Closed z0Kng closed 1 month ago
Auto Created VSMGWL-74530 for further analysis.
I have more information, My setup is:
Furthermore, the following error messages occur when using gst-inspect or using the plugins:
0:00:00.450089145 85446 0x55a544aeb950 ERROR vamemory gstvaallocator.c:616:_va_create_surface_and_export_to_dmabuf: Different objects have different modifier
(gst-plugin-scanner:85446): GStreamer-CRITICAL **: 08:01:31.514: gst_caps_append: assertion 'GST_IS_CAPS (caps2)' failed
Also the pipeline crashes after some time:
gst-launch-1.0 videotestsrc ! video/x-raw,format=BGRx ! vapostproc ! "video/x-raw(memory:VAMemory),format=NV12" ! vah265lpenc ! h265parse ! matroskamux ! filesink location=test.mkv
Setting pipeline to PAUSED ...
Pipeline is PREROLLING ...
Got context from element 'vah265lpenc0': gst.va.display.handle=context, gst-display=(GstObject)"\(GstVaDisplayDrm\)\ vadisplaydrm6", description=(string)"Intel\(R\)\ Gen\ Graphics", path=(string)/dev/dri/renderD128;
Redistribute latency...
Redistribute latency...
Pipeline is PREROLLED ...
Setting pipeline to PLAYING ...
Redistribute latency...
New clock: GstSystemClock
Bus error (core dumped)
solved with patch https://github.com/intel-innersource/drivers.gpu.unified/pull/173949
Fixed by 019772d. Close it
Which component impacted?
Video Processing
Is it regression? Good in old configuration?
None
What happened?
Using this pipeline gstreamer
gst-launch-1.0 videotestsrc ! video/x-raw,format=BGRx ! vapostproc ! "video/x-raw(memory:VAMemory),format=NV12" ! vah265lpenc ! h265parse ! matroskamux ! filesink location=test.mkv
produces a tiled image on an intel flex 140. This bug also happens for BGRx, xRGB, xBGR and RGBx as input, while the formats with alpha channel work. Using msdkvpp instead of vapostproc works. Using vaapipostproc and vaapih265enc instead also.
I only could reproduce this bug on an Intel Flex GPU.
I have already opened an issue at gestreamer: https://gitlab.freedesktop.org/gstreamer/gstreamer/-/issues/3634 and was asked to open another one here as it is suspected to be a driver problem.
What's the usage scenario when you are seeing the problem?
Transcode for media delivery
What impacted?
We are currently developing an opensource vdi solution using Flex vGPUs for graphics acceleration and video stream encoding.
Debug Information
4.vainfo.log 5.libva_trace.log
Do you want to contribute a patch to fix the issue?
None