[LTP] [PATCH] controllers/cpuacct.sh: Fix timeout for cpuacct.sh

[LTP] [PATCH] controllers/cpuacct.sh: Fix timeout for cpuacct.sh

[Xiao Yang]

With commit fbea02a, cpuacct.sh always gets timeout because wait
command tries to wait all background processes including the timeout
process.  We should just wait the processes created for test.

e.g. running cpuacct_1_1 gets the following timeout:
-----------------------------------------------------------
cpuacct 1 TINFO: timeout per run is 0h 5m 0s
cpuacct 1 TINFO: cpuacct:
cpuacct 1 TINFO: Creating 1 subgroups each with 1 processes
cpuacct 1 TBROK: test killed, timeout!
-----------------------------------------------------------

Fixes: fbea02a("lib/tst_test.sh: setup timeout per test run for the shell tests")
Signed-off-by: Xiao Yang <yangx.jy@cn.fujitsu.com>
---
 testcases/kernel/controllers/cpuacct/cpuacct.sh | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/testcases/kernel/controllers/cpuacct/cpuacct.sh b/testcases/kernel/controllers/cpuacct/cpuacct.sh
index edd4a98..5e90b05 100755
--- a/testcases/kernel/controllers/cpuacct/cpuacct.sh
+++ b/testcases/kernel/controllers/cpuacct/cpuacct.sh
@@ -41,6 +41,7 @@ TST_TESTFUNC=do_test
 TST_POS_ARGS=2
 TST_USAGE=usage
 TST_NEEDS_ROOT=1
+TST_NEEDS_TMPDIR=1
 
 . tst_test.sh
 
@@ -114,10 +115,12 @@ do_test()
 	for i in `seq 1 $max`; do
 		for j in `seq 1 $nbprocess`; do
 			cpuacct_task $testpath/subgroup_$i/tasks &
+			echo $! >> task_pids
 		done
 	done
 
-	wait
+	for pid in $(cat task_pids); do wait $pid; done
+	rm -f task_pids
 
 	acc=0
 	fails=0
-- 
1.8.3.1

[LTP] [PATCH] controllers/cpuacct.sh: Fix timeout for cpuacct.sh

[Xiao Yang]

Hi,

Ping :-)

Best Regards,
Xiao Yang
On 2018/12/06 17:05, Xiao Yang wrote:
> With commit fbea02a, cpuacct.sh always gets timeout because wait
> command tries to wait all background processes including the timeout
> process.  We should just wait the processes created for test.
>
> e.g. running cpuacct_1_1 gets the following timeout:
> -----------------------------------------------------------
> cpuacct 1 TINFO: timeout per run is 0h 5m 0s
> cpuacct 1 TINFO: cpuacct:
> cpuacct 1 TINFO: Creating 1 subgroups each with 1 processes
> cpuacct 1 TBROK: test killed, timeout!
> -----------------------------------------------------------
>
> Fixes: fbea02a("lib/tst_test.sh: setup timeout per test run for the shell tests")
> Signed-off-by: Xiao Yang <yangx.jy@cn.fujitsu.com>
> ---
>  testcases/kernel/controllers/cpuacct/cpuacct.sh | 5 ++++-
>  1 file changed, 4 insertions(+), 1 deletion(-)
>
> diff --git a/testcases/kernel/controllers/cpuacct/cpuacct.sh b/testcases/kernel/controllers/cpuacct/cpuacct.sh
> index edd4a98..5e90b05 100755
> --- a/testcases/kernel/controllers/cpuacct/cpuacct.sh
> +++ b/testcases/kernel/controllers/cpuacct/cpuacct.sh
> @@ -41,6 +41,7 @@ TST_TESTFUNC=do_test
>  TST_POS_ARGS=2
>  TST_USAGE=usage
>  TST_NEEDS_ROOT=1
> +TST_NEEDS_TMPDIR=1
>  
>  . tst_test.sh
>  
> @@ -114,10 +115,12 @@ do_test()
>  	for i in `seq 1 $max`; do
>  		for j in `seq 1 $nbprocess`; do
>  			cpuacct_task $testpath/subgroup_$i/tasks &
> +			echo $! >> task_pids
>  		done
>  	done
>  
> -	wait
> +	for pid in $(cat task_pids); do wait $pid; done
> +	rm -f task_pids
>  
>  	acc=0
>  	fails=0

[LTP] [PATCH] controllers/cpuacct.sh: Fix timeout for cpuacct.sh

[Petr Vorel]

Hi Xiao,

> With commit fbea02a, cpuacct.sh always gets timeout because wait
> command tries to wait all background processes including the timeout
> process.  We should just wait the processes created for test.

> e.g. running cpuacct_1_1 gets the following timeout:
> -----------------------------------------------------------
> cpuacct 1 TINFO: timeout per run is 0h 5m 0s
> cpuacct 1 TINFO: cpuacct:
> cpuacct 1 TINFO: Creating 1 subgroups each with 1 processes
> cpuacct 1 TBROK: test killed, timeout!
> -----------------------------------------------------------

> Fixes: fbea02a("lib/tst_test.sh: setup timeout per test run for the shell tests")
> Signed-off-by: Xiao Yang <yangx.jy@cn.fujitsu.com>

Thanks for your fix, merged.

I look for other plain wait usage, this was the only one using new shell API.

Kind regards,
Petr