How to use DirectIO in PHP

Regarding PHP file operations, we will also learn through a series of articles. Today we will first learn about an extension that few people have used, and many people even don’t know about it. It is slightly different from our daily file operations. However, these differences are not intuitively visible to our naked eyes. They mainly lie in the balance between business needs and performance.

What is Direct IO

Direct IO is actually a concept in the Linux operating system. It means to directly operate the file stream. Why is it said to be direct? In fact, when our operating system performs file operations, it does not directly read and write files on the disk immediately. There is a layer of page cache in the middle.

Since it is a cache, it will certainly bring a certain performance improvement, but this is not completely absolute. The direct operation is to ignore this layer of cache operations and directly read and write files on the disk. We all know that there is a huge gap between the processing speed of disks, even solid-state drives, and CPU and memory. The default page cache is used to bridge this gap.

However, the page cache will increase the computing operations of the CPU and occupy memory. Direct operation does not have this problem, but relatively speaking, its speed is not comparable to the file reading operation with cache. Comparable.

The above is a simple understanding of Direct IO. For a more detailed explanation, you can refer to the second link in the reference document at the end of the article and conduct in-depth study. In PHP, we download the Direct IO extension directly from PECL and install it according to the normal installation method of the extension.

Create and write files

Since it is a file operation, let us first create and write some file data.

$fd = dio_open("./test", O_RDWR | O_CREAT);

echo dio_write($fd, "This is Test.I'm ZyBlog.Show me the money4i"), PHP_EOL;
// 43

print_r(dio_stat($fd));
// Array
// (
//     [device] => 64768
//     [inode] => 652548
//     [mode] => 35432
//     [nlink] => 1
//     [uid] => 0
//     [gid] => 0
//     [device_type] => 0
//     [size] => 43
//     [block_size] => 4096
//     [blocks] => 8
//     [atime] => 1602643459
//     [mtime] => 1602656963
//     [ctime] => 1602656963
// )

dio_close($fd);

Similar to the f series functions, we need to use a dio_open() function to open a file. The O_RDWR | O_CREAT parameter means to open a read-write file and create it if the file does not exist. . These two constants correspond to the constants related to direct operation of files in Linux. You can also see the explanation of these constants in the link at the end of the article.

The writing operation can also be completed using a dio_write(). The content it returns is the length of the written content. Here we wrote 43 characters.

dio_stat() returns some information about the current file handle. We can see some information such as device number device, uid, gid, atime, mtime, etc. They are similar to the information we can see in Linux. In fact, it is some simple information about this file.

Reading files

Reading files can be done very simply using a function.

$fd = dio_open("./test", O_RDWR | O_CREAT);

echo dio_read($fd), PHP_EOL;
// This is Test.I'm ZyBlog.Show me the money4i

dio_close($fd);

dio_read() function also contains another parameter, which can read the content according to the specified byte length. We will see related examples later.

File operation

During the file reading process, we may only need to read part of the content, or start reading the file content from a certain position. The following operation function is for These two aspects operate.

$fd = dio_open("./test", O_RDWR | O_CREAT);

var_dump(dio_truncate ($fd , 20)); 
// bool(true)
echo dio_read($fd), PHP_EOL;
// This is Test.I'm ZyB

dio_seek($fd, 3); 

echo dio_read($fd), PHP_EOL;
// s is Test.I'm ZyB

dio_close($fd);

In fact, it can be seen from the name that dio_truncate() is used to truncate the content of the file. Here we truncate from the 20th character, and then use dio_read() to read only the first 20 characters.

dio_seek() specifies which character to start reading the content from. After we specify the starting character position as 3, the first three characters will not be read. It should be noted that dio_truncate() will modify the contents of the original file, while dio_seek() will not.

Other settings

$fd = dio_open('./test', O_RDWR | O_NOCTTY | O_NONBLOCK);

dio_fcntl($fd, F_SETFL, O_SYNC);

dio_tcsetattr($fd, array(
  'baud' => 9600,
  'bits' => 8,
  'stop'  => 1,
  'parity' => 0
));

while (($data = dio_read($fd, 4))!=false) {
    echo $data, PHP_EOL;
}
// This
//  is
// Test
// .I'm
//  ZyB

dio_close($fd);

dio_fcntl() function is called the fcntl function in the c function library. The purpose is to perform some specified operations on the file descriptor. This operation is also fixed with some constants. Yes, here we use F_SETFL, which means to set the file descriptor flag to the specified value. This O_SYNC means that if this descriptor is set, the write operation of the file will wait until the data is written. It doesn't end until it's on disk. Of course, this function can also be set with many other operators. You can refer to PHP's official documentation for in-depth study.

dio_tcsetattr() is used to set the terminal attributes and baud rate of the open file. baud represents the baud rate, bits represents the bits, stop represents the stop bit, and parity represents the parity bit.

This aspect requires some knowledge in "Principles of Computer Composition" and "Operating System". I am not very clear about it, so I will not explain it in detail. It can be seen from here that the basic courses in college classes are really very important. I believe that students who have studied these professional basic courses will immediately understand the role of this function.

Finally, we used the second parameter in dio_read() to read the file content according to the byte length. You can see that the read content is segment by segment in units of 4 characters in length. output.

总结

函数的学习还是比较简单的，核心的还是要知道这个扩展在什么业务场景下更适合使用。在文章开头的介绍中我们已经说明了直接操作文件与普通文件操作的一些区别，在自缓存应用或者需要传输非常大的数据时，直接操作对于 CPU 和 内存 更加地友好。

而其它情况，我们还是使用系统默认的文件操作方式就可以了。其实在大部分情况下，我们基本看不出来它们的显著区别。所以在实际应用中，还是那句话，结合业务实际情况，选择最佳的方案。

测试代码：

https://github.com/zhangyue0503/dev-blog/blob/master/php/202010/source/4.PHP中DirectIO直操作文件扩展的使用.php

推荐学习：php视频教程

The above is the detailed content of How to use DirectIO in PHP. For more information, please follow other related articles on the PHP Chinese website!