cudaupload,cudadownload: Add support for dGPU NVMM

Implement NVMM <-> CUDA, GL, SYSTEM memory conversion. Jetson is
not supported yet. Note that NVMM <-> GL interop on Jetson platform
is supported by GstGL

Some example pipelines are:
- Convert NVMM to GstGL memory
  nvv4l2decoder ! "video/x-raw(memory:NVMM)" ! cudadownload ! "video/x-raw(memory:GLMemory)" ! glimagesink

- Upload system memory to NVMM and encode
  video/x-raw,format=NV12 ! cudaupload ! "video/x-raw(memory:NVMM)" ! nvv4l2h264enc

- Convert NVMM to GstCUDA memory and encode
  nvvideoconvert ! "video/x-raw(memory:NVMM)" ! cudaupload ! "video/x-raw(memory:CUDAMemory)" ! nvh264enc

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1865>

Update wrap files from latest wrapdb version

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1877>

va: encoder: Don't preallocate reconstruct buffers.

It's not required by VA to register the reconstruct buffers at context
creation, just as in decoders.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1857>

v4l2bufferpool: Fix race condition between qbuf and pool streamoff

There is a chance that pool->buffers[index] sets BUFFER_STATE_QUEUED, but
it has not been queued yet which makes pool->buffers[index] still NULL.
At this time, if pool_streamff release all buffers with BUFFER_STATE_QUEUED
state regardless of whether the buffer is NULL or not, it will cause segfault.

To fix this, also check buffer when streamoff release buffer.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1842>

gst-rtsp-server: Plug a few memory leaks in tests

Found and fixed a few memory leaks in the gst_rtspserver, gst_onvif and
gst_stream tests by running the tests in valgrind.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1742>

flvmux: Add protection when unref GstFlvMuxPad

This is to avoid gst_object_unref: assertion 'object != NULL' failed.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1843>

gstplay: tests: Keep track of errors/warnings

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1869>

gstplay: Do not error out on message parsing failures

Specially when parsing errors and warnings, the details field can be NULL and
the gst_structure_get() call would return FALSE in such cases, triggering false
positive errors.

Follow-up for #1063

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1869>

gstplay: Fix warning parsing API

The GError is an out parameter, so should be a ** parameter, like the details
parameter.

See also #1063

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1869>

cudamemorycopy: Fix build when gl is disabled

Fixes: https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/issues/1707
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1868>

gst-rtsp-server: fix race in rtsp-client

When tunneling over HTTP, if connection on the second channel happens
before the control timer is created we may trigger an assert in
rtsp_ctrl_timeout_remove(). Avoid that by taking the priv->lock before
attaching the client thread to the context.

Fixes #1025

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1867>

msdkdec: Fix to not trigger SFC scaling when vpp at downstream

When there is vpp scaling downstream, we need to make sure SFC is not
triggered because vpp may fall into passthrough mode which causes
the decoder negotiation to create src caps with vpp scaled width/height.

This patch includes bitstream's original size in first query with
downstream in gst_msdkdec_src_caps, which is the same for what we do for
color format in this query. This is to ensure SFC scaling starts to
work only when downstream directly asks for a different size instead of
through vpp.

Note that here SFC scaling follows the same behavior as msdkvpp:
if user only changes width or height, e.g. dec ! video/x-raw,width=xx !,
the height will be modified to the value which fits the original DAR.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1838>

encodebasebin: Use GST_DEBUG instead of GST_ERROR when skipping muxer

_get_muxer() skips the muxers that do not satisfy the requirement and select
the desired one. It should not print error log, so use debug log instead.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1830>

nvcodec: Add missing null check in context sharing code

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1834>

nvcodec: Refactor cudaupload/download and add support for GL memory

* Implement new baseclass GstCudaMemoryCopy to remove duplicated
  cudaupload/download code
* Add support for CUDA <-> GL memory conversion via cudaupload/download

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1834>

nvcodec: Refactor GstCudaMemory abstraction

* Hide GstCudaMemory member variables
* Make GstCudaAllocator object GstCudaContext independent
* Set offset/stride of memory correctly via video meta
* Drop GST_BUFFER_POOL_OPTION_VIDEO_ALIGNMENT support.
  This implementation actually does not support custom alignment
  because we allocate device memory via cuMemAllocPitch
  of which alignment is almost uncontrollable

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1834>

nvcodec: Refactor basetransform subclasses

* cudaupload/download
  - Specify only formats actually we can deal with
    nvcodec elements, not all video formats
  - Supports CUDA output for download and input for upload in order
    to make passthrough possible, like other upload/download elements.
* cudabasetransform
  - Reset conversion element if upstream CUDA memory
    holds different CUDA context and the element can accept it.
    This is the same behavior as corresponding d3d11 filter elements.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1834>

cudabufferpool: Move GstCudaContext object to public member

... so that plugins can understand assosicated cuda context with the
pool.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1834>

cudacontext: Require explicit gpu id instead of auto (-1)

Sync up with GstD3D11Device implementation. The auto stuff should
be handled in context sharing step, not device creation.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1834>

nvcodec: Indent update

... to prevent gnu indent version dependent diffs.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1834>

cudacontext: Remove unnecessary cuInit() call

It's already called at plugin init time

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1834>

av1parse: Add a comment when the detection of TU fails.

Also update the print message.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1863>

camerabin: Remove drain query

This was to support very old V4L2 kernel. As we moved to DMABuf and can now
detach buffers on renegotiation, the buffer it tries to fix no longer exist.
The risk to blocking indefinitly the application does still exist though.

Fixes #1070

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1861>

doc: AV1 demuxers now expose their alignment

Update the chache accordingly.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1837>

av1parse: check the error return when parse_alignment_from_caps()

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1837>

av1parse: Use the intersected caps to decide the alignment.

When we negotiate with downstream, We should use the intersected
caps of input and output to decide the alignment and stream format.
The current code just uses the input caps which may lack the stream
format.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1837>

ivfparse: Set the alignment to tu for AV1 stream.

Just do the same as other demux.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1837>

av1parse: Only detect the stream format when input alignment is tu.

The demux now outputs the AV1 stream in "tu" alignment, so we do not need
to detect the input alignment. But the annex b stream format is not recognized
by the demux, we still need to detect that stream format for the first input.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1837>

matroska: Fix AV1 alignment to TU

Matroska stores AV1 in temporal unit, so that all OBU sharing the same
timestamp are put together. This was previously just assumed, which isn't
safe now that we have more alignments.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1837>

isomp4: Fix AV1 default alignment

ISOMP4 store TU (temporal units) worth of AV1. Expose this in the
caps to reduce overhead in the parser, and in the muxer to avoid
storing frames split in the wrong way.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1837>

av1parse: Save the default alignment

Otherwise the transformation is not applied and decoders may not
get the appropriate format despite the caps being negotiated
correctly.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1837>

av1parse: Remove duplicated check

The case of both strings being empty is already checked in the
helper function. No functional changes.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1837>

vp9parse: Fix auto-plugging of HW frame decoder

Decoders that required frame aligmment and didn't have an associated
alpha decoder were skipped. This is because the parser was constructing
caps based on the software alpha decoder, which specify super-frame
alignment.

Iterate over the caps to filter the one that have a matching codec-alpha, with
the semantic the no codec-alpha field means codec-alpha=false. Then if
everything was removed, callback to the original, so that the first non-alpha
decoder will be picked.

Fixes #820

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1855>

matroskamux: allow width+height caps changes for VP8/9

For VP8 and VP9, width+height changes are signalled inband.

Refs https://github.com/Kurento/bugtracker/issues/535 and
https://gitlab.freedesktop.org/gstreamer/gst-plugins-good/-/merge_requests/1047/diffs?commit

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1657>

matroskamux: allow width + height changes for avc3|hev1

For avc3 and hev1, the intent was to allow more flexibility for caps changes
(see https://gitlab.freedesktop.org/gstreamer/gst-plugins-good/-/merge_requests/1047/diffs?commit_id=9bd8d608d5bae27ec5ff09e733f76ca32b17420c)
however width and resolution were previously omitted.

avc3 and hev1 specifically support changing stream-parameters on the fly, whereas avc1/hvc1 disallow in-band SPS.

This commit allows for changes to width and height for these which is in line with matroskamux's behaviour prior to 1.14.0.

Practically speaking, one use case where this is commonly seen is when capturing a WebRTC stream, as the browser will adapt the resolution live.

Suggested-by: Mathieu Duponchelle "<mathieu@centricular.com>"
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1657>

deinterlace: Prevent race between _set_method and latency query

It's possible that the method is being manipulated while downstream
queries our latency, leading to crashes.

Prevent that from happening.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1854>

tsdemux: Handle PES headers bigger than a mpeg-ts packet

While the actual PES header parser could notify us that it needed more data, we
would never actually act on it.

This commit will accumulate incoming packets in such situation and re-attempt
the header parsing.

Fixes #1027

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1852>

va: basetransform: Add device-path read-only property.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1844>

va: basedec: Add device-path read-only property.

And elements will notify the used device-path if display is DRM.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1844>

gldisplay: Reorder GST_GL_WINDOW check for egl-device

"egl-device" should be checked before the "egl", otherwise unexpected egl will be picked

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1853>

vah264enc: Remove GST_PARAM_CONDITIONALLY_AVAILABLE.

Since it's used only when the property is conditionally installed, and
these parameters are always installed.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1845>

va: encoder: Clean dead code.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1845>

va: encoder: Fix error code path when open.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1845>

vah264enc: Remove preferred_output_delay.

In the current state of the encoder it's just dead code.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1845>

vah264enc: Avoid mapping coded buffer twice.

Currently for copying the coded buffer onto a GStreamer buffer, the
coded buffer is mapped two times: one for getting the size, and later
for do the actual copy. We can avoid this by doing directly in the
element rather than in the general encoder object.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1845>

vah264enc: Single parameter array per picture.

There's no need, at least in H.264, to hold a global parameter array,
since there's no need to submit SPS, trellis, hdr or quality in every
buffer, but only on IDR ones.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1845>

va: Fix log message

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1845>

play: Fix error parsing API

The GError is an out parameter, so should be a ** parameter, like the details
parameter.

Fixes #1063

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1849>

webrtc: support renegotiating adding/removing RTX

We need to always add the RTX/RED/ULPFEC elements as rtpbin will only
call us once to request aux/fec senders/receivers.

We also need to regenerate the media section of the SDP instead of
blindly copying from the previous offer.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1762>

msdkenc: Change default ref frames number as 0

When user does not set ref-frames in pipeline, we should let MediaSDK
decide the reference frame number via setting default value as 0 at
gstreamer side.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1839>

meson: wic: Fix typo

Should check WIC build option, not MediaFoundation

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1847>

soup: Fix static build with MSVC

../ext/soup/gstsouploader.c(818): error C4098: '_soup_session_send_async': 'void' function returning a value

It's technically a false warning, but that's how MSVC works, so fix
it.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1805>

soup: Fix pkgconfig generation and documentation

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1805>

soup: Fix static build when default_library=both

Fixes https://gitlab.freedesktop.org/gstreamer/gstreamer/-/issues/1007

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1805>

soup: Don't error out in static build unless option is enabled

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1805>

soup: Lookup libsoup dylib files on Apple platforms

Fixes #1007

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1805>

v4l2src: Reset the compose window to the default after setting format

When the size of V4L2 capture or output is changes with VIDIOC_S_FMT,
the device is only required to update the compisition window to fit
inside the new frame size.  This can result in captured data only being
updated on a portion of the frame after a resize.

Update the composition window to the default value determined by the
V4L2 device driver whenever the format is changed to make sure that
all image data is composed to its full size.

Fixes #765

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1806>

gst-play: Allow switching to previous tracks

Implement case sensitive key-press handling,
and make 'V', 'A', 'S' switch to the previous
video, audio, or subtitle track. The lower-case
keypress cycles to the next track, as before

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1841>

qsv: Remove strcpy for DRM device path

Simplify code. It's just one-time allocation, so don't need to worry
about leaking.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1835>

msdk: Don't print error log for missing DRM device path

It's expected in case that such DRM device is unavailable for some reasons,
specifically non-Intel platform or so

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1836>

sdp: support multiple rid parameters

As specified formally in RFC8851

Each rid description is placed in its own caps field in the structure.
This is very similar to the already existing extmap-$id sdp<->caps
transformations that already exists.

The mapping is as follows:

  a=rid:0 direction ';'-separated params

where direction is either 'send' or 'recv'

gets put into a caps structure like so:

   rid-0=(string)<"direction","param1","param2",etc>

If there are no rid parameters then the caps structure is generated to
only contain the direction as a single string like:

   rid-0=(string)direction

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1760>

wic: Add support for JPEG and PNG decoding

Adding Windows Imaging Component (WIC) plugin with JPEG/PNG decoding
support.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1605>

directshow: Fix for division by zero

The AvgTimePerFrame value may be unknown. Use arbitrary
value (30 fps) instead of crashing.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1828>

matroska-mux: Handle pixel-aspect-ratio caps field correctly when checking caps equality

Not having this field is equivalent with it being 1/1 so consider
it like that. The generic caps functions are not aware of these
semantics and would consider the caps different, causing a negotiation
failure when caps are changing from caps with to caps without or the
other way around.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1826>

matroska-mux: Handle multiview-mode/flags caps fields correctly when checking caps equality

Not having these fields is equivalent with them being mono/0 so consider
them like that. The generic caps functions are not aware of these
semantics and would consider the caps different, causing a negotiation
failure when caps are changing from caps with to caps without or the
other way around.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1826>

Revert "ci: Disable windows jobs for now"

This reverts commit be2815a07a0df7a33d136a2be9cccd0969808642.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1827>

qsv: Add H.264 decoder

Initial decoder implementation with baseclass

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1786>

webrtc_sendrecv.py: Sync element props with C version

Also add indentation to make it easier to read

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1821>

webrtc_sendrecv.py: Ensure that gst-python overrides are installed

Otherwise fetching of the offer will fail with a cryptic error:

```
Traceback (most recent call last):
  File "/../gstreamer/subprojects/gst-examples/webrtc/sendrecv/gst/webrtc_sendrecv.py", line 56, in on_offer_created
    offer = reply['offer']
TypeError: 'Structure' object is not subscriptable
```

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1821>

webrtc_sendrecv.py: Don't try to set state on a None pipe

```
ERROR peer '5762' not found
Traceback (most recent call last):
  File "/../gstreamer/subprojects/gst-examples/webrtc/sendrecv/gst/webrtc_sendrecv.py", line 190, in <module>
    res = loop.run_until_complete(c.loop())
  File "/usr/lib64/python3.10/asyncio/base_events.py", line 641, in run_until_complete
    return future.result()
  File "/../gstreamer/subprojects/gst-examples/webrtc/sendrecv/gst/webrtc_sendrecv.py", line 155, in loop
    self.close_pipeline()
  File "/../gstreamer/subprojects/gst-examples/webrtc/sendrecv/gst/webrtc_sendrecv.py", line 142, in close_pipeline
    self.pipe.set_state(Gst.State.NULL)
AttributeError: 'NoneType' object has no attribute 'set_state'
```

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1821>

webrtc_sendrecv.py: Fix SSLError when connecting to websocket server

```
  File "/../gstreamer/subprojects/gst-examples/webrtc/sendrecv/gst/webrtc_sendrecv.py", line 189, in <module>
    loop.run_until_complete(c.connect())
  File "/usr/lib64/python3.10/asyncio/base_events.py", line 641, in run_until_complete
    return future.result()
  File "/../gstreamer/subprojects/gst-examples/webrtc/sendrecv/gst/webrtc_sendrecv.py", line 40, in connect
    self.conn = await websockets.connect(self.server, ssl=sslctx)
  File "/home/nirbheek/.local/lib/python3.10/site-packages/websockets/legacy/client.py", line 650, in __await_impl_timeout__
    return await asyncio.wait_for(self.__await_impl__(), self.open_timeout)
  File "/usr/lib64/python3.10/asyncio/tasks.py", line 445, in wait_for
    return fut.result()
  File "/home/nirbheek/.local/lib/python3.10/site-packages/websockets/legacy/client.py", line 654, in __await_impl__
    transport, protocol = await self._create_connection()
  File "/usr/lib64/python3.10/asyncio/base_events.py", line 1080, in create_connection
    transport, protocol = await self._create_connection_transport(
  File "/usr/lib64/python3.10/asyncio/base_events.py", line 1110, in _create_connection_transport
    await waiter
  File "/usr/lib64/python3.10/asyncio/sslproto.py", line 631, in _on_handshake_complete
    raise handshake_exc
  File "/usr/lib64/python3.10/asyncio/sslproto.py", line 676, in _process_write_backlog
    ssldata = self._sslpipe.do_handshake(
  File "/usr/lib64/python3.10/asyncio/sslproto.py", line 116, in do_handshake
    self._sslobj = self._context.wrap_bio(
  File "/usr/lib64/python3.10/ssl.py", line 526, in wrap_bio
    return self.sslobject_class._create(
  File "/usr/lib64/python3.10/ssl.py", line 865, in _create
    sslobj = context._wrap_bio(
ssl.SSLError: Cannot create a client socket with a PROTOCOL_TLS_SERVER context (_ssl.c:801)
```

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1821>

webrtc_sendrecv.py: Fix deprecation warning with Python 3.10

asyncio.get_event_loop() will not implicitly create a new event loop
in a future version of Python, so we need to do that explicitly.

```
webrtc_sendrecv.py:188: DeprecationWarning: There is no current event loop
  loop = asyncio.get_event_loop()
```

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1821>

webrtc_sendrecv.py: Fix styling errors

These are now enforced by the pre-commit python style hook.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1821>

webrtc: Update Makefile for building webrtc-sendrecv

This now needs the RTP library.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1821>

uridecodebin3: Remove dead variables

Leftover junk from original port

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1820>

matroska-mux: If a stream has a TITLE tag, use it for the name.

If a title tag is pushed to a pad, store it as the Track name.
This means that players will use it as the human readable
description of the track, instead of something generic like 'Video'
or 'Subtitle'

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1798>

matroskademux: Don't parse Tracks element twice

If the tracks element was parsed from the SeekEntry, don't
parse it a second time and recreate tracks, as this
loses any tags that were read using the seek table.

If a genuinely new Tracks element is found, do read that
as it is needed for MSE support.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1798>

meson: Adjust ninja usage restriction info for devenv

The documentation within meson highlights, that `ninja` can be used
without problems for any case except `ninja reconfigure`. The warning
however describes to the user, that `ninja` shouldn't be used at all.
Fix the warning.

Fixes: 01eb252f678 Don't allow people to run meson inside the uninstalled env
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1743>

docs: Fix typos in documentation

In building-from-source-using-meson.md:
s/implicitely/implicitly/

In README.md:
s/uncompatible/incompatible/

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1743>

docs: Extend documentation for the GStreamer development environment

Add more extensive documentation for the development environment.
Document how the tool works, how to use it and common use cases.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1743>

Remove the uninstalled term

Remove the symbolic link `gst-uninstalled` which points to `gst-env`.
The `uninstalled` is the old name and the project should stick to a
single name for the procedure.
Remove the term from all the files, exceptions are variables from
dependencies like `uninstalled_variables` from pkgconfig and
`meson-uninstalled`.
Adjust mentions of the script in the documentation and README.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1743>

Allow running gst-uninstalled when GStreamer is a subproject

As described in Merge request 222, the previous solution is not the
best possible solution and was also missing documentation.
Adjust the suggestion to the current GStreamer mono-repository.
And apply this change after reverting the previous commit.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1743>

tests: Remove deleted gst-env option from test

The `gstbuilddir` option has been removed by reverting the commit
879126a3. But the later commit 1babccfe503 uses the option within a test.
Remove the mentioned option from the test.

Fixes: 1babccfe503  Fix using overrides when not building PyGObject
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1743>

Revert "Allow running gst-uninstalled when gst-build is a subproject"

This reverts commit 879126a31c77622f594e70f205fcefd4556d9865.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1743>

Maintain build instructions at a single location

Do not maintain similar build instructions within each gst-plugins-*
subproject and the subproject/gstreamer subproject. Use the build
instructions from the mono-repository and link to them via hyperlink.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1743>

Add documentation for GST_VALIDATE_APPS_DIR

Add documentation for the environment variable, explaining what it is
used for and the default search locations.

Fixes: 4d569b51ed2 add a way to specify an application directory.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1743>

Improve environment variable documentation

At GST_VALIDATE_FILE:
s/will be outputed/are output/

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1743>

Add documentation for GST_VALIDATE_PLUGIN_PATH

Add documentation for the environment variable, explaining what it is
used for and the default search locations.

Fixes: 83d6978f80 Implement fault_injection as a Gs(tValidate)Plugin
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1743>

gst-env: Improve coding style and cleanup

* Remove unused variables
* Remove unused imports
* Apply pycodestyle style suggestions
  - Missing newlines
  - spaces before brackets
  - Wrong indentations

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1743>

docs: Add vah264enc metadata.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1051>

va: enable the H264 encoder.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1051>

va: Add H264 encoder.

This a new VA-API implementation of a H264 encoder.

It can control the GOP and parameter settings, while the MV searching,
VCL and the rate control algorithm are implemented by VA drivers and HW.
It supports most of the common usage options in H264, but still lacks
of look ahead, field, B frame weighted prediction, etc.

Co-authored-by: Victor Jaquez <vjaquez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1051>

va: Add a common encoder object.

As the counterpart of the va decoder, this class handles all the
common logic for the encoding routine and miscellaneous queries about
encoding.

Co-authored-by: Victor Jaquez <vjaquez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1051>

va: Add vacompat.h to wrap glib functions.

The g_queue_clear_full() and g_array_copy() functions in the glib
may not be available for the current glib version check, so we add
helper functions to wrap it.
This should be deleted after the glib version bumps.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1051>

va: Add the profile string name into the profile_map.

We also add a helper function of gst_va_profile_from_name to get
the VA profile value by its profile string name.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1051>

va: Change the H264 profile string order in the profile_map.

The first one should be the one that matches the VA profile's name
most precisely.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1051>

va: caps: Expose gst_va_create_coded_caps as helper function.

And allow free indentation for array declaration.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1051>

test: Add test cases for the H264 bitwriter.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1051>

codecparsers: bitwriter: Add the common bit writer functions for H264.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1051>

codecparsers: nalutils: Add nal_writer_reset_and_get_data help function.

We not only want to create a NAL gstmemory, but also need to create and
get the raw data of a NAL writer for the later usage.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1051>