Re: Attacks based on predictable process IDs??

[Christopher Allene <cwis () nerim fr>]

Brett Hutley (brett () hutley net) wrote on 2003-11-26 at 11:32:
> Folks, does anyone know why predictable process IDs are considered harmful?
 
Predictable process IDs can be used as a vector to attack programs
vulnerable to race conditions in /tmp file creation, in case those
programs use their PID to create a file, meaning you could possibly
create one (or, for "practical" uses, more often a link or a named pipe)
first.

Programs which uses the following pseudo-code are also vulnerable:

    srand (getppid ());

because the sequence of the so-called rand()om numbers is predictible.
(Arguably, calling srand() passing a xor of your PID and the current
time is no better. See perldoc -f srand for a discussion on this, I'm
getting offtopic.)

Thus, I remember a really weird situation where predictable PIDs were
used to compromise security, it was discussed on BugTraq a while ago,
but I couldn't find a track of it in my BT archive... anyone?

-- 
Christopher Allène

_______________________________________________
Full-Disclosure - We believe in it.
Charter: http://lists.netsys.com/full-disclosure-charter.html