CVE-2026-46165

*** Pendiente de traducción *** In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> openvswitch: vport: fix self-deadlock on release of tunnel ports<br /> <br /> vports are used concurrently and protected by RCU, so netdev_put()<br /> must happen after the RCU grace period. So, either in an RCU call or<br /> after the synchronize_net(). The rtnl_delete_link() must happen under<br /> RTNL and so can&amp;#39;t be executed in RCU context. Calling synchronize_net()<br /> while holding RTNL is not a good idea for performance and system<br /> stability under load in general, so calling netdev_put() in RCU call<br /> is the right solution here.<br /> <br /> However,<br /> when the device is deleted, rtnl_unlock() will call netdev_run_todo()<br /> and block until all the references are gone. In the current code this<br /> means that we never reach the call_rcu() and the vport is never freed<br /> and the reference is never released, causing a self-deadlock on device<br /> removal.<br /> <br /> Fix that by moving the rcu_call() before the rtnl_unlock(), so the<br /> scheduled RCU callback will be executed when synchronize_net() is<br /> called from the rtnl_unlock()-&gt;netdev_run_todo() while the RTNL itself<br /> is already released.