Date: Wed, 10 Oct 2012 18:55:07 +0200 From: Marc Heuse <mh@...sec.de> To: oss-security@...ts.openwall.com CC: Solar Designer <solar@...nwall.com> Subject: Fwd: IPv6 DOS vulnerabilities Alexander kicked me several times in the ass to finally forward this email to oss-security as the embargo time is gone ... so here it is. -------- Original-Nachricht -------- Betreff: IPv6 DOS vulnerabilities Von: Marc Heuse <mh@...sec.de> An: Microsoft Security Response Center <secure@...rosoft.com>, security-officer@...eBSD.org, product-security@...le.com Kopie (CC): distros@...openwall.org Hi folks, this is just a short, quick email about two unspectecular IPv6 implementation weaknesses that result in local network denial-of-service issues in Windows, *BSD (Free and Net, Open not tested) and OS X. distros@ is in cc: for information purposes, although it seems that Linux is not affected, you might want to test though as I have only tested this with a 2.6.x kernel. Issue #1: Flooding the local target with ICMPv6 Neighbor Solicitation messages. As this is handled by the kernel, it consumes all CPU power that is there, leaving no or too little CPU for the user space. All except of OS/X went to 100% CPU, OS X went to 60%+ on a QuadCore Macbook Pro. But my test machine was not able to produce enough packets to even closely get to the satturation point of the network, so the 100% CPU might be reachable there too. In short: a fast multicore CPU helps to negate the impact (unless you are Windows, then this does not help). Test tool: flood_solicate eth0 <IPv6-Linklocal-Address-of-Target> (from the package at www.thc.org/thc-ipv6) Issue #2: Flooding the local network with ICMPv6 Router Advertisement packets containing multiple Routing entries result in either 100% CPU (Windows all Versions with IPv6 enabled) or some noticable CPU impact however IPv6 seem to break for *BSD and OS X. The BSD based systems do not reply to any ICMPv6 Neighbor Solicitation requests anymore, when trying to send locally from the victim systems you get errors (e.g. "connect failed" or "no multicast address on interface") (yes, this is basically a similar issue like RA flooding with autoconfig prefixes from two years ago) I have an unreleased test tool for this attack, if necessary I can package it and send it if needed. (I am sitting on this for over a half year now, sorry for that) Greets, Marc -- Marc Heuse www.mh-sec.de PGP: FEDD 5B50 C087 F8DF 5CB9 876F 7FDD E533 BF4F 891A
Powered by blists - more mailing lists
Please check out the Open Source Software Security Wiki, which is counterpart to this mailing list.
Powered by Openwall GNU/*/Linux - Powered by OpenVZ