At 2:04 AM 3/14/97, Mason Loring Bliss wrote:
>On 03/13/1997 at 8:37 PM -0800, you wrote:
>
>> So basically, for NetBSD to EVER run on an 040LC, we have to either compile
>> the silly thing with soft-float (sp?) or stick NOP's in front of all of
>> the floating point instructions?
>
>Well... This isn't the ideal programmatic approach, but I'd be in favour of
>the soft-float compile, with a notice about which versions of the chip are
>supported and which aren't. The SoftwareFPU documentation lists the number,
>making it an easy determination.
>
>On the other hand, supporting all 68LC040s is the ultimate goal... I'd
>think that the best solution (for now) would be to allow for either option
>when compiling. I don't know how difficult this would be to implement,
>however.

I got the impression from the SoftwareFPU doc's that it was only the first
released version of the chip that had the problem and that Motorola had
acknowledged it and would replace it if you could find your way through
channels.  Perhaps I'm reading too much into what they said, but it's
basically the same kind of situation as the famous Pentium FPU bug.  It's
complicated by the age of the chip.

I'm not opposed to software workarounds, but I don't think they should
clutter up the main distribution.  I would rather that people figure out
how to get someone to fix the problem properly by replacing the chip.  Then
we document the chip rev levels and who to contact for the replacement.  I
hope I'm not being too idealistic here.

Signature failed Preliminary Design Review.
Feasibility of a new signature is currently being evaluated.
h.b.hotz@jpl.nasa.gov, or hbhotz@oxy.edu