[bardliao]

@plbossart @slawblauciak This is a RFC to explain my idea how multi dai work on kernel side. I verified that sdw_params_stream() will be invoked probably. The only issue I met is that sdw stream didn't support multi cpu yet. ( I haven't tried #1802 yet )

[bardliao]

@plbossart I just tested it with #1802. Only rt1308-1 is enabled. I will check what's going on.

[bardliao]

@plbossart I just tested it with #1802. Only rt1308-1 is enabled. I will check what's going on.

Sorry, I forget to update my machine driver. After updating my machine driver, both ALH0x102 and ALH0x202 are enabled. But I will see inconsistent state error on sdw_eneble_stream() with or without #1802. I will look into it. We may still miss something on kernel side.

[slawblauciak]

Needs a stream map. Also, DAI configurations for the children, unless those are supposed to be passes separately.

[bardliao]

Now I can play with multi cpu without error with some hacks. My test branch is https://github.com/bardliao/linux/commits/multi-dai-test

dapm_on.txt is the list of dapm widgets which state is On. We can see that two rt1308 amplifiers are on and ALH0x102.OUT is on. I can hear left channel, but not right channel. @slawblauciak Is it possible to copy the data from BUF3 to ALH0x202.OUT? ie. Copy the data from sdw link 1 to sdw link 2.
I tried connecting ALH0x202 to the same PCM as ALH0x102 on topology. But I met DSP Panic if I do so.
After today's experiment, I think we may not need multi dai on FW side. I believe that I can get the same result without multi dai.

[ranj063]

@bardliao the PR looks fine. What topology file did you use for your test?

[plbossart]

@bardliao I think the discussion for today is that we will go with the multi-cpu DAI, add a demux and check later if we need further changes for pause_push/pause_release cases.
In short we don't need the multi-dai changes for now, so you probably want to close this PR?

[bardliao]

Yes, we don't need multi dai :)