[dpdk-dev] [PATCH v2] app/testpmd: fix physic port socket initialization

Ferruh Yigit ferruh.yigit at intel.com
Wed Oct 17 15:04:38 CEST 2018


On 10/17/2018 12:02 PM, Ananyev, Konstantin wrote:
> 
> 
>> -----Original Message-----
>> From: phil.yang at arm.com [mailto:phil.yang at arm.com]
>> Sent: Wednesday, October 17, 2018 2:37 AM
>> To: dev at dpdk.org
>> Cc: Ananyev, Konstantin <konstantin.ananyev at intel.com>; Yigit, Ferruh <ferruh.yigit at intel.com>
>> Subject: [PATCH v2] app/testpmd: fix physic port socket initialization
>>
>> Once the lcore list setting excluded the socket which physic device
>> attached, it will cause failure. Meanwhile, it will disable Testpmd
>> cross NUMA scenario.
>>
>> Fixes: dbfb8ec ("app/testpmd: optimize mbuf pool allocation")
>>
>> Signed-off-by: Phil Yang <phil.yang at arm.com>
>> Reviewed-by: Gavin Hu <Gavin.Hu at arm.com>>
> Acked-by: Konstantin Ananyev <konstantin.ananyev at intel.com>

Applied to dpdk-next-net/master, thanks.



What about 3rd item discussed before,
`port-numa-config` and `rxring-numa-config`

Currently if all cores and devices are in socket 0, there is no way for user to
ask allocating memory from socket 1. Again this happened after optimization patch.

For above case, when `port-numa-config` used to ask memory from socket 1:
`port-numa-config=(0,1)` testpmd will give error because no memory can be
allocated from socket 1.

What do you think, if user explicitly requested memory to be allocated from a
socket via `port-numa-config` and `rxring-numa-config`, and if that socket is
valid, add that socket into socket_ids[] so that mempool allocated for that socket?


More information about the dev mailing list