Re: CVE request: python-pyrad insecurities

[Vincent Danen <vdanen () redhat com>]

* [2013-02-15 23:36:21 -0700] Kurt Seifried wrote:

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> On 02/15/2013 04:53 PM, Vincent Danen wrote:
> > * [2013-02-15 19:51:07 +0000] Christey, Steven M. wrote:
> >
> > > These two issues were fixed in the same diff and reflect poor
> > > randomness - should we have only assigned one CVE?  (If the
> > > RADIUS feature was introduced in different versions than the
> > > authenticator-password feature, then maybe the SPLIT is
> > > acceptable.)
> >
> > I'm not sure.  I didn't go digging to see when they were introduced
> > -- both features may have been introduced at the same time (or
> > not).
> >
> > Ok, so doing a quick peek at the first full blob of it in git:
> >
> > https://github.com/wichert/pyrad/blob/c206b1dfc362db8b0ef9c256814377bde8ed91cf/pyrad/packet.py
> >
> >
> >
> > The use of random.randrange() is in both the CreateAuthenticator()
> > and CreateID() functions, so I would bet that they've been like
> > that the whole time (that blob is from Sept 2007).  So I guess one
> > CVE is probably sufficient.
> >
> > I only noted them as two issues as we had two separate bug reports
> > about them.
> >
> > > -----Original Message----- From: Kurt Seifried
> > > [mailto:kseifried () redhat com] Sent: Friday, February 15, 2013
> > > 2:37 PM To: oss-security () lists openwall com Cc: Vincent Danen
> > > Subject: Re: [oss-security] CVE request: python-pyrad
> > > insecurities
> > On 02/15/2013 09:14 AM, Vincent Danen wrote:
> > > > > Could a CVE be assigned to the following two issues please?
> > > > >
> > > > > #1: https://bugzilla.redhat.com/show_bug.cgi?id=911682
> > > > >
> > > > > Nathaniel McCallum of Red Hat reported that pyrad was using
> > > > > Python's random module in a number of places to generate
> > > > > pseudo-random data.  In the case of the authenticator data,
> > > > > it was being used to secure a password sent over the wire.
> > > > > Because Python's random module is not really suited for this
> > > > > purpose (not random enough), it could lead to password
> > > > > hashing that may be predictable.
> >
> > Please use CVE-2013-0294 for this issue.
> >
> >
> > > > > #2: https://bugzilla.redhat.com/show_bug.cgi?id=911685
> > > > >
> > > > > Nathaniel McCallum of Red Hat reported that pyrad was
> > > > > creating serialized RADIUS packet IDs in the CreateID()
> > > > > function in packet.py. This is not suitable for RADIUS as the
> > > > > RFC specifies that the ID must not be predictable.  As a
> > > > > result, the ID of the next packet sent can be spoofed.
> >
> > Please use CVE-2013-0295 for this issue.
>
> Please REJECT CVE-2013-0295 and use CVE-2013-0294 for both issues
> (same code issue, same version, same reporter).

Ok, so the reporter indicated that the patch noted does not fix both
issues.  It fixed one issue and the other CreateID() function referenced
wasn't actually the CreateID() function he meant:

https://github.com/wichert/pyrad/blob/38f74b36814ca5b1a27d9898141126af4953bee5/pyrad/packet.py#L518

Different function and you can see that:

    CurrentID = (CurrentID + 1) % 256
    return CurrentID

isn't that great.

Now that CVE-2012-0295 has been rejected, I suppose we cannot re-use it
for this, but I think we need another CVE for the "real" CreateID()
sequential RADIUS packet ID issue.

Sorry about this.  I hadn't realized there was more than one CreateID()
function in there and the original report was short on details.

--
Vincent Danen / Red Hat Security Response Team