Accessing a variable or memory location by both core(dual core m7 & m4).

[manohara]

my requirement is i have to initialise a variable in M7 core and access that variable in M4 core, please let me know the process t achieve this one. 
board - NUCLEO-H745ZI-Q.
ide - stm32cube(1.13.1 version).

[mƎALLEm]

Hello and welcome to the community.

I can give a raw steps:

1. Define a shared memory region: Choose a memory region that is accessible by both cores (for example DTCM is not accessible by CM4). You can for example use SRAM3 for this purpose.

2. Place your variable in the shared memory: define the variable in a specific section of the memory that is designated for shared access. Ex: SRAM3. This is done using linker script or pragmas depending on your development environment.

3. Ensure Cache coherency: Since CM7 core has a cache, it's important to manage cache coherency for the shared memory region. You may need to disable caching or explicitly manage cache operations using the MPU.

4. Synchronize access: Use a synchronization mechanism to prevent race conditions. The Hardware Semaphore (HSEM) can be used to ensure that only one core accesses the shared memory at a time.

5. Initialize variable in CM7 Core: Write the initial value to the shared variable using the CM7 core.

6. Access variable in CM4 Core: After ensuring that the CM7 core has released the semaphore, the CM4 core can read the shared variable.

Here is an example of how you might define and access a shared variable in C code:

 

// Define the shared variable (place in shared memory section using linker script or pragmas)
volatile int shared_variable __attribute__((section(".shared_memory")));

void M7_Core_Init(void) {
 // Initialize the shared variable
 shared_variable = 42;

 // Ensure the variable is written to shared memory before semaphore release
 __DSB();

 // Release the semaphore to allow M4 to access the shared variable
 HAL_HSEM_Release(Your_HSEM_ID, 0);
}

void M4_Core_Access(void) {
 // Take the semaphore to ensure safe access to the shared variable
 HAL_HSEM_Take(Your_HSEM_ID, 0);

 // Access the shared variable
 int value = shared_variable;

 // Release the semaphore after accessing the shared variable
 HAL_HSEM_Release(Your_HSEM_ID, 0);
}

 

In the above code, replace Your_HSEM_ID with the actual semaphore ID you are using. Also, ensure that the shared memory section (.shared_memory) is properly defined in your linker script to be in the SRAM3 region.

PS: you can also use OpenAMP for more advanced data sharing between cores. Some examples are available on STM32H7Cube on this link: https://github.com/STMicroelectronics/STM32CubeH7/tree/master/Projects/STM32H747I-EVAL/Applications/OpenAMP

Hope I answered your question.