Root/kernel/Kconfig.preempt

Source at commit fbf123cd4cc0c097fe9a99c90109ebb2a5e94a50 created 10 years 3 months ago.
By Lars-Peter Clausen, dma: jz4740: Dequeue descriptor from active list before completing it
1
2choice
3    prompt "Preemption Model"
4    default PREEMPT_NONE
5
6config PREEMPT_NONE
7    bool "No Forced Preemption (Server)"
8    help
9      This is the traditional Linux preemption model, geared towards
10      throughput. It will still provide good latencies most of the
11      time, but there are no guarantees and occasional longer delays
12      are possible.
13
14      Select this option if you are building a kernel for a server or
15      scientific/computation system, or if you want to maximize the
16      raw processing power of the kernel, irrespective of scheduling
17      latencies.
18
19config PREEMPT_VOLUNTARY
20    bool "Voluntary Kernel Preemption (Desktop)"
21    help
22      This option reduces the latency of the kernel by adding more
23      "explicit preemption points" to the kernel code. These new
24      preemption points have been selected to reduce the maximum
25      latency of rescheduling, providing faster application reactions,
26      at the cost of slightly lower throughput.
27
28      This allows reaction to interactive events by allowing a
29      low priority process to voluntarily preempt itself even if it
30      is in kernel mode executing a system call. This allows
31      applications to run more 'smoothly' even when the system is
32      under load.
33
34      Select this if you are building a kernel for a desktop system.
35
36config PREEMPT
37    bool "Preemptible Kernel (Low-Latency Desktop)"
38    select PREEMPT_COUNT
39    select UNINLINE_SPIN_UNLOCK if !ARCH_INLINE_SPIN_UNLOCK
40    help
41      This option reduces the latency of the kernel by making
42      all kernel code (that is not executing in a critical section)
43      preemptible. This allows reaction to interactive events by
44      permitting a low priority process to be preempted involuntarily
45      even if it is in kernel mode executing a system call and would
46      otherwise not be about to reach a natural preemption point.
47      This allows applications to run more 'smoothly' even when the
48      system is under load, at the cost of slightly lower throughput
49      and a slight runtime overhead to kernel code.
50
51      Select this if you are building a kernel for a desktop or
52      embedded system with latency requirements in the milliseconds
53      range.
54
55endchoice
56
57config PREEMPT_COUNT
58       bool

Archive Download this file



interactive