Idiot-proof APIC?

Idiot-proof APIC?

[Allan Sandfeld]

Hi, I just want to share some of my stupidity and my experience with it with 
you. 
I recently had the misfortune to try to put two celerons on an SMP-board. The 
bios correctly ignored the second cpu, but the linux-kernel(2.4.17). Would 
boot almost normally then emit two APIC-errors to the console(error 2 and 
6?), and shortly after freeze completely. After one of the celerons was 
removed linux was completely stable. Something inside makes me question 
whether or not the APIC people have taken idiots into consideration. The 
kernel should detect two cpu, detect they are not SMP and then operate using 
just one. Not very importent, but correct behavior.
Anyway for specs the board it was an Acorp 6A815EPD. Proberbly the one of the 
only SMP i815 mortherboards in the world.

-Allan

Re: Idiot-proof APIC?

[Mike Fedyk]

On Wed, Feb 20, 2002 at 08:07:47PM +0100, Allan Sandfeld wrote:
> Hi, I just want to share some of my stupidity and my experience with it with 
> you. 
> I recently had the misfortune to try to put two celerons on an SMP-board. The 
> bios correctly ignored the second cpu, but the linux-kernel(2.4.17). Would 
> boot almost normally then emit two APIC-errors to the console(error 2 and 
> 6?), and shortly after freeze completely. After one of the celerons was 
> removed linux was completely stable. Something inside makes me question 
> whether or not the APIC people have taken idiots into consideration. The 
> kernel should detect two cpu, detect they are not SMP and then operate using 
> just one. Not very importent, but correct behavior.
> Anyway for specs the board it was an Acorp 6A815EPD. Proberbly the one of the 
> only SMP i815 mortherboards in the world.
> 

Actually, with the correct adapters, celerons can be made SMP safe.

I would immagine that it is hard to detect if there is a adapter operating
correctly from the APIC code...

Mike

Re: Idiot-proof APIC?

[Mark Cooke]

On Wed, 20 Feb 2002, Mike Fedyk wrote:

> On Wed, Feb 20, 2002 at 08:07:47PM +0100, Allan Sandfeld wrote:
> > Hi, I just want to share some of my stupidity and my experience with it with 
> > you. 
> > I recently had the misfortune to try to put two celerons on an SMP-board. The 
> > bios correctly ignored the second cpu, but the linux-kernel(2.4.17). Would 
> > boot almost normally then emit two APIC-errors to the console(error 2 and 
> > 6?), and shortly after freeze completely.

<snip>
> 
> Actually, with the correct adapters, celerons can be made SMP safe.
> 
> I would immagine that it is hard to detect if there is a adapter operating
> correctly from the APIC code...

Only have to agree with Mike here.  I made the hardware mod to two 
slot-A celerons and haven't had any trouble at all from the machine.

If this is 'fixed', hopefully a command-line override will be added!

Mark

-- 
+-------------------------------------------------------------------------+
Mark Cooke                  The views expressed above are mine and are not
Systems Programmer          necessarily representative of university policy
University Of Birmingham    URL: http://www.sr.bham.ac.uk/~mpc/
+-------------------------------------------------------------------------+

Re: Idiot-proof APIC?

[Andrey Panin]

[-- Attachment #1: Type: text/plain, Size: 573 bytes --]

On Срд, Фев 20, 2002 at 08:07:47 +0100, Allan Sandfeld wrote:
> I recently had the misfortune to try to put two celerons on an SMP-board. The 

If you have Celerons based on Mendocino core you can easily hack
them to work in SMP even with Socket370 motherboard.
Celerons based on Copermine core are not SMP capable and can cause 
any brain-damaged behavior. 

And Linux can't work aroud all crazy combinations of SMP incapable CPUs :))

-- 
Andrey Panin            | Embedded systems software engineer
pazke@orbita1.ru        | PGP key: wwwkeys.eu.pgp.net

[-- Attachment #2: Type: application/pgp-signature, Size: 232 bytes --]

Re: Idiot-proof APIC?

[Maciej W. Rozycki]

On Wed, 20 Feb 2002, Allan Sandfeld wrote:

> I recently had the misfortune to try to put two celerons on an SMP-board. The 
> bios correctly ignored the second cpu, but the linux-kernel(2.4.17). Would 
> boot almost normally then emit two APIC-errors to the console(error 2 and 
> 6?), and shortly after freeze completely. After one of the celerons was 
> removed linux was completely stable. Something inside makes me question 
> whether or not the APIC people have taken idiots into consideration. The 
> kernel should detect two cpu, detect they are not SMP and then operate using 
> just one. Not very importent, but correct behavior.

 Please report a full bootstrap log when started with the "debug" option
if you want anything to be done about it.  Actually it's possible we can
handle such a configuration, but given the BIOS was unable to initialize
the second CPU properly (possibly due to a bug or a misfeature) we may
need to add some sanitizing code.  APIC errors are relatively
insignificant issue here, especially if you only get a few of them --
people report lots of them on systems that should be completely stable (I
have yet to see one myself, but maybe I'm just using weird systems that
completely refuse to generate them... ;-) ). 

-- 
+  Maciej W. Rozycki, Technical University of Gdansk, Poland   +
+--------------------------------------------------------------+
+        e-mail: macro@ds2.pg.gda.pl, PGP key available        +