Acpi / timing problem with kernel 2.6.21

Lone_Wolf · 2007-05-08 20:36:45

when kernel 2.6.21 is used on my acer laptop (amd turion X2 50) the system gets very slow from the point udev uevents are started.
Pressing enter repeatedly allows the boot process to continue (although a lot slower than normal) and allows me to login to console.

The logs show several acpi related errors and this message :
clocksource tsc unstable (Delta=9372044968 ns)

With none of the 2.6.20 kernels (including 2.6.20.7 and later) i had these problems.
With Kernel 2.6.21.1-7 the problem was very large, 2.6.21.1-8 seems a bit better.

I am only using standard acpi modules.

Apart from the kernel, fglrx and madwifi everything is up to date on the partition where i have the kernel 2.6.20 installed.

oli · 2007-05-09 18:00:30

Maybe it has anything to do with HPET. If have to disable this functionality in bios of my Asus M2N-MX motherboard, to run Linux at all.

HPET is the high resolution timer, introduced with the latest kernel.

kishd · 2007-05-10 08:45:59

There is a boot parameter in kernel 2.6.21which may help

hpet=disable

tomeil · 2007-05-11 09:22:39

I confirm that with the new kernel 2.6.21.1-8 the arch boot process on my laptop (HP-NC4200) is extremely slow, while with the previous versions it was extremely fast...
The symptoms are the same described by Lon_Wolf.

Lone_Wolf · 2007-05-13 12:35:00

Since this is a laptop, i can change very little in the bios.
I upgraded to the latest available bios and saw some improvement.

The hpet=disable boot parameter doesn't have any effect.

I've done some searching and there are several reports that hpet=disabled is sometimes ignored.

It definitely looks like this is a kernel problem, not an archlinux problem.

check your /var/cache/pacman/pkg for kernel26-2.6.20.7-2.pkg.tar and install it with pacman -U , that is the last 2.6.20 kernel released by archlinux.

Edit :

Bug 7124 in flyspray looks like it is about the same problem : http://bugs.archlinux.org/task/7124

I also added a feature request to include kernel version dependency in the pkgbuild, see flypsray 7150 .

Last edited by Lone_Wolf (2007-05-13 13:18:42)

stinkytofoo · 2007-05-20 03:58:55

I'm having the exact same problem.
But the weird thing is that two days ago, when I installed arch in my new laptop, everything seemed fine (it booted up as it should). Only when today I started it up again, I experienced the problem.

Now I'm neither sure about the following or whether it has anything to do with the issue. In contrast with the previous times, this time I booted up my laptop with the battery and without the ac adapter.

Now for the time being I guess the solution will be downgrading the kernel.

Can anyone direct me to where I could get the package for the stock 2.6.20 kernel?

stinkytofoo · 2007-05-21 01:50:16

Weird!
I haven't done any relevant (AFAIK) update. But since last night, the acpi/timing problem seems to have solved by itself. My laptop is booting nicely and fast.

stinkytofoo · 2007-05-21 04:08:00

Crap!

Just for the heck of it, I booted my laptop with the battery (and no ac adapter) again. And now the problem is back.
This is getting really annoying.

Still in need of the 2.6.20 kernel (stock) package.

stinkytofoo · 2007-05-21 14:44:32

Ok... I'm back again. And this time I'm sure neither the battery or the ac adapter should be blamed.
It's the usb mouse. Yes, I'm not joking.
Oh, trust me. I did quite a lot of plug-reboot and unplug-reboot. And I'm empirically certain that the culprit is the usb mouse.

Now I check what dsmeg spitted out for both cases--with mouse and without it--and couldn't find any significant difference, if there was one at all.

I guess it's not really the end of the world to have to boot my laptop with my mouse hanging... but it would be nice if I could do it otherwise.

Doehni · 2007-05-24 01:25:44

Here you can get the 2.6.20 stock kernel: http://phraktured.net/archmirror/current/os/i686/

stinkytofoo · 2007-05-29 10:25:02

thanks!

Lone_Wolf · 2007-05-30 22:04:02

I've tried booting with my usb mouse plugged in, and it indeed solves the problem.

As soon as i unplug the mouse the problems are back, so it's not the modules that make the difference.

I'll try to check the logs to see what is activated/deactivated when the mouse is plugged in/out.

Arch Linux

#1 2007-05-08 20:36:45

Acpi / timing problem with kernel 2.6.21

#2 2007-05-09 18:00:30

Re: Acpi / timing problem with kernel 2.6.21

#3 2007-05-10 08:45:59

Re: Acpi / timing problem with kernel 2.6.21

#4 2007-05-11 09:22:39

Re: Acpi / timing problem with kernel 2.6.21

#5 2007-05-13 12:35:00

Re: Acpi / timing problem with kernel 2.6.21

#6 2007-05-20 03:58:55

Re: Acpi / timing problem with kernel 2.6.21

#7 2007-05-21 01:50:16

Re: Acpi / timing problem with kernel 2.6.21

#8 2007-05-21 04:08:00

Re: Acpi / timing problem with kernel 2.6.21

#9 2007-05-21 14:44:32

Re: Acpi / timing problem with kernel 2.6.21

#10 2007-05-24 01:25:44

Re: Acpi / timing problem with kernel 2.6.21

#11 2007-05-29 10:25:02

Re: Acpi / timing problem with kernel 2.6.21

#12 2007-05-30 22:04:02

Re: Acpi / timing problem with kernel 2.6.21

Board footer