xfs
[Top] [All Lists]

Re: Kernel Oops .....

To: Simon Matter <simon.matter@xxxxxxxxxxxxxxxx>
Subject: Re: Kernel Oops .....
From: "Michael Wahlbrink" <miw@xxxxxxxxxxxxxxxx>
Date: Tue, 4 Sep 2001 10:18:09 +0200
Cc: Keith Owens <kaos@xxxxxxxxxxxxxxxxx>, Seth Mos <knuffie@xxxxxxxxx>, linux-xfs@xxxxxxxxxxx, Michael Wahlbrink <miw@xxxxxxxxxxxxxxxx>, owner-linux-xfs@xxxxxxxxxxx
Sender: owner-linux-xfs@xxxxxxxxxxx
On 04.09.2001 09:02:20 Simon Matter wrote:
>Michael Wahlbrink schrieb:
>>
>> On 02.09.2001 00:07:42 "Michael Wahlbrink" wrote:
>> >Ok, no problem ;-)
>> >
>> >I'll make my tests with ext2 to see if the kernel 'oopses' 
>too.........
>> >And the PII is still compiling th xfs kernel....
>> >
>> >cu
>> >michael
>> >
>> Ok, I think I've found the error! :-))
>> I've made the tests also with ext2 and run into IO errors...... I 
>checked
>> the drives and burned my fingers...... The drives are getting to hot 
>:-(
>> So I installed a additional cooler and all was nearly fine...
>> Because of all the compilation tries and tests my system was 
>unusuable and
>> i get more and more strange errors (also from xfs)...... So i 
>installed a
>> new clean system (2.4.10-pre4-xfs), and now its working.......!
>> On the PII its also working fine.....
>>
>> But there is one thing, why does ext2 says that it have IO Problems 
>when
>> writing to disk and xfs simply dies with an oops when you try to 
>acess the
>> corrupt data ????
>
>Just imagine you have bad memory, it is flipping bits sometimes. Now can
>you ask why does it crash when using this or another filesystem? No, I
>mean the filesystems just expect properly working hardware. If there is
>a problem with corruption, you can not expect reproducable effects. My
>own experience with corruption was that most time xfs survived without
>oopsing. But anyway such a system is unusable. 1 must be 1, and 0 must
>be 0.

I think it was the temperature, cause the SDRAM was checked one night long 
with memtest86 without errors, and also for a test replaced with other 
DIMMs which worked fine in another Mashine and are also correct........

cu
micha

Attachment: smime.p7s
Description: S/MIME Cryptographic Signature

<Prev in Thread] Current Thread [Next in Thread>