platform/upstream/gstreamer.git
3 years agod3d11memory: Protect view object with lock
Seungha Yang [Sat, 24 Oct 2020 11:59:55 +0000 (20:59 +0900)]
d3d11memory: Protect view object with lock

Make resource allocation more thread-safe

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1718>

3 years agod3d11convert: Add support for conversion using ID3D11VideoProcessor
Seungha Yang [Fri, 23 Oct 2020 17:47:22 +0000 (02:47 +0900)]
d3d11convert: Add support for conversion using ID3D11VideoProcessor

Output texture of d3d11 decoder cannot have the bind flag
D3D11_BIND_SHADER_RESOURCE (meaning that it cannot be used for shader
input resource). So d3d11convert (and it's subclasses) was copying
texture into another internal texture to use d3d11 shader.
It's obviously overhead and we can avoid texture copy for
colorspace conversion or resizing via ID3D11VideoProcessor
as it supports decoder output texture.
This commit would be a visible optimization for d3d11 decoder with
d3d11compositor use case because we can avoid texture copy per frame.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1718>

3 years agod3d11memory: Store ID3D11VideoProcessorOutputView object
Seungha Yang [Fri, 23 Oct 2020 17:33:29 +0000 (02:33 +0900)]
d3d11memory: Store ID3D11VideoProcessorOutputView object

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1718>

3 years agocodecs: h265decoder: Fix picture leaks because of reference set.
He Junyan [Fri, 23 Oct 2020 14:29:57 +0000 (22:29 +0800)]
codecs: h265decoder: Fix picture leaks because of reference set.

The last frame's reference set has no one to cleanup. We need to
clean all pictures in the stop() func.
We also add a helper function to cleanup all the pictures in the
reference picture set.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1717>

3 years agocodecs: h265decoder: Fix 3 ref array leaks in finalize.
He Junyan [Fri, 23 Oct 2020 13:21:05 +0000 (21:21 +0800)]
codecs: h265decoder: Fix 3 ref array leaks in finalize.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1717>

3 years agod3d11window: Reuse ID3D11VideoProcessorInputView if possible
Seungha Yang [Fri, 23 Oct 2020 07:59:00 +0000 (16:59 +0900)]
d3d11window: Reuse ID3D11VideoProcessorInputView if possible

GstMemory object could be disposed if GstBuffer is not allocated
by GstD3D11BufferPool such as via gst_buffer_copy() and/or
gst_buffer_make_writable(). So attaching qdata on GstMemory
object would cause unnecessary view alloc/free.
By using view pool which is implemented in GstD3D11Allocator,
we can avoid redundant view alloc/free.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1716>

3 years agod3d11memory: Implement ID3D11VideoProcessorInputView pool
Seungha Yang [Wed, 21 Oct 2020 07:28:11 +0000 (16:28 +0900)]
d3d11memory: Implement ID3D11VideoProcessorInputView pool

Similar to ID3D11VideoDecoderOutputView pool implementation

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1716>

3 years agoMeson: Use pkg-config generator
Xavier Claessens [Sat, 7 Apr 2018 20:33:47 +0000 (16:33 -0400)]
Meson: Use pkg-config generator

3 years agoh265decoder: Add support for l0/l1
Nicolas Dufresne [Fri, 31 Jul 2020 00:23:37 +0000 (20:23 -0400)]
h265decoder: Add support for l0/l1

Add support for reference list needed for VA-API and some V4L2 decoders.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1713>

3 years agoh265decoder: Sync with the H264 implementation
Nicolas Dufresne [Tue, 28 Jul 2020 22:37:38 +0000 (18:37 -0400)]
h265decoder: Sync with the H264 implementation

This ensures that we get the last reference to picture being outputed,
avoiding GstBuffer structure copies and simplifying the buffer management.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1713>

3 years agod3d11decoder: Directly access ID3D11VideoDecoderOutputView for decoding
Seungha Yang [Tue, 20 Oct 2020 08:31:17 +0000 (17:31 +0900)]
d3d11decoder: Directly access ID3D11VideoDecoderOutputView for decoding

Decoder output view is stored in GstD3D11Memory object instead of
wrapper struct now. So qdata is no more required.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1712>

3 years agod3d11memory: Implement ID3D11VideoDecoderOutputView pool
Seungha Yang [Mon, 19 Oct 2020 16:59:35 +0000 (01:59 +0900)]
d3d11memory: Implement ID3D11VideoDecoderOutputView pool

Similar to texture-array pool, we can reuse decoder output view
since the life time of output view is identical to that of texture-array.
In this way, we can avoid frequent output view alloc/free.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1712>

3 years agod3d11memory: Move to GArray to store texture-array status
Seungha Yang [Sun, 4 Oct 2020 14:39:05 +0000 (23:39 +0900)]
d3d11memory: Move to GArray to store texture-array status

The size D3D11_REQ_TEXTURE2D_ARRAY_AXIS_DIMENSION is 2048
which is too large in practice especially for a texture
of dpb

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1712>

3 years agova: Add VP9 decoder
Víctor Manuel Jáquez Leal [Mon, 12 Oct 2020 17:20:10 +0000 (19:20 +0200)]
va: Add VP9 decoder

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1702>

3 years agowpe: Convert launch lines to markdown and move since tag
Philippe Normand [Fri, 16 Oct 2020 14:46:20 +0000 (15:46 +0100)]
wpe: Convert launch lines to markdown and move since tag

Seems like the examples don't appear in the generated docs because the Since tag
was badly positioned in the doc blurb.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1706>

3 years agocodecs: vp9decoder: Add segmentation to picture.
Víctor Manuel Jáquez Leal [Fri, 16 Oct 2020 08:35:36 +0000 (10:35 +0200)]
codecs: vp9decoder: Add segmentation to picture.

VA-API needs AC and DC quant scales for both luma and chroma, and the loop
filter level for current frame, but these values are not available outside
the private GstVp9Parser structure. And these values may change from frame
to frame, so they are picture specific.

This patch add GstVp9Segmentation structure array to GstVp9Picture to expose
it to derived classes. This approach is safer than passing the parser at
picture handling flow.

Also, this patch, in order to solve Documentation CI, mark as private the
GstVp9Picture structure.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1700>

3 years agocodecs: vp9decoder: Pass parser as new_sequence() parameter.
Víctor Manuel Jáquez Leal [Mon, 12 Oct 2020 09:07:47 +0000 (11:07 +0200)]
codecs: vp9decoder: Pass parser as new_sequence() parameter.

In order to know the chroma format, besides profile, subsampling_x and
subsampling_y are needed (Spec 7.2.2 Color config semantics). These values are
in GstVp9Parser but not in  GstVp9Framehdr.

Also, bit_depth is available in parser but not frame header. Evenmore, those
values are copied to picture structure later.

In case of VA-API, to configure the pipeline, it is require to know the chroma
format and depth.

It is possible to know chroma and depth through caps coming from vp9parser, but
it requires string parsing. It would be less error prone to get these values
through the parser structure at new_sequence() virtual method.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1700>

3 years agortpsrc: Cleanup on BYE, timeout or when pad is reused
Nicolas Dufresne [Wed, 23 Sep 2020 20:43:30 +0000 (16:43 -0400)]
rtpsrc: Cleanup on BYE, timeout or when pad is reused

In this patch, we enabled 'autoremove' feature of rtpbin and also call
'clear-ssrc' on the rtpssrcdemux element when a pad is being reused. This
ensure that the jitterbuffer is removed and no threads accumulates.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1575>

3 years agortpsrc: re-use the same src pad for streams that have the same payload type
George Kiagiadakis [Fri, 4 Sep 2020 11:18:13 +0000 (14:18 +0300)]
rtpsrc: re-use the same src pad for streams that have the same payload type

Also use payload type when naming pads, this will make it easier to identify
pads and simplify the code.

Fixes #1395

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1575>

3 years agod3d11: Introduce d3d11compositor element
Seungha Yang [Tue, 2 Jun 2020 16:26:12 +0000 (01:26 +0900)]
d3d11: Introduce d3d11compositor element

Add new video composition element which is equivalent to compositor
and glvideomixer elements. When d3d11 decoder elements are used,
d3d11compositor can do efficient graphics memory handling
(zero copying or at least copying memory on GPU memory space).

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1323>

3 years agod3d11shader: Allow drawing without shader resource view
Seungha Yang [Wed, 5 Aug 2020 08:27:30 +0000 (17:27 +0900)]
d3d11shader: Allow drawing without shader resource view

... for the case that we are rendering on target without input texture.
For example, we might want to draw arbitrary shape on render target view
without shader resource view.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1323>

3 years agod3d11convert: Add new subclasses for only color convert or resize
Seungha Yang [Sun, 2 Aug 2020 18:19:34 +0000 (03:19 +0900)]
d3d11convert: Add new subclasses for only color convert or resize

New d3d11colorconvert and d3d11scale elements will perform only
colorspace conversion and rescale, respectively. Those new elements
would be useful when only colorspace conversion or rescale is required
and the other part should be done by another elements.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1323>

3 years agod3d11colorconverter: Allow setting alpha value to use
Seungha Yang [Sat, 1 Aug 2020 11:16:52 +0000 (20:16 +0900)]
d3d11colorconverter: Allow setting alpha value to use

... used for reordering case for now. In other words, non-alpha formats
such as NV12 is not supported case yet.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1323>

3 years agod3d11colorconverter: Add support conversion with blending
Seungha Yang [Tue, 2 Jun 2020 16:20:41 +0000 (01:20 +0900)]
d3d11colorconverter: Add support conversion with blending

This is pre-work for d3d11compositor support

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1323>

3 years agod3d11colorconverter: Add method to support updating destination rect
Seungha Yang [Tue, 2 Jun 2020 15:59:15 +0000 (00:59 +0900)]
d3d11colorconverter: Add method to support updating destination rect

It's equivalent to GST_VIDEO_CONVERTER_OPT_DEST_* options of GstVideoConverter

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1323>

3 years agod3d11: Clarify target rect to be updated
Seungha Yang [Tue, 2 Jun 2020 15:46:13 +0000 (00:46 +0900)]
d3d11: Clarify target rect to be updated

Rename internal methods to clarify which rect (i.e., input or output)
should be updated

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1323>

3 years agonvcodec: Report latency in decoder based on max-display-delay
Julian Bouzas [Fri, 2 Oct 2020 14:02:38 +0000 (10:02 -0400)]
nvcodec: Report latency in decoder based on max-display-delay

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

3 years agonvcodec: Add max-display-delay decoder property
Julian Bouzas [Fri, 2 Oct 2020 13:22:34 +0000 (09:22 -0400)]
nvcodec: Add max-display-delay decoder property

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

3 years agonvcodec: Fix compiler error if OpenGL is not enabled
Julian Bouzas [Thu, 24 Sep 2020 15:25:33 +0000 (11:25 -0400)]
nvcodec: Fix compiler error if OpenGL is not enabled

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

3 years agonvcodec: Add missing CUDAMemory src caps in h264 decoder
Julian Bouzas [Thu, 24 Sep 2020 14:33:58 +0000 (10:33 -0400)]
nvcodec: Add missing CUDAMemory src caps in h264 decoder

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

3 years agonvcodec: Add missing CUDAMemory sink caps in h264 and h265 encoders
Julian Bouzas [Wed, 23 Sep 2020 17:49:43 +0000 (13:49 -0400)]
nvcodec: Add missing CUDAMemory sink caps in h264 and h265 encoders

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

3 years agonvcodec: Fix description of cudadownload element
Julian Bouzas [Tue, 22 Sep 2020 17:07:19 +0000 (13:07 -0400)]
nvcodec: Fix description of cudadownload element

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

3 years agotests: Add CUDA filter unit tests
Seungha Yang [Sun, 11 Aug 2019 06:02:04 +0000 (15:02 +0900)]
tests: Add CUDA filter unit tests

Adding a test for buffer meta and colorspace conversion

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

3 years agonvcodec: Add CUDA video scale element
Seungha Yang [Wed, 16 Oct 2019 13:43:09 +0000 (22:43 +0900)]
nvcodec: Add CUDA video scale element

Add new element for video resizing using CUDA

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

3 years agonvcodec: Add CUDA video convert element
Seungha Yang [Wed, 16 Oct 2019 13:42:55 +0000 (22:42 +0900)]
nvcodec: Add CUDA video convert element

Add new element for colorspace conversion using CUDA.

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

3 years agonvcodec: Add generic CUDA video convert object
Seungha Yang [Wed, 16 Oct 2019 13:42:39 +0000 (22:42 +0900)]
nvcodec: Add generic CUDA video convert object

Introducing generic video convert object similar to video-converter
but using CUDA.

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

3 years agonvcodec: Add support runtime CUDA kernel source compilation
Seungha Yang [Wed, 16 Oct 2019 13:42:24 +0000 (22:42 +0900)]
nvcodec: Add support runtime CUDA kernel source compilation

Add util functions for runtime CUDA kernel source compilation
using NVRTC library. Like other nvcodec dependent libraries,
NVRTC library will be loaded via g_module_open.

Note that the NVRTC library naming is not g_module_open friendly
on Windows.
(i.e., nvrtc64_{CUDA major version}{CUDA minor version}.dll).
So users can specify the dll name using GST_NVCODEC_NVRTC_LIBNAME
environment.

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

3 years agonvcodec: Add CUDA upload/download elements with base class for CUDA filters
Seungha Yang [Wed, 16 Oct 2019 13:42:06 +0000 (22:42 +0900)]
nvcodec: Add CUDA upload/download elements with base class for CUDA filters

Similar to glupload/gldownload elements but for CUDA memory.
It will help transfer memory between system and nvidia GPU

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

3 years agonvcodec: Peer direct access support
Seungha Yang [Wed, 16 Oct 2019 13:21:05 +0000 (22:21 +0900)]
nvcodec: Peer direct access support

If support direct access each other, use device to device memory copy
without staging host memory

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

3 years agocudacontext: Enable direct CUDA memory access over multiple GPUs
Seungha Yang [Fri, 30 Aug 2019 08:19:44 +0000 (17:19 +0900)]
cudacontext: Enable direct CUDA memory access over multiple GPUs

If each device context can access each other, enable peer access
for better interoperability.

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

3 years agonvenc: Support CUDA buffer pool
Seungha Yang [Fri, 30 Aug 2019 04:57:15 +0000 (13:57 +0900)]
nvenc: Support CUDA buffer pool

When upstream support CUDA memory (only nvdec for now), we will create
CUDA buffer pool.

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

3 years agonvdec: Support CUDA buffer pool
Seungha Yang [Fri, 30 Aug 2019 04:55:25 +0000 (13:55 +0900)]
nvdec: Support CUDA buffer pool

If downstream can accept CUDA memory caps feature (currently nvenc only),
always CUDA memory is preferred.

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

3 years agonvcodec: Add CUDA specific memory and bufferpool
Seungha Yang [Mon, 19 Aug 2019 09:02:56 +0000 (18:02 +0900)]
nvcodec: Add CUDA specific memory and bufferpool

Introducing CUDA buffer pool with generic CUDA memory support.
Likewise GL memory, any elements which are able to access CUDA device
memory directly can map this CUDA memory without upload/download
overhead via the "GST_MAP_CUDA" map flag.
Also usual GstMemory map/unmap is also possible with internal staging memory.

For staging, CUDA Host allocated memory is used (see CuMemAllocHost API).
The memory is allowing system access but has lower overhead
during GPU upload/download than normal system memory.

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

3 years agocodecparsers: h264parser: guard against ref_pic_markings overflow
Andrew Wesie [Fri, 16 Oct 2020 11:29:02 +0000 (12:29 +0100)]
codecparsers: h264parser: guard against ref_pic_markings overflow

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1703>

3 years agohlssink2: fix and flesh out docs
Tim-Philipp Müller [Thu, 15 Oct 2020 23:48:01 +0000 (00:48 +0100)]
hlssink2: fix and flesh out docs

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1699>

3 years agomeson: update glib minimum version to 2.56
Stéphane Cerveau [Thu, 15 Oct 2020 16:26:48 +0000 (18:26 +0200)]
meson: update glib minimum version to 2.56

In order to support the symbol g_enum_to_string in various
project using GStreamer ( gst-validate etc.), the glib minimum
version should be 2.56.0.

Remove compat code as glib requirement
is now > 2.56

Version used by Ubuntu 18.04 LTS

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1695>

3 years agod3d11: vp8dec: No need to check show_frame flag when output_picture.
He Junyan [Sun, 11 Oct 2020 17:04:13 +0000 (01:04 +0800)]
d3d11: vp8dec: No need to check show_frame flag when output_picture.

The VP8 base class has already handled it for us.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1670>

3 years agocodecs: vp8decoder: handle the show_frame check in base class.
He Junyan [Sun, 11 Oct 2020 16:57:24 +0000 (00:57 +0800)]
codecs: vp8decoder: handle the show_frame check in base class.

Move the show_frame check from sub class to vp8 decoder's base class.
Calling the sub class' output_picture() function only when the frame
is displayed and marking the other automatically as decode only.

This is done to avoid logic and code repetition in subclasses.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1670>

3 years agocodecs: vp9decoder: handle the show_frame check in base class
Seungha Yang [Thu, 15 Oct 2020 17:06:49 +0000 (02:06 +0900)]
codecs: vp9decoder: handle the show_frame check in base class

Same as vp8 decoder update https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1670

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1697>

3 years agov4l2codecs: vp8: Allow baseclass from skipping frames
Nicolas Dufresne [Thu, 15 Oct 2020 16:08:19 +0000 (12:08 -0400)]
v4l2codecs: vp8: Allow baseclass from skipping frames

In preparation for !1670, this will allow the base class from skipping frames
that should not be displayed. Previously it would complain about unordered
decoding taking place in the driver.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1696>

3 years agov4l2codecs: decoder: Unmark previously pending request
Nicolas Dufresne [Thu, 15 Oct 2020 16:05:45 +0000 (12:05 -0400)]
v4l2codecs: decoder: Unmark previously pending request

requests are executed in order, so while dequeuing sink buffers for previous
request, also mark these request as no longer pending. This will allow reusing
the request later.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1696>

3 years agov4l2codecs: decoder: Properly remove pending requests
Nicolas Dufresne [Thu, 15 Oct 2020 15:35:04 +0000 (11:35 -0400)]
v4l2codecs: decoder: Properly remove pending requests

Pass the pointer instead of NULL in order to find and remove properly any
pending request from the queue. This coding error was leading to use after
free in error and early exit cases.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1696>

3 years agova: basedec: Create the other pool anyway.
He Junyan [Wed, 14 Oct 2020 11:04:44 +0000 (19:04 +0800)]
va: basedec: Create the other pool anyway.

Fix a bug in _create_other_pool(). The old way of checking the
base->other_pool make that other_pool never be changed until the
gst_va_base_dec_stop() to stop the current decoding context.
But in some stream, the resolution may change during the decoding
process, and we need to re-negotiate the buffer pool. Then, the
old other_pool can not be clean correctly and the new correct one
can not be created.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1692>

3 years agova: basedec: Should unmap src frame when dst frame map failing.
He Junyan [Wed, 14 Oct 2020 08:54:54 +0000 (16:54 +0800)]
va: basedec: Should unmap src frame when dst frame map failing.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1692>

3 years agova: bufferpool: use release_buffer to clean the mem.
He Junyan [Tue, 13 Oct 2020 07:28:24 +0000 (15:28 +0800)]
va: bufferpool: use release_buffer to clean the mem.

The current bufferpool wastes all pre-allocate buffers when the
buffer pool is actived.
The pool->priv->size is 0 for va buffer pool. And every time, the
reset_buffer() will clean all mem and make the buffer size 0, that
can cache the gst_buffer in the buffer pool.
But when the buffer pool is activing, the default_start() just
allocate the buffer and release_buffer() immediately, all the pre
allocated buffers and surfaces are destroyed because of
gst_buffer_get_size (buffer) != pool->priv->size.
We need to use release_buffer() to do the clean job at the pool
start time.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1686>

3 years agoh265parse: Don't enable passthrough by default
Seungha Yang [Sat, 3 Oct 2020 17:02:16 +0000 (02:02 +0900)]
h265parse: Don't enable passthrough by default

SEI messages contain various information which wouldn't be conveyed
by using upstream CAPS (HDR, timecode for example).

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1639>

3 years agocameracalibrate: Improve gst-inspect documentation
Vivia Nikolaidou [Tue, 13 Oct 2020 10:11:06 +0000 (13:11 +0300)]
cameracalibrate: Improve gst-inspect documentation

Thanks to @kazz_naka on Twitter

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1691>

3 years agowpesrc: add some debug logging around WPEView creation/destruction
Matthew Waters [Wed, 7 Oct 2020 10:13:09 +0000 (21:13 +1100)]
wpesrc: add some debug logging around WPEView creation/destruction

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1663>

3 years agowpesrc: fix a memory leak of the bytes
Matthew Waters [Wed, 7 Oct 2020 10:14:55 +0000 (21:14 +1100)]
wpesrc: fix a memory leak of the bytes

free the previous GBytes if load-bytes is called multiple times
before view creation.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1663>

3 years agowpesrc: only create webview if not already created
Matthew Waters [Tue, 6 Oct 2020 11:19:21 +0000 (22:19 +1100)]
wpesrc: only create webview if not already created

e.g. _decide_allocation() can be called multiple times throughout the
element's lifetime and we only want to create the view once rather than
overwriting.

Fixes a leak of the WPEView under certain circumstances.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1663>

3 years agowpe: free a previous pending image/shm buffer
Matthew Waters [Fri, 2 Oct 2020 02:06:59 +0000 (12:06 +1000)]
wpe: free a previous pending image/shm buffer

Don't blindly overwrite a possibly previously set buffer.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1663>

3 years agosrtsrc: Prevent `delay` from being negative
Jan Alexander Steffens (heftig) [Mon, 12 Oct 2020 12:15:49 +0000 (14:15 +0200)]
srtsrc: Prevent `delay` from being negative

`delay` should be a GstClockTimeDiff since SRT time is int64_t.

All values are in local time so we should never see a srctime that's in
the future. If we do, clamp the delay to 0 and warn about it.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1674>

3 years agosrtsrc: Don't calculate a delay if the srctime is 0
Jan Alexander Steffens (heftig) [Mon, 12 Oct 2020 12:12:24 +0000 (14:12 +0200)]
srtsrc: Don't calculate a delay if the srctime is 0

A zero srctime is a missing srctime. Apparently this can happen when
["the connection is not between SRT peers or if Timestamp-Based Packet
Delivery mode (TSBPDMODE) is not enabled"][1] so it may not apply to us,
but it's best to be defensive.

[1]: https://github.com/Haivision/srt/blob/v1.4.2/docs/API.md#sending-and-receiving

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1674>

3 years agosrtsrc: Defend against missing clock
Jan Alexander Steffens (heftig) [Mon, 12 Oct 2020 12:09:28 +0000 (14:09 +0200)]
srtsrc: Defend against missing clock

If we don't have a clock, stop the source instead of asserting in
gst_clock_get_time. This can happen when the element is removed from the
pipeline while it's playing.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1674>

3 years agortpmanagerbad: remove duplicate parent declaration
Marc Leeman [Mon, 12 Oct 2020 11:56:50 +0000 (13:56 +0200)]
rtpmanagerbad: remove duplicate parent declaration

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1689>

3 years agortmp2sink: fix since marker on new "stop-commands" property
Tim-Philipp Müller [Mon, 12 Oct 2020 10:55:46 +0000 (11:55 +0100)]
rtmp2sink: fix since marker on new "stop-commands" property

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1687>

3 years agova: basedec: copy frames logic to decide_allocation()
Víctor Manuel Jáquez Leal [Fri, 9 Oct 2020 14:00:18 +0000 (16:00 +0200)]
va: basedec: copy frames logic to decide_allocation()

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1673>

3 years agova: basedec: refactor the other video pool instantiation
Víctor Manuel Jáquez Leal [Fri, 9 Oct 2020 13:47:43 +0000 (15:47 +0200)]
va: basedec: refactor the other video pool instantiation

Just a code clean up

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1673>

3 years agova: basedec: add gstvabasedec helper
Víctor Manuel Jáquez Leal [Thu, 8 Oct 2020 17:39:56 +0000 (19:39 +0200)]
va: basedec: add gstvabasedec helper

This is a helper for all decoders.

It is not an abstract subclass, just merely a helper that avoids code
duplication among the decoders.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1673>

3 years agova: vp8dec: add element documentation
Víctor Manuel Jáquez Leal [Fri, 9 Oct 2020 08:33:58 +0000 (10:33 +0200)]
va: vp8dec: add element documentation

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1673>

3 years agova: h264dec: set video alignment definition earlier
Víctor Manuel Jáquez Leal [Fri, 9 Oct 2020 10:27:12 +0000 (12:27 +0200)]
va: h264dec: set video alignment definition earlier

This patch renames need_cropping variable to need_videoalign which is clearer
with its function. And now GstVideoAlignment is part of GstVaH264Dec structure,
so it can be set earlier.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1673>

3 years agodtlsconnection: Ignore OpenSSL system call errors
Olivier Crête [Mon, 5 Oct 2020 20:40:55 +0000 (16:40 -0400)]
dtlsconnection: Ignore OpenSSL system call errors

OpenSSL shouldn't be making real system calls, so we can safely
ignore syscall errors. System interactions should happen through
our BIO. So especially don't look at the system's errno, as it
should be meaningless.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1656>

3 years agotests: svthevcenc: Fix test_encode_simple
Jan Alexander Steffens (heftig) [Sat, 12 Sep 2020 00:48:43 +0000 (02:48 +0200)]
tests: svthevcenc: Fix test_encode_simple

Pick the same I420 format the other test use. Without this, the source
picks AYUV64, which fails.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1573>

3 years agova: allocator: add _set_format() and _get_format()
Víctor Manuel Jáquez Leal [Wed, 7 Oct 2020 16:03:20 +0000 (18:03 +0200)]
va: allocator: add _set_format() and _get_format()

Since allocators keep an available memory queue to reuse, video format and usage
hint are now persistant while allocator's memories are around.

This patch adds _set_format() and _get_format() for both VA allocators.

_set_format() validates if given format can be used or reused. If no allocated
surface previously it creates a dummy one to fetch its offsets and
strides. Updated info is returned to callee.

GstVaPool uses _set_format() at config to verify the allocator capacity and to
get the surfaces offsets and strides, which are going to be used by the video
meta.

Allocator extracted caps are compared with caps from config and if they have
different strides or offsets, force_videometa is set.

A new bufferpool method gst_va_pool_requires_video_meta() is added return the
value of force_videometa. This value is checked in order to know if decoders
need to copy the surface if downstream doesn't announce video meta support.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1667>

3 years agova: caps: added gst_caps_is_raw()
Víctor Manuel Jáquez Leal [Thu, 8 Oct 2020 12:10:41 +0000 (14:10 +0200)]
va: caps: added gst_caps_is_raw()

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1667>

3 years agova: pool: call parent's start() method
Víctor Manuel Jáquez Leal [Thu, 8 Oct 2020 08:26:54 +0000 (10:26 +0200)]
va: pool: call parent's start() method

Without preallocating buffers and memories a deadlock in pool allocator is
highly probably since it might hit the case were buffer is returned to the pool
but their memories are still hold by a copy downstream, without other
preallocated buffers available.

This kind of a hack, where buffer_reset() follow the normal path if it's called
from start().

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1667>

3 years agova: pool: fix log's object
Víctor Manuel Jáquez Leal [Wed, 7 Oct 2020 14:18:30 +0000 (16:18 +0200)]
va: pool: fix log's object

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1667>

3 years agova: allocator: remove noisy log message
Víctor Manuel Jáquez Leal [Thu, 8 Oct 2020 08:30:28 +0000 (10:30 +0200)]
va: allocator: remove noisy log message

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1667>

3 years agova: allocator: add a todo for dmabuf_memories_setup()
Víctor Manuel Jáquez Leal [Wed, 7 Oct 2020 09:08:49 +0000 (11:08 +0200)]
va: allocator: add a todo for dmabuf_memories_setup()

It would be nice to add a surface pool for this type of surface allocation in
order to have a better control of them.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1667>

3 years agova: allocator: add a surface counter
Víctor Manuel Jáquez Leal [Wed, 7 Oct 2020 08:16:27 +0000 (10:16 +0200)]
va: allocator: add a surface counter

Every time a new surface is created the counter increases by one, and when it is
destroyed (or will be destroyed in case of GstVaAllocator), the counter is
decreased. Then, at allocator dispose, it is warning if the counter is not zero.

This counter will be also used to check if the allocator can change its
configuration if the counter is zero or can not.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1667>

3 years agova: allocator: remove GstVideoInfo from GstVaBufferSurface
Víctor Manuel Jáquez Leal [Tue, 6 Oct 2020 18:01:04 +0000 (20:01 +0200)]
va: allocator: remove GstVideoInfo from GstVaBufferSurface

Don't store it them anymore since it is related with the negotiated stream and
not the concrete buffer.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1667>

3 years agova: remove GstVideoInfo parameter from _get_surface() functions
Víctor Manuel Jáquez Leal [Tue, 6 Oct 2020 17:54:26 +0000 (19:54 +0200)]
va: remove GstVideoInfo parameter from _get_surface() functions

There shouldn't be need to retrieve GstVideoInfo per buffer or memory since it
is the same for all the negotiated stream.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1667>

3 years agova: vpp: don't fetch video info from buffer
Víctor Manuel Jáquez Leal [Tue, 6 Oct 2020 17:40:16 +0000 (19:40 +0200)]
va: vpp: don't fetch video info from buffer

Instead of fetching video info from the buffer, use the already set ones.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1667>

3 years agova: dec, vpp: don't get buffer size from allocators
Víctor Manuel Jáquez Leal [Wed, 7 Oct 2020 10:49:44 +0000 (12:49 +0200)]
va: dec, vpp: don't get buffer size from allocators

Since buffer size is now ignored by bufferpool there's no need to get tha value
from the allocator.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1667>

3 years agova: pool: ignore size in config
Víctor Manuel Jáquez Leal [Sun, 4 Oct 2020 09:14:38 +0000 (11:14 +0200)]
va: pool: ignore size in config

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1667>

3 years agoh265parse: set interlace-mode=interleaved on interlaced content
Guillaume Desmottes [Mon, 5 Oct 2020 09:07:25 +0000 (11:07 +0200)]
h265parse: set interlace-mode=interleaved on interlaced content

interlace-mode=alternate is a special case of interlace-mode=interleaved
where the fields are split using two different buffers.

We should use the latter instead of the former to no break compat with
elements supporting only 'interleaved'.
Decoders producing alternate, such as OMX on the Zynq, should change the
interlace-mode on their output caps.

Fix https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/issues/825

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1655>

3 years agoReplace LGPL v2 with LGPL v2.1 in COPYING and remove COPYING.LIB
Jacek Tomaszewski [Fri, 9 Oct 2020 08:24:50 +0000 (10:24 +0200)]
Replace LGPL v2 with LGPL v2.1 in COPYING and remove COPYING.LIB

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1631>

3 years agoReplace GPL v2 with LGPL v2 in COPYING file
Jacek Tomaszewski [Fri, 2 Oct 2020 09:42:07 +0000 (11:42 +0200)]
Replace GPL v2 with LGPL v2 in COPYING file

Fixes #1422
https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/issues/1422

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1631>

3 years agosrt: Consume the error from gst_srt_object_write
Jan Alexander Steffens (heftig) [Thu, 8 Oct 2020 15:52:05 +0000 (17:52 +0200)]
srt: Consume the error from gst_srt_object_write

Instead of leaking it.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1668>

3 years agosrt: Check socket state before retrieving payload size
Jan Alexander Steffens (heftig) [Thu, 8 Oct 2020 15:48:20 +0000 (17:48 +0200)]
srt: Check socket state before retrieving payload size

The connection might be broken, which we should detect instead of just
aborting the write.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1669>

3 years agox265enc: fix deadlock on reconfig
Jakub Adam [Thu, 8 Oct 2020 16:25:59 +0000 (18:25 +0200)]
x265enc: fix deadlock on reconfig

Don't attempt to obtain encoder lock that is already held by
gst_x265_enc_encode_frame().

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1671>

3 years agowebrtc: Require gstreamer-sdp in the pkg-config file
Sebastian Dröge [Wed, 7 Oct 2020 08:04:30 +0000 (11:04 +0300)]
webrtc: Require gstreamer-sdp in the pkg-config file

Some headers include API from it.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1660>

3 years agosrtsrc: Fix timestamping
Edward Hervey [Tue, 6 Oct 2020 09:45:36 +0000 (11:45 +0200)]
srtsrc: Fix timestamping

SRT provides the original timestamp of a packet (with drift/skew corrected for
local clock), which is what should be used for timestamping the outgoing
buffers. This ensures that we output the packets with the same timestamp (and by
extension rate) as the original feed.

Also detect if packets were dropped (by checking the sequence number) and
properly set DISCONT flag on the outgoing buffer.

Finally answer the latency queries

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1658>

3 years agomfvideosrc: Use only the first video stream per device
Seungha Yang [Tue, 6 Oct 2020 20:05:25 +0000 (05:05 +0900)]
mfvideosrc: Use only the first video stream per device

Non-first video stream might not be working with current
implementation. It could be non-video (e.g., photo source) and then
ReadSample() might be blocked forever.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1661>

3 years agodecklink: Update doc
Seungha Yang [Wed, 7 Oct 2020 18:15:21 +0000 (03:15 +0900)]
decklink: Update doc

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1665>

3 years agodecklink: Update Windows headers with SDK 11.2
Seungha Yang [Wed, 7 Oct 2020 16:39:42 +0000 (01:39 +0900)]
decklink: Update Windows headers with SDK 11.2

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1665>

3 years agodecklink: Update OSX headers with SDK 11.2
Seungha Yang [Wed, 7 Oct 2020 16:33:35 +0000 (01:33 +0900)]
decklink: Update OSX headers with SDK 11.2

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1665>

3 years agodecklink: Updated DeckLink SDK to 11.2 to support DeckLink 8K Pro
Tim [Tue, 25 Jun 2019 09:51:32 +0000 (11:51 +0200)]
decklink: Updated DeckLink SDK to 11.2 to support DeckLink 8K Pro

Updated Decklink SDK to version 11.2 in order to support newer cards like the Decklink 8K Pro.
This required to replace the duplex property by a profile property.

Profile values can be the following:
-  bmdProfileOneSubDeviceFullDuplex
-  bmdProfileOneSubDeviceHalfDuplex
-  bmdProfileTwoSubDevicesFullDuplex
-  bmdProfileTwoSubDevicesHalfDuplex
-  bmdProfileFourSubDevicesHalfDuplex

Fixes #987

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1665>

3 years agomfvideosrc: Fix invalid memory access when outputting jpeg
Seungha Yang [Wed, 7 Oct 2020 08:37:25 +0000 (17:37 +0900)]
mfvideosrc: Fix invalid memory access when outputting jpeg

Don't access unknown-dangerous-nonsense address

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1662>

3 years agoRevert "webrtc: Save the media kind in the transceiver"
Sebastian Dröge [Thu, 8 Oct 2020 15:50:12 +0000 (18:50 +0300)]
Revert "webrtc: Save the media kind in the transceiver"

This reverts commit f54d8e99457996303b8477b1f3a710f0fabd1cc6.

It breaks the CI until the C# bindings are fixed.