Merge commit '69ab9f53f901eac6a649e22d28cf093357870627'
authorMichael Niedermayer <michaelni@gmx.at>
Thu, 9 Jul 2015 03:04:13 +0000 (05:04 +0200)
committerMichael Niedermayer <michaelni@gmx.at>
Thu, 9 Jul 2015 03:06:48 +0000 (05:06 +0200)
* commit '69ab9f53f901eac6a649e22d28cf093357870627':
  hevc: split bitstream unescaping to a separate file

Conflicts:
libavcodec/Makefile
libavcodec/hevc.c

See: afa93d198aaf2cc661c4df6d4095cd030265d30a
Merged-by: Michael Niedermayer <michaelni@gmx.at>
1  2 
libavcodec/Makefile
libavcodec/hevc.c
libavcodec/hevc.h
libavcodec/hevc_parse.c

diff --combined libavcodec/Makefile
@@@ -1,5 -1,3 +1,5 @@@
 +include $(SUBDIR)../config.mak
 +
  NAME = avcodec
  
  HEADERS = avcodec.h                                                     \
@@@ -7,7 -5,6 +7,7 @@@
            dv_profile.h                                                  \
            d3d11va.h                                                     \
            dxva2.h                                                       \
 +          old_codec_ids.h                                               \
            qsv.h                                                         \
            vaapi.h                                                       \
            vda.h                                                         \
@@@ -17,8 -14,6 +17,8 @@@
            xvmc.h                                                        \
  
  OBJS = allcodecs.o                                                      \
 +       audioconvert.o                                                   \
 +       avdct.o                                                          \
         avpacket.o                                                       \
         avpicture.o                                                      \
         bitstream.o                                                      \
         codec_desc.o                                                     \
         dv_profile.o                                                     \
         imgconvert.o                                                     \
 -       log2_tab.o                                                       \
         mathtables.o                                                     \
         options.o                                                        \
         parser.o                                                         \
         qsv_api.o                                                        \
         raw.o                                                            \
 +       resample.o                                                       \
 +       resample2.o                                                      \
         utils.o                                                          \
         vorbis_parser.o                                                  \
         xiph.o                                                           \
@@@ -45,16 -39,13 +45,16 @@@ OBJS-$(CONFIG_AUDIODSP)                
  OBJS-$(CONFIG_BLOCKDSP)                += blockdsp.o
  OBJS-$(CONFIG_BSWAPDSP)                += bswapdsp.o
  OBJS-$(CONFIG_CABAC)                   += cabac.o
 +OBJS-$(CONFIG_CRYSTALHD)               += crystalhd.o
  OBJS-$(CONFIG_DCT)                     += dct.o dct32_fixed.o dct32_float.o
  OBJS-$(CONFIG_ERROR_RESILIENCE)        += error_resilience.o
 +OBJS-$(CONFIG_EXIF)                    += exif.o tiff_common.o
  OBJS-$(CONFIG_FAANDCT)                 += faandct.o
  OBJS-$(CONFIG_FAANIDCT)                += faanidct.o
  OBJS-$(CONFIG_FDCTDSP)                 += fdctdsp.o jfdctfst.o jfdctint.o
  FFT-OBJS-$(CONFIG_HARDCODED_TABLES)    += cos_tables.o cos_fixed_tables.o
  OBJS-$(CONFIG_FFT)                     += avfft.o fft_fixed.o fft_float.o \
 +                                          fft_fixed_32.o fft_init_table.o \
                                            $(FFT-OBJS-yes)
  OBJS-$(CONFIG_FMTCONVERT)              += fmtconvert.o
  OBJS-$(CONFIG_GOLOMB)                  += golomb.o
@@@ -73,11 -64,9 +73,11 @@@ OBJS-$(CONFIG_IMDCT15)                 
  OBJS-$(CONFIG_INTRAX8)                 += intrax8.o intrax8dsp.o
  OBJS-$(CONFIG_JPEGTABLES)              += jpegtables.o
  OBJS-$(CONFIG_LIBXVID)                 += libxvid_rc.o
 +OBJS-$(CONFIG_LLAUDDSP)                += lossless_audiodsp.o
 +OBJS-$(CONFIG_LLVIDDSP)                += lossless_videodsp.o
  OBJS-$(CONFIG_LPC)                     += lpc.o
  OBJS-$(CONFIG_LSP)                     += lsp.o
 -OBJS-$(CONFIG_MDCT)                    += mdct_fixed.o mdct_float.o
 +OBJS-$(CONFIG_MDCT)                    += mdct_fixed.o mdct_float.o mdct_fixed_32.o
  OBJS-$(CONFIG_ME_CMP)                  += me_cmp.o
  OBJS-$(CONFIG_MPEG_ER)                 += mpeg_er.o
  OBJS-$(CONFIG_MPEGAUDIO)               += mpegaudio.o mpegaudiodata.o   \
@@@ -92,8 -81,8 +92,8 @@@ OBJS-$(CONFIG_MPEGVIDEO)               
  OBJS-$(CONFIG_MPEGVIDEOENC)            += mpegvideo_enc.o mpeg12data.o  \
                                            motion_est.o ratecontrol.o    \
                                            mpegvideoencdsp.o
 +OBJS-$(CONFIG_NVENC)                   += nvenc.o
  OBJS-$(CONFIG_PIXBLOCKDSP)             += pixblockdsp.o
 -OBJS-${CONFIG_NVENC}                   += nvenc.o
  OBJS-$(CONFIG_QPELDSP)                 += qpeldsp.o
  OBJS-$(CONFIG_QSV)                     += qsv.o
  OBJS-$(CONFIG_QSVDEC)                  += qsvdec.o
@@@ -101,7 -90,6 +101,7 @@@ OBJS-$(CONFIG_QSVENC)                  
  OBJS-$(CONFIG_RANGECODER)              += rangecoder.o
  RDFT-OBJS-$(CONFIG_HARDCODED_TABLES)   += sin_tables.o
  OBJS-$(CONFIG_RDFT)                    += rdft.o $(RDFT-OBJS-yes)
 +OBJS-$(CONFIG_SHARED)                  += log2_tab.o
  OBJS-$(CONFIG_SINEWIN)                 += sinewin.o
  OBJS-$(CONFIG_SNAPPY)                  += snappy.o
  OBJS-$(CONFIG_STARTCODE)               += startcode.o
@@@ -113,7 -101,6 +113,7 @@@ OBJS-$(CONFIG_VP3DSP)                  
  OBJS-$(CONFIG_WMA_FREQS)               += wma_freqs.o
  
  # decoders/encoders
 +OBJS-$(CONFIG_ZERO12V_DECODER)         += 012v.o
  OBJS-$(CONFIG_A64MULTI_ENCODER)        += a64multienc.o elbg.o
  OBJS-$(CONFIG_A64MULTI5_ENCODER)       += a64multienc.o elbg.o
  OBJS-$(CONFIG_AAC_DECODER)             += aacdec.o aactab.o aacsbr.o aacps.o \
@@@ -123,8 -110,7 +123,8 @@@ OBJS-$(CONFIG_AAC_ENCODER)             
                                            aacpsy.o aactab.o      \
                                            psymodel.o mpeg4audio.o kbdwin.o
  OBJS-$(CONFIG_AASC_DECODER)            += aasc.o msrledec.o
 -OBJS-$(CONFIG_AC3_DECODER)             += ac3dec.o ac3dec_data.o ac3.o kbdwin.o
 +OBJS-$(CONFIG_AC3_DECODER)             += ac3dec_float.o ac3dec_data.o ac3.o kbdwin.o
 +OBJS-$(CONFIG_AC3_FIXED_DECODER)       += ac3dec_fixed.o ac3dec_data.o ac3.o kbdwin.o
  OBJS-$(CONFIG_AC3_ENCODER)             += ac3enc_float.o ac3enc.o ac3tab.o \
                                            ac3.o kbdwin.o
  OBJS-$(CONFIG_AC3_FIXED_ENCODER)       += ac3enc_fixed.o ac3enc.o ac3tab.o ac3.o
@@@ -142,18 -128,10 +142,18 @@@ OBJS-$(CONFIG_AMRWB_DECODER)           
                                            celp_math.o acelp_filters.o \
                                            acelp_vectors.o             \
                                            acelp_pitch_delay.o
 +OBJS-$(CONFIG_AMV_ENCODER)             += mjpegenc.o mjpegenc_common.o \
 +                                          mpegvideo_enc.o motion_est.o \
 +                                          ratecontrol.o mpeg12data.o   \
 +                                          mpegvideo.o
  OBJS-$(CONFIG_ANM_DECODER)             += anm.o
  OBJS-$(CONFIG_ANSI_DECODER)            += ansi.o cga_data.o
  OBJS-$(CONFIG_APE_DECODER)             += apedec.o
 -OBJS-$(CONFIG_ASS_DECODER)             += assdec.o ass.o
 +OBJS-$(CONFIG_APNG_DECODER)            += png.o pngdec.o pngdsp.o
 +OBJS-$(CONFIG_APNG_ENCODER)            += png.o pngenc.o
 +OBJS-$(CONFIG_SSA_DECODER)             += assdec.o ass.o ass_split.o
 +OBJS-$(CONFIG_SSA_ENCODER)             += assenc.o ass.o
 +OBJS-$(CONFIG_ASS_DECODER)             += assdec.o ass.o ass_split.o
  OBJS-$(CONFIG_ASS_ENCODER)             += assenc.o ass.o
  OBJS-$(CONFIG_ASV1_DECODER)            += asvdec.o asv.o mpeg12data.o
  OBJS-$(CONFIG_ASV1_ENCODER)            += asvenc.o asv.o mpeg12data.o
@@@ -165,20 -143,12 +165,20 @@@ OBJS-$(CONFIG_ATRAC3P_DECODER)         
                                            atrac3plusdsp.o atrac.o
  OBJS-$(CONFIG_AURA_DECODER)            += cyuv.o
  OBJS-$(CONFIG_AURA2_DECODER)           += aura.o
 +OBJS-$(CONFIG_AVRN_DECODER)            += avrndec.o mjpegdec.o
 +OBJS-$(CONFIG_AVRP_DECODER)            += r210dec.o
 +OBJS-$(CONFIG_AVRP_ENCODER)            += r210enc.o
  OBJS-$(CONFIG_AVS_DECODER)             += avs.o
 +OBJS-$(CONFIG_AVUI_DECODER)            += avuidec.o
 +OBJS-$(CONFIG_AVUI_ENCODER)            += avuienc.o
 +OBJS-$(CONFIG_AYUV_DECODER)            += v408dec.o
 +OBJS-$(CONFIG_AYUV_ENCODER)            += v408enc.o
  OBJS-$(CONFIG_BETHSOFTVID_DECODER)     += bethsoftvideo.o
  OBJS-$(CONFIG_BFI_DECODER)             += bfi.o
  OBJS-$(CONFIG_BINK_DECODER)            += bink.o binkdsp.o
  OBJS-$(CONFIG_BINKAUDIO_DCT_DECODER)   += binkaudio.o
  OBJS-$(CONFIG_BINKAUDIO_RDFT_DECODER)  += binkaudio.o
 +OBJS-$(CONFIG_BINTEXT_DECODER)         += bintext.o cga_data.o
  OBJS-$(CONFIG_BMP_DECODER)             += bmp.o msrledec.o
  OBJS-$(CONFIG_BMP_ENCODER)             += bmpenc.o
  OBJS-$(CONFIG_BMV_AUDIO_DECODER)       += bmvaudio.o
@@@ -187,36 -157,26 +187,36 @@@ OBJS-$(CONFIG_BRENDER_PIX_DECODER)     
  OBJS-$(CONFIG_C93_DECODER)             += c93.o
  OBJS-$(CONFIG_CAVS_DECODER)            += cavs.o cavsdec.o cavsdsp.o \
                                            cavsdata.o mpeg12data.o
 +OBJS-$(CONFIG_CCAPTION_DECODER)        += ccaption_dec.o
  OBJS-$(CONFIG_CDGRAPHICS_DECODER)      += cdgraphics.o
  OBJS-$(CONFIG_CDXL_DECODER)            += cdxl.o
  OBJS-$(CONFIG_CINEPAK_DECODER)         += cinepak.o
 +OBJS-$(CONFIG_CINEPAK_ENCODER)         += cinepakenc.o elbg.o
  OBJS-$(CONFIG_CLJR_DECODER)            += cljrdec.o
  OBJS-$(CONFIG_CLJR_ENCODER)            += cljrenc.o
  OBJS-$(CONFIG_CLLC_DECODER)            += cllc.o canopus.o
  OBJS-$(CONFIG_COOK_DECODER)            += cook.o
  OBJS-$(CONFIG_COMFORTNOISE_DECODER)    += cngdec.o celp_filters.o
  OBJS-$(CONFIG_COMFORTNOISE_ENCODER)    += cngenc.o
 +OBJS-$(CONFIG_CPIA_DECODER)            += cpia.o
  OBJS-$(CONFIG_CSCD_DECODER)            += cscd.o
  OBJS-$(CONFIG_CYUV_DECODER)            += cyuv.o
  OBJS-$(CONFIG_DCA_DECODER)             += dcadec.o dca.o dcadsp.o      \
                                            dcadata.o dca_exss.o         \
                                            dca_xll.o synth_filter.o
 +OBJS-$(CONFIG_DCA_ENCODER)             += dcaenc.o dca.o dcadata.o
  OBJS-$(CONFIG_DDS_DECODER)             += dds.o
 +OBJS-$(CONFIG_DIRAC_DECODER)           += diracdec.o dirac.o diracdsp.o \
 +                                          dirac_arith.o mpeg12data.o dirac_dwt.o
  OBJS-$(CONFIG_DFA_DECODER)             += dfa.o
  OBJS-$(CONFIG_DNXHD_DECODER)           += dnxhddec.o dnxhddata.o
  OBJS-$(CONFIG_DNXHD_ENCODER)           += dnxhdenc.o dnxhddata.o
  OBJS-$(CONFIG_DPX_DECODER)             += dpx.o
  OBJS-$(CONFIG_DPX_ENCODER)             += dpxenc.o
 +OBJS-$(CONFIG_DSD_LSBF_DECODER)        += dsddec.o
 +OBJS-$(CONFIG_DSD_MSBF_DECODER)        += dsddec.o
 +OBJS-$(CONFIG_DSD_LSBF_PLANAR_DECODER) += dsddec.o
 +OBJS-$(CONFIG_DSD_MSBF_PLANAR_DECODER) += dsddec.o
  OBJS-$(CONFIG_DSICINAUDIO_DECODER)     += dsicinaudio.o
  OBJS-$(CONFIG_DSICINVIDEO_DECODER)     += dsicinvideo.o
  OBJS-$(CONFIG_DSS_SP_DECODER)          += dss_sp.o
@@@ -228,7 -188,7 +228,7 @@@ OBJS-$(CONFIG_DVVIDEO_DECODER)         
  OBJS-$(CONFIG_DVVIDEO_ENCODER)         += dvenc.o dv.o dvdata.o
  OBJS-$(CONFIG_DXA_DECODER)             += dxa.o
  OBJS-$(CONFIG_DXTORY_DECODER)          += dxtory.o
 -OBJS-$(CONFIG_EAC3_DECODER)            += eac3dec.o eac3_data.o
 +OBJS-$(CONFIG_EAC3_DECODER)            += eac3_data.o
  OBJS-$(CONFIG_EAC3_ENCODER)            += eac3enc.o eac3_data.o
  OBJS-$(CONFIG_EACMV_DECODER)           += eacmv.o
  OBJS-$(CONFIG_EAMAD_DECODER)           += eamad.o eaidct.o mpeg12.o \
@@@ -241,17 -201,14 +241,17 @@@ OBJS-$(CONFIG_EIGHTSVX_EXP_DECODER)    
  OBJS-$(CONFIG_EIGHTSVX_FIB_DECODER)    += 8svx.o
  OBJS-$(CONFIG_ESCAPE124_DECODER)       += escape124.o
  OBJS-$(CONFIG_ESCAPE130_DECODER)       += escape130.o
 +OBJS-$(CONFIG_EVRC_DECODER)            += evrcdec.o acelp_vectors.o lsp.o
  OBJS-$(CONFIG_EXR_DECODER)             += exr.o
  OBJS-$(CONFIG_FFV1_DECODER)            += ffv1dec.o ffv1.o
  OBJS-$(CONFIG_FFV1_ENCODER)            += ffv1enc.o ffv1.o
 +OBJS-$(CONFIG_FFWAVESYNTH_DECODER)     += ffwavesynth.o
  OBJS-$(CONFIG_FIC_DECODER)             += fic.o
  OBJS-$(CONFIG_FLAC_DECODER)            += flacdec.o flacdata.o flac.o flacdsp.o
 -OBJS-$(CONFIG_FLAC_ENCODER)            += flacenc.o flacdata.o flac.o flacdsp.o
 +OBJS-$(CONFIG_FLAC_ENCODER)            += flacenc.o flacdata.o flac.o flacdsp.o vorbis_data.o
  OBJS-$(CONFIG_FLASHSV_DECODER)         += flashsv.o
  OBJS-$(CONFIG_FLASHSV_ENCODER)         += flashsvenc.o
 +OBJS-$(CONFIG_FLASHSV2_ENCODER)        += flashsv2enc.o
  OBJS-$(CONFIG_FLASHSV2_DECODER)        += flashsv.o
  OBJS-$(CONFIG_FLIC_DECODER)            += flicvideo.o
  OBJS-$(CONFIG_FOURXM_DECODER)          += 4xm.o
@@@ -259,9 -216,7 +259,9 @@@ OBJS-$(CONFIG_FRAPS_DECODER)           
  OBJS-$(CONFIG_FRWU_DECODER)            += frwu.o
  OBJS-$(CONFIG_G2M_DECODER)             += g2meet.o elsdec.o
  OBJS-$(CONFIG_G723_1_DECODER)          += g723_1.o acelp_vectors.o \
 -                                          celp_filters.o
 +                                          celp_filters.o celp_math.o
 +OBJS-$(CONFIG_G723_1_ENCODER)          += g723_1.o acelp_vectors.o celp_math.o
 +OBJS-$(CONFIG_G729_DECODER)            += g729dec.o lsp.o celp_math.o acelp_filters.o acelp_pitch_delay.o acelp_vectors.o g729postfilter.o
  OBJS-$(CONFIG_GIF_DECODER)             += gifdec.o lzw.o
  OBJS-$(CONFIG_GIF_ENCODER)             += gif.o lzwenc.o
  OBJS-$(CONFIG_GSM_DECODER)             += gsmdec.o gsmdec_data.o msgsmdec.o
@@@ -278,14 -233,15 +278,14 @@@ OBJS-$(CONFIG_H264_DECODER)            
                                            h264_mb.o h264_picture.o h264_ps.o \
                                            h264_refs.o h264_sei.o h264_slice.o
  OBJS-$(CONFIG_H264_MMAL_DECODER)       += mmaldec.o
 -OBJS-$(CONFIG_H264_NVENC_ENCODER)      += nvenc_h264.o
 +OBJS-$(CONFIG_H264_VDA_DECODER)        += vda_h264_dec.o
  OBJS-$(CONFIG_H264_QSV_DECODER)        += qsvdec_h264.o
  OBJS-$(CONFIG_H264_QSV_ENCODER)        += qsvenc_h264.o
  OBJS-$(CONFIG_HAP_DECODER)             += hapdec.o
  OBJS-$(CONFIG_HAP_ENCODER)             += hapenc.o
  OBJS-$(CONFIG_HEVC_DECODER)            += hevc.o hevc_mvs.o hevc_ps.o hevc_sei.o \
                                            hevc_cabac.o hevc_refs.o hevcpred.o    \
-                                           hevcdsp.o hevc_filter.o
+                                           hevcdsp.o hevc_filter.o hevc_parse.o
 -OBJS-$(CONFIG_HEVC_NVENC_ENCODER)      += nvenc_hevc.o
  OBJS-$(CONFIG_HNM4_VIDEO_DECODER)      += hnm4video.o
  OBJS-$(CONFIG_HQ_HQA_DECODER)          += hq_hqa.o hq_hqadata.o hq_hqadsp.o \
                                            canopus.o
@@@ -293,7 -249,6 +293,7 @@@ OBJS-$(CONFIG_HQX_DECODER)             
  OBJS-$(CONFIG_HUFFYUV_DECODER)         += huffyuv.o huffyuvdec.o
  OBJS-$(CONFIG_HUFFYUV_ENCODER)         += huffyuv.o huffyuvenc.o
  OBJS-$(CONFIG_IDCIN_DECODER)           += idcinvideo.o
 +OBJS-$(CONFIG_IDF_DECODER)             += bintext.o cga_data.o
  OBJS-$(CONFIG_IFF_BYTERUN1_DECODER)    += iff.o
  OBJS-$(CONFIG_IFF_ILBM_DECODER)        += iff.o
  OBJS-$(CONFIG_IMC_DECODER)             += imc.o
@@@ -303,9 -258,6 +303,9 @@@ OBJS-$(CONFIG_INDEO4_DECODER)          
  OBJS-$(CONFIG_INDEO5_DECODER)          += indeo5.o ivi.o ivi_dsp.o
  OBJS-$(CONFIG_INTERPLAY_DPCM_DECODER)  += dpcm.o
  OBJS-$(CONFIG_INTERPLAY_VIDEO_DECODER) += interplayvideo.o
 +OBJS-$(CONFIG_JACOSUB_DECODER)         += jacosubdec.o ass.o
 +OBJS-$(CONFIG_JPEG2000_ENCODER)        += j2kenc.o mqcenc.o mqc.o jpeg2000.o \
 +                                          jpeg2000dwt.o
  OBJS-$(CONFIG_JPEG2000_DECODER)        += jpeg2000dec.o jpeg2000.o jpeg2000dsp.o \
                                            jpeg2000dwt.o mqcdec.o mqc.o
  OBJS-$(CONFIG_JPEGLS_DECODER)          += jpeglsdec.o jpegls.o
@@@ -321,7 -273,6 +321,7 @@@ OBJS-$(CONFIG_MACE6_DECODER)           
  OBJS-$(CONFIG_MDEC_DECODER)            += mdec.o mpeg12.o mpeg12data.o
  OBJS-$(CONFIG_METASOUND_DECODER)       += metasound.o metasound_data.o \
                                            twinvq.o
 +OBJS-$(CONFIG_MICRODVD_DECODER)        += microdvddec.o ass.o
  OBJS-$(CONFIG_MIMIC_DECODER)           += mimic.o
  OBJS-$(CONFIG_MJPEG_DECODER)           += mjpegdec.o
  OBJS-$(CONFIG_MJPEG_ENCODER)           += mjpegenc.o mjpegenc_common.o
@@@ -329,14 -280,10 +329,14 @@@ OBJS-$(CONFIG_MJPEGB_DECODER)          
  OBJS-$(CONFIG_MLP_DECODER)             += mlpdec.o mlpdsp.o
  OBJS-$(CONFIG_MMVIDEO_DECODER)         += mmvideo.o
  OBJS-$(CONFIG_MOTIONPIXELS_DECODER)    += motionpixels.o
 +OBJS-$(CONFIG_MOVTEXT_DECODER)         += movtextdec.o ass.o
 +OBJS-$(CONFIG_MOVTEXT_ENCODER)         += movtextenc.o ass_split.o
  OBJS-$(CONFIG_MP1_DECODER)             += mpegaudiodec_fixed.o
  OBJS-$(CONFIG_MP1FLOAT_DECODER)        += mpegaudiodec_float.o
  OBJS-$(CONFIG_MP2_DECODER)             += mpegaudiodec_fixed.o
 -OBJS-$(CONFIG_MP2_ENCODER)             += mpegaudioenc.o mpegaudio.o \
 +OBJS-$(CONFIG_MP2_ENCODER)             += mpegaudioenc_float.o mpegaudio.o \
 +                                          mpegaudiodata.o mpegaudiodsp_data.o
 +OBJS-$(CONFIG_MP2FIXED_ENCODER)        += mpegaudioenc_fixed.o mpegaudio.o \
                                            mpegaudiodata.o mpegaudiodsp_data.o
  OBJS-$(CONFIG_MP2FLOAT_DECODER)        += mpegaudiodec_float.o
  OBJS-$(CONFIG_MP3_DECODER)             += mpegaudiodec_fixed.o
@@@ -347,13 -294,12 +347,13 @@@ OBJS-$(CONFIG_MP3ON4_DECODER)          
  OBJS-$(CONFIG_MP3ON4FLOAT_DECODER)     += mpegaudiodec_float.o mpeg4audio.o
  OBJS-$(CONFIG_MPC7_DECODER)            += mpc7.o mpc.o
  OBJS-$(CONFIG_MPC8_DECODER)            += mpc8.o mpc.o
 -OBJS-$(CONFIG_MPEG_XVMC_DECODER)       += mpegvideo_xvmc.o
 +OBJS-$(CONFIG_MPEGVIDEO_DECODER)       += mpeg12dec.o mpeg12.o mpeg12data.o
  OBJS-$(CONFIG_MPEG1VIDEO_DECODER)      += mpeg12dec.o mpeg12.o mpeg12data.o
  OBJS-$(CONFIG_MPEG1VIDEO_ENCODER)      += mpeg12enc.o mpeg12.o
  OBJS-$(CONFIG_MPEG2VIDEO_DECODER)      += mpeg12dec.o mpeg12.o mpeg12data.o
  OBJS-$(CONFIG_MPEG2VIDEO_ENCODER)      += mpeg12enc.o mpeg12.o
  OBJS-$(CONFIG_MPEG4_DECODER)           += xvididct.o
 +OBJS-$(CONFIG_MPL2_DECODER)            += mpl2dec.o ass.o
  OBJS-$(CONFIG_MSMPEG4V1_DECODER)       += msmpeg4dec.o msmpeg4.o msmpeg4data.o
  OBJS-$(CONFIG_MSMPEG4V2_DECODER)       += msmpeg4dec.o msmpeg4.o msmpeg4data.o
  OBJS-$(CONFIG_MSMPEG4V2_ENCODER)       += msmpeg4enc.o msmpeg4.o msmpeg4data.o
@@@ -364,7 -310,6 +364,7 @@@ OBJS-$(CONFIG_MSA1_DECODER)            
  OBJS-$(CONFIG_MSS1_DECODER)            += mss1.o mss12.o
  OBJS-$(CONFIG_MSS2_DECODER)            += mss2.o mss12.o mss2dsp.o
  OBJS-$(CONFIG_MSVIDEO1_DECODER)        += msvideo1.o
 +OBJS-$(CONFIG_MSVIDEO1_ENCODER)        += msvideo1enc.o elbg.o
  OBJS-$(CONFIG_MSZH_DECODER)            += lcldec.o
  OBJS-$(CONFIG_MTS2_DECODER)            += mss4.o mss34dsp.o
  OBJS-$(CONFIG_MVC1_DECODER)            += mvcdec.o
@@@ -390,16 -335,12 +390,16 @@@ OBJS-$(CONFIG_PGMYUV_DECODER)          
  OBJS-$(CONFIG_PGMYUV_ENCODER)          += pnmenc.o
  OBJS-$(CONFIG_PGSSUB_DECODER)          += pgssubdec.o
  OBJS-$(CONFIG_PICTOR_DECODER)          += pictordec.o cga_data.o
 +OBJS-$(CONFIG_PJS_DECODER)             += textdec.o ass.o
  OBJS-$(CONFIG_PNG_DECODER)             += png.o pngdec.o pngdsp.o
  OBJS-$(CONFIG_PNG_ENCODER)             += png.o pngenc.o
  OBJS-$(CONFIG_PPM_DECODER)             += pnmdec.o pnm.o
  OBJS-$(CONFIG_PPM_ENCODER)             += pnmenc.o
 -OBJS-$(CONFIG_PRORES_DECODER)          += proresdec.o proresdata.o proresdsp.o
 -OBJS-$(CONFIG_PRORES_ENCODER)          += proresenc.o proresdata.o
 +OBJS-$(CONFIG_PRORES_DECODER)          += proresdec2.o proresdsp.o proresdata.o
 +OBJS-$(CONFIG_PRORES_LGPL_DECODER)     += proresdec_lgpl.o proresdsp.o proresdata.o
 +OBJS-$(CONFIG_PRORES_ENCODER)          += proresenc_anatoliy.o
 +OBJS-$(CONFIG_PRORES_AW_ENCODER)       += proresenc_anatoliy.o
 +OBJS-$(CONFIG_PRORES_KS_ENCODER)       += proresenc_kostya.o proresdata.o
  OBJS-$(CONFIG_PTX_DECODER)             += ptx.o
  OBJS-$(CONFIG_QCELP_DECODER)           += qcelpdec.o                     \
                                            celp_filters.o acelp_vectors.o \
@@@ -410,16 -351,13 +410,16 @@@ OBJS-$(CONFIG_QPEG_DECODER)            
  OBJS-$(CONFIG_QTRLE_DECODER)           += qtrle.o
  OBJS-$(CONFIG_QTRLE_ENCODER)           += qtrleenc.o
  OBJS-$(CONFIG_R10K_DECODER)            += r210dec.o
 +OBJS-$(CONFIG_R10K_ENCODER)            += r210enc.o
  OBJS-$(CONFIG_R210_DECODER)            += r210dec.o
 +OBJS-$(CONFIG_R210_ENCODER)            += r210enc.o
  OBJS-$(CONFIG_RA_144_DECODER)          += ra144dec.o ra144.o celp_filters.o
  OBJS-$(CONFIG_RA_144_ENCODER)          += ra144enc.o ra144.o celp_filters.o
  OBJS-$(CONFIG_RA_288_DECODER)          += ra288.o celp_filters.o
  OBJS-$(CONFIG_RALF_DECODER)            += ralf.o
  OBJS-$(CONFIG_RAWVIDEO_DECODER)        += rawdec.o
  OBJS-$(CONFIG_RAWVIDEO_ENCODER)        += rawenc.o
 +OBJS-$(CONFIG_REALTEXT_DECODER)        += realtextdec.o ass.o
  OBJS-$(CONFIG_RL2_DECODER)             += rl2.o
  OBJS-$(CONFIG_ROQ_DECODER)             += roqvideodec.o roqvideo.o
  OBJS-$(CONFIG_ROQ_ENCODER)             += roqvideoenc.o roqvideo.o elbg.o
@@@ -432,9 -370,7 +432,9 @@@ OBJS-$(CONFIG_RV20_DECODER)            
  OBJS-$(CONFIG_RV20_ENCODER)            += rv20enc.o
  OBJS-$(CONFIG_RV30_DECODER)            += rv30.o rv34.o rv30dsp.o rv34dsp.o
  OBJS-$(CONFIG_RV40_DECODER)            += rv40.o rv34.o rv34dsp.o rv40dsp.o
 +OBJS-$(CONFIG_SAMI_DECODER)            += samidec.o ass.o
  OBJS-$(CONFIG_S302M_DECODER)           += s302m.o
 +OBJS-$(CONFIG_S302M_ENCODER)           += s302menc.o
  OBJS-$(CONFIG_SANM_DECODER)            += sanm.o
  OBJS-$(CONFIG_SGI_DECODER)             += sgidec.o
  OBJS-$(CONFIG_SGI_ENCODER)             += sgienc.o rle.o
@@@ -447,46 -383,29 +447,46 @@@ OBJS-$(CONFIG_SIPR_DECODER)            
  OBJS-$(CONFIG_SMACKAUD_DECODER)        += smacker.o
  OBJS-$(CONFIG_SMACKER_DECODER)         += smacker.o
  OBJS-$(CONFIG_SMC_DECODER)             += smc.o
 +OBJS-$(CONFIG_SMVJPEG_DECODER)         += smvjpegdec.o
 +OBJS-$(CONFIG_SNOW_DECODER)            += snowdec.o snow.o snow_dwt.o
 +OBJS-$(CONFIG_SNOW_ENCODER)            += snowenc.o snow.o snow_dwt.o             \
 +                                          h263.o ituh263enc.o
  OBJS-$(CONFIG_SOL_DPCM_DECODER)        += dpcm.o
 +OBJS-$(CONFIG_SONIC_DECODER)           += sonic.o
 +OBJS-$(CONFIG_SONIC_ENCODER)           += sonic.o
 +OBJS-$(CONFIG_SONIC_LS_ENCODER)        += sonic.o
  OBJS-$(CONFIG_SP5X_DECODER)            += sp5xdec.o
  OBJS-$(CONFIG_SRT_DECODER)             += srtdec.o ass.o
 +OBJS-$(CONFIG_SRT_ENCODER)             += srtenc.o ass_split.o
 +OBJS-$(CONFIG_STL_DECODER)             += textdec.o ass.o
 +OBJS-$(CONFIG_SUBRIP_DECODER)          += srtdec.o ass.o
 +OBJS-$(CONFIG_SUBRIP_ENCODER)          += srtenc.o ass_split.o
 +OBJS-$(CONFIG_SUBVIEWER1_DECODER)      += textdec.o ass.o
 +OBJS-$(CONFIG_SUBVIEWER_DECODER)       += subviewerdec.o ass.o
  OBJS-$(CONFIG_SUNRAST_DECODER)         += sunrast.o
  OBJS-$(CONFIG_SUNRAST_ENCODER)         += sunrastenc.o
  OBJS-$(CONFIG_SVQ1_DECODER)            += svq1dec.o svq1.o svq13.o h263.o
  OBJS-$(CONFIG_SVQ1_ENCODER)            += svq1enc.o svq1.o    \
                                            h263.o ituh263enc.o
  OBJS-$(CONFIG_SVQ3_DECODER)            += svq3.o svq13.o mpegutils.o
 +OBJS-$(CONFIG_TEXT_DECODER)            += textdec.o ass.o
  OBJS-$(CONFIG_TAK_DECODER)             += takdec.o tak.o
  OBJS-$(CONFIG_TARGA_DECODER)           += targa.o
  OBJS-$(CONFIG_TARGA_ENCODER)           += targaenc.o rle.o
 +OBJS-$(CONFIG_TARGA_Y216_DECODER)      += targa_y216dec.o
  OBJS-$(CONFIG_TDSC_DECODER)            += tdsc.o
  OBJS-$(CONFIG_TIERTEXSEQVIDEO_DECODER) += tiertexseqv.o
 -OBJS-$(CONFIG_TIFF_DECODER)            += tiff.o lzw.o faxcompr.o
 -OBJS-$(CONFIG_TIFF_ENCODER)            += tiffenc.o rle.o lzwenc.o
 +OBJS-$(CONFIG_TIFF_DECODER)            += tiff.o lzw.o faxcompr.o tiff_data.o tiff_common.o
 +OBJS-$(CONFIG_TIFF_ENCODER)            += tiffenc.o rle.o lzwenc.o tiff_data.o
  OBJS-$(CONFIG_TMV_DECODER)             += tmv.o cga_data.o
 +OBJS-$(CONFIG_TRUEHD_DECODER)          += mlpdec.o mlpdsp.o
  OBJS-$(CONFIG_TRUEMOTION1_DECODER)     += truemotion1.o
  OBJS-$(CONFIG_TRUEMOTION2_DECODER)     += truemotion2.o
  OBJS-$(CONFIG_TRUESPEECH_DECODER)      += truespeech.o
  OBJS-$(CONFIG_TSCC_DECODER)            += tscc.o msrledec.o
  OBJS-$(CONFIG_TSCC2_DECODER)           += tscc2.o
 -OBJS-$(CONFIG_TTA_DECODER)             += tta.o
 +OBJS-$(CONFIG_TTA_DECODER)             += tta.o ttadata.o ttadsp.o
 +OBJS-$(CONFIG_TTA_ENCODER)             += ttaenc.o ttadata.o
  OBJS-$(CONFIG_TWINVQ_DECODER)          += twinvqdec.o twinvq.o
  OBJS-$(CONFIG_TXD_DECODER)             += txd.o
  OBJS-$(CONFIG_ULTI_DECODER)            += ulti.o
@@@ -494,10 -413,6 +494,10 @@@ OBJS-$(CONFIG_UTVIDEO_DECODER)         
  OBJS-$(CONFIG_UTVIDEO_ENCODER)         += utvideoenc.o utvideo.o
  OBJS-$(CONFIG_V210_DECODER)            += v210dec.o
  OBJS-$(CONFIG_V210_ENCODER)            += v210enc.o
 +OBJS-$(CONFIG_V308_DECODER)            += v308dec.o
 +OBJS-$(CONFIG_V308_ENCODER)            += v308enc.o
 +OBJS-$(CONFIG_V408_DECODER)            += v408dec.o
 +OBJS-$(CONFIG_V408_ENCODER)            += v408enc.o
  OBJS-$(CONFIG_V410_DECODER)            += v410dec.o
  OBJS-$(CONFIG_V410_ENCODER)            += v410enc.o
  OBJS-$(CONFIG_V210X_DECODER)           += v210x.o
@@@ -506,8 -421,7 +506,8 @@@ OBJS-$(CONFIG_VBLE_DECODER)            
  OBJS-$(CONFIG_VC1_DECODER)             += vc1dec.o vc1_block.o vc1_loopfilter.o \
                                            vc1_mc.o vc1_pred.o vc1.o vc1data.o \
                                            vc1dsp.o \
 -                                          msmpeg4dec.o msmpeg4.o msmpeg4data.o
 +                                          msmpeg4dec.o msmpeg4.o msmpeg4data.o \
 +                                          wmv2dsp.o
  OBJS-$(CONFIG_VCR1_DECODER)            += vcr1.o
  OBJS-$(CONFIG_VMDAUDIO_DECODER)        += vmdaudio.o
  OBJS-$(CONFIG_VMDVIDEO_DECODER)        += vmdvideo.o
@@@ -523,16 -437,11 +523,16 @@@ OBJS-$(CONFIG_VP6_DECODER)             
                                            vp6dsp.o vp56rac.o
  OBJS-$(CONFIG_VP7_DECODER)             += vp8.o vp8dsp.o vp56rac.o
  OBJS-$(CONFIG_VP8_DECODER)             += vp8.o vp8dsp.o vp56rac.o
 -OBJS-$(CONFIG_VP9_DECODER)             += vp9.o vp9data.o vp9dsp.o \
 -                                          vp9block.o vp9prob.o vp9mvs.o vp56rac.o
 +OBJS-$(CONFIG_VP9_DECODER)             += vp9.o vp9dsp.o vp56rac.o vp9dsp_8bpp.o \
 +                                          vp9dsp_10bpp.o vp9dsp_12bpp.o
 +OBJS-$(CONFIG_VPLAYER_DECODER)         += textdec.o ass.o
  OBJS-$(CONFIG_VQA_DECODER)             += vqavideo.o
  OBJS-$(CONFIG_WAVPACK_DECODER)         += wavpack.o
 -OBJS-$(CONFIG_WEBP_DECODER)            += webp.o
 +OBJS-$(CONFIG_WAVPACK_ENCODER)         += wavpackenc.o
 +OBJS-$(CONFIG_WEBP_DECODER)            += vp8.o vp8dsp.o vp56rac.o
 +OBJS-$(CONFIG_WEBP_DECODER)            += webp.o exif.o tiff_common.o
 +OBJS-$(CONFIG_WEBVTT_DECODER)          += webvttdec.o ass.o
 +OBJS-$(CONFIG_WEBVTT_ENCODER)          += webvttenc.o ass_split.o
  OBJS-$(CONFIG_WMALOSSLESS_DECODER)     += wmalosslessdec.o wma_common.o
  OBJS-$(CONFIG_WMAPRO_DECODER)          += wmaprodec.o wma.o wma_common.o
  OBJS-$(CONFIG_WMAV1_DECODER)           += wmadec.o wma.o wma_common.o aactab.o
@@@ -543,7 -452,6 +543,7 @@@ OBJS-$(CONFIG_WMAVOICE_DECODER)        
                                            celp_filters.o \
                                            acelp_vectors.o acelp_filters.o
  OBJS-$(CONFIG_WMV1_DECODER)            += msmpeg4dec.o msmpeg4.o msmpeg4data.o
 +OBJS-$(CONFIG_WMV1_ENCODER)            += msmpeg4enc.o
  OBJS-$(CONFIG_WMV2_DECODER)            += wmv2dec.o wmv2.o wmv2dsp.o \
                                            msmpeg4dec.o msmpeg4.o msmpeg4data.o
  OBJS-$(CONFIG_WMV2_ENCODER)            += wmv2enc.o wmv2.o wmv2dsp.o \
@@@ -553,21 -461,14 +553,21 @@@ OBJS-$(CONFIG_WS_SND1_DECODER)         
  OBJS-$(CONFIG_XAN_DPCM_DECODER)        += dpcm.o
  OBJS-$(CONFIG_XAN_WC3_DECODER)         += xan.o
  OBJS-$(CONFIG_XAN_WC4_DECODER)         += xxan.o
 +OBJS-$(CONFIG_XBIN_DECODER)            += bintext.o cga_data.o
  OBJS-$(CONFIG_XBM_DECODER)             += xbmdec.o
  OBJS-$(CONFIG_XBM_ENCODER)             += xbmenc.o
 +OBJS-$(CONFIG_XFACE_DECODER)           += xfacedec.o xface.o
 +OBJS-$(CONFIG_XFACE_ENCODER)           += xfaceenc.o xface.o
  OBJS-$(CONFIG_XL_DECODER)              += xl.o
  OBJS-$(CONFIG_XSUB_DECODER)            += xsubdec.o
  OBJS-$(CONFIG_XSUB_ENCODER)            += xsubenc.o
  OBJS-$(CONFIG_XWD_DECODER)             += xwddec.o
  OBJS-$(CONFIG_XWD_ENCODER)             += xwdenc.o
 +OBJS-$(CONFIG_Y41P_DECODER)            += y41pdec.o
 +OBJS-$(CONFIG_Y41P_ENCODER)            += y41penc.o
  OBJS-$(CONFIG_YOP_DECODER)             += yop.o
 +OBJS-$(CONFIG_YUV4_DECODER)            += yuv4dec.o
 +OBJS-$(CONFIG_YUV4_ENCODER)            += yuv4enc.o
  OBJS-$(CONFIG_ZEROCODEC_DECODER)       += zerocodec.o
  OBJS-$(CONFIG_ZLIB_DECODER)            += lcldec.o
  OBJS-$(CONFIG_ZLIB_ENCODER)            += lclenc.o
@@@ -592,16 -493,13 +592,16 @@@ OBJS-$(CONFIG_PCM_MULAW_DECODER
  OBJS-$(CONFIG_PCM_MULAW_ENCODER)          += pcm.o
  OBJS-$(CONFIG_PCM_S8_DECODER)             += pcm.o
  OBJS-$(CONFIG_PCM_S8_ENCODER)             += pcm.o
 -OBJS-$(CONFIG_PCM_S8_PLANAR_DECODER)      += 8svx.o
 +OBJS-$(CONFIG_PCM_S8_PLANAR_DECODER)      += pcm.o
 +OBJS-$(CONFIG_PCM_S8_PLANAR_ENCODER)      += pcm.o
  OBJS-$(CONFIG_PCM_S16BE_DECODER)          += pcm.o
  OBJS-$(CONFIG_PCM_S16BE_ENCODER)          += pcm.o
  OBJS-$(CONFIG_PCM_S16BE_PLANAR_DECODER)   += pcm.o
 +OBJS-$(CONFIG_PCM_S16BE_PLANAR_ENCODER)   += pcm.o
  OBJS-$(CONFIG_PCM_S16LE_DECODER)          += pcm.o
  OBJS-$(CONFIG_PCM_S16LE_ENCODER)          += pcm.o
  OBJS-$(CONFIG_PCM_S16LE_PLANAR_DECODER)   += pcm.o
 +OBJS-$(CONFIG_PCM_S16LE_PLANAR_ENCODER)   += pcm.o
  OBJS-$(CONFIG_PCM_S24BE_DECODER)          += pcm.o
  OBJS-$(CONFIG_PCM_S24BE_ENCODER)          += pcm.o
  OBJS-$(CONFIG_PCM_S24DAUD_DECODER)        += pcm.o
@@@ -609,13 -507,11 +609,13 @@@ OBJS-$(CONFIG_PCM_S24DAUD_ENCODER
  OBJS-$(CONFIG_PCM_S24LE_DECODER)          += pcm.o
  OBJS-$(CONFIG_PCM_S24LE_ENCODER)          += pcm.o
  OBJS-$(CONFIG_PCM_S24LE_PLANAR_DECODER)   += pcm.o
 +OBJS-$(CONFIG_PCM_S24LE_PLANAR_ENCODER)   += pcm.o
  OBJS-$(CONFIG_PCM_S32BE_DECODER)          += pcm.o
  OBJS-$(CONFIG_PCM_S32BE_ENCODER)          += pcm.o
  OBJS-$(CONFIG_PCM_S32LE_DECODER)          += pcm.o
  OBJS-$(CONFIG_PCM_S32LE_ENCODER)          += pcm.o
  OBJS-$(CONFIG_PCM_S32LE_PLANAR_DECODER)   += pcm.o
 +OBJS-$(CONFIG_PCM_S32LE_PLANAR_ENCODER)   += pcm.o
  OBJS-$(CONFIG_PCM_U8_DECODER)             += pcm.o
  OBJS-$(CONFIG_PCM_U8_ENCODER)             += pcm.o
  OBJS-$(CONFIG_PCM_U16BE_DECODER)          += pcm.o
@@@ -635,9 -531,7 +635,9 @@@ OBJS-$(CONFIG_PCM_ZORK_DECODER
  OBJS-$(CONFIG_ADPCM_4XM_DECODER)          += adpcm.o adpcm_data.o
  OBJS-$(CONFIG_ADPCM_ADX_DECODER)          += adxdec.o adx.o
  OBJS-$(CONFIG_ADPCM_ADX_ENCODER)          += adxenc.o adx.o
 +OBJS-$(CONFIG_ADPCM_AFC_DECODER)          += adpcm.o adpcm_data.o
  OBJS-$(CONFIG_ADPCM_CT_DECODER)           += adpcm.o adpcm_data.o
 +OBJS-$(CONFIG_ADPCM_DTK_DECODER)          += adpcm.o adpcm_data.o
  OBJS-$(CONFIG_ADPCM_EA_DECODER)           += adpcm.o adpcm_data.o
  OBJS-$(CONFIG_ADPCM_EA_MAXIS_XA_DECODER)  += adpcm.o adpcm_data.o
  OBJS-$(CONFIG_ADPCM_EA_R1_DECODER)        += adpcm.o adpcm_data.o
@@@ -648,7 -542,6 +648,7 @@@ OBJS-$(CONFIG_ADPCM_G722_DECODER
  OBJS-$(CONFIG_ADPCM_G722_ENCODER)         += g722.o g722dsp.o g722enc.o
  OBJS-$(CONFIG_ADPCM_G726_DECODER)         += g726.o
  OBJS-$(CONFIG_ADPCM_G726_ENCODER)         += g726.o
 +OBJS-$(CONFIG_ADPCM_G726LE_DECODER)       += g726.o
  OBJS-$(CONFIG_ADPCM_IMA_AMV_DECODER)      += adpcm.o adpcm_data.o
  OBJS-$(CONFIG_ADPCM_IMA_APC_DECODER)      += adpcm.o adpcm_data.o
  OBJS-$(CONFIG_ADPCM_IMA_DK3_DECODER)      += adpcm.o adpcm_data.o
@@@ -656,10 -549,8 +656,10 @@@ OBJS-$(CONFIG_ADPCM_IMA_DK4_DECODER
  OBJS-$(CONFIG_ADPCM_IMA_EA_EACS_DECODER)  += adpcm.o adpcm_data.o
  OBJS-$(CONFIG_ADPCM_IMA_EA_SEAD_DECODER)  += adpcm.o adpcm_data.o
  OBJS-$(CONFIG_ADPCM_IMA_ISS_DECODER)      += adpcm.o adpcm_data.o
 +OBJS-$(CONFIG_ADPCM_IMA_OKI_DECODER)      += adpcm.o adpcm_data.o
  OBJS-$(CONFIG_ADPCM_IMA_QT_DECODER)       += adpcm.o adpcm_data.o
  OBJS-$(CONFIG_ADPCM_IMA_QT_ENCODER)       += adpcmenc.o adpcm_data.o
 +OBJS-$(CONFIG_ADPCM_IMA_RAD_DECODER)      += adpcm.o adpcm_data.o
  OBJS-$(CONFIG_ADPCM_IMA_SMJPEG_DECODER)   += adpcm.o adpcm_data.o
  OBJS-$(CONFIG_ADPCM_IMA_WAV_DECODER)      += adpcm.o adpcm_data.o
  OBJS-$(CONFIG_ADPCM_IMA_WAV_ENCODER)      += adpcmenc.o adpcm_data.o
@@@ -676,7 -567,6 +676,7 @@@ OBJS-$(CONFIG_ADPCM_VIMA_DECODER
  OBJS-$(CONFIG_ADPCM_XA_DECODER)           += adpcm.o adpcm_data.o
  OBJS-$(CONFIG_ADPCM_YAMAHA_DECODER)       += adpcm.o adpcm_data.o
  OBJS-$(CONFIG_ADPCM_YAMAHA_ENCODER)       += adpcmenc.o adpcm_data.o
 +OBJS-$(CONFIG_VIMA_DECODER)               += vima.o adpcm_data.o
  
  # hardware accelerators
  OBJS-$(CONFIG_D3D11VA)                    += dxva2.o
@@@ -694,14 -584,11 +694,14 @@@ OBJS-$(CONFIG_H264_VDA_HWACCEL
  OBJS-$(CONFIG_H264_VDPAU_HWACCEL)         += vdpau_h264.o
  OBJS-$(CONFIG_HEVC_D3D11VA_HWACCEL)       += dxva2_hevc.o
  OBJS-$(CONFIG_HEVC_DXVA2_HWACCEL)         += dxva2_hevc.o
 +OBJS-$(CONFIG_HEVC_VDPAU_HWACCEL)         += vdpau_hevc.o
  OBJS-$(CONFIG_MPEG1_VDPAU_HWACCEL)        += vdpau_mpeg12.o
 +OBJS-$(CONFIG_MPEG1_XVMC_HWACCEL)         += mpegvideo_xvmc.o
  OBJS-$(CONFIG_MPEG2_D3D11VA_HWACCEL)      += dxva2_mpeg2.o
  OBJS-$(CONFIG_MPEG2_DXVA2_HWACCEL)        += dxva2_mpeg2.o
  OBJS-$(CONFIG_MPEG2_VAAPI_HWACCEL)        += vaapi_mpeg2.o
  OBJS-$(CONFIG_MPEG2_VDPAU_HWACCEL)        += vdpau_mpeg12.o
 +OBJS-$(CONFIG_MPEG2_XVMC_HWACCEL)         += mpegvideo_xvmc.o
  OBJS-$(CONFIG_MPEG4_VAAPI_HWACCEL)        += vaapi_mpeg4.o
  OBJS-$(CONFIG_MPEG4_VDPAU_HWACCEL)        += vdpau_mpeg4.o
  OBJS-$(CONFIG_VC1_D3D11VA_HWACCEL)        += dxva2_vc1.o
@@@ -713,47 -600,36 +713,47 @@@ OBJS-$(CONFIG_VC1_VDPAU_HWACCEL
  OBJS-$(CONFIG_ADTS_MUXER)              += mpeg4audio.o
  OBJS-$(CONFIG_CAF_DEMUXER)             += mpeg4audio.o mpegaudiodata.o  \
                                            ac3tab.o
 -OBJS-$(CONFIG_FLAC_MUXER)              += flac.o flacdata.o
 +OBJS-$(CONFIG_FLAC_DEMUXER)            += flac.o flacdata.o vorbis_data.o
 +OBJS-$(CONFIG_FLAC_MUXER)              += flac.o flacdata.o vorbis_data.o
  OBJS-$(CONFIG_FLV_DEMUXER)             += mpeg4audio.o
  OBJS-$(CONFIG_GXF_DEMUXER)             += mpeg12data.o
  OBJS-$(CONFIG_IFF_DEMUXER)             += iff.o
  OBJS-$(CONFIG_ISMV_MUXER)              += mpeg4audio.o mpegaudiodata.o
  OBJS-$(CONFIG_LATM_MUXER)              += mpeg4audio.o
 -OBJS-$(CONFIG_MATROSKA_AUDIO_MUXER)    += mpeg4audio.o                  \
 +OBJS-$(CONFIG_MATROSKA_AUDIO_MUXER)    += mpeg4audio.o vorbis_data.o    \
                                            flac.o flacdata.o
  OBJS-$(CONFIG_MATROSKA_DEMUXER)        += mpeg4audio.o mpegaudiodata.o
  OBJS-$(CONFIG_MATROSKA_MUXER)          += mpeg4audio.o mpegaudiodata.o  \
 -                                          flac.o flacdata.o
 +                                          flac.o flacdata.o vorbis_data.o
  OBJS-$(CONFIG_MP2_MUXER)               += mpegaudiodata.o mpegaudiodecheader.o
  OBJS-$(CONFIG_MP3_MUXER)               += mpegaudiodata.o mpegaudiodecheader.o
  OBJS-$(CONFIG_MOV_DEMUXER)             += mpeg4audio.o mpegaudiodata.o ac3tab.o
  OBJS-$(CONFIG_MOV_MUXER)               += mpeg4audio.o mpegaudiodata.o
  OBJS-$(CONFIG_MPEGTS_MUXER)            += mpeg4audio.o
  OBJS-$(CONFIG_MPEGTS_DEMUXER)          += mpeg4audio.o mpegaudiodata.o
 +OBJS-$(CONFIG_MXF_MUXER)               += dnxhddata.o
  OBJS-$(CONFIG_NUT_MUXER)               += mpegaudiodata.o
 +OBJS-$(CONFIG_NUT_DEMUXER)             += mpegaudiodata.o mpeg4audio.o
 +OBJS-$(CONFIG_OGA_MUXER)               += flac.o flacdata.o
  OBJS-$(CONFIG_OGG_DEMUXER)             += mpeg12data.o \
 -                                          dirac.o
 -OBJS-$(CONFIG_OGG_MUXER)               += flac.o flacdata.o
 +                                          dirac.o vorbis_data.o
 +OBJS-$(CONFIG_OGG_MUXER)               += flac.o flacdata.o \
 +                                          vorbis_data.o
  OBJS-$(CONFIG_RTP_MUXER)               += mpeg4audio.o
  OBJS-$(CONFIG_SPDIF_DEMUXER)           += aacadtsdec.o mpeg4audio.o
  OBJS-$(CONFIG_SPDIF_MUXER)             += dca.o
  OBJS-$(CONFIG_TAK_DEMUXER)             += tak.o
  OBJS-$(CONFIG_WEBM_MUXER)              += mpeg4audio.o mpegaudiodata.o  \
 -                                          flac.o flacdata.o
 +                                          flac.o flacdata.o \
 +                                          vorbis_data.o
  OBJS-$(CONFIG_WTV_DEMUXER)             += mpeg4audio.o mpegaudiodata.o
  
 +# libavfilter dependencies
 +OBJS-$(CONFIG_ELBG_FILTER)             += elbg.o
 +
  # external codec libraries
 +OBJS-$(CONFIG_LIBAACPLUS_ENCODER)         += libaacplus.o
 +OBJS-$(CONFIG_LIBCELT_DECODER)            += libcelt_dec.o
  OBJS-$(CONFIG_LIBDCADEC_DECODER)          += libdcadec.o dca.o
  OBJS-$(CONFIG_LIBFAAC_ENCODER)            += libfaac.o
  OBJS-$(CONFIG_LIBFDK_AAC_DECODER)         += libfdk-aacdec.o
@@@ -779,31 -655,24 +779,31 @@@ OBJS-$(CONFIG_LIBSCHROEDINGER_DECODER
                                               libschroedinger.o
  OBJS-$(CONFIG_LIBSCHROEDINGER_ENCODER)    += libschroedingerenc.o \
                                               libschroedinger.o
 +OBJS-$(CONFIG_LIBSHINE_ENCODER)           += libshine.o
  OBJS-$(CONFIG_LIBSPEEX_DECODER)           += libspeexdec.o
  OBJS-$(CONFIG_LIBSPEEX_ENCODER)           += libspeexenc.o
 +OBJS-$(CONFIG_LIBSTAGEFRIGHT_H264_DECODER)+= libstagefright.o
  OBJS-$(CONFIG_LIBTHEORA_ENCODER)          += libtheoraenc.o
  OBJS-$(CONFIG_LIBTWOLAME_ENCODER)         += libtwolame.o
 +OBJS-$(CONFIG_LIBUTVIDEO_DECODER)         += libutvideodec.o
 +OBJS-$(CONFIG_LIBUTVIDEO_ENCODER)         += libutvideoenc.o
  OBJS-$(CONFIG_LIBVO_AACENC_ENCODER)       += libvo-aacenc.o mpeg4audio.o
  OBJS-$(CONFIG_LIBVO_AMRWBENC_ENCODER)     += libvo-amrwbenc.o
 -OBJS-$(CONFIG_LIBVORBIS_ENCODER)          += libvorbis.o \
 +OBJS-$(CONFIG_LIBVORBIS_DECODER)          += libvorbisdec.o
 +OBJS-$(CONFIG_LIBVORBIS_ENCODER)          += libvorbisenc.o \
                                               vorbis_data.o
 -OBJS-$(CONFIG_LIBVPX_VP8_DECODER)         += libvpxdec.o libvpx.o
 -OBJS-$(CONFIG_LIBVPX_VP8_ENCODER)         += libvpxenc.o libvpx.o
 +OBJS-$(CONFIG_LIBVPX_VP8_DECODER)         += libvpxdec.o
 +OBJS-$(CONFIG_LIBVPX_VP8_ENCODER)         += libvpxenc.o
  OBJS-$(CONFIG_LIBVPX_VP9_DECODER)         += libvpxdec.o libvpx.o
  OBJS-$(CONFIG_LIBVPX_VP9_ENCODER)         += libvpxenc.o libvpx.o
  OBJS-$(CONFIG_LIBWAVPACK_ENCODER)         += libwavpackenc.o
 -OBJS-$(CONFIG_LIBWEBP_ENCODER)            += libwebpenc.o
 +OBJS-$(CONFIG_LIBWEBP_ENCODER)            += libwebpenc_common.o libwebpenc.o
 +OBJS-$(CONFIG_LIBWEBP_ANIM_ENCODER)       += libwebpenc_common.o libwebpenc_animencoder.o
  OBJS-$(CONFIG_LIBX264_ENCODER)            += libx264.o
  OBJS-$(CONFIG_LIBX265_ENCODER)            += libx265.o
  OBJS-$(CONFIG_LIBXAVS_ENCODER)            += libxavs.o
  OBJS-$(CONFIG_LIBXVID_ENCODER)            += libxvid.o
 +OBJS-$(CONFIG_LIBZVBI_TELETEXT_DECODER)   += libzvbi-teletextdec.o
  
  # parsers
  OBJS-$(CONFIG_AAC_PARSER)              += aac_parser.o aac_ac3_parser.o \
@@@ -820,10 -689,8 +820,10 @@@ OBJS-$(CONFIG_DIRAC_PARSER)            
  OBJS-$(CONFIG_DNXHD_PARSER)            += dnxhd_parser.o
  OBJS-$(CONFIG_DPX_PARSER)              += dpx_parser.o
  OBJS-$(CONFIG_DVBSUB_PARSER)           += dvbsub_parser.o
 +OBJS-$(CONFIG_DVD_NAV_PARSER)          += dvd_nav_parser.o
  OBJS-$(CONFIG_DVDSUB_PARSER)           += dvdsub_parser.o
 -OBJS-$(CONFIG_FLAC_PARSER)             += flac_parser.o flacdata.o flac.o
 +OBJS-$(CONFIG_FLAC_PARSER)             += flac_parser.o flacdata.o flac.o \
 +                                          vorbis_data.o
  OBJS-$(CONFIG_GSM_PARSER)              += gsm_parser.o
  OBJS-$(CONFIG_H261_PARSER)             += h261_parser.o
  OBJS-$(CONFIG_H263_PARSER)             += h263_parser.o
@@@ -834,7 -701,6 +834,7 @@@ OBJS-$(CONFIG_MLP_PARSER)              
  OBJS-$(CONFIG_MPEG4VIDEO_PARSER)       += mpeg4video_parser.o h263.o \
                                            mpeg4videodec.o mpeg4video.o \
                                            ituh263dec.o h263dec.o
 +OBJS-$(CONFIG_PNG_PARSER)              += png_parser.o
  OBJS-$(CONFIG_MPEGAUDIO_PARSER)        += mpegaudio_parser.o \
                                            mpegaudiodecheader.o mpegaudiodata.o
  OBJS-$(CONFIG_MPEGVIDEO_PARSER)        += mpegvideo_parser.o    \
@@@ -848,7 -714,6 +848,7 @@@ OBJS-$(CONFIG_TAK_PARSER)              
  OBJS-$(CONFIG_VC1_PARSER)              += vc1_parser.o
  OBJS-$(CONFIG_VP3_PARSER)              += vp3_parser.o
  OBJS-$(CONFIG_VP8_PARSER)              += vp8_parser.o
 +OBJS-$(CONFIG_VP9_PARSER)              += vp9_parser.o
  
  # bitstream filters
  OBJS-$(CONFIG_AAC_ADTSTOASC_BSF)          += aac_adtstoasc_bsf.o aacadtsdec.o \
@@@ -859,10 -724,7 +859,10 @@@ OBJS-$(CONFIG_H264_MP4TOANNEXB_BSF
  OBJS-$(CONFIG_IMX_DUMP_HEADER_BSF)        += imx_dump_header_bsf.o
  OBJS-$(CONFIG_MJPEG2JPEG_BSF)             += mjpeg2jpeg_bsf.o
  OBJS-$(CONFIG_MJPEGA_DUMP_HEADER_BSF)     += mjpega_dump_header_bsf.o
 +OBJS-$(CONFIG_MPEG4_UNPACK_BFRAMES_BSF)   += mpeg4_unpack_bframes_bsf.o
  OBJS-$(CONFIG_MOV2TEXTSUB_BSF)            += movsub_bsf.o
 +OBJS-$(CONFIG_MP3_HEADER_DECOMPRESS_BSF)  += mp3_header_decompress_bsf.o \
 +                                             mpegaudiodata.o
  OBJS-$(CONFIG_NOISE_BSF)                  += noise_bsf.o
  OBJS-$(CONFIG_REMOVE_EXTRADATA_BSF)       += remove_extradata_bsf.o
  OBJS-$(CONFIG_TEXT2MOVSUB_BSF)            += movsub_bsf.o
  OBJS-$(HAVE_LIBC_MSVCRT)               += file_open.o
  OBJS-$(HAVE_THREADS)                   += pthread.o pthread_slice.o pthread_frame.o
  
 +OBJS-$(CONFIG_FRAME_THREAD_ENCODER)    += frame_thread_encoder.o
 +
 +# Windows resource file
 +SLIBOBJS-$(HAVE_GNU_WINDRES)           += avcodecres.o
 +
  SKIPHEADERS                            += %_tablegen.h                  \
                                            %_tables.h                    \
                                            aac_tablegen_decl.h           \
                                            fft-internal.h                \
 +                                          libutvideo.h                  \
 +                                          old_codec_ids.h               \
                                            tableprint.h                  \
 +                                          tableprint_vlc.h              \
                                            $(ARCH)/vp56_arith.h          \
  
  SKIPHEADERS-$(CONFIG_D3D11VA)          += d3d11va.h dxva2_internal.h
  SKIPHEADERS-$(CONFIG_DXVA2)            += dxva2.h dxva2_internal.h
  SKIPHEADERS-$(CONFIG_LIBSCHROEDINGER)  += libschroedinger.h
 -SKIPHEADERS-$(CONFIG_MPEG_XVMC_DECODER) += xvmc.h
 +SKIPHEADERS-$(CONFIG_LIBUTVIDEO)       += libutvideo.h
  SKIPHEADERS-$(CONFIG_QSV)              += qsv.h qsv_internal.h
  SKIPHEADERS-$(CONFIG_QSVDEC)           += qsvdec.h
  SKIPHEADERS-$(CONFIG_QSVENC)           += qsvenc.h
 +SKIPHEADERS-$(CONFIG_XVMC)             += xvmc.h
  SKIPHEADERS-$(CONFIG_VAAPI)            += vaapi_internal.h
  SKIPHEADERS-$(CONFIG_VDA)              += vda.h vda_internal.h
  SKIPHEADERS-$(CONFIG_VDPAU)            += vdpau.h vdpau_internal.h
  
 -TESTPROGS-$(CONFIG_FFT)                   += fft fft-fixed
 +TESTPROGS = imgconvert                                                  \
 +            jpeg2000dwt                                                 \
 +            mathops                                                    \
 +            options                                                     \
 +            avfft                                                       \
 +
 +TESTPROGS-$(CONFIG_CABAC)                 += cabac
 +TESTPROGS-$(CONFIG_FFT)                   += fft fft-fixed fft-fixed32
  TESTPROGS-$(CONFIG_IDCTDSP)               += dct
  TESTPROGS-$(CONFIG_IIRFILTER)             += iirfilter
 +TESTPROGS-$(HAVE_MMX)                     += motion
  TESTPROGS-$(CONFIG_GOLOMB)                += golomb
  TESTPROGS-$(CONFIG_RANGECODER)            += rangecoder
 +TESTPROGS-$(CONFIG_SNOW_ENCODER)          += snowenc
  
  TESTOBJS = dctref.o
  
 +TOOLS = fourcc2pixfmt
 +
  HOSTPROGS = aac_tablegen                                                \
              aacps_tablegen                                              \
 +            aacsbr_tablegen                                             \
 +            cabac_tablegen                                              \
              cbrt_tablegen                                               \
              cos_tablegen                                                \
 +            dsd_tablegen                                                \
              dv_tablegen                                                 \
              motionpixels_tablegen                                       \
              mpegaudio_tablegen                                          \
@@@ -948,7 -787,7 +948,7 @@@ els
  $(SUBDIR)%_tablegen$(HOSTEXESUF): HOSTCFLAGS += -DCONFIG_SMALL=0
  endif
  
 -GEN_HEADERS = cbrt_tables.h aacps_tables.h aac_tables.h dv_tables.h     \
 +GEN_HEADERS = cabac_tables.h cbrt_tables.h aacps_tables.h aacsbr_tables.h aac_tables.h dsd_tables.h dv_tables.h     \
                sinewin_tables.h mpegaudio_tables.h motionpixels_tables.h \
                pcm_tables.h qdm2_tables.h
  GEN_HEADERS := $(addprefix $(SUBDIR), $(GEN_HEADERS))
@@@ -959,10 -798,7 +959,10 @@@ $(GEN_HEADERS): $(SUBDIR)%_tables.h: $(
  ifdef CONFIG_HARDCODED_TABLES
  $(SUBDIR)aacdec.o: $(SUBDIR)cbrt_tables.h
  $(SUBDIR)aacps.o: $(SUBDIR)aacps_tables.h
 +$(SUBDIR)aacsbr.o: $(SUBDIR)aacsbr_tables.h
  $(SUBDIR)aactab.o: $(SUBDIR)aac_tables.h
 +$(SUBDIR)cabac.o: $(SUBDIR)cabac_tables.h
 +$(SUBDIR)dsddec.o: $(SUBDIR)dsd_tables.h
  $(SUBDIR)dvenc.o: $(SUBDIR)dv_tables.h
  $(SUBDIR)sinewin.o: $(SUBDIR)sinewin_tables.h
  $(SUBDIR)mpegaudiodec_fixed.o: $(SUBDIR)mpegaudio_tables.h
diff --combined libavcodec/hevc.c
@@@ -1,29 -1,28 +1,29 @@@
  /*
 - * HEVC video decoder
 + * HEVC video Decoder
   *
   * Copyright (C) 2012 - 2013 Guillaume Martres
   * Copyright (C) 2012 - 2013 Mickael Raulet
   * Copyright (C) 2012 - 2013 Gildas Cocherel
   * Copyright (C) 2012 - 2013 Wassim Hamidouche
   *
 - * This file is part of Libav.
 + * This file is part of FFmpeg.
   *
 - * Libav is free software; you can redistribute it and/or
 + * FFmpeg is free software; you can redistribute it and/or
   * modify it under the terms of the GNU Lesser General Public
   * License as published by the Free Software Foundation; either
   * version 2.1 of the License, or (at your option) any later version.
   *
 - * Libav is distributed in the hope that it will be useful,
 + * 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
   * Lesser General Public License for more details.
   *
   * You should have received a copy of the GNU Lesser General Public
 - * License along with Libav; if not, write to the Free Software
 + * License along with FFmpeg; if not, write to the Free Software
   * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
   */
  
 +#include "libavutil/atomic.h"
  #include "libavutil/attributes.h"
  #include "libavutil/common.h"
  #include "libavutil/display.h"
  #include "golomb.h"
  #include "hevc.h"
  
 -const uint8_t ff_hevc_qpel_extra_before[4] = { 0, 3, 3, 2 };
 -const uint8_t ff_hevc_qpel_extra_after[4]  = { 0, 3, 4, 4 };
 -const uint8_t ff_hevc_qpel_extra[4]        = { 0, 6, 7, 6 };
 -
 -static const uint8_t scan_1x1[1] = { 0 };
 -
 -static const uint8_t horiz_scan2x2_x[4] = { 0, 1, 0, 1 };
 -
 -static const uint8_t horiz_scan2x2_y[4] = { 0, 0, 1, 1 };
 -
 -static const uint8_t horiz_scan4x4_x[16] = {
 -    0, 1, 2, 3,
 -    0, 1, 2, 3,
 -    0, 1, 2, 3,
 -    0, 1, 2, 3,
 -};
 -
 -static const uint8_t horiz_scan4x4_y[16] = {
 -    0, 0, 0, 0,
 -    1, 1, 1, 1,
 -    2, 2, 2, 2,
 -    3, 3, 3, 3,
 -};
 -
 -static const uint8_t horiz_scan8x8_inv[8][8] = {
 -    {  0,  1,  2,  3, 16, 17, 18, 19, },
 -    {  4,  5,  6,  7, 20, 21, 22, 23, },
 -    {  8,  9, 10, 11, 24, 25, 26, 27, },
 -    { 12, 13, 14, 15, 28, 29, 30, 31, },
 -    { 32, 33, 34, 35, 48, 49, 50, 51, },
 -    { 36, 37, 38, 39, 52, 53, 54, 55, },
 -    { 40, 41, 42, 43, 56, 57, 58, 59, },
 -    { 44, 45, 46, 47, 60, 61, 62, 63, },
 -};
 -
 -static const uint8_t diag_scan2x2_x[4] = { 0, 0, 1, 1 };
 -
 -static const uint8_t diag_scan2x2_y[4] = { 0, 1, 0, 1 };
 -
 -static const uint8_t diag_scan2x2_inv[2][2] = {
 -    { 0, 2, },
 -    { 1, 3, },
 -};
 -
 -const uint8_t ff_hevc_diag_scan4x4_x[16] = {
 -    0, 0, 1, 0,
 -    1, 2, 0, 1,
 -    2, 3, 1, 2,
 -    3, 2, 3, 3,
 -};
 -
 -const uint8_t ff_hevc_diag_scan4x4_y[16] = {
 -    0, 1, 0, 2,
 -    1, 0, 3, 2,
 -    1, 0, 3, 2,
 -    1, 3, 2, 3,
 -};
 -
 -static const uint8_t diag_scan4x4_inv[4][4] = {
 -    { 0,  2,  5,  9, },
 -    { 1,  4,  8, 12, },
 -    { 3,  7, 11, 14, },
 -    { 6, 10, 13, 15, },
 -};
 -
 -const uint8_t ff_hevc_diag_scan8x8_x[64] = {
 -    0, 0, 1, 0,
 -    1, 2, 0, 1,
 -    2, 3, 0, 1,
 -    2, 3, 4, 0,
 -    1, 2, 3, 4,
 -    5, 0, 1, 2,
 -    3, 4, 5, 6,
 -    0, 1, 2, 3,
 -    4, 5, 6, 7,
 -    1, 2, 3, 4,
 -    5, 6, 7, 2,
 -    3, 4, 5, 6,
 -    7, 3, 4, 5,
 -    6, 7, 4, 5,
 -    6, 7, 5, 6,
 -    7, 6, 7, 7,
 -};
 -
 -const uint8_t ff_hevc_diag_scan8x8_y[64] = {
 -    0, 1, 0, 2,
 -    1, 0, 3, 2,
 -    1, 0, 4, 3,
 -    2, 1, 0, 5,
 -    4, 3, 2, 1,
 -    0, 6, 5, 4,
 -    3, 2, 1, 0,
 -    7, 6, 5, 4,
 -    3, 2, 1, 0,
 -    7, 6, 5, 4,
 -    3, 2, 1, 7,
 -    6, 5, 4, 3,
 -    2, 7, 6, 5,
 -    4, 3, 7, 6,
 -    5, 4, 7, 6,
 -    5, 7, 6, 7,
 -};
 -
 -static const uint8_t diag_scan8x8_inv[8][8] = {
 -    {  0,  2,  5,  9, 14, 20, 27, 35, },
 -    {  1,  4,  8, 13, 19, 26, 34, 42, },
 -    {  3,  7, 12, 18, 25, 33, 41, 48, },
 -    {  6, 11, 17, 24, 32, 40, 47, 53, },
 -    { 10, 16, 23, 31, 39, 46, 52, 57, },
 -    { 15, 22, 30, 38, 45, 51, 56, 60, },
 -    { 21, 29, 37, 44, 50, 55, 59, 62, },
 -    { 28, 36, 43, 49, 54, 58, 61, 63, },
 -};
 +const uint8_t ff_hevc_pel_weight[65] = { [2] = 0, [4] = 1, [6] = 2, [8] = 3, [12] = 4, [16] = 5, [24] = 6, [32] = 7, [48] = 8, [64] = 9 };
  
  /**
   * NOTE: Each function hls_foo correspond to the function foo in the
@@@ -70,10 -181,6 +70,10 @@@ static void pic_arrays_free(HEVCContex
      av_freep(&s->horizontal_bs);
      av_freep(&s->vertical_bs);
  
 +    av_freep(&s->sh.entry_point_offset);
 +    av_freep(&s->sh.size);
 +    av_freep(&s->sh.offset);
 +
      av_buffer_pool_uninit(&s->tab_mvf_pool);
      av_buffer_pool_uninit(&s->rpl_tab_pool);
  }
@@@ -89,40 -196,40 +89,40 @@@ static int pic_arrays_init(HEVCContext 
      int ctb_count        = sps->ctb_width * sps->ctb_height;
      int min_pu_size      = sps->min_pu_width * sps->min_pu_height;
  
 -    s->bs_width  = width  >> 3;
 -    s->bs_height = height >> 3;
 +    s->bs_width  = (width  >> 2) + 1;
 +    s->bs_height = (height >> 2) + 1;
  
      s->sao           = av_mallocz_array(ctb_count, sizeof(*s->sao));
      s->deblock       = av_mallocz_array(ctb_count, sizeof(*s->deblock));
      if (!s->sao || !s->deblock)
          goto fail;
  
 -    s->skip_flag    = av_malloc(pic_size_in_ctb);
 -    s->tab_ct_depth = av_malloc(sps->min_cb_height * sps->min_cb_width);
 +    s->skip_flag    = av_malloc_array(sps->min_cb_height, sps->min_cb_width);
 +    s->tab_ct_depth = av_malloc_array(sps->min_cb_height, sps->min_cb_width);
      if (!s->skip_flag || !s->tab_ct_depth)
          goto fail;
  
 -    s->cbf_luma = av_malloc(sps->min_tb_width * sps->min_tb_height);
 +    s->cbf_luma = av_malloc_array(sps->min_tb_width, sps->min_tb_height);
      s->tab_ipm  = av_mallocz(min_pu_size);
 -    s->is_pcm   = av_malloc(min_pu_size);
 +    s->is_pcm   = av_malloc_array(sps->min_pu_width + 1, sps->min_pu_height + 1);
      if (!s->tab_ipm || !s->cbf_luma || !s->is_pcm)
          goto fail;
  
 -    s->filter_slice_edges = av_malloc(ctb_count);
 -    s->tab_slice_address  = av_malloc(pic_size_in_ctb *
 +    s->filter_slice_edges = av_mallocz(ctb_count);
 +    s->tab_slice_address  = av_malloc_array(pic_size_in_ctb,
                                        sizeof(*s->tab_slice_address));
 -    s->qp_y_tab           = av_malloc(pic_size_in_ctb *
 +    s->qp_y_tab           = av_malloc_array(pic_size_in_ctb,
                                        sizeof(*s->qp_y_tab));
      if (!s->qp_y_tab || !s->filter_slice_edges || !s->tab_slice_address)
          goto fail;
  
 -    s->horizontal_bs = av_mallocz(2 * s->bs_width * (s->bs_height + 1));
 -    s->vertical_bs   = av_mallocz(2 * s->bs_width * (s->bs_height + 1));
 +    s->horizontal_bs = av_mallocz_array(s->bs_width, s->bs_height);
 +    s->vertical_bs   = av_mallocz_array(s->bs_width, s->bs_height);
      if (!s->horizontal_bs || !s->vertical_bs)
          goto fail;
  
      s->tab_mvf_pool = av_buffer_pool_init(min_pu_size * sizeof(MvField),
 -                                          av_buffer_alloc);
 +                                          av_buffer_allocz);
      s->rpl_tab_pool = av_buffer_pool_init(ctb_count * sizeof(RefPicListTab),
                                            av_buffer_allocz);
      if (!s->tab_mvf_pool || !s->rpl_tab_pool)
@@@ -143,15 -250,11 +143,15 @@@ static void pred_weight_table(HEVCConte
      uint8_t chroma_weight_l0_flag[16];
      uint8_t luma_weight_l1_flag[16];
      uint8_t chroma_weight_l1_flag[16];
 +    int luma_log2_weight_denom;
  
 -    s->sh.luma_log2_weight_denom = av_clip(get_ue_golomb_long(gb), 0, 7);
 +    luma_log2_weight_denom = get_ue_golomb_long(gb);
 +    if (luma_log2_weight_denom < 0 || luma_log2_weight_denom > 7)
 +        av_log(s->avctx, AV_LOG_ERROR, "luma_log2_weight_denom %d is invalid\n", luma_log2_weight_denom);
 +    s->sh.luma_log2_weight_denom = av_clip_uintp2(luma_log2_weight_denom, 3);
      if (s->sps->chroma_format_idc != 0) {
          int delta = get_se_golomb(gb);
 -        s->sh.chroma_log2_weight_denom = av_clip(s->sh.luma_log2_weight_denom + delta, 0, 7);
 +        s->sh.chroma_log2_weight_denom = av_clip_uintp2(s->sh.luma_log2_weight_denom + delta, 3);
      }
  
      for (i = 0; i < s->sh.nb_refs[L0]; i++) {
              s->sh.luma_offset_l0[i] = 0;
          }
      }
 -    if (s->sps->chroma_format_idc != 0) { // FIXME: invert "if" and "for"
 +    if (s->sps->chroma_format_idc != 0) {
          for (i = 0; i < s->sh.nb_refs[L0]; i++)
              chroma_weight_l0_flag[i] = get_bits1(gb);
      } else {
@@@ -244,7 -347,7 +244,7 @@@ static int decode_lt_rps(HEVCContext *s
          nb_sps = get_ue_golomb_long(gb);
      nb_sh = get_ue_golomb_long(gb);
  
 -    if (nb_sh + nb_sps > FF_ARRAY_ELEMS(rps->poc))
 +    if (nb_sh + (uint64_t)nb_sps > FF_ARRAY_ELEMS(rps->poc))
          return AVERROR_INVALIDDATA;
  
      rps->nb_refs = nb_sh + nb_sps;
@@@ -326,11 -429,11 +326,11 @@@ static void export_stream_params(AVCode
                    num, den, 1 << 30);
  }
  
 -static int set_sps(HEVCContext *s, const HEVCSPS *sps)
 +static int set_sps(HEVCContext *s, const HEVCSPS *sps, enum AVPixelFormat pix_fmt)
  {
 -    #define HWACCEL_MAX (CONFIG_HEVC_DXVA2_HWACCEL + CONFIG_HEVC_D3D11VA_HWACCEL)
 +    #define HWACCEL_MAX (CONFIG_HEVC_DXVA2_HWACCEL + CONFIG_HEVC_D3D11VA_HWACCEL + CONFIG_HEVC_VDPAU_HWACCEL)
      enum AVPixelFormat pix_fmts[HWACCEL_MAX + 2], *fmt = pix_fmts;
 -    int ret;
 +    int ret, i;
  
      export_stream_params(s->avctx, s, sps);
  
  #endif
  #if CONFIG_HEVC_D3D11VA_HWACCEL
          *fmt++ = AV_PIX_FMT_D3D11VA_VLD;
 +#endif
 +#if CONFIG_HEVC_VDPAU_HWACCEL
 +        *fmt++ = AV_PIX_FMT_VDPAU;
  #endif
      }
  
 -    *fmt++ = sps->pix_fmt;
 -    *fmt = AV_PIX_FMT_NONE;
 +    if (pix_fmt == AV_PIX_FMT_NONE) {
 +        *fmt++ = sps->pix_fmt;
 +        *fmt = AV_PIX_FMT_NONE;
  
 -    ret = ff_get_format(s->avctx, pix_fmts);
 -    if (ret < 0)
 -        goto fail;
 -    s->avctx->pix_fmt = ret;
 +        ret = ff_thread_get_format(s->avctx, pix_fmts);
 +        if (ret < 0)
 +            goto fail;
 +        s->avctx->pix_fmt = ret;
 +    }
 +    else {
 +        s->avctx->pix_fmt = pix_fmt;
 +    }
  
      ff_hevc_pred_init(&s->hpc,     sps->bit_depth);
      ff_hevc_dsp_init (&s->hevcdsp, sps->bit_depth);
      ff_videodsp_init (&s->vdsp,    sps->bit_depth);
  
 +    for (i = 0; i < 3; i++) {
 +        av_freep(&s->sao_pixel_buffer_h[i]);
 +        av_freep(&s->sao_pixel_buffer_v[i]);
 +    }
 +
      if (sps->sao_enabled && !s->avctx->hwaccel) {
 -        av_frame_unref(s->tmp_frame);
 -        ret = ff_get_buffer(s->avctx, s->tmp_frame, AV_GET_BUFFER_FLAG_REF);
 -        if (ret < 0)
 -            goto fail;
 -        s->frame = s->tmp_frame;
 +        int c_count = (sps->chroma_format_idc != 0) ? 3 : 1;
 +        int c_idx;
 +
 +        for(c_idx = 0; c_idx < c_count; c_idx++) {
 +            int w = sps->width >> sps->hshift[c_idx];
 +            int h = sps->height >> sps->vshift[c_idx];
 +            s->sao_pixel_buffer_h[c_idx] =
 +                av_malloc((w * 2 * sps->ctb_height) <<
 +                          sps->pixel_shift);
 +            s->sao_pixel_buffer_v[c_idx] =
 +                av_malloc((h * 2 * sps->ctb_width) <<
 +                          sps->pixel_shift);
 +        }
      }
  
      s->sps = sps;
@@@ -402,7 -484,7 +402,7 @@@ fail
  
  static int hls_slice_header(HEVCContext *s)
  {
 -    GetBitContext *gb = &s->HEVClc.gb;
 +    GetBitContext *gb = &s->HEVClc->gb;
      SliceHeader *sh   = &s->sh;
      int i, ret;
  
          if (IS_IDR(s))
              ff_hevc_clear_refs(s);
      }
 +    sh->no_output_of_prior_pics_flag = 0;
      if (IS_IRAP(s))
          sh->no_output_of_prior_pics_flag = get_bits1(gb);
  
          return AVERROR_INVALIDDATA;
      }
      s->pps = (HEVCPPS*)s->pps_list[sh->pps_id]->data;
 +    if (s->nal_unit_type == NAL_CRA_NUT && s->last_eos == 1)
 +        sh->no_output_of_prior_pics_flag = 1;
  
      if (s->sps != (HEVCSPS*)s->sps_list[s->pps->sps_id]->data) {
 +        const HEVCSPS* last_sps = s->sps;
          s->sps = (HEVCSPS*)s->sps_list[s->pps->sps_id]->data;
 -
 +        if (last_sps && IS_IRAP(s) && s->nal_unit_type != NAL_CRA_NUT) {
 +            if (s->sps->width !=  last_sps->width || s->sps->height != last_sps->height ||
 +                s->sps->temporal_layer[s->sps->max_sub_layers - 1].max_dec_pic_buffering !=
 +                last_sps->temporal_layer[last_sps->max_sub_layers - 1].max_dec_pic_buffering)
 +                sh->no_output_of_prior_pics_flag = 0;
 +        }
          ff_hevc_clear_refs(s);
 -        ret = set_sps(s, s->sps);
 +        ret = set_sps(s, s->sps, AV_PIX_FMT_NONE);
          if (ret < 0)
              return ret;
  
              sh->colour_plane_id = get_bits(gb, 2);
  
          if (!IS_IDR(s)) {
 -            int poc;
 +            int poc, pos;
  
              sh->pic_order_cnt_lsb = get_bits(gb, s->sps->log2_max_poc_lsb);
              poc = ff_hevc_compute_poc(s, sh->pic_order_cnt_lsb);
              s->poc = poc;
  
              sh->short_term_ref_pic_set_sps_flag = get_bits1(gb);
 +            pos = get_bits_left(gb);
              if (!sh->short_term_ref_pic_set_sps_flag) {
 -                int pos = get_bits_left(gb);
                  ret = ff_hevc_decode_short_term_rps(gb, s->avctx, &sh->slice_rps, s->sps, 1);
                  if (ret < 0)
                      return ret;
  
 -                sh->short_term_ref_pic_set_size = pos - get_bits_left(gb);
                  sh->short_term_rps = &sh->slice_rps;
              } else {
                  int numbits, rps_idx;
                  rps_idx = numbits > 0 ? get_bits(gb, numbits) : 0;
                  sh->short_term_rps = &s->sps->st_rps[rps_idx];
              }
 +            sh->short_term_ref_pic_set_size = pos - get_bits_left(gb);
  
 +            pos = get_bits_left(gb);
              ret = decode_lt_rps(s, &sh->long_term_rps, gb);
              if (ret < 0) {
                  av_log(s->avctx, AV_LOG_WARNING, "Invalid long term RPS.\n");
                  if (s->avctx->err_recognition & AV_EF_EXPLODE)
                      return AVERROR_INVALIDDATA;
              }
 +            sh->long_term_ref_pic_set_size = pos - get_bits_left(gb);
  
              if (s->sps->sps_temporal_mvp_enabled_flag)
                  sh->slice_temporal_mvp_enabled_flag = get_bits1(gb);
  
          if (s->sps->sao_enabled) {
              sh->slice_sample_adaptive_offset_flag[0] = get_bits1(gb);
 -            sh->slice_sample_adaptive_offset_flag[1] =
 -            sh->slice_sample_adaptive_offset_flag[2] = get_bits1(gb);
 +            if (s->sps->chroma_format_idc) {
 +                sh->slice_sample_adaptive_offset_flag[1] =
 +                sh->slice_sample_adaptive_offset_flag[2] = get_bits1(gb);
 +            }
          } else {
              sh->slice_sample_adaptive_offset_flag[0] = 0;
              sh->slice_sample_adaptive_offset_flag[1] = 0;
              sh->slice_cr_qp_offset = 0;
          }
  
 +        if (s->pps->chroma_qp_offset_list_enabled_flag)
 +            sh->cu_chroma_qp_offset_enabled_flag = get_bits1(gb);
 +        else
 +            sh->cu_chroma_qp_offset_enabled_flag = 0;
 +
          if (s->pps->deblocking_filter_control_present_flag) {
              int deblocking_filter_override_flag = 0;
  
  
      sh->num_entry_point_offsets = 0;
      if (s->pps->tiles_enabled_flag || s->pps->entropy_coding_sync_enabled_flag) {
 -        sh->num_entry_point_offsets = get_ue_golomb_long(gb);
 +        unsigned num_entry_point_offsets = get_ue_golomb_long(gb);
 +        // It would be possible to bound this tighter but this here is simpler
 +        if (num_entry_point_offsets > get_bits_left(gb)) {
 +            av_log(s->avctx, AV_LOG_ERROR, "num_entry_point_offsets %d is invalid\n", num_entry_point_offsets);
 +            return AVERROR_INVALIDDATA;
 +        }
 +
 +        sh->num_entry_point_offsets = num_entry_point_offsets;
          if (sh->num_entry_point_offsets > 0) {
              int offset_len = get_ue_golomb_long(gb) + 1;
  
 -            for (i = 0; i < sh->num_entry_point_offsets; i++)
 -                skip_bits(gb, offset_len);
 -        }
 +            if (offset_len < 1 || offset_len > 32) {
 +                sh->num_entry_point_offsets = 0;
 +                av_log(s->avctx, AV_LOG_ERROR, "offset_len %d is invalid\n", offset_len);
 +                return AVERROR_INVALIDDATA;
 +            }
 +
 +            av_freep(&sh->entry_point_offset);
 +            av_freep(&sh->offset);
 +            av_freep(&sh->size);
 +            sh->entry_point_offset = av_malloc_array(sh->num_entry_point_offsets, sizeof(int));
 +            sh->offset = av_malloc_array(sh->num_entry_point_offsets, sizeof(int));
 +            sh->size = av_malloc_array(sh->num_entry_point_offsets, sizeof(int));
 +            if (!sh->entry_point_offset || !sh->offset || !sh->size) {
 +                sh->num_entry_point_offsets = 0;
 +                av_log(s->avctx, AV_LOG_ERROR, "Failed to allocate memory\n");
 +                return AVERROR(ENOMEM);
 +            }
 +            for (i = 0; i < sh->num_entry_point_offsets; i++) {
 +                unsigned val = get_bits_long(gb, offset_len);
 +                sh->entry_point_offset[i] = val + 1; // +1; // +1 to get the size
 +            }
 +            if (s->threads_number > 1 && (s->pps->num_tile_rows > 1 || s->pps->num_tile_columns > 1)) {
 +                s->enable_parallel_tiles = 0; // TODO: you can enable tiles in parallel here
 +                s->threads_number = 1;
 +            } else
 +                s->enable_parallel_tiles = 0;
 +        } else
 +            s->enable_parallel_tiles = 0;
      }
  
      if (s->pps->slice_header_extension_present_flag) {
          unsigned int length = get_ue_golomb_long(gb);
 +        if (length*8LL > get_bits_left(gb)) {
 +            av_log(s->avctx, AV_LOG_ERROR, "too many slice_header_extension_data_bytes\n");
 +            return AVERROR_INVALIDDATA;
 +        }
          for (i = 0; i < length; i++)
              skip_bits(gb, 8);  // slice_header_extension_data_byte
      }
  
      // Inferred parameters
 -    sh->slice_qp = 26 + s->pps->pic_init_qp_minus26 + sh->slice_qp_delta;
 +    sh->slice_qp = 26U + s->pps->pic_init_qp_minus26 + sh->slice_qp_delta;
      if (sh->slice_qp > 51 ||
          sh->slice_qp < -s->sps->qp_bd_offset) {
          av_log(s->avctx, AV_LOG_ERROR,
          return AVERROR_INVALIDDATA;
      }
  
 -    s->HEVClc.first_qp_group = !s->sh.dependent_slice_segment_flag;
 +    if (get_bits_left(gb) < 0) {
 +        av_log(s->avctx, AV_LOG_ERROR,
 +               "Overread slice header by %d bits\n", -get_bits_left(gb));
 +        return AVERROR_INVALIDDATA;
 +    }
 +
 +    s->HEVClc->first_qp_group = !s->sh.dependent_slice_segment_flag;
  
      if (!s->pps->cu_qp_delta_enabled_flag)
 -        s->HEVClc.qp_y = FFUMOD(s->sh.slice_qp + 52 + 2 * s->sps->qp_bd_offset,
 -                                52 + s->sps->qp_bd_offset) - s->sps->qp_bd_offset;
 +        s->HEVClc->qp_y = s->sh.slice_qp;
  
      s->slice_initialized = 1;
 +    s->HEVClc->tu.cu_qp_offset_cb = 0;
 +    s->HEVClc->tu.cu_qp_offset_cr = 0;
  
      return 0;
  }
@@@ -819,9 -840,10 +819,9 @@@ do 
  
  static void hls_sao_param(HEVCContext *s, int rx, int ry)
  {
 -    HEVCLocalContext *lc    = &s->HEVClc;
 +    HEVCLocalContext *lc    = s->HEVClc;
      int sao_merge_left_flag = 0;
      int sao_merge_up_flag   = 0;
 -    int shift               = s->sps->bit_depth - FFMIN(s->sps->bit_depth, 10);
      SAOParams *sao          = &CTB(s->sao, rx, ry);
      int c_idx, i;
  
          }
      }
  
 -    for (c_idx = 0; c_idx < 3; c_idx++) {
 +    for (c_idx = 0; c_idx < (s->sps->chroma_format_idc ? 3 : 1); c_idx++) {
 +        int log2_sao_offset_scale = c_idx == 0 ? s->pps->log2_sao_offset_scale_luma :
 +                                                 s->pps->log2_sao_offset_scale_chroma;
 +
          if (!s->sh.slice_sample_adaptive_offset_flag[c_idx]) {
              sao->type_idx[c_idx] = SAO_NOT_APPLIED;
              continue;
          // Inferred parameters
          sao->offset_val[c_idx][0] = 0;
          for (i = 0; i < 4; i++) {
 -            sao->offset_val[c_idx][i + 1] = sao->offset_abs[c_idx][i] << shift;
 +            sao->offset_val[c_idx][i + 1] = sao->offset_abs[c_idx][i];
              if (sao->type_idx[c_idx] == SAO_EDGE) {
                  if (i > 1)
                      sao->offset_val[c_idx][i + 1] = -sao->offset_val[c_idx][i + 1];
              } else if (sao->offset_sign[c_idx][i]) {
                  sao->offset_val[c_idx][i + 1] = -sao->offset_val[c_idx][i + 1];
              }
 +            sao->offset_val[c_idx][i + 1] *= 1 << log2_sao_offset_scale;
          }
      }
  }
  #undef SET_SAO
  #undef CTB
  
 -static void hls_residual_coding(HEVCContext *s, int x0, int y0,
 -                                int log2_trafo_size, enum ScanType scan_idx,
 -                                int c_idx)
 -{
 -#define GET_COORD(offset, n)                                    \
 -    do {                                                        \
 -        x_c = (scan_x_cg[offset >> 4] << 2) + scan_x_off[n];    \
 -        y_c = (scan_y_cg[offset >> 4] << 2) + scan_y_off[n];    \
 -    } while (0)
 -    HEVCLocalContext *lc    = &s->HEVClc;
 -    int transform_skip_flag = 0;
 -
 -    int last_significant_coeff_x, last_significant_coeff_y;
 -    int last_scan_pos;
 -    int n_end;
 -    int num_coeff    = 0;
 -    int greater1_ctx = 1;
 -
 -    int num_last_subset;
 -    int x_cg_last_sig, y_cg_last_sig;
 -
 -    const uint8_t *scan_x_cg, *scan_y_cg, *scan_x_off, *scan_y_off;
 -
 -    ptrdiff_t stride = s->frame->linesize[c_idx];
 -    int hshift       = s->sps->hshift[c_idx];
 -    int vshift       = s->sps->vshift[c_idx];
 -    uint8_t *dst     = &s->frame->data[c_idx][(y0 >> vshift) * stride +
 -                                              ((x0 >> hshift) << s->sps->pixel_shift)];
 -    DECLARE_ALIGNED(16, int16_t, coeffs[MAX_TB_SIZE * MAX_TB_SIZE]) = { 0 };
 -    DECLARE_ALIGNED(8, uint8_t, significant_coeff_group_flag[8][8]) = { { 0 } };
 -
 -    int trafo_size = 1 << log2_trafo_size;
 -    int i, qp, shift, add, scale, scale_m;
 -    const uint8_t level_scale[] = { 40, 45, 51, 57, 64, 72 };
 -    const uint8_t *scale_matrix;
 -    uint8_t dc_scale;
 -
 -    // Derive QP for dequant
 -    if (!lc->cu.cu_transquant_bypass_flag) {
 -        static const int qp_c[] = {
 -            29, 30, 31, 32, 33, 33, 34, 34, 35, 35, 36, 36, 37, 37
 -        };
 -
 -        static const uint8_t rem6[51 + 2 * 6 + 1] = {
 -            0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2,
 -            3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5,
 -            0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3,
 -        };
 -
 -        static const uint8_t div6[51 + 2 * 6 + 1] = {
 -            0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2,  3,  3,  3,
 -            3, 3, 3, 4, 4, 4, 4, 4, 4, 5, 5, 5, 5, 5, 5, 6, 6, 6,  6,  6,  6,
 -            7, 7, 7, 7, 7, 7, 8, 8, 8, 8, 8, 8, 9, 9, 9, 9, 9, 9, 10, 10, 10, 10,
 -        };
 -        int qp_y = lc->qp_y;
 -
 -        if (c_idx == 0) {
 -            qp = qp_y + s->sps->qp_bd_offset;
 -        } else {
 -            int qp_i, offset;
 -
 -            if (c_idx == 1)
 -                offset = s->pps->cb_qp_offset + s->sh.slice_cb_qp_offset;
 -            else
 -                offset = s->pps->cr_qp_offset + s->sh.slice_cr_qp_offset;
 -
 -            qp_i = av_clip(qp_y + offset, -s->sps->qp_bd_offset, 57);
 -            if (qp_i < 30)
 -                qp = qp_i;
 -            else if (qp_i > 43)
 -                qp = qp_i - 6;
 -            else
 -                qp = qp_c[qp_i - 30];
 -
 -            qp += s->sps->qp_bd_offset;
 -        }
 -
 -        shift    = s->sps->bit_depth + log2_trafo_size - 5;
 -        add      = 1 << (shift - 1);
 -        scale    = level_scale[rem6[qp]] << (div6[qp]);
 -        scale_m  = 16; // default when no custom scaling lists.
 -        dc_scale = 16;
 -
 -        if (s->sps->scaling_list_enable_flag) {
 -            const ScalingList *sl = s->pps->scaling_list_data_present_flag ?
 -                                    &s->pps->scaling_list : &s->sps->scaling_list;
 -            int matrix_id = lc->cu.pred_mode != MODE_INTRA;
 -
 -            if (log2_trafo_size != 5)
 -                matrix_id = 3 * matrix_id + c_idx;
 +static int hls_cross_component_pred(HEVCContext *s, int idx) {
 +    HEVCLocalContext *lc    = s->HEVClc;
 +    int log2_res_scale_abs_plus1 = ff_hevc_log2_res_scale_abs(s, idx);
  
 -            scale_matrix = sl->sl[log2_trafo_size - 2][matrix_id];
 -            if (log2_trafo_size >= 4)
 -                dc_scale = sl->sl_dc[log2_trafo_size - 4][matrix_id];
 -        }
 -    }
 -
 -    if (s->pps->transform_skip_enabled_flag &&
 -        !lc->cu.cu_transquant_bypass_flag   &&
 -        log2_trafo_size == 2) {
 -        transform_skip_flag = ff_hevc_transform_skip_flag_decode(s, c_idx);
 -    }
 -
 -    last_significant_coeff_x =
 -        ff_hevc_last_significant_coeff_x_prefix_decode(s, c_idx, log2_trafo_size);
 -    last_significant_coeff_y =
 -        ff_hevc_last_significant_coeff_y_prefix_decode(s, c_idx, log2_trafo_size);
 -
 -    if (last_significant_coeff_x > 3) {
 -        int suffix = ff_hevc_last_significant_coeff_suffix_decode(s, last_significant_coeff_x);
 -        last_significant_coeff_x = (1 << ((last_significant_coeff_x >> 1) - 1)) *
 -                                   (2 + (last_significant_coeff_x & 1)) +
 -                                   suffix;
 -    }
 -
 -    if (last_significant_coeff_y > 3) {
 -        int suffix = ff_hevc_last_significant_coeff_suffix_decode(s, last_significant_coeff_y);
 -        last_significant_coeff_y = (1 << ((last_significant_coeff_y >> 1) - 1)) *
 -                                   (2 + (last_significant_coeff_y & 1)) +
 -                                   suffix;
 -    }
 -
 -    if (scan_idx == SCAN_VERT)
 -        FFSWAP(int, last_significant_coeff_x, last_significant_coeff_y);
 -
 -    x_cg_last_sig = last_significant_coeff_x >> 2;
 -    y_cg_last_sig = last_significant_coeff_y >> 2;
 -
 -    switch (scan_idx) {
 -    case SCAN_DIAG: {
 -        int last_x_c = last_significant_coeff_x & 3;
 -        int last_y_c = last_significant_coeff_y & 3;
 -
 -        scan_x_off = ff_hevc_diag_scan4x4_x;
 -        scan_y_off = ff_hevc_diag_scan4x4_y;
 -        num_coeff  = diag_scan4x4_inv[last_y_c][last_x_c];
 -        if (trafo_size == 4) {
 -            scan_x_cg = scan_1x1;
 -            scan_y_cg = scan_1x1;
 -        } else if (trafo_size == 8) {
 -            num_coeff += diag_scan2x2_inv[y_cg_last_sig][x_cg_last_sig] << 4;
 -            scan_x_cg  = diag_scan2x2_x;
 -            scan_y_cg  = diag_scan2x2_y;
 -        } else if (trafo_size == 16) {
 -            num_coeff += diag_scan4x4_inv[y_cg_last_sig][x_cg_last_sig] << 4;
 -            scan_x_cg  = ff_hevc_diag_scan4x4_x;
 -            scan_y_cg  = ff_hevc_diag_scan4x4_y;
 -        } else { // trafo_size == 32
 -            num_coeff += diag_scan8x8_inv[y_cg_last_sig][x_cg_last_sig] << 4;
 -            scan_x_cg  = ff_hevc_diag_scan8x8_x;
 -            scan_y_cg  = ff_hevc_diag_scan8x8_y;
 -        }
 -        break;
 -    }
 -    case SCAN_HORIZ:
 -        scan_x_cg  = horiz_scan2x2_x;
 -        scan_y_cg  = horiz_scan2x2_y;
 -        scan_x_off = horiz_scan4x4_x;
 -        scan_y_off = horiz_scan4x4_y;
 -        num_coeff  = horiz_scan8x8_inv[last_significant_coeff_y][last_significant_coeff_x];
 -        break;
 -    default: //SCAN_VERT
 -        scan_x_cg  = horiz_scan2x2_y;
 -        scan_y_cg  = horiz_scan2x2_x;
 -        scan_x_off = horiz_scan4x4_y;
 -        scan_y_off = horiz_scan4x4_x;
 -        num_coeff  = horiz_scan8x8_inv[last_significant_coeff_x][last_significant_coeff_y];
 -        break;
 +    if (log2_res_scale_abs_plus1 !=  0) {
 +        int res_scale_sign_flag = ff_hevc_res_scale_sign_flag(s, idx);
 +        lc->tu.res_scale_val = (1 << (log2_res_scale_abs_plus1 - 1)) *
 +                               (1 - 2 * res_scale_sign_flag);
 +    } else {
 +        lc->tu.res_scale_val = 0;
      }
 -    num_coeff++;
 -    num_last_subset = (num_coeff - 1) >> 4;
 -
 -    for (i = num_last_subset; i >= 0; i--) {
 -        int n, m;
 -        int x_cg, y_cg, x_c, y_c;
 -        int implicit_non_zero_coeff = 0;
 -        int64_t trans_coeff_level;
 -        int prev_sig = 0;
 -        int offset   = i << 4;
 -
 -        uint8_t significant_coeff_flag_idx[16];
 -        uint8_t nb_significant_coeff_flag = 0;
 -
 -        x_cg = scan_x_cg[i];
 -        y_cg = scan_y_cg[i];
 -
 -        if (i < num_last_subset && i > 0) {
 -            int ctx_cg = 0;
 -            if (x_cg < (1 << (log2_trafo_size - 2)) - 1)
 -                ctx_cg += significant_coeff_group_flag[x_cg + 1][y_cg];
 -            if (y_cg < (1 << (log2_trafo_size - 2)) - 1)
 -                ctx_cg += significant_coeff_group_flag[x_cg][y_cg + 1];
 -
 -            significant_coeff_group_flag[x_cg][y_cg] =
 -                ff_hevc_significant_coeff_group_flag_decode(s, c_idx, ctx_cg);
 -            implicit_non_zero_coeff = 1;
 -        } else {
 -            significant_coeff_group_flag[x_cg][y_cg] =
 -                ((x_cg == x_cg_last_sig && y_cg == y_cg_last_sig) ||
 -                 (x_cg == 0 && y_cg == 0));
 -        }
  
 -        last_scan_pos = num_coeff - offset - 1;
  
 -        if (i == num_last_subset) {
 -            n_end                         = last_scan_pos - 1;
 -            significant_coeff_flag_idx[0] = last_scan_pos;
 -            nb_significant_coeff_flag     = 1;
 -        } else {
 -            n_end = 15;
 -        }
 -
 -        if (x_cg < ((1 << log2_trafo_size) - 1) >> 2)
 -            prev_sig = significant_coeff_group_flag[x_cg + 1][y_cg];
 -        if (y_cg < ((1 << log2_trafo_size) - 1) >> 2)
 -            prev_sig += significant_coeff_group_flag[x_cg][y_cg + 1] << 1;
 -
 -        for (n = n_end; n >= 0; n--) {
 -            GET_COORD(offset, n);
 -
 -            if (significant_coeff_group_flag[x_cg][y_cg] &&
 -                (n > 0 || implicit_non_zero_coeff == 0)) {
 -                if (ff_hevc_significant_coeff_flag_decode(s, c_idx, x_c, y_c,
 -                                                          log2_trafo_size,
 -                                                          scan_idx,
 -                                                          prev_sig) == 1) {
 -                    significant_coeff_flag_idx[nb_significant_coeff_flag] = n;
 -                    nb_significant_coeff_flag++;
 -                    implicit_non_zero_coeff = 0;
 -                }
 -            } else {
 -                int last_cg = (x_c == (x_cg << 2) && y_c == (y_cg << 2));
 -                if (last_cg && implicit_non_zero_coeff && significant_coeff_group_flag[x_cg][y_cg]) {
 -                    significant_coeff_flag_idx[nb_significant_coeff_flag] = n;
 -                    nb_significant_coeff_flag++;
 -                }
 -            }
 -        }
 -
 -        n_end = nb_significant_coeff_flag;
 -
 -        if (n_end) {
 -            int first_nz_pos_in_cg = 16;
 -            int last_nz_pos_in_cg = -1;
 -            int c_rice_param = 0;
 -            int first_greater1_coeff_idx = -1;
 -            uint8_t coeff_abs_level_greater1_flag[16] = { 0 };
 -            uint16_t coeff_sign_flag;
 -            int sum_abs = 0;
 -            int sign_hidden = 0;
 -
 -            // initialize first elem of coeff_bas_level_greater1_flag
 -            int ctx_set = (i > 0 && c_idx == 0) ? 2 : 0;
 -
 -            if (!(i == num_last_subset) && greater1_ctx == 0)
 -                ctx_set++;
 -            greater1_ctx      = 1;
 -            last_nz_pos_in_cg = significant_coeff_flag_idx[0];
 -
 -            for (m = 0; m < (n_end > 8 ? 8 : n_end); m++) {
 -                int n_idx = significant_coeff_flag_idx[m];
 -                int inc   = (ctx_set << 2) + greater1_ctx;
 -                coeff_abs_level_greater1_flag[n_idx] =
 -                    ff_hevc_coeff_abs_level_greater1_flag_decode(s, c_idx, inc);
 -                if (coeff_abs_level_greater1_flag[n_idx]) {
 -                    greater1_ctx = 0;
 -                } else if (greater1_ctx > 0 && greater1_ctx < 3) {
 -                    greater1_ctx++;
 -                }
 -
 -                if (coeff_abs_level_greater1_flag[n_idx] &&
 -                    first_greater1_coeff_idx == -1)
 -                    first_greater1_coeff_idx = n_idx;
 -            }
 -            first_nz_pos_in_cg = significant_coeff_flag_idx[n_end - 1];
 -            sign_hidden        = last_nz_pos_in_cg - first_nz_pos_in_cg >= 4 &&
 -                                 !lc->cu.cu_transquant_bypass_flag;
 -
 -            if (first_greater1_coeff_idx != -1) {
 -                coeff_abs_level_greater1_flag[first_greater1_coeff_idx] += ff_hevc_coeff_abs_level_greater2_flag_decode(s, c_idx, ctx_set);
 -            }
 -            if (!s->pps->sign_data_hiding_flag || !sign_hidden) {
 -                coeff_sign_flag = ff_hevc_coeff_sign_flag(s, nb_significant_coeff_flag) << (16 - nb_significant_coeff_flag);
 -            } else {
 -                coeff_sign_flag = ff_hevc_coeff_sign_flag(s, nb_significant_coeff_flag - 1) << (16 - (nb_significant_coeff_flag - 1));
 -            }
 -
 -            for (m = 0; m < n_end; m++) {
 -                n = significant_coeff_flag_idx[m];
 -                GET_COORD(offset, n);
 -                trans_coeff_level = 1 + coeff_abs_level_greater1_flag[n];
 -                if (trans_coeff_level == ((m < 8) ?
 -                                          ((n == first_greater1_coeff_idx) ? 3 : 2) : 1)) {
 -                    int last_coeff_abs_level_remaining = ff_hevc_coeff_abs_level_remaining(s, trans_coeff_level, c_rice_param);
 -
 -                    trans_coeff_level += last_coeff_abs_level_remaining;
 -                    if ((trans_coeff_level) > (3 * (1 << c_rice_param)))
 -                        c_rice_param = FFMIN(c_rice_param + 1, 4);
 -                }
 -                if (s->pps->sign_data_hiding_flag && sign_hidden) {
 -                    sum_abs += trans_coeff_level;
 -                    if (n == first_nz_pos_in_cg && ((sum_abs & 1) == 1))
 -                        trans_coeff_level = -trans_coeff_level;
 -                }
 -                if (coeff_sign_flag >> 15)
 -                    trans_coeff_level = -trans_coeff_level;
 -                coeff_sign_flag <<= 1;
 -                if (!lc->cu.cu_transquant_bypass_flag) {
 -                    if (s->sps->scaling_list_enable_flag) {
 -                        if (y_c || x_c || log2_trafo_size < 4) {
 -                            int pos;
 -                            switch (log2_trafo_size) {
 -                            case 3:  pos = (y_c        << 3) +  x_c;       break;
 -                            case 4:  pos = ((y_c >> 1) << 3) + (x_c >> 1); break;
 -                            case 5:  pos = ((y_c >> 2) << 3) + (x_c >> 2); break;
 -                            default: pos = (y_c        << 2) +  x_c;
 -                            }
 -                            scale_m = scale_matrix[pos];
 -                        } else {
 -                            scale_m = dc_scale;
 -                        }
 -                    }
 -                    trans_coeff_level = (trans_coeff_level * (int64_t)scale * (int64_t)scale_m + add) >> shift;
 -                    if(trans_coeff_level < 0) {
 -                        if((~trans_coeff_level) & 0xFffffffffff8000)
 -                            trans_coeff_level = -32768;
 -                    } else {
 -                        if (trans_coeff_level & 0xffffffffffff8000)
 -                            trans_coeff_level = 32767;
 -                    }
 -                }
 -                coeffs[y_c * trafo_size + x_c] = trans_coeff_level;
 -            }
 -        }
 -    }
 -
 -    if (lc->cu.cu_transquant_bypass_flag) {
 -        s->hevcdsp.transquant_bypass[log2_trafo_size - 2](dst, coeffs, stride);
 -    } else {
 -        if (transform_skip_flag)
 -            s->hevcdsp.transform_skip(dst, coeffs, stride);
 -        else if (lc->cu.pred_mode == MODE_INTRA && c_idx == 0 &&
 -                 log2_trafo_size == 2)
 -            s->hevcdsp.transform_4x4_luma_add(dst, coeffs, stride);
 -        else
 -            s->hevcdsp.transform_add[log2_trafo_size - 2](dst, coeffs, stride);
 -    }
 +    return 0;
  }
  
  static int hls_transform_unit(HEVCContext *s, int x0, int y0,
                                int xBase, int yBase, int cb_xBase, int cb_yBase,
                                int log2_cb_size, int log2_trafo_size,
 -                              int blk_idx, int cbf_luma, int cbf_cb, int cbf_cr)
 +                              int blk_idx, int cbf_luma, int *cbf_cb, int *cbf_cr)
  {
 -    HEVCLocalContext *lc = &s->HEVClc;
 +    HEVCLocalContext *lc = s->HEVClc;
 +    const int log2_trafo_size_c = log2_trafo_size - s->sps->hshift[1];
 +    int i;
  
      if (lc->cu.pred_mode == MODE_INTRA) {
          int trafo_size = 1 << log2_trafo_size;
          ff_hevc_set_neighbour_available(s, x0, y0, trafo_size, trafo_size);
  
          s->hpc.intra_pred[log2_trafo_size - 2](s, x0, y0, 0);
 -        if (log2_trafo_size > 2) {
 -            trafo_size = trafo_size << (s->sps->hshift[1] - 1);
 -            ff_hevc_set_neighbour_available(s, x0, y0, trafo_size, trafo_size);
 -            s->hpc.intra_pred[log2_trafo_size - 3](s, x0, y0, 1);
 -            s->hpc.intra_pred[log2_trafo_size - 3](s, x0, y0, 2);
 -        } else if (blk_idx == 3) {
 -            trafo_size = trafo_size << s->sps->hshift[1];
 -            ff_hevc_set_neighbour_available(s, xBase, yBase,
 -                                            trafo_size, trafo_size);
 -            s->hpc.intra_pred[log2_trafo_size - 2](s, xBase, yBase, 1);
 -            s->hpc.intra_pred[log2_trafo_size - 2](s, xBase, yBase, 2);
 -        }
      }
  
 -    if (cbf_luma || cbf_cb || cbf_cr) {
 +    if (cbf_luma || cbf_cb[0] || cbf_cr[0] ||
 +        (s->sps->chroma_format_idc == 2 && (cbf_cb[1] || cbf_cr[1]))) {
          int scan_idx   = SCAN_DIAG;
          int scan_idx_c = SCAN_DIAG;
 +        int cbf_chroma = cbf_cb[0] || cbf_cr[0] ||
 +                         (s->sps->chroma_format_idc == 2 &&
 +                         (cbf_cb[1] || cbf_cr[1]));
  
          if (s->pps->cu_qp_delta_enabled_flag && !lc->tu.is_cu_qp_delta_coded) {
              lc->tu.cu_qp_delta = ff_hevc_cu_qp_delta_abs(s);
                  return AVERROR_INVALIDDATA;
              }
  
 -            ff_hevc_set_qPy(s, x0, y0, cb_xBase, cb_yBase, log2_cb_size);
 +            ff_hevc_set_qPy(s, cb_xBase, cb_yBase, log2_cb_size);
 +        }
 +
 +        if (s->sh.cu_chroma_qp_offset_enabled_flag && cbf_chroma &&
 +            !lc->cu.cu_transquant_bypass_flag  &&  !lc->tu.is_cu_chroma_qp_offset_coded) {
 +            int cu_chroma_qp_offset_flag = ff_hevc_cu_chroma_qp_offset_flag(s);
 +            if (cu_chroma_qp_offset_flag) {
 +                int cu_chroma_qp_offset_idx  = 0;
 +                if (s->pps->chroma_qp_offset_list_len_minus1 > 0) {
 +                    cu_chroma_qp_offset_idx = ff_hevc_cu_chroma_qp_offset_idx(s);
 +                    av_log(s->avctx, AV_LOG_ERROR,
 +                        "cu_chroma_qp_offset_idx not yet tested.\n");
 +                }
 +                lc->tu.cu_qp_offset_cb = s->pps->cb_qp_offset_list[cu_chroma_qp_offset_idx];
 +                lc->tu.cu_qp_offset_cr = s->pps->cr_qp_offset_list[cu_chroma_qp_offset_idx];
 +            } else {
 +                lc->tu.cu_qp_offset_cb = 0;
 +                lc->tu.cu_qp_offset_cr = 0;
 +            }
 +            lc->tu.is_cu_chroma_qp_offset_coded = 1;
          }
  
          if (lc->cu.pred_mode == MODE_INTRA && log2_trafo_size < 4) {
 -            if (lc->tu.cur_intra_pred_mode >= 6 &&
 -                lc->tu.cur_intra_pred_mode <= 14) {
 +            if (lc->tu.intra_pred_mode >= 6 &&
 +                lc->tu.intra_pred_mode <= 14) {
                  scan_idx = SCAN_VERT;
 -            } else if (lc->tu.cur_intra_pred_mode >= 22 &&
 -                       lc->tu.cur_intra_pred_mode <= 30) {
 +            } else if (lc->tu.intra_pred_mode >= 22 &&
 +                       lc->tu.intra_pred_mode <= 30) {
                  scan_idx = SCAN_HORIZ;
              }
  
 -            if (lc->pu.intra_pred_mode_c >=  6 &&
 -                lc->pu.intra_pred_mode_c <= 14) {
 +            if (lc->tu.intra_pred_mode_c >=  6 &&
 +                lc->tu.intra_pred_mode_c <= 14) {
                  scan_idx_c = SCAN_VERT;
 -            } else if (lc->pu.intra_pred_mode_c >= 22 &&
 -                       lc->pu.intra_pred_mode_c <= 30) {
 +            } else if (lc->tu.intra_pred_mode_c >= 22 &&
 +                       lc->tu.intra_pred_mode_c <= 30) {
                  scan_idx_c = SCAN_HORIZ;
              }
          }
  
 +        lc->tu.cross_pf = 0;
 +
          if (cbf_luma)
 -            hls_residual_coding(s, x0, y0, log2_trafo_size, scan_idx, 0);
 -        if (log2_trafo_size > 2) {
 -            if (cbf_cb)
 -                hls_residual_coding(s, x0, y0, log2_trafo_size - 1, scan_idx_c, 1);
 -            if (cbf_cr)
 -                hls_residual_coding(s, x0, y0, log2_trafo_size - 1, scan_idx_c, 2);
 +            ff_hevc_hls_residual_coding(s, x0, y0, log2_trafo_size, scan_idx, 0);
 +        if (s->sps->chroma_format_idc && (log2_trafo_size > 2 || s->sps->chroma_format_idc == 3)) {
 +            int trafo_size_h = 1 << (log2_trafo_size_c + s->sps->hshift[1]);
 +            int trafo_size_v = 1 << (log2_trafo_size_c + s->sps->vshift[1]);
 +            lc->tu.cross_pf  = (s->pps->cross_component_prediction_enabled_flag && cbf_luma &&
 +                                (lc->cu.pred_mode == MODE_INTER ||
 +                                 (lc->tu.chroma_mode_c ==  4)));
 +
 +            if (lc->tu.cross_pf) {
 +                hls_cross_component_pred(s, 0);
 +            }
 +            for (i = 0; i < (s->sps->chroma_format_idc == 2 ? 2 : 1); i++) {
 +                if (lc->cu.pred_mode == MODE_INTRA) {
 +                    ff_hevc_set_neighbour_available(s, x0, y0 + (i << log2_trafo_size_c), trafo_size_h, trafo_size_v);
 +                    s->hpc.intra_pred[log2_trafo_size_c - 2](s, x0, y0 + (i << log2_trafo_size_c), 1);
 +                }
 +                if (cbf_cb[i])
 +                    ff_hevc_hls_residual_coding(s, x0, y0 + (i << log2_trafo_size_c),
 +                                                log2_trafo_size_c, scan_idx_c, 1);
 +                else
 +                    if (lc->tu.cross_pf) {
 +                        ptrdiff_t stride = s->frame->linesize[1];
 +                        int hshift = s->sps->hshift[1];
 +                        int vshift = s->sps->vshift[1];
 +                        int16_t *coeffs_y = (int16_t*)lc->edge_emu_buffer;
 +                        int16_t *coeffs   = (int16_t*)lc->edge_emu_buffer2;
 +                        int size = 1 << log2_trafo_size_c;
 +
 +                        uint8_t *dst = &s->frame->data[1][(y0 >> vshift) * stride +
 +                                                              ((x0 >> hshift) << s->sps->pixel_shift)];
 +                        for (i = 0; i < (size * size); i++) {
 +                            coeffs[i] = ((lc->tu.res_scale_val * coeffs_y[i]) >> 3);
 +                        }
 +                        s->hevcdsp.transform_add[log2_trafo_size_c-2](dst, coeffs, stride);
 +                    }
 +            }
 +
 +            if (lc->tu.cross_pf) {
 +                hls_cross_component_pred(s, 1);
 +            }
 +            for (i = 0; i < (s->sps->chroma_format_idc == 2 ? 2 : 1); i++) {
 +                if (lc->cu.pred_mode == MODE_INTRA) {
 +                    ff_hevc_set_neighbour_available(s, x0, y0 + (i << log2_trafo_size_c), trafo_size_h, trafo_size_v);
 +                    s->hpc.intra_pred[log2_trafo_size_c - 2](s, x0, y0 + (i << log2_trafo_size_c), 2);
 +                }
 +                if (cbf_cr[i])
 +                    ff_hevc_hls_residual_coding(s, x0, y0 + (i << log2_trafo_size_c),
 +                                                log2_trafo_size_c, scan_idx_c, 2);
 +                else
 +                    if (lc->tu.cross_pf) {
 +                        ptrdiff_t stride = s->frame->linesize[2];
 +                        int hshift = s->sps->hshift[2];
 +                        int vshift = s->sps->vshift[2];
 +                        int16_t *coeffs_y = (int16_t*)lc->edge_emu_buffer;
 +                        int16_t *coeffs   = (int16_t*)lc->edge_emu_buffer2;
 +                        int size = 1 << log2_trafo_size_c;
 +
 +                        uint8_t *dst = &s->frame->data[2][(y0 >> vshift) * stride +
 +                                                          ((x0 >> hshift) << s->sps->pixel_shift)];
 +                        for (i = 0; i < (size * size); i++) {
 +                            coeffs[i] = ((lc->tu.res_scale_val * coeffs_y[i]) >> 3);
 +                        }
 +                        s->hevcdsp.transform_add[log2_trafo_size_c-2](dst, coeffs, stride);
 +                    }
 +            }
 +        } else if (s->sps->chroma_format_idc && blk_idx == 3) {
 +            int trafo_size_h = 1 << (log2_trafo_size + 1);
 +            int trafo_size_v = 1 << (log2_trafo_size + s->sps->vshift[1]);
 +            for (i = 0; i < (s->sps->chroma_format_idc == 2 ? 2 : 1); i++) {
 +                if (lc->cu.pred_mode == MODE_INTRA) {
 +                    ff_hevc_set_neighbour_available(s, xBase, yBase + (i << log2_trafo_size),
 +                                                    trafo_size_h, trafo_size_v);
 +                    s->hpc.intra_pred[log2_trafo_size - 2](s, xBase, yBase + (i << log2_trafo_size), 1);
 +                }
 +                if (cbf_cb[i])
 +                    ff_hevc_hls_residual_coding(s, xBase, yBase + (i << log2_trafo_size),
 +                                                log2_trafo_size, scan_idx_c, 1);
 +            }
 +            for (i = 0; i < (s->sps->chroma_format_idc == 2 ? 2 : 1); i++) {
 +                if (lc->cu.pred_mode == MODE_INTRA) {
 +                    ff_hevc_set_neighbour_available(s, xBase, yBase + (i << log2_trafo_size),
 +                                                trafo_size_h, trafo_size_v);
 +                    s->hpc.intra_pred[log2_trafo_size - 2](s, xBase, yBase + (i << log2_trafo_size), 2);
 +                }
 +                if (cbf_cr[i])
 +                    ff_hevc_hls_residual_coding(s, xBase, yBase + (i << log2_trafo_size),
 +                                                log2_trafo_size, scan_idx_c, 2);
 +            }
 +        }
 +    } else if (s->sps->chroma_format_idc && lc->cu.pred_mode == MODE_INTRA) {
 +        if (log2_trafo_size > 2 || s->sps->chroma_format_idc == 3) {
 +            int trafo_size_h = 1 << (log2_trafo_size_c + s->sps->hshift[1]);
 +            int trafo_size_v = 1 << (log2_trafo_size_c + s->sps->vshift[1]);
 +            ff_hevc_set_neighbour_available(s, x0, y0, trafo_size_h, trafo_size_v);
 +            s->hpc.intra_pred[log2_trafo_size_c - 2](s, x0, y0, 1);
 +            s->hpc.intra_pred[log2_trafo_size_c - 2](s, x0, y0, 2);
 +            if (s->sps->chroma_format_idc == 2) {
 +                ff_hevc_set_neighbour_available(s, x0, y0 + (1 << log2_trafo_size_c),
 +                                                trafo_size_h, trafo_size_v);
 +                s->hpc.intra_pred[log2_trafo_size_c - 2](s, x0, y0 + (1 << log2_trafo_size_c), 1);
 +                s->hpc.intra_pred[log2_trafo_size_c - 2](s, x0, y0 + (1 << log2_trafo_size_c), 2);
 +            }
          } else if (blk_idx == 3) {
 -            if (cbf_cb)
 -                hls_residual_coding(s, xBase, yBase, log2_trafo_size, scan_idx_c, 1);
 -            if (cbf_cr)
 -                hls_residual_coding(s, xBase, yBase, log2_trafo_size, scan_idx_c, 2);
 +            int trafo_size_h = 1 << (log2_trafo_size + 1);
 +            int trafo_size_v = 1 << (log2_trafo_size + s->sps->vshift[1]);
 +            ff_hevc_set_neighbour_available(s, xBase, yBase,
 +                                            trafo_size_h, trafo_size_v);
 +            s->hpc.intra_pred[log2_trafo_size - 2](s, xBase, yBase, 1);
 +            s->hpc.intra_pred[log2_trafo_size - 2](s, xBase, yBase, 2);
 +            if (s->sps->chroma_format_idc == 2) {
 +                ff_hevc_set_neighbour_available(s, xBase, yBase + (1 << (log2_trafo_size)),
 +                                                trafo_size_h, trafo_size_v);
 +                s->hpc.intra_pred[log2_trafo_size - 2](s, xBase, yBase + (1 << (log2_trafo_size)), 1);
 +                s->hpc.intra_pred[log2_trafo_size - 2](s, xBase, yBase + (1 << (log2_trafo_size)), 2);
 +            }
          }
      }
 +
      return 0;
  }
  
@@@ -1132,34 -1363,17 +1132,34 @@@ static int hls_transform_tree(HEVCConte
                                int xBase, int yBase, int cb_xBase, int cb_yBase,
                                int log2_cb_size, int log2_trafo_size,
                                int trafo_depth, int blk_idx,
 -                              int cbf_cb, int cbf_cr)
 +                              const int *base_cbf_cb, const int *base_cbf_cr)
  {
 -    HEVCLocalContext *lc = &s->HEVClc;
 +    HEVCLocalContext *lc = s->HEVClc;
      uint8_t split_transform_flag;
 +    int cbf_cb[2];
 +    int cbf_cr[2];
      int ret;
  
 +    cbf_cb[0] = base_cbf_cb[0];
 +    cbf_cb[1] = base_cbf_cb[1];
 +    cbf_cr[0] = base_cbf_cr[0];
 +    cbf_cr[1] = base_cbf_cr[1];
 +
      if (lc->cu.intra_split_flag) {
 -        if (trafo_depth == 1)
 -            lc->tu.cur_intra_pred_mode = lc->pu.intra_pred_mode[blk_idx];
 +        if (trafo_depth == 1) {
 +            lc->tu.intra_pred_mode   = lc->pu.intra_pred_mode[blk_idx];
 +            if (s->sps->chroma_format_idc == 3) {
 +                lc->tu.intra_pred_mode_c = lc->pu.intra_pred_mode_c[blk_idx];
 +                lc->tu.chroma_mode_c     = lc->pu.chroma_mode_c[blk_idx];
 +            } else {
 +                lc->tu.intra_pred_mode_c = lc->pu.intra_pred_mode_c[0];
 +                lc->tu.chroma_mode_c     = lc->pu.chroma_mode_c[0];
 +            }
 +        }
      } else {
 -        lc->tu.cur_intra_pred_mode = lc->pu.intra_pred_mode[0];
 +        lc->tu.intra_pred_mode   = lc->pu.intra_pred_mode[0];
 +        lc->tu.intra_pred_mode_c = lc->pu.intra_pred_mode_c[0];
 +        lc->tu.chroma_mode_c     = lc->pu.chroma_mode_c[0];
      }
  
      if (log2_trafo_size <= s->sps->log2_max_trafo_size &&
                                 inter_split;
      }
  
 -    if (log2_trafo_size > 2 && (trafo_depth == 0 || cbf_cb))
 -        cbf_cb = ff_hevc_cbf_cb_cr_decode(s, trafo_depth);
 -    else if (log2_trafo_size > 2 || trafo_depth == 0)
 -        cbf_cb = 0;
 -    if (log2_trafo_size > 2 && (trafo_depth == 0 || cbf_cr))
 -        cbf_cr = ff_hevc_cbf_cb_cr_decode(s, trafo_depth);
 -    else if (log2_trafo_size > 2 || trafo_depth == 0)
 -        cbf_cr = 0;
 +    if (s->sps->chroma_format_idc && (log2_trafo_size > 2 || s->sps->chroma_format_idc == 3)) {
 +        if (trafo_depth == 0 || cbf_cb[0]) {
 +            cbf_cb[0] = ff_hevc_cbf_cb_cr_decode(s, trafo_depth);
 +            if (s->sps->chroma_format_idc == 2 && (!split_transform_flag || log2_trafo_size == 3)) {
 +                cbf_cb[1] = ff_hevc_cbf_cb_cr_decode(s, trafo_depth);
 +            }
 +        }
 +
 +        if (trafo_depth == 0 || cbf_cr[0]) {
 +            cbf_cr[0] = ff_hevc_cbf_cb_cr_decode(s, trafo_depth);
 +            if (s->sps->chroma_format_idc == 2 && (!split_transform_flag || log2_trafo_size == 3)) {
 +                cbf_cr[1] = ff_hevc_cbf_cb_cr_decode(s, trafo_depth);
 +            }
 +        }
 +    }
  
      if (split_transform_flag) {
          const int trafo_size_split = 1 << (log2_trafo_size - 1);
@@@ -1221,10 -1428,8 +1221,10 @@@ do 
          int cbf_luma         = 1;
  
          if (lc->cu.pred_mode == MODE_INTRA || trafo_depth != 0 ||
 -            cbf_cb || cbf_cr)
 +            cbf_cb[0] || cbf_cr[0] ||
 +            (s->sps->chroma_format_idc == 2 && (cbf_cb[1] || cbf_cr[1]))) {
              cbf_luma = ff_hevc_cbf_luma_decode(s, trafo_depth);
 +        }
  
          ret = hls_transform_unit(s, x0, y0, xBase, yBase, cb_xBase, cb_yBase,
                                   log2_cb_size, log2_trafo_size,
  
  static int hls_pcm_sample(HEVCContext *s, int x0, int y0, int log2_cb_size)
  {
 -    //TODO: non-4:2:0 support
 -    HEVCLocalContext *lc = &s->HEVClc;
 +    HEVCLocalContext *lc = s->HEVClc;
      GetBitContext gb;
      int cb_size   = 1 << log2_cb_size;
      int stride0   = s->frame->linesize[0];
      int   stride2 = s->frame->linesize[2];
      uint8_t *dst2 = &s->frame->data[2][(y0 >> s->sps->vshift[2]) * stride2 + ((x0 >> s->sps->hshift[2]) << s->sps->pixel_shift)];
  
 -    int length         = cb_size * cb_size * s->sps->pcm.bit_depth + ((cb_size * cb_size) >> 1) * s->sps->pcm.bit_depth_chroma;
 +    int length         = cb_size * cb_size * s->sps->pcm.bit_depth +
 +                         (((cb_size >> s->sps->hshift[1]) * (cb_size >> s->sps->vshift[1])) +
 +                          ((cb_size >> s->sps->hshift[2]) * (cb_size >> s->sps->vshift[2]))) *
 +                          s->sps->pcm.bit_depth_chroma;
      const uint8_t *pcm = skip_bytes(&lc->cc, (length + 7) >> 3);
      int ret;
  
      if (ret < 0)
          return ret;
  
 -    s->hevcdsp.put_pcm(dst0, stride0, cb_size,     &gb, s->sps->pcm.bit_depth);
 -    s->hevcdsp.put_pcm(dst1, stride1, cb_size / 2, &gb, s->sps->pcm.bit_depth_chroma);
 -    s->hevcdsp.put_pcm(dst2, stride2, cb_size / 2, &gb, s->sps->pcm.bit_depth_chroma);
 -    return 0;
 -}
 -
 -static void hls_mvd_coding(HEVCContext *s, int x0, int y0, int log2_cb_size)
 -{
 -    HEVCLocalContext *lc = &s->HEVClc;
 -    int x = ff_hevc_abs_mvd_greater0_flag_decode(s);
 -    int y = ff_hevc_abs_mvd_greater0_flag_decode(s);
 -
 -    if (x)
 -        x += ff_hevc_abs_mvd_greater1_flag_decode(s);
 -    if (y)
 -        y += ff_hevc_abs_mvd_greater1_flag_decode(s);
 -
 -    switch (x) {
 -    case 2: lc->pu.mvd.x = ff_hevc_mvd_decode(s);           break;
 -    case 1: lc->pu.mvd.x = ff_hevc_mvd_sign_flag_decode(s); break;
 -    case 0: lc->pu.mvd.x = 0;                               break;
 +    s->hevcdsp.put_pcm(dst0, stride0, cb_size, cb_size,     &gb, s->sps->pcm.bit_depth);
 +    if (s->sps->chroma_format_idc) {
 +        s->hevcdsp.put_pcm(dst1, stride1,
 +                           cb_size >> s->sps->hshift[1],
 +                           cb_size >> s->sps->vshift[1],
 +                           &gb, s->sps->pcm.bit_depth_chroma);
 +        s->hevcdsp.put_pcm(dst2, stride2,
 +                           cb_size >> s->sps->hshift[2],
 +                           cb_size >> s->sps->vshift[2],
 +                           &gb, s->sps->pcm.bit_depth_chroma);
      }
  
 -    switch (y) {
 -    case 2: lc->pu.mvd.y = ff_hevc_mvd_decode(s);           break;
 -    case 1: lc->pu.mvd.y = ff_hevc_mvd_sign_flag_decode(s); break;
 -    case 0: lc->pu.mvd.y = 0;                               break;
 -    }
 +    return 0;
  }
  
  /**
 - * 8.5.3.2.2.1 Luma sample interpolation process
 + * 8.5.3.2.2.1 Luma sample unidirectional interpolation process
   *
   * @param s HEVC decoding context
   * @param dst target buffer for block data at block position
   * @param y_off vertical position of block from origin (0, 0)
   * @param block_w width of block
   * @param block_h height of block
 + * @param luma_weight weighting factor applied to the luma prediction
 + * @param luma_offset additive offset applied to the luma prediction value
   */
 -static void luma_mc(HEVCContext *s, int16_t *dst, ptrdiff_t dststride,
 -                    AVFrame *ref, const Mv *mv, int x_off, int y_off,
 -                    int block_w, int block_h)
 +
 +static void luma_mc_uni(HEVCContext *s, uint8_t *dst, ptrdiff_t dststride,
 +                        AVFrame *ref, const Mv *mv, int x_off, int y_off,
 +                        int block_w, int block_h, int luma_weight, int luma_offset)
  {
 -    HEVCLocalContext *lc = &s->HEVClc;
 +    HEVCLocalContext *lc = s->HEVClc;
      uint8_t *src         = ref->data[0];
      ptrdiff_t srcstride  = ref->linesize[0];
      int pic_width        = s->sps->width;
      int pic_height       = s->sps->height;
 -
 -    int mx         = mv->x & 3;
 -    int my         = mv->y & 3;
 -    int extra_left = ff_hevc_qpel_extra_before[mx];
 -    int extra_top  = ff_hevc_qpel_extra_before[my];
 +    int mx               = mv->x & 3;
 +    int my               = mv->y & 3;
 +    int weight_flag      = (s->sh.slice_type == P_SLICE && s->pps->weighted_pred_flag) ||
 +                           (s->sh.slice_type == B_SLICE && s->pps->weighted_bipred_flag);
 +    int idx              = ff_hevc_pel_weight[block_w];
  
      x_off += mv->x >> 2;
      y_off += mv->y >> 2;
 -    src   += y_off * srcstride + (x_off << s->sps->pixel_shift);
 +    src   += y_off * srcstride + x_off * (1 << s->sps->pixel_shift);
  
 -    if (x_off < extra_left || y_off < extra_top ||
 -        x_off >= pic_width - block_w - ff_hevc_qpel_extra_after[mx] ||
 -        y_off >= pic_height - block_h - ff_hevc_qpel_extra_after[my]) {
 +    if (x_off < QPEL_EXTRA_BEFORE || y_off < QPEL_EXTRA_AFTER ||
 +        x_off >= pic_width - block_w - QPEL_EXTRA_AFTER ||
 +        y_off >= pic_height - block_h - QPEL_EXTRA_AFTER) {
          const int edge_emu_stride = EDGE_EMU_BUFFER_STRIDE << s->sps->pixel_shift;
 -        int offset = extra_top * srcstride + (extra_left << s->sps->pixel_shift);
 -        int buf_offset = extra_top *
 -                         edge_emu_stride + (extra_left << s->sps->pixel_shift);
 +        int offset     = QPEL_EXTRA_BEFORE * srcstride       + (QPEL_EXTRA_BEFORE << s->sps->pixel_shift);
 +        int buf_offset = QPEL_EXTRA_BEFORE * edge_emu_stride + (QPEL_EXTRA_BEFORE << s->sps->pixel_shift);
  
          s->vdsp.emulated_edge_mc(lc->edge_emu_buffer, src - offset,
                                   edge_emu_stride, srcstride,
 -                                 block_w + ff_hevc_qpel_extra[mx],
 -                                 block_h + ff_hevc_qpel_extra[my],
 -                                 x_off - extra_left, y_off - extra_top,
 +                                 block_w + QPEL_EXTRA,
 +                                 block_h + QPEL_EXTRA,
 +                                 x_off - QPEL_EXTRA_BEFORE, y_off - QPEL_EXTRA_BEFORE,
                                   pic_width, pic_height);
          src = lc->edge_emu_buffer + buf_offset;
          srcstride = edge_emu_stride;
      }
 -    s->hevcdsp.put_hevc_qpel[my][mx](dst, dststride, src, srcstride, block_w,
 -                                     block_h, lc->mc_buffer);
 +
 +    if (!weight_flag)
 +        s->hevcdsp.put_hevc_qpel_uni[idx][!!my][!!mx](dst, dststride, src, srcstride,
 +                                                      block_h, mx, my, block_w);
 +    else
 +        s->hevcdsp.put_hevc_qpel_uni_w[idx][!!my][!!mx](dst, dststride, src, srcstride,
 +                                                        block_h, s->sh.luma_log2_weight_denom,
 +                                                        luma_weight, luma_offset, mx, my, block_w);
 +}
 +
 +/**
 + * 8.5.3.2.2.1 Luma sample bidirectional interpolation process
 + *
 + * @param s HEVC decoding context
 + * @param dst target buffer for block data at block position
 + * @param dststride stride of the dst buffer
 + * @param ref0 reference picture0 buffer at origin (0, 0)
 + * @param mv0 motion vector0 (relative to block position) to get pixel data from
 + * @param x_off horizontal position of block from origin (0, 0)
 + * @param y_off vertical position of block from origin (0, 0)
 + * @param block_w width of block
 + * @param block_h height of block
 + * @param ref1 reference picture1 buffer at origin (0, 0)
 + * @param mv1 motion vector1 (relative to block position) to get pixel data from
 + * @param current_mv current motion vector structure
 + */
 + static void luma_mc_bi(HEVCContext *s, uint8_t *dst, ptrdiff_t dststride,
 +                       AVFrame *ref0, const Mv *mv0, int x_off, int y_off,
 +                       int block_w, int block_h, AVFrame *ref1, const Mv *mv1, struct MvField *current_mv)
 +{
 +    HEVCLocalContext *lc = s->HEVClc;
 +    ptrdiff_t src0stride  = ref0->linesize[0];
 +    ptrdiff_t src1stride  = ref1->linesize[0];
 +    int pic_width        = s->sps->width;
 +    int pic_height       = s->sps->height;
 +    int mx0              = mv0->x & 3;
 +    int my0              = mv0->y & 3;
 +    int mx1              = mv1->x & 3;
 +    int my1              = mv1->y & 3;
 +    int weight_flag      = (s->sh.slice_type == P_SLICE && s->pps->weighted_pred_flag) ||
 +                           (s->sh.slice_type == B_SLICE && s->pps->weighted_bipred_flag);
 +    int x_off0           = x_off + (mv0->x >> 2);
 +    int y_off0           = y_off + (mv0->y >> 2);
 +    int x_off1           = x_off + (mv1->x >> 2);
 +    int y_off1           = y_off + (mv1->y >> 2);
 +    int idx              = ff_hevc_pel_weight[block_w];
 +
 +    uint8_t *src0  = ref0->data[0] + y_off0 * src0stride + (int)((unsigned)x_off0 << s->sps->pixel_shift);
 +    uint8_t *src1  = ref1->data[0] + y_off1 * src1stride + (int)((unsigned)x_off1 << s->sps->pixel_shift);
 +
 +    if (x_off0 < QPEL_EXTRA_BEFORE || y_off0 < QPEL_EXTRA_AFTER ||
 +        x_off0 >= pic_width - block_w - QPEL_EXTRA_AFTER ||
 +        y_off0 >= pic_height - block_h - QPEL_EXTRA_AFTER) {
 +        const int edge_emu_stride = EDGE_EMU_BUFFER_STRIDE << s->sps->pixel_shift;
 +        int offset     = QPEL_EXTRA_BEFORE * src0stride       + (QPEL_EXTRA_BEFORE << s->sps->pixel_shift);
 +        int buf_offset = QPEL_EXTRA_BEFORE * edge_emu_stride + (QPEL_EXTRA_BEFORE << s->sps->pixel_shift);
 +
 +        s->vdsp.emulated_edge_mc(lc->edge_emu_buffer, src0 - offset,
 +                                 edge_emu_stride, src0stride,
 +                                 block_w + QPEL_EXTRA,
 +                                 block_h + QPEL_EXTRA,
 +                                 x_off0 - QPEL_EXTRA_BEFORE, y_off0 - QPEL_EXTRA_BEFORE,
 +                                 pic_width, pic_height);
 +        src0 = lc->edge_emu_buffer + buf_offset;
 +        src0stride = edge_emu_stride;
 +    }
 +
 +    if (x_off1 < QPEL_EXTRA_BEFORE || y_off1 < QPEL_EXTRA_AFTER ||
 +        x_off1 >= pic_width - block_w - QPEL_EXTRA_AFTER ||
 +        y_off1 >= pic_height - block_h - QPEL_EXTRA_AFTER) {
 +        const int edge_emu_stride = EDGE_EMU_BUFFER_STRIDE << s->sps->pixel_shift;
 +        int offset     = QPEL_EXTRA_BEFORE * src1stride       + (QPEL_EXTRA_BEFORE << s->sps->pixel_shift);
 +        int buf_offset = QPEL_EXTRA_BEFORE * edge_emu_stride + (QPEL_EXTRA_BEFORE << s->sps->pixel_shift);
 +
 +        s->vdsp.emulated_edge_mc(lc->edge_emu_buffer2, src1 - offset,
 +                                 edge_emu_stride, src1stride,
 +                                 block_w + QPEL_EXTRA,
 +                                 block_h + QPEL_EXTRA,
 +                                 x_off1 - QPEL_EXTRA_BEFORE, y_off1 - QPEL_EXTRA_BEFORE,
 +                                 pic_width, pic_height);
 +        src1 = lc->edge_emu_buffer2 + buf_offset;
 +        src1stride = edge_emu_stride;
 +    }
 +
 +    s->hevcdsp.put_hevc_qpel[idx][!!my0][!!mx0](lc->tmp, src0, src0stride,
 +                                                block_h, mx0, my0, block_w);
 +    if (!weight_flag)
 +        s->hevcdsp.put_hevc_qpel_bi[idx][!!my1][!!mx1](dst, dststride, src1, src1stride, lc->tmp,
 +                                                       block_h, mx1, my1, block_w);
 +    else
 +        s->hevcdsp.put_hevc_qpel_bi_w[idx][!!my1][!!mx1](dst, dststride, src1, src1stride, lc->tmp,
 +                                                         block_h, s->sh.luma_log2_weight_denom,
 +                                                         s->sh.luma_weight_l0[current_mv->ref_idx[0]],
 +                                                         s->sh.luma_weight_l1[current_mv->ref_idx[1]],
 +                                                         s->sh.luma_offset_l0[current_mv->ref_idx[0]],
 +                                                         s->sh.luma_offset_l1[current_mv->ref_idx[1]],
 +                                                         mx1, my1, block_w);
 +
  }
  
  /**
 - * 8.5.3.2.2.2 Chroma sample interpolation process
 + * 8.5.3.2.2.2 Chroma sample uniprediction interpolation process
   *
   * @param s HEVC decoding context
   * @param dst1 target buffer for block data at block position (U plane)
   * @param y_off vertical position of block from origin (0, 0)
   * @param block_w width of block
   * @param block_h height of block
 + * @param chroma_weight weighting factor applied to the chroma prediction
 + * @param chroma_offset additive offset applied to the chroma prediction value
   */
 -static void chroma_mc(HEVCContext *s, int16_t *dst1, int16_t *dst2,
 -                      ptrdiff_t dststride, AVFrame *ref, const Mv *mv,
 -                      int x_off, int y_off, int block_w, int block_h)
 +
 +static void chroma_mc_uni(HEVCContext *s, uint8_t *dst0,
 +                          ptrdiff_t dststride, uint8_t *src0, ptrdiff_t srcstride, int reflist,
 +                          int x_off, int y_off, int block_w, int block_h, struct MvField *current_mv, int chroma_weight, int chroma_offset)
  {
 -    HEVCLocalContext *lc = &s->HEVClc;
 -    uint8_t *src1        = ref->data[1];
 -    uint8_t *src2        = ref->data[2];
 -    ptrdiff_t src1stride = ref->linesize[1];
 -    ptrdiff_t src2stride = ref->linesize[2];
 -    int pic_width        = s->sps->width >> 1;
 -    int pic_height       = s->sps->height >> 1;
 -
 -    int mx = mv->x & 7;
 -    int my = mv->y & 7;
 -
 -    x_off += mv->x >> 3;
 -    y_off += mv->y >> 3;
 -    src1  += y_off * src1stride + (x_off << s->sps->pixel_shift);
 -    src2  += y_off * src2stride + (x_off << s->sps->pixel_shift);
 +    HEVCLocalContext *lc = s->HEVClc;
 +    int pic_width        = s->sps->width >> s->sps->hshift[1];
 +    int pic_height       = s->sps->height >> s->sps->vshift[1];
 +    const Mv *mv         = &current_mv->mv[reflist];
 +    int weight_flag      = (s->sh.slice_type == P_SLICE && s->pps->weighted_pred_flag) ||
 +                           (s->sh.slice_type == B_SLICE && s->pps->weighted_bipred_flag);
 +    int idx              = ff_hevc_pel_weight[block_w];
 +    int hshift           = s->sps->hshift[1];
 +    int vshift           = s->sps->vshift[1];
 +    intptr_t mx          = av_mod_uintp2(mv->x, 2 + hshift);
 +    intptr_t my          = av_mod_uintp2(mv->y, 2 + vshift);
 +    intptr_t _mx         = mx << (1 - hshift);
 +    intptr_t _my         = my << (1 - vshift);
 +
 +    x_off += mv->x >> (2 + hshift);
 +    y_off += mv->y >> (2 + vshift);
 +    src0  += y_off * srcstride + x_off * (1 << s->sps->pixel_shift);
  
      if (x_off < EPEL_EXTRA_BEFORE || y_off < EPEL_EXTRA_AFTER ||
          x_off >= pic_width - block_w - EPEL_EXTRA_AFTER ||
          y_off >= pic_height - block_h - EPEL_EXTRA_AFTER) {
          const int edge_emu_stride = EDGE_EMU_BUFFER_STRIDE << s->sps->pixel_shift;
 +        int offset0 = EPEL_EXTRA_BEFORE * (srcstride + (1 << s->sps->pixel_shift));
 +        int buf_offset0 = EPEL_EXTRA_BEFORE *
 +                          (edge_emu_stride + (1 << s->sps->pixel_shift));
 +        s->vdsp.emulated_edge_mc(lc->edge_emu_buffer, src0 - offset0,
 +                                 edge_emu_stride, srcstride,
 +                                 block_w + EPEL_EXTRA, block_h + EPEL_EXTRA,
 +                                 x_off - EPEL_EXTRA_BEFORE,
 +                                 y_off - EPEL_EXTRA_BEFORE,
 +                                 pic_width, pic_height);
 +
 +        src0 = lc->edge_emu_buffer + buf_offset0;
 +        srcstride = edge_emu_stride;
 +    }
 +    if (!weight_flag)
 +        s->hevcdsp.put_hevc_epel_uni[idx][!!my][!!mx](dst0, dststride, src0, srcstride,
 +                                                  block_h, _mx, _my, block_w);
 +    else
 +        s->hevcdsp.put_hevc_epel_uni_w[idx][!!my][!!mx](dst0, dststride, src0, srcstride,
 +                                                        block_h, s->sh.chroma_log2_weight_denom,
 +                                                        chroma_weight, chroma_offset, _mx, _my, block_w);
 +}
 +
 +/**
 + * 8.5.3.2.2.2 Chroma sample bidirectional interpolation process
 + *
 + * @param s HEVC decoding context
 + * @param dst target buffer for block data at block position
 + * @param dststride stride of the dst buffer
 + * @param ref0 reference picture0 buffer at origin (0, 0)
 + * @param mv0 motion vector0 (relative to block position) to get pixel data from
 + * @param x_off horizontal position of block from origin (0, 0)
 + * @param y_off vertical position of block from origin (0, 0)
 + * @param block_w width of block
 + * @param block_h height of block
 + * @param ref1 reference picture1 buffer at origin (0, 0)
 + * @param mv1 motion vector1 (relative to block position) to get pixel data from
 + * @param current_mv current motion vector structure
 + * @param cidx chroma component(cb, cr)
 + */
 +static void chroma_mc_bi(HEVCContext *s, uint8_t *dst0, ptrdiff_t dststride, AVFrame *ref0, AVFrame *ref1,
 +                         int x_off, int y_off, int block_w, int block_h, struct MvField *current_mv, int cidx)
 +{
 +    HEVCLocalContext *lc = s->HEVClc;
 +    uint8_t *src1        = ref0->data[cidx+1];
 +    uint8_t *src2        = ref1->data[cidx+1];
 +    ptrdiff_t src1stride = ref0->linesize[cidx+1];
 +    ptrdiff_t src2stride = ref1->linesize[cidx+1];
 +    int weight_flag      = (s->sh.slice_type == P_SLICE && s->pps->weighted_pred_flag) ||
 +                           (s->sh.slice_type == B_SLICE && s->pps->weighted_bipred_flag);
 +    int pic_width        = s->sps->width >> s->sps->hshift[1];
 +    int pic_height       = s->sps->height >> s->sps->vshift[1];
 +    Mv *mv0              = &current_mv->mv[0];
 +    Mv *mv1              = &current_mv->mv[1];
 +    int hshift = s->sps->hshift[1];
 +    int vshift = s->sps->vshift[1];
 +
 +    intptr_t mx0 = av_mod_uintp2(mv0->x, 2 + hshift);
 +    intptr_t my0 = av_mod_uintp2(mv0->y, 2 + vshift);
 +    intptr_t mx1 = av_mod_uintp2(mv1->x, 2 + hshift);
 +    intptr_t my1 = av_mod_uintp2(mv1->y, 2 + vshift);
 +    intptr_t _mx0 = mx0 << (1 - hshift);
 +    intptr_t _my0 = my0 << (1 - vshift);
 +    intptr_t _mx1 = mx1 << (1 - hshift);
 +    intptr_t _my1 = my1 << (1 - vshift);
 +
 +    int x_off0 = x_off + (mv0->x >> (2 + hshift));
 +    int y_off0 = y_off + (mv0->y >> (2 + vshift));
 +    int x_off1 = x_off + (mv1->x >> (2 + hshift));
 +    int y_off1 = y_off + (mv1->y >> (2 + vshift));
 +    int idx = ff_hevc_pel_weight[block_w];
 +    src1  += y_off0 * src1stride + (int)((unsigned)x_off0 << s->sps->pixel_shift);
 +    src2  += y_off1 * src2stride + (int)((unsigned)x_off1 << s->sps->pixel_shift);
 +
 +    if (x_off0 < EPEL_EXTRA_BEFORE || y_off0 < EPEL_EXTRA_AFTER ||
 +        x_off0 >= pic_width - block_w - EPEL_EXTRA_AFTER ||
 +        y_off0 >= pic_height - block_h - EPEL_EXTRA_AFTER) {
 +        const int edge_emu_stride = EDGE_EMU_BUFFER_STRIDE << s->sps->pixel_shift;
          int offset1 = EPEL_EXTRA_BEFORE * (src1stride + (1 << s->sps->pixel_shift));
          int buf_offset1 = EPEL_EXTRA_BEFORE *
                            (edge_emu_stride + (1 << s->sps->pixel_shift));
 -        int offset2 = EPEL_EXTRA_BEFORE * (src2stride + (1 << s->sps->pixel_shift));
 -        int buf_offset2 = EPEL_EXTRA_BEFORE *
 -                          (edge_emu_stride + (1 << s->sps->pixel_shift));
  
          s->vdsp.emulated_edge_mc(lc->edge_emu_buffer, src1 - offset1,
                                   edge_emu_stride, src1stride,
                                   block_w + EPEL_EXTRA, block_h + EPEL_EXTRA,
 -                                 x_off - EPEL_EXTRA_BEFORE,
 -                                 y_off - EPEL_EXTRA_BEFORE,
 +                                 x_off0 - EPEL_EXTRA_BEFORE,
 +                                 y_off0 - EPEL_EXTRA_BEFORE,
                                   pic_width, pic_height);
  
          src1 = lc->edge_emu_buffer + buf_offset1;
          src1stride = edge_emu_stride;
 -        s->hevcdsp.put_hevc_epel[!!my][!!mx](dst1, dststride, src1, src1stride,
 -                                             block_w, block_h, mx, my, lc->mc_buffer);
 +    }
  
 -        s->vdsp.emulated_edge_mc(lc->edge_emu_buffer, src2 - offset2,
 +    if (x_off1 < EPEL_EXTRA_BEFORE || y_off1 < EPEL_EXTRA_AFTER ||
 +        x_off1 >= pic_width - block_w - EPEL_EXTRA_AFTER ||
 +        y_off1 >= pic_height - block_h - EPEL_EXTRA_AFTER) {
 +        const int edge_emu_stride = EDGE_EMU_BUFFER_STRIDE << s->sps->pixel_shift;
 +        int offset1 = EPEL_EXTRA_BEFORE * (src2stride + (1 << s->sps->pixel_shift));
 +        int buf_offset1 = EPEL_EXTRA_BEFORE *
 +                          (edge_emu_stride + (1 << s->sps->pixel_shift));
 +
 +        s->vdsp.emulated_edge_mc(lc->edge_emu_buffer2, src2 - offset1,
                                   edge_emu_stride, src2stride,
                                   block_w + EPEL_EXTRA, block_h + EPEL_EXTRA,
 -                                 x_off - EPEL_EXTRA_BEFORE,
 -                                 y_off - EPEL_EXTRA_BEFORE,
 +                                 x_off1 - EPEL_EXTRA_BEFORE,
 +                                 y_off1 - EPEL_EXTRA_BEFORE,
                                   pic_width, pic_height);
 -        src2 = lc->edge_emu_buffer + buf_offset2;
 -        src2stride = edge_emu_stride;
  
 -        s->hevcdsp.put_hevc_epel[!!my][!!mx](dst2, dststride, src2, src2stride,
 -                                             block_w, block_h, mx, my,
 -                                             lc->mc_buffer);
 -    } else {
 -        s->hevcdsp.put_hevc_epel[!!my][!!mx](dst1, dststride, src1, src1stride,
 -                                             block_w, block_h, mx, my,
 -                                             lc->mc_buffer);
 -        s->hevcdsp.put_hevc_epel[!!my][!!mx](dst2, dststride, src2, src2stride,
 -                                             block_w, block_h, mx, my,
 -                                             lc->mc_buffer);
 +        src2 = lc->edge_emu_buffer2 + buf_offset1;
 +        src2stride = edge_emu_stride;
      }
 +
 +    s->hevcdsp.put_hevc_epel[idx][!!my0][!!mx0](lc->tmp, src1, src1stride,
 +                                                block_h, _mx0, _my0, block_w);
 +    if (!weight_flag)
 +        s->hevcdsp.put_hevc_epel_bi[idx][!!my1][!!mx1](dst0, s->frame->linesize[cidx+1],
 +                                                       src2, src2stride, lc->tmp,
 +                                                       block_h, _mx1, _my1, block_w);
 +    else
 +        s->hevcdsp.put_hevc_epel_bi_w[idx][!!my1][!!mx1](dst0, s->frame->linesize[cidx+1],
 +                                                         src2, src2stride, lc->tmp,
 +                                                         block_h,
 +                                                         s->sh.chroma_log2_weight_denom,
 +                                                         s->sh.chroma_weight_l0[current_mv->ref_idx[0]][cidx],
 +                                                         s->sh.chroma_weight_l1[current_mv->ref_idx[1]][cidx],
 +                                                         s->sh.chroma_offset_l0[current_mv->ref_idx[0]][cidx],
 +                                                         s->sh.chroma_offset_l1[current_mv->ref_idx[1]][cidx],
 +                                                         _mx1, _my1, block_w);
  }
  
  static void hevc_await_progress(HEVCContext *s, HEVCFrame *ref,
                                  const Mv *mv, int y0, int height)
  {
 -    int y = (mv->y >> 2) + y0 + height + 9;
 -    ff_thread_await_progress(&ref->tf, y, 0);
 +    int y = FFMAX(0, (mv->y >> 2) + y0 + height + 9);
 +
 +    if (s->threads_type == FF_THREAD_FRAME )
 +        ff_thread_await_progress(&ref->tf, y, 0);
  }
  
 -static void hevc_luma_mv_mpv_mode(HEVCContext *s, int x0, int y0, int nPbW,
 +static void hevc_luma_mv_mvp_mode(HEVCContext *s, int x0, int y0, int nPbW,
                                    int nPbH, int log2_cb_size, int part_idx,
                                    int merge_idx, MvField *mv)
  {
 -    HEVCLocalContext *lc             = &s->HEVClc;
 +    HEVCLocalContext *lc s->HEVClc;
      enum InterPredIdc inter_pred_idc = PRED_L0;
      int mvp_flag;
  
      ff_hevc_set_neighbour_available(s, x0, y0, nPbW, nPbH);
 +    mv->pred_flag = 0;
      if (s->sh.slice_type == B_SLICE)
          inter_pred_idc = ff_hevc_inter_pred_idc_decode(s, nPbW, nPbH);
  
          if (s->sh.nb_refs[L0])
              mv->ref_idx[0]= ff_hevc_ref_idx_lx_decode(s, s->sh.nb_refs[L0]);
  
 -        mv->pred_flag[0] = 1;
 -        hls_mvd_coding(s, x0, y0, 0);
 +        mv->pred_flag = PF_L0;
 +        ff_hevc_hls_mvd_coding(s, x0, y0, 0);
          mvp_flag = ff_hevc_mvp_lx_flag_decode(s);
          ff_hevc_luma_mv_mvp_mode(s, x0, y0, nPbW, nPbH, log2_cb_size,
                                   part_idx, merge_idx, mv, mvp_flag, 0);
          if (s->sh.mvd_l1_zero_flag == 1 && inter_pred_idc == PRED_BI) {
              AV_ZERO32(&lc->pu.mvd);
          } else {
 -            hls_mvd_coding(s, x0, y0, 1);
 +            ff_hevc_hls_mvd_coding(s, x0, y0, 1);
          }
  
 -        mv->pred_flag[1] = 1;
 +        mv->pred_flag += PF_L1;
          mvp_flag = ff_hevc_mvp_lx_flag_decode(s);
          ff_hevc_luma_mv_mvp_mode(s, x0, y0, nPbW, nPbH, log2_cb_size,
                                   part_idx, merge_idx, mv, mvp_flag, 1);
  
  static void hls_prediction_unit(HEVCContext *s, int x0, int y0,
                                  int nPbW, int nPbH,
 -                                int log2_cb_size, int partIdx)
 +                                int log2_cb_size, int partIdx, int idx)
  {
  #define POS(c_idx, x, y)                                                              \
      &s->frame->data[c_idx][((y) >> s->sps->vshift[c_idx]) * s->frame->linesize[c_idx] + \
                             (((x) >> s->sps->hshift[c_idx]) << s->sps->pixel_shift)]
 -    HEVCLocalContext *lc = &s->HEVClc;
 +    HEVCLocalContext *lc = s->HEVClc;
      int merge_idx = 0;
      struct MvField current_mv = {{{ 0 }}};
  
  
      MvField *tab_mvf = s->ref->tab_mvf;
      RefPicList  *refPicList = s->ref->refPicList;
 -    HEVCFrame *ref0, *ref1;
 -
 -    int tmpstride = MAX_PB_SIZE;
 -
 +    HEVCFrame *ref0 = NULL, *ref1 = NULL;
      uint8_t *dst0 = POS(0, x0, y0);
      uint8_t *dst1 = POS(1, x0, y0);
      uint8_t *dst2 = POS(2, x0, y0);
          ff_hevc_luma_mv_merge_mode(s, x0, y0, nPbW, nPbH, log2_cb_size,
                                     partIdx, merge_idx, &current_mv);
      } else {
 -        hevc_luma_mv_mpv_mode(s, x0, y0, nPbW, nPbH, log2_cb_size,
 +        hevc_luma_mv_mvp_mode(s, x0, y0, nPbW, nPbH, log2_cb_size,
                                partIdx, merge_idx, &current_mv);
      }
  
          for (i = 0; i < nPbW >> s->sps->log2_min_pu_size; i++)
              tab_mvf[(y_pu + j) * min_pu_width + x_pu + i] = current_mv;
  
 -    if (current_mv.pred_flag[0]) {
 +    if (current_mv.pred_flag & PF_L0) {
          ref0 = refPicList[0].ref[current_mv.ref_idx[0]];
          if (!ref0)
              return;
          hevc_await_progress(s, ref0, &current_mv.mv[0], y0, nPbH);
      }
 -    if (current_mv.pred_flag[1]) {
 +    if (current_mv.pred_flag & PF_L1) {
          ref1 = refPicList[1].ref[current_mv.ref_idx[1]];
          if (!ref1)
              return;
          hevc_await_progress(s, ref1, &current_mv.mv[1], y0, nPbH);
      }
  
 -    if (current_mv.pred_flag[0] && !current_mv.pred_flag[1]) {
 -        DECLARE_ALIGNED(16, int16_t,  tmp[MAX_PB_SIZE * MAX_PB_SIZE]);
 -        DECLARE_ALIGNED(16, int16_t, tmp2[MAX_PB_SIZE * MAX_PB_SIZE]);
 +    if (current_mv.pred_flag == PF_L0) {
 +        int x0_c = x0 >> s->sps->hshift[1];
 +        int y0_c = y0 >> s->sps->vshift[1];
 +        int nPbW_c = nPbW >> s->sps->hshift[1];
 +        int nPbH_c = nPbH >> s->sps->vshift[1];
  
 -        luma_mc(s, tmp, tmpstride, ref0->frame,
 -                &current_mv.mv[0], x0, y0, nPbW, nPbH);
 +        luma_mc_uni(s, dst0, s->frame->linesize[0], ref0->frame,
 +                    &current_mv.mv[0], x0, y0, nPbW, nPbH,
 +                    s->sh.luma_weight_l0[current_mv.ref_idx[0]],
 +                    s->sh.luma_offset_l0[current_mv.ref_idx[0]]);
  
 -        if ((s->sh.slice_type == P_SLICE && s->pps->weighted_pred_flag) ||
 -            (s->sh.slice_type == B_SLICE && s->pps->weighted_bipred_flag)) {
 -            s->hevcdsp.weighted_pred(s->sh.luma_log2_weight_denom,
 -                                     s->sh.luma_weight_l0[current_mv.ref_idx[0]],
 -                                     s->sh.luma_offset_l0[current_mv.ref_idx[0]],
 -                                     dst0, s->frame->linesize[0], tmp,
 -                                     tmpstride, nPbW, nPbH);
 -        } else {
 -            s->hevcdsp.put_unweighted_pred(dst0, s->frame->linesize[0], tmp, tmpstride, nPbW, nPbH);
 -        }
 -        chroma_mc(s, tmp, tmp2, tmpstride, ref0->frame,
 -                  &current_mv.mv[0], x0 / 2, y0 / 2, nPbW / 2, nPbH / 2);
 -
 -        if ((s->sh.slice_type == P_SLICE && s->pps->weighted_pred_flag) ||
 -            (s->sh.slice_type == B_SLICE && s->pps->weighted_bipred_flag)) {
 -            s->hevcdsp.weighted_pred(s->sh.chroma_log2_weight_denom,
 -                                     s->sh.chroma_weight_l0[current_mv.ref_idx[0]][0],
 -                                     s->sh.chroma_offset_l0[current_mv.ref_idx[0]][0],
 -                                     dst1, s->frame->linesize[1], tmp, tmpstride,
 -                                     nPbW / 2, nPbH / 2);
 -            s->hevcdsp.weighted_pred(s->sh.chroma_log2_weight_denom,
 -                                     s->sh.chroma_weight_l0[current_mv.ref_idx[0]][1],
 -                                     s->sh.chroma_offset_l0[current_mv.ref_idx[0]][1],
 -                                     dst2, s->frame->linesize[2], tmp2, tmpstride,
 -                                     nPbW / 2, nPbH / 2);
 -        } else {
 -            s->hevcdsp.put_unweighted_pred(dst1, s->frame->linesize[1], tmp, tmpstride, nPbW/2, nPbH/2);
 -            s->hevcdsp.put_unweighted_pred(dst2, s->frame->linesize[2], tmp2, tmpstride, nPbW/2, nPbH/2);
 -        }
 -    } else if (!current_mv.pred_flag[0] && current_mv.pred_flag[1]) {
 -        DECLARE_ALIGNED(16, int16_t, tmp [MAX_PB_SIZE * MAX_PB_SIZE]);
 -        DECLARE_ALIGNED(16, int16_t, tmp2[MAX_PB_SIZE * MAX_PB_SIZE]);
 -
 -        luma_mc(s, tmp, tmpstride, ref1->frame,
 -                &current_mv.mv[1], x0, y0, nPbW, nPbH);
 -
 -        if ((s->sh.slice_type == P_SLICE && s->pps->weighted_pred_flag) ||
 -            (s->sh.slice_type == B_SLICE && s->pps->weighted_bipred_flag)) {
 -            s->hevcdsp.weighted_pred(s->sh.luma_log2_weight_denom,
 -                                      s->sh.luma_weight_l1[current_mv.ref_idx[1]],
 -                                      s->sh.luma_offset_l1[current_mv.ref_idx[1]],
 -                                      dst0, s->frame->linesize[0], tmp, tmpstride,
 -                                      nPbW, nPbH);
 -        } else {
 -            s->hevcdsp.put_unweighted_pred(dst0, s->frame->linesize[0], tmp, tmpstride, nPbW, nPbH);
 -        }
 -
 -        chroma_mc(s, tmp, tmp2, tmpstride, ref1->frame,
 -                  &current_mv.mv[1], x0/2, y0/2, nPbW/2, nPbH/2);
 -
 -        if ((s->sh.slice_type == P_SLICE && s->pps->weighted_pred_flag) ||
 -            (s->sh.slice_type == B_SLICE && s->pps->weighted_bipred_flag)) {
 -            s->hevcdsp.weighted_pred(s->sh.chroma_log2_weight_denom,
 -                                     s->sh.chroma_weight_l1[current_mv.ref_idx[1]][0],
 -                                     s->sh.chroma_offset_l1[current_mv.ref_idx[1]][0],
 -                                     dst1, s->frame->linesize[1], tmp, tmpstride, nPbW/2, nPbH/2);
 -            s->hevcdsp.weighted_pred(s->sh.chroma_log2_weight_denom,
 -                                     s->sh.chroma_weight_l1[current_mv.ref_idx[1]][1],
 -                                     s->sh.chroma_offset_l1[current_mv.ref_idx[1]][1],
 -                                     dst2, s->frame->linesize[2], tmp2, tmpstride, nPbW/2, nPbH/2);
 -        } else {
 -            s->hevcdsp.put_unweighted_pred(dst1, s->frame->linesize[1], tmp, tmpstride, nPbW/2, nPbH/2);
 -            s->hevcdsp.put_unweighted_pred(dst2, s->frame->linesize[2], tmp2, tmpstride, nPbW/2, nPbH/2);
 -        }
 -    } else if (current_mv.pred_flag[0] && current_mv.pred_flag[1]) {
 -        DECLARE_ALIGNED(16, int16_t, tmp [MAX_PB_SIZE * MAX_PB_SIZE]);
 -        DECLARE_ALIGNED(16, int16_t, tmp2[MAX_PB_SIZE * MAX_PB_SIZE]);
 -        DECLARE_ALIGNED(16, int16_t, tmp3[MAX_PB_SIZE * MAX_PB_SIZE]);
 -        DECLARE_ALIGNED(16, int16_t, tmp4[MAX_PB_SIZE * MAX_PB_SIZE]);
 -
 -        luma_mc(s, tmp, tmpstride, ref0->frame,
 -                &current_mv.mv[0], x0, y0, nPbW, nPbH);
 -        luma_mc(s, tmp2, tmpstride, ref1->frame,
 -                &current_mv.mv[1], x0, y0, nPbW, nPbH);
 -
 -        if ((s->sh.slice_type == P_SLICE && s->pps->weighted_pred_flag) ||
 -            (s->sh.slice_type == B_SLICE && s->pps->weighted_bipred_flag)) {
 -            s->hevcdsp.weighted_pred_avg(s->sh.luma_log2_weight_denom,
 -                                         s->sh.luma_weight_l0[current_mv.ref_idx[0]],
 -                                         s->sh.luma_weight_l1[current_mv.ref_idx[1]],
 -                                         s->sh.luma_offset_l0[current_mv.ref_idx[0]],
 -                                         s->sh.luma_offset_l1[current_mv.ref_idx[1]],
 -                                         dst0, s->frame->linesize[0],
 -                                         tmp, tmp2, tmpstride, nPbW, nPbH);
 -        } else {
 -            s->hevcdsp.put_weighted_pred_avg(dst0, s->frame->linesize[0],
 -                                             tmp, tmp2, tmpstride, nPbW, nPbH);
 -        }
 -
 -        chroma_mc(s, tmp, tmp2, tmpstride, ref0->frame,
 -                  &current_mv.mv[0], x0 / 2, y0 / 2, nPbW / 2, nPbH / 2);
 -        chroma_mc(s, tmp3, tmp4, tmpstride, ref1->frame,
 -                  &current_mv.mv[1], x0 / 2, y0 / 2, nPbW / 2, nPbH / 2);
 -
 -        if ((s->sh.slice_type == P_SLICE && s->pps->weighted_pred_flag) ||
 -            (s->sh.slice_type == B_SLICE && s->pps->weighted_bipred_flag)) {
 -            s->hevcdsp.weighted_pred_avg(s->sh.chroma_log2_weight_denom,
 -                                         s->sh.chroma_weight_l0[current_mv.ref_idx[0]][0],
 -                                         s->sh.chroma_weight_l1[current_mv.ref_idx[1]][0],
 -                                         s->sh.chroma_offset_l0[current_mv.ref_idx[0]][0],
 -                                         s->sh.chroma_offset_l1[current_mv.ref_idx[1]][0],
 -                                         dst1, s->frame->linesize[1], tmp, tmp3,
 -                                         tmpstride, nPbW / 2, nPbH / 2);
 -            s->hevcdsp.weighted_pred_avg(s->sh.chroma_log2_weight_denom,
 -                                         s->sh.chroma_weight_l0[current_mv.ref_idx[0]][1],
 -                                         s->sh.chroma_weight_l1[current_mv.ref_idx[1]][1],
 -                                         s->sh.chroma_offset_l0[current_mv.ref_idx[0]][1],
 -                                         s->sh.chroma_offset_l1[current_mv.ref_idx[1]][1],
 -                                         dst2, s->frame->linesize[2], tmp2, tmp4,
 -                                         tmpstride, nPbW / 2, nPbH / 2);
 -        } else {
 -            s->hevcdsp.put_weighted_pred_avg(dst1, s->frame->linesize[1], tmp, tmp3, tmpstride, nPbW/2, nPbH/2);
 -            s->hevcdsp.put_weighted_pred_avg(dst2, s->frame->linesize[2], tmp2, tmp4, tmpstride, nPbW/2, nPbH/2);
 +        if (s->sps->chroma_format_idc) {
 +            chroma_mc_uni(s, dst1, s->frame->linesize[1], ref0->frame->data[1], ref0->frame->linesize[1],
 +                          0, x0_c, y0_c, nPbW_c, nPbH_c, &current_mv,
 +                          s->sh.chroma_weight_l0[current_mv.ref_idx[0]][0], s->sh.chroma_offset_l0[current_mv.ref_idx[0]][0]);
 +            chroma_mc_uni(s, dst2, s->frame->linesize[2], ref0->frame->data[2], ref0->frame->linesize[2],
 +                          0, x0_c, y0_c, nPbW_c, nPbH_c, &current_mv,
 +                          s->sh.chroma_weight_l0[current_mv.ref_idx[0]][1], s->sh.chroma_offset_l0[current_mv.ref_idx[0]][1]);
 +        }
 +    } else if (current_mv.pred_flag == PF_L1) {
 +        int x0_c = x0 >> s->sps->hshift[1];
 +        int y0_c = y0 >> s->sps->vshift[1];
 +        int nPbW_c = nPbW >> s->sps->hshift[1];
 +        int nPbH_c = nPbH >> s->sps->vshift[1];
 +
 +        luma_mc_uni(s, dst0, s->frame->linesize[0], ref1->frame,
 +                    &current_mv.mv[1], x0, y0, nPbW, nPbH,
 +                    s->sh.luma_weight_l1[current_mv.ref_idx[1]],
 +                    s->sh.luma_offset_l1[current_mv.ref_idx[1]]);
 +
 +        if (s->sps->chroma_format_idc) {
 +            chroma_mc_uni(s, dst1, s->frame->linesize[1], ref1->frame->data[1], ref1->frame->linesize[1],
 +                          1, x0_c, y0_c, nPbW_c, nPbH_c, &current_mv,
 +                          s->sh.chroma_weight_l1[current_mv.ref_idx[1]][0], s->sh.chroma_offset_l1[current_mv.ref_idx[1]][0]);
 +
 +            chroma_mc_uni(s, dst2, s->frame->linesize[2], ref1->frame->data[2], ref1->frame->linesize[2],
 +                          1, x0_c, y0_c, nPbW_c, nPbH_c, &current_mv,
 +                          s->sh.chroma_weight_l1[current_mv.ref_idx[1]][1], s->sh.chroma_offset_l1[current_mv.ref_idx[1]][1]);
 +        }
 +    } else if (current_mv.pred_flag == PF_BI) {
 +        int x0_c = x0 >> s->sps->hshift[1];
 +        int y0_c = y0 >> s->sps->vshift[1];
 +        int nPbW_c = nPbW >> s->sps->hshift[1];
 +        int nPbH_c = nPbH >> s->sps->vshift[1];
 +
 +        luma_mc_bi(s, dst0, s->frame->linesize[0], ref0->frame,
 +                   &current_mv.mv[0], x0, y0, nPbW, nPbH,
 +                   ref1->frame, &current_mv.mv[1], &current_mv);
 +
 +        if (s->sps->chroma_format_idc) {
 +            chroma_mc_bi(s, dst1, s->frame->linesize[1], ref0->frame, ref1->frame,
 +                         x0_c, y0_c, nPbW_c, nPbH_c, &current_mv, 0);
 +
 +            chroma_mc_bi(s, dst2, s->frame->linesize[2], ref0->frame, ref1->frame,
 +                         x0_c, y0_c, nPbW_c, nPbH_c, &current_mv, 1);
          }
      }
  }
  static int luma_intra_pred_mode(HEVCContext *s, int x0, int y0, int pu_size,
                                  int prev_intra_luma_pred_flag)
  {
 -    HEVCLocalContext *lc = &s->HEVClc;
 +    HEVCLocalContext *lc = s->HEVClc;
      int x_pu             = x0 >> s->sps->log2_min_pu_size;
      int y_pu             = y0 >> s->sps->log2_min_pu_size;
      int min_pu_width     = s->sps->min_pu_width;
      int size_in_pus      = pu_size >> s->sps->log2_min_pu_size;
 -    int x0b              = x0 & ((1 << s->sps->log2_ctb_size) - 1);
 -    int y0b              = y0 & ((1 << s->sps->log2_ctb_size) - 1);
 +    int x0b              = av_mod_uintp2(x0, s->sps->log2_ctb_size);
 +    int y0b              = av_mod_uintp2(y0, s->sps->log2_ctb_size);
  
      int cand_up   = (lc->ctb_up_flag || y0b) ?
                      s->tab_ipm[(y_pu - 1) * min_pu_width + x_pu] : INTRA_DC;
                 intra_pred_mode, size_in_pus);
  
          for (j = 0; j < size_in_pus; j++) {
 -            tab_mvf[(y_pu + j) * min_pu_width + x_pu + i].is_intra     = 1;
 -            tab_mvf[(y_pu + j) * min_pu_width + x_pu + i].pred_flag[0] = 0;
 -            tab_mvf[(y_pu + j) * min_pu_width + x_pu + i].pred_flag[1] = 0;
 -            tab_mvf[(y_pu + j) * min_pu_width + x_pu + i].ref_idx[0]   = 0;
 -            tab_mvf[(y_pu + j) * min_pu_width + x_pu + i].ref_idx[1]   = 0;
 -            tab_mvf[(y_pu + j) * min_pu_width + x_pu + i].mv[0].x      = 0;
 -            tab_mvf[(y_pu + j) * min_pu_width + x_pu + i].mv[0].y      = 0;
 -            tab_mvf[(y_pu + j) * min_pu_width + x_pu + i].mv[1].x      = 0;
 -            tab_mvf[(y_pu + j) * min_pu_width + x_pu + i].mv[1].y      = 0;
 +            tab_mvf[(y_pu + j) * min_pu_width + x_pu + i].pred_flag = PF_INTRA;
          }
      }
  
@@@ -1888,14 -1988,10 +1888,14 @@@ static av_always_inline void set_ct_dep
                 ct_depth, length);
  }
  
 +static const uint8_t tab_mode_idx[] = {
 +     0,  1,  2,  2,  2,  2,  3,  5,  7,  8, 10, 12, 13, 15, 17, 18, 19, 20,
 +    21, 22, 23, 23, 24, 24, 25, 25, 26, 27, 27, 28, 28, 29, 29, 30, 31};
 +
  static void intra_prediction_unit(HEVCContext *s, int x0, int y0,
                                    int log2_cb_size)
  {
 -    HEVCLocalContext *lc = &s->HEVClc;
 +    HEVCLocalContext *lc = s->HEVClc;
      static const uint8_t intra_chroma_table[4] = { 0, 26, 10, 1 };
      uint8_t prev_intra_luma_pred_flag[4];
      int split   = lc->cu.part_mode == PART_NxN;
          }
      }
  
 -    chroma_mode = ff_hevc_intra_chroma_pred_mode_decode(s);
 -    if (chroma_mode != 4) {
 -        if (lc->pu.intra_pred_mode[0] == intra_chroma_table[chroma_mode])
 -            lc->pu.intra_pred_mode_c = 34;
 -        else
 -            lc->pu.intra_pred_mode_c = intra_chroma_table[chroma_mode];
 -    } else {
 -        lc->pu.intra_pred_mode_c = lc->pu.intra_pred_mode[0];
 +    if (s->sps->chroma_format_idc == 3) {
 +        for (i = 0; i < side; i++) {
 +            for (j = 0; j < side; j++) {
 +                lc->pu.chroma_mode_c[2 * i + j] = chroma_mode = ff_hevc_intra_chroma_pred_mode_decode(s);
 +                if (chroma_mode != 4) {
 +                    if (lc->pu.intra_pred_mode[2 * i + j] == intra_chroma_table[chroma_mode])
 +                        lc->pu.intra_pred_mode_c[2 * i + j] = 34;
 +                    else
 +                        lc->pu.intra_pred_mode_c[2 * i + j] = intra_chroma_table[chroma_mode];
 +                } else {
 +                    lc->pu.intra_pred_mode_c[2 * i + j] = lc->pu.intra_pred_mode[2 * i + j];
 +                }
 +            }
 +        }
 +    } else if (s->sps->chroma_format_idc == 2) {
 +        int mode_idx;
 +        lc->pu.chroma_mode_c[0] = chroma_mode = ff_hevc_intra_chroma_pred_mode_decode(s);
 +        if (chroma_mode != 4) {
 +            if (lc->pu.intra_pred_mode[0] == intra_chroma_table[chroma_mode])
 +                mode_idx = 34;
 +            else
 +                mode_idx = intra_chroma_table[chroma_mode];
 +        } else {
 +            mode_idx = lc->pu.intra_pred_mode[0];
 +        }
 +        lc->pu.intra_pred_mode_c[0] = tab_mode_idx[mode_idx];
 +    } else if (s->sps->chroma_format_idc != 0) {
 +        chroma_mode = ff_hevc_intra_chroma_pred_mode_decode(s);
 +        if (chroma_mode != 4) {
 +            if (lc->pu.intra_pred_mode[0] == intra_chroma_table[chroma_mode])
 +                lc->pu.intra_pred_mode_c[0] = 34;
 +            else
 +                lc->pu.intra_pred_mode_c[0] = intra_chroma_table[chroma_mode];
 +        } else {
 +            lc->pu.intra_pred_mode_c[0] = lc->pu.intra_pred_mode[0];
 +        }
      }
  }
  
@@@ -1964,7 -2032,7 +1964,7 @@@ static void intra_prediction_unit_defau
                                                  int x0, int y0,
                                                  int log2_cb_size)
  {
 -    HEVCLocalContext *lc = &s->HEVClc;
 +    HEVCLocalContext *lc = s->HEVClc;
      int pb_size          = 1 << log2_cb_size;
      int size_in_pus      = pb_size >> s->sps->log2_min_pu_size;
      int min_pu_width     = s->sps->min_pu_width;
  
      if (size_in_pus == 0)
          size_in_pus = 1;
 -    for (j = 0; j < size_in_pus; j++) {
 +    for (j = 0; j < size_in_pus; j++)
          memset(&s->tab_ipm[(y_pu + j) * min_pu_width + x_pu], INTRA_DC, size_in_pus);
 -        for (k = 0; k < size_in_pus; k++)
 -            tab_mvf[(y_pu + j) * min_pu_width + x_pu + k].is_intra = lc->cu.pred_mode == MODE_INTRA;
 -    }
 +    if (lc->cu.pred_mode == MODE_INTRA)
 +        for (j = 0; j < size_in_pus; j++)
 +            for (k = 0; k < size_in_pus; k++)
 +                tab_mvf[(y_pu + j) * min_pu_width + x_pu + k].pred_flag = PF_INTRA;
  }
  
  static int hls_coding_unit(HEVCContext *s, int x0, int y0, int log2_cb_size)
  {
      int cb_size          = 1 << log2_cb_size;
 -    HEVCLocalContext *lc = &s->HEVClc;
 +    HEVCLocalContext *lc = s->HEVClc;
      int log2_min_cb_size = s->sps->log2_min_cb_size;
      int length           = cb_size >> log2_min_cb_size;
      int min_cb_width     = s->sps->min_cb_width;
      int x_cb             = x0 >> log2_min_cb_size;
      int y_cb             = y0 >> log2_min_cb_size;
 +    int idx              = log2_cb_size - 2;
 +    int qp_block_mask    = (1<<(s->sps->log2_ctb_size - s->pps->diff_cu_qp_delta_depth)) - 1;
      int x, y, ret;
  
      lc->cu.x                = x0;
              x += min_cb_width;
          }
          lc->cu.pred_mode = skip_flag ? MODE_SKIP : MODE_INTER;
 +    } else {
 +        x = y_cb * min_cb_width + x_cb;
 +        for (y = 0; y < length; y++) {
 +            memset(&s->skip_flag[x], 0, length);
 +            x += min_cb_width;
 +        }
      }
  
      if (SAMPLE_CTB(s->skip_flag, x_cb, y_cb)) {
 -        hls_prediction_unit(s, x0, y0, cb_size, cb_size, log2_cb_size, 0);
 +        hls_prediction_unit(s, x0, y0, cb_size, cb_size, log2_cb_size, 0, idx);
          intra_prediction_unit_default_value(s, x0, y0, log2_cb_size);
  
          if (!s->sh.disable_deblocking_filter_flag)
              intra_prediction_unit_default_value(s, x0, y0, log2_cb_size);
              switch (lc->cu.part_mode) {
              case PART_2Nx2N:
 -                hls_prediction_unit(s, x0, y0, cb_size, cb_size, log2_cb_size, 0);
 +                hls_prediction_unit(s, x0, y0, cb_size, cb_size, log2_cb_size, 0, idx);
                  break;
              case PART_2NxN:
 -                hls_prediction_unit(s, x0, y0,               cb_size, cb_size / 2, log2_cb_size, 0);
 -                hls_prediction_unit(s, x0, y0 + cb_size / 2, cb_size, cb_size / 2, log2_cb_size, 1);
 +                hls_prediction_unit(s, x0, y0,               cb_size, cb_size / 2, log2_cb_size, 0, idx);
 +                hls_prediction_unit(s, x0, y0 + cb_size / 2, cb_size, cb_size / 2, log2_cb_size, 1, idx);
                  break;
              case PART_Nx2N:
 -                hls_prediction_unit(s, x0,               y0, cb_size / 2, cb_size, log2_cb_size, 0);
 -                hls_prediction_unit(s, x0 + cb_size / 2, y0, cb_size / 2, cb_size, log2_cb_size, 1);
 +                hls_prediction_unit(s, x0,               y0, cb_size / 2, cb_size, log2_cb_size, 0, idx - 1);
 +                hls_prediction_unit(s, x0 + cb_size / 2, y0, cb_size / 2, cb_size, log2_cb_size, 1, idx - 1);
                  break;
              case PART_2NxnU:
 -                hls_prediction_unit(s, x0, y0,               cb_size, cb_size     / 4, log2_cb_size, 0);
 -                hls_prediction_unit(s, x0, y0 + cb_size / 4, cb_size, cb_size * 3 / 4, log2_cb_size, 1);
 +                hls_prediction_unit(s, x0, y0,               cb_size, cb_size     / 4, log2_cb_size, 0, idx);
 +                hls_prediction_unit(s, x0, y0 + cb_size / 4, cb_size, cb_size * 3 / 4, log2_cb_size, 1, idx);
                  break;
              case PART_2NxnD:
 -                hls_prediction_unit(s, x0, y0,                   cb_size, cb_size * 3 / 4, log2_cb_size, 0);
 -                hls_prediction_unit(s, x0, y0 + cb_size * 3 / 4, cb_size, cb_size     / 4, log2_cb_size, 1);
 +                hls_prediction_unit(s, x0, y0,                   cb_size, cb_size * 3 / 4, log2_cb_size, 0, idx);
 +                hls_prediction_unit(s, x0, y0 + cb_size * 3 / 4, cb_size, cb_size     / 4, log2_cb_size, 1, idx);
                  break;
              case PART_nLx2N:
 -                hls_prediction_unit(s, x0,               y0, cb_size     / 4, cb_size, log2_cb_size, 0);
 -                hls_prediction_unit(s, x0 + cb_size / 4, y0, cb_size * 3 / 4, cb_size, log2_cb_size, 1);
 +                hls_prediction_unit(s, x0,               y0, cb_size     / 4, cb_size, log2_cb_size, 0, idx - 2);
 +                hls_prediction_unit(s, x0 + cb_size / 4, y0, cb_size * 3 / 4, cb_size, log2_cb_size, 1, idx - 2);
                  break;
              case PART_nRx2N:
 -                hls_prediction_unit(s, x0,                   y0, cb_size * 3 / 4, cb_size, log2_cb_size, 0);
 -                hls_prediction_unit(s, x0 + cb_size * 3 / 4, y0, cb_size     / 4, cb_size, log2_cb_size, 1);
 +                hls_prediction_unit(s, x0,                   y0, cb_size * 3 / 4, cb_size, log2_cb_size, 0, idx - 2);
 +                hls_prediction_unit(s, x0 + cb_size * 3 / 4, y0, cb_size     / 4, cb_size, log2_cb_size, 1, idx - 2);
                  break;
              case PART_NxN:
 -                hls_prediction_unit(s, x0,               y0,               cb_size / 2, cb_size / 2, log2_cb_size, 0);
 -                hls_prediction_unit(s, x0 + cb_size / 2, y0,               cb_size / 2, cb_size / 2, log2_cb_size, 1);
 -                hls_prediction_unit(s, x0,               y0 + cb_size / 2, cb_size / 2, cb_size / 2, log2_cb_size, 2);
 -                hls_prediction_unit(s, x0 + cb_size / 2, y0 + cb_size / 2, cb_size / 2, cb_size / 2, log2_cb_size, 3);
 +                hls_prediction_unit(s, x0,               y0,               cb_size / 2, cb_size / 2, log2_cb_size, 0, idx - 1);
 +                hls_prediction_unit(s, x0 + cb_size / 2, y0,               cb_size / 2, cb_size / 2, log2_cb_size, 1, idx - 1);
 +                hls_prediction_unit(s, x0,               y0 + cb_size / 2, cb_size / 2, cb_size / 2, log2_cb_size, 2, idx - 1);
 +                hls_prediction_unit(s, x0 + cb_size / 2, y0 + cb_size / 2, cb_size / 2, cb_size / 2, log2_cb_size, 3, idx - 1);
                  break;
              }
          }
                  rqt_root_cbf = ff_hevc_no_residual_syntax_flag_decode(s);
              }
              if (rqt_root_cbf) {
 +                const static int cbf[2] = { 0 };
                  lc->cu.max_trafo_depth = lc->cu.pred_mode == MODE_INTRA ?
                                           s->sps->max_transform_hierarchy_depth_intra + lc->cu.intra_split_flag :
                                           s->sps->max_transform_hierarchy_depth_inter;
                  ret = hls_transform_tree(s, x0, y0, x0, y0, x0, y0,
                                           log2_cb_size,
 -                                         log2_cb_size, 0, 0, 0, 0);
 +                                         log2_cb_size, 0, 0, cbf, cbf);
                  if (ret < 0)
                      return ret;
              } else {
      }
  
      if (s->pps->cu_qp_delta_enabled_flag && lc->tu.is_cu_qp_delta_coded == 0)
 -        ff_hevc_set_qPy(s, x0, y0, x0, y0, log2_cb_size);
 +        ff_hevc_set_qPy(s, x0, y0, log2_cb_size);
  
      x = y_cb * min_cb_width + x_cb;
      for (y = 0; y < length; y++) {
          x += min_cb_width;
      }
  
 -    set_ct_depth(s, x0, y0, log2_cb_size, lc->ct.depth);
 +    if(((x0 + (1<<log2_cb_size)) & qp_block_mask) == 0 &&
 +       ((y0 + (1<<log2_cb_size)) & qp_block_mask) == 0) {
 +        lc->qPy_pred = lc->qp_y;
 +    }
 +
 +    set_ct_depth(s, x0, y0, log2_cb_size, lc->ct_depth);
  
      return 0;
  }
  static int hls_coding_quadtree(HEVCContext *s, int x0, int y0,
                                 int log2_cb_size, int cb_depth)
  {
 -    HEVCLocalContext *lc = &s->HEVClc;
 +    HEVCLocalContext *lc = s->HEVClc;
      const int cb_size    = 1 << log2_cb_size;
 +    int ret;
      int split_cu;
  
 -    lc->ct.depth = cb_depth;
 +    lc->ct_depth = cb_depth;
      if (x0 + cb_size <= s->sps->width  &&
          y0 + cb_size <= s->sps->height &&
          log2_cb_size > s->sps->log2_min_cb_size) {
          lc->tu.cu_qp_delta          = 0;
      }
  
 +    if (s->sh.cu_chroma_qp_offset_enabled_flag &&
 +        log2_cb_size >= s->sps->log2_ctb_size - s->pps->diff_cu_chroma_qp_offset_depth) {
 +        lc->tu.is_cu_chroma_qp_offset_coded = 0;
 +    }
 +
      if (split_cu) {
 +        int qp_block_mask = (1<<(s->sps->log2_ctb_size - s->pps->diff_cu_qp_delta_depth)) - 1;
          const int cb_size_split = cb_size >> 1;
          const int x1 = x0 + cb_size_split;
          const int y1 = y0 + cb_size_split;
  
 -        log2_cb_size--;
 -        cb_depth++;
 +        int more_data = 0;
  
 -#define SUBDIVIDE(x, y)                                                \
 -do {                                                                   \
 -    if (x < s->sps->width && y < s->sps->height) {                     \
 -        int ret = hls_coding_quadtree(s, x, y, log2_cb_size, cb_depth);\
 -        if (ret < 0)                                                   \
 -            return ret;                                                \
 -    }                                                                  \
 -} while (0)
 +        more_data = hls_coding_quadtree(s, x0, y0, log2_cb_size - 1, cb_depth + 1);
 +        if (more_data < 0)
 +            return more_data;
  
 -        SUBDIVIDE(x0, y0);
 -        SUBDIVIDE(x1, y0);
 -        SUBDIVIDE(x0, y1);
 -        SUBDIVIDE(x1, y1);
 +        if (more_data && x1 < s->sps->width) {
 +            more_data = hls_coding_quadtree(s, x1, y0, log2_cb_size - 1, cb_depth + 1);
 +            if (more_data < 0)
 +                return more_data;
 +        }
 +        if (more_data && y1 < s->sps->height) {
 +            more_data = hls_coding_quadtree(s, x0, y1, log2_cb_size - 1, cb_depth + 1);
 +            if (more_data < 0)
 +                return more_data;
 +        }
 +        if (more_data && x1 < s->sps->width &&
 +            y1 < s->sps->height) {
 +            more_data = hls_coding_quadtree(s, x1, y1, log2_cb_size - 1, cb_depth + 1);
 +            if (more_data < 0)
 +                return more_data;
 +        }
 +
 +        if(((x0 + (1<<log2_cb_size)) & qp_block_mask) == 0 &&
 +            ((y0 + (1<<log2_cb_size)) & qp_block_mask) == 0)
 +            lc->qPy_pred = lc->qp_y;
 +
 +        if (more_data)
 +            return ((x1 + cb_size_split) < s->sps->width ||
 +                    (y1 + cb_size_split) < s->sps->height);
 +        else
 +            return 0;
      } else {
 -        int ret = hls_coding_unit(s, x0, y0, log2_cb_size);
 +        ret = hls_coding_unit(s, x0, y0, log2_cb_size);
          if (ret < 0)
              return ret;
 +        if ((!((x0 + cb_size) %
 +               (1 << (s->sps->log2_ctb_size))) ||
 +             (x0 + cb_size >= s->sps->width)) &&
 +            (!((y0 + cb_size) %
 +               (1 << (s->sps->log2_ctb_size))) ||
 +             (y0 + cb_size >= s->sps->height))) {
 +            int end_of_slice_flag = ff_hevc_end_of_slice_flag_decode(s);
 +            return !end_of_slice_flag;
 +        } else {
 +            return 1;
 +        }
      }
  
      return 0;
  static void hls_decode_neighbour(HEVCContext *s, int x_ctb, int y_ctb,
                                   int ctb_addr_ts)
  {
 -    HEVCLocalContext *lc  = &s->HEVClc;
 +    HEVCLocalContext *lc  = s->HEVClc;
      int ctb_size          = 1 << s->sps->log2_ctb_size;
      int ctb_addr_rs       = s->pps->ctb_addr_ts_to_rs[ctb_addr_ts];
      int ctb_addr_in_slice = ctb_addr_rs - s->sh.slice_addr;
      } else if (s->pps->tiles_enabled_flag) {
          if (ctb_addr_ts && s->pps->tile_id[ctb_addr_ts] != s->pps->tile_id[ctb_addr_ts - 1]) {
              int idxX = s->pps->col_idxX[x_ctb >> s->sps->log2_ctb_size];
 -            lc->start_of_tiles_x = x_ctb;
              lc->end_of_tiles_x   = x_ctb + (s->pps->column_width[idxX] << s->sps->log2_ctb_size);
              lc->first_qp_group   = 1;
          }
          if (y_ctb > 0 && s->tab_slice_address[ctb_addr_rs] != s->tab_slice_address[ctb_addr_rs - s->sps->ctb_width])
              lc->boundary_flags |= BOUNDARY_UPPER_SLICE;
      } else {
 -        if (!ctb_addr_in_slice > 0)
 +        if (ctb_addr_in_slice <= 0)
              lc->boundary_flags |= BOUNDARY_LEFT_SLICE;
          if (ctb_addr_in_slice < s->sps->ctb_width)
              lc->boundary_flags |= BOUNDARY_UPPER_SLICE;
      lc->ctb_up_left_flag = ((x_ctb > 0) && (y_ctb > 0)  && (ctb_addr_in_slice-1 >= s->sps->ctb_width) && (s->pps->tile_id[ctb_addr_ts] == s->pps->tile_id[s->pps->ctb_addr_rs_to_ts[ctb_addr_rs-1 - s->sps->ctb_width]]));
  }
  
 -static int hls_slice_data(HEVCContext *s)
 +static int hls_decode_entry(AVCodecContext *avctxt, void *isFilterThread)
  {
 +    HEVCContext *s  = avctxt->priv_data;
      int ctb_size    = 1 << s->sps->log2_ctb_size;
      int more_data   = 1;
      int x_ctb       = 0;
      int y_ctb       = 0;
      int ctb_addr_ts = s->pps->ctb_addr_rs_to_ts[s->sh.slice_ctb_addr_rs];
 -    int ret;
 +
 +    if (!ctb_addr_ts && s->sh.dependent_slice_segment_flag) {
 +        av_log(s->avctx, AV_LOG_ERROR, "Impossible initial tile.\n");
 +        return AVERROR_INVALIDDATA;
 +    }
 +
 +    if (s->sh.dependent_slice_segment_flag) {
 +        int prev_rs = s->pps->ctb_addr_ts_to_rs[ctb_addr_ts - 1];
 +        if (s->tab_slice_address[prev_rs] != s->sh.slice_addr) {
 +            av_log(s->avctx, AV_LOG_ERROR, "Previous slice segment missing\n");
 +            return AVERROR_INVALIDDATA;
 +        }
 +    }
  
      while (more_data && ctb_addr_ts < s->sps->ctb_size) {
          int ctb_addr_rs = s->pps->ctb_addr_ts_to_rs[ctb_addr_ts];
          s->deblock[ctb_addr_rs].tc_offset   = s->sh.tc_offset;
          s->filter_slice_edges[ctb_addr_rs]  = s->sh.slice_loop_filter_across_slices_enabled_flag;
  
 -        ret = hls_coding_quadtree(s, x_ctb, y_ctb, s->sps->log2_ctb_size, 0);
 -        if (ret < 0)
 -            return ret;
 -        more_data = !ff_hevc_end_of_slice_flag_decode(s);
 +        more_data = hls_coding_quadtree(s, x_ctb, y_ctb, s->sps->log2_ctb_size, 0);
 +        if (more_data < 0) {
 +            s->tab_slice_address[ctb_addr_rs] = -1;
 +            return more_data;
 +        }
 +
  
          ctb_addr_ts++;
          ff_hevc_save_states(s, ctb_addr_ts);
  
      if (x_ctb + ctb_size >= s->sps->width &&
          y_ctb + ctb_size >= s->sps->height)
 -        ff_hevc_hls_filter(s, x_ctb, y_ctb);
 +        ff_hevc_hls_filter(s, x_ctb, y_ctb, ctb_size);
  
      return ctb_addr_ts;
  }
  
 +static int hls_slice_data(HEVCContext *s)
 +{
 +    int arg[2];
 +    int ret[2];
 +
 +    arg[0] = 0;
 +    arg[1] = 1;
 +
 +    s->avctx->execute(s->avctx, hls_decode_entry, arg, ret , 1, sizeof(int));
 +    return ret[0];
 +}
 +static int hls_decode_entry_wpp(AVCodecContext *avctxt, void *input_ctb_row, int job, int self_id)
 +{
 +    HEVCContext *s1  = avctxt->priv_data, *s;
 +    HEVCLocalContext *lc;
 +    int ctb_size    = 1<< s1->sps->log2_ctb_size;
 +    int more_data   = 1;
 +    int *ctb_row_p    = input_ctb_row;
 +    int ctb_row = ctb_row_p[job];
 +    int ctb_addr_rs = s1->sh.slice_ctb_addr_rs + ctb_row * ((s1->sps->width + ctb_size - 1) >> s1->sps->log2_ctb_size);
 +    int ctb_addr_ts = s1->pps->ctb_addr_rs_to_ts[ctb_addr_rs];
 +    int thread = ctb_row % s1->threads_number;
 +    int ret;
 +
 +    s = s1->sList[self_id];
 +    lc = s->HEVClc;
 +
 +    if(ctb_row) {
 +        ret = init_get_bits8(&lc->gb, s->data + s->sh.offset[ctb_row - 1], s->sh.size[ctb_row - 1]);
 +
 +        if (ret < 0)
 +            return ret;
 +        ff_init_cabac_decoder(&lc->cc, s->data + s->sh.offset[(ctb_row)-1], s->sh.size[ctb_row - 1]);
 +    }
 +
 +    while(more_data && ctb_addr_ts < s->sps->ctb_size) {
 +        int x_ctb = (ctb_addr_rs % s->sps->ctb_width) << s->sps->log2_ctb_size;
 +        int y_ctb = (ctb_addr_rs / s->sps->ctb_width) << s->sps->log2_ctb_size;
 +
 +        hls_decode_neighbour(s, x_ctb, y_ctb, ctb_addr_ts);
 +
 +        ff_thread_await_progress2(s->avctx, ctb_row, thread, SHIFT_CTB_WPP);
 +
 +        if (avpriv_atomic_int_get(&s1->wpp_err)){
 +            ff_thread_report_progress2(s->avctx, ctb_row , thread, SHIFT_CTB_WPP);
 +            return 0;
 +        }
 +
 +        ff_hevc_cabac_init(s, ctb_addr_ts);
 +        hls_sao_param(s, x_ctb >> s->sps->log2_ctb_size, y_ctb >> s->sps->log2_ctb_size);
 +        more_data = hls_coding_quadtree(s, x_ctb, y_ctb, s->sps->log2_ctb_size, 0);
 +
 +        if (more_data < 0) {
 +            s->tab_slice_address[ctb_addr_rs] = -1;
 +            return more_data;
 +        }
 +
 +        ctb_addr_ts++;
 +
 +        ff_hevc_save_states(s, ctb_addr_ts);
 +        ff_thread_report_progress2(s->avctx, ctb_row, thread, 1);
 +        ff_hevc_hls_filters(s, x_ctb, y_ctb, ctb_size);
 +
 +        if (!more_data && (x_ctb+ctb_size) < s->sps->width && ctb_row != s->sh.num_entry_point_offsets) {
 +            avpriv_atomic_int_set(&s1->wpp_err,  1);
 +            ff_thread_report_progress2(s->avctx, ctb_row ,thread, SHIFT_CTB_WPP);
 +            return 0;
 +        }
 +
 +        if ((x_ctb+ctb_size) >= s->sps->width && (y_ctb+ctb_size) >= s->sps->height ) {
 +            ff_hevc_hls_filter(s, x_ctb, y_ctb, ctb_size);
 +            ff_thread_report_progress2(s->avctx, ctb_row , thread, SHIFT_CTB_WPP);
 +            return ctb_addr_ts;
 +        }
 +        ctb_addr_rs       = s->pps->ctb_addr_ts_to_rs[ctb_addr_ts];
 +        x_ctb+=ctb_size;
 +
 +        if(x_ctb >= s->sps->width) {
 +            break;
 +        }
 +    }
 +    ff_thread_report_progress2(s->avctx, ctb_row ,thread, SHIFT_CTB_WPP);
 +
 +    return 0;
 +}
 +
 +static int hls_slice_data_wpp(HEVCContext *s, const uint8_t *nal, int length)
 +{
 +    HEVCLocalContext *lc = s->HEVClc;
 +    int *ret = av_malloc_array(s->sh.num_entry_point_offsets + 1, sizeof(int));
 +    int *arg = av_malloc_array(s->sh.num_entry_point_offsets + 1, sizeof(int));
 +    int offset;
 +    int startheader, cmpt = 0;
 +    int i, j, res = 0;
 +
 +    if (!ret || !arg) {
 +        av_free(ret);
 +        av_free(arg);
 +        return AVERROR(ENOMEM);
 +    }
 +
 +
 +    if (!s->sList[1]) {
 +        ff_alloc_entries(s->avctx, s->sh.num_entry_point_offsets + 1);
 +
 +
 +        for (i = 1; i < s->threads_number; i++) {
 +            s->sList[i] = av_malloc(sizeof(HEVCContext));
 +            memcpy(s->sList[i], s, sizeof(HEVCContext));
 +            s->HEVClcList[i] = av_mallocz(sizeof(HEVCLocalContext));
 +            s->sList[i]->HEVClc = s->HEVClcList[i];
 +        }
 +    }
 +
 +    offset = (lc->gb.index >> 3);
 +
 +    for (j = 0, cmpt = 0, startheader = offset + s->sh.entry_point_offset[0]; j < s->skipped_bytes; j++) {
 +        if (s->skipped_bytes_pos[j] >= offset && s->skipped_bytes_pos[j] < startheader) {
 +            startheader--;
 +            cmpt++;
 +        }
 +    }
 +
 +    for (i = 1; i < s->sh.num_entry_point_offsets; i++) {
 +        offset += (s->sh.entry_point_offset[i - 1] - cmpt);
 +        for (j = 0, cmpt = 0, startheader = offset
 +             + s->sh.entry_point_offset[i]; j < s->skipped_bytes; j++) {
 +            if (s->skipped_bytes_pos[j] >= offset && s->skipped_bytes_pos[j] < startheader) {
 +                startheader--;
 +                cmpt++;
 +            }
 +        }
 +        s->sh.size[i - 1] = s->sh.entry_point_offset[i] - cmpt;
 +        s->sh.offset[i - 1] = offset;
 +
 +    }
 +    if (s->sh.num_entry_point_offsets != 0) {
 +        offset += s->sh.entry_point_offset[s->sh.num_entry_point_offsets - 1] - cmpt;
 +        s->sh.size[s->sh.num_entry_point_offsets - 1] = length - offset;
 +        s->sh.offset[s->sh.num_entry_point_offsets - 1] = offset;
 +
 +    }
 +    s->data = nal;
 +
 +    for (i = 1; i < s->threads_number; i++) {
 +        s->sList[i]->HEVClc->first_qp_group = 1;
 +        s->sList[i]->HEVClc->qp_y = s->sList[0]->HEVClc->qp_y;
 +        memcpy(s->sList[i], s, sizeof(HEVCContext));
 +        s->sList[i]->HEVClc = s->HEVClcList[i];
 +    }
 +
 +    avpriv_atomic_int_set(&s->wpp_err, 0);
 +    ff_reset_entries(s->avctx);
 +
 +    for (i = 0; i <= s->sh.num_entry_point_offsets; i++) {
 +        arg[i] = i;
 +        ret[i] = 0;
 +    }
 +
 +    if (s->pps->entropy_coding_sync_enabled_flag)
 +        s->avctx->execute2(s->avctx, (void *) hls_decode_entry_wpp, arg, ret, s->sh.num_entry_point_offsets + 1);
 +
 +    for (i = 0; i <= s->sh.num_entry_point_offsets; i++)
 +        res += ret[i];
 +    av_free(ret);
 +    av_free(arg);
 +    return res;
 +}
 +
  /**
   * @return AVERROR_INVALIDDATA if the packet is not a valid NAL unit,
   * 0 if the unit should be skipped, 1 otherwise
   */
  static int hls_nal_unit(HEVCContext *s)
  {
 -    GetBitContext *gb = &s->HEVClc.gb;
 +    GetBitContext *gb = &s->HEVClc->gb;
      int nuh_layer_id;
  
      if (get_bits1(gb) != 0)
          return AVERROR_INVALIDDATA;
  
      av_log(s->avctx, AV_LOG_DEBUG,
 -           "nal_unit_type: %d, nuh_layer_id: %dtemporal_id: %d\n",
 +           "nal_unit_type: %d, nuh_layer_id: %dtemporal_id: %d\n",
             s->nal_unit_type, nuh_layer_id, s->temporal_id);
  
      return nuh_layer_id == 0;
  }
  
 -static void restore_tqb_pixels(HEVCContext *s)
 -{
 -    int min_pu_size = 1 << s->sps->log2_min_pu_size;
 -    int x, y, c_idx;
 -
 -    for (c_idx = 0; c_idx < 3; c_idx++) {
 -        ptrdiff_t stride = s->frame->linesize[c_idx];
 -        int hshift       = s->sps->hshift[c_idx];
 -        int vshift       = s->sps->vshift[c_idx];
 -        for (y = 0; y < s->sps->min_pu_height; y++) {
 -            for (x = 0; x < s->sps->min_pu_width; x++) {
 -                if (s->is_pcm[y * s->sps->min_pu_width + x]) {
 -                    int n;
 -                    int len      = min_pu_size >> hshift;
 -                    uint8_t *src = &s->frame->data[c_idx][((y << s->sps->log2_min_pu_size) >> vshift) * stride + (((x << s->sps->log2_min_pu_size) >> hshift) << s->sps->pixel_shift)];
 -                    uint8_t *dst = &s->sao_frame->data[c_idx][((y << s->sps->log2_min_pu_size) >> vshift) * stride + (((x << s->sps->log2_min_pu_size) >> hshift) << s->sps->pixel_shift)];
 -                    for (n = 0; n < (min_pu_size >> vshift); n++) {
 -                        memcpy(dst, src, len);
 -                        src += stride;
 -                        dst += stride;
 -                    }
 -                }
 -            }
 -        }
 -    }
 -}
 -
  static int set_side_data(HEVCContext *s)
  {
      AVFrame *out = s->ref->frame;
  
  static int hevc_frame_start(HEVCContext *s)
  {
 -    HEVCLocalContext *lc = &s->HEVClc;
 +    HEVCLocalContext *lc = s->HEVClc;
 +    int pic_size_in_ctb  = ((s->sps->width  >> s->sps->log2_min_cb_size) + 1) *
 +                           ((s->sps->height >> s->sps->log2_min_cb_size) + 1);
      int ret;
  
 -    memset(s->horizontal_bs, 0, 2 * s->bs_width * (s->bs_height + 1));
 -    memset(s->vertical_bs,   0, 2 * s->bs_width * (s->bs_height + 1));
 +    memset(s->horizontal_bs, 0, s->bs_width * s->bs_height);
 +    memset(s->vertical_bs,   0, s->bs_width * s->bs_height);
      memset(s->cbf_luma,      0, s->sps->min_tb_width * s->sps->min_tb_height);
 -    memset(s->is_pcm,        0, s->sps->min_pu_width * s->sps->min_pu_height);
 +    memset(s->is_pcm,        0, (s->sps->min_pu_width + 1) * (s->sps->min_pu_height + 1));
 +    memset(s->tab_slice_address, -1, pic_size_in_ctb * sizeof(*s->tab_slice_address));
  
 -    lc->start_of_tiles_x = 0;
      s->is_decoded        = 0;
      s->first_nal_type    = s->nal_unit_type;
  
      if (s->pps->tiles_enabled_flag)
          lc->end_of_tiles_x = s->pps->column_width[0] << s->sps->log2_ctb_size;
  
 -    ret = ff_hevc_set_new_ref(s, s->sps->sao_enabled ? &s->sao_frame : &s->frame,
 -                              s->poc);
 +    ret = ff_hevc_set_new_ref(s, &s->frame, s->poc);
      if (ret < 0)
          goto fail;
  
      if (ret < 0)
          goto fail;
  
 +    s->frame->pict_type = 3 - s->sh.slice_type;
 +
 +    if (!IS_IRAP(s))
 +        ff_hevc_bump_frame(s);
 +
      av_frame_unref(s->output_frame);
      ret = ff_hevc_output_frame(s, s->output_frame, 0);
      if (ret < 0)
          goto fail;
  
 -    ff_thread_finish_setup(s->avctx);
 +    if (!s->avctx->hwaccel)
 +        ff_thread_finish_setup(s->avctx);
  
      return 0;
  
@@@ -2639,7 -2495,7 +2639,7 @@@ fail
  
  static int decode_nal_unit(HEVCContext *s, const HEVCNAL *nal)
  {
 -    HEVCLocalContext *lc = &s->HEVClc;
 +    HEVCLocalContext *lc = s->HEVClc;
      GetBitContext *gb    = &lc->gb;
      int ctb_addr_ts, ret;
  
              if (ret < 0)
                  goto fail;
          } else {
 -            ctb_addr_ts = hls_slice_data(s);
 +            if (s->threads_number > 1 && s->sh.num_entry_point_offsets > 0)
 +                ctb_addr_ts = hls_slice_data_wpp(s, nal->data, nal->size);
 +            else
 +                ctb_addr_ts = hls_slice_data(s);
              if (ctb_addr_ts >= (s->sps->ctb_width * s->sps->ctb_height)) {
                  s->is_decoded = 1;
 -                if ((s->pps->transquant_bypass_enable_flag ||
 -                     (s->sps->pcm.loop_filter_disable_flag && s->sps->pcm_enabled_flag)) &&
 -                    s->sps->sao_enabled)
 -                    restore_tqb_pixels(s);
              }
  
              if (ctb_addr_ts < 0) {
      return 0;
  }
  
- /* FIXME: This is adapted from ff_h264_decode_nal, avoiding duplication
-  * between these functions would be nice. */
- int ff_hevc_extract_rbsp(HEVCContext *s, const uint8_t *src, int length,
-                          HEVCNAL *nal)
- {
-     int i, si, di;
-     uint8_t *dst;
-     s->skipped_bytes = 0;
- #define STARTCODE_TEST                                                  \
-         if (i + 2 < length && src[i + 1] == 0 && src[i + 2] <= 3) {     \
-             if (src[i + 2] != 3) {                                      \
-                 /* startcode, so we must be past the end */             \
-                 length = i;                                             \
-             }                                                           \
-             break;                                                      \
-         }
- #if HAVE_FAST_UNALIGNED
- #define FIND_FIRST_ZERO                                                 \
-         if (i > 0 && !src[i])                                           \
-             i--;                                                        \
-         while (src[i])                                                  \
-             i++
- #if HAVE_FAST_64BIT
-     for (i = 0; i + 1 < length; i += 9) {
-         if (!((~AV_RN64A(src + i) &
-                (AV_RN64A(src + i) - 0x0100010001000101ULL)) &
-               0x8000800080008080ULL))
-             continue;
-         FIND_FIRST_ZERO;
-         STARTCODE_TEST;
-         i -= 7;
-     }
- #else
-     for (i = 0; i + 1 < length; i += 5) {
-         if (!((~AV_RN32A(src + i) &
-                (AV_RN32A(src + i) - 0x01000101U)) &
-               0x80008080U))
-             continue;
-         FIND_FIRST_ZERO;
-         STARTCODE_TEST;
-         i -= 3;
-     }
- #endif /* HAVE_FAST_64BIT */
- #else
-     for (i = 0; i + 1 < length; i += 2) {
-         if (src[i])
-             continue;
-         if (i > 0 && src[i - 1] == 0)
-             i--;
-         STARTCODE_TEST;
-     }
- #endif /* HAVE_FAST_UNALIGNED */
-     if (i >= length - 1) { // no escaped 0
-         nal->data     =
-         nal->raw_data = src;
-         nal->size     =
-         nal->raw_size = length;
-         return length;
-     }
-     av_fast_malloc(&nal->rbsp_buffer, &nal->rbsp_buffer_size,
-                    length + FF_INPUT_BUFFER_PADDING_SIZE);
-     if (!nal->rbsp_buffer)
-         return AVERROR(ENOMEM);
-     dst = nal->rbsp_buffer;
-     memcpy(dst, src, i);
-     si = di = i;
-     while (si + 2 < length) {
-         // remove escapes (very rare 1:2^22)
-         if (src[si + 2] > 3) {
-             dst[di++] = src[si++];
-             dst[di++] = src[si++];
-         } else if (src[si] == 0 && src[si + 1] == 0) {
-             if (src[si + 2] == 3) { // escape
-                 dst[di++] = 0;
-                 dst[di++] = 0;
-                 si       += 3;
-                 s->skipped_bytes++;
-                 if (s->skipped_bytes_pos_size < s->skipped_bytes) {
-                     s->skipped_bytes_pos_size *= 2;
-                     av_reallocp_array(&s->skipped_bytes_pos,
-                             s->skipped_bytes_pos_size,
-                             sizeof(*s->skipped_bytes_pos));
-                     if (!s->skipped_bytes_pos)
-                         return AVERROR(ENOMEM);
-                 }
-                 if (s->skipped_bytes_pos)
-                     s->skipped_bytes_pos[s->skipped_bytes-1] = di - 1;
-                 continue;
-             } else // next start code
-                 goto nsc;
-         }
-         dst[di++] = src[si++];
-     }
-     while (si < length)
-         dst[di++] = src[si++];
- nsc:
-     memset(dst + di, 0, FF_INPUT_BUFFER_PADDING_SIZE);
-     nal->data = dst;
-     nal->size = di;
-     nal->raw_data = src;
-     nal->raw_size = si;
-     return si;
- }
  static int decode_nal_units(HEVCContext *s, const uint8_t *buf, int length)
  {
      int i, consumed, ret = 0;
  
      s->ref = NULL;
 +    s->last_eos = s->eos;
      s->eos = 0;
  
      /* split the input packet into NAL units, so we know the upper bound on the
                  goto fail;
              }
          } else {
 -            if (buf[2] == 0) {
 -                length--;
 -                buf++;
 -                continue;
 -            }
 -            if (buf[0] != 0 || buf[1] != 0 || buf[2] != 1) {
 -                ret = AVERROR_INVALIDDATA;
 -                goto fail;
 +            /* search start code */
 +            while (buf[0] != 0 || buf[1] != 0 || buf[2] != 1) {
 +                ++buf;
 +                --length;
 +                if (length < 4) {
 +                    av_log(s->avctx, AV_LOG_ERROR, "No start code is found.\n");
 +                    ret = AVERROR_INVALIDDATA;
 +                    goto fail;
 +                }
              }
  
              buf           += 3;
              length        -= 3;
 -            extract_length = length;
          }
  
 +        if (!s->is_nalff)
 +            extract_length = length;
 +
          if (s->nals_allocated < s->nb_nals + 1) {
              int new_size = s->nals_allocated + 1;
 -            HEVCNAL *tmp = av_realloc_array(s->nals, new_size, sizeof(*tmp));
 +            void *tmp = av_realloc_array(s->nals, new_size, sizeof(*s->nals));
 +            ret = AVERROR(ENOMEM);
              if (!tmp) {
 -                ret = AVERROR(ENOMEM);
                  goto fail;
              }
              s->nals = tmp;
              memset(s->nals + s->nals_allocated, 0,
 -                   (new_size - s->nals_allocated) * sizeof(*tmp));
 +                   (new_size - s->nals_allocated) * sizeof(*s->nals));
 +
 +            tmp = av_realloc_array(s->skipped_bytes_nal, new_size, sizeof(*s->skipped_bytes_nal));
 +            if (!tmp)
 +                goto fail;
 +            s->skipped_bytes_nal = tmp;
 +
 +            tmp = av_realloc_array(s->skipped_bytes_pos_size_nal, new_size, sizeof(*s->skipped_bytes_pos_size_nal));
 +            if (!tmp)
 +                goto fail;
 +            s->skipped_bytes_pos_size_nal = tmp;
 +
 +            tmp = av_realloc_array(s->skipped_bytes_pos_nal, new_size, sizeof(*s->skipped_bytes_pos_nal));
 +            if (!tmp)
 +                goto fail;
 +            s->skipped_bytes_pos_nal = tmp;
 +
 +            s->skipped_bytes_pos_size_nal[s->nals_allocated] = 1024; // initial buffer size
 +            s->skipped_bytes_pos_nal[s->nals_allocated] = av_malloc_array(s->skipped_bytes_pos_size_nal[s->nals_allocated], sizeof(*s->skipped_bytes_pos));
 +            if (!s->skipped_bytes_pos_nal[s->nals_allocated])
 +                goto fail;
              s->nals_allocated = new_size;
          }
 -        nal = &s->nals[s->nb_nals++];
 +        s->skipped_bytes_pos_size = s->skipped_bytes_pos_size_nal[s->nb_nals];
 +        s->skipped_bytes_pos = s->skipped_bytes_pos_nal[s->nb_nals];
 +        nal = &s->nals[s->nb_nals];
  
 -        consumed = ff_hevc_extract_rbsp(buf, extract_length, nal);
 +        consumed = ff_hevc_extract_rbsp(s, buf, extract_length, nal);
-         s->skipped_bytes_nal[s->nb_nals] = s->skipped_bytes;
-         s->skipped_bytes_pos_size_nal[s->nb_nals] = s->skipped_bytes_pos_size;
-         s->skipped_bytes_pos_nal[s->nb_nals++] = s->skipped_bytes_pos;
          if (consumed < 0) {
              ret = consumed;
              goto fail;
          }
  
 -        ret = init_get_bits8(&s->HEVClc.gb, nal->data, nal->size);
++        s->skipped_bytes_nal[s->nb_nals] = s->skipped_bytes;
++        s->skipped_bytes_pos_size_nal[s->nb_nals] = s->skipped_bytes_pos_size;
++        s->skipped_bytes_pos_nal[s->nb_nals++] = s->skipped_bytes_pos;
++
 +        ret = init_get_bits8(&s->HEVClc->gb, nal->data, nal->size);
          if (ret < 0)
              goto fail;
          hls_nal_unit(s);
  
      /* parse the NAL units */
      for (i = 0; i < s->nb_nals; i++) {
 +        s->skipped_bytes = s->skipped_bytes_nal[i];
 +        s->skipped_bytes_pos = s->skipped_bytes_pos_nal[i];
 +
          ret = decode_nal_unit(s, &s->nals[i]);
          if (ret < 0) {
              av_log(s->avctx, AV_LOG_WARNING,
      }
  
  fail:
 -    if (s->ref)
 +    if (s->ref && s->threads_type == FF_THREAD_FRAME)
          ff_thread_report_progress(&s->ref->tf, INT_MAX, 0);
  
      return ret;
@@@ -3151,9 -2860,7 +3036,9 @@@ static int hevc_decode_frame(AVCodecCon
  
  static int hevc_ref_frame(HEVCContext *s, HEVCFrame *dst, HEVCFrame *src)
  {
 -    int ret = ff_thread_ref_frame(&dst->tf, &src->tf);
 +    int ret;
 +
 +    ret = ff_thread_ref_frame(&dst->tf, &src->tf);
      if (ret < 0)
          return ret;
  
@@@ -3199,19 -2906,7 +3084,19 @@@ static av_cold int hevc_decode_free(AVC
  
      av_freep(&s->md5_ctx);
  
 -    av_frame_free(&s->tmp_frame);
 +    for(i=0; i < s->nals_allocated; i++) {
 +        av_freep(&s->skipped_bytes_pos_nal[i]);
 +    }
 +    av_freep(&s->skipped_bytes_pos_size_nal);
 +    av_freep(&s->skipped_bytes_nal);
 +    av_freep(&s->skipped_bytes_pos_nal);
 +
 +    av_freep(&s->cabac_state);
 +
 +    for (i = 0; i < 3; i++) {
 +        av_freep(&s->sao_pixel_buffer_h[i]);
 +        av_freep(&s->sao_pixel_buffer_v[i]);
 +    }
      av_frame_free(&s->output_frame);
  
      for (i = 0; i < FF_ARRAY_ELEMS(s->DPB); i++) {
          av_buffer_unref(&s->sps_list[i]);
      for (i = 0; i < FF_ARRAY_ELEMS(s->pps_list); i++)
          av_buffer_unref(&s->pps_list[i]);
 +    s->sps = NULL;
 +    s->pps = NULL;
 +    s->vps = NULL;
 +
 +    av_freep(&s->sh.entry_point_offset);
 +    av_freep(&s->sh.offset);
 +    av_freep(&s->sh.size);
 +
 +    for (i = 1; i < s->threads_number; i++) {
 +        HEVCLocalContext *lc = s->HEVClcList[i];
 +        if (lc) {
 +            av_freep(&s->HEVClcList[i]);
 +            av_freep(&s->sList[i]);
 +        }
 +    }
 +    if (s->HEVClc == s->HEVClcList[0])
 +        s->HEVClc = NULL;
 +    av_freep(&s->HEVClcList[0]);
  
      for (i = 0; i < s->nals_allocated; i++)
          av_freep(&s->nals[i].rbsp_buffer);
@@@ -3259,14 -2936,8 +3144,14 @@@ static av_cold int hevc_init_context(AV
  
      s->avctx = avctx;
  
 -    s->tmp_frame = av_frame_alloc();
 -    if (!s->tmp_frame)
 +    s->HEVClc = av_mallocz(sizeof(HEVCLocalContext));
 +    if (!s->HEVClc)
 +        goto fail;
 +    s->HEVClcList[0] = s->HEVClc;
 +    s->sList[0] = s;
 +
 +    s->cabac_state = av_malloc(HEVC_CONTEXTS);
 +    if (!s->cabac_state)
          goto fail;
  
      s->output_frame = av_frame_alloc();
      ff_bswapdsp_init(&s->bdsp);
  
      s->context_initialized = 1;
 +    s->eos = 0;
  
      return 0;
  
@@@ -3320,8 -2990,6 +3205,8 @@@ static int hevc_update_thread_context(A
          }
      }
  
 +    if (s->sps != s0->sps)
 +        s->sps = NULL;
      for (i = 0; i < FF_ARRAY_ELEMS(s->vps_list); i++) {
          av_buffer_unref(&s->vps_list[i]);
          if (s0->vps_list[i]) {
      }
  
      if (s->sps != s0->sps)
 -        ret = set_sps(s, s0->sps);
 +        if ((ret = set_sps(s, s0->sps, src->pix_fmt)) < 0)
 +            return ret;
  
      s->seq_decode = s0->seq_decode;
      s->seq_output = s0->seq_output;
      s->pocTid0    = s0->pocTid0;
      s->max_ra     = s0->max_ra;
 +    s->eos        = s0->eos;
  
      s->is_nalff        = s0->is_nalff;
      s->nal_length_size = s0->nal_length_size;
  
 +    s->threads_number      = s0->threads_number;
 +    s->threads_type        = s0->threads_type;
 +
      if (s0->eos) {
          s->seq_decode = (s->seq_decode + 1) & 0xff;
          s->max_ra = INT_MAX;
@@@ -3460,14 -3123,6 +3345,14 @@@ static av_cold int hevc_decode_init(AVC
      if (ret < 0)
          return ret;
  
 +    s->enable_parallel_tiles = 0;
 +    s->picture_struct = 0;
 +
 +    if(avctx->active_thread_type & FF_THREAD_SLICE)
 +        s->threads_number = avctx->thread_count;
 +    else
 +        s->threads_number = 1;
 +
      if (avctx->extradata_size > 0 && avctx->extradata) {
          ret = hevc_decode_extradata(s);
          if (ret < 0) {
          }
      }
  
 +    if((avctx->active_thread_type & FF_THREAD_FRAME) && avctx->thread_count > 1)
 +            s->threads_type = FF_THREAD_FRAME;
 +        else
 +            s->threads_type = FF_THREAD_SLICE;
 +
      return 0;
  }
  
@@@ -3512,15 -3162,12 +3397,15 @@@ static const AVProfile profiles[] = 
      { FF_PROFILE_HEVC_MAIN,                 "Main"                },
      { FF_PROFILE_HEVC_MAIN_10,              "Main 10"             },
      { FF_PROFILE_HEVC_MAIN_STILL_PICTURE,   "Main Still Picture"  },
 +    { FF_PROFILE_HEVC_REXT,                 "Rext"  },
      { FF_PROFILE_UNKNOWN },
  };
  
  static const AVOption options[] = {
      { "apply_defdispwin", "Apply default display window from VUI", OFFSET(apply_defdispwin),
          AV_OPT_TYPE_INT, {.i64 = 0}, 0, 1, PAR },
 +    { "strict-displaywin", "stricly apply default display window size", OFFSET(apply_defdispwin),
 +        AV_OPT_TYPE_INT, {.i64 = 0}, 0, 1, PAR },
      { NULL },
  };
  
@@@ -3545,6 -3192,6 +3430,6 @@@ AVCodec ff_hevc_decoder = 
      .update_thread_context = hevc_update_thread_context,
      .init_thread_copy      = hevc_init_thread_copy,
      .capabilities          = CODEC_CAP_DR1 | CODEC_CAP_DELAY |
 -                             CODEC_CAP_FRAME_THREADS,
 +                             CODEC_CAP_SLICE_THREADS | CODEC_CAP_FRAME_THREADS,
      .profiles              = NULL_IF_CONFIG_SMALL(profiles),
  };
diff --combined libavcodec/hevc.h
@@@ -3,26 -3,29 +3,26 @@@
   *
   * Copyright (C) 2012 - 2013 Guillaume Martres
   *
 - * This file is part of Libav.
 + * This file is part of FFmpeg.
   *
 - * Libav is free software; you can redistribute it and/or
 + * FFmpeg is free software; you can redistribute it and/or
   * modify it under the terms of the GNU Lesser General Public
   * License as published by the Free Software Foundation; either
   * version 2.1 of the License, or (at your option) any later version.
   *
 - * Libav is distributed in the hope that it will be useful,
 + * 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
   * Lesser General Public License for more details.
   *
   * You should have received a copy of the GNU Lesser General Public
 - * License along with Libav; if not, write to the Free Software
 + * License along with FFmpeg; if not, write to the Free Software
   * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
   */
  
  #ifndef AVCODEC_HEVC_H
  #define AVCODEC_HEVC_H
  
 -#include <stddef.h>
 -#include <stdint.h>
 -
  #include "libavutil/buffer.h"
  #include "libavutil/md5.h"
  
@@@ -30,7 -33,6 +30,7 @@@
  #include "bswapdsp.h"
  #include "cabac.h"
  #include "get_bits.h"
 +#include "hevcpred.h"
  #include "hevcdsp.h"
  #include "internal.h"
  #include "thread.h"
@@@ -39,9 -41,6 +39,9 @@@
  #define MAX_DPB_SIZE 16 // A.4.1
  #define MAX_REFS 16
  
 +#define MAX_NB_THREADS 16
 +#define SHIFT_CTB_WPP 2
 +
  /**
   * 7.4.2.1
   */
  #define MAX_TRANSFORM_DEPTH 5
  
  #define MAX_TB_SIZE 32
 -#define MAX_PB_SIZE 64
  #define MAX_LOG2_CTB_SIZE 6
  #define MAX_QP 51
  #define DEFAULT_INTRA_TC_OFFSET 2
  
 -#define HEVC_CONTEXTS 183
 +#define HEVC_CONTEXTS 199
  
  #define MRG_MAX_NUM_CANDS     5
  
@@@ -70,9 -70,6 +70,9 @@@
  #define EPEL_EXTRA_BEFORE 1
  #define EPEL_EXTRA_AFTER  2
  #define EPEL_EXTRA        3
 +#define QPEL_EXTRA_BEFORE 3
 +#define QPEL_EXTRA_AFTER  4
 +#define QPEL_EXTRA        7
  
  #define EDGE_EMU_BUFFER_STRIDE 80
  
  #define SAMPLE(tab, x, y) ((tab)[(y) * s->sps->width + (x)])
  #define SAMPLE_CTB(tab, x, y) ((tab)[(y) * min_cb_width + (x)])
  
 -#define IS_IDR(s) (s->nal_unit_type == NAL_IDR_W_RADL || s->nal_unit_type == NAL_IDR_N_LP)
 -#define IS_BLA(s) (s->nal_unit_type == NAL_BLA_W_RADL || s->nal_unit_type == NAL_BLA_W_LP || \
 -                   s->nal_unit_type == NAL_BLA_N_LP)
 -#define IS_IRAP(s) (s->nal_unit_type >= 16 && s->nal_unit_type <= 23)
 -
 -#define FFUDIV(a,b) (((a) > 0 ? (a) : (a) - (b) + 1) / (b))
 -#define FFUMOD(a,b) ((a) - (b) * FFUDIV(a,b))
 +#define IS_IDR(s) ((s)->nal_unit_type == NAL_IDR_W_RADL || (s)->nal_unit_type == NAL_IDR_N_LP)
 +#define IS_BLA(s) ((s)->nal_unit_type == NAL_BLA_W_RADL || (s)->nal_unit_type == NAL_BLA_W_LP || \
 +                   (s)->nal_unit_type == NAL_BLA_N_LP)
 +#define IS_IRAP(s) ((s)->nal_unit_type >= 16 && (s)->nal_unit_type <= 23)
  
  /**
   * Table 7-3: NAL unit type codes
@@@ -167,8 -167,6 +167,8 @@@ enum SyntaxElement 
      CBF_LUMA,
      CBF_CB_CR,
      TRANSFORM_SKIP_FLAG,
 +    EXPLICIT_RDPCM_FLAG,
 +    EXPLICIT_RDPCM_DIR_FLAG,
      LAST_SIGNIFICANT_COEFF_X_PREFIX,
      LAST_SIGNIFICANT_COEFF_Y_PREFIX,
      LAST_SIGNIFICANT_COEFF_X_SUFFIX,
      COEFF_ABS_LEVEL_GREATER2_FLAG,
      COEFF_ABS_LEVEL_REMAINING,
      COEFF_SIGN_FLAG,
 +    LOG2_RES_SCALE_ABS,
 +    RES_SCALE_SIGN_FLAG,
 +    CU_CHROMA_QP_OFFSET_FLAG,
 +    CU_CHROMA_QP_OFFSET_IDX,
  };
  
  enum PartMode {
@@@ -208,13 -202,6 +208,13 @@@ enum InterPredIdc 
      PRED_BI,
  };
  
 +enum PredFlag {
 +    PF_INTRA = 0,
 +    PF_L0,
 +    PF_L1,
 +    PF_BI,
 +};
 +
  enum IntraPredMode {
      INTRA_PLANAR = 0,
      INTRA_DC,
@@@ -257,7 -244,6 +257,7 @@@ enum SAOType 
      SAO_NOT_APPLIED = 0,
      SAO_BAND,
      SAO_EDGE,
 +    SAO_APPLIED
  };
  
  enum SAOEOClass {
@@@ -395,7 -381,7 +395,7 @@@ typedef struct ScalingList 
  } ScalingList;
  
  typedef struct HEVCSPS {
 -    int vps_id;
 +    unsigned vps_id;
      int chroma_format_idc;
      uint8_t separate_colour_plane_flag;
  
      int max_transform_hierarchy_depth_inter;
      int max_transform_hierarchy_depth_intra;
  
 +    int transform_skip_rotation_enabled_flag;
 +    int transform_skip_context_enabled_flag;
 +    int implicit_rdpcm_enabled_flag;
 +    int explicit_rdpcm_enabled_flag;
 +    int intra_smoothing_disabled_flag;
 +    int persistent_rice_adaptation_enabled_flag;
 +
      ///< coded frame dimension in various units
      int width;
      int height;
      int min_tb_height;
      int min_pu_width;
      int min_pu_height;
 +    int tb_mask;
  
      int hshift[3];
      int vshift[3];
@@@ -532,15 -510,6 +532,15 @@@ typedef struct HEVCPPS 
      int log2_parallel_merge_level; ///< log2_parallel_merge_level_minus2 + 2
      int num_extra_slice_header_bits;
      uint8_t slice_header_extension_present_flag;
 +    uint8_t log2_max_transform_skip_block_size;
 +    uint8_t cross_component_prediction_enabled_flag;
 +    uint8_t chroma_qp_offset_list_enabled_flag;
 +    uint8_t diff_cu_chroma_qp_offset_depth;
 +    uint8_t chroma_qp_offset_list_len_minus1;
 +    int8_t  cb_qp_offset_list[5];
 +    int8_t  cr_qp_offset_list[5];
 +    uint8_t log2_sao_offset_scale_luma;
 +    uint8_t log2_sao_offset_scale_chroma;
  
      // Inferred parameters
      unsigned int *column_width;  ///< ColumnWidth
      int *tile_id;           ///< TileId
      int *tile_pos_rs;       ///< TilePosRS
      int *min_tb_addr_zs;    ///< MinTbAddrZS
 +    int *min_tb_addr_zs_tab;///< MinTbAddrZS
  } HEVCPPS;
  
  typedef struct SliceHeader {
      int short_term_ref_pic_set_size;
      ShortTermRPS slice_rps;
      const ShortTermRPS *short_term_rps;
 +    int long_term_ref_pic_set_size;
      LongTermRPS long_term_rps;
      unsigned int list_entry_lx[2][32];
  
      int slice_cb_qp_offset;
      int slice_cr_qp_offset;
  
 +    uint8_t cu_chroma_qp_offset_enabled_flag;
 +
      int beta_offset;    ///< beta_offset_div2 * 2
      int tc_offset;      ///< tc_offset_div2 * 2
  
      unsigned int max_num_merge_cand; ///< 5 - 5_minus_max_num_merge_cand
  
 +    int *entry_point_offset;
 +    int * offset;
 +    int * size;
      int num_entry_point_offsets;
  
      int8_t slice_qp;
      int slice_ctb_addr_rs;
  } SliceHeader;
  
 -typedef struct CodingTree {
 -    int depth; ///< ctDepth
 -} CodingTree;
 -
  typedef struct CodingUnit {
      int x;
      int y;
@@@ -655,7 -621,8 +655,7 @@@ typedef struct Mv 
  typedef struct MvField {
      DECLARE_ALIGNED(4, Mv, mv)[2];
      int8_t ref_idx[2];
 -    int8_t pred_flag[2];
 -    uint8_t is_intra;
 +    int8_t pred_flag;
  } MvField;
  
  typedef struct NeighbourAvailable {
@@@ -673,24 -640,15 +673,24 @@@ typedef struct PredictionUnit 
      uint8_t intra_pred_mode[4];
      Mv mvd;
      uint8_t merge_flag;
 -    uint8_t intra_pred_mode_c;
 +    uint8_t intra_pred_mode_c[4];
 +    uint8_t chroma_mode_c[4];
  } PredictionUnit;
  
  typedef struct TransformUnit {
      int cu_qp_delta;
  
 +    int res_scale_val;
 +
      // Inferred parameters;
 -    int cur_intra_pred_mode;
 +    int intra_pred_mode;
 +    int intra_pred_mode_c;
 +    int chroma_mode_c;
      uint8_t is_cu_qp_delta_coded;
 +    uint8_t is_cu_chroma_qp_offset_coded;
 +    int8_t  cu_qp_offset_cb;
 +    int8_t  cu_qp_offset_cr;
 +    uint8_t cross_pf;
  } TransformUnit;
  
  typedef struct DBParams {
  #define HEVC_FRAME_FLAG_OUTPUT    (1 << 0)
  #define HEVC_FRAME_FLAG_SHORT_REF (1 << 1)
  #define HEVC_FRAME_FLAG_LONG_REF  (1 << 2)
 +#define HEVC_FRAME_FLAG_BUMPING   (1 << 3)
  
  typedef struct HEVCFrame {
      AVFrame *frame;
@@@ -745,11 -702,24 +745,11 @@@ typedef struct HEVCNAL 
      const uint8_t *raw_data;
  } HEVCNAL;
  
 -struct HEVCContext;
 -
 -typedef struct HEVCPredContext {
 -    void (*intra_pred[4])(struct HEVCContext *s, int x0, int y0, int c_idx);
 -
 -    void (*pred_planar[4])(uint8_t *src, const uint8_t *top,
 -                           const uint8_t *left, ptrdiff_t stride);
 -    void (*pred_dc)(uint8_t *src, const uint8_t *top, const uint8_t *left,
 -                    ptrdiff_t stride, int log2_size, int c_idx);
 -    void (*pred_angular[4])(uint8_t *src, const uint8_t *top,
 -                            const uint8_t *left, ptrdiff_t stride,
 -                            int c_idx, int mode);
 -} HEVCPredContext;
 -
  typedef struct HEVCLocalContext {
 -    DECLARE_ALIGNED(16, int16_t, mc_buffer[(MAX_PB_SIZE + 7) * MAX_PB_SIZE]);
      uint8_t cabac_state[HEVC_CONTEXTS];
  
 +    uint8_t stat_coeff[4];
 +
      uint8_t first_qp_group;
  
      GetBitContext gb;
      int8_t qp_y;
      int8_t curr_qp_y;
  
 +    int qPy_pred;
 +
      TransformUnit tu;
  
      uint8_t ctb_left_flag;
      uint8_t ctb_up_flag;
      uint8_t ctb_up_right_flag;
      uint8_t ctb_up_left_flag;
 -    int     start_of_tiles_x;
      int     end_of_tiles_x;
      int     end_of_tiles_y;
      /* +7 is for subpixel interpolation, *2 for high bit depths */
      DECLARE_ALIGNED(32, uint8_t, edge_emu_buffer)[(MAX_PB_SIZE + 7) * EDGE_EMU_BUFFER_STRIDE * 2];
 -    CodingTree ct;
 +    /* The extended size between the new edge emu buffer is abused by SAO */
 +    DECLARE_ALIGNED(32, uint8_t, edge_emu_buffer2)[(MAX_PB_SIZE + 7) * EDGE_EMU_BUFFER_STRIDE * 2];
 +    DECLARE_ALIGNED(32, int16_t, tmp [MAX_PB_SIZE * MAX_PB_SIZE]);
 +
 +    int ct_depth;
      CodingUnit cu;
      PredictionUnit pu;
      NeighbourAvailable na;
@@@ -792,26 -757,17 +792,26 @@@ typedef struct HEVCContext 
      const AVClass *c;  // needed by private avoptions
      AVCodecContext *avctx;
  
 -    HEVCLocalContext HEVClc;
 +    struct HEVCContext  *sList[MAX_NB_THREADS];
  
 -    uint8_t cabac_state[HEVC_CONTEXTS];
 +    HEVCLocalContext    *HEVClcList[MAX_NB_THREADS];
 +    HEVCLocalContext    *HEVClc;
 +
 +    uint8_t             threads_type;
 +    uint8_t             threads_number;
 +
 +    int                 width;
 +    int                 height;
 +
 +    uint8_t *cabac_state;
  
      /** 1 if the independent slice segment header was successfully parsed */
      uint8_t slice_initialized;
  
      AVFrame *frame;
 -    AVFrame *sao_frame;
 -    AVFrame *tmp_frame;
      AVFrame *output_frame;
 +    uint8_t *sao_pixel_buffer_h[3];
 +    uint8_t *sao_pixel_buffer_v[3];
  
      const HEVCVPS *vps;
      const HEVCSPS *sps;
      int pocTid0;
      int slice_idx; ///< number of the slice being currently decoded
      int eos;       ///< current packet contains an EOS/EOB NAL
 +    int last_eos;  ///< last packet contains an EOS/EOB NAL
      int max_ra;
      int bs_width;
      int bs_height;
      uint16_t seq_decode;
      uint16_t seq_output;
  
 +    int enable_parallel_tiles;
 +    int wpp_err;
 +    int skipped_bytes;
 +    int *skipped_bytes_pos;
 +    int skipped_bytes_pos_size;
 +
 +    int *skipped_bytes_nal;
 +    int **skipped_bytes_pos_nal;
 +    int *skipped_bytes_pos_size_nal;
 +
 +    const uint8_t *data;
 +
      HEVCNAL *nals;
      int nb_nals;
      int nals_allocated;
                              ///< as a format defined in 14496-15
      int apply_defdispwin;
  
 +    int active_seq_parameter_set_id;
 +
      int nal_length_size;    ///< Number of bytes used for nal length (1, 2 or 4)
      int nuh_layer_id;
  
      int sei_display_orientation_present;
      int sei_anticlockwise_rotation;
      int sei_hflip, sei_vflip;
 +
 +    int picture_struct;
  } HEVCContext;
  
  int ff_hevc_decode_short_term_rps(GetBitContext *gb, AVCodecContext *avctx,
@@@ -944,9 -883,6 +944,6 @@@ int ff_hevc_decode_nal_sps(HEVCContext 
  int ff_hevc_decode_nal_pps(HEVCContext *s);
  int ff_hevc_decode_nal_sei(HEVCContext *s);
  
- int ff_hevc_extract_rbsp(HEVCContext *s, const uint8_t *src, int length,
-                          HEVCNAL *nal);
  /**
   * Mark all frames in DPB as unused for reference.
   */
@@@ -1002,11 -938,32 +999,11 @@@ int ff_hevc_inter_pred_idc_decode(HEVCC
  int ff_hevc_ref_idx_lx_decode(HEVCContext *s, int num_ref_idx_lx);
  int ff_hevc_mvp_lx_flag_decode(HEVCContext *s);
  int ff_hevc_no_residual_syntax_flag_decode(HEVCContext *s);
 -int ff_hevc_abs_mvd_greater0_flag_decode(HEVCContext *s);
 -int ff_hevc_abs_mvd_greater1_flag_decode(HEVCContext *s);
 -int ff_hevc_mvd_decode(HEVCContext *s);
 -int ff_hevc_mvd_sign_flag_decode(HEVCContext *s);
  int ff_hevc_split_transform_flag_decode(HEVCContext *s, int log2_trafo_size);
  int ff_hevc_cbf_cb_cr_decode(HEVCContext *s, int trafo_depth);
  int ff_hevc_cbf_luma_decode(HEVCContext *s, int trafo_depth);
 -int ff_hevc_transform_skip_flag_decode(HEVCContext *s, int c_idx);
 -int ff_hevc_last_significant_coeff_x_prefix_decode(HEVCContext *s, int c_idx,
 -                                                   int log2_size);
 -int ff_hevc_last_significant_coeff_y_prefix_decode(HEVCContext *s, int c_idx,
 -                                                   int log2_size);
 -int ff_hevc_last_significant_coeff_suffix_decode(HEVCContext *s,
 -                                                 int last_significant_coeff_prefix);
 -int ff_hevc_significant_coeff_group_flag_decode(HEVCContext *s, int c_idx,
 -                                                int ctx_cg);
 -int ff_hevc_significant_coeff_flag_decode(HEVCContext *s, int c_idx, int x_c,
 -                                          int y_c, int log2_trafo_size,
 -                                          int scan_idx, int prev_sig);
 -int ff_hevc_coeff_abs_level_greater1_flag_decode(HEVCContext *s, int c_idx,
 -                                                 int ctx_set);
 -int ff_hevc_coeff_abs_level_greater2_flag_decode(HEVCContext *s, int c_idx,
 -                                                 int inc);
 -int ff_hevc_coeff_abs_level_remaining(HEVCContext *s, int base_level,
 -                                      int rc_rice_param);
 -int ff_hevc_coeff_sign_flag(HEVCContext *s, uint8_t nb);
 +int ff_hevc_log2_res_scale_abs(HEVCContext *s, int idx);
 +int ff_hevc_res_scale_sign_flag(HEVCContext *s, int idx);
  
  /**
   * Get the number of candidate references for the current frame.
@@@ -1021,8 -978,6 +1018,8 @@@ int ff_hevc_set_new_ref(HEVCContext *s
   */
  int ff_hevc_output_frame(HEVCContext *s, AVFrame *frame, int flush);
  
 +void ff_hevc_bump_frame(HEVCContext *s);
 +
  void ff_hevc_unref_frame(HEVCContext *s, HEVCFrame *frame, int flags);
  
  void ff_hevc_set_neighbour_available(HEVCContext *s, int x0, int y0,
@@@ -1034,23 -989,25 +1031,29 @@@ void ff_hevc_luma_mv_mvp_mode(HEVCConte
                                int nPbW, int nPbH, int log2_cb_size,
                                int part_idx, int merge_idx,
                                MvField *mv, int mvp_lx_flag, int LX);
 -void ff_hevc_set_qPy(HEVCContext *s, int xC, int yC, int xBase, int yBase,
 +void ff_hevc_set_qPy(HEVCContext *s, int xBase, int yBase,
                       int log2_cb_size);
  void ff_hevc_deblocking_boundary_strengths(HEVCContext *s, int x0, int y0,
                                             int log2_trafo_size);
  int ff_hevc_cu_qp_delta_sign_flag(HEVCContext *s);
  int ff_hevc_cu_qp_delta_abs(HEVCContext *s);
 -void ff_hevc_hls_filter(HEVCContext *s, int x, int y);
 +int ff_hevc_cu_chroma_qp_offset_flag(HEVCContext *s);
 +int ff_hevc_cu_chroma_qp_offset_idx(HEVCContext *s);
 +void ff_hevc_hls_filter(HEVCContext *s, int x, int y, int ctb_size);
  void ff_hevc_hls_filters(HEVCContext *s, int x_ctb, int y_ctb, int ctb_size);
 +void ff_hevc_hls_residual_coding(HEVCContext *s, int x0, int y0,
 +                                 int log2_trafo_size, enum ScanType scan_idx,
 +                                 int c_idx);
  
 -void ff_hevc_pps_free(HEVCPPS **ppps);
 +void ff_hevc_hls_mvd_coding(HEVCContext *s, int x0, int y0, int log2_cb_size);
  
 -void ff_hevc_pred_init(HEVCPredContext *hpc, int bit_depth);
  
 -int ff_hevc_extract_rbsp(const uint8_t *src, int length,
+ /**
+  * Extract the raw (unescaped) HEVC bitstream.
+  */
++int ff_hevc_extract_rbsp(HEVCContext *s, const uint8_t *src, int length,
+                          HEVCNAL *nal);
  extern const uint8_t ff_hevc_qpel_extra_before[4];
  extern const uint8_t ff_hevc_qpel_extra_after[4];
  extern const uint8_t ff_hevc_qpel_extra[4];
diff --combined libavcodec/hevc_parse.c
index 0000000,32a440b..260f1e1
mode 000000,100644..100644
--- /dev/null
@@@ -1,0 -1,129 +1,145 @@@
 - * This file is part of Libav.
+ /*
+  * HEVC common code
+  *
 - * Libav is free software; you can redistribute it and/or
++ * This file is part of FFmpeg.
+  *
 - * Libav is distributed in the hope that it will be useful,
++ * FFmpeg is free software; you can redistribute it and/or
+  * modify it under the terms of the GNU Lesser General Public
+  * License as published by the Free Software Foundation; either
+  * version 2.1 of the License, or (at your option) any later version.
+  *
 - * License along with Libav; if not, write to the Free Software
++ * 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
+  * Lesser General Public License for more details.
+  *
+  * You should have received a copy of the GNU Lesser General Public
 -int ff_hevc_extract_rbsp(const uint8_t *src, int length,
++ * License along with FFmpeg; if not, write to the Free Software
+  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+  */
+ #include <string.h>
+ #include "config.h"
+ #include "libavutil/intreadwrite.h"
+ #include "libavutil/mem.h"
+ #include "hevc.h"
+ /* FIXME: This is adapted from ff_h264_decode_nal, avoiding duplication
+  * between these functions would be nice. */
++int ff_hevc_extract_rbsp(HEVCContext *s, const uint8_t *src, int length,
+                          HEVCNAL *nal)
+ {
+     int i, si, di;
+     uint8_t *dst;
++    if (s)
++        s->skipped_bytes = 0;
+ #define STARTCODE_TEST                                                  \
+         if (i + 2 < length && src[i + 1] == 0 && src[i + 2] <= 3) {     \
+             if (src[i + 2] != 3) {                                      \
+                 /* startcode, so we must be past the end */             \
+                 length = i;                                             \
+             }                                                           \
+             break;                                                      \
+         }
+ #if HAVE_FAST_UNALIGNED
+ #define FIND_FIRST_ZERO                                                 \
+         if (i > 0 && !src[i])                                           \
+             i--;                                                        \
+         while (src[i])                                                  \
+             i++
+ #if HAVE_FAST_64BIT
+     for (i = 0; i + 1 < length; i += 9) {
+         if (!((~AV_RN64A(src + i) &
+                (AV_RN64A(src + i) - 0x0100010001000101ULL)) &
+               0x8000800080008080ULL))
+             continue;
+         FIND_FIRST_ZERO;
+         STARTCODE_TEST;
+         i -= 7;
+     }
+ #else
+     for (i = 0; i + 1 < length; i += 5) {
+         if (!((~AV_RN32A(src + i) &
+                (AV_RN32A(src + i) - 0x01000101U)) &
+               0x80008080U))
+             continue;
+         FIND_FIRST_ZERO;
+         STARTCODE_TEST;
+         i -= 3;
+     }
+ #endif /* HAVE_FAST_64BIT */
+ #else
+     for (i = 0; i + 1 < length; i += 2) {
+         if (src[i])
+             continue;
+         if (i > 0 && src[i - 1] == 0)
+             i--;
+         STARTCODE_TEST;
+     }
+ #endif /* HAVE_FAST_UNALIGNED */
+     if (i >= length - 1) { // no escaped 0
+         nal->data     =
+         nal->raw_data = src;
+         nal->size     =
+         nal->raw_size = length;
+         return length;
+     }
+     av_fast_malloc(&nal->rbsp_buffer, &nal->rbsp_buffer_size,
+                    length + FF_INPUT_BUFFER_PADDING_SIZE);
+     if (!nal->rbsp_buffer)
+         return AVERROR(ENOMEM);
+     dst = nal->rbsp_buffer;
+     memcpy(dst, src, i);
+     si = di = i;
+     while (si + 2 < length) {
+         // remove escapes (very rare 1:2^22)
+         if (src[si + 2] > 3) {
+             dst[di++] = src[si++];
+             dst[di++] = src[si++];
+         } else if (src[si] == 0 && src[si + 1] == 0) {
+             if (src[si + 2] == 3) { // escape
+                 dst[di++] = 0;
+                 dst[di++] = 0;
+                 si       += 3;
++                if (s) {
++                        s->skipped_bytes++;
++                        if (s->skipped_bytes_pos_size < s->skipped_bytes) {
++                        s->skipped_bytes_pos_size *= 2;
++                        av_reallocp_array(&s->skipped_bytes_pos,
++                                s->skipped_bytes_pos_size,
++                                sizeof(*s->skipped_bytes_pos));
++                        if (!s->skipped_bytes_pos)
++                                return AVERROR(ENOMEM);
++                        }
++                        if (s->skipped_bytes_pos)
++                        s->skipped_bytes_pos[s->skipped_bytes-1] = di - 1;
++                }
+                 continue;
+             } else // next start code
+                 goto nsc;
+         }
+         dst[di++] = src[si++];
+     }
+     while (si < length)
+         dst[di++] = src[si++];
+ nsc:
+     memset(dst + di, 0, FF_INPUT_BUFFER_PADDING_SIZE);
+     nal->data = dst;
+     nal->size = di;
+     nal->raw_data = src;
+     nal->raw_size = si;
+     return si;
+ }
++