[Piglit] [PATCH 01/16] fbo-clear-formats: fix the stencil clear test

[Marek Olšák]

I think the problem was that the source and destination depth-stencil
buffers must be of the same format, which was failing with Gallium.
The window framebuffer has D24S8, but the texture is
DEPTH32F_STENCIL8. I wonder why you didn't see the error. The test
had been failing for half a year at least!

Marek

On Tue, Dec 3, 2013 at 10:06 PM, Eric Anholt <eric at anholt.net> wrote:
> Marek Olšák <maraeo at gmail.com> writes:
>
>> From: Marek Olšák <marek.olsak at amd.com>
>>
>> glBlitFramebuffer generates a GL error, good old glCopyPixels doesn't.
>
> When does it generate an error?  I tried to hit this path and failed to
> generate an error, and I don't see what error you're trying to avoid.
>
> I'm in favor of this series in general.  Actually, I'd really like to
> switch things around so that concurrent and fbo is the default, and
> winsys tests explicitly disale fbo, winsys frontbuffer tests disable
> concurrent, and timing-dependent tests disable concurrent.