Do not read padding or out-of-buffer values when computing the output
value for a pixel close to the image buffer edge.

This avoids non visible artifacts which affected the output checksum.

See thread:
Subject: [FFmpeg-devel] [PATCH] lavfi/mp/mcdeint: avoid uninited data read
Date: Thu, 30 May 2013 18:57:14 +0200

Fixes ticket #2614.

Currently the WebVTT demuxer parses the cues but throws away
the cue id (the optional first line of the cue) and cue
settings (the optional rendering instructions that follow
the timestamp).

However, in order to write inband text tracks (to WebM
files), the entire cue payload from the WebVTT source must
be preserved.

This commit makes no change to the data part of the output
buffer packet (where the actual cue text is stored), but
does add the cue id and settings as a side data items, if
they're present in the cue. Existing code that cares only
about the data part of the packet can continue to ignore the
side data.

There are two new packet data type flags,
AV_PKT_DATA_WEBVTT_IDENTIFIER and
AV_PKT_DATA_WEBVTT_SETTINGS.

The padding data is assumed to be 0 in several places, notably in
subtitles. This problem was not detected with fate-sub-srt test because
the first element of the side data (x1) is 0 in the test, so the
trailing side data present in the packet wasn't read by the decoder. The
issue can be observed with a large enough x1.

It is also noted in FF_INPUT_BUFFER_PADDING_SIZE doxy that MPEG
bitstreams require that padding with 0, so it might fix other issues.

This keeps the structure fields more consistent after cleanup
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

This makes no difference in practice since w & h are hardcoded to 512.

* cehoyos/master:
  Fix compilation with --disable-everything --enable-encoder=jpeg2000
Merged-by: Michael Niedermayer <michaelni@gmx.at>

4400->2800 cycles
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

4700 -> 2700 cycles (sandybridge  i7)
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

The JPEG2000 encoder depends on the discrete wavelet transform.

Currently, standard tables like AV_CRC_32_IEEE and such are being generated (or
provided in case the user compiles with hardcoded tables) with only 257 elements.
We're missing a considerable boost in performance by not making them with a size
of 1024 elements instead.
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

* qatar/master:
  indeo4: expand allowed quantiser range
  configure: icl: Merge -Qdiag-error parameters
Merged-by: Michael Niedermayer <michaelni@gmx.at>

* commit '14fb9d3d':
  configure: Separate commonalities in msvc and icl flags

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

Indeo 4 has quantiser range 0-31 instead of 0-23 for Indeo 5, and clipping
quantiser leads to incorrect quantisation and DC prediction on low-quality
videos.

This fixes bug 259.

It's redundant now as check in done in demuxer instead.
Signed-off-by: Paul B Mahol <onemda@gmail.com>

Since we don't check for seek table integrity in the decoder anymore,
check it in the demuxer instead (Only tta files are guaranteed to have
one, and it should be valid there).

Check also for header integrity, since the check done in the decoder
is ignored because matroska doesn't store the header at all.
Signed-off-by: James Almer <jamrial@gmail.com>

Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>

Allows for easier handling of flags that may be specific to icl or msvc.
Furthermore, simplify the handling of warnings and remarks thanks to
icl's support of -Wall on Windows.
Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>

* lukaszmluki/master:
  ftp: enhanced error handling
  ftp: reconnect on seek
  ftp: move common commands code to function
  ftp: enhanced status code handling
  ftp: reconnect on read
  ftp: move create control connection to function
  ftp: credentials moved into FTPContext
  ftp: rename function name
  ftp: formatting and typos fixes
Merged-by: Michael Niedermayer <michaelni@gmx.at>

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

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

Add error codes to expected codes to make responses faster.
Success of a command is validated by comparing code to a success code.

ABOR command usually takes long (FTP server implementation dependent).
It also produces different response codes from different servers.

It is save for ffmpeg to reconnect both connection during seek for two reasons:
1. Alreay mentioned heavy manner of ABOR command.
2. Server may disconnected due to no transfer (seek after a long pause in ffmpeg client)

Each send command routine requires the same steps.
This commit moves repeated code into one function.

Reimplementation of ftp_status function.
New version requires explicit list of expected codes.
It flush data connection input before sending a command
and blocks until expected result appears.

FTP server may disconnect after some period of time when no transfer is done.

FTP protocol should reconnect and seek to last position.

Move common code that opens control connection to function.
This code can be reused when reconnecting to FTP server
after inactivity.

FTP server may disconnect client.

This commit stores credentials for future reconnect.

ftp_send_command was used only once.

This commit makes this function specific, not generic

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

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

* qatar/master:
  proresenc: alpha coding support
Merged-by: Michael Niedermayer <michaelni@gmx.at>