tcp/dccp: remove twchain

TCP listener refactoring, part 3 :

Our goal is to hash SYN_RECV sockets into main ehash for fast lookup,
and parallel SYN processing.

Current inet_ehash_bucket contains two chains, one for ESTABLISH (and
friend states) sockets, another for TIME_WAIT sockets only.

As the hash table is sized to get at most one socket per bucket, it
makes little sense to have separate twchain, as it makes the lookup
slightly more complicated, and doubles hash table memory usage.

If we make sure all socket types have the lookup keys at the same
offsets, we can use a generic and faster lookup. It turns out TIME_WAIT
and ESTABLISHED sockets already have common lookup fields for IPv4.

[ INET_TW_MATCH() is no longer needed ]

I'll provide a follow-up to factorize IPv6 lookup as well, to remove
INET6_TW_MATCH()

This way, SYN_RECV pseudo sockets will be supported the same.

A new sock_gen_put() helper is added, doing either a sock_put() or
inet_twsk_put() [ and will support SYN_RECV later ].

Note this helper should only be called in real slow path, when rcu
lookup found a socket that was moved to another identity (freed/reused
immediately), but could eventually be used in other contexts, like
sock_edemux()

Before patch :

dmesg | grep "TCP established"

TCP established hash table entries: 524288 (order: 11, 8388608 bytes)

After patch :

TCP established hash table entries: 524288 (order: 10, 4194304 bytes)

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

1 files changed, 13 insertions, 35 deletions

diff --git a/net/ipv4/inet_diag.c b/net/ipv4/inet_diag.c
index 22000279efc8..8e1e40653357 100644
--- a/net/ipv4/inet_diag.c
+++ b/net/ipv4/inet_diag.c
@@ -635,12 +635,14 @@ static int inet_csk_diag_dump(struct sock *sk,
 				  cb->nlh->nlmsg_seq, NLM_F_MULTI, cb->nlh);
 }
 
-static int inet_twsk_diag_dump(struct inet_timewait_sock *tw,
+static int inet_twsk_diag_dump(struct sock *sk,
 			       struct sk_buff *skb,
 			       struct netlink_callback *cb,
 			       struct inet_diag_req_v2 *r,
 			       const struct nlattr *bc)
 {
+	struct inet_timewait_sock *tw = inet_twsk(sk);
+
 	if (bc != NULL) {
 		struct inet_diag_entry entry;
 
@@ -911,8 +913,7 @@ skip_listen_ht:
 
 		num = 0;
 
-		if (hlist_nulls_empty(&head->chain) &&
-			hlist_nulls_empty(&head->twchain))
+		if (hlist_nulls_empty(&head->chain))
 			continue;
 
 		if (i > s_i)
@@ -920,7 +921,7 @@ skip_listen_ht:
 
 		spin_lock_bh(lock);
 		sk_nulls_for_each(sk, node, &head->chain) {
-			struct inet_sock *inet = inet_sk(sk);
+			int res;
 
 			if (!net_eq(sock_net(sk), net))
 				continue;
@@ -929,15 +930,19 @@ skip_listen_ht:
 			if (!(r->idiag_states & (1 << sk->sk_state)))
 				goto next_normal;
 			if (r->sdiag_family != AF_UNSPEC &&
-					sk->sk_family != r->sdiag_family)
+			    sk->sk_family != r->sdiag_family)
 				goto next_normal;
-			if (r->id.idiag_sport != inet->inet_sport &&
+			if (r->id.idiag_sport != htons(sk->sk_num) &&
 			    r->id.idiag_sport)
 				goto next_normal;
-			if (r->id.idiag_dport != inet->inet_dport &&
+			if (r->id.idiag_dport != sk->sk_dport &&
 			    r->id.idiag_dport)
 				goto next_normal;
-			if (inet_csk_diag_dump(sk, skb, cb, r, bc) < 0) {
+			if (sk->sk_state == TCP_TIME_WAIT)
+				res = inet_twsk_diag_dump(sk, skb, cb, r, bc);
+			else
+				res = inet_csk_diag_dump(sk, skb, cb, r, bc);
+			if (res < 0) {
 				spin_unlock_bh(lock);
 				goto done;
 			}
@@ -945,33 +950,6 @@ next_normal:
 			++num;
 		}
 
-		if (r->idiag_states & TCPF_TIME_WAIT) {
-			struct inet_timewait_sock *tw;
-
-			inet_twsk_for_each(tw, node,
-				    &head->twchain) {
-				if (!net_eq(twsk_net(tw), net))
-					continue;
-
-				if (num < s_num)
-					goto next_dying;
-				if (r->sdiag_family != AF_UNSPEC &&
-						tw->tw_family != r->sdiag_family)
-					goto next_dying;
-				if (r->id.idiag_sport != tw->tw_sport &&
-				    r->id.idiag_sport)
-					goto next_dying;
-				if (r->id.idiag_dport != tw->tw_dport &&
-				    r->id.idiag_dport)
-					goto next_dying;
-				if (inet_twsk_diag_dump(tw, skb, cb, r, bc) < 0) {
-					spin_unlock_bh(lock);
-					goto done;
-				}
-next_dying:
-				++num;
-			}
-		}
 		spin_unlock_bh(lock);
 	}