Skip to contentSkip to conversation

net/ipv6/route.c: packets originating on device match lo

PATCH Superseded David McCullough 2 participants 3 messages Jun 25, 2012 – Jun 26, 2012

Subsystems (3)

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 net/ipv6/route.c: packets originating on device match lo

Hi all,

Seems like the special case of "iif lo" when using rules to control packets is broken for IPv6 traffic. The section from the ip man page is:

iif NAME
select the incoming device to match. If the interface is loop‐ back, the rule only matches packets originating from this host. This means that you may create separate routing tables for for‐ warded and local packets and, hence, completely segregate them.

If I run the following configuration:

    ifconfig eth0 192.168.0.1 netmask 255.255.255.0
    ip addr add fec0:0:0:2::1/64 dev eth0

    ip route add table 200 unreachable default
    ip -6 route add table 200 unreachable default

    ip rule add to 192.168.0.2 table 200 iif lo
    ip -6 rule add to fec0:0:0:2::2 table 200 iif lo

and then test with:

    # ping 192.168.0.2
    connect: Network is unreachable
    # ping6 fec0:0:0:2::2
    PING fec0:0:0:2::2(fec0:0:0:2::2) 56 data bytes
    64 bytes from fec0:0:0:2::2: icmp_seq=1 ttl=64 time=0.648 ms
    ...

The patch below is against linux-3.4. I am not sure if its the best solution but it appears to fix the case above:

    # ping6 fec0:0:0:2::2
    connect: Network is unreachable

and does not appear to affect forwarded packets is the few cases I have tried. Tested on linux-3.4 and iproute2-3.4.0, Feedback appreciated :-)

Thanks,
Davidm

net/ipv6/route.c +3 -0
diff -p -u -r1.1.1.59 route.c
@@ -931,6 +931,9 @@ struct dst_entry * ip6_route_output(stru
 {
 	int flags = 0;
 
+	if (fl6->flowi6_iif == 0)
+		fl6->flowi6_iif = net->loopback_dev->ifindex;
+		
 	if ((sk && sk->sk_bound_dev_if) || rt6_need_strict(&fl6->daddr))
 		flags |= RT6_LOOKUP_F_IFACE;
 
Signature · 2 lines
David McCullough,      [email protected],  Ph:+61 734352815
McAfee - SnapGear      http://www.mcafee.com         http://www.uCdot.org
maintainer
#1 lore raw

From: David McCullough <[email protected]>
Date: Mon, 25 Jun 2012 16:50:30 +1000

net/ipv6/route.c struct dst_entry * ip6_route_output(stru · line 936

Quoted 7 lines
net/ipv6/route.c +3 -0
@@ -931,6 +931,9 @@ struct dst_entry * ip6_route_output(stru
 {
 	int flags = 0;
 
+	if (fl6->flowi6_iif == 0)
+		fl6->flowi6_iif = net->loopback_dev->ifindex;
+		

Like ipv4, you should make this assignment unconditionally.

Thanks.

#2 lore raw

Jivin David Miller lays it down ...

Quoted 7 lines

From: David McCullough <[email protected]>
Date: Mon, 25 Jun 2012 16:50:30 +1000

Quoted 4 lines
net/ipv6/route.c +0 -0
@@ -931,6 +931,9 @@ struct dst_entry * ip6_route_output(stru
 {
 	int flags = 0;
 
Quoted 5 lines

net/ipv6/route.c struct dst_entry * ip6_route_output(stru · line 936

Quoted 3 lines
net/ipv6/route.c +3 -0
+	if (fl6->flowi6_iif == 0)
+		fl6->flowi6_iif = net->loopback_dev->ifindex;
+		

Like ipv4, you should make this assignment unconditionally.

No problems, will resend, I wasn't so sure that I could do that safely with the flowi getting passed into IPv6 rather than created as it is in IPv4. Did some extra testing, looks good so patch to follow,

Thanks,
Davidm

Signature · 2 lines
David McCullough,      [email protected],  Ph:+61 734352815
McAfee - SnapGear      http://www.mcafee.com         http://www.uCdot.org