FS#7124 - Kernel 2.6.21 boot problem

Attached to Project: Arch Linux
Opened by Roman (Riot777) - Friday, 11 May 2007, 10:43 GMT
Last edited by Tobias Powalowski (tpowa) - Friday, 17 August 2007, 18:01 GMT
Task Type Bug Report
Category Kernel
Status Closed
Assigned To Tobias Powalowski (tpowa)
Architecture i686
Severity High
Priority Normal
Reported Version 0.8 Voodoo
Due in Version Undecided
Due Date Undecided
Percent Complete 100%
Votes 4
Private No

Details

I have upgraded few days ago kernel 2.6.20 to latest 2.6.21 from current repository. After rebooting the machine I have notice that it hangs at load of "Udev uevents" process. After whole day of tests, and downgrading random packages udev and kernel etc. I discovered that the problem lays in the newest build of Kernel 2.6.21 for arch and is related probably with ACPI. But it worked fine on 2.6.20 and I didn't had any problems there. If I used acpi=off command the system boots little slower than always and some modules that I use didn't loaded properly but loaded. Now I can't ofcourse turn off computer automaticly so it's a big issue for me at current state. I got a pc with AMD processor and motherboard based on SiS 735 chipset that can be some sort for tip for ppl who will work to fix this.
This task depends upon

Closed by  Tobias Powalowski (tpowa)
Friday, 17 August 2007, 18:01 GMT
Reason for closing:  Won't fix
Comment by Roman (Riot777) - Friday, 11 May 2007, 14:49 GMT
I have also compared modules for acpi available for both kernels:

ls -l /lib/modules/2.6.21-ARCH/kernel/drivers/acpi
razem 244
-rw-r--r-- 1 root root 5968 maj 6 18:39 ac.ko
-rw-r--r-- 1 root root 21216 maj 6 18:39 asus_acpi.ko
-rw-r--r-- 1 root root 11504 maj 6 18:39 battery.ko
-rw-r--r-- 1 root root 7256 maj 6 18:39 bay.ko
-rw-r--r-- 1 root root 9048 maj 6 18:39 button.ko
-rw-r--r-- 1 root root 5464 maj 6 18:39 container.ko
-rw-r--r-- 1 root root 9868 maj 6 18:39 dock.ko
-rw-r--r-- 1 root root 5884 maj 6 18:39 fan.ko
-rw-r--r-- 1 root root 6476 maj 6 18:39 i2c_ec.ko
-rw-r--r-- 1 root root 39172 maj 6 18:39 ibm_acpi.ko
-rw-r--r-- 1 root root 32180 maj 6 18:39 processor.ko
-rw-r--r-- 1 root root 16684 maj 6 18:39 sbs.ko
-rw-r--r-- 1 root root 15128 maj 6 18:39 thermal.ko
-rw-r--r-- 1 root root 14800 maj 6 18:39 toshiba_acpi.ko
-rw-r--r-- 1 root root 19148 maj 6 18:39 video.ko

ls -l /lib/modules/2.6.20-ARCH/kernel/drivers/acpi
razem 192
-rw-r--r-- 1 root root 5796 kwi 28 11:20 ac.ko
-rw-r--r-- 1 root root 21072 kwi 28 11:20 asus_acpi.ko
-rw-r--r-- 1 root root 11160 kwi 28 11:20 battery.ko
-rw-r--r-- 1 root root 8748 kwi 28 11:20 button.ko
-rw-r--r-- 1 root root 9808 kwi 28 11:20 dock.ko
-rw-r--r-- 1 root root 5584 kwi 28 11:20 fan.ko
-rw-r--r-- 1 root root 12144 kwi 28 11:20 hotkey.ko
-rw-r--r-- 1 root root 6176 kwi 28 11:20 i2c_ec.ko
-rw-r--r-- 1 root root 38892 kwi 28 11:20 ibm_acpi.ko
-rw-r--r-- 1 root root 16512 kwi 28 11:20 sbs.ko
-rw-r--r-- 1 root root 14688 kwi 28 11:20 toshiba_acpi.ko
-rw-r--r-- 1 root root 18220 kwi 28 11:20 video.ko

They differ, there was removed hotkey.ko and there were added bay.ko, container.ko, processor.ko and thermal.ko.
Maybe I should load the modules which differ to make it work, can it help somehow to get acpi back?
Comment by Tobias Powalowski (tpowa) - Friday, 11 May 2007, 16:21 GMT
acpi subsystem was upgraded during .21 difficult to say where it breaks your system :/
Comment by Roman Kyrylych (Romashka) - Saturday, 12 May 2007, 13:27 GMT
Just in case...
Did you upgrade full system?
initscripts-0.8-8 (IIRC) added automatical loading of these modules: ac battery button fan processor thermal
Comment by Roman (Riot777) - Saturday, 12 May 2007, 14:00 GMT
Yes, updated it fully, loading those modules manually didn't helped.
Comment by Lone_Wolf (Lone_Wolf) - Sunday, 13 May 2007, 13:06 GMT
I have similar problems and it seems they are related to the new hpet timer used in 2.6.21 (see http://bbs.archlinux.org/viewtopic.php?id=32786 )

A key error message in dmesg seems to be "clocksource tsc unstable (delta = xxxxxxx) " .

The boot parameter hpet=disable has no effect.
Comment by Philip Nilsson (leffe) - Monday, 14 May 2007, 10:22 GMT
Does the computer hang forever, or will it resume after a while? I often hang at the uevents, but after five minutes (or more perhaps), it will resume. However, /dev/ will be severely underpopulated, and be missing the HD nodes.
Comment by Victor Román (TecnoVM64) - Saturday, 26 May 2007, 18:04 GMT
Well my problem is quite similar but sometimes it'll boot and last a few seconds running in console and hangs instantly if X is started, I've tried checking the logs but I can't find anything.
My chipset is NVIDIA MCP51
I'm using kernel2.6.20-ck at the moment, this problem began with the 2.6.21 kernel.
Comment by GCN (gilcn) - Tuesday, 29 May 2007, 11:58 GMT
Same problem for me since 2.6.21.x entered Arch repos!

MoBo: ASUS A8N-E, CPU: Athlon64 4000+, Chipset: nVidia nForce 4 Ultra !

I had to put kernel26 in my IgnorePkg's list for now... :(
Comment by Victor Román (TecnoVM64) - Tuesday, 29 May 2007, 13:48 GMT
Well, got it fixed here with hpet=disable
Comment by Lone_Wolf (Lone_Wolf) - Sunday, 03 June 2007, 09:07 GMT
When i plugin an usb mouse , the problem instantly goes away and the laptop works better then with kernel 2.6.20 .
Unplugging the mouse brings it back instantly, so i think it's a driver related issue, not module-related.

Processor is AMD Turion XL-50, Chipset is ATI.
Comment by Felix Müller (Goldenphoenix) - Wednesday, 06 June 2007, 14:00 GMT
Same problem. hpet=disable doesn't help.
ICH6 Chipset, CPU=pentium-m!
Comment by Philip Nilsson (leffe) - Thursday, 07 June 2007, 12:39 GMT
I managed to solve my problem by modifying my initcpio. I suspect the USB part was at fault here.

MODULES="ext2 ext3 usbhid"
HOOKS="base udev autodetect pata"
to
MODULES=""
HOOKS="base udev autodetect pata usbinput filesystems"
Comment by Tobias Powalowski (tpowa) - Sunday, 24 June 2007, 09:26 GMT
status on 22rc kernel?
[kernel]
http://www.archlinux.org/~tpowa/rc-kernel/
Comment by Jesus Lazaro (srabd) - Friday, 29 June 2007, 01:19 GMT
Same problem here.

AMD Turion64x2 processor and nVidia C51D + MCP51 chipset.

kernel26-2.6.22-4 doesn't solve the problem.
Comment by Lone_Wolf (Lone_Wolf) - Tuesday, 10 July 2007, 10:44 GMT
Kernel26-2.6.22-6 from testing doesn't solve the problem.
the boot process hangs on 'checking filesystems' . i waited for 10 minutes, still [busy].
pressing enter : goes on , hangs at 'mounting local filesystems'
enter : hangs at 'removing leftover files'
enter : hangs at 'updating module dependencies'
enter : hangs at network start
repeatedly pressing enter brings me to console login
login from console works, but needs enter to show prompt.
starting mc : needs enter to become active
kernel.log shows clocksource tsc unstable

on shutdown i again have to press enter to make it continue

booting on battery or adapter doesn't make a difference
As with kernel 2.6.21 booting with usb-mouse attached solves all problems.
kernel.log shows no clocksource unstable message.

adding usbinput to mkinitcpio doesn't solve the problem, but it shows the clocksource unstable message on screen.


Comment by Felix Müller (Goldenphoenix) - Tuesday, 10 July 2007, 11:04 GMT
2.6.22 solves the Problem for me.
Comment by Jesus Lazaro (srabd) - Wednesday, 18 July 2007, 13:35 GMT
2.6.22.1 doesn't solve the problem.

I've installed Arch64 and works perfectly, 32 bits version still have the problem.

Loading...