If we fail to unbind and so abort the change in tiling, we will have removed the VMA for the object for no reason. The likelihood of unbind failing is slim (other than ERESTARTSYS which will cause userspace to try again), so the change is mostly for the principle. Also improve the slightly stale comment. Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk> --- drivers/gpu/drm/i915/i915_gem_tiling.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/i915/i915_gem_tiling.c b/drivers/gpu/drm/i915/i915_gem_tiling.c index c9786cd..b964df5 100644 --- a/drivers/gpu/drm/i915/i915_gem_tiling.c +++ b/drivers/gpu/drm/i915/i915_gem_tiling.c @@ -354,14 +354,15 @@ i915_gem_set_tiling(struct drm_device *dev, void *data, /* We need to rebind the object if its current allocation * no longer meets the alignment restrictions for its new * tiling mode. Otherwise we can just leave it alone, but - * need to ensure that any fence register is cleared. + * need to ensure that any fence register is updated before + * the next fenced (either through the GTT or by the BLT unit + * on older GPUs) access. * * After updating the tiling parameters, we then flag whether * we need to update an associated fence register. Note this * has to also include the unfenced register the GPU uses * whilst executing a fenced command for an untiled object. */ - i915_gem_release_mmap(obj); obj->map_and_fenceable = obj->gtt_space == NULL || @@ -385,6 +386,9 @@ i915_gem_set_tiling(struct drm_device *dev, void *data, obj->tiling_mode = args->tiling_mode; obj->stride = args->stride; + + /* Force the fence to be reacquired for GTT access */ + i915_gem_release_mmap(obj); } } /* we have to maintain this existing ABI... */ -- 1.7.10
