Subject: Re: Massive changes in progress...
To: Charles J. Williams <chas@volt.nrl.navy.mil>
From: Allen Briggs <briggs@puma.bevd.blacksburg.va.us>
List: macbsd-development
Date: 06/21/1995 10:03:09
> >Things may still be a bit broken. I've changed the way that memory is
> >mapped at boot time--if you want the details, I'll post 'em, but the
> I would like them if you dont mind
[ This was in a message just to me, but it's probably of interest to
others if I got a response back so quickly ;-) ]
I've noticed a few other changes that you'll need to make if you update
to the new kernel. Besides mountd, a number of the networking status
programs (like netstat) have been modified to match new kernel data
structures. Also, I found identd sucking up resources, so I had to
rebuild it. I'm going to build a snapshot as soon as I get a fresh
kernel build, so you might want to hang around for that... ;-)
Changes that I remember...
* Raised the default process limits as follows:
max text size: 6MB -> 8MB
default data size limit: 8MB -> 16MB
max data size limit: 16MB -> 64MB
max stack size: 16MB -> 64MB
* A few other parameters were modified.
* Modified page mapping to be done in C instead of assembly--
moving page maps for Nubus, ROM and I/O from VA==PA to be
mapped right after the kernel and before Sysmap. There were,
of course, a few more changes around the drivers to accomodate
this.
* The C code for the above was modified from the HP300 code
and placed in pmap_bootstrap.c. mac68k_init.c was obsoleted
by this code as that contained the memory bootstrap mapping
for the 040.
* The machine-dependent VM handling was modified for the hp300
68040 model instead of the Amiga 68040 model. This spread from
pmap_bootstrap.c through locore.s, trap.c, vm_machdep.c, pmap.c
and machdep.c (a tad).
* Cleaned up the nubus probe and internal video. It now
recognizes the ethernet card and internal video on my IIsi
(which doesn't have an FPU so I can't test more than booting ;-).
* Some general cleanup stuff. There's a lot more to do.
* Got rid of the BROKEN_SWAP define and moved to a new method of
forking children (including pids 1 and 2). Swapping should work.
* A few other changes that I can't recall.
It's been a busy few weeks--I'd hoped to get these in sooner. My
mailbox has been queuing up problems that need looking into...
Later,
-allen
--
Allen Briggs - end killing - allen.briggs@bev.net ** MacBSD == NetBSD/mac68k **