[dpdk-dev] [RFC] ethdev: add new offload flag to keep CRC

Ferruh Yigit ferruh.yigit at intel.com
Mon Jun 11 11:18:18 CEST 2018


On 6/9/2018 11:11 AM, Andrew Rybchenko wrote:
> On 06/09/2018 01:57 AM, Ferruh Yigit wrote:
>> DEV_RX_OFFLOAD_KEEP_CRC offload flag added.
>>
>> DEV_RX_OFFLOAD_CRC_STRIP flag will remain one more release
>> default behavior in PMDs is to keep the CRC until this flag removed
>>
>> Until DEV_RX_OFFLOAD_CRC_STRIP flag is removed:
>> - Setting both KEEP_CRC & CRC_STRIP is INVALID
> 
> ethdev layer should enforce it.

OK, will add this check.

> 
>> - Setting only CRC_STRIP PMD should strip the CRC
>> - Setting only KEEP_CRC PMD should keep the CRC
>> - Not setting both PMD should keep the CRC
>>
>> Signed-off-by: Ferruh Yigit <ferruh.yigit at intel.com>
> 
> - rte_rx_offload_names should be updated as well
> - testpmd should be updated
> 
> I'm not sure that I understand the transition plan. I think PMDs which support
> KEEP_CRC should be updated by the patch to advertise the offload. Otherwise,
> generic checks in ethdev will not allow to enable it. 

Right, PMDs should be updated to advertise this offload, will do it.

> Other option is to simply
> drop it on ethdev layer (since basically it changes nothing for PMD now) and
> encourage PMD maintainers to advertise and take it into account if the feature
> is supported (however, if the bit is dropped on ethdev layer, the code would
> be unused/dead regardless application request). Too complicated. I guess
> there are not so many PMDs which support KEEP_CRC. Better to update them

> The patch should encourage applications which would like to keep CRC to
> use the offload since the next release will drop CRC_STRIP and it will
> change behaviour (no KEEP_CRC => strip it)

+1, will check apps.


More information about the dev mailing list