[PATCH net 0/5] mptcp: Fixes for v6.6

[PATCH net 0/5] mptcp: Fixes for v6.6

[Mat Martineau]

Patch 1 corrects the logic for MP_JOIN tests where 0 RSTs are expected.

Patch 2 ensures MPTCP packets are not incorrectly coalesced in the TCP
backlog queue.

Patch 3 avoids a zero-window probe and associated WARN_ON_ONCE() in an
expected MPTCP reinjection scenario.

Patches 4 & 5 allow an initial MPTCP subflow to be closed cleanly
instead of always sending RST. Associated selftest is updated.

Signed-off-by: Mat Martineau <martineau@kernel.org>
---
Geliang Tang (1):
      mptcp: avoid sending RST when closing the initial subflow

Matthieu Baerts (2):
      selftests: mptcp: join: correctly check for no RST
      selftests: mptcp: join: no RST when rm subflow/addr

Paolo Abeni (2):
      tcp: check mptcp-level constraints for backlog coalescing
      mptcp: more conservative check for zero probes

 net/ipv4/tcp_ipv4.c                             |  1 +
 net/mptcp/protocol.c                            | 36 ++++++++++++++++---------
 tools/testing/selftests/net/mptcp/mptcp_join.sh | 21 +++++++++++++--
 3 files changed, 43 insertions(+), 15 deletions(-)
---
base-commit: 2915240eddba96b37de4c7e9a3d0ac6f9548454b
change-id: 20231018-send-net-20231018-ac6b38df05e2

Best regards,
-- 
Mat Martineau <martineau@kernel.org>

[PATCH net 1/5] selftests: mptcp: join: correctly check for no RST

[Mat Martineau]

From: Matthieu Baerts <matttbe@kernel.org>

The commit mentioned below was more tolerant with the number of RST seen
during a test because in some uncontrollable situations, multiple RST
can be generated.

But it was not taking into account the case where no RST are expected:
this validation was then no longer reporting issues for the 0 RST case
because it is not possible to have less than 0 RST in the counter. This
patch fixes the issue by adding a specific condition.

Fixes: 6bf41020b72b ("selftests: mptcp: update and extend fastclose test-cases")
Cc: stable@vger.kernel.org
Reviewed-by: Mat Martineau <martineau@kernel.org>
Signed-off-by: Matthieu Baerts <matttbe@kernel.org>
Signed-off-by: Mat Martineau <martineau@kernel.org>
---
 tools/testing/selftests/net/mptcp/mptcp_join.sh | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/tools/testing/selftests/net/mptcp/mptcp_join.sh b/tools/testing/selftests/net/mptcp/mptcp_join.sh
index ee1f89a872b3..27953670206e 100755
--- a/tools/testing/selftests/net/mptcp/mptcp_join.sh
+++ b/tools/testing/selftests/net/mptcp/mptcp_join.sh
@@ -1432,7 +1432,9 @@ chk_rst_nr()
 	count=$(get_counter ${ns_tx} "MPTcpExtMPRstTx")
 	if [ -z "$count" ]; then
 		print_skip
-	elif [ $count -lt $rst_tx ]; then
+	# accept more rst than expected except if we don't expect any
+	elif { [ $rst_tx -ne 0 ] && [ $count -lt $rst_tx ]; } ||
+	     { [ $rst_tx -eq 0 ] && [ $count -ne 0 ]; }; then
 		fail_test "got $count MP_RST[s] TX expected $rst_tx"
 	else
 		print_ok
@@ -1442,7 +1444,9 @@ chk_rst_nr()
 	count=$(get_counter ${ns_rx} "MPTcpExtMPRstRx")
 	if [ -z "$count" ]; then
 		print_skip
-	elif [ "$count" -lt "$rst_rx" ]; then
+	# accept more rst than expected except if we don't expect any
+	elif { [ $rst_rx -ne 0 ] && [ $count -lt $rst_rx ]; } ||
+	     { [ $rst_rx -eq 0 ] && [ $count -ne 0 ]; }; then
 		fail_test "got $count MP_RST[s] RX expected $rst_rx"
 	else
 		print_ok

-- 
2.41.0

[PATCH net 2/5] tcp: check mptcp-level constraints for backlog coalescing

[Mat Martineau]

From: Paolo Abeni <pabeni@redhat.com>

The MPTCP protocol can acquire the subflow-level socket lock and
cause the tcp backlog usage. When inserting new skbs into the
backlog, the stack will try to coalesce them.

Currently, we have no check in place to ensure that such coalescing
will respect the MPTCP-level DSS, and that may cause data stream
corruption, as reported by Christoph.

Address the issue by adding the relevant admission check for coalescing
in tcp_add_backlog().

Note the issue is not easy to reproduce, as the MPTCP protocol tries
hard to avoid acquiring the subflow-level socket lock.

Fixes: 648ef4b88673 ("mptcp: Implement MPTCP receive path")
Cc: stable@vger.kernel.org
Reported-by: Christoph Paasch <cpaasch@apple.com>
Closes: https://github.com/multipath-tcp/mptcp_net-next/issues/420
Reviewed-by: Mat Martineau <martineau@kernel.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Mat Martineau <martineau@kernel.org>
---
 net/ipv4/tcp_ipv4.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c
index 27140e5cdc06..4167e8a48b60 100644
--- a/net/ipv4/tcp_ipv4.c
+++ b/net/ipv4/tcp_ipv4.c
@@ -1869,6 +1869,7 @@ bool tcp_add_backlog(struct sock *sk, struct sk_buff *skb,
 #ifdef CONFIG_TLS_DEVICE
 	    tail->decrypted != skb->decrypted ||
 #endif
+	    !mptcp_skb_can_collapse(tail, skb) ||
 	    thtail->doff != th->doff ||
 	    memcmp(thtail + 1, th + 1, hdrlen - sizeof(*th)))
 		goto no_coalesce;

-- 
2.41.0

[PATCH net 3/5] mptcp: more conservative check for zero probes

[Mat Martineau]

From: Paolo Abeni <pabeni@redhat.com>

Christoph reported that the MPTCP protocol can find the subflow-level
write queue unexpectedly not empty while crafting a zero-window probe,
hitting a warning:

------------[ cut here ]------------
WARNING: CPU: 0 PID: 188 at net/mptcp/protocol.c:1312 mptcp_sendmsg_frag+0xc06/0xe70
Modules linked in:
CPU: 0 PID: 188 Comm: kworker/0:2 Not tainted 6.6.0-rc2-g1176aa719d7a #47
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.11.0-2.el7 04/01/2014
Workqueue: events mptcp_worker
RIP: 0010:mptcp_sendmsg_frag+0xc06/0xe70 net/mptcp/protocol.c:1312
RAX: 47d0530de347ff6a RBX: 47d0530de347ff6b RCX: ffff8881015d3c00
RDX: ffff8881015d3c00 RSI: 47d0530de347ff6b RDI: 47d0530de347ff6b
RBP: 47d0530de347ff6b R08: ffffffff8243c6a8 R09: ffffffff82042d9c
R10: 0000000000000002 R11: ffffffff82056850 R12: ffff88812a13d580
R13: 0000000000000001 R14: ffff88812b375e50 R15: ffff88812bbf3200
FS:  0000000000000000(0000) GS:ffff88813bc00000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 0000000000695118 CR3: 0000000115dfc001 CR4: 0000000000170ef0
Call Trace:
 <TASK>
 __subflow_push_pending+0xa4/0x420 net/mptcp/protocol.c:1545
 __mptcp_push_pending+0x128/0x3b0 net/mptcp/protocol.c:1614
 mptcp_release_cb+0x218/0x5b0 net/mptcp/protocol.c:3391
 release_sock+0xf6/0x100 net/core/sock.c:3521
 mptcp_worker+0x6e8/0x8f0 net/mptcp/protocol.c:2746
 process_scheduled_works+0x341/0x690 kernel/workqueue.c:2630
 worker_thread+0x3a7/0x610 kernel/workqueue.c:2784
 kthread+0x143/0x180 kernel/kthread.c:388
 ret_from_fork+0x4d/0x60 arch/x86/kernel/process.c:147
 ret_from_fork_asm+0x1b/0x30 arch/x86/entry/entry_64.S:304
 </TASK>

The root cause of the issue is that expectations are wrong: e.g. due
to MPTCP-level re-injection we can hit the critical condition.

Explicitly avoid the zero-window probe when the subflow write queue
is not empty and drop the related warnings.

Reported-by: Christoph Paasch <cpaasch@apple.com>
Closes: https://github.com/multipath-tcp/mptcp_net-next/issues/444
Fixes: f70cad1085d1 ("mptcp: stop relying on tcp_tx_skb_cache")
Cc: stable@vger.kernel.org
Reviewed-by: Mat Martineau <martineau@kernel.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Mat Martineau <martineau@kernel.org>
---
 net/mptcp/protocol.c | 8 +-------
 1 file changed, 1 insertion(+), 7 deletions(-)

diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c
index d1902373c974..4e30e5ba3795 100644
--- a/net/mptcp/protocol.c
+++ b/net/mptcp/protocol.c
@@ -1298,7 +1298,7 @@ static int mptcp_sendmsg_frag(struct sock *sk, struct sock *ssk,
 	if (copy == 0) {
 		u64 snd_una = READ_ONCE(msk->snd_una);
 
-		if (snd_una != msk->snd_nxt) {
+		if (snd_una != msk->snd_nxt || tcp_write_queue_tail(ssk)) {
 			tcp_remove_empty_skb(ssk);
 			return 0;
 		}
@@ -1306,11 +1306,6 @@ static int mptcp_sendmsg_frag(struct sock *sk, struct sock *ssk,
 		zero_window_probe = true;
 		data_seq = snd_una - 1;
 		copy = 1;
-
-		/* all mptcp-level data is acked, no skbs should be present into the
-		 * ssk write queue
-		 */
-		WARN_ON_ONCE(reuse_skb);
 	}
 
 	copy = min_t(size_t, copy, info->limit - info->sent);
@@ -1339,7 +1334,6 @@ static int mptcp_sendmsg_frag(struct sock *sk, struct sock *ssk,
 	if (reuse_skb) {
 		TCP_SKB_CB(skb)->tcp_flags &= ~TCPHDR_PSH;
 		mpext->data_len += copy;
-		WARN_ON_ONCE(zero_window_probe);
 		goto out;
 	}
 

-- 
2.41.0

[PATCH net 4/5] mptcp: avoid sending RST when closing the initial subflow

[Mat Martineau]

From: Geliang Tang <geliang.tang@suse.com>

When closing the first subflow, the MPTCP protocol unconditionally
calls tcp_disconnect(), which in turn generates a reset if the subflow
is established.

That is unexpected and different from what MPTCP does with MPJ
subflows, where resets are generated only on FASTCLOSE and other edge
scenarios.

We can't reuse for the first subflow the same code in place for MPJ
subflows, as MPTCP clean them up completely via a tcp_close() call,
while must keep the first subflow socket alive for later re-usage, due
to implementation constraints.

This patch adds a new helper __mptcp_subflow_disconnect() that
encapsulates, a logic similar to tcp_close, issuing a reset only when
the MPTCP_CF_FASTCLOSE flag is set, and performing a clean shutdown
otherwise.

Fixes: c2b2ae3925b6 ("mptcp: handle correctly disconnect() failures")
Cc: stable@vger.kernel.org
Reviewed-by: Matthieu Baerts <matttbe@kernel.org>
Co-developed-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Geliang Tang <geliang.tang@suse.com>
Signed-off-by: Mat Martineau <martineau@kernel.org>
---
 net/mptcp/protocol.c | 28 ++++++++++++++++++++++------
 1 file changed, 22 insertions(+), 6 deletions(-)

diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c
index 4e30e5ba3795..886ab689a8ae 100644
--- a/net/mptcp/protocol.c
+++ b/net/mptcp/protocol.c
@@ -2348,6 +2348,26 @@ bool __mptcp_retransmit_pending_data(struct sock *sk)
 #define MPTCP_CF_PUSH		BIT(1)
 #define MPTCP_CF_FASTCLOSE	BIT(2)
 
+/* be sure to send a reset only if the caller asked for it, also
+ * clean completely the subflow status when the subflow reaches
+ * TCP_CLOSE state
+ */
+static void __mptcp_subflow_disconnect(struct sock *ssk,
+				       struct mptcp_subflow_context *subflow,
+				       unsigned int flags)
+{
+	if (((1 << ssk->sk_state) & (TCPF_CLOSE | TCPF_LISTEN)) ||
+	    (flags & MPTCP_CF_FASTCLOSE)) {
+		/* The MPTCP code never wait on the subflow sockets, TCP-level
+		 * disconnect should never fail
+		 */
+		WARN_ON_ONCE(tcp_disconnect(ssk, 0));
+		mptcp_subflow_ctx_reset(subflow);
+	} else {
+		tcp_shutdown(ssk, SEND_SHUTDOWN);
+	}
+}
+
 /* subflow sockets can be either outgoing (connect) or incoming
  * (accept).
  *
@@ -2385,7 +2405,7 @@ static void __mptcp_close_ssk(struct sock *sk, struct sock *ssk,
 	lock_sock_nested(ssk, SINGLE_DEPTH_NESTING);
 
 	if ((flags & MPTCP_CF_FASTCLOSE) && !__mptcp_check_fallback(msk)) {
-		/* be sure to force the tcp_disconnect() path,
+		/* be sure to force the tcp_close path
 		 * to generate the egress reset
 		 */
 		ssk->sk_lingertime = 0;
@@ -2395,11 +2415,7 @@ static void __mptcp_close_ssk(struct sock *sk, struct sock *ssk,
 
 	need_push = (flags & MPTCP_CF_PUSH) && __mptcp_retransmit_pending_data(sk);
 	if (!dispose_it) {
-		/* The MPTCP code never wait on the subflow sockets, TCP-level
-		 * disconnect should never fail
-		 */
-		WARN_ON_ONCE(tcp_disconnect(ssk, 0));
-		mptcp_subflow_ctx_reset(subflow);
+		__mptcp_subflow_disconnect(ssk, subflow, flags);
 		release_sock(ssk);
 
 		goto out;

-- 
2.41.0

[PATCH net 5/5] selftests: mptcp: join: no RST when rm subflow/addr

[Mat Martineau]

From: Matthieu Baerts <matttbe@kernel.org>

Recently, we noticed that some RST were wrongly generated when removing
the initial subflow.

This patch makes sure RST are not sent when removing any subflows or any
addresses.

Fixes: c2b2ae3925b6 ("mptcp: handle correctly disconnect() failures")
Cc: stable@vger.kernel.org
Acked-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Matthieu Baerts <matttbe@kernel.org>
Signed-off-by: Mat Martineau <martineau@kernel.org>
---
 tools/testing/selftests/net/mptcp/mptcp_join.sh | 13 +++++++++++++
 1 file changed, 13 insertions(+)

diff --git a/tools/testing/selftests/net/mptcp/mptcp_join.sh b/tools/testing/selftests/net/mptcp/mptcp_join.sh
index 27953670206e..dc895b7b94e1 100755
--- a/tools/testing/selftests/net/mptcp/mptcp_join.sh
+++ b/tools/testing/selftests/net/mptcp/mptcp_join.sh
@@ -2309,6 +2309,7 @@ remove_tests()
 		chk_join_nr 1 1 1
 		chk_rm_tx_nr 1
 		chk_rm_nr 1 1
+		chk_rst_nr 0 0
 	fi
 
 	# multiple subflows, remove
@@ -2321,6 +2322,7 @@ remove_tests()
 			run_tests $ns1 $ns2 10.0.1.1
 		chk_join_nr 2 2 2
 		chk_rm_nr 2 2
+		chk_rst_nr 0 0
 	fi
 
 	# single address, remove
@@ -2333,6 +2335,7 @@ remove_tests()
 		chk_join_nr 1 1 1
 		chk_add_nr 1 1
 		chk_rm_nr 1 1 invert
+		chk_rst_nr 0 0
 	fi
 
 	# subflow and signal, remove
@@ -2346,6 +2349,7 @@ remove_tests()
 		chk_join_nr 2 2 2
 		chk_add_nr 1 1
 		chk_rm_nr 1 1
+		chk_rst_nr 0 0
 	fi
 
 	# subflows and signal, remove
@@ -2360,6 +2364,7 @@ remove_tests()
 		chk_join_nr 3 3 3
 		chk_add_nr 1 1
 		chk_rm_nr 2 2
+		chk_rst_nr 0 0
 	fi
 
 	# addresses remove
@@ -2374,6 +2379,7 @@ remove_tests()
 		chk_join_nr 3 3 3
 		chk_add_nr 3 3
 		chk_rm_nr 3 3 invert
+		chk_rst_nr 0 0
 	fi
 
 	# invalid addresses remove
@@ -2388,6 +2394,7 @@ remove_tests()
 		chk_join_nr 1 1 1
 		chk_add_nr 3 3
 		chk_rm_nr 3 1 invert
+		chk_rst_nr 0 0
 	fi
 
 	# subflows and signal, flush
@@ -2402,6 +2409,7 @@ remove_tests()
 		chk_join_nr 3 3 3
 		chk_add_nr 1 1
 		chk_rm_nr 1 3 invert simult
+		chk_rst_nr 0 0
 	fi
 
 	# subflows flush
@@ -2421,6 +2429,7 @@ remove_tests()
 		else
 			chk_rm_nr 3 3
 		fi
+		chk_rst_nr 0 0
 	fi
 
 	# addresses flush
@@ -2435,6 +2444,7 @@ remove_tests()
 		chk_join_nr 3 3 3
 		chk_add_nr 3 3
 		chk_rm_nr 3 3 invert simult
+		chk_rst_nr 0 0
 	fi
 
 	# invalid addresses flush
@@ -2449,6 +2459,7 @@ remove_tests()
 		chk_join_nr 1 1 1
 		chk_add_nr 3 3
 		chk_rm_nr 3 1 invert
+		chk_rst_nr 0 0
 	fi
 
 	# remove id 0 subflow
@@ -2460,6 +2471,7 @@ remove_tests()
 			run_tests $ns1 $ns2 10.0.1.1
 		chk_join_nr 1 1 1
 		chk_rm_nr 1 1
+		chk_rst_nr 0 0
 	fi
 
 	# remove id 0 address
@@ -2472,6 +2484,7 @@ remove_tests()
 		chk_join_nr 1 1 1
 		chk_add_nr 1 1
 		chk_rm_nr 1 1 invert
+		chk_rst_nr 0 0 invert
 	fi
 }
 

-- 
2.41.0

Re: [PATCH net 0/5] mptcp: Fixes for v6.6

[patchwork-bot+netdevbpf]

Hello:

This series was applied to netdev/net.git (main)
by Jakub Kicinski <kuba@kernel.org>:

On Wed, 18 Oct 2023 11:23:51 -0700 you wrote:
> Patch 1 corrects the logic for MP_JOIN tests where 0 RSTs are expected.
> 
> Patch 2 ensures MPTCP packets are not incorrectly coalesced in the TCP
> backlog queue.
> 
> Patch 3 avoids a zero-window probe and associated WARN_ON_ONCE() in an
> expected MPTCP reinjection scenario.
> 
> [...]

Here is the summary with links:
  - [net,1/5] selftests: mptcp: join: correctly check for no RST
    https://git.kernel.org/netdev/net/c/b134a5805455
  - [net,2/5] tcp: check mptcp-level constraints for backlog coalescing
    https://git.kernel.org/netdev/net/c/6db8a37dfc54
  - [net,3/5] mptcp: more conservative check for zero probes
    https://git.kernel.org/netdev/net/c/72377ab2d671
  - [net,4/5] mptcp: avoid sending RST when closing the initial subflow
    https://git.kernel.org/netdev/net/c/14c56686a64c
  - [net,5/5] selftests: mptcp: join: no RST when rm subflow/addr
    https://git.kernel.org/netdev/net/c/2cfaa8b3b7ae

You are awesome, thank you!
-- 
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html