In linux-2.1.116-1 a new system call sys_umount is introduced
and the old one is rechristened sys_oldumount.
Introducing the call was a very good idea, but I am unhappy
about the name change.
Changing the name will cause all kinds of minor problems
in the transition period, where the meaning of umount
depends on kernel and libc, and nobody knows how many
arguments this system call has.
Moreover, the name change does not serve any useful purpose.
No present call of umount() can become a call of the new function
because it has the wrong number of arguments.
If one argues that the old umount will eventually become obsolete:
I agree. But just like seek became obsolete and only lseek
and possibly llseek are left, so umount can disappear
some years from now and leave umount2.
I think Solaris and Irix have a 1-argument umount().
Therefore, I would like to suggest to call the new system call
umount2 (for `umount with 2 arguments').
Furthermore, I would like to suggest that the availability of
MNT_FORCE in <sys/mount.h> be an indication for the availability
of umount2() in glibc.
[Possibly this string should also be changed? Should it not
start with U?]
Andries
P.S. I made mount-2.8b-alpha, where the -alpha means that it
compiles but is entirely untested. It mounts ext2 partitions
by label or by UUID, and knows about umount -f.
Not for production use! Details may change when linux-2.1.116
comes out.
Will upload mount-2.8b-alpha.tar.gz later tonight to the usual place
(ftp.win.tue.nl:/pub/linux/util/mount).
-
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