|
1 /* |
|
2 * Utility compute operations used by translated code. |
|
3 * |
|
4 * Copyright (c) 2007 Thiemo Seufer |
|
5 * Copyright (c) 2007 Jocelyn Mayer |
|
6 * |
|
7 * Permission is hereby granted, free of charge, to any person obtaining a copy |
|
8 * of this software and associated documentation files (the "Software"), to deal |
|
9 * in the Software without restriction, including without limitation the rights |
|
10 * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell |
|
11 * copies of the Software, and to permit persons to whom the Software is |
|
12 * furnished to do so, subject to the following conditions: |
|
13 * |
|
14 * The above copyright notice and this permission notice shall be included in |
|
15 * all copies or substantial portions of the Software. |
|
16 * |
|
17 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR |
|
18 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, |
|
19 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL |
|
20 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER |
|
21 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, |
|
22 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN |
|
23 * THE SOFTWARE. |
|
24 */ |
|
25 |
|
26 #include "osdep.h" |
|
27 |
|
28 #if defined(__x86_64__) |
|
29 #define __HAVE_FAST_MULU64__ |
|
30 static always_inline void mulu64 (uint64_t *plow, uint64_t *phigh, |
|
31 uint64_t a, uint64_t b) |
|
32 { |
|
33 __asm__ ("mul %0\n\t" |
|
34 : "=d" (*phigh), "=a" (*plow) |
|
35 : "a" (a), "0" (b)); |
|
36 } |
|
37 #define __HAVE_FAST_MULS64__ |
|
38 static always_inline void muls64 (uint64_t *plow, uint64_t *phigh, |
|
39 int64_t a, int64_t b) |
|
40 { |
|
41 __asm__ ("imul %0\n\t" |
|
42 : "=d" (*phigh), "=a" (*plow) |
|
43 : "a" (a), "0" (b)); |
|
44 } |
|
45 #else |
|
46 void muls64(uint64_t *phigh, uint64_t *plow, int64_t a, int64_t b); |
|
47 void mulu64(uint64_t *phigh, uint64_t *plow, uint64_t a, uint64_t b); |
|
48 #endif |
|
49 |
|
50 /* Binary search for leading zeros. */ |
|
51 |
|
52 static always_inline int clz32(uint32_t val) |
|
53 { |
|
54 #if QEMU_GNUC_PREREQ(3, 4) |
|
55 if (val) |
|
56 return __builtin_clz(val); |
|
57 else |
|
58 return 32; |
|
59 #else |
|
60 int cnt = 0; |
|
61 |
|
62 if (!(val & 0xFFFF0000U)) { |
|
63 cnt += 16; |
|
64 val <<= 16; |
|
65 } |
|
66 if (!(val & 0xFF000000U)) { |
|
67 cnt += 8; |
|
68 val <<= 8; |
|
69 } |
|
70 if (!(val & 0xF0000000U)) { |
|
71 cnt += 4; |
|
72 val <<= 4; |
|
73 } |
|
74 if (!(val & 0xC0000000U)) { |
|
75 cnt += 2; |
|
76 val <<= 2; |
|
77 } |
|
78 if (!(val & 0x80000000U)) { |
|
79 cnt++; |
|
80 val <<= 1; |
|
81 } |
|
82 if (!(val & 0x80000000U)) { |
|
83 cnt++; |
|
84 } |
|
85 return cnt; |
|
86 #endif |
|
87 } |
|
88 |
|
89 static always_inline int clo32(uint32_t val) |
|
90 { |
|
91 return clz32(~val); |
|
92 } |
|
93 |
|
94 static always_inline int clz64(uint64_t val) |
|
95 { |
|
96 #if QEMU_GNUC_PREREQ(3, 4) |
|
97 if (val) |
|
98 return __builtin_clzll(val); |
|
99 else |
|
100 return 64; |
|
101 #else |
|
102 int cnt = 0; |
|
103 |
|
104 if (!(val >> 32)) { |
|
105 cnt += 32; |
|
106 } else { |
|
107 val >>= 32; |
|
108 } |
|
109 |
|
110 return cnt + clz32(val); |
|
111 #endif |
|
112 } |
|
113 |
|
114 static always_inline int clo64(uint64_t val) |
|
115 { |
|
116 return clz64(~val); |
|
117 } |
|
118 |
|
119 static always_inline int ctz32(uint32_t val) |
|
120 { |
|
121 #if QEMU_GNUC_PREREQ(3, 4) |
|
122 if (val) |
|
123 return __builtin_ctz(val); |
|
124 else |
|
125 return 32; |
|
126 #else |
|
127 int cnt; |
|
128 |
|
129 cnt = 0; |
|
130 if (!(val & 0x0000FFFFUL)) { |
|
131 cnt += 16; |
|
132 val >>= 16; |
|
133 } |
|
134 if (!(val & 0x000000FFUL)) { |
|
135 cnt += 8; |
|
136 val >>= 8; |
|
137 } |
|
138 if (!(val & 0x0000000FUL)) { |
|
139 cnt += 4; |
|
140 val >>= 4; |
|
141 } |
|
142 if (!(val & 0x00000003UL)) { |
|
143 cnt += 2; |
|
144 val >>= 2; |
|
145 } |
|
146 if (!(val & 0x00000001UL)) { |
|
147 cnt++; |
|
148 val >>= 1; |
|
149 } |
|
150 if (!(val & 0x00000001UL)) { |
|
151 cnt++; |
|
152 } |
|
153 |
|
154 return cnt; |
|
155 #endif |
|
156 } |
|
157 |
|
158 static always_inline int cto32(uint32_t val) |
|
159 { |
|
160 return ctz32(~val); |
|
161 } |
|
162 |
|
163 static always_inline int ctz64(uint64_t val) |
|
164 { |
|
165 #if QEMU_GNUC_PREREQ(3, 4) |
|
166 if (val) |
|
167 return __builtin_ctz(val); |
|
168 else |
|
169 return 64; |
|
170 #else |
|
171 int cnt; |
|
172 |
|
173 cnt = 0; |
|
174 if (!((uint32_t)val)) { |
|
175 cnt += 32; |
|
176 val >>= 32; |
|
177 } |
|
178 |
|
179 return cnt + ctz32(val); |
|
180 #endif |
|
181 } |
|
182 |
|
183 static always_inline int cto64(uint64_t val) |
|
184 { |
|
185 return ctz64(~val); |
|
186 } |
|
187 |
|
188 static always_inline int ctpop8(uint8_t val) |
|
189 { |
|
190 val = (val & 0x55) + ((val >> 1) & 0x55); |
|
191 val = (val & 0x33) + ((val >> 2) & 0x33); |
|
192 val = (val & 0x0f) + ((val >> 4) & 0x0f); |
|
193 |
|
194 return val; |
|
195 } |
|
196 |
|
197 static always_inline int ctpop16(uint16_t val) |
|
198 { |
|
199 val = (val & 0x5555) + ((val >> 1) & 0x5555); |
|
200 val = (val & 0x3333) + ((val >> 2) & 0x3333); |
|
201 val = (val & 0x0f0f) + ((val >> 4) & 0x0f0f); |
|
202 val = (val & 0x00ff) + ((val >> 8) & 0x00ff); |
|
203 |
|
204 return val; |
|
205 } |
|
206 |
|
207 static always_inline int ctpop32(uint32_t val) |
|
208 { |
|
209 #if QEMU_GNUC_PREREQ(3, 4) |
|
210 return __builtin_popcount(val); |
|
211 #else |
|
212 val = (val & 0x55555555) + ((val >> 1) & 0x55555555); |
|
213 val = (val & 0x33333333) + ((val >> 2) & 0x33333333); |
|
214 val = (val & 0x0f0f0f0f) + ((val >> 4) & 0x0f0f0f0f); |
|
215 val = (val & 0x00ff00ff) + ((val >> 8) & 0x00ff00ff); |
|
216 val = (val & 0x0000ffff) + ((val >> 16) & 0x0000ffff); |
|
217 |
|
218 return val; |
|
219 #endif |
|
220 } |
|
221 |
|
222 static always_inline int ctpop64(uint64_t val) |
|
223 { |
|
224 #if QEMU_GNUC_PREREQ(3, 4) |
|
225 return __builtin_popcountll(val); |
|
226 #else |
|
227 val = (val & 0x5555555555555555ULL) + ((val >> 1) & 0x5555555555555555ULL); |
|
228 val = (val & 0x3333333333333333ULL) + ((val >> 2) & 0x3333333333333333ULL); |
|
229 val = (val & 0x0f0f0f0f0f0f0f0fULL) + ((val >> 4) & 0x0f0f0f0f0f0f0f0fULL); |
|
230 val = (val & 0x00ff00ff00ff00ffULL) + ((val >> 8) & 0x00ff00ff00ff00ffULL); |
|
231 val = (val & 0x0000ffff0000ffffULL) + ((val >> 16) & 0x0000ffff0000ffffULL); |
|
232 val = (val & 0x00000000ffffffffULL) + ((val >> 32) & 0x00000000ffffffffULL); |
|
233 |
|
234 return val; |
|
235 #endif |
|
236 } |