Hi,
On Tue, Feb 18, 2014 at 07:16:17AM +0100, Tore Anderson wrote:
> > Could you run my test program to see whether that gets it right?
>
> It doesn't:
Meh!
> This is 64-bit Linux, yes:
I'll re-test. I *did* test it on mine...
Linux gentoo 3.3.8-gentoo #1 SMP Fri Jun 29 14:10:01 CEST 2012 x8
Hi Gert,
> Could you run my test program to see whether that gets it right?
It doesn't:
server$ mhome
AF_INET6/IPV6_RECVPKTINFO enabled
setsockopt(IPV6_V6ONLY=0)
Socket bound to local address [AF_INET6][undef]:50001
--
CMSG_NXTHDR=(nil), level=41, type=50
IPV6_PKTINFO
read: fromlen=28, r_len=5
>> This happens both with today's git master and with 2.3.2. BTW Arne's
>> patch to socket.c that fixes the assert crash we originally discussed
>> still isn't included in git master, has it been forgotten about?
> ISTR that Arne wrote something about "this is a quick fix, will send a
> proper pat
Hi,
On Mon, Feb 17, 2014 at 08:28:59PM +0100, Tore Anderson wrote:
> Now that has been fixed upstream (Linux 3.14), I tried again. Here I am
> connecting from 84.209.94.36 to 87.238.35.254, which is a secondary
> address on the server (87.238.35.145 is the primary). The OpenVPN
> server logs:
>
>
* Gert Doering
> On Sat, Jan 11, 2014 at 12:59:02AM +0100, Tore Anderson wrote:
>> FWIW: Now I also got to test git master with what I wanted to in the
>> first place, namely to see whether or not all the nice dualstack changes
>> had also made "multihome" work for IPv4-mapped clients coming in to
Hi,
On Sun, Jan 19, 2014 at 03:04:17PM +0100, Gert Doering wrote:
> In case you want to test yourself, the code is appended. It's not pretty,
> but gets the job done :-)
Here's the latest iteration of the code. The changes compared to
openvpn's socket.c are actually bigger than I expected, as w
Hi,
On Sun, Jan 19, 2014 at 03:04:17PM +0100, Gert Doering wrote:
> In case you want to test yourself, the code is appended.
... it was not. As always when people say "I have attached..." :-/
gert
--
USENET is *not* the non-clickable part of WWW!
Hi,
On Sat, Jan 11, 2014 at 12:59:02AM +0100, Tore Anderson wrote:
> FWIW: Now I also got to test git master with what I wanted to in the
> first place, namely to see whether or not all the nice dualstack changes
> had also made "multihome" work for IPv4-mapped clients coming in to the
> server on
Am 11.01.14 14:07, schrieb Gert Doering:
> Hi,
>
> On Sat, Jan 11, 2014 at 02:00:59PM +0100, Arne Schwabe wrote:
>>> Jan 11 01:07:47 greed ovpn-server[10222]: :::84.209.244.191 TLS:
>>> Initial packet from [AF_INET6]:::84.209.244.191:38878 (via
>>> 2a02:c0:1001:100::253%eth0), sid=98e4314
Hi,
On Sat, Jan 11, 2014 at 02:00:59PM +0100, Arne Schwabe wrote:
> > Jan 11 01:07:47 greed ovpn-server[10222]: :::84.209.244.191 TLS:
> > Initial packet from [AF_INET6]:::84.209.244.191:38878 (via
> > 2a02:c0:1001:100::253%eth0), sid=98e4314f 9ea08578
> >
> > I find the "via 2a02:c0:100
Am 11.01.14 01:12, schrieb Tore Anderson:
> * Arne Schwabe
>
>> Listening on multiple sockets not yet implemented. But a server having
>> proto udp6 should get IPv6 mapped IPv4 clients working. What is
>> happening/not working for you?
> "multihome" doesn't work, the OpenVPN server responds from it
Hi,
On Sat, Jan 11, 2014 at 11:01:58AM +0100, Tore Anderson wrote:
> > I *thought* that both proto udp and proto udp6 would create an IPv6 socket
> > bound to :: with IPV6_V6ONLY set to 0 (= accepting IPv4 and IPv6 packets
> > to any address). Only "proto udp4" should listen to 0.0.0.0.
>
> This
* Gert Doering
> I *thought* that both proto udp and proto udp6 would create an IPv6 socket
> bound to :: with IPV6_V6ONLY set to 0 (= accepting IPv4 and IPv6 packets
> to any address). Only "proto udp4" should listen to 0.0.0.0.
This is not how git master behaves for me. "proto udp" and "proto
Hi,
On Sat, Jan 11, 2014 at 12:59:02AM +0100, Tore Anderson wrote:
> FWIW: Now I also got to test git master with what I wanted to in the
> first place, namely to see whether or not all the nice dualstack changes
> had also made "multihome" work for IPv4-mapped clients coming in to the
> server on
* Arne Schwabe
> Listening on multiple sockets not yet implemented. But a server having
> proto udp6 should get IPv6 mapped IPv4 clients working. What is
> happening/not working for you?
"multihome" doesn't work, the OpenVPN server responds from its primary
IPv4 address (the one the OS selects as
Am 11.01.14 00:59, schrieb Tore Anderson:
> * Arne Schwabe
>
>> Can you try the following one line patch? I will do a better cleanup
>> patch later but this should fix the problem
> Yep, this works. Thanks!
>
> FWIW: Now I also got to test git master with what I wanted to in the
> first place, name
* Arne Schwabe
> Can you try the following one line patch? I will do a better cleanup
> patch later but this should fix the problem
Yep, this works. Thanks!
FWIW: Now I also got to test git master with what I wanted to in the
first place, namely to see whether or not all the nice dualstack chang
Am 10.01.14 15:03, schrieb Tore Anderson:
> * Arne Schwabe
>
>> That is very weird. I would expect it to fail for both cases in the same
>> way. The codepath should trigger the same ASSERT(0) in both cases (or
>> work in both). Can you forward a copy of your server configuration to
>> me? Then I wi
Hi,
On Fri, Jan 10, 2014 at 03:30:10PM +0100, Arne Schwabe wrote:
> But
>
> [AF_INET6]:::87.238.42.150:47707
>
> and
>
> [AF_INET6]2a02:c0:100:0:9e8e:99ff:fed1:5243:56277
>
> should be same the same size and handled both the same way from OpenVPN since
> they are both IPv6.
Good point.
Am 10.01.14 15:07, schrieb Gert Doering:
> Hi,
>
> On Fri, Jan 10, 2014 at 02:51:48PM +0100, Arne Schwabe wrote:
>> That is very weird. I would expect it to fail for both cases in the same
>> way. The codepath should trigger the same ASSERT(0) in both cases (or
>> work in both).
> ... unless somet
Hi,
On Fri, Jan 10, 2014 at 02:51:48PM +0100, Arne Schwabe wrote:
> That is very weird. I would expect it to fail for both cases in the same
> way. The codepath should trigger the same ASSERT(0) in both cases (or
> work in both).
... unless something is overwriting to->ai_family, for example by
* Arne Schwabe
> That is very weird. I would expect it to fail for both cases in the same
> way. The codepath should trigger the same ASSERT(0) in both cases (or
> work in both). Can you forward a copy of your server configuration to
> me? Then I will try to reproduce that problem.
Sure thing, he
* Gert Doering
> Just to be sure I understand: it works for "proto udp6" if you do *not*
> use --multihome on the server?
Correct. Apologies for being unclear.
Tore
Am 10.01.14 14:32, schrieb Tore Anderson:
> If I connect to an openvpn server using "proto udp6" and "multihome", it
> instantly crashes with an assertion failure:
>
> Jan 10 14:25:58 greed ovpn-server[9905]: 2a02:c0:100:0:9e8e:99ff:fed1:5243
> TLS: Initial packet from [AF_INET6]2a02:c0:100:0:9e8e
Hi,
On Fri, Jan 10, 2014 at 02:32:23PM +0100, Tore Anderson wrote:
> If I connect to an openvpn server using "proto udp6" and "multihome", it
> instantly crashes with an assertion failure:
Just to be sure I understand: it works for "proto udp6" if you do *not*
use --multihome on the server?
> Th
If I connect to an openvpn server using "proto udp6" and "multihome", it
instantly crashes with an assertion failure:
Jan 10 14:25:58 greed ovpn-server[9905]: 2a02:c0:100:0:9e8e:99ff:fed1:5243 TLS:
Initial packet from [AF_INET6]2a02:c0:100:0:9e8e:99ff:fed1:5243:56277 (via
2a02:c0:1001:100::145%e
26 matches
Mail list logo
