net: add sk_wake_async_rcu() helper

While looking at UDP receive performance, I saw sk_wake_async()
was no longer inlined.

This matters at least on AMD Zen1-4 platforms (see SRSO)

This might be because rcu_read_lock() and rcu_read_unlock()
are no longer nops in recent kernels ?

Add sk_wake_async_rcu() variant, which must be called from
contexts already holding rcu lock.

As SOCK_FASYNC is deprecated in modern days, use unlikely()
to give a hint to the compiler.

sk_wake_async_rcu() is properly inlined from
__udp_enqueue_schedule_skb() and sock_def_readable().
Signed-off-by: Eric Dumazet <edumazet@google.com>
Link: https://lore.kernel.org/r/20240328144032.1864988-5-edumazet@google.comSigned-off-by: Jakub Kicinski <kuba@kernel.org>

crypto/af_alg.c

@@ -847,7 +847,7 @@ void af_alg_wmem_wakeup(struct sock *sk)
 		wake_up_interruptible_sync_poll(&wq->wait, EPOLLIN |
 							   EPOLLRDNORM |
 							   EPOLLRDBAND);
-	sk_wake_async(sk, SOCK_WAKE_WAITD, POLL_IN);
+	sk_wake_async_rcu(sk, SOCK_WAKE_WAITD, POLL_IN);
 	rcu_read_unlock();
 }
 EXPORT_SYMBOL_GPL(af_alg_wmem_wakeup);
@@ -914,7 +914,7 @@ static void af_alg_data_wakeup(struct sock *sk)
 		wake_up_interruptible_sync_poll(&wq->wait, EPOLLOUT |
 							   EPOLLRDNORM |
 							   EPOLLRDBAND);
-	sk_wake_async(sk, SOCK_WAKE_SPACE, POLL_OUT);
+	sk_wake_async_rcu(sk, SOCK_WAKE_SPACE, POLL_OUT);
 	rcu_read_unlock();
 }
 

include/net/sock.h

@@ -2513,6 +2513,12 @@ static inline void sk_wake_async(const struct sock *sk, int how, int band)
 	}
 }
 
+static inline void sk_wake_async_rcu(const struct sock *sk, int how, int band)
+{
+	if (unlikely(sock_flag(sk, SOCK_FASYNC)))
+		sock_wake_async(rcu_dereference(sk->sk_wq), how, band);
+}
+
 /* Since sk_{r,w}mem_alloc sums skb->truesize, even a small frame might
  * need sizeof(sk_buff) + MTU + padding, unless net driver perform copybreak.
  * Note: for send buffers, TCP works better if we can build two skbs at

net/atm/common.c

@@ -116,7 +116,7 @@ static void vcc_write_space(struct sock *sk)
 		if (skwq_has_sleeper(wq))
 			wake_up_interruptible(&wq->wait);
 
-		sk_wake_async(sk, SOCK_WAKE_SPACE, POLL_OUT);
+		sk_wake_async_rcu(sk, SOCK_WAKE_SPACE, POLL_OUT);
 	}
 
 	rcu_read_unlock();

net/core/sock.c

@@ -3338,7 +3338,7 @@ static void sock_def_error_report(struct sock *sk)
 	wq = rcu_dereference(sk->sk_wq);
 	if (skwq_has_sleeper(wq))
 		wake_up_interruptible_poll(&wq->wait, EPOLLERR);
-	sk_wake_async(sk, SOCK_WAKE_IO, POLL_ERR);
+	sk_wake_async_rcu(sk, SOCK_WAKE_IO, POLL_ERR);
 	rcu_read_unlock();
 }
 
@@ -3353,7 +3353,7 @@ void sock_def_readable(struct sock *sk)
 	if (skwq_has_sleeper(wq))
 		wake_up_interruptible_sync_poll(&wq->wait, EPOLLIN | EPOLLPRI |
 						EPOLLRDNORM | EPOLLRDBAND);
-	sk_wake_async(sk, SOCK_WAKE_WAITD, POLL_IN);
+	sk_wake_async_rcu(sk, SOCK_WAKE_WAITD, POLL_IN);
 	rcu_read_unlock();
 }
 
@@ -3373,7 +3373,7 @@ static void sock_def_write_space(struct sock *sk)
 						EPOLLWRNORM | EPOLLWRBAND);
 
 		/* Should agree with poll, otherwise some programs break */
-		sk_wake_async(sk, SOCK_WAKE_SPACE, POLL_OUT);
+		sk_wake_async_rcu(sk, SOCK_WAKE_SPACE, POLL_OUT);
 	}
 
 	rcu_read_unlock();
@@ -3398,7 +3398,7 @@ static void sock_def_write_space_wfree(struct sock *sk)
 						EPOLLWRNORM | EPOLLWRBAND);
 
 		/* Should agree with poll, otherwise some programs break */
-		sk_wake_async(sk, SOCK_WAKE_SPACE, POLL_OUT);
+		sk_wake_async_rcu(sk, SOCK_WAKE_SPACE, POLL_OUT);
 	}
 }
 

net/dccp/output.c

@@ -204,7 +204,7 @@ void dccp_write_space(struct sock *sk)
 		wake_up_interruptible(&wq->wait);
 	/* Should agree with poll, otherwise some programs break */
 	if (sock_writeable(sk))
-		sk_wake_async(sk, SOCK_WAKE_SPACE, POLL_OUT);
+		sk_wake_async_rcu(sk, SOCK_WAKE_SPACE, POLL_OUT);
 
 	rcu_read_unlock();
 }

net/ipv4/udp.c

@@ -1544,7 +1544,7 @@ int __udp_enqueue_schedule_skb(struct sock *sk, struct sk_buff *skb)
 			INDIRECT_CALL_1(sk->sk_data_ready,
 					sock_def_readable, sk);
 		else
-			sk_wake_async(sk, SOCK_WAKE_WAITD, POLL_IN);
+			sk_wake_async_rcu(sk, SOCK_WAKE_WAITD, POLL_IN);
 	}
 	busylock_release(busy);
 	return 0;

net/iucv/af_iucv.c

@@ -184,7 +184,7 @@ static void iucv_sock_wake_msglim(struct sock *sk)
 	wq = rcu_dereference(sk->sk_wq);
 	if (skwq_has_sleeper(wq))
 		wake_up_interruptible_all(&wq->wait);
-	sk_wake_async(sk, SOCK_WAKE_SPACE, POLL_OUT);
+	sk_wake_async_rcu(sk, SOCK_WAKE_SPACE, POLL_OUT);
 	rcu_read_unlock();
 }
 

net/rxrpc/af_rxrpc.c

@@ -65,7 +65,7 @@ static void rxrpc_write_space(struct sock *sk)
 
 		if (skwq_has_sleeper(wq))
 			wake_up_interruptible(&wq->wait);
-		sk_wake_async(sk, SOCK_WAKE_SPACE, POLL_OUT);
+		sk_wake_async_rcu(sk, SOCK_WAKE_SPACE, POLL_OUT);
 	}
 	rcu_read_unlock();
 }

net/sctp/socket.c

@@ -9276,7 +9276,7 @@ void sctp_data_ready(struct sock *sk)
 	if (skwq_has_sleeper(wq))
 		wake_up_interruptible_sync_poll(&wq->wait, EPOLLIN |
 						EPOLLRDNORM | EPOLLRDBAND);
-	sk_wake_async(sk, SOCK_WAKE_WAITD, POLL_IN);
+	sk_wake_async_rcu(sk, SOCK_WAKE_WAITD, POLL_IN);
 	rcu_read_unlock();
 }
 

net/smc/smc_rx.c

@@ -42,10 +42,10 @@ static void smc_rx_wake_up(struct sock *sk)
 	if (skwq_has_sleeper(wq))
 		wake_up_interruptible_sync_poll(&wq->wait, EPOLLIN | EPOLLPRI |
 						EPOLLRDNORM | EPOLLRDBAND);
-	sk_wake_async(sk, SOCK_WAKE_WAITD, POLL_IN);
+	sk_wake_async_rcu(sk, SOCK_WAKE_WAITD, POLL_IN);
 	if ((sk->sk_shutdown == SHUTDOWN_MASK) ||
 	    (sk->sk_state == SMC_CLOSED))
-		sk_wake_async(sk, SOCK_WAKE_WAITD, POLL_HUP);
+		sk_wake_async_rcu(sk, SOCK_WAKE_WAITD, POLL_HUP);
 	rcu_read_unlock();
 }
 

net/unix/af_unix.c

@@ -546,7 +546,7 @@ static void unix_write_space(struct sock *sk)
 		if (skwq_has_sleeper(wq))
 			wake_up_interruptible_sync_poll(&wq->wait,
 				EPOLLOUT | EPOLLWRNORM | EPOLLWRBAND);
-		sk_wake_async(sk, SOCK_WAKE_SPACE, POLL_OUT);
+		sk_wake_async_rcu(sk, SOCK_WAKE_SPACE, POLL_OUT);
 	}
 	rcu_read_unlock();
 }