Zen-Kernel Discussion/Support Thread - Part 4

[Ant P.]

commit 94400120 - no more random X hangs on my laptop? Yay

[devl547]

cpufreq: interactive: New 'interactive' governor

New interactive governor.

This governor is designed for latency sensitive workloads, UI interaction for
example.

http://github.com/CyanogenMod/cm-kernel/commit/255f13bf41f368aa51638a854ed69cfc60f39120


Maybe include in zen?

[kernelOfTruth]

[devl547]

/*
* drivers/cpufreq/cpufreq_interactive.c
*
* Copyright (C) 2010 Google, Inc.
*
* This software is licensed under the terms of the GNU General Public
* License version 2, as published by the Free Software Foundation, and
* may be copied, distributed, and modified under those terms.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* Author: Mike Chan (mike@android.com)
*
*/

[Ant P.]

My idea of a perfect cpufreq driver would be one like that, but going straight to full speed when there's processes to run, and dropping to minimum speed as soon as they're done.

ondemand has never really made sense to me, surely the kernel knows exactly when work starts and stops so it shouldn't need to poll cpu usage all the time.

[kernelOfTruth]

it fails - and I'm not the only one:

[devl547]

--- a/kernel/sched.c
+++ b/kernel/sched.c
@@ -2910,6 +2910,7 @@ unsigned long nr_running(void)

return sum;
}
+EXPORT_SYMBOL_GPL(nr_running);

and in drivers/cpufreq/cpufreq_interactive.c:185

-cpumask_t tmp_mask = work_cpumask;
+cpumask_t *tmp_mask = &work_cpumask;

[broch]

[kernelOfTruth]

[Colt45]

Decided to give .34-zen1 a shot on my eeepc. Works great! Its too bad it has some sort of issue with my Dell XPS.
_________________
"Dash" - 2x Intel Xeon 5148 2.33GHz, 4GB DDR2-667F, 500GB, (Gentoo amd64; gcc-4.5.4; 3.4.9-gentoo)
"Sarah" - AMD Athlon XP-M "3500+", 1GB DDR-400, 80GB + 40GB, 2x Intel PRO/100, (Gentoo Hardened x86; gcc-4.5.4; 3.5.4-hardened-r1)

[kernelOfTruth]

seems like it doesn't work that well yet:

[devl547]

Hmm... Use it for 4 days - no problems =\

[kernelOfTruth]

*bump* to final 2.6.35 please

many thanks in advance !
_________________
Unofficial minimal livecd x86/amd64 w/reiser4+truecrypt (by Neo2)
2.6.37.2_plus_v1: BFS, CFS,THP,compaction, zcache or TOI
Hardcore Linux user since 2004

[rahulthewall]

[Ant P.]

Hopefully by then the ACPI poweroff/suspend bugs will all be gone too.

[kernelOfTruth]

[kernelOfTruth]

sorry to report, but there's a BUG in BFQ:

[kernelOfTruth]

check that patch out:
http://forums.gentoo.org/viewtopic-t-793263-start-50-postdays-0-postorder-asc-highlight-.html
_________________
Unofficial minimal livecd x86/amd64 w/reiser4+truecrypt (by Neo2)
2.6.37.2_plus_v1: BFS, CFS,THP,compaction, zcache or TOI
Hardcore Linux user since 2004

[kernelOfTruth]

now those are already 7:
http://lkml.org/lkml/2010/8/5/38

I noticed some latencies and jerky behavior in a non-composited environment, e.g. stuttering for several seconds of webradio (in firefox) while preparing a emerge-queue of the xorg-server

so please also apply the additional ones

thanks !
_________________
Unofficial minimal livecd x86/amd64 w/reiser4+truecrypt (by Neo2)
2.6.37.2_plus_v1: BFS, CFS,THP,compaction, zcache or TOI
Hardcore Linux user since 2004

[wrc1944]

Anyone taken a look at this (brand new slub)?
http://lwn.net/Articles/398650/ Here's a brief quote:

[kernelOfTruth]

this definitely would be interesting

zen has SLQB but that (afaik) is less efficient in memory usage but faster than slub & slab

so a slab like "SLUB: The Unified slab allocator (V3)" would be very welcome: just a little less efficient than slub but much better than it (especially on NUMA)
_________________
Unofficial minimal livecd x86/amd64 w/reiser4+truecrypt (by Neo2)
2.6.37.2_plus_v1: BFS, CFS,THP,compaction, zcache or TOI
Hardcore Linux user since 2004

[wrc1944]

Patch request : Would the zen devs please consider including the 2900_xconfig-with-qt4.patch for 2.6.35 zen? http://sourcestest.gentoo.org/cgi-bin/viewvc.cgi/linux-patches/genpatches-2.6/trunk/2.6.35/

It would be very convenient for those preferring make xconfig, and not wishing to deal with running the kde-sunset (qt3) overlay. It's very small.

BTW, bfs-323 for 2.6.35 is out. I just gave it a try (also applied the xconfig qt4 patch), and it's working well. I usually do a vanilla kernel with bfs with each major kernel release, before ck1 or zen1 is released.
_________________
Main box- ASRock 880GM-LE AM3
Phenom II x6 1090T, 3.2 GHz, 8GB GSkill DDR3 1333mhz
Samsung SATA 500GB, Radeon HD 6570 2GB DDR3
Gentoo ~x86, ~amd64, glibc-2.17, gcc-4.7.2, kernel 3.8.11, 3.9.0

[Gef]

zen.git (ec6ae446a5b9c82c6efff327716f4b66f9b07be1)

Compilation dies with:

[bollucks]

2.6.35-ck1:
http://lkml.org/lkml/2010/8/6/627

[kernelOfTruth]