/* * Copyright (C) 2001-2003 Michael Niedermayer <michaelni@gmx.at> * * This file is part of FFmpeg. * * FFmpeg is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by * the Free Software Foundation; either version 2 of the License, or * (at your option) any later version. * * FFmpeg 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 for more details. * * You should have received a copy of the GNU General Public License * along with FFmpeg; if not, write to the Free Software * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA * * the C code (not assembly, mmx, ...) of this file can be used * under the LGPL license too *//*  supported Input formats: YV12, I420/IYUV, YUY2, UYVY, BGR32, BGR24, BGR16, BGR15, RGB32, RGB24, Y8/Y800, YVU9/IF09, PAL8  supported output formats: YV12, I420/IYUV, YUY2, UYVY, {BGR,RGB}{1,4,8,15,16,24,32}, Y8/Y800, YVU9/IF09  {BGR,RGB}{1,4,8,15,16} support dithering  unscaled special converters (YV12=I420=IYUV, Y800=Y8)  YV12 -> {BGR,RGB}{1,4,8,15,16,24,32}  x -> x  YUV9 -> YV12  YUV9/YV12 -> Y800  Y800 -> YUV9/YV12  BGR24 -> BGR32 & RGB24 -> RGB32  BGR32 -> BGR24 & RGB32 -> RGB24  BGR15 -> BGR16*//*tested special converters (most are tested actually, but I did not write it down ...) YV12 -> BGR16 YV12 -> YV12 BGR15 -> BGR16 BGR16 -> BGR16 YVU9 -> YV12untested special converters  YV12/I420 -> BGR15/BGR24/BGR32 (it is the yuv2rgb stuff, so it should be ok)  YV12/I420 -> YV12/I420  YUY2/BGR15/BGR24/BGR32/RGB24/RGB32 -> same format  BGR24 -> BGR32 & RGB24 -> RGB32  BGR32 -> BGR24 & RGB32 -> RGB24  BGR24 -> YV12*/#include <inttypes.h>#include <string.h>#include <math.h>#include <stdio.h>#include <unistd.h>#include "config.h"#include <assert.h>#ifdef HAVE_SYS_MMAN_H#include <sys/mman.h>#if defined(MAP_ANON) && !defined(MAP_ANONYMOUS)#define MAP_ANONYMOUS MAP_ANON#endif#endif#include "swscale.h"#include "swscale_internal.h"#include "rgb2rgb.h"#include "libavutil/x86_cpu.h"#include "libavutil/bswap.h"#include "libavcodec/opt.h"#undef MOVNTQ#undef PAVGB//#undef HAVE_MMX2//#define HAVE_3DNOW//#undef HAVE_MMX//#undef ARCH_X86//#define WORDS_BIGENDIAN#define DITHER1XBPP#define FAST_BGR2YV12 // use 7 bit coeffs instead of 15bit#define RET 0xC3 //near return opcode for X86#ifdef M_PI#define PI M_PI#else#define PI 3.14159265358979323846#endif#define isSupportedIn(x)    (       \           (x)==PIX_FMT_YUV420P     \        || (x)==PIX_FMT_YUVA420P    \        || (x)==PIX_FMT_YUYV422     \        || (x)==PIX_FMT_UYVY422     \        || (x)==PIX_FMT_RGB32       \        || (x)==PIX_FMT_BGR24       \        || (x)==PIX_FMT_BGR565      \        || (x)==PIX_FMT_BGR555      \        || (x)==PIX_FMT_BGR32       \        || (x)==PIX_FMT_RGB24       \        || (x)==PIX_FMT_RGB565      \        || (x)==PIX_FMT_RGB555      \        || (x)==PIX_FMT_GRAY8       \        || (x)==PIX_FMT_YUV410P     \        || (x)==PIX_FMT_GRAY16BE    \        || (x)==PIX_FMT_GRAY16LE    \        || (x)==PIX_FMT_YUV444P     \        || (x)==PIX_FMT_YUV422P     \        || (x)==PIX_FMT_YUV411P     \        || (x)==PIX_FMT_PAL8        \        || (x)==PIX_FMT_BGR8        \        || (x)==PIX_FMT_RGB8        \        || (x)==PIX_FMT_BGR4_BYTE   \        || (x)==PIX_FMT_RGB4_BYTE   \        || (x)==PIX_FMT_YUV440P     \    )#define isSupportedOut(x)   (       \           (x)==PIX_FMT_YUV420P     \        || (x)==PIX_FMT_YUYV422     \        || (x)==PIX_FMT_UYVY422     \        || (x)==PIX_FMT_YUV444P     \        || (x)==PIX_FMT_YUV422P     \        || (x)==PIX_FMT_YUV411P     \        || isRGB(x)                 \        || isBGR(x)                 \        || (x)==PIX_FMT_NV12        \        || (x)==PIX_FMT_NV21        \        || (x)==PIX_FMT_GRAY16BE    \        || (x)==PIX_FMT_GRAY16LE    \        || (x)==PIX_FMT_GRAY8       \        || (x)==PIX_FMT_YUV410P     \    )#define isPacked(x)         (       \           (x)==PIX_FMT_PAL8        \        || (x)==PIX_FMT_YUYV422     \        || (x)==PIX_FMT_UYVY422     \        || isRGB(x)                 \        || isBGR(x)                 \    )#define RGB2YUV_SHIFT 16#define BY ((int)( 0.098*(1<<RGB2YUV_SHIFT)+0.5))#define BV ((int)(-0.071*(1<<RGB2YUV_SHIFT)+0.5))#define BU ((int)( 0.439*(1<<RGB2YUV_SHIFT)+0.5))#define GY ((int)( 0.504*(1<<RGB2YUV_SHIFT)+0.5))#define GV ((int)(-0.368*(1<<RGB2YUV_SHIFT)+0.5))#define GU ((int)(-0.291*(1<<RGB2YUV_SHIFT)+0.5))#define RY ((int)( 0.257*(1<<RGB2YUV_SHIFT)+0.5))#define RV ((int)( 0.439*(1<<RGB2YUV_SHIFT)+0.5))#define RU ((int)(-0.148*(1<<RGB2YUV_SHIFT)+0.5))extern const int32_t Inverse_Table_6_9[8][4];/*NOTESSpecial versions: fast Y 1:1 scaling (no interpolation in y direction)TODOmore intelligent misalignment avoidance for the horizontal scalerwrite special vertical cubic upscale versionOptimize C code (yv12 / minmax)add support for packed pixel yuv input & outputadd support for Y8 outputoptimize bgr24 & bgr32add BGR4 output supportwrite special BGR->BGR scaler*/#if defined(ARCH_X86) && defined (CONFIG_GPL)DECLARE_ASM_CONST(8, uint64_t, bF8)=       0xF8F8F8F8F8F8F8F8LL;DECLARE_ASM_CONST(8, uint64_t, bFC)=       0xFCFCFCFCFCFCFCFCLL;DECLARE_ASM_CONST(8, uint64_t, w10)=       0x0010001000100010LL;DECLARE_ASM_CONST(8, uint64_t, w02)=       0x0002000200020002LL;DECLARE_ASM_CONST(8, uint64_t, bm00001111)=0x00000000FFFFFFFFLL;DECLARE_ASM_CONST(8, uint64_t, bm00000111)=0x0000000000FFFFFFLL;DECLARE_ASM_CONST(8, uint64_t, bm11111000)=0xFFFFFFFFFF000000LL;DECLARE_ASM_CONST(8, uint64_t, bm01010101)=0x00FF00FF00FF00FFLL;static volatile uint64_t attribute_used __attribute__((aligned(8))) b5Dither;static volatile uint64_t attribute_used __attribute__((aligned(8))) g5Dither;static volatile uint64_t attribute_used __attribute__((aligned(8))) g6Dither;static volatile uint64_t attribute_used __attribute__((aligned(8))) r5Dither;const DECLARE_ALIGNED(8, uint64_t, ff_dither4[2]) = {        0x0103010301030103LL,        0x0200020002000200LL,};const DECLARE_ALIGNED(8, uint64_t, ff_dither8[2]) = {        0x0602060206020602LL,        0x0004000400040004LL,};DECLARE_ASM_CONST(8, uint64_t, b16Mask)=   0x001F001F001F001FLL;DECLARE_ASM_CONST(8, uint64_t, g16Mask)=   0x07E007E007E007E0LL;DECLARE_ASM_CONST(8, uint64_t, r16Mask)=   0xF800F800F800F800LL;DECLARE_ASM_CONST(8, uint64_t, b15Mask)=   0x001F001F001F001FLL;DECLARE_ASM_CONST(8, uint64_t, g15Mask)=   0x03E003E003E003E0LL;DECLARE_ASM_CONST(8, uint64_t, r15Mask)=   0x7C007C007C007C00LL;DECLARE_ALIGNED(8, const uint64_t, ff_M24A)         = 0x00FF0000FF0000FFLL;DECLARE_ALIGNED(8, const uint64_t, ff_M24B)         = 0xFF0000FF0000FF00LL;DECLARE_ALIGNED(8, const uint64_t, ff_M24C)         = 0x0000FF0000FF0000LL;#ifdef FAST_BGR2YV12DECLARE_ALIGNED(8, const uint64_t, ff_bgr2YCoeff)   = 0x000000210041000DULL;DECLARE_ALIGNED(8, const uint64_t, ff_bgr2UCoeff)   = 0x0000FFEEFFDC0038ULL;DECLARE_ALIGNED(8, const uint64_t, ff_bgr2VCoeff)   = 0x00000038FFD2FFF8ULL;#elseDECLARE_ALIGNED(8, const uint64_t, ff_bgr2YCoeff)   = 0x000020E540830C8BULL;DECLARE_ALIGNED(8, const uint64_t, ff_bgr2UCoeff)   = 0x0000ED0FDAC23831ULL;DECLARE_ALIGNED(8, const uint64_t, ff_bgr2VCoeff)   = 0x00003831D0E6F6EAULL;#endif /* FAST_BGR2YV12 */DECLARE_ALIGNED(8, const uint64_t, ff_bgr2YOffset)  = 0x1010101010101010ULL;DECLARE_ALIGNED(8, const uint64_t, ff_bgr2UVOffset) = 0x8080808080808080ULL;DECLARE_ALIGNED(8, const uint64_t, ff_w1111)        = 0x0001000100010001ULL;#endif /* defined(ARCH_X86) */// clipping helper table for C implementations:static unsigned char clip_table[768];static SwsVector *sws_getConvVec(SwsVector *a, SwsVector *b);extern const uint8_t dither_2x2_4[2][8];extern const uint8_t dither_2x2_8[2][8];extern const uint8_t dither_8x8_32[8][8];extern const uint8_t dither_8x8_73[8][8];extern const uint8_t dither_8x8_220[8][8];static const char * sws_context_to_name(void * ptr) {    return "swscaler";}#define OFFSET(x) offsetof(SwsContext, x)#define DEFAULT 0#define VE AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAMstatic const AVOption options[] = {    { "sws_flags", "scaler/cpu flags", OFFSET(flags), FF_OPT_TYPE_FLAGS, DEFAULT, 0, UINT_MAX, VE, "sws_flags" },    { "fast_bilinear", "fast bilinear", 0, FF_OPT_TYPE_CONST, SWS_FAST_BILINEAR, INT_MIN, INT_MAX, VE, "sws_flags" },    { "bilinear", "bilinear", 0, FF_OPT_TYPE_CONST, SWS_BILINEAR, INT_MIN, INT_MAX, VE, "sws_flags" },    { "bicubic", "bicubic", 0, FF_OPT_TYPE_CONST, SWS_BICUBIC, INT_MIN, INT_MAX, VE, "sws_flags" },    { "experimental", "experimental", 0, FF_OPT_TYPE_CONST, SWS_X, INT_MIN, INT_MAX, VE, "sws_flags" },    { "neighbor", "nearest neighbor", 0, FF_OPT_TYPE_CONST, SWS_POINT, INT_MIN, INT_MAX, VE, "sws_flags" },    { "area", "averaging area", 0, FF_OPT_TYPE_CONST, SWS_AREA, INT_MIN, INT_MAX, VE, "sws_flags" },    { "bicublin", "luma bicubic, chroma bilinear", 0, FF_OPT_TYPE_CONST, SWS_BICUBLIN, INT_MIN, INT_MAX, VE, "sws_flags" },    { "gauss", "gaussian", 0, FF_OPT_TYPE_CONST, SWS_GAUSS, INT_MIN, INT_MAX, VE, "sws_flags" },    { "sinc", "sinc", 0, FF_OPT_TYPE_CONST, SWS_SINC, INT_MIN, INT_MAX, VE, "sws_flags" },    { "lanczos", "lanczos", 0, FF_OPT_TYPE_CONST, SWS_LANCZOS, INT_MIN, INT_MAX, VE, "sws_flags" },    { "spline", "natural bicubic spline", 0, FF_OPT_TYPE_CONST, SWS_SPLINE, INT_MIN, INT_MAX, VE, "sws_flags" },    { "print_info", "print info", 0, FF_OPT_TYPE_CONST, SWS_PRINT_INFO, INT_MIN, INT_MAX, VE, "sws_flags" },    { "accurate_rnd", "accurate rounding", 0, FF_OPT_TYPE_CONST, SWS_ACCURATE_RND, INT_MIN, INT_MAX, VE, "sws_flags" },    { "mmx", "MMX SIMD acceleration", 0, FF_OPT_TYPE_CONST, SWS_CPU_CAPS_MMX, INT_MIN, INT_MAX, VE, "sws_flags" },    { "mmx2", "MMX2 SIMD acceleration", 0, FF_OPT_TYPE_CONST, SWS_CPU_CAPS_MMX2, INT_MIN, INT_MAX, VE, "sws_flags" },    { "3dnow", "3DNOW SIMD acceleration", 0, FF_OPT_TYPE_CONST, SWS_CPU_CAPS_3DNOW, INT_MIN, INT_MAX, VE, "sws_flags" },    { "altivec", "AltiVec SIMD acceleration", 0, FF_OPT_TYPE_CONST, SWS_CPU_CAPS_ALTIVEC, INT_MIN, INT_MAX, VE, "sws_flags" },    { "bfin", "Blackfin SIMD acceleration", 0, FF_OPT_TYPE_CONST, SWS_CPU_CAPS_BFIN, INT_MIN, INT_MAX, VE, "sws_flags" },    { "full_chroma_int", "full chroma interpolation", 0 , FF_OPT_TYPE_CONST, SWS_FULL_CHR_H_INT, INT_MIN, INT_MAX, VE, "sws_flags" },    { "full_chroma_inp", "full chroma input", 0 , FF_OPT_TYPE_CONST, SWS_FULL_CHR_H_INP, INT_MIN, INT_MAX, VE, "sws_flags" },    { NULL }};#undef VE#undef DEFAULTstatic const AVClass sws_context_class = { "SWScaler", sws_context_to_name, options };const char *sws_format_name(enum PixelFormat format){    switch (format) {        case PIX_FMT_YUV420P:            return "yuv420p";        case PIX_FMT_YUVA420P:            return "yuva420p";        case PIX_FMT_YUYV422:            return "yuyv422";        case PIX_FMT_RGB24:            return "rgb24";        case PIX_FMT_BGR24:            return "bgr24";        case PIX_FMT_YUV422P:            return "yuv422p";        case PIX_FMT_YUV444P:            return "yuv444p";        case PIX_FMT_RGB32:            return "rgb32";        case PIX_FMT_YUV410P:            return "yuv410p";        case PIX_FMT_YUV411P:            return "yuv411p";        case PIX_FMT_RGB565:            return "rgb565";        case PIX_FMT_RGB555:            return "rgb555";        case PIX_FMT_GRAY16BE:            return "gray16be";        case PIX_FMT_GRAY16LE:            return "gray16le";        case PIX_FMT_GRAY8:            return "gray8";        case PIX_FMT_MONOWHITE:            return "mono white";        case PIX_FMT_MONOBLACK:            return "mono black";        case PIX_FMT_PAL8:            return "Palette";        case PIX_FMT_YUVJ420P:            return "yuvj420p";        case PIX_FMT_YUVJ422P:            return "yuvj422p";        case PIX_FMT_YUVJ444P:            return "yuvj444p";        case PIX_FMT_XVMC_MPEG2_MC:            return "xvmc_mpeg2_mc";        case PIX_FMT_XVMC_MPEG2_IDCT:            return "xvmc_mpeg2_idct";        case PIX_FMT_UYVY422:            return "uyvy422";        case PIX_FMT_UYYVYY411:            return "uyyvyy411";        case PIX_FMT_RGB32_1:            return "rgb32x";        case PIX_FMT_BGR32_1:            return "bgr32x";        case PIX_FMT_BGR32:            return "bgr32";        case PIX_FMT_BGR565:            return "bgr565";        case PIX_FMT_BGR555:            return "bgr555";        case PIX_FMT_BGR8:            return "bgr8";        case PIX_FMT_BGR4:            return "bgr4";        case PIX_FMT_BGR4_BYTE:            return "bgr4 byte";        case PIX_FMT_RGB8:            return "rgb8";        case PIX_FMT_RGB4:            return "rgb4";        case PIX_FMT_RGB4_BYTE:            return "rgb4 byte";        case PIX_FMT_NV12:            return "nv12";        case PIX_FMT_NV21:            return "nv21";        case PIX_FMT_YUV440P:            return "yuv440p";        default:            return "Unknown format";    }}static inline void yuv2yuvXinC(int16_t *lumFilter, int16_t **lumSrc, int lumFilterSize,                               int16_t *chrFilter, int16_t **chrSrc, int chrFilterSize,                               uint8_t *dest, uint8_t *uDest, uint8_t *vDest, int dstW, int chrDstW){    //FIXME Optimize (just quickly writen not opti..)    int i;    for (i=0; i<dstW; i++)    {        int val=1<<18;        int j;        for (j=0; j<lumFilterSize; j++)            val += lumSrc[j][i] * lumFilter[j];        dest[i]= av_clip_uint8(val>>19);    }    if (uDest)        for (i=0; i<chrDstW; i++)        {            int u=1<<18;            int v=1<<18;            int j;            for (j=0; j<chrFilterSize; j++)            {                u += chrSrc[j][i] * chrFilter[j];