[alsa-devel] [PATCH 2/5] [RFC]intel_hdmi_audio: interface module

[Takashi Iwai]

At Mon, 22 Nov 2010 13:58:12 +0000,
Mark Brown wrote:
> 
> On Mon, Nov 22, 2010 at 07:09:43PM +0530, ramesh.babu at intel.com wrote:
> 
> > +/**
> > + * snd_intelhad_dev_free - to free alsa card instance
> > + *
> > + * @device: pointer to device
> > + *
> > + * This function is called when the hdmi cable is un plugged
> > + */
> > +static int snd_intelhad_dev_free(struct snd_device *device)
> > +{
> > +	struct snd_intelhad *intelhaddata;
> > +
> > +	BUG_ON(!device);
> > +	pr_debug("had: snd_intelhad_dev_free called\n");
> > +	intelhaddata = device->device_data;
> > +	return 0;
> > +}
> 
> This function doesn't appear to do anything like what either the
> description or name would suggest?  I'd expect to see some deallocation
> going on...

Reading through the code, the construct doesn't seem to allocate any
private stuff but only standard components, so this callback doesn't
have to do anything much, I suppose.

> > +/**
> > + * had_hot_plug - to create sound card instance for HDMI audio playabck
> > + *
> > + *
> > + * This function is called when the hdmi cable is plugged in. This function
> > + * creates and registers the sound card with ALSA
> > + */
> > +static int had_hot_plug(void)
> > +{
> 
> It would seem more natural for the HDMI device to be created and
> instatiated via the normal Linux device model - when a HDMI connection
> is detected the thing doing the detection creates and registers a new
> device of some kind which the driver core then binds to a separate
> driver.

Yeah, the driver structure here is somehow wacky, although it's not
to hard to trace.


Takashi