On Wed, Nov 16, 2011 at 9:13 PM, <valdis.kletni...@vt.edu> wrote: > On Thu, 17 Nov 2011 11:51:09 EST, "Larry W. Cashdollar" said: >> Hello list, >> I am wondering if anyone has more details on the bind9 DoS that just came >> out? (CVE-2011-4313) from what I can tell it appears a negative cached DNS >> object with a valid RR response associated with it(which shouldn't exist) >> will cause a vulnerabile bind9 server to crash. >> >> See lines 1890 - 1896 of query.c >> 1890 if (result == DNS_R_NCACHENXRRSET) { >> 1891 dns_rdataset_disassociate(rdataset); >> 1892 /* >> 1893 * Negative cache entries don't have sigrdatasets. >> 1894 */ >> 1895 INSIST(! dns_rdataset_isassociated(sigrdataset)); >> 1896 } >> >> >> Since allowing recursive queries must be enabled for this to work the >> attacker must force a vulnerable dns server to query a malicous DNS >> server by asking it to look up an NXrecord for a domain the attacker >> controls dns for. Sending a response of NXdomain but having actual DNS >> results in the response. >> >> I am wondering if someone has seen a good write up out there? > > See above. :) (Seriously, that's a good one-para summary analysis of the > issue, better than a lot of 3rd-party advisories we see...) Am I the only guy wondering why an assert is present in production code? Was -DNDEBUG accidentally omitted? Or is a crash expected behavior? I don't believe its the later, since its a DoS with CVE.
Jeff _______________________________________________ Full-Disclosure - We believe in it. Charter: http://lists.grok.org.uk/full-disclosure-charter.html Hosted and sponsored by Secunia - http://secunia.com/