FreeRTOS + Softdevice HARDFAULT:ERROR:Hardfault PC:4

[Mehdi]

Hello,

I'm experiencing a hardware fault followed by a reset when running the nRF5_SDK_12.1.0_0d23e2a ble_app_hrs_freertos with the following configuration :

• PCA10040 V1.1.0 with nRF52832
• In sdk_config.h, I set NRF_LOG_ENABLED 1 and NRF_LOG_DEFERRED 0
• Eclipse with GCC 5.4.3
• Without optimization in Makefile ( -O0).

The error occurs very rapidly if I press button 2 ( two or three times) or every few seconds if nothing is pressed.

The logs shown on the terminal are :

SDH:INFO:sd_ble_enable: RAM START at 0x20002128

APP:INFO:Fast Adverstising

HARDFAULT:ERROR:Hardfault PC:4"

I've been digging into all the questions about freertos in the forum and I did not find a resolution. I was working on an already modified version of the example, put the DEBUG_NRF flag in Makefile, had the debugger stop and show vPortStartFirstTask() at port.c .... but I found no clue. I got back to a clean SDK version and retried the same configuration ( here before) and the problem was the same.

With optimization -O3, it does not appear on the example ( but still on my modified code).

Thank you for your interest.

[www.FreeRTOS.org]

I would recommend trying to locate the instruction causing the hard fault. Information on how this can be done can be found here www.freertos.org/Debugging-Hard-Faults-On-Cortex-M-Microcontrollers.html

[Susheel Nuguru]

Normally you cannot execute instruction at address 0x4 as this is where vector table reset handlers address is located(unless you used VTOR). Seems like you have function pointer pointing to wrong address or a POP instruction gone wrong by invalid stack pointer. You say this is reproducible on clean SDK? if so please give clear instructions to reproduce it.

[Mehdi]

Yes, It is reproducible on a clear sdk with the configuration given :

• No optimization ( -O0)
• NRF_LOG_ENABLED 1 and NRF_LOG_DEFERRED 0
• GCC and eclipse.
• PCA10040, S132, SDK12.1
• ble_app_hrs_freertos

Thanx

[Mehdi]

I tried that but Nordic has SVC in starter file and it was not hit when debugging. It should be something with the interrupts and the softdevice that is a Nordic recipe.

[Susheel Nuguru]

Not sure if this is related, but i found a bug in the default sdk_config.h file. Are you using UART as debugging out? if so Please change the below value

#ifndef NRF_LOG_BACKEND_SERIAL_UART_BAUDRATE
//#define NRF_LOG_BACKEND_SERIAL_UART_BAUDRATE 30801920
#define NRF_LOG_BACKEND_SERIAL_UART_BAUDRATE 30924800
#endif