381 lines
10 KiB
C
381 lines
10 KiB
C
/*
|
|
* Linux/PA-RISC Project (http://www.parisc-linux.org/)
|
|
*
|
|
* Floating-point emulation code
|
|
* Copyright (C) 2001 Hewlett-Packard (Paul Bame) <bame@debian.org>
|
|
*
|
|
* This program is free software; you can redistribute it and/or modify
|
|
* it under the terms of the GNU General Public License as published by
|
|
* the Free Software Foundation; either version 2, or (at your option)
|
|
* any later version.
|
|
*
|
|
* This program is distributed in the hope that it will be useful,
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
* GNU General Public License for more details.
|
|
*
|
|
* You should have received a copy of the GNU General Public License
|
|
* along with this program; if not, write to the Free Software
|
|
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
|
|
*/
|
|
/*
|
|
* BEGIN_DESC
|
|
*
|
|
* File:
|
|
* @(#) pa/spmath/sfmpy.c $Revision: 1.1 $
|
|
*
|
|
* Purpose:
|
|
* Single Precision Floating-point Multiply
|
|
*
|
|
* External Interfaces:
|
|
* sgl_fmpy(srcptr1,srcptr2,dstptr,status)
|
|
*
|
|
* Internal Interfaces:
|
|
*
|
|
* Theory:
|
|
* <<please update with a overview of the operation of this file>>
|
|
*
|
|
* END_DESC
|
|
*/
|
|
|
|
|
|
#include "float.h"
|
|
#include "sgl_float.h"
|
|
|
|
/*
|
|
* Single Precision Floating-point Multiply
|
|
*/
|
|
|
|
int
|
|
sgl_fmpy(
|
|
sgl_floating_point *srcptr1,
|
|
sgl_floating_point *srcptr2,
|
|
sgl_floating_point *dstptr,
|
|
unsigned int *status)
|
|
{
|
|
register unsigned int opnd1, opnd2, opnd3, result;
|
|
register int dest_exponent, count;
|
|
register boolean inexact = FALSE, guardbit = FALSE, stickybit = FALSE;
|
|
boolean is_tiny;
|
|
|
|
opnd1 = *srcptr1;
|
|
opnd2 = *srcptr2;
|
|
/*
|
|
* set sign bit of result
|
|
*/
|
|
if (Sgl_sign(opnd1) ^ Sgl_sign(opnd2)) Sgl_setnegativezero(result);
|
|
else Sgl_setzero(result);
|
|
/*
|
|
* check first operand for NaN's or infinity
|
|
*/
|
|
if (Sgl_isinfinity_exponent(opnd1)) {
|
|
if (Sgl_iszero_mantissa(opnd1)) {
|
|
if (Sgl_isnotnan(opnd2)) {
|
|
if (Sgl_iszero_exponentmantissa(opnd2)) {
|
|
/*
|
|
* invalid since operands are infinity
|
|
* and zero
|
|
*/
|
|
if (Is_invalidtrap_enabled())
|
|
return(INVALIDEXCEPTION);
|
|
Set_invalidflag();
|
|
Sgl_makequietnan(result);
|
|
*dstptr = result;
|
|
return(NOEXCEPTION);
|
|
}
|
|
/*
|
|
* return infinity
|
|
*/
|
|
Sgl_setinfinity_exponentmantissa(result);
|
|
*dstptr = result;
|
|
return(NOEXCEPTION);
|
|
}
|
|
}
|
|
else {
|
|
/*
|
|
* is NaN; signaling or quiet?
|
|
*/
|
|
if (Sgl_isone_signaling(opnd1)) {
|
|
/* trap if INVALIDTRAP enabled */
|
|
if (Is_invalidtrap_enabled())
|
|
return(INVALIDEXCEPTION);
|
|
/* make NaN quiet */
|
|
Set_invalidflag();
|
|
Sgl_set_quiet(opnd1);
|
|
}
|
|
/*
|
|
* is second operand a signaling NaN?
|
|
*/
|
|
else if (Sgl_is_signalingnan(opnd2)) {
|
|
/* trap if INVALIDTRAP enabled */
|
|
if (Is_invalidtrap_enabled())
|
|
return(INVALIDEXCEPTION);
|
|
/* make NaN quiet */
|
|
Set_invalidflag();
|
|
Sgl_set_quiet(opnd2);
|
|
*dstptr = opnd2;
|
|
return(NOEXCEPTION);
|
|
}
|
|
/*
|
|
* return quiet NaN
|
|
*/
|
|
*dstptr = opnd1;
|
|
return(NOEXCEPTION);
|
|
}
|
|
}
|
|
/*
|
|
* check second operand for NaN's or infinity
|
|
*/
|
|
if (Sgl_isinfinity_exponent(opnd2)) {
|
|
if (Sgl_iszero_mantissa(opnd2)) {
|
|
if (Sgl_iszero_exponentmantissa(opnd1)) {
|
|
/* invalid since operands are zero & infinity */
|
|
if (Is_invalidtrap_enabled())
|
|
return(INVALIDEXCEPTION);
|
|
Set_invalidflag();
|
|
Sgl_makequietnan(opnd2);
|
|
*dstptr = opnd2;
|
|
return(NOEXCEPTION);
|
|
}
|
|
/*
|
|
* return infinity
|
|
*/
|
|
Sgl_setinfinity_exponentmantissa(result);
|
|
*dstptr = result;
|
|
return(NOEXCEPTION);
|
|
}
|
|
/*
|
|
* is NaN; signaling or quiet?
|
|
*/
|
|
if (Sgl_isone_signaling(opnd2)) {
|
|
/* trap if INVALIDTRAP enabled */
|
|
if (Is_invalidtrap_enabled()) return(INVALIDEXCEPTION);
|
|
|
|
/* make NaN quiet */
|
|
Set_invalidflag();
|
|
Sgl_set_quiet(opnd2);
|
|
}
|
|
/*
|
|
* return quiet NaN
|
|
*/
|
|
*dstptr = opnd2;
|
|
return(NOEXCEPTION);
|
|
}
|
|
/*
|
|
* Generate exponent
|
|
*/
|
|
dest_exponent = Sgl_exponent(opnd1) + Sgl_exponent(opnd2) - SGL_BIAS;
|
|
|
|
/*
|
|
* Generate mantissa
|
|
*/
|
|
if (Sgl_isnotzero_exponent(opnd1)) {
|
|
/* set hidden bit */
|
|
Sgl_clear_signexponent_set_hidden(opnd1);
|
|
}
|
|
else {
|
|
/* check for zero */
|
|
if (Sgl_iszero_mantissa(opnd1)) {
|
|
Sgl_setzero_exponentmantissa(result);
|
|
*dstptr = result;
|
|
return(NOEXCEPTION);
|
|
}
|
|
/* is denormalized, adjust exponent */
|
|
Sgl_clear_signexponent(opnd1);
|
|
Sgl_leftshiftby1(opnd1);
|
|
Sgl_normalize(opnd1,dest_exponent);
|
|
}
|
|
/* opnd2 needs to have hidden bit set with msb in hidden bit */
|
|
if (Sgl_isnotzero_exponent(opnd2)) {
|
|
Sgl_clear_signexponent_set_hidden(opnd2);
|
|
}
|
|
else {
|
|
/* check for zero */
|
|
if (Sgl_iszero_mantissa(opnd2)) {
|
|
Sgl_setzero_exponentmantissa(result);
|
|
*dstptr = result;
|
|
return(NOEXCEPTION);
|
|
}
|
|
/* is denormalized; want to normalize */
|
|
Sgl_clear_signexponent(opnd2);
|
|
Sgl_leftshiftby1(opnd2);
|
|
Sgl_normalize(opnd2,dest_exponent);
|
|
}
|
|
|
|
/* Multiply two source mantissas together */
|
|
|
|
Sgl_leftshiftby4(opnd2); /* make room for guard bits */
|
|
Sgl_setzero(opnd3);
|
|
/*
|
|
* Four bits at a time are inspected in each loop, and a
|
|
* simple shift and add multiply algorithm is used.
|
|
*/
|
|
for (count=1;count<SGL_P;count+=4) {
|
|
stickybit |= Slow4(opnd3);
|
|
Sgl_rightshiftby4(opnd3);
|
|
if (Sbit28(opnd1)) Sall(opnd3) += (Sall(opnd2) << 3);
|
|
if (Sbit29(opnd1)) Sall(opnd3) += (Sall(opnd2) << 2);
|
|
if (Sbit30(opnd1)) Sall(opnd3) += (Sall(opnd2) << 1);
|
|
if (Sbit31(opnd1)) Sall(opnd3) += Sall(opnd2);
|
|
Sgl_rightshiftby4(opnd1);
|
|
}
|
|
/* make sure result is left-justified */
|
|
if (Sgl_iszero_sign(opnd3)) {
|
|
Sgl_leftshiftby1(opnd3);
|
|
}
|
|
else {
|
|
/* result mantissa >= 2. */
|
|
dest_exponent++;
|
|
}
|
|
/* check for denormalized result */
|
|
while (Sgl_iszero_sign(opnd3)) {
|
|
Sgl_leftshiftby1(opnd3);
|
|
dest_exponent--;
|
|
}
|
|
/*
|
|
* check for guard, sticky and inexact bits
|
|
*/
|
|
stickybit |= Sgl_all(opnd3) << (SGL_BITLENGTH - SGL_EXP_LENGTH + 1);
|
|
guardbit = Sbit24(opnd3);
|
|
inexact = guardbit | stickybit;
|
|
|
|
/* re-align mantissa */
|
|
Sgl_rightshiftby8(opnd3);
|
|
|
|
/*
|
|
* round result
|
|
*/
|
|
if (inexact && (dest_exponent>0 || Is_underflowtrap_enabled())) {
|
|
Sgl_clear_signexponent(opnd3);
|
|
switch (Rounding_mode()) {
|
|
case ROUNDPLUS:
|
|
if (Sgl_iszero_sign(result))
|
|
Sgl_increment(opnd3);
|
|
break;
|
|
case ROUNDMINUS:
|
|
if (Sgl_isone_sign(result))
|
|
Sgl_increment(opnd3);
|
|
break;
|
|
case ROUNDNEAREST:
|
|
if (guardbit) {
|
|
if (stickybit || Sgl_isone_lowmantissa(opnd3))
|
|
Sgl_increment(opnd3);
|
|
}
|
|
}
|
|
if (Sgl_isone_hidden(opnd3)) dest_exponent++;
|
|
}
|
|
Sgl_set_mantissa(result,opnd3);
|
|
|
|
/*
|
|
* Test for overflow
|
|
*/
|
|
if (dest_exponent >= SGL_INFINITY_EXPONENT) {
|
|
/* trap if OVERFLOWTRAP enabled */
|
|
if (Is_overflowtrap_enabled()) {
|
|
/*
|
|
* Adjust bias of result
|
|
*/
|
|
Sgl_setwrapped_exponent(result,dest_exponent,ovfl);
|
|
*dstptr = result;
|
|
if (inexact)
|
|
if (Is_inexacttrap_enabled())
|
|
return(OVERFLOWEXCEPTION | INEXACTEXCEPTION);
|
|
else Set_inexactflag();
|
|
return(OVERFLOWEXCEPTION);
|
|
}
|
|
inexact = TRUE;
|
|
Set_overflowflag();
|
|
/* set result to infinity or largest number */
|
|
Sgl_setoverflow(result);
|
|
}
|
|
/*
|
|
* Test for underflow
|
|
*/
|
|
else if (dest_exponent <= 0) {
|
|
/* trap if UNDERFLOWTRAP enabled */
|
|
if (Is_underflowtrap_enabled()) {
|
|
/*
|
|
* Adjust bias of result
|
|
*/
|
|
Sgl_setwrapped_exponent(result,dest_exponent,unfl);
|
|
*dstptr = result;
|
|
if (inexact)
|
|
if (Is_inexacttrap_enabled())
|
|
return(UNDERFLOWEXCEPTION | INEXACTEXCEPTION);
|
|
else Set_inexactflag();
|
|
return(UNDERFLOWEXCEPTION);
|
|
}
|
|
|
|
/* Determine if should set underflow flag */
|
|
is_tiny = TRUE;
|
|
if (dest_exponent == 0 && inexact) {
|
|
switch (Rounding_mode()) {
|
|
case ROUNDPLUS:
|
|
if (Sgl_iszero_sign(result)) {
|
|
Sgl_increment(opnd3);
|
|
if (Sgl_isone_hiddenoverflow(opnd3))
|
|
is_tiny = FALSE;
|
|
Sgl_decrement(opnd3);
|
|
}
|
|
break;
|
|
case ROUNDMINUS:
|
|
if (Sgl_isone_sign(result)) {
|
|
Sgl_increment(opnd3);
|
|
if (Sgl_isone_hiddenoverflow(opnd3))
|
|
is_tiny = FALSE;
|
|
Sgl_decrement(opnd3);
|
|
}
|
|
break;
|
|
case ROUNDNEAREST:
|
|
if (guardbit && (stickybit ||
|
|
Sgl_isone_lowmantissa(opnd3))) {
|
|
Sgl_increment(opnd3);
|
|
if (Sgl_isone_hiddenoverflow(opnd3))
|
|
is_tiny = FALSE;
|
|
Sgl_decrement(opnd3);
|
|
}
|
|
break;
|
|
}
|
|
}
|
|
|
|
/*
|
|
* denormalize result or set to signed zero
|
|
*/
|
|
stickybit = inexact;
|
|
Sgl_denormalize(opnd3,dest_exponent,guardbit,stickybit,inexact);
|
|
|
|
/* return zero or smallest number */
|
|
if (inexact) {
|
|
switch (Rounding_mode()) {
|
|
case ROUNDPLUS:
|
|
if (Sgl_iszero_sign(result)) {
|
|
Sgl_increment(opnd3);
|
|
}
|
|
break;
|
|
case ROUNDMINUS:
|
|
if (Sgl_isone_sign(result)) {
|
|
Sgl_increment(opnd3);
|
|
}
|
|
break;
|
|
case ROUNDNEAREST:
|
|
if (guardbit && (stickybit ||
|
|
Sgl_isone_lowmantissa(opnd3))) {
|
|
Sgl_increment(opnd3);
|
|
}
|
|
break;
|
|
}
|
|
if (is_tiny) Set_underflowflag();
|
|
}
|
|
Sgl_set_exponentmantissa(result,opnd3);
|
|
}
|
|
else Sgl_set_exponent(result,dest_exponent);
|
|
*dstptr = result;
|
|
|
|
/* check for inexact */
|
|
if (inexact) {
|
|
if (Is_inexacttrap_enabled()) return(INEXACTEXCEPTION);
|
|
else Set_inexactflag();
|
|
}
|
|
return(NOEXCEPTION);
|
|
}
|