Secure Coding mailing list archives

Re: DJB's students release 44 poorly-worded, overblown advisories


From: dtalk-ml () prairienet org
Date: Mon, 20 Dec 2004 21:53:22 +0000

Paco Hope wrote:

Somebody's gotta come up with a  reasonable definition of "remotely
exploitable."

Agreed; this lack of an agreed-upon vocabulary is the cause of all sorts
of misunderstandings that don't make the news.

These days, with everything networked, I'm not sure proximity is a
usually a meaningful distinction, so "local" and "remote" aren't useful
terms.  In my experience, many people tend to use the term "remote" to
mean "no prior account access", while "local" tends to be used to mean
"privilege escalation" (an existing account is required to take
advantage of the flaw).

Brain droppings:  It seems to me that there are at least three different
dimensions of unauthorized access.

Interactivity:

1. Gaining access to some security context by means of social
engineering or predictable user behavior.

2. Gaining access to some security context without user interaction or
social engineering.

Proximity:

a. Gaining privilege access by means of physical access, or

b. Gaining privilege access over a network.

Level:

I. Gaining access to an interesting, but "unprivileged", context, or

II. Gaining access to a privileged "system" context.

So what's usually called a "remote root exploit" is an event with
characteristics 2, b, and II.

Therefore, is a browser bug a "remote" exploit?  Well, yes, strictly
speaking; the author didn't need physical access to run it.  However, it
is also interactive and gains a only user context (which on some poorly
configured systems might mean system privilege, but that's beside the
point).  As Paco points out, an "exploit" which is interactive and gains
only the context granted by the (presumably ignorant) user isn't
necessarily an "exploit", or even a technical problem, though it might
grant the malware access to an actual flaw that will allow escalation.

- --
David Talkington
[EMAIL PROTECTED]
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.6 (FreeBSD)

iD8DBQFBxzZC5FKhdwBLj4sRAnIRAKCvVYGoOLmXtpheyIprkUEZYkNl+ACghXlY
pMrdb8AYA82Nz4Iw8x6/od8=3D
=3DkxLN
-----END PGP SIGNATURE-----





Current thread: