On Sun Mar 07 2010 at 11:18:57 +0000, David Laight wrote:
> On Fri, Mar 05, 2010 at 06:35:02PM +0000, Antti Kantee wrote:
> >
> > Log Message:
> > Move builtin modules to a list in init and load them from there
> > instead of using linksets directly. This has two implications:
>
> 3) It must now be almost possible to do an 'ld -r' of the kernel
> (excluding any modules) to generate a 'netbsd.o' and then do a
> final link to include a desired set of modules into a kernel image.
I doubt it really has an effect on that. As I mentioned earlier,
I've been able to link and use rump kernels pretty much that way for
over a year now (i.e. cc -o foo ${*.o} foo.kmod bar.kmod -lrumpxyzzy).
In fact, it was supported way before I added support for dynamically
loading and linking modules via rump_sys_modctl().
What, specifically, is the problem with the netbsd.o approach?
(changing the default link that way would be good step, especially for
people wanting an easy-to-grow-at-home-from-sets MONOLITHIC)
