[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#606237: linux-kernel: poweroff-button fails on 2.6.32-5-openvz-686



Richard Landsman - Rimote Media <richard@rimote.nl> writes:

> The bug you mentioned looks relevant. I did not use X. But I've tried to
> install acpi-support from both the testing and unstable repo and it
> doesn't seem to maken any difference. 

You currently need acpi-support-base from unstable.  Doesn't look like
acpi-support will pull that in.  It is satisfied with acpi-support-base
from testing. 

> Maybe the update is not yet in the
> repo? It looks like it still needs X and it is installed now so it looks
> like missing X does not have to do with the broken ACPI support.

acpi-support-base version 0.138-4  Works For Me (tm)

But the code is still doing a lot of weird stuff to verify that you are
a console user and to look for a number of power event handlers.

You could verify that you are using the default powerbtn script by
looking at the acpi config:

 frtest5:~# cat /etc/acpi/events/powerbtn-acpi-support 
 event=button[ /]power
 action=/etc/acpi/powerbtn-acpi-support.sh

and then run the script with sh -x to see where (and if it fails).
Beware:  This will poweroff your host if it doesn't fail....


frtest5:~# sh -x /etc/acpi/powerbtn-acpi-support.sh
+ [ -f /etc/acpi/events/powerbtn -o -f /etc/acpi/events/powerbtn.dpkg-bak ]
+ [ -e /usr/share/acpi-support/policy-funcs ]
+ . /usr/share/acpi-support/policy-funcs
+ . /usr/share/acpi-support/power-funcs
+ umask 022
+ PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/bin/X11
+ POWERSTATE=/var/lib/acpi-support/powerstate
+ HDPARM=/sbin/hdparm -q
+ LIDSTATE=/var/lib/acpi-support/lidstate
+ CheckPolicy
+ local PMS
+ getXconsole
+ fgconsole
+ activeVT=serial
+ return
+ PMS=/usr/bin/gnome-power-manager /usr/bin/kpowersave /usr/bin/xfce4-power-manager
+ PMS=/usr/bin/gnome-power-manager /usr/bin/kpowersave /usr/bin/xfce4-power-manager /usr/bin/guidance-power-manager /usr/lib/dalston/dalston-power-applet
+ pidof -x /usr/bin/gnome-power-manager /usr/bin/kpowersave /usr/bin/xfce4-power-manager /usr/bin/guidance-power-manager /usr/lib/dalston/dalston-power-applet
+ test  != 
+ PowerDevilRunning
+ test -x /usr/bin/dbus-send
+ local DBUS_SESS
+ pidof kded4
+ return 1
+ [ -x /etc/acpi/powerbtn.sh ]
+ [ -x /etc/acpi/powerbtn.sh.dpkg-bak ]
+ /sbin/shutdown -h -P now Power button pressed

Broadcast message from root@frtest5 (pts/0) (Fri Dec 17 13:40:32 2010):

Power button pressed 
The system is going down for system halt NOW!
frtest5:~# Connection to frtest5.lab closed by remote host.
Connection to frtest5.lab closed.


As you can see, it now works for me because "activeVT=serial" has been
added as a special case in /usr/share/acpi-support/power-funcs:

# getXconsole gets the foreground (console) X user
getXconsole() {
        activeVT=$(fgconsole) || pwf_error "can't get active VT"
        case $activeVT in
                serial)
                        return
                        ;;
                ''|*[!0-9]*)
                        pwf_error "invalid active VT '$activeVT'"
                        ;;
        esac


But I assume this will still fail for any unknown console type.  Run
"fgconsole" to see what yours is.


Bjørn



Reply to: