2 * Only give sleepers 50% of their service deficit. This allows
3 * them to run sooner, but does not allow tons of sleepers to
4 * rip the spread apart.
6 SCHED_FEAT(GENTLE_FAIR_SLEEPERS, 1)
9 * Place new tasks ahead so that they do not starve already running
12 SCHED_FEAT(START_DEBIT, 1)
15 * Should wakeups try to preempt running tasks.
17 SCHED_FEAT(WAKEUP_PREEMPT, 1)
20 * Based on load and program behaviour, see if it makes sense to place
21 * a newly woken task on the same cpu as the task that woke it --
22 * improve cache locality. Typically used with SYNC wakeups as
23 * generated by pipes and the like, see also SYNC_WAKEUPS.
25 SCHED_FEAT(AFFINE_WAKEUPS, 1)
28 * Prefer to schedule the task we woke last (assuming it failed
29 * wakeup-preemption), since its likely going to consume data we
30 * touched, increases cache locality.
32 SCHED_FEAT(NEXT_BUDDY, 0)
35 * Prefer to schedule the task that ran last (when we did
36 * wake-preempt) as that likely will touch the same data, increases
39 SCHED_FEAT(LAST_BUDDY, 1)
42 * Consider buddies to be cache hot, decreases the likelyness of a
43 * cache buddy being migrated away, increases cache locality.
45 SCHED_FEAT(CACHE_HOT_BUDDY, 1)
48 * Use arch dependent cpu power functions
50 SCHED_FEAT(ARCH_POWER, 0)
53 SCHED_FEAT(DOUBLE_TICK, 0)
54 SCHED_FEAT(LB_BIAS, 1)
55 SCHED_FEAT(LB_SHARES_UPDATE, 1)
56 SCHED_FEAT(ASYM_EFF_LOAD, 1)
59 * Spin-wait on mutex acquisition when the mutex owner is running on
60 * another cpu -- assumes that when the owner is running, it will soon
61 * release the lock. Decreases scheduling overhead.
63 SCHED_FEAT(OWNER_SPIN, 1)
66 * Decrement CPU power based on irq activity
68 SCHED_FEAT(NONIRQ_POWER, 1)