ = tolower(*s++);
1522 *t = 0;
1523
1524 /* If the host has aliases, build a copy of the alias list */
1525
1526 if (hosts->h_aliases != NULL)
1527   {
1528   int count = 1;
1529   uschar **aliases, **ptr;
1530   for (aliases = USS hosts->h_aliases; *aliases != NULL; aliases++) count++;
1531   ptr = sender_host_aliases = store_get_perm(count * sizeof(uschar *));
1532   for (aliases = USS hosts->h_aliases; *aliases != NULL; aliases++)
1533     {
1534     uschar *s = *aliases;
1535     int len = Ustrlen(s) + 1;
1536     uschar *t = *ptr++ = store_get_perm(len);
1537     while (*s != 0) *t++ = tolower(*s++);
1538     *t = 0;
1539     }
1540   *ptr = NULL;
1541   }
1542
1543 return OK;
1544 }
1545
1546
1547
1548 /*************************************************
1549 *        Find host name for incoming call        *
1550 *************************************************/
1551
1552 /* Put the name in permanent store, pointed to by sender_host_name. We also set
1553 up a list of alias names, pointed to by sender_host_alias. The list is
1554 NULL-terminated. The incoming address is in sender_host_address, either in
1555 dotted-quad form for IPv4 or in colon-separated form for IPv6.
1556
1557 This function does a thorough check that the names it finds point back to the
1558 incoming IP address. Any that do not are discarded. Note that this is relied on
1559 by the ACL reverse_host_lookup check.
1560
1561 On some systems, get{host,ipnode}byaddr() appears to do this internally, but
1562 this it not universally true. Also, for release 4.30, this function was changed
1563 to do a direct DNS lookup first, by default[1], because it turns out that that
1564 is the only guaranteed way to find all the aliases on some systems. My
1565 experiments indicate that Solaris gethostbyaddr() gives the aliases for but
1566 Linux does not.
1567
1568 [1] The actual order is controlled by the host_lookup_order option.
1569
1570 Arguments:    none
1571 Returns:      OK on success, the answer being placed in the global variable
1572                 sender_host_name, with any aliases in a list hung off
1573                 sender_host_aliases
1574               FAIL if no host name can be found
1575               DEFER if a temporary error was encountered
1576
1577 The variable host_lookup_msg is set to an empty string on sucess, or to a
1578 reason for the failure otherwise, in a form suitable for tagging onto an error
1579 message, and also host_lookup_failed is set TRUE if the lookup failed. If there
1580 was a defer, host_lookup_deferred is set TRUE.
1581
1582 Any dynamically constructed string for host_lookup_msg must be in permanent
1583 store, because it might be used for several incoming messages on the same SMTP
1584 connection. */
1585
1586 int
1587 host_name_lookup(void)
1588 {
1589 int old_pool, rc;
1590 int sep = 0;
1591 uschar *hname, *save_hostname;
1592 uschar **aliases;
1593 uschar buffer[256];
1594 uschar *ordername;
1595 uschar *list = host_lookup_order;
1596 dns_record *rr;
1597 dns_answer dnsa;
1598 dns_scan dnss;
1599
1600 host_lookup_deferred = host_lookup_failed = FALSE;
1601
1602 HDEBUG(D_host_lookup)
1603   debug_printf("looking up host name for %s\n", sender_host_address);
1604
1605 /* For testing the case when a lookup does not complete, we have a special
1606 reserved IP address. */
1607
1608 if (running_in_test_harness &&
1609     Ustrcmp(sender_host_address, "99.99.99.99") == 0)
1610   {
1611   HDEBUG(D_host_lookup)
1612     debug_printf("Test harness: host name lookup returns DEFER\n");
1613   host_lookup_deferred = TRUE;
1614   return DEFER;
1615   }
1616
1617 /* Do lookups directly in the DNS or via gethostbyaddr() (or equivalent), in
1618 the order specified by the host_lookup_order option. */
1619
1620 while ((ordername = string_nextinlist(&list, &sep, buffer, sizeof(buffer)))
1621         != NULL)
1622   {
1623   if (strcmpic(ordername, US"bydns") == 0)
1624     {
1625     dns_init(FALSE, FALSE);
1626     dns_build_reverse(sender_host_address, buffer);
1627     rc = dns_lookup(&dnsa, buffer, T_PTR, NULL);
1628
1629     /* The first record we come across is used for the name; others are
1630     considered to be aliases. We have to scan twice, in order to find out the
1631     number of aliases. However, if all the names are empty, we will behave as
1632     if failure. (PTR records that yield empty names have been encountered in
1633     the DNS.) */
1634
1635     if (rc == DNS_SUCCEED)
1636       {
1637       uschar **aptr = NULL;
1638       int ssize = 264;
1639       int count = 0;
1640       int old_pool = store_pool;
1641
1642       store_pool = POOL_PERM;        /* Save names in permanent storage */
1643
1644       for (rr = dns_next_rr(&dnsa, &dnss, RESET_ANSWERS);
1645            rr != NULL;
1646            rr = dns_next_rr(&dnsa, &dnss, RESET_NEXT))
1647         {
1648         if (rr->type == T_PTR) count++;
1649         }
1650
1651       /* Get store for the list of aliases. For compatibility with
1652       gethostbyaddr, we make an empty list if there are none. */
1653
1654       aptr = sender_host_aliases = store_get(count * sizeof(uschar *));
1655
1656       /* Re-scan and extract the names */
1657
1658       for (rr = dns_next_rr(&dnsa, &dnss, RESET_ANSWERS);
1659            rr != NULL;
1660            rr = dns_next_rr(&dnsa, &dnss, RESET_NEXT))
1661         {
1662         uschar *s = NULL;
1663         if (rr->type != T_PTR) continue;
1664         s = store_get(ssize);
1665
1666         /* If an overlong response was received, the data will have been
1667         truncated and dn_expand may fail. */
1668
1669         if (dn_expand(dnsa.answer, dnsa.answer + dnsa.answerlen,
1670              (uschar *)(rr->data), (DN_EXPAND_ARG4_TYPE)(s), ssize) < 0)
1671           {
1672           log_write(0, LOG_MAIN, "host name alias list truncated for %s",
1673             sender_host_address);
1674           break;
1675           }
1676
1677         store_reset(s + Ustrlen(s) + 1);
1678         if (s[0] == 0)
1679           {
1680           HDEBUG(D_host_lookup) debug_printf("IP address lookup yielded an "
1681             "empty name: treated as non-existent host name\n");
1682           continue;
1683           }
1684         if (sender_host_name == NULL) sender_host_name = s;
1685           else *aptr++ = s;
1686         while (*s != 0) { *s = tolower(*s); s++; }
1687         }
1688
1689       *aptr = NULL;            /* End of alias list */
1690       store_pool = old_pool;   /* Reset store pool */
1691
1692       /* If we've found a names, break out of the "order" loop */
1693
1694       if (sender_host_name != NULL) break;
1695       }
1696
1697     /* If the DNS lookup deferred, we must also defer. */
1698
1699     if (rc == DNS_AGAIN)
1700       {
1701       HDEBUG(D_host_lookup)
1702         debug_printf("IP address PTR lookup gave temporary error\n");
1703       host_lookup_deferred = TRUE;
1704       return DEFER;
1705       }
1706     }
1707
1708   /* Do a lookup using gethostbyaddr() - or equivalent */
1709
1710   else if (strcmpic(ordername, US"byaddr") == 0)
1711     {
1712     HDEBUG(D_host_lookup)
1713       debug_printf("IP address lookup using gethostbyaddr()\n");
1714     rc = host_name_lookup_byaddr();
1715     if (rc == DEFER)
1716       {
1717       host_lookup_deferred = TRUE;
1718       return rc;                       /* Can't carry on */
1719       }
1720     if (rc == OK) break;               /* Found a name */
1721     }
1722   }      /* Loop for bydns/byaddr scanning */
1723
1724 /* If we have failed to find a name, return FAIL and log when required.
1725 NB host_lookup_msg must be in permanent store.  */
1726
1727 if (sender_host_name == NULL)
1728   {
1729   if (host_checking || !log_testing_mode)
1730     log_write(L_host_lookup_failed, LOG_MAIN, "no host name found for IP "
1731       "address %s", sender_host_address);
1732   host_lookup_msg = US" (failed to find host name from IP address)";
1733   host_lookup_failed = TRUE;
1734   return FAIL;
1735   }
1736
1737 HDEBUG(D_host_lookup)
1738   {
1739   uschar **aliases = sender_host_aliases;
1740   debug_printf("IP address lookup yielded %s\n", sender_host_name);
1741   while (*aliases != NULL) debug_printf("  alias %s\n", *aliases++);
1742   }
1743
1744 /* We need to verify that a forward lookup on the name we found does indeed
1745 correspond to the address. This is for security: in principle a malefactor who
1746 happened to own a reverse zone could set it to point to any names at all.
1747
1748 This code was present in versions of Exim before 3.20. At that point I took it
1749 out because I thought that gethostbyaddr() did the check anyway. It turns out
1750 that this isn't always the case, so it's coming back in at 4.01. This version
1751 is actually better, because it also checks aliases.
1752
1753 The code was made more robust at release 4.21. Prior to that, it accepted all
1754 the names if any of them had the correct IP address. Now the code checks all
1755 the names, and accepts only those that have the correct IP address. */
1756
1757 save_hostname = sender_host_name;   /* Save for error messages */
1758 aliases = sender_host_aliases;
1759 for (hname = sender_host_name; hname != NULL; hname = *aliases++)
1760   {
1761   int rc;
1762   BOOL ok = FALSE;
1763   host_item h;
1764   h.next = NULL;
1765   h.name = hname;
1766   h.mx = MX_NONE;
1767   h.address = NULL;
1768
1769   /* When called with the last argument FALSE, host_find_byname() won't return
1770   HOST_FOUND_LOCAL. If the incoming address is an IPv4 address expressed in
1771   IPv6 format, we must compare the IPv4 part to any IPv4 addresses. */
1772
1773   if ((rc = host_find_byname(&h, NULL, 0, NULL, FALSE)) == HOST_FOUND)
1774     {
1775     host_item *hh;
1776     HDEBUG(D_host_lookup) debug_printf("checking addresses for %s\n", hname);
1777     for (hh = &h; hh != NULL; hh = hh->next)
1778       {
1779       if (host_is_in_net(hh->address, sender_host_address, 0))
1780         {
1781         HDEBUG(D_host_lookup) debug_printf("  %s OK\n", hh->address);
1782         ok = TRUE;
1783         break;
1784         }
1785       else
1786         {
1787         HDEBUG(D_host_lookup) debug_printf("  %s\n", hh->address);
1788         }
1789       }
1790     if (!ok) HDEBUG(D_host_lookup)
1791       debug_printf("no IP address for %s matched %s\n", hname,
1792         sender_host_address);
1793     }
1794   else if (rc == HOST_FIND_AGAIN)
1795     {
1796     HDEBUG(D_host_lookup) debug_printf("temporary error for host name lookup\n");
1797     host_lookup_deferred = TRUE;
1798     sender_host_name = NULL;
1799     return DEFER;
1800     }
1801   else
1802     {
1803     HDEBUG(D_host_lookup) debug_printf("no IP addresses found for %s\n", hname);
1804     }
1805
1806   /* If this name is no good, and it's the sender name, set it null pro tem;
1807   if it's an alias, just remove it from the list. */
1808
1809   if (!ok)
1810     {
1811     if (hname == sender_host_name) sender_host_name = NULL; else
1812       {
1813       uschar **a;                              /* Don't amalgamate - some */
1814       a = --aliases;                           /* compilers grumble */
1815       while (*a != NULL) { *a = a[1]; a++; }
1816       }
1817     }
1818   }
1819
1820 /* If sender_host_name == NULL, it means we didn't like the name. Replace
1821 it with the first alias, if there is one. */
1822
1823 if (sender_host_name == NULL && *sender_host_aliases != NULL)
1824   sender_host_name = *sender_host_aliases++;
1825
1826 /* If we now have a main name, all is well. */
1827
1828 if (sender_host_name != NULL) return OK;
1829
1830 /* We have failed to find an address that matches. */
1831
1832 HDEBUG(D_host_lookup)
1833   debug_printf("%s does not match any IP address for %s\n",
1834     sender_host_address, save_hostname);
1835
1836 /* This message must be in permanent store */
1837
1838 old_pool = store_pool;
1839 store_pool = POOL_PERM;
1840 host_lookup_msg = string_sprintf(" (%s does not match any IP address for %s)",
1841   sender_host_address, save_hostname);
1842 store_pool = old_pool;
1843 host_lookup_failed = TRUE;
1844 return FAIL;
1845 }
1846
1847
1848
1849
1850 /*************************************************
1851 *    Find IP address(es) for host by name        *
1852 *************************************************/
1853
1854 /* The input is a host_item structure with the name filled in and the address
1855 field set to NULL. We use gethostbyname() or getipnodebyname() or
1856 gethostbyname2(), as appropriate. Of course, these functions may use the DNS,
1857 but they do not do MX processing. It appears, however, that in some systems the
1858 current setting of resolver options is used when one of these functions calls
1859 the resolver. For this reason, we call dns_init() at the start, with arguments
1860 influenced by bits in "flags", just as we do for host_find_bydns().
1861
1862 The second argument provides a host list (usually an IP list) of hosts to
1863 ignore. This makes it possible to ignore IPv6 link-local addresses or loopback
1864 addresses in unreasonable places.
1865
1866 The lookup may result in a change of name. For compatibility with the dns
1867 lookup, return this via fully_qualified_name as well as updating the host item.
1868 The lookup may also yield more than one IP address, in which case chain on
1869 subsequent host_item structures.
1870
1871 Arguments:
1872   host                   a host item with the name and MX filled in;
1873                            the address is to be filled in;
1874                            multiple IP addresses cause other host items to be
1875                              chained on.
1876   ignore_target_hosts    a list of hosts to ignore
1877   flags                  HOST_FIND_QUALIFY_SINGLE   ) passed to
1878                          HOST_FIND_SEARCH_PARENTS   )   dns_init()
1879   fully_qualified_name   if not NULL, set to point to host name for
1880                          compatibility with host_find_bydns
1881   local_host_check       TRUE if a check for the local host is wanted
1882
1883 Returns:                 HOST_FIND_FAILED  Failed to find the host or domain
1884                          HOST_FIND_AGAIN   Try again later
1885                          HOST_FOUND        Host found - data filled in
1886                          HOST_FOUND_LOCAL  Host found and is the local host
1887 */
1888
1889 int
1890 host_find_byname(host_item *host, uschar *ignore_target_hosts, int flags,
1891   uschar **fully_qualified_name, BOOL local_host_check)
1892 {
1893 int i, yield, times;
1894 uschar **addrlist;
1895 host_item *last = NULL;
1896 BOOL temp_error = FALSE;
1897 #if HAVE_IPV6
1898 int af;
1899 #endif
1900
1901 /* If we are in the test harness, a name ending in .test.again.dns always
1902 forces a temporary error response, unless the name is in
1903 dns_again_means_nonexist. */
1904
1905 if (running_in_test_harness)
1906   {
1907   uschar *endname = host->name + Ustrlen(host->name);
1908   if (Ustrcmp(endname - 14, "test.again.dns") == 0) goto RETURN_AGAIN;
1909   }
1910
1911 /* Make sure DNS options are set as required. This appears to be necessary in
1912 some circumstances when the get..byname() function actually calls the DNS. */
1913
1914 dns_init((flags & HOST_FIND_QUALIFY_SINGLE) != 0,
1915          (flags & HOST_FIND_SEARCH_PARENTS) != 0);
1916
1917 /* In an IPv6 world, unless IPv6 has been disabled, we need to scan for both
1918 kinds of address, so go round the loop twice. Note that we have ensured that
1919 AF_INET6 is defined even in an IPv4 world, which makes for slightly tidier
1920 code. However, if dns_ipv4_lookup matches the domain, we also just do IPv4
1921 lookups here (except when testing standalone). */
1922
1923 #if HAVE_IPV6
1924   #ifdef STAND_ALONE
1925   if (disable_ipv6)
1926   #else
1927   if (disable_ipv6 ||
1928     (dns_ipv4_lookup != NULL &&
1929         match_isinlist(host->name, &dns_ipv4_lookup, 0, NULL, NULL, MCL_DOMAIN,
1930           TRUE, NULL) == OK))
1931   #endif
1932
1933     { af = AF_INET; times = 1; }
1934   else
1935     { af = AF_INET6; times = 2; }
1936
1937 /* No IPv6 support */
1938
1939 #else   /* HAVE_IPV6 */
1940   times = 1;
1941 #endif  /* HAVE_IPV6 */
1942
1943 /* Initialize the flag that gets set for DNS syntax check errors, so that the
1944 interface to this function can be similar to host_find_bydns. */
1945
1946 host_find_failed_syntax = FALSE;
1947
1948 /* Loop to look up both kinds of address in an IPv6 world */
1949
1950 for (i = 1; i <= times;
1951      #if HAVE_IPV6
1952        af = AF_INET,     /* If 2 passes, IPv4 on the second */
1953      #endif
1954      i++)
1955   {
1956   BOOL ipv4_addr;
1957   int error_num = 0;
1958   struct hostent *hostdata;
1959
1960   #ifdef STAND_ALONE
1961   printf("Looking up: %s\n", host->name);
1962   #endif
1963
1964   #if HAVE_IPV6
1965   if (running_in_test_harness)
1966     hostdata = host_fake_gethostbyname(host->name, af, &error_num);
1967   else
1968     {
1969     #if HAVE_GETIPNODEBYNAME
1970     hostdata = getipnodebyname(CS host->name, af, 0, &error_num);
1971     #else
1972     hostdata = gethostbyname2(CS host->name, af);
1973     error_num = h_errno;
1974     #endif
1975     }
1976
1977   #else    /* not HAVE_IPV6 */
1978   if (running_in_test_harness)
1979     hostdata = host_fake_gethostbyname(host->name, AF_INET, &error_num);
1980   else
1981     {
1982     hostdata = gethostbyname(CS host->name);
1983     error_num = h_errno;
1984     }
1985   #endif   /* HAVE_IPV6 */
1986
1987   if (hostdata == NULL)
1988     {
1989     uschar *error;
1990     switch (error_num)
1991       {
1992       case HOST_NOT_FOUND: error = US"HOST_NOT_FOUND"; break;
1993       case TRY_AGAIN: error = US"TRY_AGAIN"; break;
1994       case NO_RECOVERY: error = US"NO_RECOVERY"; break;
1995       case NO_DATA: error = US"NO_DATA"; break;
1996       #if NO_DATA != NO_ADDRESS
1997       case NO_ADDRESS: error = US"NO_ADDRESS"; break;
1998       #endif
1999       default: error = US"?"; break;
2000       }
2001
2002     DEBUG(D_host_lookup) debug_printf("%s returned %d (%s)\n",
2003       #if HAVE_IPV6
2004         #if HAVE_GETIPNODEBYNAME
2005         (af == AF_INET6)? "getipnodebyname(af=inet6)" : "getipnodebyname(af=inet)",
2006         #else
2007         (af == AF_INET6)? "gethostbyname2(af=inet6)" : "gethostbyname2(af=inet)",
2008         #endif
2009       #else
2010       "gethostbyname",
2011       #endif
2012       error_num, error);
2013
2014     if (error_num == TRY_AGAIN || error_num == NO_RECOVERY) temp_error = TRUE;
2015     continue;
2016     }
2017   if ((hostdata->h_addr_list)[0] == NULL) continue;
2018
2019   /* Replace the name with the fully qualified one if necessary, and fill in
2020   the fully_qualified_name pointer. */
2021
2022   if (hostdata->h_name[0] != 0 &&
2023       Ustrcmp(host->name, hostdata->h_name) != 0)
2024     host->name = string_copy_dnsdomain((uschar *)hostdata->h_name);
2025   if (fully_qualified_name != NULL) *fully_qualified_name = host->name;
2026
2027   /* Get the list of addresses. IPv4 and IPv6 addresses can be distinguished
2028   by their different lengths. Scan the list, ignoring any that are to be
2029   ignored, and build a chain from the rest. */
2030
2031   ipv4_addr = hostdata->h_length == sizeof(struct in_addr);
2032
2033   for (addrlist = USS hostdata->h_addr_list; *addrlist != NULL; addrlist++)
2034     {
2035     uschar *text_address =
2036       host_ntoa(ipv4_addr? AF_INET:AF_INET6, *addrlist, NULL, NULL);
2037
2038     #ifndef STAND_ALONE
2039     if (ignore_target_hosts != NULL &&
2040         verify_check_this_host(&ignore_target_hosts, NULL, host->name,
2041           text_address, NULL) == OK)
2042       {
2043       DEBUG(D_host_lookup)
2044         debug_printf("ignored host %s [%s]\n", host->name, text_address);
2045       continue;
2046       }
2047     #endif
2048
2049     /* If this is the first address, last == NULL and we put the data in the
2050     original block. */
2051
2052     if (last == NULL)
2053       {
2054       host->address = text_address;
2055       host->port = PORT_NONE;
2056       host->status = hstatus_unknown;
2057       host->why = hwhy_unknown;
2058       last = host;
2059       }
2060
2061     /* Else add further host item blocks for any other addresses, keeping
2062     the order. */
2063
2064     else
2065       {
2066       host_item *next = store_get(sizeof(host_item));
2067       next->name = host->name;
2068       next->mx = host->mx;
2069       next->address = text_address;
2070       next->port = PORT_NONE;
2071       next->status = hstatus_unknown;
2072       next->why = hwhy_unknown;
2073       next->last_try = 0;
2074       next->next = last->next;
2075       last->next = next;
2076       last = next;
2077       }
2078     }
2079   }
2080
2081 /* If no hosts were found, the address field in the original host block will be
2082 NULL. If temp_error is set, at least one of the lookups gave a temporary error,
2083 so we pass that back. */
2084
2085 if (host->address == NULL)
2086   {
2087   uschar *msg =
2088     #ifndef STAND_ALONE
2089     (message_id[0] == 0 && smtp_in != NULL)?
2090       string_sprintf("no IP address found for host %s (during %s)", host->name,
2091           smtp_get_connection_info()) :
2092     #endif
2093     string_sprintf("no IP address found for host %s", host->name);
2094
2095   HDEBUG(D_host_lookup) debug_printf("%s\n", msg);
2096   if (temp_error) goto RETURN_AGAIN;
2097   if (host_checking || !log_testing_mode)
2098     log_write(L_host_lookup_failed, LOG_MAIN, "%s", msg);
2099   return HOST_FIND_FAILED;
2100   }
2101
2102 /* Remove any duplicate IP addresses, then check to see if this is the local
2103 host if required. */
2104
2105 host_remove_duplicates(host, &last);
2106 yield = local_host_check?
2107   host_scan_for_local_hosts(host, &last, NULL) : HOST_FOUND;
2108
2109 HDEBUG(D_host_lookup)
2110   {
2111   host_item *h;
2112   if (fully_qualified_name != NULL)
2113     debug_printf("fully qualified name = %s\n", *fully_qualified_name);
2114   debug_printf("%s looked up these IP addresses:\n",
2115     #if HAVE_IPV6
2116       #if HAVE_GETIPNODEBYNAME
2117       "getipnodebyname"
2118       #else
2119       "gethostbyname2"
2120       #endif
2121     #else
2122     "gethostbyname"
2123     #endif
2124     );
2125   for (h = host; h != last->next; h = h->next)
2126     debug_printf("  name=%s address=%s\n", h->name,
2127       (h->address == NULL)? US"<null>" : h->address);
2128   }
2129
2130 /* Return the found status. */
2131
2132 return yield;
2133
2134 /* Handle the case when there is a temporary error. If the name matches
2135 dns_again_means_nonexist, return permanent rather than temporary failure. */
2136
2137 RETURN_AGAIN:
2138   {
2139   #ifndef STAND_ALONE
2140   int rc;
2141   uschar *save = deliver_domain;
2142   deliver_domain = host->name;  /* set $domain */
2143   rc = match_isinlist(host->name, &dns_again_means_nonexist, 0, NULL, NULL,
2144     MCL_DOMAIN, TRUE, NULL);
2145   deliver_domain = save;
2146   if (rc == OK)
2147     {
2148     DEBUG(D_host_lookup) debug_printf("%s is in dns_again_means_nonexist: "
2149       "returning HOST_FIND_FAILED\n", host->name);
2150     return HOST_FIND_FAILED;
2151     }
2152   #endif
2153   return HOST_FIND_AGAIN;
2154   }
2155 }
2156
2157
2158
2159 /*************************************************
2160 *        Fill in a host address from the DNS     *
2161 *************************************************/
2162
2163 /* Given a host item, with its name, port and mx fields set, and its address
2164 field set to NULL, fill in its IP address from the DNS. If it is multi-homed,
2165 create additional host items for the additional addresses, copying all the
2166 other fields, and randomizing the order.
2167
2168 On IPv6 systems, A6 records are sought first (but only if support for A6 is
2169 configured - they may never become mainstream), then AAAA records are sought,
2170 and finally A records are sought as well.
2171
2172 The host name may be changed if the DNS returns a different name - e.g. fully
2173 qualified or changed via CNAME. If fully_qualified_name is not NULL, dns_lookup
2174 ensures that it points to the fully qualified name. However, this is the fully
2175 qualified version of the original name; if a CNAME is involved, the actual
2176 canonical host name may be different again, and so we get it directly from the
2177 relevant RR. Note that we do NOT change the mx field of the host item in this
2178 function as it may be called to set the addresses of hosts taken from MX
2179 records.
2180
2181 Arguments:
2182   host                  points to the host item we're filling in
2183   lastptr               points to pointer to last host item in a chain of
2184                           host items (may be updated if host is last and gets
2185                           extended because multihomed)
2186   ignore_target_hosts   list of hosts to ignore
2187   allow_ip              if TRUE, recognize an IP address and return it
2188   fully_qualified_name  if not NULL, return fully qualified name here if
2189                           the contents are different (i.e. it must be preset
2190                           to something)
2191
2192 Returns:       HOST_FIND_FAILED     couldn't find A record
2193                HOST_FIND_AGAIN      try again later
2194                HOST_FOUND           found AAAA and/or A record(s)
2195                HOST_IGNORED         found, but all IPs ignored
2196 */
2197
2198 static int
2199 set_address_from_dns(host_item *host, host_item **lastptr,
2200   uschar *ignore_target_hosts, BOOL allow_ip, uschar **fully_qualified_name)
2201 {
2202 dns_record *rr;
2203 host_item *thishostlast = NULL;    /* Indicates not yet filled in anything */
2204 BOOL v6_find_again = FALSE;
2205 int i;
2206
2207 /* If allow_ip is set, a name which is an IP address returns that value
2208 as its address. This is used for MX records when allow_mx_to_ip is set, for
2209 those sites that feel they have to flaunt the RFC rules. */
2210
2211 if (allow_ip && string_is_ip_address(host->name, NULL) != 0)
2212   {
2213   #ifndef STAND_ALONE
2214   if (ignore_target_hosts != NULL &&
2215         verify_check_this_host(&ignore_target_hosts, NULL, host->name,
2216         host->name, NULL) == OK)
2217     return HOST_IGNORED;
2218   #endif
2219
2220   host->address = host->name;
2221   return HOST_FOUND;
2222   }
2223
2224 /* On an IPv6 system, unless IPv6 is disabled, go round the loop up to three
2225 times, looking for A6 and AAAA records the first two times. However, unless
2226 doing standalone testing, we force an IPv4 lookup if the domain matches
2227 dns_ipv4_lookup is set. Since A6 records look like being abandoned, support
2228 them only if explicitly configured to do so. On an IPv4 system, go round the
2229 loop once only, looking only for A records. */
2230
2231 #if HAVE_IPV6
2232   #ifndef STAND_ALONE
2233     if (disable_ipv6 || (dns_ipv4_lookup != NULL &&
2234         match_isinlist(host->name, &dns_ipv4_lookup, 0, NULL, NULL, MCL_DOMAIN,
2235         TRUE, NULL) == OK))
2236       i = 0;    /* look up A records only */
2237     else
2238   #endif        /* STAND_ALONE */
2239
2240   #ifdef SUPPORT_A6
2241   i = 2;        /* look up A6 and AAAA and A records */
2242   #else
2243   i = 1;        /* look up AAAA and A records */
2244   #endif        /* SUPPORT_A6 */
2245
2246 /* The IPv4 world */
2247
2248 #else           /* HAVE_IPV6 */
2249   i = 0;        /* look up A records only */
2250 #endif          /* HAVE_IPV6 */
2251
2252 for (; i >= 0; i--)
2253   {
2254   static int types[] = { T_A, T_AAAA, T_A6 };
2255   int type = types[i];
2256   int randoffset = (i == 0)? 500 : 0;  /* Ensures v6 sorts before v4 */
2257   dns_answer dnsa;
2258   dns_scan dnss;
2259
2260   int rc = dns_lookup(&dnsa, host->name, type, fully_qualified_name);
2261
2262   /* We want to return HOST_FIND_AGAIN if one of the A, A6, or AAAA lookups
2263   fails or times out, but not if another one succeeds. (In the early
2264   IPv6 days there are name servers that always fail on AAAA, but are happy
2265   to give out an A record. We want to proceed with that A record.) */
2266
2267   if (rc != DNS_SUCCEED)
2268     {
2269     if (i == 0)  /* Just tried for an A record, i.e. end of loop */
2270       {
2271       if (host->address != NULL) return HOST_FOUND;  /* A6 or AAAA was found */
2272       if (rc == DNS_AGAIN || rc == DNS_FAIL || v6_find_again)
2273         return HOST_FIND_AGAIN;
2274       return HOST_FIND_FAILED;    /* DNS_NOMATCH or DNS_NODATA */
2275       }
2276
2277     /* Tried for an A6 or AAAA record: remember if this was a temporary
2278     error, and look for the next record type. */
2279
2280     if (rc != DNS_NOMATCH && rc != DNS_NODATA) v6_find_again = TRUE;
2281     continue;
2282     }
2283
2284   /* Lookup succeeded: fill in the given host item with the first non-ignored
2285   address found; create additional items for any others. A single A6 record
2286   may generate more than one address. */
2287
2288   for (rr = dns_next_rr(&dnsa, &dnss, RESET_ANSWERS);
2289        rr != NULL;
2290        rr = dns_next_rr(&dnsa, &dnss, RESET_NEXT))
2291     {
2292     if (rr->type == type)
2293       {
2294       /* dns_address *da = dns_address_from_rr(&dnsa, rr); */
2295
2296       dns_address *da;
2297       da = dns_address_from_rr(&dnsa, rr);
2298
2299       DEBUG(D_host_lookup)
2300         {
2301         if (da == NULL)
2302           debug_printf("no addresses extracted from A6 RR for %s\n",
2303             host->name);
2304         }
2305
2306       /* This loop runs only once for A and AAAA records, but may run
2307       several times for an A6 record that generated multiple addresses. */
2308
2309       for (; da != NULL; da = da->next)
2310         {
2311         #ifndef STAND_ALONE
2312         if (ignore_target_hosts != NULL &&
2313               verify_check_this_host(&ignore_target_hosts, NULL,
2314                 host->name, da->address, NULL) == OK)
2315           {
2316           DEBUG(D_host_lookup)
2317             debug_printf("ignored host %s [%s]\n", host->name, da->address);
2318           continue;
2319           }
2320         #endif
2321
2322         /* If this is the first address, stick it in the given host block,
2323         and change the name if the returned RR has a different name. */
2324
2325         if (thishostlast == NULL)
2326           {
2327           if (strcmpic(host->name, rr->name) != 0)
2328             host->name = string_copy_dnsdomain(rr->name);
2329           host->address = da->address;
2330           host->sort_key = host->mx * 1000 + random_number(500) + randoffset;
2331           host->status = hstatus_unknown;
2332           host->why = hwhy_unknown;
2333           thishostlast = host;
2334           }
2335
2336         /* Not the first address. Check for, and ignore, duplicates. Then
2337         insert in the chain at a random point. */
2338
2339         else
2340           {
2341           int new_sort_key;
2342           host_item *next;
2343
2344           /* End of our local chain is specified by "thishostlast". */
2345
2346           for (next = host;; next = next->next)
2347             {
2348             if (Ustrcmp(CS da->address, next->address) == 0) break;
2349             if (next == thishostlast) { next = NULL; break; }
2350             }
2351           if (next != NULL) continue;  /* With loop for next address */
2352
2353           /* Not a duplicate */
2354
2355           new_sort_key = host->mx * 1000 + random_number(500) + randoffset;
2356           next = store_get(sizeof(host_item));
2357
2358           /* New address goes first: insert the new block after the first one
2359           (so as not to disturb the original pointer) but put the new address
2360           in the original block. */
2361
2362           if (new_sort_key < host->sort_key)
2363             {
2364             *next = *host;                                  /* Copies port */
2365             host->next = next;
2366             host->address = da->address;
2367             host->sort_key = new_sort_key;
2368             if (thishostlast == host) thishostlast = next;  /* Local last */
2369             if (*lastptr == host) *lastptr = next;          /* Global last */
2370             }
2371
2372           /* Otherwise scan down the addresses for this host to find the
2373           one to insert after. */
2374
2375           else
2376             {
2377             host_item *h = host;
2378             while (h != thishostlast)
2379               {
2380               if (new_sort_key < h->next->sort_key) break;
2381               h = h->next;
2382               }
2383             *next = *h;                                 /* Copies port */
2384             h->next = next;
2385             next->address = da->address;
2386             next->sort_key = new_sort_key;
2387             if (h == thishostlast) thishostlast = next; /* Local last */
2388             if (h == *lastptr) *lastptr = next;         /* Global last */
2389             }
2390           }
2391         }
2392       }
2393     }
2394   }
2395
2396 /* Control gets here only if the third lookup (the A record) succeeded.
2397 However, the address may not be filled in if it was ignored. */
2398
2399 return (host->address == NULL)? HOST_IGNORED : HOST_FOUND;
2400 }
2401
2402
2403
2404
2405 /*************************************************
2406 *    Find IP addresses and host names via DNS    *
2407 *************************************************/
2408
2409 /* The input is a host_item structure with the name field filled in and the
2410 address field set to NULL. This may be in a chain of other host items. The
2411 lookup may result in more than one IP address, in which case we must created
2412 new host blocks for the additional addresses, and insert them into the chain.
2413 The original name may not be fully qualified. Use the fully_qualified_name
2414 argument to return the official name, as returned by the resolver.
2415
2416 Arguments:
2417   host                  point to initial host item
2418   ignore_target_hosts   a list of hosts to ignore
2419   whichrrs              flags indicating which RRs to look for:
2420                           HOST_FIND_BY_SRV  => look for SRV
2421                           HOST_FIND_BY_MX   => look for MX
2422                           HOST_FIND_BY_A    => look for A or AAAA
2423                         also flags indicating how the lookup is done
2424                           HOST_FIND_QUALIFY_SINGLE   ) passed to the
2425                           HOST_FIND_SEARCH_PARENTS   )   resolver
2426   srv_service           when SRV used, the service name
2427   srv_fail_domains      DNS errors for these domains => assume nonexist
2428   mx_fail_domains       DNS errors for these domains => assume nonexist
2429   fully_qualified_name  if not NULL, return fully-qualified name
2430   removed               set TRUE if local host was removed from the list
2431
2432 Returns:                HOST_FIND_FAILED  Failed to find the host or domain;
2433                                           if there was a syntax error,
2434                                           host_find_failed_syntax is set.
2435                         HOST_FIND_AGAIN   Could not resolve at this time
2436                         HOST_FOUND        Host found
2437                         HOST_FOUND_LOCAL  The lowest MX record points to this
2438                                           machine, if MX records were found, or
2439                                           an A record that was found contains
2440                                           an address of the local host
2441 */
2442
2443 int
2444 host_find_bydns(host_item *host, uschar *ignore_target_hosts, int whichrrs,
2445   uschar *srv_service, uschar *srv_fail_domains, uschar *mx_fail_domains,
2446   uschar **fully_qualified_name, BOOL *removed)
2447 {
2448 host_item *h, *last;
2449 dns_record *rr;
2450 int rc = DNS_FAIL;
2451 int ind_type = 0;
2452 int yield;
2453 dns_answer dnsa;
2454 dns_scan dnss;
2455
2456 /* Set the default fully qualified name to the incoming name, initialize the
2457 resolver if necessary, set up the relevant options, and initialize the flag
2458 that gets set for DNS syntax check errors. */
2459
2460 if (fully_qualified_name != NULL) *fully_qualified_name = host->name;
2461 dns_init((whichrrs & HOST_FIND_QUALIFY_SINGLE) != 0,
2462          (whichrrs & HOST_FIND_SEARCH_PARENTS) != 0);
2463 host_find_failed_syntax = FALSE;
2464
2465 /* First, if requested, look for SRV records. The service name is given; we
2466 assume TCP progocol. DNS domain names are constrained to a maximum of 256
2467 characters, so the code below should be safe. */
2468
2469 if ((whichrrs & HOST_FIND_BY_SRV) != 0)
2470   {
2471   uschar buffer[300];
2472   uschar *temp_fully_qualified_name = buffer;
2473   int prefix_length;
2474
2475   (void)sprintf(CS buffer, "_%s._tcp.%n%.256s", srv_service, &prefix_length,
2476     host->name);
2477   ind_type = T_SRV;
2478
2479   /* Search for SRV records. If the fully qualified name is different to
2480   the input name, pass back the new original domain, without the prepended
2481   magic. */
2482
2483   rc = dns_lookup(&dnsa, buffer, ind_type, &temp_fully_qualified_name);
2484   if (temp_fully_qualified_name != buffer && fully_qualified_name != NULL)
2485     *fully_qualified_name = temp_fully_qualified_name + prefix_length;
2486
2487   /* On DNS failures, we give the "try again" error unless the domain is
2488   listed as one for which we continue. */
2489
2490   if (rc == DNS_FAIL || rc == DNS_AGAIN)
2491     {
2492     #ifndef STAND_ALONE
2493     if (match_isinlist(host->name, &srv_fail_domains, 0, NULL, NULL, MCL_DOMAIN,
2494         TRUE, NULL) != OK)
2495     #endif
2496       return HOST_FIND_AGAIN;
2497     DEBUG(D_host_lookup) debug_printf("DNS_%s treated as DNS_NODATA "
2498       "(domain in srv_fail_domains)\n", (rc == DNS_FAIL)? "FAIL":"AGAIN");
2499     }
2500   }
2501
2502 /* If we did not find any SRV records, search the DNS for MX records, if
2503 requested to do so. If the result is DNS_NOMATCH, it means there is no such
2504 domain, and there's no point in going on to look for address records with the
2505 same domain. The result will be DNS_NODATA if the domain exists but has no MX
2506 records. On DNS failures, we give the "try again" error unless the domain is
2507 listed as one for which we continue. */
2508
2509 if (rc != DNS_SUCCEED && (whichrrs & HOST_FIND_BY_MX) != 0)
2510   {
2511   ind_type = T_MX;
2512   rc = dns_lookup(&dnsa, host->name, ind_type, fully_qualified_name);
2513   if (rc == DNS_NOMATCH) return HOST_FIND_FAILED;
2514   if (rc == DNS_FAIL || rc == DNS_AGAIN)
2515     {
2516     #ifndef STAND_ALONE
2517     if (match_isinlist(host->name, &mx_fail_domains, 0, NULL, NULL, MCL_DOMAIN,
2518         TRUE, NULL) != OK)
2519     #endif
2520       return HOST_FIND_AGAIN;
2521     DEBUG(D_host_lookup) debug_printf("DNS_%s treated as DNS_NODATA "
2522       "(domain in mx_fail_domains)\n", (rc == DNS_FAIL)? "FAIL":"AGAIN");
2523     }
2524   }
2525
2526 /* If we haven't found anything yet, and we are requested to do so, try for an
2527 A or AAAA record. If we find it (or them) check to see that it isn't the local
2528 host. */
2529
2530 if (rc != DNS_SUCCEED)
2531   {
2532   if ((whichrrs & HOST_FIND_BY_A) == 0)
2533     {
2534     DEBUG(D_host_lookup) debug_printf("Address records are not being sought\n");
2535     return HOST_FIND_FAILED;
2536     }
2537
2538   last = host;        /* End of local chainlet */
2539   host->mx = MX_NONE;
2540   host->port = PORT_NONE;
2541   rc = set_address_from_dns(host, &last, ignore_target_hosts, FALSE,
2542     fully_qualified_name);
2543
2544   /* If one or more address records have been found, check that none of them
2545   are local. Since we know the host items all have their IP addresses
2546   inserted, host_scan_for_local_hosts() can only return HOST_FOUND or
2547   HOST_FOUND_LOCAL. We do not need to scan for duplicate IP addresses here,
2548   because set_address_from_dns() removes them. */
2549
2550   if (rc == HOST_FOUND)
2551     rc = host_scan_for_local_hosts(host, &last, removed);
2552   else
2553     if (rc == HOST_IGNORED) rc = HOST_FIND_FAILED;  /* No special action */
2554
2555   DEBUG(D_host_lookup)
2556     {
2557     host_item *h;
2558     if (host->address != NULL)
2559       {
2560       if (fully_qualified_name != NULL)
2561         debug_printf("fully qualified name = %s\n", *fully_qualified_name);
2562       for (h = host; h != last->next; h = h->next)
2563         debug_printf("%s %s mx=%d sort=%d %s\n", h->name,
2564           (h->address == NULL)? US"<null>" : h->address, h->mx, h->sort_key,
2565           (h->status >= hstatus_unusable)? US"*" : US"");
2566       }
2567     }
2568
2569   return rc;
2570   }
2571
2572 /* We have found one or more MX or SRV records. Sort them according to
2573 precedence. Put the data for the first one into the existing host block, and
2574 insert new host_item blocks into the chain for the remainder. For equal
2575 precedences one is supposed to randomize the order. To make this happen, the
2576 sorting is actually done on the MX value * 1000 + a random number. This is put
2577 into a host field called sort_key.
2578
2579 In the case of hosts with both IPv6 and IPv4 addresses, we want to choose the
2580 IPv6 address in preference. At this stage, we don't know what kind of address
2581 the host has. We choose a random number < 500; if later we find an A record
2582 first, we add 500 to the random number. Then for any other address records, we
2583 use random numbers in the range 0-499 for AAAA records and 500-999 for A
2584 records.
2585
2586 At this point we remove any duplicates that point to the same host, retaining
2587 only the one with the lowest precedence. We cannot yet check for precedence
2588 greater than that of the local host, because that test cannot be properly done
2589 until the addresses have been found - an MX record may point to a name for this
2590 host which is not the primary hostname. */
2591
2592 last = NULL;    /* Indicates that not even the first item is filled yet */
2593
2594 for (rr = dns_next_rr(&dnsa, &dnss, RESET_ANSWERS);
2595      rr != NULL;
2596      rr = dns_next_rr(&dnsa, &dnss, RESET_NEXT))
2597   {
2598   int precedence;
2599   int weight = 0;        /* For SRV records */
2600   int port = PORT_NONE;
2601   uschar *s;             /* MUST be unsigned for GETSHORT */
2602   uschar data[256];
2603
2604   if (rr->type != ind_type) continue;
2605   s = rr->data;
2606   GETSHORT(precedence, s);      /* Pointer s is advanced */
2607
2608   /* For MX records, we use a random "weight" which causes multiple records of
2609   the same precedence to sort randomly. */
2610
2611   if (ind_type == T_MX)
2612     {
2613     weight = random_number(500);
2614     }
2615
2616   /* SRV records are specified with a port and a weight. The weight is used
2617   in a special algorithm. However, to start with, we just use it to order the
2618   records of equal priority (precedence). */
2619
2620   else
2621     {
2622     GETSHORT(weight, s);
2623     GETSHORT(port, s);
2624     }
2625
2626   /* Get the name of the host pointed to. */
2627
2628   (void)dn_expand(dnsa.answer, dnsa.answer + dnsa.answerlen, s,
2629     (DN_EXPAND_ARG4_TYPE)data, sizeof(data));
2630
2631   /* Check that we haven't already got this host on the chain; if we have,
2632   keep only the lower precedence. This situation shouldn't occur, but you
2633   never know what junk might get into the DNS (and this case has been seen on
2634   more than one occasion). */
2635
2636   if (last != NULL)       /* This is not the first record */
2637     {
2638     host_item *prev = NULL;
2639
2640     for (h = host; h != last->next; prev = h, h = h->next)
2641       {
2642       if (strcmpic(h->name, data) == 0)
2643         {
2644         DEBUG(D_host_lookup)
2645           debug_printf("discarded duplicate host %s (MX=%d)\n", data,
2646             (precedence > h->mx)? precedence : h->mx);
2647         if (precedence >= h->mx) goto NEXT_MX_RR; /* Skip greater precedence */
2648         if (h == host)                            /* Override first item */
2649           {
2650           h->mx = precedence;
2651           host->sort_key = precedence * 1000 + weight;
2652           goto NEXT_MX_RR;
2653           }
2654
2655         /* Unwanted host item is not the first in the chain, so we can get
2656         get rid of it by cutting it out. */
2657
2658         prev->next = h->next;
2659         if (h == last) last = prev;
2660         break;
2661         }
2662       }
2663     }
2664
2665   /* If this is the first MX or SRV record, put the data into the existing host
2666   block. Otherwise, add a new block in the correct place; if it has to be
2667   before the first block, copy the first block's data to a new second block. */
2668
2669   if (last == NULL)
2670     {
2671     host->name = string_copy_dnsdomain(data);
2672     host->address = NULL;
2673     host->port = port;
2674     host->mx = precedence;
2675     host->sort_key = precedence * 1000 + weight;
2676     host->status = hstatus_unknown;
2677     host->why = hwhy_unknown;
2678     last = host;
2679     }
2680
2681   /* Make a new host item and seek the correct insertion place */
2682
2683   else
2684     {
2685     int sort_key = precedence * 1000 + weight;
2686     host_item *next = store_get(sizeof(host_item));
2687     next->name = string_copy_dnsdomain(data);
2688     next->address = NULL;
2689     next->port = port;
2690     next->mx = precedence;
2691     next->sort_key = sort_key;
2692     next->status = hstatus_unknown;
2693     next->why = hwhy_unknown;
2694     next->last_try = 0;
2695
2696     /* Handle the case when we have to insert before the first item. */
2697
2698     if (sort_key < host->sort_key)
2699       {
2700       host_item htemp;
2701       htemp = *host;
2702       *host = *next;
2703       *next = htemp;
2704       host->next = next;
2705       if (last == host) last = next;
2706       }
2707
2708     /* Else scan down the items we have inserted as part of this exercise;
2709     don't go further. */
2710
2711     else
2712       {
2713       for (h = host; h != last; h = h->next)
2714         {
2715         if (sort_key < h->next->sort_key)
2716           {
2717           next->next = h->next;
2718           h->next = next;
2719           break;
2720           }
2721         }
2722
2723       /* Join on after the last host item that's part of this
2724       processing if we haven't stopped sooner. */
2725
2726       if (h == last)
2727         {
2728         next->next = last->next;
2729         last->next = next;
2730         last = next;
2731         }
2732       }
2733     }
2734
2735   NEXT_MX_RR: continue;
2736   }
2737
2738 /* If the list of hosts was obtained from SRV records, there are two things to
2739 do. First, if there is only one host, and it's name is ".", it means there is
2740 no SMTP service at this domain. Otherwise, we have to sort the hosts of equal
2741 priority according to their weights, using an algorithm that is defined in RFC
2742 2782. The hosts are currently sorted by priority and weight. For each priority
2743 group we have to pick off one host and put it first, and then repeat for any
2744 remaining in the same priority group. */
2745
2746 if (ind_type == T_SRV)
2747   {
2748   host_item **pptr;
2749
2750   if (host == last && host->name[0] == 0)
2751     {
2752     DEBUG(D_host_lookup) debug_printf("the single SRV record is \".\"\n");
2753     return HOST_FIND_FAILED;
2754     }
2755
2756   DEBUG(D_host_lookup)
2757     {
2758     debug_printf("original ordering of hosts from SRV records:\n");
2759     for (h = host; h != last->next; h = h->next)
2760       debug_printf("  %s P=%d W=%d\n", h->name, h->mx, h->sort_key % 1000);
2761     }
2762
2763   for (pptr = &host, h = host; h != last; pptr = &(h->next), h = h->next)
2764     {
2765     int sum = 0;
2766     host_item *hh;
2767
2768     /* Find the last following host that has the same precedence. At the same
2769     time, compute the sum of the weights and the running totals. These can be
2770     stored in the sort_key field. */
2771
2772     for (hh = h; hh != last; hh = hh->next)
2773       {
2774       int weight = hh->sort_key % 1000;   /* was precedence * 1000 + weight */
2775       sum += weight;
2776       hh->sort_key = sum;
2777       if (hh->mx != hh->next->mx) break;
2778       }
2779
2780     /* If there's more than one host at this precedence (priority), we need to
2781     pick one to go first. */
2782
2783     if (hh != h)
2784       {
2785       host_item *hhh;
2786       host_item **ppptr;
2787       int randomizer = random_number(sum + 1);
2788
2789       for (ppptr = pptr, hhh = h;
2790            hhh != hh;
2791            ppptr = &(hhh->next), hhh = hhh->next)
2792         {
2793         if (hhh->sort_key >= randomizer) break;
2794         }
2795
2796       /* hhh now points to the host that should go first; ppptr points to the
2797       place that points to it. Unfortunately, if the start of the minilist is
2798       the start of the entire list, we can't just swap the items over, because
2799       we must not change the value of host, since it is passed in from outside.
2800       One day, this could perhaps be changed.
2801
2802       The special case is fudged by putting the new item *second* in the chain,
2803       and then transferring the data between the first and second items. We
2804       can't just swap the first and the chosen item, because that would mean
2805       that an item with zero weight might no longer be first. */
2806
2807       if (hhh != h)
2808         {
2809         *ppptr = hhh->next;          /* Cuts it out of the chain */
2810
2811         if (h == host)
2812           {
2813           host_item temp = *h;
2814           *h = *hhh;
2815           *hhh = temp;
2816           hhh->next = temp.next;
2817           h->next = hhh;
2818           }
2819
2820         else
2821           {
2822           hhh->next = h;               /* The rest of the chain follows it */
2823           *pptr = hhh;                 /* It takes the place of h */
2824           h = hhh;                     /* It's now the start of this minilist */
2825           }
2826         }
2827       }
2828
2829     /* A host has been chosen to be first at this priority and h now points
2830     to this host. There may be others at the same priority, or others at a
2831     different priority. Before we leave this host, we need to put back a sort
2832     key of the traditional MX kind, in case this host is multihomed, because
2833     the sort key is used for ordering the multiple IP addresses. We do not need
2834     to ensure that these new sort keys actually reflect the order of the hosts,
2835     however. */
2836
2837     h->sort_key = h->mx * 1000 + random_number(500);
2838     }   /* Move on to the next host */
2839   }
2840
2841 /* Now we have to find IP addresses for all the hosts. We have ensured above
2842 that the names in all the host items are unique. Before release 4.61 we used to
2843 process records from the additional section in the DNS packet that returned the
2844 MX or SRV records. However, a DNS name server is free to drop any resource
2845 records from the additional section. In theory, this has always been a
2846 potential problem, but it is exacerbated by the advent of IPv6. If a host had
2847 several IPv4 addresses and some were not in the additional section, at least
2848 Exim would try the others. However, if a host had both IPv4 and IPv6 addresses
2849 and all the IPv4 (say) addresses were absent, Exim would try only for a IPv6
2850 connection, and never try an IPv4 address. When there was only IPv4
2851 connectivity, this was a disaster that did in practice occur.
2852
2853 So, from release 4.61 onwards, we always search for A and AAAA records
2854 explicitly. The names shouldn't point to CNAMES, but we use the general lookup
2855 function that handles them, just in case. If any lookup gives a soft error,
2856 change the default yield.
2857
2858 For these DNS lookups, we must disable qualify_single and search_parents;
2859 otherwise invalid host names obtained from MX or SRV records can cause trouble
2860 if they happen to match something local. */
2861
2862 yield = HOST_FIND_FAILED;    /* Default yield */
2863 dns_init(FALSE, FALSE);      /* Disable qualify_single and search_parents */
2864
2865 for (h = host; h != last->next; h = h->next)
2866   {
2867   if (h->address != NULL) continue;  /* Inserted by a multihomed host */
2868   rc = set_address_from_dns(h, &last, ignore_target_hosts, allow_mx_to_ip, NULL);
2869   if (rc != HOST_FOUND)
2870     {
2871     h->status = hstatus_unusable;
2872     if (rc == HOST_FIND_AGAIN)
2873       {
2874       yield = rc;
2875       h->why = hwhy_deferred;
2876       }
2877     else
2878       h->why = (rc == HOST_IGNORED)? hwhy_ignored : hwhy_failed;
2879     }
2880   }
2881
2882 /* Scan the list for any hosts that are marked unusable because they have
2883 been explicitly ignored, and remove them from the list, as if they did not
2884 exist. If we end up with just a single, ignored host, flatten its fields as if
2885 nothing was found. */
2886
2887 if (ignore_target_hosts != NULL)
2888   {
2889   host_item *prev = NULL;
2890   for (h = host; h != last->next; h = h->next)
2891     {
2892     REDO:
2893     if (h->why != hwhy_ignored)        /* Non ignored host, just continue */
2894       prev = h;
2895     else if (prev == NULL)             /* First host is ignored */
2896       {
2897       if (h != last)                   /* First is not last */
2898         {
2899         if (h->next == last) last = h; /* Overwrite it with next */
2900         *h = *(h->next);               /* and reprocess it. */
2901         goto REDO;                     /* C should have redo, like Perl */
2902         }
2903       }
2904     else                               /* Ignored host is not first - */
2905       {                                /*   cut it out */
2906       prev->next = h->next;
2907       if (h == last) last = prev;
2908       }
2909     }
2910
2911   if (host->why == hwhy_ignored) host->address = NULL;
2912   }
2913
2914 /* There is still one complication in the case of IPv6. Although the code above
2915 arranges that IPv6 addresses take precedence over IPv4 addresses for multihomed
2916 hosts, it doesn't do this for addresses that apply to different hosts with the
2917 same MX precedence, because the sorting on MX precedence happens first. So we
2918 have to make another pass to check for this case. We ensure that, within a
2919 single MX preference value, IPv6 addresses come first. This can separate the
2920 addresses of a multihomed host, but that should not matter. */
2921
2922 #if HAVE_IPV6
2923 if (h != last && !disable_ipv6)
2924   {
2925   for (h = host; h != last; h = h->next)
2926     {
2927     host_item temp;
2928     host_item *next = h->next;
2929     if (h->mx != next->mx ||                   /* If next is different MX */
2930         h->address == NULL ||                  /* OR this one is unset */
2931         Ustrchr(h->address, ':') != NULL ||    /* OR this one is IPv6 */
2932         (next->address != NULL &&
2933          Ustrchr(next->address, ':') == NULL)) /* OR next is IPv4 */
2934       continue;                                /* move on to next */
2935     temp = *h;                                 /* otherwise, swap */
2936     temp.next = next->next;
2937     *h = *next;
2938     h->next = next;
2939     *next = temp;
2940     }
2941   }
2942 #endif
2943
2944 /* Remove any duplicate IP addresses and then scan the list of hosts for any
2945 whose IP addresses are on the local host. If any are found, all hosts with the
2946 same or higher MX values are removed. However, if the local host has the lowest
2947 numbered MX, then HOST_FOUND_LOCAL is returned. Otherwise, if at least one host
2948 with an IP address is on the list, HOST_FOUND is returned. Otherwise,
2949 HOST_FIND_FAILED is returned, but in this case do not update the yield, as it
2950 might have been set to HOST_FIND_AGAIN just above here. If not, it will already
2951 be HOST_FIND_FAILED. */
2952
2953 host_remove_duplicates(host, &last);
2954 rc = host_scan_for_local_hosts(host, &last, removed);
2955 if (rc != HOST_FIND_FAILED) yield = rc;
2956
2957 DEBUG(D_host_lookup)
2958   {
2959   if (fully_qualified_name != NULL)
2960     debug_printf("fully qualified name = %s\n", *fully_qualified_name);
2961   debug_printf("host_find_bydns yield = %s (%d); returned hosts:\n",
2962     (yield == HOST_FOUND)? "HOST_FOUND" :
2963     (yield == HOST_FOUND_LOCAL)? "HOST_FOUND_LOCAL" :
2964     (yield == HOST_FIND_AGAIN)? "HOST_FIND_AGAIN" :
2965     (yield == HOST_FIND_FAILED)? "HOST_FIND_FAILED" : "?",
2966     yield);
2967   for (h = host; h != last->next; h = h->next)
2968     {
2969     debug_printf("  %s %s MX=%d ", h->name,
2970       (h->address == NULL)? US"<null>" : h->address, h->mx);
2971     if (h->port != PORT_NONE) debug_printf("port=%d ", h->port);
2972     if (h->status >= hstatus_unusable) debug_printf("*");
2973     debug_printf("\n");
2974     }
2975   }
2976
2977 return yield;
2978 }
2979
2980
2981
2982
2983 /*************************************************
2984 **************************************************
2985 *             Stand-alone test program           *
2986 **************************************************
2987 *************************************************/
2988
2989 #ifdef STAND_ALONE
2990
2991 int main(int argc, char **cargv)
2992 {
2993 host_item h;
2994 int whichrrs = HOST_FIND_BY_MX | HOST_FIND_BY_A;
2995 BOOL byname = FALSE;
2996 BOOL qualify_single = TRUE;
2997 BOOL search_parents = FALSE;
2998 uschar **argv = USS cargv;
2999 uschar buffer[256];
3000
3001 disable_ipv6 = FALSE;
3002 primary_hostname = US"";
3003 store_pool = POOL_MAIN;
3004 debug_selector = D_host_lookup|D_interface;
3005 debug_file = stdout;
3006 debug_fd = fileno(debug_file);
3007
3008 printf("Exim stand-alone host functions test\n");
3009
3010 host_find_interfaces();
3011 debug_selector = D_host_lookup | D_dns;
3012
3013 if (argc > 1) primary_hostname = argv[1];
3014
3015 /* So that debug level changes can be done first */
3016
3017 dns_init(qualify_single, search_parents);
3018
3019 printf("Testing host lookup\n");
3020 printf("> ");
3021 while (Ufgets(buffer, 256, stdin) != NULL)
3022   {
3023   int rc;
3024   int len = Ustrlen(buffer);
3025   uschar *fully_qualified_name;
3026
3027   while (len > 0 && isspace(buffer[len-1])) len--;
3028   buffer[len] = 0;
3029
3030   if (Ustrcmp(buffer, "q") == 0) break;
3031
3032   if (Ustrcmp(buffer, "byname") == 0) byname = TRUE;
3033   else if (Ustrcmp(buffer, "no_byname") == 0) byname = FALSE;
3034   else if (Ustrcmp(buffer, "a_only") == 0) whichrrs = HOST_FIND_BY_A;
3035   else if (Ustrcmp(buffer, "mx_only") == 0) whichrrs = HOST_FIND_BY_MX;
3036   else if (Ustrcmp(buffer, "srv_only") == 0) whichrrs = HOST_FIND_BY_SRV;
3037   else if (Ustrcmp(buffer, "srv+a") == 0)
3038     whichrrs = HOST_FIND_BY_SRV | HOST_FIND_BY_A;
3039   else if (Ustrcmp(buffer, "srv+mx") == 0)
3040     whichrrs = HOST_FIND_BY_SRV | HOST_FIND_BY_MX;
3041   else if (Ustrcmp(buffer, "srv+mx+a") == 0)
3042     whichrrs = HOST_FIND_BY_SRV | HOST_FIND_BY_MX | HOST_FIND_BY_A;
3043   else if (Ustrcmp(buffer, "qualify_single") == 0) qualify_single = TRUE;
3044   else if (Ustrcmp(buffer, "no_qualify_single") == 0) qualify_single = FALSE;
3045   else if (Ustrcmp(buffer, "search_parents") == 0) search_parents = TRUE;
3046   else if (Ustrcmp(buffer, "no_search_parents") == 0) search_parents = FALSE;
3047   else if (Ustrcmp(buffer, "test_harness") == 0)
3048     running_in_test_harness = !running_in_test_harness;
3049   else if (Ustrcmp(buffer, "ipv6") == 0) disable_ipv6 = !disable_ipv6;
3050   else if (Ustrcmp(buffer, "res_debug") == 0)
3051     {
3052     _res.options ^= RES_DEBUG;
3053     }
3054   else if (Ustrncmp(buffer, "retrans", 7) == 0)
3055     {
3056     (void)sscanf(CS(buffer+8), "%d", &dns_retrans);
3057     _res.retrans = dns_retrans;
3058     }
3059   else if (Ustrncmp(buffer, "retry", 5) == 0)
3060     {
3061     (void)sscanf(CS(buffer+6), "%d", &dns_retry);
3062     _res.retry = dns_retry;
3063     }
3064   else
3065     {
3066     int flags = whichrrs;
3067
3068     h.name = buffer;
3069     h.next = NULL;
3070     h.mx = MX_NONE;
3071     h.port = PORT_NONE;
3072     h.status = hstatus_unknown;
3073     h.why = hwhy_unknown;
3074     h.address = NULL;
3075
3076     if (qualify_single) flags |= HOST_FIND_QUALIFY_SINGLE;
3077     if (search_parents) flags |= HOST_FIND_SEARCH_PARENTS;
3078
3079     rc = byname?
3080       host_find_byname(&h, NULL, flags, &fully_qualified_name, TRUE)
3081       :
3082       host_find_bydns(&h, NULL, flags, US"smtp", NULL, NULL,
3083         &fully_qualified_name, NULL);
3084
3085     if (rc == HOST_FIND_FAILED) printf("Failed\n");
3086       else if (rc == HOST_FIND_AGAIN) printf("Again\n");
3087         else if (rc == HOST_FOUND_LOCAL) printf("Local\n");
3088     }
3089
3090   printf("\n> ");
3091   }
3092
3093 printf("Testing host_aton\n");
3094 printf("> ");
3095 while (Ufgets(buffer, 256, stdin) != NULL)
3096   {
3097   int i;
3098   int x[4];
3099   int len = Ustrlen(buffer);
3100
3101   while (len > 0 && isspace(buffer[len-1])) len--;
3102   buffer[len] = 0;
3103
3104   if (Ustrcmp(buffer, "q") == 0) break;
3105
3106   len = host_aton(buffer, x);
3107   printf("length = %d ", len);
3108   for (i = 0; i < len; i++)
3109     {
3110     printf("%04x ", (x[i] >> 16) & 0xffff);
3111     printf("%04x ", x[i] & 0xffff);
3112     }
3113   printf("\n> ");
3114   }
3115
3116 printf("\n");
3117
3118 printf("Testing host_name_lookup\n");
3119 printf("> ");
3120 while (Ufgets(buffer, 256, stdin) != NULL)
3121   {
3122   int len = Ustrlen(buffer);
3123   while (len > 0 && isspace(buffer[len-1])) len--;
3124   buffer[len] = 0;
3125   if (Ustrcmp(buffer, "q") == 0) break;
3126   sender_host_address = buffer;
3127   sender_host_name = NULL;
3128   sender_host_aliases = NULL;
3129   host_lookup_msg = US"";
3130   host_lookup_failed = FALSE;
3131   if (host_name_lookup() == FAIL)  /* Debug causes printing */
3132     printf("Lookup failed:%s\n", host_lookup_msg);
3133   printf("\n> ");
3134   }
3135
3136 printf("\n");
3137
3138 return 0;
3139 }
3140 #endif  /* STAND_ALONE */
3141
3142 /* End of host.c */