Bug 148636

Created attachment 66684 [details]
boot.msg after a failed boot

Holger, sounds to me like a Thermal Zone problem, I thought we had this fixed?

hp bug. We can do nothing about this, only thomas can.
I had exactly the same here on a nc6230 until i patched the DSDT. My critical trip point now is at 6280C :-)

This also happens on older hp machines (nx5000) and always with TZ3.
Also interesting is that the values seem similar on all machines (3428 was also my "magic temperature").

Again: powersave can do nothing about it, it is the kernel who is shutting the machine down.

Please attach acpidump output.
All these strange trip points and temperature stuff happening very early seem all to be HP specific. I got quite far on the nx6125, but still have no solution...

These are i386 machines? nx6125 is a x86_64 and according to joe the trip points are exported correctly with a 32-bit kernel...

Created attachment 66929 [details]
acpidump output from hp nx8220

Find attached the acpidump output of my nx8220 Notebook. This is a Pentium-M based notebook. So yes, it's a 32bit i386 machine.

This is not 100% reproducable, right? Only here and there it happens?
If not, maybe stressing the system by key strokes helps to reproduce, or whatever?

If it is reproduceable, please try:
acpi_serialize boot param
Hmm, not sure, but there should be a acpi_not_serialized boot param to get this fixed... as the _TMP function already is serialized...
However, if this is reproducable, please try it. If it does not work I'd like to test whether setting the _TMP function NotSerialized helps, I can provide you with a patched DSDT then.
Please also have a look at ACPI semaphore/mutex errors with/without the boot param in dmesg.

(In reply to comment #6)
> This is not 100% reproducable, right?
Correct.

> Only here and there it happens?
Correct. 

> If not, maybe stressing the system by key strokes helps to reproduce, or
> whatever?
Haven't tested this yet. I'll try.
 
> If it is reproduceable, please try:
> acpi_serialize boot param
> Hmm, not sure, but there should be a acpi_not_serialized boot param to get this
> fixed... as the _TMP function already is serialized...
> However, if this is reproducable, please try it. If it does not work I'd like
> to test whether setting the _TMP function NotSerialized helps, I can provide
> you with a patched DSDT then.
At the moment it is not really reproducable.

> Please also have a look at ACPI semaphore/mutex errors with/without the boot
> param in dmesg.
 

(In reply to comment #6)
> If not, maybe stressing the system by key strokes helps to reproduce, or
> whatever?
This does not make it reproducable as well. But since I installed Beta4 on that machine I haven't seen the problem a single time.

In Beta4 there is new ACPI stuff in, also more ACPI error/warning messages.
Would be nice if you can post boot.msg again if it happens, I expect some semaphore/mutex warnings shortly before this happens.
I close the bug for now, maybe it really got fixed by the ACPI upgrade.
Please reopen if you should see this again.

Created attachment 85018 [details]
Boot message of HP DV8288EA

Having exactly the same experience: laptop shuts of during boot with temperature related messages. Specs:
- HP Pavilion DV8288EA, T2400 Core Duo laptop
- Running Suse 10.1.
- Kernel 2.6.16.13-4-bigsmp (bigsmp apparantly necessary to enable my 3945 Intel Pro Wireless)
- Installed fine doing a normal (i.e. not an ACPI disabled) install. Nevertheless it only boots with acpi=off.

Can you try the latest kernel of the day from here:
ftp.suse.com/pub/projects/kernel/kotd/i386/HEAD/kernel-default.i586.rpm

Possibly this changelog could fix it:
* Mon May 22 2006 - trenn@suse.de
- patches.fixes/acpi_ec_dummy.patch: Remove fake EC driver
  unconditionally (kernel #6111).

But it's only a guess, would be nice if you could report back whether you still are able to run into it with this kernel.

I see you are trying to use lm_sensors?
lm_sensors and ACPI is evil. I could imagine this only happens when you try to load lm_sensors modules?

(In reply to comment #13)
> I see you are trying to use lm_sensors?
> lm_sensors and ACPI is evil. I could imagine this only happens when you try to
> load lm_sensors modules?
> 

I disabled lm_sensors in yast2/System/System Services but unfortunately that didn't help.

(In reply to comment #12)
> Can you try the latest kernel of the day from here:
> ftp.suse.com/pub/projects/kernel/kotd/i386/HEAD/kernel-default.i586.rpm

Will try that later today or tomorrow. It's a production machine and I don't want to trash it today ;)

By the way, isn't this (more or less??) the same bug as bug #175702?

(In reply to comment #12)
> Can you try the latest kernel of the day from here:
> ftp.suse.com/pub/projects/kernel/kotd/i386/HEAD/kernel-default.i586.rpm

I tried the KOTD (bigsmp) but that didn't solve the problem either. Additionally, my IPW3945 wireless card didn't work with that kernel, although I didn't really try very hard to figure out why it didn't work. With 2.6.16.13-4-bigsmp it works out of the box.

> 
> Possibly this changelog could fix it:
> * Mon May 22 2006 - trenn@suse.de
> - patches.fixes/acpi_ec_dummy.patch: Remove fake EC driver
>   unconditionally (kernel #6111).

Sorry, I don't know what you are referring to. Where can I get that patch and how do I install?

> 
> But it's only a guess, would be nice if you could report back whether you still
> are able to run into it with this kernel.
> 

comment #15:
It's similar. But your machine seems to shutdown with an uninitalised temp value (something with 3000 C) and at bug #175702 it shutsdown with 0 C. Both values totally make no sense, but something is different here...

You find a recent kernel of the day here (i386/single core):
ftp.suse.com/pub/projects/kernel/kotd/i386/HEAD/kernel-default.i586.rpm