drm/i915: Always convert incoming exec offsets to non-canonical
authorMichał Winiarski <michal.winiarski@intel.com>
Tue, 7 Feb 2017 19:55:59 +0000 (20:55 +0100)
committerChris Wilson <chris@chris-wilson.co.uk>
Wed, 8 Feb 2017 09:22:31 +0000 (09:22 +0000)
commit038c95a313e4ca954ee5ab8a0c7559a646b0f462
tree4605140e40ff5a2d83f1fbba06760bffd26db602
parentc0dcb203fb009678e5be9e7782329dcfbbf16439
drm/i915: Always convert incoming exec offsets to non-canonical

We're using non-canonical addresses in drm_mm, and we're making sure that
userspace is using canonical addressing - both in case of softpin
(verifying incoming offset) and when relocating (converting to canonical
when updating offset returned to userspace).
Unfortunately when considering the need for relocations, we're comparing
offset from userspace (in canonical form) with drm_mm node (in
non-canonical form), and as a result, we end up always relocating if our
offsets are in the "problematic" range.
Let's always convert the offsets to avoid the performance impact of
relocations.

Fixes: a5f0edf63bdf ("drm/i915: Avoid writing relocs with addresses in non-canonical form")
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Michel Thierry <michel.thierry@intel.com>
Reported-by: Michał Pyrzowski <michal.pyrzowski@intel.com>
Signed-off-by: Michał Winiarski <michal.winiarski@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170207195559.18798-1-michal.winiarski@intel.com
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
drivers/gpu/drm/i915/i915_gem_execbuffer.c