I am trying to persuade an stm32F412 to provide Periodic PC sample packets.
Reference manual RM0402 implies it is doable (Configure DWT as a PC sampler)
The "Cortex-M4 Technical Reference Manual r0p0" says
"You can configure the DWT to generate PC samples at defined intervals" and DWT registers are described in the ARMv7M Architecture Reference Manual
That reference manual contains what I thought were reasonable instructions.
I can see the CYCCNT register changing value as expected.
I have confirmed that ITM and DWT registers can be written to.
I'm obviously missing something!
The below code does generate ITM packets over the SWO interface of my stlink v2-mini e, picked up via openocd, so I believe that the serial and clock speeds are set correctly.
DBGMCU->CR |= DBGMCU_CR_TRACE_IOEN;
CoreDebug->DEMCR |= CoreDebug_DEMCR_TRCENA_Msk;
DWT->CYCCNT = 0;
ITM->LAR = 0xC5ACCE55; // unlock the ITM registers
ITM->TCR |= ITM_TCR_ITMENA_Msk; // enable the ITM
ITM->TCR |= ITM_TCR_DWTENA_Msk; //manual described as necessary
ITM->TCR |= ITM_TCR_SYNCENA_Msk; //This plus DWT syntap should mean I see synch packets
ITM->TPR = 0xFFFFFFFF; // enable unprivileged acc.
ITM->TER = 0xFFFFFFFF; // enable all ports
TPI->CSPSR = 1; // port width = 1 bit
TPI->SPPR = 2; // UART encoding
TPI->ACPR = (5-1);
TPI->FFCR &= ~TPI_FFCR_EnFCont_Msk; // no formatter (unclear what this function really means, more reading required)
//TPI->FFCR |= TPI_FFCR_EnFCont_Msk; //with this turned on, ITM does seem to emit more readable packets
DWT->CTRL = 0;
DWT->CTRL |= (0b1111 << DWT_CTRL_POSTINIT_Pos ); //Try to start the countdown from a high point
DWT->CTRL |= DWT_CTRL_CYCTAP_Msk
| (0xFUL << DWT_CTRL_POSTPRESET_Pos); //Reset countdown
DWT->CTRL &= ~DWT_CTRL_CYCTAP_Msk; //Changing how frequent the counter tics
DWT->CTRL |= (DWT_CTRL_CYCCNTENA_Msk); //enable counter
DWT->CTRL |=(DWT_CTRL_PCSAMPLENA_Msk); //enable pcsampling
I am trying to use the itm-tools code found at https://github.com/japaric/itm-tools to decode the packets. I have no reason to believe that tool is flawed; I can see it recognising other packet types.
