net: avoid RCU for NOCACHE dst

There is no point using RCU for dst we allocate for a very short time
(used once).

Change dst_release() to take DST_NOCACHE into account, but also change
skb_dst_set_noref() to force a refcount increment for such dst.

This is a _huge_ gain, because we dont waste memory to store xx thousand
of dsts. Instead of queueing them to RCU, we can free them instantly.

CPU caches can stay hot, re-using same memory blocks to hold temporary
dsts.

Note : remove unneeded smp_mb__before_atomic_dec(); in dst_release(),
since atomic_dec_return() implies a full memory barrier.

Stress test, 160.000.000 udp frames sent, IP route cache disabled
(DDOS).

Before:

real    0m38.091s
user    0m13.189s
sys     7m53.018s

After:

real	0m29.946s
user	0m12.157s
sys	7m40.605s

For reference, if IP route cache was enabled :

real	0m32.030s
user	0m10.521s
sys	8m15.243s

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

1 files changed, 4 insertions, 5 deletions

diff --git a/net/ipv4/route.c b/net/ipv4/route.c
index ff98983d2a45..d6cb2bfcd8e1 100644
--- a/net/ipv4/route.c
+++ b/net/ipv4/route.c
@@ -1105,9 +1105,9 @@ restart:
 		 * Note that we do rt_free on this new route entry, so that
 		 * once its refcount hits zero, we are still able to reap it
 		 * (Thanks Alexey)
-		 * Note also the rt_free uses call_rcu.  We don't actually
-		 * need rcu protection here, this is just our path to get
-		 * on the route gc list.
+		 * Note: To avoid expensive rcu stuff for this uncached dst,
+		 * we set DST_NOCACHE so that dst_release() can free dst without
+		 * waiting a grace period.
 		 */
 
 		rt->dst.flags |= DST_NOCACHE;
@@ -1117,12 +1117,11 @@ restart:
 				if (net_ratelimit())
 					printk(KERN_WARNING
 					    "Neighbour table failure & not caching routes.\n");
-				rt_drop(rt);
+				ip_rt_put(rt);
 				return err;
 			}
 		}
 
-		rt_free(rt);
 		goto skip_hashing;
 	}