Messages in this thread | | | Date | Fri, 07 Mar 2014 19:32:25 +0800 | From | Jet Chen <> | Subject | Re: [sched/idle] c365c292d05: ltp.sched_rr_get_interval02.1.TFAIL |
| |
Hi Peter,
> Does: http://lkml.kernel.org/r/20140301191838.d15d03112b2598a671dac22c@gmail.com > > Fix it?
Both patches can fix the LTP regression.
We applied each of Juri's and Kirill's patches on top of Thomas' commit, and here are the comparisons:
test case: ltp/syscalls
Thomas' Juri's c365c292d05908c ffc78ca31d51f90c516bb50cf --------------- ------------------------- 1 ~ 0% -100.0% 0 ~ 0% TOTAL ltp.sched_rr_get_interval02.1.TFAIL 73790 ~ 5% -14.1% 63363 ~ 6% TOTAL interrupts.IWI 3040 ~ 4% -6.7% 2836 ~ 5% TOTAL slabinfo.kmalloc-128.num_objs
Thomas' Kirill's c365c292d05908c f4262311faf6f326bf27fb1f4 --------------- ------------------------- 1 ~ 0% -100.0% 0 ~ 0% TOTAL ltp.sched_rr_get_interval02.1.TFAIL 3618 ~ 6% -13.2% 3141 ~ 6% TOTAL slabinfo.anon_vma.active_objs 3671 ~ 5% -11.3% 3258 ~ 4% TOTAL slabinfo.anon_vma.num_objs 8.41 ~ 2% -5.1% 7.98 ~ 3% TOTAL boottime.dhcp 15.90 ~ 2% -3.2% 15.40 ~ 2% TOTAL boottime.boot 55.18 ~ 1% -2.6% 53.73 ~ 1% TOTAL boottime.idle
Thanks, -Jet
| |