this is 10-20cpu cycles faster on duron (whole is about 50-60 cpu cylses)
I wonder why gcc isnt doing this on its own ...

Originally committed as revision 23097 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 23096 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 23095 to svn://svn.ffmpeg.org/ffmpeg/trunk

Most coder functions read it. Carting this around in the context may be
suboptimal; a refactor should be considered.

Originally committed as revision 23094 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 23093 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 23092 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 23091 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 23090 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 23089 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 23088 to svn://svn.ffmpeg.org/ffmpeg/trunk

The latter does not exist and thus compilation fails.
patch by Janne Grunau, janne-ffmpeg jannau.net

Originally committed as revision 23087 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 23086 to svn://svn.ffmpeg.org/ffmpeg/trunk

Patch by Nick Brereton <nick at nbrereton dot net>

Originally committed as revision 23085 to svn://svn.ffmpeg.org/ffmpeg/trunk

was encountered with a channel range that overlapped the previous substreams,
and the code assumed no such overlap was possible.
Patch by Nick Brereton <nick at nbrereton dot net>

Originally committed as revision 23084 to svn://svn.ffmpeg.org/ffmpeg/trunk

long was being incorrectly used as an x86-sized register, both for 32 and 64
bits, but this is not the case in win64.

Originally committed as revision 31153 to svn://svn.mplayerhq.hu/mplayer/trunk/libswscale

Originally committed as revision 23083 to svn://svn.ffmpeg.org/ffmpeg/trunk

with x86_32 gcc 4.4.4 and -fPIC.

Originally committed as revision 23082 to svn://svn.ffmpeg.org/ffmpeg/trunk

Patch by Sebastian Vater <cdgs basty googlemail com>.

Originally committed as revision 23081 to svn://svn.ffmpeg.org/ffmpeg/trunk

they are >= 0.

Patch by Sebastian Vater <cdgs basty googlemail com>.

Originally committed as revision 23080 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 23079 to svn://svn.ffmpeg.org/ffmpeg/trunk

This fixes issue1907

Patch by Thomas Devanneaux gmail(thomdev)

Originally committed as revision 23078 to svn://svn.ffmpeg.org/ffmpeg/trunk

Fix issue 1924.

Originally committed as revision 23077 to svn://svn.ffmpeg.org/ffmpeg/trunk

Patch by Francesco Lavra, francescolavra interfree it

Originally committed as revision 23076 to svn://svn.ffmpeg.org/ffmpeg/trunk

such.

Also put the [3][0][0][0] codec tag, mapped to rgb565le, in a special
section. It needs to be specified *after* the nut RGB[16] codec tag,
otherwise it will be used by default when encoding normal non-flipped
rgb565le, and will be decoded like a flipped format (see
rawdec.c:raw_init_decoder()).

Originally committed as revision 23075 to svn://svn.ffmpeg.org/ffmpeg/trunk

Add codec tags for the formats:
[15]BGR Packed RGB 5:5:5, 16bpp, (msb)1A 5R 5G 5B(lsb), big-endian [NOT in AVI]
[15]RGB Packed BGR 5:5:5, 16bpp, (msb)1A 5B 5G 5R(lsb), big-endian [NOT in AVI]
[16]BGR Packed RGB 5:6:5, 16bpp, (msb)   5R 6G 5B(lsb), big-endian [NOT in AVI]
[16]RGB Packed BGR 5:6:5, 16bpp, (msb)   5B 6G 5R(lsb), big-endian [NOT in AVI]
RGBA    Packed RGBA 8:8:8:8, 32bpp, RGBA [NOT in AVI]
BGRA    Packed RGBA 8:8:8:8, 32bpp, BGRA [NOT in AVI]
ABGR    Packed RGBA 8:8:8:8, 32bpp, ABGR [NOT in AVI]
ARGB    Packed RGBA 8:8:8:8, 32bpp, ARGB [NOT in AVI]
RGB[24] Packed RGB 8:8:8, 24bpp RGB [NOT in AVI]
BGR[24] Packed RGB 8:8:8, 24bpp BGR [NOT in AVI]

which are listed in the nut4cc.txt spec.

Originally committed as revision 23074 to svn://svn.ffmpeg.org/ffmpeg/trunk

corresponding ones for the yuvXXX pixel formats.

Originally committed as revision 23073 to svn://svn.ffmpeg.org/ffmpeg/trunk

format uyyvyy411.

The codec tag is referenced in fourcc.org.

Originally committed as revision 23072 to svn://svn.ffmpeg.org/ffmpeg/trunk

cannot recognize the provided codec tag.

Originally committed as revision 23071 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 23070 to svn://svn.ffmpeg.org/ffmpeg/trunk

Patch by Sebastian Vater <cdgs basty googlemail com>.

Originally committed as revision 23069 to svn://svn.ffmpeg.org/ffmpeg/trunk

Patch by Sebastian Vater <cdgs basty googlemail com>.

Originally committed as revision 23068 to svn://svn.ffmpeg.org/ffmpeg/trunk

addition.

Originally committed as revision 23067 to svn://svn.ffmpeg.org/ffmpeg/trunk

top_field_first in AVFilterPicRef, done in r23044.

Originally committed as revision 23066 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 23065 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 23064 to svn://svn.ffmpeg.org/ffmpeg/trunk

The decodeplane8() function processes one byte of input less than
it should.  Also, the for loop has an unusual style with side-effects
in the controlling expression; replaced with a more intuitive while
loop.

10l to Basty.

Originally committed as revision 23063 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 23062 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 23061 to svn://svn.ffmpeg.org/ffmpeg/trunk

Use the numeric value assigned to sws_flags for the sws_flags set in
the graph, rather than the string "bilinear", which is not even
parsable by the scale filter.

Originally committed as revision 23060 to svn://svn.ffmpeg.org/ffmpeg/trunk

Originally committed as revision 23059 to svn://svn.ffmpeg.org/ffmpeg/trunk