On Sat, 1 Jul 2000, Horst von Brand wrote:
> Jamie Lokier <lk@tantalophile.demon.co.uk> said:
> > Werner Almesberger wrote:
> > > Your original proposal, moving to nullfs, gives you simply the possibility
> > > to have a "hard" or a "soft" failure. The basic assumption is still that
> > > the program didn't really need to access the file in the first place.
>
> > > So I'm still wondering if a clean exit/kill wouldn't be better in almost
> > > all cases.
>
> > revoke() without killing the process would be quite useful for floppies
> > removed from the drive...
>
> Which presuposes applications capable of handling said event gracefully.
> I.e., zilch.
>
no, incorrect. It doesn't presuppose anything _at all_!
The point of revoking or forcibly umounting is _not_ to satisfy some nasty
applications (if it were we wouldn't be SIGKILLing them first, see fuser
-k manpage to see what it does (yes, there is an non-portable -signal
parameter but probably nobody uses it)). The point is to be able to free
the underlying resources in a consistent manner, consistent
filesystem-wise and obviously not necessarily application-wise. The kernel
can never guarantee application consistency and shouldn't even try
to.
Therefore, both BSD-style revoke(2) and Solaris8 style generic forced
umount will be generally useful and require no assumptions on applications
side (for a simple reason that "we can't and therefore don't care" ;)
Regards,
Tigran
-
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.tux.org/lkml/
This archive was generated by hypermail 2b29 : Fri Jul 07 2000 - 21:00:11 EST