Ken Nakata wrote:
> On Tue, 10 Mar 1998 09:06:52 -0800 (PST),
> Colin Wood <cwood@ichips.intel.com> writes:
> > 
> > That message is pretty normal, I think.  I believe that Ken compiled the
> > server using -current sources, so his libraries were a bit newer than
> > yours.
> 
> Oops, that's right.  My libc is libc.so.12.25.  But I used to get the
> similar error message with Colin's color server (my previous libc was
> 12.20).  It (or more precisely, ld.so) says it expects something like
> version > 12.21, but it also says "using it anyway".  So I figured
> minor version number wouldn't be much of a problem.  I may be wrong.

Oh, the versioning shouldn't cause any problems unless you happen to use
one of the symbols that changed.  The version of the color server I built
was probably with 1.3_BETA libraries, but I don't think there was a
version number bump between that and the 1.3 release libs.
 
> > My guess is that the 'LC040 is the problem,
> > tho, and if you try it a dozen times, it might work at least once.
> 
> And that is definitely not right... (I mean, "that" being FPE
> exhibiting such indeterministic behavior)  Is there any fundamental
> difference between kernel-mode programming model on 030s and 040s that
> I'm not aware of?

Ah....well, you would certainly know more about this than I, but the above
behavior is what I've seen described by a number of people on the list
(i.e. things work about 1 time out of 10).  I wouldn't expect
non-deterministic behavior, either, but perhaps something is happening
where a context switch occurs and FP state isn't being properly
saved/restored?  (I'm really grasping at straws here :-)

Later.

-- 
Colin Wood                                 cwood@ichips.intel.com
Component Design Engineer - MD6                 Intel Corporation
-----------------------------------------------------------------
I speak only on my own behalf, not for my employer.