[PATCH 1/1] sched/rt: add rt throttled/unthrottled log

Always print logs when rt throttle/unthrottle, it is much easier to locate the throttled period. Signed-off-by: Jing-Ting Wu --- kernel/sched/rt.c |5 - 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/kernel/sched/rt.c b/kernel/sched/rt.c index f215eea..598046c 100644 --- a

[PATCH v2 1/1] sched/rt: avoid contend with CFS task

Changes since v2: We have revised the implementation to select a better idle CPU in the same sched_domain of prev_cpu (with the same cache hotness) when the RT task wakeup.

[PATCH v2 1/1] sched/rt: avoid contend with CFS task

2.41649.24 -53.17 -7.57 After apply this RT patch, launch time decrease about 8%. Change-Id: Ia0a7a61d38cb406d82b7049787c62b95dfa0a69f Signed-off-by: Jing-Ting Wu --- kernel/sched/rt.c | 56 + 1 file changed, 31 insert

Re: [PATCH 1/1] sched/rt: avoid contend with CFS task

uldn't prove yet this being a serious problem. > I was hoping the use case presented here is based on a real workload, but it's > synthetic. So I agree we need stronger reasons, but I think conceptually we do > have a conflict of interest where RT task could unnecessarily hurt the > performance of CFS task. > > Another way to look at the problem is that the system is not partitioned > correctly and the admin could do a better job to prevent this. > > -- > Qais Yousef I use some third-party application, such as weibo and others, to test the application launch time. I apply this RT patch, and compare it with original design. Both RT patch test case and original design test case are already apply the patch:https://lore.kernel.org/patchwork/patch/1129117/ After apply the RT patch, launch time of weibo from 1325.72ms to 1214.88 ms, its launch time decreases 110.84ms(about 8.36%). Other applications also decrease 7~13%. At original design test case, RT tasks(surfaceflinger) could preempt some CFS tasks, if we add all these CFS tasks runnable time, it may have some impact on app launch time. So even if we already use the load balance patch and reduce a lot of CFS runnable time, I think choose idle CPU at RT scheduler could also reduce the some CFS runnable time. Best regards, Jing-Ting Wu

Re: [PATCH 1/1] sched/rt: avoid contend with CFS task

On Thu, 2019-09-05 at 16:01 +0200, Vincent Guittot wrote: > Hi Jing-Ting, > > On Thu, 5 Sep 2019 at 15:26, Jing-Ting Wu wrote: > > > > On Fri, 2019-08-30 at 15:55 +0100, Qais Yousef wrote: > > > On 08/29/19 11:38, Valentin Schneider wrote: > > > &g

Re: [PATCH 1/1] sched/rt: avoid contend with CFS task

On Fri, 2019-08-30 at 15:55 +0100, Qais Yousef wrote: > On 08/29/19 11:38, Valentin Schneider wrote: > > On 29/08/2019 04:15, Jing-Ting Wu wrote: > > > At original linux design, RT & CFS scheduler are independent. > > > Current RT task placement poli

[PATCH 1/1] sched/rt: avoid contend with CFS task

t to avoid preempting CFS task. Signed-off-by: Jing-Ting Wu --- kernel/sched/rt.c | 42 +- 1 file changed, 17 insertions(+), 25 deletions(-) diff --git a/kernel/sched/rt.c b/kernel/sched/rt.c index a532558..626ca27 100644 --- a/kernel/sched/rt.c +++ b/ke