2009/6/11 Daniel Ribeiro drwyrm@gmail.com:
Em Qui, 2009-06-11 às 21:34 +0800, Eric Miao escreveu:
We already use FSRT for DSP_A, and if this works on littleton I2S we should just stick with FSRT (and frame_width = sample_width * channels) to keep the code simple.
I hope so, but the assumption of frame_width == sample_width * 2 should hold true first.
Ok, here is what I think that should work for I2S after my 2 patches to sort the TDM thing.
2 Patches are inline, first version assumes that frame_width = sample_width * 2(channels), and just increases the SFRM duration to emulate the LRCLK.
Second version uses Eric and Paul code only for pxa3xx. In this version, frame_width = sample_width * 2(channels) * 2(envelope).
This was only compile tested, I dont have PXA3XX hardware to test this. It applies after the 3 patches I sent earlier.
Well, I'm completely lost in this thread. Can anyone give a summary on this issue? And it looks like set_tdm_slot() is used to generalize the issue of envelop (or the actual frame/sample width), and the DAI format setting code here will be generalized??
Daniel,
Could you please send out all the four patches? Sorry late on this, busy with the merging stuff.