[RFC][PATCH 2/2] introduce proc_mnt for pid_ns

[Eric W. Biederman]

Dave Hansen <hansendc at us.ibm.com> writes:

> The following patch completes the removal of the global proc_mnt.
> It fetches the mnt on which to do dentry invalidations from the
> pid_namespace in which the task appears.
>
> For now, there is only one pid namespace in mainline so this is
> straightforward.  In the -lxc tree we'll have to do something
> more complex.
>
> Note that the new proc_compare_super() enforces the "one proc sb
> per pid_namespace" limit.
>
> /proc currently has some special code to make sure that the root
> directory gets set up correctly.  It proc_mnt variable in order
> to find its way to the root inode.
>
> Now that we don't have the global proc_mnt, we can fill in the
> root inode's data in proc_fill_super(), where it takes a wee bit
> less work than in proc_get_sb().

Nothing jumps out at me as an obvious problem.

Eric