Re: [PATCH] perf stat: Honour --timeout for forked workloads

From: Jiri Olsa
Date: Wed Apr 15 2020 - 12:25:42 EST


On Wed, Apr 15, 2020 at 12:38:03PM -0300, Arnaldo Carvalho de Melo wrote:
> Hi guys,
>
> Please take a look and give this your acks, Tested-by, etc.
>
> Thanks,
>
> - Arnaldo
>
> ----
>
> When --timeout is used and a workload is specified to be started by
> 'perf stat', i.e.
>
> $ perf stat --timeout 1000 sleep 1h
>
> The --timeout wasn't being honoured, i.e. the workload, 'sleep 1h' in
> the above example, should be terminated after 1000ms, but it wasn't,
> 'perf stat' was waiting for it to finish.
>
> Fix it by sending a SIGTERM when the timeout expires.
>
> Now it works:
>
> # perf stat -e cycles --timeout 1234 sleep 1h
> sleep: Terminated
>
> Performance counter stats for 'sleep 1h':
>
> 1,066,692 cycles
>
> 1.234314838 seconds time elapsed
>
> 0.000750000 seconds user
> 0.000000000 seconds sys
>
> #
>
> Reported-by: Konstantin Kharlamov <hi-angel@xxxxxxxxx>
> Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=207243
> Cc: Adrian Hunter <adrian.hunter@xxxxxxxxx>
> Cc: Jiri Olsa <jolsa@xxxxxxxxxx>
> Cc: Namhyung Kim <namhyung@xxxxxxxxxx>
> Signed-off-by: Arnaldo Carvalho de Melo <acme@xxxxxxxxxx>

Acked/Tested-by: Jiri Olsa <jolsa@xxxxxxxxxx>

cc-ing yuzhoujian, author of the --timeout

jirka

> ---
> tools/perf/builtin-stat.c | 5 ++++-
> 1 file changed, 4 insertions(+), 1 deletion(-)
>
> diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c
> index ec053dc1e35c..9207b6c45475 100644
> --- a/tools/perf/builtin-stat.c
> +++ b/tools/perf/builtin-stat.c
> @@ -686,8 +686,11 @@ static int __run_perf_stat(int argc, const char **argv, int run_idx)
> break;
> }
> }
> - if (child_pid != -1)
> + if (child_pid != -1) {
> + if (timeout)
> + kill(child_pid, SIGTERM);
> wait4(child_pid, &status, 0, &stat_config.ru_data);
> + }
>
> if (workload_exec_errno) {
> const char *emsg = str_error_r(workload_exec_errno, msg, sizeof(msg));
> --
> 2.25.2
>