/**
* ntfs_file_fsync - sync a file to disk
* @filp: file to be synced
* @dentry: dentry describing the file to sync
* @datasync: if non-zero only flush user data and not metadata
*
* Data integrity sync of a file to disk. Used for fsync, fdatasync, and msync
* system calls. This function is inspired by fs/buffer.c::file_fsync().
*
* If @datasync is false, write the mft record and all associated extent mft
* records as well as the $DATA attribute and then sync the block device.
*
* If @datasync is true and the attribute is non-resident, we skip the writing
* of the mft record and all associated extent mft records (this might still
* happen due to the write_inode_now() call).
*
* Also, if @datasync is true, we do not wait on the inode to be written out
* but we always wait on the page cache pages to be written out.
*
* Note: In the past @filp could be NULL so we ignore it as we don't need it
* anyway.
*
* Locking: Caller must hold i_sem on the inode.
*
* TODO: We should probably also write all attribute/index inodes associated
* with this inode but since we have no simple way of getting to them we ignore
* this problem for now.
*/
static int ntfs_file_fsync(struct file *filp, struct dentry *dentry,
int datasync)
{
struct inode *vi = dentry->d_inode;
int err, ret = 0;
ntfs_debug("Entering for inode 0x%lx.", vi->i_ino);
BUG_ON(S_ISDIR(vi->i_mode));
if (!datasync || !NInoNonResident(NTFS_I(vi)))
ret = ntfs_write_inode(vi, 1);
write_inode_now(vi, !datasync);
/*
* NOTE: If we were to use mapping->private_list (see ext2 and
* fs/buffer.c) for dirty blocks then we could optimize the below to be
* sync_mapping_buffers(vi->i_mapping).
*/
err = sync_blockdev(vi->i_sb->s_bdev);
if (unlikely(err && !ret))
ret = err;
if (likely(!ret))
ntfs_debug("Done.");
else
ntfs_warning(vi->i_sb, "Failed to f%ssync inode 0x%lx. Error "
"%u.", datasync ? "data" : "", vi->i_ino, -ret);
return ret;
}
static struct dentry *ntfs_get_parent(struct dentry *child_dent)
{
struct inode *vi = child_dent->d_inode;
ntfs_inode *ni = NTFS_I(vi);
MFT_RECORD *mrec;
ntfs_attr_search_ctx *ctx;
ATTR_RECORD *attr;
FILE_NAME_ATTR *fn;
unsigned long parent_ino;
int err;
ntfs_debug("Entering for inode 0x%lx.", vi->i_ino);
mrec = map_mft_record(ni);
if (IS_ERR(mrec))
return (struct dentry *)mrec;
ctx = ntfs_attr_get_search_ctx(ni, mrec);
if (unlikely(!ctx)) {
unmap_mft_record(ni);
return ERR_PTR(-ENOMEM);
}
try_next:
err = ntfs_attr_lookup(AT_FILE_NAME, NULL, 0, CASE_SENSITIVE, 0, NULL,
0, ctx);
if (unlikely(err)) {
ntfs_attr_put_search_ctx(ctx);
unmap_mft_record(ni);
if (err == -ENOENT)
ntfs_error(vi->i_sb, "Inode 0x%lx does not have a "
"file name attribute. Run chkdsk.",
vi->i_ino);
return ERR_PTR(err);
}
attr = ctx->attr;
if (unlikely(attr->non_resident))
goto try_next;
fn = (FILE_NAME_ATTR *)((u8 *)attr +
le16_to_cpu(attr->data.resident.value_offset));
if (unlikely((u8 *)fn + le32_to_cpu(attr->data.resident.value_length) >
(u8*)attr + le32_to_cpu(attr->length)))
goto try_next;
parent_ino = MREF_LE(fn->parent_directory);
ntfs_attr_put_search_ctx(ctx);
unmap_mft_record(ni);
return d_obtain_alias(ntfs_iget(vi->i_sb, parent_ino));
}
/**
* ntfs_file_fsync - sync a file to disk
* @filp: file to be synced
* @dentry: dentry describing the file to sync
* @datasync: if non-zero only flush user data and not metadata
*
* Data integrity sync of a file to disk. Used for fsync, fdatasync, and msync
* system calls. This function is inspired by fs/buffer.c::file_fsync().
*
* If @datasync is false, write the mft record and all associated extent mft
* records as well as the $DATA attribute and then sync the block device.
*
* If @datasync is true and the attribute is non-resident, we skip the writing
* of the mft record and all associated extent mft records (this might still
* happen due to the write_inode_now() call).
*
* Also, if @datasync is true, we do not wait on the inode to be written out
* but we always wait on the page cache pages to be written out.
*
* Note: In the past @filp could be NULL so we ignore it as we don't need it
* anyway.
*
* Locking: Caller must hold i_sem on the inode.
*
* TODO: We should probably also write all attribute/index inodes associated
* with this inode but since we have no simple way of getting to them we ignore
* this problem for now.
*/
static int ntfs_file_fsync(struct file *filp, struct dentry *dentry,
int datasync)
{
struct inode *vi = dentry->d_inode;
int err, ret = 0;
ntfs_debug("Entering for inode 0x%lx.", vi->i_ino);
BUG_ON(S_ISDIR(vi->i_mode));
if (!datasync || !NInoNonResident(NTFS_I(vi)))
ret = ntfs_write_inode(vi, 1);
write_inode_now(vi, !datasync);
err = sync_blockdev(vi->i_sb->s_bdev);
if (unlikely(err && !ret))
ret = err;
if (likely(!ret))
ntfs_debug("Done.");
else
ntfs_warning(vi->i_sb, "Failed to f%ssync inode 0x%lx. Error "
"%u.", datasync ? "data" : "", vi->i_ino, -ret);
return ret;
}
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);
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))) {
if (is_bad_inode(dent_inode) || MSEQNO(mref) ==
NTFS_I(dent_inode)->seq_no ||
dent_ino == FILE_MFT) {
if (!name) {
ntfs_debug("Done. (Case 1.)");
return d_splice_alias(dent_inode, dent);
}
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 (struct dentry *)dent_inode;
}
if (MREF_ERR(mref) == -ENOENT) {
ntfs_debug("Entry was not found, adding negative dentry.");
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));
handle_name:
{
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) {
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 {
FILE_NAME_ATTR *fn;
ntfs_debug("Case 3.");
kfree(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;
}
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);
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;
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);
dent = d_add_ci(dent, dent_inode, &nls_name);
kfree(nls_name.name);
return 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);
}
}
/**
* ntfs_get_parent - find the dentry of the parent of a given directory dentry
* @child_dent: dentry of the directory whose parent directory to find
*
* Find the dentry for the parent directory of the directory specified by the
* dentry @child_dent. This function is called from
* fs/exportfs/expfs.c::find_exported_dentry() which in turn is called from the
* default ->decode_fh() which is export_decode_fh() in the same file.
*
* The code is based on the ext3 ->get_parent() implementation found in
* fs/ext3/namei.c::ext3_get_parent().
*
* Note: ntfs_get_parent() is called with @child_dent->d_inode->i_mutex down.
*
* Return the dentry of the parent directory on success or the error code on
* error (IS_ERR() is true).
*/
static struct dentry *ntfs_get_parent(struct dentry *child_dent)
{
struct inode *vi = child_dent->d_inode;
ntfs_inode *ni = NTFS_I(vi);
MFT_RECORD *mrec;
ntfs_attr_search_ctx *ctx;
ATTR_RECORD *attr;
FILE_NAME_ATTR *fn;
struct inode *parent_vi;
struct dentry *parent_dent;
unsigned long parent_ino;
int err;
ntfs_debug("Entering for inode 0x%lx.", vi->i_ino);
/* Get the mft record of the inode belonging to the child dentry. */
mrec = map_mft_record(ni);
if (IS_ERR(mrec))
return (struct dentry *)mrec;
/* Find the first file name attribute in the mft record. */
ctx = ntfs_attr_get_search_ctx(ni, mrec);
if (unlikely(!ctx)) {
unmap_mft_record(ni);
return ERR_PTR(-ENOMEM);
}
try_next:
err = ntfs_attr_lookup(AT_FILE_NAME, NULL, 0, CASE_SENSITIVE, 0, NULL,
0, ctx);
if (unlikely(err)) {
ntfs_attr_put_search_ctx(ctx);
unmap_mft_record(ni);
if (err == -ENOENT)
ntfs_error(vi->i_sb, "Inode 0x%lx does not have a "
"file name attribute. Run chkdsk.",
vi->i_ino);
return ERR_PTR(err);
}
attr = ctx->attr;
if (unlikely(attr->non_resident))
goto try_next;
fn = (FILE_NAME_ATTR *)((u8 *)attr +
le16_to_cpu(attr->data.resident.value_offset));
if (unlikely((u8 *)fn + le32_to_cpu(attr->data.resident.value_length) >
(u8*)attr + le32_to_cpu(attr->length)))
goto try_next;
/* Get the inode number of the parent directory. */
parent_ino = MREF_LE(fn->parent_directory);
/* Release the search context and the mft record of the child. */
ntfs_attr_put_search_ctx(ctx);
unmap_mft_record(ni);
/* Get the inode of the parent directory. */
parent_vi = ntfs_iget(vi->i_sb, parent_ino);
if (IS_ERR(parent_vi) || unlikely(is_bad_inode(parent_vi))) {
if (!IS_ERR(parent_vi))
iput(parent_vi);
ntfs_error(vi->i_sb, "Failed to get parent directory inode "
"0x%lx of child inode 0x%lx.", parent_ino,
vi->i_ino);
return ERR_PTR(-EACCES);
}
/* Finally get a dentry for the parent directory and return it. */
parent_dent = d_alloc_anon(parent_vi);
if (unlikely(!parent_dent)) {
iput(parent_vi);
return ERR_PTR(-ENOMEM);
}
ntfs_debug("Done for inode 0x%lx.", vi->i_ino);
return parent_dent;
}
/**
* 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);
}
}
/**
* 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
* @flags: lookup flags
*
* 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,
unsigned int flags)
{
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 %pd in directory inode 0x%lx.",
dent, 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 ERR_CAST(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:
{
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(dent, nls_name.name, nls_name.len);
dent = d_add_ci(dent, dent_inode, &nls_name);
kfree(nls_name.name);
return 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);
}
}
/**
* ntfs_mark_quotas_out_of_date - mark the quotas out of date on an ntfs volume
* @vol: ntfs volume on which to mark the quotas out of date
*
* Mark the quotas out of date on the ntfs volume @vol and return TRUE on
* success and FALSE on error.
*/
BOOL ntfs_mark_quotas_out_of_date(ntfs_volume *vol)
{
ntfs_index_context *ictx;
QUOTA_CONTROL_ENTRY *qce;
const le32 qid = QUOTA_DEFAULTS_ID;
int err;
ntfs_debug("Entering.");
if (NVolQuotaOutOfDate(vol))
goto done;
if (!vol->quota_ino || !vol->quota_q_ino) {
ntfs_error(vol->sb, "Quota inodes are not open.");
return FALSE;
}
down(&vol->quota_q_ino->i_sem);
ictx = ntfs_index_ctx_get(NTFS_I(vol->quota_q_ino));
if (!ictx) {
ntfs_error(vol->sb, "Failed to get index context.");
goto err_out;
}
err = ntfs_index_lookup(&qid, sizeof(qid), ictx);
if (err) {
if (err == -ENOENT)
ntfs_error(vol->sb, "Quota defaults entry is not "
"present.");
else
ntfs_error(vol->sb, "Lookup of quota defaults entry "
"failed.");
goto err_out;
}
if (ictx->data_len < offsetof(QUOTA_CONTROL_ENTRY, sid)) {
ntfs_error(vol->sb, "Quota defaults entry size is invalid. "
"Run chkdsk.");
goto err_out;
}
qce = (QUOTA_CONTROL_ENTRY*)ictx->data;
if (le32_to_cpu(qce->version) != QUOTA_VERSION) {
ntfs_error(vol->sb, "Quota defaults entry version 0x%x is not "
"supported.", le32_to_cpu(qce->version));
goto err_out;
}
ntfs_debug("Quota defaults flags = 0x%x.", le32_to_cpu(qce->flags));
/* If quotas are already marked out of date, no need to do anything. */
if (qce->flags & QUOTA_FLAG_OUT_OF_DATE)
goto set_done;
/*
* If quota tracking is neither requested, nor enabled and there are no
* pending deletes, no need to mark the quotas out of date.
*/
if (!(qce->flags & (QUOTA_FLAG_TRACKING_ENABLED |
QUOTA_FLAG_TRACKING_REQUESTED |
QUOTA_FLAG_PENDING_DELETES)))
goto set_done;
/*
* Set the QUOTA_FLAG_OUT_OF_DATE bit thus marking quotas out of date.
* This is verified on WinXP to be sufficient to cause windows to
* rescan the volume on boot and update all quota entries.
*/
qce->flags |= QUOTA_FLAG_OUT_OF_DATE;
/* Ensure the modified flags are written to disk. */
ntfs_index_entry_flush_dcache_page(ictx);
ntfs_index_entry_mark_dirty(ictx);
set_done:
ntfs_index_ctx_put(ictx);
up(&vol->quota_q_ino->i_sem);
/*
* We set the flag so we do not try to mark the quotas out of date
* again on remount.
*/
NVolSetQuotaOutOfDate(vol);
done:
ntfs_debug("Done.");
return TRUE;
err_out:
if (ictx)
ntfs_index_ctx_put(ictx);
up(&vol->quota_q_ino->i_sem);
return FALSE;
}
/**
* 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_add().
*
* If the name is not found in the directory, a NULL inode is inserted into the
* dentry @dent. 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 maintining
* 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_add() @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. 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_add() 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.
*/
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;
uchar_t *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) {
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 (MSEQNO(mref) == NTFS_I(dent_inode)->seq_no ||
dent_ino == FILE_MFT) {
/* Perfect WIN32/POSIX match. -- Case 1. */
if (!name) {
d_add(dent, dent_inode);
ntfs_debug("Done.");
return NULL;
}
/*
* 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));
if (name)
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;
MFT_RECORD *m;
attr_search_context *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. */
nls_name.len = (unsigned)ntfs_ucstonls(vol,
(uchar_t*)&name->name, name->len,
(unsigned char**)&nls_name.name,
name->len * 3 + 1);
kfree(name);
} else /* if (name->type == FILE_NAME_DOS) */ { /* Case 3. */
FILE_NAME_ATTR *fn;
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 = get_attr_search_ctx(ni, m);
if (!ctx) {
err = -ENOMEM;
goto err_out;
}
do {
ATTR_RECORD *a;
u32 val_len;
if (!lookup_attr(AT_FILE_NAME, NULL, 0, 0, 0, NULL, 0,
ctx)) {
ntfs_error(vol->sb, "Inode corrupt: No WIN32 "
"namespace counterpart to DOS "
"file name. Run chkdsk.");
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(uchar_t) +
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,
(uchar_t*)&fn->file_name, fn->file_name_length,
(unsigned char**)&nls_name.name,
fn->file_name_length * 3 + 1);
put_attr_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_sem 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;
}
d_add(real_dent, dent_inode);
return real_dent;
}
kfree(nls_name.name);
/* Matching dentry exists, check if it is negative. */
if (real_dent->d_inode) {
BUG_ON(real_dent->d_inode != dent_inode);
/*
* Already have the inode and the dentry attached, decrement
* the reference count to balance the ntfs_iget() we did
* earlier on.
*/
iput(dent_inode);
return real_dent;
}
/* Negative dentry: instantiate it. */
d_instantiate(real_dent, dent_inode);
return real_dent;
eio_err_out:
ntfs_error(vol->sb, "Illegal file name attribute. Run chkdsk.");
err = -EIO;
err_out:
if (ctx)
put_attr_search_ctx(ctx);
if (m)
unmap_mft_record(ni);
iput(dent_inode);
return ERR_PTR(err);
}
}
/**
* ntfs_mft_writepage - check if a metadata page contains dirty mft records
* @page: metadata page possibly containing dirty mft records
* @wbc: writeback control structure
*
* This is called from the VM when it wants to have a dirty $MFT/$DATA metadata
* page cache page cleaned. The VM has already locked the page and marked it
* clean. Instead of writing the page as a conventional ->writepage function
* would do, we check if the page still contains any dirty mft records (it must
* have done at some point in the past since the page was marked dirty) and if
* none are found, i.e. all mft records are clean, we unlock the page and
* return. The VM is then free to do with the page as it pleases. If on the
* other hand we do find any dirty mft records in the page, we redirty the page
* before unlocking it and returning so the VM knows that the page is still
* busy and cannot be thrown out.
*
* Note, we do not actually write any dirty mft records here because they are
* dirty inodes and hence will be written by the VFS inode dirty code paths.
* There is no need to write them from the VM page dirty code paths, too and in
* fact once we implement journalling it would be a complete nightmare having
* two code paths leading to mft record writeout.
*/
static int ntfs_mft_writepage(struct page *page, struct writeback_control *wbc)
{
struct inode *mft_vi = page->mapping->host;
struct super_block *sb = mft_vi->i_sb;
ntfs_volume *vol = NTFS_SB(sb);
u8 *maddr;
MFT_RECORD *m;
ntfs_inode **extent_nis;
unsigned long mft_no;
int nr, i, j;
BOOL is_dirty = FALSE;
BUG_ON(!PageLocked(page));
BUG_ON(PageWriteback(page));
BUG_ON(mft_vi != vol->mft_ino);
/* The first mft record number in the page. */
mft_no = page->index << (PAGE_CACHE_SHIFT - vol->mft_record_size_bits);
/* Number of mft records in the page. */
nr = PAGE_CACHE_SIZE >> vol->mft_record_size_bits;
BUG_ON(!nr);
ntfs_debug("Entering for %i inodes starting at 0x%lx.", nr, mft_no);
/* Iterate over the mft records in the page looking for a dirty one. */
maddr = (u8*)kmap(page);
for (i = 0; i < nr; ++i, ++mft_no, maddr += vol->mft_record_size) {
struct inode *vi;
ntfs_inode *ni, *eni;
ntfs_attr na;
na.mft_no = mft_no;
na.name = NULL;
na.name_len = 0;
na.type = AT_UNUSED;
/*
* Check if the inode corresponding to this mft record is in
* the VFS inode cache and obtain a reference to it if it is.
*/
ntfs_debug("Looking for inode 0x%lx in icache.", mft_no);
/*
* For inode 0, i.e. $MFT itself, we cannot use ilookup5() from
* here or we deadlock because the inode is already locked by
* the kernel (fs/fs-writeback.c::__sync_single_inode()) and
* ilookup5() waits until the inode is unlocked before
* returning it and it never gets unlocked because
* ntfs_mft_writepage() never returns. )-: Fortunately, we
* have inode 0 pinned in icache for the duration of the mount
* so we can access it directly.
*/
if (!mft_no) {
/* Balance the below iput(). */
vi = igrab(mft_vi);
BUG_ON(vi != mft_vi);
} else
vi = ilookup5(sb, mft_no, (test_t)ntfs_test_inode, &na);
if (vi) {
ntfs_debug("Inode 0x%lx is in icache.", mft_no);
/* The inode is in icache. Check if it is dirty. */
ni = NTFS_I(vi);
if (!NInoDirty(ni)) {
/* The inode is not dirty, skip this record. */
ntfs_debug("Inode 0x%lx is not dirty, "
"continuing search.", mft_no);
iput(vi);
continue;
}
ntfs_debug("Inode 0x%lx is dirty, aborting search.",
mft_no);
/* The inode is dirty, no need to search further. */
iput(vi);
is_dirty = TRUE;
break;
}
ntfs_debug("Inode 0x%lx is not in icache.", mft_no);
/* The inode is not in icache. */
/* Skip the record if it is not a mft record (type "FILE"). */
if (!ntfs_is_mft_recordp(maddr)) {
ntfs_debug("Mft record 0x%lx is not a FILE record, "
"continuing search.", mft_no);
continue;
}
m = (MFT_RECORD*)maddr;
/*
* Skip the mft record if it is not in use. FIXME: What about
* deleted/deallocated (extent) inodes? (AIA)
*/
if (!(m->flags & MFT_RECORD_IN_USE)) {
ntfs_debug("Mft record 0x%lx is not in use, "
"continuing search.", mft_no);
continue;
}
/* Skip the mft record if it is a base inode. */
if (!m->base_mft_record) {
ntfs_debug("Mft record 0x%lx is a base record, "
"continuing search.", mft_no);
continue;
}
/*
* This is an extent mft record. Check if the inode
* corresponding to its base mft record is in icache.
*/
na.mft_no = MREF_LE(m->base_mft_record);
ntfs_debug("Mft record 0x%lx is an extent record. Looking "
"for base inode 0x%lx in icache.", mft_no,
na.mft_no);
vi = ilookup5(sb, na.mft_no, (test_t)ntfs_test_inode,
&na);
if (!vi) {
/*
* The base inode is not in icache. Skip this extent
* mft record.
*/
ntfs_debug("Base inode 0x%lx is not in icache, "
"continuing search.", na.mft_no);
continue;
}
ntfs_debug("Base inode 0x%lx is in icache.", na.mft_no);
/*
* The base inode is in icache. Check if it has the extent
* inode corresponding to this extent mft record attached.
*/
ni = NTFS_I(vi);
down(&ni->extent_lock);
if (ni->nr_extents <= 0) {
/*
* The base inode has no attached extent inodes. Skip
* this extent mft record.
*/
up(&ni->extent_lock);
iput(vi);
continue;
}
/* Iterate over the attached extent inodes. */
extent_nis = ni->ext.extent_ntfs_inos;
for (eni = NULL, j = 0; j < ni->nr_extents; ++j) {
if (mft_no == extent_nis[j]->mft_no) {
/*
* Found the extent inode corresponding to this
* extent mft record.
*/
eni = extent_nis[j];
break;
}
}
/*
* If the extent inode was not attached to the base inode, skip
* this extent mft record.
*/
if (!eni) {
up(&ni->extent_lock);
iput(vi);
continue;
}
/*
* Found the extent inode corrsponding to this extent mft
* record. If it is dirty, no need to search further.
*/
if (NInoDirty(eni)) {
up(&ni->extent_lock);
iput(vi);
is_dirty = TRUE;
break;
}
/* The extent inode is not dirty, so do the next record. */
up(&ni->extent_lock);
iput(vi);
}
kunmap(page);
/* If a dirty mft record was found, redirty the page. */
if (is_dirty) {
ntfs_debug("Inode 0x%lx is dirty. Redirtying the page "
"starting at inode 0x%lx.", mft_no,
page->index << (PAGE_CACHE_SHIFT -
vol->mft_record_size_bits));
redirty_page_for_writepage(wbc, page);
unlock_page(page);
} else {
/*
* Keep the VM happy. This must be done otherwise the
* radix-tree tag PAGECACHE_TAG_DIRTY remains set even though
* the page is clean.
*/
BUG_ON(PageWriteback(page));
set_page_writeback(page);
unlock_page(page);
end_page_writeback(page);
}
ntfs_debug("Done.");
return 0;
}
