Chapter 4: Threads &

Concurrency

Operating System Concepts – 10th Edition Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 Chapter 4: Threads
 Overview
 Multi-core Programming
 Multi-threading Models
 Thread Libraries
 Implicit Threading
 Threading Issues
 Operating System Examples

Operating System Concepts – 10th Edition 4.2 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 Objectives
 Identify the basic components of a thread, and contrast threads
and processes
 Describe the benefits and challenges of designing multi-
threaded applications
 Illustrate different approaches to implicit threading, including
thread pools and fork-join
 Describe how the Linux operating system represents threads
 Explore multi-threaded applications using the Pthreads, Java,
and Windows threading APIs

Operating System Concepts – 10th Edition 4.3 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 Motivation

 Kernels are generally multi-threaded

 Most modern applications are multi-threaded
 Whereas cooperating processes are independent, cooperating
threads run within the same process (think application)
 Multiple functions or tasks within an application can be
implemented by separate threads. Example decomposition:
 A thread to update display
 A thread to fetch data from a database
 A thread to run a tool such as a spell-checker
 A thread to respond to a network request
 Process creation is costly and slow, whereas thread creation
is light-weight
 Proper threading can simplify code, increase efficiency

Operating System Concepts – 10th Edition 4.4 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 Single versus Multi-threaded Processes

Each thread has its own register set, stack, and PC.

Operating System Concepts – 10th Edition 4.5 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 Multi-threaded Server Architecture

1) client sends request to server;

2) server creates a thread to process the request, and
3) immediately returns to listening for the next request
from a client in the same main thread.

Operating System Concepts – 10th Edition 4.6 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 Benefits of Threads

 Responsiveness – may allow continued execution if part of

process is blocked, or some slow operation in a different thread -
especially important for user interfaces
 Resource Sharing – threads share same address space in single
process, easier than processes using shared memory or message
passing
 Cost – cheaper than process creation, thread switching lower
overhead than context switching
 Scalability – process can take advantage of multi-core
architectures

Operating System Concepts – 10th Edition 4.7 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 Concurrency vs. Parallelism
 Two or more sequences of instructions are said to be concurrent if no
matter what order they are executed in relation to each other, the final
result of their combined computation is the same.
 This means that they can be executed simultaneously on different
processors, or interleaved on a single processor in any order, and
whatever outputs they produce will be the same.
 A system with two ore more concurrent processes is called a concurrent
program or a concurrent system.
 Two processes or threads execute in parallel if they execute at the same
time on different processors.
 Parallel programs are those containing instruction sequences that can be
executed in parallel. A parallel program is always a concurrent program,
but a system can have concurrency even though it is not a parallel
program.

Operating System Concepts – 10th Edition 4.8 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 Concurrency vs. Parallelism
 Concurrent execution on single-core system:

 Parallelism on a multi-core system:

Operating System Concepts – 10th Edition 4.9 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 Multi-core Programming

 Multi-core or multi-processor systems challenge programmers to

take advantage of hardware, but it is not easy:
 How to decompose a single task into many independent
parallel tasks
 How to load-balance the tasks
 How to split data onto separate cores/processors
 How to identify data dependency and handle synchronization
 How to test and debug parallel programs

Operating System Concepts – 10th Edition 4.10 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 Multi-core Programming (cont.)

 Types of inherent parallelism:

 Data parallelism – distributes subsets of the same data
across multiple cores, same operation on each
 an image on which the same operation is applied to all
pixels
 a payroll with taxes to be calculated for all individuals
 a set of points to be rotated through same angle in space
 Task parallelism – distributing threads across cores, each
thread performing unique operation
 same data set evaluated by multiple algorithms for some
property (census data analyzed for demographics,
financials, geographic, etc)

Operating System Concepts – 10th Edition 4.11 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 Data and Task Parallelism

Operating System Concepts – 10th Edition 4.12 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 Amdahl’s Law
 In 1967, Gene Amdahl argued that there was an inherent limitation to the amount of speedup
that could be obtained by performing a computation using more processors. His argument is
known as “Amdahl’s Law”. If
 S, 0 <= S <= 1, is the fraction of operations that must be executed serially (in sequence),
and
 N is the number of processing cores, then the speed-up is bounded above:

 Example: if program is 75% parallel / 25% serial, (S=0.25) moving from 1 to 2 cores (N=2)
results in speedup of 1/((1/4) + (3/4)/2)) = 1.6
 As N approaches infinity, speedup approaches 1 / S

Serial portion of an application limits maximum performance gained by adding

additional cores

Operating System Concepts – 10th Edition 4.13 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 Amdahl’s Law Graphically

Operating System Concepts – 10th Edition 4.14 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 User Threads and Kernel Threads

 User threads are supported by user-level libraries

 Three primary user thread libraries:
 POSIX Pthreads
 Windows threads
 Java threads
 Kernel threads are supported directly by the kernel
 Examples – virtually all modern operating systems, including:
 Windows
 Linux
 Mac OS X
 iOS
 Android

Operating System Concepts – 10th Edition 4.15 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 User and Kernel Threads
 When threads are provided as user threads, they still must be
mapped onto kernel threads.
 There is not necessarily an equal number of user and kernel
threads.

Operating System Concepts – 10th Edition 4.16 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 Multi-threading Models

 How to map user threads to kernel threads?

 Three different models:
 Many-to-One: many user-level threads map to single kernel thread

 One-to-One: one user-level thread maps to one kernel thread

 Many-to-Many: many user-level threads map to many kernel threads

Operating System Concepts – 10th Edition 4.17 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 Many-to-One

 Many user-level threads mapped to single kernel thread.

 Weaknesses:
 One thread blocking causes all to block
 Multiple threads may not run in parallel on multi-core
system because only one may be in kernel at a time
 Few systems currently use this model because modern
systems have many cores which are not utilized well.
 Examples:
 Solaris Green Threads
 GNU Portable Threads

Operating System Concepts – 10th Edition 4.18 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 One-to-One
 Each user-level thread maps to one 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:
 Creating a user thread requires creating a kernel thread,
and too many kernel threads can degrade performance of
system.
 Examples
 Windows
 Linux

Operating System Concepts – 10th Edition 4.19 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 Many-to-Many Model
 Allows many user level threads to be multiplexed onto an
equal or smaller number of kernel threads
 Allows the operating system to create a sufficient number of
kernel threads
 Program can have as many user threads as necessary, and
the corresponding kernel threads can run in parallel on a
multiprocessor. If thread blocks, kernel can schedule a different
thread.

 Windows with the

ThreadFiber package
 Otherwise not very common

Operating System Concepts – 10th Edition 4.20 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 Two-level Model
 Similar to the many-to-many, except that it allows a user
thread to be bound to a kernel thread.

Operating System Concepts – 10th Edition 4.21 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 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 prevalent libraries: POSIX threads (Pthreads),
Windows, and Java threads.

Operating System Concepts – 10th Edition 4.22 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 Pthreads

 May be provided either as user-level or kernel-level

 A POSIX standard (IEEE 1003.1c) 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 (Linux & Mac OS X)

Operating System Concepts – 10th Edition 4.23 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 Pthreads Example 1
#include <unistd.h>
#include <stdio.h>
#include <stdlib.h>
#include <pthread.h>

void* hello_world( void* unused)

{
printf("The child says, \"Hello world!\"\n");
pthread_exit(NULL) ;
}

int main( int argc, char *argv[])

{
pthread_t child_thread;

/* Create the thread and launch it. */

if ( 0 != pthread_create(&child_thread, NULL,
hello_world, NULL ) ){
printf("pthread_create failed.\n");
exit(1);
}
printf("This is the parent thread.\n");
/* Wait for the child thread to terminate. */
pthread_join(child_thread, NULL);
return 0;
}
Operating System Concepts – 10th Edition 4.24 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 Pthreads Example 2

Operating System Concepts – 10th Edition 4.25 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 Pthreads Example 2 (cont)

Operating System Concepts – 10th Edition 4.26 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 Pthreads Code for Joining 10 Threads

perating System Concepts – 9 th Edition 4. 21 Silberschatz, Galvin and Gagne ©2013

Operating System Concepts – 10th Edition 4.27 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 Windows Multi-threaded C Program

Operating System Concepts – 10th Edition 4.28 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 Windows Multi-threaded C Program (Cont.)

Operating System Concepts – 10th Edition 4.29 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 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
 Five methods explored
 Thread Pools
 Fork-Join
 OpenMP
 Grand Central Dispatch
 Intel Threading Building Blocks

Operating System Concepts – 10th Edition 4.30 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 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
 Allows the number of threads in the application(s) to be
bound to the size of the pool
 Separating task to be performed from mechanics of
creating task allows different strategies for running task
 i.e.Tasks could be scheduled to run periodically
 Windows API supports thread pools:

Operating System Concepts – 10th Edition 4.31 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 Fork-Join Parallelism
 Multiple threads (tasks) are forked, and then joined.

Operating System Concepts – 10th Edition 4.32 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 Fork-Join Parallelism
 General algorithm for fork-join strategy:

Operating System Concepts – 10th Edition 4.33 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 Fork-Join Parallelism

Operating System Concepts – 10th Edition 4.34 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 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

Operating System Concepts – 10th Edition 4.35 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 OpenMP Example
 Run the for loop in parallel

Operating System Concepts – 10th Edition 4.36 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 Grand Central Dispatch

 Apple technology for macOS and iOS operating systems

 Extensions to C, C++ and Objective-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

Operating System Concepts – 10th Edition 4.37 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 Grand Central Dispatch (cont)
 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
 Four system wide queues divided by quality of service:
o QOS_CLASS_USER_INTERACTIVE
o QOS_CLASS_USER_INITIATED
o QOS_CLASS_USER_UTILITY
o QOS_CLASS_USER_BACKGROUND

Operating System Concepts – 10th Edition 4.38 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 Grand Central Dispatch (3)
 For the Swift language a task is defined as a closure – similar to a block,
minus the caret
 Closures are submitted to the queue using the dispatch_async()
function:

Operating System Concepts – 10th Edition 4.39 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 Threading Issues
 Semantics of fork() and exec() system calls
 Signal handling
 Synchronous and asynchronous
 Thread cancellation of target thread
 Asynchronous or deferred
 Thread-local storage
 Scheduler Activations

Operating System Concepts – 10th Edition 4.40 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 Semantics of fork() and exec()

 Does fork()duplicate only the calling thread or all

threads?
 Some UNIXes have two versions of fork
 exec() usually works as normal – replace the running
process including all threads

Operating System Concepts – 10th Edition 4.41 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 Signal Handling
 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
 User-defined signal handler can override default
 For single-threaded, signal delivered to process

Operating System Concepts – 10th Edition 4.42 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 Signal Handling (Cont.)
 Where should a signal be delivered for multi-threaded?
 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

Operating System Concepts – 10th Edition 4.43 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 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:

Operating System Concepts – 10th Edition 4.44 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 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

Operating System Concepts – 10th Edition 4.45 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 Thread-Local Storage

 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

Operating System Concepts – 10th Edition 4.46 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 Scheduler Activations
 Both M:M and Two-level models require
communication to maintain the appropriate
number of kernel threads allocated to the
application
 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
 This communication allows an application to
maintain the correct number kernel threads

Operating System Concepts – 10th Edition 4.47 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 Operating System Examples

 Windows Threads
 Linux Threads

Operating System Concepts – 10th Edition 4.48 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 Windows Threads

 Windows API – primary API for Windows applications

 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

Operating System Concepts – 10th Edition 4.49 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 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

Operating System Concepts – 10th Edition 4.50 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 Windows Threads Data Structures

Operating System Concepts – 10th Edition 4.51 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 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)

Operating System Concepts – 10th Edition 4.52 Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020
 End of Chapter 4

Operating System Concepts – 10th Edition Silberschatz, Galvin and Gagne ©2018, revised by S. Weiss 2020