At 20:27 Uhr +0200 17.7.2004, Julio M. Merino Vidal wrote:
>To get this working, one needs to fix PR 26250, which this patch tries
>to, too.  This is what the changes to machdep.c and fpu.c do, but I'm not
>sure they are right (mainly because I've done them ;).  It works with a
>fpu-less computer, but I don't know what will happen with a full 68040 chip.

The changes you sent me work fine with the Duo 280. WRT the proper booting
sequence, they should be fine ("start messing with the fpu _after_ you've
detected its presence"); whether this meets the expectations of the
scheduler activation code is probably an issue for the tech-kern audience.

I am running a GENERICSBC build now to find out what my Quadra 700 thinks
of the change.

>Anyway, that code is bogus because the check fputype != FPU_NONE is
>always false at that point, due to fputype's default value.

Definitely. And while we're here, let's change the initial value of
'fputype' as set in locore.s to something that indicates 'uninitialised'.

	hauke

--
/~\  The ASCII Ribbon Campaign
\ /    No HTML/RTF in email
 X     No Word docs in email
/ \  Respect for open standards