[virglrenderer-devel] [PATCH] vrend_clear: clear and restore rasterizer discard and depth clamp

[Gert Wollny]

Am Montag, den 14.05.2018, 08:17 -0700 schrieb Gurchetan Singh:
> On Sun, May 13, 2018 at 9:51 PM Gert Wollny
> <gert.wollny at collabora.com> 
> wrote:
> 
> > 
> It's possible the default gl-config-name doesn't have any depth bits.
> Can you test with different configs?

I tried, and whenever I give d24 explicitely, it fails, I was not able
to hunt down what the actual default configuration is.   

> 
> > When I emit a clear rasterizer state from the guest before the
> > actual clear, then the tests pass, same when I emit an
> > st_validate_state(...RENDER) in the guest st_Clear.
> 
> Ideally, a guest side patch won't be required since we can infer the
> desired state from the semantics of a pipe->clear().  What does
> sending out a default rasterizer state from the guest change that our
> current logic doesn't?

I was trying to figure that out, then I went back to the beginning,
using only current master of mesa and virglrenderer and voila,
disableling the DEPTH from by former parts now worked. I guess I was
mislead because on R600 some tests are unstable, and also by the
somewhat surprising fact that the default config for deqp-gles3 doesn't
seem to have depth bits, which made me thing the tests pass. 

I'm currently rerunning the tests with a patch that I will send to the
list once the results are confirmed.

Best,
Getr