Re: [Codel] Codel on DPDK

[Dave Taht]

On Tue, Oct 18, 2016 at 8:43 AM, Dave Taht  wrote:
> On Tue, Oct 18, 2016 at 4:42 AM,   wrote:
>> Hello
>>
>> I'm trying to implement CoDel-like implementation on Intel's DPDK library,
>> but I have trouble with measuring packet sojourn time on RX/TX queues
>> because of DMA usage. When I call send() function, DPDK puts the packets
>> into DMA-region and I have no idea when the packets sent really into
>> network. Likewise, when I call recv(), I don't have any idea when the
>> packets arrived, so I'm not sure totally but I think I have no chance to
>> measure sojourn time for dma-like implementation. Could you please correct
>> me if I'm wrong. Also any comments/suggestions would be greatly appreciated.
>
> I am under the impression there is a dpdk implementation already, somewhere.

codel, pie, etc, for dpdk: https://git.scc.kit.edu/TM/DPDK_AQM_Switch.git

I have not looked this over for accuracy or completeness.

> As for the amount of time things spend in dma, you are hopefully
> talking about a few mere
> us, and the default for codel is only worrying about ms, so you can do
> the codel work before the send and after the recv.
>
>> Thanks in advance..
>>
>> - Volkan
>>
>>
>> ___
>> Codel mailing list
>> Codel@lists.bufferbloat.net
>> https://lists.bufferbloat.net/listinfo/codel
>>
>
>
>
> --
> Dave Täht
> Let's go make home routers and wifi faster! With better software!
> http://blog.cerowrt.org



-- 
Dave Täht
Let's go make home routers and wifi faster! With better software!
http://blog.cerowrt.org
___
Codel mailing list
Codel@lists.bufferbloat.net
https://lists.bufferbloat.net/listinfo/codel

Re: [Codel] Codel on DPDK

[Dave Taht]

On Tue, Oct 18, 2016 at 4:42 AM,   wrote:
> Hello
>
> I'm trying to implement CoDel-like implementation on Intel's DPDK library,
> but I have trouble with measuring packet sojourn time on RX/TX queues
> because of DMA usage. When I call send() function, DPDK puts the packets
> into DMA-region and I have no idea when the packets sent really into
> network. Likewise, when I call recv(), I don't have any idea when the
> packets arrived, so I'm not sure totally but I think I have no chance to
> measure sojourn time for dma-like implementation. Could you please correct
> me if I'm wrong. Also any comments/suggestions would be greatly appreciated.

I am under the impression there is a dpdk implementation already, somewhere.

As for the amount of time things spend in dma, you are hopefully
talking about a few mere
us, and the default for codel is only worrying about ms, so you can do
the codel work before the send and after the recv.

> Thanks in advance..
>
> - Volkan
>
>
> ___
> Codel mailing list
> Codel@lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/codel
>



-- 
Dave Täht
Let's go make home routers and wifi faster! With better software!
http://blog.cerowrt.org
___
Codel mailing list
Codel@lists.bufferbloat.net
https://lists.bufferbloat.net/listinfo/codel