/* this routine links an IS_ROOT dentry into the dcache tree. It gains "parent"
* as a parent and "name" as a name
* It should possibly go in dcache.c
*/
int d_splice(struct dentry *target, struct dentry *parent, struct qstr *name)
{
struct dentry *tdentry;
#ifdef NFSD_PARANOIA
if (!IS_ROOT(target))
printk("nfsd: d_splice with no-root target: %s/%s\n", parent->d_name.name, name->name);
if (!(target->d_flags & DCACHE_NFSD_DISCONNECTED))
printk("nfsd: d_splice with non-DISCONNECTED target: %s/%s\n", parent->d_name.name, name->name);
#endif
tdentry = d_alloc(parent, name);
if (tdentry == NULL)
return -ENOMEM;
d_move(target, tdentry);
/* tdentry will have been made a "child" of target (the parent of target)
* make it an IS_ROOT instead
*/
spin_lock(&dcache_lock);
list_del_init(&tdentry->d_child);
tdentry->d_parent = tdentry;
spin_unlock(&dcache_lock);
d_rehash(target);
dput(tdentry);
/* if parent is properly connected, then we can assert that
* the children are connected, but it must be a singluar (non-forking)
* branch
*/
if (!(parent->d_flags & DCACHE_NFSD_DISCONNECTED)) {
while (target) {
target->d_flags &= ~DCACHE_NFSD_DISCONNECTED;
parent = target;
spin_lock(&dcache_lock);
if (list_empty(&parent->d_subdirs))
target = NULL;
else {
target = list_entry(parent->d_subdirs.next, struct dentry, d_child);
#ifdef NFSD_PARANOIA
/* must be only child */
if (target->d_child.next != &parent->d_subdirs
|| target->d_child.prev != &parent->d_subdirs)
printk("nfsd: d_splice found non-singular disconnected branch: %s/%s\n",
parent->d_name.name, target->d_name.name);
#endif
}
spin_unlock(&dcache_lock);
}
}
return 0;
}
void sysfs_rename_dir(struct kobject * kobj, const char *new_name)
{
struct dentry * new_dentry, * parent;
if (!strcmp(kobject_name(kobj), new_name))
return;
if (!kobj->parent)
return;
parent = kobj->parent->dentry;
down(&parent->d_inode->i_sem);
new_dentry = sysfs_get_dentry(parent, new_name);
d_move(kobj->dentry, new_dentry);
kobject_set_name(kobj,new_name);
up(&parent->d_inode->i_sem);
}
/**
* debugfs_rename - rename a file/directory in the debugfs filesystem
* @old_dir: a pointer to the parent dentry for the renamed object. This
* should be a directory dentry.
* @old_dentry: dentry of an object to be renamed.
* @new_dir: a pointer to the parent dentry where the object should be
* moved. This should be a directory dentry.
* @new_name: a pointer to a string containing the target name.
*
* This function renames a file/directory in debugfs. The target must not
* exist for rename to succeed.
*
* This function will return a pointer to old_dentry (which is updated to
* reflect renaming) if it succeeds. If an error occurs, %NULL will be
* returned.
*
* If debugfs is not enabled in the kernel, the value -%ENODEV will be
* returned.
*/
struct dentry *debugfs_rename(struct dentry *old_dir, struct dentry *old_dentry,
struct dentry *new_dir, const char *new_name)
{
int error;
struct dentry *dentry = NULL, *trap;
const char *old_name;
trap = lock_rename(new_dir, old_dir);
/* Source or destination directories don't exist? */
if (!old_dir->d_inode || !new_dir->d_inode)
goto exit;
/* Source does not exist, cyclic rename, or mountpoint? */
if (!old_dentry->d_inode || old_dentry == trap ||
d_mountpoint(old_dentry))
goto exit;
dentry = lookup_one_len(new_name, new_dir, strlen(new_name));
/* Lookup failed, cyclic rename or target exists? */
if (IS_ERR(dentry) || dentry == trap || dentry->d_inode)
goto exit;
old_name = fsnotify_oldname_init(old_dentry->d_name.name);
error = simple_rename(old_dir->d_inode, old_dentry, new_dir->d_inode,
dentry);
if (error) {
fsnotify_oldname_free(old_name);
goto exit;
}
d_move(old_dentry, dentry);
fsnotify_move(old_dir->d_inode, new_dir->d_inode, old_name,
old_dentry->d_name.name, S_ISDIR(old_dentry->d_inode->i_mode),
NULL, old_dentry);
fsnotify_oldname_free(old_name);
unlock_rename(new_dir, old_dir);
dput(dentry);
return old_dentry;
exit:
if (dentry && !IS_ERR(dentry))
dput(dentry);
unlock_rename(new_dir, old_dir);
return NULL;
}
static struct dentry * sysfs_lookup(struct inode *dir, struct dentry *dentry,
struct nameidata *nd)
{
struct dentry *ret = NULL;
struct sysfs_dirent *parent_sd = dentry->d_parent->d_fsdata;
struct sysfs_dirent *sd;
struct inode *inode;
mutex_lock(&sysfs_mutex);
sd = sysfs_find_dirent(parent_sd, dentry->d_name.name);
/* no such entry */
if (!sd) {
ret = ERR_PTR(-ENOENT);
goto out_unlock;
}
/* attach dentry and inode */
inode = sysfs_get_inode(dir->i_sb, sd);
if (!inode) {
ret = ERR_PTR(-ENOMEM);
goto out_unlock;
}
/* instantiate and hash dentry */
ret = d_find_alias(inode);
if (!ret) {
dentry->d_op = &sysfs_dentry_ops;
dentry->d_fsdata = sysfs_get(sd);
d_add(dentry, inode);
} else {
d_move(ret, dentry);
iput(inode);
}
out_unlock:
mutex_unlock(&sysfs_mutex);
return ret;
}
/*
* process, that is going to call fix_nodes/do_balance must hold only
* one path. If it holds 2 or more, it can get into endless waiting in
* get_empty_nodes or its clones
*/
static int do_reiserfs_rename (struct reiserfs_transaction_handle *th, struct inode * old_dir, struct dentry *old_dentry,
struct inode * new_dir, struct dentry *new_dentry)
{
int retval;
struct path old_entry_path, new_entry_path, dot_dot_entry_path;
struct reiserfs_dir_entry old_de, new_de, dot_dot_de;
struct inode * old_inode, * new_inode;
int new_entry_added = 0;
init_path (&old_entry_path);
init_path (&new_entry_path);
init_path (&dot_dot_entry_path);
goto start_up;
try_again:
current->policy |= SCHED_YIELD;
schedule();
start_up:
old_inode = new_inode = NULL;
dot_dot_de.de_bh = 0;
new_de.de_bh = 0;
/*
* look for the old name in old directory
*/
retval = -ENOENT;
old_de.de_gen_number_bit_string = 0;
if (reiserfs_find_entry (old_dir, old_dentry->d_name.name, old_dentry->d_name.len, &old_entry_path, &old_de) == POSITION_NOT_FOUND)
goto end_rename;
pathrelse (&old_entry_path);
old_inode = old_dentry->d_inode;
retval = -EPERM;
if ((old_dir->i_mode & S_ISVTX) &&
current->fsuid != old_inode->i_uid &&
current->fsuid != old_dir->i_uid && !fsuser())
goto end_rename;
new_inode = new_dentry->d_inode;
/* look for the new entry in target directory */
new_de.de_gen_number_bit_string = 0;
if (reiserfs_find_entry (new_dir, new_dentry->d_name.name, new_dentry->d_name.len, &new_entry_path, &new_de) == POSITION_FOUND) {
if (!new_inode) {
printk ("do_reiserfs_rename: new entry found, inode == 0 though\n");
}
/* this entry already exists, we can just set key of object */
new_entry_added = 1;
} else {
#ifdef REISERFS_CHECK
if (new_entry_added) {
if (new_de.de_namelen != new_dentry->d_name.len || memcmp (new_de.de_name, new_dentry->d_name.name, new_de.de_namelen) ||
de_visible (new_de.de_deh))
reiserfs_panic (old_dir->i_sb, "vs-7045: reiserfs_rename: suspicious entry found");
}
#endif /* REISERFS_CHECK */
}
pathrelse (&new_entry_path);
if (new_inode == old_inode) {
retval = 0;
goto end_rename;
}
if (new_inode && S_ISDIR(new_inode->i_mode)) {
/* new name exists and points to directory */
retval = -EISDIR;
if (!S_ISDIR(old_inode->i_mode))
goto end_rename;
retval = -EINVAL;
if (is_subdir (new_dentry, old_dentry))
goto end_rename;
retval = -ENOTEMPTY;
if (!reiserfs_empty_dir (new_inode))
goto end_rename;
retval = -EBUSY;
if (new_inode->i_count > 1)
goto end_rename;
}
retval = -EPERM;
if (new_inode && (new_dir->i_mode & S_ISVTX) &&
current->fsuid != new_inode->i_uid &&
current->fsuid != new_dir->i_uid && !fsuser())
goto end_rename;
if (S_ISDIR(old_inode->i_mode)) {
/* old name points to directory */
retval = -ENOTDIR;
if (new_inode && !S_ISDIR(new_inode->i_mode))
goto end_rename;
retval = -EINVAL;
if (is_subdir(new_dentry, old_dentry))
goto end_rename;
retval = -EIO;
/* directory is renamed, its parent directory will be changed, so find ".." entry */
dot_dot_de.de_gen_number_bit_string = 0;
if (reiserfs_find_entry (old_inode, "..", 2, &dot_dot_entry_path, &dot_dot_de) == POSITION_NOT_FOUND)
goto end_rename;
if (dot_dot_de.de_objectid != old_dir->i_ino)
goto end_rename;
pathrelse (&dot_dot_entry_path);
retval = -EMLINK;
if (!new_inode && new_dir->i_nlink >= REISERFS_LINK_MAX)
goto end_rename;
}
if (new_entry_added == 0) {
/* add new entry if we did not do it, but do not mark it as visible */
retval = reiserfs_add_entry (th, new_dir, new_dentry->d_name.name, new_dentry->d_name.len, INODE_PKEY (old_inode), &new_de, 0);
if (retval)
goto end_rename;
if_in_ram_update_sd (th, new_dir);
new_entry_added = 1;
goto try_again;
}
/*
* look for old name, new name and ".." when renaming directories again
*/
if (reiserfs_find_entry (old_dir, old_dentry->d_name.name, old_dentry->d_name.len, &old_entry_path, &old_de) == POSITION_NOT_FOUND)
reiserfs_panic (old_dir->i_sb, "vs-7050: reiserfs_rename: old name not found");
if (reiserfs_find_entry (new_dir, new_dentry->d_name.name, new_dentry->d_name.len, &new_entry_path, &new_de) == POSITION_NOT_FOUND)
reiserfs_panic (old_dir->i_sb, "vs-7055: reiserfs_rename: new name not found");
if (S_ISDIR(old_inode->i_mode) && reiserfs_find_entry (old_inode, "..", 2, &dot_dot_entry_path, &dot_dot_de) == POSITION_NOT_FOUND)
reiserfs_panic (old_dir->i_sb, "vs-7060: reiserfs_rename: \"..\" name not found");
/* sanity checking before doing the rename - avoid races */
if (!entry_points_to_object (new_dentry->d_name.name, new_dentry->d_name.len, &new_de, new_inode))
goto try_again;
if (!entry_points_to_object (old_dentry->d_name.name, old_dentry->d_name.len, &old_de, old_inode))
/* go to re-looking for old entry */
goto try_again;
if (S_ISDIR(old_inode->i_mode) && !entry_points_to_object ("..", 2, &dot_dot_de, old_dir))
/* go to re-looking for ".." entry of renamed directory */
goto try_again;
/* ok, all the changes can be done in one fell swoop when we have
claimed all the buffers needed.*/
/* make old name hidden */
mark_de_hidden (old_de.de_deh);
journal_mark_dirty(th, old_dir->i_sb, old_de.de_bh) ;
/* make new name visible and set key of old object (if entry
existed, it is already visible, if not, key is correct already) */
mark_de_visible (new_de.de_deh);
new_de.de_deh->deh_dir_id = INODE_PKEY (old_inode)->k_dir_id;
new_de.de_deh->deh_objectid = INODE_PKEY (old_inode)->k_objectid;
journal_mark_dirty(th, old_dir->i_sb, new_de.de_bh) ;
old_dir->i_ctime = old_dir->i_mtime = CURRENT_TIME;
if_in_ram_update_sd (th, old_dir);
new_dir->i_ctime = new_dir->i_mtime = CURRENT_TIME;
if_in_ram_update_sd (th, new_dir);
if (new_inode) {
new_inode->i_nlink--;
new_inode->i_ctime = CURRENT_TIME;
if_in_ram_update_sd (th, new_inode);
}
if (dot_dot_de.de_bh) {
set_ino_in_dir_entry (&dot_dot_de, INODE_PKEY (new_dir));
journal_mark_dirty(th, old_dir->i_sb, dot_dot_de.de_bh) ;
old_dir->i_nlink--;
if_in_ram_update_sd (th, old_dir);
if (new_inode) {
new_inode->i_nlink--;
if_in_ram_update_sd (th, new_inode);
} else {
new_dir->i_nlink++;
if_in_ram_update_sd (th, new_dir);
}
}
/* ok, renaming done */
decrement_counters_in_path (&new_entry_path);
decrement_counters_in_path (&dot_dot_entry_path);
/* remove old name (it is hidden now) */
if (reiserfs_cut_from_item (th, old_dir, old_dir->i_sb, &old_entry_path, &(old_de.de_entry_num),
&(old_de.de_entry_key), 0, PRESERVE_RENAMING) == 0)
printk ("reiserfs_rename: could not remove old name\n");
else {
old_dir->i_size -= DEH_SIZE + old_de.de_entrylen;
old_dir->i_blocks = old_dir->i_size / 512 + ((old_dir->i_size % 512) ? 1 : 0);
if_in_ram_update_sd (th, old_dir);
}
/* Update the dcache */
d_move(old_dentry, new_dentry);
retval = 0;
end_rename:
pathrelse (&old_entry_path);
return retval;
}
int aufs_rename(struct inode *_src_dir, struct dentry *_src_dentry,
struct inode *_dst_dir, struct dentry *_dst_dentry)
{
int err, flags;
/* reduce stack space */
struct au_ren_args *a;
AuDbg("%.*s, %.*s\n", AuDLNPair(_src_dentry), AuDLNPair(_dst_dentry));
IMustLock(_src_dir);
IMustLock(_dst_dir);
err = -ENOMEM;
BUILD_BUG_ON(sizeof(*a) > PAGE_SIZE);
a = kzalloc(sizeof(*a), GFP_NOFS);
if (unlikely(!a))
goto out;
a->src_dir = _src_dir;
a->src_dentry = _src_dentry;
a->src_inode = a->src_dentry->d_inode;
a->src_parent = a->src_dentry->d_parent; /* dir inode is locked */
a->dst_dir = _dst_dir;
a->dst_dentry = _dst_dentry;
a->dst_inode = a->dst_dentry->d_inode;
a->dst_parent = a->dst_dentry->d_parent; /* dir inode is locked */
if (a->dst_inode) {
IMustLock(a->dst_inode);
au_igrab(a->dst_inode);
}
err = -ENOTDIR;
flags = AuLock_FLUSH | AuLock_NOPLM | AuLock_GEN;
if (S_ISDIR(a->src_inode->i_mode)) {
au_fset_ren(a->flags, ISDIR);
if (unlikely(a->dst_inode && !S_ISDIR(a->dst_inode->i_mode)))
goto out_free;
err = aufs_read_and_write_lock2(a->dst_dentry, a->src_dentry,
AuLock_DIR | flags);
} else
err = aufs_read_and_write_lock2(a->dst_dentry, a->src_dentry,
flags);
if (unlikely(err))
goto out_free;
err = au_d_hashed_positive(a->src_dentry);
if (unlikely(err))
goto out_unlock;
err = -ENOENT;
if (a->dst_inode) {
/*
* If it is a dir, VFS unhash dst_dentry before this
* function. It means we cannot rely upon d_unhashed().
*/
if (unlikely(!a->dst_inode->i_nlink))
goto out_unlock;
if (!S_ISDIR(a->dst_inode->i_mode)) {
err = au_d_hashed_positive(a->dst_dentry);
if (unlikely(err))
goto out_unlock;
} else if (unlikely(IS_DEADDIR(a->dst_inode)))
goto out_unlock;
} else if (unlikely(d_unhashed(a->dst_dentry)))
goto out_unlock;
au_fset_ren(a->flags, ISSAMEDIR); /* temporary */
di_write_lock_parent(a->dst_parent);
/* which branch we process */
err = au_ren_wbr(a);
if (unlikely(err < 0))
goto out_parent;
a->br = au_sbr(a->dst_dentry->d_sb, a->btgt);
a->h_path.mnt = a->br->br_mnt;
/* are they available to be renamed */
err = au_ren_may_dir(a);
if (unlikely(err))
goto out_children;
/* prepare the writable parent dir on the same branch */
if (a->dst_bstart == a->btgt) {
au_fset_ren(a->flags, WHDST);
} else {
err = au_cpup_dirs(a->dst_dentry, a->btgt);
if (unlikely(err))
goto out_children;
}
if (a->src_dir != a->dst_dir) {
/*
* this temporary unlock is safe,
* because both dir->i_mutex are locked.
*/
di_write_unlock(a->dst_parent);
di_write_lock_parent(a->src_parent);
err = au_wr_dir_need_wh(a->src_dentry,
au_ftest_ren(a->flags, ISDIR),
&a->btgt);
di_write_unlock(a->src_parent);
di_write_lock2_parent(a->src_parent, a->dst_parent, /*isdir*/1);
au_fclr_ren(a->flags, ISSAMEDIR);
} else
err = au_wr_dir_need_wh(a->src_dentry,
au_ftest_ren(a->flags, ISDIR),
&a->btgt);
if (unlikely(err < 0))
goto out_children;
if (err)
au_fset_ren(a->flags, WHSRC);
/* lock them all */
err = au_ren_lock(a);
if (unlikely(err))
goto out_children;
if (!au_opt_test(au_mntflags(a->dst_dir->i_sb), UDBA_NONE))
err = au_may_ren(a);
else if (unlikely(a->dst_dentry->d_name.len > AUFS_MAX_NAMELEN))
err = -ENAMETOOLONG;
if (unlikely(err))
goto out_hdir;
/* store timestamps to be revertible */
au_ren_dt(a);
/* here we go */
err = do_rename(a);
if (unlikely(err))
goto out_dt;
/* update dir attributes */
au_ren_refresh_dir(a);
/* dput/iput all lower dentries */
au_ren_refresh(a);
goto out_hdir; /* success */
out_dt:
au_ren_rev_dt(err, a);
out_hdir:
au_ren_unlock(a);
out_children:
au_nhash_wh_free(&a->whlist);
if (err && a->dst_inode && a->dst_bstart != a->btgt) {
AuDbg("bstart %d, btgt %d\n", a->dst_bstart, a->btgt);
au_set_h_dptr(a->dst_dentry, a->btgt, NULL);
au_set_dbstart(a->dst_dentry, a->dst_bstart);
}
out_parent:
if (!err)
d_move(a->src_dentry, a->dst_dentry);
else {
au_update_dbstart(a->dst_dentry);
if (!a->dst_inode)
d_drop(a->dst_dentry);
}
if (au_ftest_ren(a->flags, ISSAMEDIR))
di_write_unlock(a->dst_parent);
else
di_write_unlock2(a->src_parent, a->dst_parent);
out_unlock:
aufs_read_and_write_unlock2(a->dst_dentry, a->src_dentry);
out_free:
iput(a->dst_inode);
if (a->thargs)
au_whtmp_rmdir_free(a->thargs);
kfree(a);
out:
AuTraceErr(err);
return err;
}
/**
* nfs_sillyrename - Perform a silly-rename of a dentry
* @dir: inode of directory that contains dentry
* @dentry: dentry to be sillyrenamed
*
* NFSv2/3 is stateless and the server doesn't know when the client is
* holding a file open. To prevent application problems when a file is
* unlinked while it's still open, the client performs a "silly-rename".
* That is, it renames the file to a hidden file in the same directory,
* and only performs the unlink once the last reference to it is put.
*
* The final cleanup is done during dentry_iput.
*
* (Note: NFSv4 is stateful, and has opens, so in theory an NFSv4 server
* could take responsibility for keeping open files referenced. The server
* would also need to ensure that opened-but-deleted files were kept over
* reboots. However, we may not assume a server does so. (RFC 5661
* does provide an OPEN4_RESULT_PRESERVE_UNLINKED flag that a server can
* use to advertise that it does this; some day we may take advantage of
* it.))
*/
int
nfs_sillyrename(struct inode *dir, struct dentry *dentry)
{
static unsigned int sillycounter;
const int fileidsize = sizeof(NFS_FILEID(dentry->d_inode))*2;
const int countersize = sizeof(sillycounter)*2;
const int slen = sizeof(".nfs")+fileidsize+countersize-1;
char silly[slen+1];
struct dentry *sdentry;
struct rpc_task *task;
int error = -EIO;
dfprintk(VFS, "NFS: silly-rename(%s/%s, ct=%d)\n",
dentry->d_parent->d_name.name, dentry->d_name.name,
dentry->d_count);
nfs_inc_stats(dir, NFSIOS_SILLYRENAME);
/*
* We don't allow a dentry to be silly-renamed twice.
*/
error = -EBUSY;
if (dentry->d_flags & DCACHE_NFSFS_RENAMED)
goto out;
sprintf(silly, ".nfs%*.*Lx",
fileidsize, fileidsize,
(unsigned long long)NFS_FILEID(dentry->d_inode));
/* Return delegation in anticipation of the rename */
NFS_PROTO(dentry->d_inode)->return_delegation(dentry->d_inode);
sdentry = NULL;
do {
char *suffix = silly + slen - countersize;
dput(sdentry);
sillycounter++;
sprintf(suffix, "%*.*x", countersize, countersize, sillycounter);
dfprintk(VFS, "NFS: trying to rename %s to %s\n",
dentry->d_name.name, silly);
sdentry = lookup_one_len(silly, dentry->d_parent, slen);
/*
* N.B. Better to return EBUSY here ... it could be
* dangerous to delete the file while it's in use.
*/
if (IS_ERR(sdentry))
goto out;
} while (sdentry->d_inode != NULL); /* need negative lookup */
/* queue unlink first. Can't do this from rpc_release as it
* has to allocate memory
*/
error = nfs_async_unlink(dir, dentry);
if (error)
goto out_dput;
/* populate unlinkdata with the right dname */
error = nfs_copy_dname(sdentry,
(struct nfs_unlinkdata *)dentry->d_fsdata);
if (error) {
nfs_cancel_async_unlink(dentry);
goto out_dput;
}
/* run the rename task, undo unlink if it fails */
task = nfs_async_rename(dir, dir, dentry, sdentry);
if (IS_ERR(task)) {
error = -EBUSY;
nfs_cancel_async_unlink(dentry);
goto out_dput;
}
/* wait for the RPC task to complete, unless a SIGKILL intervenes */
error = rpc_wait_for_completion_task(task);
if (error == 0)
error = task->tk_status;
switch (error) {
case 0:
/* The rename succeeded */
nfs_set_verifier(dentry, nfs_save_change_attribute(dir));
d_move(dentry, sdentry);
break;
case -ERESTARTSYS:
/* The result of the rename is unknown. Play it safe by
* forcing a new lookup */
d_drop(dentry);
d_drop(sdentry);
}
rpc_put_task(task);
out_dput:
dput(sdentry);
out:
return error;
}
/**
* ntfs_lookup - find the inode represented by a dentry in a directory inode
* @dir_ino: directory inode in which to look for the inode
* @dent: dentry representing the inode to look for
* @nd: lookup nameidata
*
* In short, ntfs_lookup() looks for the inode represented by the dentry @dent
* in the directory inode @dir_ino and if found attaches the inode to the
* dentry @dent.
*
* In more detail, the dentry @dent specifies which inode to look for by
* supplying the name of the inode in @dent->d_name.name. ntfs_lookup()
* converts the name to Unicode and walks the contents of the directory inode
* @dir_ino looking for the converted Unicode name. If the name is found in the
* directory, the corresponding inode is loaded by calling ntfs_iget() on its
* inode number and the inode is associated with the dentry @dent via a call to
* d_splice_alias().
*
* If the name is not found in the directory, a NULL inode is inserted into the
* dentry @dent via a call to d_add(). The dentry is then termed a negative
* dentry.
*
* Only if an actual error occurs, do we return an error via ERR_PTR().
*
* In order to handle the case insensitivity issues of NTFS with regards to the
* dcache and the dcache requiring only one dentry per directory, we deal with
* dentry aliases that only differ in case in ->ntfs_lookup() while maintaining
* a case sensitive dcache. This means that we get the full benefit of dcache
* speed when the file/directory is looked up with the same case as returned by
* ->ntfs_readdir() but that a lookup for any other case (or for the short file
* name) will not find anything in dcache and will enter ->ntfs_lookup()
* instead, where we search the directory for a fully matching file name
* (including case) and if that is not found, we search for a file name that
* matches with different case and if that has non-POSIX semantics we return
* that. We actually do only one search (case sensitive) and keep tabs on
* whether we have found a case insensitive match in the process.
*
* To simplify matters for us, we do not treat the short vs long filenames as
* two hard links but instead if the lookup matches a short filename, we
* return the dentry for the corresponding long filename instead.
*
* There are three cases we need to distinguish here:
*
* 1) @dent perfectly matches (i.e. including case) a directory entry with a
* file name in the WIN32 or POSIX namespaces. In this case
* ntfs_lookup_inode_by_name() will return with name set to NULL and we
* just d_splice_alias() @dent.
* 2) @dent matches (not including case) a directory entry with a file name in
* the WIN32 namespace. In this case ntfs_lookup_inode_by_name() will return
* with name set to point to a kmalloc()ed ntfs_name structure containing
* the properly cased little endian Unicode name. We convert the name to the
* current NLS code page, search if a dentry with this name already exists
* and if so return that instead of @dent. At this point things are
* complicated by the possibility of 'disconnected' dentries due to NFS
* which we deal with appropriately (see the code comments). The VFS will
* then destroy the old @dent and use the one we returned. If a dentry is
* not found, we allocate a new one, d_splice_alias() it, and return it as
* above.
* 3) @dent matches either perfectly or not (i.e. we don't care about case) a
* directory entry with a file name in the DOS namespace. In this case
* ntfs_lookup_inode_by_name() will return with name set to point to a
* kmalloc()ed ntfs_name structure containing the mft reference (cpu endian)
* of the inode. We use the mft reference to read the inode and to find the
* file name in the WIN32 namespace corresponding to the matched short file
* name. We then convert the name to the current NLS code page, and proceed
* searching for a dentry with this name, etc, as in case 2), above.
*
* Locking: Caller must hold i_mutex on the directory.
*/
static struct dentry *ntfs_lookup(struct inode *dir_ino, struct dentry *dent,
struct nameidata *nd)
{
ntfs_volume *vol = NTFS_SB(dir_ino->i_sb);
struct inode *dent_inode;
ntfschar *uname;
ntfs_name *name = NULL;
MFT_REF mref;
unsigned long dent_ino;
int uname_len;
ntfs_debug("Looking up %s in directory inode 0x%lx.",
dent->d_name.name, dir_ino->i_ino);
/* Convert the name of the dentry to Unicode. */
uname_len = ntfs_nlstoucs(vol, dent->d_name.name, dent->d_name.len,
&uname);
if (uname_len < 0) {
if (uname_len != -ENAMETOOLONG)
ntfs_error(vol->sb, "Failed to convert name to "
"Unicode.");
return ERR_PTR(uname_len);
}
mref = ntfs_lookup_inode_by_name(NTFS_I(dir_ino), uname, uname_len,
&name);
kmem_cache_free(ntfs_name_cache, uname);
if (!IS_ERR_MREF(mref)) {
dent_ino = MREF(mref);
ntfs_debug("Found inode 0x%lx. Calling ntfs_iget.", dent_ino);
dent_inode = ntfs_iget(vol->sb, dent_ino);
if (likely(!IS_ERR(dent_inode))) {
/* Consistency check. */
if (is_bad_inode(dent_inode) || MSEQNO(mref) ==
NTFS_I(dent_inode)->seq_no ||
dent_ino == FILE_MFT) {
/* Perfect WIN32/POSIX match. -- Case 1. */
if (!name) {
ntfs_debug("Done. (Case 1.)");
return d_splice_alias(dent_inode, dent);
}
/*
* We are too indented. Handle imperfect
* matches and short file names further below.
*/
goto handle_name;
}
ntfs_error(vol->sb, "Found stale reference to inode "
"0x%lx (reference sequence number = "
"0x%x, inode sequence number = 0x%x), "
"returning -EIO. Run chkdsk.",
dent_ino, MSEQNO(mref),
NTFS_I(dent_inode)->seq_no);
iput(dent_inode);
dent_inode = ERR_PTR(-EIO);
} else
ntfs_error(vol->sb, "ntfs_iget(0x%lx) failed with "
"error code %li.", dent_ino,
PTR_ERR(dent_inode));
kfree(name);
/* Return the error code. */
return (struct dentry *)dent_inode;
}
/* It is guaranteed that @name is no longer allocated at this point. */
if (MREF_ERR(mref) == -ENOENT) {
ntfs_debug("Entry was not found, adding negative dentry.");
/* The dcache will handle negative entries. */
d_add(dent, NULL);
ntfs_debug("Done.");
return NULL;
}
ntfs_error(vol->sb, "ntfs_lookup_ino_by_name() failed with error "
"code %i.", -MREF_ERR(mref));
return ERR_PTR(MREF_ERR(mref));
// TODO: Consider moving this lot to a separate function! (AIA)
handle_name:
{
struct dentry *real_dent, *new_dent;
MFT_RECORD *m;
ntfs_attr_search_ctx *ctx;
ntfs_inode *ni = NTFS_I(dent_inode);
int err;
struct qstr nls_name;
nls_name.name = NULL;
if (name->type != FILE_NAME_DOS) { /* Case 2. */
ntfs_debug("Case 2.");
nls_name.len = (unsigned)ntfs_ucstonls(vol,
(ntfschar*)&name->name, name->len,
(unsigned char**)&nls_name.name, 0);
kfree(name);
} else /* if (name->type == FILE_NAME_DOS) */ { /* Case 3. */
FILE_NAME_ATTR *fn;
ntfs_debug("Case 3.");
kfree(name);
/* Find the WIN32 name corresponding to the matched DOS name. */
ni = NTFS_I(dent_inode);
m = map_mft_record(ni);
if (IS_ERR(m)) {
err = PTR_ERR(m);
m = NULL;
ctx = NULL;
goto err_out;
}
ctx = ntfs_attr_get_search_ctx(ni, m);
if (unlikely(!ctx)) {
err = -ENOMEM;
goto err_out;
}
do {
ATTR_RECORD *a;
u32 val_len;
err = ntfs_attr_lookup(AT_FILE_NAME, NULL, 0, 0, 0,
NULL, 0, ctx);
if (unlikely(err)) {
ntfs_error(vol->sb, "Inode corrupt: No WIN32 "
"namespace counterpart to DOS "
"file name. Run chkdsk.");
if (err == -ENOENT)
err = -EIO;
goto err_out;
}
/* Consistency checks. */
a = ctx->attr;
if (a->non_resident || a->flags)
goto eio_err_out;
val_len = le32_to_cpu(a->data.resident.value_length);
if (le16_to_cpu(a->data.resident.value_offset) +
val_len > le32_to_cpu(a->length))
goto eio_err_out;
fn = (FILE_NAME_ATTR*)((u8*)ctx->attr + le16_to_cpu(
ctx->attr->data.resident.value_offset));
if ((u32)(fn->file_name_length * sizeof(ntfschar) +
sizeof(FILE_NAME_ATTR)) > val_len)
goto eio_err_out;
} while (fn->file_name_type != FILE_NAME_WIN32);
/* Convert the found WIN32 name to current NLS code page. */
nls_name.len = (unsigned)ntfs_ucstonls(vol,
(ntfschar*)&fn->file_name, fn->file_name_length,
(unsigned char**)&nls_name.name, 0);
ntfs_attr_put_search_ctx(ctx);
unmap_mft_record(ni);
}
m = NULL;
ctx = NULL;
/* Check if a conversion error occurred. */
if ((signed)nls_name.len < 0) {
err = (signed)nls_name.len;
goto err_out;
}
nls_name.hash = full_name_hash(nls_name.name, nls_name.len);
/*
* Note: No need for dent->d_lock lock as i_mutex is held on the
* parent inode.
*/
/* Does a dentry matching the nls_name exist already? */
real_dent = d_lookup(dent->d_parent, &nls_name);
/* If not, create it now. */
if (!real_dent) {
real_dent = d_alloc(dent->d_parent, &nls_name);
kfree(nls_name.name);
if (!real_dent) {
err = -ENOMEM;
goto err_out;
}
new_dent = d_splice_alias(dent_inode, real_dent);
if (new_dent)
dput(real_dent);
else
new_dent = real_dent;
ntfs_debug("Done. (Created new dentry.)");
return new_dent;
}
kfree(nls_name.name);
/* Matching dentry exists, check if it is negative. */
if (real_dent->d_inode) {
if (unlikely(real_dent->d_inode != dent_inode)) {
/* This can happen because bad inodes are unhashed. */
BUG_ON(!is_bad_inode(dent_inode));
BUG_ON(!is_bad_inode(real_dent->d_inode));
}
/*
* Already have the inode and the dentry attached, decrement
* the reference count to balance the ntfs_iget() we did
* earlier on. We found the dentry using d_lookup() so it
* cannot be disconnected and thus we do not need to worry
* about any NFS/disconnectedness issues here.
*/
iput(dent_inode);
ntfs_debug("Done. (Already had inode and dentry.)");
return real_dent;
}
/*
* Negative dentry: instantiate it unless the inode is a directory and
* has a 'disconnected' dentry (i.e. IS_ROOT and DCACHE_DISCONNECTED),
* in which case d_move() that in place of the found dentry.
*/
if (!S_ISDIR(dent_inode->i_mode)) {
/* Not a directory; everything is easy. */
d_instantiate(real_dent, dent_inode);
ntfs_debug("Done. (Already had negative file dentry.)");
return real_dent;
}
spin_lock(&dcache_lock);
if (list_empty(&dent_inode->i_dentry)) {
/*
* Directory without a 'disconnected' dentry; we need to do
* d_instantiate() by hand because it takes dcache_lock which
* we already hold.
*/
list_add(&real_dent->d_alias, &dent_inode->i_dentry);
real_dent->d_inode = dent_inode;
spin_unlock(&dcache_lock);
security_d_instantiate(real_dent, dent_inode);
ntfs_debug("Done. (Already had negative directory dentry.)");
return real_dent;
}
/*
* Directory with a 'disconnected' dentry; get a reference to the
* 'disconnected' dentry.
*/
new_dent = list_entry(dent_inode->i_dentry.next, struct dentry,
d_alias);
dget_locked(new_dent);
spin_unlock(&dcache_lock);
/* Do security vodoo. */
security_d_instantiate(real_dent, dent_inode);
/* Move new_dent in place of real_dent. */
d_move(new_dent, real_dent);
/* Balance the ntfs_iget() we did above. */
iput(dent_inode);
/* Throw away real_dent. */
dput(real_dent);
/* Use new_dent as the actual dentry. */
ntfs_debug("Done. (Already had negative, disconnected directory "
"dentry.)");
return new_dent;
eio_err_out:
ntfs_error(vol->sb, "Illegal file name attribute. Run chkdsk.");
err = -EIO;
err_out:
if (ctx)
ntfs_attr_put_search_ctx(ctx);
if (m)
unmap_mft_record(ni);
iput(dent_inode);
ntfs_error(vol->sb, "Failed, returning error code %i.", err);
return ERR_PTR(err);
}
}
int sysfs_move_dir(struct kobject *kobj, struct kobject *new_parent_kobj)
{
struct sysfs_dirent *sd = kobj->sd;
struct sysfs_dirent *new_parent_sd;
struct dentry *old_parent, *new_parent = NULL;
struct dentry *old_dentry = NULL, *new_dentry = NULL;
int error;
BUG_ON(!sd->s_parent);
new_parent_sd = new_parent_kobj->sd ? new_parent_kobj->sd : &sysfs_root;
/* get dentries */
old_dentry = sysfs_get_dentry(sd);
if (IS_ERR(old_dentry)) {
error = PTR_ERR(old_dentry);
goto out_dput;
}
old_parent = sd->s_parent->s_dentry;
new_parent = sysfs_get_dentry(new_parent_sd);
if (IS_ERR(new_parent)) {
error = PTR_ERR(new_parent);
goto out_dput;
}
if (old_parent->d_inode == new_parent->d_inode) {
error = 0;
goto out_dput; /* nothing to move */
}
again:
mutex_lock(&old_parent->d_inode->i_mutex);
if (!mutex_trylock(&new_parent->d_inode->i_mutex)) {
mutex_unlock(&old_parent->d_inode->i_mutex);
goto again;
}
new_dentry = lookup_one_len(kobj->name, new_parent, strlen(kobj->name));
if (IS_ERR(new_dentry)) {
error = PTR_ERR(new_dentry);
goto out_unlock;
} else
error = 0;
d_add(new_dentry, NULL);
d_move(sd->s_dentry, new_dentry);
dput(new_dentry);
/* Remove from old parent's list and insert into new parent's list. */
mutex_lock(&sysfs_mutex);
sysfs_unlink_sibling(sd);
sysfs_get(new_parent_sd);
sysfs_put(sd->s_parent);
sd->s_parent = new_parent_sd;
sysfs_link_sibling(sd);
mutex_unlock(&sysfs_mutex);
out_unlock:
mutex_unlock(&new_parent->d_inode->i_mutex);
mutex_unlock(&old_parent->d_inode->i_mutex);
out_dput:
dput(new_parent);
dput(old_dentry);
dput(new_dentry);
return error;
}
int
nfs_sillyrename(struct inode *dir, struct dentry *dentry)
{
static unsigned int sillycounter;
const int fileidsize = sizeof(NFS_FILEID(dentry->d_inode))*2;
const int countersize = sizeof(sillycounter)*2;
const int slen = sizeof(".nfs")+fileidsize+countersize-1;
char silly[slen+1];
struct dentry *sdentry;
struct rpc_task *task;
int error = -EIO;
dfprintk(VFS, "NFS: silly-rename(%s/%s, ct=%d)\n",
dentry->d_parent->d_name.name, dentry->d_name.name,
dentry->d_count);
nfs_inc_stats(dir, NFSIOS_SILLYRENAME);
error = -EBUSY;
if (dentry->d_flags & DCACHE_NFSFS_RENAMED)
goto out;
sprintf(silly, ".nfs%*.*Lx",
fileidsize, fileidsize,
(unsigned long long)NFS_FILEID(dentry->d_inode));
nfs_inode_return_delegation(dentry->d_inode);
sdentry = NULL;
do {
char *suffix = silly + slen - countersize;
dput(sdentry);
sillycounter++;
sprintf(suffix, "%*.*x", countersize, countersize, sillycounter);
dfprintk(VFS, "NFS: trying to rename %s to %s\n",
dentry->d_name.name, silly);
sdentry = lookup_one_len(silly, dentry->d_parent, slen);
if (IS_ERR(sdentry))
goto out;
} while (sdentry->d_inode != NULL);
error = nfs_async_unlink(dir, dentry);
if (error)
goto out_dput;
error = nfs_copy_dname(sdentry,
(struct nfs_unlinkdata *)dentry->d_fsdata);
if (error) {
nfs_cancel_async_unlink(dentry);
goto out_dput;
}
task = nfs_async_rename(dir, dir, dentry, sdentry);
if (IS_ERR(task)) {
error = -EBUSY;
nfs_cancel_async_unlink(dentry);
goto out_dput;
}
error = rpc_wait_for_completion_task(task);
if (error == 0)
error = task->tk_status;
switch (error) {
case 0:
/* The rename succeeded */
nfs_set_verifier(dentry, nfs_save_change_attribute(dir));
d_move(dentry, sdentry);
break;
case -ERESTARTSYS:
/* The result of the rename is unknown. Play it safe by
* forcing a new lookup */
d_drop(dentry);
d_drop(sdentry);
}
rpc_put_task(task);
out_dput:
dput(sdentry);
out:
return error;
}
/*
* rename uses retrying to avoid race-conditions: at least they should be minimal.
* it tries to allocate all the blocks, then sanity-checks, and if the sanity-
* checks fail, it tries to restart itself again. Very practical - no changes
* are done until we know everything works ok.. and then all the changes can be
* done in one fell swoop when we have claimed all the buffers needed.
*
* Anybody can rename anything with this: the permission checks are left to the
* higher-level routines.
*/
static int do_minix_rename(struct inode * old_dir, struct dentry *old_dentry,
struct inode * new_dir, struct dentry *new_dentry)
{
struct inode * old_inode, * new_inode;
struct buffer_head * old_bh, * new_bh, * dir_bh;
struct minix_dir_entry * old_de, * new_de;
struct minix_sb_info * info;
int retval;
info = &old_dir->i_sb->u.minix_sb;
goto start_up;
try_again:
brelse(old_bh);
brelse(new_bh);
brelse(dir_bh);
current->counter = 0;
schedule();
start_up:
old_inode = new_inode = NULL;
old_bh = new_bh = dir_bh = NULL;
old_bh = minix_find_entry(old_dir, old_dentry->d_name.name,
old_dentry->d_name.len, &old_de);
retval = -ENOENT;
if (!old_bh)
goto end_rename;
old_inode = old_dentry->d_inode;
retval = -EPERM;
new_inode = new_dentry->d_inode;
new_bh = minix_find_entry(new_dir, new_dentry->d_name.name,
new_dentry->d_name.len, &new_de);
if (new_bh) {
if (!new_inode) {
brelse(new_bh);
new_bh = NULL;
}
}
if (new_inode == old_inode) {
retval = 0;
goto end_rename;
}
if (S_ISDIR(old_inode->i_mode)) {
retval = -EINVAL;
if (is_subdir(new_dentry, old_dentry))
goto end_rename;
if (new_inode) {
/* Prune any children before testing for busy */
if (new_dentry->d_count > 1)
shrink_dcache_parent(new_dentry);
retval = -EBUSY;
if (new_dentry->d_count > 1)
retval = -ENOTEMPTY;
if (!empty_dir(new_inode))
goto end_rename;
retval = -EBUSY;
}
retval = -EIO;
dir_bh = minix_bread(old_inode,0,0);
if (!dir_bh)
goto end_rename;
if (PARENT_INO(dir_bh->b_data) != old_dir->i_ino)
goto end_rename;
retval = -EMLINK;
if (!new_inode && new_dir->i_nlink >= info->s_link_max)
goto end_rename;
}
if (!new_bh) {
retval = minix_add_entry(new_dir,
new_dentry->d_name.name,
new_dentry->d_name.len,
&new_bh, &new_de);
if (retval)
goto end_rename;
}
/* sanity checking before doing the rename - avoid races */
if (new_inode && (new_de->inode != new_inode->i_ino))
goto try_again;
if (new_de->inode && !new_inode)
goto try_again;
if (old_de->inode != old_inode->i_ino)
goto try_again;
/* ok, that's it */
old_de->inode = 0;
new_de->inode = old_inode->i_ino;
old_dir->i_ctime = old_dir->i_mtime = CURRENT_TIME;
mark_inode_dirty(old_dir);
old_dir->i_version = ++event;
new_dir->i_ctime = new_dir->i_mtime = CURRENT_TIME;
mark_inode_dirty(new_dir);
new_dir->i_version = ++event;
if (new_inode) {
new_inode->i_nlink--;
new_inode->i_ctime = CURRENT_TIME;
mark_inode_dirty(new_inode);
}
mark_buffer_dirty(old_bh, 1);
mark_buffer_dirty(new_bh, 1);
if (dir_bh) {
PARENT_INO(dir_bh->b_data) = new_dir->i_ino;
mark_buffer_dirty(dir_bh, 1);
old_dir->i_nlink--;
mark_inode_dirty(old_dir);
if (new_inode) {
new_inode->i_nlink--;
mark_inode_dirty(new_inode);
} else {
new_dir->i_nlink++;
mark_inode_dirty(new_dir);
}
}
/* Update the dcache */
d_move(old_dentry, new_dentry);
retval = 0;
end_rename:
brelse(dir_bh);
brelse(old_bh);
brelse(new_bh);
return retval;
}
DENT_T *
vnode_iop_lookup(
INODE_T *dir,
struct dentry *dent,
struct nameidata *nd
)
{
char *name;
mdki_boolean_t rele = FALSE;
int err;
VNODE_T *dvp;
VNODE_T *rt_vnode; /* returned vnode */
INODE_T *rt_inode = NULL; /* returned inode ptr */
DENT_T * real_dentry;
DENT_T *found_dentry = dent;
VATTR_T *vap;
struct lookup_ctx ctx;
CALL_DATA_T cd;
ASSERT_I_SEM_MINE(dir);
/* We can find our parent entry via the dentry provided to us. */
ASSERT(dent->d_parent->d_inode == dir);
if (dent->d_name.len > NAME_MAX)
return ERR_PTR(-ENAMETOOLONG);
name = /* drop the const */(char *) dent->d_name.name;
mdki_linux_init_call_data(&cd);
/* We pass along the dentry, as well as the parent inode so that
* mvop_linux_lookup_* has everything it needs, even if it is passed in
* the realvp, and it gets back a negative dentry.
*/
dvp = ITOV(dir);
ctx.dentrypp = &found_dentry;
ctx.flags = LOOKUP_CTX_VALID;
err = VOP_LOOKUP(dvp, name, &rt_vnode, (struct pathname *)NULL,
VNODE_LF_LOOKUP, NULL, &cd, &ctx);
err = mdki_errno_unix_to_linux(err);
if (!err) {
ASSERT(rt_vnode != NULL);
if (MDKI_INOISCLRVN(VTOI(rt_vnode))) {
/* unwrap to the real object */
ASSERT(CVN_TO_DENT(rt_vnode));
rt_inode = CVN_TO_INO(rt_vnode);
if (MDKI_INOISMVFS(rt_inode)) {
VN_HOLD(ITOV(rt_inode));
VN_RELE(rt_vnode);
rt_vnode = ITOV(rt_inode);
} else {
igrab(rt_inode);
VN_RELE(rt_vnode);
rt_vnode = NULL;
}
} else
rt_inode = VTOI(rt_vnode);
}
if (!err && (found_dentry != dent)) {
mdki_linux_destroy_call_data(&cd);
/* The hold was granted in makeloopnode() in the 'nocover' case. */
if (rt_vnode != NULL)
VN_RELE(rt_vnode);
else
iput(rt_inode);
/*
* found_dentry is the real socket/block/char device node's dentry.
* See mvop_linux_lookup_component().
*
* For sockets, we use a dentry in our tree (we fill in the
* provided dentry "dent") linked to the inode of the real
* object. This lets file name operations work in our
* namespace, and lets socket connections all work (as they're
* keyed off of the inode address) from inside to outside &
* v.v.
*
* We also do this for VCHR, VBLK devices, and it seems to work OK
* (e.g. make a node the same as /dev/tty, you can write to it)
*/
switch (found_dentry->d_inode->i_mode & S_IFMT) {
case S_IFSOCK:
case S_IFCHR:
case S_IFBLK:
ASSERT(dent->d_inode == NULL);
MDKI_SET_DOPS(dent, &vnode_shadow_dentry_ops);
igrab(found_dentry->d_inode);
VNODE_D_ADD(dent, found_dentry->d_inode);
VNODE_DPUT(found_dentry);
found_dentry = NULL; /* tell caller to use original dentry */
break;
default:
/* use returned dentry */
break;
}
return(found_dentry);
}
/* We need to pass back dentry ops even for negative dentries, I think.
* Shadow inodes will have been taken care of in lookup_component.
*/
if (dent->d_op != &vnode_shadow_dentry_ops) {
if (dent->d_parent->d_op == &vnode_setview_dentry_ops)
MDKI_SET_DOPS(dent, &vnode_setview_dentry_ops);
else
MDKI_SET_DOPS(dent, &vnode_dentry_ops);
}
vap = VATTR_ALLOC();
if (vap == NULL) {
err = -ENOMEM;
goto alloc_err;
}
if (!err && MDKI_INOISMVFS(rt_inode)) {
/* fetch attributes & place in inode */
VATTR_SET_MASK(vap, AT_ALL);
err = VOP_GETATTR(rt_vnode, vap, GETATTR_FLAG_UPDATE_ATTRS, &cd);
err = mdki_errno_unix_to_linux(err);
if (err == -EOPNOTSUPP) /* ignore it */
err = 0;
else if (err)
rele = TRUE;
else if ((rt_vnode->v_flag & VLOOPROOT) != 0 &&
rt_inode == vnlayer_get_urdir_inode())
{
/* return the real root */
VN_RELE(rt_vnode);
VATTR_FREE(vap);
mdki_linux_destroy_call_data(&cd);
return VNODE_DGET(vnlayer_get_root_dentry());
}
else if (vnlayer_looproot_vp != NULL &&
rt_vnode == vnlayer_looproot_vp &&
(real_dentry = MVOP_DENT(rt_inode,
&vnode_dentry_ops)) != NULL)
{
/* return the real /view */
VN_RELE(rt_vnode);
VATTR_FREE(vap);
mdki_linux_destroy_call_data(&cd);
return real_dentry;
}
}
VATTR_FREE(vap);
alloc_err:
mdki_linux_destroy_call_data(&cd);
/* It's an mnode-based object, set up a dentry for it */
/* We don't return ENOENT. For Linux, the negative dentry is enough */
switch (err) {
case -ENOENT:
err = 0;
ASSERT(rt_inode == NULL);
VNODE_D_ADD(dent, rt_inode);
break;
case 0:
/* We will consume the count on rt_inode as a reference for dent */
/*
* For VOB vnodes, we maintain two separate dentry trees for
* the vnodes. One tree is for setview-mode names (process
* sets to a view context, then looks directly at the VOB
* mountpoint without any cover vnodes in the path). The
* other tree is for view-extended naming into a VOB, with
* dentries starting at the view tag and covering non-VOB
* objects until crossing a mount point into a VOB.
*
* Mostly the system doesn't care, as long as it goes down the
* tree from parent to child, since it will be traversing only one
* of the dentry trees. But when the cache misses, the system calls
* this lookup method and wants to get a dentry in return.
* There are standard interfaces ( d_splice_alias() in 2.6)
* which can find a good dentry referencing the inode returned
* by the file system's lookup method, but these methods don't
* work right when we have VOB directory vnodes with both setview
* and view-extended dentries. We implement our own function
* [vnlayer_inode2dentry_internal()] which knows the
* distinctions and the rules for determining that an existing
* attached dentry is valid for the lookup request.
*
* We have our own d_compare() function which forces all VOB
* lookups to come to the inode lookup method (this function),
* and then we get to choose the right dentry to return. We
* have our own lookup cache inside MVFS so we don't care that
* the dentry cache is always missing on our names.
*
* If we have to make a new dentry, we may need to merge it
* with an NFS-created temporary dentry using d_move()
* (d_splice_alias() would do this for us, but we can't use it
* for reasons listed above).
*/
/*
* We want to find the "right" dentry (if there is one), so
* look for one that has a d_parent with the same dentry ops
* (indicating it's in the same dentry tree).
*/
if (S_ISDIR(rt_inode->i_mode)) {
/*
* It has been empirically shown that we have to check the
* parent of the dentry. If the parent has been checked out
* it is possible for the cache lookup to return an inode
* from the tree below the old parent directory. If this
* happens on a rename, the system will panic because the
* Linux rename code checks the parent of the returned
* dentry to see that it matches what it has for a parent.
*/
found_dentry = vnlayer_inode2dentry_internal(rt_inode,
dent->d_parent, NULL,
dent->d_op);
} else {
/*
* For non-directories, we also need to consider the
* parent & the requested name so that
* vnlayer_inode2dentry_internal() finds the right dentry.
* (There may be multiple hard links; we want the one in
* the same directory with the same name)
*/
found_dentry = vnlayer_inode2dentry_internal(rt_inode,
dent->d_parent,
&dent->d_name,
dent->d_op);
}
if (found_dentry != NULL) {
ASSERT(found_dentry->d_inode == rt_inode);
/*
* If the existing one is a disconnected dentry, we need
* to move the old one to the new one (just like
* d_splice_alias) to get the proper name/parent attached
* in the dcache.
*/
if ((found_dentry->d_flags & DCACHE_DISCONNECTED) != 0) {
#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,7)
ASSERT((dent->d_flags & DCACHE_UNHASHED) != 0);
#else
ASSERT((dent->d_vfs_flags & DCACHE_UNHASHED) != 0);
#endif
d_rehash(dent);
d_move(found_dentry, dent);
}
/* Release our count. found_dentry also references inode. */
iput(rt_inode);
return found_dentry;
}
/*
* Nothing suitable, wire it up to the proposed dentry.
*/
VNODE_D_ADD(dent, rt_inode);
break;
default:
/* some other error case */
if (rele)
VN_RELE(rt_vnode);
break;
}
if (err)
return ERR_PTR(err);
else
return NULL;
}
/*
* rename uses retrying to avoid race-conditions: at least they should be minimal.
* it tries to allocate all the blocks, then sanity-checks, and if the sanity-
* checks fail, it tries to restart itself again. Very practical - no changes
* are done until we know everything works ok.. and then all the changes can be
* done in one fell swoop when we have claimed all the buffers needed.
*
* Anybody can rename anything with this: the permission checks are left to the
* higher-level routines.
*/
static int do_sysv_rename(struct inode * old_dir, struct dentry * old_dentry,
struct inode * new_dir, struct dentry * new_dentry)
{
struct inode * old_inode, * new_inode;
struct buffer_head * old_bh, * new_bh, * dir_bh;
struct sysv_dir_entry * old_de, * new_de;
int retval;
goto start_up;
try_again:
brelse(old_bh);
brelse(new_bh);
brelse(dir_bh);
current->counter = 0;
schedule();
start_up:
old_inode = new_inode = NULL;
old_bh = new_bh = dir_bh = NULL;
old_bh = sysv_find_entry(old_dir, old_dentry->d_name.name,
old_dentry->d_name.len, &old_de);
retval = -ENOENT;
if (!old_bh)
goto end_rename;
old_inode = old_dentry->d_inode; /* don't cross mnt-points */
retval = -EPERM;
new_inode = new_dentry->d_inode;
new_bh = sysv_find_entry(new_dir, new_dentry->d_name.name,
new_dentry->d_name.len, &new_de);
if (new_bh) {
if (!new_inode) {
brelse(new_bh);
new_bh = NULL;
}
}
if (new_inode == old_inode) {
retval = 0;
goto end_rename;
}
if (S_ISDIR(old_inode->i_mode)) {
retval = -EINVAL;
if (is_subdir(new_dentry, old_dentry))
goto end_rename;
if (new_inode) {
if (new_dentry->d_count > 1)
shrink_dcache_parent(new_dentry);
retval = -EBUSY;
if (new_dentry->d_count > 1)
goto end_rename;
retval = -ENOTEMPTY;
if (!empty_dir(new_inode))
goto end_rename;
}
retval = -EIO;
dir_bh = sysv_file_bread(old_inode, 0, 0);
if (!dir_bh)
goto end_rename;
if (PARENT_INO(dir_bh->b_data) != old_dir->i_ino)
goto end_rename;
retval = -EMLINK;
if (!new_inode && new_dir->i_nlink >= new_dir->i_sb->sv_link_max)
goto end_rename;
}
if (!new_bh) {
retval = sysv_add_entry(new_dir, new_dentry->d_name.name,
new_dentry->d_name.len, &new_bh, &new_de);
if (retval)
goto end_rename;
}
/* sanity checking before doing the rename - avoid races */
if (new_inode && (new_de->inode != new_inode->i_ino))
goto try_again;
if (new_de->inode && !new_inode)
goto try_again;
if (old_de->inode != old_inode->i_ino)
goto try_again;
/* ok, that's it */
old_de->inode = 0;
new_de->inode = old_inode->i_ino;
old_dir->i_ctime = old_dir->i_mtime = CURRENT_TIME;
mark_inode_dirty(old_dir);
new_dir->i_ctime = new_dir->i_mtime = CURRENT_TIME;
mark_inode_dirty(new_dir);
if (new_inode) {
new_inode->i_nlink--;
new_inode->i_ctime = CURRENT_TIME;
mark_inode_dirty(new_inode);
}
mark_buffer_dirty(old_bh, 1);
mark_buffer_dirty(new_bh, 1);
if (dir_bh) {
PARENT_INO(dir_bh->b_data) = new_dir->i_ino;
mark_buffer_dirty(dir_bh, 1);
old_dir->i_nlink--;
mark_inode_dirty(old_dir);
if (new_inode) {
new_inode->i_nlink--;
mark_inode_dirty(new_inode);
} else {
new_dir->i_nlink++;
mark_inode_dirty(new_dir);
}
}
d_move(old_dentry, new_dentry);
retval = 0;
end_rename:
brelse(dir_bh);
brelse(old_bh);
brelse(new_bh);
return retval;
}
static int pvfs2_rename(
struct inode *old_dir,
struct dentry *old_dentry,
struct inode *new_dir,
struct dentry *new_dentry)
{
int ret = -EINVAL, are_directories = 0;
pvfs2_inode_t *pvfs2_old_parent_inode = PVFS2_I(old_dir);
pvfs2_inode_t *pvfs2_new_parent_inode = PVFS2_I(new_dir);
pvfs2_kernel_op_t *new_op = NULL;
struct super_block *sb = NULL;
gossip_debug(GOSSIP_NAME_DEBUG, "pvfs2_rename: called (%s/%s => %s/%s) ct=%d\n",
old_dentry->d_parent->d_name.name, old_dentry->d_name.name,
new_dentry->d_parent->d_name.name, new_dentry->d_name.name,
atomic_read(&new_dentry->d_count));
are_directories = S_ISDIR(old_dentry->d_inode->i_mode);
#if 0
/* NOTE: we have no good way to keep nlink consistent for directories
* across clients; keep constant at 1 -Phil
*/
if (are_directories && (new_dir->i_nlink >= PVFS2_LINK_MAX))
{
gossip_err("pvfs2_rename: directory %s surpassed "
"PVFS2_LINK_MAX\n",
new_dentry->d_name.name);
return -EMLINK;
}
#endif
new_op = op_alloc(PVFS2_VFS_OP_RENAME);
if (!new_op)
{
return ret;
}
/*
if no handle/fs_id is available in the parent,
use the root handle/fs_id as specified by the
inode's corresponding superblock
*/
if (pvfs2_old_parent_inode &&
pvfs2_old_parent_inode->refn.handle != PVFS_HANDLE_NULL &&
pvfs2_old_parent_inode->refn.fs_id != PVFS_FS_ID_NULL)
{
new_op->upcall.req.rename.old_parent_refn =
pvfs2_old_parent_inode->refn;
}
else
{
sb = old_dir->i_sb;
new_op->upcall.req.rename.old_parent_refn.handle =
PVFS2_SB(sb)->root_handle;
new_op->upcall.req.rename.old_parent_refn.fs_id =
PVFS2_SB(sb)->fs_id;
}
/* do the same for the new parent */
if (pvfs2_new_parent_inode &&
pvfs2_new_parent_inode->refn.handle != PVFS_HANDLE_NULL &&
pvfs2_new_parent_inode->refn.fs_id != PVFS_FS_ID_NULL)
{
new_op->upcall.req.rename.new_parent_refn =
pvfs2_new_parent_inode->refn;
}
else
{
sb = new_dir->i_sb;
new_op->upcall.req.rename.new_parent_refn.handle =
PVFS2_SB(sb)->root_handle;
new_op->upcall.req.rename.new_parent_refn.fs_id =
PVFS2_SB(sb)->fs_id;
}
strncpy(new_op->upcall.req.rename.d_old_name,
old_dentry->d_name.name, PVFS2_NAME_LEN);
strncpy(new_op->upcall.req.rename.d_new_name,
new_dentry->d_name.name, PVFS2_NAME_LEN);
ret = service_operation(
new_op, "pvfs2_rename",
get_interruptible_flag(old_dentry->d_inode));
gossip_debug(GOSSIP_NAME_DEBUG, "pvfs2_rename: got downcall status %d\n", ret);
if (new_dentry->d_inode)
{
new_dentry->d_inode->i_ctime = CURRENT_TIME;
#if 0
/* NOTE: we have no good way to keep nlink consistent for directories
* across clients; keep constant at 1 -Phil
*/
if (are_directories)
{
new_dentry->d_inode->i_nlink--;
}
#endif
}
#if 0
/* NOTE: we have no good way to keep nlink consistent for directories
* across clients; keep constant at 1 -Phil
*/
else if (are_directories)
{
new_dir->i_nlink++;
old_dir->i_nlink--;
}
#endif
op_release(new_op);
#ifdef PVFS2_LINUX_KERNEL_2_4
if (ret == 0)
{
d_move(old_dentry, new_dentry);
}
#endif
return ret;
}
static int nfs_sillyrename(struct inode *dir, struct dentry *dentry)
{
static unsigned int sillycounter;
const int i_inosize = sizeof(dir->i_ino)*2;
const int countersize = sizeof(sillycounter)*2;
const int slen = strlen(".nfs") + i_inosize + countersize;
char silly[slen+1];
struct qstr qsilly;
struct dentry *sdentry;
int error = -EIO;
dfprintk(VFS, "NFS: silly-rename(%s/%s, ct=%d)\n",
dentry->d_parent->d_name.name, dentry->d_name.name,
atomic_read(&dentry->d_count));
if (atomic_read(&dentry->d_count) == 1)
goto out; /* No need to silly rename. */
#ifdef NFS_PARANOIA
if (!dentry->d_inode)
printk("NFS: silly-renaming %s/%s, negative dentry??\n",
dentry->d_parent->d_name.name, dentry->d_name.name);
#endif
/*
* We don't allow a dentry to be silly-renamed twice.
*/
error = -EBUSY;
if (dentry->d_flags & DCACHE_NFSFS_RENAMED)
goto out;
sprintf(silly, ".nfs%*.*lx",
i_inosize, i_inosize, dentry->d_inode->i_ino);
sdentry = NULL;
do {
char *suffix = silly + slen - countersize;
dput(sdentry);
sillycounter++;
sprintf(suffix, "%*.*x", countersize, countersize, sillycounter);
dfprintk(VFS, "trying to rename %s to %s\n",
dentry->d_name.name, silly);
sdentry = lookup_one_len(silly, dentry->d_parent, slen);
/*
* N.B. Better to return EBUSY here ... it could be
* dangerous to delete the file while it's in use.
*/
if (IS_ERR(sdentry))
goto out;
} while(sdentry->d_inode != NULL); /* need negative lookup */
nfs_zap_caches(dir);
qsilly.name = silly;
qsilly.len = strlen(silly);
error = NFS_PROTO(dir)->rename(dir, &dentry->d_name, dir, &qsilly);
if (!error) {
nfs_renew_times(dentry);
d_move(dentry, sdentry);
error = nfs_async_unlink(dentry);
/* If we return 0 we don't unlink */
}
dput(sdentry);
out:
return error;
}
/*
* RENAME
* FIXME: Some nfsds, like the Linux user space nfsd, may generate a
* different file handle for the same inode after a rename (e.g. when
* moving to a different directory). A fail-safe method to do so would
* be to look up old_dir/old_name, create a link to new_dir/new_name and
* rename the old file using the sillyrename stuff. This way, the original
* file in old_dir will go away when the last process iput()s the inode.
*
* FIXED.
*
* It actually works quite well. One needs to have the possibility for
* at least one ".nfs..." file in each directory the file ever gets
* moved or linked to which happens automagically with the new
* implementation that only depends on the dcache stuff instead of
* using the inode layer
*
* Unfortunately, things are a little more complicated than indicated
* above. For a cross-directory move, we want to make sure we can get
* rid of the old inode after the operation. This means there must be
* no pending writes (if it's a file), and the use count must be 1.
* If these conditions are met, we can drop the dentries before doing
* the rename.
*/
static int nfs_rename(struct inode *old_dir, struct dentry *old_dentry,
struct inode *new_dir, struct dentry *new_dentry)
{
struct inode *old_inode = old_dentry->d_inode;
struct inode *new_inode = new_dentry->d_inode;
struct dentry *dentry = NULL, *rehash = NULL;
int error = -EBUSY;
/*
* To prevent any new references to the target during the rename,
* we unhash the dentry and free the inode in advance.
*/
if (!d_unhashed(new_dentry)) {
d_drop(new_dentry);
rehash = new_dentry;
}
dfprintk(VFS, "NFS: rename(%s/%s -> %s/%s, ct=%d)\n",
old_dentry->d_parent->d_name.name, old_dentry->d_name.name,
new_dentry->d_parent->d_name.name, new_dentry->d_name.name,
atomic_read(&new_dentry->d_count));
/*
* First check whether the target is busy ... we can't
* safely do _any_ rename if the target is in use.
*
* For files, make a copy of the dentry and then do a
* silly-rename. If the silly-rename succeeds, the
* copied dentry is hashed and becomes the new target.
*/
if (!new_inode)
goto go_ahead;
if (S_ISDIR(new_inode->i_mode))
goto out;
else if (atomic_read(&new_dentry->d_count) > 1) {
int err;
/* copy the target dentry's name */
dentry = d_alloc(new_dentry->d_parent,
&new_dentry->d_name);
if (!dentry)
goto out;
/* silly-rename the existing target ... */
err = nfs_sillyrename(new_dir, new_dentry);
if (!err) {
new_dentry = rehash = dentry;
new_inode = NULL;
/* instantiate the replacement target */
d_instantiate(new_dentry, NULL);
}
/* dentry still busy? */
if (atomic_read(&new_dentry->d_count) > 1) {
#ifdef NFS_PARANOIA
printk("nfs_rename: target %s/%s busy, d_count=%d\n",
new_dentry->d_parent->d_name.name,
new_dentry->d_name.name,
atomic_read(&new_dentry->d_count));
#endif
goto out;
}
}
go_ahead:
/*
* ... prune child dentries and writebacks if needed.
*/
if (atomic_read(&old_dentry->d_count) > 1) {
nfs_wb_all(old_inode);
shrink_dcache_parent(old_dentry);
}
if (new_inode)
d_delete(new_dentry);
nfs_zap_caches(new_dir);
nfs_zap_caches(old_dir);
error = NFS_PROTO(old_dir)->rename(old_dir, &old_dentry->d_name,
new_dir, &new_dentry->d_name);
out:
if (rehash)
d_rehash(rehash);
if (!error && !S_ISDIR(old_inode->i_mode))
d_move(old_dentry, new_dentry);
/* new dentry created? */
if (dentry)
dput(dentry);
return error;
}
int sysfs_rename_dir(struct kobject *kobj, struct sysfs_dirent *new_parent_sd,
const char *new_name)
{
struct sysfs_dirent *sd = kobj->sd;
struct dentry *new_parent = NULL;
struct dentry *old_dentry = NULL, *new_dentry = NULL;
const char *dup_name = NULL;
int error;
/* get dentries */
old_dentry = sysfs_get_dentry(sd);
if (IS_ERR(old_dentry)) {
error = PTR_ERR(old_dentry);
goto out_dput;
}
new_parent = sysfs_get_dentry(new_parent_sd);
if (IS_ERR(new_parent)) {
error = PTR_ERR(new_parent);
goto out_dput;
}
/* lock new_parent and get dentry for new name */
mutex_lock(&new_parent->d_inode->i_mutex);
new_dentry = lookup_one_len(new_name, new_parent, strlen(new_name));
if (IS_ERR(new_dentry)) {
error = PTR_ERR(new_dentry);
goto out_unlock;
}
/* By allowing two different directories with the same
* d_parent we allow this routine to move between different
* shadows of the same directory
*/
error = -EINVAL;
if (old_dentry->d_parent->d_inode != new_parent->d_inode ||
new_dentry->d_parent->d_inode != new_parent->d_inode ||
old_dentry == new_dentry)
goto out_unlock;
error = -EEXIST;
if (new_dentry->d_inode)
goto out_unlock;
/* rename kobject and sysfs_dirent */
error = -ENOMEM;
new_name = dup_name = kstrdup(new_name, GFP_KERNEL);
if (!new_name)
goto out_drop;
error = kobject_set_name(kobj, "%s", new_name);
if (error)
goto out_drop;
mutex_lock(&sysfs_mutex);
dup_name = sd->s_name;
sd->s_name = new_name;
/* move under the new parent */
d_add(new_dentry, NULL);
d_move(sd->s_dentry, new_dentry);
sysfs_unlink_sibling(sd);
sysfs_get(new_parent_sd);
sysfs_put(sd->s_parent);
sd->s_parent = new_parent_sd;
sysfs_link_sibling(sd);
mutex_unlock(&sysfs_mutex);
error = 0;
goto out_unlock;
out_drop:
d_drop(new_dentry);
out_unlock:
mutex_unlock(&new_parent->d_inode->i_mutex);
out_dput:
kfree(dup_name);
dput(new_parent);
dput(old_dentry);
dput(new_dentry);
return error;
}
