[PATCH v4 0/2] perf stat: Add interval-count and time support

[PATCH v4 0/2] perf stat: Add interval-count and time support

[ufo19890607]

From: yuzhoujian <yuzhoujian@didichuxing.com>

Introduce two new options for perf stat and update perf-stat documentation
accordingly.

The interval-count option can be used to print counts for fixed number of
times, and it should be used specifically with "-I" option.

Show below is the output of the interval-count option for perf stat.

        $ perf stat -I 1000 --interval-count 2 -e cycles -a
        #           time             counts unit events
             1.002827089         93,884,870      cycles
             2.004231506         56,573,446      cycles
    
The time option can be used to print counts after a period of time, and it
should not be used with "-I" option.

Show below is the output of the time option for perf stat.

        $ perf stat --time 2000 -e cycles -a
        Performance counter stats for 'system wide':

                157,260,423      cycles

                2.003060766 seconds time elapsed

yuzhoujian (2):
  perf stat: Add support to print counts for fixed times
  perf stat: Add support to print counts after a period of time

Changes since v3: 
- merge interval_count check and times check to one line.
- fix the wrong indent in stat.h
- use stat_config.times instead of 'times' in cmd_stat function.

Changes since v2: 
- modify the time check in __run_perf_stat func to keep some consistency
  with the workload case.
- add the warning when the time is set between 10ms to 100ms.
- add the pr_err when the time is set below 10ms.

Changes since v1:
- change the name of the new option "times-print" to "interval-count".
- keep the interval-count option interval specifically.


 tools/perf/Documentation/perf-stat.txt | 10 +++++++
 tools/perf/builtin-stat.c              | 53 ++++++++++++++++++++++++++++++++--
 tools/perf/util/stat.h                 |  2 ++
 3 files changed, 62 insertions(+), 3 deletions(-)

-- 
2.14.1

[PATCH v4 1/2] perf stat: Add support to print counts for fixed times

[ufo19890607]

From: yuzhoujian <yuzhoujian@didichuxing.com>

Introduce a new option to print counts for fixed number of times
and update perf-stat documentation accordingly.

Show below is the output of the new option for perf stat.

        $ perf stat -I 1000 --interval-count 2 -e cycles -a
        #           time             counts unit events
             1.002827089         93,884,870      cycles
             2.004231506         56,573,446      cycles

We can just print the counts for several times with this newly introduced
option. The usage of it is a little like vmstat, and it should be used
together with "-I" option.

        $ vmstat -n 1 2
procs -----------memory---------- ---swap-- -----io---- -system-- ------cpu-----
 r  b   swpd   free   buff  cache   si   so    bi    bo   in   cs us sy id wa st
 0  0      0 78270544 547484 51732076    0    0     0    20    1    1  1  0 99  0  0
 0  0      0 78270512 547484 51732080    0    0     0    16  477 1555  0  0 100  0  0

Changes since v3:
- merge interval_count check and times check to one line.
- fix the wrong indent in stat.h
- use stat_config.times instead of 'times' in cmd_stat function.

Changes since v2:
- none.

Changes since v1:
- change the name of the new option "times-print" to "interval-count".
- keep the new option interval specifically.

Signed-off-by: yuzhoujian <yuzhoujian@didichuxing.com>
---
 tools/perf/Documentation/perf-stat.txt |  5 +++++
 tools/perf/builtin-stat.c              | 20 +++++++++++++++++++-
 tools/perf/util/stat.h                 |  1 +
 3 files changed, 25 insertions(+), 1 deletion(-)

diff --git a/tools/perf/Documentation/perf-stat.txt b/tools/perf/Documentation/perf-stat.txt
index 823fce7674bb..47a21645f60c 100644
--- a/tools/perf/Documentation/perf-stat.txt
+++ b/tools/perf/Documentation/perf-stat.txt
@@ -146,6 +146,11 @@ Print count deltas every N milliseconds (minimum: 10ms)
 The overhead percentage could be high in some cases, for instance with small, sub 100ms intervals.  Use with caution.
 	example: 'perf stat -I 1000 -e cycles -a sleep 5'
 
+--interval-count times::
+Print count deltas for fixed number of times.
+This option should be used together with "-I" option.
+	example: 'perf stat -I 1000 --interval-count 2 -e cycles -a'
+
 --metric-only::
 Only print computed metrics. Print them in a single line.
 Don't show any raw values. Not supported with --per-thread.
diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c
index 98bf9d32f222..7d1d7613bf56 100644
--- a/tools/perf/builtin-stat.c
+++ b/tools/perf/builtin-stat.c
@@ -168,6 +168,7 @@ static struct timespec		ref_time;
 static struct cpu_map		*aggr_map;
 static aggr_get_id_t		aggr_get_id;
 static bool			append_file;
+static bool			interval_count;
 static const char		*output_name;
 static int			output_fd;
 static int			print_free_counters_hint;
@@ -571,6 +572,7 @@ static struct perf_evsel *perf_evsel__reset_weak_group(struct perf_evsel *evsel)
 static int __run_perf_stat(int argc, const char **argv)
 {
 	int interval = stat_config.interval;
+	int times = stat_config.times;
 	char msg[BUFSIZ];
 	unsigned long long t0, t1;
 	struct perf_evsel *counter;
@@ -700,6 +702,8 @@ static int __run_perf_stat(int argc, const char **argv)
 			while (!waitpid(child_pid, &status, WNOHANG)) {
 				nanosleep(&ts, NULL);
 				process_interval();
+				if (interval_count && !(--times))
+					break;
 			}
 		}
 		waitpid(child_pid, &status, 0);
@@ -716,8 +720,11 @@ static int __run_perf_stat(int argc, const char **argv)
 		enable_counters();
 		while (!done) {
 			nanosleep(&ts, NULL);
-			if (interval)
+			if (interval) {
 				process_interval();
+				if (interval_count && !(--times))
+					break;
+			}
 		}
 	}
 
@@ -1891,6 +1898,8 @@ static const struct option stat_options[] = {
 			"command to run after to the measured command"),
 	OPT_UINTEGER('I', "interval-print", &stat_config.interval,
 		    "print counts at regular interval in ms (>= 10)"),
+	OPT_INTEGER(0, "interval-count", &stat_config.times,
+		    "print counts for fixed number of times"),
 	OPT_SET_UINT(0, "per-socket", &stat_config.aggr_mode,
 		     "aggregate counts per processor socket", AGGR_SOCKET),
 	OPT_SET_UINT(0, "per-core", &stat_config.aggr_mode,
@@ -2870,6 +2879,15 @@ int cmd_stat(int argc, const char **argv)
 				   "The overhead percentage could be high in some cases. "
 				   "Please proceed with caution.\n");
 	}
+	if (stat_config.times && interval)
+		interval_count = true;
+	else if (stat_config.times && !interval) {
+		pr_err("interval-count option should be used together with "
+				"interval-print.\n");
+		parse_options_usage(stat_usage, stat_options, "interval-count", 0);
+		parse_options_usage(stat_usage, stat_options, "I", 1);
+		goto out;
+	}
 
 	if (perf_evlist__alloc_stats(evsel_list, interval))
 		goto out;
diff --git a/tools/perf/util/stat.h b/tools/perf/util/stat.h
index dbc6f7134f61..540fbb350e53 100644
--- a/tools/perf/util/stat.h
+++ b/tools/perf/util/stat.h
@@ -90,6 +90,7 @@ struct perf_stat_config {
 	bool		scale;
 	FILE		*output;
 	unsigned int	interval;
+	int		times;
 	struct runtime_stat *stats;
 	int		stats_num;
 };
-- 
2.14.1

[PATCH v4 2/2] perf stat: Add support to print counts after a period of time

[ufo19890607]

From: yuzhoujian <yuzhoujian@didichuxing.com>

Introduce a new option to print counts after N milliseconds
and update perf-stat documentation accordingly.

Show below is the output of the new option for perf stat.

        $ perf stat --time 2000 -e cycles -a
        Performance counter stats for 'system wide':

                157,260,423      cycles

                2.003060766 seconds time elapsed

We can print the count deltas after N milliseconds with this new
introduced option. This option is not supported with "-I" option.
In addition, according to Kangliang's patch(19afd10410957), the
monitoring overhead for system-wide core event could be very high
if the interval-print parameter was below 100ms, and the limitation
value is 10ms. So the same warning will be displayed when the time
is set between 10ms to 100ms, and the minimal time is limited to
10ms. Users can make a decision according to their spcific cases.

Changes since v3:
- none.

Changes since v2:
- modify the time check in __run_perf_stat func to keep some consistency
  with the workload case.
- add the warning when the time is set between 10ms to 100ms.
- add the pr_err when the time is set below 10ms.

Changes since v1:
- none.

Signed-off-by: yuzhoujian <yuzhoujian@didichuxing.com>
---
 tools/perf/Documentation/perf-stat.txt |  5 +++++
 tools/perf/builtin-stat.c              | 33 +++++++++++++++++++++++++++++++--
 tools/perf/util/stat.h                 |  1 +
 3 files changed, 37 insertions(+), 2 deletions(-)

diff --git a/tools/perf/Documentation/perf-stat.txt b/tools/perf/Documentation/perf-stat.txt
index 47a21645f60c..c822f374c99a 100644
--- a/tools/perf/Documentation/perf-stat.txt
+++ b/tools/perf/Documentation/perf-stat.txt
@@ -151,6 +151,11 @@ Print count deltas for fixed number of times.
 This option should be used together with "-I" option.
 	example: 'perf stat -I 1000 --interval-count 2 -e cycles -a'
 
+--time msecs::
+Print count deltas after N milliseconds (minimum: 10 ms).
+This option is not supported with "-I" option.
+	example: 'perf stat --time 2000 -e cycles -a'
+
 --metric-only::
 Only print computed metrics. Print them in a single line.
 Don't show any raw values. Not supported with --per-thread.
diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c
index 7d1d7613bf56..582db3897374 100644
--- a/tools/perf/builtin-stat.c
+++ b/tools/perf/builtin-stat.c
@@ -573,6 +573,7 @@ static int __run_perf_stat(int argc, const char **argv)
 {
 	int interval = stat_config.interval;
 	int times = stat_config.times;
+	int time = stat_config.time;
 	char msg[BUFSIZ];
 	unsigned long long t0, t1;
 	struct perf_evsel *counter;
@@ -586,6 +587,9 @@ static int __run_perf_stat(int argc, const char **argv)
 	if (interval) {
 		ts.tv_sec  = interval / USEC_PER_MSEC;
 		ts.tv_nsec = (interval % USEC_PER_MSEC) * NSEC_PER_MSEC;
+	} else if (time) {
+		ts.tv_sec  = time / USEC_PER_MSEC;
+		ts.tv_nsec = (time % USEC_PER_MSEC) * NSEC_PER_MSEC;
 	} else {
 		ts.tv_sec  = 1;
 		ts.tv_nsec = 0;
@@ -698,9 +702,11 @@ static int __run_perf_stat(int argc, const char **argv)
 		perf_evlist__start_workload(evsel_list);
 		enable_counters();
 
-		if (interval) {
+		if (interval || time) {
 			while (!waitpid(child_pid, &status, WNOHANG)) {
 				nanosleep(&ts, NULL);
+				if (time)
+					break;
 				process_interval();
 				if (interval_count && !(--times))
 					break;
@@ -720,6 +726,8 @@ static int __run_perf_stat(int argc, const char **argv)
 		enable_counters();
 		while (!done) {
 			nanosleep(&ts, NULL);
+			if (time)
+				break;
 			if (interval) {
 				process_interval();
 				if (interval_count && !(--times))
@@ -1900,6 +1908,8 @@ static const struct option stat_options[] = {
 		    "print counts at regular interval in ms (>= 10)"),
 	OPT_INTEGER(0, "interval-count", &stat_config.times,
 		    "print counts for fixed number of times"),
+	OPT_UINTEGER(0, "time", &stat_config.time,
+		    "print counts after a period of time in ms (>= 10)"),
 	OPT_SET_UINT(0, "per-socket", &stat_config.aggr_mode,
 		     "aggregate counts per processor socket", AGGR_SOCKET),
 	OPT_SET_UINT(0, "per-core", &stat_config.aggr_mode,
@@ -2697,7 +2707,7 @@ int cmd_stat(int argc, const char **argv)
 	int status = -EINVAL, run_idx;
 	const char *mode;
 	FILE *output = stderr;
-	unsigned int interval;
+	unsigned int interval, time;
 	const char * const stat_subcommands[] = { "record", "report" };
 
 	setlocale(LC_ALL, "");
@@ -2728,6 +2738,7 @@ int cmd_stat(int argc, const char **argv)
 		return __cmd_report(argc, argv);
 
 	interval = stat_config.interval;
+	time = stat_config.time;
 
 	/*
 	 * For record command the -o is already taken care of.
@@ -2879,6 +2890,7 @@ int cmd_stat(int argc, const char **argv)
 				   "The overhead percentage could be high in some cases. "
 				   "Please proceed with caution.\n");
 	}
+
 	if (stat_config.times && interval)
 		interval_count = true;
 	else if (stat_config.times && !interval) {
@@ -2889,6 +2901,23 @@ int cmd_stat(int argc, const char **argv)
 		goto out;
 	}
 
+	if (time && time < 100) {
+		if (time < 10) {
+			pr_err("time must be >= 10ms.\n");
+			parse_options_usage(stat_usage, stat_options, "time", 0);
+			goto out;
+		} else
+			pr_warning("time < 100ms. "
+				   "The overhead percentage could be high in some cases. "
+				   "Please proceed with caution.\n");
+	}
+	if (time && interval) {
+		pr_err("time option is not supported with interval-print.\n");
+		parse_options_usage(stat_usage, stat_options, "time", 0);
+		parse_options_usage(stat_usage, stat_options, "I", 1);
+		goto out;
+	}
+
 	if (perf_evlist__alloc_stats(evsel_list, interval))
 		goto out;
 
diff --git a/tools/perf/util/stat.h b/tools/perf/util/stat.h
index 540fbb350e53..fc1ab635f7e0 100644
--- a/tools/perf/util/stat.h
+++ b/tools/perf/util/stat.h
@@ -90,6 +90,7 @@ struct perf_stat_config {
 	bool		scale;
 	FILE		*output;
 	unsigned int	interval;
+	unsigned int	time;
 	int		times;
 	struct runtime_stat *stats;
 	int		stats_num;
-- 
2.14.1

Re: [PATCH v4 0/2] perf stat: Add interval-count and time support

[Jiri Olsa]

On Mon, Jan 29, 2018 at 10:25:21AM +0100, ufo19890607 wrote:
> From: yuzhoujian <yuzhoujian@didichuxing.com>
> 
> Introduce two new options for perf stat and update perf-stat documentation
> accordingly.
> 
> The interval-count option can be used to print counts for fixed number of
> times, and it should be used specifically with "-I" option.
> 
> Show below is the output of the interval-count option for perf stat.
> 
>         $ perf stat -I 1000 --interval-count 2 -e cycles -a
>         #           time             counts unit events
>              1.002827089         93,884,870      cycles
>              2.004231506         56,573,446      cycles
>     
> The time option can be used to print counts after a period of time, and it
> should not be used with "-I" option.
> 
> Show below is the output of the time option for perf stat.
> 
>         $ perf stat --time 2000 -e cycles -a
>         Performance counter stats for 'system wide':
> 
>                 157,260,423      cycles
> 
>                 2.003060766 seconds time elapsed
> 
> yuzhoujian (2):
>   perf stat: Add support to print counts for fixed times
>   perf stat: Add support to print counts after a period of time
> 
> Changes since v3: 
> - merge interval_count check and times check to one line.
> - fix the wrong indent in stat.h
> - use stat_config.times instead of 'times' in cmd_stat function.

Acked-by: Jiri Olsa <jolsa@kernel.org>

thanks,
jirka

> 
> Changes since v2: 
> - modify the time check in __run_perf_stat func to keep some consistency
>   with the workload case.
> - add the warning when the time is set between 10ms to 100ms.
> - add the pr_err when the time is set below 10ms.
> 
> Changes since v1:
> - change the name of the new option "times-print" to "interval-count".
> - keep the interval-count option interval specifically.
> 
> 
>  tools/perf/Documentation/perf-stat.txt | 10 +++++++
>  tools/perf/builtin-stat.c              | 53 ++++++++++++++++++++++++++++++++--
>  tools/perf/util/stat.h                 |  2 ++
>  3 files changed, 62 insertions(+), 3 deletions(-)
> 
> -- 
> 2.14.1
> 

Re: [PATCH v4 0/2] perf stat: Add interval-count and time support

[Jiri Olsa]

On Thu, Feb 15, 2018 at 05:33:07PM +0800, 禹舟键 wrote:
> PING
> 
> 2018-01-29 18:25 GMT+08:00 Jiri Olsa <jolsa@redhat.com>:

Arnaldo, could you pelase check on this?

thanks,
jirka

> 
> > On Mon, Jan 29, 2018 at 10:25:21AM +0100, ufo19890607 wrote:
> > > From: yuzhoujian <yuzhoujian@didichuxing.com>
> > >
> > > Introduce two new options for perf stat and update perf-stat
> > documentation
> > > accordingly.
> > >
> > > The interval-count option can be used to print counts for fixed number of
> > > times, and it should be used specifically with "-I" option.
> > >
> > > Show below is the output of the interval-count option for perf stat.
> > >
> > >         $ perf stat -I 1000 --interval-count 2 -e cycles -a
> > >         #           time             counts unit events
> > >              1.002827089         93,884,870      cycles
> > >              2.004231506         56,573,446      cycles
> > >
> > > The time option can be used to print counts after a period of time, and
> > it
> > > should not be used with "-I" option.
> > >
> > > Show below is the output of the time option for perf stat.
> > >
> > >         $ perf stat --time 2000 -e cycles -a
> > >         Performance counter stats for 'system wide':
> > >
> > >                 157,260,423      cycles
> > >
> > >                 2.003060766 seconds time elapsed
> > >
> > > yuzhoujian (2):
> > >   perf stat: Add support to print counts for fixed times
> > >   perf stat: Add support to print counts after a period of time
> > >
> > > Changes since v3:
> > > - merge interval_count check and times check to one line.
> > > - fix the wrong indent in stat.h
> > > - use stat_config.times instead of 'times' in cmd_stat function.
> >
> > Acked-by: Jiri Olsa <jolsa@kernel.org>
> >
> > thanks,
> > jirka
> >
> > >
> > > Changes since v2:
> > > - modify the time check in __run_perf_stat func to keep some consistency
> > >   with the workload case.
> > > - add the warning when the time is set between 10ms to 100ms.
> > > - add the pr_err when the time is set below 10ms.
> > >
> > > Changes since v1:
> > > - change the name of the new option "times-print" to "interval-count".
> > > - keep the interval-count option interval specifically.
> > >
> > >
> > >  tools/perf/Documentation/perf-stat.txt | 10 +++++++
> > >  tools/perf/builtin-stat.c              | 53
> > ++++++++++++++++++++++++++++++++--
> > >  tools/perf/util/stat.h                 |  2 ++
> > >  3 files changed, 62 insertions(+), 3 deletions(-)
> > >
> > > --
> > > 2.14.1
> > >
> >

Re: [PATCH v4 2/2] perf stat: Add support to print counts after a period of time

[Arnaldo Carvalho de Melo]

Em Mon, Jan 29, 2018 at 10:25:23AM +0100, ufo19890607 escreveu:
> From: yuzhoujian <yuzhoujian@didichuxing.com>
> 
> Introduce a new option to print counts after N milliseconds

This doesn't just print counts after N ms, it _stops_ the workload after
that time _and_ prints the counts, right?

Can you please send a followup patch fixing the description and
documentation?

I've applied this already, but clarifying what this option does is in
demand.

- Arnaldo

> and update perf-stat documentation accordingly.
> 
> Show below is the output of the new option for perf stat.
> 
>         $ perf stat --time 2000 -e cycles -a
>         Performance counter stats for 'system wide':
> 
>                 157,260,423      cycles
> 
>                 2.003060766 seconds time elapsed
> 
> We can print the count deltas after N milliseconds with this new
> introduced option. This option is not supported with "-I" option.
> In addition, according to Kangliang's patch(19afd10410957), the
> monitoring overhead for system-wide core event could be very high
> if the interval-print parameter was below 100ms, and the limitation
> value is 10ms. So the same warning will be displayed when the time
> is set between 10ms to 100ms, and the minimal time is limited to
> 10ms. Users can make a decision according to their spcific cases.
> 
> Changes since v3:
> - none.
> 
> Changes since v2:
> - modify the time check in __run_perf_stat func to keep some consistency
>   with the workload case.
> - add the warning when the time is set between 10ms to 100ms.
> - add the pr_err when the time is set below 10ms.
> 
> Changes since v1:
> - none.
> 
> Signed-off-by: yuzhoujian <yuzhoujian@didichuxing.com>
> ---
>  tools/perf/Documentation/perf-stat.txt |  5 +++++
>  tools/perf/builtin-stat.c              | 33 +++++++++++++++++++++++++++++++--
>  tools/perf/util/stat.h                 |  1 +
>  3 files changed, 37 insertions(+), 2 deletions(-)
> 
> diff --git a/tools/perf/Documentation/perf-stat.txt b/tools/perf/Documentation/perf-stat.txt
> index 47a21645f60c..c822f374c99a 100644
> --- a/tools/perf/Documentation/perf-stat.txt
> +++ b/tools/perf/Documentation/perf-stat.txt
> @@ -151,6 +151,11 @@ Print count deltas for fixed number of times.
>  This option should be used together with "-I" option.
>  	example: 'perf stat -I 1000 --interval-count 2 -e cycles -a'
>  
> +--time msecs::
> +Print count deltas after N milliseconds (minimum: 10 ms).
> +This option is not supported with "-I" option.
> +	example: 'perf stat --time 2000 -e cycles -a'
> +
>  --metric-only::
>  Only print computed metrics. Print them in a single line.
>  Don't show any raw values. Not supported with --per-thread.
> diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c
> index 7d1d7613bf56..582db3897374 100644
> --- a/tools/perf/builtin-stat.c
> +++ b/tools/perf/builtin-stat.c
> @@ -573,6 +573,7 @@ static int __run_perf_stat(int argc, const char **argv)
>  {
>  	int interval = stat_config.interval;
>  	int times = stat_config.times;
> +	int time = stat_config.time;
>  	char msg[BUFSIZ];
>  	unsigned long long t0, t1;
>  	struct perf_evsel *counter;
> @@ -586,6 +587,9 @@ static int __run_perf_stat(int argc, const char **argv)
>  	if (interval) {
>  		ts.tv_sec  = interval / USEC_PER_MSEC;
>  		ts.tv_nsec = (interval % USEC_PER_MSEC) * NSEC_PER_MSEC;
> +	} else if (time) {
> +		ts.tv_sec  = time / USEC_PER_MSEC;
> +		ts.tv_nsec = (time % USEC_PER_MSEC) * NSEC_PER_MSEC;
>  	} else {
>  		ts.tv_sec  = 1;
>  		ts.tv_nsec = 0;
> @@ -698,9 +702,11 @@ static int __run_perf_stat(int argc, const char **argv)
>  		perf_evlist__start_workload(evsel_list);
>  		enable_counters();
>  
> -		if (interval) {
> +		if (interval || time) {
>  			while (!waitpid(child_pid, &status, WNOHANG)) {
>  				nanosleep(&ts, NULL);
> +				if (time)
> +					break;
>  				process_interval();
>  				if (interval_count && !(--times))
>  					break;
> @@ -720,6 +726,8 @@ static int __run_perf_stat(int argc, const char **argv)
>  		enable_counters();
>  		while (!done) {
>  			nanosleep(&ts, NULL);
> +			if (time)
> +				break;
>  			if (interval) {
>  				process_interval();
>  				if (interval_count && !(--times))
> @@ -1900,6 +1908,8 @@ static const struct option stat_options[] = {
>  		    "print counts at regular interval in ms (>= 10)"),
>  	OPT_INTEGER(0, "interval-count", &stat_config.times,
>  		    "print counts for fixed number of times"),
> +	OPT_UINTEGER(0, "time", &stat_config.time,
> +		    "print counts after a period of time in ms (>= 10)"),
>  	OPT_SET_UINT(0, "per-socket", &stat_config.aggr_mode,
>  		     "aggregate counts per processor socket", AGGR_SOCKET),
>  	OPT_SET_UINT(0, "per-core", &stat_config.aggr_mode,
> @@ -2697,7 +2707,7 @@ int cmd_stat(int argc, const char **argv)
>  	int status = -EINVAL, run_idx;
>  	const char *mode;
>  	FILE *output = stderr;
> -	unsigned int interval;
> +	unsigned int interval, time;
>  	const char * const stat_subcommands[] = { "record", "report" };
>  
>  	setlocale(LC_ALL, "");
> @@ -2728,6 +2738,7 @@ int cmd_stat(int argc, const char **argv)
>  		return __cmd_report(argc, argv);
>  
>  	interval = stat_config.interval;
> +	time = stat_config.time;
>  
>  	/*
>  	 * For record command the -o is already taken care of.
> @@ -2879,6 +2890,7 @@ int cmd_stat(int argc, const char **argv)
>  				   "The overhead percentage could be high in some cases. "
>  				   "Please proceed with caution.\n");
>  	}
> +
>  	if (stat_config.times && interval)
>  		interval_count = true;
>  	else if (stat_config.times && !interval) {
> @@ -2889,6 +2901,23 @@ int cmd_stat(int argc, const char **argv)
>  		goto out;
>  	}
>  
> +	if (time && time < 100) {
> +		if (time < 10) {
> +			pr_err("time must be >= 10ms.\n");
> +			parse_options_usage(stat_usage, stat_options, "time", 0);
> +			goto out;
> +		} else
> +			pr_warning("time < 100ms. "
> +				   "The overhead percentage could be high in some cases. "
> +				   "Please proceed with caution.\n");
> +	}
> +	if (time && interval) {
> +		pr_err("time option is not supported with interval-print.\n");
> +		parse_options_usage(stat_usage, stat_options, "time", 0);
> +		parse_options_usage(stat_usage, stat_options, "I", 1);
> +		goto out;
> +	}
> +
>  	if (perf_evlist__alloc_stats(evsel_list, interval))
>  		goto out;
>  
> diff --git a/tools/perf/util/stat.h b/tools/perf/util/stat.h
> index 540fbb350e53..fc1ab635f7e0 100644
> --- a/tools/perf/util/stat.h
> +++ b/tools/perf/util/stat.h
> @@ -90,6 +90,7 @@ struct perf_stat_config {
>  	bool		scale;
>  	FILE		*output;
>  	unsigned int	interval;
> +	unsigned int	time;
>  	int		times;
>  	struct runtime_stat *stats;
>  	int		stats_num;
> -- 
> 2.14.1

Re: [PATCH v4 2/2] perf stat: Add support to print counts after a period of time

[Arnaldo Carvalho de Melo]

Em Thu, Feb 15, 2018 at 10:39:21AM -0300, Arnaldo Carvalho de Melo escreveu:
> Em Mon, Jan 29, 2018 at 10:25:23AM +0100, ufo19890607 escreveu:
> > From: yuzhoujian <yuzhoujian@didichuxing.com>
> > 
> > Introduce a new option to print counts after N milliseconds
> 
> This doesn't just print counts after N ms, it _stops_ the workload after
> that time _and_ prints the counts, right?
> 
> Can you please send a followup patch fixing the description and
> documentation?
> 
> I've applied this already, but clarifying what this option does is in
> demand.

So, this fails to build on centos:5 and centos:6 and similar systems
where the use of 'time' for a variable name breaks the build:

  CC       /tmp/build/perf/builtin-stat.o
cc1: warnings being treated as errors
builtin-stat.c: In function '__run_perf_stat':
builtin-stat.c:576: warning: declaration of 'time' shadows a global declaration
/usr/include/time.h:187: warning: shadowed declaration is here
builtin-stat.c: In function 'cmd_stat':
builtin-stat.c:2710: warning: declaration of 'time' shadows a global declaration
/usr/include/time.h:187: warning: shadowed declaration is here
mv: cannot stat `/tmp/build/perf/.builtin-stat.o.tmp': No such file or directory
make[3]: *** [/tmp/build/perf/builtin-stat.o] Error 1
make[3]: *** Waiting for unfinished jobs....
  MKDIR    /tmp/build/perf/util/
  CC       /tmp/build/perf/util/evsel_fprintf.o
make[2]: *** [/tmp/build/perf/perf-in.o] Error 2
make[2]: *** Waiting for unfinished jobs....

So since I had to rename that I replaced it with 'timeout', so that it
gets clear that when you specify this option you are setting a timeout
period that if hit will stop the 'perf stat' session and then print the
specified counters.

Please let us know if you have any strong objection to this,

- Arnaldo

[tip:perf/core] perf stat: Add support to print counts for fixed times

[tip-bot for yuzhoujian]

Commit-ID:  db06a269ecbb1d71d534fc5713624eeeee0b8f92
Gitweb:     https://git.kernel.org/tip/db06a269ecbb1d71d534fc5713624eeeee0b8f92
Author:     yuzhoujian <yuzhoujian@didichuxing.com>
AuthorDate: Mon, 29 Jan 2018 10:25:22 +0100
Committer:  Arnaldo Carvalho de Melo <acme@redhat.com>
CommitDate: Fri, 16 Feb 2018 10:09:24 -0300

perf stat: Add support to print counts for fixed times

Introduce a new option to print counts for fixed number of times and
update 'perf stat' documentation accordingly.

Show below is the output of the new option for perf stat.

  $ perf stat -I 1000 --interval-count 2 -e cycles -a
  #           time             counts unit events
           1.002827089         93,884,870      cycles
           2.004231506         56,573,446      cycles

We can just print the counts for several times with this newly
introduced option. The usage of it is a little like 'vmstat', and it
should be used together with "-I" option.

  $ vmstat -n 1 2
  procs ---------memory-------------- --swap- ----io-- -system-- ------cpu---
   r  b swpd   free   buff   cache    si   so  bi   bo  in   cs us sy id wa st
   0  0    0 78270544 547484 51732076  0   0   0   20    1    1  1  0 99  0 0
   0  0    0 78270512 547484 51732080  0   0   0   16  477 1555  0  0 100 0 0

Changes since v3:
- merge interval_count check and times check to one line.
- fix the wrong indent in stat.h
- use stat_config.times instead of 'times' in cmd_stat function.

Changes since v2:
- none.

Changes since v1:
- change the name of the new option "times-print" to "interval-count".
- keep the new option interval specifically.

Signed-off-by: yuzhoujian <yuzhoujian@didichuxing.com>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Milian Wolff <milian.wolff@kdab.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Wang Nan <wangnan0@huawei.com>
Link: http://lkml.kernel.org/r/1517217923-8302-2-git-send-email-ufo19890607@gmail.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
 tools/perf/Documentation/perf-stat.txt |  5 +++++
 tools/perf/builtin-stat.c              | 20 +++++++++++++++++++-
 tools/perf/util/stat.h                 |  1 +
 3 files changed, 25 insertions(+), 1 deletion(-)

diff --git a/tools/perf/Documentation/perf-stat.txt b/tools/perf/Documentation/perf-stat.txt
index 823fce7..47a2164 100644
--- a/tools/perf/Documentation/perf-stat.txt
+++ b/tools/perf/Documentation/perf-stat.txt
@@ -146,6 +146,11 @@ Print count deltas every N milliseconds (minimum: 10ms)
 The overhead percentage could be high in some cases, for instance with small, sub 100ms intervals.  Use with caution.
 	example: 'perf stat -I 1000 -e cycles -a sleep 5'
 
+--interval-count times::
+Print count deltas for fixed number of times.
+This option should be used together with "-I" option.
+	example: 'perf stat -I 1000 --interval-count 2 -e cycles -a'
+
 --metric-only::
 Only print computed metrics. Print them in a single line.
 Don't show any raw values. Not supported with --per-thread.
diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c
index 98bf9d3..7d1d761 100644
--- a/tools/perf/builtin-stat.c
+++ b/tools/perf/builtin-stat.c
@@ -168,6 +168,7 @@ static struct timespec		ref_time;
 static struct cpu_map		*aggr_map;
 static aggr_get_id_t		aggr_get_id;
 static bool			append_file;
+static bool			interval_count;
 static const char		*output_name;
 static int			output_fd;
 static int			print_free_counters_hint;
@@ -571,6 +572,7 @@ static struct perf_evsel *perf_evsel__reset_weak_group(struct perf_evsel *evsel)
 static int __run_perf_stat(int argc, const char **argv)
 {
 	int interval = stat_config.interval;
+	int times = stat_config.times;
 	char msg[BUFSIZ];
 	unsigned long long t0, t1;
 	struct perf_evsel *counter;
@@ -700,6 +702,8 @@ try_again:
 			while (!waitpid(child_pid, &status, WNOHANG)) {
 				nanosleep(&ts, NULL);
 				process_interval();
+				if (interval_count && !(--times))
+					break;
 			}
 		}
 		waitpid(child_pid, &status, 0);
@@ -716,8 +720,11 @@ try_again:
 		enable_counters();
 		while (!done) {
 			nanosleep(&ts, NULL);
-			if (interval)
+			if (interval) {
 				process_interval();
+				if (interval_count && !(--times))
+					break;
+			}
 		}
 	}
 
@@ -1891,6 +1898,8 @@ static const struct option stat_options[] = {
 			"command to run after to the measured command"),
 	OPT_UINTEGER('I', "interval-print", &stat_config.interval,
 		    "print counts at regular interval in ms (>= 10)"),
+	OPT_INTEGER(0, "interval-count", &stat_config.times,
+		    "print counts for fixed number of times"),
 	OPT_SET_UINT(0, "per-socket", &stat_config.aggr_mode,
 		     "aggregate counts per processor socket", AGGR_SOCKET),
 	OPT_SET_UINT(0, "per-core", &stat_config.aggr_mode,
@@ -2870,6 +2879,15 @@ int cmd_stat(int argc, const char **argv)
 				   "The overhead percentage could be high in some cases. "
 				   "Please proceed with caution.\n");
 	}
+	if (stat_config.times && interval)
+		interval_count = true;
+	else if (stat_config.times && !interval) {
+		pr_err("interval-count option should be used together with "
+				"interval-print.\n");
+		parse_options_usage(stat_usage, stat_options, "interval-count", 0);
+		parse_options_usage(stat_usage, stat_options, "I", 1);
+		goto out;
+	}
 
 	if (perf_evlist__alloc_stats(evsel_list, interval))
 		goto out;
diff --git a/tools/perf/util/stat.h b/tools/perf/util/stat.h
index dbc6f71..540fbb3 100644
--- a/tools/perf/util/stat.h
+++ b/tools/perf/util/stat.h
@@ -90,6 +90,7 @@ struct perf_stat_config {
 	bool		scale;
 	FILE		*output;
 	unsigned int	interval;
+	int		times;
 	struct runtime_stat *stats;
 	int		stats_num;
 };

[tip:perf/core] perf stat: Add support to print counts after a period of time

[tip-bot for yuzhoujian]

Commit-ID:  f1f8ad52f8bf1239282737a2a5c3bd450300cc78
Gitweb:     https://git.kernel.org/tip/f1f8ad52f8bf1239282737a2a5c3bd450300cc78
Author:     yuzhoujian <yuzhoujian@didichuxing.com>
AuthorDate: Mon, 29 Jan 2018 10:25:23 +0100
Committer:  Arnaldo Carvalho de Melo <acme@redhat.com>
CommitDate: Fri, 16 Feb 2018 10:18:06 -0300

perf stat: Add support to print counts after a period of time

Introduce a new option to print counts after N milliseconds and update
'perf stat' documentation accordingly.

Show below is the output of the new option for perf stat.

  $ perf stat --time 2000 -e cycles -a
  Performance counter stats for 'system wide':

        157,260,423      cycles

        2.003060766 seconds time elapsed

We can print the count deltas after N milliseconds with this new
introduced option. This option is not supported with "-I" option.

In addition, according to Kangliang's patch(19afd10410957), the
monitoring overhead for system-wide core event could be very high if the
interval-print parameter was below 100ms, and the limitation value is
10ms.

So the same warning will be displayed when the time is set between 10ms
to 100ms, and the minimal time is limited to 10ms. Users can make a
decision according to their spcific cases.

Committer notes:

This actually stops the workload after the specified time, then prints
the counts.

So I renamed the option to --timeout and updated the documentation to
state that it will not just print the counts after the specified time,
but will really stop the 'perf stat' session and print the counts.

The rename from 'time' to 'timeout' also fixes the build in systems
where 'time' is used by glibc and can't be used as a name of a variable,
such as centos:5 and centos:6.

Changes since v3:
- none.

Changes since v2:
- modify the time check in __run_perf_stat func to keep some consistency
  with the workload case.
- add the warning when the time is set between 10ms to 100ms.
- add the pr_err when the time is set below 10ms.

Changes since v1:
- none.

Signed-off-by: yuzhoujian <yuzhoujian@didichuxing.com>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Milian Wolff <milian.wolff@kdab.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Wang Nan <wangnan0@huawei.com>
Link: http://lkml.kernel.org/r/1517217923-8302-3-git-send-email-ufo19890607@gmail.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
 tools/perf/Documentation/perf-stat.txt |  5 +++++
 tools/perf/builtin-stat.c              | 33 +++++++++++++++++++++++++++++++--
 tools/perf/util/stat.h                 |  1 +
 3 files changed, 37 insertions(+), 2 deletions(-)

diff --git a/tools/perf/Documentation/perf-stat.txt b/tools/perf/Documentation/perf-stat.txt
index 47a2164..2bbe79a 100644
--- a/tools/perf/Documentation/perf-stat.txt
+++ b/tools/perf/Documentation/perf-stat.txt
@@ -151,6 +151,11 @@ Print count deltas for fixed number of times.
 This option should be used together with "-I" option.
 	example: 'perf stat -I 1000 --interval-count 2 -e cycles -a'
 
+--timeout msecs::
+Stop the 'perf stat' session and print count deltas after N milliseconds (minimum: 10 ms).
+This option is not supported with the "-I" option.
+	example: 'perf stat --time 2000 -e cycles -a'
+
 --metric-only::
 Only print computed metrics. Print them in a single line.
 Don't show any raw values. Not supported with --per-thread.
diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c
index 7d1d761..2d49ecc 100644
--- a/tools/perf/builtin-stat.c
+++ b/tools/perf/builtin-stat.c
@@ -573,6 +573,7 @@ static int __run_perf_stat(int argc, const char **argv)
 {
 	int interval = stat_config.interval;
 	int times = stat_config.times;
+	int timeout = stat_config.timeout;
 	char msg[BUFSIZ];
 	unsigned long long t0, t1;
 	struct perf_evsel *counter;
@@ -586,6 +587,9 @@ static int __run_perf_stat(int argc, const char **argv)
 	if (interval) {
 		ts.tv_sec  = interval / USEC_PER_MSEC;
 		ts.tv_nsec = (interval % USEC_PER_MSEC) * NSEC_PER_MSEC;
+	} else if (timeout) {
+		ts.tv_sec  = timeout / USEC_PER_MSEC;
+		ts.tv_nsec = (timeout % USEC_PER_MSEC) * NSEC_PER_MSEC;
 	} else {
 		ts.tv_sec  = 1;
 		ts.tv_nsec = 0;
@@ -698,9 +702,11 @@ try_again:
 		perf_evlist__start_workload(evsel_list);
 		enable_counters();
 
-		if (interval) {
+		if (interval || timeout) {
 			while (!waitpid(child_pid, &status, WNOHANG)) {
 				nanosleep(&ts, NULL);
+				if (timeout)
+					break;
 				process_interval();
 				if (interval_count && !(--times))
 					break;
@@ -720,6 +726,8 @@ try_again:
 		enable_counters();
 		while (!done) {
 			nanosleep(&ts, NULL);
+			if (timeout)
+				break;
 			if (interval) {
 				process_interval();
 				if (interval_count && !(--times))
@@ -1900,6 +1908,8 @@ static const struct option stat_options[] = {
 		    "print counts at regular interval in ms (>= 10)"),
 	OPT_INTEGER(0, "interval-count", &stat_config.times,
 		    "print counts for fixed number of times"),
+	OPT_UINTEGER(0, "timeout", &stat_config.timeout,
+		    "stop workload and print counts after a timeout period in ms (>= 10ms)"),
 	OPT_SET_UINT(0, "per-socket", &stat_config.aggr_mode,
 		     "aggregate counts per processor socket", AGGR_SOCKET),
 	OPT_SET_UINT(0, "per-core", &stat_config.aggr_mode,
@@ -2697,7 +2707,7 @@ int cmd_stat(int argc, const char **argv)
 	int status = -EINVAL, run_idx;
 	const char *mode;
 	FILE *output = stderr;
-	unsigned int interval;
+	unsigned int interval, timeout;
 	const char * const stat_subcommands[] = { "record", "report" };
 
 	setlocale(LC_ALL, "");
@@ -2728,6 +2738,7 @@ int cmd_stat(int argc, const char **argv)
 		return __cmd_report(argc, argv);
 
 	interval = stat_config.interval;
+	timeout = stat_config.timeout;
 
 	/*
 	 * For record command the -o is already taken care of.
@@ -2879,6 +2890,7 @@ int cmd_stat(int argc, const char **argv)
 				   "The overhead percentage could be high in some cases. "
 				   "Please proceed with caution.\n");
 	}
+
 	if (stat_config.times && interval)
 		interval_count = true;
 	else if (stat_config.times && !interval) {
@@ -2889,6 +2901,23 @@ int cmd_stat(int argc, const char **argv)
 		goto out;
 	}
 
+	if (timeout && timeout < 100) {
+		if (timeout < 10) {
+			pr_err("timeout must be >= 10ms.\n");
+			parse_options_usage(stat_usage, stat_options, "timeout", 0);
+			goto out;
+		} else
+			pr_warning("timeout < 100ms. "
+				   "The overhead percentage could be high in some cases. "
+				   "Please proceed with caution.\n");
+	}
+	if (timeout && interval) {
+		pr_err("timeout option is not supported with interval-print.\n");
+		parse_options_usage(stat_usage, stat_options, "timeout", 0);
+		parse_options_usage(stat_usage, stat_options, "I", 1);
+		goto out;
+	}
+
 	if (perf_evlist__alloc_stats(evsel_list, interval))
 		goto out;
 
diff --git a/tools/perf/util/stat.h b/tools/perf/util/stat.h
index 540fbb3..2f44e38 100644
--- a/tools/perf/util/stat.h
+++ b/tools/perf/util/stat.h
@@ -90,6 +90,7 @@ struct perf_stat_config {
 	bool		scale;
 	FILE		*output;
 	unsigned int	interval;
+	unsigned int	timeout;
 	int		times;
 	struct runtime_stat *stats;
 	int		stats_num;