[v2] clk: qcom: gcc-sc8280xp: Allow PCIe GDSCs to enter retention state

[PATCH v2] clk: qcom: gcc-sc8280xp: Allow PCIe GDSCs to enter retention state

Posted by Manivannan Sadhasivam 2 years, 7 months ago

With the minimal system suspend support in place for the PCIe driver that
keeps the interconnect path voted, the ALWAYS_ON flag can now be dropped.

Also, the PWRSTS_RET_ON flag should be used to allow the GDSCs to enter the
retention state when the parent domain get's turned off during system
suspend.

Signed-off-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
---

Changes in v2:

* Changed the patch from simple revert to changing the ALWAYS_ON flag to
  PWRSTS_RET_ON.

 drivers/clk/qcom/gcc-sc8280xp.c | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/drivers/clk/qcom/gcc-sc8280xp.c b/drivers/clk/qcom/gcc-sc8280xp.c
index 04a99dbaa57e..c59b0f91c87d 100644
--- a/drivers/clk/qcom/gcc-sc8280xp.c
+++ b/drivers/clk/qcom/gcc-sc8280xp.c
@@ -6786,7 +6786,7 @@ static struct gdsc pcie_2a_gdsc = {
 		.name = "pcie_2a_gdsc",
 	},
 	.pwrsts = PWRSTS_OFF_ON,
-	.flags = VOTABLE | ALWAYS_ON,
+	.flags = VOTABLE | PWRSTS_RET_ON,
 };
 
 static struct gdsc pcie_2b_gdsc = {
@@ -6797,7 +6797,7 @@ static struct gdsc pcie_2b_gdsc = {
 		.name = "pcie_2b_gdsc",
 	},
 	.pwrsts = PWRSTS_OFF_ON,
-	.flags = VOTABLE | ALWAYS_ON,
+	.flags = VOTABLE | PWRSTS_RET_ON,
 };
 
 static struct gdsc pcie_3a_gdsc = {
@@ -6808,7 +6808,7 @@ static struct gdsc pcie_3a_gdsc = {
 		.name = "pcie_3a_gdsc",
 	},
 	.pwrsts = PWRSTS_OFF_ON,
-	.flags = VOTABLE | ALWAYS_ON,
+	.flags = VOTABLE | PWRSTS_RET_ON,
 };
 
 static struct gdsc pcie_3b_gdsc = {
@@ -6819,7 +6819,7 @@ static struct gdsc pcie_3b_gdsc = {
 		.name = "pcie_3b_gdsc",
 	},
 	.pwrsts = PWRSTS_OFF_ON,
-	.flags = VOTABLE | ALWAYS_ON,
+	.flags = VOTABLE | PWRSTS_RET_ON,
 };
 
 static struct gdsc pcie_4_gdsc = {
@@ -6830,7 +6830,7 @@ static struct gdsc pcie_4_gdsc = {
 		.name = "pcie_4_gdsc",
 	},
 	.pwrsts = PWRSTS_OFF_ON,
-	.flags = VOTABLE | ALWAYS_ON,
+	.flags = VOTABLE | PWRSTS_RET_ON,
 };
 
 static struct gdsc ufs_card_gdsc = {
-- 
2.25.1

Re: [PATCH v2] clk: qcom: gcc-sc8280xp: Allow PCIe GDSCs to enter retention state

Posted by Johan Hovold 2 years, 7 months ago

On Thu, Jul 06, 2023 at 07:38:42PM +0530, Manivannan Sadhasivam wrote:
> With the minimal system suspend support in place for the PCIe driver that
> keeps the interconnect path voted, the ALWAYS_ON flag can now be dropped.
> 
> Also, the PWRSTS_RET_ON flag should be used to allow the GDSCs to enter the
> retention state when the parent domain get's turned off during system
> suspend.
> 
> Signed-off-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
> ---
> 
> Changes in v2:
> 
> * Changed the patch from simple revert to changing the ALWAYS_ON flag to
>   PWRSTS_RET_ON.
> 
>  drivers/clk/qcom/gcc-sc8280xp.c | 10 +++++-----
>  1 file changed, 5 insertions(+), 5 deletions(-)
> 
> diff --git a/drivers/clk/qcom/gcc-sc8280xp.c b/drivers/clk/qcom/gcc-sc8280xp.c
> index 04a99dbaa57e..c59b0f91c87d 100644
> --- a/drivers/clk/qcom/gcc-sc8280xp.c
> +++ b/drivers/clk/qcom/gcc-sc8280xp.c
> @@ -6786,7 +6786,7 @@ static struct gdsc pcie_2a_gdsc = {
>  		.name = "pcie_2a_gdsc",
>  	},
>  	.pwrsts = PWRSTS_OFF_ON,
> -	.flags = VOTABLE | ALWAYS_ON,
> +	.flags = VOTABLE | PWRSTS_RET_ON,
>  };

This is not correct either. PWRSTS_RET_ON is a pwrsts mask...

Johan

Re: [PATCH v2] clk: qcom: gcc-sc8280xp: Allow PCIe GDSCs to enter retention state

Posted by Manivannan Sadhasivam 2 years, 7 months ago

On Thu, Jul 06, 2023 at 04:17:06PM +0200, Johan Hovold wrote:
> On Thu, Jul 06, 2023 at 07:38:42PM +0530, Manivannan Sadhasivam wrote:
> > With the minimal system suspend support in place for the PCIe driver that
> > keeps the interconnect path voted, the ALWAYS_ON flag can now be dropped.
> > 
> > Also, the PWRSTS_RET_ON flag should be used to allow the GDSCs to enter the
> > retention state when the parent domain get's turned off during system
> > suspend.
> > 
> > Signed-off-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
> > ---
> > 
> > Changes in v2:
> > 
> > * Changed the patch from simple revert to changing the ALWAYS_ON flag to
> >   PWRSTS_RET_ON.
> > 
> >  drivers/clk/qcom/gcc-sc8280xp.c | 10 +++++-----
> >  1 file changed, 5 insertions(+), 5 deletions(-)
> > 
> > diff --git a/drivers/clk/qcom/gcc-sc8280xp.c b/drivers/clk/qcom/gcc-sc8280xp.c
> > index 04a99dbaa57e..c59b0f91c87d 100644
> > --- a/drivers/clk/qcom/gcc-sc8280xp.c
> > +++ b/drivers/clk/qcom/gcc-sc8280xp.c
> > @@ -6786,7 +6786,7 @@ static struct gdsc pcie_2a_gdsc = {
> >  		.name = "pcie_2a_gdsc",
> >  	},
> >  	.pwrsts = PWRSTS_OFF_ON,
> > -	.flags = VOTABLE | ALWAYS_ON,
> > +	.flags = VOTABLE | PWRSTS_RET_ON,
> >  };
> 
> This is not correct either. PWRSTS_RET_ON is a pwrsts mask...
> 

I should stop sending patches while feeling sleepy...

- Mani

> Johan

-- 
மணிவண்ணன் சதாசிவம்

Re: [PATCH v2] clk: qcom: gcc-sc8280xp: Allow PCIe GDSCs to enter retention state

Posted by Konrad Dybcio 2 years, 7 months ago

On 6.07.2023 16:17, Johan Hovold wrote:
> On Thu, Jul 06, 2023 at 07:38:42PM +0530, Manivannan Sadhasivam wrote:
>> With the minimal system suspend support in place for the PCIe driver that
>> keeps the interconnect path voted, the ALWAYS_ON flag can now be dropped.
>>
>> Also, the PWRSTS_RET_ON flag should be used to allow the GDSCs to enter the
>> retention state when the parent domain get's turned off during system
>> suspend.
>>
>> Signed-off-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
>> ---
>>
>> Changes in v2:
>>
>> * Changed the patch from simple revert to changing the ALWAYS_ON flag to
>>   PWRSTS_RET_ON.
>>
>>  drivers/clk/qcom/gcc-sc8280xp.c | 10 +++++-----
>>  1 file changed, 5 insertions(+), 5 deletions(-)
>>
>> diff --git a/drivers/clk/qcom/gcc-sc8280xp.c b/drivers/clk/qcom/gcc-sc8280xp.c
>> index 04a99dbaa57e..c59b0f91c87d 100644
>> --- a/drivers/clk/qcom/gcc-sc8280xp.c
>> +++ b/drivers/clk/qcom/gcc-sc8280xp.c
>> @@ -6786,7 +6786,7 @@ static struct gdsc pcie_2a_gdsc = {
>>  		.name = "pcie_2a_gdsc",
>>  	},
>>  	.pwrsts = PWRSTS_OFF_ON,
>> -	.flags = VOTABLE | ALWAYS_ON,
>> +	.flags = VOTABLE | PWRSTS_RET_ON,
>>  };
> 
> This is not correct either. PWRSTS_RET_ON is a pwrsts mask...
This + please rebase atop that:

https://lore.kernel.org/linux-arm-msm/20230620-topic-sc8280_gccgdsc-v2-0-562c1428c10d@linaro.org/

Konrad
> 
> Johan

Re: [PATCH v2] clk: qcom: gcc-sc8280xp: Allow PCIe GDSCs to enter retention state

Posted by Konrad Dybcio 2 years, 7 months ago

On 6.07.2023 16:40, Konrad Dybcio wrote:
> On 6.07.2023 16:17, Johan Hovold wrote:
>> On Thu, Jul 06, 2023 at 07:38:42PM +0530, Manivannan Sadhasivam wrote:
>>> With the minimal system suspend support in place for the PCIe driver that
>>> keeps the interconnect path voted, the ALWAYS_ON flag can now be dropped.
>>>
>>> Also, the PWRSTS_RET_ON flag should be used to allow the GDSCs to enter the
>>> retention state when the parent domain get's turned off during system
>>> suspend.
>>>
>>> Signed-off-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
>>> ---
>>>
>>> Changes in v2:
>>>
>>> * Changed the patch from simple revert to changing the ALWAYS_ON flag to
>>>   PWRSTS_RET_ON.
>>>
>>>  drivers/clk/qcom/gcc-sc8280xp.c | 10 +++++-----
>>>  1 file changed, 5 insertions(+), 5 deletions(-)
>>>
>>> diff --git a/drivers/clk/qcom/gcc-sc8280xp.c b/drivers/clk/qcom/gcc-sc8280xp.c
>>> index 04a99dbaa57e..c59b0f91c87d 100644
>>> --- a/drivers/clk/qcom/gcc-sc8280xp.c
>>> +++ b/drivers/clk/qcom/gcc-sc8280xp.c
>>> @@ -6786,7 +6786,7 @@ static struct gdsc pcie_2a_gdsc = {
>>>  		.name = "pcie_2a_gdsc",
>>>  	},
>>>  	.pwrsts = PWRSTS_OFF_ON,
>>> -	.flags = VOTABLE | ALWAYS_ON,
>>> +	.flags = VOTABLE | PWRSTS_RET_ON,
>>>  };
>>
>> This is not correct either. PWRSTS_RET_ON is a pwrsts mask...
> This + please rebase atop that:
> 
> https://lore.kernel.org/linux-arm-msm/20230620-topic-sc8280_gccgdsc-v2-0-562c1428c10d@linaro.org/
Moreover, I tried a similar patch out.

Johan's v6.4 branch + the patchset I linked + removing ALWAYS_ON from gcc
(so keeping PWRSTS_OFF_ON) and PCIe survives a wakeup-s2idle-wakeup cycle.

I am however running with clk_ignore_unused, so I don't know how this
impacts the RPMh system suspend situation.

Konrad

> 
> Konrad
>>
>> Johan