vsock: fix lock inversion in vsock_assign_transport()

jira VULN-80686
cve-bf CVE-2025-38461
commit-author Stefano Garzarella <[email protected]>
commit f7c877e

Syzbot reported a potential lock inversion deadlock between
vsock_register_mutex and sk_lock-AF_VSOCK when vsock_linger() is called.

The issue was introduced by commit 687aa0c ("vsock: Fix
transport_* TOCTOU") which added vsock_register_mutex locking in
vsock_assign_transport() around the transport->release() call, that can
call vsock_linger(). vsock_assign_transport() can be called with sk_lock
held. vsock_linger() calls sk_wait_event() that temporarily releases and
re-acquires sk_lock. During this window, if another thread hold
vsock_register_mutex while trying to acquire sk_lock, a circular
dependency is created.

Fix this by releasing vsock_register_mutex before calling
transport->release() and vsock_deassign_transport(). This is safe
because we don't need to hold vsock_register_mutex while releasing the
old transport, and we ensure the new transport won't disappear by
obtaining a module reference first via try_module_get().

	Reported-by: [email protected]
	Tested-by: [email protected]
Fixes: 687aa0c ("vsock: Fix transport_* TOCTOU")
	Cc: [email protected]
	Cc: [email protected]
	Signed-off-by: Stefano Garzarella <[email protected]>
Link: https://patch.msgid.link/[email protected]
	Signed-off-by: Paolo Abeni <[email protected]>

(cherry picked from commit f7c877e)
	Signed-off-by: Shreeya Patel <[email protected]>

Author: shreeya-patel98

net/vmw_vsock/af_vsock.c

@@ -479,20 +479,9 @@ int vsock_assign_transport(struct vsock_sock *vsk, struct vsock_sock *psk)
 		goto err;
 	}
 
-	if (vsk->transport) {
-		if (vsk->transport == new_transport) {
-			ret = 0;
-			goto err;
-		}
-
-		/* transport->release() must be called with sock lock acquired.
-		 * This path can only be taken during vsock_connect(), where we
-		 * have already held the sock lock. In the other cases, this
-		 * function is called on a new socket which is not assigned to
-		 * any transport.
-		 */
-		vsk->transport->release(vsk);
-		vsock_deassign_transport(vsk);
+	if (vsk->transport && vsk->transport == new_transport) {
+		ret = 0;
+		goto err;
 	}
 
 	/* We increase the module refcnt to prevent the transport unloading
@@ -509,6 +498,17 @@ int vsock_assign_transport(struct vsock_sock *vsk, struct vsock_sock *psk)
 	 */
 	mutex_unlock(&vsock_register_mutex);
 
+	if (vsk->transport) {
+		/* transport->release() must be called with sock lock acquired.
+		 * This path can only be taken during vsock_connect(), where we
+		 * have already held the sock lock. In the other cases, this
+		 * function is called on a new socket which is not assigned to
+		 * any transport.
+		 */
+		vsk->transport->release(vsk);
+		vsock_deassign_transport(vsk);
+	}
+
 	if (sk->sk_type == SOCK_SEQPACKET) {
 		if (!new_transport->seqpacket_allow ||
 		    !new_transport->seqpacket_allow(remote_cid)) {