On Thu, Apr 21, 2005 at 12:41:03AM +0200, Bernhard Schmidt wrote:
> JuanJo Ciarlante schrieb:
>
> >>I've been just trying the OpenVPN IPv6 transport patch you posted ...
> > I've finished 2.0 final port of my patch; I've tested under FreeBSD-5.3
> > (compilation and run OK).
> >
> > Get from:
> > http://www.irrigacion.gov.ar/juanjo/openvpn/
>
> Great, thanks ... Now it got just another problem :-)
>
> When you do something with routes (--route for example, or
> --redirect-gateway) you get
>
> Thu Apr 21 00:38:53 2005 OpenVPN 2.0 i386-unknown-freebsd5.3 [SSL]
> [PF_INET6] built on Apr 21 2005
> Thu Apr 21 00:38:53 2005 Static Encrypt: Cipher 'BF-CBC' initialized
> with 128 bit key
> Thu Apr 21 00:38:53 2005 Static Encrypt: Using 160 bit message hash
> 'SHA1' for HMAC authentication
> Thu Apr 21 00:38:53 2005 Static Decrypt: Cipher 'BF-CBC' initialized
> with 128 bit key
> Thu Apr 21 00:38:53 2005 Static Decrypt: Using 160 bit message hash
> 'SHA1' for HMAC authentication
> Thu Apr 21 00:38:53 2005 Assertion failed at socket.c:1495
> Thu Apr 21 00:38:53 2005 Exiting
>
> same on linux. Any idea? :-)
Yep .. redirect-default-gateway semantics (even if NOT used ) are triggering
this, please try attached patch
Feedbacck very welcome!
Regards
--
--Juanjo
# Juan Jose Ciarlante (JuanJo) jjo ;at; mendoza.gov.ar #
# GnuPG Public Key: gpg --keyserver wwwkeys.eu.pgp.net --recv-key 66727177 #
# Key fingerprint: 0D2F 3E5D 8B5C 729E 0560 F453 A3F7 E249 6672 7177 #
? .deps
? .gdb_history
? .socket.c.swp
? Makefile
? autom4te.cache
? config.h
? config.log
? config.status
? occ.d
? openvpn
? out.k
? stamp-h1
? tspc.log
Index: socket.c
===================================================================
RCS file: /home/jjo/cvs.juanjo/openvpn/openvpn-2/socket.c,v
retrieving revision 1.4.2.1
diff -u -r1.4.2.1 socket.c
--- socket.c 19 Apr 2005 18:25:11 -0000 1.4.2.1
+++ socket.c 21 Apr 2005 17:32:58 -0000
@@ -1492,7 +1492,20 @@
{
const struct link_socket_addr *lsa = info->lsa;
+/*
+ * This logic supports "redirect-gateway" semantic, which
+ * makes sense only for PF_INET routes over PF_INET endpoints
+ *
+ * Maybe in the future consider PF_INET6 endpoints also ...
+ * by now just ignore it
+ *
+ */
+#if defined ( USE_PF_INET6 ) || defined ( USE_PF_UNIX )
+ if(lsa->actual.addr.sa_family != AF_INET)
+ return 0;
+#else
ASSERT(lsa->actual.addr.sa_family == AF_INET);
+#endif
if (addr_defined (&lsa->actual.addr))
return ntohl (lsa->actual.in.sin_addr.s_addr);
else if (addr_defined (&lsa->remote.addr))
Index: socket.h
===================================================================
RCS file: /home/jjo/cvs.juanjo/openvpn/openvpn-2/socket.h,v
retrieving revision 1.4.2.1
diff -u -r1.4.2.1 socket.h
--- socket.h 19 Apr 2005 18:25:11 -0000 1.4.2.1
+++ socket.h 21 Apr 2005 17:32:58 -0000
@@ -84,7 +84,9 @@
union sockaddr_ovpn {
struct sockaddr addr;
struct sockaddr_in in;
+#ifdef USE_PF_INET6
struct sockaddr_in6 in6;
+#endif
#ifdef USE_PF_UNIX
struct sockaddr_un un;
#endif
