[OOB mbed-os-5.4] Run time error (stack overflow) on NCS36510

[mray190]

Target: NCS36510 Toolchain: Any Error: Stack overflow. Adjusting the stack and heap sizes prints out varying debug messages but regardless of what size they get set to, it will always fail

Attempted mbed_app.json config:

    "target_overrides": {
        "*": {
            "target.features_add": ["NANOSTACK", "LOWPAN_HOST", "COMMON_PAL"],
            "mbed-mesh-api.6lowpan-nd-panid-filter": "0xffff",            
            "mbed-mesh-api.6lowpan-nd-channel-page": 0,
            "mbed-mesh-api.6lowpan-nd-channel": 12,
            "mbed-mesh-api.6lowpan-nd-device-type": "NET_6LOWPAN_HOST",
            "mbed-mesh-api.6lowpan-nd-channel-mask": "(1<<12)",
            "mbed-mesh-api.thread-config-panid": "0x0700",
            "mbed-mesh-api.thread-master-key": "{0x10, 0x11, 0x22, 0x33, 0x44, 0x55, 0x66, 0x77, 0x88, 0x99, 0xaa, 0xbb, 0xcc, 0xdd, 0xee, 0xff}",
            "mbed-mesh-api.thread-config-channel": 22,
            "mbed-mesh-api.heap-size": 8000,
            "nanostack-hal.event_loop_thread_stack_size": 10000,
            "mbed-trace.enable": true,
            "platform.stdio-convert-newlines": true,
            "platform.stdio-baud-rate": 115200
        }
    }

[MarceloSalazar]

@SeppoTakalo could you please have a look?

[0xc0170]

bump

[SeppoTakalo]

I cannot replicate this.

Without mbed_app.json modifications

Connecting...
connected. IP = 2001:999:10:bbf4:200::

With supplied mbed_app.json

Connecting...
[DBG ][nslp]: connect()
[DBG ][evlp]: event_loop_thread
[DBG ][m6LND]: Link-layer security NOT enabled.
[DBG ][m6LND]: Channel: 12
[DBG ][m6LND]: Channel page: 0
[DBG ][m6LND]: Channel mask: 4096
[INFO][m6LND]: Start 6LoWPAN ND Bootstrap
[DBG ][m6LND]: app_parse_network_event() 0
[INFO][m6LND]: 6LoWPAN ND bootstrap ready
[DBG ][m6LND]: ND Access Point: 2001:999:10:bbf4:0:ff:fe00:8454
[DBG ][m6LND]: ND Prefix 64: 20:01:09:99:00:10:bb:f4
[DBG ][m6LND]: GP IPv6: 2001:999:10:bbf4:200::
[DBG ][m6LND]: MAC 16-bit: ff:ff
[DBG ][m6LND]: PAN ID: 06:91
[DBG ][m6LND]: MAC 64-bit: 00:00:00:00:00:00:00:00
[DBG ][m6LND]: IID (Based on MAC 64-bit address): 02:00:00:00:00:00:00:00
[DBG ][m6LND]: Channel: 12
[DBG ][nslp]: getOwnIpAddress()
[DBG ][nslp]: getOwnIpAddress()
connected. IP = 2001:999:10:bbf4:200::

NOTE: Device I'm using does not have MAC address. Therefore the IPv6 address ends in :: (all zeroes)

Can you provide more information. Back trace from stack perhaps. What compiler were used? I'm using

$ arm-none-eabi-gcc --version
arm-none-eabi-gcc (GNU Tools for ARM Embedded Processors) 4.9.3 20150529 (release) [ARM/embedded-4_9-branch revision 227977]

[mray190]

Compiler used:

$ arm-none-eabi-gcc --version
arm-none-eabi-gcc.exe (GNU Tools for ARM Embedded Processors) 5.4.1 20160919 (release) [ARM/embedded-5-branch revision 240496]

Utilizes mbed-os 5.4 (commit: https://github.com/ARMmbed/mbed-os/commit/3a27568a505bbc0bb8eeb973109d3d39ce823d1c)

Used the OOB test branch: https://github.com/ARMmbed/mbed-os-example-mesh-minimal/tree/oob_test_mbed-os-5.4

If you used all of these things to run what we saw above, Ill go back and grab some stack traces

[MarceloSalazar]

@mray19027 could you please try arm-none-eabi-gcc 4.9.3 (the version that it's officially supported)?

[SeppoTakalo]

I also tried again with arm-none-eabi-gcc (GNU Tools for ARM Embedded Processors) 5.4.1 20160919

I still cannot replicate this.

[DBG ][m6LND]: Restart bootstrap
[DBG ][m6LND]: Link-layer security NOT enabled.
[DBG ][m6LND]: Channel: 12
[DBG ][m6LND]: Channel page: 0
[DBG ][m6LND]: Channel mask: 4096
[INFO][m6LND]: Start 6LoWPAN ND Bootstrap
[DBG ][m6LND]: app_parse_network_event() 3
[DBG ][m6LND]: Link Layer Scan Fail: No Beacons

NOTE: My border router is not running now.

[SeppoTakalo]

Did you flash the device by copying the .bin image to device, or did you try to flash using GDB?

What debugger did you use to verify the Stack overflow?

I believe flashing through GDB's load command does not work.

[MarceloSalazar]

@mray19027 please help to provide more info

[mray190]

@SeppoTakalo I flashed the device by copying over the .bin image to the device

Currently having issues with our border router but will get additional information as soon as possible

@maclobdell Could use your help

[mray190]

Conclusion: All of our boards in lab had incorrect Flash settings - actually no settings at all as the flash memory was empty.

This caused all of the radio stacks to overflow because the rxRam variable was incrementing too far. See:

        /* Initialize frame status */
        for (uint8_t i=0; i < length; i++) {
            PHYPAYLOAD[i] = *rxRam++;
        }

in NanostackRfPhyNcs36510.cpp

Flash settings were restored on all of our devices and no further action is required.

[SeppoTakalo]

@mray19027 Thanks for investing this.