lists.openwall.net   lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  linux-hardening  linux-cve-announce  PHC 
Open Source and information security mailing list archives
 
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [day] [month] [year] [list]
Message-Id: <1465904596-20619-1-git-send-email-bjorn.topel@gmail.com>
Date:	Tue, 14 Jun 2016 13:43:16 +0200
From:	Björn Töpel <bjorn.topel@...il.com>
To:	acme@...nel.org
Cc:	alexander.shishkin@...ux.intel.com, linux-kernel@...r.kernel.org,
	Björn Töpel <bjorn.topel@...il.com>,
	Björn Töpel <bjorn.topel@...el.com>
Subject: [PATCH] perf stat: Added '--polled-interval-print/-P' mode

When enabled, 'perf stat' will print count deltas, similar to what
'--interval-print' does. However, instead of printing every N
milliseconds, the delta will be output when perf receives a SIGUSR2
signal.

The perf inferior can, e.g. use a function like

  void tell_perf(void)
  {
          kill(getppid(), SIGUSR2);
  }

to make perf output the count delta.

Run example:

  # ./perf stat -a -P -e cycles &
  [1] 7055
  # kill -USR2 7055
  #           time             counts unit events
       6.987756397     21 085 307 064      cycles
  # fg
  ./perf stat -a -P -e cycles
  ^C     9.755385729     25 164 351 653      cycles

Signed-off-by: Björn Töpel <bjorn.topel@...el.com>
---
 tools/perf/builtin-stat.c | 69 ++++++++++++++++++++++++++++++++++++++++-------
 1 file changed, 60 insertions(+), 9 deletions(-)

diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c
index ee7ada7..af7dde1 100644
--- a/tools/perf/builtin-stat.c
+++ b/tools/perf/builtin-stat.c
@@ -61,6 +61,7 @@
 #include "util/counts.h"
 #include "util/session.h"
 #include "util/tool.h"
+#include "util/trigger.h"
 #include "asm/bug.h"
 
 #include <stdlib.h>
@@ -130,6 +131,7 @@ static aggr_get_id_t		aggr_get_id;
 static bool			append_file;
 static const char		*output_name;
 static int			output_fd;
+static bool			polled_interval			= false;
 
 struct perf_stat {
 	bool			 record;
@@ -148,6 +150,8 @@ static struct perf_stat		perf_stat;
 
 static volatile int done = 0;
 
+static DEFINE_TRIGGER(polled_interval_trigger);
+
 static struct perf_stat_config stat_config = {
 	.aggr_mode	= AGGR_GLOBAL,
 	.scale		= true,
@@ -614,10 +618,25 @@ try_again:
 		perf_evlist__start_workload(evsel_list);
 		enable_counters();
 
-		if (interval) {
+		if (interval || polled_interval) {
+			struct timespec rem, *req = &ts;
+
 			while (!waitpid(child_pid, &status, WNOHANG)) {
-				nanosleep(&ts, NULL);
-				process_interval();
+				int err = nanosleep(req, &rem);
+
+				if (trigger_is_hit(&polled_interval_trigger)) {
+					trigger_ready(&polled_interval_trigger);
+
+					if (err && errno == EINTR)
+						req = &rem;
+
+					process_interval();
+					continue;
+				}
+
+				req = &ts;
+				if (interval)
+					process_interval();
 			}
 		}
 		wait(&status);
@@ -631,11 +650,26 @@ try_again:
 		if (WIFSIGNALED(status))
 			psignal(WTERMSIG(status), argv[0]);
 	} else {
+		struct timespec rem, *req = &ts;
+
 		enable_counters();
 		while (!done) {
-			nanosleep(&ts, NULL);
-			if (interval)
-				process_interval();
+			int err = nanosleep(req, &rem);
+			if (interval || polled_interval) {
+				if (trigger_is_hit(&polled_interval_trigger)) {
+					trigger_ready(&polled_interval_trigger);
+
+					if (err && errno == EINTR)
+						req = &rem;
+
+					process_interval();
+					continue;
+				}
+
+				req = &ts;
+				if (interval)
+					process_interval();
+			}
 		}
 	}
 
@@ -1412,7 +1446,7 @@ static void print_footer(void)
 
 static void print_counters(struct timespec *ts, int argc, const char **argv)
 {
-	int interval = stat_config.interval;
+	int interval = stat_config.interval || polled_interval;
 	struct perf_evsel *counter;
 	char buf[64], *prefix = NULL;
 
@@ -1472,11 +1506,20 @@ static void print_counters(struct timespec *ts, int argc, const char **argv)
 
 static volatile int signr = -1;
 
+
+static void polled_interval_signal(int sig __maybe_unused)
+{
+	if (trigger_is_ready(&polled_interval_trigger))
+		trigger_hit(&polled_interval_trigger);
+}
+
 static void skip_signal(int signo)
 {
-	if ((child_pid == -1) || stat_config.interval)
+	if ((child_pid == -1) || stat_config.interval || polled_interval)
 		done = 1;
 
+	polled_interval_signal(signo);
+
 	signr = signo;
 	/*
 	 * render child_pid harmless
@@ -1580,6 +1623,8 @@ static const struct option stat_options[] = {
 		     "ms to wait before starting measurement after program start"),
 	OPT_BOOLEAN(0, "metric-only", &metric_only,
 			"Only print computed metrics. No raw values"),
+	OPT_BOOLEAN('P', "polled-interval-print", &polled_interval,
+		    "print counts when receive SIGUSR2"),
 	OPT_END()
 };
 
@@ -2368,6 +2413,12 @@ int cmd_stat(int argc, const char **argv, const char *prefix __maybe_unused)
 	signal(SIGALRM, skip_signal);
 	signal(SIGABRT, skip_signal);
 
+	if (polled_interval) {
+		signal(SIGUSR2, polled_interval_signal);
+		trigger_on(&polled_interval_trigger);
+		trigger_ready(&polled_interval_trigger);
+	}
+
 	status = 0;
 	for (run_idx = 0; forever || run_idx < run_count; run_idx++) {
 		if (run_count != 1 && verbose)
@@ -2381,7 +2432,7 @@ int cmd_stat(int argc, const char **argv, const char *prefix __maybe_unused)
 		}
 	}
 
-	if (!forever && status != -1 && !interval)
+	if (!forever && status != -1 && !interval && !polled_interval)
 		print_counters(NULL, argc, argv);
 
 	if (STAT_RECORD) {
-- 
2.7.4

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ