open()系统调用用来打开一个文件，本文就VFS层，对open系统调用的过程进行一个简单的分析。
SYSCALL_DEFINE3(open, constchar __user *, filename, int, flags, int, mode)
{
long ret;
if (force_o_largefile())
flags |= O_LARGEFILE;
ret = do_sys_open(AT_FDCWD, filename, flags, mode);
/* avoid REGPARM breakage on x86: */
asmlinkage_protect(3, ret, filename, flags, mode);
return ret;
}
force_o_largefile()用来判断系统是否为32位的，如果不是32位，也就是说为64位，则将O_LARGEFILE置位，主体工作由do_sys_open()来做
long do_sys_open(int dfd, constchar __user *filename, int flags, int mode)
{
char *tmp = getname(filename);//拷贝文件名字符串到内核空间
int fd = PTR_ERR(tmp);
if (!IS_ERR(tmp)) {
fd = get_unused_fd_flags(flags);//为文件分配一个文件描述符
if (fd >= 0) {
//实际的OPEN操作处理
struct file *f = do_filp_open(dfd, tmp, flags, mode, 0);
if (IS_ERR(f)) {
put_unused_fd(fd);
fd = PTR_ERR(f);
} else {
fsnotify_open(f->f_path.dentry);
fd_install(fd, f);
}
}
putname(tmp);
}
return fd;
}
open操作是特定于某个进程进行的，因此涉及到了VFS中特定于进程的结构，这里简单的介绍下
struct files_struct {
/*
* read mostly part
*/
atomic_t count;
struct fdtable *fdt;
struct fdtable fdtab;
/*
* written part on a separate cache line in SMP
*/
spinlock_t file_lock ____cacheline_aligned_in_smp;
int next_fd;
struct embedded_fd_set close_on_exec_init;
struct embedded_fd_set open_fds_init;
struct file * fd_array[NR_OPEN_DEFAULT];
};
count表示共享该结构的进程数
fdtable是该进程的文件描述符数组
fdt指向fdtable
next_fd表示最大文件描述符号+1
embedded_fd_set是一个位图结构，用来标记文件描述符，close_on_exec_init用来标记那些执行exec时要关闭的文件的文件描述符，open_fds_init用来标记已经分配出去了的文件描述符
fd_array用来存储进程打开的文件的struct file指针
do_sys_open()的一个重要任务就是调用get_unused_fd_flags()为即将打开的文件分配一个文件描述符
#define get_unused_fd_flags(flags) alloc_fd(0, (flags))
int alloc_fd(unsigned start, unsigned flags)
{
struct files_struct *files = current->files;//获取当前进程的files_struct
unsigned int fd;
int error;
struct fdtable *fdt;
spin_lock(&files->file_lock);
repeat:
fdt = files_fdtable(files);//获取进程的fdtable
fd = start;
if (fd < files->next_fd)
fd = files->next_fd;
if (fd < fdt->max_fds)
fd = find_next_zero_bit(fdt->open_fds->fds_bits,
fdt->max_fds, fd);//从位图中获取一个空闲位
error = expand_files(files, fd);//这里根据需要扩充文件描述符数组
if (error < 0)
goto out;
/*
* If we needed to expand the fs array we
* might have blocked - try again.
*/
if (error)//之前进行了扩充操作，重新进行一次空闲bit的搜索
goto repeat;
if (start <= files->next_fd)
files->next_fd = fd + 1;
FD_SET(fd, fdt->open_fds);//在open_fds的位图上置位
if (flags & O_CLOEXEC)//如果设定了O_CLOEXEC，则在close_on_exec位图上将相应位置位
FD_SET(fd, fdt->close_on_exec);
else
FD_CLR(fd, fdt->close_on_exec);
error = fd;
#if 1
/* Sanity check */
if (rcu_dereference(fdt->fd[fd]) != NULL) {
printk(KERN_WARNING "alloc_fd: slot %d not NULL!\n", fd);
rcu_assign_pointer(fdt->fd[fd], NULL);
}
#endif
out:
spin_unlock(&files->file_lock);
return error;
}
int alloc_fd(unsigned start, unsigned flags)
{
struct files_struct *files = current->files;//获取当前进程的files_struct
unsigned int fd;
int error;
struct fdtable *fdt;
spin_lock(&files->file_lock);
repeat:
fdt = files_fdtable(files);//获取进程的fdtable
fd = start;
if (fd < files->next_fd)
fd = files->next_fd;
if (fd < fdt->max_fds)
fd = find_next_zero_bit(fdt->open_fds->fds_bits,
fdt->max_fds, fd);//从位图中获取一个空闲位
error = expand_files(files, fd);//这里根据需要扩充文件描述符数组
if (error < 0)
goto out;
/*
* If we needed to expand the fs array we
* might have blocked - try again.
*/
if (error)//之前进行了扩充操作，重新进行一次空闲bit的搜索
goto repeat;
if (start <= files->next_fd)
files->next_fd = fd + 1;
FD_SET(fd, fdt->open_fds);//在open_fds的位图上置位
if (flags & O_CLOEXEC)//如果设定了O_CLOEXEC，则在close_on_exec位图上将相应位置位
FD_SET(fd, fdt->close_on_exec);
else
FD_CLR(fd, fdt->close_on_exec);
error = fd;
#if 1
/* Sanity check */
if (rcu_dereference(fdt->fd[fd]) != NULL) {
printk(KERN_WARNING "alloc_fd: slot %d not NULL!\n", fd);
rcu_assign_pointer(fdt->fd[fd], NULL);
}
#endif
out:
spin_unlock(&files->file_lock);
return error;
}
实际的扩充操作：
static int expand_fdtable(struct files_struct *files, int nr)
__releases(files->file_lock)
__acquires(files->file_lock)
{
struct fdtable *new_fdt, *cur_fdt;
spin_unlock(&files->file_lock);
new_fdt = alloc_fdtable(nr);//根据nr重新创建一个新的fdtable
spin_lock(&files->file_lock);
if (!new_fdt)
return -ENOMEM;
/*
* extremely unlikely race - sysctl_nr_open decreased between the check in
* caller and alloc_fdtable(). Cheaper to catch it here...
*/
/*这里为了防止因为竞争，在alloc_fdtable调用之前systl_nr_open减小了新创建的fdtable小于nr*/
if (unlikely(new_fdt->max_fds <= nr)) {
free_fdarr(new_fdt);
free_fdset(new_fdt);
kfree(new_fdt);
return -EMFILE;
}
/*
* Check again since another task may have expanded the fd table while
* we dropped the lock
*/
cur_fdt = files_fdtable(files);//获取旧的fdtable
if (nr >= cur_fdt->max_fds) {//新的nr必须大于旧的fdtable的大小
/* Continue as planned */
copy_fdtable(new_fdt, cur_fdt);//将旧的fdtable中的内容拷贝至新的fdtable
rcu_assign_pointer(files->fdt, new_fdt);//用新的fdtable替换旧的fdtable
if (cur_fdt->max_fds > NR_OPEN_DEFAULT)
free_fdtable(cur_fdt);//释放旧的fdtable
} else {
/* Somebody else expanded, so undo our attempt */
free_fdarr(new_fdt);
free_fdset(new_fdt);
kfree(new_fdt);
}
return 1;
}
到此为止，分配新的fd的工作完成，如果分配fd成功，接下来do_sys_open()就要通过do_filp_open()函数查找文件并执行相应的打开操作
do_filp_open的工作针对两种情况进行：
1.flag中未标识O_CREAT，也就是只进行单纯的搜索打开，如果没有搜索到目标文件的话，不会进行创建，这种情况处理起来比较简单，主要工作就是通过路径解析来查找文件，查找到了的话再根据文件系统定义的open方式进行打开
2.flag中标识了O_CREAT，也就是说如果没找到目标文件要进行创建。这种情况要先查找目标文件的父目录(通过将LOOKUP_PARENT标识置位然后进行路径解析来实现)，因为假如没查找到目标文件的话，创建工作需要在父目录下完成；然后再查找最后一个文件分量，也就是目标文件，并进行打开操作
struct file *do_filp_open(int dfd, const char *pathname,
int open_flag, int mode, int acc_mode)
{
struct file *filp;
struct nameidata nd;
int error;
struct path path;
struct dentry *dir;
int count = 0;
int will_write;
int flag = open_to_namei_flags(open_flag);
if (!acc_mode)
acc_mode = MAY_OPEN | ACC_MODE(flag);
/* O_TRUNC implies we need access checks for write permissions */
if (flag & O_TRUNC)
acc_mode |= MAY_WRITE;
/* Allow the LSM permission hook to distinguish append
access from general write access. */
if (flag & O_APPEND)
acc_mode |= MAY_APPEND;
/*
* The simplest case - just a plain lookup.
*/
/*如果没有设置O_CREAT，则在未找到文件的情况下不用创建文件，直接通过查找来打开文件*/
if (!(flag & O_CREAT)) {
error = path_lookup_open(dfd, pathname, lookup_flags(flag),
&nd, flag);
if (error)
return ERR_PTR(error);
goto ok; //成功查找到了目标文件的话，就跳转到ok去执行后续操作
}
/*
* Create - we need to know the parent.
*/
/*如果需要creat,那么就要知道目标文件的父目录，因此需要设置LOOKUP_PARENT标识*/
error = path_init(dfd, pathname, LOOKUP_PARENT, &nd);
if (error)
return ERR_PTR(error);
/*进行路径名的解析，父目录将保存到nd中*/
error = path_walk(pathname, &nd);
if (error) {
if (nd.root.mnt)
path_put(&nd.root);
return ERR_PTR(error);
}
if (unlikely(!audit_dummy_context()))
audit_inode(pathname, nd.path.dentry);
/*
* We have the parent and last component. First of all, check
* that we are not asked to creat(2) an obvious directory - that
* will not do.
*/
error = -EISDIR;
/*这里要先保证路径名的最后一个分量是普通文件名(不为.和..)，并且长度不为0*/
if (nd.last_type != LAST_NORM || nd.last.name[nd.last.len])
goto exit_parent;
error = -ENFILE;
filp = get_empty_filp();//分配一个struct file
if (filp == NULL)
goto exit_parent;
/*将打开文件的信息保存在nd.intent中*/
nd.intent.open.file = filp;
nd.intent.open.flags = flag;
nd.intent.open.create_mode = mode;
dir = nd.path.dentry;//获取父目录
nd.flags &= ~LOOKUP_PARENT;//取消LOOKUP_PARENT标识
nd.flags |= LOOKUP_CREATE | LOOKUP_OPEN;//设置CREATE和OPEN标识
if (flag & O_EXCL)
nd.flags |= LOOKUP_EXCL;
mutex_lock(&dir->d_inode->i_mutex);
//lookup_hash进行最终分量的查找，先查找dentry缓存，没找到的话再通过特定于文件系统的lookup方式从磁盘查找
path.dentry = lookup_hash(&nd);
path.mnt = nd.path.mnt;
do_last:
error = PTR_ERR(path.dentry);//检查目标dentry是否有效
if (IS_ERR(path.dentry)) {
mutex_unlock(&dir->d_inode->i_mutex);
goto exit;
}
if (IS_ERR(nd.intent.open.file)) {//检查file是否有效
error = PTR_ERR(nd.intent.open.file);
goto exit_mutex_unlock;
}
/* Negative dentry, just create the file */
if (!path.dentry->d_inode) {//dentry没有对应上inode，创建之，可能的情况就是该文件被删除了
/*
* This write is needed to ensure that a
* ro->rw transition does not occur between
* the time when the file is created and when
* a permanent write count is taken through
* the 'struct file' in nameidata_to_filp().
*/
error = mnt_want_write(nd.path.mnt);
if (error)
goto exit_mutex_unlock;
/*__open_namei_create将会调用到父目录所属文件系统中定义的create方式创建文件*/
error = __open_namei_create(&nd, &path, flag, mode);
if (error) {
mnt_drop_write(nd.path.mnt);
goto exit;
}
/*nameidata_to_filp将会调用目标文件的inode对应的open函数进行打开操作*/
filp = nameidata_to_filp(&nd, open_flag);
if (IS_ERR(filp))
ima_counts_put(&nd.path,
acc_mode & (MAY_READ | MAY_WRITE |
MAY_EXEC));
mnt_drop_write(nd.path.mnt);
if (nd.root.mnt)
path_put(&nd.root);
return filp;
}
/*
* 下面的情况对应目标文件存在
*/
mutex_unlock(&dir->d_inode->i_mutex);
audit_inode(pathname, path.dentry);
error = -EEXIST;
if (flag & O_EXCL)
goto exit_dput;
/*下面要做一些必要的检查*/
if (__follow_mount(&path)) {//检测目标对象上是否挂载了文件系统
error = -ELOOP;
if (flag & O_NOFOLLOW)
goto exit_dput;
}
error = -ENOENT;
if (!path.dentry->d_inode)//检测目标对象的inode是否存在
goto exit_dput;
if (path.dentry->d_inode->i_op->follow_link)//检测目标对象是否为链接文件
goto do_link;
/*检查OK，将path保存至nd*/
path_to_nameidata(&path, &nd);
error = -EISDIR;
if (path.dentry->d_inode && S_ISDIR(path.dentry->d_inode->i_mode))
goto exit;
ok:
/*
* Consider:
* 1. may_open() truncates a file
* 2. a rw->ro mount transition occurs
* 3. nameidata_to_filp() fails due to
* the ro mount.
* That would be inconsistent, and should
* be avoided. Taking this mnt write here
* ensures that (2) can not occur.
*/
will_write = open_will_write_to_fs(flag, nd.path.dentry->d_inode);
if (will_write) {
error = mnt_want_write(nd.path.mnt);
if (error)
goto exit;
}
/*may_open()会做一些检测*/
error = may_open(&nd.path, acc_mode, flag);
if (error) {
if (will_write)
mnt_drop_write(nd.path.mnt);
goto exit;
}
//执行文件系统定义的打开操作，并保存信息至filp
filp = nameidata_to_filp(&nd, open_flag);
if (IS_ERR(filp))
ima_counts_put(&nd.path,
acc_mode & (MAY_READ | MAY_WRITE | MAY_EXEC));
/*
* It is now safe to drop the mnt write
* because the filp has had a write taken
* on its behalf.
*/
if (will_write)
mnt_drop_write(nd.path.mnt);
if (nd.root.mnt)
path_put(&nd.root);
return filp;
exit_mutex_unlock:
mutex_unlock(&dir->d_inode->i_mutex);
exit_dput:
path_put_conditional(&path, &nd);
exit:
if (!IS_ERR(nd.intent.open.file))
release_open_intent(&nd);
exit_parent:
if (nd.root.mnt)
path_put(&nd.root);
path_put(&nd.path);
return ERR_PTR(error);
do_link://目标文件为符号链接的处理，前文已经分析过
error = -ELOOP;
if (flag & O_NOFOLLOW)
goto exit_dput;
/*
* This is subtle. Instead of calling do_follow_link() we do the
* thing by hands. The reason is that this way we have zero link_count
* and path_walk() (called from ->follow_link) honoring LOOKUP_PARENT.
* After that we have the parent and last component, i.e.
* we are in the same situation as after the first path_walk().
* Well, almost - if the last component is normal we get its copy
* stored in nd->last.name and we will have to putname() it when we
* are done. Procfs-like symlinks just set LAST_BIND.
*/
nd.flags |= LOOKUP_PARENT;
error = security_inode_follow_link(path.dentry, &nd);
if (error)
goto exit_dput;
error = __do_follow_link(&path, &nd);
if (error) {
/* Does someone understand code flow here? Or it is only
* me so stupid? Anathema to whoever designed this non-sense
* with "intent.open".
*/
release_open_intent(&nd);
if (nd.root.mnt)
path_put(&nd.root);
return ERR_PTR(error);
}
nd.flags &= ~LOOKUP_PARENT;
if (nd.last_type == LAST_BIND)
goto ok;
error = -EISDIR;
if (nd.last_type != LAST_NORM)
goto exit;
if (nd.last.name[nd.last.len]) {
__putname(nd.last.name);
goto exit;
}
error = -ELOOP;
if (count++==32) {
__putname(nd.last.name);
goto exit;
}
dir = nd.path.dentry;
mutex_lock(&dir->d_inode->i_mutex);
path.dentry = lookup_hash(&nd);
path.mnt = nd.path.mnt;
__putname(nd.last.name);
goto do_last;
}