Clarification: OpenSSH entropy sources?
From: Jonathan Sturges (jonathan_at_sprintmail.com)
To: <email@example.com> Date: Fri, 2 May 2003 11:45:48 -0400
I got no reply from comp.security.ssh so I'm forwarding it to the list.
Any straight answers on this tremendously appreciated.
> I am building OpenSSL 0.9.7b and OpenSSH 3.6.1p1 under Solaris 8. My
> OpenSSL is configured to use the Solaris /dev/random supplied in
> With OpenSSH 3.6.1p1, how are entropy sources handled? During
> configuration, it identifies that OpenSSL's PRNG is internally seeded;
> is true, due to the existence of /dev/random.
> But, my confusion comes from the use of OpenSSH's rand-helper. If you run
> "configure" with "--with-rand-helper", at the end of the configuration you
> Random number source: ssh-rand-helper
> ssh-rand-helper collects from: Command hashing (timeout 200)
> WARNING: you are using the builtin random number collection
> service. Please read WARNING.RNG and request that your OS
> vendor includes kernel-based random number collection in
> future versions of your OS.
> ...which sounds worrisome. It sounds like it will never use OpenSSL's
> internal source (ie, /dev/random) at all.
> But if you use "--without-rand-helper" you get:
> Random number source: OpenSSL internal ONLY
> So the question is, is OpenSSH with rand-helper enabled ignoring the fact
> that OpenSSL has an entropy source, or will it try to use OpenSSL's
> source and just "fall back" to ssh-rand-helper if need be? I saw a
> post where someone says the latter is true; however, due to the warning
> get at configuration time it makes me think it's only trying to use
> ssh-rand-helper and never tries OpenSSL's internal source. OTOH, I never
> get "PRNG not seeded" messages, so maybe it isn't starving for entropy.
> Can someone explain what is really happening?