via oss-sec http://seclists.org/oss-sec/2015/q2/260

>     * [Bug 2797] ntp-keygen trapped in endless loop for MD5 keys on big-endian machines.
>     https://bugs.ntp.org/show_bug.cgi?id=2797
> 
>     Patch: http://bk1.ntp.org/ntp-stable/?PAGE=patch&REV=55199296N2gFqH1Hm5GOnhrk9Ypygg
> 
> 
>     While the endless loop is not a security flaw per se
> 
> 
> The unstated rationale here seems to be "ntp-keygen is a command-line
> program that is not normally exposed in a way that crosses privilege
> boundaries."
> 
> The documentation mentions:
> 
>   After setting up the environment it is advisable to update certificates
>   from time to time, if only to extend the validity interval.
>   Simply run
>   @code{ntp-keygen}
>   with the same flags as before to generate new certificates
> 
> It seems plausible that some sites may have created a web interface so
> that operations staff can occasionally do a certificate update (maybe
> with a new key), but these staff don't have login access to the
> machine running NTP. The flaw would give them the new ability to
> (sometimes) launch a CPU consumption attack. However, we have not
> actually heard of anyone with a web-based ntp-keygen arrangement, so
> we currently don't want to assign a CVE ID for that.
> 
>     the fact that
>     ntp-keygen generates non-random keys is. If the lowest byte of the temp
>     variable happens to be between 0x20 and 0x7f and not #, the generated
>     MD5 key will consist of 20 identical characters, meaning only 93
>     possible keys can be generated.
> 
> 
> Use CVE-2015-3405 for this code error that results in a key space
> that's much smaller than expected.


References:
http://seclists.org/oss-sec/2015/q2/260
https://bugzilla.redhat.com/show_bug.cgi?id=1210324
http://web.nvd.nist.gov/view/vuln/detail?vulnId=CVE-2015-3405