Re: [LTP] [PATCH v1] Refactor fork12 using new LTP API

[Richard Palethorpe <rpalethorpe@suse.de>]

Hello,

Cyril Hrubis <chrubis@suse.cz> writes:

> Hi!
>> This test also randomly fails outside of a container. Also other tests
>> that are testing the limits. This makes me think more that setting lower
>> prlimits is needed. Also this rewrite gets higher priority.
>
> Just note that this test is not in syscalls runtest file but in the
> crashme runtest file, which contains highly questionable stuff.
>
> I guess that the original test does not really take things like
> overcommit and OMM into an account, so shifting the test goals by
> setting the RLIMIT_NPROC so that we effectively check that the
> limits are enforced is probably reasonable way how to fix the test.
> Either we do that or we remove fork12.c.

Looking at the setrlimit tests we already do this in setrlimit01 as
well.

I guess someone might want to test a fork bomb. However I don't see how
it could be a reliable or meaningful test unless you set reasonable
limits for the particular system that the test is running on.

Just a thought; IMO stress tests are better handled by a tool like
stress-ng and some bespoke scripts for a particular system. Or else we
have to create a framework inside LTP for deciding on and implementing
reasonable limits.


-- 
Thank you,
Richard.

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp