On Wed, 2009-11-04 at 18:55 +0000, Graeme Gregory wrote:
On Wed, Nov 04, 2009 at 05:53:55PM +0000, Liam Girdwood wrote:
From: Graeme Gregory gg@slimlogic.co.uk
This patch increases the number of supported audio channels from 4 to 16 and was sponsored by Shotspotter inc.
Signed-off-by: Graeme Gregory gg@slimlogic.co.uk Signed-off-by: Liam Girdwood lrg@slimlogic.co.uk
- /* Default div to 1 if it wasn't set by machine driver, otherwise
* use set div as the maximum clock value*/- div = mcbsp_data->clk_div ? mcbsp_data->clk_div : 1;
- /* calc best frame size for rate and clock divider */
- do {
frame_size = (mcbsp_data->in_freq / div) / params_rate(params);pr_debug("freq %d, rate %d, frame size %d, div %d\n",mcbsp_data->in_freq, params_rate(params), frame_size, div);if (frame_size > 256)div++;- } while (frame_size > 256);
- /* Check we can fit the requested number of channels into our
* calculated frame size*/- if ((channels * wlen) > frame_size) {
printk(KERN_ERR"OMAP-MCBSP: cannot fit channels in frame size\n");return -EINVAL;- }
- /* Set the actual clkdiv to use for this samplerate */
- regs->srgr1 &= ~CLKGDV(0xFF);
- regs->srgr1 |= CLKGDV(div - 1);
This chunk changes the semantics of other devices which I have never tested.
I also dont know how much damage it does if it does to slave mode. In fact I think it might break it in cases which are actually allowable as it uses the omap as its clock constraint and not the clock source.
Btw, this has been reworked to avoid all I2S paths since you last worked on it.
So it looks like we need a conditional around this block for when the codec is mastering BCLK/FRM and MCBSP is slave (as you stated above). This will then use the original SRGR2 values for SND_SOC_DAIFMT_DSP_A/B in codec master mode.
V2 will follow.
Liam