-
Notifications
You must be signed in to change notification settings - Fork 1
/
drvrsmem.c
973 lines (791 loc) · 37.6 KB
/
drvrsmem.c
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
/* S H A R E D M E M O R Y D R I V E R
=======================================
09-Mar-98 : initial version 1.0 released
23-Mar-98 : shared_malloc now accepts new handle as an argument
23-Mar-98 : shmem://0, shmem://1, etc changed to shmem://h0, etc due to bug
in url parser.
10-Apr-98 : code cleanup
13-May-99 : delayed initialization added, global table deleted on exit when
no shmem segments remain, and last process terminates
*/
#ifdef HAVE_SHMEM_SERVICES
#include "fitsio2.h" /* drvrsmem.h is included by it */
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <errno.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>
#if defined(unix) || defined(__unix__) || defined(__unix) || defined(HAVE_UNISTD_H)
#include <unistd.h>
#endif
static int shared_kbase = 0; /* base for shared memory handles */
static int shared_maxseg = 0; /* max number of shared memory blocks */
static int shared_range = 0; /* max number of tried entries */
static int shared_fd = SHARED_INVALID; /* handle of global access lock file */
static int shared_gt_h = SHARED_INVALID; /* handle of global table segment */
static SHARED_LTAB *shared_lt = NULL; /* local table pointer */
static SHARED_GTAB *shared_gt = NULL; /* global table pointer */
static int shared_create_mode = 0666; /* permission flags for created objects */
static int shared_debug = 1; /* simple debugging tool, set to 0 to disable messages */
static int shared_init_called = 0; /* flag whether shared_init() has been called, used for delayed init */
/* static support routines prototypes */
static int shared_clear_entry(int idx); /* unconditionally clear entry */
static int shared_destroy_entry(int idx); /* unconditionally destroy sema & shseg and clear entry */
static int shared_mux(int idx, int mode); /* obtain exclusive access to specified segment */
static int shared_demux(int idx, int mode); /* free exclusive access to specified segment */
static int shared_process_count(int sem); /* valid only for time of invocation */
static int shared_delta_process(int sem, int delta); /* change number of processes hanging on segment */
static int shared_attach_process(int sem);
static int shared_detach_process(int sem);
static int shared_get_free_entry(int newhandle); /* get free entry in shared_key, or -1, entry is set rw locked */
static int shared_get_hash(long size, int idx);/* return hash value for malloc */
static long shared_adjust_size(long size); /* size must be >= 0 !!! */
static int shared_check_locked_index(int idx); /* verify that given idx is valid */
static int shared_map(int idx); /* map all tables for given idx, check for validity */
static int shared_validate(int idx, int mode); /* use intrnally inside crit.sect !!! */
/* support routines - initialization */
static int shared_clear_entry(int idx) /* unconditionally clear entry */
{ if ((idx < 0) || (idx >= shared_maxseg)) return(SHARED_BADARG);
shared_gt[idx].key = SHARED_INVALID; /* clear entries in global table */
shared_gt[idx].handle = SHARED_INVALID;
shared_gt[idx].sem = SHARED_INVALID;
shared_gt[idx].semkey = SHARED_INVALID;
shared_gt[idx].nprocdebug = 0;
shared_gt[idx].size = 0;
shared_gt[idx].attr = 0;
return(SHARED_OK);
}
static int shared_destroy_entry(int idx) /* unconditionally destroy sema & shseg and clear entry */
{ int r, r2;
union semun filler;
if ((idx < 0) || (idx >= shared_maxseg)) return(SHARED_BADARG);
r2 = r = SHARED_OK;
filler.val = 0; /* this is to make cc happy (warning otherwise) */
if (SHARED_INVALID != shared_gt[idx].sem) r = semctl(shared_gt[idx].sem, 0, IPC_RMID, filler); /* destroy semaphore */
if (SHARED_INVALID != shared_gt[idx].handle) r2 = shmctl(shared_gt[idx].handle, IPC_RMID, 0); /* destroy shared memory segment */
if (SHARED_OK == r) r = r2; /* accumulate error code in r, free r2 */
r2 = shared_clear_entry(idx);
return((SHARED_OK == r) ? r2 : r);
}
void shared_cleanup(void) /* this must (should) be called during exit/abort */
{ int i, j, r, oktodelete, filelocked, segmentspresent;
flock_t flk;
struct shmid_ds ds;
if (shared_debug) printf("shared_cleanup:");
if (NULL != shared_lt)
{ if (shared_debug) printf(" deleting segments:");
for (i=0; i<shared_maxseg; i++)
{ if (0 == shared_lt[i].tcnt) continue; /* we're not using this segment, skip this ... */
if (-1 != shared_lt[i].lkcnt) continue; /* seg not R/W locked by us, skip this ... */
r = shared_destroy_entry(i); /* destroy unconditionally sema & segment */
if (shared_debug)
{ if (SHARED_OK == r) printf(" [%d]", i);
else printf(" [error on %d !!!!]", i);
}
}
free((void *)shared_lt); /* free local table */
shared_lt = NULL;
}
if (NULL != shared_gt) /* detach global index table */
{ oktodelete = 0;
filelocked = 0;
if (shared_debug) printf(" detaching globalsharedtable");
if (SHARED_INVALID != shared_fd)
flk.l_type = F_WRLCK; /* lock whole lock file */
flk.l_whence = 0;
flk.l_start = 0;
flk.l_len = shared_maxseg;
if (-1 != fcntl(shared_fd, F_SETLK, &flk))
{ filelocked = 1; /* success, scan global table, to see if there are any segs */
segmentspresent = 0; /* assume, there are no segs in the system */
for (j=0; j<shared_maxseg; j++)
{ if (SHARED_INVALID != shared_gt[j].key)
{ segmentspresent = 1; /* yes, there is at least one */
break;
}
}
if (0 == segmentspresent) /* if there are no segs ... */
if (0 == shmctl(shared_gt_h, IPC_STAT, &ds)) /* get number of processes attached to table */
{ if (ds.shm_nattch <= 1) oktodelete = 1; /* if only one (we), then it is safe (but see text 4 lines later) to unlink */
}
}
shmdt((char *)shared_gt); /* detach global table */
if (oktodelete) /* delete global table from system, if no shm seg present */
{ shmctl(shared_gt_h, IPC_RMID, 0); /* there is a race condition here - time window between shmdt and shmctl */
shared_gt_h = SHARED_INVALID;
}
shared_gt = NULL;
if (filelocked) /* if we locked, we need to unlock */
{ flk.l_type = F_UNLCK;
flk.l_whence = 0;
flk.l_start = 0;
flk.l_len = shared_maxseg;
fcntl(shared_fd, F_SETLK, &flk);
}
}
shared_gt_h = SHARED_INVALID;
if (SHARED_INVALID != shared_fd) /* close lock file */
{ if (shared_debug) printf(" closing lockfile");
close(shared_fd);
shared_fd = SHARED_INVALID;
}
shared_kbase = 0;
shared_maxseg = 0;
shared_range = 0;
shared_init_called = 0;
if (shared_debug) printf(" <<done>>\n");
return;
}
int shared_init(int debug_msgs) /* initialize shared memory stuff, you have to call this routine once */
{ int i;
char buf[1000], *p;
mode_t oldumask;
shared_init_called = 1; /* tell everybody no need to call us for the 2nd time */
shared_debug = debug_msgs; /* set required debug mode */
if (shared_debug) printf("shared_init:");
shared_kbase = 0; /* adapt to current env. settings */
if (NULL != (p = getenv(SHARED_ENV_KEYBASE))) shared_kbase = atoi(p);
if (0 == shared_kbase) shared_kbase = SHARED_KEYBASE;
if (shared_debug) printf(" keybase=%d", shared_kbase);
shared_maxseg = 0;
if (NULL != (p = getenv(SHARED_ENV_MAXSEG))) shared_maxseg = atoi(p);
if (0 == shared_maxseg) shared_maxseg = SHARED_MAXSEG;
if (shared_debug) printf(" maxseg=%d", shared_maxseg);
shared_range = 3 * shared_maxseg;
if (SHARED_INVALID == shared_fd) /* create rw locking file (this file is never deleted) */
{ if (shared_debug) printf(" lockfileinit=");
snprintf(buf, 1000,"%s.%d.%d", SHARED_FDNAME, shared_kbase, shared_maxseg);
oldumask = umask(0);
shared_fd = open(buf, O_TRUNC | O_EXCL | O_CREAT | O_RDWR, shared_create_mode);
umask(oldumask);
if (SHARED_INVALID == shared_fd) /* or just open rw locking file, in case it already exists */
{ shared_fd = open(buf, O_TRUNC | O_RDWR, shared_create_mode);
if (SHARED_INVALID == shared_fd) return(SHARED_NOFILE);
if (shared_debug) printf("slave");
}
else
{ if (shared_debug) printf("master");
}
}
if (SHARED_INVALID == shared_gt_h) /* global table not attached, try to create it in shared memory */
{ if (shared_debug) printf(" globalsharedtableinit=");
shared_gt_h = shmget(shared_kbase, shared_maxseg * sizeof(SHARED_GTAB), IPC_CREAT | IPC_EXCL | shared_create_mode); /* try open as a master */
if (SHARED_INVALID == shared_gt_h) /* if failed, try to open as a slave */
{ shared_gt_h = shmget(shared_kbase, shared_maxseg * sizeof(SHARED_GTAB), shared_create_mode);
if (SHARED_INVALID == shared_gt_h) return(SHARED_IPCERR); /* means deleted ID residing in system, shared mem unusable ... */
shared_gt = (SHARED_GTAB *)shmat(shared_gt_h, 0, 0); /* attach segment */
if (((SHARED_GTAB *)SHARED_INVALID) == shared_gt) return(SHARED_IPCERR);
if (shared_debug) printf("slave");
}
else
{ shared_gt = (SHARED_GTAB *)shmat(shared_gt_h, 0, 0); /* attach segment */
if (((SHARED_GTAB *)SHARED_INVALID) == shared_gt) return(SHARED_IPCERR);
for (i=0; i<shared_maxseg; i++) shared_clear_entry(i); /* since we are master, init data */
if (shared_debug) printf("master");
}
}
if (NULL == shared_lt) /* initialize local table */
{ if (shared_debug) printf(" localtableinit=");
if (NULL == (shared_lt = (SHARED_LTAB *)malloc(shared_maxseg * sizeof(SHARED_LTAB)))) return(SHARED_NOMEM);
for (i=0; i<shared_maxseg; i++)
{ shared_lt[i].p = NULL; /* not mapped */
shared_lt[i].tcnt = 0; /* unused (or zero threads using this seg) */
shared_lt[i].lkcnt = 0; /* segment is unlocked */
shared_lt[i].seekpos = 0L; /* r/w pointer at the beginning of file */
}
if (shared_debug) printf("ok");
}
atexit(shared_cleanup); /* we want shared_cleanup to be called at exit or abort */
if (shared_debug) printf(" <<done>>\n");
return(SHARED_OK);
}
int shared_recover(int id) /* try to recover dormant segments after applic crash */
{ int i, r, r2;
if (NULL == shared_gt) return(SHARED_NOTINIT); /* not initialized */
if (NULL == shared_lt) return(SHARED_NOTINIT); /* not initialized */
r = SHARED_OK;
for (i=0; i<shared_maxseg; i++)
{ if (-1 != id) if (i != id) continue;
if (shared_lt[i].tcnt) continue; /* somebody (we) is using it */
if (SHARED_INVALID == shared_gt[i].key) continue; /* unused slot */
if (shared_mux(i, SHARED_NOWAIT | SHARED_RDWRITE)) continue; /* acquire exclusive access to segment, but do not wait */
r2 = shared_process_count(shared_gt[i].sem);
if ((shared_gt[i].nprocdebug > r2) || (0 == r2))
{ if (shared_debug) printf("Bogus handle=%d nproc=%d sema=%d:", i, shared_gt[i].nprocdebug, r2);
r = shared_destroy_entry(i);
if (shared_debug)
{ printf("%s", r ? "error couldn't clear handle" : "handle cleared");
}
}
shared_demux(i, SHARED_RDWRITE);
}
return(r); /* table full */
}
/* API routines - mutexes and locking */
static int shared_mux(int idx, int mode) /* obtain exclusive access to specified segment */
{ flock_t flk;
int r;
if (0 == shared_init_called) /* delayed initialization */
{ if (SHARED_OK != (r = shared_init(0))) return(r);
}
if (SHARED_INVALID == shared_fd) return(SHARED_NOTINIT);
if ((idx < 0) || (idx >= shared_maxseg)) return(SHARED_BADARG);
flk.l_type = ((mode & SHARED_RDWRITE) ? F_WRLCK : F_RDLCK);
flk.l_whence = 0;
flk.l_start = idx;
flk.l_len = 1;
if (shared_debug) printf(" [mux (%d): ", idx);
if (-1 == fcntl(shared_fd, ((mode & SHARED_NOWAIT) ? F_SETLK : F_SETLKW), &flk))
{ switch (errno)
{ case EAGAIN: ;
case EACCES: if (shared_debug) printf("again]");
return(SHARED_AGAIN);
default: if (shared_debug) printf("err]");
return(SHARED_IPCERR);
}
}
if (shared_debug) printf("ok]");
return(SHARED_OK);
}
static int shared_demux(int idx, int mode) /* free exclusive access to specified segment */
{ flock_t flk;
if (SHARED_INVALID == shared_fd) return(SHARED_NOTINIT);
if ((idx < 0) || (idx >= shared_maxseg)) return(SHARED_BADARG);
flk.l_type = F_UNLCK;
flk.l_whence = 0;
flk.l_start = idx;
flk.l_len = 1;
if (shared_debug) printf(" [demux (%d): ", idx);
if (-1 == fcntl(shared_fd, F_SETLKW, &flk))
{ switch (errno)
{ case EAGAIN: ;
case EACCES: if (shared_debug) printf("again]");
return(SHARED_AGAIN);
default: if (shared_debug) printf("err]");
return(SHARED_IPCERR);
}
}
if (shared_debug) printf("mode=%d ok]", mode);
return(SHARED_OK);
}
static int shared_process_count(int sem) /* valid only for time of invocation */
{ union semun su;
su.val = 0; /* to force compiler not to give warning messages */
return(semctl(sem, 0, GETVAL, su)); /* su is unused here */
}
static int shared_delta_process(int sem, int delta) /* change number of processes hanging on segment */
{ struct sembuf sb;
if (SHARED_INVALID == sem) return(SHARED_BADARG); /* semaphore not attached */
sb.sem_num = 0;
sb.sem_op = delta;
sb.sem_flg = SEM_UNDO;
return((-1 == semop(sem, &sb, 1)) ? SHARED_IPCERR : SHARED_OK);
}
static int shared_attach_process(int sem)
{ if (shared_debug) printf(" [attach process]");
return(shared_delta_process(sem, 1));
}
static int shared_detach_process(int sem)
{ if (shared_debug) printf(" [detach process]");
return(shared_delta_process(sem, -1));
}
/* API routines - hashing and searching */
static int shared_get_free_entry(int newhandle) /* get newhandle, or -1, entry is set rw locked */
{
if (NULL == shared_gt) return(-1); /* not initialized */
if (NULL == shared_lt) return(-1); /* not initialized */
if (newhandle < 0) return(-1);
if (newhandle >= shared_maxseg) return(-1);
if (shared_lt[newhandle].tcnt) return(-1); /* somebody (we) is using it */
if (shared_mux(newhandle, SHARED_NOWAIT | SHARED_RDWRITE)) return(-1); /* used by others */
if (SHARED_INVALID == shared_gt[newhandle].key) return(newhandle); /* we have found free slot, lock it and return index */
shared_demux(newhandle, SHARED_RDWRITE);
if (shared_debug) printf("[free_entry - ERROR - entry unusable]");
return(-1); /* table full */
}
static int shared_get_hash(long size, int idx) /* return hash value for malloc */
{ static int counter = 0;
int hash;
hash = (counter + size * idx) % shared_range;
counter = (counter + 1) % shared_range;
return(hash);
}
static long shared_adjust_size(long size) /* size must be >= 0 !!! */
{ return(((size + sizeof(BLKHEAD) + SHARED_GRANUL - 1) / SHARED_GRANUL) * SHARED_GRANUL); }
/* API routines - core : malloc/realloc/free/attach/detach/lock/unlock */
int shared_malloc(long size, int mode, int newhandle) /* return idx or SHARED_INVALID */
{ int h, i, r, idx, key;
union semun filler;
BLKHEAD *bp;
if (0 == shared_init_called) /* delayed initialization */
{ if (SHARED_OK != (r = shared_init(0))) return(r);
}
if (shared_debug) printf("malloc (size = %ld, mode = %d):", size, mode);
if (size < 0) return(SHARED_INVALID);
if (-1 == (idx = shared_get_free_entry(newhandle))) return(SHARED_INVALID);
if (shared_debug) printf(" idx=%d", idx);
for (i = 0; ; i++)
{ if (i >= shared_range) /* table full, signal error & exit */
{ shared_demux(idx, SHARED_RDWRITE);
return(SHARED_INVALID);
}
key = shared_kbase + ((i + shared_get_hash(size, idx)) % shared_range);
if (shared_debug) printf(" key=%d", key);
h = shmget(key, shared_adjust_size(size), IPC_CREAT | IPC_EXCL | shared_create_mode);
if (shared_debug) printf(" handle=%d", h);
if (SHARED_INVALID == h) continue; /* segment already accupied */
bp = (BLKHEAD *)shmat(h, 0, 0); /* try attach */
if (shared_debug) printf(" p=%p", bp);
if (((BLKHEAD *)SHARED_INVALID) == bp) /* cannot attach, delete segment, try with another key */
{ shmctl(h, IPC_RMID, 0);
continue;
} /* now create semaphor counting number of processes attached */
if (SHARED_INVALID == (shared_gt[idx].sem = semget(key, 1, IPC_CREAT | IPC_EXCL | shared_create_mode)))
{ shmdt((void *)bp); /* cannot create segment, delete everything */
shmctl(h, IPC_RMID, 0);
continue; /* try with another key */
}
if (shared_debug) printf(" sem=%d", shared_gt[idx].sem);
if (shared_attach_process(shared_gt[idx].sem)) /* try attach process */
{ semctl(shared_gt[idx].sem, 0, IPC_RMID, filler); /* destroy semaphore */
shmdt((char *)bp); /* detach shared mem segment */
shmctl(h, IPC_RMID, 0); /* destroy shared mem segment */
continue; /* try with another key */
}
bp->s.tflag = BLOCK_SHARED; /* fill in data in segment's header (this is really not necessary) */
bp->s.ID[0] = SHARED_ID_0;
bp->s.ID[1] = SHARED_ID_1;
bp->s.handle = idx; /* used in yorick */
if (mode & SHARED_RESIZE)
{ if (shmdt((char *)bp)) r = SHARED_IPCERR; /* if segment is resizable, then detach segment */
shared_lt[idx].p = NULL;
}
else { shared_lt[idx].p = bp; }
shared_lt[idx].tcnt = 1; /* one thread using segment */
shared_lt[idx].lkcnt = 0; /* no locks at the moment */
shared_lt[idx].seekpos = 0L; /* r/w pointer positioned at beg of block */
shared_gt[idx].handle = h; /* fill in data in global table */
shared_gt[idx].size = size;
shared_gt[idx].attr = mode;
shared_gt[idx].semkey = key;
shared_gt[idx].key = key;
shared_gt[idx].nprocdebug = 0;
break;
}
shared_demux(idx, SHARED_RDWRITE); /* hope this will not fail */
return(idx);
}
int shared_attach(int idx)
{ int r, r2;
if (SHARED_OK != (r = shared_mux(idx, SHARED_RDWRITE | SHARED_WAIT))) return(r);
if (SHARED_OK != (r = shared_map(idx)))
{ shared_demux(idx, SHARED_RDWRITE);
return(r);
}
if (shared_attach_process(shared_gt[idx].sem)) /* try attach process */
{ shmdt((char *)(shared_lt[idx].p)); /* cannot attach process, detach everything */
shared_lt[idx].p = NULL;
shared_demux(idx, SHARED_RDWRITE);
return(SHARED_BADARG);
}
shared_lt[idx].tcnt++; /* one more thread is using segment */
if (shared_gt[idx].attr & SHARED_RESIZE) /* if resizeable, detach and return special pointer */
{ if (shmdt((char *)(shared_lt[idx].p))) r = SHARED_IPCERR; /* if segment is resizable, then detach segment */
shared_lt[idx].p = NULL;
}
shared_lt[idx].seekpos = 0L; /* r/w pointer positioned at beg of block */
r2 = shared_demux(idx, SHARED_RDWRITE);
return(r ? r : r2);
}
static int shared_check_locked_index(int idx) /* verify that given idx is valid */
{ int r;
if (0 == shared_init_called) /* delayed initialization */
{ if (SHARED_OK != (r = shared_init(0))) return(r);
}
if ((idx < 0) || (idx >= shared_maxseg)) return(SHARED_BADARG);
if (NULL == shared_lt[idx].p) return(SHARED_BADARG); /* NULL pointer, not attached ?? */
if (0 == shared_lt[idx].lkcnt) return(SHARED_BADARG); /* not locked ?? */
if ((SHARED_ID_0 != (shared_lt[idx].p)->s.ID[0]) || (SHARED_ID_1 != (shared_lt[idx].p)->s.ID[1]) ||
(BLOCK_SHARED != (shared_lt[idx].p)->s.tflag)) /* invalid data in segment */
return(SHARED_BADARG);
return(SHARED_OK);
}
static int shared_map(int idx) /* map all tables for given idx, check for validity */
{ int h; /* have to obtain excl. access before calling shared_map */
BLKHEAD *bp;
if ((idx < 0) || (idx >= shared_maxseg)) return(SHARED_BADARG);
if (SHARED_INVALID == shared_gt[idx].key) return(SHARED_BADARG);
if (SHARED_INVALID == (h = shmget(shared_gt[idx].key, 1, shared_create_mode))) return(SHARED_BADARG);
if (((BLKHEAD *)SHARED_INVALID) == (bp = (BLKHEAD *)shmat(h, 0, 0))) return(SHARED_BADARG);
if ((SHARED_ID_0 != bp->s.ID[0]) || (SHARED_ID_1 != bp->s.ID[1]) || (BLOCK_SHARED != bp->s.tflag) || (h != shared_gt[idx].handle))
{ shmdt((char *)bp); /* invalid segment, detach everything */
return(SHARED_BADARG);
}
if (shared_gt[idx].sem != semget(shared_gt[idx].semkey, 1, shared_create_mode)) /* check if sema is still there */
{ shmdt((char *)bp); /* cannot attach semaphore, detach everything */
return(SHARED_BADARG);
}
shared_lt[idx].p = bp; /* store pointer to shmem data */
return(SHARED_OK);
}
static int shared_validate(int idx, int mode) /* use intrnally inside crit.sect !!! */
{ int r;
if (SHARED_OK != (r = shared_mux(idx, mode))) return(r); /* idx checked by shared_mux */
if (NULL == shared_lt[idx].p)
if (SHARED_OK != (r = shared_map(idx)))
{ shared_demux(idx, mode);
return(r);
}
if ((SHARED_ID_0 != (shared_lt[idx].p)->s.ID[0]) || (SHARED_ID_1 != (shared_lt[idx].p)->s.ID[1]) || (BLOCK_SHARED != (shared_lt[idx].p)->s.tflag))
{ shared_demux(idx, mode);
return(r);
}
return(SHARED_OK);
}
SHARED_P shared_realloc(int idx, long newsize) /* realloc shared memory segment */
{ int h, key, i, r;
BLKHEAD *bp;
long transfersize;
r = SHARED_OK;
if (newsize < 0) return(NULL);
if (shared_check_locked_index(idx)) return(NULL);
if (0 == (shared_gt[idx].attr & SHARED_RESIZE)) return(NULL);
if (-1 != shared_lt[idx].lkcnt) return(NULL); /* check for RW lock */
if (shared_adjust_size(shared_gt[idx].size) == shared_adjust_size(newsize))
{ shared_gt[idx].size = newsize;
return((SHARED_P)((shared_lt[idx].p) + 1));
}
for (i = 0; ; i++)
{ if (i >= shared_range) return(NULL); /* table full, signal error & exit */
key = shared_kbase + ((i + shared_get_hash(newsize, idx)) % shared_range);
h = shmget(key, shared_adjust_size(newsize), IPC_CREAT | IPC_EXCL | shared_create_mode);
if (SHARED_INVALID == h) continue; /* segment already accupied */
bp = (BLKHEAD *)shmat(h, 0, 0); /* try attach */
if (((BLKHEAD *)SHARED_INVALID) == bp) /* cannot attach, delete segment, try with another key */
{ shmctl(h, IPC_RMID, 0);
continue;
}
*bp = *(shared_lt[idx].p); /* copy header, then data */
transfersize = ((newsize < shared_gt[idx].size) ? newsize : shared_gt[idx].size);
if (transfersize > 0)
memcpy((void *)(bp + 1), (void *)((shared_lt[idx].p) + 1), transfersize);
if (shmdt((char *)(shared_lt[idx].p))) r = SHARED_IPCERR; /* try to detach old segment */
if (shmctl(shared_gt[idx].handle, IPC_RMID, 0)) if (SHARED_OK == r) r = SHARED_IPCERR; /* destroy old shared memory segment */
shared_gt[idx].size = newsize; /* signal new size */
shared_gt[idx].handle = h; /* signal new handle */
shared_gt[idx].key = key; /* signal new key */
shared_lt[idx].p = bp;
break;
}
return((SHARED_P)(bp + 1));
}
int shared_free(int idx) /* detach segment, if last process & !PERSIST, destroy segment */
{ int cnt, r, r2;
if (SHARED_OK != (r = shared_validate(idx, SHARED_RDWRITE | SHARED_WAIT))) return(r);
if (SHARED_OK != (r = shared_detach_process(shared_gt[idx].sem))) /* update number of processes using segment */
{ shared_demux(idx, SHARED_RDWRITE);
return(r);
}
shared_lt[idx].tcnt--; /* update number of threads using segment */
if (shared_lt[idx].tcnt > 0) return(shared_demux(idx, SHARED_RDWRITE)); /* if more threads are using segment we are done */
if (shmdt((char *)(shared_lt[idx].p))) /* if, we are the last thread, try to detach segment */
{ shared_demux(idx, SHARED_RDWRITE);
return(SHARED_IPCERR);
}
shared_lt[idx].p = NULL; /* clear entry in local table */
shared_lt[idx].seekpos = 0L; /* r/w pointer positioned at beg of block */
if (-1 == (cnt = shared_process_count(shared_gt[idx].sem))) /* get number of processes hanging on segment */
{ shared_demux(idx, SHARED_RDWRITE);
return(SHARED_IPCERR);
}
if ((0 == cnt) && (0 == (shared_gt[idx].attr & SHARED_PERSIST))) r = shared_destroy_entry(idx); /* no procs on seg, destroy it */
r2 = shared_demux(idx, SHARED_RDWRITE);
return(r ? r : r2);
}
SHARED_P shared_lock(int idx, int mode) /* lock given segment for exclusive access */
{ int r;
if (shared_mux(idx, mode)) return(NULL); /* idx checked by shared_mux */
if (0 != shared_lt[idx].lkcnt) /* are we already locked ?? */
if (SHARED_OK != (r = shared_map(idx)))
{ shared_demux(idx, mode);
return(NULL);
}
if (NULL == shared_lt[idx].p) /* stupid pointer ?? */
if (SHARED_OK != (r = shared_map(idx)))
{ shared_demux(idx, mode);
return(NULL);
}
if ((SHARED_ID_0 != (shared_lt[idx].p)->s.ID[0]) || (SHARED_ID_1 != (shared_lt[idx].p)->s.ID[1]) || (BLOCK_SHARED != (shared_lt[idx].p)->s.tflag))
{ shared_demux(idx, mode);
return(NULL);
}
if (mode & SHARED_RDWRITE)
{ shared_lt[idx].lkcnt = -1;
shared_gt[idx].nprocdebug++;
}
else shared_lt[idx].lkcnt++;
shared_lt[idx].seekpos = 0L; /* r/w pointer positioned at beg of block */
return((SHARED_P)((shared_lt[idx].p) + 1));
}
int shared_unlock(int idx) /* unlock given segment, assumes seg is locked !! */
{ int r, r2, mode;
if (SHARED_OK != (r = shared_check_locked_index(idx))) return(r);
if (shared_lt[idx].lkcnt > 0)
{ shared_lt[idx].lkcnt--; /* unlock read lock */
mode = SHARED_RDONLY;
}
else
{ shared_lt[idx].lkcnt = 0; /* unlock write lock */
shared_gt[idx].nprocdebug--;
mode = SHARED_RDWRITE;
}
if (0 == shared_lt[idx].lkcnt) if (shared_gt[idx].attr & SHARED_RESIZE)
{ if (shmdt((char *)(shared_lt[idx].p))) r = SHARED_IPCERR; /* segment is resizable, then detach segment */
shared_lt[idx].p = NULL; /* signal detachment in local table */
}
r2 = shared_demux(idx, mode); /* unlock segment, rest is only parameter checking */
return(r ? r : r2);
}
/* API routines - support and info routines */
int shared_attr(int idx) /* get the attributes of the shared memory segment */
{ int r;
if (shared_check_locked_index(idx)) return(SHARED_INVALID);
r = shared_gt[idx].attr;
return(r);
}
int shared_set_attr(int idx, int newattr) /* get the attributes of the shared memory segment */
{ int r;
if (shared_check_locked_index(idx)) return(SHARED_INVALID);
if (-1 != shared_lt[idx].lkcnt) return(SHARED_INVALID); /* ADDED - check for RW lock */
r = shared_gt[idx].attr;
shared_gt[idx].attr = newattr;
return(r);
}
int shared_set_debug(int mode) /* set/reset debug mode */
{ int r = shared_debug;
shared_debug = mode;
return(r);
}
int shared_set_createmode(int mode) /* set/reset debug mode */
{ int r = shared_create_mode;
shared_create_mode = mode;
return(r);
}
int shared_list(int id)
{ int i, r;
if (NULL == shared_gt) return(SHARED_NOTINIT); /* not initialized */
if (NULL == shared_lt) return(SHARED_NOTINIT); /* not initialized */
if (shared_debug) printf("shared_list:");
r = SHARED_OK;
printf(" Idx Key Nproc Size Flags\n");
printf("==============================================\n");
for (i=0; i<shared_maxseg; i++)
{ if (-1 != id) if (i != id) continue;
if (SHARED_INVALID == shared_gt[i].key) continue; /* unused slot */
switch (shared_mux(i, SHARED_NOWAIT | SHARED_RDONLY)) /* acquire exclusive access to segment, but do not wait */
{ case SHARED_AGAIN:
printf("!%3d %08lx %4d %8d", i, (unsigned long int)shared_gt[i].key,
shared_gt[i].nprocdebug, shared_gt[i].size);
if (SHARED_RESIZE & shared_gt[i].attr) printf(" RESIZABLE");
if (SHARED_PERSIST & shared_gt[i].attr) printf(" PERSIST");
printf("\n");
break;
case SHARED_OK:
printf(" %3d %08lx %4d %8d", i, (unsigned long int)shared_gt[i].key,
shared_gt[i].nprocdebug, shared_gt[i].size);
if (SHARED_RESIZE & shared_gt[i].attr) printf(" RESIZABLE");
if (SHARED_PERSIST & shared_gt[i].attr) printf(" PERSIST");
printf("\n");
shared_demux(i, SHARED_RDONLY);
break;
default:
continue;
}
}
if (shared_debug) printf(" done\n");
return(r); /* table full */
}
int shared_getaddr(int id, char **address)
{ int i;
char segname[10];
if (NULL == shared_gt) return(SHARED_NOTINIT); /* not initialized */
if (NULL == shared_lt) return(SHARED_NOTINIT); /* not initialized */
strcpy(segname,"h");
snprintf(segname+1,9,"%d", id);
if (smem_open(segname,0,&i)) return(SHARED_BADARG);
*address = ((char *)(((DAL_SHM_SEGHEAD *)(shared_lt[i].p + 1)) + 1));
/* smem_close(i); */
return(SHARED_OK);
}
int shared_uncond_delete(int id)
{ int i, r;
if (NULL == shared_gt) return(SHARED_NOTINIT); /* not initialized */
if (NULL == shared_lt) return(SHARED_NOTINIT); /* not initialized */
if (shared_debug) printf("shared_uncond_delete:");
r = SHARED_OK;
for (i=0; i<shared_maxseg; i++)
{ if (-1 != id) if (i != id) continue;
if (shared_attach(i))
{ if (-1 != id) printf("no such handle\n");
continue;
}
printf("handle %d:", i);
if (NULL == shared_lock(i, SHARED_RDWRITE | SHARED_NOWAIT))
{ printf(" cannot lock in RW mode, not deleted\n");
continue;
}
if (shared_set_attr(i, SHARED_RESIZE) >= SHARED_ERRBASE)
{ printf(" cannot clear PERSIST attribute");
}
if (shared_free(i))
{ printf(" delete failed\n");
}
else
{ printf(" deleted\n");
}
}
if (shared_debug) printf(" done\n");
return(r); /* table full */
}
/************************* CFITSIO DRIVER FUNCTIONS ***************************/
int smem_init(void)
{ return(0);
}
int smem_shutdown(void)
{ if (shared_init_called) shared_cleanup();
return(0);
}
int smem_setoptions(int option)
{ option = 0;
return(0);
}
int smem_getoptions(int *options)
{ if (NULL == options) return(SHARED_NULPTR);
*options = 0;
return(0);
}
int smem_getversion(int *version)
{ if (NULL == version) return(SHARED_NULPTR);
*version = 10;
return(0);
}
int smem_open(char *filename, int rwmode, int *driverhandle)
{ int h, nitems, r;
DAL_SHM_SEGHEAD *sp;
if (NULL == filename) return(SHARED_NULPTR);
if (NULL == driverhandle) return(SHARED_NULPTR);
nitems = sscanf(filename, "h%d", &h);
if (1 != nitems) return(SHARED_BADARG);
if (SHARED_OK != (r = shared_attach(h))) return(r);
if (NULL == (sp = (DAL_SHM_SEGHEAD *)shared_lock(h,
((READWRITE == rwmode) ? SHARED_RDWRITE : SHARED_RDONLY))))
{ shared_free(h);
return(SHARED_BADARG);
}
if ((h != sp->h) || (DAL_SHM_SEGHEAD_ID != sp->ID))
{ shared_unlock(h);
shared_free(h);
return(SHARED_BADARG);
}
*driverhandle = h;
return(0);
}
int smem_create(char *filename, int *driverhandle)
{ DAL_SHM_SEGHEAD *sp;
int h, sz, nitems;
if (NULL == filename) return(SHARED_NULPTR); /* currently ignored */
if (NULL == driverhandle) return(SHARED_NULPTR);
nitems = sscanf(filename, "h%d", &h);
if (1 != nitems) return(SHARED_BADARG);
if (SHARED_INVALID == (h = shared_malloc(sz = 2880 + sizeof(DAL_SHM_SEGHEAD),
SHARED_RESIZE | SHARED_PERSIST, h)))
return(SHARED_NOMEM);
if (NULL == (sp = (DAL_SHM_SEGHEAD *)shared_lock(h, SHARED_RDWRITE)))
{ shared_free(h);
return(SHARED_BADARG);
}
sp->ID = DAL_SHM_SEGHEAD_ID;
sp->h = h;
sp->size = sz;
sp->nodeidx = -1;
*driverhandle = h;
return(0);
}
int smem_close(int driverhandle)
{ int r;
if (SHARED_OK != (r = shared_unlock(driverhandle))) return(r);
return(shared_free(driverhandle));
}
int smem_remove(char *filename)
{ int nitems, h, r;
if (NULL == filename) return(SHARED_NULPTR);
nitems = sscanf(filename, "h%d", &h);
if (1 != nitems) return(SHARED_BADARG);
if (0 == shared_check_locked_index(h)) /* are we locked ? */
{ if (-1 != shared_lt[h].lkcnt) /* are we locked RO ? */
{ if (SHARED_OK != (r = shared_unlock(h))) return(r); /* yes, so relock in RW */
if (NULL == shared_lock(h, SHARED_RDWRITE)) return(SHARED_BADARG);
}
}
else /* not locked */
{ if (SHARED_OK != (r = smem_open(filename, READWRITE, &h)))
return(r); /* so open in RW mode */
}
shared_set_attr(h, SHARED_RESIZE); /* delete PERSIST attribute */
return(smem_close(h)); /* detach segment (this will delete it) */
}
int smem_size(int driverhandle, LONGLONG *size)
{
if (NULL == size) return(SHARED_NULPTR);
if (shared_check_locked_index(driverhandle)) return(SHARED_INVALID);
*size = (LONGLONG) (shared_gt[driverhandle].size - sizeof(DAL_SHM_SEGHEAD));
return(0);
}
int smem_flush(int driverhandle)
{
if (shared_check_locked_index(driverhandle)) return(SHARED_INVALID);
return(0);
}
int smem_seek(int driverhandle, LONGLONG offset)
{
if (offset < 0) return(SHARED_BADARG);
if (shared_check_locked_index(driverhandle)) return(SHARED_INVALID);
shared_lt[driverhandle].seekpos = offset;
return(0);
}
int smem_read(int driverhandle, void *buffer, long nbytes)
{
if (NULL == buffer) return(SHARED_NULPTR);
if (shared_check_locked_index(driverhandle)) return(SHARED_INVALID);
if (nbytes < 0) return(SHARED_BADARG);
if ((shared_lt[driverhandle].seekpos + nbytes) > shared_gt[driverhandle].size)
return(SHARED_BADARG); /* read beyond EOF */
memcpy(buffer,
((char *)(((DAL_SHM_SEGHEAD *)(shared_lt[driverhandle].p + 1)) + 1)) +
shared_lt[driverhandle].seekpos,
nbytes);
shared_lt[driverhandle].seekpos += nbytes;
return(0);
}
int smem_write(int driverhandle, void *buffer, long nbytes)
{
if (NULL == buffer) return(SHARED_NULPTR);
if (shared_check_locked_index(driverhandle)) return(SHARED_INVALID);
if (-1 != shared_lt[driverhandle].lkcnt) return(SHARED_INVALID); /* are we locked RW ? */
if (nbytes < 0) return(SHARED_BADARG);
if ((unsigned long)(shared_lt[driverhandle].seekpos + nbytes) > (unsigned long)(shared_gt[driverhandle].size - sizeof(DAL_SHM_SEGHEAD)))
{ /* need to realloc shmem */
if (NULL == shared_realloc(driverhandle, shared_lt[driverhandle].seekpos + nbytes + sizeof(DAL_SHM_SEGHEAD)))
return(SHARED_NOMEM);
}
memcpy(((char *)(((DAL_SHM_SEGHEAD *)(shared_lt[driverhandle].p + 1)) + 1)) +
shared_lt[driverhandle].seekpos,
buffer,
nbytes);
shared_lt[driverhandle].seekpos += nbytes;
return(0);
}
#endif