1. Patchew
  2. MPTCP
  3. View series

[PATCH 6.1.y] mptcp: Fix proto fallback detection with BPF

Matthieu Baerts (NGI0) posted 1 patch 1 week, 5 days ago
Failed in applying to current master (apply log)
There is a newer version of this series
net/mptcp/protocol.c | 9 +++++++--
1 file changed, 7 insertions(+), 2 deletions(-)
[PATCH 6.1.y] mptcp: Fix proto fallback detection with BPF
Posted by Matthieu Baerts (NGI0) 1 week, 5 days ago 
From: Jiayuan Chen <jiayuan.chen@linux.dev>

commit c77b3b79a92e3345aa1ee296180d1af4e7031f8f upstream.

The sockmap feature allows bpf syscall from userspace, or based
on bpf sockops, replacing the sk_prot of sockets during protocol stack
processing with sockmap's custom read/write interfaces.
'''
tcp_rcv_state_process()
  syn_recv_sock()/subflow_syn_recv_sock()
    tcp_init_transfer(BPF_SOCK_OPS_PASSIVE_ESTABLISHED_CB)
      bpf_skops_established       <== sockops
        bpf_sock_map_update(sk)   <== call bpf helper
          tcp_bpf_update_proto()  <== update sk_prot
'''

When the server has MPTCP enabled but the client sends a TCP SYN
without MPTCP, subflow_syn_recv_sock() performs a fallback on the
subflow, replacing the subflow sk's sk_prot with the native sk_prot.
'''
subflow_syn_recv_sock()
  subflow_ulp_fallback()
    subflow_drop_ctx()
      mptcp_subflow_ops_undo_override()
'''

Then, this subflow can be normally used by sockmap, which replaces the
native sk_prot with sockmap's custom sk_prot. The issue occurs when the
user executes accept::mptcp_stream_accept::mptcp_fallback_tcp_ops().
Here, it uses sk->sk_prot to compare with the native sk_prot, but this
is incorrect when sockmap is used, as we may incorrectly set
sk->sk_socket->ops.

This fix uses the more generic sk_family for the comparison instead.

Additionally, this also prevents a WARNING from occurring:

result from ./scripts/decode_stacktrace.sh:
------------[ cut here ]------------
WARNING: CPU: 0 PID: 337 at net/mptcp/protocol.c:68 mptcp_stream_accept \
(net/mptcp/protocol.c:4005)
Modules linked in:
...

PKRU: 55555554
Call Trace:
<TASK>
do_accept (net/socket.c:1989)
__sys_accept4 (net/socket.c:2028 net/socket.c:2057)
__x64_sys_accept (net/socket.c:2067)
x64_sys_call (arch/x86/entry/syscall_64.c:41)
do_syscall_64 (arch/x86/entry/syscall_64.c:63 arch/x86/entry/syscall_64.c:94)
entry_SYSCALL_64_after_hwframe (arch/x86/entry/entry_64.S:130)
RIP: 0033:0x7f87ac92b83d

---[ end trace 0000000000000000 ]---

Fixes: 0b4f33def7bb ("mptcp: fix tcp fallback crash")
Signed-off-by: Jiayuan Chen <jiayuan.chen@linux.dev>
Signed-off-by: Martin KaFai Lau <martin.lau@kernel.org>
Reviewed-by: Jakub Sitnicki <jakub@cloudflare.com>
Reviewed-by: Matthieu Baerts (NGI0) <matttbe@kernel.org>
Cc: <stable@vger.kernel.org>
Link: https://patch.msgid.link/20251111060307.194196-3-jiayuan.chen@linux.dev
[ Conflicts in protocol.c, because commit 8e2b8a9fa512 ("mptcp: don't
  overwrite sock_ops in mptcp_is_tcpsk()") is not in this version. It
  changes the logic on how and where the sock_ops is overridden in case
  of passive fallback. To fix this, mptcp_is_tcpsk() is modified to use
  the family, but first, a check of the protocol is required to continue
  returning 'false' in case of MPTCP socket. ]
Signed-off-by: Matthieu Baerts (NGI0) <matttbe@kernel.org>
---
 net/mptcp/protocol.c | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)

diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c
index e2908add97d3..10844f08752c 100644
--- a/net/mptcp/protocol.c
+++ b/net/mptcp/protocol.c
@@ -79,8 +79,13 @@ static u64 mptcp_wnd_end(const struct mptcp_sock *msk)
 static bool mptcp_is_tcpsk(struct sock *sk)
 {
 	struct socket *sock = sk->sk_socket;
+	unsigned short family;
 
-	if (unlikely(sk->sk_prot == &tcp_prot)) {
+	if (likely(sk->sk_protocol == IPPROTO_MPTCP))
+		return false;
+
+	family = READ_ONCE(sk->sk_family);
+	if (unlikely(family == AF_INET)) {
 		/* we are being invoked after mptcp_accept() has
 		 * accepted a non-mp-capable flow: sk is a tcp_sk,
 		 * not an mptcp one.
@@ -91,7 +96,7 @@ static bool mptcp_is_tcpsk(struct sock *sk)
 		sock->ops = &inet_stream_ops;
 		return true;
 #if IS_ENABLED(CONFIG_MPTCP_IPV6)
-	} else if (unlikely(sk->sk_prot == &tcpv6_prot)) {
+	} else if (unlikely(family == AF_INET6)) {
 		sock->ops = &inet6_stream_ops;
 		return true;
 #endif
-- 
2.51.0
Re: [PATCH 6.1.y] mptcp: Fix proto fallback detection with BPF
Posted by Jiayuan Chen 1 week, 5 days ago 
December 1, 2025 at 18:45, "Matthieu Baerts (NGI0)" <matttbe@kernel.org mailto:matttbe@kernel.org?to=%22Matthieu%20Baerts%20(NGI0)%22%20%3Cmatttbe%40kernel.org%3E > wrote:


> 
> From: Jiayuan Chen <jiayuan.chen@linux.dev>
> 
> commit c77b3b79a92e3345aa1ee296180d1af4e7031f8f upstream.
> 
> The sockmap feature allows bpf syscall from userspace, or based
> on bpf sockops, replacing the sk_prot of sockets during protocol stack
> processing with sockmap's custom read/write interfaces.
> '''
> tcp_rcv_state_process()
>  syn_recv_sock()/subflow_syn_recv_sock()
>  tcp_init_transfer(BPF_SOCK_OPS_PASSIVE_ESTABLISHED_CB)
>  bpf_skops_established <== sockops
>  bpf_sock_map_update(sk) <== call bpf helper
>  tcp_bpf_update_proto() <== update sk_prot
> '''
> 
> When the server has MPTCP enabled but the client sends a TCP SYN
> without MPTCP, subflow_syn_recv_sock() performs a fallback on the
> subflow, replacing the subflow sk's sk_prot with the native sk_prot.
> '''
> subflow_syn_recv_sock()
>  subflow_ulp_fallback()
>  subflow_drop_ctx()
>  mptcp_subflow_ops_undo_override()
> '''
> 
> Then, this subflow can be normally used by sockmap, which replaces the
> native sk_prot with sockmap's custom sk_prot. The issue occurs when the
> user executes accept::mptcp_stream_accept::mptcp_fallback_tcp_ops().
> Here, it uses sk->sk_prot to compare with the native sk_prot, but this
> is incorrect when sockmap is used, as we may incorrectly set
> sk->sk_socket->ops.
> 
> This fix uses the more generic sk_family for the comparison instead.
> 
> Additionally, this also prevents a WARNING from occurring:
> 
> result from ./scripts/decode_stacktrace.sh:
> ------------[ cut here ]------------
> WARNING: CPU: 0 PID: 337 at net/mptcp/protocol.c:68 mptcp_stream_accept \
> (net/mptcp/protocol.c:4005)
> Modules linked in:
> ...
> 
> PKRU: 55555554
> Call Trace:
> <TASK>
> do_accept (net/socket.c:1989)
> __sys_accept4 (net/socket.c:2028 net/socket.c:2057)
> __x64_sys_accept (net/socket.c:2067)
> x64_sys_call (arch/x86/entry/syscall_64.c:41)
> do_syscall_64 (arch/x86/entry/syscall_64.c:63 arch/x86/entry/syscall_64.c:94)
> entry_SYSCALL_64_after_hwframe (arch/x86/entry/entry_64.S:130)
> RIP: 0033:0x7f87ac92b83d
> 
> ---[ end trace 0000000000000000 ]---
> 
> Fixes: 0b4f33def7bb ("mptcp: fix tcp fallback crash")
> Signed-off-by: Jiayuan Chen <jiayuan.chen@linux.dev>
> Signed-off-by: Martin KaFai Lau <martin.lau@kernel.org>
> Reviewed-by: Jakub Sitnicki <jakub@cloudflare.com>
> Reviewed-by: Matthieu Baerts (NGI0) <matttbe@kernel.org>
> Cc: <stable@vger.kernel.org>
> Link: https://patch.msgid.link/20251111060307.194196-3-jiayuan.chen@linux.dev
> [ Conflicts in protocol.c, because commit 8e2b8a9fa512 ("mptcp: don't
>  overwrite sock_ops in mptcp_is_tcpsk()") is not in this version. It
>  changes the logic on how and where the sock_ops is overridden in case
>  of passive fallback. To fix this, mptcp_is_tcpsk() is modified to use
>  the family, but first, a check of the protocol is required to continue
>  returning 'false' in case of MPTCP socket. ]
> Signed-off-by: Matthieu Baerts (NGI0) <matttbe@kernel.org>
> ---
>  net/mptcp/protocol.c | 9 +++++++--
>  1 file changed, 7 insertions(+), 2 deletions(-)
> 
> diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c
> index e2908add97d3..10844f08752c 100644
> --- a/net/mptcp/protocol.c
> +++ b/net/mptcp/protocol.c
> @@ -79,8 +79,13 @@ static u64 mptcp_wnd_end(const struct mptcp_sock *msk)
>  static bool mptcp_is_tcpsk(struct sock *sk)
>  {
>  struct socket *sock = sk->sk_socket;
> + unsigned short family;
>  
> - if (unlikely(sk->sk_prot == &tcp_prot)) {
> + if (likely(sk->sk_protocol == IPPROTO_MPTCP))
> + return false;
> +
> + family = READ_ONCE(sk->sk_family);
> + if (unlikely(family == AF_INET)) {
>  /* we are being invoked after mptcp_accept() has
>  * accepted a non-mp-capable flow: sk is a tcp_sk,
>  * not an mptcp one.
> @@ -91,7 +96,7 @@ static bool mptcp_is_tcpsk(struct sock *sk)
>  sock->ops = &inet_stream_ops;
>  return true;
>  #if IS_ENABLED(CONFIG_MPTCP_IPV6)
> - } else if (unlikely(sk->sk_prot == &tcpv6_prot)) {
> + } else if (unlikely(family == AF_INET6)) {
>  sock->ops = &inet6_stream_ops;
>  return true;
>  #endif
> -- 
> 2.51.0
>

Thank you, Matthieu. I’ve tested the patch and confirmed it resolves the issue.
Re: [PATCH 6.1.y] mptcp: Fix proto fallback detection with BPF
Posted by Matthieu Baerts 1 week, 5 days ago 
Hi Jiayuan,

On 01/12/2025 12:15, Jiayuan Chen wrote:
> December 1, 2025 at 18:45, "Matthieu Baerts (NGI0)" <matttbe@kernel.org mailto:matttbe@kernel.org?to=%22Matthieu%20Baerts%20(NGI0)%22%20%3Cmatttbe%40kernel.org%3E > wrote:
> 
> 
>>
>> From: Jiayuan Chen <jiayuan.chen@linux.dev>
>>
>> commit c77b3b79a92e3345aa1ee296180d1af4e7031f8f upstream.

(...)
> Thank you, Matthieu. I’ve tested the patch and confirmed it resolves the issue.

Thank you for having checked! Then I will also send the same patch for
v5.15 and v5.10.

Note: regarding your other patch fbade4bd08ba ("mptcp: Disallow MPTCP
subflows from sockmap"): it cannot be applied on v5.10. I have a draft,
but the modifications have to be done in sockmap code, not in MPTCP
side, and I don't want to introduce regressions for "normal" TCP sockmap
use. By chance, are you also able to test that one on a v5.10 kernel?

Cheers,
Matt
-- 
Sponsored by the NGI0 Core fund.
Patch "mptcp: Fix proto fallback detection with BPF" has been added to the 6.1-stable tree
Posted by gregkh@linuxfoundation.org 1 week, 4 days ago 

This is a note to let you know that I've just added the patch titled

    mptcp: Fix proto fallback detection with BPF

to the 6.1-stable tree which can be found at:
    http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=summary

The filename of the patch is:
     mptcp-fix-proto-fallback-detection-with-bpf.patch
and it can be found in the queue-6.1 subdirectory.

If you, or anyone else, feels it should not be added to the stable tree,
please let <stable@vger.kernel.org> know about it.


From stable+bounces-197700-greg=kroah.com@vger.kernel.org Mon Dec  1 11:45:35 2025
From: "Matthieu Baerts (NGI0)" <matttbe@kernel.org>
Date: Mon,  1 Dec 2025 11:45:00 +0100
Subject: mptcp: Fix proto fallback detection with BPF
To: stable@vger.kernel.org, gregkh@linuxfoundation.org
Cc: MPTCP Upstream <mptcp@lists.linux.dev>, Jiayuan Chen <jiayuan.chen@linux.dev>, Martin KaFai Lau <martin.lau@kernel.org>, Jakub Sitnicki <jakub@cloudflare.com>, "Matthieu Baerts (NGI0)" <matttbe@kernel.org>
Message-ID: <20251201104459.3440448-2-matttbe@kernel.org>

From: Jiayuan Chen <jiayuan.chen@linux.dev>

commit c77b3b79a92e3345aa1ee296180d1af4e7031f8f upstream.

The sockmap feature allows bpf syscall from userspace, or based
on bpf sockops, replacing the sk_prot of sockets during protocol stack
processing with sockmap's custom read/write interfaces.
'''
tcp_rcv_state_process()
  syn_recv_sock()/subflow_syn_recv_sock()
    tcp_init_transfer(BPF_SOCK_OPS_PASSIVE_ESTABLISHED_CB)
      bpf_skops_established       <== sockops
        bpf_sock_map_update(sk)   <== call bpf helper
          tcp_bpf_update_proto()  <== update sk_prot
'''

When the server has MPTCP enabled but the client sends a TCP SYN
without MPTCP, subflow_syn_recv_sock() performs a fallback on the
subflow, replacing the subflow sk's sk_prot with the native sk_prot.
'''
subflow_syn_recv_sock()
  subflow_ulp_fallback()
    subflow_drop_ctx()
      mptcp_subflow_ops_undo_override()
'''

Then, this subflow can be normally used by sockmap, which replaces the
native sk_prot with sockmap's custom sk_prot. The issue occurs when the
user executes accept::mptcp_stream_accept::mptcp_fallback_tcp_ops().
Here, it uses sk->sk_prot to compare with the native sk_prot, but this
is incorrect when sockmap is used, as we may incorrectly set
sk->sk_socket->ops.

This fix uses the more generic sk_family for the comparison instead.

Additionally, this also prevents a WARNING from occurring:

result from ./scripts/decode_stacktrace.sh:
------------[ cut here ]------------
WARNING: CPU: 0 PID: 337 at net/mptcp/protocol.c:68 mptcp_stream_accept \
(net/mptcp/protocol.c:4005)
Modules linked in:
...

PKRU: 55555554
Call Trace:
<TASK>
do_accept (net/socket.c:1989)
__sys_accept4 (net/socket.c:2028 net/socket.c:2057)
__x64_sys_accept (net/socket.c:2067)
x64_sys_call (arch/x86/entry/syscall_64.c:41)
do_syscall_64 (arch/x86/entry/syscall_64.c:63 arch/x86/entry/syscall_64.c:94)
entry_SYSCALL_64_after_hwframe (arch/x86/entry/entry_64.S:130)
RIP: 0033:0x7f87ac92b83d

---[ end trace 0000000000000000 ]---

Fixes: 0b4f33def7bb ("mptcp: fix tcp fallback crash")
Signed-off-by: Jiayuan Chen <jiayuan.chen@linux.dev>
Signed-off-by: Martin KaFai Lau <martin.lau@kernel.org>
Reviewed-by: Jakub Sitnicki <jakub@cloudflare.com>
Reviewed-by: Matthieu Baerts (NGI0) <matttbe@kernel.org>
Cc: <stable@vger.kernel.org>
Link: https://patch.msgid.link/20251111060307.194196-3-jiayuan.chen@linux.dev
[ Conflicts in protocol.c, because commit 8e2b8a9fa512 ("mptcp: don't
  overwrite sock_ops in mptcp_is_tcpsk()") is not in this version. It
  changes the logic on how and where the sock_ops is overridden in case
  of passive fallback. To fix this, mptcp_is_tcpsk() is modified to use
  the family, but first, a check of the protocol is required to continue
  returning 'false' in case of MPTCP socket. ]
Signed-off-by: Matthieu Baerts (NGI0) <matttbe@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
 net/mptcp/protocol.c |    9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)

--- a/net/mptcp/protocol.c
+++ b/net/mptcp/protocol.c
@@ -79,8 +79,13 @@ static u64 mptcp_wnd_end(const struct mp
 static bool mptcp_is_tcpsk(struct sock *sk)
 {
 	struct socket *sock = sk->sk_socket;
+	unsigned short family;
 
-	if (unlikely(sk->sk_prot == &tcp_prot)) {
+	if (likely(sk->sk_protocol == IPPROTO_MPTCP))
+		return false;
+
+	family = READ_ONCE(sk->sk_family);
+	if (unlikely(family == AF_INET)) {
 		/* we are being invoked after mptcp_accept() has
 		 * accepted a non-mp-capable flow: sk is a tcp_sk,
 		 * not an mptcp one.
@@ -91,7 +96,7 @@ static bool mptcp_is_tcpsk(struct sock *
 		sock->ops = &inet_stream_ops;
 		return true;
 #if IS_ENABLED(CONFIG_MPTCP_IPV6)
-	} else if (unlikely(sk->sk_prot == &tcpv6_prot)) {
+	} else if (unlikely(family == AF_INET6)) {
 		sock->ops = &inet6_stream_ops;
 		return true;
 #endif


Patches currently in stable-queue which might be from matttbe@kernel.org are

queue-6.1/selftests-mptcp-join-endpoints-longer-transfer.patch
queue-6.1/mptcp-disallow-mptcp-subflows-from-sockmap.patch
queue-6.1/selftests-mptcp-connect-trunc-read-all-recv-data.patch
queue-6.1/mptcp-fix-race-condition-in-mptcp_schedule_work.patch
queue-6.1/mptcp-restore-window-probe.patch
queue-6.1/mptcp-fix-proto-fallback-detection-with-bpf.patch
queue-6.1/selftests-mptcp-join-mark-delete-re-add-signal-as-skipped-if-not-supported.patch
queue-6.1/mptcp-do-not-fallback-when-ooo-is-present.patch
queue-6.1/mptcp-decouple-mptcp-fastclose-from-tcp-close.patch
queue-6.1/mptcp-fix-duplicate-reset-on-fastclose.patch
queue-6.1/mptcp-drop-bogus-optimization-in-__mptcp_check_push.patch
queue-6.1/selftests-mptcp-join-rm-set-backup-flag.patch
queue-6.1/selftests-mptcp-connect-fix-fallback-note-due-to-ooo.patch
queue-6.1/selftests-mptcp-disable-add_addr-retrans-in-endpoint_tests.patch
queue-6.1/mptcp-fix-ack-generation-for-fallback-msk.patch
queue-6.1/mptcp-pm-in-kernel-c-flag-handle-late-add_addr.patch
queue-6.1/mptcp-fix-premature-close-in-case-of-fallback.patch
queue-6.1/mptcp-fix-a-race-in-mptcp_pm_del_add_timer.patch
queue-6.1/gcov-add-support-for-gcc-15.patch
queue-6.1/mptcp-avoid-unneeded-subflow-level-drops.patch

Patchew 2.1-devel-b67e4c2

© 2016 - 2025 Red Hat, Inc.