… | |
… | |
38 | typedef signed int int32_t; |
38 | typedef signed int int32_t; |
39 | typedef unsigned int uint32_t; |
39 | typedef unsigned int uint32_t; |
40 | #if __GNUC__ |
40 | #if __GNUC__ |
41 | typedef signed long long int64_t; |
41 | typedef signed long long int64_t; |
42 | typedef unsigned long long uint64_t; |
42 | typedef unsigned long long uint64_t; |
43 | #else |
43 | #else /* _MSC_VER || __BORLANDC__ */ |
44 | typedef signed __int64 int64_t; |
44 | typedef signed __int64 int64_t; |
45 | typedef unsigned __int64 uint64_t; |
45 | typedef unsigned __int64 uint64_t; |
46 | #endif |
46 | #endif |
47 | #else |
47 | #else |
48 | #include <inttypes.h> |
48 | #include <inttypes.h> |
… | |
… | |
54 | * or so. |
54 | * or so. |
55 | * we try to detect these and simply assume they are not gcc - if they have |
55 | * we try to detect these and simply assume they are not gcc - if they have |
56 | * an issue with that they should have done it right in the first place. |
56 | * an issue with that they should have done it right in the first place. |
57 | */ |
57 | */ |
58 | #ifndef ECB_GCC_VERSION |
58 | #ifndef ECB_GCC_VERSION |
59 | #if defined(__INTEL_COMPILER) || defined(__SUNPRO_C) || defined(__SUNPRO_CC) || defined(__llvm__) || defined(__clang__) |
59 | #if !defined(__GNUC_MINOR__) || defined(__INTEL_COMPILER) || defined(__SUNPRO_C) || defined(__SUNPRO_CC) || defined(__llvm__) || defined(__clang__) |
60 | #define ECB_GCC_VERSION(major,minor) 0 |
60 | #define ECB_GCC_VERSION(major,minor) 0 |
61 | #else |
61 | #else |
62 | #define ECB_GCC_VERSION(major,minor) (__GNUC__ > (major) || (__GNUC__ == (major) && __GNUC_MINOR__ >= (minor))) |
62 | #define ECB_GCC_VERSION(major,minor) (__GNUC__ > (major) || (__GNUC__ == (major) && __GNUC_MINOR__ >= (minor))) |
63 | #endif |
63 | #endif |
64 | #endif |
64 | #endif |
|
|
65 | |
|
|
66 | /*****************************************************************************/ |
|
|
67 | |
|
|
68 | #ifndef ECB_MEMORY_FENCE |
|
|
69 | #if ECB_GCC_VERSION(2,5) |
|
|
70 | #if __x86 |
|
|
71 | #define ECB_MEMORY_FENCE __asm__ __volatile__ ("lock; or.b $0, -1(%%esp)" : : : "memory") |
|
|
72 | #define ECB_MEMORY_FENCE_ACQUIRE ECB_MEMORY_FENCE |
|
|
73 | #define ECB_MEMORY_FENCE_RELEASE ECB_MEMORY_FENCE /* better be safe than sorry */ |
|
|
74 | #elif __amd64 |
|
|
75 | #define ECB_MEMORY_FENCE __asm__ __volatile__ ("mfence" : : : "memory") |
|
|
76 | #define ECB_MEMORY_FENCE_ACQUIRE __asm__ __volatile__ ("lfence" : : : "memory") |
|
|
77 | #define ECB_MEMORY_FENCE_RELEASE __asm__ __volatile__ ("sfence") |
|
|
78 | #endif |
|
|
79 | #endif |
|
|
80 | #endif |
|
|
81 | |
|
|
82 | #ifndef ECB_MEMORY_FENCE |
|
|
83 | #if ECB_GCC_VERSION(4,4) |
|
|
84 | #define ECB_MEMORY_FENCE __sync_synchronize () |
|
|
85 | #define ECB_MEMORY_FENCE_ACQUIRE ({ char dummy = 0; __sync_lock_test_and_set (&dummy, 1); }) |
|
|
86 | #define ECB_MEMORY_FENCE_RELEASE ({ char dummy = 1; __sync_lock_release (&dummy ); }) |
|
|
87 | #elif _MSC_VER >= 1400 |
|
|
88 | #define ECB_MEMORY_FENCE do { } while (0) |
|
|
89 | #define ECB_MEMORY_FENCE_ACQUIRE ECB_MEMORY_FENCE |
|
|
90 | #define ECB_MEMORY_FENCE_RELEASE ECB_MEMORY_FENCE |
|
|
91 | #elif defined(_WIN32) && defined(MemoryBarrier) |
|
|
92 | #define ECB_MEMORY_FENCE MemoryBarrier () |
|
|
93 | #define ECB_MEMORY_FENCE_ACQUIRE ECB_MEMORY_FENCE |
|
|
94 | #define ECB_MEMORY_FENCE_RELEASE ECB_MEMORY_FENCE |
|
|
95 | #endif |
|
|
96 | #endif |
|
|
97 | |
|
|
98 | #ifndef ECB_MEMORY_FENCE |
|
|
99 | #include <pthread.h> |
|
|
100 | |
|
|
101 | static pthread_mutex_t ecb_mf_lock = PTHREAD_MUTEX_INITIALIZER; |
|
|
102 | #define ECB_MEMORY_FENCE do { pthread_mutex_lock (&ecb_mf_lock); pthread_mutex_unlock (&ecb_mf_lock); } while (0) |
|
|
103 | #define ECB_MEMORY_FENCE_ACQUIRE ECB_MEMORY_FENCE |
|
|
104 | #define ECB_MEMORY_FENCE_RELEASE ECB_MEMORY_FENCE |
|
|
105 | #endif |
|
|
106 | |
|
|
107 | /*****************************************************************************/ |
65 | |
108 | |
66 | #define ECB_C99 (__STDC_VERSION__ >= 199901L) |
109 | #define ECB_C99 (__STDC_VERSION__ >= 199901L) |
67 | |
110 | |
68 | #if __cplusplus |
111 | #if __cplusplus |
69 | #define ecb_inline static inline |
112 | #define ecb_inline static inline |
… | |
… | |
136 | #define ecb_likely(expr) ecb_expect_true (expr) |
179 | #define ecb_likely(expr) ecb_expect_true (expr) |
137 | #define ecb_unlikely(expr) ecb_expect_false (expr) |
180 | #define ecb_unlikely(expr) ecb_expect_false (expr) |
138 | |
181 | |
139 | /* count trailing zero bits and count # of one bits */ |
182 | /* count trailing zero bits and count # of one bits */ |
140 | #if ECB_GCC_VERSION(3,4) |
183 | #if ECB_GCC_VERSION(3,4) |
|
|
184 | /* we assume int == 32 bit, long == 32 or 64 bit and long long == 64 bit */ |
|
|
185 | #define ecb_ld32(x) (__builtin_clz (x) ^ 31) |
|
|
186 | #define ecb_ld64(x) (__builtin_clzll (x) ^ 63) |
141 | #define ecb_ctz32(x) __builtin_ctz (x) |
187 | #define ecb_ctz32(x) __builtin_ctz (x) |
|
|
188 | #define ecb_ctz64(x) __builtin_ctzll (x) |
142 | #define ecb_popcount32(x) __builtin_popcount (x) |
189 | #define ecb_popcount32(x) __builtin_popcount (x) |
|
|
190 | /* no popcountll */ |
143 | #else |
191 | #else |
144 | ecb_function_ int ecb_ctz32 (uint32_t x) ecb_const; |
192 | ecb_function_ int ecb_ctz32 (uint32_t x) ecb_const; |
145 | ecb_function_ int |
193 | ecb_function_ int |
146 | ecb_ctz32 (uint32_t x) |
194 | ecb_ctz32 (uint32_t x) |
147 | { |
195 | { |
148 | int r = 0; |
196 | int r = 0; |
149 | |
197 | |
150 | x &= (uint32_t)-(int32_t)x; /* this isolates the lowest bit */ |
198 | x &= ~x + 1; /* this isolates the lowest bit */ |
151 | |
199 | |
|
|
200 | #if ECB_branchless_on_i386 |
|
|
201 | r += !!(x & 0xaaaaaaaa) << 0; |
|
|
202 | r += !!(x & 0xcccccccc) << 1; |
|
|
203 | r += !!(x & 0xf0f0f0f0) << 2; |
|
|
204 | r += !!(x & 0xff00ff00) << 3; |
|
|
205 | r += !!(x & 0xffff0000) << 4; |
|
|
206 | #else |
152 | if (x & 0xaaaaaaaa) r += 1; |
207 | if (x & 0xaaaaaaaa) r += 1; |
153 | if (x & 0xcccccccc) r += 2; |
208 | if (x & 0xcccccccc) r += 2; |
154 | if (x & 0xf0f0f0f0) r += 4; |
209 | if (x & 0xf0f0f0f0) r += 4; |
155 | if (x & 0xff00ff00) r += 8; |
210 | if (x & 0xff00ff00) r += 8; |
156 | if (x & 0xffff0000) r += 16; |
211 | if (x & 0xffff0000) r += 16; |
|
|
212 | #endif |
157 | |
213 | |
158 | return r; |
214 | return r; |
|
|
215 | } |
|
|
216 | |
|
|
217 | ecb_function_ int ecb_ctz64 (uint64_t x) ecb_const; |
|
|
218 | ecb_function_ int |
|
|
219 | ecb_ctz64 (uint64_t x) |
|
|
220 | { |
|
|
221 | int shift = x & 0xffffffffU ? 0 : 32; |
|
|
222 | return ecb_ctz32 (x >> shift) + shift; |
159 | } |
223 | } |
160 | |
224 | |
161 | ecb_function_ int ecb_popcount32 (uint32_t x) ecb_const; |
225 | ecb_function_ int ecb_popcount32 (uint32_t x) ecb_const; |
162 | ecb_function_ int |
226 | ecb_function_ int |
163 | ecb_popcount32 (uint32_t x) |
227 | ecb_popcount32 (uint32_t x) |
… | |
… | |
167 | x = ((x >> 4) + x) & 0x0f0f0f0f; |
231 | x = ((x >> 4) + x) & 0x0f0f0f0f; |
168 | x *= 0x01010101; |
232 | x *= 0x01010101; |
169 | |
233 | |
170 | return x >> 24; |
234 | return x >> 24; |
171 | } |
235 | } |
|
|
236 | |
|
|
237 | /* you have the choice beetween something with a table lookup, */ |
|
|
238 | /* something using lots of bit arithmetic and a simple loop */ |
|
|
239 | /* we went for the loop */ |
|
|
240 | ecb_function_ int ecb_ld32 (uint32_t x) ecb_const; |
|
|
241 | ecb_function_ int ecb_ld32 (uint32_t x) |
|
|
242 | { |
|
|
243 | int r = 0; |
|
|
244 | |
|
|
245 | if (x >> 16) { x >>= 16; r += 16; } |
|
|
246 | if (x >> 8) { x >>= 8; r += 8; } |
|
|
247 | if (x >> 4) { x >>= 4; r += 4; } |
|
|
248 | if (x >> 2) { x >>= 2; r += 2; } |
|
|
249 | if (x >> 1) { r += 1; } |
|
|
250 | |
|
|
251 | return r; |
|
|
252 | } |
|
|
253 | |
|
|
254 | ecb_function_ int ecb_ld64 (uint64_t x) ecb_const; |
|
|
255 | ecb_function_ int ecb_ld64 (uint64_t x) |
|
|
256 | { |
|
|
257 | int r = 0; |
|
|
258 | |
|
|
259 | if (x >> 32) { x >>= 32; r += 32; } |
|
|
260 | |
|
|
261 | return r + ecb_ld32 (x); |
|
|
262 | } |
172 | #endif |
263 | #endif |
|
|
264 | |
|
|
265 | /* popcount64 is only available on 64 bit cpus as gcc builtin */ |
|
|
266 | /* so for this version we are lazy */ |
|
|
267 | ecb_function_ int ecb_popcount64 (uint64_t x) ecb_const; |
|
|
268 | ecb_function_ int |
|
|
269 | ecb_popcount64 (uint64_t x) |
|
|
270 | { |
|
|
271 | return ecb_popcount32 (x) + ecb_popcount32 (x >> 32); |
|
|
272 | } |
|
|
273 | |
|
|
274 | ecb_inline uint8_t ecb_rotl8 (uint8_t x, unsigned int count) ecb_const; |
|
|
275 | ecb_inline uint8_t ecb_rotr8 (uint8_t x, unsigned int count) ecb_const; |
|
|
276 | ecb_inline uint16_t ecb_rotl16 (uint16_t x, unsigned int count) ecb_const; |
|
|
277 | ecb_inline uint16_t ecb_rotr16 (uint16_t x, unsigned int count) ecb_const; |
|
|
278 | ecb_inline uint32_t ecb_rotl32 (uint32_t x, unsigned int count) ecb_const; |
|
|
279 | ecb_inline uint32_t ecb_rotr32 (uint32_t x, unsigned int count) ecb_const; |
|
|
280 | ecb_inline uint64_t ecb_rotl64 (uint64_t x, unsigned int count) ecb_const; |
|
|
281 | ecb_inline uint64_t ecb_rotr64 (uint64_t x, unsigned int count) ecb_const; |
|
|
282 | |
|
|
283 | ecb_inline uint8_t ecb_rotl8 (uint8_t x, unsigned int count) { return (x >> ( 8 - count)) | (x << count); } |
|
|
284 | ecb_inline uint8_t ecb_rotr8 (uint8_t x, unsigned int count) { return (x << ( 8 - count)) | (x >> count); } |
|
|
285 | ecb_inline uint16_t ecb_rotl16 (uint16_t x, unsigned int count) { return (x >> (16 - count)) | (x << count); } |
|
|
286 | ecb_inline uint16_t ecb_rotr16 (uint16_t x, unsigned int count) { return (x << (16 - count)) | (x >> count); } |
|
|
287 | ecb_inline uint32_t ecb_rotl32 (uint32_t x, unsigned int count) { return (x >> (32 - count)) | (x << count); } |
|
|
288 | ecb_inline uint32_t ecb_rotr32 (uint32_t x, unsigned int count) { return (x << (32 - count)) | (x >> count); } |
|
|
289 | ecb_inline uint64_t ecb_rotl64 (uint64_t x, unsigned int count) { return (x >> (64 - count)) | (x << count); } |
|
|
290 | ecb_inline uint64_t ecb_rotr64 (uint64_t x, unsigned int count) { return (x << (64 - count)) | (x >> count); } |
173 | |
291 | |
174 | #if ECB_GCC_VERSION(4,3) |
292 | #if ECB_GCC_VERSION(4,3) |
175 | #define ecb_bswap32(x) __builtin_bswap32 (x) |
|
|
176 | #define ecb_bswap16(x) (__builtin_bswap32(x) >> 16) |
293 | #define ecb_bswap16(x) (__builtin_bswap32 (x) >> 16) |
|
|
294 | #define ecb_bswap32(x) __builtin_bswap32 (x) |
|
|
295 | #define ecb_bswap64(x) __builtin_bswap64 (x) |
177 | #else |
296 | #else |
|
|
297 | ecb_function_ uint16_t ecb_bswap16 (uint16_t x) ecb_const; |
|
|
298 | ecb_function_ uint16_t |
|
|
299 | ecb_bswap16 (uint16_t x) |
|
|
300 | { |
|
|
301 | return ecb_rotl16 (x, 8); |
|
|
302 | } |
|
|
303 | |
178 | ecb_function_ uint32_t ecb_bswap32 (uint32_t x) ecb_const; |
304 | ecb_function_ uint32_t ecb_bswap32 (uint32_t x) ecb_const; |
179 | ecb_function_ uint32_t |
305 | ecb_function_ uint32_t |
180 | ecb_bswap32 (uint32_t x) |
306 | ecb_bswap32 (uint32_t x) |
181 | { |
307 | { |
182 | return (x >> 24) |
308 | return (((uint32_t)ecb_bswap16 (x)) << 16) | ecb_bswap16 (x >> 16); |
183 | | ((x >> 8) & 0x0000ff00) |
|
|
184 | | ((x << 8) & 0x00ff0000) |
|
|
185 | | (x << 24); |
|
|
186 | } |
309 | } |
187 | |
310 | |
188 | ecb_function_ uint32_t ecb_bswap16 (uint32_t x) ecb_const; |
311 | ecb_function_ uint64_t ecb_bswap64 (uint64_t x) ecb_const; |
189 | ecb_function_ uint32_t |
312 | ecb_function_ uint64_t |
190 | ecb_bswap16 (uint32_t x) |
313 | ecb_bswap64 (uint64_t x) |
191 | { |
314 | { |
192 | return ((x >> 8) & 0xff) |
315 | return (((uint64_t)ecb_bswap32 (x)) << 32) | ecb_bswap32 (x >> 32); |
193 | | ((x << 8) & 0x00ff0000) |
|
|
194 | | (x << 24); |
|
|
195 | } |
316 | } |
196 | #endif |
317 | #endif |
197 | |
318 | |
198 | #if ECB_GCC_VERSION(4,5) |
319 | #if ECB_GCC_VERSION(4,5) |
199 | #define ecb_unreachable() __builtin_unreachable () |
320 | #define ecb_unreachable() __builtin_unreachable () |
… | |
… | |
234 | } |
355 | } |
235 | #else |
356 | #else |
236 | #define ecb_array_length(name) (sizeof (name) / sizeof (name [0])) |
357 | #define ecb_array_length(name) (sizeof (name) / sizeof (name [0])) |
237 | #endif |
358 | #endif |
238 | |
359 | |
239 | ecb_inline uint32_t ecb_rotr32 (uint32_t x, unsigned int count) ecb_const; |
|
|
240 | ecb_inline uint32_t |
|
|
241 | ecb_rotr32 (uint32_t x, unsigned int count) |
|
|
242 | { |
|
|
243 | return (x << (32 - count)) | (x >> count); |
|
|
244 | } |
|
|
245 | |
|
|
246 | ecb_inline uint32_t ecb_rotl32 (uint32_t x, unsigned int count) ecb_const; |
|
|
247 | ecb_inline uint32_t |
|
|
248 | ecb_rotl32 (uint32_t x, unsigned int count) |
|
|
249 | { |
|
|
250 | return (x >> (32 - count)) | (x << count); |
|
|
251 | } |
|
|
252 | |
|
|
253 | #endif |
360 | #endif |
254 | |
361 | |