Re: [PATCH 5.15.y v1 1/1] mptcp: fix UAF of child socket on listener close

From: Matthieu Baerts

Date: Thu Nov 19 2026 - 03:39:12 EST


Hi Jiayuan,

On 18/11/2026 13:50, Jiayuan Chen wrote:
> mptcp_close() orphans the listener subflow before inet_csk_listen_stop()
> runs on it. A child created in that window inherits SOCK_DEAD, so a
> following RST frees it while it is still in the accept queue:
>
> close(listen_fd): 3-way handshake & rst (plain TCP client)
>
> mptcp_close
> // mark ssk as SOCK_DEAD
> sock_orphan(ssk);
> __mptcp_destroy_sock
> // ssk in conn_list is listen sk
> __mptcp_close_ssk
> // in 3-way handshake
> tcp_v4_syn_recv_sock
> // inherit SOCK_DEAD state from listen ssk
> inet_csk_clone_lock
> // process RST
> tcp_reset
> tcp_done
> // sk is SOCK_DEAD
> inet_csk_destroy_sock
> sock_put(sk); // freed
> __tcp_close(ssk, 0)
> // free child
> inet_csk_listen_stop()
> process child <==== process freed sk
>
> refcount_t: addition on 0; use-after-free.
> WARNING: CPU: 59 PID: 404175 at lib/refcount.c:25 refcount_warn_saturate+0x12e/0x150
> RIP: 0010:refcount_warn_saturate+0x12e/0x150
> Call Trace:
> <TASK>
> inet_csk_listen_stop+0x3e1/0x530
> __tcp_close+0x3bc/0x440
> __mptcp_close_ssk+0xb9/0x210
> __mptcp_destroy_sock+0x11c/0x280
> mptcp_close+0x25d/0x3b0
> inet_release+0x44/0x90
> __sock_release+0x42/0xb0
> __fput+0x98/0x270
> ____fput+0xe/0x20
> task_work_run+0x6d/0xb0
> do_exit+0x1e7/0x3c0
> do_group_exit+0x3b/0xb0
> get_signal+0x17a/0x940
> arch_do_signal_or_restart+0xba/0x130
> exit_to_user_mode_loop+0xae/0x160
> exit_to_user_mode_prepare+0xad/0xc0
> irqentry_exit_to_user_mode+0x9/0x20
> irqentry_exit+0x21/0x40
> exc_page_fault+0x93/0x190
> asm_exc_page_fault+0x27/0x30
>
> Unhash the listener subflow and drain its accept queue before it is
> orphaned. No new child can be queued after the unhash, and the ones
> already queued are not SOCK_DEAD.
>
> Upstream fixed this in commit 57fc0f1ceaa4 ("mptcp: ensure listener is
> unhashed before updating the sk status"), which does not fit 5.15.

This patch looks good to me for v5.15:

Reviewed-by: Matthieu Baerts (NGI0) <matttbe@xxxxxxxxxx>

> This may become much easier to hit once services move to a newer Go.

I hope they will switch to a newer kernel first, v5.15 is supposed to be
EOL next month :)

Cheers,
Matt
--
Sponsored by the NGI0 Core fund.