[gnu.emacs.bug] floating pt problems on Sun-3/50 w.o. mc68881

wolfgang@MGM.MIT.EDU (Wolfgang Rupprecht) (04/18/89)

In GNU FloatEmacs 18.52.3 of Fri Apr 14 1989 on Sun3 (berkeley-unix)

It appears that adding floating point variables to GnuEmacs will
confuse the sun linker, and cause the _fp_save() and _fp_restore() to
be sucked in.  This seems to happen on matter what the setting of the
FLOAT_ENVIRONMENT or the 'cc -fsoft' compiler flag is telling it.

This normally doesn't matter on machines that have a real Mc68881;
it just causes the floating point chip to be used.  This is a real
problem on bottom-of-the-line Sun 3/50s.  In this case emacs loops 
endlessly (while catching all signals).  Running under gdb shows a 
SIGEMT being sent.

It appears that Sun's "official" crt0.o has some variables that may
influence which floating point option is actually linked in.  Does
anyone know what needs to be added to gnuemacs/src/crt0.c to achieve
this functionality?

-wolfgang
---
Wolfgang Rupprecht	ARPA:  wolfgang@mgm.mit.edu (IP 18.82.0.114)
TEL: (703) 768-2640	UUCP:  mit-eddie!mgm.mit.edu!wolfgang