[ViewVC] Diff of: cvs/libecb/ecb.h

Comparing libecb/ecb.h (file contents):
Revision 1.43 by root, Sun Jun 12 08:45:06 2011 UTC vs.
Revision 1.71 by root, Sat Dec 10 11:59:19 2011 UTC

…		…
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(__SUNPRO_CC) \|\| defined(__llvm__) \|\| defined(__clang__)	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)))
47	#endif	63	#endif
48	#endif	64	#endif
49		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	#endif
		74
		75	#ifndef ECB_MEMORY_FENCE
		76	#if ECB_GCC_VERSION(2,5) \|\| defined(__INTEL_COMPILER) \|\| defined(__clang__)
		77	#if __i386__
		78	#define ECB_MEMORY_FENCE __asm__ __volatile__ ("lock; orb $0, -1(%%esp)" : : : "memory")
		79	#define ECB_MEMORY_FENCE_ACQUIRE ECB_MEMORY_FENCE /* non-lock xchg might be enough */
		80	#define ECB_MEMORY_FENCE_RELEASE do { } while (0) /* unlikely to change in future cpus */
		81	#elif __amd64
		82	#define ECB_MEMORY_FENCE __asm__ __volatile__ ("mfence" : : : "memory")
		83	#define ECB_MEMORY_FENCE_ACQUIRE __asm__ __volatile__ ("lfence" : : : "memory")
		84	#define ECB_MEMORY_FENCE_RELEASE __asm__ __volatile__ ("sfence") /* play safe - not needed in any current cpu */
		85	#elif __powerpc__ \|\| __ppc__ \|\| __powerpc64__ \|\| __ppc64__
		86	#define ECB_MEMORY_FENCE __asm__ __volatile__ ("sync" : : : "memory")
		87	#elif defined(__ARM_ARCH_6__ ) \|\| defined(__ARM_ARCH_6J__ ) \
		88	\|\| defined(__ARM_ARCH_6K__) \|\| defined(__ARM_ARCH_6ZK__)
		89	#define ECB_MEMORY_FENCE __asm__ __volatile__ ("mcr p15,0,%0,c7,c10,5" : : "r" (0) : "memory")
		90	#elif defined(__ARM_ARCH_7__ ) \|\| defined(__ARM_ARCH_7A__ ) \
		91	\|\| defined(__ARM_ARCH_7M__) \|\| defined(__ARM_ARCH_7R__ )
		92	#define ECB_MEMORY_FENCE __asm__ __volatile__ ("dmb" : : : "memory")
		93	#endif
		94	#endif
		95	#endif
		96
		97	#ifndef ECB_MEMORY_FENCE
		98	#if ECB_GCC_VERSION(4,4) \|\| defined(__INTEL_COMPILER) \|\| defined(__clang__)
		99	#define ECB_MEMORY_FENCE __sync_synchronize ()
		100	/#define ECB_MEMORY_FENCE_ACQUIRE ({ char dummy = 0; __sync_lock_test_and_set (&dummy, 1); }) /
		101	/#define ECB_MEMORY_FENCE_RELEASE ({ char dummy = 1; __sync_lock_release (&dummy ); }) /
		102	#elif _MSC_VER >= 1400 /* VC++ 2005 */
		103	#pragma intrinsic(_ReadBarrier,_WriteBarrier,_ReadWriteBarrier)
		104	#define ECB_MEMORY_FENCE _ReadWriteBarrier ()
		105	#define ECB_MEMORY_FENCE_ACQUIRE _ReadWriteBarrier () /* according to msdn, _ReadBarrier is not a load fence */
		106	#define ECB_MEMORY_FENCE_RELEASE _WriteBarrier ()
		107	#elif defined(_WIN32)
		108	#include <WinNT.h>
		109	#define ECB_MEMORY_FENCE MemoryBarrier () /* actually just xchg on x86... scary */
		110	#endif
		111	#endif
		112
		113	#ifndef ECB_MEMORY_FENCE
		114	#if !ECB_AVOID_PTHREADS
		115	/*
		116	* if you get undefined symbol references to pthread_mutex_lock,
		117	* or failure to find pthread.h, then you should implement
		118	* the ECB_MEMORY_FENCE operations for your cpu/compiler
		119	* OR provide pthread.h and link against the posix thread library
		120	* of your system.
		121	*/
		122	#include <pthread.h>
		123	#define ECB_NEEDS_PTHREADS 1
		124	#define ECB_MEMORY_FENCE_NEEDS_PTHREADS 1
		125
		126	static pthread_mutex_t ecb_mf_lock = PTHREAD_MUTEX_INITIALIZER;
		127	#define ECB_MEMORY_FENCE do { pthread_mutex_lock (&ecb_mf_lock); pthread_mutex_unlock (&ecb_mf_lock); } while (0)
		128	#endif
		129	#endif
		130
		131	#if !defined(ECB_MEMORY_FENCE_ACQUIRE) && defined(ECB_MEMORY_FENCE)
		132	#define ECB_MEMORY_FENCE_ACQUIRE ECB_MEMORY_FENCE
		133	#endif
		134
		135	#if !defined(ECB_MEMORY_FENCE_RELEASE) && defined(ECB_MEMORY_FENCE)
		136	#define ECB_MEMORY_FENCE_RELEASE ECB_MEMORY_FENCE
		137	#endif
		138
		139	/*****************************************************************************/
		140
50	#define ECB_C99 (__STDC_VERSION__ >= 199901L)	141	#define ECB_C99 (__STDC_VERSION__ >= 199901L)
51		142
52	#if __cplusplus	143	#if __cplusplus
53	#define ECB_INLINE static inline	144	#define ecb_inline static inline
54	#elif ECB_GCC_VERSION(2,5)	145	#elif ECB_GCC_VERSION(2,5)
55	#define ECB_INLINE static __inline__	146	#define ecb_inline static __inline__
56	#elif ECB_C99	147	#elif ECB_C99
57	#define ECB_INLINE static inline	148	#define ecb_inline static inline
58	#else	149	#else
59	#define ECB_INLINE static	150	#define ecb_inline static
60	#endif	151	#endif
61		152
62	#if ECB_GCC_VERSION(3,3)	153	#if ECB_GCC_VERSION(3,3)
63	#define ecb_restrict __restrict__	154	#define ecb_restrict __restrict__
64	#elif ECB_C99	155	#elif ECB_C99
…		…
72	#define ECB_CONCAT_(a, b) a ## b	163	#define ECB_CONCAT_(a, b) a ## b
73	#define ECB_CONCAT(a, b) ECB_CONCAT_(a, b)	164	#define ECB_CONCAT(a, b) ECB_CONCAT_(a, b)
74	#define ECB_STRINGIFY_(a) # a	165	#define ECB_STRINGIFY_(a) # a
75	#define ECB_STRINGIFY(a) ECB_STRINGIFY_(a)	166	#define ECB_STRINGIFY(a) ECB_STRINGIFY_(a)
76		167
77	#define ecb_function_ ECB_INLINE	168	#define ecb_function_ ecb_inline
78		169
79	#if ECB_GCC_VERSION(3,1)	170	#if ECB_GCC_VERSION(3,1)
80	#define ecb_attribute(attrlist) __attribute__(attrlist)	171	#define ecb_attribute(attrlist) __attribute__(attrlist)
81	#define ecb_is_constant(expr) __builtin_constant_p (expr)	172	#define ecb_is_constant(expr) __builtin_constant_p (expr)
82	#define ecb_expect(expr,value) __builtin_expect ((expr),(value))	173	#define ecb_expect(expr,value) __builtin_expect ((expr),(value))
…		…
120	#define ecb_likely(expr) ecb_expect_true (expr)	211	#define ecb_likely(expr) ecb_expect_true (expr)
121	#define ecb_unlikely(expr) ecb_expect_false (expr)	212	#define ecb_unlikely(expr) ecb_expect_false (expr)
122		213
123	/* count trailing zero bits and count # of one bits */	214	/* count trailing zero bits and count # of one bits */
124	#if ECB_GCC_VERSION(3,4)	215	#if ECB_GCC_VERSION(3,4)
		216	/* we assume int == 32 bit, long == 32 or 64 bit and long long == 64 bit */
		217	#define ecb_ld32(x) (__builtin_clz (x) ^ 31)
		218	#define ecb_ld64(x) (__builtin_clzll (x) ^ 63)
125	#define ecb_ctz32(x) __builtin_ctz (x)	219	#define ecb_ctz32(x) __builtin_ctz (x)
		220	#define ecb_ctz64(x) __builtin_ctzll (x)
126	#define ecb_popcount32(x) __builtin_popcount (x)	221	#define ecb_popcount32(x) __builtin_popcount (x)
		222	/* no popcountll */
127	#else	223	#else
128	ecb_function_ int ecb_ctz32 (uint32_t x) ecb_const;	224	ecb_function_ int ecb_ctz32 (uint32_t x) ecb_const;
129	ecb_function_ int	225	ecb_function_ int
130	ecb_ctz32 (uint32_t x)	226	ecb_ctz32 (uint32_t x)
131	{	227	{
132	int r = 0;	228	int r = 0;
133		229
134	x &= -x; /* this isolates the lowest bit */	230	x &= ~x + 1; /* this isolates the lowest bit */
135		231
		232	#if ECB_branchless_on_i386
		233	r += !!(x & 0xaaaaaaaa) << 0;
		234	r += !!(x & 0xcccccccc) << 1;
		235	r += !!(x & 0xf0f0f0f0) << 2;
		236	r += !!(x & 0xff00ff00) << 3;
		237	r += !!(x & 0xffff0000) << 4;
		238	#else
136	if (x & 0xaaaaaaaa) r += 1;	239	if (x & 0xaaaaaaaa) r += 1;
137	if (x & 0xcccccccc) r += 2;	240	if (x & 0xcccccccc) r += 2;
138	if (x & 0xf0f0f0f0) r += 4;	241	if (x & 0xf0f0f0f0) r += 4;
139	if (x & 0xff00ff00) r += 8;	242	if (x & 0xff00ff00) r += 8;
140	if (x & 0xffff0000) r += 16;	243	if (x & 0xffff0000) r += 16;
		244	#endif
141		245
142	return r;	246	return r;
		247	}
		248
		249	ecb_function_ int ecb_ctz64 (uint64_t x) ecb_const;
		250	ecb_function_ int
		251	ecb_ctz64 (uint64_t x)
		252	{
		253	int shift = x & 0xffffffffU ? 0 : 32;
		254	return ecb_ctz32 (x >> shift) + shift;
143	}	255	}
144		256
145	ecb_function_ int ecb_popcount32 (uint32_t x) ecb_const;	257	ecb_function_ int ecb_popcount32 (uint32_t x) ecb_const;
146	ecb_function_ int	258	ecb_function_ int
147	ecb_popcount32 (uint32_t x)	259	ecb_popcount32 (uint32_t x)
…		…
151	x = ((x >> 4) + x) & 0x0f0f0f0f;	263	x = ((x >> 4) + x) & 0x0f0f0f0f;
152	x *= 0x01010101;	264	x *= 0x01010101;
153		265
154	return x >> 24;	266	return x >> 24;
155	}	267	}
		268
		269	ecb_function_ int ecb_ld32 (uint32_t x) ecb_const;
		270	ecb_function_ int ecb_ld32 (uint32_t x)
		271	{
		272	int r = 0;
		273
		274	if (x >> 16) { x >>= 16; r += 16; }
		275	if (x >> 8) { x >>= 8; r += 8; }
		276	if (x >> 4) { x >>= 4; r += 4; }
		277	if (x >> 2) { x >>= 2; r += 2; }
		278	if (x >> 1) { r += 1; }
		279
		280	return r;
		281	}
		282
		283	ecb_function_ int ecb_ld64 (uint64_t x) ecb_const;
		284	ecb_function_ int ecb_ld64 (uint64_t x)
		285	{
		286	int r = 0;
		287
		288	if (x >> 32) { x >>= 32; r += 32; }
		289
		290	return r + ecb_ld32 (x);
		291	}
156	#endif	292	#endif
		293
		294	ecb_function_ uint8_t ecb_bitrev8 (uint8_t x) ecb_const;
		295	ecb_function_ uint8_t ecb_bitrev8 (uint8_t x)
		296	{
		297	return ( (x * 0x0802U & 0x22110U)
		298	\| (x * 0x8020U & 0x88440U)) * 0x10101U >> 16;
		299	}
		300
		301	ecb_function_ uint16_t ecb_bitrev16 (uint16_t x) ecb_const;
		302	ecb_function_ uint16_t ecb_bitrev16 (uint16_t x)
		303	{
		304	x = ((x >> 1) & 0x5555) \| ((x & 0x5555) << 1);
		305	x = ((x >> 2) & 0x3333) \| ((x & 0x3333) << 2);
		306	x = ((x >> 4) & 0x0f0f) \| ((x & 0x0f0f) << 4);
		307	x = ( x >> 8 ) \| ( x << 8);
		308
		309	return x;
		310	}
		311
		312	ecb_function_ uint32_t ecb_bitrev32 (uint32_t x) ecb_const;
		313	ecb_function_ uint32_t ecb_bitrev32 (uint32_t x)
		314	{
		315	x = ((x >> 1) & 0x55555555) \| ((x & 0x55555555) << 1);
		316	x = ((x >> 2) & 0x33333333) \| ((x & 0x33333333) << 2);
		317	x = ((x >> 4) & 0x0f0f0f0f) \| ((x & 0x0f0f0f0f) << 4);
		318	x = ((x >> 8) & 0x00ff00ff) \| ((x & 0x00ff00ff) << 8);
		319	x = ( x >> 16 ) \| ( x << 16);
		320
		321	return x;
		322	}
		323
		324	/* popcount64 is only available on 64 bit cpus as gcc builtin */
		325	/* so for this version we are lazy */
		326	ecb_function_ int ecb_popcount64 (uint64_t x) ecb_const;
		327	ecb_function_ int
		328	ecb_popcount64 (uint64_t x)
		329	{
		330	return ecb_popcount32 (x) + ecb_popcount32 (x >> 32);
		331	}
		332
		333	ecb_inline uint8_t ecb_rotl8 (uint8_t x, unsigned int count) ecb_const;
		334	ecb_inline uint8_t ecb_rotr8 (uint8_t x, unsigned int count) ecb_const;
		335	ecb_inline uint16_t ecb_rotl16 (uint16_t x, unsigned int count) ecb_const;
		336	ecb_inline uint16_t ecb_rotr16 (uint16_t x, unsigned int count) ecb_const;
		337	ecb_inline uint32_t ecb_rotl32 (uint32_t x, unsigned int count) ecb_const;
		338	ecb_inline uint32_t ecb_rotr32 (uint32_t x, unsigned int count) ecb_const;
		339	ecb_inline uint64_t ecb_rotl64 (uint64_t x, unsigned int count) ecb_const;
		340	ecb_inline uint64_t ecb_rotr64 (uint64_t x, unsigned int count) ecb_const;
		341
		342	ecb_inline uint8_t ecb_rotl8 (uint8_t x, unsigned int count) { return (x >> ( 8 - count)) \| (x << count); }
		343	ecb_inline uint8_t ecb_rotr8 (uint8_t x, unsigned int count) { return (x << ( 8 - count)) \| (x >> count); }
		344	ecb_inline uint16_t ecb_rotl16 (uint16_t x, unsigned int count) { return (x >> (16 - count)) \| (x << count); }
		345	ecb_inline uint16_t ecb_rotr16 (uint16_t x, unsigned int count) { return (x << (16 - count)) \| (x >> count); }
		346	ecb_inline uint32_t ecb_rotl32 (uint32_t x, unsigned int count) { return (x >> (32 - count)) \| (x << count); }
		347	ecb_inline uint32_t ecb_rotr32 (uint32_t x, unsigned int count) { return (x << (32 - count)) \| (x >> count); }
		348	ecb_inline uint64_t ecb_rotl64 (uint64_t x, unsigned int count) { return (x >> (64 - count)) \| (x << count); }
		349	ecb_inline uint64_t ecb_rotr64 (uint64_t x, unsigned int count) { return (x << (64 - count)) \| (x >> count); }
157		350
158	#if ECB_GCC_VERSION(4,3)	351	#if ECB_GCC_VERSION(4,3)
159	#define ecb_bswap32(x) __builtin_bswap32 (x)
160	#define ecb_bswap16(x) (__builtin_bswap32(x) >> 16)	352	#define ecb_bswap16(x) (__builtin_bswap32 (x) >> 16)
		353	#define ecb_bswap32(x) __builtin_bswap32 (x)
		354	#define ecb_bswap64(x) __builtin_bswap64 (x)
161	#else	355	#else
		356	ecb_function_ uint16_t ecb_bswap16 (uint16_t x) ecb_const;
		357	ecb_function_ uint16_t
		358	ecb_bswap16 (uint16_t x)
		359	{
		360	return ecb_rotl16 (x, 8);
		361	}
		362
162	ecb_function_ uint32_t ecb_bswap32 (uint32_t x) ecb_const;	363	ecb_function_ uint32_t ecb_bswap32 (uint32_t x) ecb_const;
163	ecb_function_ uint32_t	364	ecb_function_ uint32_t
164	ecb_bswap32 (uint32_t x)	365	ecb_bswap32 (uint32_t x)
165	{	366	{
166	return (x >> 24)	367	return (((uint32_t)ecb_bswap16 (x)) << 16) \| ecb_bswap16 (x >> 16);
167	\| ((x >> 8) & 0x0000ff00)
168	\| ((x << 8) & 0x00ff0000)
169	\| (x << 24);
170	}	368	}
171		369
172	ecb_function_ uint32_t ecb_bswap16 (uint32_t x) ecb_const;	370	ecb_function_ uint64_t ecb_bswap64 (uint64_t x) ecb_const;
173	ecb_function_ uint32_t	371	ecb_function_ uint64_t
174	ecb_bswap16 (uint32_t x)	372	ecb_bswap64 (uint64_t x)
175	{	373	{
176	return ((x >> 8) & 0xff)	374	return (((uint64_t)ecb_bswap32 (x)) << 32) \| ecb_bswap32 (x >> 32);
177	\| ((x << 8) & 0x00ff0000)
178	\| (x << 24);
179	}	375	}
180	#endif	376	#endif
181		377
182	#if ECB_GCC_VERSION(4,5)	378	#if ECB_GCC_VERSION(4,5)
183	#define ecb_unreachable() __builtin_unreachable ()	379	#define ecb_unreachable() __builtin_unreachable ()
…		…
207	#define ecb_mod(m,n) ((m) % (n) + ((m) % (n) < 0 ? (n) : 0))	403	#define ecb_mod(m,n) ((m) % (n) + ((m) % (n) < 0 ? (n) : 0))
208	#else	404	#else
209	#define ecb_mod(m,n) ((m) < 0 ? ((n) - 1 - ((-1 - (m)) % (n))) : ((m) % (n)))	405	#define ecb_mod(m,n) ((m) < 0 ? ((n) - 1 - ((-1 - (m)) % (n))) : ((m) % (n)))
210	#endif	406	#endif
211		407
		408	#if __cplusplus
		409	template<typename T>
		410	static inline T ecb_div_rd (T val, T div)
		411	{
		412	return val < 0 ? - ((-val + div - 1) / div) : (val ) / div;
		413	}
		414	template<typename T>
		415	static inline T ecb_div_ru (T val, T div)
		416	{
		417	return val < 0 ? - ((-val ) / div) : (val + div - 1) / div;
		418	}
		419	#else
		420	#define ecb_div_rd(val,div) ((val) < 0 ? - ((-(val) + (div) - 1) / (div)) : ((val) ) / (div))
		421	#define ecb_div_ru(val,div) ((val) < 0 ? - ((-(val) ) / (div)) : ((val) + (div) - 1) / (div))
		422	#endif
		423
212	#if ecb_cplusplus_does_not_suck	424	#if ecb_cplusplus_does_not_suck
213	/* does not work for local types (http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2008/n2657.htm) */	425	/* does not work for local types (http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2008/n2657.htm) */
214	template<typename T, int N>	426	template<typename T, int N>
215	static inline int ecb_array_length (const T (&arr)[N])	427	static inline int ecb_array_length (const T (&arr)[N])
216	{	428	{
…		…
218	}	430	}
219	#else	431	#else
220	#define ecb_array_length(name) (sizeof (name) / sizeof (name [0]))	432	#define ecb_array_length(name) (sizeof (name) / sizeof (name [0]))
221	#endif	433	#endif
222		434
223	ECB_INLINE uint32_t ecb_rotr32 (uint32_t x, unsigned int count) ecb_const;
224	ECB_INLINE uint32_t
225	ecb_rotr32 (uint32_t x, unsigned int count)
226	{
227	return (x << (32 - count)) \| (x >> count);
228	}
229
230	ECB_INLINE uint32_t ecb_rotl32 (uint32_t x, unsigned int count) ecb_const;
231	ECB_INLINE uint32_t
232	ecb_rotl32 (uint32_t x, unsigned int count)
233	{
234	return (x >> (32 - count)) \| (x << count);
235	}
236
237	#endif	435	#endif
238		436

Diff Legend

-–
+Removed lines
-+
+Added lines
-<
+Changed lines
->
+Changed lines

Comparing libecb/ecb.h (file contents): Revision 1.43 by root, Sun Jun 12 08:45:06 2011 UTC vs. Revision 1.71 by root, Sat Dec 10 11:59:19 2011 UTC

Diff Legend

Comparing libecb/ecb.h (file contents):
Revision 1.43 by root, Sun Jun 12 08:45:06 2011 UTC vs.
Revision 1.71 by root, Sat Dec 10 11:59:19 2011 UTC