1 /* 2 * iterator/iter_utils.c - iterative resolver module utility functions. 3 * 4 * Copyright (c) 2007, NLnet Labs. All rights reserved. 5 * 6 * This software is open source. 7 * 8 * Redistribution and use in source and binary forms, with or without 9 * modification, are permitted provided that the following conditions 10 * are met: 11 * 12 * Redistributions of source code must retain the above copyright notice, 13 * this list of conditions and the following disclaimer. 14 * 15 * Redistributions in binary form must reproduce the above copyright notice, 16 * this list of conditions and the following disclaimer in the documentation 17 * and/or other materials provided with the distribution. 18 * 19 * Neither the name of the NLNET LABS nor the names of its contributors may 20 * be used to endorse or promote products derived from this software without 21 * specific prior written permission. 22 * 23 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS 24 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 25 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 26 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE 27 * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 28 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 29 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 30 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 31 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 32 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 33 * POSSIBILITY OF SUCH DAMAGE. 34 */ 35 36 /** 37 * \file 38 * 39 * This file contains functions to assist the iterator module. 40 * Configuration options. Forward zones. 41 */ 42 #include "config.h" 43 #include "iterator/iter_utils.h" 44 #include "iterator/iterator.h" 45 #include "iterator/iter_hints.h" 46 #include "iterator/iter_fwd.h" 47 #include "iterator/iter_donotq.h" 48 #include "iterator/iter_delegpt.h" 49 #include "iterator/iter_priv.h" 50 #include "services/cache/infra.h" 51 #include "services/cache/dns.h" 52 #include "services/cache/rrset.h" 53 #include "util/net_help.h" 54 #include "util/module.h" 55 #include "util/log.h" 56 #include "util/config_file.h" 57 #include "util/regional.h" 58 #include "util/data/msgparse.h" 59 #include "util/data/dname.h" 60 #include "util/random.h" 61 #include "util/fptr_wlist.h" 62 #include "validator/val_anchor.h" 63 #include "validator/val_kcache.h" 64 #include "validator/val_kentry.h" 65 #include "validator/val_utils.h" 66 67 /** time when nameserver glue is said to be 'recent' */ 68 #define SUSPICION_RECENT_EXPIRY 86400 69 /** penalty to validation failed blacklisted IPs */ 70 #define BLACKLIST_PENALTY (USEFUL_SERVER_TOP_TIMEOUT*4) 71 72 /** fillup fetch policy array */ 73 static void 74 fetch_fill(struct iter_env* ie, const char* str) 75 { 76 char* s = (char*)str, *e; 77 int i; 78 for(i=0; i<ie->max_dependency_depth+1; i++) { 79 ie->target_fetch_policy[i] = strtol(s, &e, 10); 80 if(s == e) 81 fatal_exit("cannot parse fetch policy number %s", s); 82 s = e; 83 } 84 } 85 86 /** Read config string that represents the target fetch policy */ 87 static int 88 read_fetch_policy(struct iter_env* ie, const char* str) 89 { 90 int count = cfg_count_numbers(str); 91 if(count < 1) { 92 log_err("Cannot parse target fetch policy: \"%s\"", str); 93 return 0; 94 } 95 ie->max_dependency_depth = count - 1; 96 ie->target_fetch_policy = (int*)calloc( 97 (size_t)ie->max_dependency_depth+1, sizeof(int)); 98 if(!ie->target_fetch_policy) { 99 log_err("alloc fetch policy: out of memory"); 100 return 0; 101 } 102 fetch_fill(ie, str); 103 return 1; 104 } 105 106 int 107 iter_apply_cfg(struct iter_env* iter_env, struct config_file* cfg) 108 { 109 int i; 110 /* target fetch policy */ 111 if(!read_fetch_policy(iter_env, cfg->target_fetch_policy)) 112 return 0; 113 for(i=0; i<iter_env->max_dependency_depth+1; i++) 114 verbose(VERB_QUERY, "target fetch policy for level %d is %d", 115 i, iter_env->target_fetch_policy[i]); 116 117 if(!iter_env->donotq) 118 iter_env->donotq = donotq_create(); 119 if(!iter_env->donotq || !donotq_apply_cfg(iter_env->donotq, cfg)) { 120 log_err("Could not set donotqueryaddresses"); 121 return 0; 122 } 123 if(!iter_env->priv) 124 iter_env->priv = priv_create(); 125 if(!iter_env->priv || !priv_apply_cfg(iter_env->priv, cfg)) { 126 log_err("Could not set private addresses"); 127 return 0; 128 } 129 iter_env->supports_ipv6 = cfg->do_ip6; 130 iter_env->supports_ipv4 = cfg->do_ip4; 131 return 1; 132 } 133 134 /** filter out unsuitable targets 135 * @param iter_env: iterator environment with ipv6-support flag. 136 * @param env: module environment with infra cache. 137 * @param name: zone name 138 * @param namelen: length of name 139 * @param qtype: query type (host order). 140 * @param now: current time 141 * @param a: address in delegation point we are examining. 142 * @return an integer that signals the target suitability. 143 * as follows: 144 * -1: The address should be omitted from the list. 145 * Because: 146 * o The address is bogus (DNSSEC validation failure). 147 * o Listed as donotquery 148 * o is ipv6 but no ipv6 support (in operating system). 149 * o is ipv4 but no ipv4 support (in operating system). 150 * o is lame 151 * Otherwise, an rtt in milliseconds. 152 * 0 .. USEFUL_SERVER_TOP_TIMEOUT-1 153 * The roundtrip time timeout estimate. less than 2 minutes. 154 * Note that util/rtt.c has a MIN_TIMEOUT of 50 msec, thus 155 * values 0 .. 49 are not used, unless that is changed. 156 * USEFUL_SERVER_TOP_TIMEOUT 157 * This value exactly is given for unresponsive blacklisted. 158 * USEFUL_SERVER_TOP_TIMEOUT+1 159 * For non-blacklisted servers: huge timeout, but has traffic. 160 * USEFUL_SERVER_TOP_TIMEOUT*1 .. 161 * parent-side lame servers get this penalty. A dispreferential 162 * server. (lame in delegpt). 163 * USEFUL_SERVER_TOP_TIMEOUT*2 .. 164 * dnsseclame servers get penalty 165 * USEFUL_SERVER_TOP_TIMEOUT*3 .. 166 * recursion lame servers get penalty 167 * UNKNOWN_SERVER_NICENESS 168 * If no information is known about the server, this is 169 * returned. 376 msec or so. 170 * +BLACKLIST_PENALTY (of USEFUL_TOP_TIMEOUT*4) for dnssec failed IPs. 171 * 172 * When a final value is chosen that is dnsseclame ; dnsseclameness checking 173 * is turned off (so we do not discard the reply). 174 * When a final value is chosen that is recursionlame; RD bit is set on query. 175 * Because of the numbers this means recursionlame also have dnssec lameness 176 * checking turned off. 177 */ 178 static int 179 iter_filter_unsuitable(struct iter_env* iter_env, struct module_env* env, 180 uint8_t* name, size_t namelen, uint16_t qtype, uint32_t now, 181 struct delegpt_addr* a) 182 { 183 int rtt, lame, reclame, dnsseclame; 184 if(a->bogus) 185 return -1; /* address of server is bogus */ 186 if(donotq_lookup(iter_env->donotq, &a->addr, a->addrlen)) { 187 log_addr(VERB_ALGO, "skip addr on the donotquery list", 188 &a->addr, a->addrlen); 189 return -1; /* server is on the donotquery list */ 190 } 191 if(!iter_env->supports_ipv6 && addr_is_ip6(&a->addr, a->addrlen)) { 192 return -1; /* there is no ip6 available */ 193 } 194 if(!iter_env->supports_ipv4 && !addr_is_ip6(&a->addr, a->addrlen)) { 195 return -1; /* there is no ip4 available */ 196 } 197 /* check lameness - need zone , class info */ 198 if(infra_get_lame_rtt(env->infra_cache, &a->addr, a->addrlen, 199 name, namelen, qtype, &lame, &dnsseclame, &reclame, 200 &rtt, now)) { 201 log_addr(VERB_ALGO, "servselect", &a->addr, a->addrlen); 202 verbose(VERB_ALGO, " rtt=%d%s%s%s%s", rtt, 203 lame?" LAME":"", 204 dnsseclame?" DNSSEC_LAME":"", 205 reclame?" REC_LAME":"", 206 a->lame?" ADDR_LAME":""); 207 if(lame) 208 return -1; /* server is lame */ 209 else if(rtt >= USEFUL_SERVER_TOP_TIMEOUT) 210 /* server is unresponsive, 211 * we used to return TOP_TIMOUT, but fairly useless, 212 * because if == TOP_TIMEOUT is dropped because 213 * blacklisted later, instead, remove it here, so 214 * other choices (that are not blacklisted) can be 215 * tried */ 216 return -1; 217 /* select remainder from worst to best */ 218 else if(reclame) 219 return rtt+USEFUL_SERVER_TOP_TIMEOUT*3; /* nonpref */ 220 else if(dnsseclame ) 221 return rtt+USEFUL_SERVER_TOP_TIMEOUT*2; /* nonpref */ 222 else if(a->lame) 223 return rtt+USEFUL_SERVER_TOP_TIMEOUT+1; /* nonpref */ 224 else return rtt; 225 } 226 /* no server information present */ 227 if(a->lame) 228 return USEFUL_SERVER_TOP_TIMEOUT+1+UNKNOWN_SERVER_NICENESS; /* nonpref */ 229 return UNKNOWN_SERVER_NICENESS; 230 } 231 232 /** lookup RTT information, and also store fastest rtt (if any) */ 233 static int 234 iter_fill_rtt(struct iter_env* iter_env, struct module_env* env, 235 uint8_t* name, size_t namelen, uint16_t qtype, uint32_t now, 236 struct delegpt* dp, int* best_rtt, struct sock_list* blacklist) 237 { 238 int got_it = 0; 239 struct delegpt_addr* a; 240 if(dp->bogus) 241 return 0; /* NS bogus, all bogus, nothing found */ 242 for(a=dp->result_list; a; a = a->next_result) { 243 a->sel_rtt = iter_filter_unsuitable(iter_env, env, 244 name, namelen, qtype, now, a); 245 if(a->sel_rtt != -1) { 246 if(sock_list_find(blacklist, &a->addr, a->addrlen)) 247 a->sel_rtt += BLACKLIST_PENALTY; 248 249 if(!got_it) { 250 *best_rtt = a->sel_rtt; 251 got_it = 1; 252 } else if(a->sel_rtt < *best_rtt) { 253 *best_rtt = a->sel_rtt; 254 } 255 } 256 } 257 return got_it; 258 } 259 260 /** filter the addres list, putting best targets at front, 261 * returns number of best targets (or 0, no suitable targets) */ 262 static int 263 iter_filter_order(struct iter_env* iter_env, struct module_env* env, 264 uint8_t* name, size_t namelen, uint16_t qtype, uint32_t now, 265 struct delegpt* dp, int* selected_rtt, int open_target, 266 struct sock_list* blacklist) 267 { 268 int got_num = 0, low_rtt = 0, swap_to_front; 269 struct delegpt_addr* a, *n, *prev=NULL; 270 271 /* fillup sel_rtt and find best rtt in the bunch */ 272 got_num = iter_fill_rtt(iter_env, env, name, namelen, qtype, now, dp, 273 &low_rtt, blacklist); 274 if(got_num == 0) 275 return 0; 276 if(low_rtt >= USEFUL_SERVER_TOP_TIMEOUT && 277 (delegpt_count_missing_targets(dp) > 0 || open_target > 0)) { 278 verbose(VERB_ALGO, "Bad choices, trying to get more choice"); 279 return 0; /* we want more choice. The best choice is a bad one. 280 return 0 to force the caller to fetch more */ 281 } 282 283 got_num = 0; 284 a = dp->result_list; 285 while(a) { 286 /* skip unsuitable targets */ 287 if(a->sel_rtt == -1) { 288 prev = a; 289 a = a->next_result; 290 continue; 291 } 292 /* classify the server address and determine what to do */ 293 swap_to_front = 0; 294 if(a->sel_rtt >= low_rtt && a->sel_rtt - low_rtt <= RTT_BAND) { 295 got_num++; 296 swap_to_front = 1; 297 } else if(a->sel_rtt<low_rtt && low_rtt-a->sel_rtt<=RTT_BAND) { 298 got_num++; 299 swap_to_front = 1; 300 } 301 /* swap to front if necessary, or move to next result */ 302 if(swap_to_front && prev) { 303 n = a->next_result; 304 prev->next_result = n; 305 a->next_result = dp->result_list; 306 dp->result_list = a; 307 a = n; 308 } else { 309 prev = a; 310 a = a->next_result; 311 } 312 } 313 *selected_rtt = low_rtt; 314 return got_num; 315 } 316 317 struct delegpt_addr* 318 iter_server_selection(struct iter_env* iter_env, 319 struct module_env* env, struct delegpt* dp, 320 uint8_t* name, size_t namelen, uint16_t qtype, int* dnssec_lame, 321 int* chase_to_rd, int open_target, struct sock_list* blacklist) 322 { 323 int sel; 324 int selrtt; 325 struct delegpt_addr* a, *prev; 326 int num = iter_filter_order(iter_env, env, name, namelen, qtype, 327 *env->now, dp, &selrtt, open_target, blacklist); 328 329 if(num == 0) 330 return NULL; 331 verbose(VERB_ALGO, "selrtt %d", selrtt); 332 if(selrtt > BLACKLIST_PENALTY) { 333 if(selrtt-BLACKLIST_PENALTY > USEFUL_SERVER_TOP_TIMEOUT*3) { 334 verbose(VERB_ALGO, "chase to " 335 "blacklisted recursion lame server"); 336 *chase_to_rd = 1; 337 } 338 if(selrtt-BLACKLIST_PENALTY > USEFUL_SERVER_TOP_TIMEOUT*2) { 339 verbose(VERB_ALGO, "chase to " 340 "blacklisted dnssec lame server"); 341 *dnssec_lame = 1; 342 } 343 } else { 344 if(selrtt > USEFUL_SERVER_TOP_TIMEOUT*3) { 345 verbose(VERB_ALGO, "chase to recursion lame server"); 346 *chase_to_rd = 1; 347 } 348 if(selrtt > USEFUL_SERVER_TOP_TIMEOUT*2) { 349 verbose(VERB_ALGO, "chase to dnssec lame server"); 350 *dnssec_lame = 1; 351 } 352 if(selrtt == USEFUL_SERVER_TOP_TIMEOUT) { 353 verbose(VERB_ALGO, "chase to blacklisted lame server"); 354 return NULL; 355 } 356 } 357 358 if(num == 1) { 359 a = dp->result_list; 360 if(++a->attempts < OUTBOUND_MSG_RETRY) 361 return a; 362 dp->result_list = a->next_result; 363 return a; 364 } 365 366 /* randomly select a target from the list */ 367 log_assert(num > 1); 368 /* grab secure random number, to pick unexpected server. 369 * also we need it to be threadsafe. */ 370 sel = ub_random_max(env->rnd, num); 371 a = dp->result_list; 372 prev = NULL; 373 while(sel > 0 && a) { 374 prev = a; 375 a = a->next_result; 376 sel--; 377 } 378 if(!a) /* robustness */ 379 return NULL; 380 if(++a->attempts < OUTBOUND_MSG_RETRY) 381 return a; 382 /* remove it from the delegation point result list */ 383 if(prev) 384 prev->next_result = a->next_result; 385 else dp->result_list = a->next_result; 386 return a; 387 } 388 389 struct dns_msg* 390 dns_alloc_msg(ldns_buffer* pkt, struct msg_parse* msg, 391 struct regional* region) 392 { 393 struct dns_msg* m = (struct dns_msg*)regional_alloc(region, 394 sizeof(struct dns_msg)); 395 if(!m) 396 return NULL; 397 memset(m, 0, sizeof(*m)); 398 if(!parse_create_msg(pkt, msg, NULL, &m->qinfo, &m->rep, region)) { 399 log_err("malloc failure: allocating incoming dns_msg"); 400 return NULL; 401 } 402 return m; 403 } 404 405 struct dns_msg* 406 dns_copy_msg(struct dns_msg* from, struct regional* region) 407 { 408 struct dns_msg* m = (struct dns_msg*)regional_alloc(region, 409 sizeof(struct dns_msg)); 410 if(!m) 411 return NULL; 412 m->qinfo = from->qinfo; 413 if(!(m->qinfo.qname = regional_alloc_init(region, from->qinfo.qname, 414 from->qinfo.qname_len))) 415 return NULL; 416 if(!(m->rep = reply_info_copy(from->rep, NULL, region))) 417 return NULL; 418 return m; 419 } 420 421 void 422 iter_dns_store(struct module_env* env, struct query_info* msgqinf, 423 struct reply_info* msgrep, int is_referral, uint32_t leeway, int pside, 424 struct regional* region) 425 { 426 if(!dns_cache_store(env, msgqinf, msgrep, is_referral, leeway, 427 pside, region)) 428 log_err("out of memory: cannot store data in cache"); 429 } 430 431 int 432 iter_ns_probability(struct ub_randstate* rnd, int n, int m) 433 { 434 int sel; 435 if(n == m) /* 100% chance */ 436 return 1; 437 /* we do not need secure random numbers here, but 438 * we do need it to be threadsafe, so we use this */ 439 sel = ub_random_max(rnd, m); 440 return (sel < n); 441 } 442 443 /** detect dependency cycle for query and target */ 444 static int 445 causes_cycle(struct module_qstate* qstate, uint8_t* name, size_t namelen, 446 uint16_t t, uint16_t c) 447 { 448 struct query_info qinf; 449 qinf.qname = name; 450 qinf.qname_len = namelen; 451 qinf.qtype = t; 452 qinf.qclass = c; 453 fptr_ok(fptr_whitelist_modenv_detect_cycle( 454 qstate->env->detect_cycle)); 455 return (*qstate->env->detect_cycle)(qstate, &qinf, 456 (uint16_t)(BIT_RD|BIT_CD), qstate->is_priming); 457 } 458 459 void 460 iter_mark_cycle_targets(struct module_qstate* qstate, struct delegpt* dp) 461 { 462 struct delegpt_ns* ns; 463 for(ns = dp->nslist; ns; ns = ns->next) { 464 if(ns->resolved) 465 continue; 466 /* see if this ns as target causes dependency cycle */ 467 if(causes_cycle(qstate, ns->name, ns->namelen, 468 LDNS_RR_TYPE_AAAA, qstate->qinfo.qclass) || 469 causes_cycle(qstate, ns->name, ns->namelen, 470 LDNS_RR_TYPE_A, qstate->qinfo.qclass)) { 471 log_nametypeclass(VERB_QUERY, "skipping target due " 472 "to dependency cycle (harden-glue: no may " 473 "fix some of the cycles)", 474 ns->name, LDNS_RR_TYPE_A, 475 qstate->qinfo.qclass); 476 ns->resolved = 1; 477 } 478 } 479 } 480 481 void 482 iter_mark_pside_cycle_targets(struct module_qstate* qstate, struct delegpt* dp) 483 { 484 struct delegpt_ns* ns; 485 for(ns = dp->nslist; ns; ns = ns->next) { 486 if(ns->done_pside4 && ns->done_pside6) 487 continue; 488 /* see if this ns as target causes dependency cycle */ 489 if(causes_cycle(qstate, ns->name, ns->namelen, 490 LDNS_RR_TYPE_A, qstate->qinfo.qclass)) { 491 log_nametypeclass(VERB_QUERY, "skipping target due " 492 "to dependency cycle", ns->name, 493 LDNS_RR_TYPE_A, qstate->qinfo.qclass); 494 ns->done_pside4 = 1; 495 } 496 if(causes_cycle(qstate, ns->name, ns->namelen, 497 LDNS_RR_TYPE_AAAA, qstate->qinfo.qclass)) { 498 log_nametypeclass(VERB_QUERY, "skipping target due " 499 "to dependency cycle", ns->name, 500 LDNS_RR_TYPE_AAAA, qstate->qinfo.qclass); 501 ns->done_pside6 = 1; 502 } 503 } 504 } 505 506 int 507 iter_dp_is_useless(struct query_info* qinfo, uint16_t qflags, 508 struct delegpt* dp) 509 { 510 struct delegpt_ns* ns; 511 /* check: 512 * o RD qflag is on. 513 * o no addresses are provided. 514 * o all NS items are required glue. 515 * OR 516 * o RD qflag is on. 517 * o no addresses are provided. 518 * o the query is for one of the nameservers in dp, 519 * and that nameserver is a glue-name for this dp. 520 */ 521 if(!(qflags&BIT_RD)) 522 return 0; 523 /* either available or unused targets */ 524 if(dp->usable_list || dp->result_list) 525 return 0; 526 527 /* see if query is for one of the nameservers, which is glue */ 528 if( (qinfo->qtype == LDNS_RR_TYPE_A || 529 qinfo->qtype == LDNS_RR_TYPE_AAAA) && 530 dname_subdomain_c(qinfo->qname, dp->name) && 531 delegpt_find_ns(dp, qinfo->qname, qinfo->qname_len)) 532 return 1; 533 534 for(ns = dp->nslist; ns; ns = ns->next) { 535 if(ns->resolved) /* skip failed targets */ 536 continue; 537 if(!dname_subdomain_c(ns->name, dp->name)) 538 return 0; /* one address is not required glue */ 539 } 540 return 1; 541 } 542 543 int 544 iter_indicates_dnssec(struct module_env* env, struct delegpt* dp, 545 struct dns_msg* msg, uint16_t dclass) 546 { 547 struct trust_anchor* a; 548 /* information not available, !env->anchors can be common */ 549 if(!env || !env->anchors || !dp || !dp->name) 550 return 0; 551 /* a trust anchor exists with this name, RRSIGs expected */ 552 if((a=anchor_find(env->anchors, dp->name, dp->namelabs, dp->namelen, 553 dclass))) { 554 lock_basic_unlock(&a->lock); 555 return 1; 556 } 557 /* see if DS rrset was given, in AUTH section */ 558 if(msg && msg->rep && 559 reply_find_rrset_section_ns(msg->rep, dp->name, dp->namelen, 560 LDNS_RR_TYPE_DS, dclass)) 561 return 1; 562 /* look in key cache */ 563 if(env->key_cache) { 564 struct key_entry_key* kk = key_cache_obtain(env->key_cache, 565 dp->name, dp->namelen, dclass, env->scratch, *env->now); 566 if(kk) { 567 if(query_dname_compare(kk->name, dp->name) == 0) { 568 if(key_entry_isgood(kk) || key_entry_isbad(kk)) { 569 regional_free_all(env->scratch); 570 return 1; 571 } else if(key_entry_isnull(kk)) { 572 regional_free_all(env->scratch); 573 return 0; 574 } 575 } 576 regional_free_all(env->scratch); 577 } 578 } 579 return 0; 580 } 581 582 int 583 iter_msg_has_dnssec(struct dns_msg* msg) 584 { 585 size_t i; 586 if(!msg || !msg->rep) 587 return 0; 588 for(i=0; i<msg->rep->an_numrrsets + msg->rep->ns_numrrsets; i++) { 589 if(((struct packed_rrset_data*)msg->rep->rrsets[i]-> 590 entry.data)->rrsig_count > 0) 591 return 1; 592 } 593 /* empty message has no DNSSEC info, with DNSSEC the reply is 594 * not empty (NSEC) */ 595 return 0; 596 } 597 598 int iter_msg_from_zone(struct dns_msg* msg, struct delegpt* dp, 599 enum response_type type, uint16_t dclass) 600 { 601 if(!msg || !dp || !msg->rep || !dp->name) 602 return 0; 603 /* SOA RRset - always from reply zone */ 604 if(reply_find_rrset_section_an(msg->rep, dp->name, dp->namelen, 605 LDNS_RR_TYPE_SOA, dclass) || 606 reply_find_rrset_section_ns(msg->rep, dp->name, dp->namelen, 607 LDNS_RR_TYPE_SOA, dclass)) 608 return 1; 609 if(type == RESPONSE_TYPE_REFERRAL) { 610 size_t i; 611 /* if it adds a single label, i.e. we expect .com, 612 * and referral to example.com. NS ... , then origin zone 613 * is .com. For a referral to sub.example.com. NS ... then 614 * we do not know, since example.com. may be in between. */ 615 for(i=0; i<msg->rep->an_numrrsets+msg->rep->ns_numrrsets; 616 i++) { 617 struct ub_packed_rrset_key* s = msg->rep->rrsets[i]; 618 if(ntohs(s->rk.type) == LDNS_RR_TYPE_NS && 619 ntohs(s->rk.rrset_class) == dclass) { 620 int l = dname_count_labels(s->rk.dname); 621 if(l == dp->namelabs + 1 && 622 dname_strict_subdomain(s->rk.dname, 623 l, dp->name, dp->namelabs)) 624 return 1; 625 } 626 } 627 return 0; 628 } 629 log_assert(type==RESPONSE_TYPE_ANSWER || type==RESPONSE_TYPE_CNAME); 630 /* not a referral, and not lame delegation (upwards), so, 631 * any NS rrset must be from the zone itself */ 632 if(reply_find_rrset_section_an(msg->rep, dp->name, dp->namelen, 633 LDNS_RR_TYPE_NS, dclass) || 634 reply_find_rrset_section_ns(msg->rep, dp->name, dp->namelen, 635 LDNS_RR_TYPE_NS, dclass)) 636 return 1; 637 /* a DNSKEY set is expected at the zone apex as well */ 638 /* this is for 'minimal responses' for DNSKEYs */ 639 if(reply_find_rrset_section_an(msg->rep, dp->name, dp->namelen, 640 LDNS_RR_TYPE_DNSKEY, dclass)) 641 return 1; 642 return 0; 643 } 644 645 /** 646 * check equality of two rrsets 647 * @param k1: rrset 648 * @param k2: rrset 649 * @return true if equal 650 */ 651 static int 652 rrset_equal(struct ub_packed_rrset_key* k1, struct ub_packed_rrset_key* k2) 653 { 654 struct packed_rrset_data* d1 = (struct packed_rrset_data*) 655 k1->entry.data; 656 struct packed_rrset_data* d2 = (struct packed_rrset_data*) 657 k2->entry.data; 658 size_t i, t; 659 if(k1->rk.dname_len != k2->rk.dname_len || 660 k1->rk.flags != k2->rk.flags || 661 k1->rk.type != k2->rk.type || 662 k1->rk.rrset_class != k2->rk.rrset_class || 663 query_dname_compare(k1->rk.dname, k2->rk.dname) != 0) 664 return 0; 665 if(d1->ttl != d2->ttl || 666 d1->count != d2->count || 667 d1->rrsig_count != d2->rrsig_count || 668 d1->trust != d2->trust || 669 d1->security != d2->security) 670 return 0; 671 t = d1->count + d1->rrsig_count; 672 for(i=0; i<t; i++) { 673 if(d1->rr_len[i] != d2->rr_len[i] || 674 d1->rr_ttl[i] != d2->rr_ttl[i] || 675 memcmp(d1->rr_data[i], d2->rr_data[i], 676 d1->rr_len[i]) != 0) 677 return 0; 678 } 679 return 1; 680 } 681 682 int 683 reply_equal(struct reply_info* p, struct reply_info* q, ldns_buffer* scratch) 684 { 685 size_t i; 686 if(p->flags != q->flags || 687 p->qdcount != q->qdcount || 688 p->ttl != q->ttl || 689 p->prefetch_ttl != q->prefetch_ttl || 690 p->security != q->security || 691 p->an_numrrsets != q->an_numrrsets || 692 p->ns_numrrsets != q->ns_numrrsets || 693 p->ar_numrrsets != q->ar_numrrsets || 694 p->rrset_count != q->rrset_count) 695 return 0; 696 for(i=0; i<p->rrset_count; i++) { 697 if(!rrset_equal(p->rrsets[i], q->rrsets[i])) { 698 /* fallback procedure: try to sort and canonicalize */ 699 ldns_rr_list* pl, *ql; 700 pl = packed_rrset_to_rr_list(p->rrsets[i], scratch); 701 ql = packed_rrset_to_rr_list(q->rrsets[i], scratch); 702 if(!pl || !ql) { 703 ldns_rr_list_deep_free(pl); 704 ldns_rr_list_deep_free(ql); 705 return 0; 706 } 707 ldns_rr_list2canonical(pl); 708 ldns_rr_list2canonical(ql); 709 ldns_rr_list_sort(pl); 710 ldns_rr_list_sort(ql); 711 if(ldns_rr_list_compare(pl, ql) != 0) { 712 ldns_rr_list_deep_free(pl); 713 ldns_rr_list_deep_free(ql); 714 return 0; 715 } 716 ldns_rr_list_deep_free(pl); 717 ldns_rr_list_deep_free(ql); 718 continue; 719 } 720 } 721 return 1; 722 } 723 724 void 725 iter_store_parentside_rrset(struct module_env* env, 726 struct ub_packed_rrset_key* rrset) 727 { 728 struct rrset_ref ref; 729 rrset = packed_rrset_copy_alloc(rrset, env->alloc, *env->now); 730 if(!rrset) { 731 log_err("malloc failure in store_parentside_rrset"); 732 return; 733 } 734 rrset->rk.flags |= PACKED_RRSET_PARENT_SIDE; 735 rrset->entry.hash = rrset_key_hash(&rrset->rk); 736 ref.key = rrset; 737 ref.id = rrset->id; 738 /* ignore ret: if it was in the cache, ref updated */ 739 (void)rrset_cache_update(env->rrset_cache, &ref, env->alloc, *env->now); 740 } 741 742 /** fetch NS record from reply, if any */ 743 static struct ub_packed_rrset_key* 744 reply_get_NS_rrset(struct reply_info* rep) 745 { 746 size_t i; 747 for(i=0; i<rep->rrset_count; i++) { 748 if(rep->rrsets[i]->rk.type == htons(LDNS_RR_TYPE_NS)) { 749 return rep->rrsets[i]; 750 } 751 } 752 return NULL; 753 } 754 755 void 756 iter_store_parentside_NS(struct module_env* env, struct reply_info* rep) 757 { 758 struct ub_packed_rrset_key* rrset = reply_get_NS_rrset(rep); 759 if(rrset) { 760 log_rrset_key(VERB_ALGO, "store parent-side NS", rrset); 761 iter_store_parentside_rrset(env, rrset); 762 } 763 } 764 765 void iter_store_parentside_neg(struct module_env* env, 766 struct query_info* qinfo, struct reply_info* rep) 767 { 768 /* TTL: NS from referral in iq->deleg_msg, 769 * or first RR from iq->response, 770 * or servfail5secs if !iq->response */ 771 uint32_t ttl = NORR_TTL; 772 struct ub_packed_rrset_key* neg; 773 struct packed_rrset_data* newd; 774 if(rep) { 775 struct ub_packed_rrset_key* rrset = reply_get_NS_rrset(rep); 776 if(!rrset && rep->rrset_count != 0) rrset = rep->rrsets[0]; 777 if(rrset) ttl = ub_packed_rrset_ttl(rrset); 778 } 779 /* create empty rrset to store */ 780 neg = (struct ub_packed_rrset_key*)regional_alloc(env->scratch, 781 sizeof(struct ub_packed_rrset_key)); 782 if(!neg) { 783 log_err("out of memory in store_parentside_neg"); 784 return; 785 } 786 memset(&neg->entry, 0, sizeof(neg->entry)); 787 neg->entry.key = neg; 788 neg->rk.type = htons(qinfo->qtype); 789 neg->rk.rrset_class = htons(qinfo->qclass); 790 neg->rk.flags = 0; 791 neg->rk.dname = regional_alloc_init(env->scratch, qinfo->qname, 792 qinfo->qname_len); 793 if(!neg->rk.dname) { 794 log_err("out of memory in store_parentside_neg"); 795 return; 796 } 797 neg->rk.dname_len = qinfo->qname_len; 798 neg->entry.hash = rrset_key_hash(&neg->rk); 799 newd = (struct packed_rrset_data*)regional_alloc_zero(env->scratch, 800 sizeof(struct packed_rrset_data) + sizeof(size_t) + 801 sizeof(uint8_t*) + sizeof(uint32_t) + sizeof(uint16_t)); 802 if(!newd) { 803 log_err("out of memory in store_parentside_neg"); 804 return; 805 } 806 neg->entry.data = newd; 807 newd->ttl = ttl; 808 /* entry must have one RR, otherwise not valid in cache. 809 * put in one RR with empty rdata: those are ignored as nameserver */ 810 newd->count = 1; 811 newd->rrsig_count = 0; 812 newd->trust = rrset_trust_ans_noAA; 813 newd->rr_len = (size_t*)((uint8_t*)newd + 814 sizeof(struct packed_rrset_data)); 815 newd->rr_len[0] = 0 /* zero len rdata */ + sizeof(uint16_t); 816 packed_rrset_ptr_fixup(newd); 817 newd->rr_ttl[0] = newd->ttl; 818 ldns_write_uint16(newd->rr_data[0], 0 /* zero len rdata */); 819 /* store it */ 820 log_rrset_key(VERB_ALGO, "store parent-side negative", neg); 821 iter_store_parentside_rrset(env, neg); 822 } 823 824 int 825 iter_lookup_parent_NS_from_cache(struct module_env* env, struct delegpt* dp, 826 struct regional* region, struct query_info* qinfo) 827 { 828 struct ub_packed_rrset_key* akey; 829 akey = rrset_cache_lookup(env->rrset_cache, dp->name, 830 dp->namelen, LDNS_RR_TYPE_NS, qinfo->qclass, 831 PACKED_RRSET_PARENT_SIDE, *env->now, 0); 832 if(akey) { 833 log_rrset_key(VERB_ALGO, "found parent-side NS in cache", akey); 834 dp->has_parent_side_NS = 1; 835 /* and mark the new names as lame */ 836 if(!delegpt_rrset_add_ns(dp, region, akey, 1)) { 837 lock_rw_unlock(&akey->entry.lock); 838 return 0; 839 } 840 lock_rw_unlock(&akey->entry.lock); 841 } 842 return 1; 843 } 844 845 int iter_lookup_parent_glue_from_cache(struct module_env* env, 846 struct delegpt* dp, struct regional* region, struct query_info* qinfo) 847 { 848 struct ub_packed_rrset_key* akey; 849 struct delegpt_ns* ns; 850 size_t num = delegpt_count_targets(dp); 851 for(ns = dp->nslist; ns; ns = ns->next) { 852 /* get cached parentside A */ 853 akey = rrset_cache_lookup(env->rrset_cache, ns->name, 854 ns->namelen, LDNS_RR_TYPE_A, qinfo->qclass, 855 PACKED_RRSET_PARENT_SIDE, *env->now, 0); 856 if(akey) { 857 log_rrset_key(VERB_ALGO, "found parent-side", akey); 858 ns->done_pside4 = 1; 859 /* a negative-cache-element has no addresses it adds */ 860 if(!delegpt_add_rrset_A(dp, region, akey, 1)) 861 log_err("malloc failure in lookup_parent_glue"); 862 lock_rw_unlock(&akey->entry.lock); 863 } 864 /* get cached parentside AAAA */ 865 akey = rrset_cache_lookup(env->rrset_cache, ns->name, 866 ns->namelen, LDNS_RR_TYPE_AAAA, qinfo->qclass, 867 PACKED_RRSET_PARENT_SIDE, *env->now, 0); 868 if(akey) { 869 log_rrset_key(VERB_ALGO, "found parent-side", akey); 870 ns->done_pside6 = 1; 871 /* a negative-cache-element has no addresses it adds */ 872 if(!delegpt_add_rrset_AAAA(dp, region, akey, 1)) 873 log_err("malloc failure in lookup_parent_glue"); 874 lock_rw_unlock(&akey->entry.lock); 875 } 876 } 877 /* see if new (but lame) addresses have become available */ 878 return delegpt_count_targets(dp) != num; 879 } 880 881 int 882 iter_get_next_root(struct iter_hints* hints, struct iter_forwards* fwd, 883 uint16_t* c) 884 { 885 uint16_t c1 = *c, c2 = *c; 886 int r1 = hints_next_root(hints, &c1); 887 int r2 = forwards_next_root(fwd, &c2); 888 if(!r1 && !r2) /* got none, end of list */ 889 return 0; 890 else if(!r1) /* got one, return that */ 891 *c = c2; 892 else if(!r2) 893 *c = c1; 894 else if(c1 < c2) /* got both take smallest */ 895 *c = c1; 896 else *c = c2; 897 return 1; 898 } 899 900 void 901 iter_scrub_ds(struct dns_msg* msg, struct ub_packed_rrset_key* ns, uint8_t* z) 902 { 903 /* Only the DS record for the delegation itself is expected. 904 * We allow DS for everything between the bailiwick and the 905 * zonecut, thus DS records must be at or above the zonecut. 906 * And the DS records must be below the server authority zone. 907 * The answer section is already scrubbed. */ 908 size_t i = msg->rep->an_numrrsets; 909 while(i < (msg->rep->an_numrrsets + msg->rep->ns_numrrsets)) { 910 struct ub_packed_rrset_key* s = msg->rep->rrsets[i]; 911 if(ntohs(s->rk.type) == LDNS_RR_TYPE_DS && 912 (!ns || !dname_subdomain_c(ns->rk.dname, s->rk.dname) 913 || query_dname_compare(z, s->rk.dname) == 0)) { 914 log_nametypeclass(VERB_ALGO, "removing irrelevant DS", 915 s->rk.dname, ntohs(s->rk.type), 916 ntohs(s->rk.rrset_class)); 917 memmove(msg->rep->rrsets+i, msg->rep->rrsets+i+1, 918 sizeof(struct ub_packed_rrset_key*) * 919 (msg->rep->rrset_count-i-1)); 920 msg->rep->ns_numrrsets--; 921 msg->rep->rrset_count--; 922 /* stay at same i, but new record */ 923 continue; 924 } 925 i++; 926 } 927 } 928 929 void iter_dec_attempts(struct delegpt* dp, int d) 930 { 931 struct delegpt_addr* a; 932 for(a=dp->target_list; a; a = a->next_target) { 933 if(a->attempts >= OUTBOUND_MSG_RETRY) { 934 /* add back to result list */ 935 a->next_result = dp->result_list; 936 dp->result_list = a; 937 } 938 if(a->attempts > d) 939 a->attempts -= d; 940 else a->attempts = 0; 941 } 942 } 943 944 void iter_merge_retry_counts(struct delegpt* dp, struct delegpt* old) 945 { 946 struct delegpt_addr* a, *o, *prev; 947 for(a=dp->target_list; a; a = a->next_target) { 948 o = delegpt_find_addr(old, &a->addr, a->addrlen); 949 if(o) { 950 log_addr(VERB_ALGO, "copy attempt count previous dp", 951 &a->addr, a->addrlen); 952 a->attempts = o->attempts; 953 } 954 } 955 prev = NULL; 956 a = dp->usable_list; 957 while(a) { 958 if(a->attempts >= OUTBOUND_MSG_RETRY) { 959 log_addr(VERB_ALGO, "remove from usable list dp", 960 &a->addr, a->addrlen); 961 /* remove from result list */ 962 if(prev) 963 prev->next_usable = a->next_usable; 964 else dp->usable_list = a->next_usable; 965 /* prev stays the same */ 966 a = a->next_usable; 967 continue; 968 } 969 prev = a; 970 a = a->next_usable; 971 } 972 } 973 974 int 975 iter_ds_toolow(struct dns_msg* msg, struct delegpt* dp) 976 { 977 /* if for query example.com, there is example.com SOA or a subdomain 978 * of example.com, then we are too low and need to fetch NS. */ 979 size_t i; 980 /* if we have a DNAME or CNAME we are probably wrong */ 981 /* if we have a qtype DS in the answer section, its fine */ 982 for(i=0; i < msg->rep->an_numrrsets; i++) { 983 struct ub_packed_rrset_key* s = msg->rep->rrsets[i]; 984 if(ntohs(s->rk.type) == LDNS_RR_TYPE_DNAME || 985 ntohs(s->rk.type) == LDNS_RR_TYPE_CNAME) { 986 /* not the right answer, maybe too low, check the 987 * RRSIG signer name (if there is any) for a hint 988 * that it is from the dp zone anyway */ 989 uint8_t* sname; 990 size_t slen; 991 val_find_rrset_signer(s, &sname, &slen); 992 if(sname && query_dname_compare(dp->name, sname)==0) 993 return 0; /* it is fine, from the right dp */ 994 return 1; 995 } 996 if(ntohs(s->rk.type) == LDNS_RR_TYPE_DS) 997 return 0; /* fine, we have a DS record */ 998 } 999 for(i=msg->rep->an_numrrsets; 1000 i < msg->rep->an_numrrsets + msg->rep->ns_numrrsets; i++) { 1001 struct ub_packed_rrset_key* s = msg->rep->rrsets[i]; 1002 if(ntohs(s->rk.type) == LDNS_RR_TYPE_SOA) { 1003 if(dname_subdomain_c(s->rk.dname, msg->qinfo.qname)) 1004 return 1; /* point is too low */ 1005 if(query_dname_compare(s->rk.dname, dp->name)==0) 1006 return 0; /* right dp */ 1007 } 1008 if(ntohs(s->rk.type) == LDNS_RR_TYPE_NSEC || 1009 ntohs(s->rk.type) == LDNS_RR_TYPE_NSEC3) { 1010 uint8_t* sname; 1011 size_t slen; 1012 val_find_rrset_signer(s, &sname, &slen); 1013 if(sname && query_dname_compare(dp->name, sname)==0) 1014 return 0; /* it is fine, from the right dp */ 1015 return 1; 1016 } 1017 } 1018 /* we do not know */ 1019 return 1; 1020 } 1021 1022 int iter_dp_cangodown(struct query_info* qinfo, struct delegpt* dp) 1023 { 1024 /* no delegation point, do not see how we can go down, 1025 * robust check, it should really exist */ 1026 if(!dp) return 0; 1027 1028 /* see if dp equals the qname, then we cannot go down further */ 1029 if(query_dname_compare(qinfo->qname, dp->name) == 0) 1030 return 0; 1031 /* if dp is one label above the name we also cannot go down further */ 1032 if(dname_count_labels(qinfo->qname) == dp->namelabs+1) 1033 return 0; 1034 return 1; 1035 } 1036