|
1 /* |
|
2 * Copyright (c) 2003, Oracle and/or its affiliates. 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. Oracle designates this |
|
8 * particular file as subject to the "Classpath" exception as provided |
|
9 * by Oracle 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 Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA |
|
22 * or visit www.oracle.com if you need additional information or have any |
|
23 * 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 |
|
42 /*#define MLIB_VIS2*/ |
|
43 |
|
44 /***************************************************************/ |
|
45 #define DTYPE mlib_s16 |
|
46 |
|
47 #define FUN_NAME(CHAN) mlib_ImageAffine_s16_##CHAN##_bl |
|
48 |
|
49 /***************************************************************/ |
|
50 static mlib_status FUN_NAME(2ch_na)(mlib_affine_param *param); |
|
51 static mlib_status FUN_NAME(4ch_na)(mlib_affine_param *param); |
|
52 |
|
53 /***************************************************************/ |
|
54 const mlib_u64 mlib_dmask_arr[] = { |
|
55 0x0000000000000000, 0x000000000000FFFF, 0x00000000FFFF0000, 0x00000000FFFFFFFF, |
|
56 0x0000FFFF00000000, 0x0000FFFF0000FFFF, 0x0000FFFFFFFF0000, 0x0000FFFFFFFFFFFF, |
|
57 0xFFFF000000000000, 0xFFFF00000000FFFF, 0xFFFF0000FFFF0000, 0xFFFF0000FFFFFFFF, |
|
58 0xFFFFFFFF00000000, 0xFFFFFFFF0000FFFF, 0xFFFFFFFFFFFF0000, 0xFFFFFFFFFFFFFFFF |
|
59 }; |
|
60 |
|
61 /***************************************************************/ |
|
62 #define XOR_8000(x) |
|
63 |
|
64 /***************************************************************/ |
|
65 #ifdef MLIB_VIS2 |
|
66 #define MLIB_WRITE_BMASK(bmask) vis_write_bmask(bmask, 0) |
|
67 #else |
|
68 #define MLIB_WRITE_BMASK(bmask) |
|
69 #endif |
|
70 |
|
71 /***************************************************************/ |
|
72 #undef DECLAREVAR |
|
73 #define DECLAREVAR() \ |
|
74 DECLAREVAR0(); \ |
|
75 mlib_s32 *warp_tbl = param -> warp_tbl; \ |
|
76 mlib_s32 srcYStride = param -> srcYStride; \ |
|
77 mlib_u8 *dl; \ |
|
78 mlib_s32 i, size; \ |
|
79 /*mlib_d64 mask_8000 = vis_to_double_dup(0x80008000);*/ \ |
|
80 mlib_d64 mask_7fff = vis_to_double_dup(0x7FFF7FFF); \ |
|
81 mlib_d64 dx64, dy64, deltax, deltay, delta1_x, delta1_y; \ |
|
82 mlib_d64 s0, s1, s2, s3; \ |
|
83 mlib_d64 d0, d1, d2, d3, dd |
|
84 |
|
85 /***************************************************************/ |
|
86 |
|
87 /* arguments (x, y) are swapped to prevent overflow */ |
|
88 #define FMUL_16x16(x, y) \ |
|
89 vis_fpadd16(vis_fmul8sux16(y, x), \ |
|
90 vis_fmul8ulx16(y, x)) |
|
91 |
|
92 /***************************************************************/ |
|
93 #define BUF_SIZE 512 |
|
94 |
|
95 /***************************************************************/ |
|
96 #define DOUBLE_4U16(x0, x1, x2, x3) \ |
|
97 vis_to_double(((((x0) & 0xFFFE) << 15) | (((x1) & 0xFFFE) >> 1)), \ |
|
98 ((((x2) & 0xFFFE) << 15) | (((x3) & 0xFFFE) >> 1))) |
|
99 |
|
100 /***************************************************************/ |
|
101 #define BL_SUM() \ |
|
102 XOR_8000(s0); \ |
|
103 XOR_8000(s1); \ |
|
104 XOR_8000(s2); \ |
|
105 XOR_8000(s3); \ |
|
106 \ |
|
107 delta1_x = vis_fpsub16(mask_7fff, deltax); \ |
|
108 delta1_y = vis_fpsub16(mask_7fff, deltay); \ |
|
109 \ |
|
110 d0 = FMUL_16x16(s0, delta1_x); \ |
|
111 d1 = FMUL_16x16(s1, deltax); \ |
|
112 d0 = vis_fpadd16(d0, d1); \ |
|
113 d0 = vis_fpadd16(d0, d0); \ |
|
114 d0 = FMUL_16x16(d0, delta1_y); \ |
|
115 \ |
|
116 d2 = FMUL_16x16(s2, delta1_x); \ |
|
117 d3 = FMUL_16x16(s3, deltax); \ |
|
118 d2 = vis_fpadd16(d2, d3); \ |
|
119 d2 = vis_fpadd16(d2, d2); \ |
|
120 d2 = FMUL_16x16(d2, deltay); \ |
|
121 \ |
|
122 dd = vis_fpadd16(d0, d2); \ |
|
123 dd = vis_fpadd16(dd, dd); \ |
|
124 XOR_8000(dd); \ |
|
125 \ |
|
126 deltax = vis_fpadd16(deltax, dx64); \ |
|
127 deltay = vis_fpadd16(deltay, dy64); \ |
|
128 deltax = vis_fand(deltax, mask_7fff); \ |
|
129 deltay = vis_fand(deltay, mask_7fff) |
|
130 |
|
131 /***************************************************************/ |
|
132 #define BL_SUM_3CH() \ |
|
133 XOR_8000(s0); \ |
|
134 XOR_8000(s1); \ |
|
135 XOR_8000(s2); \ |
|
136 XOR_8000(s3); \ |
|
137 \ |
|
138 delta1_x = vis_fpsub16(mask_7fff, deltax); \ |
|
139 delta1_y = vis_fpsub16(mask_7fff, deltay); \ |
|
140 \ |
|
141 d0 = FMUL_16x16(s0, delta1_y); \ |
|
142 d2 = FMUL_16x16(s2, deltay); \ |
|
143 d0 = vis_fpadd16(d0, d2); \ |
|
144 d0 = vis_fpadd16(d0, d0); \ |
|
145 d0 = FMUL_16x16(d0, delta1_x); \ |
|
146 \ |
|
147 d1 = FMUL_16x16(s1, delta1_y); \ |
|
148 d3 = FMUL_16x16(s3, deltay); \ |
|
149 d1 = vis_fpadd16(d1, d3); \ |
|
150 d1 = vis_fpadd16(d1, d1); \ |
|
151 d1 = FMUL_16x16(d1, deltax); \ |
|
152 \ |
|
153 vis_alignaddr((void*)0, 2); \ |
|
154 d0 = vis_faligndata(d0, d0); \ |
|
155 dd = vis_fpadd16(d0, d1); \ |
|
156 dd = vis_fpadd16(dd, dd); \ |
|
157 XOR_8000(dd); \ |
|
158 \ |
|
159 deltax = vis_fpadd16(deltax, dx64); \ |
|
160 deltay = vis_fpadd16(deltay, dy64); \ |
|
161 deltax = vis_fand(deltax, mask_7fff); \ |
|
162 deltay = vis_fand(deltay, mask_7fff) |
|
163 |
|
164 /***************************************************************/ |
|
165 #define LD_U16(sp, ind) vis_ld_u16(sp + ind) |
|
166 |
|
167 /***************************************************************/ |
|
168 #ifndef MLIB_VIS2 |
|
169 |
|
170 #define LOAD_1CH() \ |
|
171 s0 = vis_faligndata(LD_U16(sp3, 0), mask_7fff); \ |
|
172 s1 = vis_faligndata(LD_U16(sp3, 2), mask_7fff); \ |
|
173 s2 = vis_faligndata(LD_U16(sp3, srcYStride), mask_7fff); \ |
|
174 s3 = vis_faligndata(LD_U16(sp3, srcYStride + 2), mask_7fff); \ |
|
175 \ |
|
176 s0 = vis_faligndata(LD_U16(sp2, 0), s0); \ |
|
177 s1 = vis_faligndata(LD_U16(sp2, 2), s1); \ |
|
178 s2 = vis_faligndata(LD_U16(sp2, srcYStride), s2); \ |
|
179 s3 = vis_faligndata(LD_U16(sp2, srcYStride + 2), s3); \ |
|
180 \ |
|
181 s0 = vis_faligndata(LD_U16(sp1, 0), s0); \ |
|
182 s1 = vis_faligndata(LD_U16(sp1, 2), s1); \ |
|
183 s2 = vis_faligndata(LD_U16(sp1, srcYStride), s2); \ |
|
184 s3 = vis_faligndata(LD_U16(sp1, srcYStride + 2), s3); \ |
|
185 \ |
|
186 s0 = vis_faligndata(LD_U16(sp0, 0), s0); \ |
|
187 s1 = vis_faligndata(LD_U16(sp0, 2), s1); \ |
|
188 s2 = vis_faligndata(LD_U16(sp0, srcYStride), s2); \ |
|
189 s3 = vis_faligndata(LD_U16(sp0, srcYStride + 2), s3) |
|
190 |
|
191 #else |
|
192 |
|
193 #define LOAD_1CH() \ |
|
194 s0 = vis_bshuffle(LD_U16(sp0, 0), LD_U16(sp2, 0)); \ |
|
195 s1 = vis_bshuffle(LD_U16(sp0, 2), LD_U16(sp2, 2)); \ |
|
196 s2 = vis_bshuffle(LD_U16(sp0, srcYStride), LD_U16(sp2, srcYStride)); \ |
|
197 s3 = vis_bshuffle(LD_U16(sp0, srcYStride + 2), LD_U16(sp2, srcYStride + 2)); \ |
|
198 \ |
|
199 t0 = vis_bshuffle(LD_U16(sp1, 0), LD_U16(sp3, 0)); \ |
|
200 t1 = vis_bshuffle(LD_U16(sp1, 2), LD_U16(sp3, 2)); \ |
|
201 t2 = vis_bshuffle(LD_U16(sp1, srcYStride), LD_U16(sp3, srcYStride)); \ |
|
202 t3 = vis_bshuffle(LD_U16(sp1, srcYStride + 2), LD_U16(sp3, srcYStride + 2)); \ |
|
203 \ |
|
204 s0 = vis_bshuffle(s0, t0); \ |
|
205 s1 = vis_bshuffle(s1, t1); \ |
|
206 s2 = vis_bshuffle(s2, t2); \ |
|
207 s3 = vis_bshuffle(s3, t3) |
|
208 |
|
209 #endif |
|
210 |
|
211 /***************************************************************/ |
|
212 #define GET_POINTER(sp) \ |
|
213 sp = *(mlib_u8**)((mlib_u8*)lineAddr + PTR_SHIFT(Y)) + 2*(X >> MLIB_SHIFT); \ |
|
214 X += dX; \ |
|
215 Y += dY |
|
216 |
|
217 /***************************************************************/ |
|
218 #undef PREPARE_DELTAS |
|
219 #define PREPARE_DELTAS \ |
|
220 if (warp_tbl != NULL) { \ |
|
221 dX = warp_tbl[2*j ]; \ |
|
222 dY = warp_tbl[2*j + 1]; \ |
|
223 dx64 = vis_to_double_dup((((dX << 1) & 0xFFFF) << 16) | ((dX << 1) & 0xFFFF)); \ |
|
224 dy64 = vis_to_double_dup((((dY << 1) & 0xFFFF) << 16) | ((dY << 1) & 0xFFFF)); \ |
|
225 } |
|
226 |
|
227 /***************************************************************/ |
|
228 mlib_status FUN_NAME(1ch)(mlib_affine_param *param) |
|
229 { |
|
230 DECLAREVAR(); |
|
231 mlib_s32 off; |
|
232 mlib_s32 x0, x1, x2, x3, y0, y1, y2, y3; |
|
233 #ifdef MLIB_VIS2 |
|
234 mlib_d64 t0, t1, t2, t3; |
|
235 vis_write_bmask(0x45CD67EF, 0); |
|
236 #else |
|
237 vis_alignaddr((void*)0, 6); |
|
238 #endif |
|
239 |
|
240 dx64 = vis_to_double_dup((((dX << 1) & 0xFFFF) << 16) | ((dX << 1) & 0xFFFF)); |
|
241 dy64 = vis_to_double_dup((((dY << 1) & 0xFFFF) << 16) | ((dY << 1) & 0xFFFF)); |
|
242 |
|
243 for (j = yStart; j <= yFinish; j++) { |
|
244 mlib_u8 *sp0, *sp1, *sp2, *sp3; |
|
245 mlib_d64 *dp, dmask; |
|
246 |
|
247 NEW_LINE(1); |
|
248 |
|
249 off = (mlib_s32)dl & 7; |
|
250 dp = (mlib_d64*)(dl - off); |
|
251 off >>= 1; |
|
252 |
|
253 x0 = X - off*dX; y0 = Y - off*dY; |
|
254 x1 = x0 + dX; y1 = y0 + dY; |
|
255 x2 = x1 + dX; y2 = y1 + dY; |
|
256 x3 = x2 + dX; y3 = y2 + dY; |
|
257 |
|
258 deltax = DOUBLE_4U16(x0, x1, x2, x3); |
|
259 deltay = DOUBLE_4U16(y0, y1, y2, y3); |
|
260 |
|
261 if (off) { |
|
262 mlib_s32 emask = vis_edge16((void*)(2*off), (void*)(2*(off + size - 1))); |
|
263 |
|
264 off = 4 - off; |
|
265 GET_POINTER(sp3); |
|
266 sp0 = sp1 = sp2 = sp3; |
|
267 |
|
268 if (off > 1 && size > 1) { |
|
269 GET_POINTER(sp3); |
|
270 } |
|
271 |
|
272 if (off > 2) { |
|
273 sp2 = sp3; |
|
274 |
|
275 if (size > 2) { |
|
276 GET_POINTER(sp3); |
|
277 } |
|
278 } |
|
279 |
|
280 LOAD_1CH(); |
|
281 BL_SUM(); |
|
282 |
|
283 dmask = ((mlib_d64*)mlib_dmask_arr)[emask]; |
|
284 *dp++ = vis_for (vis_fand(dmask, dd), vis_fandnot(dmask, dp[0])); |
|
285 |
|
286 size -= off; |
|
287 |
|
288 if (size < 0) size = 0; |
|
289 } |
|
290 |
|
291 #pragma pipeloop(0) |
|
292 for (i = 0; i < size/4; i++) { |
|
293 GET_POINTER(sp0); |
|
294 GET_POINTER(sp1); |
|
295 GET_POINTER(sp2); |
|
296 GET_POINTER(sp3); |
|
297 |
|
298 LOAD_1CH(); |
|
299 BL_SUM(); |
|
300 |
|
301 dp[i] = dd; |
|
302 } |
|
303 |
|
304 off = size & 3; |
|
305 |
|
306 if (off) { |
|
307 GET_POINTER(sp0); |
|
308 sp1 = sp2 = sp3 = sp0; |
|
309 |
|
310 if (off > 1) { |
|
311 GET_POINTER(sp1); |
|
312 } |
|
313 |
|
314 if (off > 2) { |
|
315 GET_POINTER(sp2); |
|
316 } |
|
317 |
|
318 LOAD_1CH(); |
|
319 BL_SUM(); |
|
320 |
|
321 dmask = ((mlib_d64*)mlib_dmask_arr)[(0xF0 >> off) & 0x0F]; |
|
322 dp[i] = vis_for (vis_fand(dmask, dd), vis_fandnot(dmask, dp[i])); |
|
323 } |
|
324 } |
|
325 |
|
326 return MLIB_SUCCESS; |
|
327 } |
|
328 |
|
329 /***************************************************************/ |
|
330 #undef GET_POINTER |
|
331 #define GET_POINTER(sp) \ |
|
332 sp = *(mlib_f32**)((mlib_u8*)lineAddr + PTR_SHIFT(Y)) + (X >> MLIB_SHIFT); \ |
|
333 X += dX; \ |
|
334 Y += dY |
|
335 |
|
336 /***************************************************************/ |
|
337 #define LOAD_2CH() \ |
|
338 s0 = vis_freg_pair(sp0[0], sp1[0]); \ |
|
339 s1 = vis_freg_pair(sp0[1], sp1[1]); \ |
|
340 s2 = vis_freg_pair(sp0[srcYStride], sp1[srcYStride]); \ |
|
341 s3 = vis_freg_pair(sp0[srcYStride + 1], sp1[srcYStride + 1]) |
|
342 |
|
343 /***************************************************************/ |
|
344 #undef PREPARE_DELTAS |
|
345 #define PREPARE_DELTAS \ |
|
346 if (warp_tbl != NULL) { \ |
|
347 dX = warp_tbl[2*j ]; \ |
|
348 dY = warp_tbl[2*j + 1]; \ |
|
349 dx64 = vis_to_double_dup(((dX & 0xFFFF) << 16) | (dX & 0xFFFF)); \ |
|
350 dy64 = vis_to_double_dup(((dY & 0xFFFF) << 16) | (dY & 0xFFFF)); \ |
|
351 } |
|
352 |
|
353 /***************************************************************/ |
|
354 mlib_status FUN_NAME(2ch)(mlib_affine_param *param) |
|
355 { |
|
356 DECLAREVAR(); |
|
357 mlib_s32 off; |
|
358 mlib_s32 x0, x1, y0, y1; |
|
359 |
|
360 if (((mlib_s32)lineAddr[0] | (mlib_s32)dstData | srcYStride | dstYStride) & 3) { |
|
361 return FUN_NAME(2ch_na)(param); |
|
362 } |
|
363 |
|
364 srcYStride >>= 2; |
|
365 |
|
366 dx64 = vis_to_double_dup(((dX & 0xFFFF) << 16) | (dX & 0xFFFF)); |
|
367 dy64 = vis_to_double_dup(((dY & 0xFFFF) << 16) | (dY & 0xFFFF)); |
|
368 |
|
369 for (j = yStart; j <= yFinish; j++) { |
|
370 mlib_f32 *sp0, *sp1; |
|
371 mlib_d64 *dp; |
|
372 |
|
373 NEW_LINE(2); |
|
374 |
|
375 off = (mlib_s32)dl & 7; |
|
376 dp = (mlib_d64*)(dl - off); |
|
377 |
|
378 if (off) { |
|
379 x0 = X - dX; y0 = Y - dY; |
|
380 x1 = X; y1 = Y; |
|
381 } else { |
|
382 x0 = X; y0 = Y; |
|
383 x1 = X + dX; y1 = Y + dY; |
|
384 } |
|
385 |
|
386 deltax = DOUBLE_4U16(x0, x0, x1, x1); |
|
387 deltay = DOUBLE_4U16(y0, y0, y1, y1); |
|
388 |
|
389 if (off) { |
|
390 GET_POINTER(sp1); |
|
391 sp0 = sp1; |
|
392 LOAD_2CH(); |
|
393 |
|
394 BL_SUM(); |
|
395 |
|
396 ((mlib_f32*)dp)[1] = vis_read_lo(dd); |
|
397 dp++; |
|
398 size--; |
|
399 } |
|
400 |
|
401 #pragma pipeloop(0) |
|
402 for (i = 0; i < size/2; i++) { |
|
403 GET_POINTER(sp0); |
|
404 GET_POINTER(sp1); |
|
405 LOAD_2CH(); |
|
406 |
|
407 BL_SUM(); |
|
408 |
|
409 *dp++ = dd; |
|
410 } |
|
411 |
|
412 if (size & 1) { |
|
413 GET_POINTER(sp0); |
|
414 sp1 = sp0; |
|
415 LOAD_2CH(); |
|
416 |
|
417 BL_SUM(); |
|
418 |
|
419 ((mlib_f32*)dp)[0] = vis_read_hi(dd); |
|
420 } |
|
421 } |
|
422 |
|
423 return MLIB_SUCCESS; |
|
424 } |
|
425 |
|
426 /***************************************************************/ |
|
427 #undef GET_POINTER |
|
428 #define GET_POINTER(sp) \ |
|
429 sp = *(mlib_u8**)((mlib_u8*)lineAddr + PTR_SHIFT(Y)) + 4*(X >> MLIB_SHIFT); \ |
|
430 X += dX; \ |
|
431 Y += dY |
|
432 |
|
433 /***************************************************************/ |
|
434 #ifndef MLIB_VIS2 |
|
435 |
|
436 #define LOAD_2CH_NA() \ |
|
437 s0 = vis_faligndata(LD_U16(sp1, 2), mask_7fff); \ |
|
438 s1 = vis_faligndata(LD_U16(sp1, 6), mask_7fff); \ |
|
439 s2 = vis_faligndata(LD_U16(sp1, srcYStride + 2), mask_7fff); \ |
|
440 s3 = vis_faligndata(LD_U16(sp1, srcYStride + 6), mask_7fff); \ |
|
441 \ |
|
442 s0 = vis_faligndata(LD_U16(sp1, 0), s0); \ |
|
443 s1 = vis_faligndata(LD_U16(sp1, 4), s1); \ |
|
444 s2 = vis_faligndata(LD_U16(sp1, srcYStride), s2); \ |
|
445 s3 = vis_faligndata(LD_U16(sp1, srcYStride + 4), s3); \ |
|
446 \ |
|
447 s0 = vis_faligndata(LD_U16(sp0, 2), s0); \ |
|
448 s1 = vis_faligndata(LD_U16(sp0, 6), s1); \ |
|
449 s2 = vis_faligndata(LD_U16(sp0, srcYStride + 2), s2); \ |
|
450 s3 = vis_faligndata(LD_U16(sp0, srcYStride + 6), s3); \ |
|
451 \ |
|
452 s0 = vis_faligndata(LD_U16(sp0, 0), s0); \ |
|
453 s1 = vis_faligndata(LD_U16(sp0, 4), s1); \ |
|
454 s2 = vis_faligndata(LD_U16(sp0, srcYStride), s2); \ |
|
455 s3 = vis_faligndata(LD_U16(sp0, srcYStride + 4), s3) |
|
456 |
|
457 #else |
|
458 |
|
459 #define LOAD_2CH_NA() \ |
|
460 s0 = vis_bshuffle(LD_U16(sp0, 0), LD_U16(sp1, 0)); \ |
|
461 s1 = vis_bshuffle(LD_U16(sp0, 4), LD_U16(sp1, 4)); \ |
|
462 s2 = vis_bshuffle(LD_U16(sp0, srcYStride), LD_U16(sp1, srcYStride)); \ |
|
463 s3 = vis_bshuffle(LD_U16(sp0, srcYStride + 4), LD_U16(sp1, srcYStride + 4)); \ |
|
464 \ |
|
465 t0 = vis_bshuffle(LD_U16(sp0, 2), LD_U16(sp1, 2)); \ |
|
466 t1 = vis_bshuffle(LD_U16(sp0, 6), LD_U16(sp1, 6)); \ |
|
467 t2 = vis_bshuffle(LD_U16(sp0, srcYStride + 2), LD_U16(sp1, srcYStride + 2)); \ |
|
468 t3 = vis_bshuffle(LD_U16(sp0, srcYStride + 6), LD_U16(sp1, srcYStride + 6)); \ |
|
469 \ |
|
470 s0 = vis_bshuffle(s0, t0); \ |
|
471 s1 = vis_bshuffle(s1, t1); \ |
|
472 s2 = vis_bshuffle(s2, t2); \ |
|
473 s3 = vis_bshuffle(s3, t3) |
|
474 |
|
475 #endif |
|
476 |
|
477 /***************************************************************/ |
|
478 mlib_status FUN_NAME(2ch_na)(mlib_affine_param *param) |
|
479 { |
|
480 DECLAREVAR(); |
|
481 mlib_s32 max_xsize = param -> max_xsize, bsize; |
|
482 mlib_s32 x0, x1, y0, y1; |
|
483 mlib_d64 buff[BUF_SIZE], *pbuff = buff; |
|
484 #ifdef MLIB_VIS2 |
|
485 mlib_d64 t0, t1, t2, t3; |
|
486 #endif |
|
487 |
|
488 bsize = (max_xsize + 1)/2; |
|
489 |
|
490 if (bsize > BUF_SIZE) { |
|
491 pbuff = mlib_malloc(bsize*sizeof(mlib_d64)); |
|
492 |
|
493 if (pbuff == NULL) return MLIB_FAILURE; |
|
494 } |
|
495 |
|
496 MLIB_WRITE_BMASK(0x45CD67EF); |
|
497 |
|
498 dx64 = vis_to_double_dup(((dX & 0xFFFF) << 16) | (dX & 0xFFFF)); |
|
499 dy64 = vis_to_double_dup(((dY & 0xFFFF) << 16) | (dY & 0xFFFF)); |
|
500 |
|
501 for (j = yStart; j <= yFinish; j++) { |
|
502 mlib_u8 *sp0, *sp1; |
|
503 |
|
504 #ifndef MLIB_VIS2 |
|
505 vis_alignaddr((void*)0, 6); |
|
506 #endif |
|
507 |
|
508 NEW_LINE(2); |
|
509 |
|
510 x0 = X; y0 = Y; |
|
511 x1 = X + dX; y1 = Y + dY; |
|
512 |
|
513 deltax = DOUBLE_4U16(x0, x0, x1, x1); |
|
514 deltay = DOUBLE_4U16(y0, y0, y1, y1); |
|
515 |
|
516 #pragma pipeloop(0) |
|
517 for (i = 0; i < size/2; i++) { |
|
518 GET_POINTER(sp0); |
|
519 GET_POINTER(sp1); |
|
520 LOAD_2CH_NA(); |
|
521 |
|
522 BL_SUM(); |
|
523 |
|
524 pbuff[i] = dd; |
|
525 } |
|
526 |
|
527 if (size & 1) { |
|
528 GET_POINTER(sp0); |
|
529 sp1 = sp0; |
|
530 LOAD_2CH_NA(); |
|
531 |
|
532 BL_SUM(); |
|
533 |
|
534 pbuff[i] = dd; |
|
535 } |
|
536 |
|
537 mlib_ImageCopy_na((mlib_u8*)pbuff, dl, 4*size); |
|
538 } |
|
539 |
|
540 if (pbuff != buff) { |
|
541 mlib_free(pbuff); |
|
542 } |
|
543 |
|
544 return MLIB_SUCCESS; |
|
545 } |
|
546 |
|
547 /***************************************************************/ |
|
548 #undef PREPARE_DELTAS |
|
549 #define PREPARE_DELTAS \ |
|
550 if (warp_tbl != NULL) { \ |
|
551 dX = warp_tbl[2*j ]; \ |
|
552 dY = warp_tbl[2*j + 1]; \ |
|
553 dX = (dX - (dX >> 31)) &~ 1; /* rounding towards ZERO */ \ |
|
554 dY = (dY - (dY >> 31)) &~ 1; /* rounding towards ZERO */ \ |
|
555 dx64 = vis_to_double_dup((((dX >> 1) & 0xFFFF) << 16) | ((dX >> 1) & 0xFFFF)); \ |
|
556 dy64 = vis_to_double_dup((((dY >> 1) & 0xFFFF) << 16) | ((dY >> 1) & 0xFFFF)); \ |
|
557 } |
|
558 |
|
559 /***************************************************************/ |
|
560 mlib_status FUN_NAME(3ch)(mlib_affine_param *param) |
|
561 { |
|
562 DECLAREVAR(); |
|
563 mlib_s32 max_xsize = param -> max_xsize; |
|
564 mlib_d64 buff[BUF_SIZE], *pbuff = buff; |
|
565 |
|
566 if (max_xsize > BUF_SIZE) { |
|
567 pbuff = mlib_malloc(max_xsize*sizeof(mlib_d64)); |
|
568 |
|
569 if (pbuff == NULL) return MLIB_FAILURE; |
|
570 } |
|
571 |
|
572 dX = (dX - (dX >> 31)) &~ 1; /* rounding towards ZERO */ |
|
573 dY = (dY - (dY >> 31)) &~ 1; /* rounding towards ZERO */ |
|
574 dx64 = vis_to_double_dup((((dX >> 1) & 0xFFFF) << 16) | ((dX >> 1) & 0xFFFF)); |
|
575 dy64 = vis_to_double_dup((((dY >> 1) & 0xFFFF) << 16) | ((dY >> 1) & 0xFFFF)); |
|
576 |
|
577 for (j = yStart; j <= yFinish; j++) { |
|
578 mlib_u8 *sp; |
|
579 mlib_d64 *sp0, *sp1; |
|
580 |
|
581 NEW_LINE(3); |
|
582 |
|
583 deltax = DOUBLE_4U16(X, X, X, X); |
|
584 deltay = DOUBLE_4U16(Y, Y, Y, Y); |
|
585 |
|
586 #pragma pipeloop(0) |
|
587 for (i = 0; i < size; i++) { |
|
588 sp = *(mlib_u8**)((mlib_u8*)lineAddr + PTR_SHIFT(Y)) + 6*(X >> MLIB_SHIFT) - 2; |
|
589 |
|
590 vis_alignaddr(sp, 0); |
|
591 sp0 = AL_ADDR(sp, 0); |
|
592 s0 = vis_faligndata(sp0[0], sp0[1]); |
|
593 s1 = vis_faligndata(sp0[1], sp0[2]); |
|
594 |
|
595 vis_alignaddr(sp, srcYStride); |
|
596 sp1 = AL_ADDR(sp, srcYStride); |
|
597 s2 = vis_faligndata(sp1[0], sp1[1]); |
|
598 s3 = vis_faligndata(sp1[1], sp1[2]); |
|
599 |
|
600 BL_SUM_3CH(); |
|
601 |
|
602 pbuff[i] = dd; |
|
603 X += dX; |
|
604 Y += dY; |
|
605 } |
|
606 |
|
607 mlib_v_ImageChannelExtract_S16_43L_D1((void *)pbuff, (void *)dl, size); |
|
608 } |
|
609 |
|
610 if (pbuff != buff) { |
|
611 mlib_free(pbuff); |
|
612 } |
|
613 |
|
614 return MLIB_SUCCESS; |
|
615 } |
|
616 |
|
617 /***************************************************************/ |
|
618 mlib_status FUN_NAME(4ch)(mlib_affine_param *param) |
|
619 { |
|
620 DECLAREVAR(); |
|
621 |
|
622 if (((mlib_s32)lineAddr[0] | (mlib_s32)dstData | srcYStride | dstYStride) & 7) { |
|
623 return FUN_NAME(4ch_na)(param); |
|
624 } |
|
625 |
|
626 srcYStride >>= 3; |
|
627 |
|
628 dX = (dX - (dX >> 31)) &~ 1; /* rounding towards ZERO */ |
|
629 dY = (dY - (dY >> 31)) &~ 1; /* rounding towards ZERO */ |
|
630 dx64 = vis_to_double_dup((((dX >> 1) & 0xFFFF) << 16) | ((dX >> 1) & 0xFFFF)); |
|
631 dy64 = vis_to_double_dup((((dY >> 1) & 0xFFFF) << 16) | ((dY >> 1) & 0xFFFF)); |
|
632 |
|
633 for (j = yStart; j <= yFinish; j++) { |
|
634 mlib_d64 *sp; |
|
635 |
|
636 NEW_LINE(4); |
|
637 |
|
638 deltax = DOUBLE_4U16(X, X, X, X); |
|
639 deltay = DOUBLE_4U16(Y, Y, Y, Y); |
|
640 |
|
641 #pragma pipeloop(0) |
|
642 for (i = 0; i < size; i++) { |
|
643 sp = *(mlib_d64**)((mlib_u8*)lineAddr + PTR_SHIFT(Y)) + (X >> MLIB_SHIFT); |
|
644 s0 = sp[0]; |
|
645 s1 = sp[1]; |
|
646 s2 = sp[srcYStride]; |
|
647 s3 = sp[srcYStride + 1]; |
|
648 |
|
649 BL_SUM(); |
|
650 |
|
651 ((mlib_d64*)dl)[i] = dd; |
|
652 X += dX; |
|
653 Y += dY; |
|
654 } |
|
655 } |
|
656 |
|
657 return MLIB_SUCCESS; |
|
658 } |
|
659 |
|
660 /***************************************************************/ |
|
661 mlib_status FUN_NAME(4ch_na)(mlib_affine_param *param) |
|
662 { |
|
663 DECLAREVAR(); |
|
664 mlib_s32 max_xsize = param -> max_xsize; |
|
665 mlib_d64 buff[BUF_SIZE], *pbuff = buff; |
|
666 |
|
667 if (max_xsize > BUF_SIZE) { |
|
668 pbuff = mlib_malloc(max_xsize*sizeof(mlib_d64)); |
|
669 |
|
670 if (pbuff == NULL) return MLIB_FAILURE; |
|
671 } |
|
672 |
|
673 dX = (dX - (dX >> 31)) &~ 1; /* rounding towards ZERO */ |
|
674 dY = (dY - (dY >> 31)) &~ 1; /* rounding towards ZERO */ |
|
675 dx64 = vis_to_double_dup((((dX >> 1) & 0xFFFF) << 16) | ((dX >> 1) & 0xFFFF)); |
|
676 dy64 = vis_to_double_dup((((dY >> 1) & 0xFFFF) << 16) | ((dY >> 1) & 0xFFFF)); |
|
677 |
|
678 for (j = yStart; j <= yFinish; j++) { |
|
679 mlib_u8 *sp; |
|
680 mlib_d64 *sp0, *sp1; |
|
681 |
|
682 NEW_LINE(4); |
|
683 |
|
684 deltax = DOUBLE_4U16(X, X, X, X); |
|
685 deltay = DOUBLE_4U16(Y, Y, Y, Y); |
|
686 |
|
687 #pragma pipeloop(0) |
|
688 for (i = 0; i < size; i++) { |
|
689 sp = *(mlib_u8**)((mlib_u8*)lineAddr + PTR_SHIFT(Y)) + 8*(X >> MLIB_SHIFT); |
|
690 |
|
691 vis_alignaddr(sp, 0); |
|
692 sp0 = AL_ADDR(sp, 0); |
|
693 s0 = vis_faligndata(sp0[0], sp0[1]); |
|
694 s1 = vis_faligndata(sp0[1], sp0[2]); |
|
695 |
|
696 vis_alignaddr(sp, srcYStride); |
|
697 sp1 = AL_ADDR(sp, srcYStride); |
|
698 s2 = vis_faligndata(sp1[0], sp1[1]); |
|
699 s3 = vis_faligndata(sp1[1], sp1[2]); |
|
700 |
|
701 BL_SUM(); |
|
702 |
|
703 pbuff[i] = dd; |
|
704 X += dX; |
|
705 Y += dY; |
|
706 } |
|
707 |
|
708 mlib_ImageCopy_na((mlib_u8*)pbuff, dl, 8*size); |
|
709 } |
|
710 |
|
711 if (pbuff != buff) { |
|
712 mlib_free(pbuff); |
|
713 } |
|
714 |
|
715 return MLIB_SUCCESS; |
|
716 } |
|
717 |
|
718 /***************************************************************/ |
|
719 #define LUT(x) plut[x] |
|
720 |
|
721 mlib_status FUN_NAME(s16_i)(mlib_affine_param *param, |
|
722 const void *colormap) |
|
723 { |
|
724 DECLAREVAR(); |
|
725 mlib_s32 nchan = mlib_ImageGetLutChannels(colormap); |
|
726 mlib_s32 lut_off = mlib_ImageGetLutOffset(colormap); |
|
727 mlib_d64 *plut = (mlib_d64*)mlib_ImageGetLutNormalTable(colormap) - lut_off; |
|
728 mlib_s32 max_xsize = param -> max_xsize; |
|
729 mlib_d64 buff[BUF_SIZE], *pbuff = buff; |
|
730 |
|
731 srcYStride /= sizeof(DTYPE); |
|
732 |
|
733 if (max_xsize > BUF_SIZE) { |
|
734 pbuff = mlib_malloc(max_xsize*sizeof(mlib_d64)); |
|
735 |
|
736 if (pbuff == NULL) return MLIB_FAILURE; |
|
737 } |
|
738 |
|
739 dX = (dX - (dX >> 31)) &~ 1; /* rounding towards ZERO */ |
|
740 dY = (dY - (dY >> 31)) &~ 1; /* rounding towards ZERO */ |
|
741 dx64 = vis_to_double_dup((((dX >> 1) & 0xFFFF) << 16) | ((dX >> 1) & 0xFFFF)); |
|
742 dy64 = vis_to_double_dup((((dY >> 1) & 0xFFFF) << 16) | ((dY >> 1) & 0xFFFF)); |
|
743 |
|
744 for (j = yStart; j <= yFinish; j++) { |
|
745 DTYPE *sp; |
|
746 |
|
747 NEW_LINE(1); |
|
748 |
|
749 deltax = DOUBLE_4U16(X, X, X, X); |
|
750 deltay = DOUBLE_4U16(Y, Y, Y, Y); |
|
751 |
|
752 #pragma pipeloop(0) |
|
753 for (i = 0; i < size; i++) { |
|
754 sp = *(DTYPE**)((mlib_u8*)lineAddr + PTR_SHIFT(Y)) + (X >> MLIB_SHIFT); |
|
755 s0 = LUT(sp[0]); |
|
756 s1 = LUT(sp[1]); |
|
757 s2 = LUT(sp[srcYStride]); |
|
758 s3 = LUT(sp[srcYStride + 1]); |
|
759 |
|
760 BL_SUM(); |
|
761 |
|
762 pbuff[i] = dd; |
|
763 X += dX; |
|
764 Y += dY; |
|
765 } |
|
766 |
|
767 if (nchan == 3) { |
|
768 mlib_ImageColorTrue2IndexLine_S16_S16_3_in_4((void*)pbuff, (void*)dl, size, colormap); |
|
769 } else { |
|
770 mlib_ImageColorTrue2IndexLine_S16_S16_4((void*)pbuff, (void*)dl, size, colormap); |
|
771 } |
|
772 } |
|
773 |
|
774 if (pbuff != buff) { |
|
775 mlib_free(pbuff); |
|
776 } |
|
777 |
|
778 return MLIB_SUCCESS; |
|
779 } |
|
780 |
|
781 /***************************************************************/ |
|
782 #undef DTYPE |
|
783 #define DTYPE mlib_u8 |
|
784 |
|
785 mlib_status FUN_NAME(u8_i)(mlib_affine_param *param, |
|
786 const void *colormap) |
|
787 { |
|
788 DECLAREVAR(); |
|
789 mlib_s32 nchan = mlib_ImageGetLutChannels(colormap); |
|
790 mlib_s32 lut_off = mlib_ImageGetLutOffset(colormap); |
|
791 mlib_d64 *plut = (mlib_d64*)mlib_ImageGetLutNormalTable(colormap) - lut_off; |
|
792 mlib_s32 max_xsize = param -> max_xsize; |
|
793 mlib_d64 buff[BUF_SIZE], *pbuff = buff; |
|
794 |
|
795 if (max_xsize > BUF_SIZE) { |
|
796 pbuff = mlib_malloc(max_xsize*sizeof(mlib_d64)); |
|
797 |
|
798 if (pbuff == NULL) return MLIB_FAILURE; |
|
799 } |
|
800 |
|
801 dX = (dX - (dX >> 31)) &~ 1; /* rounding towards ZERO */ |
|
802 dY = (dY - (dY >> 31)) &~ 1; /* rounding towards ZERO */ |
|
803 dx64 = vis_to_double_dup((((dX >> 1) & 0xFFFF) << 16) | ((dX >> 1) & 0xFFFF)); |
|
804 dy64 = vis_to_double_dup((((dY >> 1) & 0xFFFF) << 16) | ((dY >> 1) & 0xFFFF)); |
|
805 |
|
806 for (j = yStart; j <= yFinish; j++) { |
|
807 DTYPE *sp; |
|
808 |
|
809 NEW_LINE(1); |
|
810 |
|
811 deltax = DOUBLE_4U16(X, X, X, X); |
|
812 deltay = DOUBLE_4U16(Y, Y, Y, Y); |
|
813 |
|
814 #pragma pipeloop(0) |
|
815 for (i = 0; i < size; i++) { |
|
816 sp = *(DTYPE**)((mlib_u8*)lineAddr + PTR_SHIFT(Y)) + (X >> MLIB_SHIFT); |
|
817 s0 = LUT(sp[0]); |
|
818 s1 = LUT(sp[1]); |
|
819 s2 = LUT(sp[srcYStride]); |
|
820 s3 = LUT(sp[srcYStride + 1]); |
|
821 |
|
822 BL_SUM(); |
|
823 |
|
824 pbuff[i] = dd; |
|
825 X += dX; |
|
826 Y += dY; |
|
827 } |
|
828 |
|
829 if (nchan == 3) { |
|
830 mlib_ImageColorTrue2IndexLine_S16_U8_3_in_4((void*)pbuff, (void*)dl, size, colormap); |
|
831 } else { |
|
832 mlib_ImageColorTrue2IndexLine_S16_U8_4((void*)pbuff, (void*)dl, size, colormap); |
|
833 } |
|
834 } |
|
835 |
|
836 if (pbuff != buff) { |
|
837 mlib_free(pbuff); |
|
838 } |
|
839 |
|
840 return MLIB_SUCCESS; |
|
841 } |
|
842 |
|
843 /***************************************************************/ |