1. Patchew
  2. linux
  3. [PATCH v1 0/9] coresight: Fix and improve clock usage
  4. View patch

[PATCH v1 7/9] coresight: Make clock sequence consistent

Leo Yan posted 9 patches 8 months, 3 weeks ago
There is a newer version of this series
[PATCH v1 7/9] coresight: Make clock sequence consistent
Posted by Leo Yan 8 months, 3 weeks ago 
Since atclk is enabled after pclk during the probe phase, this commit
maintains the same sequence for the runtime resume flow.

Signed-off-by: Leo Yan <leo.yan@arm.com>
---
 drivers/hwtracing/coresight/coresight-funnel.c     | 6 +++---
 drivers/hwtracing/coresight/coresight-replicator.c | 6 +++---
 drivers/hwtracing/coresight/coresight-stm.c        | 6 +++---
 drivers/hwtracing/coresight/coresight-tpiu.c       | 6 +++---
 4 files changed, 12 insertions(+), 12 deletions(-)

diff --git a/drivers/hwtracing/coresight/coresight-funnel.c b/drivers/hwtracing/coresight/coresight-funnel.c
index ec6d3e656548..e378c2fffca9 100644
--- a/drivers/hwtracing/coresight/coresight-funnel.c
+++ b/drivers/hwtracing/coresight/coresight-funnel.c
@@ -299,11 +299,11 @@ static int funnel_runtime_resume(struct device *dev)
 {
 	struct funnel_drvdata *drvdata = dev_get_drvdata(dev);
 
-	if (drvdata && !IS_ERR(drvdata->atclk))
-		clk_prepare_enable(drvdata->atclk);
-
 	if (drvdata && !IS_ERR_OR_NULL(drvdata->pclk))
 		clk_prepare_enable(drvdata->pclk);
+
+	if (drvdata && !IS_ERR(drvdata->atclk))
+		clk_prepare_enable(drvdata->atclk);
 	return 0;
 }
 #endif
diff --git a/drivers/hwtracing/coresight/coresight-replicator.c b/drivers/hwtracing/coresight/coresight-replicator.c
index 460af0f7b537..78854f586e89 100644
--- a/drivers/hwtracing/coresight/coresight-replicator.c
+++ b/drivers/hwtracing/coresight/coresight-replicator.c
@@ -337,11 +337,11 @@ static int replicator_runtime_resume(struct device *dev)
 {
 	struct replicator_drvdata *drvdata = dev_get_drvdata(dev);
 
-	if (drvdata && !IS_ERR(drvdata->atclk))
-		clk_prepare_enable(drvdata->atclk);
-
 	if (drvdata && !IS_ERR_OR_NULL(drvdata->pclk))
 		clk_prepare_enable(drvdata->pclk);
+
+	if (drvdata && !IS_ERR(drvdata->atclk))
+		clk_prepare_enable(drvdata->atclk);
 	return 0;
 }
 #endif
diff --git a/drivers/hwtracing/coresight/coresight-stm.c b/drivers/hwtracing/coresight/coresight-stm.c
index f13fbab4d7a2..ddb4f6678efd 100644
--- a/drivers/hwtracing/coresight/coresight-stm.c
+++ b/drivers/hwtracing/coresight/coresight-stm.c
@@ -972,11 +972,11 @@ static int stm_runtime_resume(struct device *dev)
 {
 	struct stm_drvdata *drvdata = dev_get_drvdata(dev);
 
-	if (drvdata && !IS_ERR(drvdata->atclk))
-		clk_prepare_enable(drvdata->atclk);
-
 	if (drvdata && !IS_ERR_OR_NULL(drvdata->pclk))
 		clk_prepare_enable(drvdata->pclk);
+
+	if (drvdata && !IS_ERR(drvdata->atclk))
+		clk_prepare_enable(drvdata->atclk);
 	return 0;
 }
 #endif
diff --git a/drivers/hwtracing/coresight/coresight-tpiu.c b/drivers/hwtracing/coresight/coresight-tpiu.c
index cac1b5bba086..8212959f60d4 100644
--- a/drivers/hwtracing/coresight/coresight-tpiu.c
+++ b/drivers/hwtracing/coresight/coresight-tpiu.c
@@ -220,11 +220,11 @@ static int tpiu_runtime_resume(struct device *dev)
 {
 	struct tpiu_drvdata *drvdata = dev_get_drvdata(dev);
 
-	if (drvdata && !IS_ERR(drvdata->atclk))
-		clk_prepare_enable(drvdata->atclk);
-
 	if (drvdata && !IS_ERR_OR_NULL(drvdata->pclk))
 		clk_prepare_enable(drvdata->pclk);
+
+	if (drvdata && !IS_ERR(drvdata->atclk))
+		clk_prepare_enable(drvdata->atclk);
 	return 0;
 }
 #endif
-- 
2.34.1
Re: [PATCH v1 7/9] coresight: Make clock sequence consistent
Posted by Anshuman Khandual 8 months, 2 weeks ago 
On 3/27/25 17:08, Leo Yan wrote:
> Since atclk is enabled after pclk during the probe phase, this commit
> maintains the same sequence for the runtime resume flow.
> 
> Signed-off-by: Leo Yan <leo.yan@arm.com>
> ---
>  drivers/hwtracing/coresight/coresight-funnel.c     | 6 +++---
>  drivers/hwtracing/coresight/coresight-replicator.c | 6 +++---
>  drivers/hwtracing/coresight/coresight-stm.c        | 6 +++---
>  drivers/hwtracing/coresight/coresight-tpiu.c       | 6 +++---
>  4 files changed, 12 insertions(+), 12 deletions(-)
> 
> diff --git a/drivers/hwtracing/coresight/coresight-funnel.c b/drivers/hwtracing/coresight/coresight-funnel.c
> index ec6d3e656548..e378c2fffca9 100644
> --- a/drivers/hwtracing/coresight/coresight-funnel.c
> +++ b/drivers/hwtracing/coresight/coresight-funnel.c
> @@ -299,11 +299,11 @@ static int funnel_runtime_resume(struct device *dev)
>  {
>  	struct funnel_drvdata *drvdata = dev_get_drvdata(dev);
>  
> -	if (drvdata && !IS_ERR(drvdata->atclk))
> -		clk_prepare_enable(drvdata->atclk);
> -
>  	if (drvdata && !IS_ERR_OR_NULL(drvdata->pclk))
>  		clk_prepare_enable(drvdata->pclk);
> +
> +	if (drvdata && !IS_ERR(drvdata->atclk))
> +		clk_prepare_enable(drvdata->atclk);
>  	return 0;
>  }

funnel_probe() enables pclk after atclk though - which needs to be
reversed as well ?

static int funnel_probe(struct device *dev, struct resource *res)
{
	..................
        drvdata->atclk = devm_clk_get_optional_enabled(dev, "atclk");
        if (IS_ERR(drvdata->atclk))
                return PTR_ERR(drvdata->atclk);

        drvdata->pclk = coresight_get_enable_apb_pclk(dev);
        if (IS_ERR(drvdata->pclk))
                return PTR_ERR(drvdata->pclk);

>  #endif
> diff --git a/drivers/hwtracing/coresight/coresight-replicator.c b/drivers/hwtracing/coresight/coresight-replicator.c
> index 460af0f7b537..78854f586e89 100644
> --- a/drivers/hwtracing/coresight/coresight-replicator.c
> +++ b/drivers/hwtracing/coresight/coresight-replicator.c
> @@ -337,11 +337,11 @@ static int replicator_runtime_resume(struct device *dev)
>  {
>  	struct replicator_drvdata *drvdata = dev_get_drvdata(dev);
>  
> -	if (drvdata && !IS_ERR(drvdata->atclk))
> -		clk_prepare_enable(drvdata->atclk);
> -
>  	if (drvdata && !IS_ERR_OR_NULL(drvdata->pclk))
>  		clk_prepare_enable(drvdata->pclk);
> +
> +	if (drvdata && !IS_ERR(drvdata->atclk))
> +		clk_prepare_enable(drvdata->atclk);
>  	return 0;
>  }

replicator_probe() enables pclk after atclk though - which needs to be
reversed as well ?

static int replicator_probe(struct device *dev, struct resource *res)
{
	..................
        drvdata->atclk = devm_clk_get_optional_enabled(dev, "atclk");
        if (IS_ERR(drvdata->atclk))
                return PTR_ERR(drvdata->atclk);

        drvdata->pclk = coresight_get_enable_apb_pclk(dev);
        if (IS_ERR(drvdata->pclk))
                return PTR_ERR(drvdata->pclk);



>  #endif
> diff --git a/drivers/hwtracing/coresight/coresight-stm.c b/drivers/hwtracing/coresight/coresight-stm.c
> index f13fbab4d7a2..ddb4f6678efd 100644
> --- a/drivers/hwtracing/coresight/coresight-stm.c
> +++ b/drivers/hwtracing/coresight/coresight-stm.c
> @@ -972,11 +972,11 @@ static int stm_runtime_resume(struct device *dev)
>  {
>  	struct stm_drvdata *drvdata = dev_get_drvdata(dev);
>  
> -	if (drvdata && !IS_ERR(drvdata->atclk))
> -		clk_prepare_enable(drvdata->atclk);
> -
>  	if (drvdata && !IS_ERR_OR_NULL(drvdata->pclk))
>  		clk_prepare_enable(drvdata->pclk);
> +
> +	if (drvdata && !IS_ERR(drvdata->atclk))
> +		clk_prepare_enable(drvdata->atclk);
>  	return 0;
>  }
>  #endif

same here as well.

> diff --git a/drivers/hwtracing/coresight/coresight-tpiu.c b/drivers/hwtracing/coresight/coresight-tpiu.c
> index cac1b5bba086..8212959f60d4 100644
> --- a/drivers/hwtracing/coresight/coresight-tpiu.c
> +++ b/drivers/hwtracing/coresight/coresight-tpiu.c
> @@ -220,11 +220,11 @@ static int tpiu_runtime_resume(struct device *dev)
>  {
>  	struct tpiu_drvdata *drvdata = dev_get_drvdata(dev);
>  
> -	if (drvdata && !IS_ERR(drvdata->atclk))
> -		clk_prepare_enable(drvdata->atclk);
> -
>  	if (drvdata && !IS_ERR_OR_NULL(drvdata->pclk))
>  		clk_prepare_enable(drvdata->pclk);
> +
> +	if (drvdata && !IS_ERR(drvdata->atclk))
> +		clk_prepare_enable(drvdata->atclk);
>  	return 0;
>  }
>  #endif

same here as well.

I assume this patch is trying to have the same clock sequence enablement
during original probe and resume path and then just the reverse sequence
during suspend path.
Re: [PATCH v1 7/9] coresight: Make clock sequence consistent
Posted by Leo Yan 7 months, 4 weeks ago 
On Thu, Apr 03, 2025 at 12:40:39PM +0530, Anshuman Khandual wrote:
> On 3/27/25 17:08, Leo Yan wrote:
> > Since atclk is enabled after pclk during the probe phase, this commit
> > maintains the same sequence for the runtime resume flow.
> > 
> > Signed-off-by: Leo Yan <leo.yan@arm.com>
> > ---
> >  drivers/hwtracing/coresight/coresight-funnel.c     | 6 +++---
> >  drivers/hwtracing/coresight/coresight-replicator.c | 6 +++---
> >  drivers/hwtracing/coresight/coresight-stm.c        | 6 +++---
> >  drivers/hwtracing/coresight/coresight-tpiu.c       | 6 +++---
> >  4 files changed, 12 insertions(+), 12 deletions(-)
> > 
> > diff --git a/drivers/hwtracing/coresight/coresight-funnel.c b/drivers/hwtracing/coresight/coresight-funnel.c
> > index ec6d3e656548..e378c2fffca9 100644
> > --- a/drivers/hwtracing/coresight/coresight-funnel.c
> > +++ b/drivers/hwtracing/coresight/coresight-funnel.c
> > @@ -299,11 +299,11 @@ static int funnel_runtime_resume(struct device *dev)
> >  {
> >  	struct funnel_drvdata *drvdata = dev_get_drvdata(dev);
> >  
> > -	if (drvdata && !IS_ERR(drvdata->atclk))
> > -		clk_prepare_enable(drvdata->atclk);
> > -
> >  	if (drvdata && !IS_ERR_OR_NULL(drvdata->pclk))
> >  		clk_prepare_enable(drvdata->pclk);
> > +
> > +	if (drvdata && !IS_ERR(drvdata->atclk))
> > +		clk_prepare_enable(drvdata->atclk);
> >  	return 0;
> >  }
> 
> funnel_probe() enables pclk after atclk though - which needs to be
> reversed as well ?

Good point!

The key point is a dynamic probe enables pclk clock in AMBA bus driver,
which is anyway prior to enable atclk.

We need to keep consistent flow for all flows (static probe, dynamic
probe, runtime PM resume).  The patch 09 consolidates clock enabling for
static and dynamic probe, and this patch is for runtime PM.

For a better organization, I will place this patch after the patch 09
in the next spin.

[...]

> I assume this patch is trying to have the same clock sequence enablement
> during original probe and resume path and then just the reverse sequence
> during suspend path.

Correct.  As said, the patch 09 is for clock enabling sequence in
probe, and this patch is for the resume path.

Thanks,
Leo

Patchew 2.1-devel-b67e4c2

© 2016 - 2025 Red Hat, Inc.