Merge commit 'f61272f0efd80da437570aad2c40e00f9d3f4fe6'

* commit 'f61272f0efd80da437570aad2c40e00f9d3f4fe6':
  ratecontrol: K&R cosmetic formatting
  rtpdec: Remove a useless todo comment

Conflicts:
libavcodec/ratecontrol.c

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

Merge commit '54cb096ee4558b3bfc28c2fcd6418ce82dc39fe1'

* commit '54cb096ee4558b3bfc28c2fcd6418ce82dc39fe1':
  rtsp: Remove an outdated comment
  rtsp: Remove references to weirdly named variables in other files

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

Merge commit 'c44784c9bb9d0ddf5d39d0dfa640816a57b8f457'

* commit 'c44784c9bb9d0ddf5d39d0dfa640816a57b8f457':
  rtp: Rename a static variable to normal naming conventions
  rtp: Cosmetic cleanup

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

ratecontrol: K&R cosmetic formatting

Signed-off-by: Diego Biurrun <diego@biurrun.de>

doc/protocols: improve wording of a sentence in http docs

Suggested-By: Alexander Strasser

doc/ffserver: remove painfully outdated "What do I need?" section

doc/ffserver: remove paragraph in the introductory blurb

The paragraph is about what ffserver is not and where to look for other
information, but is pretty redundant and distracting, especially
considering the new organization of the documentation.

doc/ffserver: rework introducing paragraphs of the "description" chapter

In particular, add some markup, fix empty line, and merge it with "How
does it work?" section.

doc/protocols: document http protocol options

lavf/http: fix/extend option descriptions

In particular, favor predicative form over nominal description of the set
parameter. This is more globally consistent.

mpegvideo: dont leave stale pointers in next/last picture

Fixes out of array reads

Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

vc1dec: ensure cbpcy_vlc has been set before decoding a frame.

Fixes null pointer dereference

Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

fate: try to fix dirac

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

h264: reset first_field when current_picture_ptr is reset

Fixes NULL pointer dereference

Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

rtpdec: Remove a useless todo comment

The question can be answered: No, we do not know the initial sequence
number from the SDP. In certain cases, it can be known from the
RTP-Info response header in RTSP though. (In that case, we use it as
timestamp origin, but not for rtp receiver statistics.)

Signed-off-by: Martin Storsjö <martin@martin.st>

rtsp: Remove an outdated comment

It is unclear what the bug exactly was and if it ever was fixed,
and we don't even support decoding via faad any longer. The
comment has been present since d0deedcb in 2006.

Signed-off-by: Martin Storsjö <martin@martin.st>

rtsp: Remove references to weirdly named variables in other files

One of them is renamed now, but mentioning it by name serves
no purpose here.  The other table mentioned ceased to exist
under that name in 4934884a1 in 2006.

Signed-off-by: Martin Storsjö <martin@martin.st>

rtp: Rename a static variable to normal naming conventions

Signed-off-by: Martin Storsjö <martin@martin.st>

rtp: Cosmetic cleanup

Remove leftover debug comments, fix brace placement and
add whitespace, remove unnecessary and weirdly placed braces.

Signed-off-by: Martin Storsjö <martin@martin.st>

fate: add dirac decoding test (main profile).

mjpegdec: Fix out of array read in unescaping code

Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

mxfdec: Fix integer overflow with many channels

Fixes division by zero

Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Reviewed-by: Matthieu Bouron <matthieu.bouron@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

wavpack: check pointer to avoid overreading input buffer

Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

ra144enc: fix two "may be used uninitialized in this function" warnings

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

au: switch to ff_pcm_read_packet() again, after the merge

This fixes G722 durations

Suggested-by: durandal_1707
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Merge remote-tracking branch 'qatar/master'

* qatar/master:
  configure: enable pic for shared libs on AArch64
  zmbv: Reset the decoder on keyframe errors

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

Merge commit '7b8c5b263bc680eff5710bee5994de39d47fc15e'

* commit '7b8c5b263bc680eff5710bee5994de39d47fc15e':
  vc1dec: prevent a crash due missing pred_flag parameter
  matroska: Fix use after free

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

Merge commit 'ec86ba57312745fd7ad9771e3121e79c6aacba30'

* commit 'ec86ba57312745fd7ad9771e3121e79c6aacba30':
  vp3: Fix double free in vp3_decode_end()
  fate: Split fate-siff test into demuxer and VB video decoder test

Conflicts:
tests/fate/demux.mak
tests/fate/video.mak

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

lavf: remove nonexistent symbols

Signed-off-by: Paul B Mahol <onemda@gmail.com>

configure: enable pic for shared libs on AArch64

Signed-off-by: Marcin Juszkiewicz <marcin.juszkiewicz@linaro.org>
Signed-off-by: Martin Storsjö <martin@martin.st>

zmbv: Reset the decoder on keyframe errors

Prevent the crash on fuzzed files as reported in bug 63.

vc1dec: prevent a crash due missing pred_flag parameter

Handle pred_flag parameter not given to get_mvdata_interlaced()

Signed-off-by: Vladimir Pantelic <vladoman@gmail.com>
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>

error_concealment: Check that the picture is not in a half setup state.

Fixes state becoming inconsistent
Fixes a null pointer dereference

Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

ass_split: fix out of array access in ass_split()

Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

ass_split: fix out of array access

Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

vorbisdec: Check bark_map_size.

This fixes potential divisions by zero and out of array accesses.

Reported-by: Dale Curtis <dalecurtis@chromium.org>
Found-by: inferno@chromium.org
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

matroska: Fix use after free

Signed-off-by: Dale Curtis <dalecurtis@chromium.org>
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>

lavfi/vf_overlay: check that filter_frame does not return EAGAIN.

EAGAIN is used internally in the code; it currently does not make
sense as a return code from the next filter.

Should fix CID966655.

swr: check that allocation of in/out_convert succeeded

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

vp3: Fix double free in vp3_decode_end()

Signed-off-by: Dale Curtis <dalecurtis@chromium.org>
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>

swr: Use a temporary buffer for dither/Noise shaping when the input is read only

This avoids copying the input

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

swr: mark noise data for noise shaping as const

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

swr: add seperate in/out pointers to the noise shaping code

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

swr: rename dither_pos to noise_pos

THis more closely matches what the field represents

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

fate: Split fate-siff test into demuxer and VB video decoder test

idcin: fix memleak

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

Merge remote-tracking branch 'qatar/master'

* qatar/master:
  rtpdec_vp8: Don't trim too much data from broken frames
  rtpdec_vp8: Simplify code by using an existing helper function

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

Do not allow arbitrary packet size in aac decoder.

Fixes a crash with a file containing a 400MB packet reported in bug 420.

Merge commit 'ed79093222ceb42f0c3a39095a69af0b32be5450'

* commit 'ed79093222ceb42f0c3a39095a69af0b32be5450':
  rtpdec: Add a terminating null byte at the end of the SDES/CNAME
  yuv4mpeg: do not use deprecated functions
  oggdec: fix faulty cleanup prototype
  idcin: return 0 from idcin_read_packet() on success.

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

Merge commit '5d0450461ff729be5f531d333d29754155e406c5'

* commit '5d0450461ff729be5f531d333d29754155e406c5':
  idcin: better error handling
  idcin: check for integer overflow when calling av_get_packet()

Conflicts:
libavformat/idcin.c

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

Merge commit '7040e479a1530b2eda4b89a182d5eb50a77bd907'

* commit '7040e479a1530b2eda4b89a182d5eb50a77bd907':
  idcin: allow seeking back to the first packet
  idcin: set AV_PKT_FLAG_KEY for video packets with a palette

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

Merge commit 'ccc0ffb1ba3fc1adb05a9f56dfc26131e61db3fb'

* commit 'ccc0ffb1ba3fc1adb05a9f56dfc26131e61db3fb':
  idcin: set start_time and packet duration instead of manually tracking pts.
  idcin: set channel_layout

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

Merge commit '12c2530b1d87fa94f81ea97df575b77c825e6f4f'

* commit '12c2530b1d87fa94f81ea97df575b77c825e6f4f':
  idcin: fix check for presence of an audio stream
  idcin: validate header parameters

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

Merge commit 'f7bf72a4a1146a7583577c9bdc066767e1ba3c6a'

* commit 'f7bf72a4a1146a7583577c9bdc066767e1ba3c6a':
  idcinvideo: correctly set AVFrame defaults
  yadif: Port inline assembly to yasm
  au: remove unnecessary casts
  au: return AVERROR codes instead of -1

Conflicts:
libavcodec/idcinvideo.c
libavfilter/x86/yadif_template.c
libavformat/au.c

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

Merge commit 'fd9147f11456a7e39a998d7270684922a2a46e6d'

* commit 'fd9147f11456a7e39a998d7270684922a2a46e6d':
  au: cosmetics: pretty-print and remove pointless comments

Conflicts:
libavformat/au.c

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

Merge commit 'c88d245c9866e48cb8a238b7564964c1fcf3315f'

* commit 'c88d245c9866e48cb8a238b7564964c1fcf3315f':
  au: use ff_raw_write_packet()
  au: set stream start time and packet durations

Conflicts:
libavformat/au.c

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

Merge commit 'af68a2baae6761044cbed95575e8bcfebf55c6f1'

* commit 'af68a2baae6761044cbed95575e8bcfebf55c6f1':
  au: use %u when printing id and channels since they are unsigned
  au: validate sample rate

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

Merge commit 'c837b38dd33a11c3810e988a60193a858eb4f58c'

* commit 'c837b38dd33a11c3810e988a60193a858eb4f58c':
  au: move skipping of unused data to before parameter validation
  au: do not arbitrarily limit channel count

Conflicts:
libavformat/au.c

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

lavfi/yadif: fix shorthand/option mismatch

Fix trac ticket #2128.

Signed-off-by: Stefano Sabatini <stefasab@gmail.com>

Merge commit '2613de88051818abafccb98646394341887acb3a'

We dont use au_read_packet() anymore and block_align is already set.
Thus nothing had to be merged from these.

* commit '2613de88051818abafccb98646394341887acb3a':
  au: do not set pkt->size directly
  au: set block_align and use it in au_read_packet()

Conflicts:
libavformat/au.c

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

Merge commit '9a7b56883d1333cdfcdf0fa7584a333841b86114'

* commit '9a7b56883d1333cdfcdf0fa7584a333841b86114':
  au: set bit rate
  au: validate bits-per-sample separately from codec tag
  rtpdec_vp8: Mark broken packets with AV_PKT_FLAG_CORRUPT

Conflicts:
libavformat/au.c

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

Fix compilation with --disable-everything.

rtpdec_vp8: Don't trim too much data from broken frames

Previously, for broken frames, we only returned the first partition
of the frame (we would append all the received packets to the packet
buffer, then set pkt->size to the size of the first partition, since
the rest of the frame could have lost data inbetween) - now instead
return the full buffered data we have, but don't append anything more
to the buffer after the lost packet discontinuity. Decoding the
truncated packet should hopefully get better quality than trimming out
everything after the first partition.

Signed-off-by: Martin Storsjö <martin@martin.st>

rtpdec_vp8: Simplify code by using an existing helper function

Signed-off-by: Martin Storsjö <martin@martin.st>

rtpdec: Add a terminating null byte at the end of the SDES/CNAME

This is required by RFC 3550 (section 6.5):

   The list of items in each chunk MUST be terminated by one or more
   null octets, the first of which is interpreted as an item type of
   zero to denote the end of the list.

This was implicitly added as padding before, unless the host name
length matched up so no padding was added.

This makes wireshark parse the packets properly if other RTCP items
are appended to the same packet.

Signed-off-by: Martin Storsjö <martin@martin.st>

wmalosslessdec: make arrays indexed by ch large enough for maximum number of channels

Fixes out of array accesses

Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

targa: use checked bytestream read

Fix out of array read

Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

qdm2: fix out of stack array read.

The read value is not used when its outside the array.

Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

swr: scale data down in noise shaping to avoid cliping

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

swr: reorder the operations in the noise shaping code

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

swr: add noise_scale field to seperate it from the noise shaping scales

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

swr: fix the noise sample type for noise shaping

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

swr: split swri_dither_init() out

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

doc: add noise shaping dithers

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

Add -movflags rtphint documentation.

yuv4mpeg: do not use deprecated functions

Use the libavutil replacement.

oggdec: fix faulty cleanup prototype

idcin: better error handling

Add some additional checks for EOF and print error messages on an incomplete
header or packet.

FATE reference updated for id-cin-video due to the demuxer no longer
returning a partial video packet at EOF.

idcin: return 0 from idcin_read_packet() on success.

This matches the AVInputFormat.read_packet() API.

idcin: allow seeking back to the first packet

Also, do not allow seek-by-byte, as there is no way to find the next packet
boundary.

idcin: set start_time and packet duration instead of manually tracking pts.

Also, use 1 / sample_rate for audio stream time_base.

idcin: fix check for presence of an audio stream

idcinvideo: correctly set AVFrame defaults

idcin: check for integer overflow when calling av_get_packet()

chunk_size is unsigned 32-bit, but av_get_packet() takes a signed int as the
packet size.

idcin: set AV_PKT_FLAG_KEY for video packets with a palette

idcin: set channel_layout

idcin: validate header parameters

Avoids using unsupported parameters and signed integer overflows.

swr: remove unused variable

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

swr: move dither related fields to their own context

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

yadif: Port inline assembly to yasm

Signed-off-by: Luca Barbato <lu_zero@gentoo.org>

swr: update copyright years of a few files iam working on

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

swr: Implement Noise shaping dither

The following variants are implemented:
lipshitz noise shaping dither
shibata noise shaping dither
low shibata noise shaping dither
high shibata noise shaping dither
f-weighted noise shaping dither
modified-e-weighted noise shaping dither
improved-e-weighted noise shaping dither

Data tables taken from SOX

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

au: cosmetics: pretty-print and remove pointless comments

au: use ff_raw_write_packet()

au: use %u when printing id and channels since they are unsigned

au: move skipping of unused data to before parameter validation

Also do not unnecessarily skip 0 bytes.

au: do not set pkt->size directly

It is already set by av_get_packet() even for partial reads.

au: remove unnecessary casts

au: set stream start time and packet durations

au: validate sample rate