Bugzilla – Bug 295083
hang on boot because of unhandled ACPI event
Last modified: 2010-11-15 21:31:35 UTC
The system hangs on boot repeatedly printing the following error message if the VGA out is connected to a certain monitor (PAT-LCD); it works well with a different LCD monitor: ACPI Error (evgpe-0688): No handler or method for GPE[17], disabling event [20060127]. ACPI Error (evgpe-0688): No handler or method for GPE[17], disabling event [20060127]. ACPI Error (evgpe-0688): No handler or method for GPE[17], disabling event [20060127]. ACPI Error (evgpe-0688): No handler or method for GPE[17], disabling event [20060127]. ACPI Error (evgpe-0688): No handler or method for GPE[17], disabling event [20060127]. ACPI Error (evgpe-0688): No handler or method for GPE[17], disabling event [20060127]. ... Looks like if disabling that ACPI event would not succeed. Yet it continues to boot if the PAT-LCD is unplugged. As soon as X11 loads the Nvidia driver it needs to be plugged in again, because otherwise it will not be possible to make use of the PAT-LCD monitor unless the X server is restarted. It would be a pleasure if plugging and unplugging an external monitor would not require me to restart the X-server. SaX2 configures the monitor like the following (xorg.conf): Section "Monitor" Option "CalcAlgorithm" "XServerPool" DisplaySize 340 270 HorizSync 30-82 Identifier "Monitor[0]" ModelName "LCD_MONITOR" Option "DPMS" VendorName "PAT" VertRefresh 43-75 UseModes "Modes[0]" EndSection
It is the same for Suse10.1 and Suse10.2, for the rescue system or any installed system.
Could you please test with latest available NVIDIA driver and attach the result of running nvidia-bug-report.sh if the problem still occurs?
Still no response. :-( Please reopen, once you can provide the requested information. Thanks.
(Sorry for my late reply. It has been a little bit arduous for me to get that particular screen and computer with a working burn of openSUSE-10.3-GM-KDE-Live-i386.iso together again :|. Wanna get a working installation of 10.3 on that particular computer soon.) Unfortunately although the initial error has gone a similar one has occurred: ACPI Error (evgpe-0705): No handler or method for GPE[17], disabling event [20070126].
I don`t think the bug has anything to do with X11 or a 3rd party X11 driver. In a fact that one has coped well with this particular screen. As posted in comment #1 this happens on boot while the kernel is still doing its initialisation, possibly long before /sbin/init gets a chance to run. In a fact I do not want to run X11, I just wanna boot into a text console while that particular monitor is plugged on. The basic question is on why something of that little importance like these ACPI event (be it evgpe-0705 or evgpe-068) can halt the whole boot process and make the machine hang.
s
Ok. But make sure the nvidia driver is *not* installed. Otherwise this issue won't be investigated by our kernel developers.
It should not be installed on the openSUSE-10.3-GM-KDE-Live-i386.iso; or should it?
???
In deed no nvidia kernel-module is loaded by the Live-CD: > lsmod | g nv > lsmod | g agp intel_agp 27156 0 agpgart 35764 1 intel_agp
Using the new LifeCD the boot process stops in a fact twice: * at the very beginning (before I could press ESC to view the boot messages) - i.e. somewhere around 'loading kernel modules' * directly after the output: 'loading image' temporarily plugging off that monitor helps i.e. boot process continued normally
Alexey - Any ideas here? I'm not certain the ACPI event messages are related.
Please attach acpidump output.
Created attachment 189214 [details] acpidump >acpi.dump (diff.ext.scr.plugged)
This system declares WMI interface on top of ACPI. As a proprietary MS extension it is not yet supported with mainline kernel. If you machine is from Acer, you could try to load acer_wmi driver. Please also check if your BIOS is recent.
Could you also try if 2.6.24 kernel works better?
Unfortunately this is a problem. RPM refuses to install kernel 2.6.24 on my Suse10.1 in addition to kernels 2.6.16.54 and 2.6.18 because of dependency conflicts. There is not enough place left on my hard disk for an additional installation of 10.3. Installing the kernel into a plain directory causes rpm to hang irresponsively (kill -9 required): >ps ax|g rpm shell-init: error retrieving current directory: getcwd: cannot access parent directories: Datei oder Verzeichnis nicht gefunden 437 pts/10 S+ 0:00 rpm --nodeps -ivh --root /xen/b10.3 /dld/system/kernelUpdate/10.3/kernel-default-2.6.24-133.1.i586.rpm > kill -9 437 Simply unpacking the files with rpm2cpio does not produce an initrd. How should I proceed?
Identical message this time: hang on boot because of unhandled ACPI event The error in deed occurs in a very early stage: At first: boot process is halted at blue Suse-screen. Then after unplugging the besaid monitor esc can be pressed to view this error message. That is while the kernel is waiting for root=/dev/sda7 to get ready. In a fact /dev/sda7 never occurs so that initrd lets bash prompt for user commands (At this stage the monitor can be plugged in again). In its current state of installation vmlinuz-2.6.24-133-default could neither mount the root partition /dev/sda7 nor could any /dev/sda* be accessed from the initial shell (have merely unpacked the kernel under a 10.1-installation and issued /sbin/mkinitrd -k /boot/vmlinuz-2.6.24-133-default -i /boot/initrd-2.6.24-133-default ).
Unfortunately it is really not possible to get a bios update for that computer. In spite of various complaints the hardware vendor still provides a broken bios update (http://download.gericom.com/NOTEBOOK/Win98-Win2k-WinXP/Supersonic Series/1stSupersonicPCIE-KN1/BIOS/ - machine architecture spec and fetch util is missing; the spec from similar notebooks can not be used). Even esupport.com could not provide a bios update for that computer.
I can confirm this bug on both Ubunutu 7.10 and 8.04 on the very same notebook. I've had the external monitor working once after the update from 7.10 to 8.04, but have never been able to get the external monitor working again. This bug not only shows up when the monitor is plugged in during boot but will also happen bring cpu usage right up to 100% right when I plug in the monitor after booting without the monitor attachted. Also, kern.log will instantly start to fill with the aforemented ACPI error lines. If I can help with any debug output or the like, I'd be glad to help. Being a novice linux user, I'd appreciate having copy-and-paste terminal lines to create the needed information, if possible.
Last comment from 2008 -> resolve won't fix. If you can still reproduce the issue with 11.3, please reopen.
as tested with kernel 2.6.36 the problem no more seems to exist.