Merge nuts.ninka.net:/home/davem/src/BK/network-2.5

into nuts.ninka.net:/home/davem/src/BK/net-2.5

include/net/ip6_route.h

@@ -61,7 +61,11 @@ extern struct rt6_info		*rt6_lookup(struct in6_addr *daddr,
 					    struct in6_addr *saddr,
 					    int oif, int flags);
 
-extern struct rt6_info		*ip6_dst_alloc(void);
+extern struct dst_entry *ndisc_dst_alloc(struct net_device *dev,
+					 struct neighbour *neigh,
+					 int (*output)(struct sk_buff *));
+extern int ndisc_dst_gc(int *more);
+extern void fib6_force_start_gc(void);
 
 /*
  *	support functions for ND

net/compat.c

@@ -405,34 +405,6 @@ static int do_set_attach_filter(int fd, int level, int optname,
 			      sizeof(struct sock_fprog));
 }
 
-static int do_set_icmpv6_filter(int fd, int level, int optname,
-				char *optval, int optlen)
-{
-	struct icmp6_filter kfilter;
-	mm_segment_t old_fs;
-	int ret, i;
-
-	if (optlen < sizeof(kfilter))
-		return -EINVAL;
-	if (copy_from_user(&kfilter, optval, sizeof(kfilter)))
-		return -EFAULT;
-
-	for (i = 0; i < 8; i += 2) {
-		u32 tmp = kfilter.data[i];
-
-		kfilter.data[i] = kfilter.data[i + 1];
-		kfilter.data[i + 1] = tmp;
-	}
-
-	old_fs = get_fs();
-	set_fs(KERNEL_DS);
-	ret = sys_setsockopt(fd, level, optname,
-			     (char *) &kfilter, sizeof(kfilter));
-	set_fs(old_fs);
-
-	return ret;
-}
-
 static int do_set_sock_timeout(int fd, int level, int optname, char *optval, int optlen)
 {
 	struct compat_timeval *up = (struct compat_timeval *) optval;
@@ -465,9 +437,6 @@ asmlinkage long compat_sys_setsockopt(int fd, int level, int optname,
 					    optval, optlen);
 	if (optname == SO_RCVTIMEO || optname == SO_SNDTIMEO)
 		return do_set_sock_timeout(fd, level, optname, optval, optlen);
-	if (level == SOL_ICMPV6 && optname == ICMPV6_FILTER)
-		return do_set_icmpv6_filter(fd, level, optname,
-					    optval, optlen);
 
 	return sys_setsockopt(fd, level, optname, optval, optlen);
 }

net/ipv6/ip6_fib.c

@@ -496,6 +496,12 @@ static __inline__ void fib6_start_gc(struct rt6_info *rt)
 		mod_timer(&ip6_fib_timer, jiffies + ip6_rt_gc_interval);
 }
 
+void fib6_force_start_gc(void)
+{
+	if (ip6_fib_timer.expires == 0)
+		mod_timer(&ip6_fib_timer, jiffies + ip6_rt_gc_interval);
+}
+
 /*
  *	Add routing information to the routing tree.
  *	<destination addr>/<source addr>
@@ -1214,6 +1220,7 @@ void fib6_run_gc(unsigned long dummy)
 
 
 	write_lock_bh(&rt6_lock);
+	ndisc_dst_gc(&gc_args.more);
 	fib6_clean_tree(&ip6_routing_table, fib6_age, 0, NULL);
 	write_unlock_bh(&rt6_lock);
 

net/ipv6/ndisc.c

@@ -402,25 +402,6 @@ static int ndisc_output(struct sk_buff *skb)
 	return -EINVAL;
 }
 
-static inline struct dst_entry *ndisc_dst_alloc(struct net_device *dev, 
-						struct neighbour *neigh)
-{
-	struct rt6_info *rt = ip6_dst_alloc();
-
-	if (unlikely(rt == NULL))
-		goto out;
-
-	rt->rt6i_dev	  = dev;
-	rt->rt6i_nexthop  = neigh;
-	rt->rt6i_expires  = 0;
-	rt->rt6i_flags    = RTF_LOCAL;
-	rt->rt6i_metric   = 0;
-	rt->u.dst.metrics[RTAX_HOPLIMIT-1] = 255;
-	rt->u.dst.output  = ndisc_output;
-out:
-	return (struct dst_entry *)rt;
-}
-
 static inline void ndisc_flow_init(struct flowi *fl, u8 type,
 			    struct in6_addr *saddr, struct in6_addr *daddr)
 {
@@ -463,13 +444,13 @@ static void ndisc_send_na(struct net_device *dev, struct neighbour *neigh,
 
 	ndisc_flow_init(&fl, NDISC_NEIGHBOUR_ADVERTISEMENT, src_addr, daddr);
 
-	dst = ndisc_dst_alloc(dev, neigh);	
+	dst = ndisc_dst_alloc(dev, neigh, ndisc_output);
 	if (!dst)
 		return;
 
 	err = xfrm_lookup(&dst, &fl, NULL, 0);
 	if (err < 0) {
-		dst_free(dst);
+		dst_release(dst);
 		return;
 	}
 
@@ -485,7 +466,7 @@ static void ndisc_send_na(struct net_device *dev, struct neighbour *neigh,
 
 	if (skb == NULL) {
 		ND_PRINTK1("send_na: alloc skb failed\n");
-		dst_free(dst);
+		dst_release(dst);
 		return;
 	}
 
@@ -515,7 +496,6 @@ static void ndisc_send_na(struct net_device *dev, struct neighbour *neigh,
 						 csum_partial((__u8 *) msg, 
 							      len, 0));
 
-	dst_clone(dst);
 	skb->dst = dst;
 	idev = in6_dev_get(dst->dev);
 	dst_output(skb);
@@ -550,10 +530,9 @@ void ndisc_send_ns(struct net_device *dev, struct neighbour *neigh,
 
 	ndisc_flow_init(&fl, NDISC_NEIGHBOUR_SOLICITATION, saddr, daddr);
 
-	dst = ndisc_dst_alloc(dev, neigh);
+	dst = ndisc_dst_alloc(dev, neigh, ndisc_output);
 	if (!dst)
 		return;
-	dst_clone(dst);
 
 	err = xfrm_lookup(&dst, &fl, NULL, 0);
 	if (err < 0) {
@@ -570,6 +549,7 @@ void ndisc_send_ns(struct net_device *dev, struct neighbour *neigh,
 				  1, &err);
 	if (skb == NULL) {
 		ND_PRINTK1("send_ns: alloc skb failed\n");
+		dst_release(dst);
 		return;
 	}
 
@@ -595,7 +575,6 @@ void ndisc_send_ns(struct net_device *dev, struct neighbour *neigh,
 						 csum_partial((__u8 *) msg, 
 							      len, 0));
 	/* send it! */
-	dst_clone(dst);
 	skb->dst = dst;
 	idev = in6_dev_get(dst->dev);
 	dst_output(skb);
@@ -622,10 +601,9 @@ void ndisc_send_rs(struct net_device *dev, struct in6_addr *saddr,
 
 	ndisc_flow_init(&fl, NDISC_ROUTER_SOLICITATION, saddr, daddr);
 
-	dst = ndisc_dst_alloc(dev, NULL);
+	dst = ndisc_dst_alloc(dev, NULL, ndisc_output);
 	if (!dst)
 		return;
-	dst_clone(dst);
 
 	err = xfrm_lookup(&dst, &fl, NULL, 0);
 	if (err < 0) {
@@ -664,7 +642,6 @@ void ndisc_send_rs(struct net_device *dev, struct in6_addr *saddr,
 					   csum_partial((__u8 *) hdr, len, 0));
 
 	/* send it! */
-	dst_clone(dst);
 	skb->dst = dst;
 	idev = in6_dev_get(dst->dev);
 	dst_output(skb);
@@ -1321,7 +1298,6 @@ void ndisc_send_redirect(struct sk_buff *skb, struct neighbour *neigh,
 	if (rt == NULL)
 		return;
 	dst = &rt->u.dst;
-	dst_clone(dst);
 
 	err = xfrm_lookup(&dst, &fl, NULL, 0);
 	if (err) {
@@ -1329,16 +1305,17 @@ void ndisc_send_redirect(struct sk_buff *skb, struct neighbour *neigh,
 		return;
 	}
 
+	rt = (struct rt6_info *) dst;
+
 	if (rt->rt6i_flags & RTF_GATEWAY) {
 		ND_PRINTK1("ndisc_send_redirect: not a neighbour\n");
-		dst_release(&rt->u.dst);
+		dst_release(dst);
 		return;
 	}
-	if (!xrlim_allow(&rt->u.dst, 1*HZ)) {
-		dst_release(&rt->u.dst);
+	if (!xrlim_allow(dst, 1*HZ)) {
+		dst_release(dst);
 		return;
 	}
-	dst_release(&rt->u.dst);
 
 	if (dev->addr_len) {
 		if (neigh->nud_state&NUD_VALID) {
@@ -1348,6 +1325,7 @@ void ndisc_send_redirect(struct sk_buff *skb, struct neighbour *neigh,
 			   We will make it later, when will be sure,
 			   that it is alive.
 			 */
+			dst_release(dst);
 			return;
 		}
 	}
@@ -1366,11 +1344,11 @@ void ndisc_send_redirect(struct sk_buff *skb, struct neighbour *neigh,
 
 	hlen = 0;
 
-	skb_reserve(skb, (dev->hard_header_len + 15) & ~15);
+	skb_reserve(buff, (dev->hard_header_len + 15) & ~15);
 	ip6_nd_hdr(sk, buff, dev, &saddr_buf, &skb->nh.ipv6h->saddr,
 		   IPPROTO_ICMPV6, len);
 
-	skb->h.raw = (unsigned char*) icmph = (struct icmp6hdr *) skb_put(buff, len);
+	buff->h.raw = (unsigned char*) icmph = (struct icmp6hdr *) skb_put(buff, len);
 
 	memset(icmph, 0, sizeof(struct icmp6hdr));
 	icmph->icmp6_type = NDISC_REDIRECT;
@@ -1408,9 +1386,9 @@ void ndisc_send_redirect(struct sk_buff *skb, struct neighbour *neigh,
 					     len, IPPROTO_ICMPV6,
 					     csum_partial((u8 *) icmph, len, 0));
 
-	skb->dst = dst;
+	buff->dst = dst;
 	idev = in6_dev_get(dst->dev);
-	dst_output(skb);
+	dst_output(buff);
 
 	ICMP6_INC_STATS(idev, Icmp6OutRedirects);
 	ICMP6_INC_STATS(idev, Icmp6OutMsgs);

net/ipv6/route.c

@@ -131,16 +131,11 @@ rwlock_t rt6_lock = RW_LOCK_UNLOCKED;
 
 
 /* allocate dst with ip6_dst_ops */
-static __inline__ struct rt6_info *__ip6_dst_alloc(void)
+static __inline__ struct rt6_info *ip6_dst_alloc(void)
 {
 	return dst_alloc(&ip6_dst_ops);
 }
 
-struct rt6_info *ip6_dst_alloc(void)
-{
-	return __ip6_dst_alloc();
-}
-
 /*
  *	Route lookup. Any rt6_lock is implied.
  */
@@ -560,6 +555,60 @@ static void ip6_rt_update_pmtu(struct dst_entry *dst, u32 mtu)
 	}
 }
 
+/* Protected by rt6_lock.  */
+static struct dst_entry *ndisc_dst_gc_list;
+
+struct dst_entry *ndisc_dst_alloc(struct net_device *dev, 
+				  struct neighbour *neigh,
+				  int (*output)(struct sk_buff *))
+{
+	struct rt6_info *rt = ip6_dst_alloc();
+
+	if (unlikely(rt == NULL))
+		goto out;
+
+	rt->rt6i_dev	  = dev;
+	rt->rt6i_nexthop  = neigh;
+	rt->rt6i_expires  = 0;
+	rt->rt6i_flags    = RTF_LOCAL;
+	rt->rt6i_metric   = 0;
+	atomic_set(&rt->u.dst.__refcnt, 1);
+	rt->u.dst.metrics[RTAX_HOPLIMIT-1] = 255;
+	rt->u.dst.output  = output;
+
+	write_lock_bh(&rt6_lock);
+	rt->u.dst.next = ndisc_dst_gc_list;
+	ndisc_dst_gc_list = &rt->u.dst;
+	write_unlock_bh(&rt6_lock);
+
+	fib6_force_start_gc();
+
+out:
+	return (struct dst_entry *)rt;
+}
+
+int ndisc_dst_gc(int *more)
+{
+	struct dst_entry *dst, *next, **pprev;
+	int freed;
+
+	next = NULL;
+	pprev = &ndisc_dst_gc_list;
+	freed = 0;
+	while ((dst = *pprev) != NULL) {
+		if (!atomic_read(&dst->__refcnt)) {
+			*pprev = dst->next;
+			dst_free(dst);
+			freed++;
+		} else {
+			pprev = &dst->next;
+			(*more)++;
+		}
+	}
+
+	return freed;
+}
+
 static int ip6_dst_gc(void)
 {
 	static unsigned expire = 30*HZ;
@@ -655,7 +704,7 @@ int ip6_route_add(struct in6_rtmsg *rtmsg, struct nlmsghdr *nlh, void *_rtattr)
 	if (rtmsg->rtmsg_metric == 0)
 		rtmsg->rtmsg_metric = IP6_RT_PRIO_USER;
 
-	rt = __ip6_dst_alloc();
+	rt = ip6_dst_alloc();
 
 	if (rt == NULL)
 		return -ENOMEM;
@@ -1066,7 +1115,7 @@ void rt6_pmtu_discovery(struct in6_addr *daddr, struct in6_addr *saddr,
 
 static struct rt6_info * ip6_rt_copy(struct rt6_info *ort)
 {
-	struct rt6_info *rt = __ip6_dst_alloc();
+	struct rt6_info *rt = ip6_dst_alloc();
 
 	if (rt) {
 		rt->u.dst.input = ort->u.dst.input;
@@ -1209,7 +1258,7 @@ int ip6_pkt_discard(struct sk_buff *skb)
 
 int ip6_rt_addr_add(struct in6_addr *addr, struct net_device *dev)
 {
-	struct rt6_info *rt = __ip6_dst_alloc();
+	struct rt6_info *rt = ip6_dst_alloc();
 
 	if (rt == NULL)
 		return -ENOMEM;