1 | /* |
1 | /* |
2 | * libecb - http://software.schmorp.de/pkg/libecb |
2 | * libecb - http://software.schmorp.de/pkg/libecb |
3 | * |
3 | * |
4 | * Copyright (©) 2009-2014 Marc Alexander Lehmann <libecb@schmorp.de> |
4 | * Copyright (©) 2009-2015 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: |
… | |
… | |
74 | #else |
74 | #else |
75 | #define ECB_PTRSIZE 4 |
75 | #define ECB_PTRSIZE 4 |
76 | #endif |
76 | #endif |
77 | #endif |
77 | #endif |
78 | |
78 | |
|
|
79 | #define ECB_GCC_AMD64 (__amd64 || __amd64__ || __x86_64 || __x86_64__) |
|
|
80 | #define ECB_MSVC_AMD64 (_M_AMD64 || _M_X64) |
|
|
81 | |
79 | /* work around x32 idiocy by defining proper macros */ |
82 | /* work around x32 idiocy by defining proper macros */ |
80 | #if __amd64 || __x86_64 || _M_AMD64 || _M_X64 |
83 | #if ECB_GCC_AMD64 || ECB_MSVC_AMD64 |
81 | #if _ILP32 |
84 | #if _ILP32 |
82 | #define ECB_AMD64_X32 1 |
85 | #define ECB_AMD64_X32 1 |
83 | #else |
86 | #else |
84 | #define ECB_AMD64 1 |
87 | #define ECB_AMD64 1 |
85 | #endif |
88 | #endif |
… | |
… | |
98 | #define ECB_GCC_VERSION(major,minor) (__GNUC__ > (major) || (__GNUC__ == (major) && __GNUC_MINOR__ >= (minor))) |
101 | #define ECB_GCC_VERSION(major,minor) (__GNUC__ > (major) || (__GNUC__ == (major) && __GNUC_MINOR__ >= (minor))) |
99 | #endif |
102 | #endif |
100 | |
103 | |
101 | #define ECB_CLANG_VERSION(major,minor) (__clang_major__ > (major) || (__clang_major__ == (major) && __clang_minor__ >= (minor))) |
104 | #define ECB_CLANG_VERSION(major,minor) (__clang_major__ > (major) || (__clang_major__ == (major) && __clang_minor__ >= (minor))) |
102 | |
105 | |
103 | #if __clang__ && defined(__has_builtin) |
106 | #if __clang__ && defined __has_builtin |
104 | #define ECB_CLANG_BUILTIN(x) __has_builtin(x) |
107 | #define ECB_CLANG_BUILTIN(x) __has_builtin (x) |
105 | #else |
108 | #else |
106 | #define ECB_CLANG_BUILTIN(x) 0 |
109 | #define ECB_CLANG_BUILTIN(x) 0 |
107 | #endif |
110 | #endif |
108 | |
111 | |
109 | #if __clang__ && defined(__has_extension) |
112 | #if __clang__ && defined __has_extension |
110 | #define ECB_CLANG_EXTENSION(x) __has_extension(x) |
113 | #define ECB_CLANG_EXTENSION(x) __has_extension (x) |
111 | #else |
114 | #else |
112 | #define ECB_CLANG_EXTENSION(x) 0 |
115 | #define ECB_CLANG_EXTENSION(x) 0 |
113 | #endif |
116 | #endif |
114 | |
117 | |
115 | #define ECB_CPP (__cplusplus+0) |
118 | #define ECB_CPP (__cplusplus+0) |
… | |
… | |
145 | #define ECB_NO_SMP 1 |
148 | #define ECB_NO_SMP 1 |
146 | #endif |
149 | #endif |
147 | |
150 | |
148 | #if ECB_NO_SMP |
151 | #if ECB_NO_SMP |
149 | #define ECB_MEMORY_FENCE do { } while (0) |
152 | #define ECB_MEMORY_FENCE do { } while (0) |
|
|
153 | #endif |
|
|
154 | |
|
|
155 | /* http://www-01.ibm.com/support/knowledgecenter/SSGH3R_13.1.0/com.ibm.xlcpp131.aix.doc/compiler_ref/compiler_builtins.html */ |
|
|
156 | #if __xlC__ && ECB_CPP |
|
|
157 | #include <builtins.h> |
150 | #endif |
158 | #endif |
151 | |
159 | |
152 | #ifndef ECB_MEMORY_FENCE |
160 | #ifndef ECB_MEMORY_FENCE |
153 | #if ECB_GCC_VERSION(2,5) || defined __INTEL_COMPILER || (__llvm__ && __GNUC__) || __SUNPRO_C >= 0x5110 || __SUNPRO_CC >= 0x5110 |
161 | #if ECB_GCC_VERSION(2,5) || defined __INTEL_COMPILER || (__llvm__ && __GNUC__) || __SUNPRO_C >= 0x5110 || __SUNPRO_CC >= 0x5110 |
154 | #if __i386 || __i386__ |
162 | #if __i386 || __i386__ |
155 | #define ECB_MEMORY_FENCE __asm__ __volatile__ ("lock; orb $0, -1(%%esp)" : : : "memory") |
163 | #define ECB_MEMORY_FENCE __asm__ __volatile__ ("lock; orb $0, -1(%%esp)" : : : "memory") |
156 | #define ECB_MEMORY_FENCE_ACQUIRE __asm__ __volatile__ ("" : : : "memory") |
164 | #define ECB_MEMORY_FENCE_ACQUIRE __asm__ __volatile__ ("" : : : "memory") |
157 | #define ECB_MEMORY_FENCE_RELEASE __asm__ __volatile__ ("") |
165 | #define ECB_MEMORY_FENCE_RELEASE __asm__ __volatile__ ("") |
158 | #elif __amd64 || __amd64__ || __x86_64 || __x86_64__ |
166 | #elif ECB_GCC_AMD64 |
159 | #define ECB_MEMORY_FENCE __asm__ __volatile__ ("mfence" : : : "memory") |
167 | #define ECB_MEMORY_FENCE __asm__ __volatile__ ("mfence" : : : "memory") |
160 | #define ECB_MEMORY_FENCE_ACQUIRE __asm__ __volatile__ ("" : : : "memory") |
168 | #define ECB_MEMORY_FENCE_ACQUIRE __asm__ __volatile__ ("" : : : "memory") |
161 | #define ECB_MEMORY_FENCE_RELEASE __asm__ __volatile__ ("") |
169 | #define ECB_MEMORY_FENCE_RELEASE __asm__ __volatile__ ("") |
162 | #elif __powerpc__ || __ppc__ || __powerpc64__ || __ppc64__ |
170 | #elif __powerpc__ || __ppc__ || __powerpc64__ || __ppc64__ |
163 | #define ECB_MEMORY_FENCE __asm__ __volatile__ ("sync" : : : "memory") |
171 | #define ECB_MEMORY_FENCE __asm__ __volatile__ ("sync" : : : "memory") |
… | |
… | |
278 | #define ECB_MEMORY_FENCE_RELEASE ECB_MEMORY_FENCE |
286 | #define ECB_MEMORY_FENCE_RELEASE ECB_MEMORY_FENCE |
279 | #endif |
287 | #endif |
280 | |
288 | |
281 | /*****************************************************************************/ |
289 | /*****************************************************************************/ |
282 | |
290 | |
283 | #if __cplusplus |
291 | #if ECB_CPP |
284 | #define ecb_inline static inline |
292 | #define ecb_inline static inline |
285 | #elif ECB_GCC_VERSION(2,5) |
293 | #elif ECB_GCC_VERSION(2,5) |
286 | #define ecb_inline static __inline__ |
294 | #define ecb_inline static __inline__ |
287 | #elif ECB_C99 |
295 | #elif ECB_C99 |
288 | #define ecb_inline static inline |
296 | #define ecb_inline static inline |
… | |
… | |
302 | |
310 | |
303 | #define ECB_CONCAT_(a, b) a ## b |
311 | #define ECB_CONCAT_(a, b) a ## b |
304 | #define ECB_CONCAT(a, b) ECB_CONCAT_(a, b) |
312 | #define ECB_CONCAT(a, b) ECB_CONCAT_(a, b) |
305 | #define ECB_STRINGIFY_(a) # a |
313 | #define ECB_STRINGIFY_(a) # a |
306 | #define ECB_STRINGIFY(a) ECB_STRINGIFY_(a) |
314 | #define ECB_STRINGIFY(a) ECB_STRINGIFY_(a) |
|
|
315 | #define ECB_STRINGIFY_EXPR(expr) ((expr), ECB_STRINGIFY_ (expr)) |
307 | |
316 | |
308 | #define ecb_function_ ecb_inline |
317 | #define ecb_function_ ecb_inline |
309 | |
318 | |
310 | #if ECB_GCC_VERSION(3,1) || ECB_CLANG_VERSION(2,8) |
319 | #if ECB_GCC_VERSION(3,1) || ECB_CLANG_VERSION(2,8) |
311 | #define ecb_attribute(attrlist) __attribute__ (attrlist) |
320 | #define ecb_attribute(attrlist) __attribute__ (attrlist) |
… | |
… | |
336 | #endif |
345 | #endif |
337 | |
346 | |
338 | /* no emulation for ecb_decltype */ |
347 | /* no emulation for ecb_decltype */ |
339 | #if ECB_CPP11 |
348 | #if ECB_CPP11 |
340 | // older implementations might have problems with decltype(x)::type, work around it |
349 | // older implementations might have problems with decltype(x)::type, work around it |
341 | struct ecb_decltype_t { typedef T type; }; |
350 | template<class T> struct ecb_decltype_t { typedef T type; }; |
342 | #define ecb_decltype(x) ecb_decltype_t<decltype(x)>::type |
351 | #define ecb_decltype(x) ecb_decltype_t<decltype (x)>::type |
343 | #elif ECB_GCC_VERSION(3,0) || ECB_CLANG_VERSION(2,8) |
352 | #elif ECB_GCC_VERSION(3,0) || ECB_CLANG_VERSION(2,8) |
344 | #define ecb_decltype(x) __typeof__ (x) |
353 | #define ecb_decltype(x) __typeof__ (x) |
345 | #endif |
354 | #endif |
346 | |
355 | |
347 | #if _MSC_VER >= 1300 |
356 | #if _MSC_VER >= 1300 |
348 | #define ecb_deprecated __declspec(deprecated) |
357 | #define ecb_deprecated __declspec (deprecated) |
349 | #else |
358 | #else |
350 | #define ecb_deprecated ecb_attribute ((__deprecated__)) |
359 | #define ecb_deprecated ecb_attribute ((__deprecated__)) |
351 | #endif |
360 | #endif |
352 | |
361 | |
|
|
362 | #if _MSC_VER >= 1500 |
|
|
363 | #define ecb_deprecated_message(msg) __declspec (deprecated (msg)) |
|
|
364 | #elif ECB_GCC_VERSION(4,5) |
|
|
365 | #define ecb_deprecated_message(msg) ecb_attribute ((__deprecated__ (msg)) |
|
|
366 | #else |
|
|
367 | #define ecb_deprecated_message(msg) ecb_deprecated |
|
|
368 | #endif |
|
|
369 | |
|
|
370 | #if _MSC_VER >= 1400 |
|
|
371 | #define ecb_noinline __declspec (noinline) |
|
|
372 | #else |
353 | #define ecb_noinline ecb_attribute ((__noinline__)) |
373 | #define ecb_noinline ecb_attribute ((__noinline__)) |
|
|
374 | #endif |
|
|
375 | |
354 | #define ecb_unused ecb_attribute ((__unused__)) |
376 | #define ecb_unused ecb_attribute ((__unused__)) |
355 | #define ecb_const ecb_attribute ((__const__)) |
377 | #define ecb_const ecb_attribute ((__const__)) |
356 | #define ecb_pure ecb_attribute ((__pure__)) |
378 | #define ecb_pure ecb_attribute ((__pure__)) |
357 | |
379 | |
358 | /* http://msdn.microsoft.com/en-us/library/k6ktzx3s.aspx __declspec(noreturn) */ |
380 | #if ECB_C11 || __IBMC_NORETURN |
359 | #if ECB_C11 |
381 | /* http://www-01.ibm.com/support/knowledgecenter/SSGH3R_13.1.0/com.ibm.xlcpp131.aix.doc/language_ref/noreturn.html */ |
360 | #define ecb_noreturn _Noreturn |
382 | #define ecb_noreturn _Noreturn |
|
|
383 | #elif ECB_CPP11 |
|
|
384 | #define ecb_noreturn [[noreturn]] |
|
|
385 | #elif _MSC_VER >= 1200 |
|
|
386 | /* http://msdn.microsoft.com/en-us/library/k6ktzx3s.aspx */ |
|
|
387 | #define ecb_noreturn __declspec (noreturn) |
361 | #else |
388 | #else |
362 | #define ecb_noreturn ecb_attribute ((__noreturn__)) |
389 | #define ecb_noreturn ecb_attribute ((__noreturn__)) |
363 | #endif |
390 | #endif |
364 | |
391 | |
365 | #if ECB_GCC_VERSION(4,3) |
392 | #if ECB_GCC_VERSION(4,3) |
… | |
… | |
392 | #define ecb_ctz32(x) __builtin_ctz (x) |
419 | #define ecb_ctz32(x) __builtin_ctz (x) |
393 | #define ecb_ctz64(x) __builtin_ctzll (x) |
420 | #define ecb_ctz64(x) __builtin_ctzll (x) |
394 | #define ecb_popcount32(x) __builtin_popcount (x) |
421 | #define ecb_popcount32(x) __builtin_popcount (x) |
395 | /* no popcountll */ |
422 | /* no popcountll */ |
396 | #else |
423 | #else |
397 | ecb_function_ int ecb_ctz32 (uint32_t x) ecb_const; |
424 | ecb_function_ ecb_const int ecb_ctz32 (uint32_t x); |
398 | ecb_function_ int |
425 | ecb_function_ ecb_const int |
399 | ecb_ctz32 (uint32_t x) |
426 | ecb_ctz32 (uint32_t x) |
400 | { |
427 | { |
401 | int r = 0; |
428 | int r = 0; |
402 | |
429 | |
403 | x &= ~x + 1; /* this isolates the lowest bit */ |
430 | x &= ~x + 1; /* this isolates the lowest bit */ |
… | |
… | |
417 | #endif |
444 | #endif |
418 | |
445 | |
419 | return r; |
446 | return r; |
420 | } |
447 | } |
421 | |
448 | |
422 | ecb_function_ int ecb_ctz64 (uint64_t x) ecb_const; |
449 | ecb_function_ ecb_const int ecb_ctz64 (uint64_t x); |
423 | ecb_function_ int |
450 | ecb_function_ ecb_const int |
424 | ecb_ctz64 (uint64_t x) |
451 | ecb_ctz64 (uint64_t x) |
425 | { |
452 | { |
426 | int shift = x & 0xffffffffU ? 0 : 32; |
453 | int shift = x & 0xffffffffU ? 0 : 32; |
427 | return ecb_ctz32 (x >> shift) + shift; |
454 | return ecb_ctz32 (x >> shift) + shift; |
428 | } |
455 | } |
429 | |
456 | |
430 | ecb_function_ int ecb_popcount32 (uint32_t x) ecb_const; |
457 | ecb_function_ ecb_const int ecb_popcount32 (uint32_t x); |
431 | ecb_function_ int |
458 | ecb_function_ ecb_const int |
432 | ecb_popcount32 (uint32_t x) |
459 | ecb_popcount32 (uint32_t x) |
433 | { |
460 | { |
434 | x -= (x >> 1) & 0x55555555; |
461 | x -= (x >> 1) & 0x55555555; |
435 | x = ((x >> 2) & 0x33333333) + (x & 0x33333333); |
462 | x = ((x >> 2) & 0x33333333) + (x & 0x33333333); |
436 | x = ((x >> 4) + x) & 0x0f0f0f0f; |
463 | x = ((x >> 4) + x) & 0x0f0f0f0f; |
437 | x *= 0x01010101; |
464 | x *= 0x01010101; |
438 | |
465 | |
439 | return x >> 24; |
466 | return x >> 24; |
440 | } |
467 | } |
441 | |
468 | |
442 | ecb_function_ int ecb_ld32 (uint32_t x) ecb_const; |
469 | ecb_function_ ecb_const int ecb_ld32 (uint32_t x); |
443 | ecb_function_ int ecb_ld32 (uint32_t x) |
470 | ecb_function_ ecb_const int ecb_ld32 (uint32_t x) |
444 | { |
471 | { |
445 | int r = 0; |
472 | int r = 0; |
446 | |
473 | |
447 | if (x >> 16) { x >>= 16; r += 16; } |
474 | if (x >> 16) { x >>= 16; r += 16; } |
448 | if (x >> 8) { x >>= 8; r += 8; } |
475 | if (x >> 8) { x >>= 8; r += 8; } |
… | |
… | |
451 | if (x >> 1) { r += 1; } |
478 | if (x >> 1) { r += 1; } |
452 | |
479 | |
453 | return r; |
480 | return r; |
454 | } |
481 | } |
455 | |
482 | |
456 | ecb_function_ int ecb_ld64 (uint64_t x) ecb_const; |
483 | ecb_function_ ecb_const int ecb_ld64 (uint64_t x); |
457 | ecb_function_ int ecb_ld64 (uint64_t x) |
484 | ecb_function_ ecb_const int ecb_ld64 (uint64_t x) |
458 | { |
485 | { |
459 | int r = 0; |
486 | int r = 0; |
460 | |
487 | |
461 | if (x >> 32) { x >>= 32; r += 32; } |
488 | if (x >> 32) { x >>= 32; r += 32; } |
462 | |
489 | |
463 | return r + ecb_ld32 (x); |
490 | return r + ecb_ld32 (x); |
464 | } |
491 | } |
465 | #endif |
492 | #endif |
466 | |
493 | |
467 | ecb_function_ ecb_bool ecb_is_pot32 (uint32_t x) ecb_const; |
494 | ecb_function_ ecb_const ecb_bool ecb_is_pot32 (uint32_t x); |
468 | ecb_function_ ecb_bool ecb_is_pot32 (uint32_t x) { return !(x & (x - 1)); } |
495 | ecb_function_ ecb_const ecb_bool ecb_is_pot32 (uint32_t x) { return !(x & (x - 1)); } |
469 | ecb_function_ ecb_bool ecb_is_pot64 (uint64_t x) ecb_const; |
496 | ecb_function_ ecb_const ecb_bool ecb_is_pot64 (uint64_t x); |
470 | ecb_function_ ecb_bool ecb_is_pot64 (uint64_t x) { return !(x & (x - 1)); } |
497 | ecb_function_ ecb_const ecb_bool ecb_is_pot64 (uint64_t x) { return !(x & (x - 1)); } |
471 | |
498 | |
472 | ecb_function_ uint8_t ecb_bitrev8 (uint8_t x) ecb_const; |
499 | ecb_function_ ecb_const uint8_t ecb_bitrev8 (uint8_t x); |
473 | ecb_function_ uint8_t ecb_bitrev8 (uint8_t x) |
500 | ecb_function_ ecb_const uint8_t ecb_bitrev8 (uint8_t x) |
474 | { |
501 | { |
475 | return ( (x * 0x0802U & 0x22110U) |
502 | return ( (x * 0x0802U & 0x22110U) |
476 | | (x * 0x8020U & 0x88440U)) * 0x10101U >> 16; |
503 | | (x * 0x8020U & 0x88440U)) * 0x10101U >> 16; |
477 | } |
504 | } |
478 | |
505 | |
479 | ecb_function_ uint16_t ecb_bitrev16 (uint16_t x) ecb_const; |
506 | ecb_function_ ecb_const uint16_t ecb_bitrev16 (uint16_t x); |
480 | ecb_function_ uint16_t ecb_bitrev16 (uint16_t x) |
507 | ecb_function_ ecb_const uint16_t ecb_bitrev16 (uint16_t x) |
481 | { |
508 | { |
482 | x = ((x >> 1) & 0x5555) | ((x & 0x5555) << 1); |
509 | x = ((x >> 1) & 0x5555) | ((x & 0x5555) << 1); |
483 | x = ((x >> 2) & 0x3333) | ((x & 0x3333) << 2); |
510 | x = ((x >> 2) & 0x3333) | ((x & 0x3333) << 2); |
484 | x = ((x >> 4) & 0x0f0f) | ((x & 0x0f0f) << 4); |
511 | x = ((x >> 4) & 0x0f0f) | ((x & 0x0f0f) << 4); |
485 | x = ( x >> 8 ) | ( x << 8); |
512 | x = ( x >> 8 ) | ( x << 8); |
486 | |
513 | |
487 | return x; |
514 | return x; |
488 | } |
515 | } |
489 | |
516 | |
490 | ecb_function_ uint32_t ecb_bitrev32 (uint32_t x) ecb_const; |
517 | ecb_function_ ecb_const uint32_t ecb_bitrev32 (uint32_t x); |
491 | ecb_function_ uint32_t ecb_bitrev32 (uint32_t x) |
518 | ecb_function_ ecb_const uint32_t ecb_bitrev32 (uint32_t x) |
492 | { |
519 | { |
493 | x = ((x >> 1) & 0x55555555) | ((x & 0x55555555) << 1); |
520 | x = ((x >> 1) & 0x55555555) | ((x & 0x55555555) << 1); |
494 | x = ((x >> 2) & 0x33333333) | ((x & 0x33333333) << 2); |
521 | x = ((x >> 2) & 0x33333333) | ((x & 0x33333333) << 2); |
495 | x = ((x >> 4) & 0x0f0f0f0f) | ((x & 0x0f0f0f0f) << 4); |
522 | x = ((x >> 4) & 0x0f0f0f0f) | ((x & 0x0f0f0f0f) << 4); |
496 | x = ((x >> 8) & 0x00ff00ff) | ((x & 0x00ff00ff) << 8); |
523 | x = ((x >> 8) & 0x00ff00ff) | ((x & 0x00ff00ff) << 8); |
… | |
… | |
499 | return x; |
526 | return x; |
500 | } |
527 | } |
501 | |
528 | |
502 | /* popcount64 is only available on 64 bit cpus as gcc builtin */ |
529 | /* popcount64 is only available on 64 bit cpus as gcc builtin */ |
503 | /* so for this version we are lazy */ |
530 | /* so for this version we are lazy */ |
504 | ecb_function_ int ecb_popcount64 (uint64_t x) ecb_const; |
531 | ecb_function_ ecb_const int ecb_popcount64 (uint64_t x); |
505 | ecb_function_ int |
532 | ecb_function_ ecb_const int |
506 | ecb_popcount64 (uint64_t x) |
533 | ecb_popcount64 (uint64_t x) |
507 | { |
534 | { |
508 | return ecb_popcount32 (x) + ecb_popcount32 (x >> 32); |
535 | return ecb_popcount32 (x) + ecb_popcount32 (x >> 32); |
509 | } |
536 | } |
510 | |
537 | |
511 | ecb_inline uint8_t ecb_rotl8 (uint8_t x, unsigned int count) ecb_const; |
538 | ecb_inline ecb_const uint8_t ecb_rotl8 (uint8_t x, unsigned int count); |
512 | ecb_inline uint8_t ecb_rotr8 (uint8_t x, unsigned int count) ecb_const; |
539 | ecb_inline ecb_const uint8_t ecb_rotr8 (uint8_t x, unsigned int count); |
513 | ecb_inline uint16_t ecb_rotl16 (uint16_t x, unsigned int count) ecb_const; |
540 | ecb_inline ecb_const uint16_t ecb_rotl16 (uint16_t x, unsigned int count); |
514 | ecb_inline uint16_t ecb_rotr16 (uint16_t x, unsigned int count) ecb_const; |
541 | ecb_inline ecb_const uint16_t ecb_rotr16 (uint16_t x, unsigned int count); |
515 | ecb_inline uint32_t ecb_rotl32 (uint32_t x, unsigned int count) ecb_const; |
542 | ecb_inline ecb_const uint32_t ecb_rotl32 (uint32_t x, unsigned int count); |
516 | ecb_inline uint32_t ecb_rotr32 (uint32_t x, unsigned int count) ecb_const; |
543 | ecb_inline ecb_const uint32_t ecb_rotr32 (uint32_t x, unsigned int count); |
517 | ecb_inline uint64_t ecb_rotl64 (uint64_t x, unsigned int count) ecb_const; |
544 | ecb_inline ecb_const uint64_t ecb_rotl64 (uint64_t x, unsigned int count); |
518 | ecb_inline uint64_t ecb_rotr64 (uint64_t x, unsigned int count) ecb_const; |
545 | ecb_inline ecb_const uint64_t ecb_rotr64 (uint64_t x, unsigned int count); |
519 | |
546 | |
520 | ecb_inline uint8_t ecb_rotl8 (uint8_t x, unsigned int count) { return (x >> ( 8 - count)) | (x << count); } |
547 | ecb_inline ecb_const uint8_t ecb_rotl8 (uint8_t x, unsigned int count) { return (x >> ( 8 - count)) | (x << count); } |
521 | ecb_inline uint8_t ecb_rotr8 (uint8_t x, unsigned int count) { return (x << ( 8 - count)) | (x >> count); } |
548 | ecb_inline ecb_const uint8_t ecb_rotr8 (uint8_t x, unsigned int count) { return (x << ( 8 - count)) | (x >> count); } |
522 | ecb_inline uint16_t ecb_rotl16 (uint16_t x, unsigned int count) { return (x >> (16 - count)) | (x << count); } |
549 | ecb_inline ecb_const uint16_t ecb_rotl16 (uint16_t x, unsigned int count) { return (x >> (16 - count)) | (x << count); } |
523 | ecb_inline uint16_t ecb_rotr16 (uint16_t x, unsigned int count) { return (x << (16 - count)) | (x >> count); } |
550 | ecb_inline ecb_const uint16_t ecb_rotr16 (uint16_t x, unsigned int count) { return (x << (16 - count)) | (x >> count); } |
524 | ecb_inline uint32_t ecb_rotl32 (uint32_t x, unsigned int count) { return (x >> (32 - count)) | (x << count); } |
551 | ecb_inline ecb_const uint32_t ecb_rotl32 (uint32_t x, unsigned int count) { return (x >> (32 - count)) | (x << count); } |
525 | ecb_inline uint32_t ecb_rotr32 (uint32_t x, unsigned int count) { return (x << (32 - count)) | (x >> count); } |
552 | ecb_inline ecb_const uint32_t ecb_rotr32 (uint32_t x, unsigned int count) { return (x << (32 - count)) | (x >> count); } |
526 | ecb_inline uint64_t ecb_rotl64 (uint64_t x, unsigned int count) { return (x >> (64 - count)) | (x << count); } |
553 | ecb_inline ecb_const uint64_t ecb_rotl64 (uint64_t x, unsigned int count) { return (x >> (64 - count)) | (x << count); } |
527 | ecb_inline uint64_t ecb_rotr64 (uint64_t x, unsigned int count) { return (x << (64 - count)) | (x >> count); } |
554 | ecb_inline ecb_const uint64_t ecb_rotr64 (uint64_t x, unsigned int count) { return (x << (64 - count)) | (x >> count); } |
528 | |
555 | |
529 | #if ECB_GCC_VERSION(4,3) || (ECB_CLANG_BUILTIN(__builtin_bswap32) && ECB_CLANG_BUILTIN(__builtin_bswap64)) |
556 | #if ECB_GCC_VERSION(4,3) || (ECB_CLANG_BUILTIN(__builtin_bswap32) && ECB_CLANG_BUILTIN(__builtin_bswap64)) |
|
|
557 | #if ECB_GCC_VERSION(4,8) || ECB_CLANG_BUILTIN(__builtin_bswap16) |
|
|
558 | #define ecb_bswap16(x) __builtin_bswap16 (x) |
|
|
559 | #else |
530 | #define ecb_bswap16(x) (__builtin_bswap32 (x) >> 16) |
560 | #define ecb_bswap16(x) (__builtin_bswap32 (x) >> 16) |
|
|
561 | #endif |
531 | #define ecb_bswap32(x) __builtin_bswap32 (x) |
562 | #define ecb_bswap32(x) __builtin_bswap32 (x) |
532 | #define ecb_bswap64(x) __builtin_bswap64 (x) |
563 | #define ecb_bswap64(x) __builtin_bswap64 (x) |
|
|
564 | #elif _MSC_VER |
|
|
565 | #include <stdlib.h> |
|
|
566 | #define ecb_bswap16(x) ((uint16_t)_byteswap_ushort ((uint16_t)(x))) |
|
|
567 | #define ecb_bswap32(x) ((uint32_t)_byteswap_ulong ((uint32_t)(x))) |
|
|
568 | #define ecb_bswap64(x) ((uint64_t)_byteswap_uint64 ((uint64_t)(x))) |
533 | #else |
569 | #else |
534 | ecb_function_ uint16_t ecb_bswap16 (uint16_t x) ecb_const; |
570 | ecb_function_ ecb_const uint16_t ecb_bswap16 (uint16_t x); |
535 | ecb_function_ uint16_t |
571 | ecb_function_ ecb_const uint16_t |
536 | ecb_bswap16 (uint16_t x) |
572 | ecb_bswap16 (uint16_t x) |
537 | { |
573 | { |
538 | return ecb_rotl16 (x, 8); |
574 | return ecb_rotl16 (x, 8); |
539 | } |
575 | } |
540 | |
576 | |
541 | ecb_function_ uint32_t ecb_bswap32 (uint32_t x) ecb_const; |
577 | ecb_function_ ecb_const uint32_t ecb_bswap32 (uint32_t x); |
542 | ecb_function_ uint32_t |
578 | ecb_function_ ecb_const uint32_t |
543 | ecb_bswap32 (uint32_t x) |
579 | ecb_bswap32 (uint32_t x) |
544 | { |
580 | { |
545 | return (((uint32_t)ecb_bswap16 (x)) << 16) | ecb_bswap16 (x >> 16); |
581 | return (((uint32_t)ecb_bswap16 (x)) << 16) | ecb_bswap16 (x >> 16); |
546 | } |
582 | } |
547 | |
583 | |
548 | ecb_function_ uint64_t ecb_bswap64 (uint64_t x) ecb_const; |
584 | ecb_function_ ecb_const uint64_t ecb_bswap64 (uint64_t x); |
549 | ecb_function_ uint64_t |
585 | ecb_function_ ecb_const uint64_t |
550 | ecb_bswap64 (uint64_t x) |
586 | ecb_bswap64 (uint64_t x) |
551 | { |
587 | { |
552 | return (((uint64_t)ecb_bswap32 (x)) << 32) | ecb_bswap32 (x >> 32); |
588 | return (((uint64_t)ecb_bswap32 (x)) << 32) | ecb_bswap32 (x >> 32); |
553 | } |
589 | } |
554 | #endif |
590 | #endif |
555 | |
591 | |
556 | #if ECB_GCC_VERSION(4,5) || ECB_CLANG_BUILTIN(__builtin_unreachable) |
592 | #if ECB_GCC_VERSION(4,5) || ECB_CLANG_BUILTIN(__builtin_unreachable) |
557 | #define ecb_unreachable() __builtin_unreachable () |
593 | #define ecb_unreachable() __builtin_unreachable () |
558 | #else |
594 | #else |
559 | /* this seems to work fine, but gcc always emits a warning for it :/ */ |
595 | /* this seems to work fine, but gcc always emits a warning for it :/ */ |
560 | ecb_inline void ecb_unreachable (void) ecb_noreturn; |
596 | ecb_inline ecb_noreturn void ecb_unreachable (void); |
561 | ecb_inline void ecb_unreachable (void) { } |
597 | ecb_inline ecb_noreturn void ecb_unreachable (void) { } |
562 | #endif |
598 | #endif |
563 | |
599 | |
564 | /* try to tell the compiler that some condition is definitely true */ |
600 | /* try to tell the compiler that some condition is definitely true */ |
565 | #define ecb_assume(cond) if (!(cond)) ecb_unreachable (); else 0 |
601 | #define ecb_assume(cond) if (!(cond)) ecb_unreachable (); else 0 |
566 | |
602 | |
567 | ecb_inline unsigned char ecb_byteorder_helper (void) ecb_const; |
603 | ecb_inline ecb_const unsigned char ecb_byteorder_helper (void); |
568 | ecb_inline unsigned char |
604 | ecb_inline ecb_const unsigned char |
569 | ecb_byteorder_helper (void) |
605 | ecb_byteorder_helper (void) |
570 | { |
606 | { |
571 | /* the union code still generates code under pressure in gcc, */ |
607 | /* the union code still generates code under pressure in gcc, */ |
572 | /* but less than using pointers, and always seems to */ |
608 | /* but less than using pointers, and always seems to */ |
573 | /* successfully return a constant. */ |
609 | /* successfully return a constant. */ |
574 | /* the reason why we have this horrible preprocessor mess */ |
610 | /* the reason why we have this horrible preprocessor mess */ |
575 | /* is to avoid it in all cases, at least on common architectures */ |
611 | /* is to avoid it in all cases, at least on common architectures */ |
576 | /* or when using a recent enough gcc version (>= 4.6) */ |
612 | /* or when using a recent enough gcc version (>= 4.6) */ |
577 | #if __i386 || __i386__ || _M_X86 || __amd64 || __amd64__ || _M_X64 |
613 | #if ((__i386 || __i386__) && !__VOS__) || _M_IX86 || ECB_GCC_AMD64 || ECB_MSVC_AMD64 |
578 | return 0x44; |
614 | return 0x44; |
579 | #elif __BYTE_ORDER__ && __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__ |
615 | #elif __BYTE_ORDER__ && __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__ |
580 | return 0x44; |
616 | return 0x44; |
581 | #elif __BYTE_ORDER__ && __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__ |
617 | #elif __BYTE_ORDER__ && __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__ |
582 | return 0x11; |
618 | return 0x11; |
… | |
… | |
588 | } u = { 0x11223344 }; |
624 | } u = { 0x11223344 }; |
589 | return u.c; |
625 | return u.c; |
590 | #endif |
626 | #endif |
591 | } |
627 | } |
592 | |
628 | |
593 | ecb_inline ecb_bool ecb_big_endian (void) ecb_const; |
629 | ecb_inline ecb_const ecb_bool ecb_big_endian (void); |
594 | ecb_inline ecb_bool ecb_big_endian (void) { return ecb_byteorder_helper () == 0x11; } |
630 | ecb_inline ecb_const ecb_bool ecb_big_endian (void) { return ecb_byteorder_helper () == 0x11; } |
595 | ecb_inline ecb_bool ecb_little_endian (void) ecb_const; |
631 | ecb_inline ecb_const ecb_bool ecb_little_endian (void); |
596 | ecb_inline ecb_bool ecb_little_endian (void) { return ecb_byteorder_helper () == 0x44; } |
632 | ecb_inline ecb_const ecb_bool ecb_little_endian (void) { return ecb_byteorder_helper () == 0x44; } |
597 | |
633 | |
598 | #if ECB_GCC_VERSION(3,0) || ECB_C99 |
634 | #if ECB_GCC_VERSION(3,0) || ECB_C99 |
599 | #define ecb_mod(m,n) ((m) % (n) + ((m) % (n) < 0 ? (n) : 0)) |
635 | #define ecb_mod(m,n) ((m) % (n) + ((m) % (n) < 0 ? (n) : 0)) |
600 | #else |
636 | #else |
601 | #define ecb_mod(m,n) ((m) < 0 ? ((n) - 1 - ((-1 - (m)) % (n))) : ((m) % (n))) |
637 | #define ecb_mod(m,n) ((m) < 0 ? ((n) - 1 - ((-1 - (m)) % (n))) : ((m) % (n))) |
602 | #endif |
638 | #endif |
603 | |
639 | |
604 | #if __cplusplus |
640 | #if ECB_CPP |
605 | template<typename T> |
641 | template<typename T> |
606 | static inline T ecb_div_rd (T val, T div) |
642 | static inline T ecb_div_rd (T val, T div) |
607 | { |
643 | { |
608 | return val < 0 ? - ((-val + div - 1) / div) : (val ) / div; |
644 | return val < 0 ? - ((-val + div - 1) / div) : (val ) / div; |
609 | } |
645 | } |
… | |
… | |
633 | |
669 | |
634 | /* basically, everything uses "ieee pure-endian" floating point numbers */ |
670 | /* basically, everything uses "ieee pure-endian" floating point numbers */ |
635 | /* the only noteworthy exception is ancient armle, which uses order 43218765 */ |
671 | /* the only noteworthy exception is ancient armle, which uses order 43218765 */ |
636 | #if 0 \ |
672 | #if 0 \ |
637 | || __i386 || __i386__ \ |
673 | || __i386 || __i386__ \ |
638 | || __amd64 || __amd64__ || __x86_64 || __x86_64__ \ |
674 | || ECB_GCC_AMD64 \ |
639 | || __powerpc__ || __ppc__ || __powerpc64__ || __ppc64__ \ |
675 | || __powerpc__ || __ppc__ || __powerpc64__ || __ppc64__ \ |
640 | || defined __s390__ || defined __s390x__ \ |
676 | || defined __s390__ || defined __s390x__ \ |
641 | || defined __mips__ \ |
677 | || defined __mips__ \ |
642 | || defined __alpha__ \ |
678 | || defined __alpha__ \ |
643 | || defined __hppa__ \ |
679 | || defined __hppa__ \ |
644 | || defined __ia64__ \ |
680 | || defined __ia64__ \ |
645 | || defined __m68k__ \ |
681 | || defined __m68k__ \ |
646 | || defined __m88k__ \ |
682 | || defined __m88k__ \ |
647 | || defined __sh__ \ |
683 | || defined __sh__ \ |
648 | || defined _M_IX86 || defined _M_AMD64 || defined _M_IA64 \ |
684 | || defined _M_IX86 || defined ECB_MSVC_AMD64 || defined _M_IA64 \ |
649 | || (defined __arm__ && (defined __ARM_EABI__ || defined __EABI__ || defined __VFP_FP__ || defined _WIN32_WCE || defined __ANDROID__)) \ |
685 | || (defined __arm__ && (defined __ARM_EABI__ || defined __EABI__ || defined __VFP_FP__ || defined _WIN32_WCE || defined __ANDROID__)) \ |
650 | || defined __aarch64__ |
686 | || defined __aarch64__ |
651 | #define ECB_STDFP 1 |
687 | #define ECB_STDFP 1 |
652 | #include <string.h> /* for memcpy */ |
688 | #include <string.h> /* for memcpy */ |
653 | #else |
689 | #else |
… | |
… | |
669 | #define ECB_NAN NAN |
705 | #define ECB_NAN NAN |
670 | #else |
706 | #else |
671 | #define ECB_NAN ECB_INFINITY |
707 | #define ECB_NAN ECB_INFINITY |
672 | #endif |
708 | #endif |
673 | |
709 | |
|
|
710 | #if ECB_C99 || _XOPEN_VERSION >= 600 || _POSIX_VERSION >= 200112L |
|
|
711 | #define ecb_ldexpf(x,e) ldexpf ((x), (e)) |
|
|
712 | #define ecb_frexpf(x,e) frexpf ((x), (e)) |
|
|
713 | #else |
|
|
714 | #define ecb_ldexpf(x,e) (float) ldexp ((double) (x), (e)) |
|
|
715 | #define ecb_frexpf(x,e) (float) frexp ((double) (x), (e)) |
|
|
716 | #endif |
|
|
717 | |
674 | /* converts an ieee half/binary16 to a float */ |
718 | /* converts an ieee half/binary16 to a float */ |
675 | ecb_function_ float ecb_binary16_to_float (uint16_t x) ecb_const; |
719 | ecb_function_ ecb_const float ecb_binary16_to_float (uint16_t x); |
676 | ecb_function_ float |
720 | ecb_function_ ecb_const float |
677 | ecb_binary16_to_float (uint16_t x) |
721 | ecb_binary16_to_float (uint16_t x) |
678 | { |
722 | { |
679 | int e = (x >> 10) & 0x1f; |
723 | int e = (x >> 10) & 0x1f; |
680 | int m = x & 0x3ff; |
724 | int m = x & 0x3ff; |
681 | float r; |
725 | float r; |
682 | |
726 | |
683 | if (!e ) r = ldexpf (m , -24); |
727 | if (!e ) r = ecb_ldexpf (m , -24); |
684 | else if (e != 31) r = ldexpf (m + 0x400, e - 25); |
728 | else if (e != 31) r = ecb_ldexpf (m + 0x400, e - 25); |
685 | else if (m ) r = ECB_NAN; |
729 | else if (m ) r = ECB_NAN; |
686 | else r = ECB_INFINITY; |
730 | else r = ECB_INFINITY; |
687 | |
731 | |
688 | return x & 0x8000 ? -r : r; |
732 | return x & 0x8000 ? -r : r; |
689 | } |
733 | } |
690 | |
734 | |
691 | /* convert a float to ieee single/binary32 */ |
735 | /* convert a float to ieee single/binary32 */ |
692 | ecb_function_ uint32_t ecb_float_to_binary32 (float x) ecb_const; |
736 | ecb_function_ ecb_const uint32_t ecb_float_to_binary32 (float x); |
693 | ecb_function_ uint32_t |
737 | ecb_function_ ecb_const uint32_t |
694 | ecb_float_to_binary32 (float x) |
738 | ecb_float_to_binary32 (float x) |
695 | { |
739 | { |
696 | uint32_t r; |
740 | uint32_t r; |
697 | |
741 | |
698 | #if ECB_STDFP |
742 | #if ECB_STDFP |
… | |
… | |
705 | if (x == 0e0f ) return 0x00000000U; |
749 | if (x == 0e0f ) return 0x00000000U; |
706 | if (x > +3.40282346638528860e+38f) return 0x7f800000U; |
750 | if (x > +3.40282346638528860e+38f) return 0x7f800000U; |
707 | if (x < -3.40282346638528860e+38f) return 0xff800000U; |
751 | if (x < -3.40282346638528860e+38f) return 0xff800000U; |
708 | if (x != x ) return 0x7fbfffffU; |
752 | if (x != x ) return 0x7fbfffffU; |
709 | |
753 | |
710 | m = frexpf (x, &e) * 0x1000000U; |
754 | m = ecb_frexpf (x, &e) * 0x1000000U; |
711 | |
755 | |
712 | r = m & 0x80000000U; |
756 | r = m & 0x80000000U; |
713 | |
757 | |
714 | if (r) |
758 | if (r) |
715 | m = -m; |
759 | m = -m; |
… | |
… | |
727 | |
771 | |
728 | return r; |
772 | return r; |
729 | } |
773 | } |
730 | |
774 | |
731 | /* converts an ieee single/binary32 to a float */ |
775 | /* converts an ieee single/binary32 to a float */ |
732 | ecb_function_ float ecb_binary32_to_float (uint32_t x) ecb_const; |
776 | ecb_function_ ecb_const float ecb_binary32_to_float (uint32_t x); |
733 | ecb_function_ float |
777 | ecb_function_ ecb_const float |
734 | ecb_binary32_to_float (uint32_t x) |
778 | ecb_binary32_to_float (uint32_t x) |
735 | { |
779 | { |
736 | float r; |
780 | float r; |
737 | |
781 | |
738 | #if ECB_STDFP |
782 | #if ECB_STDFP |
… | |
… | |
748 | x |= 0x800000U; |
792 | x |= 0x800000U; |
749 | else |
793 | else |
750 | e = 1; |
794 | e = 1; |
751 | |
795 | |
752 | /* we distrust ldexpf a bit and do the 2**-24 scaling by an extra multiply */ |
796 | /* we distrust ldexpf a bit and do the 2**-24 scaling by an extra multiply */ |
753 | r = ldexpf (x * (0.5f / 0x800000U), e - 126); |
797 | r = ecb_ldexpf (x * (0.5f / 0x800000U), e - 126); |
754 | |
798 | |
755 | r = neg ? -r : r; |
799 | r = neg ? -r : r; |
756 | #endif |
800 | #endif |
757 | |
801 | |
758 | return r; |
802 | return r; |
759 | } |
803 | } |
760 | |
804 | |
761 | /* convert a double to ieee double/binary64 */ |
805 | /* convert a double to ieee double/binary64 */ |
762 | ecb_function_ uint64_t ecb_double_to_binary64 (double x) ecb_const; |
806 | ecb_function_ ecb_const uint64_t ecb_double_to_binary64 (double x); |
763 | ecb_function_ uint64_t |
807 | ecb_function_ ecb_const uint64_t |
764 | ecb_double_to_binary64 (double x) |
808 | ecb_double_to_binary64 (double x) |
765 | { |
809 | { |
766 | uint64_t r; |
810 | uint64_t r; |
767 | |
811 | |
768 | #if ECB_STDFP |
812 | #if ECB_STDFP |
… | |
… | |
797 | |
841 | |
798 | return r; |
842 | return r; |
799 | } |
843 | } |
800 | |
844 | |
801 | /* converts an ieee double/binary64 to a double */ |
845 | /* converts an ieee double/binary64 to a double */ |
802 | ecb_function_ double ecb_binary64_to_double (uint64_t x) ecb_const; |
846 | ecb_function_ ecb_const double ecb_binary64_to_double (uint64_t x); |
803 | ecb_function_ double |
847 | ecb_function_ ecb_const double |
804 | ecb_binary64_to_double (uint64_t x) |
848 | ecb_binary64_to_double (uint64_t x) |
805 | { |
849 | { |
806 | double r; |
850 | double r; |
807 | |
851 | |
808 | #if ECB_STDFP |
852 | #if ECB_STDFP |