about '__attribute__((section(".non_init"))) ' problem.

[etomatoes]

Hi ,

   SDK:  nRF5_SDK_15.2.0_9412b96

   IDE:   SEGGER Embedded Studio V4.12(64bit)

   C language standard: gnu99

    code:

    u32 nums_hf  __attribute__((section(".non_init")));

    

int main(void)
{
    nums_hf=0;
    log_init();

    while(1){

        index=events_get_next_event(&flag);

       ......

       output_events_device_start_running(event);

       .........

}

void output_events_device_start_running(S_EVENTS_LIST *p_s_events)
{
.......
nums_hf++;
NRF_LOG_INFO("start running ......(%d)\n\r",nums_hf);//s_comm.u_comm_process.s_process.u_record_points.record_points);
......

}

output log:

<info> app: start running ......(1)

<info> app: start running ......(19792205)

<info> app: start running ......(19792206)

‘’nums_hf‘’ Used in only one place.

If change ‘’nums_hf‘’ defined as “u32 nums_hf”. It works will. 

Please help me to see what is going on.

[Vidar Berg]

Hi, 

Please try to clarify what the desired behavior is. I see that you are placing the  nums_hf variable in .non_init but also initializing it to '0' in main().  

[etomatoes]

Hi ,

The function I want to implement is as follows:

When a non-power-down reset (such as a watchdog reset) occurs, this value can be kept as it is.

Use “__attribute__((section(".non_init")));”  can achieve the function I want, but there is a problem like the above.

[Vidar Berg]

But how can it be retained through a reset when you initialize it to zero at the beginning of main? Also not sure what the "problem above" is, are you referring to the output log? 

Please note that RAM is not guaranteed to be retained through WD reset, see the Power chapter of the product specification for more details .  

[etomatoes]

Hi，

I see,“nums_hf=0;” is a mistake.

int main(void)
{
    //nums_hf=0;
    log_init();

    ....

}

Thank you.I will see the Power chapter for detail.

[awneil]

Vidar Berg said:

Please note that RAM is not guaranteed to be retained through WD reset, see the Power chapter of the product specification for more details

So what does an 'x' in Table 18.8, "Reset behaviour", actually mean?

[Vidar Berg]

The 'x' indicates that the RAM content may become corrupted. In other words, we do not guarantee that the content is correctly retained. It's not reset as stated in the note below table 18.8.