Hi Daniel,
On 04/09/2014 03:12 PM, Daniel Mack wrote:
Hi Peter,
On 04/04/2014 06:38 PM, Peter Ujfalusi wrote:
On 04/04/2014 04:24 PM, Daniel Mack wrote:
The cleanups look sane, though. I can test them on AM33xx based hardware early next week. If you don't want to hold them off until then, no problem. I can also send fixups in case I spot a regression.
That would be great if you could also test these. It seams you have quite good array of codecs available.
I tested this series with a CS4271 codec in I2S mode, and things still work as expected, except for the fact that the LRCLK is now inverted in comparison to the old state.
Yes, this is expected. I think you were asking for NB_IF with the old code, right? The old code was - wrongly IMHO - configured the I2S mode's FS in inverted mode: NB_NF would end up to send the left channel when the FS is high. This is not correct. NB_NF means in I2S is that left channel is when FS is low, right channel is when FS is high. Most of the daVinci device used DSP_B, which was correct in terms of FS. The NB_IF with cs4271 was fine since the codec driver did not checked for invert it only cares about the protocol. Page 20 of the cs4271 shows the I2S. With this series we can use the correct DAIFMT_I2S + NB_NF with standard compliant codecs.
After some debugging, however, I have confidence that the behaviour with the patches applied is in fact correct, we just need to flip the bits in our ASoC machine driver, which is no big deal.
Thanks for the cleanups again!
Thanks for testing it!