[gentoo-dev] Re: [gentoo-user] a preempt patch lament

[gentoo-dev] Re: [gentoo-user] a preempt patch lament

[Bruce A. Locke]

Yeah, the preempt patch seems to have some rather nasty rejects against 
-aa's rather large (?!?!) patch set.  I myself am either sticking
2.4.17-r4 or building my own because I depend on preempt and refuse to
use any kernel for my workstation without it.  And the fact -aa's patch
is huge and tested by a smaller audience then even -ac's patch set
doesn't sit well with me either.

Anyone know if some kind soul is maintaining a preempt + tested XFS
patch?  With the dozens of different kernel trees these days someone has
to be doing something like that :)


On Wed, 2002-02-27 at 22:06, Matthew Kennedy wrote:
> I realise the Robert Love preemption patch is no longer in our new
> kernel sources. After upgrading to gentoo's 2.4.18 though, I am really
> seeing some performance loss. Currently I am emerging postgresql (a
> fairly heavy ebuild), the mouse *definitely lags* while this is
> happening. Last time I emerged postgresql was a few days ago under the
> preempt patched 2.4.17 and I saw no such lag under the same postgresql
> ebuild. 
> 
> Also XMMS clicks and lags constantly when playing oggs (don't know about
> mp3 -- don't use em) during the same emerge. This didn't happen before
> either. I can actually *see* a correlation between each cc invocation in
> my xterm running the emerge and the lagging in XMMS. I can solve the
> XMMS problem by playing with the ALSA buffer settings in the XMMS
> preferences menu though.
> 
> I guess the preempt patch wasn't the only variable to change between
> 2.4.17 and 2.4.18, but the symptoms here are exactly what preempt is
> supposed to cure. I did try manually patching the gentoo kernel sources
> with preempt, but it's not compatible with the new VM stuff.
> 
> Anyway... end of bitching. I might look into applying xfs and preempt
> patches to a vanilla kernel manually, but it would be cool to see
> preempt back in for 2.4.19 :) 
> 
> _______________________________________________
> gentoo-user mailing list
> gentoo-user@gentoo.org
> http://lists.gentoo.org/mailman/listinfo/gentoo-user
> 
-- 

Bruce A. Locke
blocke@shivan.org

Re: [gentoo-dev] Re: [gentoo-user] a preempt patch lament

[Matthew Kennedy]

On Wed, 2002-02-27 at 22:14, Bruce A. Locke wrote:
> Anyone know if some kind soul is maintaining a preempt + tested XFS
> patch?  With the dozens of different kernel trees these days someone has
> to be doing something like that :)

After playing around with a few patch sets, I think I'll move back to
gentoo's 2.4.17-r4 and wait for SGI to create a 2.4.18 patch set for XFS
(should be out any time). Then I'll apply the preempt patch to that and
see what happens. I too can't live with out that preempt patch... can't
imagine how I lived before it.

Re: [gentoo-dev] Re: [gentoo-user] a preempt patch lament

[Daniel Robbins]

On Wed, 2002-02-27 at 21:14, Bruce A. Locke wrote:
> 
> Yeah, the preempt patch seems to have some rather nasty rejects against 
> -aa's rather large (?!?!) patch set.  I myself am either sticking
> 2.4.17-r4 or building my own because I depend on preempt and refuse to
> use any kernel for my workstation without it.  And the fact -aa's patch
> is huge and tested by a smaller audience then even -ac's patch set
> doesn't sit well with me either.
> 
> Anyone know if some kind soul is maintaining a preempt + tested XFS
> patch?  With the dozens of different kernel trees these days someone has
> to be doing something like that :)

Michael Cohen (mjc) and Shawn (shawnX) in #kernelnewbies are working on
this as we speak.  -aa doesn't really give us any viable options for
good desktop performance, so we're moving to an -ac kernel augmented
with XFS and -preempt (as an option for desktops).  So far, we have
everything but -preempt working.  It applies cleanly but does not
actually do anything.  Michael Cohen is working on a fix, and when he
has one I'll roll out a new kernel.  -ac also includes -rmap which is a
great performance enhancer for the desktop.

Best Regards,

-- 
Daniel Robbins                                  <drobbins@gentoo.org>
Chief Architect/President                       http://www.gentoo.org 
Gentoo Technologies, Inc.

Re: [gentoo-dev] Re: [gentoo-user] a preempt patch lament

[Tod M.Neidt]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 868 bytes --]

> Michael Cohen (mjc) and Shawn (shawnX) in #kernelnewbies are working on
> this as we speak.  -aa doesn't really give us any viable options for
> good desktop performance, so we're moving to an -ac kernel augmented
> with XFS and -preempt (as an option for desktops).  So far, we have
> everything but -preempt working.  It applies cleanly but does not
> actually do anything.  Michael Cohen is working on a fix, and when he
> has one I'll roll out a new kernel.  -ac also includes -rmap which is a
> great performance enhancer for the desktop.
> 
Hi Daniel!

You may have already see this, but the Jörg Prante tree (2.4.18-jp6) looks 
interesting.  I might give it a spin.  See

http://lwn.net/daily/#t27

Regards,

tod

---------------------------------------------
This message was sent using Fidelity Networks, Inc Mailman Professional.
http://www.fidnet.com/