Re: corrupt modversions.h built in 2.2.13

Olaf Titz (olaf@bigred.inka.de)
Sat, 23 Oct 1999 17:03:25 +0300


> > CONFIG_MODVERSIONS as currently implemented is ugly as hell but it is
> > an _absolutely necessary_ feature at least if you have externally
> > compiled modules.
>
> Wrong. As you can see it's introducing more problems that it actually
> solves.

It solves at least 9.5 out of 10 "opendev: alloc: <some error>" or
"CIPE module crashes kernel" problems on the CIPE mailing list. In
fact I've never got one of those reports which was verified _not_ due
to version mismatch.

As soon as people start packaging distributions with externally
compiled modules, some day someone will configure their own kernel and
forget about recompiling {CIPE, ALSA, lm_sensors}, or they will
compile the modules with the wrong (distro default) headers. The
chance of the old module crashing the new kernel approaches 1 very
closely when a module accesses internal config-dependent structures.

MODVERSIONS guards against these problems. I don't like how it is
implemented but it is better than risking crashes.

> Wrong the read the doc's for ftape. The statement there says:
>
> "Don't bother me with versioned modules please".

There is a bit more context in the docs, which basically boils down to
"...if you don't know what you're doing" and it tells me that the most
problems relating to MODVERSIONS come from wrong compiler options
(iow., again a mismatch between kernel and module configuration).
FWIW, ftape works perfectly for me and I've not used non-MODVERSIONS
kernels for a long time now.

Olaf

-
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/