drm/radeon: rework page flip handling v4

Instead of trying to flip inside the vblank period when
the buffer is idle, offload blocking for idle to a kernel
thread and program the flip directly into the hardware.

v2: add error handling, fix EBUSY handling
v3: add proper exclusive_lock handling
v4: update crtc->primary->fb when the flip actually happens
Signed-off-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>

drivers/gpu/drm/radeon/radeon.h

@@ -676,14 +676,16 @@ void radeon_doorbell_free(struct radeon_device *rdev, u32 doorbell);
  * IRQS.
  */
 
-struct radeon_unpin_work {
-	struct work_struct work;
+struct radeon_flip_work {
+	struct work_struct		flip_work;
+	struct work_struct		unpin_work;
 	struct radeon_device		*rdev;
 	int				crtc_id;
-	struct radeon_fence *fence;
+	struct drm_framebuffer		*fb;
 	struct drm_pending_vblank_event *event;
 	struct radeon_bo		*old_rbo;
-	u64 new_crtc_base;
+	struct radeon_bo		*new_rbo;
+	struct radeon_fence		*fence;
 };
 
 struct r500_irq_stat_regs {

drivers/gpu/drm/radeon/radeon_mode.h

@@ -325,8 +325,8 @@ struct radeon_crtc {
 	struct drm_display_mode native_mode;
 	int pll_id;
 	/* page flipping */
-	struct radeon_unpin_work *unpin_work;
-	int deferred_flip_completion;
+	struct workqueue_struct *flip_queue;
+	struct radeon_flip_work *flip_work;
 	/* pll sharing */
 	struct radeon_atom_ss ss;
 	bool ss_enabled;