STM32H563: No TRACECK present

[B1sh0p]

Hi,

I have configured my STM32H563AI CubeMX-project to use "JTAG with Trace  Synchro(4bits)" and assigned the corresponding pins. Also, I explicitly activated the corresponding registers in DBGMCU and DCB by adding:

  SET_BIT(DBGMCU->CR, DBGMCU_CR_TRACE_IOEN);
  SET_BIT(DBGMCU->CR, DBGMCU_CR_TRACE_CLKEN);
  SET_BIT(DCB->DEMCR, DCB_DEMCR_TRCENA_Msk);

to MX_GPIO_Init.

However, when I try to trace using Ozone V3.38g I always get a "No trace clock present, trace may not work correctly."-warning and nothing is read out from the instruction cache.

My trace settings in Ozone are: 

Trace Source: Trace Pins

CPU Frequency: 250 MHz

Trace Port width: 4 bit

Maximum Instruction Count: 100M

I used an oscilloscope to look at the TRACECK-signal but found nothing there, it only ever has 0V.

Also, I verified my tracing setup, using a SEGGER Cortex M-Trace Reference Board and with this it was just working fine.

Also, I verified, that the corresponding registers get set in MX_GPIO_Init using ozone and just debugging via JTAG.

Any ideas how to solve this?

-Best regards

[Pavel A.]

Have you asked on the Segger forum?  IIRC they do not have trace eval boards based on Cortex-M33.

[B1sh0p]

Although the reference board is M4 (STM32F407), Segger states that Cortex-M33, and in fact the particular device I am using is supported and tested: Tested Devices - J-Trace PRO

A sample project for the corresponding NUCLEO board is provided: NUCLEO-H563ZI - SEGGER Knowledge Base

This is what I am looking into at the moment for further clues.

[B1sh0p · Accepted solution]

I now have solved this issue:

The MCU-registers that I tried to set in the application code have to be written by the probe over the debug-interface. This has to be done in a JLinkScript-file by using the hook OnTraceStart().

I have attached a corresponding script for the STM32H563.