Date: Wed, 5 Aug 2020 08:43:18 +0200 From: Gert Doering Message-ID: <20200805064318.GY75094@...1296...> References: <20200729113835.2415-1-arne@...1227...> <20200804202010.GV75094@...1296...> MIME-Version: 1.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="VaKJWhUROU/xPxjb" Content-Disposition: inline In-Reply-To: Subject: Re: [Openvpn-devel] [PATCH v2] Rework NCP compability logic and drop BF-CBC support by default List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Arne Schwabe Cc: Gert Doering , openvpn-devel@lists.sourceforge.net --VaKJWhUROU/xPxjb Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Content-Disposition: inline HI, On Wed, Aug 05, 2020 at 12:20:54AM +0200, Arne Schwabe wrote: > > Is that intentional? >=20 > Yes. That is intentional. If you do not have any cipher option in the > config, there is nowadays a very high change that you allow BF-CBC by > "accident". I encountered this first-hand ("I do want to put as few > option in a config as possible"). OK, I can see that line of reasoning. This needs to be put very prominently into the release notes. Updating on server test success - with the core dump fix, but still *no* "--cipher" in the server configs, I get 22... Test sets succeeded: 8. Test sets failed: 1 2 3 4 6. 23.small... Test sets succeeded: none. Test sets failed: 1 2 3 4. 23... Test sets succeeded: 8 8a 9. Test sets failed: 1 1a 1b 1d 2 2a 2b 2c 2d 3 4 5 6. 24... Test sets succeeded: 1 1a 1b 1c 1d 1e 2 2b 2d 2e 3 5 6 8 8a 9. Test sets failed: 2a 2c 4 4a. master... Test sets succeeded: 1 1a 1b 1c 1d 1e 2 2b 2c 2d 2e 3 5 5a 5b 5c 5d 5v1 5v2= 5v3 5w1 5w2 5w3 5w4 5x1 5x2 5x3 5x4 6 7 7x 8 8a 9 2f. Test sets failed: 2a 4 4b. so the 2.2/2.3-small/2.3 failures are expected. 2a and 2c (for 2.4) is expected, because that's "--ncp-disable" 4 got broken somewhat accidently - this is tap tests, relying on a=20 secondary client to be connected to ping "across the tap". That client is using pushed ccd ciphers, which fails in interesting ways Aug 5 08:39:33 gentoo tap-udp-p2mp[2418]: freebsd-74-amd64/2001:608:0:814:= :f000:3 PUSH: No common cipher between server and client. Server data-ciphe= rs: 'CAMELLIA-128-CBC', client supported ciphers 'AES-256-GCM:AES-128-GCM' but this error message is misleading - the client has=20 ncp-ciphers CAMELLIA-128-CBC:AES-256-GCM:AES-128-GCM in its config, but it's a 2.4 client so cannot signal this to the master. So we *will* break pushed ciphers to 2.4 client swith non-AEAD ciphers=20 here. Any idea how to tackle this? Test run with "cipher bf-cbc" in all server configs next... gert --=20 "If was one thing all people took for granted, was conviction that if you=20 feed honest figures into a computer, honest figures come out. Never doubte= d=20 it myself till I met a computer with a sense of humor." Robert A. Heinlein, The Moon is a Harsh Mistre= ss Gert Doering - Munich, Germany gert@...1296... --VaKJWhUROU/xPxjb Content-Type: application/pgp-signature; name="signature.asc" -----BEGIN PGP SIGNATURE----- iQGcBAEBAgAGBQJfKlUFAAoJEB2Cnv7KVigSgu8L/RgvdvK3QhLor8qKZOIHCcj0 TPstofjMaed+/i855ldksVccvDOqBu014dT2OKVIwOqT2oLu8QdNFpyg5iZmvnjz +44bdbDqDwsHbl/GYXQUqj6joZi0BdxQEPs4CcZhn/o6H+ZHw57rVkNSn6UsNmwX FIVpYy1hzejpHWRbZyH0rgiD8NQ7rwasyMZR3lZEyEUvLKs4a7vdWY3MttMmnp+c a1NmjjQjYDDHInNIMcmyQ1amwOm9gX16hJHQSd3RJc+UYWxaQI2KnKNnxRRnTLGB zK1syO4G6kpmO7t31eJZ9sBg3Sew8GJlw13N5no35ifJwyOyniZfxGN0MbFdxGBn vaObj/uQ8Y/V2+mQmVRVSOBXNaMiuV1pi6dHdyAVIX3QEGGfGI6w/cxQ33RTKUi5 qchF0twUvO2Z9QUw+I1MfEc4j5KEmKODyZuaeMLdrb/vNMKfK1mrZJ4fhncO3FGk hso8hFGKgSXnV88WBImf4do3aioGUgGyorsgTVgfgg== =I14O -----END PGP SIGNATURE----- --VaKJWhUROU/xPxjb--