BIND security update protects against serious server crash
Updated The Internet Systems Consortium is advising BIND users to update immediately to protect against a bug that may already be under attack to crash vulnerable servers.
The ISC says an unidentified network event caused BIND 9 resolvers to cache an invalid record, and when subsequent queries requested the invalid record, the servers crashed with the following assertion failure:
It’s also apparently being exploited to attack networks, with multiple members of the BIND users email list from Germany, France and the US reporting simultaneous crashes across multiple servers.
The ISC describes the bug as a potential zero-day exploit with no workaround, and urges immediate upgrade to BIND 9.8.1-P1, 9.7.4-P1, 9.6-ESV-R5-P1, or 9.4-ESV-R5-P1.
The patched code is designed to handle both the cache and the crash. The ISC’s advisory states:
“When a client query is handled, the code which processes the response to the client has to ask the cache for the records for the name that is being queried. The first component of the patch prevents the cache from returning the inconsistent data. The second component prevents named from crashing if it detects that it has been given an inconsistent answer of this nature.”
According to Blair Strang, a security consultant at Australian company SenseOfSecurity, while the situation is “serious”, the “sky is not falling”.
Because past incidents have made BIND developers paranoid, Strang explained to The Register, they’ve ramped up the safety checks inside their code over the years – and this crash is due to the safety checks.
“The name server knows ‘something has gone wrong’ and exits as a defensive measure,” Strang told us. “The security checks were added in BIND 9 because BIND 8 got hammered; several exploits were released for it, which is several too many for one of the essential services holding the Internet together.”
He added that the current vulnerability is still mysterious, and added that “the patch from the BIND guys doesn’t actually fix the bug – it just papers over the crash caused by the assertion (this situation could probably change during the day as people find out more).”
Since the BIND developers don’t yet know what payload triggered the crash, remote code execution is feasible, he said. “Let’s hope it remains a denial of service condition”, he added.
BIND – the Berkeley Internet Name Daemon – has been a favourite target for black-hats. When an exploitable bug is discovered, miscreants can (for example) redirect users to counterfeit sites to harvest data like account IDs and passwords.
Unlike a phishing email attack, which is merely an IQ test that too many people fail, attacks on DNS resolvers are a different animal entirely. If users have typed their bank’s URL in the browser, they’re likely to believe that they’ve arrived at the bank’s site, even if they haven’t. ®
This post was updated to add comment from Strang.