Chapter 4: Threads. Operating System Concepts 9 th Edition

Transcription

1 Chapter 4: Threads Silberschatz, Galvin and Gagne 2013

2 Objectives To introduce the notion of a thread a fundamental unit of CPU utilization that forms the basis of multithreaded computer systems To discuss the APIs for the Pthreads, Windows, and Java thread libraries To explore several strategies that provide implicit threading To examine issues related to multithreaded programming To cover operating system support for threads in Windows and Linux 4.2 Silberschatz, Galvin and Gagne 2013

3 Chapter 4: Threads Overview Multicore Programming Multithreading Models Thread Libraries Implicit Threading Threading Issues Operating System Examples 4.3 Silberschatz, Galvin and Gagne 2013

4 Warm-up Discussion: What is abstraction? What is our first abstraction? Why we need it? A new abstraction for a single running process: the thread. Why we need this abstraction? Thinking of: Some programs have more than one point of execution (i.e., multiple PCs). Example #1: adding two large arrays together Discussion: how to speed up execution? Example #2: some elements need input Discussion: how to speed up execution? Discussion: Why multiple threads, not multiple processes? 4.4 Silberschatz, Galvin and Gagne 2013

5 Chapter 4: Threads Overview Multicore Programming Multithreading Models Thread Libraries Implicit Threading Threading Issues Operating System Examples 4.5 Silberschatz, Galvin and Gagne 2013

6 Motivation Most modern applications are multithreaded Threads run within application Multiple tasks with the application can be implemented by separate threads Update display Fetch data Spell checking Answer a network request Process creation is heavy-weight while thread creation is light-weight Can simplify code, increase efficiency Kernels are generally multithreaded 4.6 Silberschatz, Galvin and Gagne 2013

7 Benefits Responsiveness may allow continued execution if part of process is blocked, especially important for user interfaces Resource Sharing threads share resources of process, easier than shared memory or message passing Economy cheaper than process creation, thread switching lower overhead than context switching Scalability process can take advantage of multiprocessor architectures 4.7 Silberschatz, Galvin and Gagne 2013

8 Single and Multithreaded Processes 4.8 Silberschatz, Galvin and Gagne 2013

9 Multithreaded Server Architecture 4.9 Silberschatz, Galvin and Gagne 2013

10 Multithreading MS-DOS JRE UNIX Most modern OS 4.10 Silberschatz, Galvin and Gagne 2013

11 Chapter 4: Threads Overview Multicore Programming Multithreading Models Thread Libraries Implicit Threading Threading Issues Operating System Examples 4.11 Silberschatz, Galvin and Gagne 2013

12 Multicore Programming Multicore or multiprocessor systems putting pressure on programmers, challenges include: Dividing activities Balance Data splitting Data dependency Testing and debugging Parallelism implies a system can perform more than one task simultaneously Concurrency supports more than one task making progress Single processor / core, scheduler providing concurrency 4.12 Silberschatz, Galvin and Gagne 2013

13 Concurrency vs. Parallelism Concurrent execution on single-core system: Parallelism on a multi-core system: 4.13 Silberschatz, Galvin and Gagne 2013

14 Multicore Programming (Cont.) Types of parallelism Data parallelism distributes subsets of the same data across multiple cores, same operation on each Task parallelism distributing threads across cores, each thread performing unique operation As # of threads grows, so does architectural support for threading CPUs have cores as well as hardware threads Consider Oracle SPARC T4 with 8 cores, and 8 hardware threads per core Discussion: How much performance can we gain from adding additional cores? Suppose we have N processing cores Silberschatz, Galvin and Gagne 2013

15 Amdahl s Law Identifies performance gains from adding additional cores to an application that has both serial and parallel components S is serial portion N processing cores That is, if application is 75% parallel / 25% serial, moving from 1 to 2 cores results in speedup of 1.6 times As N approaches infinity, speedup approaches 1 / S Serial portion of an application has disproportionate effect on performance gained by adding additional cores But does the law take into account contemporary multicore systems? 4.15 Silberschatz, Galvin and Gagne 2013

16 Chapter 4: Threads Overview Multicore Programming Multithreading Models Thread Libraries Implicit Threading Threading Issues Operating System Examples 4.16 Silberschatz, Galvin and Gagne 2013

17 User Threads and Kernel Threads Who/How to provide support for threads? User threads - management done by user-level threads library Three primary thread libraries: POSIX Pthreads Windows threads Java threads Kernel threads - Supported by the Kernel Examples virtually all general purpose operating systems, including: Windows Solaris Linux Tru64 UNIX Mac OS X 4.17 Silberschatz, Galvin and Gagne 2013

18 Implementing Threads in Kernel/User Space 4.18 Silberschatz, Galvin and Gagne 2013

19 Motivating Kernel/User Threads Why we need user threads? There is no kernel intervention, and, hence, user threads are usually more efficient. Why user threads are not enough? Unfortunately, since the kernel only recognizes the containing process (of the threads), if one thread is blocked, every other threads of the same process are also blocked because the containing process is blocked. Why we need kernel threads? However, blocking one thread will not cause other threads of the same process to block. The kernel simply runs other threads. In a multiprocessor environment, the kernel can schedule threads on different processors Why kernel threads are not enough? Kernel threads are usually slower than the user threads Silberschatz, Galvin and Gagne 2013

20 Multithreading Models Many-to-One One-to-One Many-to-Many 4.20 Silberschatz, Galvin and Gagne 2013

21 Many-to-One Many user-level threads mapped to single kernel thread One thread blocking causes all to block Question: reason? Multiple threads may not run in parallel on multicore system Question: reason? Used on systems that do not support kernel threads. Few systems currently use this model Examples: Solaris Green Threads GNU Portable Threads 4.21 Silberschatz, Galvin and Gagne 2013

22 One-to-One Each user-level thread maps to kernel thread Creating a user-level thread creates a kernel thread More concurrency than many-to-one Number of threads per process sometimes restricted due to overhead Examples Windows Linux Solaris 9 and later 4.22 Silberschatz, Galvin and Gagne 2013

23 Many-to-Many Model Allows many user level threads to be mapped to many kernel threads Allows the operating system to create a sufficient number of kernel threads Examples Solaris prior to version 9 Windows with the ThreadFiber package 4.23 Silberschatz, Galvin and Gagne 2013

24 Two-level Model Similar to M:M, except that it allows a user thread to be bound to kernel thread Examples IRIX HP-UX Tru64 UNIX Solaris 8 and earlier 4.24 Silberschatz, Galvin and Gagne 2013

25 Aside Question: Among the three models, which model(s) are general in modern systems? M:M for server 1:1 for PC To improve LinuxThreads, two competing projects were started to develop a replacement; NGPT (Next Generation POSIX Threads from IBM) and NPTL (Native POSIX Thread Library from Red Hat). NPTL won out and is today shipped with the vast majority of Linux systems NGPT M:M NPTL 1: Silberschatz, Galvin and Gagne 2013

26 Chapter 4: Threads Overview Multicore Programming Multithreading Models Thread Libraries Implicit Threading Threading Issues Operating System Examples 4.26 Silberschatz, Galvin and Gagne 2013

27 Thread Libraries Thread library provides programmer with API for creating and managing threads Two primary ways of implementing Library entirely in user space Kernel-level library supported by the OS Three main thread libraries POSIX Pthreads Windows Java 4.27 Silberschatz, Galvin and Gagne 2013

28 Pthreads May be provided either as user-level or kernel-level A POSIX standard (IEEE c) API for thread creation and synchronization Specification, not implementation API specifies behavior of the thread library, implementation is up to development of the library Common in UNIX operating systems (Solaris, Linux, Mac OS X) 4.28 Silberschatz, Galvin and Gagne 2013

29 Pthread Creation 1. Which thread is to be created? A pointer to structure of this thread. 2. What kind of thread is it? The stack size and priority information, etc. 3. How to run the thread? The thread entry function (function pointer in C). 4. With what arguments? Arg. Aside: why void *? void * itself is an abstraction. Having a void pointer as an argument to the function start routine allows us to pass in any type of argument; Having it as a return value allows the thread to return any type of result Silberschatz, Galvin and Gagne 2013

30 Pthread Creation Discussion: What is the output? 4.30 Silberschatz, Galvin and Gagne 2013

31 Pthread Completion Wait for completion 1. Which thread is to be completed? The pointer to structure of this thread. 2. What is the return value? void ** Silberschatz, Galvin and Gagne 2013

32 Pthread Completion Discussion: What is the output? 4.32 Silberschatz, Galvin and Gagne 2013

33 Pthreads Example 4.33 Silberschatz, Galvin and Gagne 2013

34 Pthreads Example (Cont.) 4.34 Silberschatz, Galvin and Gagne 2013

35 Pthreads Code for Joining 10 Threads 4.35 Silberschatz, Galvin and Gagne 2013

36 Windows Multithreaded C Program 4.36 Silberschatz, Galvin and Gagne 2013

37 Windows Multithreaded C Program (Cont.) 4.37 Silberschatz, Galvin and Gagne 2013

38 Java Threads Java threads are managed by the JVM Typically implemented using the threads model provided by underlying OS Java threads may be created by: Extending Thread class Implementing the Runnable interface 4.38 Silberschatz, Galvin and Gagne 2013

39 Java Multithreaded Program 4.39 Silberschatz, Galvin and Gagne 2013

40 Java Multithreaded Program (Cont.) Discussion: When this program runs, how many threads are created in the JVM? 4.40 Silberschatz, Galvin and Gagne 2013

41 Chapter 4: Threads Overview Multicore Programming Multithreading Models Thread Libraries Implicit Threading Threading Issues Operating System Examples 4.41 Silberschatz, Galvin and Gagne 2013

42 Implicit Threading Growing in popularity as numbers of threads increase, program correctness more difficult with explicit threads Creation and management of threads done by compilers and run-time libraries rather than programmers Why in this way difficulties can be solved? Three methods explored Thread Pools OpenMP Grand Central Dispatch Other methods include Microsoft Threading Building Blocks (TBB), java.util.concurrent package 4.42 Silberschatz, Galvin and Gagne 2013

43 Thread Pools Create a number of threads in a pool where they await work Advantages: Usually slightly faster to service a request with an existing thread than create a new thread (Reason?) Allows the number of threads in the application(s) to be bound to the size of the pool (Reason?) Separating task to be performed from mechanics of creating task allows different strategies for running task (Reason?) i.e. Tasks could be scheduled to run periodically Windows API supports thread pools: 4.43 Silberschatz, Galvin and Gagne 2013

44 OpenMP Set of compiler directives and an API for C, C++, FORTRAN Provides support for parallel programming in shared-memory environments Identifies parallel regions blocks of code that can run in parallel #pragma omp parallel Create as many threads as there are cores #pragma omp parallel for for(i=0;i<n;i++) { c[i] = a[i] + b[i]; } Run for loop in parallel 4.44 Silberschatz, Galvin and Gagne 2013

45 Grand Central Dispatch Apple technology for Mac OS X and ios operating systems Extensions to C, C++ languages, API, and run-time library Allows identification of parallel sections Manages most of the details of threading Block is in ^{ } - ˆ{ printf("i am a block"); } Blocks placed in dispatch queue Assigned to available thread in thread pool when removed from queue 4.45 Silberschatz, Galvin and Gagne 2013

46 Grand Central Dispatch Two types of dispatch queues: serial blocks removed in FIFO order, queue is per process, called main queue Programmers can create additional serial queues within program concurrent removed in FIFO order but several may be removed at a time Three system wide queues with priorities low, default, high 4.46 Silberschatz, Galvin and Gagne 2013

47 Chapter 4: Threads Overview Multicore Programming Multithreading Models Thread Libraries Implicit Threading Threading Issues Operating System Examples 4.47 Silberschatz, Galvin and Gagne 2013

48 Threading Issues Semantics of fork() and exec() system calls Thread cancellation of target thread Asynchronous or deferred Signal handling Synchronous and asynchronous Thread-local storage Scheduler Activations 4.48 Silberschatz, Galvin and Gagne 2013

49 Semantics of fork() and exec() If one thread in a program calls fork(), does the new process duplicate all threads, or is the new process single-threaded? Two versions of fork(). What about exec()? Replace the entire process, including all thread. Discussion: If exec() is called immediately after forking, which version of fork() do you prefer? If exec() is called immediately after forking, then duplicating all threads is unnecessary 4.49 Silberschatz, Galvin and Gagne 2013

50 Thread Cancellation Terminating a thread before it has finished Thread to be canceled is target thread Two general approaches: Asynchronous cancellation terminates the target thread immediately Deferred cancellation allows the target thread to periodically check if it should be cancelled Pthread code to create and cancel a thread: 4.50 Silberschatz, Galvin and Gagne 2013

51 Thread Cancellation (Cont.) Invoking thread cancellation requests cancellation, but actual cancellation depends on thread state If thread has cancellation disabled, cancellation remains pending until thread enables it Default type is deferred Cancellation only occurs when thread reaches cancellation point I.e. pthread_testcancel() Then cleanup handler is invoked On Linux systems, thread cancellation is handled through signals 4.51 Silberschatz, Galvin and Gagne 2013

52 Signal Handling n n n Signals are used in UNIX systems to notify a process that a particular event has occurred. A signal handler is used to process signals 1. Signal is generated by particular event 2. Signal is delivered to a process 3. Signal is handled by one of two signal handlers: 1. default 2. user-defined Every signal has default handler that kernel runs when handling signal l l User-defined signal handler can override default For single-threaded, signal delivered to process 4.52 Silberschatz, Galvin and Gagne 2013

53 Portable Signal Default Action Description number SIGABRT 6 Terminate (core Signal dump) Handling Process abort signal SIGALRM 14 Terminate Alarm clock SIGBUS N/A Terminate (core dump) Access to an undefined portion of a memory object. SIGCHLD N/A Ignore Child process terminated, stopped, or continued. SIGCONT N/A Continue Continue executing, if stopped. SIGFPE N/A Terminate (core dump) Erroneous arithmetic operation. SIGHUP 1 Terminate Hangup. SIGILL N/A Terminate (core dump) Illegal instruction. SIGINT 2 Terminate Terminal interrupt signal. SIGKILL 9 Terminate Kill (cannot be caught or ignored). SIGPIPE N/A Terminate Write on a pipe with no one to read it. SIGPOLL N/A Terminate Pollable event. SIGPROF N/A Terminate Profiling timer expired. SIGQUIT 3 Terminate (core dump) Terminal quit signal. SIGSEGV N/A Terminate (core dump) Invalid memory reference. SIGSTOP N/A Stop Stop executing (cannot be caught or ignored). SIGSYS N/A Terminate (core dump) Bad system call. SIGTERM 15 Terminate Termination signal. SIGTRAP 5 Terminate (core dump) Trace/breakpoint trap. SIGTSTP N/A Stop Terminal stop signal. SIGTTIN N/A Stop Background process attempting read. SIGTTOU N/A Stop Background process attempting write. SIGURG N/A Ignore High bandwidth data is available at a socket. SIGUSR1 N/A Terminate User-defined signal 1. SIGUSR2 N/A Terminate User-defined signal 2. SIGVTALRM N/A Terminate Virtual timer expired. SIGWINCH N/A Ignore Terminal window size changed SIGXCPU N/A Terminate (core dump) CPU time limit exceeded. SIGXFSZ Operating System Concepts N/A 9 th Edition Terminate (core dump) 4.53 File size limit exceeded Silberschatz, Galvin and Gagne 2013

54 Signal Handling (Cont.) n Where should a signal be delivered for multi-threaded? l l l l Deliver the signal to the thread to which the signal applies Deliver the signal to every thread in the process Deliver the signal to certain threads in the process Assign a specific thread to receive all signals for the process l Discussion: Can a thread have its own handler? l All threads in a process share a same signal handler, which can be either a default one or a user-defined one (overridden) Silberschatz, Galvin and Gagne 2013

55 Thread-Local Storage Question: How to achieve parallelism in executing counter ++? Each thread counts, and summarize when needed. Question: Is counter shared by all threads in the process, or owned by each thread? Not shared within the process, also not in thread stacks. Thread-local storage (TLS) allows each thread to have its own copy of data Useful when you do not have control over the thread creation process (i.e., when using a thread pool) Different from local variables Local variables visible only during single function invocation TLS visible across function invocations Similar to static data TLS is unique to each thread Discussion: Why not use the thread stack? Life cycle reasons. Local variables are in thread stack 4.55 Silberschatz, Galvin and Gagne 2013

56 Scheduler Activations In M:M or Two-level model, how to maintain a appropriate number of kernel threads allocated to the application? Communication, but how? Typically use an intermediate data structure between user and kernel threads lightweight process (LWP) Appears to be a virtual processor on which process can schedule user thread to run Each LWP attached to kernel thread How many LWPs to create? Scheduler activations provide upcalls -a communication mechanism from the kernel to the upcall handler in the thread library. (vs. downcalls) This communication allows an application to maintain the correct number kernel threads 4.56 Silberschatz, Galvin and Gagne 2013

57 Scheduler Activations Question: Suppose an I/O-bound application which has five different file-read requests occur simultaneously. At some moment, two kernel threads are allocated to the application. How many LWPs are created? At some other moment, one kernel thread blocks (waiting for I/O completion) T/F? This kernel thread makes an upcall, and the attached LWP blocks the current user thread, and schedules another user thread to run. T/F? The attached LWP also blocks. The kernel makes an upcall and then allocates a new LWP to the application. After returning from upcall handling, how many LWPs now? If we wish to run this application efficiently, how many LWPs are needed for this application Silberschatz, Galvin and Gagne 2013

58 Exercise 4.1 是资源分配的最小单位 是 CPU 调度的最小单位 A B C D E 进程线程轻量级进程 (LWP) 内核线程用户线程 4.58 Silberschatz, Galvin and Gagne 2013

59 Chapter 4: Threads Overview Multicore Programming Multithreading Models Thread Libraries Implicit Threading Threading Issues Operating System Examples 4.59 Silberschatz, Galvin and Gagne 2013

60 Operating System Examples Windows Threads Linux Threads 4.60 Silberschatz, Galvin and Gagne 2013

61 Windows Threads Windows implements the Windows API primary API for Win 98, Win NT, Win 2000, Win XP, and Win 7 Implements the one-to-one mapping, kernel-level Each thread contains A thread id Register set representing state of processor Separate user and kernel stacks for when thread runs in user mode or kernel mode Private data storage area used by run-time libraries and dynamic link libraries (DLLs) The register set, stacks, and private storage area are known as the context of the thread 4.61 Silberschatz, Galvin and Gagne 2013

62 Windows Threads (Cont.) The primary data structures of a thread include: ETHREAD (executive thread block) includes pointer to process to which thread belongs and to KTHREAD, in kernel space KTHREAD (kernel thread block) scheduling and synchronization info, kernel-mode stack, pointer to TEB, in kernel space TEB (thread environment block) thread id, user-mode stack, thread-local storage, in user space 4.62 Silberschatz, Galvin and Gagne 2013

63 Windows Threads Data Structures 4.63 Silberschatz, Galvin and Gagne 2013

64 Linux Threads Linux refers to them as tasks rather than threads Thread creation is done through clone() system call clone() allows a child task to share the address space of the parent task (process) Flags control behavior struct task_struct points to process data structures (shared or unique) 4.64 Silberschatz, Galvin and Gagne 2013

65 Review What are the differences and relations between a process and a thread? What are the differences between a kernel thread and a user thread? What are M:1, 1:1, M:M models? Compare signal and interrupt. 实验 : 多进程 / 线程编程实现矩阵乘法 1. Linux 下进程的创建 : 函数 fork() 和 exec( ) 的使用 2. Linux/Windows 下多线程编程实现矩阵乘法 3. 每个乘积矩阵元素的计算过程均由一个独立线程实现 4.65 Silberschatz, Galvin and Gagne 2013

66 More Complex Example What are possible outputs of the program? void * hellofunc ( void * ptr ) { int *data; data = (int *) ptr; printf( I m Thread %d \n, *data); } int main() { pthread_t hthread[4]; for (int i = 0; i < 4; i++) pthread_create(&hthread[i], NULL, hellofunc, (void *)&i); for (int i = 0; i < 4; i++) pthread_join(hthread[i], NULL); return 0; } 4.66 Silberschatz, Galvin and Gagne 2013

67 More Complex Example Non-decreasing sequence satisfying Output Line #1 Output Line #2 Output Line #3 Output Line #4 Possible Outputs 4.67 Silberschatz, Galvin and Gagne 2013

68 End of Chapter 4 Silberschatz, Galvin and Gagne 2013