Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Mon, 4 Jun 2018 23:09:20 -0500
From:   "Serge E. Hallyn" <serge@hallyn.com>
To:     Kees Cook <keescook@chromium.org>
Cc:     Mimi Zohar <zohar@linux.vnet.ibm.com>,
        Casey Schaufler <casey@schaufler-ca.com>,
        James Morris <james.l.morris@oracle.com>,
        Paul Moore <paul@paul-moore.com>,
        "Serge E. Hallyn" <serge@hallyn.com>,
        linux-integrity <linux-integrity@vger.kernel.org>,
        linux-security-module <linux-security-module@vger.kernel.org>,
        LKML <linux-kernel@vger.kernel.org>,
        David Howells <dhowells@redhat.com>,
        "Luis R . Rodriguez" <mcgrof@kernel.org>,
        Eric Biederman <ebiederm@xmission.com>,
        Kexec Mailing List <kexec@lists.infradead.org>,
        Andres Rodriguez <andresx7@gmail.com>,
        Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
        Ard Biesheuvel <ard.biesheuvel@linaro.org>,
        Jessica Yu <jeyu@kernel.org>
Subject: Re: [PATCH v4 0/8] kexec/firmware: support system wide policy
 requiring signatures
Message-ID: <20180605040920.GA19747@mail.hallyn.com>
References: <1527616920-5415-1-git-send-email-zohar@linux.vnet.ibm.com>
 <1528121025.3237.116.camel@linux.vnet.ibm.com>
 <CAGXu5jJGrPtOUa8f4pMiZTiYBEP_eVJ1oEUqum8YBr9JbfuLmg@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <CAGXu5jJGrPtOUa8f4pMiZTiYBEP_eVJ1oEUqum8YBr9JbfuLmg@mail.gmail.com>
User-Agent: Mutt/1.5.21 (2010-09-15)
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

Quoting Kees Cook (keescook@chromium.org):
> On Mon, Jun 4, 2018 at 7:03 AM, Mimi Zohar <zohar@linux.vnet.ibm.com> wrote:
> > On Tue, 2018-05-29 at 14:01 -0400, Mimi Zohar wrote:
> >> Instead of adding the security_kernel_read_file LSM hook - or defining a
> >> wrapper for security_kernel_read_file LSM hook and adding it, or
> >> renaming the existing hook to security_kernel_read_data() and adding it
> >> - in places where the kernel isn't reading a file, this version of the
> >> patch set defines a new LSM hook named security_kernel_load_data().
> >>
> >> The new LSM hook does not replace the existing security_kernel_read_file
> >> LSM hook, which is still needed, but defines a new LSM hook allowing
> >> LSMs and IMA-appraisal the opportunity to fail loading userspace
> >> provided file/data.
> >>
> >> The only difference between the two LSM hooks is the LSM hook name and a
> >> file descriptor.  Whether this is cause enough for requiring a new LSM
> >> hook, is left to the security community.
> >
> > Paul does not have a preference as to adding a new LSM hook or calling
> > the existing hook.  Either way is fine, as long as both the new and
> > existing hooks call the existing function.
> >
> > Casey didn't like the idea of a wrapper.
> > James suggested renaming the LSM hook.
> >
> > The maintainers for the callers of the LSM hook prefer a meaningful
> > LSM hook name.  The "null" argument is not as much of a concern.  Only
> > Eric seems to be asking for a separate, new LSM hook, without the
> > "null" argument.
> >
> > Unless someone really objects, to accommodate Eric we'll define a new
> > LSM hook named security_kernel_load_data.  Eric, are you planning on
> > Ack'ing patches 1 & 2?
> 
> I'm sorry I'm late to review this series. Reading through what you
> have, it seems like the existing hook is fine. If the name has
> slipped, we can rename it, but I think adding another hook for the
> same logical action (loading something into the kernel) is confusing.

Personally I agree with Eric and prefer a new hook.  I don't feel strongly
enough about it to keep bikeshedding, but since this set already exists,
it seems like the way to go.

> It seems that only patches needed are 2 & 4 (new hook callsites), 5, 6
> & 7 (IMA coverage and policy). 1 and 8 seem needless to me. If the
> objection is that isn't use on non-file objects, sure, rename it. But
> I don't see a _logical_ difference between the proposed and existing
> callsites. enum kernel_read_file_id covers the "type" already....
> 
> -Kees
> 
> -- 
> Kees Cook
> Pixel Security