Watchdog task pattern in FreeRTOS

Introduction

A watchdog task helps keep your system safe by checking if other tasks are working properly. If something goes wrong, it can restart or fix the problem.

When you want to make sure your program does not freeze or get stuck.

When you have multiple tasks and want to monitor their health.

When you want to automatically recover from errors without manual intervention.

Syntax

FreeRTOS

void WatchdogTask(void *pvParameters) {
    for (;;) {
        // Check other tasks' status
        if (tasksAreHealthy()) {
            // Reset watchdog timer
            ResetWatchdogTimer();
        } else {
            // Take recovery action
            HandleError();
        }
        vTaskDelay(pdMS_TO_TICKS(1000));
    }
}

The watchdog task runs in an infinite loop to continuously monitor other tasks.

Use vTaskDelay() to wait between checks without blocking the system.

Examples

This example checks tasks every 500 milliseconds and restarts the system if a problem is found.

FreeRTOS

void WatchdogTask(void *pvParameters) {
    for (;;) {
        if (AllTasksResponding()) {
            ResetWatchdogTimer();
        } else {
            RestartSystem();
        }
        vTaskDelay(pdMS_TO_TICKS(500));
    }
}

This version logs errors and tries to recover instead of restarting immediately.

FreeRTOS

void WatchdogTask(void *pvParameters) {
    for (;;) {
        if (CheckTaskStatus()) {
            ResetWatchdogTimer();
        } else {
            LogError();
            AttemptRecovery();
        }
        vTaskDelay(pdMS_TO_TICKS(2000));
    }
}

Sample Program

This program simulates a watchdog task checking other tasks. It prints a message when resetting the watchdog timer or when an error is detected. The main function runs the check 7 times to show both normal and error cases.

FreeRTOS

#include "FreeRTOS.h"
#include "task.h"
#include <stdio.h>

// Simulated function to check if tasks are healthy
int tasksAreHealthy() {
    static int count = 0;
    count++;
    // Simulate a failure every 5 checks
    return (count % 5) != 0;
}

void ResetWatchdogTimer() {
    printf("Watchdog timer reset.\n");
}

void HandleError() {
    printf("Error detected! Taking recovery action.\n");
}

void WatchdogTask(void *pvParameters) {
    for (;;) {
        if (tasksAreHealthy()) {
            ResetWatchdogTimer();
        } else {
            HandleError();
        }
        vTaskDelay(pdMS_TO_TICKS(1000));
    }
}

int main() {
    // Normally FreeRTOS scheduler would be started here
    // For demonstration, run WatchdogTask loop 7 times manually
    for (int i = 0; i < 7; i++) {
        if (tasksAreHealthy()) {
            ResetWatchdogTimer();
        } else {
            HandleError();
        }
    }
    return 0;
}

Important Notes

Make sure the watchdog task runs with enough priority to check other tasks regularly.

Use simple and fast checks inside the watchdog task to avoid delays.

Resetting the watchdog timer usually prevents the system from resetting.

Summary

The watchdog task pattern helps keep your system running smoothly by monitoring other tasks.

It runs in a loop, checking tasks and resetting a timer or taking action if something is wrong.

This pattern improves system reliability and helps recover from errors automatically.

Practice

1. What is the main purpose of a watchdog task in FreeRTOS?

easy

A. To create new tasks dynamically during runtime

B. To monitor other tasks and reset the system if a task fails

C. To handle user input events

D. To manage memory allocation for tasks

Solution

1. Step 1: Understand the watchdog task role

   The watchdog task monitors the health of other tasks to ensure they are running correctly.

2. Step 2: Identify the action taken on failure

   If a task fails or stops responding, the watchdog task can reset the system or take corrective action.

3. Final Answer:

   To monitor other tasks and reset the system if a task fails -> Option B

4. Quick Check:

   Watchdog monitors and resets = A [OK]

Hint: Watchdog watches tasks and resets on failure [OK]

Common Mistakes:

• Confusing watchdog with task creation
• Thinking watchdog handles user input
• Assuming watchdog manages memory

2. Which of the following is the correct way to reset a watchdog timer in a FreeRTOS task?

easy

A. Call watchdog_reset() periodically from monitored tasks

B. Use xTaskCreate() to reset the timer

C. Call vTaskDelay() inside the watchdog task

D. Call vTaskDelete() on the watchdog task

Solution

1. Step 1: Identify how watchdog timers are reset

   Monitored tasks must periodically signal the watchdog by calling a reset function like watchdog_reset().

2. Step 2: Eliminate incorrect options

   vTaskDelay() delays a task but does not reset the watchdog timer. xTaskCreate() creates tasks, not reset timers. vTaskDelete() deletes tasks, not reset timers.

3. Final Answer:

   Call watchdog_reset() periodically from monitored tasks -> Option A

4. Quick Check:

   Reset watchdog by calling reset function = C [OK]

Hint: Reset watchdog by calling its reset function in tasks [OK]

Common Mistakes:

• Using delay functions to reset watchdog
• Trying to reset watchdog by creating or deleting tasks
• Confusing task management functions with watchdog reset

3. Consider this simplified watchdog task code snippet in FreeRTOS:

void WatchdogTask(void *pvParameters) {
    for (;;) {
        if (xSemaphoreTake(watchdogSemaphore, pdMS_TO_TICKS(1000)) == pdFALSE) {
            // No reset signal received
            vTaskDelay(pdMS_TO_TICKS(500));
            // Take recovery action
        } else {
            // Reset signal received
            vTaskDelay(pdMS_TO_TICKS(100));
        }
    }
}

What happens if the monitored task never calls xSemaphoreGive(watchdogSemaphore)?

medium

A. The watchdog task will detect no reset signal and take recovery action repeatedly

B. The watchdog task will keep waiting forever without action

C. The watchdog task will crash due to semaphore misuse

D. The watchdog task will reset the semaphore automatically

Solution

1. Step 1: Analyze semaphore wait behavior

   The watchdog task waits 1000 ms for a semaphore signal. If none arrives, xSemaphoreTake returns pdFALSE.

2. Step 2: Understand the consequence of no signal

   When no signal is received, the watchdog delays 500 ms and then takes recovery action repeatedly in the loop.

3. Final Answer:

   The watchdog task will detect no reset signal and take recovery action repeatedly -> Option A

4. Quick Check:

   No semaphore signal triggers recovery = A [OK]

Hint: No semaphore signal means watchdog triggers recovery [OK]

Common Mistakes:

• Assuming watchdog waits forever without action
• Thinking semaphore misuse causes crash here
• Believing semaphore resets automatically

4. Given this watchdog task snippet:

void WatchdogTask(void *pvParameters) {
    for (;;) {
        if (xSemaphoreTake(watchdogSemaphore, 0) == pdTRUE) {
            // Reset received
        } else {
            // No reset, take action
        }
        vTaskDelay(pdMS_TO_TICKS(1000));
    }
}

What is the main problem with this code?

medium

A. The task deletes itself after one loop

B. The task delays too long after receiving a reset signal

C. The semaphore wait time is zero, so it never blocks to wait for reset signals

D. The semaphore is given incorrectly in the monitored task

Solution

1. Step 1: Examine semaphore wait time parameter

   The call xSemaphoreTake(watchdogSemaphore, 0) means it does not wait and returns immediately.

2. Step 2: Understand impact on watchdog behavior

   Because it never blocks, the watchdog may miss reset signals if they arrive after the check, causing false recovery actions.

3. Final Answer:

   The semaphore wait time is zero, so it never blocks to wait for reset signals -> Option C

4. Quick Check:

   Zero wait means no blocking = D [OK]

Hint: Semaphore wait time zero means no blocking, causing missed signals [OK]

Common Mistakes:

• Thinking delay length is the problem
• Assuming semaphore is given incorrectly without evidence
• Believing task deletes itself here

5. You want to implement a watchdog task that monitors three tasks: TaskA, TaskB, and TaskC. Each task must send a heartbeat signal every 500 ms. Which approach best ensures the watchdog detects a failure in any task within 2 seconds?

hard

A. Use one binary semaphore shared by all tasks; each task gives it every 500 ms

B. Use a single counting semaphore incremented by all tasks; watchdog resets counter every 2 seconds

C. Use three separate semaphores; watchdog waits up to 2 seconds for each semaphore in sequence

D. Use three separate semaphores; watchdog checks each semaphore with a 700 ms timeout and flags failure if any miss

Solution

1. Step 1: Understand heartbeat timing requirements

   Each task sends a heartbeat every 500 ms. The watchdog must detect failure within 2 seconds.

2. Step 2: Evaluate semaphore usage options

   Using separate semaphores allows independent monitoring. Checking each with a 700 ms timeout ensures missing any heartbeat triggers failure detection quickly.

3. Step 3: Compare with other options

   One shared semaphore (Use one binary semaphore shared by all tasks; each task gives it every 500 ms) cannot distinguish which task failed. Waiting 2 seconds per semaphore (Use three separate semaphores; watchdog waits up to 2 seconds for each semaphore in sequence) delays detection. Counting semaphore (Use a single counting semaphore incremented by all tasks; watchdog resets counter every 2 seconds) loses individual task info.

4. Final Answer:

   Use three separate semaphores; watchdog checks each semaphore with a 700 ms timeout and flags failure if any miss -> Option D

5. Quick Check:

   Separate semaphores + short timeout = fast failure detection = B [OK]

Hint: Separate semaphores with short timeout catch failures fast [OK]

Common Mistakes:

• Using one semaphore for all tasks losing failure detail
• Waiting too long per semaphore delaying detection
• Using counting semaphore losing individual task info