Re: Set eapd to follow stream on or off

[Takashi Iwai <tiwai@suse.de>]

At Wed, 21 May 2014 13:58:14 +0800,
Kailang wrote:
> 
> 
> > At Tue, 20 May 2014 18:05:42 +0800,
> > Kailang wrote:
> >>
> >> > At Tue, 20 May 2014 17:20:44 +0800,
> >> > Kailang wrote:
> >> >>
> >> >>
> >> >> > At Tue, 20 May 2014 16:52:43 +0800,
> >> >> > Kailang wrote:
> >> >> >>
> >> >> >>
> >> >> >> > At Tue, 20 May 2014 15:15:49 +0800,
> >> >> >> > Kailang wrote:
> >> >> >> >>
> >> >> >> >> Hi Takashi,
> >> >> >> >>
> >> >> >> >> HP machine had pop noise via speaker.
> >> >> >> >> If EAPD on or off, speaker will have pop noise.
> >> >> >> >> So, they want to set eapd to follow with stream on or off.
> >> >> >> >
> >> >> >> > I can apply this as is, but still I wonder whether this results 
> >> >> >> > in
> >> >> >> > other problems.  Doing EAPD on/off there means that it happens 
> >> >> >> > far
> >> >> >> > more often.  For example, what happens if you do like below?
> >> >> >> >
> >> >> >> >     while true; do
> >> >> >> >       echo -n > /dev/snd/pcmC0D0p
> >> >> >> >     done
> >> >> >> >
> >> >> >> > And, what if the device is resumed from power-saving by an audio
> >> >> >> > playback?  Immediately after the power up, it'll start the
> >> >> >> > streaming,
> >> >> >> > so it'll turn on EAPD.
> >> >> >> >
> >> >> >> > That said, isn't it just a need for some delay before EAPD 
> >> >> >> > on/off?
> >> >> >>
> >> >> >> No, eapd on or off will cause pop noise.
> >> >> >
> >> >> > Does it do always?
> >> >> No, pop noise just one pop sound. It can not remove by software.
> >> >> But this machine need to do this always.
> >> >
> >> > So, doing EAPD on/off often is rather worse.  It's what your patch
> >> > does.
> >> >
> >> > Once again: did you try the script above?  If the pop noise always
> >>
> >> I had try it on my develop machine. I have no HP issue machine on my 
> >> site.
> >>
> >> > happens, the operation like above is the worst scenario.
> >> >
> >> > Then, let's think again how to *reduce* it.  We need one EAPD on at
> >> > least after each power up.  But it's needed only when you actually
> >> > plays a stream.  That is, we need only enable EAPD at playback stream
> >> > open, but leave EAPD as is at close.
> >> >
> >>
> >> [we need only enable EAPD at playback stream open, but leave EAPD as is 
> >> at
> >> close.]
> >> Yes.
> >> So, I need to create eapd low in alc269_shutup. Right?
> >
> > Better to create own shutup callback (you can call alc269_shutup()
> > from there, too) and set it to spec->shutup in the fixup.  The pincfg
> > cleanup isn't often needed or may have bad effect.  You should test
> > with real machines what are needed and what are not.
> 
> static void alc290_fixup_pop_noise(struct hda_codec *codec,
>            const struct hda_fixup *fix, int action)
> {
>  struct alc_spec *spec = codec->spec;
> 
>  if (action == HDA_FIXUP_ACT_PRE_PROBE) {
>   spec->gen.pcm_playback_hook = ALC290_playback_pcm_hook;
>  }
> if (action ==   HDA_FIXUP_ACT_INIT) {
>   set_eapd(codec, 0x14, false);
>  }
> }
> 
> Could I modify as upper code?
> To set eapd low in HDA_FIXUP_ACT_INIT state.
> If system resume, it will initial eapd to low.
> To write eapd to same value, it will have no pop noise.
> It will keep always low for eapd, only stream open and close have change 
> state for eapd.

Yes, it looks OK.


Takashi


> 
> >
> >
> > Takashi
> >
> >>
> >> >
> >> > Takashi
> >> >
> >> >> >> Our customer just want to do eapd on when stream is open.
> >> >> >> The pop noise will not clear. Because music gain was biger than pop
> >> >> >> noise.
> >> >> >
> >> >> > Opening a stream doesn't mean that it follows the loud sound
> >> >> > immediately.  The script above wouldn't produce any sound but just
> >> >> > does open/close the stream.  Did you try that?
> >> >> >
> >> >> >
> >> >> > Takashi
> >> >> >
> >> >> > ------Please consider the environment before printing this e-mail.
> >> >>
> >> 
>