Comment 14 for bug 11370

Message-Id: <email address hidden>
Date: Mon, 13 Dec 2004 11:07:25 -0500
From: Andres Salomon <email address hidden>
To: Horms <email address hidden>, <email address hidden>
Subject: Re: Bug#284356: New release changed symbols thus rendering modules unloadable

On Mon, 13 Dec 2004 21:46:56 +0900, Horms wrote:

[...]
>=20
> Hi,
>=20
> I think that I have discovered the cause of the problem.
>=20
> It seems to be caused by the 093_tty_lockup.diff patch which was applie=
d
> to resolve CAN-2004-081, a security bug relating to race
> conditions in the TTY subsystem. The patch was sourced from
> Jason Baron from Red Hat. I have attached it for reference.
>=20
> In testing I found that the problem lies in the following change
> in include/linux/tty.h
>=20
> -extern struct tty_ldisc ldiscs[];
> +extern struct tty_ldisc tty_ldiscs[];
>=20
> Backing out this change then runing something along the lines of
> make mrproper
> cp some_path/some_config .config
> make oldconfig
> make dep
>=20
> yields the same symbols.
>=20
> (This test was done using the 686-smp config from=20
> kernel-image-i386-2.4.27 -6 and examining the=20
> proc_mkdir symbol which is found in include/linux/modules/root.ver)
>=20
> Of course this does not compile because my change is bogus,
> but I think it does establish the cause of the symbol problem.
>=20
>=20
> I believe that there are serveral different fixes for this problem, and
> in particular the one that was incoporated in 2.6 is somewhat different=
.
> I am not sure if they cause the problem.

As far as 2.6 is concerned, I can successfully compile images against
kernel-source 2.6.8-8 that work with the official kernel-image modules
that are built against 2.6.8-10 (that is, kernel-image 2.6.8-10).=20
However, using the official kernel-image package built against 2.6.8-8
(that is, kernel-image 2.6.8-5) does not work w/ modules built against
2.6.8-10. This leads me to believe that the 2.6 problems are toolchain
related, some change made to sid between Nov 12 and Nov 22, or a change i=
n
the build environment used to create -9 and -10 (joshk, mind doing
some testing?). There was a new module-init-tools package uploaded on No=
v
20; I'll start looking there.