前言
驱动程序中实现file_operations中的open函数，在应用层中使用open来调用到内核中实现的file_operations中的open函数，那么如何实现的呢？
inode
内核中字符设备相应文件实现file_operations后生成ko文件，每个文件有一个特定的inode号，每个文件会对应一个inode结构体
struct inode { umode_t i_mode; // 打开方式 kuid_t i_uid; // 用户id kgid_t i_gid; // 组id unsigned int i_flags; // 权限标志位，阻塞，非阻塞... 标志位信息 const struct inode_operations*i_op; // inode 操作方法集 struct super_block*i_sb; struct address_space*i_mapping; unsigned longi_ino; // inode 号码 union { const unsigned int i_nlink; unsigned int __i_nlink; // 链接数 }; dev_ti_rdev; // 设备号 struct timespeci_atime; // 访问时间 struct timespeci_mtime; // 修改时间 struct timespeci_ctime; // 创建时间 // 存储信息 unsigned short i_bytes; unsigned int i_blkbits; blkcnt_t i_blocks; const struct file_operations*i_fop; /* 文件操作方法集合 */ struct list_headi_devices; // 内核循环双链表节点 union { struct pipe_inode_info*i_pipe; struct block_device*i_bdev; struct cdev *i_cdev; // 字符设备对象指针 }; ... };
inode结构体包含很多信息，如文件的权限信息，文件的时间信息等等。在驱动文件中，我们主要关注标红的信息，inode号，设备号，文件操作集合。使用mknod创建设备节点时，会将inode号和设备号关联起来。若此时内核中已经注册了该设备号的对象，此时会将设备对象的操作方法集合赋值到此处。
应用层open->底层驱动xxx_open
文件操作集合i_fop指向了def_chr_fops，这个结构中的open函数指向了chrdev_open。chrdev_open完成的主要工作是：首先根据设备号找到添加在内核中代表字符设备的cdev。找到对应的cedv对象后，用cedv关联的操作方法集和赋值给新建的file结构体中的文件操作集合。并调用操作集合中的open函数，完成真正的打开操作。
一个驱动可能会被多应用打开，那多次打开时是生成多个inode号吗？答案是否定的。inode号是唯一的，那如何表示多次打开一个驱动文件呢？
系统中每个打开的文件在内核空间都有一个对应的file结构体。
struct file { struct path f_path; // 路径名称 struct inode *f_inode; /*inode 指针*/ const struct file_operations *f_op; // 文件操作集合 unsigned int f_flags; // 标志位信息 fmode_t f_mode; // 操作方式 loff_t f_pos; struct fown_struct f_owner; // 异步通知 ，需要用到此结构体 void *private_data; ... };
应用层open->底层驱动xxx_open
应用层open->软中断指令->底层驱动xxx_open
- 应用层中使用open时，触发软中断（int $0x80）
int open(const char *pathname, int flags, mode_t mode);
- 触发软中断后更根据参数的数量触发不同的函数，利用0x80中断劫持system_call->sys_call_table（系统调用表）
static inline long stub_syscall3(long syscall, long arg1, long arg2, long arg3) { long ret; __asm__ volatile ("int $0x80" : "=a" (ret) : "0" (syscall), "b" (arg1), "c" (arg2), "d" (arg3)); return ret; }
sys_call_table（系统调用表）
#include <linux/syscalls.h> #include <linux/signal.h> #include <linux/unistd.h> #include <asm/syscall.h> #undef __SYSCALL #define __SYSCALL(nr, call) [nr] = (call), void *sys_call_table[__NR_syscalls] = { #include <asm/unistd.h> }; • 1 • 2 • 3 • 4 • 5 • 6 • 7 • 8 • 9 • 10 • 11 • 12 • 13
/* * arch/sh/kernel/syscalls_64.S * * Copyright (C) 2000, 2001 Paolo Alberelli * Copyright (C) 2004 - 2007 Paul Mundt * Copyright (C) 2003, 2004 Richard Curnow * * This file is subject to the terms and conditions of the GNU General Public * License. See the file "COPYING" in the main directory of this archive * for more details. */ #include <linux/sys.h> .section .data, "aw" .balign 32 /* * System calls jump table */ .globl sys_call_table[] sys_call_table: .long sys_restart_syscall /* 0 - old "setup()" system call */ .long sys_exit .long sys_fork .long sys_read .long sys_write .long sys_open /* 5 */ .long sys_close .long sys_waitpid .long sys_creat .long sys_link .long sys_unlink /* 10 */ .long sys_execve .long sys_chdir .long sys_time .long sys_mknod .long sys_chmod /* 15 */ .long sys_lchown16 .long sys_ni_syscall /* old break syscall holder */ .long sys_stat .long sys_lseek .long sys_getpid /* 20 */ .long sys_mount .long sys_oldumount .long sys_setuid16 .long sys_getuid16 .long sys_stime /* 25 */ .long sh64_ptrace .long sys_alarm .long sys_fstat .long sys_pause .long sys_utime /* 30 */ .long sys_ni_syscall /* old stty syscall holder */ .long sys_ni_syscall /* old gtty syscall holder */ .long sys_access .long sys_nice .long sys_ni_syscall /* 35 */ /* old ftime syscall holder */ .long sys_sync .long sys_kill .long sys_rename .long sys_mkdir .long sys_rmdir /* 40 */ .long sys_dup .long sys_pipe .long sys_times .long sys_ni_syscall /* old prof syscall holder */ .long sys_brk /* 45 */ .long sys_setgid16 .long sys_getgid16 .long sys_signal .long sys_geteuid16 .long sys_getegid16 /* 50 */ .long sys_acct .long sys_umount /* recycled never used phys( */ .long sys_ni_syscall /* old lock syscall holder */ .long sys_ioctl .long sys_fcntl /* 55 */ .long sys_ni_syscall /* old mpx syscall holder */ .long sys_setpgid .long sys_ni_syscall /* old ulimit syscall holder */ .long sys_ni_syscall /* sys_olduname */ .long sys_umask /* 60 */ .long sys_chroot .long sys_ustat .long sys_dup2 .long sys_getppid .long sys_getpgrp /* 65 */ .long sys_setsid .long sys_sigaction .long sys_sgetmask .long sys_ssetmask .long sys_setreuid16 /* 70 */ .long sys_setregid16 .long sys_sigsuspend .long sys_sigpending .long sys_sethostname .long sys_setrlimit /* 75 */ .long sys_old_getrlimit .long sys_getrusage .long sys_gettimeofday .long sys_settimeofday .long sys_getgroups16 /* 80 */ .long sys_setgroups16 .long sys_ni_syscall /* sys_oldselect */ .long sys_symlink .long sys_lstat .long sys_readlink /* 85 */ .long sys_uselib .long sys_swapon .long sys_reboot .long sys_old_readdir .long old_mmap /* 90 */ .long sys_munmap .long sys_truncate .long sys_ftruncate .long sys_fchmod .long sys_fchown16 /* 95 */ .long sys_getpriority .long sys_setpriority .long sys_ni_syscall /* old profil syscall holder */ .long sys_statfs .long sys_fstatfs /* 100 */ .long sys_ni_syscall /* ioperm */ .long sys_socketcall /* Obsolete implementation of socket syscall */ .long sys_syslog .long sys_setitimer .long sys_getitimer /* 105 */ .long sys_newstat .long sys_newlstat .long sys_newfstat .long sys_uname .long sys_ni_syscall /* 110 */ /* iopl */ .long sys_vhangup .long sys_ni_syscall /* idle */ .long sys_ni_syscall /* vm86old */ .long sys_wait4 .long sys_swapoff /* 115 */ .long sys_sysinfo .long sys_ipc /* Obsolete ipc syscall implementation */ .long sys_fsync .long sys_sigreturn .long sys_clone /* 120 */ .long sys_setdomainname .long sys_newuname .long sys_cacheflush /* x86: sys_modify_ldt */ .long sys_adjtimex .long sys_mprotect /* 125 */ .long sys_sigprocmask .long sys_ni_syscall /* old "create_module" */ .long sys_init_module .long sys_delete_module .long sys_ni_syscall /* 130: old "get_kernel_syms" */ .long sys_quotactl .long sys_getpgid .long sys_fchdir .long sys_bdflush .long sys_sysfs /* 135 */ .long sys_personality .long sys_ni_syscall /* for afs_syscall */ .long sys_setfsuid16 .long sys_setfsgid16 .long sys_llseek /* 140 */ .long sys_getdents .long sys_select .long sys_flock .long sys_msync .long sys_readv /* 145 */ .long sys_writev .long sys_getsid .long sys_fdatasync .long sys_sysctl .long sys_mlock /* 150 */ .long sys_munlock .long sys_mlockall .long sys_munlockall .long sys_sched_setparam .long sys_sched_getparam /* 155 */ .long sys_sched_setscheduler .long sys_sched_getscheduler .long sys_sched_yield .long sys_sched_get_priority_max .long sys_sched_get_priority_min /* 160 */ .long sys_sched_rr_get_interval .long sys_nanosleep .long sys_mremap .long sys_setresuid16 .long sys_getresuid16 /* 165 */ .long sys_ni_syscall /* vm86 */ .long sys_ni_syscall /* old "query_module" */ .long sys_poll .long sys_ni_syscall /* was nfsservctl */ .long sys_setresgid16 /* 170 */ .long sys_getresgid16 .long sys_prctl .long sys_rt_sigreturn .long sys_rt_sigaction .long sys_rt_sigprocmask /* 175 */ .long sys_rt_sigpending .long sys_rt_sigtimedwait .long sys_rt_sigqueueinfo .long sys_rt_sigsuspend .long sys_pread64 /* 180 */ .long sys_pwrite64 .long sys_chown16 .long sys_getcwd .long sys_capget .long sys_capset /* 185 */ .long sys_sigaltstack .long sys_sendfile .long sys_ni_syscall /* getpmsg */ .long sys_ni_syscall /* putpmsg */ .long sys_vfork /* 190 */ .long sys_getrlimit .long sys_mmap2 .long sys_truncate64 .long sys_ftruncate64 .long sys_stat64 /* 195 */ .long sys_lstat64 .long sys_fstat64 .long sys_lchown .long sys_getuid .long sys_getgid /* 200 */ .long sys_geteuid .long sys_getegid .long sys_setreuid .long sys_setregid .long sys_getgroups /* 205 */ .long sys_setgroups .long sys_fchown .long sys_setresuid .long sys_getresuid .long sys_setresgid /* 210 */ .long sys_getresgid .long sys_chown .long sys_setuid .long sys_setgid .long sys_setfsuid /* 215 */ .long sys_setfsgid .long sys_pivot_root .long sys_mincore .long sys_madvise /* Broken-out socket family (maintain backwards compatibility in syscall numbering with 2.4) */ .long sys_socket /* 220 */ .long sys_bind .long sys_connect .long sys_listen .long sys_accept .long sys_getsockname /* 225 */ .long sys_getpeername .long sys_socketpair .long sys_send .long sys_sendto .long sys_recv /* 230*/ .long sys_recvfrom .long sys_shutdown .long sys_setsockopt .long sys_getsockopt .long sys_sendmsg /* 235 */ .long sys_recvmsg /* Broken-out IPC family (maintain backwards compatibility in syscall numbering with 2.4) */ .long sys_semop .long sys_semget .long sys_semctl .long sys_msgsnd /* 240 */ .long sys_msgrcv .long sys_msgget .long sys_msgctl .long sys_shmat .long sys_shmdt /* 245 */ .long sys_shmget .long sys_shmctl /* Rest of syscalls listed in 2.4 i386 unistd.h */ .long sys_getdents64 .long sys_fcntl64 .long sys_ni_syscall /* 250 reserved for TUX */ .long sys_ni_syscall /* Reserved for Security */ .long sys_gettid .long sys_readahead .long sys_setxattr .long sys_lsetxattr /* 255 */ .long sys_fsetxattr .long sys_getxattr .long sys_lgetxattr .long sys_fgetxattr .long sys_listxattr /* 260 */ .long sys_llistxattr .long sys_flistxattr .long sys_removexattr .long sys_lremovexattr .long sys_fremovexattr /* 265 */ .long sys_tkill .long sys_sendfile64 .long sys_futex .long sys_sched_setaffinity .long sys_sched_getaffinity /* 270 */ .long sys_ni_syscall /* reserved for set_thread_area */ .long sys_ni_syscall /* reserved for get_thread_area */ .long sys_io_setup .long sys_io_destroy .long sys_io_getevents /* 275 */ .long sys_io_submit .long sys_io_cancel .long sys_fadvise64 .long sys_ni_syscall .long sys_exit_group /* 280 */ /* Rest of new 2.6 syscalls */ .long sys_lookup_dcookie .long sys_epoll_create .long sys_epoll_ctl .long sys_epoll_wait .long sys_remap_file_pages /* 285 */ .long sys_set_tid_address .long sys_timer_create .long sys_timer_settime .long sys_timer_gettime .long sys_timer_getoverrun /* 290 */ .long sys_timer_delete .long sys_clock_settime .long sys_clock_gettime .long sys_clock_getres .long sys_clock_nanosleep /* 295 */ .long sys_statfs64 .long sys_fstatfs64 .long sys_tgkill .long sys_utimes .long sys_fadvise64_64 /* 300 */ .long sys_ni_syscall /* Reserved for vserver */ .long sys_ni_syscall /* Reserved for mbind */ .long sys_ni_syscall /* get_mempolicy */ .long sys_ni_syscall /* set_mempolicy */ .long sys_mq_open /* 305 */ .long sys_mq_unlink .long sys_mq_timedsend .long sys_mq_timedreceive .long sys_mq_notify .long sys_mq_getsetattr /* 310 */ .long sys_ni_syscall /* Reserved for kexec */ .long sys_waitid .long sys_add_key .long sys_request_key .long sys_keyctl /* 315 */ .long sys_ioprio_set .long sys_ioprio_get .long sys_inotify_init .long sys_inotify_add_watch .long sys_inotify_rm_watch /* 320 */ .long sys_ni_syscall .long sys_migrate_pages .long sys_openat .long sys_mkdirat .long sys_mknodat /* 325 */ .long sys_fchownat .long sys_futimesat .long sys_fstatat64 .long sys_unlinkat .long sys_renameat /* 330 */ .long sys_linkat .long sys_symlinkat .long sys_readlinkat .long sys_fchmodat .long sys_faccessat /* 335 */ .long sys_pselect6 .long sys_ppoll .long sys_unshare .long sys_set_robust_list .long sys_get_robust_list /* 340 */ .long sys_splice .long sys_sync_file_range .long sys_tee .long sys_vmsplice .long sys_move_pages /* 345 */ .long sys_getcpu .long sys_epoll_pwait .long sys_utimensat .long sys_signalfd .long sys_timerfd_create /* 350 */ .long sys_eventfd .long sys_fallocate .long sys_timerfd_settime .long sys_timerfd_gettime .long sys_signalfd4 /* 355 */ .long sys_eventfd2 .long sys_epoll_create1 .long sys_dup3 .long sys_pipe2 .long sys_inotify_init1 /* 360 */ .long sys_preadv .long sys_pwritev .long sys_rt_tgsigqueueinfo .long sys_perf_event_open .long sys_recvmmsg /* 365 */ .long sys_accept4 .long sys_fanotify_init .long sys_fanotify_mark .long sys_prlimit64 .long sys_name_to_handle_at /* 370 */ .long sys_open_by_handle_at .long sys_clock_adjtime .long sys_syncfs .long sys_sendmmsg .long sys_setns /* 375 */ .long sys_process_vm_readv .long sys_process_vm_writev .long sys_kcmp .long sys_finit_module
- 然后执行汇编指令跳转到内核态，最终调用到do_sys_open函数
- 在do_sys_open函数中，经过一系列的调用最终会打开对应驱动，调用到对应驱动中的xxx_open函数
do_sys_open
long do_sys_open(int dfd, const char __user *filename, int flags, int mode) { /*把用户空间的存在内存的文件名字符串复制到内核空间*/ char *tmp = getname(filename); int fd = PTR_ERR(tmp); if (!IS_ERR(tmp)) { /*寻找一个暂时没用的文件描述符，或者没有的话返回错误*/ fd = get_unused_fd(); if (fd >= 0) { /*如果找到的话，就执行打开操作*/ struct file *f = do_filp_open(dfd, tmp, flags, mode); if (IS_ERR(f)) { /*打开失败，就释放文件描述符*/ put_unused_fd(fd); fd = PTR_ERR(f); } else { fsnotify_open(f->f_path.dentry); fd_install(fd, f); } } putname(tmp); } return fd; }
