This is a requirement for GstPlayer when using the default overlay interface
provided by the pipeline. The GstPlayerWrappedVideoRenderer requires a valid
pipeline, but that's available only after the GstPlay thread has successfully
started.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1345>
ShowWindow() could be blocked while doing gst_d3d11_window_win32_unprepare
when external window handle provided to d3d11videosink in multi-threaded
environment.
The condition that issue happened is, UI thread is waiting for a
background thread that changes d3d11videosink state to NULL, and the
background thread would try to send a window message to the queue.
The queue is already occupied by the UI thread, so the background
thread will be blocked.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1366>
Make sure the EGLImage we're rendering to the GL memory stays alive long enough,
until the the GL memory has been destroyed.
This change fixes tearing and black flashes artefacts that were happening
because the EGLImage was sometimes destroyed before the sink actually rendered
the associated texture.
Fixes#889
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1354>
The average_period should always represent the time between two
events. The specification defines the event time as the time
between audio samples, video frame sync, video line sync, etc.
In case of one timestamp per PDU the timestamp_interval identifies
the amount of events between the timestamp of one PDU and the
timestamp of the next PDU.
As described in IEEE 1722-2016 chapter
"10.4.12 timestamp_interval field" timestamp_interval shall be
nonzero.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1076>
Upstream caps might for example be
application/x-rtp,media=audio,encoding-name={OPUS, X-GST-OPUS-DRAFT-SPITTKA-00, multiopus}
and while that is not fixed caps it is enough to match it with a media.
Only caps structures that have the correct structure name and that have
the media and encoding-name field are preserved, but if both are present
then these caps are used as "codec preferences".
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1291>
Some decoding APIs support delayed output for performance reasons.
One example would be to request decoding for multiple frames and
then query for the oldest frame in the output queue.
This also increases throughput for transcoding and improves seek
performance when supported by the underlying backend.
Introduce support in the mpeg2 base class, so that backends that
support render delays can actually implement it.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1013>
Downstream might need the start code offset when decoding.
Previously this computation would be scattered in multiple sites. This
is error prone, so move it to the base class. Subclasses can access
slice->sc_offset directly without computing the address themselves
knowing that the size will also take the start code into account.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1013>
The GstV4l2CodecAllocator dispose function clears `self->decoder` but
the finalize function then tries to use it if the allocator has no been
detached yet.
Fix by detaching in the dispose function before we clear
`self->decoder`.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1220>
Standard interlace handling:
* If we have interlace-mode=interleaved and the field order, we just
set it when creating the session
* If we have interlace-mode=(interleaved|mixed) and no field order, we
set the field order on the first buffer
The encoder session does not support changing the FieldDetail after it
has started encoding frames, so we cannot support mixed streams
correctly.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1214>
Delay decoders downstream negotiation just before an output frame
needs to be allocated.
This is required, are least for H.264 and H.265 decoders, since
codec_data might trigger a new sequence before finishing upstream
negotiation, and sink pad caps need to set before setting source pad
caps, particularly to forward HDR fields. The other decoders are
changed too in order to keep the same structure among them.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1257>
Using GstBaseDec hack to access the parent_object of each element in
the element itself is a bit fragile. It would be better to keep its
own parent object as the usual global variable. It would make it
resistant to code changes.
The GstBaseDec macro to access the parent object now it's internal to
base decoder.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1257>
.. if a current direction has already been set
When `webrtcbin` has created an offer based on codec_preferences,
it might not have received caps on its sinkpads by the time a
remote description is set, in which case we want to connect the
input stream upon actual reception of the caps instead.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1233>
With mpeg4videoparse drop=false config-interval=N|-1 we might be
trying to insert a config before we have actually received one,
in which case we'll try to map a NULL buffer which will generate
lots of criticals.
Fixes#855
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1265>
gst_va_fixate_format() will iterate all othercaps' structures to find
the one with less information lost at color conversion. If a structure
with same color format is found, the iteration stops. It's like a
smart truncation. Then, this function also will choose the caps
feature.
Later this structure is used fixate its size and no further truncation
is needed.
Don't intersect at fixate, since it kills possible resizing.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1261>
Detected while reading the code, cccombiner must set
self->current_video_buffer to NULL *after* emitting selected-samples
in order for the application to get a useful return when peeking
the next video sample.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1252>
When schedule is true (as is the case by default), we insert padding
when no caption data is present in the schedule queue, and previously
weren't checking whether the caption pad had gone EOS, leading to
infinite scheduling of padding after EOS on the caption pad.
Rectify that by adding a "drain" parameter to dequeue_caption()
In addition, update the captions_and_eos test to push valid cc_data
in: without this cccombiner was attaching padding buffers it had
generated itself, and with that patch would now stop attaching
said padding to the second buffer. By pushing valid, non-padding
cc_data we ensure a caption buffer is indeed attached to the first
and second video buffers.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1252>
At gst_va_dmabuf_allocator_setup_buffer_full, static code analysis tool
does not know number of objects in descriptor is always larger than 0 if
export_surface_to_dmabuf succeeds. Thus, the tool will assume buf is
allocated with mem but not released when desc.num_objects equals to 0
and raise a mem leak issue.
For gst_va_dambuf_memories_setup, we should also inform the tool that
n_planes will be larger than 0 by checking the value at very beginning.
Then, the defect similar to above will not be raised during static analysis.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1241>
This will only affect individual/tarball module builds, as the
options yield to the parent project which was set to gpl=disabled
by default already. We kept it as auto in the original commit
to accommodate the need to update cerbero as well, which had to
be done separately after the initial commit.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1217>
Previously gst_structure_has_name was used to get a string to compare with supported mimetypes.
This is incorrect as above function returns a user defined structure name which is
not the structure mimetype value.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1206>
Trying to reset before the pads have been deactivated races with the
streaming thread. There was also a buggy buffer clear leaving a dangling
`stored_frame` pointer around. Use `gst_interlace_reset` so this happens
properly.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1039>
We don't support D3D11 interop for UWP because some APIs
(specifically MFTEnum2) are desktop application only.
However, the code for symbol loading is commonly used by both UWP and WIN32.
Just link GModule unconditionally which is UWP compatible, and simply don't
try to load any library/symbol dynamically when D3D11 interop is unavailable.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1216>
... in favour of dep.get_variable('foo', ..) which in some
cases allows for further cleanups in future since we can
extract variables from pkg-config dependencies as well as
internal dependencies using this mechanism.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1183>
When ass hinting value is set to anything other than NONE,
subtitles cannot use smooth scaling, thus all animations will jitter.
The libass author warns about possibility of breaking some scripts when it is enabled,
so lets do what is recommended and disable it to get the smooth scaling working.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1201>
The code in the aes elements assumes OpenSSL >= 1.1.0:
- implicit library initialization;
- version retrieved with OpenSSL_version(OPENSSL_VERSION);
and it fails to build with older versions.
Specify the required OpenSSL version explicitly in meson.build so that
the elements are excluded on older systems (e.g. Ubuntu 16.04) and the
rest of GStreamer can still build.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1067>
gst_va_vpp_complete_caps_features() now receives the @feature_name to
add and return if @caps doesn't provide it.
So, instead of two nested loops, now the function is a single loop,
traversing @caps to find if each structure already contains the requested
@features_name.
It's important to add missing caps features with @caps, in order to
not lost information.
The function caller does the external loop by calling per each
available caps feature.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1024>
In order to make more readable the caps transformation, the operation
was split in two phases:
1. Rangify the supported caps structures.
2. Add the missing (and supported) caps features.
Step 1 modified its logic, by copying any unrecognized structure.
It's a previous step required for allowing ANY caps feature as
passthrough.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1024>
For monorepo build and ugly/bad, for advanced feature
option API like get_option('xyz').required(..) which
we use in combination with the 'gpl' option.
For rest of modules for consistency (people will likely
use newer features based on the top-level requirement).
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1084>
There are streams in the wild that have to add a SCTE-35 trigger in
another e.g. GA94 stream. Most encoders would replace the GA94
descriptor ID with the CUEI one temporarily, but there are some that
will add two registration ID descriptors, one with GA94 and one with
CUEI.
Failing to parse the CUEI registration ID in that case would return
FALSE in _stream_is_private_section , therefore setting it as known PES
and pushing packets downstream instead of calling handle_psi.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/979>
We should also take into account whether data is currently pending when checking
for gap on streams. It could very well be that some streams have very low
bitrate (and spread out) data. For those we don't want to push out a gap event.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1179>
This is only enabled in push time mode. Furthermore it's only enabled for now if
PCR is to be ignored.
The problem is dealing with streams where the initial PTS/DTS observation might
be greater than following ones (from other PID for example). Before this patch,
this would result in sending buffers without any timestamp which would cause a
wide variety of issues.
Instead, pad segment and buffer timestamps with an extra
value (packetizer->extra_shift, default to 2s), to ensure that we can get valid
timestamps on outgoing buffers (even if that means they are before the segment
start).
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1179>
Introduces a `libraries` variable that contains all libraries in a
list with the following format:
``` meson
libraries = [
[pkg_name, {
'lib': library_object
'gir': [ {full gir definition in a dict } ]
],
....
]
```
It therefore refactors the way we build the gir so that we can reuse the
same information to build them against 'gstreamer-full' in gst-build
when linking statically
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1093>
Making a null check in gst_va_decode_picture_free () indicates pic->buffers or pic->slices
can be null, then in _destroy_buffers () the pointers are dereferenced, which is detected
as dereference after null check by Coverity. Thus, modify the code to do null check in
_detroy_buffers ().
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1143>
The first approach to fixate was simply a copy&paste of both
videoconvert and videoscale, trying to keep their logic as isolated
as possible. But that brought duplicated and sparse logic.
This patch merge both approaches simplifying the fixate operation.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1109>
Add a helper function to get, from GstVideoInfo and GstBuffers flags,
the VA interlace surface flags. This is used currently by vainterlace
element, but it will be used in vapostproc too if it can process
interlaced frames.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1109>
This new class is a helper for fast/tricky copy of surfaces. First it
tries to copy using the function vaCopy in libva 1.12. If it fails, or
it's not available, a GstVaFilter is tried to be instantiated with the
allocator's parameters, and if succeed, it's used for copying the
source surface.
This is required for dmabuf surfaces with drm modifier.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1023>
Initially GstVaSample processed its GstBuffer member to get the
VASurfaceID. But it might cases where we already have the VASurfaceID
to process by the filter.
This patch enables the possibility to pass the surfaces rather than
the buffers. In order to validate the surfaces a function to check
surfaces were added.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1023>
... instead of index of DXGI adapter.
The order of IDXGIAdapter1 enumerated via IDXGIFactory1::EnumAdapters1
can be varying even there's no rebooting in case that GPU preference order
is updated by user (for example, it can be done by using NVIDIA Control Panel
in case of multi-GPU laptop system) and eGPU is another possible case.
So, for an element which requires fixed target GPU requirement,
index based device enumeration is unreliable.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1098>
* gst_d3d11_device_new_for_adapter_luid()
Used for creating D3D11 device for a DXGI adapter (i.e., GPU)
corresponding to a LUID (Locally Unique Identifier).
This method can be useful for interop with other APIs such as
Direct3D12, MediaFoundation, CUDA, etc.
* gst_d3d11_device_new_wrapped()
Allows creating a new GstD3D11Device object by using already
configured ID3D11Device. This is conceptually equivalent to
gst_gl_context_new_wrapped()
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1098>
The GST_VIDEO_DECODER_ERROR() should be used only for robust/error-resilient
decoding purpose. Any other error codes such as not-negotiated or flushing
should be returned without modified for upstream to be able to handle
it immediately. (for example, application might want to try other
decoder element on not-negotiated)
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1070>
tests/check/meson.build uses the openjpeg_dep variable
unconditionally, and the subdir_done() is useless anyway, since the
plugin is only built if openjpeg_dep.found() is true. Fixes:
..\tests\check\meson.build:23:0: ERROR: Unknown variable "openjpeg_dep".
In particular, this fixes the build on UWP since we disable openjpeg
explicitly in Cerbero when building for UWP.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1069>
boolean return value is not sufficient for representing the reason
of error in most cases. For instance, any errors around new_sequence()
would mean negotiation error, not just *ERROR*.
And some subclasses will allocate buffer/memory/surface on new_picture()
but it could be failed because of expected error, likely flushing
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1019>
boolean return value is not sufficient for representing the reason
of error in most cases. For instance, any errors around new_sequence()
would mean negotiation error, not just *ERROR*.
And some subclasses will allocate buffer/memory/surface on new_picture()
but it could be failed because of expected error, likely flushing
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1019>
boolean return value is not sufficient for representing the reason
of error in most cases. For instance, any errors around new_sequence()
would mean negotiation error, not just *ERROR*.
And some subclasses will allocate buffer/memory/surface on new_picture()
but it could be failed because of expected error, likely flushing
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1019>
warning C4003: not enough arguments for function-like macro invocation 'warning'
G_STMT_END macro is extended to the below form with MSVC
__pragma(warning(push)) \
__pragma(warning(disable:4127)) \
while(0) \
__pragma(warning(pop))
So MSVC preprocessor will extend it further to
__pragma(VBI_CAT_LEVEL_LOG(push)) ...
Should rename warning() debug macro function therefore.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1018>
libgudev is a problematic dependency, particularly in sandboxed
environments, such as flatpak.
This patch implements a way to get the available VA devices using
brute-forced traverse of /dev/drm/renderD* directory. Thus usable in
those sandboxed environments.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1027>
When move the libgstva, libgudev dependency was moved as part of the
library, though it's not use by the library but the plugin. This patch
moves back libgudev dependency to the plugin.
Also HAVE_LIBDRM is move to the library which is the one who use it.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1027>
Some decoding APIs support delayed output for performance reasons.
One example would be to request decoding for multiple frames and
then query for the oldest frame in the output queue.
This also increases throughput for transcoding and improves seek
performance when supported by the underlying backend.
Introduce support in the vp9 base class, so that backends that
support render delays can actually implement it.
Co-authored by Seungha Yang <seungha@centricular.com>
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/987>
This patch contains two updates:
1. Instead of checking for dependency already checked just to verify a
version, we use the dependency version API.
2. Update the deprecated function get_pkgconfig_variable.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/997>
It's possible to have installed MediaSDK environment
package (libmfx-dev in Debian) without libva environment package. This
setup will lead to a breakage of meson configuration.
The fix is to get the libva's driver directory variable after the
dependency is validated as found.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/998>
When using the following setup (the error can be reproduced using
simpler sender pipelines), the receiver resynchronises the clock on RTCP
packets. The effect was that a couple seconds were cut out of the
playback because an initial RTCP packet was dropped.
When sending out all RTCP packets (setting sync=FALSE on the RTCP
updsink), the playback is fine.
This syncs rtpsink with rtpsrc (where this property was already set).
gst-launch-1.0 filesrc location=899-en.mp3 \
! mpegaudioparse \
! mpg123audiodec \
! audioconvert \
! audioresample \
! avenc_g722 \
! rtpg722pay
! rtpsink uri=rtp://239.1.2.3:1234
gst-launch-1.0 uridecodebin rtp://239.1.2.3:1234?encoding-name=G722 \
! autoaudiosink
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/993>
When there are elements between the demuxer and the muxer that
introduce an offset to the running time, or when offsets are
set on pads by the application, this shift must be taken into
account when calculating the final pts_adjustement.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/913>
mpegtsmux can receive SCTE sections from two origins: events
created by the application, and events forwarded downstream by
mpegtsdemux, containing sections that may not have been fully
parsed, and additional data to help tsmux translate times to
the correct domain, both for requesting keyframes and calculating
an accurate pts_adjustment.
The complete approach is documented further in a comment above
the relevant function.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/913>
Instead of modifying the splice times in the incoming sections
to running time and expecting eg mpegtsmux to convert those back
to its local PES time domain, which might be impossible when
those splice times are encrypted or the specification is extended,
transmit the needed information to the muxer as separate fields in
the event:
* A pts offset field can be used by the muxer in order to calculate
a final pts_adjustment
* A rtime_map can be used by the muxer to determine the correct
running times at which it should request keyframes
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/913>
Makes it possible to support passing SCTE 35 cue points from
demuxer to muxer, while preserving correct timing.
This will also improve ex nihilo cue points injection, as splice
times and durations are now interpreted as running time values,
and may trigger key unit requests.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/913>
Main differences with previous setup are:
- No manifest creation
- gst-indent is executed only when the bot is assigned (instead of the manifest task)
- Cerbero jobs are triggered in the cerbero repo
- Remove cerbero and android related files as they now are in cerbero
itself.
- Update `container.ps1` to the new file layout
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/891>
