- Fix skipsize on _update_backlog failure.
- Add robustness to AU completion detection by using AUD when present. If we've
received a AUD we overwrite the first VCL NAL detection when the result was
negative. VCL following AUD is the first VCL of next AU.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5862>
In case of tier 1 decoder, always use reference-only picture to avoid
fixed-size pool limitation and output decoded picture without
copy even for negative rate. Also do not use copy queue for GPU to GPU
copy. Copy queue is specialized for upload/download and may occupy
PCIE bandwidth. Use direct queue as recommended by vendors.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5877>
On flush event, baseclass will discard all pictures from DPB
but there can be still in-flight commands not finished yet.
Use our command queue, allocator and fence data helper objects
to keep resource available during command execution.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5877>
Instead of only supporting writing SPU data directly to YUV frames,
render the SPU data to an intermediate AYUV overlay buffer. The overlay
data is then blended to the video frame.
For the PGS format, the overlay buffer size is set to the size of the
Composition Window, and its position in the overlay composition is set
to the window position. The objects to render are now cropped when the
cropping flag is set.
For the Vobsub format, the overlay buffer size is set to the size of the
Display Area.
Once rendered, the overlay composition rectangle is now moved and scaled
to fit the video output size, to avoid clipping.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5827>
Shader visible descriptors occupy GPU resource and there are hardware
limits. Thus, in order to minimize the amount of shader visible heaps,
only non shader visible descriptor heap (staging) will be held by d3d12memory.
Then converter will copy the staging descriptor to shader visible
descriptor heap per draw.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5875>
Zero initialization would have overhead and it's not required
most cases except for textures. Use CREATE_NOT_ZEROED flag
in case of buffer resource or if a texture will be rendered without any
prior read operation.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5875>
Conversion will happen when constructed command list is executed,
not by converter element. Thus this object should not map output buffer
with write flag which will result in error if multiple threads
are building commands for the same output target frame.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5875>
Interlaced MJPEG is a big hack. Most of the streams we've found are from old
AVID tools. There are two methods to detect interlaced stream: the container
offers a height bigger (or double) than the image's height in SOF. The other
is from a APP0 marker.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5838>
Add macro which converts picture frame number to suitable timestamp in
nanoseconds for use in V4L2 VB2 buffer lookup. Since multiple codecs do
the same operation and almost all got it wrong, do it in one place so it
can be fixed in one place again, if needed.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5791>
The GstMpeg2Picture system_frame_number is guint32, constant 1000 is guint32,
GstV4l2CodecMpeg2Dec *_ref_ts multiplication result is u64 .
```
u64 result = (u32)((u32)system_frame_number * (u32)1000);
```
behaves the same as
```
u64 result = (u32)(((u32)system_frame_number * (u32)1000) & 0xffffffff);
```
so in case `system_frame_number > 4294967295 / 1000`, the `result` will
wrap around. Since the `result` is really used as a cookie used to look
up V4L2 buffers related to the currently decoded frame, this wraparound
leads to visible corruption during MPEG2 decoding. At 30 FPS this occurs
after cca. 40 hours of playback .
Fix this by changing the 1000 from u32 to u64, i.e.:
```
u64 result = (u64)((u32)system_frame_number * (u64)1000ULL);
```
this way, the wraparound is prevented and the correct cookie is used.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5791>
The GstAV1Picture system_frame_number is guint32, constant 1000 is guint32,
GstV4l2CodecAV1Dec v4l2_av1_frame.*_frame_ts multiplication result is u64 .
```
u64 result = (u32)((u32)system_frame_number * (u32)1000);
```
behaves the same as
```
u64 result = (u32)(((u32)system_frame_number * (u32)1000) & 0xffffffff);
```
so in case `system_frame_number > 4294967295 / 1000`, the `result` will
wrap around. Since the `result` is really used as a cookie used to look
up V4L2 buffers related to the currently decoded frame, this wraparound
leads to visible corruption during AV1 decoding. At 30 FPS this occurs
after cca. 40 hours of playback .
Fix this by changing the 1000 from u32 to u64, i.e.:
```
u64 result = (u64)((u32)system_frame_number * (u64)1000ULL);
```
this way, the wraparound is prevented and the correct cookie is used.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5791>
The GstVp9Picture system_frame_number is guint32, constant 1000 is guint32,
GstV4l2CodecVp9Dec v4l2_vp9_frame.*_frame_ts multiplication result is u64 .
```
u64 result = (u32)((u32)system_frame_number * (u32)1000);
```
behaves the same as
```
u64 result = (u32)(((u32)system_frame_number * (u32)1000) & 0xffffffff);
```
so in case `system_frame_number > 4294967295 / 1000`, the `result` will
wrap around. Since the `result` is really used as a cookie used to look
up V4L2 buffers related to the currently decoded frame, this wraparound
leads to visible corruption during VP9 decoding. At 30 FPS this occurs
after cca. 40 hours of playback .
Fix this by changing the 1000 from u32 to u64, i.e.:
```
u64 result = (u64)((u32)system_frame_number * (u64)1000ULL);
```
this way, the wraparound is prevented and the correct cookie is used.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5791>
The GstVp8Picture system_frame_number is guint32, constant 1000 is guint32,
GstV4l2CodecVp8Dec v4l2_vp8_frame.*_frame_ts multiplication result is u64 .
```
u64 result = (u32)((u32)system_frame_number * (u32)1000);
```
behaves the same as
```
u64 result = (u32)(((u32)system_frame_number * (u32)1000) & 0xffffffff);
```
so in case `system_frame_number > 4294967295 / 1000`, the `result` will
wrap around. Since the `result` is really used as a cookie used to look
up V4L2 buffers related to the currently decoded frame, this wraparound
leads to visible corruption during VP8 decoding. At 30 FPS this occurs
after cca. 40 hours of playback .
Fix this by changing the 1000 from u32 to u64, i.e.:
```
u64 result = (u64)((u32)system_frame_number * (u64)1000ULL);
```
this way, the wraparound is prevented and the correct cookie is used.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5791>
This reverts a part of de92a6c7f2. Unlike `image_filter` and
`video_filter`, `viewfinder_filter` does not get linked to `src` but
`viewfinderbin_queue`. Thus the fix in the mentioned commit does not
apply for it and should be reverted.
This was not spotted earlier as only the other filters are used in
the project that uncovered the issue.
Fixes: de92a6c7f2 ("camerabin: Fix source updates with user filters")
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5689>
Since the AV1 specification is not explicitly mentioning about
the array size bounds, array sizes in scalability structure
should be defined as possible maximum sizes that can have.
Also, this commit removes GST_AV1_MAX_SPATIAL_LAYERS define from
public header which is API break but the define is misleading
and this patch is introducing ABI break already
ZDI-CAN-22300
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5823>
- AU boundary detection reviewed to follow more closely H.264 spec. and more
specifically clauses 7.4.1.2.3 and 7.4.1.2.4.
- The gist of the changes is a look-a-head in then next AU required identify the
last vcl-nal of current AU and firt vcl-nal of next AU (according to
7.4.1.2.4) followed by the identification of the first nal of next AU
(according to 7.4.1.2.3).
- A backlog of all nals of current AU and next AU up to the point where current
AU can identified completed is kept.
- In NAL alignement mode vcl-nal are sent immediatly but the history is kept to
allow AU boundary detection. Non-vcl-nal can be delayed up to the reception of
the next vcl-nal to allow a correct AUD insertion.
- Based on this improved AU boudary detection we can avoid erronous AUD
insertion, like the one highlighted by test
test_parse_sliced_with_prefix_and_sei_nal_au.
- Add support for MVC AU boundary detection. (H.7.4.1.2.4)
- Explicitly report SVC not supported. We don't have the SVC NAL parsing
required to identify boundary. (missing dependency_id and quality_id fields
from SVC, see G.7.4.1.2.4)
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5741>
Serialize every GstMeta that supports serialization into the NEW_BUFFER
payload. This is especially important for GstVideoMeta in the case of
multiplanar buffers, or if stride!=width.
Sponsored-by: Netflix Inc.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5355>
The input of the vacompositor may be DMA buffers. And in this case, the input
caps has the format=DMA_DRM, which can not be recognized by base video
aggregator class' find_best_format() function. So we need to override the
update_caps() virtual function.
Also we consider the DMA kind caps in negotiated_src_caps() for output.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5160>
To achieve maximum throughput, waiting on command commit thread
is not ideal. And render-delay will introduce unwanted latency.
Best is to split thread and wait finished decoding job in a dedicated
output thread
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5812>
When creating a new VA pool set config size to zero because it's not used.
Also, given the potential different sizes from software buffer pools and VA
buffer pools, this patch handle that potential different values.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5805>
When creating a new VA pool set config size to zero because it's not used.
Also, given the potential different sizes from software buffer pools and VA
buffer pools, this patch handle that potential different values.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5805>
When creating a new VA pool set config size to zero because it's not used.
Also, given the potential different sizes from software buffer pools and VA
buffer pools, this patch handle that potential different values.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5805>
VA drivers allocate surfaces given their properties, so there's no need to
provide a buffer size to the VA pool.
Though, the buffer size is provided by the driver, or the canonical size
is used for single planed surfaces.
This patch removes the need to provide a size for the function
gst_va_pool_new_with_config() and adds a helper method to retrieve the surface
size, gst_va_pool_get_buffer_size(). Also change the callers accordingly.
Changes for custom VA pool creation will be addressed in the following commits.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5805>
If the ladspa plugin is enabled explicitly or via auto-features, the
liblrdf dependency can not be disabled.
As the RDF parsing currently provides hardly any features, the possibility
to disable it fairly useful.
Fixes: #3168
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5794>
With the way the runtime checks are currently set up, every single
openh264 release, no matter how minor, is considered an ABI break and
requires gst-plugins-bad recompilation. This is unnecessarily strict
because it doesn't allow downstream distributions to ship any openh264
bug fix version updates without breaking gstreamer's openh264 support.
Years ago, at the time when gstreamer's openh264 support was merged,
openh264 releases were done without a versioned soname (the library was
just libopenh264.so, unversioned). Since then, starting with version
1.3.0, openh264 has started using versioned sonames and the intent has
been to bump the soname every time there's a new release with an ABI
change.
This patch drops the strict version check. meson.build already has a
minimum requirement on openh264 version 1.3.0 where soname versioning
was added, which should be good enough to ensure that the library is
using soname versioning.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5780>
This avoid a build failure when compiling against OpenSSL 3.2.0. The
problem is when windows.h is included before WinSock2.h. Because
windows.h includes winsock.h[1]. Defining _WINSOCKAPI_ stops windows.h
including winsock.h.
Error:
```
[748/1041] Compiling C object ext/dtls/gstdtls.dll.p/gstdtlscertificate.c.obj
FAILED: ext/dtls/gstdtls.dll.p/gstdtlscertificate.c.obj
[...]
Windows Kits\10\include\10.0.17763.0\shared\ws2def.h(235): error C2011: 'sockaddr': 'struct' type redefinition
Windows Kits\10\include\10.0.17763.0\um\winsock.h(482): note: see declaration of 'sockaddr'
```
[1] https://stackoverflow.com/a/1372836
Closes: https://gitlab.freedesktop.org/gstreamer/gstreamer/-/issues/3167
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5770>
The original idea was to select the type of mapping (either using derive images
or downloading the image) in runtime, under the assumption that both methods
shared the same memory layout (offsets and strides), because a single
GstVideoMeta is assigned by the buffer pool at allocation time. Nonetheless, in
recent hardware this assumption is invalid, raising memory access errors.
This patch removes completely the mapping type selection at runtime, using the
method selected when the allocator is configured, synced with the bufferpool
allocation.
This problem was fixed originally for iHD driver only. But now it makes sense to
remove all of it.
Original-patch-by: Mengkejiergeli Ba <mengkejiergeli.ba@intel.com>
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5760>
When exporting a DMABuf from a VASurface the user might tell that the surface
was allocated with certain fourcc, but the returned VADRMPRIMESurfaceDescriptor
migth tell a different fourcc, as in the case or radeonsi driver, for duplicated
fourcc, such as YUY2 and YUYV.
Originally it was supposed to be a failed exportation. This patch relax this
validation by allowing different fourcc.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5760>
In multi-card scenario, user can set GST_MSDK_DRM_DEVICE env variable to
choose the device. This patch can align vpl's queried results with the
users' choice by passing deviceID when creating mfx implementation.
Co-authored-by: Víctor Manuel Jáquez Leal <vjaquez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5697>
This is a bit of a hack solution has I think the correct solution is to
expose model caps on sinkpad (eventually sinkpads). Till then I think
this is reasonable.
- Add a property to onnxinference to set datatype.
- Fix internal buffer allocation size based on datatype.
- Extract method to remove alphe channel and convert to planar image
when requested. Also template the method to support writing to buffers
of different datatype.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5761>
There is an existing PMT mapping between PCR_%s and an mpegtsmux sink
pad name, where %s equals the program number that the PCR corresponds
to. We re-purpose this functionality to also support a mapping between
PCR_%s and an arbitrary PID. If this mapping is set, then the header PCR
PID is set to this value, and PCR is attached to the stream with this
PID.
Note: the current implementation also attaches PCR to the video stream,
so this may be inefficient.
Co-authored-by: Jordan Yelloz <jordan.yelloz@collabora.com>
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5726>
Clip tile rows and cols to 64 as describe in AV1 specification
to avoid writing outside array range but preserve sb_cols
and sb_rows value which are used to futher computation.
Fixes ZDI-CAN-22226 / CVE-2023-44429
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5702>
in the case of an upstream element proposing a buffer pool,
use it to allocate the buffer image with the given parameters
set by the upstream element.
Besides the buffer pool handling is sync'd with GstBaseTransform
base class.
See the case of vulkanupload ! vulkanh264enc
Co-authored-by: Victor Jaquez <vjaquez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5651>
Direct3D feature level 10 supported GPUs were released
more than 15 years ago, around the time when Windows
Vista / 7 were released. Also our d3d11 plugin/library
does not support feature level 9.x very well already.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5709>
We were previously always querying index 0, and while the number of planes per
buffer will never change, it seems more proper to query the right buffer rather
than always the first one.
This was found while reading strace logs, and wondering why the
V4L2_BUF_FLAG_MAPPED flag was present on all ¬0 indices even though that
happened before VIDIOC_EXPBUF.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5647>
- GstAnalyticRelationMeta is a base class for analytics
meta. It's able to store analytics results (GstAnalyticRelatableMtd)
and describe the relation between each analysis results.
- GstAnalysisRelationMeta also contain an algorithm able to explore
analysis results relation using a bfs.
- Relation(edge) between analysis results (vertice) are stored in an adjacency-matrix
that allow to quickly identify if two analysis results are related and by
which relation they related. It also work for indirect relation
and can provide the path of analysis results by which two
analysis results are related.
- One allocation per buffer to store analysis results. Here we rely on
the application to guess how much space will be required to store all
analysis results. This is something that could be improved
significantly but it's a starting point.
- Define common analysis results, classification, object-detection,
tracking that are subclass of GstAnalyticRelatableMtd. The also
provide exemple of how to extend GstAnalyticRelatableMtd to have them
benefit for the mechanim to express relation with other analysis
results.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4962>
During the video session memory allocation, the property flags can
be different from the expected ones, so do not expect all the
property flags and test it with G_MAXUINT32
It's failing with driver 525.47.26 and NVidia HW NVIDIA GeForce
RTX 3050 and 2060
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4850>
Take the case into account when user filters have been set before the
source gets updated.
Note that the further linking of the filters, if present, happens below
in the `gst_camera_bin_check_and_replace_filter()` calls.
The audio filter is still affected by the same issue but left out for
now.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5527>
Even if IDXGIOutput6 says current display colorspace is HDR,
captured texture via IDXGIOutputDuplication::AcquireNextFrame()
is converted frame by OS unless we use IDXGIOutput5::DuplicateOutput1()
with DXGI_FORMAT_R16G16B16A16_FLOAT format, in order for captured
frame to be scRGB color space. Then application should perform
tonemap operation based on reported display white level, color primaries, etc.
Since we don't have any tonemapping implementation, ignores colorimetry
reported by IDXGIOutput6.
Fixes: https://gitlab.freedesktop.org/gstreamer/gstreamer/-/issues/3128
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5671>
This is how it was documented and how it worked before the port to GstPlay.
Without this, applications expecting signals to be emitted directly
without anything running the main context will simply not receive any
signals.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5672>
d2d runtime seems to execute pending GPU command list
when DXGI ID2D1RenderTarget is being released, and it will invoke
d3d11 immediate context APIs. Should protect all rendering operations
and DXGI resources with lock.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/5659>
The code seems to validate that the media-level fingerprint matches
the fingerprint of the previous media or of the whole session. There
is no such requirement in any RFC I found. The session-session one
is just meant to act as a fallback when there is no media-level
fingerprint.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1118>
