OpenSSH and S/Key (Re: Another shapshot)

[Ben Lindstrom]

I'm tired.. but I think I see the problem.

Should S/Key and OpenSSL SHA1 code be inter-changeable? =)  Because that i
what it seems happened.  OpenSSL functions are being used instead of S/Key
libraries.

- Ben


On Wed, 8 Nov 2000, Ben Lindstrom wrote:

> On Wed, 8 Nov 2000, Gert Doering wrote:
> > On Wed, Nov 08, 2000 at 01:34:04AM -0600, Ben Lindstrom wrote:
> > > > No.  Tried using up[0]+up[1] instead, but that also gives me a constant
> > > > number - the first always gives me "58", the second "85", but neither
> > > > gives the current challenge.
> > > > 
> > > > So - is anybody working on S/Key in OpenSSH?  Is somebody using it on a
> > > > non-BSD platform?  If yes, what did you do, which s/key library, which
> > > > CPU endianness, ...?
> > > > 
> > > If I knew of a S/Key library outside of the code in the OpenBSD tree I'd
> > > be happy to compile it up under Linux and see if I can mimic this problem.
> > 
> > There is one that has been ported from OpenBSD (similar to OpenSSH), I
> > found it by looking in "www.google.com" for "skey source".  I found it
> > at http://www.sparc.spb.su/solaris/skey/, which, as I understand, is
> > also its home site.  I use version 1.1.
> > 
> > The S/Key library itself seems to work fine, that is:
> > 
> Hmm... I see.  I finally have it running after tweaking the skey and 
> and OpenSSH.  
> 
> And I see what you mean.. And this is on a linux/i86 box.
> 
> Challenge states: 40 etoh54947
> It should state: 98 etoh39309
> 
> I doubt it's any type of endedness.  I think it's a problem with
> OpenSSH portable code.  (doing a diff -u of auth-skey.c and you'll
> notice large blocks of function changes..Which I'm attempting to track
> down)
> 
> I think there is a good reason why it was disabled. =)
> 
> Does anyone have this code working correctly on any platform (besides the
> given OpenBSD =)??
> 
> - Ben
> 
>