Merge branch 'net-cleanup-datagram-receive-helpers'

Paolo Abeni says:

====================
net: cleanup datagram receive helpers

Several receive helpers have an optional destructor argument, which uglify
the code a bit and is taxed by retpoline overhead.

This series refactor the code so that we can drop such optional argument,
cleaning the helpers a bit and avoiding an indirect call in fast path.

The first patch refactor a bit the caller, so that the second patch
actually dropping the argument is more straight-forward

v1 -> v2:
 - call scm_stat_del() only when not peeking - Kirill
 - fix build issue with CONFIG_INET_ESPINTCP
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

include/linux/skbuff.h

@@ -3514,23 +3514,15 @@ int __skb_wait_for_more_packets(struct sock *sk, struct sk_buff_head *queue,
 struct sk_buff *__skb_try_recv_from_queue(struct sock *sk,
 					  struct sk_buff_head *queue,
 					  unsigned int flags,
-					  void (*destructor)(struct sock *sk,
-							   struct sk_buff *skb),
 					  int *off, int *err,
 					  struct sk_buff **last);
 struct sk_buff *__skb_try_recv_datagram(struct sock *sk,
 					struct sk_buff_head *queue,
-					unsigned int flags,
-					void (*destructor)(struct sock *sk,
-							   struct sk_buff *skb),
-					int *off, int *err,
+					unsigned int flags, int *off, int *err,
 					struct sk_buff **last);
 struct sk_buff *__skb_recv_datagram(struct sock *sk,
 				    struct sk_buff_head *sk_queue,
-				    unsigned int flags,
-				    void (*destructor)(struct sock *sk,
-						       struct sk_buff *skb),
-				    int *off, int *err);
+				    unsigned int flags, int *off, int *err);
 struct sk_buff *skb_recv_datagram(struct sock *sk, unsigned flags, int noblock,
 				  int *err);
 __poll_t datagram_poll(struct file *file, struct socket *sock,

include/net/af_unix.h

@@ -42,7 +42,7 @@ struct unix_skb_parms {
 } __randomize_layout;
 
 struct scm_stat {
-	u32 nr_fds;
+	atomic_t nr_fds;
 };
 
 #define UNIXCB(skb)	(*(struct unix_skb_parms *)&((skb)->cb))

net/core/datagram.c

@@ -166,8 +166,6 @@ static struct sk_buff *skb_set_peeked(struct sk_buff *skb)
 struct sk_buff *__skb_try_recv_from_queue(struct sock *sk,
 					  struct sk_buff_head *queue,
 					  unsigned int flags,
-					  void (*destructor)(struct sock *sk,
-							   struct sk_buff *skb),
 					  int *off, int *err,
 					  struct sk_buff **last)
 {
@@ -198,8 +196,6 @@ struct sk_buff *__skb_try_recv_from_queue(struct sock *sk,
 			refcount_inc(&skb->users);
 		} else {
 			__skb_unlink(skb, queue);
-			if (destructor)
-				destructor(sk, skb);
 		}
 		*off = _off;
 		return skb;
@@ -212,7 +208,6 @@ struct sk_buff *__skb_try_recv_from_queue(struct sock *sk,
  *	@sk: socket
  *	@queue: socket queue from which to receive
  *	@flags: MSG\_ flags
- *	@destructor: invoked under the receive lock on successful dequeue
  *	@off: an offset in bytes to peek skb from. Returns an offset
  *	      within an skb where data actually starts
  *	@err: error code returned
@@ -245,10 +240,7 @@ struct sk_buff *__skb_try_recv_from_queue(struct sock *sk,
  */
 struct sk_buff *__skb_try_recv_datagram(struct sock *sk,
 					struct sk_buff_head *queue,
-					unsigned int flags,
-					void (*destructor)(struct sock *sk,
-							   struct sk_buff *skb),
-					int *off, int *err,
+					unsigned int flags, int *off, int *err,
 					struct sk_buff **last)
 {
 	struct sk_buff *skb;
@@ -269,8 +261,8 @@ struct sk_buff *__skb_try_recv_datagram(struct sock *sk,
 		 * However, this function was correct in any case. 8)
 		 */
 		spin_lock_irqsave(&queue->lock, cpu_flags);
-		skb = __skb_try_recv_from_queue(sk, queue, flags, destructor,
-						off, &error, last);
+		skb = __skb_try_recv_from_queue(sk, queue, flags, off, &error,
+						last);
 		spin_unlock_irqrestore(&queue->lock, cpu_flags);
 		if (error)
 			goto no_packet;
@@ -293,10 +285,7 @@ EXPORT_SYMBOL(__skb_try_recv_datagram);
 
 struct sk_buff *__skb_recv_datagram(struct sock *sk,
 				    struct sk_buff_head *sk_queue,
-				    unsigned int flags,
-				    void (*destructor)(struct sock *sk,
-						       struct sk_buff *skb),
-				    int *off, int *err)
+				    unsigned int flags, int *off, int *err)
 {
 	struct sk_buff *skb, *last;
 	long timeo;
@@ -304,8 +293,8 @@ struct sk_buff *__skb_recv_datagram(struct sock *sk,
 	timeo = sock_rcvtimeo(sk, flags & MSG_DONTWAIT);
 
 	do {
-		skb = __skb_try_recv_datagram(sk, sk_queue, flags, destructor,
-					      off, err, &last);
+		skb = __skb_try_recv_datagram(sk, sk_queue, flags, off, err,
+					      &last);
 		if (skb)
 			return skb;
 
@@ -326,7 +315,7 @@ struct sk_buff *skb_recv_datagram(struct sock *sk, unsigned int flags,
 
 	return __skb_recv_datagram(sk, &sk->sk_receive_queue,
 				   flags | (noblock ? MSG_DONTWAIT : 0),
-				   NULL, &off, err);
+				   &off, err);
 }
 EXPORT_SYMBOL(skb_recv_datagram);
 

net/ipv4/udp.c

@@ -1671,10 +1671,11 @@ struct sk_buff *__skb_recv_udp(struct sock *sk, unsigned int flags,
 		error = -EAGAIN;
 		do {
 			spin_lock_bh(&queue->lock);
-			skb = __skb_try_recv_from_queue(sk, queue, flags,
-							udp_skb_destructor,
-							off, err, &last);
+			skb = __skb_try_recv_from_queue(sk, queue, flags, off,
+							err, &last);
 			if (skb) {
+				if (!(flags & MSG_PEEK))
+					udp_skb_destructor(sk, skb);
 				spin_unlock_bh(&queue->lock);
 				return skb;
 			}
@@ -1692,9 +1693,10 @@ struct sk_buff *__skb_recv_udp(struct sock *sk, unsigned int flags,
 			spin_lock(&sk_queue->lock);
 			skb_queue_splice_tail_init(sk_queue, queue);
 
-			skb = __skb_try_recv_from_queue(sk, queue, flags,
-							udp_skb_dtor_locked,
-							off, err, &last);
+			skb = __skb_try_recv_from_queue(sk, queue, flags, off,
+							err, &last);
+			if (skb && !(flags & MSG_PEEK))
+				udp_skb_dtor_locked(sk, skb);
 			spin_unlock(&sk_queue->lock);
 			spin_unlock_bh(&queue->lock);
 			if (skb)

net/unix/af_unix.c

@@ -690,7 +690,8 @@ static void unix_show_fdinfo(struct seq_file *m, struct socket *sock)
 
 	if (sk) {
 		u = unix_sk(sock->sk);
-		seq_printf(m, "scm_fds: %u\n", READ_ONCE(u->scm_stat.nr_fds));
+		seq_printf(m, "scm_fds: %u\n",
+			   atomic_read(&u->scm_stat.nr_fds));
 	}
 }
 #else
@@ -1602,10 +1603,8 @@ static void scm_stat_add(struct sock *sk, struct sk_buff *skb)
 	struct scm_fp_list *fp = UNIXCB(skb).fp;
 	struct unix_sock *u = unix_sk(sk);
 
-	lockdep_assert_held(&sk->sk_receive_queue.lock);
-
 	if (unlikely(fp && fp->count))
-		u->scm_stat.nr_fds += fp->count;
+		atomic_add(fp->count, &u->scm_stat.nr_fds);
 }
 
 static void scm_stat_del(struct sock *sk, struct sk_buff *skb)
@@ -1613,10 +1612,8 @@ static void scm_stat_del(struct sock *sk, struct sk_buff *skb)
 	struct scm_fp_list *fp = UNIXCB(skb).fp;
 	struct unix_sock *u = unix_sk(sk);
 
-	lockdep_assert_held(&sk->sk_receive_queue.lock);
-
 	if (unlikely(fp && fp->count))
-		u->scm_stat.nr_fds -= fp->count;
+		atomic_sub(fp->count, &u->scm_stat.nr_fds);
 }
 
 /*
@@ -1805,10 +1802,8 @@ static int unix_dgram_sendmsg(struct socket *sock, struct msghdr *msg,
 	if (sock_flag(other, SOCK_RCVTSTAMP))
 		__net_timestamp(skb);
 	maybe_add_creds(skb, sock, other);
-	spin_lock(&other->sk_receive_queue.lock);
 	scm_stat_add(other, skb);
-	__skb_queue_tail(&other->sk_receive_queue, skb);
-	spin_unlock(&other->sk_receive_queue.lock);
+	skb_queue_tail(&other->sk_receive_queue, skb);
 	unix_state_unlock(other);
 	other->sk_data_ready(other);
 	sock_put(other);
@@ -1910,10 +1905,8 @@ static int unix_stream_sendmsg(struct socket *sock, struct msghdr *msg,
 			goto pipe_err_free;
 
 		maybe_add_creds(skb, sock, other);
-		spin_lock(&other->sk_receive_queue.lock);
 		scm_stat_add(other, skb);
-		__skb_queue_tail(&other->sk_receive_queue, skb);
-		spin_unlock(&other->sk_receive_queue.lock);
+		skb_queue_tail(&other->sk_receive_queue, skb);
 		unix_state_unlock(other);
 		other->sk_data_ready(other);
 		sent += size;
@@ -2113,9 +2106,12 @@ static int unix_dgram_recvmsg(struct socket *sock, struct msghdr *msg,
 
 		skip = sk_peek_offset(sk, flags);
 		skb = __skb_try_recv_datagram(sk, &sk->sk_receive_queue, flags,
-					      scm_stat_del, &skip, &err, &last);
-		if (skb)
+					      &skip, &err, &last);
+		if (skb) {
+			if (!(flags & MSG_PEEK))
+				scm_stat_del(sk, skb);
 			break;
+		}
 
 		mutex_unlock(&u->iolock);
 
@@ -2409,9 +2405,7 @@ static int unix_stream_read_generic(struct unix_stream_read_state *state,
 			sk_peek_offset_bwd(sk, chunk);
 
 			if (UNIXCB(skb).fp) {
-				spin_lock(&sk->sk_receive_queue.lock);
 				scm_stat_del(sk, skb);
-				spin_unlock(&sk->sk_receive_queue.lock);
 				unix_detach_fds(&scm, skb);
 			}
 

net/xfrm/espintcp.c

@@ -100,7 +100,7 @@ static int espintcp_recvmsg(struct sock *sk, struct msghdr *msg, size_t len,
 
 	flags |= nonblock ? MSG_DONTWAIT : 0;
 
-	skb = __skb_recv_datagram(sk, &ctx->ike_queue, flags, NULL, &off, &err);
+	skb = __skb_recv_datagram(sk, &ctx->ike_queue, flags, &off, &err);
 	if (!skb)
 		return err;