[Pkg-openssl-devel] Bug#793565: Bug#793565: libssl1.0.0: HMAC broken after upgrade to 1.0.2d-1

[Kurt Roeckx]

On Sat, Jul 25, 2015 at 06:15:10PM +0200, Marc Lehmann wrote:
> On Sat, Jul 25, 2015 at 10:48:51AM +0200, Kurt Roeckx <kurt at roeckx.be> wrote:
> > > upgrading libssl1.0.0 from 1.0.1k-3+deb8u1 to 1.0.2d-1 breaks HMAC
> > > authentication in a gvpe compiled with 1.0.1k-3.
> > 
> > I will need more information other than that it doesn't work.
> 
> Just ask, but without knowing what you want to know (you haven't said
> anything), I can only guess.
> 
> > I don't have any idea who gvpe works.
> 
> Well, many people "work gvpe". Maybe you meant "how"? gvpe uses openssl's
> HMAC (by default hmac-sha512) to verify packet integrity, and when
> upgrading libssl to 1.0.2d-1, for some connections, every packet gets a
> HMAC authentication error (causing complete loss of connectivity) that
> goes away once libssl is downgraded again.

I tried some of the test vectors in rfc4231 with the version from
squeeze, wheezy, jessie and stretch and they all produce the same
correct output.

For instance case 2:
echo -n "what do ya want for nothing?" | openssl dgst -hmac Jefe -sha512
(stdin)= 164b7a7bfcf819e2e395fbe73b56e0a387bd64222e831fd610270cd7ea2505549758bf75c05a994a6d034f65f8f0e6fdcaeab1a34d4a6b4b636e070a38bce737

So it seems more likely that this is either a bug in gvpe or
something cpu specific.

> > > Since the ABI was apparently broken before (#788511), chances are high
> > > that the fix in 1.0.2d-1 isn't effective and 1.0.2d-1 is still ABI
> > > incompatible to the version in jessie.
> > 
> > This is very unlikely.  But if it's really the case rebuilding
> > against that version should fix the issue.
> 
> I think you are not understanding the problem on a very basic level here
> - rebuilding a program using a shared library will not and can not fix
> bugs in that library. If libssl 1.0.2d-1 is incompatible to 1.0.1k-3
> w.r.t. HMAC generation, then the only way to fix this is to patch and fix
> libssl, OR bump the soname, to indicate an incompatible version.

Please note that I'm not saying that this is how you should fix
it, it's a test to see if that's the issue or not.  I might not
have worded it the best way.


Kurt