mbox series

[0/2] IPA support on qcm2290

Message ID 20241220073540.37631-1-wojciech.slenska@gmail.com
Headers show
Series IPA support on qcm2290 | expand

Message

Wojciech Slenska Dec. 20, 2024, 7:35 a.m. UTC 
Patches added IPA support for qcm2290.
Configuration is based on sc7180, becouse both has the same
IPA version 4.2.

Wojciech Slenska (2):
  dt-bindings: net: qcom,ipa: document qcm2290 compatible
  arm64: dts: qcom: qcm2290: Add IPA nodes

 .../devicetree/bindings/net/qcom,ipa.yaml     |  4 ++
 arch/arm64/boot/dts/qcom/qcm2290.dtsi         | 52 +++++++++++++++++++
 2 files changed, 56 insertions(+)

Comments

Krzysztof Kozlowski Dec. 21, 2024, 8:44 p.m. UTC | #1 
On 20/12/2024 08:35, Wojciech Slenska wrote:
> Document that ipa on qcm2290 uses version 4.2, the same
> as sc7180.
> 
> Signed-off-by: Wojciech Slenska <wojciech.slenska@gmail.com>
> ---
>  Documentation/devicetree/bindings/net/qcom,ipa.yaml | 4 ++++
>  1 file changed, 4 insertions(+)
> 
> diff --git a/Documentation/devicetree/bindings/net/qcom,ipa.yaml b/Documentation/devicetree/bindings/net/qcom,ipa.yaml
> index 53cae71d9957..ea44d02d1e5c 100644
> --- a/Documentation/devicetree/bindings/net/qcom,ipa.yaml
> +++ b/Documentation/devicetree/bindings/net/qcom,ipa.yaml
> @@ -58,6 +58,10 @@ properties:
>            - enum:
>                - qcom,sm8650-ipa
>            - const: qcom,sm8550-ipa
> +      - items:
> +          - enum:
> +              - qcom,qcm2290-ipa
> +          - const: qcom,sc7180-ipa
>  
We usually keep such lists between each other ordered by fallback, so
this should go before sm8550-fallback-list.

With that change:

Acked-by: Krzysztof Kozlowski <krzk@kernel.org>

Best regards,
Krzysztof
Alex Elder Dec. 24, 2024, 3:11 p.m. UTC | #2 
On 12/20/24 7:25 AM, Konrad Dybcio wrote:
> On 20.12.2024 8:35 AM, Wojciech Slenska wrote:
>> Document that ipa on qcm2290 uses version 4.2, the same
>> as sc7180.
>>
>> Signed-off-by: Wojciech Slenska <wojciech.slenska@gmail.com>
>> ---
> 
> FWIW this needs some more work on the Linux side, the IPA driver
> currently hardcodes a reference to IMEM, which has a different
> base between these two SoCs.

I have only glanced at this so far.  At the moment I don't
know whether this device uses a different range in IMEM, but
Konrad's message suggests it does.  And if so, he's correct:
the IMEM range needs to be defined differently (perhaps in
Device Tree) so that different SoCs using the same version
of IPA but different IMEM ranges can function correctly.

Downstream code should be consulted to determine this, and
as of now I have not done that.

					-Alex

> The IMEM region doesn't seem to be used as of current, but things
> will explode the second it is.
> 
> A long overdue update would be to make the IPA driver consume
> a syscon/memory-region-like property pointing to IMEM (or a slice
> of it, maybe Alex knows what it was supposed to be used for).
> 
> Konrad