tools/perf/builtin-stat.c

/*
 * 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/evsel.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>

#define DEFAULT_SEPARATOR	" "

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 struct cpu_map		*cpus;
static int			run_idx				=  0;

static int			run_count			=  1;
static bool			no_inherit			= false;
static bool			scale				=  true;
static bool			no_aggr				= false;
static pid_t			target_pid			= -1;
static pid_t			target_tid			= -1;
static struct thread_map	*threads;
static pid_t			child_pid			= -1;
static bool			null_run			=  false;
static bool			big_num				=  true;
static int			big_num_opt			=  -1;
static const char		*cpu_list;
static const char		*csv_sep			= NULL;
static bool			csv_output			= false;

static volatile int done = 0;

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

struct perf_stat {
	struct stats	  res_stats[3];
};

static int perf_evsel__alloc_stat_priv(struct perf_evsel *evsel)
{
	evsel->priv = zalloc(sizeof(struct perf_stat));
	return evsel->priv == NULL ? -ENOMEM : 0;
}

static void perf_evsel__free_stat_priv(struct perf_evsel *evsel)
{
	free(evsel->priv);
	evsel->priv = NULL;
}

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			runtime_nsecs_stats[MAX_NR_CPUS];
struct stats			runtime_cycles_stats[MAX_NR_CPUS];
struct stats			runtime_branches_stats[MAX_NR_CPUS];
struct stats			walltime_nsecs_stats;

static int create_perf_stat_counter(struct perf_evsel *evsel)
{
	struct perf_event_attr *attr = &evsel->attr;

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

	if (system_wide)
		return perf_evsel__open_per_cpu(evsel, cpus);

	attr->inherit = !no_inherit;
	if (target_pid == -1 && target_tid == -1) {
		attr->disabled = 1;
		attr->enable_on_exec = 1;
	}

	return perf_evsel__open_per_thread(evsel, threads);
}

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

	return 0;
}

/*
 * Read out the results of a single counter:
 * aggregate counts across CPUs in system-wide mode
 */
static int read_counter_aggr(struct perf_evsel *counter)
{
	struct perf_stat *ps = counter->priv;
	u64 *count = counter->counts->aggr.values;
	int i;

	if (__perf_evsel__read(counter, cpus->nr, threads->nr, scale) < 0)
		return -1;

	for (i = 0; i < 3; i++)
		update_stats(&ps->res_stats[i], count[i]);

	if (verbose) {
		fprintf(stderr, "%s: %" PRIu64 " %" PRIu64 " %" PRIu64 "\n",
			event_name(counter), count[0], count[1], count[2]);
	}

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

	return 0;
}

/*
 * Read out the results of a single counter:
 * do not aggregate counts across CPUs in system-wide mode
 */
static int read_counter(struct perf_evsel *counter)
{
	u64 *count;
	int cpu;

	for (cpu = 0; cpu < cpus->nr; cpu++) {
		if (__perf_evsel__read_on_cpu(counter, cpu, 0, scale) < 0)
			return -1;

		count = counter->counts->cpu[cpu].values;

		if (perf_evsel__match(counter, SOFTWARE, SW_TASK_CLOCK))
			update_stats(&runtime_nsecs_stats[cpu], count[0]);
		if (perf_evsel__match(counter, HARDWARE, HW_CPU_CYCLES))
			update_stats(&runtime_cycles_stats[cpu], count[0]);
		if (perf_evsel__match(counter, HARDWARE, HW_BRANCH_INSTRUCTIONS))
			update_stats(&runtime_branches_stats[cpu], count[0]);
	}

	return 0;
}

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

	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)
			threads->map[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]);
	}

	list_for_each_entry(counter, &evsel_list, node) {
		if (create_perf_stat_counter(counter) < 0) {
			if (errno == -EPERM || errno == -EACCES) {
				error("You may not have permission to collect %sstats.\n"
				      "\t Consider tweaking"
				      " /proc/sys/kernel/perf_event_paranoid or running as root.",
				      system_wide ? "system-wide " : "");
			} else if (errno == ENOENT) {
				error("%s event is not supported. ", event_name(counter));
			} else {
				error("open_counter returned with %d (%s). "
				      "/bin/dmesg may provide additional information.\n",
				       errno, strerror(errno));
			}
			if (child_pid != -1)
				kill(child_pid, SIGTERM);
			die("Not all events could be opened.\n");
			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);

	if (no_aggr) {
		list_for_each_entry(counter, &evsel_list, node) {
			read_counter(counter);
			perf_evsel__close_fd(counter, cpus->nr, 1);
		}
	} else {
		list_for_each_entry(counter, &evsel_list, node) {
			read_counter_aggr(counter);
			perf_evsel__close_fd(counter, cpus->nr, threads->nr);
		}
	}

	return WEXITSTATUS(status);
}

static void print_noise(struct perf_evsel *evsel, double avg)
{
	struct perf_stat *ps;

	if (run_count == 1)
		return;

	ps = evsel->priv;
	fprintf(stderr, "   ( +- %7.3f%% )",
			100 * stddev_stats(&ps->res_stats[0]) / avg);
}

static void nsec_printout(int cpu, struct perf_evsel *evsel, double avg)
{
	double msecs = avg / 1e6;
	char cpustr[16] = { '\0', };
	const char *fmt = csv_output ? "%s%.6f%s%s" : "%s%18.6f%s%-24s";

	if (no_aggr)
		sprintf(cpustr, "CPU%*d%s",
			csv_output ? 0 : -4,
			cpus->map[cpu], csv_sep);

	fprintf(stderr, fmt, cpustr, msecs, csv_sep, event_name(evsel));

	if (csv_output)
		return;

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

static void abs_printout(int cpu, struct perf_evsel *evsel, double avg)
{
	double total, ratio = 0.0;
	char cpustr[16] = { '\0', };
	const char *fmt;

	if (csv_output)
		fmt = "%s%.0f%s%s";
	else if (big_num)
		fmt = "%s%'18.0f%s%-24s";
	else
		fmt = "%s%18.0f%s%-24s";

	if (no_aggr)
		sprintf(cpustr, "CPU%*d%s",
			csv_output ? 0 : -4,
			cpus->map[cpu], csv_sep);
	else
		cpu = 0;

	fprintf(stderr, fmt, cpustr, avg, csv_sep, event_name(evsel));

	if (csv_output)
		return;

	if (perf_evsel__match(evsel, HARDWARE, HW_INSTRUCTIONS)) {
		total = avg_stats(&runtime_cycles_stats[cpu]);

		if (total)
			ratio = avg / total;

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

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

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

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

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

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

/*
 * Print out the results of a single counter:
 * aggregated counts in system-wide mode
 */
static void print_counter_aggr(struct perf_evsel *counter)
{
	struct perf_stat *ps = counter->priv;
	double avg = avg_stats(&ps->res_stats[0]);
	int scaled = counter->counts->scaled;

	if (scaled == -1) {
		fprintf(stderr, "%*s%s%-24s\n",
			csv_output ? 0 : 18,
			"<not counted>", csv_sep, event_name(counter));
		return;
	}

	if (nsec_counter(counter))
		nsec_printout(-1, counter, avg);
	else
		abs_printout(-1, counter, avg);

	if (csv_output) {
		fputc('\n', stderr);
		return;
	}

	print_noise(counter, avg);

	if (scaled) {
		double avg_enabled, avg_running;

		avg_enabled = avg_stats(&ps->res_stats[1]);
		avg_running = avg_stats(&ps->res_stats[2]);

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

	fprintf(stderr, "\n");
}

/*
 * Print out the results of a single counter:
 * does not use aggregated count in system-wide
 */
static void print_counter(struct perf_evsel *counter)
{
	u64 ena, run, val;
	int cpu;

	for (cpu = 0; cpu < cpus->nr; cpu++) {
		val = counter->counts->cpu[cpu].val;
		ena = counter->counts->cpu[cpu].ena;
		run = counter->counts->cpu[cpu].run;
		if (run == 0 || ena == 0) {
			fprintf(stderr, "CPU%*d%s%*s%s%-24s",
				csv_output ? 0 : -4,
				cpus->map[cpu], csv_sep,
				csv_output ? 0 : 18,
				"<not counted>", csv_sep,
				event_name(counter));

			fprintf(stderr, "\n");
			continue;
		}

		if (nsec_counter(counter))
			nsec_printout(cpu, counter, val);
		else
			abs_printout(cpu, counter, val);

		if (!csv_output) {
			print_noise(counter, 1.0);

			if (run != ena) {
				fprintf(stderr, "  (scaled from %.2f%%)",
					100.0 * run / ena);
			}
		}
		fprintf(stderr, "\n");
	}
}

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

	fflush(stdout);

	if (!csv_output) {
		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");
	}

	if (no_aggr) {
		list_for_each_entry(counter, &evsel_list, node)
			print_counter(counter);
	} else {
		list_for_each_entry(counter, &evsel_list, node)
			print_counter_aggr(counter);
	}

	if (!csv_output) {
		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 int stat__set_big_num(const struct option *opt __used,
			     const char *s __used, int unset)
{
	big_num_opt = unset ? 0 : 1;
	return 0;
}

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_CALLBACK_NOOPT('B', "big-num", NULL, NULL,
			   "print large numbers with thousands\' separators",
			   stat__set_big_num),
	OPT_STRING('C', "cpu", &cpu_list, "cpu",
		    "list of cpus to monitor in system-wide"),
	OPT_BOOLEAN('A', "no-aggr", &no_aggr,
		    "disable CPU count aggregation"),
	OPT_STRING('x', "field-separator", &csv_sep, "separator",
		   "print counts with custom separator"),
	OPT_END()
};

int cmd_stat(int argc, const char **argv, const char *prefix __used)
{
	struct perf_evsel *pos;
	int status = -ENOMEM;

	setlocale(LC_ALL, "");

	argc = parse_options(argc, argv, options, stat_usage,
		PARSE_OPT_STOP_AT_NON_OPTION);

	if (csv_sep)
		csv_output = true;
	else
		csv_sep = DEFAULT_SEPARATOR;

	/*
	 * let the spreadsheet do the pretty-printing
	 */
	if (csv_output) {
		/* User explicitely passed -B? */
		if (big_num_opt == 1) {
			fprintf(stderr, "-B option not supported with -x\n");
			usage_with_options(stat_usage, options);
		} else /* Nope, so disable big number formatting */
			big_num = false;
	} else if (big_num_opt == 0) /* User passed --no-big-num */
		big_num = false;

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

	/* no_aggr is for system-wide only */
	if (no_aggr && !system_wide)
		usage_with_options(stat_usage, options);

	/* Set attrs and nr_counters if no event is selected and !null_run */
	if (!null_run && !nr_counters) {
		size_t c;

		nr_counters = ARRAY_SIZE(default_attrs);

		for (c = 0; c < ARRAY_SIZE(default_attrs); ++c) {
			pos = perf_evsel__new(&default_attrs[c],
					      nr_counters);
			if (pos == NULL)
				goto out;
			list_add(&pos->node, &evsel_list);
		}
	}

	if (target_pid != -1)
		target_tid = target_pid;

	threads = thread_map__new(target_pid, target_tid);
	if (threads == NULL) {
		pr_err("Problems finding threads of monitor\n");
		usage_with_options(stat_usage, options);
	}

	if (system_wide)
		cpus = cpu_map__new(cpu_list);
	else
		cpus = cpu_map__dummy_new();

	if (cpus == NULL) {
		perror("failed to parse CPUs map");
		usage_with_options(stat_usage, options);
		return -1;
	}

	list_for_each_entry(pos, &evsel_list, node) {
		if (perf_evsel__alloc_stat_priv(pos) < 0 ||
		    perf_evsel__alloc_counts(pos, cpus->nr) < 0 ||
		    perf_evsel__alloc_fd(pos, cpus->nr, threads->nr) < 0)
			goto out_free_fd;
	}

	/*
	 * 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);
out_free_fd:
	list_for_each_entry(pos, &evsel_list, node)
		perf_evsel__free_stat_priv(pos);
	perf_evsel_list__delete();
out:
	thread_map__delete(threads);
	threads = NULL;
	return status;
}