[igt-dev] [PATCH i-g-t] lib/igt_core: Don't kill the world after a failed fork

[Hebenstreit, Michael]

Ok, thanks

-----Original Message-----
From: Hiler, Arkadiusz <arkadiusz.hiler at intel.com> 
Sent: Wednesday, June 03, 2020 10:25
To: Hebenstreit, Michael <michael.hebenstreit at intel.com>
Cc: igt-dev at lists.freedesktop.org; Yang, Fei <fei.yang at intel.com>; Latvala, Petri <petri.latvala at intel.com>
Subject: Re: [PATCH i-g-t] lib/igt_core: Don't kill the world after a failed fork


On Wed, Jun 03, 2020 at 06:43:45PM +0300, Hebenstreit, Michael wrote:
> Confirmed the patch work - but leaves a question
> 
> Shouldn't this test case then report a fail as I do not allow all forks?
> 
> 10/293 lib: igt_fork                           OK       3.37 s

Thank's for the verification! I'll add 'Tested-by: you' when merging :-)

The test that was failing is subtest_leak.

It forks 4096/sizeof(pid_t) times and makes sure that we have killed all the forked children successfully when exiting igt_subtest block, so we don't leak any of them.

We don't manage to fork that exact number but that's fine, we fork as many as the limits allows. I think this is much better behavior than having this test fail on a more limited systems.

Please keep in mind that this is a "metatest" of the framework. There's an assert inside igt_fork() which would cause any proper IGT test to fail if we cannot fork() anymore.

--
Cheers,
Arek