Hi, I have problems to connect from ipv4-only thin clients to our xdm server. There is a bug in the novell-bugzilla:
http://bugzilla.novell.com/show_bug.cgi?id=546632 The reason is that ipv4 addresses are discarded if they are received over a ipv6 socket. The attached patch fixes this. I did not test it yet. I will do it on monday when I have access to the thin clients. However I wanted to get feedback whether such a patch is wanted and what you think about the approach before I invest more time: Christoph Bartoschek --- xdmcp.c | 21 ++++++++++++++++++--- 1 files changed, 18 insertions(+), 3 deletions(-) diff --git a/xdmcp.c b/xdmcp.c index 7c91d1e..447b744 100644 --- a/xdmcp.c +++ b/xdmcp.c @@ -791,8 +791,8 @@ forward_respond ( { struct sockaddr_in6 in6_addr; - if (clientAddress.length != 16 || - clientPort.length != 2) + if ((clientAddress.length != 16 && clientAddress.length != 4) || + (clientPort.length != 2)) { goto badAddress; } @@ -801,7 +801,22 @@ forward_respond ( in6_addr.sin6_len = sizeof(in6_addr); # endif in6_addr.sin6_family = AF_INET6; - memmove(&in6_addr.sin6_addr,clientAddress.data,clientAddress.length); + if (clientAddress.length == 16) { + memmove(in6_addr.sin6_addr.s6_addr, clientAddress.data, 16); + } else { + /* If the client wants to forward the xdm server to an + ipv4 hosts it sends an ipv4 address in the forward + packet. On dual-stack hosts the packet arrives as a + ipv6 packet. To respond to the ipv4 host one has + to create an ipv4-mapped address. + + One example where this is necessary is an ipv4-only + thin client that connects to a dual-stacked xdm. + */ + in6_addr.sin6_addr.s6_addr[10] = 0xff; + in6_addr.sin6_addr.s6_addr[11] = 0xff; + memmove(in6_addr.sin6_addr.s6_addr + 12, clientAddress.data, 4); + } memmove((char *) &in6_addr.sin6_port, clientPort.data, 2); client = (struct sockaddr *) &in6_addr; clientlen = sizeof (in6_addr); -- 1.6.5.3 _______________________________________________ xorg-devel mailing list [email protected] http://lists.x.org/mailman/listinfo/xorg-devel
