Commit fd2e3ee3 authored Jul 23, 2018 by Abhijeet Dharmapurikar Committed by Satya Durga Srinivasu Prabhala May 29, 2019

sched/fair: improve big task placement



When a task that should run on the highest cluster is being placed and
if all the cpus in the highest cluster are either not available or
cannot accommodate this task, we could end up placing this task on
the smallest cpu.

Instead place this task on the most spare capacity cpu.

Change-Id: I10256918a7d6f016e7fce31af61c314b8784e22d
Signed-off-by: Abhijeet Dharmapurikar <adharmap@codeaurora.org>

parent f18dbda7

kernel/sched/fair.c

+7 −0

Original line number	Original line	Diff line number	Diff line
	@@ -6949,6 +6949,13 @@ static void find_best_target(struct sched_domain sd, cpumask_t cpus,
	continue;		continue;
	}		}

			/*
			* Skip processing placement further if we are visiting
			* cpus with lower capacity than start cpu
			*/
			if (capacity_orig < capacity_orig_of(start_cpu))
			continue;

	/*		/*
	* Case B) Non latency sensitive tasks on IDLE CPUs.		* Case B) Non latency sensitive tasks on IDLE CPUs.
	*		*