[PATCH v2] drm/msm/dpu: handle pm_runtime_get_sync() errors in bind path

[PATCH v2] drm/msm/dpu: handle pm_runtime_get_sync() errors in bind path

[Abhinav Kumar]

If there are errors while trying to enable the pm in the
bind path, it will lead to unclocked access of hw revision
register thereby crashing the device.

This will not address why the pm_runtime_get_sync() fails
but at the very least we should be able to prevent the
crash by handling the error and bailing out earlier.

changes in v2:
	- use pm_runtime_resume_and_get() instead of
	  pm_runtime_get_sync()

Signed-off-by: Abhinav Kumar <quic_abhinavk@quicinc.com>
---
 drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c b/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
index 2b9d931474e0..bce47647d891 100644
--- a/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
+++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
@@ -1089,7 +1089,9 @@ static int dpu_kms_hw_init(struct msm_kms *kms)
 
 	dpu_kms_parse_data_bus_icc_path(dpu_kms);
 
-	pm_runtime_get_sync(&dpu_kms->pdev->dev);
+	rc = pm_runtime_resume_and_get(&dpu_kms->pdev->dev);
+	if (rc < 0)
+		goto error;
 
 	dpu_kms->core_rev = readl_relaxed(dpu_kms->mmio + 0x0);
 
-- 
2.35.1

Re: [PATCH v2] drm/msm/dpu: handle pm_runtime_get_sync() errors in bind path

[Rob Clark]

On Wed, May 18, 2022 at 3:34 PM Abhinav Kumar <quic_abhinavk@quicinc.com> wrote:
>
> If there are errors while trying to enable the pm in the
> bind path, it will lead to unclocked access of hw revision
> register thereby crashing the device.
>
> This will not address why the pm_runtime_get_sync() fails
> but at the very least we should be able to prevent the
> crash by handling the error and bailing out earlier.
>
> changes in v2:
>         - use pm_runtime_resume_and_get() instead of
>           pm_runtime_get_sync()
>
> Signed-off-by: Abhinav Kumar <quic_abhinavk@quicinc.com>

Reviewed-by: Rob Clark <robdclark@gmail.com>

> ---
>  drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c | 4 +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c b/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
> index 2b9d931474e0..bce47647d891 100644
> --- a/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
> +++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c
> @@ -1089,7 +1089,9 @@ static int dpu_kms_hw_init(struct msm_kms *kms)
>
>         dpu_kms_parse_data_bus_icc_path(dpu_kms);
>
> -       pm_runtime_get_sync(&dpu_kms->pdev->dev);
> +       rc = pm_runtime_resume_and_get(&dpu_kms->pdev->dev);
> +       if (rc < 0)
> +               goto error;
>
>         dpu_kms->core_rev = readl_relaxed(dpu_kms->mmio + 0x0);
>
> --
> 2.35.1
>

Re: [PATCH v2] drm/msm/dpu: handle pm_runtime_get_sync() errors in bind path

[Stephen Boyd]

Quoting Abhinav Kumar (2022-05-18 15:34:07)
> If there are errors while trying to enable the pm in the
> bind path, it will lead to unclocked access of hw revision
> register thereby crashing the device.
>
> This will not address why the pm_runtime_get_sync() fails
> but at the very least we should be able to prevent the
> crash by handling the error and bailing out earlier.
>
> changes in v2:
>         - use pm_runtime_resume_and_get() instead of
>           pm_runtime_get_sync()
>
> Signed-off-by: Abhinav Kumar <quic_abhinavk@quicinc.com>
> ---

Any Fixes tag? When did pm errors start happening in the bind path?

Reviewed-by: Stephen Boyd <swboyd@chromium.org>

Re: [Freedreno] [PATCH v2] drm/msm/dpu: handle pm_runtime_get_sync() errors in bind path

[Abhinav Kumar]

On 5/18/2022 5:40 PM, Stephen Boyd wrote:
> Quoting Abhinav Kumar (2022-05-18 15:34:07)
>> If there are errors while trying to enable the pm in the
>> bind path, it will lead to unclocked access of hw revision
>> register thereby crashing the device.
>>
>> This will not address why the pm_runtime_get_sync() fails
>> but at the very least we should be able to prevent the
>> crash by handling the error and bailing out earlier.
>>
>> changes in v2:
>>          - use pm_runtime_resume_and_get() instead of
>>            pm_runtime_get_sync()
>>
>> Signed-off-by: Abhinav Kumar <quic_abhinavk@quicinc.com>
>> ---
> 
> Any Fixes tag? When did pm errors start happening in the bind path?
> 
> Reviewed-by: Stephen Boyd <swboyd@chromium.org>

This error got exposed with PANEL_EDP=m and DRM_MSM=y. We were not 
testing this combination previously. This combination causes "clk stuck 
at OFF" from the pm_runtime_get_sync() path which means we shouldnt 
proceed with the next register access since it failed.

We are still debugging the root-cause of why "clk stuck at OFF" error is 
present, this is just resolving the crash.

Fixes : 25fdd5933e4c ("drm/msm: Add SDM845 DPU support")