On 11/27/2023 3:17 PM, Dmitry Baryshkov wrote:
On Tue, 28 Nov 2023 at 00:00, Abhinav Kumar <quic_abhinavk@xxxxxxxxxxx> wrote:
On 11/27/2023 1:54 PM, Dmitry Baryshkov wrote:
Flush queued events when disabling the crtc. This avoids timeouts when
we come back and wait for dependencies (like the previous frame's
flip_done).
Fixes: c8afe684c95c ("drm/msm: basic KMS driver for snapdragon")
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@xxxxxxxxxx>
---
drivers/gpu/drm/msm/disp/mdp4/mdp4_crtc.c | 9 +++++++++
1 file changed, 9 insertions(+)
diff --git a/drivers/gpu/drm/msm/disp/mdp4/mdp4_crtc.c b/drivers/gpu/drm/msm/disp/mdp4/mdp4_crtc.c
index 169f9de4a12a..3100957225a7 100644
--- a/drivers/gpu/drm/msm/disp/mdp4/mdp4_crtc.c
+++ b/drivers/gpu/drm/msm/disp/mdp4/mdp4_crtc.c
@@ -269,6 +269,7 @@ static void mdp4_crtc_atomic_disable(struct drm_crtc *crtc,
{
struct mdp4_crtc *mdp4_crtc = to_mdp4_crtc(crtc);
struct mdp4_kms *mdp4_kms = get_kms(crtc);
+ unsigned long flags;
DBG("%s", mdp4_crtc->name);
@@ -281,6 +282,14 @@ static void mdp4_crtc_atomic_disable(struct drm_crtc *crtc,
mdp_irq_unregister(&mdp4_kms->base, &mdp4_crtc->err);
mdp4_disable(mdp4_kms);
+ if (crtc->state->event && !crtc->state->active) {
+ WARN_ON(mdp4_crtc->event);
Do you need a WARN_ON() here? Just wondering how often this might happen
and spam.
Well, it is atomic_disable(), so it happens only on modesets or output
disable. Also, this is a port of mdp5 code, see commit e765ea77b0fd
("drm/msm: Issue queued events when disabling crtc"). We don't have
reports from MDP5 code, so there should be no comparable reports for
MDP4.
Ok, I thought this was ported from dpu code, even that has the same
logic but not the WARN_ON().
But I am fine with this,
Reviewed-by: Abhinav Kumar <quic_abhinavk@xxxxxxxxxxx>
otherwise LGTM.
+ spin_lock_irqsave(&mdp4_kms->dev->event_lock, flags);
+ drm_crtc_send_vblank_event(crtc, crtc->state->event);
+ crtc->state->event = NULL;
+ spin_unlock_irqrestore(&mdp4_kms->dev->event_lock, flags);
+ }
+
mdp4_crtc->enabled = false;
}
