kmem_cache_alloc_one_tail BUG

I am porting my Linux 2.4 code from an evaluation board with an MPC8275
processor and 8 Mbytes of flash, consisting of four Sharp
LH28F016SCT-L95 flash chips, to a custom board with an MPC8270
processor and 8 Mbytes of flash, consisting of two Intel 28F320C3 flash
chips.  The custom board is crashing when mounting the JFFS2
filesystem.  Specifically, with debugging on, in
kmem_cache_alloc_one_tail, I reach the following BUG:

#if DEBUG
if (cachep->flags & SLAB_POISON)
if (kmem_check_poison_obj(cachep, objp))
BUG();

Anyone seen this before and solved this problem?

Thank you.

On Fri, 20 May 2005 14:48:36 -0700, Bill wrote:

> #if DEBUG
> if (cachep->flags & SLAB_POISON)
> if (kmem_check_poison_obj(cachep, objp))
> BUG();
> 
> Anyone seen this before and solved this problem?

If you look in "mm/slab.c" you'll find there are three ways to die of
a poisoned slab cache:

POISON_INUSE	- Using an uninitialized slab 
POISON_FREE	- Use after free
POISON_END	- Store-past-end-byte 

The OOPS traceback in "/var/log/messages" should help determine how 
you got here.

Cheers