Programming Model

Defines how instructions are organized, how control flow works and how multiple tasks are executed at the hardware level.

Types of Programs

Traditional Programs

Control flow via:

• Branching (if/else, loops)
• Function calls (sub-routines)

Interruptable Programs

Can be paused and resumed. Paused through interrupts. In that case, CPU context-switches to Interrupt Service Routine.

Multi-Threaded Programs

Contains 1 or more threads running in parallel.

• Shared memory → memory safety needed
• Threads from different libraries, languages, or modules

Multi-Core and Heterogeneous Systems

Multi-Core Programming

• Homogeneous multi-core: all cores are identical.
• Challenges:
  • How to split work across cores
  • Avoid race conditions
  • Synchronization and memory consistency

Heterogeneous Systems

• Mix of CPU + GPU + accelerators.

• GPU executes:

  • Large-scale parallel tasks
  • Single Instruction, Multiple Data (SIMD)

• CPU executes:

  • Control-heavy, serial tasks

System-on-Chip (SoC)

• Multiple components on a single chip:

  • CPU cores
  • GPU
  • DSP
  • Memory controllers
  • I/O

• Requires different programming models for each subsystem.

Virtualization Architectures

• One hardware → multiple virtual machines.
• Hypervisor enforces memory safety.
• Each VM sees its own virtual hardware.

Memory Safety

Memory safety ensures:

• One program cannot illegally access another’s memory.
• Prevents corruption, attacks, or accidental bugs.

Role of Virtual Memory

Based on segmentation + paging:

• Virtual pages → physical frames mapping
• Unauthorized access triggers page fault
• OS/hardware stops programs from corrupting each other

Why Needed?

If a program accesses another program’s memory:

• Could be intended (IPC)
• Could corrupt data
• Could be malicious

Who Enforces It?

• CPU hardware: MMU, page tables, permission bits
• OS kernel: manages virtual address space, allocates pages
• User program cannot bypass these mechanisms

---

Memory Segmentation & Paging

Introduction

Segmentation and paging originally built for:

• Caching data between main memory and disk
• Reducing address-translation overhead
• Providing memory protection

How It Works

• Logical address → virtual address
• MMU translates to physical frame
• If not present: page fault
• OS loads page from disk

As a Security Mechanism

Virtual memory doubles as:

• Isolation barrier
• Boundary around each program
• Prevents illegal reads/writes

---

Kernels (OS Kernels)

Introduction

A kernel controls:

• CPU scheduling
• Memory management
• Interrupt handling
• Page tables and MMU configuration

Privileged Instructions

Kernels manage:

• Virtual memory
• Hardware interrupts
• Device drivers
• Resource sharing

They run in privileged mode as defined in ISA (e.g., RISC-V Privileged Architecture).

---

System-on-Chip Programming Model

Introduction

SoCs pack various compute blocks together. Programming requires:

• Understanding of heterogeneous compute units
• Separate drivers, runtimes, firmware
• Coordination between CPU, GPU, peripherals

Key Ideas

• Work partitioning: which core does what?
• Communication and synchronization across components
• Memory safety between many sub-systems (DMA engines, accelerators, network units)

---

Summary

• Modern systems require models for:

  • Multi-threading
  • Multi-core execution
  • Heterogeneous devices (CPU+GPU)
  • Virtualized hardware
  • SoC platforms

• Memory safety is enforced through virtual memory, segmentation, paging, and OS kernel control.

• Virtual memory is both a performance and security mechanism.

• Key Concepts:

  • Single Program: A simple, linear flow of instructions executed by a processor.
  • Branching: Introduces decision points, allowing different instructions to be executed based on conditions.
  • Sub-routines/Function Calls: Enables modularization of code for better readability and reusability.

Multi-threaded and Multi-core Programming

• Multi-threaded Programs: Programs that execute multiple threads of control within a single program, improving efficiency and performance.
• Memory Safety: Ensuring that a program does not access unauthorized memory regions, preventing data corruption and system crashes.
• Multi-core Programming:
  • Homogeneous Multi-core Processors: Multiple identical cores, requiring synchronization and load balancing for optimal performance.
  • Heterogeneous Systems (GPU-CPU Co-processing): Combining CPUs and GPUs for parallel processing, offering significant performance improvements for specific workloads (e.g., graphics rendering, machine learning).
  • System On Chip (SoC): An integrated circuit that combines all components of a computer system into a single chip, often used in embedded systems and mobile devices.
  • Virtualization Architectures: Supports multiple virtual machines (VMs) running on the same hardware, enabling efficient resource sharing and isolation.

Memory Safety

• What is Memory Safety?: Refers to mechanisms that prevent a program from accessing memory regions that it is not authorized to access.

• Issues:

  • Unauthorized access can corrupt programs and tamper with data.

• Protection: Managed by the operating system through techniques like virtual memory and page protection.

• Virtual Memory: Virtual memory provides an abstraction of physical memory, allowing programs to use more memory than is physically available by swapping data between RAM and disk storage.

Memory Segmentation and Paging (Virtual Memory)

• Segmentation: Divides memory into segments based on logical divisions (e.g., code, data, stack).

• Paging: Divides memory into fixed-size pages, improving memory management and allocation.

  • Advantages:
    • Efficient memory use.
    • Security (each program can be isolated from others by mapping to different virtual pages).

• Caching Mechanism: Used between main memory and hard disk to speed up access and improve system performance.

• Security Mechanism: Helps prevent unauthorized memory access, ensuring program integrity.

Kernel and System Management

• Role of the Kernel: The kernel is the core component of an operating system that manages system resources and facilitates communication between hardware and software.
• System On Chip (SoC) Designs: In SoC architectures, the kernel manages the integration of processors, memory, and peripherals on a single chip.