SMBALERT# thermal sensor signal for Intel PII in the kernel ?

SMBALERT# thermal sensor signal for Intel PII in the kernel ?

[Florent CHANTRET]

Hi geeks and guru's ;o)

Even if I'm a developper, I'm not there for the moment to help you
developping this wonderful Linux Kernel. Later, perhaps ;o)

But, I've a problem with my laptop Sony VAIO powered by a bugged as hell
Intel PII Celeron (lot of customers from Sony have the same problem and
there is no support from this fuckin' company, nor from Intel, and I don't
think contacting Micro$$$$oft is the good thing to do).

My laptop randomly shutdown due to a bug in the thermal sensor of the PII
sending a SMBALERT# signal. There is no solution for NT fuckin' OS (only a
trick on "WinDaube" 98 (french expression for WinShit ;o))

I don't want to be on an unstable fuckin' system, so I decided to
permanently switched to Linux (I was always reticent, I was considering
Linux was the best stuff for server but not so ready yet for desktop :
crappy fonts, some applications missing or not so user-friendly, but I'm
changing my mind), cause there is a real support, and tha ability to see the
source code. Open Source ruleeezzzzzzzzzz ;o)

But let's go to the problem :

Here is the Intel stuff about the bug :
http://cipsa.physik.uni-freiburg.de/~zwerger/Vaio/Intel_Mobile_Temp-Prob.pdf
Here is the fix on fuckin' Windows 98 (as a curiosity) :
http://cipsa.physik.uni-freiburg.de/~zwerger/Vaio/indexeng.htm

So there is a software way to fix this hardware bug.

So I've build a 2.4.20 kernel without  all the APM,ACPI, PM at boot, CPU
Idle and "Machine Check Exception" but I've still the problem. The kernel
still answer to an SMBALERT# which cause the machine to shutdown.

So, before asking in the ML, I've searched on Google and Google Groups
without success (there is quite nothing about SMBALERT#).

I've grepped all the 2.4.20 source code but there is anywhere the presence
of SMBALERT# in a comment. I've noticed some codes about SMBus (which I
think include the SMBALERT functionnality) but I don't know if I can
deactivate SMBus in the kernel config and if it is safe to do it.

There is several functions in the kernel for SMBus so if I must patch the
code to deactivate this precise signal, any help would be appreciated ;o)

If I can fix it, I will try to convert all the consumers of the F-serie VAIO
bugged to Linux ;o)

Thanks for your answer and sorry if this is not the good place to post but
I've searched a lot before. I don't want to erase Linux to put a "Windaube
98" help help help ;o)

Regards and respect for your great work !
Florent CHANTRET

Re: SMBALERT# thermal sensor signal for Intel PII in the kernel ?

[jw schultz]

On Tue, Jan 14, 2003 at 03:43:41PM +0100, Florent CHANTRET wrote:
> Hi geeks and guru's ;o)
> 
> But, I've a problem with my laptop Sony VAIO powered by a bugged as hell
> Intel PII Celeron (lot of customers from Sony have the same problem and
> there is no support from this fuckin' company, nor from Intel, and I don't
> think contacting Micro$$$$oft is the good thing to do).
> 
> My laptop randomly shutdown due to a bug in the thermal sensor of the PII
> sending a SMBALERT# signal. There is no solution for NT fuckin' OS (only a
> trick on "WinDaube" 98 (french expression for WinShit ;o))
> 
> Here is the Intel stuff about the bug :
> http://cipsa.physik.uni-freiburg.de/~zwerger/Vaio/Intel_Mobile_Temp-Prob.pdf
> Here is the fix on fuckin' Windows 98 (as a curiosity) :
> http://cipsa.physik.uni-freiburg.de/~zwerger/Vaio/indexeng.htm
> 
> So there is a software way to fix this hardware bug.
> 
> So I've build a 2.4.20 kernel without  all the APM,ACPI, PM at boot, CPU
> Idle and "Machine Check Exception" but I've still the problem. The kernel
> still answer to an SMBALERT# which cause the machine to shutdown.
> 
> So, before asking in the ML, I've searched on Google and Google Groups
> without success (there is quite nothing about SMBALERT#).
> 
> I've grepped all the 2.4.20 source code but there is anywhere the presence
> of SMBALERT# in a comment. I've noticed some codes about SMBus (which I
> think include the SMBALERT functionnality) but I don't know if I can
> deactivate SMBus in the kernel config and if it is safe to do it.
> 
> There is several functions in the kernel for SMBus so if I must patch the
> code to deactivate this precise signal, any help would be appreciated ;o)
> 
> If I can fix it, I will try to convert all the consumers of the F-serie VAIO
> bugged to Linux ;o)
> 
> Thanks for your answer and sorry if this is not the good place to post but
> I've searched a lot before. I don't want to erase Linux to put a "Windaube
> 98" help help help ;o)

You don't even state which model you have.  Which limits my
ability to comment.  I know i've had no problem with mine.

The best things for you to do is check the specific model
info from linux-on-laptops.com and search, then ask, on the
linux-sony list http://returntonature.com/mailman/listinfo/linux-sony

Good luck.

-- 
________________________________________________________________
	J.W. Schultz            Pegasystems Technologies
	email address:		jw@pegasys.ws

		Remember Cernan and Schmitt

Re: SMBALERT# thermal sensor signal for Intel PII in the kernel ?

[Florent CHANTRET]

This is a PCG-F305 but all the old serie F of VAIO is concerned. The problem
don't appear immediately but later when the laptop isn't covered by the
guarantee.

Thanks for your link. Anyway, this is not really a "Linux on VAIO" related
problem, but an hardware bug on the Intel PII that can be solved by a
software trick cause the laptop doesn't crash on MS-DOS, WinShit 98 (by
disabling some stuffs in the or to be open-source, on the LILO boot. So I
think I can disable something in the kernel config or I can find some lines
of code in the minimal kernel to solve this problem.

But as soon the kernel begin to boot, my laptop crah (when it is cold,
powered off for a night for example), after several reboot, when it is hot,
it can be powered for a long time (1 day, 2 day sometimes) but I've still
random shutdown.

Disabling all power managment, ACPI, APM, PM at boot, CPU, Idle and "Machine
Check Exception" and now I2C still don't solve the problem. I've compiled a
new kernel without WatchDog and I'll try soon but I'm sceptic. I think
SMBALERT# is in the SMBus protocol that is a subset of I2C and this is still
not right.

So, if someone know something about the SMBALERT# signal or where in the
kernel anywhere that in the stuff exposed above could be the order to
shutdown immediately (without the proper task of a normal linux shutdown)
the laptop, any help would be appreciated.

I repeat the link from the Intel SPEC :
http://cipsa.physik.uni-freiburg.de/~zwerger/Vaio/Intel_Mobile_Temp-Prob.pdf

This previous PDF is a subset of this one :
http://cipsa.physik.uni-freiburg.de/~zwerger/Vaio/24388741.pdf

I've suspend my searches to fix it on a Windows NT, don't want to switch to
Windows 98 so it's time for me to stay on Linux and I'm sure I can find some
option or some lines of code where I can fix the stuff.

Regards,
Florent CHANTRET



> You don't even state which model you have.  Which limits my
> ability to comment.  I know i've had no problem with mine.
>
> The best things for you to do is check the specific model
> info from linux-on-laptops.com and search, then ask, on the
> linux-sony list http://returntonature.com/mailman/listinfo/linux-sony
>
> Good luck.
>
> --
> ________________________________________________________________
> J.W. Schultz            Pegasystems Technologies
> email address: jw@pegasys.ws
>
> Remember Cernan and Schmitt
>