[PATCH] vmwgfx: Avoid fb VRAM fragmenting.

[Thomas Hellstrom]

If the soon-to-be scanout buffer is partly covering the intended
VRAM region, move and pin will fail. In that case, just move it out
to system before attempting to move it in again.

Signed-off-by: Thomas Hellstrom <thellstrom at vmware.com>
---
 drivers/gpu/drm/vmwgfx/vmwgfx_fb.c |    5 +++++
 1 files changed, 5 insertions(+), 0 deletions(-)

diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_fb.c b/drivers/gpu/drm/vmwgfx/vmwgfx_fb.c
index cac0db0..b27a9f2 100644
--- a/drivers/gpu/drm/vmwgfx/vmwgfx_fb.c
+++ b/drivers/gpu/drm/vmwgfx/vmwgfx_fb.c
@@ -625,6 +625,11 @@ int vmw_dmabuf_to_start_of_vram(struct vmw_private *vmw_priv,
 	if (unlikely(ret != 0))
 		goto err_unlock;
 
+	if (bo->mem.mem_type == TTM_PL_VRAM &&
+	    bo->mem.mm_node->start < bo->num_pages)
+		(void) ttm_bo_validate(bo, &vmw_sys_placement, false,
+				       false, false);
+
 	ret = ttm_bo_validate(bo, &ne_placement, false, false, false);
 
 	/* Could probably bug on */
-- 
1.6.2.5