forked from bloomberg/comdb2
-
Notifications
You must be signed in to change notification settings - Fork 0
/
handle_buf.c
1318 lines (1167 loc) · 43.4 KB
/
handle_buf.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
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
/*
Copyright 2015, 2017, Bloomberg Finance L.P.
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
*/
/* comdb index front end */
#include <stdio.h>
#include <pthread.h>
#include <stddef.h>
#include <errno.h>
#include <string.h>
#include <strings.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <assert.h>
#include <unistd.h>
#include <epochlib.h>
#include <list.h>
#include <pool.h>
#include <time.h>
#include "debug_switches.h"
#include "lockmacros.h"
#include "comdb2.h"
#include "block_internal.h"
#include "util.h"
#include "translistener.h"
#include "socket_interfaces.h"
#include "osqlsession.h"
#include "sql.h"
#include "osqlblockproc.h"
#include "intern_strings.h"
#include "logmsg.h"
#include "transactionstate_systable.h"
#include <poll.h>
#ifdef MONITOR_STACK
#include "comdb2_pthread_create.h"
#endif
extern int free_it(void *obj, void *arg);
extern void destroy_hash(hash_t *h, int (*free_func)(void *, void *));
pthread_key_t thd_info_key;
void (*comdb2_ipc_sndbak)(int *, int) = 0;
enum THD_EV { THD_EV_END = 0, THD_EV_START = 1 };
/* request pool & queue */
static pool_t *p_reqs; /* request pool */
struct dbq_entry_t {
LINKC_T(struct dbq_entry_t) qlnk;
LINKC_T(struct dbq_entry_t) rqlnk;
time_t queue_time_ms;
void *obj;
};
static pool_t *pq_reqs; /* queue entry pool */
pool_t *p_bufs; /* buffer pool for socket requests */
pool_t *p_slocks; /* pool of socket locks*/
LISTC_T(struct dbq_entry_t) q_reqs; /* all queued requests */
static LISTC_T(struct dbq_entry_t) rq_reqs; /* queue of read requests */
/* thread pool */
/* thread associated with this request */
struct thd {
pthread_t tid;
pthread_cond_t wakeup;
struct ireq *iq;
LINKC_T(struct thd) lnk;
};
static pool_t *p_thds;
static LISTC_T(struct thd) idle; /*idle thread.*/
static LISTC_T(struct thd) busy; /*busy thread.*/
static int write_thd_count = 0;
static int is_req_write(int opcode);
static pthread_mutex_t lock;
pthread_mutex_t buf_lock = PTHREAD_MUTEX_INITIALIZER;
static pthread_attr_t attr;
#ifdef MONITOR_STACK
static size_t stack_sz;
static comdb2ma stack_alloc;
#endif
static pthread_cond_t coalesce_wakeup;
static unsigned coalesce_waiters = 0;
static unsigned coalesce_reqthd_waiters = 0;
/* stats */
enum { MAXSTAT = 200 };
static int nreqs;
static int nthdcreates;
static int nwaits;
static int nqfulls;
static int nerrs;
static int nretire;
static int bkt_thd[MAXSTAT];
static int bkt_queue[MAXSTAT];
static int iothreads = 0, waitthreads = 0;
void test_the_lock(void)
{
LOCK(&lock) {}
UNLOCK(&lock)
}
static void thd_io_start(void)
{
LOCK(&lock) { iothreads++; }
UNLOCK(&lock);
}
static void thd_io_complete(void)
{
LOCK(&lock) { iothreads--; }
UNLOCK(&lock);
}
#define LISTC_CLEAN(listp, lnk, dofree, type) { \
type *item, *tmp; \
/* free each item */ \
LISTC_FOR_EACH_SAFE(listp, item, tmp, lnk) { \
/* remove and potentially free item */ \
listc_rfl(listp, item); \
if (dofree) \
free(item); \
} \
}
void thd_cleanup()
{
int counter = 0;
while (nthdcreates > nretire && counter++ < gbl_thd_linger)
sleep(1);
if (nthdcreates > nretire)
abort();
LISTC_CLEAN(&busy, lnk, 0, struct thd);
LISTC_CLEAN(&idle, lnk, 0, struct thd);
LISTC_CLEAN(&rq_reqs, rqlnk, 0, struct dbq_entry_t);
LISTC_CLEAN(&q_reqs, qlnk, 0, struct dbq_entry_t);
pool_clear(pq_reqs);
pool_clear(p_slocks);
pool_clear(p_bufs);
pool_clear(p_reqs);
pool_clear(p_thds);
Pthread_cond_destroy(&coalesce_wakeup);
Pthread_attr_destroy(&attr);
Pthread_mutex_destroy(&lock);
}
int thd_init(void)
{
Pthread_mutex_init(&lock, 0);
Pthread_attr_init(&attr);
Pthread_attr_setdetachstate(&attr, PTHREAD_CREATE_DETACHED);
Pthread_cond_init(&coalesce_wakeup, NULL);
p_thds = pool_setalloc_init(sizeof(struct thd), 0, malloc, free);
if (p_thds == 0) {
logmsg(LOGMSG_ERROR, "thd_init:failed thd pool init");
return -1;
}
p_reqs = pool_setalloc_init(sizeof(struct ireq), 0, malloc, free);
if (p_reqs == 0) {
logmsg(LOGMSG_ERROR, "thd_init:failed req pool init");
return -1;
}
p_bufs = pool_setalloc_init(MAX_BUFFER_SIZE, 64, malloc, free);
if (p_bufs == 0) {
logmsg(LOGMSG_ERROR, "thd_init:failed buf pool init");
return -1;
}
p_slocks = pool_setalloc_init(sizeof(struct buf_lock_t), 64, malloc, free);
if (p_slocks == 0) {
logmsg(LOGMSG_ERROR, "thd_init:failed sock lock pool init");
return -1;
}
pq_reqs = pool_setalloc_init(sizeof(struct dbq_entry_t), 64, malloc, free);
if (pq_reqs == 0) {
logmsg(LOGMSG_ERROR, "thd_init:failed queue req pool init");
return -1;
}
listc_init(&q_reqs, offsetof(struct dbq_entry_t, qlnk));
listc_init(&rq_reqs, offsetof(struct dbq_entry_t, rqlnk));
#ifdef MONITOR_STACK
stack_sz = 4096 * 1024;
stack_alloc = comdb2ma_create_with_scope(0, 0, "stack", "tag", 1);
if (stack_alloc == NULL) {
logmsg(LOGMSG_ERROR, "thd_init: failed to initialize thread subsystem\n");
return 1;
}
#endif
listc_init(&idle, offsetof(struct thd, lnk));
listc_init(&busy, offsetof(struct thd, lnk));
bdb_set_io_control(thd_io_start, thd_io_complete);
Pthread_key_create(&thd_info_key, free);
Pthread_attr_setstacksize(&attr, 4096 * 1024);
logmsg(LOGMSG_INFO, "thd_init: thread subsystem initialized\n");
return 0;
}
void thd_stats(void)
{
int ii, jj;
logmsg(LOGMSG_USER, "num reqs %d\n", nreqs);
logmsg(LOGMSG_USER, "num waits %d\n", nwaits);
logmsg(LOGMSG_USER, "num items on queue %d\n", q_reqs.count);
logmsg(LOGMSG_USER, "num reads on queue %d\n", rq_reqs.count);
logmsg(LOGMSG_USER, "num threads wrt busy/busy/idle %d/%d/%d\n", write_thd_count,
busy.count, idle.count);
logmsg(LOGMSG_USER, "num threads in i/o %d\n", iothreads);
logmsg(LOGMSG_USER, "num threads waiting %d\n", waitthreads);
logmsg(LOGMSG_USER, "---\n");
logmsg(LOGMSG_USER, "config:MAXTHREADS %d\n", gbl_maxthreads);
logmsg(LOGMSG_USER, "config:MAXWRTTHREADS %d\n", gbl_maxwthreads);
logmsg(LOGMSG_USER, "gbl_maxwthreadpenalty %d\n", gbl_maxwthreadpenalty);
logmsg(LOGMSG_USER, "penaltyincpercent %d\n", gbl_penaltyincpercent);
logmsg(LOGMSG_USER, "config:MAXQUEUE %d\n", gbl_maxqueue);
logmsg(LOGMSG_USER, "---\n");
logmsg(LOGMSG_USER, "num queue fulls %d\n", nqfulls);
logmsg(LOGMSG_USER, "num errors %d\n", nerrs);
logmsg(LOGMSG_USER, "num thread creates %d\n", nthdcreates);
logmsg(LOGMSG_USER, "num retires %d\n", nretire);
logmsg(LOGMSG_USER, "---\n");
logmsg(LOGMSG_USER, "#threads:count\n");
for (jj = 0, ii = 0; ii < MAXSTAT; ii++) {
if (ii < 4 || bkt_thd[ii] > 0) {
logmsg(LOGMSG_USER, " %-3d:%-8d", ii + 1, bkt_thd[ii]);
jj++;
if ((jj % 4) == 0)
logmsg(LOGMSG_USER, "\n");
}
}
if ((jj % 4) != 0)
logmsg(LOGMSG_USER, "\n");
logmsg(LOGMSG_USER, "#queue:count\n");
for (jj = 0, ii = 0; ii < MAXSTAT; ii++) {
if (ii < 4 || bkt_queue[ii] > 0) {
logmsg(LOGMSG_USER, " %-3d:%-8d", ii + 1, bkt_queue[ii]);
jj++;
if ((jj % 4) == 0)
logmsg(LOGMSG_USER, "\n");
}
}
if ((jj % 4) != 0)
logmsg(LOGMSG_USER, "\n");
}
void thd_dbinfo2_stats(struct db_info2_stats *stats)
{
int ii;
uint32_t queue_sum = 0, queue_mean = 0;
stats->thr_max = gbl_maxthreads;
stats->thr_maxwr = gbl_maxwthreads;
stats->thr_cur = nthdcreates - nretire;
stats->q_max_conf = gbl_maxqueue;
for (ii = 0; ii < MAXSTAT; ii++) {
if (bkt_queue[ii] > 0) {
stats->q_max_reached = ii + 1;
queue_mean += bkt_queue[ii] * stats->q_max_reached;
queue_sum += bkt_queue[ii];
}
}
if (stats->q_max_reached > 0) {
double f = (double)queue_mean / (double)queue_sum;
f += 0.5;
stats->q_mean_reached = (uint32_t)f;
} else
stats->q_mean_reached = 0;
}
static void thd_coalesce_check_ll(void)
{
if (coalesce_waiters && busy.count <= coalesce_reqthd_waiters &&
q_reqs.count == 0) {
Pthread_cond_broadcast(&coalesce_wakeup);
}
}
static void thd_dump_nolock(void)
{
struct thd *thd;
uint64_t nowus;
int opc, cnt = 0;
nowus = comdb2_time_epochus();
{
for (thd = busy.top; thd; thd = thd->lnk.next) {
cnt++;
opc = thd->iq->opcode;
logmsg(LOGMSG_USER,
"busy tid %p time %5d ms %-6s (%-3d) "
"%-20s where %s %s\n",
(void *)thd->tid, U2M(nowus - thd->iq->nowus), req2a(opc), opc, getorigin(thd->iq), thd->iq->where,
thd->iq->gluewhere);
}
for (thd = idle.top; thd; thd = thd->lnk.next) {
cnt++;
logmsg(LOGMSG_USER, "idle tid %p \n", (void *)thd->tid);
}
}
if (cnt == 0)
logmsg(LOGMSG_USER, "no active threads\n");
}
int thd_queue_depth(void)
{
return q_reqs.count;
}
void thd_coalesce(struct dbenv *dbenv)
{
LOCK(&lock)
{
struct thd *thd;
int am_req_thd = 0;
int num_wait = 0;
/* fstsnd based fastinit can lead to us waiting for ourself.. check if
* this is one of the request threads and if so that's one less
* thread to wait for. */
LISTC_FOR_EACH(&busy, thd, lnk)
{
if (thd->tid == pthread_self()) {
am_req_thd = 1;
break;
}
}
coalesce_waiters++;
coalesce_reqthd_waiters += am_req_thd;
while (busy.count > coalesce_reqthd_waiters || q_reqs.count > 0) {
int rc;
struct timespec ts;
++num_wait;
logmsg(LOGMSG_USER, "waiting for threads %d/%d/%d num queue %d\n",
write_thd_count, busy.count, idle.count, q_reqs.count);
if (num_wait > 5)
thd_dump_nolock();
rc = clock_gettime(CLOCK_REALTIME, &ts);
ts.tv_sec += 1;
rc = pthread_cond_timedwait(&coalesce_wakeup, &lock, &ts);
if (rc != 0 && rc != ETIMEDOUT)
logmsg(LOGMSG_ERROR, "%s:pthread_cond_timedwait: %d %s\n", __func__,
rc, strerror(rc));
}
coalesce_waiters--;
coalesce_reqthd_waiters -= am_req_thd;
}
UNLOCK(&lock);
}
void thd_dump(void)
{
struct thd *thd;
uint64_t nowus;
int cnt = 0;
nowus = comdb2_time_epochus();
LOCK(&lock)
{
for (thd = busy.top; thd; thd = thd->lnk.next) {
cnt++;
logmsg(LOGMSG_USER,
"busy tid %p time %5d ms %-6s (%-3d) %-20s where %s "
"%s\n",
(void *)thd->tid, U2M(nowus - thd->iq->nowus), req2a(thd->iq->opcode), thd->iq->opcode,
getorigin(thd->iq), thd->iq->where, thd->iq->gluewhere);
}
for (thd = idle.top; thd; thd = thd->lnk.next) {
cnt++;
logmsg(LOGMSG_USER, "idle tid %p \n", (void *)thd->tid);
}
}
UNLOCK(&lock);
if (cnt == 0)
logmsg(LOGMSG_USER, "no active threads\n");
}
int get_thd_info(thd_info **data, int *npoints) {
struct thd_info *tinfo;
struct thd *thd;
uint64_t nowus;
nowus = comdb2_time_epochus();
LOCK(&lock)
{
*npoints = busy.count + idle.count;
*data = tinfo = malloc((*npoints)*sizeof(thd_info));
LISTC_FOR_EACH(&busy, thd, lnk)
{
tinfo->state = strdup("busy");
tinfo->time = U2M(nowus - thd->iq->nowus);
tinfo->machine = strdup(thd->iq->frommach);
tinfo->opcode = strdup(thd->iq->where);
tinfo->function = strdup(thd->iq->gluewhere);
tinfo->isIdle = 0;
++tinfo;
}
LISTC_FOR_EACH(&idle, thd, lnk)
{
tinfo->state = strdup("idle");
tinfo->isIdle = 1;
++tinfo;
}
}
UNLOCK(&lock);
return 0;
}
void free_thd_info(thd_info *data, int npoints) {
thd_info *tinfo = data;
for (int i=0; i<npoints; ++i) {
if (!tinfo->isIdle) {
free(tinfo->machine);
free(tinfo->opcode);
free(tinfo->function);
}
free(tinfo->state);
++tinfo;
}
free(data);
}
uint8_t *get_bigbuf()
{
uint8_t *p_buf = NULL;
LOCK(&buf_lock) { p_buf = pool_getablk(p_bufs); }
UNLOCK(&buf_lock);
return p_buf;
}
int free_bigbuf_nosignal(uint8_t *p_buf)
{
LOCK(&buf_lock) { pool_relablk(p_bufs, p_buf); }
UNLOCK(&buf_lock);
return 0;
}
int free_bigbuf(uint8_t *p_buf, struct buf_lock_t *p_slock)
{
if (p_slock == NULL)
return 0;
p_slock->reply_state = REPLY_STATE_DONE;
LOCK(&buf_lock) { pool_relablk(p_bufs, p_buf); }
UNLOCK(&buf_lock);
Pthread_cond_signal(&(p_slock->wait_cond));
return 0;
}
int signal_buflock(struct buf_lock_t *p_slock)
{
p_slock->reply_state = REPLY_STATE_DONE;
Pthread_cond_signal(&(p_slock->wait_cond));
return 0;
}
/* request handler */
static void *thd_req(void *vthd)
{
comdb2_name_thread(__func__);
struct thd *thd = (struct thd *)vthd;
struct dbenv *dbenv;
struct timespec ts;
int rc;
int iamwriter = 0;
struct thread_info *thdinfo;
struct thr_handle *thr_self;
struct reqlogger *logger;
int numwriterthreads;
thread_started("request");
ENABLE_PER_THREAD_MALLOC(__func__);
thr_self = thrman_register(THRTYPE_REQ);
logger = thrman_get_reqlogger(thr_self);
dbenv = thd->iq->dbenv;
backend_thread_event(dbenv, COMDB2_THR_EVENT_START_RDWR);
/* thdinfo is assigned to thread specific variable thd_info_key which
* will automatically free it when the thread exits. */
thdinfo = malloc(sizeof(struct thread_info));
if (thdinfo == NULL) {
logmsg(LOGMSG_FATAL, "**aborting due malloc failure thd %p\n", (void *)pthread_self());
abort();
}
thdinfo->uniquetag = 0;
thdinfo->ct_id_key = 0LL;
thdinfo->ct_add_table = create_constraint_table();
if (thdinfo->ct_add_table == NULL) {
logmsg(LOGMSG_FATAL,
"**aborting: cannot allocate constraint add table thd "
"%p\n",
(void *)pthread_self());
abort();
}
thdinfo->ct_del_table = create_constraint_table();
if (thdinfo->ct_del_table == NULL) {
logmsg(LOGMSG_FATAL,
"**aborting: cannot allocate constraint delete table "
"thd %p\n",
(void *)pthread_self());
abort();
}
thdinfo->ct_add_index = create_constraint_index_table();
if (thdinfo->ct_add_index == NULL) {
logmsg(LOGMSG_FATAL,
"**aborting: cannot allocate constraint add index table "
"thd %p\n",
(void *)pthread_self());
abort();
}
thdinfo->ct_add_table_genid_hash = hash_init(sizeof(unsigned long long));
thdinfo->ct_add_table_genid_pool =
pool_setalloc_init(sizeof(unsigned long long), 0, malloc, free);
/* Initialize the sql statement cache */
thdinfo->stmt_cache = stmt_cache_new(NULL);
if (thdinfo->stmt_cache == NULL) {
logmsg(LOGMSG_ERROR, "%s:%d failed to create sql statement cache\n",
__func__, __LINE__);
}
Pthread_setspecific(thd_info_key, thdinfo);
/*printf("started handler %ld thd %p thd->id %p\n", pthread_self(), thd,
* thd->tid);*/
do {
if (thd->tid != pthread_self()) /*sanity check*/
{
logmsg(LOGMSG_FATAL, "**aborting due thd_req mismatch thd id %p (my thd %p)\n", (void *)thd->tid,
(void *)pthread_self());
abort();
}
thd->iq->startus = comdb2_time_epochus();
thd->iq->where = "executing";
/*PROCESS REQUEST*/
thd->iq->reqlogger = logger;
iamwriter = is_req_write(thd->iq->opcode) ? 1 : 0;
dbenv = thd->iq->dbenv;
thrman_where(thr_self, req2a(thd->iq->opcode));
thrman_origin(thr_self, getorigin(thd->iq));
user_request_begin(REQUEST_TYPE_QTRAP, FLAG_REQUEST_TRACK_EVERYTHING);
handle_ireq(thd->iq);
if (debug_this_request(gbl_debug_until) ||
(gbl_who > 0 && !gbl_sdebug)) {
struct per_request_stats *st;
st = user_request_get_stats();
if (st)
logmsg(LOGMSG_USER, "nreads %d (%lld bytes) nwrites %d (%lld bytes) nfsyncs "
"%d nmempgets %d\n",
st->nreads, st->readbytes, st->nwrites, st->writebytes,
st->nfsyncs, st->mempgets);
}
thread_util_donework();
thrman_origin(thr_self, NULL);
thrman_where(thr_self, "idle");
thd->iq->where = "done executing";
// before acquiring next request, yield
comdb2bma_yield_all();
/*NEXT REQUEST*/
LOCK(&lock)
{
struct dbq_entry_t *nxtrq = NULL;
int newrqwriter = 0;
if (iamwriter) {
write_thd_count--;
}
if (thd->iq->usedb && thd->iq->ixused >= 0 &&
thd->iq->ixused < thd->iq->usedb->nix &&
thd->iq->usedb->ixuse) {
thd->iq->usedb->ixuse[thd->iq->ixused] += thd->iq->ixstepcnt;
}
thd->iq->ixused = -1;
thd->iq->ixstepcnt = 0;
if (thd->iq->dbglog_file) {
sbuf2close(thd->iq->dbglog_file);
thd->iq->dbglog_file = NULL;
}
if (thd->iq->nwrites) {
free(thd->iq->nwrites);
thd->iq->nwrites = NULL;
}
if (thd->iq->vfy_genid_hash) {
hash_free(thd->iq->vfy_genid_hash);
thd->iq->vfy_genid_hash = NULL;
}
if (thd->iq->vfy_genid_pool) {
pool_free(thd->iq->vfy_genid_pool);
thd->iq->vfy_genid_pool = NULL;
}
thd->iq->vfy_genid_track = 0;
if (thd->iq->vfy_idx_hash) {
destroy_hash(thd->iq->vfy_idx_hash, free_it);
thd->iq->vfy_idx_hash = NULL;
}
thd->iq->vfy_idx_track = 0;
thd->iq->dup_key_insert = 0;
#if 0
fprintf(stderr, "%s:%d: THD=%p relablk iq=%p\n", __func__, __LINE__, pthread_self(), thd->iq);
#endif
pool_relablk(p_reqs, thd->iq); /* this request is done, so release
* resource. */
/* get next item off hqueue */
nxtrq = (struct dbq_entry_t *)listc_rtl(&q_reqs);
thd->iq = 0;
if (nxtrq != 0) {
thd->iq = nxtrq->obj;
newrqwriter = is_req_write(thd->iq->opcode) ? 1 : 0;
numwriterthreads = gbl_maxwthreads - gbl_maxwthreadpenalty;
if (numwriterthreads < 1)
numwriterthreads = 1;
if (newrqwriter &&
(write_thd_count - iothreads) >= numwriterthreads) {
/* dont process next request as it goes over
the write limit..put it back on queue and grab
next read */
listc_atl(&q_reqs, nxtrq);
nxtrq = (struct dbq_entry_t *)listc_rtl(&rq_reqs);
if (nxtrq != NULL) {
listc_rfl(&q_reqs, nxtrq);
/* release the memory block of the link */
thd->iq = nxtrq->obj;
pool_relablk(pq_reqs, nxtrq);
newrqwriter = 0;
} else {
thd->iq = 0;
}
} else {
if (!newrqwriter) {
/*get rid of new request from read queue */
listc_rfl(&rq_reqs, nxtrq);
}
/* release the memory block of the link */
pool_relablk(pq_reqs, nxtrq);
}
if (newrqwriter && thd->iq != 0) {
write_thd_count++;
}
}
if (thd->iq == 0) {
/*wait for something to do, or go away after a while */
listc_rfl(&busy, thd);
thd_coalesce_check_ll();
listc_atl(&idle, thd);
rc = clock_gettime(CLOCK_REALTIME, &ts);
if (rc != 0) {
logmsg(LOGMSG_ERROR, "thd_req:clock_gettime bad rc %d:%s\n", rc,
strerror(errno));
memset(&ts, 0, sizeof(ts)); /*force failure later*/
}
rc = 0;
int ii = 0;
do { /* wait gbl_thd_linger seconds via one second increments */
ts.tv_sec += 1;
ii++;
/*waitft thread will deposit a request in thd->iq*/
rc = pthread_cond_timedwait(&thd->wakeup, &lock, &ts);
} while ((thd->iq == 0 && rc == 0) ||
(rc == ETIMEDOUT && ii < gbl_thd_linger && !db_is_exiting()));
if (rc != 0 && rc != ETIMEDOUT) {
logmsg(LOGMSG_ERROR, "thd_req:pthread_cond_timedwait failed:%s\n",
strerror(rc));
/* error'd out, so i still have lock: errLOCK(&lock);*/
}
if (thd->iq == 0) /*nothing to do. this thread retires.*/
{
nretire++;
listc_rfl(&idle, thd);
Pthread_cond_destroy(&thd->wakeup);
thd->tid = (pthread_t)-2; /*returned. this is just for info & debugging*/
pool_relablk(p_thds, thd); /*release this struct*/
/**/
retUNLOCK(&lock);
/**/
/*printf("ending handler %p\n", pthread_self());*/
delete_constraint_table(thdinfo->ct_add_table);
delete_constraint_table(thdinfo->ct_del_table);
delete_constraint_table(thdinfo->ct_add_index);
hash_free(thdinfo->ct_add_table_genid_hash);
if (thdinfo->ct_add_table_genid_pool) {
pool_free(thdinfo->ct_add_table_genid_pool);
}
delete_defered_index_tbl();
backend_thread_event(dbenv, COMDB2_THR_EVENT_DONE_RDWR);
return 0;
}
}
thd_coalesce_check_ll();
}
UNLOCK(&lock);
/* Should not be done under lock - might be expensive */
truncate_constraint_table(thdinfo->ct_add_table);
truncate_constraint_table(thdinfo->ct_del_table);
truncate_constraint_table(thdinfo->ct_add_index);
hash_clear(thdinfo->ct_add_table_genid_hash);
if (thdinfo->ct_add_table_genid_pool) {
pool_clear(thdinfo->ct_add_table_genid_pool);
}
truncate_defered_index_tbl();
} while (1);
}
/* sndbak error code & return resources.*/
static int reterr(intptr_t curswap, struct thd *thd, struct ireq *iq, int rc)
/* 040307dh: 64bits */
{
if (thd || iq) {
LOCK(&lock)
{
if (thd) {
if (thd->iq) {
int iamwriter = 0;
iamwriter = is_req_write(thd->iq->opcode) ? 1 : 0;
listc_rfl(&busy, thd); /*this means busy*/
thd_coalesce_check_ll();
if (iamwriter) {
write_thd_count--;
}
}
thd->iq = 0;
thd->tid = (pthread_t)-1;
pool_relablk(p_thds, thd);
}
if (iq) {
if (iq->is_fromsocket) {
if (iq->is_socketrequest) {
sndbak_open_socket(iq->sb, NULL, 0, ERR_INTERNAL);
} else {
sndbak_socket(iq->sb, NULL, 0, ERR_INTERNAL);
iq->sb = NULL;
}
}
pool_relablk(p_reqs, iq);
}
}
UNLOCK(&lock);
}
if (comdb2_ipc_sndbak && curswap) {
/* curswap is just a pointer to the buffer */
int *ibuf = (int *)curswap;
ibuf += 2;
comdb2_ipc_sndbak(ibuf, ERR_INTERNAL);
}
if (rc == ERR_INTERNAL) /*qfull hits this code too, so differentiate*/
nerrs++;
return rc;
}
static int reterr_withfree(struct ireq *iq, int rc)
{
if (iq->is_fromsocket || iq->sorese) {
if (iq->is_fromsocket) {
/* process socket end request */
if (iq->is_socketrequest) {
sndbak_open_socket(iq->sb, NULL, 0, rc);
free_bigbuf(iq->p_buf_out_start, iq->request_data);
iq->request_data = iq->p_buf_out_start = NULL;
} else {
sndbak_socket(iq->sb, NULL, 0, rc);
iq->sb = NULL;
}
} else {
/* we don't do this anymore for sorese requests */
abort();
}
if (iq->p_buf_out_start) {
free(iq->p_buf_out_start);
}
iq->p_buf_out_end = iq->p_buf_out_start = iq->p_buf_out = NULL;
iq->p_buf_in_end = iq->p_buf_in = NULL;
LOCK(&lock)
{
#if 0
fprintf(stderr, "%s:%d: THD=%p relablk iq=%p\n", __func__, __LINE__, pthread_self(), iq);
#endif
pool_relablk(p_reqs, iq);
}
UNLOCK(&lock);
return 0;
} else {
return reterr(iq->curswap, NULL, iq, rc);
}
}
int handle_buf_block_offload(struct dbenv *dbenv, uint8_t *p_buf,
const uint8_t *p_buf_end, int debug,
char *frommach, unsigned long long rqid)
{
int length = p_buf_end - p_buf;
uint8_t *p_bigbuf = get_bigbuf();
memcpy(p_bigbuf, p_buf, length);
int rc = handle_buf_main(dbenv, NULL, p_bigbuf, p_bigbuf + length, debug,
frommach, 0, NULL, NULL, REQ_SOCKREQUEST, NULL, 0,
rqid);
return rc;
}
int handle_socket_long_transaction(struct dbenv *dbenv, SBUF2 *sb,
uint8_t *p_buf, const uint8_t *p_buf_end,
int debug, char *frommach, int frompid,
char *fromtask)
{
return handle_buf_main(dbenv, sb, p_buf, p_buf_end, debug, frommach,
frompid, fromtask, NULL, REQ_SOCKET, NULL, 0, 0);
}
void cleanup_lock_buffer(struct buf_lock_t *lock_buffer)
{
if (lock_buffer == NULL)
return;
/* sbuf2 is owned by the appsock. Don't close it here. */
Pthread_cond_destroy(&lock_buffer->wait_cond);
Pthread_mutex_destroy(&lock_buffer->req_lock);
LOCK(&buf_lock)
{
if (lock_buffer->bigbuf != NULL)
pool_relablk(p_bufs, lock_buffer->bigbuf);
pool_relablk(p_slocks, lock_buffer);
}
UNLOCK(&buf_lock);
}
/* handle a buffer from waitft */
int handle_buf(struct dbenv *dbenv, uint8_t *p_buf, const uint8_t *p_buf_end,
int debug, char *frommach) /* 040307dh: 64bits */
{
return handle_buf_main(dbenv, NULL, p_buf, p_buf_end, debug, frommach, 0,
NULL, NULL, REQ_WAITFT, NULL, 0, 0);
}
int handled_queue;
int q_reqs_len(void) { return q_reqs.count; }
static int init_ireq_legacy(struct dbenv *dbenv, struct ireq *iq, SBUF2 *sb,
uint8_t *p_buf, const uint8_t *p_buf_end, int debug,
char *frommach, int frompid, char *fromtask,
osql_sess_t *sorese, int qtype, void *data_hndl,
int luxref, unsigned long long rqid, void *p_sinfo,
intptr_t curswap, int comdbg_flags)
{
struct req_hdr hdr;
uint64_t nowus;
nowus = comdb2_time_epochus();
/* set up request */
const size_t len = sizeof(*iq) - offsetof(struct ireq, region3);
bzero(&iq->region3, len);
iq->corigin[0] = '\0';
iq->debug_buf[0] = '\0';
iq->tzname[0] = '\0';
iq->where = "setup";
iq->frommach = frommach ? intern(frommach) : NULL;
iq->frompid = frompid;
iq->gluewhere = "-";
iq->debug = debug_this_request(gbl_debug_until) || (debug && gbl_debug);
iq->debug_now = iq->nowus = nowus;
iq->dbenv = dbenv;
iq->fwd_tag_rqid = rqid;
iq->p_buf_orig =
p_buf; /* need this for optimized fast fail (skip blkstate) */
iq->p_buf_in = p_buf;
iq->p_buf_in_end = p_buf_end;
iq->p_buf_out = p_buf + REQ_HDR_LEN;
iq->p_buf_out_start = p_buf;
iq->p_buf_out_end = p_buf_end - RESERVED_SZ;
/* IPC stuff */
iq->p_sinfo = p_sinfo;
iq->curswap = curswap;
iq->comdbg_flags = comdbg_flags;
/* HERE: unpack and get the proper lux - req_hdr_get_and_fixup_lux */
if (!(iq->p_buf_in = req_hdr_get(&hdr, iq->p_buf_in, iq->p_buf_in_end, iq->comdbg_flags))) {
logmsg(LOGMSG_ERROR, "handle_buf:failed to unpack req header\n");
return ERR_BADREQ;
}
iq->opcode = hdr.opcode;
if (qtype == REQ_PQREQUEST) {
iq->is_fake = 1;
iq->is_dumpresponse = 1;
iq->request_data = data_hndl;
}
if (qtype == REQ_SOCKET || qtype == REQ_SOCKREQUEST) {
iq->sb = sb;
iq->is_fromsocket = 1;
iq->is_socketrequest = (qtype == REQ_SOCKREQUEST) ? 1 : 0;
} else {
iq->sb = NULL;
}
if (iq->is_socketrequest) {
iq->request_data = data_hndl;
}
iq->__limits.maxcost = gbl_querylimits_maxcost;
iq->__limits.tablescans_ok = gbl_querylimits_tablescans_ok;
iq->__limits.temptables_ok = gbl_querylimits_temptables_ok;
iq->__limits.maxcost_warn = gbl_querylimits_maxcost_warn;
iq->__limits.tablescans_warn = gbl_querylimits_tablescans_warn;
iq->__limits.temptables_warn = gbl_querylimits_temptables_warn;
iq->cost = 0;
iq->luxref = luxref;
if (iq->is_fromsocket) {
if (iq->frommach == gbl_myhostname)
snprintf(iq->corigin, sizeof(iq->corigin), "SLCL %.8s PID %6.6d",
fromtask ? fromtask : "null", frompid);
else
snprintf(iq->corigin, sizeof(iq->corigin), "SRMT# %s PID %6.6d",
iq->frommach, frompid);
} else if (sorese) {
iq->sorese = sorese;
snprintf(iq->corigin, sizeof(iq->corigin), "SORESE# %15s %s RQST %llx",
iq->sorese->target.host,
osql_sorese_type_to_str(iq->sorese->type), iq->sorese->rqid);
iq->timings.req_received = osql_log_time();
/* cache these things so we don't change too much code */
iq->tranddl = iq->sorese->is_tranddl;
iq->tptlock = iq->sorese->is_tptlock;
iq->sorese->iq = iq;
if (!iq->debug) {
if (gbl_who > 0) {
gbl_who--;
iq->debug = 1;
}
}
}
if (dbenv->num_dbs > 0 && (luxref < 0 || luxref >= dbenv->num_dbs)) {
logmsg(LOGMSG_ERROR, "handle_buf:luxref out of range %d max %d\n",
luxref, dbenv->num_dbs);
return ERR_REJECTED;