With updated egl in L4T 35.1.0, we no longer have to workaround
missing support for EGL_NV_stream_consumer_eglimage and broken
eglQueryStreamConsumerEventNV.
Signed-off-by: Kurt Kiefer <kekiefer@gmail.com>
Both the tegra-udrm-gbm and tegra-libraries-gbm-backend work to
provide buffers for libgbm (they are gbm backends).
This change adds a provider so that either implementation can be
chosen, preferring NVIDIA's official binary version.
Signed-off-by: Kurt Kiefer <kekiefer@gmail.com>
Drops the older 1.14 recipe variant for the couple of
plugins that had split 1.14/1.20 recipes.
Signed-off-by: Matt Madison <matt@madison.systems>
gstreamer fixup 1
more gstreamer fixup
Since we've dropped our own recipe for xserver-xorg, restore the
bbappend we used to use to adjust the PACKAGECONFIG settings for
the tegra platforms.
Signed-off-by: Matt Madison <matt@madison.systems>
The VPI libraries link against the CUDA driver library and
the NVSCI library without using the proper SONAME. Add
the unnumbered shared object to RPROVIDES in the relevant
recipes so they can be found without hacking the ELF headers
in VPI libraries.
Signed-off-by: Matt Madison <matt@madison.systems>
Both Xavier NX and AGX Orin devkits use the same wifi module and
as such can share a common include to add wifi support.
This also keeps the devkit-specific wifi configuration for NX modules
out of the common xavier-nx.inc so that it is more configurable for
custom machines.
Signed-off-by: Kurt Kiefer <kekiefer@gmail.com>
Similar to the xavier nx devkit, the agx orin devkit includes a
rtl8822ce m.2 pcie wifi+bluetooth radio.
This change adds by default the necessary bits to enable it.
Signed-off-by: Kurt Kiefer <kekiefer@gmail.com>
- update LICENSE
- refresh overrides conversion
- introduce workaround to fix issues during debug-splitting process for
the .ko files
Signed-off-by: Ilies CHERGUI <ilies.chergui@gmail.com>
Versioning for some of the CUDA libraries is diverging from the
main CUDA version number, leading to incorrectly named symlinks
in the lib/stubs directory using the symlink hacking we were
doing.
Instead of trying to second-guess the naming, drop the symlink
hacking altogether, and just package the stubs separately.
We can then disable shlibs processing on the stubs by default,
and make the -dev package rdepend on the stubs package.
The cuda-driver and cuda-nvml recipes need to override some of
these defaults, due to them being -dev packages that refer to
libraries provided elsewhere by driver packages.
Signed-off-by: Matt Madison <matt@madison.systems>
Looks like NVIDIA has updated the NvDrmRender to support the newer
pseudo-DRM interface now, so we can drop this.
Signed-off-by: Matt Madison <matt@madison.systems>
The nvvidconv plugin now makes CUDA driver/runtime calls, so
add those dependencies and fix the makefile to locate the
headers and libraries.
Signed-off-by: Matt Madison <matt@madison.systems>
