Messages in this thread | | | From | Barry Song <> | Date | Thu, 25 Nov 2021 09:49:19 +1300 | Subject | Re: [PATCH v2] sched/fair: Remove the cost of a redundant cpumask_next_wrap in select_idle_cpu |
| |
On Thu, Nov 25, 2021 at 4:02 AM Peter Zijlstra <peterz@infradead.org> wrote: > > On Thu, Nov 25, 2021 at 01:02:00AM +1300, Barry Song wrote: > > On Thu, Nov 25, 2021 at 12:57 AM Barry Song <21cnbao@gmail.com> wrote: > > > > Let me make it clearer. if nr=5, the original code will loop 5 times, > > > but in the 5th loop, it returns directly, so __select_idle_cpu is > > > only done 4 times. > > > > > > if nr=1, the original code will loop 1 time, but in the 1st loop, > > > it returns directly, so __select_idle_cpu is done 0 times. > > > > this is also why in the first version of patch, i did this: > > span_avg = sd->span_weight * avg_idle; > > if (span_avg > 4*avg_cost) > > - nr = div_u64(span_avg, avg_cost); > > + nr = div_u64(span_avg, avg_cost) - 1; > > else > > - nr = 4; > > + nr = 3; > > > > because we are actually scanning 3 times or div_u64(span_avg, avg_cost) - 1 > > times but not 4 times or div_u64(span_avg, avg_cost) times. > > It still is confusing, because > 4*span -> nr = avg/span, very much > implies we want to bottom out at 4. > > > this is not confusing at all. the only thing which is confusing is the original > > code. > > But yes, it seems a whole lot of confusion stacked together. Let make it > sane and say that we do 'nr' iterations, because clearly that was the > intent :-)
yes. It seems this is much more sensible to do iterations in the number of nr rather than nr-1. we can achieve this goal by two ways:
(1) nr--;
for_each_cpu_wrap(cpu, cpus, target + 1) { _select_idle_cpu().... if (!nr--) return; }
(2) for_each_cpu_wrap(cpu, cpus, target + 1) { _select_idle_cpu().... if (!--nr) return; }
it seems the second way is still better as we don't need the "nr--" before for_each_cpu_wrap() ?
Thanks Barry
| |