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
|
/*
* Copyright (c) 2005-2008, The Android Open Source Project
* Copyright (c) 2010, Code Aurora Forum. All rights reserved.
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
.text
.global S32A_Opaque_BlitRow32_arm
.func S32A_Opaque_BlitRow32_arm
S32A_Opaque_BlitRow32_arm:
push {r4-r11}
#if __ARM_ARCH__ == 7 || defined(__ARM_NEON__)
cmp r2,#24
blt .Lless_than_24
vpush {Q4-Q7}
vmov.i16 q14,#0x100 //Q14.16 = 256
//prefix
vld4.8 {d0, d1, d2, d3}, [r1]! //d0,d1,d2,d3 = sourc rgb(0,1,2,3) A(0,1,2,3)
//update source ptr but not dst ptr
vld4.8 {d4, d5, d6, d7}, [r0] //d4,d5,d6,d7 = dst rgb(0,1,2,3) A(0,1,2,3)
add r3, r0, #32 // minus 16 to pretend the last round
mov r5, #64
sub r2,r2,#8
.Lloop:
pld [r1, #256]
pld [r0, #256]
sub r2,r2,#16
vsubw.u8 q4,q14,d3 //Q4.16 = 256-d3
//It has to be 24 since we pre-load 8 word for the next rounds
cmp r2,#16
vmovl.u8 q6,d4 //Q6 = vmovl.u8 d4
vmovl.u8 q7,d5 //Q7 = vmovl.u8 d5
vmovl.u8 q8,d6 //Q8 = vmovl.u8 d6
vmovl.u8 q9,d7 //Q9 = vmovl.u8 d7
vmul.i16 q6,q6,q4 //Q6 = Q6 * Q4
vmul.i16 q7,q7,q4 //Q7 = Q7 * Q4
vld4.8 {d20, d21, d22, d23}, [r1]! //d0,d1,d2,d3 = sourc rgb(0,1,2,3) A(0,1,2,3)
vmul.i16 q8,q8,q4 //Q8 = Q8 * Q4
vmul.i16 q9,q9,q4 //Q9 = Q9 * Q4
vld4.8 {d24, d25, d26, d27}, [r3] //d4,d5,d6,d7 = dst rgb(0,1,2,3) A(0,1,2,3)
vshrn.i16 d4,q6,#8 //d4 = Q6.16 shrn 8
vshrn.i16 d5,q7,#8 //d5 = Q7.16 shrn 8
vshrn.i16 d6,q8,#8 //d6 = Q8.16 shrn 8
vshrn.i16 d7,q9,#8 //d7 = Q9.16 shrn 8
vadd.i8 d4,d4,d0 //d4 = d4+d0
vadd.i8 d5,d5,d1 //d5 = d5+d1
vadd.i8 d6,d6,d2 //d6 = d6+d2
vadd.i8 d7,d7,d3 //d7 = d7+d3
vst4.8 {d4, d5, d6, d7}, [r0], r5 //dst rgb(0,1,2,3) A(0,1,2,3) = d4,d5,d6,d7
//add r0, r0, r5
//The next 4 words
vsubW.u8 q4,q14,d23 //Q4.16 = 256-d3
vmovl.u8 q6,d24 //Q6 = vmovl.u8 d4
vmovl.u8 q7,d25 //Q7 = vmovl.u8 d5
vmovl.u8 q8,d26 //Q8 = vmovl.u8 d6
vmovl.u8 q9,d27 //Q9 = vmovl.u8 d7
vmul.i16 q6,q6,q4 //Q6 = Q6 * Q4
vmul.i16 q7,q7,q4 //Q7 = Q7 * Q4
vld4.8 {d0, d1, d2, d3}, [r1]! //d0,d1,d2,d3 = sourc rgb(0,1,2,3) A(0,1,2,3)
vmul.i16 q8,q8,q4 //Q8 = Q8 * Q4
vmul.i16 q9,q9,q4 //Q9 = Q9 * Q4
vld4.8 {d4, d5, d6, d7}, [r0] //d4,d5,d6,d7 = dst rgb(0,1,2,3) A(0,1,2,3)
vshrn.i16 d24,q6,#8 //d4 = Q6.16 shrn 8
vshrn.i16 d25,q7,#8 //d5 = Q7.16 shrn 8
vshrn.i16 d26,q8,#8 //d6 = Q8.16 shrn 8
vshrn.i16 d27,q9,#8 //d7 = Q9.16 shrn 8
vadd.i8 d24,d24,d20 //d4 = d4+d0
vadd.i8 d25,d25,d21 //d5 = d5+d1
vadd.i8 d26,d26,d22 //d6 = d6+d2
vadd.i8 d27,d27,d23 //d7 = d7+d3
vst4.8 {d24, d25, d26, d27}, [r3], r5 //dst rgb(0,1,2,3) A(0,1,2,3) = d4,d5,d6,d7
//add r3, r3, r5
bge .Lloop
//There are 8 words left unprocessed from previous round
vsubw.u8 q4,q14,d3 //Q4.16 = 256-d3
cmp r2,#8
vmovl.u8 q6,d4 //Q6 = vmovl.u8 d4
vmovl.u8 q7,d5 //Q7 = vmovl.u8 d5
vmovl.u8 q8,d6 //Q8 = vmovl.u8 d6
vmovl.u8 q9,d7 //Q9 = vmovl.u8 d7
vmul.i16 q6,q6,q4 //Q6 = Q6 * Q4
vmul.i16 q7,q7,q4 //Q7 = Q7 * Q4
vmul.i16 q8,q8,q4 //Q8 = Q8 * Q4
vmul.i16 q9,q9,q4 //Q9 = Q9 * Q4
vshrn.i16 d4,q6,#8 //d4 = Q6.16 shrn 8
vshrn.i16 d5,q7,#8 //d5 = Q7.16 shrn 8
vshrn.i16 d6,q8,#8 //d6 = Q8.16 shrn 8
vshrn.i16 d7,q9,#8 //d7 = Q9.16 shrn 8
vadd.i8 d4,d4,d0 //d4 = d4+d0
vadd.i8 d5,d5,d1 //d5 = d5+d1
vadd.i8 d6,d6,d2 //d6 = d6+d2
vadd.i8 d7,d7,d3 //d7 = d7+d3
vst4.8 {d4, d5, d6, d7}, [r0]! //dst rgb(0,1,2,3) A(0,1,2,3) = d4,d5,d6,d7
.Lless_than_16:
cmp r2,#8
blt .Lless_than_8
sub r2,r2,#8
vld4.8 {d0, d1, d2, d3}, [r1]! //d0,d1,d2,d3 = sourc rgb(0,1,2,3) A(0,1,2,3)
//update source ptr but not dst ptr
vld4.8 {d4, d5, d6, d7}, [r0] //d4,d5,d6,d7 = dst rgb(0,1,2,3) A(0,1,2,3)
vsubw.u8 q4,q14,d3 //Q4.16 = 256-d3
vmovl.u8 q6,d4 //Q6 = vmovl.u8 d4
vmovl.u8 q7,d5 //Q7 = vmovl.u8 d5
vmovl.u8 q8,d6 //Q8 = vmovl.u8 d6
vmovl.u8 q9,d7 //Q9 = vmovl.u8 d7
vmul.i16 q6,q6,q4 //Q6 = Q6 * Q4
vmul.i16 q7,q7,q4 //Q7 = Q7 * Q4
vmul.i16 q8,q8,q4 //Q8 = Q8 * Q4
vmul.i16 q9,q9,q4 //Q9 = Q9 * Q4
vshrn.i16 d4,q6,#8 //d4 = Q6.16 shrn 8
vshrn.i16 d5,q7,#8 //d5 = Q7.16 shrn 8
vshrn.i16 d6,q8,#8 //d6 = Q8.16 shrn 8
vshrn.i16 d7,q9,#8 //d7 = Q9.16 shrn 8
vadd.i8 d4,d4,d0 //d4 = d4+d0
vadd.i8 d5,d5,d1 //d5 = d5+d1
vadd.i8 d6,d6,d2 //d6 = d6+d2
vadd.i8 d7,d7,d3 //d7 = d7+d3
vst4.8 {d4, d5, d6, d7}, [r0]! //dst rgb(0,1,2,3) A(0,1,2,3) = d4,d5,d6,d7
.Lless_than_8:
vpop {Q4-Q7}
.Lless_than_4:
cmp r2, #1
bmi .Lexit
b .Lresidual_loop
.Lless_than_24:
cmp r2,#8
blt .Lless_than_4
.Lloop_8:
sub r2,r2,#8
// We already read the 8 words from the previous pipe line
vld4.8 {d0, d1, d2, d3}, [r1]! //d0,d1,d2,d3 = sourc rgb(0,1,2,3) A(0,1,2,3)
//update source ptr but not dst ptr
vld4.8 {d4, d5, d6, d7}, [r0] //d4,d5,d6,d7 = dst rgb(0,1,2,3) A(0,1,2,3)
vmov.i16 q10,#0x100 //Q4.16 = 256
vsubW.u8 q10,q10,d3 //Q4.16 = 256-d3
cmp r2,#8
vmovl.u8 q12,d4 //Q6 = vmovl.u8 d4
vmovl.u8 q13,d5 //Q7 = vmovl.u8 d5
vmovl.u8 q8,d6 //Q8 = vmovl.u8 d6
vmovl.u8 q9,d7 //Q9 = vmovl.u8 d7
vmul.i16 q12,q12,q10 //Q6 = Q6 * Q4
vmul.i16 q13,q13,q10 //Q7 = Q7 * Q4
vmul.i16 q8,q8,q10 //Q8 = Q8 * Q4
vmul.i16 q9,q9,q10 //Q9 = Q9 * Q4
vshrn.i16 d4,q12,#8 //d4 = Q6.16 shrn 8
vshrn.i16 d5,q13,#8 //d5 = Q7.16 shrn 8
vshrn.i16 d6,q8,#8 //d6 = Q8.16 shrn 8
vshrn.i16 d7,q9,#8 //d7 = Q9.16 shrn 8
vadd.i8 d4,d4,d0 //d4 = d4+d0
vadd.i8 d5,d5,d1 //d5 = d5+d1
vadd.i8 d6,d6,d2 //d6 = d6+d2
vadd.i8 d7,d7,d3 //d7 = d7+d3
vst4.8 {d4, d5, d6, d7}, [r0]! //dst rgb(0,1,2,3) A(0,1,2,3) = d4,d5,d6,d7
bge .Lloop_8
b .Lless_than_4
#endif
/*
* r0 - dst
* r1 - src
* r2 - count
*/
.Lresidual_loop:
mov r10, #0xFF
orr r10, r10, r10, lsl #16 //mask = r10 = 0x00FF00FF
subs r2, r2, #2
blt .Lblitrow32_single_loop
.Lblitrow32_double_loop:
ldm r0, {r3, r4}
ldm r1!, {r5, r6}
orrs r9, r3, r4
beq .Lblitrow32_loop_cond
// First iteration
lsr r7, r5, #24 //extract alpha
and r8, r3, r10 //rb = (dst & mask)
rsb r7, r7, #256 //r5 = scale = (255-alpha)+1
and r9, r10, r3, lsr #8 //ag = (dst>>8) & mask
mul r11, r8, r7 //RB = rb * scale
mul r3, r9, r7 //AG = ag * scale
// combine RB and AG
and r11, r10, r11, lsr #8 //r8 = (RB>>8) & mask
and r3, r3, r10, lsl #8 //r9 = AG & ~mask
lsr r7, r6, #24 //extract alpha for second iteration
orr r3, r3, r11
// Second iteration
and r8, r4, r10 //rb = (dst & mask)
rsb r7, r7, #256 //r5 = scale = (255-alpha)+1
and r9, r10, r4, lsr #8 //ag = (dst>>8) & mask
mul r11, r8, r7 //RB = rb * scale
mul r4, r9, r7 //AG = ag * scale
// combine RB and AG
and r11, r10, r11, lsr #8 //r8 = (RB>>8) & mask
and r4, r4, r10, lsl #8 //r9 = AG & ~mask
orr r4, r4, r11
// add src to combined value
add r5, r5, r3
add r6, r6, r4
.Lblitrow32_loop_cond:
subs r2, r2, #2
stm r0!, {r5, r6}
bge .Lblitrow32_double_loop
.Lblitrow32_single_loop:
adds r2, #1
blo .Lexit
ldr r3, [r0]
ldr r5, [r1], #4
cmp r3, #0
beq .Lblitrow32_single_store
lsr r7, r5, #24 //extract alpha
and r8, r3, r10 //rb = (dst & mask)
rsb r7, r7, #256 //r5 = scale = (255-alpha)+1
and r9, r10, r3, lsr #8 //ag = (dst>>8) & mask
mul r8, r8, r7 //RB = rb * scale
mul r9, r9, r7 //AG = ag * scale
// combine RB and AG
and r8, r10, r8, lsr #8 //r8 = (RB>>8) & mask
and r9, r9, r10, lsl #8 //r9 = AG & ~mask
orr r3, r8, r9
add r5, r5, r3 //add src to combined value
.Lblitrow32_single_store:
str r5, [r0], #4
.Lexit:
pop {r4-r11}
bx lr
|
