[Intel-gfx] [PATCH 1/2] drm/i915: Defer assignment of obj->gtt_space until after all possible mallocs

[Chris Wilson]

On Wed, 21 Nov 2012 14:27:06 +0100, Daniel Vetter <daniel at ffwll.ch> wrote:
> Ok, I now see what's going on. Can't we do the bind together with the
> mappable bind, i.e.
> 
> if (!has_global_mapping && (map_and_fenceable || !aliasing_ppgtt))
>         i915_gem_gtt_bind_object()

You can indeed do it that way, it just took up too many lines... Imo, I
think we want to eventually push the i915_gem_gtt_bind_object() to the
callers. (You'll see why later if we ever get the current 3.7 regression
resolved.)
-Chris

-- 
Chris Wilson, Intel Open Source Technology Centre