Suggest a socket read buffer of 4KiB to limit message count

...and switch the same in `lightning-net-tokio`

Author: TheBlueMatt

lightning-net-tokio/src/lib.rs

@@ -176,8 +176,9 @@ impl Connection {
 		let (event_waker, event_receiver) = mpsc::channel(1);
 		tokio::spawn(Self::poll_event_process(peer_manager.clone(), event_receiver));
 
-		// 8KB is nice and big but also should never cause any issues with stack overflowing.
-		let mut buf = [0; 8192];
+		// 4KiB is nice and big without handling too many messages all at once, giving other peers
+		// a chance to do some work.
+		let mut buf = [0; 4096];
 
 		let mut our_descriptor = SocketDescriptor::new(us.clone());
 		// An enum describing why we did/are disconnecting:

lightning/src/ln/peer_handler.rs

@@ -1005,6 +1005,9 @@ impl<Descriptor: SocketDescriptor, CM: Deref, RM: Deref, OM: Deref, L: Deref, CM
 	/// [`send_data`] call on this descriptor has `resume_read` set (preventing DoS issues in the
 	/// send buffer).
 	///
+	/// In order to avoid processing too many messages at once per peer, `data` should be on the
+	/// order of 4KiB.
+	///
 	/// [`send_data`]: SocketDescriptor::send_data
 	/// [`process_events`]: PeerManager::process_events
 	pub fn read_event(&self, peer_descriptor: &mut Descriptor, data: &[u8]) -> Result<bool, PeerHandleError> {