[igt-dev] [PATCH i-g-t] i915/gem_request_retire: Removal of test

Dandamudi, Priyanka priyanka.dandamudi at intel.com
Fri Feb 21 11:25:04 UTC 2020


Thanks Chris, 

"gem_ctx_param" AFAIU, in this program it relies on a quirk of execbuf that it does not try to move a VMA , and so that  having used an object in one context, it will have the same address in the next context that shared the VM. 
Having said that, I would also go through gem_ctx_persistance as per your suggestion.

Regards,
Priyanka



-----Original Message-----
From: Chris Wilson <chris at chris-wilson.co.uk> 
Sent: 20 February 2020 01:18 PM
To: igt-dev at lists.freedesktop.org; Dandamudi, Priyanka <priyanka.dandamudi at intel.com>
Subject: Re: [igt-dev] [PATCH i-g-t] i915/gem_request_retire: Removal of test

Quoting priyanka.dandamudi at intel.com (2020-02-20 05:14:32)
> From: Priyanka Dandamudi <priyanka.dandamudi at intel.com>
> 
> Test is trying to hit a race condition by submitting same buffer 
> object between two contexts on different rings.Similar scenario can be 
> found in gem_ctx_param and gem_exec_ctx.This test doesn't add anything 
> to other tests.So,removing the test.

Neither of those try to replicate the bug found in this regression test.
gem_ctx_persistence does have similar patterns. Are you sure we do have it covered?
-Chris


More information about the igt-dev mailing list