Training courses

Kernel and Embedded Linux

Bootlin training courses

Embedded Linux, kernel,
Yocto Project, Buildroot, real-time,
graphics, boot time, debugging...

Bootlin logo

Elixir Cross Referencer

   1
   2
   3
   4
   5
   6
   7
   8
   9
  10
  11
  12
  13
  14
  15
  16
  17
  18
  19
  20
  21
  22
  23
  24
  25
  26
  27
  28
  29
  30
  31
  32
  33
  34
  35
  36
  37
  38
  39
  40
  41
  42
  43
  44
  45
  46
  47
  48
  49
  50
  51
  52
  53
  54
  55
  56
  57
  58
  59
  60
  61
  62
  63
  64
  65
  66
  67
  68
  69
  70
  71
  72
  73
  74
  75
  76
  77
  78
  79
  80
  81
  82
  83
  84
  85
  86
  87
  88
  89
  90
  91
  92
  93
  94
  95
  96
  97
  98
  99
 100
 101
 102
 103
 104
 105
 106
 107
 108
 109
 110
 111
 112
 113
 114
 115
 116
 117
 118
 119
 120
 121
 122
 123
 124
 125
 126
 127
 128
 129
 130
 131
 132
 133
 134
 135
 136
 137
 138
 139
 140
 141
 142
 143
 144
 145
 146
 147
 148
 149
 150
 151
 152
 153
 154
 155
 156
 157
 158
 159
 160
 161
 162
 163
 164
 165
 166
 167
 168
 169
 170
 171
 172
 173
 174
 175
 176
 177
 178
 179
 180
 181
 182
 183
 184
 185
 186
 187
 188
 189
 190
 191
 192
 193
 194
 195
 196
 197
 198
 199
 200
 201
 202
 203
 204
 205
 206
 207
 208
 209
 210
 211
 212
 213
 214
 215
 216
 217
 218
 219
 220
 221
 222
 223
 224
 225
 226
 227
 228
 229
 230
 231
 232
 233
 234
 235
 236
 237
 238
 239
 240
 241
 242
 243
 244
 245
 246
 247
 248
 249
 250
 251
 252
 253
 254
 255
 256
 257
 258
 259
 260
 261
 262
 263
 264
 265
 266
 267
 268
 269
 270
 271
 272
 273
 274
 275
 276
 277
 278
 279
 280
 281
 282
 283
 284
 285
 286
 287
 288
 289
 290
 291
 292
 293
 294
 295
 296
 297
 298
 299
 300
 301
 302
 303
 304
 305
 306
 307
 308
 309
 310
 311
 312
 313
 314
 315
 316
 317
 318
 319
 320
 321
 322
 323
 324
 325
 326
 327
 328
 329
 330
 331
 332
 333
 334
 335
 336
 337
 338
 339
 340
 341
 342
 343
 344
 345
 346
 347
 348
 349
 350
 351
 352
 353
 354
 355
 356
 357
 358
 359
 360
 361
 362
 363
 364
 365
 366
 367
 368
 369
 370
 371
 372
 373
 374
 375
 376
 377
 378
 379
 380
 381
 382
 383
 384
 385
 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
 982
 983
 984
 985
 986
 987
 988
 989
 990
 991
 992
 993
 994
 995
 996
 997
 998
 999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
/*
 * ntp_intres.c - Implements a generic blocking worker child or thread,
 *		  initially to provide a nonblocking solution for DNS
 *		  name to address lookups available with getaddrinfo().
 *
 * This is a new implementation as of 2009 sharing the filename and
 * very little else with the prior implementation, which used a
 * temporary file to receive a single set of requests from the parent,
 * and a NTP mode 7 authenticated request to push back responses.
 *
 * A primary goal in rewriting this code was the need to support the
 * pool configuration directive's requirement to retrieve multiple
 * addresses resolving a single name, which has previously been
 * satisfied with blocking resolver calls from the ntpd mainline code.
 *
 * A secondary goal is to provide a generic mechanism for other
 * blocking operations to be delegated to a worker using a common
 * model for both Unix and Windows ntpd.  ntp_worker.c, work_fork.c,
 * and work_thread.c implement the generic mechanism.  This file
 * implements the two current consumers, getaddrinfo_sometime() and the
 * presently unused getnameinfo_sometime().
 *
 * Both routines deliver results to a callback and manage memory
 * allocation, meaning there is no freeaddrinfo_sometime().
 *
 * The initial implementation for Unix uses a pair of unidirectional
 * pipes, one each for requests and responses, connecting the forked
 * blocking child worker with the ntpd mainline.  The threaded code
 * uses arrays of pointers to queue requests and responses.
 *
 * The parent drives the process, including scheduling sleeps between
 * retries.
 *
 * Memory is managed differently for a child process, which mallocs
 * request buffers to read from the pipe into, whereas the threaded
 * code mallocs a copy of the request to hand off to the worker via
 * the queueing array.  The resulting request buffer is free()d by
 * platform-independent code.  A wrinkle is the request needs to be
 * available to the requestor during response processing.
 *
 * Response memory allocation is also platform-dependent.  With a
 * separate process and pipes, the response is free()d after being
 * written to the pipe.  With threads, the same memory is handed
 * over and the requestor frees it after processing is completed.
 *
 * The code should be generalized to support threads on Unix using
 * much of the same code used for Windows initially.
 *
 */
#ifdef HAVE_CONFIG_H
# include <config.h>
#endif

#include "ntp_workimpl.h"

#ifdef WORKER

#include <stdio.h>
#include <ctype.h>
#include <signal.h>

/**/
#ifdef HAVE_SYS_TYPES_H
# include <sys/types.h>
#endif
#ifdef HAVE_NETINET_IN_H
#include <netinet/in.h>
#endif
#include <arpa/inet.h>
/**/
#ifdef HAVE_SYS_PARAM_H
# include <sys/param.h>
#endif

#if !defined(HAVE_RES_INIT) && defined(HAVE___RES_INIT)
# define HAVE_RES_INIT
#endif

#if defined(HAVE_RESOLV_H) && defined(HAVE_RES_INIT)
# ifdef HAVE_ARPA_NAMESER_H
#  include <arpa/nameser.h> /* DNS HEADER struct */
# endif
# ifdef HAVE_NETDB_H
#  include <netdb.h>
# endif
# include <resolv.h>
# ifdef HAVE_INT32_ONLY_WITH_DNS
#  define HAVE_INT32
# endif
# ifdef HAVE_U_INT32_ONLY_WITH_DNS
#  define HAVE_U_INT32
# endif
#endif

#include "ntp.h"
#include "ntp_debug.h"
#include "ntp_malloc.h"
#include "ntp_syslog.h"
#include "ntp_unixtime.h"
#include "ntp_intres.h"
#include "intreswork.h"


/*
 * Following are implementations of getaddrinfo_sometime() and
 * getnameinfo_sometime().  Each is implemented in three routines:
 *
 * getaddrinfo_sometime()		getnameinfo_sometime()
 * blocking_getaddrinfo()		blocking_getnameinfo()
 * getaddrinfo_sometime_complete()	getnameinfo_sometime_complete()
 *
 * The first runs in the parent and marshalls (or serializes) request
 * parameters into a request blob which is processed in the child by
 * the second routine, blocking_*(), which serializes the results into
 * a response blob unpacked by the third routine, *_complete(), which
 * calls the callback routine provided with the request and frees
 * _request_ memory allocated by the first routine.  Response memory
 * is managed by the code which calls the *_complete routines.
 */


/* === typedefs === */
typedef struct blocking_gai_req_tag {	/* marshalled args */
	size_t			octets;
	u_int			dns_idx;
	time_t			scheduled;
	time_t			earliest;
	int			retry;
	struct addrinfo		hints;
	u_int			qflags;
	gai_sometime_callback	callback;
	void *			context;
	size_t			nodesize;
	size_t			servsize;
} blocking_gai_req;

typedef struct blocking_gai_resp_tag {
	size_t			octets;
	int			retcode;
	int			retry;
	int			gai_errno; /* for EAI_SYSTEM case */
	int			ai_count;
	/*
	 * Followed by ai_count struct addrinfo and then ai_count
	 * sockaddr_u and finally the canonical name strings.
	 */
} blocking_gai_resp;

typedef struct blocking_gni_req_tag {
	size_t			octets;
	u_int			dns_idx;
	time_t			scheduled;
	time_t			earliest;
	int			retry;
	size_t			hostoctets;
	size_t			servoctets;
	int			flags;
	gni_sometime_callback	callback;
	void *			context;
	sockaddr_u		socku;
} blocking_gni_req;

typedef struct blocking_gni_resp_tag {
	size_t			octets;
	int			retcode;
	int			gni_errno; /* for EAI_SYSTEM case */
	int			retry;
	size_t			hostoctets;
	size_t			servoctets;
	/*
	 * Followed by hostoctets bytes of null-terminated host,
	 * then servoctets bytes of null-terminated service.
	 */
} blocking_gni_resp;

/* per-DNS-worker state in parent */
typedef struct dnschild_ctx_tag {
	u_int	index;
	time_t	next_dns_timeslot;
} dnschild_ctx;

/* per-DNS-worker state in worker */
typedef struct dnsworker_ctx_tag {
	blocking_child *	c;
	time_t			ignore_scheduled_before;
#ifdef HAVE_RES_INIT
	time_t	next_res_init;
#endif
} dnsworker_ctx;


/* === variables === */
dnschild_ctx **		dnschild_contexts;		/* parent */
u_int			dnschild_contexts_alloc;
dnsworker_ctx **	dnsworker_contexts;		/* child */
u_int			dnsworker_contexts_alloc;

#ifdef HAVE_RES_INIT
static	time_t		next_res_init;
#endif


/* === forward declarations === */
static	u_int		reserve_dnschild_ctx(void);
static	u_int		get_dnschild_ctx(void);
static	dnsworker_ctx *	get_worker_context(blocking_child *, u_int);
static	void		scheduled_sleep(time_t, time_t,
					dnsworker_ctx *);
static	void		manage_dns_retry_interval(time_t *, time_t *,
						  int *, time_t *,
						  int/*BOOL*/);
static	int		should_retry_dns(int, int);
#ifdef HAVE_RES_INIT
static	void		reload_resolv_conf(dnsworker_ctx *);
#else
# define		reload_resolv_conf(wc)		\
	do {						\
		(void)(wc);				\
	} while (FALSE)
#endif
static	void		getaddrinfo_sometime_complete(blocking_work_req,
						      void *, size_t,
						      void *);
static	void		getnameinfo_sometime_complete(blocking_work_req,
						      void *, size_t,
						      void *);


/* === functions === */
/*
 * getaddrinfo_sometime - uses blocking child to call getaddrinfo then
 *			  invokes provided callback completion function.
 */
int
getaddrinfo_sometime_ex(
	const char *		node,
	const char *		service,
	const struct addrinfo *	hints,
	int			retry,
	gai_sometime_callback	callback,
	void *			context,
	u_int			qflags
	)
{
	blocking_gai_req *	gai_req;
	u_int			idx;
	dnschild_ctx *		child_ctx;
	size_t			req_size;
	size_t			nodesize;
	size_t			servsize;
	time_t			now;
	
	REQUIRE(NULL != node);
	if (NULL != hints) {
		REQUIRE(0 == hints->ai_addrlen);
		REQUIRE(NULL == hints->ai_addr);
		REQUIRE(NULL == hints->ai_canonname);
		REQUIRE(NULL == hints->ai_next);
	}

	idx = get_dnschild_ctx();
	child_ctx = dnschild_contexts[idx];

	nodesize = strlen(node) + 1;
	servsize = strlen(service) + 1;
	req_size = sizeof(*gai_req) + nodesize + servsize;

	gai_req = emalloc_zero(req_size);

	gai_req->octets = req_size;
	gai_req->dns_idx = idx;
	now = time(NULL);
	gai_req->scheduled = now;
	gai_req->earliest = max(now, child_ctx->next_dns_timeslot);
	child_ctx->next_dns_timeslot = gai_req->earliest;
	if (hints != NULL)
		gai_req->hints = *hints;
	gai_req->retry = retry;
	gai_req->callback = callback;
	gai_req->context = context;
	gai_req->nodesize = nodesize;
	gai_req->servsize = servsize;
	gai_req->qflags = qflags;

	memcpy((char *)gai_req + sizeof(*gai_req), node, nodesize);
	memcpy((char *)gai_req + sizeof(*gai_req) + nodesize, service,
	       servsize);

	if (queue_blocking_request(
		BLOCKING_GETADDRINFO,
		gai_req,
		req_size, 
		&getaddrinfo_sometime_complete, 
		gai_req)) {

		msyslog(LOG_ERR, "unable to queue getaddrinfo request");
		errno = EFAULT;
		return -1;
	}

	return 0;
}

int
blocking_getaddrinfo(
	blocking_child *	c,
	blocking_pipe_header *	req
	)
{
	blocking_gai_req *	gai_req;
	dnsworker_ctx *		worker_ctx;
	blocking_pipe_header *	resp;
	blocking_gai_resp *	gai_resp;
	char *			node;
	char *			service;
	struct addrinfo *	ai_res;
	struct addrinfo *	ai;
	struct addrinfo *	serialized_ai;
	size_t			canons_octets;
	size_t			this_octets;
	size_t			resp_octets;
	char *			cp;
	time_t			time_now;

	gai_req = (void *)((char *)req + sizeof(*req));
	node = (char *)gai_req + sizeof(*gai_req);
	service = node + gai_req->nodesize;

	worker_ctx = get_worker_context(c, gai_req->dns_idx);
	scheduled_sleep(gai_req->scheduled, gai_req->earliest,
			worker_ctx);
	reload_resolv_conf(worker_ctx);

	/*
	 * Take a shot at the final size, better to overestimate
	 * at first and then realloc to a smaller size.
	 */

	resp_octets = sizeof(*resp) + sizeof(*gai_resp) +
		      16 * (sizeof(struct addrinfo) +
			    sizeof(sockaddr_u)) +
		      256;
	resp = emalloc_zero(resp_octets);
	gai_resp = (void *)(resp + 1);

	TRACE(2, ("blocking_getaddrinfo given node %s serv %s fam %d flags %x\n", 
		  node, service, gai_req->hints.ai_family,
		  gai_req->hints.ai_flags));
#ifdef DEBUG
	if (debug >= 2)
		fflush(stdout);
#endif	
	ai_res = NULL;
	gai_resp->retcode = getaddrinfo(node, service, &gai_req->hints,
					&ai_res);
	gai_resp->retry = gai_req->retry;
#ifdef EAI_SYSTEM
	if (EAI_SYSTEM == gai_resp->retcode)
		gai_resp->gai_errno = errno;
#endif
	canons_octets = 0;

	if (0 == gai_resp->retcode) {
		ai = ai_res;
		while (NULL != ai) {
			gai_resp->ai_count++;
			if (ai->ai_canonname)
				canons_octets += strlen(ai->ai_canonname) + 1;
			ai = ai->ai_next;
		}
		/*
		 * If this query succeeded only after retrying, DNS may have
		 * just become responsive.  Ignore previously-scheduled
		 * retry sleeps once for each pending request, similar to
		 * the way scheduled_sleep() does when its worker_sleep()
		 * is interrupted.
		 */
		if (gai_resp->retry > INITIAL_DNS_RETRY) {
			time_now = time(NULL);
			worker_ctx->ignore_scheduled_before = time_now;
			TRACE(1, ("DNS success after retry, ignoring sleeps scheduled before now (%s)\n",
				  humantime(time_now)));
		}
	}

	/*
	 * Our response consists of a header, followed by ai_count 
	 * addrinfo structs followed by ai_count sockaddr_storage 
	 * structs followed by the canonical names.
	 */
	gai_resp->octets = sizeof(*gai_resp)
			    + gai_resp->ai_count
				* (sizeof(gai_req->hints)
				   + sizeof(sockaddr_u))
			    + canons_octets;

	resp_octets = sizeof(*resp) + gai_resp->octets;
	resp = erealloc(resp, resp_octets);
	gai_resp = (void *)(resp + 1);

	/* cp serves as our current pointer while serializing */
	cp = (void *)(gai_resp + 1);
	canons_octets = 0;

	if (0 == gai_resp->retcode) {
		ai = ai_res;
		while (NULL != ai) {
			memcpy(cp, ai, sizeof(*ai));
			serialized_ai = (void *)cp;
			cp += sizeof(*ai);

			/* transform ai_canonname into offset */
			if (NULL != ai->ai_canonname) {
				serialized_ai->ai_canonname = (char *)canons_octets;
				canons_octets += strlen(ai->ai_canonname) + 1;
			}
			
			/* leave fixup of ai_addr pointer for receiver */

			ai = ai->ai_next;
		}

		ai = ai_res;
		while (NULL != ai) {
			INSIST(ai->ai_addrlen <= sizeof(sockaddr_u));
			memcpy(cp, ai->ai_addr, ai->ai_addrlen);
			cp += sizeof(sockaddr_u);

			ai = ai->ai_next;
		}

		ai = ai_res;
		while (NULL != ai) {
			if (NULL != ai->ai_canonname) {
				this_octets = strlen(ai->ai_canonname) + 1;
				memcpy(cp, ai->ai_canonname, this_octets);
				cp += this_octets;
			}

			ai = ai->ai_next;
		}
		freeaddrinfo(ai_res);
	}

	/*
	 * make sure our walk and earlier calc match
	 */
	DEBUG_INSIST((size_t)(cp - (char *)resp) == resp_octets);

	if (queue_blocking_response(c, resp, resp_octets, req)) {
		msyslog(LOG_ERR, "blocking_getaddrinfo can not queue response");
		return -1;
	}

	return 0;
}

int
getaddrinfo_sometime(
	const char *		node,
	const char *		service,
	const struct addrinfo *	hints,
	int			retry,
	gai_sometime_callback	callback,
	void *			context
	)
{
	return getaddrinfo_sometime_ex(node, service, hints, retry,
				       callback, context, 0);
}


static void
getaddrinfo_sometime_complete(
	blocking_work_req	rtype,
	void *			context,
	size_t			respsize,
	void *			resp
	)
{
	blocking_gai_req *	gai_req;
	blocking_gai_resp *	gai_resp;
	dnschild_ctx *		child_ctx;
	struct addrinfo *	ai;
	struct addrinfo *	next_ai;
	sockaddr_u *		psau;
	char *			node;
	char *			service;
	char *			canon_start;
	time_t			time_now;
	int			again, noerr;
	int			af;
	const char *		fam_spec;
	int			i;

	gai_req = context;
	gai_resp = resp;

	DEBUG_REQUIRE(BLOCKING_GETADDRINFO == rtype);
	DEBUG_REQUIRE(respsize == gai_resp->octets);

	node = (char *)gai_req + sizeof(*gai_req);
	service = node + gai_req->nodesize;

	child_ctx = dnschild_contexts[gai_req->dns_idx];

	if (0 == gai_resp->retcode) {
		/*
		 * If this query succeeded only after retrying, DNS may have
		 * just become responsive.
		 */
		if (gai_resp->retry > INITIAL_DNS_RETRY) {
			time_now = time(NULL);
			child_ctx->next_dns_timeslot = time_now;
			TRACE(1, ("DNS success after retry, %u next_dns_timeslot reset (%s)\n",
				  gai_req->dns_idx, humantime(time_now)));
		}
	} else {
		noerr = !!(gai_req->qflags & GAIR_F_IGNDNSERR);
		again = noerr || should_retry_dns(
					gai_resp->retcode, gai_resp->gai_errno);
		/*
		 * exponential backoff of DNS retries to 64s
		 */
		if (gai_req->retry > 0 && again) {
			/* log the first retry only */
			if (INITIAL_DNS_RETRY == gai_req->retry)
				NLOG(NLOG_SYSINFO) {
					af = gai_req->hints.ai_family;
					fam_spec = (AF_INET6 == af)
						       ? " (AAAA)"
						       : (AF_INET == af)
							     ? " (A)"
							     : "";
#ifdef EAI_SYSTEM
					if (EAI_SYSTEM == gai_resp->retcode) {
						errno = gai_resp->gai_errno;
						msyslog(LOG_INFO,
							"retrying DNS %s%s: EAI_SYSTEM %d: %m",
							node, fam_spec,
							gai_resp->gai_errno);
					} else
#endif
						msyslog(LOG_INFO,
							"retrying DNS %s%s: %s (%d)",
							node, fam_spec,
							gai_strerror(gai_resp->retcode),
							gai_resp->retcode);
				}
			manage_dns_retry_interval(
				&gai_req->scheduled, &gai_req->earliest,
				&gai_req->retry, &child_ctx->next_dns_timeslot,
				noerr);
			if (!queue_blocking_request(
					BLOCKING_GETADDRINFO,
					gai_req,
					gai_req->octets,
					&getaddrinfo_sometime_complete,
					gai_req))
				return;
			else
				msyslog(LOG_ERR,
					"unable to retry hostname %s",
					node);
		}
	}

	/*
	 * fixup pointers in returned addrinfo array
	 */
	ai = (void *)((char *)gai_resp + sizeof(*gai_resp));
	next_ai = NULL;
	for (i = gai_resp->ai_count - 1; i >= 0; i--) {
		ai[i].ai_next = next_ai;
		next_ai = &ai[i];
	}

	psau = (void *)((char *)ai + gai_resp->ai_count * sizeof(*ai));
	canon_start = (char *)psau + gai_resp->ai_count * sizeof(*psau);

	for (i = 0; i < gai_resp->ai_count; i++) {
		if (NULL != ai[i].ai_addr)
			ai[i].ai_addr = &psau->sa;
		psau++;
		if (NULL != ai[i].ai_canonname)
			ai[i].ai_canonname += (size_t)canon_start;
	}

	ENSURE((char *)psau == canon_start);

	if (!gai_resp->ai_count)
		ai = NULL;
	
	(*gai_req->callback)(gai_resp->retcode, gai_resp->gai_errno,
			     gai_req->context, node, service, 
			     &gai_req->hints, ai);

	free(gai_req);
	/* gai_resp is part of block freed by process_blocking_resp() */
}


#ifdef TEST_BLOCKING_WORKER
void gai_test_callback(int rescode, int gai_errno, void *context, const char *name, const char *service, const struct addrinfo *hints, const struct addrinfo *ai_res)
{
	sockaddr_u addr;

	if (rescode) {
		TRACE(1, ("gai_test_callback context %p error rescode %d %s serv %s\n",
			  context, rescode, name, service));
		return;
	}
	while (!rescode && NULL != ai_res) {
		ZERO_SOCK(&addr);
		memcpy(&addr, ai_res->ai_addr, ai_res->ai_addrlen);
		TRACE(1, ("ctx %p fam %d addr %s canon '%s' type %s at %p ai_addr %p ai_next %p\n", 
			  context,
			  AF(&addr),
			  stoa(&addr), 
			  (ai_res->ai_canonname)
			      ? ai_res->ai_canonname
			      : "",
			  (SOCK_DGRAM == ai_res->ai_socktype) 
			      ? "DGRAM" 
			      : (SOCK_STREAM == ai_res->ai_socktype) 
				    ? "STREAM" 
				    : "(other)",
			  ai_res,
			  ai_res->ai_addr,
			  ai_res->ai_next));

		getnameinfo_sometime((sockaddr_u *)ai_res->ai_addr, 128, 32, 0, gni_test_callback, context);

		ai_res = ai_res->ai_next;
	}
}
#endif	/* TEST_BLOCKING_WORKER */


int
getnameinfo_sometime(
	sockaddr_u *		psau,
	size_t			hostoctets,
	size_t			servoctets,
	int			flags,
	gni_sometime_callback	callback,
	void *			context
	)
{
	blocking_gni_req *	gni_req;
	u_int			idx;
	dnschild_ctx *		child_ctx;
	time_t			time_now;
	
	REQUIRE(hostoctets);
	REQUIRE(hostoctets + servoctets < 1024);

	idx = get_dnschild_ctx();
	child_ctx = dnschild_contexts[idx];

	gni_req = emalloc_zero(sizeof(*gni_req));

	gni_req->octets = sizeof(*gni_req);
	gni_req->dns_idx = idx;
	time_now = time(NULL);
	gni_req->scheduled = time_now;
	gni_req->earliest = max(time_now, child_ctx->next_dns_timeslot);
	child_ctx->next_dns_timeslot = gni_req->earliest;
	memcpy(&gni_req->socku, psau, SOCKLEN(psau));
	gni_req->hostoctets = hostoctets;
	gni_req->servoctets = servoctets;
	gni_req->flags = flags;
	gni_req->retry = INITIAL_DNS_RETRY;
	gni_req->callback = callback;
	gni_req->context = context;

	if (queue_blocking_request(
		BLOCKING_GETNAMEINFO,
		gni_req,
		sizeof(*gni_req), 
		&getnameinfo_sometime_complete, 
		gni_req)) {

		msyslog(LOG_ERR, "unable to queue getnameinfo request");
		errno = EFAULT;
		return -1;
	}

	return 0;
}


int
blocking_getnameinfo(
	blocking_child *	c,
	blocking_pipe_header *	req
	)
{
	blocking_gni_req *	gni_req;
	dnsworker_ctx *		worker_ctx;
	blocking_pipe_header *	resp;
	blocking_gni_resp *	gni_resp;
	size_t			octets;
	size_t			resp_octets;
	char *			service;
	char *			cp;
	int			rc;
	time_t			time_now;
	char			host[1024];

	gni_req = (void *)((char *)req + sizeof(*req));

	octets = gni_req->hostoctets + gni_req->servoctets;

	/*
	 * Some alloca() implementations are fragile regarding
	 * large allocations.  We only need room for the host
	 * and service names.
	 */
	REQUIRE(octets < sizeof(host));
	service = host + gni_req->hostoctets;

	worker_ctx = get_worker_context(c, gni_req->dns_idx);
	scheduled_sleep(gni_req->scheduled, gni_req->earliest,
			worker_ctx);
	reload_resolv_conf(worker_ctx);

	/*
	 * Take a shot at the final size, better to overestimate
	 * then realloc to a smaller size.
	 */

	resp_octets = sizeof(*resp) + sizeof(*gni_resp) + octets;
	resp = emalloc_zero(resp_octets);
	gni_resp = (void *)((char *)resp + sizeof(*resp));

	TRACE(2, ("blocking_getnameinfo given addr %s flags 0x%x hostlen %lu servlen %lu\n",
		  stoa(&gni_req->socku), gni_req->flags,
		  (u_long)gni_req->hostoctets, (u_long)gni_req->servoctets));
	
	gni_resp->retcode = getnameinfo(&gni_req->socku.sa,
					SOCKLEN(&gni_req->socku),
					host,
					gni_req->hostoctets,
					service,
					gni_req->servoctets,
					gni_req->flags);
	gni_resp->retry = gni_req->retry;
#ifdef EAI_SYSTEM
	if (EAI_SYSTEM == gni_resp->retcode)
		gni_resp->gni_errno = errno;
#endif

	if (0 != gni_resp->retcode) {
		gni_resp->hostoctets = 0;
		gni_resp->servoctets = 0;
	} else {
		gni_resp->hostoctets = strlen(host) + 1;
		gni_resp->servoctets = strlen(service) + 1;
		/*
		 * If this query succeeded only after retrying, DNS may have
		 * just become responsive.  Ignore previously-scheduled
		 * retry sleeps once for each pending request, similar to
		 * the way scheduled_sleep() does when its worker_sleep()
		 * is interrupted.
		 */
		if (gni_req->retry > INITIAL_DNS_RETRY) {
			time_now = time(NULL);
			worker_ctx->ignore_scheduled_before = time_now;
			TRACE(1, ("DNS success after retrying, ignoring sleeps scheduled before now (%s)\n",
				humantime(time_now)));
		}
	}
	octets = gni_resp->hostoctets + gni_resp->servoctets;
	/*
	 * Our response consists of a header, followed by the host and
	 * service strings, each null-terminated.
	 */
	resp_octets = sizeof(*resp) + sizeof(*gni_resp) + octets;

	resp = erealloc(resp, resp_octets);
	gni_resp = (void *)(resp + 1);

	gni_resp->octets = sizeof(*gni_resp) + octets;

	/* cp serves as our current pointer while serializing */
	cp = (void *)(gni_resp + 1);

	if (0 == gni_resp->retcode) {
		memcpy(cp, host, gni_resp->hostoctets);
		cp += gni_resp->hostoctets;
		memcpy(cp, service, gni_resp->servoctets);
		cp += gni_resp->servoctets;
	}

	INSIST((size_t)(cp - (char *)resp) == resp_octets);
	INSIST(resp_octets - sizeof(*resp) == gni_resp->octets);

	rc = queue_blocking_response(c, resp, resp_octets, req);
	if (rc)
		msyslog(LOG_ERR, "blocking_getnameinfo unable to queue response");
	return rc;
}


static void
getnameinfo_sometime_complete(
	blocking_work_req	rtype,
	void *			context,
	size_t			respsize,
	void *			resp
	)
{
	blocking_gni_req *	gni_req;
	blocking_gni_resp *	gni_resp;
	dnschild_ctx *		child_ctx;
	char *			host;
	char *			service;
	time_t			time_now;
	int			again;

	gni_req = context;
	gni_resp = resp;

	DEBUG_REQUIRE(BLOCKING_GETNAMEINFO == rtype);
	DEBUG_REQUIRE(respsize == gni_resp->octets);

	child_ctx = dnschild_contexts[gni_req->dns_idx];

	if (0 == gni_resp->retcode) {
		/*
		 * If this query succeeded only after retrying, DNS may have
		 * just become responsive.
		 */
		if (gni_resp->retry > INITIAL_DNS_RETRY) {
			time_now = time(NULL);
			child_ctx->next_dns_timeslot = time_now;
			TRACE(1, ("DNS success after retry, %u next_dns_timeslot reset (%s)\n",
				  gni_req->dns_idx, humantime(time_now)));
		}
	} else {
		again = should_retry_dns(gni_resp->retcode, gni_resp->gni_errno);
		/*
		 * exponential backoff of DNS retries to 64s
		 */
		if (gni_req->retry > 0)
			manage_dns_retry_interval(&gni_req->scheduled,
			    &gni_req->earliest, &gni_req->retry,
						  &child_ctx->next_dns_timeslot, FALSE);

		if (gni_req->retry > 0 && again) {
			if (!queue_blocking_request(
				BLOCKING_GETNAMEINFO,
				gni_req,
				gni_req->octets, 
				&getnameinfo_sometime_complete, 
				gni_req))
				return;

			msyslog(LOG_ERR, "unable to retry reverse lookup of %s", stoa(&gni_req->socku));
		}
	}

	if (!gni_resp->hostoctets) {
		host = NULL;
		service = NULL;
	} else {
		host = (char *)gni_resp + sizeof(*gni_resp);
		service = (gni_resp->servoctets) 
			      ? host + gni_resp->hostoctets
			      : NULL;
	}

	(*gni_req->callback)(gni_resp->retcode, gni_resp->gni_errno,
			     &gni_req->socku, gni_req->flags, host,
			     service, gni_req->context);

	free(gni_req);
	/* gni_resp is part of block freed by process_blocking_resp() */
}


#ifdef TEST_BLOCKING_WORKER
void gni_test_callback(int rescode, int gni_errno, sockaddr_u *psau, int flags, const char *host, const char *service, void *context)
{
	if (!rescode)
		TRACE(1, ("gni_test_callback got host '%s' serv '%s' for addr %s context %p\n", 
			  host, service, stoa(psau), context));
	else
		TRACE(1, ("gni_test_callback context %p rescode %d gni_errno %d flags 0x%x addr %s\n",
			  context, rescode, gni_errno, flags, stoa(psau)));
}
#endif	/* TEST_BLOCKING_WORKER */


#ifdef HAVE_RES_INIT
static void
reload_resolv_conf(
	dnsworker_ctx *	worker_ctx
	)
{
	time_t	time_now;

	/*
	 * This is ad-hoc.  Reload /etc/resolv.conf once per minute
	 * to pick up on changes from the DHCP client.  [Bug 1226]
	 * When using threads for the workers, this needs to happen
	 * only once per minute process-wide.
	 */
	time_now = time(NULL);
# ifdef WORK_THREAD
	worker_ctx->next_res_init = next_res_init;
# endif
	if (worker_ctx->next_res_init <= time_now) {
		if (worker_ctx->next_res_init != 0)
			res_init();
		worker_ctx->next_res_init = time_now + 60;
# ifdef WORK_THREAD
		next_res_init = worker_ctx->next_res_init;
# endif
	}
}
#endif	/* HAVE_RES_INIT */


static u_int
reserve_dnschild_ctx(void)
{
	const size_t	ps = sizeof(dnschild_contexts[0]);
	const size_t	cs = sizeof(*dnschild_contexts[0]);
	u_int		c;
	u_int		new_alloc;
	size_t		octets;
	size_t		new_octets;

	c = 0;
	while (TRUE) {
		for ( ; c < dnschild_contexts_alloc; c++) {
			if (NULL == dnschild_contexts[c]) {
				dnschild_contexts[c] = emalloc_zero(cs);

				return c;
			}
		}
		new_alloc = dnschild_contexts_alloc + 20;
		new_octets = new_alloc * ps;
		octets = dnschild_contexts_alloc * ps;
		dnschild_contexts = erealloc_zero(dnschild_contexts,
						  new_octets, octets);
		dnschild_contexts_alloc = new_alloc;
	}
}


static u_int
get_dnschild_ctx(void)
{
	static u_int	shared_ctx = UINT_MAX;

	if (worker_per_query)
		return reserve_dnschild_ctx();

	if (UINT_MAX == shared_ctx)
		shared_ctx = reserve_dnschild_ctx();

	return shared_ctx;
}


static dnsworker_ctx *
get_worker_context(
	blocking_child *	c,
	u_int			idx
	)
{
	u_int		min_new_alloc;
	u_int		new_alloc;
	size_t		octets;
	size_t		new_octets;
	dnsworker_ctx *	retv;

	worker_global_lock(TRUE);
	
	if (dnsworker_contexts_alloc <= idx) {
		min_new_alloc = 1 + idx;
		/* round new_alloc up to nearest multiple of 4 */
		new_alloc = (min_new_alloc + 4) & ~(4 - 1);
		new_octets = new_alloc * sizeof(dnsworker_ctx*);
		octets = dnsworker_contexts_alloc * sizeof(dnsworker_ctx*);
		dnsworker_contexts = erealloc_zero(dnsworker_contexts,
						   new_octets, octets);
		dnsworker_contexts_alloc = new_alloc;
		retv = emalloc_zero(sizeof(dnsworker_ctx));
		dnsworker_contexts[idx] = retv;
	} else if (NULL == (retv = dnsworker_contexts[idx])) {
		retv = emalloc_zero(sizeof(dnsworker_ctx));
		dnsworker_contexts[idx] = retv;
	}
	
	worker_global_lock(FALSE);
	
	ZERO(*retv);
	retv->c = c;
	return retv;
}


static void
scheduled_sleep(
	time_t		scheduled,
	time_t		earliest,
	dnsworker_ctx *	worker_ctx
	)
{
	time_t now;

	if (scheduled < worker_ctx->ignore_scheduled_before) {
		TRACE(1, ("ignoring sleep until %s scheduled at %s (before %s)\n",
			  humantime(earliest), humantime(scheduled),
			  humantime(worker_ctx->ignore_scheduled_before)));
		return;
	}

	now = time(NULL);

	if (now < earliest) {
		TRACE(1, ("sleep until %s scheduled at %s (>= %s)\n",
			  humantime(earliest), humantime(scheduled),
			  humantime(worker_ctx->ignore_scheduled_before)));
		if (-1 == worker_sleep(worker_ctx->c, earliest - now)) {
			/* our sleep was interrupted */
			now = time(NULL);
			worker_ctx->ignore_scheduled_before = now;
#ifdef HAVE_RES_INIT
			worker_ctx->next_res_init = now + 60;
			next_res_init = worker_ctx->next_res_init;
			res_init();
#endif
			TRACE(1, ("sleep interrupted by daemon, ignoring sleeps scheduled before now (%s)\n",
				  humantime(worker_ctx->ignore_scheduled_before)));
		}
	}
}


/*
 * manage_dns_retry_interval is a helper used by
 * getaddrinfo_sometime_complete and getnameinfo_sometime_complete
 * to calculate the new retry interval and schedule the next query.
 */
static void
manage_dns_retry_interval(
	time_t *	pscheduled,
	time_t *	pwhen,
	int *		pretry,
	time_t *	pnext_timeslot,
	int		forever
	)
{
	time_t	now;
	time_t	when;
	int	retry;
	int	retmax;
		
	now = time(NULL);
	retry = *pretry;
	when = max(now + retry, *pnext_timeslot);
	*pnext_timeslot = when;

	/* this exponential backoff is slower than doubling up: The
	 * sequence goes 2-3-4-6-8-12-16-24-32... and the upper limit is
	 * 64 seconds for things that should not repeat forever, and
	 * 1024 when repeated forever.
	 */
	retmax = forever ? 1024 : 64;
	retry <<= 1;
	if (retry & (retry - 1))
		retry &= (retry - 1);
	else
		retry -= (retry >> 2);
	retry = min(retmax, retry);

	*pscheduled = now;
	*pwhen = when;
	*pretry = retry;
}

/*
 * should_retry_dns is a helper used by getaddrinfo_sometime_complete
 * and getnameinfo_sometime_complete which implements ntpd's DNS retry
 * policy.
 */
static int
should_retry_dns(
	int	rescode,
	int	res_errno
	)
{
	static int	eai_again_seen;
	int		again;
#if defined (EAI_SYSTEM) && defined(DEBUG)
	char		msg[256];
#endif

	/*
	 * If the resolver failed, see if the failure is
	 * temporary. If so, return success.
	 */
	again = 0;

	switch (rescode) {

	case EAI_FAIL:
		again = 1;
		break;

	case EAI_AGAIN:
		again = 1;
		eai_again_seen = 1;		/* [Bug 1178] */
		break;

	case EAI_NONAME:
#if defined(EAI_NODATA) && (EAI_NODATA != EAI_NONAME)
	case EAI_NODATA:
#endif
		again = !eai_again_seen;	/* [Bug 1178] */
		break;

#ifdef EAI_SYSTEM
	case EAI_SYSTEM:
		/* 
		 * EAI_SYSTEM means the real error is in errno.  We should be more
		 * discriminating about which errno values require retrying, but
		 * this matches existing behavior.
		 */
		again = 1;
# ifdef DEBUG
		errno_to_str(res_errno, msg, sizeof(msg));
		TRACE(1, ("intres: EAI_SYSTEM errno %d (%s) means try again, right?\n",
			  res_errno, msg));
# endif
		break;
#endif
	}

	TRACE(2, ("intres: resolver returned: %s (%d), %sretrying\n",
		  gai_strerror(rescode), rescode, again ? "" : "not "));

	return again;
}

#else	/* !WORKER follows */
int ntp_intres_nonempty_compilation_unit;
#endif