Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 8 Oct 1995 13:52:07 +0100 (MET)
From:      Andreas Klemm <andreas@knobel.gun.de>
To:        Jake Hamby <jehamby@lightside.com>
Cc:        John Fieber <jfieber@indiana.edu>, doc@freebsd.org, hackers@freebsd.org
Subject:   Re: Kernelconfig.sgml FINALLY FINISHED!
Message-ID:  <Pine.BSF.3.91.951008130416.2618C-100000@knobel.gun.de>
In-Reply-To: <Pine.BSF.3.91.951005112134.766B-100000@localhost>

next in thread | previous in thread | raw e-mail | index | archive | help
-hackers people ... see comment/wish at XXX)

On Thu, 5 Oct 1995, Jake Hamby wrote:

> Well, after many more hours of hard work, kernelconfig.sgml is finally
> finished (all 34K of it)!  I hope it is as good as that new Printing 
> chapter I hear so much about.  The copy is available at:
> 
> http://cs.sci.csupomona.edu/~jehamby/handbook/kernelconfig.sgml

Hi !

Here a few suggestions concerning your great kernel doc:

>>------------------------------------------------------------------------<<

5.2 Building and Installing a Custom Kernel

	"...
	Next, move to the i386/conf directory and copy the GENERIC 
	configuration file to the name you want to give your kernel.
	For example: 
		# cd /usr/src/sys/i386/conf
		# cp GENERIC MYKERNEL
	..."

	I would add a note, like "don't name your kernel DEBUG,
	since that would cause debugging statements to be compiled
	into the kernel, which is only useful for kernel developers."

	Additionally one should avoid names, that conflict with
	other kernel definitions. To get an idea, what names are
	reserved, one should browse through /sys/i386/conf/LINT,
	which is a list of all possible configuration file
	definitions.

	I've heard several times, that some people fail in this
	area :-)

>>------------------------------------------------------------------------<<

5.2 Building and Installing a Custom Kernel

	"...
	When you're finished, type the following to compile and 
	 install your kernel: 

		# /usr/sbin/config MYKERNEL
		# cd ../../compile/MYKERNEL
		# make
		# make install
	..."

		# make depend			<<== what about this ?!
		# make
		# make install

	I always thought, that a "make depend" is important here ?!

>>------------------------------------------------------------------------<<

5.2 Building and Installing a Custom Kernel

	I would add this note after "# make install"

	The last stage of kernel compilation is loading the kernel 
	executable. This is done by the program ld. If ld complains
	about missing or unresolved symbols in some kernel modules
	(".o"-files) then it might be the case, that the kernel library
	and your kernel sources in /usr/src/sys are a bit out of
	sync. Then try the following to cure the problem:

	# cd /usr/src/sys
	# make clean

	After that retry loading the kernel executable with the following
	commands:

		# cd /sys/compile/MYKERNEL
		# make

	After that proceed with

		# make install

>>------------------------------------------------------------------------<<

	"...
	5.3.3. Filesystem Options
	[...]
	but feel free to comment out support for filesystems you use less 
	often (perhaps the MS-DOS filesystem?), since they will be dynamically
	loaded from the Loadable Kernel Module directory /lkm the first time 
	you mount a partition of that type.
	..."

XXX)

I think it's a good strategy, to comment out filesystems, which 
can be loaded dynamically. It saves possibly much kernel memory.
Since it _will_ be loaded on demand, it's _there_ after you 
demanded it. So why carrying the filesystem stuff with you in kernel, 
if you don't really need it ?! So why not recommending the people,
to comment out things that can be loaded on demand ?!

Most people have machines with less memory. Many people like to
run X11 although they have perhaps only 8 MB RAM or so ...
Commenting out less often used Filesystems should be standard 
manner in my eyes. The GENERIC kernel profile should reflect
this, too !!!

One feature I'd like to see in the future would be a feature,
to free kernel memory from NFS or other dynamically loaded things,
after it's not needed anymore... Would that be possible ?

Imagine, NFS wasn't used after NFS_UNLOAD_THRESHOLD seconds ...
it would be fine, if it could be unloaded to free kernel space ?!?!

(this idea is the reason for "crossmailing into hackers" sorry for
the manual-overhead)

>>------------------------------------------------------------------------<<

Ok, this for now ;-)

Thanks for that wonderful docu

	Andreas ///



--
$$ apsfilter - magic print filter 4lpd      @home : andreas@knobel.gun.de
$$ ftp://sunsite.unc.edu                    @work : andreas@sunny.wup.de
$$ /pub/Linux/system/Printing/aps-491.tgz   knobel: >>> powered by FreeBSD <<<




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.3.91.951008130416.2618C-100000>