Serial wire output on STM32L471ZGTx is not working - it hangs

[Nix Wayne]

I am using MDK5 (5.22) with STM32CubeMX (4.26.1). I generate blank project within CubeMX for STM32L471ZGTx (144pin MCU). Clock configuration uses internal HSI and LSI and it's set to 16MHz. For testing purposes I configured clocks so all other clocks are 16MHz (SYSCLK, APB1, Timers, Cortex etc...)

Under SYS menu I enable TRACE asynchronous SW to use SW for programming and SWO for retargeting printf.

I generate project and open it in MDK5 then under RTE editor I set up Compiler->I/O and select all options to ITM. So retargeting of printf goes to ITM.

Under options for target I enable TRACE - ITM channel 0.

Then I try to use printf in my main function

int main(void)
{
  /* Reset of all peripherals, Initializes the Flash interface and the Systick. */
  HAL_Init();
 
  /* Configure the system clock */
  SystemClock_Config();
 
   /* Initialize all configured peripherals */
  MX_GPIO_Init();
  printf("Hello world\n");
 
...

But I get stuck on that command. If I press Stop code execution it points to file retarget_io.c to line 99. Obviously I am stuck within that while-loop:

I can get anywhere with this... so far I tried different clock configurations, I checked for HW failure and there is solid connection to pin PB3 (SWO). I checked my ULINK2 programmer and tried with others... same problem.

I use same method for other projects with other MCU families which supports SWO and it works except with this one. I tried same approach on discovery board with STM32L475 and it works.

Any help/ideas would be much appreciated!

Thank you!

[Tesla DeLorean]

I normally check Privilege Port 0..7

[Nix Wayne]

I forgot to mention this but I also tried checking privilege for Port 0..7 and it's same.

[Tesla DeLorean]

Using SWO on several L4 series parts here, not got a L471 to try.​

[Nix Wayne]

It's quite straight forward how to enable and use SWO therfore it's strange to me because I have experiences with other MCUs and it's not problem at all except with this one.

[Nix Wayne]

For your information... I found out that problem was in Keil Ulink2 programmer. I have two of them and with both I have same problem, then I borrowed Segger's J-link and I connected it to my HW and ITM and printf works now without a problem. I guess I will use J-link with this project. Thanks for your contribution.

[Nix Wayne]

Additionaly I found out that if I try to debug first with J-link then I switch to ulink2 printf starts working with ulink (until next hardware reset). I guess that J-link puts MCU's registers in different startup states comparing to Ulink2. So what I did I made screen capture of registers when using JLink and registers when using Ulink2... and I found differences in DBGMCU registers. Then I forced those register values to same as with JLink and printf works. (see attached image for differences)

I think I will make a configuration ini file for debug mode which will setup those registers on startup when using Ulink2.

[Nix Wayne]

My ini file looks like this below:

// Manually enable debugging in DBGMCU module (DBG_SLEEP, DBG_STOP, DBG_STANDBY, TRACE_IOEN, TRACE_MODE)
_WWORD(0xE0042004, 0x27) // CR register of DBGMCU
 
// Enable fields in freeze register APB1FZR1 
_WDWORD(0xE0042008, 0x1800) // APB1FZR1 register of DBGMCU