Problem with stm32f745 work

[kitty7c6]

After the firmwared (or power off-on) and after the reset the same program takes a different amount of time in ticks. Find it in my oun code. Then tried to create code in Cube and check it - same thing. Then I checked clocking settings and frequency - it was same and after power off-on and after reset (HSE 8MHz, Sys 216M). Then I checked systic interruption systic (setting for 1ms), and it always was 1ms.

What to do and why it can be?

For example the simplest program:

volatile uint32_t *DWT_CONTROL = (uint32_t *)0xE0001000;

volatile uint32_t *DWT_CYCCNT = (uint32_t *)0xE0001004; 

volatile uint32_t *DEMCR = (uint32_t *)0xE000EDFC; 

volatile unsigned int *DWT_LAR   = (volatile unsigned int *)0xE0001FB0;

uint32_t Mcounter, count;

main()

{

 *DWT_LAR = 0xC5ACCE55; // unlock (CM7)

 // enable the use DWT

*DEMCR = *DEMCR | 0x01000000;

// Reset cycle counter

*DWT_CYCCNT = 0; 

// enable cycle counter

*DWT_CONTROL = *DWT_CONTROL | 1 ;

SystemClock_Config();

while(1)

{

u8 i=0;

Mcounter=*DWT_CYCCNT;

i =1;

count=*DWT_CYCCNT-Mcounter;

float op_time=count/216.0f;// count/F_CPU

SendToUart(count);

}

}

count after power off-on = 0;

count after reset = 34 ticks;

[kitty7c6]

help...somebody..please

[Tesla DeLorean]

What tool chain? What optimization level?

Have you looked at the generated code?

[Tesla DeLorean]

Quick test on F746G-DISCO, cleaned code up, but seems not to clock after power cycle

Hello World!
    5         12
    3  200190208
    3  400390211
    3  600590210
    3  800790208
    3 1000990210
    3 1201190208
    3 1401390212
    3 1601590211
    3 1801790208
    3 2001990208
    0          0 << Stopped Debugging in Keil
    0          0
    0          0
    0          0
    0          0
    0          0
    0          0
    0          0
    0          0
    0          0
    0          0
    0          0
    0          0
    0          0
    0          0
    0          0
    0          0
    0          0
    0          0
    0          0
Hello World! << Reset Button
    3         11
    3  200190212
    3  400390212
    3  600590215
    3  800790212
    3 1000990215
    3 1201190212
    3 1401390216
    3 1601590215
    3 1801790212
    3 2001990215
    3 2202190212
    3 2402390214
    3 2602590216
    3 2802790215
    3 3002990214
    3 3203190215
    3 3403390216
    3 3603590212
    3 3803790216
Hello World!  << Pulling off power jumper and reinstalling
    0          0
Hello World!
    0          0
    0          0
    0          0
    0          0
    0          0
    0          0
    0          0
    0          0
    0          0
    0          0
    0          0

[Tesla DeLorean]

where are you?

[kitty7c6]

RU, Siberia and unclear_how_the_code_works

[kitty7c6]

IAR7.5->FlashLoaderDemonstrator->mcu. My code (and tried cube's code on LL and on HAL). No optimization. I tried this code on stm32f407, it takes less ticks (11). But it disassembling equally on 407 and on 745

[Tesla DeLorean]

Changing the ordering helped,

volatile unsigned int *DWT_CYCCNT   = (volatile unsigned int *)0xE0001004; //address of the register
volatile unsigned int *DWT_CONTROL  = (volatile unsigned int *)0xE0001000; //address of the register
volatile unsigned int *DWT_LAR      = (volatile unsigned int *)0xE0001FB0; //address of the register
volatile unsigned int *SCB_DEMCR    = (volatile unsigned int *)0xE000EDFC; //address of the register
 
void DWT_Check(void)
{
  *SCB_DEMCR |= 0x01000000; // enable scb/debug first
  *DWT_LAR = 0xC5ACCE55; // unlock
  *DWT_CYCCNT = 0; // reset the counter
  *DWT_CONTROL |= 1 ; // enable the counter
 
  while(1)
  {
    volatile uint8_t i = 0;
    uint32_t count;
    uint32_t start = *DWT_CYCCNT;
 
    i = 1;
 
    count = *DWT_CYCCNT - start;
 
    printf("%5d %10u\n", count, *DWT_CYCCNT);
 
    HAL_Delay(1000);
  }
}

[kitty7c6]

o, thanks, it and helped and didn't,

Now it works like this:

download mcu -> count = 53

reset mcu -> count = 97

power off-on -> count = 97.

Problem with strange behavior after downloading remained.

It may be due to:

*FLASH->ACR? (latency 7, ART and prefetch disable, axim in icf)

*cache?

[Tesla DeLorean]

My test consistently reports 3 cycles in Keil.

The counter didn't work on a power cycle, unless I ran the register write code twice, or changed the ordering to put the SCB enable first.

Not sure what your problem is now.