[PATCH V2] dmaengine: tegra-apb: Really fix runtime-pm usage

[PATCH V2] dmaengine: tegra-apb: Really fix runtime-pm usage

[Jon Hunter]

Commit edd3bdbe9db1 ("dmaengine: tegra-apb: Correct runtime-pm usage")
added pm_runtime_get/put() calls to the tegra-apb DMA system suspend
callbacks. Runtime PM is disabled during system suspend and so these
APIs cannot be used. Fix the suspend handling for the tegra-apb DMA by
moving the save and restore of the DMA register context into the
runtime PM suspend and resume callbacks, and then use the
pm_runtime_force_suspend/resume() APIs to invoke the runtime PM
callbacks during system suspend.

Fixes: edd3bdbe9db1 ("dmaengine: tegra-apb: Correct runtime-pm usage")

Signed-off-by: Jon Hunter <jonathanh-DDmLM1+adcrQT0dZR+AlfA@public.gmane.org>
---

Changes since V1:
- Drop the custom suspend/resume callbacks and use
  pm_runtime_force_suspend/resume directly in SET_SYSTEM_SLEEP_PM_OPS.

 drivers/dma/tegra20-apb-dma.c | 50 +++++++++----------------------------------
 1 file changed, 10 insertions(+), 40 deletions(-)

diff --git a/drivers/dma/tegra20-apb-dma.c b/drivers/dma/tegra20-apb-dma.c
index 3722b9d8d9fe..b9d75a54c896 100644
--- a/drivers/dma/tegra20-apb-dma.c
+++ b/drivers/dma/tegra20-apb-dma.c
@@ -1494,35 +1494,7 @@ static int tegra_dma_remove(struct platform_device *pdev)
 static int tegra_dma_runtime_suspend(struct device *dev)
 {
 	struct tegra_dma *tdma = dev_get_drvdata(dev);
-
-	clk_disable_unprepare(tdma->dma_clk);
-	return 0;
-}
-
-static int tegra_dma_runtime_resume(struct device *dev)
-{
-	struct tegra_dma *tdma = dev_get_drvdata(dev);
-	int ret;
-
-	ret = clk_prepare_enable(tdma->dma_clk);
-	if (ret < 0) {
-		dev_err(dev, "clk_enable failed: %d\n", ret);
-		return ret;
-	}
-	return 0;
-}
-
-#ifdef CONFIG_PM_SLEEP
-static int tegra_dma_pm_suspend(struct device *dev)
-{
-	struct tegra_dma *tdma = dev_get_drvdata(dev);
 	int i;
-	int ret;
-
-	/* Enable clock before accessing register */
-	ret = pm_runtime_get_sync(dev);
-	if (ret < 0)
-		return ret;
 
 	tdma->reg_gen = tdma_read(tdma, TEGRA_APBDMA_GENERAL);
 	for (i = 0; i < tdma->chip_data->nr_channels; i++) {
@@ -1543,21 +1515,21 @@ static int tegra_dma_pm_suspend(struct device *dev)
 						  TEGRA_APBDMA_CHAN_WCOUNT);
 	}
 
-	/* Disable clock */
-	pm_runtime_put(dev);
+	clk_disable_unprepare(tdma->dma_clk);
+
 	return 0;
 }
 
-static int tegra_dma_pm_resume(struct device *dev)
+static int tegra_dma_runtime_resume(struct device *dev)
 {
 	struct tegra_dma *tdma = dev_get_drvdata(dev);
-	int i;
-	int ret;
+	int i, ret;
 
-	/* Enable clock before accessing register */
-	ret = pm_runtime_get_sync(dev);
-	if (ret < 0)
+	ret = clk_prepare_enable(tdma->dma_clk);
+	if (ret < 0) {
+		dev_err(dev, "clk_enable failed: %d\n", ret);
 		return ret;
+	}
 
 	tdma_write(tdma, TEGRA_APBDMA_GENERAL, tdma->reg_gen);
 	tdma_write(tdma, TEGRA_APBDMA_CONTROL, 0);
@@ -1582,16 +1554,14 @@ static int tegra_dma_pm_resume(struct device *dev)
 			(ch_reg->csr & ~TEGRA_APBDMA_CSR_ENB));
 	}
 
-	/* Disable clock */
-	pm_runtime_put(dev);
 	return 0;
 }
-#endif
 
 static const struct dev_pm_ops tegra_dma_dev_pm_ops = {
 	SET_RUNTIME_PM_OPS(tegra_dma_runtime_suspend, tegra_dma_runtime_resume,
 			   NULL)
-	SET_SYSTEM_SLEEP_PM_OPS(tegra_dma_pm_suspend, tegra_dma_pm_resume)
+	SET_SYSTEM_SLEEP_PM_OPS(pm_runtime_force_suspend,
+				pm_runtime_force_resume)
 };
 
 static const struct of_device_id tegra_dma_of_match[] = {
-- 
2.7.4

Re: [PATCH V2] dmaengine: tegra-apb: Really fix runtime-pm usage

[Thierry Reding]

[-- Attachment #1: Type: text/plain, Size: 3821 bytes --]

On Tue, Jun 06, 2017 at 01:49:29PM +0100, Jon Hunter wrote:
> Commit edd3bdbe9db1 ("dmaengine: tegra-apb: Correct runtime-pm usage")
> added pm_runtime_get/put() calls to the tegra-apb DMA system suspend
> callbacks. Runtime PM is disabled during system suspend and so these
> APIs cannot be used. Fix the suspend handling for the tegra-apb DMA by
> moving the save and restore of the DMA register context into the
> runtime PM suspend and resume callbacks, and then use the
> pm_runtime_force_suspend/resume() APIs to invoke the runtime PM
> callbacks during system suspend.
> 
> Fixes: edd3bdbe9db1 ("dmaengine: tegra-apb: Correct runtime-pm usage")
> 
> Signed-off-by: Jon Hunter <jonathanh-DDmLM1+adcrQT0dZR+AlfA@public.gmane.org>
> ---
> 
> Changes since V1:
> - Drop the custom suspend/resume callbacks and use
>   pm_runtime_force_suspend/resume directly in SET_SYSTEM_SLEEP_PM_OPS.
> 
>  drivers/dma/tegra20-apb-dma.c | 50 +++++++++----------------------------------
>  1 file changed, 10 insertions(+), 40 deletions(-)
> 
> diff --git a/drivers/dma/tegra20-apb-dma.c b/drivers/dma/tegra20-apb-dma.c
> index 3722b9d8d9fe..b9d75a54c896 100644
> --- a/drivers/dma/tegra20-apb-dma.c
> +++ b/drivers/dma/tegra20-apb-dma.c
> @@ -1494,35 +1494,7 @@ static int tegra_dma_remove(struct platform_device *pdev)
>  static int tegra_dma_runtime_suspend(struct device *dev)
>  {
>  	struct tegra_dma *tdma = dev_get_drvdata(dev);
> -
> -	clk_disable_unprepare(tdma->dma_clk);
> -	return 0;
> -}
> -
> -static int tegra_dma_runtime_resume(struct device *dev)
> -{
> -	struct tegra_dma *tdma = dev_get_drvdata(dev);
> -	int ret;
> -
> -	ret = clk_prepare_enable(tdma->dma_clk);
> -	if (ret < 0) {
> -		dev_err(dev, "clk_enable failed: %d\n", ret);
> -		return ret;
> -	}
> -	return 0;
> -}
> -
> -#ifdef CONFIG_PM_SLEEP
> -static int tegra_dma_pm_suspend(struct device *dev)
> -{
> -	struct tegra_dma *tdma = dev_get_drvdata(dev);
>  	int i;
> -	int ret;
> -
> -	/* Enable clock before accessing register */
> -	ret = pm_runtime_get_sync(dev);
> -	if (ret < 0)
> -		return ret;
>  
>  	tdma->reg_gen = tdma_read(tdma, TEGRA_APBDMA_GENERAL);
>  	for (i = 0; i < tdma->chip_data->nr_channels; i++) {
> @@ -1543,21 +1515,21 @@ static int tegra_dma_pm_suspend(struct device *dev)
>  						  TEGRA_APBDMA_CHAN_WCOUNT);
>  	}
>  
> -	/* Disable clock */
> -	pm_runtime_put(dev);
> +	clk_disable_unprepare(tdma->dma_clk);
> +
>  	return 0;
>  }
>  
> -static int tegra_dma_pm_resume(struct device *dev)
> +static int tegra_dma_runtime_resume(struct device *dev)
>  {
>  	struct tegra_dma *tdma = dev_get_drvdata(dev);
> -	int i;
> -	int ret;
> +	int i, ret;
>  
> -	/* Enable clock before accessing register */
> -	ret = pm_runtime_get_sync(dev);
> -	if (ret < 0)
> +	ret = clk_prepare_enable(tdma->dma_clk);
> +	if (ret < 0) {
> +		dev_err(dev, "clk_enable failed: %d\n", ret);
>  		return ret;
> +	}
>  
>  	tdma_write(tdma, TEGRA_APBDMA_GENERAL, tdma->reg_gen);
>  	tdma_write(tdma, TEGRA_APBDMA_CONTROL, 0);
> @@ -1582,16 +1554,14 @@ static int tegra_dma_pm_resume(struct device *dev)
>  			(ch_reg->csr & ~TEGRA_APBDMA_CSR_ENB));
>  	}
>  
> -	/* Disable clock */
> -	pm_runtime_put(dev);
>  	return 0;
>  }
> -#endif
>  
>  static const struct dev_pm_ops tegra_dma_dev_pm_ops = {
>  	SET_RUNTIME_PM_OPS(tegra_dma_runtime_suspend, tegra_dma_runtime_resume,
>  			   NULL)
> -	SET_SYSTEM_SLEEP_PM_OPS(tegra_dma_pm_suspend, tegra_dma_pm_resume)
> +	SET_SYSTEM_SLEEP_PM_OPS(pm_runtime_force_suspend,
> +				pm_runtime_force_resume)

Is that even necessary? I thought runtime PM was going to be triggered
for system sleep anyway, but it looks like there are other examples of
this usage, so maybe I'm mistaken.

Thierry

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [PATCH V2] dmaengine: tegra-apb: Really fix runtime-pm usage

[Jon Hunter]

On 06/06/17 15:48, Thierry Reding wrote:
> * PGP Signed by an unknown key
> 
> On Tue, Jun 06, 2017 at 01:49:29PM +0100, Jon Hunter wrote:
>> Commit edd3bdbe9db1 ("dmaengine: tegra-apb: Correct runtime-pm usage")
>> added pm_runtime_get/put() calls to the tegra-apb DMA system suspend
>> callbacks. Runtime PM is disabled during system suspend and so these
>> APIs cannot be used. Fix the suspend handling for the tegra-apb DMA by
>> moving the save and restore of the DMA register context into the
>> runtime PM suspend and resume callbacks, and then use the
>> pm_runtime_force_suspend/resume() APIs to invoke the runtime PM
>> callbacks during system suspend.
>>
>> Fixes: edd3bdbe9db1 ("dmaengine: tegra-apb: Correct runtime-pm usage")
>>
>> Signed-off-by: Jon Hunter <jonathanh-DDmLM1+adcrQT0dZR+AlfA@public.gmane.org>
>> ---
>>
>> Changes since V1:
>> - Drop the custom suspend/resume callbacks and use
>>   pm_runtime_force_suspend/resume directly in SET_SYSTEM_SLEEP_PM_OPS.
>>
>>  drivers/dma/tegra20-apb-dma.c | 50 +++++++++----------------------------------
>>  1 file changed, 10 insertions(+), 40 deletions(-)
>>
>> diff --git a/drivers/dma/tegra20-apb-dma.c b/drivers/dma/tegra20-apb-dma.c
>> index 3722b9d8d9fe..b9d75a54c896 100644
>> --- a/drivers/dma/tegra20-apb-dma.c
>> +++ b/drivers/dma/tegra20-apb-dma.c
>> @@ -1494,35 +1494,7 @@ static int tegra_dma_remove(struct platform_device *pdev)
>>  static int tegra_dma_runtime_suspend(struct device *dev)
>>  {
>>  	struct tegra_dma *tdma = dev_get_drvdata(dev);
>> -
>> -	clk_disable_unprepare(tdma->dma_clk);
>> -	return 0;
>> -}
>> -
>> -static int tegra_dma_runtime_resume(struct device *dev)
>> -{
>> -	struct tegra_dma *tdma = dev_get_drvdata(dev);
>> -	int ret;
>> -
>> -	ret = clk_prepare_enable(tdma->dma_clk);
>> -	if (ret < 0) {
>> -		dev_err(dev, "clk_enable failed: %d\n", ret);
>> -		return ret;
>> -	}
>> -	return 0;
>> -}
>> -
>> -#ifdef CONFIG_PM_SLEEP
>> -static int tegra_dma_pm_suspend(struct device *dev)
>> -{
>> -	struct tegra_dma *tdma = dev_get_drvdata(dev);
>>  	int i;
>> -	int ret;
>> -
>> -	/* Enable clock before accessing register */
>> -	ret = pm_runtime_get_sync(dev);
>> -	if (ret < 0)
>> -		return ret;
>>  
>>  	tdma->reg_gen = tdma_read(tdma, TEGRA_APBDMA_GENERAL);
>>  	for (i = 0; i < tdma->chip_data->nr_channels; i++) {
>> @@ -1543,21 +1515,21 @@ static int tegra_dma_pm_suspend(struct device *dev)
>>  						  TEGRA_APBDMA_CHAN_WCOUNT);
>>  	}
>>  
>> -	/* Disable clock */
>> -	pm_runtime_put(dev);
>> +	clk_disable_unprepare(tdma->dma_clk);
>> +
>>  	return 0;
>>  }
>>  
>> -static int tegra_dma_pm_resume(struct device *dev)
>> +static int tegra_dma_runtime_resume(struct device *dev)
>>  {
>>  	struct tegra_dma *tdma = dev_get_drvdata(dev);
>> -	int i;
>> -	int ret;
>> +	int i, ret;
>>  
>> -	/* Enable clock before accessing register */
>> -	ret = pm_runtime_get_sync(dev);
>> -	if (ret < 0)
>> +	ret = clk_prepare_enable(tdma->dma_clk);
>> +	if (ret < 0) {
>> +		dev_err(dev, "clk_enable failed: %d\n", ret);
>>  		return ret;
>> +	}
>>  
>>  	tdma_write(tdma, TEGRA_APBDMA_GENERAL, tdma->reg_gen);
>>  	tdma_write(tdma, TEGRA_APBDMA_CONTROL, 0);
>> @@ -1582,16 +1554,14 @@ static int tegra_dma_pm_resume(struct device *dev)
>>  			(ch_reg->csr & ~TEGRA_APBDMA_CSR_ENB));
>>  	}
>>  
>> -	/* Disable clock */
>> -	pm_runtime_put(dev);
>>  	return 0;
>>  }
>> -#endif
>>  
>>  static const struct dev_pm_ops tegra_dma_dev_pm_ops = {
>>  	SET_RUNTIME_PM_OPS(tegra_dma_runtime_suspend, tegra_dma_runtime_resume,
>>  			   NULL)
>> -	SET_SYSTEM_SLEEP_PM_OPS(tegra_dma_pm_suspend, tegra_dma_pm_resume)
>> +	SET_SYSTEM_SLEEP_PM_OPS(pm_runtime_force_suspend,
>> +				pm_runtime_force_resume)
> 
> Is that even necessary? I thought runtime PM was going to be triggered
> for system sleep anyway, but it looks like there are other examples of
> this usage, so maybe I'm mistaken.

Yes this is necessary. No RPM is not automatically trigger by system
suspend AFAICT.

Cheers
Jon

-- 
nvpublic

Re: [PATCH V2] dmaengine: tegra-apb: Really fix runtime-pm usage

[Vinod Koul]

On Tue, Jun 27, 2017 at 12:44:58PM +0100, Jon Hunter wrote:

> >>  static const struct dev_pm_ops tegra_dma_dev_pm_ops = {
> >>  	SET_RUNTIME_PM_OPS(tegra_dma_runtime_suspend, tegra_dma_runtime_resume,
> >>  			   NULL)
> >> -	SET_SYSTEM_SLEEP_PM_OPS(tegra_dma_pm_suspend, tegra_dma_pm_resume)
> >> +	SET_SYSTEM_SLEEP_PM_OPS(pm_runtime_force_suspend,
> >> +				pm_runtime_force_resume)
> > 
> > Is that even necessary? I thought runtime PM was going to be triggered
> > for system sleep anyway, but it looks like there are other examples of
> > this usage, so maybe I'm mistaken.
> 
> Yes this is necessary. No RPM is not automatically trigger by system
> suspend AFAICT.

Yes I was earlier under the same impression but later did realize that the
behaviour seems to be arch specific and we don't have guarantee on this

-- 
~Vinod

Re: [PATCH V2] dmaengine: tegra-apb: Really fix runtime-pm usage

[Vinod Koul]

On Tue, Jun 06, 2017 at 01:49:29PM +0100, Jon Hunter wrote:
> Commit edd3bdbe9db1 ("dmaengine: tegra-apb: Correct runtime-pm usage")
> added pm_runtime_get/put() calls to the tegra-apb DMA system suspend
> callbacks. Runtime PM is disabled during system suspend and so these
> APIs cannot be used. Fix the suspend handling for the tegra-apb DMA by
> moving the save and restore of the DMA register context into the
> runtime PM suspend and resume callbacks, and then use the
> pm_runtime_force_suspend/resume() APIs to invoke the runtime PM
> callbacks during system suspend.

Applied, thanks

-- 
~Vinod