> What happens when you put a file there?  What does df then
> show?
> 
> say, 
> 
> dd if=/dev/zero  of=/blah/blah1 bs=1000000 count=1

Well, I tried just copying the kernel to my /tmp partition last night,
and the only thing that happened is that df showed the freespace had
gone even more negative.

I was reminded that newfs had left a coredump behind when I tried to
reformat my /home partition the other. That's a reason I'm wondering
whether maybe lack of an FPU has something to do with it.

Maybe this weekend I should hook up a serial console and capture the
bootloog and the output of dumpfs?

BTW, Bruce would your soft float para-port for the LCs be likely to work
for an FPU-less '030? It seemed like I read something in your pages that
indicated it might, but I haven't been able to find that again. (One of
the reasons I'm trying to get this running is to take a stub at doing a
soft-float port of my own.)

> cheers
> 
> bruce
> 
> Joel Rees <joel@alpsgiken.gr.jp> wrote:
> > > 
> > > On Monday, June 2, 2003, at 10:38  PM, Joel Rees wrote:
> > > 
> > > > <http://mail-index.netbsd.org/port-mac68k/2001/01/12/0007.html>
> > > 
> > > ...
> > > 
> > > > Is it likely that the bug from January 2001 has been re-introduced or
> > > > something? I'm not getting any more recent positives on this from marc
> > > > or google.
> > > 
> > > I don't understand... there is no bug here.
> > 
> > Was the incorrect initialization in the response then a user error?
> > 
> > If so, then I've made the same error three times, and it would be nice
> > if someone would hit me over the head with a clue-by-four so I can avoid
> > doing it a fourth time. Installs on mac68k keep my wife awake. :-/
> > 
> > As I said in my first post, 
> > 
> >     http://mail-index.netbsd.org/port-mac68k/2003/06/02/0000.html
> > 
> > (which was trimmed) this is on a fresh install on a performa 550 w/out
> > FPU, the three partitions that show the negative free space start above
> > 700M in a 2G HD, all are 60M+, and none have more than 200K in use. 
> > 
> > (And I apologize that I hit "paste without quoting" on my reply.)
> > 
> > > The description of what is happening in the URL you cite is slightly 
> > > wrong.  Yes, there is %5 that is reserved for the superuser, but "100%" 
> > > is adjusted to compensate.  This means that the actual "all space used" 
> > > value is "105%".
> > > 
> > > Unix has done this for a .. very long time.
> > 
> > Thanks to everyone who pointed this out. If I am missing the reason why
> > df-ing a 60MB partition with nothing but an empty lost+found (du reports
> > 8k in use) tells me I have negative free space, please tell me.


-- 
Joel Rees <joel@alpsgiken.gr.jp>