[PATCH 16/17] ALSA: ARM: add Raumfeld audio support

[broonie@opensource.wolfsonmicro.com (Mark Brown)]

On Wed, Nov 25, 2009 at 01:24:36PM +0100, Daniel Mack wrote:
> On Wed, Nov 25, 2009 at 11:02:18AM +0000, Mark Brown wrote:

> > > +static struct i2c_board_info max9486_hwmon_info = {
> > > +	I2C_BOARD_INFO("max9485", 0x63),
> > > +};

> > This should be in the board file under arch/arm.

> Well, the problem is that if I do it there, I don't get a handle for the
> actual data transfer, which I now get from i2c_new_device(). There is no
> driver matching this device (it wasn't taken because it's 'too simple').

> How would I get a handle to pass to i2c_master_send() or an equivalent
> function?

In the same way as any other I2C driver would?

> > > +extern void raumfeld_enable_audio(bool en);

> > This should be in a proper header file somewhere.

> Hmm, I thought so too, but it would be the only thing to add there.
> Hence I decided to not do that. You really prefer that?

Well, see my other comments once I found the actual implementation of
the function but yes - bad practice is bad practice.

> > Remove this and the other empty functions.

> Hmm, I tried that and it crashed the kernel. I will check again as that
> was some month ago.

If this were required then almost all machine drivers would be buggy...

> > look like a boolean but that's not what's really going on here.  It may
> > be better to do this as a proper driver, there was at least one driver I
> > remember being posted for a TDM clock generator which I think got
> > merged.

> No, it wasn't.

There was at least one unrelated driver for a different part - I
remember the discussion since someone needed to explain to people (Alan
Cox, I think) that this wasn't an RTC but rather a TDM clock generator.

>   http://marc.info/?l=linux-i2c&m=122457836326525&w=2

> Jean Delvare's last comment on this was:

> > Honestly I don't see any value in this driver. There's nothing you can
> > do with it that you couldn't already do without it.

> The driver itself would do the right thing, but I doubt that
> resubmitting will help much.

Looking at the thread in the archive I don't see any effort to answer
Jean's question there - the reply from Jon talks about device tree
binding which is, as Jean says, pretty much irrelevant to the question:

   http://marc.info/?l=linux-i2c&m=122465761327694&w=2

The question seemed to be more about what the driver was supposed to
accomplish - talking about the functionality is provided by the driver
once it's bound to the device should address that.