[RFCv1 11/12] drm: Atomic modeset ioctl

[Rob Clark]

On Mon, Oct 7, 2013 at 1:56 PM, Daniel Vetter <daniel at ffwll.ch> wrote:
> On Mon, Oct 07, 2013 at 11:20:54AM -0400, Rob Clark wrote:
>> yeah, current atomic patch is almost an afterthought in this series.
>> I still need to get planes working on msm and find some test code (I
>> guess you have some somewhere?)..  so other than just rebasing it
>> enough to compile, I didn't really spend any time on it yet.  Right
>> now I was more interested in getting folks to have a look at the
>> atomic funcs and helpers.
>
> One in-kernel test user we have is the fbdev restore code - it needs to
> get rid of all planes, cursors and set up its new config on all crtcs. I
> think this would be a fairly nice (albeit really basic testcase): You can
> set up a bunch of planes/cursors, then vt-switch to the fbcon - this way
> we can exercise the atomic stuff in drivers a bit better even without
> needing to add any ioctl.

right..  well, we do have code that disables the planes on fbdev
restore, and fbdev restore (and initial modeset) still hit
crtc->set_config directly.  But once those are converted it would be
cleaner / more elegant to do fbdev restore w/ atomic.

BR,
-R

> -Daniel
> --
> Daniel Vetter
> Software Engineer, Intel Corporation
> +41 (0) 79 365 57 48 - http://blog.ffwll.ch