drm: Improve drm_mm search (and fix topdown allocation) with rbtrees

The drm_mm range manager claimed to support top-down insertion, but it
was neither searching for the top-most hole that could fit the
allocation request nor fitting the request to the hole correctly.

In order to search the range efficiently, we create a secondary index
for the holes using either their size or their address. This index
allows us to find the smallest hole or the hole at the bottom or top of
the range efficiently, whilst keeping the hole stack to rapidly service
evictions.

v2: Search for holes both high and low. Rename flags to mode.
v3: Discover rb_entry_safe() and use it!
v4: Kerneldoc for enum drm_mm_insert_mode.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: "Christian König" <christian.koenig@amd.com>
Cc: David Airlie <airlied@linux.ie>
Cc: Russell King <rmk+kernel@armlinux.org.uk>
Cc: Daniel Vetter <daniel.vetter@intel.com>
Cc: Jani Nikula <jani.nikula@linux.intel.com>
Cc: Sean Paul <seanpaul@chromium.org>
Cc: Lucas Stach <l.stach@pengutronix.de>
Cc: Christian Gmeiner <christian.gmeiner@gmail.com>
Cc: Rob Clark <robdclark@gmail.com>
Cc: Thierry Reding <thierry.reding@gmail.com>
Cc: Stephen Warren <swarren@wwwdotorg.org>
Cc: Alexandre Courbot <gnurou@gmail.com>
Cc: Eric Anholt <eric@anholt.net>
Cc: Sinclair Yeh <syeh@vmware.com>
Cc: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Sinclair Yeh <syeh@vmware.com> # vmwgfx
Reviewed-by: Lucas Stach <l.stach@pengutronix.de> #etnaviv
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/20170202210438.28702-1-chris@chris-wilson.co.uk

drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c

@@ -97,8 +97,7 @@ int amdgpu_gtt_mgr_alloc(struct ttm_mem_type_manager *man,
 {
 	struct amdgpu_gtt_mgr *mgr = man->priv;
 	struct drm_mm_node *node = mem->mm_node;
-	enum drm_mm_search_flags sflags = DRM_MM_SEARCH_BEST;
-	enum drm_mm_allocator_flags aflags = DRM_MM_CREATE_DEFAULT;
+	enum drm_mm_insert_mode mode;
 	unsigned long fpfn, lpfn;
 	int r;
 
@@ -115,15 +114,14 @@ int amdgpu_gtt_mgr_alloc(struct ttm_mem_type_manager *man,
 	else
 		lpfn = man->size;
 
-	if (place && place->flags & TTM_PL_FLAG_TOPDOWN) {
-		sflags = DRM_MM_SEARCH_BELOW;
-		aflags = DRM_MM_CREATE_TOP;
-	}
+	mode = DRM_MM_INSERT_BEST;
+	if (place && place->flags & TTM_PL_FLAG_TOPDOWN)
+		mode = DRM_MM_INSERT_HIGH;
 
 	spin_lock(&mgr->lock);
-	r = drm_mm_insert_node_in_range_generic(&mgr->mm, node, mem->num_pages,
-						mem->page_alignment, 0,
-						fpfn, lpfn, sflags, aflags);
+	r = drm_mm_insert_node_in_range(&mgr->mm, node,
+					mem->num_pages, mem->page_alignment, 0,
+					fpfn, lpfn, mode);
 	spin_unlock(&mgr->lock);
 
 	if (!r) {

drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c

@@ -97,8 +97,7 @@ static int amdgpu_vram_mgr_new(struct ttm_mem_type_manager *man,
 	struct amdgpu_vram_mgr *mgr = man->priv;
 	struct drm_mm *mm = &mgr->mm;
 	struct drm_mm_node *nodes;
-	enum drm_mm_search_flags sflags = DRM_MM_SEARCH_DEFAULT;
-	enum drm_mm_allocator_flags aflags = DRM_MM_CREATE_DEFAULT;
+	enum drm_mm_insert_mode mode;
 	unsigned long lpfn, num_nodes, pages_per_node, pages_left;
 	unsigned i;
 	int r;
@@ -121,10 +120,9 @@ static int amdgpu_vram_mgr_new(struct ttm_mem_type_manager *man,
 	if (!nodes)
 		return -ENOMEM;
 
-	if (place->flags & TTM_PL_FLAG_TOPDOWN) {
-		sflags = DRM_MM_SEARCH_BELOW;
-		aflags = DRM_MM_CREATE_TOP;
-	}
+	mode = DRM_MM_INSERT_BEST;
+	if (place->flags & TTM_PL_FLAG_TOPDOWN)
+		mode = DRM_MM_INSERT_HIGH;
 
 	pages_left = mem->num_pages;
 
@@ -135,13 +133,11 @@ static int amdgpu_vram_mgr_new(struct ttm_mem_type_manager *man,
 
 		if (pages == pages_per_node)
 			alignment = pages_per_node;
-		else
-			sflags |= DRM_MM_SEARCH_BEST;
 
-		r = drm_mm_insert_node_in_range_generic(mm, &nodes[i], pages,
-							alignment, 0,
-							place->fpfn, lpfn,
-							sflags, aflags);
+		r = drm_mm_insert_node_in_range(mm, &nodes[i],
+						pages, alignment, 0,
+						place->fpfn, lpfn,
+						mode);
 		if (unlikely(r))
 			goto error;
 

drivers/gpu/drm/armada/armada_gem.c

@@ -148,8 +148,8 @@ armada_gem_linear_back(struct drm_device *dev, struct armada_gem_object *obj)
 			return -ENOSPC;
 
 		mutex_lock(&priv->linear_lock);
-		ret = drm_mm_insert_node(&priv->linear, node, size, align,
-					 DRM_MM_SEARCH_DEFAULT);
+		ret = drm_mm_insert_node_generic(&priv->linear, node,
+						 size, align, 0, 0);
 		mutex_unlock(&priv->linear_lock);
 		if (ret) {
 			kfree(node);