Merge branch 'release/0.8' into release/0.7

* release/0.8:
  shorten: Fix invalid free()
  j2kdec: Fix crash in get_qcx
  j2kdec: Check curtileno for validity
  atrac3: Fix crash in tonal component decoding. Fixes Ticket780 Bug Found by: cosminamironesei
  h264: check chroma_format_idc range. Fixes Ticket758 Bug found by: Diana Elena Muscalu
  aacsbr: Fix memory corruption. Fixes Ticket760 and Ticket761 Bug Found by: Diana Elena Muscalu
  j2kdec: Fix integer overflow leading to a segfault Fixes Ticket776 Bug found by: Diana Elena Muscalu
  ws_snd1: Fix wrong samples count and crash.
  lavfi: add missing check in avfilter_filter_samples()
  Update Changelog for 0.7.4 release
  Update RELEASE file for 0.7.4
  swscale: fix crash in fast_bilinear code when compiled with -mred-zone.
  vorbis: An additional defense in the Vorbis codec.
  vorbisdec: Fix decoding bug with channel handling

Merged-by: Michael Niedermayer <michaelni@gmx.at>

11 files changed, 875 insertions, 24 deletions

diff --git a/Changelog b/Changelog
new file mode 100644
index 0000000000..30c5b8f8de
--- /dev/null
+++ b/Changelog
@@ -0,0 +1,769 @@
+Entries are sorted chronologically from oldest to youngest within each release,
+releases are sorted from youngest to oldest.
+
+version 0.7.4:
+
+- vorbis: An additional defense in the Vorbis codec. (CVE-2011-3895)
+- vorbisdec: Fix decoding bug with channel handling.
+- matroskadec: Fix a bug where a pointer was cached to an array that might
+  later move due to a realloc(). (CVE-2011-3893)
+- vorbis: Avoid some out-of-bounds reads. (CVE-2011-3893)
+- vp3: fix oob read for negative tokens and memleaks on error, (CVE-2011-3892)
+- avserver: Fix a bug where the socket is IPv4, but IPv6 is autoselected
+  for the loopback address.
+- vp3: fix streams with non-zero last coefficient.
+- swscale: fix crash in fast_bilinear code when compiled with -mred-zone.
+
+
+version 0.7.3:
+
+- check buffer and input values in various parts of the code:
+  vmd (CVE-2011-4364), qdm2 (CVE-2011-4351), imgutils (several codecs),
+  vp6 (CVE-2011-4353), svq1 (CVE-2011-4579), vp3 (CVE-2011-4352), wma, 4xm
+- backport avcodec_open2() as a replacement for avcodec_open()
+- backport avformat_find_stream_info()
+
+
+version 0.7.2:
+
+- check buffer and input values in various parts of the code:
+  H.264, VC-1, APE, FLV, Indeo 2, XAN, Ogg, MXF, wavpack, ffv1, MOV,
+  cavs (OCERT-2011-002, CVE-2011-3362), Smacker, cpu detection, lavf,
+  Matroska (CVE-2011-3504), RV10, RV30/RV40
+- memory leaks: vf_scale, eval
+
+- ARM: workaround for bug in GNU assembler
+- AVOptions: fix av_set_string3() doxy to match reality. (Bug #28)
+- Reintroduce AVInputStream.nb_streams to avoid crashes
+- aac: Only output configure if audio was found
+- aac: Remove some suspicious illegal memcpy()s from LTP
+- aacps: skip some memcpy() if src and dst would be equal
+- adts: fix PCE copying
+- alsa: fallback to buffer_size/4 for period_size
+- alsa: limit buffer_size to 32768 frames
+- cljr, indeo2: init_get_bits size in bits instead of bytes
+- configure: add missing CFLAGS to fix building on the HURD
+- dca: set AVCodecContext frame_size for DTS audio
+- fate: allow testing with libavfilter disabled
+- gxf: fix 25 fps DV material in GXF being misdetected as 50 fps
+- h264: correct implicit weight table computation for long ref pics
+- h264: correct the check for invalid long term frame index in MMCO decode
+- h264: fix PCM intra-coded blocks in monochrome case
+- jpegdec: actually search for and parse RSTn
+- lavc: fix type for thread_type option
+- lavf: fix context pointer in av_open_input_stream when avformat_open_input fails
+- lavf: do not set codec_tag for rawvideo
+- libx264: do not set pic quality if no frame is output
+- movenc: create an alternate group for each media type
+- mpegts: fix Continuity Counter error detection
+- mxfenc: fix ignored drop flag in binary timecode representation
+- fix crashes in 32-bit PIC builds (cf e.g. http://bugs.debian.org/639948)
+- ppc64: fix cast related random failures
+- riff: Add mpgv MPEG-2 fourcc
+- swscale: don't use planar output functions to write to NV12/21
+- vc1: properly zero coded_block[] edges on new slice entry
+- vp3/theora: flush after seek
+
+- various bug other fixes
+
+
+version 0.7.1:
+
+- added various additional FOURCC codec identifiers
+- H.264 4:4:4 fixes
+- build system and compilation fixes
+- Doxygen and general documentation corrections and improvements
+- fixed segfault in ffprobe
+- behavioral fix in av_open_input_stream()
+- Licensing clarification for LGPL'ed vf_gradfun
+- bugfixes while seeking in multithreaded decoding
+- support newer versions of OpenCV
+- ffmpeg: fix operation with --disable-avfilter
+- fixed integer underflow in matroska decoder
+
+
+version 0.7:
+
+- E-AC-3 audio encoder
+- ac3enc: add channel coupling support
+- floating-point sample format support for (E-)AC-3, DCA, AAC, Vorbis decoders
+- H.264/MPEG frame-level multithreading
+- av_metadata_* functions renamed to av_dict_* and moved to libavutil
+- 4:4:4 H.264 decoding support
+- 10-bit H.264 optimizations for x86
+- bump libswscale for recently reported ABI break
+
+
+version 0.7_beta2:
+
+- VP8 frame-level multithreading
+- NEON optimizations for VP8
+- removed a lot of deprecated API cruft
+- FFT and IMDCT optimizations for AVX (Sandy Bridge) processors
+- DPX image encoder
+- SMPTE 302M AES3 audio decoder
+- ffmpeg no longer quits after the 'q' key is pressed; use 'ctrl+c' instead
+- 9bit and 10bit per sample support in the H.264 decoder
+
+
+version 0.7_beta1:
+
+- WebM support in Matroska de/muxer
+- low overhead Ogg muxing
+- MMS-TCP support
+- VP8 de/encoding via libvpx
+- Demuxer for On2's IVF format
+- Pictor/PC Paint decoder
+- HE-AAC v2 decoder
+- libfaad2 wrapper removed
+- DTS-ES extension (XCh) decoding support
+- native VP8 decoder
+- RTSP tunneling over HTTP
+- RTP depacketization of SVQ3
+- -strict inofficial replaced by -strict unofficial
+- ffplay -exitonkeydown and -exitonmousedown options added
+- native GSM / GSM MS decoder
+- RTP depacketization of QDM2
+- ANSI/ASCII art playback system
+- Lego Mindstorms RSO de/muxer
+- SubRip subtitle file muxer and demuxer
+- Chinese AVS encoding via libxavs
+- ffprobe -show_packets option added
+- RTP packetization of Theora and Vorbis
+- RTP depacketization of MP4A-LATM
+- RTP packetization and depacketization of VP8
+- hflip filter
+- Apple HTTP Live Streaming demuxer
+- a64 codec
+- MMS-HTTP support
+- G.722 ADPCM audio encoder/decoder
+- R10k video decoder
+- ocv_smooth filter
+- frei0r wrapper filter
+- change crop filter syntax to width:height:x:y
+- make the crop filter accept parametric expressions
+- make ffprobe accept AVFormatContext options
+- yadif filter
+- blackframe filter
+- Demuxer for Leitch/Harris' VR native stream format (LXF)
+- RTP depacketization of the X-QT QuickTime format
+- SAP (Session Announcement Protocol, RFC 2974) muxer and demuxer
+- cropdetect filter
+- ffmpeg -crop* options removed
+- transpose filter added
+- ffmpeg -force_key_frames option added
+- demuxer for receiving raw rtp:// URLs without an SDP description
+- single stream LATM/LOAS decoder
+- setpts filter added
+- Win64 support for optimized x86 assembly functions
+- MJPEG/AVI1 to JPEG/JFIF bitstream filter
+- ASS subtitle encoder and decoder
+- IEC 61937 encapsulation for E-AC-3, TrueHD, DTS-HD (for HDMI passthrough)
+- overlay filter added
+- rename aspect filter to setdar, and pixelaspect to setsar
+- IEC 61937 demuxer
+- Mobotix .mxg demuxer
+- frei0r source added
+- hqdn3d filter added
+- RTP depacketization of QCELP
+- FLAC parser added
+- gradfun filter added
+- AMR-WB decoder
+- replace the ocv_smooth filter with a more generic ocv filter
+- Windows Televison (WTV) demuxer
+- FFmpeg metadata format muxer and demuxer
+- SubRip (srt) subtitle decoder
+- floating-point AC-3 encoder added
+- Lagarith decoder
+- ffmpeg -copytb option added
+- IVF muxer added
+- Wing Commander IV movies decoder added
+- movie source added
+- Bink version 'b' audio and video decoder
+- Bitmap Brothers JV playback system
+- Apple HTTP Live Streaming protocol handler
+- sndio support for playback and record
+- Linux framebuffer input device added
+- Chronomaster DFA decoder
+- Mobotix MxPEG decoder
+- AAC encoding via libvo-aacenc
+- AMR-WB encoding via libvo-amrwbenc
+- xWMA demuxer
+- fieldorder video filter added
+
+
+version 0.6:
+
+- PB-frame decoding for H.263
+- deprecated vhook subsystem removed
+- deprecated old scaler removed
+- VQF demuxer
+- Alpha channel scaler
+- PCX encoder
+- RTP packetization of H.263
+- RTP packetization of AMR
+- RTP depacketization of Vorbis
+- CorePNG decoding support
+- Cook multichannel decoding support
+- introduced avlanguage helpers in libavformat
+- 8088flex TMV demuxer and decoder
+- per-stream language-tags extraction in asfdec
+- V210 decoder and encoder
+- remaining GPL parts in AC-3 decoder converted to LGPL
+- QCP demuxer
+- SoX native format muxer and demuxer
+- AMR-NB decoding/encoding, AMR-WB decoding via OpenCORE libraries
+- DPX image decoder
+- Electronic Arts Madcow decoder
+- DivX (XSUB) subtitle encoder
+- nonfree libamr support for AMR-NB/WB decoding/encoding removed
+- experimental AAC encoder
+- RTP depacketization of ASF and RTSP from WMS servers
+- RTMP support in libavformat
+- noX handling for OPT_BOOL X options
+- Wave64 demuxer
+- IEC-61937 compatible Muxer
+- TwinVQ decoder
+- Bluray (PGS) subtitle decoder
+- LPCM support in MPEG-TS (HDMV RID as found on Blu-ray disks)
+- WMA Pro decoder
+- Core Audio Format demuxer
+- Atrac1 decoder
+- MD STUDIO audio demuxer
+- RF64 support in WAV demuxer
+- MPEG-4 Audio Lossless Coding (ALS) decoder
+- -formats option split into -formats, -codecs, -bsfs, and -protocols
+- IV8 demuxer
+- CDG demuxer and decoder
+- R210 decoder
+- Auravision Aura 1 and 2 decoders
+- Deluxe Paint Animation playback system
+- SIPR decoder
+- Adobe Filmstrip muxer and demuxer
+- RTP depacketization of H.263
+- Bink demuxer and audio/video decoders
+- enable symbol versioning by default for linkers that support it
+- IFF PBM/ILBM bitmap decoder
+- concat protocol
+- Indeo 5 decoder
+- RTP depacketization of AMR
+- WMA Voice decoder
+- ffprobe tool
+- AMR-NB decoder
+- RTSP muxer
+- HE-AAC v1 decoder
+- Kega Game Video (KGV1) decoder
+- VorbisComment writing for FLAC, Ogg FLAC and Ogg Speex files
+- RTP depacketization of Theora
+- HTTP Digest authentication
+- RTMP/RTMPT/RTMPS/RTMPE/RTMPTE protocol support via librtmp
+- Psygnosis YOP demuxer and video decoder
+- spectral extension support in the E-AC-3 decoder
+- unsharp video filter
+- RTP hinting in the mov/3gp/mp4 muxer
+- Dirac in Ogg demuxing
+- seek to keyframes in Ogg
+- 4:2:2 and 4:4:4 Theora decoding
+- 35% faster VP3/Theora decoding
+- faster AAC decoding
+- faster H.264 decoding
+- RealAudio 1.0 (14.4K) encoder
+
+
+version 0.5:
+
+- DV50 AKA DVCPRO50 encoder, decoder, muxer and demuxer
+- TechSmith Camtasia (TSCC) video decoder
+- IBM Ultimotion (ULTI) video decoder
+- Sierra Online audio file demuxer and decoder
+- Apple QuickDraw (qdrw) video decoder
+- Creative ADPCM audio decoder (16 bits as well as 8 bits schemes)
+- Electronic Arts Multimedia (WVE/UV2/etc.) file demuxer
+- Miro VideoXL (VIXL) video decoder
+- H.261 video encoder
+- QPEG video decoder
+- Nullsoft Video (NSV) file demuxer
+- Shorten audio decoder
+- LOCO video decoder
+- Apple Lossless Audio Codec (ALAC) decoder
+- Winnov WNV1 video decoder
+- Autodesk Animator Studio Codec (AASC) decoder
+- Indeo 2 video decoder
+- Fraps FPS1 video decoder
+- Snow video encoder/decoder
+- Sonic audio encoder/decoder
+- Vorbis audio decoder
+- Macromedia ADPCM decoder
+- Duck TrueMotion 2 video decoder
+- support for decoding FLX and DTA extensions in FLIC files
+- H.264 custom quantization matrices support
+- ffserver fixed, it should now be usable again
+- QDM2 audio decoder
+- Real Cooker audio decoder
+- TrueSpeech audio decoder
+- WMA2 audio decoder fixed, now all files should play correctly
+- RealAudio 14.4 and 28.8 decoders fixed
+- JPEG-LS decoder
+- build system improvements
+- tabs and trailing whitespace removed from the codebase
+- CamStudio video decoder
+- AIFF/AIFF-C audio format, encoding and decoding
+- ADTS AAC file reading and writing
+- Creative VOC file reading and writing
+- American Laser Games multimedia (*.mm) playback system
+- Zip Motion Blocks Video decoder
+- improved Theora/VP3 decoder
+- True Audio (TTA) decoder
+- AVS demuxer and video decoder
+- JPEG-LS encoder
+- Smacker demuxer and decoder
+- NuppelVideo/MythTV demuxer and RTjpeg decoder
+- KMVC decoder
+- MPEG-2 intra VLC support
+- MPEG-2 4:2:2 encoder
+- Flash Screen Video decoder
+- GXF demuxer
+- Chinese AVS decoder
+- GXF muxer
+- MXF demuxer
+- VC-1/WMV3/WMV9 video decoder
+- MacIntel support
+- AVISynth support
+- VMware video decoder
+- VP5 video decoder
+- VP6 video decoder
+- WavPack lossless audio decoder
+- Targa (.TGA) picture decoder
+- Vorbis audio encoder
+- Delphine Software .cin demuxer/audio and video decoder
+- Tiertex .seq demuxer/video decoder
+- MTV demuxer
+- TIFF picture encoder and decoder
+- GIF picture decoder
+- Intel Music Coder decoder
+- Zip Motion Blocks Video encoder
+- Musepack decoder
+- Flash Screen Video encoder
+- Theora encoding via libtheora
+- BMP encoder
+- WMA encoder
+- GSM-MS encoder and decoder
+- DCA decoder
+- DXA demuxer and decoder
+- DNxHD decoder
+- Gamecube movie (.THP) playback system
+- Blackfin optimizations
+- Interplay C93 demuxer and video decoder
+- Bethsoft VID demuxer and video decoder
+- CRYO APC demuxer
+- Atrac3 decoder
+- V.Flash PTX decoder
+- RoQ muxer, RoQ audio encoder
+- Renderware TXD demuxer and decoder
+- extern C declarations for C++ removed from headers
+- sws_flags command line option
+- codebook generator
+- RoQ video encoder
+- QTRLE encoder
+- OS/2 support removed and restored again
+- AC-3 decoder
+- NUT muxer
+- additional SPARC (VIS) optimizations
+- Matroska muxer
+- slice-based parallel H.264 decoding
+- Monkey's Audio demuxer and decoder
+- AMV audio and video decoder
+- DNxHD encoder
+- H.264 PAFF decoding
+- Nellymoser ASAO decoder
+- Beam Software SIFF demuxer and decoder
+- libvorbis Vorbis decoding removed in favor of native decoder
+- IntraX8 (J-Frame) subdecoder for WMV2 and VC-1
+- Ogg (Theora, Vorbis and FLAC) muxer
+- The "device" muxers and demuxers are now in a new libavdevice library
+- PC Paintbrush PCX decoder
+- Sun Rasterfile decoder
+- TechnoTrend PVA demuxer
+- Linux Media Labs MPEG-4 (LMLM4) demuxer
+- AVM2 (Flash 9) SWF muxer
+- QT variant of IMA ADPCM encoder
+- VFW grabber
+- iPod/iPhone compatible mp4 muxer
+- Mimic decoder
+- MSN TCP Webcam stream demuxer
+- RL2 demuxer / decoder
+- IFF demuxer
+- 8SVX audio decoder
+- non-recursive Makefiles
+- BFI demuxer
+- MAXIS EA XA (.xa) demuxer / decoder
+- BFI video decoder
+- OMA demuxer
+- MLP/TrueHD decoder
+- Electronic Arts CMV decoder
+- Motion Pixels Video decoder
+- Motion Pixels MVI demuxer
+- removed animated GIF decoder/demuxer
+- D-Cinema audio muxer
+- Electronic Arts TGV decoder
+- Apple Lossless Audio Codec (ALAC) encoder
+- AAC decoder
+- floating point PCM encoder/decoder
+- MXF muxer
+- DV100 AKA DVCPRO HD decoder and demuxer
+- E-AC-3 support added to AC-3 decoder
+- Nellymoser ASAO encoder
+- ASS and SSA demuxer and muxer
+- liba52 wrapper removed
+- SVQ3 watermark decoding support
+- Speex decoding via libspeex
+- Electronic Arts TGQ decoder
+- RV40 decoder
+- QCELP / PureVoice decoder
+- RV30 decoder
+- hybrid WavPack support
+- R3D REDCODE demuxer
+- ALSA support for playback and record
+- Electronic Arts TQI decoder
+- OpenJPEG based JPEG 2000 decoder
+- NC (NC4600) camera file demuxer
+- Gopher client support
+- MXF D-10 muxer
+- generic metadata API
+
+
+version 0.4.9-pre1:
+
+- DV encoder, DV muxer
+- Microsoft RLE video decoder
+- Microsoft Video-1 decoder
+- Apple Animation (RLE) decoder
+- Apple Graphics (SMC) decoder
+- Apple Video (RPZA) decoder
+- Cinepak decoder
+- Sega FILM (CPK) file demuxer
+- Westwood multimedia support (VQA & AUD files)
+- Id Quake II CIN playback support
+- 8BPS video decoder
+- FLIC playback support
+- RealVideo 2.0 (RV20) decoder
+- Duck TrueMotion v1 (DUCK) video decoder
+- Sierra VMD demuxer and video decoder
+- MSZH and ZLIB decoder support
+- SVQ1 video encoder
+- AMR-WB support
+- PPC optimizations
+- rate distortion optimal cbp support
+- rate distorted optimal ac prediction for MPEG-4
+- rate distorted optimal lambda->qp support
+- AAC encoding with libfaac
+- Sunplus JPEG codec (SP5X) support
+- use Lagrange multipler instead of QP for ratecontrol
+- Theora/VP3 decoding support
+- XA and ADX ADPCM codecs
+- export MPEG-2 active display area / pan scan
+- Add support for configuring with IBM XLC
+- floating point AAN DCT
+- initial support for zygo video (not complete)
+- RGB ffv1 support
+- new audio/video parser API
+- av_log() system
+- av_read_frame() and av_seek_frame() support
+- missing last frame fixes
+- seek by mouse in ffplay
+- noise reduction of DCT coefficients
+- H.263 OBMC & 4MV support
+- H.263 alternative inter vlc support
+- H.263 loop filter
+- H.263 slice structured mode
+- interlaced DCT support for MPEG-2 encoding
+- stuffing to stay above min_bitrate
+- MB type & QP visualization
+- frame stepping for ffplay
+- interlaced motion estimation
+- alternate scantable support
+- SVCD scan offset support
+- closed GOP support
+- SSE2 FDCT
+- quantizer noise shaping
+- G.726 ADPCM audio codec
+- MS ADPCM encoding
+- multithreaded/SMP motion estimation
+- multithreaded/SMP encoding for MPEG-1/MPEG-2/MPEG-4/H.263
+- multithreaded/SMP decoding for MPEG-2
+- FLAC decoder
+- Metrowerks CodeWarrior suppport
+- H.263+ custom pcf support
+- nicer output for 'ffmpeg -formats'
+- Matroska demuxer
+- SGI image format, encoding and decoding
+- H.264 loop filter support
+- H.264 CABAC support
+- nicer looking arrows for the motion vector visualization
+- improved VCD support
+- audio timestamp drift compensation
+- MPEG-2 YUV 422/444 support
+- polyphase kaiser windowed sinc and blackman nuttall windowed sinc audio resample
+- better image scaling
+- H.261 support
+- correctly interleave packets during encoding
+- VIS optimized motion compensation
+- intra_dc_precision>0 encoding support
+- support reuse of motion vectors/MB types/field select values of the source video
+- more accurate deblock filter
+- padding support
+- many optimizations and bugfixes
+- FunCom ISS audio file demuxer and according ADPCM decoding
+
+
+version 0.4.8:
+
+- MPEG-2 video encoding (Michael)
+- Id RoQ playback subsystem (Mike Melanson and Tim Ferguson)
+- Wing Commander III Movie (.mve) file playback subsystem (Mike Melanson
+  and Mario Brito)
+- Xan DPCM audio decoder (Mario Brito)
+- Interplay MVE playback subsystem (Mike Melanson)
+- Duck DK3 and DK4 ADPCM audio decoders (Mike Melanson)
+
+
+version 0.4.7:
+
+- RealAudio 1.0 (14_4) and 2.0 (28_8) native decoders. Author unknown, code from mplayerhq
+  (originally from public domain player for Amiga at http://www.honeypot.net/audio)
+- current version now also compiles with older GCC (Fabrice)
+- 4X multimedia playback system including 4xm file demuxer (Mike
+  Melanson), and 4X video and audio codecs (Michael)
+- Creative YUV (CYUV) decoder (Mike Melanson)
+- FFV1 codec (our very simple lossless intra only codec, compresses much better
+  than HuffYUV) (Michael)
+- ASV1 (Asus), H.264, Intel indeo3 codecs have been added (various)
+- tiny PNG encoder and decoder, tiny GIF decoder, PAM decoder (PPM with
+  alpha support), JPEG YUV colorspace support. (Fabrice Bellard)
+- ffplay has been replaced with a newer version which uses SDL (optionally)
+  for multiplatform support (Fabrice)
+- Sorenson Version 3 codec (SVQ3) support has been added (decoding only) - donated
+  by anonymous
+- AMR format has been added (Johannes Carlsson)
+- 3GP support has been added (Johannes Carlsson)
+- VP3 codec has been added (Mike Melanson)
+- more MPEG-1/2 fixes
+- better multiplatform support, MS Visual Studio fixes (various)
+- AltiVec optimizations (Magnus Damn and others)
+- SH4 processor support has been added (BERO)
+- new public interfaces (avcodec_get_pix_fmt) (Roman Shaposhnick)
+- VOB streaming support (Brian Foley)
+- better MP3 autodetection (Andriy Rysin)
+- qpel encoding (Michael)
+- 4mv+b frames encoding finally fixed (Michael)
+- chroma ME (Michael)
+- 5 comparison functions for ME (Michael)
+- B-frame encoding speedup (Michael)
+- WMV2 codec (unfinished - Michael)
+- user specified diamond size for EPZS (Michael)
+- Playstation STR playback subsystem, still experimental (Mike and Michael)
+- ASV2 codec (Michael)
+- CLJR decoder (Alex)
+
+.. And lots more new enhancements and fixes.
+
+
+version 0.4.6:
+
+- completely new integer only MPEG audio layer 1/2/3 decoder rewritten
+  from scratch
+- Recoded DCT and motion vector search with gcc (no longer depends on nasm)
+- fix quantization bug in AC3 encoder
+- added PCM codecs and format. Corrected WAV/AVI/ASF PCM issues
+- added prototype ffplay program
+- added GOB header parsing on H.263/H.263+ decoder (Juanjo)
+- bug fix on MCBPC tables of H.263 (Juanjo)
+- bug fix on DC coefficients of H.263 (Juanjo)
+- added Advanced Prediction Mode on H.263/H.263+ decoder (Juanjo)
+- now we can decode H.263 streams found in QuickTime files (Juanjo)
+- now we can decode H.263 streams found in VIVO v1 files(Juanjo)
+- preliminary RTP "friendly" mode for H.263/H.263+ coding. (Juanjo)
+- added GOB header for H.263/H.263+ coding on RTP mode (Juanjo)
+- now H.263 picture size is returned on the first decoded frame (Juanjo)
+- added first regression tests
+- added MPEG-2 TS demuxer
+- new demux API for libav
+- more accurate and faster IDCT (Michael)
+- faster and entropy-controlled motion search (Michael)
+- two pass video encoding (Michael)
+- new video rate control (Michael)
+- added MSMPEG4V1, MSMPEGV2 and WMV1 support (Michael)
+- great performance improvement of video encoders and decoders (Michael)
+- new and faster bit readers and vlc parsers (Michael)
+- high quality encoding mode: tries all macroblock/VLC types (Michael)
+- added DV video decoder
+- preliminary RTP/RTSP support in ffserver and libavformat
+- H.263+ AIC decoding/encoding support (Juanjo)
+- VCD MPEG-PS mode (Juanjo)
+- PSNR stuff (Juanjo)
+- simple stats output (Juanjo)
+- 16-bit and 15-bit RGB/BGR/GBR support (Bisqwit)
+
+
+version 0.4.5:
+
+- some header fixes (Zdenek Kabelac <kabi at informatics.muni.cz>)
+- many MMX optimizations (Nick Kurshev <nickols_k at mail.ru>)
+- added configure system (actually a small shell script)
+- added MPEG audio layer 1/2/3 decoding using LGPL'ed mpglib by
+  Michael Hipp (temporary solution - waiting for integer only
+  decoder)
+- fixed VIDIOCSYNC interrupt
+- added Intel H.263 decoding support ('I263' AVI fourCC)
+- added Real Video 1.0 decoding (needs further testing)
+- simplified image formats again. Added PGM format (=grey
+  pgm). Renamed old PGM to PGMYUV.
+- fixed msmpeg4 slice issues (tell me if you still find problems)
+- fixed OpenDivX bugs with newer versions (added VOL header decoding)
+- added support for MPlayer interface
+- added macroblock skip optimization
+- added MJPEG decoder
+- added mmx/mmxext IDCT from libmpeg2
+- added pgmyuvpipe, ppm, and ppm_pipe formats (original patch by Celer
+  <celer at shell.scrypt.net>)
+- added pixel format conversion layer (e.g. for MJPEG or PPM)
+- added deinterlacing option
+- MPEG-1/2 fixes
+- MPEG-4 vol header fixes (Jonathan Marsden <snmjbm at pacbell.net>)
+- ARM optimizations (Lionel Ulmer <lionel.ulmer at free.fr>).
+- Windows porting of file converter
+- added MJPEG raw format (input/ouput)
+- added JPEG image format support (input/output)
+
+
+version 0.4.4:
+
+- fixed some std header definitions (Bjorn Lindgren
+  <bjorn.e.lindgren at telia.com>).
+- added MPEG demuxer (MPEG-1 and 2 compatible).
+- added ASF demuxer
+- added prototype RM demuxer
+- added AC3 decoding (done with libac3 by Aaron Holtzman)
+- added decoding codec parameter guessing (.e.g. for MPEG, because the
+  header does not include them)
+- fixed header generation in MPEG-1, AVI and ASF muxer: wmplayer can now
+  play them (only tested video)
+- fixed H.263 white bug
+- fixed phase rounding in img resample filter
+- add MMX code for polyphase img resample filter
+- added CPU autodetection
+- added generic title/author/copyright/comment string handling (ASF and RM
+  use them)
+- added SWF demux to extract MP3 track (not usable yet because no MP3
+  decoder)
+- added fractional frame rate support
+- codecs are no longer searched by read_header() (should fix ffserver
+  segfault)
+
+
+version 0.4.3:
+
+- BGR24 patch (initial patch by Jeroen Vreeken <pe1rxq at amsat.org>)
+- fixed raw yuv output
+- added motion rounding support in MPEG-4
+- fixed motion bug rounding in MSMPEG4
+- added B-frame handling in video core
+- added full MPEG-1 decoding support
+- added partial (frame only) MPEG-2 support
+- changed the FOURCC code for H.263 to "U263" to be able to see the
+  +AVI/H.263 file with the UB Video H.263+ decoder. MPlayer works with
+  this +codec ;) (JuanJo).
+- Halfpel motion estimation after MB type selection (JuanJo)
+- added pgm and .Y.U.V output format
+- suppressed 'img:' protocol. Simply use: /tmp/test%d.[pgm|Y] as input or
+  output.
+- added pgmpipe I/O format (original patch from Martin Aumueller
+  <lists at reserv.at>, but changed completely since we use a format
+  instead of a protocol)
+
+
+version 0.4.2:
+
+- added H.263/MPEG-4/MSMPEG4 decoding support. MPEG-4 decoding support
+  (for OpenDivX) is almost complete: 8x8 MVs and rounding are
+  missing. MSMPEG4 support is complete.
+- added prototype MPEG-1 decoder. Only I- and P-frames handled yet (it
+  can decode ffmpeg MPEGs :-)).
+- added libavcodec API documentation (see apiexample.c).
+- fixed image polyphase bug (the bottom of some images could be
+  greenish)
+- added support for non clipped motion vectors (decoding only)
+  and image sizes non-multiple of 16
+- added support for AC prediction (decoding only)
+- added file overwrite confirmation (can be disabled with -y)
+- added custom size picture to H.263 using H.263+ (Juanjo)
+
+
+version 0.4.1:
+
+- added MSMPEG4 (aka DivX) compatible encoder. Changed default codec
+  of AVI and ASF to DIV3.
+- added -me option to set motion estimation method
+  (default=log). suppressed redundant -hq option.
+- added options -acodec and -vcodec to force a given codec (useful for
+  AVI for example)
+- fixed -an option
+- improved dct_quantize speed
+- factorized some motion estimation code
+
+
+version 0.4.0:
+
+- removing grab code from ffserver and moved it to ffmpeg. Added
+  multistream support to ffmpeg.
+- added timeshifting support for live feeds (option ?date=xxx in the
+  URL)
+- added high quality image resize code with polyphase filter (need
+  mmx/see optimization). Enable multiple image size support in ffserver.
+- added multi live feed support in ffserver
+- suppressed master feature from ffserver (it should be done with an
+  external program which opens the .ffm url and writes it to another
+  ffserver)
+- added preliminary support for video stream parsing (WAV and AVI half
+  done). Added proper support for audio/video file conversion in
+  ffmpeg.
+- added preliminary support for video file sending from ffserver
+- redesigning I/O subsystem: now using URL based input and output
+  (see avio.h)
+- added WAV format support
+- added "tty user interface" to ffmpeg to stop grabbing gracefully
+- added MMX/SSE optimizations to SAD (Sums of Absolutes Differences)
+  (Juan J. Sierralta P. a.k.a. "Juanjo" <juanjo at atmlab.utfsm.cl>)
+- added MMX DCT from mpeg2_movie 1.5 (Juanjo)
+- added new motion estimation algorithms, log and phods (Juanjo)
+- changed directories: libav for format handling, libavcodec for
+  codecs
+
+
+version 0.3.4:
+
+- added stereo in MPEG audio encoder
+
+
+version 0.3.3:
+
+- added 'high quality' mode which use motion vectors. It can be used in
+  real time at low resolution.
+- fixed rounding problems which caused quality problems at high
+  bitrates and large GOP size
+
+
+version 0.3.2: small fixes
+
+- ASF fixes
+- put_seek bug fix
+
+
+version 0.3.1: added avi/divx support
+
+- added AVI support
+- added MPEG-4 codec compatible with OpenDivX. It is based on the H.263 codec
+- added sound for flash format (not tested)
+
+
+version 0.3: initial public release
diff --git a/libavcodec/aacsbr.c b/libavcodec/aacsbr.c
index 10b8daf280..866482aac3 100644
--- a/libavcodec/aacsbr.c
+++ b/libavcodec/aacsbr.c
@@ -1185,7 +1185,7 @@ static void sbr_qmf_synthesis(DSPContext *dsp, FFTContext *mdct,
     const float *sbr_qmf_window = div ? sbr_qmf_window_ds : sbr_qmf_window_us;
     float *v;
     for (i = 0; i < 32; i++) {
-        if (*v_off == 0) {
+        if (*v_off < 128 >> div) {
             int saved_samples = (1280 - 128) >> div;
             memcpy(&v0[SBR_SYNTHESIS_BUF_SIZE - saved_samples], v0, saved_samples * sizeof(float));
             *v_off = SBR_SYNTHESIS_BUF_SIZE - saved_samples - (128 >> div);
diff --git a/libavcodec/atrac3.c b/libavcodec/atrac3.c
index 20ab75dfd7..f16630450d 100644
--- a/libavcodec/atrac3.c
+++ b/libavcodec/atrac3.c
@@ -395,6 +395,8 @@ static int decodeTonalComponents (GetBitContext *gb, tonal_component *pComponent
 
             for (k=0; k<coded_components; k++) {
                 sfIndx = get_bits(gb,6);
+                if(component_count>=64)
+                    return AVERROR_INVALIDDATA;
                 pComponent[component_count].pos = j * 64 + (get_bits(gb,6));
                 max_coded_values = 1024 - pComponent[component_count].pos;
                 coded_values = coded_values_per_component + 1;
diff --git a/libavcodec/h264_ps.c b/libavcodec/h264_ps.c
index 89e2502e0d..680db1e5a6 100644
--- a/libavcodec/h264_ps.c
+++ b/libavcodec/h264_ps.c
@@ -342,6 +342,10 @@ int ff_h264_decode_seq_parameter_set(H264Context *h){
 
     if(sps->profile_idc >= 100){ //high profile
         sps->chroma_format_idc= get_ue_golomb_31(&s->gb);
+        if (sps->chroma_format_idc > 3U) {
+            av_log(h->s.avctx, AV_LOG_ERROR, "chroma_format_idc %d is illegal\n", sps->chroma_format_idc);
+            goto fail;
+        }
         if(sps->chroma_format_idc == 3)
             sps->residual_color_transform_flag = get_bits1(&s->gb);
         sps->bit_depth_luma   = get_ue_golomb(&s->gb) + 8;
diff --git a/libavcodec/j2k_dwt.c b/libavcodec/j2k_dwt.c
index ab7a1ab757..48aa33735e 100644
--- a/libavcodec/j2k_dwt.c
+++ b/libavcodec/j2k_dwt.c
@@ -321,7 +321,7 @@ int ff_j2k_dwt_init(DWTContext *s, uint16_t border[2][2], int decomp_levels, int
     int i, j, lev = decomp_levels, maxlen,
         b[2][2];
 
-    if (decomp_levels >= FF_DWT_MAX_DECLVLS)
+    if ((unsigned)decomp_levels >= FF_DWT_MAX_DECLVLS)
         return AVERROR_INVALIDDATA;
     s->ndeclevels = decomp_levels;
     s->type = type;
diff --git a/libavcodec/j2kdec.c b/libavcodec/j2kdec.c
index 96b4f64098..3315a835d7 100644
--- a/libavcodec/j2kdec.c
+++ b/libavcodec/j2kdec.c
@@ -359,7 +359,7 @@ static int get_qcx(J2kDecoderContext *s, int n, J2kQuantStyle *q)
 
     if (q->quantsty == J2K_QSTY_NONE){
         n -= 3;
-        if (s->buf_end - s->buf < n)
+        if (s->buf_end - s->buf < n || 32*3 < n)
             return AVERROR(EINVAL);
         for (i = 0; i < n; i++)
             q->expn[i] = bytestream_get_byte(&s->buf) >> 3;
@@ -376,7 +376,7 @@ static int get_qcx(J2kDecoderContext *s, int n, J2kQuantStyle *q)
         }
     } else{
         n = (n - 3) >> 1;
-        if (s->buf_end - s->buf < n)
+        if (s->buf_end - s->buf < n || 32*3 < n)
             return AVERROR(EINVAL);
         for (i = 0; i < n; i++){
             x = bytestream_get_be16(&s->buf);
@@ -421,6 +421,10 @@ static uint8_t get_sot(J2kDecoderContext *s)
         return AVERROR(EINVAL);
 
     s->curtileno = bytestream_get_be16(&s->buf); ///< Isot
+    if((unsigned)s->curtileno >= s->numXtiles * s->numYtiles){
+        s->curtileno=0;
+        return AVERROR(EINVAL);
+    }
 
     s->buf += 4; ///< Psot (ignored)
 
diff --git a/libavcodec/shorten.c b/libavcodec/shorten.c
index 0b9d420d86..621281fc75 100644
--- a/libavcodec/shorten.c
+++ b/libavcodec/shorten.c
@@ -81,6 +81,7 @@ typedef struct ShortenContext {
     int channels;
 
     int32_t *decoded[MAX_CHANNELS];
+    int32_t *decoded_base[MAX_CHANNELS];
     int32_t *offset[MAX_CHANNELS];
     int *coeffs;
     uint8_t *bitstream;
@@ -130,13 +131,13 @@ static int allocate_buffers(ShortenContext *s)
             return AVERROR(ENOMEM);
         s->offset[chan] = tmp_ptr;
 
-        tmp_ptr = av_realloc(s->decoded[chan], sizeof(int32_t)*(s->blocksize + s->nwrap));
+        tmp_ptr = av_realloc(s->decoded_base[chan], sizeof(int32_t)*(s->blocksize + s->nwrap));
         if (!tmp_ptr)
             return AVERROR(ENOMEM);
-        s->decoded[chan] = tmp_ptr;
+        s->decoded_base[chan] = tmp_ptr;
         for (i=0; i<s->nwrap; i++)
-            s->decoded[chan][i] = 0;
-        s->decoded[chan] += s->nwrap;
+            s->decoded_base[chan][i] = 0;
+        s->decoded[chan] = s->decoded_base[chan] + s->nwrap;
     }
 
     coeffs = av_realloc(s->coeffs, s->nwrap * sizeof(*s->coeffs));
@@ -548,8 +549,8 @@ static av_cold int shorten_decode_close(AVCodecContext *avctx)
     int i;
 
     for (i = 0; i < s->channels; i++) {
-        s->decoded[i] -= s->nwrap;
-        av_freep(&s->decoded[i]);
+        s->decoded[i] = NULL;
+        av_freep(&s->decoded_base[i]);
         av_freep(&s->offset[i]);
     }
     av_freep(&s->bitstream);
diff --git a/libavcodec/vorbisdec.c b/libavcodec/vorbisdec.c
index 8f16d3a5a5..ddb9bd6d5d 100644
--- a/libavcodec/vorbisdec.c
+++ b/libavcodec/vorbisdec.c
@@ -660,7 +660,7 @@ static int vorbis_parse_setup_hdr_residues(vorbis_context *vc)
         res_setup->partition_size = get_bits(gb, 24) + 1;
         /* Validations to prevent a buffer overflow later. */
         if (res_setup->begin>res_setup->end ||
-            res_setup->end > vc->avccontext->channels * vc->blocksize[1] / 2 ||
+            res_setup->end > (res_setup->type == 2 ? vc->avccontext->channels : 1) * vc->blocksize[1] / 2 ||
             (res_setup->end-res_setup->begin) / res_setup->partition_size > V_MAX_PARTITIONS) {
             av_log(vc->avccontext, AV_LOG_ERROR,
                    "partition out of bounds: type, begin, end, size, blocksize: %"PRIu16", %"PRIu32", %"PRIu32", %u, %"PRIu32"\n",
@@ -1269,6 +1269,7 @@ static av_always_inline int vorbis_residue_decode_internal(vorbis_context *vc,
                                                            uint8_t *do_not_decode,
                                                            float *vec,
                                                            unsigned vlen,
+                                                           unsigned ch_left,
                                                            int vr_type)
 {
     GetBitContext *gb = &vc->gb;
@@ -1276,6 +1277,7 @@ static av_always_inline int vorbis_residue_decode_internal(vorbis_context *vc,
     unsigned ptns_to_read = vr->ptns_to_read;
     uint8_t *classifs = vr->classifs;
     unsigned pass, ch_used, i, j, k, l;
+    unsigned max_output = (ch - 1) * vlen;
 
     if (vr_type == 2) {
         for (j = 1; j < ch; ++j)
@@ -1283,8 +1285,15 @@ static av_always_inline int vorbis_residue_decode_internal(vorbis_context *vc,
         if (do_not_decode[0])
             return 0;
         ch_used = 1;
+        max_output += vr->end / ch;
     } else {
         ch_used = ch;
+        max_output += vr->end;
+    }
+
+    if (max_output > ch_left * vlen) {
+        av_log(vc->avccontext, AV_LOG_ERROR, "Insufficient output buffer\n");
+        return -1;
     }
 
     av_dlog(NULL, " residue type 0/1/2 decode begin, ch: %d  cpc %d  \n", ch, c_p_c);
@@ -1411,14 +1420,15 @@ static av_always_inline int vorbis_residue_decode_internal(vorbis_context *vc,
 static inline int vorbis_residue_decode(vorbis_context *vc, vorbis_residue *vr,
                                         unsigned ch,
                                         uint8_t *do_not_decode,
-                                        float *vec, unsigned vlen)
+                                        float *vec, unsigned vlen,
+                                        unsigned ch_left)
 {
     if (vr->type == 2)
-        return vorbis_residue_decode_internal(vc, vr, ch, do_not_decode, vec, vlen, 2);
+        return vorbis_residue_decode_internal(vc, vr, ch, do_not_decode, vec, vlen, ch_left, 2);
     else if (vr->type == 1)
-        return vorbis_residue_decode_internal(vc, vr, ch, do_not_decode, vec, vlen, 1);
+        return vorbis_residue_decode_internal(vc, vr, ch, do_not_decode, vec, vlen, ch_left, 1);
     else if (vr->type == 0)
-        return vorbis_residue_decode_internal(vc, vr, ch, do_not_decode, vec, vlen, 0);
+        return vorbis_residue_decode_internal(vc, vr, ch, do_not_decode, vec, vlen, ch_left, 0);
     else {
         av_log(vc->avccontext, AV_LOG_ERROR, " Invalid residue type while residue decode?! \n");
         return -1;
@@ -1466,6 +1476,8 @@ static int vorbis_parse_audio_packet(vorbis_context *vc)
     uint8_t res_chan[255];
     unsigned res_num = 0;
     int retlen  = 0;
+    unsigned ch_left = vc->audio_channels;
+    unsigned vlen;
 
     if (get_bits1(gb)) {
         av_log(vc->avccontext, AV_LOG_ERROR, "Not a Vorbis I audio packet.\n");
@@ -1485,11 +1497,12 @@ static int vorbis_parse_audio_packet(vorbis_context *vc)
 
     blockflag = vc->modes[mode_number].blockflag;
     blocksize = vc->blocksize[blockflag];
+    vlen = blocksize / 2;
     if (blockflag)
         skip_bits(gb, 2); // previous_window, next_window
 
-    memset(ch_res_ptr,   0, sizeof(float) * vc->audio_channels * blocksize / 2); //FIXME can this be removed ?
-    memset(ch_floor_ptr, 0, sizeof(float) * vc->audio_channels * blocksize / 2); //FIXME can this be removed ?
+    memset(ch_res_ptr,   0, sizeof(float) * vc->audio_channels * vlen); //FIXME can this be removed ?
+    memset(ch_floor_ptr, 0, sizeof(float) * vc->audio_channels * vlen); //FIXME can this be removed ?
 
 // Decode floor
 
@@ -1509,7 +1522,7 @@ static int vorbis_parse_audio_packet(vorbis_context *vc)
             return -1;
         }
         no_residue[i] = ret;
-        ch_floor_ptr += blocksize / 2;
+        ch_floor_ptr += vlen;
     }
 
 // Nonzero vector propagate
@@ -1526,6 +1539,7 @@ static int vorbis_parse_audio_packet(vorbis_context *vc)
     for (i = 0; i < mapping->submaps; ++i) {
         vorbis_residue *residue;
         unsigned ch = 0;
+        int ret;
 
         for (j = 0; j < vc->audio_channels; ++j) {
             if ((mapping->submaps == 1) || (i == mapping->mux[j])) {
@@ -1540,9 +1554,18 @@ static int vorbis_parse_audio_packet(vorbis_context *vc)
             }
         }
         residue = &vc->residues[mapping->submap_residue[i]];
-        vorbis_residue_decode(vc, residue, ch, do_not_decode, ch_res_ptr, blocksize/2);
+        if (ch_left < ch) {
+            av_log(vc->avccontext, AV_LOG_ERROR, "Too many channels in vorbis_floor_decode.\n");
+            return -1;
+        }
+        if (ch) {
+            ret = vorbis_residue_decode(vc, residue, ch, do_not_decode, ch_res_ptr, vlen, ch_left);
+            if (ret < 0)
+                return ret;
+        }
 
-        ch_res_ptr += ch * blocksize / 2;
+        ch_res_ptr += ch * vlen;
+        ch_left -= ch;
     }
 
 // Inverse coupling
diff --git a/libavcodec/ws-snd1.c b/libavcodec/ws-snd1.c
index f92c3531e0..c28d1a8d6a 100644
--- a/libavcodec/ws-snd1.c
+++ b/libavcodec/ws-snd1.c
@@ -100,8 +100,8 @@ static int ws_snd_decode_frame(AVCodecContext *avctx,
 
         /* make sure we don't write more than out_size samples */
         switch (code) {
-        case 0:  smp = 4;                              break;
-        case 1:  smp = 2;                              break;
+        case 0:  smp = 4*(count+1);                    break;
+        case 1:  smp = 2*(count+1);                    break;
         case 2:  smp = (count & 0x20) ? 1 : count + 1; break;
         default: smp = count + 1;                      break;
         }
diff --git a/libavfilter/avfilter.c b/libavfilter/avfilter.c
index a57677c0e4..d2b4986cef 100644
--- a/libavfilter/avfilter.c
+++ b/libavfilter/avfilter.c
@@ -614,7 +614,7 @@ void avfilter_filter_samples(AVFilterLink *link, AVFilterBufferRef *samplesref)
         link->cur_buf->audio->sample_rate = samplesref->audio->sample_rate;
 
         /* Copy actual data into new samples buffer */
-        for (i = 0; samplesref->data[i]; i++)
+        for (i = 0; samplesref->data[i] && i < 8; i++)
             memcpy(link->cur_buf->data[i], samplesref->data[i], samplesref->linesize[0]);
 
         avfilter_unref_buffer(samplesref);
diff --git a/libswscale/x86/swscale_template.c b/libswscale/x86/swscale_template.c
index 25399fadef..ec1b15f118 100644
--- a/libswscale/x86/swscale_template.c
+++ b/libswscale/x86/swscale_template.c
@@ -2238,12 +2238,24 @@ static void RENAME(hyscale_fast)(SwsContext *c, int16_t *dst,
     void    *mmx2FilterCode= c->lumMmx2FilterCode;
     int i;
 #if defined(PIC)
-    DECLARE_ALIGNED(8, uint64_t, ebxsave);
+    uint64_t ebxsave;
+#endif
+#if ARCH_X86_64
+    uint64_t retsave;
 #endif
 
     __asm__ volatile(
 #if defined(PIC)
         "mov               %%"REG_b", %5        \n\t"
+#if ARCH_X86_64
+        "mov               -8(%%rsp), %%"REG_a" \n\t"
+        "mov               %%"REG_a", %6        \n\t"
+#endif
+#else
+#if ARCH_X86_64
+        "mov               -8(%%rsp), %%"REG_a" \n\t"
+        "mov               %%"REG_a", %5        \n\t"
+#endif
 #endif
         "pxor                  %%mm7, %%mm7     \n\t"
         "mov                      %0, %%"REG_c" \n\t"
@@ -2285,12 +2297,24 @@ static void RENAME(hyscale_fast)(SwsContext *c, int16_t *dst,
 
 #if defined(PIC)
         "mov                      %5, %%"REG_b" \n\t"
+#if ARCH_X86_64
+        "mov                      %6, %%"REG_a" \n\t"
+        "mov               %%"REG_a", -8(%%rsp) \n\t"
+#endif
+#else
+#if ARCH_X86_64
+        "mov                      %5, %%"REG_a" \n\t"
+        "mov               %%"REG_a", -8(%%rsp) \n\t"
+#endif
 #endif
         :: "m" (src), "m" (dst), "m" (filter), "m" (filterPos),
            "m" (mmx2FilterCode)
 #if defined(PIC)
           ,"m" (ebxsave)
 #endif
+#if ARCH_X86_64
+          ,"m"(retsave)
+#endif
         : "%"REG_a, "%"REG_c, "%"REG_d, "%"REG_S, "%"REG_D
 #if !defined(PIC)
          ,"%"REG_b
@@ -2312,10 +2336,22 @@ static void RENAME(hcscale_fast)(SwsContext *c, int16_t *dst1, int16_t *dst2,
 #if defined(PIC)
     DECLARE_ALIGNED(8, uint64_t, ebxsave);
 #endif
+#if ARCH_X86_64
+    DECLARE_ALIGNED(8, uint64_t, retsave);
+#endif
 
     __asm__ volatile(
 #if defined(PIC)
         "mov          %%"REG_b", %7         \n\t"
+#if ARCH_X86_64
+        "mov          -8(%%rsp), %%"REG_a"  \n\t"
+        "mov          %%"REG_a", %8         \n\t"
+#endif
+#else
+#if ARCH_X86_64
+        "mov          -8(%%rsp), %%"REG_a"  \n\t"
+        "mov          %%"REG_a", %7         \n\t"
+#endif
 #endif
         "pxor             %%mm7, %%mm7      \n\t"
         "mov                 %0, %%"REG_c"  \n\t"
@@ -2345,12 +2381,24 @@ static void RENAME(hcscale_fast)(SwsContext *c, int16_t *dst1, int16_t *dst2,
 
 #if defined(PIC)
         "mov %7, %%"REG_b"    \n\t"
+#if ARCH_X86_64
+        "mov                 %8, %%"REG_a"  \n\t"
+        "mov          %%"REG_a", -8(%%rsp)  \n\t"
+#endif
+#else
+#if ARCH_X86_64
+        "mov                 %7, %%"REG_a"  \n\t"
+        "mov          %%"REG_a", -8(%%rsp)  \n\t"
+#endif
 #endif
         :: "m" (src1), "m" (dst1), "m" (filter), "m" (filterPos),
            "m" (mmx2FilterCode), "m" (src2), "m"(dst2)
 #if defined(PIC)
           ,"m" (ebxsave)
 #endif
+#if ARCH_X86_64
+          ,"m"(retsave)
+#endif
         : "%"REG_a, "%"REG_c, "%"REG_d, "%"REG_S, "%"REG_D
 #if !defined(PIC)
          ,"%"REG_b