The stringprep_utf8_to_ucs4 function in libin before 1.31, as used in jabberd2, allows context-dependent attackers to read system memory and possibly have other unspecified impact via invalid UTF-8 characters in a string, which triggers an out-of-bounds read.
https://github.com/jabberd2/jabberd2/issues/85
http://www.ubuntu.com/usn/USN-3068-1
http://www.securityfocus.com/bid/72736
http://www.openwall.com/lists/oss-security/2015/02/23/25
http://www.debian.org/security/2016/dsa-3578
http://lists.opensuse.org/opensuse-updates/2016-08/msg00098.html
http://lists.opensuse.org/opensuse-updates/2015-07/msg00042.html
http://lists.fedoraproject.org/pipermail/package-announce/2015-July/162549.html
http://lists.fedoraproject.org/pipermail/package-announce/2015-July/162537.html
http://git.savannah.gnu.org/cgit/libidn.git/commit/?id=2e97c279