ccconverter: ignore padding cea608 data even if marked as 'valid'

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2019>

v4l2codecs: Detect missing M2M_HOLD_CAPTURE_BUF

Produce an error if we try to use the feature of holding capture buffer
but it is not supported by the driver. Ingoring this can lead to stalls
as the driver will run-out of capture buffer to decode into. This
affects slice decoders but also split-field interlaced decoding.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2009>

v4l2codecs: h264: Set other field poc if available

Whenever the other field POC is available, pass it back to the
driver. This should not be strictly required, but it makes things
easier to debug and more forgiven.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2009>

v4l2codecs: Trace request by it FD

This helps when crossing the debug logs with the kernel logs.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2009>

v4l2codecs: h264: Fix dpb entry flag FLAG_FIELD

The logic to guess back this value was broken. Use the value
now saved into the picture instead. This was tested using
LibreELEC patched 5.15 kernel, and fixed the interlaced decode
issues.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2009>

codecs: h264: Save the field_pic_flag

This flag is set when the stream is interlaced and the specific
slice is made of single parity fields rather the paired at the
macroblock layer. This is rarely needed in late decoding process
but the Rockchip RKVDEC HW interface requires it, hence needs to
be passed through V4L2 Stateless interface.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2009>

d3d11: Fix example build error

```
../subprojects/gst-plugins-bad/tests/examples/d3d11/d3d11device.cpp:129:50: error: invalid conversion from ‘int’ to ‘DXGI_FORMAT’ [-fpermissive]
  129 |     D3D11_SHADER_RESOURCE_VIEW_DESC srv_desc = { 0, };
```

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2029>

gl: cocoa: fix warnings of unused variables

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2026>

Fetch libxml2 subproject using http instead of ftp

The official releases of libxml2 have been migrated to gitlab where
they are published for download via HTTP instead of FTP. Besides
adapting to the new location we now also get the benefit that the
tarball can be downloaded in restricted networks where FTP might be
blocked.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2020>

navigationtest: Add some support for modifiers

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2010>

navigation: Add support for key Modifiers in all relevant events

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2010>

va: h265dec: Fix a crash because of missing reference frame.

Some problematic H265 stream may miss the reference frame in the DPB,
and get some message like: "No short term reference picture for xxx".
So there may be empty entries in ref_pic_list0/1 when passing to
decode_slice() function of sub class. We need to check the NULL pointer.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2018>

openexr: Specify modules when finding OpenEXR.

Specify modules to look for OpenEXR when CMake is used, as we may have
CMake config files instead of pkg-config files that result from building
OpenEXR, which may be built with CMake which is typically the case on Visual
Studio builds.

In this case, Meson does seem to find the 'OpenEXR' package with CMake
after trying pkg-config, but does not consider it enough without the
'modules:' argument.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2014>

webrtcbin: Update documentation of 'get-stats' action signal

Some stats fields are updated according to the current implementation.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2005>

rtpptdemux: fix leak of caps when ignoring a pt

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2025>

Revert "webrtcbin: fix msid line and allow customization"

This reverts commit 3cad3455377d5a22faa138d9df840257059776c8.

That commit was breaking the association between an audio and
a video track in the standard case.

In practice, to support carrying separate MediaStream, we are
going a way to map what MediaStreamTrack belong to what MediaStream,
but that will require some thinking about the API.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2023>

tsdemux: Don't check SCTE events for being too far from the PCR

Otherwise it happens that SCTE events can only be scheduled up to 15
seconds since the last seen PCR, which is a useless restriction.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2021>

navigation: Add missing annotation to send_event_simple

Adds the missing "transfer full" annotation for the event argument.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2017>

webrtcbin: fix msid line and allow customization

From https://datatracker.ietf.org/doc/html/draft-ietf-mmusic-msid-16:

> Multiple media descriptions with the same value for msid-id and
> msid-appdata are not permitted.

Our previous implementation of simply using the CNAME as the msid
identifier and the name of the transceiver as the msid appdata was
misguided and incorrect, and created issues when bundling multiple
video streams together: the ontrack event was emitted with the same
streams for the two bundled medias, at least in Firefox.

Instead, use the transceiver name as the identifier, and expose
a msid-appdata property on transceivers to allow for further
customization by the application. When the property is not set,
msid-appdata can be left empty as it is specified as optional.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2003>

webrtc: Update dependencies of the Rust examples

And also clean up code a bit while updating to new APIs.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2016>

Fix license as GES is released under the LGPL2+ license

This was a mistake that some of the licensing notice in a few files was
referring to GPL3+ and it needs fixing

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1956>

wpe: Mark first audio buffer as discont

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1492>

wpe: Use `about:blank` as default URL to support only using load-bytes

WebKit is not going to render anything until a URI is set, leading to a
WPE posting a `WPE View did not render a buffer` error message. To avoid
requiring the user to know it if they only want to use
`wpesrc::load-bytes` we can just use `about:blank` as default and
everything will work as users would expect.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1492>

validate:scenario: Support args in the `emit-signal` action

This was never needed before but it is generally useful and there
was a todo about it

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1492>

webrtc_sendrecv.py: Link pads instead of elements

This was not a problem here because even if we end up accidentally
linking to the wrong pad, things will work out eventually as long as
one pad-added is emitted for each pad that is added.

But it will be a huge problem if someone copies this code and changes
something that requires different handling for different sorts of
pads. The resultant code will be racy. Let's not do this, it's a bad
example.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2008>

README: update documentation for macOS

Give hints regarding the certificates
needed by Python.

urllib.error.URLError: urlopen error [SSL ..

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1794>

fakevideosink,fakeaudiosink: Proxy handoff and preroll-handoff signals

Proxy signals for application to be able to consume them

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1585>

aom: av1enc: Specify Temporal Unit alignment

Encoded bitstream consists of leading Temporal delimiter OBU
with frame, that's Temporal Unit alignment.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1471>

qsvencoder: Clear caps after debug print

Otherwise debug function will print (null) for alread cleared pointer

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2012>

qsvh264enc: Fix profile and level setting in codec data

The profile field in SPS is located after header byte

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2012>

nvcodec: Add new Direct3D11/CUDA mode encoder implementation

Adding new encoder elements nvd3d11{h264,h265}enc for Direct3D11
input support and re-written nvcuda{h264,h265}enc elements.
Newly writeen elements have some differences compared with old
nv{h264,h265}enc including non-backward compatible changes.

* RGBA is not a supported input format any more:
  New elements will support only YUV formats to avoid implicit conversion
  done by hardware. Ideally it should be done by upstream element
  in order to have more control on it. Moreover, RGBA support can cause
  redundant RGBA -> YUV conversion if multiple encoders are
  used for the same RGBA input
* Subsampled planar format support is dropped:
  I420 and YV12 format are not supported formats for Direct3D11.
  Although it's supported in CUDA mode, it's not a hardware friendly
  memory layout and it will waste GPU memory since UV planes
  will have large padding due to the memory layout requirement of NVENC.
* GL support is dropped: Similar to the RGBA case,
  GL support in encoder would be suboptimal if GL input is
  used by multiple encoders, because each encoder will copy GL memory
  into CUDA memory.
  Upstream cudaupload element can be used for GL <-> CUDA
  interop instead.
* No more pre-allocation of encoder input surfaces. New implementation
  will use input CUDA memory without copy (zero-copy) or
  will copy into a NVENC's input buffer struct in case of
  system memory input.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1997>

qt: Add touch event support

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1633>

gtk: Add touch event support

Add a handler for touch events to gtkbasewidget.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1633>

wpevideosrc: Add touch event support

Dispatches a list of active touch events to the wpe view on each
received TOUCH_FRAME event. Touch inputs currently only move the cursor,
since wpe doesn't seem to support clicking/scrolling or zooming with
touch input.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1633>

xvimagesink: Add touch event support

Send touch events for XI_TouchBegin, XI_TouchEnd, and XI_TouchUpdate
events, grouping events with identical timestamps into one TOUCH_FRAME.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1633>

ximagesink: Add touch event support

Send touch events for XI_TouchBegin, XI_TouchEnd, and XI_TouchUpdate
events, grouping events with identical timestamps into one TOUCH_FRAME.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1633>

navigationtest: Display touchscreen events, log all events

Represents touchscreen events as a trail of black squares, one for each
reported position. Additionally, this adds the `display-mouse` and
`display-touch` properties to toggle visibility of mouse/touchscreen
events, since touchscreens often emulate mouse events, as well as
logging for all received navigation events.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1633>

navigation: Add API for touchscreen events

Add 5 new navigation event types for touchscreen events, with the same
naming and meaning as in libinput - touch-down, touch-motion, touch-up,
touch-frame and touch-cancel - as well as constructors and parse
functions for them.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1633>

all: Use new navigation interface and API

Use and implement the new navigation interface in all relevant sink elements,
and use API functions everywhere instead of directy accessing the event structure.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1633>

navigation: Add coordinate helper functions

Add a function to get x/y coordinates from suitable navigation events,
and one to create a copy with given coordinate values.

For e.g. translating event coordinates, this avoids having to either
switch on the event type to select the right parse function, or
having to rely on implementation details of the underlying event
structure.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1633>

navigation: Improve interface to avoid exposing implementation details

This deprecates the current send_event interface, and the wrapper
functions based on it, replacing it with a send_event_simple interface and
wrapper function. Together with the new event constructors, this avoids
implementations having to directly access the underlying structure.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1633>

msdkenc: add macro definitions to avoid the gst-inspect error

Since the strings are empty for GST_MSDK_CAPS_MAKE_WITH_DMABUF_FEATURE
and GST_MSDK_CAPS_MAKE_WITH_VA_FEATURE, when excuting
gst-inspect-1.0.exe msdkh265enc, there will be convert static caps error
because of the extra semicolon between two empty strings. Now macro
definitions are added to avoid this issue.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2004>

wavparse: handle query in any parse state

In order to create the stream_id, we need to
pass the query to the default query handler.

If the parse state is different from GST_WAVPARSE_DATA
the query should be passed to the default query
handler.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1987>

flvmux: Clean up aggregate's control flow

This unifies exits to go through a single out label. It mostly
simplifies how EOS is handled.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1035>

msdkdec: Check width and height of mfxVideoParam before allocation

DecodeHeader must be called to fill the mfxVideoParam before allocation,
and thus the check for width and height in mfxVideoParam is necessary.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1817>

doc: Update cache after NV12_8L128 addition

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1379>

v4l2: Add NV12_8L128 and NV12_10BE_8L128

These formats are used by i.MX 8QXP/8QM VPU.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1379>

video: Add support for linear 8x128 NV12 tiles and 10bit BE tiles

This adds linear 8x128 NV12 based tiles and NV12 10bit big endian tiles.
These formats are used by i.MX 8QXP/8QM VPU and exposed in V4L2.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1379>

v4l2codecs: Fix memory leak

Free pending_requests array when releasing decoder to avoid memory leak

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2000>

av1decoder: Sync duplicate_picture with VP9 one

Pass the current frame to the duplicate_picture callback. This makes it easier
to set the frame's output_buffer if we already have one available. Also
documented that unlike VP9, it is not optional to implement this as the
picture will populate the DPB if it is a key-frame. To ensure this, remove the
default implementation.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1992>

vp9decoder: Copy system_frame_number in duplicate_picture

Just like AV1, copy the system_frame_number from the original picture to make
it clear they reference the same data.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1992>

cdoecs: av1: Copy system_frame_number in duplicate_picture

The system_frame_number is notably used by V4L2 decoder as a unique
indentifier for the frame that was decoded. This value is used to tell driver
which frame to reference, as V4L2 does not have an efficient mechanism to
otherwise pass back the frames.

For this reason, and because it is more ligical, copy the original
system_frame_number into the duplicate picture instead of using the current
frame.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1992>

meson: va: Define GST_USE_UNSTABLE_API

Remove extra_c_args which is not defined in this meson file at all,
and define GST_USE_UNSTABLE_API to avoid build warnings

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1999>

rtpmanager/rtx: implement initial support for reading/writing rid extensions

Two RTP Header extensions are very relevant for rtprtxsend/receive.
1. "urn:ietf:params:rtp-hdrext:sdes:rtp-stream-id": will always be removed
2. "urn:ietf:params:rtp-hdrext:sdes:repaired-rtp-stream-id": will be written
    instead of the "rtp-stream-id" header extension.

Currently it's only a simple replacement of one header extension for
another however a future change would only add the relevant extension
based on some heuristics (like, video frames only on one of the rtp key
frame buffers, or only until the rtx ssrc has been validated by the peer)
in order to reduce the required bandwidth.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1759>

test: add tests for sdes-based RTP header extensions

mid, stream id and repaired stream id.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1759>

rtpmanager: add support for RFC8852 (rid) RTP header extensions

Both for regular RID and for adding on a repaired (RTX) etc stream.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1759>

rtpmanager: add support for writing RFC8843 (BUNDLE mid) RTP header extension

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1759>

codecs: av1: Fix state when we show existing keyframe

Showing existing keyframe have special meaning in AV1. All the references
frame will be refreshed with the original keyframe information. The refresh
process (7.20) is implemented by saving data from the frame_header into the
state. To fix this special case, load all the relevant information into the
frame_header.

As there is nothing happening in between this and the loading of the key-frame
into the state, this patch also remove the separate API function, using it
internally instead.

Fixes #1090

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1971>

videocrop: Add support for v210

Like UYVY and similar formats this is rounding down to the start of the
previous macro-pixel to not mix up the different components.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1988>

videocrop: Use GST_ROUND_DOWN_2 instead of re-defining a local version

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1988>

videocrop: Rename PACKED_COMPLEX to PACKED_YVYU

It's not handling any kind of complex packed format, only formats that
are like YVYU.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1988>

d3d11: Fix some typos

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1985>

validate: Totally hide media descriptor node types from the API

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1986>

meson: Fix warning about build_root() and source_root()

```
../meson.build:384: WARNING: Project targeting '>= 0.60.0' but tried to use feature deprecated since '0.56.0': meson.build_root. use meson.project_build_root() or meson.global_build_root() instead.
../meson.build:385: WARNING: Project targeting '>= 0.60.0' but tried to use feature deprecated since '0.56.0': meson.source_root. use meson.project_source_root() or meson.global_source_root() instead.
...
WARNING: Deprecated features used:
 * 0.56.0: {'meson.source_root', 'meson.build_root'}
 ```

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1977>

meson: Bump all meson requirements to 0.60

Lots of new warnings ever since
https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1934

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1977>

openexr: Fix some warnings

```
../subprojects/gst-plugins-bad/ext/openexr/gstopenexrdec.cpp:46:24: warning: ‘Imf_3_1::Int64’ is deprecated: use uint64_t [-Wdeprecated-declarations]
   46 |   virtual Int64 tellg ();
      |                        ^
In file included from ../subprojects/gst-plugins-bad/ext/openexr/gstopenexrdec.cpp:32:
/usr/include/OpenEXR/ImfInt64.h:23:32: note: declared here
   23 | typedef IMATH_NAMESPACE::Int64 Int64;
      |                                ^~~~~
../subprojects/gst-plugins-bad/ext/openexr/gstopenexrdec.cpp:47:32: warning: ‘Imf_3_1::Int64’ is deprecated: use uint64_t [-Wdeprecated-declarations]
   47 |   virtual void seekg (Int64 pos);
      |                                ^
In file included from ../subprojects/gst-plugins-bad/ext/openexr/gstopenexrdec.cpp:32:
/usr/include/OpenEXR/ImfInt64.h:23:32: note: declared here
   23 | typedef IMATH_NAMESPACE::Int64 Int64;
      |                                ^~~~~
../subprojects/gst-plugins-bad/ext/openexr/gstopenexrdec.cpp:67:26: warning: ‘Imf_3_1::Int64’ is deprecated: use uint64_t [-Wdeprecated-declarations]
   67 | Int64 MemIStream::tellg ()
      |                          ^
In file included from ../subprojects/gst-plugins-bad/ext/openexr/gstopenexrdec.cpp:32:
/usr/include/OpenEXR/ImfInt64.h:23:32: note: declared here
   23 | typedef IMATH_NAMESPACE::Int64 Int64;
      |                                ^~~~~
../subprojects/gst-plugins-bad/ext/openexr/gstopenexrdec.cpp:73:29: warning: ‘Imf_3_1::Int64’ is deprecated: use uint64_t [-Wdeprecated-declarations]
   73 | MemIStream::seekg (Int64 pos)
      |                             ^
In file included from ../subprojects/gst-plugins-bad/ext/openexr/gstopenexrdec.cpp:32:
/usr/include/OpenEXR/ImfInt64.h:23:32: note: declared here
   23 | typedef IMATH_NAMESPACE::Int64 Int64;
      |                                ^~~~~
```

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1977>

subprojects/avtp: Update to latest avtp 0.2.0

Fixes some meson warnings too.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1977>

codecparsers: av1parser: Fix a typo in FilmGrainParams comment.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1974>

discoverer: chain up to parent finalize methods in all our types

Fixes a memory leak:

Direct leak of 32 byte(s) in 1 object(s) allocated from:
    #0 0x5ac5cd in malloc /src/llvm-project/compiler-rt/lib/asan/asan_malloc_linux.cpp:129:3
    #1 0x1007007 in g_malloc /work/glib-2.72.0/_builddir/../glib/gmem.c:125:13
    #2 0xf82e82 in g_data_set_internal /work/glib-2.72.0/_builddir/../glib/gdataset.c:464:8
    #3 0xf833f7 in g_datalist_id_set_data_full /work/glib-2.72.0/_builddir/../glib/gdataset.c:670:3
    #4 0xef81be in g_object_notify_queue_freeze /work/glib-2.72.0/_builddir/../gobject/gobject.c:295:7
    #5 0xef79c6 in g_object_unref /work/glib-2.72.0/_builddir/../gobject/gobject.c:3632:16
    #6 0x5e58bf in LLVMFuzzerTestOneInput /src/gstreamer/ci/fuzzing/gst-discoverer.c:132:5
    #7 0x4dd1a2 in fuzzer::Fuzzer::ExecuteCallback(unsigned char const*, unsigned long) /src/llvm-project/compiler-rt/lib/fuzzer/FuzzerLoop.cpp:611:15
    #8 0x4dc98a in fuzzer::Fuzzer::RunOne(unsigned char const*, unsigned long, bool, fuzzer::InputInfo*, bool, bool*) /src/llvm-project/compiler-rt/lib/fuzzer/FuzzerLoop.cpp:514:3
    #9 0x4de6c4 in fuzzer::Fuzzer::ReadAndExecuteSeedCorpora(std::__Fuzzer::vector<fuzzer::SizedFile, std::__Fuzzer::allocator<fuzzer::SizedFile> >&) /src/llvm-project/compiler-rt/lib/fuzzer/FuzzerLoop.cpp:809:5
    #10 0x4dea29 in fuzzer::Fuzzer::Loop(std::__Fuzzer::vector<fuzzer::SizedFile, std::__Fuzzer::allocator<fuzzer::SizedFile> >&) /src/llvm-project/compiler-rt/lib/fuzzer/FuzzerLoop.cpp:857:3
    #11 0x4ce4a0 in fuzzer::FuzzerDriver(int*, char***, int (*)(unsigned char const*, unsigned long)) /src/llvm-project/compiler-rt/lib/fuzzer/FuzzerDriver.cpp:912:6
    #12 0x4f6f52 in main /src/llvm-project/compiler-rt/lib/fuzzer/FuzzerMain.cpp:20:10
    #13 0x7f1c709980b2 in __libc_start_main (/lib/x86_64-linux-gnu/libc.so.6+0x240b2)

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1984>

d3d11screencapturesrc: Allow capturing screen sub-area

Adds crop-x, crop-y, crop-width, crop-height properties specifying the
screen area to capture.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1932>

webrtcbin: Warn when offer didn't intersect with transceiver caps

We were silently falling back to creating a recvonly offer if the caps
didn't intersect.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1864>

webrtc-sendrecv: Fix create-answer caps negotiation

We need to parse the payload type map provided by the offer SDP and
set those values on the payloader, otherwise webrtcbin will create
a recvonly answer SDP and we won't send anything to the browser.

Fixed it for both C and Python sendrecv examples.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1864>

webrtc_sendrecv.py: Add picture-id-mode to rtpvp8pay

This doesn't just make TWCC stats perform better, it also fixes
stuttery video playback in Chrome.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1864>

webrtc_sendrecv.py: Print an error on unknown JSON message

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1864>

webrtc_sendrecv.py: Add missing copyright headers

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1864>

webrtc_sendrecv.py: Implement all negotiation modes

Earlier, the example only supported one negotiation mode:
* Browser client is running, gstreamer starts a call and sends offer

Now these three modes are also supported:
* Browser client is running, gstreamer starts a call and sends an
  offer request
* gstreamer connects and waits for browser client to start a call and
  send an offer
* gstreamer connects and waits for browser client to start a call and
  send an offer request

The following features are still missing:
* Data channel support
* TWCC support + stats logging

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1864>

webrtc_sendrecv.py: Make it executable

Why wasn't it already. Tired of typing 'python webrtc_sendrecv.py'.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1864>

webrtc_sendrecv.py: Fix event loop usage for messages

Instead of creating a new loop, we should just be fetching the running
loop, then doing a blocking network call inside the callback, schedule
it on the event loop. This is what the C example does too.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1864>

webrtc_sendrecv.py: Use default arg for server URL

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1864>

ci/fuzzing: update glib version to 2.72.0

Fixes an issue where on configure, glib's internal gnulib was tripping
up the undefined behaviour sanitizer with a divide by 0.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1976>

glvideomixerelement: send translated navigation events to the relevant sink pads

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1495>

compositor: send translated navigation events to the relevant sink pads

Fixes #888

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1495>

ci/fuzzling: fix oss-fuzz build

gst-plugins-base doesn't have a cairo option anymore.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1975>

validate: Cleanup ABI and make it ready to be stabilized this cycle

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1810>

validate: Introspect gst-validate-runner.c

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1810>

validate: Add getters for ValidateReports

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1810>

validate: Change IssueId type to guint32

Which is a bit more bindings-friendly.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1810>

validate: ssim: Respect requested dump recurrence parameter

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1810>

validate: Keep scenario alive enough when executing actions

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1810>

validate: Introspection annotation cleanups

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1810>

validate: Fix padding and constants

Without this the Rust -sys validate crate tests don't pass.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1810>

validate: Expose all public headers in main header

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1810>

validate: Fix pkgconfig name

Making it consistent with the reset of GStreamer. This also means that
anyone wanting to target -validate >= 1.22 needs to change the way they
link against it.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1810>

rtpjitterbuffer: Fix invalid memory access in rtp_jitter_buffer_pop()

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1973>

documentation: improve misleading wording

The documentation for several gst_*_writable_structure functions stated
that they would never return NULL, without making clear that the passed
object is required to be writable. This changes the wording in those
cases to make that requirement more clear.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1784>

ccconverter: drop data with a warning if scratch buffers overflow

Instead of asserting which could bring down the entire application.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1972>

v4l2videodec: set frame duration according to framerate

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1953>

uri: Build doubly-linked list by prepending items

As outlined in the API documentation, g_list_append() iterates over the whole
list, which can quickly introduce performance issues when the list becomes very
big, such as for data URIs for instance.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1909>

typefind: Skip parsing of data URIs

Commit a46ab2ced20d757e0e770d4de1edc3a152cc4f2f introduced a regression,
breaking typefinding for media content muxed in mp4 container and serialized to
data URIs. For this case it doesn't make sense to look for a file extension, so
skip URI parsing.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1909>

basesink: Fix annotations

We should annotate the Class and not the object itself.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1965>