route.generic

diff -ru -x*~ linux-lt-2.3.99-pre3.prev/Documentation/networking/ip-sysctl.txt linux-lt-2.3.99-pre3/Documentation/networking/ip-sysctl.txt
--- linux-lt-2.3.99-pre3.prev/Documentation/networking/ip-sysctl.txt	Sun Jan 23 03:54:56 2000
+++ linux-lt-2.3.99-pre3/Documentation/networking/ip-sysctl.txt	Tue Mar 28 19:40:59 2000
@@ -262,13 +262,21 @@
 	Do proxy arp.
 
 shared_media - BOOLEAN
-	Send(router) or accept(host) RFC1620 shared media redirects.
+	Do not check the new gateway specified in incoming ICMP redirect
+	messages for belonging to a directly attached network (i.e. the
+	routing table has for this address an entry pointing to the given
+	device, doesn't have a gateway, and with scope not wider SCOPE_LINK).
+	If this variable is TRUE then new gateways are only checked for being a
+	unicast addresses.  If it is FALSE then the full check described
+	above is performed.  See RFC1620 for background information about
+	shared media.
 	Overrides ip_secure_redirects.
 	default TRUE
 
 secure_redirects - BOOLEAN
-	Accept ICMP redirect messages only for gateways,
-	listed in default gateway list.
+	Accept ICMP redirect messages only for gateways already listed as
+	gateways in the routing tables.  This check is performed only when
+	`shared_media' is FALSE.
 	default TRUE
 
 send_redirects - BOOLEAN
@@ -287,6 +295,19 @@
 	default TRUE (router)
 		FALSE (host)
 
+source_check - BOOLEAN
+	Check source address for outgoing packets.
+	If source_check is turned on all outgoing packets (including going
+	through a loopback interface) are checked for the source address
+	being local.  An address is considered as local for this purposes if
+	a route lookup in the opposite direction (i.e. with source and
+	destination addresses being reversed) gives a unicast local route
+	entry.
+	Note: source addresses are always checked for being not a multicast,
+	limited broadcast, zero net, or loopback (for non-loopback
+	interfaces) independetly of the setting of the option.
+	default TRUE
+
 rp_filter - BOOLEAN
 	1 - do source validation by reversed path, as specified in RFC1812
 	    Recommended option for single homed hosts and stub network
@@ -305,4 +326,8 @@
 Updated by:
 Andi Kleen
 ak@muc.de
+
+Andrey Savochkin
+saw@msu.ru
+
 $Id: route.generic,v 1.1.1.1 2002/09/09 14:51:13 imagestream Exp $
diff -ru -x*~ linux-lt-2.3.99-pre3.prev/include/linux/in_route.h linux-lt-2.3.99-pre3/include/linux/in_route.h
--- linux-lt-2.3.99-pre3.prev/include/linux/in_route.h	Fri Jun 12 13:52:33 1998
+++ linux-lt-2.3.99-pre3/include/linux/in_route.h	Tue Mar 28 19:39:49 2000
@@ -4,6 +4,7 @@
 /* IPv4 routing cache flags */
 
 #define RTCF_DEAD	RTNH_F_DEAD
+#define RTCF_PERVASIVE	RTNH_F_PERVASIVE
 #define RTCF_ONLINK	RTNH_F_ONLINK
 
 /* Obsolete flag. About to be deleted */
diff -ru -x*~ linux-lt-2.3.99-pre3.prev/include/linux/inetdevice.h linux-lt-2.3.99-pre3/include/linux/inetdevice.h
--- linux-lt-2.3.99-pre3.prev/include/linux/inetdevice.h	Tue Aug 24 01:01:02 1999
+++ linux-lt-2.3.99-pre3/include/linux/inetdevice.h	Tue Mar 28 19:39:49 2000
@@ -9,6 +9,7 @@
 	int	send_redirects;
 	int	secure_redirects;
 	int	shared_media;
+	int	source_check;
 	int	accept_source_route;
 	int	rp_filter;
 	int	proxy_arp;
@@ -46,6 +47,7 @@
 #define IN_DEV_SHARED_MEDIA(in_dev)	(ipv4_devconf.shared_media || (in_dev)->cnf.shared_media)
 #define IN_DEV_TX_REDIRECTS(in_dev)	(ipv4_devconf.send_redirects || (in_dev)->cnf.send_redirects)
 #define IN_DEV_SEC_REDIRECTS(in_dev)	(ipv4_devconf.secure_redirects || (in_dev)->cnf.secure_redirects)
+#define IN_DEV_SRC_CHECK(in_dev)	(ipv4_devconf.source_check || (in_dev)->cnf.source_check)
 #define IN_DEV_IDTAG(in_dev)		((in_dev)->cnf.tag)
 
 #define IN_DEV_RX_REDIRECTS(in_dev) \
@@ -73,7 +75,6 @@
 extern int unregister_inetaddr_notifier(struct notifier_block *nb);
 
 extern struct net_device 	*ip_dev_find(u32 addr);
-extern int		inet_addr_onlink(struct in_device *in_dev, u32 a, u32 b);
 extern int		devinet_ioctl(unsigned int cmd, void *);
 extern void		devinet_init(void);
 extern struct in_device *inetdev_init(struct net_device *dev);
diff -ru -x*~ linux-lt-2.3.99-pre3.prev/include/linux/rtnetlink.h linux-lt-2.3.99-pre3/include/linux/rtnetlink.h
--- linux-lt-2.3.99-pre3.prev/include/linux/rtnetlink.h	Thu Feb 10 12:08:09 2000
+++ linux-lt-2.3.99-pre3/include/linux/rtnetlink.h	Tue Mar 28 19:39:49 2000
@@ -224,9 +224,11 @@
 
 /* rtnh_flags */
 
-#define RTNH_F_DEAD		1	/* Nexthop is dead (used by multipath)	*/
-#define RTNH_F_PERVASIVE	2	/* Do recursive gateway lookup	*/
-#define RTNH_F_ONLINK		4	/* Gateway is forced on link	*/
+#define RTNH_F_DEAD		0x01	/* Nexthop is dead (used by multipath)	*/
+#define RTNH_F_PERVASIVE	0x02	/* Omit gateway & pref_src test	*/
+#define RTNH_F_ONLINK		0x04	/* Gateway is forced on link	*/
+#define RTNH_F_GLUE		0x08	/* Nexthop is glued		*/
+#define RTNH_F_USEFIRST		0x10	/* Use only it (for multipath)	*/
 
 /* Macros to handle hexthops */
 
diff -ru -x*~ linux-lt-2.3.99-pre3.prev/include/linux/sysctl.h linux-lt-2.3.99-pre3/include/linux/sysctl.h
--- linux-lt-2.3.99-pre3.prev/include/linux/sysctl.h	Thu Mar  9 01:16:24 2000
+++ linux-lt-2.3.99-pre3/include/linux/sysctl.h	Tue Mar 28 19:39:49 2000
@@ -302,7 +302,8 @@
 	NET_IPV4_CONF_ACCEPT_SOURCE_ROUTE=9,
 	NET_IPV4_CONF_BOOTP_RELAY=10,
 	NET_IPV4_CONF_LOG_MARTIANS=11,
-	NET_IPV4_CONF_TAG=12
+	NET_IPV4_CONF_TAG=12,
+	NET_IPV4_CONF_SRC_CHECK=13,
 };
 
 /* /proc/sys/net/ipv6 */
diff -ru -x*~ linux-lt-2.3.99-pre3.prev/include/net/ip_fib.h linux-lt-2.3.99-pre3/include/net/ip_fib.h
--- linux-lt-2.3.99-pre3.prev/include/net/ip_fib.h	Tue Aug 24 01:01:02 1999
+++ linux-lt-2.3.99-pre3/include/net/ip_fib.h	Tue Mar 28 19:39:49 2000
@@ -217,7 +217,8 @@
 extern int fib_dump_info(struct sk_buff *skb, u32 pid, u32 seq, int event,
 			 u8 tb_id, u8 type, u8 scope, void *dst, int dst_len, u8 tos,
 			 struct fib_info *fi);
-extern int fib_sync_down(u32 local, struct net_device *dev, int force);
+extern int fib_sync_addr_down(u32 local);
+extern int fib_sync_dev_down(struct net_device *dev, int force);
 extern int fib_sync_up(struct net_device *dev);
 extern int fib_convert_rtentry(int cmd, struct nlmsghdr *nl, struct rtmsg *rtm,
 			       struct kern_rta *rta, struct rtentry *r);
diff -ru -x*~ linux-lt-2.3.99-pre3.prev/include/net/route.h linux-lt-2.3.99-pre3/include/net/route.h
--- linux-lt-2.3.99-pre3.prev/include/net/route.h	Sun Mar 19 04:11:22 2000
+++ linux-lt-2.3.99-pre3/include/net/route.h	Tue Mar 28 19:39:49 2000
@@ -106,6 +106,9 @@
 extern void		ip_rt_send_redirect(struct sk_buff *skb);
 
 extern unsigned		inet_addr_type(u32 addr);
+extern int		inet_addr_onlink(struct net_device *, u32 dst, u32 src, u8 tos);
+extern int		fib_local_source(u32 saddr, u32 daddr, u8 tos, struct net_device *);
+extern u32		fib_select_addr(struct net_device *, u32 dst, int scope);
 extern void		ip_rt_multicast_event(struct in_device *);
 extern int		ip_rt_ioctl(unsigned int cmd, void *arg);
 extern void		ip_rt_get_source(u8 *src, struct rtable *rt);
diff -ru -x*~ linux-lt-2.3.99-pre3.prev/net/ipv4/af_inet.c linux-lt-2.3.99-pre3/net/ipv4/af_inet.c
--- linux-lt-2.3.99-pre3.prev/net/ipv4/af_inet.c	Tue Feb 22 09:35:06 2000
+++ linux-lt-2.3.99-pre3/net/ipv4/af_inet.c	Tue Mar 28 19:43:30 2000
@@ -463,6 +463,15 @@
 		return -EINVAL;
 		
 	chk_addr_ret = inet_addr_type(addr->sin_addr.s_addr);
+	/* The source address check is omitted here.
+	 * We may allow to bind sockets to any address for listening purposes.
+	 * Such sockets will get only those packets which were considered as
+	 * "local" by routing (i.e. configured to go locally by the
+	 * administrator).
+	 * Outgoing packets are checked by output routing (see
+	 * ip_route_output_slow and outrt_check_src in net/ipv4/route.c).
+	 * 1999/11/13  SAW
+	 */
 
 	snum = ntohs(addr->sin_port);
 	if (snum && snum < PROT_SOCK && !capable(CAP_NET_BIND_SERVICE))
diff -ru -x*~ linux-lt-2.3.99-pre3.prev/net/ipv4/arp.c linux-lt-2.3.99-pre3/net/ipv4/arp.c
--- linux-lt-2.3.99-pre3.prev/net/ipv4/arp.c	Sun Jan 23 03:54:57 2000
+++ linux-lt-2.3.99-pre3/net/ipv4/arp.c	Tue Mar 28 19:39:50 2000
@@ -333,10 +333,11 @@
 	u32 target = *(u32*)neigh->primary_key;
 	int probes = atomic_read(&neigh->probes);
 
-	if (skb && inet_addr_type(skb->nh.iph->saddr) == RTN_LOCAL)
+	if (skb && fib_local_source(skb->nh.iph->saddr, target,
+			skb->nh.iph->tos, dev) == 0)
 		saddr = skb->nh.iph->saddr;
 	else
-		saddr = inet_select_addr(dev, target, RT_SCOPE_LINK);
+		saddr = fib_select_addr(dev, target, RT_SCOPE_LINK);
 
 	if ((probes -= neigh->parms->ucast_probes) < 0) {
 		if (!(neigh->nud_state&NUD_VALID))
diff -ru -x*~ linux-lt-2.3.99-pre3.prev/net/ipv4/devinet.c linux-lt-2.3.99-pre3/net/ipv4/devinet.c
--- linux-lt-2.3.99-pre3.prev/net/ipv4/devinet.c	Sun Jan  9 13:36:20 2000
+++ linux-lt-2.3.99-pre3/net/ipv4/devinet.c	Tue Mar 28 19:39:50 2000
@@ -58,8 +58,8 @@
 #include <net/route.h>
 #include <net/ip_fib.h>
 
-struct ipv4_devconf ipv4_devconf = { 1, 1, 1, 1, 0, };
-static struct ipv4_devconf ipv4_devconf_dflt = { 1, 1, 1, 1, 1, };
+struct ipv4_devconf ipv4_devconf = { 1, 1, 1, 1, 1, 0, };
+static struct ipv4_devconf ipv4_devconf_dflt = { 1, 1, 1, 1, 1, 1, };
 
 #ifdef CONFIG_RTNETLINK
 static void rtmsg_ifa(int event, struct in_ifaddr *);
@@ -186,21 +186,6 @@
 	in_dev_put(in_dev);
 }
 
-int inet_addr_onlink(struct in_device *in_dev, u32 a, u32 b)
-{
-	read_lock(&in_dev->lock);
-	for_primary_ifa(in_dev) {
-		if (inet_ifa_match(a, ifa)) {
-			if (!b || inet_ifa_match(b, ifa)) {
-				read_unlock(&in_dev->lock);
-				return 1;
-			}
-		}
-	} endfor_ifa(in_dev);
-	read_unlock(&in_dev->lock);
-	return 0;
-} 
-
 static void
 inet_del_ifa(struct in_device *in_dev, struct in_ifaddr **ifap, int destroy)
 {
@@ -1027,7 +1012,7 @@
 static struct devinet_sysctl_table
 {
 	struct ctl_table_header *sysctl_header;
-	ctl_table devinet_vars[13];
+	ctl_table devinet_vars[14];
 	ctl_table devinet_dev[2];
 	ctl_table devinet_conf_dir[2];
 	ctl_table devinet_proto_dir[2];
@@ -1066,6 +1051,9 @@
          &proc_dointvec},
         {NET_IPV4_CONF_LOG_MARTIANS, "log_martians",
          &ipv4_devconf.log_martians, sizeof(int), 0644, NULL,
+         &proc_dointvec},
+        {NET_IPV4_CONF_SRC_CHECK, "source_check",
+         &ipv4_devconf.source_check, sizeof(int), 0644, NULL,
          &proc_dointvec},
 	{NET_IPV4_CONF_TAG, "tag",
 	 &ipv4_devconf.tag, sizeof(int), 0644, NULL,
diff -ru -x*~ linux-lt-2.3.99-pre3.prev/net/ipv4/fib_frontend.c linux-lt-2.3.99-pre3/net/ipv4/fib_frontend.c
--- linux-lt-2.3.99-pre3.prev/net/ipv4/fib_frontend.c	Thu Dec 23 11:55:38 1999
+++ linux-lt-2.3.99-pre3/net/ipv4/fib_frontend.c	Tue Mar 28 19:39:50 2000
@@ -30,6 +30,7 @@
 #include <linux/in.h>
 #include <linux/inet.h>
 #include <linux/netdevice.h>
+#include <linux/inetdevice.h>
 #include <linux/if_arp.h>
 #include <linux/proc_fs.h>
 #include <linux/skbuff.h>
@@ -168,11 +169,31 @@
 	return dev;
 }
 
+int fib_local_source(u32 saddr, u32 daddr, u8 tos, struct net_device *dev)
+{
+	struct rt_key		key;
+	struct fib_result	res;
+
+	memset(&key, 0, sizeof(key));
+	key.src = daddr;
+	key.dst = saddr;
+	key.tos = tos;
+	key.iif = dev->ifindex;
+	if (fib_lookup(&key, &res) == 0) {
+		unsigned ret;
+		ret = res.type;
+		fib_res_put(&res);
+		if (ret != RTN_LOCAL)
+			return -EINVAL;
+	}
+	return 0;
+}
+
 unsigned inet_addr_type(u32 addr)
 {
 	struct rt_key		key;
 	struct fib_result	res;
-	unsigned ret = RTN_BROADCAST;
+	unsigned		ret;
 
 	if (ZERONET(addr) || BADCLASS(addr))
 		return RTN_BROADCAST;
@@ -180,21 +201,57 @@
 		return RTN_MULTICAST;
 
 	memset(&key, 0, sizeof(key));
+	key.src = addr;
 	key.dst = addr;
-#ifdef CONFIG_IP_MULTIPLE_TABLES
-	res.r = NULL;
-#endif
 	
-	if (local_table) {
-		ret = RTN_UNICAST;
-		if (local_table->tb_lookup(local_table, &key, &res) == 0) {
-			ret = res.type;
-			fib_res_put(&res);
-		}
+	ret = RTN_UNICAST;
+	if (fib_lookup(&key, &res) == 0) {
+		ret = res.type;
+		fib_res_put(&res);
 	}
 	return ret;
 }
 
+u32 fib_select_addr(struct net_device *dev, u32 dst, int scope)
+{
+	struct rt_key		key;
+	struct fib_result	res;
+	u32			ret;
+
+	memset(&key, 0, sizeof(key));
+	key.src = dst;
+	key.dst = dst;
+	key.oif = dev->ifindex;
+	key.scope = scope;
+	
+	if (fib_lookup(&key, &res) == 0) {
+		ret = FIB_RES_PREFSRC(res);
+		fib_res_put(&res);
+	} else
+		ret = inet_select_addr(dev, dst, scope);
+	return ret;
+}
+
+/* Check if dst is a UNICAST address and reachable via device dev */
+int inet_addr_onlink(struct net_device *dev, u32 dst, u32 src, u8 tos)
+{
+	struct rt_key key;
+	struct fib_result res;
+	int ret;
+
+	key.src = src;
+	key.dst = dst;
+	key.tos = tos;
+	key.iif = 0;
+	key.oif = 0;
+	key.scope = RT_SCOPE_LINK;
+	if (fib_lookup(&key, &res) != 0)
+		return 0;
+	ret = (res.type == RTN_UNICAST && FIB_RES_DEV(res) == dev);
+	fib_res_put(&res);
+	return ret;
+}
+
 /* Given (packet source, input interface) and optional (dst, oif, tos):
    - (main) check, that source is valid i.e. not broadcast or our local
      address.
@@ -559,7 +616,7 @@
 			   First of all, we scan fib_info list searching
 			   for stray nexthop entries, then ignite fib_flush.
 			*/
-			if (fib_sync_down(ifa->ifa_local, NULL, 0))
+			if (fib_sync_addr_down(ifa->ifa_local))
 				fib_flush();
 		}
 	}
@@ -571,7 +628,7 @@
 
 static void fib_disable_ip(struct net_device *dev, int force)
 {
-	if (fib_sync_down(0, dev, force))
+	if (fib_sync_dev_down(dev, force))
 		fib_flush();
 	rt_cache_flush(0);
 	arp_ifdown(dev);
@@ -591,8 +648,10 @@
 			/* Last address was deleted from this interface.
 			   Disable IP.
 			 */
+			printk("fib_inetaddr_event: dev down, fib_disable_ip(1)\n");
 			fib_disable_ip(ifa->ifa_dev->dev, 1);
 		} else {
+			printk("fib_inetaddr_event: dev down, fib_del_ifaddr\n");
 			fib_del_ifaddr(ifa);
 			rt_cache_flush(-1);
 		}
@@ -606,11 +665,10 @@
 	struct net_device *dev = ptr;
 	struct in_device *in_dev = __in_dev_get(dev);
 
-	if (!in_dev)
-		return NOTIFY_DONE;
-
 	switch (event) {
 	case NETDEV_UP:
+		if (!in_dev)
+			break;
 		for_ifa(in_dev) {
 			fib_add_ifaddr(ifa);
 		} endfor_ifa(in_dev);
@@ -620,9 +678,18 @@
 		rt_cache_flush(-1);
 		break;
 	case NETDEV_DOWN:
+		printk("fib_netdev_event: dev down, fib_disable_ip(0)\n");
 		fib_disable_ip(dev, 0);
 		break;
 	case NETDEV_UNREGISTER:
+		/* Routes pointing to dev may still exists even when IP has
+		 * been shut down.  It may happen because routes of local type
+		 * has special nexthop (see fib_create_info() and
+		 * fib_sync_dev_down()).  I don't know if this state is valid.
+		 * Now I call fib_disable_ip() independently from if the device
+		 * has IP because otherwise stale device references are left.
+		 * 1999/11/28  SAW */
+		printk("fib_netdev_event: dev unregister, fib_disable_ip(1)\n");
 		fib_disable_ip(dev, 1);
 		break;
 	case NETDEV_CHANGEMTU:
diff -ru -x*~ linux-lt-2.3.99-pre3.prev/net/ipv4/fib_semantics.c linux-lt-2.3.99-pre3/net/ipv4/fib_semantics.c
--- linux-lt-2.3.99-pre3.prev/net/ipv4/fib_semantics.c	Tue Aug 24 01:01:02 1999
+++ linux-lt-2.3.99-pre3/net/ipv4/fib_semantics.c	Tue Mar 28 19:39:50 2000
@@ -345,14 +345,15 @@
 {
 	int err;
 
+	if (nh->nh_flags&RTNH_F_PERVASIVE) {
+		fi->fib_flags |= RTCF_PERVASIVE;
+		return 0;
+	}
+
 	if (nh->nh_gw) {
 		struct rt_key key;
 		struct fib_result res;
 
-#ifdef CONFIG_IP_ROUTE_PERVASIVE
-		if (nh->nh_flags&RTNH_F_PERVASIVE)
-			return 0;
-#endif
 		if (nh->nh_flags&RTNH_F_ONLINK) {
 			struct net_device *dev;
 
@@ -389,7 +390,7 @@
 	} else {
 		struct in_device *in_dev;
 
-		if (nh->nh_flags&(RTNH_F_PERVASIVE|RTNH_F_ONLINK))
+		if (nh->nh_flags&RTNH_F_ONLINK)
 			return -EINVAL;
 
 		in_dev = inetdev_by_index(nh->nh_oif);
@@ -528,7 +529,7 @@
 		} endfor_nexthops(fi)
 	}
 
-	if (fi->fib_prefsrc) {
+	if (fi->fib_prefsrc && !(fi->fib_flags&RTCF_PERVASIVE)) {
 		if (r->rtm_type != RTN_LOCAL || rta->rta_dst == NULL ||
 		    memcmp(&fi->fib_prefsrc, rta->rta_dst, 4))
 			if (inet_addr_type(fi->fib_prefsrc) != RTN_LOCAL)
@@ -857,19 +858,34 @@
    - device went down -> we must shutdown all nexthops going via it.
  */
 
-int fib_sync_down(u32 local, struct net_device *dev, int force)
+int fib_sync_addr_down(u32 local)
+{
+	int ret = 0;
+
+	if (!local)
+		goto out;
+
+	for_fib_info() {
+		if (fi->fib_prefsrc == local &&
+				!(fi->fib_flags&RTCF_PERVASIVE)) {
+			fi->fib_flags |= RTCF_DEAD;
+			ret++;
+		}
+	} endfor_fib_info();
+out:
+	return ret;
+}
+
+int fib_sync_dev_down(struct net_device *dev, int force)
 {
 	int ret = 0;
 	int scope = RT_SCOPE_NOWHERE;
-	
+
 	if (force)
 		scope = -1;
 
 	for_fib_info() {
-		if (local && fi->fib_prefsrc == local) {
-			fi->fib_flags |= RTNH_F_DEAD;
-			ret++;
-		} else if (dev && fi->fib_nhs) {
+		if (fi->fib_nhs) {
 			int dead = 0;
 
 			change_nexthops(fi) {
@@ -886,7 +902,7 @@
 				}
 			} endfor_nexthops(fi)
 			if (dead == fi->fib_nhs) {
-				fi->fib_flags |= RTNH_F_DEAD;
+				fi->fib_flags |= RTCF_DEAD;
 				ret++;
 			}
 		}
@@ -947,6 +963,10 @@
 		int power = 0;
 		change_nexthops(fi) {
 			if (!(nh->nh_flags&RTNH_F_DEAD)) {
+				if (nh->nh_flags&RTNH_F_USEFIRST) {
+					res->nh_sel = nhsel;
+					return;
+				}
 				power += nh->nh_weight;
 				nh->nh_power = nh->nh_weight;
 			}
diff -ru -x*~ linux-lt-2.3.99-pre3.prev/net/ipv4/route.c linux-lt-2.3.99-pre3/net/ipv4/route.c
--- linux-lt-2.3.99-pre3.prev/net/ipv4/route.c	Mon Mar 27 18:25:56 2000
+++ linux-lt-2.3.99-pre3/net/ipv4/route.c	Tue Mar 28 19:57:19 2000
@@ -711,22 +711,28 @@
 	struct rtable *rth, **rthp;
 	u32  skeys[2] = { saddr, 0 };
 	int  ikeys[2] = { dev->ifindex, 0 };
+	char *reason;
 
 	tos &= IPTOS_TOS_MASK;
 
 	if (!in_dev)
 		return;
 
+	reason = "bad gateway";
 	if (new_gw == old_gw || !IN_DEV_RX_REDIRECTS(in_dev)
 	    || MULTICAST(new_gw) || BADCLASS(new_gw) || ZERONET(new_gw))
 		goto reject_redirect;
 
 	if (!IN_DEV_SHARED_MEDIA(in_dev)) {
-		if (!inet_addr_onlink(in_dev, new_gw, old_gw))
+		reason = "gateway not onlink";
+		if (!inet_addr_onlink(dev, new_gw, saddr, tos))
 			goto reject_redirect;
-		if (IN_DEV_SEC_REDIRECTS(in_dev) && ip_fib_check_default(new_gw, dev))
+		reason = "insecure gateway";
+		if (IN_DEV_SEC_REDIRECTS(in_dev) &&
+				ip_fib_check_default(new_gw, dev))
 			goto reject_redirect;
 	} else {
+		reason = "unacceptable gateway";
 		if (inet_addr_type(new_gw) != RTN_UNICAST)
 			goto reject_redirect;
 	}
@@ -816,9 +822,9 @@
 reject_redirect:
 #ifdef CONFIG_IP_ROUTE_VERBOSE
 	if (IN_DEV_LOG_MARTIANS(in_dev) && net_ratelimit())
-		printk(KERN_INFO "Redirect from %X/%s to %X ignored."
+		printk(KERN_INFO "Redirect from %X/%s to %X ignored (%s). "
 		       "Path = %X -> %X, tos %02x\n",
-		       ntohl(old_gw), dev->name, ntohl(new_gw),
+		       ntohl(old_gw), dev->name, ntohl(new_gw), reason,
 		       ntohl(saddr), ntohl(daddr), tos);
 #endif
 	in_dev_put(in_dev);
@@ -836,7 +842,7 @@
 		if ((rt->rt_flags&RTCF_REDIRECTED) || rt->u.dst.expires) {
 			unsigned hash = rt_hash_code(rt->key.dst, rt->key.src^(rt->key.oif<<5), rt->key.tos);
 #if RT_CACHE_DEBUG >= 1
-			printk(KERN_DEBUG "ip_rt_advice: redirect to %d.%d.%d.%d/%02x dropped\n", NIPQUAD(rt->rt_dst), rt->key.tos);
+			printk(KERN_DEBUG "ip_rt_advice: cache entry to %d.%d.%d.%d/%02x dropped\n", NIPQUAD(rt->rt_dst), rt->key.tos);
 #endif
 			rt_del(hash, rt);
 			return NULL;
@@ -1106,6 +1112,10 @@
 }
 #endif
 
+/*********************************************************************
+			 Input/output routing
+ *********************************************************************/
+
 static void rt_set_nexthop(struct rtable *rt, struct fib_result *res, u32 itag)
 {
 	struct fib_info *fi = res->fi;
@@ -1145,6 +1155,10 @@
         rt->rt_type = res->type;
 }
 
+/*********************************************************************
+				Input
+ *********************************************************************/
+
 static int
 ip_route_input_mc(struct sk_buff *skb, u32 daddr, u32 saddr,
 		  u8 tos, struct net_device *dev, int our)
@@ -1362,9 +1376,7 @@
 	if (err)
 		flags |= RTCF_DIRECTSRC;
 
-	if (out_dev == in_dev && err && !(flags&(RTCF_NAT|RTCF_MASQ)) &&
-	    (IN_DEV_SHARED_MEDIA(out_dev)
-	     || inet_addr_onlink(out_dev, saddr, FIB_RES_GW(res))))
+	if (out_dev == in_dev && err && !(flags&(RTCF_NAT|RTCF_MASQ)))
 		flags |= RTCF_DOREDIRECT;
 
 	if (skb->protocol != __constant_htons(ETH_P_IP)) {
@@ -1592,22 +1604,183 @@
 	return ip_route_input_slow(skb, daddr, saddr, tos, dev);
 }
 
+/*********************************************************************
+				Output
+ *********************************************************************/
+
+/* TODO:
+    - Check if CONFIG_IP_MROUTE ifdef makes any sense.
+ */
+
+/* User supplied source address verification for output packets.
+   Such a verification can't be considered as a security measure.
+   It's rather an additional Internet protection against bugs in applications
+   (like using an uninitialized garbage as a source for UDP packets).
+ */
+static int outrt_check_src(u32 saddr, u32 daddr, u32 tos, struct net_device *dev_out)
+{
+	struct in_device	*in_dev;
+	int			src_check;
+
+	if (MULTICAST(saddr) || BADCLASS(saddr) || ZERONET(saddr))
+		return -EINVAL;
+	if (saddr == htonl(INADDR_BROADCAST))
+		return -EINVAL;
+
+	in_dev = in_dev_get(dev_out);
+	src_check = IN_DEV_SRC_CHECK(in_dev);
+	in_dev_put(in_dev);
+	if (!src_check)
+		return 0;
+
+	if (LOOPBACK(saddr) && !(dev_out->flags&IFF_LOOPBACK))
+		return -EINVAL;
+
+	return fib_local_source(saddr, daddr, tos, dev_out);
+}
+
+static int outrt_make_route(struct rtable **rp,
+	/* route lookup key */
+	struct rt_key *key,
+	/* path */
+	u32 daddr, u32 saddr, struct net_device *dev_out,
+	/* FIB lookup results (type, fi, nh.gw, nh.scope) */
+	struct fib_result *res
+	)
+{
+	struct rtable *rth;
+	unsigned hash;
+	unsigned flags;
+
+	rth = dst_alloc(&ipv4_dst_ops);
+	if (!rth)
+		return -ENOBUFS;
+
+	atomic_set(&rth->u.dst.__refcnt, 1);
+	rth->u.dst.flags= DST_HOST;
+	rth->key	= *key;
+	rth->key.iif	= 0; /* output route */
+	rth->rt_dst	= daddr;
+	rth->rt_src	= saddr;
+#ifdef CONFIG_IP_ROUTE_NAT
+	rth->rt_dst_map	= daddr;
+	rth->rt_src_map	= saddr;
+#endif
+
+	/* Set input, output ROUTINES and rt_spec_dst */
+	switch (res->type) {
+		case RTN_LOCAL:
+			/* Use loopback interface for unicast local traffic */
+			fib_res_put(res);
+			res->fi = NULL;
+#ifdef CONFIG_IP_MULTIPLE_TABLES
+			res->r = NULL;
+#endif
+			dev_out = &loopback_dev;
+			flags = RTCF_LOCAL;
+			rth->u.dst.input  = ip_local_deliver;
+			rth->u.dst.output = ip_output;
+			rth->rt_spec_dst  = daddr; /* local side of the path */
+			break;
+		case RTN_UNICAST:
+			flags = 0;
+			rth->u.dst.output = ip_output;
+			rth->rt_spec_dst  = saddr;
+#ifdef CONFIG_IP_ROUTE_MULTIPATH
+			if (res->fi->fib_nhs > 1 && key->oif == 0)
+				/* Set the proper res->nh_sel. */
+				fib_select_multipath(key, res);
+			else
+#endif
+			if (res->prefixlen == 0 && res->type == RTN_UNICAST &&
+					key->oif == 0)
+				fib_select_default(key, res);
+			break;
+		case RTN_BROADCAST:
+			flags = RTCF_BROADCAST|RTCF_LOCAL;
+			rth->u.dst.input  = ip_local_deliver;
+			if (!(dev_out->flags&IFF_LOOPBACK))
+				rth->u.dst.output = ip_mc_output;
+			else
+				rth->u.dst.output = ip_output;
+			rth->rt_spec_dst  = saddr;
+			break;
+		case RTN_MULTICAST:
+			{
+			/* Please note that all ancient band-aids were removed.
+			   I don't try to catch route table deficient for
+			   multicast or 255.255.255.255 routes and "smartly"
+			   replace a gatewayed default by the corresponding
+			   route.  1999/11/06  SAW
+			 */
+			struct in_device *in_dev = in_dev_get(dev_out);
+			rth->u.dst.input  = ip_local_deliver;
+			rth->u.dst.output = ip_output;
+			flags = RTCF_MULTICAST;
+			if (in_dev && ip_check_mc(in_dev, daddr)) {
+				/* Note: I preserve the original behaviour
+				   here.  It means that users after joining and
+				   leaving a multicast group have to flush
+				   the route cache.  I hope they know about it
+				   :-)  1999/11/06  SAW
+				 */
+				flags = RTCF_MULTICAST|RTCF_LOCAL;
+				if (!(dev_out->flags&IFF_LOOPBACK))
+					rth->u.dst.output = ip_mc_output;
+			}
+#ifdef CONFIG_IP_MROUTE
+			if (in_dev && !(dev_out->flags&IFF_LOOPBACK)) {
+				if (IN_DEV_MFORWARD(in_dev) &&
+						!LOCAL_MCAST(daddr))
+				{
+					rth->u.dst.input = ip_mr_input;
+					rth->u.dst.output = ip_mc_output;
+				}
+			}
+#endif
+			if (in_dev)
+				in_dev_put(in_dev);
+			rth->rt_spec_dst  = saddr;
+			}
+			break;
+		case RTN_NAT:
+			dst_free(&rth->u.dst);
+			return -EINVAL;
+		default:
+			printk(KERN_CRIT "bad lookup result type in route output\n");
+			return -EINVAL;
+	}
+
+	/* INTERFACE */
+	/* Store the interface information to allow users to get it via
+	 * [SOL_IP, IP_PKTINFO] conrol message for locally seen packets
+	 * (including broadcast and multicast ones).  --SAW */
+	rth->rt_iif	= key->oif ? : dev_out->ifindex;
+	/* Set output device */
+	rth->u.dst.dev	= dev_out;
+	dev_hold(dev_out);
+
+	/* Set GATEWAY */
+	rth->rt_gateway = daddr;
+	/* if res->fi != NULL set the real gateway */
+	rt_set_nexthop(rth, res, 0);
+
+	rth->rt_flags = flags;
+
+	hash = rt_hash_code(key->dst, key->src^(key->oif<<5), key->tos);
+	return rt_intern_hash(hash, rth, rp);
+}
+
 /*
  * Major route resolver routine.
  */
-
 int ip_route_output_slow(struct rtable **rp, u32 daddr, u32 saddr, u32 tos, int oif)
 {
 	struct rt_key key;
 	struct fib_result res;
-	unsigned flags = 0;
-	struct rtable *rth;
 	struct net_device *dev_out = NULL;
-	unsigned hash;
-	int free_res = 0;
 	int err;
 
-	tos &= IPTOS_TOS_MASK|RTO_ONLINK;
 	key.dst = daddr;
 	key.src = saddr;
 	key.tos = tos&IPTOS_TOS_MASK;
@@ -1619,252 +1792,100 @@
 	res.r = NULL;
 #endif
 
-	if (saddr) {
-		if (MULTICAST(saddr) || BADCLASS(saddr) || ZERONET(saddr))
-			return -EINVAL;
-
-		/* It is equivalent to inet_addr_type(saddr) == RTN_LOCAL */
-		dev_out = ip_dev_find(saddr);
-		if (dev_out == NULL)
-			return -EINVAL;
-
-		/* I removed check for oif == dev_out->oif here.
-		   It was wrong by three reasons:
-		   1. ip_dev_find(saddr) can return wrong iface, if saddr is
-		      assigned to multiple interfaces.
-		   2. Moreover, we are allowed to send packets with saddr
-		      of another iface. --ANK
-		 */
-
-		if (oif == 0 &&
-			(MULTICAST(daddr) || daddr == 0xFFFFFFFF)) {
-			/* Special hack: user can direct multicasts
-			   and limited broadcast via necessary interface
-			   without fiddling with IP_MULTICAST_IF or IP_PKTINFO.
-			   This hack is not just for fun, it allows
-			   vic,vat and friends to work.
-			   They bind socket to loopback, set ttl to zero
-			   and expect that it will work.
-			   From the viewpoint of routing cache they are broken,
-			   because we are not allowed to build multicast path
-			   with loopback source addr (look, routing cache
-			   cannot know, that ttl is zero, so that packet
-			   will not leave this host and route is valid).
-			   Luckily, this hack is good workaround.
-			 */
-
-			key.oif = dev_out->ifindex;
-			goto make_route;
-		}
-		if (dev_out)
-			dev_put(dev_out);
-		dev_out = NULL;
-	}
-	if (oif) {
-		dev_out = dev_get_by_index(oif);
-		if (dev_out == NULL)
-			return -ENODEV;
-		if (__in_dev_get(dev_out) == NULL) {
-			dev_put(dev_out);
-			return -ENODEV;	/* Wrong error code */
-		}
-
-		if (LOCAL_MCAST(daddr) || daddr == 0xFFFFFFFF) {
-			if (!key.src)
-				key.src = inet_select_addr(dev_out, 0, RT_SCOPE_LINK);
-			goto make_route;
-		}
-		if (!key.src) {
-			if (MULTICAST(daddr))
-				key.src = inet_select_addr(dev_out, 0, key.scope);
-			else if (!daddr)
-				key.src = inet_select_addr(dev_out, 0, RT_SCOPE_HOST);
-		}
-	}
+	if (!daddr)
+		goto dest_insanity;
 
-	if (!key.dst) {
-		key.dst = key.src;
-		if (!key.dst)
-			key.dst = key.src = htonl(INADDR_LOOPBACK);
-		if (dev_out)
-			dev_put(dev_out);
-		dev_out = &loopback_dev;
+	err = fib_lookup(&key, &res);
+	if (!err) {
+		dev_out = FIB_RES_DEV(res);
 		dev_hold(dev_out);
-		key.oif = loopback_dev.ifindex;
-		res.type = RTN_LOCAL;
-		flags |= RTCF_LOCAL;
-		goto make_route;
-	}
-
-	if (fib_lookup(&key, &res)) {
-		res.fi = NULL;
-		if (oif) {
-			/* Apparently, routing tables are wrong. Assume,
-			   that the destination is on link.
-
-			   WHY? DW.
-			   Because we are allowed to send to iface
-			   even if it has NO routes and NO assigned
-			   addresses. When oif is specified, routing
-			   tables are looked up with only one purpose:
-			   to catch if destination is gatewayed, rather than
-			   direct. Moreover, if MSG_DONTROUTE is set,
-			   we send packet, ignoring both routing tables
-			   and ifaddr state. --ANK
-
-
-			   We could make it even if oif is unknown,
-			   likely IPv6, but we do not.
+		if (saddr) {
+			/* Verify user supplied source address */
+			err = outrt_check_src(saddr, daddr, tos, dev_out);
+		} else {
+			/* Obtain path source from routing table */
+			saddr = FIB_RES_PREFSRC(res);
+			/* We don't verify source address obtained from routing
+			 * table.  It's a task of administrators to keep it
+			 * sane.
 			 */
-
-			if (key.src == 0)
-				key.src = inet_select_addr(dev_out, 0, RT_SCOPE_LINK);
-			res.type = RTN_UNICAST;
-			goto make_route;
 		}
-		if (dev_out)
-			dev_put(dev_out);
-		return -ENETUNREACH;
-	}
-	free_res = 1;
-
-	if (res.type == RTN_NAT)
-		goto e_inval;
-
-	if (res.type == RTN_LOCAL) {
-		if (!key.src)
-			key.src = key.dst;
-		if (dev_out)
-			dev_put(dev_out);
-		dev_out = &loopback_dev;
-		dev_hold(dev_out);
-		key.oif = dev_out->ifindex;
-		if (res.fi)
-			fib_info_put(res.fi);
-		res.fi = NULL;
-		flags |= RTCF_LOCAL;
-		goto make_route;
-	}
-
-#ifdef CONFIG_IP_ROUTE_MULTIPATH
-	if (res.fi->fib_nhs > 1 && key.oif == 0)
-		fib_select_multipath(&key, &res);
-	else
-#endif
-	if (res.prefixlen==0 && res.type == RTN_UNICAST && key.oif == 0)
-		fib_select_default(&key, &res);
-
-	if (!key.src)
-		key.src = FIB_RES_PREFSRC(res);
-
-	if (dev_out)
-		dev_put(dev_out);
-	dev_out = FIB_RES_DEV(res);
-	dev_hold(dev_out);
-	key.oif = dev_out->ifindex;
-
-make_route:
-	if (LOOPBACK(key.src) && !(dev_out->flags&IFF_LOOPBACK))
-		goto e_inval;
-
-	if (key.dst == 0xFFFFFFFF)
-		res.type = RTN_BROADCAST;
-	else if (MULTICAST(key.dst))
-		res.type = RTN_MULTICAST;
-	else if (BADCLASS(key.dst) || ZERONET(key.dst))
-		goto e_inval;
+		if (!err)
+			err = outrt_make_route(rp, &key, daddr, saddr,
+					dev_out, &res);
+		fib_res_put(&res);
+		/* The usual code path ends here */
 
-	if (dev_out->flags&IFF_LOOPBACK)
-		flags |= RTCF_LOCAL;
+	} else if (err == -ENETUNREACH) {
 
-	if (res.type == RTN_BROADCAST) {
-		flags |= RTCF_BROADCAST|RTCF_LOCAL;
-		if (res.fi) {
-			fib_info_put(res.fi);
-			res.fi = NULL;
-		}
-	} else if (res.type == RTN_MULTICAST) {
-		flags |= RTCF_MULTICAST|RTCF_LOCAL;
-		read_lock(&inetdev_lock);
-		if (!__in_dev_get(dev_out) || !ip_check_mc(__in_dev_get(dev_out), daddr))
-			flags &= ~RTCF_LOCAL;
-		read_unlock(&inetdev_lock);
-		/* If multicast route do not exist use
-		   default one, but do not gateway in this case.
-		   Yes, it is hack.
+		/* Just return if the access is prohibited etc.
+		   If the routing table doesn't have both an appropriate route
+		   and a default assume that the destination is on link.  --SAW
+
+		   WHY? DW.
+		   Because we are allowed to send to iface
+		   even if it has NO routes and NO assigned
+		   addresses. When oif is specified, routing
+		   tables are looked up with only one purpose:
+		   to catch if destination is gatewayed, rather than
+		   direct. Moreover, if MSG_DONTROUTE is set,
+		   we send packet, ignoring both routing tables
+		   and ifaddr state. --ANK
+
+		   We could make it even if oif is unknown,
+		   likely IPv6, but we do not.
+
+		   The above statements aren't exactly correct.
+		   Routing tables contain a lot of useful information (like
+		   preferred source, for instance).  But the general idea is
+		   right.  --SAW
 		 */
-		if (res.fi && res.prefixlen < 4) {
-			fib_info_put(res.fi);
-			res.fi = NULL;
-		}
-	}
-
-	rth = dst_alloc(&ipv4_dst_ops);
-	if (!rth)
-		goto e_nobufs;
-
-	atomic_set(&rth->u.dst.__refcnt, 1);
-	rth->u.dst.flags= DST_HOST;
-	rth->key.dst	= daddr;
-	rth->key.tos	= tos;
-	rth->key.src	= saddr;
-	rth->key.iif	= 0;
-	rth->key.oif	= oif;
-	rth->rt_dst	= key.dst;
-	rth->rt_src	= key.src;
-#ifdef CONFIG_IP_ROUTE_NAT
-	rth->rt_dst_map	= key.dst;
-	rth->rt_src_map	= key.src;
-#endif
-	rth->rt_iif	= oif ? : dev_out->ifindex;
-	rth->u.dst.dev	= dev_out;
-	dev_hold(dev_out);
-	rth->rt_gateway = key.dst;
-	rth->rt_spec_dst= key.src;
-
-	rth->u.dst.output=ip_output;
-
-	if (flags&RTCF_LOCAL) {
-		rth->u.dst.input = ip_local_deliver;
-		rth->rt_spec_dst = key.dst;
-	}
-	if (flags&(RTCF_BROADCAST|RTCF_MULTICAST)) {
-		rth->rt_spec_dst = key.src;
-		if (flags&RTCF_LOCAL && !(dev_out->flags&IFF_LOOPBACK))
-			rth->u.dst.output = ip_mc_output;
-#ifdef CONFIG_IP_MROUTE
-		if (res.type == RTN_MULTICAST) {
-			struct in_device *in_dev = in_dev_get(dev_out);
-			if (in_dev) {
-				if (IN_DEV_MFORWARD(in_dev) && !LOCAL_MCAST(daddr)) {
-					rth->u.dst.input = ip_mr_input;
-					rth->u.dst.output = ip_mc_output;
-				}
-				in_dev_put(in_dev);
+		if (oif) {
+			err = -ENODEV;
+			dev_out = dev_get_by_index(oif);
+			if (dev_out == NULL)
+				goto out;
+			if (__in_dev_get(dev_out) == NULL)
+				goto out;	/* Wrong error code */
+			if (saddr) {
+				/* Verify user supplied source address */
+				err = outrt_check_src(saddr, daddr, tos, dev_out);
+				if (err)
+					goto out;
+			} else {
+				int scope;
+				if (LOCAL_MCAST(daddr) || daddr == 0xFFFFFFFF)
+					scope = RT_SCOPE_LINK;
+				else if (MULTICAST(daddr))
+					scope = key.scope;
+				else
+					scope = RT_SCOPE_HOST;
+				saddr = inet_select_addr(dev_out, 0, scope);
 			}
-		}
-#endif
+			res.type = RTN_UNICAST;
+			res.fi = NULL;
+			err = outrt_make_route(rp, &key, daddr, saddr, dev_out,
+					&res);
+		} else
+			err = -ENETUNREACH;
 	}
-
-	rt_set_nexthop(rth, &res, 0);
-
-	rth->rt_flags = flags;
-
-	hash = rt_hash_code(daddr, saddr^(oif<<5), tos);
-	err = rt_intern_hash(hash, rth, rp);
-done:
-	if (free_res)
-		fib_res_put(&res);
+out:
 	if (dev_out)
 		dev_put(dev_out);
 	return err;
 
-e_inval:
-	err = -EINVAL;
-	goto done;
-e_nobufs:
-	err = -ENOBUFS;
-	goto done;
+	/* I don't know what reason this hack was for */
+dest_insanity:
+	daddr = saddr;
+	if (!daddr)
+		daddr = saddr = htonl(INADDR_LOOPBACK);
+	dev_out = &loopback_dev;
+	dev_hold(dev_out);
+	key.oif = loopback_dev.ifindex;
+	res.type = RTN_LOCAL;
+	res.fi = NULL;
+	err = outrt_make_route(rp, &key, daddr, saddr, dev_out, &res);
+	goto out;
 }
 
 int ip_route_output(struct rtable **rp, u32 daddr, u32 saddr, u32 tos, int oif)