Import of Linux 2.2.14 subset (ipv4 stack and related)

16 files changed, 207 insertions, 77 deletions

diff --git a/pfinet/linux-src/net/core/dev.c b/pfinet/linux-src/net/core/dev.c
index 75fae54a..0baa1371 100644
--- a/pfinet/linux-src/net/core/dev.c
+++ b/pfinet/linux-src/net/core/dev.c
@@ -57,6 +57,8 @@
  *					A network device unload needs to purge
  *					the backlog queue.
  *	Paul Rusty Russel	:	SIOCSIFNAME
+ *	Andrea Arcangeli	:	dev_clear_backlog() needs the
+ *					skb_queue_lock held.
  */
 
 #include <asm/uaccess.h>
@@ -711,7 +713,8 @@ static void netdev_wakeup(void)
 
 static void dev_clear_backlog(struct device *dev)
 {
-	struct sk_buff *prev, *curr;
+	struct sk_buff *curr;
+	unsigned long flags;
 
 	/*
 	 *
@@ -719,27 +722,24 @@ static void dev_clear_backlog(struct device *dev)
 	 *
 	 *  We are competing here both with netif_rx() and net_bh().
 	 *  We don't want either of those to mess with skb ptrs
-	 *  while we work on them, thus cli()/sti().
-	 *
-	 *  It looks better to use net_bh trick, at least
-	 *  to be sure, that we keep interrupt latency really low. --ANK (980727)
+	 *  while we work on them, thus we must grab the
+	 *  skb_queue_lock.
 	 */ 
 
 	if (backlog.qlen) {
-		start_bh_atomic();
-		curr = backlog.next;
-		while ( curr != (struct sk_buff *)(&backlog) ) {
-			unsigned long flags;
-			curr=curr->next;
-			if ( curr->prev->dev == dev ) {
-				prev = curr->prev;
-				spin_lock_irqsave(&skb_queue_lock, flags);
-				__skb_unlink(prev, &backlog);
+	repeat:
+		spin_lock_irqsave(&skb_queue_lock, flags);
+		for (curr = backlog.next;
+		     curr != (struct sk_buff *)(&backlog);
+		     curr = curr->next)
+			if (curr->dev == dev)
+			{
+				__skb_unlink(curr, &backlog);
 				spin_unlock_irqrestore(&skb_queue_lock, flags);
-				kfree_skb(prev);
+				kfree_skb(curr);
+				goto repeat;
 			}
-		}
-		end_bh_atomic();
+		spin_unlock_irqrestore(&skb_queue_lock, flags);
 #ifdef CONFIG_NET_HW_FLOWCONTROL
 		if (netdev_dropping)
 			netdev_wakeup();
@@ -796,7 +796,11 @@ void netif_rx(struct sk_buff *skb)
 #ifdef CONFIG_BRIDGE
 static inline void handle_bridge(struct sk_buff *skb, unsigned short type)
 {
-	if (br_stats.flags & BR_UP && br_protocol_ok(ntohs(type)))
+	/* 
+	 * The br_stats.flags is checked here to save the expense of a 
+	 * function call.
+	 */
+	if ((br_stats.flags & BR_UP) && br_call_bridge(skb, type))
 	{
 		/*
 		 *	We pass the bridge a complete frame. This means
@@ -820,7 +824,6 @@ static inline void handle_bridge(struct sk_buff *skb, unsigned short type)
 }
 #endif
 
-
 /*
  *	When we are called the queue is ready to grab, the interrupts are
  *	on and hardware can interrupt and queue to the receive queue as we
@@ -1275,8 +1278,9 @@ static int sprintf_wireless_stats(char *buffer, struct device *dev)
 	int size;
 
 	if(stats != (struct iw_statistics *) NULL)
+	{
 		size = sprintf(buffer,
-			       "%6s: %02x  %3d%c %3d%c  %3d%c %5d %5d %5d\n",
+			       "%6s: %04x  %3d%c  %3d%c  %3d%c  %6d %6d %6d\n",
 			       dev->name,
 			       stats->status,
 			       stats->qual.qual,
@@ -1288,6 +1292,8 @@ static int sprintf_wireless_stats(char *buffer, struct device *dev)
 			       stats->discard.nwid,
 			       stats->discard.code,
 			       stats->discard.misc);
+		stats->qual.updated = 0;
+	}
 	else
 		size = 0;
 
@@ -1309,8 +1315,9 @@ int dev_get_wireless_info(char * buffer, char **start, off_t offset,
 	struct device *	dev;
 
 	size = sprintf(buffer,
-		       "Inter-|sta|  Quality       |  Discarded packets\n"
-		       " face |tus|link level noise| nwid crypt  misc\n");
+		       "Inter-| sta-|   Quality        |   Discarded packets\n"
+		       " face | tus | link level noise |  nwid  crypt   misc\n"
+			);
 	
 	pos+=size;
 	len+=size;
@@ -1846,6 +1853,7 @@ extern int lance_init(void);
 extern int bpq_init(void);
 extern int scc_init(void);
 extern void sdla_setup(void);
+extern void sdla_c_setup(void);
 extern void dlci_setup(void);
 extern int dmascc_init(void);
 extern int sm_init(void);
@@ -1855,6 +1863,7 @@ extern int baycom_ser_hdx_init(void);
 extern int baycom_par_init(void);
 
 extern int lapbeth_init(void);
+extern int comx_init(void);
 extern void arcnet_init(void);
 extern void ip_auto_config(void);
 #ifdef CONFIG_8xx
@@ -1922,7 +1931,7 @@ __initfunc(int net_dev_init(void))
 	dlci_setup();
 #endif
 #if defined(CONFIG_SDLA)
-	sdla_setup();
+	sdla_c_setup();
 #endif
 #if defined(CONFIG_BAYCOM_PAR)
 	baycom_par_init();
@@ -1948,6 +1957,9 @@ __initfunc(int net_dev_init(void))
 #if defined(CONFIG_8xx)
         cpm_enet_init();
 #endif
+#if defined(CONFIG_COMX)
+	comx_init();
+#endif
 	/*
 	 *	SLHC if present needs attaching so other people see it
 	 *	even if not opened.
@@ -2018,6 +2030,13 @@ __initfunc(int net_dev_init(void))
 
 	dev_mcast_init();
 
+#ifdef CONFIG_BRIDGE
+	/*
+	 * Register any statically linked ethernet devices with the bridge
+	 */
+	br_spacedevice_register();
+#endif
+
 #ifdef CONFIG_IP_PNP
 	ip_auto_config();
 #endif
diff --git a/pfinet/linux-src/net/ethernet/eth.c b/pfinet/linux-src/net/ethernet/eth.c
index bce35d48..945f4a06 100644
--- a/pfinet/linux-src/net/ethernet/eth.c
+++ b/pfinet/linux-src/net/ethernet/eth.c
@@ -201,7 +201,7 @@ unsigned short eth_type_trans(struct sk_buff *skb, struct device *dev)
 	 *	seems to set IFF_PROMISC.
 	 */
 	 
-	else if(dev->flags&(IFF_PROMISC/*|IFF_ALLMULTI*/))
+	else if(1 /*dev->flags&IFF_PROMISC*/)
 	{
 		if(memcmp(eth->h_dest,dev->dev_addr, ETH_ALEN))
 			skb->pkt_type=PACKET_OTHERHOST;
diff --git a/pfinet/linux-src/net/ipv4/arp.c b/pfinet/linux-src/net/ipv4/arp.c
index 1e5cf1b1..d81c1bee 100644
--- a/pfinet/linux-src/net/ipv4/arp.c
+++ b/pfinet/linux-src/net/ipv4/arp.c
@@ -1,6 +1,6 @@
 /* linux/net/inet/arp.c
  *
- * Version:	$Id: arp.c,v 1.77.2.1 1999/06/28 10:39:23 davem Exp $
+ * Version:	$Id: arp.c,v 1.77.2.4 1999/09/23 19:03:36 davem Exp $
  *
  * Copyright (C) 1994 by Florian  La Roche
  *
@@ -65,6 +65,8 @@
  *					clean up the APFDDI & gen. FDDI bits.
  *		Alexey Kuznetsov:	new arp state machine;
  *					now it is in net/core/neighbour.c.
+ *		Julian Anastasov:	"hidden" flag: hide the
+ *					interface and don't reply for it
  */
 
 /* RFC1122 Status:
@@ -308,10 +310,15 @@ static void arp_solicit(struct neighbour *neigh, struct sk_buff *skb)
 	u32 saddr;
 	u8  *dst_ha = NULL;
 	struct device *dev = neigh->dev;
+	struct device *dev2;
+	struct in_device *in_dev2;
 	u32 target = *(u32*)neigh->primary_key;
 	int probes = neigh->probes;
 
-	if (skb && inet_addr_type(skb->nh.iph->saddr) == RTN_LOCAL)
+	if (skb &&
+	    (dev2 = ip_dev_find(skb->nh.iph->saddr)) != NULL &&
+	    (in_dev2 = dev2->ip_ptr) != NULL &&
+	    !IN_DEV_HIDDEN(in_dev2))
 		saddr = skb->nh.iph->saddr;
 	else
 		saddr = inet_select_addr(dev, target, RT_SCOPE_LINK);
@@ -653,8 +660,14 @@ int arp_rcv(struct sk_buff *skb, struct device *dev, struct packet_type *pt)
 
 	/* Special case: IPv4 duplicate address detection packet (RFC2131) */
 	if (sip == 0) {
+		struct device *dev2;
+		struct in_device *in_dev2;
+
 		if (arp->ar_op == __constant_htons(ARPOP_REQUEST) &&
-		    inet_addr_type(tip) == RTN_LOCAL)
+		    (dev2 = ip_dev_find(tip)) != NULL &&
+		    (dev2 == dev ||
+		     ((in_dev2 = dev2->ip_ptr) != NULL &&
+		      !IN_DEV_HIDDEN(in_dev2))))
 			arp_send(ARPOP_REPLY,ETH_P_ARP,tip,dev,tip,sha,dev->dev_addr,dev->dev_addr);
 		goto out;
 	}
@@ -668,6 +681,20 @@ int arp_rcv(struct sk_buff *skb, struct device *dev, struct packet_type *pt)
 		if (addr_type == RTN_LOCAL) {
 			n = neigh_event_ns(&arp_tbl, sha, &sip, dev);
 			if (n) {
+				if (ipv4_devconf.hidden &&
+				    skb->pkt_type != PACKET_HOST) {
+					struct device *dev2;
+					struct in_device *in_dev2;
+
+					if ((dev2 = ip_dev_find(tip)) != NULL &&
+					    dev2 != dev &&
+					    (in_dev2 = dev2->ip_ptr) != NULL &&
+					    IN_DEV_HIDDEN(in_dev2)) {
+						neigh_release(n);
+						goto out;
+					}
+				}
+
 				arp_send(ARPOP_REPLY,ETH_P_ARP,sip,dev,tip,sha,dev->dev_addr,sha);
 				neigh_release(n);
 			}
diff --git a/pfinet/linux-src/net/ipv4/devinet.c b/pfinet/linux-src/net/ipv4/devinet.c
index a50ee3bd..41b42a52 100644
--- a/pfinet/linux-src/net/ipv4/devinet.c
+++ b/pfinet/linux-src/net/ipv4/devinet.c
@@ -658,7 +658,7 @@ u32 inet_select_addr(struct device *dev, u32 dst, int scope)
 			addr = ifa->ifa_local;
 	} endfor_ifa(in_dev);
 	
-	if (addr || scope >= RT_SCOPE_LINK)
+	if (addr)
 		return addr;
 
 	/* Not loopback addresses on loopback should be preferred
@@ -670,7 +670,9 @@ u32 inet_select_addr(struct device *dev, u32 dst, int scope)
 			continue;
 
 		for_primary_ifa(in_dev) {
-			if (ifa->ifa_scope <= scope)
+			if (!IN_DEV_HIDDEN(in_dev) &&
+			    ifa->ifa_scope <= scope &&
+			    ifa->ifa_scope != RT_SCOPE_LINK)
 				return ifa->ifa_local;
 		} endfor_ifa(in_dev);
 	}
@@ -923,7 +925,7 @@ int devinet_sysctl_forward(ctl_table *ctl, int write, struct file * filp,
 static struct devinet_sysctl_table
 {
 	struct ctl_table_header *sysctl_header;
-	ctl_table devinet_vars[12];
+	ctl_table devinet_vars[13];
 	ctl_table devinet_dev[2];
 	ctl_table devinet_conf_dir[2];
 	ctl_table devinet_proto_dir[2];
@@ -963,6 +965,9 @@ static struct devinet_sysctl_table
         {NET_IPV4_CONF_LOG_MARTIANS, "log_martians",
          &ipv4_devconf.log_martians, sizeof(int), 0644, NULL,
          &proc_dointvec},
+	{NET_IPV4_CONF_HIDDEN, "hidden",
+         &ipv4_devconf.hidden, sizeof(int), 0644, NULL,
+         &proc_dointvec},
 	 {0}},
 
 	{{NET_PROTO_CONF_ALL, "all", NULL, 0, 0555, devinet_sysctl.devinet_vars},{0}},
diff --git a/pfinet/linux-src/net/ipv4/icmp.c b/pfinet/linux-src/net/ipv4/icmp.c
index 78c54be8..6c1edfd7 100644
--- a/pfinet/linux-src/net/ipv4/icmp.c
+++ b/pfinet/linux-src/net/ipv4/icmp.c
@@ -3,7 +3,7 @@
  *	
  *		Alan Cox, <alan@redhat.com>
  *
- *	Version: $Id: icmp.c,v 1.52.2.3 1999/09/22 16:33:02 davem Exp $
+ *	Version: $Id: icmp.c,v 1.52.2.4 1999/11/16 02:28:40 davem Exp $
  *
  *	This program is free software; you can redistribute it and/or
  *	modify it under the terms of the GNU General Public License
diff --git a/pfinet/linux-src/net/ipv4/ip_forward.c b/pfinet/linux-src/net/ipv4/ip_forward.c
index 08ebbc2f..fd8c0435 100644
--- a/pfinet/linux-src/net/ipv4/ip_forward.c
+++ b/pfinet/linux-src/net/ipv4/ip_forward.c
@@ -5,11 +5,12 @@
  *
  *		The IP forwarding functionality.
  *		
- * Version:	$Id: ip_forward.c,v 1.43 1999/03/21 05:22:37 davem Exp $
+ * Version:	$Id: ip_forward.c,v 1.43.2.1 1999/11/16 06:33:43 davem Exp $
  *
  * Authors:	see ip.c
  *
  * Fixes:
+ *		Joseph Gooch	:	Removed maddr selection for ip_masq, now done in ip_masq.c
  *		Many		:	Split from ip.c , see ip_input.c for 
  *					history.
  *		Dave Gregorich	:	NULL ip_rt_put fix for multicast 
@@ -168,7 +169,6 @@ int ip_forward(struct sk_buff *skb)
 		 *	and skip the firewall checks
 		 */
 		if (iph->protocol == IPPROTO_ICMP) {
-			__u32 maddr;
 #ifdef CONFIG_IP_MASQUERADE_ICMP
 			struct icmphdr *icmph = (struct icmphdr *)((char*)iph + (iph->ihl << 2));
 			if ((icmph->type==ICMP_DEST_UNREACH)||
@@ -176,8 +176,7 @@ int ip_forward(struct sk_buff *skb)
 			    (icmph->type==ICMP_TIME_EXCEEDED))
 			{
 #endif
-				maddr = inet_select_addr(dev2, rt->rt_gateway, RT_SCOPE_UNIVERSE);
-				fw_res = ip_fw_masq_icmp(&skb, maddr);
+				fw_res = ip_fw_masquerade(&skb, 0);
 			        if (fw_res < 0) {
 					kfree_skb(skb);
 					return -1;
@@ -187,7 +186,7 @@ int ip_forward(struct sk_buff *skb)
 					/* ICMP matched - skip firewall */
 					goto skip_call_fw_firewall;
 #ifdef CONFIG_IP_MASQUERADE_ICMP
-			       }
+			}
 #endif				
 		}
 		if (rt->rt_flags&RTCF_MASQ)
@@ -219,15 +218,11 @@ skip_call_fw_firewall:
 	 */
 	if (!(IPCB(skb)->flags&IPSKB_MASQUERADED) &&
 	    (fw_res==FW_MASQUERADE || rt->rt_flags&RTCF_MASQ)) {
-		u32 maddr;
+		u32 maddr = 0;
 
 #ifdef CONFIG_IP_ROUTE_NAT
 		maddr = (rt->rt_flags&RTCF_MASQ) ? rt->rt_src_map : 0;
-
-		if (maddr == 0)
 #endif
-			maddr = inet_select_addr(dev2, rt->rt_gateway, RT_SCOPE_UNIVERSE);
-
 			if (ip_fw_masquerade(&skb, maddr) < 0) {
 				kfree_skb(skb);
 				return -1;
diff --git a/pfinet/linux-src/net/ipv4/ip_fw.c b/pfinet/linux-src/net/ipv4/ip_fw.c
index 6319b89e..400f46c0 100644
--- a/pfinet/linux-src/net/ipv4/ip_fw.c
+++ b/pfinet/linux-src/net/ipv4/ip_fw.c
@@ -40,11 +40,12 @@
  * 23-Jul-1999: Fixed small fragment security exposure opened on 15-May-1998.
  *              John McDonald <jm@dataprotect.com>
  *              Thomas Lopatic <tl@dataprotect.com>
+ * 21-Oct-1999: Applied count fix by Emanuele Caratti <wiz@iol.it> --RR
  */
 
 /*
  *
- * The origina Linux port was done Alan Cox, with changes/fixes from
+ * The original Linux port was done Alan Cox, with changes/fixes from
  * Pauline Middlelink, Jos Vos, Thomas Quinot, Wouter Gadeyne, Juan
  * Jose Ciarlante, Bernd Eckenfels, Keith Owens and others.
  * 
@@ -86,6 +87,7 @@
 #include <net/udp.h>
 #include <net/sock.h>
 #include <net/icmp.h>
+#include <net/ip_masq.h>
 #include <linux/netlink.h>
 #include <linux/init.h>
 #include <linux/firewall.h>
@@ -228,6 +230,7 @@ struct ip_reent
 {
 	struct ip_chain *prevchain;	/* Pointer to referencing chain */
 	struct ip_fwkernel *prevrule;	/* Pointer to referencing rule */
+	unsigned int count;
 	struct ip_counters counters;
 };
 
@@ -488,7 +491,10 @@ static int find_special(ip_chainlabel label, int *answer)
                 static int enabled = 0;
 
                 if(!enabled)
+                {
+                	enabled=1;
                         sysctl_ip_always_defrag++;
+                }
 		*answer = FW_REDIRECT;
 		return 1;
 #endif
@@ -729,8 +735,8 @@ ip_fw_check(struct iphdr *ip,
 	else FWC_HAVE_LOCK(fwc_rlocks);
 
 	f = chain->chain;
+	count = 0;
 	do {
-		count = 0;
 		for (; f; f = f->next) {
 			count++;
 			if (ip_rule_match(f,rif,ip,
@@ -768,10 +774,12 @@ ip_fw_check(struct iphdr *ip,
 				else {
 					f->branch->reent[slot].prevchain 
 						= chain;
+					f->branch->reent[slot].count = count;
 					f->branch->reent[slot].prevrule 
 						= f->next;
 					chain = f->branch;
 					f = chain->chain;
+					count = 0;
 				}
 			}
 			else if (f->simplebranch == FW_SKIP) 
@@ -790,6 +798,7 @@ ip_fw_check(struct iphdr *ip,
 			if (chain->reent[slot].prevchain) {
 				struct ip_chain *tmp = chain;
 				f = chain->reent[slot].prevrule;
+				count = chain->reent[slot].count;
 				chain = chain->reent[slot].prevchain;
 				tmp->reent[slot].prevchain = NULL;
 			}
diff --git a/pfinet/linux-src/net/ipv4/ip_input.c b/pfinet/linux-src/net/ipv4/ip_input.c
index b1725fa3..545f093c 100644
--- a/pfinet/linux-src/net/ipv4/ip_input.c
+++ b/pfinet/linux-src/net/ipv4/ip_input.c
@@ -97,6 +97,7 @@
  *		Alan Cox	:	Multicast routing hooks
  *		Jos Vos		:	Do accounting *before* call_in_firewall
  *	Willy Konynenberg	:	Transparent proxying support
+ *             Stephan Uphoff   :       Check IP header length field
  *
  *  
  *
@@ -421,6 +422,10 @@ int ip_rcv(struct sk_buff *skb, struct device *dev, struct packet_type *pt)
 
 	if (skb->len < sizeof(struct iphdr))
 		goto inhdr_error; 
+
+	if (skb->len < (iph->ihl << 2))
+		goto inhdr_error;
+
 	if (iph->ihl < 5 || iph->version != 4 || ip_fast_csum((u8 *)iph, iph->ihl) != 0)
 		goto inhdr_error; 
 
@@ -429,6 +434,9 @@ int ip_rcv(struct sk_buff *skb, struct device *dev, struct packet_type *pt)
 	if (skb->len < len)
 		goto inhdr_error; 
 
+	if (len <  (iph->ihl << 2))
+		goto inhdr_error; 
+
 	/*
 	 *	Our transport medium may have padded the buffer out. Now we know it
 	 *	is IP we can trim to the true length of the frame.
diff --git a/pfinet/linux-src/net/ipv4/ip_masq.c b/pfinet/linux-src/net/ipv4/ip_masq.c
index 6d0588c0..8f00409f 100644
--- a/pfinet/linux-src/net/ipv4/ip_masq.c
+++ b/pfinet/linux-src/net/ipv4/ip_masq.c
@@ -10,6 +10,9 @@
  *	See ip_fw.c for original log
  *
  * Fixes:
+ *	Joseph Gooch		:	Modified ip_fw_masquerade() to do a ip_route_output()
+ *	 (help by Dan Drown)	:	to choose the proper local address.
+ *	 (and Alexey)		:
  *	Juan Jose Ciarlante	:	Modularized application masquerading (see ip_masq_app.c)
  *	Juan Jose Ciarlante	:	New struct ip_masq_seq that holds output/input delta seq.
  *	Juan Jose Ciarlante	:	Added hashed lookup by proto,maddr,mport and proto,saddr,sport
@@ -787,6 +790,8 @@ void ip_masq_put(struct ip_masq *ms)
 	}
 }
 
+extern int sysctl_ip_always_defrag;
+
 static void masq_expire(unsigned long data)
 {
 	struct ip_masq *ms = (struct ip_masq *)data;
@@ -839,6 +844,7 @@ static void masq_expire(unsigned long data)
 	 */
 	if (atomic_read(&ms->refcnt) == 1) {
 		kfree_s(ms,sizeof(*ms));
+		sysctl_ip_always_defrag--;
 		MOD_DEC_USE_COUNT;
 		goto masq_expire_out;
 	}
@@ -880,8 +886,6 @@ static __u16 get_next_mport(void)
  * 	Be careful, it can be called from u-space
  */
 
-extern int sysctl_ip_always_defrag;
-
 struct ip_masq * ip_masq_new(int proto, __u32 maddr, __u16 mport, __u32 saddr, __u16 sport, __u32 daddr, __u16 dport, unsigned mflags)
 {
         struct ip_masq *ms, *mst;
@@ -1141,6 +1145,23 @@ int ip_fw_masquerade(struct sk_buff **skb_p, __u32 maddr)
 		return -1;
 	}
 
+	/* Lets determine our maddr now, shall we? */
+	if (maddr == 0) {
+		struct rtable *rt;
+		struct rtable *skb_rt = (struct rtable*)skb->dst;
+		struct device *skb_dev = skb_rt->u.dst.dev;
+
+		if (ip_route_output(&rt, iph->daddr, 0, RT_TOS(iph->tos)|RTO_CONN, skb_dev?skb_dev->ifindex:0)) {
+			/* Fallback on old method */
+			/* This really shouldn't happen... */
+			maddr = inet_select_addr(skb_dev, skb_rt->rt_gateway, RT_SCOPE_UNIVERSE);
+		} else {
+			/* Route lookup succeeded */
+			maddr = rt->rt_src;
+			ip_rt_put(rt);
+		}
+	}
+
 	switch (iph->protocol) {
 	case IPPROTO_ICMP:
 		return(ip_fw_masq_icmp(skb_p, maddr));
diff --git a/pfinet/linux-src/net/ipv4/ip_masq_user.c b/pfinet/linux-src/net/ipv4/ip_masq_user.c
index f369f03d..848cbfe8 100644
--- a/pfinet/linux-src/net/ipv4/ip_masq_user.c
+++ b/pfinet/linux-src/net/ipv4/ip_masq_user.c
@@ -2,7 +2,7 @@
  *	IP_MASQ_USER user space control module
  *
  *
- *	$Id: ip_masq_user.c,v 1.1.2.2 1999/08/13 18:26:33 davem Exp $
+ *	$Id: ip_masq_user.c,v 1.1.2.3 1999/11/16 06:33:51 davem Exp $
  */
 
 #include <linux/config.h>
@@ -100,7 +100,7 @@ static int ip_masq_user_maddr(struct ip_masq_user *ums)
 		return ret;
 	}
 	dev = rt->u.dst.dev;
-	ums->maddr = ip_masq_select_addr(dev, rt->rt_gateway, RT_SCOPE_UNIVERSE);
+	ums->maddr = rt->rt_src;  /* Per Alexey */
 
 	IP_MASQ_DEBUG(1-debug, "did setup maddr=%lX\n", ntohl(ums->maddr));
 	ip_rt_put(rt);
diff --git a/pfinet/linux-src/net/ipv4/route.c b/pfinet/linux-src/net/ipv4/route.c
index b308995b..2b114b3e 100644
--- a/pfinet/linux-src/net/ipv4/route.c
+++ b/pfinet/linux-src/net/ipv4/route.c
@@ -5,7 +5,7 @@
  *
  *		ROUTE - implementation of the IP router.
  *
- * Version:	$Id: route.c,v 1.67.2.3 1999/08/08 08:43:12 davem Exp $
+ * Version:	$Id: route.c,v 1.67.2.4 1999/11/16 02:28:43 davem Exp $
  *
  * Authors:	Ross Biro, <bir7@leland.Stanford.Edu>
  *		Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
diff --git a/pfinet/linux-src/net/ipv4/syncookies.c b/pfinet/linux-src/net/ipv4/syncookies.c
index fb4e8f80..62884aae 100644
--- a/pfinet/linux-src/net/ipv4/syncookies.c
+++ b/pfinet/linux-src/net/ipv4/syncookies.c
@@ -9,7 +9,7 @@
  *      as published by the Free Software Foundation; either version
  *      2 of the License, or (at your option) any later version.
  * 
- *  $Id: syncookies.c,v 1.7.2.1 1999/08/08 08:43:13 davem Exp $
+ *  $Id: syncookies.c,v 1.7.2.3 1999/12/07 03:11:07 davem Exp $
  *
  *  Missing: IPv6 support. 
  */
@@ -184,8 +184,10 @@ cookie_v4_check(struct sock *sk, struct sk_buff *skb, struct ip_options *opt)
 			    req->af.v4_req.loc_addr,
 			    sk->ip_tos | RTO_CONN,
 			    0)) { 
-	    tcp_openreq_free(req);
-	    return NULL; 
+		if (req->af.v4_req.opt)
+			kfree(req->af.v4_req.opt);
+		tcp_openreq_free(req);
+		return NULL; 
 	}
 
 	/* Try to redo what tcp_v4_send_synack did. */
diff --git a/pfinet/linux-src/net/ipv4/tcp.c b/pfinet/linux-src/net/ipv4/tcp.c
index 65763215..68b7e5e7 100644
--- a/pfinet/linux-src/net/ipv4/tcp.c
+++ b/pfinet/linux-src/net/ipv4/tcp.c
@@ -5,7 +5,7 @@
  *
  *		Implementation of the Transmission Control Protocol(TCP).
  *
- * Version:	$Id: tcp.c,v 1.140.2.4 1999/08/09 03:13:12 davem Exp $
+ * Version:	$Id: tcp.c,v 1.140.2.5 1999/09/23 19:21:16 davem Exp $
  *
  * Authors:	Ross Biro, <bir7@leland.Stanford.Edu>
  *		Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
@@ -1798,6 +1798,8 @@ extern void __skb_cb_too_small_for_tcp(int, int);
 void __init tcp_init(void)
 {
 	struct sk_buff *skb = NULL;
+	unsigned long goal;
+	int order;
 
 	if(sizeof(struct tcp_skb_cb) > sizeof(skb->cb))
 		__skb_cb_too_small_for_tcp(sizeof(struct tcp_skb_cb),
@@ -1823,4 +1825,43 @@ void __init tcp_init(void)
 						NULL, NULL);
 	if(!tcp_timewait_cachep)
 		panic("tcp_init: Cannot alloc tcp_tw_bucket cache.");
+
+	/* Size and allocate TCP hash tables. */
+	goal = num_physpages >> (20 - PAGE_SHIFT);
+	for (order = 0; (1UL << order) < goal; order++)
+		;
+	do {
+		tcp_ehash_size = (1UL << order) * PAGE_SIZE /
+			sizeof(struct sock *);
+		tcp_ehash = (struct sock **)
+			__get_free_pages(GFP_ATOMIC, order);
+	} while (tcp_ehash == NULL && --order >= 0);
+
+	if (!tcp_ehash)
+		panic("Failed to allocate TCP established hash table\n");
+	memset(tcp_ehash, 0, tcp_ehash_size * sizeof(struct sock *));
+
+	goal = (((1UL << order) * PAGE_SIZE) / sizeof(struct tcp_bind_bucket *));
+	if (goal > (64 * 1024)) {
+		/* Don't size the bind-hash larger than the port
+		 * space, that is just silly.
+		 */
+		goal = (((64 * 1024) * sizeof(struct tcp_bind_bucket *)) / PAGE_SIZE);
+		for (order = 0; (1UL << order) < goal; order++)
+			;
+	}
+
+	do {
+		tcp_bhash_size = (1UL << order) * PAGE_SIZE /
+			sizeof(struct tcp_bind_bucket *);
+		tcp_bhash = (struct tcp_bind_bucket **)
+			__get_free_pages(GFP_ATOMIC, order);
+	} while (tcp_bhash == NULL && --order >= 0);
+
+	if (!tcp_bhash)
+		panic("Failed to allocate TCP bind hash table\n");
+	memset(tcp_bhash, 0, tcp_bhash_size * sizeof(struct tcp_bind_bucket *));
+
+	printk("TCP: Hash tables configured (ehash %d bhash %d)\n",
+	       tcp_ehash_size, tcp_bhash_size);
 }
diff --git a/pfinet/linux-src/net/ipv4/tcp_input.c b/pfinet/linux-src/net/ipv4/tcp_input.c
index a753b128..7f5cc4e7 100644
--- a/pfinet/linux-src/net/ipv4/tcp_input.c
+++ b/pfinet/linux-src/net/ipv4/tcp_input.c
@@ -5,7 +5,7 @@
  *
  *		Implementation of the Transmission Control Protocol(TCP).
  *
- * Version:	$Id: tcp_input.c,v 1.164.2.7 1999/08/13 16:14:27 davem Exp $
+ * Version:	$Id: tcp_input.c,v 1.164.2.8 1999/09/23 19:21:23 davem Exp $
  *
  * Authors:	Ross Biro, <bir7@leland.Stanford.Edu>
  *		Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
@@ -1043,7 +1043,7 @@ static __inline__ void tcp_tw_hashdance(struct sock *sk, struct tcp_tw_bucket *t
 	sk->prot->inuse--;
 
 	/* Step 4: Hash TW into TIMEWAIT half of established hash table. */
-	head = &tcp_established_hash[sk->hashent + (TCP_HTABLE_SIZE/2)];
+	head = &tcp_ehash[sk->hashent + (tcp_ehash_size/2)];
 	sktw = (struct sock *)tw;
 	if((sktw->next = *head) != NULL)
 		(*head)->pprev = &sktw->next;
diff --git a/pfinet/linux-src/net/ipv4/tcp_ipv4.c b/pfinet/linux-src/net/ipv4/tcp_ipv4.c
index 904f8b5d..1864f096 100644
--- a/pfinet/linux-src/net/ipv4/tcp_ipv4.c
+++ b/pfinet/linux-src/net/ipv4/tcp_ipv4.c
@@ -5,7 +5,7 @@
  *
  *		Implementation of the Transmission Control Protocol(TCP).
  *
- * Version:	$Id: tcp_ipv4.c,v 1.175.2.10 1999/08/13 16:14:35 davem Exp $
+ * Version:	$Id: tcp_ipv4.c,v 1.175.2.13 1999/11/16 06:33:53 davem Exp $
  *
  *		IPv4 specific functions
  *
@@ -90,12 +90,14 @@ void tcp_v4_send_check(struct sock *sk, struct tcphdr *th, int len,
  * First half of the table is for sockets not in TIME_WAIT, second half
  * is for TIME_WAIT sockets only.
  */
-struct sock *tcp_established_hash[TCP_HTABLE_SIZE];
+unsigned int tcp_ehash_size;
+struct sock **tcp_ehash;
 
 /* Ok, let's try this, I give up, we do need a local binding
  * TCP hash as well as the others for fast bind/connect.
  */
-struct tcp_bind_bucket *tcp_bound_hash[TCP_BHTABLE_SIZE];
+unsigned int tcp_bhash_size;
+struct tcp_bind_bucket **tcp_bhash;
 
 /* All sockets in TCP_LISTEN state will be in here.  This is the only table
  * where wildcard'd TCP sockets can exist.  Hash function here is just local
@@ -117,7 +119,7 @@ int tcp_port_rover = (1024 - 1);
 static __inline__ int tcp_hashfn(__u32 laddr, __u16 lport,
 				 __u32 faddr, __u16 fport)
 {
-	return ((laddr ^ lport) ^ (faddr ^ fport)) & ((TCP_HTABLE_SIZE/2) - 1);
+	return ((laddr ^ lport) ^ (faddr ^ fport)) & ((tcp_ehash_size/2) - 1);
 }
 
 static __inline__ int tcp_sk_hashfn(struct sock *sk)
@@ -140,7 +142,7 @@ struct tcp_bind_bucket *tcp_bucket_create(unsigned short snum)
 	tb = kmem_cache_alloc(tcp_bucket_cachep, SLAB_ATOMIC);
 	if(tb != NULL) {
 		struct tcp_bind_bucket **head =
-			&tcp_bound_hash[tcp_bhashfn(snum)];
+			&tcp_bhash[tcp_bhashfn(snum)];
 		tb->port = snum;
 		tb->fastreuse = 0;
 		tb->owners = NULL;
@@ -162,7 +164,7 @@ static __inline__ int __tcp_bucket_check(unsigned short snum)
 {
 	struct tcp_bind_bucket *tb;
 
-	tb = tcp_bound_hash[tcp_bhashfn(snum)];
+	tb = tcp_bhash[tcp_bhashfn(snum)];
 	for( ; (tb && (tb->port != snum)); tb = tb->next)
 		;
 	if (tb == NULL) {
@@ -180,8 +182,8 @@ static __inline__ void __tcp_inherit_port(struct sock *sk, struct sock *child)
 
 #ifdef CONFIG_IP_TRANSPARENT_PROXY
 	if (child->num != sk->num) {
-		unsigned short snum = ntohs(child->num);
-		for(tb = tcp_bound_hash[tcp_bhashfn(snum)];
+		unsigned short snum = child->num;
+		for(tb = tcp_bhash[tcp_bhashfn(snum)];
 		    tb && tb->port != snum;
 		    tb = tb->next)
 			;
@@ -220,7 +222,7 @@ static int tcp_v4_get_port(struct sock *sk, unsigned short snum)
 		do {	rover++;
 			if ((rover < low) || (rover > high))
 				rover = low;
-			tb = tcp_bound_hash[tcp_bhashfn(rover)];
+			tb = tcp_bhash[tcp_bhashfn(rover)];
 			for ( ; tb; tb = tb->next)
 				if (tb->port == rover)
 					goto next;
@@ -237,7 +239,7 @@ static int tcp_v4_get_port(struct sock *sk, unsigned short snum)
 		snum = rover;
 		tb = NULL;
 	} else {
-		for (tb = tcp_bound_hash[tcp_bhashfn(snum)];
+		for (tb = tcp_bhash[tcp_bhashfn(snum)];
 		     tb != NULL;
 		     tb = tb->next)
 			if (tb->port == snum)
@@ -328,7 +330,7 @@ static __inline__ void __tcp_v4_hash(struct sock *sk)
 	if(sk->state == TCP_LISTEN)
 		skp = &tcp_listening_hash[tcp_sk_listen_hashfn(sk)];
 	else
-		skp = &tcp_established_hash[(sk->hashent = tcp_sk_hashfn(sk))];
+		skp = &tcp_ehash[(sk->hashent = tcp_sk_hashfn(sk))];
 
 	if((sk->next = *skp) != NULL)
 		(*skp)->pprev = &sk->next;
@@ -421,7 +423,7 @@ static inline struct sock *__tcp_v4_lookup(struct tcphdr *th,
 	 * have wildcards anyways.
 	 */
 	hash = tcp_hashfn(daddr, hnum, saddr, sport);
-	for(sk = tcp_established_hash[hash]; sk; sk = sk->next) {
+	for(sk = tcp_ehash[hash]; sk; sk = sk->next) {
 		if(TCP_IPV4_MATCH(sk, acookie, saddr, daddr, ports, dif)) {
 			if (sk->state == TCP_ESTABLISHED)
 				TCP_RHASH(sport) = sk;
@@ -429,7 +431,7 @@ static inline struct sock *__tcp_v4_lookup(struct tcphdr *th,
 		}
 	}
 	/* Must check for a TIME_WAIT'er before going to listener hash. */
-	for(sk = tcp_established_hash[hash+(TCP_HTABLE_SIZE/2)]; sk; sk = sk->next)
+	for(sk = tcp_ehash[hash+(tcp_ehash_size/2)]; sk; sk = sk->next)
 		if(TCP_IPV4_MATCH(sk, acookie, saddr, daddr, ports, dif))
 			goto hit;
 	sk = tcp_v4_lookup_listener(daddr, hnum, dif);
@@ -469,7 +471,7 @@ static struct sock *tcp_v4_proxy_lookup(unsigned short num, unsigned long raddr,
 
 	/* This code must run only from NET_BH. */
 	{
-		struct tcp_bind_bucket *tb = tcp_bound_hash[tcp_bhashfn(hnum)];
+		struct tcp_bind_bucket *tb = tcp_bhash[tcp_bhashfn(hnum)];
 		for( ; (tb && tb->port != hnum); tb = tb->next)
 			;
 		if(tb == NULL)
@@ -510,7 +512,7 @@ pass2:
 	}
 next:
 	if(firstpass--) {
-		struct tcp_bind_bucket *tb = tcp_bound_hash[tcp_bhashfn(hpnum)];
+		struct tcp_bind_bucket *tb = tcp_bhash[tcp_bhashfn(hpnum)];
 		for( ; (tb && tb->port != hpnum); tb = tb->next)
 			;
 		if(tb) {
@@ -547,7 +549,7 @@ static int tcp_v4_unique_address(struct sock *sk)
 
 	/* Freeze the hash while we snoop around. */
 	SOCKHASH_LOCK();
-	tb = tcp_bound_hash[tcp_bhashfn(snum)];
+	tb = tcp_bhash[tcp_bhashfn(snum)];
 	for(; tb; tb = tb->next) {
 		if(tb->port == snum && tb->owners != NULL) {
 			/* Almost certainly the re-use port case, search the real hashes
@@ -1615,7 +1617,8 @@ static inline struct sock *tcp_v4_hnd_req(struct sock *sk,struct sk_buff *skb)
 			sk = tcp_check_req(sk, skb, req);
 		}
 #ifdef CONFIG_SYN_COOKIES
-		else if (flg == __constant_htonl(0x00120000))  {
+		else if ((flg & __constant_htonl(0x00120000))==__constant_htonl(0x00100000))
+		{
 			sk = cookie_v4_check(sk, skb, &(IPCB(skb)->opt));
 		}
 #endif
@@ -1793,7 +1796,7 @@ do_time_wait:
 
 static void __tcp_v4_rehash(struct sock *sk)
 {
-	struct sock **skp = &tcp_established_hash[(sk->hashent = tcp_sk_hashfn(sk))];
+	struct sock **skp = &tcp_ehash[(sk->hashent = tcp_sk_hashfn(sk))];
 
 	SOCKHASH_LOCK();
 	if(sk->pprev) {
diff --git a/pfinet/linux-src/net/ipv4/tcp_timer.c b/pfinet/linux-src/net/ipv4/tcp_timer.c
index 21029f8e..7529a441 100644
--- a/pfinet/linux-src/net/ipv4/tcp_timer.c
+++ b/pfinet/linux-src/net/ipv4/tcp_timer.c
@@ -5,7 +5,7 @@
  *
  *		Implementation of the Transmission Control Protocol(TCP).
  *
- * Version:	$Id: tcp_timer.c,v 1.62.2.3 1999/06/20 20:14:30 davem Exp $
+ * Version:	$Id: tcp_timer.c,v 1.62.2.4 1999/09/23 19:21:39 davem Exp $
  *
  * Authors:	Ross Biro, <bir7@leland.Stanford.Edu>
  *		Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
@@ -363,8 +363,8 @@ static void tcp_keepalive(unsigned long data)
 	int count = 0;
 	int i;
 	
-	for(i = chain_start; i < (chain_start + ((TCP_HTABLE_SIZE/2) >> 2)); i++) {
-		struct sock *sk = tcp_established_hash[i];
+	for(i = chain_start; i < (chain_start + ((tcp_ehash_size/2) >> 2)); i++) {
+		struct sock *sk = tcp_ehash[i];
 		while(sk) {
 			if(!atomic_read(&sk->sock_readers) && sk->keepopen) {
 				count += tcp_keepopen_proc(sk);
@@ -375,8 +375,8 @@ static void tcp_keepalive(unsigned long data)
 		}
 	}
 out:
-	chain_start = ((chain_start + ((TCP_HTABLE_SIZE/2)>>2)) &
-		       ((TCP_HTABLE_SIZE/2) - 1));
+	chain_start = ((chain_start + ((tcp_ehash_size/2)>>2)) &
+		       ((tcp_ehash_size/2) - 1));
 }
 
 /*