By popular demand, here is release -v24 of the CFS scheduler patch.
It is a full backport of the latest & greatest scheduler code to

v2.6.24-rc3, v2.6.23.8, v2.6.22.13, v2.6.21.7. The patches can be

downloaded from the usual place:
    http://people.redhat.com/mingo/cfs-scheduler/
There's tons of changes since v22 was released:
    36 files changed, 2359 insertions(+), 1082 deletions(-)
that's 187 individual commits from 32 authors.
So even if CFS v22 worked well for you, please try this release too and

report regressions (if any).
There are countless improvements in -v24 (see the shortlog further below

for details), but here are a few highlights:
 - improved interactivity via Peter Ziljstra's "virtual slices" feature.

   As load increases, the scheduler shortens the virtual timeslices that

   tasks get, so that applications observe the same constant latency for

   getting on the CPU. (This goes on until the slices reach a minimum

   granularity value)
 - CONFIG_FAIR_USER_SCHED is now available across all backported

   kernels and the per user weights are configurable via

   /sys/kernel/uids/. Group scheduling got refined all around.
 - performance improvements
 - bugfixes
99% of these changes are already upstream in Linus's git tree and they

will be released as part of v2.6.24. (there are 4 pending commits that

are in the small 2.6.24-rc3-v24 patch.)
As usual, any sort of feedback, bugreport, fix and suggestion is more

than welcome!
	Ingo
------------------>

Adrian Bunk (3):

      sched: make kernel/sched.c:account_guest_time() static

      sched: proper prototype for kernel/sched.c:migration_init()

      sched: make sched_nr_latency static
Alexey Dobriyan (1):

      sched: uninline scheduler
Andi Kleen (5):

      sched: cleanup: remove unnecessary gotos

      sched: cleanup: refactor common code of sleep_on / wait_for_completion

      sched: cleanup: refactor normalize_rt_tasks

      sched: remove stale comment from sc...
[ message continues ]

update: there's now a CFS-v24 patch available for v2.6.23.9 as well.
	Ingo

-

Hi.

First, thanks for this patch. I don't have numbers but working with my

laptop feels much better.

Just a question: does the patch include the fix (divide by zero....)

you just posted in the stable review for 2.6.23.9?
Thanks and regards,

Fabio
-

yeah.
	Ingo

-

Hi
OK, thanks. Have you already tried to apply the patch to 2.6.23.9?
Regards,

Fabio

-

are you sure? The last -ck patch i can find is for .22:
  http://www.kernel.org/pub/linux/kernel/people/ck/patches/2.6/
or have they been forward ported? (if yes, do you have an URL for that)
(my guess is you used the 2.6.23.1 scheduler (CFS), so the improvement

you felt on the laptop is relative to 2.6.23-vanilla in essence - which 
hm:
The latest stable version of the Linux kernel is:           2.6.23.8
so there's no 23.9 yet.
	Ingo

-

Hi
On Nov 20, 2007 10:53 PM, Ingo Molnar <mingo@elte.hu> wrote:
Well, it seems that Matthew has been faster than me :-) Yes, I got -ck
I meant the 23.9 patches that Greg KH posted some hours ago for

review. I guess I will find myself in a couple of days.
To answer your latest mail, again I don't have numbers but from my

point ov view:
23.1 < 23.1+ck < 23.8+cfs.24
where 23.1+ck is slightly better than vanilla 23.1 and 23.8+cfs.24 is

much better than 23.1+ck.
My workload is pretty standard, F8+kde+firefox+kmail+amsn, some kernel

compile and so on.
Regards,

Fabio

-

do you remember which particular workload improved the most, and in

roughly whatway? (no need for numbers - subjective impressions are

accurate enough for this)
	Ingo

-

Hi
Firefox, especially when multiple tabs are open. Sometimes I used to
Regards,

Fabio

-

Hi Ingo,
Tested latest v24 patch against 686 & x86_64 and it runs smoothly, at

least as well as previous v22 release.
As always, Thnx very much for this backport.
- vin
-

Hello,
Testing sched-cfs-v2.6.24-rc3-v24.patch on top of 2.6.24-rc3-git1

(ignored the two "already applied" messages coming from git1 commits),

I get a 1.00 minimum load in top, coming from the load_balance_mo thread

staying in D-state. I get this on 2 different computers with similar

configs, so I am attaching one of them here.
--

Damien Wyart

does the patch below fix that problem for you?
	Ingo
Index: linux/kernel/sched.c

===================================================================

--- linux.orig/kernel/sched.c

+++ linux/kernel/sched.c

@@ -7138,7 +7138,7 @@ static int load_balance_monitor(void *un

 		else if (timeout < sysctl_sched_max_bal_int_shares)

 			timeout *= 2;
-		msleep(timeout);

+		msleep_interruptible(timeout);

 	}
 	return 0;

-

Yes, it does. Thanks for your answer.
-

Ingo,

	I am reworking the fair-group load balance patches. Can you drop

what you have until I resubmit?
--

Regards,

vatsa

-

Doesn't compile with 2.6.21.7:
kernel/sched.c: In function ‘cpu_to_core_group’:

kernel/sched.c:6056: error: ‘per_cpu__cpu_sibling_map’ undeclared (first use in this function)

kernel/sched.c:6056: error: (Each undeclared identifier is reported only once

kernel/sched.c:6056: error: for each function it appears in.)

kernel/sched.c:6056: warning: type defaults to ‘int’ in declaration of ‘type name’

kernel/sched.c:6056: error: invalid type argument of ‘unary *’

kernel/sched.c: In function ‘build_sched_domains’:

kernel/sched.c:6319: error: ‘per_cpu__cpu_sibling_map’ undeclared (first use in this function)

kernel/sched.c:6319: warning: type defaults to ‘int’ in declaration of ‘type name’

kernel/sched.c:6319: error: invalid type argument of ‘unary *’

kernel/sched.c:6330: warning: type defaults to ‘int’ in declaration of ‘type name’

kernel/sched.c:6330: error: invalid type argument of ‘unary *’

make[1]: *** [kernel/sched.o] Error 1

make[1]: *** Waiting for unfinished jobs....
 -Erik
--

Erik B. Andersen             http://codepoet-consulting.com/

--This message was written using 73% post-consumer electrons--

-

does the patch below help?
	Ingo
Index: linux-cfs-2.6.21.7.q/include/linux/sched.h

===================================================================

--- linux-cfs-2.6.21.7.q.orig/include/linux/sched.h

+++ linux-cfs-2.6.21.7.q/include/linux/sched.h

@@ -10,7 +10,7 @@

 #define COMPAT_REGISTER_SYSCTL
 /* backporting helper macro: */

-#define cpu_sibling_map(cpu) per_cpu(cpu_sibling_map, cpu)

+#define cpu_sibling_map(cpu) cpu_sibling_map[cpu]
 /*

  * cloning flags:
-

Looks good.  Compiling.
- vin

-

Yes, compiles and boots with that change,
 -Erik
--

Erik B. Andersen             http://codepoet-consulting.com/

--This message was written using 73% post-consumer electrons--

-

Hi Ingo,
Thnx a lot for theses backports...
Ran into this while compiling a 2.6.22.13 with CFS v24
  CC      kernel/sched.o

kernel/sched.c: In function 'cpu_to_core_group':

kernel/sched.c:6056: error: 'per_cpu__cpu_sibling_map' undeclared (first

use in this function)

kernel/sched.c:6056: error: (Each undeclared identifier is reported only

once

kernel/sched.c:6056: error: for each function it appears in.)

kernel/sched.c:6056: warning: type defaults to 'int' in declaration of

'type name'

kernel/sched.c:6056: error: invalid type argument of 'unary *'

kernel/sched.c: In function 'build_sched_domains':

kernel/sched.c:6319: error: 'per_cpu__cpu_sibling_map' undeclared (first

use in this function)

kernel/sched.c:6319: warning: type defaults to 'int' in declaration of

'type name'

kernel/sched.c:6319: error: invalid type argument of 'unary *'

kernel/sched.c:6330: warning: type defaults to 'int' in declaration of

'type name'

kernel/sched.c:6330: error: invalid type argument of 'unary *'

make[2]: *** [kernel/sched.o] Error 1

make[1]: *** [kernel] Error 2

make[1]: Leaving directory

`/usr/src/linux-headers-2.6.22.13-cfs-sarge-686-envcan'

make: *** [debian/stamp-build-kernel] Error 2
- vin

-

find below the current set of fixlets - it should fix the

cpu_sibling_map build error too. (or re-download the v24 patch again to

get the fixes)
	Ingo
--- linux.orig/kernel/sysctl.c

+++ linux/kernel/sysctl.c

@@ -309,6 +309,7 @@ static struct ctl_table kern_table[] = {

 		.mode		= 644,

 		.proc_handler	= &proc_dointvec,

 	},

+#ifdef CONFIG_FAIR_GROUP_SCHED

 	{

 		.ctl_name	= CTL_UNNUMBERED,

 		.procname	= "sched_min_bal_int_shares",

@@ -326,6 +327,7 @@ static struct ctl_table kern_table[] = {

 		.proc_handler	= &proc_dointvec,

 	},

 #endif

+#endif

 	{

 		.ctl_name	= CTL_UNNUMBERED,

 		.procname	= "sched_compat_yield",

--- linux.orig/include/linux/sched.h

+++ linux/include/linux/sched.h

@@ -10,7 +10,7 @@

 #define COMPAT_REGISTER_SYSCTL
 /* backporting helper macro: */

-#define cpu_sibling_map(cpu) per_cpu(cpu_sibling_map, cpu)

+#define cpu_sibling_map(cpu) cpu_sibling_map[cpu]
 /*

  * cloning flags:

-