# File Descriptor
> 延伸閱讀: [「一切皆為檔案」的理念與解讀](https://hackmd.io/@sysprog/io-universality)
檔案描述符（file descriptor）就是一個整數，它對每個程序（process）都是私有的，在 UNIX 系統中用來存取檔案。因此，一旦檔案被開啟，只要你有權限，你就可以透過這個檔案描述符去讀取或寫入檔案。 從這個角度來看，file descriptor 是一種「能力」(capability) —— 它是一個不透明的操作控制代碼，可以賦予你執行某些特定操作的能力。
也可以把 file descriptor 想像成一個指向 file 類型物件的指標，一旦你持有這個物件，你就能使用像 read() 和 write() 這些「方法」來存取檔案。在 UNIX 系統中，每個 process 的 proc 結構中都會維護一個簡單的資料結構（例如陣列）透過 file descriptor 作為索引，來追蹤目前有哪些檔案是被這個程序開啟的。這個陣列的每一個元素其實就是一個指向 struct file 的指標，用來儲存目前正在讀寫的檔案的詳細資訊
> [source](https://man7.org/training/download/lusp_fileio_slides.pdf)
### tips
可以用 strace 來追蹤程式呼叫了哪些系統呼叫（system call）、傳遞了什麼參數、回傳了什麼值
- -f 可以追蹤 fork 出來的子程序；
- -t 會列出每個呼叫發生的時間；
- -e trace=open,close,read,write 只追蹤這幾個系統呼叫，忽略其他的。
範例
```clike
#include <fcntl.h> // open()
#include <unistd.h> // read(), write(), close()
#include <stdio.h> // perror()
int main() {
char buffer[128];
int fd = open("test.txt", O_RDONLY);
if (fd < 0) {
perror("open");
return 1;
}
ssize_t bytes_read = read(fd, buffer, sizeof(buffer) - 1);
if (bytes_read < 0) {
perror("read");
return 1;
}
buffer[bytes_read] = '\0'; // null terminate
write(STDOUT_FILENO, buffer, bytes_read);
close(fd);
return 0;
}
```
接著執行
``` bash
echo "Hello from file descriptor!" > test.txt
gcc -o fd_example fd_example.c
```
```bash
$ strace -e trace=openat,read,write,close ./fd_example
openat(AT_FDCWD, "/usr/local/cuda-11.8/lib64/glibc-hwcaps/x86-64-v3/libc.so.6", O_RDONLY|O_CLOEXEC) = -1 ENOENT (No such file or directory)
openat(AT_FDCWD, "/usr/local/cuda-11.8/lib64/glibc-hwcaps/x86-64-v2/libc.so.6", O_RDONLY|O_CLOEXEC) = -1 ENOENT (No such file or directory)
openat(AT_FDCWD, "/usr/local/cuda-11.8/lib64/libc.so.6", O_RDONLY|O_CLOEXEC) = -1 ENOENT (No such file or directory)
openat(AT_FDCWD, "glibc-hwcaps/x86-64-v3/libc.so.6", O_RDONLY|O_CLOEXEC) = -1 ENOENT (No such file or directory)
openat(AT_FDCWD, "glibc-hwcaps/x86-64-v2/libc.so.6", O_RDONLY|O_CLOEXEC) = -1 ENOENT (No such file or directory)
openat(AT_FDCWD, "libc.so.6", O_RDONLY|O_CLOEXEC) = -1 ENOENT (No such file or directory)
openat(AT_FDCWD, "/etc/ld.so.cache", O_RDONLY|O_CLOEXEC) = 3
close(3) = 0
openat(AT_FDCWD, "/lib/x86_64-linux-gnu/libc.so.6", O_RDONLY|O_CLOEXEC) = 3
read(3, "\177ELF\2\1\1\3\0\0\0\0\0\0\0\0\3\0>\0\1\0\0\0\220\243\2\0\0\0\0\0"..., 832) = 832
close(3) = 0
openat(AT_FDCWD, "test.txt", O_RDONLY) = 3
read(3, "Hello from file descriptor!\n", 127) = 28
write(1, "Hello from file descriptor!\n", 28Hello from file descriptor!
) = 28
close(3) = 0
+++ exited with 0 +++
```
open() 被底層實作成 openat()，所以 trace 參數要寫 openat。
- `int fd = open("test.txt", O_RDONLY);` => `openat(AT_FDCWD, "test.txt", O_RDONLY) = 3` : = 3 表示這次開啟成功，系統回傳 file descriptor 3
- `read(fd, buffer, sizeof(buffer) - 1);` => `read(3, "Hello from file descriptor!\n", 127) = 28` : fd = 3，成功從 test.txt 讀了 28 byte； 存到 buffer[] 裡
- `write(STDOUT_FILENO, buffer, bytes_read);` => `write(1, "Hello from file descriptor!\n", 28) = 28` : STDOUT_FILENO 是 1；把剛剛讀到的 28 個字元寫到標準輸出（螢幕）；成功寫了 28 bytes。
- `close(fd);` => `close(3) = 0` : 關閉 file descriptor 3；回傳 0 表示關閉成功。
為了要觀察更詳細資訊，在 `close(fd)` 之前加一行 `sleep(120)`
```shell
$ ls -l /proc/<PID>/fd/
total 0
lrwx------ 1 neat neat 64 Apr 23 11:27 0 -> /dev/pts/3
lrwx------ 1 neat neat 64 Apr 23 11:27 1 -> /dev/pts/3
l-wx------ 1 neat neat 64 Apr 23 11:27 19 -> /home/neat/.vscode-server/data/logs/20250423T093438/remoteagent.log
lrwx------ 1 neat neat 64 Apr 23 11:27 2 -> /dev/pts/3
l-wx------ 1 neat neat 64 Apr 23 11:27 20 -> /home/neat/.vscode-server/data/logs/20250423T093438/ptyhost.log
lrwx------ 1 neat neat 64 Apr 23 11:27 21 -> /dev/ptmx
l-wx------ 1 neat neat 64 Apr 23 11:27 22 -> /home/neat/.vscode-server/data/logs/20250423T093438/remoteTelemetry.log
lrwx------ 1 neat neat 64 Apr 23 11:27 23 -> /dev/ptmx
lrwx------ 1 neat neat 64 Apr 23 11:27 24 -> /dev/ptmx
lr-x------ 1 neat neat 64 Apr 23 11:27 3 -> /home/neat/YCL/Workspace/c_test/test.txt
```
可看到最後一行 fd = 3，即是我在程式中 open() 的檔案
```shell
$ cat /proc/1609218/fdinfo/3
pos: 28
flags: 0100000
mnt_id: 31
ino: 37501928
```
- pos: 目前的檔案位移位置（offset），這裡是 28，代表你的程式已經從檔案讀了 28 bytes。
- flags:
- mnt_id:
- ino: 檔案的 inode number
```shell
```
範例:
```bash
prompt> echo hello > foo
prompt> cat foo
hello
prompt> strace cat foo
...
openat(AT_FDCWD, "foo", O_RDONLY) = 3
fstat(3, {st_mode=S_IFREG|0664, st_size=6, ...}) = 0
fadvise64(3, 0, 0, POSIX_FADV_SEQUENTIAL) = 0
mmap(NULL, 139264, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x70950e2bd000
read(3, "hello\n", 131072) = 6
write(1, "hello\n", 6hello
) = 6
read(3, "", 131072) = 0
munmap(0x70950e2bd000, 139264) = 0
close(3) = 0
...
```
`cat` 程式第一步會開啟檔案來進行讀取。我們可以發現: 檔案是以唯讀（read-only）方式開啟的，這由 O_RDONLY 旗標所示；open() 成功並回傳了一個檔案描述符（fd）3。
當 open() 一個新的檔案時（像 cat 這樣），它幾乎一定會得到 fd 3，這是因為每個正在執行的 process 都有三個預設已開啟的檔案: stdin（fd 0）; stdout（fd 1）; stderr（fd 2）
### Linux code
```c
struct fdtable {
unsigned int max_fds;
struct file __rcu **fd; /* current fd array */
unsigned long *close_on_exec;
unsigned long *open_fds;
unsigned long *full_fds_bits;
struct rcu_head rcu;
};
```
>file: https://github.com/torvalds/linux/blob/master/include/linux/fdtable.h
```c
/*
* Open file table structure
*/
struct files_struct {
/*
* read mostly part
*/
atomic_t count;
bool resize_in_progress;
wait_queue_head_t resize_wait;
struct fdtable __rcu *fdt;
struct fdtable fdtab;
/*
* written part on a separate cache line in SMP
*/
spinlock_t file_lock ____cacheline_aligned_in_smp;
unsigned int next_fd;
unsigned long close_on_exec_init[1];
unsigned long open_fds_init[1];
unsigned long full_fds_bits_init[1];
struct file __rcu * fd_array[NR_OPEN_DEFAULT];
};
```
>file: https://github.com/torvalds/linux/blob/master/include/linux/fdtable.h
```c
/**
* struct file - Represents a file
* @f_lock: Protects f_ep, f_flags. Must not be taken from IRQ context.
* @f_mode: FMODE_* flags often used in hotpaths
* @f_op: file operations
* @f_mapping: Contents of a cacheable, mappable object.
* @private_data: filesystem or driver specific data
* @f_inode: cached inode
* @f_flags: file flags
* @f_iocb_flags: iocb flags
* @f_cred: stashed credentials of creator/opener
* @f_owner: file owner
* @f_path: path of the file
* @f_pos_lock: lock protecting file position
* @f_pipe: specific to pipes
* @f_pos: file position
* @f_security: LSM security context of this file
* @f_wb_err: writeback error
* @f_sb_err: per sb writeback errors
* @f_ep: link of all epoll hooks for this file
* @f_task_work: task work entry point
* @f_llist: work queue entrypoint
* @f_ra: file's readahead state
* @f_freeptr: Pointer used by SLAB_TYPESAFE_BY_RCU file cache (don't touch.)
* @f_ref: reference count
*/
struct file {
spinlock_t f_lock;
fmode_t f_mode;
const struct file_operations *f_op;
struct address_space *f_mapping;
void *private_data;
struct inode *f_inode;
unsigned int f_flags;
unsigned int f_iocb_flags;
const struct cred *f_cred;
struct fown_struct *f_owner;
/* --- cacheline 1 boundary (64 bytes) --- */
struct path f_path;
union {
/* regular files (with FMODE_ATOMIC_POS) and directories */
struct mutex f_pos_lock;
/* pipes */
u64 f_pipe;
};
loff_t f_pos;
#ifdef CONFIG_SECURITY
void *f_security;
#endif
/* --- cacheline 2 boundary (128 bytes) --- */
errseq_t f_wb_err;
errseq_t f_sb_err;
#ifdef CONFIG_EPOLL
struct hlist_head *f_ep;
#endif
union {
struct callback_head f_task_work;
struct llist_node f_llist;
struct file_ra_state f_ra;
freeptr_t f_freeptr;
};
file_ref_t f_ref;
/* --- cacheline 3 boundary (192 bytes) --- */
} __randomize_layout
__attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
```
> file: https://github.com/torvalds/linux/blob/master/include/linux/fs.h
```clike
```
>file:https://github.com/torvalds/linux/blob/master/include/linux/sched.h
## reference
https://wiyi.org/linux-file-descriptor.html
Sign in with Wallet
Connect another wallet