On 2020-09-07 10:04, Rob Clark wrote:
From: Rob Clark <robdclark@xxxxxxxxxxxx>
We could get a vblank event racing with the current atomic commit,
resulting in sending the pageflip event to userspace early, causing
tearing. On the other hand, complete_commit() ensures that the
pending flush is complete.
Signed-off-by: Rob Clark <robdclark@xxxxxxxxxxxx>
I checked our downstream code as well and yes we are not signaling
page flips inside the vblank_cb and are doing it after
wait_for_commit_done
This aligns with that.
Hence,
Reviewed-by: Abhinav Kumar <abhinavk@xxxxxxxxxxxxxx>
---
drivers/gpu/drm/msm/disp/dpu1/dpu_crtc.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_crtc.c
b/drivers/gpu/drm/msm/disp/dpu1/dpu_crtc.c
index c2729f71e2fa..89c0245b5de5 100644
--- a/drivers/gpu/drm/msm/disp/dpu1/dpu_crtc.c
+++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_crtc.c
@@ -297,7 +297,6 @@ void dpu_crtc_vblank_callback(struct drm_crtc
*crtc)
dpu_crtc->vblank_cb_time = ktime_get();
else
dpu_crtc->vblank_cb_count++;
- _dpu_crtc_complete_flip(crtc);
drm_crtc_handle_vblank(crtc);
trace_dpu_crtc_vblank_cb(DRMID(crtc));
}
@@ -402,6 +401,7 @@ static void dpu_crtc_frame_event_cb(void *data, u32
event)
void dpu_crtc_complete_commit(struct drm_crtc *crtc)
{
trace_dpu_crtc_complete_commit(DRMID(crtc));
+ _dpu_crtc_complete_flip(crtc);
}
static void _dpu_crtc_setup_lm_bounds(struct drm_crtc *crtc,
