Re: [PATCH V1 3/3] ALSA: hda: add CIX IPBLOQ HDA controller support

[Takashi Iwai <tiwai@suse.de>]

On Fri, 31 Oct 2025 10:05:01 +0100,
Krzysztof Kozlowski wrote:
> 
> On Thu, Oct 30, 2025 at 07:09:28PM +0800, joakim.zhang@cixtech.com wrote:
> > From: Joakim Zhang <joakim.zhang@cixtech.com>
> > +	if (err) {
> > +		dev_err(chip->card->dev,
> > +			"unable to request IRQ %d, disabling device\n",
> 
> Why are you requesting resources outside of probe, in work item? You
> cannot handle here deferred probe.
> 
> How is it supposed to work? How deferred probe would work?

The deferred probe is used for snd-hda-intel as it may require either
the vga_switcheroo binding or the firmware loading before the actual
probe.

OTOH, in this driver case, *_hda_init() call might be better as the
part of the probe callback.  Though, we may still need some deferred
work for probing and configuring the codecs, i.e. azx_probe_codecs()
and azx_codec_configure() calls where explicit module bindings are
involved.

I guess snd-hda-tegra does like snd-hda-intel just because it followed
the code flow of snd-hda-intel.  But it doesn't mean that a new driver
should follow the same, too.  Doing the initialization in the probe
callback would be much easier to handle the errors, for example.


thanks,

Takashi