One of the jitterbuffers functions is to try and make sense of weird
network behavior.
It is quite unhelpful for the jitterbuffer to start dropping packets
itself when what you are trying to achieve is better network resilience.
In the case of a skew, this could often mean the sender has restarted
in some fashion, and then dropping the very first buffer of this "new"
stream could often mean missing valuable information, like in the case
of video and I-frames.
This patch simply reverts back to the old behavior, prior to 8d955fc32b
and includes the simplest test I could write to demonstrate the behavior,
where a single packet arrives "perfectly", then a 50ms gap happens,
and then two more packets arrive in perfect order after that.
# Conflicts:
# tests/check/elements/rtpjitterbuffer.c
The memory leak occurs in the case when the buffer has been
added to the fragment_buffers array of the current pad and
never been sent because of the push failure of the previous
buffers: moof or mdat header or fragmented buffer(s).
This commit adds custom element messages for when gstrtpjitterbuffer
drops an incoming rtp packets due to for example arriving too late.
Applications can listen to these messages on the bus which enables
actions to be taken when packets are dropped due to for example high
network jitter.
Two properties has been added, one to enable posting drop messages and
one to set a minimum time between each message to enable throttling the
posting of messages as high drop rates.
There are in the wild (mp4) streams that basically contain no tracks
but do have a redirect info[0], in which case, we won't be able
to expose any pad (there are no tracks) so we can't post anything but
an error on the bus, as:
- it can't send EOS downstream, it has no pad,
- posting an EOS message will be useless as PAUSED state can't be
reached and there is no sink in the pipeline meaning GstBin will
simply ignore it
The approach here is to to add details to the ERROR message with a
`redirect-location` field which elements like playbin handle and use right
away.
[0]: http://movietrailers.apple.com/movies/paramount/terminator-dark-fate/terminator-dark-fate-trailer-2_480p.mov
When the queue is full (and adding more packets would risk a seqnum
roll-over), the best approach is to just start pushing out packets
from the other side. Just pushing out the packets results in the
timers being left hanging with old seqnums, so it's safer to just
execute them immediately in this case. It does limit the timer space
to the time it takes to receiver about 32k packets, but without
extended sequence number, this is the best RTP can do.
This also results in the test no longer needed to have timeouts or
timers as pushing packets in drives everything.
Fixes#619
This basically add ability to choose between inserting from head, tail
or in-place in order to try and minimize the distance to walk through in
the timer queue. This removes an overhead we had seen on high drop rate.
The timer passed to update_timers may be from the stats timer. At the
moment, we could endup rescheduling (reusing) that timer onto the normal
timer queue, unschedul it as if it was from the normal timer queue or
duplicate it into the stats timer queue again. This was protected before
as the with the fact the stats timer didn't have a valid idx.
As the offset is already applied now, we need to update and reschedule
all timers each time the offset is changed. I'm not sure who expect this
to be retro-actively applied, but there was a unit test for it.
If the jitterbuffer head change, there is no need to systematically
wakeup the timer thread. The timer thread will be waken up on if
an earlier timeout has been pushed. This prevent some more spurious
wakeup when the system is loaded. As a side effect, cranking the clock
may set the clock at an earlier position.
In this patch we now make use of the new RtpTimerQueue instead of the
old GArray. This required a lot of changes all over the place, some of
the important changes are that `timer->timeout` is no longer a PTS but
the actual timeout. This was required to get the RtpTimerQueue sorting
right. The applied offset is saved as `timer->offset`, this allow
retreiving back the PTS when needed.
The clockid updates only happens once per incoming packet. If the
currently schedule timer is before the earliest timer in the queue, we
no longer wakeup the thread. This way, if other timers get setup in the
meantime, this will reduce the number of wakup.
The timer loop code has been mostly rewritten, though the behaviour of
running the lost timers first has been kept (even though there is no
test to show what would be the side effect of doing this differently).
Fixes#608
Implement a single timer queue for all timers. The goal is to always use
ordered queues for storing timers. This way, extracting timers for
execution becomes O(1). This also allow separating the clock wait
scheduling from the timer itself and ensure that we only wake up the
timer thread when strictly needed.
The knew data structure is still O(n) on insertions and reschedule,
but we now use proximity optimization so that normal cases should be
really fast. The GList structure is also embeded intot he RtpTimer
structure to reduce the number of allocations.
This moves the RtpJitterBufferStructure type, alloc, free into
rtpjitterbuffer.c/h implementation. jitterbuffer.c strictly rely on
the fact this structure is compatible with GList, and so it make more
sense to keep encapsulate it. Also, anything that could possibly
reduce the amount of code in the element is a win.
In order to support that move, a function pointer to free the data
was added. This also allow making the free function option when
flushing the jitterbuffer.
This helps understanding which function modify the Timerdata
and which one does not. This is not always obvious from thelper
name considering recalculate_timer() does not.
In file included from ../../../../dist/linux_x86_64/include/gstreamer-1.0/gst/gst.h:55,
from ../../../../dist/linux_x86_64/include/gstreamer-1.0/gst/tag/tag.h:25,
from ../gst/isomp4/qtdemux.c:56:
In function ‘qtdemux_inspect_transformation_matrix’,
inlined from ‘qtdemux_parse_trak’ at ../gst/isomp4/qtdemux.c:10676:5,
inlined from ‘qtdemux_parse_tree’ at ../gst/isomp4/qtdemux.c:14210:5:
../../../../dist/linux_x86_64/include/gstreamer-1.0/gst/gstinfo.h:645:5: error: ‘%s’ directive argument is null [-Werror=format-overflow=]
645 | gst_debug_log ((cat), (level), __FILE__, GST_FUNCTION, __LINE__, \
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
646 | (GObject *) (object), __VA_ARGS__); \
| ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
../../../../dist/linux_x86_64/include/gstreamer-1.0/gst/gstinfo.h:1062:35: note: in expansion of macro ‘GST_CAT_LEVEL_LOG’
1062 | #define GST_DEBUG_OBJECT(obj,...) GST_CAT_LEVEL_LOG (GST_CAT_DEFAULT, GST_LEVEL_DEBUG, obj, __VA_ARGS__)
| ^~~~~~~~~~~~~~~~~
../gst/isomp4/qtdemux.c:10294:5: note: in expansion of macro ‘GST_DEBUG_OBJECT’
10294 | GST_DEBUG_OBJECT (qtdemux, "Transformation matrix rotation %s",
| ^~~~~~~~~~~~~~~~
../gst/isomp4/qtdemux.c: In function ‘qtdemux_parse_tree’:
../gst/isomp4/qtdemux.c:10294:64: note: format string is defined here
10294 | GST_DEBUG_OBJECT (qtdemux, "Transformation matrix rotation %s",
| ^~
Add a property which explicitly maps splitmuxsink pads to the
muxer pads they should connect to, overriding the implicit logic
that tries to match pads but yields arbitrary names.
When running in async-finalize mode, request new pads from the muxer
using the same names as old pads, instead of letting the muxer assign
new ones based on the pad template name.
The output segment is only used in ONVIF mode.
The previous behaviour was to output a segment computed from
the Range response sent by the server.
In ONVIF mode, servers will start serving from the appropriate
synchronization point (keyframe), and the Range in response will
start at that position.
This means rtspsrc can now perform truly accurate seeks in that
mode, by clipping the output segment to the values requested in
the seek. The decoder will then discard out of segment buffers
and playback will start without artefacts at the exact requested
position, similar to the behaviour of a demuxer when an accurate
seek is requested.
In push mode (streaming), if the audio size is smaller than segment buffer size, it would be ignored.
This happens because when the plugin receives an EOS signal while a single audio chunk that is less than the segment buffer size is buffered, it does not
flush this chunk. The fix is to flush the data chunk when it receives an EOS signal and has a single (first) chunk buffered.
How to reproduce:
1. Run gst-launch with tcp source
```
gst-launch-1.0 tcpserversrc port=3000 ! wavparse ignore-length=0 ! audioconvert ! filesink location=bug.wav
```
2. Send a wav file with unspecified data chunk length (0). Attached a test file
```
cat test.wav | nc localhost 3000
```
3. Compare the length of the source file and output file
```
ls -l test.wav bug.wav
-rw-rw-r-- 1 amr amr 0 Aug 15 11:07 bug.wav
-rwxrwxr-x 1 amr amr 3564 Aug 15 11:06 test.wav
```
The expected length of the result of the gst-lauch pipeline should be the same as the test file minus the headers (44), which is ```3564 - 44 = 3520``` but the actual output length is ```0```
After the fix:
```
ls -l test.wav fix.wav
-rw-rw-r-- 1 amr amr 3520 Aug 15 11:09 fix.wav
-rwxrwxr-x 1 amr amr 3564 Aug 15 11:06 test.wav
```
If VP8 is not encoded with error resilience enabled then any packet loss
causes very bad artefacts when decoding and waiting for the next
keyframe instead improves user experience considerably.
Various audio formats require an audio lead-in to decode it properly.
Most parsers would take care of it, but when a container like matroska is
involved, the demuxer handles the seeking and without its own lead-in
handling would never even pass the lead-in data to the parser.
This commit provides an initial implementation of that for audio/mpeg,
audio/x-ac3 and audio/x-eac3 by calculating the worst case lead-in time
needed from known samplerate, potential lead-in frames need and the
maximum blocksize possible for the format (as we don't parse that out
exactly in matroskademux) and seeking that much earlier in case of
accurate seeks. This is especially important for NLE use-cases with GES.
If accurate seeking to a position that happens to have a video keyframe,
it'll go back to the previous keyframe than needed, but with typical
video files that's the best we can do anyway without falling back to
scanning the clusters, as typically only keyframes are indexed in
Cueing Data.
If the media doesn't have a CUE, then we bisect for the cluster to seek
to with the same modified time as well in case of accurate seeking,
ensuring sufficient lead-in. This code path is typically hit only with
(suboptimal) audio-only matroska files, e.g. when created with ffmpeg,
which doesn't add a CUE for audio-only mkv muxing.
The send path in rtpsession processes the buffer list along the way,
sharing info and stats between packets in the same list, because it
assumes that all packets in a buffer list are from the same frame.
However, in the receiving path packets can arrive in all sorts of
arrangements:
- different sources,
- different frames (different timestamps),
- different types (multiplexed RTP and RTCP, invalid RTP packets).
so a more general approach should be used to correctly support buffer
lists in the receive path.
It turns out that it's simpler and more robust to process buffers
individually inside the rtpsession element even if they come in a buffer
list, and then reassemble a new buffer list when pushing the buffers
downstream.
This avoids complicating the existing code to make all functions
buffer-list-aware with the risk of introducing regressions,
To support buffer lists in the receive path and reduce the "push
overhead" in the pipeline, a new private field named processed_list is
added to GstRtpSessionPrivate, it is set in the chain_list handler and
used in the process_rtp callback; this is to achieve the following:
- iterate over the incoming buffer list;
- process the packets one by one;
- add the valid ones to a new buffer list;
- push the new buffer list downstream.
The processed_list field is reset before pushing a buffer list to be on
the safe side in case a single buffer was to be pushed by upstream
at some later point.
NOTE:
The proposed modifications do not change the behavior of the send path.
The process_rtp callback is called in rtpsource.c by the push_rtp
callback (via source_push_rtp) only when the source is not internal.
So even though push_rtp is also called in the send path, it won't end up
using process_rtp in this case because the source would be internal in
the send path.
The reasoning from above may suggest a future refactoring: push_rtp
might be split to better differentiate the send and receive path.
In push mode (streaming), if the last audio payload chunk is less than the segment rate buffer size, it would be ignored since the plugin waits until it has at least segment rate bufer size of audio.
The fix is to introduce a flushing flag that indicates that no more audio will be available so that the plugin can recognize this condition and flush the data is has even if it is less
than the desired segment rate buffer size.
This is useful to support the ONVIF case: when is-live is set to
FALSE and onvif-rate-control is no, the client can control the
rate of delivery and arrange for the server to block and still
keep sending when unblocked, without requiring back and forth
PAUSE / PLAY requests. This enables, amongst other things, fast
frame stepping on the client side.
When is-live is FALSE, we don't use a manager at all. This case
was actually already pretty well handled by the current code. The
standard manager, rtpbin, is simply no longer needed in this case.
Applications can instantiate a downloadbuffer after rtspsrc if
needed.
Refactor the code for parsing and generating the Range, taking
advantage of existing API in GstRtspTimeRange.
Only use the TCP protocol in that mode, as per the specification.
Generate an accurate segment when in that mode, and signal to the
depayloader that it should not generate its own segment, through
the "onvif-mode" field in the caps, see
<https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/merge_requests/328>
for more information.
Translate trickmode seek flags to their ONVIF representation
Expose an onvif-rate-control property
Forwarding a single segment event from the pad that first gets
chained is incorrect: when that first event was sent by an element
such as x264enc, with its offset start, we end pushing out of segment
buffers for the other pad(s).
Instead, everytime the active pad changes, forward the appropriate
segment event.
Fixes https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/issues/1028
