Problem with Semaphore and Task-Delay

Hi, I am working with FreeRTOS and have encountered a problem with one task. The task in general looks this way: #define MILLISECONDS( x ) ( x / portTICKPERIODMS) void foo(void* pvParameters) { const TickTypet delay = MILLISECONDS(10); const TickTypet unblockSmphrDelay = MILLISECONDS(1000);
xSemaphoreTake(SyncStatusRegSmphr, 0);

for (;;)
{
    xSemaphoreTake(SyncStatusRegSmphr,unblockSmphrDelay);

    //Send CAN message

    vTaskDelay(delay);
}
} It runs with FreeRTOS V8.1.2 on an STM32F407VGT6 controller. The task can be unblocked from different tasks/ISRs (ISRs with xSemaphoreGiveFromISR). The intention is to send event based CAN messages, but with a maximum frequency of 10ms, no matter how many events occur. Also the message should be sent at least every second, even if there were no events that unblocked the task. The behaviour is as follows: The task works as it should for a time that seems random to me (at least for some minutes, sometimes for several hours). But sporadically it occurs, that the CAN-message is not sent anymore. The controller does not crash and several other tasks keep sending on the CAN-bus. The controller still executes anything it should except this task. My question would be, if it is ok to write the task in this way in general, and how this behavior can occur? Thanks! Simon

Problem with Semaphore and Task-Delay

First a small remark. If you want to transform milliseconds to a tick-count, it is more correct to use ‘pdMS_TO_TICKS‘. In this example 5 seconds:
TickType_t xTickCount = pdMS_TO_TICKS( 5000ul )
The example that you give looks perfect to me, nothing wrong with it. I’m afraid that the problem is in the code that you’re not showing. You are sending a CAN message, is that resource well-protected, is it “re-entrant” ? Regards.