On Fri, 31 May 2002, dakidd wrote:

> At 04:05 -0700 05/31/2002, Frederick Bruckman wrote:
> >On Wed, 29 May 2002, dakidd wrote:
> >
> >> We recommend creating a non-root account and using su(1) for root access.
> >> # /sbin/mount -uKernel FPU trap.
> >> trap type 16, code = 0x0, v=0x0
> >
> >What it means is that 1) the PowerBook170 has no FPU (Right?)
>
> Based on (1) My memory of physically eyeballing the 68882 chip on the
> bottom side of the processor board when I installed it, (2) The fact that
> TechTool (and similar programs) claim that it does indeed have an FPU (See
> the bottom of this message for a pasted TechTool report on the machine),
> and (3) The refusal of SoftFPU to load (accompanied by the message
> "Hardware FPU already installed") I'm pretty much forced to go with the
> response "Wrong" here.

Ah. I just checked on the web, and the PB 170 did ship with a 68882.

> > and 2)
> >something tried to execute an "fmul" instrustion in interrupt context.
>
> I'll have to take your word for that part. I have no idea.

The sources are browsable via ftp or cvsweb, e.g:
<http://cvsweb.netbsd.org/bsdweb.cgi/~checkout~/syssrc/sys/arch/mac68k/mac68k/trap.c
?rev=1.97&content-type=text/plain&only_with_tag=netbsd-1-6>

Your error is on line 394.

> >Another is a bug. "objdump" shows a lot of floating point instructions
> >in a GENERIC kernel, but they're all limited to the "FPSP" module
> >(which is for 68040's). One thing I'm still trying to understand:
> >".../mac68k/vectors.s" loads the FPSP functions into the vector table
> >if "options FPSP" is in the kernel config. How can this ever be
> >correct for a no-fpu-at-all machine?
>
> Now you're talking *WAY* beyond where I am :)

I didn't expect you to answer that. I would appreciate if anyone could
explain to me how a GENERIC kernel does the right thing on anything
but a 68040, because it looks to me as if the floating point exception
vectors are all loaded with the FPSP routines if 'options FPSP' is in
the kernel config.

Frederick