Understanding the Linux Kernel

17.7. Reading and Writing an Ext2 Regular File

In Chapter 12 we described how the Virtual File System recognizes the type of file being accessed by a read( ) or write( ) system call and invokes the corresponding method of the proper file operation table. We now have all the needed tools to understand how a regular file is actually read or written in the Ext2 filesystem.

There's nothing more to say about read operations, however, because they have already been completely discussed. As shown in Table 17-9, the Ext2's read method is implemented by the generic_file_read( ) function, which is described in the section Section 15.1.1 in Chapter 15.

Let's concentrate then on Ext2's write method, which is implemented by the ext2_file_write( ) function. It acts on four parameters:

fd

File descriptor of the file being written

buf

Address of a memory area containing the data to be written

count

Number of bytes to be written

ppos

Pointer to a variable storing the file offset where data must be written

The function performs the following actions:

1. Removes any superuser privilege from the file (to guard against tampering with setuid programs, described in Chapter 19).

2. If the file has been opened with the O_APPEND flag set, sets the file offset where data must be written to the end of the file.

3. If the file has been opened in synchronous mode (O_SYNC flag set), sets i_osync field in the ext2_inode_info structure of the disk inode to 1. This flag is tested when a data block is allocated for the ...