linux/arch/arm
Nick Piggin 64c7c8f885 [PATCH] sched: resched and cpu_idle rework
Make some changes to the NEED_RESCHED and POLLING_NRFLAG to reduce
confusion, and make their semantics rigid.  Improves efficiency of
resched_task and some cpu_idle routines.

* In resched_task:
- TIF_NEED_RESCHED is only cleared with the task's runqueue lock held,
  and as we hold it during resched_task, then there is no need for an
  atomic test and set there. The only other time this should be set is
  when the task's quantum expires, in the timer interrupt - this is
  protected against because the rq lock is irq-safe.

- If TIF_NEED_RESCHED is set, then we don't need to do anything. It
  won't get unset until the task get's schedule()d off.

- If we are running on the same CPU as the task we resched, then set
  TIF_NEED_RESCHED and no further action is required.

- If we are running on another CPU, and TIF_POLLING_NRFLAG is *not* set
  after TIF_NEED_RESCHED has been set, then we need to send an IPI.

Using these rules, we are able to remove the test and set operation in
resched_task, and make clear the previously vague semantics of
POLLING_NRFLAG.

* In idle routines:
- Enter cpu_idle with preempt disabled. When the need_resched() condition
  becomes true, explicitly call schedule(). This makes things a bit clearer
  (IMO), but haven't updated all architectures yet.

- Many do a test and clear of TIF_NEED_RESCHED for some reason. According
  to the resched_task rules, this isn't needed (and actually breaks the
  assumption that TIF_NEED_RESCHED is only cleared with the runqueue lock
  held). So remove that. Generally one less locked memory op when switching
  to the idle thread.

- Many idle routines clear TIF_POLLING_NRFLAG, and only set it in the inner
  most polling idle loops. The above resched_task semantics allow it to be
  set until before the last time need_resched() is checked before going into
  a halt requiring interrupt wakeup.

  Many idle routines simply never enter such a halt, and so POLLING_NRFLAG
  can be always left set, completely eliminating resched IPIs when rescheduling
  the idle task.

  POLLING_NRFLAG width can be increased, to reduce the chance of resched IPIs.

Signed-off-by: Nick Piggin <npiggin@suse.de>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Con Kolivas <kernel@kolivas.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-11-09 07:56:33 -08:00
..
boot [ARM] 3132/1: S3C2410 - reset on decompression error 2005-11-08 22:43:05 +00:00
common [ARM] 3093/1: SharpSL PCMCIA Updates for Cxx00 models 2005-11-08 19:15:43 +00:00
configs [PATCH] ARM: Reverted 2918/1: [update] Base port of Comdial MP1000 platfrom 2005-11-04 17:28:34 +00:00
kernel [PATCH] sched: resched and cpu_idle rework 2005-11-09 07:56:33 -08:00
lib [ARM] 3094/1: remove PLD stuff from old uaccess code 2005-11-04 17:15:43 +00:00
mach-aaec2000 [PATCH] fix remaining missing includes 2005-11-07 07:53:41 -08:00
mach-clps711x [PATCH] ARM: Reverted 2918/1: [update] Base port of Comdial MP1000 platfrom 2005-11-04 17:28:34 +00:00
mach-clps7500 [ARM] 2994/1: Replace map_desc.physical with map_desc.pfn: CLPS7500 2005-10-28 15:19:05 +01:00
mach-ebsa110 [ARM] Convert EBSA110 network driver to a platform driver 2005-10-31 17:14:57 +00:00
mach-epxa10db [PATCH] fix remaining missing includes 2005-11-07 07:53:41 -08:00
mach-footbridge [ARM] 2998/1: Replace map_desc.physical with map_desc.pfn: Footbridge 2005-10-28 15:19:08 +01:00
mach-h720x Create platform_device.h to contain all the platform device details. 2005-10-29 19:07:23 +01:00
mach-imx Merge master.kernel.org:/home/rmk/linux-2.6-drvmodel 2005-10-31 07:32:56 -08:00
mach-integrator [PATCH] kfree cleanup: arch 2005-11-07 07:54:06 -08:00
mach-iop3xx Merge master.kernel.org:/home/rmk/linux-2.6-drvmodel 2005-10-31 07:32:56 -08:00
mach-ixp4xx [ARM] 3089/1: ixp4xx AHB/PCI endianness fix 2005-11-06 14:34:12 +00:00
mach-ixp2000 [ARM] 3121/1: unconditionally use XCB=101 on ixp2000 2005-11-07 21:12:09 +00:00
mach-l7200 [ARM] 2980/1: Fix L7200 core.c compile 2005-10-14 12:49:15 +01:00
mach-lh7a40x Create platform_device.h to contain all the platform device details. 2005-10-29 19:07:23 +01:00
mach-omap1 [PATCH] changing CONFIG_LOCALVERSION rebuilds too much, for no good reason 2005-11-09 07:55:57 -08:00
mach-pxa [ARM] 3135/1: harden SA11x0 and PXA2xx timer init code 2005-11-08 22:43:06 +00:00
mach-realview [ARM SMP] Fix some sparse warnings in SMP code 2005-11-08 10:44:46 +00:00
mach-rpc [ARM] 2989/1: Replace map_desc.physical with map_desc.pfn: RiscPC 2005-10-28 15:19:02 +01:00
mach-s3c2410 [ARM] 3132/1: S3C2410 - reset on decompression error 2005-11-08 22:43:05 +00:00
mach-sa1100 [ARM] 3135/1: harden SA11x0 and PXA2xx timer init code 2005-11-08 22:43:06 +00:00
mach-shark [ARM] 2991/1: Replace map_desc.physical with map_desc.pfn: Shark 2005-10-28 15:19:03 +01:00
mach-versatile Merge master.kernel.org:/home/rmk/linux-2.6-drvmodel 2005-10-31 07:32:56 -08:00
mm [ARM SMP] Add Realview MPcore SMP support 2005-11-07 21:05:42 +00:00
nwfpe [ARM] 3118/1: fix and reenable nwfpe extended precision emulation for big-endian 2005-11-07 21:12:08 +00:00
oprofile [PATCH] mm: kill check_user_page_readable 2005-10-29 21:40:41 -07:00
plat-omap [PATCH] changing CONFIG_LOCALVERSION rebuilds too much, for no good reason 2005-11-09 07:55:57 -08:00
tools [ARM] Update mach-types 2005-10-10 09:48:10 +01:00
vfp kbuild: arm - use generic asm-offsets.h support 2005-09-09 21:08:59 +02:00
Kconfig [ARM SMP] Add core ARM support for local timers 2005-11-08 19:08:05 +00:00
Kconfig.debug [ARM] Fix typo in arch/arm/Kconfig.debug 2005-09-09 15:57:17 +01:00
Makefile [ARM] 3097/1: change library link ordering 2005-11-04 17:17:30 +00:00