ppc_ftod.s

powerpc 405 优化过的硬浮点库

/* fpopt/ppc_ftod.S, pl_FPE_common, pl_linux 11/24/03 16:17:39                                                                  */
/*----------------------------------------------------------------------------- */
/*  Copyright (c) 2003, IBM Corporation                                         */
/*  All rights reserved.                                                        */
/*                                                                              */
/*  Redistribution and use in source and binary forms, with or                  */
/*  without modification, are permitted provided that the following             */
/*  conditions are met:                                                         */
/*                                                                              */
/*    * Redistributions of source code must retain the above                    */
/*      copyright notice, this list of conditions and the following             */
/*      disclaimer.                                                             */
/*    * Redistributions in binary form must reproduce the above                 */
/*      copyright notice, this list of conditions and the following             */
/*      disclaimer in the documentation and/or other materials                  */
/*      provided with the distribution.                                         */
/*    * Neither the name of IBM nor the names of its contributors               */
/*      may be used to endorse or promote products derived from this            */
/*      software without specific prior written permission.                     */
/*                                                                              */
/*  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND                      */
/*  CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES,                 */
/*  INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF                    */
/*  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE                    */
/*  DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS           */
/*  BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY,         */
/*  OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,                    */
/*  PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR          */
/*  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY         */
/*  OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT                */
/*  (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE           */
/*  USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.    */
/*                                                                              */
/*----------------------------------------------------------------------------- */
/*                                                                              */
/* Function: convert single floating point to double                            */
/* Input:    r3                                                                 */
/* Output:   r3,r4                                                              */
/* Notes:   1. No stack frame is created for this function, so the following    */
/*             registers must be preserved, as required by ABI specification:   */
/*               LR, CR0, R1, R2, R13-R31                                       */
/*          2. operation performed according to IEEE754-1985 standard.          */
/*                                                                              */
/*----------------------------------------------------------------------------- */

#include <ppc4xx.inc>
#include "fpeLib.inc"
 
function_prolog(__extendsfdf2)
   
     mfcr       r0                     /* Use r0 to save CR */
/* Take single apart                                                             */
     rlwinm.    r8,r3,9,0xFF           /* extract exponent (and set CR0) */
     addi       r9,r8,-1               /* make zero look like high value */
     cmpli      cr1,r9,0xFE            /* r9 >= FE means zero/denorm/INF/NaN */
     addi       r9,r8,896              /* +1023 - 127 */
     rlwinm     r4,r3,29,0xE0000000    /* fill lower word with 3 lsb's of orig */
     srawi      r11,r3,3               /* position sign and fraction for DP */
     bge        cr1,unusual            /* not a standard normalized number */
 
insert_exp:
     rlwimi     r11,r9,20,0x7FF00000   /* insert biased exponent */
 
do_store:
     mr         r3,r11                 /* move first word */
                                       /* second word already in r4 */
     mtcr       r0                     /* restore condition register */
     blr
/*                                                                              */
INF_NaN:
     oris       r11,r11,0x7FF0         /* mark as INF/NaN */
     rlwinm.    r9,r3,0,0x007fffff     /* if NaN,  */
     beq        cr0,do_store
     oris       r11,r11,0x0008         /*   make into QNaN */
     b          do_store               /* and otherwise same as norm */
/*                                                                              */
/* here we have a INF or NaN, or else a zero or denorm                          */
/*                                                                              */
unusual:
     bne        INF_NaN                /* if original exp not zero, then INF.. */
/*                                                                              */
/* zero or denorm.                                                              */
/*                                                                              */
     rlwinm.    r10,r3,0,0x7FFFFFFF    /* clear sign.   */
     cntlzw     r5,r10                 /* get location of first bit */
     ori        r11,r3,0               /* just copy input (assume a zero) */
     beq        do_store               /* input is a zero - r4 already set */
                                       /* (assume that this branch is "free") */
/*                                                                              */
/* here for a denorm                                                            */
/* note that this path is penalized, but zeros are considered to be much        */
/* more common, and the zero argument path has been optimized                   */
/*                                                                              */
     addi       r7,r5,-11              /* get rotate amount  */
/* Note that the fraction is "pulled up" to its final position in the resulting */
/* double                                                                       */
     rlwnm      r11,r10,r7,0x000FFFFF  /* insert fraction (leading bit in bit 11) */
     rlwnm      r4,r10,r7,0xE0000000   /* low order bits to r4 */
/* The -3 is because the shift count is 3 less than the true value because      */
/* the fractions are in different positions in the source single and target     */
/* double.                                                                      */
     subfic     r9,r7,897-3            /* compute biased exp */
     rlwimi     r11,r3,0,0x80000000    /* insert sign bit (Arrgh!) */
     b          insert_exp
 

function_epilog(__extendsfdf2)