> > Hmmm. On my system, network protocols with no active sockets and unused
> > filesystems are already "swapped out" - or, at least, they do not eat my
> > memory as I compiled them as modules...
>
> Modules are arguably a better way of doing this, but for some stuff, they
> may be kernel data structures which could be swapped out.
Ok, I pay that - how much kernel data space is actually used by things
like inactive network protocols, however - and could this possibly be
solved by pushing that into the modules init code to allocate the kernel
memory, or extend that data structure?
> I expect 2.3.x will be considerably more modular, especially the networking
> code.
Well, this is probably a good thing(tm), even if I can't see where more
modularity is needed right now. Someone will probably use it, I guess
:)
[math-emu as module]
> I'm not sure how you would handle FPU code if the module couldn't be loaded.
> I guess you could alwas kill the process with a SIGILL or something, doesn't
> seem like a very nice solution.
Ooops. You caught me on an assumption - I guess that SIGILL would be an
ugly but resonable response if the module were not present, and could
not be dynamically loaded for whatever reason; I forget that some sites
may not use the kmod thread to bring in needed modules - I don't think
(but really don't know) that doing the same for math emulation would be
impossible...
Daniel
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.altern.org/andrebalsa/doc/lkml-faq.html