diff options
Diffstat (limited to 'kernel')
-rw-r--r-- | kernel/sched_fair.c | 12 |
1 files changed, 5 insertions, 7 deletions
diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index f819f943fb86..ec1592eb8d08 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c | |||
@@ -25,14 +25,12 @@ | |||
25 | * (default: 20ms, units: nanoseconds) | 25 | * (default: 20ms, units: nanoseconds) |
26 | * | 26 | * |
27 | * NOTE: this latency value is not the same as the concept of | 27 | * NOTE: this latency value is not the same as the concept of |
28 | * 'timeslice length' - timeslices in CFS are of variable length. | 28 | * 'timeslice length' - timeslices in CFS are of variable length |
29 | * (to see the precise effective timeslice length of your workload, | 29 | * and have no persistent notion like in traditional, time-slice |
30 | * run vmstat and monitor the context-switches field) | 30 | * based scheduling concepts. |
31 | * | 31 | * |
32 | * On SMP systems the value of this is multiplied by the log2 of the | 32 | * (to see the precise effective timeslice length of your workload, |
33 | * number of CPUs. (i.e. factor 2x on 2-way systems, 3x on 4-way | 33 | * run vmstat and monitor the context-switches (cs) field) |
34 | * systems, 4x on 8-way systems, 5x on 16-way systems, etc.) | ||
35 | * Targeted preemption latency for CPU-bound tasks: | ||
36 | */ | 34 | */ |
37 | const_debug unsigned int sysctl_sched_latency = 20000000ULL; | 35 | const_debug unsigned int sysctl_sched_latency = 20000000ULL; |
38 | 36 | ||