search

markqvist / Reticulum

The cryptography-based networking stack for building unstoppable networks with LoRa, Packet Radio, WiFi and everything in between.
https://reticulum.network
MIT License
2k stars 124 forks source link

Wrong FW reported in RNodeConf #497

Closed faragher closed 5 months ago

faragher commented 5 months ago

Describe the Bug

RNodeConf reports incorrect FW version. This also prevents --auto-intstall upgrades

To Reproduce

Connect to RNode using RNodeConf (tested on 1.6.1 FW version 1.59). Reports version is, at this time, 1.70 and up to date.

Expected Behavior

Reports accurate FW Version

Logs & Screenshots

OLED display reports FW 1.59.

Issue on Linux: 

rnodeconf 2.1.3
...
[00:47:37] Opening serial port /dev/ttyACM0...
[00:47:40] Device connected
[00:47:40] Current firmware version: 1.70
[00:47:40] Reading EEPROM...
[00:47:41] EEPROM checksum correct
[00:47:41] Device signature validated

Windows reports:

rnodeconf 2.1.3
...
[18:43:22] Opening serial port com15...
[18:43:26] Device connected
[18:43:26] Current firmware version: 1.59
[18:43:26] Reading EEPROM...
[18:43:26] EEPROM checksum correct
[18:43:26] Device signature validation failed

System Information See above

Additional context Issue discovered when tracking down Bluetooth bug. Debugging between the two is overlapping.

markqvist commented 5 months ago

Ok, pardon the seemingly obvious question here, but I gotta ask, since I've done this myself a few times. Can you confirm that this is not occurring because you actually have another RNode connected to the system, that is using /dev/ttyACM0, and that the RNode you are intending to communicate with is on another port? I've done this before, where I completely forgot I had another RNode going off on a long USB cable somewhere, and was seriously scratching my head for a good while.

If that is not the case, something bordering on spooky is going on, because I currently have literally no clue how it would get a firmware version value of 1.70 from the device. It's pulling it via a command from the RNode device, and the value being read in the actual hardware is hardcoded into the compiled firmware.

So if it is indeed an issue with rnodeconf or the firmware itself, this is probably going to be an interesting one to hunt down.

faragher commented 5 months ago

An excellent question, however, it's the only RNode. The ttyACM0 was the only ttyACM* device, it was the only one that was picked up on the autoinstall, and the only other USB devices attached are a camera and three SDRs, none of which could be a false positive.

I've got a bout of insomnia, so I'll see if I can find anything.

faragher commented 5 months ago

Ahem. Never mind. I won't admit what happened, but it wasn't that I missed an RNode attached. All I can say is I do almost all my work through SSH.

markqvist commented 5 months ago

Heh, we've all been there :) I once managed to drop an entire database. On the production server - not the local development copy. That required some fast thinking.

faragher commented 5 months ago

Oof. I've done some DOS undelete before, but that's a different level of "Oh no."