FreeRTOS heap implementations (heap_1 to heap_5) - Deep Dive

Overview - FreeRTOS heap implementations (heap_1 to heap_5)

What is it?

FreeRTOS provides five different heap memory management schemes named heap_1 to heap_5. These schemes control how dynamic memory allocation and deallocation happen inside the operating system. Each heap implementation offers a different balance of simplicity, efficiency, and features like memory freeing or coalescing. They help FreeRTOS manage memory safely and efficiently on small embedded devices.

Why it matters

Embedded systems often have very limited memory, so managing it well is critical to avoid crashes or wasted space. Without these heap implementations, FreeRTOS would struggle to allocate memory dynamically, leading to system instability or inefficient use of resources. Choosing the right heap scheme ensures your device runs reliably and uses memory optimally.

Where it fits

Before learning about FreeRTOS heap implementations, you should understand basic C programming, pointers, and dynamic memory concepts. After this, you can explore FreeRTOS task management and synchronization, which rely on memory allocation. Later, you might study advanced memory protection and real-time system optimization.

Mental Model

Core Idea

FreeRTOS heap implementations are different ways to manage dynamic memory allocation and freeing, each with trade-offs in simplicity, efficiency, and features.

Think of it like...

Imagine a small workshop with limited storage space. Each heap scheme is like a different way to organize tools and materials: some just add tools without removing old ones, others carefully track and rearrange tools to save space, and some combine nearby empty spots to make room for bigger tools.

┌───────────────┐
│ FreeRTOS App │
└──────┬────────┘
       │
       ▼
┌─────────────────────────────┐
│ Heap Implementations (heap_1 to heap_5) │
│                             │
│ heap_1: Simple alloc only    │
│ heap_2: Alloc + free         │
│ heap_3: Uses malloc/free     │
│ heap_4: Coalescing free      │
│ heap_5: Multiple regions     │
└─────────────────────────────┘
       │
       ▼
┌───────────────┐
│ Memory Hardware│
└───────────────┘

Build-Up - 7 Steps

1

Foundation

What is Dynamic Memory Allocation

Concept: Introduce the basic idea of dynamic memory allocation and why it is needed.

Dynamic memory allocation means reserving memory space while the program runs, not before. This lets programs use memory flexibly, like creating new tasks or buffers as needed. In C, functions like malloc() and free() handle this, but embedded systems need special care because memory is limited.

Result

You understand that dynamic memory allocation allows programs to request and release memory during execution.

Understanding dynamic memory is key because FreeRTOS heap schemes are all about how this memory is managed safely and efficiently.

2

Foundation

Why FreeRTOS Needs Custom Heaps

Concept: Explain why FreeRTOS does not just use standard malloc/free and needs its own heap implementations.

Standard malloc/free are often not suitable for embedded systems because they can be slow, non-deterministic, or use too much memory. FreeRTOS provides its own heap schemes to give predictable, efficient memory management tailored for real-time and resource-constrained environments.

Result

You see why FreeRTOS offers multiple heap options instead of relying on standard C library functions.

Knowing the limitations of standard memory functions helps you appreciate why FreeRTOS heap implementations exist.

3

Intermediate

Heap_1: Simple Allocation Only

Concept: Learn the simplest heap scheme that only allows memory allocation but no freeing.

heap_1 just moves a pointer forward to allocate memory and never frees it. This is very fast and simple but can only be used when memory is allocated once and never released, like static buffers or startup tasks.

Result

You understand heap_1 is fast but cannot reclaim memory, so it suits simple or fixed allocation needs.

Knowing heap_1's limitation helps you avoid memory leaks in systems that need dynamic freeing.

4

Intermediate

Heap_2: Allocation and Freeing

🤔Before reading on: do you think heap_2 can reuse freed memory immediately or does it wait until the end?

Concept: heap_2 adds the ability to free memory, allowing reuse of freed blocks but does not merge adjacent free blocks.

heap_2 tracks allocated and freed blocks and can reuse freed memory for new allocations. However, it does not combine adjacent free blocks, so fragmentation can occur over time.

Result

You see heap_2 supports freeing but may suffer from fragmentation, limiting long-term memory use.

Understanding heap_2's behavior helps you predict when fragmentation might cause allocation failures.

5

Intermediate

Heap_3: Using Standard malloc/free

🤔Before reading on: do you think heap_3 is deterministic and suitable for real-time systems?

Concept: heap_3 simply wraps the standard C library malloc() and free() functions for memory management.

heap_3 delegates all allocation and freeing to the system's malloc/free. This is easy to use but depends on the system's implementation, which may not be real-time safe or efficient.

Result

You realize heap_3 is convenient but may not be suitable for strict real-time or embedded constraints.

Knowing heap_3's reliance on system malloc/free helps you decide when to avoid it in critical systems.

6

Advanced

Heap_4: Coalescing Free Blocks

🤔Before reading on: do you think heap_4 can reduce fragmentation better than heap_2? Commit to yes or no.

Concept: heap_4 improves on heap_2 by merging adjacent free blocks to reduce fragmentation and improve memory reuse.

heap_4 tracks allocated and free blocks and when freeing memory, it merges neighboring free blocks into larger blocks. This reduces fragmentation and allows larger allocations later.

Result

You understand heap_4 offers better memory efficiency and longer system uptime than heap_2.

Recognizing how coalescing works helps you design systems that maintain memory health over time.

7

Expert

Heap_5: Multiple Memory Regions

🤔Before reading on: do you think heap_5 can manage memory spread across different physical areas? Commit to yes or no.

Concept: heap_5 extends heap_4 by allowing multiple separate memory regions to be managed as one heap.

heap_5 lets you define several non-contiguous memory areas that FreeRTOS treats as a single heap. This is useful for systems with fragmented or special memory types. It still coalesces free blocks within each region but cannot merge across regions.

Result

You see heap_5 provides maximum flexibility for complex memory layouts in advanced embedded systems.

Understanding heap_5's multi-region approach helps you optimize memory use on hardware with diverse memory types.

Under the Hood

Each heap implementation manages a block of memory by tracking which parts are allocated and which are free. heap_1 simply moves a pointer forward for allocation. heap_2 and heap_4 maintain linked lists of free and allocated blocks, with heap_4 merging adjacent free blocks to reduce fragmentation. heap_3 calls the system's malloc/free. heap_5 manages multiple such regions independently and combines their results logically. Internally, these schemes manipulate pointers and metadata to keep track of memory usage and ensure allocations do not overlap.

Why designed this way?

FreeRTOS heap schemes were designed to offer a range of trade-offs between simplicity, speed, and memory efficiency to suit different embedded system needs. Early embedded systems had very limited memory and CPU power, so simple schemes like heap_1 were useful. As systems grew more complex, features like freeing and coalescing became necessary. heap_3 was added for convenience on systems with robust malloc/free. heap_5 was introduced to handle complex memory layouts common in modern hardware. Alternatives like fully dynamic allocators were too heavy or unpredictable for real-time use.

┌───────────────┐
│ Memory Block  │
├───────────────┤
│ Metadata      │◄─┐
│ (size, status)│  │
├───────────────┤  │
│ User Data     │  │
└───────────────┘  │
                   │
┌───────────────┐  │
│ Linked List   │──┘
│ of Blocks     │
└───────────────┘

heap_1: Pointer moves forward
heap_2/4: Linked list with free/allocated blocks
heap_4: Merges adjacent free blocks
heap_5: Multiple such lists for regions

Myth Busters - 4 Common Misconceptions

Quick: Does heap_1 support freeing memory? Commit yes or no.

Common Belief: heap_1 supports freeing memory just like malloc/free.

Reality: heap_1 does not support freeing memory; once allocated, memory is never reclaimed.

Why it matters: Assuming heap_1 frees memory leads to memory exhaustion and system crashes in long-running applications.

Quick: Is heap_3 always real-time safe? Commit yes or no.

Common Belief: heap_3 is safe for real-time systems because it uses standard malloc/free.

Reality: heap_3 depends on the system's malloc/free, which may not be deterministic or safe for real-time use.

Why it matters: Using heap_3 in real-time systems can cause unpredictable delays and missed deadlines.

Quick: Does heap_4 merge free blocks across different memory regions in heap_5? Commit yes or no.

Common Belief: heap_4's coalescing merges free blocks across all memory regions in heap_5.

Reality: heap_4 merges free blocks only within a single memory region; heap_5 manages multiple regions separately without cross-region merging.

Why it matters: Expecting cross-region merging can cause fragmentation and allocation failures in multi-region setups.

Quick: Does heap_2 prevent fragmentation completely? Commit yes or no.

Common Belief: heap_2 prevents fragmentation by reusing freed memory blocks.

Reality: heap_2 reuses freed blocks but does not merge adjacent free blocks, so fragmentation can still occur.

Why it matters: Ignoring fragmentation risks leads to allocation failures even when free memory exists in small pieces.

Expert Zone

1

heap_4's coalescing algorithm can cause subtle delays during freeing, which may impact real-time performance if freeing happens frequently.

2

heap_5 requires careful configuration of memory regions to avoid gaps and ensure efficient use; misconfiguration can cause wasted memory.

3

heap_3's behavior depends heavily on the underlying system malloc/free implementation, which can vary widely between platforms and affect portability.

When NOT to use

Avoid heap_1 if your system needs to free memory dynamically; use heap_4 or heap_5 instead. Avoid heap_3 in hard real-time systems due to unpredictable malloc/free behavior. For systems with very limited memory and no need to free, heap_1 is best. For complex memory layouts, heap_5 is preferred. If fragmentation is a major concern, heap_4 or heap_5 are better choices.

Production Patterns

In production, heap_4 is commonly used for general-purpose embedded systems needing dynamic allocation with fragmentation control. heap_5 is chosen for advanced systems with multiple memory types like internal RAM and external memory. heap_1 is used in simple or safety-critical systems where memory is allocated once at startup. heap_3 is rare in embedded but used when porting FreeRTOS to systems with robust OS support.

Connections

Memory Fragmentation

Heap implementations manage fragmentation differently, directly affecting system stability.

Understanding fragmentation helps you choose or design heap schemes that maintain long-term memory health.

Real-Time Operating Systems (RTOS) Scheduling

Heap behavior impacts task scheduling by affecting allocation delays and determinism.

Knowing heap internals helps optimize RTOS scheduling to avoid unexpected latency caused by memory operations.

Warehouse Inventory Management

Both involve organizing limited space efficiently and reclaiming freed space to store new items.

Recognizing this similarity can inspire better memory management strategies by applying logistics principles.

Common Pitfalls

#1 Using heap_1 in a system that needs to free memory dynamically.

Wrong approach:

void *ptr = pvPortMalloc(100);
pvPortFree(ptr); // Using heap_1 which does not support free

Correct approach:

Use heap_4 or heap_5 which support pvPortFree:
void *ptr = pvPortMalloc(100);
pvPortFree(ptr);

Root cause:

Misunderstanding that heap_1 supports freeing memory leads to memory leaks and eventual exhaustion.

#2 Assuming heap_3 is deterministic and suitable for real-time tasks.

Wrong approach:

Using heap_3 in a timing-critical task expecting consistent allocation times.

Correct approach:

Use heap_4 or heap_5 for deterministic allocation times in real-time tasks.

Root cause:

Not realizing that system malloc/free can have unpredictable timing and fragmentation.

#3 Configuring heap_5 with overlapping or misaligned memory regions.

Wrong approach:

Defining heap_5 regions with overlapping addresses or gaps:
HeapRegion_t regions[] = {
{ (uint8_t *)0x20000000, 0x1000 },
{ (uint8_t *)0x20000800, 0x0800 }, // Overlaps previous
{ NULL, 0 }
};

Correct approach:

Define non-overlapping, aligned regions:
HeapRegion_t regions[] = {
{ (uint8_t *)0x20000000, 0x0800 },
{ (uint8_t *)0x20000800, 0x0800 },
{ NULL, 0 }
};

Root cause:

Lack of understanding of memory layout causes wasted memory or corruption.

Key Takeaways

FreeRTOS offers five heap implementations to manage dynamic memory with different trade-offs in complexity, efficiency, and features.

Choosing the right heap depends on your system's need for freeing memory, fragmentation control, and memory layout complexity.

heap_1 is simple and fast but cannot free memory; heap_4 and heap_5 support freeing and reduce fragmentation.

heap_3 uses standard malloc/free but may not be suitable for real-time or embedded constraints.

Understanding how each heap works internally helps avoid common pitfalls and optimize system reliability.