Latest working kernel version: Not known
Earliest failing kernel version: 2,6,26 (possibly earlier)
Distribution: openSUSE (11.0+, possibly earlier), Ubuntu (8.04+, possibly earlier)
Hardware Environment: Gigabyte GA-MA69VM-S2 motherboard, AMD Athlon 64 X2 4800+ 2.5GHz,
virtualisation AMD-V On
Software Environment: openSUSE
Problem Description: System freezes after boot on i386 versions of Ubuntu or openSUSE on GA-MA69VM-S2 if "acpi=off" append option not used.
Steps to reproduce: Boot i386 version of openSUSE or Ubuntu (custom kernel 188.8.131.52 too) without the acpi=off switch on Gigabyte GA-MA69VM-S2. System freezes, with no disk activity.
From the problem description it seems that the system can't be booted with ACPI enabled. But it can be booted with ACPI disabled.
Will you please try the following boot options and see whether the system can be booted with ACPI enabled?
b. processor.max_cstate=1 (The processor should be compiled as built-in kernel).
It will be great if you can attach the output of acpidump.
can you get a screenshot when the system freezes?
Created attachment 19377 [details]
output of acpidump on working system (kubuntu 64-bit 8.04)
Output from acpidump on working system: Kubuntu 8.04 (kernel 2.6.24)
The system (i386 Kubuntu 8.10) will not boot even with those options. Booting with the options "verbose" instead of "splash silent" shows that high up in the screen, it always freezes with a message near the top that says "clocksource tsc unstable", delta = 299 seconds, but it goes on and loads things and it stops right after loading the ioschedulers (default = cfq in this case)
Using the option "hpet=disable" fixes the problem, I can't boot without either acpi=off or hpet=disable.
This could be a known BIOS bug, if it is what I think it is you should either:
- update your BIOS
- try the patch from commit id: a6825f1c1fa83b1e92b6715ee5771a4d6524d3b9
This should be fixed by in latest kernel, best just try the latest.
This should also be fixed if you try the latest OpenSUSE update kernel.
It's not a BIOS bug (I have the latest BIOS revision by the way), it's an HPET bug in the Linux kernel, a similar bug has been filed for amd64 and has now been fixed in 2.6.28. I shall try 2.6.28 (.1) on i386 tonight with no acpi=off or hpet=disable flags and close this bug with PATCH_ALREADY_AVAILABLE if it works.
Works on self-compiled Hardy 2.6.28 system.