Hi Sergios!

Your changes look good so far. I've never tested the recording with the ar9331, so I cannot guarantee, that the setup is correct.

Which pin are you using for recording? It should be GPIO22 on the ar9331.

Please also check the i2s lines with an oscilloscope, check especially if the master clock and the input data lines are OK.

from ar9331-i2s-alsa.

Thanks Franz

You're right, I'm already using GPIO22 and the oscilloscope shows right values:
MCK=11.28MHz
CK=1.41MHz
WS=44.1KHz
BTW, the I2S input it is not showing any activity (always 0), but as far I know, this may be due to some switch/I2C in the codec, I guess that this has no influence in the overrun errors...

As far I understand it may be related to DMA interruption or so.

from ar9331-i2s-alsa.

Yes it is possible that there is something wrong in the RX path of the driver.

Could you please also try to do the following:

arecord -f cd /tmp/test.wav

Maybe the overrun errors are occuring because the data cannot be streamed as fast as the DMA is producing them.

from ar9331-i2s-alsa.

ummm! I've made a test a seems to work (but I don't understand your argumentation).

Let me activate the codec input in order to get I2S_DATA_IN working... I'll be back in a moment (I hope)

from ar9331-i2s-alsa.

That's great! That means the alsa driver is OK, also with recording.

I can explain the issue if you want:

when you do "arecord -f cd" the output will be streamed to stdout. However if stdout is too slow to print the data, the DMA on the receive side will run into a buffer overflow. The same issue could also occure if you record to the SPI flash, which is also not very fast.

When recording to /tmp the data will be streamed directly into the RAM and this is the fastest memory you have on the ar9331.

Regards,
Franz

from ar9331-i2s-alsa.

Hi Franz

Thank you for your explanation. It makes sense

Sergio

from ar9331-i2s-alsa.