v4l2-source/libcamera-source: Multi-plane formats (like NV12) don't have info about contiguous planes
Buffers from cameras providing video in formats like NV12
currently don't seem to carry any explicit info about additional contiguous planes, leaving it to apps to guess that info (e.g. about number of planes, offsets) by themselves. Notably spa_buffer.n_datas
is set to 1
.
While apps (like e.g. Gstreamer) currently can rely on a linear modifier being used and the stride containing no padding to compute the offsets of extra planes, it would make Pipewire easier to use if such info was explicitly provided.
This would be especially helpful for apps relying on e.g. GPU drivers to know how to handle the formats in question - e.g. when uploading NV12 buffers from the camera to GL/Vulkan/Wayland.