UNIX Input/Output

UNIX Input/Output
Working with Disk Files and the File System

Prof. David Bernstein
James Madison University

Computer Science Department

bernstdh@jmu.edu

Review

Files in UNIX:
- A file is a sequence of bytes
- All I/O devices are modeled as files
- A device driver is the code that allows for (the standard set of) operations to be performed on an I/O device
File Descriptors:
- A (usually small) nonnegative integer that an application can use to identify a file
Universal Operations:
- All files have some common functionality (e.g., open, read, write, close)

Some Common (though Not Universal) Operations

Change the Current File Position:
- Request a change to the file position (that is an offset from the first byte) maintained by the kernel
Control:
- Manipulate the underlying device parameters

Changing the File Offset: lseek()

off_t lseek(int fd, off_t offset, int whence);

Purpose:

Reposition the (logical) file offset of an open file descriptor (without causing any physical device access)

Details:

`fd`	The file descriptor
`offset`	The offset (in bytes)
`whence`	The base point from which offset is to be interpreted
Return	The new file offset; -1 on error

#include <unistd.h>

Changing the File Offset: whence Values

`SEEK_SET`	The base is the beginning of the file (i.e., the file offset is set to `offset`n which must be non-negative)
`SEEK_CUR`	The base is the current offset
`SEEK_END`	The base is the size of the file

Some Common Idioms

  // Obtain the current file offset
  current = lseek(fd, 0, SEEK_CUR);

  // Set the file offset to the start of the file
  lseek(fd, 0, SEEK_SET);

  // Set the file offset to the next byte after the end of the file
  lseek(fd, 0, SEEK_END);

A Common Sequence of Calls

  // Save the current file offset
  original = lseek(fd, 0, SEEK_CUR);

  // Set the file offset to a particular value
  lseek(fd, offset, SEEK_SET);

  // Read from the new file offset
  size = read(fd, buffer, length);

  // Reset the file offset
  lseek(fd, original, SEEK_SET);

Unfortunately, this sequence of calls is not atmoic (i.e., might be interrupted)!

Reading from an Offset: pread()

ssize_t pread(int fd, void* buffer, size_t count, off_t offset);

Purpose:

Atomically read from a specific file offset, without changing the offset

Details:

`fd`	The file descriptor to read from
`buffer`	The ultimate location of the bytes in memory
`count`	The maximum number of bytes to copy
`offset`	The file offset to copy from
Return	The number of bytes copied; 0 on EOF; -1 on error

#include <unistd.h>

Writing to an Offset: pwrite()

ssize_t pwrite(int fd, const void *buffer, size_t count, off_t offset

Purpose:

Atomically write to a specific file offset, without changing the offset

Details:

`fd`	The file descriptor to copy to
`buffer`	The memory to copy from
`count`	The maximum number of bytes to copy
`offset`	The file offset to copy to
Return	The number of bytes copied; -1 on error

#include <unistd.h>