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
|
/* ix87 specific implementation of pow function.
Copyright (C) 1996-2014 Free Software Foundation, Inc.
This file is part of the GNU C Library.
Contributed by Ulrich Drepper <drepper@cygnus.com>, 1996.
The GNU C Library is free software; you can redistribute it and/or
modify it under the terms of the GNU Lesser General Public
License as published by the Free Software Foundation; either
version 2.1 of the License, or (at your option) any later version.
The GNU C Library 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
Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public
License along with the GNU C Library; if not, see
<http://www.gnu.org/licenses/>. */
#include <machine/asm.h>
.section .rodata.cst8,"aM",@progbits,8
.p2align 3
.type one,@object
one: .double 1.0
ASM_SIZE_DIRECTIVE(one)
.type limit,@object
limit: .double 0.29
ASM_SIZE_DIRECTIVE(limit)
.type p63,@object
p63: .byte 0, 0, 0, 0, 0, 0, 0xe0, 0x43
ASM_SIZE_DIRECTIVE(p63)
.type p10,@object
p10: .byte 0, 0, 0, 0, 0, 0, 0x90, 0x40
ASM_SIZE_DIRECTIVE(p10)
.section .rodata.cst16,"aM",@progbits,16
.p2align 3
.type infinity,@object
inf_zero:
infinity:
.byte 0, 0, 0, 0, 0, 0, 0xf0, 0x7f
ASM_SIZE_DIRECTIVE(infinity)
.type zero,@object
zero: .double 0.0
ASM_SIZE_DIRECTIVE(zero)
.type minf_mzero,@object
minf_mzero:
minfinity:
.byte 0, 0, 0, 0, 0, 0, 0xf0, 0xff
mzero:
.byte 0, 0, 0, 0, 0, 0, 0, 0x80
ASM_SIZE_DIRECTIVE(minf_mzero)
#ifdef PIC
# define MO(op) op##@GOTOFF(%ecx)
# define MOX(op,x,f) op##@GOTOFF(%ecx,x,f)
#else
# define MO(op) op
# define MOX(op,x,f) op(,x,f)
#endif
.text
ENTRY(__ieee754_pow)
fldl 12(%esp) // y
fxam
#ifdef PIC
LOAD_PIC_REG (cx)
#endif
fnstsw
movb %ah, %dl
andb $0x45, %ah
cmpb $0x40, %ah // is y == 0 ?
je 11f
cmpb $0x05, %ah // is y == ±inf ?
je 12f
cmpb $0x01, %ah // is y == NaN ?
je 30f
fldl 4(%esp) // x : y
subl $8,%esp
cfi_adjust_cfa_offset (8)
fxam
fnstsw
movb %ah, %dh
andb $0x45, %ah
cmpb $0x40, %ah
je 20f // x is ±0
cmpb $0x05, %ah
je 15f // x is ±inf
cmpb $0x01, %ah
je 32f // x is NaN
fxch // y : x
/* fistpll raises invalid exception for |y| >= 1L<<63. */
fld %st // y : y : x
fabs // |y| : y : x
fcompl MO(p63) // y : x
fnstsw
sahf
jnc 2f
/* First see whether `y' is a natural number. In this case we
can use a more precise algorithm. */
fld %st // y : y : x
fistpll (%esp) // y : x
fildll (%esp) // int(y) : y : x
fucomp %st(1) // y : x
fnstsw
sahf
jne 3f
/* OK, we have an integer value for y. If large enough that
errors may propagate out of the 11 bits excess precision, use
the algorithm for real exponent instead. */
fld %st // y : y : x
fabs // |y| : y : x
fcompl MO(p10) // y : x
fnstsw
sahf
jnc 2f
popl %eax
cfi_adjust_cfa_offset (-4)
popl %edx
cfi_adjust_cfa_offset (-4)
orl $0, %edx
fstp %st(0) // x
jns 4f // y >= 0, jump
fdivrl MO(one) // 1/x (now referred to as x)
negl %eax
adcl $0, %edx
negl %edx
4: fldl MO(one) // 1 : x
fxch
6: shrdl $1, %edx, %eax
jnc 5f
fxch
fmul %st(1) // x : ST*x
fxch
5: fmul %st(0), %st // x*x : ST*x
shrl $1, %edx
movl %eax, %ecx
orl %edx, %ecx
jnz 6b
fstp %st(0) // ST*x
ret
/* y is ±NAN */
30: fldl 4(%esp) // x : y
fldl MO(one) // 1.0 : x : y
fucomp %st(1) // x : y
fnstsw
sahf
je 31f
fxch // y : x
31: fstp %st(1)
ret
cfi_adjust_cfa_offset (8)
32: addl $8, %esp
cfi_adjust_cfa_offset (-8)
fstp %st(1)
ret
cfi_adjust_cfa_offset (8)
.align ALIGNARG(4)
2: // y is a large integer (absolute value at least 1L<<10), but
// may be odd unless at least 1L<<64. So it may be necessary
// to adjust the sign of a negative result afterwards.
fxch // x : y
fabs // |x| : y
fxch // y : x
.align ALIGNARG(4)
3: /* y is a real number. */
fxch // x : y
fldl MO(one) // 1.0 : x : y
fldl MO(limit) // 0.29 : 1.0 : x : y
fld %st(2) // x : 0.29 : 1.0 : x : y
fsub %st(2) // x-1 : 0.29 : 1.0 : x : y
fabs // |x-1| : 0.29 : 1.0 : x : y
fucompp // 1.0 : x : y
fnstsw
fxch // x : 1.0 : y
sahf
ja 7f
fsub %st(1) // x-1 : 1.0 : y
fyl2xp1 // log2(x) : y
jmp 8f
7: fyl2x // log2(x) : y
8: fmul %st(1) // y*log2(x) : y
fst %st(1) // y*log2(x) : y*log2(x)
frndint // int(y*log2(x)) : y*log2(x)
fsubr %st, %st(1) // int(y*log2(x)) : fract(y*log2(x))
fxch // fract(y*log2(x)) : int(y*log2(x))
f2xm1 // 2^fract(y*log2(x))-1 : int(y*log2(x))
faddl MO(one) // 2^fract(y*log2(x)) : int(y*log2(x))
fscale // 2^fract(y*log2(x))*2^int(y*log2(x)) : int(y*log2(x))
fstp %st(1) // 2^fract(y*log2(x))*2^int(y*log2(x))
testb $2, %dh
jz 292f
// x is negative. If y is an odd integer, negate the result.
fldl 20(%esp) // y : abs(result)
fld %st // y : y : abs(result)
fabs // |y| : y : abs(result)
fcompl MO(p63) // y : abs(result)
fnstsw
sahf
jnc 291f
// We must find out whether y is an odd integer.
fld %st // y : y : abs(result)
fistpll (%esp) // y : abs(result)
fildll (%esp) // int(y) : y : abs(result)
fucompp // abs(result)
fnstsw
sahf
jne 292f
// OK, the value is an integer, but is it odd?
popl %eax
cfi_adjust_cfa_offset (-4)
popl %edx
cfi_adjust_cfa_offset (-4)
andb $1, %al
jz 290f // jump if not odd
// It's an odd integer.
fchs
290: ret
cfi_adjust_cfa_offset (8)
291: fstp %st(0) // abs(result)
292: addl $8, %esp
cfi_adjust_cfa_offset (-8)
ret
// pow(x,±0) = 1
.align ALIGNARG(4)
11: fstp %st(0) // pop y
fldl MO(one)
ret
// y == ±inf
.align ALIGNARG(4)
12: fstp %st(0) // pop y
fldl MO(one) // 1
fldl 4(%esp) // x : 1
fabs // abs(x) : 1
fucompp // < 1, == 1, or > 1
fnstsw
andb $0x45, %ah
cmpb $0x45, %ah
je 13f // jump if x is NaN
cmpb $0x40, %ah
je 14f // jump if |x| == 1
shlb $1, %ah
xorb %ah, %dl
andl $2, %edx
fldl MOX(inf_zero, %edx, 4)
ret
.align ALIGNARG(4)
14: fldl MO(one)
ret
.align ALIGNARG(4)
13: fldl 4(%esp) // load x == NaN
ret
cfi_adjust_cfa_offset (8)
.align ALIGNARG(4)
// x is ±inf
15: fstp %st(0) // y
testb $2, %dh
jz 16f // jump if x == +inf
// fistpll raises invalid exception for |y| >= 1L<<63, so test
// that (in which case y is certainly even) before testing
// whether y is odd.
fld %st // y : y
fabs // |y| : y
fcompl MO(p63) // y
fnstsw
sahf
jnc 16f
// We must find out whether y is an odd integer.
fld %st // y : y
fistpll (%esp) // y
fildll (%esp) // int(y) : y
fucompp // <empty>
fnstsw
sahf
jne 17f
// OK, the value is an integer.
popl %eax
cfi_adjust_cfa_offset (-4)
popl %edx
cfi_adjust_cfa_offset (-4)
andb $1, %al
jz 18f // jump if not odd
// It's an odd integer.
shrl $31, %edx
fldl MOX(minf_mzero, %edx, 8)
ret
cfi_adjust_cfa_offset (8)
.align ALIGNARG(4)
16: fcompl MO(zero)
addl $8, %esp
cfi_adjust_cfa_offset (-8)
fnstsw
shrl $5, %eax
andl $8, %eax
fldl MOX(inf_zero, %eax, 1)
ret
cfi_adjust_cfa_offset (8)
.align ALIGNARG(4)
17: shll $30, %edx // sign bit for y in right position
addl $8, %esp
cfi_adjust_cfa_offset (-8)
18: shrl $31, %edx
fldl MOX(inf_zero, %edx, 8)
ret
cfi_adjust_cfa_offset (8)
.align ALIGNARG(4)
// x is ±0
20: fstp %st(0) // y
testb $2, %dl
jz 21f // y > 0
// x is ±0 and y is < 0. We must find out whether y is an odd integer.
testb $2, %dh
jz 25f
// fistpll raises invalid exception for |y| >= 1L<<63, so test
// that (in which case y is certainly even) before testing
// whether y is odd.
fld %st // y : y
fabs // |y| : y
fcompl MO(p63) // y
fnstsw
sahf
jnc 25f
fld %st // y : y
fistpll (%esp) // y
fildll (%esp) // int(y) : y
fucompp // <empty>
fnstsw
sahf
jne 26f
// OK, the value is an integer.
popl %eax
cfi_adjust_cfa_offset (-4)
popl %edx
cfi_adjust_cfa_offset (-4)
andb $1, %al
jz 27f // jump if not odd
// It's an odd integer.
// Raise divide-by-zero exception and get minus infinity value.
fldl MO(one)
fdivl MO(zero)
fchs
ret
cfi_adjust_cfa_offset (8)
25: fstp %st(0)
26: addl $8, %esp
cfi_adjust_cfa_offset (-8)
27: // Raise divide-by-zero exception and get infinity value.
fldl MO(one)
fdivl MO(zero)
ret
cfi_adjust_cfa_offset (8)
.align ALIGNARG(4)
// x is ±0 and y is > 0. We must find out whether y is an odd integer.
21: testb $2, %dh
jz 22f
// fistpll raises invalid exception for |y| >= 1L<<63, so test
// that (in which case y is certainly even) before testing
// whether y is odd.
fcoml MO(p63) // y
fnstsw
sahf
jnc 22f
fld %st // y : y
fistpll (%esp) // y
fildll (%esp) // int(y) : y
fucompp // <empty>
fnstsw
sahf
jne 23f
// OK, the value is an integer.
popl %eax
cfi_adjust_cfa_offset (-4)
popl %edx
cfi_adjust_cfa_offset (-4)
andb $1, %al
jz 24f // jump if not odd
// It's an odd integer.
fldl MO(mzero)
ret
cfi_adjust_cfa_offset (8)
22: fstp %st(0)
23: addl $8, %esp // Don't use 2 x pop
cfi_adjust_cfa_offset (-8)
24: fldl MO(zero)
ret
END(__ieee754_pow)
strong_alias (__ieee754_pow, __pow_finite)
|