aboutsummaryrefslogtreecommitdiffstats
path: root/tools/perf/builtin-stat.c
blob: a6b4d44f950246e27d4cb6b0bc3e6d5afd27adcc (plain) (blame)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
/*
 * builtin-stat.c
 *
 * Builtin stat command: Give a precise performance counters summary
 * overview about any workload, CPU or specific PID.
 *
 * Sample output:

   $ perf stat ~/hackbench 10
   Time: 0.104

    Performance counter stats for '/home/mingo/hackbench':

       1255.538611  task clock ticks     #      10.143 CPU utilization factor
             54011  context switches     #       0.043 M/sec
               385  CPU migrations       #       0.000 M/sec
             17755  pagefaults           #       0.014 M/sec
        3808323185  CPU cycles           #    3033.219 M/sec
        1575111190  instructions         #    1254.530 M/sec
          17367895  cache references     #      13.833 M/sec
           7674421  cache misses         #       6.112 M/sec

    Wall-clock time elapsed:   123.786620 msecs

 *
 * Copyright (C) 2008, Red Hat Inc, Ingo Molnar <mingo@redhat.com>
 *
 * Improvements and fixes by:
 *
 *   Arjan van de Ven <arjan@linux.intel.com>
 *   Yanmin Zhang <yanmin.zhang@intel.com>
 *   Wu Fengguang <fengguang.wu@intel.com>
 *   Mike Galbraith <efault@gmx.de>
 *   Paul Mackerras <paulus@samba.org>
 *   Jaswinder Singh Rajput <jaswinder@kernel.org>
 *
 * Released under the GPL v2. (and only v2, not any later version)
 */

#include "perf.h"
#include "builtin.h"
#include "util/util.h"
#include "util/parse-options.h"
#include "util/parse-events.h"
#include "util/event.h"
#include "util/debug.h"
#include "util/header.h"
#include "util/cpumap.h"
#include "util/thread.h"

#include <sys/prctl.h>
#include <math.h>
#include <locale.h>

static struct perf_event_attr default_attrs[] = {

  { .type = PERF_TYPE_SOFTWARE, .config = PERF_COUNT_SW_TASK_CLOCK		},
  { .type = PERF_TYPE_SOFTWARE, .config = PERF_COUNT_SW_CONTEXT_SWITCHES	},
  { .type = PERF_TYPE_SOFTWARE, .config = PERF_COUNT_SW_CPU_MIGRATIONS		},
  { .type = PERF_TYPE_SOFTWARE, .config = PERF_COUNT_SW_PAGE_FAULTS		},

  { .type = PERF_TYPE_HARDWARE, .config = PERF_COUNT_HW_CPU_CYCLES		},
  { .type = PERF_TYPE_HARDWARE, .config = PERF_COUNT_HW_INSTRUCTIONS		},
  { .type = PERF_TYPE_HARDWARE, .config = PERF_COUNT_HW_BRANCH_INSTRUCTIONS	},
  { .type = PERF_TYPE_HARDWARE, .config = PERF_COUNT_HW_BRANCH_MISSES		},
  { .type = PERF_TYPE_HARDWARE, .config = PERF_COUNT_HW_CACHE_REFERENCES	},
  { .type = PERF_TYPE_HARDWARE, .config = PERF_COUNT_HW_CACHE_MISSES		},

};

static bool			system_wide			=  false;
static int			nr_cpus				=  0;
static int			run_idx				=  0;

static int			run_count			=  1;
static bool			no_inherit			= false;
static bool			scale				=  true;
static pid_t			target_pid			= -1;
static pid_t			target_tid			= -1;
static pid_t			*all_tids			=  NULL;
static int			thread_num			=  0;
static pid_t			child_pid			= -1;
static bool			null_run			=  false;
static bool			big_num				=  false;
static const char		*cpu_list;


static int			*fd[MAX_NR_CPUS][MAX_COUNTERS];

static int			event_scaled[MAX_COUNTERS];

static volatile int done = 0;

struct stats
{
	double n, mean, M2;
};

static void update_stats(struct stats *stats, u64 val)
{
	double delta;

	stats->n++;
	delta = val - stats->mean;
	stats->mean += delta / stats->n;
	stats->M2 += delta*(val - stats->mean);
}

static double avg_stats(struct stats *stats)
{
	return stats->mean;
}

/*
 * http://en.wikipedia.org/wiki/Algorithms_for_calculating_variance
 *
 *       (\Sum n_i^2) - ((\Sum n_i)^2)/n
 * s^2 = -------------------------------
 *                  n - 1
 *
 * http://en.wikipedia.org/wiki/Stddev
 *
 * The std dev of the mean is related to the std dev by:
 *
 *             s
 * s_mean = -------
 *          sqrt(n)
 *
 */
static double stddev_stats(struct stats *stats)
{
	double variance = stats->M2 / (stats->n - 1);
	double variance_mean = variance / stats->n;

	return sqrt(variance_mean);
}

struct stats			event_res_stats[MAX_COUNTERS][3];
struct stats			runtime_nsecs_stats;
struct stats			walltime_nsecs_stats;
struct stats			runtime_cycles_stats;
struct stats			runtime_branches_stats;

#define MATCH_EVENT(t, c, counter)			\
	(attrs[counter].type == PERF_TYPE_##t &&	\
	 attrs[counter].config == PERF_COUNT_##c)

#define ERR_PERF_OPEN \
"Error: counter %d, sys_perf_event_open() syscall returned with %d (%s)\n"

static int create_perf_stat_counter(int counter)
{
	struct perf_event_attr *attr = attrs + counter;
	int thread;
	int ncreated = 0;

	if (scale)
		attr->read_format = PERF_FORMAT_TOTAL_TIME_ENABLED |
				    PERF_FORMAT_TOTAL_TIME_RUNNING;

	if (system_wide) {
		int cpu;

		for (cpu = 0; cpu < nr_cpus; cpu++) {
			fd[cpu][counter][0] = sys_perf_event_open(attr,
					-1, cpumap[cpu], -1, 0);
			if (fd[cpu][counter][0] < 0)
				pr_debug(ERR_PERF_OPEN, counter,
					 fd[cpu][counter][0], strerror(errno));
			else
				++ncreated;
		}
	} else {
		attr->inherit = !no_inherit;
		if (target_pid == -1 && target_tid == -1) {
			attr->disabled = 1;
			attr->enable_on_exec = 1;
		}
		for (thread = 0; thread < thread_num; thread++) {
			fd[0][counter][thread] = sys_perf_event_open(attr,
				all_tids[thread], -1, -1, 0);
			if (fd[0][counter][thread] < 0)
				pr_debug(ERR_PERF_OPEN, counter,
					 fd[0][counter][thread],
					 strerror(errno));
			else
				++ncreated;
		}
	}

	return ncreated;
}

/*
 * Does the counter have nsecs as a unit?
 */
static inline int nsec_counter(int counter)
{
	if (MATCH_EVENT(SOFTWARE, SW_CPU_CLOCK, counter) ||
	    MATCH_EVENT(SOFTWARE, SW_TASK_CLOCK, counter))
		return 1;

	return 0;
}

/*
 * Read out the results of a single counter:
 */
static void read_counter(int counter)
{
	u64 count[3], single_count[3];
	int cpu;
	size_t res, nv;
	int scaled;
	int i, thread;

	count[0] = count[1] = count[2] = 0;

	nv = scale ? 3 : 1;
	for (cpu = 0; cpu < nr_cpus; cpu++) {
		for (thread = 0; thread < thread_num; thread++) {
			if (fd[cpu][counter][thread] < 0)
				continue;

			res = read(fd[cpu][counter][thread],
					single_count, nv * sizeof(u64));
			assert(res == nv * sizeof(u64));

			close(fd[cpu][counter][thread]);
			fd[cpu][counter][thread] = -1;

			count[0] += single_count[0];
			if (scale) {
				count[1] += single_count[1];
				count[2] += single_count[2];
			}
		}
	}

	scaled = 0;
	if (scale) {
		if (count[2] == 0) {
			event_scaled[counter] = -1;
			count[0] = 0;
			return;
		}

		if (count[2] < count[1]) {
			event_scaled[counter] = 1;
			count[0] = (unsigned long long)
				((double)count[0] * count[1] / count[2] + 0.5);
		}
	}

	for (i = 0; i < 3; i++)
		update_stats(&event_res_stats[counter][i], count[i]);

	if (verbose) {
		fprintf(stderr, "%s: %Ld %Ld %Ld\n", event_name(counter),
				count[0], count[1], count[2]);
	}

	/*
	 * Save the full runtime - to allow normalization during printout:
	 */
	if (MATCH_EVENT(SOFTWARE, SW_TASK_CLOCK, counter))
		update_stats(&runtime_nsecs_stats, count[0]);
	if (MATCH_EVENT(HARDWARE, HW_CPU_CYCLES, counter))
		update_stats(&runtime_cycles_stats, count[0]);
	if (MATCH_EVENT(HARDWARE, HW_BRANCH_INSTRUCTIONS, counter))
		update_stats(&runtime_branches_stats, count[0]);
}

static int run_perf_stat(int argc __used, const char **argv)
{
	unsigned long long t0, t1;
	int status = 0;
	int counter, ncreated = 0;
	int child_ready_pipe[2], go_pipe[2];
	const bool forks = (argc > 0);
	char buf;

	if (!system_wide)
		nr_cpus = 1;

	if (forks && (pipe(child_ready_pipe) < 0 || pipe(go_pipe) < 0)) {
		perror("failed to create pipes");
		exit(1);
	}

	if (forks) {
		if ((child_pid = fork()) < 0)
			perror("failed to fork");

		if (!child_pid) {
			close(child_ready_pipe[0]);
			close(go_pipe[1]);
			fcntl(go_pipe[0], F_SETFD, FD_CLOEXEC);

			/*
			 * Do a dummy execvp to get the PLT entry resolved,
			 * so we avoid the resolver overhead on the real
			 * execvp call.
			 */
			execvp("", (char **)argv);

			/*
			 * Tell the parent we're ready to go
			 */
			close(child_ready_pipe[1]);

			/*
			 * Wait until the parent tells us to go.
			 */
			if (read(go_pipe[0], &buf, 1) == -1)
				perror("unable to read pipe");

			execvp(argv[0], (char **)argv);

			perror(argv[0]);
			exit(-1);
		}

		if (target_tid == -1 && target_pid == -1 && !system_wide)
			all_tids[0] = child_pid;

		/*
		 * Wait for the child to be ready to exec.
		 */
		close(child_ready_pipe[1]);
		close(go_pipe[0]);
		if (read(child_ready_pipe[0], &buf, 1) == -1)
			perror("unable to read pipe");
		close(child_ready_pipe[0]);
	}

	for (counter = 0; counter < nr_counters; counter++)
		ncreated += create_perf_stat_counter(counter);

	if (ncreated == 0) {
		pr_err("No permission to collect %sstats.\n"
		       "Consider tweaking /proc/sys/kernel/perf_event_paranoid.\n",
		       system_wide ? "system-wide " : "");
		if (child_pid != -1)
			kill(child_pid, SIGTERM);
		return -1;
	}

	/*
	 * Enable counters and exec the command:
	 */
	t0 = rdclock();

	if (forks) {
		close(go_pipe[1]);
		wait(&status);
	} else {
		while(!done) sleep(1);
	}

	t1 = rdclock();

	update_stats(&walltime_nsecs_stats, t1 - t0);

	for (counter = 0; counter < nr_counters; counter++)
		read_counter(counter);

	return WEXITSTATUS(status);
}

static void print_noise(int counter, double avg)
{
	if (run_count == 1)
		return;

	fprintf(stderr, "   ( +- %7.3f%% )",
			100 * stddev_stats(&event_res_stats[counter][0]) / avg);
}

static void nsec_printout(int counter, double avg)
{
	double msecs = avg / 1e6;

	fprintf(stderr, " %18.6f  %-24s", msecs, event_name(counter));

	if (MATCH_EVENT(SOFTWARE, SW_TASK_CLOCK, counter)) {
		fprintf(stderr, " # %10.3f CPUs ",
				avg / avg_stats(&walltime_nsecs_stats));
	}
}

static void abs_printout(int counter, double avg)
{
	double total, ratio = 0.0;

	if (big_num)
		fprintf(stderr, " %'18.0f  %-24s", avg, event_name(counter));
	else
		fprintf(stderr, " %18.0f  %-24s", avg, event_name(counter));

	if (MATCH_EVENT(HARDWARE, HW_INSTRUCTIONS, counter)) {
		total = avg_stats(&runtime_cycles_stats);

		if (total)
			ratio = avg / total;

		fprintf(stderr, " # %10.3f IPC  ", ratio);
	} else if (MATCH_EVENT(HARDWARE, HW_BRANCH_MISSES, counter) &&
			runtime_branches_stats.n != 0) {
		total = avg_stats(&runtime_branches_stats);

		if (total)
			ratio = avg * 100 / total;

		fprintf(stderr, " # %10.3f %%    ", ratio);

	} else if (runtime_nsecs_stats.n != 0) {
		total = avg_stats(&runtime_nsecs_stats);

		if (total)
			ratio = 1000.0 * avg / total;

		fprintf(stderr, " # %10.3f M/sec", ratio);
	}
}

/*
 * Print out the results of a single counter:
 */
static void print_counter(int counter)
{
	double avg = avg_stats(&event_res_stats[counter][0]);
	int scaled = event_scaled[counter];

	if (scaled == -1) {
		fprintf(stderr, " %18s  %-24s\n",
			"<not counted>", event_name(counter));
		return;
	}

	if (nsec_counter(counter))
		nsec_printout(counter, avg);
	else
		abs_printout(counter, avg);

	print_noise(counter, avg);

	if (scaled) {
		double avg_enabled, avg_running;

		avg_enabled = avg_stats(&event_res_stats[counter][1]);
		avg_running = avg_stats(&event_res_stats[counter][2]);

		fprintf(stderr, "  (scaled from %.2f%%)",
				100 * avg_running / avg_enabled);
	}

	fprintf(stderr, "\n");
}

static void print_stat(int argc, const char **argv)
{
	int i, counter;

	fflush(stdout);

	fprintf(stderr, "\n");
	fprintf(stderr, " Performance counter stats for ");
	if(target_pid == -1 && target_tid == -1) {
		fprintf(stderr, "\'%s", argv[0]);
		for (i = 1; i < argc; i++)
			fprintf(stderr, " %s", argv[i]);
	} else if (target_pid != -1)
		fprintf(stderr, "process id \'%d", target_pid);
	else
		fprintf(stderr, "thread id \'%d", target_tid);

	fprintf(stderr, "\'");
	if (run_count > 1)
		fprintf(stderr, " (%d runs)", run_count);
	fprintf(stderr, ":\n\n");

	for (counter = 0; counter < nr_counters; counter++)
		print_counter(counter);

	fprintf(stderr, "\n");
	fprintf(stderr, " %18.9f  seconds time elapsed",
			avg_stats(&walltime_nsecs_stats)/1e9);
	if (run_count > 1) {
		fprintf(stderr, "   ( +- %7.3f%% )",
				100*stddev_stats(&walltime_nsecs_stats) /
				avg_stats(&walltime_nsecs_stats));
	}
	fprintf(stderr, "\n\n");
}

static volatile int signr = -1;

static void skip_signal(int signo)
{
	if(child_pid == -1)
		done = 1;

	signr = signo;
}

static void sig_atexit(void)
{
	if (child_pid != -1)
		kill(child_pid, SIGTERM);

	if (signr == -1)
		return;

	signal(signr, SIG_DFL);
	kill(getpid(), signr);
}

static const char * const stat_usage[] = {
	"perf stat [<options>] [<command>]",
	NULL
};

static const struct option options[] = {
	OPT_CALLBACK('e', "event", NULL, "event",
		     "event selector. use 'perf list' to list available events",
		     parse_events),
	OPT_BOOLEAN('i', "no-inherit", &no_inherit,
		    "child tasks do not inherit counters"),
	OPT_INTEGER('p', "pid", &target_pid,
		    "stat events on existing process id"),
	OPT_INTEGER('t', "tid", &target_tid,
		    "stat events on existing thread id"),
	OPT_BOOLEAN('a', "all-cpus", &system_wide,
		    "system-wide collection from all CPUs"),
	OPT_BOOLEAN('c', "scale", &scale,
		    "scale/normalize counters"),
	OPT_INCR('v', "verbose", &verbose,
		    "be more verbose (show counter open errors, etc)"),
	OPT_INTEGER('r', "repeat", &run_count,
		    "repeat command and print average + stddev (max: 100)"),
	OPT_BOOLEAN('n', "null", &null_run,
		    "null run - dont start any counters"),
	OPT_BOOLEAN('B', "big-num", &big_num,
		    "print large numbers with thousands\' separators"),
	OPT_STRING('C', "cpu", &cpu_list, "cpu",
		    "list of cpus to monitor in system-wide"),
	OPT_END()
};

int cmd_stat(int argc, const char **argv, const char *prefix __used)
{
	int status;
	int i,j;

	setlocale(LC_ALL, "");

	argc = parse_options(argc, argv, options, stat_usage,
		PARSE_OPT_STOP_AT_NON_OPTION);
	if (!argc && target_pid == -1 && target_tid == -1)
		usage_with_options(stat_usage, options);
	if (run_count <= 0)
		usage_with_options(stat_usage, options);

	/* Set attrs and nr_counters if no event is selected and !null_run */
	if (!null_run && !nr_counters) {
		memcpy(attrs, default_attrs, sizeof(default_attrs));
		nr_counters = ARRAY_SIZE(default_attrs);
	}

	if (system_wide)
		nr_cpus = read_cpu_map(cpu_list);
	else
		nr_cpus = 1;

	if (nr_cpus < 1)
		usage_with_options(stat_usage, options);

	if (target_pid != -1) {
		target_tid = target_pid;
		thread_num = find_all_tid(target_pid, &all_tids);
		if (thread_num <= 0) {
			fprintf(stderr, "Can't find all threads of pid %d\n",
					target_pid);
			usage_with_options(stat_usage, options);
		}
	} else {
		all_tids=malloc(sizeof(pid_t));
		if (!all_tids)
			return -ENOMEM;

		all_tids[0] = target_tid;
		thread_num = 1;
	}

	for (i = 0; i < MAX_NR_CPUS; i++) {
		for (j = 0; j < MAX_COUNTERS; j++) {
			fd[i][j] = malloc(sizeof(int)*thread_num);
			if (!fd[i][j])
				return -ENOMEM;
		}
	}

	/*
	 * We dont want to block the signals - that would cause
	 * child tasks to inherit that and Ctrl-C would not work.
	 * What we want is for Ctrl-C to work in the exec()-ed
	 * task, but being ignored by perf stat itself:
	 */
	atexit(sig_atexit);
	signal(SIGINT,  skip_signal);
	signal(SIGALRM, skip_signal);
	signal(SIGABRT, skip_signal);

	status = 0;
	for (run_idx = 0; run_idx < run_count; run_idx++) {
		if (run_count != 1 && verbose)
			fprintf(stderr, "[ perf stat: executing run #%d ... ]\n", run_idx + 1);
		status = run_perf_stat(argc, argv);
	}

	if (status != -1)
		print_stat(argc, argv);

	return status;
}