System Design (LLD)Concurrency Fundamentals

Threads and Processes

LevelIntermediate

Duration75 mins

TopicConcurrency Fundamentals

1 / 4

Process: The Independent Execution Environment

Understanding the Foundation of Program Execution

Before any line of code you write can manipulate data, before any algorithm can sort a list, before any design pattern can structure your logic—there must be a process. The process is the vessel that carries your program from static bytes on disk to living, breathing computation in memory.

Yet most developers treat processes as invisible infrastructure—like the air we breathe. We launch applications, run tests, deploy services, without deeply understanding the elegant machinery that makes program execution possible. This understanding becomes critical when you need to:

Debug memory issues that only appear in production
Understand why your application's performance degrades under load
Design systems that efficiently share resources across components
Implement concurrent processing without creating chaos

What You Will Master

By the end of this page, you will understand processes not as abstract operating system concepts, but as concrete execution environments with predictable structure, resource constraints, and behavioral characteristics. This understanding forms the foundation for everything that follows in concurrent programming.

What Is a Process?

A process is an instance of a program in execution. This definition, while accurate, doesn't capture the profound architectural implications. Let's unpack it systematically.

From Static to Dynamic:

A program is a passive entity—executable code stored on disk. When you double-click an application or run a command in your terminal, the operating system performs a remarkable transformation:

Loading: The executable's code and initial data are copied from disk into main memory
Allocation: The OS allocates necessary memory regions (stack, heap, data segments)
Initialization: Runtime environment is configured (environment variables, file descriptors, security context)
Scheduling: The process is placed in the ready queue, awaiting CPU time

The result is a process—a dynamic, active entity that consumes resources, executes instructions, and interacts with the operating system.

Program vs Process: The Fundamental Distinction
Aspect	Program	Process
Nature	Static, passive	Dynamic, active
Storage	On disk (filesystem)	In memory (RAM)
State	Unchanging until modified	Constantly evolving during execution
Resources	None consumed	CPU, memory, I/O, handles
Lifetime	Permanent until deleted	Exists only during execution
Instances	One file	Multiple concurrent instances possible
Identity	File path	Process ID (PID)

The Restaurant Analogy

Think of a program as a recipe written in a cookbook, and a process as the actual cooking session executing that recipe. The recipe is static—it never changes. But each cooking session (process) is unique: it uses specific ingredients (resources), follows the recipe at its own pace (CPU scheduling), and eventually produces an outcome (program output). You can have multiple chefs cooking the same recipe simultaneously—multiple processes from the same program.

Process Memory Layout: The Architecture of Execution

Every process receives its own virtual address space—a private, isolated region of memory that appears to the process as if it owns the entire machine. This virtualization is one of the operating system's most elegant illusions. Understanding the layout of this address space is essential for systems programming.

The Classical Memory Model:

A process's virtual address space is divided into distinct segments, each with specific purposes and properties:

process-memory-layout.txt
┌─────────────────────────────────────────┐ High Memory Address
│                                         │
│              KERNEL SPACE               │ ← Not accessible to user process
│         (System calls, drivers)         │
│                                         │
├─────────────────────────────────────────┤
│                                         │
│               STACK                     │ ← Grows downward ↓
│     (Local variables, return addrs,     │
│      function parameters, frames)       │
│             ↓ ↓ ↓ ↓ ↓                   │
├ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─┤
│                                         │
│          (Unused/Unmapped)              │ ← Gap between stack and heap
│                                         │
├ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─┤
│             ↑ ↑ ↑ ↑ ↑                   │
│               HEAP                      │ ← Grows upward ↑
│      (Dynamic memory allocation:        │
│       malloc, new, garbage collected)   │
│                                         │
├─────────────────────────────────────────┤
│            BSS SEGMENT                  │
│  (Uninitialized global/static vars)     │
├─────────────────────────────────────────┤
│           DATA SEGMENT                  │
│   (Initialized global/static vars)      │
├─────────────────────────────────────────┤
│           TEXT SEGMENT                  │
│    (Executable code - read-only)        │
│                                         │
└─────────────────────────────────────────┘ Low Memory Address

Understanding Each Memory Segment

•Text Segment (Code) — Contains the compiled machine instructions of your program. Marked as read-only and executable. Shared across multiple instances of the same program to conserve memory.
•Data Segment — Stores initialized global and static variables. For example: static int counter = 100; lives here. Loaded from the executable at process creation.
•BSS Segment — Contains uninitialized global and static variables. The OS zeros this memory at process creation. Named from historical "Block Started by Symbol" terminology.
•Heap — Region for dynamic memory allocation. Grows upward (toward higher addresses) as you call malloc(), new, or allocate objects. Managed either manually or by garbage collectors.
•Stack — Stores local variables, function parameters, return addresses, and saved registers. Grows downward (toward lower addresses) as functions call other functions. Each function invocation creates a new stack frame.
•Kernel Space — Reserved for the operating system. User processes cannot directly access this region—attempts trigger protection faults. System calls bridge this boundary.

Stack vs Heap: Critical Trade-offs

The stack is fast but limited (typically 1-8 MB per thread). Allocation is trivial—just move the stack pointer. The heap is larger but slower, requiring complex bookkeeping. Choosing between stack and heap allocation affects performance, locality, and lifetime management. Understanding this trade-off is fundamental to memory-conscious programming.

Process Control Block: The Process's Identity Card

The operating system manages thousands of processes simultaneously. To track each process's state, resources, and execution context, the OS maintains a data structure called the Process Control Block (PCB)—sometimes called the Task Control Block.

Think of the PCB as the complete administrative record for a process. Every time the OS needs to make a decision about a process—whether to run it, pause it, terminate it, or allocate resources to it—the PCB provides the necessary information.

process-control-block.ts
TypeScript
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
// Conceptual representation of a Process Control Block
interface ProcessControlBlock {
    // === PROCESS IDENTIFICATION ===
    processId: number;                    // Unique PID
    parentProcessId: number;              // PPID - who created this process
    userId: number;                       // Owner/creator user ID
    groupId: number;                      // Process group ID
 
    // === PROCESS STATE ===
    state: ProcessState;                  // Current execution state
    priority: number;                     // Scheduling priority
    schedulingInfo: SchedulingData;       // Quantum remaining, queue position
 
    // === CPU CONTEXT (saved during context switches) ===
    programCounter: number;               // Address of next instruction
    stackPointer: number;                 // Current top of stack
    generalRegisters: RegisterSet;        // CPU register values
    statusFlags: CPUFlags;                // Condition codes, mode bits
 
    // === MEMORY MANAGEMENT ===
    pageTableBaseRegister: number;        // Points to page table
    memoryLimits: MemoryBounds;           // Valid address range
    segmentTable: SegmentDescriptor[];    // Segment information
 
    // === I/O AND FILE SYSTEM ===
    openFileDescriptors: FileDescriptor[];// Array of open files
    currentWorkingDirectory: string;      // CWD path
    rootDirectory: string;                // (for chroot)
 
    // === ACCOUNTING AND STATISTICS ===
    cpuTimeUsed: number;                  // Total CPU time consumed
    creationTime: Date;                   // When process was created
    ioStatistics: IOStats;                // Read/write counts
 
    // === INTER-PROCESS COMMUNICATION ===
    signalMask: SignalMask;               // Blocked signals
    pendingSignals: Signal[];             // Signals awaiting delivery
    messageQueue: Message[];              // IPC messages
}
 
enum ProcessState {
    NEW = 'NEW',                          // Being created
    READY = 'READY',                      // Waiting to be scheduled
    RUNNING = 'RUNNING',                  // Currently executing
    WAITING = 'WAITING',                  // Blocked on I/O or event
    TERMINATED = 'TERMINATED'             // Finished execution
}

Why the PCB is Central to Operating System Design:

The PCB enables the OS's most critical operations:

Context Switching: When the OS switches from one process to another, it saves the current process's CPU state into its PCB, then loads the new process's state from its PCB. This happens thousands of times per second.
Scheduling Decisions: The scheduler examines PCBs to determine which process should run next based on priority, waiting time, and resource availability.
Resource Tracking: The PCB records all resources held by a process, enabling proper cleanup at termination and deadlock detection.
Process Hierarchy: Parent-child relationships tracked in PCBs enable tree-structured process management (e.g., killing a parent can cascade to children).

Real-World Inspection

On Linux, you can inspect PCB-equivalent information through /proc/<PID>/. For example, /proc/1234/status shows process state, /proc/1234/maps shows memory layout, and /proc/1234/fd/ lists open file descriptors. On Windows, tools like Process Explorer expose similar details through the Windows API.

Process States and Transitions

A process doesn't simply "run"—it moves through a well-defined state machine as it interacts with the operating system and competes for resources. Understanding these states reveals how the OS orchestrates concurrent execution.

Converting Mermaid diagram...

The Five Fundamental Process States

•NEW (Created) — The process is being created. The OS is allocating resources, setting up the PCB, and loading the program. The process cannot run yet.
•READY — The process is fully initialized and waiting to be assigned to a CPU. It has everything it needs to run—memory, file handles, initial state—except CPU time.
•RUNNING — The process is actively executing on a CPU core. Its instructions are being fetched, decoded, and executed. Only one process per CPU core can be in this state.
•WAITING (Blocked) — The process cannot proceed until some external event occurs—typically I/O completion, signal arrival, or resource availability. It's removed from scheduling consideration.
•TERMINATED — Execution has finished (normally via exit() or abnormally via signal). The process is awaiting cleanup. The PCB remains until the parent collects the exit status.

State Transitions and Their Triggers
Transition	Trigger	Who Initiates	Example
New → Ready	Initialization complete	OS/Loader	Program loaded, memory allocated
Ready → Running	Scheduler dispatch	OS Scheduler	Process selected for CPU time
Running → Ready	Preemption	OS (timer interrupt)	Time slice expired
Running → Waiting	I/O or resource request	Process	read() from disk, lock acquisition
Waiting → Ready	Event completion	Hardware/OS	Disk read complete, lock released
Running → Terminated	Exit or signal	Process or OS	exit(0), segfault, kill signal

The Critical Insight: Why Waiting Matters

The WAITING state is why multiprogramming works. When a process waits for (slow) I/O, the CPU can run another process. Without this, each I/O operation would waste millions of CPU cycles. The OS transforms idle time into productive computation—the foundation of responsive systems.

Process Creation: Bringing Processes Into Existence

How does a new process come into being? The mechanisms differ by operating system but share common patterns. Understanding process creation reveals important design decisions and performance implications.

Unix systems use a two-step process creation model that elegantly separates duplication from replacement:

Step 1: fork() — Creates an exact copy of the current process Step 2: exec() — Replaces the process image with a new program

C
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
#include <stdio.h>
#include <unistd.h>
#include <sys/wait.h>
 
int main() {
    printf("Parent process (PID: %d)\n", getpid());
    
    pid_t pid = fork();  // Create child process
    
    if (pid < 0) {
        // Error: fork failed
        perror("fork failed");
        return 1;
    } 
    else if (pid == 0) {
        // Child process: fork() returns 0
        printf("Child process (PID: %d, Parent PID: %d)\n", 
               getpid(), getppid());
        
        // Replace this process with a new program
        char *args[] = {"ls", "-la", NULL};
        execvp("ls", args);
        
        // If exec succeeds, this line never runs
        perror("exec failed");
        return 1;
    } 
    else {
        // Parent process: fork() returns child's PID
        printf("Parent: created child with PID %d\n", pid);
        
        int status;
        waitpid(pid, &status, 0);  // Wait for child to complete
        
        if (WIFEXITED(status)) {
            printf("Parent: child exited with status %d\n", 
                   WEXITSTATUS(status));
        }
    }
    
    return 0;
}

Why Two Steps?

This design provides remarkable flexibility:

File Redirection: Between fork() and exec(), you can redirect stdin/stdout to files or pipes
Environment Setup: Modify environment variables before the new program runs
Privilege Changes: Drop privileges (setuid) before executing untrusted code
Resource Limits: Adjust rlimits for the child before execution

Copy-On-Write (COW) Optimization:

Modern fork() doesn't actually copy all memory immediately. Both parent and child share the same physical pages, marked read-only. Only when either attempts to write does the OS copy that specific page. This makes fork() extremely fast, even for processes with gigabytes of memory.

Process Isolation: The Walls Between Programs

One of the most important properties of processes is isolation. Each process operates in its own protected environment, unable to directly observe or corrupt other processes. This isolation is fundamental to system stability and security.

What Isolation Provides

•Memory Protection — One process cannot read or write another's memory. Invalid memory access triggers a segmentation fault.
•CPU Protection — Time slicing ensures no process can monopolize the CPU indefinitely.
•Resource Accounting — Each process's resource usage is tracked separately for limits and billing.
•Failure Containment — A crash in one process doesn't bring down others or the system.
•Security Boundaries — Processes can run with different permissions, preventing privilege escalation.

The Cost of Isolation

•Memory Overhead — Each process needs its own page tables, kernel structures, and metadata.
•Context Switch Cost — Switching between processes is expensive (TLB flush, cache pollution).
•Communication Overhead — Inter-process communication requires explicit mechanisms (IPC).
•Creation Cost — Starting a new process is heavyweight compared to threads.
•Data Sharing Complexity — Sharing state between processes requires special facilities.

How Isolation Is Enforced:

The CPU and OS work together to enforce isolation through several mechanisms:

1. Virtual Memory and Page Tables Each process has its own page table mapping virtual to physical addresses. Even if two processes use the same virtual address, they access different physical memory. The MMU (Memory Management Unit) enforces this in hardware.

2. Privilege Rings (Protection Rings) CPUs operate in different privilege modes. User processes run in Ring 3 (least privileged), while the OS kernel runs in Ring 0 (most privileged). Privileged operations from Ring 3 trigger traps to the kernel.

3. System Call Interface Processes cannot directly execute privileged operations. They must request services through system calls, where the kernel validates and mediates all requests.

The Isolation Spectrum

Process isolation sits on a spectrum. Containers (Docker) add namespace isolation on top of processes. Virtual machines provide even stronger isolation with separate kernels. Choosing the right isolation level depends on your security requirements, performance needs, and operational constraints.

Inter-Process Communication (IPC)

Despite isolation, processes often need to cooperate. Inter-Process Communication (IPC) mechanisms provide controlled channels for data exchange across process boundaries. Each mechanism offers different trade-offs between performance, complexity, and suitability for different use cases.

IPC Mechanisms Comparison
Mechanism	Communication Type	Performance	Best For
Pipes	Unidirectional, streaming	High (in-kernel buffer)	Parent-child communication, shell pipelines
Named Pipes (FIFOs)	Bidirectional, streaming	High	Unrelated processes on same machine
Message Queues	Message-based, async	Medium	Decoupled producers/consumers
Shared Memory	Direct memory access	Highest	Large data, low latency requirements
Sockets	Bidirectional, networked	Medium-Low	Network communication, flexibility
Signals	Notifications only	Very High	Interrupts, process control
Memory-Mapped Files	File-backed shared memory	High	Persistent shared state, large datasets

ipc-example.ts
TypeScript (Node.js)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
import { fork } from 'child_process';
import { cpus } from 'os';
 
/**
 * Example: Using IPC to distribute work across child processes
 * Demonstrates Node.js's built-in IPC over pipes
 */
interface WorkResult {
    workerId: number;
    result: number;
    processId: number;
}
 
function distributedCalculation(items: number[]): Promise<WorkResult[]> {
    return new Promise((resolve) => {
        const numWorkers = cpus().length;
        const chunkSize = Math.ceil(items.length / numWorkers);
        const results: WorkResult[] = [];
        let completed = 0;
 
        for (let i = 0; i < numWorkers; i++) {
            const start = i * chunkSize;
            const chunk = items.slice(start, start + chunkSize);
            
            if (chunk.length === 0) continue;
 
            // Fork a child process - automatically sets up IPC channel
            const worker = fork('./worker.ts');
            
            // Send work to child via IPC
            worker.send({ workerId: i, data: chunk });
            
            // Receive results from child via IPC
            worker.on('message', (result: WorkResult) => {
                results.push(result);
                completed++;
                
                if (completed === numWorkers) {
                    resolve(results);
                }
            });
        }
    });
}
 
// worker.ts - runs in child process
process.on('message', (message: { workerId: number; data: number[] }) => {
    const { workerId, data } = message;
    
    // Perform computation
    const result = data.reduce((sum, n) => sum + n * n, 0);
    
    // Send result back to parent via IPC
    process.send?.({
        workerId,
        result,
        processId: process.pid
    });
    
    process.exit(0);
});

IPC Adds Complexity

Every IPC mechanism introduces challenges: synchronization overhead, potential for deadlock, serialization costs, and error handling for communication failures. The isolation that makes processes robust also makes their cooperation complex. This is a key motivation for threads, which we'll explore next.

Summary: The Process Mental Model

We've built a comprehensive understanding of processes as the fundamental unit of program execution. Let's consolidate the key concepts:

Key Takeaways

•A process is a program in execution — It transforms static code into dynamic computation with its own resources and execution context.
•Virtual address space provides isolation — Each process sees its own private memory layout (text, data, heap, stack) even when sharing physical memory.
•The PCB tracks everything — Process ID, state, CPU registers, memory mappings, open files—all maintained by the OS for management.
•Processes move through states — NEW → READY → RUNNING → WAITING → TERMINATED, driven by scheduling and events.
•Creation models differ — Unix favors fork()+exec() for flexibility; Windows uses monolithic CreateProcess().
•Isolation enables stability — Memory protection, CPU scheduling, and resource accounting prevent cascading failures.
•IPC bridges isolation — Pipes, shared memory, sockets, and other mechanisms enable controlled cooperation.

What's Next:

With processes understood, we're ready to explore their lightweight cousins: threads. Where processes provide isolation at the cost of overhead, threads provide concurrency within a single process—sharing memory, file handles, and other resources while maintaining separate execution contexts. This trade-off is fundamental to concurrent system design.

Page Complete

You now understand processes as independent execution environments—their structure, lifecycle, resource management, and communication mechanisms. This foundation is essential for grasping how threads work within processes and why the process-thread distinction matters for concurrent programming.

1 / 4

Loading learning content...

System Design (LLD)Concurrency Fundamentals

Threads and Processes

LevelIntermediate

Duration75 mins

TopicConcurrency Fundamentals

1 / 4

Process: The Independent Execution Environment

Understanding the Foundation of Program Execution

Debug memory issues that only appear in production
Understand why your application's performance degrades under load
Design systems that efficiently share resources across components
Implement concurrent processing without creating chaos

What You Will Master

What Is a Process?

A process is an instance of a program in execution. This definition, while accurate, doesn't capture the profound architectural implications. Let's unpack it systematically.

From Static to Dynamic:

A program is a passive entity—executable code stored on disk. When you double-click an application or run a command in your terminal, the operating system performs a remarkable transformation:

Loading: The executable's code and initial data are copied from disk into main memory
Allocation: The OS allocates necessary memory regions (stack, heap, data segments)
Initialization: Runtime environment is configured (environment variables, file descriptors, security context)
Scheduling: The process is placed in the ready queue, awaiting CPU time

The result is a process—a dynamic, active entity that consumes resources, executes instructions, and interacts with the operating system.

Program vs Process: The Fundamental Distinction
Aspect	Program	Process
Nature	Static, passive	Dynamic, active
Storage	On disk (filesystem)	In memory (RAM)
State	Unchanging until modified	Constantly evolving during execution
Resources	None consumed	CPU, memory, I/O, handles
Lifetime	Permanent until deleted	Exists only during execution
Instances	One file	Multiple concurrent instances possible
Identity	File path	Process ID (PID)

The Restaurant Analogy

Process Memory Layout: The Architecture of Execution

The Classical Memory Model:

A process's virtual address space is divided into distinct segments, each with specific purposes and properties:

process-memory-layout.txt
┌─────────────────────────────────────────┐ High Memory Address
│                                         │
│              KERNEL SPACE               │ ← Not accessible to user process
│         (System calls, drivers)         │
│                                         │
├─────────────────────────────────────────┤
│                                         │
│               STACK                     │ ← Grows downward ↓
│     (Local variables, return addrs,     │
│      function parameters, frames)       │
│             ↓ ↓ ↓ ↓ ↓                   │
├ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─┤
│                                         │
│          (Unused/Unmapped)              │ ← Gap between stack and heap
│                                         │
├ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─┤
│             ↑ ↑ ↑ ↑ ↑                   │
│               HEAP                      │ ← Grows upward ↑
│      (Dynamic memory allocation:        │
│       malloc, new, garbage collected)   │
│                                         │
├─────────────────────────────────────────┤
│            BSS SEGMENT                  │
│  (Uninitialized global/static vars)     │
├─────────────────────────────────────────┤
│           DATA SEGMENT                  │
│   (Initialized global/static vars)      │
├─────────────────────────────────────────┤
│           TEXT SEGMENT                  │
│    (Executable code - read-only)        │
│                                         │
└─────────────────────────────────────────┘ Low Memory Address

Understanding Each Memory Segment

•Text Segment (Code) — Contains the compiled machine instructions of your program. Marked as read-only and executable. Shared across multiple instances of the same program to conserve memory.
•Data Segment — Stores initialized global and static variables. For example: static int counter = 100; lives here. Loaded from the executable at process creation.
•BSS Segment — Contains uninitialized global and static variables. The OS zeros this memory at process creation. Named from historical "Block Started by Symbol" terminology.
•Heap — Region for dynamic memory allocation. Grows upward (toward higher addresses) as you call malloc(), new, or allocate objects. Managed either manually or by garbage collectors.
•Stack — Stores local variables, function parameters, return addresses, and saved registers. Grows downward (toward lower addresses) as functions call other functions. Each function invocation creates a new stack frame.
•Kernel Space — Reserved for the operating system. User processes cannot directly access this region—attempts trigger protection faults. System calls bridge this boundary.

Stack vs Heap: Critical Trade-offs

Process Control Block: The Process's Identity Card

process-control-block.ts
TypeScript
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
// Conceptual representation of a Process Control Block
interface ProcessControlBlock {
    // === PROCESS IDENTIFICATION ===
    processId: number;                    // Unique PID
    parentProcessId: number;              // PPID - who created this process
    userId: number;                       // Owner/creator user ID
    groupId: number;                      // Process group ID
 
    // === PROCESS STATE ===
    state: ProcessState;                  // Current execution state
    priority: number;                     // Scheduling priority
    schedulingInfo: SchedulingData;       // Quantum remaining, queue position
 
    // === CPU CONTEXT (saved during context switches) ===
    programCounter: number;               // Address of next instruction
    stackPointer: number;                 // Current top of stack
    generalRegisters: RegisterSet;        // CPU register values
    statusFlags: CPUFlags;                // Condition codes, mode bits
 
    // === MEMORY MANAGEMENT ===
    pageTableBaseRegister: number;        // Points to page table
    memoryLimits: MemoryBounds;           // Valid address range
    segmentTable: SegmentDescriptor[];    // Segment information
 
    // === I/O AND FILE SYSTEM ===
    openFileDescriptors: FileDescriptor[];// Array of open files
    currentWorkingDirectory: string;      // CWD path
    rootDirectory: string;                // (for chroot)
 
    // === ACCOUNTING AND STATISTICS ===
    cpuTimeUsed: number;                  // Total CPU time consumed
    creationTime: Date;                   // When process was created
    ioStatistics: IOStats;                // Read/write counts
 
    // === INTER-PROCESS COMMUNICATION ===
    signalMask: SignalMask;               // Blocked signals
    pendingSignals: Signal[];             // Signals awaiting delivery
    messageQueue: Message[];              // IPC messages
}
 
enum ProcessState {
    NEW = 'NEW',                          // Being created
    READY = 'READY',                      // Waiting to be scheduled
    RUNNING = 'RUNNING',                  // Currently executing
    WAITING = 'WAITING',                  // Blocked on I/O or event
    TERMINATED = 'TERMINATED'             // Finished execution
}

Why the PCB is Central to Operating System Design:

The PCB enables the OS's most critical operations:

Context Switching: When the OS switches from one process to another, it saves the current process's CPU state into its PCB, then loads the new process's state from its PCB. This happens thousands of times per second.
Scheduling Decisions: The scheduler examines PCBs to determine which process should run next based on priority, waiting time, and resource availability.
Resource Tracking: The PCB records all resources held by a process, enabling proper cleanup at termination and deadlock detection.
Process Hierarchy: Parent-child relationships tracked in PCBs enable tree-structured process management (e.g., killing a parent can cascade to children).

Real-World Inspection

Process States and Transitions

Converting Mermaid diagram...

The Five Fundamental Process States

•NEW (Created) — The process is being created. The OS is allocating resources, setting up the PCB, and loading the program. The process cannot run yet.
•READY — The process is fully initialized and waiting to be assigned to a CPU. It has everything it needs to run—memory, file handles, initial state—except CPU time.
•RUNNING — The process is actively executing on a CPU core. Its instructions are being fetched, decoded, and executed. Only one process per CPU core can be in this state.
•WAITING (Blocked) — The process cannot proceed until some external event occurs—typically I/O completion, signal arrival, or resource availability. It's removed from scheduling consideration.
•TERMINATED — Execution has finished (normally via exit() or abnormally via signal). The process is awaiting cleanup. The PCB remains until the parent collects the exit status.

State Transitions and Their Triggers
Transition	Trigger	Who Initiates	Example
New → Ready	Initialization complete	OS/Loader	Program loaded, memory allocated
Ready → Running	Scheduler dispatch	OS Scheduler	Process selected for CPU time
Running → Ready	Preemption	OS (timer interrupt)	Time slice expired
Running → Waiting	I/O or resource request	Process	read() from disk, lock acquisition
Waiting → Ready	Event completion	Hardware/OS	Disk read complete, lock released
Running → Terminated	Exit or signal	Process or OS	exit(0), segfault, kill signal

The Critical Insight: Why Waiting Matters

Process Creation: Bringing Processes Into Existence

Unix systems use a two-step process creation model that elegantly separates duplication from replacement:

Step 1: fork() — Creates an exact copy of the current process Step 2: exec() — Replaces the process image with a new program

C
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
#include <stdio.h>
#include <unistd.h>
#include <sys/wait.h>
 
int main() {
    printf("Parent process (PID: %d)\n", getpid());
    
    pid_t pid = fork();  // Create child process
    
    if (pid < 0) {
        // Error: fork failed
        perror("fork failed");
        return 1;
    } 
    else if (pid == 0) {
        // Child process: fork() returns 0
        printf("Child process (PID: %d, Parent PID: %d)\n", 
               getpid(), getppid());
        
        // Replace this process with a new program
        char *args[] = {"ls", "-la", NULL};
        execvp("ls", args);
        
        // If exec succeeds, this line never runs
        perror("exec failed");
        return 1;
    } 
    else {
        // Parent process: fork() returns child's PID
        printf("Parent: created child with PID %d\n", pid);
        
        int status;
        waitpid(pid, &status, 0);  // Wait for child to complete
        
        if (WIFEXITED(status)) {
            printf("Parent: child exited with status %d\n", 
                   WEXITSTATUS(status));
        }
    }
    
    return 0;
}

Why Two Steps?

This design provides remarkable flexibility:

File Redirection: Between fork() and exec(), you can redirect stdin/stdout to files or pipes
Environment Setup: Modify environment variables before the new program runs
Privilege Changes: Drop privileges (setuid) before executing untrusted code
Resource Limits: Adjust rlimits for the child before execution

Copy-On-Write (COW) Optimization:

Process Isolation: The Walls Between Programs

What Isolation Provides

•Memory Protection — One process cannot read or write another's memory. Invalid memory access triggers a segmentation fault.
•CPU Protection — Time slicing ensures no process can monopolize the CPU indefinitely.
•Resource Accounting — Each process's resource usage is tracked separately for limits and billing.
•Failure Containment — A crash in one process doesn't bring down others or the system.
•Security Boundaries — Processes can run with different permissions, preventing privilege escalation.

The Cost of Isolation

•Memory Overhead — Each process needs its own page tables, kernel structures, and metadata.
•Context Switch Cost — Switching between processes is expensive (TLB flush, cache pollution).
•Communication Overhead — Inter-process communication requires explicit mechanisms (IPC).
•Creation Cost — Starting a new process is heavyweight compared to threads.
•Data Sharing Complexity — Sharing state between processes requires special facilities.

How Isolation Is Enforced:

The CPU and OS work together to enforce isolation through several mechanisms:

3. System Call Interface Processes cannot directly execute privileged operations. They must request services through system calls, where the kernel validates and mediates all requests.

The Isolation Spectrum

Inter-Process Communication (IPC)

IPC Mechanisms Comparison
Mechanism	Communication Type	Performance	Best For
Pipes	Unidirectional, streaming	High (in-kernel buffer)	Parent-child communication, shell pipelines
Named Pipes (FIFOs)	Bidirectional, streaming	High	Unrelated processes on same machine
Message Queues	Message-based, async	Medium	Decoupled producers/consumers
Shared Memory	Direct memory access	Highest	Large data, low latency requirements
Sockets	Bidirectional, networked	Medium-Low	Network communication, flexibility
Signals	Notifications only	Very High	Interrupts, process control
Memory-Mapped Files	File-backed shared memory	High	Persistent shared state, large datasets

ipc-example.ts
TypeScript (Node.js)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
import { fork } from 'child_process';
import { cpus } from 'os';
 
/**
 * Example: Using IPC to distribute work across child processes
 * Demonstrates Node.js's built-in IPC over pipes
 */
interface WorkResult {
    workerId: number;
    result: number;
    processId: number;
}
 
function distributedCalculation(items: number[]): Promise<WorkResult[]> {
    return new Promise((resolve) => {
        const numWorkers = cpus().length;
        const chunkSize = Math.ceil(items.length / numWorkers);
        const results: WorkResult[] = [];
        let completed = 0;
 
        for (let i = 0; i < numWorkers; i++) {
            const start = i * chunkSize;
            const chunk = items.slice(start, start + chunkSize);
            
            if (chunk.length === 0) continue;
 
            // Fork a child process - automatically sets up IPC channel
            const worker = fork('./worker.ts');
            
            // Send work to child via IPC
            worker.send({ workerId: i, data: chunk });
            
            // Receive results from child via IPC
            worker.on('message', (result: WorkResult) => {
                results.push(result);
                completed++;
                
                if (completed === numWorkers) {
                    resolve(results);
                }
            });
        }
    });
}
 
// worker.ts - runs in child process
process.on('message', (message: { workerId: number; data: number[] }) => {
    const { workerId, data } = message;
    
    // Perform computation
    const result = data.reduce((sum, n) => sum + n * n, 0);
    
    // Send result back to parent via IPC
    process.send?.({
        workerId,
        result,
        processId: process.pid
    });
    
    process.exit(0);
});

IPC Adds Complexity

Summary: The Process Mental Model

We've built a comprehensive understanding of processes as the fundamental unit of program execution. Let's consolidate the key concepts:

Key Takeaways

•A process is a program in execution — It transforms static code into dynamic computation with its own resources and execution context.
•Virtual address space provides isolation — Each process sees its own private memory layout (text, data, heap, stack) even when sharing physical memory.
•The PCB tracks everything — Process ID, state, CPU registers, memory mappings, open files—all maintained by the OS for management.
•Processes move through states — NEW → READY → RUNNING → WAITING → TERMINATED, driven by scheduling and events.
•Creation models differ — Unix favors fork()+exec() for flexibility; Windows uses monolithic CreateProcess().
•Isolation enables stability — Memory protection, CPU scheduling, and resource accounting prevent cascading failures.
•IPC bridges isolation — Pipes, shared memory, sockets, and other mechanisms enable controlled cooperation.

What's Next:

Page Complete

1 / 4