OS Processes

Understanding OS Processes

Evolution of OS Serial Processing Simple Batch System Multiprogrammed Batch

Systems Multitasking/Time

Sharing Design

Monolithic Layered systems Microkernel

Concept of Mechanisms (what/how to do?) And Policies (which?)

Process treats an OS as a set of APIs

CPU Virtualization an illusion by the OS

Process Process Control Block Role of Dispatcher 2-state process model

Creation/Termination 5-state process model

New, Ready, Run, Blocked/Waiting, Exit Trace and dispatcher Queuing model

Suspended Process Need for Swapping 6/7-state process model

Few other concepts Process preemption Process/Context

switching kernel, trap tables System/User level CPU / I/O-bound Cooperative / NC

OS Ctrl structures OS Ctrl Tables

Process Ctrl Structures Elements of Process Image PCB Process List Structures

PCB List based on states Process Control

Modes of Execution Protect sys. structures

Steps - Process Creation Process Switching Role of schedulers

Types of schedulers Mix of CPU and I/O

bound jobs



Sharing

Design Monolithic Layered systems Microkernel






switching System/User level CPU / I/O-bound Cooperative







bound jobs

New jobInteractive logon

Created for OS serviceSpawn by existing process



Sharing














bound jobs

Normal completionTime limit exceeded

Mem unavailableProtection errorArithmetic error

Time overrun (wait)I/O failure

Privileged instructionOS intervention

Parent terminationParent request



Sharing














bound jobs

Can affect the execution of other processes

Can be affected by other processes

Share data with other processes



Sharing














bound jobs

User Data program data, user stack etc.User Program Program to be executedStack call parameters, addressesPCB Data needed by OS to control the process



Sharing














bound jobs

PCBProcess Identification

(pid, ppid, uid)

Processor State Info.(user-visible registers,

Control/Status registers,Stack pointers)

Process Control Info.(scheduling and state info.,Process data structuring,

Inter-process comm.,Process privileges,

Memory mgmt.,Resource ownership

And utilization)



Sharing














bound jobsAssign a unique pidAllocate space for the process

Initialize the PCBSet the appropriate linkages

Create/expand system data structures

With Suspend State

7-State Process Model

Ready: Process is in Main Memory and ready for execution.Blocked: Process is in Main Memory and awaiting an event I/O.Blocked, Suspended: Process is in Secondary Memory and awaiting an event I/O.Ready, Suspended: Process is in Secondary memory and available for execution as soon as loaded in main memory

Process state-based Queuing Models

Process / CPU Scheduling Policies

Scheduling Goals

Scheduling Goals

Performance

Throughput Response Time

Many context switch not good for high throughput, but may be necessary for quick response time.

Throughput - Rate at which useful work is accomplished, e.g, search transaction on Google.

Remember: It is not only CPU, efficientuse of I/O, memory efficiently, a scheduler can maximize throughput.

Scheduling MechanismNon-preemptive vs. Preemptive

Process Parameters for SchedulingArrival Time (AT)

Service Time (ST) / Execution Time (ET) / Burst Time (BT)

Priority Static vs. Dynamic (No., reverse)

Finish Time (FT)

Turnaround time (TAT) and Mean TAT

Waiting Time (WT) and Mean WT

Exact Schedule (run order) of processes

Visualization Gantt Charts (or Time scale diagram)

Uniprocessor Scheduling

Focus: State

of Ready Queue

Policy dependent

Process / CPU Scheduling Policies

First Come First Serve (FCFS) / First In First Out (FIFO)

Shortest Process Next (SPN) / Shortest Job First (SJF)

Highest Response Ratio Next (HRRN)

Priority-based

Round Robin (RR)

Shortest Remaining Time Next (SRTN) / Preemptive Shortest Process Next (P-SPN)

Multi-level Queue (MLQ)

Multi-level Feedback Queue (MLFQ) (sometimes called Feedback scheduling)

Multi-level Queue (MLQ) Scheduling Multiple queues, No feedback Every queue can have a separate scheduling function

3 queues : RR (q=2), RR(q=3), RR(q=5) 3 queues : FCFS, SPN, SRTN (possible ordering: SPN, SRTN, FCFS).

It is important to consider the queue ordering.

Unless the first queue is empty, the next queue will not be considered in scheduling other processes.

In case of 3 queues, queue 2 not processed until queue 1 gets empty, queue 3 not processed until queues 1 and 2 get empty.

It is possible to place a process in one or more queues depending on the criteria (for e.g., execution time, memory usage, ratio of CPU and I/O usage) to assign a queue. However, there should be a policy to resolve any such conflict.

For e.g., FCFS (t > 10), SPN(t < 5), SRTN(5 < t < 10) In case different queue ordering is considered, it is possible to reach to

different schedules suggesting multiple ways to schedule a given set of processes.

Scheduler scans all queues based on queue ordering.

Example 1 : Schedule Processes

FCFS,

RR (q=1), RR(q=4),

SPN/SJF,

Priority-based

SRT/SRTN,

HRRN,

Multi-level Queue (q=1,2,3)

Feedback (q=1), (q=2^n)

Process priority

A (3), B(1), C(0), D(1), E(2)

Solution Example 1: Schedule of Processes

Example 2

Example 3FCFS,

RR (q=1),

RR(q=4),

SPN/SJF,

SRT/SRTN,

HRRN,

Multi-level Feedback (q=2),

Multi-level Feedback (RR=1,SRT, HRRN)

Remember:For HRRN (is preemptive when time quantum is given) when used in MLFQ schedulingResponse Ratio = (WT + RST) / RST, where WT indicates waiting time w.r.to arrival time, andRST indicates remaining service time based on status of a process in a particular queue. RST = Total service time time for which CPU has already been assigned to that process.

When policies like SJF, SRT, HRRN are used in context of feedback scheduling, each of these policies are only used for process selection from the queue. However, CPU time is given only for a given/assumed time slot, and then the process is moved to next queue if it is not yet completed.

Example 4

b) Draw a chart showing execution of processes using SRTN. For each process, calculate the total time spent in the Ready queue.

c) Draw a chart showing execution of processes using HRRN.

d) Draw charts for the MLFQ (4 queues, q=2n+1, n=0,1,2,3)

Example 5

Example 6

FCFSRR(q=1,2,4)SPN(Shortest Process Next, Non-preemptive)SRTN (a.k.a. PSPN(Shortest Process Next, preemptive))HRRNPreemptive HRRNMLFQ (q=2n+1, n={0,0.5,1,1.5})

Solution 6

Solution 6 ...cont'd

RR (q = 1)

Solution 6...cont'd

Example 7

(a) The system uses RR scheduling with q=10 ms and q=5 ms.(b) The system uses MLFQ scheduling with 4 queues, q=2n-1, where n=1,2,3,4

Solution 7 (a)

Blocked (I/O) QueueCPU UseTime Ready Queue(end of time slot)

I/O Use

Example 8Consider the following three processes. Each process makes a CPU burst

then an I/O burst, another CPU burst, another I/O burst and terminates with a CPU burst. The lengths of the CPU burst and I/O burst times in milliseconds are given in the following table:

Process CPU-burst1 I/O-burst1 CPU-burst2 I/O-burst2 CPU-burst3 Arrival

P1 2 4 2 2 2 0

P2 2 2 3 3 1 1

P3 1 2 1 1 1 1

The processes are assumed to arrive as indicated.

Draw charts that illustrate the execution of theses processes using the (1) FCFS (2) RR (q = 1) (3) RR ( q = 2) (4) SJF (5) HRRN (6) SRTN (7) MLFQ (4 queues, q=2k, k=1,1.5,2,2.5).

Clearly indicate the queue state, and use the charts to compute FT,WT and its WT, TAT and its mean. Give your conclusion about the comparative performance of the above listed policies.

If an I/O completion and a CPU timeout of two processes occur at the same time, we treat the I/O completion first.

Soluiton 8

Solution 8 ... cont'd

Example 9 (typical objective type questions)

CPU Scheduling Algorithms

Slide 1Slide 2Slide 3Slide 4Slide 5Slide 6Slide 7Slide 8Slide 9Slide 10Slide 11Slide 12Slide 13Slide 16Slide 17Slide 20Slide 21Slide 22Slide 25Slide 27Slide 32Slide 33Slide 34Slide 35Slide 36

OS Processes

Documents

Transcript of OS Processes