Stack overflow on Wi-SUN QUECTEL EC2x initialization

[felipeLeast]

debug.error.txt

Description of defect
Intermitent stack overflow when runing default code from master on same configuration as describe on README file. When I start runing the code some times I get the following stack overflow right after initiating the cellular module. It takes a few resets before I the code start running, but eventually does.

Step to reproduce:
I just compile the master code from nanostack-border-router to F64K, and I let it run, eventually you will get that stack overflow

Target(s) affected by this defect ?
F64K with S2LP shield and EC25 demo board connected with uart, power key and reset pins to the device.

Toolchain(s) (name and version) displaying this defect ?
ARM_GCC

What version of Mbed-os are you using (tag or sha) ?
mbed-os-5.14.2

What version(s) of tools are you using. List all that apply (E.g. mbed-cli)
N/A

Can anyone shed a light on this issue please?

Issue request type

[ ] Question
[ ] Enhancement
[x] Bug

[ciarmcom]

Internal Jira reference: https://jira.arm.com/browse/MBOTRIAGE-2404

[0xc0170]

cc @ARMmbed/mbed-os-wan

[AnttiKauppila]

cc @ARMmbed/mbed-os-mesh
Has this combination ever been tested? I think there are two IP stacks in use simultaneously: Nanostack for Wi-SUN and LwIP stack (PPP mode) for Backhaul Cellular. From log it looks like when CGDATA="PPP" gets called the device runs out of memory (Stack overflow)

[felipeLeast]

Hi @AnttiKauppila. I believe it has. It is something develop for that purpouse. Follows the link for the application that uses it. https://github.com/ARMmbed/nanostack-border-router

If you see in the configs folder there's a config for Wisun_Stm_s2lp_RF_cell.json which does exactly that.

But if you think is just a stack overflow because is running out of memory I'll try to increase my heap memory to see if the stack overflow continious.

But I still don't understand why it only happens sometimes.

[mikaleppanen]

Cannot repeat the problem here locally, but I detected that some of the PPP debug traces were enabled also without debug flags. This resulted that PPP used more stack than needed.
Corrected that on:

#12195

@felipeLeast Could you check whether the stack overflow happens anymore with the above modification.

[mikter]

Closed as issue is resolved