citadel/locate_host.c

   1 /*
   2  * $Id$
   3  *
   4  * locate the originating host
   5  *
   6  */
   7
   8 #include "sysdep.h"
   9 #include <stdlib.h>
  10 #include <unistd.h>
  11 #include <stdio.h>
  12 #include <ctype.h>
  13 #include <signal.h>
  14 #include <sys/types.h>
  15 #include <sys/socket.h>
  16 #include <netinet/in.h>
  17 #include <arpa/inet.h>
  18 #include <limits.h>
  19 #include <netdb.h>
  20 #include <string.h>
  21 #include <errno.h>
  22 #include <libcitadel.h>
  23 #include "citadel.h"
  24 #include "server.h"
  25 #include "locate_host.h"
  26 #include "sysdep_decls.h"
  27 #include "config.h"
  28 #include "domain.h"
  29 #include "ctdl_module.h"
  30
  31 #ifdef HAVE_RESOLV_H
  32 #include <arpa/nameser.h>
  33 #ifdef HAVE_ARPA_NAMESER_COMPAT_H
  34 #include <arpa/nameser_compat.h>
  35 #endif
  36 #include <resolv.h>
  37 #endif
  38
  39
  40 /* Hacks to work around nameser.h declarations missing on OpenBSD
  41  * see also: http://search.cpan.org/src/MIKER/Net-DNS-ToolKit-0.30/ToolKit.h
  42  */
  43
  44 #ifndef NS_INT16SZ
  45 # ifdef INT16SZ
  46 #  define NS_INT16SZ INT16SZ
  47 # endif
  48 #endif
  49
  50 #ifndef NS_INT32SZ
  51 # ifdef INT32SZ
  52 #  define NS_INT32SZ INT32SZ
  53 # endif
  54 #endif
  55
  56 #ifndef NS_GET16
  57 # ifdef GETSHORT
  58 #  define NS_GET16 GETSHORT
  59 # endif
  60 #endif
  61
  62
  63 /***************************************************************************/
  64
  65
  66 void locate_host(char *tbuf, size_t n,
  67                 char *abuf, size_t na,
  68                 const struct in_addr *addr)
  69 {
  70         struct hostent *ch;
  71         const char *i;
  72         char *j;
  73         int a1, a2, a3, a4;
  74         char address_string[SIZ];
  75
  76
  77 #ifdef HAVE_NONREENTRANT_NETDB
  78         begin_critical_section(S_NETDB);
  79 #endif
  80
  81         i = (const char *) addr;
  82         a1 = ((*i++) & 0xff);
  83         a2 = ((*i++) & 0xff);
  84         a3 = ((*i++) & 0xff);
  85         a4 = ((*i++) & 0xff);
  86         sprintf(address_string, "%d.%d.%d.%d", a1, a2, a3, a4);
  87
  88         if (abuf != NULL) {
  89                 safestrncpy(abuf, address_string, na);
  90         }
  91
  92         if ((ch = gethostbyaddr((const char *) addr,
  93            sizeof(*addr), AF_INET)) == NULL) {
  94 bad_dns:
  95                 safestrncpy(tbuf, address_string, n);
  96                 goto end;       /* because we might need to end the critical
  97                                    section */
  98         }
  99         /* check if the forward DNS agrees; if not, they're spoofing */
 100         j = strdup(ch->h_name);
 101         ch = gethostbyname(j);
 102         free(j);
 103         if (ch == NULL)
 104                 goto bad_dns;
 105
 106         /* check address for consistency */
 107         for (; *ch->h_addr_list; ch->h_addr_list++)
 108                 if (!memcmp(*ch->h_addr_list, addr,
 109                             sizeof *addr)) {
 110                         safestrncpy(tbuf, ch->h_name, 63);
 111                         goto end;
 112                 }
 113         goto bad_dns;           /* they were spoofing. report a numeric IP
 114                                    address. */
 115
 116       end:
 117
 118 #ifdef HAVE_NONREENTRANT_NETDB
 119         end_critical_section(S_NETDB);
 120 #endif
 121
 122         tbuf[63] = 0;
 123 }
 124
 125
 126 /*
 127  * RBL check written by Edward S. Marshall [http://rblcheck.sourceforge.net]
 128  */
 129 #define RESULT_SIZE 4096 /* What is the longest result text we support? */
 130 int rblcheck_backend(char *domain, char *txtbuf, int txtbufsize) {
 131         int a, b, c;
 132         char *result = NULL;
 133         u_char fixedans[ PACKETSZ ];
 134         u_char *answer;
 135         int need_to_free_answer = 0;
 136         const u_char *cp;
 137         u_char *rp;
 138         const u_char *cend;
 139         const u_char *rend;
 140         int len;
 141         char *p = NULL;
 142
 143         /* Make our DNS query. */
 144         //res_init();
 145         answer = fixedans;
 146         if (CtdlThreadCheckStop())
 147         {
 148                 if (txtbuf != NULL)
 149                         snprintf(txtbuf, txtbufsize, "System shutting down");
 150                 return (1);
 151         }
 152         len = res_query( domain, C_IN, T_A, answer, PACKETSZ );
 153
 154         /* Was there a problem? If so, the domain doesn't exist. */
 155         if( len == -1 ) {
 156                 if (txtbuf != NULL) {
 157                         strcpy(txtbuf, "");
 158                 }
 159                 return(0);
 160         }
 161
 162         if( len > PACKETSZ )
 163         {
 164                 answer = malloc( len );
 165                 need_to_free_answer = 1;
 166                 len = res_query( domain, C_IN, T_A, answer, len );
 167                 if( len == -1 ) {
 168                         if (txtbuf != NULL) {
 169                                 snprintf(txtbuf, txtbufsize,
 170                                         "Message rejected due to known spammer source IP address");
 171                         }
 172                         if (need_to_free_answer) free(answer);
 173                         return(1);
 174                 }
 175         }
 176         if (CtdlThreadCheckStop())
 177         {
 178                 if (txtbuf != NULL)
 179                         snprintf(txtbuf, txtbufsize, "System shutting down");
 180                 if (need_to_free_answer) free(answer);
 181                 return (1);
 182         }
 183
 184         result = ( char * )malloc( RESULT_SIZE );
 185         result[ 0 ] = '\0';
 186
 187
 188         /* Make another DNS query for textual data; this shouldn't
 189            be a performance hit, since it'll now be cached at the
 190            nameserver we're using. */
 191         res_init();
 192         len = res_query( domain, C_IN, T_TXT, answer, PACKETSZ );
 193         if (CtdlThreadCheckStop())
 194         {
 195                 if (txtbuf != NULL)
 196                         snprintf(txtbuf, txtbufsize, "System shutting down");
 197                 if (need_to_free_answer) free(answer);
 198                 free(result);
 199                 return (1);
 200         }
 201
 202         /* Just in case there's no TXT record... */
 203         if( len == -1 )
 204         {
 205                 if (txtbuf != NULL) {
 206                         snprintf(txtbuf, txtbufsize,
 207                                 "Message rejected due to known spammer source IP address");
 208                 }
 209                 if (need_to_free_answer) free(answer);
 210                 free(result);
 211                 return(1);
 212         }
 213
 214         /* Skip the header and the address we queried. */
 215         cp = answer + sizeof( HEADER );
 216         while( *cp != '\0' )
 217         {
 218                 a = *cp++;
 219                 while( a-- )
 220                         cp++;
 221         }
 222
 223         /* This seems to be a bit of magic data that we need to
 224            skip. I wish there were good online documentation
 225            for programming for libresolv, so I'd know what I'm
 226            skipping here. Anyone reading this, feel free to
 227            enlighten me. */
 228         cp += 1 + NS_INT16SZ + NS_INT32SZ;
 229
 230         /* Skip the type, class and ttl. */
 231         cp += ( NS_INT16SZ * 2 ) + NS_INT32SZ;
 232
 233         /* Get the length and end of the buffer. */
 234         NS_GET16( c, cp );
 235         cend = cp + c;
 236
 237         /* Iterate over any multiple answers we might have. In
 238            this context, it's unlikely, but anyway. */
 239         rp = (u_char *) result;
 240         rend = (u_char *) result + RESULT_SIZE - 1;
 241         while( cp < cend && rp < rend )
 242         {
 243                 a = *cp++;
 244                 if( a != 0 )
 245                         for( b = a; b > 0 && cp < cend && rp < rend;
 246                           b-- )
 247                         {
 248                                 if( *cp == '\n' || *cp == '"' ||
 249                                   *cp == '\\' )
 250                                 {
 251                                         *rp++ = '\\';
 252                                 }
 253                                 *rp++ = *cp++;
 254                         }
 255         }
 256         *rp = '\0';
 257         if (txtbuf != NULL) {
 258                 snprintf(txtbuf, txtbufsize, "%s", result);
 259         }
 260         /* Remove nonprintable characters */
 261         for (p=txtbuf; *p; ++p) {
 262                 if (!isprint(*p)) strcpy(p, p+1);
 263         }
 264         if (need_to_free_answer) free(answer);
 265         free(result);
 266         return(1);
 267 }
 268
 269
 270 /*
 271  * Check to see if a host is on some sort of spam list (RBL)
 272  * If spammer, returns nonzero and places reason in 'message_to_spammer'
 273  */
 274 int rbl_check_addr(struct in_addr *addr, char *message_to_spammer)
 275 {
 276         int a1, a2, a3, a4;
 277         char tbuf[256];
 278         int rbl;
 279         int num_rbl;
 280         char rbl_domains[SIZ];
 281         char txt_answer[1024];
 282         char dotted_quad[32];
 283
 284         strcpy(message_to_spammer, "ok");
 285         safestrncpy(dotted_quad, inet_ntoa(*addr), sizeof dotted_quad);
 286         sscanf(dotted_quad, "%d.%d.%d.%d", &a1, &a2, &a3, &a4);
 287
 288         /* See if we have any RBL domains configured */
 289         num_rbl = get_hosts(rbl_domains, "rbl");
 290         if (num_rbl < 1) return(0);
 291
 292         /* Try all configured RBL's */
 293         for (rbl=0; rbl<num_rbl; ++rbl) {
 294                 snprintf(tbuf, sizeof tbuf,
 295                         "%d.%d.%d.%d.",
 296                         a4, a3, a2, a1);
 297                 extract_token(&tbuf[strlen(tbuf)], rbl_domains, rbl, '|', (sizeof tbuf - strlen(tbuf)));
 298
 299                 if (rblcheck_backend(tbuf, txt_answer, sizeof txt_answer)) {
 300                         strcpy(message_to_spammer, txt_answer);
 301                         CtdlLogPrintf(CTDL_INFO, "RBL: %s\n", txt_answer);
 302                         return(1);
 303                 }
 304         }
 305
 306         return(0);
 307 }
 308
 309
 310 /*
 311  * Check to see if the client host is on some sort of spam list (RBL)
 312  * If spammer, returns nonzero and places reason in 'message_to_spammer'
 313  *
 314  * PORTABILITY NOTE!  I've made my best effort to rewrite this in a portable fashion.
 315  * If anyone makes changes to this function, please shout-out so we can test it to
 316  * make sure it didn't break on Linux!
 317  */
 318 int rbl_check(char *message_to_spammer) {
 319         int r;
 320         struct sockaddr_in peer;
 321         socklen_t peer_len = 0;
 322
 323         peer_len = sizeof(peer);
 324         r = getpeername(CC->client_socket, &peer, &peer_len);
 325         if (r == 0) {
 326                 return(rbl_check_addr(&peer.sin_addr, message_to_spammer));
 327         }
 328         else {
 329                 CtdlLogPrintf(CTDL_INFO, "RBL getpeername() failed: %s\n", strerror(errno));
 330         }
 331         return(0);
 332 }
 333
 334 /*
 335  * Convert a host name to a dotted quad address.
 336  * Returns zero on success or nonzero on failure.
 337  */
 338 int hostname_to_dotted_quad(char *addr, char *host) {
 339         struct hostent *ch;
 340         const char *i;
 341         int a1, a2, a3, a4;
 342
 343         ch = gethostbyname(host);
 344         if (ch == NULL) {
 345                 strcpy(addr, "0.0.0.0");
 346                 return(1);
 347         }
 348
 349         i = (const char *) ch->h_addr_list[0];
 350         a1 = ((*i++) & 0xff);
 351         a2 = ((*i++) & 0xff);
 352         a3 = ((*i++) & 0xff);
 353         a4 = ((*i++) & 0xff);
 354         sprintf(addr, "%d.%d.%d.%d", a1, a2, a3, a4);
 355         return(0);
 356 }