Loading...
 

Zephyr project on STM32

   Zephyr Workbench, a VSCode extension to manage Zephyr on STM32.
It enables users to easily create, develop, and debug Zephyr applications.
Main features:
  • Install host dependencies.
  • Import toolchain and SDK.
  • Create, configure, build and manage apps.
  • Debug STM32.
You can directly download it from the VSCode marketplace
For more details, visit the Zephyr Workbench

System Workbench for STM32

Forum List Topic List

[ prev topic ]  

Issue snprintf floats

Posted by ArminOonk on 2019-05-09 16:23

Hi,
I am having an issue with floats and snprintf. I created a small demo program using CubeMX 5.2 and put it on bitbucket: https://bitbucket.org/arminoonk/printf-float-issue/src/master/Question

I have added “-u _printf_float” to the GCC Linker flags. Complete linker flags are: “-specs=nosys.specs -specs=nano.specs -u _printf_float”

In the default task I have added a small blinking led. In a separate task I have the following code:

char temp33;
snprintf(temp, 33, “%f”, 6.66)

When snprintf is run a hard fault handler occurs. The stack size for the snprintf task is 2048 words which should be enough.
I have created code using snprintf before using cubemx. This worked beautifully but now suddenly it stopped.

I am using system workbench with the latest software updates installed.
System Workbench

Is there a known issue with the newer releases of cubeMX/system workbench? Is there a work around?

Kind regards,
Armin

Reads: 7282
Link
Posted by ArminOonk on 2019-05-16 16:01

I have been looking further into this.
When the hardfault occurs the PC points to _Bfree
Hardfault

I have been looking into what the implementation of snprintf does. It seems it is using some dynamic memory using malloc().

Way back I found some resource telling me to use pvPortMalloc and vPortFree instead of the normal malloc and free. I am using FreeRTOS with heap4.

Could it be the call to malloc causing problems because I am using FreeRTOS?

Link
Posted by ArminOonk on 2019-05-21 15:13

The answer to this is, no it does not always work. I will continue this monologue.

I have used the solution from: http://www.openstm32.org/forumthread353#threadId354Question
It appears to be an incorrect _sbrk implementation

I have update the example project on bitbucket

_sbrk implementation

caddr_t _sbrk(int incr) {

extern char end asm(“end”);
extern char _min_heap_end asm(“_min_heap_end”);
static char *heap_end;
char *prev_heap_end;

if (heap_end == 0) {
heap_end = &end;
}

prev_heap_end = heap_end;
if (prev_heap_end + incr > &_min_heap_end) {
if (heap_end + incr > stack_ptr) {
errno = ENOMEM;
return (caddr_t) -1;
}
}
heap_end += incr;
return (caddr_t) prev_heap_end;

}



Update to linker file:

/* User_heap_stack section, used to check that there is enough RAM left */
._user_heap_stack :
{
. = ALIGN(4);
PROVIDE ( end = . );
PROVIDE ( _end = . );
. = . + _Min_Heap_Size;
PROVIDE ( _min_heap_end = . );
. = . + _Min_Stack_Size;
. = ALIGN(4);
} >RAM

Link