Quoting Miklos Szeredi (miklos@xxxxxxxxxx): > On Wed, 27 Aug 2008, Serge E. Hallyn wrote: > > Ok, thanks. I look forward to playing around with it when you publish > > the resulting git tree :) > > A couple of centuries later... > > ...here's the updated git tree: > > git://git.kernel.org/pub/scm/linux/kernel/git/mszeredi/vfs.git unprivileged-mounts > > Changes since the previous version: > > - update to apply against latest git > - downgrade shared mounts to slave for unprivileged binds (if owners differ) > - don't allow unprivileged recursive binds > > Serge, thanks again for testing and reviewing these patches! Well I see where a shared mount *should* be turned into a slave mount when bind-mounted as a user mount, but it doesn't seem to be happening. In particular, after doing a user mount of /mnt onto /home/hallyn/etc/mnt, /proc/self/mountinfo ends in: 22 13 3:1 /mnt /mnt rw shared:1 - ext3 /dev/root rw,errors=continue,user_xattr,acl,data=ordered 23 13 3:1 /mnt /root/mnt rw shared:1 - ext3 /dev/root rw,errors=continue,user_xattr,acl,data=ordered 24 13 3:1 /mnt /home/hallyn/etc/mnt rw,user=500 shared:1 - ext3 /dev/root rw,errors=continue,user_xattr,acl,data=ordered I assume this means that /mnt and /home/hallyn/etc/mnt are peers in peergroup 1? And indeed if hallyn does mount --bind /usr /home/hallyn/etc/mnt/usr, then /mnt/usr shows the contents of /usr. I see that in do_loopback() you are adding CL_SLAVE|CL_MAKE_SHARED to flags so I don't get what is going on. Still looking through the code. -serge -- To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
