[igt-dev] Non-dumb framebuffer tests

[igt-dev] Non-dumb framebuffer tests

[Simon Ser]

Hi,

I've been wondering about tests with non-dumb framebuffers. From what I've
understood, the create_bo_for_fb helper will use dumb framebuffers unless
the format is tiling or a specific size/stride is requested or it's YUV.

This makes it so that some framebuffer type/usage combinations are not
tested, especially for simple RGBA formats. For instance, cursor tests
seem to always use dumb framebuffers.

Would it be useful to add GEM (or GBM?) framebuffer tests?

Thanks,

--
Simon Ser
https://emersion.fr

_______________________________________________
igt-dev mailing list
igt-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/igt-dev

Re: [igt-dev] Non-dumb framebuffer tests

[Daniel Vetter]

On Fri, Jan 18, 2019 at 09:07:48AM +0000, Simon Ser wrote:
> Hi,
> 
> I've been wondering about tests with non-dumb framebuffers. From what I've
> understood, the create_bo_for_fb helper will use dumb framebuffers unless
> the format is tiling or a specific size/stride is requested or it's YUV.

Hm, maybe we need dumb_create2, like we have addfb2?

> This makes it so that some framebuffer type/usage combinations are not
> tested, especially for simple RGBA formats. For instance, cursor tests
> seem to always use dumb framebuffers.
> 
> Would it be useful to add GEM (or GBM?) framebuffer tests?

Yeah. I think longer-term we might want to have plug-in library for
allocating buffers on the various gpus, if things get too extreme. Pulling
in something like gbm is probably too much (and wont solve anything for
display-only drivers).

This also gets into areas where kms isn't all the well-specified, so
definitely good to have better tests for it.
-Daniel
-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch
_______________________________________________
igt-dev mailing list
igt-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/igt-dev