[virglrenderer-devel] [PATCH] formats: include compressed formats in the copy-compatibility check

[Gert Wollny]

Am Freitag, den 03.08.2018, 20:39 -0700 schrieb Gurchetan Singh:
> On Fri, Aug 3, 2018 at 2:02 PM Gert Wollny <gert.wollny at collabora.com
> > wrote:
> > 
> > Add the combinations that include compressed format to the check to
> > use
> > glCopyImageSubData. This fixes a number tests of "arb_copy_image-
> > formats"
> > were compressed textures are the source and uncompressed textures
> > the
> > destination (no regressions).
> > 
> > Signed-off-by: Gert Wollny <gert.wollny at collabora.com>
> > ---
> > Two remarks:
> > 
> > * Given the way the piglit is written it will be impossible to fix
> > the tests
> >   where the destination texture is a compressed texture when run on
> > a GLES
> >   host, because this would require glGetCompressedTexImage, (unless
> > of course
> >   one would want to mirror each copy operation on the backing IOVs.
> >   On a GL host the ways readbacks are done would need an overhaul
> > to give
> >   precedence to the  gettextimage code path.
> 
> 
> > * All the dEQP copy_image tests are already passing,
> 
> Interesting.  Even with this patch applied, I'm seeing many
> copy_image failures (

> see https://gitlab.freedesktop.org/virgl/virglrenderer/issu
> es/1
> for a list).  Do you not see these failures in your setup?  Those
> results are from a Nvidia blob driver.
Currently, all the astc tests are marked as NotSupported, but all the
other tests pass on the Intel host. I think unlike the piglit, instead
of comparing the conmpressed textures they actually render the texture
and compare the rendered results, otherwise they could not pass, but
this doesn't explain why they wouldn't pass with the blob. 

Best, 
Gert