aboutsummaryrefslogtreecommitdiffstats
path: root/contrib/libs/jemalloc/src/tcache.c
blob: 2c574e8f48f0a04cb9431921e278747cf591db46 (plain) (blame)
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
#define JEMALLOC_TCACHE_C_ 
#include "jemalloc/internal/jemalloc_preamble.h" 
#include "jemalloc/internal/jemalloc_internal_includes.h" 

#include "jemalloc/internal/assert.h" 
#include "jemalloc/internal/mutex.h" 
#include "jemalloc/internal/safety_check.h" 
#include "jemalloc/internal/sc.h" 
 
/******************************************************************************/
/* Data. */

bool	opt_tcache = true;
ssize_t	opt_lg_tcache_max = LG_TCACHE_MAXCLASS_DEFAULT;

cache_bin_info_t	*tcache_bin_info; 
static unsigned		stack_nelms; /* Total stack elms per tcache. */

unsigned		nhbins; 
size_t			tcache_maxclass;

tcaches_t		*tcaches; 
 
/* Index of first element within tcaches that has never been used. */ 
static unsigned		tcaches_past; 
 
/* Head of singly linked list tracking available tcaches elements. */ 
static tcaches_t	*tcaches_avail; 
 
/* Protects tcaches{,_past,_avail}. */ 
static malloc_mutex_t	tcaches_mtx; 
 
/******************************************************************************/

size_t 
tcache_salloc(tsdn_t *tsdn, const void *ptr) { 
	return arena_salloc(tsdn, ptr); 
}

void
tcache_event_hard(tsd_t *tsd, tcache_t *tcache) { 
	szind_t binind = tcache->next_gc_bin; 

	cache_bin_t *tbin; 
	if (binind < SC_NBINS) { 
		tbin = tcache_small_bin_get(tcache, binind); 
	} else { 
		tbin = tcache_large_bin_get(tcache, binind); 
	} 
	if (tbin->low_water > 0) {
		/*
		 * Flush (ceiling) 3/4 of the objects below the low water mark.
		 */
		if (binind < SC_NBINS) { 
			tcache_bin_flush_small(tsd, tcache, tbin, binind, 
			    tbin->ncached - tbin->low_water + (tbin->low_water 
			    >> 2)); 
			/* 
			 * Reduce fill count by 2X.  Limit lg_fill_div such that 
			 * the fill count is always at least 1. 
			 */ 
			cache_bin_info_t *tbin_info = &tcache_bin_info[binind]; 
			if ((tbin_info->ncached_max >> 
			     (tcache->lg_fill_div[binind] + 1)) >= 1) { 
				tcache->lg_fill_div[binind]++; 
			} 
		} else {
			tcache_bin_flush_large(tsd, tbin, binind, tbin->ncached 
			    - tbin->low_water + (tbin->low_water >> 2), tcache); 
		}
	} else if (tbin->low_water < 0) {
		/*
		 * Increase fill count by 2X for small bins.  Make sure 
		 * lg_fill_div stays greater than 0. 
		 */
		if (binind < SC_NBINS && tcache->lg_fill_div[binind] > 1) { 
			tcache->lg_fill_div[binind]--; 
		} 
	}
	tbin->low_water = tbin->ncached;

	tcache->next_gc_bin++;
	if (tcache->next_gc_bin == nhbins) { 
		tcache->next_gc_bin = 0;
	} 
}

void *
tcache_alloc_small_hard(tsdn_t *tsdn, arena_t *arena, tcache_t *tcache, 
    cache_bin_t *tbin, szind_t binind, bool *tcache_success) { 
	void *ret;

	assert(tcache->arena != NULL); 
	arena_tcache_fill_small(tsdn, arena, tcache, tbin, binind, 
	    config_prof ? tcache->prof_accumbytes : 0);
	if (config_prof) { 
		tcache->prof_accumbytes = 0;
	} 
	ret = cache_bin_alloc_easy(tbin, tcache_success); 

	return ret; 
}

/* Enabled with --enable-extra-size-check. */ 
static void 
tbin_extents_lookup_size_check(tsdn_t *tsdn, cache_bin_t *tbin, szind_t binind, 
    size_t nflush, extent_t **extents){ 
	rtree_ctx_t rtree_ctx_fallback; 
	rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn, &rtree_ctx_fallback); 
 
	/* 
	 * Verify that the items in the tcache all have the correct size; this 
	 * is useful for catching sized deallocation bugs, also to fail early 
	 * instead of corrupting metadata.  Since this can be turned on for opt 
	 * builds, avoid the branch in the loop. 
	 */ 
	szind_t szind; 
	size_t sz_sum = binind * nflush; 
	for (unsigned i = 0 ; i < nflush; i++) { 
		rtree_extent_szind_read(tsdn, &extents_rtree, 
		    rtree_ctx, (uintptr_t)*(tbin->avail - 1 - i), true, 
		    &extents[i], &szind); 
		sz_sum -= szind; 
	} 
	if (sz_sum != 0) { 
		safety_check_fail("<jemalloc>: size mismatch in thread cache " 
		    "detected, likely caused by sized deallocation bugs by " 
		    "application. Abort.\n"); 
		abort(); 
	} 
} 
 
void
tcache_bin_flush_small(tsd_t *tsd, tcache_t *tcache, cache_bin_t *tbin, 
    szind_t binind, unsigned rem) { 
	bool merged_stats = false;

	assert(binind < SC_NBINS); 
	assert((cache_bin_sz_t)rem <= tbin->ncached); 

	arena_t *arena = tcache->arena; 
	assert(arena != NULL); 
	unsigned nflush = tbin->ncached - rem; 
	VARIABLE_ARRAY(extent_t *, item_extent, nflush); 
 
	/* Look up extent once per item. */ 
	if (config_opt_safety_checks) { 
		tbin_extents_lookup_size_check(tsd_tsdn(tsd), tbin, binind, 
		    nflush, item_extent); 
	} else { 
		for (unsigned i = 0 ; i < nflush; i++) { 
			item_extent[i] = iealloc(tsd_tsdn(tsd), 
			    *(tbin->avail - 1 - i)); 
		} 
	} 
	while (nflush > 0) { 
		/* Lock the arena bin associated with the first object. */
		extent_t *extent = item_extent[0]; 
		unsigned bin_arena_ind = extent_arena_ind_get(extent); 
		arena_t *bin_arena = arena_get(tsd_tsdn(tsd), bin_arena_ind, 
		    false); 
		unsigned binshard = extent_binshard_get(extent); 
		assert(binshard < bin_infos[binind].n_shards); 
		bin_t *bin = &bin_arena->bins[binind].bin_shards[binshard]; 

		if (config_prof && bin_arena == arena) { 
			if (arena_prof_accum(tsd_tsdn(tsd), arena, 
			    tcache->prof_accumbytes)) { 
				prof_idump(tsd_tsdn(tsd)); 
			} 
			tcache->prof_accumbytes = 0;
		}

		malloc_mutex_lock(tsd_tsdn(tsd), &bin->lock); 
		if (config_stats && bin_arena == arena && !merged_stats) { 
			merged_stats = true;
			bin->stats.nflushes++;
			bin->stats.nrequests += tbin->tstats.nrequests;
			tbin->tstats.nrequests = 0;
		}
		unsigned ndeferred = 0; 
		for (unsigned i = 0; i < nflush; i++) { 
			void *ptr = *(tbin->avail - 1 - i); 
			extent = item_extent[i]; 
			assert(ptr != NULL && extent != NULL); 
 
			if (extent_arena_ind_get(extent) == bin_arena_ind 
			    && extent_binshard_get(extent) == binshard) { 
				arena_dalloc_bin_junked_locked(tsd_tsdn(tsd), 
				    bin_arena, bin, binind, extent, ptr); 
			} else {
				/*
				 * This object was allocated via a different
				 * arena bin than the one that is currently
				 * locked.  Stash the object, so that it can be
				 * handled in a future pass.
				 */
				*(tbin->avail - 1 - ndeferred) = ptr; 
				item_extent[ndeferred] = extent; 
				ndeferred++;
			}
		}
		malloc_mutex_unlock(tsd_tsdn(tsd), &bin->lock); 
		arena_decay_ticks(tsd_tsdn(tsd), bin_arena, nflush - ndeferred); 
		nflush = ndeferred; 
	}
	if (config_stats && !merged_stats) { 
		/*
		 * The flush loop didn't happen to flush to this thread's
		 * arena, so the stats didn't get merged.  Manually do so now.
		 */
		unsigned binshard; 
		bin_t *bin = arena_bin_choose_lock(tsd_tsdn(tsd), arena, binind, 
		    &binshard); 
		bin->stats.nflushes++;
		bin->stats.nrequests += tbin->tstats.nrequests;
		tbin->tstats.nrequests = 0;
		malloc_mutex_unlock(tsd_tsdn(tsd), &bin->lock); 
	}

	memmove(tbin->avail - rem, tbin->avail - tbin->ncached, rem * 
	    sizeof(void *)); 
	tbin->ncached = rem;
	if (tbin->ncached < tbin->low_water) { 
		tbin->low_water = tbin->ncached;
	} 
}

void
tcache_bin_flush_large(tsd_t *tsd, cache_bin_t *tbin, szind_t binind, 
    unsigned rem, tcache_t *tcache) { 
	bool merged_stats = false;

	assert(binind < nhbins);
	assert((cache_bin_sz_t)rem <= tbin->ncached); 

	arena_t *tcache_arena = tcache->arena; 
	assert(tcache_arena != NULL); 
	unsigned nflush = tbin->ncached - rem; 
	VARIABLE_ARRAY(extent_t *, item_extent, nflush); 
 
#ifndef JEMALLOC_EXTRA_SIZE_CHECK 
	/* Look up extent once per item. */ 
	for (unsigned i = 0 ; i < nflush; i++) { 
		item_extent[i] = iealloc(tsd_tsdn(tsd), *(tbin->avail - 1 - i)); 
	} 
#else 
	tbin_extents_lookup_size_check(tsd_tsdn(tsd), tbin, binind, nflush, 
	    item_extent); 
#endif 
	while (nflush > 0) { 
		/* Lock the arena associated with the first object. */
		extent_t *extent = item_extent[0]; 
		unsigned locked_arena_ind = extent_arena_ind_get(extent); 
		arena_t *locked_arena = arena_get(tsd_tsdn(tsd), 
		    locked_arena_ind, false); 
		bool idump; 

		if (config_prof) { 
			idump = false;
		} 
 
		bool lock_large = !arena_is_auto(locked_arena); 
		if (lock_large) { 
			malloc_mutex_lock(tsd_tsdn(tsd), &locked_arena->large_mtx); 
		} 
		for (unsigned i = 0; i < nflush; i++) { 
			void *ptr = *(tbin->avail - 1 - i); 
			assert(ptr != NULL); 
			extent = item_extent[i]; 
			if (extent_arena_ind_get(extent) == locked_arena_ind) { 
				large_dalloc_prep_junked_locked(tsd_tsdn(tsd), 
				    extent); 
			} 
		} 
		if ((config_prof || config_stats) && 
		    (locked_arena == tcache_arena)) { 
			if (config_prof) {
				idump = arena_prof_accum(tsd_tsdn(tsd), 
				    tcache_arena, tcache->prof_accumbytes); 
				tcache->prof_accumbytes = 0;
			}
			if (config_stats) {
				merged_stats = true;
				arena_stats_large_flush_nrequests_add( 
				    tsd_tsdn(tsd), &tcache_arena->stats, binind, 
				    tbin->tstats.nrequests); 
				tbin->tstats.nrequests = 0;
			}
		}
		if (lock_large) { 
			malloc_mutex_unlock(tsd_tsdn(tsd), &locked_arena->large_mtx); 
		} 
 
		unsigned ndeferred = 0; 
		for (unsigned i = 0; i < nflush; i++) { 
			void *ptr = *(tbin->avail - 1 - i); 
			extent = item_extent[i]; 
			assert(ptr != NULL && extent != NULL); 
 
			if (extent_arena_ind_get(extent) == locked_arena_ind) { 
				large_dalloc_finish(tsd_tsdn(tsd), extent); 
			} else { 
				/*
				 * This object was allocated via a different
				 * arena than the one that is currently locked.
				 * Stash the object, so that it can be handled
				 * in a future pass.
				 */
				*(tbin->avail - 1 - ndeferred) = ptr; 
				item_extent[ndeferred] = extent; 
				ndeferred++;
			}
		}
		if (config_prof && idump) { 
			prof_idump(tsd_tsdn(tsd)); 
		} 
		arena_decay_ticks(tsd_tsdn(tsd), locked_arena, nflush - 
		    ndeferred); 
		nflush = ndeferred; 
	}
	if (config_stats && !merged_stats) { 
		/*
		 * The flush loop didn't happen to flush to this thread's
		 * arena, so the stats didn't get merged.  Manually do so now.
		 */
		arena_stats_large_flush_nrequests_add(tsd_tsdn(tsd), 
		    &tcache_arena->stats, binind, tbin->tstats.nrequests); 
		tbin->tstats.nrequests = 0;
	}

	memmove(tbin->avail - rem, tbin->avail - tbin->ncached, rem * 
	    sizeof(void *)); 
	tbin->ncached = rem;
	if (tbin->ncached < tbin->low_water) { 
		tbin->low_water = tbin->ncached;
	} 
}

void
tcache_arena_associate(tsdn_t *tsdn, tcache_t *tcache, arena_t *arena) { 
	assert(tcache->arena == NULL); 
	tcache->arena = arena; 

	if (config_stats) {
		/* Link into list of extant tcaches. */
		malloc_mutex_lock(tsdn, &arena->tcache_ql_mtx); 
 
		ql_elm_new(tcache, link);
		ql_tail_insert(&arena->tcache_ql, tcache, link);
		cache_bin_array_descriptor_init( 
		    &tcache->cache_bin_array_descriptor, tcache->bins_small, 
		    tcache->bins_large); 
		ql_tail_insert(&arena->cache_bin_array_descriptor_ql, 
		    &tcache->cache_bin_array_descriptor, link); 
 
		malloc_mutex_unlock(tsdn, &arena->tcache_ql_mtx); 
	}
}

static void 
tcache_arena_dissociate(tsdn_t *tsdn, tcache_t *tcache) { 
	arena_t *arena = tcache->arena; 
	assert(arena != NULL); 
	if (config_stats) {
		/* Unlink from list of extant tcaches. */
		malloc_mutex_lock(tsdn, &arena->tcache_ql_mtx); 
		if (config_debug) { 
			bool in_ql = false; 
			tcache_t *iter; 
			ql_foreach(iter, &arena->tcache_ql, link) { 
				if (iter == tcache) { 
					in_ql = true; 
					break; 
				} 
			} 
			assert(in_ql); 
		} 
		ql_remove(&arena->tcache_ql, tcache, link); 
		ql_remove(&arena->cache_bin_array_descriptor_ql, 
		    &tcache->cache_bin_array_descriptor, link); 
		tcache_stats_merge(tsdn, tcache, arena); 
		malloc_mutex_unlock(tsdn, &arena->tcache_ql_mtx); 
	}
	tcache->arena = NULL; 
}

void 
tcache_arena_reassociate(tsdn_t *tsdn, tcache_t *tcache, arena_t *arena) { 
	tcache_arena_dissociate(tsdn, tcache); 
	tcache_arena_associate(tsdn, tcache, arena); 
} 
 
bool 
tsd_tcache_enabled_data_init(tsd_t *tsd) { 
	/* Called upon tsd initialization. */ 
	tsd_tcache_enabled_set(tsd, opt_tcache); 
	tsd_slow_update(tsd); 
 
	if (opt_tcache) { 
		/* Trigger tcache init. */ 
		tsd_tcache_data_init(tsd); 
	} 
 
	return false; 
} 
 
/* Initialize auto tcache (embedded in TSD). */ 
static void 
tcache_init(tsd_t *tsd, tcache_t *tcache, void *avail_stack) { 
	memset(&tcache->link, 0, sizeof(ql_elm(tcache_t))); 
	tcache->prof_accumbytes = 0; 
	tcache->next_gc_bin = 0; 
	tcache->arena = NULL; 
 
	ticker_init(&tcache->gc_ticker, TCACHE_GC_INCR); 
 
	size_t stack_offset = 0; 
	assert((TCACHE_NSLOTS_SMALL_MAX & 1U) == 0); 
	memset(tcache->bins_small, 0, sizeof(cache_bin_t) * SC_NBINS); 
	memset(tcache->bins_large, 0, sizeof(cache_bin_t) * (nhbins - SC_NBINS)); 
	unsigned i = 0; 
	for (; i < SC_NBINS; i++) { 
		tcache->lg_fill_div[i] = 1; 
		stack_offset += tcache_bin_info[i].ncached_max * sizeof(void *); 
		/* 
		 * avail points past the available space.  Allocations will 
		 * access the slots toward higher addresses (for the benefit of 
		 * prefetch). 
		 */ 
		tcache_small_bin_get(tcache, i)->avail = 
		    (void **)((uintptr_t)avail_stack + (uintptr_t)stack_offset); 
	} 
	for (; i < nhbins; i++) { 
		stack_offset += tcache_bin_info[i].ncached_max * sizeof(void *); 
		tcache_large_bin_get(tcache, i)->avail = 
		    (void **)((uintptr_t)avail_stack + (uintptr_t)stack_offset); 
	} 
	assert(stack_offset == stack_nelms * sizeof(void *)); 
} 
 
/* Initialize auto tcache (embedded in TSD). */ 
bool 
tsd_tcache_data_init(tsd_t *tsd) { 
	tcache_t *tcache = tsd_tcachep_get_unsafe(tsd); 
	assert(tcache_small_bin_get(tcache, 0)->avail == NULL); 
	size_t size = stack_nelms * sizeof(void *); 
	/* Avoid false cacheline sharing. */ 
	size = sz_sa2u(size, CACHELINE); 
 
	void *avail_array = ipallocztm(tsd_tsdn(tsd), size, CACHELINE, true, 
	    NULL, true, arena_get(TSDN_NULL, 0, true)); 
	if (avail_array == NULL) { 
		return true; 
	} 
 
	tcache_init(tsd, tcache, avail_array); 
	/* 
	 * Initialization is a bit tricky here.  After malloc init is done, all 
	 * threads can rely on arena_choose and associate tcache accordingly. 
	 * However, the thread that does actual malloc bootstrapping relies on 
	 * functional tsd, and it can only rely on a0.  In that case, we 
	 * associate its tcache to a0 temporarily, and later on 
	 * arena_choose_hard() will re-associate properly. 
	 */ 
	tcache->arena = NULL; 
	arena_t *arena; 
	if (!malloc_initialized()) { 
		/* If in initialization, assign to a0. */ 
		arena = arena_get(tsd_tsdn(tsd), 0, false); 
		tcache_arena_associate(tsd_tsdn(tsd), tcache, arena); 
	} else { 
		arena = arena_choose(tsd, NULL); 
		/* This may happen if thread.tcache.enabled is used. */ 
		if (tcache->arena == NULL) { 
			tcache_arena_associate(tsd_tsdn(tsd), tcache, arena); 
		} 
	} 
	assert(arena == tcache->arena); 
 
	return false; 
} 
 
/* Created manual tcache for tcache.create mallctl. */ 
tcache_t *
tcache_create_explicit(tsd_t *tsd) { 
	tcache_t *tcache;
	size_t size, stack_offset;

	size = sizeof(tcache_t); 
	/* Naturally align the pointer stacks. */
	size = PTR_CEILING(size);
	stack_offset = size;
	size += stack_nelms * sizeof(void *);
	/* Avoid false cacheline sharing. */ 
	size = sz_sa2u(size, CACHELINE); 

	tcache = ipallocztm(tsd_tsdn(tsd), size, CACHELINE, true, NULL, true, 
	    arena_get(TSDN_NULL, 0, true)); 
	if (tcache == NULL) { 
		return NULL; 
	}

	tcache_init(tsd, tcache, 
	    (void *)((uintptr_t)tcache + (uintptr_t)stack_offset)); 
	tcache_arena_associate(tsd_tsdn(tsd), tcache, arena_ichoose(tsd, NULL)); 

	return tcache; 
}

static void 
tcache_flush_cache(tsd_t *tsd, tcache_t *tcache) { 
	assert(tcache->arena != NULL); 

	for (unsigned i = 0; i < SC_NBINS; i++) { 
		cache_bin_t *tbin = tcache_small_bin_get(tcache, i); 
		tcache_bin_flush_small(tsd, tcache, tbin, i, 0); 

		if (config_stats) { 
			assert(tbin->tstats.nrequests == 0); 
		}
	}
	for (unsigned i = SC_NBINS; i < nhbins; i++) { 
		cache_bin_t *tbin = tcache_large_bin_get(tcache, i); 
		tcache_bin_flush_large(tsd, tbin, i, 0, tcache); 

		if (config_stats) { 
			assert(tbin->tstats.nrequests == 0); 
		}
	}

	if (config_prof && tcache->prof_accumbytes > 0 &&
	    arena_prof_accum(tsd_tsdn(tsd), tcache->arena, 
	    tcache->prof_accumbytes)) { 
		prof_idump(tsd_tsdn(tsd)); 
	} 
} 

void 
tcache_flush(tsd_t *tsd) { 
	assert(tcache_available(tsd)); 
	tcache_flush_cache(tsd, tsd_tcachep_get(tsd)); 
} 

static void 
tcache_destroy(tsd_t *tsd, tcache_t *tcache, bool tsd_tcache) { 
	tcache_flush_cache(tsd, tcache); 
	arena_t *arena = tcache->arena; 
	tcache_arena_dissociate(tsd_tsdn(tsd), tcache); 

	if (tsd_tcache) { 
		/* Release the avail array for the TSD embedded auto tcache. */ 
		void *avail_array = 
		    (void *)((uintptr_t)tcache_small_bin_get(tcache, 0)->avail - 
		    (uintptr_t)tcache_bin_info[0].ncached_max * sizeof(void *)); 
		idalloctm(tsd_tsdn(tsd), avail_array, NULL, NULL, true, true); 
	} else { 
		/* Release both the tcache struct and avail array. */ 
		idalloctm(tsd_tsdn(tsd), tcache, NULL, NULL, true, true); 
	} 
 
	/* 
	 * The deallocation and tcache flush above may not trigger decay since 
	 * we are on the tcache shutdown path (potentially with non-nominal 
	 * tsd).  Manually trigger decay to avoid pathological cases.  Also 
	 * include arena 0 because the tcache array is allocated from it. 
	 */ 
	arena_decay(tsd_tsdn(tsd), arena_get(tsd_tsdn(tsd), 0, false), 
	    false, false); 
 
	if (arena_nthreads_get(arena, false) == 0 && 
	    !background_thread_enabled()) { 
		/* Force purging when no threads assigned to the arena anymore. */ 
		arena_decay(tsd_tsdn(tsd), arena, false, true); 
	} else { 
		arena_decay(tsd_tsdn(tsd), arena, false, false); 
	} 
}

/* For auto tcache (embedded in TSD) only. */ 
void
tcache_cleanup(tsd_t *tsd) { 
	tcache_t *tcache = tsd_tcachep_get(tsd); 
	if (!tcache_available(tsd)) { 
		assert(tsd_tcache_enabled_get(tsd) == false); 
		if (config_debug) { 
			assert(tcache_small_bin_get(tcache, 0)->avail == NULL); 
		} 
		return; 
	} 
	assert(tsd_tcache_enabled_get(tsd)); 
	assert(tcache_small_bin_get(tcache, 0)->avail != NULL); 

	tcache_destroy(tsd, tcache, true); 
	if (config_debug) { 
		tcache_small_bin_get(tcache, 0)->avail = NULL; 
	}
}

void
tcache_stats_merge(tsdn_t *tsdn, tcache_t *tcache, arena_t *arena) { 
	unsigned i;

	cassert(config_stats);

	/* Merge and reset tcache stats. */
	for (i = 0; i < SC_NBINS; i++) { 
		cache_bin_t *tbin = tcache_small_bin_get(tcache, i); 
		unsigned binshard; 
		bin_t *bin = arena_bin_choose_lock(tsdn, arena, i, &binshard); 
		bin->stats.nrequests += tbin->tstats.nrequests;
		malloc_mutex_unlock(tsdn, &bin->lock); 
		tbin->tstats.nrequests = 0;
	}

	for (; i < nhbins; i++) {
		cache_bin_t *tbin = tcache_large_bin_get(tcache, i); 
		arena_stats_large_flush_nrequests_add(tsdn, &arena->stats, i, 
		    tbin->tstats.nrequests); 
		tbin->tstats.nrequests = 0;
	}
}

static bool 
tcaches_create_prep(tsd_t *tsd) { 
	bool err; 
 
	malloc_mutex_lock(tsd_tsdn(tsd), &tcaches_mtx); 
 
	if (tcaches == NULL) { 
		tcaches = base_alloc(tsd_tsdn(tsd), b0get(), sizeof(tcache_t *) 
		    * (MALLOCX_TCACHE_MAX+1), CACHELINE); 
		if (tcaches == NULL) { 
			err = true; 
			goto label_return; 
		} 
	} 
 
	if (tcaches_avail == NULL && tcaches_past > MALLOCX_TCACHE_MAX) { 
		err = true; 
		goto label_return; 
	} 
 
	err = false; 
label_return: 
	malloc_mutex_unlock(tsd_tsdn(tsd), &tcaches_mtx); 
	return err; 
} 
 
bool
tcaches_create(tsd_t *tsd, unsigned *r_ind) { 
	witness_assert_depth(tsdn_witness_tsdp_get(tsd_tsdn(tsd)), 0); 

	bool err; 

	if (tcaches_create_prep(tsd)) { 
		err = true; 
		goto label_return; 
	} 

	tcache_t *tcache = tcache_create_explicit(tsd); 
	if (tcache == NULL) { 
		err = true; 
		goto label_return; 
	} 
 
	tcaches_t *elm; 
	malloc_mutex_lock(tsd_tsdn(tsd), &tcaches_mtx); 
	if (tcaches_avail != NULL) { 
		elm = tcaches_avail; 
		tcaches_avail = tcaches_avail->next; 
		elm->tcache = tcache; 
		*r_ind = (unsigned)(elm - tcaches); 
	} else { 
		elm = &tcaches[tcaches_past]; 
		elm->tcache = tcache; 
		*r_ind = tcaches_past; 
		tcaches_past++; 
	} 
	malloc_mutex_unlock(tsd_tsdn(tsd), &tcaches_mtx); 
 
	err = false; 
label_return: 
	witness_assert_depth(tsdn_witness_tsdp_get(tsd_tsdn(tsd)), 0); 
	return err; 
} 
 
static tcache_t * 
tcaches_elm_remove(tsd_t *tsd, tcaches_t *elm, bool allow_reinit) { 
	malloc_mutex_assert_owner(tsd_tsdn(tsd), &tcaches_mtx); 
 
	if (elm->tcache == NULL) { 
		return NULL; 
	} 
	tcache_t *tcache = elm->tcache; 
	if (allow_reinit) { 
		elm->tcache = TCACHES_ELM_NEED_REINIT; 
	} else { 
		elm->tcache = NULL; 
	} 
 
	if (tcache == TCACHES_ELM_NEED_REINIT) { 
		return NULL; 
	} 
	return tcache; 
} 
 
void 
tcaches_flush(tsd_t *tsd, unsigned ind) { 
	malloc_mutex_lock(tsd_tsdn(tsd), &tcaches_mtx); 
	tcache_t *tcache = tcaches_elm_remove(tsd, &tcaches[ind], true); 
	malloc_mutex_unlock(tsd_tsdn(tsd), &tcaches_mtx); 
	if (tcache != NULL) { 
		/* Destroy the tcache; recreate in tcaches_get() if needed. */ 
		tcache_destroy(tsd, tcache, false); 
	} 
} 
 
void 
tcaches_destroy(tsd_t *tsd, unsigned ind) { 
	malloc_mutex_lock(tsd_tsdn(tsd), &tcaches_mtx); 
	tcaches_t *elm = &tcaches[ind]; 
	tcache_t *tcache = tcaches_elm_remove(tsd, elm, false); 
	elm->next = tcaches_avail; 
	tcaches_avail = elm; 
	malloc_mutex_unlock(tsd_tsdn(tsd), &tcaches_mtx); 
	if (tcache != NULL) { 
		tcache_destroy(tsd, tcache, false); 
	} 
} 
 
bool 
tcache_boot(tsdn_t *tsdn) { 
	/* If necessary, clamp opt_lg_tcache_max. */ 
	if (opt_lg_tcache_max < 0 || (ZU(1) << opt_lg_tcache_max) < 
	    SC_SMALL_MAXCLASS) { 
		tcache_maxclass = SC_SMALL_MAXCLASS; 
	} else { 
		tcache_maxclass = (ZU(1) << opt_lg_tcache_max); 
	} 
 
	if (malloc_mutex_init(&tcaches_mtx, "tcaches", WITNESS_RANK_TCACHES, 
	    malloc_mutex_rank_exclusive)) { 
		return true; 
	} 
 
	nhbins = sz_size2index(tcache_maxclass) + 1; 
 
	/* Initialize tcache_bin_info. */
	tcache_bin_info = (cache_bin_info_t *)base_alloc(tsdn, b0get(), nhbins 
	    * sizeof(cache_bin_info_t), CACHELINE); 
	if (tcache_bin_info == NULL) { 
		return true; 
	} 
	stack_nelms = 0;
	unsigned i; 
	for (i = 0; i < SC_NBINS; i++) { 
		if ((bin_infos[i].nregs << 1) <= TCACHE_NSLOTS_SMALL_MIN) { 
			tcache_bin_info[i].ncached_max =
			    TCACHE_NSLOTS_SMALL_MIN; 
		} else if ((bin_infos[i].nregs << 1) <= 
		    TCACHE_NSLOTS_SMALL_MAX) { 
			tcache_bin_info[i].ncached_max = 
			    (bin_infos[i].nregs << 1); 
		} else {
			tcache_bin_info[i].ncached_max =
			    TCACHE_NSLOTS_SMALL_MAX;
		}
		stack_nelms += tcache_bin_info[i].ncached_max;
	}
	for (; i < nhbins; i++) {
		tcache_bin_info[i].ncached_max = TCACHE_NSLOTS_LARGE;
		stack_nelms += tcache_bin_info[i].ncached_max;
	}

	return false; 
}

void 
tcache_prefork(tsdn_t *tsdn) { 
	if (!config_prof && opt_tcache) { 
		malloc_mutex_prefork(tsdn, &tcaches_mtx); 
	} 
} 

void 
tcache_postfork_parent(tsdn_t *tsdn) { 
	if (!config_prof && opt_tcache) { 
		malloc_mutex_postfork_parent(tsdn, &tcaches_mtx); 
	} 
} 

void 
tcache_postfork_child(tsdn_t *tsdn) { 
	if (!config_prof && opt_tcache) { 
		malloc_mutex_postfork_child(tsdn, &tcaches_mtx); 
	} 
}