der Mouse wrote:
> > The problem here is that a LKM is unloaded, it doesn't have a chance
> > to free it's [sic] resources.
> 
> > Solutions would be to add a new VFS routine, [...]
> 
> What about LKMs that aren't filesystems?  Won't we have exactly the
> same problem with them as soon as one of them has something global that
> needs freeing on unload?
> 
> IOW, shouldn't this be "a new LKM routine" rather than "a new VFS
> routine"?

I think about my proposed way as more consistent. We have
VFS "init" routine (called when the filesystem is attached,
either after loading it in as LKM, or on boot), so we should
have a "done" routine, which would clean it up so that it
can be detached.

Doing the cleanup in lkm dispatch code seems to me like
abstraction violation.

Jaromir
-- 
Jaromir Dolecek <jdolecek@NetBSD.org>      http://www.ics.muni.cz/~dolecek/
@@@@  Wanna a real operating system ? Go and get NetBSD, damn!  @@@@