[PATCH] i2c: designware: amdisp: balance genpd on probe failure

[PATCH] i2c: designware: amdisp: balance genpd on probe failure

[Guangshuo Li]

The change referenced by the Fixes tag avoids a runtime PM resume during
probe by powering the ISP domain directly with dev_pm_genpd_resume()
before calling i2c_dw_probe(). Runtime PM is enabled only after
i2c_dw_probe() has succeeded.

However, the i2c_dw_probe() failure path still uses the old runtime PM
cleanup path. At that point pm_runtime_enable() has not been called yet,
so pm_runtime_disable() is not paired with any enable. The failure path
also skips the dev_pm_genpd_suspend() needed to balance the direct
dev_pm_genpd_resume() call.

Suspend the genpd on the i2c_dw_probe() failure path instead of disabling
runtime PM. This keeps the direct genpd resume/suspend pairing symmetric
on both success and failure paths.

Fixes: e2f1ada8e089 ("i2c: designware: amdisp: Fix resume-probe race condition issue")
Signed-off-by: Guangshuo Li <lgs201920130244@gmail.com>
---
 drivers/i2c/busses/i2c-designware-amdisp.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/i2c/busses/i2c-designware-amdisp.c b/drivers/i2c/busses/i2c-designware-amdisp.c
index 9f0ec0fae6f2..724cb05e21c8 100644
--- a/drivers/i2c/busses/i2c-designware-amdisp.c
+++ b/drivers/i2c/busses/i2c-designware-amdisp.c
@@ -81,7 +81,7 @@ static int amd_isp_dw_i2c_plat_probe(struct platform_device *pdev)
 	ret = i2c_dw_probe(isp_i2c_dev);
 	if (ret) {
 		dev_err_probe(&pdev->dev, ret, "i2c_dw_probe failed\n");
-		goto error_release_rpm;
+		goto error_suspend_genpd;
 	}
 	dev_pm_genpd_suspend(&pdev->dev);
 	pm_runtime_set_suspended(&pdev->dev);
@@ -89,8 +89,8 @@ static int amd_isp_dw_i2c_plat_probe(struct platform_device *pdev)
 
 	return 0;
 
-error_release_rpm:
-	amd_isp_dw_i2c_plat_pm_cleanup(isp_i2c_dev);
+error_suspend_genpd:
+	dev_pm_genpd_suspend(&pdev->dev);
 	return ret;
 }
 
-- 
2.43.0

Re: [PATCH] i2c: designware: amdisp: balance genpd on probe failure

[Andy Shevchenko]

On Wed, Jul 08, 2026 at 05:05:52PM +0800, Guangshuo Li wrote:
> The change referenced by the Fixes tag avoids a runtime PM resume during
> probe by powering the ISP domain directly with dev_pm_genpd_resume()
> before calling i2c_dw_probe(). Runtime PM is enabled only after
> i2c_dw_probe() has succeeded.
> 
> However, the i2c_dw_probe() failure path still uses the old runtime PM
> cleanup path. At that point pm_runtime_enable() has not been called yet,
> so pm_runtime_disable() is not paired with any enable. The failure path
> also skips the dev_pm_genpd_suspend() needed to balance the direct
> dev_pm_genpd_resume() call.
> 
> Suspend the genpd on the i2c_dw_probe() failure path instead of disabling
> runtime PM. This keeps the direct genpd resume/suspend pairing symmetric
> on both success and failure paths.

Ideally this needs AMD people to Ack.

-- 
With Best Regards,
Andy Shevchenko

Re: [PATCH] i2c: designware: amdisp: balance genpd on probe failure

[Nirujogi, Pratap]

On 7/8/2026 5:05 AM, Guangshuo Li wrote:
> [You don't often get email from lgs201920130244@gmail.com. Learn why this is important at https://aka.ms/LearnAboutSenderIdentification ]
> 
> Caution: This message originated from an External Source. Use proper caution when opening attachments, clicking links, or responding.
> 
> 
> The change referenced by the Fixes tag avoids a runtime PM resume during
> probe by powering the ISP domain directly with dev_pm_genpd_resume()
> before calling i2c_dw_probe(). Runtime PM is enabled only after
> i2c_dw_probe() has succeeded.
> 
> However, the i2c_dw_probe() failure path still uses the old runtime PM
> cleanup path. At that point pm_runtime_enable() has not been called yet,
> so pm_runtime_disable() is not paired with any enable. The failure path
> also skips the dev_pm_genpd_suspend() needed to balance the direct
> dev_pm_genpd_resume() call.
> 
> Suspend the genpd on the i2c_dw_probe() failure path instead of disabling
> runtime PM. This keeps the direct genpd resume/suspend pairing symmetric
> on both success and failure paths.
> 
> Fixes: e2f1ada8e089 ("i2c: designware: amdisp: Fix resume-probe race condition issue")
> Signed-off-by: Guangshuo Li <lgs201920130244@gmail.com>
> ---
>   drivers/i2c/busses/i2c-designware-amdisp.c | 6 +++---
>   1 file changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/i2c/busses/i2c-designware-amdisp.c b/drivers/i2c/busses/i2c-designware-amdisp.c
> index 9f0ec0fae6f2..724cb05e21c8 100644
> --- a/drivers/i2c/busses/i2c-designware-amdisp.c
> +++ b/drivers/i2c/busses/i2c-designware-amdisp.c
> @@ -81,7 +81,7 @@ static int amd_isp_dw_i2c_plat_probe(struct platform_device *pdev)
>          ret = i2c_dw_probe(isp_i2c_dev);
>          if (ret) {
>                  dev_err_probe(&pdev->dev, ret, "i2c_dw_probe failed\n");
> -               goto error_release_rpm;
> +               goto error_suspend_genpd;
>          }
>          dev_pm_genpd_suspend(&pdev->dev);
>          pm_runtime_set_suspended(&pdev->dev);
> @@ -89,8 +89,8 @@ static int amd_isp_dw_i2c_plat_probe(struct platform_device *pdev)
> 
>          return 0;
> 
> -error_release_rpm:
> -       amd_isp_dw_i2c_plat_pm_cleanup(isp_i2c_dev);
> +error_suspend_genpd:
> +       dev_pm_genpd_suspend(&pdev->dev);
>          return ret;
>   }
> 
Good catch! I overlooked that earlier. Thanks.

Acked-by: Pratap Nirujogi <pratap.nirujogi@amd.com>

> --
> 2.43.0
> 

Re: [PATCH] i2c: designware: amdisp: balance genpd on probe failure

[Mukesh Savaliya]

Looks fine to me, Have someone from AMD also Ack'ing this as requested 
by Andy Shevchenko.

On 7/8/2026 2:35 PM, Guangshuo Li wrote:
> The change referenced by the Fixes tag avoids a runtime PM resume during
> probe by powering the ISP domain directly with dev_pm_genpd_resume()
> before calling i2c_dw_probe(). Runtime PM is enabled only after
> i2c_dw_probe() has succeeded.
> 
> However, the i2c_dw_probe() failure path still uses the old runtime PM
> cleanup path. At that point pm_runtime_enable() has not been called yet,
> so pm_runtime_disable() is not paired with any enable. The failure path
> also skips the dev_pm_genpd_suspend() needed to balance the direct
> dev_pm_genpd_resume() call.
> 
> Suspend the genpd on the i2c_dw_probe() failure path instead of disabling
> runtime PM. This keeps the direct genpd resume/suspend pairing symmetric
> on both success and failure paths.
> 
> Fixes: e2f1ada8e089 ("i2c: designware: amdisp: Fix resume-probe race condition issue")
> Signed-off-by: Guangshuo Li <lgs201920130244@gmail.com>
> ---
>   drivers/i2c/busses/i2c-designware-amdisp.c | 6 +++---
>   1 file changed, 3 insertions(+), 3 deletions(-)

Acked-by: Mukesh Kumar Savaliya <mukesh.savaliya@oss.qualcomm.com>

Re: [PATCH] i2c: designware: amdisp: balance genpd on probe failure

[Bin Du]

On 7/9/2026 2:04 PM, Mukesh Savaliya wrote:
> Looks fine to me, Have someone from AMD also Ack'ing this as requested
> by Andy Shevchenko.
> 
> On 7/8/2026 2:35 PM, Guangshuo Li wrote:
>> The change referenced by the Fixes tag avoids a runtime PM resume during
>> probe by powering the ISP domain directly with dev_pm_genpd_resume()
>> before calling i2c_dw_probe(). Runtime PM is enabled only after
>> i2c_dw_probe() has succeeded.
>>
>> However, the i2c_dw_probe() failure path still uses the old runtime PM
>> cleanup path. At that point pm_runtime_enable() has not been called yet,
>> so pm_runtime_disable() is not paired with any enable. The failure path
>> also skips the dev_pm_genpd_suspend() needed to balance the direct
>> dev_pm_genpd_resume() call.
>>
>> Suspend the genpd on the i2c_dw_probe() failure path instead of disabling
>> runtime PM. This keeps the direct genpd resume/suspend pairing symmetric
>> on both success and failure paths.
>>
>> Fixes: e2f1ada8e089 ("i2c: designware: amdisp: Fix resume-probe race
>> condition issue")
>> Signed-off-by: Guangshuo Li <lgs201920130244@gmail.com>
>> ---
>>   drivers/i2c/busses/i2c-designware-amdisp.c | 6 +++---
>>   1 file changed, 3 insertions(+), 3 deletions(-)
> 
> Acked-by: Mukesh Kumar Savaliya <mukesh.savaliya@oss.qualcomm.com>
> 

Acked-by: Bin Du <bin.du@amd.com>
Tested-by: Bin Du <bin.du@amd.com>