Do not use last buffer TS + buffer duration because buffer duration
might be inaccurate, especially for frame rates like 30fps where a
rounding error is observed.
https://bugzilla.gnome.org/show_bug.cgi?id=773785
When doing rtx, the jitterbuffer will always add an rtx-timer for the next
sequence number.
In the case of the packet corresponding to that sequence number arriving,
that same timer will be reused, and simply moved on to wait for the
following sequence number etc.
Once an rtx-timer expires (after all retries), it will be rescheduled as
a lost-timer instead for the same sequence number.
Now, if this particular sequence-number now arrives (after the timer has
become a lost-timer), the reuse mechanism *should* now set a new
rtx-timer for the next sequence number, but the bug is that it does
not change the timer-type, and hence schedules a lost-timer for that
following sequence number, with the result that you will have a very
early lost-event for a packet that might still arrive, and you will
never be able to send any rtx for this packet.
Found by Erlend Graff - erlend@pexip.comhttps://bugzilla.gnome.org/show_bug.cgi?id=773891
The lost-event was using a different time-domain (dts) than the outgoing
buffers (pts). Given certain network-conditions these two would become
sufficiently different and the lost-event contained timestamp/duration
that was really wrong. As an example GstAudioDecoder could produce
a stream that jumps back and forth in time after receiving a lost-event.
The previous behavior calculated the pts (based on the rtptime) inside the
rtp_jitter_buffer_insert function, but now this functionality has been
refactored into a new function rtp_jitter_buffer_calculate_pts that is
called much earlier in the _chain function to make pts available to
various calculations that wrongly used dts previously
(like the lost-event).
There are however two calculations where using dts is the right thing to
do: calculating the receive-jitter and the rtx-round-trip-time, where the
arrival time of the buffer from the network is the right metric
(and is what dts in fact is today).
The patch also adds two tests regarding B-frames or the
“rtptime-going-backwards”-scenario, as there were some concerns that this
patch might break this behavior (which the tests shows it does not).
The new timeout is always going to be (timeout + delay), however, the
old behavior compared the current timeout to just (timeout), basically
being (delay) off.
This would happen if rtx-delay == rtx-retry-timeout, with the result that
a second rtx attempt for any buffers would be scheduled immediately instead
of after rtx-delay ms.
Simply calculate (new_timeout = timeout + delay) and then use that instead.
https://bugzilla.gnome.org/show_bug.cgi?id=773905
Commit 83e718 added a pad template to splitmux request
pads, which means that GstElement now releases the pads on
dispose, but after having removed all elements in the bin
and unlinked them. Make sure we can handle cleanup in that case
without throwing assertions.
https://bugzilla.gnome.org/show_bug.cgi?id=773784
qtdemux.c: In function ‘qtdemux_parse_tree’:
qtdemux.c:10139:16: error: ‘color_table_id’ may be used uninitialized in this function [-Werror=maybe-uninitialized]
if (color_table_id != 0) {
^
qtdemux.c:10121:19: note: ‘color_table_id’ was declared here
guint16 color_table_id;
^~~~~~~~~~~~~~
The ProRes guidelines suggest an interleave of 0.5s is common, but
specifies that for ProRes at most 2MB (for SD) and 4MB (for HD) should
be used per chunk.
It might also make sense to use similar numbers in general.
https://bugzilla.gnome.org/show_bug.cgi?id=773217
Previously we were switching from one chunk to another on every single
buffer. This wastes some space in the headers and, depending on the
software, might depend in more reads (e.g. if the software is reading
multiple samples in one go if they're in the same chunk).
The ProRes guidelines suggest an interleave of 0.5s is common, but
specifies that for ProRes at most 2MB (for SD) and 4MB (for HD) should
be used per chunk. This will be handled in a follow-up commit.
https://bugzilla.gnome.org/show_bug.cgi?id=773217
It's required for ProRes to work with other software.
It is also in the MP4 standard, but inventing values here seems a bit
tricky for the general case and it does not really give any extra
information.
https://bugzilla.gnome.org/show_bug.cgi?id=769048
Some buggy payloaders, e.g. rtph263pay, may use mode B for packets
that starts with a picture (or GOB) start code although it's not
allowed. Let's be nice and not drop these packets/frames.
https://bugzilla.gnome.org/show_bug.cgi?id=773516
Bump the bitstream parsing to TRACE log level so it doesn't flood the
output when trying to read the more useful DEBUG and LOG messages.
Also use GST_DEBUG_OBJECT instead of GST_DEBUG in various places
https://bugzilla.gnome.org/show_bug.cgi?id=773514
Altough commits 6a16be7, 64f9d08 and 0c7e3a8 fixed some issues they
introduced others. This patch fixes the leak of one macroblock for every
B fragment.
Macroblock structures must not be freed immediately after finding the
boundaries as they are stored and used later. However the inital dummy
structure (used for finding the first boundary) must be freed.
CID #1212156https://bugzilla.gnome.org/show_bug.cgi?id=773512
Instead of sending EOS when a source byes we have to wait for
all the sources to be gone, which means they already sent BYE and
were removed from the session. We now handle the EOS in the rtcp
loop checking the amount of sources in the session.
https://bugzilla.gnome.org/show_bug.cgi?id=773218
Improve RFC2326 - chapter C.3 compatibility:
In case just a single stream is specified in SDP and the control attribute
is missing do not drop the stream but rather assume "a=control:*"
https://bugzilla.gnome.org/show_bug.cgi?id=770568
Use the number of milliframes per second for integral and drop-frame
framerates, as suggested by the QT file format specification and other
places. We already did that for integral framerates before, but not for
drop-frame framerates. This now keeps precision better.
For all other framerates, check if it's close to a well-known framerate
and use that instead.
https://bugzilla.gnome.org/show_bug.cgi?id=769041
We consider there's a sifnificant difference when it's larger than on second
or than half the duration of the last processed fragment in case the latter is
larger.
https://bugzilla.gnome.org/show_bug.cgi?id=754230
Modify the caps string to allow width and height greater than 4096.
There is no need to restrict it since the matroska format allows the
width and height values to be up to eight bytes long.
https://bugzilla.gnome.org/show_bug.cgi?id=773582
This solves a hanging mainloop in following scenario:
* connect to source
* network/server drops
* pipeline set to NULL (and connection to flushing as part)
* pipeline set to PAUSED/PLAYING (connection to non-flushing, but not recorded)
* [connecting still not possible]
* pipeline set to NULL => mainloop hangs (since no actual flushing is done)
The pacing of the overall muxing is controlled
by the video GOPs arriving, so we can only handle
1 video stream, and the request pad is named accordingly.
Ignore a request for a 2nd video pad if there's already
an active one.
In file included from ../subprojects/gst-plugins-good/gst/monoscope/gstmonoscope.c:42:0:
../subprojects/gst-plugins-base/gst-libs/gst/audio/audio.h:26:39: fatal error: gst/audio/audio-enumtypes.h: No such file or directory
#include <gst/audio/audio-enumtypes.h>
^
compilation terminated.
https://ci.gstreamer.net/job/GStreamer-master-meson/271/console
Found via the Jenkins CI:
FAILED: subprojects/gst-plugins-good/gst/multifile/gstmultifile@sha/gstsplitmuxsink.c.o
[...]
In file included from ../subprojects/gst-plugins-good/gst/multifile/gstsplitmuxsink.h:24:0,
from ../subprojects/gst-plugins-good/gst/multifile/gstsplitmuxsink.c:59:
../subprojects/gst-plugins-base/gst-libs/gst/pbutils/pbutils.h:30:43: fatal error: gst/pbutils/pbutils-enumtypes.h: No such file or directory
#include <gst/pbutils/pbutils-enumtypes.h>
^
compilation terminated.
https://ci.gstreamer.net/job/GStreamer-master-meson/263/console
If the seek stop point (or start, during reverse play)
was within the segment we just finished, go EOS immediately
instead of proceeding through all other parts and sending
0 length seeks to them.
https://bugzilla.gnome.org/show_bug.cgi?id=772138
When one part moves ahead of the others - due to excessive
downstream queueing, or really small input files - then
we can end up activating parts more than once. That can lead to
effects like shutting down pad tasks prematurely.
https://bugzilla.gnome.org/show_bug.cgi?id=772138
This reverts commit f1ceaab02f.
This broke atomic file writes in "buffer" mode. It did make
sure that any streamheaders are prepended to each file in
buffer mode as well, but that's not really needed in practice,
whereas atomic file writes are, so let's restore the status
quo ante for now since this was primarily a code cleanup anyway,
and if anyone needs to streamheaders in buffer mode too they
can make a patch to implement that differently. Re-implementing
the atomic writes in the element also seems way too much work.
https://bugzilla.gnome.org/show_bug.cgi?id=766990
We were just picking the timestamp of the last buffer pushed into our
adapter before we had enough data to push out.
This fixes things to figure out how large each frame is and what
duration it covers, so we can set both the timestamp and duration
correctly.
Also adds some DISCONT handling.
The basic idea is this:
1. For *larger* rtx-rtt, weigh a new measurement as before
2. For *smaller* rtx-rtt, be a bit more conservative and weigh a bit less
3. For very large measurements, consider them "outliers"
and count them a lot less
The idea being that reducing the rtx-rtt is much more harmful then
increasing it, since we don't want to be underestimating the rtt of the
network, and when using this number to estimate the latency you need for
you jitterbuffer, you would rather want it to be a bit larger then a bit
smaller, potentially losing rtx-packets. The "outlier-detector" is there
to prevent a single skewed measurement to affect the outcome too much.
On wireless networks, these are surprisingly common.
https://bugzilla.gnome.org/show_bug.cgi?id=769768
Assuming equidistant packet spacing when that's not true leads to more
loss than necessary in the case of reordering and jitter. Typically this
is true for video where one frame often consists of multiple packets
with the same rtp timestamp. In this case it's better to assume that the
missing packets have the same timestamp as the last received packet, so
that the scheduled lost timer does not time out too early causing the
packets to be considered lost even though they may arrive in time.
https://bugzilla.gnome.org/show_bug.cgi?id=769768
There is no need to schedule another EXPECTED timer if we're already
past the retry period. Under normal operation this won't happen, but if
there are more timers than the jitterbuffer is able to process in
real-time, scheduling more timers will just make the situation worse.
Instead, consider this packet as lost and move on. This scenario can
occur with high loss rate, low rtt and high configured latency.
https://bugzilla.gnome.org/show_bug.cgi?id=769768
This patch fixes an issue with the estimated gap duration when there is
a gap immediately after a lost timer has been processed. Previously
there was a discrepancy beteen the gap in seqnum and gap in dts which
would cause wrong calculated duration. The issue would only be seen with
retranmission enabled since when it's disabled lost timers are only
created when a packet is received and the actual gap length and last dts
is known.
https://bugzilla.gnome.org/show_bug.cgi?id=769768
Stats should also be collected for unsuccessful packets.
rtx-rtt is very important for determining the necessary configured
latency on the jitterbuffer. It's especially important to be able to
increase the latency when retransmitted packets arrive too late and are
considered lost. This patch includes these late packets in the
calculation of the various rtx stats, making them more correct and
useful.
Also in the case where the original packet arrives after a NACK is sent,
the received RTX packet should update the stats since it provides useful
information about RTT.
The RTT is only updated if and only if all requested retranmissions are
received. That way the RTT is guaranteed to make sense. If not we don't
know which request the packet is a response to and the RTT may be bogus.
A consequence of this patch is that RTT is not updated for a request
when one of the RTX packets for that seqnum is lost, but that since
measured RTT will be more accurate.
The implementation store the RTX information from the timed out timers
and use this when the retransmitted packet arrives. For performance
these timers are stored separately from the "normal" timers in order to
not impact performance (see attached performance test).
https://bugzilla.gnome.org/show_bug.cgi?id=769768
When disabled we can save some iterations over timers.
There is probably an argument for rtx-delay-reorder to exist, but
for normal operations, handling jitter (reordering) is something a
jitterbuffer should do, and this variable feels like functionality that
is not "in-sync" with what the jitterbuffer is trying to achieve.
Example: You have 50ms jitter on your network, and are receiving
audio packets with 10ms durations. An audio packet should not be
considered late until its rtx-timeout has expired (and hence a rtx-event
is sent), but with rtx-delay-reorder, events will be sent pretty much
all the time due to the jitter on the network.
Point being: The jitterbuffer should adapt its size to the measured network
jitter, and then rtx-delay-reorder needs to adapt as well, or simply
get out of the way and let the other (better) rtx-mechanisms do their job.
Also change find_timer to only use seqnum as an argument, since there
will only ever be one timer per seqnum at any given time. In the
one case where the type matters, the caller simply checks the type.
https://bugzilla.gnome.org/show_bug.cgi?id=769768
And actually calculate the field duration instead of a frame duration so
that we can properly timestamp output frames in fields=all mode.
This is probably still broken for reverse playback in telecine mode.
This may cause a few packets to be processed by the parser, but it's
better than never pushing out buffers from a slightly broken stream
where no marker bits are set.
To be able to cap the number of allowed streams for one session.
This is useful for preventing DoS attacks, where a sender can change
SSRC for every buffer, effectively bringing rtpbin to a halt.
https://bugzilla.gnome.org/show_bug.cgi?id=770292
Under certain conditions gst_rtp_buffer_get_payload() returns a copy of
the payload. In this case the payload modifications will not affect the
rtp buffer. So instead of modifying the payload buffer directly we
should modify the buffer that actually gets pushed on the adapter.
