[NET]: Cleanup DST metrics and abstract MSS/PMTU further.

- Changed dst named metrics, to RTAX_MAX metrics array.
- Add inline shorthands to access them
- Add update_pmtu and get_mss to DST ops.
- Add path component to DST, it is DST itself by default.

5 files changed, 46 insertions, 15 deletions

diff --git a/include/net/dst.h b/include/net/dst.h
index faf5e30316be..44d7374573f2 100644
--- a/include/net/dst.h
+++ b/include/net/dst.h
@@ -9,6 +9,7 @@
 #define _NET_DST_H
 
 #include <linux/config.h>
+#include <linux/rtnetlink.h>
 #include <net/neighbour.h>
 
 /*
@@ -22,6 +23,13 @@
 #define DST_GC_INC	(5*HZ)
 #define DST_GC_MAX	(120*HZ)
 
+/* Each dst_entry has reference count and sits in some parent list(s).
+ * When it is removed from parent list, it is "freed" (dst_free).
+ * After this it enters dead state (dst->obsolete > 0) and if its refcnt
+ * is zero, it can be destroyed immediately, otherwise it is added
+ * to gc list and garbage collector periodically checks the refcnt.
+ */
+
 struct sk_buff;
 
 struct dst_entry
@@ -39,15 +47,8 @@ struct dst_entry
 
 	unsigned		header_len;	/* more space at head required */
 
-	unsigned		mxlock;
-	unsigned		pmtu;
-	unsigned		window;
-	unsigned		rtt;
-	unsigned		rttvar;
-	unsigned		ssthresh;
-	unsigned		cwnd;
-	unsigned		advmss;
-	unsigned		reordering;
+	u32			metrics[RTAX_MAX];
+	struct dst_entry	*path;
 
 	unsigned long		rate_last;	/* rate limiting for ICMP */
 	unsigned long		rate_tokens;
@@ -81,6 +82,8 @@ struct dst_ops
 	void			(*destroy)(struct dst_entry *);
 	struct dst_entry *	(*negative_advice)(struct dst_entry *);
 	void			(*link_failure)(struct sk_buff *);
+	void			(*update_pmtu)(struct dst_entry *dst, u32 mtu);
+	int			(*get_mss)(struct dst_entry *dst, u32 mtu);
 	int			entry_size;
 
 	atomic_t		entries;
@@ -89,6 +92,33 @@ struct dst_ops
 
 #ifdef __KERNEL__
 
+static inline u32
+dst_metric(struct dst_entry *dst, int metric)
+{
+	return dst->metrics[metric-1];
+}
+
+static inline u32
+dst_path_metric(struct dst_entry *dst, int metric)
+{
+	return dst->path->metrics[metric-1];
+}
+
+static inline u32
+dst_pmtu(struct dst_entry *dst)
+{
+	u32 mtu = dst_path_metric(dst, RTAX_MTU);
+	/* Yes, _exactly_. This is paranoia. */
+	barrier();
+	return mtu;
+}
+
+static inline int
+dst_metric_locked(struct dst_entry *dst, int metric)
+{
+	return dst_metric(dst, RTAX_LOCK) & (1<<metric);
+}
+
 static inline void dst_hold(struct dst_entry * dst)
 {
 	atomic_inc(&dst->__refcnt);
diff --git a/include/net/ip.h b/include/net/ip.h
index af74bb3f79d9..7b365788fe00 100644
--- a/include/net/ip.h
+++ b/include/net/ip.h
@@ -178,7 +178,7 @@ int ip_dont_fragment(struct sock *sk, struct dst_entry *dst)
 {
 	return (inet_sk(sk)->pmtudisc == IP_PMTUDISC_DO ||
 		(inet_sk(sk)->pmtudisc == IP_PMTUDISC_WANT &&
-		 !(dst->mxlock&(1<<RTAX_MTU))));
+		 !(dst_metric(dst, RTAX_LOCK)&(1<<RTAX_MTU))));
 }
 
 extern void __ip_select_ident(struct iphdr *iph, struct dst_entry *dst, int more);
diff --git a/include/net/ip_fib.h b/include/net/ip_fib.h
index debc59c4c240..0d847352d121 100644
--- a/include/net/ip_fib.h
+++ b/include/net/ip_fib.h
@@ -67,7 +67,7 @@ struct fib_info
 	int			fib_protocol;
 	u32			fib_prefsrc;
 	u32			fib_priority;
-	unsigned		fib_metrics[RTAX_MAX];
+	u32			fib_metrics[RTAX_MAX];
 #define fib_mtu fib_metrics[RTAX_MTU-1]
 #define fib_window fib_metrics[RTAX_WINDOW-1]
 #define fib_rtt fib_metrics[RTAX_RTT-1]
diff --git a/include/net/route.h b/include/net/route.h
index f74195ea16ea..eea5c8f0269f 100644
--- a/include/net/route.h
+++ b/include/net/route.h
@@ -115,7 +115,6 @@ extern void		rt_cache_flush(int how);
 extern int		ip_route_output_key(struct rtable **, const struct flowi *flp);
 extern int		ip_route_input(struct sk_buff*, u32 dst, u32 src, u8 tos, struct net_device *devin);
 extern unsigned short	ip_rt_frag_needed(struct iphdr *iph, unsigned short new_mtu);
-extern void		ip_rt_update_pmtu(struct dst_entry *dst, unsigned mtu);
 extern void		ip_rt_send_redirect(struct sk_buff *skb);
 
 extern unsigned		inet_addr_type(u32 addr);
diff --git a/include/net/tcp.h b/include/net/tcp.h
index 6683b8379936..38cc0e90793b 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -921,9 +921,11 @@ static __inline__ unsigned int tcp_current_mss(struct sock *sk, int large)
 	int mss_now = large && (sk->route_caps&NETIF_F_TSO) && !tp->urg_mode ?
 		tp->mss_cache : tp->mss_cache_std;
 
-	if (dst && dst->pmtu != tp->pmtu_cookie)
- 		mss_now = tcp_sync_mss(sk, dst->pmtu);
-
+	if (dst) {
+		u32 mtu = dst_pmtu(dst);
+		if (mtu != tp->pmtu_cookie)
+			mss_now = tcp_sync_mss(sk, mtu);
+	}
 	if (tp->eff_sacks)
 		mss_now -= (TCPOLEN_SACK_BASE_ALIGNED +
 			    (tp->eff_sacks * TCPOLEN_SACK_PERBLOCK));