Unik3 (ePu Scheduling ) ___ full oxpsotnark on? : = Agenda.) objective : a 4.J CPU Scheduding and Critevt. for CPU gche duiio , 21 Objectives of Cpu scheduling. = al “Preemptive VS Non Pie -ey mptive scheduling, 4} Operation of mulsitevel schedulira . S.J Short. term 2 long tev Schedulin. 6J State of procere T1 process state! diag ram. 3] Procers convo blook and T's campolfenis. “l9.J_ Difference biw Schedwters. 9} Precers addvere space. to] Process Mertification tnfovmation, 11] Thead 2 TS aduantoges , implementation YoCes , types. Z 12.) User Sevel thveads yg Kernel Jeval thread. 13.] “Thiead us Process 19.3 Thread Gncellation. 15.) FCFS scheduling and tts advantage, 16.\ Shortest Job Fi¥et tes) scheduling algo, odu. 8 disadu.. = 12) Priority Scheduling Algorithm , adu+ dad. $0.\_ found “Robin Schedulivig AlgovttAm om 19.|_ Mult Deve feedback queue” scheduling. do. Mulffprocessoy scheduling .. u ete oe 1-7 Varfant of ound Robin Khedullng. a 22. Deadlock. | Resource allocatfow a graphs Banker's ah, : Oo247] Dead.dock V8 Sievoadon= —— oy Like y Shave, Gubgcribe- . 4.J cpu Scheduling - > Management of CPU Yex,ourvceg. 2 Sechéduding mechanism fe the part of he _ process ‘mansgey. 3 Hordies Removal & Selection of POCO. > Tk is the base of mutt’- praqtammin ch System. Jd we Tncveases cpu usthigadkion. Scheduting Critevia- 4.CPL_UtitiaccHon e Average functton af me. * Range rom ay. t© 1007, 2. Thro aghput : At Yefers 40 the amount of work completed in a. umit of time. «the highey the no. the move -wotk is done by ‘the “aystom, 3. woot Time & f 4. Turnelound Times Submissionef <> Comphetion of Process Proce Turnaround Thme.=*Chaman Tam Waise Tr RoF ron Book, — _____Fadte? fn _ a Okjectivet [needy of CPU scheduding Se | 4 Eptfcfeney 3 | 3 Meimut Throughput |__ a _Min?mige #5 Refpowe Tine . a | 3 Minimige overhead ~ Maximige Rerource ase 3 Avoid ftnde finite postponement > _Enterice prfavittes. y La Pre- -emptive Scheduting yb Nen- Pie-emplie Scheduling - a J £ fre-emp tive Non pre-emptive 7 Processov can tun [> One Lxeaatioy, aathey proces white | staves then after QuiEnd PTOCEIG ex Corr Compledron othey kon. PYocere can dun, = 2 efy wWgation #8 [> os ehu_udttagton. — more.> bess wetting x ~ Move walting # | Respome erie. Yerpome tne” > fdexfhie 3 _nigid. | 2.89 SRTF, Prieiity | > &Y. Fors, sat Round - Robin eke prtovity eke. tobe used fn computer conive 2 whe non- pve -empitve scheduling fe nity Operation of Muttt levet scheduding : I ~_Heve p¥ocess queues ave usthey ‘divided irks Peveral A queues. : [> Each queue hoe ite own scheduling eulgovfthm . [> V for example : diffevent-o queus can.’ be_used foy foreground ¥ background proces $a. Aiqhest prior te Oe Zz System Procpse, — — Trevactive Processe3 —5 — Trtevactive edittra pfoces — I —5 batch process —> il — Sssudent proces — Jowat priority - Kubseribe tos Sant Fraginess Babu:Bate} Page: — as] 35 — Shott Term Schedueling Objective & Tmplem entation — a > __Tnevease Sustern Pevformmance wrth the chosen set Lot ext teva. > Tk is the change of Yeady state to Aunning state of the pvoceks. 2 Te 's aio _eathed CPU scheduler, (alakulelabebast Long tevin che dull YQ _obteckive Z: fmpleme. — ond md me 7 Main obfectfue fe 46 provide balanced — execution. 3 Covtyatly the d eqvee af multe proqya — mming, Zz 3 They detexming w hich pYoqiams ave — admitted to the system for predentin 5 = Tob Scheduler JU q Select pYoceva from queue & Soad sthem Frio mah Memory for eXeation, : § 1 Skates of Procae. | Procesg Trantdton Dfagyomn: J = 1. New 1 Creation a Running 2 creator of Mstructong 8. waiting Woaiting for other procws ereauden 4 Ready + assigned ~6 processor 3. Teva noted < ercutton tnished.Sch : e [ile oF waa sheduley_dpadch = xlo at event Completion wart Gntert Switching A 4. Tt © a mechanism +o stove and Yaire ‘the state ox contert af a CPU in process ‘biock., So that a plore? ereation can be yesumed foo the Same point or a JateY Line. 2. Tt aflowe mult? tasking , 3. Tt oecuys when prams cxeusion & chang ed fron one to they. Process tnuciwed im contert Switchira - od te Save the cordert of the pyoceisoy, Mmetuding proqyam coureY_and athey veqistes. J oT U 2, Updoxe the proces conties . block of the proces selected. This included changing the state of this process 40 dunnine J 3. Move the pyocas contiod block of Ais il procms te the Oppo priake quente ,e pincers waa ptaced fn the Swipended state by Hselt ov oS. > Parent procars YeQuet ° > Swap ph > Other Xeason S.J _Procers Ortial block > 4 Tt Ba datactiuokwe used © steve the information about the ptocers, > “his info. is wed by CPU at tantime- the faltowteng ave the information that 7 cont ained by cintial block. 2s Namiig the process + State “Of the pracers * Resources allocated Jo the procos. * Memory oocated to the ptaces 2 Scheditling fnfermahon * Lilo J ofp’ devices associated with proces, Process Id SS Process State Progam Counted ister Mnformetion NQ_thfovmaton Memory tated infovmadion Aca oa thfo, | Stak tae eS 6+shot “Term _Mid Ter GY tong Term. schedulers schedulers Scheduler de [Tab Shedtudor | CPU gchodulery | P-1oce3 of Swapp- ing scheduley Q. | Leis Speed Fast Speed Medium speed. - 3. | Gctvats degree | Les conttad over} Reduced the of mulit- pirag- degree of mutt | deqice of ams Yamming . ping varnrnvng. | ptoqiaynrning - q. fibsent ov'minimal Minimal in-time| Tire shay ira, in-tme- shaving. sharing system. |sustem we md so System. [term mheduler, 5. (oak) RAM @) 2) | P. can be veintyodued. AJ PYecexs addvess space.” > Process addvexe space means a space thot is allocated in mematy fox a process” 2 _Add¥es pace fe a space fh computer memory, Cg a fuety proces Fos an addvas space. 2 Phusicat addrere spi Addvexs space i Cx SUTURE RIATETS TpaAcS eatedfy. UM. } LAL 8) Procas Adentiticatton Information . 4 as uu Tt ’s 4 of process. 3's fafo, te tecovded fight after the process [Lis _eveated o¥ boxn, i see ess) ff _ — - ‘ 3 4. 2 > Ws fd fs very umrque_t = _ ait we fattdwing numerte tdevftioye, ~— a bs UTTGRE provers CdewRttey—— B User fdenkihrey, a 9.) Thread — Ss x Tis a flaw of eveeutin though the process code. a *& Tt has (U's own proqiom counter, a system , Yeqistey, stack. 7 = UJ « Fath thread belongs to exactly one process. v oe _No thieod cant exist cutside the praca. * “They ave wed 46 achieve mutica , | Tete Yieia Ftey | Cor = i ace [sine Register) aE I Sead s) a | a : Rie a | GB Sigle Aead = | Adi Jo} — = | > Thieod minimize contert swttohing Une. 7 Use of Thieads provides conoeuvreney witht _ G process, S > Efficient Gommunicadéon, x 2 Economie friendly . i > we con achieve nuiltiterking, J— ta ene > Threads can he »_be tmpsernendod aif Aevet. 3 _ User devel thread % Known or Josey actuodiu. : | Aryplementedion. of “Thread; - oe. _@ Thread implemented on fn user space- 7 i n the kernel. & it simp. 2 Thveads Tmpllemontation tn tevned 5 : Th thie procedure thveads package enticy ~) In -his no thiead -tebse 1s proort > Th a thiead wants to do any operation Tipsy do kernel call. = Here the? mformation % the sare @ [utth the user Jevel thveads. 7 G Hy bid ?mptementasion + Combination of beths = | { to.J | wer Level Threads tleynel Level Thread) Qtower Jo create and Ae fous to evoate and mana, | manage: a. Mim plemented Pad | Kevnek Revert thyeads neat Sip tox at ate spect 4o the wer Love! - mg Sustem. ' 8. an tun on ox. Se asin specific. 4. crit, proutded at the|_tleve Keviveh a tee Jeuvel called Support ( user Jovel thread. |4 11 “Types of “Thread - 1. User dever thread - > Alt of the work of thYead managemert- done by appureation. v : a Created, destYoyed ay we can Say that impoemented ustrgq Fhiead Librovy. 2 i user Space T read (4 ig Kernek Space GD ; Advantage ¢ a > "thread switching does not require KeYnel mode privilage. J 5 > _e@ &r kan any Operating System. YD Scheduling cay be” applicakend spec Kio, > fast (© create 2 manage. a Disadvantage: 2 Ine Eypreat O8 , mest system cals axe _blacking. 3 “Mult threaded appltoation cannot take eovanege of mult! pro cesoing. J 2. Kernel Level thread: Hee thread manag ere rst ig done dy Kein eh (Self, > Kernel threads O%€ Supported divectly dy 08. : SS U> Sehedutig by-the kernel! fg done on o ahread basis’ = Imptementation fs done_in Kernek space. Advantages § ~ Tk con stmuttaneo gchedule mbtiple threads from the same neat on raultiple pYocesse. > Tf one thread fn a proceds 7s blocked, the kernel eon schedute another thvead of the same process. > Kernel voutines thenelve can mutt- thread. ie _||_Digad vantages - > _ Slowey to create and manage. | 3 Traméer of contval from on’ thread & another within Same piocBs Yequeyeys & mode switch to the Kerner. 12.5 Thieod Process 4s [Tread ts called Light | Process fe caste d-heavy | eight process. weight preeas. weight p ight proce 2. | Thad switching doe Trot need 40 caWan | operating gygtem ard Process awitehing Needs frtevfece woth eause an imtevvupt 40 operating system. TU the kevnek. 3. | AgLthYeads ean shave Tn muuttiple pyocas Sayne_ Set of open fale) fmphementotion ead child procers. process executes the Game code."> blacked and wal rg ; is_b§ocked no other second thvead_fh the same | sotvey pracas can task coutd yun. execute Unt the feet process unhlo. eked. &._One +hyvead ean Yead , In multiple process witte oy ever) Completely | each pYocers aperortes wipe out another thyeady” frd ependevesla of Stack. the o-they. : ~~ 4. while one sevuer thread | Tf one serve PYocer, Lobel cheat ‘| votes Thread do nat Yequive any cman to be execute. Thread Cancellation t Task of terménesting a thread before tt has Completed. > Fey example, If multipte thveads ave con- cwrentty searching through o database and one trend returns the Tue. the temaining @ +thyeads might be canceled. > often a eb page f Loaded using several thveods and each image & toaded’' in a separate thread. > when a wer presses the elep button on the browsey , att threads dJoading +he page ave canceled. v a o v cf P— Siieuthe oF Sail Frances Simplest scheduling alg ovithmm . 3 Te tobiged with“ FTF queue. Lt Enola : Jike- % when @ process enters the Yeady gueve, 7S Peb is Littted onto the tail af Me Guene ® As goon ah cpu fe free ,it fs allocated t© the fist procels of queue. & Aifter_completion process 78 temoved from dhe_queue. _fidua ntage? Divadvanitagg : Q fetler for Lo79 procera. > Very emadl process should 7 Simple mefted wate for hie tur + No slarvation 3 Throughput t¢ not emphias 'ged- 14.) Shortest Job Frist (SIDE > Tt velUiG on the Length of proces. > Gmabl procerser ave run Prt Aduavetog a: ot Disadvardagel? 7 High Theapa: a > Blopsed Ye tecording 3 Ottves putovieg Wehort > Starvation be g 2 “ibs. possible fot Lord p roceise Foltaw won Instaqvayn (a Gast engineey_ babe. Jfr 15) Priory scheduling aulgotitm= mon. pie-empeue algorithm.» x mast eammonty wed aloori thm. a each proces (Y aisi@ned “a_priovitq » process with high” picority yun Hist. og wi; biLiuyL I Advaxtagel- |S UPriotty ean be selected . based on ‘ema vy | vequirement’, time Tequuitement usoy preference. | 9 Assignes_fmpartance to pYocesse? Disadvanfagg, 7- : | 3 Te Wher priovtey procera) we up alot af CPU thle , Lower priority proce) 9 may state. [4 Assigning preovitey fe W sevsitive work do, - [46.3 Round - Rabin schedudiva 2 U > Stmttoy to FCFS algorithm. | 2 T's fe wed fov Rme shaving gustem. a Hele the teady queue fe cfvdutdy j queue. > pv scheduler start tiawersing The. queue and ausign 4 quantum Yme to eoah process . > Then there processes ereeltes on thely time_hasis.Adu 23 ¢ : > effective on_tme- shaving syetert o¥ —— eae piocessing system. 2 faty-Heatment fol all the piocesces. > Overhead an processor 1¢ Low. > God yesporse time. Oserdvanieg e- > Cakefeully chooging of quartam value. > Procossnd overhedd ts theve in-handting look inter sup. > Throughput fs Sow Ff time quantum 8 00 Small 1-3 ePu Scheduling > akyedy covered: Tt Mutitlevel feedback queue schedui7g - +. overcomes the problem of multilevel queue scheduling alaovithone 3 __ separate Yroclses with diffevert cpu burct_tme. =) thone pYocas user too Much timed move tt to ~owoy p¥iovity queme. > TE allows procas 6 rave b/wW-quamg, > Tt implements twe oy mote scheduling quevgs,. J £8.J Mult? Processor Scheduling - | 2 _one a multtprocerey Scheduling fe Zp- 2 Scheduling done wind mnuublipte procesas.TTeda/! Numericals- ; ‘ 19.J what would be effect of pusknd too pointers %6 the same process tn the ready ~ ques 9 7 20.1 Peadsock eet : shy deadtock axige - Mutual exctuston Hodd and watt. No- pye -ernptten LILilie Ctrewley wast- Mutual excturton Hold and weft appvoacher of preventad Con : > > 3 Ne pre-emption > Crveuley weit -Dead tock avotdance D__ Tt allowg the three Necessary condHors but makes judicious chotee to assure deadlock oint ig nevereached- ; > Tt allows ma ve conayvenoy then preventatton does. 2 “hwo Approaches ave used to avoide dead - Lock — *& Do mot start a process (ft (L's demand A0eads to deadlock. * Do Not grant an nee rental ete yequart 0 K process ff thie allocator mignt Lead to deadlock. : ue System can be fs following gtatoy~ U J Safe. fax uate | 24) Resource atoratton giaiph: 2 Guiaphitgat Yeprerentation ww ONIG TRoUVCA SF procusg. Resource 3 [ay Ry Procas— ©) =da) deadlock detecton’, = (ETS a process of deteminiva eldhe deadlock fn _a system: J deadjock defection appioaches - > Single frstance af each vesource type. 2 Sereval_imtance of a Yaource Jtype, 23] Bonker's angasithem for avoiding deadsock : oO ~ Tt ts an yoourde allocation and deadlock avotdance algorithm. > followiig date stiuckive ave uaed bo ?mptement “the Banker's algorithm. 4. Available! 1P aviay of available varourcg. J & Max! 2d arvay of maximum devnand of earch process. 3. PUocmtion: 20 aviay that define the mo, of Yesourees atlocated each proca 4 Need? 2D array that inateatey the aemaining yvesomicad meed af each proces. F >: Fase} ——] | toner go. wer bon alge lo aie deod.lock: 2 —— | 4. Safely ataortthn | Qu. Ressiee vquat algor"thm, \ “lechniques to Recover fiom deadlock! i O_Pocar Riminaton bs Abort all deadlocked procescal Ey Rott one pvocess ak atmo, GRaovice pie-empuon, 3 Selecting a vietin + Refback 7 iM the procerg |_ Like , Shave, Gubecribe,