[alsa-devel] [PATCH 1/4] ASoC: pxa-ssp: enhance I2S and add Left_J support

[Mark Brown]

On Mon, Jun 08, 2009 at 01:03:20PM -0300, Daniel Ribeiro wrote:
> Em Seg, 2009-06-08 às 13:40 +0100, Mark Brown escreveu:

> > I'm inclined to go with sample here since it seems harder for the
> > machine drivers to get wrong but I've not really thought it through yet?

> But sample wouldn't cover all cases. eg, magician and ezx uses the same
> sample size, but magician needs 16bit frames and ezx 32bits frames.

See my reply to Philip - even with frame size his case is going to be
very hard to cover in a standard fashion since it's not clear what to do
when your frame clock is run too fast relative to your bit clock.  On
the receive side a lot of devices are just going to discard incomplete
stereo frames.

> Just to confirm that I'm understanding the terminology here..
> Sample size is SNDRV_PCM_FORMAT_*, and frame size is the actual distance
> in clocks between each SSPFRM assertion. Is this correct?

That's pretty much my understanding - sample size is the number of bits
clocked per mono sample on the wire and frame size is the number of bits
clocked per cycle of the frame clock.

> > OTOH I don't really see much difference between the two cases - it's
> > just an extra couple of parameters on the end of the call.  That said,

> I agree with pHilipp, I don't need to set network mode for normal
> operation, but i want to be able to use real network mode.

Network mode is just a detail of the implementation of the PXA here - it
should not be visible outside the pxa-ssp driver.  Or to put it another
way setting a tdm_slot of 1, 1 ought to result in the same behaviour as
disabling TDM as far as the user is concerned.

> Using set_tdm_slot to setup this also seems awkward, because the dma
> configuration to use is also affected by the frame size.

I'd not expect it to be?

> And finally, set_tdm_mode doesn't cover all use cases, I may want to
> receive on a timeslot and keep the TX line high impedance, or I may want
> to use a different timeslot for TX...

Yes, set_tdm_mode() needs separate RX and TX configuration regardless -
if we added the framing configuration in there we should definitely make
that change at that point.

> My suggestion here is to make a clear distinction between normal mode
> and network mode. We should not abuse SSCR0_MOD when we don't need it.
> We should not require the machine driver to call set_tdm_slot when it
> doesn't need to. And for frame sizes that really need SSCR0_MOD to be
> set we should issue a warning telling that network mode was
> "automatically" enabled.

I'd like to see all these details handled within the driver - knowing if
and when network mode are to be set up is the sort of thing that users
ought to be able to rely on the driver for.