lists.openwall.net   lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  linux-hardening  linux-cve-announce  PHC 
Open Source and information security mailing list archives
 
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [day] [month] [year] [list]
Message-Id: <20241107173134.542802-1-jchapman@katalix.com>
Date: Thu,  7 Nov 2024 17:31:34 +0000
From: James Chapman <jchapman@...alix.com>
To: netdev@...r.kernel.org
Subject: [PATCH RFC] l2tp: fix warning in l2tp_exit_net found by syzbot

The following patch fixes a WARN_ON in l2tp found by syzbot but I am
not confident that it's the right approach, hence this RFC.

In l2tp's net exit handler, we check that an IDR is empty before
destroying it:

	WARN_ON_ONCE(!idr_is_empty(&pn->l2tp_tunnel_idr));
	idr_destroy(&pn->l2tp_tunnel_idr);

However, syzbot is able to provoke a condition where idr_is_empty
returns false despite there being no items in the IDR. This turns out
to be because the radix tree of the IDR contains one or more internal
radix-tree nodes and these cause idr_is_empty to return false. These
internal nodes are cleaned by idr_destroy.

Using idr_for_each to check that the IDR is empty instead of
idr_is_empty, as per the included patch, avoids the problem. But the
idr_is_empty pattern is used in other code too,
e.g. drivers/net/ethernet/mellanox/mlxsw/spectrum_pgt.c

Should idr_is_empty be returning false if the radix tree contains only
internal nodes?

Ref: https://syzkaller.appspot.com/bug?extid=332fe1e67018625f63c9

---
 net/l2tp/l2tp_core.c | 22 +++++++++++++++++++---
 1 file changed, 19 insertions(+), 3 deletions(-)

diff --git a/net/l2tp/l2tp_core.c b/net/l2tp/l2tp_core.c
index 3eec23ac5ab1..369a2f2e459c 100644
--- a/net/l2tp/l2tp_core.c
+++ b/net/l2tp/l2tp_core.c
@@ -1870,15 +1870,31 @@ static __net_exit void l2tp_pre_exit_net(struct net *net)
 	}
 }
 
+static int l2tp_idr_item_unexpected(int id, void *p, void *data)
+{
+	const char *idr_name = data;
+
+	pr_err("l2tp: %s IDR not empty at net %d exit\n", idr_name, id);
+	WARN_ON_ONCE(1);
+	return 1;
+}
+
 static __net_exit void l2tp_exit_net(struct net *net)
 {
 	struct l2tp_net *pn = l2tp_pernet(net);
 
-	WARN_ON_ONCE(!idr_is_empty(&pn->l2tp_v2_session_idr));
+	/* Our per-net IDRs should be empty. Check that is so, to
+	 * help catch cleanup races or refcnt leaks.
+	 */
+	idr_for_each(&pn->l2tp_v2_session_idr, l2tp_idr_item_unexpected,
+		     "v2_session");
+	idr_for_each(&pn->l2tp_v3_session_idr, l2tp_idr_item_unexpected,
+		     "v3_session");
+	idr_for_each(&pn->l2tp_tunnel_idr, l2tp_idr_item_unexpected,
+		     "tunnel");
+
 	idr_destroy(&pn->l2tp_v2_session_idr);
-	WARN_ON_ONCE(!idr_is_empty(&pn->l2tp_v3_session_idr));
 	idr_destroy(&pn->l2tp_v3_session_idr);
-	WARN_ON_ONCE(!idr_is_empty(&pn->l2tp_tunnel_idr));
 	idr_destroy(&pn->l2tp_tunnel_idr);
 }
 
-- 
2.34.1


Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ