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
|
.arch armv7-a
.text
.global blitAntiH_NEON
.type blitAntiH_NEON, %function
blitAntiH_NEON:
.fnstart
@r0 antialias
@r1 device
@r2 runs
@r3 srcExpanded
@r4 scale
@r5 count
@r6 temp
@r7 0x7E0
@r8 ~(0x7E0)
@r9 scale5
@r10 src32
@r11 temp2
@r12 temp3
@r14 count
@d16 = {0x07E0, 0x07E0, 0x07E0, 0x07E0}
@d17 = {0xF81F, 0XF81F, 0XF81F, 0XF81F}
@q12 = d16
@q13 = d17
stmfd sp!, {r4-r12,r14}
mov r14, r13
add r14, #40
ldr r4, [r14]
mov r5, #0x7E
vdup.16 d16, r5
vshl.u16 d16, #4
vmov.u16 d17, d16
vmvn.u16 d18, d17
vmov.u16 d19, d18
vmovl.u16 q12, d16
vmvn.u32 q13, q12
default:
ldrh r5, [r2] @r5 = runs[0]
cmp r5, #0 @if (r5 <=0)
bls end_prog @return
add r2, r5, lsl #1 @ runs += (count*2)
ldrb r6, [r0] @ r6 = antialias[0]
add r0, r5 @ antialias += count
cmp r6, #0 @ if(r6 == 0)
addeq r1, r5, lsl #1 @ device += (count*2)
beq default @ goto default
add r6, #1 @ antialias[0]++
mul r7, r6, r4 @ antialias[0] * scale = r7
lsr r8, r7, #11 @ SkAlpha255To256(aa) * scale >> 11
mul r10, r3, r8 @ r10 = scale5 * srcExpanded
rsb r9, r8, #32 @ r9 = 32 - scale5
vdup.32 q10, r9 @ q10 = scale5
vdup.32 q11, r10 @ q11 = src32
mov r7, #0x7E
lsl r7, #4 @ r7 = 0x7E0
mvn r8, r7 @ r8 = ~(0x7E0)
mov r14, r5, lsr #3
cmp r14, #0
beq start_fourbytes
eightbytes:
vld1.16 {q0}, [r1] @device load
vand.u16 q1, q0, q8 @ c & 0X7E0
vand.u16 q2, q0, q9 @ c &~(0x7E0)
vmovl.u16 q3, d2
vmovl.u16 q4, d4 @ conversion from 16bits to 32bits
vshl.u32 q5, q3, #16 @ q5 = (c & 0x7E0) << 16
vorr.u32 q6, q4, q5 @ q6 = (c & 0x7E0) << 16 | (c & ~(0x7E0))
vmul.u32 q7, q6, q10 @ dst32(q7) = scale5 * SkExpand_rgb_16(*device)
vadd.u32 q14, q7, q11 @ q14 = dst32 + src32
vshr.u32 q15, q14, #5 @ q15 = (src32 + dst32) >> 5
vshr.u32 q5, q15, #16 @ q5 = ( (src32 + dst32) >> 5 ) >> 16
vand.u32 q6, q5, q12 @ q6 = (c >> 16) & (0x7E0)
vand.u32 q7, q15, q13 @ q7 = (c & ~(0x7E0))
vorr.u32 q14, q6, q7 @ q14 = ((c >> 16) & 0x7E0) | (c & ~(0x7E0))
vmovn.u32 d0, q14
vmovl.u16 q3, d3
vmovl.u16 q4, d5 @ conversion from 16bits to 32bits
vshl.u32 q5, q3, #16 @ q5 = (c & 0x7E0) << 16
vorr.u32 q6, q4, q5 @ q6 = (c & 0x7E0) << 16 | (c & ~(0x7E0))
vmul.u32 q7, q6, q10 @ dst32(q7) = scale5 * SkExpand_rgb_16(*device)
vadd.u32 q14, q7, q11 @ q14 = dst32 + src32
vshr.u32 q15, q14, #5 @ q15 = (src32 + dst32) >> 5
vshr.u32 q5, q15, #16 @ q5 = ( (src32 + dst32) >> 5 ) >> 16
vand.u32 q6, q5, q12 @ q6 = (c >> 16) & (0x7E0)
vand.u32 q7, q15, q13 @ q7 = (c & ~(0x7E0))
vorr.u32 q14, q6, q7 @ q14 = ((c >> 16) & 0x7E0) | (c & ~(0x7E0))
vmovn.u32 d1, q14
vst1.16 {q0}, [r1]
add r1, #16 @ device++
subs r14, r14, #1
bhi eightbytes
start_fourbytes:
and r14, r5, #4
cmp r14, #0
beq start_onebyte
fourbytes:
vld1.16 d0, [r1] @ device load
vand.u16 d1, d0, d16 @ c & 0x7E0
vand.u16 d2, d0, d18 @ c & ~(0x7E0)
vmovl.u16 q3, d1
vmovl.u16 q4, d2 @ conversion from 16bits to 32bits
vshl.u32 q5, q3, #16 @ q5 = (c & 0x7E0) << 16
vorr.u32 q6, q4, q5 @ q6 = (c & 0x7E0) << 16 | (c & ~(0x7E0))
vmul.u32 q7, q6, q10 @ dst32(q7) = scale5 * SkExpand_rgb_16(*device)
vadd.u32 q3, q11, q7 @ q3 = dst32 + src32
vshr.u32 q1, q3, #5 @ q1 = (src32 + dst32) >> 5
vshr.u32 q2, q1, #16 @ q2 = ( (src32 + dst32) >> 5 ) >> 16
vand.u32 q3, q2, q12 @ q3 = (c >> 16) & (0x7E0)
vand.u32 q4, q1, q13 @ q4 = (c & ~(0x7E0))
vorr.u32 q5, q4, q3 @ q6 = ((c >> 16) & 0x7E0) | (c & ~(0x7E0))
vmovn.u32 d0, q5
vst1.16 d0, [r1]
add r1, #8
start_onebyte:
and r14, r5, #3
cmp r14, #0
beq default
onebyte:
ldrh r6, [r1] @ device load
and r5, r6, r7 @ r5 = c & 0x7E0
and r11, r6, r8 @ r11 = c & (~0x7E0)
orr r12, r11, r5, lsl #16 @ r12 = (c & (~0x7E0) ) | ((c & 0x7E0) << 16)
mul r6, r9, r12 @dst32(r5) = scale5 * SkExpand_rgb_16(*device)
add r5, r6, r10 @ src32 + dst32
lsr r6, r5, #5 @ (src32 + dst32) >> 5
and r12, r7, r6, lsr #16 @ (c>>16) & 2016
and r11, r8, r6 @ (c & ~(2016))
orr r5, r11, r12 @ (c & ~(2016) | ( (c>>16) & 2016)
strh r5, [r1] @ *device = r5
add r1, #2 @ device++
subs r14, r14, #1
bhi onebyte
b default
end_prog:
ldmfd sp!, {r4-r12, r14}
mov pc, lr
.fnend
|
