How can I debug a float-point code

[xtting]

Hi,

I always see HardFault when I debug a float point code. But it can run smoothly in run-time.

Firstly, I use code Style-1.

float val
val = (int32_t)NRF_ADC->RESULT;

it go into HardFault directly.

Then I split the code into Stype-2

float f_val
int i_val

i_val = (int32_t)NRF_ADC->RESULT;
f_val = i_val;

it will fault in f_val = i_val;

Then, I changed code to Style-3

float f_val
int i_val

i_val = (int32_t)NRF_ADC->RESULT;
f_val = i_val + 0.0;

it can go smoothly.

How can I use the first style directly when debug, for I dont like to change all my code into style-3. It will waste so many time, and is not a good code style.

Thanks,

[xtting]

I guess may the debugger has something wrong when call the float point liburay, but how can I config the Keil-472 correctly? I used the official Demo code to develop my project, without any config changed.

[monpetit]

Hi,

You'd better show your development environments to get answers. Device name, compiler, debugger, SDK version, whether to use the SoftDevice or not, and so on.

[xtting]

Hi,

Thanks for your answer.

Device: 51822 IDE: Keil-472 SDK Version: 10-alpha I used soft-device.

And the debugger and complier is what the SDK-10-Demo used by default, for I don't so familiar about how to change config I don't try to any change at all.

[Susheel Nuguru]

nRF51 does not have FPU so your compiler is generating soft FPU instructions. What are the project settings you have for Target->MicroLIB, target->use Float Hardware?