[alsa-devel] [PATCH 5/5] ASoC: tpa6130a2: Make DAPM registration optional, and direct interface

[Peter Ujfalusi]

Hi,

On Wednesday 01 December 2010 08:54:20 Ujfalusi Peter (Nokia-MS/Tampere) wrote:
> On Tuesday 30 November 2010 16:30:13 ext Mark Brown wrote:
> > On Tue, Nov 30, 2010 at 04:00:04PM +0200, Peter Ujfalusi wrote:
> > > Users can choose to not add the DAPM routes provided by the
> > > amp driver, but use the direct enable/disable interface
> > > from machine driver with SND_SOC_DAPM_HP's event callback.
> > > In some cases this method must be used to make the audio
> > > path pop noise free.
> > 
> > Is there any situation where it would undesirable to do this?  If not
> > it'd seem better to just make the driver do this always.
> 
> You mean to not have DAPM widgets/routes in the tpa6130a2 driver, and only
> have a function, which can be used to turn on/off the amp?
> The original [1] (first version) of the tpa6130a2 driver only had DAPM_HP
> widget. It has been changed based on the comments.

I think the simplest thing would be to actually remove all DAPM widgets, 
routings from this driver, and provide only the funtion, which enable the amp in 
stereo mode.
As it is now, we do not have user for this amp in upstream, so I'm not that 
worried about braking machine drivers ;)
The problem with tpa6130a2 provided DAPM_HP is that at least in one machine, I 
have to introduce delay before I enable the amp, otherwise I have really bad pop 
at stream start. The codec in this case alone does not produce that bad pop, but 
it seams the sequence, and timing in the whole system makes the pop bad.
On other setups I have not encountered such a problem.

So:
I would remove all dapm related things from the tpa6130a2 driver, and offer only 
the:
int tpa6130a2_stereo_enable(snd_soc_codec, *codec, int enable);
Function for users (machine drivers).

If there's a need I can re-introduce tpa6130a2 provided DAPM_HP widget to reduce 
duplicated code in machine drivers (when we have enough users for it).

What do you think?

-- 
Péter