JDK14/Java14源码在线阅读
/*
* Copyright (c) 2003, Oracle and/or its affiliates. All rights reserved.
* DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
*
* This code is free software; you can redistribute it and/or modify it
* under the terms of the GNU General Public License version 2 only, as
* published by the Free Software Foundation. Oracle designates this
* particular file as subject to the "Classpath" exception as provided
* by Oracle in the LICENSE file that accompanied this code.
*
* This code is distributed in the hope that it will be useful, but WITHOUT
* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
* FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
* version 2 for more details (a copy is included in the LICENSE file that
* accompanied this code).
*
* You should have received a copy of the GNU General Public License version
* 2 along with this work; if not, write to the Free Software Foundation,
* Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
*
* Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
* or visit www.oracle.com if you need additional information or have any
* questions.
*/
/*
* The functions step along the lines from xLeft to xRight and apply
* the bilinear filtering.
*
*/
#include "vis_proto.h"
#include "mlib_image.h"
#include "mlib_ImageCopy.h"
#include "mlib_ImageAffine.h"
#include "mlib_v_ImageFilters.h"
#include "mlib_v_ImageChannelExtract.h"
/*#define MLIB_VIS2*/
/***************************************************************/
#define DTYPE mlib_s16
#define FUN_NAME(CHAN) mlib_ImageAffine_s16_##CHAN##_bl
/***************************************************************/
static mlib_status FUN_NAME(2ch_na)(mlib_affine_param *param);
static mlib_status FUN_NAME(4ch_na)(mlib_affine_param *param);
/***************************************************************/
const mlib_u64 mlib_dmask_arr[] = {
0x0000000000000000, 0x000000000000FFFF, 0x00000000FFFF0000, 0x00000000FFFFFFFF,
0x0000FFFF00000000, 0x0000FFFF0000FFFF, 0x0000FFFFFFFF0000, 0x0000FFFFFFFFFFFF,
0xFFFF000000000000, 0xFFFF00000000FFFF, 0xFFFF0000FFFF0000, 0xFFFF0000FFFFFFFF,
0xFFFFFFFF00000000, 0xFFFFFFFF0000FFFF, 0xFFFFFFFFFFFF0000, 0xFFFFFFFFFFFFFFFF
};
/***************************************************************/
#define XOR_8000(x)
/***************************************************************/
#ifdef MLIB_VIS2
#define MLIB_WRITE_BMASK(bmask) vis_write_bmask(bmask, 0)
#else
#define MLIB_WRITE_BMASK(bmask)
#endif
/***************************************************************/
#undef DECLAREVAR
#define DECLAREVAR() \
DECLAREVAR0(); \
mlib_s32 *warp_tbl = param -> warp_tbl; \
mlib_s32 srcYStride = param -> srcYStride; \
mlib_u8 *dl; \
mlib_s32 i, size; \
/*mlib_d64 mask_8000 = vis_to_double_dup(0x80008000);*/ \
mlib_d64 mask_7fff = vis_to_double_dup(0x7FFF7FFF); \
mlib_d64 dx64, dy64, deltax, deltay, delta1_x, delta1_y; \
mlib_d64 s0, s1, s2, s3; \
mlib_d64 d0, d1, d2, d3, dd
/***************************************************************/
/* arguments (x, y) are swapped to prevent overflow */
#define FMUL_16x16(x, y) \
vis_fpadd16(vis_fmul8sux16(y, x), \
vis_fmul8ulx16(y, x))
/***************************************************************/
#define BUF_SIZE 512
/***************************************************************/
#define DOUBLE_4U16(x0, x1, x2, x3) \
vis_to_double(((((x0) & 0xFFFE) << 15) | (((x1) & 0xFFFE) >> 1)), \
((((x2) & 0xFFFE) << 15) | (((x3) & 0xFFFE) >> 1)))
/***************************************************************/
#define BL_SUM() \
XOR_8000(s0); \
XOR_8000(s1); \
XOR_8000(s2); \
XOR_8000(s3); \
\
delta1_x = vis_fpsub16(mask_7fff, deltax); \
delta1_y = vis_fpsub16(mask_7fff, deltay); \
\
d0 = FMUL_16x16(s0, delta1_x); \
d1 = FMUL_16x16(s1, deltax); \
d0 = vis_fpadd16(d0, d1); \
d0 = vis_fpadd16(d0, d0); \
d0 = FMUL_16x16(d0, delta1_y); \
\
d2 = FMUL_16x16(s2, delta1_x); \
d3 = FMUL_16x16(s3, deltax); \
d2 = vis_fpadd16(d2, d3); \
d2 = vis_fpadd16(d2, d2); \
d2 = FMUL_16x16(d2, deltay); \
\
dd = vis_fpadd16(d0, d2); \
dd = vis_fpadd16(dd, dd); \
XOR_8000(dd); \
\
deltax = vis_fpadd16(deltax, dx64); \
deltay = vis_fpadd16(deltay, dy64); \
deltax = vis_fand(deltax, mask_7fff); \
deltay = vis_fand(deltay, mask_7fff)
/***************************************************************/
#define BL_SUM_3CH() \
XOR_8000(s0); \
XOR_8000(s1); \
XOR_8000(s2); \
XOR_8000(s3); \
\
delta1_x = vis_fpsub16(mask_7fff, deltax); \
delta1_y = vis_fpsub16(mask_7fff, deltay); \
\
d0 = FMUL_16x16(s0, delta1_y); \
d2 = FMUL_16x16(s2, deltay); \
d0 = vis_fpadd16(d0, d2); \
d0 = vis_fpadd16(d0, d0); \
d0 = FMUL_16x16(d0, delta1_x); \
\
d1 = FMUL_16x16(s1, delta1_y); \
d3 = FMUL_16x16(s3, deltay); \
d1 = vis_fpadd16(d1, d3); \
d1 = vis_fpadd16(d1, d1); \
d1 = FMUL_16x16(d1, deltax); \
\
vis_alignaddr((void*)0, 2); \
d0 = vis_faligndata(d0, d0); \
dd = vis_fpadd16(d0, d1); \
dd = vis_fpadd16(dd, dd); \
XOR_8000(dd); \
\
deltax = vis_fpadd16(deltax, dx64); \
deltay = vis_fpadd16(deltay, dy64); \
deltax = vis_fand(deltax, mask_7fff); \
deltay = vis_fand(deltay, mask_7fff)
/***************************************************************/
#define LD_U16(sp, ind) vis_ld_u16(sp + ind)
/***************************************************************/
#ifndef MLIB_VIS2
#define LOAD_1CH() \
s0 = vis_faligndata(LD_U16(sp3, 0), mask_7fff); \
s1 = vis_faligndata(LD_U16(sp3, 2), mask_7fff); \
s2 = vis_faligndata(LD_U16(sp3, srcYStride), mask_7fff); \
s3 = vis_faligndata(LD_U16(sp3, srcYStride + 2), mask_7fff); \
\
s0 = vis_faligndata(LD_U16(sp2, 0), s0); \
s1 = vis_faligndata(LD_U16(sp2, 2), s1); \
s2 = vis_faligndata(LD_U16(sp2, srcYStride), s2); \
s3 = vis_faligndata(LD_U16(sp2, srcYStride + 2), s3); \
\
s0 = vis_faligndata(LD_U16(sp1, 0), s0); \
s1 = vis_faligndata(LD_U16(sp1, 2), s1); \
s2 = vis_faligndata(LD_U16(sp1, srcYStride), s2); \
s3 = vis_faligndata(LD_U16(sp1, srcYStride + 2), s3); \
\
s0 = vis_faligndata(LD_U16(sp0, 0), s0); \
s1 = vis_faligndata(LD_U16(sp0, 2), s1); \
s2 = vis_faligndata(LD_U16(sp0, srcYStride), s2); \
s3 = vis_faligndata(LD_U16(sp0, srcYStride + 2), s3)
#else
#define LOAD_1CH() \
s0 = vis_bshuffle(LD_U16(sp0, 0), LD_U16(sp2, 0)); \
s1 = vis_bshuffle(LD_U16(sp0, 2), LD_U16(sp2, 2)); \
s2 = vis_bshuffle(LD_U16(sp0, srcYStride), LD_U16(sp2, srcYStride)); \
s3 = vis_bshuffle(LD_U16(sp0, srcYStride + 2), LD_U16(sp2, srcYStride + 2)); \
\
t0 = vis_bshuffle(LD_U16(sp1, 0), LD_U16(sp3, 0)); \
t1 = vis_bshuffle(LD_U16(sp1, 2), LD_U16(sp3, 2)); \
t2 = vis_bshuffle(LD_U16(sp1, srcYStride), LD_U16(sp3, srcYStride)); \
t3 = vis_bshuffle(LD_U16(sp1, srcYStride + 2), LD_U16(sp3, srcYStride + 2)); \
\
s0 = vis_bshuffle(s0, t0); \
s1 = vis_bshuffle(s1, t1); \
s2 = vis_bshuffle(s2, t2); \
s3 = vis_bshuffle(s3, t3)
#endif
/***************************************************************/
#define GET_POINTER(sp) \
sp = *(mlib_u8**)((mlib_u8*)lineAddr + PTR_SHIFT(Y)) + 2*(X >> MLIB_SHIFT); \
X += dX; \
Y += dY
/***************************************************************/
#undef PREPARE_DELTAS
#define PREPARE_DELTAS \
if (warp_tbl != NULL) { \
dX = warp_tbl[2*j ]; \
dY = warp_tbl[2*j + 1]; \
dx64 = vis_to_double_dup((((dX << 1) & 0xFFFF) << 16) | ((dX << 1) & 0xFFFF)); \
dy64 = vis_to_double_dup((((dY << 1) & 0xFFFF) << 16) | ((dY << 1) & 0xFFFF)); \
}
/***************************************************************/
mlib_status FUN_NAME(1ch)(mlib_affine_param *param)
{
DECLAREVAR();
mlib_s32 off;
mlib_s32 x0, x1, x2, x3, y0, y1, y2, y3;
#ifdef MLIB_VIS2
mlib_d64 t0, t1, t2, t3;
vis_write_bmask(0x45CD67EF, 0);
#else
vis_alignaddr((void*)0, 6);
#endif
dx64 = vis_to_double_dup((((dX << 1) & 0xFFFF) << 16) | ((dX << 1) & 0xFFFF));
dy64 = vis_to_double_dup((((dY << 1) & 0xFFFF) << 16) | ((dY << 1) & 0xFFFF));
for (j = yStart; j <= yFinish; j++) {
mlib_u8 *sp0, *sp1, *sp2, *sp3;
mlib_d64 *dp, dmask;
NEW_LINE(1);
off = (mlib_s32)dl & 7;
dp = (mlib_d64*)(dl - off);
off >>= 1;
x0 = X - off*dX; y0 = Y - off*dY;
x1 = x0 + dX; y1 = y0 + dY;
x2 = x1 + dX; y2 = y1 + dY;
x3 = x2 + dX; y3 = y2 + dY;
deltax = DOUBLE_4U16(x0, x1, x2, x3);
deltay = DOUBLE_4U16(y0, y1, y2, y3);
if (off) {
mlib_s32 emask = vis_edge16((void*)(2*off), (void*)(2*(off + size - 1)));
off = 4 - off;
GET_POINTER(sp3);
sp0 = sp1 = sp2 = sp3;
if (off > 1 && size > 1) {
GET_POINTER(sp3);
}
if (off > 2) {
sp2 = sp3;
if (size > 2) {
GET_POINTER(sp3);
}
}
LOAD_1CH();
BL_SUM();
dmask = ((mlib_d64*)mlib_dmask_arr)[emask];
*dp++ = vis_for (vis_fand(dmask, dd), vis_fandnot(dmask, dp[0]));
size -= off;
if (size < 0) size = 0;
}
#pragma pipeloop(0)
for (i = 0; i < size/4; i++) {
GET_POINTER(sp0);
GET_POINTER(sp1);
GET_POINTER(sp2);
GET_POINTER(sp3);
LOAD_1CH();
BL_SUM();
dp[i] = dd;
}
off = size & 3;
if (off) {
GET_POINTER(sp0);
sp1 = sp2 = sp3 = sp0;
if (off > 1) {
GET_POINTER(sp1);
}
if (off > 2) {
GET_POINTER(sp2);
}
LOAD_1CH();
BL_SUM();
dmask = ((mlib_d64*)mlib_dmask_arr)[(0xF0 >> off) & 0x0F];
dp[i] = vis_for (vis_fand(dmask, dd), vis_fandnot(dmask, dp[i]));
}
}
return MLIB_SUCCESS;
}
/***************************************************************/
#undef GET_POINTER
#define GET_POINTER(sp) \
sp = *(mlib_f32**)((mlib_u8*)lineAddr + PTR_SHIFT(Y)) + (X >> MLIB_SHIFT); \
X += dX; \
Y += dY
/***************************************************************/
#define LOAD_2CH() \
s0 = vis_freg_pair(sp0[0], sp1[0]); \
s1 = vis_freg_pair(sp0[1], sp1[1]); \
s2 = vis_freg_pair(sp0[srcYStride], sp1[srcYStride]); \
s3 = vis_freg_pair(sp0[srcYStride + 1], sp1[srcYStride + 1])
/***************************************************************/
#undef PREPARE_DELTAS
#define PREPARE_DELTAS \
if (warp_tbl != NULL) { \
dX = warp_tbl[2*j ]; \
dY = warp_tbl[2*j + 1]; \
dx64 = vis_to_double_dup(((dX & 0xFFFF) << 16) | (dX & 0xFFFF)); \
dy64 = vis_to_double_dup(((dY & 0xFFFF) << 16) | (dY & 0xFFFF)); \
}
/***************************************************************/
mlib_status FUN_NAME(2ch)(mlib_affine_param *param)
{
DECLAREVAR();
mlib_s32 off;
mlib_s32 x0, x1, y0, y1;
if (((mlib_s32)lineAddr[0] | (mlib_s32)dstData | srcYStride | dstYStride) & 3) {
return FUN_NAME(2ch_na)(param);
}
srcYStride >>= 2;
dx64 = vis_to_double_dup(((dX & 0xFFFF) << 16) | (dX & 0xFFFF));
dy64 = vis_to_double_dup(((dY & 0xFFFF) << 16) | (dY & 0xFFFF));
for (j = yStart; j <= yFinish; j++) {
mlib_f32 *sp0, *sp1;
mlib_d64 *dp;
NEW_LINE(2);
off = (mlib_s32)dl & 7;
dp = (mlib_d64*)(dl - off);
if (off) {
x0 = X - dX; y0 = Y - dY;
x1 = X; y1 = Y;
} else {
x0 = X; y0 = Y;
x1 = X + dX; y1 = Y + dY;
}
deltax = DOUBLE_4U16(x0, x0, x1, x1);
deltay = DOUBLE_4U16(y0, y0, y1, y1);
if (off) {
GET_POINTER(sp1);
sp0 = sp1;
LOAD_2CH();
BL_SUM();
((mlib_f32*)dp)[1] = vis_read_lo(dd);
dp++;
size--;
}
#pragma pipeloop(0)
for (i = 0; i < size/2; i++) {
GET_POINTER(sp0);
GET_POINTER(sp1);
LOAD_2CH();
BL_SUM();
*dp++ = dd;
}
if (size & 1) {
GET_POINTER(sp0);
sp1 = sp0;
LOAD_2CH();
BL_SUM();
((mlib_f32*)dp)[0] = vis_read_hi(dd);
}
}
return MLIB_SUCCESS;
}
/***************************************************************/
#undef GET_POINTER
#define GET_POINTER(sp) \
sp = *(mlib_u8**)((mlib_u8*)lineAddr + PTR_SHIFT(Y)) + 4*(X >> MLIB_SHIFT); \
X += dX; \
Y += dY
/***************************************************************/
#ifndef MLIB_VIS2
#define LOAD_2CH_NA() \
s0 = vis_faligndata(LD_U16(sp1, 2), mask_7fff); \
s1 = vis_faligndata(LD_U16(sp1, 6), mask_7fff); \
s2 = vis_faligndata(LD_U16(sp1, srcYStride + 2), mask_7fff); \
s3 = vis_faligndata(LD_U16(sp1, srcYStride + 6), mask_7fff); \
\
s0 = vis_faligndata(LD_U16(sp1, 0), s0); \
s1 = vis_faligndata(LD_U16(sp1, 4), s1); \
s2 = vis_faligndata(LD_U16(sp1, srcYStride), s2); \
s3 = vis_faligndata(LD_U16(sp1, srcYStride + 4), s3); \
\
s0 = vis_faligndata(LD_U16(sp0, 2), s0); \
s1 = vis_faligndata(LD_U16(sp0, 6), s1); \
s2 = vis_faligndata(LD_U16(sp0, srcYStride + 2), s2); \
s3 = vis_faligndata(LD_U16(sp0, srcYStride + 6), s3); \
\
s0 = vis_faligndata(LD_U16(sp0, 0), s0); \
s1 = vis_faligndata(LD_U16(sp0, 4), s1); \
s2 = vis_faligndata(LD_U16(sp0, srcYStride), s2); \
s3 = vis_faligndata(LD_U16(sp0, srcYStride + 4), s3)
#else
#define LOAD_2CH_NA() \
s0 = vis_bshuffle(LD_U16(sp0, 0), LD_U16(sp1, 0)); \
s1 = vis_bshuffle(LD_U16(sp0, 4), LD_U16(sp1, 4)); \
s2 = vis_bshuffle(LD_U16(sp0, srcYStride), LD_U16(sp1, srcYStride)); \
s3 = vis_bshuffle(LD_U16(sp0, srcYStride + 4), LD_U16(sp1, srcYStride + 4)); \
\
t0 = vis_bshuffle(LD_U16(sp0, 2), LD_U16(sp1, 2)); \
t1 = vis_bshuffle(LD_U16(sp0, 6), LD_U16(sp1, 6)); \
t2 = vis_bshuffle(LD_U16(sp0, srcYStride + 2), LD_U16(sp1, srcYStride + 2)); \
t3 = vis_bshuffle(LD_U16(sp0, srcYStride + 6), LD_U16(sp1, srcYStride + 6)); \
\
s0 = vis_bshuffle(s0, t0); \
s1 = vis_bshuffle(s1, t1); \
s2 = vis_bshuffle(s2, t2); \
s3 = vis_bshuffle(s3, t3)
#endif
/***************************************************************/
mlib_status FUN_NAME(2ch_na)(mlib_affine_param *param)
{
DECLAREVAR();
mlib_s32 max_xsize = param -> max_xsize, bsize;
mlib_s32 x0, x1, y0, y1;
mlib_d64 buff[BUF_SIZE], *pbuff = buff;
#ifdef MLIB_VIS2
mlib_d64 t0, t1, t2, t3;
#endif
bsize = (max_xsize + 1)/2;
if (bsize > BUF_SIZE) {
pbuff = mlib_malloc(bsize*sizeof(mlib_d64));
if (pbuff == NULL) return MLIB_FAILURE;
}
MLIB_WRITE_BMASK(0x45CD67EF);
dx64 = vis_to_double_dup(((dX & 0xFFFF) << 16) | (dX & 0xFFFF));
dy64 = vis_to_double_dup(((dY & 0xFFFF) << 16) | (dY & 0xFFFF));
for (j = yStart; j <= yFinish; j++) {
mlib_u8 *sp0, *sp1;
#ifndef MLIB_VIS2
vis_alignaddr((void*)0, 6);
#endif
NEW_LINE(2);
x0 = X; y0 = Y;
x1 = X + dX; y1 = Y + dY;
deltax = DOUBLE_4U16(x0, x0, x1, x1);
deltay = DOUBLE_4U16(y0, y0, y1, y1);
#pragma pipeloop(0)
for (i = 0; i < size/2; i++) {
GET_POINTER(sp0);
GET_POINTER(sp1);
LOAD_2CH_NA();
BL_SUM();
pbuff[i] = dd;
}
if (size & 1) {
GET_POINTER(sp0);
sp1 = sp0;
LOAD_2CH_NA();
BL_SUM();
pbuff[i] = dd;
}
mlib_ImageCopy_na((mlib_u8*)pbuff, dl, 4*size);
}
if (pbuff != buff) {
mlib_free(pbuff);
}
return MLIB_SUCCESS;
}
/***************************************************************/
#undef PREPARE_DELTAS
#define PREPARE_DELTAS \
if (warp_tbl != NULL) { \
dX = warp_tbl[2*j ]; \
dY = warp_tbl[2*j + 1]; \
dX = (dX - (dX >> 31)) &~ 1; /* rounding towards ZERO */ \
dY = (dY - (dY >> 31)) &~ 1; /* rounding towards ZERO */ \
dx64 = vis_to_double_dup((((dX >> 1) & 0xFFFF) << 16) | ((dX >> 1) & 0xFFFF)); \
dy64 = vis_to_double_dup((((dY >> 1) & 0xFFFF) << 16) | ((dY >> 1) & 0xFFFF)); \
}
/***************************************************************/
mlib_status FUN_NAME(3ch)(mlib_affine_param *param)
{
DECLAREVAR();
mlib_s32 max_xsize = param -> max_xsize;
mlib_d64 buff[BUF_SIZE], *pbuff = buff;
if (max_xsize > BUF_SIZE) {
pbuff = mlib_malloc(max_xsize*sizeof(mlib_d64));
if (pbuff == NULL) return MLIB_FAILURE;
}
dX = (dX - (dX >> 31)) &~ 1; /* rounding towards ZERO */
dY = (dY - (dY >> 31)) &~ 1; /* rounding towards ZERO */
dx64 = vis_to_double_dup((((dX >> 1) & 0xFFFF) << 16) | ((dX >> 1) & 0xFFFF));
dy64 = vis_to_double_dup((((dY >> 1) & 0xFFFF) << 16) | ((dY >> 1) & 0xFFFF));
for (j = yStart; j <= yFinish; j++) {
mlib_u8 *sp;
mlib_d64 *sp0, *sp1;
NEW_LINE(3);
deltax = DOUBLE_4U16(X, X, X, X);
deltay = DOUBLE_4U16(Y, Y, Y, Y);
#pragma pipeloop(0)
for (i = 0; i < size; i++) {
sp = *(mlib_u8**)((mlib_u8*)lineAddr + PTR_SHIFT(Y)) + 6*(X >> MLIB_SHIFT) - 2;
vis_alignaddr(sp, 0);
sp0 = AL_ADDR(sp, 0);
s0 = vis_faligndata(sp0[0], sp0[1]);
s1 = vis_faligndata(sp0[1], sp0[2]);
vis_alignaddr(sp, srcYStride);
sp1 = AL_ADDR(sp, srcYStride);
s2 = vis_faligndata(sp1[0], sp1[1]);
s3 = vis_faligndata(sp1[1], sp1[2]);
BL_SUM_3CH();
pbuff[i] = dd;
X += dX;
Y += dY;
}
mlib_v_ImageChannelExtract_S16_43L_D1((void *)pbuff, (void *)dl, size);
}
if (pbuff != buff) {
mlib_free(pbuff);
}
return MLIB_SUCCESS;
}
/***************************************************************/
/**代码未完, 请加载全部代码(NowJava.com).**/
关注时代Java
关注时代Java