Originally committed as revision 20115 to svn://svn.ffmpeg.org/ffmpeg/trunk

mode allowed for there is no code to decode it (yet).

Originally committed as revision 20114 to svn://svn.ffmpeg.org/ffmpeg/trunk

Huffman-compressed JPEG files, so make it decode those files
(samples are welcome).
This fixes issue 1420

Originally committed as revision 20113 to svn://svn.ffmpeg.org/ffmpeg/trunk

This fixes issue 1415

Originally committed as revision 20112 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 20111 to svn://svn.ffmpeg.org/ffmpeg/trunk

Fixes Issue 1426.

Originally committed as revision 20110 to svn://svn.ffmpeg.org/ffmpeg/trunk

Based on a patch by j f (gonzaz-z # operamail (d) com)

Originally committed as revision 20109 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 20108 to svn://svn.ffmpeg.org/ffmpeg/trunk

ebx_available if both "b" constraint and exb clobber code compile.
This is to fix compilation with Mac OSX 10.5.8's gcc 4.0.1 in 64
bit mode (-m64) which only fails for ebx/rbx clobbers but not
"b" constraints - this is probably a compiler bug.

Originally committed as revision 20107 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 20106 to svn://svn.ffmpeg.org/ffmpeg/trunk

This makes it possible to use check_asm in conditionals.

Originally committed as revision 20105 to svn://svn.ffmpeg.org/ffmpeg/trunk

patch by Anne-Laure de Smit, annelaure.desmit gmail com

Originally committed as revision 20104 to svn://svn.ffmpeg.org/ffmpeg/trunk

flush_put_bits ensures it is divisible 8.

Originally committed as revision 20103 to svn://svn.ffmpeg.org/ffmpeg/trunk

become part of the string.

Originally committed as revision 20102 to svn://svn.ffmpeg.org/ffmpeg/trunk

The test using clobber incorrectly indicates ebx to be available with the
default compilers on OpenBSD 4.5 and Haiku.

Originally committed as revision 20101 to svn://svn.ffmpeg.org/ffmpeg/trunk

Fixes issue 975: high db peak levels when encoding mp2

Original patch by Lasse Reinhold, lassemikkelreinhold hotmail

Originally committed as revision 20100 to svn://svn.ffmpeg.org/ffmpeg/trunk

might end up in an endless loop where the same chunk is parsed over and over.
Fixes a hang near the end for http://samples.mplayerhq.hu/A-codecs/musepack/sv8/sv8-tags.mpc

Originally committed as revision 20099 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 20098 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 20097 to svn://svn.ffmpeg.org/ffmpeg/trunk

with put_bits.

Originally committed as revision 20096 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 20095 to svn://svn.ffmpeg.org/ffmpeg/trunk

#if !HAVE_EBX_AVAILABLE, since all it does is avoid using ebx.

Originally committed as revision 20094 to svn://svn.ffmpeg.org/ffmpeg/trunk

if no header was found the parser was not initialized and thus will
crash when trying to use it.

Originally committed as revision 20093 to svn://svn.ffmpeg.org/ffmpeg/trunk

must be valid UTF-8.

Originally committed as revision 20092 to svn://svn.ffmpeg.org/ffmpeg/trunk

monochrome, so initialize picture before decoding.
This fixes decoding monochrome files produced by lavc TIFF encoder.

Originally committed as revision 20091 to svn://svn.ffmpeg.org/ffmpeg/trunk

about 1.8% faster on my system.

Originally committed as revision 20090 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 20089 to svn://svn.ffmpeg.org/ffmpeg/trunk

tests, as the use of huffyuv may cause the auto-insertion of a scale
filter, thus masking the mis-behavior of some filter.

Originally committed as revision 20088 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 20087 to svn://svn.ffmpeg.org/ffmpeg/trunk

patch by Michael Kostylev, michael.kostylev gmail com

Originally committed as revision 20086 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 20085 to svn://svn.ffmpeg.org/ffmpeg/trunk

The specification does not say which value to use for unused
parts, so fill all unused bytes with 0xff, which is consistent
with what DV usually uses for reserved or unused parts.

Originally committed as revision 20084 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 20083 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 20082 to svn://svn.ffmpeg.org/ffmpeg/trunk

The count can be a non-multiple of 4 after all.

Originally committed as revision 20081 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 20080 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 20079 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 20078 to svn://svn.ffmpeg.org/ffmpeg/trunk

GCC does stupid things if these assignments are done using floats
directly, so fill the runs using integer operations instead.  Also
unroll the loops since the length is always a multiple of 4.

Originally committed as revision 20077 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 20076 to svn://svn.ffmpeg.org/ffmpeg/trunk