Eric Lee / linux-smarc-t335x-v3.2

Download as
Browse Code ยป

Commit f8cbd99bd3a023db8d6356d19a5f6f539d367327

Authored by Ingo Molnar
Committed by Linus Torvalds
1 parent f704f56af9
Exists in master and in 4 other branches v3.2_AM335xPSP_04.06.00.11, v3.2_NEWT335xPSP_04.06.00.11, v3.2_SBC_SMARTMEN, v3.2_SMARCT335xPSP_04.06.00.11

[PATCH] sched: voluntary kernel preemption 

This patch adds a new preemption model: 'Voluntary Kernel Preemption'.  The
3 models can be selected from a new menu:

            (X) No Forced Preemption (Server)
            ( ) Voluntary Kernel Preemption (Desktop)
            ( ) Preemptible Kernel (Low-Latency Desktop)

we still default to the stock (Server) preemption model.

Voluntary preemption works by adding a cond_resched()
(reschedule-if-needed) call to every might_sleep() check.  It is lighter
than CONFIG_PREEMPT - at the cost of not having as tight latencies.  It
represents a different latency/complexity/overhead tradeoff.

It has no runtime impact at all if disabled.  Here are size stats that show
how the various preemption models impact the kernel's size:

    text    data     bss     dec     hex filename
 3618774  547184  179896 4345854  424ffe vmlinux.stock
 3626406  547184  179896 4353486  426dce vmlinux.voluntary   +0.2%
 3748414  548640  179896 4476950  445016 vmlinux.preempt     +3.5%

voluntary-preempt is +0.2% of .text, preempt is +3.5%.

This feature has been tested for many months by lots of people (and it's
also included in the RHEL4 distribution and earlier variants were in Fedora
as well), and it's intended for users and distributions who dont want to
use full-blown CONFIG_PREEMPT for one reason or another.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Showing 2 changed files with 62 additions and 13 deletions Side-by-side Diff

include/linux/kernel.h
Diff comments   View file @ f8cbd99
... ... @@ -58,14 +58,22 @@
58 58 * be biten later when the calling function happens to sleep when it is not
59 59 * supposed to.
60 60 */
  61 +#ifdef CONFIG_PREEMPT_VOLUNTARY
  62 +extern int cond_resched(void);
  63 +# define might_resched() cond_resched()
  64 +#else
  65 +# define might_resched() do { } while (0)
  66 +#endif
  67 +
61 68 #ifdef CONFIG_DEBUG_SPINLOCK_SLEEP
62   -#define might_sleep() __might_sleep(__FILE__, __LINE__)
63   -#define might_sleep_if(cond) do { if (unlikely(cond)) might_sleep(); } while (0)
64   -void __might_sleep(char *file, int line);
  69 + void __might_sleep(char *file, int line);
  70 +# define might_sleep() \
  71 + do { __might_sleep(__FILE__, __LINE__); might_resched(); } while (0)
65 72 #else
66   -#define might_sleep() do {} while(0)
67   -#define might_sleep_if(cond) do {} while (0)
  73 +# define might_sleep() do { might_resched(); } while (0)
68 74 #endif
  75 +
  76 +#define might_sleep_if(cond) do { if (unlikely(cond)) might_sleep(); } while (0)
69 77  
70 78 #define abs(x) ({ \
71 79 int __x = (x); \
kernel/Kconfig.preempt
Diff comments   View file @ f8cbd99
1 1  
2   -config PREEMPT
3   - bool "Preemptible Kernel"
  2 +choice
  3 + prompt "Preemption Model"
  4 + default PREEMPT_NONE
  5 +
  6 +config PREEMPT_NONE
  7 + bool "No Forced Preemption (Server)"
4 8 help
5   - This option reduces the latency of the kernel when reacting to
6   - real-time or interactive events by allowing a low priority process to
7   - be preempted even if it is in kernel mode executing a system call.
8   - This allows applications to run more reliably even when the system is
  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 +
  19 +config 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 slighly 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
9 32 under load.
10 33  
11   - Say Y here if you are building a kernel for a desktop, embedded
12   - or real-time system. Say N if you are unsure.
  34 + Select this if you are building a kernel for a desktop system.
  35 +
  36 +config PREEMPT
  37 + bool "Preemptible Kernel (Low-Latency Desktop)"
  38 + help
  39 + This option reduces the latency of the kernel by making
  40 + all kernel code (that is not executing in a critical section)
  41 + preemptible. This allows reaction to interactive events by
  42 + permitting a low priority process to be preempted involuntarily
  43 + even if it is in kernel mode executing a system call and would
  44 + otherwise not be about to reach a natural preemption point.
  45 + This allows applications to run more 'smoothly' even when the
  46 + system is under load, at the cost of slighly lower throughput
  47 + and a slight runtime overhead to kernel code.
  48 +
  49 + Select this if you are building a kernel for a desktop or
  50 + embedded system with latency requirements in the milliseconds
  51 + range.
  52 +
  53 +endchoice
13 54  
14 55 config PREEMPT_BKL
15 56 bool "Preempt The Big Kernel Lock"