DWT not running on STM32L5 with Debugger NOT connected

[MPetr.3]

Hello community,

I implemented a microsecond delay on a STM32L5 microprocessor using the "Data watchpoint and trace unit" (DWT). I am initialising the unit and count the DWT->CYCCNT down until my delay has passed.

This works totally fine when the Debugger (STLink) is connected. But if I run the programm without the debugger connected, than the DWT->CYCCNT is not increasing.

My initialisation code is the following:

uint32_t TM_DELAY_Init(void) {
    uint32_t c;
	
    /* Enable TRC */
    CoreDebug->DEMCR &= ~0x01000000;
    CoreDebug->DEMCR |=  0x01000000;
 
    /* Enable counter */
    DWT->CTRL &= ~0x00000001;
    DWT->CTRL |=  0x00000001;
	
    /* Reset counter */
    DWT->CYCCNT = 0;
	
    /* Check if DWT has started */
    c = DWT->CYCCNT;
	
    /* 2 dummys */
    __ASM volatile ("NOP");
    __ASM volatile ("NOP");
 
    /* Return difference, if result is zero, DWT has not started */
    return (DWT->CYCCNT - c);
}

My delay function is the following:

__STATIC_INLINE void Delay(__IO uint32_t micros) {
	uint32_t start = DWT->CYCCNT;
	
	/* Go to number of cycles for system */
	micros *= (HAL_RCC_GetHCLKFreq() / 1000000);
 
	/* Delay till end */
	while ((DWT->CYCCNT - start) < micros); // error
	// THE CODE DOENST REACH BELOW THE WHILE
}

The program .elf file is exactly the same with and without debugger.

I have read multiple online sources and noticed, that some sources also state, that you have to unlock the Lock-Access-Register DWT->LAR to make the DWT run without a debugger connected. This LAR-Register doesnt exsit for the L5-series. Adding the code

volatile uint32_t *LAR  = (uint32_t *) 0xE0001FB0;   // <-- added lock access register
*LAR = 0xC5ACCE55;                // <-- added unlock access to DWT (ITM, etc.)registers

doesn't help.

I would appreciate any help.

Thanks in advance.