The encode function is supposed to just return 0 on success.
This stems from a mixup with the return value of decode functions.
Reviewed-by: Jan Gerber <j@v2v.cc>
Signed-off-by: Martin Storsjö <martin@martin.st>
We still only support one single layer though, but this allows
receiving streams that have this structure present even for
single layer streams.
Signed-off-by: Martin Storsjö <martin@martin.st>
This was missed in e1eb0fc960, when ff_interleaved_peek was
changed to include const during the evolution of the patch.
Signed-off-by: Martin Storsjö <martin@martin.st>
As long as caller only writes packets using av_interleaved_write_frame
with no manual flushing, this should allow us to always have accurate
durations at the end of fragments, since there should be at least
one queued packet in each stream (except for the stream where the
current packet is being written, but if the muxer itself does the
cutting of fragments, it also has info about the next packet for that
stream).
Signed-off-by: Martin Storsjö <martin@martin.st>
This allows callers with avio write callbacks to get the bytestream
positions that correspond to keyframes, suitable for live streaming.
In the simplest form, a caller could expect that a header is written
to the bytestream during the avformat_write_header, and the data
output to the avio context during e.g. av_write_frame corresponds
exactly to the current packet passed in.
When combined with av_interleaved_write_frame, and with muxers that
do buffering (most muxers that do some sort of fragmenting or
clustering), the mapping from input data to bytestream positions
is nontrivial.
This allows callers to get directly information about what part
of the bytestream is what, without having to resort to assumptions
about the muxer behaviour.
One keyframe/fragment/block can still be split into multiple (if
they are larger than the aviocontext buffer), which would call
the callback with e.g. AVIO_DATA_MARKER_SYNC_POINT, followed by
AVIO_DATA_MARKER_UNKNOWN for the second time it is called with
the following data.
Signed-off-by: Martin Storsjö <martin@martin.st>
When feeding input RTP packets to the depacketizer via custom IO,
it needs to pick the right stream using the payload type for
RTP packets, and using the SSRC for RTCP packets. If the first
packet is an RTCP packet, we don't (currently) know the SSRC
yet and thus can't pick the right RTP depacketizer to handle it.
By parsing the SSRC attribute in the SDP, we can map initial
RTCP packets to the right stream.
Signed-off-by: Martin Storsjö <martin@martin.st>
It doesn't matter what the actual reason for not returning
an AVPacket was - if we didn't return any packet and we have
the next one queued, parse it immediately. (rtp_parse_queued_packet
always consumes a queued packet if one exists, so there's no risk
for infinite loops.)
Signed-off-by: Martin Storsjö <martin@martin.st>
The declarations that this comment referred to were removed
in 2439f2ca8 - there is no unbuffered IO in this header now.
Signed-off-by: Martin Storsjö <martin@martin.st>
We still only support one single layer though, but this allows
receiving streams that have this structure present even for
single layer streams.
Signed-off-by: Martin Storsjö <martin@martin.st>
This codepath isn't quite as bad as it used to sound, if fragments
are cut automatically at video packets.
Signed-off-by: Martin Storsjö <martin@martin.st>
This can only be used if the input data happens to be laid out
exactly correctly.
This might not be supported on all encoders, so only enable it
with an option, but enable it automatically on raspberry pi,
where it is known to be supported.
Signed-off-by: Martin Storsjö <martin@martin.st>
The raspberry pi uses the alternative API/ABI for OMX; this makes
such builds incompatible with all the normal OpenMAX implementations.
Since this can't easily be detected at configure time (one can
build for raspberry pi's OMX just fine using the generic, pristine
Khronos OpenMAX IL headers, no need for their own extensions),
require a separate configure switch for it instead.
The broadcom host library can't be unloaded once loaded and started;
the deinit function that it provides is a no-op, and after started,
it has got background threads running, so dlclosing it makes it
crash.
Signed-off-by: Martin Storsjö <martin@martin.st>
Previously, ff_h264_idct_add_neon (originally in the arm version) used
a non-regular transpose in order to be able to use more instructions
that deal with registers as 128 bit register pairs. The aarch64
translation doesn't do it to the same extent, but brought along the
same structure since it was a straight translation.
This reshuffles ff_h264_idct_add_neon, bringing it closer to
the C implementation, making the transpose_4x4H macro do a regular
transpose, usable for other algorithms as well.
Previously, the third and fourth output from transpose_4x4H were
swapped, and prior to cc29d96d5a, the same inputs as well. In
addition to just swapping the outputs, also renumber the intermediate
registers for better readability (making the register order match
transpose_4x8B).
This runs with the same number of cycles as before.
Signed-off-by: Martin Storsjö <martin@martin.st>
Apply the default value for timeout in code instead of via the
avoption, to allow distinguishing the default value from the user
not setting anything at all.
Signed-off-by: Martin Storsjö <martin@martin.st>
Since all URLContexts have the same AVOptions, such AVOptions
will be applied on the outermost context only and removed from the
dict, while they probably make sense on all contexts.
This makes sure that rw_timeout gets propagated to the innermost
URLContext (to make sure it gets passed to the tcp protocol, when
opening a http connection for instance).
Alternatively, such matching options would be kept in the dict
and only removed after the ffurl_connect call.
Signed-off-by: Martin Storsjö <martin@martin.st>
Using this requires setting the rw_timeout option to make it
terminate, alternatively using the interrupt callback (if used via
the API).
Signed-off-by: Martin Storsjö <martin@martin.st>
For http, this avoids spurious warnings about failed requests (e.g.
HTTP error 416 Requested Range Not Satisfiable), if the last packet
is truncated and the size read is bogus.
Signed-off-by: Martin Storsjö <martin@martin.st>
This fixes builds with --disable-vfp.
Checking for the armv6 cpu flag is incorrect, since vfpv2 isn't
armv6 specific.
Signed-off-by: Martin Storsjö <martin@martin.st>
Use two separate functions, depending on whether VFP/NEON is available.
This is set to require armv5te - it uses blx, which is only available
since armv5t, but we don't have a separate configure item for that.
(It also uses ldrd, which requires armv5te, but this could be avoided
if necessary.)
Signed-off-by: Martin Storsjö <martin@martin.st>
This fixes builds with --disable-vda, which previously failed with
undefined references to CVImageBuffer* and CVPixelBuffer* functions.
Signed-off-by: Martin Storsjö <martin@martin.st>
The XTEA algorithm operates on 32 bit numbers, not on byte sequences.
The XTEA implementation in libavutil is written assuming big endian
numbers, while the rtmpe signature encryption assumes little endian.
This fixes rtmpe communication with rtmpe servers that use signature
type 8 (XTEA), e.g. crunchyroll.
Signed-off-by: Martin Storsjö <martin@martin.st>
By writing a zero-sized packet, the caller can communicate the
start_dts/start_cts for the stream without actually writing
the first packet.
This allows doing random-access writing of fragments when the
start dts of the stream isn't zero, so that the edit list in the moov
is written based on timestamps from the nominal start time signaled
via the zero-sized packet, while the first proper packet written
corresponds to a later fragment.
To avoid potential unexpected behaviour, empty packets only set
start_dts if the frag_discont flag is set.
Signed-off-by: Martin Storsjö <martin@martin.st>
This allows producing fragments discontinously where the video
stream has b-frames (but starts at pts=0), but doesn't work for the
cases with audio with preroll.
Signed-off-by: Martin Storsjö <martin@martin.st>
Contrary to the normal fate tests that run via avconv, this tests
nontrivial call sequences that are only doable via the API
(mainly for different corner cases when using the muxer for
segmenting).
The test muxes fake packet data (with extradata that looks
enough like proper data to make the file be viewable with e.g.
boxdumper) and checks the hash of the produced files. The test also
verifies that fragments produced via different call sequences remain
identical (to avoid e.g. updating the output hashes and suddenly
having fragments that used to be identical suddenly diverging), for
fragments written with frag_discont and/or delay_moov.
Signed-off-by: Martin Storsjö <martin@martin.st>
In most other cases when writing fragmented mp4 files, the output
IO context is flushed after each fragment. Also flush it after
writing the initial moov, to have it behave in the same way.
Signed-off-by: Martin Storsjö <martin@martin.st>
Some systems may be lacking getservbyport; the previous ifdef wasn't
quite enough since it still assumed that struct servent was defined,
as pointed out by Clément Gregoire.
Simply remove the possibility to return non-numeric services in
getnameinfo; no caller of getnameinfo within libavformat
currently try to use getnameinfo for retrieving the port number without
NI_NUMERICSERV, and falling back on getservbyport may be non-threadsafe.
Signed-off-by: Martin Storsjö <martin@martin.st>
This silences warnings like this one:
libavformat/file.c(62) : warning C4996: 'read': The POSIX name for this
item is deprecated. Instead, use the ISO C++ conformant name: _read.
See online help for details.
Signed-off-by: Martin Storsjö <martin@martin.st>
This allows enabling the frame skipping, which is required for the
encoder to properly hit the target bitrate.
Signed-off-by: Martin Storsjö <martin@martin.st>
This also makes sure that a fragmented file without the empty_moov
flag (i.e. with a non-empty initial moov fragment) actually gets
written, if some of the tracks turn out to not have any samples.
Signed-off-by: Martin Storsjö <martin@martin.st>
Some RTSP servers ("HiIpcam/V100R003 VodServer/1.0.0") respond to
our keepalive GET_PARAMETER request by a truncated RTSP header
(lacking the final empty line to indicate a complete response
header). Prior to 764ec70149, this worked just fine since we
reacted to the $ as interleaved packet indicator anywhere.
Since $ is a valid character within the response header lines,
764ec70149 changed it to be ignored there. But to keep
compatibility with such broken servers, we need to at least
allow reacting to it at the start of lines.
Signed-off-by: Martin Storsjö <martin@martin.st>
Since the actual max length of the jitter buffer is restricted by
max_delay, this shouldn't harm the overall latency (assuming that
max_delay is set properly), while allowing packet reordering with
a larger number of packets (which may be required with high bitrate
video).
Signed-off-by: Martin Storsjö <martin@martin.st>
The previous restriction was partially designed to fix certain
(broken) samples from bug 215. There should be no restriction on the
number of keyframes per fragment or trun.
The spec suggests that all frames lacking MOV_FRAG_SAMPLE_FLAG_IS_NON_SYNC
are key frames, but we require the flag MOV_FRAG_SAMPLE_FLAG_DEPENDS_YES
to be unset as well. This works for (possibly broken) media that never
sets the NON_SYNC flag and should also be correct for any spec-compliant
file.
For files that never set either of the flags, all samples are marked
as keyframes.
Signed-off-by: Martin Storsjö <martin@martin.st>
Previously most of the error paths leaked.
Also add FF_CODEC_CAP_INIT_THREADSAFE while adding caps_internal;
this decoder wrapper doesn't have any static data that is initialized.
Signed-off-by: Martin Storsjö <martin@martin.st>
For ADTS streams, the output format (number of channels, frame size)
can change at any point (with the latest version of fdk-aac, the decoder
seems to change format after a handful of frames, not outputting the
right format immediately, for cases that worked fine with the earlier
version of the lib).
Previously, the decoder decoded straight into the output frame once the
number of channels and frame size was known. This obviously does not
work if the number of channels or frame size changes.
The alternative would be to allocate the AVFrame with the maximum number
of channels and frame size, and change them afterward decoding into it,
but that may cause confusion to users e.g. of the get_buffer callback.
This solution should be more robust.
CC: libav-stable@libav.org
Signed-off-by: Martin Storsjö <martin@martin.st>
In the latest version of fdk-aac, the decoder can output up to 8
channels; take this into account when preallocating buffers that
need to fit the output from any packet.
CC: libav-stable@libav.org
Signed-off-by: Martin Storsjö <martin@martin.st>
Don't try to detect the float ABI by checking at the toolchain
name or by trying to assemble and link files with eabi_attributes.
This fixes the float ABI detection when building using clang
with -fembed-bitcode, where the current eabi_attributes check
accidentally passes.
This issue was pointed out by James Howe <james.howe@hp.com>.
CC: libav-stable@libav.org
Signed-off-by: Martin Storsjö <martin@martin.st>
The double meaning of the faststart flag (moving a moov atom
to the start of files, making them streamable, for non-fragmented
files, vs inserting a global sidx index at the start of files
for fragmented files) is confusing - see 40ed1cbf1 for
explanation of its origins.
Since the second meaning of the flag hasn't been part of any
libav release yet, just rename it to get rid of the confusion
without any extra deprecation (which wouldn't get rid of the
potential confusion, of users adding -movflags faststart
even for fragmented files, where it isn't needed for making
them "streamable").
This gets back the old behaviour, where -movflags faststart
doesn't have any effect for fragmented files.
Signed-off-by: Martin Storsjö <martin@martin.st>
For fragmented files with non-empty moov, with a fragment index
(sidx), place the index after the initial moov/mdat pair.
Previously, for this pathological case, the index was written
at the start of the file.
Signed-off-by: Martin Storsjö <martin@martin.st>
The same field is also used for writing the sidx index header,
for fragmented files, when the faststart flag is used.
Signed-off-by: Martin Storsjö <martin@martin.st>
This fixes crashes with pathological cases when trying to write
a sidx index (via the -movflags faststart option, in combination
with fragmenting options), when no fragments actually have been
written. (This is possible if the empty_moov flag isn't used,
so that all actual packet data is written in the moov/mdat pair,
and no moof/mdat pairs have been written.)
In these pathological cases, no sidx should be written at all.
Signed-off-by: Martin Storsjö <martin@martin.st>
Allow $ as character anywhere within normal RTSP replies - both
within the lines, and as the first character of RTSP header lines.
(The existing old comment indicated that an inline packet could
start at any line within a RTSP reply header, but that doesn't
sound valid to me, and I'm not sure if the existing code
handled that correctly either.)
CC: libav-stable@libav.org
Signed-off-by: Martin Storsjö <martin@martin.st>
configure does check for isatty, and checkasm properly checks
HAVE_ISATTY, but on some platforms (e.g. WinRT), io.h needs to be
included for isatty to be available.
Signed-off-by: Martin Storsjö <martin@martin.st>
This avoids having to manually set _WIN32_WINNT in --extra-cflags
when targeting these API families, which only was necessary to
work around configure setting _WIN32_WINNT to an older version
by default.
Signed-off-by: Martin Storsjö <martin@martin.st>
If _WIN32_WINNT is unset, we force it to a new enough value to
make sure the necessary definitions are visible.
When targeting Windows Phone or Windows RT, _WIN32_WINNT should
be at least 0x0602 - otherwise the windows headers themselves
can cause errors (which technically are bugs in the headers).
Raising this value here shouldn't hurt; the alternative would
be to not touch it at all if WINAPI_FAMILY is set to phone/app,
or to force setting it to 0x0602 in configure if unset (for phone/app).
Signed-off-by: Martin Storsjö <martin@martin.st>
This macro identifies whether VFPv3 is available; MSVC defaults
to hardfloat (except for older MSVC versions for CE, targeting
ARMv4).
Signed-off-by: Martin Storsjö <martin@martin.st>
This is only necessary on MSVC 2010/2012 (and possibly on some
ICL versions). This both avoids an extra hack on newer MSVC
versions, and better documents what the extra compiler option
is used for.
Signed-off-by: Martin Storsjö <martin@martin.st>
Most of the fate-dds-* and fate-txd-* tests already
output into the same pixel format regardless of
platform endianness, so there's no need to force
conversion to another format.
This fixes the tests fate-txd-16bpp, fate-txd-odd,
fate-dds-rgb16, fate-dds-rgb24 and fate-dds-xrgb on
big endian, where the tests seem to fail due to issues
with certain conversion codepaths in swscale.
Those conversion codepaths should of course be fixed, but
the individual decoder tests should use as little extra
conversion steps as possible.
Signed-off-by: Martin Storsjö <martin@martin.st>
These are only necessary once/if avconv gets support for this hwaccel.
While that obviously is desireable, we don't have it yet, and they
currently only are a distraction.
Signed-off-by: Martin Storsjö <martin@martin.st>
The only need to be listed if they are to be used in ifdefs from
within the code - config items used as dependencies only within
configure don't need to be exported.
Signed-off-by: Martin Storsjö <martin@martin.st>
This fixes dxva2 detection (i.e. correctly realizes that it isn't
available) for WinRT, where dxva2api.h does exist, but these definitions
are omitted (when targeting the API subsets).
Ideally we should rather check for e.g. DXVA2_ConfigPictureDecode,
but configure might fail to find that definition due to _WIN32_WINNT
not being set to the right value during configure. (libavcodec/dxva2.h
manually overrides the _WIN32_WINNT define.)
This allows removing hardcoded --disable-dxva2 from such build
configurations.
Signed-off-by: Martin Storsjö <martin@martin.st>
mpz_import and mpz_export were added in GMP 4.1, in 2002.
This simplifies the DH code by clarifying that it only uses pure
bignum functions, no other parts of nettle/hogweed.
Signed-off-by: Martin Storsjö <martin@martin.st>
Martin StorsjöandMichael Niedermayer
