summaryrefslogtreecommitdiff
path: root/usr.bin/make/job.c
blob: ed2ea5ecf8246b8439da7d4e09ff7d5010c3103f (plain)
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
/*	$OpenBSD: job.c,v 1.128 2012/10/06 09:32:40 espie Exp $	*/
/*	$NetBSD: job.c,v 1.16 1996/11/06 17:59:08 christos Exp $	*/

/*
 * Copyright (c) 2012 Marc Espie.
 *
 * Extensive code modifications for the OpenBSD project.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice, this list of conditions and the following disclaimer.
 * 2. 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.
 *
 * THIS SOFTWARE IS PROVIDED BY THE OPENBSD PROJECT 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 OPENBSD
 * PROJECT 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.
 */
/*
 * Copyright (c) 1988, 1989, 1990 The Regents of the University of California.
 * Copyright (c) 1988, 1989 by Adam de Boor
 * Copyright (c) 1989 by Berkeley Softworks
 * All rights reserved.
 *
 * This code is derived from software contributed to Berkeley by
 * Adam de Boor.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice, this list of conditions and the following disclaimer.
 * 2. 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.
 * 3. Neither the name of the University 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 REGENTS 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 REGENTS 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.
 */

/*-
 * job.c --
 *	handle the creation etc. of our child processes.
 *
 * Interface:
 *	Job_Make		Start the creation of the given target.
 *
 *	Job_Init		Called to initialize this module. 
 *
 *	Job_Begin		execute commands attached to the .BEGIN target
 *				if any.
 *
 *	can_start_job		Return true if we can start job
 *
 *	Job_Empty		Return true if the job table is completely
 *				empty.
 *
 *	Job_Finish		Perform any final processing which needs doing.
 *				This includes the execution of any commands
 *				which have been/were attached to the .END
 *				target. 
 *
 *	Job_AbortAll		Abort all current jobs. It doesn't
 *				handle output or do anything for the jobs,
 *				just kills them. 
 *
 *	Job_Wait		Wait for all running jobs to finish.
 */

#include <sys/types.h>
#include <sys/wait.h>
#include <ctype.h>
#include <errno.h>
#include <fcntl.h>
#include <signal.h>
#include <stdarg.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>
#include "config.h"
#include "defines.h"
#include "job.h"
#include "engine.h"
#include "pathnames.h"
#include "var.h"
#include "targ.h"
#include "error.h"
#include "extern.h"
#include "lst.h"
#include "gnode.h"
#include "memory.h"
#include "make.h"
#include "buf.h"

static int	aborting = 0;	    /* why is the make aborting? */
#define ABORT_ERROR	1	    /* Because of an error */
#define ABORT_INTERRUPT 2	    /* Because it was interrupted */
#define ABORT_WAIT	3	    /* Waiting for jobs to finish */

static int	maxJobs;	/* The most children we can run at once */
static int	nJobs;		/* Number of jobs already allocated */
static bool	no_new_jobs;	/* Mark recursive shit so we shouldn't start
				 * something else at the same time
				 */
Job *runningJobs;		/* Jobs currently running a process */
Job *errorJobs;			/* Jobs in error at end */
static Job *heldJobs;		/* Jobs not running yet because of expensive */
static pid_t mypid;

static volatile sig_atomic_t got_fatal;

static volatile sig_atomic_t got_SIGINT, got_SIGHUP, got_SIGQUIT, got_SIGTERM, 
    got_SIGINFO;

static sigset_t sigset, emptyset;

static void handle_fatal_signal(int);
static void handle_siginfo(void);
static void postprocess_job(Job *, bool);
static Job *prepare_job(GNode *);
static void determine_job_next_step(Job *);
static void remove_job(Job *, bool);
static void may_continue_job(Job *);
static void continue_job(Job *);
static Job *reap_finished_job(pid_t);
static bool reap_jobs(void);

static void loop_handle_running_jobs(void);
static bool expensive_job(Job *);
static bool expensive_command(const char *);
static void setup_signal(int);
static void notice_signal(int);
static void setup_all_signals(void);
static const char *really_kill(Job *, int);
static void kill_with_sudo_maybe(pid_t, int, const char *);
static void debug_kill_printf(const char *, ...);
static void debug_vprintf(const char *, va_list);
static void may_remove_target(Job *);

static void 
kill_with_sudo_maybe(pid_t pid, int signo, const char *p)
{
	char buf[32]; /* largely enough */

	for (;*p != '\0'; p++) {
		if (*p != 's')
			continue;
		if (p[1] != 'u')
			continue;
		p++;
		if (p[1] != 'd')
			continue;
		p++;
		if (p[1] != 'o')
			continue;
		snprintf(buf, sizeof buf, "sudo -n /bin/kill -%d %ld", 
		    signo, (long)pid);
		debug_kill_printf("trying to kill with %s", buf);
		system(buf);
		return;
	}

}

static const char *
really_kill(Job *job, int signo)
{
	pid_t pid = job->pid;
	job->sent_signal = signo;
	if (getpgid(pid) != getpgrp()) {
		if (killpg(pid, signo) == 0)
			return "group got signal";
		pid = -pid;
	} else {
		if (kill(pid, signo) == 0)
			return "process got signal";
	}
	if (errno == ESRCH) {
		job->flags |= JOB_LOST;
		return "not found";
	} else if (errno == EPERM) {
		kill_with_sudo_maybe(pid, signo, job->cmd);
		return "";
	}
}

static void
may_remove_target(Job *j)
{
	if ((j->sent_signal == SIGINT || j->sent_signal == SIGQUIT ||
	    j->sent_signal == SIGHUP || j->sent_signal == SIGTERM) 
	    && !noExecute && !Targ_Precious(j->node)) {
		const char *file = Var(TARGET_INDEX, j->node);
		int r = eunlink(file);

		if (DEBUG(JOB) && r == -1)
			fprintf(stderr, " *** would unlink %s\n", file);
		if (r != -1)
			fprintf(stderr, " *** %s removed\n", file);
	}
}

static void 
print_error(Job *j, Job *k)
{
	const char *type;

	if (j->exit_type == JOB_EXIT_BAD) {
		fprintf(stderr, "*** Error %d", j->code);
		if (j->sent_signal != 0 && j->code == j->sent_signal + 128)
			fprintf(stderr, " (SIG%s in shell)", 
			    sys_signame[j->sent_signal]);
	} else if (j->exit_type == JOB_SIGNALED) {
		if (j->code < NSIG)
			fprintf(stderr, "*** SIG%s", 
			    sys_signame[j->code]);
		else
			fprintf(stderr, "*** unknown signal %d", j->code);
	} else
		fprintf(stderr, "*** Should not happen %d", j->code);
	if (DEBUG(KILL) && (j->flags & JOB_LOST))
		fprintf(stderr, "!");
	fprintf(stderr, " (line %lu", j->location->lineno);
	if (j == k)
		fprintf(stderr, " of %s,", j->location->fname);
	else
		fputs(",", stderr);
	if ((j->flags & (JOB_SILENT | JOB_IS_EXPENSIVE)) == JOB_SILENT)
		fprintf(stderr, "\n     target '%s': %s", j->node->name, 
		    j->cmd);
	else
		fprintf(stderr, " target '%s'", j->node->name);
	fputs(")\n", stderr);
	free(j->cmd);
	may_remove_target(j);
}

void
print_errors(void)
{
	Job *j, *k, *jnext;

	fprintf(stderr, "\nStop in %s%c\n", Var_Value(".CURDIR"),
	    errorJobs ? ':' : '.');

	while (errorJobs != NULL) {
		k = errorJobs;
		errorJobs = NULL;
		for (j = k; j != NULL; j = jnext) {
			jnext = j->next;
			if (j->location->fname == k->location->fname)
				print_error(j, k);
			else {
				j->next = errorJobs;
				errorJobs = j;
			}
		}
	}
}

static void
setup_signal(int sig)
{
	if (signal(sig, SIG_IGN) != SIG_IGN) {
		(void)signal(sig, notice_signal);
		sigaddset(&sigset, sig);
	}
}

static void
notice_signal(int sig)
{

	switch(sig) {
	case SIGINT:
		got_SIGINT++;
		got_fatal = 1;
		break;
	case SIGHUP:
		got_SIGHUP++;
		got_fatal = 1;
		break;
	case SIGQUIT:
		got_SIGQUIT++;
		got_fatal = 1;
		break;
	case SIGTERM:
		got_SIGTERM++;
		got_fatal = 1;
		break;
	case SIGINFO:
		got_SIGINFO++;
		break;
	case SIGCHLD:
		break;
	}
}

void
setup_all_signals(void)
{
	sigemptyset(&sigset);
	sigemptyset(&emptyset);
	/*
	 * Catch the four signals that POSIX specifies if they aren't ignored.
	 * handle_signal will take care of calling JobInterrupt if appropriate.
	 */
	setup_signal(SIGINT);
	setup_signal(SIGHUP);
	setup_signal(SIGQUIT);
	setup_signal(SIGTERM);
	/* Display running jobs on SIGINFO */
	setup_signal(SIGINFO);
	/* Have to see SIGCHLD */
	setup_signal(SIGCHLD);
	got_fatal = 0;
}

static void 
handle_siginfo(void)
{
	static BUFFER buf;
	static size_t length = 0;

	Job *job;
	bool first = true;

	got_SIGINFO = 0;
	/* we have to store the info in a buffer, because status from all
	 * makes running would get intermixed otherwise
	 */

	if (length == 0) {
		Buf_Init(&buf, 0);
		Buf_printf(&buf, "%s in %s: ", Var_Value("MAKE"), Var_Value(".CURDIR"));
		length = Buf_Size(&buf);
	} else
		Buf_Truncate(&buf, length);

	for (job = runningJobs; job != NULL ; job = job->next) {
		if (!first)
			Buf_puts(&buf, ", ");
		first = false;
		Buf_puts(&buf, job->node->name);
	}
	Buf_puts(&buf, first ? "nothing running\n" : "\n");

	fputs(Buf_Retrieve(&buf), stderr);
}

void
handle_all_signals(void)
{
	if (got_SIGINFO)
		handle_siginfo();
	while (got_fatal) {
		got_fatal = 0;
		aborting = ABORT_INTERRUPT;

		if (got_SIGINT) {
			got_SIGINT=0;
			handle_fatal_signal(SIGINT);
		}
		if (got_SIGHUP) {
			got_SIGHUP=0;
			handle_fatal_signal(SIGHUP);
		}
		if (got_SIGQUIT) {
			got_SIGQUIT=0;
			handle_fatal_signal(SIGQUIT);
		}
		if (got_SIGTERM) {
			got_SIGTERM=0;
			handle_fatal_signal(SIGTERM);
		}
	}
}

static void
debug_vprintf(const char *fmt, va_list va)
{
	(void)printf("[%ld] ", (long)mypid);
	(void)vprintf(fmt, va);
	fflush(stdout);
}

void
debug_job_printf(const char *fmt, ...)
{
	if (DEBUG(JOB)) {
		va_list va;
		va_start(va, fmt);
		debug_vprintf(fmt, va);
		va_end(va);
	}
}

static void
debug_kill_printf(const char *fmt, ...)
{
	if (DEBUG(KILL)) {
		va_list va;
		va_start(va, fmt);
		debug_vprintf(fmt, va);
		va_end(va);
	}
}

/*-
 *-----------------------------------------------------------------------
 * postprocess_job  --
 *	Do final processing for the given job including updating
 *	parents and starting new jobs as available/necessary.
 *
 * Side Effects:
 *	If we got an error and are aborting (aborting == ABORT_ERROR) and
 *	the job list is now empty, we are done for the day.
 *	If we recognized an error we set the aborting flag
 *	to ABORT_ERROR so no more jobs will be started.
 *-----------------------------------------------------------------------
 */
/*ARGSUSED*/

static void
postprocess_job(Job *job, bool okay)
{
	if (okay &&
	    aborting != ABORT_ERROR &&
	    aborting != ABORT_INTERRUPT) {
		/* As long as we aren't aborting and the job didn't return a
		 * non-zero status that we shouldn't ignore, we call
		 * Make_Update to update the parents. */
		job->node->built_status = MADE;
		Make_Update(job->node);
		free(job);
	}

	if (errorJobs != NULL && !keepgoing &&
	    aborting != ABORT_INTERRUPT)
		aborting = ABORT_ERROR;

	if (aborting == ABORT_ERROR && DEBUG(QUICKDEATH))
		handle_fatal_signal(SIGINT);
	if (aborting == ABORT_ERROR && Job_Empty())
		Finish();
}

/* expensive jobs handling: in order to avoid forking an exponential number
 * of jobs, make tries to figure out "recursive make" configurations.
 * It may err on the side of caution.
 * Basically, a command is "expensive" if it's likely to fork an extra
 * level of make: either by looking at the command proper, or if it has
 * some specific qualities ('+cmd' are likely to be recursive, as are
 * .MAKE: commands).  It's possible to explicitly say some targets are
 * expensive or cheap with .EXPENSIVE or .CHEAP.
 *
 * While an expensive command is running, no_new_jobs
 * is set, so jobs that would fork new processes are accumulated in the
 * heldJobs list instead.
 *
 * This heuristics is also used on error exit: we display silent commands
 * that failed, unless those ARE expensive commands: expensive commands
 * are likely to not be failing by themselves, but to be the result of
 * a cascade of failures in descendant makes.
 */
void
determine_expensive_job(Job *job)
{ 
	if (expensive_job(job)) {
		job->flags |= JOB_IS_EXPENSIVE;
		no_new_jobs = true;
	} else
		job->flags &= ~JOB_IS_EXPENSIVE;
	if (DEBUG(EXPENSIVE))
		fprintf(stderr, "[%ld] Target %s running %.50s: %s\n",
		    (long)mypid, job->node->name, job->cmd,
		    job->flags & JOB_IS_EXPENSIVE ? "expensive" : "cheap");
}

static bool
expensive_job(Job *job)
{
	if (job->node->type & OP_CHEAP)
		return false;
	if (job->node->type & (OP_EXPENSIVE | OP_MAKE))
		return true;
	return expensive_command(job->cmd);
}

static bool
expensive_command(const char *s)
{
	const char *p;
	bool include = false;
	bool expensive = false;

	/* okay, comments are cheap, always */
	if (*s == '#')
		return false;
	/* and commands we always execute are expensive */
	if (*s == '+')
		return true;

	for (p = s; *p != '\0'; p++) {
		if (*p == ' ' || *p == '\t') {
			include = false;
			if (p[1] == '-' && p[2] == 'I')
				include = true;
		}
		if (include)
			continue;
		/* KMP variant, avoid looking twice at the same
		 * letter.
		 */
		if (*p != 'm')
			continue;
		if (p[1] != 'a')
			continue;
		p++;
		if (p[1] != 'k')
			continue;
		p++;
		if (p[1] != 'e')
			continue;
		p++;
		expensive = true;
		while (p[1] != '\0' && p[1] != ' ' && p[1] != '\t') {
			if (p[1] == '.' || p[1] == '/') {
				expensive = false;
				break;
			}
		    	p++;
		}
		if (expensive)
			return true;
	}
	return false;
}

static Job *
prepare_job(GNode *gn)
{
	/* a new job is prepared unless its commands are bogus (we don't
	 * have anything for it), or if we're in touch mode.
	 *
	 * Note that even in noexec mode, some commands may still run
	 * thanks to the +cmd construct.
	 */
	if (node_find_valid_commands(gn)) {
		if (touchFlag) {
			Job_Touch(gn);
			return NULL;
		} else {
			Job *job;       	

			job = emalloc(sizeof(Job));
			if (job == NULL)
				Punt("can't create job: out of memory");

			job_attach_node(job, gn);
			return job;
		}
	} else {
		node_failure(gn);
		return NULL;
	}
}

static void
may_continue_job(Job *job)
{
	if (no_new_jobs) {
		if (DEBUG(EXPENSIVE))
			fprintf(stderr, "[%ld] expensive -> hold %s\n",
			    (long)mypid, job->node->name);
		job->next = heldJobs;
		heldJobs = job;
	} else
		continue_job(job);
}

static void
continue_job(Job *job)
{
	bool finished = job_run_next(job);
	if (finished)
		remove_job(job, true);
	else
		determine_expensive_job(job);
}

/*-
 *-----------------------------------------------------------------------
 * Job_Make  --
 *	Start a target-creation process going for the target described
 *	by the graph node gn.
 *
 * Side Effects:
 *	A new Job node is created and  its commands continued, which
 *	may fork the first command of that job.
 *-----------------------------------------------------------------------
 */
void
Job_Make(GNode *gn)
{
	Job *job;

	job = prepare_job(gn);
	if (!job)
		return;
	nJobs++;
	may_continue_job(job);
}

static void
determine_job_next_step(Job *job)
{
	bool okay;
	if (job->flags & JOB_IS_EXPENSIVE) {
		no_new_jobs = false;
		if (DEBUG(EXPENSIVE))
			fprintf(stderr, "[%ld] "
			    "Returning from expensive target %s, "
			    "allowing new jobs\n", (long)mypid, 
			    job->node->name);
	}

	okay = job->exit_type == JOB_EXIT_OKAY;
	if (!okay || job->next_cmd == NULL)
		remove_job(job, okay);
	else
		may_continue_job(job);
}

static void
remove_job(Job *job, bool okay)
{
	nJobs--;
	postprocess_job(job, okay);
	while (!no_new_jobs) {
		if (heldJobs != NULL) {
			job = heldJobs;
			heldJobs = heldJobs->next;
			if (DEBUG(EXPENSIVE))
				fprintf(stderr, "[%ld] cheap -> release %s\n",
				    (long)mypid, job->node->name);
			continue_job(job);
		} else
			break;
	}
}

/*
 * job = reap_finished_job(pid):
 * 	retrieve and remove a job from runningJobs, based on its pid
 *
 *	Note that we remove it right away, so that handle_signals()
 *	is accurate.
 */
static Job *
reap_finished_job(pid_t pid)
{
	Job **j, *job;

	for (j = &runningJobs; *j != NULL; j = &((*j)->next))
		if ((*j)->pid == pid) {
			job = *j;
			*j = job->next;
			return job;
		}

	return NULL;
}

/*
 * classic waitpid handler: retrieve as many dead children as possible.
 * returns true if succesful
 */
static bool
reap_jobs(void)
{
 	pid_t pid;	/* pid of dead child */
 	int status;	/* Exit/termination status */
	bool reaped = false;
	Job *job;

	while ((pid = waitpid(WAIT_ANY, &status, WNOHANG)) > 0) {
		reaped = true;
		job = reap_finished_job(pid);

		if (job == NULL) {
			Punt("Child (%ld) not in table?", (long)pid);
		} else {
			job_handle_status(job, status);
			determine_job_next_step(job);
		}
	}
	/* sanity check, should not happen */
	if (pid == -1 && errno == ECHILD && runningJobs != NULL)
		Punt("Process has no children, but runningJobs is not empty ?");
	return reaped;
}

void
handle_running_jobs(void)
{
	sigset_t old;

	/* reaping children in the presence of caught signals */

	/* first, we make sure to hold on new signals, to synchronize
	 * reception of new stuff on sigsuspend
	 */
	sigprocmask(SIG_BLOCK, &sigset, &old);
	while (runningJobs != NULL) {
		/* did we already have pending stuff that advances things ?
		 * then handle_all_signals() will not return
		 * or reap_jobs() will reap_jobs()
		 */
		handle_all_signals();
		if (reap_jobs())
			break;
		/* okay, so it's safe to suspend, we have nothing to do but
		 * wait...
		 */
		sigsuspend(&emptyset);
	}
	sigprocmask(SIG_SETMASK, &old, NULL);
}

void
handle_one_job(Job *job)
{
	int stat;
	int status;
	sigset_t old;

	sigprocmask(SIG_BLOCK, &sigset, &old);
	while (1) {
		handle_all_signals();
		stat = waitpid(job->pid, &status, WNOHANG);
		if (stat == job->pid)
			break;
		sigsuspend(&emptyset);
	}
	runningJobs = NULL;
	job_handle_status(job, status);
	sigprocmask(SIG_SETMASK, &old, NULL);
}

static void
loop_handle_running_jobs()
{
	while (runningJobs != NULL)
		handle_running_jobs();
}

void
Job_Init(int maxproc)
{
	runningJobs = NULL;
	heldJobs = NULL;
	errorJobs = NULL;
	maxJobs = maxproc;
	mypid = getpid();

	nJobs = 0;

	aborting = 0;
	setup_all_signals();
}

bool
can_start_job(void)
{
	if (aborting || nJobs >= maxJobs)
		return false;
	else
		return true;
}

bool
Job_Empty(void)
{
	return runningJobs == NULL;
}

/*-
 *-----------------------------------------------------------------------
 * handle_fatal_signal --
 *	Handle the receipt of a fatal interrupt
 *
 * Side Effects:
 *	All children are killed. Another job may be started if there
 *	is an interrupt target and the signal was SIGINT.
 *-----------------------------------------------------------------------
 */
static void
handle_fatal_signal(int signo)
{
	Job *job;

	debug_kill_printf("handle_fatal_signal(%d) called.\n", signo);


	for (job = runningJobs; job != NULL; job = job->next) {
		debug_kill_printf("passing to "
		    "child %ld running %s: %s\n", (long)job->pid,
		    job->node->name, really_kill(job, signo));
		may_remove_target(job);
	}

	if (signo == SIGINT && !touchFlag) {
		if ((interrupt_node->type & OP_DUMMY) == 0) {
			ignoreErrors = false;

			Job_Make(interrupt_node);
		}
	}
	loop_handle_running_jobs();
	print_errors();

	/* die by that signal */
	sigprocmask(SIG_BLOCK, &sigset, NULL);
	signal(signo, SIG_DFL);
	kill(getpid(), signo);
	sigprocmask(SIG_SETMASK, &emptyset, NULL);
	/*NOTREACHED*/
}

/*
 *-----------------------------------------------------------------------
 * Job_Finish --
 *	Do final processing such as the running of the commands
 *	attached to the .END target.
 *
 *	return true if fatal errors have happened.
 *-----------------------------------------------------------------------
 */
bool
Job_Finish(void)
{
	bool errors = errorJobs != NULL;

	if ((end_node->type & OP_DUMMY) == 0) {
		if (errors) {
			Error("Errors reported so .END ignored");
		} else {
			Job_Make(end_node);
			loop_handle_running_jobs();
		}
	}
	return errors;
}

void
Job_Begin(void)
{
	if ((begin_node->type & OP_DUMMY) == 0) {
		Job_Make(begin_node);
		loop_handle_running_jobs();
	}
}

/*-
 *-----------------------------------------------------------------------
 * Job_Wait --
 *	Waits for all running jobs to finish and returns. Sets 'aborting'
 *	to ABORT_WAIT to prevent other jobs from starting.
 *
 * Side Effects:
 *	Currently running jobs finish.
 *
 *-----------------------------------------------------------------------
 */
void
Job_Wait(void)
{
	aborting = ABORT_WAIT;
	loop_handle_running_jobs();
	aborting = 0;
}

/*-
 *-----------------------------------------------------------------------
 * Job_AbortAll --
 *	Abort all currently running jobs without handling output or anything.
 *	This function is to be called only in the event of a major
 *	error.
 *
 * Side Effects:
 *	All children are killed
 *-----------------------------------------------------------------------
 */
void
Job_AbortAll(void)
{
	Job *job;	/* the job descriptor in that element */
	int foo;

	aborting = ABORT_ERROR;

	for (job = runningJobs; job != NULL; job = job->next) {
		killpg(job->pid, SIGINT);
		killpg(job->pid, SIGKILL);
	}

	/*
	 * Catch as many children as want to report in at first, then give up
	 */
	while (waitpid(WAIT_ANY, &foo, WNOHANG) > 0)
		continue;
}