aboutsummaryrefslogtreecommitdiff
path: root/tools/perf/builtin-test.c
blob: 6173f780dce0a614633a35d262721bdd379f3f07 (plain)
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
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
/*
 * builtin-test.c
 *
 * Builtin regression testing command: ever growing number of sanity tests
 */
#include "builtin.h"

#include "util/cache.h"
#include "util/debug.h"
#include "util/debugfs.h"
#include "util/evlist.h"
#include "util/parse-options.h"
#include "util/parse-events.h"
#include "util/symbol.h"
#include "util/thread_map.h"
#include "../../include/linux/hw_breakpoint.h"

static int vmlinux_matches_kallsyms_filter(struct map *map __used, struct symbol *sym)
{
	bool *visited = symbol__priv(sym);
	*visited = true;
	return 0;
}

static int test__vmlinux_matches_kallsyms(void)
{
	int err = -1;
	struct rb_node *nd;
	struct symbol *sym;
	struct map *kallsyms_map, *vmlinux_map;
	struct machine kallsyms, vmlinux;
	enum map_type type = MAP__FUNCTION;
	long page_size = sysconf(_SC_PAGE_SIZE);
	struct ref_reloc_sym ref_reloc_sym = { .name = "_stext", };

	/*
	 * Step 1:
	 *
	 * Init the machines that will hold kernel, modules obtained from
	 * both vmlinux + .ko files and from /proc/kallsyms split by modules.
	 */
	machine__init(&kallsyms, "", HOST_KERNEL_ID);
	machine__init(&vmlinux, "", HOST_KERNEL_ID);

	/*
	 * Step 2:
	 *
	 * Create the kernel maps for kallsyms and the DSO where we will then
	 * load /proc/kallsyms. Also create the modules maps from /proc/modules
	 * and find the .ko files that match them in /lib/modules/`uname -r`/.
	 */
	if (machine__create_kernel_maps(&kallsyms) < 0) {
		pr_debug("machine__create_kernel_maps ");
		return -1;
	}

	/*
	 * Step 3:
	 *
	 * Load and split /proc/kallsyms into multiple maps, one per module.
	 */
	if (machine__load_kallsyms(&kallsyms, "/proc/kallsyms", type, NULL) <= 0) {
		pr_debug("dso__load_kallsyms ");
		goto out;
	}

	/*
	 * Step 4:
	 *
	 * kallsyms will be internally on demand sorted by name so that we can
	 * find the reference relocation * symbol, i.e. the symbol we will use
	 * to see if the running kernel was relocated by checking if it has the
	 * same value in the vmlinux file we load.
	 */
	kallsyms_map = machine__kernel_map(&kallsyms, type);

	sym = map__find_symbol_by_name(kallsyms_map, ref_reloc_sym.name, NULL);
	if (sym == NULL) {
		pr_debug("dso__find_symbol_by_name ");
		goto out;
	}

	ref_reloc_sym.addr = sym->start;

	/*
	 * Step 5:
	 *
	 * Now repeat step 2, this time for the vmlinux file we'll auto-locate.
	 */
	if (machine__create_kernel_maps(&vmlinux) < 0) {
		pr_debug("machine__create_kernel_maps ");
		goto out;
	}

	vmlinux_map = machine__kernel_map(&vmlinux, type);
	map__kmap(vmlinux_map)->ref_reloc_sym = &ref_reloc_sym;

	/*
	 * Step 6:
	 *
	 * Locate a vmlinux file in the vmlinux path that has a buildid that
	 * matches the one of the running kernel.
	 *
	 * While doing that look if we find the ref reloc symbol, if we find it
	 * we'll have its ref_reloc_symbol.unrelocated_addr and then
	 * maps__reloc_vmlinux will notice and set proper ->[un]map_ip routines
	 * to fixup the symbols.
	 */
	if (machine__load_vmlinux_path(&vmlinux, type,
				       vmlinux_matches_kallsyms_filter) <= 0) {
		pr_debug("machine__load_vmlinux_path ");
		goto out;
	}

	err = 0;
	/*
	 * Step 7:
	 *
	 * Now look at the symbols in the vmlinux DSO and check if we find all of them
	 * in the kallsyms dso. For the ones that are in both, check its names and
	 * end addresses too.
	 */
	for (nd = rb_first(&vmlinux_map->dso->symbols[type]); nd; nd = rb_next(nd)) {
		struct symbol *pair, *first_pair;
		bool backwards = true;

		sym  = rb_entry(nd, struct symbol, rb_node);

		if (sym->start == sym->end)
			continue;

		first_pair = machine__find_kernel_symbol(&kallsyms, type, sym->start, NULL, NULL);
		pair = first_pair;

		if (pair && pair->start == sym->start) {
next_pair:
			if (strcmp(sym->name, pair->name) == 0) {
				/*
				 * kallsyms don't have the symbol end, so we
				 * set that by using the next symbol start - 1,
				 * in some cases we get this up to a page
				 * wrong, trace_kmalloc when I was developing
				 * this code was one such example, 2106 bytes
				 * off the real size. More than that and we
				 * _really_ have a problem.
				 */
				s64 skew = sym->end - pair->end;
				if (llabs(skew) < page_size)
					continue;

				pr_debug("%#" PRIx64 ": diff end addr for %s v: %#" PRIx64 " k: %#" PRIx64 "\n",
					 sym->start, sym->name, sym->end, pair->end);
			} else {
				struct rb_node *nnd;
detour:
				nnd = backwards ? rb_prev(&pair->rb_node) :
						  rb_next(&pair->rb_node);
				if (nnd) {
					struct symbol *next = rb_entry(nnd, struct symbol, rb_node);

					if (next->start == sym->start) {
						pair = next;
						goto next_pair;
					}
				}

				if (backwards) {
					backwards = false;
					pair = first_pair;
					goto detour;
				}

				pr_debug("%#" PRIx64 ": diff name v: %s k: %s\n",
					 sym->start, sym->name, pair->name);
			}
		} else
			pr_debug("%#" PRIx64 ": %s not on kallsyms\n", sym->start, sym->name);

		err = -1;
	}

	if (!verbose)
		goto out;

	pr_info("Maps only in vmlinux:\n");

	for (nd = rb_first(&vmlinux.kmaps.maps[type]); nd; nd = rb_next(nd)) {
		struct map *pos = rb_entry(nd, struct map, rb_node), *pair;
		/*
		 * If it is the kernel, kallsyms is always "[kernel.kallsyms]", while
		 * the kernel will have the path for the vmlinux file being used,
		 * so use the short name, less descriptive but the same ("[kernel]" in
		 * both cases.
		 */
		pair = map_groups__find_by_name(&kallsyms.kmaps, type,
						(pos->dso->kernel ?
							pos->dso->short_name :
							pos->dso->name));
		if (pair)
			pair->priv = 1;
		else
			map__fprintf(pos, stderr);
	}

	pr_info("Maps in vmlinux with a different name in kallsyms:\n");

	for (nd = rb_first(&vmlinux.kmaps.maps[type]); nd; nd = rb_next(nd)) {
		struct map *pos = rb_entry(nd, struct map, rb_node), *pair;

		pair = map_groups__find(&kallsyms.kmaps, type, pos->start);
		if (pair == NULL || pair->priv)
			continue;

		if (pair->start == pos->start) {
			pair->priv = 1;
			pr_info(" %" PRIx64 "-%" PRIx64 " %" PRIx64 " %s in kallsyms as",
				pos->start, pos->end, pos->pgoff, pos->dso->name);
			if (pos->pgoff != pair->pgoff || pos->end != pair->end)
				pr_info(": \n*%" PRIx64 "-%" PRIx64 " %" PRIx64 "",
					pair->start, pair->end, pair->pgoff);
			pr_info(" %s\n", pair->dso->name);
			pair->priv = 1;
		}
	}

	pr_info("Maps only in kallsyms:\n");

	for (nd = rb_first(&kallsyms.kmaps.maps[type]);
	     nd; nd = rb_next(nd)) {
		struct map *pos = rb_entry(nd, struct map, rb_node);

		if (!pos->priv)
			map__fprintf(pos, stderr);
	}
out:
	return err;
}

#include "util/cpumap.h"
#include "util/evsel.h"
#include <sys/types.h>

static int trace_event__id(const char *evname)
{
	char *filename;
	int err = -1, fd;

	if (asprintf(&filename,
		     "%s/syscalls/%s/id",
		     tracing_events_path, evname) < 0)
		return -1;

	fd = open(filename, O_RDONLY);
	if (fd >= 0) {
		char id[16];
		if (read(fd, id, sizeof(id)) > 0)
			err = atoi(id);
		close(fd);
	}

	free(filename);
	return err;
}

static int test__open_syscall_event(void)
{
	int err = -1, fd;
	struct thread_map *threads;
	struct perf_evsel *evsel;
	struct perf_event_attr attr;
	unsigned int nr_open_calls = 111, i;
	int id = trace_event__id("sys_enter_open");

	if (id < 0) {
		pr_debug("is debugfs mounted on /sys/kernel/debug?\n");
		return -1;
	}

	threads = thread_map__new(-1, getpid());
	if (threads == NULL) {
		pr_debug("thread_map__new\n");
		return -1;
	}

	memset(&attr, 0, sizeof(attr));
	attr.type = PERF_TYPE_TRACEPOINT;
	attr.config = id;
	evsel = perf_evsel__new(&attr, 0);
	if (evsel == NULL) {
		pr_debug("perf_evsel__new\n");
		goto out_thread_map_delete;
	}

	if (perf_evsel__open_per_thread(evsel, threads, false, NULL) < 0) {
		pr_debug("failed to open counter: %s, "
			 "tweak /proc/sys/kernel/perf_event_paranoid?\n",
			 strerror(errno));
		goto out_evsel_delete;
	}

	for (i = 0; i < nr_open_calls; ++i) {
		fd = open("/etc/passwd", O_RDONLY);
		close(fd);
	}

	if (perf_evsel__read_on_cpu(evsel, 0, 0) < 0) {
		pr_debug("perf_evsel__read_on_cpu\n");
		goto out_close_fd;
	}

	if (evsel->counts->cpu[0].val != nr_open_calls) {
		pr_debug("perf_evsel__read_on_cpu: expected to intercept %d calls, got %" PRIu64 "\n",
			 nr_open_calls, evsel->counts->cpu[0].val);
		goto out_close_fd;
	}
	
	err = 0;
out_close_fd:
	perf_evsel__close_fd(evsel, 1, threads->nr);
out_evsel_delete:
	perf_evsel__delete(evsel);
out_thread_map_delete:
	thread_map__delete(threads);
	return err;
}

#include <sched.h>

static int test__open_syscall_event_on_all_cpus(void)
{
	int err = -1, fd, cpu;
	struct thread_map *threads;
	struct cpu_map *cpus;
	struct perf_evsel *evsel;
	struct perf_event_attr attr;
	unsigned int nr_open_calls = 111, i;
	cpu_set_t cpu_set;
	int id = trace_event__id("sys_enter_open");

	if (id < 0) {
		pr_debug("is debugfs mounted on /sys/kernel/debug?\n");
		return -1;
	}

	threads = thread_map__new(-1, getpid());
	if (threads == NULL) {
		pr_debug("thread_map__new\n");
		return -1;
	}

	cpus = cpu_map__new(NULL);
	if (cpus == NULL) {
		pr_debug("cpu_map__new\n");
		goto out_thread_map_delete;
	}


	CPU_ZERO(&cpu_set);

	memset(&attr, 0, sizeof(attr));
	attr.type = PERF_TYPE_TRACEPOINT;
	attr.config = id;
	evsel = perf_evsel__new(&attr, 0);
	if (evsel == NULL) {
		pr_debug("perf_evsel__new\n");
		goto out_thread_map_delete;
	}

	if (perf_evsel__open(evsel, cpus, threads, false, NULL) < 0) {
		pr_debug("failed to open counter: %s, "
			 "tweak /proc/sys/kernel/perf_event_paranoid?\n",
			 strerror(errno));
		goto out_evsel_delete;
	}

	for (cpu = 0; cpu < cpus->nr; ++cpu) {
		unsigned int ncalls = nr_open_calls + cpu;
		/*
		 * XXX eventually lift this restriction in a way that
		 * keeps perf building on older glibc installations
		 * without CPU_ALLOC. 1024 cpus in 2010 still seems
		 * a reasonable upper limit tho :-)
		 */
		if (cpus->map[cpu] >= CPU_SETSIZE) {
			pr_debug("Ignoring CPU %d\n", cpus->map[cpu]);
			continue;
		}

		CPU_SET(cpus->map[cpu], &cpu_set);
		if (sched_setaffinity(0, sizeof(cpu_set), &cpu_set) < 0) {
			pr_debug("sched_setaffinity() failed on CPU %d: %s ",
				 cpus->map[cpu],
				 strerror(errno));
			goto out_close_fd;
		}
		for (i = 0; i < ncalls; ++i) {
			fd = open("/etc/passwd", O_RDONLY);
			close(fd);
		}
		CPU_CLR(cpus->map[cpu], &cpu_set);
	}

	/*
	 * Here we need to explicitely preallocate the counts, as if
	 * we use the auto allocation it will allocate just for 1 cpu,
	 * as we start by cpu 0.
	 */
	if (perf_evsel__alloc_counts(evsel, cpus->nr) < 0) {
		pr_debug("perf_evsel__alloc_counts(ncpus=%d)\n", cpus->nr);
		goto out_close_fd;
	}

	err = 0;

	for (cpu = 0; cpu < cpus->nr; ++cpu) {
		unsigned int expected;

		if (cpus->map[cpu] >= CPU_SETSIZE)
			continue;

		if (perf_evsel__read_on_cpu(evsel, cpu, 0) < 0) {
			pr_debug("perf_evsel__read_on_cpu\n");
			err = -1;
			break;
		}

		expected = nr_open_calls + cpu;
		if (evsel->counts->cpu[cpu].val != expected) {
			pr_debug("perf_evsel__read_on_cpu: expected to intercept %d calls on cpu %d, got %" PRIu64 "\n",
				 expected, cpus->map[cpu], evsel->counts->cpu[cpu].val);
			err = -1;
		}
	}

out_close_fd:
	perf_evsel__close_fd(evsel, 1, threads->nr);
out_evsel_delete:
	perf_evsel__delete(evsel);
out_thread_map_delete:
	thread_map__delete(threads);
	return err;
}

/*
 * This test will generate random numbers of calls to some getpid syscalls,
 * then establish an mmap for a group of events that are created to monitor
 * the syscalls.
 *
 * It will receive the events, using mmap, use its PERF_SAMPLE_ID generated
 * sample.id field to map back to its respective perf_evsel instance.
 *
 * Then it checks if the number of syscalls reported as perf events by
 * the kernel corresponds to the number of syscalls made.
 */
static int test__basic_mmap(void)
{
	int err = -1;
	union perf_event *event;
	struct thread_map *threads;
	struct cpu_map *cpus;
	struct perf_evlist *evlist;
	struct perf_event_attr attr = {
		.type		= PERF_TYPE_TRACEPOINT,
		.read_format	= PERF_FORMAT_ID,
		.sample_type	= PERF_SAMPLE_ID,
		.watermark	= 0,
	};
	cpu_set_t cpu_set;
	const char *syscall_names[] = { "getsid", "getppid", "getpgrp",
					"getpgid", };
	pid_t (*syscalls[])(void) = { (void *)getsid, getppid, getpgrp,
				      (void*)getpgid };
#define nsyscalls ARRAY_SIZE(syscall_names)
	int ids[nsyscalls];
	unsigned int nr_events[nsyscalls],
		     expected_nr_events[nsyscalls], i, j;
	struct perf_evsel *evsels[nsyscalls], *evsel;
	int sample_size = __perf_evsel__sample_size(attr.sample_type);

	for (i = 0; i < nsyscalls; ++i) {
		char name[64];

		snprintf(name, sizeof(name), "sys_enter_%s", syscall_names[i]);
		ids[i] = trace_event__id(name);
		if (ids[i] < 0) {
			pr_debug("Is debugfs mounted on /sys/kernel/debug?\n");
			return -1;
		}
		nr_events[i] = 0;
		expected_nr_events[i] = random() % 257;
	}

	threads = thread_map__new(-1, getpid());
	if (threads == NULL) {
		pr_debug("thread_map__new\n");
		return -1;
	}

	cpus = cpu_map__new(NULL);
	if (cpus == NULL) {
		pr_debug("cpu_map__new\n");
		goto out_free_threads;
	}

	CPU_ZERO(&cpu_set);
	CPU_SET(cpus->map[0], &cpu_set);
	sched_setaffinity(0, sizeof(cpu_set), &cpu_set);
	if (sched_setaffinity(0, sizeof(cpu_set), &cpu_set) < 0) {
		pr_debug("sched_setaffinity() failed on CPU %d: %s ",
			 cpus->map[0], strerror(errno));
		goto out_free_cpus;
	}

	evlist = perf_evlist__new(cpus, threads);
	if (evlist == NULL) {
		pr_debug("perf_evlist__new\n");
		goto out_free_cpus;
	}

	/* anonymous union fields, can't be initialized above */
	attr.wakeup_events = 1;
	attr.sample_period = 1;

	for (i = 0; i < nsyscalls; ++i) {
		attr.config = ids[i];
		evsels[i] = perf_evsel__new(&attr, i);
		if (evsels[i] == NULL) {
			pr_debug("perf_evsel__new\n");
			goto out_free_evlist;
		}

		perf_evlist__add(evlist, evsels[i]);

		if (perf_evsel__open(evsels[i], cpus, threads, false, NULL) < 0) {
			pr_debug("failed to open counter: %s, "
				 "tweak /proc/sys/kernel/perf_event_paranoid?\n",
				 strerror(errno));
			goto out_close_fd;
		}
	}

	if (perf_evlist__mmap(evlist, 128, true) < 0) {
		pr_debug("failed to mmap events: %d (%s)\n", errno,
			 strerror(errno));
		goto out_close_fd;
	}

	for (i = 0; i < nsyscalls; ++i)
		for (j = 0; j < expected_nr_events[i]; ++j) {
			int foo = syscalls[i]();
			++foo;
		}

	while ((event = perf_evlist__mmap_read(evlist, 0)) != NULL) {
		struct perf_sample sample;

		if (event->header.type != PERF_RECORD_SAMPLE) {
			pr_debug("unexpected %s event\n",
				 perf_event__name(event->header.type));
			goto out_munmap;
		}

		err = perf_event__parse_sample(event, attr.sample_type, sample_size,
					       false, &sample, false);
		if (err) {
			pr_err("Can't parse sample, err = %d\n", err);
			goto out_munmap;
		}

		evsel = perf_evlist__id2evsel(evlist, sample.id);
		if (evsel == NULL) {
			pr_debug("event with id %" PRIu64
				 " doesn't map to an evsel\n", sample.id);
			goto out_munmap;
		}
		nr_events[evsel->idx]++;
	}

	list_for_each_entry(evsel, &evlist->entries, node) {
		if (nr_events[evsel->idx] != expected_nr_events[evsel->idx]) {
			pr_debug("expected %d %s events, got %d\n",
				 expected_nr_events[evsel->idx],
				 event_name(evsel), nr_events[evsel->idx]);
			goto out_munmap;
		}
	}

	err = 0;
out_munmap:
	perf_evlist__munmap(evlist);
out_close_fd:
	for (i = 0; i < nsyscalls; ++i)
		perf_evsel__close_fd(evsels[i], 1, threads->nr);
out_free_evlist:
	perf_evlist__delete(evlist);
out_free_cpus:
	cpu_map__delete(cpus);
out_free_threads:
	thread_map__delete(threads);
	return err;
#undef nsyscalls
}

#define TEST_ASSERT_VAL(text, cond) \
do { \
	if (!cond) { \
		pr_debug("FAILED %s:%d %s\n", __FILE__, __LINE__, text); \
		return -1; \
	} \
} while (0)

static int test__checkevent_tracepoint(struct perf_evlist *evlist)
{
	struct perf_evsel *evsel = list_entry(evlist->entries.next,
					      struct perf_evsel, node);

	TEST_ASSERT_VAL("wrong number of entries", 1 == evlist->nr_entries);
	TEST_ASSERT_VAL("wrong type", PERF_TYPE_TRACEPOINT == evsel->attr.type);
	TEST_ASSERT_VAL("wrong sample_type",
		(PERF_SAMPLE_RAW | PERF_SAMPLE_TIME | PERF_SAMPLE_CPU) ==
		evsel->attr.sample_type);
	TEST_ASSERT_VAL("wrong sample_period", 1 == evsel->attr.sample_period);
	return 0;
}

static int test__checkevent_tracepoint_multi(struct perf_evlist *evlist)
{
	struct perf_evsel *evsel;

	TEST_ASSERT_VAL("wrong number of entries", evlist->nr_entries > 1);

	list_for_each_entry(evsel, &evlist->entries, node) {
		TEST_ASSERT_VAL("wrong type",
			PERF_TYPE_TRACEPOINT == evsel->attr.type);
		TEST_ASSERT_VAL("wrong sample_type",
			(PERF_SAMPLE_RAW | PERF_SAMPLE_TIME | PERF_SAMPLE_CPU)
			== evsel->attr.sample_type);
		TEST_ASSERT_VAL("wrong sample_period",
			1 == evsel->attr.sample_period);
	}
	return 0;
}

static int test__checkevent_raw(struct perf_evlist *evlist)
{
	struct perf_evsel *evsel = list_entry(evlist->entries.next,
					      struct perf_evsel, node);

	TEST_ASSERT_VAL("wrong number of entries", 1 == evlist->nr_entries);
	TEST_ASSERT_VAL("wrong type", PERF_TYPE_RAW == evsel->attr.type);
	TEST_ASSERT_VAL("wrong config", 1 == evsel->attr.config);
	return 0;
}

static int test__checkevent_numeric(struct perf_evlist *evlist)
{
	struct perf_evsel *evsel = list_entry(evlist->entries.next,
					      struct perf_evsel, node);

	TEST_ASSERT_VAL("wrong number of entries", 1 == evlist->nr_entries);
	TEST_ASSERT_VAL("wrong type", 1 == evsel->attr.type);
	TEST_ASSERT_VAL("wrong config", 1 == evsel->attr.config);
	return 0;
}

static int test__checkevent_symbolic_name(struct perf_evlist *evlist)
{
	struct perf_evsel *evsel = list_entry(evlist->entries.next,
					      struct perf_evsel, node);

	TEST_ASSERT_VAL("wrong number of entries", 1 == evlist->nr_entries);
	TEST_ASSERT_VAL("wrong type", PERF_TYPE_HARDWARE == evsel->attr.type);
	TEST_ASSERT_VAL("wrong config",
			PERF_COUNT_HW_INSTRUCTIONS == evsel->attr.config);
	return 0;
}

static int test__checkevent_symbolic_alias(struct perf_evlist *evlist)
{
	struct perf_evsel *evsel = list_entry(evlist->entries.next,
					      struct perf_evsel, node);

	TEST_ASSERT_VAL("wrong number of entries", 1 == evlist->nr_entries);
	TEST_ASSERT_VAL("wrong type", PERF_TYPE_SOFTWARE == evsel->attr.type);
	TEST_ASSERT_VAL("wrong config",
			PERF_COUNT_SW_PAGE_FAULTS == evsel->attr.config);
	return 0;
}

static int test__checkevent_genhw(struct perf_evlist *evlist)
{
	struct perf_evsel *evsel = list_entry(evlist->entries.next,
					      struct perf_evsel, node);

	TEST_ASSERT_VAL("wrong number of entries", 1 == evlist->nr_entries);
	TEST_ASSERT_VAL("wrong type", PERF_TYPE_HW_CACHE == evsel->attr.type);
	TEST_ASSERT_VAL("wrong config", (1 << 16) == evsel->attr.config);
	return 0;
}

static int test__checkevent_breakpoint(struct perf_evlist *evlist)
{
	struct perf_evsel *evsel = list_entry(evlist->entries.next,
					      struct perf_evsel, node);

	TEST_ASSERT_VAL("wrong number of entries", 1 == evlist->nr_entries);
	TEST_ASSERT_VAL("wrong type", PERF_TYPE_BREAKPOINT == evsel->attr.type);
	TEST_ASSERT_VAL("wrong config", 0 == evsel->attr.config);
	TEST_ASSERT_VAL("wrong bp_type", (HW_BREAKPOINT_R | HW_BREAKPOINT_W) ==
					 evsel->attr.bp_type);
	TEST_ASSERT_VAL("wrong bp_len", HW_BREAKPOINT_LEN_4 ==
					evsel->attr.bp_len);
	return 0;
}

static int test__checkevent_breakpoint_x(struct perf_evlist *evlist)
{
	struct perf_evsel *evsel = list_entry(evlist->entries.next,
					      struct perf_evsel, node);

	TEST_ASSERT_VAL("wrong number of entries", 1 == evlist->nr_entries);
	TEST_ASSERT_VAL("wrong type", PERF_TYPE_BREAKPOINT == evsel->attr.type);
	TEST_ASSERT_VAL("wrong config", 0 == evsel->attr.config);
	TEST_ASSERT_VAL("wrong bp_type",
			HW_BREAKPOINT_X == evsel->attr.bp_type);
	TEST_ASSERT_VAL("wrong bp_len", sizeof(long) == evsel->attr.bp_len);
	return 0;
}

static int test__checkevent_breakpoint_r(struct perf_evlist *evlist)
{
	struct perf_evsel *evsel = list_entry(evlist->entries.next,
					      struct perf_evsel, node);

	TEST_ASSERT_VAL("wrong number of entries", 1 == evlist->nr_entries);
	TEST_ASSERT_VAL("wrong type",
			PERF_TYPE_BREAKPOINT == evsel->attr.type);
	TEST_ASSERT_VAL("wrong config", 0 == evsel->attr.config);
	TEST_ASSERT_VAL("wrong bp_type",
			HW_BREAKPOINT_R == evsel->attr.bp_type);
	TEST_ASSERT_VAL("wrong bp_len",
			HW_BREAKPOINT_LEN_4 == evsel->attr.bp_len);
	return 0;
}

static int test__checkevent_breakpoint_w(struct perf_evlist *evlist)
{
	struct perf_evsel *evsel = list_entry(evlist->entries.next,
					      struct perf_evsel, node);

	TEST_ASSERT_VAL("wrong number of entries", 1 == evlist->nr_entries);
	TEST_ASSERT_VAL("wrong type",
			PERF_TYPE_BREAKPOINT == evsel->attr.type);
	TEST_ASSERT_VAL("wrong config", 0 == evsel->attr.config);
	TEST_ASSERT_VAL("wrong bp_type",
			HW_BREAKPOINT_W == evsel->attr.bp_type);
	TEST_ASSERT_VAL("wrong bp_len",
			HW_BREAKPOINT_LEN_4 == evsel->attr.bp_len);
	return 0;
}

static struct test__event_st {
	const char *name;
	__u32 type;
	int (*check)(struct perf_evlist *evlist);
} test__events[] = {
	{
		.name  = "syscalls:sys_enter_open",
		.check = test__checkevent_tracepoint,
	},
	{
		.name  = "syscalls:*",
		.check = test__checkevent_tracepoint_multi,
	},
	{
		.name  = "r1",
		.check = test__checkevent_raw,
	},
	{
		.name  = "1:1",
		.check = test__checkevent_numeric,
	},
	{
		.name  = "instructions",
		.check = test__checkevent_symbolic_name,
	},
	{
		.name  = "faults",
		.check = test__checkevent_symbolic_alias,
	},
	{
		.name  = "L1-dcache-load-miss",
		.check = test__checkevent_genhw,
	},
	{
		.name  = "mem:0",
		.check = test__checkevent_breakpoint,
	},
	{
		.name  = "mem:0:x",
		.check = test__checkevent_breakpoint_x,
	},
	{
		.name  = "mem:0:r",
		.check = test__checkevent_breakpoint_r,
	},
	{
		.name  = "mem:0:w",
		.check = test__checkevent_breakpoint_w,
	},
};

#define TEST__EVENTS_CNT (sizeof(test__events) / sizeof(struct test__event_st))

static int test__parse_events(void)
{
	struct perf_evlist *evlist;
	u_int i;
	int ret = 0;

	for (i = 0; i < TEST__EVENTS_CNT; i++) {
		struct test__event_st *e = &test__events[i];

		evlist = perf_evlist__new(NULL, NULL);
		if (evlist == NULL)
			break;

		ret = parse_events(evlist, e->name, 0);
		if (ret) {
			pr_debug("failed to parse event '%s', err %d\n",
				 e->name, ret);
			break;
		}

		ret = e->check(evlist);
		if (ret)
			break;

		perf_evlist__delete(evlist);
	}

	return ret;
}

static int sched__get_first_possible_cpu(pid_t pid, cpu_set_t **maskp,
					 size_t *sizep)
{
	cpu_set_t *mask;
	size_t size;
	int i, cpu = -1, nrcpus = 1024;
realloc:
	mask = CPU_ALLOC(nrcpus);
	size = CPU_ALLOC_SIZE(nrcpus);
	CPU_ZERO_S(size, mask);

	if (sched_getaffinity(pid, size, mask) == -1) {
		CPU_FREE(mask);
		if (errno == EINVAL && nrcpus < (1024 << 8)) {
			nrcpus = nrcpus << 2;
			goto realloc;
		}
		perror("sched_getaffinity");
			return -1;
	}

	for (i = 0; i < nrcpus; i++) {
		if (CPU_ISSET_S(i, size, mask)) {
			if (cpu == -1) {
				cpu = i;
				*maskp = mask;
				*sizep = size;
			} else
				CPU_CLR_S(i, size, mask);
		}
	}

	if (cpu == -1)
		CPU_FREE(mask);

	return cpu;
}

static int test__PERF_RECORD(void)
{
	struct perf_record_opts opts = {
		.target_pid = -1,
		.target_tid = -1,
		.no_delay   = true,
		.freq	    = 10,
		.mmap_pages = 256,
		.sample_id_all_avail = true,
	};
	cpu_set_t *cpu_mask = NULL;
	size_t cpu_mask_size = 0;
	struct perf_evlist *evlist = perf_evlist__new(NULL, NULL);
	struct perf_evsel *evsel;
	struct perf_sample sample;
	const char *cmd = "sleep";
	const char *argv[] = { cmd, "1", NULL, };
	char *bname;
	u64 sample_type, prev_time = 0;
	bool found_cmd_mmap = false,
	     found_libc_mmap = false,
	     found_vdso_mmap = false,
	     found_ld_mmap = false;
	int err = -1, errs = 0, i, wakeups = 0, sample_size;
	u32 cpu;
	int total_events = 0, nr_events[PERF_RECORD_MAX] = { 0, };

	if (evlist == NULL || argv == NULL) {
		pr_debug("Not enough memory to create evlist\n");
		goto out;
	}

	/*
	 * We need at least one evsel in the evlist, use the default
	 * one: "cycles".
	 */
	err = perf_evlist__add_default(evlist);
	if (err < 0) {
		pr_debug("Not enough memory to create evsel\n");
		goto out_delete_evlist;
	}

	/*
	 * Create maps of threads and cpus to monitor. In this case
	 * we start with all threads and cpus (-1, -1) but then in
	 * perf_evlist__prepare_workload we'll fill in the only thread
	 * we're monitoring, the one forked there.
	 */
	err = perf_evlist__create_maps(evlist, opts.target_pid,
				       opts.target_tid, opts.cpu_list);
	if (err < 0) {
		pr_debug("Not enough memory to create thread/cpu maps\n");
		goto out_delete_evlist;
	}

	/*
	 * Prepare the workload in argv[] to run, it'll fork it, and then wait
	 * for perf_evlist__start_workload() to exec it. This is done this way
	 * so that we have time to open the evlist (calling sys_perf_event_open
	 * on all the fds) and then mmap them.
	 */
	err = perf_evlist__prepare_workload(evlist, &opts, argv);
	if (err < 0) {
		pr_debug("Couldn't run the workload!\n");
		goto out_delete_evlist;
	}

	/*
	 * Config the evsels, setting attr->comm on the first one, etc.
	 */
	evsel = list_entry(evlist->entries.next, struct perf_evsel, node);
	evsel->attr.sample_type |= PERF_SAMPLE_CPU;
	evsel->attr.sample_type |= PERF_SAMPLE_TID;
	evsel->attr.sample_type |= PERF_SAMPLE_TIME;
	perf_evlist__config_attrs(evlist, &opts);

	err = sched__get_first_possible_cpu(evlist->workload.pid, &cpu_mask,
					    &cpu_mask_size);
	if (err < 0) {
		pr_debug("sched__get_first_possible_cpu: %s\n", strerror(errno));
		goto out_delete_evlist;
	}

	cpu = err;

	/*
	 * So that we can check perf_sample.cpu on all the samples.
	 */
	if (sched_setaffinity(evlist->workload.pid, cpu_mask_size, cpu_mask) < 0) {
		pr_debug("sched_setaffinity: %s\n", strerror(errno));
		goto out_free_cpu_mask;
	}

	/*
	 * Call sys_perf_event_open on all the fds on all the evsels,
	 * grouping them if asked to.
	 */
	err = perf_evlist__open(evlist, opts.group);
	if (err < 0) {
		pr_debug("perf_evlist__open: %s\n", strerror(errno));
		goto out_delete_evlist;
	}

	/*
	 * mmap the first fd on a given CPU and ask for events for the other
	 * fds in the same CPU to be injected in the same mmap ring buffer
	 * (using ioctl(PERF_EVENT_IOC_SET_OUTPUT)).
	 */
	err = perf_evlist__mmap(evlist, opts.mmap_pages, false);
	if (err < 0) {
		pr_debug("perf_evlist__mmap: %s\n", strerror(errno));
		goto out_delete_evlist;
	}

	/*
	 * We'll need these two to parse the PERF_SAMPLE_* fields in each
	 * event.
	 */
	sample_type = perf_evlist__sample_type(evlist);
	sample_size = __perf_evsel__sample_size(sample_type);

	/*
	 * Now that all is properly set up, enable the events, they will
	 * count just on workload.pid, which will start...
	 */
	perf_evlist__enable(evlist);

	/*
	 * Now!
	 */
	perf_evlist__start_workload(evlist);

	while (1) {
		int before = total_events;

		for (i = 0; i < evlist->nr_mmaps; i++) {
			union perf_event *event;

			while ((event = perf_evlist__mmap_read(evlist, i)) != NULL) {
				const u32 type = event->header.type;
				const char *name = perf_event__name(type);

				++total_events;
				if (type < PERF_RECORD_MAX)
					nr_events[type]++;

				err = perf_event__parse_sample(event, sample_type,
							       sample_size, true,
							       &sample, false);
				if (err < 0) {
					if (verbose)
						perf_event__fprintf(event, stderr);
					pr_debug("Couldn't parse sample\n");
					goto out_err;
				}

				if (verbose) {
					pr_info("%" PRIu64" %d ", sample.time, sample.cpu);
					perf_event__fprintf(event, stderr);
				}

				if (prev_time > sample.time) {
					pr_debug("%s going backwards in time, prev=%" PRIu64 ", curr=%" PRIu64 "\n",
						 name, prev_time, sample.time);
					++errs;
				}

				prev_time = sample.time;

				if (sample.cpu != cpu) {
					pr_debug("%s with unexpected cpu, expected %d, got %d\n",
						 name, cpu, sample.cpu);
					++errs;
				}

				if ((pid_t)sample.pid != evlist->workload.pid) {
					pr_debug("%s with unexpected pid, expected %d, got %d\n",
						 name, evlist->workload.pid, sample.pid);
					++errs;
				}

				if ((pid_t)sample.tid != evlist->workload.pid) {
					pr_debug("%s with unexpected tid, expected %d, got %d\n",
						 name, evlist->workload.pid, sample.tid);
					++errs;
				}

				if ((type == PERF_RECORD_COMM ||
				     type == PERF_RECORD_MMAP ||
				     type == PERF_RECORD_FORK ||
				     type == PERF_RECORD_EXIT) &&
				     (pid_t)event->comm.pid != evlist->workload.pid) {
					pr_debug("%s with unexpected pid/tid\n", name);
					++errs;
				}

				if ((type == PERF_RECORD_COMM ||
				     type == PERF_RECORD_MMAP) &&
				     event->comm.pid != event->comm.tid) {
					pr_debug("%s with different pid/tid!\n", name);
					++errs;
				}

				switch (type) {
				case PERF_RECORD_COMM:
					if (strcmp(event->comm.comm, cmd)) {
						pr_debug("%s with unexpected comm!\n", name);
						++errs;
					}
					break;
				case PERF_RECORD_EXIT:
					goto found_exit;
				case PERF_RECORD_MMAP:
					bname = strrchr(event->mmap.filename, '/');
					if (bname != NULL) {
						if (!found_cmd_mmap)
							found_cmd_mmap = !strcmp(bname + 1, cmd);
						if (!found_libc_mmap)
							found_libc_mmap = !strncmp(bname + 1, "libc", 4);
						if (!found_ld_mmap)
							found_ld_mmap = !strncmp(bname + 1, "ld", 2);
					} else if (!found_vdso_mmap)
						found_vdso_mmap = !strcmp(event->mmap.filename, "[vdso]");
					break;

				case PERF_RECORD_SAMPLE:
					/* Just ignore samples for now */
					break;
				default:
					pr_debug("Unexpected perf_event->header.type %d!\n",
						 type);
					++errs;
				}
			}
		}

		/*
		 * We don't use poll here because at least at 3.1 times the
		 * PERF_RECORD_{!SAMPLE} events don't honour
		 * perf_event_attr.wakeup_events, just PERF_EVENT_SAMPLE does.
		 */
		if (total_events == before && false)
			poll(evlist->pollfd, evlist->nr_fds, -1);

		sleep(1);
		if (++wakeups > 5) {
			pr_debug("No PERF_RECORD_EXIT event!\n");
			break;
		}
	}

found_exit:
	if (nr_events[PERF_RECORD_COMM] > 1) {
		pr_debug("Excessive number of PERF_RECORD_COMM events!\n");
		++errs;
	}

	if (nr_events[PERF_RECORD_COMM] == 0) {
		pr_debug("Missing PERF_RECORD_COMM for %s!\n", cmd);
		++errs;
	}

	if (!found_cmd_mmap) {
		pr_debug("PERF_RECORD_MMAP for %s missing!\n", cmd);
		++errs;
	}

	if (!found_libc_mmap) {
		pr_debug("PERF_RECORD_MMAP for %s missing!\n", "libc");
		++errs;
	}

	if (!found_ld_mmap) {
		pr_debug("PERF_RECORD_MMAP for %s missing!\n", "ld");
		++errs;
	}

	if (!found_vdso_mmap) {
		pr_debug("PERF_RECORD_MMAP for %s missing!\n", "[vdso]");
		++errs;
	}
out_err:
	perf_evlist__munmap(evlist);
out_free_cpu_mask:
	CPU_FREE(cpu_mask);
out_delete_evlist:
	perf_evlist__delete(evlist);
out:
	return (err < 0 || errs > 0) ? -1 : 0;
}

static struct test {
	const char *desc;
	int (*func)(void);
} tests[] = {
	{
		.desc = "vmlinux symtab matches kallsyms",
		.func = test__vmlinux_matches_kallsyms,
	},
	{
		.desc = "detect open syscall event",
		.func = test__open_syscall_event,
	},
	{
		.desc = "detect open syscall event on all cpus",
		.func = test__open_syscall_event_on_all_cpus,
	},
	{
		.desc = "read samples using the mmap interface",
		.func = test__basic_mmap,
	},
	{
		.desc = "parse events tests",
		.func = test__parse_events,
	},
	{
		.desc = "Validate PERF_RECORD_* events & perf_sample fields",
		.func = test__PERF_RECORD,
	},
	{
		.func = NULL,
	},
};

static bool perf_test__matches(int curr, int argc, const char *argv[])
{
	int i;

	if (argc == 0)
		return true;

	for (i = 0; i < argc; ++i) {
		char *end;
		long nr = strtoul(argv[i], &end, 10);

		if (*end == '\0') {
			if (nr == curr + 1)
				return true;
			continue;
		}

		if (strstr(tests[curr].desc, argv[i]))
			return true;
	}

	return false;
}

static int __cmd_test(int argc, const char *argv[])
{
	int i = 0;

	while (tests[i].func) {
		int curr = i++, err;

		if (!perf_test__matches(curr, argc, argv))
			continue;

		pr_info("%2d: %s:", i, tests[curr].desc);
		pr_debug("\n--- start ---\n");
		err = tests[curr].func();
		pr_debug("---- end ----\n%s:", tests[curr].desc);
		pr_info(" %s\n", err ? "FAILED!\n" : "Ok");
	}

	return 0;
}

static int perf_test__list(int argc, const char **argv)
{
	int i = 0;

	while (tests[i].func) {
		int curr = i++;

		if (argc > 1 && !strstr(tests[curr].desc, argv[1]))
			continue;

		pr_info("%2d: %s\n", i, tests[curr].desc);
	}

	return 0;
}

int cmd_test(int argc, const char **argv, const char *prefix __used)
{
	const char * const test_usage[] = {
	"perf test [<options>] [{list <test-name-fragment>|[<test-name-fragments>|<test-numbers>]}]",
	NULL,
	};
	const struct option test_options[] = {
	OPT_INTEGER('v', "verbose", &verbose,
		    "be more verbose (show symbol address, etc)"),
	OPT_END()
	};

	argc = parse_options(argc, argv, test_options, test_usage, 0);
	if (argc >= 1 && !strcmp(argv[0], "list"))
		return perf_test__list(argc, argv);

	symbol_conf.priv_size = sizeof(int);
	symbol_conf.sort_by_name = true;
	symbol_conf.try_vmlinux_path = true;

	if (symbol__init() < 0)
		return -1;

	setup_pager();

	return __cmd_test(argc, argv);
}