1. Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Chapter 5: Process Synchronization

2. 5.2 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Chapter 5: Process Synchronization Background The Critical-Section Problem Peterson’s Solution Synchronization Hardware Mutex Locks Semaphores Classic Problems of Synchronization Monitors Synchronization Examples Alternative Approaches

3. 5.3 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Objectives To present the concept of process synchronization. To introduce the critical-section problem, whose solutions can be used to ensure the consistency of shared data To present both software and hardware solutions of the critical-section problem To examine several classical process-synchronization problems To explore several tools that are used to solve process synchronization problems

4. 5.4 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Background Processes can execute concurrently May be interrupted at any time, partially completing execution Concurrent access to shared data may result in data inconsistency Maintaining data consistency requires mechanisms to ensure the orderly execution of cooperating processes In this chapter, what we discuss for “processes” almost always applies to “threads” as well Can change “processes” to “threads” Illustration of the problem: Suppose that we wanted to provide a solution to the consumer- producer problem that fills all the buffers. We can do so by having an integer counter that keeps track of the number of full buffers. Initially, counter is set to 0. It is incremented by the producer after it produces a new buffer and is decremented by the consumer after it consumes a buffer.

5. 5.5 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Producer while (true) { /* produce an item in next produced */ while (counter == BUFFER_SIZE) ; /* do nothing */ buffer[in] = next_produced; in = (in + 1) % BUFFER_SIZE; counter++; }

6. 5.6 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Consumer while (true) { while (counter == 0) ; /* do nothing */ next_consumed = buffer[out]; out = (out + 1) % BUFFER_SIZE; counter--; /* consume the item in next consumed */ }

7. 5.7 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Race Condition counter++ could be implemented as register1 = counter register1 = register1 + 1 counter = register1 counter-- could be implemented as register2 = counter register2 = register2 - 1 counter = register2 Consider this execution interleaving with “count = 5” initially: S0: producer execute register1 = counter {register1 = 5} S1: producer execute register1 = register1 + 1 {register1 = 6} S2: consumer execute register2 = counter {register2 = 5} S3: consumer execute register2 = register2 – 1 {register2 = 4} S4: producer execute counter = register1 {counter = 6} S5: consumer execute counter = register2 {counter = 4} Race condition – a situation when: several processes access the same data concurrently the outcome of the execution depends on the order of the accesses

8. 5.8 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Critical Section Problem Consider system of n processes {P 0, P 1, …,P n-1 } Each process has critical section segment of code Process may be changing common variables, updating table, writing file, etc When one process in critical section, no other is allowed to be in its critical section Critical section problem is to design protocol to solve this Each process must ask permission to enter critical section in entry section, may follow critical section with exit section, then remainder section

9. 5.9 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Diagram of Critical Section Problem P0P0 P1P1 P n-1 Shared Data Each process P i has its own critical section Accesses to shared data only in CS Lock – will consider later In shared data, accessible by all Often 1 lock for all n processes Guards access to all critical sections Ensures mutual exclusivity...

10. 5.10 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Critical Section General structure of process P i

11. 5.11 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Requirements to Critical-Section Problem Assume that each process executes at a nonzero speed No assumption concerning relative speed of the n processes 1. Mutual Exclusion Only one process can be in the critical section at a time –  otherwise what critical section? 2. Progress Intuition: No process is forced to wait for an available resource –  otherwise very wasteful. Formal: If no process is executing in its critical section and there exist some processes that wish to enter their critical section, then the selection of the processes that will enter the critical section next cannot be postponed indefinitely 3. Bounded Waiting Intuition: No process can wait forever for a resource –  otherwise an easy solution: no one gets in Formal: A bound must exist on the number of times that other processes are allowed to enter their critical sections after a process has made a request to enter its critical section and before that request is granted Some material taken from http://www.cs.cmu.edu/~gkesden/412-18/fall01/ln/lecture6.html

12. 5.12 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Algorithm for Process P i do { while (turn == j); //j is not my turn => wait critical section turn = j; //change turn to j remainder section } while (true); Assume 2-process case: processes Pi and Pj Mutual exclusion? Progress? Bounded waiting?

13. 5.13 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Types of solutions to CS problem Software-based Hardware-based

14. 5.14 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Software-based solution to CS Will consider 2-process case first: Processes P i and P j Will examine various incorrect and correct solutions next Solutions assume that the load and store machine-language instructions are atomic; that is, cannot be interrupted In general, this is not true for modern architectures  Peterson’s algorithm (we will consider shortly) does not work in general  Can work on some machines correctly, but can fail on others But good algorithmic description, allows to understand various issues The two processes share two variables: l int turn; Boolean flag[2] The variable turn indicates whose turn it is to enter the CS The flag array is used to indicate if a process is ready to enter the CS flag[i] = true implies that process P i is ready!

15. 5.15 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Algorithm 1 Case: 2 processes Pi and Pj only Shared Variables:  var turn: (0..1); initially turn = 0;  turn = i  P i can enter its critical section Process P i while(true) { while (turn != i) no-op; critical section turn = j; remainder section } l Satisfies mutual exclusion, but not progress  If it is turn i and if Pi is not in its CS, then if Pj wants to enter its CS, then Pj has to wait – but why wait?: Pi is not in its CS

16. 5.16 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Algorithm 2 Shared Variables  var flag: array (0..1) of boolean; initially: flag[0] = flag[1] = false;  flag[i] = true  Pi ready to enter its critical section Process P i while(true) { flag[i] = true; while (flag[j]) no-op; critical section flag[i]= false; remainder section } Can block indefinitely… Progress requirement not met.  Both can set flag[] to true and then both block busy-waiting

17. 5.17 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Algorithm 3 Shared Variables  var flag: array (0..1) of boolean; initially flag[0] = flag[1] = false;  flag[i] = true  Pi ready to enter its critical section Process P i while(true) { while (flag[j]) no-op; flag[i] = true; critical section flag[i] = false; remainder section } Does not satisfy mutual exclusion requirement …  (1) Pj flag[j]=false;  (2) Pi moves to after while, before flag[i] = true;  (3) Pj passes its while; sets flag[j] = true; enters its CS  (4) Pi sets flag[i] = true; enters its CS => both in their CSes

18. 5.18 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Algorithm 4: “Peterson’s Solution” l Combined Shared Variables of algorithms 1 and 2 l Process Pi while(true) { flag[i] = true; turn = j; while (flag[j] && turn == j) no-op; critical section flag[i] = false; remainder section } l YES!!! Meets all three requirements, solves 2-process case But … no guarantees that Peterson’s solution (or other software-based solutions) will work correctly on modern architectures.  Because of the way modern computer architectures perform basic machine-language instructions, such as load and store

19. 5.19 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Algorithm 4: Peterson’s solution for Process P i do { flag[i] = true; turn = j; while (flag[j] && turn == j); critical section flag[i] = false; remainder section } while (true); Provable that the three CS requirement are met: 1. Mutual exclusion is preserved P i enters CS only if: either flag[j] = false or turn = i 2. Progress requirement is satisfied 3. Bounded-waiting requirement is met

20. 5.20 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition (Lamport’s) “Bakery Algorithm” n Critical section for n processes n Key idea: Before entering its CS, the process receives a number (=token) l Holder of the smallest number enters the CS l Like in a bakery, get a number at the entrance, served according to it n If processes Pi and Pj receive the same number, l if i ≤ j, then Pi is served first; else Pj is served first l Will say Pi has higher priority than Pj n The numbering scheme always generates numbers in non-decreasing order of enumeration; l For example: 1,2,3,3,3,3,4,4,5,5

21. 5.21 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Bakery Algorithm (cont.) n Notation l Lexicographic order  (ticket#, process_id#)  (a,b) < (c,d), same as: if (a<c or ((a=c) and (b<d)) n Shared Data var choosing: array[0..n-1] of boolean; (initialized to false) number: array[0..n-1] of integer; (initialized to 0)

22. 5.22 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Bakery Algorithm (cont.) while(true) { choosing[i] = true; number[i] = max(number[0], number[1],…,number[n-1]) +1; choosing[i] = false; for (int j = 0; j <= n-1; j++) { // wait until process Pj receives its number (i.e, token) while (choosing[j]) ; // no-op // wait until all processes // (a) with smaller number, or // (b) with the same number but with higher priority // finish their work while (number[j] != 0 && (number[j],j) < (number[i],i)) ; // no-op } critical section number[i]= 0; remainder section } Code for process Pi

23. 5.23 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Critical-Section Handling in OS Two approaches depending on if kernel is preemptive or non- preemptive Preemptive – allows preemption of process when running in kernel mode Non-preemptive – runs until exits kernel mode, blocks, or voluntarily yields CPU  Essentially free of race conditions in kernel mode

24. 5.24 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Peterson’s Solution Good algorithmic description of solving the problem Two process solution Assume that the load and store machine-language instructions are atomic; that is, cannot be interrupted The two processes share two variables: l int turn; l Boolean flag[2] The variable turn indicates whose turn it is to enter the critical section The flag array is used to indicate if a process is ready to enter the critical section. flag[i] = true implies that process P i is ready!

25. 5.25 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Algorithm for Process P i do { flag[i] = true; turn = j; while (flag[j] && turn = = j); critical section flag[i] = false; remainder section } while (true);

26. 5.26 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Peterson’s Solution (Cont.) Provable that the three CS requirement are met: 1. Mutual exclusion is preserved P i enters CS only if: either flag[j] = false or turn = i 2. Progress requirement is satisfied 3. Bounded-waiting requirement is met

27. 5.27 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Synchronization Hardware Many systems provide hardware support for implementing the critical section code. All solutions below based on idea of locking Protecting critical regions via locks Uniprocessors – could disable interrupts Currently running code would execute without preemption  That is, without being interrupted Generally too inefficient on multiprocessor systems  OSes using this are not broadly scalable Modern machines provide special atomic hardware instructions Atomic = non-interruptible l test_and_set instruction  test memory word and set value l compare_and_swap instruction  swap contents of two memory words

28. 5.28 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Solution to Critical-section Problem Using Locks while (true) { acquire lock critical section release lock remainder section }

29. 5.29 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition test_and_set Instruction Definition: boolean test_and_set (boolean *target) { boolean rv = *target; *target = true; return rv; } 1. Executed atomically – cannot be interrupted in the middle 2. Returns the original value of passed parameter 3. Set the new value of passed parameter to true.

30. 5.30 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Solution using test_and_set() Shared Boolean variable lock initialized lock = false unlocked initially Solution: while (true) { while (test_and_set(&lock)) ; // wait until lock becomes equal false // here lock=true /* critical section */ lock = false; /* remainder section */ } Bounded waiting is not satisfied Other processes might keep acquiring the lock first

31. 5.31 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Bounded-waiting Mutual Exclusion with test_and_set The common data structures: boolean waiting[n]; // initialized to false boolean lock; // initialized to false while (true) { waiting[i] = true; // wants the lock key = true; while (waiting[i] && key) key = test_and_set(&lock); // trying to get the lock // got the lock here waiting[i] = false; /* critical section */ j = (i + 1) % n; while ((j != i) && !waiting[j]) // find the next in sequence Pj waiting for the lock j = (j + 1) % n; if (j == i) lock = false; // no one is waiting for the lock, simply release it else waiting[j] = false; // let Pj go into its CS, “lock” is kept locked /* remainder section */ } Satisfies all the three CS requirements, less fair than Lamport’s Bakery algorithm

32. 5.32 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition compare_and_swap Instruction Definition: int compare_and_swap(int *value, int expected, int new_value) { int temp = *value; if (*value == expected) *value = new_value; return temp; } 1. Executed atomically 2. Returns the original value of passed parameter “value” 3. Set the variable value to new_value but only if value ==expected The swap takes place only under this condition

33. 5.33 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Solution using compare_and_swap Shared integer lock initialized to lock=0; // unlocked Solution: while (true) { while (compare_and_swap(&lock, 0, 1) != 0) ; // wait for lock to be 0 // here lock = 1 /* critical section */ lock = 0; /* remainder section */ } Bounded waiting is not satisfied

34. 5.34 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Mutex Locks Previous solutions are complicated and generally inaccessible to application programmers OS designers build software tools to solve critical section problem Simplest is mutex lock Protect a critical section by: first acquire() a lock then release() the lock Boolean variable indicating if lock is available or not Calls to acquire() and release() must be atomic Usually implemented via hardware atomic instructions But this solution requires busy waiting This lock therefore called a spinlock

35. 5.35 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition acquire() and release() n acquire() { while (!available) ; /* busy wait */ available = false; } n release() { available = true; } n while (true) { acquire lock critical section release lock remainder section }

36. 5.36 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Semaphore Synchronization tool that provides more sophisticated ways (than Mutex locks) for process to synchronize their activities. Semaphore S – integer variable Can only be accessed via two indivisible (atomic) operations wait() and signal() Originally called P() and V()-- important to remember, used often  P() (wait) – from Dutch proberen: “to test”  V() (signal) – from Dutch verhogen: “to increment” Definition of the wait() operation wait(S) { while (S <= 0) ; // Notice, busy waits, spending CPU cycles, not (always) good S--; } Definition of the signal() operation signal(S) { S++; }

37. 5.37 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Semaphore Usage Counting semaphore – integer value can range over an unrestricted domain Can control access to resource that has a finite number of instances Initialized to the number of resources available Binary semaphore – integer value can range only between 0 and 1 Same as a mutex lock Can solve various synchronization problems Consider P 1 and P 2 that require S 1 to happen before S 2 Create a semaphore “ synch ” initialized to 0 P1: S 1 ; signal(synch); P2: wait(synch) ; S 2 ;

38. 5.38 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Implementing a counting semaphore via binary Can implement a counting semaphore S via binary semaphores n Data Structures int S = n; // semaphore count mutex S_guard = 1; // initial value 1 (FREE) mutex delay = 0; // for delaying other processes Solution adopted from Jim Mooney of WVU.

39. 5.39 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Implementing S wait() { wait(S_guard); S --; if (S <= -1) { // check if someone's already waiting on this sem signal(S_guard); // relesase S_guard wait(delay); // join the waiting queue } else signal(S_guard); } signal() { wait(S_guard); S = ++; if (S < 0) signal(delay); signal(S_guard); }

40. 5.40 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Semaphore Implementation Operations wait() and signal() must be executed atomically Thus, the implementation becomes the critical section problem wait and signal code are placed in the critical section Hence, can now have busy waiting in critical section implementation If critical section rarely occupied  Little busy waiting But, applications may spend lots of time in critical sections  Hence, the busy-waiting approach is not a good solution

41. 5.41 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition No Busy Waiting Implementation of Semaphore A waiting queue is associated with each semaphore It stores the processes waiting on the semaphore n typedef struct{ int value; struct process *list; // waiting queue } semaphore; Two operations: block – place the process invoking the operation on the appropriate waiting queue wakeup – remove one of processes in the waiting queue and place it in the ready queue

42. 5.42 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Implementation with no Busy waiting (Cont.) wait(semaphore *S) { S->value--; if (S->value list; block(); //suspends self, sleeps, avoids CPU cycles } signal(semaphore *S) { S->value++; if (S->value list; wakeup(P); } n S->value can be negative l |S->value| is then the number of processes waiting on the semaphore l It is never negative in classical definition

43. 5.43 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Deadlock and Starvation Synchronization problems Deadlock – two or more processes are waiting indefinitely for an event that can be caused by only one of the waiting processes Let S and Q be two semaphores initialized to 1 P 0 P 1 wait(S); wait(Q); wait(Q); wait(S);...... signal(S); signal(Q); signal(Q); signal(S); Starvation (= indefinite blocking) Related to deadlocks (but different) Occurs when a process waits indefinitely in the semaphore queue For example, assume a LIFO semaphore queue  Processed pushed first into it might not get a chance to execute Priority Inversion A situation when a higher-priority process needs to wait for a lower-priority process that holds a lock

44. 5.44 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Classical Problems of Synchronization Classical problems used to test newly-proposed synchronization schemes Bounded-Buffer Problem Readers and Writers Problem Dining-Philosophers Problem

45. 5.45 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition The Bounded-Buffer Problem Producer and Consumer processes Buffer pool consists of n buffers each can hold one item Shared data structures: int n; semaphore mutex = 1; // Guards the access to the buffer pool semaphore empty = n; // Counts the number of empty buffers semaphore full = 0; // Counts the number of full buffers

46. 5.46 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Bounded Buffer Problem (Cont.) The structure of the producer process while (true) {... /* produce an item in next_produced */... wait(empty); // suspend self if the buffer is full wait(mutex); // get access to the buffer... /* add next produced to the buffer */... signal(mutex); // release the lock to the buffer signal(full); // }

47. 5.47 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Bounded Buffer Problem (Cont.) The structure of the consumer process while (true) { wait(full); //=P() suspend self if the buffer is empty wait(mutex); // get access to the buffer... /* remove an item from buffer to next_consumed */... signal(mutex); // V() release the lock to the buffer signal(empty);... /* consume the item in next consumed */... }

48. 5.48 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Discussion A symmetry Producer does: P(empty), V(full) Consumer does: P(full), V(empty) Producer producing full buffers for consumer Consumer producing empty buffers for the producer Is order of P’s important? Yes! Can cause deadlock Is order of V’s important? No, except that it might affect scheduling efficiency

49. 5.49 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition The Readers-Writers Problem A data set is shared among a number of concurrent processes: Readers – only read the data set  they do not perform any updates Writers – can both read and write Problem Allow multiple readers to read (shared data) at the same time Only one single writer can access shared data at the same time  Readers cannot read during this time  Other writers cannot write during this time

50. 5.50 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Readers-Writers Problem (Cont.) Shared Data: Data set semaphore rw_mutex = 1; // mutual exclusion for writers semaphore mutex = 1; // guards read_count int read_count = 0; // #processes currently reading the object The structure of a writer process while (true) { wait(rw_mutex);... /* writing is performed */... signal(rw_mutex); }

51. 5.51 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Readers-Writers Problem (Cont.) The structure of a reader process while (true) { wait(mutex); read_count ++; if (read_count == 1) wait(rw_mutex); signal(mutex);... /* reading is performed */... wait(mutex); read_count--; if (read_count == 0) signal(rw_mutex); signal(mutex); } Writers can starve in this solution How would you design a solution where writers don’t starve?

52. 5.52 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Readers-Writers Problem Variations First variation – no reader kept waiting unless writer has permission to use shared object Second variation – once writer is ready, it performs the write ASAP Both may have starvation leading to even more variations Problem is solved on some systems by kernel providing reader-writer locks

53. 5.53 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition The Dining-Philosophers Problem Philosophers spend their lives alternating: thinking and eating Occasionally try to pick up 2 chopsticks (left and right) to eat from bowl One chopstick at a time Need both chopsticks to eat, then release both when done Problem: not enough chopsticks for all  N philosophes and N chopsticks (not 2N) The case of 5 philosophers (and 5 chopsticks only) Shared data  Bowl of rice (data set)  Semaphore chopstick [5] initialized to 1 (free)

54. 5.54 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Dining-Philosophers Problem Algorithm The structure of Philosopher i: while (true) { wait (chopstick[i]); // wait to get the left stick wait (chopstick[(i + 1) % 5]); // get the right // eat signal (chopstick[i]); signal (chopstick[(i + 1) % 5]); // think }; What is the problem with this algorithm? A deadlock is possible!

55. 5.55 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Dining-Philosophers Problem Algorithm (Cont.) Deadlock handling possibilities: Allow at most 4 philosophers to be sitting simultaneously at the table. Allow a philosopher to pick up the chopsticks only if both are available  Picking must be done in a critical section Use an asymmetric solution:  an odd-numbered philosopher – picks up first the left and then right chopstick  even-numbered philosopher – picks up first the right and then left chopstick

56. 5.56 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Problems with Semaphores Incorrect use of semaphore operations: Can occur relatively easily and be difficult to detect  Timing errors – could occur only under certain circumstances and won’t occur otherwise – Run a program and it crashes, run it again and it doesn’t  Can be more complex to debug Wrong order: signal (mutex) …. wait (mutex) Wrong calls: wait (mutex) … wait (mutex) Omitting of wait (mutex) or signal (mutex) (or both) Deadlock and starvation are possible How to deal with this? Researchers developed high-level language constructs Monitor type – one such construct

57. 5.57 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Monitors monitor monitor-name { // shared variable declarations function P1 (…) { … } … function Pn (…) {……} initialization_code (…) { … } } Recall that: an Abstract data type (ADT) -- encapsulates data internal variables only accessible by code within the procedure A monitor type – is an ADT that provides a convenient and effective mechanism for process synchronization For several concurrent processes Encapsulation  Local variables accessed only via local functions  Local functions access only local vars and params Only one process at a time may be active within the monitor A lock guards shared data Hence, the programmer does not need to code this constraint

58. 5.58 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Schematic view of a Monitor

59. 5.59 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Condition Variables Monitors not powerful enough to model some synchr. schemes Hence, additional synchr. mechanisms are added: Provided by the condition construct condition x, y ; Two operations are allowed on a condition variable x : x.wait()  A process that invokes the operation is suspended (sleeps) – until x.signal()is called  Lets other processes enter the monitor – releases the lock to shared data, atomically with sleep x.signal()  Resumes one of processes (if any) that invoked x.wait()  If no x.wait( ) was called, then x.signal() has no effect x

60. 5.60 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Monitor with Condition Variables

61. 5.61 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Condition Variables Choices Issues with monitors: assume Process P invokes x.signal(), and Process Q is suspended in x.wait() Who proceeds next?  Both Q and P cannot execute in parallel  Because they are within a monitor Options include Signal and wait – P waits until Q either leaves the monitor or it waits for another condition Signal and continue – Q waits until P either leaves the monitor or it waits for another condition Both have pros and cons – language implementer can decide

62. 5.62 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Each philosopher i invokes the operations pickup() and putdown() in an infinite loop: DiningPhilosophers.pickup(i) ; // EAT DiningPhilosophers.putdown(i) ; No deadlocks, but starvation is possible Monitor Solution to Dining Philosophers

63. 5.63 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Solution to Dining Philosophers (Cont.) monitor DiningPhilosophers { //-- shared data, local to the monitor -- enum {THINKING; HUNGRY, EATING) state [5]; condition self[5]; // for suspending self when chopsticks are not available //-- initialization code -- initialization_code() { for (int i = 0; i < 5; i++) state[i] = THINKING; }

64. 5.64 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Solution to Dining Philosophers (cont.) //-- local functions -- void pickup (int i) { state[i] = HUNGRY; test(i); // try to start eating. if (state[i] != EATING) // if cannot start eating -- wait self[i].wait; // suspend self -- until a direct neighbor bumps you } // test(i) will set state of i to EATING only if: // (a) i is hungry and (b) its left & right neighbors are not eating void test (int i) { if ((state[i] == HUNGRY) && // access to shared data is protected automatically (state[(i + 1) % 5] != EATING) && (state[(i + 4) % 5] != EATING) ) { state[i] = EATING; self[i].signal(); //release i if it is waiting }

65. 5.65 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Solution to Dining Philosophers (cont.) void putdown (int i) { state[i] = THINKING; // test the left/right neighbors (who perhaps are waiting for your chopsticks) test((i + 1) % 5); test((i + 4) % 5); } Recall what test(i) does: // test(i) will set state of i to EATING only if: // (a) i is hungry and (b) the left & right neighbors are not eating void test (int i) { if ((state[i] == HUNGRY) && (state[(i + 1) % 5] != EATING) && (state[(i + 4) % 5] != EATING) ) { state[i] = EATING; self[i].signal(); //release i if it is waiting }

66. 5.66 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Monitor Implementation Using Semaphores Variables semaphore mutex; // (initially = 1) semaphore next; // (initially = 0) int next_count = 0; Each procedure F will be replaced by wait(mutex); … body of F; … if (next_count > 0) signal(next) else signal(mutex); Mutual exclusion within a monitor is ensured

67. 5.67 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Monitor Implementation – Condition Variables For each condition variable x, we have : semaphore x_sem; // (initially = 0) int x_count = 0; The operation x.wait can be implemented as : x_count++; if (next_count > 0) signal(next); else signal(mutex); wait(x_sem); x_count--;

68. 5.68 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Monitor Implementation (Cont.) The operation x.signal can be implemented as: if (x_count > 0) { next_count++; signal(x_sem); wait(next); next_count--; }

69. 5.69 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Resuming Processes within a Monitor If several processes queued on condition x, and x.signal() executed, which should be resumed? FCFS frequently not adequate conditional-wait construct of the form x.wait(c) Where c is priority number Process with lowest number (highest priority) is scheduled next

70. 5.70 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Allocate a single resource among competing processes using priority numbers that specify the maximum time a process plans to use the resource R.acquire(t) ;... access the resurce;... R.release; Where R is an instance of type ResourceAllocator Single Resource allocation

71. 5.71 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition A Monitor to Allocate Single Resource monitor ResourceAllocator { boolean busy; condition x; void acquire(int time) { if (busy) x.wait(time); busy = TRUE; } void release() { busy = FALSE; x.signal(); } initialization code() { busy = FALSE; }

72. 5.72 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Synchronization Examples Solaris Windows Linux Pthreads

73. 5.73 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Solaris Synchronization Implements a variety of locks to support multitasking, multithreading (including real-time threads), and multiprocessing Uses adaptive mutexes for efficiency when protecting data from short code segments Starts as a standard semaphore spin-lock If lock held, and by a thread running on another CPU, spins If lock held by non-run-state thread, block and sleep waiting for signal of lock being released Uses condition variables Uses readers-writers locks when longer sections of code need access to data Uses turnstiles to order the list of threads waiting to acquire either an adaptive mutex or reader-writer lock Turnstiles are per-lock-holding-thread, not per-object Priority-inheritance per-turnstile gives the running thread the highest of the priorities of the threads in its turnstile

74. 5.74 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Windows Synchronization Uses interrupt masks to protect access to global resources on uniprocessor systems Uses spinlocks on multiprocessor systems Spinlocking-thread will never be preempted Also provides dispatcher objects user-land which may act mutexes, semaphores, events, and timers Events  An event acts much like a condition variable Timers notify one or more thread when time expired Dispatcher objects either signaled-state (object available) or non-signaled state (thread will block)

75. 5.75 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Linux Synchronization Linux: Prior to kernel Version 2.6, disables interrupts to implement short critical sections Version 2.6 and later, fully preemptive Linux provides: Semaphores atomic integers spinlocks reader-writer versions of both On single-cpu system, spinlocks replaced by enabling and disabling kernel preemption

76. 5.76 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Pthreads Synchronization Pthreads API is OS-independent It provides: mutex locks condition variable Non-portable extensions include: read-write locks spinlocks

77. 5.77 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Alternative Approaches Transactional Memory OpenMP Functional Programming Languages

78. 5.78 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition A memory transaction is a sequence of read-write operations to memory that are performed atomically. void update() { /* read/write memory */ } Transactional Memory

79. 5.79 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition OpenMP is a set of compiler directives and API that support parallel progamming. void update(int value) { #pragma omp critical { count += value } The code contained within the #pragma omp critical directive is treated as a critical section and performed atomically. OpenMP

80. 5.80 Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition Functional programming languages offer a different paradigm than procedural languages in that they do not maintain state. Variables are treated as immutable and cannot change state once they have been assigned a value. There is increasing interest in functional languages such as Erlang and Scala for their approach in handling data races. Functional Programming Languages

81. Silberschatz, Galvin and Gagne ©2013 Operating System Concepts – 9 th Edition End of Chapter 5