linux overview coms w4118 spring 2008. slides based on phil hutto, silberschatz 2 history linux is a...
TRANSCRIPT
LinuxOverview
COMS W4118
Spring 2008
Slides based on Phil Hutto, Silberschatz2
History Linux is a modern, free operating system based on UNIX standards First developed as a small but self-contained kernel in 1991 by Linus
Torvalds, with the major design goal of UNIX compatibility Its history has been one of collaboration by many users from all
around the world, corresponding almost exclusively over the Internet It has been designed to run efficiently and reliably on common PC
hardware, but also runs on a variety of other platforms The core Linux operating system kernel is entirely original, but it can
run much existing free UNIX software, resulting in an entire UNIX-compatible operating system free from proprietary code
Many, varying Linux Distributions including the kernel, applications, and management tools
Slides based on Phil Hutto, Silberschatz3
Linux Lineage
Multics: 1964 Corbato MIT, GE/Honeywell, Bell Labs UNIX: 1969 Thompson & Ritchie AT&T Bell Labs BSD: 1978 Berkeley Software Distribution Commercial Vendors: Sun, HP, IBM, SGI, DEC GNU: 1984 Richard Stallman, FSF POSIX: 1986 IEEE Portable Operating System unIX Minix: 1987 Andy Tanenbaum SVR4/Solaris: 1989 AT&T and Sun Linux: 1991 Linus Torvalds Intel 386 (i386) Open Source: GPL, LGPL, Cathedral and Bazaar
Slides based on Phil Hutto, Silberschatz4
Linux Distributions Standard, precompiled sets of packages, or distributions,
include the basic Linux system, system installation and management utilities, and ready-to-install packages of common UNIX tools
The first distributions managed these packages by simply providing a means of unpacking all the files into the appropriate places; modern distributions include advanced package management
Early distributions included SLS and Slackware Red Hat and Debian are popular distributions from
commercial and noncommercial sources, respectively The RPM Package file format permits compatibility
among certain Linux distributions
Slides based on Phil Hutto, Silberschatz5
Linux Licensing
The Linux kernel is distributed under the GNU General Public License (GPL), the terms of which are set out by the Free Software Foundation
Anyone using Linux, or creating their own derivative of Linux, may not make the derived product proprietary; software released under the GPL may not be redistributed as a binary-only product
Slides based on Phil Hutto, Silberschatz6
Linux Features
Linux is a “UNIX-like” operating system that “aims at” standards compliance
Linus Torvalds does not feel constrained to implement things that he things are “broken” (e.g. STREAMS) even if they are part of some standard
Still most standard things are supported and Linux includes many innovative features
Lot’s of sharing of features occurs between rival operating systems; if BSD or Windows or Mac OS comes up with a good idea, Linux will quickly incorporate a similar, often improved version
Slides based on Phil Hutto, Silberschatz7
OS Components
Kernel: core OS code loaded at boot Daemons: background servers that provide system
services (e.g. init, inetd, ntpd, httpd…) Libraries: libc, libdl, etc. System software: compilers, linkers, loaders,
debuggers, virus scanners, etc. System monitors: top, df, ps, vmstat, etc. Utilities: backup, compression, recovery, etc. Interfaces: windowing (GUI), shells (CLI)
Slides based on Phil Hutto, Silberschatz8
Modules Modules were originally developed to support the conditional inclusion
of device drivers Early OS kernels would need to either:
1) include code for all possible devices or 2) be recompiled to add support for a new device
Modules provided a means to dynamically add pre-compiled device code to an existing kernel binary
Modules are basically dynamically linked libraries for the kernel! Modules could also be unloaded if the device was no longer in use (not
usually done with DLLs) Most modern OS kernels (e.g. Solaris, BSD, Windows) support kernel
modules Linux uses kernel modules extensively as a structuring technique and
allows file systems, network protocols, system calls, etc. to be defined in kernel modules
Slides based on Phil Hutto, Silberschatz9
Linux Core Kernel
CPU Management Interrupts, exceptions, context switching, timers
Memory Management Memory addressing, allocation, page frame management,
swapping Process Management
Descriptors, threads, address spaces, lifecycle, synchronization, signals, inter-process communication
Device Management Drivers, device caches and buffer, file systems, networking,
disk and packet scheduling
Slides based on Phil Hutto, Silberschatz10
Core KernelApplications
System Libraries (libc)
System Call Interface
Hardware
Architecture-Dependent Code
I/O Related Process RelatedScheduler
Memory Management
IPC
File Systems
Networking
Device Drivers
Mod
ules
Slides based on Phil Hutto, Silberschatz11
Linux Source Tree Spend some time poking around in the source tree using
lxr.linux.no You should familiarize yourself with the structure of the source
tree and get an idea about which directories contain the most code
The latest kernel contains 5 to 7 million lines of code (depending on how you count)
By comparison, Open Office is about 5 million lines of code and the latest Debian release (kernel plus all included applications) is about 200 million lines of code!
Take a look at the Wikipedia entry “Source lines of code” (SLOC) for more fun …
Slides based on Phil Hutto, Silberschatz12
Architecture Dependencies Linux uses a clever technique to deal with architecture (CPU)
dependencies (cf. device dependencies) Logically architecture dependencies comprise a hardware abstraction
layer (HAL) but effected code is spread across hundreds of files One approach would be to use #ifdef macros but this leads to
unreadable code Linux isolates and abstracts all architecture dependencies into small
functions or macros (e.g. disable interrupts) Architecture-dependent versions of each function or macro reside in the
arch/* subdirectories The kernel Makefile does all the work of linking the appropriate
architecture dependent versions without any #ifdefs The only #ifdefs that appear in the kernel source are for optional
capabilities (like support for multiple processors or hyper-threading or kernel modules) that can be compiled out during kernel configuration