Http://codeschool.org/ This work is licensed under a Creative Commons Attribution-ShareAlike 3.0 Unported License. Unix system calls (part 1) history and.

http://codeschool.org/This work is licensed under a Creative Commons Attribution-ShareAlike 3.0 Unported License.

Unix system calls (part 1)

• history and usage of Python• basic data types and the type hierarchy• syntax• modules and variable scopes


Unix system calls(part 1)


This is one part of a larger series. You may need to view previous parts to understand this material.

It’s a Unix system!

System V BSD

1980’s

Linux Mac OS X

FreeBSD, OpenBSD

today

POSIX (Portable Operating System Interface for Unix)

SUS (Single Unix Specification)

Process C

Process B

kernel

Process A

jump to system call code via special instruction

RAM

0x76 00 00 00system call 0



0xA2 22 00 10system call 3

0xFF 31 21 14system call 4




… …

stack

code

heap

heap

heap

kernel code pages only accessible in system calls

jump to system call code via

special instruction

frame of main

frame of cat

frame of dog

frame of fishstack space

frame of syscall

created

waiting running

blocked

terminated

• processes• files• networking sockets• signals• inter-process communication• terminals• threads• I/O devices

ssize_t read(int fd, void *buf, size_t count);

ssize_t read(int fd, void *buf, size_t count);

read(fd)

process:address spaceuser idsfile descriptorsenvironmentcurrent and root directory

stack

heap

code

heap

stack

code

initialized data

heap

heap

kernel code

uninitialized data

global variables with initial values

global variables without initial values

stack

code

initialized data

heap

heap

kernel code

uninitialized data

a.k.a. the “text”

global variables with initial values

global variables without initial values

starts empty, grows automatically

explicitly allocated during execution

mmap(‘memory map’ pages to the process address space)

munmap(‘memory unmap’ pages from the process address space)

mmap(‘memory map’ pages to the process address space)

munmap(‘memory unmap’ pages from the process address space)

address = mmap(5000)… # do stuff with memory at addressmunmap(address)

stack

code

initialized data

heap

heap

kernel code

uninitialized data

heap

heap

mmap fails when not enough space

garbage collection

if fork() == 0: … // new (child) processelse: … // original (parent) process

RAM

byte 0

byte n

HD

stack

heap

code

heap

fork

stack

heap

code

heap

RAM

byte 0

byte n

HD

stack

heap

code

heap

fork

stack

heap

code

heap

RAM

byte 0

byte n

HD

stack

heap

code

heap

fork

stack

heap

code

heap

RAM

byte 0

byte n

HD

stack

heap

code

heap

write

fork

stack

heap

code

heap

RAM

byte 0

byte n

HD

stack

heap

code

heap

copy

write

fork

stack

heap

code

heap

exec

code

exec

(executable)

if fork() == 0: // new (child) process exec(‘/games/pong’)else: … // original (parent) process

pid 1 (init)

pid 85 pid 17

pid 24pid 230

pid 104

pid 34

pid 50

_exit(terminate the process)

_exit(0)

wait(block the process until child process terminates)

pid = fork()if pid == 0: // new (child) process exec(‘/games/pong’)else: // original (parent) process code = wait(pid)

TERM=xtermSHELL=/bin/bashUSER=greysMAIL=/var/mail/ted PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/binPWD=/home/tedEDITOR=vim

name=value

pid 1 (init), user 0

pid 85, user 8 pid 17, user 4

pid 24, user 33pid 230, user 8

pid 104, user 33

pid 34, user 4

pid 50, user 4

user accounts:

/etc/passwd

user accounts:

/etc/passwd

superuser/root = user id 0

privileged to do anything it wants

each process has three user ids:

each file and directory is owned by a single user

“real” id:the owning user

“effective” id:determines privileges

“saved” id:set by exec to match the effective id

exec (sets effective and saved ids when binary file has setuid bit)


seteuid (sets effective user id)

setuid (sets real, effective, and saved user ids)


seteuid (sets effective user id)

setuid (sets real, effective, and saved user ids)

non-superuser can only directly set effective id to match the real or saved id


pid 3 (shell), user 1780

pid 2 (login), user 0




fork, exec




fork, exec

fork, setuid, exec

user groups:

/etc/group

• user may belong to multiple groups but has one “primary” group• each file and directory is owned by one group• each process has a real, effective, and saved group id• binary files have setgid bit• setegid and setgid

rwx rwx rwxuser group other


if file_user_id == effective_user_id:user class

else if file_group_id == effective_group_id:group class

else:other

file permissions:read: can read bytes of filewrite: can modify bytes of fileexecute: can exec file

directory permissions:read: can get names of fileswrite: can add/remove/rename filesexecute: can use in file paths


/adams/taft/garfield/eisenhower


/adams/taft/garfield/eisenhower/adams/taft/ (OK)

r-xr-xr-xrw-r-----r-x--x--xrwx------

/adams/lincoln/adams/cleveland/roosevelt/fillmore


dr-xrw-r-x /adams/


http://codeschool.org/

This work is licensed under a Creative Commons Attribution-ShareAlike 3.0 Unported License.

Http://codeschool.org/ This work is licensed under a Creative Commons Attribution-ShareAlike 3.0 Unported License. Unix system calls (part 1) history and.

Documents