Re: 2.2.20 - Possible module symbol bug

Keith Owens (kaos@ocs.com.au)
Sun, 11 Nov 2001 17:03:38 +1100


On Sat, 10 Nov 2001 20:38:58 -0800,
David Ranch <dranch@trinnet.net> wrote:
>I did start off with a clean config though I didn't do
>a "mrproper". I re-tested on the MD7.0 box and
>"mrproper" solved my issue. I haven't had to do this
>step since the old 2.0.2x days and I feel somewhat
>embarrassed by this.
>
>Thanks again for setting me strait though. I am
>curious for an explanation from the powers at be to
>why this hit me now though I've been compiling kernels
>without "mrproper" for years. Am I just lucky? ;-)

The implementation of module symbol versions is fundamentally broken.
The idea is good, to prevent incompatible modules being loaded into
your kernel and crashing it, but the implementation is built on several
assumptions that are not valid. Total redesign coming up in kernel
2.5. See the build 2.5 history package[1], expecially the html files.

Sometimes modversions work, more often then don't.

[1] http://sourceforge.net/project/showfiles.php?group_id=18813, under
History.

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