Hardfault in nrf_drv_clock_hfclk_request() with S112 softdevice/nRF52810

[Gerald Heinig]

I have the following trivial program which has almost identical functionality to the connectivity/ble_connectivity/pca10040e example:

int main(void)
{
    uint32_t err_code =0;
    
    err_code = nrf_drv_clock_init();
    if ((err_code != NRF_SUCCESS) &&
        (err_code != NRF_ERROR_MODULE_ALREADY_INITIALIZED))
    {
        APP_ERROR_CHECK(err_code);
    }
    
    nrf_drv_clock_hfclk_request(NULL);broken.zip

I am using the S112 softdevice, which I have programmed to the nRF52 Development Board (pca10040) which has an nRF52832 chip emulating an nRF52810.

I have used the ble_connectivity/pca10040e project as a template and when I compile and run the ble_connectivity project, it runs perfectly.

I have also taken the ble_connectivity project and reduced it to the bare essentials, so that it is identical to my trivial project.

ble_connectivity works, mine hardfaults.

• The sdk_config.h is identical
• The linker configuration file is identical
• After the stripdown of the ble_connectivity project, the main.c file in both projects are identical.
• The compiler settings are identical (no optimisation).

I've run out of things to try. Please help.

[Gerald Heinig]

Hi Susheel,

yes, I have set SOFTDEVICE_PRESENT.

I will get the address of the faulting instruction, check its context and get back to you.

[Gerald Heinig]

Address of the faulting instruction is 0xa64. In terms of context, do you mean interrupt/normal context or the calling function?

The calling function is sdh_stack_observers1().

[Susheel Nuguru]

0Xa64 is not the faulting address but the address of the hardfault hander inside the MBR (within the softdevice).

I will try to replicate your fault on my desk and try to find out what caused the hardfault.

[Gerald Heinig]

Hello Susheel,

I found the problem:

I had overlooked extra options in the debugger (Options->Debugger->Extra options).

The following line was set:

--drv_vector_table_base=0x0

Please close this issue.

Thank you very much for your prompt response!