Re: [PATCH] cifs: handle termination of cifs oplockd kernel thread

[Bodo Eggert <harvested.in.lkml@xxxxxxxxxxxxxxxxxxxxxxxxxx>]

Steve French <smfrench@xxxxxxxxxxxxx> wrote:

> As we try to gradually obsolete smbfs, this came up with various users
> (there was even a bugzilla bug opened for adding it) who said that they
> need the ability to unmount their own mounts for network filesystems
> without using /etc/fstab.    Unfortunately for network filesytsems,
> unlike local filesystems, it is impractical to put every possible mount
> target in /etc/fstab since servers get renamed and the universe of
> possible cifs mount targets for a user is large.
> 
> There seemed only three alternatives -
> 1) mimic the smbfs ioctl -   as can be seen from smbfs and smbumount
> source this has portability problems because apparently there is no
> guarantee that uid_t is the same size in kernel and in userspace - smbfs
> actually has two ioctls for different sizes of uid field - this seemed
> like a bad idea
> 2) export the uid in /proc/mounts - same problem as above
> 3) call into the kernel to see if current matches the uid of the mounter
> - this has no 16/32/64 bit uid portability issues since the check is
> made in kernel

4) Turn umounting own mounts into a non-privileged operation.

As (AFAI see) the only thing that needs suid privileges is the umount
operation, and it is granted if the user mounted it himself, you can as
well do this simple check in the kernel.
-- 
Funny quotes:
40. Isn't making a smoking section in a restaurant like making a peeing
    section in a swimming pool?

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/