On Wed, Oct 23, 2019 at 11:30 AM Cong Wang wrote:
>
> On Wed, Oct 23, 2019 at 11:14 AM Eric Dumazet wrote:
> > > In case you misunderstand, the CPU profiling I used is captured
> > > during 256 parallel TCP_STREAM.
> >
> > When I asked you the workload, you gave me TCP_RR output, not TCP_STREAM :
On Wed, Oct 23, 2019 at 11:14 AM Eric Dumazet wrote:
> > In case you misunderstand, the CPU profiling I used is captured
> > during 256 parallel TCP_STREAM.
>
> When I asked you the workload, you gave me TCP_RR output, not TCP_STREAM :/
>
> "A single netperf TCP_RR could _also_ confirm the improve
On Wed, Oct 23, 2019 at 10:40 AM Cong Wang wrote:
>
> On Tue, Oct 22, 2019 at 7:15 PM Eric Dumazet wrote:
> >
> > On Tue, Oct 22, 2019 at 6:10 PM Cong Wang wrote:
> > >
> > > On Tue, Oct 22, 2019 at 4:24 PM Eric Dumazet wrote:
> > > >
> > > > On Tue, Oct 22, 2019 at 4:11 PM Cong Wang
> > > >
On Tue, Oct 22, 2019 at 7:15 PM Eric Dumazet wrote:
>
> On Tue, Oct 22, 2019 at 6:10 PM Cong Wang wrote:
> >
> > On Tue, Oct 22, 2019 at 4:24 PM Eric Dumazet wrote:
> > >
> > > On Tue, Oct 22, 2019 at 4:11 PM Cong Wang
> > > wrote:
> > > >
> > > > Currently RTO, TLP and PROBE0 all share a same
On Tue, Oct 22, 2019 at 6:10 PM Cong Wang wrote:
>
> On Tue, Oct 22, 2019 at 4:24 PM Eric Dumazet wrote:
> >
> > On Tue, Oct 22, 2019 at 4:11 PM Cong Wang wrote:
> > >
> > > Currently RTO, TLP and PROBE0 all share a same timer instance
> > > in kernel and use icsk->icsk_pending to dispatch the w
On Tue, Oct 22, 2019 at 4:24 PM Eric Dumazet wrote:
>
> On Tue, Oct 22, 2019 at 4:11 PM Cong Wang wrote:
> >
> > Currently RTO, TLP and PROBE0 all share a same timer instance
> > in kernel and use icsk->icsk_pending to dispatch the work.
> > This causes spinlock contention when resetting the time
On Tue, Oct 22, 2019 at 4:11 PM Cong Wang wrote:
>
> Currently RTO, TLP and PROBE0 all share a same timer instance
> in kernel and use icsk->icsk_pending to dispatch the work.
> This causes spinlock contention when resetting the timer is
> too frequent, as clearly shown in the perf report:
>
>
Currently RTO, TLP and PROBE0 all share a same timer instance
in kernel and use icsk->icsk_pending to dispatch the work.
This causes spinlock contention when resetting the timer is
too frequent, as clearly shown in the perf report:
61.72%61.71% swapper [kernel.kallsyms]