Re: kernel BUG at mm/slab.c:1488! (2.6.13.2)

[Nate Diller <nate.diller@gmail.com>]

On 10/27/05, Ben Greear <greearb@candelatech.com> wrote:
> I was compiling with ext3 as a module, then changed to compiling static
> and un-tarred the new kernel over the old (the old ext3 module was still
> in existence.)  I re-ran the mkinitrd and rebooted.
>
> It seems that something still tries to load the ext3 module, and I get the
> BUG seen below.  If I remove the ext3 module and re-build the initrd,
> the error goes away.
>
> I was thinking that at the least, the ext3 module code should somehow
> detect
> that it is already built-in and exit it's load attempt very early (before
> triggering whatever bug it hit).
>

There was a thread about this some months ago
(http://marc.theaimsgroup.com/?l=linux-kernel&m=111639356609390&w=2),
and two patches were proposed, to either return NULL here or WARN_ON,
and keep going.  Neither seems to have been picked up.  Maybe people
think this is not a big problem?  I would certainly like to see a way
to prevent modules from being loaded multiple times, or when they are
also compiled-in, but aside from returning NULL here in
kmem_cache_create, I don't see a trivial way of doing it.

NATE