Subject: Re: Status of LKM support
To: Stephen J. Roznowski <sjr@zombie.ncsc.mil>
From: Chris G Demetriou <Chris_G_Demetriou@LAGAVULIN.PDL.CS.CMU.EDU>
List: tech-kern
Date: 12/24/1994 08:31:55
> It appears that the LKM samples no longer work, mostly due to
> nosys wanting 3 arguments now (shows up in DISPATCH from lkm.h).

yup.

> Should "nosys" be replaced with "lkmnosys", or should the "load",
> "unload", "stat" functions be modified to have 3 arguments?

neither.  use of nosys here is a 'fun' example of bad programming
style that's allowed when things like prototypes (or lint 8-) aren't
around.


i replaced the uses of nosys with use of a new function, 'lkm_nofunc',
and simplified the dispatch logic a bit.  it would also have been
reasonable to use NULL, but i picked the former, because it seemed a
bit more straightforward.


current LKM binaries will run just fine, but source may have to be
changed slightly (if they used nosys for load/unload/stat functions).
of course, if they did, they wouldn't compile anyway...  8-)


chris