Received-SPF: pass (google.com: domain of linux-kernel+bounces-84400-linux.lists.archive=gmail.com@vger.kernel.org designates 139.178.88.99 as permitted sender) client-ip=139.178.88.99;
Message-ID: <e5d3d76f-32a3-4378-8d94-06f762ffb11f@quicinc.com>
Date: Wed, 28 Feb 2024 10:03:50 +0800
Precedence: bulk
MIME-Version: 1.0
User-Agent: Mozilla Thunderbird
From: "Aiqun Yu (Maria)" <quic_aiquny@quicinc.com>
Subject: Re: [PATCH v4 1/6] dt-bindings: arm: qcom: Document QCM8550, QCS8550
 SoC and board
To: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
CC: Tengfei Fan <quic_tengfan@quicinc.com>, <andersson@kernel.org>,
        <konrad.dybcio@linaro.org>, <robh+dt@kernel.org>,
        <krzysztof.kozlowski+dt@linaro.org>, <conor+dt@kernel.org>,
        <linux-arm-msm@vger.kernel.org>, <devicetree@vger.kernel.org>,
        <linux-kernel@vger.kernel.org>, <kernel@quicinc.com>
References: <20240119100621.11788-1-quic_tengfan@quicinc.com>
 <20240119100621.11788-2-quic_tengfan@quicinc.com>
 <CAA8EJprpMjK03rKPK6wgfVuDvBikYsKZjMc0Wusa1BxFOBnXhQ@mail.gmail.com>
 <86672501-206a-49ed-8af7-2b6c332c1697@quicinc.com>
 <CAA8EJppkDDACV_sLxFW4EqKQLHfo4ivSLwa_jCde8JpeH4YfzA@mail.gmail.com>
 <88845d8d-0b58-46fa-8c52-c1bb4f2d39f8@quicinc.com>
 <CAA8EJpr_vqwpf89GvyAUYJwcrT2mZxwWUkkqmDEOGH0Lci_YYw@mail.gmail.com>
Content-Language: en-US
In-Reply-To: <CAA8EJpr_vqwpf89GvyAUYJwcrT2mZxwWUkkqmDEOGH0Lci_YYw@mail.gmail.com>
Content-Type: text/plain; charset="UTF-8"; format=flowed
Content-Transfer-Encoding: 7bit


On 2/20/2024 5:15 PM, Dmitry Baryshkov wrote:
> On Tue, 20 Feb 2024 at 11:09, Aiqun Yu (Maria) <quic_aiquny@quicinc.com> wrote:
>>
>>
>>
>> On 2/5/2024 9:48 PM, Dmitry Baryshkov wrote:
>>> On Mon, 5 Feb 2024 at 12:21, Tengfei Fan <quic_tengfan@quicinc.com> wrote:
>>>>
>>>>
>>>>
>>>> On 2/5/2024 12:29 AM, Dmitry Baryshkov wrote:
>>>>> On Fri, 19 Jan 2024 at 11:07, Tengfei Fan <quic_tengfan@quicinc.com> wrote:
>>>>>>
>>>>>> Document QCM8550, QCS8550 SoC and the AIM300 AIoT board bindings.
>>>>>> QCS8550 and QCM8550 processor combines powerful computing, extreme edge
>>>>>> AI processing, Wi-Fi 7, and robust video and graphics for a wide range
>>>>>> of use cases for the Internet of Things (IoT). QCS8550 is a QCS version
>>>>>> for QCM8550. Modem RF only in QCM8550 but not in QCS8550.
>>>>>> AIM300 Series is a highly optimized family of modules designed to
>>>>>> support AIoT applications. The module is mounted onto Qualcomm AIoT
>>>>>> carrier board to support verification, evaluation and development. It
>>>>>> integrates QCS8550 SoC, UFS and PMIC chip etc.
>>>>>> AIM stands for Artificial Intelligence Module. AIoT stands for AI IoT.
>>>>>>
>>>>>> Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
>>>>>> Signed-off-by: Tengfei Fan <quic_tengfan@quicinc.com>
>>>>>> ---
>>>>>>     Documentation/devicetree/bindings/arm/qcom.yaml | 11 +++++++++++
>>>>>>     1 file changed, 11 insertions(+)
>>>>>>
>>>>>> diff --git a/Documentation/devicetree/bindings/arm/qcom.yaml b/Documentation/devicetree/bindings/arm/qcom.yaml
>>>>>> index 1a5fb889a444..9cee874a8eae 100644
>>>>>> --- a/Documentation/devicetree/bindings/arm/qcom.yaml
>>>>>> +++ b/Documentation/devicetree/bindings/arm/qcom.yaml
>>>>>> @@ -49,8 +49,10 @@ description: |
>>>>>>             msm8996
>>>>>>             msm8998
>>>>>>             qcs404
>>>>>> +        qcs8550
>>>>>>             qcm2290
>>>>>>             qcm6490
>>>>>> +        qcm8550
>>>>>
>>>>> Drop
>>>>
>>>> we want to introduce qcm8550 here.
>>>
>>> What for. It either had to be introduced beforehand, or it should be
>>> introduced when one adds support for an actual qcm8550 device.
>>>
>>>> qcm8550.dtsi has been introduced and qcs8550-aim300.dtsi include
>>>> qcm8550.dtsi directly.
>>>>
>>>> qcs8550 is a QCS version for qcm8550. qcs8550 is a sub-series of
>>>> qcm8550. qcm8550 will be a firmware release series from qualcomm.
>>>
>>> All three names refer to the different kinds of the same platform. The
>>> base chip name is sm8550, so it is the last one. Other than that,
>>> there is no need to include any SoC compatibles other than the actual
>>> SoC name. See existing qrb devices for an inspiration.
>>>
>>>>
>>>> here is the qcm8550/qcs8550 detailed spec:
>>>> https://docs.qualcomm.com/bundle/publicresource/87-61717-1_REV_A_Qualcomm_QCS8550_QCM8550_Processors_Product_Brief.pdf
>>>>
>>>> here is the sm8550 detailed spec:
>>>> https://docs.qualcomm.com/bundle/publicresource/87-71408-1_REV_C_Snapdragon_8_gen_3_Mobile_Platform_Product_Brief.pdf
>>>
>>> Can you please summarise the _actual_ difference between qcm8550,
>>> qcs8550 and sm8550? Are they fully soft compatible? Soft compatible
>>> except the modem? Pin compatible?
>>
>> we can remove the qcm8550 compatible for now, and rename current dtsi
>> back for qcs8550.dtsi, and only introduce qcm8550 later when there is
>> qcm8550 board public-ed.
> 
> Yes, please.
> 
>>
>>   From software point of view, currently it is single firmware image
>> release for both qcm8550 and qcs8550, and the firmware is not grantee
>> for sm8550 software compatible.
> 
> I assume that modem.mbn is different for qcm and qcs devices. Or does
> qcs completely miss the modem DSP?
You are right, modem.mbn part is different. QCS have a gps only modem image.
> 
>>
>>   From hardware point of view, qcm8550, qcs8550, sm8550 are different
>> hardware socs, qcm8550, qcs8550 is derived from sm8550. We can only
>> share the public document about those soc descriptions [1]. For soc
>> itself, it is all similar difference for QCS and QCM version.
>> Currently(in current development stage) there is not notable software
>> difference needed other than memory map in kernel side needed to be
>> differentiate from qcm8550 qcs8550 to sm8550.
>>
>> [1]
>> https://docs.qualcomm.com/bundle/publicresource/87-61717-1_REV_A_Qualcomm_QCS8550_QCM8550_Processors_Product_Brief.pdf
>>
>>>
>>>>
>>>>>
>>>>>>             qdu1000
>>>>>>             qrb2210
>>>>>>             qrb4210
>>>>>> @@ -93,6 +95,7 @@ description: |
>>>>>>       The 'board' element must be one of the following strings:
>>>>>>
>>>>>>             adp
>>>>>> +        aim300-aiot
>>>>>
>>>>> We probably need to drop this list, it doesn't surve its purposes.
>>>>
>>>> I am a little confused, do you expect to just remove this "aim300-aiot"
>>>> or do you want to introduce a new patch and remove the whole list?
>>>
>>> If you were following the list, you would have seen the patch
>>> reworking the bindings.
>>>
>>>>
>>>>>
>>>>>>             cdp
>>>>>>             dragonboard
>>>>>>             idp
>>>>>> @@ -904,6 +907,14 @@ properties:
>>>>>>               - const: qcom,qcs404-evb
>>>>>>               - const: qcom,qcs404
>>>>>>
>>>>>> +      - items:
>>>>>> +          - enum:
>>>>>> +              - qcom,qcs8550-aim300-aiot
>>>>>> +          - const: qcom,qcs8550-aim300
>>>>>> +          - const: qcom,qcs8550
>>>>>> +          - const: qcom,qcm8550
>>>>>
>>>>> In the review comments for v3 you have been asked to add qcom,sm8550.
>>>>> But not the qcom,qcm8550. I don't think that there is any need to
>>>>> mention qcm8550 here.
>>>>
>>>> qcm8550 and sm8550 are different, they have different firmware release.
>>>>
>>>> AIM300 AIoT board depend on qcs8550, qcs8550 is a QCS version for
>>>> qcm8550. Modem RF only in qcm8550 but not in qcs8550.
>>>
>>> There are no 'dependecies' here. The thing is about declaring compatibility.
>>> In my opinion, the qcm8550 is an unnecesary part of the equation. You
>>> declare compatibility with the board itself,  with the SoM, with the
>>> actual SoC and with the base of the series. Anybody caring for the
>>> difference between QCM, QCS and SM will have to check for both
>>> qcom,qcs8550 and qcom,qcm8550 anyway, as there are differences on the
>>> modem side.
>>>
>>>>>> +          - const: qcom,sm8550
>>>>>> +
>>>>>>           - items:
>>>>>>               - enum:
>>>>>>                   - qcom,sa8155p-adp
>>>>>> --
>>>
>>
>> --
>> Thx and BRs,
>> Aiqun(Maria) Yu
> 
> 
> 

-- 
Thx and BRs,
Aiqun(Maria) Yu