Re: pre-patch-2.1.107 breaks kmod

Paul Gortmaker (linux@rasty.ph.unimelb.edu.au)
Sat, 27 Jun 1998 03:07:21 +1000 (EST)


> Note that for the record, I don't think calling request_module
> from init_module is a good idea. I know that the one network
> driver is doing it, but I have continually failed to see how
> what he is doing is better than having depmod and modprobe do
> the dependancies and loading.

In principle, I agree with you that depmod handling it is better.
The only justification as "better" that I can offer (FWIW) is the
fact that it is more luser-proof. i.e. I see a lot of posts and
e-mail from people like "I tried to load the ne driver but I got
ei_interrupt and some other things undefined..."

Assuming distributions ship with kmod enabled (they should, since
they did with kerneld) then a novice user struggling with initial
configuration will no longer get a meaningless message like this.

If kmod support is not enabled, then "normal" depmod behaviour
is still used (with zero extra lines of compiled code). Furthermore,
kmod-aware 8390 based drivers can be used on a non-kmod kernel and
vice-versa. And finally, as a fall-back, the whole thing can be
disabled (ie normal depmod usage in all cases) just by nuking the
one line #define LOAD_8390_BY_KMOD

So some thought did go into it and it wasn't totally a random hack :)

Paul.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu