UNIX Environment advanced programming --3-file io---3.10 file sharing

1、 The kernel is used for all IO Data structure of

Open the back of the file ：fd = open("/tmp/text.c", O_RDWR | O_CREATE, 0777); fd Behind it 3 Data structures , It is this. 3 Data structures , Determines the properties of the open file , And the impact of multiple processes on the same file （ File sharing ）;

（1） Each process has a record entry in the process table , There is an open file descriptor table in the record entry , That is, a vector , Each descriptor occupies a vector bit ; Associated with each file descriptor is ：① file descriptor flags （close_on_exec）;② Pointer to a file table entry

（2） The kernel maintains a file table for all open files , Each file table entry contains ：① Document status flag ： read 、 Write 、 Sync 、 Non blocking, etc ;② Current file offset ;③ Point to the file v node Pointer to table entry

（3）v The node pointer points to a v The node structure ,v The node contains the file type and pointers to functions that perform various operations on this file ,v The contents of the node contain the contents of the file i node

2、i node

analysis ：

①i The blocks of nodes are only stored in i Content of node （ file type , File access rights , File length and number of files pointed to ）, Data block this contains data content , Fragmented space , Catalog items, etc

② Two directory entries point to the same i node , In other words, two more links can be found i Content of node

③stat The structure contains st_nlink（ Hard links ） Member records how many times this node has been linked , So delete a directory entry using unlink Function instead of delete

④ Each file （ equipment ） Can I only correspond to one i node , that i What to do after the nodes are used up ？

⑤ When renaming a file , Just construct a pointer i New directory entry for node , And delete the old directory entries

⑥ The catalog is essentially ” file “, Creating a directory will also point to i node