This is a fix for a data race leading to:
> GLib-CRITICAL: g_hash_table_foreach:
> assertion 'version == hash_table->version' failed
Identified sequence:
* `rtp_session_on_timeout` acquires the lock on `session` and proceeds with its
processing.
* `rtp_session_process_rtcp` is called (debug log : received RTCP packet) and
attempts to acquire the lock on `session`, which is still held by
`rtp_session_on_timeout`.
* as part of an hash table iterator, `rtp_session_on_timeout` transitively
invokes `source_caps` which releases the lock on `session` so as to call
`session->callbacks.caps`.
* Since `rtp_session_process_rtcp` was waiting for the lock to be released, it
succeeds in acquiring it and proceeds with `rtp_session_process_rr` which
transitively calls `g_hash_table_insert` via `add_source`.
* After `source_caps` re-acquires the lock and gives the control flow back to
`rtp_session_on_timeout`, the hash table iterator is changed, resulting in the
assertion failure.
This commits copies `sess->ssrcs[sess->mask_idx]` and iterates on the copy so
the iterator is not affected by a concurrent change due to the lock being
released in the `source_caps` callback.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4585>
On very quick start/stop, the mainloop may never be run. As a side
effect, our idle stop function is not really being ran, so we can't rely
on that to free the main loop. Simply unref the mainloop when the
thread have completely stop.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4539>
If we don't do that, clients can rely on this signal to see the final pad
topology but it won't be the real one as some of them will disappear after
emitting that signal. This can happen after injecting a different init segment.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4557>
Since c2f890ab, element properties are gathered from the parse-launch
line and passed at object construction.
This caused the following issue to happen in videoflip:
* videoflip installed a CONSTRUCT property named method, now deprecated
* videoflip now also overrides that property with a video-direction
property
GObject construction causes method to be set first at construct time,
with the user-provided value, then video-direction with the default
value.
The user-provided value was thus overridden, causing a regression.
Fix by not installing the properties as CONSTRUCT, and explicitly
implementing constructed() instead in order to ensure that we do still
call gst_video_flip_set_method() at least once during construction.
Fixes https://gitlab.freedesktop.org/gstreamer/gstreamer/-/issues/2529
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4551>
A blocking pad probe is added on new sink pads, it's usually removed after the
caps have been negotiated or the signaling state switched to stable, but if that
never happens and the pad is released we kept the pad probe active, leaving the
pad blocked, preventing clean disposal.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4533>
While testing the [implementation for insertable streams] in `webrtcsink` &
`webrtcsrc`, I encountered critical warnings, which turned out to result from
two race conditions in `rtpsession`. Both race conditions produce:
> GLib-CRITICAL: g_hash_table_foreach:
> assertion 'version == hash_table->version' failed
This commit fixes one of the race conditions observed.
In its simplest form, the test consists in 2 pipelines and a Signalling server:
* pipelines_sink: audiotestsrc ! webrtcsink
* pipelines_src: webrtcsrc ! appsrc
1. Set `pipelines_sink` to `Playing`.
2. The Signalling server delivers the `producer_id`.
3. Initialize `pipelines_src` to establish a session with `producer_id`.
4. Set `pipelines_src` to `Playing`.
5. Wait for a buffer to be received by the `appsrc`.
6. Set `pipelines_src` to `Null`.
7. Set `pipelines_sink` to `Null`.
The race condition happens in the following sequence:
* `webrtcsink` runs a task to periodically retrieve statistics from `webrtcbin`.
This transitively ends up executing `rtp_session_create_stats`.
* `pipelines_sink` is set to `Null`.
* In `Paused` to `Ready`, `gst_rtp_session_change_state()` calls
`rtp_session_reset()`.
* The assertion failure occurs when `rtp_session_reset` is called while
`rtp_session_create_stats` is executing.
This is because `rtp_session_create_stats` acquires the lock on `session` prior
to calling `g_hash_table_foreach`, but `rtp_session_reset` doesn't acquire the
lock before calling `g_hash_table_remove_all`.
Acquiring the lock in `rtp_session_reset` fixes the issue.
[implementing insertable streams support]: https://gitlab.freedesktop.org/gstreamer/gst-plugins-rs/-/merge_requests/1176
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4532>
Unfortunately streamoff does not flush the events, and this can cause all
sort of issues. Flush events on capture queue. We also return
GST_V4L2_FLOW_RESOLUTION_CHANGE in case a resolution change was seen.
This allow skipping streamon(capture) on flush, which could lead to a
configuration miss-match, or failure if the buffers aren't of the right
size.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4525>
Let the driver detects the change and reconfigure the capture side
transparently from there. This avoid reallocation of the output buffers,
and eliminates the need to stop and restart the capture task. This is
only happening if the driver have support for this, otherwise the old
behaviour is maintained.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4525>
Stop doing capture buffer allocation based on guesses
and wait for the source change event when available.
Unlike stateless decoder, the stateful decoder is not aware of
the coded resolution, and this may lead to the wrong result
even when using TRY_FMT.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4525>
In previous implementation that job was split between handle_frame and
the processing loop and it wasn't clear if this mechanism was race
free. The capture setup would also be tried for every buffer, which was
not necessary.
This also simplify the handling of SRC_CH event, dropping the unneeded
atomic boolean.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4525>
This reverts commit f29c19be58. If this is
called for the reference context then we would run into an infinite
loop, which is not really better than an assertion.
By fixing up DTS to never be ahead of the PTS in the previous commit
this situation should be impossible to hit now.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4515>
decodebin3 will do its best to figure out whether a parsebin is required to
process the incoming stream.
The problem is that for push-based stream it could happen that the stream would
not provide any caps, resulting in nothing being linked internally.
Furthermore, there is the possibility that a stream *with* caps would not be
using a TIME segment, which is required for multiqueue to properly work.
In order to fix those two issues, we force the usage of parsebin on push-based
streams:
* When the pad is linked, if upstream can't provide any caps
* When we get a non-TIME segment
Fixes#2521
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4499>
On MacOS with homebrew, the openssl library is not
properly detected with pkg-config.
So disable the test compilation if openssl
is not properly detected along with libcrypto.
libcrypto is detected but it uses the system one
which leads to the error:
your binary is not an allowed client of /usr/lib/libcrypto.dylib for
architecture x86_64
See more details from Apple:
https://developer.apple.com/forums/thread/124782
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4483>
The addresses we get from `resolve_host_finish()` (via
`resolve_host_async()`, `resolve_host_main_cb()`, `on_resolve_host()`,
`g_resolver_lookup_by_name_finish()`) must be freed. Otherwise we leak
memory.
Leak found and confirmed fixed with GCC AddressSanitizer.
Change-Id: If32d24452d626234f01b253b77a7d6d16eac1cee
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4482>
The decoder needs to force another enumeration of the format. For
this it was clearing the v4l2object insternal list, leaving a fmtdesc
pointer pointing to freed memory. This patch clears the fmtdesc pointer
that has just been free. It also makes sure the probe function does not
use the cached formats list. The probe function will restore the current
fmtdesc pointer based on the currently configured pixelformat.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4426>
As we don't have anything smart in the fixation process, we may endup with
a format that has a lower bitdepth, even if downstream can handle higher
depth. it is notably the case when negotiating with deinterlace, which places
is non-passthrough caps before its passthrough one. This makes the generic
fixation prefer the formats natively supported by deinterlace element over
the HW 10bit format. As some HW can downscale 10bit to 8bit, this can break
10bit decoding.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4426>
The original code was:
if (!gst_structure_get (s, "srtp-key", GST_TYPE_BUFFER, &buf, NULL) || !buf) {
goto error;
} else {
stream->key = buf;
}
So use "srtp-key" if it is set so a non NULL buffer. The condition was
incorrectly inverted in ad7ffe64a6 to:
if (gst_structure_get (s, "srtp-key", GST_TYPE_BUFFER, &buf, NULL) || !buf) {
stream->key = buf;
} ...
Fix the condition so it works as originally intended and avoid accessing
'buf' uninitialised.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4423>
One race condition is the fact that the window object
can be destroyed while running some routine in the UI
thread (such as resizing). To avoid that situation we make
UI thread hold a reference on the window object while it's
running.
Other probpematic case is when the window handle is reused:
if we stop and start the pipeline very fast,
so the sink creates a new window object that is going to use
the same window handle as the previous one.
And finally the case when the pipeline is stopped immediatelly
right after starting, this one is also handled in this commit.
NOTE: a unit test that reproduces this cases have been added
in the previous commit.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4417>
g_string_free(.., FALSE) gives us ownership of the string
already, no need to duplicate that again with g_strdup(),
and doing so will leak the string returned by g_string_free()
here. Caught by compiler warnings in newer GLib versions.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4380>
The previous code would only check if two packets in a row were duplicates. If
not (i.e. a packet is a duplicate of a packet received slightly before) the code
would generate completely bogus FCI because it assumes there were no duplicates
present in the array.
In order to be efficient, just store all received packets and remove the
duplicates just before the FCI is generated once the array of observations have
been sorted by seqnum.
Fixes TWCC usage with moderate to high packet duplication.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4378>
The proxy and queue are created in the gst_gl_window_wayland_egl_open()
function and will be recreated on open. This leaks both objects, the
wayland client documentation mentions that they should be destroyed
using the appropriate destroy functions.
Found during valgrind memory leak testing, these blocks were marked as
definitely lost.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4355>
Existing codes rely on modified argc value by g_option_context_parse()
but g_option_context_parse_strv() is used in case of Windows.
Count arguments after the option parsing manually.
Fixing command "gst-inspect-1.0.exe -b"
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4322>
Moving from PLAYING to NULL will set the stop_streaming_threads to TRUE,
but when moving back upwards its not reset to FALSE (as only done in
uncalled init and resume callbacks).
Fix by reseting value in the prepare callback.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4321>
Running element_vkcolorconver test with Vulkan validation layer this error is
raised:
Code 0 : Validation Error: [ VUID-VkMappedMemoryRange-size-01390 ] Object 0:
handle = 0x100000000010, type = VK_OBJECT_TYPE_DEVICE_MEMORY;
| MessageID = 0xdd4e6d8b
| vkFlushMappedMemoryRanges: Size in pMemRanges[0] is 0x4, which is not a
multiple of VkPhysicalDeviceLimits::nonCoherentAtomSize (0x40) and offset +
size (0x0 + 0x4 = 0x4) not equal to the memory size (0xb). The Vulkan spec
states: If size is not equal to VK_WHOLE_SIZE, size must either be a multiple of
VkPhysicalDeviceLimits::nonCoherentAtomSize, or offset plus size must equal the
size of memory
The reason of is that the image size used in the test doesn't comply hardware
restrictions. In order to avoid juggling with image size and hardware
restrictions, this patch proposes to use VK_WHOLE_SIZE macro.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4311>
Running tests with Vulkan Validation enabled show an error on vkimage tests:
Code 0 : Validation Error: [ VUID-VkImageViewCreateInfo-image-04441 ]
Object 0: VK_NULL_HANDLE, type = VK_OBJECT_TYPE_COMMAND_BUFFER; Object 1: handle
= 0x50000000005, type = VK_OBJECT_TYPE_IMAGE;
| MessageID = 0xb75da543
| Invalid usage flag for VkImage 0x50000000005[] used by vkCreateImageView(). In
this case, VkImage should have VK_IMAGE_USAGE_SAMPLED_BIT |
VK_IMAGE_USAGE_STORAGE_BIT | VK_IMAGE_USAGE_COLOR_ATTACHMENT_BIT |
VK_IMAGE_USAGE_DEPTH_STENCIL_ATTACHMENT_BIT |
VK_IMAGE_USAGE_TRANSIENT_ATTACHMENT_BIT | VK_IMAGE_USAGE_INPUT_ATTACHMENT_BIT |
VK_IMAGE_USAGE_FRAGMENT_SHADING_RATE_ATTACHMENT_BIT_KHR |
VK_IMAGE_USAGE_FRAGMENT_DENSITY_MAP_BIT_EXT |
VK_IMAGE_USAGE_VIDEO_DECODE_DST_BIT_KHR |
VK_IMAGE_USAGE_VIDEO_DECODE_DPB_BIT_KHR |
VK_IMAGE_USAGE_VIDEO_ENCODE_SRC_BIT_KHR |
VK_IMAGE_USAGE_VIDEO_ENCODE_DPB_BIT_KHR | VK_IMAGE_USAGE_SAMPLE_WEIGHT_BIT_QCOM
| VK_IMAGE_USAGE_SAMPLE_BLOCK_MATCH_BIT_QCOM set during creation.
The Vulkan spec states: image must have been created with a usage value
containing at least one of the usages defined in the valid image usage list for
image views
This patch adds VK_IMAGE_USAGE_SAMPLED_BIT to the usage bits in test.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4311>
While using the validation layer with this pipeline:
gst-launch-1.0 videotestsrc num-buffers=10 ! vulkanupload ! vulkancolorconvert ! vulkansink
The validation layer throws this message:
Code 0 : Validation Error: [ VUID-VkAttachmentDescription-format-06699 ]
Object 0: handle = 0x5555562e9610, type = VK_OBJECT_TYPE_DEVICE; | MessageID = 0x52b3229e |
vkCreateRenderPass: pCreateInfo->pAttachments[0] format is
VK_FORMAT_B8G8R8A8_UNORM and loadOp is VK_ATTACHMENT_LOAD_OP_LOAD, but
initialLayout is VK_IMAGE_LAYOUT_UNDEFINED.
The Vulkan spec states: If format includes a color or depth aspect and loadOp is
VK_ATTACHMENT_LOAD_OP_LOAD, then initialLayout must not be VK_IMAGE_LAYOUT_UNDEFINED
When creating the render pass the loadOp can be either
`VK_ATTACHMENT_LOAD_OP_CLEAR` or `VK_ATTACHMENT_LOAD_OP_LOAD` depending on
`enable_clear`. While `enable_clear` is FALSE by default (which means
`VK_ATTACHMENT_LOAD_OP_LOAD`). Nonetheless, its value is explicitly changed by
`vkoverlaycompositor` to FALSE too!
This behavior was introduced in merge request #2470 where
`VK_ATTACHMENT_LOAD_OP_CLEAR` was a fixed value for loadOp. Thus, the bug
consists in a missing initialization of `enable_clear` to TRUE from that merge
request.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4311>
With GST_SEEK_FLAG_SNAP_AFTER present, the previous version would
adjust seek time based on the keyframe farthest away from desired_time.
This was incorrect, because we always want the *earliest* suitable keyframe
to seek to, not the last one.
With this fix, in case of the SNAP_AFTER, we now look for the closest keyframe
that can be found after desired_time. Behaviour for SNAP_BEFORE should remain
unchanged.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4251>
Assuming that V4L2 CAPTURE devices always use one buffer per JPEG image, we can
always mark JPEGs provided by a V4L2 element as parsed.
The V4L2 elements require that JPEG images sent to V4L2 OUTPUT devices must
always be parsed.
This is necessary to link a V4L2 CAPTURE device with a V4L2 OUTPUT device
without explicitly marking the stream as parsed or adding a jpegparse into the
pipeline.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4247>
The goal of parsebin is to figure out which elements to link together in order
to provide elementary streams given any random input.
The problem is that deciding whether a given stream should still have more
elements plugged in or not was dependent on ... the presence of compatible
decoders (sic).
Instead of that, if we can't plug anymore elements on a given stream *and* it is
detected as being an elementary stream, expose it.
Fixes#2118
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4237>
If sticky events are present on parsebin source pads, we propagate them to the
multiqueue source pads. Those will be propagated on the new urisourcebin source
pads like in the other code paths.
This ensures that STREAM_START event are present on new source pads. If CAPS
event are also present (not guaranteed), they will also be available.
Fixes#2384
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4218>
Previously, reassigning loop index l in nicestream.c
could cause a segfault if l->data was null, as it could
reassign l to a null variable, triggering the loop
postassignment l->next, which then segfaults due to
l now being null. It is instead moved into the loop.
_delete_transport already performs the reassignment
inline.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4213>
In webrtc_data_channel_send functions, both data and string,
an early return on a non-open datachannel caused it to leak
the buffer used for pushing to appsrc, meaning any buffer
sent after leaving the open state was leaked in full.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4206>
This is already done for every other calls to send_packet. The deadlock occures
since FFMPeg 6.0. The decoder tries to get a buffer from a thread during
the draining process, and blocks trying to get the video decoder stream lock
already heald by the drain function.
Fixes#2383
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4187>
Removing a meta from a buffer means one doesn't have access to it
anymore. Instead use the already reffed composition directly.
Fixes a use-after-free in the following pipeline:
... ! vulkanupload ! timeoverlay ! vulkanoverlaycompositor ! ...
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4147>
There may be garbage or some bits before a SOI comes in some problematic
mjpeg streams. For example, some network error may cause the EOI marker
of the previous frame lost, and when the new frame's SOI comes, we still
use the state of the last frame, which will generate errors.
For this kind of frames without EOI, if that frame already has some data
(the SOS segment is detected), we still push it as a frame with CORRUPTED
flag set. But if not, we just discard all the data before the new SOI.
Co-Authored-By: Víctor Jáquez <vjaquez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4130>
In tests in the rust bindings we end up with 2 thread initializing
concurrently, and it should not be a problem, -validate should be MT
safe.
Using a recursive mutex as we might recursively init for some reason
and we are not on the hot path here in any case.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4097>
Recursively invoking the NSMainLoop can cause crashes in
applications that don't expect it. Instead of waiting for
permission to be granted, move the wait later - until we
actually need device permissions when starting the capture
session. That moves the wait into the streaming thread
instead of the application thread that's setting the pipeline
state to READY.
Instead of a manual state change implementation to open
and close the device, use the basesrc start/stop methods that
are intended for the purpose.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4096>
These checks were introduced to prevent exposing ARGB64/RGBA64 in the caps
when running on M1 Pro/Max with macOS <13 because of a bug in VideoToolbox.
Unfortunately, the initial buffer size of 15 is too short when running
in a VM - the CPU brand string there looks like "Apple M1 Pro (Virtual)",
which due to its length causes sysctlbyname to return -1, resulting in
broken formats still showing up in the caps.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4082>
Fixes the following valgrind error:
==616== Conditional jump or move depends on uninitialised value(s)
==616== at 0x4900E34: gst_debug_print_object (gstinfo.c:1143)
==616== by 0x49010B6: gst_info_printf_pointer_extension_func (gstinfo.c:1215)
==616== by 0x4959FDB: __gst_printf_pointer_extension_serialize (printf-extension.c:47)
==616== by 0x495A487: printf_postprocess_args (vasnprintf.c:258)
==616== by 0x495A52C: __gst_vasnprintf (vasnprintf.c:290)
==616== by 0x4959F8F: __gst_vasprintf (printf.c:154)
==616== by 0x4901C1F: gst_debug_message_get (gstinfo.c:791)
==616== by 0x4901C75: _gst_debug_log_preamble (gstinfo.c:1431)
==616== by 0x4903208: gst_debug_log_default (gstinfo.c:1575)
==616== by 0x49020BA: gst_debug_log_full_valist (gstinfo.c:624)
==616== by 0x490211D: gst_debug_log_valist (gstinfo.c:656)
==616== by 0x49021AD: gst_debug_log (gstinfo.c:533)
==616== by 0x48DDC11: gst_buffer_copy_into (gstbuffer.c:693)
==616== by 0x48DF5F1: gst_buffer_copy_with_flags (gstbuffer.c:727)
==616== by 0x48DF640: gst_buffer_copy_deep (gstbuffer.c:756)
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4038>
The av1decoder class does not implement the ->parse() virtual function,
and we always need to add the av1parse element before it. So we should
set_needs_format of the decoder to TRUE, then if no parse before it, it
can fail with a "not-negotiated" error early, rather than go on and
generate unexpected error.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4066>
The vp9decoder class does not implement the ->parse() virtual function,
and we always need to add the vp9parse element before it. So we should
set_needs_format of the decoder to TRUE, then if no parse before it, it
can fail with a "not-negotiated" error early, rather than go on and
generate unexpected error.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4066>
The vp8decoder class does not implement the ->parse() virtual function,
it can only accepts frame aligned data. If some element such as filesrc
feed it with unaligned data, the behaviour is undecided. So we should
set_needs_format of the decoder to TRUE, then it can fail with a
"not-negotiated" error early, rather than go on and generate unexpected
error.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4066>
The mpeg2decoder class does not implement the ->parse() virtual function,
and we always need to add the mpegvideoparse element before it. So we should
set_needs_format of the decoder to TRUE, then if no parse before it, it
can fail with a "not-negotiated" error early, rather than go on and
generate unexpected error.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4066>
The h264decoder class does not implement the ->parse() virtual function,
and we always need to add the h264parse element before it. So we should
set_needs_format of the decoder to TRUE, then if no parse before it, it
can fail with a "not-negotiated" error early, rather than go on and
generate unexpected error.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4066>
The h265decoder class does not implement the ->parse() virtual function,
and we always need to add the h265parse element before it. So we should
set_needs_format of the decoder to TRUE, then if no parse before it, it
can fail with a "not-negotiated" error early, rather than go on and
generate unexpected error.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4066>
These parameters are not actually `out` parameters but must
be allocated and zero-initialized by the calling function.
Marking them as `out caller-allocates` will cause memory
corruptions when calling these APIs from e.g., Python code.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4061>
The dimension of the overlay texture directly corresponds to the size of the overlay **buffer** which is given by its video meta.
The dimension at which the overlay should be displayed directly correspond to the overlay `render_width`and `render_height`.
This match the behavior of glimagesink
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4053>
The av1C box is optional so dropping parsing does not break anything
fundamentally, and there seems to be no historical record how version 0
even looks like while the comments and the parsing disagreed with each
other.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4027>
If we have caps then we can only set exactly those caps, if we have no
caps yet then negotiating anything is not very meaningful because the
caps are defined by the application and not downstream.
Avoids, among other things, an unnecessary allocation query and spurious
useless caps being set before the first buffer.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4020>
This crept in several years ago sadly :(
The usage of accurate seeking should be reserved to use-cases where it is
essential that we seek to that position. This should not be the default.
There is a new option `--acurate-seeks/-a` to be able to force that.
Furthermore, if accurate seeks aren't required, a player should be using the
GST_SEEK_FLAG_KEY_UNIT flag to seek to the closest keyframe and provide the most
reactive experience.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4017>
Generating the source element is done when uridecodebin is doing the
READY to PAUSED state change, so it is reasonable to set the new source
element to that state.
This also allows detecting early failures with backing libraries or
hardware (checks done in NULL->READY).
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4016>
When using qtdemux in a pipeline that should only work as a pure demuxer (not
for actual playback), qtdemux shouldn't emit new GstSegments to correct
the start time (jump to the future) to ensure that the user experiences no
playback delay. By doing so, it's generating the wrong segments when an append
of data from the past happens. When that happens, downstream elements such as
parsers (eg: aacparse) may clip those buffers laying before the GstSegment and
create problems on the GStreamer client app (eg: WebKit).
Getting buffers clipped out because of the wrong GstSegments started becoming
a problen when this commit was introduced:
ab6e49e9cc audioparsers: add back segment clipping to parsers that have lost it
This clipping makes test DASH shaka 35 from MVT tests[1] to fail in
WebKitGTK/WPE (at least) and can potentially cause a number of other problems
in the WebKit Media Source Extensions (MSE) code.
Note that this new behaviour of not emitting new GstSegments only makes sense
when qtdemux is being used as a pure demuxer and not as part of a regular
pipeline. This is why the variant field has been added. When equal to
VARIANT_MSE_BYTESTREAM, it will make qtdemux behave differently in push mode,
taking decisions that meet the expectations for an MSE-like processing mode.
This kind of tweaks have been done in the past for MSS streams, for instance.
That code has been refactored to use VARIANT_MSS_FRAGMENTED now, instead of
its own dedicated boolean flag.
Co-authored by: Alicia Boya García <ntrrgc@gmail.com>
...who suggested to use "variant: mse-bytestream" in the caps to identify that
mode, as proposed in her unmerged patch:
https://gitlab.freedesktop.org/gstreamer/gst-plugins-good/-/issues/467
[1] https://github.com/rdkcentral/mvt
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3990>
This fixes a compile error with recent upstream FFmpeg.
The AV_CODEC_CAP_AUTO_THREADS was deprecated and renamed to
AV_CODEC_CAP_OTHER_THREADS in FFmpeg upstream commit
7d09579190de (lavc 58.132.100).
The AV_CODEC_CAP_AUTO_THREADS was finally removed in FFmpeg upstream
commit 10c9a0874cb3 (lavc 59.63.100).
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3964>
Fixes#1358.
Passing ARGB64/RGBA64 to vtenc caused the encoding to fail
when running on M1 Pro/Max variants with macOS 12.x, so let's
remove these formats from caps when such scenario is detected.
This issue appears to have been fixed OS-side in macOS 13.0.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3962>
This was causing incorrect output when seeking, especially
when used with a multithreaded source like `videotestsrc n-threads=2`.
It should now correctly wait for frames still being processed by VT
while vtdec is flushing.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3937>
We are using std::isspace() with one parameter. That function is defined
in the cctype header.
```
win32ipcutils.cpp(34): error C2672: 'std::isspace': no matching overloaded function found
win32ipcutils.cpp(34): error C2780: 'bool std::isspace(_Elem,const std::locale &)': expects 2 arguments - 1 provided
```
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3936>
When the task already exists, we forgot to free the passed `user_data`.
This wasn't an issue for most C code, which doesn't pass a
`GDestroyNotify`, but bindings such as gstreamer-rs do!
That said, allocating a trampoline in gstreamer-rs just for it to get
thrown away again is awkward. Maybe we need a `gst_pad_resume_task`?
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3925>
The VA API has not defined the scaling list entries for U/V planes
for the 4:4:4 stream. In fact, we do not meet the 4:4:4 format output
for H264 so far, and scaling list is not used frequently, so we just
print out some warning and ignore these scaling list values.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3877>
The gst-devtools project generates gstreamer-validate-1.0.pc, this
must match the dependency in gst-editing-services for detection
to work properly.
Fixes:
Run-time dependency gst-validate-1.0 found: NO (tried pkgconfig and cmake)
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3872>
Due to a bug in the VT API, attempting to encode interlaced content
with ProRes results in an error, halting the pipeline instead of
gracefully falling back to software encoding.
Should be removed in the future if Apple ever fixes this issue.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3878>
All the RTP src pads were sharing the same stream-id while each actually
carry a different stream.
This was causing problem for example when funneling the streams together
and then trying to split them using 'streamiddemux'.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3866>
It's not possible to annotate a in-parameter for a return value array as
the array length. Both are assumed to have the same direction and the
current annotation causes the size parameter to be considered an out
parameter.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3814>
Spec 7.1.3:
If a memory object does not have the VK_MEMORY_PROPERTY_HOST_COHERENT_BIT
property, then vkFlushMappedMemoryRanges must be called in order to guarantee
that writes to the memory object from the host are made available to the host
domain, where they can be further made available to the device domain via a
domain operation. Similarly, vkInvalidateMappedMemoryRanges must be called to
guarantee that writes which are available to the host domain are made visible to
host operations.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3817>
It is really difficult for people to figure out why nvcodec has
0 features. Even the debug log is cryptic. Also make sure the errors
go to the ERROR log level, which is more likely to be enabled by
default.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3813>
Due to the dynamic nature of multiqueue, when `use-interleave` is used we can't
report a maximum tolerated latency (when queried) since it is calculated
dynamically.
When in such live pipelines, we need to make sure multiqueue can handle the
lowest global latency (provided by this event). Failure to do that would
result in not providing enough buffering for a realtime pipeline.
Fixes#1732
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3810>
Deserialize socket control messages as GstSocketTimestampMessage only
if (level, type) is (SOL_SOCKET, SCM_TIMESTAMPNS).
Without this patch, messages with types SCM_RIGHTS or SCM_CREDENTIALS
could be deserialized as GstSocketTimestampMessage instead of
GUnixFDMessage or GUnixCredentialsMessage from gio.
Fixes#1736
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3777>
Handling mouse navigation events in glvideomixer element, if no
pixel-aspect-ratio info in the caps, an assertion error is produced
inside gst_util_fraction_multiply because default denominator is zero.
Error fixed:
```
(gst-launch-1.0:102654): GStreamer-CRITICAL **: 00:47:51.598: gst_util_fraction_multiply: assertion 'b_d != 0' failed
```
Simple pipeline to reproduce the issue:
```
gst-launch-1.0 -v glvideomixer name=mix ! glimagesinkelement gltestsrc ! mix.sink_0
```
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3766>
Allow requesting an offer from the peer if we're joining a call with a
peer, and allow the peer to request an offer from us if waiting for an
incoming call.
This implements all 4 variants the protocol allows for.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3758>
AVC-Intra is a range of H.264-compliant intra-only codecs from
Panasonic. The codes and descriptions have been taken from VLC.
The (encumbered) sample I have here produces byte-stream H.264,
including SPS and PPS and no `avcC` box.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3739>
We create a new context in `gst_gl_context_create_thread()` and then
activate it on the current thread. Thereafter we assume that the
current thread continues to be the active thread for that context and
call `gst_gl_context_fill_info()` which asserts that the current
thread is the active thread.
However, if at the same time a different thread calls
`send_message_async()`, it will call into
`gst_gl_window_cocoa_send_message_async()` which will schedule the
message to be invoked using GCD. That anonymous function will also
call `gst_gl_context_activate()`, which creates a race, which can lead
to:
```
gst_gl_context_fill_info: assertion 'context->priv->active_thread == g_thread_self ()' failed
```
Fix it by using `gst_gl_context_thread_add()` to invoke `fill_info()`
on the context.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3732>
The goal of the "global" group-id is to fix new inputs that do not come from the
same "source" as others. In order to ensure all "current" streams have the same
group-id we distribute the first valid group-id to all streams.
This commit fixes two issues with that:
* When inputs are unlinked they weren't always properly resetted (it would only
work if parsebin is used, which is no longer the default in
uridecodebin3/playbin3).
* When computing the global group-id, take into account unset
group-id (i.e. GST_GROUP_ID_INVALID).
Fixes https://gitlab.freedesktop.org/gstreamer/gstreamer/-/issues/1698
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3712>
No matter if they're allocated via GSlice or malloc(). The allocator is
completely irrelevant, all local tags need to be in the primer so they
can be handled.
This didn't have any effect in practice because all local tags that
appear in the muxer are allocated via GSlice. Only from the demuxer they
might be allocated via malloc().
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3699>
As the path to the gir file is passed to hotdoc.generate_doc() and
not the build target itself, meson doesn't know about the dependency.
In turn, as the CI doesn't build everything before building the
documentation target, some gir files might not exist, for instance
in the case of gst-rtsp-server, causing the output documentation to
be empty.
The error occurred silently because hotdoc accepts wildcards for
*-sources arguments, thus it won't warn about a missing gir file as
it is legitimate for glob matching to resolve to nothing.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3686>
It might be possible to fulfill those but not with the first caps
structure. Instead of just fixating the first caps structure, check if
the preference can be fulfilled by any of the structures as the first
step.
Without this the following pipeline negotiates to mono after the
decoder because opusenc only has a single channel in its first caps
structure.
gst-launch-1.0 audiotestsrc ! audio/x-raw,channels=2 ! opusenc \
! queue ! opusdec ! queue ! opusenc ! fakesink
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3689>
When calculating the presentation offset for CMAF input in live
playback, subtract the stream_time of the fragment from the
calculated presentation offset, so that the first fragment
is played at running time zero.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3680>
This should fix pipelines such as this one to work as expected
... ! opusenc ! capsfilter caps='audio/x-opus,
channels=1; audio/x-opus, channels=2' ! ...
The expectation is that the encoder will propose the first structure
before the second one to the source.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3673>
The number of expected pads was:
* Defaulting to 1
* Or being overriden by GST_MESSAGE_STREAMS_SELECTED
This fails if upstream isn't a selectable source and has multiple streams, and
would therefore cause failures with multi-stream gapless playback
Fixes#1672
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3658>
It is quite possible to have the blocking probe called from different streaming
threads when all expected pads are present.
* Notify all waiters by using g_cond_broadcast instead of g_cond_signal
* Properly remove the probe after waiting
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3658>
gst_element_add_pad() is supposed to activate the pad if the element
state is >= PAUSED and the pad is not already active.
Unfortunately, before this patch, the activation was performed while the
element lock was still taken, which ended causing a deadlock in
gst_pad_start_task() as it attempted to post `stream-status` message in
the element, which also requires the element lock.
Elements could work around this bug by activating the pad manually
before adding it to the element.
This patch fixes the problem by performing pad activation only after the
element lock has been released.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3635>
Commit d3a66f9851 introduced a potential deadlock with two parallel release_pad
calls, where one could release the main multiqueue lock (qlock) while still
holding the reconf_lock and then calling other routines which in some conditions
may try to acquire qlock again. The second release_pad could already acquire the
qlock and then start waiting on reconf_lock, which may never be possible because
because the first one isn't releasing it until it can acquire qlock.
Fix it by holding reconf_lock for the whole durationg of qlock, making this
particular deadlock impossible.
Fixes#1642
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3571>
This is recommended by various specifications for such framerates, while
for integer framerates we continue using centiframes to allow for some
more accuracy.
Using N means that no rounding error accumulates, eventually leading to
outputting a packet with a different duration.
Some tools such as MediaInfo determine that a stream is variable
framerate if any packet has a different duration than the others, and
there is no reason I can see for not using the full 4 bytes of
resolution that the mp4 timescale offers.
Example problematic pipeline:
```
videotestsrc num-buffers=5001 ! video/x-raw,framerate=60000/1001,width=320,height=240 ! \
videoconvert ! x264enc bitrate=80000 speed-preset=1 tune=zerolatency ! h264parse ! \
video/x-h264,profile=high-10 ! mp4mux ! filesink location="result2.mp4"
```
This results in a media file that MediaInfo detects as variable
framerate because the 5000th packet has duration 99 instead of 100.
With this patch, the timescale is 60000 and all packets have duration
1001.
Related issue for context: https://bugzilla.gnome.org/show_bug.cgi?id=769041
Co-authored-by: Sebastian Dröge <sebastian@centricular.com>
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3049>
The VAAPI vaQueryVideoProcPipelineCaps() requires the context as the
parameter. So far, we always pass VA_INVALID_ID and it can succeed.
But the API does not say that and in theory, a valid context is required.
Now the new platform really needs a valid context and so we have to
delay that query until the context is created.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3613>
NVDEC launches CUDA kernel function (ConvertNV12BLtoNV12 or so)
when CuvidMapVideoFrame() is called. Which seems to be
NVDEC's internal post-processing kernel function, maybe
to convert tiled YUV to linear YUV format or something similar.
A problem if we don't pass CUDA stream to the CuvidMapVideoFrame()
call is that the NVDEC's internel kernel function will use default CUDA stream.
Then lots of the other CUDA API calls will be blocked/serialized.
To avoid the unnecessary blocking, we should pass our own
CUDA stream object to the CuvidMapVideoFrame() call
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3605>
Using the "GstBin" flags to check if an adaptive demuxer is streams-aware isn't
a good idea since it prevents using elements which aren't bins.
Instead we see if a collection was posted by the demuxer by the time a pad is
added.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3601>
If a discontinuity is detected in push mode, we need to clear the cached section
observations since they might have potentially changed.
This was only done properly when operating with TIME segments (dvb, udp,
adaptive demuxers, ...) but not with BYTE segments (such as with custom app/fd
sources).
We still don't want to flush out the PCR observations, since this might be
needed for seeking in push-based BYTE sources.
Fixes https://gitlab.freedesktop.org/gstreamer/gstreamer/-/issues/1650
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3584>
This reverts the decision from
https://bugzilla.gnome.org/show_bug.cgi?id=754230
where it was decided that we rather play safe and only use the `tfdt` if
it is "significantly different" to the sum of sample durations.
As the specification says
If the time expressed in the track fragment decode time (‘tfdt’) box
exceeds the sum of the durations of the samples in the preceding
movie and movie fragments, then the duration of the last sample
preceding this track fragment is extended such that the sum now
equals the time given in this box.
we have to use the `tfdt` in general to allow for it to signal gaps in
the stream.
A muxer producing fragments might not yet know the full duration of the
last sample of a previous fragment if the next fragment starts with a
gap, and knowing the actual start of the next fragment would potentially
require to violate latency requirements.
Additionally, the existence of `tfdt` allows to avoid accumulating
rounding errors from summing up the durations.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3586>
... when rendering on external HWND. ShowWindow() will cause
synchronous message passing to window thread and then can be blocked.
At the same time, window thread can wait for GStreamer thread.
Instead of the synchronous call, queue the task to window message
and performs from the window thread.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3583>
The rtpjitterbuffer test drop_messages_interval uses a GstClockTime for
the message drop interval. This property is defined as a guint. On
systems with 64-bit time_t but 32-bit uint, this can cause the
g_object_set function to fail to read the arguments properly.
Fixes: #1656
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3580>
Deadlock sequence:
* From a streaming thread, d3d11videosink sends synchronous message
to the parent window, so that internal (child) window can be
constructed on the parent window's thread
* App thread (parent window thread) is waiting for pipeline's
state change (to GST_STATE_NULL) but streaming thread is
blocked and waiting for app thread
To avoid the deadlock, GstD3D11WindowWin32 should send message
to the parent window asynchronously.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3570>
If we keep the old events they can be end up being passed to the app, that could
discard the protection information because it has been seen before.
Drive by improvement: use g_queue_clear_full instead of foreach+clear for
protection events.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3547>
On macOS, a Cocoa event loop is needed in the main thread to ensure
things like opening a GL window work correctly. In the past, this was
patched into glib via Cerbero, but that prevented us from updating it.
This workaround simply runs an NSApplication and then calls the
main function on a secondary thread, allowing GStreamer to correctly
display windows and/or system permission prompts, for example.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3532>
When a session request is coming in, ERROR occurs when the callee is busy.
But peer_status is the status of the caller, which is of course None when
calling someone, while self.peers[callee_id][2] is that of the callee.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2460>
jpegdec is capable to parse input frames, but if jpegparse is before,
there's no need to reparse frames. This patch configure jpegdec as
packetized, skipping parsing, if negotiated sink caps has the boolean
field 'parsed' as true.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2464>
Systems like musl libc don't support ISO 6937 in iconv. This ensures
that the MPEG-TS plugin can cope with that. There is existing support
in the plugin for other methods, so it seems to have been the original
intent anyway.
Fixes: #1314
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3245>
According to comment in gst_pulsering_stream_latency_cb, latency updates
happen every 100 ms. The code in gst_pulsering_stream_latency_cb does
not care about the actual state of the ringbuffer, pbuf->acquired or
not.
Thus, every 100 ms the ringbuf->segdone may be set, even though the
object itself might be in 'destroyed' state. On next
gst_pulseringbuffer_acquire the segdone is not touched, so playback may
resume with invalid/wrong segdone value. This finally leads to a period
of silence playing after resuming the pipeline.
The problem was found on 'not-yet-released'-hardware and so far was not
reproducible on desktop computer.
Removing the callback as long as the ringbuffer is not in 'acquired'
state solves the problem reliably on the hardware device that the issue
was detected on.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3082>
Apparently mesa 22.3.0 has updated the egl headers, and eglplatform.h now
contains commit
3670d645f4
after which xlib headers don't get included by default anymore but are
dependent upon whether USE_X11 was defined.
This breaks headless builds of gstreamer-vaapi because we always define
an internal define USE_X11 as either 1 or 0.
Change these defines to GST_VAAPI_USE_XYZ instead to avoid this.
Fixes#1634
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3555>
This wasn't really done, and is needed in order to detect potential section
changes for sections that have got identical information (such as when switching
between streams that have the same PAT/PMT pid and subtable information).
Other checks exist in tsbase to detect if the "new" PAT/PMT really is an update or not.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3530>
gst_plugin_load_by_name() assumed a plugin has a filename,
which isn't true for static plugins, leading to criticals.
If a plugin is already loaded, just return the loaded plugin,
which makes it work for static plugins as well as saving a
moment for already-loaded dynamic plugins.
Add locking in gst_plugin_is_loaded(), as a plugin may be
still being loaded in another thread.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3552>
Currently the element calls abort when failed to prepare reference
picture set. This can happent when the input stream is somehow
corrupted, like a rtsp strem with lost packets. Now it will only
return with GST_FLOW_ERROR instead of terminating whole process.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3505>
An end packet is only produced once for the last subtitle, so multiple
GAP events between subtitles would result only in a single end packet
and nothing else otherwise. This would potentially starve downstream
then, so instead forward the GAP events in that case.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/3534>
