1 /*****************************************************************************
2 * osdep.h: platform-specific code
3 *****************************************************************************
4 * Copyright (C) 2007-2016 x264 project
6 * Authors: Loren Merritt <lorenm@u.washington.edu>
7 * Laurent Aimar <fenrir@via.ecp.fr>
8 * Henrik Gramner <henrik@gramner.com>
10 * This program is free software; you can redistribute it and/or modify
11 * it under the terms of the GNU General Public License as published by
12 * the Free Software Foundation; either version 2 of the License, or
13 * (at your option) any later version.
15 * This program is distributed in the hope that it will be useful,
16 * but WITHOUT ANY WARRANTY; without even the implied warranty of
17 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
18 * GNU General Public License for more details.
20 * You should have received a copy of the GNU General Public License
21 * along with this program; if not, write to the Free Software
22 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02111, USA.
24 * This program is also available under a commercial proprietary license.
25 * For more information, contact us at licensing@x264.com.
26 *****************************************************************************/
31 #define _LARGEFILE_SOURCE 1
32 #define _FILE_OFFSET_BITS 64
40 #ifdef __INTEL_COMPILER
47 #define log2f(x) (logf(x)/0.693147180559945f)
48 #define log2(x) (log(x)/0.693147180559945)
52 #define inline __inline
53 #define strcasecmp _stricmp
54 #define strncasecmp _strnicmp
55 #define strtok_r strtok_s
56 #define S_ISREG(x) (((x) & S_IFMT) == S_IFREG)
58 #define snprintf _snprintf
62 #if !defined(va_copy) && defined(__INTEL_COMPILER)
63 #define va_copy(dst, src) ((dst) = (src))
66 #if !defined(isfinite) && (SYS_OPENBSD || SYS_SunOS)
67 #define isfinite finite
72 #define strtok_r(str,delim,save) strtok(str,delim)
75 #define utf8_to_utf16( utf8, utf16 )\
76 MultiByteToWideChar( CP_UTF8, MB_ERR_INVALID_CHARS, utf8, -1, utf16, sizeof(utf16)/sizeof(wchar_t) )
77 FILE *x264_fopen( const char *filename, const char *mode );
78 int x264_rename( const char *oldname, const char *newname );
79 #define x264_struct_stat struct _stati64
80 #define x264_fstat _fstati64
81 int x264_stat( const char *path, x264_struct_stat *buf );
82 int x264_vfprintf( FILE *stream, const char *format, va_list arg );
83 int x264_is_pipe( const char *path );
85 #define x264_fopen fopen
86 #define x264_rename rename
87 #define x264_struct_stat struct stat
88 #define x264_fstat fstat
89 #define x264_stat stat
90 #define x264_vfprintf vfprintf
91 #define x264_is_pipe(x) 0
95 #define DECLARE_ALIGNED( var, n ) __declspec(align(n)) var
97 #define DECLARE_ALIGNED( var, n ) var __attribute__((aligned(n)))
99 #define ALIGNED_32( var ) DECLARE_ALIGNED( var, 32 )
100 #define ALIGNED_16( var ) DECLARE_ALIGNED( var, 16 )
101 #define ALIGNED_8( var ) DECLARE_ALIGNED( var, 8 )
102 #define ALIGNED_4( var ) DECLARE_ALIGNED( var, 4 )
104 // ARM compiliers don't reliably align stack variables
105 // - EABI requires only 8 byte stack alignment to be maintained
106 // - gcc can't align stack variables to more even if the stack were to be correctly aligned outside the function
107 // - armcc can't either, but is nice enough to actually tell you so
108 // - Apple gcc only maintains 4 byte alignment
109 // - llvm can align the stack, but only in svn and (unrelated) it exposes bugs in all released GNU binutils...
111 #define ALIGNED_ARRAY_EMU( mask, type, name, sub1, ... )\
112 uint8_t name##_u [sizeof(type sub1 __VA_ARGS__) + mask]; \
113 type (*name) __VA_ARGS__ = (void*)((intptr_t)(name##_u+mask) & ~mask)
115 #if ARCH_ARM && SYS_MACOSX
116 #define ALIGNED_ARRAY_8( ... ) ALIGNED_ARRAY_EMU( 7, __VA_ARGS__ )
118 #define ALIGNED_ARRAY_8( type, name, sub1, ... )\
119 ALIGNED_8( type name sub1 __VA_ARGS__ )
123 #define ALIGNED_ARRAY_16( ... ) ALIGNED_ARRAY_EMU( 15, __VA_ARGS__ )
125 #define ALIGNED_ARRAY_16( type, name, sub1, ... )\
126 ALIGNED_16( type name sub1 __VA_ARGS__ )
131 #if STACK_ALIGNMENT >= 32
132 #define ALIGNED_ARRAY_32( type, name, sub1, ... )\
133 ALIGNED_32( type name sub1 __VA_ARGS__ )
135 #define ALIGNED_ARRAY_32( ... ) EXPAND( ALIGNED_ARRAY_EMU( 31, __VA_ARGS__ ) )
138 #define ALIGNED_ARRAY_64( ... ) EXPAND( ALIGNED_ARRAY_EMU( 63, __VA_ARGS__ ) )
141 #if ARCH_X86 || ARCH_X86_64
142 #define NATIVE_ALIGN 32
143 #define ALIGNED_N ALIGNED_32
144 #define ALIGNED_ARRAY_N ALIGNED_ARRAY_32
146 #define NATIVE_ALIGN 16
147 #define ALIGNED_N ALIGNED_16
148 #define ALIGNED_ARRAY_N ALIGNED_ARRAY_16
151 #if defined(__GNUC__) && (__GNUC__ > 3 || __GNUC__ == 3 && __GNUC_MINOR__ > 0)
152 #define UNUSED __attribute__((unused))
153 #define ALWAYS_INLINE __attribute__((always_inline)) inline
154 #define NOINLINE __attribute__((noinline))
155 #define MAY_ALIAS __attribute__((may_alias))
156 #define x264_constant_p(x) __builtin_constant_p(x)
157 #define x264_nonconstant_p(x) (!__builtin_constant_p(x))
160 #define ALWAYS_INLINE __forceinline
161 #define NOINLINE __declspec(noinline)
163 #define ALWAYS_INLINE inline
168 #define x264_constant_p(x) 0
169 #define x264_nonconstant_p(x) 0
174 #include <kernel/OS.h>
175 #define x264_pthread_t thread_id
176 static inline int x264_pthread_create( x264_pthread_t *t, void *a, void *(*f)(void *), void *d )
178 *t = spawn_thread( f, "", 10, d );
179 if( *t < B_NO_ERROR )
184 #define x264_pthread_join(t,s) { long tmp; \
185 wait_for_thread(t,(s)?(long*)(s):&tmp); }
187 #elif HAVE_POSIXTHREAD
189 #define x264_pthread_t pthread_t
190 #define x264_pthread_create pthread_create
191 #define x264_pthread_join pthread_join
192 #define x264_pthread_mutex_t pthread_mutex_t
193 #define x264_pthread_mutex_init pthread_mutex_init
194 #define x264_pthread_mutex_destroy pthread_mutex_destroy
195 #define x264_pthread_mutex_lock pthread_mutex_lock
196 #define x264_pthread_mutex_unlock pthread_mutex_unlock
197 #define x264_pthread_cond_t pthread_cond_t
198 #define x264_pthread_cond_init pthread_cond_init
199 #define x264_pthread_cond_destroy pthread_cond_destroy
200 #define x264_pthread_cond_broadcast pthread_cond_broadcast
201 #define x264_pthread_cond_wait pthread_cond_wait
202 #define x264_pthread_attr_t pthread_attr_t
203 #define x264_pthread_attr_init pthread_attr_init
204 #define x264_pthread_attr_destroy pthread_attr_destroy
205 #define x264_pthread_num_processors_np pthread_num_processors_np
206 #define X264_PTHREAD_MUTEX_INITIALIZER PTHREAD_MUTEX_INITIALIZER
208 #elif HAVE_WIN32THREAD
209 #include "win32thread.h"
212 #define x264_pthread_t int
213 #define x264_pthread_create(t,u,f,d) 0
214 #define x264_pthread_join(t,s)
215 #endif //HAVE_*THREAD
217 #if !HAVE_POSIXTHREAD && !HAVE_WIN32THREAD
218 #define x264_pthread_mutex_t int
219 #define x264_pthread_mutex_init(m,f) 0
220 #define x264_pthread_mutex_destroy(m)
221 #define x264_pthread_mutex_lock(m)
222 #define x264_pthread_mutex_unlock(m)
223 #define x264_pthread_cond_t int
224 #define x264_pthread_cond_init(c,f) 0
225 #define x264_pthread_cond_destroy(c)
226 #define x264_pthread_cond_broadcast(c)
227 #define x264_pthread_cond_wait(c,m)
228 #define x264_pthread_attr_t int
229 #define x264_pthread_attr_init(a) 0
230 #define x264_pthread_attr_destroy(a)
231 #define X264_PTHREAD_MUTEX_INITIALIZER 0
234 #if HAVE_WIN32THREAD || PTW32_STATIC_LIB
235 int x264_threading_init( void );
237 #define x264_threading_init() 0
240 static ALWAYS_INLINE int x264_pthread_fetch_and_add( int *val, int add, x264_pthread_mutex_t *mutex )
243 #if defined(__GNUC__) && (__GNUC__ > 4 || __GNUC__ == 4 && __GNUC_MINOR__ > 0) && ARCH_X86
244 return __sync_fetch_and_add( val, add );
246 x264_pthread_mutex_lock( mutex );
249 x264_pthread_mutex_unlock( mutex );
259 #define WORD_SIZE sizeof(void*)
264 #define endian_fix(x) (x)
265 #define endian_fix64(x) (x)
266 #define endian_fix32(x) (x)
267 #define endian_fix16(x) (x)
269 #if HAVE_X86_INLINE_ASM && HAVE_MMX
270 static ALWAYS_INLINE uint32_t endian_fix32( uint32_t x )
272 asm("bswap %0":"+r"(x));
275 #elif defined(__GNUC__) && HAVE_ARMV6
276 static ALWAYS_INLINE uint32_t endian_fix32( uint32_t x )
278 asm("rev %0, %0":"+r"(x));
282 static ALWAYS_INLINE uint32_t endian_fix32( uint32_t x )
284 return (x<<24) + ((x<<8)&0xff0000) + ((x>>8)&0xff00) + (x>>24);
287 #if HAVE_X86_INLINE_ASM && ARCH_X86_64
288 static ALWAYS_INLINE uint64_t endian_fix64( uint64_t x )
290 asm("bswap %0":"+r"(x));
294 static ALWAYS_INLINE uint64_t endian_fix64( uint64_t x )
296 return endian_fix32(x>>32) + ((uint64_t)endian_fix32(x)<<32);
299 static ALWAYS_INLINE intptr_t endian_fix( intptr_t x )
301 return WORD_SIZE == 8 ? endian_fix64(x) : endian_fix32(x);
303 static ALWAYS_INLINE uint16_t endian_fix16( uint16_t x )
305 return (x<<8)|(x>>8);
309 /* For values with 4 bits or less. */
310 static int ALWAYS_INLINE x264_ctz_4bit( uint32_t x )
312 static uint8_t lut[16] = {4,0,1,0,2,0,1,0,3,0,1,0,2,0,1,0};
316 #if defined(__GNUC__) && (__GNUC__ > 3 || __GNUC__ == 3 && __GNUC_MINOR__ > 3)
317 #define x264_clz(x) __builtin_clz(x)
318 #define x264_ctz(x) __builtin_ctz(x)
320 static int ALWAYS_INLINE x264_clz( uint32_t x )
322 static uint8_t lut[16] = {4,3,2,2,1,1,1,1,0,0,0,0,0,0,0,0};
323 int y, z = (((x >> 16) - 1) >> 27) & 16;
325 z += y = ((x - 0x100) >> 28) & 8;
327 z += y = ((x - 0x10) >> 29) & 4;
332 static int ALWAYS_INLINE x264_ctz( uint32_t x )
334 static uint8_t lut[16] = {4,0,1,0,2,0,1,0,3,0,1,0,2,0,1,0};
335 int y, z = (((x & 0xffff) - 1) >> 27) & 16;
337 z += y = (((x & 0xff) - 1) >> 28) & 8;
339 z += y = (((x & 0xf) - 1) >> 29) & 4;
341 return z + lut[x&0xf];
345 #if HAVE_X86_INLINE_ASM && HAVE_MMX
346 /* Don't use __builtin_prefetch; even as recent as 4.3.4, GCC seems incapable of
347 * using complex address modes properly unless we use inline asm. */
348 static ALWAYS_INLINE void x264_prefetch( void *p )
350 asm volatile( "prefetcht0 %0"::"m"(*(uint8_t*)p) );
352 /* We require that prefetch not fault on invalid reads, so we only enable it on
353 * known architectures. */
354 #elif defined(__GNUC__) && (__GNUC__ > 3 || __GNUC__ == 3 && __GNUC_MINOR__ > 1) &&\
355 (ARCH_X86 || ARCH_X86_64 || ARCH_ARM || ARCH_PPC)
356 #define x264_prefetch(x) __builtin_prefetch(x)
358 #define x264_prefetch(x)
363 #define x264_lower_thread_priority(p)\
365 x264_pthread_t handle = pthread_self();\
366 struct sched_param sp;\
367 int policy = SCHED_OTHER;\
368 pthread_getschedparam( handle, &policy, &sp );\
369 sp.sched_priority -= p;\
370 pthread_setschedparam( handle, policy, &sp );\
374 #define x264_lower_thread_priority(p)\
375 { UNUSED status_t nice_ret = set_thread_priority( find_thread( NULL ), B_LOW_PRIORITY ); }
378 #define x264_lower_thread_priority(p) { UNUSED int nice_ret = nice(p); }
379 #endif /* SYS_WINDOWS */
380 #elif HAVE_WIN32THREAD
381 #define x264_lower_thread_priority(p) SetThreadPriority( GetCurrentThread(), X264_MAX( -2, -p ) )
383 #define x264_lower_thread_priority(p)
386 static inline int x264_is_regular_file( FILE *filehandle )
388 x264_struct_stat file_stat;
389 if( x264_fstat( fileno( filehandle ), &file_stat ) )
391 return S_ISREG( file_stat.st_mode );
394 static inline int x264_is_regular_file_path( const char *filename )
396 x264_struct_stat file_stat;
397 if( x264_stat( filename, &file_stat ) )
398 return !x264_is_pipe( filename );
399 return S_ISREG( file_stat.st_mode );
402 #endif /* X264_OSDEP_H */