/usr/include/tbb/machine/linux_ia32.h is in libtbb-dev 4.4~20151115-0ubuntu3.
This file is owned by root:root, with mode 0o644.
The actual contents of the file can be viewed below.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 | /*
Copyright 2005-2015 Intel Corporation. All Rights Reserved.
This file is part of Threading Building Blocks. Threading Building Blocks is free software;
you can redistribute it and/or modify it under the terms of the GNU General Public License
version 2 as published by the Free Software Foundation. Threading Building Blocks is
distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the
implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
See the GNU General Public License for more details. You should have received a copy of
the GNU General Public License along with Threading Building Blocks; if not, write to the
Free Software Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
As a special exception, you may use this file as part of a free software library without
restriction. Specifically, if other files instantiate templates or use macros or inline
functions from this file, or you compile this file and link it with other files to produce
an executable, this file does not by itself cause the resulting executable to be covered
by the GNU General Public License. This exception does not however invalidate any other
reasons why the executable file might be covered by the GNU General Public License.
*/
#if !defined(__TBB_machine_H) || defined(__TBB_machine_linux_ia32_H)
#error Do not #include this internal file directly; use public TBB headers instead.
#endif
#define __TBB_machine_linux_ia32_H
#include <stdint.h>
#include "gcc_ia32_common.h"
#define __TBB_WORDSIZE 4
#define __TBB_ENDIANNESS __TBB_ENDIAN_LITTLE
#define __TBB_compiler_fence() __asm__ __volatile__("": : :"memory")
#define __TBB_control_consistency_helper() __TBB_compiler_fence()
#define __TBB_acquire_consistency_helper() __TBB_compiler_fence()
#define __TBB_release_consistency_helper() __TBB_compiler_fence()
#define __TBB_full_memory_fence() __asm__ __volatile__("mfence": : :"memory")
#if __TBB_ICC_ASM_VOLATILE_BROKEN
#define __TBB_VOLATILE
#else
#define __TBB_VOLATILE volatile
#endif
#define __TBB_MACHINE_DEFINE_ATOMICS(S,T,X,R) \
static inline T __TBB_machine_cmpswp##S (volatile void *ptr, T value, T comparand ) \
{ \
T result; \
\
__asm__ __volatile__("lock\ncmpxchg" X " %2,%1" \
: "=a"(result), "=m"(*(__TBB_VOLATILE T*)ptr) \
: "q"(value), "0"(comparand), "m"(*(__TBB_VOLATILE T*)ptr) \
: "memory"); \
return result; \
} \
\
static inline T __TBB_machine_fetchadd##S(volatile void *ptr, T addend) \
{ \
T result; \
__asm__ __volatile__("lock\nxadd" X " %0,%1" \
: R (result), "=m"(*(__TBB_VOLATILE T*)ptr) \
: "0"(addend), "m"(*(__TBB_VOLATILE T*)ptr) \
: "memory"); \
return result; \
} \
\
static inline T __TBB_machine_fetchstore##S(volatile void *ptr, T value) \
{ \
T result; \
__asm__ __volatile__("lock\nxchg" X " %0,%1" \
: R (result), "=m"(*(__TBB_VOLATILE T*)ptr) \
: "0"(value), "m"(*(__TBB_VOLATILE T*)ptr) \
: "memory"); \
return result; \
} \
__TBB_MACHINE_DEFINE_ATOMICS(1,int8_t,"","=q")
__TBB_MACHINE_DEFINE_ATOMICS(2,int16_t,"","=r")
__TBB_MACHINE_DEFINE_ATOMICS(4,int32_t,"l","=r")
#if __INTEL_COMPILER
#pragma warning( push )
// reference to EBX in a function requiring stack alignment
#pragma warning( disable: 998 )
#endif
#if __TBB_GCC_CAS8_BUILTIN_INLINING_BROKEN
#define __TBB_IA32_CAS8_NOINLINE __attribute__ ((noinline))
#else
#define __TBB_IA32_CAS8_NOINLINE
#endif
static inline __TBB_IA32_CAS8_NOINLINE int64_t __TBB_machine_cmpswp8 (volatile void *ptr, int64_t value, int64_t comparand ) {
//TODO: remove the extra part of condition once __TBB_GCC_BUILTIN_ATOMICS_PRESENT is lowered to gcc version 4.1.2
#if (__TBB_GCC_BUILTIN_ATOMICS_PRESENT || (__TBB_GCC_VERSION >= 40102)) && !__TBB_GCC_64BIT_ATOMIC_BUILTINS_BROKEN
return __sync_val_compare_and_swap( reinterpret_cast<volatile int64_t*>(ptr), comparand, value );
#else /* !__TBB_GCC_BUILTIN_ATOMICS_PRESENT */
//TODO: look like ICC 13.0 has some issues with this code, investigate it more deeply
int64_t result;
union {
int64_t i64;
int32_t i32[2];
};
i64 = value;
#if __PIC__
/* compiling position-independent code */
// EBX register preserved for compliance with position-independent code rules on IA32
int32_t tmp;
__asm__ __volatile__ (
"movl %%ebx,%2\n\t"
"movl %5,%%ebx\n\t"
#if __GNUC__==3
"lock\n\t cmpxchg8b %1\n\t"
#else
"lock\n\t cmpxchg8b (%3)\n\t"
#endif
"movl %2,%%ebx"
: "=A"(result)
, "=m"(*(__TBB_VOLATILE int64_t *)ptr)
, "=m"(tmp)
#if __GNUC__==3
: "m"(*(__TBB_VOLATILE int64_t *)ptr)
#else
: "SD"(ptr)
#endif
, "0"(comparand)
, "m"(i32[0]), "c"(i32[1])
: "memory"
#if __INTEL_COMPILER
,"ebx"
#endif
);
#else /* !__PIC__ */
__asm__ __volatile__ (
"lock\n\t cmpxchg8b %1\n\t"
: "=A"(result), "=m"(*(__TBB_VOLATILE int64_t *)ptr)
: "m"(*(__TBB_VOLATILE int64_t *)ptr)
, "0"(comparand)
, "b"(i32[0]), "c"(i32[1])
: "memory"
);
#endif /* __PIC__ */
return result;
#endif /* !__TBB_GCC_BUILTIN_ATOMICS_PRESENT */
}
#undef __TBB_IA32_CAS8_NOINLINE
#if __INTEL_COMPILER
#pragma warning( pop )
#endif // warning 998 is back
static inline void __TBB_machine_or( volatile void *ptr, uint32_t addend ) {
__asm__ __volatile__("lock\norl %1,%0" : "=m"(*(__TBB_VOLATILE uint32_t *)ptr) : "r"(addend), "m"(*(__TBB_VOLATILE uint32_t *)ptr) : "memory");
}
static inline void __TBB_machine_and( volatile void *ptr, uint32_t addend ) {
__asm__ __volatile__("lock\nandl %1,%0" : "=m"(*(__TBB_VOLATILE uint32_t *)ptr) : "r"(addend), "m"(*(__TBB_VOLATILE uint32_t *)ptr) : "memory");
}
//TODO: Check if it possible and profitable for IA-32 architecture on (Linux* and Windows*)
//to use of 64-bit load/store via floating point registers together with full fence
//for sequentially consistent load/store, instead of CAS.
#if __clang__
#define __TBB_fildq "fildll"
#define __TBB_fistpq "fistpll"
#else
#define __TBB_fildq "fildq"
#define __TBB_fistpq "fistpq"
#endif
static inline int64_t __TBB_machine_aligned_load8 (const volatile void *ptr) {
__TBB_ASSERT(tbb::internal::is_aligned(ptr,8),"__TBB_machine_aligned_load8 should be used with 8 byte aligned locations only \n");
int64_t result;
__asm__ __volatile__ ( __TBB_fildq " %1\n\t"
__TBB_fistpq " %0" : "=m"(result) : "m"(*(const __TBB_VOLATILE uint64_t*)ptr) : "memory" );
return result;
}
static inline void __TBB_machine_aligned_store8 (volatile void *ptr, int64_t value ) {
__TBB_ASSERT(tbb::internal::is_aligned(ptr,8),"__TBB_machine_aligned_store8 should be used with 8 byte aligned locations only \n");
// Aligned store
__asm__ __volatile__ ( __TBB_fildq " %1\n\t"
__TBB_fistpq " %0" : "=m"(*(__TBB_VOLATILE int64_t*)ptr) : "m"(value) : "memory" );
}
static inline int64_t __TBB_machine_load8 (const volatile void *ptr) {
#if __TBB_FORCE_64BIT_ALIGNMENT_BROKEN
if( tbb::internal::is_aligned(ptr,8)) {
#endif
return __TBB_machine_aligned_load8(ptr);
#if __TBB_FORCE_64BIT_ALIGNMENT_BROKEN
} else {
// Unaligned load
return __TBB_machine_cmpswp8(const_cast<void*>(ptr),0,0);
}
#endif
}
//! Handles misaligned 8-byte store
/** Defined in tbb_misc.cpp */
extern "C" void __TBB_machine_store8_slow( volatile void *ptr, int64_t value );
extern "C" void __TBB_machine_store8_slow_perf_warning( volatile void *ptr );
static inline void __TBB_machine_store8(volatile void *ptr, int64_t value) {
#if __TBB_FORCE_64BIT_ALIGNMENT_BROKEN
if( tbb::internal::is_aligned(ptr,8)) {
#endif
__TBB_machine_aligned_store8(ptr,value);
#if __TBB_FORCE_64BIT_ALIGNMENT_BROKEN
} else {
// Unaligned store
#if TBB_USE_PERFORMANCE_WARNINGS
__TBB_machine_store8_slow_perf_warning(ptr);
#endif /* TBB_USE_PERFORMANCE_WARNINGS */
__TBB_machine_store8_slow(ptr,value);
}
#endif
}
// Machine specific atomic operations
#define __TBB_AtomicOR(P,V) __TBB_machine_or(P,V)
#define __TBB_AtomicAND(P,V) __TBB_machine_and(P,V)
#define __TBB_USE_GENERIC_DWORD_FETCH_ADD 1
#define __TBB_USE_GENERIC_DWORD_FETCH_STORE 1
#define __TBB_USE_FETCHSTORE_AS_FULL_FENCED_STORE 1
#define __TBB_USE_GENERIC_HALF_FENCED_LOAD_STORE 1
#define __TBB_USE_GENERIC_RELAXED_LOAD_STORE 1
#define __TBB_USE_GENERIC_SEQUENTIAL_CONSISTENCY_LOAD_STORE 1
|