Re: [PATCH 0/7] MSM8953/SDM632 rpmpd/mss fixes

Re: [PATCH 0/7] MSM8953/SDM632 rpmpd/mss fixes

[Dmitry Baryshkov]

On Fri, Mar 27, 2026 at 09:11:42PM +0100, Barnabás Czémán wrote:
> SDM632 pm domains are different from MSM8953 because MSM8953
> is defining pm8953_s1 as regulator but SDM632 is defining it
> as pm domain.
> 
> This patch series correcting the pm domains defined in rpmpd
> driver and splitting the MSS resources for the both SoC to match
> the reality.
> 
> These changes was discussed in a previous threads:
> https://lore.kernel.org/all/2b057aa5-4416-4fd4-aeab-6bc23acbb53d@oss.qualcomm.com/

Was 4.9 never expected to work on MSM8953? Or did it require firmware
changes?

> 
> Signed-off-by: Barnabás Czémán <barnabas.czeman@mainlining.org>
> ---

-- 
With best wishes
Dmitry

Re: [PATCH 4/7] arm64: dts: qcom: msm8953: fix modem pm domains

[Dmitry Baryshkov]

On Fri, Mar 27, 2026 at 09:11:46PM +0100, Barnabás Czémán wrote:
> MSM8953 MSS is using mss-supply as regulator.

"On MSM8953 MSS regulators is controlled using the voltages rather than
performance levels. Correct DT definition and model the MSS as a
regulator rather than a power domain".

Also please squash with the changes actually making use of the
regulator.

> 
> Signed-off-by: Barnabás Czémán <barnabas.czeman@mainlining.org>
> ---
>  arch/arm64/boot/dts/qcom/msm8953.dtsi | 5 ++---
>  1 file changed, 2 insertions(+), 3 deletions(-)
> 

-- 
With best wishes
Dmitry

Re: [PATCH 0/7] MSM8953/SDM632 rpmpd/mss fixes

[Barnabás Czémán]

On 2026-03-27 21:33, Dmitry Baryshkov wrote:
> On Fri, Mar 27, 2026 at 09:11:42PM +0100, Barnabás Czémán wrote:
>> SDM632 pm domains are different from MSM8953 because MSM8953
>> is defining pm8953_s1 as regulator but SDM632 is defining it
>> as pm domain.
>> 
>> This patch series correcting the pm domains defined in rpmpd
>> driver and splitting the MSS resources for the both SoC to match
>> the reality.
>> 
>> These changes was discussed in a previous threads:
>> https://lore.kernel.org/all/2b057aa5-4416-4fd4-aeab-6bc23acbb53d@oss.qualcomm.com/
> 
> Was 4.9 never expected to work on MSM8953? Or did it require firmware
> changes?
There is no official 4.9 kernel for MSM8953 devices, all of them are 
using 3.18.
There are unofficial 4.9 ports for those devices but all of them 
defining s1 as
a regulator so the regulators were changed back like it was on 3.18.
> 
>> 
>> Signed-off-by: Barnabás Czémán <barnabas.czeman@mainlining.org>
>> ---

Re: [PATCH 0/7] MSM8953/SDM632 rpmpd/mss fixes

[Dmitry Baryshkov]

On Fri, Mar 27, 2026 at 09:38:20PM +0100, Barnabás Czémán wrote:
> On 2026-03-27 21:33, Dmitry Baryshkov wrote:
> > On Fri, Mar 27, 2026 at 09:11:42PM +0100, Barnabás Czémán wrote:
> > > SDM632 pm domains are different from MSM8953 because MSM8953
> > > is defining pm8953_s1 as regulator but SDM632 is defining it
> > > as pm domain.
> > > 
> > > This patch series correcting the pm domains defined in rpmpd
> > > driver and splitting the MSS resources for the both SoC to match
> > > the reality.
> > > 
> > > These changes was discussed in a previous threads:
> > > https://lore.kernel.org/all/2b057aa5-4416-4fd4-aeab-6bc23acbb53d@oss.qualcomm.com/
> > 
> > Was 4.9 never expected to work on MSM8953? Or did it require firmware
> > changes?
> There is no official 4.9 kernel for MSM8953 devices, all of them are using
> 3.18.
> There are unofficial 4.9 ports for those devices but all of them defining s1
> as
> a regulator so the regulators were changed back like it was on 3.18.

Ack, thanks. I was looking at the msm-4.4 tree from the sdm660 branches,
which has msm8953.dtsi. Thanks for the confirmation.

> > 
> > > 
> > > Signed-off-by: Barnabás Czémán <barnabas.czeman@mainlining.org>
> > > ---

-- 
With best wishes
Dmitry

Re: [PATCH 1/7] dt-bindings: power: qcom-rpmpd: Split MSM8953 and SDM632

[Barnabás Czémán]

On 2026-03-27 21:26, Dmitry Baryshkov wrote:
> On Fri, Mar 27, 2026 at 09:11:43PM +0100, Barnabás Czémán wrote:
>> Remove modem related bindings from MSM8953 rpmpd because MSM8953 MSS
>> is using mss-supply as a regulator usually it is pm8953_s1.
>> Split SDM632 bindings from MSM8953 because SDM632 is using mss-supply
>> as a pm domain.
>> 
>> Signed-off-by: Barnabás Czémán <barnabas.czeman@mainlining.org>
>> ---
>>  .../devicetree/bindings/power/qcom,rpmpd.yaml        |  1 +
>>  include/dt-bindings/power/qcom-rpmpd.h               | 20 
>> +++++++++++++-------
>>  2 files changed, 14 insertions(+), 7 deletions(-)
>> 
>> diff --git a/Documentation/devicetree/bindings/power/qcom,rpmpd.yaml 
>> b/Documentation/devicetree/bindings/power/qcom,rpmpd.yaml
>> index 8174ceeab572..659936d6a46e 100644
>> --- a/Documentation/devicetree/bindings/power/qcom,rpmpd.yaml
>> +++ b/Documentation/devicetree/bindings/power/qcom,rpmpd.yaml
>> @@ -48,6 +48,7 @@ properties:
>>            - qcom,sc7280-rpmhpd
>>            - qcom,sc8180x-rpmhpd
>>            - qcom,sc8280xp-rpmhpd
>> +          - qcom,sdm632-rpmpd
>>            - qcom,sdm660-rpmpd
>>            - qcom,sdm670-rpmhpd
>>            - qcom,sdm845-rpmhpd
>> diff --git a/include/dt-bindings/power/qcom-rpmpd.h 
>> b/include/dt-bindings/power/qcom-rpmpd.h
>> index 4371ac941f29..2d82434b993c 100644
>> --- a/include/dt-bindings/power/qcom-rpmpd.h
>> +++ b/include/dt-bindings/power/qcom-rpmpd.h
>> @@ -84,13 +84,11 @@
>>  #define QM215_VDDMX_AO		MSM8917_VDDMX_AO
>> 
>>  /* MSM8953 Power Domain Indexes */
>> -#define MSM8953_VDDMD		0
>> -#define MSM8953_VDDMD_AO	1
>> -#define MSM8953_VDDCX		2
>> -#define MSM8953_VDDCX_AO	3
>> -#define MSM8953_VDDCX_VFL	4
>> -#define MSM8953_VDDMX		5
>> -#define MSM8953_VDDMX_AO	6
>> +#define MSM8953_VDDCX		RPMPD_VDDCX
>> +#define MSM8953_VDDCX_AO	RPMPD_VDDCX_AO
>> +#define MSM8953_VDDCX_VFL	RPMPD_VDDCX_VFL
>> +#define MSM8953_VDDMX		RPMPD_VDDMX
>> +#define MSM8953_VDDMX_AO	RPMPD_VDDMX_AO
> 
> Well, no. This is an ABI break. It will make previous DT to stop from
> working. You can drop unused indices, but you can not change the values
> used by the existing domains.
Do these indices never can be changed?
> 
>> 
>>  /* MSM8974 Power Domain Indexes */
>>  #define MSM8974_VDDCX		0
>> @@ -156,6 +154,14 @@
>>  #define QCS404_LPIMX		5
>>  #define QCS404_LPIMX_VFL	6
>> 
>> +/* SDM632 Power Domain Indexes */
>> +#define SDM632_VDDMD		0
>> +#define SDM632_VDDCX		1
>> +#define SDM632_VDDCX_AO		2
>> +#define SDM632_VDDCX_VFL	3
>> +#define SDM632_VDDMX		4
>> +#define SDM632_VDDMX_AO		5
> 
> Please use RPMHPD_* instead of introducing new entries.
I do not understand completely, should I use RPHPD bindings in rpmpd 
driver or
I should use rpmhpd driver for SDM632?
> 
>> +
>>  /* SDM660 Power Domains */
>>  #define SDM660_VDDCX		RPMPD_VDDCX
>>  #define SDM660_VDDCX_AO		RPMPD_VDDCX_AO
>> 
>> --
>> 2.53.0
>> 

Re: [PATCH 1/7] dt-bindings: power: qcom-rpmpd: Split MSM8953 and SDM632

[Krzysztof Kozlowski]

On Fri, Mar 27, 2026 at 09:11:43PM +0100, Barnabás Czémán wrote:
> Remove modem related bindings from MSM8953 rpmpd because MSM8953 MSS
> is using mss-supply as a regulator usually it is pm8953_s1.
> Split SDM632 bindings from MSM8953 because SDM632 is using mss-supply
> as a pm domain.
> 
> Signed-off-by: Barnabás Czémán <barnabas.czeman@mainlining.org>
> ---
>  .../devicetree/bindings/power/qcom,rpmpd.yaml        |  1 +
>  include/dt-bindings/power/qcom-rpmpd.h               | 20 +++++++++++++-------
>  2 files changed, 14 insertions(+), 7 deletions(-)
> 
> diff --git a/Documentation/devicetree/bindings/power/qcom,rpmpd.yaml b/Documentation/devicetree/bindings/power/qcom,rpmpd.yaml
> index 8174ceeab572..659936d6a46e 100644
> --- a/Documentation/devicetree/bindings/power/qcom,rpmpd.yaml
> +++ b/Documentation/devicetree/bindings/power/qcom,rpmpd.yaml
> @@ -48,6 +48,7 @@ properties:
>            - qcom,sc7280-rpmhpd
>            - qcom,sc8180x-rpmhpd
>            - qcom,sc8280xp-rpmhpd
> +          - qcom,sdm632-rpmpd
>            - qcom,sdm660-rpmpd
>            - qcom,sdm670-rpmhpd
>            - qcom,sdm845-rpmhpd
> diff --git a/include/dt-bindings/power/qcom-rpmpd.h b/include/dt-bindings/power/qcom-rpmpd.h
> index 4371ac941f29..2d82434b993c 100644
> --- a/include/dt-bindings/power/qcom-rpmpd.h
> +++ b/include/dt-bindings/power/qcom-rpmpd.h
> @@ -84,13 +84,11 @@
>  #define QM215_VDDMX_AO		MSM8917_VDDMX_AO
>  
>  /* MSM8953 Power Domain Indexes */
> -#define MSM8953_VDDMD		0

ABI break / impact and due to two changes combined I don't really
understand why. Why MSS using mss-supply makes this ABI invalid/wrong?

> -#define MSM8953_VDDMD_AO	1
> -#define MSM8953_VDDCX		2
> -#define MSM8953_VDDCX_AO	3
> -#define MSM8953_VDDCX_VFL	4
> -#define MSM8953_VDDMX		5
> -#define MSM8953_VDDMX_AO	6
> +#define MSM8953_VDDCX		RPMPD_VDDCX
> +#define MSM8953_VDDCX_AO	RPMPD_VDDCX_AO
> +#define MSM8953_VDDCX_VFL	RPMPD_VDDCX_VFL
> +#define MSM8953_VDDMX		RPMPD_VDDMX
> +#define MSM8953_VDDMX_AO	RPMPD_VDDMX_AO

I don't see how this is related to new compatible and SDM632.

>  
>  /* MSM8974 Power Domain Indexes */
>  #define MSM8974_VDDCX		0
> @@ -156,6 +154,14 @@
>  #define QCS404_LPIMX		5
>  #define QCS404_LPIMX_VFL	6

Best regards,
Krzysztof

Re: [PATCH 1/7] dt-bindings: power: qcom-rpmpd: Split MSM8953 and SDM632

[Dmitry Baryshkov]

On Sat, Mar 28, 2026 at 09:22:19AM +0100, Barnabás Czémán wrote:
> On 2026-03-27 21:26, Dmitry Baryshkov wrote:
> > On Fri, Mar 27, 2026 at 09:11:43PM +0100, Barnabás Czémán wrote:
> > > Remove modem related bindings from MSM8953 rpmpd because MSM8953 MSS
> > > is using mss-supply as a regulator usually it is pm8953_s1.
> > > Split SDM632 bindings from MSM8953 because SDM632 is using mss-supply
> > > as a pm domain.
> > > 
> > > Signed-off-by: Barnabás Czémán <barnabas.czeman@mainlining.org>
> > > ---
> > >  .../devicetree/bindings/power/qcom,rpmpd.yaml        |  1 +
> > >  include/dt-bindings/power/qcom-rpmpd.h               | 20
> > > +++++++++++++-------
> > >  2 files changed, 14 insertions(+), 7 deletions(-)
> > > 
> > > diff --git a/Documentation/devicetree/bindings/power/qcom,rpmpd.yaml
> > > b/Documentation/devicetree/bindings/power/qcom,rpmpd.yaml
> > > index 8174ceeab572..659936d6a46e 100644
> > > --- a/Documentation/devicetree/bindings/power/qcom,rpmpd.yaml
> > > +++ b/Documentation/devicetree/bindings/power/qcom,rpmpd.yaml
> > > @@ -48,6 +48,7 @@ properties:
> > >            - qcom,sc7280-rpmhpd
> > >            - qcom,sc8180x-rpmhpd
> > >            - qcom,sc8280xp-rpmhpd
> > > +          - qcom,sdm632-rpmpd
> > >            - qcom,sdm660-rpmpd
> > >            - qcom,sdm670-rpmhpd
> > >            - qcom,sdm845-rpmhpd
> > > diff --git a/include/dt-bindings/power/qcom-rpmpd.h
> > > b/include/dt-bindings/power/qcom-rpmpd.h
> > > index 4371ac941f29..2d82434b993c 100644
> > > --- a/include/dt-bindings/power/qcom-rpmpd.h
> > > +++ b/include/dt-bindings/power/qcom-rpmpd.h
> > > @@ -84,13 +84,11 @@
> > >  #define QM215_VDDMX_AO		MSM8917_VDDMX_AO
> > > 
> > >  /* MSM8953 Power Domain Indexes */
> > > -#define MSM8953_VDDMD		0
> > > -#define MSM8953_VDDMD_AO	1
> > > -#define MSM8953_VDDCX		2
> > > -#define MSM8953_VDDCX_AO	3
> > > -#define MSM8953_VDDCX_VFL	4
> > > -#define MSM8953_VDDMX		5
> > > -#define MSM8953_VDDMX_AO	6
> > > +#define MSM8953_VDDCX		RPMPD_VDDCX
> > > +#define MSM8953_VDDCX_AO	RPMPD_VDDCX_AO
> > > +#define MSM8953_VDDCX_VFL	RPMPD_VDDCX_VFL
> > > +#define MSM8953_VDDMX		RPMPD_VDDMX
> > > +#define MSM8953_VDDMX_AO	RPMPD_VDDMX_AO
> > 
> > Well, no. This is an ABI break. It will make previous DT to stop from
> > working. You can drop unused indices, but you can not change the values
> > used by the existing domains.
> Do these indices never can be changed?

You can add new indices and you can (with some care) drop existing
incorrecr or unused ones. You can't reassign indices though. The rule of
thumb is that old DTs should continue to work without rebuilding.

> > 
> > > 
> > >  /* MSM8974 Power Domain Indexes */
> > >  #define MSM8974_VDDCX		0
> > > @@ -156,6 +154,14 @@
> > >  #define QCS404_LPIMX		5
> > >  #define QCS404_LPIMX_VFL	6
> > > 
> > > +/* SDM632 Power Domain Indexes */
> > > +#define SDM632_VDDMD		0
> > > +#define SDM632_VDDCX		1
> > > +#define SDM632_VDDCX_AO		2
> > > +#define SDM632_VDDCX_VFL	3
> > > +#define SDM632_VDDMX		4
> > > +#define SDM632_VDDMX_AO		5
> > 
> > Please use RPMHPD_* instead of introducing new entries.
> I do not understand completely, should I use RPHPD bindings in rpmpd driver
> or
> I should use rpmhpd driver for SDM632?

Sorry, I meant RPMPD_*

> > 
> > > +
> > >  /* SDM660 Power Domains */
> > >  #define SDM660_VDDCX		RPMPD_VDDCX
> > >  #define SDM660_VDDCX_AO		RPMPD_VDDCX_AO
> > > 
> > > --
> > > 2.53.0
> > > 

-- 
With best wishes
Dmitry

Re: [PATCH 1/7] dt-bindings: power: qcom-rpmpd: Split MSM8953 and SDM632

[Konrad Dybcio]

On 3/28/26 9:22 AM, Barnabás Czémán wrote:
> On 2026-03-27 21:26, Dmitry Baryshkov wrote:
>> On Fri, Mar 27, 2026 at 09:11:43PM +0100, Barnabás Czémán wrote:
>>> Remove modem related bindings from MSM8953 rpmpd because MSM8953 MSS
>>> is using mss-supply as a regulator usually it is pm8953_s1.
>>> Split SDM632 bindings from MSM8953 because SDM632 is using mss-supply
>>> as a pm domain.
>>>
>>> Signed-off-by: Barnabás Czémán <barnabas.czeman@mainlining.org>
>>> ---
>>>  .../devicetree/bindings/power/qcom,rpmpd.yaml        |  1 +
>>>  include/dt-bindings/power/qcom-rpmpd.h               | 20 +++++++++++++-------
>>>  2 files changed, 14 insertions(+), 7 deletions(-)
>>>
>>> diff --git a/Documentation/devicetree/bindings/power/qcom,rpmpd.yaml b/Documentation/devicetree/bindings/power/qcom,rpmpd.yaml
>>> index 8174ceeab572..659936d6a46e 100644
>>> --- a/Documentation/devicetree/bindings/power/qcom,rpmpd.yaml
>>> +++ b/Documentation/devicetree/bindings/power/qcom,rpmpd.yaml
>>> @@ -48,6 +48,7 @@ properties:
>>>            - qcom,sc7280-rpmhpd
>>>            - qcom,sc8180x-rpmhpd
>>>            - qcom,sc8280xp-rpmhpd
>>> +          - qcom,sdm632-rpmpd
>>>            - qcom,sdm660-rpmpd
>>>            - qcom,sdm670-rpmhpd
>>>            - qcom,sdm845-rpmhpd
>>> diff --git a/include/dt-bindings/power/qcom-rpmpd.h b/include/dt-bindings/power/qcom-rpmpd.h
>>> index 4371ac941f29..2d82434b993c 100644
>>> --- a/include/dt-bindings/power/qcom-rpmpd.h
>>> +++ b/include/dt-bindings/power/qcom-rpmpd.h
>>> @@ -84,13 +84,11 @@
>>>  #define QM215_VDDMX_AO        MSM8917_VDDMX_AO
>>>
>>>  /* MSM8953 Power Domain Indexes */
>>> -#define MSM8953_VDDMD        0
>>> -#define MSM8953_VDDMD_AO    1
>>> -#define MSM8953_VDDCX        2
>>> -#define MSM8953_VDDCX_AO    3
>>> -#define MSM8953_VDDCX_VFL    4
>>> -#define MSM8953_VDDMX        5
>>> -#define MSM8953_VDDMX_AO    6
>>> +#define MSM8953_VDDCX        RPMPD_VDDCX
>>> +#define MSM8953_VDDCX_AO    RPMPD_VDDCX_AO
>>> +#define MSM8953_VDDCX_VFL    RPMPD_VDDCX_VFL
>>> +#define MSM8953_VDDMX        RPMPD_VDDMX
>>> +#define MSM8953_VDDMX_AO    RPMPD_VDDMX_AO
>>
>> Well, no. This is an ABI break. It will make previous DT to stop from
>> working. You can drop unused indices, but you can not change the values
>> used by the existing domains.
> Do these indices never can be changed?

Yes, values in include/dt-bindings are supposed to never change

Here you're e.g. changing the "msm8953 domain 0" from translating into
VDDMD to translating into VDDCX (because RPMPD_VDDCX is defined as 0)

Some other older platforms were converted to use these macros, because
the indices happened to match

Konrad