Poll for monitored tasks being alive in fork mode
PATCH v2 1/1 Poll for monitored tasks being alive in fork mode
Following test shows the stat keeps running even if no longer task to monitor (mgen exits at ~5s).
perf stat -e cycles -p pgrep mgen -I1000 -- sleep 10
time counts unit events
1.000148916 1,308,365,864 cycles
2.000379171 1,297,269,875 cycles
3.000556719 1,297,187,078 cycles
4.000914241 761,261,827 cycles
5.001306091 <not counted> cycles
6.001676881 <not counted> cycles
7.002046336 <not counted> cycles
8.002405651 <not counted> cycles
9.002766625 <not counted> cycles
10.001395827 <not counted> cyclesWe'd better finish stat immediately if there's no longer task to monitor.
After:
perf stat -e cycles -p pgrep mgen -I1000 -- sleep 10
time counts unit events
1.000180062 1,236,592,661 cycles
2.000421539 1,223,733,572 cycles
3.000609910 1,297,047,663 cycles
4.000807545 1,297,215,816 cycles
5.001001578 1,297,208,032 cycles
6.001390345 582,343,659 cyclessleep: Terminated
Now the stat exits immediately when the monitored tasks ends.
v2:
---
There is a potential race condition window, that is the child_pid set with -1 and kill(-1, SIGTERM) may happen. Jiri provides a simple method which stores the child_pid to a local variable and check the local variable before kill(). At least, we can avoid the kill(-1, SIGTERM) case.
--- tools/perf/builtin-stat.c | 9 +++++++++ 1 file changed, 9 insertions(+)
tools/perf/builtin-stat.c +9 -0
@@ -553,6 +553,15 @@ static int __run_perf_stat(int argc, const char **argv, int run_idx)if (interval || timeout) {while (!waitpid (child_pid, &status,WNOHANG )) { +if (!is_target_alive (&target, + evsel_list->threads)) { +int pid = child_pid; + +if (pid !=-1 ) +kill (pid,SIGTERM ); +break ; + } +nanosleep (&ts,NULL );if (timeout)break ;
Signature · 1 line
2.7.4
Quoted 41 lines · Jin Yao
Following test shows the stat keeps running even if no longer task to monitor (mgen exits at ~5s).
perf stat -e cycles -p
pgrep mgen-I1000 -- sleep 10time counts unit events 1.000148916 1,308,365,864 cycles 2.000379171 1,297,269,875 cycles 3.000556719 1,297,187,078 cycles 4.000914241 761,261,827 cycles 5.001306091 <not counted> cycles 6.001676881 <not counted> cycles 7.002046336 <not counted> cycles 8.002405651 <not counted> cycles 9.002766625 <not counted> cycles 10.001395827 <not counted> cyclesWe'd better finish stat immediately if there's no longer task to monitor.
After:
perf stat -e cycles -p
pgrep mgen-I1000 -- sleep 10time counts unit events 1.000180062 1,236,592,661 cycles 2.000421539 1,223,733,572 cycles 3.000609910 1,297,047,663 cycles 4.000807545 1,297,215,816 cycles 5.001001578 1,297,208,032 cycles 6.001390345 582,343,659 cyclessleep: Terminated
Now the stat exits immediately when the monitored tasks ends.
v2:
---
There is a potential race condition window, that is thechild_pidset with -1 andkill(-1, SIGTERM)may happen. Jiri provides a simple method which stores thechild_pidto a local variable and check the local variable beforekill(). At least, we can avoid thekill(-1, SIGTERM)case.
Quoted 1 line
Signed-off-by Jin Yao <[email protected]>
thanks,
jirka
Quoted 27 lines
--- tools/perf/builtin-stat.c | 9 +++++++++ 1 file changed, 9 insertions(+)tools/perf/builtin-stat.c +9
-0@@ -553,6 +553,15 @@ static int __run_perf_stat(int argc, const char **argv, int run_idx)if (interval || timeout) {while (!waitpid (child_pid, &status,WNOHANG )) { +if (!is_target_alive (&target, + evsel_list->threads)) { +int pid = child_pid; + +if (pid !=-1 ) +kill (pid,SIGTERM ); +break ; + } +nanosleep (&ts,NULL );if (timeout)break ;Signature · 1 line
2.7.4
Hi Arnaldo,
Is this patch acceptable?
Thanks
Jin Yao
Quoted 76 lines · Jiri Olsa
Quoted 41 lines · Jin Yao
Following test shows the stat keeps running even if no longer task to monitor (mgen exits at ~5s).
perf stat -e cycles -p
pgrep mgen-I1000 -- sleep 10time counts unit events 1.000148916 1,308,365,864 cycles 2.000379171 1,297,269,875 cycles 3.000556719 1,297,187,078 cycles 4.000914241 761,261,827 cycles 5.001306091 <not counted> cycles 6.001676881 <not counted> cycles 7.002046336 <not counted> cycles 8.002405651 <not counted> cycles 9.002766625 <not counted> cycles 10.001395827 <not counted> cyclesWe'd better finish stat immediately if there's no longer task to monitor.
After:
perf stat -e cycles -p
pgrep mgen-I1000 -- sleep 10time counts unit events 1.000180062 1,236,592,661 cycles 2.000421539 1,223,733,572 cycles 3.000609910 1,297,047,663 cycles 4.000807545 1,297,215,816 cycles 5.001001578 1,297,208,032 cycles 6.001390345 582,343,659 cyclessleep: Terminated
Now the stat exits immediately when the monitored tasks ends.
v2:
---
There is a potential race condition window, that is thechild_pidset with -1 andkill(-1, SIGTERM)may happen. Jiri provides a simple method which stores thechild_pidto a local variable and check the local variable beforekill(). At least, we can avoid thekill(-1, SIGTERM)case.Quoted 1 line
Signed-off-by Jin Yao <[email protected]>Reviewed-by Jiri Olsa <[email protected]>thanks,
jirkaQuoted 27 lines
--- tools/perf/builtin-stat.c | 9 +++++++++ 1 file changed, 9 insertions(+)tools/perf/builtin-stat.c +9
-0@@ -553,6 +553,15 @@ static int __run_perf_stat(int argc, const char **argv, int run_idx)if (interval || timeout) {while (!waitpid (child_pid, &status,WNOHANG )) { +if (!is_target_alive (&target, + evsel_list->threads)) { +int pid = child_pid; + +if (pid !=-1 ) +kill (pid,SIGTERM ); +break ; + } +nanosleep (&ts,NULL );if (timeout)break ;Signature · 1 line
2.7.4