1 | /* |
1 | /* |
2 | * libecb - http://software.schmorp.de/pkg/libecb |
2 | * libecb - http://software.schmorp.de/pkg/libecb |
3 | * |
3 | * |
4 | * Copyright (©) 2009-2011 Marc Alexander Lehmann |
4 | * Copyright (©) 2009-2011 Marc Alexander Lehmann <libecb@schmorp.de> |
5 | * Copyright (©) 2011 Emanuele Giaquinta |
5 | * Copyright (©) 2011 Emanuele Giaquinta |
6 | * All rights reserved. |
6 | * All rights reserved. |
7 | * |
7 | * |
8 | * Redistribution and use in source and binary forms, with or without modifica- |
8 | * Redistribution and use in source and binary forms, with or without modifica- |
9 | * tion, are permitted provided that the following conditions are met: |
9 | * tion, are permitted provided that the following conditions are met: |
… | |
… | |
28 | */ |
28 | */ |
29 | |
29 | |
30 | #ifndef ECB_H |
30 | #ifndef ECB_H |
31 | #define ECB_H |
31 | #define ECB_H |
32 | |
32 | |
|
|
33 | #ifdef _WIN32 |
|
|
34 | typedef signed char int8_t; |
|
|
35 | typedef unsigned char uint8_t; |
|
|
36 | typedef signed short int16_t; |
|
|
37 | typedef unsigned short uint16_t; |
|
|
38 | typedef signed int int32_t; |
|
|
39 | typedef unsigned int uint32_t; |
|
|
40 | #if __GNUC__ |
|
|
41 | typedef signed long long int64_t; |
|
|
42 | typedef unsigned long long uint64_t; |
|
|
43 | #else /* _MSC_VER || __BORLANDC__ */ |
|
|
44 | typedef signed __int64 int64_t; |
|
|
45 | typedef unsigned __int64 uint64_t; |
|
|
46 | #endif |
|
|
47 | #else |
33 | #include <inttypes.h> |
48 | #include <inttypes.h> |
|
|
49 | #endif |
34 | |
50 | |
35 | /* many compilers define _GNUC_ to some versions but then only implement |
51 | /* many compilers define _GNUC_ to some versions but then only implement |
36 | * what their idiot authors think are the "more important" extensions, |
52 | * what their idiot authors think are the "more important" extensions, |
37 | * causing enourmous grief in return for some better fake benchmark numbers. |
53 | * causing enormous grief in return for some better fake benchmark numbers. |
38 | * or so. |
54 | * or so. |
39 | * 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 |
40 | * 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. |
41 | */ |
57 | */ |
42 | #ifndef ECB_GCC_VERSION |
58 | #ifndef ECB_GCC_VERSION |
43 | # if defined(__INTEL_COMPILER) || defined(__SUNPRO_C) || defined(__llvm__) |
59 | #if !defined(__GNUC_MINOR__) || defined(__INTEL_COMPILER) || defined(__SUNPRO_C) || defined(__SUNPRO_CC) || defined(__llvm__) || defined(__clang__) |
44 | # define ECB_GCC_VERSION(major,minor) 0 |
60 | #define ECB_GCC_VERSION(major,minor) 0 |
45 | # else |
61 | #else |
46 | # 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 |
47 | # endif |
64 | #endif |
|
|
65 | |
|
|
66 | /*****************************************************************************/ |
|
|
67 | |
|
|
68 | /* ECB_NO_THREADS - ecb is not used by multiple threads, ever */ |
|
|
69 | /* ECB_NO_SMP - ecb might be used in multiple threads, but only on a single cpu */ |
|
|
70 | |
|
|
71 | #if ECB_NO_THREADS || ECB_NO_SMP |
|
|
72 | #define ECB_MEMORY_FENCE do { } while (0) |
|
|
73 | #define ECB_MEMORY_FENCE_ACQUIRE do { } while (0) |
|
|
74 | #define ECB_MEMORY_FENCE_RELEASE do { } while (0) |
|
|
75 | #endif |
|
|
76 | |
|
|
77 | #ifndef ECB_MEMORY_FENCE |
|
|
78 | #if ECB_GCC_VERSION(2,5) |
|
|
79 | #if __x86 |
|
|
80 | #define ECB_MEMORY_FENCE __asm__ __volatile__ ("lock; orb $0, -1(%%esp)" : : : "memory") |
|
|
81 | #define ECB_MEMORY_FENCE_ACQUIRE ECB_MEMORY_FENCE /* non-lock xchg might be enough */ |
|
|
82 | #define ECB_MEMORY_FENCE_RELEASE do { } while (0) /* unlikely to change in future cpus */ |
|
|
83 | #elif __amd64 |
|
|
84 | #define ECB_MEMORY_FENCE __asm__ __volatile__ ("mfence" : : : "memory") |
|
|
85 | #define ECB_MEMORY_FENCE_ACQUIRE __asm__ __volatile__ ("lfence" : : : "memory") |
|
|
86 | #define ECB_MEMORY_FENCE_RELEASE __asm__ __volatile__ ("sfence") /* play safe - not needed in any current cpu */ |
|
|
87 | #elif __powerpc__ || __ppc__ || __powerpc64__ || __ppc64__ |
|
|
88 | #define ECB_MEMORY_FENCE __asm__ __volatile__ ("sync" : : : "memory") |
|
|
89 | #elif defined(__ARM_ARCH_6__ ) || defined(__ARM_ARCH_6J__ ) \ |
|
|
90 | || defined(__ARM_ARCH_6K__) || defined(__ARM_ARCH_6ZK__) \ |
|
|
91 | || defined(__ARM_ARCH_7__ ) || defined(__ARM_ARCH_7A__ ) \ |
|
|
92 | || defined(__ARM_ARCH_7M__) || defined(__ARM_ARCH_7R__ ) |
|
|
93 | #define ECB_MEMORY_FENCE \ |
|
|
94 | do { \ |
|
|
95 | int null = 0; \ |
|
|
96 | __asm__ __volatile__ ("mcr p15,0,%0,c6,c10,5", : "=&r" (null) : : "memory"); \ |
|
|
97 | while (0) |
|
|
98 | #endif |
48 | #endif |
99 | #endif |
|
|
100 | #endif |
49 | |
101 | |
|
|
102 | #ifndef ECB_MEMORY_FENCE |
|
|
103 | #if ECB_GCC_VERSION(4,4) || defined(__INTEL_COMPILER) |
|
|
104 | #define ECB_MEMORY_FENCE __sync_synchronize () |
|
|
105 | /*#define ECB_MEMORY_FENCE_ACQUIRE ({ char dummy = 0; __sync_lock_test_and_set (&dummy, 1); }) */ |
|
|
106 | /*#define ECB_MEMORY_FENCE_RELEASE ({ char dummy = 1; __sync_lock_release (&dummy ); }) */ |
|
|
107 | #elif _MSC_VER >= 1400 /* VC++ 2005 */ |
|
|
108 | #pragma intrinsic(_ReadBarrier,_WriteBarrier,_ReadWriteBarrier) |
|
|
109 | #define ECB_MEMORY_FENCE _ReadWriteBarrier () |
|
|
110 | #define ECB_MEMORY_FENCE_ACQUIRE _ReadWriteBarrier () /* according to msdn, _ReadBarrier is not a load fence */ |
|
|
111 | #define ECB_MEMORY_FENCE_RELEASE _WriteBarrier () |
|
|
112 | #elif defined(_WIN32) |
|
|
113 | #include <WinNT.h> |
|
|
114 | #define ECB_MEMORY_FENCE MemoryBarrier () /* actually just xchg on x86... scary */ |
|
|
115 | #endif |
|
|
116 | #endif |
|
|
117 | |
|
|
118 | #ifndef ECB_MEMORY_FENCE |
|
|
119 | #if !ECB_AVOID_PTHREADS |
|
|
120 | /* |
|
|
121 | * if you get undefined symbol references to pthread_mutex_lock, |
|
|
122 | * or failure to find pthread.h, then you should implement |
|
|
123 | * the ECB_MEMORY_FENCE operations for your cpu/compiler |
|
|
124 | * OR provide pthread.h and link against the posix thread library |
|
|
125 | * of your system. |
|
|
126 | */ |
|
|
127 | #include <pthread.h> |
|
|
128 | #define ECB_NEEDS_PTHREADS 1 |
|
|
129 | #define ECB_MEMORY_FENCE_NEEDS_PTHREADS 1 |
|
|
130 | |
|
|
131 | static pthread_mutex_t ecb_mf_lock = PTHREAD_MUTEX_INITIALIZER; |
|
|
132 | #define ECB_MEMORY_FENCE do { pthread_mutex_lock (&ecb_mf_lock); pthread_mutex_unlock (&ecb_mf_lock); } while (0) |
|
|
133 | #endif |
|
|
134 | #endif |
|
|
135 | |
|
|
136 | #if !defined(ECB_MEMORY_FENCE_ACQUIRE) && defined(ECB_MEMORY_FENCE) |
|
|
137 | #define ECB_MEMORY_FENCE_ACQUIRE ECB_MEMORY_FENCE |
|
|
138 | #endif |
|
|
139 | |
|
|
140 | #if !defined(ECB_MEMORY_FENCE_RELEASE) && defined(ECB_MEMORY_FENCE) |
|
|
141 | #define ECB_MEMORY_FENCE_RELEASE ECB_MEMORY_FENCE |
|
|
142 | #endif |
|
|
143 | |
|
|
144 | /*****************************************************************************/ |
|
|
145 | |
|
|
146 | #define ECB_C99 (__STDC_VERSION__ >= 199901L) |
|
|
147 | |
50 | #ifndef __cplusplus |
148 | #if __cplusplus |
51 | # if __STDC_VERSION__ >= 199901L |
149 | #define ecb_inline static inline |
52 | # define ECB_INLINE inline |
150 | #elif ECB_GCC_VERSION(2,5) |
53 | typedef _Bool ecb_bool; |
151 | #define ecb_inline static __inline__ |
|
|
152 | #elif ECB_C99 |
|
|
153 | #define ecb_inline static inline |
54 | # else |
154 | #else |
55 | # define ECB_INLINE inline /* yeah! */ |
155 | #define ecb_inline static |
|
|
156 | #endif |
|
|
157 | |
|
|
158 | #if ECB_GCC_VERSION(3,3) |
|
|
159 | #define ecb_restrict __restrict__ |
|
|
160 | #elif ECB_C99 |
|
|
161 | #define ecb_restrict restrict |
|
|
162 | #else |
|
|
163 | #define ecb_restrict |
|
|
164 | #endif |
|
|
165 | |
56 | typedef int ecb_bool; |
166 | typedef int ecb_bool; |
57 | # endif |
|
|
58 | #endif |
|
|
59 | |
167 | |
60 | #define ECB_CONCAT_(a, b) a ## b |
168 | #define ECB_CONCAT_(a, b) a ## b |
61 | #define ECB_CONCAT(a, b) ECB_CONCAT_(a, b) |
169 | #define ECB_CONCAT(a, b) ECB_CONCAT_(a, b) |
62 | #define ECB_STRINGIFY_(a) # a |
170 | #define ECB_STRINGIFY_(a) # a |
63 | #define ECB_STRINGIFY(a) ECB_STRINGIFY_(a) |
171 | #define ECB_STRINGIFY(a) ECB_STRINGIFY_(a) |
64 | |
172 | |
65 | #define ECB_HEADER_INLINE static ECB_INLINE |
173 | #define ecb_function_ ecb_inline |
66 | |
174 | |
67 | #if ECB_GCC_VERSION(3,1) |
175 | #if ECB_GCC_VERSION(3,1) |
68 | # define ecb_attribute(attrlist) __attribute__(attrlist) |
176 | #define ecb_attribute(attrlist) __attribute__(attrlist) |
69 | # define ecb_is_constant(expr) __builtin_constant_p (expr) |
177 | #define ecb_is_constant(expr) __builtin_constant_p (expr) |
70 | # define ecb_expect(expr,value) __builtin_expect ((expr),(value)) |
178 | #define ecb_expect(expr,value) __builtin_expect ((expr),(value)) |
71 | # define ecb_prefetch(addr,rw,locality) __builtin_prefetch (addr, rw, locality) |
179 | #define ecb_prefetch(addr,rw,locality) __builtin_prefetch (addr, rw, locality) |
72 | #else |
180 | #else |
73 | # define ecb_attribute(attrlist) |
181 | #define ecb_attribute(attrlist) |
74 | # define ecb_is_constant(expr) 0 |
182 | #define ecb_is_constant(expr) 0 |
75 | # define ecb_expect(expr,value) (expr) |
183 | #define ecb_expect(expr,value) (expr) |
76 | # define ecb_prefetch(addr,rw,locality) |
184 | #define ecb_prefetch(addr,rw,locality) |
77 | #endif |
185 | #endif |
78 | |
186 | |
79 | /* no emulation for ecb_decltype */ |
187 | /* no emulation for ecb_decltype */ |
80 | #if ECB_GCC_VERSION(4,5) |
188 | #if ECB_GCC_VERSION(4,5) |
81 | # define ecb_decltype(x) __decltype(x) |
189 | #define ecb_decltype(x) __decltype(x) |
82 | #elif ECB_GCC_VERSION(3,0) |
190 | #elif ECB_GCC_VERSION(3,0) |
83 | # define ecb_decltype(x) typeof(x) |
191 | #define ecb_decltype(x) __typeof(x) |
84 | #endif |
192 | #endif |
85 | |
193 | |
86 | #define ecb_noinline ecb_attribute ((noinline)) |
194 | #define ecb_noinline ecb_attribute ((__noinline__)) |
87 | #define ecb_noreturn ecb_attribute ((noreturn)) |
195 | #define ecb_noreturn ecb_attribute ((__noreturn__)) |
88 | #define ecb_unused ecb_attribute ((unused)) |
196 | #define ecb_unused ecb_attribute ((__unused__)) |
89 | #define ecb_const ecb_attribute ((const)) |
197 | #define ecb_const ecb_attribute ((__const__)) |
90 | #define ecb_pure ecb_attribute ((pure)) |
198 | #define ecb_pure ecb_attribute ((__pure__)) |
|
|
199 | |
|
|
200 | #if ECB_GCC_VERSION(4,3) |
|
|
201 | #define ecb_artificial ecb_attribute ((__artificial__)) |
91 | #define ecb_hot ecb_attribute ((hot)) /* 4.3 */ |
202 | #define ecb_hot ecb_attribute ((__hot__)) |
92 | #define ecb_cold ecb_attribute ((cold)) /* 4.3 */ |
203 | #define ecb_cold ecb_attribute ((__cold__)) |
|
|
204 | #else |
|
|
205 | #define ecb_artificial |
|
|
206 | #define ecb_hot |
|
|
207 | #define ecb_cold |
|
|
208 | #endif |
93 | |
209 | |
94 | /* put into if's if you are very sure that the expression */ |
210 | /* put around conditional expressions if you are very sure that the */ |
95 | /* is mostly true or mosty false. note that these return */ |
211 | /* expression is mostly true or mostly false. note that these return */ |
96 | /* booleans, not the expression. */ |
212 | /* booleans, not the expression. */ |
97 | #define ecb_unlikely(expr) ecb_expect (!!(expr), 0) |
213 | #define ecb_expect_false(expr) ecb_expect (!!(expr), 0) |
|
|
214 | #define ecb_expect_true(expr) ecb_expect (!!(expr), 1) |
|
|
215 | /* for compatibility to the rest of the world */ |
98 | #define ecb_likely(expr) ecb_expect (!!(expr), 1) |
216 | #define ecb_likely(expr) ecb_expect_true (expr) |
|
|
217 | #define ecb_unlikely(expr) ecb_expect_false (expr) |
|
|
218 | |
|
|
219 | /* count trailing zero bits and count # of one bits */ |
|
|
220 | #if ECB_GCC_VERSION(3,4) |
|
|
221 | /* we assume int == 32 bit, long == 32 or 64 bit and long long == 64 bit */ |
|
|
222 | #define ecb_ld32(x) (__builtin_clz (x) ^ 31) |
|
|
223 | #define ecb_ld64(x) (__builtin_clzll (x) ^ 63) |
|
|
224 | #define ecb_ctz32(x) __builtin_ctz (x) |
|
|
225 | #define ecb_ctz64(x) __builtin_ctzll (x) |
|
|
226 | #define ecb_popcount32(x) __builtin_popcount (x) |
|
|
227 | /* no popcountll */ |
|
|
228 | #else |
|
|
229 | ecb_function_ int ecb_ctz32 (uint32_t x) ecb_const; |
|
|
230 | ecb_function_ int |
|
|
231 | ecb_ctz32 (uint32_t x) |
|
|
232 | { |
|
|
233 | int r = 0; |
|
|
234 | |
|
|
235 | x &= ~x + 1; /* this isolates the lowest bit */ |
|
|
236 | |
|
|
237 | #if ECB_branchless_on_i386 |
|
|
238 | r += !!(x & 0xaaaaaaaa) << 0; |
|
|
239 | r += !!(x & 0xcccccccc) << 1; |
|
|
240 | r += !!(x & 0xf0f0f0f0) << 2; |
|
|
241 | r += !!(x & 0xff00ff00) << 3; |
|
|
242 | r += !!(x & 0xffff0000) << 4; |
|
|
243 | #else |
|
|
244 | if (x & 0xaaaaaaaa) r += 1; |
|
|
245 | if (x & 0xcccccccc) r += 2; |
|
|
246 | if (x & 0xf0f0f0f0) r += 4; |
|
|
247 | if (x & 0xff00ff00) r += 8; |
|
|
248 | if (x & 0xffff0000) r += 16; |
|
|
249 | #endif |
|
|
250 | |
|
|
251 | return r; |
|
|
252 | } |
|
|
253 | |
|
|
254 | ecb_function_ int ecb_ctz64 (uint64_t x) ecb_const; |
|
|
255 | ecb_function_ int |
|
|
256 | ecb_ctz64 (uint64_t x) |
|
|
257 | { |
|
|
258 | int shift = x & 0xffffffffU ? 0 : 32; |
|
|
259 | return ecb_ctz32 (x >> shift) + shift; |
|
|
260 | } |
|
|
261 | |
|
|
262 | ecb_function_ int ecb_popcount32 (uint32_t x) ecb_const; |
|
|
263 | ecb_function_ int |
|
|
264 | ecb_popcount32 (uint32_t x) |
|
|
265 | { |
|
|
266 | x -= (x >> 1) & 0x55555555; |
|
|
267 | x = ((x >> 2) & 0x33333333) + (x & 0x33333333); |
|
|
268 | x = ((x >> 4) + x) & 0x0f0f0f0f; |
|
|
269 | x *= 0x01010101; |
|
|
270 | |
|
|
271 | return x >> 24; |
|
|
272 | } |
|
|
273 | |
|
|
274 | ecb_function_ int ecb_ld32 (uint32_t x) ecb_const; |
|
|
275 | ecb_function_ int ecb_ld32 (uint32_t x) |
|
|
276 | { |
|
|
277 | int r = 0; |
|
|
278 | |
|
|
279 | if (x >> 16) { x >>= 16; r += 16; } |
|
|
280 | if (x >> 8) { x >>= 8; r += 8; } |
|
|
281 | if (x >> 4) { x >>= 4; r += 4; } |
|
|
282 | if (x >> 2) { x >>= 2; r += 2; } |
|
|
283 | if (x >> 1) { r += 1; } |
|
|
284 | |
|
|
285 | return r; |
|
|
286 | } |
|
|
287 | |
|
|
288 | ecb_function_ int ecb_ld64 (uint64_t x) ecb_const; |
|
|
289 | ecb_function_ int ecb_ld64 (uint64_t x) |
|
|
290 | { |
|
|
291 | int r = 0; |
|
|
292 | |
|
|
293 | if (x >> 32) { x >>= 32; r += 32; } |
|
|
294 | |
|
|
295 | return r + ecb_ld32 (x); |
|
|
296 | } |
|
|
297 | #endif |
|
|
298 | |
|
|
299 | /* popcount64 is only available on 64 bit cpus as gcc builtin */ |
|
|
300 | /* so for this version we are lazy */ |
|
|
301 | ecb_function_ int ecb_popcount64 (uint64_t x) ecb_const; |
|
|
302 | ecb_function_ int |
|
|
303 | ecb_popcount64 (uint64_t x) |
|
|
304 | { |
|
|
305 | return ecb_popcount32 (x) + ecb_popcount32 (x >> 32); |
|
|
306 | } |
|
|
307 | |
|
|
308 | ecb_inline uint8_t ecb_rotl8 (uint8_t x, unsigned int count) ecb_const; |
|
|
309 | ecb_inline uint8_t ecb_rotr8 (uint8_t x, unsigned int count) ecb_const; |
|
|
310 | ecb_inline uint16_t ecb_rotl16 (uint16_t x, unsigned int count) ecb_const; |
|
|
311 | ecb_inline uint16_t ecb_rotr16 (uint16_t x, unsigned int count) ecb_const; |
|
|
312 | ecb_inline uint32_t ecb_rotl32 (uint32_t x, unsigned int count) ecb_const; |
|
|
313 | ecb_inline uint32_t ecb_rotr32 (uint32_t x, unsigned int count) ecb_const; |
|
|
314 | ecb_inline uint64_t ecb_rotl64 (uint64_t x, unsigned int count) ecb_const; |
|
|
315 | ecb_inline uint64_t ecb_rotr64 (uint64_t x, unsigned int count) ecb_const; |
|
|
316 | |
|
|
317 | ecb_inline uint8_t ecb_rotl8 (uint8_t x, unsigned int count) { return (x >> ( 8 - count)) | (x << count); } |
|
|
318 | ecb_inline uint8_t ecb_rotr8 (uint8_t x, unsigned int count) { return (x << ( 8 - count)) | (x >> count); } |
|
|
319 | ecb_inline uint16_t ecb_rotl16 (uint16_t x, unsigned int count) { return (x >> (16 - count)) | (x << count); } |
|
|
320 | ecb_inline uint16_t ecb_rotr16 (uint16_t x, unsigned int count) { return (x << (16 - count)) | (x >> count); } |
|
|
321 | ecb_inline uint32_t ecb_rotl32 (uint32_t x, unsigned int count) { return (x >> (32 - count)) | (x << count); } |
|
|
322 | ecb_inline uint32_t ecb_rotr32 (uint32_t x, unsigned int count) { return (x << (32 - count)) | (x >> count); } |
|
|
323 | ecb_inline uint64_t ecb_rotl64 (uint64_t x, unsigned int count) { return (x >> (64 - count)) | (x << count); } |
|
|
324 | ecb_inline uint64_t ecb_rotr64 (uint64_t x, unsigned int count) { return (x << (64 - count)) | (x >> count); } |
|
|
325 | |
|
|
326 | #if ECB_GCC_VERSION(4,3) |
|
|
327 | #define ecb_bswap16(x) (__builtin_bswap32 (x) >> 16) |
|
|
328 | #define ecb_bswap32(x) __builtin_bswap32 (x) |
|
|
329 | #define ecb_bswap64(x) __builtin_bswap64 (x) |
|
|
330 | #else |
|
|
331 | ecb_function_ uint16_t ecb_bswap16 (uint16_t x) ecb_const; |
|
|
332 | ecb_function_ uint16_t |
|
|
333 | ecb_bswap16 (uint16_t x) |
|
|
334 | { |
|
|
335 | return ecb_rotl16 (x, 8); |
|
|
336 | } |
|
|
337 | |
|
|
338 | ecb_function_ uint32_t ecb_bswap32 (uint32_t x) ecb_const; |
|
|
339 | ecb_function_ uint32_t |
|
|
340 | ecb_bswap32 (uint32_t x) |
|
|
341 | { |
|
|
342 | return (((uint32_t)ecb_bswap16 (x)) << 16) | ecb_bswap16 (x >> 16); |
|
|
343 | } |
|
|
344 | |
|
|
345 | ecb_function_ uint64_t ecb_bswap64 (uint64_t x) ecb_const; |
|
|
346 | ecb_function_ uint64_t |
|
|
347 | ecb_bswap64 (uint64_t x) |
|
|
348 | { |
|
|
349 | return (((uint64_t)ecb_bswap32 (x)) << 32) | ecb_bswap32 (x >> 32); |
|
|
350 | } |
|
|
351 | #endif |
|
|
352 | |
|
|
353 | #if ECB_GCC_VERSION(4,5) |
|
|
354 | #define ecb_unreachable() __builtin_unreachable () |
|
|
355 | #else |
|
|
356 | /* this seems to work fine, but gcc always emits a warning for it :/ */ |
|
|
357 | ecb_function_ void ecb_unreachable (void) ecb_noreturn; |
|
|
358 | ecb_function_ void ecb_unreachable (void) { } |
|
|
359 | #endif |
99 | |
360 | |
100 | /* try to tell the compiler that some condition is definitely true */ |
361 | /* try to tell the compiler that some condition is definitely true */ |
101 | #define ecb_assume(cond) do { if (!(cond)) ecb_unreachable (); } while (0) |
362 | #define ecb_assume(cond) do { if (!(cond)) ecb_unreachable (); } while (0) |
102 | |
363 | |
103 | /* count trailing zero bits and count # of one bits */ |
|
|
104 | #if ECB_GCC_VERSION(3,4) |
|
|
105 | #define ecb_ctz32 (x) __builtin_ctz (x) |
|
|
106 | #define ecb_popcount32 (x) __builtin_popcount (x) |
|
|
107 | #else |
|
|
108 | ECB_HEADER_INLINE int ecb_ctz32 (uint32_t x) ecb_const; |
|
|
109 | ECB_HEADER_INLINE |
|
|
110 | ecb_ctz32 (uint32_t x) |
|
|
111 | { |
|
|
112 | int r = 0; |
|
|
113 | |
|
|
114 | x &= -x; /* this isolates the lowest bit */ |
|
|
115 | |
|
|
116 | if (x & 0xaaaaaaaa) r += 1; |
|
|
117 | if (x & 0xcccccccc) r += 2; |
|
|
118 | if (x & 0xf0f0f0f0) r += 4; |
|
|
119 | if (x & 0xff00ff00) r += 8; |
|
|
120 | if (x & 0xffff0000) r += 16; |
|
|
121 | |
|
|
122 | return r; |
|
|
123 | } |
|
|
124 | |
|
|
125 | ECB_HEADER_INLINE int ecb_popcount32 (uint32_t x) ecb_const; |
|
|
126 | ECB_HEADER_INLINE |
|
|
127 | ecb_popcount32 (uint32_t x) |
|
|
128 | { |
|
|
129 | x -= (x >> 1) & 0x55555555; |
|
|
130 | x = ((x >> 2) & 0x33333333) + (x & 0x33333333); |
|
|
131 | x = ((x >> 4) + x) & 0x0f0f0f0f; |
|
|
132 | x *= 0x01010101; |
|
|
133 | |
|
|
134 | return x >> 24; |
|
|
135 | } |
|
|
136 | #endif |
|
|
137 | |
|
|
138 | #if ECB_GCC_VERSION(4,3) |
|
|
139 | # define ecb_bswap32(x) __builtin_bswap32 (x) |
|
|
140 | # define ecb_bswap16(x) (__builtin_bswap32(x) >> 16) |
|
|
141 | #else |
|
|
142 | ECB_HEADER_INLINE uint32_t ecb_bswap32 (uint32_t x) ecb_const; |
|
|
143 | ECB_HEADER_INLINE uint32_t |
|
|
144 | ecb_bswap32 (uint32_t x) |
|
|
145 | { |
|
|
146 | return (x >> 24) |
|
|
147 | | ((x >> 8) & 0x0000ff00) |
|
|
148 | | ((x << 8) & 0x00ff0000) |
|
|
149 | | (x << 24); |
|
|
150 | } |
|
|
151 | |
|
|
152 | ECB_HEADER_INLINE uint32_t ecb_bswap16 (uint32_t x) ecb_const; |
|
|
153 | ECB_HEADER_INLINE uint32_t |
|
|
154 | ecb_bswap16 (uint32_t x) |
|
|
155 | { |
|
|
156 | return ((x >> 8) & 0xff) |
|
|
157 | | ((x << 8) & 0x00ff0000) |
|
|
158 | | (x << 24); |
|
|
159 | } |
|
|
160 | #endif |
|
|
161 | |
|
|
162 | #if ECB_GCC_VERSION(4,5) |
|
|
163 | # define ecb_unreachable() __builtin_unreachable () |
|
|
164 | #else |
|
|
165 | /* this seems to work fine, but gcc always emits a warning for it :/ */ |
|
|
166 | ECB_HEADER_INLINE void ecb_unreachable (void) ecb_noreturn; |
|
|
167 | ECB_HEADER_INLINE void ecb_unreachable (void) { } |
|
|
168 | #endif |
|
|
169 | |
|
|
170 | ECB_HEADER_INLINE unsigned char ecb_byteorder_helper (void) ecb_const; |
364 | ecb_function_ unsigned char ecb_byteorder_helper (void) ecb_const; |
171 | ECB_HEADER_INLINE unsigned char |
365 | ecb_function_ unsigned char |
172 | ecb_byteorder_helper (void) |
366 | ecb_byteorder_helper (void) |
173 | { |
367 | { |
174 | const uint32_t u = 0x11223344; |
368 | const uint32_t u = 0x11223344; |
175 | return *(unsigned char *)&u; |
369 | return *(unsigned char *)&u; |
176 | } |
370 | } |
177 | |
371 | |
178 | ECB_HEADER_INLINE ecb_bool ecb_big_endian (void) ecb_const; |
372 | ecb_function_ ecb_bool ecb_big_endian (void) ecb_const; |
179 | ECB_HEADER_INLINE ecb_bool ecb_big_endian (void) { return ecb_byteorder_helper () == 0x11; }; |
373 | ecb_function_ ecb_bool ecb_big_endian (void) { return ecb_byteorder_helper () == 0x11; } |
180 | ECB_HEADER_INLINE ecb_bool ecb_little_endian (void) ecb_const; |
374 | ecb_function_ ecb_bool ecb_little_endian (void) ecb_const; |
181 | ECB_HEADER_INLINE ecb_bool ecb_little_endian (void) { return ecb_byteorder_helper () == 0x44; }; |
375 | ecb_function_ ecb_bool ecb_little_endian (void) { return ecb_byteorder_helper () == 0x44; } |
182 | |
376 | |
|
|
377 | #if ECB_GCC_VERSION(3,0) || ECB_C99 |
|
|
378 | #define ecb_mod(m,n) ((m) % (n) + ((m) % (n) < 0 ? (n) : 0)) |
|
|
379 | #else |
183 | #define ecb_mod(m, n) ((m) < 0 ? ((n) - (-(m) % (n))) : ((m) % (n))) |
380 | #define ecb_mod(m,n) ((m) < 0 ? ((n) - 1 - ((-1 - (m)) % (n))) : ((m) % (n))) |
|
|
381 | #endif |
184 | |
382 | |
185 | #if ecb_cplusplus_does_not_suck |
383 | #if ecb_cplusplus_does_not_suck |
186 | // does not work for local types (http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2008/n2657.htm) |
384 | /* does not work for local types (http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2008/n2657.htm) */ |
187 | template<typename T, int N> |
385 | template<typename T, int N> |
188 | static inline int ecb_array_length (const T (&arr)[N]) |
386 | static inline int ecb_array_length (const T (&arr)[N]) |
189 | { |
387 | { |
190 | return N; |
388 | return N; |
191 | } |
389 | } |
192 | #else |
390 | #else |
193 | #define ecb_array_length(name) (sizeof (name) / sizeof (name [0])) |
391 | #define ecb_array_length(name) (sizeof (name) / sizeof (name [0])) |
194 | #endif |
392 | #endif |
195 | |
393 | |
196 | ECB_INLINE uint32_t ecb_rotr32 (uint32_t x, unsigned int count) ecb_const; |
|
|
197 | ECB_INLINE uint32_t |
|
|
198 | ecb_rotr32 (uint32_t x, unsigned int count) |
|
|
199 | { |
|
|
200 | return (x << (32 - count)) | (x >> count); |
|
|
201 | } |
|
|
202 | |
|
|
203 | ECB_INLINE uint32_t ecb_rotl32 (uint32_t x, unsigned int count) ecb_const; |
|
|
204 | ECB_INLINE uint32_t |
|
|
205 | ecb_rotl32 (uint32_t x, unsigned int count) |
|
|
206 | { |
|
|
207 | return (x >> (32 - count)) | (x << count); |
|
|
208 | } |
|
|
209 | |
|
|
210 | #endif |
394 | #endif |
211 | |
395 | |