Re: [patch 2/2] ufd v1 - use unsequential O(1) fdmap

[Ulrich Drepper <drepper@redhat.com>]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Davide Libenzi wrote:
>> If randomizing each allocator is too expensive then randomize at the
>> very least the number of the first descriptor you give out.
> 
> Can you tell me how this can be a problem, and in which way making a 
> random thing would help?

In attacking an application every bit of known data can be used in an
exploit.  Be it something as simple as having a predetermined value at a
certain point in the program since it loaded a file descriptor into a
register.

But what I'm mostly thinking about is the case where I/O could be
redirected.  The intruding program could call dup2() and suddenly the
program wanting to write a password to disk could be directed to send it
over a socket.  One could imagine countless such attacks.

I don't say such an attack exists today.  But this is no reason to not
implement these extra security measures.  The cost of a randomized star
base (offset from 2^30) should be zero.

- --
➧ Ulrich Drepper ➧ Red Hat, Inc. ➧ 444 Castro St ➧ Mountain View, CA ❖
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.7 (GNU/Linux)

iD8DBQFGYyid2ijCOnn/RHQRAjRoAJ9XsAazZtc9V3AxaPjiNMjK8jPUZgCdG/Eg
KPug5Sq9REHd6H3AR0ax2aU=
=9iUM
-----END PGP SIGNATURE-----