12.md

What do you need to do to write your own kernel?

Step 1. Define the kernel philosophy (required)

Before creating folders, you must answer the 5 questions.

1. Who is the kernel for?

• educational
• research
• minimalist
• extensible

This affects:

• complexity
• flexibility
• number of abstraction levels

2. What the kernel must do

Minimum set:

• run
• manage memory
• manage execution
• respond to events

Everything else is optional.

3. What the kernel will NOT do

It is very important to explicitly state:

• no binary compatibility
• no legacy
• no automagic

---

Step 4. Define the main subsystems

Don't implement – ​​name and describe.

Required subsystems:

1. Boot / Entry

Responsible for:

• taking control
• initialization
• transitioning to the running state

2. CPU Management

Responsible for:

• processor modes
• execution context
• switching

3. Memory Management

Responsible for:

• memory accounting
• isolation
• allocation

4. Interrupt & Event System

Responsible for:

• hardware events
• timers
• system responses

5. Execution Model

Responsible for:

• tasks
• processes
• scheduling

6. Core Services

Common mechanisms:

• synchronization
• queues
• basic data structures

---

Step 5. Creating your kernel architecture

Use non-random folder names What's important for the system and for understanding you, not drt/tree, etc.

We will always use the following structure:

kernel/
├─ entry/
│ └─ start
├─ platform/
│ └─ cpu
│ └─ memory
│ └─ interrupts
├─ core/
│ └─ scheduler
│ └─ execution
│ └─ events
├─ services/
│ └─ sync
│ └─ time
│ └─ objects
├─ internal/
│ └─ types
│ └─ config
│ └─ contracts

---

Step 6. Rigidly define boundaries

The most common mistake is shifting responsibility.

You must decide in advance:

• who can call whom
• who CANNOT
• where is the single entry point

Example:

• entry → platform
• platform → core
• core → services
• services DO NOT interfere with platform