netdev - Re: [PATCH v2 net 3/3] wireguard: send: account for mtu=0 devices

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Fri, 14 Feb 2020 22:57:09 +0100
From:   "Jason A. Donenfeld" <Jason@...c4.com>
To:     Eric Dumazet <eric.dumazet@...il.com>
Cc:     Netdev <netdev@...r.kernel.org>
Subject: Re: [PATCH v2 net 3/3] wireguard: send: account for mtu=0 devices

Hey Eric,

On Fri, Feb 14, 2020 at 7:53 PM Eric Dumazet <eric.dumazet@...il.com> wrote:
> > The before passings off to the udp tunnel api, we indicate that we
> > support ip segmentation, and then it gets handled and fragmented
> > deeper down. Check out socket.c.
>
> Okay. Speaking of socket.c, I found this wg_socket_reinit() snippet :
>
> synchronize_rcu();
> synchronize_net();
>
> Which makes little sense. Please add a comment explaining why these two
> calls are needed.

Thanks, I appreciate your scrutiny here. Right again, you are. It
looks like that was added in 2017 after observing the pattern in other
drivers and seeing the documentation comment, "Wait for packets
currently being received to be done." That sounds like an important
thing to do before tearing down a socket. But here it makes no sense
at all, since synchronize_net() is just a wrapper around
synchronize_rcu() (and sometimes _expedited). And here, the
synchronize_rcu() usage makes sense to have, since this is as boring
of an rcu pattern as can be:

mutex_lock()
old = rcu_dereference_protected(x->y)
rcu_assign(x->y, new)
mutex_unlock()
synchronize_rcu()
free_it(old)

Straight out of the documentation. Having the extra synchronize_net()
in there adds nothing at all. I'll send a v3 of this 5.6-rc2 cleanup
series containing that removal.

Jason