aboutsummaryrefslogtreecommitdiff
path: root/winsup/cygwin/posix_ipc.cc
blob: e82c38526966166c230b5c5af1bf4cafe3f80a64 (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
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
/* posix_ipc.cc: POSIX IPC API for Cygwin.

   Copyright 2007, 2008, 2009, 2010, 2011, 2012, 2014, 2015 Red Hat, Inc.

This file is part of Cygwin.

This software is a copyrighted work licensed under the terms of the
Cygwin license.  Please consult the file "CYGWIN_LICENSE" for
details. */

#include "winsup.h"
#include "shared_info.h"
#include "thread.h"
#include "path.h"
#include "cygtls.h"
#include "fhandler.h"
#include "dtable.h"
#include "cygheap.h"
#include "sigproc.h"
#include "ntdll.h"
#include <sys/mman.h>
#include <sys/param.h>
#include <stdlib.h>
#include <unistd.h>
#include <mqueue.h>
#include <semaphore.h>

/* The prefix_len is the length of the path prefix ncluding trailing "/"
   (or "/sem." for semaphores) as well as the trailing NUL. */
static struct
{
  const char *prefix;
  const size_t prefix_len;
  const char *description;
} ipc_names[] = {
  { "/dev/shm", 10, "POSIX shared memory object" },
  { "/dev/mqueue", 13, "POSIX message queue" },
  { "/dev/shm", 14, "POSIX semaphore" }
};

enum ipc_type_t
{
  shmem,
  mqueue,
  semaphore
};

static bool
check_path (char *res_name, ipc_type_t type, const char *name, size_t len)
{
  /* Note that we require the existance of the appropriate /dev subdirectories
     for POSIX IPC object support, similar to Linux (which supports the
     directories, but doesn't require to mount them).  We don't create
     these directory here, that's the task of the installer.  But we check
     for existance and give ample warning. */
  path_conv path (ipc_names[type].prefix, PC_SYM_NOFOLLOW);
  if (path.error || !path.exists () || !path.isdir ())
    {
      small_printf (
	"Warning: '%s' does not exists or is not a directory.\n\n"
	"%ss require the existance of this directory.\n"
	"Create the directory '%s' and set the permissions to 01777.\n"
	"For instance on the command line: mkdir -m 01777 %s\n",
	ipc_names[type].prefix, ipc_names[type].description,
	ipc_names[type].prefix, ipc_names[type].prefix);
      set_errno (EINVAL);
      return false;
    }
  /* Name must not be empty, or just be a single slash, or start with more
     than one slash.  Same for backslash.
     Apart from handling backslash like slash, the naming rules are identical
     to Linux, including the names and requirements for subdirectories, if
     the name contains further slashes. */
  if (!name || (strchr ("/\\", name[0])
		&& (!name[1] || strchr ("/\\", name[1]))))
    {
      debug_printf ("Invalid %s name '%s'", ipc_names[type].description, name);
      set_errno (EINVAL);
      return false;
    }
  /* Skip leading (back-)slash. */
  if (strchr ("/\\", name[0]))
    ++name;
  if (len > PATH_MAX - ipc_names[type].prefix_len)
    {
      debug_printf ("%s name '%s' too long", ipc_names[type].description, name);
      set_errno (ENAMETOOLONG);
      return false;
    }
  __small_sprintf (res_name, "%s/%s%s", ipc_names[type].prefix,
					type == semaphore ? "sem." : "",
					name);
  return true;
}

static int
ipc_mutex_init (HANDLE *pmtx, const char *name)
{
  WCHAR buf[MAX_PATH];
  UNICODE_STRING uname;
  OBJECT_ATTRIBUTES attr;
  NTSTATUS status;

  __small_swprintf (buf, L"mqueue/mtx_%s", name);
  RtlInitUnicodeString (&uname, buf);
  InitializeObjectAttributes (&attr, &uname,
			      OBJ_INHERIT | OBJ_OPENIF | OBJ_CASE_INSENSITIVE,
			      get_shared_parent_dir (),
			      everyone_sd (CYG_MUTANT_ACCESS));
  status = NtCreateMutant (pmtx, CYG_MUTANT_ACCESS, &attr, FALSE);
  if (!NT_SUCCESS (status))
    {
      debug_printf ("NtCreateMutant: %y", status);
      return geterrno_from_win_error (RtlNtStatusToDosError (status));
    }
  return 0;
}

static int
ipc_mutex_lock (HANDLE mtx, bool eintr)
{
  switch (cygwait (mtx, cw_infinite, cw_cancel | cw_cancel_self
				     | (eintr ? cw_sig_eintr : cw_sig_restart)))
    {
    case WAIT_OBJECT_0:
    case WAIT_ABANDONED_0:
      return 0;
    case WAIT_SIGNALED:
      set_errno (EINTR);
      return 1;
    default:
      break;
    }
  return geterrno_from_win_error ();
}

static inline int
ipc_mutex_unlock (HANDLE mtx)
{
  return ReleaseMutex (mtx) ? 0 : geterrno_from_win_error ();
}

static inline int
ipc_mutex_close (HANDLE mtx)
{
  return CloseHandle (mtx) ? 0 : geterrno_from_win_error ();
}

static int
ipc_cond_init (HANDLE *pevt, const char *name, char sr)
{
  WCHAR buf[MAX_PATH];
  UNICODE_STRING uname;
  OBJECT_ATTRIBUTES attr;
  NTSTATUS status;

  __small_swprintf (buf, L"mqueue/evt_%s%c", name, sr);
  RtlInitUnicodeString (&uname, buf);
  InitializeObjectAttributes (&attr, &uname,
			      OBJ_INHERIT | OBJ_OPENIF | OBJ_CASE_INSENSITIVE,
			      get_shared_parent_dir (),
			      everyone_sd (CYG_EVENT_ACCESS));
  status = NtCreateEvent (pevt, CYG_EVENT_ACCESS, &attr,
			  NotificationEvent, FALSE);
  if (!NT_SUCCESS (status))
    {
      debug_printf ("NtCreateEvent: %y", status);
      return geterrno_from_win_error (RtlNtStatusToDosError (status));
    }
  return 0;
}

static int
ipc_cond_timedwait (HANDLE evt, HANDLE mtx, const struct timespec *abstime)
{
  HANDLE w4[4] = { evt, };
  DWORD cnt = 2;
  DWORD timer_idx = 0;
  int ret = 0;

  wait_signal_arrived here (w4[1]);
  if ((w4[cnt] = pthread::get_cancel_event ()) != NULL)
    ++cnt;
  if (abstime)
    {
      if (abstime->tv_sec < 0
	       || abstime->tv_nsec < 0
	       || abstime->tv_nsec > 999999999)
	return EINVAL;

      /* If a timeout is set, we create a waitable timer to wait for.
	 This is the easiest way to handle the absolute timeout value, given
	 that NtSetTimer also takes absolute times and given the double
	 dependency on evt *and* mtx, which requires to call WFMO twice. */
      NTSTATUS status;
      LARGE_INTEGER duetime;

      timer_idx = cnt++;
      status = NtCreateTimer (&w4[timer_idx], TIMER_ALL_ACCESS, NULL,
			      NotificationTimer);
      if (!NT_SUCCESS (status))
	return geterrno_from_nt_status (status);
      timespec_to_filetime (abstime, &duetime);
      status = NtSetTimer (w4[timer_idx], &duetime, NULL, NULL, FALSE, 0, NULL);
      if (!NT_SUCCESS (status))
	{
	  NtClose (w4[timer_idx]);
	  return geterrno_from_nt_status (status);
	}
    }
  ResetEvent (evt);
  if ((ret = ipc_mutex_unlock (mtx)) != 0)
    return ret;
  /* Everything's set up, so now wait for the event to be signalled. */
restart1:
  switch (WaitForMultipleObjects (cnt, w4, FALSE, INFINITE))
    {
    case WAIT_OBJECT_0:
      break;
    case WAIT_OBJECT_0 + 1:
      if (_my_tls.call_signal_handler ())
	goto restart1;
      ret = EINTR;
      break;
    case WAIT_OBJECT_0 + 2:
      if (timer_idx != 2)
	pthread::static_cancel_self ();
      /*FALLTHRU*/
    case WAIT_OBJECT_0 + 3:
      ret = ETIMEDOUT;
      break;
    default:
      ret = geterrno_from_win_error ();
      break;
    }
  if (ret == 0)
    {
      /* At this point we need to lock the mutex.  The wait is practically
	 the same as before, just that we now wait on the mutex instead of the
	 event. */
    restart2:
      w4[0] = mtx;
      switch (WaitForMultipleObjects (cnt, w4, FALSE, INFINITE))
	{
	case WAIT_OBJECT_0:
	case WAIT_ABANDONED_0:
	  break;
	case WAIT_OBJECT_0 + 1:
	  if (_my_tls.call_signal_handler ())
	    goto restart2;
	  ret = EINTR;
	  break;
	case WAIT_OBJECT_0 + 2:
	  if (timer_idx != 2)
	    pthread_testcancel ();
	  /*FALLTHRU*/
	case WAIT_OBJECT_0 + 3:
	  ret = ETIMEDOUT;
	  break;
	default:
	  ret = geterrno_from_win_error ();
	  break;
	}
    }
  if (timer_idx)
    {
      if (ret != ETIMEDOUT)
	NtCancelTimer (w4[timer_idx], NULL);
      NtClose (w4[timer_idx]);
    }
  return ret;
}

static inline void
ipc_cond_signal (HANDLE evt)
{
  SetEvent (evt);
}

static inline void
ipc_cond_close (HANDLE evt)
{
  CloseHandle (evt);
}

class ipc_flock
{
  struct flock fl;

public:
  ipc_flock () { memset (&fl, 0, sizeof fl); }

  int lock (int fd, size_t size)
  {
    fl.l_type = F_WRLCK;
    fl.l_whence = SEEK_SET;
    fl.l_start = 0;
    fl.l_len = size;
    return fcntl64 (fd, F_SETLKW, &fl);
  }
  int unlock (int fd)
  {
    if (!fl.l_len)
      return 0;
    fl.l_type = F_UNLCK;
    return fcntl64 (fd, F_SETLKW, &fl);
  }
};

/* POSIX shared memory object implementation. */

extern "C" int
shm_open (const char *name, int oflag, mode_t mode)
{
  size_t len = strlen (name);
  char shmname[ipc_names[shmem].prefix_len + len];

  if (!check_path (shmname, shmem, name, len))
    return -1;

  /* Check for valid flags. */
  if (((oflag & O_ACCMODE) != O_RDONLY && (oflag & O_ACCMODE) != O_RDWR)
      || (oflag & ~(O_ACCMODE | O_CREAT | O_EXCL | O_TRUNC)))
    {
      debug_printf ("Invalid oflag 0%o", oflag);
      set_errno (EINVAL);
      return -1;
    }

  return open (shmname, oflag | O_CLOEXEC, mode & 0777);
}

extern "C" int
shm_unlink (const char *name)
{
  size_t len = strlen (name);
  char shmname[ipc_names[shmem].prefix_len + len];

  if (!check_path (shmname, shmem, name, len))
    return -1;

  return unlink (shmname);
}

/* The POSIX message queue implementation is based on W. Richard STEVENS
   implementation, just tweaked for Cygwin.  The main change is
   the usage of Windows mutexes and events instead of using the pthread
   synchronization objects.  The pathname is massaged so that the
   files are created under /dev/mqueue.  mq_timedsend and mq_timedreceive
   are implemented additionally. */

/* The mq_attr structure is defined using long datatypes per POSIX.
   For interoperability reasons between 32 and 64 bit processes, we have
   to make sure to use a unified structure layout in the message queue file.
   That's what the mq_fattr is, the in-file representation of the mq_attr
   struct. */
#pragma pack (push, 4)
struct mq_fattr
{
  uint32_t mq_flags;
  uint32_t mq_maxmsg;
  uint32_t mq_msgsize;
  uint32_t mq_curmsgs;
};

struct mq_hdr
{
  struct mq_fattr mqh_attr;	 /* the queue's attributes */
  int32_t         mqh_head;	 /* index of first message */
  int32_t         mqh_free;	 /* index of first free message */
  int32_t         mqh_nwait;	 /* #threads blocked in mq_receive() */
  pid_t           mqh_pid;	 /* nonzero PID if mqh_event set */
  char            mqh_uname[36]; /* unique name used to identify synchronization
				    objects connected to this queue */
  union {
    struct sigevent mqh_event;	 /* for mq_notify() */
    /* Make sure sigevent takes the same space on 32 and 64 bit systems.
       Other than that, it doesn't need to be compatible since only
       one process can be notified at a time. */
    uint64_t        mqh_placeholder[8];
  };
  uint32_t        mqh_magic;	/* Expect MQI_MAGIC here, otherwise it's
				   an old-style message queue. */
};

struct msg_hdr
{
  int32_t         msg_next;	 /* index of next on linked list */
  int32_t         msg_len;	 /* actual length */
  unsigned int    msg_prio;	 /* priority */
};
#pragma pack (pop)

struct mq_info
{
  struct mq_hdr  *mqi_hdr;	 /* start of mmap'ed region */
  uint32_t        mqi_magic;	 /* magic number if open */
  int             mqi_flags;	 /* flags for this process */
  HANDLE          mqi_lock;	 /* mutex lock */
  HANDLE          mqi_waitsend;	 /* and condition variable for full queue */
  HANDLE          mqi_waitrecv;	 /* and condition variable for empty queue */
};

#define MQI_MAGIC	0x98765432UL

#define MSGSIZE(i)	roundup((i), sizeof(long))

#define	 MAX_TRIES	10	/* for waiting for initialization */

struct mq_attr defattr = { 0, 10, 8192, 0 };	/* Linux defaults. */

extern "C" off_t lseek64 (int, off_t, int);
extern "C" void *mmap64 (void *, size_t, int, int, int, off_t);

extern "C" mqd_t
mq_open (const char *name, int oflag, ...)
{
  int i, fd = -1, nonblock, created = 0;
  long msgsize, index;
  off_t filesize = 0;
  va_list ap;
  mode_t mode;
  int8_t *mptr = (int8_t *) MAP_FAILED;
  struct stat statbuff;
  struct mq_hdr *mqhdr;
  struct msg_hdr *msghdr;
  struct mq_attr *attr;
  struct mq_info *mqinfo = NULL;
  LUID luid;

  size_t len = strlen (name);
  char mqname[ipc_names[mqueue].prefix_len + len];

  if (!check_path (mqname, mqueue, name, len))
    return (mqd_t) -1;

  __try
    {
      oflag &= (O_CREAT | O_EXCL | O_NONBLOCK);
      nonblock = oflag & O_NONBLOCK;
      oflag &= ~O_NONBLOCK;

    again:
      if (oflag & O_CREAT)
	{
	  va_start (ap, oflag);		/* init ap to final named argument */
	  mode = va_arg (ap, mode_t) & ~S_IXUSR;
	  attr = va_arg (ap, struct mq_attr *);
	  va_end (ap);

	  /* Open and specify O_EXCL and user-execute */
	  fd = open (mqname, oflag | O_EXCL | O_RDWR | O_CLOEXEC,
		     mode | S_IXUSR);
	  if (fd < 0)
	    {
	      if (errno == EEXIST && (oflag & O_EXCL) == 0)
		goto exists;		/* already exists, OK */
	      return (mqd_t) -1;
	    }
	  created = 1;
	  /* First one to create the file initializes it */
	  if (attr == NULL)
	    attr = &defattr;
	  /* Check minimum and maximum values.  The max values are pretty much
	     arbitrary, taken from the linux mq_overview man page.  However,
	     these max values make sure that the internal mq_fattr structure
	     can use 32 bit types. */
	  else if (attr->mq_maxmsg <= 0 || attr->mq_maxmsg > 32768
		   || attr->mq_msgsize <= 0 || attr->mq_msgsize > 1048576)
	    {
	      set_errno (EINVAL);
	      __leave;
	    }
	  /* Calculate and set the file size */
	  msgsize = MSGSIZE (attr->mq_msgsize);
	  filesize = sizeof (struct mq_hdr)
		     + (attr->mq_maxmsg * (sizeof (struct msg_hdr) + msgsize));
	  if (lseek64 (fd, filesize - 1, SEEK_SET) == -1)
	    __leave;
	  if (write (fd, "", 1) == -1)
	    __leave;

	  /* Memory map the file */
	  mptr = (int8_t *) mmap64 (NULL, (size_t) filesize,
				    PROT_READ | PROT_WRITE,
				    MAP_SHARED, fd, 0);
	  if (mptr == (int8_t *) MAP_FAILED)
	    __leave;

	  /* Allocate one mq_info{} for the queue */
	  if (!(mqinfo = (struct mq_info *)
			 calloc (1, sizeof (struct mq_info))))
	    __leave;
	  mqinfo->mqi_hdr = mqhdr = (struct mq_hdr *) mptr;
	  mqinfo->mqi_magic = MQI_MAGIC;
	  mqinfo->mqi_flags = nonblock;

	  /* Initialize header at beginning of file */
	  /* Create free list with all messages on it */
	  mqhdr->mqh_attr.mq_flags = 0;
	  mqhdr->mqh_attr.mq_maxmsg = attr->mq_maxmsg;
	  mqhdr->mqh_attr.mq_msgsize = attr->mq_msgsize;
	  mqhdr->mqh_attr.mq_curmsgs = 0;
	  mqhdr->mqh_nwait = 0;
	  mqhdr->mqh_pid = 0;
	  NtAllocateLocallyUniqueId (&luid);
	  __small_sprintf (mqhdr->mqh_uname, "%016X%08x%08x",
			   hash_path_name (0,mqname),
			   luid.HighPart, luid.LowPart);
	  mqhdr->mqh_head = 0;
	  mqhdr->mqh_magic = MQI_MAGIC;
	  index = sizeof (struct mq_hdr);
	  mqhdr->mqh_free = index;
	  for (i = 0; i < attr->mq_maxmsg - 1; i++)
	    {
	      msghdr = (struct msg_hdr *) &mptr[index];
	      index += sizeof (struct msg_hdr) + msgsize;
	      msghdr->msg_next = index;
	    }
	  msghdr = (struct msg_hdr *) &mptr[index];
	  msghdr->msg_next = 0;		/* end of free list */

	  /* Initialize mutex & condition variables */
	  i = ipc_mutex_init (&mqinfo->mqi_lock, mqhdr->mqh_uname);
	  if (i != 0)
	    {
	      set_errno (i);
	      __leave;
	    }
	  i = ipc_cond_init (&mqinfo->mqi_waitsend, mqhdr->mqh_uname, 'S');
	  if (i != 0)
	    {
	      set_errno (i);
	      __leave;
	    }
	  i = ipc_cond_init (&mqinfo->mqi_waitrecv, mqhdr->mqh_uname, 'R');
	  if (i != 0)
	    {
	      set_errno (i);
	      __leave;
	    }
	  /* Initialization complete, turn off user-execute bit */
	  if (fchmod (fd, mode) == -1)
	    __leave;
	  close (fd);
	  return ((mqd_t) mqinfo);
	}

    exists:
      /* Open the file then memory map */
      if ((fd = open (mqname, O_RDWR | O_CLOEXEC)) < 0)
	{
	  if (errno == ENOENT && (oflag & O_CREAT))
	    goto again;
	  __leave;
	}
      /* Make certain initialization is complete */
      for (i = 0; i < MAX_TRIES; i++)
	{
	  if (stat64 (mqname, &statbuff) == -1)
	    {
	      if (errno == ENOENT && (oflag & O_CREAT))
		{
		  close (fd);
		  fd = -1;
		  goto again;
		}
	      __leave;
	    }
	  if ((statbuff.st_mode & S_IXUSR) == 0)
	    break;
	  sleep (1);
	}
      if (i == MAX_TRIES)
	{
	  set_errno (ETIMEDOUT);
	  __leave;
	}

      filesize = statbuff.st_size;
      mptr = (int8_t *) mmap64 (NULL, (size_t) filesize, PROT_READ | PROT_WRITE,
				MAP_SHARED, fd, 0);
      if (mptr == (int8_t *) MAP_FAILED)
	__leave;
      close (fd);
      fd = -1;

      /* Allocate one mq_info{} for each open */
      if (!(mqinfo = (struct mq_info *) calloc (1, sizeof (struct mq_info))))
	__leave;
      mqinfo->mqi_hdr = mqhdr = (struct mq_hdr *) mptr;
      if (mqhdr->mqh_magic != MQI_MAGIC)
	{
	  system_printf (
    "Old message queue \"%s\" detected!\n"
    "This file is not usable as message queue anymore due to changes in the "
    "internal file layout.  Please remove the file and try again.", mqname);
	  set_errno (EACCES);
	  __leave;
	}
      mqinfo->mqi_magic = MQI_MAGIC;
      mqinfo->mqi_flags = nonblock;

      /* Initialize mutex & condition variable */
      i = ipc_mutex_init (&mqinfo->mqi_lock, mqhdr->mqh_uname);
      if (i != 0)
	{
	  set_errno (i);
	  __leave;
	}
      i = ipc_cond_init (&mqinfo->mqi_waitsend, mqhdr->mqh_uname, 'S');
      if (i != 0)
	{
	  set_errno (i);
	  __leave;
	}
      i = ipc_cond_init (&mqinfo->mqi_waitrecv, mqhdr->mqh_uname, 'R');
      if (i != 0)
	{
	  set_errno (i);
	  __leave;
	}
      return (mqd_t) mqinfo;
    }
  __except (EFAULT) {}
  __endtry
  /* Don't let following function calls change errno */
  save_errno save;
  if (created)
    unlink (mqname);
  if (mptr != (int8_t *) MAP_FAILED)
    munmap((void *) mptr, (size_t) filesize);
  if (mqinfo)
    {
      if (mqinfo->mqi_lock)
	ipc_mutex_close (mqinfo->mqi_lock);
      if (mqinfo->mqi_waitsend)
	ipc_cond_close (mqinfo->mqi_waitsend);
      if (mqinfo->mqi_waitrecv)
	ipc_cond_close (mqinfo->mqi_waitrecv);
      free (mqinfo);
    }
  if (fd >= 0)
    close (fd);
  return (mqd_t) -1;
}

extern "C" int
mq_getattr (mqd_t mqd, struct mq_attr *mqstat)
{
  int n;
  struct mq_hdr *mqhdr;
  struct mq_fattr *attr;
  struct mq_info *mqinfo;

  __try
    {
      mqinfo = (struct mq_info *) mqd;
      if (mqinfo->mqi_magic != MQI_MAGIC)
	{
	  set_errno (EBADF);
	  __leave;
	}
      mqhdr = mqinfo->mqi_hdr;
      attr = &mqhdr->mqh_attr;
      if ((n = ipc_mutex_lock (mqinfo->mqi_lock, false)) != 0)
	{
	  errno = n;
	  __leave;
	}
      mqstat->mq_flags = mqinfo->mqi_flags;   /* per-open */
      mqstat->mq_maxmsg = attr->mq_maxmsg;    /* remaining three per-queue */
      mqstat->mq_msgsize = attr->mq_msgsize;
      mqstat->mq_curmsgs = attr->mq_curmsgs;

      ipc_mutex_unlock (mqinfo->mqi_lock);
      return 0;
    }
  __except (EBADF) {}
  __endtry
  return -1;
}

extern "C" int
mq_setattr (mqd_t mqd, const struct mq_attr *mqstat, struct mq_attr *omqstat)
{
  int n;
  struct mq_hdr *mqhdr;
  struct mq_fattr *attr;
  struct mq_info *mqinfo;

  __try
    {
      mqinfo = (struct mq_info *) mqd;
      if (mqinfo->mqi_magic != MQI_MAGIC)
	{
	  set_errno (EBADF);
	  __leave;
	}
      mqhdr = mqinfo->mqi_hdr;
      attr = &mqhdr->mqh_attr;
      if ((n = ipc_mutex_lock (mqinfo->mqi_lock, false)) != 0)
	{
	  errno = n;
	  __leave;
	}

      if (omqstat != NULL)
	{
	  omqstat->mq_flags = mqinfo->mqi_flags;  /* previous attributes */
	  omqstat->mq_maxmsg = attr->mq_maxmsg;
	  omqstat->mq_msgsize = attr->mq_msgsize;
	  omqstat->mq_curmsgs = attr->mq_curmsgs; /* and current status */
	}

      if (mqstat->mq_flags & O_NONBLOCK)
	mqinfo->mqi_flags |= O_NONBLOCK;
      else
	mqinfo->mqi_flags &= ~O_NONBLOCK;

      ipc_mutex_unlock (mqinfo->mqi_lock);
      return 0;
    }
  __except (EBADF) {}
  __endtry
  return -1;
}

extern "C" int
mq_notify (mqd_t mqd, const struct sigevent *notification)
{
  int n;
  pid_t pid;
  struct mq_hdr *mqhdr;
  struct mq_info *mqinfo;

  __try
    {
      mqinfo = (struct mq_info *) mqd;
      if (mqinfo->mqi_magic != MQI_MAGIC)
	{
	  set_errno (EBADF);
	  __leave;
	}
      mqhdr = mqinfo->mqi_hdr;
      if ((n = ipc_mutex_lock (mqinfo->mqi_lock, false)) != 0)
	{
	  errno = n;
	  __leave;
	}

      pid = getpid ();
      if (!notification)
	{
	  if (mqhdr->mqh_pid == pid)
	      mqhdr->mqh_pid = 0;     /* unregister calling process */
	}
      else
	{
	  if (mqhdr->mqh_pid != 0)
	    {
	      if (kill (mqhdr->mqh_pid, 0) != -1 || errno != ESRCH)
		{
		  set_errno (EBUSY);
		  ipc_mutex_unlock (mqinfo->mqi_lock);
		  __leave;
		}
	    }
	  mqhdr->mqh_pid = pid;
	  mqhdr->mqh_event = *notification;
	}
      ipc_mutex_unlock (mqinfo->mqi_lock);
      return 0;
    }
  __except (EBADF) {}
  __endtry
  return -1;
}

static int
_mq_send (mqd_t mqd, const char *ptr, size_t len, unsigned int prio,
	  const struct timespec *abstime)
{
  int n;
  long index, freeindex;
  int8_t *mptr;
  struct sigevent *sigev;
  struct mq_hdr *mqhdr;
  struct mq_fattr *attr;
  struct msg_hdr *msghdr, *nmsghdr, *pmsghdr;
  struct mq_info *mqinfo = NULL;
  bool ipc_mutex_locked = false;
  int ret = -1;

  pthread_testcancel ();

  __try
    {
      mqinfo = (struct mq_info *) mqd;
      if (mqinfo->mqi_magic != MQI_MAGIC)
	{
	  set_errno (EBADF);
	  __leave;
	}
      if (prio > MQ_PRIO_MAX)
	{
	  set_errno (EINVAL);
	  __leave;
	}

      mqhdr = mqinfo->mqi_hdr;        /* struct pointer */
      mptr = (int8_t *) mqhdr;        /* byte pointer */
      attr = &mqhdr->mqh_attr;
      if ((n = ipc_mutex_lock (mqinfo->mqi_lock, true)) != 0)
	{
	  errno = n;
	  __leave;
	}
      ipc_mutex_locked = true;
      if (len > (size_t) attr->mq_msgsize)
	{
	  set_errno (EMSGSIZE);
	  __leave;
	}
      if (attr->mq_curmsgs == 0)
	{
	  if (mqhdr->mqh_pid != 0 && mqhdr->mqh_nwait == 0)
	    {
	      sigev = &mqhdr->mqh_event;
	      if (sigev->sigev_notify == SIGEV_SIGNAL)
		sigqueue (mqhdr->mqh_pid, sigev->sigev_signo,
			  sigev->sigev_value);
	      mqhdr->mqh_pid = 0;             /* unregister */
	    }
	}
      else if (attr->mq_curmsgs >= attr->mq_maxmsg)
	{
	  /* Queue is full */
	  if (mqinfo->mqi_flags & O_NONBLOCK)
	    {
	      set_errno (EAGAIN);
	      __leave;
	    }
	  /* Wait for room for one message on the queue */
	  while (attr->mq_curmsgs >= attr->mq_maxmsg)
	    {
	      int ret = ipc_cond_timedwait (mqinfo->mqi_waitsend,
					    mqinfo->mqi_lock, abstime);
	      if (ret != 0)
		{
		  set_errno (ret);
		  __leave;
		}
	    }
	}

      /* nmsghdr will point to new message */
      if ((freeindex = mqhdr->mqh_free) == 0)
	api_fatal ("mq_send: curmsgs = %ld; free = 0", attr->mq_curmsgs);

      nmsghdr = (struct msg_hdr *) &mptr[freeindex];
      nmsghdr->msg_prio = prio;
      nmsghdr->msg_len = len;
      memcpy (nmsghdr + 1, ptr, len);         /* copy message from caller */
      mqhdr->mqh_free = nmsghdr->msg_next;    /* new freelist head */

      /* Find right place for message in linked list */
      index = mqhdr->mqh_head;
      pmsghdr = (struct msg_hdr *) &(mqhdr->mqh_head);
      while (index)
	{
	  msghdr = (struct msg_hdr *) &mptr[index];
	  if (prio > msghdr->msg_prio)
	    {
	      nmsghdr->msg_next = index;
	      pmsghdr->msg_next = freeindex;
	      break;
	    }
	  index = msghdr->msg_next;
	  pmsghdr = msghdr;
	}
      if (index == 0)
	{
	  /* Queue was empty or new goes at end of list */
	  pmsghdr->msg_next = freeindex;
	  nmsghdr->msg_next = 0;
	}
      /* Wake up anyone blocked in mq_receive waiting for a message */
      if (attr->mq_curmsgs == 0)
	ipc_cond_signal (mqinfo->mqi_waitrecv);
      attr->mq_curmsgs++;

      ipc_mutex_unlock (mqinfo->mqi_lock);
      ret = 0;
    }
  __except (EBADF) {}
  __endtry
  if (ipc_mutex_locked)
    ipc_mutex_unlock (mqinfo->mqi_lock);
  return ret;
}

extern "C" int
mq_send (mqd_t mqd, const char *ptr, size_t len, unsigned int prio)
{
  return _mq_send (mqd, ptr, len, prio, NULL);
}

extern "C" int
mq_timedsend (mqd_t mqd, const char *ptr, size_t len, unsigned int prio,
	      const struct timespec *abstime)
{
  return _mq_send (mqd, ptr, len, prio, abstime);
}

static ssize_t
_mq_receive (mqd_t mqd, char *ptr, size_t maxlen, unsigned int *priop,
	     const struct timespec *abstime)
{
  int n;
  long index;
  int8_t *mptr;
  ssize_t len = -1;
  struct mq_hdr *mqhdr;
  struct mq_fattr *attr;
  struct msg_hdr *msghdr;
  struct mq_info *mqinfo = (struct mq_info *) mqd;
  bool ipc_mutex_locked = false;

  pthread_testcancel ();

  __try
    {
      if (mqinfo->mqi_magic != MQI_MAGIC)
	{
	  set_errno (EBADF);
	  __leave;
	}
      mqhdr = mqinfo->mqi_hdr;        /* struct pointer */
      mptr = (int8_t *) mqhdr;        /* byte pointer */
      attr = &mqhdr->mqh_attr;
      if ((n = ipc_mutex_lock (mqinfo->mqi_lock, true)) != 0)
	{
	  errno = n;
	  __leave;
	}
      ipc_mutex_locked = true;
      if (maxlen < (size_t) attr->mq_msgsize)
	{
	  set_errno (EMSGSIZE);
	  __leave;
	}
      if (attr->mq_curmsgs == 0)	/* queue is empty */
	{
	  if (mqinfo->mqi_flags & O_NONBLOCK)
	    {
	      set_errno (EAGAIN);
	      __leave;
	    }
	  /* Wait for a message to be placed onto queue */
	  mqhdr->mqh_nwait++;
	  while (attr->mq_curmsgs == 0)
	    {
	      int ret = ipc_cond_timedwait (mqinfo->mqi_waitrecv,
					    mqinfo->mqi_lock, abstime);
	      if (ret != 0)
		{
		  set_errno (ret);
		  __leave;
		}
	    }
	  mqhdr->mqh_nwait--;
	}

      if ((index = mqhdr->mqh_head) == 0)
	api_fatal ("mq_receive: curmsgs = %ld; head = 0", attr->mq_curmsgs);

      msghdr = (struct msg_hdr *) &mptr[index];
      mqhdr->mqh_head = msghdr->msg_next;     /* new head of list */
      len = msghdr->msg_len;
      memcpy(ptr, msghdr + 1, len);           /* copy the message itself */
      if (priop != NULL)
	*priop = msghdr->msg_prio;

      /* Just-read message goes to front of free list */
      msghdr->msg_next = mqhdr->mqh_free;
      mqhdr->mqh_free = index;

      /* Wake up anyone blocked in mq_send waiting for room */
      if (attr->mq_curmsgs == attr->mq_maxmsg)
	ipc_cond_signal (mqinfo->mqi_waitsend);
      attr->mq_curmsgs--;

      ipc_mutex_unlock (mqinfo->mqi_lock);
    }
  __except (EBADF) {}
  __endtry
  if (ipc_mutex_locked)
    ipc_mutex_unlock (mqinfo->mqi_lock);
  return len;
}

extern "C" ssize_t
mq_receive (mqd_t mqd, char *ptr, size_t maxlen, unsigned int *priop)
{
  return _mq_receive (mqd, ptr, maxlen, priop, NULL);
}

extern "C" ssize_t
mq_timedreceive (mqd_t mqd, char *ptr, size_t maxlen, unsigned int *priop,
		 const struct timespec *abstime)
{
  return _mq_receive (mqd, ptr, maxlen, priop, abstime);
}

extern "C" int
mq_close (mqd_t mqd)
{
  long msgsize, filesize;
  struct mq_hdr *mqhdr;
  struct mq_fattr *attr;
  struct mq_info *mqinfo;

  __try
    {
      mqinfo = (struct mq_info *) mqd;
      if (mqinfo->mqi_magic != MQI_MAGIC)
	{
	  set_errno (EBADF);
	  __leave;
	}
      mqhdr = mqinfo->mqi_hdr;
      attr = &mqhdr->mqh_attr;

      if (mq_notify (mqd, NULL))	/* unregister calling process */
	__leave;

      msgsize = MSGSIZE (attr->mq_msgsize);
      filesize = sizeof (struct mq_hdr)
		 + (attr->mq_maxmsg * (sizeof (struct msg_hdr) + msgsize));
      if (munmap (mqinfo->mqi_hdr, filesize) == -1)
	__leave;

      mqinfo->mqi_magic = 0;          /* just in case */
      ipc_cond_close (mqinfo->mqi_waitsend);
      ipc_cond_close (mqinfo->mqi_waitrecv);
      ipc_mutex_close (mqinfo->mqi_lock);
      free (mqinfo);
      return 0;
    }
  __except (EBADF) {}
  __endtry
  return -1;
}

extern "C" int
mq_unlink (const char *name)
{
  size_t len = strlen (name);
  char mqname[ipc_names[mqueue].prefix_len + len];

  if (!check_path (mqname, mqueue, name, len))
    return -1;
  if (unlink (mqname) == -1)
    return -1;
  return 0;
}

/* POSIX named semaphore implementation.  Loosely based on Richard W. STEPHENS
   implementation as far as sem_open is concerned, but under the hood using
   the already existing semaphore class in thread.cc.  Using a file backed
   solution allows to implement kernel persistent named semaphores.  */

struct sem_finfo
{
  unsigned int       value;
  unsigned long long hash;
  LUID               luid;
};

extern "C" sem_t *
sem_open (const char *name, int oflag, ...)
{
  int i, fd = -1, created = 0;
  va_list ap;
  mode_t mode = 0;
  unsigned int value = 0;
  struct stat statbuff;
  sem_t *sem = SEM_FAILED;
  sem_finfo sf;
  bool wasopen = false;
  ipc_flock file;

  size_t len = strlen (name);
  char semname[ipc_names[semaphore].prefix_len + len];

  if (!check_path (semname, semaphore, name, len))
    return SEM_FAILED;

  __try
    {
      oflag &= (O_CREAT | O_EXCL);

    again:
      if (oflag & O_CREAT)
	{
	  va_start (ap, oflag);		/* init ap to final named argument */
	  mode = va_arg (ap, mode_t) & ~S_IXUSR;
	  value = va_arg (ap, unsigned int);
	  va_end (ap);

	  /* Open and specify O_EXCL and user-execute */
	  fd = open (semname, oflag | O_EXCL | O_RDWR | O_CLOEXEC,
		     mode | S_IXUSR);
	  if (fd < 0)
	    {
	      if (errno == EEXIST && (oflag & O_EXCL) == 0)
		goto exists;		/* already exists, OK */
	      return SEM_FAILED;
	    }
	  created = 1;
	  /* First one to create the file initializes it. */
	  NtAllocateLocallyUniqueId (&sf.luid);
	  sf.value = value;
	  sf.hash = hash_path_name (0, semname);
	  if (write (fd, &sf, sizeof sf) != sizeof sf)
	    __leave;
	  sem = semaphore::open (sf.hash, sf.luid, fd, oflag, mode, value,
				 wasopen);
	  if (sem == SEM_FAILED)
	    __leave;
	  /* Initialization complete, turn off user-execute bit */
	  if (fchmod (fd, mode) == -1)
	    __leave;
	  /* Don't close (fd); */
	  return sem;
	}

    exists:
      /* Open the file and fetch the semaphore name. */
      if ((fd = open (semname, O_RDWR | O_CLOEXEC)) < 0)
	{
	  if (errno == ENOENT && (oflag & O_CREAT))
	    goto again;
	  __leave;
	}
      /* Make certain initialization is complete */
      for (i = 0; i < MAX_TRIES; i++)
	{
	  if (stat64 (semname, &statbuff) == -1)
	    {
	      if (errno == ENOENT && (oflag & O_CREAT))
		{
		  close (fd);
		  fd = -1;
		  goto again;
		}
	      __leave;
	    }
	  if ((statbuff.st_mode & S_IXUSR) == 0)
	    break;
	  sleep (1);
	}
      if (i == MAX_TRIES)
	{
	  set_errno (ETIMEDOUT);
	  __leave;
	}
      if (file.lock (fd, sizeof sf))
	__leave;
      if (read (fd, &sf, sizeof sf) != sizeof sf)
	__leave;
      sem = semaphore::open (sf.hash, sf.luid, fd, oflag, mode, sf.value,
			     wasopen);
      file.unlock (fd);
      if (sem == SEM_FAILED)
	__leave;
      /* If wasopen is set, the semaphore was already opened and we already have
	 an open file descriptor pointing to the file.  This means, we have to
	 close the file descriptor created in this call.  It won't be stored
	 anywhere anyway. */
      if (wasopen)
	close (fd);
      return sem;
    }
  __except (EFAULT) {}
  __endtry
  /* Don't let following function calls change errno */
  save_errno save;

  if (fd >= 0)
    file.unlock (fd);
  if (created)
    unlink (semname);
  if (sem != SEM_FAILED)
    semaphore::close (sem);
  if (fd >= 0)
    close (fd);
  return SEM_FAILED;
}

int
_sem_close (sem_t *sem, bool do_close)
{
  sem_finfo sf;
  int fd, ret = -1;
  ipc_flock file;

  if (semaphore::getinternal (sem, &fd, &sf.hash, &sf.luid, &sf.value) == -1)
    return -1;
  if (!file.lock (fd, sizeof sf)
      && lseek64 (fd, 0LL, SEEK_SET) != (off_t) -1
      && write (fd, &sf, sizeof sf) == sizeof sf)
    ret = do_close ? semaphore::close (sem) : 0;

  /* Don't let following function calls change errno */
  save_errno save;
  file.unlock (fd);
  close (fd);

  return ret;
}

extern "C" int
sem_close (sem_t *sem)
{
  return _sem_close (sem, true);
}

extern "C" int
sem_unlink (const char *name)
{
  size_t len = strlen (name);
  char semname[ipc_names[semaphore].prefix_len + len];

  if (!check_path (semname, semaphore, name, len))
    return -1;
  if (unlink (semname) == -1)
    return -1;
  return 0;
}