1 /* 2 * iterator/iterator.h - iterative resolver DNS query response module 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 25 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR 26 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT 27 * HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, 28 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED 29 * TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR 30 * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF 31 * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING 32 * NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS 33 * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 34 */ 35 36 /** 37 * \file 38 * 39 * This file contains a module that performs recursive iterative DNS query 40 * processing. 41 */ 42 43 #ifndef ITERATOR_ITERATOR_H 44 #define ITERATOR_ITERATOR_H 45 #include "services/outbound_list.h" 46 #include "util/data/msgreply.h" 47 #include "util/module.h" 48 struct delegpt; 49 struct iter_hints; 50 struct iter_forwards; 51 struct iter_donotq; 52 struct iter_prep_list; 53 struct iter_priv; 54 struct rbtree_type; 55 56 /** max number of targets spawned for a query and its subqueries */ 57 #define MAX_TARGET_COUNT 64 58 /** max number of target lookups per qstate, per delegation point */ 59 #define MAX_DP_TARGET_COUNT 16 60 /** max number of nxdomains allowed for target lookups for a query and 61 * its subqueries */ 62 #define MAX_TARGET_NX 5 63 /** max number of nxdomains allowed for target lookups for a query and 64 * its subqueries when fallback has kicked in */ 65 #define MAX_TARGET_NX_FALLBACK (MAX_TARGET_NX*2) 66 /** max number of referrals. Makes sure resolver does not run away */ 67 #define MAX_REFERRAL_COUNT 130 68 /** max number of queries for which to perform dnsseclameness detection, 69 * (rrsigs missing detection) after that, just pick up that response */ 70 #define DNSSEC_LAME_DETECT_COUNT 4 71 /** 72 * max number of QNAME minimisation iterations. Limits number of queries for 73 * QNAMEs with a lot of labels. 74 */ 75 #define MAX_MINIMISE_COUNT 10 76 /* max number of time-outs for minimised query. Prevents resolving failures 77 * when the QNAME minimisation QTYPE is blocked. */ 78 #define MAX_MINIMISE_TIMEOUT_COUNT 3 79 /** 80 * number of labels from QNAME that are always send individually when using 81 * QNAME minimisation, even when the number of labels of the QNAME is bigger 82 * than MAX_MINIMISE_COUNT */ 83 #define MINIMISE_ONE_LAB 4 84 #define MINIMISE_MULTIPLE_LABS (MAX_MINIMISE_COUNT - MINIMISE_ONE_LAB) 85 /** at what query-sent-count to stop target fetch policy */ 86 #define TARGET_FETCH_STOP 3 87 /** how nice is a server without further information, in msec 88 * Equals rtt initial timeout value. 89 */ 90 extern int UNKNOWN_SERVER_NICENESS; 91 /** maximum timeout before a host is deemed unsuitable, in msec. 92 * After host_ttl this will be timed out and the host will be tried again. 93 * Equals RTT_MAX_TIMEOUT, and thus when RTT_MAX_TIMEOUT is overwritten by 94 * config infra_cache_max_rtt, it will be overwritten as well. */ 95 extern int USEFUL_SERVER_TOP_TIMEOUT; 96 /** penalty to validation failed blacklisted IPs 97 * Equals USEFUL_SERVER_TOP_TIMEOUT*4, and thus when RTT_MAX_TIMEOUT is 98 * overwritten by config infra_cache_max_rtt, it will be overwritten as well. */ 99 extern int BLACKLIST_PENALTY; 100 /** RTT band, within this amount from the best, servers are chosen randomly. 101 * Chosen so that the UNKNOWN_SERVER_NICENESS falls within the band of a 102 * fast server, this causes server exploration as a side benefit. msec. */ 103 #define RTT_BAND 400 104 105 /** 106 * Global state for the iterator. 107 */ 108 struct iter_env { 109 /** A flag to indicate whether or not we have an IPv6 route */ 110 int supports_ipv6; 111 112 /** A flag to indicate whether or not we have an IPv4 route */ 113 int supports_ipv4; 114 115 /** A flag to locally apply NAT64 to make IPv4 addrs into IPv6 */ 116 int use_nat64; 117 118 /** NAT64 prefix address, cf. dns64_env->prefix_addr */ 119 struct sockaddr_storage nat64_prefix_addr; 120 121 /** sizeof(sockaddr_in6) */ 122 socklen_t nat64_prefix_addrlen; 123 124 /** CIDR mask length of NAT64 prefix */ 125 int nat64_prefix_net; 126 127 /** A set of inetaddrs that should never be queried. */ 128 struct iter_donotq* donotq; 129 130 /** private address space and private domains */ 131 struct iter_priv* priv; 132 133 /** whitelist for capsforid names */ 134 struct rbtree_type* caps_white; 135 136 /** The maximum dependency depth that this resolver will pursue. */ 137 int max_dependency_depth; 138 139 /** 140 * The target fetch policy for each dependency level. This is 141 * described as a simple number (per dependency level): 142 * negative numbers (usually just -1) mean fetch-all, 143 * 0 means only fetch on demand, and 144 * positive numbers mean to fetch at most that many targets. 145 * array of max_dependency_depth+1 size. 146 */ 147 int* target_fetch_policy; 148 149 /** lock on ratelimit counter */ 150 lock_basic_type queries_ratelimit_lock; 151 /** number of queries that have been ratelimited */ 152 size_t num_queries_ratelimited; 153 154 /** number of retries on outgoing queries */ 155 int outbound_msg_retry; 156 157 /** number of queries_sent */ 158 int max_sent_count; 159 160 /** max number of query restarts to limit length of CNAME chain */ 161 int max_query_restarts; 162 }; 163 164 /** 165 * QNAME minimisation state 166 */ 167 enum minimisation_state { 168 /** 169 * (Re)start minimisation. Outgoing QNAME should be set to dp->name. 170 * State entered on new query or after following referral or CNAME. 171 */ 172 INIT_MINIMISE_STATE = 0, 173 /** 174 * QNAME minimisation ongoing. Increase QNAME on every iteration. 175 */ 176 MINIMISE_STATE, 177 /** 178 * Don't increment QNAME this iteration 179 */ 180 SKIP_MINIMISE_STATE, 181 /** 182 * Send out full QNAME + original QTYPE 183 */ 184 DONOT_MINIMISE_STATE, 185 }; 186 187 /** 188 * State of the iterator for a query. 189 */ 190 enum iter_state { 191 /** 192 * Externally generated queries start at this state. Query restarts are 193 * reset to this state. 194 */ 195 INIT_REQUEST_STATE = 0, 196 197 /** 198 * Root priming events reactivate here, most other events pass 199 * through this naturally as the 2nd part of the INIT_REQUEST_STATE. 200 */ 201 INIT_REQUEST_2_STATE, 202 203 /** 204 * Stub priming events reactivate here, most other events pass 205 * through this naturally as the 3rd part of the INIT_REQUEST_STATE. 206 */ 207 INIT_REQUEST_3_STATE, 208 209 /** 210 * Each time a delegation point changes for a given query or a 211 * query times out and/or wakes up, this state is (re)visited. 212 * This state is responsible for iterating through a list of 213 * nameserver targets. 214 */ 215 QUERYTARGETS_STATE, 216 217 /** 218 * Responses to queries start at this state. This state handles 219 * the decision tree associated with handling responses. 220 */ 221 QUERY_RESP_STATE, 222 223 /** Responses to priming queries finish at this state. */ 224 PRIME_RESP_STATE, 225 226 /** Collecting query class information, for qclass=ANY, when 227 * it spawns off queries for every class, it returns here. */ 228 COLLECT_CLASS_STATE, 229 230 /** Find NS record to resolve DS record from, walking to the right 231 * NS spot until we find it */ 232 DSNS_FIND_STATE, 233 234 /** Responses that are to be returned upstream end at this state. 235 * As well as responses to target queries. */ 236 FINISHED_STATE 237 }; 238 239 /** 240 * Shared counters for queries. 241 */ 242 enum target_count_variables { 243 /** Reference count for the shared iter_qstate->target_count. */ 244 TARGET_COUNT_REF = 0, 245 /** Number of target queries spawned for the query and subqueries. */ 246 TARGET_COUNT_QUERIES, 247 /** Number of nxdomain responses encountered. */ 248 TARGET_COUNT_NX, 249 250 /** This should stay last here, it is used for the allocation */ 251 TARGET_COUNT_MAX, 252 }; 253 254 /** 255 * Per query state for the iterator module. 256 */ 257 struct iter_qstate { 258 /** 259 * State of the iterator module. 260 * This is the state that event is in or should sent to -- all 261 * requests should start with the INIT_REQUEST_STATE. All 262 * responses should start with QUERY_RESP_STATE. Subsequent 263 * processing of the event will change this state. 264 */ 265 enum iter_state state; 266 267 /** 268 * Final state for the iterator module. 269 * This is the state that responses should be routed to once the 270 * response is final. For externally initiated queries, this 271 * will be FINISHED_STATE, locally initiated queries will have 272 * different final states. 273 */ 274 enum iter_state final_state; 275 276 /** 277 * The depth of this query, this means the depth of recursion. 278 * This address is needed for another query, which is an address 279 * needed for another query, etc. Original client query has depth 0. 280 */ 281 int depth; 282 283 /** 284 * The response 285 */ 286 struct dns_msg* response; 287 288 /** 289 * This is a list of RRsets that must be prepended to the 290 * ANSWER section of a response before being sent upstream. 291 */ 292 struct iter_prep_list* an_prepend_list; 293 /** Last element of the prepend list */ 294 struct iter_prep_list* an_prepend_last; 295 296 /** 297 * This is the list of RRsets that must be prepended to the 298 * AUTHORITY section of the response before being sent upstream. 299 */ 300 struct iter_prep_list* ns_prepend_list; 301 /** Last element of the authority prepend list */ 302 struct iter_prep_list* ns_prepend_last; 303 304 /** query name used for chasing the results. Initially the same as 305 * the state qinfo, but after CNAMEs this will be different. 306 * The query info used to elicit the results needed. */ 307 struct query_info qchase; 308 /** query flags to use when chasing the answer (i.e. RD flag) */ 309 uint16_t chase_flags; 310 /** true if we set RD bit because of last resort recursion lame query*/ 311 int chase_to_rd; 312 313 /** 314 * This is the current delegation point for an in-progress query. This 315 * object retains state as to which delegation targets need to be 316 * (sub)queried for vs which ones have already been visited. 317 */ 318 struct delegpt* dp; 319 320 /** state for 0x20 fallback when capsfail happens, 0 not a fallback */ 321 int caps_fallback; 322 /** state for capsfail: current server number to try */ 323 size_t caps_server; 324 /** state for capsfail: stored query for comparisons. Can be NULL if 325 * no response had been seen prior to starting the fallback. */ 326 struct reply_info* caps_reply; 327 struct dns_msg* caps_response; 328 329 /** Current delegation message - returned for non-RD queries */ 330 struct dns_msg* deleg_msg; 331 332 /** number of outstanding target sub queries */ 333 int num_target_queries; 334 335 /** outstanding direct queries */ 336 int num_current_queries; 337 338 /** the number of times this query has been restarted. */ 339 int query_restart_count; 340 341 /** the number of times this query has followed a referral. */ 342 int referral_count; 343 344 /** number of queries fired off */ 345 int sent_count; 346 347 /** malloced-array shared with this query and its subqueries. It keeps 348 * track of the defined enum target_count_variables counters. */ 349 int* target_count; 350 351 /** number of target lookups per delegation point. Reset to 0 after 352 * receiving referral answer. Not shared with subqueries. */ 353 int dp_target_count; 354 355 /** Delegation point that triggered the NXNS fallback; shared with 356 * this query and its subqueries, count-referenced by the reference 357 * counter in target_count. 358 * This also marks the fallback activation. */ 359 uint8_t** nxns_dp; 360 361 /** if true, already tested for ratelimiting and passed the test */ 362 int ratelimit_ok; 363 364 /** 365 * The query must store NS records from referrals as parentside RRs 366 * Enabled once it hits resolution problems, to throttle retries. 367 * If enabled it is the pointer to the old delegation point with 368 * the old retry counts for bad-nameserver-addresses. 369 */ 370 struct delegpt* store_parent_NS; 371 372 /** 373 * The query is for parent-side glue(A or AAAA) for a nameserver. 374 * If the item is seen as glue in a referral, and pside_glue is NULL, 375 * then it is stored in pside_glue for later. 376 * If it was never seen, at the end, then a negative caching element 377 * must be created. 378 * The (data or negative) RR cache element then throttles retries. 379 */ 380 int query_for_pside_glue; 381 /** the parent-side-glue element (NULL if none, its first match) */ 382 struct ub_packed_rrset_key* pside_glue; 383 384 /** If nonNULL we are walking upwards from DS query to find NS */ 385 uint8_t* dsns_point; 386 /** length of the dname in dsns_point */ 387 size_t dsns_point_len; 388 389 /** 390 * expected dnssec information for this iteration step. 391 * If dnssec rrsigs are expected and not given, the server is marked 392 * lame (dnssec-lame). 393 */ 394 int dnssec_expected; 395 396 /** 397 * We are expecting dnssec information, but we also know the server 398 * is DNSSEC lame. The response need not be marked dnssec-lame again. 399 */ 400 int dnssec_lame_query; 401 402 /** 403 * This is flag that, if true, means that this event is 404 * waiting for a stub priming query. 405 */ 406 int wait_priming_stub; 407 408 /** 409 * This is a flag that, if true, means that this query is 410 * for (re)fetching glue from a zone. Since the address should 411 * have been glue, query again to the servers that should have 412 * been returning it as glue. 413 * The delegation point must be set to the one that should *not* 414 * be used when creating the state. A higher one will be attempted. 415 */ 416 int refetch_glue; 417 418 /** list of pending queries to authoritative servers. */ 419 struct outbound_list outlist; 420 421 /** QNAME minimisation state, RFC9156 */ 422 enum minimisation_state minimisation_state; 423 424 /** State for capsfail: QNAME minimisation state for comparisons. */ 425 enum minimisation_state caps_minimisation_state; 426 427 /** 428 * The query info that is sent upstream. Will be a subset of qchase 429 * when qname minimisation is enabled. 430 */ 431 struct query_info qinfo_out; 432 433 /** 434 * Count number of QNAME minimisation iterations. Used to limit number of 435 * outgoing queries when QNAME minimisation is enabled. 436 */ 437 int minimise_count; 438 439 /** 440 * Count number of time-outs. Used to prevent resolving failures when 441 * the QNAME minimisation QTYPE is blocked. Used to determine if 442 * capsforid fallback should be started.*/ 443 int timeout_count; 444 445 /** True if the current response is from auth_zone */ 446 int auth_zone_response; 447 /** True if the auth_zones should not be consulted for the query */ 448 int auth_zone_avoid; 449 /** true if there have been scrubbing failures of reply packets */ 450 int scrub_failures; 451 /** true if there have been parse failures of reply packets */ 452 int parse_failures; 453 /** a failure printout address for last received answer */ 454 union { 455 struct in_addr in; 456 #ifdef AF_INET6 457 struct in6_addr in6; 458 #endif 459 } fail_addr; 460 /** which fail_addr, 0 is nothing, 4 or 6 */ 461 int fail_addr_type; 462 }; 463 464 /** 465 * List of prepend items 466 */ 467 struct iter_prep_list { 468 /** next in list */ 469 struct iter_prep_list* next; 470 /** rrset */ 471 struct ub_packed_rrset_key* rrset; 472 }; 473 474 /** 475 * Get the iterator function block. 476 * @return: function block with function pointers to iterator methods. 477 */ 478 struct module_func_block* iter_get_funcblock(void); 479 480 /** 481 * Get iterator state as a string 482 * @param state: to convert 483 * @return constant string that is printable. 484 */ 485 const char* iter_state_to_string(enum iter_state state); 486 487 /** 488 * See if iterator state is a response state 489 * @param s: to inspect 490 * @return true if response state. 491 */ 492 int iter_state_is_responsestate(enum iter_state s); 493 494 /** iterator init */ 495 int iter_init(struct module_env* env, int id); 496 497 /** iterator deinit */ 498 void iter_deinit(struct module_env* env, int id); 499 500 /** iterator operate on a query */ 501 void iter_operate(struct module_qstate* qstate, enum module_ev event, int id, 502 struct outbound_entry* outbound); 503 504 /** 505 * Return priming query results to interested super querystates. 506 * 507 * Sets the delegation point and delegation message (not nonRD queries). 508 * This is a callback from walk_supers. 509 * 510 * @param qstate: query state that finished. 511 * @param id: module id. 512 * @param super: the qstate to inform. 513 */ 514 void iter_inform_super(struct module_qstate* qstate, int id, 515 struct module_qstate* super); 516 517 /** iterator cleanup query state */ 518 void iter_clear(struct module_qstate* qstate, int id); 519 520 /** iterator alloc size routine */ 521 size_t iter_get_mem(struct module_env* env, int id); 522 523 #endif /* ITERATOR_ITERATOR_H */ 524