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
|
/*-
* Copyright (c) 2005-2007 Joseph Koshy
* Copyright (c) 2007 The FreeBSD Foundation
* All rights reserved.
*
* Portions of this software were developed by A. Joseph Koshy under
* sponsorship from the FreeBSD Foundation and Google, Inc.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
*/
/*
* Logging code for hwpmc(4)
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include <sys/param.h>
#include <sys/file.h>
#include <sys/kernel.h>
#include <sys/kthread.h>
#include <sys/lock.h>
#include <sys/module.h>
#include <sys/mutex.h>
#include <sys/pmc.h>
#include <sys/pmckern.h>
#include <sys/pmclog.h>
#include <sys/proc.h>
#include <sys/signalvar.h>
#include <sys/sysctl.h>
#include <sys/systm.h>
#include <sys/uio.h>
#include <sys/unistd.h>
#include <sys/vnode.h>
/*
* Sysctl tunables
*/
SYSCTL_DECL(_kern_hwpmc);
/*
* kern.hwpmc.logbuffersize -- size of the per-cpu owner buffers.
*/
static int pmclog_buffer_size = PMC_LOG_BUFFER_SIZE;
TUNABLE_INT(PMC_SYSCTL_NAME_PREFIX "logbuffersize", &pmclog_buffer_size);
SYSCTL_INT(_kern_hwpmc, OID_AUTO, logbuffersize, CTLFLAG_TUN|CTLFLAG_RD,
&pmclog_buffer_size, 0, "size of log buffers in kilobytes");
/*
* kern.hwpmc.nbuffer -- number of global log buffers
*/
static int pmc_nlogbuffers = PMC_NLOGBUFFERS;
TUNABLE_INT(PMC_SYSCTL_NAME_PREFIX "nbuffers", &pmc_nlogbuffers);
SYSCTL_INT(_kern_hwpmc, OID_AUTO, nbuffers, CTLFLAG_TUN|CTLFLAG_RD,
&pmc_nlogbuffers, 0, "number of global log buffers");
/*
* Global log buffer list and associated spin lock.
*/
TAILQ_HEAD(, pmclog_buffer) pmc_bufferlist =
TAILQ_HEAD_INITIALIZER(pmc_bufferlist);
static struct mtx pmc_bufferlist_mtx; /* spin lock */
static struct mtx pmc_kthread_mtx; /* sleep lock */
#define PMCLOG_INIT_BUFFER_DESCRIPTOR(D) do { \
const int __roundup = roundup(sizeof(*D), \
sizeof(uint32_t)); \
(D)->plb_fence = ((char *) (D)) + \
1024*pmclog_buffer_size; \
(D)->plb_base = (D)->plb_ptr = ((char *) (D)) + \
__roundup; \
} while (0)
/*
* Log file record constructors.
*/
#define _PMCLOG_TO_HEADER(T,L) \
((PMCLOG_HEADER_MAGIC << 24) | \
(PMCLOG_TYPE_ ## T << 16) | \
((L) & 0xFFFF))
/* reserve LEN bytes of space and initialize the entry header */
#define _PMCLOG_RESERVE(PO,TYPE,LEN,ACTION) do { \
uint32_t *_le; \
int _len = roundup((LEN), sizeof(uint32_t)); \
if ((_le = pmclog_reserve((PO), _len)) == NULL) { \
ACTION; \
} \
*_le = _PMCLOG_TO_HEADER(TYPE,_len); \
_le += 3 /* skip over timestamp */
#define PMCLOG_RESERVE(P,T,L) _PMCLOG_RESERVE(P,T,L,return)
#define PMCLOG_RESERVE_WITH_ERROR(P,T,L) _PMCLOG_RESERVE(P,T,L, \
error=ENOMEM;goto error)
#define PMCLOG_EMIT32(V) do { *_le++ = (V); } while (0)
#define PMCLOG_EMIT64(V) do { \
*_le++ = (uint32_t) ((V) & 0xFFFFFFFF); \
*_le++ = (uint32_t) (((V) >> 32) & 0xFFFFFFFF); \
} while (0)
/* Emit a string. Caution: does NOT update _le, so needs to be last */
#define PMCLOG_EMITSTRING(S,L) do { bcopy((S), _le, (L)); } while (0)
#define PMCLOG_DESPATCH(PO) \
pmclog_release((PO)); \
} while (0)
/*
* Assertions about the log file format.
*/
CTASSERT(sizeof(struct pmclog_callchain) == 6*4 +
PMC_CALLCHAIN_DEPTH_MAX*sizeof(uintfptr_t));
CTASSERT(sizeof(struct pmclog_closelog) == 3*4);
CTASSERT(sizeof(struct pmclog_dropnotify) == 3*4);
CTASSERT(sizeof(struct pmclog_map_in) == PATH_MAX +
4*4 + sizeof(uintfptr_t));
CTASSERT(offsetof(struct pmclog_map_in,pl_pathname) ==
4*4 + sizeof(uintfptr_t));
CTASSERT(sizeof(struct pmclog_map_out) == 4*4 + 2*sizeof(uintfptr_t));
CTASSERT(sizeof(struct pmclog_pcsample) == 6*4 + sizeof(uintfptr_t));
CTASSERT(sizeof(struct pmclog_pmcallocate) == 6*4);
CTASSERT(sizeof(struct pmclog_pmcattach) == 5*4 + PATH_MAX);
CTASSERT(offsetof(struct pmclog_pmcattach,pl_pathname) == 5*4);
CTASSERT(sizeof(struct pmclog_pmcdetach) == 5*4);
CTASSERT(sizeof(struct pmclog_proccsw) == 5*4 + 8);
CTASSERT(sizeof(struct pmclog_procexec) == 5*4 + PATH_MAX +
sizeof(uintfptr_t));
CTASSERT(offsetof(struct pmclog_procexec,pl_pathname) == 5*4 +
sizeof(uintfptr_t));
CTASSERT(sizeof(struct pmclog_procexit) == 5*4 + 8);
CTASSERT(sizeof(struct pmclog_procfork) == 5*4);
CTASSERT(sizeof(struct pmclog_sysexit) == 4*4);
CTASSERT(sizeof(struct pmclog_userdata) == 4*4);
/*
* Log buffer structure
*/
struct pmclog_buffer {
TAILQ_ENTRY(pmclog_buffer) plb_next;
char *plb_base;
char *plb_ptr;
char *plb_fence;
};
/*
* Prototypes
*/
static int pmclog_get_buffer(struct pmc_owner *po);
static void pmclog_loop(void *arg);
static void pmclog_release(struct pmc_owner *po);
static uint32_t *pmclog_reserve(struct pmc_owner *po, int length);
static void pmclog_schedule_io(struct pmc_owner *po);
static void pmclog_stop_kthread(struct pmc_owner *po);
/*
* Helper functions
*/
/*
* Get a log buffer
*/
static int
pmclog_get_buffer(struct pmc_owner *po)
{
struct pmclog_buffer *plb;
mtx_assert(&po->po_mtx, MA_OWNED);
KASSERT(po->po_curbuf == NULL,
("[pmclog,%d] po=%p current buffer still valid", __LINE__, po));
mtx_lock_spin(&pmc_bufferlist_mtx);
if ((plb = TAILQ_FIRST(&pmc_bufferlist)) != NULL)
TAILQ_REMOVE(&pmc_bufferlist, plb, plb_next);
mtx_unlock_spin(&pmc_bufferlist_mtx);
PMCDBG(LOG,GTB,1, "po=%p plb=%p", po, plb);
#ifdef DEBUG
if (plb)
KASSERT(plb->plb_ptr == plb->plb_base &&
plb->plb_base < plb->plb_fence,
("[pmclog,%d] po=%p buffer invariants: ptr=%p "
"base=%p fence=%p", __LINE__, po, plb->plb_ptr,
plb->plb_base, plb->plb_fence));
#endif
po->po_curbuf = plb;
/* update stats */
atomic_add_int(&pmc_stats.pm_buffer_requests, 1);
if (plb == NULL)
atomic_add_int(&pmc_stats.pm_buffer_requests_failed, 1);
return (plb ? 0 : ENOMEM);
}
/*
* Log handler loop.
*
* This function is executed by each pmc owner's helper thread.
*/
static void
pmclog_loop(void *arg)
{
int error, last_buffer;
struct pmc_owner *po;
struct pmclog_buffer *lb;
struct proc *p;
struct ucred *ownercred;
struct ucred *mycred;
struct thread *td;
struct uio auio;
struct iovec aiov;
size_t nbytes;
po = (struct pmc_owner *) arg;
p = po->po_owner;
td = curthread;
mycred = td->td_ucred;
last_buffer = 0;
PROC_LOCK(p);
ownercred = crhold(p->p_ucred);
PROC_UNLOCK(p);
PMCDBG(LOG,INI,1, "po=%p kt=%p", po, po->po_kthread);
KASSERT(po->po_kthread == curthread->td_proc,
("[pmclog,%d] proc mismatch po=%p po/kt=%p curproc=%p", __LINE__,
po, po->po_kthread, curthread->td_proc));
lb = NULL;
/*
* Loop waiting for I/O requests to be added to the owner
* struct's queue. The loop is exited when the log file
* is deconfigured.
*/
mtx_lock(&pmc_kthread_mtx);
for (;;) {
/* check if we've been asked to exit */
if ((po->po_flags & PMC_PO_OWNS_LOGFILE) == 0)
break;
if (lb == NULL) { /* look for a fresh buffer to write */
mtx_lock_spin(&po->po_mtx);
if ((lb = TAILQ_FIRST(&po->po_logbuffers)) == NULL) {
mtx_unlock_spin(&po->po_mtx);
(void) msleep(po, &pmc_kthread_mtx, PWAIT,
"pmcloop", 0);
continue;
}
TAILQ_REMOVE(&po->po_logbuffers, lb, plb_next);
if (po->po_flags & PMC_PO_SHUTDOWN)
last_buffer = TAILQ_EMPTY(&po->po_logbuffers);
mtx_unlock_spin(&po->po_mtx);
}
mtx_unlock(&pmc_kthread_mtx);
/* process the request */
PMCDBG(LOG,WRI,2, "po=%p base=%p ptr=%p", po,
lb->plb_base, lb->plb_ptr);
/* change our thread's credentials before issuing the I/O */
aiov.iov_base = lb->plb_base;
aiov.iov_len = nbytes = lb->plb_ptr - lb->plb_base;
auio.uio_iov = &aiov;
auio.uio_iovcnt = 1;
auio.uio_offset = -1;
auio.uio_resid = nbytes;
auio.uio_rw = UIO_WRITE;
auio.uio_segflg = UIO_SYSSPACE;
auio.uio_td = td;
/* switch thread credentials -- see kern_ktrace.c */
td->td_ucred = ownercred;
error = fo_write(po->po_file, &auio, ownercred, 0, td);
td->td_ucred = mycred;
if (error) {
/* XXX some errors are recoverable */
/* send a SIGIO to the owner and exit */
PROC_LOCK(p);
psignal(p, SIGIO);
PROC_UNLOCK(p);
mtx_lock(&pmc_kthread_mtx);
po->po_error = error; /* save for flush log */
PMCDBG(LOG,WRI,2, "po=%p error=%d", po, error);
break;
}
if (last_buffer) {
/*
* Close the file to get PMCLOG_EOF error
* in pmclog(3).
*/
fo_close(po->po_file, curthread);
}
mtx_lock(&pmc_kthread_mtx);
/* put the used buffer back into the global pool */
PMCLOG_INIT_BUFFER_DESCRIPTOR(lb);
mtx_lock_spin(&pmc_bufferlist_mtx);
TAILQ_INSERT_HEAD(&pmc_bufferlist, lb, plb_next);
mtx_unlock_spin(&pmc_bufferlist_mtx);
lb = NULL;
}
po->po_kthread = NULL;
mtx_unlock(&pmc_kthread_mtx);
/* return the current I/O buffer to the global pool */
if (lb) {
PMCLOG_INIT_BUFFER_DESCRIPTOR(lb);
mtx_lock_spin(&pmc_bufferlist_mtx);
TAILQ_INSERT_HEAD(&pmc_bufferlist, lb, plb_next);
mtx_unlock_spin(&pmc_bufferlist_mtx);
}
/*
* Exit this thread, signalling the waiter
*/
crfree(ownercred);
kproc_exit(0);
}
/*
* Release and log entry and schedule an I/O if needed.
*/
static void
pmclog_release(struct pmc_owner *po)
{
KASSERT(po->po_curbuf->plb_ptr >= po->po_curbuf->plb_base,
("[pmclog,%d] buffer invariants po=%p ptr=%p base=%p", __LINE__,
po, po->po_curbuf->plb_ptr, po->po_curbuf->plb_base));
KASSERT(po->po_curbuf->plb_ptr <= po->po_curbuf->plb_fence,
("[pmclog,%d] buffer invariants po=%p ptr=%p fenc=%p", __LINE__,
po, po->po_curbuf->plb_ptr, po->po_curbuf->plb_fence));
/* schedule an I/O if we've filled a buffer */
if (po->po_curbuf->plb_ptr >= po->po_curbuf->plb_fence)
pmclog_schedule_io(po);
mtx_unlock_spin(&po->po_mtx);
PMCDBG(LOG,REL,1, "po=%p", po);
}
/*
* Attempt to reserve 'length' bytes of space in an owner's log
* buffer. The function returns a pointer to 'length' bytes of space
* if there was enough space or returns NULL if no space was
* available. Non-null returns do so with the po mutex locked. The
* caller must invoke pmclog_release() on the pmc owner structure
* when done.
*/
static uint32_t *
pmclog_reserve(struct pmc_owner *po, int length)
{
uintptr_t newptr, oldptr;
uint32_t *lh;
struct timespec ts;
PMCDBG(LOG,ALL,1, "po=%p len=%d", po, length);
KASSERT(length % sizeof(uint32_t) == 0,
("[pmclog,%d] length not a multiple of word size", __LINE__));
mtx_lock_spin(&po->po_mtx);
/* No more data when shutdown in progress. */
if (po->po_flags & PMC_PO_SHUTDOWN) {
mtx_unlock_spin(&po->po_mtx);
return (NULL);
}
if (po->po_curbuf == NULL)
if (pmclog_get_buffer(po) != 0) {
mtx_unlock_spin(&po->po_mtx);
return (NULL);
}
KASSERT(po->po_curbuf != NULL,
("[pmclog,%d] po=%p no current buffer", __LINE__, po));
KASSERT(po->po_curbuf->plb_ptr >= po->po_curbuf->plb_base &&
po->po_curbuf->plb_ptr <= po->po_curbuf->plb_fence,
("[pmclog,%d] po=%p buffer invariants: ptr=%p base=%p fence=%p",
__LINE__, po, po->po_curbuf->plb_ptr, po->po_curbuf->plb_base,
po->po_curbuf->plb_fence));
oldptr = (uintptr_t) po->po_curbuf->plb_ptr;
newptr = oldptr + length;
KASSERT(oldptr != (uintptr_t) NULL,
("[pmclog,%d] po=%p Null log buffer pointer", __LINE__, po));
/*
* If we have space in the current buffer, return a pointer to
* available space with the PO structure locked.
*/
if (newptr <= (uintptr_t) po->po_curbuf->plb_fence) {
po->po_curbuf->plb_ptr = (char *) newptr;
goto done;
}
/*
* Otherwise, schedule the current buffer for output and get a
* fresh buffer.
*/
pmclog_schedule_io(po);
if (pmclog_get_buffer(po) != 0) {
mtx_unlock_spin(&po->po_mtx);
return (NULL);
}
KASSERT(po->po_curbuf != NULL,
("[pmclog,%d] po=%p no current buffer", __LINE__, po));
KASSERT(po->po_curbuf->plb_ptr != NULL,
("[pmclog,%d] null return from pmc_get_log_buffer", __LINE__));
KASSERT(po->po_curbuf->plb_ptr == po->po_curbuf->plb_base &&
po->po_curbuf->plb_ptr <= po->po_curbuf->plb_fence,
("[pmclog,%d] po=%p buffer invariants: ptr=%p base=%p fence=%p",
__LINE__, po, po->po_curbuf->plb_ptr, po->po_curbuf->plb_base,
po->po_curbuf->plb_fence));
oldptr = (uintptr_t) po->po_curbuf->plb_ptr;
done:
lh = (uint32_t *) oldptr;
lh++; /* skip header */
getnanotime(&ts); /* fill in the timestamp */
*lh++ = ts.tv_sec & 0xFFFFFFFF;
*lh++ = ts.tv_nsec & 0xFFFFFFF;
return ((uint32_t *) oldptr);
}
/*
* Schedule an I/O.
*
* Transfer the current buffer to the helper kthread.
*/
static void
pmclog_schedule_io(struct pmc_owner *po)
{
KASSERT(po->po_curbuf != NULL,
("[pmclog,%d] schedule_io with null buffer po=%p", __LINE__, po));
KASSERT(po->po_curbuf->plb_ptr >= po->po_curbuf->plb_base,
("[pmclog,%d] buffer invariants po=%p ptr=%p base=%p", __LINE__,
po, po->po_curbuf->plb_ptr, po->po_curbuf->plb_base));
KASSERT(po->po_curbuf->plb_ptr <= po->po_curbuf->plb_fence,
("[pmclog,%d] buffer invariants po=%p ptr=%p fenc=%p", __LINE__,
po, po->po_curbuf->plb_ptr, po->po_curbuf->plb_fence));
PMCDBG(LOG,SIO, 1, "po=%p", po);
mtx_assert(&po->po_mtx, MA_OWNED);
/*
* Add the current buffer to the tail of the buffer list and
* wakeup the helper.
*/
TAILQ_INSERT_TAIL(&po->po_logbuffers, po->po_curbuf, plb_next);
po->po_curbuf = NULL;
wakeup_one(po);
}
/*
* Stop the helper kthread.
*/
static void
pmclog_stop_kthread(struct pmc_owner *po)
{
/*
* Close the file to force the thread out of fo_write,
* unset flag, wakeup the helper thread,
* wait for it to exit
*/
if (po->po_file != NULL)
fo_close(po->po_file, curthread);
mtx_lock(&pmc_kthread_mtx);
po->po_flags &= ~PMC_PO_OWNS_LOGFILE;
wakeup_one(po);
if (po->po_kthread)
msleep(po->po_kthread, &pmc_kthread_mtx, PPAUSE, "pmckstp", 0);
mtx_unlock(&pmc_kthread_mtx);
}
/*
* Public functions
*/
/*
* Configure a log file for pmc owner 'po'.
*
* Parameter 'logfd' is a file handle referencing an open file in the
* owner process. This file needs to have been opened for writing.
*/
int
pmclog_configure_log(struct pmc_mdep *md, struct pmc_owner *po, int logfd)
{
int error;
struct proc *p;
/*
* As long as it is possible to get a LOR between pmc_sx lock and
* proctree/allproc sx locks used for adding a new process, assure
* the former is not held here.
*/
sx_assert(&pmc_sx, SA_UNLOCKED);
PMCDBG(LOG,CFG,1, "config po=%p logfd=%d", po, logfd);
p = po->po_owner;
/* return EBUSY if a log file was already present */
if (po->po_flags & PMC_PO_OWNS_LOGFILE)
return (EBUSY);
KASSERT(po->po_kthread == NULL,
("[pmclog,%d] po=%p kthread (%p) already present", __LINE__, po,
po->po_kthread));
KASSERT(po->po_file == NULL,
("[pmclog,%d] po=%p file (%p) already present", __LINE__, po,
po->po_file));
/* get a reference to the file state */
error = fget_write(curthread, logfd, &po->po_file);
if (error)
goto error;
/* mark process as owning a log file */
po->po_flags |= PMC_PO_OWNS_LOGFILE;
error = kproc_create(pmclog_loop, po, &po->po_kthread,
RFHIGHPID, 0, "hwpmc: proc(%d)", p->p_pid);
if (error)
goto error;
/* mark process as using HWPMCs */
PROC_LOCK(p);
p->p_flag |= P_HWPMC;
PROC_UNLOCK(p);
/* create a log initialization entry */
PMCLOG_RESERVE_WITH_ERROR(po, INITIALIZE,
sizeof(struct pmclog_initialize));
PMCLOG_EMIT32(PMC_VERSION);
PMCLOG_EMIT32(md->pmd_cputype);
PMCLOG_DESPATCH(po);
return (0);
error:
/* shutdown the thread */
if (po->po_kthread)
pmclog_stop_kthread(po);
KASSERT(po->po_kthread == NULL, ("[pmclog,%d] po=%p kthread not "
"stopped", __LINE__, po));
if (po->po_file)
(void) fdrop(po->po_file, curthread);
po->po_file = NULL; /* clear file and error state */
po->po_error = 0;
return (error);
}
/*
* De-configure a log file. This will throw away any buffers queued
* for this owner process.
*/
int
pmclog_deconfigure_log(struct pmc_owner *po)
{
int error;
struct pmclog_buffer *lb;
PMCDBG(LOG,CFG,1, "de-config po=%p", po);
if ((po->po_flags & PMC_PO_OWNS_LOGFILE) == 0)
return (EINVAL);
KASSERT(po->po_sscount == 0,
("[pmclog,%d] po=%p still owning SS PMCs", __LINE__, po));
KASSERT(po->po_file != NULL,
("[pmclog,%d] po=%p no log file", __LINE__, po));
/* stop the kthread, this will reset the 'OWNS_LOGFILE' flag */
if (po->po_kthread)
pmclog_stop_kthread(po);
KASSERT(po->po_kthread == NULL,
("[pmclog,%d] po=%p kthread not stopped", __LINE__, po));
/* return all queued log buffers to the global pool */
while ((lb = TAILQ_FIRST(&po->po_logbuffers)) != NULL) {
TAILQ_REMOVE(&po->po_logbuffers, lb, plb_next);
PMCLOG_INIT_BUFFER_DESCRIPTOR(lb);
mtx_lock_spin(&pmc_bufferlist_mtx);
TAILQ_INSERT_HEAD(&pmc_bufferlist, lb, plb_next);
mtx_unlock_spin(&pmc_bufferlist_mtx);
}
/* return the 'current' buffer to the global pool */
if ((lb = po->po_curbuf) != NULL) {
PMCLOG_INIT_BUFFER_DESCRIPTOR(lb);
mtx_lock_spin(&pmc_bufferlist_mtx);
TAILQ_INSERT_HEAD(&pmc_bufferlist, lb, plb_next);
mtx_unlock_spin(&pmc_bufferlist_mtx);
}
/* drop a reference to the fd */
error = fdrop(po->po_file, curthread);
po->po_file = NULL;
po->po_error = 0;
return (error);
}
/*
* Flush a process' log buffer.
*/
int
pmclog_flush(struct pmc_owner *po)
{
int error;
PMCDBG(LOG,FLS,1, "po=%p", po);
/*
* If there is a pending error recorded by the logger thread,
* return that.
*/
if (po->po_error)
return (po->po_error);
error = 0;
/*
* Check that we do have an active log file.
*/
mtx_lock(&pmc_kthread_mtx);
if ((po->po_flags & PMC_PO_OWNS_LOGFILE) == 0) {
error = EINVAL;
goto error;
}
/*
* Schedule the current buffer if any.
*/
mtx_lock_spin(&po->po_mtx);
if (po->po_curbuf)
pmclog_schedule_io(po);
mtx_unlock_spin(&po->po_mtx);
/*
* Initiate shutdown: no new data queued,
* thread will close file on last block.
*/
po->po_flags |= PMC_PO_SHUTDOWN;
error:
mtx_unlock(&pmc_kthread_mtx);
return (error);
}
void
pmclog_process_callchain(struct pmc *pm, struct pmc_sample *ps)
{
int n, recordlen;
uint32_t flags;
struct pmc_owner *po;
PMCDBG(LOG,SAM,1,"pm=%p pid=%d n=%d", pm, ps->ps_pid,
ps->ps_nsamples);
recordlen = offsetof(struct pmclog_callchain, pl_pc) +
ps->ps_nsamples * sizeof(uintfptr_t);
po = pm->pm_owner;
flags = PMC_CALLCHAIN_TO_CPUFLAGS(ps->ps_cpu,ps->ps_flags);
PMCLOG_RESERVE(po, CALLCHAIN, recordlen);
PMCLOG_EMIT32(ps->ps_pid);
PMCLOG_EMIT32(pm->pm_id);
PMCLOG_EMIT32(flags);
for (n = 0; n < ps->ps_nsamples; n++)
PMCLOG_EMITADDR(ps->ps_pc[n]);
PMCLOG_DESPATCH(po);
}
void
pmclog_process_closelog(struct pmc_owner *po)
{
PMCLOG_RESERVE(po,CLOSELOG,sizeof(struct pmclog_closelog));
PMCLOG_DESPATCH(po);
}
void
pmclog_process_dropnotify(struct pmc_owner *po)
{
PMCLOG_RESERVE(po,DROPNOTIFY,sizeof(struct pmclog_dropnotify));
PMCLOG_DESPATCH(po);
}
void
pmclog_process_map_in(struct pmc_owner *po, pid_t pid, uintfptr_t start,
const char *path)
{
int pathlen, recordlen;
KASSERT(path != NULL, ("[pmclog,%d] map-in, null path", __LINE__));
pathlen = strlen(path) + 1; /* #bytes for path name */
recordlen = offsetof(struct pmclog_map_in, pl_pathname) +
pathlen;
PMCLOG_RESERVE(po, MAP_IN, recordlen);
PMCLOG_EMIT32(pid);
PMCLOG_EMITADDR(start);
PMCLOG_EMITSTRING(path,pathlen);
PMCLOG_DESPATCH(po);
}
void
pmclog_process_map_out(struct pmc_owner *po, pid_t pid, uintfptr_t start,
uintfptr_t end)
{
KASSERT(start <= end, ("[pmclog,%d] start > end", __LINE__));
PMCLOG_RESERVE(po, MAP_OUT, sizeof(struct pmclog_map_out));
PMCLOG_EMIT32(pid);
PMCLOG_EMITADDR(start);
PMCLOG_EMITADDR(end);
PMCLOG_DESPATCH(po);
}
void
pmclog_process_pmcallocate(struct pmc *pm)
{
struct pmc_owner *po;
po = pm->pm_owner;
PMCDBG(LOG,ALL,1, "pm=%p", pm);
PMCLOG_RESERVE(po, PMCALLOCATE, sizeof(struct pmclog_pmcallocate));
PMCLOG_EMIT32(pm->pm_id);
PMCLOG_EMIT32(pm->pm_event);
PMCLOG_EMIT32(pm->pm_flags);
PMCLOG_DESPATCH(po);
}
void
pmclog_process_pmcattach(struct pmc *pm, pid_t pid, char *path)
{
int pathlen, recordlen;
struct pmc_owner *po;
PMCDBG(LOG,ATT,1,"pm=%p pid=%d", pm, pid);
po = pm->pm_owner;
pathlen = strlen(path) + 1; /* #bytes for the string */
recordlen = offsetof(struct pmclog_pmcattach, pl_pathname) + pathlen;
PMCLOG_RESERVE(po, PMCATTACH, recordlen);
PMCLOG_EMIT32(pm->pm_id);
PMCLOG_EMIT32(pid);
PMCLOG_EMITSTRING(path, pathlen);
PMCLOG_DESPATCH(po);
}
void
pmclog_process_pmcdetach(struct pmc *pm, pid_t pid)
{
struct pmc_owner *po;
PMCDBG(LOG,ATT,1,"!pm=%p pid=%d", pm, pid);
po = pm->pm_owner;
PMCLOG_RESERVE(po, PMCDETACH, sizeof(struct pmclog_pmcdetach));
PMCLOG_EMIT32(pm->pm_id);
PMCLOG_EMIT32(pid);
PMCLOG_DESPATCH(po);
}
/*
* Log a context switch event to the log file.
*/
void
pmclog_process_proccsw(struct pmc *pm, struct pmc_process *pp, pmc_value_t v)
{
struct pmc_owner *po;
KASSERT(pm->pm_flags & PMC_F_LOG_PROCCSW,
("[pmclog,%d] log-process-csw called gratuitously", __LINE__));
PMCDBG(LOG,SWO,1,"pm=%p pid=%d v=%jx", pm, pp->pp_proc->p_pid,
v);
po = pm->pm_owner;
PMCLOG_RESERVE(po, PROCCSW, sizeof(struct pmclog_proccsw));
PMCLOG_EMIT32(pm->pm_id);
PMCLOG_EMIT64(v);
PMCLOG_EMIT32(pp->pp_proc->p_pid);
PMCLOG_DESPATCH(po);
}
void
pmclog_process_procexec(struct pmc_owner *po, pmc_id_t pmid, pid_t pid,
uintfptr_t startaddr, char *path)
{
int pathlen, recordlen;
PMCDBG(LOG,EXC,1,"po=%p pid=%d path=\"%s\"", po, pid, path);
pathlen = strlen(path) + 1; /* #bytes for the path */
recordlen = offsetof(struct pmclog_procexec, pl_pathname) + pathlen;
PMCLOG_RESERVE(po, PROCEXEC, recordlen);
PMCLOG_EMIT32(pid);
PMCLOG_EMITADDR(startaddr);
PMCLOG_EMIT32(pmid);
PMCLOG_EMITSTRING(path,pathlen);
PMCLOG_DESPATCH(po);
}
/*
* Log a process exit event (and accumulated pmc value) to the log file.
*/
void
pmclog_process_procexit(struct pmc *pm, struct pmc_process *pp)
{
int ri;
struct pmc_owner *po;
ri = PMC_TO_ROWINDEX(pm);
PMCDBG(LOG,EXT,1,"pm=%p pid=%d v=%jx", pm, pp->pp_proc->p_pid,
pp->pp_pmcs[ri].pp_pmcval);
po = pm->pm_owner;
PMCLOG_RESERVE(po, PROCEXIT, sizeof(struct pmclog_procexit));
PMCLOG_EMIT32(pm->pm_id);
PMCLOG_EMIT64(pp->pp_pmcs[ri].pp_pmcval);
PMCLOG_EMIT32(pp->pp_proc->p_pid);
PMCLOG_DESPATCH(po);
}
/*
* Log a fork event.
*/
void
pmclog_process_procfork(struct pmc_owner *po, pid_t oldpid, pid_t newpid)
{
PMCLOG_RESERVE(po, PROCFORK, sizeof(struct pmclog_procfork));
PMCLOG_EMIT32(oldpid);
PMCLOG_EMIT32(newpid);
PMCLOG_DESPATCH(po);
}
/*
* Log a process exit event of the form suitable for system-wide PMCs.
*/
void
pmclog_process_sysexit(struct pmc_owner *po, pid_t pid)
{
PMCLOG_RESERVE(po, SYSEXIT, sizeof(struct pmclog_sysexit));
PMCLOG_EMIT32(pid);
PMCLOG_DESPATCH(po);
}
/*
* Write a user log entry.
*/
int
pmclog_process_userlog(struct pmc_owner *po, struct pmc_op_writelog *wl)
{
int error;
PMCDBG(LOG,WRI,1, "writelog po=%p ud=0x%x", po, wl->pm_userdata);
error = 0;
PMCLOG_RESERVE_WITH_ERROR(po, USERDATA,
sizeof(struct pmclog_userdata));
PMCLOG_EMIT32(wl->pm_userdata);
PMCLOG_DESPATCH(po);
error:
return (error);
}
/*
* Initialization.
*
* Create a pool of log buffers and initialize mutexes.
*/
void
pmclog_initialize()
{
int n;
struct pmclog_buffer *plb;
if (pmclog_buffer_size <= 0) {
(void) printf("hwpmc: tunable logbuffersize=%d must be "
"greater than zero.\n", pmclog_buffer_size);
pmclog_buffer_size = PMC_LOG_BUFFER_SIZE;
}
if (pmc_nlogbuffers <= 0) {
(void) printf("hwpmc: tunable nlogbuffers=%d must be greater "
"than zero.\n", pmc_nlogbuffers);
pmc_nlogbuffers = PMC_NLOGBUFFERS;
}
/* create global pool of log buffers */
for (n = 0; n < pmc_nlogbuffers; n++) {
plb = malloc(1024 * pmclog_buffer_size, M_PMC,
M_WAITOK|M_ZERO);
PMCLOG_INIT_BUFFER_DESCRIPTOR(plb);
TAILQ_INSERT_HEAD(&pmc_bufferlist, plb, plb_next);
}
mtx_init(&pmc_bufferlist_mtx, "pmc-buffer-list", "pmc-leaf",
MTX_SPIN);
mtx_init(&pmc_kthread_mtx, "pmc-kthread", "pmc-sleep", MTX_DEF);
}
/*
* Shutdown logging.
*
* Destroy mutexes and release memory back the to free pool.
*/
void
pmclog_shutdown()
{
struct pmclog_buffer *plb;
mtx_destroy(&pmc_kthread_mtx);
mtx_destroy(&pmc_bufferlist_mtx);
while ((plb = TAILQ_FIRST(&pmc_bufferlist)) != NULL) {
TAILQ_REMOVE(&pmc_bufferlist, plb, plb_next);
free(plb, M_PMC);
}
}
|