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 <limits.h>
  18 #include <netdb.h>
  19 #include <string.h>
  20 #include <libcitadel.h>
  21 #include "citadel.h"
  22 #include "server.h"
  23 #include "locate_host.h"
  24 #include "sysdep_decls.h"
  25 #include "config.h"
  26 #include "domain.h"
  27 #include "ctdl_module.h"
  28
  29 #ifdef HAVE_RESOLV_H
  30 #include <arpa/nameser.h>
  31 #ifdef HAVE_ARPA_NAMESER_COMPAT_H
  32 #include <arpa/nameser_compat.h>
  33 #endif
  34 #include <resolv.h>
  35 #endif
  36
  37
  38 /* Hacks to work around nameser.h declarations missing on OpenBSD
  39  * see also: http://search.cpan.org/src/MIKER/Net-DNS-ToolKit-0.30/ToolKit.h
  40  */
  41
  42 #ifndef NS_INT16SZ
  43 # ifdef INT16SZ
  44 #  define NS_INT16SZ INT16SZ
  45 # endif
  46 #endif
  47
  48 #ifndef NS_INT32SZ
  49 # ifdef INT32SZ
  50 #  define NS_INT32SZ INT32SZ
  51 # endif
  52 #endif
  53
  54 #ifndef NS_GET16
  55 # ifdef GETSHORT
  56 #  define NS_GET16 GETSHORT
  57 # endif
  58 #endif
  59
  60
  61 /***************************************************************************/
  62
  63
  64 void locate_host(char *tbuf, size_t n,
  65                 char *abuf, size_t na,
  66                 const struct in_addr *addr)
  67 {
  68         struct hostent *ch;
  69         const char *i;
  70         char *j;
  71         int a1, a2, a3, a4;
  72         char address_string[SIZ];
  73
  74
  75 #ifdef HAVE_NONREENTRANT_NETDB
  76         begin_critical_section(S_NETDB);
  77 #endif
  78
  79         i = (const char *) addr;
  80         a1 = ((*i++) & 0xff);
  81         a2 = ((*i++) & 0xff);
  82         a3 = ((*i++) & 0xff);
  83         a4 = ((*i++) & 0xff);
  84         sprintf(address_string, "%d.%d.%d.%d", a1, a2, a3, a4);
  85
  86         if (abuf != NULL) {
  87                 safestrncpy(abuf, address_string, na);
  88         }
  89
  90         if ((ch = gethostbyaddr((const char *) addr,
  91            sizeof(*addr), AF_INET)) == NULL) {
  92 bad_dns:
  93                 safestrncpy(tbuf, address_string, n);
  94                 goto end;       /* because we might need to end the critical
  95                                    section */
  96         }
  97         /* check if the forward DNS agrees; if not, they're spoofing */
  98         j = strdup(ch->h_name);
  99         ch = gethostbyname(j);
 100         free(j);
 101         if (ch == NULL)
 102                 goto bad_dns;
 103
 104         /* check address for consistency */
 105         for (; *ch->h_addr_list; ch->h_addr_list++)
 106                 if (!memcmp(*ch->h_addr_list, addr,
 107                             sizeof *addr)) {
 108                         safestrncpy(tbuf, ch->h_name, 63);
 109                         goto end;
 110                 }
 111         goto bad_dns;           /* they were spoofing. report a numeric IP
 112                                    address. */
 113
 114       end:
 115
 116 #ifdef HAVE_NONREENTRANT_NETDB
 117         end_critical_section(S_NETDB);
 118 #endif
 119
 120         tbuf[63] = 0;
 121 }
 122
 123
 124 /*
 125  * RBL check written by Edward S. Marshall [http://rblcheck.sourceforge.net]
 126  */
 127 #define RESULT_SIZE 4096 /* What is the longest result text we support? */
 128 int rblcheck_backend(char *domain, char *txtbuf, int txtbufsize) {
 129         int a, b, c;
 130         char *result = NULL;
 131         u_char fixedans[ PACKETSZ ];
 132         u_char *answer;
 133         int need_to_free_answer = 0;
 134         const u_char *cp;
 135         u_char *rp;
 136         const u_char *cend;
 137         const u_char *rend;
 138         int len;
 139         char *p = NULL;
 140
 141         /* Make our DNS query. */
 142         //res_init();
 143         answer = fixedans;
 144         if (CtdlThreadCheckStop())
 145         {
 146                 if (txtbuf != NULL)
 147                         snprintf(txtbuf, txtbufsize, "System shutting down");
 148                 return (1);
 149         }
 150         len = res_query( domain, C_IN, T_A, answer, PACKETSZ );
 151
 152         /* Was there a problem? If so, the domain doesn't exist. */
 153         if( len == -1 ) {
 154                 if (txtbuf != NULL) {
 155                         strcpy(txtbuf, "");
 156                 }
 157                 return(0);
 158         }
 159
 160         if( len > PACKETSZ )
 161         {
 162                 answer = malloc( len );
 163                 need_to_free_answer = 1;
 164                 len = res_query( domain, C_IN, T_A, answer, len );
 165                 if( len == -1 ) {
 166                         if (txtbuf != NULL) {
 167                                 snprintf(txtbuf, txtbufsize,
 168                                         "Message rejected due to known spammer source IP address");
 169                         }
 170                         if (need_to_free_answer) free(answer);
 171                         return(1);
 172                 }
 173         }
 174         if (CtdlThreadCheckStop())
 175         {
 176                 if (txtbuf != NULL)
 177                         snprintf(txtbuf, txtbufsize, "System shutting down");
 178                 if (need_to_free_answer) free(answer);
 179                 return (1);
 180         }
 181
 182         result = ( char * )malloc( RESULT_SIZE );
 183         result[ 0 ] = '\0';
 184
 185
 186         /* Make another DNS query for textual data; this shouldn't
 187            be a performance hit, since it'll now be cached at the
 188            nameserver we're using. */
 189         res_init();
 190         len = res_query( domain, C_IN, T_TXT, answer, PACKETSZ );
 191         if (CtdlThreadCheckStop())
 192         {
 193                 if (txtbuf != NULL)
 194                         snprintf(txtbuf, txtbufsize, "System shutting down");
 195                 if (need_to_free_answer) free(answer);
 196                 free(result);
 197                 return (1);
 198         }
 199
 200         /* Just in case there's no TXT record... */
 201         if( len == -1 )
 202         {
 203                 if (txtbuf != NULL) {
 204                         snprintf(txtbuf, txtbufsize,
 205                                 "Message rejected due to known spammer source IP address");
 206                 }
 207                 if (need_to_free_answer) free(answer);
 208                 free(result);
 209                 return(1);
 210         }
 211
 212         /* Skip the header and the address we queried. */
 213         cp = answer + sizeof( HEADER );
 214         while( *cp != '\0' )
 215         {
 216                 a = *cp++;
 217                 while( a-- )
 218                         cp++;
 219         }
 220
 221         /* This seems to be a bit of magic data that we need to
 222            skip. I wish there were good online documentation
 223            for programming for libresolv, so I'd know what I'm
 224            skipping here. Anyone reading this, feel free to
 225            enlighten me. */
 226         cp += 1 + NS_INT16SZ + NS_INT32SZ;
 227
 228         /* Skip the type, class and ttl. */
 229         cp += ( NS_INT16SZ * 2 ) + NS_INT32SZ;
 230
 231         /* Get the length and end of the buffer. */
 232         NS_GET16( c, cp );
 233         cend = cp + c;
 234
 235         /* Iterate over any multiple answers we might have. In
 236            this context, it's unlikely, but anyway. */
 237         rp = (u_char *) result;
 238         rend = (u_char *) result + RESULT_SIZE - 1;
 239         while( cp < cend && rp < rend )
 240         {
 241                 a = *cp++;
 242                 if( a != 0 )
 243                         for( b = a; b > 0 && cp < cend && rp < rend;
 244                           b-- )
 245                         {
 246                                 if( *cp == '\n' || *cp == '"' ||
 247                                   *cp == '\\' )
 248                                 {
 249                                         *rp++ = '\\';
 250                                 }
 251                                 *rp++ = *cp++;
 252                         }
 253         }
 254         *rp = '\0';
 255         if (txtbuf != NULL) {
 256                 snprintf(txtbuf, txtbufsize, "%s", result);
 257         }
 258         /* Remove nonprintable characters */
 259         for (p=txtbuf; *p; ++p) {
 260                 if (!isprint(*p)) strcpy(p, p+1);
 261         }
 262         if (need_to_free_answer) free(answer);
 263         free(result);
 264         return(1);
 265 }
 266
 267
 268 /*
 269  * Check to see if a host is on some sort of spam list (RBL)
 270  * If spammer, returns nonzero and places reason in 'message_to_spammer'
 271  */
 272 int rbl_check_addr(struct in_addr *addr, char *message_to_spammer)
 273 {
 274         const char *i;
 275         int a1, a2, a3, a4;
 276         char tbuf[256];
 277         int rbl;
 278         int num_rbl;
 279         char rbl_domains[SIZ];
 280         char txt_answer[1024];
 281
 282         strcpy(message_to_spammer, "ok");
 283
 284         i = (const char *) addr;
 285         a1 = ((*i++) & 0xff);
 286         a2 = ((*i++) & 0xff);
 287         a3 = ((*i++) & 0xff);
 288         a4 = ((*i++) & 0xff);
 289
 290         /* See if we have any RBL domains configured */
 291         num_rbl = get_hosts(rbl_domains, "rbl");
 292         if (num_rbl < 1) return(0);
 293
 294         /* Try all configured RBL's */
 295         for (rbl=0; rbl<num_rbl; ++rbl) {
 296                 snprintf(tbuf, sizeof tbuf,
 297                         "%d.%d.%d.%d.",
 298                         a4, a3, a2, a1);
 299                 extract_token(&tbuf[strlen(tbuf)], rbl_domains, rbl, '|', (sizeof tbuf - strlen(tbuf)));
 300
 301                 if (rblcheck_backend(tbuf, txt_answer, sizeof txt_answer)) {
 302                         sprintf(message_to_spammer, "5.7.1 %s", txt_answer);
 303                         CtdlLogPrintf(CTDL_INFO, "RBL: %s\n", txt_answer);
 304                         return(1);
 305                 }
 306         }
 307
 308         return(0);
 309 }
 310
 311
 312 /*
 313  * Check to see if the client host is on some sort of spam list (RBL)
 314  * If spammer, returns nonzero and places reason in 'message_to_spammer'
 315  */
 316 int rbl_check(char *message_to_spammer) {
 317         struct sockaddr_in sin;
 318         int len;        /* should be socklen_t but doesn't work on Macintosh */
 319         struct timeval tv1, tv2;
 320         suseconds_t total_time = 0;
 321
 322         gettimeofday(&tv1, NULL);
 323         len = 0;
 324         memset (&sin, 0, sizeof (struct sockaddr_in));
 325         if (!getpeername(CC->client_socket, (struct sockaddr *) &sin, (socklen_t *)&len)) {
 326                 return(rbl_check_addr(&sin.sin_addr, message_to_spammer));
 327         }
 328
 329         gettimeofday(&tv2, NULL);
 330         total_time = (tv2.tv_usec + (tv2.tv_sec * 1000000)) - (tv1.tv_usec + (tv1.tv_sec * 1000000));
 331         CtdlLogPrintf(CTDL_DEBUG, "RBL check completed in %ld.%ld seconds\n",
 332                 (total_time / 1000000),
 333                 (total_time % 1000000)
 334         );
 335         return(0);
 336 }
 337
 338 /*
 339  * Convert a host name to a dotted quad address.
 340  * Returns zero on success or nonzero on failure.
 341  */
 342 int hostname_to_dotted_quad(char *addr, char *host) {
 343         struct hostent *ch;
 344         const char *i;
 345         int a1, a2, a3, a4;
 346
 347         ch = gethostbyname(host);
 348         if (ch == NULL) {
 349                 strcpy(addr, "0.0.0.0");
 350                 return(1);
 351         }
 352
 353         i = (const char *) ch->h_addr_list[0];
 354         a1 = ((*i++) & 0xff);
 355         a2 = ((*i++) & 0xff);
 356         a3 = ((*i++) & 0xff);
 357         a4 = ((*i++) & 0xff);
 358         sprintf(addr, "%d.%d.%d.%d", a1, a2, a3, a4);
 359         return(0);
 360 }