src/src/host.c

   1 /* $Cambridge: exim/src/src/host.c,v 1.30 2009/10/16 09:10:40 tom Exp $ */
   2
   3 /*************************************************
   4 *     Exim - an Internet mail transport agent    *
   5 *************************************************/
   6
   7 /* Copyright (c) University of Cambridge 1995 - 2007 */
   8 /* See the file NOTICE for conditions of use and distribution. */
   9
  10 /* Functions for finding hosts, either by gethostbyname(), gethostbyaddr(), or
  11 directly via the DNS. When IPv6 is supported, getipnodebyname() and
  12 getipnodebyaddr() may be used instead of gethostbyname() and gethostbyaddr(),
  13 if the newer functions are available. This module also contains various other
  14 functions concerned with hosts and addresses, and a random number function,
  15 used for randomizing hosts with equal MXs but available for use in other parts
  16 of Exim. */
  17
  18
  19 #include "exim.h"
  20
  21
  22 /* Static variable for preserving the list of interface addresses in case it is
  23 used more than once. */
  24
  25 static ip_address_item *local_interface_data = NULL;
  26
  27
  28 #ifdef USE_INET_NTOA_FIX
  29 /*************************************************
  30 *         Replacement for broken inet_ntoa()     *
  31 *************************************************/
  32
  33 /* On IRIX systems, gcc uses a different structure passing convention to the
  34 native libraries. This causes inet_ntoa() to always yield 0.0.0.0 or
  35 255.255.255.255. To get round this, we provide a private version of the
  36 function here. It is used only if USE_INET_NTOA_FIX is set, which should happen
  37 only when gcc is in use on an IRIX system. Code send to me by J.T. Breitner,
  38 with these comments:
  39
  40   code by Stuart Levy
  41   as seen in comp.sys.sgi.admin
  42
  43 August 2005: Apparently this is also needed for AIX systems; USE_INET_NTOA_FIX
  44 should now be set for them as well.
  45
  46 Arguments:  sa  an in_addr structure
  47 Returns:        pointer to static text string
  48 */
  49
  50 char *
  51 inet_ntoa(struct in_addr sa)
  52 {
  53 static uschar addr[20];
  54 sprintf(addr, "%d.%d.%d.%d",
  55         (US &sa.s_addr)[0],
  56         (US &sa.s_addr)[1],
  57         (US &sa.s_addr)[2],
  58         (US &sa.s_addr)[3]);
  59   return addr;
  60 }
  61 #endif
  62
  63
  64
  65 /*************************************************
  66 *              Random number generator           *
  67 *************************************************/
  68
  69 /* This is a simple pseudo-random number generator. It does not have to be
  70 very good for the uses to which it is put. When running the regression tests,
  71 start with a fixed seed.
  72
  73 If you need better, see pseudo_random_number() which is potentially stronger,
  74 if a crypto library is available, but might end up just calling this instead.
  75
  76 Arguments:
  77   limit:    one more than the largest number required
  78
  79 Returns:    a pseudo-random number in the range 0 to limit-1
  80 */
  81
  82 int
  83 random_number(int limit)
  84 {
  85 if (limit < 1)
  86   return 0;
  87 if (random_seed == 0)
  88   {
  89   if (running_in_test_harness) random_seed = 42; else
  90     {
  91     int p = (int)getpid();
  92     random_seed = (int)time(NULL) ^ ((p << 16) | p);
  93     }
  94   }
  95 random_seed = 1103515245 * random_seed + 12345;
  96 return (unsigned int)(random_seed >> 16) % limit;
  97 }
  98
  99
 100
 101 /*************************************************
 102 *       Replace gethostbyname() when testing     *
 103 *************************************************/
 104
 105 /* This function is called instead of gethostbyname(), gethostbyname2(), or
 106 getipnodebyname() when running in the test harness. It recognizes the name
 107 "manyhome.test.ex" and generates a humungous number of IP addresses. It also
 108 recognizes an unqualified "localhost" and forces it to the appropriate loopback
 109 address. IP addresses are treated as literals. For other names, it uses the DNS
 110 to find the host name. In the test harness, this means it will access only the
 111 fake DNS resolver.
 112
 113 Arguments:
 114   name          the host name or a textual IP address
 115   af            AF_INET or AF_INET6
 116   error_num     where to put an error code:
 117                 HOST_NOT_FOUND/TRY_AGAIN/NO_RECOVERY/NO_DATA
 118
 119 Returns:        a hostent structure or NULL for an error
 120 */
 121
 122 static struct hostent *
 123 host_fake_gethostbyname(uschar *name, int af, int *error_num)
 124 {
 125 #if HAVE_IPV6
 126 int alen = (af == AF_INET)? sizeof(struct in_addr):sizeof(struct in6_addr);
 127 #else
 128 int alen = sizeof(struct in_addr);
 129 #endif
 130
 131 int ipa;
 132 uschar *lname = name;
 133 uschar *adds;
 134 uschar **alist;
 135 struct hostent *yield;
 136 dns_answer dnsa;
 137 dns_scan dnss;
 138 dns_record *rr;
 139
 140 DEBUG(D_host_lookup)
 141   debug_printf("using host_fake_gethostbyname for %s (%s)\n", name,
 142     (af == AF_INET)? "IPv4" : "IPv6");
 143
 144 /* Handle the name that needs a vast number of IP addresses */
 145
 146 if (Ustrcmp(name, "manyhome.test.ex") == 0 && af == AF_INET)
 147   {
 148   int i, j;
 149   yield = store_get(sizeof(struct hostent));
 150   alist = store_get(2049 * sizeof(char *));
 151   adds  = store_get(2048 * alen);
 152   yield->h_name = CS name;
 153   yield->h_aliases = NULL;
 154   yield->h_addrtype = af;
 155   yield->h_length = alen;
 156   yield->h_addr_list = CSS alist;
 157   for (i = 104; i <= 111; i++)
 158     {
 159     for (j = 0; j <= 255; j++)
 160       {
 161       *alist++ = adds;
 162       *adds++ = 10;
 163       *adds++ = 250;
 164       *adds++ = i;
 165       *adds++ = j;
 166       }
 167     }
 168   *alist = NULL;
 169   return yield;
 170   }
 171
 172 /* Handle unqualified "localhost" */
 173
 174 if (Ustrcmp(name, "localhost") == 0)
 175   lname = (af == AF_INET)? US"127.0.0.1" : US"::1";
 176
 177 /* Handle a literal IP address */
 178
 179 ipa = string_is_ip_address(lname, NULL);
 180 if (ipa != 0)
 181   {
 182   if ((ipa == 4 && af == AF_INET) ||
 183       (ipa == 6 && af == AF_INET6))
 184     {
 185     int i, n;
 186     int x[4];
 187     yield = store_get(sizeof(struct hostent));
 188     alist = store_get(2 * sizeof(char *));
 189     adds  = store_get(alen);
 190     yield->h_name = CS name;
 191     yield->h_aliases = NULL;
 192     yield->h_addrtype = af;
 193     yield->h_length = alen;
 194     yield->h_addr_list = CSS alist;
 195     *alist++ = adds;
 196     n = host_aton(lname, x);
 197     for (i = 0; i < n; i++)
 198       {
 199       int y = x[i];
 200       *adds++ = (y >> 24) & 255;
 201       *adds++ = (y >> 16) & 255;
 202       *adds++ = (y >> 8) & 255;
 203       *adds++ = y & 255;
 204       }
 205     *alist = NULL;
 206     }
 207
 208   /* Wrong kind of literal address */
 209
 210   else
 211     {
 212     *error_num = HOST_NOT_FOUND;
 213     return NULL;
 214     }
 215   }
 216
 217 /* Handle a host name */
 218
 219 else
 220   {
 221   int type = (af == AF_INET)? T_A:T_AAAA;
 222   int rc = dns_lookup(&dnsa, lname, type, NULL);
 223   int count = 0;
 224
 225   switch(rc)
 226     {
 227     case DNS_SUCCEED: break;
 228     case DNS_NOMATCH: *error_num = HOST_NOT_FOUND; return NULL;
 229     case DNS_NODATA:  *error_num = NO_DATA; return NULL;
 230     case DNS_AGAIN:   *error_num = TRY_AGAIN; return NULL;
 231     default:
 232     case DNS_FAIL:    *error_num = NO_RECOVERY; return NULL;
 233     }
 234
 235   for (rr = dns_next_rr(&dnsa, &dnss, RESET_ANSWERS);
 236        rr != NULL;
 237        rr = dns_next_rr(&dnsa, &dnss, RESET_NEXT))
 238     {
 239     if (rr->type == type) count++;
 240     }
 241
 242   yield = store_get(sizeof(struct hostent));
 243   alist = store_get((count + 1) * sizeof(char **));
 244   adds  = store_get(count *alen);
 245
 246   yield->h_name = CS name;
 247   yield->h_aliases = NULL;
 248   yield->h_addrtype = af;
 249   yield->h_length = alen;
 250   yield->h_addr_list = CSS alist;
 251
 252   for (rr = dns_next_rr(&dnsa, &dnss, RESET_ANSWERS);
 253        rr != NULL;
 254        rr = dns_next_rr(&dnsa, &dnss, RESET_NEXT))
 255     {
 256     int i, n;
 257     int x[4];
 258     dns_address *da;
 259     if (rr->type != type) continue;
 260     da = dns_address_from_rr(&dnsa, rr);
 261     *alist++ = adds;
 262     n = host_aton(da->address, x);
 263     for (i = 0; i < n; i++)
 264       {
 265       int y = x[i];
 266       *adds++ = (y >> 24) & 255;
 267       *adds++ = (y >> 16) & 255;
 268       *adds++ = (y >> 8) & 255;
 269       *adds++ = y & 255;
 270       }
 271     }
 272   *alist = NULL;
 273   }
 274
 275 return yield;
 276 }
 277
 278
 279
 280 /*************************************************
 281 *       Build chain of host items from list      *
 282 *************************************************/
 283
 284 /* This function builds a chain of host items from a textual list of host
 285 names. It does not do any lookups. If randomize is true, the chain is build in
 286 a randomized order. There may be multiple groups of independently randomized
 287 hosts; they are delimited by a host name consisting of just "+".
 288
 289 Arguments:
 290   anchor      anchor for the chain
 291   list        text list
 292   randomize   TRUE for randomizing
 293
 294 Returns:      nothing
 295 */
 296
 297 void
 298 host_build_hostlist(host_item **anchor, uschar *list, BOOL randomize)
 299 {
 300 int sep = 0;
 301 int fake_mx = MX_NONE;          /* This value is actually -1 */
 302 uschar *name;
 303 uschar buffer[1024];
 304
 305 if (list == NULL) return;
 306 if (randomize) fake_mx--;       /* Start at -2 for randomizing */
 307
 308 *anchor = NULL;
 309
 310 while ((name = string_nextinlist(&list, &sep, buffer, sizeof(buffer))) != NULL)
 311   {
 312   host_item *h;
 313
 314   if (name[0] == '+' && name[1] == 0)   /* "+" delimits a randomized group */
 315     {                                   /* ignore if not randomizing */
 316     if (randomize) fake_mx--;
 317     continue;
 318     }
 319
 320   h = store_get(sizeof(host_item));
 321   h->name = string_copy(name);
 322   h->address = NULL;
 323   h->port = PORT_NONE;
 324   h->mx = fake_mx;
 325   h->sort_key = randomize? (-fake_mx)*1000 + random_number(1000) : 0;
 326   h->status = hstatus_unknown;
 327   h->why = hwhy_unknown;
 328   h->last_try = 0;
 329
 330   if (*anchor == NULL)
 331     {
 332     h->next = NULL;
 333     *anchor = h;
 334     }
 335   else
 336     {
 337     host_item *hh = *anchor;
 338     if (h->sort_key < hh->sort_key)
 339       {
 340       h->next = hh;
 341       *anchor = h;
 342       }
 343     else
 344       {
 345       while (hh->next != NULL && h->sort_key >= (hh->next)->sort_key)
 346         hh = hh->next;
 347       h->next = hh->next;
 348       hh->next = h;
 349       }
 350     }
 351   }
 352 }
 353
 354
 355
 356
 357
 358 /*************************************************
 359 *        Extract port from address string        *
 360 *************************************************/
 361
 362 /* In the spool file, and in the -oMa and -oMi options, a host plus port is
 363 given as an IP address followed by a dot and a port number. This function
 364 decodes this.
 365
 366 An alternative format for the -oMa and -oMi options is [ip address]:port which
 367 is what Exim 4 uses for output, because it seems to becoming commonly used,
 368 whereas the dot form confuses some programs/people. So we recognize that form
 369 too.
 370
 371 Argument:
 372   address    points to the string; if there is a port, the '.' in the string
 373              is overwritten with zero to terminate the address; if the string
 374              is in the [xxx]:ppp format, the address is shifted left and the
 375              brackets are removed
 376
 377 Returns:     0 if there is no port, else the port number. If there's a syntax
 378              error, leave the incoming address alone, and return 0.
 379 */
 380
 381 int
 382 host_address_extract_port(uschar *address)
 383 {
 384 int port = 0;
 385 uschar *endptr;
 386
 387 /* Handle the "bracketed with colon on the end" format */
 388
 389 if (*address == '[')
 390   {
 391   uschar *rb = address + 1;
 392   while (*rb != 0 && *rb != ']') rb++;
 393   if (*rb++ == 0) return 0;        /* Missing ]; leave invalid address */
 394   if (*rb == ':')
 395     {
 396     port = Ustrtol(rb + 1, &endptr, 10);
 397     if (*endptr != 0) return 0;    /* Invalid port; leave invalid address */
 398     }
 399   else if (*rb != 0) return 0;     /* Bad syntax; leave invalid address */
 400   memmove(address, address + 1, rb - address - 2);
 401   rb[-2] = 0;
 402   }
 403
 404 /* Handle the "dot on the end" format */
 405
 406 else
 407   {
 408   int skip = -3;                   /* Skip 3 dots in IPv4 addresses */
 409   address--;
 410   while (*(++address) != 0)
 411     {
 412     int ch = *address;
 413     if (ch == ':') skip = 0;       /* Skip 0 dots in IPv6 addresses */
 414       else if (ch == '.' && skip++ >= 0) break;
 415     }
 416   if (*address == 0) return 0;
 417   port = Ustrtol(address + 1, &endptr, 10);
 418   if (*endptr != 0) return 0;      /* Invalid port; leave invalid address */
 419   *address = 0;
 420   }
 421
 422 return port;
 423 }
 424
 425
 426 /*************************************************
 427 *         Get port from a host item's name       *
 428 *************************************************/
 429
 430 /* This function is called when finding the IP address for a host that is in a
 431 list of hosts explicitly configured, such as in the manualroute router, or in a
 432 fallback hosts list. We see if there is a port specification at the end of the
 433 host name, and if so, remove it. A minimum length of 3 is required for the
 434 original name; nothing shorter is recognized as having a port.
 435
 436 We test for a name ending with a sequence of digits; if preceded by colon we
 437 have a port if the character before the colon is ] and the name starts with [
 438 or if there are no other colons in the name (i.e. it's not an IPv6 address).
 439
 440 Arguments:  pointer to the host item
 441 Returns:    a port number or PORT_NONE
 442 */
 443
 444 int
 445 host_item_get_port(host_item *h)
 446 {
 447 uschar *p;
 448 int port, x;
 449 int len = Ustrlen(h->name);
 450
 451 if (len < 3 || (p = h->name + len - 1, !isdigit(*p))) return PORT_NONE;
 452
 453 /* Extract potential port number */
 454
 455 port = *p-- - '0';
 456 x = 10;
 457
 458 while (p > h->name + 1 && isdigit(*p))
 459   {
 460   port += (*p-- - '0') * x;
 461   x *= 10;
 462   }
 463
 464 /* The smallest value of p at this point is h->name + 1. */
 465
 466 if (*p != ':') return PORT_NONE;
 467
 468 if (p[-1] == ']' && h->name[0] == '[')
 469   h->name = string_copyn(h->name + 1, p - h->name - 2);
 470 else if (Ustrchr(h->name, ':') == p)
 471   h->name = string_copyn(h->name, p - h->name);
 472 else return PORT_NONE;
 473
 474 DEBUG(D_route|D_host_lookup) debug_printf("host=%s port=%d\n", h->name, port);
 475 return port;
 476 }
 477
 478
 479
 480 #ifndef STAND_ALONE    /* Omit when standalone testing */
 481
 482 /*************************************************
 483 *     Build sender_fullhost and sender_rcvhost   *
 484 *************************************************/
 485
 486 /* This function is called when sender_host_name and/or sender_helo_name
 487 have been set. Or might have been set - for a local message read off the spool
 488 they won't be. In that case, do nothing. Otherwise, set up the fullhost string
 489 as follows:
 490
 491 (a) No sender_host_name or sender_helo_name: "[ip address]"
 492 (b) Just sender_host_name: "host_name [ip address]"
 493 (c) Just sender_helo_name: "(helo_name) [ip address]" unless helo is IP
 494             in which case: "[ip address}"
 495 (d) The two are identical: "host_name [ip address]" includes helo = IP
 496 (e) The two are different: "host_name (helo_name) [ip address]"
 497
 498 If log_incoming_port is set, the sending host's port number is added to the IP
 499 address.
 500
 501 This function also builds sender_rcvhost for use in Received: lines, whose
 502 syntax is a bit different. This value also includes the RFC 1413 identity.
 503 There wouldn't be two different variables if I had got all this right in the
 504 first place.
 505
 506 Because this data may survive over more than one incoming SMTP message, it has
 507 to be in permanent store.
 508
 509 Arguments:  none
 510 Returns:    nothing
 511 */
 512
 513 void
 514 host_build_sender_fullhost(void)
 515 {
 516 BOOL show_helo = TRUE;
 517 uschar *address;
 518 int len;
 519 int old_pool = store_pool;
 520
 521 if (sender_host_address == NULL) return;
 522
 523 store_pool = POOL_PERM;
 524
 525 /* Set up address, with or without the port. After discussion, it seems that
 526 the only format that doesn't cause trouble is [aaaa]:pppp. However, we can't
 527 use this directly as the first item for Received: because it ain't an RFC 2822
 528 domain. Sigh. */
 529
 530 address = string_sprintf("[%s]:%d", sender_host_address, sender_host_port);
 531 if ((log_extra_selector & LX_incoming_port) == 0 || sender_host_port <= 0)
 532   *(Ustrrchr(address, ':')) = 0;
 533
 534 /* If there's no EHLO/HELO data, we can't show it. */
 535
 536 if (sender_helo_name == NULL) show_helo = FALSE;
 537
 538 /* If HELO/EHLO was followed by an IP literal, it's messy because of two
 539 features of IPv6. Firstly, there's the "IPv6:" prefix (Exim is liberal and
 540 doesn't require this, for historical reasons). Secondly, IPv6 addresses may not
 541 be given in canonical form, so we have to canonicize them before comparing. As
 542 it happens, the code works for both IPv4 and IPv6. */
 543
 544 else if (sender_helo_name[0] == '[' &&
 545          sender_helo_name[(len=Ustrlen(sender_helo_name))-1] == ']')
 546   {
 547   int offset = 1;
 548   uschar *helo_ip;
 549
 550   if (strncmpic(sender_helo_name + 1, US"IPv6:", 5) == 0) offset += 5;
 551   if (strncmpic(sender_helo_name + 1, US"IPv4:", 5) == 0) offset += 5;
 552
 553   helo_ip = string_copyn(sender_helo_name + offset, len - offset - 1);
 554
 555   if (string_is_ip_address(helo_ip, NULL) != 0)
 556     {
 557     int x[4], y[4];
 558     int sizex, sizey;
 559     uschar ipx[48], ipy[48];    /* large enough for full IPv6 */
 560
 561     sizex = host_aton(helo_ip, x);
 562     sizey = host_aton(sender_host_address, y);
 563
 564     (void)host_nmtoa(sizex, x, -1, ipx, ':');
 565     (void)host_nmtoa(sizey, y, -1, ipy, ':');
 566
 567     if (strcmpic(ipx, ipy) == 0) show_helo = FALSE;
 568     }
 569   }
 570
 571 /* Host name is not verified */
 572
 573 if (sender_host_name == NULL)
 574   {
 575   uschar *portptr = Ustrstr(address, "]:");
 576   int size = 0;
 577   int ptr = 0;
 578   int adlen;    /* Sun compiler doesn't like ++ in initializers */
 579
 580   adlen = (portptr == NULL)? Ustrlen(address) : (++portptr - address);
 581   sender_fullhost = (sender_helo_name == NULL)? address :
 582     string_sprintf("(%s) %s", sender_helo_name, address);
 583
 584   sender_rcvhost = string_cat(NULL, &size, &ptr, address, adlen);
 585
 586   if (sender_ident != NULL || show_helo || portptr != NULL)
 587     {
 588     int firstptr;
 589     sender_rcvhost = string_cat(sender_rcvhost, &size, &ptr, US" (", 2);
 590     firstptr = ptr;
 591
 592     if (portptr != NULL)
 593       sender_rcvhost = string_append(sender_rcvhost, &size, &ptr, 2, US"port=",
 594         portptr + 1);
 595
 596     if (show_helo)
 597       sender_rcvhost = string_append(sender_rcvhost, &size, &ptr, 2,
 598         (firstptr == ptr)? US"helo=" : US" helo=", sender_helo_name);
 599
 600     if (sender_ident != NULL)
 601       sender_rcvhost = string_append(sender_rcvhost, &size, &ptr, 2,
 602         (firstptr == ptr)? US"ident=" : US" ident=", sender_ident);
 603
 604     sender_rcvhost = string_cat(sender_rcvhost, &size, &ptr, US")", 1);
 605     }
 606
 607   sender_rcvhost[ptr] = 0;   /* string_cat() always leaves room */
 608
 609   /* Release store, because string_cat allocated a minimum of 100 bytes that
 610   are rarely completely used. */
 611
 612   store_reset(sender_rcvhost + ptr + 1);
 613   }
 614
 615 /* Host name is known and verified. Unless we've already found that the HELO
 616 data matches the IP address, compare it with the name. */
 617
 618 else
 619   {
 620   if (show_helo && strcmpic(sender_host_name, sender_helo_name) == 0)
 621     show_helo = FALSE;
 622
 623   if (show_helo)
 624     {
 625     sender_fullhost = string_sprintf("%s (%s) %s", sender_host_name,
 626       sender_helo_name, address);
 627     sender_rcvhost = (sender_ident == NULL)?
 628       string_sprintf("%s (%s helo=%s)", sender_host_name,
 629         address, sender_helo_name) :
 630       string_sprintf("%s\n\t(%s helo=%s ident=%s)", sender_host_name,
 631         address, sender_helo_name, sender_ident);
 632     }
 633   else
 634     {
 635     sender_fullhost = string_sprintf("%s %s", sender_host_name, address);
 636     sender_rcvhost = (sender_ident == NULL)?
 637       string_sprintf("%s (%s)", sender_host_name, address) :
 638       string_sprintf("%s (%s ident=%s)", sender_host_name, address,
 639         sender_ident);
 640     }
 641   }
 642
 643 store_pool = old_pool;
 644
 645 DEBUG(D_host_lookup) debug_printf("sender_fullhost = %s\n", sender_fullhost);
 646 DEBUG(D_host_lookup) debug_printf("sender_rcvhost = %s\n", sender_rcvhost);
 647 }
 648
 649
 650
 651 /*************************************************
 652 *          Build host+ident message              *
 653 *************************************************/
 654
 655 /* Used when logging rejections and various ACL and SMTP incidents. The text
 656 return depends on whether sender_fullhost and sender_ident are set or not:
 657
 658   no ident, no host   => U=unknown
 659   no ident, host set  => H=sender_fullhost
 660   ident set, no host  => U=ident
 661   ident set, host set => H=sender_fullhost U=ident
 662
 663 Arguments:
 664   useflag   TRUE if first item to be flagged (H= or U=); if there are two
 665               items, the second is always flagged
 666
 667 Returns:    pointer to a string in big_buffer
 668 */
 669
 670 uschar *
 671 host_and_ident(BOOL useflag)
 672 {
 673 if (sender_fullhost == NULL)
 674   {
 675   (void)string_format(big_buffer, big_buffer_size, "%s%s", useflag? "U=" : "",
 676      (sender_ident == NULL)? US"unknown" : sender_ident);
 677   }
 678 else
 679   {
 680   uschar *flag = useflag? US"H=" : US"";
 681   uschar *iface = US"";
 682   if ((log_extra_selector & LX_incoming_interface) != 0 &&
 683        interface_address != NULL)
 684     iface = string_sprintf(" I=[%s]:%d", interface_address, interface_port);
 685   if (sender_ident == NULL)
 686     (void)string_format(big_buffer, big_buffer_size, "%s%s%s",
 687       flag, sender_fullhost, iface);
 688   else
 689     (void)string_format(big_buffer, big_buffer_size, "%s%s%s U=%s",
 690       flag, sender_fullhost, iface, sender_ident);
 691   }
 692 return big_buffer;
 693 }
 694
 695 #endif   /* STAND_ALONE */
 696
 697
 698
 699
 700 /*************************************************
 701 *         Build list of local interfaces         *
 702 *************************************************/
 703
 704 /* This function interprets the contents of the local_interfaces or
 705 extra_local_interfaces options, and creates an ip_address_item block for each
 706 item on the list. There is no special interpretation of any IP addresses; in
 707 particular, 0.0.0.0 and ::0 are returned without modification. If any address
 708 includes a port, it is set in the block. Otherwise the port value is set to
 709 zero.
 710
 711 Arguments:
 712   list        the list
 713   name        the name of the option being expanded
 714
 715 Returns:      a chain of ip_address_items, each containing to a textual
 716               version of an IP address, and a port number (host order) or
 717               zero if no port was given with the address
 718 */
 719
 720 ip_address_item *
 721 host_build_ifacelist(uschar *list, uschar *name)
 722 {
 723 int sep = 0;
 724 uschar *s;
 725 uschar buffer[64];
 726 ip_address_item *yield = NULL;
 727 ip_address_item *last = NULL;
 728 ip_address_item *next;
 729
 730 while ((s = string_nextinlist(&list, &sep, buffer, sizeof(buffer))) != NULL)
 731   {
 732   int ipv;
 733   int port = host_address_extract_port(s);            /* Leaves just the IP address */
 734   if ((ipv = string_is_ip_address(s, NULL)) == 0)
 735     log_write(0, LOG_MAIN|LOG_PANIC_DIE, "Malformed IP address \"%s\" in %s",
 736       s, name);
 737
 738   /* Skip IPv6 addresses if IPv6 is disabled. */
 739
 740   if (disable_ipv6 && ipv == 6) continue;
 741
 742   /* This use of strcpy() is OK because we have checked that s is a valid IP
 743   address above. The field in the ip_address_item is large enough to hold an
 744   IPv6 address. */
 745
 746   next = store_get(sizeof(ip_address_item));
 747   next->next = NULL;
 748   Ustrcpy(next->address, s);
 749   next->port = port;
 750   next->v6_include_v4 = FALSE;
 751
 752   if (yield == NULL) yield = last = next; else
 753     {
 754     last->next = next;
 755     last = next;
 756     }
 757   }
 758
 759 return yield;
 760 }
 761
 762
 763
 764
 765
 766 /*************************************************
 767 *         Find addresses on local interfaces     *
 768 *************************************************/
 769
 770 /* This function finds the addresses of local IP interfaces. These are used
 771 when testing for routing to the local host. As the function may be called more
 772 than once, the list is preserved in permanent store, pointed to by a static
 773 variable, to save doing the work more than once per process.
 774
 775 The generic list of interfaces is obtained by calling host_build_ifacelist()
 776 for local_interfaces and extra_local_interfaces. This list scanned to remove
 777 duplicates (which may exist with different ports - not relevant here). If
 778 either of the wildcard IP addresses (0.0.0.0 and ::0) are encountered, they are
 779 replaced by the appropriate (IPv4 or IPv6) list of actual local interfaces,
 780 obtained from os_find_running_interfaces().
 781
 782 Arguments:    none
 783 Returns:      a chain of ip_address_items, each containing to a textual
 784               version of an IP address; the port numbers are not relevant
 785 */
 786
 787
 788 /* First, a local subfunction to add an interface to a list in permanent store,
 789 but only if there isn't a previous copy of that address on the list. */
 790
 791 static ip_address_item *
 792 add_unique_interface(ip_address_item *list, ip_address_item *ipa)
 793 {
 794 ip_address_item *ipa2;
 795 for (ipa2 = list; ipa2 != NULL; ipa2 = ipa2->next)
 796   if (Ustrcmp(ipa2->address, ipa->address) == 0) return list;
 797 ipa2 = store_get_perm(sizeof(ip_address_item));
 798 *ipa2 = *ipa;
 799 ipa2->next = list;
 800 return ipa2;
 801 }
 802
 803
 804 /* This is the globally visible function */
 805
 806 ip_address_item *
 807 host_find_interfaces(void)
 808 {
 809 ip_address_item *running_interfaces = NULL;
 810
 811 if (local_interface_data == NULL)
 812   {
 813   void *reset_item = store_get(0);
 814   ip_address_item *dlist = host_build_ifacelist(local_interfaces,
 815     US"local_interfaces");
 816   ip_address_item *xlist = host_build_ifacelist(extra_local_interfaces,
 817     US"extra_local_interfaces");
 818   ip_address_item *ipa;
 819
 820   if (dlist == NULL) dlist = xlist; else
 821     {
 822     for (ipa = dlist; ipa->next != NULL; ipa = ipa->next);
 823     ipa->next = xlist;
 824     }
 825
 826   for (ipa = dlist; ipa != NULL; ipa = ipa->next)
 827     {
 828     if (Ustrcmp(ipa->address, "0.0.0.0") == 0 ||
 829         Ustrcmp(ipa->address, "::0") == 0)
 830       {
 831       ip_address_item *ipa2;
 832       BOOL ipv6 = ipa->address[0] == ':';
 833       if (running_interfaces == NULL)
 834         running_interfaces = os_find_running_interfaces();
 835       for (ipa2 = running_interfaces; ipa2 != NULL; ipa2 = ipa2->next)
 836         {
 837         if ((Ustrchr(ipa2->address, ':') != NULL) == ipv6)
 838           local_interface_data = add_unique_interface(local_interface_data,
 839           ipa2);
 840         }
 841       }
 842     else
 843       {
 844       local_interface_data = add_unique_interface(local_interface_data, ipa);
 845       DEBUG(D_interface)
 846         {
 847         debug_printf("Configured local interface: address=%s", ipa->address);
 848         if (ipa->port != 0) debug_printf(" port=%d", ipa->port);
 849         debug_printf("\n");
 850         }
 851       }
 852     }
 853   store_reset(reset_item);
 854   }
 855
 856 return local_interface_data;
 857 }
 858
 859
 860
 861
 862
 863 /*************************************************
 864 *        Convert network IP address to text      *
 865 *************************************************/
 866
 867 /* Given an IPv4 or IPv6 address in binary, convert it to a text
 868 string and return the result in a piece of new store. The address can
 869 either be given directly, or passed over in a sockaddr structure. Note
 870 that this isn't the converse of host_aton() because of byte ordering
 871 differences. See host_nmtoa() below.
 872
 873 Arguments:
 874   type       if < 0 then arg points to a sockaddr, else
 875              either AF_INET or AF_INET6
 876   arg        points to a sockaddr if type is < 0, or
 877              points to an IPv4 address (32 bits), or
 878              points to an IPv6 address (128 bits),
 879              in both cases, in network byte order
 880   buffer     if NULL, the result is returned in gotten store;
 881              else points to a buffer to hold the answer
 882   portptr    points to where to put the port number, if non NULL; only
 883              used when type < 0
 884
 885 Returns:     pointer to character string
 886 */
 887
 888 uschar *
 889 host_ntoa(int type, const void *arg, uschar *buffer, int *portptr)
 890 {
 891 uschar *yield;
 892
 893 /* The new world. It is annoying that we have to fish out the address from
 894 different places in the block, depending on what kind of address it is. It
 895 is also a pain that inet_ntop() returns a const uschar *, whereas the IPv4
 896 function inet_ntoa() returns just uschar *, and some picky compilers insist
 897 on warning if one assigns a const uschar * to a uschar *. Hence the casts. */
 898
 899 #if HAVE_IPV6
 900 uschar addr_buffer[46];
 901 if (type < 0)
 902   {
 903   int family = ((struct sockaddr *)arg)->sa_family;
 904   if (family == AF_INET6)
 905     {
 906     struct sockaddr_in6 *sk = (struct sockaddr_in6 *)arg;
 907     yield = (uschar *)inet_ntop(family, &(sk->sin6_addr), CS addr_buffer,
 908       sizeof(addr_buffer));
 909     if (portptr != NULL) *portptr = ntohs(sk->sin6_port);
 910     }
 911   else
 912     {
 913     struct sockaddr_in *sk = (struct sockaddr_in *)arg;
 914     yield = (uschar *)inet_ntop(family, &(sk->sin_addr), CS addr_buffer,
 915       sizeof(addr_buffer));
 916     if (portptr != NULL) *portptr = ntohs(sk->sin_port);
 917     }
 918   }
 919 else
 920   {
 921   yield = (uschar *)inet_ntop(type, arg, CS addr_buffer, sizeof(addr_buffer));
 922   }
 923
 924 /* If the result is a mapped IPv4 address, show it in V4 format. */
 925
 926 if (Ustrncmp(yield, "::ffff:", 7) == 0) yield += 7;
 927
 928 #else  /* HAVE_IPV6 */
 929
 930 /* The old world */
 931
 932 if (type < 0)
 933   {
 934   yield = US inet_ntoa(((struct sockaddr_in *)arg)->sin_addr);
 935   if (portptr != NULL) *portptr = ntohs(((struct sockaddr_in *)arg)->sin_port);
 936   }
 937 else
 938   yield = US inet_ntoa(*((struct in_addr *)arg));
 939 #endif
 940
 941 /* If there is no buffer, put the string into some new store. */
 942
 943 if (buffer == NULL) return string_copy(yield);
 944
 945 /* Callers of this function with a non-NULL buffer must ensure that it is
 946 large enough to hold an IPv6 address, namely, at least 46 bytes. That's what
 947 makes this use of strcpy() OK. */
 948
 949 Ustrcpy(buffer, yield);
 950 return buffer;
 951 }
 952
 953
 954
 955
 956 /*************************************************
 957 *         Convert address text to binary         *
 958 *************************************************/
 959
 960 /* Given the textual form of an IP address, convert it to binary in an
 961 array of ints. IPv4 addresses occupy one int; IPv6 addresses occupy 4 ints.
 962 The result has the first byte in the most significant byte of the first int. In
 963 other words, the result is not in network byte order, but in host byte order.
 964 As a result, this is not the converse of host_ntoa(), which expects network
 965 byte order. See host_nmtoa() below.
 966
 967 Arguments:
 968   address    points to the textual address, checked for syntax
 969   bin        points to an array of 4 ints
 970
 971 Returns:     the number of ints used
 972 */
 973
 974 int
 975 host_aton(uschar *address, int *bin)
 976 {
 977 int x[4];
 978 int v4offset = 0;
 979
 980 /* Handle IPv6 address, which may end with an IPv4 address. It may also end
 981 with a "scope", introduced by a percent sign. This code is NOT enclosed in #if
 982 HAVE_IPV6 in order that IPv6 addresses are recognized even if IPv6 is not
 983 supported. */
 984
 985 if (Ustrchr(address, ':') != NULL)
 986   {
 987   uschar *p = address;
 988   uschar *component[8];
 989   BOOL ipv4_ends = FALSE;
 990   int ci = 0;
 991   int nulloffset = 0;
 992   int v6count = 8;
 993   int i;
 994
 995   /* If the address starts with a colon, it will start with two colons.
 996   Just lose the first one, which will leave a null first component. */
 997
 998   if (*p == ':') p++;
 999
1000   /* Split the address into components separated by colons. The input address
1001   is supposed to be checked for syntax. There was a case where this was
1002   overlooked; to guard against that happening again, check here and crash if
1003   there are too many components. */
1004
1005   while (*p != 0 && *p != '%')
1006     {
1007     int len = Ustrcspn(p, ":%");
1008     if (len == 0) nulloffset = ci;
1009     if (ci > 7) log_write(0, LOG_MAIN|LOG_PANIC_DIE,
1010       "Internal error: invalid IPv6 address \"%s\" passed to host_aton()",
1011       address);
1012     component[ci++] = p;
1013     p += len;
1014     if (*p == ':') p++;
1015     }
1016
1017   /* If the final component contains a dot, it is a trailing v4 address.
1018   As the syntax is known to be checked, just set up for a trailing
1019   v4 address and restrict the v6 part to 6 components. */
1020
1021   if (Ustrchr(component[ci-1], '.') != NULL)
1022     {
1023     address = component[--ci];
1024     ipv4_ends = TRUE;
1025     v4offset = 3;
1026     v6count = 6;
1027     }
1028
1029   /* If there are fewer than 6 or 8 components, we have to insert some
1030   more empty ones in the middle. */
1031
1032   if (ci < v6count)
1033     {
1034     int insert_count = v6count - ci;
1035     for (i = v6count-1; i > nulloffset + insert_count; i--)
1036       component[i] = component[i - insert_count];
1037     while (i > nulloffset) component[i--] = US"";
1038     }
1039
1040   /* Now turn the components into binary in pairs and bung them
1041   into the vector of ints. */
1042
1043   for (i = 0; i < v6count; i += 2)
1044     bin[i/2] = (Ustrtol(component[i], NULL, 16) << 16) +
1045       Ustrtol(component[i+1], NULL, 16);
1046
1047   /* If there was no terminating v4 component, we are done. */
1048
1049   if (!ipv4_ends) return 4;
1050   }
1051
1052 /* Handle IPv4 address */
1053
1054 (void)sscanf(CS address, "%d.%d.%d.%d", x, x+1, x+2, x+3);
1055 bin[v4offset] = (x[0] << 24) + (x[1] << 16) + (x[2] << 8) + x[3];
1056 return v4offset+1;
1057 }
1058
1059
1060 /*************************************************
1061 *           Apply mask to an IP address          *
1062 *************************************************/
1063
1064 /* Mask an address held in 1 or 4 ints, with the ms bit in the ms bit of the
1065 first int, etc.
1066
1067 Arguments:
1068   count        the number of ints
1069   binary       points to the ints to be masked
1070   mask         the count of ms bits to leave, or -1 if no masking
1071
1072 Returns:       nothing
1073 */
1074
1075 void
1076 host_mask(int count, int *binary, int mask)
1077 {
1078 int i;
1079 if (mask < 0) mask = 99999;
1080 for (i = 0; i < count; i++)
1081   {
1082   int wordmask;
1083   if (mask == 0) wordmask = 0;
1084   else if (mask < 32)
1085     {
1086     wordmask = (-1) << (32 - mask);
1087     mask = 0;
1088     }
1089   else
1090     {
1091     wordmask = -1;
1092     mask -= 32;
1093     }
1094   binary[i] &= wordmask;
1095   }
1096 }
1097
1098
1099
1100
1101 /*************************************************
1102 *     Convert masked IP address in ints to text  *
1103 *************************************************/
1104
1105 /* We can't use host_ntoa() because it assumes the binary values are in network
1106 byte order, and these are the result of host_aton(), which puts them in ints in
1107 host byte order. Also, we really want IPv6 addresses to be in a canonical
1108 format, so we output them with no abbreviation. In a number of cases we can't
1109 use the normal colon separator in them because it terminates keys in lsearch
1110 files, so we want to use dot instead. There's an argument that specifies what
1111 to use for IPv6 addresses.
1112
1113 Arguments:
1114   count       1 or 4 (number of ints)
1115   binary      points to the ints
1116   mask        mask value; if < 0 don't add to result
1117   buffer      big enough to hold the result
1118   sep         component separator character for IPv6 addresses
1119
1120 Returns:      the number of characters placed in buffer, not counting
1121               the final nul.
1122 */
1123
1124 int
1125 host_nmtoa(int count, int *binary, int mask, uschar *buffer, int sep)
1126 {
1127 int i, j;
1128 uschar *tt = buffer;
1129
1130 if (count == 1)
1131   {
1132   j = binary[0];
1133   for (i = 24; i >= 0; i -= 8)
1134     {
1135     sprintf(CS tt, "%d.", (j >> i) & 255);
1136     while (*tt) tt++;
1137     }
1138   }
1139 else
1140   {
1141   for (i = 0; i < 4; i++)
1142     {
1143     j = binary[i];
1144     sprintf(CS tt, "%04x%c%04x%c", (j >> 16) & 0xffff, sep, j & 0xffff, sep);
1145     while (*tt) tt++;
1146     }
1147   }
1148
1149 tt--;   /* lose final separator */
1150
1151 if (mask < 0)
1152   *tt = 0;
1153 else
1154   {
1155   sprintf(CS tt, "/%d", mask);
1156   while (*tt) tt++;
1157   }
1158
1159 return tt - buffer;
1160 }
1161
1162
1163
1164 /*************************************************
1165 *        Check port for tls_on_connect           *
1166 *************************************************/
1167
1168 /* This function checks whether a given incoming port is configured for tls-
1169 on-connect. It is called from the daemon and from inetd handling. If the global
1170 option tls_on_connect is already set, all ports operate this way. Otherwise, we
1171 check the tls_on_connect_ports option for a list of ports.
1172
1173 Argument:  a port number
1174 Returns:   TRUE or FALSE
1175 */
1176
1177 BOOL
1178 host_is_tls_on_connect_port(int port)
1179 {
1180 int sep = 0;
1181 uschar buffer[32];
1182 uschar *list = tls_on_connect_ports;
1183 uschar *s;
1184
1185 if (tls_on_connect) return TRUE;
1186
1187 while ((s = string_nextinlist(&list, &sep, buffer, sizeof(buffer))) != NULL)
1188   {
1189   uschar *end;
1190   int lport = Ustrtol(s, &end, 10);
1191   if (*end != 0) log_write(0, LOG_MAIN|LOG_PANIC_DIE, "tls_on_connect_ports "
1192     "contains \"%s\", which is not a port number: exim abandoned", s);
1193   if (lport == port) return TRUE;
1194   }
1195
1196 return FALSE;
1197 }
1198
1199
1200
1201 /*************************************************
1202 *        Check whether host is in a network      *
1203 *************************************************/
1204
1205 /* This function checks whether a given IP address matches a pattern that
1206 represents either a single host, or a network (using CIDR notation). The caller
1207 of this function must check the syntax of the arguments before calling it.
1208
1209 Arguments:
1210   host        string representation of the ip-address to check
1211   net         string representation of the network, with optional CIDR mask
1212   maskoffset  offset to the / that introduces the mask in the key
1213               zero if there is no mask
1214
1215 Returns:
1216   TRUE   the host is inside the network
1217   FALSE  the host is NOT inside the network
1218 */
1219
1220 BOOL
1221 host_is_in_net(uschar *host, uschar *net, int maskoffset)
1222 {
1223 int i;
1224 int address[4];
1225 int incoming[4];
1226 int mlen;
1227 int size = host_aton(net, address);
1228 int insize;
1229
1230 /* No mask => all bits to be checked */
1231
1232 if (maskoffset == 0) mlen = 99999;    /* Big number */
1233   else mlen = Uatoi(net + maskoffset + 1);
1234
1235 /* Convert the incoming address to binary. */
1236
1237 insize = host_aton(host, incoming);
1238
1239 /* Convert IPv4 addresses given in IPv6 compatible mode, which represent
1240    connections from IPv4 hosts to IPv6 hosts, that is, addresses of the form
1241    ::ffff:<v4address>, to IPv4 format. */
1242
1243 if (insize == 4 && incoming[0] == 0 && incoming[1] == 0 &&
1244     incoming[2] == 0xffff)
1245   {
1246   insize = 1;
1247   incoming[0] = incoming[3];
1248   }
1249
1250 /* No match if the sizes don't agree. */
1251
1252 if (insize != size) return FALSE;
1253
1254 /* Else do the masked comparison. */
1255
1256 for (i = 0; i < size; i++)
1257   {
1258   int mask;
1259   if (mlen == 0) mask = 0;
1260   else if (mlen < 32)
1261     {
1262     mask = (-1) << (32 - mlen);
1263     mlen = 0;
1264     }
1265   else
1266     {
1267     mask = -1;
1268     mlen -= 32;
1269     }
1270   if ((incoming[i] & mask) != (address[i] & mask)) return FALSE;
1271   }
1272
1273 return TRUE;
1274 }
1275
1276
1277
1278 /*************************************************
1279 *       Scan host list for local hosts           *
1280 *************************************************/
1281
1282 /* Scan through a chain of addresses and check whether any of them is the
1283 address of an interface on the local machine. If so, remove that address and
1284 any previous ones with the same MX value, and all subsequent ones (which will
1285 have greater or equal MX values) from the chain. Note: marking them as unusable
1286 is NOT the right thing to do because it causes the hosts not to be used for
1287 other domains, for which they may well be correct.
1288
1289 The hosts may be part of a longer chain; we only process those between the
1290 initial pointer and the "last" pointer.
1291
1292 There is also a list of "pseudo-local" host names which are checked against the
1293 host names. Any match causes that host item to be treated the same as one which
1294 matches a local IP address.
1295
1296 If the very first host is a local host, then all MX records had a precedence
1297 greater than or equal to that of the local host. Either there's a problem in
1298 the DNS, or an apparently remote name turned out to be an abbreviation for the
1299 local host. Give a specific return code, and let the caller decide what to do.
1300 Otherwise, give a success code if at least one host address has been found.
1301
1302 Arguments:
1303   host        pointer to the first host in the chain
1304   lastptr     pointer to pointer to the last host in the chain (may be updated)
1305   removed     if not NULL, set TRUE if some local addresses were removed
1306                 from the list
1307
1308 Returns:
1309   HOST_FOUND       if there is at least one host with an IP address on the chain
1310                      and an MX value less than any MX value associated with the
1311                      local host
1312   HOST_FOUND_LOCAL if a local host is among the lowest-numbered MX hosts; when
1313                      the host addresses were obtained from A records or
1314                      gethostbyname(), the MX values are set to -1.
1315   HOST_FIND_FAILED if no valid hosts with set IP addresses were found
1316 */
1317
1318 int
1319 host_scan_for_local_hosts(host_item *host, host_item **lastptr, BOOL *removed)
1320 {
1321 int yield = HOST_FIND_FAILED;
1322 host_item *last = *lastptr;
1323 host_item *prev = NULL;
1324 host_item *h;
1325
1326 if (removed != NULL) *removed = FALSE;
1327
1328 if (local_interface_data == NULL) local_interface_data = host_find_interfaces();
1329
1330 for (h = host; h != last->next; h = h->next)
1331   {
1332   #ifndef STAND_ALONE
1333   if (hosts_treat_as_local != NULL)
1334     {
1335     int rc;
1336     uschar *save = deliver_domain;
1337     deliver_domain = h->name;   /* set $domain */
1338     rc = match_isinlist(string_copylc(h->name), &hosts_treat_as_local, 0,
1339       &domainlist_anchor, NULL, MCL_DOMAIN, TRUE, NULL);
1340     deliver_domain = save;
1341     if (rc == OK) goto FOUND_LOCAL;
1342     }
1343   #endif
1344
1345   /* It seems that on many operating systems, 0.0.0.0 is treated as a synonym
1346   for 127.0.0.1 and refers to the local host. We therefore force it always to
1347   be treated as local. */
1348
1349   if (h->address != NULL)
1350     {
1351     ip_address_item *ip;
1352     if (Ustrcmp(h->address, "0.0.0.0") == 0) goto FOUND_LOCAL;
1353     for (ip = local_interface_data; ip != NULL; ip = ip->next)
1354       if (Ustrcmp(h->address, ip->address) == 0) goto FOUND_LOCAL;
1355     yield = HOST_FOUND;  /* At least one remote address has been found */
1356     }
1357
1358   /* Update prev to point to the last host item before any that have
1359   the same MX value as the one we have just considered. */
1360
1361   if (h->next == NULL || h->next->mx != h->mx) prev = h;
1362   }
1363
1364 return yield;  /* No local hosts found: return HOST_FOUND or HOST_FIND_FAILED */
1365
1366 /* A host whose IP address matches a local IP address, or whose name matches
1367 something in hosts_treat_as_local has been found. */
1368
1369 FOUND_LOCAL:
1370
1371 if (prev == NULL)
1372   {
1373   HDEBUG(D_host_lookup) debug_printf((h->mx >= 0)?
1374     "local host has lowest MX\n" :
1375     "local host found for non-MX address\n");
1376   return HOST_FOUND_LOCAL;
1377   }
1378
1379 HDEBUG(D_host_lookup)
1380   {
1381   debug_printf("local host in host list - removed hosts:\n");
1382   for (h = prev->next; h != last->next; h = h->next)
1383     debug_printf("  %s %s %d\n", h->name, h->address, h->mx);
1384   }
1385
1386 if (removed != NULL) *removed = TRUE;
1387 prev->next = last->next;
1388 *lastptr = prev;
1389 return yield;
1390 }
1391
1392
1393
1394
1395 /*************************************************
1396 *        Remove duplicate IPs in host list       *
1397 *************************************************/
1398
1399 /* You would think that administrators could set up their DNS records so that
1400 one ended up with a list of unique IP addresses after looking up A or MX
1401 records, but apparently duplication is common. So we scan such lists and
1402 remove the later duplicates. Note that we may get lists in which some host
1403 addresses are not set.
1404
1405 Arguments:
1406   host        pointer to the first host in the chain
1407   lastptr     pointer to pointer to the last host in the chain (may be updated)
1408
1409 Returns:      nothing
1410 */
1411
1412 static void
1413 host_remove_duplicates(host_item *host, host_item **lastptr)
1414 {
1415 while (host != *lastptr)
1416   {
1417   if (host->address != NULL)
1418     {
1419     host_item *h = host;
1420     while (h != *lastptr)
1421       {
1422       if (h->next->address != NULL &&
1423           Ustrcmp(h->next->address, host->address) == 0)
1424         {
1425         DEBUG(D_host_lookup) debug_printf("duplicate IP address %s (MX=%d) "
1426           "removed\n", host->address, h->next->mx);
1427         if (h->next == *lastptr) *lastptr = h;
1428         h->next = h->next->next;
1429         }
1430       else h = h->next;
1431       }
1432     }
1433   /* If the last item was removed, host may have become == *lastptr */
1434   if (host != *lastptr) host = host->next;
1435   }
1436 }
1437
1438
1439
1440
1441 /*************************************************
1442 *    Find sender host name by gethostbyaddr()    *
1443 *************************************************/
1444
1445 /* This used to be the only way it was done, but it turns out that not all
1446 systems give aliases for calls to gethostbyaddr() - or one of the modern
1447 equivalents like getipnodebyaddr(). Fortunately, multiple PTR records are rare,
1448 but they can still exist. This function is now used only when a DNS lookup of
1449 the IP address fails, in order to give access to /etc/hosts.
1450
1451 Arguments:   none
1452 Returns:     OK, DEFER, FAIL
1453 */
1454
1455 static int
1456 host_name_lookup_byaddr(void)
1457 {
1458 int len;
1459 uschar *s, *t;
1460 struct hostent *hosts;
1461 struct in_addr addr;
1462
1463 /* Lookup on IPv6 system */
1464
1465 #if HAVE_IPV6
1466 if (Ustrchr(sender_host_address, ':') != NULL)
1467   {
1468   struct in6_addr addr6;
1469   if (inet_pton(AF_INET6, CS sender_host_address, &addr6) != 1)
1470     log_write(0, LOG_MAIN|LOG_PANIC_DIE, "unable to parse \"%s\" as an "
1471       "IPv6 address", sender_host_address);
1472   #if HAVE_GETIPNODEBYADDR
1473   hosts = getipnodebyaddr(CS &addr6, sizeof(addr6), AF_INET6, &h_errno);
1474   #else
1475   hosts = gethostbyaddr(CS &addr6, sizeof(addr6), AF_INET6);
1476   #endif
1477   }
1478 else
1479   {
1480   if (inet_pton(AF_INET, CS sender_host_address, &addr) != 1)
1481     log_write(0, LOG_MAIN|LOG_PANIC_DIE, "unable to parse \"%s\" as an "
1482       "IPv4 address", sender_host_address);
1483   #if HAVE_GETIPNODEBYADDR
1484   hosts = getipnodebyaddr(CS &addr, sizeof(addr), AF_INET, &h_errno);
1485   #else
1486   hosts = gethostbyaddr(CS &addr, sizeof(addr), AF_INET);
1487   #endif
1488   }
1489
1490 /* Do lookup on IPv4 system */
1491
1492 #else
1493 addr.s_addr = (S_ADDR_TYPE)inet_addr(CS sender_host_address);
1494 hosts = gethostbyaddr(CS(&addr), sizeof(addr), AF_INET);
1495 #endif
1496
1497 /* Failed to look up the host. */
1498
1499 if (hosts == NULL)
1500   {
1501   HDEBUG(D_host_lookup) debug_printf("IP address lookup failed: h_errno=%d\n",
1502     h_errno);
1503   return (h_errno == TRY_AGAIN || h_errno == NO_RECOVERY) ? DEFER : FAIL;
1504   }
1505
1506 /* It seems there are some records in the DNS that yield an empty name. We
1507 treat this as non-existent. In some operating systems, this is returned as an
1508 empty string; in others as a single dot. */
1509
1510 if (hosts->h_name == NULL || hosts->h_name[0] == 0 || hosts->h_name[0] == '.')
1511   {
1512   HDEBUG(D_host_lookup) debug_printf("IP address lookup yielded an empty name: "
1513     "treated as non-existent host name\n");
1514   return FAIL;
1515   }
1516
1517 /* Copy and lowercase the name, which is in static storage in many systems.
1518 Put it in permanent memory. */
1519
1520 s = (uschar *)hosts->h_name;
1521 len = Ustrlen(s) + 1;
1522 t = sender_host_name = store_get_perm(len);
1523 while (*s != 0) *t++ = tolower(*s++);
1524 *t = 0;
1525
1526 /* If the host has aliases, build a copy of the alias list */
1527
1528 if (hosts->h_aliases != NULL)
1529   {
1530   int count = 1;
1531   uschar **aliases, **ptr;
1532   for (aliases = USS hosts->h_aliases; *aliases != NULL; aliases++) count++;
1533   ptr = sender_host_aliases = store_get_perm(count * sizeof(uschar *));
1534   for (aliases = USS hosts->h_aliases; *aliases != NULL; aliases++)
1535     {
1536     uschar *s = *aliases;
1537     int len = Ustrlen(s) + 1;
1538     uschar *t = *ptr++ = store_get_perm(len);
1539     while (*s != 0) *t++ = tolower(*s++);
1540     *t = 0;
1541     }
1542   *ptr = NULL;
1543   }
1544
1545 return OK;
1546 }
1547
1548
1549
1550 /*************************************************
1551 *        Find host name for incoming call        *
1552 *************************************************/
1553
1554 /* Put the name in permanent store, pointed to by sender_host_name. We also set
1555 up a list of alias names, pointed to by sender_host_alias. The list is
1556 NULL-terminated. The incoming address is in sender_host_address, either in
1557 dotted-quad form for IPv4 or in colon-separated form for IPv6.
1558
1559 This function does a thorough check that the names it finds point back to the
1560 incoming IP address. Any that do not are discarded. Note that this is relied on
1561 by the ACL reverse_host_lookup check.
1562
1563 On some systems, get{host,ipnode}byaddr() appears to do this internally, but
1564 this it not universally true. Also, for release 4.30, this function was changed
1565 to do a direct DNS lookup first, by default[1], because it turns out that that
1566 is the only guaranteed way to find all the aliases on some systems. My
1567 experiments indicate that Solaris gethostbyaddr() gives the aliases for but
1568 Linux does not.
1569
1570 [1] The actual order is controlled by the host_lookup_order option.
1571
1572 Arguments:    none
1573 Returns:      OK on success, the answer being placed in the global variable
1574                 sender_host_name, with any aliases in a list hung off
1575                 sender_host_aliases
1576               FAIL if no host name can be found
1577               DEFER if a temporary error was encountered
1578
1579 The variable host_lookup_msg is set to an empty string on sucess, or to a
1580 reason for the failure otherwise, in a form suitable for tagging onto an error
1581 message, and also host_lookup_failed is set TRUE if the lookup failed. If there
1582 was a defer, host_lookup_deferred is set TRUE.
1583
1584 Any dynamically constructed string for host_lookup_msg must be in permanent
1585 store, because it might be used for several incoming messages on the same SMTP
1586 connection. */
1587
1588 int
1589 host_name_lookup(void)
1590 {
1591 int old_pool, rc;
1592 int sep = 0;
1593 uschar *hname, *save_hostname;
1594 uschar **aliases;
1595 uschar buffer[256];
1596 uschar *ordername;
1597 uschar *list = host_lookup_order;
1598 dns_record *rr;
1599 dns_answer dnsa;
1600 dns_scan dnss;
1601
1602 host_lookup_deferred = host_lookup_failed = FALSE;
1603
1604 HDEBUG(D_host_lookup)
1605   debug_printf("looking up host name for %s\n", sender_host_address);
1606
1607 /* For testing the case when a lookup does not complete, we have a special
1608 reserved IP address. */
1609
1610 if (running_in_test_harness &&
1611     Ustrcmp(sender_host_address, "99.99.99.99") == 0)
1612   {
1613   HDEBUG(D_host_lookup)
1614     debug_printf("Test harness: host name lookup returns DEFER\n");
1615   host_lookup_deferred = TRUE;
1616   return DEFER;
1617   }
1618
1619 /* Do lookups directly in the DNS or via gethostbyaddr() (or equivalent), in
1620 the order specified by the host_lookup_order option. */
1621
1622 while ((ordername = string_nextinlist(&list, &sep, buffer, sizeof(buffer)))
1623         != NULL)
1624   {
1625   if (strcmpic(ordername, US"bydns") == 0)
1626     {
1627     dns_init(FALSE, FALSE);
1628     dns_build_reverse(sender_host_address, buffer);
1629     rc = dns_lookup(&dnsa, buffer, T_PTR, NULL);
1630
1631     /* The first record we come across is used for the name; others are
1632     considered to be aliases. We have to scan twice, in order to find out the
1633     number of aliases. However, if all the names are empty, we will behave as
1634     if failure. (PTR records that yield empty names have been encountered in
1635     the DNS.) */
1636
1637     if (rc == DNS_SUCCEED)
1638       {
1639       uschar **aptr = NULL;
1640       int ssize = 264;
1641       int count = 0;
1642       int old_pool = store_pool;
1643
1644       store_pool = POOL_PERM;        /* Save names in permanent storage */
1645
1646       for (rr = dns_next_rr(&dnsa, &dnss, RESET_ANSWERS);
1647            rr != NULL;
1648            rr = dns_next_rr(&dnsa, &dnss, RESET_NEXT))
1649         {
1650         if (rr->type == T_PTR) count++;
1651         }
1652
1653       /* Get store for the list of aliases. For compatibility with
1654       gethostbyaddr, we make an empty list if there are none. */
1655
1656       aptr = sender_host_aliases = store_get(count * sizeof(uschar *));
1657
1658       /* Re-scan and extract the names */
1659
1660       for (rr = dns_next_rr(&dnsa, &dnss, RESET_ANSWERS);
1661            rr != NULL;
1662            rr = dns_next_rr(&dnsa, &dnss, RESET_NEXT))
1663         {
1664         uschar *s = NULL;
1665         if (rr->type != T_PTR) continue;
1666         s = store_get(ssize);
1667
1668         /* If an overlong response was received, the data will have been
1669         truncated and dn_expand may fail. */
1670
1671         if (dn_expand(dnsa.answer, dnsa.answer + dnsa.answerlen,
1672              (uschar *)(rr->data), (DN_EXPAND_ARG4_TYPE)(s), ssize) < 0)
1673           {
1674           log_write(0, LOG_MAIN, "host name alias list truncated for %s",
1675             sender_host_address);
1676           break;
1677           }
1678
1679         store_reset(s + Ustrlen(s) + 1);
1680         if (s[0] == 0)
1681           {
1682           HDEBUG(D_host_lookup) debug_printf("IP address lookup yielded an "
1683             "empty name: treated as non-existent host name\n");
1684           continue;
1685           }
1686         if (sender_host_name == NULL) sender_host_name = s;
1687           else *aptr++ = s;
1688         while (*s != 0) { *s = tolower(*s); s++; }
1689         }
1690
1691       *aptr = NULL;            /* End of alias list */
1692       store_pool = old_pool;   /* Reset store pool */
1693
1694       /* If we've found a names, break out of the "order" loop */
1695
1696       if (sender_host_name != NULL) break;
1697       }
1698
1699     /* If the DNS lookup deferred, we must also defer. */
1700
1701     if (rc == DNS_AGAIN)
1702       {
1703       HDEBUG(D_host_lookup)
1704         debug_printf("IP address PTR lookup gave temporary error\n");
1705       host_lookup_deferred = TRUE;
1706       return DEFER;
1707       }
1708     }
1709
1710   /* Do a lookup using gethostbyaddr() - or equivalent */
1711
1712   else if (strcmpic(ordername, US"byaddr") == 0)
1713     {
1714     HDEBUG(D_host_lookup)
1715       debug_printf("IP address lookup using gethostbyaddr()\n");
1716     rc = host_name_lookup_byaddr();
1717     if (rc == DEFER)
1718       {
1719       host_lookup_deferred = TRUE;
1720       return rc;                       /* Can't carry on */
1721       }
1722     if (rc == OK) break;               /* Found a name */
1723     }
1724   }      /* Loop for bydns/byaddr scanning */
1725
1726 /* If we have failed to find a name, return FAIL and log when required.
1727 NB host_lookup_msg must be in permanent store.  */
1728
1729 if (sender_host_name == NULL)
1730   {
1731   if (host_checking || !log_testing_mode)
1732     log_write(L_host_lookup_failed, LOG_MAIN, "no host name found for IP "
1733       "address %s", sender_host_address);
1734   host_lookup_msg = US" (failed to find host name from IP address)";
1735   host_lookup_failed = TRUE;
1736   return FAIL;
1737   }
1738
1739 HDEBUG(D_host_lookup)
1740   {
1741   uschar **aliases = sender_host_aliases;
1742   debug_printf("IP address lookup yielded %s\n", sender_host_name);
1743   while (*aliases != NULL) debug_printf("  alias %s\n", *aliases++);
1744   }
1745
1746 /* We need to verify that a forward lookup on the name we found does indeed
1747 correspond to the address. This is for security: in principle a malefactor who
1748 happened to own a reverse zone could set it to point to any names at all.
1749
1750 This code was present in versions of Exim before 3.20. At that point I took it
1751 out because I thought that gethostbyaddr() did the check anyway. It turns out
1752 that this isn't always the case, so it's coming back in at 4.01. This version
1753 is actually better, because it also checks aliases.
1754
1755 The code was made more robust at release 4.21. Prior to that, it accepted all
1756 the names if any of them had the correct IP address. Now the code checks all
1757 the names, and accepts only those that have the correct IP address. */
1758
1759 save_hostname = sender_host_name;   /* Save for error messages */
1760 aliases = sender_host_aliases;
1761 for (hname = sender_host_name; hname != NULL; hname = *aliases++)
1762   {
1763   int rc;
1764   BOOL ok = FALSE;
1765   host_item h;
1766   h.next = NULL;
1767   h.name = hname;
1768   h.mx = MX_NONE;
1769   h.address = NULL;
1770
1771   /* When called with the last argument FALSE, host_find_byname() won't return
1772   HOST_FOUND_LOCAL. If the incoming address is an IPv4 address expressed in
1773   IPv6 format, we must compare the IPv4 part to any IPv4 addresses. */
1774
1775   if ((rc = host_find_byname(&h, NULL, 0, NULL, FALSE)) == HOST_FOUND)
1776     {
1777     host_item *hh;
1778     HDEBUG(D_host_lookup) debug_printf("checking addresses for %s\n", hname);
1779     for (hh = &h; hh != NULL; hh = hh->next)
1780       {
1781       if (host_is_in_net(hh->address, sender_host_address, 0))
1782         {
1783         HDEBUG(D_host_lookup) debug_printf("  %s OK\n", hh->address);
1784         ok = TRUE;
1785         break;
1786         }
1787       else
1788         {
1789         HDEBUG(D_host_lookup) debug_printf("  %s\n", hh->address);
1790         }
1791       }
1792     if (!ok) HDEBUG(D_host_lookup)
1793       debug_printf("no IP address for %s matched %s\n", hname,
1794         sender_host_address);
1795     }
1796   else if (rc == HOST_FIND_AGAIN)
1797     {
1798     HDEBUG(D_host_lookup) debug_printf("temporary error for host name lookup\n");
1799     host_lookup_deferred = TRUE;
1800     sender_host_name = NULL;
1801     return DEFER;
1802     }
1803   else
1804     {
1805     HDEBUG(D_host_lookup) debug_printf("no IP addresses found for %s\n", hname);
1806     }
1807
1808   /* If this name is no good, and it's the sender name, set it null pro tem;
1809   if it's an alias, just remove it from the list. */
1810
1811   if (!ok)
1812     {
1813     if (hname == sender_host_name) sender_host_name = NULL; else
1814       {
1815       uschar **a;                              /* Don't amalgamate - some */
1816       a = --aliases;                           /* compilers grumble */
1817       while (*a != NULL) { *a = a[1]; a++; }
1818       }
1819     }
1820   }
1821
1822 /* If sender_host_name == NULL, it means we didn't like the name. Replace
1823 it with the first alias, if there is one. */
1824
1825 if (sender_host_name == NULL && *sender_host_aliases != NULL)
1826   sender_host_name = *sender_host_aliases++;
1827
1828 /* If we now have a main name, all is well. */
1829
1830 if (sender_host_name != NULL) return OK;
1831
1832 /* We have failed to find an address that matches. */
1833
1834 HDEBUG(D_host_lookup)
1835   debug_printf("%s does not match any IP address for %s\n",
1836     sender_host_address, save_hostname);
1837
1838 /* This message must be in permanent store */
1839
1840 old_pool = store_pool;
1841 store_pool = POOL_PERM;
1842 host_lookup_msg = string_sprintf(" (%s does not match any IP address for %s)",
1843   sender_host_address, save_hostname);
1844 store_pool = old_pool;
1845 host_lookup_failed = TRUE;
1846 return FAIL;
1847 }
1848
1849
1850
1851
1852 /*************************************************
1853 *    Find IP address(es) for host by name        *
1854 *************************************************/
1855
1856 /* The input is a host_item structure with the name filled in and the address
1857 field set to NULL. We use gethostbyname() or getipnodebyname() or
1858 gethostbyname2(), as appropriate. Of course, these functions may use the DNS,
1859 but they do not do MX processing. It appears, however, that in some systems the
1860 current setting of resolver options is used when one of these functions calls
1861 the resolver. For this reason, we call dns_init() at the start, with arguments
1862 influenced by bits in "flags", just as we do for host_find_bydns().
1863
1864 The second argument provides a host list (usually an IP list) of hosts to
1865 ignore. This makes it possible to ignore IPv6 link-local addresses or loopback
1866 addresses in unreasonable places.
1867
1868 The lookup may result in a change of name. For compatibility with the dns
1869 lookup, return this via fully_qualified_name as well as updating the host item.
1870 The lookup may also yield more than one IP address, in which case chain on
1871 subsequent host_item structures.
1872
1873 Arguments:
1874   host                   a host item with the name and MX filled in;
1875                            the address is to be filled in;
1876                            multiple IP addresses cause other host items to be
1877                              chained on.
1878   ignore_target_hosts    a list of hosts to ignore
1879   flags                  HOST_FIND_QUALIFY_SINGLE   ) passed to
1880                          HOST_FIND_SEARCH_PARENTS   )   dns_init()
1881   fully_qualified_name   if not NULL, set to point to host name for
1882                          compatibility with host_find_bydns
1883   local_host_check       TRUE if a check for the local host is wanted
1884
1885 Returns:                 HOST_FIND_FAILED  Failed to find the host or domain
1886                          HOST_FIND_AGAIN   Try again later
1887                          HOST_FOUND        Host found - data filled in
1888                          HOST_FOUND_LOCAL  Host found and is the local host
1889 */
1890
1891 int
1892 host_find_byname(host_item *host, uschar *ignore_target_hosts, int flags,
1893   uschar **fully_qualified_name, BOOL local_host_check)
1894 {
1895 int i, yield, times;
1896 uschar **addrlist;
1897 host_item *last = NULL;
1898 BOOL temp_error = FALSE;
1899 #if HAVE_IPV6
1900 int af;
1901 #endif
1902
1903 /* If we are in the test harness, a name ending in .test.again.dns always
1904 forces a temporary error response, unless the name is in
1905 dns_again_means_nonexist. */
1906
1907 if (running_in_test_harness)
1908   {
1909   uschar *endname = host->name + Ustrlen(host->name);
1910   if (Ustrcmp(endname - 14, "test.again.dns") == 0) goto RETURN_AGAIN;
1911   }
1912
1913 /* Make sure DNS options are set as required. This appears to be necessary in
1914 some circumstances when the get..byname() function actually calls the DNS. */
1915
1916 dns_init((flags & HOST_FIND_QUALIFY_SINGLE) != 0,
1917          (flags & HOST_FIND_SEARCH_PARENTS) != 0);
1918
1919 /* In an IPv6 world, unless IPv6 has been disabled, we need to scan for both
1920 kinds of address, so go round the loop twice. Note that we have ensured that
1921 AF_INET6 is defined even in an IPv4 world, which makes for slightly tidier
1922 code. However, if dns_ipv4_lookup matches the domain, we also just do IPv4
1923 lookups here (except when testing standalone). */
1924
1925 #if HAVE_IPV6
1926   #ifdef STAND_ALONE
1927   if (disable_ipv6)
1928   #else
1929   if (disable_ipv6 ||
1930     (dns_ipv4_lookup != NULL &&
1931         match_isinlist(host->name, &dns_ipv4_lookup, 0, NULL, NULL, MCL_DOMAIN,
1932           TRUE, NULL) == OK))
1933   #endif
1934
1935     { af = AF_INET; times = 1; }
1936   else
1937     { af = AF_INET6; times = 2; }
1938
1939 /* No IPv6 support */
1940
1941 #else   /* HAVE_IPV6 */
1942   times = 1;
1943 #endif  /* HAVE_IPV6 */
1944
1945 /* Initialize the flag that gets set for DNS syntax check errors, so that the
1946 interface to this function can be similar to host_find_bydns. */
1947
1948 host_find_failed_syntax = FALSE;
1949
1950 /* Loop to look up both kinds of address in an IPv6 world */
1951
1952 for (i = 1; i <= times;
1953      #if HAVE_IPV6
1954        af = AF_INET,     /* If 2 passes, IPv4 on the second */
1955      #endif
1956      i++)
1957   {
1958   BOOL ipv4_addr;
1959   int error_num = 0;
1960   struct hostent *hostdata;
1961
1962   #ifdef STAND_ALONE
1963   printf("Looking up: %s\n", host->name);
1964   #endif
1965
1966   #if HAVE_IPV6
1967   if (running_in_test_harness)
1968     hostdata = host_fake_gethostbyname(host->name, af, &error_num);
1969   else
1970     {
1971     #if HAVE_GETIPNODEBYNAME
1972     hostdata = getipnodebyname(CS host->name, af, 0, &error_num);
1973     #else
1974     hostdata = gethostbyname2(CS host->name, af);
1975     error_num = h_errno;
1976     #endif
1977     }
1978
1979   #else    /* not HAVE_IPV6 */
1980   if (running_in_test_harness)
1981     hostdata = host_fake_gethostbyname(host->name, AF_INET, &error_num);
1982   else
1983     {
1984     hostdata = gethostbyname(CS host->name);
1985     error_num = h_errno;
1986     }
1987   #endif   /* HAVE_IPV6 */
1988
1989   if (hostdata == NULL)
1990     {
1991     uschar *error;
1992     switch (error_num)
1993       {
1994       case HOST_NOT_FOUND: error = US"HOST_NOT_FOUND"; break;
1995       case TRY_AGAIN: error = US"TRY_AGAIN"; break;
1996       case NO_RECOVERY: error = US"NO_RECOVERY"; break;
1997       case NO_DATA: error = US"NO_DATA"; break;
1998       #if NO_DATA != NO_ADDRESS
1999       case NO_ADDRESS: error = US"NO_ADDRESS"; break;
2000       #endif
2001       default: error = US"?"; break;
2002       }
2003
2004     DEBUG(D_host_lookup) debug_printf("%s returned %d (%s)\n",
2005       #if HAVE_IPV6
2006         #if HAVE_GETIPNODEBYNAME
2007         (af == AF_INET6)? "getipnodebyname(af=inet6)" : "getipnodebyname(af=inet)",
2008         #else
2009         (af == AF_INET6)? "gethostbyname2(af=inet6)" : "gethostbyname2(af=inet)",
2010         #endif
2011       #else
2012       "gethostbyname",
2013       #endif
2014       error_num, error);
2015
2016     if (error_num == TRY_AGAIN || error_num == NO_RECOVERY) temp_error = TRUE;
2017     continue;
2018     }
2019   if ((hostdata->h_addr_list)[0] == NULL) continue;
2020
2021   /* Replace the name with the fully qualified one if necessary, and fill in
2022   the fully_qualified_name pointer. */
2023
2024   if (hostdata->h_name[0] != 0 &&
2025       Ustrcmp(host->name, hostdata->h_name) != 0)
2026     host->name = string_copy_dnsdomain((uschar *)hostdata->h_name);
2027   if (fully_qualified_name != NULL) *fully_qualified_name = host->name;
2028
2029   /* Get the list of addresses. IPv4 and IPv6 addresses can be distinguished
2030   by their different lengths. Scan the list, ignoring any that are to be
2031   ignored, and build a chain from the rest. */
2032
2033   ipv4_addr = hostdata->h_length == sizeof(struct in_addr);
2034
2035   for (addrlist = USS hostdata->h_addr_list; *addrlist != NULL; addrlist++)
2036     {
2037     uschar *text_address =
2038       host_ntoa(ipv4_addr? AF_INET:AF_INET6, *addrlist, NULL, NULL);
2039
2040     #ifndef STAND_ALONE
2041     if (ignore_target_hosts != NULL &&
2042         verify_check_this_host(&ignore_target_hosts, NULL, host->name,
2043           text_address, NULL) == OK)
2044       {
2045       DEBUG(D_host_lookup)
2046         debug_printf("ignored host %s [%s]\n", host->name, text_address);
2047       continue;
2048       }
2049     #endif
2050
2051     /* If this is the first address, last == NULL and we put the data in the
2052     original block. */
2053
2054     if (last == NULL)
2055       {
2056       host->address = text_address;
2057       host->port = PORT_NONE;
2058       host->status = hstatus_unknown;
2059       host->why = hwhy_unknown;
2060       last = host;
2061       }
2062
2063     /* Else add further host item blocks for any other addresses, keeping
2064     the order. */
2065
2066     else
2067       {
2068       host_item *next = store_get(sizeof(host_item));
2069       next->name = host->name;
2070       next->mx = host->mx;
2071       next->address = text_address;
2072       next->port = PORT_NONE;
2073       next->status = hstatus_unknown;
2074       next->why = hwhy_unknown;
2075       next->last_try = 0;
2076       next->next = last->next;
2077       last->next = next;
2078       last = next;
2079       }
2080     }
2081   }
2082
2083 /* If no hosts were found, the address field in the original host block will be
2084 NULL. If temp_error is set, at least one of the lookups gave a temporary error,
2085 so we pass that back. */
2086
2087 if (host->address == NULL)
2088   {
2089   uschar *msg =
2090     #ifndef STAND_ALONE
2091     (message_id[0] == 0 && smtp_in != NULL)?
2092       string_sprintf("no IP address found for host %s (during %s)", host->name,
2093           smtp_get_connection_info()) :
2094     #endif
2095     string_sprintf("no IP address found for host %s", host->name);
2096
2097   HDEBUG(D_host_lookup) debug_printf("%s\n", msg);
2098   if (temp_error) goto RETURN_AGAIN;
2099   if (host_checking || !log_testing_mode)
2100     log_write(L_host_lookup_failed, LOG_MAIN, "%s", msg);
2101   return HOST_FIND_FAILED;
2102   }
2103
2104 /* Remove any duplicate IP addresses, then check to see if this is the local
2105 host if required. */
2106
2107 host_remove_duplicates(host, &last);
2108 yield = local_host_check?
2109   host_scan_for_local_hosts(host, &last, NULL) : HOST_FOUND;
2110
2111 HDEBUG(D_host_lookup)
2112   {
2113   host_item *h;
2114   if (fully_qualified_name != NULL)
2115     debug_printf("fully qualified name = %s\n", *fully_qualified_name);
2116   debug_printf("%s looked up these IP addresses:\n",
2117     #if HAVE_IPV6
2118       #if HAVE_GETIPNODEBYNAME
2119       "getipnodebyname"
2120       #else
2121       "gethostbyname2"
2122       #endif
2123     #else
2124     "gethostbyname"
2125     #endif
2126     );
2127   for (h = host; h != last->next; h = h->next)
2128     debug_printf("  name=%s address=%s\n", h->name,
2129       (h->address == NULL)? US"<null>" : h->address);
2130   }
2131
2132 /* Return the found status. */
2133
2134 return yield;
2135
2136 /* Handle the case when there is a temporary error. If the name matches
2137 dns_again_means_nonexist, return permanent rather than temporary failure. */
2138
2139 RETURN_AGAIN:
2140   {
2141   #ifndef STAND_ALONE
2142   int rc;
2143   uschar *save = deliver_domain;
2144   deliver_domain = host->name;  /* set $domain */
2145   rc = match_isinlist(host->name, &dns_again_means_nonexist, 0, NULL, NULL,
2146     MCL_DOMAIN, TRUE, NULL);
2147   deliver_domain = save;
2148   if (rc == OK)
2149     {
2150     DEBUG(D_host_lookup) debug_printf("%s is in dns_again_means_nonexist: "
2151       "returning HOST_FIND_FAILED\n", host->name);
2152     return HOST_FIND_FAILED;
2153     }
2154   #endif
2155   return HOST_FIND_AGAIN;
2156   }
2157 }
2158
2159
2160
2161 /*************************************************
2162 *        Fill in a host address from the DNS     *
2163 *************************************************/
2164
2165 /* Given a host item, with its name, port and mx fields set, and its address
2166 field set to NULL, fill in its IP address from the DNS. If it is multi-homed,
2167 create additional host items for the additional addresses, copying all the
2168 other fields, and randomizing the order.
2169
2170 On IPv6 systems, A6 records are sought first (but only if support for A6 is
2171 configured - they may never become mainstream), then AAAA records are sought,
2172 and finally A records are sought as well.
2173
2174 The host name may be changed if the DNS returns a different name - e.g. fully
2175 qualified or changed via CNAME. If fully_qualified_name is not NULL, dns_lookup
2176 ensures that it points to the fully qualified name. However, this is the fully
2177 qualified version of the original name; if a CNAME is involved, the actual
2178 canonical host name may be different again, and so we get it directly from the
2179 relevant RR. Note that we do NOT change the mx field of the host item in this
2180 function as it may be called to set the addresses of hosts taken from MX
2181 records.
2182
2183 Arguments:
2184   host                  points to the host item we're filling in
2185   lastptr               points to pointer to last host item in a chain of
2186                           host items (may be updated if host is last and gets
2187                           extended because multihomed)
2188   ignore_target_hosts   list of hosts to ignore
2189   allow_ip              if TRUE, recognize an IP address and return it
2190   fully_qualified_name  if not NULL, return fully qualified name here if
2191                           the contents are different (i.e. it must be preset
2192                           to something)
2193
2194 Returns:       HOST_FIND_FAILED     couldn't find A record
2195                HOST_FIND_AGAIN      try again later
2196                HOST_FOUND           found AAAA and/or A record(s)
2197                HOST_IGNORED         found, but all IPs ignored
2198 */
2199
2200 static int
2201 set_address_from_dns(host_item *host, host_item **lastptr,
2202   uschar *ignore_target_hosts, BOOL allow_ip, uschar **fully_qualified_name)
2203 {
2204 dns_record *rr;
2205 host_item *thishostlast = NULL;    /* Indicates not yet filled in anything */
2206 BOOL v6_find_again = FALSE;
2207 int i;
2208
2209 /* If allow_ip is set, a name which is an IP address returns that value
2210 as its address. This is used for MX records when allow_mx_to_ip is set, for
2211 those sites that feel they have to flaunt the RFC rules. */
2212
2213 if (allow_ip && string_is_ip_address(host->name, NULL) != 0)
2214   {
2215   #ifndef STAND_ALONE
2216   if (ignore_target_hosts != NULL &&
2217         verify_check_this_host(&ignore_target_hosts, NULL, host->name,
2218         host->name, NULL) == OK)
2219     return HOST_IGNORED;
2220   #endif
2221
2222   host->address = host->name;
2223   return HOST_FOUND;
2224   }
2225
2226 /* On an IPv6 system, unless IPv6 is disabled, go round the loop up to three
2227 times, looking for A6 and AAAA records the first two times. However, unless
2228 doing standalone testing, we force an IPv4 lookup if the domain matches
2229 dns_ipv4_lookup is set. Since A6 records look like being abandoned, support
2230 them only if explicitly configured to do so. On an IPv4 system, go round the
2231 loop once only, looking only for A records. */
2232
2233 #if HAVE_IPV6
2234   #ifndef STAND_ALONE
2235     if (disable_ipv6 || (dns_ipv4_lookup != NULL &&
2236         match_isinlist(host->name, &dns_ipv4_lookup, 0, NULL, NULL, MCL_DOMAIN,
2237         TRUE, NULL) == OK))
2238       i = 0;    /* look up A records only */
2239     else
2240   #endif        /* STAND_ALONE */
2241
2242   #ifdef SUPPORT_A6
2243   i = 2;        /* look up A6 and AAAA and A records */
2244   #else
2245   i = 1;        /* look up AAAA and A records */
2246   #endif        /* SUPPORT_A6 */
2247
2248 /* The IPv4 world */
2249
2250 #else           /* HAVE_IPV6 */
2251   i = 0;        /* look up A records only */
2252 #endif          /* HAVE_IPV6 */
2253
2254 for (; i >= 0; i--)
2255   {
2256   static int types[] = { T_A, T_AAAA, T_A6 };
2257   int type = types[i];
2258   int randoffset = (i == 0)? 500 : 0;  /* Ensures v6 sorts before v4 */
2259   dns_answer dnsa;
2260   dns_scan dnss;
2261
2262   int rc = dns_lookup(&dnsa, host->name, type, fully_qualified_name);
2263
2264   /* We want to return HOST_FIND_AGAIN if one of the A, A6, or AAAA lookups
2265   fails or times out, but not if another one succeeds. (In the early
2266   IPv6 days there are name servers that always fail on AAAA, but are happy
2267   to give out an A record. We want to proceed with that A record.) */
2268
2269   if (rc != DNS_SUCCEED)
2270     {
2271     if (i == 0)  /* Just tried for an A record, i.e. end of loop */
2272       {
2273       if (host->address != NULL) return HOST_FOUND;  /* A6 or AAAA was found */
2274       if (rc == DNS_AGAIN || rc == DNS_FAIL || v6_find_again)
2275         return HOST_FIND_AGAIN;
2276       return HOST_FIND_FAILED;    /* DNS_NOMATCH or DNS_NODATA */
2277       }
2278
2279     /* Tried for an A6 or AAAA record: remember if this was a temporary
2280     error, and look for the next record type. */
2281
2282     if (rc != DNS_NOMATCH && rc != DNS_NODATA) v6_find_again = TRUE;
2283     continue;
2284     }
2285
2286   /* Lookup succeeded: fill in the given host item with the first non-ignored
2287   address found; create additional items for any others. A single A6 record
2288   may generate more than one address. */
2289
2290   for (rr = dns_next_rr(&dnsa, &dnss, RESET_ANSWERS);
2291        rr != NULL;
2292        rr = dns_next_rr(&dnsa, &dnss, RESET_NEXT))
2293     {
2294     if (rr->type == type)
2295       {
2296       /* dns_address *da = dns_address_from_rr(&dnsa, rr); */
2297
2298       dns_address *da;
2299       da = dns_address_from_rr(&dnsa, rr);
2300
2301       DEBUG(D_host_lookup)
2302         {
2303         if (da == NULL)
2304           debug_printf("no addresses extracted from A6 RR for %s\n",
2305             host->name);
2306         }
2307
2308       /* This loop runs only once for A and AAAA records, but may run
2309       several times for an A6 record that generated multiple addresses. */
2310
2311       for (; da != NULL; da = da->next)
2312         {
2313         #ifndef STAND_ALONE
2314         if (ignore_target_hosts != NULL &&
2315               verify_check_this_host(&ignore_target_hosts, NULL,
2316                 host->name, da->address, NULL) == OK)
2317           {
2318           DEBUG(D_host_lookup)
2319             debug_printf("ignored host %s [%s]\n", host->name, da->address);
2320           continue;
2321           }
2322         #endif
2323
2324         /* If this is the first address, stick it in the given host block,
2325         and change the name if the returned RR has a different name. */
2326
2327         if (thishostlast == NULL)
2328           {
2329           if (strcmpic(host->name, rr->name) != 0)
2330             host->name = string_copy_dnsdomain(rr->name);
2331           host->address = da->address;
2332           host->sort_key = host->mx * 1000 + random_number(500) + randoffset;
2333           host->status = hstatus_unknown;
2334           host->why = hwhy_unknown;
2335           thishostlast = host;
2336           }
2337
2338         /* Not the first address. Check for, and ignore, duplicates. Then
2339         insert in the chain at a random point. */
2340
2341         else
2342           {
2343           int new_sort_key;
2344           host_item *next;
2345
2346           /* End of our local chain is specified by "thishostlast". */
2347
2348           for (next = host;; next = next->next)
2349             {
2350             if (Ustrcmp(CS da->address, next->address) == 0) break;
2351             if (next == thishostlast) { next = NULL; break; }
2352             }
2353           if (next != NULL) continue;  /* With loop for next address */
2354
2355           /* Not a duplicate */
2356
2357           new_sort_key = host->mx * 1000 + random_number(500) + randoffset;
2358           next = store_get(sizeof(host_item));
2359
2360           /* New address goes first: insert the new block after the first one
2361           (so as not to disturb the original pointer) but put the new address
2362           in the original block. */
2363
2364           if (new_sort_key < host->sort_key)
2365             {
2366             *next = *host;                                  /* Copies port */
2367             host->next = next;
2368             host->address = da->address;
2369             host->sort_key = new_sort_key;
2370             if (thishostlast == host) thishostlast = next;  /* Local last */
2371             if (*lastptr == host) *lastptr = next;          /* Global last */
2372             }
2373
2374           /* Otherwise scan down the addresses for this host to find the
2375           one to insert after. */
2376
2377           else
2378             {
2379             host_item *h = host;
2380             while (h != thishostlast)
2381               {
2382               if (new_sort_key < h->next->sort_key) break;
2383               h = h->next;
2384               }
2385             *next = *h;                                 /* Copies port */
2386             h->next = next;
2387             next->address = da->address;
2388             next->sort_key = new_sort_key;
2389             if (h == thishostlast) thishostlast = next; /* Local last */
2390             if (h == *lastptr) *lastptr = next;         /* Global last */
2391             }
2392           }
2393         }
2394       }
2395     }
2396   }
2397
2398 /* Control gets here only if the third lookup (the A record) succeeded.
2399 However, the address may not be filled in if it was ignored. */
2400
2401 return (host->address == NULL)? HOST_IGNORED : HOST_FOUND;
2402 }
2403
2404
2405
2406
2407 /*************************************************
2408 *    Find IP addresses and host names via DNS    *
2409 *************************************************/
2410
2411 /* The input is a host_item structure with the name field filled in and the
2412 address field set to NULL. This may be in a chain of other host items. The
2413 lookup may result in more than one IP address, in which case we must created
2414 new host blocks for the additional addresses, and insert them into the chain.
2415 The original name may not be fully qualified. Use the fully_qualified_name
2416 argument to return the official name, as returned by the resolver.
2417
2418 Arguments:
2419   host                  point to initial host item
2420   ignore_target_hosts   a list of hosts to ignore
2421   whichrrs              flags indicating which RRs to look for:
2422                           HOST_FIND_BY_SRV  => look for SRV
2423                           HOST_FIND_BY_MX   => look for MX
2424                           HOST_FIND_BY_A    => look for A or AAAA
2425                         also flags indicating how the lookup is done
2426                           HOST_FIND_QUALIFY_SINGLE   ) passed to the
2427                           HOST_FIND_SEARCH_PARENTS   )   resolver
2428   srv_service           when SRV used, the service name
2429   srv_fail_domains      DNS errors for these domains => assume nonexist
2430   mx_fail_domains       DNS errors for these domains => assume nonexist
2431   fully_qualified_name  if not NULL, return fully-qualified name
2432   removed               set TRUE if local host was removed from the list
2433
2434 Returns:                HOST_FIND_FAILED  Failed to find the host or domain;
2435                                           if there was a syntax error,
2436                                           host_find_failed_syntax is set.
2437                         HOST_FIND_AGAIN   Could not resolve at this time
2438                         HOST_FOUND        Host found
2439                         HOST_FOUND_LOCAL  The lowest MX record points to this
2440                                           machine, if MX records were found, or
2441                                           an A record that was found contains
2442                                           an address of the local host
2443 */
2444
2445 int
2446 host_find_bydns(host_item *host, uschar *ignore_target_hosts, int whichrrs,
2447   uschar *srv_service, uschar *srv_fail_domains, uschar *mx_fail_domains,
2448   uschar **fully_qualified_name, BOOL *removed)
2449 {
2450 host_item *h, *last;
2451 dns_record *rr;
2452 int rc = DNS_FAIL;
2453 int ind_type = 0;
2454 int yield;
2455 dns_answer dnsa;
2456 dns_scan dnss;
2457
2458 /* Set the default fully qualified name to the incoming name, initialize the
2459 resolver if necessary, set up the relevant options, and initialize the flag
2460 that gets set for DNS syntax check errors. */
2461
2462 if (fully_qualified_name != NULL) *fully_qualified_name = host->name;
2463 dns_init((whichrrs & HOST_FIND_QUALIFY_SINGLE) != 0,
2464          (whichrrs & HOST_FIND_SEARCH_PARENTS) != 0);
2465 host_find_failed_syntax = FALSE;
2466
2467 /* First, if requested, look for SRV records. The service name is given; we
2468 assume TCP progocol. DNS domain names are constrained to a maximum of 256
2469 characters, so the code below should be safe. */
2470
2471 if ((whichrrs & HOST_FIND_BY_SRV) != 0)
2472   {
2473   uschar buffer[300];
2474   uschar *temp_fully_qualified_name = buffer;
2475   int prefix_length;
2476
2477   (void)sprintf(CS buffer, "_%s._tcp.%n%.256s", srv_service, &prefix_length,
2478     host->name);
2479   ind_type = T_SRV;
2480
2481   /* Search for SRV records. If the fully qualified name is different to
2482   the input name, pass back the new original domain, without the prepended
2483   magic. */
2484
2485   rc = dns_lookup(&dnsa, buffer, ind_type, &temp_fully_qualified_name);
2486   if (temp_fully_qualified_name != buffer && fully_qualified_name != NULL)
2487     *fully_qualified_name = temp_fully_qualified_name + prefix_length;
2488
2489   /* On DNS failures, we give the "try again" error unless the domain is
2490   listed as one for which we continue. */
2491
2492   if (rc == DNS_FAIL || rc == DNS_AGAIN)
2493     {
2494     #ifndef STAND_ALONE
2495     if (match_isinlist(host->name, &srv_fail_domains, 0, NULL, NULL, MCL_DOMAIN,
2496         TRUE, NULL) != OK)
2497     #endif
2498       return HOST_FIND_AGAIN;
2499     DEBUG(D_host_lookup) debug_printf("DNS_%s treated as DNS_NODATA "
2500       "(domain in srv_fail_domains)\n", (rc == DNS_FAIL)? "FAIL":"AGAIN");
2501     }
2502   }
2503
2504 /* If we did not find any SRV records, search the DNS for MX records, if
2505 requested to do so. If the result is DNS_NOMATCH, it means there is no such
2506 domain, and there's no point in going on to look for address records with the
2507 same domain. The result will be DNS_NODATA if the domain exists but has no MX
2508 records. On DNS failures, we give the "try again" error unless the domain is
2509 listed as one for which we continue. */
2510
2511 if (rc != DNS_SUCCEED && (whichrrs & HOST_FIND_BY_MX) != 0)
2512   {
2513   ind_type = T_MX;
2514   rc = dns_lookup(&dnsa, host->name, ind_type, fully_qualified_name);
2515   if (rc == DNS_NOMATCH) return HOST_FIND_FAILED;
2516   if (rc == DNS_FAIL || rc == DNS_AGAIN)
2517     {
2518     #ifndef STAND_ALONE
2519     if (match_isinlist(host->name, &mx_fail_domains, 0, NULL, NULL, MCL_DOMAIN,
2520         TRUE, NULL) != OK)
2521     #endif
2522       return HOST_FIND_AGAIN;
2523     DEBUG(D_host_lookup) debug_printf("DNS_%s treated as DNS_NODATA "
2524       "(domain in mx_fail_domains)\n", (rc == DNS_FAIL)? "FAIL":"AGAIN");
2525     }
2526   }
2527
2528 /* If we haven't found anything yet, and we are requested to do so, try for an
2529 A or AAAA record. If we find it (or them) check to see that it isn't the local
2530 host. */
2531
2532 if (rc != DNS_SUCCEED)
2533   {
2534   if ((whichrrs & HOST_FIND_BY_A) == 0)
2535     {
2536     DEBUG(D_host_lookup) debug_printf("Address records are not being sought\n");
2537     return HOST_FIND_FAILED;
2538     }
2539
2540   last = host;        /* End of local chainlet */
2541   host->mx = MX_NONE;
2542   host->port = PORT_NONE;
2543   rc = set_address_from_dns(host, &last, ignore_target_hosts, FALSE,
2544     fully_qualified_name);
2545
2546   /* If one or more address records have been found, check that none of them
2547   are local. Since we know the host items all have their IP addresses
2548   inserted, host_scan_for_local_hosts() can only return HOST_FOUND or
2549   HOST_FOUND_LOCAL. We do not need to scan for duplicate IP addresses here,
2550   because set_address_from_dns() removes them. */
2551
2552   if (rc == HOST_FOUND)
2553     rc = host_scan_for_local_hosts(host, &last, removed);
2554   else
2555     if (rc == HOST_IGNORED) rc = HOST_FIND_FAILED;  /* No special action */
2556
2557   DEBUG(D_host_lookup)
2558     {
2559     host_item *h;
2560     if (host->address != NULL)
2561       {
2562       if (fully_qualified_name != NULL)
2563         debug_printf("fully qualified name = %s\n", *fully_qualified_name);
2564       for (h = host; h != last->next; h = h->next)
2565         debug_printf("%s %s mx=%d sort=%d %s\n", h->name,
2566           (h->address == NULL)? US"<null>" : h->address, h->mx, h->sort_key,
2567           (h->status >= hstatus_unusable)? US"*" : US"");
2568       }
2569     }
2570
2571   return rc;
2572   }
2573
2574 /* We have found one or more MX or SRV records. Sort them according to
2575 precedence. Put the data for the first one into the existing host block, and
2576 insert new host_item blocks into the chain for the remainder. For equal
2577 precedences one is supposed to randomize the order. To make this happen, the
2578 sorting is actually done on the MX value * 1000 + a random number. This is put
2579 into a host field called sort_key.
2580
2581 In the case of hosts with both IPv6 and IPv4 addresses, we want to choose the
2582 IPv6 address in preference. At this stage, we don't know what kind of address
2583 the host has. We choose a random number < 500; if later we find an A record
2584 first, we add 500 to the random number. Then for any other address records, we
2585 use random numbers in the range 0-499 for AAAA records and 500-999 for A
2586 records.
2587
2588 At this point we remove any duplicates that point to the same host, retaining
2589 only the one with the lowest precedence. We cannot yet check for precedence
2590 greater than that of the local host, because that test cannot be properly done
2591 until the addresses have been found - an MX record may point to a name for this
2592 host which is not the primary hostname. */
2593
2594 last = NULL;    /* Indicates that not even the first item is filled yet */
2595
2596 for (rr = dns_next_rr(&dnsa, &dnss, RESET_ANSWERS);
2597      rr != NULL;
2598      rr = dns_next_rr(&dnsa, &dnss, RESET_NEXT))
2599   {
2600   int precedence;
2601   int weight = 0;        /* For SRV records */
2602   int port = PORT_NONE;
2603   uschar *s;             /* MUST be unsigned for GETSHORT */
2604   uschar data[256];
2605
2606   if (rr->type != ind_type) continue;
2607   s = rr->data;
2608   GETSHORT(precedence, s);      /* Pointer s is advanced */
2609
2610   /* For MX records, we use a random "weight" which causes multiple records of
2611   the same precedence to sort randomly. */
2612
2613   if (ind_type == T_MX)
2614     {
2615     weight = random_number(500);
2616     }
2617
2618   /* SRV records are specified with a port and a weight. The weight is used
2619   in a special algorithm. However, to start with, we just use it to order the
2620   records of equal priority (precedence). */
2621
2622   else
2623     {
2624     GETSHORT(weight, s);
2625     GETSHORT(port, s);
2626     }
2627
2628   /* Get the name of the host pointed to. */
2629
2630   (void)dn_expand(dnsa.answer, dnsa.answer + dnsa.answerlen, s,
2631     (DN_EXPAND_ARG4_TYPE)data, sizeof(data));
2632
2633   /* Check that we haven't already got this host on the chain; if we have,
2634   keep only the lower precedence. This situation shouldn't occur, but you
2635   never know what junk might get into the DNS (and this case has been seen on
2636   more than one occasion). */
2637
2638   if (last != NULL)       /* This is not the first record */
2639     {
2640     host_item *prev = NULL;
2641
2642     for (h = host; h != last->next; prev = h, h = h->next)
2643       {
2644       if (strcmpic(h->name, data) == 0)
2645         {
2646         DEBUG(D_host_lookup)
2647           debug_printf("discarded duplicate host %s (MX=%d)\n", data,
2648             (precedence > h->mx)? precedence : h->mx);
2649         if (precedence >= h->mx) goto NEXT_MX_RR; /* Skip greater precedence */
2650         if (h == host)                            /* Override first item */
2651           {
2652           h->mx = precedence;
2653           host->sort_key = precedence * 1000 + weight;
2654           goto NEXT_MX_RR;
2655           }
2656
2657         /* Unwanted host item is not the first in the chain, so we can get
2658         get rid of it by cutting it out. */
2659
2660         prev->next = h->next;
2661         if (h == last) last = prev;
2662         break;
2663         }
2664       }
2665     }
2666
2667   /* If this is the first MX or SRV record, put the data into the existing host
2668   block. Otherwise, add a new block in the correct place; if it has to be
2669   before the first block, copy the first block's data to a new second block. */
2670
2671   if (last == NULL)
2672     {
2673     host->name = string_copy_dnsdomain(data);
2674     host->address = NULL;
2675     host->port = port;
2676     host->mx = precedence;
2677     host->sort_key = precedence * 1000 + weight;
2678     host->status = hstatus_unknown;
2679     host->why = hwhy_unknown;
2680     last = host;
2681     }
2682
2683   /* Make a new host item and seek the correct insertion place */
2684
2685   else
2686     {
2687     int sort_key = precedence * 1000 + weight;
2688     host_item *next = store_get(sizeof(host_item));
2689     next->name = string_copy_dnsdomain(data);
2690     next->address = NULL;
2691     next->port = port;
2692     next->mx = precedence;
2693     next->sort_key = sort_key;
2694     next->status = hstatus_unknown;
2695     next->why = hwhy_unknown;
2696     next->last_try = 0;
2697
2698     /* Handle the case when we have to insert before the first item. */
2699
2700     if (sort_key < host->sort_key)
2701       {
2702       host_item htemp;
2703       htemp = *host;
2704       *host = *next;
2705       *next = htemp;
2706       host->next = next;
2707       if (last == host) last = next;
2708       }
2709
2710     /* Else scan down the items we have inserted as part of this exercise;
2711     don't go further. */
2712
2713     else
2714       {
2715       for (h = host; h != last; h = h->next)
2716         {
2717         if (sort_key < h->next->sort_key)
2718           {
2719           next->next = h->next;
2720           h->next = next;
2721           break;
2722           }
2723         }
2724
2725       /* Join on after the last host item that's part of this
2726       processing if we haven't stopped sooner. */
2727
2728       if (h == last)
2729         {
2730         next->next = last->next;
2731         last->next = next;
2732         last = next;
2733         }
2734       }
2735     }
2736
2737   NEXT_MX_RR: continue;
2738   }
2739
2740 /* If the list of hosts was obtained from SRV records, there are two things to
2741 do. First, if there is only one host, and it's name is ".", it means there is
2742 no SMTP service at this domain. Otherwise, we have to sort the hosts of equal
2743 priority according to their weights, using an algorithm that is defined in RFC
2744 2782. The hosts are currently sorted by priority and weight. For each priority
2745 group we have to pick off one host and put it first, and then repeat for any
2746 remaining in the same priority group. */
2747
2748 if (ind_type == T_SRV)
2749   {
2750   host_item **pptr;
2751
2752   if (host == last && host->name[0] == 0)
2753     {
2754     DEBUG(D_host_lookup) debug_printf("the single SRV record is \".\"\n");
2755     return HOST_FIND_FAILED;
2756     }
2757
2758   DEBUG(D_host_lookup)
2759     {
2760     debug_printf("original ordering of hosts from SRV records:\n");
2761     for (h = host; h != last->next; h = h->next)
2762       debug_printf("  %s P=%d W=%d\n", h->name, h->mx, h->sort_key % 1000);
2763     }
2764
2765   for (pptr = &host, h = host; h != last; pptr = &(h->next), h = h->next)
2766     {
2767     int sum = 0;
2768     host_item *hh;
2769
2770     /* Find the last following host that has the same precedence. At the same
2771     time, compute the sum of the weights and the running totals. These can be
2772     stored in the sort_key field. */
2773
2774     for (hh = h; hh != last; hh = hh->next)
2775       {
2776       int weight = hh->sort_key % 1000;   /* was precedence * 1000 + weight */
2777       sum += weight;
2778       hh->sort_key = sum;
2779       if (hh->mx != hh->next->mx) break;
2780       }
2781
2782     /* If there's more than one host at this precedence (priority), we need to
2783     pick one to go first. */
2784
2785     if (hh != h)
2786       {
2787       host_item *hhh;
2788       host_item **ppptr;
2789       int randomizer = random_number(sum + 1);
2790
2791       for (ppptr = pptr, hhh = h;
2792            hhh != hh;
2793            ppptr = &(hhh->next), hhh = hhh->next)
2794         {
2795         if (hhh->sort_key >= randomizer) break;
2796         }
2797
2798       /* hhh now points to the host that should go first; ppptr points to the
2799       place that points to it. Unfortunately, if the start of the minilist is
2800       the start of the entire list, we can't just swap the items over, because
2801       we must not change the value of host, since it is passed in from outside.
2802       One day, this could perhaps be changed.
2803
2804       The special case is fudged by putting the new item *second* in the chain,
2805       and then transferring the data between the first and second items. We
2806       can't just swap the first and the chosen item, because that would mean
2807       that an item with zero weight might no longer be first. */
2808
2809       if (hhh != h)
2810         {
2811         *ppptr = hhh->next;          /* Cuts it out of the chain */
2812
2813         if (h == host)
2814           {
2815           host_item temp = *h;
2816           *h = *hhh;
2817           *hhh = temp;
2818           hhh->next = temp.next;
2819           h->next = hhh;
2820           }
2821
2822         else
2823           {
2824           hhh->next = h;               /* The rest of the chain follows it */
2825           *pptr = hhh;                 /* It takes the place of h */
2826           h = hhh;                     /* It's now the start of this minilist */
2827           }
2828         }
2829       }
2830
2831     /* A host has been chosen to be first at this priority and h now points
2832     to this host. There may be others at the same priority, or others at a
2833     different priority. Before we leave this host, we need to put back a sort
2834     key of the traditional MX kind, in case this host is multihomed, because
2835     the sort key is used for ordering the multiple IP addresses. We do not need
2836     to ensure that these new sort keys actually reflect the order of the hosts,
2837     however. */
2838
2839     h->sort_key = h->mx * 1000 + random_number(500);
2840     }   /* Move on to the next host */
2841   }
2842
2843 /* Now we have to find IP addresses for all the hosts. We have ensured above
2844 that the names in all the host items are unique. Before release 4.61 we used to
2845 process records from the additional section in the DNS packet that returned the
2846 MX or SRV records. However, a DNS name server is free to drop any resource
2847 records from the additional section. In theory, this has always been a
2848 potential problem, but it is exacerbated by the advent of IPv6. If a host had
2849 several IPv4 addresses and some were not in the additional section, at least
2850 Exim would try the others. However, if a host had both IPv4 and IPv6 addresses
2851 and all the IPv4 (say) addresses were absent, Exim would try only for a IPv6
2852 connection, and never try an IPv4 address. When there was only IPv4
2853 connectivity, this was a disaster that did in practice occur.
2854
2855 So, from release 4.61 onwards, we always search for A and AAAA records
2856 explicitly. The names shouldn't point to CNAMES, but we use the general lookup
2857 function that handles them, just in case. If any lookup gives a soft error,
2858 change the default yield.
2859
2860 For these DNS lookups, we must disable qualify_single and search_parents;
2861 otherwise invalid host names obtained from MX or SRV records can cause trouble
2862 if they happen to match something local. */
2863
2864 yield = HOST_FIND_FAILED;    /* Default yield */
2865 dns_init(FALSE, FALSE);      /* Disable qualify_single and search_parents */
2866
2867 for (h = host; h != last->next; h = h->next)
2868   {
2869   if (h->address != NULL) continue;  /* Inserted by a multihomed host */
2870   rc = set_address_from_dns(h, &last, ignore_target_hosts, allow_mx_to_ip, NULL);
2871   if (rc != HOST_FOUND)
2872     {
2873     h->status = hstatus_unusable;
2874     if (rc == HOST_FIND_AGAIN)
2875       {
2876       yield = rc;
2877       h->why = hwhy_deferred;
2878       }
2879     else
2880       h->why = (rc == HOST_IGNORED)? hwhy_ignored : hwhy_failed;
2881     }
2882   }
2883
2884 /* Scan the list for any hosts that are marked unusable because they have
2885 been explicitly ignored, and remove them from the list, as if they did not
2886 exist. If we end up with just a single, ignored host, flatten its fields as if
2887 nothing was found. */
2888
2889 if (ignore_target_hosts != NULL)
2890   {
2891   host_item *prev = NULL;
2892   for (h = host; h != last->next; h = h->next)
2893     {
2894     REDO:
2895     if (h->why != hwhy_ignored)        /* Non ignored host, just continue */
2896       prev = h;
2897     else if (prev == NULL)             /* First host is ignored */
2898       {
2899       if (h != last)                   /* First is not last */
2900         {
2901         if (h->next == last) last = h; /* Overwrite it with next */
2902         *h = *(h->next);               /* and reprocess it. */
2903         goto REDO;                     /* C should have redo, like Perl */
2904         }
2905       }
2906     else                               /* Ignored host is not first - */
2907       {                                /*   cut it out */
2908       prev->next = h->next;
2909       if (h == last) last = prev;
2910       }
2911     }
2912
2913   if (host->why == hwhy_ignored) host->address = NULL;
2914   }
2915
2916 /* There is still one complication in the case of IPv6. Although the code above
2917 arranges that IPv6 addresses take precedence over IPv4 addresses for multihomed
2918 hosts, it doesn't do this for addresses that apply to different hosts with the
2919 same MX precedence, because the sorting on MX precedence happens first. So we
2920 have to make another pass to check for this case. We ensure that, within a
2921 single MX preference value, IPv6 addresses come first. This can separate the
2922 addresses of a multihomed host, but that should not matter. */
2923
2924 #if HAVE_IPV6
2925 if (h != last && !disable_ipv6)
2926   {
2927   for (h = host; h != last; h = h->next)
2928     {
2929     host_item temp;
2930     host_item *next = h->next;
2931     if (h->mx != next->mx ||                   /* If next is different MX */
2932         h->address == NULL ||                  /* OR this one is unset */
2933         Ustrchr(h->address, ':') != NULL ||    /* OR this one is IPv6 */
2934         (next->address != NULL &&
2935          Ustrchr(next->address, ':') == NULL)) /* OR next is IPv4 */
2936       continue;                                /* move on to next */
2937     temp = *h;                                 /* otherwise, swap */
2938     temp.next = next->next;
2939     *h = *next;
2940     h->next = next;
2941     *next = temp;
2942     }
2943   }
2944 #endif
2945
2946 /* Remove any duplicate IP addresses and then scan the list of hosts for any
2947 whose IP addresses are on the local host. If any are found, all hosts with the
2948 same or higher MX values are removed. However, if the local host has the lowest
2949 numbered MX, then HOST_FOUND_LOCAL is returned. Otherwise, if at least one host
2950 with an IP address is on the list, HOST_FOUND is returned. Otherwise,
2951 HOST_FIND_FAILED is returned, but in this case do not update the yield, as it
2952 might have been set to HOST_FIND_AGAIN just above here. If not, it will already
2953 be HOST_FIND_FAILED. */
2954
2955 host_remove_duplicates(host, &last);
2956 rc = host_scan_for_local_hosts(host, &last, removed);
2957 if (rc != HOST_FIND_FAILED) yield = rc;
2958
2959 DEBUG(D_host_lookup)
2960   {
2961   if (fully_qualified_name != NULL)
2962     debug_printf("fully qualified name = %s\n", *fully_qualified_name);
2963   debug_printf("host_find_bydns yield = %s (%d); returned hosts:\n",
2964     (yield == HOST_FOUND)? "HOST_FOUND" :
2965     (yield == HOST_FOUND_LOCAL)? "HOST_FOUND_LOCAL" :
2966     (yield == HOST_FIND_AGAIN)? "HOST_FIND_AGAIN" :
2967     (yield == HOST_FIND_FAILED)? "HOST_FIND_FAILED" : "?",
2968     yield);
2969   for (h = host; h != last->next; h = h->next)
2970     {
2971     debug_printf("  %s %s MX=%d ", h->name,
2972       (h->address == NULL)? US"<null>" : h->address, h->mx);
2973     if (h->port != PORT_NONE) debug_printf("port=%d ", h->port);
2974     if (h->status >= hstatus_unusable) debug_printf("*");
2975     debug_printf("\n");
2976     }
2977   }
2978
2979 return yield;
2980 }
2981
2982
2983
2984
2985 /*************************************************
2986 **************************************************
2987 *             Stand-alone test program           *
2988 **************************************************
2989 *************************************************/
2990
2991 #ifdef STAND_ALONE
2992
2993 int main(int argc, char **cargv)
2994 {
2995 host_item h;
2996 int whichrrs = HOST_FIND_BY_MX | HOST_FIND_BY_A;
2997 BOOL byname = FALSE;
2998 BOOL qualify_single = TRUE;
2999 BOOL search_parents = FALSE;
3000 uschar **argv = USS cargv;
3001 uschar buffer[256];
3002
3003 disable_ipv6 = FALSE;
3004 primary_hostname = US"";
3005 store_pool = POOL_MAIN;
3006 debug_selector = D_host_lookup|D_interface;
3007 debug_file = stdout;
3008 debug_fd = fileno(debug_file);
3009
3010 printf("Exim stand-alone host functions test\n");
3011
3012 host_find_interfaces();
3013 debug_selector = D_host_lookup | D_dns;
3014
3015 if (argc > 1) primary_hostname = argv[1];
3016
3017 /* So that debug level changes can be done first */
3018
3019 dns_init(qualify_single, search_parents);
3020
3021 printf("Testing host lookup\n");
3022 printf("> ");
3023 while (Ufgets(buffer, 256, stdin) != NULL)
3024   {
3025   int rc;
3026   int len = Ustrlen(buffer);
3027   uschar *fully_qualified_name;
3028
3029   while (len > 0 && isspace(buffer[len-1])) len--;
3030   buffer[len] = 0;
3031
3032   if (Ustrcmp(buffer, "q") == 0) break;
3033
3034   if (Ustrcmp(buffer, "byname") == 0) byname = TRUE;
3035   else if (Ustrcmp(buffer, "no_byname") == 0) byname = FALSE;
3036   else if (Ustrcmp(buffer, "a_only") == 0) whichrrs = HOST_FIND_BY_A;
3037   else if (Ustrcmp(buffer, "mx_only") == 0) whichrrs = HOST_FIND_BY_MX;
3038   else if (Ustrcmp(buffer, "srv_only") == 0) whichrrs = HOST_FIND_BY_SRV;
3039   else if (Ustrcmp(buffer, "srv+a") == 0)
3040     whichrrs = HOST_FIND_BY_SRV | HOST_FIND_BY_A;
3041   else if (Ustrcmp(buffer, "srv+mx") == 0)
3042     whichrrs = HOST_FIND_BY_SRV | HOST_FIND_BY_MX;
3043   else if (Ustrcmp(buffer, "srv+mx+a") == 0)
3044     whichrrs = HOST_FIND_BY_SRV | HOST_FIND_BY_MX | HOST_FIND_BY_A;
3045   else if (Ustrcmp(buffer, "qualify_single") == 0) qualify_single = TRUE;
3046   else if (Ustrcmp(buffer, "no_qualify_single") == 0) qualify_single = FALSE;
3047   else if (Ustrcmp(buffer, "search_parents") == 0) search_parents = TRUE;
3048   else if (Ustrcmp(buffer, "no_search_parents") == 0) search_parents = FALSE;
3049   else if (Ustrcmp(buffer, "test_harness") == 0)
3050     running_in_test_harness = !running_in_test_harness;
3051   else if (Ustrcmp(buffer, "ipv6") == 0) disable_ipv6 = !disable_ipv6;
3052   else if (Ustrcmp(buffer, "res_debug") == 0)
3053     {
3054     _res.options ^= RES_DEBUG;
3055     }
3056   else if (Ustrncmp(buffer, "retrans", 7) == 0)
3057     {
3058     (void)sscanf(CS(buffer+8), "%d", &dns_retrans);
3059     _res.retrans = dns_retrans;
3060     }
3061   else if (Ustrncmp(buffer, "retry", 5) == 0)
3062     {
3063     (void)sscanf(CS(buffer+6), "%d", &dns_retry);
3064     _res.retry = dns_retry;
3065     }
3066   else
3067     {
3068     int flags = whichrrs;
3069
3070     h.name = buffer;
3071     h.next = NULL;
3072     h.mx = MX_NONE;
3073     h.port = PORT_NONE;
3074     h.status = hstatus_unknown;
3075     h.why = hwhy_unknown;
3076     h.address = NULL;
3077
3078     if (qualify_single) flags |= HOST_FIND_QUALIFY_SINGLE;
3079     if (search_parents) flags |= HOST_FIND_SEARCH_PARENTS;
3080
3081     rc = byname?
3082       host_find_byname(&h, NULL, flags, &fully_qualified_name, TRUE)
3083       :
3084       host_find_bydns(&h, NULL, flags, US"smtp", NULL, NULL,
3085         &fully_qualified_name, NULL);
3086
3087     if (rc == HOST_FIND_FAILED) printf("Failed\n");
3088       else if (rc == HOST_FIND_AGAIN) printf("Again\n");
3089         else if (rc == HOST_FOUND_LOCAL) printf("Local\n");
3090     }
3091
3092   printf("\n> ");
3093   }
3094
3095 printf("Testing host_aton\n");
3096 printf("> ");
3097 while (Ufgets(buffer, 256, stdin) != NULL)
3098   {
3099   int i;
3100   int x[4];
3101   int len = Ustrlen(buffer);
3102
3103   while (len > 0 && isspace(buffer[len-1])) len--;
3104   buffer[len] = 0;
3105
3106   if (Ustrcmp(buffer, "q") == 0) break;
3107
3108   len = host_aton(buffer, x);
3109   printf("length = %d ", len);
3110   for (i = 0; i < len; i++)
3111     {
3112     printf("%04x ", (x[i] >> 16) & 0xffff);
3113     printf("%04x ", x[i] & 0xffff);
3114     }
3115   printf("\n> ");
3116   }
3117
3118 printf("\n");
3119
3120 printf("Testing host_name_lookup\n");
3121 printf("> ");
3122 while (Ufgets(buffer, 256, stdin) != NULL)
3123   {
3124   int len = Ustrlen(buffer);
3125   while (len > 0 && isspace(buffer[len-1])) len--;
3126   buffer[len] = 0;
3127   if (Ustrcmp(buffer, "q") == 0) break;
3128   sender_host_address = buffer;
3129   sender_host_name = NULL;
3130   sender_host_aliases = NULL;
3131   host_lookup_msg = US"";
3132   host_lookup_failed = FALSE;
3133   if (host_name_lookup() == FAIL)  /* Debug causes printing */
3134     printf("Lookup failed:%s\n", host_lookup_msg);
3135   printf("\n> ");
3136   }
3137
3138 printf("\n");
3139
3140 return 0;
3141 }
3142 #endif  /* STAND_ALONE */
3143
3144 /* End of host.c */