Re: [PATCH net-next v2] ip_tunnel: Move stats update to iptunnel_xmit()

[Nicolas Dichtel <nicolas.dichtel@6wind.com>]

Le 24/12/2015 13:46, Thadeu Lima de Souza Cascardo a écrit :
> On Thu, Dec 24, 2015 at 10:21:27AM +0100, Nicolas Dichtel wrote:
>> Le 24/12/2015 00:52, Pravin B Shelar a écrit :
>> [snip]
>>> diff --git a/net/tipc/udp_media.c b/net/tipc/udp_media.c
>>> index 6af78c6..d63a911 100644
>>> --- a/net/tipc/udp_media.c
>>> +++ b/net/tipc/udp_media.c
>>> @@ -182,15 +182,9 @@ static int tipc_udp_send_msg(struct net *net, struct sk_buff *skb,
>>>   			goto tx_error;
>>>   		}
>>>   		ttl = ip4_dst_hoplimit(&rt->dst);
>>> -		err = udp_tunnel_xmit_skb(rt, ub->ubsock->sk, skb,
>>> -					  src->ipv4.s_addr,
>>> -					  dst->ipv4.s_addr, 0, ttl, 0,
>>> -					  src->udp_port, dst->udp_port,
>>> -					  false, true);
>>> -		if (err < 0) {
>>> -			ip_rt_put(rt);
>>> -			goto tx_error;
>>> -		}
>>> +		udp_tunnel_xmit_skb(rt, ub->ubsock->sk, skb, src->ipv4.s_addr,
>>> +				    dst->ipv4.s_addr, 0, ttl, 0, src->udp_port,
>>> +				    dst->udp_port, false, true);
>> I don't know how tipc works, but this change is clearly suspect. What make the
>> error path not needed anymore after your patch?
>
> I looked into it as well. As far as I see, err could only be positive or 0, so
> if there is a tipc bug here, Pravin's patch introduces no regression. Or did I
> fail to see how udp_tunnel_xmit_skb could return a negative value?
You're probably right. But I think it needs a separate patch in that case.


Regards,
Nicolas