ESP32 TheThingsNetwork.h Causing Crashes

[kevin192291]

Hi All,
I have been having a massive headache trying to get TheThingsNetwork.h (https://github.com/TheThingsNetwork/arduino-device-lib) working on the ESP32, I really want to use the ESP32 for my project.
When I attempt to get this up and running on my esp32, it will crash, and reboot, I have attempted to use the ESP Exception Decoder, but it isn't working, or I am not entering the correct info.
I have found another library (https://github.com/jpmeijers/RN2483-Arduino-Library) that does at very least get a boot, but has a restriction of 11 bytes that may be sent

Here is the most simple example I have tried:

Code: [Select all] [Expand/Collapse]

1. #include <TheThingsNetwork.h>
2.  
3. #define loraSerial Serial2
4. #define debugSerial Serial
5.  
6. // Replace REPLACE_ME with TTN_FP_EU868 or TTN_FP_US915
7. #define freqPlan TTN_FP_US915
8.  
9. TheThingsNetwork ttn(loraSerial, debugSerial, freqPlan);
10.  
11. void setup()
12. {
13.   loraSerial.begin(57600);
14.   debugSerial.begin(9600);
15. }
16.  
17. void loop()
18. {
19.   debugSerial.println("Device Information");
20.   debugSerial.println();
21.   ttn.showStatus();
22.   debugSerial.println();
23.   debugSerial.println("Use the EUI to register the device for OTAA");
24.   debugSerial.println("-------------------------------------------");
25.   debugSerial.println();
26.  
27.   delay(10000);
28. }

GeSHi © Codebox Plus Extension

The TTN library works great with ATMEL chips, I have tried it on a few different ones, Just not on the ESP line...
Any help at all would be greatly appreciated, and if you have a way to get this up and working, I can throw you a few
bucks over paypal or something.

Thanks,
-Kevin

[PeterR]

I'm not an arduino expert but in IDF 9/10 weird issues are typically down to stack or rouge pointers.
You say that backtrace does not work. A quick sanity check is to add (char *)0 = 1 in/after initialisation. If you get a reasonable backtrace then that adds weight to stack corruption.
Something to think about is that variables are not automatically 0 initialised in ESP. I have had a few 'working' arduino libraries which fail because of initialisation or lack off. ESP are correct btw. EDIT: Just less forgiving!
The fact that you do not get a reasonable backtrace suggests too small a stack. Rouge pointer is also possible, see above.
Past that my suggestion would be to add printf/sleep & so isolate the fault. Removing the connection (cable/wifi) would help show if initialisation or frame processing.