[alsa-devel] snd-hda-intel: Power management not enabled until audio is played
Dear ALSA folks,
playing with PowerTOP 1.97 I notice that it lists 100 % usage for the Audio device on several systems [1].
I made sure that power management is enabled.
$ grep snd /etc/modprobe.d/* …: options snd_hda_intel power_save=1 $ more /sys/module/snd_hda_intel/parameters/power_save_controller Y $ more /sys/module/snd_hda_intel/parameters/power_save 0
Since I was on AC power pm-utils seems to reset that parameter to not save power. :/
$ sudo su # echo "1" > /sys/module/snd_hda_intel/parameters/power_save # exit
But enabling that does not change anything right away. Only after using the device by playing some audio – `speaker-test` in this case, the power management seems to kick in.
$ more /sys/module/snd_hda_intel/parameters/power_save* :::::::::::::: /sys/module/snd_hda_intel/parameters/power_save :::::::::::::: 1 :::::::::::::: /sys/module/snd_hda_intel/parameters/power_save_controller :::::::::::::: Y $ more /sys/devices/pci0000:00/0000:00:1b.0/sound/card0/hwC0D0/power_*_acct :::::::::::::: /sys/devices/pci0000:00/0000:00:1b.0/sound/card0/hwC0D0/power_off_acct :::::::::::::: 9640 :::::::::::::: /sys/devices/pci0000:00/0000:00:1b.0/sound/card0/hwC0D0/power_on_acct :::::::::::::: 405056 $ more /sys/devices/pci0000:00/0000:00:1b.0/sound/card0/hwC0D0/power_*_acct :::::::::::::: /sys/devices/pci0000:00/0000:00:1b.0/sound/card0/hwC0D0/power_off_acct :::::::::::::: 9640 :::::::::::::: /sys/devices/pci0000:00/0000:00:1b.0/sound/card0/hwC0D0/power_on_acct :::::::::::::: 409664 $ more /sys/devices/pci0000:00/0000:00:1b.0/sound/card0/hwC0D0/power_*_acct :::::::::::::: /sys/devices/pci0000:00/0000:00:1b.0/sound/card0/hwC0D0/power_off_acct :::::::::::::: 9640 :::::::::::::: /sys/devices/pci0000:00/0000:00:1b.0/sound/card0/hwC0D0/power_on_acct :::::::::::::: 413672 $ date && more /sys/devices/pci0000:00/0000:00:1b.0/sound/card0/hwC0D0/power_*_acct Sa 11. Feb 10:08:46 CET 2012 :::::::::::::: /sys/devices/pci0000:00/0000:00:1b.0/sound/card0/hwC0D0/power_off_acct :::::::::::::: 9640 :::::::::::::: /sys/devices/pci0000:00/0000:00:1b.0/sound/card0/hwC0D0/power_on_acct :::::::::::::: 464644 $ date && more /sys/devices/pci0000:00/0000:00:1b.0/sound/card0/hwC0D0/power_*_acct Sa 11. Feb 10:08:52 CET 2012 :::::::::::::: /sys/devices/pci0000:00/0000:00:1b.0/sound/card0/hwC0D0/power_off_acct :::::::::::::: 9640 :::::::::::::: /sys/devices/pci0000:00/0000:00:1b.0/sound/card0/hwC0D0/power_on_acct :::::::::::::: 470076 $ date && more /sys/devices/pci0000:00/0000:00:1b.0/sound/card0/hwC0D0/power_*_acct Sa 11. Feb 10:09:04 CET 2012 :::::::::::::: /sys/devices/pci0000:00/0000:00:1b.0/sound/card0/hwC0D0/power_off_acct :::::::::::::: 9640 :::::::::::::: /sys/devices/pci0000:00/0000:00:1b.0/sound/card0/hwC0D0/power_on_acct :::::::::::::: 482604 $ date && more /sys/devices/pci0000:00/0000:00:1b.0/sound/card0/hwC0D0/power_*_acct Sa 11. Feb 10:10:41 CET 2012 :::::::::::::: /sys/devices/pci0000:00/0000:00:1b.0/sound/card0/hwC0D0/power_off_acct :::::::::::::: 9640 :::::::::::::: /sys/devices/pci0000:00/0000:00:1b.0/sound/card0/hwC0D0/power_on_acct :::::::::::::: 578904 $ date && speaker-test Sa 11. Feb 10:12:58 CET 2012
speaker-test 1.0.24.2
Wiedergabe-Gerät ist default Stream-Parameter sind 48000 Hz, S16_LE, 1 Kanäle Verwende 16 Oktaven rosa Rauschen Rate ist 48000 Hz (angefordert: 48000 Hz) Puffergröße von 2048 bis 8192 Periodengröße von 1024 bis 1024 Verwende maximale Puffergröße 8192 Perioden = 4 gesetzt: period_size = 1024 gesetzt: buffer_size = 8192 0 - Vorne links Zeit pro Periode = 2,828717 0 - Vorne links Zeit pro Periode = 2,985785 0 - Vorne links Zeit pro Periode = 2,985784 0 - Vorne links Zeit pro Periode = 2,985806 0 - Vorne links Zeit pro Periode = 2,985834 0 - Vorne links Zeit pro Periode = 2,985798 0 - Vorne links Zeit pro Periode = 2,985838 0 - Vorne links ^C $ date && more /sys/devices/pci0000:00/0000:00:1b.0/sound/card0/hwC0D0/power_*_acct Sa 11. Feb 10:13:24 CET 2012 :::::::::::::: /sys/devices/pci0000:00/0000:00:1b.0/sound/card0/hwC0D0/power_off_acct :::::::::::::: 100732 :::::::::::::: /sys/devices/pci0000:00/0000:00:1b.0/sound/card0/hwC0D0/power_on_acct :::::::::::::: 651372 $ date && more /sys/devices/pci0000:00/0000:00:1b.0/sound/card0/hwC0D0/power_*_acct Sa 11. Feb 10:13:29 CET 2012 :::::::::::::: /sys/devices/pci0000:00/0000:00:1b.0/sound/card0/hwC0D0/power_off_acct :::::::::::::: 105384 :::::::::::::: /sys/devices/pci0000:00/0000:00:1b.0/sound/card0/hwC0D0/power_on_acct :::::::::::::: 651372 $ date && more /sys/devices/pci0000:00/0000:00:1b.0/sound/card0/hwC0D0/power_*_acct Sa 11. Feb 10:24:18 CET 2012 :::::::::::::: /sys/devices/pci0000:00/0000:00:1b.0/sound/card0/hwC0D0/power_off_acct :::::::::::::: 753148 :::::::::::::: /sys/devices/pci0000:00/0000:00:1b.0/sound/card0/hwC0D0/power_on_acct :::::::::::::: 652876
Is that an error in the driver? I have seen that with Realtek and Conexant devices. Please also find the output of `alsa-info.sh` [2] attached. There is no difference before and after playing some audio.
$ diff -u alsa-info.txt.beforeplayingaudio alsa-info.txt.afterplayingaudio --- alsa-info.txt.beforeplayingaudio 2012-02-11 10:11:31.000000000 +0100 +++ alsa-info.txt.afterplayingaudio 2012-02-11 10:13:52.000000000 +0100 @@ -3,7 +3,7 @@ !!ALSA Information Script v 0.4.60 !!################################
-!!Script ran on: Sat Feb 11 09:11:29 UTC 2012 +!!Script ran on: Sat Feb 11 09:13:51 UTC 2012
!!Linux Distribution @@ -379,7 +379,7 @@ crw-rw----+ 1 root audio 116, 5 Feb 11 10:00 /dev/snd/controlC0 crw-rw----+ 1 root audio 116, 4 Feb 11 10:00 /dev/snd/hwC0D0 crw-rw----+ 1 root audio 116, 3 Feb 11 10:00 /dev/snd/pcmC0D0c -crw-rw----+ 1 root audio 116, 2 Feb 11 10:00 /dev/snd/pcmC0D0p +crw-rw----+ 1 root audio 116, 2 Feb 11 10:12 /dev/snd/pcmC0D0p crw-rw----+ 1 root audio 116, 1 Feb 11 10:00 /dev/snd/seq crw-rw----+ 1 root audio 116, 33 Feb 11 10:00 /dev/snd/timer
Thanks,
Paul
[1] http://lists.lesswatts.org/pipermail/discuss/2012-February/000687.html [2] http://alsa-project.org/main/index.php/Help_To_Debug
At Sat, 11 Feb 2012 10:59:04 +0100, Paul Menzel wrote:
Dear ALSA folks,
playing with PowerTOP 1.97 I notice that it lists 100 % usage for the Audio device on several systems [1].
I made sure that power management is enabled.
$ grep snd /etc/modprobe.d/* …: options snd_hda_intel power_save=1 $ more /sys/module/snd_hda_intel/parameters/power_save_controller Y $ more /sys/module/snd_hda_intel/parameters/power_save 0Since I was on AC power pm-utils seems to reset that parameter to not save power. :/
$ sudo su # echo "1" > /sys/module/snd_hda_intel/parameters/power_save # exitBut enabling that does not change anything right away. Only after using the device by playing some audio – `speaker-test` in this case, the power management seems to kick in.
Right, it's a known issue. You don't have to play actually but just opening/closing a device should suffice.
The reason is that changing the sysfs parameter file would change only the variable but doesn't trigger the event by itself. It's possible to add some triggering action, but it'd become fairly messy. So I've postponed the implementation, so far.
Takashi
Am Montag, den 13.02.2012, 15:16 +0100 schrieb Takashi Iwai:
At Sat, 11 Feb 2012 10:59:04 +0100, Paul Menzel wrote:
playing with PowerTOP 1.97 I notice that it lists 100 % usage for the Audio device on several systems [1].
I made sure that power management is enabled.
$ grep snd /etc/modprobe.d/* …: options snd_hda_intel power_save=1 $ more /sys/module/snd_hda_intel/parameters/power_save_controller Y $ more /sys/module/snd_hda_intel/parameters/power_save 0Since I was on AC power pm-utils seems to reset that parameter to not save power. :/
$ sudo su # echo "1" > /sys/module/snd_hda_intel/parameters/power_save # exitBut enabling that does not change anything right away. Only after using the device by playing some audio – `speaker-test` in this case, the power management seems to kick in.
Right, it's a known issue.
Should I submit a bug report to track that issue?
You don't have to play actually but just opening/closing a device should suffice.
Is `touch /dev/dsp` enough? I am thinking about adding that command to some start script.
The reason is that changing the sysfs parameter file would change only the variable but doesn't trigger the event by itself. It's possible to add some triggering action, but it'd become fairly messy. So I've postponed the implementation, so far.
Understood. I guess most users do not experience that since nowadays the login manager plays some sound opening the device every start up. Mostly advanced users will suffer from this not using such a login manager or desktop environment like GNOME or KDE.
Thanks,
Paul
At Mon, 13 Feb 2012 20:51:26 +0100, Paul Menzel wrote:
Am Montag, den 13.02.2012, 15:16 +0100 schrieb Takashi Iwai:
At Sat, 11 Feb 2012 10:59:04 +0100, Paul Menzel wrote:
playing with PowerTOP 1.97 I notice that it lists 100 % usage for the Audio device on several systems [1].
I made sure that power management is enabled.
$ grep snd /etc/modprobe.d/* …: options snd_hda_intel power_save=1 $ more /sys/module/snd_hda_intel/parameters/power_save_controller Y $ more /sys/module/snd_hda_intel/parameters/power_save 0Since I was on AC power pm-utils seems to reset that parameter to not save power. :/
$ sudo su # echo "1" > /sys/module/snd_hda_intel/parameters/power_save # exitBut enabling that does not change anything right away. Only after using the device by playing some audio – `speaker-test` in this case, the power management seems to kick in.
Right, it's a known issue.
Should I submit a bug report to track that issue?
No need for that. It's a missing feature but no bug.
You don't have to play actually but just opening/closing a device should suffice.
Is `touch /dev/dsp` enough? I am thinking about adding that command to some start script.
I'm not sure whether "touch" program actually works for a device file. Playing a sound like "echo -n | aplay" would work.
Takashi
participants (2)
-
Paul Menzel -
Takashi Iwai