linux-kernel - Re: [PATCH 2/2] arm64: dts: qcom: sm8750-mtp: Add sound (speakers, headset codec, dmics)

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <3a29e34c-d286-4673-adac-1fd8627c3eff@oss.qualcomm.com>
Date: Wed, 30 Apr 2025 12:48:24 +0200
From: Konrad Dybcio <konrad.dybcio@....qualcomm.com>
To: Krzysztof Kozlowski <krzysztof.kozlowski@...aro.org>,
        Bjorn Andersson <andersson@...nel.org>,
        Konrad Dybcio <konradybcio@...nel.org>, Rob Herring <robh@...nel.org>,
        Krzysztof Kozlowski <krzk+dt@...nel.org>,
        Conor Dooley <conor+dt@...nel.org>
Cc: linux-arm-msm@...r.kernel.org, devicetree@...r.kernel.org,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH 2/2] arm64: dts: qcom: sm8750-mtp: Add sound (speakers,
 headset codec, dmics)

On 4/30/25 8:19 AM, Krzysztof Kozlowski wrote:
> On 29/04/2025 21:11, Konrad Dybcio wrote:
>> On 4/28/25 4:41 PM, Krzysztof Kozlowski wrote:
>>> On 25/04/2025 11:30, Konrad Dybcio wrote:
>>>> On 4/24/25 11:40 AM, Krzysztof Kozlowski wrote:
>>>>> Add device nodes for most of the sound support - WSA883x smart speakers,
>>>>> WCD9395 audio codec (headset) and sound card - which allows sound
>>>>> playback via speakers and recording via DMIC microphones.  Changes bring
>>>>> necessary foundation for headset playback/recording via USB, but that
>>>>> part is not yet ready.
>>>>>
>>>>> Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@...aro.org>
>>>>> ---
>>>>
>>>> [...]
>>>>
>>>>> +	sound {
>>>>> +		compatible = "qcom,sm8750-sndcard", "qcom,sm8450-sndcard";
>>>>> +		model = "SM8750-MTP";
>>>>> +		audio-routing = "SpkrLeft IN", "WSA_SPK1 OUT",
>>>>> +				"SpkrRight IN", "WSA_SPK2 OUT",
>>>>> +				"IN1_HPHL", "HPHL_OUT",
>>>>> +				"IN2_HPHR", "HPHR_OUT",
>>>>> +				"AMIC2", "MIC BIAS2",
>>>>> +				"VA DMIC0", "MIC BIAS3", /* MIC4 on schematics */
>>>>> +				"VA DMIC1", "MIC BIAS3", /* MIC1 on schematics */
>>>>
>>>> Is this a mistake in what the codec driver exposes, or just a fumble
>>>> in numbering $somewhere?
>>>
>>> Which mistake? MIC4? Schematics call name things differently. They
>>> always were, so to make it clear for people without schematics I wrote
>>> which MIC it actually is.
>>
>> I'm not sure how to parse your response
>>
>> are you saying that there are MIC[0..4] that are/may be connected
>> to different codec ports, and that the MIC4/1 lines are plumbed to
>> VA DMIC0/1 respectively?
> 
> Yes, as always. Nothing weird here.
> 
>>
>> I think I got confused about the MIC BIAS3 going to both and none
> 
> What is both and none?

missing Oxford comma I suppose.. MIC BIAS3 goes to both MIC4 and MIC1
and neither of them has a '3' in the name. I was wondering whether
that's intentional.

> 
>> matching the index, but perhaps that's just because it comes from
>> the WCD (which is the third piece of hw involved beyond VA and the
>> mic itself)
> 
> Again, what is the mistake you are pointing here?

I'm not necessarily saying this is a mistake, just thinking out loud
how I understand the non-obvious plumbing

[...]

>>>>> +		/*
>>>>> +		 * WCD9395 RX Port 1 (HPH_L/R)      <=> SWR1 Port 1 (HPH_L/R)
>>>>> +		 * WCD9395 RX Port 2 (CLSH)         <=> SWR1 Port 2 (CLSH)
>>>>> +		 * WCD9395 RX Port 3 (COMP_L/R)     <=> SWR1 Port 3 (COMP_L/R)
>>>>> +		 * WCD9395 RX Port 4 (LO)           <=> SWR1 Port 4 (LO)
>>>>> +		 * WCD9395 RX Port 5 (DSD_L/R)      <=> SWR1 Port 5 (DSD_L/R)
>>>>> +		 * WCD9395 RX Port 6 (HIFI_PCM_L/R) <=> SWR1 Port 9 (HIFI_PCM_L/R)
>>>>> +		 */
>>>>> +		qcom,rx-port-mapping = <1 2 3 4 5 9>;
>>>>
>>>> Does this deserve some dt-bindings constants?
>>>
>>> No, because these are hardware details/constants. Drivers do not use them.
>>
>> I'd argue it makes sense here - it makes more sense to pass meaningfully
>> named constants to the driver, rather than blobs with a comment
> 
> Sense of what? You want to make it a binding then answer what does it
> bind, what part of ABI for driver is here a binding (answer none:
> because driver does not use it)?

Sense of the magic numbers that otherwise require a comment.

dt-bindings don't exclusively contain enums-turned-defines that are
indices of C arrays - some contain various forms of hardware ABI, be
it register addresses, or names for magic values

Konrad