[Intel-xe] [PATCH] drm/xe/display: Append dma_buf fences to framebuffer fences

[José Roberto de Souza]

This is required to comply with KMS page flip requirements, and it is
the only way to synchronize framebuffer scanouts in Xe KMD.

UMDs are expected to attach a syncobj that will be signaled at the end
of the GPU job that is rendering in the framebuffer.
This way, the compositor can send jobs to the GPU to render the
framebuffer and without having to wait for it to complete, queue the
page flip in KMS.

This fixes display glitches.

Link: https://gitlab.freedesktop.org/mesa/mesa/-/issues/8923
Signed-off-by: José Roberto de Souza <jose.souza at intel.com>
---
 drivers/gpu/drm/i915/display/intel_atomic_plane.c | 13 ++++++++++++-
 1 file changed, 12 insertions(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/i915/display/intel_atomic_plane.c b/drivers/gpu/drm/i915/display/intel_atomic_plane.c
index b756561f08bd1..a3f30a319e4c0 100644
--- a/drivers/gpu/drm/i915/display/intel_atomic_plane.c
+++ b/drivers/gpu/drm/i915/display/intel_atomic_plane.c
@@ -32,6 +32,7 @@
  */
 
 #include <drm/drm_atomic_helper.h>
+#include <drm/drm_gem_atomic_helper.h>
 #include <drm/drm_blend.h>
 #include <drm/drm_fourcc.h>
 
@@ -1120,10 +1121,20 @@ intel_prepare_plane_fb(struct drm_plane *_plane,
 
 	return ret;
 #else
+	int ret;
+
 	if (!intel_fb_obj(new_plane_state->hw.fb))
 		return 0;
 
-	return intel_plane_pin_fb(new_plane_state);
+	ret = intel_plane_pin_fb(new_plane_state);
+	if (ret)
+		return ret;
+
+	ret = drm_gem_plane_helper_prepare_fb(_plane, _new_plane_state);
+	if (ret)
+		intel_plane_unpin_fb(new_plane_state);
+
+	return ret;
 #endif
 }
 
-- 
2.41.0