[ovs-dev] [PATCH 0/5] Quicker pmd threads reloads

Ilya Maximets i.maximets at samsung.com
Thu Jun 6 07:57:08 UTC 2019


On 06.06.2019 10:35, David Marchand wrote:
> Hello guys,
> 
> On Thu, May 23, 2019 at 4:27 PM David Marchand <david.marchand at redhat.com <mailto:david.marchand at redhat.com>> wrote:
> 
>     We have been testing the rebalance code in different situations while
>     having traffic going through OVS.
>     Those tests have shown that part of the observed packets losses is due to
>     some time wasted in signaling/waiting for the pmd threads to reload their
>     polling configurations.
> 
>     This series is an attempt at getting pmd threads reloads quicker and
>     more deterministic.
> 
>     Example of number of cycles spent by a pmd between two polling
>     configurations (in cycles minimum/average/maximum of 1000 changes):
>     - d58b59c17c70: 126822/312103/756580
>     - patch1:       113658/296157/741688
>     - patch2:        49198/167206/466108
>     - patch3:        13032/120730/341163
>     - patch4:        12803/112964/323455
>     - patch5:        13633/ 20373/ 47410
> 
>     Changelog since RFC v2:
>     - added ack from Eelco
>     -
> 
>     Changelog since RFC v1:
>     - added numbers per patch in cover letter
>     - added memory ordering for explicit synchronisations between threads
>       in patch 1 and patch 2
> 
> 
> I did not get feedback on this series apart from Eelco.
> Did you have a chance to look at it?

Hi David.
Sorry for delays. Patches seems fine at the first glance.
I hope to have some time next week to look closer and, probably,
test something.

Best regards, Ilya Maximets.


More information about the dev mailing list