[alsa-devel] [PATCH 15/15] ASoC: hdac_hdmi: Keep display active while enumerating codec

[Takashi Iwai]

On Tue, 01 Dec 2015 17:48:43 +0100,
Babu, Ramesh wrote:
> 
> > On Tue, 01 Dec 2015 22:48:41 +0100,
> > Subhransu S. Prusty wrote:
> > >
> > > On Tue, Dec 01, 2015 at 01:57:01PM +0100, Takashi Iwai wrote:
> > > > On Tue, 01 Dec 2015 18:47:11 +0100,
> > > > Subhransu S. Prusty wrote:
> > > > >
> > > > > From: Ramesh Babu <ramesh.babu at intel.com>
> > > > >
> > > > > Signed-off-by: Ramesh Babu <ramesh.babu at intel.com>
> > > > > Signed-off-by: Subhransu S. Prusty <subhransu.s.prusty at intel.com>
> > > > > Signed-off-by: Vinod Koul <vinod.koul at intel.com>
> > > > > ---
> > > > >  sound/soc/codecs/hdac_hdmi.c | 8 ++++++++
> > > > >  1 file changed, 8 insertions(+)
> > > > >
> > > > > diff --git a/sound/soc/codecs/hdac_hdmi.c
> > b/sound/soc/codecs/hdac_hdmi.c
> > > > > index 46aa5cc..fa3cfa5 100644
> > > > > --- a/sound/soc/codecs/hdac_hdmi.c
> > > > > +++ b/sound/soc/codecs/hdac_hdmi.c
> > > > > @@ -1136,6 +1136,14 @@ static int hdac_hdmi_dev_probe(struct
> > hdac_ext_device *edev)
> > > > >  	INIT_LIST_HEAD(&hdmi_priv->pin_list);
> > > > >  	INIT_LIST_HEAD(&hdmi_priv->cvt_list);
> > > > >
> > > > > +	ret = snd_hdac_display_power(edev->hdac.bus, true);
> > > > > +	if (ret < 0) {
> > > > > +		dev_err(&edev->hdac.dev,
> > > > > +			"Cannot turn on display power on i915 err: %d\n",
> > > > > +			ret);
> > > > > +		return ret;
> > > > > +	}
> > > > > +
> > > > >  	ret = hdac_hdmi_parse_and_map_nid(edev, &hdmi_dais, &num_dais);
> > > > >  	if (ret < 0) {
> > > > >  		dev_err(&codec->dev, "Failed in parse and map nid with err:
> > %d\n", ret);
> > > > > --
> > > > > 1.9.1
> > > >
> > > > No counterpart to turn off?
> > > turned off in runtime suspend during first explicit suspend call.
> > 
> > It's a refcount, hence it does matter how many times it's called.
> > Does it really balance well?  It smells suspicious if you add only a
> > call to turn on without turning off...
> 
> Display power is turned ON during device probe. When soc codec probe
> completes, runtime suspend call is invoked and display power is turned OFF.
> So refcount is balanced. 

The runtime suspend of the codec, or of the controller?  I haven't
seen the codec suspend/resume doing it in this patchset, so I wonder.

if it's really well balanced, you should add the proper comment in the
code who turns off at when and where.  Otherwise reader will be lost.


Takashi