Re: Compiling untrusted source -- what are the risks?

From: Karsten W. Rohrbach (karsten@rohrbach.de)
Date: 06/14/01 

Date: Thu, 14 Jun 2001 21:04:27 +0200
From: "Karsten W. Rohrbach" <karsten@rohrbach.de>
To: Kris Kennaway <kris@obsecurity.org>

Kris Kennaway(kris@obsecurity.org)@2001.06.13 13:03:13 +0000:
> On Wed, Jun 13, 2001 at 09:24:02AM +0300, Alex Popa wrote:
>
> > The step I am worried about is the compiling, since I do need to have
> > the include files and libraries available. The output should be a
> > statically linked file, which would run in a jail (separate one per
> > source file) which contains nothing more than the compiled binary, and
> > the input file. The evaluation program will run in a separate jail,
> > given only the output file from the program, and maybe an "expected
> > results" file. I plan on using ipfw to block all traffic on that
> > machine (will be a dedicated machine) not coming from a few trusted
> > uids (like root and the evaluation process). I also plan setting up
> > resource limits, and not running more evaluation jobs at the same time
> > (ruins timing).
>
> You could do this step in a jail if you wanted to. If you're using
> user-supplied makefiles, then they can run arbitrary commands. If
> you're using a fixed set of compiler invocations and the standard
> toolchain then it should probably be okay (I don't know of any ways to
> cause the compiler toolchain to execute arbitrary commands during
> compilation).
>

although, being a paranoid *** myself, i would reconstruct the whole
jail after creating a backup of the work environment only when the
evaluation process for one package is finished. this gives you a clean
slate point of start for everything again.

/k 

-- 
> Only wimps use tape backups; real men put their software on ftp-servers
> and let the rest of the world mirror it. --Linus Torvalds
KR433/KR11-RIPE -- WebMonster Community Founder -- nGENn GmbH Senior Techie
http://www.webmonster.de/ -- ftp://ftp.webmonster.de/ -- http://www.ngenn.net/
karsten&rohrbach.de -- alpha&ngenn.net -- alpha&scene.org -- catch@spam.de
GnuPG 0x2964BF46 2001-03-15 42F9 9FFF 50D4 2F38 DBEE  DF22 3340 4F4E 2964 BF46

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-security" in the body of the message

Quantcast