Understanding the Linux Kernel

13.6. Page I/O Operations

Block devices transfer information one block at a time, while process address spaces (or to be more precise, memory regions allocated for the process) are defined as sets of pages. This mismatch can be hidden to some extent by using page I/O operations (see the section Section 13.5). They may be activated in the following cases:

• A process issues a read( ) or write( ) system call on a regular file (see Section 15.1 in Chapter 15).

• A process reads a location of a page that maps a file in memory (see Section 15.2 in Chapter 15).

• The kernel flushes some dirty pages related to a file memory mapping to disk (see Section 15.2.6 in Chapter 15).

• When swapping in or swapping out, the kernel loads from disk or saves to disk the contents of whole page frames (see Chapter 16).

We'll use the rest of this chapter to describe how these operations are carried out.

13.6.1. Starting Page I/O Operations

A page I/O operation is activated by invoking the brw_page( ) function, which receives the following parameters:

rw

Type of I/O operation (READ or WRITE)

page

Address of a page descriptor

dev

Block device number

b

Array of logical block numbers

size

Block size

bmap

Flag specifying whether the block numbers in b were computed by using the bmap method of the inode operations (see Section 12.2.2 in Chapter 12)

The page descriptor refers to the page involved in the page I/O operation. It must already be locked (PG_locked flag on) before invoking brw_page( ) so that no other kernel ...