dmaengine: sh: rz-dmac: Add system sleep PM support

[PATCH 2/4] dmaengine: sh: rz-dmac: Use devm_add_action_or_reset()

Posted by Tommaso Merciai 5 months, 1 week ago

Slightly simplify rz_dmac_probe() by using devm_add_action_or_reset()
for reset cleanup.

Signed-off-by: Tommaso Merciai <tommaso.merciai.xr@bp.renesas.com>
---
 drivers/dma/sh/rz-dmac.c | 13 +++++++++++--
 1 file changed, 11 insertions(+), 2 deletions(-)

diff --git a/drivers/dma/sh/rz-dmac.c b/drivers/dma/sh/rz-dmac.c
index 0b526cc4d24be..0bc11a6038383 100644
--- a/drivers/dma/sh/rz-dmac.c
+++ b/drivers/dma/sh/rz-dmac.c
@@ -905,6 +905,11 @@ static int rz_dmac_parse_of(struct device *dev, struct rz_dmac *dmac)
 	return rz_dmac_parse_of_icu(dev, dmac);
 }
 
+static void rz_dmac_reset_control_assert(void *data)
+{
+	reset_control_assert(data);
+}
+
 static int rz_dmac_probe(struct platform_device *pdev)
 {
 	const char *irqname = "error";
@@ -977,6 +982,12 @@ static int rz_dmac_probe(struct platform_device *pdev)
 	if (ret)
 		goto err_pm_runtime_put;
 
+	ret = devm_add_action_or_reset(&pdev->dev,
+				       rz_dmac_reset_control_assert,
+				       dmac->rstc);
+	if (ret)
+		goto err_pm_runtime_put;
+
 	for (i = 0; i < dmac->n_channels; i++) {
 		ret = rz_dmac_chan_probe(dmac, &dmac->channels[i], i);
 		if (ret < 0)
@@ -1031,7 +1042,6 @@ static int rz_dmac_probe(struct platform_device *pdev)
 				  channel->lmdesc.base_dma);
 	}
 
-	reset_control_assert(dmac->rstc);
 err_pm_runtime_put:
 	pm_runtime_put(&pdev->dev);
 
@@ -1053,7 +1063,6 @@ static void rz_dmac_remove(struct platform_device *pdev)
 				  channel->lmdesc.base,
 				  channel->lmdesc.base_dma);
 	}
-	reset_control_assert(dmac->rstc);
 	pm_runtime_put(&pdev->dev);
 
 	platform_device_put(dmac->icu.pdev);
-- 
2.43.0

Re: [PATCH 2/4] dmaengine: sh: rz-dmac: Use devm_add_action_or_reset()

Posted by Philipp Zabel 5 months, 1 week ago

On Fr, 2025-09-05 at 16:44 +0200, Tommaso Merciai wrote:
> Slightly simplify rz_dmac_probe() by using devm_add_action_or_reset()
> for reset cleanup.
> 
> Signed-off-by: Tommaso Merciai <tommaso.merciai.xr@bp.renesas.com>
> ---
>  drivers/dma/sh/rz-dmac.c | 13 +++++++++++--
>  1 file changed, 11 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/dma/sh/rz-dmac.c b/drivers/dma/sh/rz-dmac.c
> index 0b526cc4d24be..0bc11a6038383 100644
> --- a/drivers/dma/sh/rz-dmac.c
> +++ b/drivers/dma/sh/rz-dmac.c
> @@ -905,6 +905,11 @@ static int rz_dmac_parse_of(struct device *dev, struct rz_dmac *dmac)
>  	return rz_dmac_parse_of_icu(dev, dmac);
>  }
>  
> +static void rz_dmac_reset_control_assert(void *data)
> +{
> +	reset_control_assert(data);
> +}
> +
>  static int rz_dmac_probe(struct platform_device *pdev)
>  {
>  	const char *irqname = "error";
> @@ -977,6 +982,12 @@ static int rz_dmac_probe(struct platform_device *pdev)
>  	if (ret)
>  		goto err_pm_runtime_put;
>  
> +	ret = devm_add_action_or_reset(&pdev->dev,
> +				       rz_dmac_reset_control_assert,
> +				       dmac->rstc);
> +	if (ret)
> +		goto err_pm_runtime_put;
> +
>  	for (i = 0; i < dmac->n_channels; i++) {
>  		ret = rz_dmac_chan_probe(dmac, &dmac->channels[i], i);
>  		if (ret < 0)
> @@ -1031,7 +1042,6 @@ static int rz_dmac_probe(struct platform_device *pdev)
>  				  channel->lmdesc.base_dma);
>  	}
>  
> -	reset_control_assert(dmac->rstc);
>  err_pm_runtime_put:
>  	pm_runtime_put(&pdev->dev);
>  
> @@ -1053,7 +1063,6 @@ static void rz_dmac_remove(struct platform_device *pdev)
>  				  channel->lmdesc.base,
>  				  channel->lmdesc.base_dma);
>  	}
> -	reset_control_assert(dmac->rstc);

This patch changes cleanup order by effectively moving the
reset_control_assert() after pm_runtime_put(). The commit message does
not explain that this is safe to do.

If this is ok, I'd move the reset_control_assert() up before
pm_runtime_enable/resume_and_get().

regards
Philipp

Re: [PATCH 2/4] dmaengine: sh: rz-dmac: Use devm_add_action_or_reset()

Posted by Tommaso Merciai 5 months, 1 week ago

Hi Philipp,
Thank you for your review!

On Fri, Sep 05, 2025 at 04:53:54PM +0200, Philipp Zabel wrote:
> On Fr, 2025-09-05 at 16:44 +0200, Tommaso Merciai wrote:
> > Slightly simplify rz_dmac_probe() by using devm_add_action_or_reset()
> > for reset cleanup.
> > 
> > Signed-off-by: Tommaso Merciai <tommaso.merciai.xr@bp.renesas.com>
> > ---
> >  drivers/dma/sh/rz-dmac.c | 13 +++++++++++--
> >  1 file changed, 11 insertions(+), 2 deletions(-)
> > 
> > diff --git a/drivers/dma/sh/rz-dmac.c b/drivers/dma/sh/rz-dmac.c
> > index 0b526cc4d24be..0bc11a6038383 100644
> > --- a/drivers/dma/sh/rz-dmac.c
> > +++ b/drivers/dma/sh/rz-dmac.c
> > @@ -905,6 +905,11 @@ static int rz_dmac_parse_of(struct device *dev, struct rz_dmac *dmac)
> >  	return rz_dmac_parse_of_icu(dev, dmac);
> >  }
> >  
> > +static void rz_dmac_reset_control_assert(void *data)
> > +{
> > +	reset_control_assert(data);
> > +}
> > +
> >  static int rz_dmac_probe(struct platform_device *pdev)
> >  {
> >  	const char *irqname = "error";
> > @@ -977,6 +982,12 @@ static int rz_dmac_probe(struct platform_device *pdev)
> >  	if (ret)
> >  		goto err_pm_runtime_put;
> >  
> > +	ret = devm_add_action_or_reset(&pdev->dev,
> > +				       rz_dmac_reset_control_assert,
> > +				       dmac->rstc);
> > +	if (ret)
> > +		goto err_pm_runtime_put;
> > +
> >  	for (i = 0; i < dmac->n_channels; i++) {
> >  		ret = rz_dmac_chan_probe(dmac, &dmac->channels[i], i);
> >  		if (ret < 0)
> > @@ -1031,7 +1042,6 @@ static int rz_dmac_probe(struct platform_device *pdev)
> >  				  channel->lmdesc.base_dma);
> >  	}
> >  
> > -	reset_control_assert(dmac->rstc);
> >  err_pm_runtime_put:
> >  	pm_runtime_put(&pdev->dev);
> >  
> > @@ -1053,7 +1063,6 @@ static void rz_dmac_remove(struct platform_device *pdev)
> >  				  channel->lmdesc.base,
> >  				  channel->lmdesc.base_dma);
> >  	}
> > -	reset_control_assert(dmac->rstc);
> 
> This patch changes cleanup order by effectively moving the
> reset_control_assert() after pm_runtime_put(). The commit message does
> not explain that this is safe to do.

Agreed. Thanks.

> 
> If this is ok, I'd move the reset_control_assert() up before
> pm_runtime_enable/resume_and_get().

You mean having in the end the following calls:

...
	dmac->rstc = devm_reset_control_array_get_optional_exclusive(&pdev->dev);
	if (IS_ERR(dmac->rstc))
		return dev_err_probe(&pdev->dev, PTR_ERR(dmac->rstc),
				     "failed to get resets\n");

	ret = reset_control_deassert(dmac->rstc);
	if (ret)
		return dev_err_probe(&pdev->dev, ret,
				     "failed to deassert resets\n");

	ret = devm_add_action_or_reset(&pdev->dev,
				       rz_dmac_reset_control_assert,
				       dmac->rstc);
	if (ret)
		return dev_err_probe(&pdev->dev, ret,
				     "failed to register reset cleanup action\n");

	ret = devm_pm_runtime_enable(&pdev->dev);
	if (ret < 0)
		return dev_err_probe(&pdev->dev, ret,
				     "Failed to enable runtime PM\n");
...

Right?
Thanks in advance.


Kind Regards,
Tommaso

> 
> regards
> Philipp

Re: [PATCH 2/4] dmaengine: sh: rz-dmac: Use devm_add_action_or_reset()

Posted by Philipp Zabel 5 months, 1 week ago

On Fr, 2025-09-05 at 17:22 +0200, Tommaso Merciai wrote:
> Hi Philipp,
> Thank you for your review!
> 
> On Fri, Sep 05, 2025 at 04:53:54PM +0200, Philipp Zabel wrote:
> > On Fr, 2025-09-05 at 16:44 +0200, Tommaso Merciai wrote:
> > > Slightly simplify rz_dmac_probe() by using devm_add_action_or_reset()
> > > for reset cleanup.
> > > 
> > > Signed-off-by: Tommaso Merciai <tommaso.merciai.xr@bp.renesas.com>
> > > ---
> > >  drivers/dma/sh/rz-dmac.c | 13 +++++++++++--
> > >  1 file changed, 11 insertions(+), 2 deletions(-)
> > > 
> > > diff --git a/drivers/dma/sh/rz-dmac.c b/drivers/dma/sh/rz-dmac.c
> > > index 0b526cc4d24be..0bc11a6038383 100644
> > > --- a/drivers/dma/sh/rz-dmac.c
> > > +++ b/drivers/dma/sh/rz-dmac.c
> > > @@ -905,6 +905,11 @@ static int rz_dmac_parse_of(struct device *dev, struct rz_dmac *dmac)
> > >  	return rz_dmac_parse_of_icu(dev, dmac);
> > >  }
> > >  
> > > +static void rz_dmac_reset_control_assert(void *data)
> > > +{
> > > +	reset_control_assert(data);
> > > +}
> > > +
> > >  static int rz_dmac_probe(struct platform_device *pdev)
> > >  {
> > >  	const char *irqname = "error";
> > > @@ -977,6 +982,12 @@ static int rz_dmac_probe(struct platform_device *pdev)
> > >  	if (ret)
> > >  		goto err_pm_runtime_put;
> > >  
> > > +	ret = devm_add_action_or_reset(&pdev->dev,
> > > +				       rz_dmac_reset_control_assert,
> > > +				       dmac->rstc);
> > > +	if (ret)
> > > +		goto err_pm_runtime_put;
> > > +
> > >  	for (i = 0; i < dmac->n_channels; i++) {
> > >  		ret = rz_dmac_chan_probe(dmac, &dmac->channels[i], i);
> > >  		if (ret < 0)
> > > @@ -1031,7 +1042,6 @@ static int rz_dmac_probe(struct platform_device *pdev)
> > >  				  channel->lmdesc.base_dma);
> > >  	}
> > >  
> > > -	reset_control_assert(dmac->rstc);
> > >  err_pm_runtime_put:
> > >  	pm_runtime_put(&pdev->dev);
> > >  
> > > @@ -1053,7 +1063,6 @@ static void rz_dmac_remove(struct platform_device *pdev)
> > >  				  channel->lmdesc.base,
> > >  				  channel->lmdesc.base_dma);
> > >  	}
> > > -	reset_control_assert(dmac->rstc);
> > 
> > This patch changes cleanup order by effectively moving the
> > reset_control_assert() after pm_runtime_put(). The commit message does
> > not explain that this is safe to do.
> 
> Agreed. Thanks.
> 
> > 
> > If this is ok, I'd move the reset_control_assert() up before
> > pm_runtime_enable/resume_and_get().
> 
> You mean having in the end the following calls:
> 
> ...
> 	dmac->rstc = devm_reset_control_array_get_optional_exclusive(&pdev->dev);
> 	if (IS_ERR(dmac->rstc))
> 		return dev_err_probe(&pdev->dev, PTR_ERR(dmac->rstc),
> 				     "failed to get resets\n");
> 
> 	ret = reset_control_deassert(dmac->rstc);
> 	if (ret)
> 		return dev_err_probe(&pdev->dev, ret,
> 				     "failed to deassert resets\n");
> 
> 	ret = devm_add_action_or_reset(&pdev->dev,
> 				       rz_dmac_reset_control_assert,
> 				       dmac->rstc);
> 	if (ret)
> 		return dev_err_probe(&pdev->dev, ret,
> 				     "failed to register reset cleanup action\n");
> 
> 	ret = devm_pm_runtime_enable(&pdev->dev);
> 	if (ret < 0)
> 		return dev_err_probe(&pdev->dev, ret,
> 				     "Failed to enable runtime PM\n");
> ...
> 
> Right?

Right.

regards
Philipp