On Wed, 19 Dec 2012 08:10:24 +1000, Dave Airlie <airl...@gmail.com> wrote:
> > As the shrinker may be invoked for the allocation, and it may reap
> > neighbouring objects in the offset range mm, we need to be careful in
> > the order in which we allocate the node, search for free space and then
> > insert the node into the mmap offset range manager.
>
> Maybe I'm being a bit stupid here, but this seems like a pointless
> micro optimisation thrown in with a
> deinlining.
The actual bug fix is the removal of the allocation between searching
for a free node and inserting it into the tree:
diff --git a/drivers/gpu/drm/drm_gem.c b/drivers/gpu/drm/drm_gem.c
index 24efae4..b885f2c 100644
--- a/drivers/gpu/drm/drm_gem.c
+++ b/drivers/gpu/drm/drm_gem.c
@@ -353,8 +353,8 @@ drm_gem_create_mmap_offset(struct drm_gem_object *obj)
map->handle = obj;
/* Get a DRM GEM mmap offset allocated... */
- list->file_offset_node = drm_mm_search_free(&mm->offset_manager,
- obj->size / PAGE_SIZE, 0, false);
+ list->file_offset_node = kzalloc(sizeof(*list->file_offset_node),
+ GFP_KERNEL);
if (!list->file_offset_node) {
DRM_ERROR("failed to allocate offset for bo %d\n", obj->name);
@@ -362,12 +362,11 @@ drm_gem_create_mmap_offset(struct drm_gem_object *obj)
goto out_free_list;
}
- list->file_offset_node = drm_mm_get_block(list->file_offset_node,
- obj->size / PAGE_SIZE, 0);
- if (!list->file_offset_node) {
- ret = -ENOMEM;
+ ret = drm_mm_insert_node(&mm->offset_manage,
+ list->file_offset_node,
+ obj->size / PAGE_SIZE, 0);
+ if (ret)
goto out_free_list;
- }
list->hash.key = list->file_offset_node->start;
ret = drm_ht_insert_item(&mm->offset_hash, &list->hash);
--
Chris Wilson, Intel Open Source Technology Centre
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
