power management in alsa

power management in alsa

[Andi Kleen]

Hallo,

I noticed ALSA hda seems to only change PCI power management states
of the sound devices on suspend/resume. Is there a reason the device
cannot be automatically suspended when nobody has sound opened? 

-Andi

Re: power management in alsa

[Takashi Iwai]

At Wed, 25 Jul 2007 14:36:58 +0200,
Andi Kleen wrote:
> 
> 
> Hallo,
> 
> I noticed ALSA hda seems to only change PCI power management states
> of the sound devices on suspend/resume. Is there a reason the device
> cannot be automatically suspended when nobody has sound opened? 

I made once the automatic power-control patch for hd-audio, but it was
possibly racy and a bit messy implementation, so I abandoned it.
Since then, no one has worked on this issue.


Takashi

Re: Re: power management in alsa

[Pavel Machek]

Hi!

> > I noticed ALSA hda seems to only change PCI power management states
> > of the sound devices on suspend/resume. Is there a reason the device
> > cannot be automatically suspended when nobody has sound opened? 
> 
> I made once the automatic power-control patch for hd-audio, but it was
> possibly racy and a bit messy implementation, so I abandoned it.
> Since then, no one has worked on this issue.

IIRC that patch saved .3W, which is quite significant -- about half of
spinning harddrive. I'd say it is worth the effort.

-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

Re: Re: power management in alsa

[Takashi Iwai]

At Wed, 25 Jul 2007 18:41:50 +0000,
Pavel Machek wrote:
> 
> Hi!
> 
> > > I noticed ALSA hda seems to only change PCI power management states
> > > of the sound devices on suspend/resume. Is there a reason the device
> > > cannot be automatically suspended when nobody has sound opened? 
> > 
> > I made once the automatic power-control patch for hd-audio, but it was
> > possibly racy and a bit messy implementation, so I abandoned it.
> > Since then, no one has worked on this issue.
> 
> IIRC that patch saved .3W, which is quite significant -- about half of
> spinning harddrive. I'd say it is worth the effort.

OK, then I'll take a look again (and try reimplementing a bit better :)


thanks,

Takashi

power management in ALSA

[Tehn Yit Chin]

[-- Attachment #1: Type: text/plain, Size: 898 bytes --]

Hi all,

I am trying to put power management awareness into a driver that I wrote,
but I am having a little trouble understanding which model to use.

Looking at the other drivers, I noticed that the suspend/resume callbacks
can be registered via snd_card_set_pm_callback(). The power management is
then handled via pm.c and friends.

The other way is to initialise the suspend/resume callbacks via the linux
device driver model like the following, and then registered via
driver_register(). I think that the power management is then handled by
sysfs.

static struct device_driver codec_driver = {
        .name = "codec",
        .bus = &platform_bus_type,
        .probe = codec_probe,
        .remove = codec_remove,
        .suspend = codec_suspend,
        .resume = codec_resume
};

Any suggestions on which is a better/correct method is welcome.

many thanks,
tyc

[-- Attachment #2: Type: text/html, Size: 1167 bytes --]

Re: power management in ALSA

[Takashi Iwai]

At Wed, 14 Dec 2005 10:50:05 +1100,
Tehn Yit Chin wrote:
> 
> Hi all,
> 
> I am trying to put power management awareness into a driver that I wrote, but I am
> having a little trouble understanding which model to use.
> 
> Looking at the other drivers, I noticed that the suspend/resume callbacks can be
> registered via snd_card_set_pm_callback(). The power management is then handled via
> pm.c and friends.
> 
> The other way is to initialise the suspend/resume callbacks via the linux device
> driver model like the following, and then registered via driver_register(). I think
> that the power management is then handled by sysfs.
> 
> static struct device_driver codec_driver = {
>         .name = "codec",
>         .bus = &platform_bus_type,
>         .probe = codec_probe,
>         .remove = codec_remove,
>         .suspend = codec_suspend,
>         .resume = codec_resume
> };
> 
> Any suggestions on which is a better/correct method is welcome.

The PM stuff has been completely changed in the recent version.
There is no longer the common PM layer in ALSA (except for
snd_power_change_state()).  Please take a look at ALSA 1.0.11rc1.

BTW, the correct PM implementation is the latter, using suspend/resume
callbacks of corresponding driver structs.


Takashi


-------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc. Do you grep through log files
for problems?  Stop!  Download the new AJAX search engine that makes
searching your log files as easy as surfing the  web.  DOWNLOAD SPLUNK!
http://ads.osdn.com/?ad_id=7637&alloc_id=16865&op=click