[PATCH net] l2tp: avoid one data-race in l2tp_tunnel_del_work()

[PATCH net] l2tp: avoid one data-race in l2tp_tunnel_del_work()

[Eric Dumazet]

We should read sk->sk_socket only when dealing with kernel sockets.

syzbot reported the following data-race:

BUG: KCSAN: data-race in l2tp_tunnel_del_work / sk_common_release

write to 0xffff88811c182b20 of 8 bytes by task 5365 on cpu 0:
  sk_set_socket include/net/sock.h:2092 [inline]
  sock_orphan include/net/sock.h:2118 [inline]
  sk_common_release+0xae/0x230 net/core/sock.c:4003
  udp_lib_close+0x15/0x20 include/net/udp.h:325
  inet_release+0xce/0xf0 net/ipv4/af_inet.c:437
  __sock_release net/socket.c:662 [inline]
  sock_close+0x6b/0x150 net/socket.c:1455
  __fput+0x29b/0x650 fs/file_table.c:468
  ____fput+0x1c/0x30 fs/file_table.c:496
  task_work_run+0x131/0x1a0 kernel/task_work.c:233
  resume_user_mode_work include/linux/resume_user_mode.h:50 [inline]
  __exit_to_user_mode_loop kernel/entry/common.c:44 [inline]
  exit_to_user_mode_loop+0x1fe/0x740 kernel/entry/common.c:75
  __exit_to_user_mode_prepare include/linux/irq-entry-common.h:226 [inline]
  syscall_exit_to_user_mode_prepare include/linux/irq-entry-common.h:256 [inline]
  syscall_exit_to_user_mode_work include/linux/entry-common.h:159 [inline]
  syscall_exit_to_user_mode include/linux/entry-common.h:194 [inline]
  do_syscall_64+0x1e1/0x2b0 arch/x86/entry/syscall_64.c:100
 entry_SYSCALL_64_after_hwframe+0x77/0x7f

read to 0xffff88811c182b20 of 8 bytes by task 827 on cpu 1:
  l2tp_tunnel_del_work+0x2f/0x1a0 net/l2tp/l2tp_core.c:1418
  process_one_work kernel/workqueue.c:3257 [inline]
  process_scheduled_works+0x4ce/0x9d0 kernel/workqueue.c:3340
  worker_thread+0x582/0x770 kernel/workqueue.c:3421
  kthread+0x489/0x510 kernel/kthread.c:463
  ret_from_fork+0x149/0x290 arch/x86/kernel/process.c:158
  ret_from_fork_asm+0x1a/0x30 arch/x86/entry/entry_64.S:246

value changed: 0xffff88811b818000 -> 0x0000000000000000

Fixes: d00fa9adc528 ("l2tp: fix races with tunnel socket close")
Reported-by: syzbot+7312e82745f7fa2526db@syzkaller.appspotmail.com
Closes: https://lore.kernel.org/netdev/6968b029.050a0220.58bed.0016.GAE@google.com/T/#u
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: James Chapman <jchapman@katalix.com>
---
 net/l2tp/l2tp_core.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/net/l2tp/l2tp_core.c b/net/l2tp/l2tp_core.c
index 687c1366a4d0f4afb5ba2800e8faff3cb9c5437b..3eed63d8c9cdcb10cad0eb260dcddc4ac4272634 100644
--- a/net/l2tp/l2tp_core.c
+++ b/net/l2tp/l2tp_core.c
@@ -1414,8 +1414,6 @@ static void l2tp_tunnel_del_work(struct work_struct *work)
 {
 	struct l2tp_tunnel *tunnel = container_of(work, struct l2tp_tunnel,
 						  del_work);
-	struct sock *sk = tunnel->sock;
-	struct socket *sock = sk->sk_socket;
 
 	l2tp_tunnel_closeall(tunnel);
 
@@ -1423,6 +1421,8 @@ static void l2tp_tunnel_del_work(struct work_struct *work)
 	 * the sk API to release it here.
 	 */
 	if (tunnel->fd < 0) {
+		struct socket *sock = tunnel->sock->sk_socket;
+
 		if (sock) {
 			kernel_sock_shutdown(sock, SHUT_RDWR);
 			sock_release(sock);
-- 
2.52.0.457.g6b5491de43-goog

Re: [PATCH net] l2tp: avoid one data-race in l2tp_tunnel_del_work()

[Guillaume Nault]

On Thu, Jan 15, 2026 at 09:21:39AM +0000, Eric Dumazet wrote:
> We should read sk->sk_socket only when dealing with kernel sockets.
> 
> syzbot reported the following data-race:
> 
> BUG: KCSAN: data-race in l2tp_tunnel_del_work / sk_common_release
> 
> write to 0xffff88811c182b20 of 8 bytes by task 5365 on cpu 0:
>   sk_set_socket include/net/sock.h:2092 [inline]
>   sock_orphan include/net/sock.h:2118 [inline]
>   sk_common_release+0xae/0x230 net/core/sock.c:4003
>   udp_lib_close+0x15/0x20 include/net/udp.h:325
>   inet_release+0xce/0xf0 net/ipv4/af_inet.c:437
>   __sock_release net/socket.c:662 [inline]
>   sock_close+0x6b/0x150 net/socket.c:1455
>   __fput+0x29b/0x650 fs/file_table.c:468
>   ____fput+0x1c/0x30 fs/file_table.c:496
>   task_work_run+0x131/0x1a0 kernel/task_work.c:233
>   resume_user_mode_work include/linux/resume_user_mode.h:50 [inline]
>   __exit_to_user_mode_loop kernel/entry/common.c:44 [inline]
>   exit_to_user_mode_loop+0x1fe/0x740 kernel/entry/common.c:75
>   __exit_to_user_mode_prepare include/linux/irq-entry-common.h:226 [inline]
>   syscall_exit_to_user_mode_prepare include/linux/irq-entry-common.h:256 [inline]
>   syscall_exit_to_user_mode_work include/linux/entry-common.h:159 [inline]
>   syscall_exit_to_user_mode include/linux/entry-common.h:194 [inline]
>   do_syscall_64+0x1e1/0x2b0 arch/x86/entry/syscall_64.c:100
>  entry_SYSCALL_64_after_hwframe+0x77/0x7f
> 
> read to 0xffff88811c182b20 of 8 bytes by task 827 on cpu 1:
>   l2tp_tunnel_del_work+0x2f/0x1a0 net/l2tp/l2tp_core.c:1418
>   process_one_work kernel/workqueue.c:3257 [inline]
>   process_scheduled_works+0x4ce/0x9d0 kernel/workqueue.c:3340
>   worker_thread+0x582/0x770 kernel/workqueue.c:3421
>   kthread+0x489/0x510 kernel/kthread.c:463
>   ret_from_fork+0x149/0x290 arch/x86/kernel/process.c:158
>   ret_from_fork_asm+0x1a/0x30 arch/x86/entry/entry_64.S:246
> 
> value changed: 0xffff88811b818000 -> 0x0000000000000000
> 
> Fixes: d00fa9adc528 ("l2tp: fix races with tunnel socket close")

Reviewed-by: Guillaume Nault <gnault@redhat.com>

Re: [PATCH net] l2tp: avoid one data-race in l2tp_tunnel_del_work()

[patchwork-bot+netdevbpf]

Hello:

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

On Thu, 15 Jan 2026 09:21:39 +0000 you wrote:
> We should read sk->sk_socket only when dealing with kernel sockets.
> 
> syzbot reported the following data-race:
> 
> BUG: KCSAN: data-race in l2tp_tunnel_del_work / sk_common_release
> 
> write to 0xffff88811c182b20 of 8 bytes by task 5365 on cpu 0:
>   sk_set_socket include/net/sock.h:2092 [inline]
>   sock_orphan include/net/sock.h:2118 [inline]
>   sk_common_release+0xae/0x230 net/core/sock.c:4003
>   udp_lib_close+0x15/0x20 include/net/udp.h:325
>   inet_release+0xce/0xf0 net/ipv4/af_inet.c:437
>   __sock_release net/socket.c:662 [inline]
>   sock_close+0x6b/0x150 net/socket.c:1455
>   __fput+0x29b/0x650 fs/file_table.c:468
>   ____fput+0x1c/0x30 fs/file_table.c:496
>   task_work_run+0x131/0x1a0 kernel/task_work.c:233
>   resume_user_mode_work include/linux/resume_user_mode.h:50 [inline]
>   __exit_to_user_mode_loop kernel/entry/common.c:44 [inline]
>   exit_to_user_mode_loop+0x1fe/0x740 kernel/entry/common.c:75
>   __exit_to_user_mode_prepare include/linux/irq-entry-common.h:226 [inline]
>   syscall_exit_to_user_mode_prepare include/linux/irq-entry-common.h:256 [inline]
>   syscall_exit_to_user_mode_work include/linux/entry-common.h:159 [inline]
>   syscall_exit_to_user_mode include/linux/entry-common.h:194 [inline]
>   do_syscall_64+0x1e1/0x2b0 arch/x86/entry/syscall_64.c:100
>  entry_SYSCALL_64_after_hwframe+0x77/0x7f
> 
> [...]

Here is the summary with links:
  - [net] l2tp: avoid one data-race in l2tp_tunnel_del_work()
    https://git.kernel.org/netdev/net/c/7a29f6bf60f2

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