Date: Wed, 11 Jan 2023 13:18:46 +0100 From: Frank Lichtenheld Message-ID: References: <20230110170257.113527-1-frank@...2641...> MIME-Version: 1.0 In-Reply-To: Subject: Re: [Openvpn-devel] [PATCH] check_engine_keys: make pass with OpenSSL 3 List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 8bit Content-Disposition: inline To: Gert Doering Cc: openvpn-devel@lists.sourceforge.net On Wed, Jan 11, 2023 at 08:39:51AM +0100, Gert Doering wrote: > Hi, > > On Tue, Jan 10, 2023 at 06:02:57PM +0100, Frank Lichtenheld wrote: > > @@ -27,7 +27,7 @@ ${top_builddir}/src/openvpn/openvpn --cd ${top_srcdir}/sample --config sample-co > > # first off check we died because of a key mismatch. If this doesn't > > # pass, suspect openssl of returning different messages and update the > > # test accordingly > > -loggrep '(X509_check_private_key:key values mismatch|func\(128\):reason\(116\))' log.txt || { echo "Key mismatch not detected"; exit 1; } > > +loggrep '(x509 certificate routines:(X509_check_private_key)?:key values mismatch|func\(128\):reason\(116\))' log.txt || { echo "Key mismatch not detected"; exit 1; } > > This change does not convince me - assuming normal regex here, the > original grep would have found any lines with > > X509_check_private_key:key values mismatch > > in them, while the new one would find > > X509_check_private_key::key values mismatch > X509_check_private_key:X509_check_private_key:key values mismatch > > but not "the old one with just one :" > > Should this be > > (x509 certificate routines:(X509_check_private_key:)?key va... > > with the second ":" part of the (...:)? zero-or-once match? So, OpenSSL 3 prints: x509 certificate routines::key values mismatch yes, two colons! OpenSSL 1.1 prints: x509 certificate routines:X509_check_private_key:key values mismatch My patch matches for both. But you are right that I made it slightly stricter, since it previously we only matched on X509_check_private_key:key values mismatch for OpenSSL 1.1. If you would prefer that we could go for (x509 certificate routines:|X509_check_private_key):key values mismatch But I like my version better. If that passes buildbot, I think it should be "good enough". > (... as a side note, I hate this test... it's easily the test with the > most commits to make it behave across platforms and SSL library versions) yeah, definitely. Regards, -- Frank Lichtenheld