On Fri, 16 Oct 1998, Steven Hascall wrote:

> The log file system is completely hosed under NetBSD, isn't it? I brought
> the system down no less than 6 times before I decided to give up on it.
> There isn't any special rule that the lfs can be the only fs on a disk, is
> there? 

There have been changes made to the lfs drivers and tools in -current,
only this month. Unfourtunately, no one's been able to build a working
mac68k kernel since about Sep. 1.

> Also I noticed that any mfs I try to mount (from the fstab at least) can
> be no larger than about 60 megabytes or so. Is there a way around this? Is

If you mount it on the swap partition, it comes out to be the same
size as the swap partition. You can mount more than one that way. I'd
hate to think what would happen if they all filled up at once, though,
so I use only one, and limit it to 24M, like so

/dev/sd0a	/		ffs	rw		1 1
/dev/sd0b	none		swap	sw		0 0
/dev/sd0b	/tmp		mfs	rw,-s=48000	0 0
/dev/sd0g	/usr		ffs	rw		1 2
/dev/sd2a	/var		ffs	rw		1 2
/dev/sd2b	none		swap	sw		0 0
/dev/sd1a	/mnt/1		ffs	rw,async	0 2
/dev/sd1b	none		swap	sw,-priority=1	0 0
kern		/kern		kernfs	rw		0 0
proc		/proc		procfs	rw		0 0

The "async" partition is linked to /usr/obj. This works a lot like a
ram disk, except that it is written out when memory becomes low.
Haven't had any problems due to mounting async particularly, but it's
a removable drive, which has it's own problems. I wouldn't want to
trust that for my /home partition.

You mentioned, in an earlier post, that you were looking to speed up a
Quadra ???. I find that having /var on a separate drive makes the
system feel much snappier, especially when I'm using "pine" while my
huge mail spool is still downloading.

> this an option I can set if I recompile my own kernel? Is there any
> advantage to compiling a kernel with the -O3 and -m68040 flags. I seem to
> remember my system actually ran *slower* the last time I built a kernel
> myself, and I used those flags thinking I would actually see an
> improvement. Thanks for any help.

I noticed a minor, subjective improvement with -m68040 for the
Xserver. With the kernel, it doesn't seem to make much difference.
Kernels compiled with -O3 aren't very much smaller than kernels
compiled with -O2. I can't imagine why it would be slower, though,
unless you're timing "make". Using -O3 as your default will obviously
slow down all of your "make"s.

For what it's worth, I've had no problems with using -02 -m68040 for
everything. The last kernel that works was build on Aug, 30; last tree
Sep. 11, just before the signal changes.