Bug 129954

Summary:	processor and thermal modules in mkinitrd freeze IBM thinkpad R40e
Product:	[openSUSE] SUSE LINUX 10.0	Reporter:	Gerald Weber <geraldweber>
Component:	Kernel	Assignee:	Olaf Kirch <okir>
Status:	RESOLVED FIXED	QA Contact:	E-mail List <qa-bugs>
Severity:	Normal
Priority:	P5 - None
Version:	Final
Target Milestone:	---
Hardware:	i686
OS:	SuSE Linux 10.0
Whiteboard:
Found By:	Customer	Services Priority:
Business Priority:		Blocker:	---
Marketing QA Status:	---	IT Deployment:	---
Attachments:	result of dmidecode Thinkpad R40e blacklist all BIOS versions

Description Gerald Weber 2005-10-21 11:03:02 UTC

If the processor & thermal modules are loaded 
via /etc/sysconfig/system/kernel,INITRD_MODULES the system will freeze on 
startup on a IBM laptop Thinkpad R40e (model 2684-HVG). 
 
If started with acpi=off, the system runs but will freeze on any acpi event 
such as closing the lid or disconnecting AC power. Reboot/halt will not work. 
 
Upgrade from 9.3 had to be performed with failsafe option, but system would 
not restart after installing the first CD. 
 
Additionally the system performs very badly when processor module is running  
(/etc/sysconfig/powermanagement, ACPI_MODULES), it will easily freeze in this 
case.

Comment 1 Hubert Mantel 2005-10-21 12:53:06 UTC

Product ships already, so BLOCKER is not appropriate.

Comment 2 Thomas Renninger 2005-10-21 13:16:28 UTC

Can you provide a backtrace when the laptop is freezing via serial console or sysreq-T?

You can also try:
  - boot with acpi=off
  - throw out thermal, processor, fan modules in /etc/sysconfig/kernel for 
    INITRD_MODULES
  - mkinitrd
  - "chkconfig powersaved off"
    disabled powersaved (so that acpi modules are not loaded later)
  - reboot without acpi=off (does the system boot properly?)
  - then try to load the acpi modules and try to catch and send the backtrace
    "modprobe processor" ...

Comment 3 Gerald Weber 2005-10-21 15:56:57 UTC

I've removed the modules thermal and processor from INITRD_MODULES, system now boots correctly also with powersaved running.

I'll provide the other backtraces as soon as I can find some time.

Comment 4 Gerald Weber 2005-10-21 18:51:29 UTC

(In reply to comment #2)
> You can also try:
>   - boot with acpi=off
>   - throw out thermal, processor, fan modules in /etc/sysconfig/kernel for 
>     INITRD_MODULES
>   - mkinitrd
>   - "chkconfig powersaved off"
>     disabled powersaved (so that acpi modules are not loaded later)

I did this, but lsmod still shows modules fan, battery, button and ibm_acpi running.

>   - reboot without acpi=off (does the system boot properly?)

It does.

>   - then try to load the acpi modules and try to catch and send the backtrace
>     "modprobe processor" ...

First echo 1 > /proc/sys/kernel/sysreq, then loaded module, computer freezes, but no backtrace on <alt><sysrq>t. Any ideas?

Let me tell you that the processor module never worked for this laptop (SuSE 8.2, 9.0-.3), so this is probably hopeless. I reported this bug because it took me ages to figure out that I had to remove the processor&hermal modules from INITRD_MODULES. Never had this problem with SuSE before. Just booting with acpi=off is unfortunately not an option because laptop then freezes, e.g. when disconnecting the AC plug.

Many thanks, Gerald

Comment 5 Thomas Renninger 2005-10-24 07:57:34 UTC

You probably have one of the systems where C-states freeze the machine?
Try to load the processor module with nocst=1 or max_cstate=0. Does the machine still lock up? If no you might want to try with max_cstate=1 (maybe at least C1 is working?).
If this is the case, please attach dmidecode, so that we can blacklist your machine.

Comment 6 Gerald Weber 2005-10-24 16:49:38 UTC

Created attachment 55244 [details]
result of dmidecode

Comment 7 Gerald Weber 2005-10-24 16:58:30 UTC

(In reply to comment #5)
> Try to load the processor module with nocst=1 or max_cstate=0. Does the machine
> still lock up? If no you might want to try with max_cstate=1 (maybe at least C1
> is working?).

nocst=1 freezes
max_cstate=0 OK 
max_cstate=1 OK
max_cstate=2 freezes (don't know if it makes sense, but tried anyway)

for max_cstate=0,1 I have only loaded the module to see if it runs, have not
tried to use the machine with it.

> If this is the case, please attach dmidecode, so that we can blacklist your
> machine.
done.

Meanwhile, I also found that using powersaved was destabilising this machine.
Symptoms: X windows suddenly starts running at 100% and after a few minutes the system, would freeze. When I managed to restart X windows before freezing the symptom persisted, only a reboot would bring it back. When I stopped powersaved from running the problem went away. I was having this problem since upgrade to SuSE 9.3 (even with all updates). 
Should I register this as a separate bug?

Comment 8 Thomas Renninger 2005-10-25 07:02:05 UTC

Your machine is already blacklisted.
Please upgrade BIOS and report back if there should be a newer version than the already blacklisted one (currently blacklisted BIOS: 1SET60WW).

> powersaved was destabilising this machine
-> Hmm, maybe it is throttling? Best is you open yast->System->Powermanagement and disable throttling for the powersave and performance scheme.
Could also be something else, first upgrade BIOS, please.

Comment 9 Thomas Renninger 2005-10-31 16:12:45 UTC

Created attachment 56014 [details]
Thinkpad R40e blacklist all BIOS versions

This one has been posted on acpi-devel lately (by akpm@osdl.org - will probably go mainline?) (be careful for whitespaces...).

Olaf, Andi, should we also add it?

Comment 10 Thomas Renninger 2005-10-31 16:15:30 UTC

Olaf, could you comment on comment #8 and the patch in comment #9 please and add it if you think we should do, thanks.

Comment 11 Olaf Kirch 2005-11-02 11:07:41 UTC

According to the latest update from kernel.org bug 3549, the 1SET69WW BIOS
needs to be listed too.

I've added the patch to SL100; will add to HEAD as well.

Comment 12 Olaf Kirch 2005-11-02 11:41:05 UTC

Done.