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
|
/*
* Written by J.T. Conklin <jtc@netbsd.org>.
* Adapted for use as log2 by Ulrich Drepper <drepper@cygnus.com>.
* Public domain.
*
* Changed to use fyl2xp1 for values near 1, <drepper@cygnus.com>.
* Adapted for x86-64 by Andreas Jaeger <aj@suse.de>.
*/
#include <machine/asm.h>
#include <libm-alias-finite.h>
.section .rodata.cst8,"aM",@progbits,8
.p2align 3
.type one,@object
one: .double 1.0
ASM_SIZE_DIRECTIVE(one)
/* It is not important that this constant is precise. It is only
a value which is known to be on the safe side for using the
fyl2xp1 instruction. */
.type limit,@object
limit: .double 0.29
ASM_SIZE_DIRECTIVE(limit)
#ifdef PIC
# define MO(op) op##(%rip)
#else
# define MO(op) op
#endif
.text
ENTRY(__ieee754_log2l)
fldl MO(one)
fldt 8(%rsp) // x : 1
fxam
fnstsw
fld %st // x : x : 1
testb $1, %ah
jnz 3f // in case x is NaN or ±Inf
4: fsub %st(2), %st // x-1 : x : 1
fld %st // x-1 : x-1 : x : 1
fabs // |x-1| : x-1 : x : 1
fcompl MO(limit) // x-1 : x : 1
fnstsw // x-1 : x : 1
andb $0x45, %ah
jz 2f
fxam
fnstsw
andb $0x45, %ah
cmpb $0x40, %ah
jne 5f
fabs // log2(1) is +0 in all rounding modes.
5: fstp %st(1) // x-1 : 1
fyl2xp1 // log(x)
ret
2: fstp %st(0) // x : 1
fyl2x // log(x)
ret
3: testb $4, %ah
jnz 4b // in case x is ±Inf
fstp %st(1)
fstp %st(1)
fadd %st(0)
ret
END (__ieee754_log2l)
ENTRY(__log2l_finite)
fldl MO(one)
fldt 8(%rsp) // x : 1
fld %st // x : x : 1
fsub %st(2), %st // x-1 : x : 1
fld %st // x-1 : x-1 : x : 1
fabs // |x-1| : x-1 : x : 1
fcompl MO(limit) // x-1 : x : 1
fnstsw // x-1 : x : 1
andb $0x45, %ah
jz 2b
fxam
fnstsw
andb $0x45, %ah
cmpb $0x40, %ah
jne 6f
fabs // log2(1) is +0 in all rounding modes.
6: fstp %st(1) // x-1 : 1
fyl2xp1 // log(x)
ret
END (__log2l_finite)
libm_alias_finite (__log2l_finite, __log2l)
|