Skip to contentSkip to conversation

netfilter netbios-ns on ethernet bridge firewall

DISCUSSION Jaco Kroon 2 participants 2 messages Nov 12, 2005

lore
Share & tools
lore

Jaco Kroon

author
#0 lore raw

Hello all,

I'm trying to make the ip_conntrack_netbios_ns module work on my ethernet bridge firewall. The module works extremely nicely on a single host when packets are locally generated, but doesn't even look at the packets if they are generated elsewhere and merely pass through our box.

I'm currently running a vanilla kernel (2.6.14) with some of my own printk statements and some minor modifications (diff below).

My configuration is essentially a device called br0 with eth0 and eth1 as it's two physical interfaces. I've got an IP of 192.168.0.4 assigned to br0 (and an alias of 192.168.2.2), neither eth0 nor eth1 has an IP.

With the patch below applied when I send out a netbios-ns packet from 192.168.0.3 (coming in off eth0) I get the following in my logs:

Nov 12 09:38:17 xacatecas netbios packet seen
Nov 12 09:38:17 xacatecas sk is null
Nov 12 09:38:17 xacatecas rtable dst: 00000000
Nov 12 09:38:17 xacatecas rtable src: 00000000
Nov 12 09:38:17 xacatecas rtable iif: 00000000
Nov 12 09:38:17 xacatecas netbios packet seen
Nov 12 09:38:17 xacatecas sk is null
Nov 12 09:38:17 xacatecas rtable dst: ff00a8c0
Nov 12 09:38:17 xacatecas rtable src: 0300a8c0
Nov 12 09:38:17 xacatecas rtable iif: 00000004
Nov 12 09:38:17 xacatecas is a broadcast
Nov 12 09:38:17 xacatecas packet accepted!
Nov 12 09:38:17 xacatecas checking ...
Nov 12 09:38:17 xacatecas (lo) 00000000 vs ff00a8c0
Nov 12 09:38:17 xacatecas You almost certainly have a misconfigured

broadcast address.

Quoted 1 line

From what I understand the packet passes through the conntrack module

twice, once on PREROUTING and once on POSTROUTING, which is why the first instance has rtable all zeroed. The second time round I expect to see br0s details in the rtable, not lo though! The routing table looks as follows:

Kernel IP routing table
Destination Gateway Genmask Flags MSS Window irtt
Iface

192.168.2.0     0.0.0.0         255.255.255.0   U         0 0          0 br0
192.168.0.0     0.0.0.0         255.255.255.0   U         0 0          0 br0
127.0.0.0       127.0.0.1       255.0.0.0       UG        0 0          0 lo
0.0.0.0         192.168.0.1     0.0.0.0         UG        0 0          0 br0

In essense I've got two choices now:

1. Hardcode the mask as in the commented out mask = 0x00FFFFFF below, which will only work for myself. (Or pass it as a parameter or something - what happens in cases where the netmasks differ for multiple interfaces?)

2. Cycle through the devices as provided by rtable, but also cycle through _all_ inet devices. This would be more generic, but somehow I don't think this is the correct solution, and it would eat a lot more CPU.

Thanks for any and all help/advice.

Jaco

--- net/ipv4/netfilter/ip_conntrack_netbios_ns.c.orig 2005-11-12
09:39:35.000000000 +0200
+++ net/ipv4/netfilter/ip_conntrack_netbios_ns.c 2005-11-12
09:33:18.000000000 +0200
@@ -50,32 +50,52 @@

        u_int32_t mask = 0;

        /* we're only interested in locally generated packets */
+       printk("netbios packet seen\n");
        if ((*pskb)->sk == NULL)

- goto out;
+ printk("sk is null\n");
+ if (rt == NULL)
+ printk("rt == NULL!\n");
+ else {

+               printk("rtable dst: %08x\n", rt->rt_dst);
+               printk("rtable src: %08x\n", rt->rt_src);
+               printk("rtable iif: %08x\n", rt->rt_iif);
+       }
        if (rt == NULL || !(rt->rt_flags & RTCF_BROADCAST))
                goto out;
+       printk("is a broadcast\n");
        if (CTINFO2DIR(ctinfo) != IP_CT_DIR_ORIGINAL)
                goto out;
-
+       printk("packet accepted!\n");
+
+//     mask = 0x00FFFFFF;
        rcu_read_lock();
        in_dev = __in_dev_get_rcu(rt->u.dst.dev);
        if (in_dev != NULL) {
+               printk("checking ...\n");
                for_primary_ifa(in_dev) {
+                       printk("(%s) %08x vs %08x\n", ifa->ifa_label,
ifa->ifa_broadcast, iph->daddr);
                        if (ifa->ifa_broadcast == iph->daddr) {
                                mask = ifa->ifa_mask;
-                               break;
+                               printk(":)\n");
+//                             break;
                        }
                } endfor_ifa(in_dev);
        }
        rcu_read_unlock();

- if (mask == 0)
+ if (mask == 0) {
+ printk(KERN_INFO "You almost certainly have a

misconfigured broadcast address.\n");
                goto out;

+ }

+ printk("Yes, we are delivering to a local broadcast addr!\n");
+

        exp = ip_conntrack_expect_alloc(ct);
        if (exp == NULL)
                goto out;

+ printk("We have managed to allocate an expect structure,
proceeding to add expect.\n");
+

        exp->tuple                = ct->tuplehash[IP_CT_DIR_REPLY].tuple;
        exp->tuple.src.u.udp.port = ntohs(NMBD_PORT);
Signature · 3 lines
There are only 10 kinds of people in this world,
  those that understand binary and those that don't.
http://www.kroon.co.za/
#1 lore raw
Quoted 1 line · Jaco Kroon

I'm trying to make the ip_conntrack_netbios_ns module work on my

JFYI: I've passed this on to [email protected], that's where questions/comments/patches regarding netfilter code should go.

I'm confident the original author of the netbios helper will contact you shortly.

Signature · 5 lines
- Harald Welte <[email protected]>                 http://netfilter.org/
============================================================================
  "Fragmentation is like classful addressing -- an interesting early
   architectural error that shows how much experimentation was going
   on while IP was being designed."                    -- Paul Vixie