Re: Path to kmods

[Robert Elz <kre%munnari.OZ.AU@localhost>]

    Date:        Thu, 13 Nov 2008 15:09:05 +0000
    From:        Andrew Doran <ad%NetBSD.org@localhost>
    Message-ID:  <20081113150905.GD8195%hairylemon.org@localhost>

  | Rather than waste any more time arguing
  | about whether or not we should take the path of least resistance, I will
  | resolve any differences that matter.

The case that interests me most would be making NFS a module (which along
with the other *fs's is, I would assume, a reasonable candidate), having a
"generic" compile of the NFS module (which would include defining INET6,
and INET) and then loading it into a kernel where one (or perhaps even both)
of INET6 and INET has been disabled at compile time.

If you can make that work without adding considerable overhead to the
module (in the cases where INET and INET6 are in fact present in the kernel)
then go for it.  Otherwise, I really don't see a way to have kernel compile
independent modules.

kre

ps: bias alert - I am one of those who doesn't much like dynamically linked
kernels (whether they're called LKM's or MODULES or anything else).   I also
much favour end users doing kernel source compiles (not mandatory but highly
recommended).   I do believe that the current config(8) needs a hell of a
lot of work to make that objective rational however, at the minute it involves
way too much black magic.