LCOV - code coverage report
Current view: top level - src/libswscale - swscale.c (source / functions) Hit Total Coverage
Test: coverage.info Lines: 444 527 84.3 %
Date: 2017-01-19 23:52:33 Functions: 19 22 86.4 %

          Line data    Source code
       1             : /*
       2             :  * Copyright (C) 2001-2011 Michael Niedermayer <michaelni@gmx.at>
       3             :  *
       4             :  * This file is part of FFmpeg.
       5             :  *
       6             :  * FFmpeg is free software; you can redistribute it and/or
       7             :  * modify it under the terms of the GNU Lesser General Public
       8             :  * License as published by the Free Software Foundation; either
       9             :  * version 2.1 of the License, or (at your option) any later version.
      10             :  *
      11             :  * FFmpeg is distributed in the hope that it will be useful,
      12             :  * but WITHOUT ANY WARRANTY; without even the implied warranty of
      13             :  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
      14             :  * Lesser General Public License for more details.
      15             :  *
      16             :  * You should have received a copy of the GNU Lesser General Public
      17             :  * License along with FFmpeg; if not, write to the Free Software
      18             :  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
      19             :  */
      20             : 
      21             : #include <inttypes.h>
      22             : #include <math.h>
      23             : #include <stdio.h>
      24             : #include <string.h>
      25             : 
      26             : #include "libavutil/avassert.h"
      27             : #include "libavutil/avutil.h"
      28             : #include "libavutil/bswap.h"
      29             : #include "libavutil/cpu.h"
      30             : #include "libavutil/imgutils.h"
      31             : #include "libavutil/intreadwrite.h"
      32             : #include "libavutil/mathematics.h"
      33             : #include "libavutil/pixdesc.h"
      34             : #include "config.h"
      35             : #include "rgb2rgb.h"
      36             : #include "swscale_internal.h"
      37             : #include "swscale.h"
      38             : 
      39             : DECLARE_ALIGNED(8, const uint8_t, ff_dither_8x8_128)[9][8] = {
      40             :     {  36, 68,  60, 92,  34, 66,  58, 90, },
      41             :     { 100,  4, 124, 28,  98,  2, 122, 26, },
      42             :     {  52, 84,  44, 76,  50, 82,  42, 74, },
      43             :     { 116, 20, 108, 12, 114, 18, 106, 10, },
      44             :     {  32, 64,  56, 88,  38, 70,  62, 94, },
      45             :     {  96,  0, 120, 24, 102,  6, 126, 30, },
      46             :     {  48, 80,  40, 72,  54, 86,  46, 78, },
      47             :     { 112, 16, 104,  8, 118, 22, 110, 14, },
      48             :     {  36, 68,  60, 92,  34, 66,  58, 90, },
      49             : };
      50             : 
      51             : DECLARE_ALIGNED(8, static const uint8_t, sws_pb_64)[8] = {
      52             :     64, 64, 64, 64, 64, 64, 64, 64
      53             : };
      54             : 
      55         710 : static av_always_inline void fillPlane(uint8_t *plane, int stride, int width,
      56             :                                        int height, int y, uint8_t val)
      57             : {
      58             :     int i;
      59         710 :     uint8_t *ptr = plane + stride * y;
      60      126590 :     for (i = 0; i < height; i++) {
      61      125880 :         memset(ptr, val, width);
      62      125880 :         ptr += stride;
      63             :     }
      64         710 : }
      65             : 
      66       26523 : static void hScale16To19_c(SwsContext *c, int16_t *_dst, int dstW,
      67             :                            const uint8_t *_src, const int16_t *filter,
      68             :                            const int32_t *filterPos, int filterSize)
      69             : {
      70       26523 :     const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(c->srcFormat);
      71             :     int i;
      72       26523 :     int32_t *dst        = (int32_t *) _dst;
      73       26523 :     const uint16_t *src = (const uint16_t *) _src;
      74       26523 :     int bits            = desc->comp[0].depth - 1;
      75       26523 :     int sh              = bits - 4;
      76             : 
      77       26523 :     if((isAnyRGB(c->srcFormat) || c->srcFormat==AV_PIX_FMT_PAL8) && desc->comp[0].depth<16)
      78           0 :         sh= 9;
      79             : 
      80     4980474 :     for (i = 0; i < dstW; i++) {
      81             :         int j;
      82     4953951 :         int srcPos = filterPos[i];
      83     4953951 :         int val    = 0;
      84             : 
      85    39629502 :         for (j = 0; j < filterSize; j++) {
      86    34675551 :             val += src[srcPos + j] * filter[filterSize * i + j];
      87             :         }
      88             :         // filter=14 bit, input=16 bit, output=30 bit, >> 11 makes 19 bit
      89     4953951 :         dst[i] = FFMIN(val >> sh, (1 << 19) - 1);
      90             :     }
      91       26523 : }
      92             : 
      93     5569778 : static void hScale16To15_c(SwsContext *c, int16_t *dst, int dstW,
      94             :                            const uint8_t *_src, const int16_t *filter,
      95             :                            const int32_t *filterPos, int filterSize)
      96             : {
      97     5569778 :     const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(c->srcFormat);
      98             :     int i;
      99     5569778 :     const uint16_t *src = (const uint16_t *) _src;
     100     5569778 :     int sh              = desc->comp[0].depth - 1;
     101             : 
     102     5569778 :     if(sh<15)
     103     5118730 :         sh= isAnyRGB(c->srcFormat) || c->srcFormat==AV_PIX_FMT_PAL8 ? 13 : (desc->comp[0].depth - 1);
     104             : 
     105  1548783534 :     for (i = 0; i < dstW; i++) {
     106             :         int j;
     107  1543213756 :         int srcPos = filterPos[i];
     108  1543213756 :         int val    = 0;
     109             : 
     110  4015890432 :         for (j = 0; j < filterSize; j++) {
     111  2472676676 :             val += src[srcPos + j] * filter[filterSize * i + j];
     112             :         }
     113             :         // filter=14 bit, input=16 bit, output=30 bit, >> 15 makes 15 bit
     114  1543213756 :         dst[i] = FFMIN(val >> sh, (1 << 15) - 1);
     115             :     }
     116     5569778 : }
     117             : 
     118             : // bilinear / bicubic scaling
     119    10014249 : static void hScale8To15_c(SwsContext *c, int16_t *dst, int dstW,
     120             :                           const uint8_t *src, const int16_t *filter,
     121             :                           const int32_t *filterPos, int filterSize)
     122             : {
     123             :     int i;
     124  2901243149 :     for (i = 0; i < dstW; i++) {
     125             :         int j;
     126  2891228900 :         int srcPos = filterPos[i];
     127  2891228900 :         int val    = 0;
     128 11087470954 :         for (j = 0; j < filterSize; j++) {
     129  8196242054 :             val += ((int)src[srcPos + j]) * filter[filterSize * i + j];
     130             :         }
     131  2891228900 :         dst[i] = FFMIN(val >> 7, (1 << 15) - 1); // the cubic equation does overflow ...
     132             :     }
     133    10014249 : }
     134             : 
     135      612352 : static void hScale8To19_c(SwsContext *c, int16_t *_dst, int dstW,
     136             :                           const uint8_t *src, const int16_t *filter,
     137             :                           const int32_t *filterPos, int filterSize)
     138             : {
     139             :     int i;
     140      612352 :     int32_t *dst = (int32_t *) _dst;
     141   182784120 :     for (i = 0; i < dstW; i++) {
     142             :         int j;
     143   182171768 :         int srcPos = filterPos[i];
     144   182171768 :         int val    = 0;
     145   452714056 :         for (j = 0; j < filterSize; j++) {
     146   270542288 :             val += ((int)src[srcPos + j]) * filter[filterSize * i + j];
     147             :         }
     148   182171768 :         dst[i] = FFMIN(val >> 3, (1 << 19) - 1); // the cubic equation does overflow ...
     149             :     }
     150      612352 : }
     151             : 
     152             : // FIXME all pal and rgb srcFormats could do this conversion as well
     153             : // FIXME all scalers more complex than bilinear could do half of this transform
     154      160874 : static void chrRangeToJpeg_c(int16_t *dstU, int16_t *dstV, int width)
     155             : {
     156             :     int i;
     157    34728854 :     for (i = 0; i < width; i++) {
     158    34567980 :         dstU[i] = (FFMIN(dstU[i], 30775) * 4663 - 9289992) >> 12; // -264
     159    34567980 :         dstV[i] = (FFMIN(dstV[i], 30775) * 4663 - 9289992) >> 12; // -264
     160             :     }
     161      160874 : }
     162             : 
     163      162334 : static void chrRangeFromJpeg_c(int16_t *dstU, int16_t *dstV, int width)
     164             : {
     165             :     int i;
     166    27939132 :     for (i = 0; i < width; i++) {
     167    27776798 :         dstU[i] = (dstU[i] * 1799 + 4081085) >> 11; // 1469
     168    27776798 :         dstV[i] = (dstV[i] * 1799 + 4081085) >> 11; // 1469
     169             :     }
     170      162334 : }
     171             : 
     172      480729 : static void lumRangeToJpeg_c(int16_t *dst, int width)
     173             : {
     174             :     int i;
     175   162264222 :     for (i = 0; i < width; i++)
     176   161783493 :         dst[i] = (FFMIN(dst[i], 30189) * 19077 - 39057361) >> 14;
     177      480729 : }
     178             : 
     179      230260 : static void lumRangeFromJpeg_c(int16_t *dst, int width)
     180             : {
     181             :     int i;
     182    78578780 :     for (i = 0; i < width; i++)
     183    78348520 :         dst[i] = (dst[i] * 14071 + 33561947) >> 14;
     184      230260 : }
     185             : 
     186           0 : static void chrRangeToJpeg16_c(int16_t *_dstU, int16_t *_dstV, int width)
     187             : {
     188             :     int i;
     189           0 :     int32_t *dstU = (int32_t *) _dstU;
     190           0 :     int32_t *dstV = (int32_t *) _dstV;
     191           0 :     for (i = 0; i < width; i++) {
     192           0 :         dstU[i] = (FFMIN(dstU[i], 30775 << 4) * 4663 - (9289992 << 4)) >> 12; // -264
     193           0 :         dstV[i] = (FFMIN(dstV[i], 30775 << 4) * 4663 - (9289992 << 4)) >> 12; // -264
     194             :     }
     195           0 : }
     196             : 
     197           0 : static void chrRangeFromJpeg16_c(int16_t *_dstU, int16_t *_dstV, int width)
     198             : {
     199             :     int i;
     200           0 :     int32_t *dstU = (int32_t *) _dstU;
     201           0 :     int32_t *dstV = (int32_t *) _dstV;
     202           0 :     for (i = 0; i < width; i++) {
     203           0 :         dstU[i] = (dstU[i] * 1799 + (4081085 << 4)) >> 11; // 1469
     204           0 :         dstV[i] = (dstV[i] * 1799 + (4081085 << 4)) >> 11; // 1469
     205             :     }
     206           0 : }
     207             : 
     208       15840 : static void lumRangeToJpeg16_c(int16_t *_dst, int width)
     209             : {
     210             :     int i;
     211       15840 :     int32_t *dst = (int32_t *) _dst;
     212     5591520 :     for (i = 0; i < width; i++) {
     213     5575680 :         dst[i] = ((int)(FFMIN(dst[i], 30189 << 4) * 4769U - (39057361 << 2))) >> 12;
     214             :     }
     215       15840 : }
     216             : 
     217           0 : static void lumRangeFromJpeg16_c(int16_t *_dst, int width)
     218             : {
     219             :     int i;
     220           0 :     int32_t *dst = (int32_t *) _dst;
     221           0 :     for (i = 0; i < width; i++)
     222           0 :         dst[i] = (dst[i]*(14071/4) + (33561947<<4)/4)>>12;
     223           0 : }
     224             : 
     225             : 
     226             : #define DEBUG_SWSCALE_BUFFERS 0
     227             : #define DEBUG_BUFFERS(...)                      \
     228             :     if (DEBUG_SWSCALE_BUFFERS)                  \
     229             :         av_log(c, AV_LOG_DEBUG, __VA_ARGS__)
     230             : 
     231       29099 : static int swscale(SwsContext *c, const uint8_t *src[],
     232             :                    int srcStride[], int srcSliceY,
     233             :                    int srcSliceH, uint8_t *dst[], int dstStride[])
     234             : {
     235             :     /* load a few things into local vars to make the code more readable?
     236             :      * and faster */
     237       29099 :     const int dstW                   = c->dstW;
     238       29099 :     const int dstH                   = c->dstH;
     239             : 
     240       29099 :     const enum AVPixelFormat dstFormat = c->dstFormat;
     241       29099 :     const int flags                  = c->flags;
     242       29099 :     int32_t *vLumFilterPos           = c->vLumFilterPos;
     243       29099 :     int32_t *vChrFilterPos           = c->vChrFilterPos;
     244             : 
     245       29099 :     const int vLumFilterSize         = c->vLumFilterSize;
     246       29099 :     const int vChrFilterSize         = c->vChrFilterSize;
     247             : 
     248       29099 :     yuv2planar1_fn yuv2plane1        = c->yuv2plane1;
     249       29099 :     yuv2planarX_fn yuv2planeX        = c->yuv2planeX;
     250       29099 :     yuv2interleavedX_fn yuv2nv12cX   = c->yuv2nv12cX;
     251       29099 :     yuv2packed1_fn yuv2packed1       = c->yuv2packed1;
     252       29099 :     yuv2packed2_fn yuv2packed2       = c->yuv2packed2;
     253       29099 :     yuv2packedX_fn yuv2packedX       = c->yuv2packedX;
     254       29099 :     yuv2anyX_fn yuv2anyX             = c->yuv2anyX;
     255       29099 :     const int chrSrcSliceY           =                srcSliceY >> c->chrSrcVSubSample;
     256       29099 :     const int chrSrcSliceH           = AV_CEIL_RSHIFT(srcSliceH,   c->chrSrcVSubSample);
     257       56139 :     int should_dither                = isNBPS(c->srcFormat) ||
     258       27040 :                                        is16BPS(c->srcFormat);
     259             :     int lastDstY;
     260             : 
     261             :     /* vars which will change and which we need to store back in the context */
     262       29099 :     int dstY         = c->dstY;
     263       29099 :     int lumBufIndex  = c->lumBufIndex;
     264       29099 :     int chrBufIndex  = c->chrBufIndex;
     265       29099 :     int lastInLumBuf = c->lastInLumBuf;
     266       29099 :     int lastInChrBuf = c->lastInChrBuf;
     267             : 
     268             : 
     269       29099 :     int lumStart = 0;
     270       29099 :     int lumEnd = c->descIndex[0];
     271       29099 :     int chrStart = lumEnd;
     272       29099 :     int chrEnd = c->descIndex[1];
     273       29099 :     int vStart = chrEnd;
     274       29099 :     int vEnd = c->numDesc;
     275       29099 :     SwsSlice *src_slice = &c->slice[lumStart];
     276       29099 :     SwsSlice *hout_slice = &c->slice[c->numSlice-2];
     277       29099 :     SwsSlice *vout_slice = &c->slice[c->numSlice-1];
     278       29099 :     SwsFilterDescriptor *desc = c->desc;
     279             : 
     280             : 
     281       29099 :     int needAlpha = c->needAlpha;
     282             : 
     283       29099 :     int hasLumHoles = 1;
     284       29099 :     int hasChrHoles = 1;
     285             : 
     286             : 
     287       29099 :     if (isPacked(c->srcFormat)) {
     288        5903 :         src[0] =
     289       11806 :         src[1] =
     290       11806 :         src[2] =
     291        5903 :         src[3] = src[0];
     292        5903 :         srcStride[0] =
     293       11806 :         srcStride[1] =
     294       11806 :         srcStride[2] =
     295        5903 :         srcStride[3] = srcStride[0];
     296             :     }
     297       29099 :     srcStride[1] <<= c->vChrDrop;
     298       29099 :     srcStride[2] <<= c->vChrDrop;
     299             : 
     300             :     DEBUG_BUFFERS("swscale() %p[%d] %p[%d] %p[%d] %p[%d] -> %p[%d] %p[%d] %p[%d] %p[%d]\n",
     301             :                   src[0], srcStride[0], src[1], srcStride[1],
     302             :                   src[2], srcStride[2], src[3], srcStride[3],
     303             :                   dst[0], dstStride[0], dst[1], dstStride[1],
     304             :                   dst[2], dstStride[2], dst[3], dstStride[3]);
     305             :     DEBUG_BUFFERS("srcSliceY: %d srcSliceH: %d dstY: %d dstH: %d\n",
     306             :                   srcSliceY, srcSliceH, dstY, dstH);
     307             :     DEBUG_BUFFERS("vLumFilterSize: %d vChrFilterSize: %d\n",
     308             :                   vLumFilterSize, vChrFilterSize);
     309             : 
     310       58198 :     if (dstStride[0]&15 || dstStride[1]&15 ||
     311       58198 :         dstStride[2]&15 || dstStride[3]&15) {
     312             :         static int warnedAlready = 0; // FIXME maybe move this into the context
     313           0 :         if (flags & SWS_PRINT_INFO && !warnedAlready) {
     314           0 :             av_log(c, AV_LOG_WARNING,
     315             :                    "Warning: dstStride is not aligned!\n"
     316             :                    "         ->cannot do aligned memory accesses anymore\n");
     317           0 :             warnedAlready = 1;
     318             :         }
     319             :     }
     320             : 
     321       29099 :     if (   (uintptr_t)dst[0]&15 || (uintptr_t)dst[1]&15 || (uintptr_t)dst[2]&15
     322       28803 :         || (uintptr_t)src[0]&15 || (uintptr_t)src[1]&15 || (uintptr_t)src[2]&15
     323       26728 :         || dstStride[0]&15 || dstStride[1]&15 || dstStride[2]&15 || dstStride[3]&15
     324       26728 :         || srcStride[0]&15 || srcStride[1]&15 || srcStride[2]&15 || srcStride[3]&15
     325             :     ) {
     326             :         static int warnedAlready=0;
     327        2623 :         int cpu_flags = av_get_cpu_flags();
     328        2623 :         if (HAVE_MMXEXT && (cpu_flags & AV_CPU_FLAG_SSE2) && !warnedAlready){
     329           0 :             av_log(c, AV_LOG_WARNING, "Warning: data is not aligned! This can lead to a speedloss\n");
     330           0 :             warnedAlready=1;
     331             :         }
     332             :     }
     333             : 
     334             :     /* Note the user might start scaling the picture in the middle so this
     335             :      * will not get executed. This is not really intended but works
     336             :      * currently, so people might do it. */
     337       29099 :     if (srcSliceY == 0) {
     338       29099 :         lumBufIndex  = -1;
     339       29099 :         chrBufIndex  = -1;
     340       29099 :         dstY         = 0;
     341       29099 :         lastInLumBuf = -1;
     342       29099 :         lastInChrBuf = -1;
     343             :     }
     344             : 
     345       29099 :     if (!should_dither) {
     346       26202 :         c->chrDither8 = c->lumDither8 = sws_pb_64;
     347             :     }
     348       29099 :     lastDstY = dstY;
     349             : 
     350       29099 :     ff_init_vscale_pfn(c, yuv2plane1, yuv2planeX, yuv2nv12cX,
     351             :                    yuv2packed1, yuv2packed2, yuv2packedX, yuv2anyX, c->use_mmx_vfilter);
     352             : 
     353       29099 :     ff_init_slice_from_src(src_slice, (uint8_t**)src, srcStride, c->srcW,
     354             :             srcSliceY, srcSliceH, chrSrcSliceY, chrSrcSliceH, 1);
     355             : 
     356       58198 :     ff_init_slice_from_src(vout_slice, (uint8_t**)dst, dstStride, c->dstW,
     357       29099 :             dstY, dstH, dstY >> c->chrDstVSubSample,
     358       29099 :             AV_CEIL_RSHIFT(dstH, c->chrDstVSubSample), 0);
     359       29099 :     if (srcSliceY == 0) {
     360       29099 :         hout_slice->plane[0].sliceY = lastInLumBuf + 1;
     361       29099 :         hout_slice->plane[1].sliceY = lastInChrBuf + 1;
     362       29099 :         hout_slice->plane[2].sliceY = lastInChrBuf + 1;
     363       29099 :         hout_slice->plane[3].sliceY = lastInLumBuf + 1;
     364             : 
     365       29099 :         hout_slice->plane[0].sliceH =
     366       29099 :         hout_slice->plane[1].sliceH =
     367       29099 :         hout_slice->plane[2].sliceH =
     368       29099 :         hout_slice->plane[3].sliceH = 0;
     369       29099 :         hout_slice->width = dstW;
     370             :     }
     371             : 
     372     9616294 :     for (; dstY < dstH; dstY++) {
     373     9587195 :         const int chrDstY = dstY >> c->chrDstVSubSample;
     374     9587195 :         int use_mmx_vfilter= c->use_mmx_vfilter;
     375             : 
     376             :         // First line needed as input
     377     9587195 :         const int firstLumSrcY  = FFMAX(1 - vLumFilterSize, vLumFilterPos[dstY]);
     378     9587195 :         const int firstLumSrcY2 = FFMAX(1 - vLumFilterSize, vLumFilterPos[FFMIN(dstY | ((1 << c->chrDstVSubSample) - 1), dstH - 1)]);
     379             :         // First line needed as input
     380     9587195 :         const int firstChrSrcY  = FFMAX(1 - vChrFilterSize, vChrFilterPos[chrDstY]);
     381             : 
     382             :         // Last line needed as input
     383     9587195 :         int lastLumSrcY  = FFMIN(c->srcH,    firstLumSrcY  + vLumFilterSize) - 1;
     384     9587195 :         int lastLumSrcY2 = FFMIN(c->srcH,    firstLumSrcY2 + vLumFilterSize) - 1;
     385     9587195 :         int lastChrSrcY  = FFMIN(c->chrSrcH, firstChrSrcY  + vChrFilterSize) - 1;
     386             :         int enough_lines;
     387             : 
     388             :         int i;
     389             :         int posY, cPosY, firstPosY, lastPosY, firstCPosY, lastCPosY;
     390             : 
     391             :         // handle holes (FAST_BILINEAR & weird filters)
     392     9587195 :         if (firstLumSrcY > lastInLumBuf) {
     393             : 
     394     6138730 :             hasLumHoles = lastInLumBuf != firstLumSrcY - 1;
     395     6138730 :             if (hasLumHoles) {
     396       94300 :                 hout_slice->plane[0].sliceY = firstLumSrcY;
     397       94300 :                 hout_slice->plane[3].sliceY = firstLumSrcY;
     398       94300 :                 hout_slice->plane[0].sliceH =
     399       94300 :                 hout_slice->plane[3].sliceH = 0;
     400             :             }
     401             : 
     402     6138730 :             lastInLumBuf = firstLumSrcY - 1;
     403             :         }
     404     9587195 :         if (firstChrSrcY > lastInChrBuf) {
     405             : 
     406     1149395 :             hasChrHoles = lastInChrBuf != firstChrSrcY - 1;
     407     1149395 :             if (hasChrHoles) {
     408      230850 :                 hout_slice->plane[1].sliceY = firstChrSrcY;
     409      230850 :                 hout_slice->plane[2].sliceY = firstChrSrcY;
     410      230850 :                 hout_slice->plane[1].sliceH =
     411      230850 :                 hout_slice->plane[2].sliceH = 0;
     412             :             }
     413             : 
     414     1149395 :             lastInChrBuf = firstChrSrcY - 1;
     415             :         }
     416             : 
     417             :         DEBUG_BUFFERS("dstY: %d\n", dstY);
     418             :         DEBUG_BUFFERS("\tfirstLumSrcY: %d lastLumSrcY: %d lastInLumBuf: %d\n",
     419             :                       firstLumSrcY, lastLumSrcY, lastInLumBuf);
     420             :         DEBUG_BUFFERS("\tfirstChrSrcY: %d lastChrSrcY: %d lastInChrBuf: %d\n",
     421             :                       firstChrSrcY, lastChrSrcY, lastInChrBuf);
     422             : 
     423             :         // Do we have enough lines in this slice to output the dstY line
     424    19174390 :         enough_lines = lastLumSrcY2 < srcSliceY + srcSliceH &&
     425     9587195 :                        lastChrSrcY < AV_CEIL_RSHIFT(srcSliceY + srcSliceH, c->chrSrcVSubSample);
     426             : 
     427     9587195 :         if (!enough_lines) {
     428           0 :             lastLumSrcY = srcSliceY + srcSliceH - 1;
     429           0 :             lastChrSrcY = chrSrcSliceY + chrSrcSliceH - 1;
     430             :             DEBUG_BUFFERS("buffering slice: lastLumSrcY %d lastChrSrcY %d\n",
     431             :                           lastLumSrcY, lastChrSrcY);
     432             :         }
     433             : 
     434     9587195 :         av_assert0((lastLumSrcY - firstLumSrcY + 1) <= hout_slice->plane[0].available_lines);
     435     9587195 :         av_assert0((lastChrSrcY - firstChrSrcY + 1) <= hout_slice->plane[1].available_lines);
     436             : 
     437             : 
     438     9587195 :         posY = hout_slice->plane[0].sliceY + hout_slice->plane[0].sliceH;
     439     9587195 :         if (posY <= lastLumSrcY && !hasLumHoles) {
     440     1200824 :             firstPosY = FFMAX(firstLumSrcY, posY);
     441     1200824 :             lastPosY = FFMIN(firstLumSrcY + hout_slice->plane[0].available_lines - 1, srcSliceY + srcSliceH - 1);
     442             :         } else {
     443     8386371 :             firstPosY = posY;
     444     8386371 :             lastPosY = lastLumSrcY;
     445             :         }
     446             : 
     447     9587195 :         cPosY = hout_slice->plane[1].sliceY + hout_slice->plane[1].sliceH;
     448     9587195 :         if (cPosY <= lastChrSrcY && !hasChrHoles) {
     449      825931 :             firstCPosY = FFMAX(firstChrSrcY, cPosY);
     450      825931 :             lastCPosY = FFMIN(firstChrSrcY + hout_slice->plane[1].available_lines - 1, AV_CEIL_RSHIFT(srcSliceY + srcSliceH, c->chrSrcVSubSample) - 1);
     451             :         } else {
     452     8761264 :             firstCPosY = cPosY;
     453     8761264 :             lastCPosY = lastChrSrcY;
     454             :         }
     455             : 
     456     9587195 :         ff_rotate_slice(hout_slice, lastPosY, lastCPosY);
     457             : 
     458     9587195 :         if (posY < lastLumSrcY + 1) {
     459     2834056 :             for (i = lumStart; i < lumEnd; ++i)
     460     1538932 :                 desc[i].process(c, &desc[i], firstPosY, lastPosY - firstPosY + 1);
     461             :         }
     462             : 
     463     9587195 :         lumBufIndex += lastLumSrcY - lastInLumBuf;
     464     9587195 :         lastInLumBuf = lastLumSrcY;
     465             : 
     466     9587195 :         if (cPosY < lastChrSrcY + 1) {
     467     2420318 :             for (i = chrStart; i < chrEnd; ++i)
     468     1363537 :                 desc[i].process(c, &desc[i], firstCPosY, lastCPosY - firstCPosY + 1);
     469             :         }
     470             : 
     471     9587195 :         chrBufIndex += lastChrSrcY - lastInChrBuf;
     472     9587195 :         lastInChrBuf = lastChrSrcY;
     473             : 
     474             :         // wrap buf index around to stay inside the ring buffer
     475     9587195 :         if (lumBufIndex >= vLumFilterSize)
     476     6318957 :             lumBufIndex -= vLumFilterSize;
     477     9587195 :         if (chrBufIndex >= vChrFilterSize)
     478     1737166 :             chrBufIndex -= vChrFilterSize;
     479     9587195 :         if (!enough_lines)
     480           0 :             break;  // we can't output a dstY line so let's try with the next slice
     481             : 
     482             : #if HAVE_MMX_INLINE
     483     9587195 :         ff_updateMMXDitherTables(c, dstY, lumBufIndex, chrBufIndex,
     484             :                               lastInLumBuf, lastInChrBuf);
     485             : #endif
     486     9587195 :         if (should_dither) {
     487     2761263 :             c->chrDither8 = ff_dither_8x8_128[chrDstY & 7];
     488     2761263 :             c->lumDither8 = ff_dither_8x8_128[dstY    & 7];
     489             :         }
     490     9587195 :         if (dstY >= dstH - 2) {
     491             :             /* hmm looks like we can't use MMX here without overwriting
     492             :              * this array's tail */
     493       58198 :             ff_sws_init_output_funcs(c, &yuv2plane1, &yuv2planeX, &yuv2nv12cX,
     494             :                                      &yuv2packed1, &yuv2packed2, &yuv2packedX, &yuv2anyX);
     495       58198 :             use_mmx_vfilter= 0;
     496       58198 :             ff_init_vscale_pfn(c, yuv2plane1, yuv2planeX, yuv2nv12cX,
     497             :                            yuv2packed1, yuv2packed2, yuv2packedX, yuv2anyX, use_mmx_vfilter);
     498             :         }
     499             : 
     500             :         {
     501    26426826 :             for (i = vStart; i < vEnd; ++i)
     502    16839631 :                 desc[i].process(c, &desc[i], dstY, 1);
     503             :         }
     504             :     }
     505       29099 :     if (isPlanar(dstFormat) && isALPHA(dstFormat) && !needAlpha) {
     506         990 :         int length = dstW;
     507         990 :         int height = dstY - lastDstY;
     508             : 
     509        1270 :         if (is16BPS(dstFormat) || isNBPS(dstFormat)) {
     510         280 :             const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(dstFormat);
     511         280 :             fillPlane16(dst[3], dstStride[3], length, height, lastDstY,
     512             :                     1, desc->comp[3].depth,
     513             :                     isBE(dstFormat));
     514             :         } else
     515         710 :             fillPlane(dst[3], dstStride[3], length, height, lastDstY, 255);
     516             :     }
     517             : 
     518             : #if HAVE_MMXEXT_INLINE
     519       29099 :     if (av_get_cpu_flags() & AV_CPU_FLAG_MMXEXT)
     520        1448 :         __asm__ volatile ("sfence" ::: "memory");
     521             : #endif
     522       29099 :     emms_c();
     523             : 
     524             :     /* store changed local vars back in the context */
     525       29099 :     c->dstY         = dstY;
     526       29099 :     c->lumBufIndex  = lumBufIndex;
     527       29099 :     c->chrBufIndex  = chrBufIndex;
     528       29099 :     c->lastInLumBuf = lastInLumBuf;
     529       29099 :     c->lastInChrBuf = lastInChrBuf;
     530             : 
     531       29099 :     return dstY - lastDstY;
     532             : }
     533             : 
     534        2636 : av_cold void ff_sws_init_range_convert(SwsContext *c)
     535             : {
     536        2636 :     c->lumConvertRange = NULL;
     537        2636 :     c->chrConvertRange = NULL;
     538        2636 :     if (c->srcRange != c->dstRange && !isAnyRGB(c->dstFormat)) {
     539         201 :         if (c->dstBpc <= 14) {
     540         178 :             if (c->srcRange) {
     541          29 :                 c->lumConvertRange = lumRangeFromJpeg_c;
     542          29 :                 c->chrConvertRange = chrRangeFromJpeg_c;
     543             :             } else {
     544         149 :                 c->lumConvertRange = lumRangeToJpeg_c;
     545         149 :                 c->chrConvertRange = chrRangeToJpeg_c;
     546             :             }
     547             :         } else {
     548          23 :             if (c->srcRange) {
     549           0 :                 c->lumConvertRange = lumRangeFromJpeg16_c;
     550           0 :                 c->chrConvertRange = chrRangeFromJpeg16_c;
     551             :             } else {
     552          23 :                 c->lumConvertRange = lumRangeToJpeg16_c;
     553          23 :                 c->chrConvertRange = chrRangeToJpeg16_c;
     554             :             }
     555             :         }
     556             :     }
     557        2636 : }
     558             : 
     559        2247 : static av_cold void sws_init_swscale(SwsContext *c)
     560             : {
     561        2247 :     enum AVPixelFormat srcFormat = c->srcFormat;
     562             : 
     563        2247 :     ff_sws_init_output_funcs(c, &c->yuv2plane1, &c->yuv2planeX,
     564             :                              &c->yuv2nv12cX, &c->yuv2packed1,
     565             :                              &c->yuv2packed2, &c->yuv2packedX, &c->yuv2anyX);
     566             : 
     567        2247 :     ff_sws_init_input_funcs(c);
     568             : 
     569             : 
     570        2247 :     if (c->srcBpc == 8) {
     571        1935 :         if (c->dstBpc <= 14) {
     572        1636 :             c->hyScale = c->hcScale = hScale8To15_c;
     573        1636 :             if (c->flags & SWS_FAST_BILINEAR) {
     574           0 :                 c->hyscale_fast = ff_hyscale_fast_c;
     575           0 :                 c->hcscale_fast = ff_hcscale_fast_c;
     576             :             }
     577             :         } else {
     578         299 :             c->hyScale = c->hcScale = hScale8To19_c;
     579             :         }
     580             :     } else {
     581         312 :         c->hyScale = c->hcScale = c->dstBpc > 14 ? hScale16To19_c
     582             :                                                  : hScale16To15_c;
     583             :     }
     584             : 
     585        2247 :     ff_sws_init_range_convert(c);
     586             : 
     587        2247 :     if (!(isGray(srcFormat) || isGray(c->dstFormat) ||
     588             :           srcFormat == AV_PIX_FMT_MONOBLACK || srcFormat == AV_PIX_FMT_MONOWHITE))
     589        2162 :         c->needs_hcscale = 1;
     590        2247 : }
     591             : 
     592        2247 : SwsFunc ff_getSwsFunc(SwsContext *c)
     593             : {
     594        2247 :     sws_init_swscale(c);
     595             : 
     596             :     if (ARCH_PPC)
     597             :         ff_sws_init_swscale_ppc(c);
     598             :     if (ARCH_X86)
     599        2247 :         ff_sws_init_swscale_x86(c);
     600             :     if (ARCH_AARCH64)
     601             :         ff_sws_init_swscale_aarch64(c);
     602             :     if (ARCH_ARM)
     603             :         ff_sws_init_swscale_arm(c);
     604             : 
     605        2247 :     return swscale;
     606             : }
     607             : 
     608       71938 : static void reset_ptr(const uint8_t *src[], enum AVPixelFormat format)
     609             : {
     610       71938 :     if (!isALPHA(format))
     611       61702 :         src[3] = NULL;
     612       71938 :     if (!isPlanar(format)) {
     613       26733 :         src[3] = src[2] = NULL;
     614             : 
     615       26733 :         if (!usePal(format))
     616       21247 :             src[1] = NULL;
     617             :     }
     618       71938 : }
     619             : 
     620       71938 : static int check_image_pointers(const uint8_t * const data[4], enum AVPixelFormat pix_fmt,
     621             :                                 const int linesizes[4])
     622             : {
     623       71938 :     const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(pix_fmt);
     624             :     int i;
     625             : 
     626             :     av_assert2(desc);
     627             : 
     628      359690 :     for (i = 0; i < 4; i++) {
     629      287752 :         int plane = desc->comp[i].plane;
     630      287752 :         if (!data[plane] || !linesizes[plane])
     631           0 :             return 0;
     632             :     }
     633             : 
     634       71938 :     return 1;
     635             : }
     636             : 
     637           2 : static void xyz12Torgb48(struct SwsContext *c, uint16_t *dst,
     638             :                          const uint16_t *src, int stride, int h)
     639             : {
     640             :     int xp,yp;
     641           2 :     const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(c->srcFormat);
     642             : 
     643         578 :     for (yp=0; yp<h; yp++) {
     644      203328 :         for (xp=0; xp+2<stride; xp+=3) {
     645             :             int x, y, z, r, g, b;
     646             : 
     647      202752 :             if (desc->flags & AV_PIX_FMT_FLAG_BE) {
     648      101376 :                 x = AV_RB16(src + xp + 0);
     649      101376 :                 y = AV_RB16(src + xp + 1);
     650      101376 :                 z = AV_RB16(src + xp + 2);
     651             :             } else {
     652      101376 :                 x = AV_RL16(src + xp + 0);
     653      101376 :                 y = AV_RL16(src + xp + 1);
     654      101376 :                 z = AV_RL16(src + xp + 2);
     655             :             }
     656             : 
     657      202752 :             x = c->xyzgamma[x>>4];
     658      202752 :             y = c->xyzgamma[y>>4];
     659      202752 :             z = c->xyzgamma[z>>4];
     660             : 
     661             :             // convert from XYZlinear to sRGBlinear
     662      608256 :             r = c->xyz2rgb_matrix[0][0] * x +
     663      405504 :                 c->xyz2rgb_matrix[0][1] * y +
     664      202752 :                 c->xyz2rgb_matrix[0][2] * z >> 12;
     665      608256 :             g = c->xyz2rgb_matrix[1][0] * x +
     666      405504 :                 c->xyz2rgb_matrix[1][1] * y +
     667      202752 :                 c->xyz2rgb_matrix[1][2] * z >> 12;
     668      608256 :             b = c->xyz2rgb_matrix[2][0] * x +
     669      405504 :                 c->xyz2rgb_matrix[2][1] * y +
     670      202752 :                 c->xyz2rgb_matrix[2][2] * z >> 12;
     671             : 
     672             :             // limit values to 12-bit depth
     673      202752 :             r = av_clip_uintp2(r, 12);
     674      202752 :             g = av_clip_uintp2(g, 12);
     675      202752 :             b = av_clip_uintp2(b, 12);
     676             : 
     677             :             // convert from sRGBlinear to RGB and scale from 12bit to 16bit
     678      202752 :             if (desc->flags & AV_PIX_FMT_FLAG_BE) {
     679      101376 :                 AV_WB16(dst + xp + 0, c->rgbgamma[r] << 4);
     680      101376 :                 AV_WB16(dst + xp + 1, c->rgbgamma[g] << 4);
     681      101376 :                 AV_WB16(dst + xp + 2, c->rgbgamma[b] << 4);
     682             :             } else {
     683      101376 :                 AV_WL16(dst + xp + 0, c->rgbgamma[r] << 4);
     684      101376 :                 AV_WL16(dst + xp + 1, c->rgbgamma[g] << 4);
     685      101376 :                 AV_WL16(dst + xp + 2, c->rgbgamma[b] << 4);
     686             :             }
     687             :         }
     688         576 :         src += stride;
     689         576 :         dst += stride;
     690             :     }
     691           2 : }
     692             : 
     693          31 : static void rgb48Toxyz12(struct SwsContext *c, uint16_t *dst,
     694             :                          const uint16_t *src, int stride, int h)
     695             : {
     696             :     int xp,yp;
     697          31 :     const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(c->dstFormat);
     698             : 
     699        8583 :     for (yp=0; yp<h; yp++) {
     700     2833384 :         for (xp=0; xp+2<stride; xp+=3) {
     701             :             int x, y, z, r, g, b;
     702             : 
     703     2824832 :             if (desc->flags & AV_PIX_FMT_FLAG_BE) {
     704     1338688 :                 r = AV_RB16(src + xp + 0);
     705     1338688 :                 g = AV_RB16(src + xp + 1);
     706     1338688 :                 b = AV_RB16(src + xp + 2);
     707             :             } else {
     708     1486144 :                 r = AV_RL16(src + xp + 0);
     709     1486144 :                 g = AV_RL16(src + xp + 1);
     710     1486144 :                 b = AV_RL16(src + xp + 2);
     711             :             }
     712             : 
     713     2824832 :             r = c->rgbgammainv[r>>4];
     714     2824832 :             g = c->rgbgammainv[g>>4];
     715     2824832 :             b = c->rgbgammainv[b>>4];
     716             : 
     717             :             // convert from sRGBlinear to XYZlinear
     718     8474496 :             x = c->rgb2xyz_matrix[0][0] * r +
     719     5649664 :                 c->rgb2xyz_matrix[0][1] * g +
     720     2824832 :                 c->rgb2xyz_matrix[0][2] * b >> 12;
     721     8474496 :             y = c->rgb2xyz_matrix[1][0] * r +
     722     5649664 :                 c->rgb2xyz_matrix[1][1] * g +
     723     2824832 :                 c->rgb2xyz_matrix[1][2] * b >> 12;
     724     8474496 :             z = c->rgb2xyz_matrix[2][0] * r +
     725     5649664 :                 c->rgb2xyz_matrix[2][1] * g +
     726     2824832 :                 c->rgb2xyz_matrix[2][2] * b >> 12;
     727             : 
     728             :             // limit values to 12-bit depth
     729     2824832 :             x = av_clip_uintp2(x, 12);
     730     2824832 :             y = av_clip_uintp2(y, 12);
     731     2824832 :             z = av_clip_uintp2(z, 12);
     732             : 
     733             :             // convert from XYZlinear to X'Y'Z' and scale from 12bit to 16bit
     734     2824832 :             if (desc->flags & AV_PIX_FMT_FLAG_BE) {
     735     1338688 :                 AV_WB16(dst + xp + 0, c->xyzgammainv[x] << 4);
     736     1338688 :                 AV_WB16(dst + xp + 1, c->xyzgammainv[y] << 4);
     737     1338688 :                 AV_WB16(dst + xp + 2, c->xyzgammainv[z] << 4);
     738             :             } else {
     739     1486144 :                 AV_WL16(dst + xp + 0, c->xyzgammainv[x] << 4);
     740     1486144 :                 AV_WL16(dst + xp + 1, c->xyzgammainv[y] << 4);
     741     1486144 :                 AV_WL16(dst + xp + 2, c->xyzgammainv[z] << 4);
     742             :             }
     743             :         }
     744        8552 :         src += stride;
     745        8552 :         dst += stride;
     746             :     }
     747          31 : }
     748             : 
     749             : /**
     750             :  * swscale wrapper, so we don't need to export the SwsContext.
     751             :  * Assumes planar YUV to be in YUV order instead of YVU.
     752             :  */
     753       35969 : int attribute_align_arg sws_scale(struct SwsContext *c,
     754             :                                   const uint8_t * const srcSlice[],
     755             :                                   const int srcStride[], int srcSliceY,
     756             :                                   int srcSliceH, uint8_t *const dst[],
     757             :                                   const int dstStride[])
     758             : {
     759             :     int i, ret;
     760             :     const uint8_t *src2[4];
     761             :     uint8_t *dst2[4];
     762       35969 :     uint8_t *rgb0_tmp = NULL;
     763       35969 :     int macro_height = isBayer(c->srcFormat) ? 2 : (1 << c->chrSrcVSubSample);
     764             :     // copy strides, so they can safely be modified
     765             :     int srcStride2[4];
     766             :     int dstStride2[4];
     767       35969 :     int srcSliceY_internal = srcSliceY;
     768             : 
     769       35969 :     if (!srcStride || !dstStride || !dst || !srcSlice) {
     770           0 :         av_log(c, AV_LOG_ERROR, "One of the input parameters to sws_scale() is NULL, please check the calling code\n");
     771           0 :         return 0;
     772             :     }
     773             : 
     774      179845 :     for (i=0; i<4; i++) {
     775      143876 :         srcStride2[i] = srcStride[i];
     776      143876 :         dstStride2[i] = dstStride[i];
     777             :     }
     778             : 
     779       71938 :     if ((srcSliceY & (macro_height-1)) ||
     780       71991 :         ((srcSliceH& (macro_height-1)) && srcSliceY + srcSliceH != c->srcH) ||
     781       35969 :         srcSliceY + srcSliceH > c->srcH) {
     782           0 :         av_log(c, AV_LOG_ERROR, "Slice parameters %d, %d are invalid\n", srcSliceY, srcSliceH);
     783           0 :         return AVERROR(EINVAL);
     784             :     }
     785             : 
     786       35969 :     if (c->gamma_flag && c->cascaded_context[0]) {
     787             : 
     788             : 
     789           0 :         ret = sws_scale(c->cascaded_context[0],
     790             :                     srcSlice, srcStride, srcSliceY, srcSliceH,
     791           0 :                     c->cascaded_tmp, c->cascaded_tmpStride);
     792             : 
     793           0 :         if (ret < 0)
     794           0 :             return ret;
     795             : 
     796           0 :         if (c->cascaded_context[2])
     797           0 :             ret = sws_scale(c->cascaded_context[1], (const uint8_t * const *)c->cascaded_tmp, c->cascaded_tmpStride, srcSliceY, srcSliceH, c->cascaded1_tmp, c->cascaded1_tmpStride);
     798             :         else
     799           0 :             ret = sws_scale(c->cascaded_context[1], (const uint8_t * const *)c->cascaded_tmp, c->cascaded_tmpStride, srcSliceY, srcSliceH, dst, dstStride);
     800             : 
     801           0 :         if (ret < 0)
     802           0 :             return ret;
     803             : 
     804           0 :         if (c->cascaded_context[2]) {
     805           0 :             ret = sws_scale(c->cascaded_context[2],
     806           0 :                         (const uint8_t * const *)c->cascaded1_tmp, c->cascaded1_tmpStride, c->cascaded_context[1]->dstY - ret, c->cascaded_context[1]->dstY,
     807             :                         dst, dstStride);
     808             :         }
     809           0 :         return ret;
     810             :     }
     811             : 
     812       35969 :     if (c->cascaded_context[0] && srcSliceY == 0 && srcSliceH == c->cascaded_context[0]->srcH) {
     813           0 :         ret = sws_scale(c->cascaded_context[0],
     814             :                         srcSlice, srcStride, srcSliceY, srcSliceH,
     815           0 :                         c->cascaded_tmp, c->cascaded_tmpStride);
     816           0 :         if (ret < 0)
     817           0 :             return ret;
     818           0 :         ret = sws_scale(c->cascaded_context[1],
     819           0 :                         (const uint8_t * const * )c->cascaded_tmp, c->cascaded_tmpStride, 0, c->cascaded_context[0]->dstH,
     820             :                         dst, dstStride);
     821           0 :         return ret;
     822             :     }
     823             : 
     824       35969 :     memcpy(src2, srcSlice, sizeof(src2));
     825       35969 :     memcpy(dst2, dst, sizeof(dst2));
     826             : 
     827             :     // do not mess up sliceDir if we have a "trailing" 0-size slice
     828       35969 :     if (srcSliceH == 0)
     829           0 :         return 0;
     830             : 
     831       35969 :     if (!check_image_pointers(srcSlice, c->srcFormat, srcStride)) {
     832           0 :         av_log(c, AV_LOG_ERROR, "bad src image pointers\n");
     833           0 :         return 0;
     834             :     }
     835       35969 :     if (!check_image_pointers((const uint8_t* const*)dst, c->dstFormat, dstStride)) {
     836           0 :         av_log(c, AV_LOG_ERROR, "bad dst image pointers\n");
     837           0 :         return 0;
     838             :     }
     839             : 
     840       35969 :     if (c->sliceDir == 0 && srcSliceY != 0 && srcSliceY + srcSliceH != c->srcH) {
     841           0 :         av_log(c, AV_LOG_ERROR, "Slices start in the middle!\n");
     842           0 :         return 0;
     843             :     }
     844       35969 :     if (c->sliceDir == 0) {
     845       35969 :         if (srcSliceY == 0) c->sliceDir = 1; else c->sliceDir = -1;
     846             :     }
     847             : 
     848       35969 :     if (usePal(c->srcFormat)) {
     849     1006669 :         for (i = 0; i < 256; i++) {
     850     1002752 :             int r, g, b, y, u, v, a = 0xff;
     851     1002752 :             if (c->srcFormat == AV_PIX_FMT_PAL8) {
     852     1000704 :                 uint32_t p = ((const uint32_t *)(srcSlice[1]))[i];
     853     1000704 :                 a = (p >> 24) & 0xFF;
     854     1000704 :                 r = (p >> 16) & 0xFF;
     855     1000704 :                 g = (p >>  8) & 0xFF;
     856     1000704 :                 b =  p        & 0xFF;
     857        2048 :             } else if (c->srcFormat == AV_PIX_FMT_RGB8) {
     858         256 :                 r = ( i >> 5     ) * 36;
     859         256 :                 g = ((i >> 2) & 7) * 36;
     860         256 :                 b = ( i       & 3) * 85;
     861        1792 :             } else if (c->srcFormat == AV_PIX_FMT_BGR8) {
     862         256 :                 b = ( i >> 6     ) * 85;
     863         256 :                 g = ((i >> 3) & 7) * 36;
     864         256 :                 r = ( i       & 7) * 36;
     865        1536 :             } else if (c->srcFormat == AV_PIX_FMT_RGB4_BYTE) {
     866         256 :                 r = ( i >> 3     ) * 255;
     867         256 :                 g = ((i >> 1) & 3) * 85;
     868         256 :                 b = ( i       & 1) * 255;
     869        1280 :             } else if (c->srcFormat == AV_PIX_FMT_GRAY8 || c->srcFormat == AV_PIX_FMT_GRAY8A) {
     870        1024 :                 r = g = b = i;
     871             :             } else {
     872             :                 av_assert1(c->srcFormat == AV_PIX_FMT_BGR4_BYTE);
     873         256 :                 b = ( i >> 3     ) * 255;
     874         256 :                 g = ((i >> 1) & 3) * 85;
     875         256 :                 r = ( i       & 1) * 255;
     876             :             }
     877             : #define RGB2YUV_SHIFT 15
     878             : #define BY ( (int) (0.114 * 219 / 255 * (1 << RGB2YUV_SHIFT) + 0.5))
     879             : #define BV (-(int) (0.081 * 224 / 255 * (1 << RGB2YUV_SHIFT) + 0.5))
     880             : #define BU ( (int) (0.500 * 224 / 255 * (1 << RGB2YUV_SHIFT) + 0.5))
     881             : #define GY ( (int) (0.587 * 219 / 255 * (1 << RGB2YUV_SHIFT) + 0.5))
     882             : #define GV (-(int) (0.419 * 224 / 255 * (1 << RGB2YUV_SHIFT) + 0.5))
     883             : #define GU (-(int) (0.331 * 224 / 255 * (1 << RGB2YUV_SHIFT) + 0.5))
     884             : #define RY ( (int) (0.299 * 219 / 255 * (1 << RGB2YUV_SHIFT) + 0.5))
     885             : #define RV ( (int) (0.500 * 224 / 255 * (1 << RGB2YUV_SHIFT) + 0.5))
     886             : #define RU (-(int) (0.169 * 224 / 255 * (1 << RGB2YUV_SHIFT) + 0.5))
     887             : 
     888     1002752 :             y = av_clip_uint8((RY * r + GY * g + BY * b + ( 33 << (RGB2YUV_SHIFT - 1))) >> RGB2YUV_SHIFT);
     889     1002752 :             u = av_clip_uint8((RU * r + GU * g + BU * b + (257 << (RGB2YUV_SHIFT - 1))) >> RGB2YUV_SHIFT);
     890     1002752 :             v = av_clip_uint8((RV * r + GV * g + BV * b + (257 << (RGB2YUV_SHIFT - 1))) >> RGB2YUV_SHIFT);
     891     1002752 :             c->pal_yuv[i]= y + (u<<8) + (v<<16) + ((unsigned)a<<24);
     892             : 
     893     1002752 :             switch (c->dstFormat) {
     894             :             case AV_PIX_FMT_BGR32:
     895             : #if !HAVE_BIGENDIAN
     896             :             case AV_PIX_FMT_RGB24:
     897             : #endif
     898      851712 :                 c->pal_rgb[i]=  r + (g<<8) + (b<<16) + ((unsigned)a<<24);
     899      851712 :                 break;
     900             :             case AV_PIX_FMT_BGR32_1:
     901             : #if HAVE_BIGENDIAN
     902             :             case AV_PIX_FMT_BGR24:
     903             : #endif
     904           0 :                 c->pal_rgb[i]= a + (r<<8) + (g<<16) + ((unsigned)b<<24);
     905           0 :                 break;
     906             :             case AV_PIX_FMT_RGB32_1:
     907             : #if HAVE_BIGENDIAN
     908             :             case AV_PIX_FMT_RGB24:
     909             : #endif
     910           0 :                 c->pal_rgb[i]= a + (b<<8) + (g<<16) + ((unsigned)r<<24);
     911           0 :                 break;
     912             :             case AV_PIX_FMT_RGB32:
     913             : #if !HAVE_BIGENDIAN
     914             :             case AV_PIX_FMT_BGR24:
     915             : #endif
     916             :             default:
     917      151040 :                 c->pal_rgb[i]=  b + (g<<8) + (r<<16) + ((unsigned)a<<24);
     918             :             }
     919             :         }
     920             :     }
     921             : 
     922       35969 :     if (c->src0Alpha && !c->dst0Alpha && isALPHA(c->dstFormat)) {
     923             :         uint8_t *base;
     924             :         int x,y;
     925           6 :         rgb0_tmp = av_malloc(FFABS(srcStride[0]) * srcSliceH + 32);
     926           6 :         if (!rgb0_tmp)
     927           0 :             return AVERROR(ENOMEM);
     928             : 
     929           6 :         base = srcStride[0] < 0 ? rgb0_tmp - srcStride[0] * (srcSliceH-1) : rgb0_tmp;
     930        4668 :         for (y=0; y<srcSliceH; y++){
     931        4662 :             memcpy(base + srcStride[0]*y, src2[0] + srcStride[0]*y, 4*c->srcW);
     932     2873864 :             for (x=c->src0Alpha-1; x<4*c->srcW; x+=4) {
     933     2869202 :                 base[ srcStride[0]*y + x] = 0xFF;
     934             :             }
     935             :         }
     936           6 :         src2[0] = base;
     937             :     }
     938             : 
     939       35969 :     if (c->srcXYZ && !(c->dstXYZ && c->srcW==c->dstW && c->srcH==c->dstH)) {
     940             :         uint8_t *base;
     941           2 :         rgb0_tmp = av_malloc(FFABS(srcStride[0]) * srcSliceH + 32);
     942           2 :         if (!rgb0_tmp)
     943           0 :             return AVERROR(ENOMEM);
     944             : 
     945           2 :         base = srcStride[0] < 0 ? rgb0_tmp - srcStride[0] * (srcSliceH-1) : rgb0_tmp;
     946             : 
     947           2 :         xyz12Torgb48(c, (uint16_t*)base, (const uint16_t*)src2[0], srcStride[0]/2, srcSliceH);
     948           2 :         src2[0] = base;
     949             :     }
     950             : 
     951       35969 :     if (!srcSliceY && (c->flags & SWS_BITEXACT) && c->dither == SWS_DITHER_ED && c->dither_error[0])
     952          25 :         for (i = 0; i < 4; i++)
     953          20 :             memset(c->dither_error[i], 0, sizeof(c->dither_error[0][0]) * (c->dstW+2));
     954             : 
     955       35969 :     if (c->sliceDir != 1) {
     956             :         // slices go from bottom to top => we flip the image internally
     957           0 :         for (i=0; i<4; i++) {
     958           0 :             srcStride2[i] *= -1;
     959           0 :             dstStride2[i] *= -1;
     960             :         }
     961             : 
     962           0 :         src2[0] += (srcSliceH - 1) * srcStride[0];
     963           0 :         if (!usePal(c->srcFormat))
     964           0 :             src2[1] += ((srcSliceH >> c->chrSrcVSubSample) - 1) * srcStride[1];
     965           0 :         src2[2] += ((srcSliceH >> c->chrSrcVSubSample) - 1) * srcStride[2];
     966           0 :         src2[3] += (srcSliceH - 1) * srcStride[3];
     967           0 :         dst2[0] += ( c->dstH                         - 1) * dstStride[0];
     968           0 :         dst2[1] += ((c->dstH >> c->chrDstVSubSample) - 1) * dstStride[1];
     969           0 :         dst2[2] += ((c->dstH >> c->chrDstVSubSample) - 1) * dstStride[2];
     970           0 :         dst2[3] += ( c->dstH                         - 1) * dstStride[3];
     971             : 
     972           0 :         srcSliceY_internal = c->srcH-srcSliceY-srcSliceH;
     973             :     }
     974       35969 :     reset_ptr(src2, c->srcFormat);
     975       35969 :     reset_ptr((void*)dst2, c->dstFormat);
     976             : 
     977             :     /* reset slice direction at end of frame */
     978       35969 :     if (srcSliceY_internal + srcSliceH == c->srcH)
     979       35969 :         c->sliceDir = 0;
     980       35969 :     ret = c->swscale(c, src2, srcStride2, srcSliceY_internal, srcSliceH, dst2, dstStride2);
     981             : 
     982             : 
     983       35969 :     if (c->dstXYZ && !(c->srcXYZ && c->srcW==c->dstW && c->srcH==c->dstH)) {
     984          31 :         int dstY = c->dstY ? c->dstY : srcSliceY + srcSliceH;
     985          31 :         uint16_t *dst16 = (uint16_t*)(dst2[0] + (dstY - ret) * dstStride2[0]);
     986          31 :         av_assert0(dstY >= ret);
     987          31 :         av_assert0(ret >= 0);
     988          31 :         av_assert0(c->dstH >= dstY);
     989             : 
     990             :         /* replace on the same data */
     991          31 :         rgb48Toxyz12(c, dst16, dst16, dstStride2[0]/2, ret);
     992             :     }
     993             : 
     994       35969 :     av_free(rgb0_tmp);
     995       35969 :     return ret;
     996             : }

Generated by: LCOV version 1.12