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

From: Dmitry Safonov <dima@arista.com>
To: Eric Dumazet <edumazet@google.com>
Cc: linux-kernel@vger.kernel.org, David Ahern <dsahern@kernel.org>,
	Bob Gilligan <gilligan@arista.com>,
	"David S. Miller" <davem@davemloft.net>,
	Dmitry Safonov <0x7f454c46@gmail.com>,
	Francesco Ruggeri <fruggeri@arista.com>,
	Hideaki YOSHIFUJI <yoshfuji@linux-ipv6.org>,
	Jakub Kicinski <kuba@kernel.org>, Paolo Abeni <pabeni@redhat.com>,
	Salam Noureddine <noureddine@arista.com>,
	netdev@vger.kernel.org
Subject: Re: [PATCH 2/2] net/tcp: Disable TCP-MD5 static key on tcp_md5sig_info destruction
Date: Thu, 3 Nov 2022 15:40:11 +0000	[thread overview]
Message-ID: <ad9be7ae-aeec-b208-8252-7a566534ded4@arista.com> (raw)
In-Reply-To: <CANn89i+UxgHwm9apzBXV-afpcfXfuX2S+6i4vPzF2ec4Dr6X0A@mail.gmail.com>

On 11/2/22 21:53, Eric Dumazet wrote:
> On Wed, Nov 2, 2022 at 2:49 PM Eric Dumazet <edumazet@google.com> wrote:
> 
>>
>> Are you sure ?
>>
>> static_branch_inc() is what we want here, it is a nice wrapper around
>> the correct internal details,
>> and ultimately boils to an atomic_inc(). It is safe for all contexts.
>>
>> But if/when jump labels get refcount_t one day, we will not have to
>> change TCP stack because
>> it made some implementation assumptions.
> 
> Oh, I think I understand this better now.
> 
> Please provide a helper like
> 
> static inline void static_key_fast_inc(struct static_key *key)
> {
>        atomic_inc(&key->enabled);
> }
> 
> Something like that.

Sure, that sounds like a better thing to do, rather than the hack I had.

Thanks, will send v2 soon,
          Dmitry