Just got a rock64 board and tried the armbian testing image. Is ethernet working out-of-the-box with this image? My ethernet isn't working. Also the normal password didn't work (root / 1234). Would be nice to have some infos about that.

18 minutes ago, Goerlitzer said:

Also the normal password didn't work (root / 1234).

To my knowledge, root 1234 should work on every board. That's often a sign that something on your system is corrupt. Did you follow the getting started recommendations? (e.g. etcher instead of dd/win32, reliable SD card & psu).

2 hours ago, Goerlitzer said:

Is ethernet working out-of-the-box with this image?

 

AFAIK not. Please see here for a workaround and progress: https://github.com/armbian/build/commit/d5c6bf69db29fd8cd0b929aa15d7fc056cb02441#commitcomment-23724640

 

 

18 minutes ago, tkaiser said:

AFAIK not. Please see here for a workaround and progress: https://github.com/armbian/build/commit/d5c6bf69db29fd8cd0b929aa15d7fc056cb02441#commitcomment-23724640

AFAIR I heard about different issues (not related to tx/rx offloading) on the production boards.

5 minutes ago, zador.blood.stained said:

I heard about different issues (not related to tx/rx offloading) on the production boards.

 

Yes, ayufan said he had problems with Ethernet -- we talked today in #ROCK64 channel and I tested with 1GB production board (and offloading active): https://pastebin.com/aAzkEYEB (somewhat low performance in one direction but IMO not critical) and then next batch of boards got DRAM 'downgraded' which seems to cause problems for him.

 

Anyway: I tested with production board and without tx/rx offloading fix/workaround Ethernet stopped to work after ~30 seconds as it's the case with the pre-production boards too.

1 minute ago, tkaiser said:

Anyway: I tested with production board and without tx/rx offloading fix/workaround Ethernet stopped to work after ~30 seconds as it's the case with the pre-production boards too.

Unfortunately udev rules don't work since they are executed too early, so we may need a simple systemd service to execute the ethtool command at the right time.

6 minutes ago, zador.blood.stained said:

we may need a simple systemd service to execute the ethtool command at the right time.

 

Hmm... and what about simply re-using what's known to work: https://github.com/ayufan-rock64/linux-build/blob/master/package/root/etc/network/if-up.d/rock64-offload?

Just now, tkaiser said:

Hmm... and what about simply re-using what's known to work: https://github.com/ayufan-rock64/linux-build/blob/master/package/root/etc/network/if-up.d/rock64-offload?

Do these hooks work with NM too?

Thanks for the quick answers! I will try the workaround.

1 minute ago, zador.blood.stained said:

Do these hooks work with NM too?

 

At least as long as the interface is not controlled by NM it works (tested 20 days ago when I benchmarked my first Armbian build on ROCK64). So we might add a NM hook too but since their way of doing things isn't that persistent a systemd service might be the better idea...

 

What about this in the meantime?

diff --git a/packages/bsp/common/etc/init.d/armhwinfo b/packages/bsp/common/etc/init.d/armhwinfo
index 96345fa..1900085 100755
--- a/packages/bsp/common/etc/init.d/armhwinfo
+++ b/packages/bsp/common/etc/init.d/armhwinfo
@@ -253,6 +253,8 @@ prepare_board() {
                                        echo 7 >/sys/class/net/eth0/queues/rx-0/rps_cpus
                                        echo 32768 >/proc/sys/net/core/rps_sock_flow_entries
                                        echo 32768 >/sys/class/net/eth0/queues/rx-0/rps_flow_cnt
+                                       # preliminary rx/tx offloading workaround
+                                       /sbin/ethtool -K eth0 rx off tx off
                                        ;;
                                nanopim3) # dw-mci on cpu1, USB host on cpu2, GbE on cpu3, USB OTG on cpu4, video-codec on cpu5
                                        for i in $(awk -F':' '/dw-mci/{print $1}' </proc/interrupts | sed 's/\ //g'); do

 

Can someone from armbian please allow for the testing image the root login with the default password 1234? Thanks.

2 hours ago, Goerlitzer said:

root login with the default password 1234?

 

Initial root login on Armbian is always 1234 unless changed within build script.

Sorry, I wasn't clear. I can't login as user "root".

Please describe your situation in details, perhaps add screenshot or video.

 

Are you trying to log in via serial console? In case it's not added to /etc/securetty ... it's not possible to login in. In case of this board, this is quite possible since it's WIP.

No, I didn't use a serial console. I use a hdmi screen and a usb keyboard. I boot the board, the login screen appears and I try to login. But the system doesn't accept "root" as user, the error message is "incorrect login", the "password" prompt doesn't appear. It's possible to tape "rock64", than the password prompt appears.

Are you sure it is an Armbian image and not a ayufan one ?

This should fix the issue: https://github.com/armbian/build/commit/0b1e00c3942cb0578dd36518c5e6ef2c9ec5c693

And fixed image for Rock64 was just rebuilt.

https://dl.armbian.com/rock64/Ubuntu_xenial_default_nightly.7z

18 hours ago, Igor said:

And fixed image for Rock64 was just rebuilt.

https://dl.armbian.com/rock64/Ubuntu_xenial_default_nightly.7z

It works, thank you very much!

On 7.9.2017 at 6:19 PM, tkaiser said:

 

Hmm... and what about simply re-using what's known to work: https://github.com/ayufan-rock64/linux-build/blob/master/package/root/etc/network/if-up.d/rock64-offload?

Sorry, I'm a nob, but maybe it's not so complicated to use the script from ayufan. I put it in the directory like in his github repo (/etc/network/if-up.d/rock64-offload) and made it executable. But it's not working on my rock64 (Version 2.0, 2 GB). Can anyone help me? Or is there a possibility to solve the ethernet problem with one of the next nightly builds. That woud be great :-)