[Intel-gfx] [PATCH v3] drm/i915: Do not use ggtt_view with (aliasing) PPGTT

[Joonas Lahtinen]

Hi,

Regression testing completed without problems for BYT, HSW and BDW
already.

On ma, 2015-03-16 at 13:26 +0000, Tvrtko Ursulin wrote:
> Hi,
> 
> On 03/16/2015 12:11 PM, Joonas Lahtinen wrote:
> > GGTT views are only applicable when dealing with GGTT. Change the code to
> > reject ggtt_view where it should not be used and require it when it should
> > be.
> >
> > v2:
> > - Dropped _ppgtt_ infixes, allow both types to be passed
> > - Disregard other but normal views when no view is specified
> > - More checks that valid parameters are passed
> > - More readable error checking
> >
> > v3:
> > - Prefer WARN_ONCE over BUG_ON when there is code path for failure
> 
> [snip]
> 
> > +i915_gem_obj_to_vma(struct drm_i915_gem_object *obj,
> > +		    struct i915_address_space *vm);
> > +struct i915_vma *
> > +i915_gem_obj_to_ggtt_view(struct drm_i915_gem_object *obj,
> > +			  const struct i915_ggtt_view *view);
> 
> Would i915_gem_obj_to_ggtt_vma be a better name? At least should have 
> vma in the name I think.
> 

The i915_gem_obj_to_ggtt functions doesn't mention _vma either (and
would cause a lot of changes all around code to change), so I decided to
stay with the same convention. In that sense it would add more
confusion, compared to the current *_view function being the same as
without _view, but with explicitly specified view.

> > +struct i915_vma *i915_gem_obj_to_vma(struct drm_i915_gem_object *obj,
> > +				     struct i915_address_space *vm)
> >   {
> >   	struct i915_vma *vma;
> > -	list_for_each_entry(vma, &obj->vma_list, vma_link)
> > -		if (vma->vm == vm && vma->ggtt_view.type == view->type)
> > +	list_for_each_entry(vma, &obj->vma_list, vma_link) {
> > +		if (i915_is_ggtt(vma->vm) &&
> > +		    vma->ggtt_view.type != I915_GGTT_VIEW_NORMAL)
> 
> Since there are 4-5 instances of this check it may make sense to add a 
> helper like i915_is_normal_ggtt_view(vma), but it is not that important 
> for me.
> 

This will be done in following patch that makes the view struct (minus
implementation parts like the pages sg_table) define the view.

> The rest looks good to me.
> 

Sound like you could R-B this then?

Best Regards,
Joonas

> Regards,
> 
> Tvrtko