Regression on TX throughput when using bonding

On Thu, 2012-06-14 at 10:58 +0200, Jean-Michel Hautbois wrote:
> Hi all,
> 
> I have bisected a regression which concerns TX throughput when using bonding.
> I tested only with 10Gbps cards, as it appears when bandwidth need is
> over 1Gbps on my machine.
> I send UDP multicast packets over bonding and observe the tc result.
> 
> When KO :
> $>tc -s -d qdisc show dev eth1
> qdisc pfifo_fast 0: root refcnt 2 bands 3 priomap  1 2 2 2 1 2 0 0 1 1
> 1 1 1 1 1 1
>  Sent 1106527591 bytes 273802 pkt (dropped 306419, overlimits 0 requeues 223)
>  backlog 0b 0p requeues 223
> 
> Ok course, when OK, dropped is 0.
> $>tc -s -d qdisc show dev eth1
> qdisc pfifo_fast 0: root refcnt 2 bands 3 priomap  1 2 2 2 1 2 0 0 1 1
> 1 1 1 1 1 1
>  Sent 1648662087 bytes 408009 pkt (dropped 0, overlimits 0 requeues 0)
>  backlog 0b 0p requeues 0
> 
> 
> Here is the incriminated commit:
> 
> fc6055a5ba31e2c14e36e8939f9bf2b6d586a7f5 is the first bad commit
> commit fc6055a5ba31e2c14e36e8939f9bf2b6d586a7f5

>     net: Introduce skb_orphan_try()

So you are saying that if you make skb_orphan_try() doing nothing, it
solves your problem ?



--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

2012/6/14 Eric Dumazet <eric.dumazet@gmail.com>:
> On Thu, 2012-06-14 at 10:58 +0200, Jean-Michel Hautbois wrote:
>> Hi all,
>>
>> I have bisected a regression which concerns TX throughput when using bonding.
>> I tested only with 10Gbps cards, as it appears when bandwidth need is
>> over 1Gbps on my machine.
>> I send UDP multicast packets over bonding and observe the tc result.
>>
>> When KO :
>> $>tc -s -d qdisc show dev eth1
>> qdisc pfifo_fast 0: root refcnt 2 bands 3 priomap  1 2 2 2 1 2 0 0 1 1
>> 1 1 1 1 1 1
>>  Sent 1106527591 bytes 273802 pkt (dropped 306419, overlimits 0 requeues 223)
>>  backlog 0b 0p requeues 223
>>
>> Ok course, when OK, dropped is 0.
>> $>tc -s -d qdisc show dev eth1
>> qdisc pfifo_fast 0: root refcnt 2 bands 3 priomap  1 2 2 2 1 2 0 0 1 1
>> 1 1 1 1 1 1
>>  Sent 1648662087 bytes 408009 pkt (dropped 0, overlimits 0 requeues 0)
>>  backlog 0b 0p requeues 0
>>
>>
>> Here is the incriminated commit:
>>
>> fc6055a5ba31e2c14e36e8939f9bf2b6d586a7f5 is the first bad commit
>> commit fc6055a5ba31e2c14e36e8939f9bf2b6d586a7f5
>
>>     net: Introduce skb_orphan_try()
>
> So you are saying that if you make skb_orphan_try() doing nothing, it
> solves your problem ?
>
> diff --git a/net/core/dev.c b/net/core/dev.c
> index cd09819..6df40dd 100644
> --- a/net/core/dev.c
> +++ b/net/core/dev.c
> @@ -2096,6 +2096,7 @@ static int dev_gso_segment(struct sk_buff *skb, netdev_features_t features)
>  */
>  static inline void skb_orphan_try(struct sk_buff *skb)
>  {
> +#if 0
>        struct sock *sk = skb->sk;
>
>        if (sk && !skb_shinfo(skb)->tx_flags) {
> @@ -2106,6 +2107,7 @@ static inline void skb_orphan_try(struct sk_buff *skb)
>                        skb->rxhash = sk->sk_hash;
>                skb_orphan(skb);
>        }
> +#endif
>  }
>
>  static bool can_checksum_protocol(netdev_features_t features, __be16 protocol)
>

Yes :

[09:44:46]root@debian:~# uname -r
3.2.0
[09:44:50]root@debian:~# tc -s -d qdisc show dev eth1
qdisc mq 0: root
 Sent 11786861758 bytes 2916943 pkt (dropped 0, overlimits 0 requeues 0)
 backlog 0b 0p requeues 0

JM
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html