Re: [RFC PATCH 5/8] KEYS: exec request-key within the requesting task's init namespace

[Benjamin Coddington <bcodding@redhat.com>]

On Tue, 24 Feb 2015, J. Bruce Fields wrote:

> On Mon, Feb 23, 2015 at 05:22:12PM -0800, Benjamin Coddington wrote:
> > That sounds a lot closer to some of the work I've been doing to see if I can
> > come up with a way to solve the "where's the namespace I need?" problem.
> >
> > I agree with Greg's very early comments that the easiest way to determine
> > which namespace context a process should use is to keep it as a copy of
> > the task -- and the place that copy should be done is fork().
>
> So you're suggesting that the key_agent could be that copy?  But:
>
> > ... If not, then the calling process itself is forked/execve-ed into a
> > new persistent key_agent that is installed on the calling process'
> > keyrings just like a key, and with the same lifetime and GC
> > expectations of a key.
> >
> > A key_agent is a user-space process...
>
> If the key_agent can die before it's needed, then we have to keep around
> some other context information to allow regenerating a new one.  So what
> is that piece of information?  Aren't we back where we started?

Yes.  It would seem to make sense then to keep a copy of task which would be
used to create the user-space key_agent.  If that's the standard behavior,
it doubles the number of tasks for the average use case.  The average use
case (I anticipate) would be to just always fork the caller if a key_agent
is not found.

It would probably make sense provide the key_agent_type the option of
binding a key_agent reference task to another structure -- like a mount.
And, if so, that reservied reference task would be used to create/re-create the
user-space key_agent.  The trouble there is how to have the caller
communicate the object bound to the task.. a reference to the object could
be guessed, which would give any caller access to a that reserved task..

In order to keep further verbose speculation to a minimum, I'll say: good
point.. there may be a way to authorize the usage of a reference task by
means of the possession of an authkey.  I'll keep hacking at this.

Ben