netfilter: ip_tables: remove clusterip target
authorFlorian Westphal <fw@strlen.de>
Thu, 5 Jan 2023 19:22:02 +0000 (20:22 +0100)
committerFlorian Westphal <fw@strlen.de>
Wed, 18 Jan 2023 12:05:24 +0000 (13:05 +0100)
Marked as 'to be removed soon' since kernel 4.1 (2015).
Functionality was superseded by the 'cluster' match, added in kernel
2.6.30 (2009).

clusterip_tg_check still has races that can give

 proc_dir_entry 'ipt_CLUSTERIP/10.1.1.2' already registered

followed by a WARN splat.

Remove it instead of trying to fix this up again.
clusterip uapi header is left as-is for now.

Signed-off-by: Florian Westphal <fw@strlen.de>
net/ipv4/netfilter/Kconfig
net/ipv4/netfilter/Makefile
net/ipv4/netfilter/ipt_CLUSTERIP.c [deleted file]

index aab3841..f71a7e9 100644 (file)
@@ -259,20 +259,6 @@ config IP_NF_MANGLE
 
          To compile it as a module, choose M here.  If unsure, say N.
 
-config IP_NF_TARGET_CLUSTERIP
-       tristate "CLUSTERIP target support"
-       depends on IP_NF_MANGLE
-       depends on NF_CONNTRACK
-       depends on NETFILTER_ADVANCED
-       select NF_CONNTRACK_MARK
-       select NETFILTER_FAMILY_ARP
-       help
-         The CLUSTERIP target allows you to build load-balancing clusters of
-         network servers without having a dedicated load-balancing
-         router/server/switch.
-
-         To compile it as a module, choose M here.  If unsure, say N.
-
 config IP_NF_TARGET_ECN
        tristate "ECN target support"
        depends on IP_NF_MANGLE
index 93bad11..5a26f9d 100644 (file)
@@ -39,7 +39,6 @@ obj-$(CONFIG_IP_NF_MATCH_AH) += ipt_ah.o
 obj-$(CONFIG_IP_NF_MATCH_RPFILTER) += ipt_rpfilter.o
 
 # targets
-obj-$(CONFIG_IP_NF_TARGET_CLUSTERIP) += ipt_CLUSTERIP.o
 obj-$(CONFIG_IP_NF_TARGET_ECN) += ipt_ECN.o
 obj-$(CONFIG_IP_NF_TARGET_REJECT) += ipt_REJECT.o
 obj-$(CONFIG_IP_NF_TARGET_SYNPROXY) += ipt_SYNPROXY.o
diff --git a/net/ipv4/netfilter/ipt_CLUSTERIP.c b/net/ipv4/netfilter/ipt_CLUSTERIP.c
deleted file mode 100644 (file)
index b3cc416..0000000
+++ /dev/null
@@ -1,929 +0,0 @@
-// SPDX-License-Identifier: GPL-2.0-only
-/* Cluster IP hashmark target
- * (C) 2003-2004 by Harald Welte <laforge@netfilter.org>
- * based on ideas of Fabio Olive Leite <olive@unixforge.org>
- *
- * Development of this code funded by SuSE Linux AG, https://www.suse.com/
- */
-#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
-#include <linux/module.h>
-#include <linux/proc_fs.h>
-#include <linux/jhash.h>
-#include <linux/bitops.h>
-#include <linux/skbuff.h>
-#include <linux/slab.h>
-#include <linux/ip.h>
-#include <linux/tcp.h>
-#include <linux/udp.h>
-#include <linux/icmp.h>
-#include <linux/if_arp.h>
-#include <linux/seq_file.h>
-#include <linux/refcount.h>
-#include <linux/netfilter_arp.h>
-#include <linux/netfilter/x_tables.h>
-#include <linux/netfilter_ipv4/ip_tables.h>
-#include <linux/netfilter_ipv4/ipt_CLUSTERIP.h>
-#include <net/netfilter/nf_conntrack.h>
-#include <net/net_namespace.h>
-#include <net/netns/generic.h>
-#include <net/checksum.h>
-#include <net/ip.h>
-
-#define CLUSTERIP_VERSION "0.8"
-
-MODULE_LICENSE("GPL");
-MODULE_AUTHOR("Harald Welte <laforge@netfilter.org>");
-MODULE_DESCRIPTION("Xtables: CLUSTERIP target");
-
-struct clusterip_config {
-       struct list_head list;                  /* list of all configs */
-       refcount_t refcount;                    /* reference count */
-       refcount_t entries;                     /* number of entries/rules
-                                                * referencing us */
-
-       __be32 clusterip;                       /* the IP address */
-       u_int8_t clustermac[ETH_ALEN];          /* the MAC address */
-       int ifindex;                            /* device ifindex */
-       u_int16_t num_total_nodes;              /* total number of nodes */
-       unsigned long local_nodes;              /* node number array */
-
-#ifdef CONFIG_PROC_FS
-       struct proc_dir_entry *pde;             /* proc dir entry */
-#endif
-       enum clusterip_hashmode hash_mode;      /* which hashing mode */
-       u_int32_t hash_initval;                 /* hash initialization */
-       struct rcu_head rcu;                    /* for call_rcu */
-       struct net *net;                        /* netns for pernet list */
-       char ifname[IFNAMSIZ];                  /* device ifname */
-};
-
-#ifdef CONFIG_PROC_FS
-static const struct proc_ops clusterip_proc_ops;
-#endif
-
-struct clusterip_net {
-       struct list_head configs;
-       /* lock protects the configs list */
-       spinlock_t lock;
-
-       bool clusterip_deprecated_warning;
-#ifdef CONFIG_PROC_FS
-       struct proc_dir_entry *procdir;
-       /* mutex protects the config->pde*/
-       struct mutex mutex;
-#endif
-       unsigned int hook_users;
-};
-
-static unsigned int clusterip_arp_mangle(void *priv, struct sk_buff *skb, const struct nf_hook_state *state);
-
-static const struct nf_hook_ops cip_arp_ops = {
-       .hook = clusterip_arp_mangle,
-       .pf = NFPROTO_ARP,
-       .hooknum = NF_ARP_OUT,
-       .priority = -1
-};
-
-static unsigned int clusterip_net_id __read_mostly;
-static inline struct clusterip_net *clusterip_pernet(struct net *net)
-{
-       return net_generic(net, clusterip_net_id);
-}
-
-static inline void
-clusterip_config_get(struct clusterip_config *c)
-{
-       refcount_inc(&c->refcount);
-}
-
-static void clusterip_config_rcu_free(struct rcu_head *head)
-{
-       struct clusterip_config *config;
-       struct net_device *dev;
-
-       config = container_of(head, struct clusterip_config, rcu);
-       dev = dev_get_by_name(config->net, config->ifname);
-       if (dev) {
-               dev_mc_del(dev, config->clustermac);
-               dev_put(dev);
-       }
-       kfree(config);
-}
-
-static inline void
-clusterip_config_put(struct clusterip_config *c)
-{
-       if (refcount_dec_and_test(&c->refcount))
-               call_rcu(&c->rcu, clusterip_config_rcu_free);
-}
-
-/* decrease the count of entries using/referencing this config.  If last
- * entry(rule) is removed, remove the config from lists, but don't free it
- * yet, since proc-files could still be holding references */
-static inline void
-clusterip_config_entry_put(struct clusterip_config *c)
-{
-       struct clusterip_net *cn = clusterip_pernet(c->net);
-
-       local_bh_disable();
-       if (refcount_dec_and_lock(&c->entries, &cn->lock)) {
-               list_del_rcu(&c->list);
-               spin_unlock(&cn->lock);
-               local_bh_enable();
-               /* In case anyone still accesses the file, the open/close
-                * functions are also incrementing the refcount on their own,
-                * so it's safe to remove the entry even if it's in use. */
-#ifdef CONFIG_PROC_FS
-               mutex_lock(&cn->mutex);
-               if (cn->procdir)
-                       proc_remove(c->pde);
-               mutex_unlock(&cn->mutex);
-#endif
-               return;
-       }
-       local_bh_enable();
-}
-
-static struct clusterip_config *
-__clusterip_config_find(struct net *net, __be32 clusterip)
-{
-       struct clusterip_config *c;
-       struct clusterip_net *cn = clusterip_pernet(net);
-
-       list_for_each_entry_rcu(c, &cn->configs, list) {
-               if (c->clusterip == clusterip)
-                       return c;
-       }
-
-       return NULL;
-}
-
-static inline struct clusterip_config *
-clusterip_config_find_get(struct net *net, __be32 clusterip, int entry)
-{
-       struct clusterip_config *c;
-
-       rcu_read_lock_bh();
-       c = __clusterip_config_find(net, clusterip);
-       if (c) {
-#ifdef CONFIG_PROC_FS
-               if (!c->pde)
-                       c = NULL;
-               else
-#endif
-               if (unlikely(!refcount_inc_not_zero(&c->refcount)))
-                       c = NULL;
-               else if (entry) {
-                       if (unlikely(!refcount_inc_not_zero(&c->entries))) {
-                               clusterip_config_put(c);
-                               c = NULL;
-                       }
-               }
-       }
-       rcu_read_unlock_bh();
-
-       return c;
-}
-
-static void
-clusterip_config_init_nodelist(struct clusterip_config *c,
-                              const struct ipt_clusterip_tgt_info *i)
-{
-       int n;
-
-       for (n = 0; n < i->num_local_nodes; n++)
-               set_bit(i->local_nodes[n] - 1, &c->local_nodes);
-}
-
-static int
-clusterip_netdev_event(struct notifier_block *this, unsigned long event,
-                      void *ptr)
-{
-       struct net_device *dev = netdev_notifier_info_to_dev(ptr);
-       struct net *net = dev_net(dev);
-       struct clusterip_net *cn = clusterip_pernet(net);
-       struct clusterip_config *c;
-
-       spin_lock_bh(&cn->lock);
-       list_for_each_entry_rcu(c, &cn->configs, list) {
-               switch (event) {
-               case NETDEV_REGISTER:
-                       if (!strcmp(dev->name, c->ifname)) {
-                               c->ifindex = dev->ifindex;
-                               dev_mc_add(dev, c->clustermac);
-                       }
-                       break;
-               case NETDEV_UNREGISTER:
-                       if (dev->ifindex == c->ifindex) {
-                               dev_mc_del(dev, c->clustermac);
-                               c->ifindex = -1;
-                       }
-                       break;
-               case NETDEV_CHANGENAME:
-                       if (!strcmp(dev->name, c->ifname)) {
-                               c->ifindex = dev->ifindex;
-                               dev_mc_add(dev, c->clustermac);
-                       } else if (dev->ifindex == c->ifindex) {
-                               dev_mc_del(dev, c->clustermac);
-                               c->ifindex = -1;
-                       }
-                       break;
-               }
-       }
-       spin_unlock_bh(&cn->lock);
-
-       return NOTIFY_DONE;
-}
-
-static struct clusterip_config *
-clusterip_config_init(struct net *net, const struct ipt_clusterip_tgt_info *i,
-                     __be32 ip, const char *iniface)
-{
-       struct clusterip_net *cn = clusterip_pernet(net);
-       struct clusterip_config *c;
-       struct net_device *dev;
-       int err;
-
-       if (iniface[0] == '\0') {
-               pr_info("Please specify an interface name\n");
-               return ERR_PTR(-EINVAL);
-       }
-
-       c = kzalloc(sizeof(*c), GFP_ATOMIC);
-       if (!c)
-               return ERR_PTR(-ENOMEM);
-
-       dev = dev_get_by_name(net, iniface);
-       if (!dev) {
-               pr_info("no such interface %s\n", iniface);
-               kfree(c);
-               return ERR_PTR(-ENOENT);
-       }
-       c->ifindex = dev->ifindex;
-       strcpy(c->ifname, dev->name);
-       memcpy(&c->clustermac, &i->clustermac, ETH_ALEN);
-       dev_mc_add(dev, c->clustermac);
-       dev_put(dev);
-
-       c->clusterip = ip;
-       c->num_total_nodes = i->num_total_nodes;
-       clusterip_config_init_nodelist(c, i);
-       c->hash_mode = i->hash_mode;
-       c->hash_initval = i->hash_initval;
-       c->net = net;
-       refcount_set(&c->refcount, 1);
-
-       spin_lock_bh(&cn->lock);
-       if (__clusterip_config_find(net, ip)) {
-               err = -EBUSY;
-               goto out_config_put;
-       }
-
-       list_add_rcu(&c->list, &cn->configs);
-       spin_unlock_bh(&cn->lock);
-
-#ifdef CONFIG_PROC_FS
-       {
-               char buffer[16];
-
-               /* create proc dir entry */
-               sprintf(buffer, "%pI4", &ip);
-               mutex_lock(&cn->mutex);
-               c->pde = proc_create_data(buffer, 0600,
-                                         cn->procdir,
-                                         &clusterip_proc_ops, c);
-               mutex_unlock(&cn->mutex);
-               if (!c->pde) {
-                       err = -ENOMEM;
-                       goto err;
-               }
-       }
-#endif
-
-       refcount_set(&c->entries, 1);
-       return c;
-
-#ifdef CONFIG_PROC_FS
-err:
-#endif
-       spin_lock_bh(&cn->lock);
-       list_del_rcu(&c->list);
-out_config_put:
-       spin_unlock_bh(&cn->lock);
-       clusterip_config_put(c);
-       return ERR_PTR(err);
-}
-
-#ifdef CONFIG_PROC_FS
-static int
-clusterip_add_node(struct clusterip_config *c, u_int16_t nodenum)
-{
-
-       if (nodenum == 0 ||
-           nodenum > c->num_total_nodes)
-               return 1;
-
-       /* check if we already have this number in our bitfield */
-       if (test_and_set_bit(nodenum - 1, &c->local_nodes))
-               return 1;
-
-       return 0;
-}
-
-static bool
-clusterip_del_node(struct clusterip_config *c, u_int16_t nodenum)
-{
-       if (nodenum == 0 ||
-           nodenum > c->num_total_nodes)
-               return true;
-
-       if (test_and_clear_bit(nodenum - 1, &c->local_nodes))
-               return false;
-
-       return true;
-}
-#endif
-
-static inline u_int32_t
-clusterip_hashfn(const struct sk_buff *skb,
-                const struct clusterip_config *config)
-{
-       const struct iphdr *iph = ip_hdr(skb);
-       unsigned long hashval;
-       u_int16_t sport = 0, dport = 0;
-       int poff;
-
-       poff = proto_ports_offset(iph->protocol);
-       if (poff >= 0) {
-               const u_int16_t *ports;
-               u16 _ports[2];
-
-               ports = skb_header_pointer(skb, iph->ihl * 4 + poff, 4, _ports);
-               if (ports) {
-                       sport = ports[0];
-                       dport = ports[1];
-               }
-       } else {
-               net_info_ratelimited("unknown protocol %u\n", iph->protocol);
-       }
-
-       switch (config->hash_mode) {
-       case CLUSTERIP_HASHMODE_SIP:
-               hashval = jhash_1word(ntohl(iph->saddr),
-                                     config->hash_initval);
-               break;
-       case CLUSTERIP_HASHMODE_SIP_SPT:
-               hashval = jhash_2words(ntohl(iph->saddr), sport,
-                                      config->hash_initval);
-               break;
-       case CLUSTERIP_HASHMODE_SIP_SPT_DPT:
-               hashval = jhash_3words(ntohl(iph->saddr), sport, dport,
-                                      config->hash_initval);
-               break;
-       default:
-               /* to make gcc happy */
-               hashval = 0;
-               /* This cannot happen, unless the check function wasn't called
-                * at rule load time */
-               pr_info("unknown mode %u\n", config->hash_mode);
-               BUG();
-               break;
-       }
-
-       /* node numbers are 1..n, not 0..n */
-       return reciprocal_scale(hashval, config->num_total_nodes) + 1;
-}
-
-static inline int
-clusterip_responsible(const struct clusterip_config *config, u_int32_t hash)
-{
-       return test_bit(hash - 1, &config->local_nodes);
-}
-
-/***********************************************************************
- * IPTABLES TARGET
- ***********************************************************************/
-
-static unsigned int
-clusterip_tg(struct sk_buff *skb, const struct xt_action_param *par)
-{
-       const struct ipt_clusterip_tgt_info *cipinfo = par->targinfo;
-       struct nf_conn *ct;
-       enum ip_conntrack_info ctinfo;
-       u_int32_t hash;
-
-       /* don't need to clusterip_config_get() here, since refcount
-        * is only decremented by destroy() - and ip_tables guarantees
-        * that the ->target() function isn't called after ->destroy() */
-
-       ct = nf_ct_get(skb, &ctinfo);
-       if (ct == NULL)
-               return NF_DROP;
-
-       /* special case: ICMP error handling. conntrack distinguishes between
-        * error messages (RELATED) and information requests (see below) */
-       if (ip_hdr(skb)->protocol == IPPROTO_ICMP &&
-           (ctinfo == IP_CT_RELATED ||
-            ctinfo == IP_CT_RELATED_REPLY))
-               return XT_CONTINUE;
-
-       /* nf_conntrack_proto_icmp guarantees us that we only have ICMP_ECHO,
-        * TIMESTAMP, INFO_REQUEST or ICMP_ADDRESS type icmp packets from here
-        * on, which all have an ID field [relevant for hashing]. */
-
-       hash = clusterip_hashfn(skb, cipinfo->config);
-
-       switch (ctinfo) {
-       case IP_CT_NEW:
-               WRITE_ONCE(ct->mark, hash);
-               break;
-       case IP_CT_RELATED:
-       case IP_CT_RELATED_REPLY:
-               /* FIXME: we don't handle expectations at the moment.
-                * They can arrive on a different node than
-                * the master connection (e.g. FTP passive mode) */
-       case IP_CT_ESTABLISHED:
-       case IP_CT_ESTABLISHED_REPLY:
-               break;
-       default:                        /* Prevent gcc warnings */
-               break;
-       }
-
-#ifdef DEBUG
-       nf_ct_dump_tuple_ip(&ct->tuplehash[IP_CT_DIR_ORIGINAL].tuple);
-#endif
-       pr_debug("hash=%u ct_hash=%u ", hash, READ_ONCE(ct->mark));
-       if (!clusterip_responsible(cipinfo->config, hash)) {
-               pr_debug("not responsible\n");
-               return NF_DROP;
-       }
-       pr_debug("responsible\n");
-
-       /* despite being received via linklayer multicast, this is
-        * actually a unicast IP packet. TCP doesn't like PACKET_MULTICAST */
-       skb->pkt_type = PACKET_HOST;
-
-       return XT_CONTINUE;
-}
-
-static int clusterip_tg_check(const struct xt_tgchk_param *par)
-{
-       struct ipt_clusterip_tgt_info *cipinfo = par->targinfo;
-       struct clusterip_net *cn = clusterip_pernet(par->net);
-       const struct ipt_entry *e = par->entryinfo;
-       struct clusterip_config *config;
-       int ret, i;
-
-       if (par->nft_compat) {
-               pr_err("cannot use CLUSTERIP target from nftables compat\n");
-               return -EOPNOTSUPP;
-       }
-
-       if (cn->hook_users == UINT_MAX)
-               return -EOVERFLOW;
-
-       if (cipinfo->hash_mode != CLUSTERIP_HASHMODE_SIP &&
-           cipinfo->hash_mode != CLUSTERIP_HASHMODE_SIP_SPT &&
-           cipinfo->hash_mode != CLUSTERIP_HASHMODE_SIP_SPT_DPT) {
-               pr_info("unknown mode %u\n", cipinfo->hash_mode);
-               return -EINVAL;
-
-       }
-       if (e->ip.dmsk.s_addr != htonl(0xffffffff) ||
-           e->ip.dst.s_addr == 0) {
-               pr_info("Please specify destination IP\n");
-               return -EINVAL;
-       }
-       if (cipinfo->num_local_nodes > ARRAY_SIZE(cipinfo->local_nodes)) {
-               pr_info("bad num_local_nodes %u\n", cipinfo->num_local_nodes);
-               return -EINVAL;
-       }
-       for (i = 0; i < cipinfo->num_local_nodes; i++) {
-               if (cipinfo->local_nodes[i] - 1 >=
-                   sizeof(config->local_nodes) * 8) {
-                       pr_info("bad local_nodes[%d] %u\n",
-                               i, cipinfo->local_nodes[i]);
-                       return -EINVAL;
-               }
-       }
-
-       config = clusterip_config_find_get(par->net, e->ip.dst.s_addr, 1);
-       if (!config) {
-               if (!(cipinfo->flags & CLUSTERIP_FLAG_NEW)) {
-                       pr_info("no config found for %pI4, need 'new'\n",
-                               &e->ip.dst.s_addr);
-                       return -EINVAL;
-               } else {
-                       config = clusterip_config_init(par->net, cipinfo,
-                                                      e->ip.dst.s_addr,
-                                                      e->ip.iniface);
-                       if (IS_ERR(config))
-                               return PTR_ERR(config);
-               }
-       } else if (memcmp(&config->clustermac, &cipinfo->clustermac, ETH_ALEN)) {
-               clusterip_config_entry_put(config);
-               clusterip_config_put(config);
-               return -EINVAL;
-       }
-
-       ret = nf_ct_netns_get(par->net, par->family);
-       if (ret < 0) {
-               pr_info("cannot load conntrack support for proto=%u\n",
-                       par->family);
-               clusterip_config_entry_put(config);
-               clusterip_config_put(config);
-               return ret;
-       }
-
-       if (cn->hook_users == 0) {
-               ret = nf_register_net_hook(par->net, &cip_arp_ops);
-
-               if (ret < 0) {
-                       clusterip_config_entry_put(config);
-                       clusterip_config_put(config);
-                       nf_ct_netns_put(par->net, par->family);
-                       return ret;
-               }
-       }
-
-       cn->hook_users++;
-
-       if (!cn->clusterip_deprecated_warning) {
-               pr_info("ipt_CLUSTERIP is deprecated and it will removed soon, "
-                       "use xt_cluster instead\n");
-               cn->clusterip_deprecated_warning = true;
-       }
-
-       cipinfo->config = config;
-       return ret;
-}
-
-/* drop reference count of cluster config when rule is deleted */
-static void clusterip_tg_destroy(const struct xt_tgdtor_param *par)
-{
-       const struct ipt_clusterip_tgt_info *cipinfo = par->targinfo;
-       struct clusterip_net *cn = clusterip_pernet(par->net);
-
-       /* if no more entries are referencing the config, remove it
-        * from the list and destroy the proc entry */
-       clusterip_config_entry_put(cipinfo->config);
-
-       clusterip_config_put(cipinfo->config);
-
-       nf_ct_netns_put(par->net, par->family);
-       cn->hook_users--;
-
-       if (cn->hook_users == 0)
-               nf_unregister_net_hook(par->net, &cip_arp_ops);
-}
-
-#ifdef CONFIG_NETFILTER_XTABLES_COMPAT
-struct compat_ipt_clusterip_tgt_info
-{
-       u_int32_t       flags;
-       u_int8_t        clustermac[6];
-       u_int16_t       num_total_nodes;
-       u_int16_t       num_local_nodes;
-       u_int16_t       local_nodes[CLUSTERIP_MAX_NODES];
-       u_int32_t       hash_mode;
-       u_int32_t       hash_initval;
-       compat_uptr_t   config;
-};
-#endif /* CONFIG_NETFILTER_XTABLES_COMPAT */
-
-static struct xt_target clusterip_tg_reg __read_mostly = {
-       .name           = "CLUSTERIP",
-       .family         = NFPROTO_IPV4,
-       .target         = clusterip_tg,
-       .checkentry     = clusterip_tg_check,
-       .destroy        = clusterip_tg_destroy,
-       .targetsize     = sizeof(struct ipt_clusterip_tgt_info),
-       .usersize       = offsetof(struct ipt_clusterip_tgt_info, config),
-#ifdef CONFIG_NETFILTER_XTABLES_COMPAT
-       .compatsize     = sizeof(struct compat_ipt_clusterip_tgt_info),
-#endif /* CONFIG_NETFILTER_XTABLES_COMPAT */
-       .me             = THIS_MODULE
-};
-
-
-/***********************************************************************
- * ARP MANGLING CODE
- ***********************************************************************/
-
-/* hardcoded for 48bit ethernet and 32bit ipv4 addresses */
-struct arp_payload {
-       u_int8_t src_hw[ETH_ALEN];
-       __be32 src_ip;
-       u_int8_t dst_hw[ETH_ALEN];
-       __be32 dst_ip;
-} __packed;
-
-#ifdef DEBUG
-static void arp_print(struct arp_payload *payload)
-{
-#define HBUFFERLEN 30
-       char hbuffer[HBUFFERLEN];
-       int j, k;
-
-       for (k = 0, j = 0; k < HBUFFERLEN - 3 && j < ETH_ALEN; j++) {
-               hbuffer[k++] = hex_asc_hi(payload->src_hw[j]);
-               hbuffer[k++] = hex_asc_lo(payload->src_hw[j]);
-               hbuffer[k++] = ':';
-       }
-       hbuffer[--k] = '\0';
-
-       pr_debug("src %pI4@%s, dst %pI4\n",
-                &payload->src_ip, hbuffer, &payload->dst_ip);
-}
-#endif
-
-static unsigned int
-clusterip_arp_mangle(void *priv, struct sk_buff *skb,
-                    const struct nf_hook_state *state)
-{
-       struct arphdr *arp = arp_hdr(skb);
-       struct arp_payload *payload;
-       struct clusterip_config *c;
-       struct net *net = state->net;
-
-       /* we don't care about non-ethernet and non-ipv4 ARP */
-       if (arp->ar_hrd != htons(ARPHRD_ETHER) ||
-           arp->ar_pro != htons(ETH_P_IP) ||
-           arp->ar_pln != 4 || arp->ar_hln != ETH_ALEN)
-               return NF_ACCEPT;
-
-       /* we only want to mangle arp requests and replies */
-       if (arp->ar_op != htons(ARPOP_REPLY) &&
-           arp->ar_op != htons(ARPOP_REQUEST))
-               return NF_ACCEPT;
-
-       payload = (void *)(arp+1);
-
-       /* if there is no clusterip configuration for the arp reply's
-        * source ip, we don't want to mangle it */
-       c = clusterip_config_find_get(net, payload->src_ip, 0);
-       if (!c)
-               return NF_ACCEPT;
-
-       /* normally the linux kernel always replies to arp queries of
-        * addresses on different interfacs.  However, in the CLUSTERIP case
-        * this wouldn't work, since we didn't subscribe the mcast group on
-        * other interfaces */
-       if (c->ifindex != state->out->ifindex) {
-               pr_debug("not mangling arp reply on different interface: cip'%d'-skb'%d'\n",
-                        c->ifindex, state->out->ifindex);
-               clusterip_config_put(c);
-               return NF_ACCEPT;
-       }
-
-       /* mangle reply hardware address */
-       memcpy(payload->src_hw, c->clustermac, arp->ar_hln);
-
-#ifdef DEBUG
-       pr_debug("mangled arp reply: ");
-       arp_print(payload);
-#endif
-
-       clusterip_config_put(c);
-
-       return NF_ACCEPT;
-}
-
-/***********************************************************************
- * PROC DIR HANDLING
- ***********************************************************************/
-
-#ifdef CONFIG_PROC_FS
-
-struct clusterip_seq_position {
-       unsigned int pos;       /* position */
-       unsigned int weight;    /* number of bits set == size */
-       unsigned int bit;       /* current bit */
-       unsigned long val;      /* current value */
-};
-
-static void *clusterip_seq_start(struct seq_file *s, loff_t *pos)
-{
-       struct clusterip_config *c = s->private;
-       unsigned int weight;
-       u_int32_t local_nodes;
-       struct clusterip_seq_position *idx;
-
-       /* FIXME: possible race */
-       local_nodes = c->local_nodes;
-       weight = hweight32(local_nodes);
-       if (*pos >= weight)
-               return NULL;
-
-       idx = kmalloc(sizeof(struct clusterip_seq_position), GFP_KERNEL);
-       if (!idx)
-               return ERR_PTR(-ENOMEM);
-
-       idx->pos = *pos;
-       idx->weight = weight;
-       idx->bit = ffs(local_nodes);
-       idx->val = local_nodes;
-       clear_bit(idx->bit - 1, &idx->val);
-
-       return idx;
-}
-
-static void *clusterip_seq_next(struct seq_file *s, void *v, loff_t *pos)
-{
-       struct clusterip_seq_position *idx = v;
-
-       *pos = ++idx->pos;
-       if (*pos >= idx->weight) {
-               kfree(v);
-               return NULL;
-       }
-       idx->bit = ffs(idx->val);
-       clear_bit(idx->bit - 1, &idx->val);
-       return idx;
-}
-
-static void clusterip_seq_stop(struct seq_file *s, void *v)
-{
-       if (!IS_ERR(v))
-               kfree(v);
-}
-
-static int clusterip_seq_show(struct seq_file *s, void *v)
-{
-       struct clusterip_seq_position *idx = v;
-
-       if (idx->pos != 0)
-               seq_putc(s, ',');
-
-       seq_printf(s, "%u", idx->bit);
-
-       if (idx->pos == idx->weight - 1)
-               seq_putc(s, '\n');
-
-       return 0;
-}
-
-static const struct seq_operations clusterip_seq_ops = {
-       .start  = clusterip_seq_start,
-       .next   = clusterip_seq_next,
-       .stop   = clusterip_seq_stop,
-       .show   = clusterip_seq_show,
-};
-
-static int clusterip_proc_open(struct inode *inode, struct file *file)
-{
-       int ret = seq_open(file, &clusterip_seq_ops);
-
-       if (!ret) {
-               struct seq_file *sf = file->private_data;
-               struct clusterip_config *c = pde_data(inode);
-
-               sf->private = c;
-
-               clusterip_config_get(c);
-       }
-
-       return ret;
-}
-
-static int clusterip_proc_release(struct inode *inode, struct file *file)
-{
-       struct clusterip_config *c = pde_data(inode);
-       int ret;
-
-       ret = seq_release(inode, file);
-
-       if (!ret)
-               clusterip_config_put(c);
-
-       return ret;
-}
-
-static ssize_t clusterip_proc_write(struct file *file, const char __user *input,
-                               size_t size, loff_t *ofs)
-{
-       struct clusterip_config *c = pde_data(file_inode(file));
-#define PROC_WRITELEN  10
-       char buffer[PROC_WRITELEN+1];
-       unsigned long nodenum;
-       int rc;
-
-       if (size > PROC_WRITELEN)
-               return -EIO;
-       if (copy_from_user(buffer, input, size))
-               return -EFAULT;
-       buffer[size] = 0;
-
-       if (*buffer == '+') {
-               rc = kstrtoul(buffer+1, 10, &nodenum);
-               if (rc)
-                       return rc;
-               if (clusterip_add_node(c, nodenum))
-                       return -ENOMEM;
-       } else if (*buffer == '-') {
-               rc = kstrtoul(buffer+1, 10, &nodenum);
-               if (rc)
-                       return rc;
-               if (clusterip_del_node(c, nodenum))
-                       return -ENOENT;
-       } else
-               return -EIO;
-
-       return size;
-}
-
-static const struct proc_ops clusterip_proc_ops = {
-       .proc_open      = clusterip_proc_open,
-       .proc_read      = seq_read,
-       .proc_write     = clusterip_proc_write,
-       .proc_lseek     = seq_lseek,
-       .proc_release   = clusterip_proc_release,
-};
-
-#endif /* CONFIG_PROC_FS */
-
-static int clusterip_net_init(struct net *net)
-{
-       struct clusterip_net *cn = clusterip_pernet(net);
-
-       INIT_LIST_HEAD(&cn->configs);
-
-       spin_lock_init(&cn->lock);
-
-#ifdef CONFIG_PROC_FS
-       cn->procdir = proc_mkdir("ipt_CLUSTERIP", net->proc_net);
-       if (!cn->procdir) {
-               pr_err("Unable to proc dir entry\n");
-               return -ENOMEM;
-       }
-       mutex_init(&cn->mutex);
-#endif /* CONFIG_PROC_FS */
-
-       return 0;
-}
-
-static void clusterip_net_exit(struct net *net)
-{
-#ifdef CONFIG_PROC_FS
-       struct clusterip_net *cn = clusterip_pernet(net);
-
-       mutex_lock(&cn->mutex);
-       proc_remove(cn->procdir);
-       cn->procdir = NULL;
-       mutex_unlock(&cn->mutex);
-#endif
-}
-
-static struct pernet_operations clusterip_net_ops = {
-       .init = clusterip_net_init,
-       .exit = clusterip_net_exit,
-       .id   = &clusterip_net_id,
-       .size = sizeof(struct clusterip_net),
-};
-
-static struct notifier_block cip_netdev_notifier = {
-       .notifier_call = clusterip_netdev_event
-};
-
-static int __init clusterip_tg_init(void)
-{
-       int ret;
-
-       ret = register_pernet_subsys(&clusterip_net_ops);
-       if (ret < 0)
-               return ret;
-
-       ret = xt_register_target(&clusterip_tg_reg);
-       if (ret < 0)
-               goto cleanup_subsys;
-
-       ret = register_netdevice_notifier(&cip_netdev_notifier);
-       if (ret < 0)
-               goto unregister_target;
-
-       pr_info("ClusterIP Version %s loaded successfully\n",
-               CLUSTERIP_VERSION);
-
-       return 0;
-
-unregister_target:
-       xt_unregister_target(&clusterip_tg_reg);
-cleanup_subsys:
-       unregister_pernet_subsys(&clusterip_net_ops);
-       return ret;
-}
-
-static void __exit clusterip_tg_exit(void)
-{
-       pr_info("ClusterIP Version %s unloading\n", CLUSTERIP_VERSION);
-
-       unregister_netdevice_notifier(&cip_netdev_notifier);
-       xt_unregister_target(&clusterip_tg_reg);
-       unregister_pernet_subsys(&clusterip_net_ops);
-
-       /* Wait for completion of call_rcu()'s (clusterip_config_rcu_free) */
-       rcu_barrier();
-}
-
-module_init(clusterip_tg_init);
-module_exit(clusterip_tg_exit);