[PATCH net v2] net: kcm: Fix race condition in kcm_unattach()

[PATCH net v2] net: kcm: Fix race condition in kcm_unattach()

[Sven Stegemann]

syzbot found a race condition when kcm_unattach(psock)
and kcm_release(kcm) are executed at the same time.

kcm_unattach() is missing a check of the flag
kcm->tx_stopped before calling queue_work().

If the kcm has a reserved psock, kcm_unattach() might get executed
between cancel_work_sync() and unreserve_psock() in kcm_release(),
requeuing kcm->tx_work right before kcm gets freed in kcm_done().

Remove kcm->tx_stopped and replace it by the less
error-prone disable_work_sync().

Fixes: ab7ac4eb9832 ("kcm: Kernel Connection Multiplexor module")
Reported-by: syzbot+e62c9db591c30e174662@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=e62c9db591c30e174662
Reported-by: syzbot+d199b52665b6c3069b94@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=d199b52665b6c3069b94
Reported-by: syzbot+be6b1fdfeae512726b4e@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=be6b1fdfeae512726b4e
Signed-off-by: Sven Stegemann <sven@stegemann.de>
---
v2:
 - merge disable_work() and cancel_work_sync() into disable_work_sync()
v1: https://lore.kernel.org/netdev/20250809063622.117420-1-sven@stegemann.de/
---
 include/net/kcm.h |  1 -
 net/kcm/kcmsock.c | 10 ++--------
 2 files changed, 2 insertions(+), 9 deletions(-)

diff --git a/include/net/kcm.h b/include/net/kcm.h
index 441e993be634..d9c35e71ecea 100644
--- a/include/net/kcm.h
+++ b/include/net/kcm.h
@@ -71,7 +71,6 @@ struct kcm_sock {
 	struct list_head wait_psock_list;
 	struct sk_buff *seq_skb;
 	struct mutex tx_mutex;
-	u32 tx_stopped : 1;
 
 	/* Don't use bit fields here, these are set under different locks */
 	bool tx_wait;
diff --git a/net/kcm/kcmsock.c b/net/kcm/kcmsock.c
index a4971e6fa943..b4f01cb07561 100644
--- a/net/kcm/kcmsock.c
+++ b/net/kcm/kcmsock.c
@@ -430,7 +430,7 @@ static void psock_write_space(struct sock *sk)
 
 	/* Check if the socket is reserved so someone is waiting for sending. */
 	kcm = psock->tx_kcm;
-	if (kcm && !unlikely(kcm->tx_stopped))
+	if (kcm)
 		queue_work(kcm_wq, &kcm->tx_work);
 
 	spin_unlock_bh(&mux->lock);
@@ -1693,12 +1693,6 @@ static int kcm_release(struct socket *sock)
 	 */
 	__skb_queue_purge(&sk->sk_write_queue);
 
-	/* Set tx_stopped. This is checked when psock is bound to a kcm and we
-	 * get a writespace callback. This prevents further work being queued
-	 * from the callback (unbinding the psock occurs after canceling work.
-	 */
-	kcm->tx_stopped = 1;
-
 	release_sock(sk);
 
 	spin_lock_bh(&mux->lock);
@@ -1714,7 +1708,7 @@ static int kcm_release(struct socket *sock)
 	/* Cancel work. After this point there should be no outside references
 	 * to the kcm socket.
 	 */
-	cancel_work_sync(&kcm->tx_work);
+	disable_work_sync(&kcm->tx_work);
 
 	lock_sock(sk);
 	psock = kcm->tx_psock;
-- 
2.50.1

Re: [PATCH net v2] net: kcm: Fix race condition in kcm_unattach()

[patchwork-bot+netdevbpf]

Hello:

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

On Tue, 12 Aug 2025 21:18:03 +0200 you wrote:
> syzbot found a race condition when kcm_unattach(psock)
> and kcm_release(kcm) are executed at the same time.
> 
> kcm_unattach() is missing a check of the flag
> kcm->tx_stopped before calling queue_work().
> 
> If the kcm has a reserved psock, kcm_unattach() might get executed
> between cancel_work_sync() and unreserve_psock() in kcm_release(),
> requeuing kcm->tx_work right before kcm gets freed in kcm_done().
> 
> [...]

Here is the summary with links:
  - [net,v2] net: kcm: Fix race condition in kcm_unattach()
    https://git.kernel.org/netdev/net/c/52565a935213

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