My bind (9.2.5-r6) keeps crashing on me with this kind of log messages: rbt.c:1261: INSIST(order < 0) failed Judging after http://www.isc.org/index.pl?/sw/bind/bind9.3-beta.php , there are several security bugs involving this keyword solved in the latest beta release. Please bump it (and mark it stable since it is a security problem).
> there are several security bugs unfortunately, i'm unable to find any security issues on link you provided. can you please paste security-related changeLog-entries here? anyway, 9.3.2b1 will be committed today.
Anything with INSIST looks as potential DoS to me: 1867. [bug] It was possible to trigger a INSIST in dlv_validatezonekey(). [RT #14846] 1861. [bug] dig could trigger a INSIST on certain malformed responses. [RT #14801] My named daemon keeps crashing once at every 3 days or so, with following log lines: hades log # egrep 'INSIST|assertion' /var/log/by.host/hades.log Oct 1 23:59:12 hades named[24656]: rbt.c:1261: INSIST(order < 0) failed Oct 1 23:59:12 hades named[24656]: exiting (due to assertion failure) Oct 11 10:16:14 hades named[28434]: rbt.c:1261: INSIST(order < 0) failed Oct 11 10:16:14 hades named[28434]: exiting (due to assertion failure) Oct 14 12:39:01 hades named[7407]: rbt.c:1261: INSIST(order < 0) failed Oct 14 12:39:01 hades named[7407]: exiting (due to assertion failure) Oct 17 19:23:01 hades named[1692]: rbt.c:1261: INSIST(order < 0) failed Oct 17 19:23:01 hades named[1692]: exiting (due to assertion failure) Maybe I'm the first to notice such problem, but what are the odds?
i'm not sure about possibility of DoS there, but 9.3.2b1 is in portage now.
Is 9.3.2b1 ready to be marked stable or should i reset status to upstream?
to upstream, probably
thanks for the bumping. I've just installed the new version. I'll keep you informed if this version still have same problem, but I think we should wait for at least a week before I counclude it is solved.
FYI: idn support is broken there.
Not sure this is a security issue... It's a bug alright. But for it to become a vulnerability it needs to be triggerable by an attacker, and I can't see any evidence of this possibility.
Please reopen the bug if you find some way to trigger those DoS, because afaict they are only stability bugs.