drm/i915: fix order of fence release wrt flushing

i915_gem_object_unbind had the ordering wrong. The other user,
i915_gem_object_put_fence_reg already has the correct ordering.

Results was usually corrupted pixmaps, especially garbled font glyphs
after a suspend/resume (because this evicts everything).

I'm still waiting for the feedback from the bug-reporters, but
because this obviously fixes a bug (at least for me) I'm already
submitting it.

Bugzilla: http://bugs.freedesktop.org/show_bug.cgi?id=25406
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Eric Anholt <eric@anholt.net>
CC: stable@kernel.org

drivers/gpu/drm/i915/i915_gem.c

@@ -2021,9 +2021,6 @@ i915_gem_object_unbind(struct drm_gem_object *obj)
 	/* blow away mappings if mapped through GTT */
 	i915_gem_release_mmap(obj);
 
-	if (obj_priv->fence_reg != I915_FENCE_REG_NONE)
-		i915_gem_clear_fence_reg(obj);
-
 	/* Move the object to the CPU domain to ensure that
 	 * any possible CPU writes while it's not in the GTT
 	 * are flushed when we go to remap it. This will
@@ -2039,6 +2036,10 @@ i915_gem_object_unbind(struct drm_gem_object *obj)
 
 	BUG_ON(obj_priv->active);
 
+	/* release the fence reg _after_ flushing */
+	if (obj_priv->fence_reg != I915_FENCE_REG_NONE)
+		i915_gem_clear_fence_reg(obj);
+
 	if (obj_priv->agp_mem != NULL) {
 		drm_unbind_agp(obj_priv->agp_mem);
 		drm_free_agp(obj_priv->agp_mem, obj->size / PAGE_SIZE);