Hello,

I have installed the currently provided image "Armbian_24.11.1_Nanopct6_bookworm_current_6.12.0_minimal.img.xz" on my NanoPC-T6 (not LTS). Unfortuantely the only one USB3 port is not working. So no local access is possible and no external USB storage device can be used. I wonder if no one else has such problem reported yet since 3 weeks now (released 2024-11-21).

The USB3 port is working with the FriendlyELEC image (rk3588-sd-debian-bookworm-core-6.1-arm64-20241112.img.gz), so maybe there is a small problem with a kernel module or the device tree.

If someone can guide me to a starting point for investigation, I can do some debugging. I have little experience with device tree decompiling/compiling etc.

Thanks Thomas

Edited December 14, 20241 yr by g2t

mainline support for rk3588 isn't complete yet.

I suggest to try vendor 6.1.y kernel based image.

Thanks, I will try this now: https://dl.armbian.com/nanopct6/Bookworm_vendor_minimal

Edited December 14, 20241 yr by g2t

It is working, also for the USB-C port now.

 

Many Thanks for the quick help!

Crashing the party here - I also have the same issue with my NanoPC-T6. Unfortunately USB3 (5 Gbps) is for me still not working with Armbian and the suggested kernel (vendor-6.1.y).

 

I can only get USB2 or slower to work (480Mbps or 12 Mbps). Have tried many devices and if I use a USB2 cable I can get a USB3 device to work but only at USB2 speed. The same device with a USB3 cable will not be detected at all.

 

During booting I get the following error if a USB3 device is connected:

usb6-port1: Cannot enable. Maybe the USB cable is bad?

 

The problem is the same with other distros, see e.g.:

https://github.com/Joshua-Riek/ubuntu-rockchip/issues/1123

 

The USB-C port works fine at all speeds. Any suggestions?

Edited December 17, 20241 yr by ZenSay
Spelling error

1 hour ago, ZenSay said:

I can only get USB2 or slower to work (480Mbps or 12 Mbps). Have tried many devices and if I use a USB2 cable I can get a USB3 device to work but only at USB2 speed. The same device with a USB3 cable will not be detected at all.

 

Stability of USB stack is quite a problem and should be responsibility of Rockchip. It can't be ours - they don't support us, you don't support us, other distributions just keep leeching and don't help us - Joshua's Ubuntu like OS, if you are mentioning, was the only example, we shared the burden, but users and some Chinese capitalists greed managed to destroy him  

 

54 minutes ago, ZenSay said:

is for me

 

For everyone. But there are at least two different hardware revisions of T6. Which means, the problem could manifest on one revision but not on another. Sadly some of such things are beyond repair. (I am speculating, as this is second hand info)

While you wait that we resolve other 1000 problems, so we can start working on this one, try to be help out https://forum.armbian.com/staffapplications/ 

 

1 hour ago, ZenSay said:

Any suggestions?

 

Helping us around general task such us reviewing the code, writing documentation, promoting the project, ... If you are asking us for the most demanding and expensive work, at least help us to "clean around the house" and other generic maintenance work. This way you will some day perhaps understand the severity of what you are asking. Rockchip and board vendor engineers were unable to fix this (or there is a HW design issue and problem can't be resolved).

On my NanoPi-R6C (also FriendlyElec) I also discovered that USB3 did not work when running  6.12.2-current-rockchip-rk3588

Updated and running:

Linux ranc 6.12.5-current-rockchip-rk3588 #5 SMP PREEMPT Sat Dec 14 19:04:17 UTC 2024 aarch64 GNU/Linux

 

Now the USB3 port works again. Speeds about 400MByte/s for USB3-SATA SSD, so fine.
 

 

And just rebooted with:

Linux ranc 6.1.84-vendor-rk35xx #1 SMP Sun Dec 15 09:25:38 UTC 2024 aarch64 GNU/Linux
 

Also working USB3.

Edited December 17, 20241 yr by eselarm

Igor:

 

Sorry, not trying to make unreasonable demands here, it just seemed that there was already a solution and that I was simply doing something wrong.

 

So if this is a general Rockchip (RK3588) issue then I guess it would be the same across all the different SBCs?

 

I'm a software developer so can absolutely try to work on this - or other issues. Will take a look!

 

1 hour ago, ZenSay said:

unreasonable

 

End users already have a huge depth toward developers. Every 3 months we make a summary what was accomplished https://docs.armbian.com/Release_Changelog/#v24111-2024-11-28 (example). And your demands / expectations never stops ... Think again. You / users "only" want that everything work, ignoring / not understanding the path to get there. And what is needed to keep things there - operational.

 

1 hour ago, ZenSay said:

it just seemed that there was already a solution and that I was simply doing something wrong.

 

Few ideas for thinking - regressions of low level functions (world we live in) happens all the time. In some cases boards HW design prevents to get to the satisfaction stability level or the software stack / driver was done sloppy and we don't have needed resources to get it right. When its down to software maintaining, most of people doesn't have luxury of time. We steal that time from our families, which is bad. We can only patch the driver even done terrible wrong. Very little people have the luxury of time and needed experiences to do things properly. That is serious problem and sadly common practice. And at the end of the day, open source developers are often irreplaceable and things collapses. There is new and new hardware out ...

 

One possible example (i don't remember details anymore) of this is Odroid C2, where we keep loosing USB support.

 

Then people are insulting us also to emails all the time - from last week:
 

Quote

"most of the builds are a full shit and you are waisting time resources.to give you an example: the odroid c2/bookworm minimal has no usb support"
fuckers

 

Within years, nobody managed to stabilize this properly. Sometimes it works, sometimes it doesn't. There are plenty of similar cases you are experiencing across other boards containing other SoCs. In theory, SoC defines how low level serial protocols will work, but board HW and SW is a complex field of engineering with lots of signals going around. And those signals can interfere between each other. This means board integrator, especially if they are differing to far from reference design (which could also be done sloppy sometimes). if they just place things on PCB in a critical way (perfect storm), it becomes too costly or impossible to adopt software in order that function works. When kernel is upgraded, things usually broke down again - and you want most recent kernel all the time for security and other general updates. Sometimes they even forget to wire something or similar HW level mistakes. Like I said, this is possible to happened with T6 v1. There is a second revision of this board - with a reason. Something is wrong with v1 (I don't know exactly what).

 

1 hour ago, ZenSay said:

I'm a software developer

 

This problem is in the domain of embedded Linux software developer specialized in communication protocols. If you are not that, help with what you master or be open for long-term learning. That is what I am trying to convey.

Igor:

 

Thanks for the info, I will order a new T6 which I assume will be the v2 revision. Perhaps things will work out better - let's see.