1. Operating Systems Unit 3: – Concurrent execution mutual exclusion – Concurrent programming semaphore monitor Operating Systems

2. COP 5994 - Operating Systems2 Concurrent execution System has more than one thread/process either independent or in cooperation: –mostly independent –occasionally need to communicate or synchronize

3. COP 5994 - Operating Systems3 Communication/synchronizati on Threads may access resource simultaneously –resource can be put in inconsistent state Context switch can occur at anytime, such as before a thread finishes modifying value Solution: mutual exclusion –idea: serialized access –Only one thread allowed access at one time –Others must wait until resource is available Must be managed such that wait time not unreasonable

4. COP 5994 - Operating Systems4 Critical section a Section of code –where shared resource is modified –only one thread can be in its critical section –avoid infinite loops and blocking inside Provides mutual exclusion Rest of code is safe to run concurrently

5. COP 5994 - Operating Systems5 Mutual Exclusion properties Mutual Exclusion If process is executing in its critical section, then no other processes can be executing in their critical sections Progress 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 Assume that each process executes at a nonzero speed No assumption concerning relative speed of the n processes Bounded Waiting A limit 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

6. COP 5994 - Operating Systems6 Mutual exclusion algorithm General structure of process P i (vs. P j ) do { entry protocol critical section exit protocol remainder section } while (true);

7. COP 5994 - Operating Systems7 Dekker’s Algorithm Idea: processes share common variables to synchronize their actions: int turn; –denotes which process is favored to enter its critical section boolean flag[2]; –denotes whether process is ready to enter its critical section

8. COP 5994 - Operating Systems8 Dekker’s Algorithm: Process P i do { flag[i] = true; while (flag[j]) { if (turn == j) { flag[i] = false; while ( turn == j ) ; flag[i] = true; } critical section turn = j; flag[i] = false; remainder section } while(true);

9. COP 5994 - Operating Systems9 Dekker’s Algorithm Guarantees mutual exclusion for 2 processes Uses notion of favored thread to resolve conflict over which thread should execute first Each thread temporarily yields to other thread Favored status alternates between threads

10. COP 5994 - Operating Systems10 Peterson’s Algorithm Less complicated than Dekker’s Algorithm –Still uses busy waiting, favored threads –Requires fewer steps to perform mutual exclusion primitives –Easier to demonstrate its correctness

11. COP 5994 - Operating Systems11 Peterson’s Algorithm: Process P i do { flag[i] = true; turn = j; while (flag[j] and turn == j) ; critical section flag[i] = false; remainder section } while (true);

12. COP 5994 - Operating Systems12 Lamport’s Bakery Algorithm N-Thread Mutual Exclusion –Creates a queue of waiting threads by distributing numbered “tickets” –Each thread executes when its ticket’s number is the lowest of all threads –Unlike Dekker’s and Peterson’s Algorithms, the Bakery Algorithm works in multiprocessor systems and for n threads –Relatively simple to understand due to its real-world analog

13. COP 5994 - Operating Systems13 Bakery Algorithm Critical section for n processes: Before entering its critical section process receives a number Holder of the smallest number enters critical section If processes P i and P j receive the same number, process with lower process id is served first The numbering scheme always generates numbers in increasing order of enumeration; i.e., 1,2,3,3,3,3,4,5...

14. COP 5994 - Operating Systems14 Bakery Algorithm Shared data boolean choosing[n]; int number[n]; Data structures are initialized to false and 0 respectively

15. COP 5994 - Operating Systems15 Bakery Algorithm do { choosing[i] = true; number[i] = max(number[0], number[1], …, number [n – 1])+1; choosing[i] = false; for (j = 0; j < n; j++) { while (choosing[j]) ; while ((number[j] != 0) && (number[j],j) < number[i],i)) ; } critical section number[i] = 0; remainder section } while (true);

16. COP 5994 - Operating Systems16 Mutual Exclusion via Hardware Disabling interrupts Special instructions –test and set –swap

17. COP 5994 - Operating Systems17 Disabling Interrupts mask interrupts while in critical section –current thread cannot be preempted could result in deadlock –e.g.: thread does I/O block in critical section works only on uniprocessor systems used rarely

18. COP 5994 - Operating Systems18 Test-and-Set Instruction boolean testAndSet(var) returns the value of var and sets var to true atomic instruction simplifies mutual exclusion algorithm algorithm must use instruction correctly

19. COP 5994 - Operating Systems19 Mutual exclusion algorithm do { while (testAndSet(lock)) ; critical section lock = false; remainder section } while (true);

20. COP 5994 - Operating Systems20 Swap Instruction swap(a, b) exchanges the values of a and b atomically Similar in functionality to test-and-set –more common

21. COP 5994 - Operating Systems21 Mutual exclusion algorithm do { myTurn = true; do { swap(lock, myTurn); } while (myTurn) ; critical section lock = false; remainder section } while (true);

22. COP 5994 - Operating Systems22 Semaphore Software construct to enforce mutual exclusion Contains a protected variable: –accessed via wait and signal commands P V binary semaphore: 0 or one counting semaphore

23. COP 5994 - Operating Systems23 Binary Semaphores only one thread allowed in critical section –Wait operation: P If no other thread in critical section: –thread enters critical section –Decrement protected variable (to 0 in this case) Otherwise place in waiting queue –Signal operation: V Indicate that thread has left its critical section Increment protected variable (from 0 to 1) A waiting thread (if there is one) may now enter

24. COP 5994 - Operating Systems24 Mutual exclusion algorithm do { P(lock) ; critical section V(lock); remainder section } while (true);

25. COP 5994 - Operating Systems25 Synchronization with Semaphores Semaphores can be used to notify other threads that events have occurred –Producer-consumer relationship: Producer enters its critical section to produce value Consumer is blocked until producer finishes Consumer enters its critical section to read value Producer cannot update value until it is consumed –Semaphores offer a clear, easy-to- implement solution to this problem

26. COP 5994 - Operating Systems26 Counting Semaphores Initialized with values greater than one Can be used to control access to a pool of identical resources –Decrement the semaphore’s counter when taking resource from pool –Increment the semaphore’s counter when returning it to pool –If no resources are available, thread is blocked until a resource becomes available

27. COP 5994 - Operating Systems27 Implementing Semaphores Application level –typically implemented by busy waiting –inefficient Kernel implementations block waiting threads via locks disable interrupts via masks –must avoid poor performance and deadlock –hard to implement on multiprocessor systems

28. COP 5994 - Operating Systems28 Linux Synchronization Tools to protect kernel data structures: –spin lock reader/writer lock seqlock –kernel semaphore general thread synchronization –System V Semaphores

29. COP 5994 - Operating Systems29 Linux Kernel Spin Locks Protects critical sections on SMP systems –Once acquired, all subsequent requests to the spin lock cause busy waiting (spinning) until the lock is released Unnecessary in uniprocessor systems –code removed for speed

30. COP 5994 - Operating Systems30 Linux Kernel Reader/Writer Locks optimize concurrency for read/write of kernel data –Allow multiple kernel control paths to hold a read lock, but permit only one kernel control path to hold a write lock with no concurrent readers –A kernel control path that holds a read lock on a critical section must release its read lock and acquire a write lock if it wishes to modify data –An attempt to acquire a write lock succeeds only if there are no other readers or writers concurrently executing inside their critical sections.

31. COP 5994 - Operating Systems31 Linux Kernel Seqlocks Seqlocks –Allow writers to access data immediately without waiting for readers to release the lock –Combines spinlock with a sequence counter –Requires readers to detect if a writer has modified the value of the data protected by the seqlock by examining the value of the seqlock’s sequence counter –Appropriate for interrupt handling

32. COP 5994 - Operating Systems32 Linux Kernel Semaphores Counting semaphores –before entering critical section, must call function down If the value of the counter is greater than 0, decrement the counter, allow the process to execute. If the value of the counter is less than or equal to 0, down decrements the counter, and the process is added to the wait queue and enters the sleeping state. –Reduces the overhead due to busy waiting –when a process exits its critical section, must call function up If the value of the counter is greater than or equal to 0, increments counter If the value of the counter is less than 0, up increments the counter, and a process from the wait queue is awakened to execute its critical section

33. COP 5994 - Operating Systems33 Linux System V Semaphores accessible via the system call interface Semaphore arrays –Protect a group of related resources –Before a process can access resources protected by a semaphore array, the kernel requires that there be sufficient available resources to satisfy the process’s request –Otherwise, kernel blocks requesting process until resources become available

34. COP 5994 - Operating Systems34 Windows XP Synchronization Tools Dispatcher objects –states: signaled vs. unsignaled –operation: wait specify maximum waiting time –variations Event, Mutex, Semaphore, Waitable timer

35. COP 5994 - Operating Systems35 Windows XP kernel synchronization Spin lock Queued spin lock –Guarantees FIFO ordering of requests Fast mutex –Like a mutex, but more efficient –Cannot specify maximum wait time –Reacquisition by owning thread causes deadlock Executive resource lock –One lock holder in exclusive mode –Many lock holders in shared mode

36. COP 5994 - Operating Systems36 Windows XP: Other synchronization tools Critical section object –Like mutex, but only for threads of same process –Faster than mutex, no maximum wait time Timer-queue timer –Waitable timer objects combined with thread pool Interlocked variable access –Atomic operations on variables Interlocked singly-linked lists –Atomic insertion and deletion

37. COP 5994 - Operating Systems37 Monitor Programming language construct –Contains data and procedures needed to access shared resource resource accessible only within the monitor Supports: –mutual exclusion –synchronization Dijkstra, Brinch-Hansen, Hoare

38. COP 5994 - Operating Systems38 Monitor: mutual exclusion Entry to monitor is controlled Resource access using monitor: –thread must call monitor entry routine –only one thread is allowed into monitor –other threads must wait

39. COP 5994 - Operating Systems39 Monitor: mutual exclusion Resource release using monitor: –Monitor entry routine alerts one waiting thread to acquire resource and enter monitor –Higher priority given to waiting threads than ones newly arrived Avoids indefinite postponement

40. COP 5994 - Operating Systems40 Monitor: synchronization Special monitor variable: condition variable every condition variable has associated queue operations: –wait(condVar) thread is suspended –signal(condVar) suspended thread may resume

41. COP 5994 - Operating Systems41 Monitor: condition variables Before a thread can reenter the monitor, the thread calling signal must first exit monitor –Signal-and-exit monitor Requires thread to exit the monitor immediately upon signaling

42. COP 5994 - Operating Systems42 Monitor: condition variables Signal-and-continue monitor –Allows thread inside monitor to signal that the monitor will soon become available –Still maintain lock on the monitor until thread exits monitor –Thread can exit monitor by waiting on a condition variable or by completing execution of code protected by monitor

43. COP 5994 - Operating Systems43 Dining Philosophers Example monitor dp { enum {thinking, hungry, eating} state[5]; condition self[5]; void pickup(int i) // following slides void putdown(int i) // following slides void test(int i) // following slides void init() { for (int i = 0; i < 5; i++) state[i] = thinking; }

44. COP 5994 - Operating Systems44 Dining Philosophers void pickup(int i) { state[i] = hungry; test(i); if (state[i] != eating) self[i].wait(); }

45. COP 5994 - Operating Systems45 Dining Philosophers void putdown(int i) { state[i] = thinking; // test left and right neighbors test((i+4) % 5); test((i+1) % 5); }

46. COP 5994 - Operating Systems46 Dining Philosophers void test(int i) { if ( (state[(i + 4) % 5] != eating) && (state[i] == hungry) && (state[(i + 1) % 5] != eating)) { state[i] = eating; self[i].signal(); }

47. COP 5994 - Operating Systems47 Java Monitors enables thread mutual exclusion and synchronization Signal-and-continue monitors –Allow a thread to signal that the monitor will soon become available –Maintain a lock on monitor until thread exits monitor method keyword synchronized

48. COP 5994 - Operating Systems48 Java Monitors wait method –releases lock on monitor, thread is placed in wait set –condition variable is “this” object –when thread reenters monitor, reason for waiting may not be met notify and notifyAll –signal waiting thread(s)

49. COP 5994 - Operating Systems49 Agenda for next week: Homework: –implement Dining Philosophers with Java monitor Chapter 7 –Deadlock Chapter 8 –Processor scheduling Read ahead !