Re: [RFC] dealing with proc_ns_follow_link() and "namespace" dentries

[Al Viro <viro@ZenIV.linux.org.uk>]

On Sat, Nov 01, 2014 at 03:06:20PM +0000, Al Viro wrote:
> On Sat, Nov 01, 2014 at 08:38:04AM +0000, Al Viro wrote:
> > OK, interim branch (_completely_ untested, and there's quite a bit of
> > work remaining) is in vfs.git#nsfs.
> 
> ... except that what got pushed was completely buggered - the last changes
> not committed *and* include/linux/ns_common.h not git-add'ed at the very
> beginning.  Oh, well - it wasn't hard to reconstruct its history...
> 
> Anyway, that much got fixed and pushed; sorry about the noise - shouldn't
> have posted before grabbing some sleep...

And now it even seems to work.  Poking in /proc/*/ns/*, mount --bind of those
suckers to regular files, unshare and nsenter (both by PID and by file,
including the ones we'd bound somewhere).  Eric, could you hit it with
whatever testsuite you are using and see if it survives?

It's vfs.git#nsfs (head should be at 2e64120), branched at -rc2.  The meat
of that sucker is in the next-to-last commit; the rest is preparation and
a minor cleanup.