EDUP EP-RT2983 (AX1800) with netis N6 firmware - No LEDs

Hardware Questions and Recommendations
1

I flashed an EDUP EP-RT2983 with the netis N6 firmware, and it works with the given caveat that the ports are labelled LAN2-LAN4.

The only thing that seems to be missing is that the LEDs don’t light with the triggers. There’s no mention of that on the N6 page, does anyone have any suggestions on how to add support for the LEDs?

2

Good that it boots at all....
LEDs are of low priority - is wifi MAC and tx power correct (like in iw phy)?

3

Not 100% sure it's the same device...

4

Now you mention it, the MAC address did change - I’m using it as an AP and it didn’t get the IP address I’d allocated when it was stock firmware and I had to change my LAN’s DHCP allocation for the new address.

I don’t think the power is reported properly either:

image
image980×364 45.2 KB

5

Those issues do look the same.

6

I guess it’s not as simple as changing the LED definitions, because:

  • That patch mapped one of the LEDs (led-1) to gpio13
  • mt7621_netis_n6.dts already has led-1 on gpio13 for LED_FUNCTION_WPS
  • Adding an LED action for green:wps default-on doesn’t produce a light.
7

You have to explore gpio-s first - expose them, observe values, flip values, check which is right for led.

8

I installed gpiod-tools and started gpioset -c gpiochip1 -p 1s 20=1 etc but all I got was “Resource busy” on the ones I’d tried to associate with network activity etc, or hangs on the lines that weren’t being used by anything.

I suspect this is beyond my reverse-engineering abilities, and I’ll just live with the dead looking panel.

9

go to sys/class/gpio

echo all numbers to expose
read them all
and again
blinking ones will flip
ones at 1 should be turnes to zero - up to you - one per second with mobile camera filming or half at once like bisecting.
usually functional gpios are at start of block
now at one point wifi or usb or ethernet goes off
obviously dont touch those again.
then set them to one (except those being used by smth else...) using the previous algorithm.

good luck (and nice crashes)

10

I’m probably not doing it right (especially since I didn’t get any crashes).

I removed all the LED rules, and now the wifi light is flashing with what I could believe is traffic on one of the two phy interfaces).

With AI assistance, I arrived at this script, but no light-flashing or crashing happened, just the occasional “ash: write error: Resource busy”

for i in $(seq 512 608); do
  echo "GPIO $i"
  echo $i > /sys/class/gpio/export
  if [ -d /sys/class/gpio/gpio$i ]; then
    echo out > /sys/class/gpio/gpio$i/direction
    echo 1 > /sys/class/gpio/gpio$i/value
    echo "GPIO $i set to 1 - check LEDs"
    sleep 1
    echo 0 > /sys/class/gpio/gpio$i/value
    echo "GPIO $i set to 0 - check LEDs"
    sleep 1
  fi
done
11

Where did you get base number 512?

12

/sys/class/gpio contained gpiochip512, gpiochip544, gpiochip576

1 Like
13

That sounds about right.... (the for ... seq loop is OK)

Lets start more structured approach

Boot the device
Step 1 (encouraging you to try tu understand underlying logic, knowingly withholding details on howto)
EXPORT all gpios (nothing happened)

Step 2 : read all values
echo $i : cat /sys/class/gpio$i/value

Step 3 : set all pipes to output - then values that were 1 to 0 (use list of numbers ipo seq)

Step 4 : is it off ? now reboot, try half of those and so on until you find actual thing switching it off

14

Reading the value of all the exported gpios gives 0 for each of them.

I don’t seem to be able to export 516, or 525-531, I get ‘write error: Resource busy” ( I get the same message if I try to export a gpio that’s already been exported, but 516/525-531 don’t have entries in /sys/class/gpio).

15

I have submitted a pull request (#22197) for this device.

16

Hi @untilscour , I built and flashed from main + your PR and it worked almost perfectly (after forcing the upgrade because the previous firmware was for the Netis N6).

I say almost because the WAN light doesn’t work - the Power, WIFI, LAN1-3 all light up correctly but WAN doesn’t (although the WAN socket itself works). I’m not sure how I would test the MESH light.

Thanks for making this PR, it’s a great advance from using the N6 version!

2 Likes
17

Thanks for the testing and feedback!

The WAN light works for me on 25.12.0-rc4. I'm not sure why it's not working for you on snapshot (main branch). Did you discard all settings when upgrading?

Stable v25.12.0 has recently been tagged and should be announced in a few days. In the meantime I have forked it and applied the PR here. If you would like to test it you can clone my fork and checkout the branch v25.12.0-edup_ep-rt2983 to build and test.

I haven't had time to test mesh (WPS) light and button, it would require installing wpad and hostapd-utils and adding the button to wifi settings.

18

Ah, no I didn’t delete the settings because it took me too long to get the dhcp, access points and routing working correctly (I’m still a networking novice really).

Since you’ve tested and found it ok, I’m happy to trust that I’ve just messed something up during my hacking about earlier in this thread.

19

In that case you could delete the existing WAN LED setting at LuCI, System, LED Configuration and then add again according to the screenshot:

847×422 17.4 KB

And then reboot.

Or alternatively use uci to delete the existing system.led_wan properties:

uci delete system.led_wan
uci commit system

and then add the settings back again:

uci set system.led_wan=led
uci set system.led_wan.name='wan'
uci set system.led_wan.sysfs='green:wan'
uci set system.led_wan.trigger='netdev'
uci set system.led_wan.mode='link tx rx'
uci set system.led_wan.dev='wan'
uci commit system
reboot

If the WAN LED still doesn't work then it will have to wait until your next settings reset, perhaps at the next network reconfiguration.

20

That worked, thanks so much!

1 Like