On Mon, Jul 04, 2016 at 03:48:30AM -0700, Jarno Rajahalme wrote:
>
> > On Jun 29, 2016, at 4:58 PM, Flavio Leitner wrote:
> >
> > On Wed, Jun 29, 2016 at 05:04:06AM -0700, Jarno Rajahalme wrote:
> >>
> >>> On Jun 23, 2016, at 11:40 AM, Flavio Leitner wrote:
> >>>
> ...
> >
> >>> +
> >>> +
> On Jun 29, 2016, at 4:58 PM, Flavio Leitner wrote:
>
> On Wed, Jun 29, 2016 at 05:04:06AM -0700, Jarno Rajahalme wrote:
>>
>>> On Jun 23, 2016, at 11:40 AM, Flavio Leitner wrote:
>>>
...
>
>>> +
>>> +ovs_assert(!single_threaded());
>>> +perthread = ovsrcu_perthread_get();
>>
>> Is
On Wed, Jun 29, 2016 at 05:04:06AM -0700, Jarno Rajahalme wrote:
>
> > On Jun 23, 2016, at 11:40 AM, Flavio Leitner wrote:
> >
> > The PMD thread needs to keep processing RX queues in order
> > to achieve maximum throughput. It also needs to sweep emc
> > cache and quiesce which use seq_mutex. T
> On Jun 23, 2016, at 11:40 AM, Flavio Leitner wrote:
>
> The PMD thread needs to keep processing RX queues in order
> to achieve maximum throughput. It also needs to sweep emc
> cache and quiesce which use seq_mutex. That mutex can
> eventually block the PMD thread causing latency spikes and
>
The PMD thread needs to keep processing RX queues in order
to achieve maximum throughput. It also needs to sweep emc
cache and quiesce which use seq_mutex. That mutex can
eventually block the PMD thread causing latency spikes and
affecting the throughput.
Since there is no requirement for running
