Re: [PATCH v2 5/6] ASoC: qcom: sm8250: shut down MI2S/TDM AFE port clocks

On 5/7/26 5:57 AM, Srinivas Kandagatla wrote:
> On 5/6/26 8:33 PM, Val Packett wrote:
>> These port clocks were never being turned off, leading to increased
>> power consumption after a sound was played through any of these ports
>> for the first time. Use enable counters to disable the clocks in the
>> shutdown callback, similar to how it's done for sdm845.
>>
>> Signed-off-by: Val Packett <[email protected]>
>> ---
>>   sound/soc/qcom/sm8250.c | 150 +++++++++++++++++++++++++++++++---------
>>   1 file changed, 119 insertions(+), 31 deletions(-)
>>
>> diff --git a/sound/soc/qcom/sm8250.c b/sound/soc/qcom/sm8250.c
>> index b64fd3970ba1..9f0f846a14bb 100644
>> --- a/sound/soc/qcom/sm8250.c
>> +++ b/sound/soc/qcom/sm8250.c
>> @@ -24,6 +24,7 @@ static unsigned int tdm_slot_offset[8] = {0, 4, 8, 12, 16, 20, 24, 28};
>>   
>>   struct sm8250_snd_data {
>>   	bool stream_prepared[AFE_PORT_MAX];
>> +	uint32_t clk_count[AFE_PORT_MAX];
> In your setup are you sharing BE dais to multiple FE?
>
> if not you would really not need the count, startup and shutdown are
> paired for single PCM stream.
>
> If yes, then you need some protection against these counters.

Hmm, possibly not? I don't entirely understand how the q6 hardware 
works, sorry.. Who's the BE in this situation?

Basically what we have is e.g.:

- a dai-link between speakers (<&aw88261_l>, <&aw88261_r>) and the 
digital interface they're connected to (<&q6afedai SENARY_MI2S_RX>) 
defined in the device tree
- one q6asmdai frontend (MultiMedia1) connected to it in the mixer, i.e. 
cset name='QUIN_MI2S_RX Audio Mixer MultiMedia1' 1
- VoiceMMode1 (currently only via an out-of-tree hacky patchset but 
someone's reworking that) also connected to it like that during voice 
calls, but we can definitely "unplug" the MultiMedia when that happens

> [..]

Thanks,
~val