From: H.J. Lu (hlu@yoda.eecs.wsu.edu)
Date: 06/30/92


From: hlu@yoda.eecs.wsu.edu (H.J. Lu)
Subject: Re: Can your 386/387 or 486 handle 1.0/0.0?
Date: 1 Jul 1992 04:03:00 GMT

In article <CORYWEST.92Jun30204255@rio-grande.rice.edu> corywest@rice.edu (Cory West) writes:
>In article <__zlm3+.kehcheng@netcom.com> kehcheng@netcom.com
>(Keh-Cheng Chu) writes:
>
>> It seems that my 25 MHz 386/387 cannot handle floating point
>> exceptions correctly. Please try compiling the following code
>[etc...]
>

Some motherboards may fails the test. There was a story about it
in comp.unix.bsd. You may ask for a list of motherboards there.

> Under gcc 2.2.2 on a 486-33C, I get a Floating Point
>Exception and the core dumped.

That is because in crt0.s I used 0x0262(?) for control word of 387. We
can change it to mask out "Floating Point Exception", But I am not sure
if the stdio can handle "NaN". I think we will get it eventually. Is
that a big problem?

H.J.
=====
gcc/libc maintainer for Linux