[Intel-gfx] [PATCH 3/3] drm/i915: Use insert_page for pwrite_fast

[ankitprasad.r.sharma at intel.com]

From: Ankitprasad Sharma <ankitprasad.r.sharma at intel.com>

In pwrite_fast, map an object page by page if obj_ggtt_pin fails. First,
we try a nonblocking pin for the whole object (since that is fastest if
reused), then failing that we try to grab one page in the mappable
aperture. It also allows us to handle objects larger than the mappable
aperture (e.g. if we need to pwrite with vGPU restricting the aperture
to a measely 8MiB or something like that).

v2: Pin pages before starting pwrite, Combined duplicate loops (Chris)

Signed-off-by: Ankitprasad Sharma <ankitprasad.r.sharma at intel.com>
---
 drivers/gpu/drm/i915/i915_gem.c | 68 ++++++++++++++++++++++++++++++++---------
 1 file changed, 53 insertions(+), 15 deletions(-)

diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index bf5ef7a..2556936 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -766,14 +766,26 @@ i915_gem_gtt_pwrite_fast(struct drm_device *dev,
 			 struct drm_file *file)
 {
 	struct drm_i915_private *dev_priv = dev->dev_private;
+	struct drm_mm_node node;
 	ssize_t remain;
 	loff_t offset, page_base;
 	char __user *user_data;
-	int page_offset, page_length, ret;
+	int page_offset, page_length, ret, i = 0;
+	bool pinned = true;
 
 	ret = i915_gem_obj_ggtt_pin(obj, 0, PIN_MAPPABLE | PIN_NONBLOCK);
-	if (ret)
-		goto out;
+	if (ret) {
+		pinned = false;
+		memset(&node, 0, sizeof(node));
+		ret = drm_mm_insert_node_in_range_generic(&dev_priv->gtt.base.mm,
+							  &node, 4096, 0,
+							  I915_CACHE_NONE, 0,
+							  dev_priv->gtt.mappable_end,
+							  DRM_MM_SEARCH_DEFAULT,
+							  DRM_MM_CREATE_DEFAULT);
+		if (ret)
+			goto out;
+	}
 
 	ret = i915_gem_object_set_to_gtt_domain(obj, true);
 	if (ret)
@@ -783,26 +795,41 @@ i915_gem_gtt_pwrite_fast(struct drm_device *dev,
 	if (ret)
 		goto out_unpin;
 
+	i915_gem_object_pin_pages(obj);
+
 	user_data = to_user_ptr(args->data_ptr);
 	remain = args->size;
 
-	offset = i915_gem_obj_ggtt_offset(obj) + args->offset;
-
 	intel_fb_obj_invalidate(obj, ORIGIN_GTT);
 
-	while (remain > 0) {
-		/* Operation in this page
-		 *
-		 * page_base = page offset within aperture
-		 * page_offset = offset within page
-		 * page_length = bytes to copy for this page
-		 */
+	if (pinned) {
+		offset = i915_gem_obj_ggtt_offset(obj) + args->offset;
+		/* page_base = page offset within aperture */
 		page_base = offset & PAGE_MASK;
-		page_offset = offset_in_page(offset);
+	} else {
+		offset = args->offset;
+		page_base = node.start;
+		i = args->offset / PAGE_SIZE;
+	}
+
+	/* page_offset = offset within page */
+	page_offset = offset_in_page(offset);
+	while (remain > 0) {
+		/* page_length = bytes to copy for the page */
 		page_length = remain;
 		if ((page_offset + remain) > PAGE_SIZE)
 			page_length = PAGE_SIZE - page_offset;
 
+		if (!pinned) {
+			wmb();
+			dev_priv->gtt.base.insert_page(&dev_priv->gtt.base,
+					i915_gem_object_get_dma_address(obj, i),
+					node.start,
+					I915_CACHE_NONE,
+					0);
+			wmb();
+			i++;
+		}
 		/* If we get a fault while copying data, then (presumably) our
 		 * source page isn't available.  Return the error and we'll
 		 * retry in the slow path.
@@ -815,13 +842,24 @@ i915_gem_gtt_pwrite_fast(struct drm_device *dev,
 
 		remain -= page_length;
 		user_data += page_length;
-		offset += page_length;
+		page_offset = 0;
 	}
 
 out_flush:
+	if (!pinned) {
+		wmb();
+		dev_priv->gtt.base.clear_range(&dev_priv->gtt.base,
+				node.start, node.size,
+				true);
+	}
+
+	i915_gem_object_unpin_pages(obj);
 	intel_fb_obj_flush(obj, false, ORIGIN_GTT);
 out_unpin:
-	i915_gem_object_ggtt_unpin(obj);
+	if (pinned)
+		i915_gem_object_ggtt_unpin(obj);
+	else
+		drm_mm_remove_node(&node);
 out:
 	return ret;
 }
-- 
1.9.1