Skip to content

Sched load balance backport

[email protected] linux-rt-users lkml 1 version New

Get series with b4
b4 am [email protected]

Versions newest first

  1. v1 Latest17 patches 0 repliesNew

Patches in v1 No applies recorded yet

#Subject and filesSizeRepliesApplied
1/17sched: restore __cpu_power to a straight sum of powerkernel/sched.c+12 -160 replies
2/17sched: SD_PREFER_SIBLINGinclude/linux/sched.h kernel/sched.c+28 -150 replies
3/17sched: update the cpu_power sum during load-balancekernel/sched.c+29 -40 replies
4/17sched: add smt_gaininclude/linux/sched.h include/linux/topology.h kernel/sched.c+9 -10 replies
5/17sched: dynamic cpu_powerkernel/sched.c+35 -30 replies
6/17sched: scale down cpu_power due to RT tasksinclude/linux/sched.h kernel/sched.c kernel/sched_rt.c +1 more+72 -70 replies
7/17sched: try to deal with low capacitykernel/sched.c+29 -60 replies
8/17sched: remove reciprocal for cpu_powerinclude/linux/sched.h kernel/sched.c+36 -740 replies
9/17x86: move APERF/MPERF into a X86_FEATUREarch/x86/include/asm/cpufeature.h arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c arch/x86/kernel/cpu/intel.c+9 -70 replies
10/17x86: Add generic aperf/mperf codearch/x86/include/asm/processor.h arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c+39 -700 replies
11/17Provide an arch specific hook for cpufreq based scaling of cpu_power.kernel/sched.c+19 -20 replies
12/17x86: sched: provide arch implementations using aperf/mperfarch/x86/kernel/cpu/Makefile arch/x86/kernel/cpu/sched.c include/linux/sched.h+63 -10 replies
13/17sched: cleanup wake_idle power savingkernel/sched_fair.c+39 -190 replies
14/17sched: cleanup wake_idlekernel/sched_fair.c+29 -200 replies
15/17sched: Add a missing =kernel/sched.c+1 -10 replies
16/17sched: Deal with low-load in wake_affine()kernel/sched_fair.c+11 -10 replies
17/17sched: Fix dynamic power-balancing crashkernel/sched.c+5 -20 replies

Reviews trailers per patch

No Reviewed-by, Acked-by or Tested-by trailers for v1 yet.

Read the v1 thread

Cover letter [patch -rt 00/17] [patch -rt] Sched load balance backport

Problem: The current -rt, 2.6.31.4-rt14, has several load balancing issues. It can easily be seen with dbench running with the same number of processes as the number of cpus on a SMP system. The dbench threads are running as SCHED_OTHER processes. Several threads end up running on the same cpu even though other cpus are idling. This results in severe throughput regression.

Peter Zijlstra posted several load balance patches meant for mainline a while ago. The following patch series is a backport of the same to 2.6.31.4-rt14. Peter's original patches can be found at
http://marc.info/?l=linux-kernel&m=125198436208787&w=2

Patches 1 through 14 are backports of the original patches except patch 10 which has the original code + bug fixes. Patches 15 - 17 are various relevant fixes that have gone on top of the original patches.

With these patches, load balancing improves considerably on -rt. However it does not completely resolve the problem. This is still under investigation.

This has been stress tested on x86_64 and i686

-Dinakar

--