blob: 67059c33a250a9e45bcdb4c18e84ad295fb9de6a [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001/* IEEE754 floating point arithmetic
2 * single precision square root
3 */
4/*
5 * MIPS floating point support
6 * Copyright (C) 1994-2000 Algorithmics Ltd.
Linus Torvalds1da177e2005-04-16 15:20:36 -07007 *
Linus Torvalds1da177e2005-04-16 15:20:36 -07008 * This program is free software; you can distribute it and/or modify it
9 * under the terms of the GNU General Public License (Version 2) as
10 * published by the Free Software Foundation.
11 *
12 * This program is distributed in the hope it will be useful, but WITHOUT
13 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
14 * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
15 * for more details.
16 *
17 * You should have received a copy of the GNU General Public License along
18 * with this program; if not, write to the Free Software Foundation, Inc.,
Ralf Baechle3f7cac42014-04-26 01:49:14 +020019 * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
Linus Torvalds1da177e2005-04-16 15:20:36 -070020 */
21
Linus Torvalds1da177e2005-04-16 15:20:36 -070022#include "ieee754sp.h"
23
Ralf Baechle2209bcb2014-04-16 01:31:11 +020024union ieee754sp ieee754sp_sqrt(union ieee754sp x)
Linus Torvalds1da177e2005-04-16 15:20:36 -070025{
26 int ix, s, q, m, t, i;
27 unsigned int r;
28 COMPXSP;
29
30 /* take care of Inf and NaN */
31
32 EXPLODEXSP;
Ralf Baechle9e8bad12014-04-19 00:36:32 +020033 ieee754_clearcx();
Linus Torvalds1da177e2005-04-16 15:20:36 -070034 FLUSHXSP;
35
36 /* x == INF or NAN? */
37 switch (xc) {
Maciej W. Rozyckid5afa7e2015-04-03 23:25:34 +010038 case IEEE754_CLASS_SNAN:
39 return ieee754sp_nanxcpt(x);
40
Linus Torvalds1da177e2005-04-16 15:20:36 -070041 case IEEE754_CLASS_QNAN:
42 /* sqrt(Nan) = Nan */
Maciej W. Rozycki539bfb52015-04-03 23:25:30 +010043 return x;
Ralf Baechle3f7cac42014-04-26 01:49:14 +020044
Linus Torvalds1da177e2005-04-16 15:20:36 -070045 case IEEE754_CLASS_ZERO:
46 /* sqrt(0) = 0 */
47 return x;
Ralf Baechle3f7cac42014-04-26 01:49:14 +020048
Linus Torvalds1da177e2005-04-16 15:20:36 -070049 case IEEE754_CLASS_INF:
50 if (xs) {
51 /* sqrt(-Inf) = Nan */
Ralf Baechle9e8bad12014-04-19 00:36:32 +020052 ieee754_setcx(IEEE754_INVALID_OPERATION);
Maciej W. Rozycki539bfb52015-04-03 23:25:30 +010053 return ieee754sp_indef();
Linus Torvalds1da177e2005-04-16 15:20:36 -070054 }
55 /* sqrt(+Inf) = Inf */
56 return x;
Ralf Baechle3f7cac42014-04-26 01:49:14 +020057
Linus Torvalds1da177e2005-04-16 15:20:36 -070058 case IEEE754_CLASS_DNORM:
59 case IEEE754_CLASS_NORM:
60 if (xs) {
61 /* sqrt(-x) = Nan */
Ralf Baechle9e8bad12014-04-19 00:36:32 +020062 ieee754_setcx(IEEE754_INVALID_OPERATION);
Maciej W. Rozycki539bfb52015-04-03 23:25:30 +010063 return ieee754sp_indef();
Linus Torvalds1da177e2005-04-16 15:20:36 -070064 }
65 break;
66 }
67
68 ix = x.bits;
69
70 /* normalize x */
71 m = (ix >> 23);
72 if (m == 0) { /* subnormal x */
73 for (i = 0; (ix & 0x00800000) == 0; i++)
74 ix <<= 1;
75 m -= i - 1;
76 }
77 m -= 127; /* unbias exponent */
78 ix = (ix & 0x007fffff) | 0x00800000;
79 if (m & 1) /* odd m, double x to make it even */
80 ix += ix;
81 m >>= 1; /* m = [m/2] */
82
83 /* generate sqrt(x) bit by bit */
84 ix += ix;
85 q = s = 0; /* q = sqrt(x) */
86 r = 0x01000000; /* r = moving bit from right to left */
87
88 while (r != 0) {
89 t = s + r;
90 if (t <= ix) {
91 s = t + r;
92 ix -= t;
93 q += r;
94 }
95 ix += ix;
96 r >>= 1;
97 }
98
99 if (ix != 0) {
Ralf Baechle9e8bad12014-04-19 00:36:32 +0200100 ieee754_setcx(IEEE754_INEXACT);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700101 switch (ieee754_csr.rm) {
Ralf Baechle56a64732014-04-30 11:21:55 +0200102 case FPU_CSR_RU:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700103 q += 2;
104 break;
Ralf Baechle56a64732014-04-30 11:21:55 +0200105 case FPU_CSR_RN:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700106 q += (q & 1);
107 break;
108 }
109 }
110 ix = (q >> 1) + 0x3f000000;
111 ix += (m << 23);
112 x.bits = ix;
113 return x;
114}