[PATCH 0/3] PM / Domains: Add initial PM clock support to genpd

[PATCH 0/3] PM / Domains: Add initial PM clock support to genpd

[Ulf Hansson]

It's quite common for PM domains to use PM clocks. Typically from SOC specific
code, the per device PM clock list is created and pm_clk_suspend|resume() are
invoked to handle clock gating/ungating.

A step towards consolidation is to integrate PM clock support into genpd, which
is what this patchset does.

In this initial step, the calls to the pm_clk_suspend|resume() are handled
within genpd, but the per device PM clock list still needs to be created from
SOC specific code. It seems reasonable to have gendp to handle that as well, but
that left to future patchsets to address.

It's not every users of genpd that are keen on using PM clocks thus we need
to provide this a configuration option for genpd.

In patch 2 and patch 3, we convert some of the SH-MOBILE SOCs to start using the
new PM clock support in genpd.


Ulf Hansson (3):
  PM / Domains: Initial PM clock support for genpd
  ARM: shmobile: Convert to genpd flags for PM clocks for r8a7779
  ARM: shmobile: Convert to genpd flags for PM clocks for R-mobile

 arch/arm/mach-shmobile/pm-r8a7779.c | 3 +--
 arch/arm/mach-shmobile/pm-rmobile.c | 3 +--
 drivers/base/power/domain.c         | 7 +++++++
 include/linux/pm_domain.h           | 3 +++
 4 files changed, 12 insertions(+), 4 deletions(-)

-- 
1.9.1

[PATCH 1/3] PM / Domains: Initial PM clock support for genpd

[Ulf Hansson]

It's quite common for PM domains to use PM clocks. Typically from SOC
specific code, the per device PM clock list is created and
pm_clk_suspend|resume() are invoked to handle clock gating/ungating.

A step towards consolidation is to integrate PM clock support into
genpd, which is what this patch does.

In this initial step, the calls to the pm_clk_suspend|resume() are
handled within genpd, but the per device PM clock list still needs to
be created from SOC specific code. It seems reasonable to have gendp to
handle that as well, but that left to future patches to address.

It's not every users of genpd that are keen on using PM clocks thus we
need to provide this a configuration option for genpd. Therefore let's
add flag field in the genpd struct to keep this information and define
a new PM_DOMAIN_PM_CLK bit can then be set at initialization.

Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
---
 drivers/base/power/domain.c | 7 +++++++
 include/linux/pm_domain.h   | 3 +++
 2 files changed, 10 insertions(+)

diff --git a/drivers/base/power/domain.c b/drivers/base/power/domain.c
index 3989eb6..42e328c 100644
--- a/drivers/base/power/domain.c
+++ b/drivers/base/power/domain.c
@@ -12,6 +12,7 @@
 #include <linux/pm_runtime.h>
 #include <linux/pm_domain.h>
 #include <linux/pm_qos.h>
+#include <linux/pm_clock.h>
 #include <linux/slab.h>
 #include <linux/err.h>
 #include <linux/sched.h>
@@ -1948,6 +1949,12 @@ void pm_genpd_init(struct generic_pm_domain *genpd,
 	genpd->domain.ops.complete = pm_genpd_complete;
 	genpd->dev_ops.save_state = pm_genpd_default_save_state;
 	genpd->dev_ops.restore_state = pm_genpd_default_restore_state;
+
+	if (genpd->flags & PM_DOMAIN_PM_CLK) {
+		genpd->dev_ops.stop = pm_clk_suspend;
+		genpd->dev_ops.start = pm_clk_suspend;
+	}
+
 	mutex_lock(&gpd_list_lock);
 	list_add(&genpd->gpd_list_node, &gpd_list);
 	mutex_unlock(&gpd_list_lock);
diff --git a/include/linux/pm_domain.h b/include/linux/pm_domain.h
index 9d254e2..44c6931 100644
--- a/include/linux/pm_domain.h
+++ b/include/linux/pm_domain.h
@@ -14,6 +14,7 @@
 #include <linux/pm.h>
 #include <linux/err.h>
 #include <linux/of.h>
+#include <linux/bitops.h>
 #include <linux/notifier.h>
 #include <linux/cpuidle.h>
 
@@ -76,6 +77,8 @@ struct generic_pm_domain {
 			  struct device *dev);
 	void (*detach_dev)(struct generic_pm_domain *domain,
 			   struct device *dev);
+	unsigned int flags;		/* Bit field of configs for genpd */
+#define PM_DOMAIN_PM_CLK	BIT(0)	/* PM domain use PM clk */
 };
 
 static inline struct generic_pm_domain *pd_to_genpd(struct dev_pm_domain *pd)
-- 
1.9.1

[PATCH 1/3] PM / Domains: Initial PM clock support for genpd

[Dmitry Torokhov]

On Wed, Nov 19, 2014 at 03:00:36PM +0100, Ulf Hansson wrote:
> It's quite common for PM domains to use PM clocks. Typically from SOC
> specific code, the per device PM clock list is created and
> pm_clk_suspend|resume() are invoked to handle clock gating/ungating.
> 
> A step towards consolidation is to integrate PM clock support into
> genpd, which is what this patch does.
> 
> In this initial step, the calls to the pm_clk_suspend|resume() are
> handled within genpd, but the per device PM clock list still needs to
> be created from SOC specific code. It seems reasonable to have gendp to
> handle that as well, but that left to future patches to address.
> 
> It's not every users of genpd that are keen on using PM clocks thus we
> need to provide this a configuration option for genpd. Therefore let's
> add flag field in the genpd struct to keep this information and define
> a new PM_DOMAIN_PM_CLK bit can then be set at initialization.
> 
> Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
> ---
>  drivers/base/power/domain.c | 7 +++++++
>  include/linux/pm_domain.h   | 3 +++
>  2 files changed, 10 insertions(+)
> 
> diff --git a/drivers/base/power/domain.c b/drivers/base/power/domain.c
> index 3989eb6..42e328c 100644
> --- a/drivers/base/power/domain.c
> +++ b/drivers/base/power/domain.c
> @@ -12,6 +12,7 @@
>  #include <linux/pm_runtime.h>
>  #include <linux/pm_domain.h>
>  #include <linux/pm_qos.h>
> +#include <linux/pm_clock.h>
>  #include <linux/slab.h>
>  #include <linux/err.h>
>  #include <linux/sched.h>
> @@ -1948,6 +1949,12 @@ void pm_genpd_init(struct generic_pm_domain *genpd,
>  	genpd->domain.ops.complete = pm_genpd_complete;
>  	genpd->dev_ops.save_state = pm_genpd_default_save_state;
>  	genpd->dev_ops.restore_state = pm_genpd_default_restore_state;
> +
> +	if (genpd->flags & PM_DOMAIN_PM_CLK) {
> +		genpd->dev_ops.stop = pm_clk_suspend;
> +		genpd->dev_ops.start = pm_clk_suspend;

The 2nd one is wrong.

> +	}
> +
>  	mutex_lock(&gpd_list_lock);
>  	list_add(&genpd->gpd_list_node, &gpd_list);
>  	mutex_unlock(&gpd_list_lock);
> diff --git a/include/linux/pm_domain.h b/include/linux/pm_domain.h
> index 9d254e2..44c6931 100644
> --- a/include/linux/pm_domain.h
> +++ b/include/linux/pm_domain.h
> @@ -14,6 +14,7 @@
>  #include <linux/pm.h>
>  #include <linux/err.h>
>  #include <linux/of.h>
> +#include <linux/bitops.h>
>  #include <linux/notifier.h>
>  #include <linux/cpuidle.h>
>  
> @@ -76,6 +77,8 @@ struct generic_pm_domain {
>  			  struct device *dev);
>  	void (*detach_dev)(struct generic_pm_domain *domain,
>  			   struct device *dev);
> +	unsigned int flags;		/* Bit field of configs for genpd */
> +#define PM_DOMAIN_PM_CLK	BIT(0)	/* PM domain use PM clk */

s/use/uses ?

Are you planning on adding a separate flag for collecting clocks from OF
on attach/detach?

>  };
>  
>  static inline struct generic_pm_domain *pd_to_genpd(struct dev_pm_domain *pd)
> -- 
> 1.9.1
> 

Thanks.

-- 
Dmitry

[PATCH 1/3] PM / Domains: Initial PM clock support for genpd

[Ulf Hansson]

On 19 November 2014 18:25, Dmitry Torokhov <dmitry.torokhov@gmail.com> wrote:
> On Wed, Nov 19, 2014 at 03:00:36PM +0100, Ulf Hansson wrote:
>> It's quite common for PM domains to use PM clocks. Typically from SOC
>> specific code, the per device PM clock list is created and
>> pm_clk_suspend|resume() are invoked to handle clock gating/ungating.
>>
>> A step towards consolidation is to integrate PM clock support into
>> genpd, which is what this patch does.
>>
>> In this initial step, the calls to the pm_clk_suspend|resume() are
>> handled within genpd, but the per device PM clock list still needs to
>> be created from SOC specific code. It seems reasonable to have gendp to
>> handle that as well, but that left to future patches to address.
>>
>> It's not every users of genpd that are keen on using PM clocks thus we
>> need to provide this a configuration option for genpd. Therefore let's
>> add flag field in the genpd struct to keep this information and define
>> a new PM_DOMAIN_PM_CLK bit can then be set at initialization.
>>
>> Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
>> ---
>>  drivers/base/power/domain.c | 7 +++++++
>>  include/linux/pm_domain.h   | 3 +++
>>  2 files changed, 10 insertions(+)
>>
>> diff --git a/drivers/base/power/domain.c b/drivers/base/power/domain.c
>> index 3989eb6..42e328c 100644
>> --- a/drivers/base/power/domain.c
>> +++ b/drivers/base/power/domain.c
>> @@ -12,6 +12,7 @@
>>  #include <linux/pm_runtime.h>
>>  #include <linux/pm_domain.h>
>>  #include <linux/pm_qos.h>
>> +#include <linux/pm_clock.h>
>>  #include <linux/slab.h>
>>  #include <linux/err.h>
>>  #include <linux/sched.h>
>> @@ -1948,6 +1949,12 @@ void pm_genpd_init(struct generic_pm_domain *genpd,
>>       genpd->domain.ops.complete = pm_genpd_complete;
>>       genpd->dev_ops.save_state = pm_genpd_default_save_state;
>>       genpd->dev_ops.restore_state = pm_genpd_default_restore_state;
>> +
>> +     if (genpd->flags & PM_DOMAIN_PM_CLK) {
>> +             genpd->dev_ops.stop = pm_clk_suspend;
>> +             genpd->dev_ops.start = pm_clk_suspend;
>
> The 2nd one is wrong.

Huh, thanks for spotting this.

>
>> +     }
>> +
>>       mutex_lock(&gpd_list_lock);
>>       list_add(&genpd->gpd_list_node, &gpd_list);
>>       mutex_unlock(&gpd_list_lock);
>> diff --git a/include/linux/pm_domain.h b/include/linux/pm_domain.h
>> index 9d254e2..44c6931 100644
>> --- a/include/linux/pm_domain.h
>> +++ b/include/linux/pm_domain.h
>> @@ -14,6 +14,7 @@
>>  #include <linux/pm.h>
>>  #include <linux/err.h>
>>  #include <linux/of.h>
>> +#include <linux/bitops.h>
>>  #include <linux/notifier.h>
>>  #include <linux/cpuidle.h>
>>
>> @@ -76,6 +77,8 @@ struct generic_pm_domain {
>>                         struct device *dev);
>>       void (*detach_dev)(struct generic_pm_domain *domain,
>>                          struct device *dev);
>> +     unsigned int flags;             /* Bit field of configs for genpd */
>> +#define PM_DOMAIN_PM_CLK     BIT(0)  /* PM domain use PM clk */
>
> s/use/uses ?
>
> Are you planning on adding a separate flag for collecting clocks from OF
> on attach/detach?

Yes, I am thinking on that approach. Do you have any suggestions around this?

Kind regards
Uffe

[PATCH 2/3] ARM: shmobile: Convert to genpd flags for PM clocks for r8a7779

[Ulf Hansson]

Instead of using the dev_ops ->stop|start() callbacks for genpd, let's
convert to use genpd's flag field and set it to PM_DOMAIN_PM_CLK.

Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
---
 arch/arm/mach-shmobile/pm-r8a7779.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/arch/arm/mach-shmobile/pm-r8a7779.c b/arch/arm/mach-shmobile/pm-r8a7779.c
index 82fe3d7..6d1c011 100644
--- a/arch/arm/mach-shmobile/pm-r8a7779.c
+++ b/arch/arm/mach-shmobile/pm-r8a7779.c
@@ -83,9 +83,8 @@ static void r8a7779_init_pm_domain(struct r8a7779_pm_domain *r8a7779_pd)
 {
 	struct generic_pm_domain *genpd = &r8a7779_pd->genpd;
 
+	genpd->flags = PM_DOMAIN_PM_CLK;
 	pm_genpd_init(genpd, NULL, false);
-	genpd->dev_ops.stop = pm_clk_suspend;
-	genpd->dev_ops.start = pm_clk_resume;
 	genpd->dev_ops.active_wakeup = pd_active_wakeup;
 	genpd->power_off = pd_power_down;
 	genpd->power_on = pd_power_up;
-- 
1.9.1

[PATCH 3/3] ARM: shmobile: Convert to genpd flags for PM clocks for R-mobile

[Ulf Hansson]

Instead of using the dev_ops ->stop|start() callbacks for genpd, let's
convert to use genpd's flag field and set it to PM_DOMAIN_PM_CLK.

Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
---
 arch/arm/mach-shmobile/pm-rmobile.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/arch/arm/mach-shmobile/pm-rmobile.c b/arch/arm/mach-shmobile/pm-rmobile.c
index 717e641..42af9bd 100644
--- a/arch/arm/mach-shmobile/pm-rmobile.c
+++ b/arch/arm/mach-shmobile/pm-rmobile.c
@@ -106,9 +106,8 @@ static void rmobile_init_pm_domain(struct rmobile_pm_domain *rmobile_pd)
 	struct generic_pm_domain *genpd = &rmobile_pd->genpd;
 	struct dev_power_governor *gov = rmobile_pd->gov;
 
+	genpd->flags = PM_DOMAIN_PM_CLK;
 	pm_genpd_init(genpd, gov ? : &simple_qos_governor, false);
-	genpd->dev_ops.stop		= pm_clk_suspend;
-	genpd->dev_ops.start		= pm_clk_resume;
 	genpd->dev_ops.active_wakeup	= rmobile_pd_active_wakeup;
 	genpd->power_off		= rmobile_pd_power_down;
 	genpd->power_on			= rmobile_pd_power_up;
-- 
1.9.1

[PATCH 0/3] PM / Domains: Add initial PM clock support to genpd

[Geert Uytterhoeven]

Hi Ulf,

On Wed, Nov 19, 2014 at 3:00 PM, Ulf Hansson <ulf.hansson@linaro.org> wrote:
> It's quite common for PM domains to use PM clocks. Typically from SOC specific
> code, the per device PM clock list is created and pm_clk_suspend|resume() are
> invoked to handle clock gating/ungating.
>
> A step towards consolidation is to integrate PM clock support into genpd, which
> is what this patchset does.

Thanks, this sounds like a good idea.

> In this initial step, the calls to the pm_clk_suspend|resume() are handled
> within genpd, but the per device PM clock list still needs to be created from
> SOC specific code. It seems reasonable to have gendp to handle that as well, but

s/gendp/genpd/

> that left to future patchsets to address.

If/when genpd will handle the clock creation, perhaps the flag can just
be replaced by a pointer to a list of clocks?
Or do you have other uses in mind for adding the flags field now?

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert at linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds

[PATCH 0/3] PM / Domains: Add initial PM clock support to genpd

[Kevin Hilman]

Ulf Hansson <ulf.hansson@linaro.org> writes:

> It's quite common for PM domains to use PM clocks. Typically from SOC specific
> code, the per device PM clock list is created and pm_clk_suspend|resume() are
> invoked to handle clock gating/ungating.
>
> A step towards consolidation is to integrate PM clock support into genpd, which
> is what this patchset does.
>
> In this initial step, the calls to the pm_clk_suspend|resume() are handled
> within genpd, but the per device PM clock list still needs to be created from
> SOC specific code. It seems reasonable to have gendp to handle that as well, but
> that left to future patchsets to address.

I think we need to get rid of the SoC specific code already.  For
example, we're already seeing SoCs where the arm32 core is being
replaced by an arm64 core but the other IPs, and power-domain logic is
staying more or less the same.  

> It's not every users of genpd that are keen on using PM clocks thus we need
> to provide this a configuration option for genpd.

Or more likely, probably some compatible string, or property in the
domain node.  Grygorii, Arnd and myself were discussing this elsewhere
in the context of the TI Keystone2 PM domain support[1].

Kevin

[1] http://lists.infradead.org/pipermail/linux-arm-kernel/2014-November/304099.html

[PATCH 0/3] PM / Domains: Add initial PM clock support to genpd

[Ulf Hansson]

On 20 November 2014 01:33, Kevin Hilman <khilman@kernel.org> wrote:
> Ulf Hansson <ulf.hansson@linaro.org> writes:
>
>> It's quite common for PM domains to use PM clocks. Typically from SOC specific
>> code, the per device PM clock list is created and pm_clk_suspend|resume() are
>> invoked to handle clock gating/ungating.
>>
>> A step towards consolidation is to integrate PM clock support into genpd, which
>> is what this patchset does.
>>
>> In this initial step, the calls to the pm_clk_suspend|resume() are handled
>> within genpd, but the per device PM clock list still needs to be created from
>> SOC specific code. It seems reasonable to have gendp to handle that as well, but
>> that left to future patchsets to address.
>
> I think we need to get rid of the SoC specific code already.  For
> example, we're already seeing SoCs where the arm32 core is being
> replaced by an arm64 core but the other IPs, and power-domain logic is
> staying more or less the same.
>
>> It's not every users of genpd that are keen on using PM clocks thus we need
>> to provide this a configuration option for genpd.
>
> Or more likely, probably some compatible string, or property in the
> domain node.  Grygorii, Arnd and myself were discussing this elsewhere
> in the context of the TI Keystone2 PM domain support[1].

Thanks for pointing that out. It was actually the reason to why I
posted this patchset now, I have been keeping the patches locally in
my tree for too long. :-)

Let me comment of that thread, instead of here.

Kind regards
Uffe