On 08/11/2016 09:30, Noam Camus wrote:

quoted

From: Daniel Lezcano [mailto:daniel.lezcano@linaro.org] 
Sent: Tuesday, November 1, 2016 10:02 PM

...

quoted

Assuming cpu0 and cpu1 are sibling, does

quoted

taskset 0x1 time sleep 2 & taskset 0x2 time sleep 3

I will use 16,17 instead of 0,1

quoted

give a correct result without a dmesg log ?

[root@192.168.8.2 /]$ [root@192.168.8.2 /]$ taskset 65536 time sleep 2 & taskset 131072 time sleep 3

Thanks for providing the numbers.

So there is no weird messages in dmesg ?

real    0m 2.54s
user    0m 0.04s
sys     0m 0.14s
real    0m 3.47s
user    0m 0.00s
sys     0m 0.15s
[1]+  Done                       taskset 65536 time sleep 2

Seem OK to me.

I'm not sure of that.

2.54 instead of 2
3.47 instead of 3

0.02 time drift could be acceptable but 0.54 or 0.47 is too much.

And 0.15 in sys also a big number AFAICT.

Is the system in NO_HZ ?



-- 
 <http://www.linaro.org/> Linaro.org │ Open source software for ARM SoCs

Follow Linaro:  <http://www.facebook.com/pages/Linaro> Facebook |
<http://twitter.com/#!/linaroorg> Twitter |
<http://www.linaro.org/linaro-blog/> Blog