diff mbox series

[net-next,v3,4/7] tcp: rstreason: introduce SK_RST_REASON_TCP_STATE for active reset

Message ID 20240801145444.22988-5-kerneljasonxing@gmail.com (mailing list archive)
State Superseded
Delegated to: Netdev Maintainers
Headers show
Series tcp: completely support active reset | expand

Checks

Context Check Description
netdev/series_format success Posting correctly formatted
netdev/tree_selection success Clearly marked for net-next
netdev/ynl success Generated files up to date; no warnings/errors; no diff in generated;
netdev/fixes_present success Fixes tag not required for -next series
netdev/header_inline success No static functions without inline keyword in header files
netdev/build_32bit success Errors and warnings before: 59 this patch: 59
netdev/build_tools success Errors and warnings before: 10 this patch: 10
netdev/cc_maintainers success CCed 5 of 5 maintainers
netdev/build_clang success Errors and warnings before: 84 this patch: 84
netdev/verify_signedoff success Signed-off-by tag matches author and committer
netdev/deprecated_api success None detected
netdev/check_selftest success No net selftest shell script
netdev/verify_fixes success No Fixes tag
netdev/build_allmodconfig_warn success Errors and warnings before: 2229 this patch: 2229
netdev/checkpatch success total: 0 errors, 0 warnings, 0 checks, 48 lines checked
netdev/build_clang_rust success No Rust files in patch. Skipping build
netdev/kdoc success Errors and warnings before: 4 this patch: 4
netdev/source_inline success Was 0 now: 0

Commit Message

Jason Xing Aug. 1, 2024, 2:54 p.m. UTC 
From: Jason Xing <kernelxing@tencent.com>

Introducing a new type TCP_STATE to handle some reset conditions
appearing in RFC 793 due to its socket state. Actually, we can look
into RFC 9293 which has no discrepancy about this part.

Signed-off-by: Jason Xing <kernelxing@tencent.com>
---
v3
Link: https://lore.kernel.org/all/20240731120955.23542-5-kerneljasonxing@gmail.com/
1. remove one case from tcp_disconnect, which will be separately
categorized as another reason in the later patch (Eric)

V2
Link: https://lore.kernel.org/all/20240730200633.93761-1-kuniyu@amazon.com/
1. use RFC 9293 instead of RFC 793 which is too old (Kuniyuki)
---
 include/net/rstreason.h |  6 ++++++
 net/ipv4/tcp.c          | 10 ++++++----
 net/ipv4/tcp_timer.c    |  2 +-
 3 files changed, 13 insertions(+), 5 deletions(-)

Comments

Eric Dumazet Aug. 2, 2024, 9:41 a.m. UTC | #1 
On Thu, Aug 1, 2024 at 4:55 PM Jason Xing <kerneljasonxing@gmail.com> wrote:
>
> From: Jason Xing <kernelxing@tencent.com>
>
> Introducing a new type TCP_STATE to handle some reset conditions
> appearing in RFC 793 due to its socket state. Actually, we can look
> into RFC 9293 which has no discrepancy about this part.
>
> Signed-off-by: Jason Xing <kernelxing@tencent.com>

I really think this SK_RST_REASON_TCP_STATE is weak.

'Please see RFC 9293' does not help, this RFC has more than 5000 lines in it :/

Reviewed-by: Eric Dumazet <edumazet@google.com>
Jason Xing Aug. 2, 2024, 9:49 a.m. UTC | #2 
Hello Eric,

On Fri, Aug 2, 2024 at 5:41 PM Eric Dumazet <edumazet@google.com> wrote:
>
> On Thu, Aug 1, 2024 at 4:55 PM Jason Xing <kerneljasonxing@gmail.com> wrote:
> >
> > From: Jason Xing <kernelxing@tencent.com>
> >
> > Introducing a new type TCP_STATE to handle some reset conditions
> > appearing in RFC 793 due to its socket state. Actually, we can look
> > into RFC 9293 which has no discrepancy about this part.
> >
> > Signed-off-by: Jason Xing <kernelxing@tencent.com>
>
> I really think this SK_RST_REASON_TCP_STATE is weak.
>
> 'Please see RFC 9293' does not help, this RFC has more than 5000 lines in it :/

Yes, there are various reasons and conditions written in RFC 9293
nearly all over the place. I'm unable to conclude and get an union
name. Sorry about that. If I figure out a better name in the future,
I'll let you know.

>
> Reviewed-by: Eric Dumazet <edumazet@google.com>

Thanks for all your help.

Thanks,
Jason
diff mbox series

Patch

diff --git a/include/net/rstreason.h b/include/net/rstreason.h
index eef658da8952..bbf20d0bbde7 100644
--- a/include/net/rstreason.h
+++ b/include/net/rstreason.h
@@ -20,6 +20,7 @@ 
 	FN(TCP_ABORT_ON_CLOSE)		\
 	FN(TCP_ABORT_ON_LINGER)		\
 	FN(TCP_ABORT_ON_MEMORY)		\
+	FN(TCP_STATE)			\
 	FN(MPTCP_RST_EUNSPEC)		\
 	FN(MPTCP_RST_EMPTCP)		\
 	FN(MPTCP_RST_ERESOURCE)		\
@@ -102,6 +103,11 @@  enum sk_rst_reason {
 	 * 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,
 
 	/* Copy from include/uapi/linux/mptcp.h.
 	 * These reset fields will not be changed since they adhere to
diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c
index fd928c447ce8..24777e48bcc8 100644
--- a/net/ipv4/tcp.c
+++ b/net/ipv4/tcp.c
@@ -3025,9 +3025,11 @@  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
 		 */
@@ -4649,7 +4651,7 @@  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);
 	}
 
diff --git a/net/ipv4/tcp_timer.c b/net/ipv4/tcp_timer.c
index 0fba4a4fb988..3910f6d8614e 100644
--- a/net/ipv4/tcp_timer.c
+++ b/net/ipv4/tcp_timer.c
@@ -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;
 	}