[playground & fun] 2.6.12-rc5-vivid2

[fallow]

--- 2.6.12-rc5-vivid2 info ---

this is patchset for 2.6.12-rc5-mm2.
aplly it on top of mm2.
its just like CK for mm + vesa-tng and fbsplash.

and ps. this is just for fun and its a kind of playground , and also to see & feel new mm features but with the Con's Staircase CPU Scheduler policy(ies)

whats in it?

Code: Select all

1gb_lowmem.diff (rc5ck2) 
sched-run_normal_with_rt_on_sibling.diff (rc5ck2)
2.6.12-rc5_to_staircase11.2.diff (rc5ck2)
s11.2_test1.diff (rc5ck2-testpatch)
320-irqbal_fast.txt - http://www.kernel.org/pub/linux/kernel/people/mbligh/
def_iosched_select_menu.diff - my trivial def. io scheduler selection via Kconfig
fbsplash-0.9.2-r3-2.6.12-rc3.patch (Spock)
inotify-44-update-2.patch - reversed to avoid remoun/umount oopsing.
isobatch_ionice2.diff (rc5ck2) 
mapped_watermark3.diff (rc5ck2)
rt_ionice.diff (rc5ck2) 
schedbatch2.7.diff (rc5ck2)
schediso2.12.diff (rc5ck2)
schedrange.diff (rc5ck2) 
smp-nice-support3.diff (rc5ck2)
mm-IKtimer-set-again-to-1000.diff - in mm def is now 250, again to 1000 for destkops.
usbhid-readd-kconfig-r1.patch (Krejler)
vesafb-tng-0.9-rc6-r2-2.6.12-rc1.patch (Spock) 

dont forget to install schedtool to using sched_iso and sched_batch staircase policies

NO ANY ATI / NVIDIA fixes here , its working with my NVIDIA card and drivers from portage with AGPGART
Possibly fixing should be on driver side, not kernel space.


Everybody know that mm is just exactly a kind of playground for 2.6 so by this it can be potentially bugged. If You have some problem of course You can post here , but if You have errors and it seems that its mm fault, test pure mm then , and then if error still happens , report it to LKML.

http://vivid.dat.pl/fallow/pub/index.ph ... c5-vivid2/

http://vivid.dat.pl/fallow/pub/vivid-so ... -r2.ebuild
patch : http://vivid.dat.pl/fallow/pub/index.ph ... 2.diff.bz2
info http://vivid.dat.pl/fallow/pub/index.ph ... =notes.txt

[vipernicus]

finally, been waiting patiently for this one.

Which patches from 12-rc5-mm2 did you have to reverse to apply staircase and policies?

[vipernicus]

rc5-ck2 is out already. looks like there is a 11.2-test.diff, which adds this:

+ * Tasks that have marked their sleep as noninteractive get
+ * woken up without their sleep counting.

[fallow]

rc5-ck2 is out already. looks like there is a 11.2-test.diff, which adds this:

+ * Tasks that have marked their sleep as noninteractive get
+ * woken up without their sleep counting.

updated to ck2 .

cheers.

[bzcat_-|patch-p1]

Hiya fallow looks like a pretty interesting patchset. Good job.

[pwe]

hey fallow!
on x86 working fine, but on x86-64 i cant run it on my nvidia

Code: Select all

*  media-video/nvidia-kernel
      Latest version available: 1.0.7174
      Latest version installed: 1.0.7174

pci-e card. maybe mm2+pci-e + x86-64 bug ?

[vipernicus]

This patchset really does work well. NOt sure if you saw my previous question:
Which patches did you have to reverse in -mm to apply staircase?

[pwe]

new nvidia-kernel in portage, vivid2 working on gentoo64. im curios how its works, because its my "first time" with vivid

[Evil Dark Archon]

This patchset really does work well. NOt sure if you saw my previous question:
Which patches did you have to reverse in -mm to apply staircase?

Code: Select all

#
# CPU scheduler
#
sched2-cleanup-wake_idle.patch
sched2-improve-load-balancing-pinned-tasks.patch
sched2-reduce-active-load-balancing.patch
sched2-fix-smt-scheduling-problems.patch
sched2-fix-smt-scheduling-problems-fix.patch
sched2-add-debugging.patch
sched2-less-aggressive-idle-balancing.patch
sched2-balance-timers.patch
sched2-tweak-affine-wakeups.patch
sched2-no-aggressive-idle-balancing.patch
sched2-balance-on-fork.patch
sched2-schedstats-update-for-balance-on-fork.patch
sched2-sched-tuning.patch
sched2-sched-tuning-fix.patch
sched2-sched-domain-sysctl.patch
sched-uninline-task_timeslice.patch
sched-cleanup-context-switch-locking.patch
sched-null-domains.patch
sched-remove-degenerate-domains.patch
sched-remove-degenerate-domains-fix.patch
sched-multilevel-sbe-sbf.patch
sched-rcu-domains.patch
sched-rcu-domains-fix.patch
sched-consolidate-sbe-sbf.patch
sched-consolidate-sbe-sbf-fix-2.patch
sched-relax-pinned-balancing.patch
sched-micro-optimize-task-requeueing-in-schedule.patch
sched-changing-rt-priority-without-cap_sys_nice.patch
dynamic-sched-domains-sched-changes.patch
dynamic-sched-domains-sched-changes-fix.patch
dynamic-sched-domains-cpuset-changes.patch
dynamic-sched-domains-ia64-changes.patch
sched-implement-nice-support-across-physical-cpus-on-smp.patch
sched-change_prio_bias_only_if_queued.patch
sched-account_rt_tasks_in_prio_bias.patch
consolidate-preempt-options-into-kernel-kconfigpreempt.patch
enable-preempt_bkl-on-preemptsmp-too.patch
sched-tweak-idle-thread-setup-semantics.patch
sched-voluntary-kernel-preemption.patch
sched-smp-nice-bias-busy-queues-on-idle-rebalance.patch
sched-task_noninteractive.patch

those are the cpu scheduler patches from the -mm series file, reversing (or not applying in the first place if you use quilt) those allows me to apply staircase in my own patchset.

[COiN3D]

Hm fallow, no ext3-low-latency patch inside? I thought you like this patch so much.. or are you including this low-latency patch just in love-sources?

Bye, coin

[vipernicus]

Hm fallow, no ext3-low-latency patch inside? I thought you like this patch so much.. or are you including this low-latency patch just in love-sources?

Bye, coin

Those patches are part of -mm. So they are most likely in there.

[COiN3D]

Hm fallow, no ext3-low-latency patch inside? I thought you like this patch so much.. or are you including this low-latency patch just in love-sources?

Bye, coin

Those patches are part of -mm. So they are most likely in there.

Ok, didn't know that. By the way, can you use splashimages? I activated it in the kernel, and I also installed a theme, and appended the correct kernel-boot-option, but there's no splash image while the kernel is booting.

What about you?

[fallow]

Hiya fallow looks like a pretty interesting patchset. Good job.

just for fun

new nvidia-kernel in portage, vivid2 working on gentoo64. im curios how its works, because its my "first time" with vivid

its just mm + ck

This patchset really does work well. NOt sure if you saw my previous question:
Which patches did you have to reverse in -mm to apply staircase?

Code: Select all

#
# CPU scheduler
#
sched2-cleanup-wake_idle.patch
sched2-improve-load-balancing-pinned-tasks.patch
sched2-reduce-active-load-balancing.patch
sched2-fix-smt-scheduling-problems.patch
sched2-fix-smt-scheduling-problems-fix.patch
sched2-add-debugging.patch
sched2-less-aggressive-idle-balancing.patch
sched2-balance-timers.patch
sched2-tweak-affine-wakeups.patch
sched2-no-aggressive-idle-balancing.patch
sched2-balance-on-fork.patch
sched2-schedstats-update-for-balance-on-fork.patch
sched2-sched-tuning.patch
sched2-sched-tuning-fix.patch
sched2-sched-domain-sysctl.patch
sched-uninline-task_timeslice.patch
sched-cleanup-context-switch-locking.patch
sched-null-domains.patch
sched-remove-degenerate-domains.patch
sched-remove-degenerate-domains-fix.patch
sched-multilevel-sbe-sbf.patch
sched-rcu-domains.patch
sched-rcu-domains-fix.patch
sched-consolidate-sbe-sbf.patch
sched-consolidate-sbe-sbf-fix-2.patch
sched-relax-pinned-balancing.patch
sched-micro-optimize-task-requeueing-in-schedule.patch
sched-changing-rt-priority-without-cap_sys_nice.patch
dynamic-sched-domains-sched-changes.patch
dynamic-sched-domains-sched-changes-fix.patch
dynamic-sched-domains-cpuset-changes.patch
dynamic-sched-domains-ia64-changes.patch
sched-implement-nice-support-across-physical-cpus-on-smp.patch
sched-change_prio_bias_only_if_queued.patch
sched-account_rt_tasks_in_prio_bias.patch
consolidate-preempt-options-into-kernel-kconfigpreempt.patch
enable-preempt_bkl-on-preemptsmp-too.patch
sched-tweak-idle-thread-setup-semantics.patch
sched-voluntary-kernel-preemption.patch
sched-smp-nice-bias-busy-queues-on-idle-rebalance.patch
sched-task_noninteractive.patch

those are the cpu scheduler patches from the -mm series file, reversing (or not applying in the first place if you use quilt) those allows me to apply staircase in my own patchset.

exactly

cheers.

[maxpayne]

hi - i've been trying a number of kernels, but so far none run both the areca raid drivers (they're in the latest mm) and the adaptec i2o (dpt_i2o) drivers.

do you think you could include these two?

btw, latest kernel with mm patch works fine with the areca adapter, but doesn't want to work with the adaptec i2o on-board adapter - on the other side, gentoo live cd 2005.0 runs perfectly the adaptec i2o adapter, though no areca adapter in sight... heeeelp lol

well, i thought as a kernel devel you might be able to include these from their respective sources? that'd be awesome!

[grzewho]

great sources, made me switch from nitro to vivid. thx fallow !

[fallow]

IMHO not. these things are just personal. vivid is just like mm+ck and splash/vesatng. nothing more

great sources, made me switch from nitro to vivid. thx fallow !

nice , I have intend to releasing vivid if mm will be not so bugged

cheers

[vipernicus]

I have actually seen a slight performance increase with 2.6.12-rc6+ck1+mm1. Not sure if it is -mm1 or staircase 11.3, but I feel it. Dynamic Ticking is also working well for me, my computer was always loud during idle, but with dynamic ticking enabled, it is not. What utilities can I use to generate the latency and performance numbers that I have seen fallow post in the past?