Complete.Org: Mailing Lists: Archives: discussion: February 2002:
[aclug-L] Re: An idea -- any volunteers/helpers?
Home

[aclug-L] Re: An idea -- any volunteers/helpers?

[Top] [All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index] [Thread Index]
To: discussion@xxxxxxxxx
Subject: [aclug-L] Re: An idea -- any volunteers/helpers?
From: james l <james@xxxxxxxxxxxxxxxxxx>
Date: Fri, 22 Feb 2002 08:18:20 -0600
Reply-to: discussion@xxxxxxxxx

On Thursday 21 February 2002 23:19, you wrote:
> > -----Original Message-----
> > From: discussion-bounce@xxxxxxxxx [mailto:discussion-bounce@xxxxxxxxx]On
> > Behalf Of Anne McCadden
> > Sent: Thursday, February 21, 2002 8:02 PM
> >
> > I have linux servers also and I probably have unneeded services
> > running.  I know that to shut these off is in rc.d, but it has been a
> > while since I have worked with this.  I certainly don't want to shutdown
> > services that I am using, I just want to streamline the services.  ~Anne
> >
> > Luke Wahlmeier wrote:
> > > I would love to recomple-edit the kernel, i have read all about it, and
> > > understand the steps very well, but the linux computers that i have are
> > > all servers, and i can aford to spend time on them if i do something
> > > wrong my 1st time.  Once i learn how to and get comfertable doing it i
> > > would do it on the servers to help speen them up by taking all the
> > > unneeded process out of it.  If anyone is going to be doing this soon I
> > > would love to learn.
>
> Recompiling the kernel has ZERO effect on running processes. The most
> common reason for recompiling is to gain some new feature or hardware
> support offered by a later kernel revision.  The kernels that are shipped
> with modern distribuitons have 'hooks' built in to support modules for a
> wide array of devices.  It is possible to reduce the base 'footprint in
> ram' of the kernel by only building in the 'hooks' that are needed to
> support your exact hardware. However, if you should later change hardware,
> you will need to recompile the kernel to support the new hardware. It is
> also possible to optimize the kernel code for your particular model of
> processor, but the same caveats apply.
>

There are some things that do require a rebuild of the kernel (or a kernel 
module) For example: how many people have usb CompactFlash readers? While 
there is support (now in the kernel, when I first got it, it was a patch) it 
does require going into the kernel's config to make work. (btw, it is a 
sub-option of usb-storage.o if you care, and requires EXPEREMENTAL support to 
be enabled.)

I would disagree that it has zero effect on running processes. If the kernel 
is more efficient (say via optimizations such as specific ide chipset 
drivers, processor specific optimizations (SSE, 3dnow, MMX, etc) and other 
such things) then the processes will have to spend less time in the kernel, 
so more time is spent in the processes. Granted an individual thing may save 
only a few cycles, but if it is in something called lots of times in the 
kernel (say the ide driver) that can add up.

That being said, generally the distros will have better kernel setups
 than most people will be able to create, but they may lack a specific piece 
of hardware, so you may need to go in and complile modules/recompile the 
kernel. 

Plus there can be some wierd errors. For example: chosing to complile in the 
dri (Direct Rendering I? for XF86) for both the 3dfx and ati card will result 
in the ati driver overwriting the 3dfx driver in the kernel when it loads. I 
found this out the hard way (I thought I might get an ati card, and wanted to 
have it in already, but I have a voodoo3 so it kind of messed it up)

James L.

> --dwh
>
> ---
> Dale W Hodge - dwh@xxxxxxxxxxxxxxxx
> Secretary - info@xxxxxxxxx
> Air Capital Linux User's Group  (ACLUG)
> ---
>
>
>
>
> -- This is the discussion@xxxxxxxxx list.  To unsubscribe,
> visit http://www.complete.org/cgi-bin/listargate-aclug.cgi
-- This is the discussion@xxxxxxxxx list.  To unsubscribe,
visit http://www.complete.org/cgi-bin/listargate-aclug.cgi


[Prev in Thread] Current Thread [Next in Thread]