0N/A/*
2362N/A * Copyright (c) 1998, 2003, Oracle and/or its affiliates. All rights reserved.
0N/A * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
0N/A *
0N/A * This code is free software; you can redistribute it and/or modify it
0N/A * under the terms of the GNU General Public License version 2 only, as
2362N/A * published by the Free Software Foundation. Oracle designates this
0N/A * particular file as subject to the "Classpath" exception as provided
2362N/A * by Oracle in the LICENSE file that accompanied this code.
0N/A *
0N/A * This code is distributed in the hope that it will be useful, but WITHOUT
0N/A * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
0N/A * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
0N/A * version 2 for more details (a copy is included in the LICENSE file that
0N/A * accompanied this code).
0N/A *
0N/A * You should have received a copy of the GNU General Public License version
0N/A * 2 along with this work; if not, write to the Free Software Foundation,
0N/A * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
0N/A *
2362N/A * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
2362N/A * or visit www.oracle.com if you need additional information or have any
2362N/A * questions.
0N/A */
0N/A
0N/A
0N/A
0N/A/*
0N/A * The functions step along the lines from xLeft to xRight and apply
0N/A * the bilinear filtering.
0N/A *
0N/A */
0N/A
0N/A#include "vis_proto.h"
0N/A#include "mlib_image.h"
0N/A#include "mlib_ImageColormap.h"
0N/A#include "mlib_ImageCopy.h"
0N/A#include "mlib_ImageAffine.h"
0N/A#include "mlib_v_ImageFilters.h"
0N/A#include "mlib_v_ImageChannelExtract.h"
0N/A
0N/A/***************************************************************/
0N/A/*#define MLIB_VIS2*/
0N/A
0N/A/***************************************************************/
0N/A#define DTYPE mlib_u8
0N/A
0N/A#define FUN_NAME(CHAN) mlib_ImageAffine_u8_##CHAN##_bl
0N/A
0N/A/***************************************************************/
0N/Astatic mlib_status FUN_NAME(2ch_na)(mlib_affine_param *param);
0N/Astatic mlib_status FUN_NAME(4ch_na)(mlib_affine_param *param);
0N/A
0N/A/***************************************************************/
0N/A#ifdef MLIB_VIS2
0N/A#define MLIB_WRITE_BMASK(bmask) vis_write_bmask(bmask, 0)
0N/A#else
0N/A#define MLIB_WRITE_BMASK(bmask)
0N/A#endif /* MLIB_VIS2 */
0N/A
0N/A/***************************************************************/
0N/A#define FILTER_BITS 8
0N/A
0N/A/***************************************************************/
0N/A#undef DECLAREVAR
0N/A#define DECLAREVAR() \
0N/A DECLAREVAR0(); \
0N/A mlib_s32 *warp_tbl = param -> warp_tbl; \
0N/A mlib_s32 srcYStride = param -> srcYStride; \
0N/A mlib_u8 *dl; \
0N/A mlib_s32 i, size; \
0N/A mlib_d64 k05 = vis_to_double_dup(0x00080008); \
0N/A mlib_d64 d0, d1, d2, d3, dd
0N/A
0N/A/***************************************************************/
0N/A#define FMUL_16x16(x, y) \
0N/A vis_fpadd16(vis_fmul8sux16(x, y), vis_fmul8ulx16(x, y))
0N/A
0N/A/***************************************************************/
0N/A#define BUF_SIZE 512
0N/A
0N/A/***************************************************************/
0N/Aconst mlib_u32 mlib_fmask_arr[] = {
0N/A 0x00000000, 0x000000FF, 0x0000FF00, 0x0000FFFF,
0N/A 0x00FF0000, 0x00FF00FF, 0x00FFFF00, 0x00FFFFFF,
0N/A 0xFF000000, 0xFF0000FF, 0xFF00FF00, 0xFF00FFFF,
0N/A 0xFFFF0000, 0xFFFF00FF, 0xFFFFFF00, 0xFFFFFFFF
0N/A};
0N/A
0N/A/***************************************************************/
0N/A#define DOUBLE_4U16(x0, x1, x2, x3) \
0N/A vis_to_double((((x0 & 0xFFFE) << 15) | ((x1 & 0xFFFE) >> 1)), \
0N/A (((x2 & 0xFFFE) << 15) | ((x3 & 0xFFFE) >> 1)))
0N/A
0N/A/***************************************************************/
0N/A#define BL_SUM(HL) \
0N/A delta1_x = vis_fpsub16(mask_7fff, deltax); \
0N/A delta1_y = vis_fpsub16(mask_7fff, deltay); \
0N/A \
0N/A d0 = vis_fmul8x16(vis_read_##HL(s0), delta1_x); \
0N/A d1 = vis_fmul8x16(vis_read_##HL(s1), deltax); \
0N/A d0 = vis_fpadd16(d0, d1); \
0N/A d0 = FMUL_16x16(d0, delta1_y); \
0N/A d2 = vis_fmul8x16(vis_read_##HL(s2), delta1_x); \
0N/A d3 = vis_fmul8x16(vis_read_##HL(s3), deltax); \
0N/A d2 = vis_fpadd16(d2, d3); \
0N/A d2 = FMUL_16x16(d2, deltay); \
0N/A dd = vis_fpadd16(d0, d2); \
0N/A dd = vis_fpadd16(dd, k05); \
0N/A df = vis_fpack16(dd); \
0N/A \
0N/A deltax = vis_fpadd16(deltax, dx64); \
0N/A deltay = vis_fpadd16(deltay, dy64); \
0N/A deltax = vis_fand(deltax, mask_7fff); \
0N/A deltay = vis_fand(deltay, mask_7fff)
0N/A
0N/A/***************************************************************/
0N/A#define GET_FILTER_XY() \
0N/A mlib_d64 filterx, filtery, filterxy; \
0N/A mlib_s32 filterpos; \
0N/A filterpos = (X >> FILTER_SHIFT) & FILTER_MASK; \
0N/A filterx = *((mlib_d64 *) ((mlib_u8 *) mlib_filters_u8_bl + \
0N/A filterpos)); \
0N/A filterpos = (Y >> FILTER_SHIFT) & FILTER_MASK; \
0N/A filtery = *((mlib_d64 *) ((mlib_u8 *) mlib_filters_u8_bl + \
0N/A filterpos + 8*FILTER_SIZE)); \
0N/A filterxy = FMUL_16x16(filterx, filtery)
0N/A
0N/A/***************************************************************/
0N/A#define LD_U8(sp, ind) vis_read_lo(vis_ld_u8(sp + ind))
0N/A#define LD_U16(sp, ind) vis_ld_u16(sp + ind)
0N/A
0N/A/***************************************************************/
0N/A#define LOAD_1CH() \
0N/A s0 = vis_fpmerge(LD_U8(sp0, 0), LD_U8(sp2, 0)); \
0N/A s1 = vis_fpmerge(LD_U8(sp0, 1), LD_U8(sp2, 1)); \
0N/A s2 = vis_fpmerge(LD_U8(sp0, srcYStride), LD_U8(sp2, srcYStride)); \
0N/A s3 = vis_fpmerge(LD_U8(sp0, srcYStride + 1), \
0N/A LD_U8(sp2, srcYStride + 1)); \
0N/A \
0N/A t0 = vis_fpmerge(LD_U8(sp1, 0), LD_U8(sp3, 0)); \
0N/A t1 = vis_fpmerge(LD_U8(sp1, 1), LD_U8(sp3, 1)); \
0N/A t2 = vis_fpmerge(LD_U8(sp1, srcYStride), LD_U8(sp3, srcYStride)); \
0N/A t3 = vis_fpmerge(LD_U8(sp1, srcYStride + 1), \
0N/A LD_U8(sp3, srcYStride + 1)); \
0N/A \
0N/A s0 = vis_fpmerge(vis_read_lo(s0), vis_read_lo(t0)); \
0N/A s1 = vis_fpmerge(vis_read_lo(s1), vis_read_lo(t1)); \
0N/A s2 = vis_fpmerge(vis_read_lo(s2), vis_read_lo(t2)); \
0N/A s3 = vis_fpmerge(vis_read_lo(s3), vis_read_lo(t3))
0N/A
0N/A/***************************************************************/
0N/A#define GET_POINTER(sp) \
0N/A sp = *(mlib_u8**)((mlib_u8*)lineAddr + PTR_SHIFT(Y)) + \
0N/A (X >> MLIB_SHIFT); \
0N/A X += dX; \
0N/A Y += dY
0N/A
0N/A/***************************************************************/
0N/A#undef PREPARE_DELTAS
0N/A#define PREPARE_DELTAS \
0N/A if (warp_tbl != NULL) { \
0N/A dX = warp_tbl[2*j ]; \
0N/A dY = warp_tbl[2*j + 1]; \
0N/A dx64 = vis_to_double_dup((((dX << 1) & 0xFFFF) << 16) | ((dX << 1) & 0xFFFF)); \
0N/A dy64 = vis_to_double_dup((((dY << 1) & 0xFFFF) << 16) | ((dY << 1) & 0xFFFF)); \
0N/A }
0N/A
0N/A/***************************************************************/
0N/Amlib_status FUN_NAME(1ch)(mlib_affine_param *param)
0N/A{
0N/A DECLAREVAR();
0N/A mlib_d64 mask_7fff = vis_to_double_dup(0x7FFF7FFF);
0N/A mlib_d64 dx64, dy64, deltax, deltay, delta1_x, delta1_y;
0N/A mlib_s32 off, x0, x1, x2, x3, y0, y1, y2, y3;
0N/A mlib_f32 *dp, fmask;
0N/A
0N/A vis_write_gsr((1 << 3) | 7);
0N/A
0N/A dx64 = vis_to_double_dup((((dX << 1) & 0xFFFF) << 16) | ((dX << 1) & 0xFFFF));
0N/A dy64 = vis_to_double_dup((((dY << 1) & 0xFFFF) << 16) | ((dY << 1) & 0xFFFF));
0N/A
0N/A for (j = yStart; j <= yFinish; j++) {
0N/A mlib_u8 *sp0, *sp1, *sp2, *sp3;
0N/A mlib_d64 s0, s1, s2, s3, t0, t1, t2, t3;
0N/A mlib_f32 df;
0N/A
0N/A NEW_LINE(1);
0N/A
0N/A off = (mlib_s32)dl & 3;
0N/A dp = (mlib_f32*)(dl - off);
0N/A
0N/A x0 = X - off*dX; y0 = Y - off*dY;
0N/A x1 = x0 + dX; y1 = y0 + dY;
0N/A x2 = x1 + dX; y2 = y1 + dY;
0N/A x3 = x2 + dX; y3 = y2 + dY;
0N/A
0N/A deltax = DOUBLE_4U16(x0, x1, x2, x3);
0N/A deltay = DOUBLE_4U16(y0, y1, y2, y3);
0N/A
0N/A if (off) {
0N/A mlib_s32 emask = vis_edge16((void*)(2*off), (void*)(2*(off + size - 1)));
0N/A
0N/A off = 4 - off;
0N/A GET_POINTER(sp3);
0N/A sp0 = sp1 = sp2 = sp3;
0N/A
0N/A if (off > 1 && size > 1) {
0N/A GET_POINTER(sp3);
0N/A }
0N/A
0N/A if (off > 2) {
0N/A sp2 = sp3;
0N/A
0N/A if (size > 2) {
0N/A GET_POINTER(sp3);
0N/A }
0N/A }
0N/A
0N/A LOAD_1CH();
0N/A BL_SUM(lo);
0N/A
0N/A fmask = ((mlib_f32*)mlib_fmask_arr)[emask];
0N/A *dp++ = vis_fors(vis_fands(fmask, df), vis_fandnots(fmask, dp[0]));
0N/A
0N/A size -= off;
0N/A
0N/A if (size < 0) size = 0;
0N/A }
0N/A
0N/A#pragma pipeloop(0)
0N/A for (i = 0; i < size/4; i++) {
0N/A GET_POINTER(sp0);
0N/A GET_POINTER(sp1);
0N/A GET_POINTER(sp2);
0N/A GET_POINTER(sp3);
0N/A
0N/A LOAD_1CH();
0N/A BL_SUM(lo);
0N/A
0N/A dp[i] = df;
0N/A }
0N/A
0N/A off = size & 3;
0N/A
0N/A if (off) {
0N/A GET_POINTER(sp0);
0N/A sp1 = sp2 = sp3 = sp0;
0N/A
0N/A if (off > 1) {
0N/A GET_POINTER(sp1);
0N/A }
0N/A
0N/A if (off > 2) {
0N/A GET_POINTER(sp2);
0N/A }
0N/A
0N/A LOAD_1CH();
0N/A BL_SUM(lo);
0N/A
0N/A fmask = ((mlib_f32*)mlib_fmask_arr)[(0xF0 >> off) & 0x0F];
0N/A dp[i] = vis_fors(vis_fands(fmask, df), vis_fandnots(fmask, dp[i]));
0N/A }
0N/A }
0N/A
0N/A return MLIB_SUCCESS;
0N/A}
0N/A
0N/A/***************************************************************/
0N/A#undef GET_POINTER
0N/A#define GET_POINTER(sp) \
0N/A sp = *(mlib_u8**)((mlib_u8*)lineAddr + PTR_SHIFT(Y)) + \
0N/A 2*(X >> MLIB_SHIFT); \
0N/A X += dX; \
0N/A Y += dY
0N/A
0N/A/***************************************************************/
0N/A#ifndef MLIB_VIS2
0N/A
0N/A#define LOAD_2CH() \
0N/A s0 = vis_faligndata(LD_U16(sp1, 0), k05); \
0N/A s1 = vis_faligndata(LD_U16(sp1, 2), k05); \
0N/A s2 = vis_faligndata(LD_U16(sp1, srcYStride), k05); \
0N/A s3 = vis_faligndata(LD_U16(sp1, srcYStride + 2), k05); \
0N/A \
0N/A s0 = vis_faligndata(LD_U16(sp0, 0), s0); \
0N/A s1 = vis_faligndata(LD_U16(sp0, 2), s1); \
0N/A s2 = vis_faligndata(LD_U16(sp0, srcYStride), s2); \
0N/A s3 = vis_faligndata(LD_U16(sp0, srcYStride + 2), s3)
0N/A
0N/A#define BL_SUM_2CH() BL_SUM(hi)
0N/A
0N/A#else
0N/A
0N/A#define LOAD_2CH() \
0N/A s0 = vis_bshuffle(LD_U16(sp0, 0), LD_U16(sp1, 0)); \
0N/A s1 = vis_bshuffle(LD_U16(sp0, 2), LD_U16(sp1, 2)); \
0N/A s2 = vis_bshuffle(LD_U16(sp0, srcYStride), \
0N/A LD_U16(sp1, srcYStride)); \
0N/A s3 = vis_bshuffle(LD_U16(sp0, srcYStride + 2), \
0N/A LD_U16(sp1, srcYStride + 2))
0N/A
0N/A#define BL_SUM_2CH() BL_SUM(lo)
0N/A
0N/A#endif /* MLIB_VIS2 */
0N/A
0N/A/***************************************************************/
0N/A#undef PREPARE_DELTAS
0N/A#define PREPARE_DELTAS \
0N/A if (warp_tbl != NULL) { \
0N/A dX = warp_tbl[2*j ]; \
0N/A dY = warp_tbl[2*j + 1]; \
0N/A dx64 = vis_to_double_dup(((dX & 0xFFFF) << 16) | (dX & 0xFFFF)); \
0N/A dy64 = vis_to_double_dup(((dY & 0xFFFF) << 16) | (dY & 0xFFFF)); \
0N/A }
0N/A
0N/A/***************************************************************/
0N/Amlib_status FUN_NAME(2ch)(mlib_affine_param *param)
0N/A{
0N/A DECLAREVAR();
0N/A mlib_d64 mask_7fff = vis_to_double_dup(0x7FFF7FFF);
0N/A mlib_d64 dx64, dy64, deltax, deltay, delta1_x, delta1_y;
0N/A mlib_s32 off, x0, x1, y0, y1;
0N/A
0N/A if (((mlib_s32)lineAddr[0] | (mlib_s32)dstData | srcYStride | dstYStride) & 1) {
0N/A return FUN_NAME(2ch_na)(param);
0N/A }
0N/A
0N/A vis_write_gsr((1 << 3) | 6);
0N/A MLIB_WRITE_BMASK(0x45cd67ef);
0N/A
0N/A dx64 = vis_to_double_dup(((dX & 0xFFFF) << 16) | (dX & 0xFFFF));
0N/A dy64 = vis_to_double_dup(((dY & 0xFFFF) << 16) | (dY & 0xFFFF));
0N/A
0N/A for (j = yStart; j <= yFinish; j++) {
0N/A mlib_u8 *sp0, *sp1;
0N/A mlib_d64 s0, s1, s2, s3;
0N/A mlib_f32 *dp, df, fmask;
0N/A
0N/A NEW_LINE(2);
0N/A
0N/A off = (mlib_s32)dl & 3;
0N/A dp = (mlib_f32*)(dl - off);
0N/A
0N/A if (off) {
0N/A x0 = X - dX; y0 = Y - dY;
0N/A x1 = X; y1 = Y;
0N/A } else {
0N/A x0 = X; y0 = Y;
0N/A x1 = X + dX; y1 = Y + dY;
0N/A }
0N/A
0N/A deltax = DOUBLE_4U16(x0, x0, x1, x1);
0N/A deltay = DOUBLE_4U16(y0, y0, y1, y1);
0N/A
0N/A if (off) {
0N/A GET_POINTER(sp1);
0N/A sp0 = sp1;
0N/A LOAD_2CH();
0N/A
0N/A BL_SUM_2CH();
0N/A
0N/A fmask = ((mlib_f32*)mlib_fmask_arr)[0x3];
0N/A *dp++ = vis_fors(vis_fands(fmask, df), vis_fandnots(fmask, dp[0]));
0N/A
0N/A size--;
0N/A }
0N/A
0N/A if (size >= 2) {
0N/A GET_POINTER(sp0);
0N/A GET_POINTER(sp1);
0N/A LOAD_2CH();
0N/A
0N/A#pragma pipeloop(0)
0N/A for (i = 0; i < (size - 2)/2; i++) {
0N/A BL_SUM_2CH();
0N/A
0N/A GET_POINTER(sp0);
0N/A GET_POINTER(sp1);
0N/A LOAD_2CH();
0N/A
0N/A *dp++ = df;
0N/A }
0N/A
0N/A BL_SUM_2CH();
0N/A *dp++ = df;
0N/A }
0N/A
0N/A if (size & 1) {
0N/A GET_POINTER(sp0);
0N/A sp1 = sp0;
0N/A LOAD_2CH();
0N/A
0N/A BL_SUM_2CH();
0N/A
0N/A fmask = ((mlib_f32*)mlib_fmask_arr)[0x0C];
0N/A *dp = vis_fors(vis_fands(fmask, df), vis_fandnots(fmask, *dp));
0N/A }
0N/A }
0N/A
0N/A return MLIB_SUCCESS;
0N/A}
0N/A
0N/A/***************************************************************/
0N/A#ifndef MLIB_VIS2
0N/A
0N/A#define LOAD_2CH_NA() \
0N/A s0 = vis_fpmerge(LD_U8(sp0, 0), LD_U8(sp1, 0)); \
0N/A s1 = vis_fpmerge(LD_U8(sp0, 2), LD_U8(sp1, 2)); \
0N/A s2 = vis_fpmerge(LD_U8(sp0, srcYStride), \
0N/A LD_U8(sp1, srcYStride)); \
0N/A s3 = vis_fpmerge(LD_U8(sp0, srcYStride + 2), \
0N/A LD_U8(sp1, srcYStride + 2)); \
0N/A \
0N/A t0 = vis_fpmerge(LD_U8(sp0, 1), LD_U8(sp1, 1)); \
0N/A t1 = vis_fpmerge(LD_U8(sp0, 3), LD_U8(sp1, 3)); \
0N/A t2 = vis_fpmerge(LD_U8(sp0, srcYStride + 1), \
0N/A LD_U8(sp1, srcYStride + 1)); \
0N/A t3 = vis_fpmerge(LD_U8(sp0, srcYStride + 3), \
0N/A LD_U8(sp1, srcYStride + 3)); \
0N/A \
0N/A s0 = vis_fpmerge(vis_read_lo(s0), vis_read_lo(t0)); \
0N/A s1 = vis_fpmerge(vis_read_lo(s1), vis_read_lo(t1)); \
0N/A s2 = vis_fpmerge(vis_read_lo(s2), vis_read_lo(t2)); \
0N/A s3 = vis_fpmerge(vis_read_lo(s3), vis_read_lo(t3))
0N/A
0N/A#define BL_SUM_2CH_NA() BL_SUM(lo)
0N/A
0N/A#else
0N/A
0N/A#define LOAD_2CH_NA() \
0N/A vis_alignaddr(sp0, 0); \
0N/A spa = AL_ADDR(sp0, 0); \
0N/A s0 = vis_faligndata(spa[0], spa[1]); \
0N/A \
0N/A vis_alignaddr(sp1, 0); \
0N/A spa = AL_ADDR(sp1, 0); \
0N/A s1 = vis_faligndata(spa[0], spa[1]); \
0N/A \
0N/A vis_alignaddr(sp0, srcYStride); \
0N/A spa = AL_ADDR(sp0, srcYStride); \
0N/A s2 = vis_faligndata(spa[0], spa[1]); \
0N/A \
0N/A vis_alignaddr(sp1, srcYStride); \
0N/A spa = AL_ADDR(sp1, srcYStride); \
0N/A s3 = vis_faligndata(spa[0], spa[1]); \
0N/A \
0N/A s0 = vis_bshuffle(s0, s1); \
0N/A s2 = vis_bshuffle(s2, s3)
0N/A
0N/A#define BL_SUM_2CH_NA() \
0N/A delta1_x = vis_fpsub16(mask_7fff, deltax); \
0N/A delta1_y = vis_fpsub16(mask_7fff, deltay); \
0N/A \
0N/A d0 = vis_fmul8x16(vis_read_hi(s0), delta1_x); \
0N/A d1 = vis_fmul8x16(vis_read_lo(s0), deltax); \
0N/A d0 = vis_fpadd16(d0, d1); \
0N/A d0 = FMUL_16x16(d0, delta1_y); \
0N/A d2 = vis_fmul8x16(vis_read_hi(s2), delta1_x); \
0N/A d3 = vis_fmul8x16(vis_read_lo(s2), deltax); \
0N/A d2 = vis_fpadd16(d2, d3); \
0N/A d2 = FMUL_16x16(d2, deltay); \
0N/A dd = vis_fpadd16(d0, d2); \
0N/A dd = vis_fpadd16(dd, k05); \
0N/A df = vis_fpack16(dd); \
0N/A \
0N/A deltax = vis_fpadd16(deltax, dx64); \
0N/A deltay = vis_fpadd16(deltay, dy64); \
0N/A deltax = vis_fand(deltax, mask_7fff); \
0N/A deltay = vis_fand(deltay, mask_7fff)
0N/A
0N/A#endif /* MLIB_VIS2 */
0N/A
0N/A/***************************************************************/
0N/Amlib_status FUN_NAME(2ch_na)(mlib_affine_param *param)
0N/A{
0N/A DECLAREVAR();
0N/A mlib_d64 mask_7fff = vis_to_double_dup(0x7FFF7FFF);
0N/A mlib_d64 dx64, dy64, deltax, deltay, delta1_x, delta1_y;
0N/A mlib_s32 max_xsize = param -> max_xsize, bsize;
0N/A mlib_s32 x0, x1, y0, y1;
0N/A mlib_f32 buff[BUF_SIZE], *pbuff = buff;
0N/A
0N/A bsize = (max_xsize + 1)/2;
0N/A
0N/A if (bsize > BUF_SIZE) {
0N/A pbuff = mlib_malloc(bsize*sizeof(mlib_f32));
0N/A
0N/A if (pbuff == NULL) return MLIB_FAILURE;
0N/A }
0N/A
0N/A vis_write_gsr((1 << 3) | 6);
0N/A MLIB_WRITE_BMASK(0x018923AB);
0N/A
0N/A dx64 = vis_to_double_dup(((dX & 0xFFFF) << 16) | (dX & 0xFFFF));
0N/A dy64 = vis_to_double_dup(((dY & 0xFFFF) << 16) | (dY & 0xFFFF));
0N/A
0N/A for (j = yStart; j <= yFinish; j++) {
0N/A mlib_u8 *sp0, *sp1;
0N/A mlib_d64 s0, s1, s2, s3;
0N/A#ifndef MLIB_VIS2
0N/A mlib_d64 t0, t1, t2, t3;
0N/A#else
0N/A mlib_d64 *spa;
0N/A#endif /* MLIB_VIS2 */
0N/A mlib_f32 *dp, df;
0N/A
0N/A NEW_LINE(2);
0N/A
0N/A dp = pbuff;
0N/A
0N/A x0 = X; y0 = Y;
0N/A x1 = X + dX; y1 = Y + dY;
0N/A
0N/A deltax = DOUBLE_4U16(x0, x0, x1, x1);
0N/A deltay = DOUBLE_4U16(y0, y0, y1, y1);
0N/A
0N/A#pragma pipeloop(0)
0N/A for (i = 0; i < size/2; i++) {
0N/A GET_POINTER(sp0);
0N/A GET_POINTER(sp1);
0N/A LOAD_2CH_NA();
0N/A
0N/A BL_SUM_2CH_NA();
0N/A
0N/A *dp++ = df;
0N/A }
0N/A
0N/A if (size & 1) {
0N/A GET_POINTER(sp0);
0N/A sp1 = sp0;
0N/A LOAD_2CH_NA();
0N/A
0N/A BL_SUM_2CH_NA();
0N/A
0N/A *dp++ = df;
0N/A }
0N/A
0N/A mlib_ImageCopy_na((mlib_u8*)pbuff, dl, 2*size);
0N/A }
0N/A
0N/A if (pbuff != buff) {
0N/A mlib_free(pbuff);
0N/A }
0N/A
0N/A return MLIB_SUCCESS;
0N/A}
0N/A
0N/A/***************************************************************/
0N/A#undef PREPARE_DELTAS
0N/A#define PREPARE_DELTAS \
0N/A if (warp_tbl != NULL) { \
0N/A dX = warp_tbl[2*j ]; \
0N/A dY = warp_tbl[2*j + 1]; \
0N/A }
0N/A
0N/A/***************************************************************/
0N/Amlib_status FUN_NAME(3ch)(mlib_affine_param *param)
0N/A{
0N/A DECLAREVAR();
0N/A mlib_s32 max_xsize = param -> max_xsize;
0N/A mlib_f32 buff[BUF_SIZE], *pbuff = buff;
0N/A
0N/A if (max_xsize > BUF_SIZE) {
0N/A pbuff = mlib_malloc(max_xsize*sizeof(mlib_f32));
0N/A
0N/A if (pbuff == NULL) return MLIB_FAILURE;
0N/A }
0N/A
0N/A vis_write_gsr(3 << 3);
0N/A
0N/A for (j = yStart; j <= yFinish; j++) {
0N/A mlib_d64 *sp0, *sp1, s0, s1;
0N/A mlib_u8 *sp;
0N/A
0N/A NEW_LINE(3);
0N/A
0N/A#pragma pipeloop(0)
0N/A for (i = 0; i < size; i++) {
0N/A GET_FILTER_XY();
0N/A
0N/A sp = *(mlib_u8**)((mlib_u8*)lineAddr + PTR_SHIFT(Y)) + 3*(X >> MLIB_SHIFT) - 1;
0N/A
0N/A vis_alignaddr(sp, 0);
0N/A sp0 = AL_ADDR(sp, 0);
0N/A s0 = vis_faligndata(sp0[0], sp0[1]);
0N/A d0 = vis_fmul8x16au(vis_read_hi(s0), vis_read_hi(filterxy));
0N/A d1 = vis_fmul8x16al(vis_read_lo(s0), vis_read_hi(filterxy));
0N/A
0N/A vis_alignaddr(sp, srcYStride);
0N/A sp1 = AL_ADDR(sp, srcYStride);
0N/A s1 = vis_faligndata(sp1[0], sp1[1]);
0N/A d2 = vis_fmul8x16au(vis_read_hi(s1), vis_read_lo(filterxy));
0N/A d3 = vis_fmul8x16al(vis_read_lo(s1), vis_read_lo(filterxy));
0N/A
0N/A vis_alignaddr((void*)0, 2);
0N/A d0 = vis_fpadd16(d0, d2);
0N/A dd = vis_fpadd16(k05, d1);
0N/A dd = vis_fpadd16(dd, d3);
0N/A d0 = vis_faligndata(d0, d0);
0N/A dd = vis_fpadd16(dd, d0);
0N/A
0N/A pbuff[i] = vis_fpack16(dd);
0N/A X += dX;
0N/A Y += dY;
0N/A }
0N/A
0N/A mlib_v_ImageChannelExtract_U8_43L_D1((mlib_u8*)pbuff, dl, size);
0N/A }
0N/A
0N/A if (pbuff != buff) {
0N/A mlib_free(pbuff);
0N/A }
0N/A
0N/A return MLIB_SUCCESS;
0N/A}
0N/A
0N/A/***************************************************************/
0N/A#define PROCESS_4CH(s0, s1, s2, s3) \
0N/A d0 = vis_fmul8x16au(s0, vis_read_hi(filterxy)); \
0N/A d1 = vis_fmul8x16al(s1, vis_read_hi(filterxy)); \
0N/A d2 = vis_fmul8x16au(s2, vis_read_lo(filterxy)); \
0N/A d3 = vis_fmul8x16al(s3, vis_read_lo(filterxy)); \
0N/A \
0N/A dd = vis_fpadd16(d0, k05); \
0N/A d1 = vis_fpadd16(d1, d2); \
0N/A dd = vis_fpadd16(dd, d3); \
0N/A dd = vis_fpadd16(dd, d1)
0N/A
0N/A/***************************************************************/
0N/Amlib_status FUN_NAME(4ch)(mlib_affine_param *param)
0N/A{
0N/A DECLAREVAR();
0N/A
0N/A if (((mlib_s32)lineAddr[0] | (mlib_s32)dstData | srcYStride | dstYStride) & 3) {
0N/A return FUN_NAME(4ch_na)(param);
0N/A }
0N/A
0N/A vis_write_gsr(3 << 3);
0N/A
0N/A srcYStride >>= 2;
0N/A
0N/A for (j = yStart; j <= yFinish; j++) {
0N/A mlib_f32 *sp, s0, s1, s2, s3;
0N/A
0N/A NEW_LINE(4);
0N/A
0N/A#pragma pipeloop(0)
0N/A for (i = 0; i < size; i++) {
0N/A GET_FILTER_XY();
0N/A
0N/A sp = *(mlib_f32**)((mlib_u8*)lineAddr + PTR_SHIFT(Y)) + (X >> MLIB_SHIFT);
0N/A s0 = sp[0];
0N/A s1 = sp[1];
0N/A s2 = sp[srcYStride];
0N/A s3 = sp[srcYStride + 1];
0N/A
0N/A PROCESS_4CH(s0, s1, s2, s3);
0N/A
0N/A ((mlib_f32*)dl)[i] = vis_fpack16(dd);
0N/A X += dX;
0N/A Y += dY;
0N/A }
0N/A }
0N/A
0N/A return MLIB_SUCCESS;
0N/A}
0N/A
0N/A/***************************************************************/
0N/Amlib_status FUN_NAME(4ch_na)(mlib_affine_param *param)
0N/A{
0N/A DECLAREVAR();
0N/A mlib_s32 max_xsize = param -> max_xsize;
0N/A mlib_f32 buff[BUF_SIZE], *pbuff = buff;
0N/A
0N/A if (max_xsize > BUF_SIZE) {
0N/A pbuff = mlib_malloc(max_xsize*sizeof(mlib_f32));
0N/A
0N/A if (pbuff == NULL) return MLIB_FAILURE;
0N/A }
0N/A
0N/A vis_write_gsr(3 << 3);
0N/A
0N/A for (j = yStart; j <= yFinish; j++) {
0N/A mlib_d64 *sp0, *sp1, s0, s1;
0N/A mlib_u8 *sp;
0N/A
0N/A NEW_LINE(4);
0N/A
0N/A#pragma pipeloop(0)
0N/A for (i = 0; i < size; i++) {
0N/A GET_FILTER_XY();
0N/A
0N/A sp = *(mlib_u8**)((mlib_u8*)lineAddr + PTR_SHIFT(Y)) + 4*(X >> MLIB_SHIFT);
0N/A
0N/A vis_alignaddr(sp, 0);
0N/A sp0 = AL_ADDR(sp, 0);
0N/A s0 = vis_faligndata(sp0[0], sp0[1]);
0N/A
0N/A vis_alignaddr(sp, srcYStride);
0N/A sp1 = AL_ADDR(sp, srcYStride);
0N/A s1 = vis_faligndata(sp1[0], sp1[1]);
0N/A
0N/A PROCESS_4CH(vis_read_hi(s0), vis_read_lo(s0), vis_read_hi(s1), vis_read_lo(s1));
0N/A
0N/A pbuff[i] = vis_fpack16(dd);
0N/A X += dX;
0N/A Y += dY;
0N/A }
0N/A
0N/A mlib_ImageCopy_na((mlib_u8*)pbuff, dl, 4*size);
0N/A }
0N/A
0N/A if (pbuff != buff) {
0N/A mlib_free(pbuff);
0N/A }
0N/A
0N/A return MLIB_SUCCESS;
0N/A}
0N/A
0N/A/***************************************************************/
0N/A#define LUT(x) plut[x]
0N/A
0N/Amlib_status FUN_NAME(u8_i)(mlib_affine_param *param,
0N/A const void *colormap)
0N/A{
0N/A DECLAREVAR();
0N/A mlib_s32 nchan = mlib_ImageGetLutChannels(colormap);
0N/A mlib_s32 lut_off = mlib_ImageGetLutOffset(colormap);
0N/A mlib_f32 *plut = (mlib_f32*)mlib_ImageGetLutNormalTable(colormap) - lut_off;
0N/A mlib_s32 max_xsize = param -> max_xsize;
0N/A mlib_f32 buff[BUF_SIZE], *pbuff = buff;
0N/A
0N/A if (max_xsize > BUF_SIZE) {
0N/A pbuff = mlib_malloc(max_xsize*sizeof(mlib_f32));
0N/A
0N/A if (pbuff == NULL) return MLIB_FAILURE;
0N/A }
0N/A
0N/A vis_write_gsr(3 << 3);
0N/A
0N/A for (j = yStart; j <= yFinish; j++) {
0N/A mlib_f32 s0, s1, s2, s3;
0N/A DTYPE *sp;
0N/A
0N/A NEW_LINE(1);
0N/A
0N/A#pragma pipeloop(0)
0N/A for (i = 0; i < size; i++) {
0N/A GET_FILTER_XY();
0N/A
0N/A sp = *(DTYPE**)((mlib_u8*)lineAddr + PTR_SHIFT(Y)) + (X >> MLIB_SHIFT);
0N/A s0 = LUT(sp[0]);
0N/A s1 = LUT(sp[1]);
0N/A s2 = LUT(sp[srcYStride]);
0N/A s3 = LUT(sp[srcYStride + 1]);
0N/A
0N/A PROCESS_4CH(s0, s1, s2, s3);
0N/A
0N/A pbuff[i] = vis_fpack16(dd);
0N/A X += dX;
0N/A Y += dY;
0N/A }
0N/A
0N/A if (nchan == 3) {
0N/A mlib_ImageColorTrue2IndexLine_U8_U8_3_in_4((void*)pbuff, (void*)dl, size, colormap);
0N/A } else {
0N/A mlib_ImageColorTrue2IndexLine_U8_U8_4((void*)pbuff, (void*)dl, size, colormap);
0N/A }
0N/A }
0N/A
0N/A if (pbuff != buff) {
0N/A mlib_free(pbuff);
0N/A }
0N/A
0N/A return MLIB_SUCCESS;
0N/A}
0N/A
0N/A/***************************************************************/
0N/A#undef DTYPE
0N/A#define DTYPE mlib_s16
0N/A
0N/Amlib_status FUN_NAME(s16_i)(mlib_affine_param *param,
0N/A const void *colormap)
0N/A{
0N/A DECLAREVAR();
0N/A mlib_s32 nchan = mlib_ImageGetLutChannels(colormap);
0N/A mlib_s32 lut_off = mlib_ImageGetLutOffset(colormap);
0N/A mlib_f32 *plut = (mlib_f32*)mlib_ImageGetLutNormalTable(colormap) - lut_off;
0N/A mlib_s32 max_xsize = param -> max_xsize;
0N/A mlib_f32 buff[BUF_SIZE], *pbuff = buff;
0N/A
0N/A srcYStride /= sizeof(DTYPE);
0N/A
0N/A if (max_xsize > BUF_SIZE) {
0N/A pbuff = mlib_malloc(max_xsize*sizeof(mlib_f32));
0N/A
0N/A if (pbuff == NULL) return MLIB_FAILURE;
0N/A }
0N/A
0N/A vis_write_gsr(3 << 3);
0N/A
0N/A for (j = yStart; j <= yFinish; j++) {
0N/A mlib_f32 s0, s1, s2, s3;
0N/A DTYPE *sp;
0N/A
0N/A NEW_LINE(1);
0N/A
0N/A#pragma pipeloop(0)
0N/A for (i = 0; i < size; i++) {
0N/A GET_FILTER_XY();
0N/A
0N/A sp = *(DTYPE**)((mlib_u8*)lineAddr + PTR_SHIFT(Y)) + (X >> MLIB_SHIFT);
0N/A s0 = LUT(sp[0]);
0N/A s1 = LUT(sp[1]);
0N/A s2 = LUT(sp[srcYStride]);
0N/A s3 = LUT(sp[srcYStride + 1]);
0N/A
0N/A PROCESS_4CH(s0, s1, s2, s3);
0N/A
0N/A pbuff[i] = vis_fpack16(dd);
0N/A X += dX;
0N/A Y += dY;
0N/A }
0N/A
0N/A if (nchan == 3) {
0N/A mlib_ImageColorTrue2IndexLine_U8_S16_3_in_4((void*)pbuff, (void*)dl, size, colormap);
0N/A } else {
0N/A mlib_ImageColorTrue2IndexLine_U8_S16_4((void*)pbuff, (void*)dl, size, colormap);
0N/A }
0N/A }
0N/A
0N/A if (pbuff != buff) {
0N/A mlib_free(pbuff);
0N/A }
0N/A
0N/A return MLIB_SUCCESS;
0N/A}
0N/A
0N/A/***************************************************************/
0N/Aconst type_affine_i_fun mlib_AffineFunArr_bl_i[] = {
0N/A mlib_ImageAffine_u8_u8_i_bl,
0N/A mlib_ImageAffine_u8_u8_i_bl,
0N/A mlib_ImageAffine_u8_s16_i_bl,
0N/A mlib_ImageAffine_u8_s16_i_bl,
0N/A mlib_ImageAffine_s16_u8_i_bl,
0N/A mlib_ImageAffine_s16_u8_i_bl,
0N/A mlib_ImageAffine_s16_s16_i_bl,
0N/A mlib_ImageAffine_s16_s16_i_bl
0N/A};
0N/A
0N/A/***************************************************************/