Asset Tracker Program hangs before main on Thingy91

[ImagiSam]

Hi,

We are having trouble getting the Thingy91 firmware running as is. We have achieved this in the past but I had made too many modifications and I decided to scrap it and start over. But I didn't realise that I could've just reused the cmake file so I uninstalled and reinstalled the SDK. 

We are trying to run the Asset Tracker project from the 1.3.0 toolchain.

So what happens is that when I build and debug instead of going to main() it stops here:

Then when it runs it hangs here:

I have tried to modify it to change the floating point configuration but I changed it back. I didn't try running it before the modifications

[Simon]

I see the exact same behaviour as you, and am able to debug without problems. First I would recommend you to set

CONFIG_DEBUG_OPTIMIZATIONS=y in ncs\v1.3.0\nrf\applications\asset_tracker\prj.conf to optimize the code for debugging, then reopen the project and set the nRF Connect Options again (if you only build it, the changes in prj.conf will not be taken into effect).

Regarding the first image, the one where it stops at arch_kernel_init(), in my case it always stops there and always has. I'm not sure why this is, but I can look into it. However, it has never affected the application in any way. After I continue the debugging, the program runs as expected.

Regarding the second image, this is expected behaviour. If I click the break sign while the application is running I will see that the program is there as well. You can see that you're in cpu_idle and that wfi has just been run, which means the idle thread is running, the chip is sleeping, and there is currently nothing happening. Every once in a while the device will wake up from sleep and handle different kinds of interrupts, like cellular-related stuff, timers, callback handler for peripherals, but when that is handled it will go back to the idle thread an into sleep again. Take a look at this ticket about the System ON sleep and the idle thread.

Could you check that the Asset Tracker application is actually working by checking the serial output and if it connects to your account on https://nrfcloud.com?

By the way, I would urge you to open public cases instead of private ones, if you're not sharing any confidential information. Then other people may benefit from the answers given here and you may get answers from other DevZone users as well

Best regards,
Simon

[ImagiSam]

Ok, I'll look more into this. This didn't happen to me before and it always started at main() and stopped there. So, I thought that this might be unexpected behaviour. Currently the sim card is out of credit so I haven't been connecting from the cloud but I can inspect the serial output and get back to you.

Regarding the public vs. private I figured it's easier to get it sorted out then convert to public if there's nothing confidential

[Simon]

Sounds good, look forward to hearing from you.

You could also simply put a breakpoint in main() to verify that it goes there.

[ImagiSam]

So, I figured out the problem. when I added an extra source file to the cmake file (that's how I add extra files, correct?) it was deleting the main.c file and this was because I neglected to add a proper CMakeLists.txt file to the directory. So now it works but I have 2 questions.

1. First we think that perhaps running from the code has a different sampling frequency from the other version he ran (out of the box). IS this possible?
2. When he was collecting data he was able to obtain the air_quality value but for me that code is absent. Is it possible to obtain this segment or get help on how to fix it?
   env_sensors.c
   int env_sensors_get_air_quality(env_sensor_data_t *sensor_data)
   {
   return -1;
   }
   
   this code is called from env_data_send(void) in main.c

[ImagiSam]

I tried to dig into the configuration and adjust the settings for the BSEC library but I get this:

with this reason: