[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <20171128.095837.899409439582426451.davem@davemloft.net>
Date: Tue, 28 Nov 2017 09:58:37 -0500 (EST)
From: David Miller <davem@...emloft.net>
To: tommi.t.rantala@...ia.com
Cc: jon.maloy@...csson.com, ying.xue@...driver.com,
netdev@...r.kernel.org, tipc-discussion@...ts.sourceforge.net,
linux-kernel@...r.kernel.org
Subject: Re: [PATCH] tipc: call tipc_rcv() only if bearer is up in
tipc_udp_recv()
From: Tommi Rantala <tommi.t.rantala@...ia.com>
Date: Tue, 28 Nov 2017 14:53:15 +0200
> Call tipc_rcv() only if bearer is up in tipc_udp_recv().
> Fixes a rare TIPC div-by-zero crash in tipc_node_calculate_timer():
>
> We're enabling a bearer, but it's not yet up and fully initialized.
> At the same time we receive a discovery packet, and in tipc_udp_recv()
> we end up calling tipc_rcv() with the not-yet-initialized bearer,
> causing later a div-by-zero crash in tipc_node_calculate_timer().
You're also now ignoring any error being returned by tipc_udp_rcast_disc().
> -
> - if (unlikely(msg_user(hdr) == LINK_CONFIG)) {
> - err = tipc_udp_rcast_disc(b, skb);
> - if (err)
> - goto rcu_out;
> + } else {
> + if (unlikely(b && msg_user(hdr) == LINK_CONFIG))
> + tipc_udp_rcast_disc(b, skb);
> + kfree_skb(skb);
> }
Either put the 'err' propagation back or clearly explain in your
commit log message why this part of the change if absolutely essential
for this bug fix.
Thank you.
Powered by blists - more mailing lists