vivid-sources

[fallow]

[EDIT]
now vivid-sources also on sourceforge.net:
http://sourceforge.net/projects/vivid-sources
[/EDIT]



Maybe someone will be interested in trying a Peter Williams runtime cpu scheduler elevation tool called Hydra I`ve add some ohter patches to this also.Hydra is based on others single prior array schedulers like staircase , spa and ebs, and it can be control at runtime via GUI or mode string in /proc.
It`s simply just for fun

list is here :

Code: Select all

2.6.7-spa_hydra_FULL-v1.5 - cpu scheduler with runtime elevation 
from_2.6.7-staircase7.8_to_staircase7.9 from ck,update for hydra
voluntary-preempt-2.6.7-bk20-H3 - additional kernel preemption patch  
 /option in Processor Type & Features-->Voluntary Kernel Preemption/
wli_preempttest -         non-preemptible critical section timing / also in Proc Type & Feat menu / 
reiser4 snapshot from ck 
gcloop-2.6-20040510
9000_SuSE-117-writeback-lat
BadRAM-2.6.5.2
autotune_inactivation 
autotune_swappiness 
bootsplash
config-nr-tty-devices
config_hz 
defaultcfq
linux-2.6.0.dxr3.20031229
lirc-2.6.5-mm1-20040406
lufs-0.9.7-2.6.0-test9
vesafb-rrc-0.1.6
patch-i386-irq_enable_spinlocks2
patch-ool-spinlocks
rsbac-v1.2.3-2.6.7 - compiled but I`m not sure about good works;)
squashfs1.3r3
supermount-ng204 from ck

vesafb_modes.h in /kernel_scr/arch/i386/boot is filled for 1024x768@85Hz for example and CFLAGS in /kernel_src/arch/i386/Makefile is -03 -ftracer -pipe -msoft-float

screenshot of the control GUI is here -> http://amnesty.scene.pl/~sqward/fallow/2.jpg

there are total basic scripts to set the cpu scheduling policy at startup :
this into /etc/conf.d - > http://amnesty.scene.pl/~sqward/fallow/ ... /cpu_sched
and this into /etc/init.d - > http://amnesty.scene.pl/~sqward/fallow/ ... /cpu_sched
The choice is :
"sc" - Staircase based scheduler
"pb" - Priority based scheduler
"eb" - Entitlement based scheduler
don`t forget

Code: Select all

 # rc-update add cpu_sched default 

Of course You can swith and control cpu sched at runtime via this for ex.
GUI is here -> http://amnesty.scene.pl/~sqward/fallow/ ... 1.1.tar.gz

and ebuild is here -> http://amnesty.scene.pl/~sqward/fallow/ ... -r1.ebuild

greetings

[MrApples]

how does this campare to staircase, nicksched, etc as far as performance?

[fallow]

It`s single priority array scheduler so it`s similar to staircase/spa/ebs.
scheduler latency is also similar to the staircase .
Hydra might be slightly less efficient than each of the constituent schedulers would be on its own in some aspects.I don`t feel this much
Overall performance is depends of scheduling policy that are You using at the moment of course .
Some people has opinion that Staricase with his add-ons patches from ck is more agrresive than this.

greetings

[oberyno]

Up and running. The gui scheduler selection tool is pretty cool. If only more of the /proc interface had this kind of user-friendliness.

Anyway, great work.

[fallow]

Tx and I also like this GUI. Initials and Troughput Bonus settings are good also ; at Con Kolivas site are new preempt_timing and truncate_inode_pages_latency_fix patches also

greetings

[primero.gentoo]

ok, up and running ... i've also tried the gui ... but where can i find info about tuning?

would not to be possible to include also nick scheduler?

bye

[fallow]

std 2.6 kernel O(1) Ingo Molnar`s scheduler and Nick`s scheduler using 2 prior tables , stair,spa,ebs one .

this is info

> base_promotion_interval -- (milliseconds) controls the interval between successive promotions (is multiplied by the number of active tasks on the CPU in question) NB no promotion occurs if there are less than 2 active tasks

> time_slice -- (milliseconds) the size of the time slice (i.e. how long it will be allowed to hold the CPU before it is kicked off to allow other tasks a chance to run) that is allocated to a task when it becomes active or finishes a time slice. (min is 1 millisec and max is 1 second).

> max_ia_bonus -- (a value between 0 and 10) that determines the maximum interactive bonus that a task can acquire

> initial_ia_bonus -- (a value between 0 and 10) that determines the initial interactive bonus that a newly forked task will be given. This value will be capped by the max_ia_bonus.

> ia_threshold -- (parts per thousand) is the sleep to (sleep + on_cpu) ratio above which a task will have its interactive bonus increased asymptotically towards the maximum

> cpu_hog_threshold -- (parts per thousand) is the usage rate above which a task will be considered a CPU hog and start to lose interactive bonus points if it has any

> max_tpt_bonus -- (a value between 0 and 9) that determines the maximum throughput bonus that tasks may be awarded

> log_at_exit - (0 or 1) turns off/on the logging of tasks' scheduling statistics at exit. This feature is useful for determining the scheduling characteristics of relatively short lived tasks that run as part of some larger job such as a kernel build where trying to get time series data is impractical.

> compute -- (0 or 1) turn on/off the staircase schedulers "compute" switch

> interactive -- (0 or 1) turn on/off the staircase schedulers "interactive" mode

( switching off Staircase interactive mode ) will disable interactive tasks from having bursts, thus being even stricter about nice levels (suitable for non gui desktop usage or a server)

( switching on Staircase compute mode ) makes round robin intervals much longer, delays task preemption and disables interactive mode to optimise cpu cache usage - suitable for computational intensive tasks.

greetings

[Woocash]

I've got this error

Code: Select all

Calculating dependencies ...done!
>>> emerge (1 of 1) sys-kernel/vivid-sources-2.6.8_rc3-r1 to /
>>> Downloading http://www.kernel.org/pub/linux/kernel/v2.6/testing/patch-2.6.8-rc3.bz2
--15:25:07--  http://www.kernel.org/pub/linux/kernel/v2.6/testing/patch-2.6.8-rc3.bz2
           => `/usr/portage/distfiles/patch-2.6.8-rc3.bz2'
Translacja www.kernel.org... 204.152.189.116
£±czenie siê z www.kernel.org[204.152.189.116]:80... po³±czono.
¯±danie HTTP wys³ano, oczekiwanie na odpowied¼... 404 Not Found
15:25:07 B£¡D 404: Not Found.

>>> Downloading http://www.de.kernel.org/pub/linux/kernel/v2.6/testing/patch-2.6.8-rc3.bz2
--15:25:07--  http://www.de.kernel.org/pub/linux/kernel/v2.6/testing/patch-2.6.8-rc3.bz2
           => `/usr/portage/distfiles/patch-2.6.8-rc3.bz2'
Translacja www.de.kernel.org... 129.143.116.10, 80.252.110.38
£±czenie siê z www.de.kernel.org[129.143.116.10]:80... po³±czono.
¯±danie HTTP wys³ano, oczekiwanie na odpowied¼... 404 Not Found
15:25:08 B£¡D 404: Not Found.

>>> Downloading http://www.us.kernel.org/pub/linux/kernel/v2.6/testing/patch-2.6.8-rc3.bz2
--15:25:08--  http://www.us.kernel.org/pub/linux/kernel/v2.6/testing/patch-2.6.8-rc3.bz2
           => `/usr/portage/distfiles/patch-2.6.8-rc3.bz2'
Translacja www.us.kernel.org... 207.45.221.24, 209.61.158.75, 209.140.211.22, ...
£±czenie siê z www.us.kernel.org[207.45.221.24]:80... po³±czono.
¯±danie HTTP wys³ano, oczekiwanie na odpowied¼... 404 Not Found
15:25:08 B£¡D 404: Not Found.

>>> Downloading http://www.fr.kernel.org/pub/linux/kernel/v2.6/testing/patch-2.6.8-rc3.bz2
--15:25:08--  http://www.fr.kernel.org/pub/linux/kernel/v2.6/testing/patch-2.6.8-rc3.bz2
           => `/usr/portage/distfiles/patch-2.6.8-rc3.bz2'
Translacja www.fr.kernel.org... 212.180.1.39, 213.186.33.38, 81.220.255.20
£±czenie siê z www.fr.kernel.org[212.180.1.39]:80... po³±czono.
¯±danie HTTP wys³ano, oczekiwanie na odpowied¼... 404 Not Found
15:25:08 B£¡D 404: Not Found.

>>> Downloading http://www.uk.kernel.org/pub/linux/kernel/v2.6/testing/patch-2.6.8-rc3.bz2
--15:25:08--  http://www.uk.kernel.org/pub/linux/kernel/v2.6/testing/patch-2.6.8-rc3.bz2
           => `/usr/portage/distfiles/patch-2.6.8-rc3.bz2'
Translacja www.uk.kernel.org... 212.219.56.150, 212.219.56.151, 212.219.56.156, ...
£±czenie siê z www.uk.kernel.org[212.219.56.150]:80... po³±czono.
¯±danie HTTP wys³ano, oczekiwanie na odpowied¼... 404 Not Found
15:25:09 B£¡D 404: Not Found.

>>> Downloading http://www.at.kernel.org/pub/linux/kernel/v2.6/testing/patch-2.6.8-rc3.bz2
--15:25:09--  http://www.at.kernel.org/pub/linux/kernel/v2.6/testing/patch-2.6.8-rc3.bz2
           => `/usr/portage/distfiles/patch-2.6.8-rc3.bz2'
Translacja www.at.kernel.org... 81.223.20.167, 62.116.33.4
£±czenie siê z www.at.kernel.org[81.223.20.167]:80... po³±czono.
¯±danie HTTP wys³ano, oczekiwanie na odpowied¼... 404 Not Found
15:25:09 B£¡D 404: Not Found.

!!! Couldn't download patch-2.6.8-rc3.bz2. Aborting.

Who helps me ?

[fallow]

>>> emerge (1 of 1) sys-kernel/vivid-sources-2.6.8_rc3-r1 to /
>>> Downloading http://www.kernel.org/pub/linux/kernel/ ... .8-rc3.bz2
--15:25:07-- http://www.kernel.org/pub/linux/kernel/ ... .8-rc3.bz2
=> `/usr/portage/distfiles/patch-2.6.8-rc3.bz2'

2.6.8_rc3 ? ? ? - the latest is 2.6.8_rc1 at now.

I wrote at the top of thread , ebuild name is - >

and ebuild is here -> http://amnesty.scene.pl/~sqward/fallow/ ... -r1.ebuild

not 2.6.8_rc3

with correct ebuild`s name , and ebuild - digest and merging is ok .

greetz

[Woocash]

Ok, thx My fault

[fallow]

no problem, Yo`re welcome

I`ve add some latency-fix patches from ck,and new bz2 and ebuild is here
list is here :

Code: Select all

2.6.7-spa_hydra_FULL-v1.5 - cpu scheduler with runtime elevation 
from_2.6.7-staircase7.8_to_staircase7.9 from ck,update for hydra
voluntary-preempt-2.6.7-bk20-H3 - additional kernel preemption patch  
 /option in Processor Type & Features-->Voluntary Kernel Preemption/
wli_preempttest -         non-preemptible critical section timing / also in Proc Type & Feat menu / 
reiser4 snapshot from ck 
gcloop-2.6-20040510
9000_SuSE-117-writeback-lat
BadRAM-2.6.5.2
autotune_inactivation 
autotune_swappiness 
bootsplash
config-nr-tty-devices
config_hz 
defaultcfq
linux-2.6.0.dxr3.20031229
lirc-2.6.5-mm1-20040406
lufs-0.9.7-2.6.0-test9
vesafb-rrc-0.1.6
patch-i386-irq_enable_spinlocks2
patch-ool-spinlocks
rsbac-v1.2.3-2.6.7 - compiled but I`m not sure about good works;)
squashfs1.3r3
supermount-ng204 from ck
__cleanup_transaction-latency-fix from ck
crq-fixes from ck
filemap_sync-latency-fix from ck
get_user_pages-latency-fix from ck
jbd-recovery-latency-fix from ck
journal_clean_checkpoint_list-latency-fix from ck
kjournald-smp-latency-fix from ck
prune_dcache-latency-fix from ck
slab-latency-fix from ck
truncate_inode_pages-latency-fix from ck
unmap_vmas-smp-latency-fix from ck

--end below is the same as in previous --
vesafb_modes.h in /kernel_scr/arch/i386/boot is filled for 1024x768@85Hz for example and CFLAGS in /kernel_src/arch/i386/Makefile is -03 -ftracer -pipe -msoft-float

screenshot of the control GUI is here -> http://amnesty.scene.pl/~sqward/fallow/2.jpg

there are total basic scripts to set the cpu scheduling policy at startup :
this into /etc/conf.d - > http://amnesty.scene.pl/~sqward/fallow/ ... /cpu_sched
and this into /etc/init.d - > http://amnesty.scene.pl/~sqward/fallow/ ... /cpu_sched
The choice is :
"sc" - Staircase based scheduler
"pb" - Priority based scheduler
"eb" - Entitlement based scheduler
don`t forget

Code: Select all

 # rc-update add cpu_sched default 

Of course You can swith and control cpu sched at runtime via this for ex.
GUI is here -> http://amnesty.scene.pl/~sqward/fallow/ ... 1.1.tar.gz

and ebuild is here -> http://amnesty.scene.pl/~sqward/fallow/ ... -r2.ebuild

greetings

[fallow]

Hydra 2.0 is out and 2.6.8_rc2 also

I`ve updated Hydra from 1.5 to 2.0 , GUI is new also .patches are the same as in previous version , but for 2.6.8_rc2 now.

screenshot of the new control GUI v1.2 is here -> http://amnesty.scene.pl/~sqward/fallow/3.jpg

in 2.0 version - staircase interacitve mode is off by default , so it can be switched on by basic script below.

there are total basic scripts to set the cpu scheduling policy at startup :
this into /etc/conf.d - > http://amnesty.scene.pl/~sqward/fallow/ ... /cpu_sched
and this into /etc/init.d - > http://amnesty.scene.pl/~sqward/fallow/ ... /cpu_sched
don`t forget

Code: Select all

 # rc-update add cpu_sched default 

new GUI is here -> http://amnesty.scene.pl/~sqward/fallow/ ... 1.2.tar.gz
and ebuild -> http://amnesty.scene.pl/~sqward/fallow/ ... -r1.ebuild

greetings

[nmcsween]

Fallow think you could get swsusp2 in with this?

[superjaded]

Hmm.. I got a compile error on reiser4 on 2.6.8-rc2-vivid1.

Code: Select all

   CC      fs/reiser4/plugin/file/file.o
fs/reiser4/plugin/file/file.c: In function `sendfile_common':
fs/reiser4/plugin/file/file.c:2453: error: structure has no member named `buf'
make[2]: *** [fs/reiser4/plugin/file/file.o] Error 1
make[1]: *** [fs/reiser4] Error 2
make: *** [fs] Error 2

Using almost the same config (especially when it comes to Reiser4, oldconfig didn't ask about any new Reiser4 options) from my perfectly fine running 2.6.8-rc1-love1 kernel. Here's the config I was using, if it helps anyway.

Since I don't have anything TOO important on my Reiser4 partition, I'll probably compile it without Reiser4 support after I get home from work and see how that goes.

[fallow]

superjaded: sorry for that , i don`t use reiser4 at the moment , I forgot to check it.

i think this sholud fix this

Code: Select all

diff -Naur a/fs/reiser4/plugin/file/file.c b/fs/reiser4/plugin/file/file.c
--- a/fs/reiser4/plugin/file/file.c	2004-07-19 00:28:14.000000000 +0200
+++ b/fs/reiser4/plugin/file/file.c	2004-07-19 00:25:30.000000000 +0200
@@ -2450,7 +2450,7 @@
 
 	desc.error = 0;
 	desc.written = 0;
-	desc.buf = target;
+	desc.arg.buf = target;
 	desc.count = count;
 
 	fplug = inode_file_plugin(inode);

nmcsween : I`m going to work ... , I try with swsup2 tommorow If Ypu`re interested

edit --

also fix for possible redefine error with dcookies with some .config`s - sorry

Code: Select all

diff -Naur b/fs/dcookies.c a/fs/dcookies.c
--- b/fs/dcookies.c	2004-07-18 23:51:24.000000000 +0200
+++ a/fs/dcookies.c	2004-07-19 00:48:36.255903440 +0200
@@ -105,38 +105,6 @@
 }
 
 
-/* This is the main kernel-side routine that retrieves the cookie
- * value for a dentry/vfsmnt pair.
- */
-int get_dcookie(struct dentry * dentry, struct vfsmount * vfsmnt,
-	unsigned long * cookie)
-{
-	int err = 0;
-	struct dcookie_struct * dcs;
-
-	down(&dcookie_sem);
-
-	if (!is_live()) {
-		err = -EINVAL;
-		goto out;
-	}
-
-	dcs = dentry->d_cookie;
-
-	if (!dcs)
-		dcs = alloc_dcookie(dentry, vfsmnt);
-
-	if (!dcs) {
-		err = -ENOMEM;
-		goto out;
-	}
-
-	*cookie = dcookie_value(dcs);
-
-out:
-	up(&dcookie_sem);
-	return err;
-}
 
 
 /* And here is where the userspace process can look up the cookie value
@@ -286,43 +254,6 @@
 	struct list_head next;
 };
  
-struct dcookie_user * dcookie_register(void)
-{
-	struct dcookie_user * user;
-
-	down(&dcookie_sem);
-
-	user = kmalloc(sizeof(struct dcookie_user), GFP_KERNEL);
-	if (!user)
-		goto out;
-
-	if (!is_live() && dcookie_init())
-		goto out_free;
-
-	list_add(&user->next, &dcookie_users);
-
-out:
-	up(&dcookie_sem);
-	return user;
-out_free:
-	kfree(user);
-	user = NULL;
-	goto out;
-}
-
-
-void dcookie_unregister(struct dcookie_user * user)
-{
-	down(&dcookie_sem);
-
-	list_del(&user->next);
-	kfree(user);
-
-	if (!is_live())
-		dcookie_exit();
-
-	up(&dcookie_sem);
-}
 
 EXPORT_SYMBOL_GPL(dcookie_register);
 EXPORT_SYMBOL_GPL(dcookie_unregister);

greetings

[nmcsween]

Fallow I wouldn't mind working with you to make releases of our own. I have already worked on getting swsusp2 to work you 2.6.8-rc1 mm1 and a handful of other patches, seeing as we both like the staircase scheduler wouldn't it be in both of our interests if we worked together?

[fallow]

I think it is a good idea and I am agree - it`s our both interest
It would be nice
I fixed rejects with swsup2 and 2.6.8-rc2-vivid1 but i have to fix compile errors now , I haven`t enough time at the moment , but I`m try later.
Or ... If You are interested in togheter patch / work and You applied swsup2 to 2.6.8-rc1 so please do that with 2.6.8-rc2-vivid1 and it would be our "patchset" I Can send You the actual diff after fixing rejects if it might be a some kind of help.
However - the idea is nice

greetings

[nmcsween]

I'll need the other patches you have to apply swsusp2 since it's horrific to get to work with when putting it on last. I have a couple of sites that could offer space for our patches or even a subversion repository. I also wouldn't mind taking a few patches from 2.6.8-rc1-mm1 and adding them as well.

[oberyno]

Works great after applying the reiser4 patch of course.

Code: Select all

oberyno> uname -r                                                  /usr/portage
2.6.8-rc2-vivid1

[fallow]

Oberyno : great

I`ve fixed rejects and compile errors with 2.6.8-rc1-vivid1 and swusp2.0.0.100 version , also reiser4 plugin error and dcookies.Now kernel with swsup2 compiles fine and booting , I don`t know hot to test swsup2 , I never had this feature before so it`s testing version

ebuild,bz2,init.d.conf.d is as previous everytning in
-->> http://amnesty.scene.pl/~sqward/fallow/268-rc2-vivid2


Nmcsween : if this version doesn`t work good , so all hope in You
I think some patches from mm might be a good idea also
I don`t have enough space on the server , so i deleted one of previous versions to upload new. is 268-rc2-vivid1 and testing 268-rc2-vivid2 in dirs at the moment
I`m going to upload all patches for You , please give my adress for that

greetings

[nmcsween]

Nmcsween[AT]Gmail.com. I'll also get the subversion repository set up asap. I'll also set up a htaccess file for you

[fallow]

Ello new version is here based on 2.6.8-rc3
list is here :

Code: Select all

---"main" patches---------------
HYDRA3.1 CPU scheduler with runtime elevation
bootsplash-3.1.4-sp3-2.6.7
* bootsplash modified for rc3 , temporary working without user font selection*
cdfs-2.6.3a
config-nr-tty-devices
config_hz
defaultcfq
gcloop-2.6-20040527
linux-2.6.0.dxr3
lirc - long working version 
lufs-0.9.7
squashfs2.0
supermount-ng204
vesafb-rrc-0.1.6
reiser4-2004.07.23

---LKML patches-----------------
io_performance_fix_with_2.6.7_and_2.6.8_rc
mpgage_readpage_fix

---latency fixes----------------
SuSE-117-writeback-lat
cleanup_transaction-latency-fix
truncate_inode_pages-latency-fix
unmap_vmas-smp-latency-fix
filemap_sync-latency-fix
get_user_pages-latency-fix
get_user_pages-latency-fix
jbd-recovery-latency-fix
journal_clean_checkpoint_list-latency-fix
kjournald-smp-latency-fix
prune_dcache-latency-fix
slab-latency-fix

---additional patches from mm---
patch-i386-irq_enable_spinlocks2
patch-ool-spinlocks
reiserfs-v3-barrier-support
sound-control-build-fix
bk-acpi.patch
bk-agpgart.patch
bk-alsa.patch
bk-drm.patch
bk-ieee1394.patch
bk-libata.patch
bk-ntfs.patch
bk-pci.patch
disk-barrier-core.patch
disk-barrier-dm.patch
disk-barrier-ide.patch
disk-barrier-md.patch
disk-barrier-scsi.patch
gcc35 fixes from mm.

vesafb_modes.h in /kernel_scr/arch/i386/boot is filled for 1024x768@85Hz for example and CFLAGS in /kernel_src/arch/i386/Makefile is -03 -ftracer -pipe -msoft-float

screenshot of the control GUI is here -> http://amnesty.scene.pl/~sqward/fallow/4.jpg

there are total basic scripts to set the cpu scheduling policy at startup :
this into /etc/conf.d - > http://amnesty.scene.pl/~sqward/fallow/ ... /cpu_sched
and this into /etc/init.d - > http://amnesty.scene.pl/~sqward/fallow/ ... /cpu_sched
The choice is :
"sc" - Staircase based scheduler
"pb" - Priority based scheduler
"eb" - Entitlement based scheduler
don`t forget

Code: Select all

 # rc-update add cpu_sched default 

Of course You can swith and control cpu sched at runtime via this for ex.
GUI is here -> http://amnesty.scene.pl/~sqward/fallow/ ... 1.3.tar.gz

and ebuild is here -> http://amnesty.scene.pl/~sqward/fallow/ ... -r1.ebuild

greetings

[fallow]

ok , by the some request , PPC patches for 2.6.8-rc3-vivid1 from bk`s are here .

http://amnesty.scene.pl/~sqward/fallow/ ... /ppcv1.bz2

and the same ebuild , but with this patches are here :
http://amnesty.scene.pl/~sqward/fallow/ ... -r1.ebuild

I`m going on vacations
greetings

[_troll_]

ok , by the some request , PPC patches for 2.6.8-rc3-vivid1 from bk`s are here .

THX!!!! PPC rulez!

And of course have fun

Reagrds,
Przemek

[Woocash]

I've got this error when I wrote this :

Code: Select all

rc-update add cpu_sched default

woocash ~ # rc-update add cpu_sched default
* cpu_sched not executable; skipping