Board: Toradex Luna SL1680
Torizon OS: 7.5.0-devel-20260123143454+build.0 (scarthgap), refspec
torizon:7/winglet/common-torizon/torizon-docker/testing
I’m trying to bring up an Arducam OV9782 MIPI CSI-2 camera on CSI0 (with the
expander adapter board). v4l2-ctl --list-devices only shows the SoC’s own
ISP/HDMI-in/encoder/decoder nodes (vvcam-video.0.0-3, syna-hrx,
syna-h1-encoder, syna-v4g-vdec) — no sensor subdev, no i2c client for the
sensor.
The image ships with camera overlays already present at
/boot/ostree//dtb/overlays/:
dolphin-bothcsi-without-expander.dtbo
dolphin-csi0-with-expander.dtbo
dolphin-csi1-without-expander.dtbo
dolphin-csi0-with-expander-csi1-without-expander.dtbo
selectable via fdt_overlays=<name>.dtbo in overlays.txt next to them
(empty by default). However, /boot/loader.1/uEnv.txt’s uenv_extra_configs
unconditionally does env set skip_fdt_overlays 1, which disables overlay
application regardless of overlays.txt.
I flipped that flag to 0 (single-line edit) and applied an overlay, and
tested twice with different overlays:
- dolphin-csi0-with-expander.dtbo
- dolphin-bothcsi-without-expander.dtbo
Both times, on the next boot the board hung completely: power LED on, HDMI
stays black/no signal, unreachable on the network, indefinitely. No serial
console is broken out on this board that I could use to see U-Boot/kernel
output. Only recovery was a full reflash via the USB_BOOT jumper + GCT port
- flash-image.sh procedure from the getting-started guide, both times.
Since two different overlays produced the identical hang, this doesn’t look
like a bad overlay choice — more like overlay application itself is
broken/unverified on this testing-channel build, and skip_fdt_overlays=1
might be an intentional safety default rather than an oversight.
Questions:
- Is CSI camera overlay support known-working on a different
Torizon OS build for the Luna SL1680, or still WIP? - Is there a supported way to bring up the OV9782 today (e.g. via
TorizonCore Builder instead of a raw uEnv.txt edit), or a recommended
sensor/overlay combination that’s actually been validated? - Is there a debug UART/serial console header on this board I’m missing,
for next time this happens?
Happy to share full dmesg/journalctl output, exact uEnv.txt diff, or
try anything else that’d help narrow this down.