Re: kmem_alloc: Bad slab magic (corrupt) (name=size-128)

Craig B Agricola (agricolc@btv.ibm.com)
Tue, 3 Aug 1999 16:56:19 -0400

Hi Peter,
This looks suspiciously like a problem that was plaguing the Token Ring
code around kernels 2.2.3-2.2.8 or so. How long does it take before you start
getting this message after you initialize your eth0 interface? In the Token
Ring case, the problem was traced down to lots of net packets that were lying
about thier length, so when they were being free'd, they were freeing up less
space than they were actually taking, so before long there was no more real
memory available to the kernel. You may want to take a look with a packet
sniffer to see if there are any bad packets flying around (maybe some bad
NetBUI junk or something...)

-Craig

On Tue, Aug 03, 1999 at 10:23:37PM +0200, Peter Gervai wrote:
> Hello,
>
> Aug 3 21:04:46 my kernel: kmem_alloc: Bad slab magic (corrupt) (name=size-128)
> Aug 3 21:04:46 my kernel: eth0: Couldn't allocate a sk_buff of size 60.
> Aug 3 21:04:47 my kernel: kmem_alloc: Bad slab magic (corrupt) (name=size-128)
> Aug 3 21:04:47 my kernel: eth0: Couldn't allocate a sk_buff of size 67.
> Aug 3 21:04:47 my kernel: kmem_alloc: Bad slab magic (corrupt) (name=size-128)
> (last messages repeated zillion times till reboot)
>
> As far as I am able to see this is some kind of very serious memory
> corruption, severity level equals to panic. (Results the same, anyway.)
> (The last messages before this agony were three proftpd sessions closing,
> with probably lots of net activities in the back as well.)
>
> Any idea why this could have happened? The machine is 256M RAM, some jumbo
> ide disks, and quite an amount of processes (sometimes hits the 512 limit
> in the default kernel). As far as I know there should not be any hardware
> (RAM) problems, at least there wasn't any sign for quite a long time.
>
> I'd appreciate at least a "we don't have a clue" or "sorry, too little
> input" answer, to close this problem report of mine. :-) (A cc to my
> email would be much appreciated as well.)
>
> thanks,
> grin
>
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.rutgers.edu
> Please read the FAQ at http://www.tux.org/lkml/

-- 
 Craig B. Agricola    W: (802) 769-4636
  Work Related Email: agricolc@btv.ibm.com
  Personal Email:     cba@po.cwru.edu


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/