This is a hand-tuned version of the code with impossible parts of
the FASTDIV function ommitted.

2-5% faster overall on Cortex-A8.

Signed-off-by: Mans Rullgard <mans@mansr.com>

In particular, make the json writer write size values, fix regression
introduced with the addition of the print_val() macro.

This makes the two pass encoding with x264 working with -f null - as
first pass.

This is required to fulfill avcodec_decode_video2() promise to return
the number of consumed bytes on success.

Add all seven test bitstreams of Coding Technologies "aacPlus Decoder
Check Package". The streams cover different ways to signal SBR and PS
in different formats.

As the filter uses the default start frame, the cleanup done by the
default end frame is needed.

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

* qatar/master:
  snow: split snow in snowdec and snowenc
  tiffenc: deprecate using compression_level
  swscale: fix failing fate tests.
  swscale: add support for planar RGB input.
  h264: add support for decoding planar RGB images.
  Clean up swscale pixfmt macros using av_pix_fmt_descriptors[].
  pixfmt: add planar RGB formats.

Conflicts:
	libavcodec/h264.c
	libavcodec/snow.c
	libavcodec/utils.c
	libavutil/avutil.h
	libavutil/pixdesc.c
	libavutil/pixfmt.h
	libswscale/swscale.c
	libswscale/swscale_internal.h
	libswscale/swscale_unscaled.c
	libswscale/utils.c

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

Casting the left-most byte to unsigned avoids an undefined
result of the shift by 24 if bit 7 is set.  This affects
the rm demuxer.

Signed-off-by: Mans Rullgard <mans@mansr.com>

Also reduce verbosity for the unsupported stream message, use
an AVFormatContext for av_log and and print the tag of the
unknown stream.

Improves ticket #672.

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

Signed-off-by: Alex Zhukov <zhukov.alex@gmail.com>
Reviewed-by: Jean First <jeanfirst@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

The common non inlined code goes in snow.c, the common inlined code in
snow.h, tables move in snowdata.h (included only by snow.c)

isGray() is left as a FIXME for later.

If the sample format is invalid, av_get_sample_fmt_name() will always
return NULL.

The number of used channels need to be reset to zero when
swr_alloc_set_opts() is called successive times.

This is a regression introduced in c8136ebd.

Fixes null pointer dereference / http://www.google-melange.com/gci/task/view/google/gci2011/7120335

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

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

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

The qatar implementation makes no sense.
a muxer without timestamps is constant fps thus needs vsync.
the crc/mp5 are special cases that have timestamps yet allow any
nonsensical timestamps.
raw (yuv/rgb) video is constant fps thus needs vsync too.

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

* qatar/master:
  libavutil: add utility functions to simplify allocation of audio buffers.
  libavutil: add planar sample formats and av_sample_fmt_is_planar()
  avconv: fix segfault at EOF with delayed pictures
  pcmdec: remove unneeded resetting of samples pointer
  avconv: remove a now unused parameter from output_packet().
  avconv: formatting fixes in output_packet()
  avconv: declare some variables in blocks where they are used
  avconv: use the same behavior when decoding audio/video/subs
  bethsoftvideo: return proper consumed size for palette packets.
  cdg: skip packets that don't contain a cdg command.
  crcenc: add flags
  avconv: use vsync 0 for AVFMT_NOTIMESTAMPS formats.
  tiffenc: add a private option for selecting compression algorithm
  md5enc: add flags
  ARM: remove needless .text/.align directives

Conflicts:
	doc/APIchanges
	libavcodec/tiffenc.c
	libavutil/avutil.h
	libavutil/samplefmt.c
	libavutil/samplefmt.h
	tests/ref/fate/bethsoft-vid
	tests/ref/fate/cdgraphics
	tests/ref/fate/film-cvid-pcm-stereo-8bit
	tests/ref/fate/mpeg2-field-enc
	tests/ref/fate/nuv
	tests/ref/fate/tiertex-seq
	tests/ref/fate/tscc-32bit
	tests/ref/fate/vmnc-32bit

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

Based on code by Stefano Sabatini.

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

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

At EOF it makes no sense to modify avpkt.{data,size} in output_packet
since no data is consumed. Frame threading with more than 1 threads
hits the segfault.

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

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

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

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

I.e. if the packet was only partially consumed, pass the rest of it into
the decoder again.

Also simplify the code so it's the same for video/audio/subs.

Also check for sufficient packet size.

Signed-off-by: Anton Khirnov <anton@khirnov.net>

tag that is larger than the probe buffer.
Fixes Ticket336

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

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

AVFMT_NOTIMESTAMPS for crc, as it ignores the timestamps.
AVFMT_VARIABLE_FPS for framecrc, as it prints dts.

Many FATE changes, because avconv is no longer duplicating frames in
those tests.
Also added -vsync 0 for some tests to prevent avconv from dropping
frames until it can be fixed more properly.

Prevent avconv from duplicating or dropping frames for formats where it
makes even less sense than usual, e.g. image2.

AVFMT_NOTIMESTAMPS for md5, as it ignores the timestamps.
AVFMT_VARIABLE_FPS for framemd5, as it prints dts.

-vsync 0 for the vp8 test is needed because with vsync 2 the timestamp
guessing code gets confused by an altref frame that is never displayed
and drops a frame later.