Skip to contentSkip to conversation

Problem sending skb built from scratch with IPv6

DISCUSSION Pierre Rondou 2 participants 3 messages Mar 22, 2011 – Apr 18, 2011 netfilter

lore
Share & tools
lore
#0 lore raw

Hi all,

I'm a student at the University of Liege (Belgium) and for my master thesis, I have to devellop a netfilter module implementing NatIvI and Nat64.

Basically, those modules ask to "translate" an Ipv4 packet to an IPv6 one (and in the other side aswell), you can find more detailled information on the IETF website.

For my modules, I use xtables-addons.

I have been able to do every step correctly, but one:
Sending the newly created IPv6 packet to the network.

The modules, as stated by the protocols, need to be able to send packet with IP that doesn't belong to them (i.e, sort of spoofing). Everything works fine for the newly created IPv4 skb (I can see the packets in wireshark), you can see the source code at the end.

But for Ipv6, the only think I see is neighbour solicitation messages (ICMPv6) for both the source IP and the dest IP. There is one only case when the program works: when the source IP of the packet is the same as the interface's one.

Even more curious, the two function I call for my IPv6 sending return 0 (ip6_route_output and ip6_local_out), meaning that they were successfull.

So, my question is simple:
"How can I send an IPv6 paquet, built from scratch (new skb) to the network?"

Thanks in advance for your help,

Pierre Rondou

Finaly, here are some info you might find usefull:

The source code used for the new IPv4 packets (working 100%), it's copied from the NAT64 module available in GPL:

   newskb->protocol = htons(ETH_P_IP);

   newip = ip_hdr(newskb);

   memset(&fl, 0, sizeof(fl));
   fl.fl4_dst = newip->daddr;
   fl.fl4_tos = RT_TOS(newip->tos);
   fl.proto = newskb->protocol;
   if (ip_route_output_key(&init_net, &rt, &fl))
   {
      printk("error: ip_route_output_key failed\n");
      return NF_DROP;
   }

   if (!rt)
   {
      printk("error: rt null\n");
      return NF_DROP;
   }


   newskb->dev = rt->u.dst.dev;
   skb_dst_set(newskb, (struct dst_entry *)rt);
   if(ip_local_out(newskb)) {
          printk("error: ip_local_out failed\n");
          return NF_DROP;
   }

Here the IPv6 code, NOT WORKING:

   newskb->protocol = htons(ETH_P_IPV6);

   newip = ipv6_hdr(newskb);

   memset(&fl, 0, sizeof(fl));
   fl.fl6_src = newip->saddr;
   fl.fl6_dst = newip->daddr;
   fl.fl6_flowlabel = 0;
   fl.proto = newskb->protocol;
   dst = ip6_route_output(&init_net, NULL, &fl);
   if (!dst)
   {
      printk("error: ip_route_output_key failed\n");
      return NF_DROP;
   }

   skb_dst_set(newskb, dst);
   newskb->dev = dst->dev;
   skb_dst_set(newskb, dst);
   if(ip6_local_out(newskb)) {
      printk("error: ip_local_out failed\n");
      return NF_DROP;
   }

   pr_info("Packet sent \n");

return NF_DROP;

my network configuration:

# The primary network interface
allow-hotplug eth1
iface eth1 inet static

        address 192.168.1.40
        netmask 255.255.255.0
        gateway 192.168.1.1

#IPV6 static configuration
iface eth1 inet6 static

        address 2001:6a8:2d80:128::0001
        netmask 64
        gateway 2001:6a8:2d80:128::0002

# END IPV6 configuration

route result:

# route --inet
Table de routage IP du noyau
Destination     Passerelle      Genmask         Indic Metric Ref    Use Iface
192.168.1.0     *               255.255.255.0   U     0      0        0 eth1
default         192.168.1.1     0.0.0.0         UG    0      0        0 eth1
# route --inet6
Table de routage IPv6 du noyau
Destination                    Next Hop                   Flag Met Ref Use If
2001:6a8:2d80:128::/64         ::                         U    256 0     1 eth1
fe80::/64                      ::                         U    256 0     0 eth1
::/0                           2001:6a8:2d80:128::2       UG   1   0    35 eth1
::/0                           ::                         !n   -1  1    61 lo
::1/128                        ::                         Un   0   1    80 lo
2001:6a8:2d80:128::/128        ::                         Un   0   1     0 lo
2001:6a8:2d80:128::1/128       ::                         Un   0   1    20 lo
fe80::/128                     ::                         Un   0   1     0 lo
fe80::20e:a6ff:feb0:e1a2/128   ::                         Un   0   1    14 lo
ff00::/8                       ::                         U    256 0     0 eth1
::/0                           ::                         !n   -1  1    61 lo
#1 lore raw
Quoted 5 lines · Pierre Rondou

Hi all,

I'm a student at the University of Liege (Belgium) and for my master thesis, I have to devellop a netfilter module implementing NatIvI and Nat64.

Let me just make aware of preexisting software, such as Ecdysis nat64, because too often have I encountered students who just reimplemented everything and then felt like it was for nothing.

Quoted 9 lines

The modules, as stated by the protocols, need to be able to send packet with IP that doesn't belong to them (i.e, sort of spoofing). Everything works fine for the newly created IPv4 skb (I can see the packets in wireshark), you can see the source code at the end.

But for Ipv6, the only think I see is neighbour solicitation messages (ICMPv6) for both the source IP and the dest IP. There is one only case when the program works: when the source IP of the packet is the same as the interface's one.

Of course a connected router will do NDISC if the address is directly reachable according to its routing table.

Quoted 2 lines

The source code used for the new IPv4 packets (working 100%), it's copied from the NAT64 module available in GPL:

"The NAT64 module", but which?

Quoted 3 lines
  skb_dst_set(newskb, dst);
  newskb->dev = dst->dev;
  skb_dst_set(newskb, dst);

Why set dst twice?

Quoted 5 lines
# route --inet
Table de routage IP du noyau
Destination     Passerelle      Genmask         Indic Metric Ref    Use Iface
192.168.1.0     *               255.255.255.0   U     0      0        0 eth1
default         192.168.1.1     0.0.0.0         UG    0      0        0 eth1

For future reference, stick to iproute2.

Quoted 14 lines
# route --inet6
Table de routage IPv6 du noyau
Destination                    Next Hop                   Flag Met Ref Use If
2001:6a8:2d80:128::/64         ::                         U    256 0     1 eth1
fe80::/64                      ::                         U    256 0     0 eth1
::/0                           2001:6a8:2d80:128::2       UG   1   0    35 eth1
::/0                           ::                         !n   -1  1    61 lo
::1/128                        ::                         Un   0   1    80 lo
2001:6a8:2d80:128::/128        ::                         Un   0   1     0 lo
2001:6a8:2d80:128::1/128       ::                         Un   0   1    20 lo
fe80::/128                     ::                         Un   0   1     0 lo
fe80::20e:a6ff:feb0:e1a2/128   ::                         Un   0   1    14 lo
ff00::/8                       ::                         U    256 0     0 eth1
::/0                           ::                         !n   -1  1    61 lo
#2 lore raw

Long time since my previous mail, I've been looking on other issues.

The problem solved itself by upgrading to debian squeeze (I previously was on a testing version between Lenny and squeeze).

Thanks anyway :-)

Quoted 42 lines · Jan Engelhardt
Quoted 6 lines · Pierre Rondou

Hi all,

I'm a student at the University of Liege (Belgium) and for my master thesis, I have to devellop a netfilter module implementing NatIvI and Nat64.

Let me just make aware of preexisting software, such as Ecdysis nat64, because too often have I encountered students who just reimplemented everything and then felt like it was for nothing.

Quoted 10 lines

The modules, as stated by the protocols, need to be able to send packet with IP that doesn't belong to them (i.e, sort of spoofing). Everything works fine for the newly created IPv4 skb (I can see the packets in wireshark), you can see the source code at the end.

But for Ipv6, the only think I see is neighbour solicitation messages (ICMPv6) for both the source IP and the dest IP. There is one only case when the program works: when the source IP of the packet is the same as the interface's one.

Of course a connected router will do NDISC if the address is directly reachable according to its routing table.

Quoted 3 lines

The source code used for the new IPv4 packets (working 100%), it's copied from the NAT64 module available in GPL:

"The NAT64 module", but which?

Quoted 4 lines
   skb_dst_set(newskb, dst);
   newskb->dev = dst->dev;
   skb_dst_set(newskb, dst);

Why set dst twice?

Quoted 24 lines
Quoted 6 lines
# route --inet
Table de routage IP du noyau
Destination     Passerelle      Genmask         Indic Metric Ref    Use Iface
192.168.1.0     *               255.255.255.0   U     0      0        0 eth1
default         192.168.1.1     0.0.0.0         UG    0      0        0 eth1

For future reference, stick to iproute2.

Quoted 15 lines
# route --inet6
Table de routage IPv6 du noyau
Destination                    Next Hop                   Flag Met Ref Use If
2001:6a8:2d80:128::/64         ::                         U    256 0     1 eth1
fe80::/64                      ::                         U    256 0     0 eth1
::/0                           2001:6a8:2d80:128::2       UG   1   0    35 eth1
::/0                           ::                         !n   -1  1    61 lo
::1/128                        ::                         Un   0   1    80 lo
2001:6a8:2d80:128::/128        ::                         Un   0   1     0 lo
2001:6a8:2d80:128::1/128       ::                         Un   0   1    20 lo
fe80::/128                     ::                         Un   0   1     0 lo
fe80::20e:a6ff:feb0:e1a2/128   ::                         Un   0   1    14 lo
ff00::/8                       ::                         U    256 0     0 eth1
::/0                           ::                         !n   -1  1    61 lo

--
To unsubscribe from this list: send the line "unsubscribe netfilter-devel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html