2005-10-28 11:18:35

by Con Kolivas

[permalink] [raw]
Subject: 2.6.14-ck1

These are patches designed to improve system responsiveness and interactivity.
It is configurable to any workload but the default ck* patch is aimed at the
desktop and ck*-server is available with more emphasis on serverspace.

Apply to 2.6.14
http://ck.kolivas.org/patches/2.6/2.6.14/2.6.14-ck1/patch-2.6.14-ck1.bz2

or server version
http://ck.kolivas.org/patches/2.6/2.6.14/2.6.14-ck1/patch-2.6.14-ck1-server.bz2

web:
http://kernel.kolivas.org
all patches:
http://ck.kolivas.org/patches/
Split patches available.


Changes:

Added:
+adaptive-readahead-4.patch
We Fengguang's adaptive readahead patch. Please test and report experiences -
Wu has been cc'ed on this email, please keep him cc'ed for reports.

+set_workqueue_nice-1.patch
Add a function to set workqueue nice levels

+dm-drop_kcryptd_prio.patch
Use the above function to set kcryptd priority to nice 0 since kcryptd
consumes massive amounts of cpu. This code is safe but I don't even know if
it's successfully changing priority yet.


Updated:
~2.6.14_to_staircase12.1.diff
Rolled up staircase tweaks

~mm-swap_prefetch-18.patch
Current version of swap prefetching. Stable!


Full patchlist:
2.6.14_to_staircase12.1.diff
schedrange.diff
schedbatch2.9.diff
sched-iso3.2.patch
smp-nice-support7.diff
1g_lowmem1_i386.diff
defaultcfq.diff
isobatch_ionice2.diff
rt_ionice.diff
pdflush-tweaks.patch
hz-default_values.patch
hz-no250.patch
vm-mapped.diff
vm-lots_watermark.diff
vm-background_scan-1.diff
mm-swap_prefetch-18.patch
set_workqueue_nice-1.patch
dm-drop_kcryptd_prio.patch
adaptive-readahead-4.patch
2614ck1-version.diff


Cheers,
Con


Attachments:
(No filename) (1.57 kB)
(No filename) (189.00 B)
Download all attachments

2005-10-28 11:53:39

by Wu Fengguang

[permalink] [raw]
Subject: Re: 2.6.14-ck1

On Fri, Oct 28, 2005 at 09:18:09PM +1000, Con Kolivas wrote:
> These are patches designed to improve system responsiveness and interactivity.
> It is configurable to any workload but the default ck* patch is aimed at the
> desktop and ck*-server is available with more emphasis on serverspace.
> ...
> Added:
> +adaptive-readahead-4.patch
> We Fengguang's adaptive readahead patch. Please test and report experiences -
> Wu has been cc'ed on this email, please keep him cc'ed for reports.
Thanks. I've done a lot testings and improvements these days, and will release a
new version soon.

Regards,
Wu Fengguang

2005-11-02 21:35:03

by Predrag Ivanovic

[permalink] [raw]
Subject: Re: [ck] 2.6.14-ck1

On Fri, 28 Oct 2005 21:18:09 +1000
Con Kolivas wrote:

<snip>
> Changes:
>
> Added:
> +adaptive-readahead-4.patch
> We Fengguang's adaptive readahead patch. Please test and report
> experiences - Wu has been cc'ed on this email, please keep him cc'ed
> for reports.
>
Con,any recommended value for /proc/sys/kernel/readahead_ratio,
or is it automagicly set?It's value is 0 ATM.

Great work with ck.BTW :-)

Pedja
--
Data is a lot like humans: It is born. Matures. Gets married to other
data, divorced. Gets old. One thing that it doesn't do is die. It has
to be killed. -- Arthur Miller

2005-11-02 22:48:49

by Con Kolivas

[permalink] [raw]
Subject: Re: [ck] 2.6.14-ck1

On Thu, 3 Nov 2005 07:38 am, Predrag Ivanovic wrote:
> On Fri, 28 Oct 2005 21:18:09 +1000
> Con Kolivas wrote:
>
> <snip>
>
> > Changes:
> >
> > Added:
> > +adaptive-readahead-4.patch
> > We Fengguang's adaptive readahead patch. Please test and report
> > experiences - Wu has been cc'ed on this email, please keep him cc'ed
> > for reports.
>
> Con,any recommended value for /proc/sys/kernel/readahead_ratio,
> or is it automagicly set?It's value is 0 ATM.

Yes. First it's supposed to be in /proc/sys/vm (my fault on the merge), and it
should be set to about 50. All this is corrected in 2.6.14-ck2 which has the
new readahead code, the tunable in the correct location, and the default set
to 50.

> Great work with ck.BTW :-)

Thanks!

Cheers,
Con

2005-11-03 08:34:38

by Rudo Thomas

[permalink] [raw]
Subject: Re: 2.6.14-ck1

> > Con,any recommended value for /proc/sys/kernel/readahead_ratio, or
> > is it automagicly set?It's value is 0 ATM.
>
> Yes. First it's supposed to be in /proc/sys/vm (my fault on the
> merge), and it should be set to about 50. All this is corrected in
> 2.6.14-ck2 which has the new readahead code, the tunable in the
> correct location, and the default set to 50.

Con, it's still in /proc/sys/kernel in ck2, AFAICT.

Rudo.

2005-11-03 08:46:04

by Con Kolivas

[permalink] [raw]
Subject: Re: 2.6.14-ck1

On Thu, 3 Nov 2005 07:34 pm, Rudo Thomas wrote:
> > > Con,any recommended value for /proc/sys/kernel/readahead_ratio, or
> > > is it automagicly set?It's value is 0 ATM.
> >
> > Yes. First it's supposed to be in /proc/sys/vm (my fault on the
> > merge), and it should be set to about 50. All this is corrected in
> > 2.6.14-ck2 which has the new readahead code, the tunable in the
> > correct location, and the default set to 50.
>
> Con, it's still in /proc/sys/kernel in ck2, AFAICT.

Bah you're right. Thanks for the heads up. It will work the same there; it's
just not meant to be there. Anyway I don't think you'll need to be fiddling
with the default value generally.

Cheers,
Con