Introduction OS act as an interface between users program and computer hardware Purpose : to provide environment in which we can execute program It is the only program running at all times on the computer called kernel OS also called- control, monitor, executive, supervisor

A program modules that acts as an intermediately between user of the computer and computer hardware Denotes program modules that govern the control of equipment resources such as processors, main storage, I/O devices and files Computer system divided into four components Hardware Operating system Application program UsersImportance of OS To control the computer operation To allocate resources To improve efficiency Decrease the cost Basic concepts and terminology Computer hardware terminology Programming terminology Operating system terminologyComputer hardware terminology Instruction and data stored in core memory or main memory Connected to main memory is processors that interpret the instructions and perform the operation CPU- a processor that manipulate and performs arithmetic operation upon data Control unit: h/w requires control circuitry OS decides the accessibility to resourcesProgramming terminology Software: collection of programs or data that used to perform certain task Program: sequence of instructions , placed into memory and interpreted by processor Stored in secondary storage when not in use. Prepackaged programs routines to perform predefined task e.g. square roots, sortingOperating system terminology User Job Process Job step Address space Pure code Multiprogramming Privileged instructions Protection hardware Interrupt hardware User : anybody that desires work to be done by a computer Job: collection of activities to do the work required Job steps: job divided into several steps- units of work that must be done sequentially Process or task: computation that may be done concurrently with other computations Address space: collection of programs and data that are accessed in a process forms an address space A user submit the job to the operating system Job divided into several job steps Once OS accepts the users job it create several process Refer diagram 1.2- depicts two sample address space- i/o process and CPU process Actual code for file system portion is same Code is shared so locks must be set to prevent race conditions Pure code/ reentrant code : code that does no modify itself OS maps the address space of processes into physical memory This is done by special H/W- paged system and S/W swapping system Multiprogramming: a system that have several processes in the states of execution at the same time State of execution: computation has started but not completed or terminated. i.e. Processor is not currently working on the process Privileged instructions: instructions that are not available to ordinary users- two states of execution in computer Problem state: user state, slave stateSupervisor state: executive state, master state ( can execute privileged instructions) Protection h/w: to control access to the parts of memory e.g. To prohibit from altering the OS program Interrupt h/w: to coordinate the operation going on simultaneously Interrupt: mechanism by which a processor is forced to take note of an eventAn OS Resource manager View framework for study of OS 3 views Process view : sequencing of the management of resources Hierarchical view Demonstrates interrelationship and primitive function Extended view These modals are used extension of the work of Dijkstra, Donovan, Hansen, Saltzer and Madnick Primary view OS- collection of programs (algorithm) designed to manage the system resource Memory, processor, devices, information (programs and Data) OS should keep track of Status of each resource Decide which process get it Allocate it Reclaim itViewing OS as resource manager Each manager should do1. Keep track of resources2. Enforce policy that determines who gets what, when and how much3. Allocate resource4. Reclaim the resourcesMemory management functions1. Keep track of resource ( memory). What parts are in use and by whom? What part are not in use (free)?.2. If multiprogramming, decide which process gets memory, when and how much3. Allocate the resource ( memory) when the process request it and the policy of 2 above allows itReclaim the resources when the process no longer needs it or has been terminated

Processor management functions1. Keep track of the resource ( processor and the status of processor)- done by traffic controller program2. Decide who will use the processor-done by job scheduler that chooses the job from submitted job. If multiprogramming, decides which process gets the processor, when and how much done by processor scheduler3. Allocate resources (processor) to process by setting the hardware register- called dispatcher4. Reclaim the resources( processor) when process finishes, terminates or exceeds allowed amount of usage.Device management functions1. Keep track of resources (devices, channels, control unit) I/O traffic controller2. Decide what is an efficient way to allocate resources ( device) . If it is to be shared , then decide who gets it, how much he gets it I/O scheduling3. Allocate resource and initiate I/O operation4. Reclaim the resources- I/O terminates automatically5. If cannot be shared then dedicated to a process (card readers)6. Shared devices- complication in allocation virtual devices 7. Example: punching on a punch card will be transferred to disk, at later time a routine will copy the information 8. Virtualizing card reader, card punch, is done by SPOOLing routines9. Virtual devices approach1. More flexibility in allocating dedicated devices2. Flexibility in job scheduling3. Better match of speed of device and speed of requestInformation management functions1. Keep track of the resource ( information), its location, use, status- called as file system2. Decide who gets it, enforce protection requirements and provide accessing routines3. Allocate the resources ( information) e.g. open a file4. Deallocate the resource e.g. Close a file An OS- process view point Life of process Represented by 3 transition states Run: process is assigned a processor and its program are being executed Wait: process is waiting for some event (e.g. an I/O operation to be completed) Ready : process is ready but more process than processor so must wait for its turn on processorSimple states of a process

Assumption : all process already existing in the system and will continue running forever. But complete and realistic life cycle will not be so. It needs 3 more state to be complete and realistic Submit Hold Complete Submit : a user submit a job and the system must respond to the users request Hold: the users job has been converted to internal machine readable form . But no resources is allocated. Job has been spooled onto the disk. Resource must be allocated to move it to the ready state. Complete : process has completed its computation and all its assigned resources may be reclaimed. Modal of process states

States and transitions in a process life cycle Circles-> states, clouds-> transitions Explained with an example- processing deck into card reader.1. User submit job by placing the deck into a card reader -> submit state2. Job consist of many deck of programs proceeded by job control cards.3. Job control cards pass information to OS about what resources will be needed.4. SPOOLing routine reads the job places it into disk -> Hold state5. The SPOOLing routine call information management to find storage space 6. Job scheduling routine scans all SPOOLed files and picks a job to be admitted7. In doing this , it will call- memory management for sufficient memory, device management for requested devices8. Then job scheduler call traffic controller to create associated process information and memory management to allocate main storage.9. The job is then loaded into memory and process ready to run. -> ready state10. When a processor is free- process scheduler scans list of ready to run process and choose, assign processor-> running state11. If running process request access to information, the information management will call device management to initiate the reading of file12. Device management initiate I/O operation and 13. then call process management to indicate that the process is waiting for the completion of I/O operation -> wait state14. When I/O is completed- h/w sends signal to traffic controller in process management, 15. then puts back process back in the ready state16. If process complete its computation, then completed state and17. allocated resources are freed. Operating system hierarchical and extended machine viewExtended machine view

These instructions provided by the operating system User request it by using a special supervisor call instructions. Similar to subroutine call but transfer the control to the operating system rather to the users subroutine Statement 2 and 3 are legal but do not correspond to the instruction of the bare machine Basic hardware instruction + additional instructions=> instruction set called as extended machine Refer diagram 1.7 Kernel of the operating system runs on the bare machine User programs runs on the extended machine

Hierarchical machine concept OS as a one big program- unmanageable Extended machine approach could be applied in 2 ways Key functions needed by many system module could be separated into an inner extended machine. Certain modules could be separated out and run on the extended machine is same way as user processes Inner and outer extended machine generalized into levels of extended machine Layers of process- generalized operating system process Kernel : all modules resides in extended machine as opposed to those that operate as a process layers No rule for number of layers and which modules goes into which layersHierarchical operating system structure All processes uses kernel and share all the system resources Some process are parent or controller- denoted by wavy lines- indicate separate process layer Given level is allowed to call upon services of lower level but not higher levels Example : if a interprocess message management call service of memory management , the memory module must be in lower level, but should not call upon process message management In the lower level- function used by all resource manager keeping track of resource allocation This requires a synchronization of resources allocation done by primitive operators P operator- resources seized V operator- resources released These operator upon a S/W called semaphore- a switch( binary semaphore) or a integer value ( counting semaphore) When resources is requested , test the P operator if off turns it on If already on then make a note that the requesting process has been placed in WAIT state for resources If V operator issued by another process releases the resources, then turn the semaphore OFF and awake the process waiting for that resources Examples of function in various levels Level 1: processor management lower level P, V synchronization primitive process scheduling Level 2: memory management Allocate memory Release memory Level 3: processor management upper level Create/ destroy process Send/receive messages between processes Stop process Start process Level 4 :Device management Keep track of all status of I/ O devices Schedule I/O Initiate I/O Level 5: information management create / destroy file Open/close file Read/ write file

Unit 1: Memory managementIntroduction Concerned with the management of primary memory or core memory Processors directly access for instructions and data Four functions Keep track of the status of each location of memory- either allocated or unallocated Determining allocation policy for memory- decide how much, when, where the memory shoul d be allocated- if shared then determine which process gets the memory Allocation technique selt the specific location and allocation information updated Dellocation technique and policy may explicitly release or unilaterally reclaim the memory .- must be updatedMemory management techniques1. Single contiguous memory management2. Partioned memory management3. Relocation partioned memory management4. Paged memory management5. Demand-paged memory management6. Segmented memory management7. Segmented and demand-paged memory management8. Other memory managementAnalysis of technique An overview of the approach and concepts employed A description of special hardware A description of software algorithm Advantages and disadvantages

Single contiguous allocation Simple memory mangement scheme No special hardware Associated with small stand alone computers IBM OS/360, Primary conrol Program No multi programming One-to-one correspondance exists between user, job, job step, process Memory is allocated to a job Conceptually divided into 3 contiguous regions Portion of memory permanently allocated to the OS All of the remainder of memory is availableto single process Job uses only a portion of memory other are unused region Example : if 256K bytes- allocated but 32K only used by a process then remining are unused It cannot be returned With respect to four functions1. Kep track of memory- it is allocated to one job2. Determining factor- the job gets all the memory3. Allocation all of its allocated to the job4. Deallocation when the job is done- memory is returned to free status Hardware support No special hardware Only a protection mechanism- to ensure that users program do not accidently tamper the OS Consists of bounds register and supervisor-mode of CPU If in the user mode- on each reference , hardware checks no acces is made If access made then an iterrupt must occur and control is transfered to the OS In supervisor mode can access the protected, and execute the privileged instructions Mode is changed when the control is transferred to the OSSoftware support A flowchart depicts single contiguous allocation Algorithm is called when job scheduler of processor management schedules a job to run Called only when no other job is using the memory Refer diagram 3.2AdvantagesUNIT ONE OVEr

UNIT 3: Job and processor schedulingProcessor management Concerned with management of physical processor Assignment if processors to processes Modules from unit 1 Job scheduler- creates processes (non-multi programmed) Processor scheduler-which process receives processor (in multiprogramming) Traffic controller- keep track of status of the process Job scheduler can be viewed as macro scheduler- choosing which job will run Process scheduler can be viewed as micro scheduler assigning processors to the process associated with scheduled jobs User sub divide the job into job steps- processed sub tasks System creates processes to do the computation of job steps Job scheduling concerned with management of jobs Processor scheduling concerned with management of processors No distinction between process and job scheduling in non multiprogramming system One to one correspondence- one job create one process and assigned a processor in non multiprogramming For multiprogramming systems- job scheduler creates processes for the jobs and assign the processor Process scheduler decides which process will be assigned a processor at what time and how long 1.State model Submit state Hold state Ready state Running state Blocked state / wait state Completed

5. Keep track of status of all processes6. Provide mechanism for changing process state7. Coordinate inter process synchronization and communication 1.1 job scheduler2 Keep track of status of all jobs. It must note which jobs are trying to get some service (hold state) and status of all jobs being serviced (ready, running, or blocked)3 Choose the policy to enter the system based on priority, resources requested4 Allocate the necessary resources for the scheduled job by use of memory , device and processor management 5 Deallocate the resources when the job is done Process scheduling Once job moved form hold to ready it creates one or more processes Functions 1. Keep track of the status of the process traffic controller (running, ready, blocked)2. Deciding which process gets a processor for how long processor scheduler 3. Allocation of processor to process traffic controller4. Deallocation of processor- when running process exceeds the current quantum or must wait for an I/O completion traffic controller Job and process synchronization Mechanism to prevent race condition Example: request a printer while it is printing P and V operator Semaphore Deadlock a situation- two processes , each waiting for resources that the other has and will not give up Structure of processor management Processor management operates in two levels- assigning a processors to jobs and assigning processors to processes On the job level processor management concerned which job will run first Not concerned with multiprogramming Assumes once job is scheduled it will run Job scheduling can run concurrently with other user program Creation, destroying, sending messages between process common to all address space Refer diagram 4.2 So in processor management upper level In multiprogramming environment the process scheduling and synchronization called by all modules So center of the kernel ( lower level )Job scheduling Focus on job scheduling policies, implications Dealing systems with and without multiprogramming Time sharing system- Compatible Time Sharing Systems (CTSS)- priority based Batch systems-OS/VS-2- arrival and priority basedJob scheduling Functions Policies Job scheduling in non-multiprogrammed environment Job scheduling using FIFO Job scheduling using shortest job first Job scheduling using future knowledge Measure of scheduling performance Job scheduling in multiprogrammed environment Job scheduling with multiprogramming but no i/o overlap Job scheduling with multiprogramming and i/o overlap Job scheduling with memory requirements and no i/o overlap Job scheduling with memory and tape constraints and no i/o overlap Job scheduling summaryFunctions Job scheduler overall supervisor Assigns system resources to certain jobs Keep track of jobs Invoke policies to decide which job gets resources Allocate resources Deallocate resources

Policies Job scheduler must choose from the HOLD state and put it into READY to RUN Choosing may be arbitrarily or shortest job Key concept Policy issue Typical considerationPolicies- Key concept More job wish to run Most resources are finite Many resources cannot be shared or easily reassigned to another processPolicies- policy issue Running as many jobs as possible- only short jobs Keeping the processor busy Fairness to all jobsPolicies- typical consideration To determine a job scheduling policy Availability of special resources With preference Without preference Cost-higher rates for faster service System commitments- processor time and memory-the more you want the longer you wait System balancing- mixing I/O intensive and CPU intensive Guaranteed service Completing the job by or at specific time job scheduler has selected a collection of jobs to run the process scheduler schedules dynamically Job and process scheduler may interact. Process scheduler postpone a process and sends it back to the macro-level job scheduling

Job scheduling in nonmutiprogrammed environment No multiprogramming Once a processor has been assigned a processor it does not release the processor until it is finished One CPU process is created for each job Policy to reduce average turnaround timeJob scheduling using FIFO

Job 1 arrived at 10 am and run for 2 hrs If we use First In First Out then, jobs will run as depicted Turnaround time = finish time- arrival time Average turnaround = sum of turnaround time divided by total number of jobs Job 1 arrived at 10 am and run for 2 hrs If we use First In First Out then, jobs will run as depicted Turnaround time = finish time- arrival time Average turnaround = sum of turnaround time divided by total number of jobs TABLE-plain sheet 31Job scheduling using shortest job first To reduce the average turnaround time Scheduling algorithm runs the HOLD job with shortest job run time first When job 1 arrives it is run While job 1 is running job 2 and 3 arrive Choose job 3 next shorter job than job 2 ( run time ) Job scheduling using shortest job firstTABLE_-plain sheet 33Job scheduling using future knowledge To improve the average turnaround time based on future knowledge If at 10 am we know that two short jobs will arrive then we wont run job 1 Reduce average turnaround time but wasted .25 hr of CPU time TABLE-sheet 35Problems in these job scheduling algorithm Future knowledge is rare Run times are usually estimated approximately Other resources must be considered such as memory requirements, I/O devices

Measure of scheduling performance Turnaround time- a measure of performance Another measure- Weighted turnaround time (W) W= T/ R T-> turnaround time R-> run timeJob scheduling in multiprogrammed environment Job scheduler selects the job to run Simple process scheduling algorithm- round robin Round robin- each job is assigned a processor for small quantum of time If n jobs are running simultaneously, they each get an equal share of run time CONTEXT SWITCH -39,40,41 These above tables represents FIFO scheduling with no multiprogramming Refer diagram 4.9a for graphic representation If multiprogramming is used then- CPU headway amount of CPU spent on a job equal to the half of the clock time elapsed Refer diagram 4.10a, 4.10b, 4.10c Job 1 arrived at 10 and run for 0.3hrs After job 1 had run for 0.2 hrs job 2 arrives so the processor was time sliced between the two jobs job 1 and job 2 Even though job 1 has only 0.1hrs of execution it will complete only at 10.4 Unit 3- process schedulingProcess scheduling Concerned with transitions 3 and 4 (hold-> ready, ready-> running) Job scheduler selects a collection of jobs to run ( submit-> hold) Process scheduler/ dispatcher or low-level scheduler assigns process to processor Process uses processor for only for 100ms Job and process scheduler may interact Process scheduler will roll out the process to undergo a job scheduling again time sharing systemSingle processor interlock

It is a single processor multiprogramming system Processes A and D need a tape drive Only one tape Process A request it first So tape assigned to process A When process D request it, it must be blocked until process A release it Processes are blocked but not the processor Functions process scheduler Keep track of the state of the process Decide which process gets a processor, when and for how long Allocate processors to processes Deallocate processors from processes Process control BlockProcess identification

Current state

Priority

Copy of active registers

Pointers to list of other processes in same state

Etc.

Keeping track of status of processes are done by traffic controller Using PCB- Process Control Block a database associated with each process Separate PCB for each process All PCBs with same state ( ready, blocked) are linked Resulting list is called ready list/ block list

Traffic controller is called whenever a status of a resource is changed If a process request a device and that device is already is in use, the process is linked to the block list When the device is released the traffic controller checks any process is waiting for that device If so then processes are placed in ready state.Policies