Bug 156115

Created attachment 71798 [details]
boot.msg from first boot on CD1

Probably ACPI, reassigning to Thomas Renninger

Created attachment 71908 [details]
boot.msg with bios 1.13

I checked for the latest bios, so I upgraded from 1.06 to 1.13 with embeded controler going from 1.00a to 1.01. This does not solve the problem.

Created attachment 71910 [details]
boot.msg from running system opensuse10.1 beta6

Is there a special kernel (debug?) which I could try to get some messages on console ?

What's odd, is that the system freezes also if left alone for some time, screen blanks and no response, even ctrl-alt-SysRq doesn't work.

Created attachment 71940 [details]
last lines from /var/log/messages

saw these lines in /var/log/messages before logging restarted:
Mar  9 07:54:33 linux zmd-bin: ShutdownManager (WARN): Going to sleep...
Mar  9 08:48:25 linux syslog-ng[1786]: STATS: dropped 0
Mar  9 08:54:43 linux zmd-bin: ShutdownManager (WARN): Going to sleep...

The machine should work again fine in Beta7 (oh that one won't be public AFAIK).
You can verify by just installing the newest kernel... you find it somewhere here:
ftp://ftp.suse.com/pub/projects/kernel/kotd/i386/HEAD/kernel-default.i586.rpm

*** This bug has been marked as a duplicate of 154964 ***

I tried the suggested kernel: 2.6.16-rc5-git9-20060309130028-default
System locks when pushing volume keys, I tried the nolapic boot option, did not change anything.
But debug kernel works fine: 2.6.16-rc5-git9-20060309130028-debug

Created attachment 72213 [details]
boot.msg for default kotd

Created attachment 72214 [details]
boot.msg for default kotd whith nolapic

Created attachment 72215 [details]
boot.msg for debug kotd

That is weird.
No time right now, I will have a look at dmesg output on Monday.
Thanks for the info.

Created attachment 72285 [details]
output from acpidump

I am quite sure that the problem is solved. Maybe the kernel was not built/synced yet, but the debug kernel was? Please try again the very current kotd. If it still does not work, please also try noapic option.

The laptop work fine with previous Betas and/or SL10.0?

Stop, Kernel from comment #9 already has the fix (local apic disabled by BIOS and not reenabled). Maybe you mixed up kernels when testing?

I checked with latest kotd : default and debug version 20060323*

with the default kernel, the system freezes when pushing a volume key
choosing noapic or nolapic does not solve the problem
The debug kernel works fine
The laptop worked also fine with SL9.2 (kernel 2.6.8-24.18-default)

I am attaching boot.log files for the different combinations

Created attachment 74904 [details]
boot.msg for several combinations of kernel/options

I see it:
You get this line with default kernel (non-working, right?):
<6>IO/L-APIC allowed because system is MP or new enough

and this line with debug(SMP) kernel (working):
<6>Using APIC driver default

According to ak the debug kernel must be an SMP kernel which seem to cause the grief with APIC initialisation here.

However the APIC configs are quite fragile... Assigning Andi:
Andi: Here debug kernel works using APIC initialisation through
mach-generic/probe.c (search for "Using APIC driver default")

and default kernel does not work (going through CONFIG_X86_APIC_OFF in
arch/i386/kernel/apic dmi_check_apic (search for "IO/L-APIC allowed because
system is MP or new enough").

This machine needs APIC, but setup the way the smp kernel sets it up (For verification you can try to boot the -smp kernel).

You should boot with apic=verbose or even apic=debug to get a bit more data out of the kernel.

Created attachment 75882 [details]
boot.msg with apic=debug for several combinations of kernel/options

Created attachment 75898 [details]
boot.msg with apic=debug for several combinations of kernel/options

I tried the smp 20060329 kernel, it locks when pushing the volume keys, so smp is not working

Created attachment 75914 [details]
boot.msg with apic=debug for several combinations of kernel/options

As I had no idea, I tried removing noacpi from the boot line
Now the smp kernel works, so it seems this system needs acpi
Next, I'll retry the default kernel with acpi enabled

Created attachment 75928 [details]
boot.msg with apic=debug for several combinations of kernel/options

The default kernel works also but it needs acpi.

Now I rechecked booting the various alternatives for the very first boot (boot from CD1 opensuse10.1 beta8):
Installation : OK
Installation - ACPI disabled : kernel locks even if not pushing the keys after "Starting udev ..." is displayed
Installation - Local apic disabled : OK
Installation - Safe Settings : pushing volume up / volume down keys locks the kernel
Rescue System : OK

there is no noacpi option

you're rigth but there is acpi=off

PS: I'll be offline for one week

This is a won't fix IMO. Default config works. I agree that on these old machines apci=off and apm=on should work, but I think it is too late to change anything there and risk beakage of other machines.

CC'ing Seife from mobile team, maybe they have seen similar things with old machines were it's worth to fix acpi=off.

Can you test the latest kotd? I fixed some issues in this area now.

I tried the kernel 2.6.16-20060411175326-default with following combinations:
acpi=on apm=on ->ok
acpi=on apm=off ->ok
acpi=off apm=off -> locks when pushing on volume keys
acpi=off apm=on -> locks while booting, last line displayed at console is:
"input: AT Translated Set 2 keyboard as /class/input/input0"
The debug kernel is working with acpi=off
conclusion : not fixed by latest kotd

Can you try again and not load the ibm_acpi module.
Delete the line in /var/lib/acpi/laptop_modules (do not delete the file).
Be sure it did not get loaded accidently somehow (lsmod |grep ibm_acpi).

Ah yes, I mixed up apm/acpi is working and had a bug related to the A21m and ACPI recently. So acpi_ibm module is probably unrelated.

I tried it anyway : it was loaded previously
but commenting it out does'nt fix the problem (checked with lsmod).
Then I saw 3 ACPI related modules loaded: processor, thermal and fan are listed in INITRD_MODULES variable in /etc/sysconfig/kernel, so I removed them from this line and reinstalled the kernel.
Now only the processor module gets loaded but now it is first in lsmod list, so it is loaded very late in boot process, kernel can be blocked before.
I tried blacklisting it in /etc/modprobe.d/blacklist but it is load anyway.
When trying to rmmod the module, I got an oops.
Should I open a new bugzilla entry as it seems not to be related ?

Does the following line from boot.msg trigger some thought ?
IBM machine detected. Enabling interrupts during APM calls.

You could try with boot param: apm=no_allow_ints

Maybe apm=debug prints out an event when hitting the key?

apm=no_allow_ints does'nt solve anything

apm=debug blocks also and displays these lines:
Freeing initrd memory: 2230k freed
NET: Registered protocol family 16
PCI: PCI BIOS revision 2.10 entry at 0xfd9af, last bus=3
PCI: Using configuration type 1
ACPI: Subsystem revision 20060127
ACPI: Interpreter disabled.
Linux Plug and Play Support v0.97 (c) Adam Belay
pnp: PnP ACPI: disabled
PCI: Probing PCI hardware
PCI: Probing PCI hardware (bus 00)
Boot video device is 0000:00:04.0
PCI: Using IRQ router PIIX/ICH [8086/7198] at 0000:00:07.0
TC classifier action (bugs to netdev@vger.kernel.org cc hadi@cyberus.ca)
Setting up standard PCI resources
PCI: Bus 1, cardbus bridge: 0000:00:02.0
  IO window: 00001000-000010ff
  IO window: 00001400-000014ff
  PREFETCH window: 10000000-11ffffff
  MEM window: 12000000-13ffffff
PCI: Found IRQ 11 for device 0000:00:02.0
PCI: Sharing IRQ 11 with 0000:00:04.0
IBM machine detected. Enabling interrupts during APM calls.
apm: BIOS version 1.2 Flags 0x03 (Driver version 1.16ac)
apm: entry f000:3f98 cseg16 f000 dseg 40 cseg len ffff, dseg len ffff cseg16 len ffff
apm: Connection version 1.2

same with apm=no_allow_ints,debug

Does this still happen with RC3/RC4?

yes, I just retried with RC3.
It still locks if acpi=off boot option specified

Maybe you could try with init=/bin/bash.
If it still happens it must be somewhere in the kernel core.

If machine still runs, it is in some hardware specific module.
Then try to boot into runlevel 1 (/etc/init.d/boot.* services started).
Does it freeze then?
Try to identify which module/service could cause this if init=/bin/bash does not freeze the machine, but in higher runlevels it freezes.
If this is the case you might also want to send lsmod of minimal loaded modules loaded that freeze the machine, maybe we can identify a canditate ...

Created attachment 82509 [details]
dmesg with "init=/bin/bash"

I tried the suggested "init=/bin/bash", if I don't push the volume keys, the shell comes up: I saved the dmesg (attached). As soon as I push one of the volume keys, I locks.
I retried with the same arguments, this time pushing repeatedly the volume keys until the system locks: the last displayed line was line 84 (I reproduced it several times):
isapnp: Scanning for PnP cards...
Line 71 prompts me to ask the following question:
Is it possible to dump interrupt vectors to compare locking and non locking configs ? 
(IBM machine detected. Enabling interrupts during APM calls.)

Bug and machine are quite old...