aboutsummaryrefslogblamecommitdiff
path: root/external/unbound/libunbound/libunbound.c
blob: b3a4c2ba77f6c8ce86ca7dbf7eb18f177d37d5b4 (plain) (tree)
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981






























































                                                                            
                          












































































































































































































































































































                                                                               



















                                                                           



















































































































































































































































































































































































































































































































































































































                                                                               
                                                                                            















































                                                                           





















                                                                    
                                                                                        
                                

                                  












                                                                           

                                                                             

















































































































































































                                                                               
/*
 * unbound.c - unbound validating resolver public API implementation
 *
 * Copyright (c) 2007, NLnet Labs. All rights reserved.
 *
 * This software is open source.
 * 
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 
 * Redistributions of source code must retain the above copyright notice,
 * this list of conditions and the following disclaimer.
 * 
 * Redistributions in binary form must reproduce the above copyright notice,
 * this list of conditions and the following disclaimer in the documentation
 * and/or other materials provided with the distribution.
 * 
 * Neither the name of the NLNET LABS nor the names of its contributors may
 * be used to endorse or promote products derived from this software without
 * specific prior written permission.
 * 
 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
 * HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED
 * TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
 * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
 * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
 * NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
 * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 */

/**
 * \file
 *
 * This file contains functions to resolve DNS queries and 
 * validate the answers. Synchonously and asynchronously.
 *
 */

/* include the public api first, it should be able to stand alone */
#include "libunbound/unbound.h"
#include "libunbound/unbound-event.h"
#include "config.h"
#include <ctype.h>
#include "libunbound/context.h"
#include "libunbound/libworker.h"
#include "util/locks.h"
#include "util/config_file.h"
#include "util/alloc.h"
#include "util/module.h"
#include "util/regional.h"
#include "util/log.h"
#include "util/random.h"
#include "util/net_help.h"
#include "util/tube.h"
#include "services/modstack.h"
#include "services/localzone.h"
#include "services/cache/infra.h"
#include "services/cache/rrset.h"
#include "sldns/sbuffer.h"
#ifdef HAVE_PTHREAD
#include <signal.h>
#endif

#if defined(UB_ON_WINDOWS) && defined (HAVE_WINDOWS_H)
#include <windows.h>
#include <iphlpapi.h>
#endif /* UB_ON_WINDOWS */

/** create context functionality, but no pipes */
static struct ub_ctx* ub_ctx_create_nopipe(void)
{
	struct ub_ctx* ctx;
	unsigned int seed;
#ifdef USE_WINSOCK
	int r;
	WSADATA wsa_data;
#endif
	
	log_init(NULL, 0, NULL); /* logs to stderr */
	log_ident_set("libunbound");
#ifdef USE_WINSOCK
	if((r = WSAStartup(MAKEWORD(2,2), &wsa_data)) != 0) {
		log_err("could not init winsock. WSAStartup: %s",
			wsa_strerror(r));
		return NULL;
	}
#endif
	verbosity = 0; /* errors only */
	checklock_start();
	ctx = (struct ub_ctx*)calloc(1, sizeof(*ctx));
	if(!ctx) {
		errno = ENOMEM;
		return NULL;
	}
	alloc_init(&ctx->superalloc, NULL, 0);
	seed = (unsigned int)time(NULL) ^ (unsigned int)getpid();
	if(!(ctx->seed_rnd = ub_initstate(seed, NULL))) {
		seed = 0;
		ub_randfree(ctx->seed_rnd);
		free(ctx);
		errno = ENOMEM;
		return NULL;
	}
	seed = 0;
	lock_basic_init(&ctx->qqpipe_lock);
	lock_basic_init(&ctx->rrpipe_lock);
	lock_basic_init(&ctx->cfglock);
	ctx->env = (struct module_env*)calloc(1, sizeof(*ctx->env));
	if(!ctx->env) {
		ub_randfree(ctx->seed_rnd);
		free(ctx);
		errno = ENOMEM;
		return NULL;
	}
	ctx->env->cfg = config_create_forlib();
	if(!ctx->env->cfg) {
		free(ctx->env);
		ub_randfree(ctx->seed_rnd);
		free(ctx);
		errno = ENOMEM;
		return NULL;
	}
	ctx->env->alloc = &ctx->superalloc;
	ctx->env->worker = NULL;
	ctx->env->need_to_validate = 0;
	modstack_init(&ctx->mods);
	rbtree_init(&ctx->queries, &context_query_cmp);
	return ctx;
}

struct ub_ctx* 
ub_ctx_create(void)
{
	struct ub_ctx* ctx = ub_ctx_create_nopipe();
	if(!ctx)
		return NULL;
	if((ctx->qq_pipe = tube_create()) == NULL) {
		int e = errno;
		ub_randfree(ctx->seed_rnd);
		config_delete(ctx->env->cfg);
		modstack_desetup(&ctx->mods, ctx->env);
		free(ctx->env);
		free(ctx);
		errno = e;
		return NULL;
	}
	if((ctx->rr_pipe = tube_create()) == NULL) {
		int e = errno;
		tube_delete(ctx->qq_pipe);
		ub_randfree(ctx->seed_rnd);
		config_delete(ctx->env->cfg);
		modstack_desetup(&ctx->mods, ctx->env);
		free(ctx->env);
		free(ctx);
		errno = e;
		return NULL;
	}
	return ctx;
}

struct ub_ctx* 
ub_ctx_create_event(struct event_base* eb)
{
	struct ub_ctx* ctx = ub_ctx_create_nopipe();
	if(!ctx)
		return NULL;
	/* no pipes, but we have the locks to make sure everything works */
	ctx->created_bg = 0;
	ctx->dothread = 1; /* the processing is in the same process,
		makes ub_cancel and ub_ctx_delete do the right thing */
	ctx->event_base = eb;
	return ctx;
}
	
/** delete q */
static void
delq(rbnode_t* n, void* ATTR_UNUSED(arg))
{
	struct ctx_query* q = (struct ctx_query*)n;
	context_query_delete(q);
}

/** stop the bg thread */
static void ub_stop_bg(struct ub_ctx* ctx)
{
	/* stop the bg thread */
	lock_basic_lock(&ctx->cfglock);
	if(ctx->created_bg) {
		uint8_t* msg;
		uint32_t len;
		uint32_t cmd = UB_LIBCMD_QUIT;
		lock_basic_unlock(&ctx->cfglock);
		lock_basic_lock(&ctx->qqpipe_lock);
		(void)tube_write_msg(ctx->qq_pipe, (uint8_t*)&cmd, 
			(uint32_t)sizeof(cmd), 0);
		lock_basic_unlock(&ctx->qqpipe_lock);
		lock_basic_lock(&ctx->rrpipe_lock);
		while(tube_read_msg(ctx->rr_pipe, &msg, &len, 0)) {
			/* discard all results except a quit confirm */
			if(context_serial_getcmd(msg, len) == UB_LIBCMD_QUIT) {
				free(msg);
				break;
			}
			free(msg);
		}
		lock_basic_unlock(&ctx->rrpipe_lock);

		/* if bg worker is a thread, wait for it to exit, so that all
	 	 * resources are really gone. */
		lock_basic_lock(&ctx->cfglock);
		if(ctx->dothread) {
			lock_basic_unlock(&ctx->cfglock);
			ub_thread_join(ctx->bg_tid);
		} else {
			lock_basic_unlock(&ctx->cfglock);
		}
	}
	else {
		lock_basic_unlock(&ctx->cfglock);
	}
}

void 
ub_ctx_delete(struct ub_ctx* ctx)
{
	struct alloc_cache* a, *na;
	int do_stop = 1;
	if(!ctx) return;

	/* see if bg thread is created and if threads have been killed */
	/* no locks, because those may be held by terminated threads */
	/* for processes the read pipe is closed and we see that on read */
#ifdef HAVE_PTHREAD
	if(ctx->created_bg && ctx->dothread) {
		if(pthread_kill(ctx->bg_tid, 0) == ESRCH) {
			/* thread has been killed */
			do_stop = 0;
		}
	}
#endif /* HAVE_PTHREAD */
	if(do_stop)
		ub_stop_bg(ctx);
	libworker_delete_event(ctx->event_worker);

	modstack_desetup(&ctx->mods, ctx->env);
	a = ctx->alloc_list;
	while(a) {
		na = a->super;
		a->super = &ctx->superalloc;
		alloc_clear(a);
		free(a);
		a = na;
	}
	local_zones_delete(ctx->local_zones);
	lock_basic_destroy(&ctx->qqpipe_lock);
	lock_basic_destroy(&ctx->rrpipe_lock);
	lock_basic_destroy(&ctx->cfglock);
	tube_delete(ctx->qq_pipe);
	tube_delete(ctx->rr_pipe);
	if(ctx->env) {
		slabhash_delete(ctx->env->msg_cache);
		rrset_cache_delete(ctx->env->rrset_cache);
		infra_delete(ctx->env->infra_cache);
		config_delete(ctx->env->cfg);
		free(ctx->env);
	}
	ub_randfree(ctx->seed_rnd);
	alloc_clear(&ctx->superalloc);
	traverse_postorder(&ctx->queries, delq, NULL);
	free(ctx);
#ifdef USE_WINSOCK
	WSACleanup();
#endif
}

int 
ub_ctx_set_option(struct ub_ctx* ctx, const char* opt, const char* val)
{
	lock_basic_lock(&ctx->cfglock);
	if(ctx->finalized) {
		lock_basic_unlock(&ctx->cfglock);
		return UB_AFTERFINAL;
	}
	if(!config_set_option(ctx->env->cfg, opt, val)) {
		lock_basic_unlock(&ctx->cfglock);
		return UB_SYNTAX;
	}
	lock_basic_unlock(&ctx->cfglock);
	return UB_NOERROR;
}

int
ub_ctx_get_option(struct ub_ctx* ctx, const char* opt, char** str)
{
	int r;
	lock_basic_lock(&ctx->cfglock);
	r = config_get_option_collate(ctx->env->cfg, opt, str);
	lock_basic_unlock(&ctx->cfglock);
	if(r == 0) r = UB_NOERROR;
	else if(r == 1) r = UB_SYNTAX;
	else if(r == 2) r = UB_NOMEM;
	return r;
}

int 
ub_ctx_config(struct ub_ctx* ctx, const char* fname)
{
	lock_basic_lock(&ctx->cfglock);
	if(ctx->finalized) {
		lock_basic_unlock(&ctx->cfglock);
		return UB_AFTERFINAL;
	}
	if(!config_read(ctx->env->cfg, fname, NULL)) {
		lock_basic_unlock(&ctx->cfglock);
		return UB_SYNTAX;
	}
	lock_basic_unlock(&ctx->cfglock);
	return UB_NOERROR;
}

int 
ub_ctx_add_ta(struct ub_ctx* ctx, const char* ta)
{
	char* dup = strdup(ta);
	if(!dup) return UB_NOMEM;
	lock_basic_lock(&ctx->cfglock);
	if(ctx->finalized) {
		lock_basic_unlock(&ctx->cfglock);
		free(dup);
		return UB_AFTERFINAL;
	}
	if(!cfg_strlist_insert(&ctx->env->cfg->trust_anchor_list, dup)) {
		lock_basic_unlock(&ctx->cfglock);
		free(dup);
		return UB_NOMEM;
	}
	lock_basic_unlock(&ctx->cfglock);
	return UB_NOERROR;
}

int 
ub_ctx_add_ta_file(struct ub_ctx* ctx, const char* fname)
{
	char* dup = strdup(fname);
	if(!dup) return UB_NOMEM;
	lock_basic_lock(&ctx->cfglock);
	if(ctx->finalized) {
		lock_basic_unlock(&ctx->cfglock);
		free(dup);
		return UB_AFTERFINAL;
	}
	if(!cfg_strlist_insert(&ctx->env->cfg->trust_anchor_file_list, dup)) {
		lock_basic_unlock(&ctx->cfglock);
		free(dup);
		return UB_NOMEM;
	}
	lock_basic_unlock(&ctx->cfglock);
	return UB_NOERROR;
}

int ub_ctx_add_ta_autr(struct ub_ctx* ctx, const char* fname)
{
	char* dup = strdup(fname);
	if(!dup) return UB_NOMEM;
	lock_basic_lock(&ctx->cfglock);
	if(ctx->finalized) {
		lock_basic_unlock(&ctx->cfglock);
		free(dup);
		return UB_AFTERFINAL;
	}
	if(!cfg_strlist_insert(&ctx->env->cfg->auto_trust_anchor_file_list,
		dup)) {
		lock_basic_unlock(&ctx->cfglock);
		free(dup);
		return UB_NOMEM;
	}
	lock_basic_unlock(&ctx->cfglock);
	return UB_NOERROR;
}

int 
ub_ctx_trustedkeys(struct ub_ctx* ctx, const char* fname)
{
	char* dup = strdup(fname);
	if(!dup) return UB_NOMEM;
	lock_basic_lock(&ctx->cfglock);
	if(ctx->finalized) {
		lock_basic_unlock(&ctx->cfglock);
		free(dup);
		return UB_AFTERFINAL;
	}
	if(!cfg_strlist_insert(&ctx->env->cfg->trusted_keys_file_list, dup)) {
		lock_basic_unlock(&ctx->cfglock);
		free(dup);
		return UB_NOMEM;
	}
	lock_basic_unlock(&ctx->cfglock);
	return UB_NOERROR;
}

int
ub_ctx_debuglevel(struct ub_ctx* ctx, int d)
{
	lock_basic_lock(&ctx->cfglock);
	verbosity = d;
	ctx->env->cfg->verbosity = d;
	lock_basic_unlock(&ctx->cfglock);
	return UB_NOERROR;
}

int ub_ctx_debugout(struct ub_ctx* ctx, void* out)
{
	lock_basic_lock(&ctx->cfglock);
	log_file((FILE*)out);
	ctx->logfile_override = 1;
	ctx->log_out = out;
	lock_basic_unlock(&ctx->cfglock);
	return UB_NOERROR;
}

int 
ub_ctx_async(struct ub_ctx* ctx, int dothread)
{
#ifdef THREADS_DISABLED
	if(dothread) /* cannot do threading */
		return UB_NOERROR;
#endif
	lock_basic_lock(&ctx->cfglock);
	if(ctx->finalized) {
		lock_basic_unlock(&ctx->cfglock);
		return UB_AFTERFINAL;
	}
	ctx->dothread = dothread;
	lock_basic_unlock(&ctx->cfglock);
	return UB_NOERROR;
}

int 
ub_poll(struct ub_ctx* ctx)
{
	/* no need to hold lock while testing for readability. */
	return tube_poll(ctx->rr_pipe);
}

int 
ub_fd(struct ub_ctx* ctx)
{
	return tube_read_fd(ctx->rr_pipe);
}

/** process answer from bg worker */
static int
process_answer_detail(struct ub_ctx* ctx, uint8_t* msg, uint32_t len,
	ub_callback_t* cb, void** cbarg, int* err,
	struct ub_result** res)
{
	struct ctx_query* q;
	if(context_serial_getcmd(msg, len) != UB_LIBCMD_ANSWER) {
		log_err("error: bad data from bg worker %d",
			(int)context_serial_getcmd(msg, len));
		return 0;
	}

	lock_basic_lock(&ctx->cfglock);
	q = context_deserialize_answer(ctx, msg, len, err);
	if(!q) {
		lock_basic_unlock(&ctx->cfglock);
		/* probably simply the lookup that failed, i.e.
		 * response returned before cancel was sent out, so noerror */
		return 1;
	}
	log_assert(q->async);

	/* grab cb while locked */
	if(q->cancelled) {
		*cb = NULL;
		*cbarg = NULL;
	} else {
		*cb = q->cb;
		*cbarg = q->cb_arg;
	}
	if(*err) {
		*res = NULL;
		ub_resolve_free(q->res);
	} else {
		/* parse the message, extract rcode, fill result */
		sldns_buffer* buf = sldns_buffer_new(q->msg_len);
		struct regional* region = regional_create();
		*res = q->res;
		(*res)->rcode = LDNS_RCODE_SERVFAIL;
		if(region && buf) {
			sldns_buffer_clear(buf);
			sldns_buffer_write(buf, q->msg, q->msg_len);
			sldns_buffer_flip(buf);
			libworker_enter_result(*res, buf, region,
				q->msg_security);
		}
		(*res)->answer_packet = q->msg;
		(*res)->answer_len = (int)q->msg_len;
		q->msg = NULL;
		sldns_buffer_free(buf);
		regional_destroy(region);
	}
	q->res = NULL;
	/* delete the q from list */
	(void)rbtree_delete(&ctx->queries, q->node.key);
	ctx->num_async--;
	context_query_delete(q);
	lock_basic_unlock(&ctx->cfglock);

	if(*cb) return 2;
	ub_resolve_free(*res);
	return 1;
}

/** process answer from bg worker */
static int
process_answer(struct ub_ctx* ctx, uint8_t* msg, uint32_t len)
{
	int err;
	ub_callback_t cb;
	void* cbarg;
	struct ub_result* res;
	int r;

	r = process_answer_detail(ctx, msg, len, &cb, &cbarg, &err, &res);

	/* no locks held while calling callback, so that library is
	 * re-entrant. */
	if(r == 2)
		(*cb)(cbarg, err, res);

	return r;
}

int 
ub_process(struct ub_ctx* ctx)
{
	int r;
	uint8_t* msg;
	uint32_t len;
	while(1) {
		msg = NULL;
		lock_basic_lock(&ctx->rrpipe_lock);
		r = tube_read_msg(ctx->rr_pipe, &msg, &len, 1);
		lock_basic_unlock(&ctx->rrpipe_lock);
		if(r == 0)
			return UB_PIPE;
		else if(r == -1)
			break;
		if(!process_answer(ctx, msg, len)) {
			free(msg);
			return UB_PIPE;
		}
		free(msg);
	}
	return UB_NOERROR;
}

int 
ub_wait(struct ub_ctx* ctx)
{
	int err;
	ub_callback_t cb;
	void* cbarg;
	struct ub_result* res;
	int r;
	uint8_t* msg;
	uint32_t len;
	/* this is basically the same loop as _process(), but with changes.
	 * holds the rrpipe lock and waits with tube_wait */
	while(1) {
		lock_basic_lock(&ctx->rrpipe_lock);
		lock_basic_lock(&ctx->cfglock);
		if(ctx->num_async == 0) {
			lock_basic_unlock(&ctx->cfglock);
			lock_basic_unlock(&ctx->rrpipe_lock);
			break;
		}
		lock_basic_unlock(&ctx->cfglock);

		/* keep rrpipe locked, while
		 * 	o waiting for pipe readable
		 * 	o parsing message
		 * 	o possibly decrementing num_async
		 * do callback without lock
		 */
		r = tube_wait(ctx->rr_pipe);
		if(r) {
			r = tube_read_msg(ctx->rr_pipe, &msg, &len, 1);
			if(r == 0) {
				lock_basic_unlock(&ctx->rrpipe_lock);
				return UB_PIPE;
			}
			if(r == -1) {
				lock_basic_unlock(&ctx->rrpipe_lock);
				continue;
			}
			r = process_answer_detail(ctx, msg, len, 
				&cb, &cbarg, &err, &res);
			lock_basic_unlock(&ctx->rrpipe_lock);
			free(msg);
			if(r == 0)
				return UB_PIPE;
			if(r == 2)
				(*cb)(cbarg, err, res);
		} else {
			lock_basic_unlock(&ctx->rrpipe_lock);
		}
	}
	return UB_NOERROR;
}

int 
ub_resolve(struct ub_ctx* ctx, const char* name, int rrtype, 
	int rrclass, struct ub_result** result)
{
	struct ctx_query* q;
	int r;
	*result = NULL;

	lock_basic_lock(&ctx->cfglock);
	if(!ctx->finalized) {
		r = context_finalize(ctx);
		if(r) {
			lock_basic_unlock(&ctx->cfglock);
			return r;
		}
	}
	/* create new ctx_query and attempt to add to the list */
	lock_basic_unlock(&ctx->cfglock);
	q = context_new(ctx, name, rrtype, rrclass, NULL, NULL);
	if(!q)
		return UB_NOMEM;
	/* become a resolver thread for a bit */

	r = libworker_fg(ctx, q);
	if(r) {
		lock_basic_lock(&ctx->cfglock);
		(void)rbtree_delete(&ctx->queries, q->node.key);
		context_query_delete(q);
		lock_basic_unlock(&ctx->cfglock);
		return r;
	}
	q->res->answer_packet = q->msg;
	q->res->answer_len = (int)q->msg_len;
	q->msg = NULL;
	*result = q->res;
	q->res = NULL;

	lock_basic_lock(&ctx->cfglock);
	(void)rbtree_delete(&ctx->queries, q->node.key);
	context_query_delete(q);
	lock_basic_unlock(&ctx->cfglock);
	return UB_NOERROR;
}

int 
ub_resolve_event(struct ub_ctx* ctx, const char* name, int rrtype, 
	int rrclass, void* mydata, ub_event_callback_t callback, int* async_id)
{
	struct ctx_query* q;
	int r;

	if(async_id)
		*async_id = 0;
	lock_basic_lock(&ctx->cfglock);
	if(!ctx->finalized) {
		int r = context_finalize(ctx);
		if(r) {
			lock_basic_unlock(&ctx->cfglock);
			return r;
		}
	}
	lock_basic_unlock(&ctx->cfglock);
	if(!ctx->event_worker) {
		ctx->event_worker = libworker_create_event(ctx,
			ctx->event_base);
		if(!ctx->event_worker) {
			return UB_INITFAIL;
		}
	}

	/* create new ctx_query and attempt to add to the list */
	q = context_new(ctx, name, rrtype, rrclass, (ub_callback_t)callback,
		mydata);
	if(!q)
		return UB_NOMEM;

	/* attach to mesh */
	if((r=libworker_attach_mesh(ctx, q, async_id)) != 0)
		return r;
	return UB_NOERROR;
}


int 
ub_resolve_async(struct ub_ctx* ctx, const char* name, int rrtype, 
	int rrclass, void* mydata, ub_callback_t callback, int* async_id)
{
	struct ctx_query* q;
	uint8_t* msg = NULL;
	uint32_t len = 0;

	if(async_id)
		*async_id = 0;
	lock_basic_lock(&ctx->cfglock);
	if(!ctx->finalized) {
		int r = context_finalize(ctx);
		if(r) {
			lock_basic_unlock(&ctx->cfglock);
			return r;
		}
	}
	if(!ctx->created_bg) {
		int r;
		ctx->created_bg = 1;
		lock_basic_unlock(&ctx->cfglock);
		r = libworker_bg(ctx);
		if(r) {
			lock_basic_lock(&ctx->cfglock);
			ctx->created_bg = 0;
			lock_basic_unlock(&ctx->cfglock);
			return r;
		}
	} else {
		lock_basic_unlock(&ctx->cfglock);
	}

	/* create new ctx_query and attempt to add to the list */
	q = context_new(ctx, name, rrtype, rrclass, callback, mydata);
	if(!q)
		return UB_NOMEM;

	/* write over pipe to background worker */
	lock_basic_lock(&ctx->cfglock);
	msg = context_serialize_new_query(q, &len);
	if(!msg) {
		(void)rbtree_delete(&ctx->queries, q->node.key);
		ctx->num_async--;
		context_query_delete(q);
		lock_basic_unlock(&ctx->cfglock);
		return UB_NOMEM;
	}
	if(async_id)
		*async_id = q->querynum;
	lock_basic_unlock(&ctx->cfglock);
	
	lock_basic_lock(&ctx->qqpipe_lock);
	if(!tube_write_msg(ctx->qq_pipe, msg, len, 0)) {
		lock_basic_unlock(&ctx->qqpipe_lock);
		free(msg);
		return UB_PIPE;
	}
	lock_basic_unlock(&ctx->qqpipe_lock);
	free(msg);
	return UB_NOERROR;
}

int 
ub_cancel(struct ub_ctx* ctx, int async_id)
{
	struct ctx_query* q;
	uint8_t* msg = NULL;
	uint32_t len = 0;
	lock_basic_lock(&ctx->cfglock);
	q = (struct ctx_query*)rbtree_search(&ctx->queries, &async_id);
	if(!q || !q->async) {
		/* it is not there, so nothing to do */
		lock_basic_unlock(&ctx->cfglock);
		return UB_NOID;
	}
	log_assert(q->async);
	q->cancelled = 1;
	
	/* delete it */
	if(!ctx->dothread) { /* if forked */
		(void)rbtree_delete(&ctx->queries, q->node.key);
		ctx->num_async--;
		msg = context_serialize_cancel(q, &len);
		context_query_delete(q);
		lock_basic_unlock(&ctx->cfglock);
		if(!msg) {
			return UB_NOMEM;
		}
		/* send cancel to background worker */
		lock_basic_lock(&ctx->qqpipe_lock);
		if(!tube_write_msg(ctx->qq_pipe, msg, len, 0)) {
			lock_basic_unlock(&ctx->qqpipe_lock);
			free(msg);
			return UB_PIPE;
		}
		lock_basic_unlock(&ctx->qqpipe_lock);
		free(msg);
	} else {
		lock_basic_unlock(&ctx->cfglock);
	}
	return UB_NOERROR;
}

void 
ub_resolve_free(struct ub_result* result)
{
	char** p;
	if(!result) return;
	free(result->qname);
	if(result->canonname != result->qname)
		free(result->canonname);
	if(result->data)
		for(p = result->data; *p; p++)
			free(*p);
	free(result->data);
	free(result->len);
	free(result->answer_packet);
	free(result->why_bogus);
	free(result);
}

const char* 
ub_strerror(int err)
{
	switch(err) {
		case UB_NOERROR: return "no error";
		case UB_SOCKET: return "socket io error";
		case UB_NOMEM: return "out of memory";
		case UB_SYNTAX: return "syntax error";
		case UB_SERVFAIL: return "server failure";
		case UB_FORKFAIL: return "could not fork";
		case UB_INITFAIL: return "initialization failure";
		case UB_AFTERFINAL: return "setting change after finalize";
		case UB_PIPE: return "error in pipe communication with async";
		case UB_READFILE: return "error reading file";
		case UB_NOID: return "error async_id does not exist";
		default: return "unknown error";
	}
}

int 
ub_ctx_set_fwd(struct ub_ctx* ctx, const char* addr)
{
	struct sockaddr_storage storage;
	socklen_t stlen;
	struct config_stub* s;
	char* dupl;
	lock_basic_lock(&ctx->cfglock);
	if(ctx->finalized) {
		lock_basic_unlock(&ctx->cfglock);
		errno=EINVAL;
		return UB_AFTERFINAL;
	}
	if(!addr) {
		/* disable fwd mode - the root stub should be first. */
		if(ctx->env->cfg->forwards &&
			strcmp(ctx->env->cfg->forwards->name, ".") == 0) {
			s = ctx->env->cfg->forwards;
			ctx->env->cfg->forwards = s->next;
			s->next = NULL;
			config_delstubs(s);
		}
		lock_basic_unlock(&ctx->cfglock);
		return UB_NOERROR;
	}
	lock_basic_unlock(&ctx->cfglock);

	/* check syntax for addr */
	if(!extstrtoaddr(addr, &storage, &stlen)) {
		errno=EINVAL;
		return UB_SYNTAX;
	}
	
	/* it parses, add root stub in front of list */
	lock_basic_lock(&ctx->cfglock);
	if(!ctx->env->cfg->forwards ||
		strcmp(ctx->env->cfg->forwards->name, ".") != 0) {
		s = calloc(1, sizeof(*s));
		if(!s) {
			lock_basic_unlock(&ctx->cfglock);
			errno=ENOMEM;
			return UB_NOMEM;
		}
		s->name = strdup(".");
		if(!s->name) {
			free(s);
			lock_basic_unlock(&ctx->cfglock);
			errno=ENOMEM;
			return UB_NOMEM;
		}
		s->next = ctx->env->cfg->forwards;
		ctx->env->cfg->forwards = s;
	} else {
		log_assert(ctx->env->cfg->forwards);
		s = ctx->env->cfg->forwards;
	}
	dupl = strdup(addr);
	if(!dupl) {
		lock_basic_unlock(&ctx->cfglock);
		errno=ENOMEM;
		return UB_NOMEM;
	}
	if(!cfg_strlist_insert(&s->addrs, dupl)) {
		free(dupl);
		lock_basic_unlock(&ctx->cfglock);
		errno=ENOMEM;
		return UB_NOMEM;
	}
	lock_basic_unlock(&ctx->cfglock);
	return UB_NOERROR;
}

int 
ub_ctx_resolvconf(struct ub_ctx* ctx, const char* fname)
{
	FILE* in;
	int numserv = 0;
	char buf[1024];
	char* parse, *addr;
	int r;

	if(fname == NULL) {
#if !defined(UB_ON_WINDOWS) || !defined(HAVE_WINDOWS_H)
		fname = "/etc/resolv.conf";
#else
		FIXED_INFO *info;
		ULONG buflen = sizeof(*info);
		IP_ADDR_STRING *ptr;

		info = (FIXED_INFO *) malloc(sizeof (FIXED_INFO));
		if (info == NULL) 
			return UB_READFILE;

		if (GetNetworkParams(info, &buflen) == ERROR_BUFFER_OVERFLOW) {
			free(info);
			info = (FIXED_INFO *) malloc(buflen);
			if (info == NULL)
				return UB_READFILE;
		}

		if (GetNetworkParams(info, &buflen) == NO_ERROR) {
			int retval=0;
			ptr = &(info->DnsServerList);
			while (ptr) {
				numserv++;
				if((retval=ub_ctx_set_fwd(ctx, 
					ptr->IpAddress.String)!=0)) {
					free(info);
					return retval;
				}
				ptr = ptr->Next;
			}
			free(info);
			if (numserv==0)
				return UB_READFILE;
			return UB_NOERROR;
		}
		free(info);
		return UB_READFILE;
#endif /* WINDOWS */
	}
	in = fopen(fname, "r");
	if(!in) {
		/* error in errno! perror(fname) */
		return UB_READFILE;
	}
	while(fgets(buf, (int)sizeof(buf), in)) {
		buf[sizeof(buf)-1] = 0;
		parse=buf;
		while(*parse == ' ' || *parse == '\t')
			parse++;
		if(strncmp(parse, "nameserver", 10) == 0) {
			numserv++;
			parse += 10; /* skip 'nameserver' */
			/* skip whitespace */
			while(*parse == ' ' || *parse == '\t')
				parse++;
			addr = parse;
			/* skip [0-9a-fA-F.:]*, i.e. IP4 and IP6 address */
			while(isxdigit((unsigned char)*parse) || *parse=='.' || *parse==':')
				parse++;
			/* terminate after the address, remove newline */
			*parse = 0;
			
			if((r = ub_ctx_set_fwd(ctx, addr)) != UB_NOERROR) {
				fclose(in);
				return r;
			}
		}
	}
	fclose(in);
	if(numserv == 0) {
		/* from resolv.conf(5) if none given, use localhost */
		return ub_ctx_set_fwd(ctx, "127.0.0.1");
	}
	return UB_NOERROR;
}

int
ub_ctx_hosts(struct ub_ctx* ctx, const char* fname)
{
	FILE* in;
	char buf[1024], ldata[1024];
	char* parse, *addr, *name, *ins;
	lock_basic_lock(&ctx->cfglock);
	if(ctx->finalized) {
		lock_basic_unlock(&ctx->cfglock);
		errno=EINVAL;
		return UB_AFTERFINAL;
	}
	lock_basic_unlock(&ctx->cfglock);
	if(fname == NULL) {
#if defined(UB_ON_WINDOWS) && defined(HAVE_WINDOWS_H)
		/*
		 * If this is Windows NT/XP/2K it's in
		 * %WINDIR%\system32\drivers\etc\hosts.
		 * If this is Windows 95/98/Me it's in %WINDIR%\hosts.
		 */
		name = getenv("WINDIR");
		if (name != NULL) {
			int retval=0;
			snprintf(buf, sizeof(buf), "%s%s", name, 
				"\\system32\\drivers\\etc\\hosts");
			if((retval=ub_ctx_hosts(ctx, buf)) !=0 ) {
				snprintf(buf, sizeof(buf), "%s%s", name, 
					"\\hosts");
				retval=ub_ctx_hosts(ctx, buf);
			}
			return retval;
		}
		return UB_READFILE;
#else
		fname = "/etc/hosts";
#endif /* WIN32 */
	}
	in = fopen(fname, "r");
	if(!in) {
		/* error in errno! perror(fname) */
		return UB_READFILE;
	}
	while(fgets(buf, (int)sizeof(buf), in)) {
		buf[sizeof(buf)-1] = 0;
		parse=buf;
		while(*parse == ' ' || *parse == '\t')
			parse++;
		if(*parse == '#')
			continue; /* skip comment */
		/* format: <addr> spaces <name> spaces <name> ... */
		addr = parse;
		/* skip addr */
		while(isxdigit((unsigned char)*parse) || *parse == '.' || *parse == ':')
			parse++;
		if(*parse == '\r')
			parse++;
		if(*parse == '\n' || *parse == 0)
			continue;
		if(*parse == '%') 
			continue; /* ignore macOSX fe80::1%lo0 localhost */
		if(*parse != ' ' && *parse != '\t') {
			/* must have whitespace after address */
			fclose(in);
			errno=EINVAL;
			return UB_SYNTAX;
		}
		*parse++ = 0; /* end delimiter for addr ... */
		/* go to names and add them */
		while(*parse) {
			while(*parse == ' ' || *parse == '\t' || *parse=='\n'
				|| *parse=='\r')
				parse++;
			if(*parse == 0 || *parse == '#')
				break;
			/* skip name, allows (too) many printable characters */
			name = parse;
			while('!' <= *parse && *parse <= '~')
				parse++;
			if(*parse)
				*parse++ = 0; /* end delimiter for name */
			snprintf(ldata, sizeof(ldata), "%s %s %s",
				name, str_is_ip6(addr)?"AAAA":"A", addr);
			ins = strdup(ldata);
			if(!ins) {
				/* out of memory */
				fclose(in);
				errno=ENOMEM;
				return UB_NOMEM;
			}
			lock_basic_lock(&ctx->cfglock);
			if(!cfg_strlist_insert(&ctx->env->cfg->local_data, 
				ins)) {
				lock_basic_unlock(&ctx->cfglock);
				fclose(in);
				free(ins);
				errno=ENOMEM;
				return UB_NOMEM;
			}
			lock_basic_unlock(&ctx->cfglock);
		}
	}
	fclose(in);
	return UB_NOERROR;
}

/** finalize the context, if not already finalized */
static int ub_ctx_finalize(struct ub_ctx* ctx)
{
	int res = 0;
	lock_basic_lock(&ctx->cfglock);
	if (!ctx->finalized) {
		res = context_finalize(ctx);
	}
	lock_basic_unlock(&ctx->cfglock);
	return res;
}

/* Print local zones and RR data */
int ub_ctx_print_local_zones(struct ub_ctx* ctx)
{   
	int res = ub_ctx_finalize(ctx);
	if (res) return res;

	local_zones_print(ctx->local_zones);

	return UB_NOERROR;
}

/* Add a new zone */
int ub_ctx_zone_add(struct ub_ctx* ctx, const char *zone_name, 
	const char *zone_type)
{
	enum localzone_type t;
	struct local_zone* z;
	uint8_t* nm;
	int nmlabs;
	size_t nmlen;

	int res = ub_ctx_finalize(ctx);
	if (res) return res;

	if(!local_zone_str2type(zone_type, &t)) {
		return UB_SYNTAX;
	}

	if(!parse_dname(zone_name, &nm, &nmlen, &nmlabs)) {
		return UB_SYNTAX;
	}

	lock_rw_wrlock(&ctx->local_zones->lock);
	if((z=local_zones_find(ctx->local_zones, nm, nmlen, nmlabs, 
		LDNS_RR_CLASS_IN))) {
		/* already present in tree */
		lock_rw_wrlock(&z->lock);
		z->type = t; /* update type anyway */
		lock_rw_unlock(&z->lock);
		lock_rw_unlock(&ctx->local_zones->lock);
		free(nm);
		return UB_NOERROR;
	}
	if(!local_zones_add_zone(ctx->local_zones, nm, nmlen, nmlabs, 
		LDNS_RR_CLASS_IN, t)) {
		lock_rw_unlock(&ctx->local_zones->lock);
		return UB_NOMEM;
	}
	lock_rw_unlock(&ctx->local_zones->lock);
	return UB_NOERROR;
}

/* Remove zone */
int ub_ctx_zone_remove(struct ub_ctx* ctx, const char *zone_name)
{   
	struct local_zone* z;
	uint8_t* nm;
	int nmlabs;
	size_t nmlen;

	int res = ub_ctx_finalize(ctx);
	if (res) return res;

	if(!parse_dname(zone_name, &nm, &nmlen, &nmlabs)) {
		return UB_SYNTAX;
	}

	lock_rw_wrlock(&ctx->local_zones->lock);
	if((z=local_zones_find(ctx->local_zones, nm, nmlen, nmlabs, 
		LDNS_RR_CLASS_IN))) {
		/* present in tree */
		local_zones_del_zone(ctx->local_zones, z);
	}
	lock_rw_unlock(&ctx->local_zones->lock);
	free(nm);
	return UB_NOERROR;
}

/* Add new RR data */
int ub_ctx_data_add(struct ub_ctx* ctx, const char *data)
{
	int res = ub_ctx_finalize(ctx);
	if (res) return res;

	res = local_zones_add_RR(ctx->local_zones, data);
	return (!res) ? UB_NOMEM : UB_NOERROR;
}

/* Remove RR data */
int ub_ctx_data_remove(struct ub_ctx* ctx, const char *data)
{
	uint8_t* nm;
	int nmlabs;
	size_t nmlen;
	int res = ub_ctx_finalize(ctx);
	if (res) return res;

	if(!parse_dname(data, &nm, &nmlen, &nmlabs)) 
		return UB_SYNTAX;

	local_zones_del_data(ctx->local_zones, nm, nmlen, nmlabs, 
		LDNS_RR_CLASS_IN);

	free(nm);
	return UB_NOERROR;
}

const char* ub_version(void)
{
	return PACKAGE_VERSION;
}

int 
ub_ctx_set_event(struct ub_ctx* ctx, struct event_base* base) {
	if (!ctx || !ctx->event_base || !base) {
		return UB_INITFAIL;
	}
	if (ctx->event_base == base) {
		/* already set */
		return UB_NOERROR;
	}
	
	lock_basic_lock(&ctx->cfglock);
	/* destroy the current worker - safe to pass in NULL */
	libworker_delete_event(ctx->event_worker);
	ctx->event_worker = NULL;
	ctx->event_base = base;	
	ctx->created_bg = 0;
	ctx->dothread = 1;
	lock_basic_unlock(&ctx->cfglock);
	return UB_NOERROR;
}