A USB-Ethernet class (CDC-ECM) device plugged into a Systemcore USB-A port gets 172.27.0.1/24 on usb0 correctly only if it's attached before Systemcore powers on.

If it's hot-plugged into an already-running Systemcore, or its link drops and reconnects for any reason mid-session, cdc_ether re-registers usb0 cleanly at the kernel level, but the static IP/admin-up is never reapplied — usb0 is left as <BROADCAST,MULTICAST>, state DOWN, no inet line, for the rest of that boot. Only a full Systemcore power cycle recovers it; replugging the device does not.

Is this intended behavior?

• If not, then hopefully the findings below help
• If it is, then do you have any guidance for how to handle hot-plug events for devices like this?

Environment

• Systemcore OS: limelightosr-2027.0.0-alpha13
• device: a CDC-ECM class USB-Ethernet device, ex. RP2350

Root cause (confirmed via SSH)

limelight_gadget.service (Type=simple, ordered after network-online.target, no Restart=) runs its gadget --systemcore binary exactly once at boot. Confirmed directly from the journal that it executes:

sudo /usr/sbin/ifconfig usb0 up 172.27.0.1 netmask 255.255.255.0
sudo /usr/sbin/ifconfig usb1 up 172.26.0.1 netmask 255.255.255.0

matched by interface name, disregarding which physical device holds that name. Nothing else re-triggers this step. If usb0/usb1 disappear and reappear after boot for any reason, the kernel re-registers them cleanly but the static IP is never reapplied.

Steps To Reproduce

1. Power on Systemcore with the device already attached → usb0 gets 172.27.0.1/24, up. Reliable across all 4 USB-A ports, with or without a second USB device attached (ex. CANivore).
2. With Systemcore still running, disconnect/reconnect the device's USB link (replug or A driver-level unbind/rebind).
3. dmesg shows a clean unregister/register of usb0 with the same MAC. ip addr afterward:

   usb0: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN group default qlen 1000
       link/ether aa:bb:cc:dd:ee:ff brd ff:ff:ff:ff:ff:ff
   

   No inet line. Persists until a full power cycle.

Attempted Fix

Added a udev rule that re-runs the same ifconfig commands on every usb0/usb1 add event (not just the first) resolves it:

# /etc/udev/rules.d/99-usb-gadget-hotplug-fix.rules
SUBSYSTEM=="net", ACTION=="add", KERNEL=="usb0", RUN+="/usr/sbin/ifconfig usb0 up 172.27.0.1 netmask 255.255.255.0"
SUBSYSTEM=="net", ACTION=="add", KERNEL=="usb1", RUN+="/usr/sbin/ifconfig usb1 up 172.26.0.1 netmask 255.255.255.0"

Verified across multiple simulated hot-plug cycles: without the rule, usb0 comes back unconfigured every time; with it, usb0 comes back UP with 172.27.0.1/24 and is reachable (ping succeeds) every time. It's additive-only (one new file, nothing existing modified) and idempotent with the boot-time step, so it doesn't change normal boot behavior.

Two small scripts are attached to apply/roll back this exact rule over SSH.

requirements.txt
systemcore_hotplug_fix_apply.py
systemcore_hotplug_fix_rollback.py