File: ppc_intrinsics.h

package info (click to toggle)
libmikmod 3.3.10-1
  • links: PTS, VCS
  • area: main
  • in suites: stretch
  • size: 5,900 kB
  • ctags: 5,175
  • sloc: ansic: 34,289; sh: 4,433; makefile: 522
file content (1038 lines) | stat: -rw-r--r-- 31,874 bytes parent folder | download | duplicates (3)
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
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
/* APPLE LOCAL file PPC_INTRINSICS */

/* Definitions for PowerPC intrinsic instructions
   Copyright (C) 2002, 2003, 2004 Free Software Foundation, Inc.

This file is part of GCC.

GCC is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free
Software Foundation; either version 2, or (at your option) any later
version.

GCC 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 GCC; see the file COPYING.  If not, write to the Free
Software Foundation, 59 Temple Place - Suite 330, Boston, MA
02111-1307, USA.  */

/* As a special exception, if you include this header file into source
   files compiled by GCC, this header 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.  */

/*
 * The following PowerPC intrinsics are provided by this header:
 *
 * Low-Level Processor Synchronization
 *   __eieio    - Enforce In-Order Execution of I/O
 *   __isync    - Instruction Synchronize
 *   __sync     - Synchronize
 *   __lwsync   - Lightweight Synchronize
 *
 * Manipulating the Contents of a Variable or Register
 *   __cntlzw   - Count Leading Zeros Word
 *   __cntlzd   - Count Leading Zeros Double Word
 *   __rlwimi   - Rotate Left Word Immediate then Mask Insert
 *   __rlwinm   - Rotate Left Word Immediate then AND with Mask
 *   __rlwnm    - Rotate Left Word then AND with Mask
 *
 * Byte-Reversing Functions
 *   __lhbrx    - Load Half Word Byte-Reverse Indexed
 *   __lwbrx    - Load Word Byte-Reverse Indexed
 *   __sthbrx   - Store Half Word Byte-Reverse Indexed
 *   __stwbrx   - Store Word Byte-Reverse Indexed
 *
 * Data Cache Manipulation
 *   __dcba     - Data Cache Block Allocate
 *   __dcbf     - Data Cache Block Flush
 *   __dcbst    - Data Cache Block Store
 *   __dcbt     - Data Cache Block Touch
 *   __dcbtst   - Data Cache Block Touch for Store
 *   __dcbzl    - Data Cache Block Set to Zero
 *   __dcbz     - Data Cache Block Set to Zero (32-bytes only)
 *
 * Setting the Floating-Point Environment
 *   __setflm   - Set Floating-point Mode
 *
 * Math Functions
 *   __fabs     - Floating-Point Absolute Value
 *   __fnabs    - Floating Negative Absolute Value
 *   __fctiw    - Floating Convert to Integer Word
 *   __fctiwz   - Floating Convert to Integer Word with Round toward Zero
 *   __fctidz   - Floating Convert to Integer Doubleword with Round toward Zero
 *   __fctid    - Floating Convert to Integer Doubleword
 *   __fcfid    - Floating Convert From Integer Doubleword
 *   __fmadd    - Floating Multiply-Add (Double-Precision)
 *   __fmadds   - Floating Multiply-Add Single
 *   __fmsub    - Floating Multiply-Subract (Double-Precision)
 *   __fmsubs   - Floating Multiply-Subract Single
 *   __fmul     - Floating Multiply (Double-Precision)
 *   __fmuls    - Floating Multiply Single
 *   __fnmadd   - Floating Negative Multiply-Add (Double-Precision)
 *   __fnmadds  - Floating Negative Multiply-Add Single
 *   __fnmsub   - Floating Negative Multiply-Subtract (Double-Precision)
 *   __fnmsubs  - Floating Negative Multiply-Subtract Single
 *   __fres     - Floating Reciprocal Estimate
 *   __frsp     - Floating Round to Single-Precision
 *   __frsqrte  - Floating Reciprocal Square Root Estimate
 *   __frsqrtes - Floating Reciprocal Square Root Estimate Single
 *   __fsel     - Floating Select
 *   __fsels    - Floating Select (Single-Precision variant)
 *   __fsqrt    - Floating-Point Square Root (Double-Precision)
 *   __fsqrts   - Floating-Point Square Root Single-Precision
 *   __mulhw    - Multiply High Word
 *   __mulhwu   - Multiply High Word Unsigned
 *   __stfiwx   - Store Floating-Point as Integer Word Indexed
 *
 * Miscellaneous Functions
 *   __nop      - PPC preferred form of no operation
 *   __astrcmp  - assembly strcmp
 *   __icbi     - Instruction Cache Block Invalidate
 *   __mffs     - Move from FPSCR
 *   __mfspr    - Move from Special Purpose Register
 *   __mtfsf    - Move to SPSCR Fields
 *   __mtspr    - Move to Special Purpose Register
 *   __OSReadSwapSInt16 - lhbrx for signed shorts
 *   __OSReadSwapUInt16 - lhbrx for unsigned shorts
 *
 * TO DO:
 * - Desired:
 *   mullw
 * - Available in CodeWarrior, not yet implemented here:
 *   abs, labs, fabsf, fnabsf
 *
 * NOTES:
 * - Some of the intrinsics need to be macros because certain 
 *   parameters MUST be integer constants and not values in registers.
 * - The declarations use __asm__ instead of asm and __inline__ instead
 *   of inline to prevent errors when -ansi is specified.
 * - Some of the intrinsic definitions use the "volatile" specifier on
 *   the "asm" statements in order to work around what appears to be
 *   a bug in the compiler/optimizer.  In general we have avoided the
 *   use of "volatile" because it suppresses optimization on the
 *   generated instructions.  The instructions to which "volatile"
 *   has been added where it appears that it should not be needed are
 *   lhbrx and lwbrx.
 *
 * Contributors: Fred Forsman (editor), Turly O'Connor, Ian Ollmann, Sanjay Patel
 * Last modified: October 6, 2004
 */

#ifndef _PPC_INTRINSICS_H_
#define _PPC_INTRINSICS_H_

#if (defined(__ppc__) || defined(__ppc64__)) && ! defined(__MWERKS__)

/*******************************************************************
 *                 Special Purpose Registers (SPRs)                *
 *******************************************************************/

#define __SPR_MQR       0               /* PPC 601 only */
#define __SPR_XER       1               
#define __SPR_RTCU      4               /* Real time clock upper. PPC 601 only.*/
#define __SPR_RTCL      5               /* Real time clock lower. PPC 601 only.*/
#define __SPR_LR        8               
#define __SPR_CTR       9               
#define __SPR_VRSAVE    256             /* AltiVec */
#define __SPR_TBL       268             /* Time-base Lower. Not on PPC 601 */
#define __SPR_TBU       269             /* Time-base Upper. Not on PPC 601 */
#define __SPR_UMMCR2    928             /* PPC 74xx */
#define __SPR_UPMC5     929             /* PPC 745x */
#define __SPR_UPMC6     930             /* PPC 745x */
#define __SPR_UBAMR     935             /* PPC 7400 and 7410 */
#define __SPR_UMMCR0    936             /* PPC 74xx and 750 */
#define __SPR_UPMC1     937             /* PPC 74xx and 750 */
#define __SPR_UPMC2     938             /* PPC 74xx and 750 */
#define __SPR_USIAR     939             /* PPC 74xx and 750 */
#define __SPR_UMMCR1    940             /* PPC 74xx and 750 */
#define __SPR_UPMC3     941             /* PPC 74xx and 750 */
#define __SPR_UPMC4     942             /* PPC 74xx and 750 */
#define __SPR_PIR       1023            /* supervisor level only! */

/*
 * Shorthand macros for some commonly used SPR's.
 */
#define __mfxer()               __mfspr(__SPR_XER)
#define __mflr()                __mfspr(__SPR_LR)
#define __mfctr()               __mfspr(__SPR_CTR)
#define __mfvrsave()            __mfspr(__SPR_VRSAVE)
#define __mftb()                __mfspr(__SPR_TBL)
#define __mftbu()               __mfspr(__SPR_TBU)

#define __mtlr(value)           __mtspr(__SPR_LR, value)
#define __mtxer(value)          __mtspr(__SPR_XER, value)
#define __mtctr(value)          __mtspr(__SPR_CTR, value)
#define __mtvrsave(value)       __mtspr(__SPR_VRSAVE, value)


/*******************************************************************
 *               Low-Level Processor Synchronization               *
 *******************************************************************/

/*
 * __eieio - Enforce In-Order Execution of I/O
 *
 *   void __eieio (void);
 */
#define __eieio() __asm__ ("eieio" : : : "memory")

/*
 * __isync - Instruction Synchronize
 *
 *   void __isync (void);
 */
#define __isync()       \
  __asm__ volatile ("isync")

/*
 * __sync - Synchronize
 *
 *  void __sync (void);
 */
#define __sync() __asm__ volatile ("sync")

/*
 * __lwsync - Lightweight Synchronize, see PPC2.01, Book 2
 *
 *  void __lwsync (void);
 */
#define __lwsync() __asm__ volatile ("sync 1")


/*******************************************************************
 *                     Byte-Reversing Functions                    *
 *******************************************************************/

/*
 * __lhbrx - Load Half Word Byte-Reverse Indexed
 *
 *   int __lhbrx(void *, int);
 */
#define __lhbrx(base, index)   \
  ({ unsigned short __ppc_i_lhbrxResult;       \
     __asm__ volatile ("lhbrx %0, %1, %2" : "=r" (__ppc_i_lhbrxResult) : "b%" (index), "r" (base) : "memory"); \
     /*return*/ __ppc_i_lhbrxResult; })

/*
 * __lwbrx - Load Word Byte-Reverse Indexed
 *
 *   int __lwbrx(void *, int);
 */
#define __lwbrx(base, index)    \
  ({ unsigned int __ppc_i_lwbrxResult; \
     __asm__ volatile ("lwbrx %0, %1, %2" : "=r" (__ppc_i_lwbrxResult) : "b%" (index), "r" (base) : "memory"); \
     /*return*/ __ppc_i_lwbrxResult; })

/*
 * __sthbrx - Store Half Word Byte-Reverse Indexed
 *
 *   int __sthbrx(unsigned short, void *, int);
 */
#define __sthbrx(value, base, index)    \
  __asm__ ("sthbrx %0, %1, %2" : : "r" (value), "b%" (index), "r" (base) : "memory")

/*
 * __stwbrx - Store Word Byte-Reverse Indexed
 *
 *   int __sthbrx(unsigned int, void *, int);
 */
#define __stwbrx(value, base, index)    \
  __asm__ ("stwbrx %0, %1, %2" : : "r" (value), "b%" (index), "r" (base) : "memory")


/*******************************************************************
 *       Manipulating the Contents of a Variable or Register       *
 *******************************************************************/

/*
 * __cntlzw - Count Leading Zeros Word
 * __cntlzd - Count Leading Zeros Double Word
 */
#if (__GNUC__ < 3) || (__GNUC__ == 3 && __GNUC_MINOR__ < 4)
static inline int __cntlzw (int value) __attribute__((always_inline));
static inline int
__cntlzw (int value)
{
  long result;
  __asm__ ("cntlzw %0, %1"
           /* outputs:  */ : "=r" (result)
           /* inputs:   */ : "r" (value));
  return result;
}
#else
#define __cntlzw(a)     __builtin_clz(a)
#define __cntlzd(a)     __builtin_clzll(a)
#endif

/*
 * __rlwimi - Rotate Left Word Immediate then Mask Insert
 *
 *   int __rlwimi(int, long, int, int, int);
 *
 * We don't mention "%1" below: operand[1] needs to be skipped as 
 * it's just a placeholder to let the compiler know that rA is read 
 * from as well as written to. 
 */
#define __rlwimi(rA, rS, cnt, mb, me)                               \
  ({ __asm__ ("rlwimi %0,%2,%3,%4,%5" : "=r" (rA)                   \
              : "0" (rA), "r" (rS), "n" (cnt), "n" (mb), "n" (me)); \
     /*return*/ rA;})

/*
 * __rlwinm - Rotate Left Word Immediate then AND with Mask
 *
 *   int __rlwinm(long, int, int, int);
 */
#define __rlwinm(rS, cnt, mb, me)						\
  ({ unsigned int __ppc_i_val;							\
     __asm__ ("rlwinm %0,%1,%2,%3,%4" : "=r" (__ppc_i_val)			\
              : "r" (rS), "n" (cnt), "n" (mb), "n" (me));			\
     /*return*/ __ppc_i_val;})

/*
 * __rlwnm - Rotate Left Word then AND with Mask
 *
 *   int __rlwnm(long, int, int, int);
 */
#define __rlwnm(value, leftRotateBits, maskStart, maskEnd)			\
  ({ unsigned int __ppc_i_result;						\
     __asm__ ("rlwnm %0, %1, %2, %3, %4" : "=r" (__ppc_i_result) :		\
              "r" (value), "r" (leftRotateBits), "n" (maskStart), "n" (maskEnd)); \
     /*return */ __ppc_i_result; })


/*******************************************************************
 *                     Data Cache Manipulation                     *
 *******************************************************************/

/* 
 * --- Data Cache Block instructions ---
 *
 * Please see Motorola's "The Programming Environments for 32-Bit 
 * Microprocessors" for a description of what these do.
 *
 *   Parameter descriptions:
 *
 *     base             starting address for figuring out where the 
 *                      cacheline is
 *
 *     index            byte count to be added to the base address for 
 *                      purposes of calculating the effective address 
 *                      of the cacheline to be operated on.  
 *                                      
 *   Effective Address of cacheline to be manipulated = 
 *     (char*) base + index
 *      
 *   WARNING: The size and alignment of cachelines are subject to 
 *     change on future processors!  Cachelines are 32 bytes in 
 *     size and are aligned to 32 bytes on PowerPC 601, 603, 604, 
 *     750, 7400, 7410, 7450, and 7455.
 *
 */
 
/*
 * __dcba - Data Cache Block Allocate
 *
 *   void __dcba(void *, int)
 *
 * WARNING: dcba is a valid instruction only on PowerPC 7400, 7410, 
 *          7450 and 7455.
 */
#define __dcba(base, index)     \
  __asm__ ("dcba %0, %1" : /*no result*/ : "b%" (index), "r" (base) : "memory")

/*
 * __dcbf - Data Cache Block Flush
 *
 *   void __dcbf(void *, int);
 */
#define __dcbf(base, index)     \
  __asm__ ("dcbf %0, %1" : /*no result*/ : "b%" (index), "r" (base) : "memory")

/*
 * __dcbst - Data Cache Block Store
 *
 *   void __dcbst(void *, int);
 */
#define __dcbst(base, index)    \
  __asm__ ("dcbst %0, %1" : /*no result*/ : "b%" (index), "r" (base) : "memory")

/*
 * __dcbt - Data Cache Block Touch
 *
 *   void __dcbt(void *, int);
 */
#define __dcbt(base, index)     \
  __asm__ ("dcbt %0, %1" : /*no result*/ : "b%" (index), "r" (base) : "memory")

/*
 * __dcbtst - Data Cache Block Touch for Store
 *
 *   void __dcbtst(void *, int);
 */
#define __dcbtst(base, index)   \
  __asm__ ("dcbtst %0, %1" : /*no result*/ : "b%" (index), "r" (base) : "memory")

/*
 * __dcbzl - Data Cache Block Set to Zero
 *
 *   void __dcbzl(void *, int);
 */
#define __dcbzl(base, index)     \
  __asm__ ("dcbzl %0, %1" : /*no result*/ : "b%" (index), "r" (base) : "memory")

/*
 * __dcbz - Data Cache Block Set to Zero (32-bytes only)
 *
 * WARNING: this is for legacy purposes only
 *
 *   void __dcbz(void *, int);
 */
#define __dcbz(base, index)     \
  __asm__ ("dcbz %0, %1" : /*no result*/ : "b%" (index), "r" (base) : "memory")


/*******************************************************************
 *              Setting the Floating-Point Environment             *
 *******************************************************************/

/*
 * __setflm - Set Floating-point Mode
 *
 * Sets the FPSCR (floating-point status and control register),
 * returning the original value.
 *
 *   ??? CW: float __setflm(float);
 */
static inline double __setflm (double newflm) __attribute__((always_inline));
static inline double 
__setflm(double newflm)
{
  double original;

  __asm__ ("mffs %0" 
           /* outputs:  */ : "=f" (original));
  __asm__ ("mtfsf 255,%0" 
           /* outputs:  */ : /* none */ 
           /* inputs:   */ : "f" (newflm));
  return original;
}


/*******************************************************************
 *                          Math Functions                         *
 *******************************************************************/

/*
 * __fabs - Floating-Point Absolute Value
 */
static inline double __fabs (double value) __attribute__((always_inline));
static inline double 
__fabs (double value)
{
  double result;
  __asm__ ("fabs %0, %1" 
           /* outputs:  */ : "=f" (result) 
           /* inputs:   */ : "f" (value));
  return result;
}

/*
 * __fnabs - Floating Negative Absolute Value
 */
static inline double __fnabs (double b) __attribute__((always_inline));
static inline double 
__fnabs (double b)
{
  double result;
  __asm__ ("fnabs %0, %1" 
           /* outputs:  */ : "=f" (result) 
           /* inputs:   */ : "f" (b));
  return result;
}

/*
 * fctiw - Floating Convert to Integer Word
 *
 * Convert the input value to a signed long and place in the low 32 
 * bits of the FP register.  Clip to LONG_MIN or LONG_MAX if the FP 
 * value exceeds the range representable by a long.  Use the rounding
 * mode indicated in the FPSCR.
 */
static inline double __fctiw (double b) __attribute__((always_inline));
static inline double 
__fctiw (double b)
{
  double result;
  __asm__ ("fctiw %0, %1" 
           /* outputs:  */ : "=f" (result) 
           /* inputs:   */ : "f" (b));
  return result;
}

/*
 * fctiwz - Floating Convert to Integer Word with Round toward Zero
 *
 * Convert the input value to a signed long and place in the low 32 
 * bits of the FP register.  Clip to LONG_MIN or LONG_MAX if the FP 
 * value exceeds the range representable by a long.
 */
static inline double __fctiwz (double b) __attribute__((always_inline));
static inline double 
__fctiwz (double b)
{
  double result;
  __asm__ ("fctiwz %0, %1" 
           /* outputs:  */ : "=f" (result) 
           /* inputs:   */ : "f" (b));
  return result;
}

/*
 * fctidz - Floating Convert to Integer Double Word with Round toward Zero
 *
 * Convert the input value to a signed 64-bit int and place in the FP
 * destination register.  Clip to LLONG_MIN (-2**63) or LLONG_MAX (2**63-1) 
 * if the FP value exceeds the range representable by a int64_t.
 * 
 * WARNING: fctidz is a valid instruction only on 64-bit PowerPC 
 */
static inline double __fctidz (double b) __attribute__((always_inline));
static inline double 
__fctidz (double b)
{
  double result;
  __asm__ ("fctidz %0, %1" 
           /* outputs:  */ : "=f" (result) 
           /* inputs:   */ : "f" (b));
  return result;
}

/*
 * fctid - Floating Convert to Integer Double Word
 *
 * Convert the input value to a signed 64-bit int and place in the FP
 * destination register.  Clip to LLONG_MIN (-2**63) or LLONG_MAX (2**63-1) 
 * if the FP value exceeds the range representable by a int64_t. Use the 
 * rounding mode indicated in the FPSCR.
 * 
 * WARNING: fctid is a valid instruction only on 64-bit PowerPC 
 */
static inline double __fctid (double b) __attribute__((always_inline));
static inline double 
__fctid (double b)
{
  double result;
  __asm__ ("fctid %0, %1" 
           /* outputs:  */ : "=f" (result) 
           /* inputs:   */ : "f" (b));
  return result;
}

/*
 * fcfid - Floating Convert From Integer Double Word
 *
 * Convert the 64-bit signed integer input value to a 64-bit FP value.  
 * Use the rounding mode indicated in the FPSCR if the integer is out of
 * double precision range.
 * 
 * WARNING: fcfid is a valid instruction only on 64-bit PowerPC 
 */
static inline double __fcfid (double b) __attribute__((always_inline));
static inline double 
__fcfid (double b)
{
  double result;
  __asm__ ("fcfid %0, %1" 
           /* outputs:  */ : "=f" (result) 
           /* inputs:   */ : "f" (b));
  return result;
}

/*
 * fmadd - Floating Multiply-Add (Double-Precision)
 *
 *   (a * c + b) double precision
 */
static inline double __fmadd (double a, double c, double b) __attribute__((always_inline));
static inline double 
__fmadd (double  a, double c, double b)
{
  double result;
  __asm__ ("fmadd %0, %1, %2, %3" 
           /* outputs:  */ : "=f" (result) 
           /* inputs:   */ : "f" (a), "f" (c), "f" (b));
  return result;
}

/*
 * fmadds - Floating Multiply-Add Single
 *
 *   (a * c + b) single precision
 *
 * Double precision arguments are used to prevent the compiler from
 * issuing frsp instructions upstream.
 */
static inline float __fmadds (double a, double c, double b) __attribute__((always_inline));
static inline float 
__fmadds (double  a, double c, double b)
{
  float result;
  __asm__ ("fmadds %0, %1, %2, %3" 
           /* outputs:  */ : "=f" (result) 
           /* inputs:   */ : "f" (a), "f" (c), "f" (b));
  return result;
}

/*
 * fmsub - Floating Multiply-Subract (Double-Precision)
 *
 *   (a * c - b) double precision
 */
static inline double __fmsub (double a, double c, double b) __attribute__((always_inline));
static inline double 
__fmsub (double  a, double c, double b)
{
  double result;
  __asm__ ("fmsub %0, %1, %2, %3" 
           /* outputs:  */ : "=f" (result) 
           /* inputs:   */ : "f" (a), "f" (c), "f" (b));
  return result;
}

/*
 * fmsubs - Floating Multiply-Subract Single
 *
 *   (a * c - b) single precision
 *
 * Double precision arguments are used to prevent the compiler from
 * issuing frsp instructions upstream.
 */
static inline float __fmsubs (double a, double c, double b) __attribute__((always_inline));
static inline float 
__fmsubs (double  a, double c, double b)
{
  float result;
  __asm__ ("fmsubs %0, %1, %2, %3" 
           /* outputs:  */ : "=f" (result) 
           /* inputs:   */ : "f" (a), "f" (c), "f" (b));
  return result;
}

/*
 * fmul - Floating Multiply (Double-Precision)
 *
 *   (a * c) double precision
 */
static inline double __fmul (double a, double c) __attribute__((always_inline));
static inline double 
__fmul (double  a, double c)
{
  double result;
  __asm__ ("fmul %0, %1, %2" 
           /* outputs:  */ : "=f" (result) 
           /* inputs:   */ : "f" (a), "f" (c));
  return result;
}

/*
 * fmuls - Floating Multiply Single
 *
 *   (a * c) single precision
 *
 * Double precision arguments are used to prevent the compiler from
 * issuing frsp instructions upstream.
 */
static inline float __fmuls (double a, double c) __attribute__((always_inline));
static inline float 
__fmuls (double  a, double c)
{
  float result;
  __asm__ ("fmuls %0, %1, %2" 
           /* outputs:  */ : "=f" (result) 
           /* inputs:   */ : "f" (a), "f" (c));
  return result;
}

/*
 * __fnmadd - Floating Negative Multiply-Add (Double-Precision)
 *
 *   -(a * c + b) double precision
 */
static inline double __fnmadd (double a, double c, double b) __attribute__((always_inline));
static inline double 
__fnmadd (double  a, double c, double b)
{
  double result;
  __asm__ ("fnmadd %0, %1, %2, %3" 
           /* outputs:  */ : "=f" (result) 
           /* inputs:   */ : "f" (a), "f" (c), "f" (b));
  return result;
}

/*
 * __fnmadds - Floating Negative Multiply-Add Single
 *
 *   -(a * c + b) single precision
 *
 * Double precision arguments are used to prevent the compiler from
 * issuing frsp instructions upstream.
 */
static inline float __fnmadds (double a, double c, double b) __attribute__((always_inline));
static inline float 
__fnmadds (double  a, double c, double b)
{
  float result;
  __asm__ ("fnmadds %0, %1, %2, %3" 
           /* outputs:  */ : "=f" (result) 
           /* inputs:   */ : "f" (a), "f" (c), "f" (b));
  return result;
}

/*
 * __fnmsub - Floating Negative Multiply-Subtract (Double-Precision)
 *
 *   -(a * c - B) double precision
 */
static inline double __fnmsub (double a, double c, double b) __attribute__((always_inline));
static inline double 
__fnmsub (double  a, double c, double b)
{
  double result;
  __asm__ ("fnmsub %0, %1, %2, %3" 
           /* outputs:  */ : "=f" (result) 
           /* inputs:   */ : "f" (a), "f" (c), "f" (b));
  return result;
}

/*
 * __fnmsubs - Floating Negative Multiply-Subtract Single
 *
 *   -(a * c - b) single precision
 *
 * Double precision arguments are used to prevent the compiler from
 * issuing frsp instructions upstream.
 */
static inline float __fnmsubs (double a, double c, double b) __attribute__((always_inline));
static inline float 
__fnmsubs (double  a, double c, double b)
{
  float result;
  __asm__ ("fnmsubs %0, %1, %2, %3" 
           /* outputs:  */ : "=f" (result) 
           /* inputs:   */ : "f" (a), "f" (c), "f" (b));
  return result;
}

/*
 * __fres - Floating Reciprocal Estimate
 *
 * Produces a double precision result with 5 bits of accuracy.
 * Note: not valid on the PowerPC 601.
 *
 * ??? CW: float __fres(float)
 */
static inline float __fres (float val) __attribute__((always_inline));
static inline float 
__fres (float val)
{
  float estimate;
  __asm__ ("fres %0,%1" 
           /* outputs:  */ : "=f" (estimate) 
           /* inputs:   */ : "f" (val));
  return estimate;
}

/*
 * __frsp - Floating Round to Single-Precision
 */
static inline float __frsp (double d) __attribute__((always_inline));
static inline float 
__frsp (double d)
{
  float result;
  __asm__ ("frsp %0, %1" 
           /* outputs:  */ : "=f" (result) 
           /* inputs:   */ : "f" (d));
  return result;
}

/*
 * __frsqrte - Floating Reciprocal Square Root Estimate
 *
 * Note: not valid on the PowerPC 601.
 */
static inline double __frsqrte (double val) __attribute__((always_inline));
static inline double 
__frsqrte (double val)
{
  double estimate;

  __asm__ ("frsqrte %0,%1" 
           /* outputs:  */ : "=f" (estimate) 
           /* inputs:   */ : "f" (val));
  return estimate;
}

/*
 * __frsqrtes - Floating Reciprocal Square Root Estimate Single
 */
static inline float __frsqrtes (double f) __attribute__((always_inline));
static inline float 
__frsqrtes (double f)
{
  float result;
  __asm__ ("frsqrte %0, %1" 
           /* outputs:  */ : "=f" (result) 
           /* inputs:   */ : "f" (f));
  return result;
}

/*
 * __fsel - Floating Select
 *
 *   if (test >= 0) return a; else return b;
 *
 * Note: not valid on the PowerPC 601.
 */
static inline double __fsel (double test, double a, double b) __attribute__((always_inline));
static inline double 
__fsel (double test, double a, double b)
{
  double result;
  __asm__ ("fsel %0,%1,%2,%3" 
           /* outputs:  */ : "=f" (result) 
           /* inputs:   */ : "f" (test), "f" (a), "f" (b));
  return result;
}

/*
 * __fsels - Floating Select (Single-Precision variant)
 *
 * An artificial single precision variant of fsel. This produces the 
 * same results as fsel, but is useful because the result is cast as 
 * a float, discouraging the compiler from issuing a frsp instruction 
 * afterward.
 */
static inline float __fsels (double test, double a, double b) __attribute__((always_inline));
static inline float 
__fsels (double test, double a, double b)
{
  float result;
  __asm__ ("fsel %0,%1,%2,%3" 
           /* outputs:  */ : "=f" (result) 
           /* inputs:   */ : "f" (test), "f" (a), "f" (b));
  return result;
}

/*
 * __fsqrt - Floating-Point Square Root (Double-Precision)
 *
 * WARNING: Illegal instruction for PowerPC 603, 604, 750, 7400, 7410, 
 * 7450, and 7455
 */
static inline double __fsqrt (double b) __attribute__((always_inline));
static inline double
__fsqrt(double d)
{
  double result;
  __asm__ ("fsqrt %0, %1" 
           /* outputs:  */ : "=f" (result) 
           /* inputs:   */ : "f" (d));
  return result;
}

/*
 * __fsqrts - Floating-Point Square Root Single-Precision
 *
 * WARNING: Illegal instruction for PowerPC 603, 604, 750, 7400, 7410, 
 * 7450, and 7455
 */
static inline float __fsqrts (float f) __attribute__((always_inline));
static inline float 
__fsqrts (float f)
{
  float result;
  __asm__ ("fsqrts %0, %1" 
           /* outputs:  */ : "=f" (result) 
           /* inputs:   */ : "f" (f));
  return result;
}

/*
 * __mulhw - Multiply High Word
 */
static inline int __mulhw (int a, int b) __attribute__((always_inline));
static inline int 
__mulhw (int a, int b)
{
  int result;
  __asm__ ("mulhw %0, %1, %2" 
           /* outputs:  */ : "=r" (result) 
           /* inputs:   */ : "r" (a), "r"(b));
  return result;
}

/*
 * __mulhwu - Multiply High Word Unsigned
 */
static inline unsigned int __mulhwu (unsigned int a, unsigned int b) __attribute__((always_inline));
static inline unsigned int 
__mulhwu (unsigned int a, unsigned int b)
{
  unsigned int result;
  __asm__ ("mulhwu %0, %1, %2" 
           /* outputs:  */ : "=r" (result) 
           /* inputs:   */ : "r" (a), "r"(b));
  return result;
}

/*
 * __stfiwx - Store Floating-Point as Integer Word Indexed
 *
 *   void x(int, void *, int);
 */
#define __stfiwx(value, base, index)    \
  __asm__ ("stfiwx %0, %1, %2" : /*no result*/  \
           : "f" (value), "b%" (index), "r" (base) : "memory")


/*******************************************************************
 *                     Miscellaneous Functions                     *
 *******************************************************************/

/*
 * __nop - no operation (PowerPC preferred form)
 *
 *   void __nop();
 */
#define __nop()    \
  __asm__ ("ori 0,0,0")

/*
 * __icbi - Instruction Cache Block Invalidate
 *
 *   void __icbi(void *, int);
 */
#define __icbi(base, index)    \
  __asm__ ("icbi %0, %1" : /*no result*/ : "b%" (index), "r" (base) : "memory")

/*
 * __mffs - Move from FPSCR
 */
static inline double __mffs (void) __attribute__((always_inline));
static inline double 
__mffs (void)
{
  double result;
  __asm__ volatile ("mffs %0" 
                    /* outputs:  */ : "=f" (result));
  return result;
}

/*
 * __mfspr - Move from Special Purpose Register
 *
 *   int __mfspr(int);
 */
#define __mfspr(spr)							\
  __extension__ ({ long __ppc_i_mfsprResult;				\
     __asm__ volatile ("mfspr %0, %1" : "=r" (__ppc_i_mfsprResult) : "n" (spr)); \
     /*return*/ __ppc_i_mfsprResult; })

/*
 * __mtfsf - Move to SPSCR Fields
 *
 *   void __mtfsf(int, int);
 */
#define __mtfsf(mask, newValue) \
  __asm__ volatile ("mtfsf %0, %1" : : "n" (mask), "f" (newValue))

/*
 * __mtspr - Move to Special Purpose Register
 *
 *   __mtspr x(int, int);
 */
#define __mtspr(spr, value)     \
  __asm__ volatile ("mtspr %0, %1" : : "n" (spr), "r" (value))

/*
 * __OSReadSwapSInt16
 *
 * lhbrx for signed shorts.  This will do the required sign 
 * extension after load and byteswap.
 */
static inline signed short __OSReadSwapSInt16 (signed short *base, int index) __attribute__((always_inline));
static inline signed short 
__OSReadSwapSInt16 (signed short *base, int index)
{
  signed long result;
  __asm__ volatile ("lhbrx %0, %1, %2" 
		    /* outputs:  */ : "=r" (result) 
		    /* inputs:   */ : "b%" (index), "r" (base) 
		    /* clobbers: */ : "memory");
  return result;
}

/*
 * __OSReadSwapUInt16
 */
static inline unsigned short __OSReadSwapUInt16 (volatile void *base, int inex) __attribute__((always_inline));
static inline unsigned short 
__OSReadSwapUInt16 (volatile void *base, int index)
{
  unsigned long result;
  __asm__ volatile ("lhbrx %0, %1, %2"
		    /* outputs:  */ : "=r" (result)
		    /* inputs:   */ : "b" (index), "r" (base)
		    /* clobbers: */ : "memory");
  return result;
}

/*
 * __astrcmp - assembly strcmp
 */
static inline int astrcmp (const char *in_s1, const char *in_s2) __attribute__((always_inline));
static inline int 
astrcmp (const char *in_s1, const char *in_s2)
{
  int result, temp;
  register const char *s1 = in_s1 - 1;
  register const char *s2 = in_s2 - 1;

  __asm__ ("1:lbzu %0,1(%1)\n"
           "\tcmpwi cr1,%0,0\n"
           "\tlbzu %3,1(%2)\n"
           "\tsubf. %0,%3,%0\n"
           "\tbeq- cr1,2f\n"
           "\tbeq+ 1b\n2:"
            /* outputs: */  : "=&r" (result), "+b" (s1), "+b" (s2), "=r" (temp)
            /* inputs: */   : 
            /* clobbers: */ : "cr0", "cr1", "memory");

  return result;

  /*
   * "=&r" (result)     means: 'result' is written on (the '='), it's any GP
   *                    register (the 'r'), and it must not be the same as
   *                    any of the input registers (the '&').
   * "+b" (s1)          means: 's1' is read from and written to (the '+'),
   *                    and it must be a base GP register (i.e., not R0.)
   * "=r" (temp)        means: 'temp' is any GP reg and it's only written to.
   * 
   * "memory"           in the 'clobbers' section means that gcc will make
   *                    sure that anything that should be in memory IS there
   *                    before calling this routine.
   */
}

#endif  /* (defined(__ppc__) || defined(__ppc64__)) && ! defined(__MWERKS__) */

#endif /* _PPC_INTRINSICS_H_ */