Re: [PATCH v4 02/15] ASoC: dt-bindings: qcom: q6dsp: add internal mi2s support

[Nickolay Goppen <setotau@mainlining.org>]

25.05.2026 14:51, Nickolay Goppen пишет:
>
> 14.05.2026 16:55, Krzysztof Kozlowski пишет:
>> On 03/05/2026 23:21, Richard Acayan wrote:
>>> On Sun, May 03, 2026 at 02:11:31PM +0200, Krzysztof Kozlowski wrote:
>>>> On Fri, May 01, 2026 at 11:31:15AM -0400, Richard Acayan wrote:
>>>>> The internal MI2S ports are found on devices with the internal sound
>>>>> card for Snapdragon 660. Add support for them.
>>>>>
>>>>> Signed-off-by: Richard Acayan <mailingradian@gmail.com>
>>>>> ---
>>>>>   .../bindings/sound/qcom,q6dsp-lpass-ports.yaml     |  4 ++--
>>>>>   include/dt-bindings/sound/qcom,q6dsp-lpass-ports.h | 14 
>>>>> ++++++++++++++
>>>>>   2 files changed, 16 insertions(+), 2 deletions(-)
>>>> Same feedback as last time. DO NOT send next versions, untill you 
>>>> reply
>>>> or address the feedback.
>>> Ok, I believe this is about [1] and [2] with reference to the patches
>>> already in v7.1-rc1 [3].
>>>
>>> My patches add INT0_MI2S_RX through INT6_MI2S_TX, which correspond to
>>> q6afe port IDs 0x102e - 0x103b. Srini's patches add LPI_MI2S_RX_0
>>> through LPI_MI2S_TX_4, but no q6afe port IDs.
>>>
>>> I asked if the LPI_MI2S ports have q6afe port IDs to check if we're
>>> referring to the same ports, but apparently q6afe is too old to have it
>>> [4]:
>>>
>>>     > Is the LPI MI2S on Q6AFE? I don't see the AFE port ID.
>>>     I think the older SOCs did not have this I guess,
>>>
>>> Even if there's already an allocation of dt-bindings ports for 
>>> LPI_MI2S,
>>> they can't just be reused for INT_MI2S. There are allegedly INT5_MI2S
>>> and INT6_MI2S (each RX and TX) but LPI_MI2S only counts to 4.
>> Then what is the difference between INT and LPI? People stuff here some
>> hardware names but this is SW ABI, so MI2S_RX_0 is only one. Just like
>> we have DISPLAY_PORT_RX_0, but no LPI_DISPLAY_PORT_RX_0 and
>> INT_DISPLAY_PORT_RX_0.
resending since this message failed to deliver.
> Then do you mean that LPI_MI2S ports need to have port ID's assigned 
> like Richard done that for INT_MI2S? +Srini, Is this a good idea?
>> Best regards,
>> Krzysztof
>
-- 
Best regards,
Nickolay