Some boards (e.g. sdm845-samsung-starqltechn) provide a cx-supply
reference for the SLPI PAS.

The Linux driver unconditionally tries getting "cx" and "px" supplies,
so it actually is used.

Fixes: 3d447dcdae53 ("dt-bindings: remoteproc: qcom,adsp: Make msm8974 use CX as power domain")
Signed-off-by: Luca Weiss <luca@lucaweiss.eu>
---
There's literally one board using this upstream, judging from that I'm
not sure this is a misuse of cx-supply or what exactly. An alternative
to this patch is of course removing the usage in
sdm845-samsung-starqltechn, but as it stands right now the patch under
"Fixes" introduces a dtbs_check warning.
---
 Documentation/devicetree/bindings/remoteproc/qcom,adsp.yaml | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/Documentation/devicetree/bindings/remoteproc/qcom,adsp.yaml b/Documentation/devicetree/bindings/remoteproc/qcom,adsp.yaml
index 137f95028313..bde138716873 100644
--- a/Documentation/devicetree/bindings/remoteproc/qcom,adsp.yaml
+++ b/Documentation/devicetree/bindings/remoteproc/qcom,adsp.yaml
@@ -32,6 +32,9 @@ properties:
   reg:
     maxItems: 1
 
+  cx-supply:
+    description: Phandle to the CX regulator
+
   px-supply:
     description: Phandle to the PX regulator
 

---
base-commit: 37bb2e7217b01404e2abf9d90d8e5705a5603b52
change-id: 20251207-adsp-cx-fixup-442882486ec5

Best regards,
-- 
Luca Weiss <luca@lucaweiss.eu>

On Sun, 07 Dec 2025 19:35:35 +0900, Luca Weiss wrote:
> Some boards (e.g. sdm845-samsung-starqltechn) provide a cx-supply
> reference for the SLPI PAS.
> 
> The Linux driver unconditionally tries getting "cx" and "px" supplies,
> so it actually is used.
> 
> Fixes: 3d447dcdae53 ("dt-bindings: remoteproc: qcom,adsp: Make msm8974 use CX as power domain")
> Signed-off-by: Luca Weiss <luca@lucaweiss.eu>
> ---
> There's literally one board using this upstream, judging from that I'm
> not sure this is a misuse of cx-supply or what exactly. An alternative
> to this patch is of course removing the usage in
> sdm845-samsung-starqltechn, but as it stands right now the patch under
> "Fixes" introduces a dtbs_check warning.
> ---
>  Documentation/devicetree/bindings/remoteproc/qcom,adsp.yaml | 3 +++
>  1 file changed, 3 insertions(+)
> 

Acked-by: Rob Herring (Arm) <robh@kernel.org>

On 12/7/25 11:35 AM, Luca Weiss wrote:
> Some boards (e.g. sdm845-samsung-starqltechn) provide a cx-supply
> reference for the SLPI PAS.
> 
> The Linux driver unconditionally tries getting "cx" and "px" supplies,
> so it actually is used.
> 
> Fixes: 3d447dcdae53 ("dt-bindings: remoteproc: qcom,adsp: Make msm8974 use CX as power domain")
> Signed-off-by: Luca Weiss <luca@lucaweiss.eu>
> ---
> There's literally one board using this upstream, judging from that I'm
> not sure this is a misuse of cx-supply or what exactly. An alternative
> to this patch is of course removing the usage in
> sdm845-samsung-starqltechn, but as it stands right now the patch under
> "Fixes" introduces a dtbs_check warning.

FWIW that's likely a hack (because IIUC it needs to power up some
regulator for the sensor devices to work) but that's "fine"

I don't know if there's a better way to handle it though.. I think it's
a board design "issue", since the DSP has access to some GPIOs but it
seems like that wasn't utilized

I'm open to suggestions

+Dzmitry-the-starqltechn-submitter

Konrad

пн, 8 дек. 2025 г. в 19:33, Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>:
>
> On 12/7/25 11:35 AM, Luca Weiss wrote:
> > Some boards (e.g. sdm845-samsung-starqltechn) provide a cx-supply
> > reference for the SLPI PAS.
> >
> > The Linux driver unconditionally tries getting "cx" and "px" supplies,
> > so it actually is used.
> >
> > Fixes: 3d447dcdae53 ("dt-bindings: remoteproc: qcom,adsp: Make msm8974 use CX as power domain")
> > Signed-off-by: Luca Weiss <luca@lucaweiss.eu>
> > ---
> > There's literally one board using this upstream, judging from that I'm
> > not sure this is a misuse of cx-supply or what exactly. An alternative
> > to this patch is of course removing the usage in
> > sdm845-samsung-starqltechn, but as it stands right now the patch under
> > "Fixes" introduces a dtbs_check warning.
>
> FWIW that's likely a hack (because IIUC it needs to power up some
> regulator for the sensor devices to work) but that's "fine"
>

I confirm this probably powers up devices, because firmware boots fine
without cx-supply.

-- 

Best regards,
Dzmitry