[Mesa-dev] V3 On disk shader cache for i965 (Now with real world results!)

[Timothy Arceri]

On Sat, 2016-07-09 at 20:21 +0300, Grazvydas Ignotas wrote:
> On Sat, Jul 9, 2016 at 10:02 AM, Timothy Arceri
> <timothy.arceri at collabora.com> wrote:
> > On Fri, 2016-07-01 at 14:12 +1000, Timothy Arceri wrote:
> > > On Thu, 2016-06-30 at 00:59 +0300, Grazvydas Ignotas wrote:
> > > > On Wed, Jun 29, 2016 at 3:11 PM, Timothy Arceri
> > > > <timothy.arceri at collabora.com> wrote:
> > > > > On Wed, 2016-06-29 at 03:47 +0300, Grazvydas Ignotas wrote:
> > > > 
> > > > Steps to reproduce:
> > > > rm -rf ~/.cache/mesa
> > > > MESA_GLSL_CACHE_ENABLE=1 glretrace -b
> > > > Soma_slow_trim.x86_64.trace
> > > > MESA_GLSL_CACHE_ENABLE=1 glretrace -b Soma.bin.x86_64.trace #
> > > > from
> > > > bug 92229
> > > > 
> > > > The first one should hit an assert due to reasons unrelated to
> > > > cache,
> > > > after that playing the second one crashes on free() due to some
> > > > corruption for me. If you remove the "bad" cache and just play
> > > > the
> > > > second one, it works with empty and it's own full cache.
> > > 
> > > I couldn't hit a crash using this method. However these traces do
> > > hit
> > > a
> > > bug with the previous hack I did that could be the problem you
> > > are
> > > seeing. I've pushed a fix to the same branch.
> > > 
> > > Now I'm trying to track down a problem where the walls and floor
> > > in
> > > the
> > > room are missing in the trace from the bug on the first run. They
> > > are
> > > there in the second run so it seems like another bug in the
> > > fallback
> > > code.
> > 
> > Not sure if you are still interested in testing but I've just
> > pushed
> > some updates to the shader-cache branch. These replace the missing
> > shader source hack with a proper solution, fix some UBO bugs and
> > fix a
> > couple of other issues noticed alone the way.
> > 
> > More than happy for you to point out further issues :)
> 
> I think I still have some more:
> - running 32bit program after 64bit version of the same thing (or
> vice
> versa) leads to segfaults and assert hits. Probably easiest to build
> 32bit and 64bit apitrace and play some traces.
> - a trace from comment 2 of bug 96624 doesn't seem to render on
> cache.

I can't seem to reproduce this. I created some traces using the 32-bit
and 64-bit versions of The Talos Principle but they seems to run with
no issues. I've pushed updates to the shader-cache branch maybe I fixed
something. 

> 
> Other than that my traces and games seem to be fine, will report in
> case I encounter anything else.

Thanks again for testing.

> 
> Gražvydas
> _______________________________________________
> mesa-dev mailing list
> mesa-dev at lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/mesa-dev