[Intel-gfx] [PATCH 00/12] drm/i915: Some cleanups and fixes to the sprite code v2

[Daniel Vetter]

On Fri, Oct 04, 2013 at 11:28:16AM +0100, Chris Wilson wrote:
> On Tue, Oct 01, 2013 at 06:02:09PM +0300, ville.syrjala at linux.intel.com wrote:
> > Chris asked for some renames and assertions during v1. While adding those I
> > noticed that what I did in the original patch 02 didn't match quite so well
> > with the assertions. So I modified patch 02 a bit, and that caused quite a bit
> > of bit of rebase issues for most of the other patches, so I figured it's better
> > to repost the whole thing.
> > 
> > Changes from v1:
> > - Move the primary disable/enable calls inside intel_crtc->active checks
> >   in intel_update_plane/intel_disable_plane. That also ate up patch 03 from
> >   the original series.
> > - Add primary_disabled WARNs
> > - Rename primary plane funcs
> > - Flush primary plane changes from sprite code
> > - Add a POSTING_READ() to intel_flush_primary_plane. This shouldn't really
> >   be necessary now that I think about it some more. So we might want to drop
> >   that change...
> 
> Looks good, very good, to me.
> 
> Even with throwing up over FBC,
> 
> Reviewed-by: Chris Wilson <chris at chris-wilson.co.uk>
> 
> except for
> 
> 08/12: drm/i915: Enable/disable IPS when primary is
>        enabled/disabled
> 
> For which the code looks ok, but only merits an
> Acked-by: Chris Wilson <chris at chris-wilson.co.uk>

All merged, thanks for patches&review. Now my problem here is that I'm
really uneasy with our complete lack of testcoverage for all these
corner-cases. I know that we can't really get full functional testing
going for sprites/planes/cursors before we have the CRC stuff all merged,
but at least exercising all this code would be great.

So can we please move "polish/create testcases for cursor/sprite/plane
corner-cases and push the to igt" to the top-spot for all things planes?
At least before we start to wreak utter havoc with atomic
pageflips/modeset I want to have some assurance that we don't break all
the low-level functions we've painstakingly beaten into shape in the past
few months ...

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