u64 start, u64 end, int *page_started,
unsigned long *nr_written, int unlock);
-static int btrfs_init_inode_security(struct inode *inode, struct inode *dir)
+static int btrfs_init_inode_security(struct btrfs_trans_handle *trans,
+ struct inode *inode, struct inode *dir)
{
int err;
- err = btrfs_init_acl(inode, dir);
+ err = btrfs_init_acl(trans, inode, dir);
if (!err)
- err = btrfs_xattr_security_init(inode, dir);
+ err = btrfs_xattr_security_init(trans, inode, dir);
return err;
}
btrfs_mark_buffer_dirty(leaf);
btrfs_free_path(path);
+ /*
+ * we're an inline extent, so nobody can
+ * extend the file past i_size without locking
+ * a page we already have locked.
+ *
+ * We must do any isize and inode updates
+ * before we unlock the pages. Otherwise we
+ * could end up racing with unlink.
+ */
BTRFS_I(inode)->disk_i_size = inode->i_size;
btrfs_update_inode(trans, root, inode);
+
return 0;
fail:
btrfs_free_path(path);
return 1;
}
- ret = btrfs_drop_extents(trans, root, inode, start,
- aligned_end, aligned_end, start,
+ ret = btrfs_drop_extents(trans, inode, start, aligned_end,
&hint_byte, 1);
BUG_ON(ret);
start, end,
total_compressed, pages);
}
- btrfs_end_transaction(trans, root);
if (ret == 0) {
/*
* inline extent creation worked, we don't need
* and free up our temp pages.
*/
extent_clear_unlock_delalloc(inode,
- &BTRFS_I(inode)->io_tree,
- start, end, NULL, 1, 0,
- 0, 1, 1, 1, 0);
- ret = 0;
+ &BTRFS_I(inode)->io_tree,
+ start, end, NULL,
+ EXTENT_CLEAR_UNLOCK_PAGE | EXTENT_CLEAR_DIRTY |
+ EXTENT_CLEAR_DELALLOC |
+ EXTENT_CLEAR_ACCOUNTING |
+ EXTENT_SET_WRITEBACK | EXTENT_END_WRITEBACK);
+
+ btrfs_end_transaction(trans, root);
goto free_pages_out;
}
+ btrfs_end_transaction(trans, root);
}
if (will_compress) {
struct btrfs_root *root = BTRFS_I(inode)->root;
struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
struct extent_io_tree *io_tree;
- int ret;
+ int ret = 0;
if (list_empty(&async_cow->extents))
return 0;
- trans = btrfs_join_transaction(root, 1);
while (!list_empty(&async_cow->extents)) {
async_extent = list_entry(async_cow->extents.next,
io_tree = &BTRFS_I(inode)->io_tree;
+retry:
/* did the compression code fall back to uncompressed IO? */
if (!async_extent->pages) {
int page_started = 0;
async_extent->ram_size - 1, GFP_NOFS);
/* allocate blocks */
- cow_file_range(inode, async_cow->locked_page,
- async_extent->start,
- async_extent->start +
- async_extent->ram_size - 1,
- &page_started, &nr_written, 0);
+ ret = cow_file_range(inode, async_cow->locked_page,
+ async_extent->start,
+ async_extent->start +
+ async_extent->ram_size - 1,
+ &page_started, &nr_written, 0);
/*
* if page_started, cow_file_range inserted an
* and IO for us. Otherwise, we need to submit
* all those pages down to the drive.
*/
- if (!page_started)
+ if (!page_started && !ret)
extent_write_locked_range(io_tree,
inode, async_extent->start,
async_extent->start +
lock_extent(io_tree, async_extent->start,
async_extent->start + async_extent->ram_size - 1,
GFP_NOFS);
+
+ trans = btrfs_join_transaction(root, 1);
+ ret = btrfs_reserve_extent(trans, root,
+ async_extent->compressed_size,
+ async_extent->compressed_size,
+ 0, alloc_hint,
+ (u64)-1, &ins, 1);
+ btrfs_end_transaction(trans, root);
+
+ if (ret) {
+ int i;
+ for (i = 0; i < async_extent->nr_pages; i++) {
+ WARN_ON(async_extent->pages[i]->mapping);
+ page_cache_release(async_extent->pages[i]);
+ }
+ kfree(async_extent->pages);
+ async_extent->nr_pages = 0;
+ async_extent->pages = NULL;
+ unlock_extent(io_tree, async_extent->start,
+ async_extent->start +
+ async_extent->ram_size - 1, GFP_NOFS);
+ goto retry;
+ }
+
/*
* here we're doing allocation and writeback of the
* compressed pages
async_extent->start +
async_extent->ram_size - 1, 0);
- ret = btrfs_reserve_extent(trans, root,
- async_extent->compressed_size,
- async_extent->compressed_size,
- 0, alloc_hint,
- (u64)-1, &ins, 1);
- BUG_ON(ret);
em = alloc_extent_map(GFP_NOFS);
em->start = async_extent->start;
em->len = async_extent->ram_size;
BTRFS_ORDERED_COMPRESSED);
BUG_ON(ret);
- btrfs_end_transaction(trans, root);
-
/*
* clear dirty, set writeback and unlock the pages.
*/
extent_clear_unlock_delalloc(inode,
- &BTRFS_I(inode)->io_tree,
- async_extent->start,
- async_extent->start +
- async_extent->ram_size - 1,
- NULL, 1, 1, 0, 1, 1, 0, 0);
+ &BTRFS_I(inode)->io_tree,
+ async_extent->start,
+ async_extent->start +
+ async_extent->ram_size - 1,
+ NULL, EXTENT_CLEAR_UNLOCK_PAGE |
+ EXTENT_CLEAR_UNLOCK |
+ EXTENT_CLEAR_DELALLOC |
+ EXTENT_CLEAR_DIRTY | EXTENT_SET_WRITEBACK);
ret = btrfs_submit_compressed_write(inode,
async_extent->start,
async_extent->nr_pages);
BUG_ON(ret);
- trans = btrfs_join_transaction(root, 1);
alloc_hint = ins.objectid + ins.offset;
kfree(async_extent);
cond_resched();
}
- btrfs_end_transaction(trans, root);
return 0;
}
start, end, 0, NULL);
if (ret == 0) {
extent_clear_unlock_delalloc(inode,
- &BTRFS_I(inode)->io_tree,
- start, end, NULL, 1, 1,
- 1, 1, 1, 1, 0);
+ &BTRFS_I(inode)->io_tree,
+ start, end, NULL,
+ EXTENT_CLEAR_UNLOCK_PAGE |
+ EXTENT_CLEAR_UNLOCK |
+ EXTENT_CLEAR_DELALLOC |
+ EXTENT_CLEAR_ACCOUNTING |
+ EXTENT_CLEAR_DIRTY |
+ EXTENT_SET_WRITEBACK |
+ EXTENT_END_WRITEBACK);
+
*nr_written = *nr_written +
(end - start + PAGE_CACHE_SIZE) / PAGE_CACHE_SIZE;
*page_started = 1;
em = search_extent_mapping(&BTRFS_I(inode)->extent_tree,
start, num_bytes);
if (em) {
- alloc_hint = em->block_start;
- free_extent_map(em);
+ /*
+ * if block start isn't an actual block number then find the
+ * first block in this inode and use that as a hint. If that
+ * block is also bogus then just don't worry about it.
+ */
+ if (em->block_start >= EXTENT_MAP_LAST_BYTE) {
+ free_extent_map(em);
+ em = search_extent_mapping(em_tree, 0, 0);
+ if (em && em->block_start < EXTENT_MAP_LAST_BYTE)
+ alloc_hint = em->block_start;
+ if (em)
+ free_extent_map(em);
+ } else {
+ alloc_hint = em->block_start;
+ free_extent_map(em);
+ }
}
read_unlock(&BTRFS_I(inode)->extent_tree.lock);
btrfs_drop_extent_cache(inode, start, start + num_bytes - 1, 0);
while (disk_num_bytes > 0) {
+ unsigned long op;
+
cur_alloc_size = min(disk_num_bytes, root->fs_info->max_extent);
ret = btrfs_reserve_extent(trans, root, cur_alloc_size,
root->sectorsize, 0, alloc_hint,
* Do set the Private2 bit so we know this page was properly
* setup for writepage
*/
+ op = unlock ? EXTENT_CLEAR_UNLOCK_PAGE : 0;
+ op |= EXTENT_CLEAR_UNLOCK | EXTENT_CLEAR_DELALLOC |
+ EXTENT_SET_PRIVATE2;
+
extent_clear_unlock_delalloc(inode, &BTRFS_I(inode)->io_tree,
start, start + ram_size - 1,
- locked_page, unlock, 1,
- 1, 0, 0, 0, 1);
+ locked_page, op);
disk_num_bytes -= cur_alloc_size;
num_bytes -= cur_alloc_size;
alloc_hint = ins.objectid + ins.offset;
u64 cur_end;
int limit = 10 * 1024 * 1042;
- clear_extent_bit(&BTRFS_I(inode)->io_tree, start, end, EXTENT_LOCKED |
- EXTENT_DELALLOC, 1, 0, NULL, GFP_NOFS);
+ clear_extent_bit(&BTRFS_I(inode)->io_tree, start, end, EXTENT_LOCKED,
+ 1, 0, NULL, GFP_NOFS);
while (start < end) {
async_cow = kmalloc(sizeof(*async_cow), GFP_NOFS);
async_cow->inode = inode;
if (found_key.offset > cur_offset) {
extent_end = found_key.offset;
+ extent_type = 0;
goto out_check;
}
BUG_ON(ret);
extent_clear_unlock_delalloc(inode, &BTRFS_I(inode)->io_tree,
- cur_offset, cur_offset + num_bytes - 1,
- locked_page, 1, 1, 1, 0, 0, 0, 1);
+ cur_offset, cur_offset + num_bytes - 1,
+ locked_page, EXTENT_CLEAR_UNLOCK_PAGE |
+ EXTENT_CLEAR_UNLOCK | EXTENT_CLEAR_DELALLOC |
+ EXTENT_SET_PRIVATE2);
cur_offset = extent_end;
if (cur_offset > end)
break;
return ret;
}
+static int btrfs_split_extent_hook(struct inode *inode,
+ struct extent_state *orig, u64 split)
+{
+ struct btrfs_root *root = BTRFS_I(inode)->root;
+ u64 size;
+
+ if (!(orig->state & EXTENT_DELALLOC))
+ return 0;
+
+ size = orig->end - orig->start + 1;
+ if (size > root->fs_info->max_extent) {
+ u64 num_extents;
+ u64 new_size;
+
+ new_size = orig->end - split + 1;
+ num_extents = div64_u64(size + root->fs_info->max_extent - 1,
+ root->fs_info->max_extent);
+
+ /*
+ * if we break a large extent up then leave oustanding_extents
+ * be, since we've already accounted for the large extent.
+ */
+ if (div64_u64(new_size + root->fs_info->max_extent - 1,
+ root->fs_info->max_extent) < num_extents)
+ return 0;
+ }
+
+ spin_lock(&BTRFS_I(inode)->accounting_lock);
+ BTRFS_I(inode)->outstanding_extents++;
+ spin_unlock(&BTRFS_I(inode)->accounting_lock);
+
+ return 0;
+}
+
+/*
+ * extent_io.c merge_extent_hook, used to track merged delayed allocation
+ * extents so we can keep track of new extents that are just merged onto old
+ * extents, such as when we are doing sequential writes, so we can properly
+ * account for the metadata space we'll need.
+ */
+static int btrfs_merge_extent_hook(struct inode *inode,
+ struct extent_state *new,
+ struct extent_state *other)
+{
+ struct btrfs_root *root = BTRFS_I(inode)->root;
+ u64 new_size, old_size;
+ u64 num_extents;
+
+ /* not delalloc, ignore it */
+ if (!(other->state & EXTENT_DELALLOC))
+ return 0;
+
+ old_size = other->end - other->start + 1;
+ if (new->start < other->start)
+ new_size = other->end - new->start + 1;
+ else
+ new_size = new->end - other->start + 1;
+
+ /* we're not bigger than the max, unreserve the space and go */
+ if (new_size <= root->fs_info->max_extent) {
+ spin_lock(&BTRFS_I(inode)->accounting_lock);
+ BTRFS_I(inode)->outstanding_extents--;
+ spin_unlock(&BTRFS_I(inode)->accounting_lock);
+ return 0;
+ }
+
+ /*
+ * If we grew by another max_extent, just return, we want to keep that
+ * reserved amount.
+ */
+ num_extents = div64_u64(old_size + root->fs_info->max_extent - 1,
+ root->fs_info->max_extent);
+ if (div64_u64(new_size + root->fs_info->max_extent - 1,
+ root->fs_info->max_extent) > num_extents)
+ return 0;
+
+ spin_lock(&BTRFS_I(inode)->accounting_lock);
+ BTRFS_I(inode)->outstanding_extents--;
+ spin_unlock(&BTRFS_I(inode)->accounting_lock);
+
+ return 0;
+}
+
/*
* extent_io.c set_bit_hook, used to track delayed allocation
* bytes in this file, and to maintain the list of inodes that
static int btrfs_set_bit_hook(struct inode *inode, u64 start, u64 end,
unsigned long old, unsigned long bits)
{
+
/*
* set_bit and clear bit hooks normally require _irqsave/restore
* but in this case, we are only testeing for the DELALLOC
*/
if (!(old & EXTENT_DELALLOC) && (bits & EXTENT_DELALLOC)) {
struct btrfs_root *root = BTRFS_I(inode)->root;
+
+ spin_lock(&BTRFS_I(inode)->accounting_lock);
+ BTRFS_I(inode)->outstanding_extents++;
+ spin_unlock(&BTRFS_I(inode)->accounting_lock);
btrfs_delalloc_reserve_space(root, inode, end - start + 1);
spin_lock(&root->fs_info->delalloc_lock);
BTRFS_I(inode)->delalloc_bytes += end - start + 1;
/*
* extent_io.c clear_bit_hook, see set_bit_hook for why
*/
-static int btrfs_clear_bit_hook(struct inode *inode, u64 start, u64 end,
- unsigned long old, unsigned long bits)
+static int btrfs_clear_bit_hook(struct inode *inode,
+ struct extent_state *state, unsigned long bits)
{
/*
* set_bit and clear bit hooks normally require _irqsave/restore
* but in this case, we are only testeing for the DELALLOC
* bit, which is only set or cleared with irqs on
*/
- if ((old & EXTENT_DELALLOC) && (bits & EXTENT_DELALLOC)) {
+ if ((state->state & EXTENT_DELALLOC) && (bits & EXTENT_DELALLOC)) {
struct btrfs_root *root = BTRFS_I(inode)->root;
+ if (bits & EXTENT_DO_ACCOUNTING) {
+ spin_lock(&BTRFS_I(inode)->accounting_lock);
+ BTRFS_I(inode)->outstanding_extents--;
+ spin_unlock(&BTRFS_I(inode)->accounting_lock);
+ btrfs_unreserve_metadata_for_delalloc(root, inode, 1);
+ }
+
spin_lock(&root->fs_info->delalloc_lock);
- if (end - start + 1 > root->fs_info->delalloc_bytes) {
+ if (state->end - state->start + 1 >
+ root->fs_info->delalloc_bytes) {
printk(KERN_INFO "btrfs warning: delalloc account "
"%llu %llu\n",
- (unsigned long long)end - start + 1,
+ (unsigned long long)
+ state->end - state->start + 1,
(unsigned long long)
root->fs_info->delalloc_bytes);
btrfs_delalloc_free_space(root, inode, (u64)-1);
BTRFS_I(inode)->delalloc_bytes = 0;
} else {
btrfs_delalloc_free_space(root, inode,
- end - start + 1);
- root->fs_info->delalloc_bytes -= end - start + 1;
- BTRFS_I(inode)->delalloc_bytes -= end - start + 1;
+ state->end -
+ state->start + 1);
+ root->fs_info->delalloc_bytes -= state->end -
+ state->start + 1;
+ BTRFS_I(inode)->delalloc_bytes -= state->end -
+ state->start + 1;
}
if (BTRFS_I(inode)->delalloc_bytes == 0 &&
!list_empty(&BTRFS_I(inode)->delalloc_inodes)) {
struct inode *inode, u64 file_pos,
u64 disk_bytenr, u64 disk_num_bytes,
u64 num_bytes, u64 ram_bytes,
- u64 locked_end,
u8 compression, u8 encryption,
u16 other_encoding, int extent_type)
{
* the caller is expected to unpin it and allow it to be merged
* with the others.
*/
- ret = btrfs_drop_extents(trans, root, inode, file_pos,
- file_pos + num_bytes, locked_end,
- file_pos, &hint, 0);
+ ret = btrfs_drop_extents(trans, inode, file_pos, file_pos + num_bytes,
+ &hint, 0);
BUG_ON(ret);
ins.objectid = inode->i_ino;
}
}
- trans = btrfs_join_transaction(root, 1);
-
if (!ordered_extent)
ordered_extent = btrfs_lookup_ordered_extent(inode, start);
BUG_ON(!ordered_extent);
- if (test_bit(BTRFS_ORDERED_NOCOW, &ordered_extent->flags))
- goto nocow;
+ if (test_bit(BTRFS_ORDERED_NOCOW, &ordered_extent->flags)) {
+ BUG_ON(!list_empty(&ordered_extent->list));
+ ret = btrfs_ordered_update_i_size(inode, 0, ordered_extent);
+ if (!ret) {
+ trans = btrfs_join_transaction(root, 1);
+ ret = btrfs_update_inode(trans, root, inode);
+ BUG_ON(ret);
+ btrfs_end_transaction(trans, root);
+ }
+ goto out;
+ }
lock_extent(io_tree, ordered_extent->file_offset,
ordered_extent->file_offset + ordered_extent->len - 1,
GFP_NOFS);
+ trans = btrfs_join_transaction(root, 1);
+
if (test_bit(BTRFS_ORDERED_COMPRESSED, &ordered_extent->flags))
compressed = 1;
if (test_bit(BTRFS_ORDERED_PREALLOC, &ordered_extent->flags)) {
BUG_ON(compressed);
- ret = btrfs_mark_extent_written(trans, root, inode,
+ ret = btrfs_mark_extent_written(trans, inode,
ordered_extent->file_offset,
ordered_extent->file_offset +
ordered_extent->len);
ordered_extent->disk_len,
ordered_extent->len,
ordered_extent->len,
- ordered_extent->file_offset +
- ordered_extent->len,
compressed, 0, 0,
BTRFS_FILE_EXTENT_REG);
unpin_extent_cache(&BTRFS_I(inode)->extent_tree,
unlock_extent(io_tree, ordered_extent->file_offset,
ordered_extent->file_offset + ordered_extent->len - 1,
GFP_NOFS);
-nocow:
add_pending_csums(trans, inode, ordered_extent->file_offset,
&ordered_extent->list);
- mutex_lock(&BTRFS_I(inode)->extent_mutex);
- btrfs_ordered_update_i_size(inode, ordered_extent);
- btrfs_update_inode(trans, root, inode);
- btrfs_remove_ordered_extent(inode, ordered_extent);
- mutex_unlock(&BTRFS_I(inode)->extent_mutex);
-
+ /* this also removes the ordered extent from the tree */
+ btrfs_ordered_update_i_size(inode, 0, ordered_extent);
+ ret = btrfs_update_inode(trans, root, inode);
+ BUG_ON(ret);
+ btrfs_end_transaction(trans, root);
+out:
/* once for us */
btrfs_put_ordered_extent(ordered_extent);
/* once for the tree */
btrfs_put_ordered_extent(ordered_extent);
- btrfs_end_transaction(trans, root);
return 0;
}
return -EIO;
}
+struct delayed_iput {
+ struct list_head list;
+ struct inode *inode;
+};
+
+void btrfs_add_delayed_iput(struct inode *inode)
+{
+ struct btrfs_fs_info *fs_info = BTRFS_I(inode)->root->fs_info;
+ struct delayed_iput *delayed;
+
+ if (atomic_add_unless(&inode->i_count, -1, 1))
+ return;
+
+ delayed = kmalloc(sizeof(*delayed), GFP_NOFS | __GFP_NOFAIL);
+ delayed->inode = inode;
+
+ spin_lock(&fs_info->delayed_iput_lock);
+ list_add_tail(&delayed->list, &fs_info->delayed_iputs);
+ spin_unlock(&fs_info->delayed_iput_lock);
+}
+
+void btrfs_run_delayed_iputs(struct btrfs_root *root)
+{
+ LIST_HEAD(list);
+ struct btrfs_fs_info *fs_info = root->fs_info;
+ struct delayed_iput *delayed;
+ int empty;
+
+ spin_lock(&fs_info->delayed_iput_lock);
+ empty = list_empty(&fs_info->delayed_iputs);
+ spin_unlock(&fs_info->delayed_iput_lock);
+ if (empty)
+ return;
+
+ down_read(&root->fs_info->cleanup_work_sem);
+ spin_lock(&fs_info->delayed_iput_lock);
+ list_splice_init(&fs_info->delayed_iputs, &list);
+ spin_unlock(&fs_info->delayed_iput_lock);
+
+ while (!list_empty(&list)) {
+ delayed = list_entry(list.next, struct delayed_iput, list);
+ list_del(&delayed->list);
+ iput(delayed->inode);
+ kfree(delayed);
+ }
+ up_read(&root->fs_info->cleanup_work_sem);
+}
+
/*
* This creates an orphan entry for the given inode in case something goes
* wrong in the middle of an unlink/truncate.
struct inode *inode;
int ret = 0, nr_unlink = 0, nr_truncate = 0;
- path = btrfs_alloc_path();
- if (!path)
+ if (!xchg(&root->clean_orphans, 0))
return;
+
+ path = btrfs_alloc_path();
+ BUG_ON(!path);
path->reada = -1;
key.objectid = BTRFS_ORPHAN_OBJECTID;
btrfs_set_key_type(&key, BTRFS_ORPHAN_ITEM_KEY);
key.offset = (u64)-1;
-
while (1) {
ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
if (ret < 0) {
root = BTRFS_I(dir)->root;
+ /*
+ * 5 items for unlink inode
+ * 1 for orphan
+ */
+ ret = btrfs_reserve_metadata_space(root, 6);
+ if (ret)
+ return ret;
+
trans = btrfs_start_transaction(root, 1);
+ if (IS_ERR(trans)) {
+ btrfs_unreserve_metadata_space(root, 6);
+ return PTR_ERR(trans);
+ }
btrfs_set_trans_block_group(trans, dir);
nr = trans->blocks_used;
btrfs_end_transaction_throttle(trans, root);
+ btrfs_unreserve_metadata_space(root, 6);
btrfs_btree_balance_dirty(root, nr);
return ret;
}
inode->i_ino == BTRFS_FIRST_FREE_OBJECTID)
return -ENOTEMPTY;
+ ret = btrfs_reserve_metadata_space(root, 5);
+ if (ret)
+ return ret;
+
trans = btrfs_start_transaction(root, 1);
+ if (IS_ERR(trans)) {
+ btrfs_unreserve_metadata_space(root, 5);
+ return PTR_ERR(trans);
+ }
+
btrfs_set_trans_block_group(trans, dir);
if (unlikely(inode->i_ino == BTRFS_EMPTY_SUBVOL_DIR_OBJECTID)) {
out:
nr = trans->blocks_used;
ret = btrfs_end_transaction_throttle(trans, root);
+ btrfs_unreserve_metadata_space(root, 5);
btrfs_btree_balance_dirty(root, nr);
if (ret && !err)
* min_type is the minimum key type to truncate down to. If set to 0, this
* will kill all the items on this inode, including the INODE_ITEM_KEY.
*/
-noinline int btrfs_truncate_inode_items(struct btrfs_trans_handle *trans,
- struct btrfs_root *root,
- struct inode *inode,
- u64 new_size, u32 min_type)
+int btrfs_truncate_inode_items(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root,
+ struct inode *inode,
+ u64 new_size, u32 min_type)
{
- int ret;
struct btrfs_path *path;
- struct btrfs_key key;
- struct btrfs_key found_key;
- u32 found_type = (u8)-1;
struct extent_buffer *leaf;
struct btrfs_file_extent_item *fi;
+ struct btrfs_key key;
+ struct btrfs_key found_key;
u64 extent_start = 0;
u64 extent_num_bytes = 0;
u64 extent_offset = 0;
u64 item_end = 0;
+ u64 mask = root->sectorsize - 1;
+ u32 found_type = (u8)-1;
int found_extent;
int del_item;
int pending_del_nr = 0;
int pending_del_slot = 0;
int extent_type = -1;
int encoding;
- u64 mask = root->sectorsize - 1;
+ int ret;
+ int err = 0;
+
+ BUG_ON(new_size > 0 && min_type != BTRFS_EXTENT_DATA_KEY);
if (root->ref_cows)
btrfs_drop_extent_cache(inode, new_size & (~mask), (u64)-1, 0);
+
path = btrfs_alloc_path();
BUG_ON(!path);
path->reada = -1;
- /* FIXME, add redo link to tree so we don't leak on crash */
key.objectid = inode->i_ino;
key.offset = (u64)-1;
key.type = (u8)-1;
search_again:
path->leave_spinning = 1;
ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
- if (ret < 0)
- goto error;
+ if (ret < 0) {
+ err = ret;
+ goto out;
+ }
if (ret > 0) {
/* there are no items in the tree for us to truncate, we're
* done
*/
- if (path->slots[0] == 0) {
- ret = 0;
- goto error;
- }
+ if (path->slots[0] == 0)
+ goto out;
path->slots[0]--;
}
}
item_end--;
}
- if (item_end < new_size) {
- if (found_type == BTRFS_DIR_ITEM_KEY)
- found_type = BTRFS_INODE_ITEM_KEY;
- else if (found_type == BTRFS_EXTENT_ITEM_KEY)
- found_type = BTRFS_EXTENT_DATA_KEY;
- else if (found_type == BTRFS_EXTENT_DATA_KEY)
- found_type = BTRFS_XATTR_ITEM_KEY;
- else if (found_type == BTRFS_XATTR_ITEM_KEY)
- found_type = BTRFS_INODE_REF_KEY;
- else if (found_type)
- found_type--;
- else
+ if (found_type > min_type) {
+ del_item = 1;
+ } else {
+ if (item_end < new_size)
break;
- btrfs_set_key_type(&key, found_type);
- goto next;
+ if (found_key.offset >= new_size)
+ del_item = 1;
+ else
+ del_item = 0;
}
- if (found_key.offset >= new_size)
- del_item = 1;
- else
- del_item = 0;
found_extent = 0;
-
/* FIXME, shrink the extent if the ref count is only 1 */
if (found_type != BTRFS_EXTENT_DATA_KEY)
goto delete;
inode->i_ino, extent_offset);
BUG_ON(ret);
}
-next:
- if (path->slots[0] == 0) {
- if (pending_del_nr)
- goto del_pending;
- btrfs_release_path(root, path);
- if (found_type == BTRFS_INODE_ITEM_KEY)
- break;
- goto search_again;
- }
- path->slots[0]--;
- if (pending_del_nr &&
- path->slots[0] + 1 != pending_del_slot) {
- struct btrfs_key debug;
-del_pending:
- btrfs_item_key_to_cpu(path->nodes[0], &debug,
- pending_del_slot);
- ret = btrfs_del_items(trans, root, path,
- pending_del_slot,
- pending_del_nr);
- BUG_ON(ret);
- pending_del_nr = 0;
+ if (found_type == BTRFS_INODE_ITEM_KEY)
+ break;
+
+ if (path->slots[0] == 0 ||
+ path->slots[0] != pending_del_slot) {
+ if (root->ref_cows) {
+ err = -EAGAIN;
+ goto out;
+ }
+ if (pending_del_nr) {
+ ret = btrfs_del_items(trans, root, path,
+ pending_del_slot,
+ pending_del_nr);
+ BUG_ON(ret);
+ pending_del_nr = 0;
+ }
btrfs_release_path(root, path);
- if (found_type == BTRFS_INODE_ITEM_KEY)
- break;
goto search_again;
+ } else {
+ path->slots[0]--;
}
}
- ret = 0;
-error:
+out:
if (pending_del_nr) {
ret = btrfs_del_items(trans, root, path, pending_del_slot,
pending_del_nr);
}
btrfs_free_path(path);
- return ret;
+ return err;
}
/*
if ((offset & (blocksize - 1)) == 0)
goto out;
+ ret = btrfs_check_data_free_space(root, inode, PAGE_CACHE_SIZE);
+ if (ret)
+ goto out;
+
+ ret = btrfs_reserve_metadata_for_delalloc(root, inode, 1);
+ if (ret)
+ goto out;
ret = -ENOMEM;
again:
page = grab_cache_page(mapping, index);
- if (!page)
+ if (!page) {
+ btrfs_free_reserved_data_space(root, inode, PAGE_CACHE_SIZE);
+ btrfs_unreserve_metadata_for_delalloc(root, inode, 1);
goto out;
+ }
page_start = page_offset(page);
page_end = page_start + PAGE_CACHE_SIZE - 1;
goto again;
}
- btrfs_set_extent_delalloc(inode, page_start, page_end);
+ clear_extent_bits(&BTRFS_I(inode)->io_tree, page_start, page_end,
+ EXTENT_DIRTY | EXTENT_DELALLOC | EXTENT_DO_ACCOUNTING,
+ GFP_NOFS);
+
+ ret = btrfs_set_extent_delalloc(inode, page_start, page_end);
+ if (ret) {
+ unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
+ goto out_unlock;
+ }
+
ret = 0;
if (offset != PAGE_CACHE_SIZE) {
kaddr = kmap(page);
unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
out_unlock:
+ if (ret)
+ btrfs_free_reserved_data_space(root, inode, PAGE_CACHE_SIZE);
+ btrfs_unreserve_metadata_for_delalloc(root, inode, 1);
unlock_page(page);
page_cache_release(page);
out:
u64 last_byte;
u64 cur_offset;
u64 hole_size;
- int err;
+ int err = 0;
if (size <= hole_start)
return 0;
- err = btrfs_check_metadata_free_space(root);
- if (err)
- return err;
-
- btrfs_truncate_page(inode->i_mapping, inode->i_size);
-
while (1) {
struct btrfs_ordered_extent *ordered;
btrfs_wait_ordered_range(inode, hole_start,
btrfs_put_ordered_extent(ordered);
}
- trans = btrfs_start_transaction(root, 1);
- btrfs_set_trans_block_group(trans, inode);
-
cur_offset = hole_start;
while (1) {
em = btrfs_get_extent(inode, NULL, 0, cur_offset,
BUG_ON(IS_ERR(em) || !em);
last_byte = min(extent_map_end(em), block_end);
last_byte = (last_byte + mask) & ~mask;
- if (test_bit(EXTENT_FLAG_VACANCY, &em->flags)) {
+ if (!test_bit(EXTENT_FLAG_PREALLOC, &em->flags)) {
u64 hint_byte = 0;
hole_size = last_byte - cur_offset;
- err = btrfs_drop_extents(trans, root, inode,
- cur_offset,
- cur_offset + hole_size,
- block_end,
- cur_offset, &hint_byte, 1);
+
+ err = btrfs_reserve_metadata_space(root, 2);
if (err)
break;
+
+ trans = btrfs_start_transaction(root, 1);
+ btrfs_set_trans_block_group(trans, inode);
+
+ err = btrfs_drop_extents(trans, inode, cur_offset,
+ cur_offset + hole_size,
+ &hint_byte, 1);
+ BUG_ON(err);
+
err = btrfs_insert_file_extent(trans, root,
inode->i_ino, cur_offset, 0,
0, hole_size, 0, hole_size,
0, 0, 0);
+ BUG_ON(err);
+
btrfs_drop_extent_cache(inode, hole_start,
last_byte - 1, 0);
+
+ btrfs_end_transaction(trans, root);
+ btrfs_unreserve_metadata_space(root, 2);
}
free_extent_map(em);
cur_offset = last_byte;
- if (err || cur_offset >= block_end)
+ if (cur_offset >= block_end)
break;
}
- btrfs_end_transaction(trans, root);
unlock_extent(io_tree, hole_start, block_end - 1, GFP_NOFS);
return err;
}
+static int btrfs_setattr_size(struct inode *inode, struct iattr *attr)
+{
+ struct btrfs_root *root = BTRFS_I(inode)->root;
+ struct btrfs_trans_handle *trans;
+ unsigned long nr;
+ int ret;
+
+ if (attr->ia_size == inode->i_size)
+ return 0;
+
+ if (attr->ia_size > inode->i_size) {
+ unsigned long limit;
+ limit = current->signal->rlim[RLIMIT_FSIZE].rlim_cur;
+ if (attr->ia_size > inode->i_sb->s_maxbytes)
+ return -EFBIG;
+ if (limit != RLIM_INFINITY && attr->ia_size > limit) {
+ send_sig(SIGXFSZ, current, 0);
+ return -EFBIG;
+ }
+ }
+
+ ret = btrfs_reserve_metadata_space(root, 1);
+ if (ret)
+ return ret;
+
+ trans = btrfs_start_transaction(root, 1);
+ btrfs_set_trans_block_group(trans, inode);
+
+ ret = btrfs_orphan_add(trans, inode);
+ BUG_ON(ret);
+
+ nr = trans->blocks_used;
+ btrfs_end_transaction(trans, root);
+ btrfs_unreserve_metadata_space(root, 1);
+ btrfs_btree_balance_dirty(root, nr);
+
+ if (attr->ia_size > inode->i_size) {
+ ret = btrfs_cont_expand(inode, attr->ia_size);
+ if (ret) {
+ btrfs_truncate(inode);
+ return ret;
+ }
+
+ i_size_write(inode, attr->ia_size);
+ btrfs_ordered_update_i_size(inode, inode->i_size, NULL);
+
+ trans = btrfs_start_transaction(root, 1);
+ btrfs_set_trans_block_group(trans, inode);
+
+ ret = btrfs_update_inode(trans, root, inode);
+ BUG_ON(ret);
+ if (inode->i_nlink > 0) {
+ ret = btrfs_orphan_del(trans, inode);
+ BUG_ON(ret);
+ }
+ nr = trans->blocks_used;
+ btrfs_end_transaction(trans, root);
+ btrfs_btree_balance_dirty(root, nr);
+ return 0;
+ }
+
+ /*
+ * We're truncating a file that used to have good data down to
+ * zero. Make sure it gets into the ordered flush list so that
+ * any new writes get down to disk quickly.
+ */
+ if (attr->ia_size == 0)
+ BTRFS_I(inode)->ordered_data_close = 1;
+
+ /* we don't support swapfiles, so vmtruncate shouldn't fail */
+ ret = vmtruncate(inode, attr->ia_size);
+ BUG_ON(ret);
+
+ return 0;
+}
+
static int btrfs_setattr(struct dentry *dentry, struct iattr *attr)
{
struct inode *inode = dentry->d_inode;
return err;
if (S_ISREG(inode->i_mode) && (attr->ia_valid & ATTR_SIZE)) {
- if (attr->ia_size > inode->i_size) {
- err = btrfs_cont_expand(inode, attr->ia_size);
- if (err)
- return err;
- } else if (inode->i_size > 0 &&
- attr->ia_size == 0) {
-
- /* we're truncating a file that used to have good
- * data down to zero. Make sure it gets into
- * the ordered flush list so that any new writes
- * get down to disk quickly.
- */
- BTRFS_I(inode)->ordered_data_close = 1;
- }
+ err = btrfs_setattr_size(inode, attr);
+ if (err)
+ return err;
}
+ attr->ia_valid &= ~ATTR_SIZE;
- err = inode_setattr(inode, attr);
+ if (attr->ia_valid)
+ err = inode_setattr(inode, attr);
if (!err && ((attr->ia_valid & ATTR_MODE)))
err = btrfs_acl_chmod(inode);
}
btrfs_wait_ordered_range(inode, 0, (u64)-1);
+ if (root->fs_info->log_root_recovering) {
+ BUG_ON(!list_empty(&BTRFS_I(inode)->i_orphan));
+ goto no_delete;
+ }
+
if (inode->i_nlink > 0) {
BUG_ON(btrfs_root_refs(&root->root_item) != 0);
goto no_delete;
}
btrfs_i_size_write(inode, 0);
- trans = btrfs_join_transaction(root, 1);
- btrfs_set_trans_block_group(trans, inode);
- ret = btrfs_truncate_inode_items(trans, root, inode, inode->i_size, 0);
- if (ret) {
- btrfs_orphan_del(NULL, inode);
- goto no_delete_lock;
- }
+ while (1) {
+ trans = btrfs_start_transaction(root, 1);
+ btrfs_set_trans_block_group(trans, inode);
+ ret = btrfs_truncate_inode_items(trans, root, inode, 0, 0);
- btrfs_orphan_del(trans, inode);
+ if (ret != -EAGAIN)
+ break;
- nr = trans->blocks_used;
- clear_inode(inode);
+ nr = trans->blocks_used;
+ btrfs_end_transaction(trans, root);
+ trans = NULL;
+ btrfs_btree_balance_dirty(root, nr);
+ }
- btrfs_end_transaction(trans, root);
- btrfs_btree_balance_dirty(root, nr);
- return;
+ if (ret == 0) {
+ ret = btrfs_orphan_del(trans, inode);
+ BUG_ON(ret);
+ }
-no_delete_lock:
nr = trans->blocks_used;
btrfs_end_transaction(trans, root);
btrfs_btree_balance_dirty(root, nr);
no_delete:
clear_inode(inode);
+ return;
}
/*
bi->generation = 0;
bi->sequence = 0;
bi->last_trans = 0;
+ bi->last_sub_trans = 0;
bi->logged_trans = 0;
bi->delalloc_bytes = 0;
bi->reserved_bytes = 0;
INIT_LIST_HEAD(&BTRFS_I(inode)->ordered_operations);
RB_CLEAR_NODE(&BTRFS_I(inode)->rb_node);
btrfs_ordered_inode_tree_init(&BTRFS_I(inode)->ordered_tree);
- mutex_init(&BTRFS_I(inode)->extent_mutex);
mutex_init(&BTRFS_I(inode)->log_mutex);
}
}
srcu_read_unlock(&root->fs_info->subvol_srcu, index);
+ if (root != sub_root) {
+ down_read(&root->fs_info->cleanup_work_sem);
+ if (!(inode->i_sb->s_flags & MS_RDONLY))
+ btrfs_orphan_cleanup(sub_root);
+ up_read(&root->fs_info->cleanup_work_sem);
+ }
+
return inode;
}
{
struct btrfs_root *root;
- if (!dentry->d_inode)
- return 0;
+ if (!dentry->d_inode && !IS_ROOT(dentry))
+ dentry = dentry->d_parent;
- root = BTRFS_I(dentry->d_inode)->root;
- if (btrfs_root_refs(&root->root_item) == 0)
- return 1;
+ if (dentry->d_inode) {
+ root = BTRFS_I(dentry->d_inode)->root;
+ if (btrfs_root_refs(&root->root_item) == 0)
+ return 1;
+ }
return 0;
}
/* Reached end of directory/root. Bump pos past the last item. */
if (key_type == BTRFS_DIR_INDEX_KEY)
- filp->f_pos = INT_LIMIT(off_t);
+ /*
+ * 32-bit glibc will use getdents64, but then strtol -
+ * so the last number we can serve is this.
+ */
+ filp->f_pos = 0x7fffffff;
else
filp->f_pos++;
nopos:
if (!new_valid_dev(rdev))
return -EINVAL;
- err = btrfs_check_metadata_free_space(root);
+ /*
+ * 2 for inode item and ref
+ * 2 for dir items
+ * 1 for xattr if selinux is on
+ */
+ err = btrfs_reserve_metadata_space(root, 5);
if (err)
- goto fail;
+ return err;
trans = btrfs_start_transaction(root, 1);
+ if (!trans)
+ goto fail;
btrfs_set_trans_block_group(trans, dir);
err = btrfs_find_free_objectid(trans, root, dir->i_ino, &objectid);
if (IS_ERR(inode))
goto out_unlock;
- err = btrfs_init_inode_security(inode, dir);
+ err = btrfs_init_inode_security(trans, inode, dir);
if (err) {
drop_inode = 1;
goto out_unlock;
nr = trans->blocks_used;
btrfs_end_transaction_throttle(trans, root);
fail:
+ btrfs_unreserve_metadata_space(root, 5);
if (drop_inode) {
inode_dec_link_count(inode);
iput(inode);
u64 objectid;
u64 index = 0;
- err = btrfs_check_metadata_free_space(root);
+ /*
+ * 2 for inode item and ref
+ * 2 for dir items
+ * 1 for xattr if selinux is on
+ */
+ err = btrfs_reserve_metadata_space(root, 5);
if (err)
- goto fail;
+ return err;
+
trans = btrfs_start_transaction(root, 1);
+ if (!trans)
+ goto fail;
btrfs_set_trans_block_group(trans, dir);
err = btrfs_find_free_objectid(trans, root, dir->i_ino, &objectid);
if (IS_ERR(inode))
goto out_unlock;
- err = btrfs_init_inode_security(inode, dir);
+ err = btrfs_init_inode_security(trans, inode, dir);
if (err) {
drop_inode = 1;
goto out_unlock;
nr = trans->blocks_used;
btrfs_end_transaction_throttle(trans, root);
fail:
+ btrfs_unreserve_metadata_space(root, 5);
if (drop_inode) {
inode_dec_link_count(inode);
iput(inode);
if (inode->i_nlink == 0)
return -ENOENT;
- btrfs_inc_nlink(inode);
- err = btrfs_check_metadata_free_space(root);
+ /* do not allow sys_link's with other subvols of the same device */
+ if (root->objectid != BTRFS_I(inode)->root->objectid)
+ return -EPERM;
+
+ /*
+ * 1 item for inode ref
+ * 2 items for dir items
+ */
+ err = btrfs_reserve_metadata_space(root, 3);
if (err)
- goto fail;
+ return err;
+
+ btrfs_inc_nlink(inode);
+
err = btrfs_set_inode_index(dir, &index);
if (err)
goto fail;
nr = trans->blocks_used;
btrfs_end_transaction_throttle(trans, root);
fail:
+ btrfs_unreserve_metadata_space(root, 3);
if (drop_inode) {
inode_dec_link_count(inode);
iput(inode);
u64 index = 0;
unsigned long nr = 1;
- err = btrfs_check_metadata_free_space(root);
+ /*
+ * 2 items for inode and ref
+ * 2 items for dir items
+ * 1 for xattr if selinux is on
+ */
+ err = btrfs_reserve_metadata_space(root, 5);
if (err)
- goto out_unlock;
+ return err;
trans = btrfs_start_transaction(root, 1);
- btrfs_set_trans_block_group(trans, dir);
-
- if (IS_ERR(trans)) {
- err = PTR_ERR(trans);
+ if (!trans) {
+ err = -ENOMEM;
goto out_unlock;
}
+ btrfs_set_trans_block_group(trans, dir);
err = btrfs_find_free_objectid(trans, root, dir->i_ino, &objectid);
if (err) {
drop_on_err = 1;
- err = btrfs_init_inode_security(inode, dir);
+ err = btrfs_init_inode_security(trans, inode, dir);
if (err)
goto out_fail;
btrfs_end_transaction_throttle(trans, root);
out_unlock:
+ btrfs_unreserve_metadata_space(root, 5);
if (drop_on_err)
iput(inode);
btrfs_btree_balance_dirty(root, nr);
*/
clear_extent_bit(tree, page_start, page_end,
EXTENT_DIRTY | EXTENT_DELALLOC |
- EXTENT_LOCKED, 1, 0, NULL, GFP_NOFS);
+ EXTENT_LOCKED | EXTENT_DO_ACCOUNTING, 1, 0,
+ NULL, GFP_NOFS);
/*
* whoever cleared the private bit is responsible
* for the finish_ordered_io
lock_extent(tree, page_start, page_end, GFP_NOFS);
}
clear_extent_bit(tree, page_start, page_end,
- EXTENT_LOCKED | EXTENT_DIRTY | EXTENT_DELALLOC,
- 1, 1, NULL, GFP_NOFS);
+ EXTENT_LOCKED | EXTENT_DIRTY | EXTENT_DELALLOC |
+ EXTENT_DO_ACCOUNTING, 1, 1, NULL, GFP_NOFS);
__btrfs_releasepage(page, GFP_NOFS);
ClearPageChecked(page);
goto out;
}
+ ret = btrfs_reserve_metadata_for_delalloc(root, inode, 1);
+ if (ret) {
+ btrfs_free_reserved_data_space(root, inode, PAGE_CACHE_SIZE);
+ ret = VM_FAULT_SIGBUS;
+ goto out;
+ }
+
ret = VM_FAULT_NOPAGE; /* make the VM retry the fault */
again:
lock_page(page);
goto again;
}
- btrfs_set_extent_delalloc(inode, page_start, page_end);
+ /*
+ * XXX - page_mkwrite gets called every time the page is dirtied, even
+ * if it was already dirty, so for space accounting reasons we need to
+ * clear any delalloc bits for the range we are fixing to save. There
+ * is probably a better way to do this, but for now keep consistent with
+ * prepare_pages in the normal write path.
+ */
+ clear_extent_bits(&BTRFS_I(inode)->io_tree, page_start, page_end,
+ EXTENT_DIRTY | EXTENT_DELALLOC | EXTENT_DO_ACCOUNTING,
+ GFP_NOFS);
+
+ ret = btrfs_set_extent_delalloc(inode, page_start, page_end);
+ if (ret) {
+ unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
+ ret = VM_FAULT_SIGBUS;
+ btrfs_free_reserved_data_space(root, inode, PAGE_CACHE_SIZE);
+ goto out_unlock;
+ }
ret = 0;
/* page is wholly or partially inside EOF */
set_page_dirty(page);
SetPageUptodate(page);
- BTRFS_I(inode)->last_trans = root->fs_info->generation + 1;
+ BTRFS_I(inode)->last_trans = root->fs_info->generation;
+ BTRFS_I(inode)->last_sub_trans = BTRFS_I(inode)->root->log_transid;
+
unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
out_unlock:
+ btrfs_unreserve_metadata_for_delalloc(root, inode, 1);
if (!ret)
return VM_FAULT_LOCKED;
unlock_page(page);
unsigned long nr;
u64 mask = root->sectorsize - 1;
- if (!S_ISREG(inode->i_mode))
+ if (!S_ISREG(inode->i_mode)) {
+ WARN_ON(1);
return;
- if (IS_APPEND(inode) || IS_IMMUTABLE(inode))
+ }
+
+ ret = btrfs_truncate_page(inode->i_mapping, inode->i_size);
+ if (ret)
return;
- btrfs_truncate_page(inode->i_mapping, inode->i_size);
btrfs_wait_ordered_range(inode, inode->i_size & (~mask), (u64)-1);
+ btrfs_ordered_update_i_size(inode, inode->i_size, NULL);
trans = btrfs_start_transaction(root, 1);
+ btrfs_set_trans_block_group(trans, inode);
/*
* setattr is responsible for setting the ordered_data_close flag,
if (inode->i_size == 0 && BTRFS_I(inode)->ordered_data_close)
btrfs_add_ordered_operation(trans, root, inode);
- btrfs_set_trans_block_group(trans, inode);
- btrfs_i_size_write(inode, inode->i_size);
+ while (1) {
+ ret = btrfs_truncate_inode_items(trans, root, inode,
+ inode->i_size,
+ BTRFS_EXTENT_DATA_KEY);
+ if (ret != -EAGAIN)
+ break;
- ret = btrfs_orphan_add(trans, inode);
- if (ret)
- goto out;
- /* FIXME, add redo link to tree so we don't leak on crash */
- ret = btrfs_truncate_inode_items(trans, root, inode, inode->i_size,
- BTRFS_EXTENT_DATA_KEY);
- btrfs_update_inode(trans, root, inode);
+ ret = btrfs_update_inode(trans, root, inode);
+ BUG_ON(ret);
+
+ nr = trans->blocks_used;
+ btrfs_end_transaction(trans, root);
+ btrfs_btree_balance_dirty(root, nr);
+
+ trans = btrfs_start_transaction(root, 1);
+ btrfs_set_trans_block_group(trans, inode);
+ }
- ret = btrfs_orphan_del(trans, inode);
+ if (ret == 0 && inode->i_nlink > 0) {
+ ret = btrfs_orphan_del(trans, inode);
+ BUG_ON(ret);
+ }
+
+ ret = btrfs_update_inode(trans, root, inode);
BUG_ON(ret);
-out:
nr = trans->blocks_used;
ret = btrfs_end_transaction_throttle(trans, root);
BUG_ON(ret);
if (!ei)
return NULL;
ei->last_trans = 0;
+ ei->last_sub_trans = 0;
ei->logged_trans = 0;
+ ei->outstanding_extents = 0;
+ ei->reserved_extents = 0;
+ ei->root = NULL;
+ spin_lock_init(&ei->accounting_lock);
btrfs_ordered_inode_tree_init(&ei->ordered_tree);
INIT_LIST_HEAD(&ei->i_orphan);
INIT_LIST_HEAD(&ei->ordered_operations);
WARN_ON(inode->i_data.nrpages);
/*
+ * This can happen where we create an inode, but somebody else also
+ * created the same inode and we need to destroy the one we already
+ * created.
+ */
+ if (!root)
+ goto free;
+
+ /*
* Make sure we're properly removed from the ordered operation
* lists.
*/
spin_lock(&root->list_lock);
if (!list_empty(&BTRFS_I(inode)->i_orphan)) {
- printk(KERN_ERR "BTRFS: inode %lu: inode still on the orphan"
- " list\n", inode->i_ino);
- dump_stack();
+ printk(KERN_INFO "BTRFS: inode %lu still on the orphan list\n",
+ inode->i_ino);
+ list_del_init(&BTRFS_I(inode)->i_orphan);
}
spin_unlock(&root->list_lock);
}
inode_tree_del(inode);
btrfs_drop_extent_cache(inode, 0, (u64)-1, 0);
+free:
kmem_cache_free(btrfs_inode_cachep, BTRFS_I(inode));
}
new_inode->i_size > BTRFS_EMPTY_DIR_SIZE)
return -ENOTEMPTY;
- ret = btrfs_check_metadata_free_space(root);
+ /*
+ * We want to reserve the absolute worst case amount of items. So if
+ * both inodes are subvols and we need to unlink them then that would
+ * require 4 item modifications, but if they are both normal inodes it
+ * would require 5 item modifications, so we'll assume their normal
+ * inodes. So 5 * 2 is 10, plus 1 for the new link, so 11 total items
+ * should cover the worst case number of items we'll modify.
+ */
+ ret = btrfs_reserve_metadata_space(root, 11);
if (ret)
return ret;
if (old_inode->i_ino == BTRFS_FIRST_FREE_OBJECTID)
up_read(&root->fs_info->subvol_sem);
+
+ btrfs_unreserve_metadata_space(root, 11);
return ret;
}
* some fairly slow code that needs optimization. This walks the list
* of all the inodes with pending delalloc and forces them to disk.
*/
-int btrfs_start_delalloc_inodes(struct btrfs_root *root)
+int btrfs_start_delalloc_inodes(struct btrfs_root *root, int delay_iput)
{
struct list_head *head = &root->fs_info->delalloc_inodes;
struct btrfs_inode *binode;
spin_unlock(&root->fs_info->delalloc_lock);
if (inode) {
filemap_flush(inode->i_mapping);
- iput(inode);
+ if (delay_iput)
+ btrfs_add_delayed_iput(inode);
+ else
+ iput(inode);
}
cond_resched();
spin_lock(&root->fs_info->delalloc_lock);
if (name_len > BTRFS_MAX_INLINE_DATA_SIZE(root))
return -ENAMETOOLONG;
- err = btrfs_check_metadata_free_space(root);
+ /*
+ * 2 items for inode item and ref
+ * 2 items for dir items
+ * 1 item for xattr if selinux is on
+ */
+ err = btrfs_reserve_metadata_space(root, 5);
if (err)
- goto out_fail;
+ return err;
trans = btrfs_start_transaction(root, 1);
+ if (!trans)
+ goto out_fail;
btrfs_set_trans_block_group(trans, dir);
err = btrfs_find_free_objectid(trans, root, dir->i_ino, &objectid);
if (IS_ERR(inode))
goto out_unlock;
- err = btrfs_init_inode_security(inode, dir);
+ err = btrfs_init_inode_security(trans, inode, dir);
if (err) {
drop_inode = 1;
goto out_unlock;
nr = trans->blocks_used;
btrfs_end_transaction_throttle(trans, root);
out_fail:
+ btrfs_unreserve_metadata_space(root, 5);
if (drop_inode) {
inode_dec_link_count(inode);
iput(inode);
return err;
}
-static int prealloc_file_range(struct btrfs_trans_handle *trans,
- struct inode *inode, u64 start, u64 end,
- u64 locked_end, u64 alloc_hint, int mode)
+static int prealloc_file_range(struct inode *inode, u64 start, u64 end,
+ u64 alloc_hint, int mode)
{
+ struct btrfs_trans_handle *trans;
struct btrfs_root *root = BTRFS_I(inode)->root;
struct btrfs_key ins;
u64 alloc_size;
while (num_bytes > 0) {
alloc_size = min(num_bytes, root->fs_info->max_extent);
+
+ trans = btrfs_start_transaction(root, 1);
+
ret = btrfs_reserve_extent(trans, root, alloc_size,
root->sectorsize, 0, alloc_hint,
(u64)-1, &ins, 1);
if (ret) {
WARN_ON(1);
- goto out;
+ goto stop_trans;
+ }
+
+ ret = btrfs_reserve_metadata_space(root, 3);
+ if (ret) {
+ btrfs_free_reserved_extent(root, ins.objectid,
+ ins.offset);
+ goto stop_trans;
}
+
ret = insert_reserved_file_extent(trans, inode,
cur_offset, ins.objectid,
ins.offset, ins.offset,
- ins.offset, locked_end,
- 0, 0, 0,
+ ins.offset, 0, 0, 0,
BTRFS_FILE_EXTENT_PREALLOC);
BUG_ON(ret);
btrfs_drop_extent_cache(inode, cur_offset,
cur_offset + ins.offset -1, 0);
+
num_bytes -= ins.offset;
cur_offset += ins.offset;
alloc_hint = ins.objectid + ins.offset;
- }
-out:
- if (cur_offset > start) {
+
inode->i_ctime = CURRENT_TIME;
BTRFS_I(inode)->flags |= BTRFS_INODE_PREALLOC;
if (!(mode & FALLOC_FL_KEEP_SIZE) &&
- cur_offset > i_size_read(inode))
- btrfs_i_size_write(inode, cur_offset);
+ cur_offset > inode->i_size) {
+ i_size_write(inode, cur_offset);
+ btrfs_ordered_update_i_size(inode, cur_offset, NULL);
+ }
+
ret = btrfs_update_inode(trans, root, inode);
BUG_ON(ret);
+
+ btrfs_end_transaction(trans, root);
+ btrfs_unreserve_metadata_space(root, 3);
}
+ return ret;
+stop_trans:
+ btrfs_end_transaction(trans, root);
return ret;
+
}
static long btrfs_fallocate(struct inode *inode, int mode,
u64 locked_end;
u64 mask = BTRFS_I(inode)->root->sectorsize - 1;
struct extent_map *em;
- struct btrfs_trans_handle *trans;
- struct btrfs_root *root;
int ret;
alloc_start = offset & ~mask;
goto out;
}
- root = BTRFS_I(inode)->root;
-
- ret = btrfs_check_data_free_space(root, inode,
+ ret = btrfs_check_data_free_space(BTRFS_I(inode)->root, inode,
alloc_end - alloc_start);
if (ret)
goto out;
while (1) {
struct btrfs_ordered_extent *ordered;
- trans = btrfs_start_transaction(BTRFS_I(inode)->root, 1);
- if (!trans) {
- ret = -EIO;
- goto out_free;
- }
-
/* the extent lock is ordered inside the running
* transaction
*/
btrfs_put_ordered_extent(ordered);
unlock_extent(&BTRFS_I(inode)->io_tree,
alloc_start, locked_end, GFP_NOFS);
- btrfs_end_transaction(trans, BTRFS_I(inode)->root);
-
/*
* we can't wait on the range with the transaction
* running or with the extent lock held
BUG_ON(IS_ERR(em) || !em);
last_byte = min(extent_map_end(em), alloc_end);
last_byte = (last_byte + mask) & ~mask;
- if (em->block_start == EXTENT_MAP_HOLE) {
- ret = prealloc_file_range(trans, inode, cur_offset,
- last_byte, locked_end + 1,
- alloc_hint, mode);
+ if (em->block_start == EXTENT_MAP_HOLE ||
+ (cur_offset >= inode->i_size &&
+ !test_bit(EXTENT_FLAG_PREALLOC, &em->flags))) {
+ ret = prealloc_file_range(inode,
+ cur_offset, last_byte,
+ alloc_hint, mode);
if (ret < 0) {
free_extent_map(em);
break;
unlock_extent(&BTRFS_I(inode)->io_tree, alloc_start, locked_end,
GFP_NOFS);
- btrfs_end_transaction(trans, BTRFS_I(inode)->root);
-out_free:
- btrfs_free_reserved_data_space(root, inode, alloc_end - alloc_start);
+ btrfs_free_reserved_data_space(BTRFS_I(inode)->root, inode,
+ alloc_end - alloc_start);
out:
mutex_unlock(&inode->i_mutex);
return ret;
.readpage_io_failed_hook = btrfs_io_failed_hook,
.set_bit_hook = btrfs_set_bit_hook,
.clear_bit_hook = btrfs_clear_bit_hook,
+ .merge_extent_hook = btrfs_merge_extent_hook,
+ .split_extent_hook = btrfs_split_extent_hook,
};
/*
.removexattr = btrfs_removexattr,
};
-struct dentry_operations btrfs_dentry_operations = {
+const struct dentry_operations btrfs_dentry_operations = {
.d_delete = btrfs_dentry_delete,
};
git://ftp.safe.ca / safe / jmp / linux-2.6 / blobdiff