[Intel-gfx] [PATCH 2/3] drm/i915/selftests: Use full release for local ppgtt allocation

[Chris Wilson]

Quoting Tvrtko Ursulin (2018-07-05 10:50:36)
> 
> On 05/07/2018 07:56, Chris Wilson wrote:
> > We can now use the full release mechanism (i915_ppgtt_put) for our local
> > ppgtt allocation in igt_ppgtt_alloc.
> > 
> > Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
> > ---
> >   drivers/gpu/drm/i915/selftests/i915_gem_gtt.c | 3 +--
> >   1 file changed, 1 insertion(+), 2 deletions(-)
> > 
> > diff --git a/drivers/gpu/drm/i915/selftests/i915_gem_gtt.c b/drivers/gpu/drm/i915/selftests/i915_gem_gtt.c
> > index 4bfb0537f9be..e108fe4e0fd9 100644
> > --- a/drivers/gpu/drm/i915/selftests/i915_gem_gtt.c
> > +++ b/drivers/gpu/drm/i915/selftests/i915_gem_gtt.c
> > @@ -202,9 +202,8 @@ static int igt_ppgtt_alloc(void *arg)
> >   
> >   err_ppgtt_cleanup:
> >       mutex_lock(&dev_priv->drm.struct_mutex);
> > -     ppgtt->vm.cleanup(&ppgtt->vm);
> > +     i915_ppgtt_put(ppgtt);
> >       mutex_unlock(&dev_priv->drm.struct_mutex);
> > -     kfree(ppgtt);
> >       return err;
> >   }
> >   
> > 
> 
> Hm, it's not obvious why this works, if it does. This ppgtt it creates 
> does not call kref_init on ppgtt->ref, or i915_address_space_init. 
> Colour me confused.

Are you sure? There were a few ordering issues that required pushing
i915_address_space_init earlier into __hw_ppgtt_create() itself that
dragged the kref_init along for the fun.
-Chris