2 * Disregards a certain amount of sleep time (sched_latency_ns) and
3 * considers the task to be running during that period. This gives it
4 * a service deficit on wakeup, allowing it to run sooner.
6 SCHED_FEAT(FAIR_SLEEPERS, 1)
9 * Only give sleepers 50% of their service deficit. This allows
10 * them to run sooner, but does not allow tons of sleepers to
11 * rip the spread apart.
13 SCHED_FEAT(GENTLE_FAIR_SLEEPERS, 1)
16 * Place new tasks ahead so that they do not starve already running
19 SCHED_FEAT(START_DEBIT, 1)
22 * Should wakeups try to preempt running tasks.
24 SCHED_FEAT(WAKEUP_PREEMPT, 1)
27 * When converting the wakeup granularity to virtual time, do it such
28 * that heavier tasks preempting a lighter task have an edge.
30 SCHED_FEAT(ASYM_GRAN, 1)
33 * Always wakeup-preempt SYNC wakeups, see SYNC_WAKEUPS.
35 SCHED_FEAT(WAKEUP_SYNC, 0)
38 * Use the SYNC wakeup hint, pipes and the likes use this to indicate
39 * the remote end is likely to consume the data we just wrote, and
40 * therefore has cache benefit from being placed on the same cpu, see
41 * also AFFINE_WAKEUPS.
43 SCHED_FEAT(SYNC_WAKEUPS, 1)
46 * Based on load and program behaviour, see if it makes sense to place
47 * a newly woken task on the same cpu as the task that woke it --
48 * improve cache locality. Typically used with SYNC wakeups as
49 * generated by pipes and the like, see also SYNC_WAKEUPS.
51 SCHED_FEAT(AFFINE_WAKEUPS, 1)
54 * Prefer to schedule the task we woke last (assuming it failed
55 * wakeup-preemption), since its likely going to consume data we
56 * touched, increases cache locality.
58 SCHED_FEAT(NEXT_BUDDY, 0)
61 * Prefer to schedule the task that ran last (when we did
62 * wake-preempt) as that likely will touch the same data, increases
65 SCHED_FEAT(LAST_BUDDY, 1)
68 * Consider buddies to be cache hot, decreases the likelyness of a
69 * cache buddy being migrated away, increases cache locality.
71 SCHED_FEAT(CACHE_HOT_BUDDY, 1)
74 * Use arch dependent cpu power functions
76 SCHED_FEAT(ARCH_POWER, 0)
79 SCHED_FEAT(DOUBLE_TICK, 0)
80 SCHED_FEAT(LB_BIAS, 1)
81 SCHED_FEAT(LB_SHARES_UPDATE, 1)
82 SCHED_FEAT(ASYM_EFF_LOAD, 1)
85 * Spin-wait on mutex acquisition when the mutex owner is running on
86 * another cpu -- assumes that when the owner is running, it will soon
87 * release the lock. Decreases scheduling overhead.
89 SCHED_FEAT(OWNER_SPIN, 1)