netfilter: nf_tables: always release netdev hooks from notifier

Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=2189550
Upstream Status: commit dc1c9fd

commit dc1c9fd
Author: Florian Westphal <fw@strlen.de>
Date:   Thu May 4 14:20:21 2023 +0200

    netfilter: nf_tables: always release netdev hooks from notifier

    This reverts "netfilter: nf_tables: skip netdev events generated on netns removal".

    The problem is that when a veth device is released, the veth release
    callback will also queue the peer netns device for removal.

    Its possible that the peer netns is also slated for removal.  In this
    case, the device memory is already released before the pre_exit hook of
    the peer netns runs:

    BUG: KASAN: slab-use-after-free in nf_hook_entry_head+0x1b8/0x1d0
    Read of size 8 at addr ffff88812c0124f0 by task kworker/u8:1/45
    Workqueue: netns cleanup_net
    Call Trace:
     nf_hook_entry_head+0x1b8/0x1d0
     __nf_unregister_net_hook+0x76/0x510
     nft_netdev_unregister_hooks+0xa0/0x220
     __nft_release_hook+0x184/0x490
     nf_tables_pre_exit_net+0x12f/0x1b0
     ..

    Order is:
    1. First netns is released, veth_dellink() queues peer netns device
       for removal
    2. peer netns is queued for removal
    3. peer netns device is released, unreg event is triggered
    4. unreg event is ignored because netns is going down
    5. pre_exit hook calls nft_netdev_unregister_hooks but device memory
       might be free'd already.

    Fixes: 68a3765 ("netfilter: nf_tables: skip netdev events generated on netns removal")
    Signed-off-by: Florian Westphal <fw@strlen.de>
    Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>

Signed-off-by: Florian Westphal <fwestpha@redhat.com>

Author: fw-strlen

net/netfilter/nft_chain_filter.c

@@ -344,6 +344,12 @@ static void nft_netdev_event(unsigned long event, struct net_device *dev,
 		return;
 	}
 
+	/* UNREGISTER events are also happening on netns exit.
+	 *
+	 * Although nf_tables core releases all tables/chains, only this event
+	 * handler provides guarantee that hook->ops.dev is still accessible,
+	 * so we cannot skip exiting net namespaces.
+	 */
 	__nft_release_basechain(ctx);
 }
 
@@ -362,9 +368,6 @@ static int nf_tables_netdev_event(struct notifier_block *this,
 	    event != NETDEV_CHANGENAME)
 		return NOTIFY_DONE;
 
-	if (!check_net(ctx.net))
-		return NOTIFY_DONE;
-
 	nft_net = nft_pernet(ctx.net);
 	mutex_lock(&nft_net->commit_mutex);
 	list_for_each_entry(table, &nft_net->tables, list) {