Replies: 1 comment 4 replies
-
Unless you really need the gobject introspection files, you could work around the problem by disabling their generation for that recipe. To fix the problem, though, you need to track down where in the gstreamer code it's calling on the nvbuf APIs, and see if you can avoid doing that for the code paths used during GI generation... assuming that's possible. Otherwise, you might have to defer the generation to on-target, since it's unlikely the NVIDIA libraries could ever be made to work in the usermode QEMU used during the build. |
Beta Was this translation helpful? Give feedback.
4 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
I'm building a kirkstone image, and trying build gstreamer1.0-plugins-base with the NVMM interop in the GL stack. Compiling requires depending on tegra-libraries-multimedia-utils and tegra-mmapi to pull in the nvbuf_utils header and lib.
Here's my
gstreamer1.0-plugins-base_%.bbappend
:The first issue was that causing a circular build dep on tegra-libraries-multimedia-utils, but I saw that was solved in master last year via #1243. I backported that, and the gstreamer1.0-plugins-base build successfully detects and tries to build the NVMM support. The problem is that it cause a gobject-introspection failure, with the nvidia library trying to initialise things that aren't available in qemu (I think):
I'm not sure how to work around that, or why it doesn't affect some of the other nvidia GStreamer packages that build against
tegra-libraries-multimedia-utils
Any suggestions appreciated.
Beta Was this translation helpful? Give feedback.
All reactions