lotsa oops - 2.6.5 (preempt + unable handle virutal address + more?)

lotsa oops - 2.6.5 (preempt + unable handle virutal address + more?)

[Pods]

Hi, sorry for the not very descriptive subject.

I have a problem with linux 2.6.5, it crashes, a lot. Some times it spews
out oops at me, sometimes i dotn know if it oopsed or not because the
machine doesnt respond.

I can reproduce this by several means, including playing a dvd, compiling
firefox, mplayer or thunderbird and i can be doing something not quite a
hard on the system such as browsing the web or chatting... It'll find some
way to piss me off.

Now, i have supplied several oops, each on containing the full output of
dmesg via the kernel logging daemon. I have also supplied my hardware specs
(mobo + lspci -vvv) and out put of ver_linux?... All these files can be
found @ http://users.quickfox.org/~pods/linux-issues/.

I didnt want to post them here because each oops IIRC is different. Just to
let you know if have tried several way in which to correct this.

* several different kernels ( 2.4.18-bf24, 2.4.26, 2.6.5 )
** with ide shareirq
** noapic
** + apic
** noacpi
** + acpi
** lots of other differences

* Recently i had turned to bios and have disabled many things including apic
(just incase you where wondering, bios had mps @ 1.4)

If you would rather me add the very large amoung of oops to this email, tell
me and i'll do so. Thanks in advance.

P.S Im not subscribed to the list, i may consider it depending on the amount
of feedback i get and how involved i need to be, but for now i'd rather if
people could CC me their response. Thank you
again.

________________________________________________

Message sent using
Dodo Internet Webmail Server

Re: lotsa oops - 2.6.5 (preempt + unable handle virutal address + more?)

[Anupam Kapoor]

looks like you are using nvidia ?

anupam

"Pods" <pods@dodo.com.au> writes:

>    Hi, sorry for the not very descriptive subject.
>
>    I have a problem with linux 2.6.5, it crashes, a lot. Some times it
>    spews
>    out oops at me, sometimes i dotn know if it oopsed or not because the
>    machine doesnt respond.
>
>    I can reproduce this by several means, including playing a dvd,
>    compiling
>    firefox, mplayer or thunderbird and i can be doing something not quite
>    a
>    hard on the system such as browsing the web or chatting... It'll find
>    some
>    way to piss me off.
>
>    Now, i have supplied several oops, each on containing the full output
>    of
>    dmesg via the kernel logging daemon. I have also supplied my hardware
>    specs
>    (mobo + lspci -vvv) and out put of ver_linux?... All these files can
>    be
>    found @ [1]http://users.quickfox.org/~pods/linux-issues/.
>
>    I didnt want to post them here because each oops IIRC is different.
>    Just to
>    let you know if have tried several way in which to correct this.
>
>    * several different kernels ( 2.4.18-bf24, 2.4.26, 2.6.5 )
>    ** with ide shareirq
>    ** noapic
>    ** + apic
>    ** noacpi
>    ** + acpi
>    ** lots of other differences
>
>    * Recently i had turned to bios and have disabled many things
>    including apic
>    (just incase you where wondering, bios had mps @ 1.4)
>
>    If you would rather me add the very large amoung of oops to this
>    email, tell
>    me and i'll do so. Thanks in advance.
>
>    P.S Im not subscribed to the list, i may consider it depending on the
>    amount
>    of feedback i get and how involved i need to be, but for now i'd
>    rather if
>    people could CC me their response. Thank you
>    again.
>
>    ________________________________________________
>
>    Message sent using
>    Dodo Internet Webmail Server
>
>    -
>    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  [2]http://vger.kernel.org/majordomo-info.html
>    Please read the FAQ at  [3]http://www.tux.org/lkml/
>
> References
>
>    1. http://users.quickfox.org/~pods/linux-issues/
>    2. http://vger.kernel.org/majordomo-info.html
>    3. http://www.tux.org/lkml/

Re: lotsa oops - 2.6.5 (preempt + unable handle virutal address + more?)

[Denis Vlasenko]

On Friday 04 June 2004 09:26, Anupam Kapoor wrote:
> looks like you are using nvidia ?

several oopses are marked as "Not tainted".

Anyway, your hardware might be flakey.

Run memtest86 overnight and some of
cpuburn tools too. Do heavy PCI traffic
(dd your disk to /dev/null, flood
your ethernet, etc).

Underclock your system and/or set lower
IDE DMA mode and see whether it stops oopsing.
--
vda

Re: lotsa oops - 2.6.5 (preempt + unable handle virutal address + more?)

[Pods]

>> Run memtest86
Tried once before, didnt boot... died :( memtest page says this happens on
some hardware and they're not sure why.

>> Underclock your system
its not over clocked :/

>> lower IDE DMA mode and see whether it stops oopsing.
tried setting dma = off in hdparm, didnt work, still got crashes compiling
firebird.. infact, it didnt even get passed the ./configure stage :(

Each time (just about, didnt do it once out of 6ish times) i set the dma to
off (either at boot or runtime) i got a "spurious 8259A interript: IRQ 7"..
apparantly, iirc that debug message has been taken out of 2.6.6

Please guys, CC me your responces, otherwise i have to look at some
archives, and reply via webmail... it took 2 firefox crashes (one of which
brought down X) to just right this message... now im going to take a chance
and hit the submit
button

________________________________________________

Message sent using
Dodo Internet Webmail Server

Re: lotsa oops - 2.6.5 (preempt + unable handle virutal address + more?)

[Denis Vlasenko]

On Saturday 05 June 2004 18:27, Pods wrote:
> >> Run memtest86
> Tried once before, didnt boot... died :( memtest page says this happens on
> some hardware and they're not sure why.

Ho hum. Looks like good starting point to debug.
Remove RAM until you have <= 2 GB, then run memtest86
(or modified versions of it with debug prints)
until you know where it dies.

> >> Underclock your system
> its not over clocked :/

I mean, underclock below normal conditions.

> >> lower IDE DMA mode and see whether it stops oopsing.
>
> tried setting dma = off in hdparm, didnt work, still got crashes compiling
> firebird.. infact, it didnt even get passed the ./configure stage :(

Use hdparm to set, say, udma0.

> Each time (just about, didnt do it once out of 6ish times) i set the dma to
> off (either at boot or runtime) i got a "spurious 8259A interript: IRQ 7"..
> apparantly, iirc that debug message has been taken out of 2.6.6

I have 'em too. Not a real issue on lots of systems.

> Please guys, CC me your responces, otherwise i have to look at some
> archives, and reply via webmail... it took 2 firefox crashes (one of which
> brought down X) to just right this message... now im going to take a chance
> and hit the submit
> button
--
vda

Re: lotsa oops - 2.6.5 (preempt + unable handle virutal address + more?)

[Pods]

i'll get to your suggestions soon, infact the RAM was my next plan of
attack. But meanwhile, why dont you have a look at some new oops output from
a new kernel with lots of debug info turned on.

http://users.quickfox.org/~pods/linux-issues/kernel-debug-oops.txt


--------- Original Message --------
From: Denis Vlasenko <vda@port.imtp.ilyichevsk.odessa.ua>
To: Pods <pods@dodo.com.au>, linux-kernel@vger.kernel.org
<linux-kernel@vger.kernel.org>
Subject: Re: lotsa oops - 2.6.5 (preempt + unable handle virutal address +
more?)
Date: 06/06/04 05:28

>
> On Saturday 05 June 2004 18:27, Pods wrote:
> &gt; &gt;&gt; Run memtest86
> &gt; Tried once before, didnt boot... died :( memtest page says this
happens on
> &gt; some hardware and they're not sure why.
>
> Ho hum. Looks like good starting point to debug.
> Remove RAM until you have &lt;= 2 GB, then run memtest86
> (or modified versions of it with debug prints)
> until you know where it dies.
>
> &gt; &gt;&gt; Underclock your system
> &gt; its not over clocked :/
>
> I mean, underclock below normal conditions.
>
> &gt; &gt;&gt; lower IDE DMA mode and see whether it stops oopsing.
> &gt;
> &gt; tried setting dma = off in hdparm, didnt work, still got crashes
compiling
> &gt; firebird.. infact, it didnt even get passed the ./configure stage :(
>
> Use hdparm to set, say, udma0.
>
> &gt; Each time (just about, didnt do it once out of 6ish times) i set the
dma to
> &gt; off (either at boot or runtime) i got a &quot;spurious 8259A
interript: IRQ 7&quot;..
> &gt; apparantly, iirc that debug message has been taken out of 2.6.6
>
> I have 'em too. Not a real issue on lots of systems.
>
> &gt; Please guys, CC me your responces, otherwise i have to look at some
> &gt; archives, and reply via webmail... it took 2 firefox crashes (one of
which
> &gt; brought down X) to just right this message... now im going to take a
chance
> &gt; and hit the submit
> &gt; button
> --
> vda
>
>
>
>
>
>
>
>
>
>
> 

________________________________________________

Message sent using
Dodo Internet Webmail Server

Re: lotsa oops - 2.6.5 (preempt + unable handle virutal address + more?)

[Pods]

Ok i fixed it. Was a RAM installation problem. I've had trouble with RAM in
this board before. When i initialy set it up, the RAM modules being in 2&3 3
was the only i could get it to boot.

However i had a few more tries and i got a single DIMM bootting the machine
and then i tried two. It didnt boot, so i wirggled it and it work? very
strange. I think the DDR slots may not be up to scratch, but its working...
o rat least, seems to be.

Thanx for your help, even though it wouldnt have been neccessary if i had
taken more care in installing the RAM. *color me red faced*

Thanx again.


--------- Original Message --------
From: Denis Vlasenko <vda@port.imtp.ilyichevsk.odessa.ua>
To: Pods <pods@dodo.com.au>, linux-kernel@vger.kernel.org
<linux-kernel@vger.kernel.org>
Subject: Re: lotsa oops - 2.6.5 (preempt + unable handle virutal address +
more?)
Date: 06/06/04 05:28

>
> On Saturday 05 June 2004 18:27, Pods wrote:
> &gt; &gt;&gt; Run memtest86
> &gt; Tried once before, didnt boot... died :( memtest page says this
happens on
> &gt; some hardware and they're not sure why.
>
> Ho hum. Looks like good starting point to debug.
> Remove RAM until you have &lt;= 2 GB, then run memtest86
> (or modified versions of it with debug prints)
> until you know where it dies.
>
> &gt; &gt;&gt; Underclock your system
> &gt; its not over clocked :/
>
> I mean, underclock below normal conditions.
>
> &gt; &gt;&gt; lower IDE DMA mode and see whether it stops oopsing.
> &gt;
> &gt; tried setting dma = off in hdparm, didnt work, still got crashes
compiling
> &gt; firebird.. infact, it didnt even get passed the ./configure stage :(
>
> Use hdparm to set, say, udma0.
>
> &gt; Each time (just about, didnt do it once out of 6ish times) i set the
dma to
> &gt; off (either at boot or runtime) i got a &quot;spurious 8259A
interript: IRQ 7&quot;..
> &gt; apparantly, iirc that debug message has been taken out of 2.6.6
>
> I have 'em too. Not a real issue on lots of systems.
>
> &gt; Please guys, CC me your responces, otherwise i have to look at some
> &gt; archives, and reply via webmail... it took 2 firefox crashes (one of
which
> &gt; brought down X) to just right this message... now im going to take a
chance
> &gt; and hit the submit
> &gt; button
> --
> vda
>
>
>
>
>
>
>
>
>
>
>
> 

________________________________________________

Message sent using
Dodo Internet Webmail Server