linux/arch/parisc/math-emu/sfsqrt.c
Linus Torvalds 1da177e4c3 Linux-2.6.12-rc2
Initial git repository build. I'm not bothering with the full history,
even though we have it. We can create a separate "historical" git
archive of that later if we want to, and in the meantime it's about
3.2GB when imported into git - space that would just make the early
git days unnecessarily complicated, when we don't have a lot of good
infrastructure for it.

Let it rip!
2005-04-16 15:20:36 -07:00

188 lines
4.7 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/sfsqrt.c $Revision: 1.1 $
*
* Purpose:
* Single Floating-point Square Root
*
* External Interfaces:
* sgl_fsqrt(srcptr,nullptr,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 Floating-point Square Root
*/
/*ARGSUSED*/
unsigned int
sgl_fsqrt(
sgl_floating_point *srcptr,
unsigned int *nullptr,
sgl_floating_point *dstptr,
unsigned int *status)
{
register unsigned int src, result;
register int src_exponent;
register unsigned int newbit, sum;
register boolean guardbit = FALSE, even_exponent;
src = *srcptr;
/*
* check source operand for NaN or infinity
*/
if ((src_exponent = Sgl_exponent(src)) == SGL_INFINITY_EXPONENT) {
/*
* is signaling NaN?
*/
if (Sgl_isone_signaling(src)) {
/* trap if INVALIDTRAP enabled */
if (Is_invalidtrap_enabled()) return(INVALIDEXCEPTION);
/* make NaN quiet */
Set_invalidflag();
Sgl_set_quiet(src);
}
/*
* Return quiet NaN or positive infinity.
* Fall thru to negative test if negative infinity.
*/
if (Sgl_iszero_sign(src) || Sgl_isnotzero_mantissa(src)) {
*dstptr = src;
return(NOEXCEPTION);
}
}
/*
* check for zero source operand
*/
if (Sgl_iszero_exponentmantissa(src)) {
*dstptr = src;
return(NOEXCEPTION);
}
/*
* check for negative source operand
*/
if (Sgl_isone_sign(src)) {
/* trap if INVALIDTRAP enabled */
if (Is_invalidtrap_enabled()) return(INVALIDEXCEPTION);
/* make NaN quiet */
Set_invalidflag();
Sgl_makequietnan(src);
*dstptr = src;
return(NOEXCEPTION);
}
/*
* Generate result
*/
if (src_exponent > 0) {
even_exponent = Sgl_hidden(src);
Sgl_clear_signexponent_set_hidden(src);
}
else {
/* normalize operand */
Sgl_clear_signexponent(src);
src_exponent++;
Sgl_normalize(src,src_exponent);
even_exponent = src_exponent & 1;
}
if (even_exponent) {
/* exponent is even */
/* Add comment here. Explain why odd exponent needs correction */
Sgl_leftshiftby1(src);
}
/*
* Add comment here. Explain following algorithm.
*
* Trust me, it works.
*
*/
Sgl_setzero(result);
newbit = 1 << SGL_P;
while (newbit && Sgl_isnotzero(src)) {
Sgl_addition(result,newbit,sum);
if(sum <= Sgl_all(src)) {
/* update result */
Sgl_addition(result,(newbit<<1),result);
Sgl_subtract(src,sum,src);
}
Sgl_rightshiftby1(newbit);
Sgl_leftshiftby1(src);
}
/* correct exponent for pre-shift */
if (even_exponent) {
Sgl_rightshiftby1(result);
}
/* check for inexact */
if (Sgl_isnotzero(src)) {
if (!even_exponent && Sgl_islessthan(result,src))
Sgl_increment(result);
guardbit = Sgl_lowmantissa(result);
Sgl_rightshiftby1(result);
/* now round result */
switch (Rounding_mode()) {
case ROUNDPLUS:
Sgl_increment(result);
break;
case ROUNDNEAREST:
/* stickybit is always true, so guardbit
* is enough to determine rounding */
if (guardbit) {
Sgl_increment(result);
}
break;
}
/* increment result exponent by 1 if mantissa overflowed */
if (Sgl_isone_hiddenoverflow(result)) src_exponent+=2;
if (Is_inexacttrap_enabled()) {
Sgl_set_exponent(result,
((src_exponent-SGL_BIAS)>>1)+SGL_BIAS);
*dstptr = result;
return(INEXACTEXCEPTION);
}
else Set_inexactflag();
}
else {
Sgl_rightshiftby1(result);
}
Sgl_set_exponent(result,((src_exponent-SGL_BIAS)>>1)+SGL_BIAS);
*dstptr = result;
return(NOEXCEPTION);
}