xfs
[Top] [All Lists]

Re: oops with 2.4.5 + xfs + md raid0

To: Jim Wright <jwright@xxxxxxxxxxxxxxxxxxxx>
Subject: Re: oops with 2.4.5 + xfs + md raid0
From: Steve Lord <lord@xxxxxxx>
Date: Mon, 18 Jun 2001 15:47:02 -0500
Cc: linux-xfs@xxxxxxxxxxx
In-reply-to: Message from Jim Wright <jwright@penguincomputing.com> of "Mon, 18 Jun 2001 13:10:09 PDT." <Pine.LNX.4.33.0106181255430.31396-100000@foo.penguincomputing.com>
Sender: owner-linux-xfs@xxxxxxxxxxx
> System:
> 
>     Tyan Tiger LE S2515ING
>     dual 1ghz PIII
>     1gb memory
>     two IBM DTLA 307075 drives
>     each drive master on a channel of the onboard raid
>     onboard raid detected as Promise PDC20267
>     onboard raid set so each drive is a single span set
>     net effect is to use the onboard raid as a pair of simple ide controllers
> 
>     operating system on separate hard drive
>     2.4.5
>       + linux-2.4.5-xfs-06042001.patch.bz2
>       + ide.2.4.5.06062001.patch (Andre Hedrick IDE patches)
>       + linux-2.4.0-sard.patch (extended /proc/partitions info)
> 
> Drives have numerous partitions.  Each pair of partitions are configured
> to use software raid 0 with varying stripe size.  Each raid 0 set has
> an xfs filesystem on it, built with the default parameters.
> 
> I left Bonnie running over night, and this morning came in to find an
> oops.  The decoding follows.  Let me know if I can be of further help.
> I'm not on the list, so please CC me on any replies.
> 
> I found the new patch bundle from 11June2001 and will use those.  Are
> the patches on an every Monday schedule?  If so, I'll use today's instead
> when they come out.


Patches are not really regular, the cvs tree is the most frequently updated
source of xfs code. This is currently at 2.4.6-pre3. There are some changes
in the later patch, but nothing I would expect to affect this.

This oops is nasty in that it appears to represent a bad pointer in a
page hash so far as I can tell. Can you report the compiler version you
used and send a disassembly of the whole of the __find_lock_page (run gdb
on vmlinux and use the disassemble command).

Can you also send me a pointer to Andre's ide patch please.

Thanks

   Steve


> 
> Jim
> 


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