Right now it seems that the included driver for MCP2515 is broken in Rockchip kernel. We tested with upstream Linux 6.1 and it worked fine. We are looking to backport it.
[230526] System Release Notice for ROCK 5B
The system I am currently working on is Linux rock-5b 5.10.110-11-rockchip #8c4e8d205.
Quite disappointed with my rock pi’s collecting dust as they are still unstable and not much of functionality is implemented. Even simple things as transferring a larger file over FTP caused network to reset on the rock pi. I recently also had to change my Cardano testnet pool from a rock pi due to stability issues as well. Great hardware on paper but the software support has been very slow.
This is a Rock 5 forum
1 Like
Yeah sorry to be specific my rock 5b boards (all 12) have had stability issues and closest I have had for usecase is simple low i/o operations with the ubuntu server image. Anything more advanced say kodi or as a book library and it fails on for example larger file transfers (suspect not good network driver implementations so far). Good luck booting with nvme+wireless with any regularity either. Just my honest impressions after well over half a year with the boards. Happy if others got it working for them.
Have you tried other distributions? I’m not naming them as this thread is about one 
Same here. While i finally got mine repaired ( after months of frustration, it came with semi-functioning power chip ) and it works, i’m afraid to trust it for actual use, or even update it as it might not work afterward. So it sits in a box. Which is a shame. But it is what it is.
I have the following configuration:
- SBC: Rock Pi 5B
- Power Supply: Radxa PD 30
- eMMC 5.1 for ROCK 4 / 3A / 5B / E 32GB
- WiFi-Modul: Rock Pi M.2 Wireless Module A2
- NVMe: WD Blue SN570 1TB
- OS: rock-5b_ubuntu_jammy_cli_b36.img.xz (Build 36)
The Ethernet connection is on the local network and the Internet comes via the wireless module. Booting is done via eMMC. The NVMe is only used for data independent of the operating system.
Problem
After a few hours, the network dies. The Rock Pi 5B can only be controlled via the directly connected keyboard. Whereby this network crash also happens without connected screen and keyboard. This problem is repeatable. The network can not be started with systemctl ..., this commano hangs too. This problem also exists with build 33 with another Rock Pi 5B.
Questions
- Am I alone with this problem?
- What are the possible solutions?
- Is it certain that a Rock Pi 5B runs over several days without problems?
First thing first.
Where is dmesg on pastebin?
@HugoGraf please run the following command to save the kernel message once the network crashed, and upload it to Debian Pastebin:
sudo dmesg > kmsg.log
There are two different Rock Pi 5B with different installed versions of the OS. More information is at the beginning of pastebin.
Build(33)
Rock Pi 5B / Died network #1
Rock Pi 5B / Died network #2
Build(36)
Rock Pi 5B / Died network #3
@Stephen brcmfmac SDIO issue again?
@HugoGraf I assume the Ethernet was still working, only the Wi-Fi was down? If so can you describe what kind of traffic went through Wi-Fi?
An ssh on the local network (Ethernet) is no longer possible. A ping to an address in local network executed on this Rock Pi still gives answers. This command can no longer be aborted. All network commands hang. Most of the time there is little activity on this WiFi to the internet. I have many SBC of different models with similar configuration in use and they work fine. Only on the Rock Pi 5B I have not been able to observe stable behavior.
This cron-job runs every 5 minutes. It publishes the ipv4 and ipv6 address to duckdns.org. For security reasons, XXXX and YYYY were used here for the domain and token, respectively.
ipv6=$(ip addr show dev "wlan0" | sed -e's/^.*inet6 \([^ ]*\)\/.*$/\1/;t;d' | grep -v '^fd00' | grep -v '^fe80' | head -1)
curl -s "https://www.duckdns.org/update?domains=XXXX&token=YYYY&ip=&ipv6=$ipv6" -o ~/duckdns/duckdns.log
echo url="https://www.duckdns.org/update?domains=XXXX&token=YYYY&ip=" | curl -k -o ~/duckdns/duck.log -K -
Otherwise, there is heavy data traffic only during a few minutes per day. However, the network did not die during this heavy traffic.
Please make sure your system has been upgraded to the latest version first. You can run sudo apt update && sudo apt full-upgrade to upgrade your existing system.
Wireless dies u mean? So it happens that I had lost access to my bt once upon a long time ago and when checked it was my bt antenna that had broken on the A8 module so yeah for troubleshooting check about them just in case. You can replace them easy to better ones even with more dbis.
The cause of this error crashes the WiFi after a while and later the entire LAN of the Rock Pi.
brcmfmac: brcmf_sdio_bus_sleep: error while changing bus sleep state -110
brcmfmac: brcmf_sdio_txfail: sdio error, abort command and terminate frame
I searched for solutions on the Internet and found this one:
RPI4 arm64 wireless issues w/ power management #3127
Since the command iwconfig wlan0 power off is executed after boot, my two Rock Pi 5B run stable. You might want to adjust the default configuration of the “Rock Pi M.2 Wireless Module A2” accordingly. In the meantime I still get
ieee80211 phy0: brcmf_cfg80211_get_station: GET STA INFO failed, -52
but this does not cause a network crash.
1 Like
Interesting. I’ll check what this does.
1 Like
Btw, where is current kernel repository? I tried to look into any kind of list of commits between kernel versions - and discovered that there is literally none
Previously it was here
But the latest commit here is from 2022
The above problems also exist on rock 3a. Can you release the corresponding spi image?