[Intel-gfx] [PATCH 10/11] drm/i915: bind objects to the global gtt only when needed

[Daniel Vetter]

And track the existence of such a binding similar to the aliasing
ppgtt case. Speeds up binding/unbinding in the common case where we
only need a ppgtt binding (which is accessed in a cpu coherent fashion
by the gpu) and no gloabl gtt binding (which needs uc writes for the
ptes).

Signed-Off-by: Daniel Vetter <daniel.vetter at ffwll.ch>
---
 drivers/gpu/drm/i915/i915_drv.h            |    1 +
 drivers/gpu/drm/i915/i915_gem.c            |   16 +++++++++++++---
 drivers/gpu/drm/i915/i915_gem_execbuffer.c |   10 ++++++++++
 drivers/gpu/drm/i915/i915_gem_gtt.c        |    4 ++++
 4 files changed, 28 insertions(+), 3 deletions(-)

diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index 51de703..462dd11 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -857,6 +857,7 @@ struct drm_i915_gem_object {
 	unsigned int cache_level:2;
 
 	unsigned int has_aliasing_ppgtt_mapping:1;
+	unsigned int has_global_gtt_mapping:1;
 
 	struct page **pages;
 
diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index 234eae8..ee907d8 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -1224,6 +1224,9 @@ int i915_gem_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
 			goto unlock;
 	}
 
+	if (!obj->has_global_gtt_mapping)
+		i915_gem_gtt_bind_object(obj, obj->cache_level);
+
 	if (obj->tiling_mode == I915_TILING_NONE)
 		ret = i915_gem_object_put_fence(obj);
 	else
@@ -2133,7 +2136,8 @@ i915_gem_object_unbind(struct drm_i915_gem_object *obj)
 
 	trace_i915_gem_object_unbind(obj);
 
-	i915_gem_gtt_unbind_object(obj);
+	if (obj->has_global_gtt_mapping)
+		i915_gem_gtt_unbind_object(obj);
 	if (obj->has_aliasing_ppgtt_mapping) {
 		i915_ppgtt_unbind_object(dev_priv->mm.aliasing_ppgtt, obj);
 		obj->has_aliasing_ppgtt_mapping = 0;
@@ -2792,7 +2796,9 @@ i915_gem_object_bind_to_gtt(struct drm_i915_gem_object *obj,
 
 		goto search_free;
 	}
-	i915_gem_gtt_bind_object(obj, obj->cache_level);
+
+	if (!dev_priv->mm.aliasing_ppgtt)
+		i915_gem_gtt_bind_object(obj, obj->cache_level);
 
 	list_add_tail(&obj->gtt_list, &dev_priv->mm.gtt_list);
 	list_add_tail(&obj->mm_list, &dev_priv->mm.inactive_list);
@@ -2986,7 +2992,8 @@ int i915_gem_object_set_cache_level(struct drm_i915_gem_object *obj,
 				return ret;
 		}
 
-		i915_gem_gtt_bind_object(obj, cache_level);
+		if (obj->has_global_gtt_mapping)
+			i915_gem_gtt_bind_object(obj, cache_level);
 		if (obj->has_aliasing_ppgtt_mapping)
 			i915_ppgtt_bind_object(dev_priv->mm.aliasing_ppgtt,
 					       obj, cache_level);
@@ -3369,6 +3376,9 @@ i915_gem_object_pin(struct drm_i915_gem_object *obj,
 			return ret;
 	}
 
+	if (!obj->has_global_gtt_mapping && map_and_fenceable)
+		i915_gem_gtt_bind_object(obj, obj->cache_level);
+
 	if (obj->pin_count++ == 0) {
 		if (!obj->active)
 			list_move_tail(&obj->mm_list,
diff --git a/drivers/gpu/drm/i915/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
index 9c81cda..c7a681a 100644
--- a/drivers/gpu/drm/i915/i915_gem_execbuffer.c
+++ b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
@@ -381,6 +381,16 @@ i915_gem_execbuffer_relocate_entry(struct drm_i915_gem_object *obj,
 		io_mapping_unmap_atomic(reloc_page);
 	}
 
+	/* Sandybridge PPGTT errata: We need a global gtt mapping for MI and
+	 * pipe_control writes because the gpu doesn't properly redirect them
+	 * through the ppgtt for non_secure batchbuffers. */
+	if (unlikely(IS_GEN6(dev) &&
+	    reloc->write_domain == I915_GEM_DOMAIN_INSTRUCTION &&
+	    !obj->has_global_gtt_mapping)) {
+		i915_gem_gtt_bind_object(obj, obj->cache_level);
+		obj->has_global_gtt_mapping = 1;
+	}
+
 	/* and update the user's relocation entry */
 	reloc->presumed_offset = target_offset;
 
diff --git a/drivers/gpu/drm/i915/i915_gem_gtt.c b/drivers/gpu/drm/i915/i915_gem_gtt.c
index bfa500d..f437d4b 100644
--- a/drivers/gpu/drm/i915/i915_gem_gtt.c
+++ b/drivers/gpu/drm/i915/i915_gem_gtt.c
@@ -388,12 +388,16 @@ void i915_gem_gtt_bind_object(struct drm_i915_gem_object *obj,
 				       obj->base.size >> PAGE_SHIFT,
 				       obj->pages,
 				       agp_type);
+
+	obj->has_global_gtt_mapping = 1;
 }
 
 void i915_gem_gtt_unbind_object(struct drm_i915_gem_object *obj)
 {
 	intel_gtt_clear_range(obj->gtt_space->start >> PAGE_SHIFT,
 			      obj->base.size >> PAGE_SHIFT);
+
+	obj->has_global_gtt_mapping = 0;
 }
 
 void i915_gem_gtt_finish_object(struct drm_i915_gem_object *obj)
-- 
1.7.7.3