L4: Input/Output Flashcards
Unix Directory Structure:
Important Directories(7)
/ - Root Directory
/bin - commonly used commands
/dev - Device Files
/etc - System Maintenance Files
/lib - System Libraries
/tmp - Temporary Files
/usr - User Files
Unix Directory Structure:
User File Directory Subdirectories(4)
/usr
/usr/include - System Include Files
/usr/lib - More Library Files
/usr/bin - More Executable Files
/usr/man - Manual Pages
Unix Innovations:
Shell
- Pipes
- Simple model for connecting together programs
- Tools
- Rich set of utilities for common text manipulation tasks
Unix Innovations:
Programming
- C Language
- As efficient and flexible as assembly
- But more readable
- OS Interface
- Convenient subroutine interface to all system services
Unix Innovations:
Processes
- Simplicity
- Unix has a simple model for spawning and coordinating processes
- Uniformity
- Single model for:
- jobs
- concurrency
- memory
- etc
- Single model for:
Two Building Blocks
of
Unix Data
- Files
- Actual storage areas on disk
- Store data
- Processes
- Manipulate data
- Running programs in memory
Unix Innovations:
File System
- Organization
- Simple hierarchical structure
- access control
- Content
- Simple linear byte streams
Unix Minimalist Philosophy
- Build it for programmers
- Keep capabilities few
- Make the capabilities powerful
- Compose complex capabilities FROM simple ones
File Structure
- Regular Files are just arrays of bytes, starting at index 0
- Uses a byte offset to track the “position” while reading and writing to the file/data
- This increases flexibility:
- Can reposition offset to overwrite or reread
- Can move offset to end+1 to append without overwriting
Unix File System Model:
3 Types of Files
- Regular Files
- Array of bytes
- Simple sequence with arbitrary file size
- Directory Files:
- Single hierarchy of files
- Very deep nesting
- Special Files
- Things that LOOK like files, but are not actually on the disk
- terminals,
- network connections,
- memory,
- pipes,
- etc
- Things that LOOK like files, but are not actually on the disk
Error Handling:
Using ERRNO
- Every system routine reports errors this way
- Returns an integer value
- Sets an extern int called errno, from errno.h
- represents the specific error that occurred
- Use the function:
- strerror(errno)
- obtain a string with info about the error
How the Shell launches a program
Example:
%ls -l -t foo.c bar.c < abc
- Creates a new process
- Finds the program file (ls)
- Uses it for instructions and initial data
- Creates an array of command line arguments
- Sets file descriptors 0,1,2 (stdin, stdout,stderr)
- Runs the Process
- First instruction is C startup routine from library
- Calls “main” with the command line arguments
- Wait for process
- Ends with the return
- Shell waits( unless pipe or & )
Library for Working with Files
fcntl.h
File Operation Methods:
Opening a File
method:
int open( char* pathname, int flags, int mode)
- pathname
- The pathname of the file
- flags
- determine how the file is to be opened
- predefined set of flags in fcntl.h
- mode
- permissions(read/write)
File Descriptors
Overview
File Descriptors
- Used to manipulate open files
- Integer numbers used as indices into a (secret) array managed by the kernel
- Each entry in the array has information about the state of an open file
File Information
in the File Descriptor array
The File Descriptor Array manages information about the state of open files
- File location on disk
- Mode(read/write) the file is open in
- Current offset
Changing the
Current Offset
of a File Descriptor
Use the “lseek()” function from the unistd.h library
lseek( fd, value, flag)
Returns a new offset, or -1 if the fd is not a disk file
The flag determines the kind of seek operation
- Set specific offset:
- lseek(fd, i, SEEK_SET)
- new offset = i
- Move by i:
- lseek(fd, i, SEEK_CUR)
- new offset = old_offset + i
- Move i bytes past end of file:
- lseek(fd, i, SEEK_END)
- new offset = size of file + i
Reading from a File
Use the read() method from the unistd.h library
actual = read(fd, buffer, attempt);
- bytes from the file will be copied into the buffer
- the buffer must be allocated beforehand
- attempt defines the expected number of bytes to read
- returns the actual number of bytes read, or 0 if no more bytes to read
- Read begins at the current offset
- After reading, the offset is incremented by “actual”
Opening a File:
Basic Flags
Used in the open(path, flag, mode_permissions)
Defined in the fcntl libary
- O_CREAT
- Create File
- O_RDONLY
- Open for reading
- O_TRUNC
- Set size to 0
- O_WRONLY
- Open for writing
- O_APPEND
- Set offset to EOF before each write
- O_RDWR
- Open for both reading and writing
Writing to a File
Use the write() method from the unistd.h
actual = write(fd, buffer, actual)
- writing begins at the current offset
- bytes in buffer are written to the file
- if “fd” was opened with the O_APPEND flag, the offset is set to EOF + 1, so each write appends
- After the write, offset is incremented by “actual”, or the number of bytes written
Layers of the Unix Architecture
- Kernel
- Essential Basic Services
- Only one kernel
- General functionality
- Libraries
- Useful additional services
- Many libraries
- Specific functionality
- Programs
- Executable files
- Lots of programs
- Specific functionality
- Shell
- Command Line environment
- Few
- Powerful
- Users
- Humans using the computer
Default File Descriptors
- 0 - stdin
- 1 - stdout
- 2 - stderr
Debuggin/Error Handling:
Using Assert
- Include the “assert.h” library
- Verifies that a specified condition is true using an assert macro
- If condition is false, program will abort
- Specifying a condition in code:
- assert( myVar < 2)
- assert( myVar == 5), etc
fprintf()
-Destination
Full function prototype
fprintf( destination, string, arg1, arg2, … )
- destination is a file descriptor
- could be stdout, stderr, fd…
- arguments are substituted into the string following general C string substitution rules
Common String Formatting
Substitution Symbols (5)
- %s - string value
- %c - char value
- %d - decimal integer
- %x - hex integer
- %f - float
Note: This is far from comprehensive
Unix Innovations:
5 Areas
- File System
- Organization
- Content
- Processes
- SImplicity
- Uniformity
- Programming
- C language
- OS Interface
- Shell
- Pipes
- Tools
- Malleability
- Portability
- Open Source
Bash command:
Pass a file to a program/process
%program < file_name
- This file supplies initial content for a new process
Unix Innovations:
Malleability
- Portability
- both OS and tools are written in C
- both can be easily ported to other machines
- Open Source
- source can be licensed and modified for experiments
Specialized File Operations
