drm/i915: Mark CPU cache as dirty on every transition for CPU writes

Currently, we only mark the CPU cache as dirty if we skip a clflush.
This leads to some confusion where we have to ask if the object is in
the write domain or missed a clflush. If we always mark the cache as
dirty, this becomes a much simply question to answer.

The goal remains to do as few clflushes as required and to do them as
late as possible, in the hope of deferring the work to a kthread and not
block the caller (e.g. execbuf, flips).

v2: Always call clflush before GPU execution when the cache_dirty flag
is set. This may cause some extra work on llc systems that migrate dirty
buffers back and forth - but we do try to limit that by only setting
cache_dirty at the end of the gpu sequence.

v3: Always mark the cache as dirty upon a level change, as we need to
invalidate any stale cachelines due to external writes.

Reported-by: Dongwon Kim <dongwon.kim@intel.com>
Fixes: a6a7cc4b7db6 ("drm/i915: Always flush the dirty CPU cache when pinning the scanout")
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Dongwon Kim <dongwon.kim@intel.com>
Cc: Matt Roper <matthew.d.roper@intel.com>
Tested-by: Dongwon Kim <dongwon.kim@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170615123850.26843-1-chris@chris-wilson.co.uk
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>

1 files changed, 7 insertions, 14 deletions

diff --git a/drivers/gpu/drm/i915/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
index 96705171e397..2a9aed5640e2 100644
--- a/drivers/gpu/drm/i915/i915_gem_execbuffer.c
+++ b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
@@ -309,7 +309,7 @@ static inline int use_cpu_reloc(struct drm_i915_gem_object *obj)
 		return DBG_USE_CPU_RELOC > 0;
 
 	return (HAS_LLC(to_i915(obj->base.dev)) ||
-		obj->base.write_domain == I915_GEM_DOMAIN_CPU ||
+		obj->cache_dirty ||
 		obj->cache_level != I915_CACHE_NONE);
 }
 
@@ -1110,10 +1110,8 @@ eb_move_to_gpu(struct i915_execbuffer *eb)
 		if (vma->exec_entry->flags & EXEC_OBJECT_ASYNC)
 			continue;
 
-		if (obj->base.write_domain & I915_GEM_DOMAIN_CPU) {
+		if (obj->cache_dirty)
 			i915_gem_clflush_object(obj, 0);
-			obj->base.write_domain = 0;
-		}
 
 		ret = i915_gem_request_await_object
 			(eb->request, obj, obj->base.pending_write_domain);
@@ -1248,12 +1246,6 @@ static int eb_select_context(struct i915_execbuffer *eb)
 	return 0;
 }
 
-static bool gpu_write_needs_clflush(struct drm_i915_gem_object *obj)
-{
-	return !(obj->cache_level == I915_CACHE_NONE ||
-		 obj->cache_level == I915_CACHE_WT);
-}
-
 void i915_vma_move_to_active(struct i915_vma *vma,
 			     struct drm_i915_gem_request *req,
 			     unsigned int flags)
@@ -1277,15 +1269,16 @@ void i915_vma_move_to_active(struct i915_vma *vma,
 	i915_gem_active_set(&vma->last_read[idx], req);
 	list_move_tail(&vma->vm_link, &vma->vm->active_list);
 
+	obj->base.write_domain = 0;
 	if (flags & EXEC_OBJECT_WRITE) {
+		obj->base.write_domain = I915_GEM_DOMAIN_RENDER;
+
 		if (intel_fb_obj_invalidate(obj, ORIGIN_CS))
 			i915_gem_active_set(&obj->frontbuffer_write, req);
 
-		/* update for the implicit flush after a batch */
-		obj->base.write_domain &= ~I915_GEM_GPU_DOMAINS;
-		if (!obj->cache_dirty && gpu_write_needs_clflush(obj))
-			obj->cache_dirty = true;
+		obj->base.read_domains = 0;
 	}
+	obj->base.read_domains |= I915_GEM_GPU_DOMAINS;
 
 	if (flags & EXEC_OBJECT_NEEDS_FENCE)
 		i915_gem_active_set(&vma->last_fence, req);