> Am Dienstag, 2. Juli 2002 19:48 schrieb Tom Rini:
> > On Tue, Jul 02, 2002 at 06:07:06PM +0200, Oliver Neukum wrote:
> > > > developing drivers and such. Aunt Tillie would no longer be able to
> > > > remove modules from her kernel, but that's not likely to bother her
> > > > too much...
> > >
> > > It would very much bother uncle John, who is in high availability.
> >
> > Then the HA kernel turns on the ability to still remove modules, along
> > with all of the other things needed in an HA environment but not
> > elsewhere. Provided removing a module doesn't become a horribly racy,
> > barely usable bit of functionality, which I hope it won't.
>
> Either there is a race or there isn't. Such a thing is unusable on a
> production system.
In a single processor, no preempt kernel, there is no race.
Turn on SMP or preempt and there is one.
Anyway, on a HA machine, how the heck are you going to be removing
modules anyway? If one of your two identical network cards fails, you
lose your HA status if you need to shut down both to remove the module
and thus be able to remove the second card, right?
I would think the HA guys would be pushing for a status of "modules are
never removed, so design around that" to make their lives easier. That
would also mean you would have a way to say, "hey driver, rescan for
devices - I think you'll find a new one that you should manage".
-- Ryan Anderson sometimes Pug Majere- 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/