[NET]: Merge in IPVS layer.

net/ipv4/Kconfig

@@ -375,4 +375,5 @@ config INET_IPCOMP
 	  If unsure, say Y.
 
 source "net/ipv4/netfilter/Kconfig"
+source "net/ipv4/ipvs/Kconfig"
 

net/ipv4/Makefile

@@ -21,5 +21,6 @@ obj-$(CONFIG_INET_ESP) += esp4.o
 obj-$(CONFIG_INET_IPCOMP) += ipcomp.o
 obj-$(CONFIG_IP_PNP) += ipconfig.o
 obj-$(CONFIG_NETFILTER)	+= netfilter/
+obj-$(CONFIG_IP_VS) += ipvs/
 
 obj-y += xfrm4_policy.o xfrm4_state.o xfrm4_input.o xfrm4_tunnel.o

net/ipv4/ipvs/Makefile

+#
+# Makefile for the IPVS modules on top of IPv4.
+#
+# Note! Dependencies are done automagically by 'make dep', which also
+# removes any old dependencies. DON'T put your own dependencies here
+# unless it's something special (ie not a .c file).
+#
+# Note 2! The CFLAGS definition is now in the main makefile...
+
+
+# IPVS transport protocol load balancing support
+ip_vs_proto-objs-y :=
+ip_vs_proto-objs-$(CONFIG_IP_VS_PROTO_TCP) += ip_vs_proto_tcp.o
+ip_vs_proto-objs-$(CONFIG_IP_VS_PROTO_UDP) += ip_vs_proto_udp.o
+ip_vs_proto-objs-$(CONFIG_IP_VS_PROTO_ESP) += ip_vs_proto_esp.o
+ip_vs_proto-objs-$(CONFIG_IP_VS_PROTO_AH) += ip_vs_proto_ah.o
+
+ip_vs-objs :=	ip_vs_conn.o ip_vs_core.o ip_vs_ctl.o ip_vs_sched.o	   \
+		ip_vs_xmit.o ip_vs_timer.o ip_vs_app.o ip_vs_sync.o	   \
+		ip_vs_est.o ip_vs_proto.o ip_vs_proto_icmp.o		   \
+		$(ip_vs_proto-objs-y)
+
+
+# IPVS core
+obj-$(CONFIG_IP_VS) += ip_vs.o
+
+# IPVS schedulers
+obj-$(CONFIG_IP_VS_RR) += ip_vs_rr.o
+obj-$(CONFIG_IP_VS_WRR) += ip_vs_wrr.o
+obj-$(CONFIG_IP_VS_LC) += ip_vs_lc.o
+obj-$(CONFIG_IP_VS_WLC) += ip_vs_wlc.o
+obj-$(CONFIG_IP_VS_LBLC) += ip_vs_lblc.o
+obj-$(CONFIG_IP_VS_LBLCR) += ip_vs_lblcr.o
+obj-$(CONFIG_IP_VS_DH) += ip_vs_dh.o
+obj-$(CONFIG_IP_VS_SH) += ip_vs_sh.o
+obj-$(CONFIG_IP_VS_SED) += ip_vs_sed.o
+obj-$(CONFIG_IP_VS_NQ) += ip_vs_nq.o
+
+# IPVS application helpers
+obj-$(CONFIG_IP_VS_FTP) += ip_vs_ftp.o

net/ipv4/ipvs/ip_vs_dh.c

+/*
+ * IPVS:        Destination Hashing scheduling module
+ *
+ * Version:     $Id: ip_vs_dh.c,v 1.5 2002/09/15 08:14:08 wensong Exp $
+ *
+ * Authors:     Wensong Zhang <wensong@gnuchina.org>
+ *
+ *              Inspired by the consistent hashing scheduler patch from
+ *              Thomas Proell <proellt@gmx.de>
+ *
+ *              This program is free software; you can redistribute it and/or
+ *              modify it under the terms of the GNU General Public License
+ *              as published by the Free Software Foundation; either version
+ *              2 of the License, or (at your option) any later version.
+ *
+ * Changes:
+ *
+ */
+
+/*
+ * The dh algorithm is to select server by the hash key of destination IP
+ * address. The pseudo code is as follows:
+ *
+ *       n <- servernode[dest_ip];
+ *       if (n is dead) OR
+ *          (n is overloaded) OR (n.weight <= 0) then
+ *                 return NULL;
+ *
+ *       return n;
+ *
+ * Notes that servernode is a 256-bucket hash table that maps the hash
+ * index derived from packet destination IP address to the current server
+ * array. If the dh scheduler is used in cache cluster, it is good to
+ * combine it with cache_bypass feature. When the statically assigned
+ * server is dead or overloaded, the load balancer can bypass the cache
+ * server and send requests to the original server directly.
+ *
+ */
+
+#include <linux/config.h>
+#include <linux/module.h>
+#include <linux/init.h>
+#include <linux/types.h>
+#include <linux/kernel.h>
+#include <linux/errno.h>
+
+#include <net/ip_vs.h>
+
+
+/*
+ *      IPVS DH bucket
+ */
+struct ip_vs_dh_bucket {
+	struct ip_vs_dest       *dest;          /* real server (cache) */
+};
+
+/*
+ *     for IPVS DH entry hash table
+ */
+#ifndef CONFIG_IP_VS_DH_TAB_BITS
+#define CONFIG_IP_VS_DH_TAB_BITS        8
+#endif
+#define IP_VS_DH_TAB_BITS               CONFIG_IP_VS_DH_TAB_BITS
+#define IP_VS_DH_TAB_SIZE               (1 << IP_VS_DH_TAB_BITS)
+#define IP_VS_DH_TAB_MASK               (IP_VS_DH_TAB_SIZE - 1)
+
+
+/*
+ *	Returns hash value for IPVS DH entry
+ */
+static inline unsigned ip_vs_dh_hashkey(__u32 addr)
+{
+	return (ntohl(addr)*2654435761UL) & IP_VS_DH_TAB_MASK;
+}
+
+
+/*
+ *      Get ip_vs_dest associated with supplied parameters.
+ */
+static inline struct ip_vs_dest *
+ip_vs_dh_get(struct ip_vs_dh_bucket *tbl, __u32 addr)
+{
+	return (tbl[ip_vs_dh_hashkey(addr)]).dest;
+}
+
+
+/*
+ *      Assign all the hash buckets of the specified table with the service.
+ */
+static int
+ip_vs_dh_assign(struct ip_vs_dh_bucket *tbl, struct ip_vs_service *svc)
+{
+	int i;
+	struct ip_vs_dh_bucket *b;
+	struct list_head *p;
+	struct ip_vs_dest *dest;
+
+	b = tbl;
+	p = &svc->destinations;
+	for (i=0; i<IP_VS_DH_TAB_SIZE; i++) {
+		if (list_empty(p)) {
+			b->dest = NULL;
+		} else {
+			if (p == &svc->destinations)
+				p = p->next;
+
+			dest = list_entry(p, struct ip_vs_dest, n_list);
+			atomic_inc(&dest->refcnt);
+			b->dest = dest;
+
+			p = p->next;
+		}
+		b++;
+	}
+	return 0;
+}
+
+
+/*
+ *      Flush all the hash buckets of the specified table.
+ */
+static void ip_vs_dh_flush(struct ip_vs_dh_bucket *tbl)
+{
+	int i;
+	struct ip_vs_dh_bucket *b;
+
+	b = tbl;
+	for (i=0; i<IP_VS_DH_TAB_SIZE; i++) {
+		if (b->dest) {
+			atomic_dec(&b->dest->refcnt);
+			b->dest = NULL;
+		}
+		b++;
+	}
+}
+
+
+static int ip_vs_dh_init_svc(struct ip_vs_service *svc)
+{
+	struct ip_vs_dh_bucket *tbl;
+
+	/* allocate the DH table for this service */
+	tbl = kmalloc(sizeof(struct ip_vs_dh_bucket)*IP_VS_DH_TAB_SIZE,
+		      GFP_ATOMIC);
+	if (tbl == NULL) {
+		IP_VS_ERR("ip_vs_dh_init_svc(): no memory\n");
+		return -ENOMEM;
+	}
+	svc->sched_data = tbl;
+	IP_VS_DBG(6, "DH hash table (memory=%dbytes) allocated for "
+		  "current service\n",
+		  sizeof(struct ip_vs_dh_bucket)*IP_VS_DH_TAB_SIZE);
+
+	/* assign the hash buckets with the updated service */
+	ip_vs_dh_assign(tbl, svc);
+
+	return 0;
+}
+
+
+static int ip_vs_dh_done_svc(struct ip_vs_service *svc)
+{
+	struct ip_vs_dh_bucket *tbl = svc->sched_data;
+
+	/* got to clean up hash buckets here */
+	ip_vs_dh_flush(tbl);
+
+	/* release the table itself */
+	kfree(svc->sched_data);
+	IP_VS_DBG(6, "DH hash table (memory=%dbytes) released\n",
+		  sizeof(struct ip_vs_dh_bucket)*IP_VS_DH_TAB_SIZE);
+
+	return 0;
+}
+
+
+static int ip_vs_dh_update_svc(struct ip_vs_service *svc)
+{
+	struct ip_vs_dh_bucket *tbl = svc->sched_data;
+
+	/* got to clean up hash buckets here */
+	ip_vs_dh_flush(tbl);
+
+	/* assign the hash buckets with the updated service */
+	ip_vs_dh_assign(tbl, svc);
+
+	return 0;
+}
+
+
+/*
+ *      If the dest flags is set with IP_VS_DEST_F_OVERLOAD,
+ *      consider that the server is overloaded here.
+ */
+static inline int is_overloaded(struct ip_vs_dest *dest)
+{
+	return dest->flags & IP_VS_DEST_F_OVERLOAD;
+}
+
+
+/*
+ *      Destination hashing scheduling
+ */
+static struct ip_vs_dest *
+ip_vs_dh_schedule(struct ip_vs_service *svc, struct iphdr *iph)
+{
+	struct ip_vs_dest *dest;
+	struct ip_vs_dh_bucket *tbl;
+
+	IP_VS_DBG(6, "ip_vs_dh_schedule(): Scheduling...\n");
+
+	tbl = (struct ip_vs_dh_bucket *)svc->sched_data;
+	dest = ip_vs_dh_get(tbl, iph->daddr);
+	if (!dest
+	    || !(dest->flags & IP_VS_DEST_F_AVAILABLE)
+	    || atomic_read(&dest->weight) <= 0
+	    || is_overloaded(dest)) {
+		return NULL;
+	}
+
+	IP_VS_DBG(6, "DH: destination IP address %u.%u.%u.%u "
+		  "--> server %u.%u.%u.%u:%d\n",
+		  NIPQUAD(iph->daddr),
+		  NIPQUAD(dest->addr),
+		  ntohs(dest->port));
+
+	return dest;
+}
+
+
+/*
+ *      IPVS DH Scheduler structure
+ */
+static struct ip_vs_scheduler ip_vs_dh_scheduler =
+{
+	.name =			"dh",
+	.refcnt =		ATOMIC_INIT(0),
+	.module =		THIS_MODULE,
+	.init_service =		ip_vs_dh_init_svc,
+	.done_service =		ip_vs_dh_done_svc,
+	.update_service =	ip_vs_dh_update_svc,
+	.schedule =		ip_vs_dh_schedule,
+};
+
+
+static int __init ip_vs_dh_init(void)
+{
+	INIT_LIST_HEAD(&ip_vs_dh_scheduler.n_list);
+	return register_ip_vs_scheduler(&ip_vs_dh_scheduler);
+}
+
+
+static void __exit ip_vs_dh_cleanup(void)
+{
+	unregister_ip_vs_scheduler(&ip_vs_dh_scheduler);
+}
+
+
+module_init(ip_vs_dh_init);
+module_exit(ip_vs_dh_cleanup);
+MODULE_LICENSE("GPL");

net/ipv4/ipvs/ip_vs_est.c

+/*
+ * ip_vs_est.c: simple rate estimator for IPVS
+ *
+ * Version:     $Id: ip_vs_est.c,v 1.4 2002/11/30 01:50:35 wensong Exp $
+ *
+ * Authors:     Wensong Zhang <wensong@linuxvirtualserver.org>
+ *
+ *              This program is free software; you can redistribute it and/or
+ *              modify it under the terms of the GNU General Public License
+ *              as published by the Free Software Foundation; either version
+ *              2 of the License, or (at your option) any later version.
+ *
+ * Changes:
+ *
+ */
+#include <linux/kernel.h>
+#include <linux/types.h>
+
+#include <net/ip_vs.h>
+
+/*
+  This code is to estimate rate in a shorter interval (such as 8
+  seconds) for virtual services and real servers. For measure rate in a
+  long interval, it is easy to implement a user level daemon which
+  periodically reads those statistical counters and measure rate.
+
+  Currently, the measurement is activated by slow timer handler. Hope
+  this measurement will not introduce too much load.
+
+  We measure rate during the last 8 seconds every 2 seconds:
+
+    avgrate = avgrate*(1-W) + rate*W
+
+    where W = 2^(-2)
+
+  NOTES.
+
+  * The stored value for average bps is scaled by 2^5, so that maximal
+    rate is ~2.15Gbits/s, average pps and cps are scaled by 2^10.
+
+  * A lot code is taken from net/sched/estimator.c
+ */
+
+
+struct ip_vs_estimator
+{
+	struct ip_vs_estimator	*next;
+	struct ip_vs_stats	*stats;
+
+	u32			last_conns;
+	u32			last_inpkts;
+	u32			last_outpkts;
+	u64			last_inbytes;
+	u64			last_outbytes;
+
+	u32			cps;
+	u32			inpps;
+	u32			outpps;
+	u32			inbps;
+	u32			outbps;
+};
+
+
+static struct ip_vs_estimator *est_list = NULL;
+static rwlock_t est_lock = RW_LOCK_UNLOCKED;
+static struct timer_list est_timer;
+
+static void estimation_timer(unsigned long arg)
+{
+	struct ip_vs_estimator *e;
+	struct ip_vs_stats *s;
+	u32 n_conns;
+	u32 n_inpkts, n_outpkts;
+	u64 n_inbytes, n_outbytes;
+	u32 rate;
+
+	read_lock(&est_lock);
+	for (e = est_list; e; e = e->next) {
+		s = e->stats;
+		n_conns = s->conns;
+		n_inpkts = s->inpkts;
+		n_outpkts = s->outpkts;
+		n_inbytes = s->inbytes;
+		n_outbytes = s->outbytes;
+
+		/* scaled by 2^10, but divided 2 seconds */
+		rate = (n_conns - e->last_conns)<<9;
+		e->last_conns = n_conns;
+		e->cps += ((long)rate - (long)e->cps)>>2;
+		s->cps = (e->cps+0x1FF)>>10;
+
+		rate = (n_inpkts - e->last_inpkts)<<9;
+		e->last_inpkts = n_inpkts;
+		e->inpps += ((long)rate - (long)e->inpps)>>2;
+		s->inpps = (e->inpps+0x1FF)>>10;
+
+		rate = (n_outpkts - e->last_outpkts)<<9;
+		e->last_outpkts = n_outpkts;
+		e->outpps += ((long)rate - (long)e->outpps)>>2;
+		s->outpps = (e->outpps+0x1FF)>>10;
+
+		rate = (n_inbytes - e->last_inbytes)<<4;
+		e->last_inbytes = n_inbytes;
+		e->inbps += ((long)rate - (long)e->inbps)>>2;
+		s->inbps = (e->inbps+0xF)>>5;
+
+		rate = (n_outbytes - e->last_outbytes)<<4;
+		e->last_outbytes = n_outbytes;
+		e->outbps += ((long)rate - (long)e->outbps)>>2;
+		s->outbps = (e->outbps+0xF)>>5;
+	}
+	read_unlock(&est_lock);
+	mod_timer(&est_timer, jiffies + 2*HZ);
+}
+
+int ip_vs_new_estimator(struct ip_vs_stats *stats)
+{
+	struct ip_vs_estimator *est;
+
+	est = kmalloc(sizeof(*est), GFP_KERNEL);
+	if (est == NULL)
+		return -ENOMEM;
+
+	memset(est, 0, sizeof(*est));
+	est->stats = stats;
+	est->last_conns = stats->conns;
+	est->cps = stats->cps<<10;
+
+	est->last_inpkts = stats->inpkts;
+	est->inpps = stats->inpps<<10;
+
+	est->last_outpkts = stats->outpkts;
+	est->outpps = stats->outpps<<10;
+
+	est->last_inbytes = stats->inbytes;
+	est->inbps = stats->inbps<<5;
+
+	est->last_outbytes = stats->outbytes;
+	est->outbps = stats->outbps<<5;
+
+	write_lock_bh(&est_lock);
+	est->next = est_list;
+	if (est->next == NULL) {
+		init_timer(&est_timer);
+		est_timer.expires = jiffies + 2*HZ;
+		est_timer.function = estimation_timer;
+		add_timer(&est_timer);
+	}
+	est_list = est;
+	write_unlock_bh(&est_lock);
+	return 0;
+}
+
+void ip_vs_kill_estimator(struct ip_vs_stats *stats)
+{
+	struct ip_vs_estimator *est, **pest;
+	int killed = 0;
+
+	write_lock_bh(&est_lock);
+	pest = &est_list;
+	while ((est=*pest) != NULL) {
+		if (est->stats != stats) {
+			pest = &est->next;
+			continue;
+		}
+		*pest = est->next;
+		kfree(est);
+		killed++;
+	}
+	if (killed && est_list == NULL)
+		del_timer_sync(&est_timer);
+	write_unlock_bh(&est_lock);
+}
+
+void ip_vs_zero_estimator(struct ip_vs_stats *stats)
+{
+	struct ip_vs_estimator *e;
+
+	write_lock_bh(&est_lock);
+	for (e = est_list; e; e = e->next) {
+		if (e->stats != stats)
+			continue;
+
+		/* set counters zero */
+		e->last_conns = 0;
+		e->last_inpkts = 0;
+		e->last_outpkts = 0;
+		e->last_inbytes = 0;
+		e->last_outbytes = 0;
+		e->cps = 0;
+		e->inpps = 0;
+		e->outpps = 0;
+		e->inbps = 0;
+		e->outbps = 0;
+	}
+	write_unlock_bh(&est_lock);
+}

net/ipv4/ipvs/ip_vs_ftp.c

+/*
+ * ip_vs_ftp.c: IPVS ftp application module
+ *
+ * Version:	$Id: ip_vs_ftp.c,v 1.13 2002/09/15 08:14:08 wensong Exp $
+ *
+ * Authors:	Wensong Zhang <wensong@linuxvirtualserver.org>
+ *
+ * Changes:
+ *
+ *
+ *	This program is free software; you can redistribute it and/or
+ *	modify it under the terms of the GNU General Public License
+ *	as published by the Free Software Foundation; either version
+ *	2 of the License, or (at your option) any later version.
+ *
+ * Most code here is taken from ip_masq_ftp.c in kernel 2.2. The difference
+ * is that ip_vs_ftp module handles the reverse direction to ip_masq_ftp.
+ *
+ *		IP_MASQ_FTP ftp masquerading module
+ *
+ * Version:	@(#)ip_masq_ftp.c 0.04   02/05/96
+ *
+ * Author:	Wouter Gadeyne
+ *
+ */
+
+#include <linux/config.h>
+#include <linux/module.h>
+#include <asm/system.h>
+#include <linux/types.h>
+#include <linux/kernel.h>
+#include <linux/skbuff.h>
+#include <linux/in.h>
+#include <linux/ip.h>
+#include <linux/init.h>
+#include <net/protocol.h>
+#include <net/tcp.h>
+
+#include <net/ip_vs.h>
+
+
+#define SERVER_STRING "227 Entering Passive Mode ("
+#define CLIENT_STRING "PORT "
+
+
+/*
+ * List of ports (up to IP_VS_APP_MAX_PORTS) to be handled by helper
+ * First port is set to the default port.
+ */
+static int ports[IP_VS_APP_MAX_PORTS] = {21, 0};
+
+/*
+ *	Debug level
+ */
+#ifdef CONFIG_IP_VS_DEBUG
+static int debug=0;
+MODULE_PARM(debug, "i");
+#endif
+
+MODULE_PARM(ports, "1-" __MODULE_STRING(IP_VS_APP_MAX_PORTS) "i");
+
+/*	Dummy variable */
+static int ip_vs_ftp_pasv;
+
+
+static int
+ip_vs_ftp_init_conn(struct ip_vs_app *app, struct ip_vs_conn *cp)
+{
+	return 0;
+}
+
+
+static int
+ip_vs_ftp_done_conn(struct ip_vs_app *app, struct ip_vs_conn *cp)
+{
+	return 0;
+}
+
+
+/*
+ * Get <addr,port> from the string "xxx.xxx.xxx.xxx,ppp,ppp", started
+ * with the "pattern" and terminated with the "term" character.
+ * <addr,port> is in network order.
+ */
+static int ip_vs_ftp_get_addrport(char *data, char *data_limit,
+				  const char *pattern, size_t plen, char term,
+				  __u32 *addr, __u16 *port,
+				  char **start, char **end)
+{
+	unsigned char p1,p2,p3,p4,p5,p6;
+
+	while (data < data_limit) {
+		if (strnicmp(data, pattern, plen) != 0) {
+			data++;
+			continue;
+		}
+		*start = data+plen;
+		p1 = simple_strtoul(data+plen, &data, 10);
+		if (*data != ',')
+			continue;
+		p2 = simple_strtoul(data+1, &data, 10);
+		if (*data != ',')
+			continue;
+		p3 = simple_strtoul(data+1, &data, 10);
+		if (*data != ',')
+			continue;
+		p4 = simple_strtoul(data+1, &data, 10);
+		if (*data != ',')
+			continue;
+		p5 = simple_strtoul(data+1, &data, 10);
+		if (*data != ',')
+			continue;
+		p6 = simple_strtoul(data+1, &data, 10);
+		if (*data != term)
+			continue;
+
+		*end = data;
+		*addr = (p4<<24) | (p3<<16) | (p2<<8) | p1;
+		*port = (p6<<8) | p5;
+		return 1;
+	}
+	return 0;
+}
+
+
+/*
+ * Look at outgoing ftp packets to catch the response to a PASV command
+ * from the server (inside-to-outside).
+ * When we see one, we build a connection entry with the client address,
+ * client port 0 (unknown at the moment), the server address and the
+ * server port.  Mark the current connection entry as a control channel
+ * of the new entry. All this work is just to make the data connection
+ * can be scheduled to the right server later.
+ *
+ * The outgoing packet should be something like
+ *   "227 Entering Passive Mode (xxx,xxx,xxx,xxx,ppp,ppp)".
+ * xxx,xxx,xxx,xxx is the server address, ppp,ppp is the server port number.
+ */
+static int ip_vs_ftp_out(struct ip_vs_app *app,
+			 struct ip_vs_conn *cp, struct sk_buff *skb)
+{
+	struct iphdr *iph;
+	struct tcphdr *th;
+	char *data, *data_limit;
+	char *start, *end;
+	__u32 from;
+	__u16 port;
+	struct ip_vs_conn *n_cp;
+	char buf[24];		/* xxx.xxx.xxx.xxx,ppp,ppp\000 */
+	unsigned buf_len;
+	int diff;
+
+	/* Only useful for established sessions */
+	if (cp->state != IP_VS_TCP_S_ESTABLISHED)
+		return 0;
+
+	if (cp->app_data == &ip_vs_ftp_pasv) {
+		iph = skb->nh.iph;
+		th = (struct tcphdr *)&(((char *)iph)[iph->ihl*4]);
+		data = (char *)th + (th->doff << 2);
+		data_limit = skb->tail;
+
+		if (ip_vs_ftp_get_addrport(data, data_limit,
+					   SERVER_STRING,
+					   sizeof(SERVER_STRING)-1, ')',
+					   &from, &port,
+					   &start, &end) == 0)
+			return 0;
+
+		IP_VS_DBG(1-debug, "PASV response (%u.%u.%u.%u:%d) -> "
+			  "%u.%u.%u.%u:%d detected\n",
+			  NIPQUAD(from), ntohs(port), NIPQUAD(cp->caddr), 0);
+
+		/*
+		 * Now update or create an connection entry for it
+		 */
+		n_cp = ip_vs_conn_out_get(iph->protocol, from, port,
+					  cp->caddr, 0);
+		if (!n_cp) {
+			n_cp = ip_vs_conn_new(IPPROTO_TCP,
+					      cp->caddr, 0,
+					      cp->vaddr, port,
+					      from, port,
+					      IP_VS_CONN_F_NO_CPORT,
+					      cp->dest);
+			if (!n_cp)
+				return 0;
+
+			/* add its controller */
+			ip_vs_control_add(n_cp, cp);
+		}
+
+		/*
+		 * Replace the old passive address with the new one
+		 */
+		from = n_cp->vaddr;
+		port = n_cp->vport;
+		sprintf(buf,"%d,%d,%d,%d,%d,%d", NIPQUAD(from),
+			port&255, port>>8&255);
+		buf_len = strlen(buf);
+
+		/*
+		 * Calculate required delta-offset to keep TCP happy
+		 */
+		diff = buf_len - (end-start);
+
+		if (diff == 0) {
+			/* simply replace it with new passive address */
+			memcpy(start, buf, buf_len);
+		} else {
+			/* fixme: return value isn't checked here */
+			ip_vs_skb_replace(skb, GFP_ATOMIC, start,
+					  end-start, buf, buf_len);
+		}
+
+		cp->app_data = NULL;
+		ip_vs_tcp_conn_listen(n_cp);
+		ip_vs_conn_put(n_cp);
+		return diff;
+	}
+	return 0;
+}
+
+
+/*
+ * Look at incoming ftp packets to catch the PASV/PORT command
+ * (outside-to-inside).
+ *
+ * The incoming packet having the PORT command should be something like
+ *      "PORT xxx,xxx,xxx,xxx,ppp,ppp\n".
+ * xxx,xxx,xxx,xxx is the client address, ppp,ppp is the client port number.
+ * In this case, we create a connection entry using the client address and
+ * port, so that the active ftp data connection from the server can reach
+ * the client.
+ */
+static int ip_vs_ftp_in(struct ip_vs_app *app,
+			struct ip_vs_conn *cp, struct sk_buff *skb)
+{
+	struct iphdr *iph;
+	struct tcphdr *th;
+	char *data, *data_start, *data_limit;
+	char *start, *end;
+	__u32 to;
+	__u16 port;
+	struct ip_vs_conn *n_cp;
+
+	/* Only useful for established sessions */
+	if (cp->state != IP_VS_TCP_S_ESTABLISHED)
+		return 0;
+
+	/*
+	 * Detecting whether it is passive
+	 */
+	iph = skb->nh.iph;
+	th = (struct tcphdr *)&(((char *)iph)[iph->ihl*4]);
+
+	/* Since there may be OPTIONS in the TCP packet and the HLEN is
+	   the length of the header in 32-bit multiples, it is accurate
+	   to calculate data address by th+HLEN*4 */
+	data = data_start = (char *)th + (th->doff << 2);
+	data_limit = skb->tail;
+
+	while (data < data_limit) {
+		if (strnicmp(data, "PASV\r\n", 6) == 0) {
+			IP_VS_DBG(1-debug, "got PASV at %d of %d\n",
+				  data - data_start,
+				  data_limit - data_start);
+			cp->app_data = &ip_vs_ftp_pasv;
+			return 0;
+		}
+		data++;
+	}
+
+	/*
+	 * To support virtual FTP server, the scenerio is as follows:
+	 *       FTP client ----> Load Balancer ----> FTP server
+	 * First detect the port number in the application data,
+	 * then create a new connection entry for the coming data
+	 * connection.
+	 */
+	data = data_start;
+	data_limit = skb->h.raw + skb->len - 18;
+
+	if (ip_vs_ftp_get_addrport(data, data_limit,
+				   CLIENT_STRING, sizeof(CLIENT_STRING)-1,
+				   '\r', &to, &port,
+				   &start, &end) == 0)
+		return 0;
+
+	IP_VS_DBG(1-debug, "PORT %u.%u.%u.%u:%d detected\n",
+		  NIPQUAD(to), ntohs(port));
+
+	/*
+	 * Now update or create a connection entry for it
+	 */
+	IP_VS_DBG(1-debug, "protocol %s %u.%u.%u.%u:%d %u.%u.%u.%u:%d\n",
+		  ip_vs_proto_name(iph->protocol),
+		  NIPQUAD(to), ntohs(port), NIPQUAD(iph->daddr), 0);
+
+	n_cp = ip_vs_conn_in_get(iph->protocol,
+				 to, port,
+				 iph->daddr, htons(ntohs(cp->vport)-1));
+	if (!n_cp) {
+		n_cp = ip_vs_conn_new(IPPROTO_TCP,
+				      to, port,
+				      cp->vaddr, htons(ntohs(cp->vport)-1),
+				      cp->daddr, htons(ntohs(cp->dport)-1),
+				      0,
+				      cp->dest);
+		if (!n_cp)
+			return 0;
+
+		/* add its controller */
+		ip_vs_control_add(n_cp, cp);
+	}
+
+	/*
+	 *	Move tunnel to listen state
+	 */
+	ip_vs_tcp_conn_listen(n_cp);
+	ip_vs_conn_put(n_cp);
+
+	/* no diff required for incoming packets */
+	return 0;
+}
+
+
+static struct ip_vs_app ip_vs_ftp = {
+	.name =		"ftp",
+	.type =		IP_VS_APP_TYPE_FTP,
+	.protocol =	IPPROTO_TCP,
+	.module =	THIS_MODULE,
+	.incs_list =	LIST_HEAD_INIT(ip_vs_ftp.incs_list),
+	.init_conn =	ip_vs_ftp_init_conn,
+	.done_conn =	ip_vs_ftp_done_conn,
+	.bind_conn =	NULL,
+	.unbind_conn =	NULL,
+	.pkt_out =	ip_vs_ftp_out,
+	.pkt_in =	ip_vs_ftp_in,
+};
+
+
+/*
+ *	ip_vs_ftp initialization
+ */
+static int __init ip_vs_ftp_init(void)
+{
+	int i, ret;
+	struct ip_vs_app *app = &ip_vs_ftp;
+
+	ret = register_ip_vs_app(app);
+	if (ret)
+		return ret;
+
+	for (i=0; i<IP_VS_APP_MAX_PORTS; i++) {
+		if (!ports[i])
+			continue;
+		ret = register_ip_vs_app_inc(app, app->protocol, ports[i]);
+		if (ret)
+			break;
+		IP_VS_DBG(1-debug, "%s: loaded support on port[%d] = %d\n",
+			  app->name, i, ports[i]);
+	}
+
+	if (ret)
+		unregister_ip_vs_app(app);
+
+	return ret;
+}
+
+
+/*
+ *	ip_vs_ftp finish.
+ */
+static void __exit ip_vs_ftp_exit(void)
+{
+	unregister_ip_vs_app(&ip_vs_ftp);
+}
+
+
+module_init(ip_vs_ftp_init);
+module_exit(ip_vs_ftp_exit);
+MODULE_LICENSE("GPL");

net/ipv4/netfilter/ip_fw_compat.c

@@ -19,6 +19,12 @@ struct notifier_block;
 
 static struct firewall_ops *fwops;
 
+#ifdef CONFIG_IP_VS
+/* From ip_vs_core.c */
+extern unsigned int
+check_for_ip_vs_out(struct sk_buff **skb_p, int (*okfn)(struct sk_buff *));
+#endif
+
 /* They call these; we do what they want. */
 int register_firewall(int pf, struct firewall_ops *fw)
 {
@@ -134,8 +140,14 @@ fw_in(unsigned int hooknum,
 		return NF_ACCEPT;
 
 	case FW_MASQUERADE:
-		if (hooknum == NF_IP_FORWARD)
+		if (hooknum == NF_IP_FORWARD) {
+#ifdef CONFIG_IP_VS
+			/* check if it is for ip_vs */
+			if (check_for_ip_vs_out(pskb, okfn) == NF_STOLEN)
+				return NF_STOLEN;
+#endif
 			return do_masquerade(pskb, out);
+		}
 		else return NF_ACCEPT;
 
 	case FW_REDIRECT:

net/netsyms.c

@@ -265,6 +265,7 @@ EXPORT_SYMBOL(inet_family_ops);
 EXPORT_SYMBOL(in_aton);
 EXPORT_SYMBOL(ip_mc_inc_group);
 EXPORT_SYMBOL(ip_mc_dec_group);
+EXPORT_SYMBOL(ip_mc_join_group);
 EXPORT_SYMBOL(ip_finish_output);
 EXPORT_SYMBOL(inet_stream_ops);
 EXPORT_SYMBOL(inet_dgram_ops);