From owner-freebsd-bugs  Thu Jun  8 17:01:28 1995
Return-Path: bugs-owner
Received: (from majordom@localhost)
          by freefall.cdrom.com (8.6.10/8.6.6) id RAA03319
          for bugs-outgoing; Thu, 8 Jun 1995 17:01:28 -0700
Received: from aristotle.algonet.se (root@aristotle.algonet.se [193.12.207.1])
          by freefall.cdrom.com (8.6.10/8.6.6) with ESMTP id RAA03312
          for <bugs@freebsd.org>; Thu, 8 Jun 1995 17:01:25 -0700
Received: from sophocles. (mal@sophocles.algonet.se [193.12.207.10]) by aristotle.algonet.se (8.6.9/hdw.1.0) with SMTP id BAA18970 for <bugs@freebsd.org>; Fri, 9 Jun 1995 01:59:09 +0200
Received: by sophocles. (5.x/SMI-SVR4)
	id AA06087; Fri, 9 Jun 1995 02:00:43 +0200
Date: Fri, 9 Jun 1995 02:00:43 +0200
From: mal@algonet.se (Mats Lofkvist)
Message-Id: <9506090000.AA06087@sophocles.>
To: bugs@FreeBSD.org
Subject: Bogus (?) SIGFPE's from divide by zero
Sender: bugs-owner@FreeBSD.org
Precedence: bulk

I don't know what the standards say, but other systems (at least SunOS)
do not issue floating point exceptions on divide by zero. Bug in FreeBSD?

Test program and outputs from FreeBSD and SunOS 4.1.4 below.

Mats Lofkvist

---testfpe.c---
#include <stdio.h>
#include <signal.h>

void handler(int foo)
{
  printf("handler\n");
}

double zero, one;

set()
{
  zero = 0.0;
  one = 1.0;
}

main()
{
  double foo;

  set();

  signal(SIGFPE, handler);

  printf("1.0/0.0\n");
  foo = 1.0/0.0;
  printf("%f\n", foo);

  printf("0.0/0.0\n");
  foo = 0.0/0.0;
  printf("%f\n", foo);

  printf("vars 1.0/0.0\n");
  foo = one/zero;
  printf("%f\n", foo);

  printf("vars 0.0/0.0\n");
  foo = zero/zero;
  printf("%f\n", foo);

  printf("done\n");
}

---

---FreBSD 2.0.5A:---
garm>a.out
1.0/0.0
Inf
0.0/0.0
NaN
vars 1.0/0.0
handler
1.000000
vars 0.0/0.0
handler
0.000000
done
---

---SunOS 4.1.4:---
mumrik>a.out
1.0/0.0
Inf
0.0/0.0
NaN
vars 1.0/0.0
Inf
vars 0.0/0.0
NaN
done
---

(The constant divisions are replaced by constants by gcc, I suppose)