Skip to contentSkip to conversation

netfilter: payload_len is be16, add size of struct rather than size of pointer

RFC-PATCH Applied Harvey Harrison 3 participants 3 messages Nov 7, 2008 – Nov 11, 2008

Subsystems (4)
lore
Share & tools
lore
#0 lore raw

payload_len is a be16 value, not cpu_endian, also the size of a ponter to a struct ipv6hdr was being added, not the size of the struct itself.

---
I'm quite supicious of the following code in net/netfilter/ipvs/ip_vs_xmit.c

Line 714:

	iph->payload_len	=	old_iph->payload_len + sizeof(old_iph);

I believe that the payload_len is a big-endian value and this is treating it as cpu-ordered. In addition, it is adding the size of a pointer to a struct ipv6hdr and not the size of the struct itself.

If I'm correct, I'd suggest the following is what _may_ have been intended.

 net/netfilter/ipvs/ip_vs_xmit.c |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)
net/netfilter/ipvs/ip_vs_xmit.c +2 -1
index 2f36721..425ab14 100644
@@ -711,7 +711,8 @@ ip_vs_tunnel_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
 	iph			=	ipv6_hdr(skb);
 	iph->version		=	6;
 	iph->nexthdr		=	IPPROTO_IPV6;
-	iph->payload_len	=	old_iph->payload_len + sizeof(old_iph);
+	iph->payload_len	=	old_iph->payload_len;
+	be16_add_cpu(&iph->payload_len, sizeof(*old_iph));
 	iph->priority		=	old_iph->priority;
 	memset(&iph->flow_lbl, 0, sizeof(iph->flow_lbl));
 	iph->daddr		=	rt->rt6i_dst.addr;
Signature · 1 line
1.6.0.3.756.gb776d
#1 lore raw
Quoted 34 lines · Harvey Harrison

payload_len is a be16 value, not cpu_endian, also the size of a ponter to a struct ipv6hdr was being added, not the size of the struct itself.

---
I'm quite supicious of the following code in net/netfilter/ipvs/ip_vs_xmit.c

Line 714:

       iph->payload_len        =       old_iph->payload_len + sizeof(old_iph);

I believe that the payload_len is a big-endian value and this is treating it as cpu-ordered. In addition, it is adding the size of a pointer to a struct ipv6hdr and not the size of the struct itself.

If I'm correct, I'd suggest the following is what _may_ have been intended.

 net/netfilter/ipvs/ip_vs_xmit.c |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)
net/netfilter/ipvs/ip_vs_xmit.c +2 -2
index 2f36721..425ab14 100644
@@ -711,7 +711,8 @@ ip_vs_tunnel_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
       iph                     =       ipv6_hdr(skb);
       iph->version            =       6;
       iph->nexthdr            =       IPPROTO_IPV6;
-       iph->payload_len        =       old_iph->payload_len + sizeof(old_iph);
+       iph->payload_len        =       old_iph->payload_len;
+       be16_add_cpu(&iph->payload_len, sizeof(*old_iph));
       iph->priority           =       old_iph->priority;
       memset(&iph->flow_lbl, 0, sizeof(iph->flow_lbl));
       iph->daddr              =       rt->rt6i_dst.addr;
--
Quoted 1 line

1.6.0.3.756.gb776d

Yes, both of that makes sense, thank you!

This function was never tested because IIRC there was/is no decapsulation support for IPv6 in IPv6 tunneling on the receiving side (please correct me if I'm wrong). I sent some packets through it which somehow arrived at the real server, but that was about it. So the whole code in that function is pretty only an idea of how the v6 tunneling should work ;)

Julius