[dpdk-dev] [PATCH v2 3/4] app/testpmd: fix packet header in txonly mode

Ferruh Yigit ferruh.yigit at intel.com
Thu Sep 17 13:16:31 CEST 2020


On 9/17/2020 8:10 AM, Chengchang Tang wrote:
> 
> 
> On 2020/9/15 0:23, Ferruh Yigit wrote:
>> On 8/20/2020 2:42 AM, Wei Hu (Xavier) wrote:
>>> From: Chengchang Tang <tangchengchang at huawei.com>
>>>
>>> In txonly forward mode, the packet header is fixed by the initial
>>> setting, including the packet length and checksum. So when the packets
>>> varies, this may cause a packet header error. Currently, there are two
>>> methods in txonly mode to randomly change the packets.
>>> 1. Set txsplit random and txpkts (x[,y]*), the number of segments
>>>     each packets will be a random value between 1 and total number of
>>>     segments determined by txpkts settings.
>>>     The step as follows:
>>>       a) ./testpmd -w xxx -l xx -n 4 -- -i --rxd xxxx --txd xxxx
>>>       b) set fwd txonly
>>>       c) set txsplit rand
>>>       d) set txpkts 2048,2048,2048,2048
>>>       e) start
>>> The nb_segs of the packets sent by testpmd will be 1~4. The real packet
>>> length will be 2048, 4096, 6144 and 8192. But in fact the packet length
>>> in ip header and udp header will be fixed by 8178 and 8158.
>>
>> Although I confirm the patch fixes the ip & udp header packet size for
>> "txsplit=rand" config,
>> I am always getting actual packet size wrong, independent from
>> 'txsplit', and it is always first segment size. Am I doing something wrong?
> 
> Yes, I miss it. If the txsplit is not set and the txpkts is set, the txonly
> fwd engine will only send single segment packets, and there will be a payload
> error for the data_len will be refreshed by the sum of the segment size.
> 

I am getting the size error when 'txsplit' is 'on' or 'rand'. In that 
case packet header are correct (after your patch) but wireshark 
complains the size of the packet on wire is different than the values in 
headers. And the size of the actual packet is reported by wireshark as 
the size of the first segment.

> May be the data_len should be the first segment size if the txsplit is not set?
> I will fix it in the next version.
>>
>> And not related to this patch but why setting 'txpkts' requires "--rxd
>> xxxx --txd xxxx" options explicitly set? If you are already there can
>> you also remove this restriction?
> 
> OK, I will fix it in next version.

Thanks, appreciated.



More information about the dev mailing list