Re: Supporting fused display configurations v4

[Daniel Vetter <daniel@ffwll.ch>]

On Mon, Jan 06, 2014 at 08:05:56PM -0200, Paulo Zanoni wrote:
> Hi
> 
> 2014/1/6 Damien Lespiau <damien.lespiau@intel.com>:
> > Follow up of the v3:
> >   http://lists.freedesktop.org/archives/intel-gfx/2013-December/037333.html
> >
> > The main change is the removal of INTEL_INFO() from the driver (patches 1 to
> > 5), a clean-up suggested by Chris: the usage of that macro hides that we go
> > from dev to dev_priv while we could use the dev_priv pointer that is usually
> > available.
> >
> > Patch 6 is also new, a minor clean-up while at it.
> 
> For patches 1 to 6:
> Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
> 
> 
> Some comments/bikesheds (I'm not requiring these changes, just
> throwing some ideas):
> 
> - You removed INTEL_INFO, but all those IS_SOMETHING and HAS_SOMETHING
> macros still accept dev as argument instead of dev_priv. Do we have
> plans to change this too? I can remember many places where I had to
> add a "dev" variable just because of these macros. Perhaps maybe the
> new goal is a series removing the to_i915 macro? I could see a lot of
> code getting almost entirely rid of "dev" with these changes.

Once we have proper subclassing casting between drm_device and i915_device
is very cheap, so I'm ok with sprinkling to_i915 all over those macros. If
we end up with tons of IS_SOMETHING(&dev_priv->base) kind of code in a
glorious future we can then switch things around.

> - It would be nice if we could come up with a clever strategy to find
> out if a still-not-initialized field of intel_device_info is being
> used. No clever ideas here, besides the obvious implementations that
> don't look good.

Same here - would be neat but no idea.

> - Quick! Merge patch 3 before it conflicts and someone needs to review
> that giant thing again!

As long as topic/ppgtt is still hanging out there and not yet merged into
dinq that can't happen due to conflict galore. And since there are still
some regressions pending that'll take a while.

Please don't blame me for this mess, I never really liked this approach of
merging ppgtt ...
-Daniel
-- 
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch