Date: Thu, 18 Aug 2022 21:52:54 +0200 From: Gert Doering Message-ID: References: <20220818092638.189632-1-a@...2181...> <20220818100953.191328-1-a@...2181...> MIME-Version: 1.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="mLMhX7hbS0pVjeV9" Content-Disposition: inline In-Reply-To: <20220818100953.191328-1-a@...2181...> Subject: Re: [Openvpn-devel] [PATCH v103 2/7] dco-win: check for incompatible options List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Antonio Quartulli Cc: openvpn-devel@lists.sourceforge.net, Lev Stipakov --mLMhX7hbS0pVjeV9 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Content-Disposition: inline Hi, On Thu, Aug 18, 2022 at 12:09:53PM +0200, Antonio Quartulli wrote: > At the moment dco-win doesn't support --persist-tun and --server, > so check for these options at startup time. >=20 > Signed-off-by: Antonio Quartulli > Signed-off-by: Lev Stipakov I was about to merge this (and thank you for bearing with a grumpy old man's preferences ;-) - but I really think that the difference in these platforms need to stand out right there...). Butbut... This breaks Linux/Non-DCO builds with --dev tun30, and I can't see an obvious reason *why* $ src/openvpn/openvpn --version OpenVPN 2.6_git [git:vw/master/f10c870c140bd3c6] x86_64-pc-linux-gnu [SSL (= mbed TLS)] [LZO] [LZ4] [EPOLL] [MH/PKTINFO] [AEAD] built on Aug 18 2022 (no DCO here, this is "master + 2/7 v103", on Linux) $ openvpn ... --dev tun30 --proto udp ... ... 2022-08-18 21:43:52 OPTIONS IMPORT: peer-id set 2022-08-18 21:43:52 OPTIONS IMPORT: data channel crypto options modified 2022-08-18 21:43:52 net_route_v4_best_gw query: dst 0.0.0.0 2022-08-18 21:43:52 net_route_v4_best_gw result: via 195.30.8.94 dev ens160 2022-08-18 21:43:52 GDG6: remote_host_ipv6=3Dn/a 2022-08-18 21:43:52 net_route_v6_best_gw query: dst :: 2022-08-18 21:43:52 net_route_v6_best_gw result: via 2001:608:1:995a::ffff = dev ens160 2022-08-18 21:43:52 DCO device tun30 opened 2022-08-18 21:43:52 net_iface_mtu_set: rtnl: cannot get ifindex for tun30: = No such device (errno=3D19) 2022-08-18 21:43:52 Linux can't set mtu (1500) on tun30 2022-08-18 21:43:52 Exiting due to fatal error why is it talking about DCO here? Why would it even *know* about DCO in a non-DCO build? Not sure why any of the hunks in this patch have an effect, but if I=20 go back to the current master, --dev tun30 works, both with and without --enable-dco. *scratch head* Ow. +#if defined(TARGET_LINUX) && defined(TARGET_FREEBSD) /* check if any option should force disabling DCO */ -#if defined(TARGET_LINUX) || defined(TARGET_FREEBSD) this is not "just moving" these lines, but requiring "LEENUX AND FREEBSD" now... You owe me a beer :-) (Changing the && to || makes the --dev tun30 test pass again - I'm=20 still not sure *what* happened here... but maybe we should reconsider the decision to have disable_dco default to "false" even on non-DCO builds...) 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... --mLMhX7hbS0pVjeV9 Content-Type: application/pgp-signature; name="signature.asc" -----BEGIN PGP SIGNATURE----- iQGcBAEBAgAGBQJi/piWAAoJEB2Cnv7KVigSLDUL/2NrrtUelL0f4EwIEwDgF+x4 9SFCu6tO6U4TzknWIwd5nQZr9WhF+AZs0rxKODqSEwf0RRJlvBg0f5+rQzlrv0pF eSMU4liJ4fq4ySJBLVVpwQuHVYVRf4T7L42d55/Nk93sI+gPB+u2UqSK4og14/0O f/Vor77UMeSqG3UuFG8ehS8FirBSRD8kUqrd846aXnpOljk2a7aGmIelHQ+p+ryz oJZv4I6VCi38kW6pbtIBI1czJYLAX5v8z7E+OHVgXRD77eyngZh3m3oLryTggyX6 RyfTdV5WFJ2Dkuwkl7C5mEDVO4zxTZIVWdCpUeU4FbTfZbJE+yX87bdG7uy7UYW4 7sfxgJsWvGe8jxQZ+1Sn1D9TOxWX8glxqK3s7OsDlSVk92D+PM3sPo7Hp3GBdjas O3+fK9rrFCP8CANfBf/5OWbeCipMyEscwh6BbZag0M7nmkXLeSmQbY4HDrFcgvs7 //AKOr2tstGXv0BtYv2YqFrOn34395G4/JkBtK0nLQ== =AfTr -----END PGP SIGNATURE----- --mLMhX7hbS0pVjeV9--