> > > set_tdm_slot()
>
> > We will register multiple codecs inside a single dai_link and differentiate these
> > codecs by of_node. And the adc_slot and the dac_slot may be different, they can
> > be decided by the user.If we use set_tdm_slot,the adc_slot and the dac_slot will
> > be same.
>
> No, the machine driver should be configuring different TDM slots for
> each device - that's the whole point of the API.
We are using multiple codecs as a single device.So we can't use set_tdm_slot to configure
different slots for multiple codecs under one device.
> > > > + ret = clk_prepare_enable(es8389->mclk);
> > > > + if (ret) {
> > > > + dev_err(codec->dev, "%s, unable to enable mclk\n", __func__);
> > > > + return ret;
> > > > + }
> > > > + }
>
> > > Making the use of a MCLK depend on the configuration of a TDM slot for
> > > the ADC seems *very* unusual, what's going on there?
>
> > Because we are associating multiple codecs under a single dai_link, we will be
> > executing probe and init many times during initialization.But MCLK only needs
> > to be used once.So we decided making the use of a MCLK depend on the configuration
> > of a TDM slot for the ADC
>
> No, each device should just get and enable the MCLK itself - the clock
> API does reference counting so there's no problem with this, it's normal
> for a clok to have multiple consumers.
ok,I'll fix it
