--- /dev/null
+commit 39e1cee3184d275fa3ec4122de39b90d0d8e9bf4
+Author: root <root@rhel6.(none)>
+Date: Thu Apr 29 19:59:33 2010 -0400
+
+ linux-2.6-522-iptables-connection-tagging.patch
+
+diff --git a/include/linux/netfilter/xt_SETXID.h b/include/linux/netfilter/xt_SETXID.h
+new file mode 100644
+index 0000000..235b9d6
+--- /dev/null
++++ b/include/linux/netfilter/xt_SETXID.h
+@@ -0,0 +1,13 @@
++#ifndef _XT_SETXID_H_target
++#define _XT_SETXID_H_target
++
++enum {
++ XT_SET_PACKET_XID=0
++};
++
++struct xt_setxid_target_info_v2 {
++ unsigned long mark;
++ u_int8_t mode;
++};
++
++#endif /*_XT_SETXID_H_target*/
+diff --git a/include/net/netfilter/nf_conntrack.h b/include/net/netfilter/nf_conntrack.h
+index 5cf7270..95a5fde 100644
+--- a/include/net/netfilter/nf_conntrack.h
++++ b/include/net/netfilter/nf_conntrack.h
+@@ -119,6 +119,9 @@ struct nf_conn {
+ /* Storage reserved for other modules: */
+ union nf_conntrack_proto proto;
+
++ /* PLANETLAB. VNET-specific */
++ int xid[IP_CT_DIR_MAX];
++
+ /* Extensions */
+ struct nf_ct_ext *ext;
+ #ifdef CONFIG_NET_NS
+diff --git a/net/netfilter/Kconfig b/net/netfilter/Kconfig
+index 634d14a..a2872f5 100644
+--- a/net/netfilter/Kconfig
++++ b/net/netfilter/Kconfig
+@@ -543,6 +543,13 @@ config NETFILTER_XT_MATCH_CLUSTER
+ If you say Y or M here, try `iptables -m cluster --help` for
+ more information.
+
++config NETFILTER_XT_TARGET_SETXID
++ tristate '"SETXID" target support'
++ depends on NETFILTER_XTABLES
++ help
++ This option adds a `SETXID' target, which allows you to alter the
++ xid of a socket.
++
+ config NETFILTER_XT_MATCH_COMMENT
+ tristate '"comment" match support'
+ depends on NETFILTER_ADVANCED
+diff --git a/net/netfilter/Makefile b/net/netfilter/Makefile
+index 49f62ee..820655e 100644
+--- a/net/netfilter/Makefile
++++ b/net/netfilter/Makefile
+@@ -41,6 +41,7 @@ obj-$(CONFIG_NETFILTER_TPROXY) += nf_tproxy_core.o
+ obj-$(CONFIG_NETFILTER_XTABLES) += x_tables.o xt_tcpudp.o
+
+ # targets
++obj-$(CONFIG_NETFILTER_XT_TARGET_SETXID) += xt_SETXID.o
+ obj-$(CONFIG_NETFILTER_XT_TARGET_CLASSIFY) += xt_CLASSIFY.o
+ obj-$(CONFIG_NETFILTER_XT_TARGET_CONNMARK) += xt_CONNMARK.o
+ obj-$(CONFIG_NETFILTER_XT_TARGET_CONNSECMARK) += xt_CONNSECMARK.o
+diff --git a/net/netfilter/nf_conntrack_core.c b/net/netfilter/nf_conntrack_core.c
+index 1e1df20..144e131 100644
+--- a/net/netfilter/nf_conntrack_core.c
++++ b/net/netfilter/nf_conntrack_core.c
+@@ -673,6 +673,9 @@ init_conntrack(struct net *net,
+ hlist_nulls_add_head_rcu(&ct->tuplehash[IP_CT_DIR_ORIGINAL].hnnode,
+ &net->ct.unconfirmed);
+
++ ct->xid[IP_CT_DIR_ORIGINAL] = -1;
++ ct->xid[IP_CT_DIR_REPLY] = -1;
++
+ spin_unlock_bh(&nf_conntrack_lock);
+
+ if (exp) {
+diff --git a/net/netfilter/xt_MARK.c b/net/netfilter/xt_MARK.c
+index 225f8d1..7513997 100644
+--- a/net/netfilter/xt_MARK.c
++++ b/net/netfilter/xt_MARK.c
+@@ -13,7 +13,13 @@
+ #include <linux/module.h>
+ #include <linux/skbuff.h>
+ #include <linux/ip.h>
++#include <net/udp.h>
+ #include <net/checksum.h>
++#include <net/route.h>
++#include <net/inet_hashtables.h>
++#include <net/net_namespace.h>
++
++#include <net/netfilter/nf_conntrack.h>
+
+ #include <linux/netfilter/x_tables.h>
+ #include <linux/netfilter/xt_MARK.h>
+@@ -24,22 +30,267 @@ MODULE_DESCRIPTION("Xtables: packet mark modification");
+ MODULE_ALIAS("ipt_MARK");
+ MODULE_ALIAS("ip6t_MARK");
+
++DECLARE_PER_CPU(int, sknid_elevator);
++
++#define PEERCRED_SET(x) ((x!=0) && (x!=(unsigned int)-1))
++
++static inline u_int16_t get_dst_port(struct nf_conntrack_tuple *tuple)
++{
++ switch (tuple->dst.protonum) {
++ case IPPROTO_GRE:
++ /* XXX Truncate 32-bit GRE key to 16 bits */
++ return tuple->dst.u.gre.key;
++ case IPPROTO_ICMP:
++ /* Bind on ICMP echo ID */
++ return tuple->src.u.icmp.id;
++ case IPPROTO_TCP:
++ return tuple->dst.u.tcp.port;
++ case IPPROTO_UDP:
++ return tuple->dst.u.udp.port;
++ default:
++ return tuple->dst.u.all;
++ }
++}
++
++static inline u_int16_t get_src_port(struct nf_conntrack_tuple *tuple)
++{
++ switch (tuple->dst.protonum) {
++ case IPPROTO_GRE:
++ /* XXX Truncate 32-bit GRE key to 16 bits */
++ return htons(ntohl(tuple->src.u.gre.key));
++ case IPPROTO_ICMP:
++ /* Bind on ICMP echo ID */
++ return tuple->src.u.icmp.id;
++ case IPPROTO_TCP:
++ return tuple->src.u.tcp.port;
++ case IPPROTO_UDP:
++ return tuple->src.u.udp.port;
++ default:
++ return tuple->src.u.all;
++ }
++}
++
++static struct sock *__udp4_lib_lookup(struct net *net, __be32 saddr,
++ __be16 sport, __be32 daddr, __be16 dport,
++ int dif, struct hlist_head udptable[])
++{
++ struct sock *sk, *result = NULL;
++ struct hlist_node *node;
++ unsigned short hnum = ntohs(dport);
++ int badness = -1;
++
++ rcu_read_lock();
++ sk_for_each_rcu(sk, node, &udptable[udp_hashfn(net, hnum)]) {
++ struct inet_sock *inet = inet_sk(sk);
++
++ if (net_eq(sock_net(sk), net) && sk->sk_hash == hnum &&
++ !ipv6_only_sock(sk)) {
++ int score = (sk->sk_family == PF_INET ? 1 : 0);
++
++ if (inet->rcv_saddr) {
++ if (inet->rcv_saddr != daddr)
++ continue;
++ score += 2;
++ } else {
++ /* block non nx_info ips */
++ if (!v4_addr_in_nx_info(sk->sk_nx_info,
++ daddr, NXA_MASK_BIND))
++ continue;
++ }
++ if (inet->daddr) {
++ if (inet->daddr != saddr)
++ continue;
++ score += 2;
++ }
++ if (inet->dport) {
++ if (inet->dport != sport)
++ continue;
++ score += 2;
++ }
++ if (sk->sk_bound_dev_if) {
++ if (sk->sk_bound_dev_if != dif)
++ continue;
++ score += 2;
++ }
++ if (score == 9) {
++ result = sk;
++ break;
++ } else if (score > badness) {
++ result = sk;
++ badness = score;
++ }
++ }
++ }
++
++ if (result)
++ sock_hold(result);
++ rcu_read_unlock();
++ return result;
++}
++
++int onceonly = 1;
++
+ static unsigned int
+ mark_tg(struct sk_buff *skb, const struct xt_target_param *par)
+ {
+ const struct xt_mark_tginfo2 *info = par->targinfo;
++ long mark = -1;
++ enum ip_conntrack_info ctinfo;
++ struct sock *connection_sk;
++ int dif;
++ struct nf_conn *ct;
++ extern struct inet_hashinfo tcp_hashinfo;
++ enum ip_conntrack_dir dir;
++ int *curtag;
++ u_int32_t src_ip;
++ u_int32_t dst_ip;
++ u_int16_t proto, src_port;
++ u_int32_t ip;
++ u_int16_t port;
++
++ if (info->mark == ~0U) {
++ // As of 2.6.27.39, Dec 8 2009,
++ // NetNS + VNET = Trouble
++ // Let's handle this as a special case
++ struct net *net = dev_net(skb->dev);
++ if (!net_eq(net, &init_net)) {
++ WARN_ON(onceonly);
++ onceonly = 0;
++ return XT_CONTINUE;
++ }
++
++ /* copy-xid */
++ dif = ((struct rtable *)(skb->dst))->rt_iif;
++
++ ct = nf_ct_get(skb, &ctinfo);
++ if (!ct)
++ goto out_mark_finish;
++
++ dir = CTINFO2DIR(ctinfo);
++ src_ip = ct->tuplehash[dir].tuple.src.u3.ip;
++ dst_ip = ct->tuplehash[dir].tuple.dst.u3.ip;
++ src_port = get_src_port(&ct->tuplehash[dir].tuple);
++ proto = ct->tuplehash[dir].tuple.dst.protonum;
++
++ ip = ct->tuplehash[dir].tuple.dst.u3.ip;
++ port = get_dst_port(&ct->tuplehash[dir].tuple);
++
++ if (proto == 1) {
++ if (skb->mark > 0)
++ /* The packet is marked, it's going out */
++ ct->xid[0] = skb->mark;
++
++ if (ct->xid[0] > 0)
++ mark = ct->xid[0];
++ } else if (proto == 17) {
++ struct sock *sk;
++ if (!skb->mark) {
++ sk = __udp4_lib_lookup(net, src_ip, src_port,
++ ip, port, dif, udp_hash);
++
++ if (sk && hooknum == NF_INET_LOCAL_IN)
++ mark = sk->sk_nid;
++
++ if (sk)
++ sock_put(sk);
++ } else if (skb->mark > 0)
++ /* The packet is marked, it's going out */
++ ct->xid[0] = skb->mark;
++ } else if (proto == 6) { /* TCP */
++ int sockettype = 0; /* Established socket */
++
++ /* Looks for an established socket or a listening
++ socket corresponding to the 4-tuple, in that order.
++ The order is important for Codemux connections
++ to be handled properly */
++
++ connection_sk = inet_lookup_established(net,
++ &tcp_hashinfo,
++ src_ip,
++ src_port, ip,
++ port, dif);
++
++ if (!connection_sk) {
++ connection_sk = inet_lookup_listener(net,
++ &tcp_hashinfo,
++ ip, port,
++ dif);
++ sockettype = 1; /* Listening socket */
++ }
++
++ if (connection_sk) {
++ if (connection_sk->sk_state == TCP_TIME_WAIT) {
++ inet_twsk_put(inet_twsk(connection_sk));
++ goto out_mark_finish;
++ }
++
++ /* The peercred is not set. We set it if the other side has an xid. */
++ if (!PEERCRED_SET
++ (connection_sk->sk_peercred.uid)
++ && ct->xid[!dir] > 0 && (sockettype == 0)) {
++ connection_sk->sk_peercred.gid =
++ connection_sk->sk_peercred.uid =
++ ct->xid[!dir];
++ }
++
++ /* The peercred is set, and is not equal to the XID of 'the other side' */
++ else if (PEERCRED_SET
++ (connection_sk->sk_peercred.uid)
++ && (connection_sk->sk_peercred.uid !=
++ ct->xid[!dir])
++ && (sockettype == 0)) {
++ mark = connection_sk->sk_peercred.uid;
++ }
++
++ /* Has this connection already been tagged? */
++ if (ct->xid[dir] < 1) {
++ /* No - let's tag it */
++ ct->xid[dir] = connection_sk->sk_nid;
++ }
++
++ if (mark == -1 && (ct->xid[dir] != 0))
++ mark = ct->xid[dir];
++
++ sock_put(connection_sk);
++ }
++
++ /* All else failed. Is this a connection over raw sockets?
++ That explains why we couldn't get anything out of skb->sk,
++ or look up a "real" connection. */
++ if (ct->xid[dir] < 1) {
++ if (skb->skb_tag)
++ ct->xid[dir] = skb->skb_tag;
++ }
++
++ /* Covers CoDemux case */
++ if (mark < 1 && (ct->xid[dir] > 0))
++ mark = ct->xid[dir];
++
++ if (mark < 1 && (ct->xid[!dir] > 0))
++ mark = ct->xid[!dir];
++ goto out_mark_finish;
++ }
++ } else
++ mark = (skb->mark & ~info->mask) ^ info->mark;
++
++out_mark_finish:
++ if (mark != -1)
++ skb->mark = mark;
++
++ curtag = &__get_cpu_var(sknid_elevator);
++ if (mark > 0 && *curtag == -2 && hooknum == NF_INET_LOCAL_IN)
++ *curtag = mark;
+
+- skb->mark = (skb->mark & ~info->mask) ^ info->mark;
+ return XT_CONTINUE;
+ }
+
+ static struct xt_target mark_tg_reg __read_mostly = {
+- .name = "MARK",
+- .revision = 2,
+- .family = NFPROTO_UNSPEC,
+- .target = mark_tg,
+- .targetsize = sizeof(struct xt_mark_tginfo2),
+- .me = THIS_MODULE,
++ .name = "MARK",
++ .revision = 2,
++ .family = NFPROTO_UNSPEC,
++ .target = mark_tg,
++ .targetsize = sizeof(struct xt_mark_tginfo2),
++ .me = THIS_MODULE,
+ };
+
+ static int __init mark_tg_init(void)
+diff --git a/net/netfilter/xt_SETXID.c b/net/netfilter/xt_SETXID.c
+new file mode 100644
+index 0000000..f8553c5
+--- /dev/null
++++ b/net/netfilter/xt_SETXID.c
+@@ -0,0 +1,77 @@
++#include <linux/module.h>
++#include <linux/skbuff.h>
++#include <linux/ip.h>
++#include <net/checksum.h>
++#include <linux/vs_network.h>
++
++#include <linux/netfilter/x_tables.h>
++#include <linux/netfilter/xt_SETXID.h>
++
++MODULE_LICENSE("GPL");
++MODULE_AUTHOR("");
++MODULE_DESCRIPTION("");
++MODULE_ALIAS("ipt_SETXID");
++
++static unsigned int
++target_v2(struct sk_buff **pskb,
++ const struct net_device *in,
++ const struct net_device *out,
++ unsigned int hooknum,
++ const struct xt_target *target, const void *targinfo)
++{
++ const struct xt_setxid_target_info_v2 *setxidinfo = targinfo;
++
++ switch (setxidinfo->mode) {
++ case XT_SET_PACKET_XID:
++ (*pskb)->skb_tag = setxidinfo->mark;
++ break;
++ }
++ return XT_CONTINUE;
++}
++
++static int
++checkentry_v2(const char *tablename,
++ const void *entry,
++ const struct xt_target *target,
++ void *targinfo, unsigned int hook_mask)
++{
++ struct xt_setxid_target_info_v2 *setxidinfo = targinfo;
++
++ if (setxidinfo->mode != XT_SET_PACKET_XID) {
++ printk(KERN_WARNING "SETXID: unknown mode %u\n",
++ setxidinfo->mode);
++ return 0;
++ }
++
++ return 1;
++}
++
++static struct xt_target xt_setxid_target[] = {
++ {
++ .name = "SETXID",
++ .family = AF_INET,
++ .revision = 2,
++ .checkentry = checkentry_v2,
++ .target = target_v2,
++ .targetsize = sizeof(struct xt_setxid_target_info_v2),
++ .table = "mangle",
++ .me = THIS_MODULE,
++ }
++};
++
++static int __init init(void)
++{
++ int err;
++
++ err =
++ xt_register_target(xt_setxid_target);
++ return err;
++}
++
++static void __exit fini(void)
++{
++ xt_unregister_target(xt_setxid_target);
++}
++
++module_init(init);
++module_exit(fini);
git://git.onelab.eu / linux-2.6.git / commitdiff