Patch suggested inside LTP memcg stress test case 'memcg_stress_test.sh'.
Hi,
We have found out a cgroup test case issue with
'ltp-full-20100331/testcases/kernel/controllers/memcg/stress/memcg_stress_test.sh'. The sniff of test case function is as below:-
Note: - Below code sniff is taken from latest LTP August month release.
run_stress()
{
do_mount;
for i in $(seq 0 $(($1-1)))
do
mkdir /dev/memcg/$i 2> /dev/null
./memcg_process_stress $2 $3 &
eval pid$i=$!
eval echo \$pid$i > /dev/memcg/$i/tasks
done
for i in $(seq 0 $(($1-1)))
do
eval /bin/kill -s SIGUSR1 \$pid$i 2> /dev/null
done
sleep $4
for i in $(seq 0 $(($1-1)))
doeval /bin/kill -s SIGINT \$pid$i 2> /dev/null *<----send SIGINT signal for a pid *
eval wait \$pid$i
rmdir /dev/memcg/$i 2> /dev/null
done
cleanup;
}
Basically, the test case is not doing a proper cleanup using SIGINT.It seems that the current test runs for much longer time than expected, thus affecting test performance. During this course of time, other dependent tests were also not proceeding further and system seemed to be in a process hang state. However we have found that SIGKILL does a better cleanup job compared to SIGINT, to overcome this situation.
So, I have created a patch to fix above test case issue. Kindly review this patch and acknowledge. I think this patch can be merged into latest LTP code to avoid this problem.
Please let me know if you need any clarifications.
Regards,
Shubham
memcg_stress_test_latest.sh renamed from memcg_stress_test.sh +2 -2
@@ -1,4 +1,4 @@ -#! /bin/sh +#! /bin/sh ################################################################################ ## ##
@@ -85,7 +85,7 @@ for i in $(seq 0 $(($1-1))) do - eval /bin/kill -s SIGINT \$pid$i 2> /dev/null + eval /bin/kill -s SIGKILL \$pid$i 2> /dev/null eval wait \$pid$i rmdir /dev/memcg/$i 2> /dev/null
------------------------------------------------------------------------------
This SF.net Dev2Dev email is sponsored by:
Show off your parallel programming skills.
Enter the Intel(R) Threading Challenge 2010.
http://p.sf.net/sfu/intel-thread-sfd
_______________________________________________
Ltp-list mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/ltp-list
Thanks. I have merged your patch. However you need to address the following things in future:
1. All patches need to have a DCO signoff like: Signed-off-by:
Shubham <[email protected]>,
2. Create all your patches from LTP root: like:
diff -uprN ltp-*/testcases/*..*/file.orig ltp-*/testcases/*..*/file
Regards--
Subrata
Quoted 56 lines · Shubham
Hi,
We have found out a cgroup test case issue with
'ltp-full-20100331/testcases/kernel/controllers/memcg/stress/memcg_stress_test.sh'. The sniff of test case function is as below:-Note: - Below code sniff is taken from latest LTP August month release.
run_stress(){ do_mount; for i in $(seq 0 $(($1-1))) do mkdir /dev/memcg/$i 2> /dev/null ./memcg_process_stress $2 $3 & eval pid$i=$! eval echo \$pid$i > /dev/memcg/$i/tasks done for i in $(seq 0 $(($1-1))) do eval /bin/kill -s SIGUSR1 \$pid$i 2> /dev/null done sleep $4 for i in $(seq 0 $(($1-1))) do eval /bin/kill -s SIGINT \$pid$i 2> /dev/null<----send SIGINT signal for a pid
eval wait \$pid$irmdir
/dev/memcg/$i2>/dev/null
donecleanup;
}Basically, the test case is not doing a proper cleanup using SIGINT.It seems that the current test runs for much longer time than expected, thus affecting test performance. During this course of time, other dependent tests were also not proceeding further and system seemed to be in a process hang state. However we have found that SIGKILL does a better cleanup job compared to SIGINT, to overcome this situation.
So, I have created a patch to fix above test case issue. Kindly review this patch and acknowledge. I think this patch can be merged into latest LTP code to avoid this problem.
Please let me know if you need any clarifications.
Regards,
Shubham
Quoted 7 lines
------------------------------------------------------------------------------
This SF.net Dev2Dev email is sponsored by:Show off your parallel programming skills.
Enter the Intel(R) Threading Challenge 2010.
http://p.sf.net/sfu/intel-thread-sfd
_______________________________________________ Ltp-list mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/ltp-list
------------------------------------------------------------------------------
Start uncovering the many advantages of virtual appliances and start using them to simplify application deployment and accelerate your shift to cloud computing.
http://p.sf.net/sfu/novell-sfdev2dev
_______________________________________________
Ltp-list mailing list [email protected]
https://lists.sourceforge.net/lists/listinfo/ltp-list
Thanks Subrata,
I will take care of below mentioned points
in future.
Regards
--Shubham
Quoted 8 lines · Subrata Modak
Thanks. I have merged your patch. However you need to address the following things in future:
1. All patches need to have a DCO signoff like: Signed-off-by:
Shubham<[email protected]>,
2. Create all your patches from LTP root: like:
diff -uprN ltp-*/testcases/*..*/file.orig ltp-*/testcases/*..*/file
Quoted 67 lines
Regards--
SubrataQuoted 63 lines · Shubham
Hi,
We have found out a cgroup test case issue with
'ltp-full-20100331/testcases/kernel/controllers/memcg/stress/memcg_stress_test.sh'. The sniff of test case function is as below:-Note: - Below code sniff is taken from latest LTP August month release.
run_stress(){ do_mount; for i in $(seq 0 $(($1-1))) do mkdir /dev/memcg/$i 2> /dev/null ./memcg_process_stress $2 $3& eval pid$i=$! eval echo \$pid$i> /dev/memcg/$i/tasks done for i in $(seq 0 $(($1-1))) do eval /bin/kill -s SIGUSR1 \$pid$i 2> /dev/null done sleep $4 for i in $(seq 0 $(($1-1))) do eval /bin/kill -s SIGINT \$pid$i 2> /dev/null<----send SIGINT signal for a pid
eval wait \$pid$irmdir
/dev/memcg/$i2>/dev/null
donecleanup;
}Basically, the test case is not doing a proper cleanup using SIGINT.It seems that the current test runs for much longer time than expected, thus affecting test performance. During this course of time, other dependent tests were also not proceeding further and system seemed to be in a process hang state. However we have found that SIGKILL does a better cleanup job compared to SIGINT, to overcome this situation.
So, I have created a patch to fix above test case issue. Kindly review this patch and acknowledge. I think this patch can be merged into latest LTP code to avoid this problem.
Please let me know if you need any clarifications.
Regards,
Shubham------------------------------------------------------------------------------
This SF.net Dev2Dev email is sponsored by:Show off your parallel programming skills.
Enter the Intel(R) Threading Challenge 2010.
http://p.sf.net/sfu/intel-thread-sfd
_______________________________________________ Ltp-list mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/ltp-list
------------------------------------------------------------------------------
Start uncovering the many advantages of virtual appliances and start using them to simplify application deployment and accelerate your shift to cloud computing.
http://p.sf.net/sfu/novell-sfdev2dev
_______________________________________________
Ltp-list mailing list [email protected]
https://lists.sourceforge.net/lists/listinfo/ltp-list