Hello, I'm implementing an IP header compression protocol as a loadable kernel module that takes an IPv4 header and compresses it to smaller header (let's call it "iphc" header). To do this, for outgoing packets, i've used the NF_IP_POST_ROUTING netfilter hook to grab the packet after routing decisions are done. Then, i'd like to replace the IP header with an IPHC header (compressed IP), change the ethertype to something else, send it down the stack (to Ethernet) and out. From what i'm seeing, when i intercept the packet at the NF_IP_POST_ROUTING hook, the MAC headers have not been added yet. A few questions i had: 1. Given that i'm replacing the IP header, will that affect Ethernet's ability to find the MAC hardware address? I saw that in the ip_finish_output2() function, they take the dst value in skbuff (using skb_dst() function) to search the neighbor table for an ARP entry. If i'm only swapping out the IP header with an IPHC header and changing the protocol, will the ARP lookup fail? I don't think i've touched the "dst" field in skbuff. 2. Is it enough for me to just replace the IP header with an IPHC header, change the Ethertype, and send it back down the stack? Or should i transmit it directly myself without it being returned to ip_finish_output2 For some reason, if i do #2, my code crashes. Below is a sample of what i'm trying to do after i hook into the NF_IP_POST_ROUTING netfilter hook (skb is passed in): struct iphc *my_hdr; /* stripping the IPv4 header from skbuff */ skb_pull(skb, sizeof (struct iphdr)); /* adding my header skb */ skb_push(skb, IPHC_HDR_SIZE); /* reset network header */ skb_reset_network_header(skb); my_hdr = (struct iphc *)skb; my_hdr->field1 = 1; my_hdr->field2 = 2; /* change ethertype */ skb->protocol = __constant_htons(ETH_P_IPHC); return NF_ACCEPT; Am i missing something here? Thanks! -- To unsubscribe from this list: send the line "unsubscribe lartc" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
