On 6/30/2020 11:57 PM, Eric Dumazet wrote:
>
> On 6/29/20 7:54 PM, we...@ucloud.cn wrote:
>> From: wenxu <we...@ucloud.cn>
>>
>> The fragment packets do defrag in act_ct module. The reassembled packet
>> over the mtu in the act_mirred. This big packet should be fragmented
>> to send out.
>>
>> Fixes: b57dc7c13ea9 ("net/sched: Introduce action ct")
>> Signed-off-by: wenxu <we...@ucloud.cn>
>> ---
>> This patch is based on
>> http://patchwork.ozlabs.org/project/netdev/patch/1593422178-26949-1-git-send-email-we...@ucloud.cn/
>>
>> include/net/sch_generic.h | 6 +-
>> net/sched/act_ct.c | 7 ++-
>> net/sched/act_mirred.c | 157
>> ++++++++++++++++++++++++++++++++++++++++++++--
>> 3 files changed, 158 insertions(+), 12 deletions(-)
>>
>> diff --git a/include/net/sch_generic.h b/include/net/sch_generic.h
>> index c510b03..3597244 100644
>> --- a/include/net/sch_generic.h
>> +++ b/include/net/sch_generic.h
>> @@ -384,6 +384,7 @@ struct qdisc_skb_cb {
>> };
>> #define QDISC_CB_PRIV_LEN 20
>> unsigned char data[QDISC_CB_PRIV_LEN];
>> + u16 mru;
>> };
>>
>
> Wow, this change is potentially a big problem.
>
> Explain why act_ct/act_mirred need to pollute qdisc_skb_cb
Hi Eric,
In the act_ct the fragment packets will defrag to a big packet and do conntrack
things.
But in the latter filter mirred action, the big packet normally send over the
mtu of outgoing device.
In this case the packet should be fragment to send And the frag size should
based on origin frag size
recored in act_ct. In the act_ct the frag size should retore in qdisc_skb_cb
and act_mirred get it to do fragment.
So there are any other good method?
BR
wenxu
>
>
>
