fasync: RCU and fine grained locking

kill_fasync() uses a central rwlock, candidate for RCU conversion, to
avoid cache line ping pongs on SMP.

fasync_remove_entry() and fasync_add_entry() can disable IRQS on a short
section instead during whole list scan.

Use a spinlock per fasync_struct to synchronize kill_fasync_rcu() and
fasync_{remove|add}_entry(). This spinlock is IRQ safe, so sock_fasync()
doesnt need its own implementation and can use fasync_helper(), to
reduce code size and complexity.

We can remove __kill_fasync() direct use in net/socket.c, and rename it
to kill_fasync_rcu().

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Lai Jiangshan <laijs@cn.fujitsu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

1 files changed, 11 insertions, 62 deletions

diff --git a/net/socket.c b/net/socket.c
index 35bc198..9822081 100644
--- a/net/socket.c
+++ b/net/socket.c
@@ -1067,78 +1067,27 @@ static int sock_close(struct inode *inode, struct file *filp)
  *	1. fasync_list is modified only under process context socket lock
  *	   i.e. under semaphore.
  *	2. fasync_list is used under read_lock(&sk->sk_callback_lock)
- *	   or under socket lock.
- *	3. fasync_list can be used from softirq context, so that
- *	   modification under socket lock have to be enhanced with
- *	   write_lock_bh(&sk->sk_callback_lock).
- *							--ANK (990710)
+ *	   or under socket lock
  */
 
 static int sock_fasync(int fd, struct file *filp, int on)
 {
-	struct fasync_struct *fa, *fna = NULL, **prev;
-	struct socket *sock;
-	struct sock *sk;
-
-	if (on) {
-		fna = kmalloc(sizeof(struct fasync_struct), GFP_KERNEL);
-		if (fna == NULL)
-			return -ENOMEM;
-	}
-
-	sock = filp->private_data;
+	struct socket *sock = filp->private_data;
+	struct sock *sk = sock->sk;
 
-	sk = sock->sk;
-	if (sk == NULL) {
-		kfree(fna);
+	if (sk == NULL)
 		return -EINVAL;
-	}
 
 	lock_sock(sk);
 
-	spin_lock(&filp->f_lock);
-	if (on)
-		filp->f_flags |= FASYNC;
-	else
-		filp->f_flags &= ~FASYNC;
-	spin_unlock(&filp->f_lock);
-
-	prev = &(sock->fasync_list);
+	fasync_helper(fd, filp, on, &sock->fasync_list);
 
-	for (fa = *prev; fa != NULL; prev = &fa->fa_next, fa = *prev)
-		if (fa->fa_file == filp)
-			break;
-
-	if (on) {
-		if (fa != NULL) {
-			write_lock_bh(&sk->sk_callback_lock);
-			fa->fa_fd = fd;
-			write_unlock_bh(&sk->sk_callback_lock);
-
-			kfree(fna);
-			goto out;
-		}
-		fna->fa_file = filp;
-		fna->fa_fd = fd;
-		fna->magic = FASYNC_MAGIC;
-		fna->fa_next = sock->fasync_list;
-		write_lock_bh(&sk->sk_callback_lock);
-		sock->fasync_list = fna;
+	if (!sock->fasync_list)
+		sock_reset_flag(sk, SOCK_FASYNC);
+	else
 		sock_set_flag(sk, SOCK_FASYNC);
-		write_unlock_bh(&sk->sk_callback_lock);
-	} else {
-		if (fa != NULL) {
-			write_lock_bh(&sk->sk_callback_lock);
-			*prev = fa->fa_next;
-			if (!sock->fasync_list)
-				sock_reset_flag(sk, SOCK_FASYNC);
-			write_unlock_bh(&sk->sk_callback_lock);
-			kfree(fa);
-		}
-	}
 
-out:
-	release_sock(sock->sk);
+	release_sock(sk);
 	return 0;
 }
 
@@ -1159,10 +1108,10 @@ int sock_wake_async(struct socket *sock, int how, int band)
 		/* fall through */
 	case SOCK_WAKE_IO:
 call_kill:
-		__kill_fasync(sock->fasync_list, SIGIO, band);
+		kill_fasync(&sock->fasync_list, SIGIO, band);
 		break;
 	case SOCK_WAKE_URG:
-		__kill_fasync(sock->fasync_list, SIGURG, band);
+		kill_fasync(&sock->fasync_list, SIGURG, band);
 	}
 	return 0;
 }