Bug 137345

Summary: computer stucks during installation
Product: [openSUSE] SUSE LINUX 10.0 Reporter: Heiko Schmidt <heiko.schmidt>
Component: KernelAssignee: Thomas Renninger <trenn>
Status: RESOLVED FIXED QA Contact: E-mail List <qa-bugs>
Severity: Normal    
Priority: P5 - None    
Version: unspecified   
Target Milestone: ---   
Hardware: x86-64   
OS: Other   
Whiteboard:
Found By: Beta-Customer Services Priority:
Business Priority: Blocker: ---
Marketing QA Status: --- IT Deployment: ---

Description Heiko Schmidt 2005-12-07 07:18:07 UTC
I tried to install SuSE10.0 64 on a laptop with a AMD Turion 64 ML-30. It was only possible to start without ACPI installation. After all packages was installed, the computer rebooted. I pressed on SuSE Linux 10.0 and received a black display. Trying the same with secure settings I got the following message on the screen:

CPU: AMD Turion 64 Mobile Technology ML-30 stepping 02
checking if image is initramfs ... it is
testing NMI watchdog ... <7> Detected 12.500 MHz APIC Timer
APIC error on CPU0 :40(40)

So can you help me ?
Comment 1 Danny Al-Gaaf 2005-12-07 10:29:45 UTC
*** Bug 137346 has been marked as a duplicate of this bug. ***
Comment 2 Stefan Hundhammer 2005-12-07 10:59:59 UTC
Can't be "critical" if only a small number of users (those with that kind of hardware) are affected - and even less so if there is a workaround with a different ACPI mode.

Besides, it is obvious that this is a kernel bug and not a bug of the installer.
Comment 3 Stefan Hundhammer 2005-12-07 11:19:43 UTC
reassigning to kernel maintainers
Comment 4 Hannes Reinecke 2005-12-07 11:25:02 UTC
Thomas, can you help here?
Comment 5 Thomas Renninger 2005-12-08 18:21:17 UTC
Please be sure you have the latest BIOS installed.
If yes and you still have problems: Does the boot param noapic help?

You should not boot with acpi=off or failsafe options (AFAIK they have acpi=off included).
If noapic does not help, try pci=noacpi. It is also worth to try both if it still does not work (noacpi pci=noacpi).
Comment 6 Heiko Schmidt 2005-12-11 18:58:04 UTC
Thanks this works.

Four more questions:
- I´ve got an 1600 MHz but my system only uses 800 GHz how can I change this
- when I try setting up my WLAN via ndiswrapper, ndiswrapper itselfs is working, but when I use modeprobe ndiswrapper t tells me that the kernel-module does´nt exist, but it´s installed, so what to do (I already tried reinstalling)
- should I take this two problems into new defects ?
- is ACPI working with AMD Turion MT-37 ?

Thanks for your help.
Comment 7 Thomas Renninger 2005-12-12 10:19:14 UTC
What works? If pci=noacpi helps, you may be able to boot without this param in the newst 10.0 YOU update kernel. Otherwise you have to live with the boot param for 10.0 (there shouldn't be any disadvantages concerning functionality).

NDiswrapper: I had the same problem.
Be sure you have installed the ndiswrapper and the km_ndiswrapper package (kernel module).

If you already have the km_ndiswrapper package installed and modprobe ndiswrapper tells you the module is not there, this is a bug. You can workaround it by installing the kernel sources, then goto /usr/src/kernel_modules/ndiswrapper (or similar) and do: make;make install (be sure you installed make and gcc packages before). If you have to do this, please open a bug and assign it to jg@suse.de.

ACPI and AMD Turion MT-37:
Depending on your BIOS/mainboard, I know following problems with a Turion (with Open Suse 10.1 Alpha):

    - It does not shutdown (you need to hit the power off button when it should 
      power off)
    - if you have an ATI chipset you need boot param: no_timer_check or your 
      timer speed is doubled

But beside these little issues, it should run OK. Please open a bug report if you find any other things.
Comment 8 Heiko Schmidt 2005-12-12 11:00:54 UTC
Thanks for the info.

I boot with 'noacpi ide=noacpi'.
I ´ve tried to boot without bows with the newest kernel, but it stucks again.

You ´ve any idea for the problem with 800 GHz instead of 1600 MHz on the cpu ?
Comment 9 Thomas Renninger 2005-12-12 11:07:24 UTC
If cpufreq is working correct you are at 800 MHz most of the time.
As soon as you use CPU load, the processor is automatically scaled up to the performance the OS needs and then scaled down if idle again.

Check by: cat /dev/zero > /dev/null &
or glxgears
this should produce 100% CPU load and your machine should run on full power then.
If you stop the processes it should go down to 800 again.

There are machines (BIOS) that only allow lower frequencies when working on battery. If BIOS changes frequencies behind the kernels back (unplugging AC adapter), this is a bit broken at the moment. Still it should work after some ACPI events (after a minute or so). But I doubt that this is what you are seeing as it would not make much sense to let such a nice CPU always run on lowest freq, even if on battery.