Your patch has been applied to the master branch.
Tested with t_client test, one binary with --enable-iproute2, one without.
"Route add"-Diffs look nicely different, including the /128 bypass...
Wed Jun 5 11:09:02 2019 /bin/ip route add 194.97.144.0/24 via x.x.x.73
Wed Jun 5 11:09:02 2019 add_route_ipv6(2001:x:x:x::f000:11/128 ->
fe80::250:43ff:fe01:dc37 metric 1) dev enp0s25
Wed Jun 5 11:09:02 2019 /bin/ip -6 route add 2001:x:x:x::f000:11/128 dev
enp0s25 via fe80::250:43ff:fe01:dc37 metric 1
Wed Jun 5 11:09:02 2019 add_route_ipv6(2001:608::/32 -> 2001:x:x:x::1 metric
-1) dev tun0
Wed Jun 5 11:09:02 2019 /bin/ip -6 route add 2001:608::/32 dev tun0
Wed Jun 5 11:09:21 2019 net_route_v4_add: 194.97.144.0/24 via x.x.x.73 dev
[NULL] table 0 metric -1
Wed Jun 5 11:09:21 2019 add_route_ipv6(2001:x:x:x::f000:11/128 ->
fe80::250:43ff:fe01:dc37 metric 1) dev enp0s25
Wed Jun 5 11:09:21 2019 net_route_v6_add: 2001:x:x:x::f000:11/128 via
fe80::250:43ff:fe01:dc37 dev enp0s25 table 0 metric 1
Wed Jun 5 11:09:21 2019 add_route_ipv6(2001:608::/32 -> 2001:x:x:x::1 metric
-1) dev tun0
Wed Jun 5 11:09:21 2019 net_route_v6_add: 2001:608::/32 via :: dev tun0 table
0 metric -1
.. but the resulting output of "ip route" + "ip -6 route" is exactly
the same. Well done :-)
OTOH... :-) - looking at all the changes needed to transport &c->net_ctx,
I wonder why you didn't just hide it in c->c1.tuntap, as we did for the
windows socket... (tt->options.msg_channel). Almost everything related
to "routing and ifconfig" needs to have the tt pointer available already
today...
Warts:
+#else
+ const char *iface;
+ int metric;
+#endif
.. I think this really should be inside the #ifdef TARGET_LINUX block,
not outside in an #else for an #if !defined(TARGET_LINUX)... so the
code further down would then become
#if defined(TARGET_LINUX)
const char *ifcace = NULL;
if (is_on_link(...))
.. having declaration and initialization in one place, and "where it
belongs". Same thing for add_route_ipv6()...
+#if defined(TARGET_LINUX)
+ int metric;
+#endif
.. this is adding an extra #ifdef and taking 3 extra lines instead of
just having "int metric = -1;" down where it's initialized.
delete_route() sports this beauty...
+#if !defined(TARGET_LINUX)
const char *network;
#if !defined(ENABLE_IPROUTE) && !defined(TARGET_AIX)
const char *netmask;
#endif
#if !defined(TARGET_LINUX) && !defined(TARGET_ANDROID)
const char *gateway;
+#endif
.. so you have two linux-specific defines (ENABLE_IPROUTE and TARGET_LINUX)
*inside* an "#if !defined(TARGET_LINUX)"... plus the "metric" thing...
commit aec4a3d1b6a9e4d9e584b368126da061c15b174b
Author: Antonio Quartulli
Date: Wed Dec 19 15:01:15 2018 +1000
route.c: use new networking API to handle routing table on Linux
Signed-off-by: Antonio Quartulli <[email protected]>
Acked-by: Arne Schwabe <[email protected]>
Message-Id: <[email protected]>
URL:
https://www.mail-archive.com/[email protected]/msg18029.html
Signed-off-by: Gert Doering <[email protected]>
--
kind regards,
Gert Doering
_______________________________________________
Openvpn-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/openvpn-devel
