深入理解Linux内核：从用户空间到内核的实践指南

需积分: 9 61 浏览量更新于2024-07-20 收藏 4.51MB PDF 举报

"深入理解Linux内核" 《Linux内核精要》是一本针对x86和PowerPC架构的深入探讨Linux内核编程的指南。作者Claudia Salzberg Rodriguez、Gordon Fischer和Steven Smolski采取了一种独特的自顶向下的方法，使内核编程变得更加易于理解，通过系统地追踪功能从用户空间到内核，并将内核内部结构与用户级编程基础紧密关联。这种方法帮助读者利用他们已有的Linux知识，深入理解内核的工作原理以及其组成部分如何相互配合。本书逐步引导读者掌握所有必要的工具和汇编语言编程技术，以便理解和控制内核的行为。作者同时对比了x86和PowerPC架构的实现，通过详尽注解的源代码示例和实际项目，揭示了晦涩的内核功能。这种方式使读者能够深入到内核的细节之中，不仅仅是理论上的学习，更是实践中的探索。这本书的主要内容可能包括但不限于以下几个方面： 1. **内核架构**：介绍Linux内核的整体架构，包括进程管理、内存管理、设备驱动、文件系统、网络协议栈等核心组件。 2. **用户空间与内核空间交互**：讲解如何通过系统调用进入内核，以及内核态和用户态的转换机制。 3. **进程管理**：详细阐述进程创建、调度、同步和通信等过程，以及相关数据结构如进程描述符。 4. **内存管理**：涵盖虚拟内存的概念、页表机制、内存分配策略以及内存回收等。 5. **设备驱动**：探讨如何编写和理解设备驱动程序，包括字符设备和块设备的驱动模型。 6. **文件系统**：分析文件系统的工作原理，如VFS（虚拟文件系统）层、文件操作函数和I/O控制。 7. **网络编程**：讲解网络协议栈的层次结构，如TCP/IP协议的实现，以及socket接口的使用。 8. **汇编语言编程**：教授如何阅读和编写内核级别的汇编代码，这对于理解底层硬件交互至关重要。 9. **调试技巧**：介绍如何使用GDB等工具对内核进行调试，以及如何使用内建的内核调试机制。 10. **性能优化**：讨论如何通过调整内核参数和代码优化来提高系统性能。通过这本书的学习，无论是开发者、系统管理员还是对操作系统感兴趣的读者，都能获得对Linux内核的深刻洞察，从而更好地驾驭这个强大的开源操作系统。

Chapter 7, "Scheduling and Kernel Synchronization," describes the operation of the scheduler, which allows

multiple processes to run as though they are the only process in the system. This chapter covers in detail how

the kernel selects which task to execute and how it interfaces with the hardware to switch from one process to

another. This chapter also describes what kernel preemption is and how it is executed. Finally, it describes

how the system clock works and its use by the kernel to keep time.

Chapter 8, "Booting the Kernel," describes what happens from Power On to Power Off. It traces how the

various processors handle the loading of the kernel, including a description of BIOS, Open Firmware, and

bootloaders. This chapter then goes through the linear order in kernel bringup and initialization, covering all

the subsystems discussed in previous chapters.

Part III deals with a more hands-on approach to building and interacting with the Linux kernel:

Chapter 9, "Building the Linux Kernel," covers the toolchain necessary to build the kernel and the format of

the object files executed. It also describes in detail how the Kernel Source Build system operates and how to

add configuration options into the kernel build system.

Chapter 10, "Adding Your Code to the Kernel," describes the operation of /dev/random, which is seen in

all Linux systems. As it traces the device, the chapter touches on previously described concepts from a more

practical perspective. It then covers how to implement your own device in the kernel.

Our Approach

This book introduces the reader to the concepts necessary to understand the kernel. We follow a top-down

approach in the following two ways:

First, we associate the kernel workings with the execution of user space operations the reader may be more

familiar with and strive to explain the kernel workings in association with this. When possible, we begin with

a user space example and trace the execution of the code down into the kernel. It is not always possible to

follow this tracing straight down since the subsystem data types and substructures need to be introduced

before the explanation of how it works can take place. In these cases, we tie in explanations of the kernel

subsystem with specific examples of how it relates to a user space program. The intent is twofold: to highlight

the layering seen in the kernel as it interfaces with user space on one side and the hardware on the other, and

to explain workings of the subsystem by tracing the code and following the order of events as they occur. We

believe this will help the reader get a sense of how the kernel workings fit in with what he knows, and will

provide him with a framed reference for how a particular functionality associates to the rest of the operating

system.

Second, we use the top-down perspective to view the data structures central to the operation of the subsystem

and see how they relate to the execution of the system's management. We strive to delineate structures central

to the subsystem operation and to keep focus on them as we follow the operation of the subsystem.

Conventions

Throughout this book, you will see listings of the source code. The top-right corner will hold the location of

the source file with respect to the root of the source code tree. The listings are shown in this font. Line

numbers are provided for the code commentary that usually follows. As we explain the kernel subsystem and

how it works, we will continually refer to the source code and explain it.

Command-line options, function names, function output, and variable names are distinguished by this

font.

Bold type is used whenever a new concept is introduced.

Chapter 1. Overview

In this chapter

1.1 History of UNIX 2•

1.2 Standards and Common Interfaces 4•

1.3 Free Software and Open Source 5•

1.4 A Quick Survey of Linux Distributions 5•

1.5 Kernel Release Information 8•

1.6 Linux on Power 8•

1.7 What Is an Operating System? 9•

1.8 Kernel Organization 11•

1.9 Overview of the Linux Kernel 11•

1.10 Portability and Architecture Dependence 26•

Summary 27•

Exercises 27•

Linux is an operating system that came into existence as the hobby of a student named Linus Torvalds in

1991. The beginnings of Linux appear humble and unassuming in comparison to what it has become. Linux

was developed to run on machines with x86 architecture microprocessors with AT hard disks. The first release

sported a bash shell and a gcc compiler. Portability was not a design concern at that time, nor was

widespread use in academia and industry its vision. There was no business plan or vision statement. However,

it has been available for free from day one.

Linux became a collaborative project under the guidance and maintenance of Linus from the early days of

beta versions. It filled a gap that existed for hackers wanting a free operating system that would run on the x86

architecture. These hackers began to contribute code that provided support for their own particular needs.

It is often said that Linux is a type of UNIX. Technically, Linux is a clone of UNIX because it implements the

POSIX UNIX Specification P1003.0. UNIX has dominated the non-Intel workstation scene since its inception

in 1969, and it is highly regarded as a powerful and elegant operating system. Relegated to high-performance

workstations, UNIX was only available at research, academic, and development institutions. Linux brought

the capabilities of a UNIX system to the Intel personal computer and into the homes of its users. Today, Linux

sees widespread use in industry and academia, and it supports numerous architectures, such as PowerPC.

This chapter provides a bird's eye view of the concepts that surround Linux. It takes you through an overview

of the components and features of the kernel and introduces some of the features that make Linux so

appealing. To understand the concepts of the Linux kernel, you need to have a basic understanding of its

intended purpose.

1.1. History of UNIX

We mentioned that Linux is a type of UNIX. Although Linux did not develop directly from an existing UNIX,

the fact that it implements common UNIX standards makes the history of UNIX relevant to our discussion.

MULTiplexed Information and Computing Service (MULTICS), which is considered the precursor of the

UNIX operating systems, came about from a joint venture between MIT, Bell Laboratories, and the General

Electric Company (GEC), which was involved in the computer-manufacturing business at that time. The

development of MULTICS was born of the desire to introduce a machine to support numerous timesharing

users. At the time of this joint venture in 1965, operating systems, although capable of multiprogramming

(timesharing between jobs), were batch systems that supported only a single user. The response time between

a user submitting a job and getting back the output was in the order of hours. The goal behind MULTICS was

to create an operating system that allowed multiuser timesharing that provided each user access to his own

terminal. Although Bell Labs and General Electric eventually abandoned the project, MULTICS eventually

ran in production settings in numerous places.

UNIX development began with the porting of a stripped-down version of MULTICS in an effort to develop an

operating system to run in the PDP-7 minicomputer that would support a new filesystem. The new filesystem

was the first version of the UNIX filesystem. This operating system, developed by Ken Thompson, supported

two users and had a command interpreter and programs that allowed file manipulation for the new filesystem.

In 1970, UNIX was ported to the PDP-11 and updated to support more users. This was technically the first

edition of UNIX.

In 1973, for the release of the fourth edition of UNIX, Ken Thompson and Dennis Ritchie rewrote UNIX in C

(a language then recently developed by Ritchie). This moved the operating system away from pure assembly

and opened the doors to the portability of the operating system. Take a moment to consider the pivotal nature

of this decision. Until then, operating systems were entirely entrenched with the system's architecture

specifications because assembly language is extremely particular and not easily ported to other architectures.

The rewrite of UNIX in C was the first step toward a more portable (and readable) operating system, a step

that contributed to UNIX's sudden rise in popularity.

1974 marked the beginning of a boost in popularity of UNIX among universities. Academics began to

collaborate with the UNIX systems group at Bell Laboratories to produce the fifth edition with many new

innovations. This version was available free of cost and with source code to universities for educational

purposes. In 1979, after many innovations, code cleanups, and an effort to improve portability, the seventh

edition (V7) of the UNIX operating system came about. This version contained a C compiler and a command

interpreter known as the Bourne shell.

The 1980s brought the advent of the personal computer. The workstation was now within the reach of

businesses and universities. A number of UNIX variants were then developed from the seventh edition. These

include Berkley UNIX (BSD), which was developed at the University of California at Berkley, and the AT&T

UNIX System III and System V. Each version was then developed into other systems, such as NetBSD and

OpenBSD (variants of BSD), and AIX (IBM's variant of System V). In fact, all commercial variants of UNIX

are derived from System V or BSD.

Linux was introduced in 1991 at a time when UNIX was extremely popular but not available for the PC. The

cost of UNIX was prohibitive and not really available to a user unless he was affiliated with a university.

Linux was first implemented as an extension of an operating system called Minix (a small operating system

written by Andrew Tanenbaum for educational purposes).

In the following years, the Linux kernel, combined with system software provided by the Free Software

Foundation's (FSF) GNU project, made Linux

[1]

develop into a sufficiently solid system that attracted

attention beyond the scope of the contributing hackers. In 1994, version 1.0 of Linux was released. From then

on, Linux has grown vastly, generating a demand for the distribution of Linux in mass quantities and to an

increasing number of universities, corporations, and individual users that require support on various

architectures.

[1]

Linux is also referred to as GNU/Linux in order to credit the component of system

software provided by the FSF's GNU project.

1.2. Standards and Common Interfaces

Common standards bridge the gaps between the different types of UNIX. The user's decision of what variant

of UNIX to use impacts its portability and, therefore, its potential market. If you are a program developer,

clearly, the market for your program is limited to the people who use the same system you developed on

unless you take the trouble to port it. Standards come about from the need for a specification of a common

programming interface that would facilitate having code developed on one operating system run on another

with minimal or no patching. Various standards organizations have set out to define specifications for UNIX.

POSIX, formed by the Institute of Electronic Engineers (IEEE), is a standard for a portable operating system

for computer environments with which Linux aims to be compliant.

1.3. Free Software and Open Source

Linux is one of the most successful examples of open-source software. Open-source software is software

whose source code is freely available such that anyone can modify, read, and redistribute it. This stands in

contrast to the closed-source software distributed only in binary form.

Open source allows a user to develop the software at will to suit his own needs. Depending on the license,

certain restrictions apply to the code. The benefit of this is that users are never limited by what has been

developed by others because they can freely alter the code to suit their needs. Linux provides an operating

system that allows anyone to develop and contribute to it. This caused a fairly rapid evolution of Linux as the

rate of involvement, whether in development, testing, or documentation, is staggering.

Various open-source licenses exist: In particular, Linux is licensed under the GNU General Public License

(GPL) version 2. A copy of the license can be found at the root of the source code in a file called COPYING.

If you plan on hacking the Linux kernel, it is a good idea to become familiar with the terms of this license so

that you know what the legal fate of your contribution will be.

There are two main camps around the conveyance of free and open-source software. The Free Software

Foundation and the open-source groups differ in ideology. The Free Software Foundation, which is the older

of the two groups, holds the ideology that the word free should be applied to software in much the same way

that the word free is applied to speech. The open-source group views free and open-source software as a

different methodology on par with proprietary software. For more information, go to http://www.fsf.org and

http://www.opensource.org.

1.4. A Quick Survey of Linux Distributions

We mentioned that the Linux kernel is only part of what is usually referred to as "Linux." A Linux distribution

is a combination of the Linux kernel, tools, window managers, and many other applications. Many of the

system programs used in Linux are developed and maintained by the FSF GNU project. With the rise in

Linux's demand and popularity, the packaging of the kernel with these and other tools has becoming a

significant and lucrative undertaking. Groups of people and corporations take on the mission of providing a

particular distribution of Linux in keeping with a particular set of objectives. Without getting into too much

detail, we review the major Linux distributions as of this writing. New Linux distributions continue to be

released.

Most Linux distributions organize the tools and applications into groups of header and executable files. These

groupings are called packages and are the major advantage of using a Linux distribution as opposed to

downloading header files and compiling everything from source. Referring to the GPL, the license gives the

freedom to charge for added value to the open-source software, such as these services provided in the code's

redistribution.

1.4.1. Debian

Debian

[2]

is a GNU/Linux operating system. Like other distributions, the majority of applications and tools

come from GNU software and the Linux kernel. Debian has one of the better package-management systems,

apt (advanced packaging tool). The major drawback of Debian is in the initial installation procedure, which

seems to cause confusion among novice Linux users. Debian is not tied to a corporation and is developed by a

community of volunteers.

[2]

http://www.debian.org.

1.4.2. Red Hat/Fedora

Red Hat

[3]

(the company) is a major player in the open-source software-development arena. Red Hat Linux

was the company's Linux distribution until recently (20022003) when it replaced its sole offering with two

separate distributions Red Hat Enterprise Linux and the Fedora Core. Red Hat Enterprise Linux is aimed at

business, government, or other industries that require a stable and supported Linux environment. The Fedora

Core is targeted to individual users and enthusiasts. The major difference between the two distributions is

stability versus features. Fedora will have newer, less stable code included in the distribution than Red Hat

Enterprise. Red Hat appears to be the Linux enterprise version of choice in America.

[3]

http://www.redhat.com.

1.4.3. Mandriva

Mandriva Linux

[4]

(formerly Mandrake Linux) originated as an easier-to-install version of Red Hat Linux, but

has since diverged into a separate distribution that targets the individual Linux user. The major features of

Mandriva Linux are easy system configuration and setup.

[4]

http://www.mandriva.com/.

1.4.4. SUSE

SUSE Linux

[5]

is another major player in the Linux arena. SUSE targets business, government, industry, and

individual users. The major advantage of SUSE is its installation and administration tool Yast2. SUSE appears

to be the Linux enterprise version of choice in Europe.

[5]

http://www.novell.com/linux/suse/.

剩余539页未读，继续阅读

admant

粉丝: 0
资源: 6

深入理解Linux内核：从用户空间到内核的实践指南

嵌入式开发实战指南：Embedded Linux Primer

《Embedded Linux Primer》：实战导向的嵌入式系统入门指南

免费Linux儿童教育软件Preschool Primer

The Linux Kernel Primer (LINUX内核编程) E文pdf版

The Linux Kernel Primer A Top-Down Approach for x86 and PowerPC Architectures

The Linux Kernel Primer A Top-Down Approach for x86 and PowerPC Architectur

The Linux Kernel Primer A Top Down Approach for x86 and PowerPC Architectures.rar

The.Linux.Kernel.Primer

Embedded Linux Primer

Embedded Linux Primer.pdf

最新资源