From: ceriel <none@none>
Date: Tue, 25 Jul 1989 14:21:09 +0000 (+0000)
Subject: many changes, to use IEEE format
X-Git-Tag: release-5-5~2325
X-Git-Url: https://git.ndcode.org/public/gitweb.cgi?a=commitdiff_plain;h=f7360f0fec6e8b77940e37566c34532af288e686;p=ack.git

many changes, to use IEEE format
---

diff --git a/mach/proto/fp/FP_bias.h b/mach/proto/fp/FP_bias.h
index fae2987ef..db17a4132 100644
--- a/mach/proto/fp/FP_bias.h
+++ b/mach/proto/fp/FP_bias.h
@@ -9,7 +9,6 @@
 	include file for floating point package
 */
 
-#ifdef	IEEEFORMAT
 		/*	FLOAT FORMAT EXPONENT BIAS	*/
 
 #define	SGL_BIAS	 127	/* excess  128 notation used	*/
@@ -21,29 +20,9 @@
 		/*	VARIOUS MAX AND MIN VALUES	*/
 		/*	1) FOR THE DIFFERENT FORMATS	*/
 
-#define	SGL_MAX		   255	/*	standard definition	*/
+#define	SGL_MAX		   254	/*	standard definition	*/
 #define	SGL_MIN		     1	/*	standard definition	*/
-#define	DBL_MAX		  2047	/*	standard definition	*/
+#define	DBL_MAX		  2046	/*	standard definition	*/
 #define	DBL_MIN		     1	/*	standard definition	*/
-#define EXT_MAX		 16384	/*	standard minimum	*/
-#define EXT_MIN		-16383	/*	standard minimum	*/
-#else
-
-		/*	FLOAT FORMAT EXPONENT BIAS	*/
-
-#define	SGL_BIAS	 127	/* excess  128 notation used	*/
-#define	DBL_BIAS	127	/* excess 128 notation used	*/
-#define	EXT_BIAS	   0	/* 2s-complement notation used	*/
-				/* this is possible because the	*/
-				/* sign is in a seperate word	*/
-		
-		/*	VARIOUS MAX AND MIN VALUES	*/
-		/*	1) FOR THE DIFFERENT FORMATS	*/
-
-#define	SGL_MAX		   255	/*	standard definition	*/
-#define	SGL_MIN		     1	/*	standard definition	*/
-#define	DBL_MAX		  255	/*	standard definition	*/
-#define	DBL_MIN		     1	/*	standard definition	*/
-#define EXT_MAX		 16384	/*	standard minimum	*/
-#define EXT_MIN		-16383	/*	standard minimum	*/
-#endif
+#define EXT_MAX		 16383	/*	standard minimum	*/
+#define EXT_MIN		-16382	/*	standard minimum	*/
diff --git a/mach/proto/fp/FP_shift.h b/mach/proto/fp/FP_shift.h
index f713b881e..5b685636f 100644
--- a/mach/proto/fp/FP_shift.h
+++ b/mach/proto/fp/FP_shift.h
@@ -26,32 +26,13 @@
 #define	SGL_MASK	0x007fffffL
 
 				/* parameters for Double Precision */
-#ifndef	IEEEFORMAT
-
-#define DBL_EXPSHIFT	SGL_EXPSHIFT
-#define DBL_M1LEFT	SGL_M1LEFT
-
-#define	DBL_LPACK	DBL_RUNPACK
-#define	DBL_RPACK	DBL_LUNPACK
-
-#define DBL_ZERO	SGL_ZERO
-#define DBL_EXACT	SGL_EXACT
-
-#define DBL_RUNPACK	DBL_M1LEFT
-#define DBL_LUNPACK	32-DBL_M1LEFT
-
-#define DBL_ROUNDUP	SGL_ROUNDUP
-#define	DBL_CARRYOUT	SGL_CARRYOUT
-#define	DBL_MASK	SGL_MASK
-
-#else
 				/* used in extend.c */
 
 #define DBL_EXPSHIFT	4
 
 #define DBL_M1LEFT	11
 
-#define	DBL_RPACK	32-DBL_M1LEFT
+#define	DBL_RPACK	(32-DBL_M1LEFT)
 #define	DBL_LPACK	DBL_M1LEFT
 
 				/* used in compact.c */
@@ -61,10 +42,8 @@
 #define DBL_EXACT	0x7ff
 
 #define DBL_RUNPACK	DBL_M1LEFT
-#define DBL_LUNPACK	32-DBL_RUNPACK
+#define DBL_LUNPACK	(32-DBL_RUNPACK)
 
 #define DBL_ROUNDUP	0x400
 #define	DBL_CARRYOUT	0x00200000L
 #define	DBL_MASK	0x000fffffL
-
-#endif	IEEEFORMAT
diff --git a/mach/proto/fp/Makefile b/mach/proto/fp/Makefile
index d8ae5cc3d..b83adde88 100644
--- a/mach/proto/fp/Makefile
+++ b/mach/proto/fp/Makefile
@@ -8,47 +8,43 @@ CFLAGS=
 #
 #	$Header$
 #
-# various flags that can be used during compilation
-# define DEBUG
-# define PRT_ADD
-# define PRT_ALL
-# define PRT_DBL
-# define PRT_ENTRY
-# define PRT_EXIT
-# define PRT_EXT
-# define PRT_EXT2
-# define PRT_LONG
-# define PRT_RNDMSG
-# define PRT_STDERR
-# define PRT_TRAP
-#
-# DFLAGS=-DPRT_ADD -DPRT_ALL -DPRT_DBL -DPRT_ENTRY -DPRT_EXIT -DPRT_EXT -DPRT_EXT2 -DPRT_LONG -DPRT_RNDMSG -DPRT_STDERR -DPRT_TRAP
-DFLAGS=
-EMFLAGS= -L -LIB -I. $(DFLAGS) -O $(CFLAGS)
+EMFLAGS= -L -LIB -I. -O $(CFLAGS)
 #	AS=ack -m$(MACH) -c.$(SUF)
 #	CC=ack -m$(MACH) -c.s
 #	CCFLAGS=$(EMFLAGS)
 CDIR=$(EMHOME)/mach/proto/fp
 
-LIST =		cff4.$(SUF) cff8.$(SUF) cfu.$(SUF) cmf4.$(SUF) cmf8.$(SUF)\
+LIST =		cff4.$(SUF) cff8.$(SUF)\
+		cfu.$(SUF)\
+		cmf4.$(SUF) cmf8.$(SUF)\
 		cuf4.$(SUF) cuf8.$(SUF)\
-		dvf4.$(SUF) dvf8.$(SUF) fef4.$(SUF) fef8.$(SUF)\
+		dvf4.$(SUF) dvf8.$(SUF)\
+		fef4.$(SUF) fef8.$(SUF)\
 		fif4.$(SUF) fif8.$(SUF)\
-		cfi.$(SUF) cif4.$(SUF) cif8.$(SUF) mlf4.$(SUF) mlf8.$(SUF)\
-		ngf4.$(SUF)\
-		ngf8.$(SUF) sbf4.$(SUF) sbf8.$(SUF) adf4.$(SUF) adf8.$(SUF)\
+		cfi.$(SUF)\
+		cif4.$(SUF) cif8.$(SUF)\
+		mlf4.$(SUF) mlf8.$(SUF)\
+		ngf4.$(SUF) ngf8.$(SUF)\
+		sbf4.$(SUF) sbf8.$(SUF)\
+		adf4.$(SUF) adf8.$(SUF)\
 		zrf4.$(SUF) zrf8.$(SUF)\
 		extend.$(SUF) compact.$(SUF)\
 		add_ext.$(SUF) div_ext.$(SUF) mul_ext.$(SUF) nrm_ext.$(SUF)\
 		sft_ext.$(SUF) sub_ext.$(SUF) zrf_ext.$(SUF)\
 		adder.$(SUF) shifter.$(SUF) fptrp.$(SUF)
-SLIST =		cff4.s cff8.s cfu.s cmf4.s cmf8.s\
+SLIST =		cff4.s cff8.s\
+		cfu.s\
+		cmf4.s cmf8.s\
 		cuf4.s cuf8.s\
-		dvf4.s dvf8.s fef4.s fef8.s\
+		dvf4.s dvf8.s\
+		fef4.s fef8.s\
 		fif4.s fif8.s\
-		cfi.s cif4.s cif8.s mlf4.s mlf8.s\
-		ngf4.s\
-		ngf8.s sbf4.s sbf8.s adf4.s adf8.s\
+		cfi.s\
+		cif4.s cif8.s\
+		mlf4.s mlf8.s\
+		ngf4.s ngf8.s\
+		sbf4.s sbf8.s\
+		adf4.s adf8.s\
 		zrf4.s zrf8.s\
 		extend.s compact.s\
 		add_ext.s div_ext.s mul_ext.s nrm_ext.s\
@@ -56,13 +52,19 @@ SLIST =		cff4.s cff8.s cfu.s cmf4.s cmf8.s\
 		adder.s shifter.s fptrp.s
 
 SRC =		FP_bias.h FP_shift.h FP_trap.h FP_types.h adder.h get_put.h\
-		cff4.c cff8.c cfu.c cmf4.c cmf8.c\
+		cff4.c cff8.c\
+		cfu.c\
+		cmf4.c cmf8.c\
 		cuf4.c cuf8.c\
-		dvf4.c dvf8.c fef4.c fef8.c\
+		dvf4.c dvf8.c\
+		fef4.c fef8.c\
 		fif4.c fif8.c\
-		cfi.c cif4.c cif8.c mlf4.c mlf8.c\
-		ngf4.c\
-		ngf8.c sbf4.c sbf8.c adf4.c adf8.c\
+		cfi.c\
+		cif4.c cif8.c\
+		mlf4.c mlf8.c\
+		ngf4.c ngf8.c\
+		sbf4.c sbf8.c\
+		adf4.c adf8.c\
 		zrf4.c zrf8.c\
 		extend.c compact.c\
 		add_ext.c div_ext.c mul_ext.c nrm_ext.c\
diff --git a/mach/proto/fp/add_ext.c b/mach/proto/fp/add_ext.c
index 07a55c43d..43f15656e 100644
--- a/mach/proto/fp/add_ext.c
+++ b/mach/proto/fp/add_ext.c
@@ -27,12 +27,15 @@ register EXTEND	*e1,*e2;
 		if (e2->m1 > e1->m1 ||
                     (e2->m1 == e1->m1 && e2->m2 > e1->m2)) {
                 	/*      abs(e2) > abs(e1) */
-                	if (e1->m2 > e2->m2) {
-                        	e2->m1 -= 1;    /* carry in */
+			EXTEND x;
+
+			x = *e1;
+			*e1 = *e2;
+                	if (x.m2 > e1->m2) {
+                        	e1->m1 -= 1;    /* carry in */
                 	}
-                	e2->m1 -= e1->m1;
-                	e2->m2 -= e1->m2;
-                	*e1 = *e2;
+                	e1->m1 -= x.m1;
+                	e1->m2 -= x.m2;
         	}
         	else {
                 	if (e2->m2 > e1->m2)
diff --git a/mach/proto/fp/adf4.c b/mach/proto/fp/adf4.c
index cb07cf452..c59b67399 100644
--- a/mach/proto/fp/adf4.c
+++ b/mach/proto/fp/adf4.c
@@ -29,5 +29,5 @@ _float	s1,s2;
 	extend((_double *)&s2,&e2,sizeof(SINGLE));
 	add_ext(&e1,&e2);
 	compact(&e1,(_double *)&s1,sizeof(SINGLE));
-	return(s1);
+	return s1;
 }
diff --git a/mach/proto/fp/adf8.c b/mach/proto/fp/adf8.c
index 540be95a3..6519f844c 100644
--- a/mach/proto/fp/adf8.c
+++ b/mach/proto/fp/adf8.c
@@ -29,5 +29,5 @@ _double	s1,s2;
 	extend(&s2,&e2,sizeof(_double));
 	add_ext(&e1,&e2);
 	compact(&e1,&s1,sizeof(_double));
-	return(s1);
+	return s1;
 }
diff --git a/mach/proto/fp/cff4.c b/mach/proto/fp/cff4.c
index d30a10509..d82ba6a72 100644
--- a/mach/proto/fp/cff4.c
+++ b/mach/proto/fp/cff4.c
@@ -17,6 +17,7 @@
 
 #include	"FP_types.h"
 
+_float
 cff4(src)
 _double	src;	/* the source itself -	THIS TIME it's DOUBLE */
 {
@@ -24,4 +25,5 @@ _double	src;	/* the source itself -	THIS TIME it's DOUBLE */
 
 	extend(&src,&buf,8);	/* no matter what */
 	compact(&buf,(_double *) &(src.__double[1]),4);
+	return *(_float *)&(src.__double[1]);
 }
diff --git a/mach/proto/fp/cff8.c b/mach/proto/fp/cff8.c
index 737d0346f..4e59d7802 100644
--- a/mach/proto/fp/cff8.c
+++ b/mach/proto/fp/cff8.c
@@ -17,11 +17,13 @@
 
 #include "FP_types.h"
 
+_double
 cff8(src)
-_float	src;	/* the space on the stack is for a double - see cg/table */
+_float	src;
 {
 	EXTEND	buf;
 
 	extend((_double *) &src,&buf,4);	/* no matter what */
-	compact(&buf,&src,8);
+	compact(&buf,(_double *) &src,8);
+	return *(_double *) &src;
 }
diff --git a/mach/proto/fp/cfi.c b/mach/proto/fp/cfi.c
index 18a9c46f4..12d6cde51 100644
--- a/mach/proto/fp/cfi.c
+++ b/mach/proto/fp/cfi.c
@@ -15,6 +15,7 @@
 
 #include "FP_trap.h"
 #include "FP_types.h"
+#include "FP_shift.h"
 
 long
 cfi(ds,ss,src)
@@ -24,21 +25,25 @@ _double	src;	/* assume worst case */
 {
 	EXTEND	buf;
 	long	new;
-	short	newint,	max_exp;
+	short	max_exp;
 
 	extend(&src,&buf,ss);	/* get extended format */
-	buf.exp--;		/* additional bias correction */
-	if (buf.exp < 1) {	/* no conversion needed */
+	if (buf.exp < 0) {	/* no conversion needed */
 		src.__double[ss == 8] = 0L;
 		return(0L);
 	}
-	max_exp = (ds << 3) - 1;	/* signed numbers */
+	max_exp = (ds << 3) - 2;	/* signed numbers */
 				/* have more limited max_exp */
 	if (buf.exp > max_exp) {
-		trap(EIOVFL);	/* integer overflow	*/
-		buf.exp %= max_exp; /* truncate	*/
+		if (buf.exp == max_exp+1 && buf.sign && buf.m1 == NORMBIT &&
+		    buf.m2 == 0L) {
+		}
+		else {
+			trap(EIOVFL);	/* integer overflow	*/
+			buf.exp %= max_exp; /* truncate	*/
+		}
 	}
-	new = buf.m1 >> (32-buf.exp);
+	new = buf.m1 >> (31-buf.exp);
 	if (buf.sign)
 		new = -new;
 done:
diff --git a/mach/proto/fp/cfu.c b/mach/proto/fp/cfu.c
index 8cb2bc8a7..ba36fa33d 100644
--- a/mach/proto/fp/cfu.c
+++ b/mach/proto/fp/cfu.c
@@ -27,17 +27,16 @@ _double	src;	/* assume worst case */
 	short	newint, max_exp;
 
 	extend(&src,&buf,ss);	/* get extended format	*/
-	buf.exp--;		/* additional bias correction */
-	if (buf.exp < 1) {	/* no conversion needed	*/
+	if (buf.exp < 0) {	/* no conversion needed	*/
 		src.__double[ss == 8] = 0L;
 		return(0L);
 	}
-	max_exp = (ds << 3);
+	max_exp = (ds << 3) - 1;
 	if (buf.exp > max_exp) {
 		trap(EIOVFL);	/* integer overflow	*/
 		buf.exp %= max_exp;
 	}
-	new = buf.m1 >> (32-buf.exp);
+	new = buf.m1 >> (31-buf.exp);
 done:
 	src.__double[ss == 8] = new;
 	return(new);
diff --git a/mach/proto/fp/cif4.c b/mach/proto/fp/cif4.c
index cec5c9c81..19d78c546 100644
--- a/mach/proto/fp/cif4.c
+++ b/mach/proto/fp/cif4.c
@@ -28,14 +28,14 @@ long	src;	/* largest possible integer to convert */
 
 	zrf_ext(&buf);
 	if (ss == sizeof(long))	{
-		buf.exp = 33;
+		buf.exp = 31;
 		i_src = src;
 		result = (_float *) &src;
 	}
 	else	{
 		ipt = (short *) &src;
 		i_src = (long) *ipt;
-		buf.exp = 17;
+		buf.exp = 15;
 		result = (_float *) &ss;
 	}
 	if (i_src == 0)	{
diff --git a/mach/proto/fp/cif8.c b/mach/proto/fp/cif8.c
index 374ae823f..545e17dae 100644
--- a/mach/proto/fp/cif8.c
+++ b/mach/proto/fp/cif8.c
@@ -29,13 +29,13 @@ long	src;	/* largest possible integer to convert */
 	result = (_double *) &ss;	/* always */
 	zrf_ext(&buf);
 	if (ss == sizeof(long))	{
-		buf.exp = 33;
+		buf.exp = 31;
 		i_src = src;
 	}
 	else	{
 		ipt = (short *) &src;
 		i_src = (long) *ipt;
-		buf.exp = 17;
+		buf.exp = 15;
 	}
 	if (i_src == 0)	{
 		zrf8(result);
diff --git a/mach/proto/fp/cmf4.c b/mach/proto/fp/cmf4.c
index b04ca6b30..0611e332b 100644
--- a/mach/proto/fp/cmf4.c
+++ b/mach/proto/fp/cmf4.c
@@ -12,6 +12,7 @@
 #include	"FP_types.h"
 #include	"get_put.h"
 
+int
 cmf4(f1,f2)
 _float	f1,f2;
 {
diff --git a/mach/proto/fp/cmf8.c b/mach/proto/fp/cmf8.c
index 85b563966..8fa51c53d 100644
--- a/mach/proto/fp/cmf8.c
+++ b/mach/proto/fp/cmf8.c
@@ -34,14 +34,13 @@ _double	d1,d2;
 		rv = l1 < l2 ? 1 : -1;
 	}
 	else	{ 		/* decide in 2nd half */
-		l1 = get4(((char *)&d1 + 4));
-		l2 = get4(((char *)&d2 + 4));
-		if (l1 == l2)
+		unsigned long u1, u2;
+		u1 = get4(((char *)&d1 + 4));
+		u2 = get4(((char *)&d2 + 4));
+		if (u1 == u2)
 			return(0);
-		if (l1 >= 0)
-			rv = l1 < l2 || l2 < 0 ? 1 : -1;
-		else if (l2 >= 0) rv = -1;
-		else rv = l1 < l2 ? 1 : -1;
+		if (u1 < u2) rv = 1;
+		else rv = -1;
 	}
 	return sign1 * rv;
 }
diff --git a/mach/proto/fp/compact.c b/mach/proto/fp/compact.c
index ffc42f29e..34b8fd579 100644
--- a/mach/proto/fp/compact.c
+++ b/mach/proto/fp/compact.c
@@ -35,18 +35,19 @@ int	size;
 		f->exp += DBL_BIAS;	/* restore proper bias	*/
 		if (f->exp > DBL_MAX)	{
 dbl_over:			trap(EFOVFL);
-			f->exp = DBL_MAX;
-			f->m1 = 0xffffffffL;
-			f->m2 = DBL_ZERO;
+			f->exp = DBL_MAX+1;
+			f->m1 = 0;
+			f->m2 = 0;
 			if (error++)
 				return;
 		}
 		else if (f->exp < DBL_MIN)	{
-			trap(EFUNFL);
-			f->exp = DBL_MIN;
-			f->m1 = f->m2 = 0L;
-			if (error++)
-				return;
+			b64_rsft(&(f->m1));
+			if (f->exp < 0) {
+				b64_sft(&(f->m1), -f->exp);
+				f->exp = 0;
+			}
+			/* underflow ??? */
 		}
 			
 		/* local CAST conversion		*/
@@ -121,18 +122,19 @@ dbl_over:			trap(EFOVFL);
 		f->exp += SGL_BIAS;	/* restore bias	*/
 		if (f->exp > SGL_MAX)	{
 sgl_over:			trap(EFOVFL);
-			f->exp = SGL_MAX;
-			f->m1 = SGL_ZERO;
+			f->exp = SGL_MAX+1;
+			f->m1 = 0L;
 			f->m2 = 0L;
 			if (error++)
 				return;
 		}
 		else if (f->exp < SGL_MIN)	{
-			trap(EFUNFL);
-			f->exp = SGL_MIN;
-			f->m1 = f->m2 = 0L;
-			if (error++)
-				return;
+			b64_rsft(&(f->m1));
+			if (f->exp < 0) {
+				b64_sft(&(f->m1), -f->exp);
+				f->exp = 0;
+			}
+			/* underflow ??? */
 		}
 
 		/* shift mantissa and store	*/
diff --git a/mach/proto/fp/cuf4.c b/mach/proto/fp/cuf4.c
index 891ac169d..a83d10126 100644
--- a/mach/proto/fp/cuf4.c
+++ b/mach/proto/fp/cuf4.c
@@ -28,14 +28,14 @@ long	src;	/* largest possible integer to convert */
 
 	zrf_ext(&buf);
 	if (ss == sizeof(long))	{
-		buf.exp = 33;
+		buf.exp = 31;
 		i_src = src;
 		result = (_float *) &src;
 	}
 	else	{
 		ipt = (short *) &src;
 		i_src = (long) *ipt;
-		buf.exp = 17;
+		buf.exp = 15;
 		result = (_float *) &ss;
 	}
 	if (i_src == 0)	{
diff --git a/mach/proto/fp/cuf8.c b/mach/proto/fp/cuf8.c
index 619ab5e2e..d4da6b997 100644
--- a/mach/proto/fp/cuf8.c
+++ b/mach/proto/fp/cuf8.c
@@ -16,6 +16,7 @@
 
 #include "FP_types.h"
 
+_double
 cuf8(ss,src)
 int	ss;	/* source size */
 long	src;	/* largest possible integer to convert */
@@ -26,13 +27,13 @@ long	src;	/* largest possible integer to convert */
 
 	zrf_ext(&buf);
 	if (ss == sizeof(long))	{
-		buf.exp = 33;
+		buf.exp = 31;
 		i_src = src;
 	}
 	else	{
 		ipt = (short *) &src;
 		i_src = (long) *ipt;
-		buf.exp = 17;
+		buf.exp = 15;
 	}
 	if (i_src == 0)	{
 		zrf8(&ss);
@@ -50,4 +51,5 @@ long	src;	/* largest possible integer to convert */
 			/* adjust mantissa field	*/
 	nrm_ext(&buf);
 	compact(&buf,(_double *) &ss,8);
+	return *((_double *) &ss);
 }
diff --git a/mach/proto/fp/div_ext.c b/mach/proto/fp/div_ext.c
index c58be1781..9f135c6b5 100644
--- a/mach/proto/fp/div_ext.c
+++ b/mach/proto/fp/div_ext.c
@@ -68,25 +68,6 @@ EXTEND	*e1,*e2;
 	/*	check for underflow, divide by zero, etc	*/
 	e1->sign ^= e2->sign;
 	e1->exp -= e2->exp;
-	e1->exp += 2;		/* bias correction	*/
-	if (e1->exp < EXT_MIN)	{
-		/*
-		 * Exception 8.4 - Underflow
-		 */
-		trap(EFUNFL);	/* underflow */
-		e1->exp = EXT_MIN;
-		e1->m1 = e1->m2 = 0L;
-		return;
-	}
-	if (e1->exp >= EXT_MAX) {
-                /*
-                 * Exception 8.3 - Overflow
-                 */
-                trap(EFOVFL);   /* overflow */
-                e1->exp = EXT_MAX;
-                e1->m1 = e1->m2 = 0L;
-                return;
-        }
 
 #ifndef USE_DIVIDE
 		/* do division of mantissas	*/
@@ -264,4 +245,22 @@ EXTEND	*e1,*e2;
 	e1->m2 = result[1];
 
 	nrm_ext(e1);
+	if (e1->exp < EXT_MIN)	{
+		/*
+		 * Exception 8.4 - Underflow
+		 */
+		trap(EFUNFL);	/* underflow */
+		e1->exp = EXT_MIN;
+		e1->m1 = e1->m2 = 0L;
+		return;
+	}
+	if (e1->exp >= EXT_MAX) {
+                /*
+                 * Exception 8.3 - Overflow
+                 */
+                trap(EFOVFL);   /* overflow */
+                e1->exp = EXT_MAX;
+                e1->m1 = e1->m2 = 0L;
+                return;
+        }
 }
diff --git a/mach/proto/fp/dvf4.c b/mach/proto/fp/dvf4.c
index 0c11e61e5..9569d42a3 100644
--- a/mach/proto/fp/dvf4.c
+++ b/mach/proto/fp/dvf4.c
@@ -11,6 +11,7 @@
 
 #include	"FP_types.h"
 
+_float
 dvf4(s2,s1)
 _float	s1,s2;
 {
@@ -22,4 +23,5 @@ _float	s1,s2;
 		/* do a divide */
 	div_ext(&e1,&e2);
 	compact(&e1,(_double *)&s1,sizeof(_float));
+	return s1;
 }
diff --git a/mach/proto/fp/dvf8.c b/mach/proto/fp/dvf8.c
index 0a40c075e..4891796e9 100644
--- a/mach/proto/fp/dvf8.c
+++ b/mach/proto/fp/dvf8.c
@@ -11,6 +11,7 @@
 
 #include	"FP_types.h"
 
+_double
 dvf8(s2,s1)
 _double	s1,s2;
 {
@@ -22,4 +23,5 @@ _double	s1,s2;
 		/* do a divide */
 	div_ext(&e1,&e2);
 	compact(&e1,&s1,sizeof(_double));
+	return s1;
 }
diff --git a/mach/proto/fp/extend.c b/mach/proto/fp/extend.c
index 9f1a40bd6..84e1559fc 100644
--- a/mach/proto/fp/extend.c
+++ b/mach/proto/fp/extend.c
@@ -80,6 +80,8 @@ zero:			zrf_ext(to);
 	}
 
 	to->m1 |= NORMBIT;				/* set bit L	*/
-	if (leadbit == 0)		/* set or clear Leading Bit	*/
+	if (leadbit == 0) {		/* set or clear Leading Bit	*/
 		to->m1 &= ~NORMBIT;			/* clear bit L	*/
+		nrm_ext(to);				/* and normalize */
+	}
 }
diff --git a/mach/proto/fp/fef4.c b/mach/proto/fp/fef4.c
index 6552ed584..b781507b8 100644
--- a/mach/proto/fp/fef4.c
+++ b/mach/proto/fp/fef4.c
@@ -16,14 +16,17 @@ struct	fef4_returns {
 	_float	f;
 };
 
-fef4(s1)
+fef4(r,s1)
 _float	s1;
+struct fef4_returns	*r;
 {
-	struct	fef4_returns	*r = (struct fef4_returns *) &s1;
 	EXTEND	buf;
+	register struct fef4_returns	*p = r;	/* make copy; r might refer
+						   to itself (see table)
+						*/
 
 	extend((_double *) &s1,&buf,sizeof(_float));
-	r->e = buf.exp-1;
-	buf.exp = 1;
-	compact(&buf,(_double *) &r->f,sizeof(_float));
+	p->e = buf.exp+1;
+	buf.exp = -1;
+	compact(&buf,(_double *) &p->f,sizeof(_float));
 }
diff --git a/mach/proto/fp/fef8.c b/mach/proto/fp/fef8.c
index f4d7da861..e261411ba 100644
--- a/mach/proto/fp/fef8.c
+++ b/mach/proto/fp/fef8.c
@@ -16,14 +16,17 @@ struct	fef8_returns	{
 	_double	f;
 };
 
-fef8(s1)
+fef8(r, s1)
 _double	s1;
+struct fef8_returns *r;
 {
 	EXTEND	buf;
-	struct fef8_returns *r = (struct fef8_returns *) &s1;
+	register struct fef8_returns *p = r;	/* make copy, r might refer
+						   to itself (see table)
+						*/
 
 	extend(&s1,&buf,sizeof(_double));
-	r->e = buf.exp - 1;
-	buf.exp = 1;
-	compact(&buf,&r->f,sizeof(_double));
+	p->e = buf.exp + 1;
+	buf.exp = -1;
+	compact(&buf,&p->f,sizeof(_double));
 }
diff --git a/mach/proto/fp/fif4.c b/mach/proto/fp/fif4.c
index 1142dbb73..aa8a3d12f 100644
--- a/mach/proto/fp/fif4.c
+++ b/mach/proto/fp/fif4.c
@@ -14,8 +14,14 @@
 
 _float sbf4();
 
-fif4(x,y)
+struct fif4_returns {
+	_float ipart;
+	_float fpart;
+};
+
+fif4(p,x,y)
 _float	x,y;
+struct fif4_returns *p;
 {
 
 	EXTEND	e1,e2;
@@ -26,19 +32,18 @@ _float	x,y;
 	mul_ext(&e1,&e2);
 	e2 = e1;
 	compact(&e2, (_double *)&y, sizeof(_float));
-	e1.exp--;			/* additional bias correction */
-	if (e1.exp < 1) {
-		x = 0;
+	if (e1.exp < 0) {
+		p->ipart = 0;
+		p->fpart = y;
 		return;
 	}
-	if (e1.exp > 31 - SGL_M1LEFT) {
-		x = y;
-		y = 0;
+	if (e1.exp > 30 - SGL_M1LEFT) {
+		p->ipart = y;
+		p->fpart = 0;
 		return;
 	}
 	b64_sft(&e1.m1, 64 - e1.exp);
 	b64_sft(&e1.m1, e1.exp - 64);	/* "loose" low order bits */
-	e1.exp++;
-	compact(&e1,(_double *) &x, sizeof(SINGLE));
-	y = sbf4(x, y);
+	compact(&e1,(_double *) &(p->ipart), sizeof(SINGLE));
+	p->fpart = sbf4(p->ipart, y);
 }
diff --git a/mach/proto/fp/fif8.c b/mach/proto/fp/fif8.c
index 6a938cc2e..40ed77534 100644
--- a/mach/proto/fp/fif8.c
+++ b/mach/proto/fp/fif8.c
@@ -14,8 +14,14 @@
 
 _double sbf8();
 
-fif8(x,y)
+struct fif8_returns {
+	_double	ipart;
+	_double fpart;
+};
+
+fif8(p,x,y)
 _double	x,y;
+struct fif8_returns *p;
 {
 
 	EXTEND	e1,e2;
@@ -26,22 +32,20 @@ _double	x,y;
 	mul_ext(&e1,&e2);
 	e2 = e1;
 	compact(&e2, &y, sizeof(_double));
-	e1.exp--;			/* additional bias correction */
-	if (e1.exp < 1) {
-		x.__double[0] = 0;
-		x.__double[1] = 0;
+	if (e1.exp < 0) {
+		p->ipart.__double[0] = 0;
+		p->ipart.__double[1] = 0;
+		p->fpart = y;
 		return;
 	}
-	if (e1.exp > 63 - DBL_M1LEFT) {
-		x.__double[0] = y.__double[0];
-		x.__double[1] = y.__double[1];
-		y.__double[0] = 0;
-		y.__double[1] = 0;
+	if (e1.exp > 62 - DBL_M1LEFT) {
+		p->ipart = y;
+		p->fpart.__double[0] = 0;
+		p->fpart.__double[1] = 0;
 		return;
 	}
 	b64_sft(&e1.m1, 64 - e1.exp);
 	b64_sft(&e1.m1, e1.exp - 64);	/* "loose" low order bits */
-	e1.exp++;
-	compact(&e1, &x, sizeof(DOUBLE));
-	y = sbf8(x, y);
+	compact(&e1, &(p->ipart), sizeof(DOUBLE));
+	p->fpart = sbf8(p->ipart, y);
 }
diff --git a/mach/proto/fp/mul_ext.c b/mach/proto/fp/mul_ext.c
index 3d30ffcd1..9a5d507a4 100644
--- a/mach/proto/fp/mul_ext.c
+++ b/mach/proto/fp/mul_ext.c
@@ -13,69 +13,22 @@
 # include "FP_bias.h"
 # include "FP_trap.h"
 # include "FP_types.h"
+# include "FP_shift.h"
 
 mul_ext(e1,e2)
 EXTEND	*e1,*e2;
 {
 	register int	i,j;		/* loop control	*/
-	short unsigned	mp[4];	/* multiplier */
-	short unsigned	mc[4];	/* multipcand */
-	short unsigned	result[8];	/* result */
-	B64		tmp64;
+	unsigned short	mp[4];		/* multiplier */
+	unsigned short	mc[4];		/* multipcand */
+	unsigned short	result[8];	/* result */
 	register unsigned short *pres;
 
 	/* first save the sign (XOR)			*/
-
 	e1->sign ^= e2->sign;
 
-	/********************************************************/
-	/* 	 	INCREASE EXPONENT BY ONE (1)	    	*/
-	/*							*/
-	/* the nature of the multiplication algorithm used	*/
-	/* results in an exponent that is small by an additive	*/
-	/* factor of one (1);					*/
-	/* if the maximum bit is set it will not be subtracted	*/
-	/* during normalization -> this is correct and can be	*/
-	/* expected often with normalized numbers		*/
-	/*	HOWEVER, it is also possible that unnormalized	*/
-	/*	numbers are used. Rather than shifting here	*/
-	/*	always(!) (unless L bit is set) I chose to	*/
-	/*	increase the exponent by one - a simple (FAST)	*/
-	/*	process - and to decrease it later during	*/
-	/*	normalization.					*/
-	/*							*/
-	/********************************************************/
-	/* The effects of bias (as used here)			*/
-	/* and the multiplication algorithm used cancel		*/
-	/* so these statements are commented out		*/
-	/* August 1985 - if changing the Leading Bit (or NORMBIT) */
-	/* this problem with the multiplication algorithm no longer */
-	/* exists - bias must be subtracted now			*/
-	/*							*/
-	/* e1->exp++;						*/
-	/********************************************************/
-
-	/* next add the exponents			*/
-
-	e1->exp += e2->exp;
-	e1->exp -= 1;			/* correction for bias	*/
-
-					/* check for overflow	*/
-	if (e1->exp >= EXT_MAX)	{
-		trap(EFOVFL);
-			/* if caught 			*/
-			/* return signed infinity	*/
-		e1->exp = EXT_MAX;
-infinity:	e1->m1 = e1->m2 =0L;
-		return;
-	}
-				/* check for underflow	*/
-	if (e1->exp < EXT_MIN)	{
-		trap(EFUNFL);
-		e1->exp = EXT_MIN;
-		goto infinity;
-	}
-
+	/* compute new exponent */
+	e1->exp += e2->exp + 1;
 	/* 128 bit multiply of mantissas			*/
 
 		/* assign unknown long formats		*/
@@ -105,7 +58,15 @@ infinity:	e1->m1 = e1->m2 =0L;
 		}
 		pres[-1] = k;
 	}
-	
+        if (! (result[0] & 0x8000)) {
+                e1->exp--;
+                for (i = 0; i <= 3; i++) {
+                        result[i] <<= 1;
+                        if (result[i+1]&0x8000) result[i] |= 1;
+                }
+                result[4] <<= 1;
+        }
+
 	/*
 	 *	combine the registers to a total
 	 */
@@ -113,8 +74,25 @@ infinity:	e1->m1 = e1->m2 =0L;
 	e1->m2 = ((unsigned long)(result[2]) << 16) + result[3];
 	if (result[4] & 0x8000) {
 		if (++e1->m2 == 0)
-			e1->m1++;
+			if (++e1->m1 == 0) {
+				e1->m1 = NORMBIT;
+				e1->exp++;
+			}
 	}
 
-	nrm_ext(e1);
+					/* check for overflow	*/
+	if (e1->exp >= EXT_MAX)	{
+		trap(EFOVFL);
+			/* if caught 			*/
+			/* return signed infinity	*/
+		e1->exp = EXT_MAX;
+infinity:	e1->m1 = e1->m2 =0L;
+		return;
+	}
+				/* check for underflow	*/
+	if (e1->exp < EXT_MIN)	{
+		trap(EFUNFL);
+		e1->exp = EXT_MIN;
+		goto infinity;
+	}
 }
diff --git a/mach/proto/fp/ngf4.c b/mach/proto/fp/ngf4.c
index 2ef0abaa6..f8633eb83 100644
--- a/mach/proto/fp/ngf4.c
+++ b/mach/proto/fp/ngf4.c
@@ -16,13 +16,15 @@
 
 #include "FP_types.h"
 
+_float
 ngf4(f)
 _float	f;
 {
-	char unsigned	*p;
+	unsigned char *p;
 
 	if (f != (_float) 0) {
-		p = (char unsigned *) &f;
+		p = (unsigned char *) &f;
 		*p ^= 0x80;
 	}
+	return f;
 }
diff --git a/mach/proto/fp/ngf8.c b/mach/proto/fp/ngf8.c
index d014d39fa..91aa46f4c 100644
--- a/mach/proto/fp/ngf8.c
+++ b/mach/proto/fp/ngf8.c
@@ -16,6 +16,7 @@
 
 #include "FP_types.h"
 
+_double
 ngf8(f)
 _double	f;
 {
@@ -25,5 +26,5 @@ _double	f;
 		p = (unsigned char *) &f;
 		*p ^= 0x80;
 	}
+	return f;
 }
-
diff --git a/mach/proto/fp/sbf4.c b/mach/proto/fp/sbf4.c
index 844880ecf..e6a3aa484 100644
--- a/mach/proto/fp/sbf4.c
+++ b/mach/proto/fp/sbf4.c
@@ -17,16 +17,13 @@ _float
 sbf4(s2,s1)
 _float	s1,s2;
 {
-				/* changing the sign directly	*/
-				/* is faster than the code:	*/
-				/*		s2 = -s2	*/
-	char unsigned *p;
+	unsigned char *p;
 	_float *result = &s1;	/* s1 may not be in a register! */
 
 	if (s2 == (_float) 0) {
 		return s1;
 	}
-	p = (char unsigned *) &s2;
+	p = (unsigned char *) &s2;
 	*p ^= 0x80;	/* change sign of s2 */
 	*result = adf4(s2,s1);
 	return(s1);	/* add and return result */
diff --git a/mach/proto/fp/sbf8.c b/mach/proto/fp/sbf8.c
index 76d437250..f2b711230 100644
--- a/mach/proto/fp/sbf8.c
+++ b/mach/proto/fp/sbf8.c
@@ -17,16 +17,13 @@ _double
 sbf8(s2,s1)
 _double	s1,s2;
 {
-				/* changing the sign directly	*/
-				/* is faster than the code line	*/
-				/*	s2 = -s2;		*/
-	char unsigned *p;		/* sufficient to access sign bit */
+	unsigned char *p;		/* sufficient to access sign bit */
 	_double *result = &s1;	/* s1 may not be in a register! */
 
 	if (s2.__double[0] == 0 && s2.__double[1] == 0) {
 		return s1;
 	}
-	p = (char unsigned *) &s2;
+	p = (unsigned char *) &s2;
 	*p ^= 0x80;	/* change sign of s2 */
 	*result = adf8(s2,s1);	/* add and return result */
 	return(s1);
diff --git a/mach/proto/fp/sub_ext.c b/mach/proto/fp/sub_ext.c
index 4a511a2c8..81e91b8c5 100644
--- a/mach/proto/fp/sub_ext.c
+++ b/mach/proto/fp/sub_ext.c
@@ -10,6 +10,7 @@
 */
 
 #include "FP_types.h"
+
 sub_ext(e1,e2)
 EXTEND	*e1,*e2;
 {
diff --git a/mach/proto/fp/zrf4.c b/mach/proto/fp/zrf4.c
index dd78c7a5b..eb02cb785 100644
--- a/mach/proto/fp/zrf4.c
+++ b/mach/proto/fp/zrf4.c
@@ -9,8 +9,6 @@
 	return a zero float (ZRF 4)
 */
 
-#include "FP_types.h"
-
 zrf4(l)
 long	*l;
 {