Operating Systems: Process Concepts

• Process scheduling involves the ready queue where every process waits until the scheduler dispatches it for execution, and I/O queues for I/O requests.
• Key terms include uniprogramming (one process in RAM), multiprogramming (multiple processes in RAM), multitasking (multiple processes from different users), multiprocessing (multiple CPUs executing processes), and multithreading (one process with multiple threads).
• There are also batch systems that handle jobs and time-shared systems that handle user programs/tasks.

What is a Process

• A process is a program in execution, executing sequentially.
• There are job queues containing all processes, ready queues for processes in main memory, and device queues for processes waiting for I/O.
• Processes transition between new, ready, running, and waiting states.
• A program becomes a process when loaded into memory.
• Programs are executed via GUI, mouse clicks, or command-line entry.

Process Components

• After loading into memory, a process contains multiple parts.
• These parts include the stack (local variables, function parameters), heap (dynamically allocated memory), data (initialized global variables), uninitialized data, and text (program code).
• The stack and heap grow/shrink dynamically.
• A Process Control Block (PCB) stores all information about a process, facilitating its efficient execution.

PCB Contents

• The PCB contains the process state (running, waiting, etc.), process identification (unique number), program counter (next instruction location), CPU registers (process-specific data), CPU scheduling information (priority, scheduling queues), memory limits, and I/O status information.

Process States

• There are five states a process can have: new, ready, running, waiting, and termination.
• Context switching involves the CPU saving the current process state into its PCB and loading the new process state from its PCB.
• Complex OSs and PCBs increase context switching time.

Schedulers

• The short-term scheduler (CPU scheduler) decides which process should execute next and allocates CPU time rapidly (milliseconds).
• The long-term scheduler (job scheduler) selects which processes to put into the ready queue slowly (seconds, minutes).
• The medium-term scheduler decreases the degree of multiprogramming by swapping processes out of memory and storing them on disk, later bringing them back.

Long Term Scheduler Facts

• The long-term scheduler balances I/O-bound and CPU-bound processes.
• Imbalance can result in either low CPU utilization (if only I/O-bound processes) or an overloaded ready queue with an unused I/O queue (if only CPU-bound processes).

Process Types

• I/O bound processes spend more time waiting for I/O with short CPU bursts.
• CPU-bound processes spend more time doing computations with few, long CPU bursts.

Operations On Process

• Process creation involves a parent process creating child processes, forming a tree structure.
• Processes are identified and managed using process identifiers (PIDs).
• Children inherit resources from the parent.

Execution Options

• The parent and children can execute concurrently (multitasking), or the parent can wait until the child terminates.

Address Space Options

• Children can duplicate parent's address space or have a program loaded in it.

Process termination has three types

• When a process finishes executing, it calls the exit system call for notifying the OS
• When a parent process terminates, the children may terminate by an abort system call because exceeding resource allocation, assigned task no longer needed, the parent exits
• Zombie is when the child terminates and there isn't a parent waiting
• Orphan is when the parent terminates and does not invoke the wait cell

Process Communication

• Processes can be independent or cooperating, with cooperating processes affecting/being affected by others.
• Reasons for the use of cooperating process includes modularity, computation speed up, etc
• Cooperating processes need Interprocess Communication (IPC).

Interprocess Communication (IPC) Models:

• Shared Memory: Where multiple cooperating processes have shared memory region
  • It is hard to implement, difficult to set up
  • Requires system call to establish the shared memory, but not to exchange data
  • Used for large data on same system but is user-controlled
• Message Passing:
  • Where communicating processes exchange data
  • Easy to implement
  • It is slow because it uses system calls to exchange data because of kernel intervention
  • Used for small data across multiple systems

Message Passing

• Is implemented with synchronization, can lack lack synchronization
• Supports both fixed and variable size messages
• It is provided by the IPC send and receive messages over communication links

Communication Links Implementation

• Communication links are implemented via direct, indirect, synchronous, asynchronous, automatic, or explicit buffering methods.

Direct vs Indirect Communication

• Direct Communication: Processes must explicitly know each other's names.
  • Links are established automatically, associated with one pair, and can be unidirectional or bidirectional.
• Indirect Communication: Processes communicate through mailboxes (ports) with unique IDs.
  • Links are established through shared mailboxes, associated with several processes, and can be unidirectional or bidirectional.

Syncronization (Blocking vs Non-Blocking)

• Blocking (Synchronous):
  • Blocking send is where the sender is blocked until the message is received.
  • Blocking receive is where the receiver is blocked until a message is available.
  • Both can be combined in syncronization making it rendezvous
• Non-Blocking (Asynchronous):
  • Non-blocking send is where the sender sends a message and continues.
  • Non-blocking receive is where the receiver receives a valid or NULL message.

Buffering

• Buffering is a queue of messages attached to a link.
  • When capacity is at zero the sender has to wait for a receiver to obtain the messsage
  • When at a bounded message queue the sender waits till it goes through
  • When unbounded they never wait

Producer-Consumer Paradigm

• It involves a producer process to produce information, and a consumer to take the information
• Involves both bounded and unbounded buffer

Explore process scheduling, including ready and I/O queues. Key concepts covered are uniprogramming, multiprogramming, multitasking, multiprocessing, and multithreading. Learn about the lifecycle and components of a process within an operating system.