Re: [PATCH 2/2] net/tcp: Disable TCP-MD5 static key on tcp_md5sig_info destruction

On 11/2/22 21:25, Eric Dumazet wrote:

On Wed, Nov 2, 2022 at 2:14 PM Dmitry Safonov [off-list ref] wrote:

[..]

quoted

@@ -337,11 +338,13 @@ EXPORT_SYMBOL(tcp_time_wait);
 void tcp_twsk_destructor(struct sock *sk)
 {
 #ifdef CONFIG_TCP_MD5SIG
-       if (static_branch_unlikely(&tcp_md5_needed)) {
+       if (static_branch_unlikely(&tcp_md5_needed.key)) {
                struct tcp_timewait_sock *twsk = tcp_twsk(sk);

-               if (twsk->tw_md5_key)
+               if (twsk->tw_md5_key) {

Orthogonal to this patch, but I wonder why we do not clear
twsk->tw_md5_key before kfree_rcu()

It seems a lookup could catch the invalid pointer.

quoted

                        kfree_rcu(twsk->tw_md5_key, rcu);
+                       static_branch_slow_dec_deferred(&tcp_md5_needed);
+               }
        }

I looked into that, it seems tcp_twsk_destructor() is called from
inet_twsk_free(), which is either called from:
1. inet_twsk_put(), protected by tw->tw_refcnt
2. sock_gen_put(), protected by the same sk->sk_refcnt

So, in result, if I understand correctly, lookups should fail on ref
counter check. Maybe I'm missing something, but clearing here seems not
necessary?

I can add rcu_assign_pointer() just in case the destruction path changes
in v2 if you think it's worth it :-)

Thanks,
          Dmitry