drm/msm: dpu: Don't set frame_busy_mask for async updates

Submitted by Sean Paul on Jan. 30, 2019, 4:32 p.m.

Details

Message ID 20190130163220.138637-1-sean@poorly.run
State New
Series "drm/msm: dpu: Don't set frame_busy_mask for async updates"
Headers show

Commit Message

Sean Paul Jan. 30, 2019, 4:32 p.m.
From: Sean Paul <seanpaul@chromium.org>

The frame_busy mask is used in frame_done event handling, which is not
invoked for async commits. So an async commit will leave the
frame_busy mask populated after it completes and future commits will start
with the busy mask incorrect.

This showed up on disable after cursor move. I was hitting the "this should
not happen" comment in the frame event worker since frame_busy was set,
we queued the event, but there were no frames pending (since async
also doesn't set that).

Signed-off-by: Sean Paul <seanpaul@chromium.org>
---
 drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

Patch hide | download patch | download mbox

diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c
index 36158b7d99cd..1a81c4daabc9 100644
--- a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c
+++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c
@@ -1558,8 +1558,14 @@  static void _dpu_encoder_kickoff_phys(struct dpu_encoder_virt *dpu_enc,
 		if (!ctl)
 			continue;
 
-		if (phys->split_role != ENC_ROLE_SLAVE)
+		/*
+		 * This is cleared in frame_done worker, which isn't invoked
+		 * for async commits. So don't set this for async, since it'll
+		 * roll over to the next commit.
+		 */
+		if (!async && phys->split_role != ENC_ROLE_SLAVE)
 			set_bit(i, dpu_enc->frame_busy_mask);
+
 		if (!phys->ops.needs_single_flush ||
 				!phys->ops.needs_single_flush(phys))
 			_dpu_encoder_trigger_flush(&dpu_enc->base, phys, 0x0,

Comments

Fritz Koenig Feb. 5, 2019, 8:11 p.m.
On Wed, Jan 30, 2019 at 8:32 AM Sean Paul <sean@poorly.run> wrote:
>
> From: Sean Paul <seanpaul@chromium.org>
>
> The frame_busy mask is used in frame_done event handling, which is not
> invoked for async commits. So an async commit will leave the
> frame_busy mask populated after it completes and future commits will start
> with the busy mask incorrect.
>
> This showed up on disable after cursor move. I was hitting the "this should
> not happen" comment in the frame event worker since frame_busy was set,
> we queued the event, but there were no frames pending (since async
> also doesn't set that).
>
> Signed-off-by: Sean Paul <seanpaul@chromium.org>
> ---
>  drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c | 8 +++++++-
>  1 file changed, 7 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c
> index 36158b7d99cd..1a81c4daabc9 100644
> --- a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c
> +++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c
> @@ -1558,8 +1558,14 @@ static void _dpu_encoder_kickoff_phys(struct dpu_encoder_virt *dpu_enc,
>                 if (!ctl)
>                         continue;
>
> -               if (phys->split_role != ENC_ROLE_SLAVE)
> +               /*
> +                * This is cleared in frame_done worker, which isn't invoked
> +                * for async commits. So don't set this for async, since it'll
> +                * roll over to the next commit.
> +                */
> +               if (!async && phys->split_role != ENC_ROLE_SLAVE)
>                         set_bit(i, dpu_enc->frame_busy_mask);
> +
>                 if (!phys->ops.needs_single_flush ||
>                                 !phys->ops.needs_single_flush(phys))
>                         _dpu_encoder_trigger_flush(&dpu_enc->base, phys, 0x0,
> --
> Sean Paul, Software Engineer, Google / Chromium OS
>

Reviewed-by: Fritz Koenig <frkoenig@google.com>

> _______________________________________________
> dri-devel mailing list
> dri-devel@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/dri-devel