[Intel-gfx] [PATCH 2/2] drm/i915: fix the racy object accounting

[Daniel Vetter]

On Wed, Jul 24, 2013 at 06:09:14PM +0100, Chris Wilson wrote:
> On Wed, Jul 24, 2013 at 03:49:43PM +0200, Daniel Vetter wrote:
> > Use an atomic_t for the count (since that is useful for leak detection
> > in tests) and just rip out the object memory check.
> > 
> > v2: Rebase onto the new object create refcount fix patch.
> > 
> > Cc: Chris Wilson <chris at chris-wilson.co.uk>
> > Signed-off-by: Daniel Vetter <daniel.vetter at ffwll.ch>
> 
> I will miss that summary of total allocation bytes, and would actually
> like to have the information provided by the patch you reverted earlier.

We already dump both the bound and unbound list, and imo adding those up
and claiming is everything is too misleading. I guess I could do a
spinlock around the object_memory instead, but since I only cared about
the object count I didn't bother.
-Daniel
-- 
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch