1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
|
//===-- lib/fp_lib.h - Floating-point utilities -------------------*- C -*-===//
//
// The LLVM Compiler Infrastructure
//
// This file is dual licensed under the MIT and the University of Illinois Open
// Source Licenses. See LICENSE.TXT for details.
//
//===----------------------------------------------------------------------===//
//
// This file is a configuration header for soft-float routines in compiler-rt.
// This file does not provide any part of the compiler-rt interface, but defines
// many useful constants and utility routines that are used in the
// implementation of the soft-float routines in compiler-rt.
//
// Assumes that float, double and long double correspond to the IEEE-754
// binary32, binary64 and binary 128 types, respectively, and that integer
// endianness matches floating point endianness on the target platform.
//
//===----------------------------------------------------------------------===//
#ifndef FP_LIB_HEADER
#define FP_LIB_HEADER
#include <stdint.h>
#include <stdbool.h>
#include <limits.h>
#include "int_lib.h"
// x86_64 FreeBSD prior v9.3 define fixed-width types incorrectly in
// 32-bit mode.
#if defined(__FreeBSD__) && defined(__i386__)
# include <sys/param.h>
# if __FreeBSD_version < 903000 // v9.3
# define uint64_t unsigned long long
# define int64_t long long
# undef UINT64_C
# define UINT64_C(c) (c ## ULL)
# endif
#endif
#if defined SINGLE_PRECISION
typedef uint32_t rep_t;
typedef int32_t srep_t;
typedef float fp_t;
#define REP_C UINT32_C
#define significandBits 23
static __inline int rep_clz(rep_t a) {
return __builtin_clz(a);
}
// 32x32 --> 64 bit multiply
static __inline void wideMultiply(rep_t a, rep_t b, rep_t *hi, rep_t *lo) {
const uint64_t product = (uint64_t)a*b;
*hi = product >> 32;
*lo = product;
}
COMPILER_RT_ABI fp_t __addsf3(fp_t a, fp_t b);
#elif defined DOUBLE_PRECISION
typedef uint64_t rep_t;
typedef int64_t srep_t;
typedef double fp_t;
#define REP_C UINT64_C
#define significandBits 52
static __inline int rep_clz(rep_t a) {
#if defined __LP64__
return __builtin_clzl(a);
#else
if (a & REP_C(0xffffffff00000000))
return __builtin_clz(a >> 32);
else
return 32 + __builtin_clz(a & REP_C(0xffffffff));
#endif
}
#define loWord(a) (a & 0xffffffffU)
#define hiWord(a) (a >> 32)
// 64x64 -> 128 wide multiply for platforms that don't have such an operation;
// many 64-bit platforms have this operation, but they tend to have hardware
// floating-point, so we don't bother with a special case for them here.
static __inline void wideMultiply(rep_t a, rep_t b, rep_t *hi, rep_t *lo) {
// Each of the component 32x32 -> 64 products
const uint64_t plolo = loWord(a) * loWord(b);
const uint64_t plohi = loWord(a) * hiWord(b);
const uint64_t philo = hiWord(a) * loWord(b);
const uint64_t phihi = hiWord(a) * hiWord(b);
// Sum terms that contribute to lo in a way that allows us to get the carry
const uint64_t r0 = loWord(plolo);
const uint64_t r1 = hiWord(plolo) + loWord(plohi) + loWord(philo);
*lo = r0 + (r1 << 32);
// Sum terms contributing to hi with the carry from lo
*hi = hiWord(plohi) + hiWord(philo) + hiWord(r1) + phihi;
}
#undef loWord
#undef hiWord
COMPILER_RT_ABI fp_t __adddf3(fp_t a, fp_t b);
#elif defined QUAD_PRECISION
#if __LDBL_MANT_DIG__ == 113
#define CRT_LDBL_128BIT
typedef __uint128_t rep_t;
typedef __int128_t srep_t;
typedef long double fp_t;
#define REP_C (__uint128_t)
// Note: Since there is no explicit way to tell compiler the constant is a
// 128-bit integer, we let the constant be casted to 128-bit integer
#define significandBits 112
static __inline int rep_clz(rep_t a) {
const union
{
__uint128_t ll;
#if _YUGA_BIG_ENDIAN
struct { uint64_t high, low; } s;
#else
struct { uint64_t low, high; } s;
#endif
} uu = { .ll = a };
uint64_t word;
uint64_t add;
if (uu.s.high){
word = uu.s.high;
add = 0;
}
else{
word = uu.s.low;
add = 64;
}
return __builtin_clzll(word) + add;
}
#define Word_LoMask UINT64_C(0x00000000ffffffff)
#define Word_HiMask UINT64_C(0xffffffff00000000)
#define Word_FullMask UINT64_C(0xffffffffffffffff)
#define Word_1(a) (uint64_t)((a >> 96) & Word_LoMask)
#define Word_2(a) (uint64_t)((a >> 64) & Word_LoMask)
#define Word_3(a) (uint64_t)((a >> 32) & Word_LoMask)
#define Word_4(a) (uint64_t)(a & Word_LoMask)
// 128x128 -> 256 wide multiply for platforms that don't have such an operation;
// many 64-bit platforms have this operation, but they tend to have hardware
// floating-point, so we don't bother with a special case for them here.
static __inline void wideMultiply(rep_t a, rep_t b, rep_t *hi, rep_t *lo) {
const uint64_t product11 = Word_1(a) * Word_1(b);
const uint64_t product12 = Word_1(a) * Word_2(b);
const uint64_t product13 = Word_1(a) * Word_3(b);
const uint64_t product14 = Word_1(a) * Word_4(b);
const uint64_t product21 = Word_2(a) * Word_1(b);
const uint64_t product22 = Word_2(a) * Word_2(b);
const uint64_t product23 = Word_2(a) * Word_3(b);
const uint64_t product24 = Word_2(a) * Word_4(b);
const uint64_t product31 = Word_3(a) * Word_1(b);
const uint64_t product32 = Word_3(a) * Word_2(b);
const uint64_t product33 = Word_3(a) * Word_3(b);
const uint64_t product34 = Word_3(a) * Word_4(b);
const uint64_t product41 = Word_4(a) * Word_1(b);
const uint64_t product42 = Word_4(a) * Word_2(b);
const uint64_t product43 = Word_4(a) * Word_3(b);
const uint64_t product44 = Word_4(a) * Word_4(b);
const __uint128_t sum0 = (__uint128_t)product44;
const __uint128_t sum1 = (__uint128_t)product34 +
(__uint128_t)product43;
const __uint128_t sum2 = (__uint128_t)product24 +
(__uint128_t)product33 +
(__uint128_t)product42;
const __uint128_t sum3 = (__uint128_t)product14 +
(__uint128_t)product23 +
(__uint128_t)product32 +
(__uint128_t)product41;
const __uint128_t sum4 = (__uint128_t)product13 +
(__uint128_t)product22 +
(__uint128_t)product31;
const __uint128_t sum5 = (__uint128_t)product12 +
(__uint128_t)product21;
const __uint128_t sum6 = (__uint128_t)product11;
const __uint128_t r0 = (sum0 & Word_FullMask) +
((sum1 & Word_LoMask) << 32);
const __uint128_t r1 = (sum0 >> 64) +
((sum1 >> 32) & Word_FullMask) +
(sum2 & Word_FullMask) +
((sum3 << 32) & Word_HiMask);
*lo = r0 + (r1 << 64);
*hi = (r1 >> 64) +
(sum1 >> 96) +
(sum2 >> 64) +
(sum3 >> 32) +
sum4 +
(sum5 << 32) +
(sum6 << 64);
}
#undef Word_1
#undef Word_2
#undef Word_3
#undef Word_4
#undef Word_HiMask
#undef Word_LoMask
#undef Word_FullMask
#endif // __LDBL_MANT_DIG__ == 113
#else
#error SINGLE_PRECISION, DOUBLE_PRECISION or QUAD_PRECISION must be defined.
#endif
#if defined(SINGLE_PRECISION) || defined(DOUBLE_PRECISION) || defined(CRT_LDBL_128BIT)
#define typeWidth (sizeof(rep_t)*CHAR_BIT)
#define exponentBits (typeWidth - significandBits - 1)
#define maxExponent ((1 << exponentBits) - 1)
#define exponentBias (maxExponent >> 1)
#define implicitBit (REP_C(1) << significandBits)
#define significandMask (implicitBit - 1U)
#define signBit (REP_C(1) << (significandBits + exponentBits))
#define absMask (signBit - 1U)
#define exponentMask (absMask ^ significandMask)
#define oneRep ((rep_t)exponentBias << significandBits)
#define infRep exponentMask
#define quietBit (implicitBit >> 1)
#define qnanRep (exponentMask | quietBit)
static __inline rep_t toRep(fp_t x) {
const union { fp_t f; rep_t i; } rep = {.f = x};
return rep.i;
}
static __inline fp_t fromRep(rep_t x) {
const union { fp_t f; rep_t i; } rep = {.i = x};
return rep.f;
}
static __inline int normalize(rep_t *significand) {
const int shift = rep_clz(*significand) - rep_clz(implicitBit);
*significand <<= shift;
return 1 - shift;
}
static __inline void wideLeftShift(rep_t *hi, rep_t *lo, int count) {
*hi = *hi << count | *lo >> (typeWidth - count);
*lo = *lo << count;
}
static __inline void wideRightShiftWithSticky(rep_t *hi, rep_t *lo, unsigned int count) {
if (count < typeWidth) {
const bool sticky = *lo << (typeWidth - count);
*lo = *hi << (typeWidth - count) | *lo >> count | sticky;
*hi = *hi >> count;
}
else if (count < 2*typeWidth) {
const bool sticky = *hi << (2*typeWidth - count) | *lo;
*lo = *hi >> (count - typeWidth) | sticky;
*hi = 0;
} else {
const bool sticky = *hi | *lo;
*lo = sticky;
*hi = 0;
}
}
#endif
#endif // FP_LIB_HEADER
|
