FreeRTOS example on nucleo-H745ZI-Q

Kernel
1

Hi,
I try run demo application for core to core communication, but I get assertion that CM4 core doesn’t see contents initialized by CM7 core.

Here is place where program stopped:

configASSERT( ( xControlMessageBuffer != NULL ) && ( xDataMessageBuffers[ 0 ] != NULL ) && ( xDataMessageBuffers[ 1 ] != NULL ) );

MessageBufferLocations.h

#ifndef MESSAGE_BUFFER_AMP_H
#define MESSAGE_BUFFER_AMP_H

/* Enough four 4 8 byte strings, plus the additional 4 bytes per message
overhead of message buffers. */
#define mbaTASK_MESSAGE_BUFFER_SIZE ( 60 )

#define mbaCONTROL_MESSAGE_BUFFER_SIZE ( 24 )

/* The number of instances of prvM4CoreTasks that are created. */
#define mbaNUMBER_OF_CORE_2_TASKS	2

/* A block time of 0 simply means, don't block. */
#define mbaDONT_BLOCK				0
/* Place the message buffers at a fixed location so it is the same for both
cores. */
MessageBufferHandle_t xControlMessageBuffer __attribute__((section(".RAM_D3_Z1")));

MessageBufferHandle_t xDataMessageBuffers[ mbaNUMBER_OF_CORE_2_TASKS ] __attribute__ ((section (".RAM_D3_Z2")));

static uint32_t ulStartSyncCounters[ mbaNUMBER_OF_CORE_2_TASKS ] __attribute__ ((section (".RAM_D3_Z3")));

/* The variable used to hold the stream buffer structure.*/
StaticStreamBuffer_t xControlMessageBufferStruct  __attribute__ ((section (".RAM_D3_Z4")));

StaticStreamBuffer_t xDataMessageBufferStructs[mbaNUMBER_OF_CORE_2_TASKS] __attribute__ ((section (".RAM_D3_Z5")));

/* Used to dimension the array used to hold the streams.*/
/* Defines the memory that will actually hold the streams within the stream buffer.*/
static uint8_t ucControlBufferStorage[ mbaCONTROL_MESSAGE_BUFFER_SIZE ] __attribute__ ((section (".RAM_D3_Z6")));

static uint8_t ucDataBufferStorage[mbaNUMBER_OF_CORE_2_TASKS][ mbaTASK_MESSAGE_BUFFER_SIZE ] __attribute__ ((section (".RAM_D3_Z7")));

#endif /* MESSAGE_BUFFER_AMP_H */

Can somone help handle this problem ?

2

It is helpful to provide context for people reading the post. In the case the assert() in question is presumably on this line: https://github.com/FreeRTOS/FreeRTOS/blob/main/FreeRTOS/Demo/CORTEX_M7_M4_AMP_STM32H745I_Discovery_IAR/CM4/main.c#L229 - so the call to prvWaitForOtherCoreToStart() on the line above has returned.

Which of the three tests in the assert is failing?

3

Thank you for your answer, I had solved problem myself, but now I have other following problem:

I have two tasks for both cores (CM7 and CM4) and if I use osDelay function with 1 ms I get different times on both cores. CM7 has right time, but in CM4 delay takes 3ms instead 1ms.
Someone know where can e problem ?

4

What else is running on the CM4 core? Is it possible that your application is masking interrupts for long duration?