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
|
/* Copyright (C) 1996, 1997 Free Software Foundation, Inc.
This file is part of the GNU C Library.
Contributed by Richard Henderson <rth@tamu.edu>.
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, write to the Free
Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
02111-1307 USA. */
#include <sysdep.h>
#ifdef __linux__
# include <asm/gentrap.h>
# include <asm/pal.h>
#else
# include <machine/pal.h>
#endif
.set noat
.align 4
.globl div
.ent div
div:
.frame sp, 0, ra
#ifdef PROF
ldgp gp, 0(pv)
lda AT, _mcount
jsr AT, (AT), _mcount
.prologue 1
#else
.prologue 0
#endif
#define divisor t1
#define mask t2
#define quotient t3
#define modulus t4
#define tmp1 t5
#define tmp2 t6
#define compare t7
/* find correct sign for input to unsigned divide loop. */
negl a1, modulus # e0 :
negl a2, divisor # .. e1 :
sextl a1, a1 # e0 :
sextl a2, a2 # .. e1 :
mov zero, quotient # e0 :
mov 1, mask # .. e1 :
cmovge a1, a1, modulus # e0 :
cmovge a2, a2, divisor # .. e1 :
beq a2, $divbyzero # e1 :
unop # :
/* shift divisor left, using 3-bit shifts for 32-bit divides as we
can't overflow. Three-bit shifts will result in looping three
times less here, but can result in two loops more later. Thus
using a large shift isn't worth it (and s8addq pairs better than
a shift). */
1: cmpult divisor, modulus, compare # e0 :
s8addq divisor, zero, divisor # .. e1 :
s8addq mask, zero, mask # e0 :
bne compare, 1b # .. e1 :
/* start to go right again. */
2: addq quotient, mask, tmp2 # e1 :
srl mask, 1, mask # .. e0 :
cmpule divisor, modulus, compare # e0 :
subq modulus, divisor, tmp1 # .. e1 :
cmovne compare, tmp2, quotient # e1 :
srl divisor, 1, divisor # .. e0 :
cmovne compare, tmp1, modulus # e0 :
bne mask, 2b # .. e1 :
/* find correct sign for result. */
xor a1, a2, compare # e0 :
negl quotient, tmp1 # .. e1 :
negl modulus, tmp2 # e0 :
cmovlt compare, tmp1, quotient # .. e1 :
cmovlt a1, tmp2, modulus # e1 :
/* and store it away in the structure. */
stl quotient, 0(a0) # .. e0 :
mov a0, v0 # e1 :
stl modulus, 4(a0) # .. e0 :
ret # e1 :
$divbyzero:
mov a0, v0
ldiq a0, GEN_INTDIV
call_pal PAL_gentrap
/* if trap returns, return zero. */
stl zero, 0(v0)
stl zero, 4(v0)
ret
.end div
|
