[PATCH v3] ARM: dts: mt8173: support arm64 cpuidle-dt

[PATCH v3] ARM: dts: mt8173: support arm64 cpuidle-dt

[Howard Chen]

This patch adds an idle-states node to describe the mt8173 idle states and
also adds references to the idle-states node in all CPU nodes.

Signed-off-by: Howard Chen <howard.chen-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
---
 arch/arm64/boot/dts/mediatek/mt8173.dtsi | 18 ++++++++++++++++++
 1 file changed, 18 insertions(+)

diff --git a/arch/arm64/boot/dts/mediatek/mt8173.dtsi b/arch/arm64/boot/dts/mediatek/mt8173.dtsi
index 924fdb6..47e8d56 100644
--- a/arch/arm64/boot/dts/mediatek/mt8173.dtsi
+++ b/arch/arm64/boot/dts/mediatek/mt8173.dtsi
@@ -49,6 +49,8 @@
 			device_type = "cpu";
 			compatible = "arm,cortex-a53";
 			reg = <0x000>;
+			enable-method = "psci";
+			cpu-idle-states = <&CPU_SLEEP_0>;
 		};
 
 		cpu1: cpu@1 {
@@ -56,6 +58,7 @@
 			compatible = "arm,cortex-a53";
 			reg = <0x001>;
 			enable-method = "psci";
+			cpu-idle-states = <&CPU_SLEEP_0>;
 		};
 
 		cpu2: cpu@100 {
@@ -63,6 +66,7 @@
 			compatible = "arm,cortex-a57";
 			reg = <0x100>;
 			enable-method = "psci";
+			cpu-idle-states = <&CPU_SLEEP_0>;
 		};
 
 		cpu3: cpu@101 {
@@ -70,6 +74,20 @@
 			compatible = "arm,cortex-a57";
 			reg = <0x101>;
 			enable-method = "psci";
+			cpu-idle-states = <&CPU_SLEEP_0>;
+		};
+
+		idle-states {
+			entry-method = "arm,psci";
+
+			CPU_SLEEP_0: cpu-sleep-0 {
+				compatible = "arm,idle-state";
+				arm,psci-suspend-param = <0x0010000>;
+				entry-latency-us = <639>;
+				exit-latency-us = <680>;
+				min-residency-us = <1088>;
+				local-timer-stop;
+			};
 		};
 	};
 
-- 
1.9.1

--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH v3] ARM: dts: mt8173: support arm64 cpuidle-dt

[Lorenzo Pieralisi]

On Fri, May 29, 2015 at 08:01:46AM +0100, Howard Chen wrote:
> This patch adds an idle-states node to describe the mt8173 idle states and
> also adds references to the idle-states node in all CPU nodes.
> 
> Signed-off-by: Howard Chen <howard.chen-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
> ---
You should have added a list of incremental changes (v1->v2->v3) here
otherwise there is no point in adding a patch version, we are not
supposed to remember what you have changed.

>  arch/arm64/boot/dts/mediatek/mt8173.dtsi | 18 ++++++++++++++++++
>  1 file changed, 18 insertions(+)
> 
> diff --git a/arch/arm64/boot/dts/mediatek/mt8173.dtsi b/arch/arm64/boot/dts/mediatek/mt8173.dtsi
> index 924fdb6..47e8d56 100644
> --- a/arch/arm64/boot/dts/mediatek/mt8173.dtsi
> +++ b/arch/arm64/boot/dts/mediatek/mt8173.dtsi
> @@ -49,6 +49,8 @@
>  			device_type = "cpu";
>  			compatible = "arm,cortex-a53";
>  			reg = <0x000>;
> +			enable-method = "psci";
> +			cpu-idle-states = <&CPU_SLEEP_0>;
>  		};
>  
>  		cpu1: cpu@1 {
> @@ -56,6 +58,7 @@
>  			compatible = "arm,cortex-a53";
>  			reg = <0x001>;
>  			enable-method = "psci";
> +			cpu-idle-states = <&CPU_SLEEP_0>;
>  		};
>  
>  		cpu2: cpu@100 {
> @@ -63,6 +66,7 @@
>  			compatible = "arm,cortex-a57";
>  			reg = <0x100>;
>  			enable-method = "psci";
> +			cpu-idle-states = <&CPU_SLEEP_0>;
>  		};
>  
>  		cpu3: cpu@101 {
> @@ -70,6 +74,20 @@
>  			compatible = "arm,cortex-a57";
>  			reg = <0x101>;
>  			enable-method = "psci";
> +			cpu-idle-states = <&CPU_SLEEP_0>;
> +		};
> +
> +		idle-states {
> +			entry-method = "arm,psci";
> +
> +			CPU_SLEEP_0: cpu-sleep-0 {
> +				compatible = "arm,idle-state";

UltraNit: please move local-timer-stop here to keep the property ordering
as in the bindings.

More importantly, IIRC the idle state was not local-timer-stop in v2,
what changed in the interim ? v1->v2 the cluster idle state disappeared,
now the local timer seems to stop when it was not before (and that is
the same HW), I hope things are settled now.

https://lkml.org/lkml/2015/4/7/76

> +				arm,psci-suspend-param = <0x0010000>;

Move this property as last on the list, let's keep the generic
properties first.

> +				entry-latency-us = <639>;
> +				exit-latency-us = <680>;
> +				min-residency-us = <1088>;
> +				local-timer-stop;
> +			};
>  		};
>  	};

I guess there is no hope of seeing the cluster state restored (?).

I would like some explanation regarding local-timer-stop change, apart
from that you can add:

Acked-by: Lorenzo Pieralisi <lorenzo.pieralisi-5wv7dgnIgG8@public.gmane.org>
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH v3] ARM: dts: mt8173: support arm64 cpuidle-dt

[Howard Chen]

On 29 May 2015 at 22:55, Lorenzo Pieralisi <lorenzo.pieralisi-5wv7dgnIgG8@public.gmane.org> wrote:
> On Fri, May 29, 2015 at 08:01:46AM +0100, Howard Chen wrote:
>> This patch adds an idle-states node to describe the mt8173 idle states and
>> also adds references to the idle-states node in all CPU nodes.
>>
>> Signed-off-by: Howard Chen <howard.chen-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
>> ---
> You should have added a list of incremental changes (v1->v2->v3) here
> otherwise there is no point in adding a patch version, we are not
> supposed to remember what you have changed.
>
OK.

>>  arch/arm64/boot/dts/mediatek/mt8173.dtsi | 18 ++++++++++++++++++
>>  1 file changed, 18 insertions(+)
>>
>> diff --git a/arch/arm64/boot/dts/mediatek/mt8173.dtsi b/arch/arm64/boot/dts/mediatek/mt8173.dtsi
>> index 924fdb6..47e8d56 100644
>> --- a/arch/arm64/boot/dts/mediatek/mt8173.dtsi
>> +++ b/arch/arm64/boot/dts/mediatek/mt8173.dtsi
>> @@ -49,6 +49,8 @@
>>                       device_type = "cpu";
>>                       compatible = "arm,cortex-a53";
>>                       reg = <0x000>;
>> +                     enable-method = "psci";
>> +                     cpu-idle-states = <&CPU_SLEEP_0>;
>>               };
>>
>>               cpu1: cpu@1 {
>> @@ -56,6 +58,7 @@
>>                       compatible = "arm,cortex-a53";
>>                       reg = <0x001>;
>>                       enable-method = "psci";
>> +                     cpu-idle-states = <&CPU_SLEEP_0>;
>>               };
>>
>>               cpu2: cpu@100 {
>> @@ -63,6 +66,7 @@
>>                       compatible = "arm,cortex-a57";
>>                       reg = <0x100>;
>>                       enable-method = "psci";
>> +                     cpu-idle-states = <&CPU_SLEEP_0>;
>>               };
>>
>>               cpu3: cpu@101 {
>> @@ -70,6 +74,20 @@
>>                       compatible = "arm,cortex-a57";
>>                       reg = <0x101>;
>>                       enable-method = "psci";
>> +                     cpu-idle-states = <&CPU_SLEEP_0>;
>> +             };
>> +
>> +             idle-states {
>> +                     entry-method = "arm,psci";
>> +
>> +                     CPU_SLEEP_0: cpu-sleep-0 {
>> +                             compatible = "arm,idle-state";
>
> UltraNit: please move local-timer-stop here to keep the property ordering
> as in the bindings.
OK.

>
> More importantly, IIRC the idle state was not local-timer-stop in v2,

My original idea is to hand over to a general purpose timer when the
local timer stops, however, Danienl Lezcano naked this and points out cases
that it may misbehave. So I fall back to the generic approach which uses the
broadcast timer on the v3 patch.

> what changed in the interim ? v1->v2 the cluster idle state disappeared,

The cluster idle won't be available in the near future and possibly never will,
while the C1 state is ready and stable, so I split the C1 part out.

> now the local timer seems to stop when it was not before (and that is
> the same HW), I hope things are settled now.
>
> https://lkml.org/lkml/2015/4/7/76
>
>> +                             arm,psci-suspend-param = <0x0010000>;
>
> Move this property as last on the list, let's keep the generic
> properties first.
OK.

>
>> +                             entry-latency-us = <639>;
>> +                             exit-latency-us = <680>;
>> +                             min-residency-us = <1088>;
>> +                             local-timer-stop;
>> +                     };
>>               };
>>       };
>
> I guess there is no hope of seeing the cluster state restored (?).
>
> I would like some explanation regarding local-timer-stop change, apart
> from that you can add:
>
> Acked-by: Lorenzo Pieralisi <lorenzo.pieralisi-5wv7dgnIgG8@public.gmane.org>

Many thanks for your review and all the great comments.


Regards,
Howard
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH v3] ARM: dts: mt8173: support arm64 cpuidle-dt

[Amit Kucheria]

On Tue, Jun 2, 2015 at 12:47 PM, Howard Chen <howard.chen-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org> wrote:

>> what changed in the interim ? v1->v2 the cluster idle state disappeared,
>
> The cluster idle won't be available in the near future and possibly never will,
> while the C1 state is ready and stable, so I split the C1 part out.

May I ask why cluster idle won't be available?
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html