Antonio Ospite wrote:
I attach the patch ported to the new code.
SND_SOC_DAIFMT_MSB is for I2S emulation in MSB mode, shouldn't the
format be called SND_SOC_DAIFMT_I2S_MSB?
Yes, but there is no DAIFMT_I2S_MSB on soc.h. It will work with
I2S_DAIFMT_I2S too, but the audio channels will be swapped.
case SND_SOC_DAIFMT_I2S:
+ SSPSP_P(port) |= SSPSP_FSRT;
This is an OR and there's actually no 'break' here, right?
Right! The only difference between I2S and MSB is that the first has
SSPSP_FSRT set.
case SND_SOC_DAIFMT_MSB:
- /* handled above */
+ SSCR0_P(port) |= SSCR0_EDSS | SCCR0_DataSize(16);
+ SSPSP_P(port) |= SSPSP_SFRMWDTH(16);
break;
+ ret = cpu_dai->dai_ops.set_fmt(cpu_dai, SND_SOC_DAIFMT_MSB |
SND_SOC_DAIFMT_IB_IF | SND_SOC_DAIFMT_CBM_CFM);
^ ^
AFAICS the Ms refer to codec point of view, so we are setting pxa as
slave here, right?
Exactly! PXA is slave when SCLKDIR and SFRMDIR are set.
--
Daniel Ribeiro
