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
|
/*
* Copyright (c) 2011 The Chromium OS Authors. All rights reserved.
* Use of this source code is governed by a BSD-style license that can be
* found in the LICENSE file.
*
* Alternatively, this software may be distributed under the terms of the
* GNU General Public License ("GPL") version 2 as published by the Free
* Software Foundation.
*/
/*
* A, Q = r0 + (r1 << 32)
* B, R = r2 + (r3 << 32)
* A / B = Q ... R
*/
.text
.global __aeabi_uldivmod
.type __aeabi_uldivmod, function
.align 0
A_0 .req r0
A_1 .req r1
B_0 .req r2
B_1 .req r3
C_0 .req r4
C_1 .req r5
D_0 .req r6
D_1 .req r7
Q_0 .req r0
Q_1 .req r1
R_0 .req r2
R_1 .req r3
__aeabi_uldivmod:
stmfd sp!, {r4, r5, r6, r7, lr}
@ Test if B == 0
orrs ip, B_0, B_1 @ Z set -> B == 0
beq L_div_by_0
@ Test if B is power of 2: (B & (B - 1)) == 0
subs C_0, B_0, #1
sbc C_1, B_1, #0
tst C_0, B_0
tsteq B_1, C_1
beq L_pow2
@ Test if A_1 == B_1 == 0
orrs ip, A_1, B_1
beq L_div_32_32
L_div_64_64:
mov C_0, #1
mov C_1, #0
@ D_0 = clz A
teq A_1, #0
clz D_0, A_1
clzeq ip, A_0
addeq D_0, D_0, ip
@ D_1 = clz B
teq B_1, #0
clz D_1, B_1
clzeq ip, B_0
addeq D_1, D_1, ip
@ if clz B - clz A > 0
subs D_0, D_1, D_0
bls L_done_shift
@ B <<= (clz B - clz A)
subs D_1, D_0, #32
rsb ip, D_0, #32
movmi B_1, B_1, lsl D_0
orrmi B_1, B_1, B_0, lsr ip
movpl B_1, B_0, lsl D_1
mov B_0, B_0, lsl D_0
@ C = 1 << (clz B - clz A)
movmi C_1, C_1, lsl D_0
orrmi C_1, C_1, C_0, lsr ip
movpl C_1, C_0, lsl D_1
mov C_0, C_0, lsl D_0
L_done_shift:
mov D_0, #0
mov D_1, #0
@ C: current bit; D: result
L_subtract:
@ if A >= B
cmp A_1, B_1
cmpeq A_0, B_0
bcc L_update
@ A -= B
subs A_0, A_0, B_0
sbc A_1, A_1, B_1
@ D |= C
orr D_0, D_0, C_0
orr D_1, D_1, C_1
L_update:
@ if A == 0: break
orrs ip, A_1, A_0
beq L_exit
@ C >>= 1
movs C_1, C_1, lsr #1
movs C_0, C_0, rrx
@ if C == 0: break
orrs ip, C_1, C_0
beq L_exit
@ B >>= 1
movs B_1, B_1, lsr #1
mov B_0, B_0, rrx
b L_subtract
L_exit:
@ Note: A, B & Q, R are aliases
mov R_0, A_0
mov R_1, A_1
mov Q_0, D_0
mov Q_1, D_1
ldmfd sp!, {r4, r5, r6, r7, pc}
L_div_32_32:
@ Note: A_0 & r0 are aliases
@ Q_1 r1
mov r1, B_0
bl __aeabi_uidivmod
mov R_0, r1
mov R_1, #0
mov Q_1, #0
ldmfd sp!, {r4, r5, r6, r7, pc}
L_pow2:
@ Note: A, B and Q, R are aliases
@ R = A & (B - 1)
and C_0, A_0, C_0
and C_1, A_1, C_1
@ Q = A >> log2(B)
@ Note: B must not be 0 here!
clz D_0, B_0
add D_1, D_0, #1
rsbs D_0, D_0, #31
bpl L_1
clz D_0, B_1
rsb D_0, D_0, #31
mov A_0, A_1, lsr D_0
add D_0, D_0, #32
L_1:
movpl A_0, A_0, lsr D_0
orrpl A_0, A_0, A_1, lsl D_1
mov A_1, A_1, lsr D_0
@ Mov back C to R
mov R_0, C_0
mov R_1, C_1
ldmfd sp!, {r4, r5, r6, r7, pc}
L_div_by_0:
bl __div0
@ As wrong as it could be
mov Q_0, #0
mov Q_1, #0
mov R_0, #0
mov R_1, #0
ldmfd sp!, {r4, r5, r6, r7, pc}
|