Deadlock prevention patterns in FreeRTOS - Time & Space Complexity

Time Complexity: Deadlock prevention patterns

O(n)

Understanding Time Complexity

When using FreeRTOS, preventing deadlocks is key to keeping tasks running smoothly.

We want to see how the time cost changes when applying deadlock prevention patterns.

Scenario Under Consideration

Analyze the time complexity of this deadlock prevention pattern using resource ordering.

// Task tries to take two mutexes in order
if(xSemaphoreTake(mutexA, portMAX_DELAY) == pdTRUE) {
    if(xSemaphoreTake(mutexB, portMAX_DELAY) == pdTRUE) {
        // Critical section
        xSemaphoreGive(mutexB);
    }
    xSemaphoreGive(mutexA);
}

This code shows a task taking two mutexes in a fixed order to avoid deadlock.

Identify Repeating Operations

Look for loops or repeated waits that affect time.

• Primary operation: Waiting to take each mutex (xSemaphoreTake calls)
• How many times: Each mutex is taken once per critical section entry

How Execution Grows With Input

As the number of tasks or mutexes grows, waiting time can increase.

Input Size (n tasks)	Approx. Operations (wait attempts)
10	About 10 waits per mutex
100	About 100 waits per mutex
1000	About 1000 waits per mutex

Pattern observation: Waiting grows roughly linearly with the number of tasks competing for mutexes.

Final Time Complexity

Time Complexity: O(n)

This means waiting time grows linearly as more tasks try to access the same resources.

Common Mistake

[X] Wrong: "Taking mutexes in any order won't affect waiting time or deadlocks."

[OK] Correct: Taking mutexes in different orders can cause deadlocks, making tasks wait forever, which is worse than linear waiting.

Interview Connect

Understanding how deadlock prevention affects waiting time shows you can write safe multitasking code that scales well.

Self-Check

"What if tasks could take mutexes in any order? How would the time complexity and system behavior change?"

Practice

1.

What is a common method to prevent deadlocks in FreeRTOS when multiple tasks need several mutexes?

Choose the best practice.

easy

A. Avoid using mutexes altogether

B. Allow tasks to take mutexes in any order

C. Use recursive mutexes without order

D. Always take mutexes in a fixed global order

Solution

1. Step 1: Understand deadlock cause in FreeRTOS

   Deadlocks happen when tasks wait forever for mutexes held by each other in a cycle.

2. Step 2: Apply fixed order mutex acquisition

   Taking mutexes always in the same order prevents circular wait, stopping deadlocks.

3. Final Answer:

   Always take mutexes in a fixed global order -> Option D

4. Quick Check:

   Fixed mutex order = deadlock prevention [OK]

Hint: Take mutexes always in the same order to avoid deadlocks [OK]

Common Mistakes:

• Taking mutexes in random order
• Ignoring priority inheritance
• Using recursive mutexes incorrectly

2.

Which of the following is the correct way to create a mutex in FreeRTOS to help prevent deadlocks?

?

easy

A. xSemaphoreCreateMutex();

B. xSemaphoreCreateCounting(1, 1);

C. xQueueCreateMutex();

D. xSemaphoreCreateRecursiveMutex();

Solution

1. Step 1: Identify FreeRTOS mutex creation functions

   FreeRTOS uses xSemaphoreCreateMutex() to create a standard mutex.

2. Step 2: Differentiate from other semaphore types

   Counting semaphores and recursive mutexes serve different purposes; standard mutex is best for deadlock prevention.

3. Final Answer:

   xSemaphoreCreateMutex(); -> Option A

4. Quick Check:

   Standard mutex creation = xSemaphoreCreateMutex() [OK]

Hint: Use xSemaphoreCreateMutex() for standard mutex creation [OK]

Common Mistakes:

• Confusing counting semaphore with mutex
• Using recursive mutex without need
• Trying to create mutex with queue functions

3.

Consider this FreeRTOS code snippet where two tasks try to take two mutexes in different orders:

Task1:
  xSemaphoreTake(mutexA, portMAX_DELAY);
  xSemaphoreTake(mutexB, portMAX_DELAY);

Task2:
  xSemaphoreTake(mutexB, portMAX_DELAY);
  xSemaphoreTake(mutexA, portMAX_DELAY);

What is the likely outcome?

medium

A. One task always fails to take mutex

B. Both tasks run without blocking

C. Deadlock occurs because tasks wait on each other's mutex

D. Tasks release mutexes immediately, no deadlock

Solution

1. Step 1: Analyze mutex acquisition order

   Task1 takes mutexA then mutexB; Task2 takes mutexB then mutexA, creating circular wait.

2. Step 2: Identify deadlock condition

   Each task waits for the other's mutex, causing deadlock with portMAX_DELAY blocking indefinitely.

3. Final Answer:

   Deadlock occurs because tasks wait on each other's mutex -> Option C

4. Quick Check:

   Opposite mutex order + infinite wait = deadlock [OK]

Hint: Opposite mutex order with infinite wait causes deadlock [OK]

Common Mistakes:

• Assuming tasks run without blocking
• Ignoring portMAX_DELAY effect
• Thinking one task fails immediately

4.

Given this FreeRTOS code snippet, identify the problem that can cause deadlock:

void Task1(void *pvParameters) {
  xSemaphoreTake(mutex1, portMAX_DELAY);
  vTaskDelay(10);
  xSemaphoreTake(mutex2, portMAX_DELAY);
  // ...
  xSemaphoreGive(mutex2);
  xSemaphoreGive(mutex1);
}

void Task2(void *pvParameters) {
  xSemaphoreTake(mutex2, portMAX_DELAY);
  vTaskDelay(10);
  xSemaphoreTake(mutex1, portMAX_DELAY);
  // ...
  xSemaphoreGive(mutex1);
  xSemaphoreGive(mutex2);
}

What is the best fix to prevent deadlock?

medium

A. Remove all delays between mutex takes

B. Make both tasks take mutexes in the same order

C. Use recursive mutexes instead

D. Increase delay time to avoid overlap

Solution

1. Step 1: Identify cause of deadlock

   Task1 takes mutex1 then mutex2; Task2 takes mutex2 then mutex1, causing circular wait and deadlock.

2. Step 2: Apply fixed mutex order to both tasks

   Changing Task2 to take mutex1 then mutex2 avoids circular wait and prevents deadlock.

3. Final Answer:

   Make both tasks take mutexes in the same order -> Option B

4. Quick Check:

   Same mutex order = no deadlock [OK]

Hint: Fix mutex order in all tasks to the same sequence [OK]

Common Mistakes:

• Removing delays does not fix deadlock
• Using recursive mutexes won't solve order problem
• Increasing delay time only delays deadlock

5.

You have three tasks that need to access three shared resources protected by mutexes: mutexX, mutexY, and mutexZ. To prevent deadlocks, which strategy is best?

1. Task A needs mutexX and mutexY
2. Task B needs mutexY and mutexZ
3. Task C needs mutexX and mutexZ

Choose the best deadlock prevention pattern.

hard

A. Each task takes mutexes in alphabetical order: mutexX, mutexY, mutexZ

B. Each task takes mutexes in random order to improve speed

C. Use recursive mutexes for all mutexes

D. Tasks take mutexes only when all are free simultaneously

Solution

1. Step 1: Analyze mutex requirements and potential deadlocks

   Tasks share mutexes in overlapping pairs, risking circular wait if order differs.

2. Step 2: Apply fixed global order for mutex acquisition

   Taking mutexes always in alphabetical order (mutexX, mutexY, mutexZ) prevents circular wait and deadlocks.

3. Final Answer:

   Each task takes mutexes in alphabetical order: mutexX, mutexY, mutexZ -> Option A

4. Quick Check:

   Global fixed mutex order = deadlock prevention [OK]

Hint: Use global fixed order for all mutex acquisitions [OK]

Common Mistakes:

• Taking mutexes randomly causes deadlocks
• Recursive mutexes don't fix order issues
• Waiting for all mutexes simultaneously is complex