On Wed, Mar 28, 2001 at 03:41:11PM -0800, Charles M. Hannum wrote:
> No.  Known problem.  Branch instructions have limited range, and the
> LKM is loaded `too far' from the main kernel.  This is the same problem
> that causes arm32 LKMs to fail as well (except on that platform the
> linker doesn't actually barf; it just DTWT and the kernel panics).
> 
> Nobody has stepped up to the plate to deal with this yet.

Not true.

I'm dealing with it now. (The reason I pulled a machine up to
-current and, in the process, made a snapshot of -current as of last
week.)

If people'd rather have LKMs than -current X packages, I won't
bother with the latter till I've fixed the former.

The bug is with gcc, btw, not really with NetBSD proper.

(Or, if you fix it the right-but-really-complicated way, with ld.
Still not with the OS proper, since we use GNU's ld too.)

       ~ g r @ eclipsed.net