apps/plugins/fractals/cpu_sh7043.h

   1 /***************************************************************************
   2  *             __________               __   ___.
   3  *   Open      \______   \ ____   ____ |  | _\_ |__   _______  ___
   4  *   Source     |       _//  _ \_/ ___\|  |/ /| __ \ /  _ \  \/  /
   5  *   Jukebox    |    |   (  <_> )  \___|    < | \_\ (  <_> > <  <
   6  *   Firmware   |____|_  /\____/ \___  >__|_ \|___  /\____/__/\_ \
   7  *                     \/            \/     \/    \/            \/
   8  * $Id$
   9  *
  10  * Copyright (C) 2009 Tomer Shalev
  11  *
  12  * This program is free software; you can redistribute it and/or
  13  * modify it under the terms of the GNU General Public License
  14  * as published by the Free Software Foundation; either version 2
  15  * of the License, or (at your option) any later version.
  16  *
  17  * This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY
  18  * KIND, either express or implied.
  19  *
  20  ****************************************************************************/
  21
  22 #ifndef _CPU_SH7043_H
  23 #define _CPU_SH7043_H
  24
  25 inline static short muls16_asr10(short a, short b)
  26 {
  27     short r;
  28     asm (
  29         "muls    %[a],%[b]   \n"
  30         "sts     macl,%[r]   \n"
  31         "shlr8   %[r]        \n"
  32         "shlr2   %[r]        \n"
  33         : /* outputs */
  34         [r]"=r"(r)
  35         : /* inputs */
  36         [a]"r"(a),
  37         [b]"r"(b)
  38     );
  39     return r;
  40 }
  41
  42 inline static long muls32_asr26(long a, long b)
  43 {
  44     long r, t1, t2, t3;
  45     asm (
  46         /* Signed 32bit * 32bit -> 64bit multiplication.
  47            Notation: xxab * xxcd, where each letter represents 16 bits.
  48            xx is the 64 bit sign extension.  */
  49         "swap.w  %[a],%[t1]  \n" /* t1 = ba */
  50         "mulu    %[t1],%[b]  \n" /* a * d */
  51         "swap.w  %[b],%[t3]  \n" /* t3 = dc */
  52         "sts     macl,%[t2]  \n" /* t2 = a * d */
  53         "mulu    %[t1],%[t3] \n" /* a * c */
  54         "sts     macl,%[r]   \n" /* hi = a * c */
  55         "mulu    %[a],%[t3]  \n" /* b * c */
  56         "clrt                \n"
  57         "sts     macl,%[t3]  \n" /* t3 = b * c */
  58         "addc    %[t2],%[t3] \n" /* t3 += t2, carry -> t2 */
  59         "movt    %[t2]       \n"
  60         "mulu    %[a],%[b]   \n" /* b * d */
  61         "mov     %[t3],%[t1] \n" /* t1t3 = t2t3 << 16 */
  62         "xtrct   %[t2],%[t1] \n"
  63         "shll16  %[t3]       \n"
  64         "sts     macl,%[t2]  \n" /* lo = b * d */
  65         "clrt                \n" /* hi.lo += t1t3 */
  66         "addc    %[t3],%[t2] \n"
  67         "addc    %[t1],%[r]  \n"
  68         "cmp/pz  %[a]        \n" /* ab >= 0 ? */
  69         "bt      1f          \n"
  70         "sub     %[b],%[r]   \n" /* no: hi -= cd (sign extension of ab is -1) */
  71     "1:                      \n"
  72         "cmp/pz  %[b]        \n" /* cd >= 0 ? */
  73         "bt      2f          \n"
  74         "sub     %[a],%[r]   \n" /* no: hi -= ab (sign extension of cd is -1) */
  75     "2:                      \n"
  76         /* Shift right by 26 and return low 32 bits */
  77         "shll2   %[r]        \n" /* hi <<= 6 */
  78         "shll2   %[r]        \n"
  79         "shll2   %[r]        \n"
  80         "shlr16  %[t2]       \n" /* (unsigned)lo >>= 26 */
  81         "shlr8   %[t2]       \n"
  82         "shlr2   %[t2]       \n"
  83         "or      %[t2],%[r]  \n" /* combine result */
  84         : /* outputs */
  85         [r] "=&r"(r),
  86         [t1]"=&r"(t1),
  87         [t2]"=&r"(t2),
  88         [t3]"=&r"(t3)
  89         : /* inputs */
  90         [a] "r"  (a),
  91         [b] "r"  (b)
  92     );
  93     return r;
  94 }
  95
  96 #endif