This adds a property to select the maximum number of threads to use for
conversion and scaling. During processing, each plane is split into
an equal number of consecutive lines that are then processed by each
thread.
During tests, this gave up to 1.8x speedup with 2 threads and up to 3.2x
speedup with 4 threads when converting e.g. 1080p to 4k in v210.
https://bugzilla.gnome.org/show_bug.cgi?id=778974
See https://bugzilla.gnome.org/show_bug.cgi?id=773666
This would ideally be solved in baseparse but that requires further
thought at this point, and in the meantime it would be good to have
rawbaseparse not assert on this but handle it gracefully instead.
Probe for MultiQueue source pad might receive EOS twice,
the first is fake-eos and the other is actual EOS.
And the slot can be freed with fake-eos/EOS if the slot has no input.
Since slot freeing is async, double free can be possible.
So, decodebin3 needs to remove the probe also with slot freeing.
https://bugzilla.gnome.org/show_bug.cgi?id=777530
"requested_selection" list might be generated by select-streams event.
And memory of stream-id(s) in select-streams is independent from that of stream-collection.
https://bugzilla.gnome.org/show_bug.cgi?id=775553
The latency query originally had a fallthrough to the default
label at the end as fallback, but that got messed up when the
DURATION and POSITION queries were added, so it then fell through
to the duration query handler instead. Restore original behaviour.
https://bugzilla.gnome.org/show_bug.cgi?id=699077
Duration query would return TRUE and duration=-1. This
worked in the unit test because the unit test implementation
was a bit broken.
Both queries need to access rate with a lock.
Fix broken duration query test as well. It relied on broken
behaviour by the videorate query handler, and also it was
implemented as a downstream query rather than an upstream
query. And we must return HANDLED from the probe so that the
query we intercept actually returns TRUE.
https://bugzilla.gnome.org/show_bug.cgi?id=699077
When the decodebin state change fails because of an error
message, we might not go through PAUSED->READY. Don't leak
a ref to decodebin pads due to pad blocking in that case.
This is because we return ASYNC going to PAUSED, and if
we fail before reaching PAUSED the only transition we'll
see is READY->NULL.
https://bugzilla.gnome.org/show_bug.cgi?id=775893
This adds some extra options that affect pattern=ball mode, allowing the
animation to be synced to running time or wall-time clock for comparing
sync across different instances / pipelines / machines.
Also added is the ability to invert the rendering colours every second,
and some different ball motion patterns.
https://bugzilla.gnome.org/show_bug.cgi?id=740557
The state of urisourcebin (and all elements contained within) can
change at any point in time, including when setting up the typefind
element.
In order to avoid ending up with typefind starting without being fully
connected, lock the state and connect to the 'have-type' signal.
Due to the special nature of adaptivedemux, reconfigure happens
frequently with seek/track-change.
In very exceptional cases, the following sequence is possible:
* EOS event is pushed to queue element and still buffers are queued
* During draining remaining buffers, reconfiguration downstream
happens due to track switch.
* The queue gets a not-linked flow return from downstream
* Because the sinkpad is EOS, the queue registers an
error on the bus, causing the pipeline to fail.
Avoid the sinkpad getting marked EOS in the first place, by using a
custom event in place of EOS.
https://bugzilla.gnome.org/show_bug.cgi?id=777009
When shutting down decodebin2 and parsebin, they set their
output pads to flushing, and there is a very small window
where elements might send a sticky event such as a tag event
(which silently fails due to flushing) and then sends a buffer,
and the buffer will return GST_FLOW_ERROR because it can't
forward sticky events. The element will then send an error
message on the bus. This can also happen when elements send EOS
just as shutdown is happening. Since we're about to destroy all
the elements inside parsebin and decodebin anyway, just discard
error messages from them.
A nicer but more difficult fix for GStreamer 2.0 is to make
all event pushing / handling in core return a GstFlowReturn
like buffers do, so we can report a FLUSHING state cleanly.
Make sure ticks start with an accumulator value of 0 by incrementing it
after filling in samples instead of before and by resetting the accumulator
every time a tick begins. This prevents it from being discontinuous at the
beginning of the tick.
https://bugzilla.gnome.org/show_bug.cgi?id=774050
When plugging and then exposing a parser, don't fail
if it fails to send sticky events. The most likely
reason is that things were flushed due to the app
immediately doing a seek, but we can't detect flushing
separately to other error conditions without a
gst_pad_send_event_full() core function that returns
a GstFlowReturn.
In some case we might have EncodingProfile that will be defined
in a way that, for example if a Preset is not present, another
profile for that stream should be used.
A test is added showing the feature.
https://bugzilla.gnome.org/show_bug.cgi?id=776188
There are cases when there is no demuxer involved that could do the
buffering, e.g. HLS with raw MP3 or AAC. In this case we want to place
the buffering multiqueue after the parser.
Before this change, we've considered the first element after the
adaptive streaming demuxer as a parser. This is not always true, e.g.
id3demux. Instead we now wait until we actually have a parser (or
decoder).
Fixes playback on such HLS streams.
Compositor does not support it currently and it needs special support
for handling this correctly, and is rather non-trivial to implement for
all formats.
Playbin3 takes lock when querying duration and handling
stream-collection message. So,to post stream-collection message,
duration query should be dropped when input pad is being unlinked.
https://bugzilla.gnome.org/show_bug.cgi?id=773341
max_buffer_usage is the index of the oldest buffer in the queue,
starting at zero, not the number of buffers queued.
find_limits returns the index of the oldest buffer that satisfies the
limits in its min_idx parameter, not the number of buffers needed. Fix
this use too in order to keep passing the tests that read
buffers-queued.
https://bugzilla.gnome.org/show_bug.cgi?id=775351
If a client gets dropped and the iteration gets restarted, bufpos is
incremented again for all clients that preceded the dropped one, causing
havoc.
Adjust the bufpos for all clients first before trying to drop any.
https://bugzilla.gnome.org/show_bug.cgi?id=774908
Optimize LE<->BE conversion by adding a dedicated fast path instead of
using the generic converter. Implement transform_ip function in order to do the
endian swap in place.
This saves buffer allocation for the intermediate format, can be done in place
and also performs the conversion in one step instead of unpack-convert-pack.
For all bit widths the naive algorithm is implemented, which provides the best
performance when compiled with -O3. ORC was considered but eventually removed
as it requires a dedicated function for in-place conversion (due to the
"restrict" parameters).
A more complex algorithm for the 24-bit conversion with unrolled loop and
32-bit processing is implemented in the #if 0 section. It performs better if
compiled with -O2. With -O3 however the naive algorithm performs better.
https://bugzilla.gnome.org/show_bug.cgi?id=773073
For frame->buffer, baseparse is doing that automatically for us. For
frame->output_buffer it doesn't and assumes that the subclass is already
doing that. Consistency!
Deterministic generation of snow and smpte is important for tests so
that it's not affected by other videotestsrc elements in current or
possibly previous tests.
https://bugzilla.gnome.org/show_bug.cgi?id=773102
find_suitable_mask() had complexity O(n^2) on the number of bits.
For common case like 2-channel audio the mask was calculated in about 4k loop
cycles.
Optimize both n_bits_set() and find_suitable_mask() to O(n) where n is the
number of bits set in the mask.
https://bugzilla.gnome.org/show_bug.cgi?id=772864
rawvideoparse wouldn't error out on not-negotiated,
but would just keep on going, because it didn't pass
the flow return value back to the parent class and
thus upstream, so the source wouldnt' stop streaming.
We have to calculate from the segment.stop, not the segment.start, as
playback goes from stop to start. This fix works around another race
condition in streamsynchronizer in my testcase.
See https://bugzilla.gnome.org/show_bug.cgi?id=771479
When connecting a demuxer through a multiqueue ensure to copy sticky
events in order to allow the following factory being properly
checked that it is functional.
https://bugzilla.gnome.org/show_bug.cgi?id=769580
https://github.com/mesonbuild/meson
With contributions from:
Tim-Philipp Müller <tim@centricular.com>
Matej Knopp <matej.knopp@gmail.com>
Jussi Pakkanen <jpakkane@gmail.com> (original port)
Highlights of the features provided are:
* Faster builds on Linux (~40-50% faster)
* The ability to build with MSVC on Windows
* Generate Visual Studio project files
* Generate XCode project files
* Much faster builds on Windows (on-par with Linux)
* Seriously fast configure and building on embedded
... and many more. For more details see:
http://blog.nirbheek.in/2016/05/gstreamer-and-meson-new-hope.htmlhttp://blog.nirbheek.in/2016/07/building-and-developing-gstreamer-using.html
Building with Meson should work on both Linux and Windows, but may
need a few more tweaks on other operating systems.
https://github.com/mesonbuild/meson
With contributions from:
Tim-Philipp Müller <tim@centricular.com>
Matej Knopp <matej.knopp@gmail.com>
Jussi Pakkanen <jpakkane@gmail.com> (original port)
Highlights of the features provided are:
* Faster builds on Linux (~40-50% faster)
* The ability to build with MSVC on Windows
* Generate Visual Studio project files
* Generate XCode project files
* Much faster builds on Windows (on-par with Linux)
* Seriously fast configure and building on embedded
... and many more. For more details see:
http://blog.nirbheek.in/2016/05/gstreamer-and-meson-new-hope.htmlhttp://blog.nirbheek.in/2016/07/building-and-developing-gstreamer-using.html
Building with Meson should work on both Linux and Windows, but may
need a few more tweaks on other operating systems.
https://github.com/mesonbuild/meson
With contributions from:
Tim-Philipp Müller <tim@centricular.com>
Jussi Pakkanen <jpakkane@gmail.com> (original port)
Highlights of the features provided are:
* Faster builds on Linux (~40-50% faster)
* The ability to build with MSVC on Windows
* Generate Visual Studio project files
* Generate XCode project files
* Much faster builds on Windows (on-par with Linux)
* Seriously fast configure and building on embedded
... and many more. For more details see:
http://blog.nirbheek.in/2016/05/gstreamer-and-meson-new-hope.htmlhttp://blog.nirbheek.in/2016/07/building-and-developing-gstreamer-using.html
Building with Meson should work on both Linux and Windows, but may
need a few more tweaks on other operating systems.
This is enough for making it work in GES, but it's unclear if all the various
property combinations are working correctly. It's an improvement over what was
there before in any case, which was to just drop all buffers if rate < 0.0.
https://bugzilla.gnome.org/show_bug.cgi?id=769624
When processing EOS for a pad, send a stream-group-done
for the pad in case downstream is waiting for more
data on this stream before it can process related
streams from the group.
https://bugzilla.gnome.org/show_bug.cgi?id=768995
My collection leak fix 83f30627cd
introduced a crash in this scenario: audiotestsrc ! decodebin3 ! fakesink
The reference handling of collection in decodebin3 wasn't very clear and
my attempt to fix the leak introduced a regression where we went one
reference short in some other scenarios.
Fixing this by:
- Giving a strong reference to DecodebinInput making things clearer
- Fixing get_merged_collection() which was sometimes returning an
existing reference and sometimes a new one.
https://bugzilla.gnome.org/show_bug.cgi?id=769080
After we reset the resampler, there is no history anymore in the resampler
and the previously calculated output size is no longer valid.
Recalculate the new output size after a reset to make sure we don't try
to convert too much.
The collection owned by GstDecodebin3 has to be unreffed when disposing.
gst_event_new_stream_collection() doesn't consume the collection passed
to it so no need to give it an extra ref.
https://bugzilla.gnome.org/show_bug.cgi?id=768811
MultiQueueSlot owns a ref on the active stream so it should release it
when being freed.
DecodebinInputStream owns ref on the active and pending stream so they
should be dropped when being freed.
https://bugzilla.gnome.org/show_bug.cgi?id=768811
gst_stream_get_caps() returns a reffed caps.
The caps passed to gst_query_set_caps_result() are not transfered.
The caps in gst_parse_pad_stream_start_event() was either acquired
using gst_pad_get_current_caps() which returns a new ref or
explicitly reffed.
https://bugzilla.gnome.org/show_bug.cgi?id=768811
When a discont buffer is processed, the state is re-initialized, which
nullifies the allowed_tags.
The problem is when a subrip string with tags is processed and allowed_tags is
NULL. The function subrip_unescape_formatting() calls g_strjoinv with a
str_array as NULL, leading to a GLib-CRITICAL.
This patch removes the allowed_tags resetting, in parser_state_init(), but
move it into gst_sub_parse_format_autodetect().
https://bugzilla.gnome.org/show_bug.cgi?id=768525
With contributions from Jan Schmidt <jan@centricular.com>
* decodebin3 and playbin3 have the same purpose as the decodebin and
playbin elements, except make usage of more 1.x features and the new
GstStream API. This allows them to be more memory/cpu efficient.
* parsebin is a new element that demuxers/depayloads/parses an incoming
stream and exposes elementary streams. It is used by decodebin3.
It also automatically creates GstStream and GstStreamCollection for
elements that don't natively create them and sends the corresponding
events and messages
* Any application using playbin can use playbin3 by setting the env
variable USE_PLAYBIN3=1 without reconfiguration/recompilation.
We take a ref before removing which was never freeded.
The element is still alive anyway because the group has its own ref as
well.
Fix a leak with the 'test_suburi_error_wrongproto' test.
https://bugzilla.gnome.org/show_bug.cgi?id=766515
As is done everywhere else, and avoids setting bogus values
And remove useless *<val> checks (we always provide valid values and
it's an internal function).
CID #1320700
This helps in cases where raw audio data is being delivered, but the
buffers do not come in sample aligned sizes. The new unalignedaudioparse
bin can be autoplugged and configures an internal audioparse element to
align the data. audioparse itself gets support for audio/x-unaligned-raw
input caps; the output caps then contain the same information, except that
the name is changed to audio/x-raw (since audioparse aligns the data).
This ensures that souphttpsrc ! audioparse still works.
https://bugzilla.gnome.org/show_bug.cgi?id=689460
When we initialize an element in decodebin, we 1) set it to PAUSED and
push sticky events on its sinkpad to trigger negotiation 2) block its
src pad(s) to detect CAPS events. We can't block before 1) as that
would lead to a deadlock.
It's possible (and common) tho that an element configures its srcpad
during 1) and before 2). Therefore before this change we would
typically block and expose an element's pad only once the element
output its first buffer, triggering sticky events to be resent. One
consequence of this behaviour is that it sometimes broke
renegotiation.
With this change now we consider a pad ready to be exposed when it's
->blocked or has fixed caps (which were set before we could block it).
https://bugzilla.gnome.org/show_bug.cgi?id=765456
If we are configured to use buffering and there is no demuxer in the chain, we
still want a multiqueue, otherwise we will ignore the use-buffering property.
In that case, we will insert a multiqueue after the parser or decoder - not
elsewhere, otherwise we won't have timestamps.
https://bugzilla.gnome.org/show_bug.cgi?id=764948
gstsubparse.c: In function ‘parse_subrip’:
gstsubparse.c:988:7: error: ignoring return value of ‘strtol’, declared with attribute warn_unused_result [-Werror=unused-result]
cc1: all warnings being treated as errors
https://bugzilla.gnome.org/show_bug.cgi?id=765042
When blocking the subtitle pad, it's expected that stream-start
is the first event, and that it can precede caps arriving on the
peer pad - in fact the caps can only have arrived on the peer
pad when it was pre-primed with sticky events previously.
Instead, just pass the stream-start and don't block, because
stream-start is sticky anyway.
Don't require a cue identifier preceding the time range line
when parsing WebVTT. We could also store the CueID, but it's
not using anywhere, so just ignore it for now.
Make writable the buffer before pushing it lead to a buffer copy. It's
because a reference is keep for the previous buffer.
The previous buffer reference is only need to duplicate the buffer. In
drop-only mode, the previous buffer is release just after pushing the
buffer so a copy is done but it's useless.
https://bugzilla.gnome.org/show_bug.cgi?id=764319
Insert extra checks for the validity of the incoming
data when parsing subrip/webvtt content and debug log
output for invalid content.
Should fix Coverity warnings.
Remove some unused variables from the inner product functions.
Make filter coefficients by interpolating if required.
Rename some fields.
Try hard to not recalculate filters when just chaging the rate.
Add more proprties to audioresample.
Remove the consumed/produced output fields from the resampler and
converter. Let the caler specify the right number of input/output
samples so we can be more optimal.
Use just one function to update the converter configuration.
Simplify some things internally.
Make it possible to use writable input as temp space in audioconvert.
WebVTT is a new subtitle format for HTML5 video. In this first
version of the parser the cue settings are parsed but only stored in
the internal parser state structure. Later on these settings could be
part of the GstBuffer metadata.
https://bugzilla.gnome.org/show_bug.cgi?id=629764
There's a small window between decodebin choosing a buffering level
to post and another thread choosing a different buffering level
where things can race. Close that window by holding a new lock
that's only for posting buffering messages - like what was done
in multiqueue.
https://bugzilla.gnome.org/show_bug.cgi?id=764020
In check_upstream_seekable function, it returns FALSE value even though
we already declare about the seekable variable. So, This patch return
result of seekable in check_upstream_seekable function.
https://bugzilla.gnome.org/show_bug.cgi?id=763975
Due to transient locked state during autoplugging, some elements might be
ignored by the GstBin::change_state() and might still be running. Which could
then cause pad-added and similar accessing decodebin state that does not exist
anymore, and crash.
https://bugzilla.gnome.org/show_bug.cgi?id=763625
And also consider HEADER buffers without DELTA_UNIT flag as sync points. This
fixes sync-mode=2 with mpegtsmux for example, which has no streamheaders but
puts the HEADER flag on its keyframes.
https://bugzilla.gnome.org/show_bug.cgi?id=763278
In other places we lock it the other way around, leading to possible
deadlocks. Also this will deadlock if analyze_pad() causes a new element to be
autoplugged that adds new pads on itself when its state is changed.
https://bugzilla.gnome.org/show_bug.cgi?id=763491
This reverts commit 0615794300.
deinterlace was ported at some point in the last 4 years and has better video
format support, and especially better negotiation than avdeinterlace. Having
avdeinterlace but not deinterlace causes various problems in zerocopy
scenarios.
https://bugzilla.gnome.org/show_bug.cgi?id=760553
libgstreamer currently exports some debug category
symbols GST_CAT_*, but those are not declared in any
public headers.
Some plugins and libgstvideo just use GST_DEBUG_CATEGORY_EXTERN()
to declare and use those, but that's just not right at
all, and it won't work on Windows with MSVC. Instead look
up the categories via the API.
Avoids some false positives leading to miss identification:
* Prevent picture start code emulation for the first 2 bytes read
* Add check for valid "picture coding type" and "PB-frames mode" combination
Additionally, change name on confusingly named TR var to what
it is, the layer's PTYPE.
https://bugzilla.gnome.org/show_bug.cgi?id=693263
When getting caps of the decode chain, in get_topology, the caps are being
checked if fixed or not. But get_topology will be called when the decode is
chain is being exposed and hence it will always be fixed. Hence removing the
check for fixed caps. Removing gst_pad_get_current_caps for the chain->pad, as
get_pad_caps will again call the same api.
And get_topology can return NULL value if currently shutting down the
pipeline, which on being passed to create message will result in assertion
error. Check if topology is valid before using it
https://bugzilla.gnome.org/show_bug.cgi?id=755918
Avoid overflow in rate calculation. This can cause the resampler to
start on the wrong phase after a rate change.
Avoid overflow in cubic fraction calculation. This can cause noise when
dealing with higher samplerates.
Allows the subclass to completely override the chosen src caps.
This is needed as videoaggregator generally has no idea exactly
what operation is being performed.
- Adds a fixate_caps vfunc for fixation
- Merges gst_video_aggregator_update_converters() into
gst_videoaggregator_update_src_caps() as we need some of its info
for proper caps handling.
- Pass the downstream caps to the update_caps vfunc
https://bugzilla.gnome.org/show_bug.cgi?id=756207
analyze_new_pad() can return a new decode chain, which might have a new
GstDecodePad in the end. We should use those two for expose_pad() and not the
original ones that were passed to analyze_new_pad().
This fails when having a demuxer element that has raw pads immediately or
if a decoder with raw caps is after an adaptive demuxer.
https://bugzilla.gnome.org/show_bug.cgi?id=760949
It's useful enough already to be used in other elements for audio aggregation,
let's give people the opportunity to use it and give it some API testing.
https://bugzilla.gnome.org/show_bug.cgi?id=760733
[..] when resetting group start time. In GES, we are usually connected
to the streamsynchronizer on one audio and one video pad.
When seeking the timeline, both nlecompositions often output their flush_start
before any of them has output its flush_stop.
The current code, when receiving the first flush stop was using the
running time of the start of the second composition, which could
be pretty much anything, and means nothing at that point.
This patch is thread-safe, as STREAM_SYNCHRONIZER_LOCK is taken
both when setting flushing and when checking it.
https://bugzilla.gnome.org/show_bug.cgi?id=750013
When blocking input pads, we also need to properly set the appropriate
pending flag.
Without this, when switching stream types after initial configuration
(like going from Audio+Video to Audio+Video+Sub) playsink would never
wait for *all* input streams to be blocked (it would just wait for the
new input pad (text in this case) to be blocked).
Since the reconfiguration might introduce unlinking/relinking of elements,
we need to ensure that *ALL* input streams are blocked.
Failure to do so would result in having some input streams pushing data
to inactive elements (returning GST_FLOW_FLUSHING) or unlinked pads
(returning GST_FLOW_NOT_LINKED).
A later optimization could involve only blocking the input pads that
might be involved in reconfiguration. But better be safe than sorry for
now :)
Elements usually require that all fields on their caps are present
on the fixed caps they receive. Using intersection won't verify it,
resort to using is_subset() checks.
https://bugzilla.gnome.org/show_bug.cgi?id=760477
Those accept caps are actually checking if downstream supports
some particular caps to check if it need to negotiate a different
format. Checking only the next element with accept-caps is not enough
to guarantee that it is supported.
Using a caps query makes it obtain the supported caps for downstream
as a whole instead of only the next element.
Pass flags in _converter_new() so that we can configure ourselves
differently depending on some options.
SOURCE_WRITABLE -> IN_WRITABLE because the array is called 'in'
Simplify the API, we don't need the consumed and produced output
arguments. The caller needs to use the _get_in_frames/get_out_frames API
to check how much input is needed and how much output will be produced.
In this specific case it wouldn't cause problems as we only ever access the
first array element, but let's make explicit what is happening here.
CID 1346530 and 1346529
The filters' floating references are sinked during set_property() already,
which means that GstBin takes a new reference when adding the filter to it.
Get rid of the additional reference after adding the filter to the bin.
Unconditionally adding the template caps when proxying the caps query will play
havoc with decoders that attempt to choose an output format based on some caps
features. Creating a sink that does not include those caps features and a
decoder/parser/etc that preferentially chooses some specific caps feature when
available, will always return the decoder/parser/etc template caps and choose a
feature that downstream will be unable to support.
Fix by limiting the addition of the template caps to when the result is actually
empty.
https://bugzilla.gnome.org/show_bug.cgi?id=758212
This reverts commit 77dc09c3a9.
It can cause the FLUSH_START/STOP events to go to the sink elements, which
then causes state changes and various other problems. We shouldn't really
flush downstream here, the idea is to do *draining*.
Apart from that the testcase for the original bug here works without this
commit now.
Since the loops increasing count from 0 are always run at least
once (if count < 1), count will always be at least one when
compared to the drop/dup conditions.
Coverity 1139674
rename gst-launch --> gst-launch-1.0
replace old elements with new elements(ffmpegcolorspace -> videoconvert, ffenc_** -> avenc_**)
fix caps in examples
https://bugzilla.gnome.org/show_bug.cgi?id=759432
Add a property and logic to send a GstNetworkMessage event containing
the message that was received from a client. This can be used to
implement simply bidirectional communication.
Add a property and logic to send a GstNetworkMessageDispatched
event upstream to notify that a buffer has been sent. This can be used
to keep track of what client received what buffers.
Add a property to handle GstNetworkMessage events. These events contain
a buffer that is sent on the socket to allow for simple bidirectional
communication.
Otherwise we'll remove that element while keeping its buffering message in our
list, and because of that never ever report buffering 100% as that element
will always be at a lower percentage.
This fixes e.g. seeking over Period boundaries in DASH and various other
issues when buffering happens between group switches.
Also use a new mutex for protecting the buffering messages. The object lock is
already used by gst_object_has_as_ancestor() and we need to use it now for
checking if the buffering message sender has the to-be-removed element as
ancestor.
When we stop sending because we need more data, still keep a GSource
around to receive data from the clients.
Also handle read and write in the same go.
Make sure that any access to the GstDecodeBin's decode_chain
field is protected using the EXPOSE_LOCK. Also add a simple
reference counter to the GstDecodeChain structure so that when
the type_found signal fires it can hold onto the decode chain
even while the EXPOSE_LOCK is not held. This should fix a
race condition if the type_found signal fires right in the
middle of a state change that messes with the same decode
chain.
https://bugzilla.gnome.org/show_bug.cgi?id=755260
Just setting the ghostpad as flushing wasn't enough. It needs to be
consistent on the internal proxypad also, otherwise you end up in
situations where:
* a pending buffer on the target pad triggers the sticky event
propagation
* the default implementation sees that the proxypad is not flushing,
so it tries to push it to the other pad (the actual ghostpad)
* the ghostpad is flushing, so returns FALSE
* the push_event function sees that pushing the event failed...
* ... and pending buffer push returns GST_FLOW_ERROR, instead of
GST_FLOW_FLUSHING
By using gst_pad_set_active(FALSE), we ensure that both the ghostpad
and the proxypad are flushing/deactivated. The situation above will
no longer occur, and a GST_FLOW_FLUSHING will be returned.
Remove the format and layout from the mix_samples function and use the
format when creating the channel mixer object. Also use a flag to handle
the unlikely case of non-interleaved samples like we do elsewhere.
Add docs for the internal audioconvert object before moving it to the
audio library.
Remove get_sizes and implement the trivial logic in the element.
Remove some unused orc functions
Move the audio quantize code from audioconvert to the audio library.
work on making an audio converter helper function similar to the video
converter.
Fold fastrandom directly into the quantizer, add some ORC code to
optimize this later.
Rename _get_default_mask() to _get_fallback_mask() to make it more
clear that the function only provides a fallback if nothing else can be
done. Also clarify this in the documentation.
API: gst_audio_channel_get_fallback_mask()
In some conditions we might process empty buffers, calling
gst_control_binding_get_value_array in that case will lead
to the assertion:
(lt-ges-launch-1.0:18859): GStreamer-CRITICAL **: gst_control_binding_get_value_array: assertion 'values' failed
Add a TRUNCATE_RANGE flag for unpack functions to fill the least
significate bits with 0 (as did the old code). Also add functions
that don't truncate. Use the TRUNC flag in audioconvert for
backwards compatibility for now.
Use (1 << 31) as the multiplier for int<->float conversions. This makes
sure that int->float conversions always end up with floats between
[-1.0, 1.0].
For the conversion from float to int, this multiplier will give the complete
int range after we perform clipping.
Change the unit test to take this into consideration.
Fixes https://bugzilla.gnome.org/show_bug.cgi?id=755301
The client-removed signal used G_INT_TYPE instead of G_SOCKET_TYPE
in its definition leading to problems on platforms where the size
of a pointer is larger than the size of an integer, It would also
not work at all with dynamic language bindings.
https://bugzilla.gnome.org/show_bug.cgi?id=757155
Rewrite audioconvert to try to make it more clear what steps are
executed during conversion.
Add passthrough step that just does a memcpy when possible.
Add ORC optimized dither and quantization functions.
Implement noise-shaping on S32 samples only and allow for arbitrary
noise shaping coefficients if we want this later.
This makes sure that they will always get SEEK events, even if we're currently
in the middle of a group switch (i.e. switching to another
representation/bitrate/etc).
https://bugzilla.gnome.org/show_bug.cgi?id=606382
As stated in GST_PAD_PROBE_HANDLED's documentation, we are
supposed to unref the event before returning.
Fixes an event leak in the validate.hls.playback.play_15s.hls_bibbop
validate scenario.
https://bugzilla.gnome.org/show_bug.cgi?id=754459
Rework the converter to use the pack/unpack functions
Because the unpack functions can only unpack to 1 format, add a separate
conversion step for doubles when the unpack function produces int.
Do conversion to S32 in the quantize function directly.
Tweak the conversion factor for doing float->int conversion slightly to
get the full range of negative samples, use clamp to make sure we don't
exceed our int range on the positive axis (see also #755301)
Send event directly to playsink instead of letting GstBin iterate
over all sink elements. The latter might send the event multiple times
in case the SEEK causes a reconfiguration of the pipeline, as can easily
happen with adaptive streaming demuxers.
What would then happen is that the iterator would be reset, we send the
event again, and on the second time it will fail in the majority of cases
because the pipeline is still being reconfigured
The logic introduced by
[d50b713: decodebin: set the decode pad target before setting elements to PAUSED]
to expose pads would only ever be able to possibly expose one (the last) pad per element.
Make it so that any exposable pads are able to be exposed rather than just the
last pad returned by connect_element.
https://bugzilla.gnome.org/show_bug.cgi?id=742924
In the case of analyzing a demuxer chain, analyze_new_pad may create
a new GstDecodeChain. This was not propagated to the calling function which as
of [d50b713f decodebin: set the decode pad target before setting elements to PAUSED]
is now required to be able to expose the correct pad.
https://bugzilla.gnome.org/show_bug.cgi?id=742924
In case of reconfiguration, text_pad should be re-connected with
stream synchronizer sink pad. Otherwise we'll leave an unlinked pad around if
there always was a streamsynchronizer text pad.
https://bugzilla.gnome.org/show_bug.cgi?id=756804
Otherwise caps and context queries will disappear into nothing and therefore
fail. With autoplug-query now actually working, users (such as playbin) can
proxy these queries to the selected video sink and be able to select an
more appropriate configuration.
https://bugzilla.gnome.org/show_bug.cgi?id=731204
In case sink implements a streamvolume interface, volume element is being got
from the sink. But this is transfer full. So the memory should be freed before
setting it to NULL. This was resulting in major memory leaks
https://bugzilla.gnome.org/show_bug.cgi?id=755867
New subclass with a similar behaviour as the old liveadder, but
a slightly different API as the latency is in nanoseconds, not
milliseconds. Also, the new liveadder has a effective latency that
is latency + output-buffer-duration. In practice, just setting a non-zero
latency with the new audiomixer gives you the right behavior in 99% of the
cases.
Build error due to wrong argument type in debug message
aagg->priv->offset and next_offset are of type int64, but uint64
formatter is being used in logs. Changing all those to int64
https://bugzilla.gnome.org/show_bug.cgi?id=756065
Allows to run such a command line :
gst-launch-1.0 uridecodebin uri=file:///home/meh/Music/sthg.mp4 ! \
encodebin profile-string="audio/x-wav|1" ! filesink location=sthg.wav
Previously the code failed because wavenc is considered as a muxer.
We still want encodebin to audio/x-wav as an AudioEncodingProfile,
so this simple fix allows that.
Ability to mux raw streams in containers such as matroskamux
is a different issue.
https://bugzilla.gnome.org/show_bug.cgi?id=751470
intersection with a downstream that accepts any video/x-raw caps
with no further detail won't create a framerate field. If it's
not in the caps, don't fixate it, just set it to 30/1
We might've queued up a GAP buffer that is only partially inside the current
output buffer (i.e. we received it too late!). In that case we should only
skip the part of the GAP buffer that is inside the current output buffer, not
also the remaining part. Otherwise we forward this pad too far into the future
and break synchronization.
We have to queue buffers based on their running time, not based on
the segment position.
Also return running time from GstAggregator::get_next_time() instead of
a segment position, as required by the API.
Also only update the segment position after we pushed a buffer, otherwise
we're going to push down a segment event with the next position already.
https://bugzilla.gnome.org/show_bug.cgi?id=753196
Casting to gpointer from gulong generates the following warning with
64bit Windows target MinGW:
gstplaybin2.c: In function 'pad_added_cb':
gstplaybin2.c:3476:7: error: cast to pointer from integer of different size [-Werror=int-to-pointer-cast]
(gpointer) group_id_probe_handler);
^
cc1: all warnings being treated as errors
We should cast to guintptr from gulong before we cast to gpointer.
https://bugzilla.gnome.org/show_bug.cgi?id=754755
x/y/w/h are signed integers. As can be seen in GstCompositorPad.
The prototype for clamp_rectangle was wrong. This commit reverts the change
and fixes the prototype.
This reverts commit bca444ea4a.
CLAMP checks both if value is '< 0' and '> max'. Value will never be a negative
number since it is an unsigned integer. Removing that check and only checking if
it is bigger than max by using MIN().
CID 1320707
The default one will just go through the internal elements which might
just be identity when it is in passthrough which will lead to the query
being handled by the downstream sink, ignoring all that playsinkconvertbin
could actually handle and convert.
https://bugzilla.gnome.org/show_bug.cgi?id=754235
The obscured check in compositor was using the dimensions of the pad to clamp
the h/w of the pad instead of the output resolution, and was doing an incorrect
calculation to do so. Fix that by simplifying the whole calculation by using
corner coordinates. Also add a test for this bug which fell through the cracks,
and just skip all the obscured tests if the pad's alpha is 0.0.
https://bugzilla.gnome.org/show_bug.cgi?id=754107
It's only relevant for each group, and by storing it in the group
we have locking and everything else like for the other buffering-related
variables. Locking looks a bit fishy still, but it was like that for a long
time already so shouldn't be worse than before.
Overview:
There are some of interleaved streams which has long-term location of audio data.
It mean the audio data is located far away more than multiqueue size.
In this case, because of multiqueue overrun, the pipeline is stopped.
To prevent hanging-like state, the decodebin needs to handle the queue size.
Caused:
The multiqueue size is not enough, the pipeline will stay being stalled status
and decodebin cannot complete to build decode chain.
In this issue file, decodebin did not receive no_more_pads signal or audio data yet.
Steps to Reproduce:
play the high-resolution(4K file) files or some streaming media(push mode).
Actual Results:
There is no audio or subtitle.
We can see only video or infinite loading.
Resolution:
Decodebin detect this problem, and add extra buffer size to multiqueue.
The multiqueue is larger than before, the next data can be pushed the downstream element.
Additional Information:
The max-preroll extra buffer size is set 8MB.
We can use total pre-roll buffer 10MB.
Only first overrun callback can handle multiqueue size.
https://bugzilla.gnome.org/show_bug.cgi?id=733235
When an upstream element wants to flush downstream, we need to take
all chains/groups into consideration.
To that effect, when a FLUSH_START event is seen, after having it
sent downstream we mark all those chains/groups as "drained" (as if
they had seen a EOS event on the endpads).
When a FLUSH_STOP event is received, we check if we need to switch groups.
This is done by checking if there are next groups. If so, we will switch
over to the latest next_group. The actual switch will be done when
that group is blocked.
https://bugzilla.gnome.org/show_bug.cgi?id=606382
When upstream events/queries reach sinkpads of unlinked groups (i.e.
no longer linked to the upstream demuxer), this patch attempts to find
the linked group and forward it upstream of that group.
This is done by adding upstream event/query probes on new group sinkpads
and then:
* Checking if the pad is linked or not (has a peer or not)
* If there is a peer, just let the event/query follow through normally
* If there is no peer, we find a pad to which to proxy it and return
GST_PROBE_HANDLED if it succeeded (allowing the event/query to be properly
returned to the initial called)
Note that this is definitely not thread-safe for the time being
https://bugzilla.gnome.org/show_bug.cgi?id=606382
When we see prefix NALs before a Subset SPS has been spotted,
it might just be because the stream was truncated at the
start, so don't count those as either 'bad' or 'good' packets.
Since the videorate element just duplicates or drops frames
to achieve the desired framerate, it can accept video/x-bayer media
(in any format), which are not present in the current caps.
Just add "video/x-bayer(ANY);" to the caps of the static pad template
(fixing line style to pass the indent commit hook).
https://bugzilla.gnome.org/show_bug.cgi?id=753483
deadend_details need not be returned when the pad is not a deadend.
Hence checking if res value is TRUE and clearing the string instead of
passing it on
https://bugzilla.gnome.org/show_bug.cgi?id=753088
In the case where you have a source giving the GstAggregator smaller
buffers than it uses, when it reaches a timeout, it will consume the
first buffer, then try to read another buffer for the pad. If the
previous element is not fast enough, it may get the next buffer even
though it may be queued just before. To prevent that race, the easiest
solution is to move the queue inside the GstAggregatorPad itself. It
also means that there is no need for strange code cause by increasing
the min latency without increasing the max latency proportionally.
This also means queuing the synchronized events and possibly acting
on them on the src task.
https://bugzilla.gnome.org/show_bug.cgi?id=745768
We need to sync the pad values before taking the aggregator and pad locks
otherwise the element will just deadlock if there's any property changes
scheduled using GstController since that involves taking the aggregator and pad
locks.
Also add a test for this.
https://bugzilla.gnome.org/show_bug.cgi?id=749574
Add logging categories for most video objects.
Remove some useless debug lines in video-info and videotestsrc.
Add a performance debug line in the video scaler.
When switching to a new chain it might be that this new chain
is not yet ready to be exposed so check it before exposing.
Can happen with mpegts that might delay adding pads or pushing data
until it has found the PMT/PAT/PCR and that may take a while depending
on the stream.
It happened frequently with HLS:
http://vevoplaylist-live.hls.adaptive.level3.net/vevo/ch1/appleman.m3u8
If the sink has properties named volume and mute, we have no idea about their
meaning. The streamvolume interface standardizes the meaning.
In the case of osxaudiosink for example, the current volume property has a
range of 0.0 to 1.0, but we need 0.0 to 10.0 or similar. Also osxaudiosink
has no mute property. As such, the volume element should be used here instead.
https://bugzilla.gnome.org/show_bug.cgi?id=752156
Have all sections in alphabetical order. Also make the macro order consistent.
This is a preparation for generating the file. Remove GET_CLASS macro for
some elements, since it is not used and the header is not installed.
We reset the group start time to the running time of the start of the other
streams that are not flushed. This fixes seeking in gapless mode after the
first track has played.
https://bugzilla.gnome.org/show_bug.cgi?id=750013
If suburidecodebin is failed to negotiate (e.g file does not exist)
then free internal suburi variable so that 'current-suburi' property
returns correct status.
https://bugzilla.gnome.org/show_bug.cgi?id=751118
6a2f017bfa changed it to check the subtitle
factory caps if there is a text-sink but we fail to get its sinkpad. What
actually should be done here is to use the factory caps if there is no
text-sink at all.
https://bugzilla.gnome.org/show_bug.cgi?id=750785
There is the GstVideoMultiviewMode enum and the
GstVideoMultiviewFramePacking, which is a subset of the
multiview modes, with the same values as the corresponding
types from the full enum. Do some casts and use the right
times to avoid implicitly using/passing GstVideoMultiviewFramePacking
when a GstVideoMultiviewMode is needed.
Add GstVideoMultiviewFramePacking enum, and the
video-multiview-mode and video-multiview-flags
properties on playbin.
Use a pad probe to replace the multiview information in
video caps sent out from uridecodebin.
This is a part implementation only - for full
correctness, it should also modify caps in caps events,
accept-caps and allocation queries.
https://bugzilla.gnome.org/show_bug.cgi?id=611157
When traversing the color balance element channel list to find the one that
matches with the playsink proxy, the assignation was set to iterator of the
playsink proxy, not the balance element. Thus, the mapping to the values of
the balance element channel was wrong.
This patch fixes the assignation of the color balance element channel, so the
mapping to the channel of the color balance element is fixed.
https://bugzilla.gnome.org/show_bug.cgi?id=750691
when text playbin is not enabled in the beginning, then
video_srcpad_stream_synchronizer gets linked to videochain->sinkpad
and when we try to enable text bin during play, since it is already linked to videochain,
text chain does not get linked properly. Hence unlinking the same
before linking to text chain
https://bugzilla.gnome.org/show_bug.cgi?id=748908
Summary:
So that the user can easily use the same encoding profile to render
with/without audio/video stream.
API:
gst_encoding_profile_is_disabled
gst_encoding_pofile_set_enabled
https://bugzilla.gnome.org/show_bug.cgi?id=749056
When a stream has a variable framerate, videorate calculates it and
forces it on the output caps. However, the code in _transform_caps()
currently also does that if the transform is going in the opposite
direction (GST_PAD_SRC), so during a renegotiation it tries to force
upstream to use the calculated framerate and it fails.
https://bugzilla.gnome.org/show_bug.cgi?id=750032
This part of pipeline is:
tee name=t ! visualizationbin ! streamsynchronizer name=s
t. ! s.
streamsynchronizer might block and it could starve the visualization
branch of the pipeline when it is enabled.
The visualization bin has queues internally but the other branch
that links the audiotee directly to the synchronizer is vulnerable
to block. Adding a queue between "t. ! s." fixes deadlocks.
https://bugzilla.gnome.org/show_bug.cgi?id=749676
From the API documentation: "Note that it is generally not
a good idea to reuse an existing cancellable for more
operations after it has been cancelled once, as this
function might tempt you to do. The recommended practice
is to drop the reference to a cancellable after cancelling
it, and let it die with the outstanding async operations.
You should create a fresh cancellable for further async
operations."
https://bugzilla.gnome.org/show_bug.cgi?id=739132
From the API documentation: "Note that it is generally not
a good idea to reuse an existing cancellable for more
operations after it has been cancelled once, as this
function might tempt you to do. The recommended practice
is to drop the reference to a cancellable after cancelling
it, and let it die with the outstanding async operations.
You should create a fresh cancellable for further async
operations."
https://bugzilla.gnome.org/show_bug.cgi?id=739132
Rather than one of the input pad video info's.
The test checking this was not constraining the output frame size
to ensure that the out of frame stream was not being displayed.
GST_VIDEO_CONVERTER_OPT_ALPHA_MODE, GST_VIDEO_CONVERTER_OPT_CHROMA_MODE,
GST_VIDEO_CONVERTER_OPT_MATRIX_MODE, GST_VIDEO_CONVERTER_OPT_GAMMA_MODE and
GST_VIDEO_CONVERTER_OPT_PRIMARIES_MODE were G_TYPE_STRING with only a few valid
options. Changed those to real enums.
https://bugzilla.gnome.org/show_bug.cgi?id=749104
Upstream might want to use it to properly map timestamps to running/stream
times, if we just override it with 0 synchronization will be just wrong.
For this we remove some old 0.10 code related to segment accumulation, and
remove some more code that is useless now, and accumulate the group start time
(aka segment.base offset) manually now.
https://bugzilla.gnome.org/show_bug.cgi?id=635701
It's a waste of resources to map it if it won't be converted
or used at all. Since we moved the frame mapping down, we need
to use the GST_VIDEO_INFO accessor macros now in the code above
that instead of the GST_VIDEO_FRAME accessor macros.
https://bugzilla.gnome.org/show_bug.cgi?id=746147
For each frame, compare the frame boundaries, check if the format contains an
alpha channel, check opacity, and skip the frame if it's going to be completely
overwritten by a higher zorder frame. The check is O(n^2), but that doesn't
matter here because the number of sinkpads is small.
More can be done to avoid needless drawing, but this covers the majority of
cases. See TODOs. Ideally, a reverse painter's algorithm should be used for
optimal drawing, but memcpy during compositing is small compared to the CPU used
for frame conversion on each pad.
https://bugzilla.gnome.org/show_bug.cgi?id=746147
Embedded systems often have limited charset conversion
functionality, so don't rely on g_convert() (i.e. iconv)
for UTF-16 to UTF-8 conversions, we can easily enough do
that ourselves by converting to native endianness and
then using GLib's helper functions.
This is a fixup for b2db18cda2
audioconvert: avoid float calculations when mixing integer-formatted channels
The int matrix was using gint and gint32 synonymously, which can theoretically
cause problems if gint and gint32 are actually different types.
https://bugzilla.gnome.org/show_bug.cgi?id=747005
Try harder to look for gvfs backend changes in the right
place, to make sure the plugin gets reloaded when backends
are removed or installed. We watch the gvfs mounts directory
because the files there contain absolute paths to the
backend executables, and those may not be in the usual gio
path.
https://bugzilla.gnome.org/show_bug.cgi?id=747841
We don't expect clients to send us any data, but if they do, just
ignore it. Web browsers might send us an HTTP request for example,
but some will still be happy if we just send them data without
a proper HTTP response.
There was a bug in the reading code path. We only have a small
read buffer and would provoke an EWOULDBLOCK trying to read
because we don't bail out of the loop early enough.
https://bugzilla.gnome.org/show_bug.cgi?id=743834
When shutting down the chain, we can get a deadlock when removing
a pad, if that chain was being busy streaming but blocked (eg, while
waiting for a queue to have free space).
https://bugzilla.gnome.org/show_bug.cgi?id=746480
In case upstream does not provide videorate with framerate information,
it will detect the current framerate from the buffer it received,
but if downstream forces the use of variable framerate (most probably
through the use of a caps filter with framerate = 0 / 1), videorate will
respect that.
And add some unit tests
https://bugzilla.gnome.org/show_bug.cgi?id=734424
In the case the framerate is variable (represented by framerate=0/1),
we currently end up loop pushing the first buffer and then recompute
diff1 and diff2 without updating the videorate->next_ts at all
leading to infinitely looping pushing that first buffer.
In the case of variable framerate, we should just compute the next_ts
as previous_pts + previous_duration.
https://bugzilla.gnome.org/show_bug.cgi?id=734424
The patch calculates a second channel mixing matrix from the current one. The
matrix contains the original values * (2^10) as integers. This matrix is used
when integer-formatted channels are mixed.
On a ARM Cortex-A8, single core, 800MHz this improves performance in a
testcase from 29s to 9s for downmixing 6 channels to stereo.
https://bugzilla.gnome.org/show_bug.cgi?id=747005
If a new pad is added after playbin has been put to READY/NULL it
should ignore new pads as it is shutting down.
This can happen when the pipeline fails to preroll (is still in READY)
and the user gives up on waiting or an error that doesn't reach
the demuxer occurs (on some event handling) and it will continue to
work and exposing pads while playbin has been put to NULL.
Without this check an input-selector is created and set to PAUSED
state, preventing playbin from properly shutting down in case it
has data blocked inside it.
audio_convert_convert unpacks to default format (signed) before calling
quantize, and the unsigned variants were equivalent to signed anyway,
so we just get rid of them.
Since range size is always 2^n, we can simply use modulo (implemented
with a bitmask).
The previous implementation used 64-bit integer division, which is
done in software on ARMv7. Although the divisor was constant, the
division could not be transformed into "multiplication by magic number"
since the dividend was 64-bit.
The now-unused and not-so-fast gst_fast_random_(u)int32_range functions
were removed.
Also, implementing bug fixes:
1) ADD_DITHER_TPDF_HF_I no longer discards bias.
2) We change TPDF's noise range to be the same as RPDF's. Previously,
RPDF's noise ranged:
{ bias - dither, bias + dither }
while TPDF's noise ranged:
{ bias/2 - dither/2, bias/2 + dither/2 - 1 } +
{ bias/2 - dither/2, bias/2 + dither/2 - 1 } =
{ bias - dither, bias + dither - 2 }
Now, both range:
{ bias - dither, bias + dither - 1 }
https://bugzilla.gnome.org/show_bug.cgi?id=746661
This fixes a race where the use-buffering property on a multiqueue was
set before the queue depth was changed from it's high preroll limits to
lower playback limits. This resulted in buffering messages being emitted
by the multiqueue in the short window between use-buffering being
set and the queue depth being reset.
https://bugzilla.gnome.org/show_bug.cgi?id=744308
Also:
- Don't modify size on early buffer
The size is the size of the buffer, not of remaining part.
- Use the input caps when manipulating the input buffer
Also store in in the sink pad
- Reply to the position query in bytes too
- Put GAP flag on output if all inputs are GAP data
- Only try to clip buffer if the incoming segment is in time or samples
- Use incoming segment with incoming timestamp
Handle non-time segments and NONE timestamps
- Don't reset the position when pushing out new caps
- Make a number of member variables private
- Correctly handle case where no pad has a buffer
If none of the pads have buffers that can be handled, don't claim to be EOS.
- Ensure proper locking
- Only support time segments
https://bugzilla.gnome.org/show_bug.cgi?id=740236
When the timeout is reached, only ignore pads with no buffers, iterate
over the other pads until all buffers have been read. This is important
in the cases where the input buffers are smaller than the output buffer.
https://bugzilla.gnome.org/show_bug.cgi?id=745768
Correctly calculate alpha in a few places by dividing by 255,
not 256.
Fix the argb and bgra blending functions to avoid an off-by-one
error in the calculations, so painting with alpha = 0xff doesn't
ever bleed through from behind
The variables could have changed when the lock was released
to push a gap event. Streamsynchronizer needs to check them
again before going to sleep.
Bonus: fix a comment typo
multisocketsink now understands the new GstNetControlMessageMeta to allow
sending control messages (ancillary data) with data when writing to Unix
domain sockets.
Thanks to glib's `GSocketControlMessage` abstraction the code introduced
in this commit is entirely portable and doesn't introduce and additional
dependencies or conditionally compiled code, even if it is unlikely to be
of much use on non-UNIX systems.
multisocketsink now understands the new GstNetControlMessageMeta to allow
sending control messages (ancillary data) with data when writing to Unix
domain sockets.
A later commit will introduce a new socketsrc element which will similarly
understand `GstNetControlMessageMeta`. This, when used with a
`GSocketControlMessage` of type `GUnixFDMessage` will allow GStreamer to
send and receive file-descriptions in ancillary data, the first step to
using memfds to implement zero-copy video IPC.
Thanks to glib's `GSocketControlMessage` abstraction the code introduced
in this commit is entirely portable and doesn't introduce and additional
dependencies or conditionally compiled code, even if it is unlikely to be
of much use on non-UNIX systems.
This provides notification that the socket in use was closed by the peer
and gives an opportunity to replace it with a new one which is not
closed, allowing reading from many sockets in order.
I use this in pulsevideo to implement reconnection logic to handle the
pulsevideo service dieing, such that is can be restarted without
disrupting downstream.
Fixes https://bugzilla.gnome.org/show_bug.cgi?id=739546
* Don't bother polling, just do a blocking read, the `GCancellable` will
take care of unlocking. This should also be faster on MS Windows where
the GIO documentation for `g_socket_get_available_bytes` states: "Note
that on Windows, this function is rather inefficient in the UDP case".
* Implement `GstPushSrc.fill` rather than `GstPushSrc.create`. This means
that we will be using the downstream allocator which may be more
efficient. It also means that socketsrc is likely to respect its
"blocksize" property (assuming that there is enough data available).
See https://bugzilla.gnome.org/show_bug.cgi?id=739546
`socketsrc` can be considered a source counterpart to `multisocketsink`.
It can be considered a generalization of `tcpclientsrc` and
`tcpserversrc`: it contains all the logic required to communicate over
the socket but none of the logic for creating the sockets/establishing
the connection in the first place, allowing the user to accomplish this
externally in whatever manner they wish making it applicable to other
types of sockets besides TCP.
This commit essentially copies the implementation directly from
tcpserversrc. Later patches will tidy the implementation up and
re-implement `tcpclientsrc` and `tcpserversrc` in terms of `socketsrc`.
See https://bugzilla.gnome.org/show_bug.cgi?id=739546
If a buffer is made up of non-contiguous `GstMemory`s `gst_buffer_map`
has to copy all the data into a new `GstMemory` which is contiguous. By
mapping all the `GstMemory`s individually and then using scatter-gather
IO we avoid this situation.
This is a preparatory step for adding support to multisocketsink for
sending file descriptors, where a GstBuffer may be made up of several
`GstMemory`s, some of which are backed by a memfd or file, but I think this
patch is valid and useful on its own.
Fixes https://bugzilla.gnome.org/show_bug.cgi?id=746150
Actually accumulate the sample counter to check the accumulated error
between actual timestamps and expected ones instead of just resetting
the error back to 0 with every new buffer.
Also don't reset discont_time whenever we don't resync. The whole point of
discont_time is to remember when we first detected a discont until we actually
act on it a bit later if the discont stayed around for discont_wait time.
https://bugzilla.gnome.org/show_bug.cgi?id=746032
This allows us to handle new segment events correctly; either by dropping
buffers or inserting silence; for example if the offset is changed on an srcpad
connected to audiomixer.
This reverts commit d387cf67df.
The analysis was wrong: The first 20ms of latency are introduced by the source
already and put into the latency query, making it only necessary to cover the
additional 20ms of audiomixer inside audiomixer.
Let's assume a source that outputs outputs 20ms buffers, and audiomixer having
a 20ms output buffer duration. However timestamps don't align perfectly, the
source buffers are offsetted by 5ms.
For our ASCII art picture, each letter is 5ms, each pipe is the start of a
20ms buffer. So what happens is the following:
0 20 40 60
OOOOOOOOOOOOOOOO
| | | |
5 25 45 65
IIIIIIIIIIIIIIII
| | | |
This means that the second output buffer (20 to 40ms) only gets its last 5ms
at time 45ms (the timestamp of the next buffer is the time when the buffer
arrives). But if we only have a latency of 20ms, we would wait until 40ms
to generate the output buffer and miss the last 5ms of the input buffer.
When we modify a GList (via g_list_delete_link), always reassign the
new head to the original GList. Otherwise we end up with
filtered_errors being corrupt (the head might have been the element
removed)
This function is static, and only ever called with the expose lock
taken. It thus has no reason to take this lock itself.
This was introduced by one of my locking fixes from 741355.
https://bugzilla.gnome.org/show_bug.cgi?id=741355
Check if dbin->decode_chain is NULL before running drain_and_switch_chains()
because if it is, we shouldn't run that function or it will segfault.
CID #1271074
Otherwise if there are multiple parsers we would most likely break negotiation
of the stream-format/alignment wanted by the decoders as parsers generally
support all possible stream-formats and alignments.
If caps on a newly added pad are NULL, analyze_new_pad will try to
acquire the chain lock to add a probe to the pad so the chain can
be built later. This comes from the streaming thread, in response
to headers or other buffers causing this pad to be added, so the
stream lock is taken.
Meanwhile, another thread might be destroying the chain from a
downward state change. This will cause the chain to be freed with
the chain lock taken, and some elements are set to NULL here, which
can include the parser. This causes pad deactivation, which tries
to take the element's pad's stream lock, deadlocking.
Fix this by keeping track of which elements need setting to NULL,
and only do this after the chain lock is released. Only the chain
manipulation needs to be locked, not the elements' state changes.
https://bugzilla.gnome.org/show_bug.cgi?id=741355
There was a deadlock between a thread changing decodebin/demuxer
state from PAUSED to READY, and another thread pushing data
when starting.
From the stack trace at
https://bug741355.bugzilla-attachments.gnome.org/attachment.cgi?id=292471,
I deduce the following is happening, though I did not reproduce the
problem so I'm not sure this patch fixes it.
The streaming thread (thread 2 in that stack trace) takes the demuxer's
sink pad's stream lock in gst_ogg_demux_perform_seek_pull and will
activate a new chain. This ends up causing the expose lock being taken
in _pad_added_cb in decodebin.
Meanwhile, a state changed is triggered on thread 1, which takes the
expose lock in decodebin in gst_decode_bin_change_state, then frees
the previous chain, which ends up calling gst_pad_stop_task on the
demuxer's task, which in turn takes the demuxer's sink pad's stream
lock, deadlocking as both threads are now waiting for each other.
https://bugzilla.gnome.org/show_bug.cgi?id=741355
Also improve the waiting condition for stream switches, which was assuming
before that the condition variable will only stop waiting once when it is
signaled. But the documentation says that there might be spurious wakeups.
https://bugzilla.gnome.org/show_bug.cgi?id=736655
Change the GAP events that are currently sent from the chain function of
the current pad to all other EOS pads. They should instead be sent from
their own streaming threads.
https://bugzilla.gnome.org/show_bug.cgi?id=736655
Wait in the event function when EOS is received until all pads are EOS
and then forward the EOS event from each pads own event function.
Also send a new GAP event for EOS pads from the event function whenever
going from PLAYING->PAUSED by shortly waking up the GCond. This is needed
to allow sinks to pre-roll again, as they did not receive EOS yet because
we blocked that, but also will never get data again.
https://bugzilla.gnome.org/show_bug.cgi?id=736655
There's no reason why audiomixer should override the segment
base of upstream with whatever value it got from a SEEK event,
or even worse... with 0 if there was no SEEK event yet. This
broke synchronization if upstream provided a segment base other
than 0, e.g. when using pad offsets.
Also that this code did things conditional on the element's state
should've been a big warning already that something is just wrong.
If this breaks anything else now, let's fix it properly :)
Also don't do fancy segment position trickery when receiving a
segment event. It's just not correct.
In gst_video_scale_fixate_caps () it can goto done without freeing the memory
of the tmp GstStructure. This makes it go out of scope and leak.
CID #1265766
Instead of using the GST_OBJECT_LOCK we should have
a dedicated mutex for the pad as it is also associated
with the mutex on the EVENT_MUTEX on which we wait
in the _chain function of the pad.
The GstAggregatorPad.segment is still protected with the
GST_OBJECT_LOCK.
Remove the gst_aggregator_pad_peak_unlocked method as it does not make
sense anymore with a private lock.
https://bugzilla.gnome.org/show_bug.cgi?id=742684
Reduce the number of locks simplify code, what is protects
is exposed, but the lock was not.
Also means adding an _unlocked version of gst_aggregator_pad_steal_buffer().
https://bugzilla.gnome.org/show_bug.cgi?id=742684
Ignore chroma subsampling and color matrix transformations like the
old videoscale used to do. This is to make the performance like it was
before.
See https://bugzilla.gnome.org/show_bug.cgi?id=741987
Speex may decide not to consume any samples because it can't write any. I've
seen a hang during draining caused by the resample loop never terminating.
In that case, resampling happened as normal until olen was 0 but ilen was
still 1. _process_native then reduced ichunk to 0, so ilen never decreased
below 1 and the loop never terminated.
Instead of reverting 684cf44 ({audioresample: don't skip input samples),
break only if all output samples have been produced and speex refuses
to consume any more input samples.
https://bugzilla.gnome.org/show_bug.cgi?id=732908
VideRate keeps 1 buffer in order to duplicate base on closest buffer
relative to targeted time. This extra buffer need to be request
otherwise the pipeline may stall when fixed size buffer pool is used.
https://bugzilla.gnome.org/show_bug.cgi?id=738302
Consider pipeline: gst-launch-1.0 playbin uri=http://example.com/a.ogg
Consider 128kbit audio stream.
As soon as uridecodebin detects the bitrate, it configures its input
queue2 max-size to 32000 bytes.
The 2MB buffer in multiqueue is nearly 2 orders of magnitude bigger.
This non-deterministically drives queue2 buffer anywhere from
100% to 0% until multiqueue is filled.
This patch sets multiqueue size to 5 buffers early in no_more_pads_cb.
Partly reverts commit db771185ed.
https://bugzilla.gnome.org/show_bug.cgi?id=740689
Decodebin has already added the element to the bin and should only
select caps compatible pads. It should disable the pad link checks
to avoid doing those again.
https://bugzilla.gnome.org/show_bug.cgi?id=742885
This can happen if this is a live pipeline and no source produced any buffer
and sent no caps until the an output buffer should've been produced according
to the latency.
When this is TRUE, we really have to produce output. This happens
in live mixing mode when we have to output something for the current
time, no matter if we have enough input or not.
When this is TRUE, we really have to produce output. This happens
in live mixing mode when we have to output something for the current
time, no matter if we have enough input or not.
Create a function to do the pad cleanup of the GstSourceCombine struct
and use it to not forget to also cleanup the sink pad and fix a memory
leak.
https://bugzilla.gnome.org/show_bug.cgi?id=741198
In some cases, the user might want the stream outputted by encodebin to
be in the exact same format during all the stream. We should let the
user specify when this is the case. This commit add some API in the
GstEncodingProfile to determine whether the format can be renegotiated
after the encoding started or not.
API:
gst_encoding_profile_set_allow_dynamic_output
gst_encoding_profile_get_allow_dynamic_output
https://bugzilla.gnome.org/show_bug.cgi?id=740214
Before we were setting them to PAUSED and (much) later connecting to
their source pad caps notify signal.
There was a race where that demuxer was pushing a caps and later a buffer
on its source pad when we were not even connected to its source pad caps notify
signal leading to decodebin missing the information and not keeping on
building the pipeline on CAPS event thus the demuxer was posting an ERROR
(not linked) message on the bus. This need to be done for 'simple
demuxers' because those have one ALWAYS source pad, not like usual demuxers
that have several dynamic source pads.
A "simple demuxer" is a demuxer that has one and only one ALWAYS source
pad.
https://bugzilla.gnome.org/show_bug.cgi?id=740693
There was a race where:
1) we would put the element to PAUSED
2) It would get data sent to it from upstream
3) It would thus send caps
3) caps_notify_cb would continue autoplugging
4) caps would flow downstream, the last pad would get exposed
5) we were still not done sending the sticky events
Taking the stream lock on the new element's sinkpad and only
releasing it when sticky events have all been sent prevents
the caps from reaching the source pad of the element before
we're all set.
https://bugzilla.gnome.org/show_bug.cgi?id=740694
