|
2
|
1 |
/*
|
|
|
2 |
* Copyright 2003 Sun Microsystems, Inc. All Rights Reserved.
|
|
|
3 |
* DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
|
|
|
4 |
*
|
|
|
5 |
* This code is free software; you can redistribute it and/or modify it
|
|
|
6 |
* under the terms of the GNU General Public License version 2 only, as
|
|
|
7 |
* published by the Free Software Foundation. Sun designates this
|
|
|
8 |
* particular file as subject to the "Classpath" exception as provided
|
|
|
9 |
* by Sun in the LICENSE file that accompanied this code.
|
|
|
10 |
*
|
|
|
11 |
* This code is distributed in the hope that it will be useful, but WITHOUT
|
|
|
12 |
* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
|
|
|
13 |
* FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
|
|
|
14 |
* version 2 for more details (a copy is included in the LICENSE file that
|
|
|
15 |
* accompanied this code).
|
|
|
16 |
*
|
|
|
17 |
* You should have received a copy of the GNU General Public License version
|
|
|
18 |
* 2 along with this work; if not, write to the Free Software Foundation,
|
|
|
19 |
* Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
|
|
|
20 |
*
|
|
|
21 |
* Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa Clara,
|
|
|
22 |
* CA 95054 USA or visit www.sun.com if you need additional information or
|
|
|
23 |
* have any questions.
|
|
|
24 |
*/
|
|
|
25 |
|
|
|
26 |
|
|
|
27 |
|
|
|
28 |
/*
|
|
|
29 |
* The functions step along the lines from xLeft to xRight and apply
|
|
|
30 |
* the bilinear filtering.
|
|
|
31 |
*
|
|
|
32 |
*/
|
|
|
33 |
|
|
|
34 |
#include "vis_proto.h"
|
|
|
35 |
#include "mlib_image.h"
|
|
|
36 |
#include "mlib_ImageColormap.h"
|
|
|
37 |
#include "mlib_ImageCopy.h"
|
|
|
38 |
#include "mlib_ImageAffine.h"
|
|
|
39 |
#include "mlib_v_ImageFilters.h"
|
|
|
40 |
#include "mlib_v_ImageChannelExtract.h"
|
|
|
41 |
#include "mlib_v_ImageAffine_BL_S16.h"
|
|
|
42 |
|
|
|
43 |
/*#define MLIB_VIS2*/
|
|
|
44 |
|
|
|
45 |
/***************************************************************/
|
|
|
46 |
#define DTYPE mlib_s16
|
|
|
47 |
|
|
|
48 |
#define FUN_NAME(CHAN) mlib_ImageAffine_u16_##CHAN##_bl
|
|
|
49 |
|
|
|
50 |
/***************************************************************/
|
|
|
51 |
mlib_status FUN_NAME(2ch_na)(mlib_affine_param *param);
|
|
|
52 |
mlib_status FUN_NAME(4ch_na)(mlib_affine_param *param);
|
|
|
53 |
|
|
|
54 |
/***************************************************************/
|
|
|
55 |
#define XOR_8000(x) x = vis_fxor(x, mask_8000)
|
|
|
56 |
|
|
|
57 |
/***************************************************************/
|
|
|
58 |
#ifdef MLIB_VIS2
|
|
|
59 |
#define MLIB_WRITE_BMASK(bmask) vis_write_bmask(bmask, 0)
|
|
|
60 |
#else
|
|
|
61 |
#define MLIB_WRITE_BMASK(bmask)
|
|
|
62 |
#endif /* MLIB_VIS2 */
|
|
|
63 |
|
|
|
64 |
/***************************************************************/
|
|
|
65 |
#undef DECLAREVAR
|
|
|
66 |
#define DECLAREVAR() \
|
|
|
67 |
DECLAREVAR0(); \
|
|
|
68 |
mlib_s32 *warp_tbl = param -> warp_tbl; \
|
|
|
69 |
mlib_s32 srcYStride = param -> srcYStride; \
|
|
|
70 |
mlib_u8 *dl; \
|
|
|
71 |
mlib_s32 i, size; \
|
|
|
72 |
mlib_d64 mask_8000 = vis_to_double_dup(0x80008000); \
|
|
|
73 |
mlib_d64 mask_7fff = vis_to_double_dup(0x7FFF7FFF); \
|
|
|
74 |
mlib_d64 dx64, dy64, deltax, deltay, delta1_x, delta1_y; \
|
|
|
75 |
mlib_d64 s0, s1, s2, s3; \
|
|
|
76 |
mlib_d64 d0, d1, d2, d3, dd
|
|
|
77 |
|
|
|
78 |
/***************************************************************/
|
|
|
79 |
|
|
|
80 |
/* arguments (x, y) are swapped to prevent overflow */
|
|
|
81 |
#define FMUL_16x16(x, y) \
|
|
|
82 |
vis_fpadd16(vis_fmul8sux16(y, x), \
|
|
|
83 |
vis_fmul8ulx16(y, x))
|
|
|
84 |
|
|
|
85 |
/***************************************************************/
|
|
|
86 |
#define BUF_SIZE 512
|
|
|
87 |
|
|
|
88 |
/***************************************************************/
|
|
|
89 |
#define DOUBLE_4U16(x0, x1, x2, x3) \
|
|
|
90 |
vis_to_double(((((x0) & 0xFFFE) << 15) | (((x1) & 0xFFFE) >> 1)), \
|
|
|
91 |
((((x2) & 0xFFFE) << 15) | (((x3) & 0xFFFE) >> 1)))
|
|
|
92 |
|
|
|
93 |
/***************************************************************/
|
|
|
94 |
#define BL_SUM() \
|
|
|
95 |
XOR_8000(s0); \
|
|
|
96 |
XOR_8000(s1); \
|
|
|
97 |
XOR_8000(s2); \
|
|
|
98 |
XOR_8000(s3); \
|
|
|
99 |
\
|
|
|
100 |
delta1_x = vis_fpsub16(mask_7fff, deltax); \
|
|
|
101 |
delta1_y = vis_fpsub16(mask_7fff, deltay); \
|
|
|
102 |
\
|
|
|
103 |
d0 = FMUL_16x16(s0, delta1_x); \
|
|
|
104 |
d1 = FMUL_16x16(s1, deltax); \
|
|
|
105 |
d0 = vis_fpadd16(d0, d1); \
|
|
|
106 |
d0 = vis_fpadd16(d0, d0); \
|
|
|
107 |
d0 = FMUL_16x16(d0, delta1_y); \
|
|
|
108 |
\
|
|
|
109 |
d2 = FMUL_16x16(s2, delta1_x); \
|
|
|
110 |
d3 = FMUL_16x16(s3, deltax); \
|
|
|
111 |
d2 = vis_fpadd16(d2, d3); \
|
|
|
112 |
d2 = vis_fpadd16(d2, d2); \
|
|
|
113 |
d2 = FMUL_16x16(d2, deltay); \
|
|
|
114 |
\
|
|
|
115 |
dd = vis_fpadd16(d0, d2); \
|
|
|
116 |
dd = vis_fpadd16(dd, dd); \
|
|
|
117 |
XOR_8000(dd); \
|
|
|
118 |
\
|
|
|
119 |
deltax = vis_fpadd16(deltax, dx64); \
|
|
|
120 |
deltay = vis_fpadd16(deltay, dy64); \
|
|
|
121 |
deltax = vis_fand(deltax, mask_7fff); \
|
|
|
122 |
deltay = vis_fand(deltay, mask_7fff)
|
|
|
123 |
|
|
|
124 |
/***************************************************************/
|
|
|
125 |
#define BL_SUM_3CH() \
|
|
|
126 |
XOR_8000(s0); \
|
|
|
127 |
XOR_8000(s1); \
|
|
|
128 |
XOR_8000(s2); \
|
|
|
129 |
XOR_8000(s3); \
|
|
|
130 |
\
|
|
|
131 |
delta1_x = vis_fpsub16(mask_7fff, deltax); \
|
|
|
132 |
delta1_y = vis_fpsub16(mask_7fff, deltay); \
|
|
|
133 |
\
|
|
|
134 |
d0 = FMUL_16x16(s0, delta1_y); \
|
|
|
135 |
d2 = FMUL_16x16(s2, deltay); \
|
|
|
136 |
d0 = vis_fpadd16(d0, d2); \
|
|
|
137 |
d0 = vis_fpadd16(d0, d0); \
|
|
|
138 |
d0 = FMUL_16x16(d0, delta1_x); \
|
|
|
139 |
\
|
|
|
140 |
d1 = FMUL_16x16(s1, delta1_y); \
|
|
|
141 |
d3 = FMUL_16x16(s3, deltay); \
|
|
|
142 |
d1 = vis_fpadd16(d1, d3); \
|
|
|
143 |
d1 = vis_fpadd16(d1, d1); \
|
|
|
144 |
d1 = FMUL_16x16(d1, deltax); \
|
|
|
145 |
\
|
|
|
146 |
vis_alignaddr((void*)0, 2); \
|
|
|
147 |
d0 = vis_faligndata(d0, d0); \
|
|
|
148 |
dd = vis_fpadd16(d0, d1); \
|
|
|
149 |
dd = vis_fpadd16(dd, dd); \
|
|
|
150 |
XOR_8000(dd); \
|
|
|
151 |
\
|
|
|
152 |
deltax = vis_fpadd16(deltax, dx64); \
|
|
|
153 |
deltay = vis_fpadd16(deltay, dy64); \
|
|
|
154 |
deltax = vis_fand(deltax, mask_7fff); \
|
|
|
155 |
deltay = vis_fand(deltay, mask_7fff)
|
|
|
156 |
|
|
|
157 |
/***************************************************************/
|
|
|
158 |
#define LD_U16(sp, ind) vis_ld_u16(sp + ind)
|
|
|
159 |
|
|
|
160 |
/***************************************************************/
|
|
|
161 |
#ifndef MLIB_VIS2
|
|
|
162 |
|
|
|
163 |
#define LOAD_1CH() \
|
|
|
164 |
s0 = vis_faligndata(LD_U16(sp3, 0), mask_7fff); \
|
|
|
165 |
s1 = vis_faligndata(LD_U16(sp3, 2), mask_7fff); \
|
|
|
166 |
s2 = vis_faligndata(LD_U16(sp3, srcYStride), mask_7fff); \
|
|
|
167 |
s3 = vis_faligndata(LD_U16(sp3, srcYStride + 2), mask_7fff); \
|
|
|
168 |
\
|
|
|
169 |
s0 = vis_faligndata(LD_U16(sp2, 0), s0); \
|
|
|
170 |
s1 = vis_faligndata(LD_U16(sp2, 2), s1); \
|
|
|
171 |
s2 = vis_faligndata(LD_U16(sp2, srcYStride), s2); \
|
|
|
172 |
s3 = vis_faligndata(LD_U16(sp2, srcYStride + 2), s3); \
|
|
|
173 |
\
|
|
|
174 |
s0 = vis_faligndata(LD_U16(sp1, 0), s0); \
|
|
|
175 |
s1 = vis_faligndata(LD_U16(sp1, 2), s1); \
|
|
|
176 |
s2 = vis_faligndata(LD_U16(sp1, srcYStride), s2); \
|
|
|
177 |
s3 = vis_faligndata(LD_U16(sp1, srcYStride + 2), s3); \
|
|
|
178 |
\
|
|
|
179 |
s0 = vis_faligndata(LD_U16(sp0, 0), s0); \
|
|
|
180 |
s1 = vis_faligndata(LD_U16(sp0, 2), s1); \
|
|
|
181 |
s2 = vis_faligndata(LD_U16(sp0, srcYStride), s2); \
|
|
|
182 |
s3 = vis_faligndata(LD_U16(sp0, srcYStride + 2), s3)
|
|
|
183 |
|
|
|
184 |
#else
|
|
|
185 |
|
|
|
186 |
#define LOAD_1CH() \
|
|
|
187 |
s0 = vis_bshuffle(LD_U16(sp0, 0), LD_U16(sp2, 0)); \
|
|
|
188 |
s1 = vis_bshuffle(LD_U16(sp0, 2), LD_U16(sp2, 2)); \
|
|
|
189 |
s2 = vis_bshuffle(LD_U16(sp0, srcYStride), LD_U16(sp2, srcYStride)); \
|
|
|
190 |
s3 = vis_bshuffle(LD_U16(sp0, srcYStride + 2), LD_U16(sp2, srcYStride + 2)); \
|
|
|
191 |
\
|
|
|
192 |
t0 = vis_bshuffle(LD_U16(sp1, 0), LD_U16(sp3, 0)); \
|
|
|
193 |
t1 = vis_bshuffle(LD_U16(sp1, 2), LD_U16(sp3, 2)); \
|
|
|
194 |
t2 = vis_bshuffle(LD_U16(sp1, srcYStride), LD_U16(sp3, srcYStride)); \
|
|
|
195 |
t3 = vis_bshuffle(LD_U16(sp1, srcYStride + 2), LD_U16(sp3, srcYStride + 2)); \
|
|
|
196 |
\
|
|
|
197 |
s0 = vis_bshuffle(s0, t0); \
|
|
|
198 |
s1 = vis_bshuffle(s1, t1); \
|
|
|
199 |
s2 = vis_bshuffle(s2, t2); \
|
|
|
200 |
s3 = vis_bshuffle(s3, t3)
|
|
|
201 |
|
|
|
202 |
#endif /* MLIB_VIS2 */
|
|
|
203 |
|
|
|
204 |
/***************************************************************/
|
|
|
205 |
#define GET_POINTER(sp) \
|
|
|
206 |
sp = *(mlib_u8**)((mlib_u8*)lineAddr + PTR_SHIFT(Y)) + 2*(X >> MLIB_SHIFT); \
|
|
|
207 |
X += dX; \
|
|
|
208 |
Y += dY
|
|
|
209 |
|
|
|
210 |
/***************************************************************/
|
|
|
211 |
#undef PREPARE_DELTAS
|
|
|
212 |
#define PREPARE_DELTAS \
|
|
|
213 |
if (warp_tbl != NULL) { \
|
|
|
214 |
dX = warp_tbl[2*j ]; \
|
|
|
215 |
dY = warp_tbl[2*j + 1]; \
|
|
|
216 |
dx64 = vis_to_double_dup((((dX << 1) & 0xFFFF) << 16) | ((dX << 1) & 0xFFFF)); \
|
|
|
217 |
dy64 = vis_to_double_dup((((dY << 1) & 0xFFFF) << 16) | ((dY << 1) & 0xFFFF)); \
|
|
|
218 |
}
|
|
|
219 |
|
|
|
220 |
/***************************************************************/
|
|
|
221 |
mlib_status FUN_NAME(1ch)(mlib_affine_param *param)
|
|
|
222 |
{
|
|
|
223 |
DECLAREVAR();
|
|
|
224 |
mlib_s32 off;
|
|
|
225 |
mlib_s32 x0, x1, x2, x3, y0, y1, y2, y3;
|
|
|
226 |
#ifdef MLIB_VIS2
|
|
|
227 |
mlib_d64 t0, t1, t2, t3;
|
|
|
228 |
vis_write_bmask(0x45CD67EF, 0);
|
|
|
229 |
#else
|
|
|
230 |
vis_alignaddr((void*)0, 6);
|
|
|
231 |
#endif /* MLIB_VIS2 */
|
|
|
232 |
|
|
|
233 |
dx64 = vis_to_double_dup((((dX << 1) & 0xFFFF) << 16) | ((dX << 1) & 0xFFFF));
|
|
|
234 |
dy64 = vis_to_double_dup((((dY << 1) & 0xFFFF) << 16) | ((dY << 1) & 0xFFFF));
|
|
|
235 |
|
|
|
236 |
for (j = yStart; j <= yFinish; j++) {
|
|
|
237 |
mlib_u8 *sp0, *sp1, *sp2, *sp3;
|
|
|
238 |
mlib_d64 *dp, dmask;
|
|
|
239 |
|
|
|
240 |
NEW_LINE(1);
|
|
|
241 |
|
|
|
242 |
off = (mlib_s32)dl & 7;
|
|
|
243 |
dp = (mlib_d64*)(dl - off);
|
|
|
244 |
off >>= 1;
|
|
|
245 |
|
|
|
246 |
x0 = X - off*dX; y0 = Y - off*dY;
|
|
|
247 |
x1 = x0 + dX; y1 = y0 + dY;
|
|
|
248 |
x2 = x1 + dX; y2 = y1 + dY;
|
|
|
249 |
x3 = x2 + dX; y3 = y2 + dY;
|
|
|
250 |
|
|
|
251 |
deltax = DOUBLE_4U16(x0, x1, x2, x3);
|
|
|
252 |
deltay = DOUBLE_4U16(y0, y1, y2, y3);
|
|
|
253 |
|
|
|
254 |
if (off) {
|
|
|
255 |
mlib_s32 emask = vis_edge16((void*)(2*off), (void*)(2*(off + size - 1)));
|
|
|
256 |
|
|
|
257 |
off = 4 - off;
|
|
|
258 |
GET_POINTER(sp3);
|
|
|
259 |
sp0 = sp1 = sp2 = sp3;
|
|
|
260 |
|
|
|
261 |
if (off > 1 && size > 1) {
|
|
|
262 |
GET_POINTER(sp3);
|
|
|
263 |
}
|
|
|
264 |
|
|
|
265 |
if (off > 2) {
|
|
|
266 |
sp2 = sp3;
|
|
|
267 |
|
|
|
268 |
if (size > 2) {
|
|
|
269 |
GET_POINTER(sp3);
|
|
|
270 |
}
|
|
|
271 |
}
|
|
|
272 |
|
|
|
273 |
LOAD_1CH();
|
|
|
274 |
BL_SUM();
|
|
|
275 |
|
|
|
276 |
dmask = ((mlib_d64*)mlib_dmask_arr)[emask];
|
|
|
277 |
*dp++ = vis_for (vis_fand(dmask, dd), vis_fandnot(dmask, dp[0]));
|
|
|
278 |
|
|
|
279 |
size -= off;
|
|
|
280 |
|
|
|
281 |
if (size < 0) size = 0;
|
|
|
282 |
}
|
|
|
283 |
|
|
|
284 |
#pragma pipeloop(0)
|
|
|
285 |
for (i = 0; i < size/4; i++) {
|
|
|
286 |
GET_POINTER(sp0);
|
|
|
287 |
GET_POINTER(sp1);
|
|
|
288 |
GET_POINTER(sp2);
|
|
|
289 |
GET_POINTER(sp3);
|
|
|
290 |
|
|
|
291 |
LOAD_1CH();
|
|
|
292 |
BL_SUM();
|
|
|
293 |
|
|
|
294 |
dp[i] = dd;
|
|
|
295 |
}
|
|
|
296 |
|
|
|
297 |
off = size & 3;
|
|
|
298 |
|
|
|
299 |
if (off) {
|
|
|
300 |
GET_POINTER(sp0);
|
|
|
301 |
sp1 = sp2 = sp3 = sp0;
|
|
|
302 |
|
|
|
303 |
if (off > 1) {
|
|
|
304 |
GET_POINTER(sp1);
|
|
|
305 |
}
|
|
|
306 |
|
|
|
307 |
if (off > 2) {
|
|
|
308 |
GET_POINTER(sp2);
|
|
|
309 |
}
|
|
|
310 |
|
|
|
311 |
LOAD_1CH();
|
|
|
312 |
BL_SUM();
|
|
|
313 |
|
|
|
314 |
dmask = ((mlib_d64*)mlib_dmask_arr)[(0xF0 >> off) & 0x0F];
|
|
|
315 |
dp[i] = vis_for (vis_fand(dmask, dd), vis_fandnot(dmask, dp[i]));
|
|
|
316 |
}
|
|
|
317 |
}
|
|
|
318 |
|
|
|
319 |
return MLIB_SUCCESS;
|
|
|
320 |
}
|
|
|
321 |
|
|
|
322 |
/***************************************************************/
|
|
|
323 |
#undef GET_POINTER
|
|
|
324 |
#define GET_POINTER(sp) \
|
|
|
325 |
sp = *(mlib_f32**)((mlib_u8*)lineAddr + PTR_SHIFT(Y)) + (X >> MLIB_SHIFT); \
|
|
|
326 |
X += dX; \
|
|
|
327 |
Y += dY
|
|
|
328 |
|
|
|
329 |
/***************************************************************/
|
|
|
330 |
#define LOAD_2CH() \
|
|
|
331 |
s0 = vis_freg_pair(sp0[0], sp1[0]); \
|
|
|
332 |
s1 = vis_freg_pair(sp0[1], sp1[1]); \
|
|
|
333 |
s2 = vis_freg_pair(sp0[srcYStride], sp1[srcYStride]); \
|
|
|
334 |
s3 = vis_freg_pair(sp0[srcYStride + 1], sp1[srcYStride + 1])
|
|
|
335 |
|
|
|
336 |
/***************************************************************/
|
|
|
337 |
#undef PREPARE_DELTAS
|
|
|
338 |
#define PREPARE_DELTAS \
|
|
|
339 |
if (warp_tbl != NULL) { \
|
|
|
340 |
dX = warp_tbl[2*j ]; \
|
|
|
341 |
dY = warp_tbl[2*j + 1]; \
|
|
|
342 |
dx64 = vis_to_double_dup(((dX & 0xFFFF) << 16) | (dX & 0xFFFF)); \
|
|
|
343 |
dy64 = vis_to_double_dup(((dY & 0xFFFF) << 16) | (dY & 0xFFFF)); \
|
|
|
344 |
}
|
|
|
345 |
|
|
|
346 |
/***************************************************************/
|
|
|
347 |
mlib_status FUN_NAME(2ch)(mlib_affine_param *param)
|
|
|
348 |
{
|
|
|
349 |
DECLAREVAR();
|
|
|
350 |
mlib_s32 off;
|
|
|
351 |
mlib_s32 x0, x1, y0, y1;
|
|
|
352 |
|
|
|
353 |
if (((mlib_s32)lineAddr[0] | (mlib_s32)dstData | srcYStride | dstYStride) & 3) {
|
|
|
354 |
return FUN_NAME(2ch_na)(param);
|
|
|
355 |
}
|
|
|
356 |
|
|
|
357 |
srcYStride >>= 2;
|
|
|
358 |
|
|
|
359 |
dx64 = vis_to_double_dup(((dX & 0xFFFF) << 16) | (dX & 0xFFFF));
|
|
|
360 |
dy64 = vis_to_double_dup(((dY & 0xFFFF) << 16) | (dY & 0xFFFF));
|
|
|
361 |
|
|
|
362 |
for (j = yStart; j <= yFinish; j++) {
|
|
|
363 |
mlib_f32 *sp0, *sp1;
|
|
|
364 |
mlib_d64 *dp;
|
|
|
365 |
|
|
|
366 |
NEW_LINE(2);
|
|
|
367 |
|
|
|
368 |
off = (mlib_s32)dl & 7;
|
|
|
369 |
dp = (mlib_d64*)(dl - off);
|
|
|
370 |
|
|
|
371 |
if (off) {
|
|
|
372 |
x0 = X - dX; y0 = Y - dY;
|
|
|
373 |
x1 = X; y1 = Y;
|
|
|
374 |
} else {
|
|
|
375 |
x0 = X; y0 = Y;
|
|
|
376 |
x1 = X + dX; y1 = Y + dY;
|
|
|
377 |
}
|
|
|
378 |
|
|
|
379 |
deltax = DOUBLE_4U16(x0, x0, x1, x1);
|
|
|
380 |
deltay = DOUBLE_4U16(y0, y0, y1, y1);
|
|
|
381 |
|
|
|
382 |
if (off) {
|
|
|
383 |
GET_POINTER(sp1);
|
|
|
384 |
sp0 = sp1;
|
|
|
385 |
LOAD_2CH();
|
|
|
386 |
|
|
|
387 |
BL_SUM();
|
|
|
388 |
|
|
|
389 |
((mlib_f32*)dp)[1] = vis_read_lo(dd);
|
|
|
390 |
dp++;
|
|
|
391 |
size--;
|
|
|
392 |
}
|
|
|
393 |
|
|
|
394 |
#pragma pipeloop(0)
|
|
|
395 |
for (i = 0; i < size/2; i++) {
|
|
|
396 |
GET_POINTER(sp0);
|
|
|
397 |
GET_POINTER(sp1);
|
|
|
398 |
LOAD_2CH();
|
|
|
399 |
|
|
|
400 |
BL_SUM();
|
|
|
401 |
|
|
|
402 |
*dp++ = dd;
|
|
|
403 |
}
|
|
|
404 |
|
|
|
405 |
if (size & 1) {
|
|
|
406 |
GET_POINTER(sp0);
|
|
|
407 |
sp1 = sp0;
|
|
|
408 |
LOAD_2CH();
|
|
|
409 |
|
|
|
410 |
BL_SUM();
|
|
|
411 |
|
|
|
412 |
((mlib_f32*)dp)[0] = vis_read_hi(dd);
|
|
|
413 |
}
|
|
|
414 |
}
|
|
|
415 |
|
|
|
416 |
return MLIB_SUCCESS;
|
|
|
417 |
}
|
|
|
418 |
|
|
|
419 |
/***************************************************************/
|
|
|
420 |
#undef GET_POINTER
|
|
|
421 |
#define GET_POINTER(sp) \
|
|
|
422 |
sp = *(mlib_u8**)((mlib_u8*)lineAddr + PTR_SHIFT(Y)) + 4*(X >> MLIB_SHIFT); \
|
|
|
423 |
X += dX; \
|
|
|
424 |
Y += dY
|
|
|
425 |
|
|
|
426 |
/***************************************************************/
|
|
|
427 |
#ifndef MLIB_VIS2
|
|
|
428 |
|
|
|
429 |
#define LOAD_2CH_NA() \
|
|
|
430 |
s0 = vis_faligndata(LD_U16(sp1, 2), mask_7fff); \
|
|
|
431 |
s1 = vis_faligndata(LD_U16(sp1, 6), mask_7fff); \
|
|
|
432 |
s2 = vis_faligndata(LD_U16(sp1, srcYStride + 2), mask_7fff); \
|
|
|
433 |
s3 = vis_faligndata(LD_U16(sp1, srcYStride + 6), mask_7fff); \
|
|
|
434 |
\
|
|
|
435 |
s0 = vis_faligndata(LD_U16(sp1, 0), s0); \
|
|
|
436 |
s1 = vis_faligndata(LD_U16(sp1, 4), s1); \
|
|
|
437 |
s2 = vis_faligndata(LD_U16(sp1, srcYStride), s2); \
|
|
|
438 |
s3 = vis_faligndata(LD_U16(sp1, srcYStride + 4), s3); \
|
|
|
439 |
\
|
|
|
440 |
s0 = vis_faligndata(LD_U16(sp0, 2), s0); \
|
|
|
441 |
s1 = vis_faligndata(LD_U16(sp0, 6), s1); \
|
|
|
442 |
s2 = vis_faligndata(LD_U16(sp0, srcYStride + 2), s2); \
|
|
|
443 |
s3 = vis_faligndata(LD_U16(sp0, srcYStride + 6), s3); \
|
|
|
444 |
\
|
|
|
445 |
s0 = vis_faligndata(LD_U16(sp0, 0), s0); \
|
|
|
446 |
s1 = vis_faligndata(LD_U16(sp0, 4), s1); \
|
|
|
447 |
s2 = vis_faligndata(LD_U16(sp0, srcYStride), s2); \
|
|
|
448 |
s3 = vis_faligndata(LD_U16(sp0, srcYStride + 4), s3)
|
|
|
449 |
|
|
|
450 |
#else
|
|
|
451 |
|
|
|
452 |
#define LOAD_2CH_NA() \
|
|
|
453 |
s0 = vis_bshuffle(LD_U16(sp0, 0), LD_U16(sp1, 0)); \
|
|
|
454 |
s1 = vis_bshuffle(LD_U16(sp0, 4), LD_U16(sp1, 4)); \
|
|
|
455 |
s2 = vis_bshuffle(LD_U16(sp0, srcYStride), LD_U16(sp1, srcYStride)); \
|
|
|
456 |
s3 = vis_bshuffle(LD_U16(sp0, srcYStride + 4), LD_U16(sp1, srcYStride + 4)); \
|
|
|
457 |
\
|
|
|
458 |
t0 = vis_bshuffle(LD_U16(sp0, 2), LD_U16(sp1, 2)); \
|
|
|
459 |
t1 = vis_bshuffle(LD_U16(sp0, 6), LD_U16(sp1, 6)); \
|
|
|
460 |
t2 = vis_bshuffle(LD_U16(sp0, srcYStride + 2), LD_U16(sp1, srcYStride + 2)); \
|
|
|
461 |
t3 = vis_bshuffle(LD_U16(sp0, srcYStride + 6), LD_U16(sp1, srcYStride + 6)); \
|
|
|
462 |
\
|
|
|
463 |
s0 = vis_bshuffle(s0, t0); \
|
|
|
464 |
s1 = vis_bshuffle(s1, t1); \
|
|
|
465 |
s2 = vis_bshuffle(s2, t2); \
|
|
|
466 |
s3 = vis_bshuffle(s3, t3)
|
|
|
467 |
|
|
|
468 |
#endif /* MLIB_VIS2 */
|
|
|
469 |
|
|
|
470 |
/***************************************************************/
|
|
|
471 |
mlib_status FUN_NAME(2ch_na)(mlib_affine_param *param)
|
|
|
472 |
{
|
|
|
473 |
DECLAREVAR();
|
|
|
474 |
mlib_s32 max_xsize = param -> max_xsize, bsize;
|
|
|
475 |
mlib_s32 x0, x1, y0, y1;
|
|
|
476 |
mlib_d64 buff[BUF_SIZE], *pbuff = buff;
|
|
|
477 |
#ifdef MLIB_VIS2
|
|
|
478 |
mlib_d64 t0, t1, t2, t3;
|
|
|
479 |
#endif /* MLIB_VIS2 */
|
|
|
480 |
|
|
|
481 |
bsize = (max_xsize + 1)/2;
|
|
|
482 |
|
|
|
483 |
if (bsize > BUF_SIZE) {
|
|
|
484 |
pbuff = mlib_malloc(bsize*sizeof(mlib_d64));
|
|
|
485 |
|
|
|
486 |
if (pbuff == NULL) return MLIB_FAILURE;
|
|
|
487 |
}
|
|
|
488 |
|
|
|
489 |
MLIB_WRITE_BMASK(0x45CD67EF);
|
|
|
490 |
|
|
|
491 |
dx64 = vis_to_double_dup(((dX & 0xFFFF) << 16) | (dX & 0xFFFF));
|
|
|
492 |
dy64 = vis_to_double_dup(((dY & 0xFFFF) << 16) | (dY & 0xFFFF));
|
|
|
493 |
|
|
|
494 |
for (j = yStart; j <= yFinish; j++) {
|
|
|
495 |
mlib_u8 *sp0, *sp1;
|
|
|
496 |
|
|
|
497 |
#ifndef MLIB_VIS2
|
|
|
498 |
vis_alignaddr((void*)0, 6);
|
|
|
499 |
#endif /* MLIB_VIS2 */
|
|
|
500 |
|
|
|
501 |
NEW_LINE(2);
|
|
|
502 |
|
|
|
503 |
x0 = X; y0 = Y;
|
|
|
504 |
x1 = X + dX; y1 = Y + dY;
|
|
|
505 |
|
|
|
506 |
deltax = DOUBLE_4U16(x0, x0, x1, x1);
|
|
|
507 |
deltay = DOUBLE_4U16(y0, y0, y1, y1);
|
|
|
508 |
|
|
|
509 |
#pragma pipeloop(0)
|
|
|
510 |
for (i = 0; i < size/2; i++) {
|
|
|
511 |
GET_POINTER(sp0);
|
|
|
512 |
GET_POINTER(sp1);
|
|
|
513 |
LOAD_2CH_NA();
|
|
|
514 |
|
|
|
515 |
BL_SUM();
|
|
|
516 |
|
|
|
517 |
pbuff[i] = dd;
|
|
|
518 |
}
|
|
|
519 |
|
|
|
520 |
if (size & 1) {
|
|
|
521 |
GET_POINTER(sp0);
|
|
|
522 |
sp1 = sp0;
|
|
|
523 |
LOAD_2CH_NA();
|
|
|
524 |
|
|
|
525 |
BL_SUM();
|
|
|
526 |
|
|
|
527 |
pbuff[i] = dd;
|
|
|
528 |
}
|
|
|
529 |
|
|
|
530 |
mlib_ImageCopy_na((mlib_u8*)pbuff, dl, 4*size);
|
|
|
531 |
}
|
|
|
532 |
|
|
|
533 |
if (pbuff != buff) {
|
|
|
534 |
mlib_free(pbuff);
|
|
|
535 |
}
|
|
|
536 |
|
|
|
537 |
return MLIB_SUCCESS;
|
|
|
538 |
}
|
|
|
539 |
|
|
|
540 |
/***************************************************************/
|
|
|
541 |
#undef PREPARE_DELTAS
|
|
|
542 |
#define PREPARE_DELTAS \
|
|
|
543 |
if (warp_tbl != NULL) { \
|
|
|
544 |
dX = warp_tbl[2*j ]; \
|
|
|
545 |
dY = warp_tbl[2*j + 1]; \
|
|
|
546 |
dX = (dX - (dX >> 31)) &~ 1; /* rounding towards ZERO */ \
|
|
|
547 |
dY = (dY - (dY >> 31)) &~ 1; /* rounding towards ZERO */ \
|
|
|
548 |
dx64 = vis_to_double_dup((((dX >> 1) & 0xFFFF) << 16) | ((dX >> 1) & 0xFFFF)); \
|
|
|
549 |
dy64 = vis_to_double_dup((((dY >> 1) & 0xFFFF) << 16) | ((dY >> 1) & 0xFFFF)); \
|
|
|
550 |
}
|
|
|
551 |
|
|
|
552 |
/***************************************************************/
|
|
|
553 |
mlib_status FUN_NAME(3ch)(mlib_affine_param *param)
|
|
|
554 |
{
|
|
|
555 |
DECLAREVAR();
|
|
|
556 |
mlib_s32 max_xsize = param -> max_xsize;
|
|
|
557 |
mlib_d64 buff[BUF_SIZE], *pbuff = buff;
|
|
|
558 |
|
|
|
559 |
if (max_xsize > BUF_SIZE) {
|
|
|
560 |
pbuff = mlib_malloc(max_xsize*sizeof(mlib_d64));
|
|
|
561 |
|
|
|
562 |
if (pbuff == NULL) return MLIB_FAILURE;
|
|
|
563 |
}
|
|
|
564 |
|
|
|
565 |
dX = (dX - (dX >> 31)) &~ 1; /* rounding towards ZERO */
|
|
|
566 |
dY = (dY - (dY >> 31)) &~ 1; /* rounding towards ZERO */
|
|
|
567 |
dx64 = vis_to_double_dup((((dX >> 1) & 0xFFFF) << 16) | ((dX >> 1) & 0xFFFF));
|
|
|
568 |
dy64 = vis_to_double_dup((((dY >> 1) & 0xFFFF) << 16) | ((dY >> 1) & 0xFFFF));
|
|
|
569 |
|
|
|
570 |
for (j = yStart; j <= yFinish; j++) {
|
|
|
571 |
mlib_u8 *sp;
|
|
|
572 |
mlib_d64 *sp0, *sp1;
|
|
|
573 |
|
|
|
574 |
NEW_LINE(3);
|
|
|
575 |
|
|
|
576 |
deltax = DOUBLE_4U16(X, X, X, X);
|
|
|
577 |
deltay = DOUBLE_4U16(Y, Y, Y, Y);
|
|
|
578 |
|
|
|
579 |
#pragma pipeloop(0)
|
|
|
580 |
for (i = 0; i < size; i++) {
|
|
|
581 |
sp = *(mlib_u8**)((mlib_u8*)lineAddr + PTR_SHIFT(Y)) + 6*(X >> MLIB_SHIFT) - 2;
|
|
|
582 |
|
|
|
583 |
vis_alignaddr(sp, 0);
|
|
|
584 |
sp0 = AL_ADDR(sp, 0);
|
|
|
585 |
s0 = vis_faligndata(sp0[0], sp0[1]);
|
|
|
586 |
s1 = vis_faligndata(sp0[1], sp0[2]);
|
|
|
587 |
|
|
|
588 |
vis_alignaddr(sp, srcYStride);
|
|
|
589 |
sp1 = AL_ADDR(sp, srcYStride);
|
|
|
590 |
s2 = vis_faligndata(sp1[0], sp1[1]);
|
|
|
591 |
s3 = vis_faligndata(sp1[1], sp1[2]);
|
|
|
592 |
|
|
|
593 |
BL_SUM_3CH();
|
|
|
594 |
|
|
|
595 |
pbuff[i] = dd;
|
|
|
596 |
X += dX;
|
|
|
597 |
Y += dY;
|
|
|
598 |
}
|
|
|
599 |
|
|
|
600 |
mlib_v_ImageChannelExtract_S16_43L_D1((void *)pbuff, (void *)dl, size);
|
|
|
601 |
}
|
|
|
602 |
|
|
|
603 |
if (pbuff != buff) {
|
|
|
604 |
mlib_free(pbuff);
|
|
|
605 |
}
|
|
|
606 |
|
|
|
607 |
return MLIB_SUCCESS;
|
|
|
608 |
}
|
|
|
609 |
|
|
|
610 |
/***************************************************************/
|
|
|
611 |
mlib_status FUN_NAME(4ch)(mlib_affine_param *param)
|
|
|
612 |
{
|
|
|
613 |
DECLAREVAR();
|
|
|
614 |
|
|
|
615 |
if (((mlib_s32)lineAddr[0] | (mlib_s32)dstData | srcYStride | dstYStride) & 7) {
|
|
|
616 |
return FUN_NAME(4ch_na)(param);
|
|
|
617 |
}
|
|
|
618 |
|
|
|
619 |
srcYStride >>= 3;
|
|
|
620 |
|
|
|
621 |
dX = (dX - (dX >> 31)) &~ 1; /* rounding towards ZERO */
|
|
|
622 |
dY = (dY - (dY >> 31)) &~ 1; /* rounding towards ZERO */
|
|
|
623 |
dx64 = vis_to_double_dup((((dX >> 1) & 0xFFFF) << 16) | ((dX >> 1) & 0xFFFF));
|
|
|
624 |
dy64 = vis_to_double_dup((((dY >> 1) & 0xFFFF) << 16) | ((dY >> 1) & 0xFFFF));
|
|
|
625 |
|
|
|
626 |
for (j = yStart; j <= yFinish; j++) {
|
|
|
627 |
mlib_d64 *sp;
|
|
|
628 |
|
|
|
629 |
NEW_LINE(4);
|
|
|
630 |
|
|
|
631 |
deltax = DOUBLE_4U16(X, X, X, X);
|
|
|
632 |
deltay = DOUBLE_4U16(Y, Y, Y, Y);
|
|
|
633 |
|
|
|
634 |
#pragma pipeloop(0)
|
|
|
635 |
for (i = 0; i < size; i++) {
|
|
|
636 |
sp = *(mlib_d64**)((mlib_u8*)lineAddr + PTR_SHIFT(Y)) + (X >> MLIB_SHIFT);
|
|
|
637 |
s0 = sp[0];
|
|
|
638 |
s1 = sp[1];
|
|
|
639 |
s2 = sp[srcYStride];
|
|
|
640 |
s3 = sp[srcYStride + 1];
|
|
|
641 |
|
|
|
642 |
BL_SUM();
|
|
|
643 |
|
|
|
644 |
((mlib_d64*)dl)[i] = dd;
|
|
|
645 |
X += dX;
|
|
|
646 |
Y += dY;
|
|
|
647 |
}
|
|
|
648 |
}
|
|
|
649 |
|
|
|
650 |
return MLIB_SUCCESS;
|
|
|
651 |
}
|
|
|
652 |
|
|
|
653 |
/***************************************************************/
|
|
|
654 |
mlib_status FUN_NAME(4ch_na)(mlib_affine_param *param)
|
|
|
655 |
{
|
|
|
656 |
DECLAREVAR();
|
|
|
657 |
mlib_s32 max_xsize = param -> max_xsize;
|
|
|
658 |
mlib_d64 buff[BUF_SIZE], *pbuff = buff;
|
|
|
659 |
|
|
|
660 |
if (max_xsize > BUF_SIZE) {
|
|
|
661 |
pbuff = mlib_malloc(max_xsize*sizeof(mlib_d64));
|
|
|
662 |
|
|
|
663 |
if (pbuff == NULL) return MLIB_FAILURE;
|
|
|
664 |
}
|
|
|
665 |
|
|
|
666 |
dX = (dX - (dX >> 31)) &~ 1; /* rounding towards ZERO */
|
|
|
667 |
dY = (dY - (dY >> 31)) &~ 1; /* rounding towards ZERO */
|
|
|
668 |
dx64 = vis_to_double_dup((((dX >> 1) & 0xFFFF) << 16) | ((dX >> 1) & 0xFFFF));
|
|
|
669 |
dy64 = vis_to_double_dup((((dY >> 1) & 0xFFFF) << 16) | ((dY >> 1) & 0xFFFF));
|
|
|
670 |
|
|
|
671 |
for (j = yStart; j <= yFinish; j++) {
|
|
|
672 |
mlib_u8 *sp;
|
|
|
673 |
mlib_d64 *sp0, *sp1;
|
|
|
674 |
|
|
|
675 |
NEW_LINE(4);
|
|
|
676 |
|
|
|
677 |
deltax = DOUBLE_4U16(X, X, X, X);
|
|
|
678 |
deltay = DOUBLE_4U16(Y, Y, Y, Y);
|
|
|
679 |
|
|
|
680 |
#pragma pipeloop(0)
|
|
|
681 |
for (i = 0; i < size; i++) {
|
|
|
682 |
sp = *(mlib_u8**)((mlib_u8*)lineAddr + PTR_SHIFT(Y)) + 8*(X >> MLIB_SHIFT);
|
|
|
683 |
|
|
|
684 |
vis_alignaddr(sp, 0);
|
|
|
685 |
sp0 = AL_ADDR(sp, 0);
|
|
|
686 |
s0 = vis_faligndata(sp0[0], sp0[1]);
|
|
|
687 |
s1 = vis_faligndata(sp0[1], sp0[2]);
|
|
|
688 |
|
|
|
689 |
vis_alignaddr(sp, srcYStride);
|
|
|
690 |
sp1 = AL_ADDR(sp, srcYStride);
|
|
|
691 |
s2 = vis_faligndata(sp1[0], sp1[1]);
|
|
|
692 |
s3 = vis_faligndata(sp1[1], sp1[2]);
|
|
|
693 |
|
|
|
694 |
BL_SUM();
|
|
|
695 |
|
|
|
696 |
pbuff[i] = dd;
|
|
|
697 |
X += dX;
|
|
|
698 |
Y += dY;
|
|
|
699 |
}
|
|
|
700 |
|
|
|
701 |
mlib_ImageCopy_na((mlib_u8*)pbuff, dl, 8*size);
|
|
|
702 |
}
|
|
|
703 |
|
|
|
704 |
if (pbuff != buff) {
|
|
|
705 |
mlib_free(pbuff);
|
|
|
706 |
}
|
|
|
707 |
|
|
|
708 |
return MLIB_SUCCESS;
|
|
|
709 |
}
|
|
|
710 |
|
|
|
711 |
/***************************************************************/
|