I would like to spawn multiple threads dynamically from inside a loop to create a simulation.
t0s: TaskA.0 runs for 4sec
t1s: TaskA.1 runs for 4sec
t2s: TaskA.2 runs for 4sec
t3s: TaskA.3 runs for 4sec
t4s: TaskA.4 runs for 4sec
...
t50s: TaskA.50 runs for 4sec
The same 4sec task begins every second, so the tasks overlap, but also expire. No more than 4 instances of the same task will run at any time. Spawning them from the loop works when the tasks don't overlap (delay between task > 4sec). CONFIG_USERSPACE=y allows me to reuse the thread.
Things crash when the threads overlap (delay between task(t) < 4sec). I seem to be able to create k_thread as an array, but it's reusing "my_stack_area" which is probably a problem.
Because this is predictable I might be able to create five separate threads that could overlap each other and use a loop 10x. But, I'm new to Zephyr/RTOS and want to know if there a way, or alternative approach, to accomplishing this with a single thread definition and without statically creating multiple threads to handle the overlap?
This code doesn't exactly work, but illustrates what I was thinking:
static int thread_count = 0;
extern void my_entry_point(void *, void *, void *)
{
int mycount = thread_count;
LOG_INF("thread: %d", mycount);
//do something that takes 4000ms
k_sleep(K_MSEC(4000));
return;
}
#define MY_STACK_SIZE 2048
#define MY_PRIORITY 5
K_THREAD_STACK_DEFINE(my_stack_area, MY_STACK_SIZE);
struct k_thread my_thread_data[50];
int main(){
for(int i = 0; i < 50; i++){
thread_count++;
LOG_INF("LOOP: %d", thread_count);
k_thread_create(&my_thread_data[i], my_stack_area,
K_THREAD_STACK_SIZEOF(my_stack_area),
my_entry_point,
NULL, NULL, NULL,
MY_PRIORITY, 0, K_NO_WAIT);
k_sleep(K_MSEC(1000));
}
}
