[KLUG Advocacy] Re: [KLUG Members] Distro distro ...

[Peter Buxton]

On Thu, Dec 04, 2003 at 08:17:26PM -0500, Robert G. Brown was only escaped
   alone to tell thee:

> On Thu, 04 Dec 2003 15:35:59 -0500, Bruce Smith <bruce@armintl.com> wrote:

> > IMO this is better than having stop scripts (w/o rmmod) for shutdown
> > and another set of stop scripts for every other situation (w/rmmod).
> 
> Actually, I was clarifying the question...and it might not stop some
> race conditions or panics in any case.

In Rusty Russell's interview on kerneltrap.org
( http://kerneltrap.org/node/view/892 ) he notes that rmmod and insmod
are for hackers: modprobe (-r) is for sysadmins and scripts. He was
speaking of the 2.6.x verisons he wrote, but it probably applies to 
<2.5 as well

Still, I think it silly. Yes, you free a bit of memory, but why unload
drivers just because you don't happen to be using that piece of hardware
at the moment?

   Our kernel isn't pageable (because that would be "a stupid idea"),
   but we have this ability to remove live parts of our kernel, which is
   a poor-man's version of the same thing! To do this properly, you need
   to know when you can discard modules, which means you need reference
   counts (lots of reference counts), and keeping reference counts isn't
   free.
           -- RR, above.

-- 
We are the 801. We are the central shaft. -- Brian Eno