]> www.infradead.org Git - users/jedix/linux-maple.git/commitdiff
drm/buddy: check range allocation matches alignment
authorMatthew Auld <matthew.auld@intel.com>
Mon, 19 Feb 2024 12:18:53 +0000 (12:18 +0000)
committerChristian König <christian.koenig@amd.com>
Wed, 28 Feb 2024 07:02:53 +0000 (08:02 +0100)
Likely not a big deal for real users, but for consistency we should
respect the min_page_size here. Main issue is that bias allocations
turns into normal range allocation if the range and size matches
exactly, and in the next patch we want to add some unit tests for this
part of the api.

Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Arunpravin Paneer Selvam <Arunpravin.PaneerSelvam@amd.com>
Cc: Christian König <christian.koenig@amd.com>
Reviewed-by: Arunpravin Paneer Selvam <Arunpravin.PaneerSelvam@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240219121851.25774-5-matthew.auld@intel.com
Signed-off-by: Christian König <christian.koenig@amd.com>
drivers/gpu/drm/drm_buddy.c

index f3a6ac908f815118ac39a4f18cf7cf0bd1955fcd..5ebdd6f8f36e6bc8d67e99a54bac3856d45ac9eb 100644 (file)
@@ -771,8 +771,12 @@ int drm_buddy_alloc_blocks(struct drm_buddy *mm,
                return -EINVAL;
 
        /* Actual range allocation */
-       if (start + size == end)
+       if (start + size == end) {
+               if (!IS_ALIGNED(start | end, min_block_size))
+                       return -EINVAL;
+
                return __drm_buddy_alloc_range(mm, start, size, NULL, blocks);
+       }
 
        original_size = size;
        original_min_size = min_block_size;