Apparently, though unproven, at 16:59 on Monday 16 May 2011, felix@crowfix.com
did opine thusly:

> On Mon, May 16, 2011 at 04:49:07PM +0200, Alan McKinnon wrote:
> > The correct way to use module-rebuild is to run once:
> >
> > module-rebuild populate
> >
> > This will search the tree to find out-of-kernel-tree module ebuilds you
> > are using and put them in a db or later use.
> >
> > Every time you emerge and build a new kernel, run:
> >
> > module-rebuild rebuild
> >
> > This will build the missing modules for the kernel you just built.
> >
> > module-rebuild add|del lets you maintain the list as you add and delete
> > stuff
>
> If populate inits the list, are add/del only there to avoid a length
> tree search? Otherwise I take it you mean run populate once, then
> rebuild after every new kernel, and otherwise do nothing?

Correct. populate is the kind of thing you run once at the beginning and never
again. add|del is run whenever you need them and rebuild after every new
kernel merge (even -r versions)


--
alan dot mckinnon at gmail dot com

felix@crowfix.com writes:

> At any rate, it seems kind of odd. What is the proper way of using
> module_rebuild? It seems to me there are two cases, and maybe that
> is why this script has this odd code. If you have just built a brand
> new kernel, you might want to rebuild the module list from scratch.
> But once you have done that, future emerges only need to keep the
> module list up to date.

Whatever. With a recent portage you just emerge @module-rebuild, that's what
I use. There's also @x11-module-rebuild, for recompiling Xorg stuff when
xorg-server is updated.

Wonko