[alsa-devel] Disabling DAPM from ASoC framework

Mark Brown broonie at kernel.org
Tue Apr 22 14:20:50 CEST 2014

On Tue, Apr 22, 2014 at 10:17:35AM +0530, Santhosh N M wrote:

> >>>>>Why do you not want to do power management?

> It is because of the following reasons,

> 1) Due the complication in hardware (where codec is inbuilt to the SoC) and
> the software architecture, we decided to leave all the AMPs, DIFFINs and
> SINGINs ON always. All the MUXs are pre-configured (in codec driver) to
> route audio to the right path. So there is no real need to switch OFF any
> of the power components on the device!

That doesn't sound like a generic upstreamable driver that'll work on
other systems...

> 2) The device is AC powered and power consumption is not a problem here
> (unlike battery operated devices)

This isn't done purely for power reasons, it's also done so that an idle
device doesn't end up playing the noise floor which generally is audible
and so that changing use cases can be done without generating noise.

> This approach worked when I tested quickly. However, I'm slightly worried
> when you say " it's likely things will break". Not sure if there could lead
> to some memory leak in the long run.

Not memory leaks particularly, but the framework does assume that DAPM
is there and may fail to do things like start the audio streams.  If
it's holding together for you it'll probably not break but I'd expect
you'll run into trouble if you try to move forward to a new kernel
version, and possibly if you change what you're doing on your current

> Do you have any recommendation to achieve this in the right way?

If you really want the audio stream up all the time I'd just play audio
all the time from your sound server.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: Digital signature
URL: <http://mailman.alsa-project.org/pipermail/alsa-devel/attachments/20140422/4823d19c/attachment.sig>

More information about the Alsa-devel mailing list