Dear Community We are currently experiencing several issues related to the CM4 camera and device detection in our lab environment. Below is a summary of the problems we've encountered: 1. **Renesas Firmware Module Missing**: The firmware fails to load as there is no external module available to enable it. We could not locate any `hci_usb_renesas.ko` or `hci_pci_renesas.ko` modules. 2. **Bus Disappearance After Power Loss**: When the camera power is disconnected without proper shutdown, USB buses 002 and 003 disappear. Running `dmesg | grep renesas` indicates that the Renesas module failed to load. 3. **Persistent Failure After Reboot**: Even after multiple reboots, the system cannot recover or re-enable the controller manually. 4. **Multiple Kernel Versions in Use**: We are testing across various kernel versions (6.6.51 / 6.6.74 / 6.6.81) but the issue persists. 5. **USB Speed Shifts**: We observed frequent transitions from USB speed 1.1 to 2.1, as shown in the attached screenshots. 6. **DepthAI Version Impact**: Updating DepthAI from version 2.28.0.0 to 2.30.0.0 resolves the `dai.Device not found` issue. 7. **Inconsistent Bus Detection with Movidius**: Often only bus 003 is visible when using Movidius devices, while bus 002 is not — we’re unsure what this behavior implies. Could you please assist or advise on how to address these issues ? [upl-image-preview uuid=af4aa875-2a9a-41f1-9560-cefa195ee2ff url=https://discuss.luxonis.com/assets/files/2025-08-08/1754658296-339496-tlc-ok-1.png alt={TEXT?}] [upl-image-preview uuid=e7c665a4-e853-404f-9c93-2cde3044d0d2 url=https://discuss.luxonis.com/assets/files/2025-08-08/1754658296-375018-tlc-fail-2.png alt={TEXT?}] [upl-image-preview uuid=21f5f04f-f76d-4761-86e6-36b57927ff22 url=https://discuss.luxonis.com/assets/files/2025-08-08/1754658296-408134-tlc-fail-1.png alt={TEXT?}]

@"martin181818"#p30804 Can you try following this thread - https://discuss.luxonis.com/d/5966-using-luxonis-preped-rpi-os-images-v10/2 Thanks, Jaka

Hello everyone, I'm having a similar problem: `(venv) pi@luxonis:/usr/bin/user $ lsusb Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub Bus 003 Device 001: ID 1d6b:0003 Linux Foundation 3.0 root hub Bus 002 Device 002: ID 03e7:2485 Intel Movidius MyriadX Bus 002 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub (venv) pi@luxonis:/usr/bin/user $ python Python 3.11.2 (main, Mar 13 2023, 12:18:29) [GCC 12.2.0] on linux import depthai as dai dai.Device.getAllAvailableDevices() [2025-08-18 12:03:04.554] [depthai] [warning] skipping X_LINK_UNBOOTED device having name "2.1" []` Following the suggestions on the link I get: `(venv) pi@luxonis:/usr/bin/user $ echo -n 0000:01:00.0 | sudo tee /sys/bus/pci/drivers/xhci-pci-renesas/bind tee: /sys/bus/pci/drivers/xhci-pci-renesas/bind: No such file or directory` '(venv) pi@luxonis:/usr/bin/user$ ls /sys/bus/pci/drivers/ 8250_pericom cn10k_rng dwc3-haps exar_serial nvme pcieport rp1 serial xhci_hcd' Do you have any info on why the renesas module does not seem to be present? Also, was this the actual solution of the original problem? because from the discussion linked above it seems like you are still checking the core cause. Thank you

@"Stefano"#p30951 In your case the device seems to be detected correctly so it is not the same issue. Are you on CM4 as well? Which image are you using? Thanks, Jaka

@"jakaskerl"#p30963 I just used 'sudo apt update && sudo apt -y upgrade' now when i call lsusb it returns: Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub Bus 003 Device 001: ID 1d6b:0003 Linux Foundation 3.0 root hub Bus 002 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub I tried again with the suggestion, but it seems the resource is busy. I rebooted the camera, but I keep having the issue. echo -n 0000:01:00.0 | sudo tee /sys/bus/pci/drivers/xhci-pci-renesas/bind 0000:01:00.0tee: /sys/bus/pci/drivers/xhci-pci-renesas/bind: Device or resource busy dai.Device.getAllAvailableDevices() keeps returning [] do you have any idea? Thank you

@"Stefano"#p30966 https://drive.google.com/drive/u/1/folders/1O50jPpGj_82jkAokdrsG--k9OBQfMXK5 I think the latest image should have this fixed cc @"OskarSonc"#3466 Thanks Jaka

@"jakaskerl"#p30997 Thank you for your help. I downloaded version V11 of the OAK image, and the Renesas loading issue is still present. I found that this is a concurrency problem between the moment the xhci-pci module (which includes the Renesas driver) actually starts and requests the file `K2026.mem` in `/lib/firmware/renesas/`, and the moment when this file is actually available. If the file is not found, the initialization fails and does not recover. Starting from this situation, which applies to all kernel versions, the following manual command: `sudo ln -sf renesas/K2026.mem /lib/firmware/renesas_usb_fw.mem` creates a symbolic link between these two files. Then the command: `echo -n 0000:01:00.0 | sudo tee /sys/bus/pci/drivers/xhci-pci-renesas/bind` is executed to force the reinitialization of the search. After this procedure, Bus 002 and Bus 003 are correctly detected, and the device can be identified using `dai.Device.getAllAvailableDevices()`. If a normal reboot is performed, or the Ethernet cable is disconnected, the device no longer detects Bus 002 and 003. To detect them again, it is sufficient to run the second command: `echo -n 0000:01:00.0 | sudo tee /sys/bus/pci/drivers/xhci-pci-renesas/bind` Is there a way to verify that the installed version of the image is indeed the correct one? Thank you

@"OskarSonc"#p31086 I am using the PoE version of CM4. Thank you.

CM4 connection issues

OskarSonc

Hey @martin181818, thanks a lot for digging into this and sharing the details.

When I tested the device I didn’t see the issue you describe, but I’ll run some more tests on my side to double-check.
Can you let me know if you’re using a PoE CM4 device or a non-PoE version? That will help narrow it down.

martin181818

OskarSonc

I am using the PoE version of CM4.
Thank you.

Stefano

jakaskerl
I'm having some problems in flashing the new image.
I start the flashing procedure, and after a while it freezes, one time at 96%, another one at 19%.
Any suggestion about this problem?

Also, did you manage to complete your test to check if the latest image actually fix the original issue of the post?

Thank you

OskarSonc

Stefano
Are you using Raspberry Pi Imager or balenaEtcher?
It's normal that it takes a lot of time, since it's USB 2.0 connection…

We are working on it. Looks like the problem is only for PoE device…
For now these two commands seem to resolve it:
sudo ln -sf renesas/K2026.mem /lib/firmware/renesas_usb_fw.mem

echo -n 0000:01:00.0 | sudo tee /sys/bus/pci/drivers/xhci-pci-renesas/bind

Stefano

OskarSonc
I tried with both PiImager and balenaEtcher, the flashing freezed after a while.
I left the process go on for a couple of days before stopping, so I don't think it to be just slow.

Let me know about the PoE actual solution as soon as possible please.

Thank you

OskarSonc

Hey @Stefano ,

Can you double-check that the switch is set to ENABLE and that you followed the steps from the tutorial:

https://docs.luxonis.com/hardware/platform/deploy/to-rpi#Deploy%20with%20Raspberry%20Pi-Flashing%20an%20image%20to%20the%20eMMC?

Also, please try a different USB cable/port if possible.

For the PoE issue, yes I will let you know…

Best
Oskar

OskarSonc

@Stefaan @martin181818 I’ve just uploaded an updated OAK_CM4_POE_V11_64bit.img.xz image.
It should fix the issue with Renesas drivers not loading.
You can download it and flash it to your CM4 PoE:
https://drive.google.com/drive/folders/1O50jPpGj_82jkAokdrsG--k9OBQfMXK5?usp=drive_link

Thanks
Oskar

Stefano

Hi OskarSonc ,
The new image was working well up to yesterday.
This morning, after a reboot, I have again the same issue, "No available device".
I tried to boot the camera many times, without starting any custom script or service.

If I call "depthai.Device.getAllAvailableDevices()" I get "[]".

I tried the "old solution":

echo -n 0000:01:00.0 | sudo tee /sys/bus/pci/drivers/xhci-pci-renesas/bind
with this result:
0000:01:00.0tee: /sys/bus/pci/drivers/xhci-pci-renesas/bind: Device or resource busy.

this is my kernel version:
Linux luxonis 6.12.34+rpt-rpi-v8 #1 SMP PREEMPT Debian 1:6.12.34-1+rpt1_bookworm (2025-06-26) aarch64 GNU/Linux

My camera is installed in a client plant and it is difficult for me to physically reach it.
What can I do to solve the issue?

I need a reliable solution, because I have many cameras installed and I cannot afford to have them randomly stop working without a clue of why and without way to fix the problem without flashing the camera.

Thank you,

OskarSonc

Sorry for late reply - tried to get my device in this state but couldn't.
“Device or resource busy” means the Renesas controller is already bound to a driver. Please unbind first, then bind again.
Thanks;
Oskar

Stefano

Hi @OskarSonc,

At some point the error changed, and I got:

pi@luxonis:~ $ echo -n "0000:01:00.0" | sudo tee /sys/bus/pci/drivers/xhci-pci/bind
tee: /sys/bus/pci/drivers/xhci-pci/bind: No such file or directory

using command dmesg | tail -n 20
I found:
xhci-pci-renesas 0000:01:00.0: hardware is in an error state. Giving up (poweroff/reboot needed).

I had to power off the camera (sudo reboot was not enough) and it started working again.
The problem is that I cannot easily power it off, and I don't understand why this happens and how to fix it.
Can you help me?

Thanks

OskarSonc

Hi @Stefano — this is a known Renesas/CM4 PCIe quirk on recent Raspberry Pi kernels.
When the controller hits a hardware error state, a warm reboot often isn’t enough — it usually needs a full power cycle to clear. We are sorry but I don't think we can fix this on our side. You could try some Raspi forums maybe...

Stefano

Ok, I get that when the issue appears, the only solution is a full power cycle, but do you have any idea about why the problem originates?
The camera was working correctly and at some point it stopped, without recognizing the internal camera anymore.

Thank you

OskarSonc

Hmm really hard to say, I guess the Renesas PCIe USB3 controller hangs (firmware/driver timing). It’s an upstream kernel/firmware quirk, not your app. My only reliable solution would be to use a PoE switch with port-power control, so you could power off the camera if the issue persists... I am really sorry, but there isn't something I could do since this is renesas/raspi driver thing...