1 /*
2 Compatibility <intrin.h> header for GCC -- GCC equivalents of intrinsic
3 Microsoft Visual C++ functions. Originally developed for the ReactOS
4 (<https://reactos.org/>) and TinyKrnl (<http://www.tinykrnl.org/>)
5 projects.
6
7 Copyright (c) 2006 KJK::Hyperion <hackbunny@reactos.com>
8
9 Permission is hereby granted, free of charge, to any person obtaining a
10 copy of this software and associated documentation files (the "Software"),
11 to deal in the Software without restriction, including without limitation
12 the rights to use, copy, modify, merge, publish, distribute, sublicense,
13 and/or sell copies of the Software, and to permit persons to whom the
14 Software is furnished to do so, subject to the following conditions:
15
16 The above copyright notice and this permission notice shall be included in
17 all copies or substantial portions of the Software.
18
19 THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
20 IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
21 FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
22 AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
23 LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
24 FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
25 DEALINGS IN THE SOFTWARE.
26 */
27
28 #ifndef KJK_INTRIN_ARM_H_
29 #define KJK_INTRIN_ARM_H_
30
31 #ifndef __GNUC__
32 #error Unsupported compiler
33 #endif
34
35 #define _ReturnAddress() (__builtin_return_address(0))
36 #define _ReadWriteBarrier() __sync_synchronize()
37
__yield(void)38 __INTRIN_INLINE void __yield(void) { __asm__ __volatile__("yield"); }
39
__break(unsigned int value)40 __INTRIN_INLINE void __break(unsigned int value) { __asm__ __volatile__("bkpt %0": : "M" (value)); }
41
_byteswap_ushort(unsigned short value)42 __INTRIN_INLINE unsigned short _byteswap_ushort(unsigned short value)
43 {
44 return (value >> 8) | (value << 8);
45 }
46
_CountLeadingZeros(long Mask)47 __INTRIN_INLINE unsigned _CountLeadingZeros(long Mask)
48 {
49 return Mask ? __builtin_clz(Mask) : 32;
50 }
51
_CountTrailingZeros(long Mask)52 __INTRIN_INLINE unsigned _CountTrailingZeros(long Mask)
53 {
54 return Mask ? __builtin_ctz(Mask) : 32;
55 }
56
_BitScanForward(unsigned long * const Index,const unsigned long Mask)57 __INTRIN_INLINE unsigned char _BitScanForward(unsigned long * const Index, const unsigned long Mask)
58 {
59 *Index = __builtin_ctz(Mask);
60 return Mask ? 1 : 0;
61 }
62
_InterlockedCompareExchange8(volatile char * const Destination,const char Exchange,const char Comperand)63 __INTRIN_INLINE char _InterlockedCompareExchange8(volatile char * const Destination, const char Exchange, const char Comperand)
64 {
65 return __sync_val_compare_and_swap(Destination, Comperand, Exchange);
66 }
67
_InterlockedCompareExchange16(volatile short * const Destination,const short Exchange,const short Comperand)68 __INTRIN_INLINE short _InterlockedCompareExchange16(volatile short * const Destination, const short Exchange, const short Comperand)
69 {
70 short a, b;
71
72 __asm__ __volatile__ ( "0:\n\t"
73 "ldr %1, [%2]\n\t"
74 "cmp %1, %4\n\t"
75 "bne 1f\n\t"
76 "swp %0, %3, [%2]\n\t"
77 "cmp %0, %1\n\t"
78 "swpne %3, %0, [%2]\n\t"
79 "bne 0b\n\t"
80 "1:"
81 : "=&r" (a), "=&r" (b)
82 : "r" (Destination), "r" (Exchange), "r" (Comperand)
83 : "cc", "memory");
84
85 return a;
86 }
87
_InterlockedExchangeAdd16(volatile short * const Addend,const short Value)88 __INTRIN_INLINE short _InterlockedExchangeAdd16(volatile short * const Addend, const short Value)
89 {
90 short a, b, c;
91
92 __asm__ __volatile__ ( "0:\n\t"
93 "ldr %0, [%3]\n\t"
94 "add %1, %0, %4\n\t"
95 "swp %2, %1, [%3]\n\t"
96 "cmp %0, %2\n\t"
97 "swpne %1, %2, [%3]\n\t"
98 "bne 0b"
99 : "=&r" (a), "=&r" (b), "=&r" (c)
100 : "r" (Value), "r" (Addend)
101 : "cc", "memory");
102
103 return a;
104 }
105
_InterlockedCompareExchange(volatile long * const dest,const long exch,const long comp)106 __INTRIN_INLINE long _InterlockedCompareExchange(volatile long * const dest, const long exch, const long comp)
107 {
108 long a, b;
109
110 __asm__ __volatile__ ( "0:\n\t"
111 "ldr %1, [%2]\n\t"
112 "cmp %1, %4\n\t"
113 "bne 1f\n\t"
114 "swp %0, %3, [%2]\n\t"
115 "cmp %0, %1\n\t"
116 "swpne %3, %0, [%2]\n\t"
117 "bne 0b\n\t"
118 "1:"
119 : "=&r" (a), "=&r" (b)
120 : "r" (dest), "r" (exch), "r" (comp)
121 : "cc", "memory");
122
123 return a;
124 }
125
_InterlockedCompareExchange64(volatile long long * const dest,const long long exch,const long long comp)126 __INTRIN_INLINE long long _InterlockedCompareExchange64(volatile long long * const dest, const long long exch, const long long comp)
127 {
128 //
129 // FIXME
130 //
131 long long result;
132 result = *dest;
133 if (*dest == comp) *dest = exch;
134 return result;
135 }
136
_InterlockedCompareExchangePointer(void * volatile * const Destination,void * const Exchange,void * const Comperand)137 __INTRIN_INLINE void * _InterlockedCompareExchangePointer(void * volatile * const Destination, void * const Exchange, void * const Comperand)
138 {
139 return (void*)_InterlockedCompareExchange((volatile long* const)Destination, (const long)Exchange, (const long)Comperand);
140 }
141
142
_InterlockedExchangeAdd(volatile long * const dest,const long add)143 __INTRIN_INLINE long _InterlockedExchangeAdd(volatile long * const dest, const long add)
144 {
145 long a, b, c;
146
147 __asm__ __volatile__ ( "0:\n\t"
148 "ldr %0, [%3]\n\t"
149 "add %1, %0, %4\n\t"
150 "swp %2, %1, [%3]\n\t"
151 "cmp %0, %2\n\t"
152 "swpne %1, %2, [%3]\n\t"
153 "bne 0b"
154 : "=&r" (a), "=&r" (b), "=&r" (c)
155 : "r" (dest), "r" (add)
156 : "cc", "memory");
157
158 return a;
159 }
160
_InterlockedExchange(volatile long * const dest,const long exch)161 __INTRIN_INLINE long _InterlockedExchange(volatile long * const dest, const long exch)
162 {
163 long a;
164
165 __asm__ __volatile__ ( "swp %0, %2, [%1]"
166 : "=&r" (a)
167 : "r" (dest), "r" (exch));
168
169 return a;
170 }
171
172
_InterlockedExchangePointer(void * volatile * const Target,void * const Value)173 __INTRIN_INLINE void * _InterlockedExchangePointer(void * volatile * const Target, void * const Value)
174 {
175 return (void *)_InterlockedExchange((volatile long * const)Target, (const long)Value);
176 }
177
178
179
_BitScanReverse(unsigned long * const Index,const unsigned long Mask)180 __INTRIN_INLINE unsigned char _BitScanReverse(unsigned long * const Index, const unsigned long Mask)
181 {
182 *Index = 31 - __builtin_clz(Mask);
183 return Mask ? 1 : 0;
184 }
185
_InterlockedAnd8(volatile char * const value,const char mask)186 __INTRIN_INLINE char _InterlockedAnd8(volatile char * const value, const char mask)
187 {
188 char x;
189 char y;
190
191 y = *value;
192
193 do
194 {
195 x = y;
196 y = _InterlockedCompareExchange8(value, x & mask, x);
197 }
198 while(y != x);
199
200 return y;
201 }
202
_InterlockedAnd16(volatile short * const value,const short mask)203 __INTRIN_INLINE short _InterlockedAnd16(volatile short * const value, const short mask)
204 {
205 short x;
206 short y;
207
208 y = *value;
209
210 do
211 {
212 x = y;
213 y = _InterlockedCompareExchange16(value, x & mask, x);
214 }
215 while(y != x);
216
217 return y;
218 }
219
_InterlockedAnd(volatile long * const value,const long mask)220 __INTRIN_INLINE long _InterlockedAnd(volatile long * const value, const long mask)
221 {
222 long x;
223 long y;
224
225 y = *value;
226
227 do
228 {
229 x = y;
230 y = _InterlockedCompareExchange(value, x & mask, x);
231 }
232 while(y != x);
233
234 return y;
235 }
236
_InterlockedOr8(volatile char * const value,const char mask)237 __INTRIN_INLINE char _InterlockedOr8(volatile char * const value, const char mask)
238 {
239 char x;
240 char y;
241
242 y = *value;
243
244 do
245 {
246 x = y;
247 y = _InterlockedCompareExchange8(value, x | mask, x);
248 }
249 while(y != x);
250
251 return y;
252 }
253
_InterlockedOr16(volatile short * const value,const short mask)254 __INTRIN_INLINE short _InterlockedOr16(volatile short * const value, const short mask)
255 {
256 short x;
257 short y;
258
259 y = *value;
260
261 do
262 {
263 x = y;
264 y = _InterlockedCompareExchange16(value, x | mask, x);
265 }
266 while(y != x);
267
268 return y;
269 }
270
_InterlockedOr(volatile long * const value,const long mask)271 __INTRIN_INLINE long _InterlockedOr(volatile long * const value, const long mask)
272 {
273 long x;
274 long y;
275
276 y = *value;
277
278 do
279 {
280 x = y;
281 y = _InterlockedCompareExchange(value, x | mask, x);
282 }
283 while(y != x);
284
285 return y;
286 }
287
_InterlockedXor8(volatile char * const value,const char mask)288 __INTRIN_INLINE char _InterlockedXor8(volatile char * const value, const char mask)
289 {
290 char x;
291 char y;
292
293 y = *value;
294
295 do
296 {
297 x = y;
298 y = _InterlockedCompareExchange8(value, x ^ mask, x);
299 }
300 while(y != x);
301
302 return y;
303 }
304
_InterlockedXor16(volatile short * const value,const short mask)305 __INTRIN_INLINE short _InterlockedXor16(volatile short * const value, const short mask)
306 {
307 short x;
308 short y;
309
310 y = *value;
311
312 do
313 {
314 x = y;
315 y = _InterlockedCompareExchange16(value, x ^ mask, x);
316 }
317 while(y != x);
318
319 return y;
320 }
321
_InterlockedXor(volatile long * const value,const long mask)322 __INTRIN_INLINE long _InterlockedXor(volatile long * const value, const long mask)
323 {
324 long x;
325 long y;
326
327 y = *value;
328
329 do
330 {
331 x = y;
332 y = _InterlockedCompareExchange(value, x ^ mask, x);
333 }
334 while(y != x);
335
336 return y;
337 }
338
_InterlockedDecrement(volatile long * const lpAddend)339 __INTRIN_INLINE long _InterlockedDecrement(volatile long * const lpAddend)
340 {
341 return _InterlockedExchangeAdd(lpAddend, -1) - 1;
342 }
343
_InterlockedIncrement(volatile long * const lpAddend)344 __INTRIN_INLINE long _InterlockedIncrement(volatile long * const lpAddend)
345 {
346 return _InterlockedExchangeAdd(lpAddend, 1) + 1;
347 }
348
_InterlockedDecrement16(volatile short * const lpAddend)349 __INTRIN_INLINE long _InterlockedDecrement16(volatile short * const lpAddend)
350 {
351 return _InterlockedExchangeAdd16(lpAddend, -1) - 1;
352 }
353
_InterlockedIncrement16(volatile short * const lpAddend)354 __INTRIN_INLINE long _InterlockedIncrement16(volatile short * const lpAddend)
355 {
356 return _InterlockedExchangeAdd16(lpAddend, 1) + 1;
357 }
358
_InterlockedAddLargeStatistic(volatile long long * const Addend,const long Value)359 __INTRIN_INLINE long _InterlockedAddLargeStatistic(volatile long long * const Addend, const long Value)
360 {
361 *Addend += Value;
362 return Value;
363 }
364
_disable(void)365 __INTRIN_INLINE void _disable(void)
366 {
367 __asm__ __volatile__
368 (
369 "cpsid i @ __cli" : : : "memory", "cc"
370 );
371 }
372
_enable(void)373 __INTRIN_INLINE void _enable(void)
374 {
375 __asm__ __volatile__
376 (
377 "cpsie i @ __sti" : : : "memory", "cc"
378 );
379 }
380
_interlockedbittestandset(volatile long * a,const long b)381 __INTRIN_INLINE unsigned char _interlockedbittestandset(volatile long * a, const long b)
382 {
383 return (_InterlockedOr(a, 1 << b) >> b) & 1;
384 }
385
_interlockedbittestandreset(volatile long * a,const long b)386 __INTRIN_INLINE unsigned char _interlockedbittestandreset(volatile long * a, const long b)
387 {
388 return (_InterlockedAnd(a, ~(1 << b)) >> b) & 1;
389 }
390
391 #ifndef __MSVCRT__
_rotl(const unsigned int value,int shift)392 __INTRIN_INLINE unsigned int _rotl(const unsigned int value, int shift)
393 {
394 return (((value) << ((int)(shift))) | ((value) >> (32 - (int)(shift))));
395 }
396 #endif
397
398 #define _clz(a) \
399 ({ ULONG __value, __arg = (a); \
400 asm ("clz\t%0, %1": "=r" (__value): "r" (__arg)); \
401 __value; })
402
403 #endif
404 /* EOF */
405