* [PATCH] l2tp: fix tunnel refcount leak on register failure
@ 2026-05-28 4:34 Wentao Liang
2026-05-28 10:31 ` Guillaume Nault
0 siblings, 1 reply; 2+ messages in thread
From: Wentao Liang @ 2026-05-28 4:34 UTC (permalink / raw)
To: James Chapman, David S . Miller, Eric Dumazet, Jakub Kicinski,
Paolo Abeni
Cc: Simon Horman, netdev, linux-kernel, Wentao Liang, stable
pppol2tp_tunnel_get() creates a new tunnel via l2tp_tunnel_create()
which initializes tunnel->ref_count to 1, then increments it to 2 via
refcount_inc(). If l2tp_tunnel_register() subsequently fails, the
error path calls kfree(tunnel) directly, bypassing the refcount
mechanism entirely. This leaves the tunnel's ref_count at 2 with no
way to properly release it through l2tp_tunnel_put().
Replace kfree(tunnel) with l2tp_tunnel_put(tunnel) to properly release
the reference through the standard refcount cleanup path, which will
call l2tp_tunnel_free() when the counter reaches zero.
Cc: stable@vger.kernel.org
Fixes: 6b9f34239b00 ("l2tp: fix races in tunnel creation")
Signed-off-by: Wentao Liang <vulab@iscas.ac.cn>
---
net/l2tp/l2tp_ppp.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/net/l2tp/l2tp_ppp.c b/net/l2tp/l2tp_ppp.c
index 99d6582f41de..16672e9df748 100644
--- a/net/l2tp/l2tp_ppp.c
+++ b/net/l2tp/l2tp_ppp.c
@@ -661,7 +661,7 @@ static struct l2tp_tunnel *pppol2tp_tunnel_get(struct net *net,
refcount_inc(&tunnel->ref_count);
error = l2tp_tunnel_register(tunnel, net, &tcfg);
if (error < 0) {
- kfree(tunnel);
+ l2tp_tunnel_put(tunnel);
return ERR_PTR(error);
}
--
2.34.1
^ permalink raw reply related [flat|nested] 2+ messages in thread* Re: [PATCH] l2tp: fix tunnel refcount leak on register failure
2026-05-28 4:34 [PATCH] l2tp: fix tunnel refcount leak on register failure Wentao Liang
@ 2026-05-28 10:31 ` Guillaume Nault
0 siblings, 0 replies; 2+ messages in thread
From: Guillaume Nault @ 2026-05-28 10:31 UTC (permalink / raw)
To: Wentao Liang
Cc: James Chapman, David S . Miller, Eric Dumazet, Jakub Kicinski,
Paolo Abeni, Simon Horman, netdev, linux-kernel, stable
On Thu, May 28, 2026 at 04:34:18AM +0000, Wentao Liang wrote:
> pppol2tp_tunnel_get() creates a new tunnel via l2tp_tunnel_create()
> which initializes tunnel->ref_count to 1, then increments it to 2 via
> refcount_inc(). If l2tp_tunnel_register() subsequently fails, the
> error path calls kfree(tunnel) directly, bypassing the refcount
> mechanism entirely.
This is the expected behaviour. Since l2tp_tunnel_register() failed,
the tunnel hasn't been exposed to the rest of the system and we're
guaranteed that nothing depends on it.
> This leaves the tunnel's ref_count at 2 with no
> way to properly release it through l2tp_tunnel_put().
The tunnel is simply freed. The refcount doesn't matter here.
> Replace kfree(tunnel) with l2tp_tunnel_put(tunnel) to properly release
> the reference through the standard refcount cleanup path, which will
> call l2tp_tunnel_free() when the counter reaches zero.
This is buggy. At this point, the refcount is 2, and l2tp_tunnel_put()
will decrement it by only one. So this patch just leaks the the tunnel
forever.
> Cc: stable@vger.kernel.org
> Fixes: 6b9f34239b00 ("l2tp: fix races in tunnel creation")
> Signed-off-by: Wentao Liang <vulab@iscas.ac.cn>
> ---
> net/l2tp/l2tp_ppp.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/net/l2tp/l2tp_ppp.c b/net/l2tp/l2tp_ppp.c
> index 99d6582f41de..16672e9df748 100644
> --- a/net/l2tp/l2tp_ppp.c
> +++ b/net/l2tp/l2tp_ppp.c
> @@ -661,7 +661,7 @@ static struct l2tp_tunnel *pppol2tp_tunnel_get(struct net *net,
> refcount_inc(&tunnel->ref_count);
> error = l2tp_tunnel_register(tunnel, net, &tcfg);
> if (error < 0) {
> - kfree(tunnel);
> + l2tp_tunnel_put(tunnel);
> return ERR_PTR(error);
> }
>
> --
> 2.34.1
>
>
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2026-05-28 10:31 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2026-05-28 4:34 [PATCH] l2tp: fix tunnel refcount leak on register failure Wentao Liang
2026-05-28 10:31 ` Guillaume Nault
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox