drm/mm: Do DRM_MM_CREATE_TOP adj_start calculation after color_adjust
authorMichel Thierry <michel.thierry@intel.com>
Sun, 16 Aug 2015 03:02:28 +0000 (04:02 +0100)
committerDaniel Vetter <daniel.vetter@ffwll.ch>
Tue, 8 Sep 2015 11:45:47 +0000 (13:45 +0200)
The adj_start calculation for DRM_MM_CREATE_TOP should happen after
mm->color_adjust. There was an inconsistency between
drm_mm_insert_helper_range
and drm_mm_insert_helper, as the later was already updating after
color_adjust.

Didn't spot it before, as color_adjust is only done in systems without
LLC. But I'm not aware of anybody using this test case yet.

Signed-off-by: Michel Thierry <michel.thierry@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
drivers/gpu/drm/drm_mm.c

index 3427b11..04de6fd 100644 (file)
@@ -267,12 +267,12 @@ static void drm_mm_insert_helper_range(struct drm_mm_node *hole_node,
        if (adj_end > end)
                adj_end = end;
 
-       if (flags & DRM_MM_CREATE_TOP)
-               adj_start = adj_end - size;
-
        if (mm->color_adjust)
                mm->color_adjust(hole_node, color, &adj_start, &adj_end);
 
+       if (flags & DRM_MM_CREATE_TOP)
+               adj_start = adj_end - size;
+
        if (alignment) {
                u64 tmp = adj_start;
                unsigned rem;