Annoying powersaving features in kernel 2.6.34?

nordog · 2010-06-27 01:51:02

I've got a Dell Inspiron E1505 laptop with the ATI X1400 card. Recently I've noticed that every few minutes my hard drive becomes completely silent and then spins up a few seconds later, and I think this behaviour started about the same time as I upgraded to the latest kernel version.

So my question is this: does the latest kernel include some new power saving features, and if so is this behaviour safe for my hard drive? I would have thought that constantly spinning up the drive would damage it in the long run. Furthermore, is there any way to turn this off?

If this is not a new feature, I guess I'm looking at some hardware failure...

Last edited by nordog (2010-06-27 01:59:11)

thtroyer · 2010-06-27 04:03:34

I don't know of any solution yet, but I noticed this on my Dell E1505 after an update... Figured I'd stop in here to see if anyone knew anything.

It is very annoying to hear the drive spin up every time it is accessed (I just had it spin up twice in less than 10 seconds) and I'm noticing some GUI slowdown when that happens...

thtroyer · 2010-06-27 04:15:35

Actually, setting this up seems to have helped so far:

http://wiki.archlinux.org/index.php/Lap … wn_problem

nordog · 2010-06-27 18:42:53

thtroyer wrote:

Actually, setting this up seems to have helped so far:
http://wiki.archlinux.org/index.php/Lap … wn_problem

Thanks for the tip.

I actually didn't have much luck with setting hdparm -B to 254 or 255, but it did set me off in the right track. Setting hdparm -S to 0 solved my problem...

I wonder if this new agressive powersaving is a feature or a bug?

demonicanima · 2010-06-29 05:46:08

I don't think it's a kernel issue...since I've been running a custom kernel for weeks, without problems, and am still using it; yet I still have this problem.

thtroyer wrote:

Actually, setting this up seems to have helped so far:
http://wiki.archlinux.org/index.php/Lap … wn_problem

While this works for me, it's only a temporary solution. If I switch to AC and then switch back to the battery, then the command needs to be used again.

If it helps any, here's what I last updated before this started to occur:

[2010-06-26 15:56] starting full system upgrade
[2010-06-26 16:33] removed kernel26-firmware (2.6.33.4-1)
[2010-06-26 16:33] Generating locales...
[2010-06-26 16:33]   en_US.UTF-8... done
[2010-06-26 16:33]   en_US.ISO-8859-1... done
[2010-06-26 16:33] Generation complete.
[2010-06-26 16:34] upgraded glibc (2.12-2 -> 2.12-4)
[2010-06-26 16:34] upgraded coreutils (8.5-1 -> 8.5-2)
[2010-06-26 16:34] installed linux-firmware (20100606-1)
[2010-06-26 16:34] upgraded udev (151-3 -> 157-1)
[2010-06-26 16:34] upgraded mkinitcpio (0.6.4-1 -> 0.6.6-1)
*snip*
[2010-06-26 16:35] upgraded kernel26 (2.6.33.4-1 -> 2.6.34-2)
[2010-06-26 16:35] upgraded aufs2 (2.6.33_20100425-2 -> 2.6.34_20100517-1)
[2010-06-26 16:35] upgraded bluez (4.65-1 -> 4.66-1)
[2010-06-26 16:35] installed boost-libs (1.43.0-1)
[2010-06-26 16:36] upgraded boost (1.41.0-1 -> 1.43.0-1)
[2010-06-26 16:37] upgraded brasero (2.30.1-1 -> 2.30.2-1)
[2010-06-26 16:37] upgraded dri2proto (2.1-2 -> 2.3-1)
[2010-06-26 16:37] installed poppler-data (0.4.2-1)
[2010-06-26 16:37] upgraded poppler (0.12.4-1 -> 0.14.0-1)
[2010-06-26 16:37] upgraded poppler-glib (0.12.4-1 -> 0.14.0-1)
[2010-06-26 16:37] upgraded evince (2.30.1-2 -> 2.30.3-1)
[2010-06-26 16:37] upgraded libsoup (2.30.1-1 -> 2.30.2-1)
[2010-06-26 16:37] upgraded libsoup-gnome (2.30.1-1 -> 2.30.2-1)
[2010-06-26 16:37] upgraded libgweather (2.30.0-1 -> 2.30.2-1)
[2010-06-26 16:37] upgraded evolution-data-server (2.30.1-1 -> 2.30.2.1-1)
[2010-06-26 16:38] upgraded file-roller (2.30.1.1-1 -> 2.30.2-1)
[2010-06-26 16:38] upgraded gedit (2.30.2-1 -> 2.30.3-1)
[2010-06-26 16:38] upgraded gnome-desktop (2.30.0-1 -> 2.30.2-1)
[2010-06-26 16:38] upgraded gnome-keyring (2.30.1-2 -> 2.30.3-1)
[2010-06-26 16:38] upgraded gnome-menus (2.30.0-1 -> 2.30.2-1)
[2010-06-26 16:38] upgraded libwnck (2.30.0-1 -> 2.30.2-1)
[2010-06-26 16:38] upgraded gnome-panel (2.30.0-1 -> 2.30.2-1)
[2010-06-26 16:38] upgraded gnome-themes (2.30.1-1 -> 2.30.2-1)
[2010-06-26 16:38] upgraded gparted (0.5.2-1 -> 0.6.0-1)
[2010-06-26 16:38] upgraded gstreamer0.10-bad (0.10.18-5 -> 0.10.19-1)
[2010-06-26 16:38] upgraded gstreamer0.10-bad-plugins (0.10.18-5 -> 0.10.19-1)
[2010-06-26 16:38] upgraded gstreamer0.10-good (0.10.22-1 -> 0.10.23-1)
[2010-06-26 16:38] upgraded gstreamer0.10-good-plugins (0.10.22-1 -> 0.10.23-1)
[2010-06-26 16:38] upgraded gstreamer0.10-ugly (0.10.14-5 -> 0.10.15-1)
[2010-06-26 16:38] upgraded gstreamer0.10-ugly-plugins (0.10.14-5 -> 0.10.15-1)
[2010-06-26 16:38] upgraded gtkhtml (3.30.1-1 -> 3.30.2-1)
[2010-06-26 16:38] upgraded gtksourceview2 (2.10.3-1 -> 2.10.4-1)
[2010-06-26 16:38] warning: /etc/rc.conf installed as /etc/rc.conf.pacnew
[2010-06-26 16:38] upgraded initscripts (2010.05-3 -> 2010.06-2)
[2010-06-26 16:39] upgraded inkscape (0.47-2 -> 0.47-3)
[2010-06-26 16:39] upgraded libdrm (2.4.19-2 -> 2.4.21-1)
[2010-06-26 16:39] upgraded libgl (7.7.1-1 -> 7.8.2-1)
[2010-06-26 16:39] upgraded intel-dri (7.7.1-1 -> 7.8.2-1)
[2010-06-26 16:40] upgraded kernel26-headers (2.6.33.4-1 -> 2.6.34-2)
[2010-06-26 16:40] upgraded libao (1.0.0-1 -> 1.0.0-2)
[2010-06-26 16:40] upgraded libfetch (2.31-1 -> 2.32-1)
[2010-06-26 16:40] upgraded libtiff (3.9.2-2 -> 3.9.4-1)
[2010-06-26 16:40] upgraded libtool (2.2.8-1 -> 2.2.10-1)
[2010-06-26 16:40] upgraded libtorrent-rasterbar (0.15.0-4 -> 0.15.0-5)
[2010-06-26 16:40] upgraded libxfont (1.4.1-1 -> 1.4.2-1)
[2010-06-26 16:40] upgraded m4 (1.4.14-1 -> 1.4.14-2)
[2010-06-26 16:40] upgraded man-pages (3.24-1 -> 3.25-1)
[2010-06-26 16:40] upgraded mesa (7.7.1-1 -> 7.8.2-1)
[2010-06-26 16:40] upgraded mingw32-runtime (3.18-1 -> 3.18-2)
[2010-06-26 16:40] warning: /etc/pacman.d/mirrorlist installed as /etc/pacman.d/mirrorlist.pacnew
[2010-06-26 16:40] upgraded pacman-mirrorlist (20100131-1 -> 20100621-1)
[2010-06-26 16:40] upgraded perl-xyne-common (2010.02.11.1-1 -> 2010.04.01.2-3)
[2010-06-26 16:40] upgraded perl-xyne-arch (2010.05.21.1-1 -> 2010.06.25.2-1)
[2010-06-26 16:40] upgraded pm-quirks (0.20100316-1 -> 0.20100619-1)
[2010-06-26 16:40] upgraded pm-utils (1.3.0-2 -> 1.4.0-1)
[2010-06-26 16:40] upgraded pmount (0.9.22-2 -> 0.9.23-1)
[2010-06-26 16:40] upgraded poppler-qt (0.12.4-1 -> 0.14.0-1)
[2010-06-26 16:40] upgraded powerpill (17.1-1 -> 2010.06.20.1-1)
[2010-06-26 16:40] upgraded pygobject (2.21.1-1 -> 2.21.3-1)
[2010-06-26 16:40] upgraded python-mpd (0.2.1-2 -> 0.2.1-3)
[2010-06-26 16:40] upgraded setuptools (0.6c11-1 -> 0.6c11-2)
[2010-06-26 16:40] upgraded twisted (10.0.0-0 -> 10.0.0-1)
[2010-06-26 16:40] upgraded udisks (1.0.1-2 -> 1.0.1-3)
[2010-06-26 16:40] upgraded vigra (1.7.0-1 -> 1.7.0-3)
[2010-06-26 16:40] upgraded vte (0.24.1-1 -> 0.24.2-1)
[2010-06-26 16:40] upgraded xf86-input-evdev (2.3.2-1 -> 2.4.0-1)
[2010-06-26 16:40] upgraded xf86-input-keyboard (1.4.0-1 -> 1.4.0-2)
[2010-06-26 16:40] upgraded xf86-input-mouse (1.5.0-1 -> 1.5.0-2)
[2010-06-26 16:40] upgraded xf86-input-synaptics (1.2.1-1 -> 1.2.2-2)
[2010-06-26 16:40] upgraded xf86-input-wacom (0.10.5-1 -> 0.10.6-2)
[2010-06-26 16:40] upgraded xf86-video-intel (2.10.0-1 -> 2.11.0-2)
[2010-06-26 16:40] upgraded xf86-video-vesa (2.3.0-1 -> 2.3.0-2)
[2010-06-26 16:40] upgraded xorg-server (1.7.6-3 -> 1.8.1.902-1)
[2010-06-26 16:40] upgraded zathura (0.0.6-1 -> 0.0.7-2)

R00KIE · 2010-06-29 19:49:09

Your problem may be that hard disks for mobile devices have very aggressive power saving settings, that is programmed into the firmware by the manufacturer.

It may happen that for some reason commits to disk are happening at a bigger time interval, this leads the hard disk to go into standby. You can try to disable the standby timeout and/or the advanced power management with hdparm even if you don't use laptop-mode-tools.

A good place to do that is in /etc/rc.local and look at [1] if you happen to suspend/hibernate because you need to set the same parameter after resuming from suspend/hibernate.

[1] http://wiki.archlinux.org/index.php/Hibernate

tashtego · 2010-06-30 20:05:10

Same here, and after the update, listed by demonicanima. Desktop computer, not a laptop.
What it can be? linux-firmware, pm-quirks, pm-utils?
It's very annoying.

magillos · 2010-07-03 18:08:17

It's pm-utils. It now does the same thing as laptop-mode-tools. (maybe it's better to disable laptop-mode entirely like I did?) You can copy /usr/lib/pm-utils/power.d/harddrive file to /etc/pm/power.d/ and setup it as you need or disable pm-utils disk management by creating empty file there (touch /etc/pm/power.d/harddrive).
You maybe want to have a look on other /usr/lib/pm-utils/power.d/ files. You can apply above to them too.
There is odd thing on my laptop when I setup /etc/pm/power.d/harddrive file; battery settings are in fact AC settings and AC settings are battery settings. Is it me only?

I found solution to the last one as well. AC section in file /etc/acpi/handler.sh need to be edited to smth like that:

ac_adapter)
        case "$2" in
            ACAD)
                case "$4" in
                    00000000)
                       echo -n $minspeed >$setspeed
                    /usr/sbin/pm-powersave true
                    ;;
                    00000001)
                        echo -n $maxspeed >$setspeed
                    /usr/sbin/pm-powersave false
                   ;;
                esac
                ;;
            *)  logger "ACPI action undefined: $2" ;;
        esac
        ;;

I added pm-powersave lines and changed AC) to ACAD) like showed in acpi_listen. Maybe some will find that useful. It's all messed little.

Last edited by magillos (2010-07-05 23:28:06)

Vardyr · 2010-07-04 19:43:39

I would've never noticed this happening on my desktop if I hadn't heard my netbook hard drive spinning up, knowing that I had disabled hard drive power management in laptop-mode. magillos's solution appears to fix the problem.

This is likely to tear up a few drives in a short period of time, and it's (stupidly) enabled by default with the latest pm-utils package. Should we file a bug?

demonicanima · 2010-07-08 01:14:42

magillos wrote:

It's pm-utils. It now does the same thing as laptop-mode-tools. (maybe it's better to disable laptop-mode entirely like I did?) You can copy /usr/lib/pm-utils/power.d/harddrive file to /etc/pm/power.d/ and setup it as you need or disable pm-utils disk management by creating empty file there (touch /etc/pm/power.d/harddrive).

This worked for me. Thanks for pointing out where those files were magillos.

As for your other problem, the closest thing I have to that is that is the battery indicator saying that I'm still on AC despite being on battery power for hours. But that's for another thread if I ever get around to asking about it.

R00KIE · 2010-07-08 20:24:49

I'm still not convinced, I have pm-utils installed and I don't see this behavior.

According to the man page of pm-utils, stuff inside power.d should only be used when issuing pm-powersave, it might be that something is doing that for you.

I would point my gaze at acpid and not pm-utils (I don't have acpid installed). Try removing it from your daemons array and see if you still have problems.

demonicanima · 2010-07-10 04:19:44

R00KIE wrote:

I would point my gaze at acpid and not pm-utils (I don't have acpid installed). Try removing it from your daemons array and see if you still have problems.

Acpid is not in my daemons list. Hal is however, and when I removed that earlier nothing changed. The hard drive would still spin down/up several times a minute.

This is the relevant section in /usr/lib/pm-utils/power.d/harddrive;

# Default values on AC
DRIVE_SPINDOWN_VALUE_AC="${DRIVE_SPINDOWN_VALUE_AC:-0}"
DRIVE_WRITE_CACHE_AC="${DRIVE_WRITE_CACHE_AC:-1}" 
DRIVE_POWER_MGMT_AC="${DRIVE_POWER_MGMT_AC:-254}"
DRIVE_ACOUSTIC_MGMT_AC="${DRIVE_ACOUSTIC_MGMT_AC:-0}"

# Default values on battery
DRIVE_SPINDOWN_VALUE_BAT="${DRIVE_SPINDOWN_VALUE_BAT:-6}"
DRIVE_WRITE_CACHE_BAT="${DRIVE_WRITE_CACHE_BAT:-0}" 
DRIVE_POWER_MGMT_BAT="${DRIVE_POWER_MGMT_BAT:-1}"
DRIVE_ACOUSTIC_MGMT_BAT="${DRIVE_ACOUSTIC_MGMT_BAT:-254}"

After changing
DRIVE_POWER_MGMT_BAT="${DRIVE_POWER_MGMT_BAT:-1}"
to
DRIVE_POWER_MGMT_BAT="${DRIVE_POWER_MGMT_BAT:-128}"
The hard drive is no longer aggressively "conserving" battery power (no real difference in runtime for me...).

But it does seem odd to me as well. I've never touched anything concerning powersaving except standby and hibernation. But it is fully possible I screwed something up along the way.:)
I'll try looking around.

R00KIE · 2010-07-10 08:16:58

My install is a couple years old and already survived one laptop death so I may have some leftover file(s) that may be keeping things working properly, however if it was working fine before for you and now it's broken then that should have happened to me and more people.

I have hal in my daemons array but like I said before I don't have acpid installed (which seems the likely offender from my perspective). I was also thinking that it might be something new your DE is doing (if it is gnome or kde) but in that case more people would have noticed I guess.

rent0n · 2010-07-10 09:35:54

As far as I understand, the pm-utils scripts in those directories should be activated only when pm-powersave is executed. That's the new feature of pm-utils 1.4, 2.6.34 is not involved directly.
So I guess you have something calling pm-powersave, either acpid or your DE's power management.
Also, keep in mind that acpid does not need to be explicitly called at startup, because hal automatically starts it (if hal is on your DAEMONS list).

Btw, you're not using laptop-mode-tools, right?

demonicanima · 2010-07-11 01:58:04

rent0n wrote:

So I guess you have something calling pm-powersave, either acpid or your DE's power management.
Also, keep in mind that acpid does not need to be explicitly called at startup, because hal automatically starts it (if hal is on your DAEMONS list).
Btw, you're not using laptop-mode-tools, right?

R00KIE wrote:

I have hal in my daemons array but like I said before I don't have acpid installed (which seems the likely offender from my perspective). I was also thinking that it might be something new your DE is doing (if it is gnome or kde) but in that case more people would have noticed I guess.

Despite having most of gnome installed, I do not use a DE; I do however use Openbox.
Laptop-mode-tools has never been installed.

I did recently change (same day that I changed the /usr/lib/pm-utils/power.d/harddrive file) from using gnome-power-manager to using batterymon, so I went and changed the value back and rebooted a few times switching between the two. And the hard drive goes back to the aggressive power saving for both.

So I tried putting !acpid in the daemons list after hal, that did nothing, so I removed hal entirely and replaced it with dbus (needed for wicd) and the aggressive power saving continues.

The only parts of pm-utils that I've (unsuccessfully) used is pm-suspend and pm-hibernate. I'll try removing pm-utils and see if the problem persists. Granted it'll probably fix it, but uninstalling something shouldn't be a solution unless it conflicts with its replacement.

edit:
Oook....well that didn't work either. Hal and upower require pm-utils. So I used -Rd in pacman, and the hard drive is still in an aggressive power save mode. I also looked through some of the upower config files and found nothing of interest.

edit2:
New "symptom"...after removing and reinstalling pm-utils it's now in powersave mode all the time (ac/dc) and editing /usr/lib/pm-utils/power.d/battery no longer has an affect.

Last edited by demonicanima (2010-07-11 05:11:21)

coolaman · 2010-07-11 06:51:11

Same issue here.
Solved for me with this awful hack: I comment the function in /usr/lib/pm-utils/sleep.d/00powersave.

#!/bin/sh

. "${PM_FUNCTIONS}"

command_exists pm-powersave || exit $NA

#case $1 in
# suspend|hibernate) pm-powersave false ;;
# resume|thaw) pm-powersave ;;
# *) exit $NA ;;
#esac
exit 0

BelgiumArcher · 2010-07-14 10:37:25

Same Problem here on laptop Dell XPS M1530.
I disabled Hal, laptop-mode-tools in rc.conf
Pm-utils is installed.
Have not acpid in deamonslist
solution descibed in the wiki doesn't help
Haven't tried yet the solution from coolaman as it awfal as he told himself.

Only solution right now is to work on ac!

Is anyone looking further to track down the real problem.

I have been using Arch for 2 years now and it is my first post as I could fix all other problems by looking in the wiki and this forum.
(expect N-speed for my wireless )

ablepharus · 2010-07-14 10:41:42

may this will help you http://bbs.archlinux.org/viewtopic.php?id=100063

BelgiumArcher · 2010-07-14 11:05:17

Thanks.
But it is more of a workaround for now.
My harddrive should spin down if I don't use it for several minutes.
Mayby if I re-enable laptop-mode-tools.
I will check it this evening.

I still don't understand why we need both pm-utils and laptop-mode-tools.

Vamp898 · 2010-07-14 11:12:58

[vamp898@VampLap vamp898]$ cat /etc/rc.d/hdparm 
#!/bin/bash

. /etc/rc.conf
. /etc/rc.d/functions

case "$1" in
        start)
        stat_busy "Setup HDParm"
        hdparm -q -B 192 /dev/sda
        hdparm -q -B 192 /dev/sda1
        add_daemon hdparm
        stat_done
        ;;
        stop)
        stat_busy "Unset HDParm"
        hdparm -q -B 128 /dev/sda
        hdparm -q -B 128 /dev/sda1
        rm_daemon hdparm
        stat_done
        ;;
        restart)
        stat_busy "Unset HDParm"         
        hdparm -q -B 128 /dev/sda
        hdparm -q -B 128 /dev/sda1
        rm_daemon hdparm
        stat_done
        stat_busy "Setup HDParm"
        hdparm -q -B 192 /dev/sda
        hdparm -q -B 192 /dev/sda1
        add_daemon hdparm
        stat_done
        ;;
esac

oh and dont forget this

touch /etc/pm/power.d/harddrive

the last command overwrites the PM-Utils settings for power-saving and so you have either to set your own settings or your harddisk will stay always at the same power-settings that means default

so with

touch /etc/pm/power.d/harddrive

you completely disable every settings.

With my RC-Script (which you should put into your /etc/rc.d and than put into the DAEMONS array) you setup the Harddisk for a state of 192

its a bit lesser than the default settings so your harddisk doesnt park the head that often and so longers the life

If you want to know why the touch command disables the stuff you should read the man-pages of pm-utils

Last edited by Vamp898 (2010-07-14 11:26:00)

BelgiumArcher · 2010-07-15 13:00:54

the solution from Vamp898 works!
Too bad we still don't know why the problem occurered in the first place.
Thank you for helping me.

BelgiumArcher · 2010-12-28 18:56:42

Same problem again.
Don't know when it happened but it must be maximum 4 weeks.

I checked the script but is hasn't be modified since juli.

Anybody still following this thread?

Arch Linux

#1 2010-06-27 01:51:02

Annoying powersaving features in kernel 2.6.34?

#2 2010-06-27 04:03:34

Re: Annoying powersaving features in kernel 2.6.34?

#3 2010-06-27 04:15:35

Re: Annoying powersaving features in kernel 2.6.34?

#4 2010-06-27 18:42:53

Re: Annoying powersaving features in kernel 2.6.34?

#5 2010-06-29 05:46:08

Re: Annoying powersaving features in kernel 2.6.34?

#6 2010-06-29 19:49:09

Re: Annoying powersaving features in kernel 2.6.34?

#7 2010-06-30 20:05:10

Re: Annoying powersaving features in kernel 2.6.34?

#8 2010-07-03 18:08:17

Re: Annoying powersaving features in kernel 2.6.34?

#9 2010-07-04 19:43:39

Re: Annoying powersaving features in kernel 2.6.34?

#10 2010-07-08 01:14:42

Re: Annoying powersaving features in kernel 2.6.34?

#11 2010-07-08 20:24:49

Re: Annoying powersaving features in kernel 2.6.34?

#12 2010-07-10 04:19:44

Re: Annoying powersaving features in kernel 2.6.34?

#13 2010-07-10 08:16:58

Re: Annoying powersaving features in kernel 2.6.34?

#14 2010-07-10 09:35:54

Re: Annoying powersaving features in kernel 2.6.34?

#15 2010-07-11 01:58:04

Re: Annoying powersaving features in kernel 2.6.34?

#16 2010-07-11 06:51:11

Re: Annoying powersaving features in kernel 2.6.34?

#17 2010-07-14 10:37:25

Re: Annoying powersaving features in kernel 2.6.34?

#18 2010-07-14 10:41:42

Re: Annoying powersaving features in kernel 2.6.34?

#19 2010-07-14 11:05:17

Re: Annoying powersaving features in kernel 2.6.34?

#20 2010-07-14 11:12:58

Re: Annoying powersaving features in kernel 2.6.34?

#21 2010-07-15 13:00:54

Re: Annoying powersaving features in kernel 2.6.34?

#22 2010-12-28 18:56:42

Re: Annoying powersaving features in kernel 2.6.34?

Board footer