summaryrefslogtreecommitdiff
path: root/libthreads/cprocs.c
blob: 5fb138d246ff3f23da93bf301339de5b29944133 (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
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
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
/*
 * Mach Operating System
 * Copyright (c) 1993-1989 Carnegie Mellon University
 * All Rights Reserved.
 *
 * Permission to use, copy, modify and distribute this software and its
 * documentation is hereby granted, provided that both the copyright
 * notice and this permission notice appear in all copies of the
 * software, derivative works or modified versions, and any portions
 * thereof, and that both notices appear in supporting documentation.
 *
 * CARNEGIE MELLON ALLOWS FREE USE OF THIS SOFTWARE IN ITS "AS IS"
 * CONDITION.  CARNEGIE MELLON DISCLAIMS ANY LIABILITY OF ANY KIND FOR
 * ANY DAMAGES WHATSOEVER RESULTING FROM THE USE OF THIS SOFTWARE.
 *
 * Carnegie Mellon requests users of this software to return to
 *
 *  Software Distribution Coordinator  or  Software.Distribution@CS.CMU.EDU
 *  School of Computer Science
 *  Carnegie Mellon University
 *  Pittsburgh PA 15213-3890
 *
 * any improvements or extensions that they make and grant Carnegie Mellon
 * the rights to redistribute these changes.
 */
/*
 * HISTORY
 * 26-Oct-94 Johannes Helander (jvh) Helsinki University of Technology
 *	Set the wait_type field.
 *
 * $Log: cprocs.c,v $
 * Revision 1.16  2002/05/27 02:50:10  roland
 * 2002-05-26  Roland McGrath  <roland@frob.com>
 *
 * 	Changes merged from CMU MK83a version:
 * 	* cthreads.h, options.h: Various cleanups.
 * 	* call.c, cthread_data.c, sync.c, mig_support.c: Likewise.
 * 	* i386/cthreads.h, i386/thread.c, i386/lock.s: Likewise.
 * 	* cthread_internals.h: Add decls for internal functions.
 * 	(struct cproc): Use vm_offset_t for stack_base and stack_size members.
 * 	Use natural_t for context member.
 * 	* cprocs.c: Use prototypes for all defns.
 * 	* cthreads.c: Likewise.
 * 	(cthread_exit): Cast any_t to integer_t before int.
 *
 * Revision 2.18  93/03/09  10:59:10  danner
 * 	Lint.
 * 	[93/03/06            af]
 *
 * Revision 2.17  93/01/19  08:55:44  danner
 * 	Added missing spin_lock_t type from cproc_list_lock decl.
 * 	[92/12/30            af]
 *
 *
 * Revision 2.16  93/01/14  18:04:46  danner
 * 	Convert file to ANSI C.
 * 	[92/12/18            pds]
 * 	64bit cleanup.
 * 	[92/12/10  21:08:32  af]
 *
 * Revision 2.15  92/03/06  14:09:31  rpd
 * 	Replaced swtch_pri with yield.
 * 	[92/03/06            rpd]
 *
 * Revision 2.14  91/08/28  11:19:16  jsb
 * 	Fixed the loop in cproc_fork_child that frees cprocs.
 * 	[91/08/23            rpd]
 *
 * Revision 2.13  91/07/31  18:33:04  dbg
 * 	Fix some more bad types.  Ints are NOT pointers.
 *
 * 	Fix argument type mismatch in cproc_create.
 * 	[91/07/30  17:32:59  dbg]
 *
 * Revision 2.12  91/05/14  17:56:11  mrt
 * 	Correcting copyright
 *
 * Revision 2.11  91/02/14  14:19:26  mrt
 * 	Added new Mach copyright
 * 	[91/02/13  12:40:50  mrt]
 *
 * Revision 2.10  90/11/05  14:36:41  rpd
 * 	Added cproc_fork_{prepare,parent,child}.
 * 	[90/11/02            rwd]
 *
 * 	Fix for positive stack growth.
 * 	[90/11/01            rwd]
 *
 * 	Add spin_lock_t.
 * 	[90/10/31            rwd]
 *
 * Revision 2.9  90/10/12  13:07:12  rpd
 * 	Fix type
 * 	[90/10/10  15:09:59  rwd]
 *
 * 	Comment code.
 * 	[90/10/02            rwd]
 *
 * Revision 2.8  90/09/09  14:34:44  rpd
 * 	Remove special mutex.  Remove thread_calls and debug_mutex
 * 	[90/08/24            rwd]
 * 	Fix up old call to cthread_msg_busy to new format.
 * 	[90/08/22            rwd]
 *
 * Revision 2.7  90/08/06  15:09:17  rwd
 * 	Fixed arguments to cthread_mach_msg.
 * 	[90/06/26            rwd]
 * 	Add additional STATISTICS.
 * 	[90/06/07            rwd]
 *
 * 	Attempt to reduce number of times a cthread is released to to a
 * 	msg_receive by adding min/max instead of single number to
 * 	cthread_msg calls.
 * 	[90/06/06            rwd]
 *
 * Revision 2.6  90/06/02  15:13:36  rpd
 * 	Converted to new IPC.
 * 	[90/03/20  20:46:16  rpd]
 *
 * Revision 2.5  90/05/29  18:40:11  rwd
 * 	Don't incr special field until the mutex grab is successful.
 * 	[90/05/09            rwd]
 *
 * Revision 2.4  90/03/14  21:12:02  rwd
 * 	Added WAIT_DEBUG code for deadlock debugging.
 * 	[90/03/01            rwd]
 * 	Insert cprocs in cproc_list as allocated.
 * 	[90/03/01  10:20:16  rwd]
 *
 * Revision 2.3  90/01/19  14:36:57  rwd
 * 	Make cthread_msg_busy only release new thread if this is still
 * 	busy.  Ie don't release two on back to back calls.
 * 	[90/01/11            rwd]
 * 	Add THREAD_CALL code.  Add CPROC_ARUN state.
 * 	[90/01/03            rwd]
 * 	Add new cthread_msg_rpc call
 * 	[89/12/20            rwd]
 * 	Change cproc_self pointer to top of stack.  Now need to change
 * 	the stack of the first thread.
 * 	[89/12/12            rwd]
 *
 * Revision 2.2  89/12/08  19:53:13  rwd
 * 	Added CPROC_CONDWAIT state to deal with lock held
 * 	across mutex_unlock problem.
 * 	[89/11/29            rwd]
 * 	Changed mutexes to not hand off.  MUTEX_EXTRA conditional is
 * 	now obsolete.
 * 	[89/11/27            rwd]
 *
 * 	Add MUTEX_EXTRA code for extra kernel threads to serve special
 * 	mutexes in time of need.
 * 	[89/11/25            rwd]
 * 	Add MUTEX_SPECIAL and DEBUG_MUTEX code
 * 	[89/11/24            rwd]
 * 	Changed mutex_lock to mutex_lock_solid.  Mutex_lock is now a
 * 	macro which tries the spin_lock before making a subroutine call.
 * 	Mutex_unlock is now a macro with mutex_unlock_solid for worst case.
 * 	[89/11/13            rwd]
 *
 * 	Rewrite most to merge coroutine and thread implementation.
 * 	New routines are cthread_set_kernel_limit, cthread_kernel_limit,
 * 	cthread_wire, cthread_unwire, and cthread_receive.
 * 	[89/10/23            rwd]
 *
 * Revision 2.1  89/08/03  17:07:10  rwd
 * Created.
 *
 * 11-Apr-89  David Golub (dbg) at Carnegie-Mellon University
 *	Made condition_yield loop break if swtch_pri returns TRUE (in
 *	case we fix it).
 *
 * 31-Mar-89  David Golub (dbg) at Carnegie-Mellon University
 *	Change cond_signal, cond_broadcast, and cproc_continue so that
 *	the condition's spin lock is not held while continuing the
 *	process.
 *
 * 16-Jan-89  David Golub (dbg) at Carnegie-Mellon University
 *	Changes for stand-alone library to run on pure kernel:
 *	. made IPC_WAIT standard, as calls that are used if IPC_WAIT == 0
 *	  vanished a year ago.
 *	. Removed (as much as possible) references to stdio or other U*X
 *	  features.
 *
 *
 * 01-Apr-88  Eric Cooper (ecc) at Carnegie Mellon University
 *	Changed condition_clear(c) to acquire c->lock,
 *	to serialize after any threads still doing condition_signal(c).
 *	Suggested by Dan Julin.
 *
 * 19-Feb-88  Eric Cooper (ecc) at Carnegie Mellon University
 * 	Extended the inline scripts to handle spin_unlock() and mutex_unlock().
 *
 * 28-Jan-88  David Golub (dbg) at Carnegie Mellon University
 *	Removed thread_data argument from thread_create
 *	and converted to new thread_set_state call.
 *
 * 01-Dec-87  Eric Cooper (ecc) at Carnegie Mellon University
 *	Added inline expansion for cthread_sp() function.
 *
 * 21-Aug-87  Eric Cooper (ecc) at Carnegie Mellon University
 *	Fixed uninitialized reply_port in cproc_alloc() (found by rds).
 *
 * 14-Aug-87  Eric Cooper (ecc) at Carnegie Mellon University
 *	Tried using return value of swtch() to guide condition_wait().
 *	Performance was worse than using a hybrid spin/yield/block
 *	scheme, so the version using swtch() was commented out.
 *	Disabled IPC_WAIT in released version.
 *
 * 13-Aug-87  Eric Cooper (ecc) at Carnegie Mellon University
 *	Added IPC_WAIT option.
 *	If defined, thread synchronization (condition_wait() and
 *	cproc_continue()) are implemented using	msg_receive() and
 *	msg_send() instead of thread_suspend() and thread_resume().
 *
 * 11-Aug-87  Eric Cooper (ecc) at Carnegie Mellon University
 *	Moved thread reply port to cproc structure in cthread_internals.h,
 *	because mig calls are made while cproc is idle (no cthread structure).
 *	Changed cproc_switch() and cproc_start (COROUTINE implementation)
 *	to use address of saved context, rather than address of enclosing cproc,
 *	to eliminate dependency on cproc layout.
 */
/*
 * 	File: 	cprocs.c
 *	Author: Eric Cooper, Carnegie Mellon University
 *	Date:	Aug, 1987
 *
 * 	Implementation of cprocs (lightweight processes)
 * 	and primitive synchronization operations.
 */


#include <cthreads.h>
#include "cthread_internals.h"
#include <mach/message.h>
#include <hurd/threadvar.h>	/* GNU */
#include <assert-backtrace.h>

/*
 * Port_entry's are used by cthread_mach_msg to store information
 * about each port/port_set for which it is managing threads
 */

typedef struct port_entry {
	struct port_entry *next;	/* next port_entry */
	mach_port_t	port;		/* which port/port_set */
	struct cthread_queue queue;	/* queue of runnable threads for
					   this port/port_set */
	int min;			/* minimum number of kernel threads
					   to be used by this port/port_set */
	int max;			/* maximum number of kernel threads
					   to be used by this port/port_set */
	int held;			/* actual number of kernel threads
					   currentlt in use */
	spin_lock_t lock;		/* lock governing all above fields */
} *port_entry_t;

#define PORT_ENTRY_NULL ((port_entry_t) 0)

/* Available to outside for statistics */

int cthread_wait_stack_size = 8192;	/* stack size for idle threads */
int cthread_max_kernel_threads = 0;	/* max kernel threads */
int cthread_kernel_threads = 0;		/* current kernel threads */
private spin_lock_t n_kern_lock = SPIN_LOCK_INITIALIZER;
					/* lock for 2 above */
#ifdef STATISTICS
int cthread_ready = 0;			/* currently runnable */
int cthread_running = 1;		/* currently running */
int cthread_waiting = 0;		/* currently waiting */
int cthread_wired = 0;			/* currently wired */
private spin_lock_t wired_lock = SPIN_LOCK_INITIALIZER;
					/* lock for above */
int cthread_wait_stacks = 0;		/* total cthread waiting stacks */
int cthread_waiters = 0;		/* total of watiers */
int cthread_wakeup = 0;			/* total times woken when starting to
					   block */
int cthread_blocked = 0;		/* total blocked */
int cthread_rnone = 0;			/* total times no cthread available
					   to meet minimum for port_entry */
int cthread_yields = 0;			/* total cthread_yields */
int cthread_none = 0;			/* total idle wakeups w/o runnable */
int cthread_switches = 0;		/* total number of cproc_switches */
int cthread_no_mutex = 0;		/* total number times woken to get
					   mutex and couldn't */
private spin_lock_t mutex_count_lock = SPIN_LOCK_INITIALIZER;
					/* lock for above */
#endif  /* STATISTICS */

cproc_t cproc_list = NO_CPROC;		/* list of all cprocs */
private spin_lock_t cproc_list_lock = SPIN_LOCK_INITIALIZER;
					/* lock for above */
private int cprocs_started = FALSE;	/* initialized? */
private struct cthread_queue ready = QUEUE_INITIALIZER;
					/* ready queue */
private int ready_count = 0;		/* number of ready threads on ready
					   queue - number of messages sent */
private spin_lock_t ready_lock = SPIN_LOCK_INITIALIZER;
					/* lock for 2 above */
private mach_port_t wait_port = MACH_PORT_NULL;
					/* port on which idle threads wait */
private int wait_count = 0;		/* number of waiters - messages pending
					   to wake them */
private struct cthread_queue waiters = QUEUE_INITIALIZER;
					/* queue of cthreads to run as idle */
private spin_lock_t waiters_lock = SPIN_LOCK_INITIALIZER;
					/* lock for 2 above */
private port_entry_t port_list = PORT_ENTRY_NULL;
					/* master list of port_entries */
private spin_lock_t port_lock = SPIN_LOCK_INITIALIZER;
					/* lock for above queue */
private mach_msg_header_t wakeup_msg;	/* prebuilt message used by idle
					   threads */

/*
 * Return current value for max kernel threads
 * Note: 0 means no limit
 */
int
cthread_kernel_limit(void)
{
	return cthread_max_kernel_threads;
}

/*
 * Set max number of kernel threads
 * Note:	This will not currently terminate existing threads
 * 		over maximum.
 */

void
cthread_set_kernel_limit(int n)
{
	cthread_max_kernel_threads = n;
}

/*
 * Wire a cthread to its current kernel thread
 */

void
cthread_wire(void)
{
	register cproc_t p = cproc_self();

	/* In GNU, we wire all threads on creation (in cproc_alloc).  */
	assert_backtrace (p->wired != MACH_PORT_NULL);
}

/*
 * Unwire a cthread.  Deallocate its wait port.
 */

void
cthread_unwire(void)
{
	register cproc_t p = cproc_self();

	/* This is bad juju in GNU, where all cthreads must be wired.  */
	abort();
#if 0
	if (p->wired != MACH_PORT_NULL) {
		MACH_CALL(mach_port_mod_refs(mach_task_self(), p->wired,
					     MACH_PORT_RIGHT_SEND, -1), r);
		MACH_CALL(mach_port_mod_refs(mach_task_self(), p->wired,
					     MACH_PORT_RIGHT_RECEIVE, -1), r);
		p->wired = MACH_PORT_NULL;
#ifdef STATISTICS
		spin_lock(&wired_lock);
		cthread_wired--;
		spin_unlock(&wired_lock);
#endif  /* STATISTICS */
	}
#endif
}

private cproc_t
cproc_alloc(void)
{
	register cproc_t p = (cproc_t) malloc(sizeof(struct cproc));
	kern_return_t r;

	p->incarnation = NO_CTHREAD;
#if 0
	/* This member is not used in GNU.  */
	p->reply_port = MACH_PORT_NULL;
#endif

	spin_lock_init(&p->lock);
	p->state = CPROC_RUNNING;
	p->busy = 0;

	/*
	 * In GNU, every cthread must be wired.  So we just
	 * initialize P->wired on creation.
	 *
	 * A wired thread has a port associated with it for all
	 * of its wait/block cases.  We also prebuild a wakeup
	 * message.
	 */

	MACH_CALL(mach_port_allocate(mach_task_self(),
				     MACH_PORT_RIGHT_RECEIVE,
				     &p->wired), r);
	MACH_CALL(mach_port_insert_right(mach_task_self(),
					 p->wired, p->wired,
					 MACH_MSG_TYPE_MAKE_SEND), r);
	p->msg.msgh_bits = MACH_MSGH_BITS(MACH_MSG_TYPE_COPY_SEND, 0);
	p->msg.msgh_size = 0; /* initialized in call */
	p->msg.msgh_remote_port = p->wired;
	p->msg.msgh_local_port = MACH_PORT_NULL;
	p->msg.msgh_kind = MACH_MSGH_KIND_NORMAL;
	p->msg.msgh_id = 0;

	spin_lock(&cproc_list_lock);
	p->list = cproc_list;
	cproc_list = p;
	spin_unlock(&cproc_list_lock);

	return p;
}

/*
 * Called by cthread_init to set up initial data structures.
 */

vm_offset_t
cproc_init(void)
{
	kern_return_t r;

	cproc_t p = cproc_alloc();

	cthread_kernel_threads = 1;

	MACH_CALL(mach_port_allocate(mach_task_self(),
				     MACH_PORT_RIGHT_RECEIVE,
				     &wait_port), r);
	MACH_CALL(mach_port_insert_right(mach_task_self(),
					 wait_port, wait_port,
					 MACH_MSG_TYPE_MAKE_SEND), r);

	wakeup_msg.msgh_bits = MACH_MSGH_BITS(MACH_MSG_TYPE_COPY_SEND, 0);
	wakeup_msg.msgh_size = 0; /* initialized in call */
	wakeup_msg.msgh_remote_port = wait_port;
	wakeup_msg.msgh_local_port = MACH_PORT_NULL;
	wakeup_msg.msgh_kind = MACH_MSGH_KIND_NORMAL;
	wakeup_msg.msgh_id = 0;

	cprocs_started = TRUE;


	/*
	 * We pass back the new stack which should be switched to
	 * by crt0.  This guarantess correct size and alignment.
	 */
	return (stack_init(p));
}

/*
 * Insert cproc on ready queue.  Make sure it is ready for queue by
 * synching on its lock.  Just send message to wired cproc.
 */

private boolean_t cproc_ready(register cproc_t p, register int preq)
{
	register cproc_t s=cproc_self();
	kern_return_t r;

	if (p->wired != MACH_PORT_NULL) {
		r = mach_msg(&p->msg, MACH_SEND_MSG,
			     sizeof p->msg, 0, MACH_PORT_NULL,
			     MACH_MSG_TIMEOUT_NONE, MACH_PORT_NULL);
#ifdef CHECK_STATUS
		if (r != MACH_MSG_SUCCESS) {
			mach_error("mach_msg", r);
			exit(1);
		}
#endif  /* CHECK_STATUS */
		return TRUE;
	}
	spin_lock(&p->lock);		/* is it ready to be queued?  It
					   can appear on a queue before
					   being switched from.  This lock
					   is released by cproc_switch as
					   its last operation. */
	if (p->state & CPROC_SWITCHING) {
		/*
		 * We caught it early on.  Just set to RUNNING
		 * and we will save a lot of time.
		 */
		p->state = (p->state & ~CPROC_SWITCHING) | CPROC_RUNNING;
		spin_unlock(&p->lock);
		return TRUE;
	}
	spin_unlock(&p->lock);

	spin_lock(&ready_lock);

	if (preq) {
		cthread_queue_preq(&ready, p);
	} else {
		cthread_queue_enq(&ready, p);
	}
#ifdef STATISTICS
	cthread_ready++;
#endif  /* STATISTICS */
	ready_count++;

	if ((s->state & CPROC_CONDWAIT) && !(s->wired)) {
		/*
		 * This is an optimiztion.  Don't bother waking anyone to grab
		 * this guy off the ready queue since my thread will block
		 * momentarily for the condition wait.
		 */

		spin_unlock(&ready_lock);
		return TRUE;
	}

	if ((ready_count > 0) && wait_count) {
		wait_count--;
		ready_count--;
		spin_unlock(&ready_lock);
		r = mach_msg(&wakeup_msg, MACH_SEND_MSG,
			     sizeof wakeup_msg, 0, MACH_PORT_NULL,
			     MACH_MSG_TIMEOUT_NONE, MACH_PORT_NULL);
#ifdef CHECK_STATUS
		if (r != MACH_MSG_SUCCESS) {
			mach_error("mach_msg", r);
			exit(1);
		}
#endif  /* CHECK_STATUS */
		return TRUE;
	}
	spin_unlock(&ready_lock);
	return FALSE;
}

/*
 * This is only run on a partial "waiting" stack and called from
 * cproc_start_wait
 */

void
cproc_waiting(cproc_t p)
{
	mach_msg_header_t msg;
	register cproc_t new;
	kern_return_t r;

#ifdef STATISTICS
	spin_lock(&ready_lock);
	cthread_waiting++;
	cthread_waiters++;
	spin_unlock(&ready_lock);
#endif  /* STATISTICS */
	for (;;) {
		MACH_CALL(mach_msg(&msg, MACH_RCV_MSG,
				   0, sizeof msg, wait_port,
				   MACH_MSG_TIMEOUT_NONE, MACH_PORT_NULL), r);
		spin_lock(&ready_lock);
		cthread_queue_deq(&ready, cproc_t, new);
		if (new != NO_CPROC) break;
		wait_count++;
		ready_count++;
#ifdef STATISTICS
		cthread_none++;
#endif  /* STATISTICS */
		spin_unlock(&ready_lock);
	}
#ifdef STATISTICS
	cthread_ready--;
	cthread_running++;
	cthread_waiting--;
#endif  /* STATISTICS */
	spin_unlock(&ready_lock);
	spin_lock(&new->lock);
	new->state = CPROC_RUNNING;
	spin_unlock(&new->lock);
	spin_lock(&waiters_lock);
	cthread_queue_enq(&waiters, p);
	spin_lock(&p->lock);
	spin_unlock(&waiters_lock);
	cproc_switch(&p->context,&new->context,&p->lock);
}

/*
 * Get a waiter with stack
 *
 */

private cproc_t
cproc_waiter(void)
{
	register cproc_t waiter;

	spin_lock(&waiters_lock);
	cthread_queue_deq(&waiters, cproc_t, waiter);
	spin_unlock(&waiters_lock);
	if (waiter == NO_CPROC) {
		vm_address_t base;
		kern_return_t r;
#ifdef STATISTICS
		spin_lock(&waiters_lock);
		cthread_wait_stacks++;
		spin_unlock(&waiters_lock);
#endif  /* STATISTICS */
		waiter = cproc_alloc();
		MACH_CALL(vm_allocate(mach_task_self(), &base,
				      cthread_wait_stack_size, TRUE), r);
		waiter->stack_base = base;
		waiter->stack_size = cthread_wait_stack_size;
	}
	return (waiter);
}


/*
 * Current cproc is blocked so switch to any ready cprocs, or, if
 * none, go into the wait state.
 *
 * You must hold cproc_self()->lock when called.
 */

void
cproc_block(void)
{
  extern unsigned int __hurd_threadvar_max; /* GNU */
	register cproc_t waiter, new, p = cproc_self();

	if (p->wired != MACH_PORT_NULL) {
		mach_msg_header_t msg;
		kern_return_t r;

		spin_unlock(&p->lock);
		MACH_CALL(mach_msg(&msg, MACH_RCV_MSG,
				   0, sizeof msg, p->wired,
				   MACH_MSG_TIMEOUT_NONE, MACH_PORT_NULL), r);
		return;
	}
	p->state = CPROC_SWITCHING;
	spin_unlock(&p->lock);
	spin_lock(&ready_lock);
#ifdef STATISTICS
	cthread_blocked++;
#endif  /* STATISTICS */
	cthread_queue_deq(&ready, cproc_t, new);
	if (new) {
#ifdef STATISTICS
		cthread_ready--;
		cthread_switches++;
#endif  /* STATISTICS */
		ready_count--;
		spin_unlock(&ready_lock);
		spin_lock(&p->lock);
		if (p->state == CPROC_RUNNING) { /* have we been saved */
			spin_unlock(&p->lock);
#ifdef STATISTICS
			spin_lock(&ready_lock);
			cthread_wakeup++;
			cthread_switches--;
			spin_unlock(&ready_lock);
#endif  /* STATISTICS */
			cproc_ready(new, 1);  /* requeue at head were it was */
		} else {
			p->state = CPROC_BLOCKED;
			spin_lock(&new->lock); /* incase still switching */
			new->state = CPROC_RUNNING;
			spin_unlock(&new->lock);
			cproc_switch(&p->context,&new->context,&p->lock);
		}
	} else {
		wait_count++;
#ifdef STATISTICS
		cthread_running--;
#endif  /* STATISTICS */
		spin_unlock(&ready_lock);
		waiter = cproc_waiter();
		spin_lock(&p->lock);
		if (p->state == CPROC_RUNNING) { /* we have been saved */
			spin_unlock(&p->lock);
			spin_lock(&ready_lock);
			wait_count--;
#ifdef STATISTICS
			cthread_running++;
			cthread_wakeup++;
#endif  /* STATISTICS */
			spin_unlock(&ready_lock);
			spin_lock(&waiters_lock);
			cthread_queue_preq(&waiters, waiter);
			spin_unlock(&waiters_lock);
		} else {
			p->state = CPROC_BLOCKED;
			spin_lock(&waiter->lock); /* in case still switching */
			spin_unlock(&waiter->lock);
			cproc_start_wait
			  (&p->context, waiter,
			   cproc_stack_base(waiter,
					    sizeof(ur_cthread_t *) +
					    /* Account for GNU per-thread
					       variables.  */
					    __hurd_threadvar_max *
					    sizeof (long int)),
			   &p->lock);
		}
	}
}

/*
 * Implement C threads using MACH threads.
 */
cproc_t
cproc_create(void)
{
	register cproc_t child = cproc_alloc();
	register kern_return_t r;
	extern void cproc_setup();
	extern void cproc_prepare();
	extern void cthread_body();
	thread_t n;

      	alloc_stack(child);
	spin_lock(&n_kern_lock);
	if (cthread_max_kernel_threads == 0 ||
	    cthread_kernel_threads < cthread_max_kernel_threads) {
		tcbhead_t *tcb = _dl_allocate_tls(NULL);
		cthread_kernel_threads++;
		spin_unlock(&n_kern_lock);
		MACH_CALL(thread_create(mach_task_self(), &n), r);
		cproc_setup(child, n, tcb, cthread_body);	/* machine dependent */
		MACH_CALL(thread_resume(n), r);
#ifdef STATISTICS
		spin_lock(&ready_lock);
		cthread_running++;
		spin_unlock(&ready_lock);
#endif  /* STATISTICS */
	} else {
	  	vm_offset_t stack;
		spin_unlock(&n_kern_lock);
		child->state = CPROC_BLOCKED;
		/* The original CMU code does the excessively clever
		   optimization of putting CHILD at the base of the stack
		   and setting up to be the argument to cthread_body in the
		   same place (by passing zero as the second arg to
		   cproc_stack_base here)..  This doesn't fly for GNU,
		   because we need some more space allocated at the base of
		   the stack, after the cproc_self pointer (where CHILD is
		   stored).  */
		stack = cproc_stack_base(child,
					 sizeof(ur_cthread_t *) +
					 /* Account for GNU per-thread
					    variables.  */
					 __hurd_threadvar_max *
					 sizeof (long int));
		cproc_prepare(child, &child->context, stack, &cthread_body);
		/* Set up the cproc_self ptr at the base of CHILD's stack.  */
		ur_cthread_ptr(stack) = (ur_cthread_t) child;
		cproc_ready(child,0);
	}
	return child;
}

void
condition_wait(condition_t c, mutex_t m)
{
	register cproc_t p = cproc_self();

	p->state = CPROC_CONDWAIT | CPROC_SWITCHING;

	spin_lock(&c->lock);
	cthread_queue_enq(&c->queue, p);
	spin_unlock(&c->lock);
#ifdef	WAIT_DEBUG
	p->waiting_for = (char *)c;
#endif	 /* WAIT_DEBUG */

	mutex_unlock(m);

	spin_lock(&p->lock);
	if (p->state & CPROC_SWITCHING) {
		cproc_block();
	} else {
		p->state = CPROC_RUNNING;
		spin_unlock(&p->lock);
	}


#ifdef	WAIT_DEBUG
	p->waiting_for = (char *)0;
#endif	 /* WAIT_DEBUG */

	/*
	 * Re-acquire the mutex and return.
	 */
	mutex_lock(m);
}

/* Declare that IMPLICATOR should consider IMPLICATAND's waiter queue
   to be an extension of its own queue.  It is an error for either
   condition to be deallocated as long as the implication persists. */
void
condition_implies (condition_t implicator, condition_t implicatand)
{
  struct cond_imp *imp;

  imp = malloc (sizeof (struct cond_imp));
  imp->implicatand = implicatand;
  imp->next = implicator->implications;
  implicator->implications = imp;
}

/* Declare that the implication relationship from IMPLICATOR to
   IMPLICATAND should cease. */
void
condition_unimplies (condition_t implicator, condition_t implicatand)
{
  struct cond_imp **impp;

  for (impp = &implicator->implications; *impp; impp = &(*impp)->next)
    {
      if ((*impp)->implicatand == implicatand)
	{
	  struct cond_imp *tmp = *impp;
	  *impp = (*impp)->next;
	  free (tmp);
	  return;
	}
    }
}

/* Signal one waiter on C.  If there were no waiters at all, return
   0, else return 1. */
int
cond_signal(condition_t c)
{
	register cproc_t p;
	struct cond_imp *imp;

	spin_lock(&c->lock);
	cthread_queue_deq(&c->queue, cproc_t, p);
	spin_unlock(&c->lock);
	if (p != NO_CPROC) {
		cproc_ready(p,0);
		return 1;
	}
	else {
		for (imp = c->implications; imp; imp = imp->next)
			if (cond_signal (imp->implicatand))
				return 1;
        }
	return 0;
}

void
cond_broadcast(condition_t c)
{
	register cproc_t p;
	struct cthread_queue blocked_queue;
	struct cond_imp *imp;

	cthread_queue_init(&blocked_queue);

	spin_lock(&c->lock);
	for (;;) {
		cthread_queue_deq(&c->queue, cproc_t, p);
		if (p == NO_CPROC)
			break;
		cthread_queue_enq(&blocked_queue, p);
	}
	spin_unlock(&c->lock);

	for(;;) {
		cthread_queue_deq(&blocked_queue, cproc_t, p);
		if (p == NO_CPROC)
			break;
		cproc_ready(p,0);
	}

	for (imp = c->implications; imp; imp = imp->next)
		condition_broadcast (imp->implicatand);
}

void
cthread_yield(void)
{
	register cproc_t new, p = cproc_self();

	if (p->wired != MACH_PORT_NULL) {
		yield();
		return;
	}
	spin_lock(&ready_lock);
#ifdef STATISTICS
	cthread_yields++;
#endif  /* STATISTICS */
	cthread_queue_deq(&ready, cproc_t, new);
	if (new) {
		cthread_queue_enq(&ready, p);
		spin_lock(&p->lock);
		p->state = CPROC_BLOCKED;
		spin_unlock(&ready_lock);
		spin_lock(&new->lock);
		new->state = CPROC_RUNNING;
		spin_unlock(&new->lock);
		cproc_switch(&p->context,&new->context,&p->lock);
	} else {
		spin_unlock(&ready_lock);
		yield();
	}
}

/*
 * Mutex objects.
 */

void
__mutex_lock_solid(void *ptr)
{
	register mutex_t m = ptr;
	register cproc_t p = cproc_self();
	register int queued;
	register int tried = 0;

#ifdef	WAIT_DEBUG
	p->waiting_for = (char *)m;
#endif	 /* WAIT_DEBUG */
	while (1) {
		spin_lock(&m->lock);
		if (cthread_queue_head(&m->queue, cproc_t) == NO_CPROC) {
			cthread_queue_enq(&m->queue, p);
			queued = 1;
		} else {
			queued = 0;
		}
		if (spin_try_lock(&m->held)) {
			if (queued) cthread_queue_deq(&m->queue, cproc_t, p);
			spin_unlock(&m->lock);
#ifdef	WAIT_DEBUG
			p->waiting_for = (char *)0;
#endif	 /* WAIT_DEBUG */
			return;
		} else {
			if (!queued) cthread_queue_enq(&m->queue, p);
			spin_lock(&p->lock);
			spin_unlock(&m->lock);
			cproc_block();
			if (spin_try_lock(&m->held)) {
#ifdef	WAIT_DEBUG
			    p->waiting_for = (char *)0;
#endif	 /* WAIT_DEBUG */
			    return;
			}
#ifdef STATISTICS
			spin_lock(&mutex_count_lock);
			cthread_no_mutex++;
			spin_unlock(&mutex_count_lock);
#endif  /* STATISTICS */
		}
	}
}

void
__mutex_unlock_solid(void *ptr)
{
	register mutex_t m = ptr;
	register cproc_t new;

	if (!spin_try_lock(&m->held))
		return;
	spin_lock(&m->lock);
	cthread_queue_deq(&m->queue, cproc_t, new);
	spin_unlock(&m->held);
	spin_unlock(&m->lock);
	if (new) {
		cproc_ready(new,0);
	}
}


/*
 * Use instead of mach_msg in a multi-threaded server so as not
 * to tie up excessive kernel threads.  This uses a simple linked list for
 * ports since this should never be more than a few.
 */

/*
 * A cthread holds a reference to a port_entry even after it receives a
 * message.  This reference is not released until the thread does a
 * cthread_msg_busy.  This allows the fast case of a single mach_msg
 * call to occur as often as is possible.
 */

private port_entry_t
get_port_entry(mach_port_t port, int min, int max)
{
	register port_entry_t i;

	spin_lock(&port_lock);
	for(i=port_list;i!=PORT_ENTRY_NULL;i=i->next)
		if (i->port == port) {
			spin_unlock(&port_lock);
			return i;
		}
	i = (port_entry_t)malloc(sizeof(struct port_entry));
	cthread_queue_init(&i->queue);
	i->port = port;
	i->next = port_list;
	port_list = i;
	i->min = min;
	i->max = max;
	i->held = 0;
	spin_lock_init(&i->lock);
	spin_unlock(&port_lock);
	return i;
}

void
cthread_msg_busy(mach_port_t port, int min, int max)
{
	register port_entry_t port_entry;
	register cproc_t new, p = cproc_self();

	if (p->busy) {
	    port_entry = get_port_entry(port, min, max);
	    spin_lock(&port_entry->lock);
	    p->busy = 0;
	    if (port_entry->held <= port_entry->min) {
		cthread_queue_deq(&port_entry->queue, cproc_t, new);
		if (new != NO_CPROC){
		    spin_unlock(&port_entry->lock);
		    cproc_ready(new,0);
		} else {
		    port_entry->held--;
		    spin_unlock(&port_entry->lock);
#ifdef STATISTICS
		    spin_lock(&port_lock);
		    cthread_rnone++;
		    spin_unlock(&port_lock);
#endif  /* STATISTICS */
		}
	    } else {
		port_entry->held--;
		spin_unlock(&port_entry->lock);
	    }
	}

}

void
cthread_msg_active(mach_port_t port, int min, int max)
{
	register cproc_t p = cproc_self();
	register port_entry_t port_entry;

	if (!p->busy) {
	    port_entry = get_port_entry(port, min, max);
	    if (port_entry == 0) return;
	    spin_lock(&port_entry->lock);
	    if (port_entry->held < port_entry->max) {
		port_entry->held++;
		p->busy = port_entry;
	    }
	    spin_unlock(&port_entry->lock);
	}
}

mach_msg_return_t
cthread_mach_msg(register mach_msg_header_t *header,
		 register mach_msg_option_t option, mach_msg_size_t send_size,
		 mach_msg_size_t rcv_size, register mach_port_t rcv_name,
		 mach_msg_timeout_t timeout, mach_port_t notify, int min,
		 int max)
{
	register port_entry_t port_entry;
	register cproc_t p = cproc_self();
	register int sent=0;
	mach_msg_return_t r;
	port_entry_t op = (port_entry_t)p->busy;

	port_entry = get_port_entry(rcv_name, min, max);

	if (op && (port_entry_t)op != port_entry)
	    cthread_msg_busy(op->port, op->min, op->max);
	spin_lock(&port_entry->lock);
	if (!(port_entry == (port_entry_t)p->busy)) {
	    if (port_entry->held >= max) {
		if (option & MACH_SEND_MSG) {
		    spin_unlock(&port_entry->lock);
		    r = mach_msg(header, option &~ MACH_RCV_MSG,
				 send_size, 0, MACH_PORT_NULL,
				 timeout, notify);
		    if (r != MACH_MSG_SUCCESS) return r;
		    spin_lock(&port_entry->lock);
		    sent=1;
		}
		if (port_entry->held >= max) {
		    spin_lock(&p->lock);
		    cthread_queue_preq(&port_entry->queue, p);
		    spin_unlock(&port_entry->lock);
#ifdef	WAIT_DEBUG
		    p->waiting_for = (char *)port_entry;
#endif	 /* WAIT_DEBUG */
		    cproc_block();
		} else {
		    port_entry->held++;
		    spin_unlock(&port_entry->lock);
		}
	    } else {
		port_entry->held++;
		spin_unlock(&port_entry->lock);
	    }
	} else {
		spin_unlock(&port_entry->lock);
	}
#ifdef	WAIT_DEBUG
	p->waiting_for = (char *)0;
#endif	 /* WAIT_DEBUG */
	p->busy = port_entry;
	if ((option & MACH_SEND_MSG) && !sent) {
	    r = mach_msg(header, option,
			 send_size, rcv_size, rcv_name,
			 timeout, notify);
	} else {
	    r = mach_msg(header, option &~ MACH_SEND_MSG,
			 0, rcv_size, rcv_name,
			 timeout, notify);
	}
	return r;
}

void
cproc_fork_prepare(void)
{
    register cproc_t p = cproc_self();

    vm_inherit(mach_task_self(),p->stack_base, p->stack_size, VM_INHERIT_COPY);
    spin_lock(&port_lock);
    spin_lock(&cproc_list_lock);
}

void
cproc_fork_parent(void)
{
    register cproc_t p = cproc_self();

    spin_unlock(&cproc_list_lock);
    spin_unlock(&port_lock);
    vm_inherit(mach_task_self(),p->stack_base, p->stack_size, VM_INHERIT_NONE);
}

void
cproc_fork_child(void)
{
    register cproc_t l,p = cproc_self();
    cproc_t m;
    register port_entry_t pe;
    port_entry_t pet;
    kern_return_t r;


    vm_inherit(mach_task_self(),p->stack_base, p->stack_size, VM_INHERIT_NONE);
    spin_lock_init(&n_kern_lock);
    cthread_kernel_threads=0;
#ifdef STATISTICS
    cthread_ready = 0;
    cthread_running = 1;
    cthread_waiting = 0;
    cthread_wired = 0;
    spin_lock_init(&wired_lock);
    cthread_wait_stacks = 0;
    cthread_waiters = 0;
    cthread_wakeup = 0;
    cthread_blocked = 0;
    cthread_rnone = 0;
    cthread_yields = 0;
    cthread_none = 0;
    cthread_switches = 0;
    cthread_no_mutex = 0;
    spin_lock_init(&mutex_count_lock);
#endif  /* STATISTICS */

    for(l=cproc_list;l!=NO_CPROC;l=m) {
	m=l->next;
	if (l!=p)
	    free(l);
    }

    cproc_list = p;
    p->next = NO_CPROC;
    spin_lock_init(&cproc_list_lock);
    cprocs_started = FALSE;
    cthread_queue_init(&ready);
    ready_count = 0;
    spin_lock_init(&ready_lock);

    MACH_CALL(mach_port_allocate(mach_task_self(),
				 MACH_PORT_RIGHT_RECEIVE,
				 &wait_port), r);
    MACH_CALL(mach_port_insert_right(mach_task_self(),
				     wait_port, wait_port,
				     MACH_MSG_TYPE_MAKE_SEND), r);
    wakeup_msg.msgh_remote_port = wait_port;
    wait_count = 0;
    cthread_queue_init(&waiters);
    spin_lock_init(&waiters_lock);
    for(pe=port_list;pe!=PORT_ENTRY_NULL;pe=pet) {
	pet = pe->next;
	free(pe);
    }
    port_list = PORT_ENTRY_NULL;
    spin_lock_init(&port_lock);

    if (p->wired) cthread_wire();
}