On 18 December 2025 15:38:24 CET, Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com> wrote:
>DPU units before 4.x don't have a separate CTL_START IRQ to mark the
>begin of the data transfer. In such a case, wait for the
Unfinished sentence. And would it make sense to note this down in a code-comment as well?
E.g.
...wait for the "transfer complete" IRQ if there's no CTL_START IRQ (or the ctl is already started).
Is there a significant timing difference on the two irqs, or does it not matter?
- Marijn
>
>Fixes: 050770cbbd26 ("drm/msm/dpu: Fix timeout issues on command mode panels")
>Reported-by: Alexey Minnekhanov <alexeymin@postmarketos.org>
>Closes: https://lore.kernel.org/r/8e1d33ff-d902-4ae9-9162-e00d17a5e6d1@postmarketos.org
>Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
>---
> drivers/gpu/drm/msm/disp/dpu1/dpu_encoder_phys_cmd.c | 3 ++-
> 1 file changed, 2 insertions(+), 1 deletion(-)
>
>diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder_phys_cmd.c b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder_phys_cmd.c
>index 0ec6d67c7c70..e6f55902e355 100644
>--- a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder_phys_cmd.c
>+++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder_phys_cmd.c
>@@ -681,7 +681,8 @@ static int dpu_encoder_phys_cmd_wait_for_commit_done(
> if (!dpu_encoder_phys_cmd_is_master(phys_enc))
> return 0;
>
>- if (phys_enc->hw_ctl->ops.is_started(phys_enc->hw_ctl))
>+ if (phys_enc->hw_ctl->ops.is_started(phys_enc->hw_ctl) ||
>+ !phys_enc->irq[INTR_IDX_CTL_START])
> return dpu_encoder_phys_cmd_wait_for_tx_complete(phys_enc);
>
> return _dpu_encoder_phys_cmd_wait_for_ctl_start(phys_enc);
>