smp freeze on 2.4.9

smp freeze on 2.4.9

[Philippe Amelant]

Hi,

I have an ABIT BP6 mobo with 2 celeron 400 running redhat 7.1 with 2.4.3
SMP (from kernel.org not from redhat)
it works well.
But when i try to update my kernel, i always get a lockups in few
minutes. sometimes kernel just freeze when booting.
I have tried other kernel such as 2.4.5, 2.4.6 and 2.4.7 and I always
get the same result.
but if i compile a non SMP kernel the box is stable.
Is it a know problem ?
are there a work around ?

thanks

Re: smp freeze on 2.4.9

[john slee]

On Thu, Aug 30, 2001 at 12:10:37PM +0200, Philippe Amelant wrote:
> I have an ABIT BP6 mobo with 2 celeron 400 running redhat 7.1 with 2.4.3

before you blame smp, try the usual bp6 stuff:
*	bigger/better power supply
*	better cooling
*	boot with 'noapic' on commandline

search a linux-kernel archive (http://marc.theaimsgroup.com)
for more info.  these boards seem to be a bit of a lucky dip.  some
never have any problems, others have heaps.  i have a vague memory of
someone mentioning flaky caps on some revisions...  also are you using
the onboard ata66 controller?  there's been a fair few reports of
trouble with those, not sure if it was fixed/hacked-around or not.

best of luck,

j.

-- 
R N G G   "Well, there it goes again... And we just sit 
 I G G G   here without opposable thumbs." -- gary larson

Re: smp freeze on 2.4.9

[Philippe Amelant]

On jeu, 2001-08-30 at 14:17, john slee wrote:
> On Thu, Aug 30, 2001 at 12:10:37PM +0200, Philippe Amelant wrote:
> > I have an ABIT BP6 mobo with 2 celeron 400 running redhat 7.1 with 2.4.3
> 
> before you blame smp, try the usual bp6 stuff:
> *	bigger/better power supply

350 W should be enough (just 1 HD and 1 DVD + mobo ) ?

> *	better cooling

I have 2 big fan :)
cpu temp is typically 35 °C

> *	boot with 'noapic' on commandline
> 

interresting, i notice that i have some error apic in kernel message
with 2.4.3
i will search that on lkml archive

thank for response

> search a linux-kernel archive (http://marc.theaimsgroup.com)
> for more info.  these boards seem to be a bit of a lucky dip.  some
> never have any problems, others have heaps.  i have a vague memory of
> someone mentioning flaky caps on some revisions...  also are you using
> the onboard ata66 controller?  there's been a fair few reports of
> trouble with those, not sure if it was fixed/hacked-around or not.
> 
> best of luck,
> 
> j.
> 

Re: smp freeze on 2.4.9

[Alan Cox]

> interresting, i notice that i have some error apic in kernel message
> with 2.4.3
> i will search that on lkml archive

Lots of apic errors imply problems on the link between the processors and
io controller. A few is basically ok (there is a checksum) but a huge number
and one day it'll checksum a bad frame ok and you are history

There is also a problem with Linus tree where apic errors causing event
replays where the erroring component was not the CPUs can cause crashes.
That is fixed in -ac.

Alan

Re: smp freeze on 2.4.9

[Philippe Amelant]

On jeu, 2001-08-30 at 15:06, Alan Cox wrote:
> > interresting, i notice that i have some error apic in kernel message
> > with 2.4.3
> > i will search that on lkml archive
> 
> Lots of apic errors imply problems on the link between the processors and
> io controller. A few is basically ok (there is a checksum) but a huge number
> and one day it'll checksum a bad frame ok and you are history
> 

So I think it's ok for my 2.4.3, I just have about 5 ~ 10 apic error by
hour

> There is also a problem with Linus tree where apic errors causing event
> replays where the erroring component was not the CPUs can cause crashes.
> That is fixed in -ac.

I have tried 2.4.9-ac4 and i still have a freeze in few minutes, but
when i configured it with noapic all seem ok, I use the box 3 hours
without freeze

Thank
> 
> Alan
>