======================================================= mpg123-0.59r buffer overflow vulnerability ======================================================= Davide Del Vecchio Adv#10 Discovered in: 16/08/2003 Date: 06/09/2003 Version affected: mpg123-0.59r and maybe mpg123-0.59s CVE: CAN-2004-0805 Tested and verified on Linux debian SID and OpenBSD. The same vulnerable code is also present in the development version 0.59s, but new and unrelated header checks have prevented the test case for 0.59r from crashing this version as well. A more carefully crafted file might hit the vulnerability on 0.59s as well. It should affect almost every OS with mpg123 package installed. Description: mpg123 reads one or more files (or standard input if
======================================================= mpg123-0.59r buffer overflow vulnerability ======================================================= Davide Del Vecchio Adv#10 Discovered in: 16/08/2003 Date: 06/09/2003 Version affected: mpg123-0.59r and maybe mpg123-0.59s CVE: CAN-2004-0805 Tested and verified on Linux debian SID and OpenBSD. The same vulnerable code is also present in the development version 0.59s, but new and unrelated header checks have prevented the test case for 0.59r from crashing this version as well. A more carefully crafted file might hit the vulnerability on 0.59s as well. It should affect almost every OS with mpg123 package installed. Description: mpg123 reads one or more files (or standard input if - is specified) or URLs and plays them on the audio device (default) or outputs them to stdout. The problem: A malicious formatted mp3/2 causes mpg123 to fail header checks, this may allow arbitrary code to be executed with the privilege of the user trying to play the mp3. For more informations read and understand the patch. Solution: Author has been contacted with no answer. A patch has been provided by Daniel Kobras, the Debian mpg123 package mantainer. The patch is attached at the end of this document. Credits: Davide Del Vecchio would like to thank all the people supporting him and his research, at Telecom Italia S2OC - Security Services Operation Center; especially Roberto Barbieri "sirius", Marcelo Borges "formica", Matteo Cantoni "goony", Demetrio Milea and Joy Gian Luigi Savioli. Daniel Kobras for his help. I love yellow cats. Disclaimer: The information within this paper may change without notice. Use of this information constitutes acceptance for use in an AS IS condition. There are NO warranties with regard to this information. In no event shall the author be liable for any damages whatsoever arising out of or in connection with the use or spread of this information. Any use of this information is at the user's own risk. ^^^^^^^^ Please send suggestions, updates, and comments to: Davide Del Vecchio "Dante Alighieri" - dante at alighieri dot org http://www.alighieri.org http://www.bluejack.it http://www.ezln.it ---[snip]--- Index: layer2.c =================================================================== RCS file: /home/kobras/cvsroot/debian/mpg123/layer2.c,v retrieving revision 1.1.1.1 diff -u -r1.1.1.1 layer2.c --- layer2.c 1999/02/10 12:13:06 1.1.1.1 +++ layer2.c 2004/09/02 21:43:58 @@ -265,6 +265,11 @@ fr->jsbound = (fr->mode == MPG_MD_JOINT_STEREO) ? (fr->mode_ext<<2)+4 : fr->II_sblimit; + if (fr->jsbound > fr->II_sblimit) { + fprintf(stderr, "Truncating stereo boundary to sideband limit.\n"); + fr->jsbound=fr->II_sblimit; + } + if(stereo == 1 || single == 3) single = 0; - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - Davide Del Vecchio "Dante Alighieri" dante@alighieri.org ~ dante@bluejack.it http://www.alighieri.org http://www.bluejack.it http://www.ezln.it - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - _______________________________________________ Full-Disclosure - We believe in it. Charter: http://lists.netsys.com/full-disclosure-charter.html Reproducible: Always Steps to Reproduce: 1. 2. 3.
This advisory says that mpg123-0.58s is "maybe" vulnerable. Can someone please verify?
layer2.c in pre0.59s looks vulnerable.
CAN-2004-0805 (under review)
Richard Johnson <thief@bugtraq.org> replied to this advisory saying: I coded an exploit for that bug for my hacker crew (GOBBLES Security) back in 1992 AD. Get with the program, lamer. ============================ His exploit can be found here: http://www.derkeiler.com/Mailing-Lists/securityfocus/bugtraq/2003-01/att-0135/01-jinglebellz.c I was unable to get this exploit to work with mpg123, it just exits saying "Can't find frame start".
Now someone on that list said that the exploit is for something different. Gah, I'm done following that thread. Let's figure it out on our own.
I don't see a reason _not_ to apply this patch. Any objections? If not, can one of the sound guys apply this patch?
sound please apply the patch.
The patch applies and builds ok on my side. However, before I go commiting this and stable mark requesting for arches, I need to know what exactly is being fixed here. The first exploit was found to be the wrong one, so I need the right exploit, that way I can go to the arches and say "This is what can exploit the program, try this and see if it runs successfully". Otherwise the patching security wise (not functionality wise) seems somewhat pointless.
There is no known exploit for 0.59s, the exploit that I posted was only for 0.59r. The advisory says that 0.59s has the same vulnerable code as 0.59r, but new and unrelated header checks have prevented the test case for 0.59r from crashing this version. A more carefully crafted file might hit the vulnerability on 0.59s as well. All of the information in the advisory still remains true, it's just that this vulnerability has not been exploited yet for our stable version.
Sound please provide an updated ebuild. Debian applied the patch according to the changelog: http://packages.debian.org/changelogs/pool/non-free/m/mpg123/mpg123_0.59r-16/changelog http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=270542
sound@ your taking to long with this one. >=8 days. Security bugs should be resolved with 48 hrs. I guess I'll do your job for you.
mpg123-0.59s-r4.ebuild added to the portage tree. KEYWORDS="~x86 ~ia64 ~amd64 ~ppc ~sparc ~alpha ~hppa ~mips" Arch maintainers please test and mark stable.
Thx Solar. Increasing to A2 (execution of arbitrary code).
Stable on x86
stable on ppc
Stable on hppa.
Stable on sparc.
stable amd64.
Stable on alpha.
GLSA 200409-20 ia64: please mark stable to benefit from GLSA