DOCSLIB.ORG

Synchronization: Semaphores, Monitors, and Condition Variables

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
Synchronization: Semaphores, Monitors, and Condition Variables Today’s Topics u Mutex Isn’t Enough COS 318: Operating Systems u Semaphores u Condition Variables Synchronization: Semaphores, u Monitors Monitors and Condition Variables u Barriers Jaswinder Pal Singh and a Fabulous Course Staff Computer Science Department Princeton University (http://www.cs.princeton.edu/courses/cos318/) 2 Revisit Mutex Producer-Consumer (Bounded Buffer) Problem u Mutex can solve the critical section problem Acquire( lock ); Producer: Consumer: while (1) { Critical section while (1) { produce an item remove an item from buffer Release( lock ); Insert item in buffer count--; u Use Mutex primitives to access shared data structures count++; consume an item E.g. shared “count” variable } } Acquire( lock ); count = 4 count++; Release( lock ); N = 12 u Are mutex primitives adequate to solve all problems? u Can we solve this problem with Mutex primitives? 3 4 1 Use Mutex, Block and Unblock Use Mutex, Block and Unblock Producer: Consumer: Producer: Consumer: while (1) { while (1) { while (1) { while (1) { produce an item if (!count) produce an item if (!count) if (count == N) Block(); if (count == N) {context switch} Block(); remove an item from buffer Block(); Block(); Insert item in buffer Acquire(lock); Insert item in buffer remove an item from buffer Acquire(lock); count--; Acquire(lock); Acquire(lock); count++; Release(lock); count++; count--; Release(lock); if (count == N-1) Release(lock); Release(lock); if (count == 1) count = 4 Unblock(Producer); if (count == 1) countcountcount = == 12 01 if (count == N-1) Unblock(Consumer); consume an item Unblock(Consumer); Unblock(Producer); } N = 12 } } N = 12 consume an item } u Does this work? u Race condition! u Ultimately, both block and never wake up u Lost the unblock; any way to “remember” them? 5 6 Limitations of Locks Semaphores (Dijkstra, 1965) u A semaphore is a synchronization variable that contains an u Provide mutual exclusion: only one process/thread can be in the integer value critical section at a time l Cannot access the integer value directly (only via semaphore operations) u Do not provide ordering or sequencing (aka event synchronization) l Initialized to some integer value l Who gets to be in critical section first? l Supports two atomic operations other than initialization l How does thread A wait for thread B (or C, D, E) to do X before A does Y? • down() (or wait() or P()) u Need additional synchronization mechanisms • up (or signal() or V()) l Semaphores u If positive value, think of value as keeping track of how many ‘resources’ or “un-activated unblocks” are available l Condition Variables u If negative, tracks how many threads are waiting for a resource l Monitors or unblock l (Higher level constructs composed from these) 7 8 2 Semaphores (Dijkstra, 1965) Bounded Buffer with Semaphores “ ” u P (or Down or Wait or Proberen (to try)) definition Producer: Consumer: l Atomic operation while (1) { while (1) { l Block version: Decrement value, and if result less than zero then block produce an item P(fullCount); l Spin version: Wait for semaphore to become positive and then decrement P(emptyCount); P(s){ P(s){ P(mutex); if (--s < 0) while (s <= 0) P(mutex); take an item from buffer block(s); ; put item in buffer V(mutex); s--; V(mutex); } } V(emptyCount); V(fullCount); consume item u V (or Up or Signal or “Verhogen” (increment)) definition } } l Atomic operation l Block version: increment, and if non-positive (which means at least one thread is blocked waiting on the sempahore) then unblock a thread u Initialization: emptyCount = N; fullCount = 0 l Spin version: Increment semaphore u Are P(mutex)and V(mutex) necessary? V(s){ V(s){ if (++s <=0) s++; unblock(s); } } 9 Uses of Semaphores in this Example Example: Interrupt Handler u Event sequencing l Don’t consume if buffer empty, wait for something to be Init(s,0); added l Don’t add if buffer full, wait for something to be removed Device thread Interrupted Thread u Mutual exclusion while (1) { P(s); … l Avoid race conditions on shared variables Acquire(m); Interrupt handler ... ... Interrupt deal with interrupt V(s); … ... ... Release(m); } 11 12 3 Bounded Buffer with Semaphores (again) Does Order Matter? producer() { consumer() { producer() { consumer() { while (1) { while (1) { while (1) { while (1) { produce an item P(fullCount); produce an item P(fullCount); P(emptyCount); P(mutex); P(mutex); P(emptyCount); P(mutex); P(mutex); take an item from buffer take an item from buffer put the item in buffer V(mutex); put the item in buffer V(mutex); V(mutex); V(mutex); V(emptyCount); V(emptyCount); V(fullCount); consume the item V(fullCount); consume the item } } } } } } } } Another Example: Are Locks Enough? Condition Variables u Make tryRemove wait until something is on the queue? u A lock provides mutual exclusion to the shared data l Can’t just sleep while holding the lock u Rules for using a lock: l Key idea: make it possible to go to sleep inside critical section, l Always acquire before accessing shared data structure by atomically releasing lock at same time we go to sleep. l Always release after finishing with shared data l Lock is initially free. u Condition variable: enables a queue of threads waiting for something inside a critical section. u Simple example: a synchronized queue l Wait() --- Release lock, go to sleep, re-acquire when woken • release lock and going to sleep is atomic bool tryRemove() bool tryInsert() { … { lock.Acquire(); l Signal() --- Wake up a waiter, if any lock.Acquire(); // lock before use if something on queue // can we wait? l Broadcast() --- Wake up all waiters … put item on queue; // ok to access remove it; lock.Release(); // unlock after done lock->Release(); return success; return success; } } 4 Synchronized Queue Condition variable design pattern u Rule: must hold lock when doing condition variable methodThatWaits() { methodThatSignals() { operations lock.acquire(); lock.acquire(); RemoveFromQueue() // Read/write shared state // Read/write shared state AddToQueue() { { lock.acquire(); lock.acquire(); while (!testSharedState()) { // If testSharedState is now true while nothing on queue cv.wait(&lock); cv.signal(&lock); put item on queue; condition.wait(&lock); } condition.signal(); // release lock; got to // sleep; reacquire lock // Read/write shared state // Read/write shared state lock.release(); // when woken } lock.release(); remove item from queue; lock.release(); } lock.release(); } return item; } Condition variables Structured synchronization u ALWAYS hold lock when calling wait, signal, broadcast u Identify objects or data structures that can be accessed l Condition variable is synchronization FOR shared state by multiple threads concurrently l Remember: ALWAYS hold lock when accessing shared state u Add locks to object/module l Obtain lock on start to every method/procedure u Unlike semaphore, condition variable is memory-less l Release lock when finished l If signal when no one is waiting, no op u If need to wait l If signal after a wait is posted, a waiter wakes up l while(needToWait()) { condition.Wait(lock); } u If do something that should wake someone up u Wait atomically releases lock l Signal or Broadcast u Always leave shared state variables in a consistent state l When lock is released, or when waiting 5 Monitors Monitors Embedded in Languages u High-level data abstraction that unifies handling of: u Monitor definition: l Shared data, operations on it, synchronization and scheduling l a lock and zero or more condition variables for managing • All operations on data structure have single (implicit) lock concurrent access to shared data • An operation can relinquish control and wait on condition // only one process at time can update instance of Q u Monitors make things easier: class Q { int head, tail; // shared data l “locks” for mutual exclusion void enqueue(v) { locked access to Q instance } l “condition variables” for scheduling constraints int dequeue() { locked access to Q instance } } l Java from Sun; Mesa/Cedar from Xerox PARC u Monitors are easy and safe l Compiler can check, lock is implicit (cannot be forgotten) Monitor: Hide Mutual Exclusion Condition Variables in A Monitor u Brinch-Hansen (73), Hoare (74) u Wait(condition) Queues associated u Procedures are mutually exclusive l Block on “condition” with x, y u Signal(condition) conditions x Shared l Wakeup a blocked process y data on “condition” Queue of waiting processes Shared trying to enter the monitor data ... procedures ... Entry queue procedures 6 Producer-Consumer with Monitors Hoare’s Signal Implementation (MOS p137) monitor ProdCons procedure Producer monitor ProdCons u Run the signaled thread condition full, empty; condition full, empty; begin immediately and suspend while true do procedure Enter; begin procedure Enter; the current one (Hoare) begin produce an item begin if (buffer is full) ProdCons.Enter(); if (buffer is full) wait(full); wait(full); end; u What if the current thread end; put item into buffer; put item into buffer; if (only one item) has more things to do? if (only one item) signal(empty); procedure Consumer signal(empty); end; end; begin if (only one item) while true do signal(empty); procedure Remove; begin procedure Remove; something else begin ProdCons.Remove(); begin end; consume an item; if (buffer is empty) if (buffer is empty) wait(empty); end; wait(empty); end; remove an item; remove an item; if (buffer was full) if (buffer was full) signal(full); signal(full); end; end; 31 Hansen’s Signal Implementation (MOS p 137) Mesa Signal Implementation u Signal must be the last monitor ProdCons u Continues its execution statement of a monitor condition full, empty; if (only one item) signal(empty); procedure procedure Enter; something else begin u end; Exit the monitor if (buffer is full) wait(full); put item into buffer; l B. W. Lampson and D. D. Redell, “Experience with Processes and u Any issue with this if (only one item) Monitors in Mesa,” Communiction of the ACM, 23(2):105-117.
Load more

Recommended publications
  • Palmetto Tatters Guild Glossary of Tatting Terms (Not All Inclusive!)
    Palmetto Tatters Guild Glossary of Tatting Terms (not all inclusive!) Tat Days Written * or or To denote repeating lines of a pattern. Example: Rep directions or # or § from * 7 times & ( ) Eg. R: 5 + 5 – 5 – 5 (last P of prev R). Modern B Bead (Also see other bead notation below) notation BTS Bare Thread Space B or +B Put bead on picot before joining BBB B or BBB|B Three beads on knotting thread, and one bead on core thread Ch Chain ^ Construction picot or very small picot CTM Continuous Thread Method D Dimple: i.e. 1st Half Double Stitch X4, 2nd Half Double Stitch X4 dds or { } daisy double stitch DNRW DO NOT Reverse Work DS Double Stitch DP Down Picot: 2 of 1st HS, P, 2 of 2nd HS DPB Down Picot with a Bead: 2 of 1st HS, B, 2 of 2nd HS HMSR Half Moon Split Ring: Fold the second half of the Split Ring inward before closing so that the second side and first side arch in the same direction HR Half Ring: A ring which is only partially closed, so it looks like half of a ring 1st HS First Half of Double Stitch 2nd HS Second Half of Double Stitch JK Josephine Knot (Ring made of only the 1st Half of Double Stitch OR only the 2nd Half of Double Stitch) LJ or Sh+ or SLJ Lock Join or Shuttle join or Shuttle Lock Join LPPCh Last Picot of Previous Chain LPPR Last Picot of Previous Ring LS Lock Stitch: First Half of DS is NOT flipped, Second Half is flipped, as usual LCh Make a series of Lock Stitches to form a Lock Chain MP or FP Mock Picot or False Picot MR Maltese Ring Pearl Ch Chain made with pearl tatting technique (picots on both sides of the chain, made using three threads) P or - Picot Page 1 of 4 PLJ or ‘PULLED LOOP’ join or ‘PULLED LOCK’ join since it is actually a lock join made after placing thread under a finished ring and pulling this thread through a picot.
    [Show full text]
  • Logix 5000 Controllers Security, 1756-PM016O-EN-P
    Logix 5000 Controllers Security 1756 ControlLogix, 1756 GuardLogix, 1769 CompactLogix, 1769 Compact GuardLogix, 1789 SoftLogix, 5069 CompactLogix, 5069 Compact GuardLogix, Studio 5000 Logix Emulate Programming Manual Original Instructions Logix 5000 Controllers Security Important User Information Read this document and the documents listed in the additional resources section about installation, configuration, and operation of this equipment before you install, configure, operate, or maintain this product. Users are required to familiarize themselves with installation and wiring instructions in addition to requirements of all applicable codes, laws, and standards. Activities including installation, adjustments, putting into service, use, assembly, disassembly, and maintenance are required to be carried out by suitably trained personnel in accordance with applicable code of practice. If this equipment is used in a manner not specified by the manufacturer, the protection provided by the equipment may be impaired. In no event will Rockwell Automation, Inc. be responsible or liable for indirect or consequential damages resulting from the use or application of this equipment. The examples and diagrams in this manual are included solely for illustrative purposes. Because of the many variables and requirements associated with any particular installation, Rockwell Automation, Inc. cannot assume responsibility or liability for actual use based on the examples and diagrams. No patent liability is assumed by Rockwell Automation, Inc. with respect to use of information, circuits, equipment, or software described in this manual. Reproduction of the contents of this manual, in whole or in part, without written permission of Rockwell Automation, Inc., is prohibited. Throughout this manual, when necessary, we use notes to make you aware of safety considerations.
    [Show full text]
  • Stclang: State Thread Composition As a Foundation for Monadic Dataflow Parallelism Sebastian Ertel∗ Justus Adam Norman A
    STCLang: State Thread Composition as a Foundation for Monadic Dataflow Parallelism Sebastian Ertel∗ Justus Adam Norman A. Rink Dresden Research Lab Chair for Compiler Construction Chair for Compiler Construction Huawei Technologies Technische Universität Dresden Technische Universität Dresden Dresden, Germany Dresden, Germany Dresden, Germany [email protected] [email protected] [email protected] Andrés Goens Jeronimo Castrillon Chair for Compiler Construction Chair for Compiler Construction Technische Universität Dresden Technische Universität Dresden Dresden, Germany Dresden, Germany [email protected] [email protected] Abstract using monad-par and LVars to expose parallelism explicitly Dataflow execution models are used to build highly scalable and reach the same level of performance, showing that our parallel systems. A programming model that targets parallel programming model successfully extracts parallelism that dataflow execution must answer the following question: How is present in an algorithm. Further evaluation shows that can parallelism between two dependent nodes in a dataflow smap is expressive enough to implement parallel reductions graph be exploited? This is difficult when the dataflow lan- and our programming model resolves short-comings of the guage or programming model is implemented by a monad, stream-based programming model for current state-of-the- as is common in the functional community, since express- art big data processing systems. ing dependence between nodes by a monadic bind suggests CCS Concepts • Software and its engineering → Func- sequential execution. Even in monadic constructs that explic- tional languages. itly separate state from computation, problems arise due to the need to reason about opaquely defined state.
    [Show full text]
  • Multithreading Design Patterns and Thread-Safe Data Structures
    Lecture 12: Multithreading Design Patterns and Thread-Safe Data Structures Principles of Computer Systems Autumn 2019 Stanford University Computer Science Department Lecturer: Chris Gregg Philip Levis PDF of this presentation 1 Review from Last Week We now have three distinct ways to coordinate between threads: mutex: mutual exclusion (lock), used to enforce critical sections and atomicity condition_variable: way for threads to coordinate and signal when a variable has changed (integrates a lock for the variable) semaphore: a generalization of a lock, where there can be n threads operating in parallel (a lock is a semaphore with n=1) 2 Mutual Exclusion (mutex) A mutex is a simple lock that is shared between threads, used to protect critical regions of code or shared data structures. mutex m; mutex.lock() mutex.unlock() A mutex is often called a lock: the terms are mostly interchangeable When a thread attempts to lock a mutex: Currently unlocked: the thread takes the lock, and continues executing Currently locked: the thread blocks until the lock is released by the current lock- holder, at which point it attempts to take the lock again (and could compete with other waiting threads). Only the current lock-holder is allowed to unlock a mutex Deadlock can occur when threads form a circular wait on mutexes (e.g. dining philosophers) Places we've seen an operating system use mutexes for us: All file system operation (what if two programs try to write at the same time? create the same file?) Process table (what if two programs call fork() at the same time?) 3 lock_guard<mutex> The lock_guard<mutex> is very simple: it obtains the lock in its constructor, and releases the lock in its destructor.
    [Show full text]
  • APPLYING MODEL-VIEW-CONTROLLER (MVC) in DESIGN and DEVELOPMENT of INFORMATION SYSTEMS an Example of Smart Assistive Script Breakdown in an E-Business Application
    APPLYING MODEL-VIEW-CONTROLLER (MVC) IN DESIGN AND DEVELOPMENT OF INFORMATION SYSTEMS An Example of Smart Assistive Script Breakdown in an e-Business Application Andreas Holzinger, Karl Heinz Struggl Institute of Information Systems and Computer Media (IICM), TU Graz, Graz, Austria Matjaž Debevc Faculty of Electrical Engineering and Computer Science, University of Maribor, Maribor, Slovenia Keywords: Information Systems, Software Design Patterns, Model-view-controller (MVC), Script Breakdown, Film Production. Abstract: Information systems are supporting professionals in all areas of e-Business. In this paper we concentrate on our experiences in the design and development of information systems for the use in film production processes. Professionals working in this area are neither computer experts, nor interested in spending much time for information systems. Consequently, to provide a useful, useable and enjoyable application the system must be extremely suited to the requirements and demands of those professionals. One of the most important tasks at the beginning of a film production is to break down the movie script into its elements and aspects, and create a solid estimate of production costs based on the resulting breakdown data. Several film production software applications provide interfaces to support this task. However, most attempts suffer from numerous usability deficiencies. As a result, many film producers still use script printouts and textmarkers to highlight script elements, and transfer the data manually into their film management software. This paper presents a novel approach for unobtrusive and efficient script breakdown using a new way of breaking down text into its relevant elements. We demonstrate how the implementation of this interface benefits from employing the Model-View-Controller (MVC) as underlying software design paradigm in terms of both software development confidence and user satisfaction.
    [Show full text]
  • Concurrent Objects and Linearizability Concurrent Computaton
    Concurrent Objects and Concurrent Computaton Linearizability memory Nir Shavit Subing for N. Lynch object Fall 2003 object © 2003 Herlihy and Shavit 2 Objectivism FIFO Queue: Enqueue Method • What is a concurrent object? q.enq( ) – How do we describe one? – How do we implement one? – How do we tell if we’re right? © 2003 Herlihy and Shavit 3 © 2003 Herlihy and Shavit 4 FIFO Queue: Dequeue Method Sequential Objects q.deq()/ • Each object has a state – Usually given by a set of fields – Queue example: sequence of items • Each object has a set of methods – Only way to manipulate state – Queue example: enq and deq methods © 2003 Herlihy and Shavit 5 © 2003 Herlihy and Shavit 6 1 Pre and PostConditions for Sequential Specifications Dequeue • If (precondition) – the object is in such-and-such a state • Precondition: – before you call the method, – Queue is non-empty • Then (postcondition) • Postcondition: – the method will return a particular value – Returns first item in queue – or throw a particular exception. • Postcondition: • and (postcondition, con’t) – Removes first item in queue – the object will be in some other state • You got a problem with that? – when the method returns, © 2003 Herlihy and Shavit 7 © 2003 Herlihy and Shavit 8 Pre and PostConditions for Why Sequential Specifications Dequeue Totally Rock • Precondition: • Documentation size linear in number –Queue is empty of methods • Postcondition: – Each method described in isolation – Throws Empty exception • Interactions among methods captured • Postcondition: by side-effects
    [Show full text]
  • Computer Science 322 Operating Systems Topic Notes: Process
    Computer Science 322 Operating Systems Mount Holyoke College Spring 2008 Topic Notes: Process Synchronization Cooperating Processes An Independent process is not affected by other running processes. Cooperating processes may affect each other, hopefully in some controlled way. Why cooperating processes? • information sharing • computational speedup • modularity or convenience It’s hard to find a computer system where processes do not cooperate. Consider the commands you type at the Unix command line. Your shell process and the process that executes your com- mand must cooperate. If you use a pipe to hook up two commands, you have even more process cooperation (See the shell lab later this semester). For the processes to cooperate, they must have a way to communicate with each other. Two com- mon methods: • shared variables – some segment of memory accessible to both processes • message passing – a process sends an explicit message that is received by another For now, we will consider shared-memory communication. We saw that threads, for example, share their global context, so that is one way to get two processes (threads) to share a variable. Producer-Consumer Problem The classic example for studying cooperating processes is the Producer-Consumer problem. Producer Consumer Buffer CS322 OperatingSystems Spring2008 One or more produces processes is “producing” data. This data is stored in a buffer to be “con- sumed” by one or more consumer processes. The buffer may be: • unbounded – We assume that the producer can continue producing items and storing them in the buffer at all times. However, the consumer must wait for an item to be inserted into the buffer before it can take one out for consumption.
    [Show full text]
  • Objective C Runtime Reference
    Objective C Runtime Reference Drawn-out Britt neighbour: he unscrambling his grosses sombrely and professedly. Corollary and spellbinding Web never nickelised ungodlily when Lon dehumidify his blowhard. Zonular and unfavourable Iago infatuate so incontrollably that Jordy guesstimate his misinstruction. Proper fixup to subclassing or if necessary, objective c runtime reference Security and objects were native object is referred objects stored in objective c, along from this means we have already. Use brake, or perform certificate pinning in there attempt to deter MITM attacks. An object which has a reference to a class It's the isa is a and that's it This is fine every hierarchy in Objective-C needs to mount Now what's. Use direct access control the man page. This function allows us to voluntary a reference on every self object. The exception handling code uses a header file implementing the generic parts of the Itanium EH ABI. If the method is almost in the cache, thanks to Medium Members. All reference in a function must not control of data with references which met. Understanding the Objective-C Runtime Logo Table Of Contents. Garbage collection was declared deprecated in OS X Mountain Lion in exercise of anxious and removed from as Objective-C runtime library in macOS Sierra. Objective-C Runtime Reference. It may not access to be used to keep objects are really calling conventions and aggregate operations. Thank has for putting so in effort than your posts. This will cut down on the alien of Objective C runtime information. Given a daily Objective-C compiler and runtime it should be relate to dent a.
    [Show full text]
  • Lock Free Data Structures Using STM in Haskell
    Lock Free Data Structures using STM in Haskell Anthony Discolo1, Tim Harris2, Simon Marlow2, Simon Peyton Jones2, Satnam Singh1 1 Microsoft, One Microsoft Way, Redmond, WA 98052, USA {adiscolo, satnams}@microsoft.com http://www.research.microsoft.com/~satnams 2 Microsoft Research, 7 JJ Thomson Avenue, Cambridge, CB3 0FB, United Kingdon {tharris, simonmar, simonpj}@microsoft.com Abstract. This paper explores the feasibility of re-expressing concurrent algo- rithms with explicit locks in terms of lock free code written using Haskell’s im- plementation of software transactional memory. Experimental results are pre- sented which show that for multi-processor systems the simpler lock free im- plementations offer superior performance when compared to their correspond- ing lock based implementations. 1 Introduction This paper explores the feasibility of re-expressing lock based data structures and their associated operations in a functional language using a lock free methodology based on Haskell’s implementation of composable software transactional memory (STM) [1]. Previous research has suggested that transactional memory may offer a simpler abstraction for concurrent programming that avoids deadlocks [4][5][6][10]. Although there is much recent research activity in the area of software transactional memories much of the work has focused on implementation. This paper explores software engineering aspects of using STM for a realistic concurrent data structure. Furthermore, we consider the runtime costs of using STM compared with a more lock-based design. To explore the software engineering aspects, we took an existing well-designed concurrent library and re-expressed part of it in Haskell, in two ways: first by using explicit locks, and second using STM.
    [Show full text]
  • Process Scheduling
    PROCESS SCHEDULING ANIRUDH JAYAKUMAR LAST TIME • Build a customized Linux Kernel from source • System call implementation • Interrupts and Interrupt Handlers TODAY’S SESSION • Process Management • Process Scheduling PROCESSES • “ a program in execution” • An active program with related resources (instructions and data) • Short lived ( “pwd” executed from terminal) or long-lived (SSH service running as a background process) • A.K.A tasks – the kernel’s point of view • Fundamental abstraction in Unix THREADS • Objects of activity within the process • One or more threads within a process • Asynchronous execution • Each thread includes a unique PC, process stack, and set of processor registers • Kernel schedules individual threads, not processes • tasks are Linux threads (a.k.a kernel threads) TASK REPRESENTATION • The kernel maintains info about each process in a process descriptor, of type task_struct • See include/linux/sched.h • Each task descriptor contains info such as run-state of process, address space, list of open files, process priority etc • The kernel stores the list of processes in a circular doubly linked list called the task list. TASK LIST • struct list_head tasks; • init the "mother of all processes” – statically allocated • extern struct task_struct init_task; • for_each_process() - iterates over the entire task list • next_task() - returns the next task in the list PROCESS STATE • TASK_RUNNING: running or on a run-queue waiting to run • TASK_INTERRUPTIBLE: sleeping, waiting for some event to happen; awakes prematurely if it receives a signal • TASK_UNINTERRUPTIBLE: identical to TASK_INTERRUPTIBLE except it ignores signals • TASK_ZOMBIE: The task has terminated, but its parent has not yet issued a wait4(). The task's process descriptor must remain in case the parent wants to access it.
    [Show full text]
  • 4. Process Synchronization
    Lecture Notes for CS347: Operating Systems Mythili Vutukuru, Department of Computer Science and Engineering, IIT Bombay 4. Process Synchronization 4.1 Race conditions and Locks • Multiprogramming and concurrency bring in the problem of race conditions, where multiple processes executing concurrently on shared data may leave the shared data in an undesirable, inconsistent state due to concurrent execution. Note that race conditions can happen even on a single processor system, if processes are context switched out by the scheduler, or are inter- rupted otherwise, while updating shared data structures. • Consider a simple example of two threads of a process incrementing a shared variable. Now, if the increments happen in parallel, it is possible that the threads will overwrite each other’s result, and the counter will not be incremented twice as expected. That is, a line of code that increments a variable is not atomic, and can be executed concurrently by different threads. • Pieces of code that must be accessed in a mutually exclusive atomic manner by the contending threads are referred to as critical sections. Critical sections must be protected with locks to guarantee the property of mutual exclusion. The code to update a shared counter is a simple example of a critical section. Code that adds a new node to a linked list is another example. A critical section performs a certain operation on a shared data structure, that may temporar- ily leave the data structure in an inconsistent state in the middle of the operation. Therefore, in order to maintain consistency and preserve the invariants of shared data structures, critical sections must always execute in a mutually exclusive fashion.
    [Show full text]
  • The Critical Section Problem Problem Description
    The Critical Section Problem Problem Description Informally, a critical section is a code segment that accesses shared variables and has to be executed as an atomic action. The critical section problem refers to the problem of how to ensure that at most one process is executing its critical section at a given time. Important: Critical sections in different threads are not necessarily the same code segment! Concurrent Software Systems 2 1 Problem Description Formally, the following requirements should be satisfied: Mutual exclusion: When a thread is executing in its critical section, no other threads can be executing in their critical sections. Progress: If no thread is executing in its critical section, and if there are some threads that wish to enter their critical sections, then one of these threads will get into the critical section. Bounded waiting: After a thread makes a request to enter its critical section, there is a bound on the number of times that other threads are allowed to enter their critical sections, before the request is granted. Concurrent Software Systems 3 Problem Description In discussion of the critical section problem, we often assume that each thread is executing the following code. It is also assumed that (1) after a thread enters a critical section, it will eventually exit the critical section; (2) a thread may terminate in the non-critical section. while (true) { entry section critical section exit section non-critical section } Concurrent Software Systems 4 2 Solution 1 In this solution, lock is a global variable initialized to false. A thread sets lock to true to indicate that it is entering the critical section.
    [Show full text]