Merge branch 'tcp-active-reset'

Jason Xing says:

===================
tcp: completely support active reset

This time the patch series finally covers all the cases in the active
reset logic. After this, we can know the related exact reason(s).

v4
Link:
1. revise the changelog to avoid future confusion in patch [5/7] (Eric)
2. revise the changelog of patch [6/7] like above.
3. add reviewed-by tags (Eric)

v3
Link: https://lore.kernel.org/all/20240731120955.23542-1-kerneljasonxing@gmail.com/
1. introduce TCP_DISCONNECT_WITH_DATA reason (Eric)
2. use a better name 'TCP_KEEPALIVE_TIMEOUT' (Eric)
3. add three reviewed-by tags (Eric)

v2
Link: https://lore.kernel.org/all/20240730133513.99986-1-kerneljasonxing@gmail.com/
1. use RFC 9293 in the comment and changelog instead of old RFC 793
2. correct the comment and changelog in patch 5
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

include/net/rstreason.h

@@ -17,6 +17,12 @@
 	FN(TCP_ABORT_ON_DATA)		\
 	FN(TCP_TIMEWAIT_SOCKET)		\
 	FN(INVALID_SYN)			\
+	FN(TCP_ABORT_ON_CLOSE)		\
+	FN(TCP_ABORT_ON_LINGER)		\
+	FN(TCP_ABORT_ON_MEMORY)		\
+	FN(TCP_STATE)			\
+	FN(TCP_KEEPALIVE_TIMEOUT)	\
+	FN(TCP_DISCONNECT_WITH_DATA)	\
 	FN(MPTCP_RST_EUNSPEC)		\
 	FN(MPTCP_RST_EMPTCP)		\
 	FN(MPTCP_RST_ERESOURCE)		\
@@ -84,6 +90,39 @@ enum sk_rst_reason {
 	 * an error, send a reset"
 	 */
 	SK_RST_REASON_INVALID_SYN,
+	/**
+	 * @SK_RST_REASON_TCP_ABORT_ON_CLOSE: abort on close
+	 * corresponding to LINUX_MIB_TCPABORTONCLOSE
+	 */
+	SK_RST_REASON_TCP_ABORT_ON_CLOSE,
+	/**
+	 * @SK_RST_REASON_TCP_ABORT_ON_LINGER: abort on linger
+	 * corresponding to LINUX_MIB_TCPABORTONLINGER
+	 */
+	SK_RST_REASON_TCP_ABORT_ON_LINGER,
+	/**
+	 * @SK_RST_REASON_TCP_ABORT_ON_MEMORY: abort on memory
+	 * corresponding to LINUX_MIB_TCPABORTONMEMORY
+	 */
+	SK_RST_REASON_TCP_ABORT_ON_MEMORY,
+	/**
+	 * @SK_RST_REASON_TCP_STATE: abort on tcp state
+	 * Please see RFC 9293 for all possible reset conditions
+	 */
+	SK_RST_REASON_TCP_STATE,
+	/**
+	 * @SK_RST_REASON_TCP_KEEPALIVE_TIMEOUT: time to timeout
+	 * When we have already run out of all the chances, which means
+	 * keepalive timeout, we have to reset the connection
+	 */
+	SK_RST_REASON_TCP_KEEPALIVE_TIMEOUT,
+	/**
+	 * @SK_RST_REASON_TCP_DISCONNECT_WITH_DATA: disconnect when write
+	 * queue is not empty
+	 * It means user has written data into the write queue when doing
+	 * disconnecting, so we have to send an RST.
+	 */
+	SK_RST_REASON_TCP_DISCONNECT_WITH_DATA,
 
 	/* Copy from include/uapi/linux/mptcp.h.
 	 * These reset fields will not be changed since they adhere to

net/ipv4/tcp.c

@@ -2833,7 +2833,7 @@ void __tcp_close(struct sock *sk, long timeout)
 		NET_INC_STATS(sock_net(sk), LINUX_MIB_TCPABORTONCLOSE);
 		tcp_set_state(sk, TCP_CLOSE);
 		tcp_send_active_reset(sk, sk->sk_allocation,
-				      SK_RST_REASON_NOT_SPECIFIED);
+				      SK_RST_REASON_TCP_ABORT_ON_CLOSE);
 	} else if (sock_flag(sk, SOCK_LINGER) && !sk->sk_lingertime) {
 		/* Check zero linger _after_ checking for unread data. */
 		sk->sk_prot->disconnect(sk, 0);
@@ -2908,7 +2908,7 @@ void __tcp_close(struct sock *sk, long timeout)
 		if (READ_ONCE(tp->linger2) < 0) {
 			tcp_set_state(sk, TCP_CLOSE);
 			tcp_send_active_reset(sk, GFP_ATOMIC,
-					      SK_RST_REASON_NOT_SPECIFIED);
+					      SK_RST_REASON_TCP_ABORT_ON_LINGER);
 			__NET_INC_STATS(sock_net(sk),
 					LINUX_MIB_TCPABORTONLINGER);
 		} else {
@@ -2927,7 +2927,7 @@ void __tcp_close(struct sock *sk, long timeout)
 		if (tcp_check_oom(sk, 0)) {
 			tcp_set_state(sk, TCP_CLOSE);
 			tcp_send_active_reset(sk, GFP_ATOMIC,
-					      SK_RST_REASON_NOT_SPECIFIED);
+					      SK_RST_REASON_TCP_ABORT_ON_MEMORY);
 			__NET_INC_STATS(sock_net(sk),
 					LINUX_MIB_TCPABORTONMEMORY);
 		} else if (!check_net(sock_net(sk))) {
@@ -3025,13 +3025,16 @@ int tcp_disconnect(struct sock *sk, int flags)
 		inet_csk_listen_stop(sk);
 	} else if (unlikely(tp->repair)) {
 		WRITE_ONCE(sk->sk_err, ECONNABORTED);
-	} else if (tcp_need_reset(old_state) ||
-		   (tp->snd_nxt != tp->write_seq &&
-		    (1 << old_state) & (TCPF_CLOSING | TCPF_LAST_ACK))) {
+	} else if (tcp_need_reset(old_state)) {
+		tcp_send_active_reset(sk, gfp_any(), SK_RST_REASON_TCP_STATE);
+		WRITE_ONCE(sk->sk_err, ECONNRESET);
+	} else if (tp->snd_nxt != tp->write_seq &&
+		   (1 << old_state) & (TCPF_CLOSING | TCPF_LAST_ACK)) {
 		/* The last check adjusts for discrepancy of Linux wrt. RFC
 		 * states
 		 */
-		tcp_send_active_reset(sk, gfp_any(), SK_RST_REASON_NOT_SPECIFIED);
+		tcp_send_active_reset(sk, gfp_any(),
+				      SK_RST_REASON_TCP_DISCONNECT_WITH_DATA);
 		WRITE_ONCE(sk->sk_err, ECONNRESET);
 	} else if (old_state == TCP_SYN_SENT)
 		WRITE_ONCE(sk->sk_err, ECONNRESET);
@@ -4649,7 +4652,7 @@ int tcp_abort(struct sock *sk, int err)
 	if (!sock_flag(sk, SOCK_DEAD)) {
 		if (tcp_need_reset(sk->sk_state))
 			tcp_send_active_reset(sk, GFP_ATOMIC,
-					      SK_RST_REASON_NOT_SPECIFIED);
+					      SK_RST_REASON_TCP_STATE);
 		tcp_done_with_error(sk, err);
 	}
 

net/ipv4/tcp_output.c

@@ -3649,7 +3649,7 @@ void tcp_send_active_reset(struct sock *sk, gfp_t priority,
 	/* skb of trace_tcp_send_reset() keeps the skb that caused RST,
 	 * skb here is different to the troublesome skb, so use NULL
 	 */
-	trace_tcp_send_reset(sk, NULL, SK_RST_REASON_NOT_SPECIFIED);
+	trace_tcp_send_reset(sk, NULL, reason);
 }
 
 /* Send a crossed SYN-ACK during socket establishment.

net/ipv4/tcp_timer.c

@@ -125,7 +125,7 @@ static int tcp_out_of_resources(struct sock *sk, bool do_reset)
 			do_reset = true;
 		if (do_reset)
 			tcp_send_active_reset(sk, GFP_ATOMIC,
-					      SK_RST_REASON_NOT_SPECIFIED);
+					      SK_RST_REASON_TCP_ABORT_ON_MEMORY);
 		tcp_done(sk);
 		__NET_INC_STATS(sock_net(sk), LINUX_MIB_TCPABORTONMEMORY);
 		return 1;
@@ -779,7 +779,7 @@ static void tcp_keepalive_timer (struct timer_list *t)
 				goto out;
 			}
 		}
-		tcp_send_active_reset(sk, GFP_ATOMIC, SK_RST_REASON_NOT_SPECIFIED);
+		tcp_send_active_reset(sk, GFP_ATOMIC, SK_RST_REASON_TCP_STATE);
 		goto death;
 	}
 
@@ -807,7 +807,7 @@ static void tcp_keepalive_timer (struct timer_list *t)
 		    (user_timeout == 0 &&
 		    icsk->icsk_probes_out >= keepalive_probes(tp))) {
 			tcp_send_active_reset(sk, GFP_ATOMIC,
-					      SK_RST_REASON_NOT_SPECIFIED);
+					      SK_RST_REASON_TCP_KEEPALIVE_TIMEOUT);
 			tcp_write_err(sk);
 			goto out;
 		}