Re: [PATCH] ASoC: qcom: sc8280xp: enable primary mi2s

[Luca Weiss]

On Tue Feb 4, 2025 at 9:19 AM CET, neil.armstrong wrote:
> On 04/02/2025 00:27, Luca Weiss wrote:
>> On Mon Feb 3, 2025 at 5:33 PM CET, Danila Tikhonov wrote:
>>> On 03/02/2025 17:23, neil.armstrong@xxxxxxxxxx wrote:
>>>> On 03/02/2025 12:38, Danila Tikhonov wrote:
>>>>> When using primary mi2s on sc8280xp-compatible SoCs, the correct clock
>>>>> needs to get enabled to be able to use the mi2s interface.
>>>>>
>>>>> Signed-off-by: Danila Tikhonov <danila@xxxxxxxxxxx>
>>>>> ---
>>>>>    sound/soc/qcom/sc8280xp.c | 13 +++++++++++++
>>>>>    1 file changed, 13 insertions(+)
>>>>>
>>>>> diff --git a/sound/soc/qcom/sc8280xp.c b/sound/soc/qcom/sc8280xp.c
>>>>> index 311377317176..03687de1ebb0 100644
>>>>> --- a/sound/soc/qcom/sc8280xp.c
>>>>> +++ b/sound/soc/qcom/sc8280xp.c
>>>>> @@ -14,6 +14,8 @@
>>>>>    #include "common.h"
>>>>>    #include "sdw.h"
>>>>>    +#define MI2S_BCLK_RATE    1536000
>>>>> +
>>>>>    struct sc8280xp_snd_data {
>>>>>        bool stream_prepared[AFE_PORT_MAX];
>>>>>        struct snd_soc_card *card;
>>>>> @@ -25,13 +27,24 @@ struct sc8280xp_snd_data {
>>>>>      static int sc8280xp_snd_init(struct snd_soc_pcm_runtime *rtd)
>>>>>    {
>>>>> +    unsigned int codec_dai_fmt = SND_SOC_DAIFMT_BC_FC;
>>>>> +    unsigned int fmt = SND_SOC_DAIFMT_BP_FP;
>>>>>        struct sc8280xp_snd_data *data =
>>>>> snd_soc_card_get_drvdata(rtd->card);
>>>>> +    struct snd_soc_dai *codec_dai = snd_soc_rtd_to_codec(rtd, 0);
>>>>>        struct snd_soc_dai *cpu_dai = snd_soc_rtd_to_cpu(rtd, 0);
>>>>>        struct snd_soc_card *card = rtd->card;
>>>>>        struct snd_soc_jack *dp_jack  = NULL;
>>>>>        int dp_pcm_id = 0;
>>>>>          switch (cpu_dai->id) {
>>>>> +    case PRIMARY_MI2S_RX:
>>>>> +        codec_dai_fmt |= SND_SOC_DAIFMT_NB_NF | SND_SOC_DAIFMT_I2S;
>>>>> +        snd_soc_dai_set_sysclk(cpu_dai,
>>>>> +            Q6AFE_LPASS_CLK_ID_PRI_MI2S_IBIT,
>>>>> +            MI2S_BCLK_RATE, SNDRV_PCM_STREAM_PLAYBACK);
>>>>
>>>> How is this possible ? sc8280xp uses the q6prm clock driver, and
>>>> there's no
>>>> way this call sets the Q6PRM_LPASS_CLK_ID_PRI_MI2S_IBIT, or I totally
>>>> missed
>>>> something.
>>>>
>>>> And prm is neither a dai nor has the set_sysclk callback.
>>>>
>>>> Neil
>>>>
>>> Oh, thanks for the answer.
>>>
>>> This comes from qcm6490 compatibility. Actually it is needed for
>>> Nothing Phone (1) and Luca Weiss just suggested me to use
>>> qcom,qcm6490-idp-sndcard like do they do it for FP5. (SM7325 is the
>>> closest to QCM6490):
>>> https://github.com/sc7280-mainline/linux/pull/5#discussion_r1848984788
>>> Actually I also think it is a bit incorrect.
>> 
>> For reference, this is coming from this and following:
>> https://lore.kernel.org/linux-arm-msm/e8a24709-de96-4d09-ba00-1e084a656c68@xxxxxxxxxx/
>
> Sure, but do you use sound/soc/qcom/sc8280xp.c ? it's designed for audioreach
> compatible audio architecture (post sm8250).

I would've also thought qcm6490-idp and qcm6490-rb3gen2 are
pre-audioreach, so sm8250.c would be a better match - but these two
devices already use sc8280xp.c driver for some reason?

Maybe someone with more insight on these devices can help clarify this?
I'll also need this for qcm6490-fairphone-fp5 soon.

Regards Luca

>
> sound/soc/qcom/sm8250.c has already support for PRIMARY_MI2S_RX via
> the Q6AFE_LPASS_CLK_ID_PRI_MI2S_IBIT, and it compatible with the previous
> audio architecture (non audioreach).
>
> Neil
>
>> 
>> Regards
>> Luca
>> 
>>>
>>> ---
>>> Regards,
>>> Danila
>>>>> +        snd_soc_dai_set_fmt(cpu_dai, fmt);
>>>>> +        snd_soc_dai_set_fmt(codec_dai, codec_dai_fmt);
>>>>> +        break;
>>>>>        case WSA_CODEC_DMA_RX_0:
>>>>>        case WSA_CODEC_DMA_RX_1:
>>>>>            /*
>>>>
>>