HP Compaq nc6320 - Booting on battery power causes softlock

Hi, I've got an issue with my HP Compaq nc6320, where it won't boot off its battery without softlocking on CPU#0. I remember it booting ok off battery occasionally in the past, but once I've seen the fault - the only way I can get it to boot is off AC. (I usually boot off AC anyway).

I'm presuming ACPI as a good starting point, since that is an area where AC / Battery may make a difference.

Can anyone suggest a debugging procedure to identify the cause of a softlock, and where would be appropriate (which category) to file a kernel bug for this?

Peter Clifton

Electrical Engineering Division,
Engineering Department,
University of Cambridge,
9, JJ Thomson Avenue,
Cambridge
CB3 0FA

Tel: +44 (0)7729 980173 - (No signal in the lab!)

The first thing is to make sure you are running the latest bios, kernel. Then, you need to sort out the kernel component/modules/drivers that cause the problem.
If you guess it is ACPI issue, pleas test boot option acpi=off to see if the problem goes away. If you guess it is battery issue, please remove battery module, etc.
Please don't forget to enter a bug in bugzilla.kernel.org.

Thanks,
Luming

try also "apm=off" -- since the ACPI/APM dependency checking is currently broken in some configs.

Ok, I've tested acpi=off and apm=off with the stock Ubuntu kernel. It didn't softlock with either, but I'll try to use one of the two when booting now to ensure it is consistent.

with acpi=off, there was a backtrace printed from an unloved IRQ11, and the sound card wasn't particularly happy. These HP laptops seem very picky - perhaps I should buy an IBM next time for linux!

I've filed a bug, #7840, http://bugzilla.kernel.org/show_bug.cgi?id=7840

It is under ACPI/Other. Attached to that bug is dmesg output from the normal AC boot, and battery boots with acpi=off, api=off, also an acpidump output from the normal AC boot.

I will check the bios, and vanilla kernel tests when I get chance. Unfortunately, it seems quite hard to get a kernel config which exactly matches a stock Ubuntu one (for keeping as few variables changed as possible). I can recompile a kernel source deb for Ubuntu quite happily, but it produces a huge number of versions, and takes many many hours (brings me back to using linux on a [34]86 :))

Thanks for your input Luming, Len.

Regards,

Peter Clifton

Ok, having tried some more, apm=off can still give a softlock. Will keep trying with acpi=off.

Peter C.

AFAIK, I'm on the current bios.. I updated it when I initially had problems with CPU Frequency scaling. I'll check.

Kernel.. still debugging with the Ubuntu one, but will try a fresh vanilla one when I get free time (not needing the machine for work).

I've tried booting with init=/bin/sh to drop me into the initrd of the system, then grabbed an acpidump when booting on battery. It was the same as AC.

Manually bootstrapping the root file-system, (not via normal init), allowed me to play about with what might cause the lockup, and it seems that it may be related to the Intel wireless driver: ipw3945.

When I loaded it, it failed to find firmware - no crash. (I'd not got udev running to help hotplug the firmware for it).

I loaded udev, and re-tried loading the wireless. Appologies that this crash dump is a type-up, I think the key points should be here - no time-stamps, but a full trace is listed.

[ loading ipw3945 module via modprobe ]
...
[ Detects ]
[ Prints module copyright ]
ACPI: PCI Interrupt 0000:08:00.00[A] -> GSI16 (level,low) -> IRQ 169
ipw3945: Detected .... 3945ABG ....

BUG: Soft lockup ... CPU#0

<c01481cf> softlockup_tick+0x9f/0xf0 <c012bee1> update_process_times
+0x31/0x80

<c0114d13> smp_apic_timer_interrupt+0x53/0x60 <c010413c>
apic_timer_interrupt+0x1c/0x30

<f8a240f7> ipw_bg_alive_start+0x67/0xa0 [ipw3945] <c0132702>
run_workqueue+0x72/0xf0

<f8a24090> ipw_bg_alive_start+0x0/0xa0 [ipw3945] <c01331d0>
worker_thread+0x0/0x140

<c0135f8b> kthread+0xab/0xe0 <c0135ee0> ktherad+0x0/0e0

<c0101005> kernel_thread_helper+0x5/0x10

Hope that helps cast some light on the problem..

Regards

Peter Clifton

Never-mind.. looks like it has been discovered elsewhere..

http://bughost.org/bugzilla/show_bug.cgi?id=1096

I'm not sure why it only happens on battery though.

Thanks for your help.

Peter Clifton