>  | it installs a syscall that ignores all its arguments,
>  | does nothing else, and then returns.  all so that i have one.  it,
>  | therefore, *CANNOT* fail to match any api in the kernel, because it
>  | doesn't use any.
>
>Hmm?   Isn't there an API in the kernel being used to install a new syscall?

um...yeah, though i would expect that if *that* changed, the version
of the lkm interface would refuse to load *any* module, not just those
that are three days too old...

that's fine.

>While I don't much believe in LKMs for available source systems, I do
>however understand what you're requesting I think - that is, a more finely
>grained way to tell just what API may have altered, so LKMs could
>determine whether they will still work, or need to be recompiled.  The
>current "nothing has changed" (which is often a lie anyway), or "anything
>might have changed" is a little crude.

maybe something like that, yes.

-- 
|-----< "CODE WARRIOR" >-----|
codewarrior@daemon.org             * "ah!  i see you have the internet
twofsonet@graffiti.com (Andrew Brown)                that goes *ping*!"
werdna@squooshy.com       * "information is power -- share the wealth."