[LTP] [PATCH 2/2] tst_test: using SIGTERM to terminate process

[Joerg Vehlow]

Hi,

On 5/31/2021 2:49 PM, Cyril Hrubis wrote:
> Hi!
>>   lib/newlib_tests/shell/test_timeout.sh | 2 +-
>>   lib/newlib_tests/shell/timeout03.sh    | 1 +
>>   testcases/lib/tst_test.sh              | 9 +++++----
>>   3 files changed, 7 insertions(+), 5 deletions(-)
>>
>> diff --git a/lib/newlib_tests/shell/test_timeout.sh b/lib/newlib_tests/shell/test_timeout.sh
>> index b05680cb1..9f31afa32 100755
>> --- a/lib/newlib_tests/shell/test_timeout.sh
>> +++ b/lib/newlib_tests/shell/test_timeout.sh
>> @@ -28,7 +28,7 @@ timeout02.sh|  -10|0|  |2
>>   timeout02.sh| -0.1|0|  |0
>>   timeout02.sh| -1.1|0|  |2
>>   timeout02.sh|-10.1|0|  |2
>> -timeout03.sh|     |0|12|137| | | |Test kill if test does not terminate by SIGINT
>> +timeout03.sh|     |0|12|137| | | |Test kill if test does not terminate by SIGTERM
>>   timeout04.sh|     |0|  |  2|0|0|1|Verify that timeout is enforced
>>   timeout02.sh|    2|1| 2|   |1|0|0|Test termination of timeout process
>>   "
>> diff --git a/lib/newlib_tests/shell/timeout03.sh b/lib/newlib_tests/shell/timeout03.sh
>> index cd548d9a2..124e96a84 100755
>> --- a/lib/newlib_tests/shell/timeout03.sh
>> +++ b/lib/newlib_tests/shell/timeout03.sh
>> @@ -30,6 +30,7 @@ TST_TIMEOUT=1
>>   
>>   do_test()
>>   {
>> +	trap "tst_res TINFO 'Sorry, timeout03 is still alive'" TERM
>>   	tst_res TINFO "testing killing test after TST_TIMEOUT"
>>   
>>   	sleep 2
>> diff --git a/testcases/lib/tst_test.sh b/testcases/lib/tst_test.sh
>> index 3a5651c01..66ffde4eb 100644
>> --- a/testcases/lib/tst_test.sh
>> +++ b/testcases/lib/tst_test.sh
>> @@ -21,7 +21,8 @@ export TST_LIB_LOADED=1
>>   . tst_security.sh
>>   
>>   # default trap function
>> -trap "tst_brk TBROK 'test interrupted or timed out'" INT
>> +trap "tst_brk TBROK 'test interrupted'" INT
>> +trap "unset _tst_setup_timer_pid; tst_brk TBROK 'test terminated'" TERM
> I've been looking at this for a while and I think that we should unset
> the _tst_setup_timer_pid at the end of the _tst_timeout_process()
> instead, right?
That won't work or am I missing something? ;) We are in a subshell in 
_tst_timeout_process, it cannot manipulate the environment of the parent 
or is my mental modal of how shell works wrong again?.
> Otherwise we will leave the timeout process sleeping if someone sends
> SIGTERM to the test process from the outside, or do I miss something?
Yes, that is one of the reasons why I initially suggested a different 
signal for timeout. That would cleanly separate the logic.
But on the other hand when will SIGTERM be sent from the outside? This 
is probably only happening on shutdown? In that case it is not really a 
problem.
If someone wants to terminate gracefully using SIGTERM, he could send it 
to the process group. That would kill the timeout process as well.

Jörg