Cyrix 6x86 Comma Bug 2.6

Cyrix 6x86 Comma Bug 2.6

[Matthias-Christian Ott]

Hi!
I have some problems with my Cyrix cpu it stop some times and ugly 
tokens appear on the screen. Is the Comma Bug the reason (I test the 
hdd, memory and pci cards -- they are all ok)? Is the comma bug "fixed"? 
If not so, how to do this?

Thanks
Matthias-Christian Ott

Re: Cyrix 6x86 Comma Bug 2.6

[Denis Vlasenko]

On Saturday 18 December 2004 13:52, Matthias-Christian Ott wrote:
> Hi!
> I have some problems with my Cyrix cpu it stop some times and ugly 
> tokens appear on the screen. Is the Comma Bug the reason (I test the 
> hdd, memory and pci cards -- they are all ok)? Is the comma bug "fixed"? 
> If not so, how to do this?

It is very unlikely that you see "Coma" bug. It can be triggered only
by deliberately coded tight endless loop. "Ugly tokens on the screen"
suggest that you see something else.

arch/i386/kernel/cpu/mtrr/changelog says:
    19990218   ZoltАn BЖszЖrmИnyi <zboszor@mail.externet.hu>
               Remove Cyrix "coma bug" workaround from here.
               Moved to linux/arch/i386/kernel/setup.c and
               linux/include/asm-i386/bugs.h

but neither of those files mention "coma"...

PS: "coma" bug won't trigger if automatic locking of
xchg insn is disabled (yes, Cyrix can be configured to do this).
This shall be done for performance reasons too.
No SMP dangers since Cyrix CPUs aren't used in SMP...

So grep the source / google Inet for this if you are still interested.
--
vda

Re: Cyrix 6x86 Comma Bug 2.6

[Alan Cox]

On Llu, 2004-12-20 at 22:26, Denis Vlasenko wrote:
> It is very unlikely that you see "Coma" bug. It can be triggered only
> by deliberately coded tight endless loop. "Ugly tokens on the screen"
> suggest that you see something else.

Presumably those tokens included "Oops" somewhere near the top and
function names. The Cyrix stuff is notoriously hard to keep cool so that
may be a good thing to check, as well as running memtest86+ to check the
RAM.

Also some very early stepping 6x86 Cyrixes simply don't run Linux
reliably and it seemed to be cache problems in the CPU.

Re: Cyrix 6x86 Comma Bug 2.6

[Matthias-Christian Ott]

Alan Cox wrote:

>On Llu, 2004-12-20 at 22:26, Denis Vlasenko wrote:
>  
>
>>It is very unlikely that you see "Coma" bug. It can be triggered only
>>by deliberately coded tight endless loop. "Ugly tokens on the screen"
>>suggest that you see something else.
>>    
>>
>
>Presumably those tokens included "Oops" somewhere near the top and
>function names. The Cyrix stuff is notoriously hard to keep cool so that
>may be a good thing to check, as well as running memtest86+ to check the
>RAM.
>
>Also some very early stepping 6x86 Cyrixes simply don't run Linux
>reliably and it seemed to be cache problems in the CPU.
>-
>To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
>the body of a message to majordomo@vger.kernel.org
>More majordomo info at  http://vger.kernel.org/majordomo-info.html
>Please read the FAQ at  http://www.tux.org/lkml/
>
>  
>
Hi!
The problem wasn't the comma bug, one of the pci slots is broken. But 
thanks for your interest.

Sincerely
Matthias-Christian Ott