[igt-dev] [PATCH i-g-t] i915/gem_mocs_settings/mocs-isolation: Filter out missing rings

Kumar Valsan, Prathap prathap.kumar.valsan at intel.com
Wed Mar 20 16:59:22 UTC 2019


On Wed, Mar 20, 2019 at 04:35:52PM +0000, Chris Wilson wrote:
> Quoting Kumar Valsan, Prathap (2019-03-20 16:40:50)
> > On Tue, Mar 19, 2019 at 07:23:20PM +0000, Chris Wilson wrote:
> > > Don't try to evaluate whether reads executed on an absent ring do
> > > anything.
> > 
> > Chris,
> > 
> > I am trying to understand this patch and has this question.
> > AFAIK every GPU engine has an assosiated ring buffer per
> > context. Absent ring is a error case?
> 
> Not all HW has the same set of engines available. And we should get a
> failure from gem_execbuf() if you try to dispatch something on a ring
> that didn't exist. What I haven't actually answered is why we actually
> hit a failure later. But it's definitely a requirement of the test that
> the engine exist.
> -Chris

I get it now. Thank you.

-Prathap


More information about the igt-dev mailing list