lkml.org 
[lkml]   [2019]   [Jul]   [31]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
SubjectRe: [PATCH 1/5] sched/deadline: Fix double accounting of rq/running bw in push_dl_task()
From
Date
On 7/29/19 10:00 AM, Dietmar Eggemann wrote:
> On 7/26/19 2:30 PM, luca abeni wrote:
>> Hi,
>>
>> On Fri, 26 Jul 2019 09:27:52 +0100
>> Dietmar Eggemann <dietmar.eggemann@arm.com> wrote:
>> [...]
>>> @@ -2121,17 +2121,13 @@ static int push_dl_task(struct rq *rq)
>>> }
>>>
>>> deactivate_task(rq, next_task, 0);
>>> - sub_running_bw(&next_task->dl, &rq->dl);
>>> - sub_rq_bw(&next_task->dl, &rq->dl);
>>> set_task_cpu(next_task, later_rq->cpu);
>>> - add_rq_bw(&next_task->dl, &later_rq->dl);
>>>
>>> /*
>>> * Update the later_rq clock here, because the clock is used
>>> * by the cpufreq_update_util() inside __add_running_bw().
>>> */
>>> update_rq_clock(later_rq);
>>> - add_running_bw(&next_task->dl, &later_rq->dl);
>>
>> Looking at the code again and thinking a little bit more about this
>> issue, I suspect a similar change is needed in pull_dl_task() too, no?
>
> The code looks the same. Let me try to test it. I will add it in v2 then.

Like you expected, it happens on the pull side as well.

[ 129.813720] --> CPU7: pull_dl_task: p=[thread0-6 1313] dl_bw=125829
src_rq->dl.running_bw=251658 this_rq->dl.running_bw=125829
[ 129.825167] <-- CPU7: pull_dl_task: p=[thread0-6 1313] dl_bw=125829
src_rq->dl.running_bw=0 this_rq->dl.running_bw=377487
...
[ 129.948528] dl_rq->running_bw > old
[ 129.948568] WARNING: CPU: 7 PID: 0 at kernel/sched/deadline.c:117
inactive_task_timer+0x5e8/0x6b8
...
[ 130.077158] inactive_task_timer+0x5e8/0x6b8
[ 130.081427] __hrtimer_run_queues+0x12c/0x398
[ 130.085782] hrtimer_interrupt+0xfc/0x330
...

\
 
 \ /
  Last update: 2019-07-31 12:33    [W:0.058 / U:1.456 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site