Re: Heads up: powerdown changes to alsa-base in Lucid/10.04

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

Re: Heads up: powerdown changes to alsa-base in Lucid/10.04

Amit Kucheria-6
Incorrect alias when cc'ing the kernel mailing list.

----- Forwarded message from Amit Kucheria <[hidden email]> -----

Date: Tue, 16 Feb 2010 13:31:05 +0200
From: Amit Kucheria <[hidden email]>
To: Daniel Chen <[hidden email]>
Cc: [hidden email]
Subject: Re: Heads up: powerdown changes to alsa-base in Lucid/10.04
User-Agent: Mutt/1.5.20 (2009-06-14)

Hi Daniel,

On 09 Dec 30, Daniel Chen wrote:
> Hi, folks,
>
> [a more complete version of my announcement in #ubuntu-devel on Freenode]
>
> Based on the abysmal feedback that we received for Karmic/9.10
> regarding powering down HDA controllers after ten idle seconds[0],

Do you have any statistics on how many machines were affected by these
changes? I see many bugs with 'HDA power_save=10' in their title but am
unable to comment on the mix of HW involved.  

What about codecs and amplifiers? Is there a pattern on which ones are most
affected or is it scattered all over the place?

>
> I've reverted the change in the most recent upload of alsa-driver
> 1.0.22.1 to Lucid/10.04 and, in doing so, restored the behavior
> shipped in Jaunty/9.04. So, at the cost of less efficient battery use
> [for unplugged laptops] one will no longer hear the annoying crackling
> and popping associated with these HDA controllers.
>
> This change was also driven by the fact that Lucid's shipped kernel is
> highly unlikely to have the necessary patches backported. The good

Can you elaborate on what patches needed porting? I believe most of the
patches from Karmic are upstream now, so it'll be a simple matter of
cherry-picking.

> news, OTOH, is that over the past month my fixes for these symptoms
> (including the infrastructure) have landed in alsa-driver 1.0.22.1 and
> will be available in crack-of-the-day alsa-driver snapshot builds soon
> and/or via linux-backports-modules-alsa-lucid-generic. Additionally,
> there is ongoing work to power down additional nodes for the HDA
> codecs; please find me on IRC (or leave a note via memoserv) if you
> wish to help roll out these fixes for your particular hardware. Please
> have your /proc/asound/card*/codec* handy/pastebinned.
>
> Thanks,
> Dan
 
Thanks for this work!

I'm trying to figure out if we can selectively enable this feature or atleast
make it very easy to enable.

> [0] https://lists.ubuntu.com/archives/ubuntu-devel-discuss/2009-May/008239.html
>

Regards,
Amit

--
----------------------------------------------------------------------
Amit Kucheria, Kernel Engineer || [hidden email]
----------------------------------------------------------------------

----- End forwarded message -----

--
----------------------------------------------------------------------
Amit Kucheria, Kernel Engineer || [hidden email]
----------------------------------------------------------------------

--
kernel-team mailing list
[hidden email]
https://lists.ubuntu.com/mailman/listinfo/kernel-team
Reply | Threaded
Open this post in threaded view
|

Re: Heads up: powerdown changes to alsa-base in Lucid/10.04

Daniel Chen-2
Hi, Amit, all,

On Tue, Feb 16, 2010 at 6:55 AM, Amit Kucheria
<[hidden email]> wrote:
> Do you have any statistics on how many machines were affected by these
> changes? I see many bugs with 'HDA power_save=10' in their title but am
> unable to comment on the mix of HW involved.

Using gmail's filters for an estimate, there are 22 reports each for
Realtek, 7 Analog Devices, 14 Conexant, 6 Sigmatel/IDT, and 2 VIA.
There are bound to be duplicates among them (also since there aren't
22 different Realtek HDA codecs, etc.). However, these are just from
Launchpad; there are far higher numbers from Ubuntu Forums, blog
spatterings, and so on. In other words, if you were unlucky enough to
have a non-Sigmatel/IDT codec, you pretty much got hammered by this
symptom; suffice to say - despite it being hand-waving - that it was
"a lot".

(Sigmatel/IDT users were spared, largely, because the fix was
backported to Karmic, so those users didn't experience the symptom.)

> What about codecs and amplifiers? Is there a pattern on which ones are most
> affected or is it scattered all over the place?

By far the most reported in Launchpad and across the forums/blogs seem
to be Realteks and Analog Devices. Lots of Dell, ASUS, Acer, HP.

> Can you elaborate on what patches needed porting? I believe most of the
> patches from Karmic are upstream now, so it'll be a simple matter of
> cherry-picking.

So there are two separate sets of fixes, one for quieting the
controller/codec on powerdown and another for powering down certain
nodes (e.g., Front/Spkr, HP) for suspend. There are some
infrastructure ones and then follow-on codec-specific ones. Roughly:

http://git.kernel.org/?p=linux/kernel/git/tiwai/sound-2.6.git;a=commitdiff;h=167eae5a17b3cd44a324dbb972c338e489413f54

http://git.kernel.org/?p=linux/kernel/git/tiwai/sound-2.6.git;a=commitdiff;h=fb8d1a344dbe963f16249d07eee8415e93f9f3c2

http://git.kernel.org/?p=linux/kernel/git/tiwai/sound-2.6.git;a=commitdiff;h=92ee6162c48fab24f0676969f0f147fc12f8f21c

http://git.kernel.org/?p=linux/kernel/git/tiwai/sound-2.6.git;a=commitdiff;h=f5de24b06aa46427500d0fdbe8616b73a71d8c28

http://git.kernel.org/?p=linux/kernel/git/tiwai/sound-2.6.git;a=commitdiff;h=a4e09aa3cf592d9f084ff4ceb216be40c4c265dc

http://git.kernel.org/?p=linux/kernel/git/tiwai/sound-2.6.git;a=commitdiff;h=c97259df3f2e163c72f4d0685c61fb2e026dc989

http://git.kernel.org/?p=linux/kernel/git/tiwai/sound-2.6.git;a=commitdiff;h=ea52bf260ecbb175339af3178c15788df21b7516

Additional changesets may be needed, but the ones above cover Realtek,
Analog Devices, and Sigmatel/IDT.

Just a note about Realtek coverage: I didn't enable the Front/Spkr &
HP coverage for all Realtek codecs, only the ones tested, which is in
contrast to Analog Devices and Sigmatel/IDT where it is globally
enabled. It's fairly trivial to enable globally for Realtek, however.

> I'm trying to figure out if we can selectively enable this feature or atleast
> make it very easy to enable.

Hmm, I've been pretty bad about announcing updates, sorry. Two weeks
ago, after discussing in #ubuntu-desktop with Martin Pitt, I
re-enabled this powerdown in pm-utils-powersave-policy 0.3 [0, 1] only
for Sigmatel/IDT when on battery power, because those have full
coverage already.

Best,
-Dan

[0] https://lists.ubuntu.com/archives/lucid-changes/2010-February/004670.html
[1] http://bazaar.launchpad.net/~ubuntu-branches/ubuntu/lucid/pm-utils-powersave-policy/lucid/revision/3

--
kernel-team mailing list
[hidden email]
https://lists.ubuntu.com/mailman/listinfo/kernel-team