please share another YUV dump with proper width/height settings, and, please dump several frames for reference.
for example, $ v4l2-ctl --device /dev/video0 --set-fmt-video=width=1920,height=2306,pixelformat=NV16 --set-ctrl bypass_mode=0 --stream-mmap --stream-to=frame.yuv --stream-count=5
Please explain why the dump is YV16.
There are data in range 0x0~0x438eff (1920x2306 bytes, Y area).
And UV area is filled with ‘0’ (0x438f00~0x871dff).
there might be some bug for surface alignment, we don’t have sensor for digging into this.
could you please examine pixel memory format,
for example, it should be TEGRA_IMAGE_FORMAT_T_Y8__V8U8_N422 = 44; in driver layer for NV16.
since NV16 has two planes (Y and UV) in a single buffer, hence we need to configure the additional surface accordingly.
please also try adding below to VI driver,