Re: [PATCH] 3 of 5 IMA: LSM-based measurement code

From: Chris Wright (chrisw@private)
Date: Wed Jun 15 2005 - 15:49:21 PDT


* Serge E. Hallyn (serue@private) wrote:
> Quoting Chris Wright (chrisw@private):
> > * serue@private (serue@private) wrote:
> > > Quoting Chris Wright (chrisw@private):
> > > > The primary purpose of the hooks is access control.  Some of them, of
> > > > course, are helpers to keep labels coherent.  IIRC, James objected
> > > > because the measurement data was simply collected from these hooks.
> > > 
> > > Ok, so to be clear, any module which does not directly impose some form
> > > of access control is not eligible for an LSM?
> > 
> > That's exactly the intention, yes.
> 
> Ok, thanks.
> 
> I thought it was intended to be more general than that - in fact I
> specifically thought it was not intended to be purely for single machine
> authentication decisions within a single kernel module, but that anything
> which would aid in enabling new security features, locally or remotely,
> would be game.  (Which - it means nothing - but I would clearly have
> preferred :)

The problem with being more general is it becomes a more attractive
target for abuse.

thanks,
-chris



This archive was generated by hypermail 2.1.3 : Wed Jun 15 2005 - 15:53:16 PDT