* Boston, MA 021110-1307, USA.
*/
-#include <linux/gfp.h>
#include <linux/slab.h>
#include <linux/blkdev.h>
+#include <linux/writeback.h>
+#include <linux/pagevec.h>
#include "ctree.h"
#include "transaction.h"
#include "btrfs_inode.h"
#include "extent_io.h"
-
static u64 entry_end(struct btrfs_ordered_extent *entry)
{
if (entry->file_offset + entry->len < entry->file_offset)
return entry->file_offset + entry->len;
}
+/* returns NULL if the insertion worked, or it returns the node it did find
+ * in the tree
+ */
static struct rb_node *tree_insert(struct rb_root *root, u64 file_offset,
struct rb_node *node)
{
- struct rb_node ** p = &root->rb_node;
- struct rb_node * parent = NULL;
+ struct rb_node **p = &root->rb_node;
+ struct rb_node *parent = NULL;
struct btrfs_ordered_extent *entry;
- while(*p) {
+ while (*p) {
parent = *p;
entry = rb_entry(parent, struct btrfs_ordered_extent, rb_node);
return NULL;
}
+/*
+ * look for a given offset in the tree, and if it can't be found return the
+ * first lesser offset
+ */
static struct rb_node *__tree_search(struct rb_root *root, u64 file_offset,
struct rb_node **prev_ret)
{
- struct rb_node * n = root->rb_node;
+ struct rb_node *n = root->rb_node;
struct rb_node *prev = NULL;
struct rb_node *test;
struct btrfs_ordered_extent *entry;
struct btrfs_ordered_extent *prev_entry = NULL;
- while(n) {
+ while (n) {
entry = rb_entry(n, struct btrfs_ordered_extent, rb_node);
prev = n;
prev_entry = entry;
if (!prev_ret)
return NULL;
- while(prev && file_offset >= entry_end(prev_entry)) {
+ while (prev && file_offset >= entry_end(prev_entry)) {
test = rb_next(prev);
if (!test)
break;
if (prev)
prev_entry = rb_entry(prev, struct btrfs_ordered_extent,
rb_node);
- while(prev && file_offset < entry_end(prev_entry)) {
+ while (prev && file_offset < entry_end(prev_entry)) {
test = rb_prev(prev);
if (!test)
break;
return NULL;
}
+/*
+ * helper to check if a given offset is inside a given entry
+ */
static int offset_in_entry(struct btrfs_ordered_extent *entry, u64 file_offset)
{
if (file_offset < entry->file_offset ||
return 1;
}
+/*
+ * look find the first ordered struct that has this offset, otherwise
+ * the first one less than this offset
+ */
static inline struct rb_node *tree_search(struct btrfs_ordered_inode_tree *tree,
u64 file_offset)
{
*
* len is the length of the extent
*
- * This also sets the EXTENT_ORDERED bit on the range in the inode.
- *
* The tree is given a single reference on the ordered extent that was
* inserted.
*/
int btrfs_add_ordered_extent(struct inode *inode, u64 file_offset,
- u64 start, u64 len)
+ u64 start, u64 len, u64 disk_len, int type)
{
struct btrfs_ordered_inode_tree *tree;
struct rb_node *node;
if (!entry)
return -ENOMEM;
- mutex_lock(&tree->mutex);
entry->file_offset = file_offset;
entry->start = start;
entry->len = len;
+ entry->disk_len = disk_len;
+ entry->bytes_left = len;
+ entry->inode = inode;
+ if (type != BTRFS_ORDERED_IO_DONE && type != BTRFS_ORDERED_COMPLETE)
+ set_bit(type, &entry->flags);
+
/* one ref for the tree */
atomic_set(&entry->refs, 1);
init_waitqueue_head(&entry->wait);
INIT_LIST_HEAD(&entry->list);
+ INIT_LIST_HEAD(&entry->root_extent_list);
+ spin_lock(&tree->lock);
node = tree_insert(&tree->tree, file_offset,
&entry->rb_node);
- if (node) {
- entry = rb_entry(node, struct btrfs_ordered_extent, rb_node);
- atomic_inc(&entry->refs);
- }
- set_extent_ordered(&BTRFS_I(inode)->io_tree, file_offset,
- entry_end(entry) - 1, GFP_NOFS);
+ BUG_ON(node);
+ spin_unlock(&tree->lock);
+
+ spin_lock(&BTRFS_I(inode)->root->fs_info->ordered_extent_lock);
+ list_add_tail(&entry->root_extent_list,
+ &BTRFS_I(inode)->root->fs_info->ordered_extents);
+ spin_unlock(&BTRFS_I(inode)->root->fs_info->ordered_extent_lock);
- mutex_unlock(&tree->mutex);
BUG_ON(node);
return 0;
}
struct btrfs_ordered_inode_tree *tree;
tree = &BTRFS_I(inode)->ordered_tree;
- mutex_lock(&tree->mutex);
+ spin_lock(&tree->lock);
list_add_tail(&sum->list, &entry->list);
- mutex_unlock(&tree->mutex);
+ spin_unlock(&tree->lock);
return 0;
}
* to make sure this function only returns 1 once for a given ordered extent.
*/
int btrfs_dec_test_ordered_pending(struct inode *inode,
+ struct btrfs_ordered_extent **cached,
u64 file_offset, u64 io_size)
{
struct btrfs_ordered_inode_tree *tree;
struct rb_node *node;
- struct btrfs_ordered_extent *entry;
- struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
+ struct btrfs_ordered_extent *entry = NULL;
int ret;
tree = &BTRFS_I(inode)->ordered_tree;
- mutex_lock(&tree->mutex);
- clear_extent_ordered(io_tree, file_offset, file_offset + io_size - 1,
- GFP_NOFS);
+ spin_lock(&tree->lock);
node = tree_search(tree, file_offset);
if (!node) {
ret = 1;
goto out;
}
- ret = test_range_bit(io_tree, entry->file_offset,
- entry->file_offset + entry->len - 1,
- EXTENT_ORDERED, 0);
- if (ret == 0)
+ if (io_size > entry->bytes_left) {
+ printk(KERN_CRIT "bad ordered accounting left %llu size %llu\n",
+ (unsigned long long)entry->bytes_left,
+ (unsigned long long)io_size);
+ }
+ entry->bytes_left -= io_size;
+ if (entry->bytes_left == 0)
ret = test_and_set_bit(BTRFS_ORDERED_IO_DONE, &entry->flags);
+ else
+ ret = 1;
out:
- mutex_unlock(&tree->mutex);
+ if (!ret && cached && entry) {
+ *cached = entry;
+ atomic_inc(&entry->refs);
+ }
+ spin_unlock(&tree->lock);
return ret == 0;
}
struct btrfs_ordered_sum *sum;
if (atomic_dec_and_test(&entry->refs)) {
- while(!list_empty(&entry->list)) {
+ while (!list_empty(&entry->list)) {
cur = entry->list.next;
sum = list_entry(cur, struct btrfs_ordered_sum, list);
list_del(&sum->list);
/*
* remove an ordered extent from the tree. No references are dropped
- * but, anyone waiting on this extent is woken up.
+ * and you must wake_up entry->wait. You must hold the tree lock
+ * while you call this function.
*/
-int btrfs_remove_ordered_extent(struct inode *inode,
+static int __btrfs_remove_ordered_extent(struct inode *inode,
struct btrfs_ordered_extent *entry)
{
struct btrfs_ordered_inode_tree *tree;
+ struct btrfs_root *root = BTRFS_I(inode)->root;
struct rb_node *node;
tree = &BTRFS_I(inode)->ordered_tree;
- mutex_lock(&tree->mutex);
node = &entry->rb_node;
rb_erase(node, &tree->tree);
tree->last = NULL;
set_bit(BTRFS_ORDERED_COMPLETE, &entry->flags);
- mutex_unlock(&tree->mutex);
+
+ spin_lock(&BTRFS_I(inode)->accounting_lock);
+ WARN_ON(!BTRFS_I(inode)->outstanding_extents);
+ BTRFS_I(inode)->outstanding_extents--;
+ spin_unlock(&BTRFS_I(inode)->accounting_lock);
+ btrfs_unreserve_metadata_for_delalloc(BTRFS_I(inode)->root,
+ inode, 1);
+
+ spin_lock(&root->fs_info->ordered_extent_lock);
+ list_del_init(&entry->root_extent_list);
+
+ /*
+ * we have no more ordered extents for this inode and
+ * no dirty pages. We can safely remove it from the
+ * list of ordered extents
+ */
+ if (RB_EMPTY_ROOT(&tree->tree) &&
+ !mapping_tagged(inode->i_mapping, PAGECACHE_TAG_DIRTY)) {
+ list_del_init(&BTRFS_I(inode)->ordered_operations);
+ }
+ spin_unlock(&root->fs_info->ordered_extent_lock);
+
+ return 0;
+}
+
+/*
+ * remove an ordered extent from the tree. No references are dropped
+ * but any waiters are woken.
+ */
+int btrfs_remove_ordered_extent(struct inode *inode,
+ struct btrfs_ordered_extent *entry)
+{
+ struct btrfs_ordered_inode_tree *tree;
+ int ret;
+
+ tree = &BTRFS_I(inode)->ordered_tree;
+ spin_lock(&tree->lock);
+ ret = __btrfs_remove_ordered_extent(inode, entry);
+ spin_unlock(&tree->lock);
wake_up(&entry->wait);
+
+ return ret;
+}
+
+/*
+ * wait for all the ordered extents in a root. This is done when balancing
+ * space between drives.
+ */
+int btrfs_wait_ordered_extents(struct btrfs_root *root,
+ int nocow_only, int delay_iput)
+{
+ struct list_head splice;
+ struct list_head *cur;
+ struct btrfs_ordered_extent *ordered;
+ struct inode *inode;
+
+ INIT_LIST_HEAD(&splice);
+
+ spin_lock(&root->fs_info->ordered_extent_lock);
+ list_splice_init(&root->fs_info->ordered_extents, &splice);
+ while (!list_empty(&splice)) {
+ cur = splice.next;
+ ordered = list_entry(cur, struct btrfs_ordered_extent,
+ root_extent_list);
+ if (nocow_only &&
+ !test_bit(BTRFS_ORDERED_NOCOW, &ordered->flags) &&
+ !test_bit(BTRFS_ORDERED_PREALLOC, &ordered->flags)) {
+ list_move(&ordered->root_extent_list,
+ &root->fs_info->ordered_extents);
+ cond_resched_lock(&root->fs_info->ordered_extent_lock);
+ continue;
+ }
+
+ list_del_init(&ordered->root_extent_list);
+ atomic_inc(&ordered->refs);
+
+ /*
+ * the inode may be getting freed (in sys_unlink path).
+ */
+ inode = igrab(ordered->inode);
+
+ spin_unlock(&root->fs_info->ordered_extent_lock);
+
+ if (inode) {
+ btrfs_start_ordered_extent(inode, ordered, 1);
+ btrfs_put_ordered_extent(ordered);
+ if (delay_iput)
+ btrfs_add_delayed_iput(inode);
+ else
+ iput(inode);
+ } else {
+ btrfs_put_ordered_extent(ordered);
+ }
+
+ spin_lock(&root->fs_info->ordered_extent_lock);
+ }
+ spin_unlock(&root->fs_info->ordered_extent_lock);
+ return 0;
+}
+
+/*
+ * this is used during transaction commit to write all the inodes
+ * added to the ordered operation list. These files must be fully on
+ * disk before the transaction commits.
+ *
+ * we have two modes here, one is to just start the IO via filemap_flush
+ * and the other is to wait for all the io. When we wait, we have an
+ * extra check to make sure the ordered operation list really is empty
+ * before we return
+ */
+int btrfs_run_ordered_operations(struct btrfs_root *root, int wait)
+{
+ struct btrfs_inode *btrfs_inode;
+ struct inode *inode;
+ struct list_head splice;
+
+ INIT_LIST_HEAD(&splice);
+
+ mutex_lock(&root->fs_info->ordered_operations_mutex);
+ spin_lock(&root->fs_info->ordered_extent_lock);
+again:
+ list_splice_init(&root->fs_info->ordered_operations, &splice);
+
+ while (!list_empty(&splice)) {
+ btrfs_inode = list_entry(splice.next, struct btrfs_inode,
+ ordered_operations);
+
+ inode = &btrfs_inode->vfs_inode;
+
+ list_del_init(&btrfs_inode->ordered_operations);
+
+ /*
+ * the inode may be getting freed (in sys_unlink path).
+ */
+ inode = igrab(inode);
+
+ if (!wait && inode) {
+ list_add_tail(&BTRFS_I(inode)->ordered_operations,
+ &root->fs_info->ordered_operations);
+ }
+ spin_unlock(&root->fs_info->ordered_extent_lock);
+
+ if (inode) {
+ if (wait)
+ btrfs_wait_ordered_range(inode, 0, (u64)-1);
+ else
+ filemap_flush(inode->i_mapping);
+ btrfs_add_delayed_iput(inode);
+ }
+
+ cond_resched();
+ spin_lock(&root->fs_info->ordered_extent_lock);
+ }
+ if (wait && !list_empty(&root->fs_info->ordered_operations))
+ goto again;
+
+ spin_unlock(&root->fs_info->ordered_extent_lock);
+ mutex_unlock(&root->fs_info->ordered_operations_mutex);
+
return 0;
}
* start IO on any dirty ones so the wait doesn't stall waiting
* for pdflush to find them
*/
-#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,22)
- do_sync_file_range(file, start, end, SYNC_FILE_RANGE_WRITE);
-#else
- do_sync_mapping_range(inode->i_mapping, start, end,
- SYNC_FILE_RANGE_WRITE);
-#endif
- if (wait)
+ filemap_fdatawrite_range(inode->i_mapping, start, end);
+ if (wait) {
wait_event(entry->wait, test_bit(BTRFS_ORDERED_COMPLETE,
&entry->flags));
+ }
}
/*
* Used to wait on ordered extents across a large range of bytes.
*/
-void btrfs_wait_ordered_range(struct inode *inode, u64 start, u64 len)
+int btrfs_wait_ordered_range(struct inode *inode, u64 start, u64 len)
{
u64 end;
u64 orig_end;
u64 wait_end;
struct btrfs_ordered_extent *ordered;
- u64 mask = BTRFS_I(inode)->root->sectorsize - 1;
+ int found;
if (start + len < start) {
- wait_end = (inode->i_size + mask) & ~mask;
- orig_end = (u64)-1;
+ orig_end = INT_LIMIT(loff_t);
} else {
orig_end = start + len - 1;
- wait_end = orig_end;
+ if (orig_end > INT_LIMIT(loff_t))
+ orig_end = INT_LIMIT(loff_t);
}
-
+ wait_end = orig_end;
+again:
/* start IO across the range first to instantiate any delalloc
* extents
*/
-#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,22)
- do_sync_file_range(file, start, wait_end, SYNC_FILE_RANGE_WRITE);
-#else
- do_sync_mapping_range(inode->i_mapping, start, wait_end,
- SYNC_FILE_RANGE_WRITE);
-#endif
- end = orig_end;
- wait_on_extent_writeback(&BTRFS_I(inode)->io_tree, start, orig_end);
+ filemap_fdatawrite_range(inode->i_mapping, start, orig_end);
- while(1) {
+ /* The compression code will leave pages locked but return from
+ * writepage without setting the page writeback. Starting again
+ * with WB_SYNC_ALL will end up waiting for the IO to actually start.
+ */
+ filemap_fdatawrite_range(inode->i_mapping, start, orig_end);
+
+ filemap_fdatawait_range(inode->i_mapping, start, orig_end);
+
+ end = orig_end;
+ found = 0;
+ while (1) {
ordered = btrfs_lookup_first_ordered_extent(inode, end);
- if (!ordered) {
+ if (!ordered)
break;
- }
if (ordered->file_offset > orig_end) {
btrfs_put_ordered_extent(ordered);
break;
btrfs_put_ordered_extent(ordered);
break;
}
+ found++;
btrfs_start_ordered_extent(inode, ordered, 1);
end = ordered->file_offset;
btrfs_put_ordered_extent(ordered);
break;
end--;
}
+ if (found || test_range_bit(&BTRFS_I(inode)->io_tree, start, orig_end,
+ EXTENT_DELALLOC, 0, NULL)) {
+ schedule_timeout(1);
+ goto again;
+ }
+ return 0;
}
/*
struct btrfs_ordered_extent *entry = NULL;
tree = &BTRFS_I(inode)->ordered_tree;
- mutex_lock(&tree->mutex);
+ spin_lock(&tree->lock);
node = tree_search(tree, file_offset);
if (!node)
goto out;
if (entry)
atomic_inc(&entry->refs);
out:
- mutex_unlock(&tree->mutex);
+ spin_unlock(&tree->lock);
return entry;
}
* if none is found
*/
struct btrfs_ordered_extent *
-btrfs_lookup_first_ordered_extent(struct inode * inode, u64 file_offset)
+btrfs_lookup_first_ordered_extent(struct inode *inode, u64 file_offset)
{
struct btrfs_ordered_inode_tree *tree;
struct rb_node *node;
struct btrfs_ordered_extent *entry = NULL;
tree = &BTRFS_I(inode)->ordered_tree;
- mutex_lock(&tree->mutex);
+ spin_lock(&tree->lock);
node = tree_search(tree, file_offset);
if (!node)
goto out;
entry = rb_entry(node, struct btrfs_ordered_extent, rb_node);
atomic_inc(&entry->refs);
out:
- mutex_unlock(&tree->mutex);
+ spin_unlock(&tree->lock);
return entry;
}
* After an extent is done, call this to conditionally update the on disk
* i_size. i_size is updated to cover any fully written part of the file.
*/
-int btrfs_ordered_update_i_size(struct inode *inode,
+int btrfs_ordered_update_i_size(struct inode *inode, u64 offset,
struct btrfs_ordered_extent *ordered)
{
struct btrfs_ordered_inode_tree *tree = &BTRFS_I(inode)->ordered_tree;
u64 disk_i_size;
u64 new_i_size;
u64 i_size_test;
+ u64 i_size = i_size_read(inode);
struct rb_node *node;
+ struct rb_node *prev = NULL;
struct btrfs_ordered_extent *test;
+ int ret = 1;
- mutex_lock(&tree->mutex);
+ if (ordered)
+ offset = entry_end(ordered);
+ else
+ offset = ALIGN(offset, BTRFS_I(inode)->root->sectorsize);
+
+ spin_lock(&tree->lock);
disk_i_size = BTRFS_I(inode)->disk_i_size;
+ /* truncate file */
+ if (disk_i_size > i_size) {
+ BTRFS_I(inode)->disk_i_size = i_size;
+ ret = 0;
+ goto out;
+ }
+
/*
* if the disk i_size is already at the inode->i_size, or
* this ordered extent is inside the disk i_size, we're done
*/
- if (disk_i_size >= inode->i_size ||
- ordered->file_offset + ordered->len <= disk_i_size) {
+ if (disk_i_size == i_size || offset <= disk_i_size) {
goto out;
}
* we can't update the disk_isize if there are delalloc bytes
* between disk_i_size and this ordered extent
*/
- if (test_range_bit(io_tree, disk_i_size,
- ordered->file_offset + ordered->len - 1,
- EXTENT_DELALLOC, 0)) {
+ if (test_range_bit(io_tree, disk_i_size, offset - 1,
+ EXTENT_DELALLOC, 0, NULL)) {
goto out;
}
/*
* if we find an ordered extent then we can't update disk i_size
* yet
*/
- node = &ordered->rb_node;
- while(1) {
- node = rb_prev(node);
- if (!node)
- break;
+ if (ordered) {
+ node = rb_prev(&ordered->rb_node);
+ } else {
+ prev = tree_search(tree, offset);
+ /*
+ * we insert file extents without involving ordered struct,
+ * so there should be no ordered struct cover this offset
+ */
+ if (prev) {
+ test = rb_entry(prev, struct btrfs_ordered_extent,
+ rb_node);
+ BUG_ON(offset_in_entry(test, offset));
+ }
+ node = prev;
+ }
+ while (node) {
test = rb_entry(node, struct btrfs_ordered_extent, rb_node);
if (test->file_offset + test->len <= disk_i_size)
break;
- if (test->file_offset >= inode->i_size)
+ if (test->file_offset >= i_size)
break;
if (test->file_offset >= disk_i_size)
goto out;
+ node = rb_prev(node);
}
- new_i_size = min_t(u64, entry_end(ordered), i_size_read(inode));
+ new_i_size = min_t(u64, offset, i_size);
/*
* at this point, we know we can safely update i_size to at least
* walk forward and see if ios from higher up in the file have
* finished.
*/
- node = rb_next(&ordered->rb_node);
+ if (ordered) {
+ node = rb_next(&ordered->rb_node);
+ } else {
+ if (prev)
+ node = rb_next(prev);
+ else
+ node = rb_first(&tree->tree);
+ }
i_size_test = 0;
if (node) {
/*
* between our ordered extent and the next one.
*/
test = rb_entry(node, struct btrfs_ordered_extent, rb_node);
- if (test->file_offset > entry_end(ordered)) {
- i_size_test = test->file_offset - 1;
- }
+ if (test->file_offset > offset)
+ i_size_test = test->file_offset;
} else {
- i_size_test = i_size_read(inode);
+ i_size_test = i_size;
}
/*
* are no delalloc bytes in this area, it is safe to update
* disk_i_size to the end of the region.
*/
- if (i_size_test > entry_end(ordered) &&
- !test_range_bit(io_tree, entry_end(ordered), i_size_test,
- EXTENT_DELALLOC, 0)) {
- new_i_size = min_t(u64, i_size_test, i_size_read(inode));
+ if (i_size_test > offset &&
+ !test_range_bit(io_tree, offset, i_size_test - 1,
+ EXTENT_DELALLOC, 0, NULL)) {
+ new_i_size = min_t(u64, i_size_test, i_size);
}
BTRFS_I(inode)->disk_i_size = new_i_size;
+ ret = 0;
out:
- mutex_unlock(&tree->mutex);
- return 0;
+ /*
+ * we need to remove the ordered extent with the tree lock held
+ * so that other people calling this function don't find our fully
+ * processed ordered entry and skip updating the i_size
+ */
+ if (ordered)
+ __btrfs_remove_ordered_extent(inode, ordered);
+ spin_unlock(&tree->lock);
+ if (ordered)
+ wake_up(&ordered->wait);
+ return ret;
}
/*
* try to find a checksum. This is used because we allow pages to
* be reclaimed before their checksum is actually put into the btree
*/
-int btrfs_find_ordered_sum(struct inode *inode, u64 offset, u32 *sum)
+int btrfs_find_ordered_sum(struct inode *inode, u64 offset, u64 disk_bytenr,
+ u32 *sum)
{
struct btrfs_ordered_sum *ordered_sum;
struct btrfs_sector_sum *sector_sums;
struct btrfs_ordered_extent *ordered;
struct btrfs_ordered_inode_tree *tree = &BTRFS_I(inode)->ordered_tree;
- struct list_head *cur;
unsigned long num_sectors;
unsigned long i;
u32 sectorsize = BTRFS_I(inode)->root->sectorsize;
if (!ordered)
return 1;
- mutex_lock(&tree->mutex);
- list_for_each_prev(cur, &ordered->list) {
- ordered_sum = list_entry(cur, struct btrfs_ordered_sum, list);
- if (offset >= ordered_sum->file_offset) {
+ spin_lock(&tree->lock);
+ list_for_each_entry_reverse(ordered_sum, &ordered->list, list) {
+ if (disk_bytenr >= ordered_sum->bytenr) {
num_sectors = ordered_sum->len / sectorsize;
- sector_sums = &ordered_sum->sums;
+ sector_sums = ordered_sum->sums;
for (i = 0; i < num_sectors; i++) {
- if (sector_sums[i].offset == offset) {
-printk("find ordered sum inode %lu offset %Lu\n", inode->i_ino, offset);
+ if (sector_sums[i].bytenr == disk_bytenr) {
*sum = sector_sums[i].sum;
ret = 0;
goto out;
}
}
out:
- mutex_unlock(&tree->mutex);
+ spin_unlock(&tree->lock);
+ btrfs_put_ordered_extent(ordered);
return ret;
}
+
+/*
+ * add a given inode to the list of inodes that must be fully on
+ * disk before a transaction commit finishes.
+ *
+ * This basically gives us the ext3 style data=ordered mode, and it is mostly
+ * used to make sure renamed files are fully on disk.
+ *
+ * It is a noop if the inode is already fully on disk.
+ *
+ * If trans is not null, we'll do a friendly check for a transaction that
+ * is already flushing things and force the IO down ourselves.
+ */
+int btrfs_add_ordered_operation(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root,
+ struct inode *inode)
+{
+ u64 last_mod;
+
+ last_mod = max(BTRFS_I(inode)->generation, BTRFS_I(inode)->last_trans);
+
+ /*
+ * if this file hasn't been changed since the last transaction
+ * commit, we can safely return without doing anything
+ */
+ if (last_mod < root->fs_info->last_trans_committed)
+ return 0;
+
+ /*
+ * the transaction is already committing. Just start the IO and
+ * don't bother with all of this list nonsense
+ */
+ if (trans && root->fs_info->running_transaction->blocked) {
+ btrfs_wait_ordered_range(inode, 0, (u64)-1);
+ return 0;
+ }
+
+ spin_lock(&root->fs_info->ordered_extent_lock);
+ if (list_empty(&BTRFS_I(inode)->ordered_operations)) {
+ list_add_tail(&BTRFS_I(inode)->ordered_operations,
+ &root->fs_info->ordered_operations);
+ }
+ spin_unlock(&root->fs_info->ordered_extent_lock);
+
+ return 0;
+}