net: sk_dst_cache RCUification
With latest CONFIG_PROVE_RCU stuff, I felt more comfortable to make this
work.
sk->sk_dst_cache is currently protected by a rwlock (sk_dst_lock)
This rwlock is readlocked for a very small amount of time, and dst
entries are already freed after RCU grace period. This calls for RCU
again :)
This patch converts sk_dst_lock to a spinlock, and use RCU for readers.
__sk_dst_get() is supposed to be called with rcu_read_lock() or if
socket locked by user, so use appropriate rcu_dereference_check()
condition (rcu_read_lock_held() || sock_owned_by_user(sk))
This patch avoids two atomic ops per tx packet on UDP connected sockets,
for example, and permits sk_dst_lock to be much less dirtied.
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
diff --git a/net/decnet/af_decnet.c b/net/decnet/af_decnet.c
index 2b494fa..55e3b6b 100644
--- a/net/decnet/af_decnet.c
+++ b/net/decnet/af_decnet.c
@@ -446,7 +446,7 @@
skb_queue_purge(&scp->other_xmit_queue);
skb_queue_purge(&scp->other_receive_queue);
- dst_release(xchg(&sk->sk_dst_cache, NULL));
+ dst_release(rcu_dereference_check(sk->sk_dst_cache, 1));
}
static int dn_memory_pressure;
@@ -1105,7 +1105,7 @@
release_sock(sk);
dst = skb_dst(skb);
- dst_release(xchg(&newsk->sk_dst_cache, dst));
+ sk_dst_set(newsk, dst);
skb_dst_set(skb, NULL);
DN_SK(newsk)->state = DN_CR;
@@ -1956,7 +1956,7 @@
}
if ((flags & MSG_TRYHARD) && sk->sk_dst_cache)
- dst_negative_advice(&sk->sk_dst_cache, sk);
+ dst_negative_advice(sk);
mss = scp->segsize_rem;
fctype = scp->services_rem & NSP_FC_MASK;