/**
* inode.c - NTFS kernel inode handling. Part of the Linux-NTFS project.
*
- * Copyright (c) 2001-2005 Anton Altaparmakov
+ * Copyright (c) 2001-2007 Anton Altaparmakov
*
* This program/include file is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License as published
* Foundation,Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
*/
-#include <linux/pagemap.h>
#include <linux/buffer_head.h>
-#include <linux/smp_lock.h>
-#include <linux/quotaops.h>
+#include <linux/fs.h>
+#include <linux/mm.h>
#include <linux/mount.h>
+#include <linux/mutex.h>
+#include <linux/pagemap.h>
+#include <linux/quotaops.h>
+#include <linux/slab.h>
#include "aops.h"
+#include "attrib.h"
+#include "bitmap.h"
#include "dir.h"
#include "debug.h"
#include "inode.h"
-#include "attrib.h"
+#include "lcnalloc.h"
#include "malloc.h"
#include "mft.h"
#include "time.h"
* If initializing the normal file/directory inode, set @na->type to AT_UNUSED.
* In that case, @na->name and @na->name_len should be set to NULL and 0,
* respectively. Although that is not strictly necessary as
- * ntfs_read_inode_locked() will fill them in later.
+ * ntfs_read_locked_inode() will fill them in later.
*
* Return 0 on success and -errno on error.
*
BUG_ON(!na->name);
i = na->name_len * sizeof(ntfschar);
- ni->name = (ntfschar*)kmalloc(i + sizeof(ntfschar), GFP_ATOMIC);
+ ni->name = kmalloc(i + sizeof(ntfschar), GFP_ATOMIC);
if (!ni->name)
return -ENOMEM;
memcpy(ni->name, na->name, i);
- ni->name[i] = 0;
+ ni->name[na->name_len] = 0;
}
return 0;
}
struct inode *ntfs_iget(struct super_block *sb, unsigned long mft_no)
{
struct inode *vi;
- ntfs_attr na;
int err;
+ ntfs_attr na;
na.mft_no = mft_no;
na.type = AT_UNUSED;
ntfschar *name, u32 name_len)
{
struct inode *vi;
- ntfs_attr na;
int err;
+ ntfs_attr na;
/* Make sure no one calls ntfs_attr_iget() for indices. */
BUG_ON(type == AT_INDEX_ALLOCATION);
u32 name_len)
{
struct inode *vi;
- ntfs_attr na;
int err;
+ ntfs_attr na;
na.mft_no = base_vi->i_ino;
na.type = AT_INDEX_ALLOCATION;
ntfs_inode *ni;
ntfs_debug("Entering.");
- ni = kmem_cache_alloc(ntfs_big_inode_cache, SLAB_NOFS);
+ ni = kmem_cache_alloc(ntfs_big_inode_cache, GFP_NOFS);
if (likely(ni != NULL)) {
ni->state = 0;
return VFS_I(ni);
ntfs_inode *ni;
ntfs_debug("Entering.");
- ni = kmem_cache_alloc(ntfs_inode_cache, SLAB_NOFS);
+ ni = kmem_cache_alloc(ntfs_inode_cache, GFP_NOFS);
if (likely(ni != NULL)) {
ni->state = 0;
return ni;
kmem_cache_free(ntfs_inode_cache, ni);
}
+/*
+ * The attribute runlist lock has separate locking rules from the
+ * normal runlist lock, so split the two lock-classes:
+ */
+static struct lock_class_key attr_list_rl_lock_class;
+
/**
* __ntfs_init_inode - initialize ntfs specific part of an inode
* @sb: super block of mounted volume
atomic_set(&ni->count, 1);
ni->vol = NTFS_SB(sb);
ntfs_init_runlist(&ni->runlist);
- init_MUTEX(&ni->mrec_lock);
+ mutex_init(&ni->mrec_lock);
ni->page = NULL;
ni->page_ofs = 0;
ni->attr_list_size = 0;
ni->attr_list = NULL;
ntfs_init_runlist(&ni->attr_list_rl);
- ni->itype.index.bmp_ino = NULL;
+ lockdep_set_class(&ni->attr_list_rl.lock,
+ &attr_list_rl_lock_class);
ni->itype.index.block_size = 0;
ni->itype.index.vcn_size = 0;
ni->itype.index.collation_rule = 0;
ni->itype.index.block_size_bits = 0;
ni->itype.index.vcn_size_bits = 0;
- init_MUTEX(&ni->extent_lock);
+ mutex_init(&ni->extent_lock);
ni->nr_extents = 0;
ni->ext.base_ntfs_ino = NULL;
}
+/*
+ * Extent inodes get MFT-mapped in a nested way, while the base inode
+ * is still mapped. Teach this nesting to the lock validator by creating
+ * a separate class for nested inode's mrec_lock's:
+ */
+static struct lock_class_key extent_inode_mrec_lock_key;
+
inline ntfs_inode *ntfs_new_extent_inode(struct super_block *sb,
unsigned long mft_no)
{
ntfs_debug("Entering.");
if (likely(ni != NULL)) {
__ntfs_init_inode(sb, ni);
+ lockdep_set_class(&ni->mrec_lock, &extent_inode_mrec_lock_key);
ni->mft_no = mft_no;
ni->type = AT_UNUSED;
ni->name = NULL;
{
ntfs_volume *vol = NTFS_SB(vi->i_sb);
ntfs_inode *ni;
+ struct inode *bvi;
MFT_RECORD *m;
ATTR_RECORD *a;
STANDARD_INFORMATION *si;
/* Setup the generic vfs inode parts now. */
- /* This is the optimal IO size (for stat), not the fs block size. */
- vi->i_blksize = PAGE_CACHE_SIZE;
/*
* This is for checking whether an inode has changed w.r.t. a file so
* that the file can be updated if necessary (compare with f_version).
ntfs_debug("Attribute list found in inode 0x%lx.", vi->i_ino);
NInoSetAttrList(ni);
a = ctx->attr;
- if (a->flags & ATTR_IS_ENCRYPTED ||
- a->flags & ATTR_COMPRESSION_MASK ||
- a->flags & ATTR_IS_SPARSE) {
+ if (a->flags & ATTR_COMPRESSION_MASK) {
ntfs_error(vi->i_sb, "Attribute list attribute is "
- "compressed/encrypted/sparse.");
+ "compressed.");
goto unm_err_out;
}
+ if (a->flags & ATTR_IS_ENCRYPTED ||
+ a->flags & ATTR_IS_SPARSE) {
+ if (a->non_resident) {
+ ntfs_error(vi->i_sb, "Non-resident attribute "
+ "list attribute is encrypted/"
+ "sparse.");
+ goto unm_err_out;
+ }
+ ntfs_warning(vi->i_sb, "Resident attribute list "
+ "attribute in inode 0x%lx is marked "
+ "encrypted/sparse which is not true. "
+ "However, Windows allows this and "
+ "chkdsk does not detect or correct it "
+ "so we will just ignore the invalid "
+ "flags and pretend they are not set.",
+ vi->i_ino);
+ }
/* Now allocate memory for the attribute list. */
ni->attr_list_size = (u32)ntfs_attr_size(a);
ni->attr_list = ntfs_malloc_nofs(ni->attr_list_size);
*/
if (S_ISDIR(vi->i_mode)) {
loff_t bvi_size;
- struct inode *bvi;
ntfs_inode *bni;
INDEX_ROOT *ir;
u8 *ir_end, *index_end;
err = PTR_ERR(bvi);
goto unm_err_out;
}
- ni->itype.index.bmp_ino = bvi;
bni = NTFS_I(bvi);
if (NInoCompressed(bni) || NInoEncrypted(bni) ||
NInoSparse(bni)) {
ntfs_error(vi->i_sb, "$BITMAP attribute is compressed "
"and/or encrypted and/or sparse.");
- goto unm_err_out;
+ goto iput_unm_err_out;
}
/* Consistency check bitmap size vs. index allocation size. */
bvi_size = i_size_read(bvi);
ntfs_error(vi->i_sb, "Index bitmap too small (0x%llx) "
"for index allocation (0x%llx).",
bvi_size << 3, vi->i_size);
- goto unm_err_out;
+ goto iput_unm_err_out;
}
+ /* No longer need the bitmap attribute inode. */
+ iput(bvi);
skip_large_dir_stuff:
/* Setup the operations for this inode. */
vi->i_op = &ntfs_dir_inode_ops;
if (a->non_resident) {
NInoSetNonResident(ni);
if (NInoCompressed(ni) || NInoSparse(ni)) {
- if (a->data.non_resident.compression_unit !=
- 4) {
+ if (NInoCompressed(ni) && a->data.non_resident.
+ compression_unit != 4) {
ntfs_error(vi->i_sb, "Found "
- "nonstandard "
+ "non-standard "
"compression unit (%u "
"instead of 4). "
"Cannot handle this.",
err = -EOPNOTSUPP;
goto unm_err_out;
}
- ni->itype.compressed.block_clusters = 1U <<
- a->data.non_resident.
- compression_unit;
- ni->itype.compressed.block_size = 1U << (
- a->data.non_resident.
- compression_unit +
- vol->cluster_size_bits);
- ni->itype.compressed.block_size_bits = ffs(
- ni->itype.compressed.
- block_size) - 1;
+ if (a->data.non_resident.compression_unit) {
+ ni->itype.compressed.block_size = 1U <<
+ (a->data.non_resident.
+ compression_unit +
+ vol->cluster_size_bits);
+ ni->itype.compressed.block_size_bits =
+ ffs(ni->itype.
+ compressed.
+ block_size) - 1;
+ ni->itype.compressed.block_clusters =
+ 1U << a->data.
+ non_resident.
+ compression_unit;
+ } else {
+ ni->itype.compressed.block_size = 0;
+ ni->itype.compressed.block_size_bits =
+ 0;
+ ni->itype.compressed.block_clusters =
+ 0;
+ }
ni->itype.compressed.size = sle64_to_cpu(
a->data.non_resident.
compressed_size);
vi->i_blocks = ni->allocated_size >> 9;
ntfs_debug("Done.");
return 0;
-
+iput_unm_err_out:
+ iput(bvi);
unm_err_out:
if (!err)
err = -EIO;
*
* Return 0 on success and -errno on error. In the error case, the inode will
* have had make_bad_inode() executed on it.
+ *
+ * Note this cannot be called for AT_INDEX_ALLOCATION.
*/
static int ntfs_read_locked_attr_inode(struct inode *base_vi, struct inode *vi)
{
base_ni = NTFS_I(base_vi);
/* Just mirror the values from the base inode. */
- vi->i_blksize = base_vi->i_blksize;
vi->i_version = base_vi->i_version;
vi->i_uid = base_vi->i_uid;
vi->i_gid = base_vi->i_gid;
}
}
/*
- * The encryption flag set in an index root just means to
- * compress all files.
+ * The compressed/sparse flag set in an index root just means
+ * to compress all files.
*/
if (NInoMstProtected(ni) && ni->type != AT_INDEX_ROOT) {
ntfs_error(vi->i_sb, "Found mst protected attribute "
"the mapping pairs array.");
goto unm_err_out;
}
- if ((NInoCompressed(ni) || NInoSparse(ni)) &&
- ni->type != AT_INDEX_ROOT) {
- if (a->data.non_resident.compression_unit != 4) {
- ntfs_error(vi->i_sb, "Found nonstandard "
+ if (NInoCompressed(ni) || NInoSparse(ni)) {
+ if (NInoCompressed(ni) && a->data.non_resident.
+ compression_unit != 4) {
+ ntfs_error(vi->i_sb, "Found non-standard "
"compression unit (%u instead "
"of 4). Cannot handle this.",
a->data.non_resident.
err = -EOPNOTSUPP;
goto unm_err_out;
}
- ni->itype.compressed.block_clusters = 1U <<
- a->data.non_resident.compression_unit;
- ni->itype.compressed.block_size = 1U << (
- a->data.non_resident.compression_unit +
- vol->cluster_size_bits);
- ni->itype.compressed.block_size_bits = ffs(
- ni->itype.compressed.block_size) - 1;
+ if (a->data.non_resident.compression_unit) {
+ ni->itype.compressed.block_size = 1U <<
+ (a->data.non_resident.
+ compression_unit +
+ vol->cluster_size_bits);
+ ni->itype.compressed.block_size_bits =
+ ffs(ni->itype.compressed.
+ block_size) - 1;
+ ni->itype.compressed.block_clusters = 1U <<
+ a->data.non_resident.
+ compression_unit;
+ } else {
+ ni->itype.compressed.block_size = 0;
+ ni->itype.compressed.block_size_bits = 0;
+ ni->itype.compressed.block_clusters = 0;
+ }
ni->itype.compressed.size = sle64_to_cpu(
a->data.non_resident.compressed_size);
}
ni->allocated_size = sle64_to_cpu(
a->data.non_resident.allocated_size);
}
- /* Setup the operations for this attribute inode. */
- vi->i_op = NULL;
- vi->i_fop = NULL;
if (NInoMstProtected(ni))
vi->i_mapping->a_ops = &ntfs_mst_aops;
else
"Run chkdsk.", err, vi->i_ino, ni->type, ni->name_len,
base_vi->i_ino);
make_bad_inode(vi);
- make_bad_inode(base_vi);
if (err != -ENOMEM)
NVolSetErrors(vol);
return err;
ni = NTFS_I(vi);
base_ni = NTFS_I(base_vi);
/* Just mirror the values from the base inode. */
- vi->i_blksize = base_vi->i_blksize;
vi->i_version = base_vi->i_version;
vi->i_uid = base_vi->i_uid;
vi->i_gid = base_vi->i_gid;
"$INDEX_ALLOCATION attribute.");
goto unm_err_out;
}
+ a = ctx->attr;
if (!a->non_resident) {
ntfs_error(vi->i_sb, "$INDEX_ALLOCATION attribute is "
"resident.");
vi->i_size);
goto iput_unm_err_out;
}
- ni->itype.index.bmp_ino = bvi;
+ iput(bvi);
skip_large_index_stuff:
/* Setup the operations for this index inode. */
vi->i_op = NULL;
ntfs_debug("Done.");
return 0;
-
iput_unm_err_out:
iput(bvi);
unm_err_out:
return err;
}
+/*
+ * The MFT inode has special locking, so teach the lock validator
+ * about this by splitting off the locking rules of the MFT from
+ * the locking rules of other inodes. The MFT inode can never be
+ * accessed from the VFS side (or even internally), only by the
+ * map_mft functions.
+ */
+static struct lock_class_key mft_ni_runlist_lock_key, mft_ni_mrec_lock_key;
+
/**
* ntfs_read_inode_mount - special read_inode for mount time use only
* @vi: inode to read
} else /* if (!err) */ {
ATTR_LIST_ENTRY *al_entry, *next_al_entry;
u8 *al_end;
+ static const char *es = " Not allowed. $MFT is corrupt. "
+ "You should run chkdsk.";
ntfs_debug("Attribute list attribute found in $MFT.");
NInoSetAttrList(ni);
a = ctx->attr;
- if (a->flags & ATTR_IS_ENCRYPTED ||
- a->flags & ATTR_COMPRESSION_MASK ||
- a->flags & ATTR_IS_SPARSE) {
+ if (a->flags & ATTR_COMPRESSION_MASK) {
ntfs_error(sb, "Attribute list attribute is "
- "compressed/encrypted/sparse. Not "
- "allowed. $MFT is corrupt. You should "
- "run chkdsk.");
+ "compressed.%s", es);
goto put_err_out;
}
+ if (a->flags & ATTR_IS_ENCRYPTED ||
+ a->flags & ATTR_IS_SPARSE) {
+ if (a->non_resident) {
+ ntfs_error(sb, "Non-resident attribute list "
+ "attribute is encrypted/"
+ "sparse.%s", es);
+ goto put_err_out;
+ }
+ ntfs_warning(sb, "Resident attribute list attribute "
+ "in $MFT system file is marked "
+ "encrypted/sparse which is not true. "
+ "However, Windows allows this and "
+ "chkdsk does not detect or correct it "
+ "so we will just ignore the invalid "
+ "flags and pretend they are not set.");
+ }
/* Now allocate memory for the attribute list. */
ni->attr_list_size = (u32)ntfs_attr_size(a);
ni->attr_list = ntfs_malloc_nofs(ni->attr_list_size);
ntfs_attr_put_search_ctx(ctx);
ntfs_debug("Done.");
ntfs_free(m);
+
+ /*
+ * Split the locking rules of the MFT inode from the
+ * locking rules of other inodes:
+ */
+ lockdep_set_class(&ni->runlist.lock, &mft_ni_runlist_lock_key);
+ lockdep_set_class(&ni->mrec_lock, &mft_ni_mrec_lock_key);
+
return 0;
em_put_err_out:
return -1;
}
-/**
- * ntfs_put_inode - handler for when the inode reference count is decremented
- * @vi: vfs inode
- *
- * The VFS calls ntfs_put_inode() every time the inode reference count (i_count)
- * is about to be decremented (but before the decrement itself.
- *
- * If the inode @vi is a directory with two references, one of which is being
- * dropped, we need to put the attribute inode for the directory index bitmap,
- * if it is present, otherwise the directory inode would remain pinned for
- * ever.
- */
-void ntfs_put_inode(struct inode *vi)
-{
- if (S_ISDIR(vi->i_mode) && atomic_read(&vi->i_count) == 2) {
- ntfs_inode *ni = NTFS_I(vi);
- if (NInoIndexAllocPresent(ni)) {
- struct inode *bvi = NULL;
- down(&vi->i_sem);
- if (atomic_read(&vi->i_count) == 2) {
- bvi = ni->itype.index.bmp_ino;
- if (bvi)
- ni->itype.index.bmp_ino = NULL;
- }
- up(&vi->i_sem);
- if (bvi)
- iput(bvi);
- }
- }
-}
-
static void __ntfs_clear_inode(ntfs_inode *ni)
{
/* Free all alocated memory. */
{
ntfs_inode *ni = NTFS_I(vi);
- /*
- * If the inode @vi is an index inode we need to put the attribute
- * inode for the index bitmap, if it is present, otherwise the index
- * inode would disappear and the attribute inode for the index bitmap
- * would no longer be referenced from anywhere and thus it would remain
- * pinned for ever.
- */
- if (NInoAttr(ni) && (ni->type == AT_INDEX_ALLOCATION) &&
- NInoIndexAllocPresent(ni) && ni->itype.index.bmp_ino) {
- iput(ni->itype.index.bmp_ino);
- ni->itype.index.bmp_ino = NULL;
- }
#ifdef NTFS_RW
if (NInoDirty(ni)) {
- BOOL was_bad = (is_bad_inode(vi));
+ bool was_bad = (is_bad_inode(vi));
/* Committing the inode also commits all extent inodes. */
ntfs_commit_inode(vi);
#ifdef NTFS_RW
+static const char *es = " Leaving inconsistent metadata. Unmount and run "
+ "chkdsk.";
+
/**
* ntfs_truncate - called when the i_size of an ntfs inode is changed
* @vi: inode for which the i_size was changed
*
- * We do not support i_size changes yet.
+ * We only support i_size changes for normal files at present, i.e. not
+ * compressed and not encrypted. This is enforced in ntfs_setattr(), see
+ * below.
*
* The kernel guarantees that @vi is a regular file (S_ISREG() is true) and
* that the change is allowed.
*
* Returns 0 on success or -errno on error.
*
- * Called with ->i_sem held. In all but one case ->i_alloc_sem is held for
- * writing. The only case where ->i_alloc_sem is not held is
+ * Called with ->i_mutex held. In all but one case ->i_alloc_sem is held for
+ * writing. The only case in the kernel where ->i_alloc_sem is not held is
* mm/filemap.c::generic_file_buffered_write() where vmtruncate() is called
- * with the current i_size as the offset which means that it is a noop as far
- * as ntfs_truncate() is concerned.
+ * with the current i_size as the offset. The analogous place in NTFS is in
+ * fs/ntfs/file.c::ntfs_file_buffered_write() where we call vmtruncate() again
+ * without holding ->i_alloc_sem.
*/
int ntfs_truncate(struct inode *vi)
{
- ntfs_inode *ni = NTFS_I(vi);
+ s64 new_size, old_size, nr_freed, new_alloc_size, old_alloc_size;
+ VCN highest_vcn;
+ unsigned long flags;
+ ntfs_inode *base_ni, *ni = NTFS_I(vi);
ntfs_volume *vol = ni->vol;
ntfs_attr_search_ctx *ctx;
MFT_RECORD *m;
ATTR_RECORD *a;
const char *te = " Leaving file length out of sync with i_size.";
- int err;
+ int err, mp_size, size_change, alloc_change;
+ u32 attr_len;
ntfs_debug("Entering for inode 0x%lx.", vi->i_ino);
BUG_ON(NInoAttr(ni));
+ BUG_ON(S_ISDIR(vi->i_mode));
+ BUG_ON(NInoMstProtected(ni));
BUG_ON(ni->nr_extents < 0);
- m = map_mft_record(ni);
+retry_truncate:
+ /*
+ * Lock the runlist for writing and map the mft record to ensure it is
+ * safe to mess with the attribute runlist and sizes.
+ */
+ down_write(&ni->runlist.lock);
+ if (!NInoAttr(ni))
+ base_ni = ni;
+ else
+ base_ni = ni->ext.base_ntfs_ino;
+ m = map_mft_record(base_ni);
if (IS_ERR(m)) {
err = PTR_ERR(m);
ntfs_error(vi->i_sb, "Failed to map mft record for inode 0x%lx "
"(error code %d).%s", vi->i_ino, err, te);
ctx = NULL;
m = NULL;
- goto err_out;
+ goto old_bad_out;
}
- ctx = ntfs_attr_get_search_ctx(ni, m);
+ ctx = ntfs_attr_get_search_ctx(base_ni, m);
if (unlikely(!ctx)) {
ntfs_error(vi->i_sb, "Failed to allocate a search context for "
"inode 0x%lx (not enough memory).%s",
vi->i_ino, te);
err = -ENOMEM;
- goto err_out;
+ goto old_bad_out;
}
err = ntfs_attr_lookup(ni->type, ni->name, ni->name_len,
CASE_SENSITIVE, 0, NULL, 0, ctx);
if (unlikely(err)) {
- if (err == -ENOENT)
+ if (err == -ENOENT) {
ntfs_error(vi->i_sb, "Open attribute is missing from "
"mft record. Inode 0x%lx is corrupt. "
- "Run chkdsk.", vi->i_ino);
- else
+ "Run chkdsk.%s", vi->i_ino, te);
+ err = -EIO;
+ } else
ntfs_error(vi->i_sb, "Failed to lookup attribute in "
- "inode 0x%lx (error code %d).",
- vi->i_ino, err);
- goto err_out;
+ "inode 0x%lx (error code %d).%s",
+ vi->i_ino, err, te);
+ goto old_bad_out;
}
+ m = ctx->mrec;
a = ctx->attr;
- /* If the size has not changed there is nothing to do. */
- if (ntfs_attr_size(a) == i_size_read(vi))
- goto done;
- // TODO: Implement the truncate...
- ntfs_error(vi->i_sb, "Inode size has changed but this is not "
- "implemented yet. Resetting inode size to old value. "
- " This is most likely a bug in the ntfs driver!");
- i_size_write(vi, ntfs_attr_size(a));
-done:
+ /*
+ * The i_size of the vfs inode is the new size for the attribute value.
+ */
+ new_size = i_size_read(vi);
+ /* The current size of the attribute value is the old size. */
+ old_size = ntfs_attr_size(a);
+ /* Calculate the new allocated size. */
+ if (NInoNonResident(ni))
+ new_alloc_size = (new_size + vol->cluster_size - 1) &
+ ~(s64)vol->cluster_size_mask;
+ else
+ new_alloc_size = (new_size + 7) & ~7;
+ /* The current allocated size is the old allocated size. */
+ read_lock_irqsave(&ni->size_lock, flags);
+ old_alloc_size = ni->allocated_size;
+ read_unlock_irqrestore(&ni->size_lock, flags);
+ /*
+ * The change in the file size. This will be 0 if no change, >0 if the
+ * size is growing, and <0 if the size is shrinking.
+ */
+ size_change = -1;
+ if (new_size - old_size >= 0) {
+ size_change = 1;
+ if (new_size == old_size)
+ size_change = 0;
+ }
+ /* As above for the allocated size. */
+ alloc_change = -1;
+ if (new_alloc_size - old_alloc_size >= 0) {
+ alloc_change = 1;
+ if (new_alloc_size == old_alloc_size)
+ alloc_change = 0;
+ }
+ /*
+ * If neither the size nor the allocation are being changed there is
+ * nothing to do.
+ */
+ if (!size_change && !alloc_change)
+ goto unm_done;
+ /* If the size is changing, check if new size is allowed in $AttrDef. */
+ if (size_change) {
+ err = ntfs_attr_size_bounds_check(vol, ni->type, new_size);
+ if (unlikely(err)) {
+ if (err == -ERANGE) {
+ ntfs_error(vol->sb, "Truncate would cause the "
+ "inode 0x%lx to %simum size "
+ "for its attribute type "
+ "(0x%x). Aborting truncate.",
+ vi->i_ino,
+ new_size > old_size ? "exceed "
+ "the max" : "go under the min",
+ le32_to_cpu(ni->type));
+ err = -EFBIG;
+ } else {
+ ntfs_error(vol->sb, "Inode 0x%lx has unknown "
+ "attribute type 0x%x. "
+ "Aborting truncate.",
+ vi->i_ino,
+ le32_to_cpu(ni->type));
+ err = -EIO;
+ }
+ /* Reset the vfs inode size to the old size. */
+ i_size_write(vi, old_size);
+ goto err_out;
+ }
+ }
+ if (NInoCompressed(ni) || NInoEncrypted(ni)) {
+ ntfs_warning(vi->i_sb, "Changes in inode size are not "
+ "supported yet for %s files, ignoring.",
+ NInoCompressed(ni) ? "compressed" :
+ "encrypted");
+ err = -EOPNOTSUPP;
+ goto bad_out;
+ }
+ if (a->non_resident)
+ goto do_non_resident_truncate;
+ BUG_ON(NInoNonResident(ni));
+ /* Resize the attribute record to best fit the new attribute size. */
+ if (new_size < vol->mft_record_size &&
+ !ntfs_resident_attr_value_resize(m, a, new_size)) {
+ /* The resize succeeded! */
+ flush_dcache_mft_record_page(ctx->ntfs_ino);
+ mark_mft_record_dirty(ctx->ntfs_ino);
+ write_lock_irqsave(&ni->size_lock, flags);
+ /* Update the sizes in the ntfs inode and all is done. */
+ ni->allocated_size = le32_to_cpu(a->length) -
+ le16_to_cpu(a->data.resident.value_offset);
+ /*
+ * Note ntfs_resident_attr_value_resize() has already done any
+ * necessary data clearing in the attribute record. When the
+ * file is being shrunk vmtruncate() will already have cleared
+ * the top part of the last partial page, i.e. since this is
+ * the resident case this is the page with index 0. However,
+ * when the file is being expanded, the page cache page data
+ * between the old data_size, i.e. old_size, and the new_size
+ * has not been zeroed. Fortunately, we do not need to zero it
+ * either since on one hand it will either already be zero due
+ * to both readpage and writepage clearing partial page data
+ * beyond i_size in which case there is nothing to do or in the
+ * case of the file being mmap()ped at the same time, POSIX
+ * specifies that the behaviour is unspecified thus we do not
+ * have to do anything. This means that in our implementation
+ * in the rare case that the file is mmap()ped and a write
+ * occured into the mmap()ped region just beyond the file size
+ * and writepage has not yet been called to write out the page
+ * (which would clear the area beyond the file size) and we now
+ * extend the file size to incorporate this dirty region
+ * outside the file size, a write of the page would result in
+ * this data being written to disk instead of being cleared.
+ * Given both POSIX and the Linux mmap(2) man page specify that
+ * this corner case is undefined, we choose to leave it like
+ * that as this is much simpler for us as we cannot lock the
+ * relevant page now since we are holding too many ntfs locks
+ * which would result in a lock reversal deadlock.
+ */
+ ni->initialized_size = new_size;
+ write_unlock_irqrestore(&ni->size_lock, flags);
+ goto unm_done;
+ }
+ /* If the above resize failed, this must be an attribute extension. */
+ BUG_ON(size_change < 0);
+ /*
+ * We have to drop all the locks so we can call
+ * ntfs_attr_make_non_resident(). This could be optimised by try-
+ * locking the first page cache page and only if that fails dropping
+ * the locks, locking the page, and redoing all the locking and
+ * lookups. While this would be a huge optimisation, it is not worth
+ * it as this is definitely a slow code path as it only ever can happen
+ * once for any given file.
+ */
ntfs_attr_put_search_ctx(ctx);
- unmap_mft_record(ni);
- NInoClearTruncateFailed(ni);
- ntfs_debug("Done.");
- return 0;
-err_out:
- if (err != -ENOMEM) {
+ unmap_mft_record(base_ni);
+ up_write(&ni->runlist.lock);
+ /*
+ * Not enough space in the mft record, try to make the attribute
+ * non-resident and if successful restart the truncation process.
+ */
+ err = ntfs_attr_make_non_resident(ni, old_size);
+ if (likely(!err))
+ goto retry_truncate;
+ /*
+ * Could not make non-resident. If this is due to this not being
+ * permitted for this attribute type or there not being enough space,
+ * try to make other attributes non-resident. Otherwise fail.
+ */
+ if (unlikely(err != -EPERM && err != -ENOSPC)) {
+ ntfs_error(vol->sb, "Cannot truncate inode 0x%lx, attribute "
+ "type 0x%x, because the conversion from "
+ "resident to non-resident attribute failed "
+ "with error code %i.", vi->i_ino,
+ (unsigned)le32_to_cpu(ni->type), err);
+ if (err != -ENOMEM)
+ err = -EIO;
+ goto conv_err_out;
+ }
+ /* TODO: Not implemented from here, abort. */
+ if (err == -ENOSPC)
+ ntfs_error(vol->sb, "Not enough space in the mft record/on "
+ "disk for the non-resident attribute value. "
+ "This case is not implemented yet.");
+ else /* if (err == -EPERM) */
+ ntfs_error(vol->sb, "This attribute type may not be "
+ "non-resident. This case is not implemented "
+ "yet.");
+ err = -EOPNOTSUPP;
+ goto conv_err_out;
+#if 0
+ // TODO: Attempt to make other attributes non-resident.
+ if (!err)
+ goto do_resident_extend;
+ /*
+ * Both the attribute list attribute and the standard information
+ * attribute must remain in the base inode. Thus, if this is one of
+ * these attributes, we have to try to move other attributes out into
+ * extent mft records instead.
+ */
+ if (ni->type == AT_ATTRIBUTE_LIST ||
+ ni->type == AT_STANDARD_INFORMATION) {
+ // TODO: Attempt to move other attributes into extent mft
+ // records.
+ err = -EOPNOTSUPP;
+ if (!err)
+ goto do_resident_extend;
+ goto err_out;
+ }
+ // TODO: Attempt to move this attribute to an extent mft record, but
+ // only if it is not already the only attribute in an mft record in
+ // which case there would be nothing to gain.
+ err = -EOPNOTSUPP;
+ if (!err)
+ goto do_resident_extend;
+ /* There is nothing we can do to make enough space. )-: */
+ goto err_out;
+#endif
+do_non_resident_truncate:
+ BUG_ON(!NInoNonResident(ni));
+ if (alloc_change < 0) {
+ highest_vcn = sle64_to_cpu(a->data.non_resident.highest_vcn);
+ if (highest_vcn > 0 &&
+ old_alloc_size >> vol->cluster_size_bits >
+ highest_vcn + 1) {
+ /*
+ * This attribute has multiple extents. Not yet
+ * supported.
+ */
+ ntfs_error(vol->sb, "Cannot truncate inode 0x%lx, "
+ "attribute type 0x%x, because the "
+ "attribute is highly fragmented (it "
+ "consists of multiple extents) and "
+ "this case is not implemented yet.",
+ vi->i_ino,
+ (unsigned)le32_to_cpu(ni->type));
+ err = -EOPNOTSUPP;
+ goto bad_out;
+ }
+ }
+ /*
+ * If the size is shrinking, need to reduce the initialized_size and
+ * the data_size before reducing the allocation.
+ */
+ if (size_change < 0) {
+ /*
+ * Make the valid size smaller (i_size is already up-to-date).
+ */
+ write_lock_irqsave(&ni->size_lock, flags);
+ if (new_size < ni->initialized_size) {
+ ni->initialized_size = new_size;
+ a->data.non_resident.initialized_size =
+ cpu_to_sle64(new_size);
+ }
+ a->data.non_resident.data_size = cpu_to_sle64(new_size);
+ write_unlock_irqrestore(&ni->size_lock, flags);
+ flush_dcache_mft_record_page(ctx->ntfs_ino);
+ mark_mft_record_dirty(ctx->ntfs_ino);
+ /* If the allocated size is not changing, we are done. */
+ if (!alloc_change)
+ goto unm_done;
+ /*
+ * If the size is shrinking it makes no sense for the
+ * allocation to be growing.
+ */
+ BUG_ON(alloc_change > 0);
+ } else /* if (size_change >= 0) */ {
+ /*
+ * The file size is growing or staying the same but the
+ * allocation can be shrinking, growing or staying the same.
+ */
+ if (alloc_change > 0) {
+ /*
+ * We need to extend the allocation and possibly update
+ * the data size. If we are updating the data size,
+ * since we are not touching the initialized_size we do
+ * not need to worry about the actual data on disk.
+ * And as far as the page cache is concerned, there
+ * will be no pages beyond the old data size and any
+ * partial region in the last page between the old and
+ * new data size (or the end of the page if the new
+ * data size is outside the page) does not need to be
+ * modified as explained above for the resident
+ * attribute truncate case. To do this, we simply drop
+ * the locks we hold and leave all the work to our
+ * friendly helper ntfs_attr_extend_allocation().
+ */
+ ntfs_attr_put_search_ctx(ctx);
+ unmap_mft_record(base_ni);
+ up_write(&ni->runlist.lock);
+ err = ntfs_attr_extend_allocation(ni, new_size,
+ size_change > 0 ? new_size : -1, -1);
+ /*
+ * ntfs_attr_extend_allocation() will have done error
+ * output already.
+ */
+ goto done;
+ }
+ if (!alloc_change)
+ goto alloc_done;
+ }
+ /* alloc_change < 0 */
+ /* Free the clusters. */
+ nr_freed = ntfs_cluster_free(ni, new_alloc_size >>
+ vol->cluster_size_bits, -1, ctx);
+ m = ctx->mrec;
+ a = ctx->attr;
+ if (unlikely(nr_freed < 0)) {
+ ntfs_error(vol->sb, "Failed to release cluster(s) (error code "
+ "%lli). Unmount and run chkdsk to recover "
+ "the lost cluster(s).", (long long)nr_freed);
NVolSetErrors(vol);
- make_bad_inode(vi);
+ nr_freed = 0;
}
+ /* Truncate the runlist. */
+ err = ntfs_rl_truncate_nolock(vol, &ni->runlist,
+ new_alloc_size >> vol->cluster_size_bits);
+ /*
+ * If the runlist truncation failed and/or the search context is no
+ * longer valid, we cannot resize the attribute record or build the
+ * mapping pairs array thus we mark the inode bad so that no access to
+ * the freed clusters can happen.
+ */
+ if (unlikely(err || IS_ERR(m))) {
+ ntfs_error(vol->sb, "Failed to %s (error code %li).%s",
+ IS_ERR(m) ?
+ "restore attribute search context" :
+ "truncate attribute runlist",
+ IS_ERR(m) ? PTR_ERR(m) : err, es);
+ err = -EIO;
+ goto bad_out;
+ }
+ /* Get the size for the shrunk mapping pairs array for the runlist. */
+ mp_size = ntfs_get_size_for_mapping_pairs(vol, ni->runlist.rl, 0, -1);
+ if (unlikely(mp_size <= 0)) {
+ ntfs_error(vol->sb, "Cannot shrink allocation of inode 0x%lx, "
+ "attribute type 0x%x, because determining the "
+ "size for the mapping pairs failed with error "
+ "code %i.%s", vi->i_ino,
+ (unsigned)le32_to_cpu(ni->type), mp_size, es);
+ err = -EIO;
+ goto bad_out;
+ }
+ /*
+ * Shrink the attribute record for the new mapping pairs array. Note,
+ * this cannot fail since we are making the attribute smaller thus by
+ * definition there is enough space to do so.
+ */
+ attr_len = le32_to_cpu(a->length);
+ err = ntfs_attr_record_resize(m, a, mp_size +
+ le16_to_cpu(a->data.non_resident.mapping_pairs_offset));
+ BUG_ON(err);
+ /*
+ * Generate the mapping pairs array directly into the attribute record.
+ */
+ err = ntfs_mapping_pairs_build(vol, (u8*)a +
+ le16_to_cpu(a->data.non_resident.mapping_pairs_offset),
+ mp_size, ni->runlist.rl, 0, -1, NULL);
+ if (unlikely(err)) {
+ ntfs_error(vol->sb, "Cannot shrink allocation of inode 0x%lx, "
+ "attribute type 0x%x, because building the "
+ "mapping pairs failed with error code %i.%s",
+ vi->i_ino, (unsigned)le32_to_cpu(ni->type),
+ err, es);
+ err = -EIO;
+ goto bad_out;
+ }
+ /* Update the allocated/compressed size as well as the highest vcn. */
+ a->data.non_resident.highest_vcn = cpu_to_sle64((new_alloc_size >>
+ vol->cluster_size_bits) - 1);
+ write_lock_irqsave(&ni->size_lock, flags);
+ ni->allocated_size = new_alloc_size;
+ a->data.non_resident.allocated_size = cpu_to_sle64(new_alloc_size);
+ if (NInoSparse(ni) || NInoCompressed(ni)) {
+ if (nr_freed) {
+ ni->itype.compressed.size -= nr_freed <<
+ vol->cluster_size_bits;
+ BUG_ON(ni->itype.compressed.size < 0);
+ a->data.non_resident.compressed_size = cpu_to_sle64(
+ ni->itype.compressed.size);
+ vi->i_blocks = ni->itype.compressed.size >> 9;
+ }
+ } else
+ vi->i_blocks = new_alloc_size >> 9;
+ write_unlock_irqrestore(&ni->size_lock, flags);
+ /*
+ * We have shrunk the allocation. If this is a shrinking truncate we
+ * have already dealt with the initialized_size and the data_size above
+ * and we are done. If the truncate is only changing the allocation
+ * and not the data_size, we are also done. If this is an extending
+ * truncate, need to extend the data_size now which is ensured by the
+ * fact that @size_change is positive.
+ */
+alloc_done:
+ /*
+ * If the size is growing, need to update it now. If it is shrinking,
+ * we have already updated it above (before the allocation change).
+ */
+ if (size_change > 0)
+ a->data.non_resident.data_size = cpu_to_sle64(new_size);
+ /* Ensure the modified mft record is written out. */
+ flush_dcache_mft_record_page(ctx->ntfs_ino);
+ mark_mft_record_dirty(ctx->ntfs_ino);
+unm_done:
+ ntfs_attr_put_search_ctx(ctx);
+ unmap_mft_record(base_ni);
+ up_write(&ni->runlist.lock);
+done:
+ /* Update the mtime and ctime on the base inode. */
+ /* normally ->truncate shouldn't update ctime or mtime,
+ * but ntfs did before so it got a copy & paste version
+ * of file_update_time. one day someone should fix this
+ * for real.
+ */
+ if (!IS_NOCMTIME(VFS_I(base_ni)) && !IS_RDONLY(VFS_I(base_ni))) {
+ struct timespec now = current_fs_time(VFS_I(base_ni)->i_sb);
+ int sync_it = 0;
+
+ if (!timespec_equal(&VFS_I(base_ni)->i_mtime, &now) ||
+ !timespec_equal(&VFS_I(base_ni)->i_ctime, &now))
+ sync_it = 1;
+ VFS_I(base_ni)->i_mtime = now;
+ VFS_I(base_ni)->i_ctime = now;
+
+ if (sync_it)
+ mark_inode_dirty_sync(VFS_I(base_ni));
+ }
+
+ if (likely(!err)) {
+ NInoClearTruncateFailed(ni);
+ ntfs_debug("Done.");
+ }
+ return err;
+old_bad_out:
+ old_size = -1;
+bad_out:
+ if (err != -ENOMEM && err != -EOPNOTSUPP)
+ NVolSetErrors(vol);
+ if (err != -EOPNOTSUPP)
+ NInoSetTruncateFailed(ni);
+ else if (old_size >= 0)
+ i_size_write(vi, old_size);
+err_out:
if (ctx)
ntfs_attr_put_search_ctx(ctx);
if (m)
- unmap_mft_record(ni);
- NInoSetTruncateFailed(ni);
+ unmap_mft_record(base_ni);
+ up_write(&ni->runlist.lock);
+out:
+ ntfs_debug("Failed. Returning error code %i.", err);
return err;
+conv_err_out:
+ if (err != -ENOMEM && err != -EOPNOTSUPP)
+ NVolSetErrors(vol);
+ if (err != -EOPNOTSUPP)
+ NInoSetTruncateFailed(ni);
+ else
+ i_size_write(vi, old_size);
+ goto out;
}
/**
* We also abort all changes of user, group, and mode as we do not implement
* the NTFS ACLs yet.
*
- * Called with ->i_sem held. For the ATTR_SIZE (i.e. ->truncate) case, also
+ * Called with ->i_mutex held. For the ATTR_SIZE (i.e. ->truncate) case, also
* called with ->i_alloc_sem held for writing.
*
* Basically this is a copy of generic notify_change() and inode_setattr()
err = inode_change_ok(vi, attr);
if (err)
- return err;
-
+ goto out;
/* We do not support NTFS ACLs yet. */
if (ia_valid & (ATTR_UID | ATTR_GID | ATTR_MODE)) {
ntfs_warning(vi->i_sb, "Changes in user/group/mode are not "
err = -EOPNOTSUPP;
goto out;
}
-
if (ia_valid & ATTR_SIZE) {
if (attr->ia_size != i_size_read(vi)) {
- ntfs_warning(vi->i_sb, "Changes in inode size are not "
- "supported yet, ignoring.");
- err = -EOPNOTSUPP;
- // TODO: Implement...
- // err = vmtruncate(vi, attr->ia_size);
+ ntfs_inode *ni = NTFS_I(vi);
+ /*
+ * FIXME: For now we do not support resizing of
+ * compressed or encrypted files yet.
+ */
+ if (NInoCompressed(ni) || NInoEncrypted(ni)) {
+ ntfs_warning(vi->i_sb, "Changes in inode size "
+ "are not supported yet for "
+ "%s files, ignoring.",
+ NInoCompressed(ni) ?
+ "compressed" : "encrypted");
+ err = -EOPNOTSUPP;
+ } else
+ err = vmtruncate(vi, attr->ia_size);
if (err || ia_valid == ATTR_SIZE)
goto out;
} else {
MFT_RECORD *m;
STANDARD_INFORMATION *si;
int err = 0;
- BOOL modified = FALSE;
+ bool modified = false;
ntfs_debug("Entering for %sinode 0x%lx.", NInoAttr(ni) ? "attr " : "",
vi->i_ino);
sle64_to_cpu(si->last_data_change_time),
(long long)sle64_to_cpu(nt));
si->last_data_change_time = nt;
- modified = TRUE;
+ modified = true;
}
nt = utc2ntfs(vi->i_ctime);
if (si->last_mft_change_time != nt) {
sle64_to_cpu(si->last_mft_change_time),
(long long)sle64_to_cpu(nt));
si->last_mft_change_time = nt;
- modified = TRUE;
+ modified = true;
}
nt = utc2ntfs(vi->i_atime);
if (si->last_access_time != nt) {
(long long)sle64_to_cpu(si->last_access_time),
(long long)sle64_to_cpu(nt));
si->last_access_time = nt;
- modified = TRUE;
+ modified = true;
}
/*
* If we just modified the standard information attribute we need to
* record will be cleaned and written out to disk below, i.e. before
* this function returns.
*/
- if (modified && !NInoTestSetDirty(ctx->ntfs_ino))
- mark_ntfs_record_dirty(ctx->ntfs_ino->page,
- ctx->ntfs_ino->page_ofs);
+ if (modified) {
+ flush_dcache_mft_record_page(ctx->ntfs_ino);
+ if (!NInoTestSetDirty(ctx->ntfs_ino))
+ mark_ntfs_record_dirty(ctx->ntfs_ino->page,
+ ctx->ntfs_ino->page_ofs);
+ }
ntfs_attr_put_search_ctx(ctx);
/* Now the access times are updated, write the base mft record. */
if (NInoDirty(ni))
err = write_mft_record(ni, m, sync);
/* Write all attached extent mft records. */
- down(&ni->extent_lock);
+ mutex_lock(&ni->extent_lock);
if (ni->nr_extents > 0) {
ntfs_inode **extent_nis = ni->ext.extent_ntfs_inos;
int i;
}
}
}
- up(&ni->extent_lock);
+ mutex_unlock(&ni->extent_lock);
unmap_mft_record(ni);
if (unlikely(err))
goto err_out;
"retries later.");
mark_inode_dirty(vi);
} else {
- ntfs_error(vi->i_sb, "Failed (error code %i): Marking inode "
- "as bad. You should run chkdsk.", -err);
- make_bad_inode(vi);
+ ntfs_error(vi->i_sb, "Failed (error %i): Run chkdsk.", -err);
NVolSetErrors(ni->vol);
}
return err;
git://ftp.safe.ca / safe / jmp / linux-2.6 / blobdiff