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
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
|
// Copyright 2009 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
// Malloc profiling.
// Patterned after tcmalloc's algorithms; shorter code.
package runtime
import (
"internal/abi"
"runtime/internal/atomic"
"runtime/internal/sys"
"unsafe"
)
// NOTE(rsc): Everything here could use cas if contention became an issue.
var (
// profInsertLock protects changes to the start of all *bucket linked lists
profInsertLock mutex
// profBlockLock protects the contents of every blockRecord struct
profBlockLock mutex
// profMemActiveLock protects the active field of every memRecord struct
profMemActiveLock mutex
// profMemFutureLock is a set of locks that protect the respective elements
// of the future array of every memRecord struct
profMemFutureLock [len(memRecord{}.future)]mutex
)
// All memory allocations are local and do not escape outside of the profiler.
// The profiler is forbidden from referring to garbage-collected memory.
const (
// profile types
memProfile bucketType = 1 + iota
blockProfile
mutexProfile
// size of bucket hash table
buckHashSize = 179999
// max depth of stack to record in bucket
maxStack = 32
)
type bucketType int
// A bucket holds per-call-stack profiling information.
// The representation is a bit sleazy, inherited from C.
// This struct defines the bucket header. It is followed in
// memory by the stack words and then the actual record
// data, either a memRecord or a blockRecord.
//
// Per-call-stack profiling information.
// Lookup by hashing call stack into a linked-list hash table.
//
// None of the fields in this bucket header are modified after
// creation, including its next and allnext links.
//
// No heap pointers.
type bucket struct {
_ sys.NotInHeap
next *bucket
allnext *bucket
typ bucketType // memBucket or blockBucket (includes mutexProfile)
hash uintptr
size uintptr
nstk uintptr
}
// A memRecord is the bucket data for a bucket of type memProfile,
// part of the memory profile.
type memRecord struct {
// The following complex 3-stage scheme of stats accumulation
// is required to obtain a consistent picture of mallocs and frees
// for some point in time.
// The problem is that mallocs come in real time, while frees
// come only after a GC during concurrent sweeping. So if we would
// naively count them, we would get a skew toward mallocs.
//
// Hence, we delay information to get consistent snapshots as
// of mark termination. Allocations count toward the next mark
// termination's snapshot, while sweep frees count toward the
// previous mark termination's snapshot:
//
// MT MT MT MT
// .·| .·| .·| .·|
// .·˙ | .·˙ | .·˙ | .·˙ |
// .·˙ | .·˙ | .·˙ | .·˙ |
// .·˙ |.·˙ |.·˙ |.·˙ |
//
// alloc → ▲ ← free
// ┠┅┅┅┅┅┅┅┅┅┅┅P
// C+2 → C+1 → C
//
// alloc → ▲ ← free
// ┠┅┅┅┅┅┅┅┅┅┅┅P
// C+2 → C+1 → C
//
// Since we can't publish a consistent snapshot until all of
// the sweep frees are accounted for, we wait until the next
// mark termination ("MT" above) to publish the previous mark
// termination's snapshot ("P" above). To do this, allocation
// and free events are accounted to *future* heap profile
// cycles ("C+n" above) and we only publish a cycle once all
// of the events from that cycle must be done. Specifically:
//
// Mallocs are accounted to cycle C+2.
// Explicit frees are accounted to cycle C+2.
// GC frees (done during sweeping) are accounted to cycle C+1.
//
// After mark termination, we increment the global heap
// profile cycle counter and accumulate the stats from cycle C
// into the active profile.
// active is the currently published profile. A profiling
// cycle can be accumulated into active once its complete.
active memRecordCycle
// future records the profile events we're counting for cycles
// that have not yet been published. This is ring buffer
// indexed by the global heap profile cycle C and stores
// cycles C, C+1, and C+2. Unlike active, these counts are
// only for a single cycle; they are not cumulative across
// cycles.
//
// We store cycle C here because there's a window between when
// C becomes the active cycle and when we've flushed it to
// active.
future [3]memRecordCycle
}
// memRecordCycle
type memRecordCycle struct {
allocs, frees uintptr
alloc_bytes, free_bytes uintptr
}
// add accumulates b into a. It does not zero b.
func (a *memRecordCycle) add(b *memRecordCycle) {
a.allocs += b.allocs
a.frees += b.frees
a.alloc_bytes += b.alloc_bytes
a.free_bytes += b.free_bytes
}
// A blockRecord is the bucket data for a bucket of type blockProfile,
// which is used in blocking and mutex profiles.
type blockRecord struct {
count float64
cycles int64
}
var (
mbuckets atomic.UnsafePointer // *bucket, memory profile buckets
bbuckets atomic.UnsafePointer // *bucket, blocking profile buckets
xbuckets atomic.UnsafePointer // *bucket, mutex profile buckets
buckhash atomic.UnsafePointer // *buckhashArray
mProfCycle mProfCycleHolder
)
type buckhashArray [buckHashSize]atomic.UnsafePointer // *bucket
const mProfCycleWrap = uint32(len(memRecord{}.future)) * (2 << 24)
// mProfCycleHolder holds the global heap profile cycle number (wrapped at
// mProfCycleWrap, stored starting at bit 1), and a flag (stored at bit 0) to
// indicate whether future[cycle] in all buckets has been queued to flush into
// the active profile.
type mProfCycleHolder struct {
value atomic.Uint32
}
// read returns the current cycle count.
func (c *mProfCycleHolder) read() (cycle uint32) {
v := c.value.Load()
cycle = v >> 1
return cycle
}
// setFlushed sets the flushed flag. It returns the current cycle count and the
// previous value of the flushed flag.
func (c *mProfCycleHolder) setFlushed() (cycle uint32, alreadyFlushed bool) {
for {
prev := c.value.Load()
cycle = prev >> 1
alreadyFlushed = (prev & 0x1) != 0
next := prev | 0x1
if c.value.CompareAndSwap(prev, next) {
return cycle, alreadyFlushed
}
}
}
// increment increases the cycle count by one, wrapping the value at
// mProfCycleWrap. It clears the flushed flag.
func (c *mProfCycleHolder) increment() {
// We explicitly wrap mProfCycle rather than depending on
// uint wraparound because the memRecord.future ring does not
// itself wrap at a power of two.
for {
prev := c.value.Load()
cycle := prev >> 1
cycle = (cycle + 1) % mProfCycleWrap
next := cycle << 1
if c.value.CompareAndSwap(prev, next) {
break
}
}
}
// newBucket allocates a bucket with the given type and number of stack entries.
func newBucket(typ bucketType, nstk int) *bucket {
size := unsafe.Sizeof(bucket{}) + uintptr(nstk)*unsafe.Sizeof(uintptr(0))
switch typ {
default:
throw("invalid profile bucket type")
case memProfile:
size += unsafe.Sizeof(memRecord{})
case blockProfile, mutexProfile:
size += unsafe.Sizeof(blockRecord{})
}
b := (*bucket)(persistentalloc(size, 0, &memstats.buckhash_sys))
b.typ = typ
b.nstk = uintptr(nstk)
return b
}
// stk returns the slice in b holding the stack.
func (b *bucket) stk() []uintptr {
stk := (*[maxStack]uintptr)(add(unsafe.Pointer(b), unsafe.Sizeof(*b)))
return stk[:b.nstk:b.nstk]
}
// mp returns the memRecord associated with the memProfile bucket b.
func (b *bucket) mp() *memRecord {
if b.typ != memProfile {
throw("bad use of bucket.mp")
}
data := add(unsafe.Pointer(b), unsafe.Sizeof(*b)+b.nstk*unsafe.Sizeof(uintptr(0)))
return (*memRecord)(data)
}
// bp returns the blockRecord associated with the blockProfile bucket b.
func (b *bucket) bp() *blockRecord {
if b.typ != blockProfile && b.typ != mutexProfile {
throw("bad use of bucket.bp")
}
data := add(unsafe.Pointer(b), unsafe.Sizeof(*b)+b.nstk*unsafe.Sizeof(uintptr(0)))
return (*blockRecord)(data)
}
// Return the bucket for stk[0:nstk], allocating new bucket if needed.
func stkbucket(typ bucketType, size uintptr, stk []uintptr, alloc bool) *bucket {
bh := (*buckhashArray)(buckhash.Load())
if bh == nil {
lock(&profInsertLock)
// check again under the lock
bh = (*buckhashArray)(buckhash.Load())
if bh == nil {
bh = (*buckhashArray)(sysAlloc(unsafe.Sizeof(buckhashArray{}), &memstats.buckhash_sys))
if bh == nil {
throw("runtime: cannot allocate memory")
}
buckhash.StoreNoWB(unsafe.Pointer(bh))
}
unlock(&profInsertLock)
}
// Hash stack.
var h uintptr
for _, pc := range stk {
h += pc
h += h << 10
h ^= h >> 6
}
// hash in size
h += size
h += h << 10
h ^= h >> 6
// finalize
h += h << 3
h ^= h >> 11
i := int(h % buckHashSize)
// first check optimistically, without the lock
for b := (*bucket)(bh[i].Load()); b != nil; b = b.next {
if b.typ == typ && b.hash == h && b.size == size && eqslice(b.stk(), stk) {
return b
}
}
if !alloc {
return nil
}
lock(&profInsertLock)
// check again under the insertion lock
for b := (*bucket)(bh[i].Load()); b != nil; b = b.next {
if b.typ == typ && b.hash == h && b.size == size && eqslice(b.stk(), stk) {
unlock(&profInsertLock)
return b
}
}
// Create new bucket.
b := newBucket(typ, len(stk))
copy(b.stk(), stk)
b.hash = h
b.size = size
var allnext *atomic.UnsafePointer
if typ == memProfile {
allnext = &mbuckets
} else if typ == mutexProfile {
allnext = &xbuckets
} else {
allnext = &bbuckets
}
b.next = (*bucket)(bh[i].Load())
b.allnext = (*bucket)(allnext.Load())
bh[i].StoreNoWB(unsafe.Pointer(b))
allnext.StoreNoWB(unsafe.Pointer(b))
unlock(&profInsertLock)
return b
}
func eqslice(x, y []uintptr) bool {
if len(x) != len(y) {
return false
}
for i, xi := range x {
if xi != y[i] {
return false
}
}
return true
}
// mProf_NextCycle publishes the next heap profile cycle and creates a
// fresh heap profile cycle. This operation is fast and can be done
// during STW. The caller must call mProf_Flush before calling
// mProf_NextCycle again.
//
// This is called by mark termination during STW so allocations and
// frees after the world is started again count towards a new heap
// profiling cycle.
func mProf_NextCycle() {
mProfCycle.increment()
}
// mProf_Flush flushes the events from the current heap profiling
// cycle into the active profile. After this it is safe to start a new
// heap profiling cycle with mProf_NextCycle.
//
// This is called by GC after mark termination starts the world. In
// contrast with mProf_NextCycle, this is somewhat expensive, but safe
// to do concurrently.
func mProf_Flush() {
cycle, alreadyFlushed := mProfCycle.setFlushed()
if alreadyFlushed {
return
}
index := cycle % uint32(len(memRecord{}.future))
lock(&profMemActiveLock)
lock(&profMemFutureLock[index])
mProf_FlushLocked(index)
unlock(&profMemFutureLock[index])
unlock(&profMemActiveLock)
}
// mProf_FlushLocked flushes the events from the heap profiling cycle at index
// into the active profile. The caller must hold the lock for the active profile
// (profMemActiveLock) and for the profiling cycle at index
// (profMemFutureLock[index]).
func mProf_FlushLocked(index uint32) {
assertLockHeld(&profMemActiveLock)
assertLockHeld(&profMemFutureLock[index])
head := (*bucket)(mbuckets.Load())
for b := head; b != nil; b = b.allnext {
mp := b.mp()
// Flush cycle C into the published profile and clear
// it for reuse.
mpc := &mp.future[index]
mp.active.add(mpc)
*mpc = memRecordCycle{}
}
}
// mProf_PostSweep records that all sweep frees for this GC cycle have
// completed. This has the effect of publishing the heap profile
// snapshot as of the last mark termination without advancing the heap
// profile cycle.
func mProf_PostSweep() {
// Flush cycle C+1 to the active profile so everything as of
// the last mark termination becomes visible. *Don't* advance
// the cycle, since we're still accumulating allocs in cycle
// C+2, which have to become C+1 in the next mark termination
// and so on.
cycle := mProfCycle.read() + 1
index := cycle % uint32(len(memRecord{}.future))
lock(&profMemActiveLock)
lock(&profMemFutureLock[index])
mProf_FlushLocked(index)
unlock(&profMemFutureLock[index])
unlock(&profMemActiveLock)
}
// Called by malloc to record a profiled block.
func mProf_Malloc(p unsafe.Pointer, size uintptr) {
var stk [maxStack]uintptr
nstk := callers(4, stk[:])
index := (mProfCycle.read() + 2) % uint32(len(memRecord{}.future))
b := stkbucket(memProfile, size, stk[:nstk], true)
mp := b.mp()
mpc := &mp.future[index]
lock(&profMemFutureLock[index])
mpc.allocs++
mpc.alloc_bytes += size
unlock(&profMemFutureLock[index])
// Setprofilebucket locks a bunch of other mutexes, so we call it outside of
// the profiler locks. This reduces potential contention and chances of
// deadlocks. Since the object must be alive during the call to
// mProf_Malloc, it's fine to do this non-atomically.
systemstack(func() {
setprofilebucket(p, b)
})
}
// Called when freeing a profiled block.
func mProf_Free(b *bucket, size uintptr) {
index := (mProfCycle.read() + 1) % uint32(len(memRecord{}.future))
mp := b.mp()
mpc := &mp.future[index]
lock(&profMemFutureLock[index])
mpc.frees++
mpc.free_bytes += size
unlock(&profMemFutureLock[index])
}
var blockprofilerate uint64 // in CPU ticks
// SetBlockProfileRate controls the fraction of goroutine blocking events
// that are reported in the blocking profile. The profiler aims to sample
// an average of one blocking event per rate nanoseconds spent blocked.
//
// To include every blocking event in the profile, pass rate = 1.
// To turn off profiling entirely, pass rate <= 0.
func SetBlockProfileRate(rate int) {
var r int64
if rate <= 0 {
r = 0 // disable profiling
} else if rate == 1 {
r = 1 // profile everything
} else {
// convert ns to cycles, use float64 to prevent overflow during multiplication
r = int64(float64(rate) * float64(tickspersecond()) / (1000 * 1000 * 1000))
if r == 0 {
r = 1
}
}
atomic.Store64(&blockprofilerate, uint64(r))
}
func blockevent(cycles int64, skip int) {
if cycles <= 0 {
cycles = 1
}
rate := int64(atomic.Load64(&blockprofilerate))
if blocksampled(cycles, rate) {
saveblockevent(cycles, rate, skip+1, blockProfile)
}
}
// blocksampled returns true for all events where cycles >= rate. Shorter
// events have a cycles/rate random chance of returning true.
func blocksampled(cycles, rate int64) bool {
if rate <= 0 || (rate > cycles && int64(fastrand())%rate > cycles) {
return false
}
return true
}
func saveblockevent(cycles, rate int64, skip int, which bucketType) {
gp := getg()
var nstk int
var stk [maxStack]uintptr
if gp.m.curg == nil || gp.m.curg == gp {
nstk = callers(skip, stk[:])
} else {
nstk = gcallers(gp.m.curg, skip, stk[:])
}
b := stkbucket(which, 0, stk[:nstk], true)
bp := b.bp()
lock(&profBlockLock)
// We want to up-scale the count and cycles according to the
// probability that the event was sampled. For block profile events,
// the sample probability is 1 if cycles >= rate, and cycles / rate
// otherwise. For mutex profile events, the sample probability is 1 / rate.
// We scale the events by 1 / (probability the event was sampled).
if which == blockProfile && cycles < rate {
// Remove sampling bias, see discussion on http://golang.org/cl/299991.
bp.count += float64(rate) / float64(cycles)
bp.cycles += rate
} else if which == mutexProfile {
bp.count += float64(rate)
bp.cycles += rate * cycles
} else {
bp.count++
bp.cycles += cycles
}
unlock(&profBlockLock)
}
var mutexprofilerate uint64 // fraction sampled
// SetMutexProfileFraction controls the fraction of mutex contention events
// that are reported in the mutex profile. On average 1/rate events are
// reported. The previous rate is returned.
//
// To turn off profiling entirely, pass rate 0.
// To just read the current rate, pass rate < 0.
// (For n>1 the details of sampling may change.)
func SetMutexProfileFraction(rate int) int {
if rate < 0 {
return int(mutexprofilerate)
}
old := mutexprofilerate
atomic.Store64(&mutexprofilerate, uint64(rate))
return int(old)
}
//go:linkname mutexevent sync.event
func mutexevent(cycles int64, skip int) {
if cycles < 0 {
cycles = 0
}
rate := int64(atomic.Load64(&mutexprofilerate))
// TODO(pjw): measure impact of always calling fastrand vs using something
// like malloc.go:nextSample()
if rate > 0 && int64(fastrand())%rate == 0 {
saveblockevent(cycles, rate, skip+1, mutexProfile)
}
}
// Go interface to profile data.
// A StackRecord describes a single execution stack.
type StackRecord struct {
Stack0 [32]uintptr // stack trace for this record; ends at first 0 entry
}
// Stack returns the stack trace associated with the record,
// a prefix of r.Stack0.
func (r *StackRecord) Stack() []uintptr {
for i, v := range r.Stack0 {
if v == 0 {
return r.Stack0[0:i]
}
}
return r.Stack0[0:]
}
// MemProfileRate controls the fraction of memory allocations
// that are recorded and reported in the memory profile.
// The profiler aims to sample an average of
// one allocation per MemProfileRate bytes allocated.
//
// To include every allocated block in the profile, set MemProfileRate to 1.
// To turn off profiling entirely, set MemProfileRate to 0.
//
// The tools that process the memory profiles assume that the
// profile rate is constant across the lifetime of the program
// and equal to the current value. Programs that change the
// memory profiling rate should do so just once, as early as
// possible in the execution of the program (for example,
// at the beginning of main).
var MemProfileRate int = 512 * 1024
// disableMemoryProfiling is set by the linker if runtime.MemProfile
// is not used and the link type guarantees nobody else could use it
// elsewhere.
var disableMemoryProfiling bool
// A MemProfileRecord describes the live objects allocated
// by a particular call sequence (stack trace).
type MemProfileRecord struct {
AllocBytes, FreeBytes int64 // number of bytes allocated, freed
AllocObjects, FreeObjects int64 // number of objects allocated, freed
Stack0 [32]uintptr // stack trace for this record; ends at first 0 entry
}
// InUseBytes returns the number of bytes in use (AllocBytes - FreeBytes).
func (r *MemProfileRecord) InUseBytes() int64 { return r.AllocBytes - r.FreeBytes }
// InUseObjects returns the number of objects in use (AllocObjects - FreeObjects).
func (r *MemProfileRecord) InUseObjects() int64 {
return r.AllocObjects - r.FreeObjects
}
// Stack returns the stack trace associated with the record,
// a prefix of r.Stack0.
func (r *MemProfileRecord) Stack() []uintptr {
for i, v := range r.Stack0 {
if v == 0 {
return r.Stack0[0:i]
}
}
return r.Stack0[0:]
}
// MemProfile returns a profile of memory allocated and freed per allocation
// site.
//
// MemProfile returns n, the number of records in the current memory profile.
// If len(p) >= n, MemProfile copies the profile into p and returns n, true.
// If len(p) < n, MemProfile does not change p and returns n, false.
//
// If inuseZero is true, the profile includes allocation records
// where r.AllocBytes > 0 but r.AllocBytes == r.FreeBytes.
// These are sites where memory was allocated, but it has all
// been released back to the runtime.
//
// The returned profile may be up to two garbage collection cycles old.
// This is to avoid skewing the profile toward allocations; because
// allocations happen in real time but frees are delayed until the garbage
// collector performs sweeping, the profile only accounts for allocations
// that have had a chance to be freed by the garbage collector.
//
// Most clients should use the runtime/pprof package or
// the testing package's -test.memprofile flag instead
// of calling MemProfile directly.
func MemProfile(p []MemProfileRecord, inuseZero bool) (n int, ok bool) {
cycle := mProfCycle.read()
// If we're between mProf_NextCycle and mProf_Flush, take care
// of flushing to the active profile so we only have to look
// at the active profile below.
index := cycle % uint32(len(memRecord{}.future))
lock(&profMemActiveLock)
lock(&profMemFutureLock[index])
mProf_FlushLocked(index)
unlock(&profMemFutureLock[index])
clear := true
head := (*bucket)(mbuckets.Load())
for b := head; b != nil; b = b.allnext {
mp := b.mp()
if inuseZero || mp.active.alloc_bytes != mp.active.free_bytes {
n++
}
if mp.active.allocs != 0 || mp.active.frees != 0 {
clear = false
}
}
if clear {
// Absolutely no data, suggesting that a garbage collection
// has not yet happened. In order to allow profiling when
// garbage collection is disabled from the beginning of execution,
// accumulate all of the cycles, and recount buckets.
n = 0
for b := head; b != nil; b = b.allnext {
mp := b.mp()
for c := range mp.future {
lock(&profMemFutureLock[c])
mp.active.add(&mp.future[c])
mp.future[c] = memRecordCycle{}
unlock(&profMemFutureLock[c])
}
if inuseZero || mp.active.alloc_bytes != mp.active.free_bytes {
n++
}
}
}
if n <= len(p) {
ok = true
idx := 0
for b := head; b != nil; b = b.allnext {
mp := b.mp()
if inuseZero || mp.active.alloc_bytes != mp.active.free_bytes {
record(&p[idx], b)
idx++
}
}
}
unlock(&profMemActiveLock)
return
}
// Write b's data to r.
func record(r *MemProfileRecord, b *bucket) {
mp := b.mp()
r.AllocBytes = int64(mp.active.alloc_bytes)
r.FreeBytes = int64(mp.active.free_bytes)
r.AllocObjects = int64(mp.active.allocs)
r.FreeObjects = int64(mp.active.frees)
if raceenabled {
racewriterangepc(unsafe.Pointer(&r.Stack0[0]), unsafe.Sizeof(r.Stack0), getcallerpc(), abi.FuncPCABIInternal(MemProfile))
}
if msanenabled {
msanwrite(unsafe.Pointer(&r.Stack0[0]), unsafe.Sizeof(r.Stack0))
}
if asanenabled {
asanwrite(unsafe.Pointer(&r.Stack0[0]), unsafe.Sizeof(r.Stack0))
}
copy(r.Stack0[:], b.stk())
for i := int(b.nstk); i < len(r.Stack0); i++ {
r.Stack0[i] = 0
}
}
func iterate_memprof(fn func(*bucket, uintptr, *uintptr, uintptr, uintptr, uintptr)) {
lock(&profMemActiveLock)
head := (*bucket)(mbuckets.Load())
for b := head; b != nil; b = b.allnext {
mp := b.mp()
fn(b, b.nstk, &b.stk()[0], b.size, mp.active.allocs, mp.active.frees)
}
unlock(&profMemActiveLock)
}
// BlockProfileRecord describes blocking events originated
// at a particular call sequence (stack trace).
type BlockProfileRecord struct {
Count int64
Cycles int64
StackRecord
}
// BlockProfile returns n, the number of records in the current blocking profile.
// If len(p) >= n, BlockProfile copies the profile into p and returns n, true.
// If len(p) < n, BlockProfile does not change p and returns n, false.
//
// Most clients should use the runtime/pprof package or
// the testing package's -test.blockprofile flag instead
// of calling BlockProfile directly.
func BlockProfile(p []BlockProfileRecord) (n int, ok bool) {
lock(&profBlockLock)
head := (*bucket)(bbuckets.Load())
for b := head; b != nil; b = b.allnext {
n++
}
if n <= len(p) {
ok = true
for b := head; b != nil; b = b.allnext {
bp := b.bp()
r := &p[0]
r.Count = int64(bp.count)
// Prevent callers from having to worry about division by zero errors.
// See discussion on http://golang.org/cl/299991.
if r.Count == 0 {
r.Count = 1
}
r.Cycles = bp.cycles
if raceenabled {
racewriterangepc(unsafe.Pointer(&r.Stack0[0]), unsafe.Sizeof(r.Stack0), getcallerpc(), abi.FuncPCABIInternal(BlockProfile))
}
if msanenabled {
msanwrite(unsafe.Pointer(&r.Stack0[0]), unsafe.Sizeof(r.Stack0))
}
if asanenabled {
asanwrite(unsafe.Pointer(&r.Stack0[0]), unsafe.Sizeof(r.Stack0))
}
i := copy(r.Stack0[:], b.stk())
for ; i < len(r.Stack0); i++ {
r.Stack0[i] = 0
}
p = p[1:]
}
}
unlock(&profBlockLock)
return
}
// MutexProfile returns n, the number of records in the current mutex profile.
// If len(p) >= n, MutexProfile copies the profile into p and returns n, true.
// Otherwise, MutexProfile does not change p, and returns n, false.
//
// Most clients should use the runtime/pprof package
// instead of calling MutexProfile directly.
func MutexProfile(p []BlockProfileRecord) (n int, ok bool) {
lock(&profBlockLock)
head := (*bucket)(xbuckets.Load())
for b := head; b != nil; b = b.allnext {
n++
}
if n <= len(p) {
ok = true
for b := head; b != nil; b = b.allnext {
bp := b.bp()
r := &p[0]
r.Count = int64(bp.count)
r.Cycles = bp.cycles
i := copy(r.Stack0[:], b.stk())
for ; i < len(r.Stack0); i++ {
r.Stack0[i] = 0
}
p = p[1:]
}
}
unlock(&profBlockLock)
return
}
// ThreadCreateProfile returns n, the number of records in the thread creation profile.
// If len(p) >= n, ThreadCreateProfile copies the profile into p and returns n, true.
// If len(p) < n, ThreadCreateProfile does not change p and returns n, false.
//
// Most clients should use the runtime/pprof package instead
// of calling ThreadCreateProfile directly.
func ThreadCreateProfile(p []StackRecord) (n int, ok bool) {
first := (*m)(atomic.Loadp(unsafe.Pointer(&allm)))
for mp := first; mp != nil; mp = mp.alllink {
n++
}
if n <= len(p) {
ok = true
i := 0
for mp := first; mp != nil; mp = mp.alllink {
p[i].Stack0 = mp.createstack
i++
}
}
return
}
//go:linkname runtime_goroutineProfileWithLabels runtime/pprof.runtime_goroutineProfileWithLabels
func runtime_goroutineProfileWithLabels(p []StackRecord, labels []unsafe.Pointer) (n int, ok bool) {
return goroutineProfileWithLabels(p, labels)
}
const go119ConcurrentGoroutineProfile = true
// labels may be nil. If labels is non-nil, it must have the same length as p.
func goroutineProfileWithLabels(p []StackRecord, labels []unsafe.Pointer) (n int, ok bool) {
if labels != nil && len(labels) != len(p) {
labels = nil
}
if go119ConcurrentGoroutineProfile {
return goroutineProfileWithLabelsConcurrent(p, labels)
}
return goroutineProfileWithLabelsSync(p, labels)
}
var goroutineProfile = struct {
sema uint32
active bool
offset atomic.Int64
records []StackRecord
labels []unsafe.Pointer
}{
sema: 1,
}
// goroutineProfileState indicates the status of a goroutine's stack for the
// current in-progress goroutine profile. Goroutines' stacks are initially
// "Absent" from the profile, and end up "Satisfied" by the time the profile is
// complete. While a goroutine's stack is being captured, its
// goroutineProfileState will be "InProgress" and it will not be able to run
// until the capture completes and the state moves to "Satisfied".
//
// Some goroutines (the finalizer goroutine, which at various times can be
// either a "system" or a "user" goroutine, and the goroutine that is
// coordinating the profile, any goroutines created during the profile) move
// directly to the "Satisfied" state.
type goroutineProfileState uint32
const (
goroutineProfileAbsent goroutineProfileState = iota
goroutineProfileInProgress
goroutineProfileSatisfied
)
type goroutineProfileStateHolder atomic.Uint32
func (p *goroutineProfileStateHolder) Load() goroutineProfileState {
return goroutineProfileState((*atomic.Uint32)(p).Load())
}
func (p *goroutineProfileStateHolder) Store(value goroutineProfileState) {
(*atomic.Uint32)(p).Store(uint32(value))
}
func (p *goroutineProfileStateHolder) CompareAndSwap(old, new goroutineProfileState) bool {
return (*atomic.Uint32)(p).CompareAndSwap(uint32(old), uint32(new))
}
func goroutineProfileWithLabelsConcurrent(p []StackRecord, labels []unsafe.Pointer) (n int, ok bool) {
semacquire(&goroutineProfile.sema)
ourg := getg()
stopTheWorld("profile")
// Using gcount while the world is stopped should give us a consistent view
// of the number of live goroutines, minus the number of goroutines that are
// alive and permanently marked as "system". But to make this count agree
// with what we'd get from isSystemGoroutine, we need special handling for
// goroutines that can vary between user and system to ensure that the count
// doesn't change during the collection. So, check the finalizer goroutine
// in particular.
n = int(gcount())
if fingStatus.Load()&fingRunningFinalizer != 0 {
n++
}
if n > len(p) {
// There's not enough space in p to store the whole profile, so (per the
// contract of runtime.GoroutineProfile) we're not allowed to write to p
// at all and must return n, false.
startTheWorld()
semrelease(&goroutineProfile.sema)
return n, false
}
// Save current goroutine.
sp := getcallersp()
pc := getcallerpc()
systemstack(func() {
saveg(pc, sp, ourg, &p[0])
})
ourg.goroutineProfiled.Store(goroutineProfileSatisfied)
goroutineProfile.offset.Store(1)
// Prepare for all other goroutines to enter the profile. Aside from ourg,
// every goroutine struct in the allgs list has its goroutineProfiled field
// cleared. Any goroutine created from this point on (while
// goroutineProfile.active is set) will start with its goroutineProfiled
// field set to goroutineProfileSatisfied.
goroutineProfile.active = true
goroutineProfile.records = p
goroutineProfile.labels = labels
// The finalizer goroutine needs special handling because it can vary over
// time between being a user goroutine (eligible for this profile) and a
// system goroutine (to be excluded). Pick one before restarting the world.
if fing != nil {
fing.goroutineProfiled.Store(goroutineProfileSatisfied)
if readgstatus(fing) != _Gdead && !isSystemGoroutine(fing, false) {
doRecordGoroutineProfile(fing)
}
}
startTheWorld()
// Visit each goroutine that existed as of the startTheWorld call above.
//
// New goroutines may not be in this list, but we didn't want to know about
// them anyway. If they do appear in this list (via reusing a dead goroutine
// struct, or racing to launch between the world restarting and us getting
// the list), they will already have their goroutineProfiled field set to
// goroutineProfileSatisfied before their state transitions out of _Gdead.
//
// Any goroutine that the scheduler tries to execute concurrently with this
// call will start by adding itself to the profile (before the act of
// executing can cause any changes in its stack).
forEachGRace(func(gp1 *g) {
tryRecordGoroutineProfile(gp1, Gosched)
})
stopTheWorld("profile cleanup")
endOffset := goroutineProfile.offset.Swap(0)
goroutineProfile.active = false
goroutineProfile.records = nil
goroutineProfile.labels = nil
startTheWorld()
// Restore the invariant that every goroutine struct in allgs has its
// goroutineProfiled field cleared.
forEachGRace(func(gp1 *g) {
gp1.goroutineProfiled.Store(goroutineProfileAbsent)
})
if raceenabled {
raceacquire(unsafe.Pointer(&labelSync))
}
if n != int(endOffset) {
// It's a big surprise that the number of goroutines changed while we
// were collecting the profile. But probably better to return a
// truncated profile than to crash the whole process.
//
// For instance, needm moves a goroutine out of the _Gdead state and so
// might be able to change the goroutine count without interacting with
// the scheduler. For code like that, the race windows are small and the
// combination of features is uncommon, so it's hard to be (and remain)
// sure we've caught them all.
}
semrelease(&goroutineProfile.sema)
return n, true
}
// tryRecordGoroutineProfileWB asserts that write barriers are allowed and calls
// tryRecordGoroutineProfile.
//
//go:yeswritebarrierrec
func tryRecordGoroutineProfileWB(gp1 *g) {
if getg().m.p.ptr() == nil {
throw("no P available, write barriers are forbidden")
}
tryRecordGoroutineProfile(gp1, osyield)
}
// tryRecordGoroutineProfile ensures that gp1 has the appropriate representation
// in the current goroutine profile: either that it should not be profiled, or
// that a snapshot of its call stack and labels are now in the profile.
func tryRecordGoroutineProfile(gp1 *g, yield func()) {
if readgstatus(gp1) == _Gdead {
// Dead goroutines should not appear in the profile. Goroutines that
// start while profile collection is active will get goroutineProfiled
// set to goroutineProfileSatisfied before transitioning out of _Gdead,
// so here we check _Gdead first.
return
}
if isSystemGoroutine(gp1, true) {
// System goroutines should not appear in the profile. (The finalizer
// goroutine is marked as "already profiled".)
return
}
for {
prev := gp1.goroutineProfiled.Load()
if prev == goroutineProfileSatisfied {
// This goroutine is already in the profile (or is new since the
// start of collection, so shouldn't appear in the profile).
break
}
if prev == goroutineProfileInProgress {
// Something else is adding gp1 to the goroutine profile right now.
// Give that a moment to finish.
yield()
continue
}
// While we have gp1.goroutineProfiled set to
// goroutineProfileInProgress, gp1 may appear _Grunnable but will not
// actually be able to run. Disable preemption for ourselves, to make
// sure we finish profiling gp1 right away instead of leaving it stuck
// in this limbo.
mp := acquirem()
if gp1.goroutineProfiled.CompareAndSwap(goroutineProfileAbsent, goroutineProfileInProgress) {
doRecordGoroutineProfile(gp1)
gp1.goroutineProfiled.Store(goroutineProfileSatisfied)
}
releasem(mp)
}
}
// doRecordGoroutineProfile writes gp1's call stack and labels to an in-progress
// goroutine profile. Preemption is disabled.
//
// This may be called via tryRecordGoroutineProfile in two ways: by the
// goroutine that is coordinating the goroutine profile (running on its own
// stack), or from the scheduler in preparation to execute gp1 (running on the
// system stack).
func doRecordGoroutineProfile(gp1 *g) {
if readgstatus(gp1) == _Grunning {
print("doRecordGoroutineProfile gp1=", gp1.goid, "\n")
throw("cannot read stack of running goroutine")
}
offset := int(goroutineProfile.offset.Add(1)) - 1
if offset >= len(goroutineProfile.records) {
// Should be impossible, but better to return a truncated profile than
// to crash the entire process at this point. Instead, deal with it in
// goroutineProfileWithLabelsConcurrent where we have more context.
return
}
// saveg calls gentraceback, which may call cgo traceback functions. When
// called from the scheduler, this is on the system stack already so
// traceback.go:cgoContextPCs will avoid calling back into the scheduler.
//
// When called from the goroutine coordinating the profile, we still have
// set gp1.goroutineProfiled to goroutineProfileInProgress and so are still
// preventing it from being truly _Grunnable. So we'll use the system stack
// to avoid schedule delays.
systemstack(func() { saveg(^uintptr(0), ^uintptr(0), gp1, &goroutineProfile.records[offset]) })
if goroutineProfile.labels != nil {
goroutineProfile.labels[offset] = gp1.labels
}
}
func goroutineProfileWithLabelsSync(p []StackRecord, labels []unsafe.Pointer) (n int, ok bool) {
gp := getg()
isOK := func(gp1 *g) bool {
// Checking isSystemGoroutine here makes GoroutineProfile
// consistent with both NumGoroutine and Stack.
return gp1 != gp && readgstatus(gp1) != _Gdead && !isSystemGoroutine(gp1, false)
}
stopTheWorld("profile")
// World is stopped, no locking required.
n = 1
forEachGRace(func(gp1 *g) {
if isOK(gp1) {
n++
}
})
if n <= len(p) {
ok = true
r, lbl := p, labels
// Save current goroutine.
sp := getcallersp()
pc := getcallerpc()
systemstack(func() {
saveg(pc, sp, gp, &r[0])
})
r = r[1:]
// If we have a place to put our goroutine labelmap, insert it there.
if labels != nil {
lbl[0] = gp.labels
lbl = lbl[1:]
}
// Save other goroutines.
forEachGRace(func(gp1 *g) {
if !isOK(gp1) {
return
}
if len(r) == 0 {
// Should be impossible, but better to return a
// truncated profile than to crash the entire process.
return
}
// saveg calls gentraceback, which may call cgo traceback functions.
// The world is stopped, so it cannot use cgocall (which will be
// blocked at exitsyscall). Do it on the system stack so it won't
// call into the schedular (see traceback.go:cgoContextPCs).
systemstack(func() { saveg(^uintptr(0), ^uintptr(0), gp1, &r[0]) })
if labels != nil {
lbl[0] = gp1.labels
lbl = lbl[1:]
}
r = r[1:]
})
}
if raceenabled {
raceacquire(unsafe.Pointer(&labelSync))
}
startTheWorld()
return n, ok
}
// GoroutineProfile returns n, the number of records in the active goroutine stack profile.
// If len(p) >= n, GoroutineProfile copies the profile into p and returns n, true.
// If len(p) < n, GoroutineProfile does not change p and returns n, false.
//
// Most clients should use the runtime/pprof package instead
// of calling GoroutineProfile directly.
func GoroutineProfile(p []StackRecord) (n int, ok bool) {
return goroutineProfileWithLabels(p, nil)
}
func saveg(pc, sp uintptr, gp *g, r *StackRecord) {
n := gentraceback(pc, sp, 0, gp, 0, &r.Stack0[0], len(r.Stack0), nil, nil, 0)
if n < len(r.Stack0) {
r.Stack0[n] = 0
}
}
// Stack formats a stack trace of the calling goroutine into buf
// and returns the number of bytes written to buf.
// If all is true, Stack formats stack traces of all other goroutines
// into buf after the trace for the current goroutine.
func Stack(buf []byte, all bool) int {
if all {
stopTheWorld("stack trace")
}
n := 0
if len(buf) > 0 {
gp := getg()
sp := getcallersp()
pc := getcallerpc()
systemstack(func() {
g0 := getg()
// Force traceback=1 to override GOTRACEBACK setting,
// so that Stack's results are consistent.
// GOTRACEBACK is only about crash dumps.
g0.m.traceback = 1
g0.writebuf = buf[0:0:len(buf)]
goroutineheader(gp)
traceback(pc, sp, 0, gp)
if all {
tracebackothers(gp)
}
g0.m.traceback = 0
n = len(g0.writebuf)
g0.writebuf = nil
})
}
if all {
startTheWorld()
}
return n
}
// Tracing of alloc/free/gc.
var tracelock mutex
func tracealloc(p unsafe.Pointer, size uintptr, typ *_type) {
lock(&tracelock)
gp := getg()
gp.m.traceback = 2
if typ == nil {
print("tracealloc(", p, ", ", hex(size), ")\n")
} else {
print("tracealloc(", p, ", ", hex(size), ", ", typ.string(), ")\n")
}
if gp.m.curg == nil || gp == gp.m.curg {
goroutineheader(gp)
pc := getcallerpc()
sp := getcallersp()
systemstack(func() {
traceback(pc, sp, 0, gp)
})
} else {
goroutineheader(gp.m.curg)
traceback(^uintptr(0), ^uintptr(0), 0, gp.m.curg)
}
print("\n")
gp.m.traceback = 0
unlock(&tracelock)
}
func tracefree(p unsafe.Pointer, size uintptr) {
lock(&tracelock)
gp := getg()
gp.m.traceback = 2
print("tracefree(", p, ", ", hex(size), ")\n")
goroutineheader(gp)
pc := getcallerpc()
sp := getcallersp()
systemstack(func() {
traceback(pc, sp, 0, gp)
})
print("\n")
gp.m.traceback = 0
unlock(&tracelock)
}
func tracegc() {
lock(&tracelock)
gp := getg()
gp.m.traceback = 2
print("tracegc()\n")
// running on m->g0 stack; show all non-g0 goroutines
tracebackothers(gp)
print("end tracegc\n")
print("\n")
gp.m.traceback = 0
unlock(&tracelock)
}
|