Subject: Re: Recognizing swap
To: Bill Merrill <wlmerril@mtu.edu>
From: Steven Campbell <scampbel@astral.magic.ca>
List: port-mac68k
Date: 10/29/1996 15:48:34
On Tue, 29 Oct 1996, Bill Merrill wrote:

> First off, obligatory thanks to all people that made it possible for my
> lcIII to boot into NetBSD this weekend.  I am so happy to be running
> Unix on my mac.
> 
> I hope to be able to contirbute back to the effort.
> 
> Though I do have my newbie question...
> 
> How do I get a swap partition to work?
> Right now I am running a usr and root partition a 100 meg drive, sd1a,
> and have partitioned sd3a, a 50 meg drive for swap (overkill I know, but
> I just wanted to get things going).  When I ran build device from
> the installer, the fstab entry for swap was "none".  I have changed it
> to sd3a, but get the following...
> 
> fugly# swapon -a
> swapon: /dev/sd3a: device not configured
> 
> >From the faq, it didn't look like I need to do anything like makeswap
> form Linux, so I am confused.
> 
> Also, before during boot, but before swapon is reached, I get a message
> something like WARNING: no swap.
> 

Most NetBSD kernels assume that your first swap partition will be on the
same drive as your root filesystem.  About a month ago, I tried building
myself a kernel to force my IIci to swap on a different drive.  It didn't
work, and I simply didn't have any swap space.  Luckily, I did have enough
ram to support all of the processes I was running.  What I had to do was
rethink how I had my filesystems laid out.  I have my root filesystem on a
Quantum ELS80 ( 80 MB ).  It takes up about 60 MB, and the other 20 MB is
given to a swap partition.  I have a 2nd drive, a 540 MB Quantum Maverick,
and it has a 350 MB /usr partition, and another 20 MB of swap, and the
rest is given to the MacOS.  I was interested in my machine swapping to a
different drive for performance reasons.  After all, if you're running
some real memory hogs, you'll take less of a performance hit if the swap
and /usr filesystems are on different drives. 

Some people may be working on this problem, but I think it may take a
lot of recoding of the kernel sources.

Steve