26. Code implementation of file using disk

【README】

1. The content of this paper is summarized from B standing 《 operating system - Li Zhijun, teacher of Harbin Institute of technology 》, The content is great , Wall crack recommendation ;

2. Review the contents of the document ：

• 1） How to abstract from a raw disk to a file ？
  • The core is to calculate the disk block number from the position of the character stream ;
• 2） How to calculate the disc number ？
  • Use sequential structure , Or chain structure , Or the mapping table from the character stream provided by the index structure to the disk block number .

---

【1】 Use the disk again , Use through file

 1） fs/read_write.c File reading and writing code

//  stay  fs/read_write.c  in  
// fd :  File descriptor  
// buf:  Memory buffer 
// count:  Number of read and write characters   
int sys_write(int fd, const char* buf, int count) 
{
	struct file *file = current->file[fd]; 
	struct m_inode *inode = file->inode; 
	if (S_ISREG(inode->i_mode)) 
		return file_write(inode, file, buf, count); // file_write  Writing documents 
}

【 Code instructions 】

• Write At the kernel level, the code is the system call   sys_write ;
• Sys_write The code does not use the disk block number , But the disk block number will be calculated through parameters ;
• Actual use of documents , Read and write to disk , Is to deal with a segment in the character stream ;

【 The illustration 】 Read write disk details （ The code level ）

• step 1： retrievable inode,i It's the index ,node refer to FCB（ File control block ）; among FCB Index disk block number stored （ namely The mapping relationship between character stream and disk block number ）;
• step 2： According to the character position and index to be operated （ The mapping relationship ） Calculate the disk block number corresponding to the character stream for reading and writing ; The code behaves as file_write(inode, file, buf, count) , among inode yes index+fcb（ Indexes + Index block ）,file The file character stream position range is stored （ Such as 200~212）,buf Memory buffer address ,count The number of characters ;  

---

【1.1】file_write The whole working process

1）file_write(inode, file, buf, count) File write function

• inode： inode Index disk block number stored （ Mapping relationship between character range and disk block number ）;
• file： Character start position 200;
• buf： Memory buffer ;
• count： The number of characters 12;

adopt file and count You can calculate the character range 200~212 ;

2） The whole process of disk reading and writing is as follows （ User program to disk drive , To disk controller ）：

• step 1： User application call file_write Function write disk , Pass in inode,file,buf,count Parameters ;
• step 2： file The starting position of the character in 200,count by 12, Can represent the character range 200~212;
• step 3： Read the index through the index disk block number （ Mapping relationship between character range and disk block number ）, The disk block number can be found by searching the index according to the starting position of the character 789;
• step 4： Disc number 789 Add to request queue , The interrupt handler in the driver calls the elevator algorithm to select the disk block number , And analyze the disk block number CHS（ Cylinder number , Head number , Sector offset ） Into the disk controller ;
• step 5： The disk controller is based on CHS Read and write to specific sectors ;

3）file_write Of Code implementation

 【 The illustration 】file_write Detailed steps for writing files

• 1） step 1： Find the file character position to operate pos;
• 2） step 2： call create_block() According to the location pos Search index （inode Stored index ） Get the disk number block;
• 3） step 3： bread(...) Add the disk block number to the request queue , Current thread sleep （ When the disk is read and written , Wake up the current thread by the interrupt handler ）;
• 4） step 4： After adding to the request queue , Read characters from disk in the loop body （ Read one character at a time ）;

4）Create_block() Calculate the disk block number code implementation

5） Of the device file inode

 【 The illustration 】

• i_mode Point to character file ;
• i_zone Point to the file content data block ;
• MAJOR ： The main equipment, ;
• MINOR  : Secondary device number ;

---

【2】 Great file view

【2.1】 Operation file

 【 The illustration 】 Operating disk steps

• step 1： Open file ; fd = open(“test.c”) ; // Get file descriptor ;
• step 2： Read the file ; write(fd) ; fd Exist in File list ;
• step 3： Find fd Corresponding inode; （ above 3 A step It's about Get by file name or file path inode）
• step 4： adopt inode Find the disk number ;
• step 5： Send the disk block number into the request queue ;
• step 6： The driver takes the disk block number from the queue , To calculate the CHS Parameters and issue out The command sends parameters to the disk controller ;
• step 7： The disk controller drives the motor to read and write disk sectors , Electricity generates magnetism and magnetism generates electricity ;

【2.2】 Operate the display and other peripherals

• Its call path is also similar to the operation disk path ;
• It's just inode It points to the display , Instead of the file block number ; The operation function is tty_write instead of file_write ;