We should move this functionality to gst-libs so that GstD3D11Converter
can be moved to gst-libs.
Another advantage is that applications can call our
HLSL compiler wrapper method without any worry about OS version
dependent system installed HLSL library.
Note that there are multiple HLSL compiler library versions
on Windows and system installed one would be OS version dependent.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2760>
We need GStreamer elements to do the bandwidth estimation as this way
they can also control the pacing of the transmission flow as specified
in the [GCC] algorithm for example.
Bandwidth estimator element are placed right before the "RTPSession" as
an "rtp-aux-sender" element. This way they can use the "Transport-wide
Congestion Control" RTCP feedback messages through the "RTPTwcc" custom
events that are sent by the rtpsession.
Applications are responsible to react to the bandwidth estimator element
and set the encoder target bitrate etc... which means that we can not
pass an estimator as an element factory, so a signal as been chosen
instead.
[GCC]: https://datatracker.ietf.org/doc/html/draft-ietf-rmcat-gcc-02
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2562>
Raw memory upload should always be the least preferred input
caps, only added by the raw memory uploader as the last thing
in the caps.
Caps negotiation should still choose raw data when it needs to,
and other upload methods that can accept raw data buffers will still do so.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2725>
The AVClass name of Animated PNG in FFmpeg 5.x is "(A)PNG"
and it will be converted to "-a-png" through
g_ascii_strdown() and g_strcanon(). But GLib disallow leading '-'
character for a GType name. Strip leading '-' to workaround it.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2724>
There might be a sequence of event and buffer flow:
- Got stream-start/caps/segment events
- Got flush events
- And then buffers with a new segment event
In the above case, stream-start and caps event might not be reached to
peer proxysrc if peer proxysrc is not ready to receive them.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1552>
gst_video_convert_scale_get_fixed_format() receives 'othercaps' from
basetransforms' fixate_caps() vmethod which explicitly mentions that
'`othercaps` may not be writable'.
The gst_caps_intersect() call just before may or may not produce new
caps. Particularly in cases like EMPTY or ANY caps on either of the
inputs, only a ref is taken and returned to the caller.
As a result, gst_video_convert_scale_fixate_format() may have attempted
to modify a non-writable caps structure.
Fix by adding a gst_caps_make_writable().
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2709>
- ssl module requires an explicit TLS_SERVER role
- asyncio throws a deprecation warning when using
asyncio.get_event_loop(). Remove custom event loop handling entirely
- No need to keep the websocket server in a member variable, can use
a future to signal exit case along with the async with context manager
of websockets.serve()
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2698>
... and don't use GstD3D11VideoProcessor. Now GstD3D11Converter will
be able to convert using videoprocessor, and texture upload is also supported by
GstD3D11Converter. All the noisy code can be removed therefore.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2697>
* Add videoprocessor feature to d3d11converter, in order to unifiy
conversion flow.
* Add convert_buffer() method to support automatic shader/videoprocessor
selection. The method also supports texture upload if input memory
cannot be used for conversion (e.g., system memory or so)
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2697>
There's no need to re-assign the return value of
g_string_append_*() functions and such to the variable
holding the GString. These return values are just for
convenience so function calls can be chained. The actual
GString pointer won't change, it's not a GList after all.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2685>
This reverts commit 6f9ae5d758.
The _transform_caps() function can't tell the difference
between the caller wanting to know the output caps
for the current method, or all possible output caps. If
it includes caps for all possible methods, glupload can
end up negotiating and sending the wrong output caps
downstream.
Partially reverts !2687Fixes#1310
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2699>
Adding nvautogpu{h264,h265}enc class which will accept upstream logical
GPU device object (GstCudaContext or GstD3D11Device) instead of
using pre-assigned GPU instance.
If upstream logical GPU device object is not NVENC compatible
(e.g., D3D11 device of non-NVIDIA GPU) or it's system memory,
then user specified "cuda-device-id" or "adapter-luid" property
will be used for GPU device selection.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2666>
GstCudaMemory supports CPU access via CUDA pinned host memory already
and it would show faster memory transfer performance between
GPU and CPU than copying from/to normal system memory.
If downstream supports video meta, we can passthrough CUDA memory.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2690>
If no filter caps are provided with a caps query, always
generate a full set of all caps from all upload methods,
not just the configured one. This is needed to handle
renegotiation when dealing with raw sysmem caps - as the upload
method might accept raw sysmem caps, but only the raw data
uploader adds those to the caps query.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2687>
This reverts commit f3292dc156.
Only the raw data uploader should add sysmem caps to the
actual caps query, because we want them to be at the
lowest priority. If upstream does select to send raw
caps, then the correct upload method will still
be chosen because the accept_caps implementation
will accept them
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2687>
When checking if we need to reconfigure when uploading, check
specifically the output caps of the current method will
result in compatible/incompatible caps, not the full set
of output caps from all upload methods.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2687>
Performs crop, scale, and color space conversion all in
a single render pipeline. Note that cropping related property is not
added in this element (which will make negotiation very complicated),
but user can configure videocrop element for crop meta to be attached
on each buffer.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2678>
Fixes warnings like:
Received a structure string that contains '="0.5"'. Reading as a gdouble value, rather than a string value. This is undesired behaviour, and with GStreamer 1.22 onward, this will be interpreted as a string value instead because it is wrapped in '"' quotes. If you want to guarantee this value is read as a string, before this change, use '=(string)"0.5"' instead. If you want to read in a gdouble value, leave its value unquoted.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2621>
* When dealing with rendition streams, we attempt to synchronize the media
playlist against the variant stream. This helps with speeding up the correct
initial fragment search and avoids issues when streams at activated at a much
later time.
* Also add checks for variant stream existence before attempting to use them
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2679>
When updating playlists, there is a possibility that the playlists don't
perfectly align, but the last entry of the previous playlist is *just* before
the first entry of the new playlist.
In those cases, we still can transfer the timing information from one playlist
to another, but we do not want to return that segment as being the matching one.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2679>
When matching playlists, there is a possibility that rendition streams will not
have been updated in time (for example because that stream started later, or
playback was paused). This would cause several playback failures and seeking
failures.
In order to still fall back on our feet, attempt to synchronize that rendition
playlist against the current variant playlist. This will attempt to match the
stream time using SN/DNS/PDT/...
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2679>
If we have been updating too slowly and have gone out of the current live
window, inform the baseclass accordingly.
This is different from the case where we have been updating quicker than what
the server provides.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2679>
* Since only flushing seeks are allowed, the "current" position is always the
global output position (and not "some" stream current position).
* In terms of figuring out to which stream to "snap" to, we can send it to any
selected stream. Removes the requirement of this function to a specific output
pad.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2679>
Remove the "pending advance" hack and instead rely on the base stream current
position to track our position (instead of a potentially NULL "current
segment").
Also ensure the media playlists are always refreshed with valid stream time,
even if there is no current segment.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2679>
The stream start and current position would be properly set when seeking or
activating a stream after playback started. But it would never be properly
initialized.
Set it to NONE initially to indicate to subclasses that no position has been
tracked yet. This will allow them to detect initial stream usage.
Futhermore, once the initial streams setup is done, make sure that it is set to
a valid initial value:
* The minimum stream time in live
* Or else the period start
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2679>
If the driver does not support VIDIOC_CREATE_BUFS ioctl, the pool
configuration may get changed, which requires a validation. This would
fail to activate a pool in a case it shouldn't normally fail unless we
are out of memory.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2456>
This example code demonstrates D3D11 device sharing between
application and GStreamer. Application can access texture
using appsink and it can be rendered on application's window without
any copy operation.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2646>
Our Direct3D11 abstraction layer has been improved and
it gained good shape from API point of view.
Also, On Windows, GstD3D11 has various advantages over GstGL
in terms of compatibility/stability/feature/performance.
Note that WGL implementation is known to be buggy for some
drivers/vendors/scenario (that's a reason why Google implemented ANGLE).
Moreover, GstGL is not fully optimized for Windows unfortunately.
It's the time to open this interface to application developers
for various optimized processing using our Direct3D11
infrastructure.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2646>
This patch adds general mechanism for handling specific hacks. In this
case for jpeg decoder in i965 driver, which cannot create surfaces
with fourcc specified.
From jpeg decoder to the allocator, which creates the surfaces,
there's a non-simple path: basedec pseudo-class adds a hacks guint32
which will be set by actual elements (vajpegdec, in this case) and
basedec will always set the hack to the allocator when the allocator
is instantiated.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1575>
Given the supported rt formats in a profile/entrypoint config it's
possible to know the supported JPEG colorspace and subsampling. This
patch adds this information in coded caps to a safer autoplugging
after jpegparser.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1575>
This base class is intented for hardware accelerated decoders, but since
only VA uses it, it will be kept internally in va plugin.
It follows the same logic as the others video decoders in the library but.
as JPEG are independet images, there's no need to handle a DBP so no need
of a picture object. Instead a scan object is added with all the structures
required to decode the image (huffman and quant tables, mcus, etc.).
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1575>
Gallium drivers historically have reported strange dmabuf sizes, from always
zero to the whole frame (multiple fds). The simplest solution is to use lseek
SEEK_END to get the prime descriptor size.
Also the allocator raises a warning if both values differ in order to report
it to driver.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2574>
If compiled with -Dgstreamer:gst_debug=false and we have
GST_REMOVE_DISABLED defined we will get the following compiler error:
```
[...]/libgstreamer-1.0.so.0.2100.0.p/gst.c.o: in function `gst_deinit':
[...]/gst/gst.c:1258: undefined reference to `_priv_gst_debug_cleanup'
[...] hidden symbol `_priv_gst_debug_cleanup' isn't defined
```
Add the missing define guard to avoid this.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2648>
The latency messages are non-deterministic and can arrive before/after
async-done or during state-changes as they are posted by e.g. sinks from
their streaming thread but bins are finishing asynchronous state changes
from a secondary helper thread.
To solve this, expect latency messages at any time and assert that we
receive one at some point during the test.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2643>
The commit b90d0274 introduces uninitialized width and height when we
consider to change the "pixel-aspect-ratio" for some interlaced stream.
We need to check the resolution in the src caps, and if no resolution
info found, there is no need to consider the aspect ratio.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2630>
Previously it was only possible to request them with the exact template
name, e.g. 'src_%s', but not with "instantiated" names that would match
this template, e.g.'src_foo_bar'.
This is now possible and a test was added for this, in addition to
fixing a previously invalid test.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2635>
Removing glvideomixer-like nuance (it was initially referenced)
and rewriting element since it's not an optimal design at all
from performance point of view.
* Remove wrapper bin (and internal conversion/upload/download elements)
which will waste CPU/GPU resources. Conversion/blending can be done by the
d3d11compositor element at once.
* Add support YUV blending without RGB conversion.
The RGB <-> YUV conversion is completely unnecessary since YUV textures
support blending as well.
* Remove complicated blending operation properties since it's hard
to use from application point of view. Instead, adding "operator" property
like what compositor element does.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2631>
Similar to and inspired by glimagesink and gtkglsink.
Using the Wayland buffer transform API allows to offload
rotate operations to the Wayland compositor. This can have
several advantages:
- The Wayland compositor may be able to use hardware plane
capabilities to do the rotation.
- In case of pre-rotated content on rotated outputs the
rotations may equal out, potentially allowing the
compositor to use hardware planes even if they don't
support rotate operations.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2543>
AVC and HEVC define crop rectangle and the x/y coordinates might
not be zero. This commit will address the non-zero x/y offset coordinates
via GstVideoCropMeta if downstream supports the meta and d3d11 memory.
Otherwise decoder will copy decoded texture into output frame.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2624>
Some mpeg-ts streams have extra data at the beginning. While it's not ideal, we
should be able to cope with it.
Therefore increase the initial search window for at least 4 consecutive
synchronization points to 1kB.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2626>
This allows the reception of streams that don't exactly match
the codec preferences. In particular, the ssrc in the codec preferences
is local sender SSRC, the other side is expected to send a different SSRC.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2615>
Some encoders (e.g. Makito) have H265 field-based interlacing, but then
also specify an 1:2 pixel aspect ratio. That makes it kind-of work with
decoders that don't properly support field-based decoding, but makes us
end up with the wrong aspect ratio if we implement everything properly.
As a workaround, detect 1:2 pixel aspect ratio for field-based
interlacing, and check if making that 1:1 would make the new display
aspect ratio common. In that case, we override it with 1:1.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2577>
Various variables were of smaller types than needed and there were no
checks for any overflows when doing additions on the sizes. This is all
checked now.
In addition the size of the decompressed data is limited to 200MB now as
any larger sizes are likely pathological and we can avoid out of memory
situations in many cases like this.
Also fix a bug where the available output size on the next iteration in
the zlib decompression code was provided too large and could
potentially lead to out of bound writes.
Thanks to Adam Doupe for analyzing and reporting the issue.
CVE: tbd
https://gstreamer.freedesktop.org/security/sa-2022-0003.html
Fixes https://gitlab.freedesktop.org/gstreamer/gstreamer/-/issues/1225
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2610>
Various variables were of smaller types than needed and there were no
checks for any overflows when doing additions on the sizes. This is all
checked now.
In addition the size of the decompressed data is limited to 120MB now as
any larger sizes are likely pathological and we can avoid out of memory
situations in many cases like this.
Also fix a bug where the available output size on the next iteration in
the zlib/bz2 decompression code was provided too large and could
potentially lead to out of bound writes.
Thanks to Adam Doupe for analyzing and reporting the issue.
CVE: CVE-2022-1922, CVE-2022-1923, CVE-2022-1924, CVE-2022-1925
https://gstreamer.freedesktop.org/security/sa-2022-0002.html
Fixes https://gitlab.freedesktop.org/gstreamer/gstreamer/-/issues/1225
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2610>
Uses prelude header files with #defines to rename DASH and MSS
symbols duplicated in their old standalone versions.
Also redefines soup-related functions when building it for
adaptivedemux2 to prevent symbol conflicts there.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2534>
Currently, video format is decided with downstream caps intersection,
but that's not correct since chroma is not considered. The video
decoders have to decide the output format given the used chroma, not
by the downstream caps negotiation.
This patch changes that. Still, caps feature is selected by caps
negotiation, then, with the preferred caps feature, the output format
is search within that caps feature.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2569>
The mq we get out of the weak ref might be NULL if we're
shutting down, which could cause assertion failures or
crashes.
It might also cause miscompilations where the compiler just
optimises away the NULL check because it jumps to a code path
that then dereferences the pointer which clearly isn't going
to work. Seems like something like this happens with gcc 11.
Fixes#1262
Co-authored-by: Doug Nazar <nazard@nazar.ca>
Co-authored-by: Sebastian Dröge <sebastian@centricular.com>
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2599>
Fixes:
../plugins/elements/gstmultiqueue.c: In function ‘gst_multi_queue_loop’:
../plugins/elements/gstmultiqueue.c:2394:19: warning: ‘is_query’ may be used uninitialized in this function [-Wmaybe-uninitialized]
2394 | if (object && !is_query)
| ^~~~~~~~~
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2593>
Rewriting GstD3D11Converter (equivalent to GstVideoConverter)
to optimize some conversion path and clean up.
* Extract YUV <-> RGB conversion matrix building method to
utils. It will be used by other implementation
* Use calculated offset values for YCbCr <-> YPbPr conversion
instead of hardcoded values
* Handle color range adjustment
* Move transform matrix building helper function to utils.
The method will be used by other elements
* Use single constant buffer. Multiple constatne buffer for
conversion pipeline is almost pointless
* Remove lots of duplicated HLSL code and split pixel shader
code path into sampling -> colorspace conversion ->
shader output packing
* Avoid floating point precision error around UV coordinates
* Optimize RGB -> YUV conversion path
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2581>
