Skip to content

Cure faux idle wreckage

Peter Zijlstra linux-pm lkml 1 version In review

Get series with b4
b4 am [email protected]

Versions newest first

  1. v1 Latest8 patches 5 repliesIn review

Patches in v1 No applies recorded yet

#Subject and filesSizeRepliesApplied
1/8x86, acpi, idle: Restructure the mwait idle routinesarch/x86/include/asm/mwait.h arch/x86/include/asm/processor.h arch/x86/kernel/acpi/cstate.c +4 more+43 -750 replies
2/8sched, preempt: Fixup missed PREEMPT_NEED_RESCHED foldingarch/x86/include/asm/mwait.h include/linux/preempt.h include/linux/sched.h +2 more+42 -100 replies
3/8idle, thermal, acpi: Remove home grown idle implementationsdrivers/acpi/acpi_pad.c drivers/thermal/intel_powerclamp.c include/linux/cpu.h +4 more+91 -1270 replies
4/8preempt, locking: Rework local_bh_{dis,en}able()include/linux/bottom_half.h include/linux/hardirq.h include/linux/preempt_mask.h +1 more+36 -330 replies
5/8locking: Optimize lock_bh functionsinclude/linux/preempt_mask.h include/linux/rwlock_api_smp.h include/linux/spinlock_api_smp.h +1 more+34 -210 replies
6/8sched, net: Clean up preempt_enable_no_resched() abusenet/ipv4/tcp.c+2 -20 replies
7/8sched, net: Fixup busy_loop_us_clock()include/net/busy_poll.h+1 -184 replies
8/8preempt: Take away preempt_enable_no_resched() from modulesinclude/linux/preempt.h include/linux/uaccess.h+24 -30 replies

Reviews trailers per patch

6 of 8 patches have a Reviewed-by or Acked-by.

PatchRafael J. WysockiThomas Gleixner
1/8A-b
2/8
3/8
4/8R-b
5/8R-b
6/8R-b
7/8R-b
8/8R-b

Trailers given in replies are filled; those carried in the patches from earlier versions are outlined with dashes.

Read the v1 thread

Cover letter [PATCH 0/8] Cure faux idle wreckage

Respin of the earlier series that tries to cure the 2 idle injection drivers and cleans up some of the preempt_enable_no_resched() mess.

The intel_powerclamp driver is tested by Jacob Pan and needs one more patch to cpuidle to work as before. I'll let him provide this patch; since he actually has it and tested it.

Jacob also said he'll try and work with the QoS people to sort out the conflict of interest between the idle injectors and the QoS framework.

Can someone please test acpi_pad? Rafael, since the original author seems MIA and you're the over-all ACPI maintainer, can you appoint a person who knows what he's doing? Alternatively, Jacob would you be willing to have a look at that thing? Better still rm drivers/acpi/acpi_pad.c ?

Thomas, can you pick this series up and merge it into -tip provided acpi_pad works?

---
Changes since the earlier version:

- fixed a few build issues; thanks Jacob for spotting them
- Added PF_IDLE so that is_idle_task() can work for the faux idle tasks, which in turn is require for RCU-idle support.
- added an rcu_sleep_check() to play_idle() to ensure we don't try and play idle while holding rcu_read_lock(), which would counter the previous point.
- changed the net busy_poll over to local_clock().