*/
#include <linux/fs.h>
+#include <linux/slab.h>
#include <linux/sched.h>
#include <linux/writeback.h>
#include <linux/pagemap.h>
cur_trans->commit_done = 0;
cur_trans->start_time = get_seconds();
- cur_trans->delayed_refs.root.rb_node = NULL;
+ cur_trans->delayed_refs.root = RB_ROOT;
cur_trans->delayed_refs.num_entries = 0;
cur_trans->delayed_refs.num_heads_ready = 0;
cur_trans->delayed_refs.num_heads = 0;
{
if (root->ref_cows && root->last_trans < trans->transid) {
WARN_ON(root == root->fs_info->extent_root);
- WARN_ON(root->root_item.refs == 0);
WARN_ON(root->commit_root != root->node);
radix_tree_tag_set(&root->fs_info->fs_roots_radix,
while (1) {
prepare_to_wait(&root->fs_info->transaction_wait, &wait,
TASK_UNINTERRUPTIBLE);
- if (cur_trans->blocked) {
- mutex_unlock(&root->fs_info->trans_mutex);
- schedule();
- mutex_lock(&root->fs_info->trans_mutex);
- finish_wait(&root->fs_info->transaction_wait,
- &wait);
- } else {
- finish_wait(&root->fs_info->transaction_wait,
- &wait);
+ if (!cur_trans->blocked)
break;
- }
+ mutex_unlock(&root->fs_info->trans_mutex);
+ schedule();
+ mutex_lock(&root->fs_info->trans_mutex);
}
+ finish_wait(&root->fs_info->transaction_wait, &wait);
put_transaction(cur_trans);
}
}
+enum btrfs_trans_type {
+ TRANS_START,
+ TRANS_JOIN,
+ TRANS_USERSPACE,
+};
+
+static int may_wait_transaction(struct btrfs_root *root, int type)
+{
+ if (!root->fs_info->log_root_recovering &&
+ ((type == TRANS_START && !root->fs_info->open_ioctl_trans) ||
+ type == TRANS_USERSPACE))
+ return 1;
+ return 0;
+}
+
static struct btrfs_trans_handle *start_transaction(struct btrfs_root *root,
- int num_blocks, int wait)
+ u64 num_items, int type)
{
- struct btrfs_trans_handle *h =
- kmem_cache_alloc(btrfs_trans_handle_cachep, GFP_NOFS);
+ struct btrfs_trans_handle *h;
+ struct btrfs_transaction *cur_trans;
+ int retries = 0;
int ret;
+again:
+ h = kmem_cache_alloc(btrfs_trans_handle_cachep, GFP_NOFS);
+ if (!h)
+ return ERR_PTR(-ENOMEM);
mutex_lock(&root->fs_info->trans_mutex);
- if (!root->fs_info->log_root_recovering &&
- ((wait == 1 && !root->fs_info->open_ioctl_trans) || wait == 2))
+ if (may_wait_transaction(root, type))
wait_current_trans(root);
+
ret = join_transaction(root);
BUG_ON(ret);
- h->transid = root->fs_info->running_transaction->transid;
- h->transaction = root->fs_info->running_transaction;
- h->blocks_reserved = num_blocks;
+ cur_trans = root->fs_info->running_transaction;
+ cur_trans->use_count++;
+ mutex_unlock(&root->fs_info->trans_mutex);
+
+ h->transid = cur_trans->transid;
+ h->transaction = cur_trans;
h->blocks_used = 0;
h->block_group = 0;
- h->alloc_exclude_nr = 0;
- h->alloc_exclude_start = 0;
+ h->bytes_reserved = 0;
h->delayed_ref_updates = 0;
+ h->block_rsv = NULL;
+
+ smp_mb();
+ if (cur_trans->blocked && may_wait_transaction(root, type)) {
+ btrfs_commit_transaction(h, root);
+ goto again;
+ }
+
+ if (num_items > 0) {
+ ret = btrfs_trans_reserve_metadata(h, root, num_items,
+ &retries);
+ if (ret == -EAGAIN) {
+ btrfs_commit_transaction(h, root);
+ goto again;
+ }
+ if (ret < 0) {
+ btrfs_end_transaction(h, root);
+ return ERR_PTR(ret);
+ }
+ }
- root->fs_info->running_transaction->use_count++;
+ mutex_lock(&root->fs_info->trans_mutex);
record_root_in_trans(h, root);
mutex_unlock(&root->fs_info->trans_mutex);
+
+ if (!current->journal_info && type != TRANS_USERSPACE)
+ current->journal_info = h;
return h;
}
struct btrfs_trans_handle *btrfs_start_transaction(struct btrfs_root *root,
- int num_blocks)
+ int num_items)
{
- return start_transaction(root, num_blocks, 1);
+ return start_transaction(root, num_items, TRANS_START);
}
struct btrfs_trans_handle *btrfs_join_transaction(struct btrfs_root *root,
int num_blocks)
{
- return start_transaction(root, num_blocks, 0);
+ return start_transaction(root, 0, TRANS_JOIN);
}
struct btrfs_trans_handle *btrfs_start_ioctl_transaction(struct btrfs_root *r,
int num_blocks)
{
- return start_transaction(r, num_blocks, 2);
+ return start_transaction(r, 0, TRANS_USERSPACE);
}
/* wait for a transaction commit to be fully complete */
mutex_unlock(&root->fs_info->trans_mutex);
}
+static int should_end_transaction(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root)
+{
+ int ret;
+ ret = btrfs_block_rsv_check(trans, root,
+ &root->fs_info->global_block_rsv, 0, 5);
+ return ret ? 1 : 0;
+}
+
+int btrfs_should_end_transaction(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root)
+{
+ struct btrfs_transaction *cur_trans = trans->transaction;
+ int updates;
+
+ if (cur_trans->blocked || cur_trans->delayed_refs.flushing)
+ return 1;
+
+ updates = trans->delayed_ref_updates;
+ trans->delayed_ref_updates = 0;
+ if (updates)
+ btrfs_run_delayed_refs(trans, root, updates);
+
+ return should_end_transaction(trans, root);
+}
+
static int __btrfs_end_transaction(struct btrfs_trans_handle *trans,
struct btrfs_root *root, int throttle)
{
- struct btrfs_transaction *cur_trans;
+ struct btrfs_transaction *cur_trans = trans->transaction;
struct btrfs_fs_info *info = root->fs_info;
int count = 0;
count++;
}
+ btrfs_trans_release_metadata(trans, root);
+
+ if (!root->fs_info->open_ioctl_trans &&
+ should_end_transaction(trans, root))
+ trans->transaction->blocked = 1;
+
+ if (cur_trans->blocked && !cur_trans->in_commit) {
+ if (throttle)
+ return btrfs_commit_transaction(trans, root);
+ else
+ wake_up_process(info->transaction_kthread);
+ }
+
mutex_lock(&info->trans_mutex);
- cur_trans = info->running_transaction;
- WARN_ON(cur_trans != trans->transaction);
+ WARN_ON(cur_trans != info->running_transaction);
WARN_ON(cur_trans->num_writers < 1);
cur_trans->num_writers--;
wake_up(&cur_trans->writer_wait);
put_transaction(cur_trans);
mutex_unlock(&info->trans_mutex);
+
+ if (current->journal_info == trans)
+ current->journal_info = NULL;
memset(trans, 0, sizeof(*trans));
kmem_cache_free(btrfs_trans_handle_cachep, trans);
+ if (throttle)
+ btrfs_run_delayed_iputs(root);
+
return 0;
}
/*
* when btree blocks are allocated, they have some corresponding bits set for
* them in one of two extent_io trees. This is used to make sure all of
- * those extents are on disk for transaction or log commit
+ * those extents are sent to disk but does not wait on them
*/
-int btrfs_write_and_wait_marked_extents(struct btrfs_root *root,
- struct extent_io_tree *dirty_pages)
+int btrfs_write_marked_extents(struct btrfs_root *root,
+ struct extent_io_tree *dirty_pages, int mark)
{
int ret;
int err = 0;
while (1) {
ret = find_first_extent_bit(dirty_pages, start, &start, &end,
- EXTENT_DIRTY);
+ mark);
if (ret)
break;
while (start <= end) {
page_cache_release(page);
}
}
+ if (err)
+ werr = err;
+ return werr;
+}
+
+/*
+ * when btree blocks are allocated, they have some corresponding bits set for
+ * them in one of two extent_io trees. This is used to make sure all of
+ * those extents are on disk for transaction or log commit. We wait
+ * on all the pages and clear them from the dirty pages state tree
+ */
+int btrfs_wait_marked_extents(struct btrfs_root *root,
+ struct extent_io_tree *dirty_pages, int mark)
+{
+ int ret;
+ int err = 0;
+ int werr = 0;
+ struct page *page;
+ struct inode *btree_inode = root->fs_info->btree_inode;
+ u64 start = 0;
+ u64 end;
+ unsigned long index;
+
while (1) {
- ret = find_first_extent_bit(dirty_pages, 0, &start, &end,
- EXTENT_DIRTY);
+ ret = find_first_extent_bit(dirty_pages, start, &start, &end,
+ mark);
if (ret)
break;
- clear_extent_dirty(dirty_pages, start, end, GFP_NOFS);
+ clear_extent_bits(dirty_pages, start, end, mark, GFP_NOFS);
while (start <= end) {
index = start >> PAGE_CACHE_SHIFT;
start = (u64)(index + 1) << PAGE_CACHE_SHIFT;
return werr;
}
+/*
+ * when btree blocks are allocated, they have some corresponding bits set for
+ * them in one of two extent_io trees. This is used to make sure all of
+ * those extents are on disk for transaction or log commit
+ */
+int btrfs_write_and_wait_marked_extents(struct btrfs_root *root,
+ struct extent_io_tree *dirty_pages, int mark)
+{
+ int ret;
+ int ret2;
+
+ ret = btrfs_write_marked_extents(root, dirty_pages, mark);
+ ret2 = btrfs_wait_marked_extents(root, dirty_pages, mark);
+ return ret || ret2;
+}
+
int btrfs_write_and_wait_transaction(struct btrfs_trans_handle *trans,
struct btrfs_root *root)
{
return filemap_write_and_wait(btree_inode->i_mapping);
}
return btrfs_write_and_wait_marked_extents(root,
- &trans->transaction->dirty_pages);
+ &trans->transaction->dirty_pages,
+ EXTENT_DIRTY);
}
/*
{
int ret;
u64 old_root_bytenr;
+ u64 old_root_used;
struct btrfs_root *tree_root = root->fs_info->tree_root;
+ old_root_used = btrfs_root_used(&root->root_item);
btrfs_write_dirty_block_groups(trans, root);
while (1) {
old_root_bytenr = btrfs_root_bytenr(&root->root_item);
- if (old_root_bytenr == root->node->start)
+ if (old_root_bytenr == root->node->start &&
+ old_root_used == btrfs_root_used(&root->root_item))
break;
btrfs_set_root_node(&root->root_item, root->node);
&root->root_item);
BUG_ON(ret);
+ old_root_used = btrfs_root_used(&root->root_item);
ret = btrfs_write_dirty_block_groups(trans, root);
BUG_ON(ret);
}
btrfs_free_log(trans, root);
btrfs_update_reloc_root(trans, root);
+ btrfs_orphan_commit_root(trans, root);
if (root->commit_root != root->node) {
switch_commit_root(root);
int btrfs_defrag_root(struct btrfs_root *root, int cacheonly)
{
struct btrfs_fs_info *info = root->fs_info;
- int ret;
struct btrfs_trans_handle *trans;
+ int ret;
unsigned long nr;
- smp_mb();
- if (root->defrag_running)
+ if (xchg(&root->defrag_running, 1))
return 0;
- trans = btrfs_start_transaction(root, 1);
+
while (1) {
- root->defrag_running = 1;
+ trans = btrfs_start_transaction(root, 0);
+ if (IS_ERR(trans))
+ return PTR_ERR(trans);
+
ret = btrfs_defrag_leaves(trans, root, cacheonly);
+
nr = trans->blocks_used;
btrfs_end_transaction(trans, root);
btrfs_btree_balance_dirty(info->tree_root, nr);
cond_resched();
- trans = btrfs_start_transaction(root, 1);
if (root->fs_info->closing || ret != -EAGAIN)
break;
}
root->defrag_running = 0;
- smp_mb();
- btrfs_end_transaction(trans, root);
- return 0;
+ return ret;
}
#if 0
struct btrfs_root_item *new_root_item;
struct btrfs_root *tree_root = fs_info->tree_root;
struct btrfs_root *root = pending->root;
+ struct btrfs_root *parent_root;
+ struct inode *parent_inode;
+ struct dentry *dentry;
struct extent_buffer *tmp;
struct extent_buffer *old;
int ret;
+ int retries = 0;
+ u64 to_reserve = 0;
+ u64 index = 0;
u64 objectid;
new_root_item = kmalloc(sizeof(*new_root_item), GFP_NOFS);
if (!new_root_item) {
- ret = -ENOMEM;
+ pending->error = -ENOMEM;
goto fail;
}
+
ret = btrfs_find_free_objectid(trans, tree_root, 0, &objectid);
- if (ret)
+ if (ret) {
+ pending->error = ret;
goto fail;
+ }
+
+ btrfs_reloc_pre_snapshot(trans, pending, &to_reserve);
+ btrfs_orphan_pre_snapshot(trans, pending, &to_reserve);
+
+ if (to_reserve > 0) {
+ ret = btrfs_block_rsv_add(trans, root, &pending->block_rsv,
+ to_reserve, &retries);
+ if (ret) {
+ pending->error = ret;
+ goto fail;
+ }
+ }
+
+ key.objectid = objectid;
+ key.offset = (u64)-1;
+ key.type = BTRFS_ROOT_ITEM_KEY;
+
+ trans->block_rsv = &pending->block_rsv;
+
+ dentry = pending->dentry;
+ parent_inode = dentry->d_parent->d_inode;
+ parent_root = BTRFS_I(parent_inode)->root;
+ record_root_in_trans(trans, parent_root);
+
+ /*
+ * insert the directory item
+ */
+ ret = btrfs_set_inode_index(parent_inode, &index);
+ BUG_ON(ret);
+ ret = btrfs_insert_dir_item(trans, parent_root,
+ dentry->d_name.name, dentry->d_name.len,
+ parent_inode->i_ino, &key,
+ BTRFS_FT_DIR, index);
+ BUG_ON(ret);
+
+ btrfs_i_size_write(parent_inode, parent_inode->i_size +
+ dentry->d_name.len * 2);
+ ret = btrfs_update_inode(trans, parent_root, parent_inode);
+ BUG_ON(ret);
record_root_in_trans(trans, root);
btrfs_set_root_last_snapshot(&root->root_item, trans->transid);
memcpy(new_root_item, &root->root_item, sizeof(*new_root_item));
- key.objectid = objectid;
- key.offset = 0;
- btrfs_set_key_type(&key, BTRFS_ROOT_ITEM_KEY);
-
old = btrfs_lock_root_node(root);
btrfs_cow_block(trans, root, old, NULL, 0, &old);
btrfs_set_lock_blocking(old);
free_extent_buffer(old);
btrfs_set_root_node(new_root_item, tmp);
- ret = btrfs_insert_root(trans, root->fs_info->tree_root, &key,
- new_root_item);
+ /* record when the snapshot was created in key.offset */
+ key.offset = trans->transid;
+ ret = btrfs_insert_root(trans, tree_root, &key, new_root_item);
btrfs_tree_unlock(tmp);
free_extent_buffer(tmp);
- if (ret)
- goto fail;
-
- key.offset = (u64)-1;
- memcpy(&pending->root_key, &key, sizeof(key));
-fail:
- kfree(new_root_item);
- return ret;
-}
-
-static noinline int finish_pending_snapshot(struct btrfs_fs_info *fs_info,
- struct btrfs_pending_snapshot *pending)
-{
- int ret;
- int namelen;
- u64 index = 0;
- struct btrfs_trans_handle *trans;
- struct inode *parent_inode;
- struct inode *inode;
- struct btrfs_root *parent_root;
-
- parent_inode = pending->dentry->d_parent->d_inode;
- parent_root = BTRFS_I(parent_inode)->root;
- trans = btrfs_join_transaction(parent_root, 1);
+ BUG_ON(ret);
/*
- * insert the directory item
+ * insert root back/forward references
*/
- namelen = strlen(pending->name);
- ret = btrfs_set_inode_index(parent_inode, &index);
- ret = btrfs_insert_dir_item(trans, parent_root,
- pending->name, namelen,
- parent_inode->i_ino,
- &pending->root_key, BTRFS_FT_DIR, index);
-
- if (ret)
- goto fail;
-
- btrfs_i_size_write(parent_inode, parent_inode->i_size + namelen * 2);
- ret = btrfs_update_inode(trans, parent_root, parent_inode);
- BUG_ON(ret);
-
- /* add the backref first */
- ret = btrfs_add_root_ref(trans, parent_root->fs_info->tree_root,
- pending->root_key.objectid,
- BTRFS_ROOT_BACKREF_KEY,
+ ret = btrfs_add_root_ref(trans, tree_root, objectid,
parent_root->root_key.objectid,
- parent_inode->i_ino, index, pending->name,
- namelen);
-
+ parent_inode->i_ino, index,
+ dentry->d_name.name, dentry->d_name.len);
BUG_ON(ret);
- /* now add the forward ref */
- ret = btrfs_add_root_ref(trans, parent_root->fs_info->tree_root,
- parent_root->root_key.objectid,
- BTRFS_ROOT_REF_KEY,
- pending->root_key.objectid,
- parent_inode->i_ino, index, pending->name,
- namelen);
+ key.offset = (u64)-1;
+ pending->snap = btrfs_read_fs_root_no_name(root->fs_info, &key);
+ BUG_ON(IS_ERR(pending->snap));
- inode = btrfs_lookup_dentry(parent_inode, pending->dentry);
- d_instantiate(pending->dentry, inode);
+ btrfs_reloc_post_snapshot(trans, pending);
+ btrfs_orphan_post_snapshot(trans, pending);
fail:
- btrfs_end_transaction(trans, fs_info->fs_root);
- return ret;
+ kfree(new_root_item);
+ btrfs_block_rsv_release(root, &pending->block_rsv, (u64)-1);
+ return 0;
}
/*
return 0;
}
-static noinline int finish_pending_snapshots(struct btrfs_trans_handle *trans,
- struct btrfs_fs_info *fs_info)
-{
- struct btrfs_pending_snapshot *pending;
- struct list_head *head = &trans->transaction->pending_snapshots;
- int ret;
-
- while (!list_empty(head)) {
- pending = list_entry(head->next,
- struct btrfs_pending_snapshot, list);
- ret = finish_pending_snapshot(fs_info, pending);
- BUG_ON(ret);
- list_del(&pending->list);
- kfree(pending->name);
- kfree(pending);
- }
- return 0;
-}
-
static void update_super_roots(struct btrfs_root *root)
{
struct btrfs_root_item *root_item;
super->root_level = root_item->level;
}
+int btrfs_transaction_in_commit(struct btrfs_fs_info *info)
+{
+ int ret = 0;
+ spin_lock(&info->new_trans_lock);
+ if (info->running_transaction)
+ ret = info->running_transaction->in_commit;
+ spin_unlock(&info->new_trans_lock);
+ return ret;
+}
+
+int btrfs_transaction_blocked(struct btrfs_fs_info *info)
+{
+ int ret = 0;
+ spin_lock(&info->new_trans_lock);
+ if (info->running_transaction)
+ ret = info->running_transaction->blocked;
+ spin_unlock(&info->new_trans_lock);
+ return ret;
+}
+
int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
struct btrfs_root *root)
{
unsigned long timeout = 1;
struct btrfs_transaction *cur_trans;
struct btrfs_transaction *prev_trans = NULL;
- struct extent_io_tree *pinned_copy;
DEFINE_WAIT(wait);
int ret;
int should_grow = 0;
ret = btrfs_run_delayed_refs(trans, root, 0);
BUG_ON(ret);
+ btrfs_trans_release_metadata(trans, root);
+
cur_trans = trans->transaction;
/*
* set the flushing flag so procs in this transaction have to
return 0;
}
- pinned_copy = kmalloc(sizeof(*pinned_copy), GFP_NOFS);
- if (!pinned_copy)
- return -ENOMEM;
-
- extent_io_tree_init(pinned_copy,
- root->fs_info->btree_inode->i_mapping, GFP_NOFS);
-
trans->transaction->in_commit = 1;
trans->transaction->blocked = 1;
if (cur_trans->list.prev != &root->fs_info->trans_list) {
snap_pending = 1;
WARN_ON(cur_trans != trans->transaction);
- prepare_to_wait(&cur_trans->writer_wait, &wait,
- TASK_UNINTERRUPTIBLE);
-
if (cur_trans->num_writers > 1)
timeout = MAX_SCHEDULE_TIMEOUT;
else if (should_grow)
mutex_unlock(&root->fs_info->trans_mutex);
- if (flush_on_commit) {
- btrfs_start_delalloc_inodes(root);
- ret = btrfs_wait_ordered_extents(root, 0);
- BUG_ON(ret);
- } else if (snap_pending) {
- ret = btrfs_wait_ordered_extents(root, 1);
+ if (flush_on_commit || snap_pending) {
+ btrfs_start_delalloc_inodes(root, 1);
+ ret = btrfs_wait_ordered_extents(root, 0, 1);
BUG_ON(ret);
}
*/
btrfs_run_ordered_operations(root, 1);
+ prepare_to_wait(&cur_trans->writer_wait, &wait,
+ TASK_UNINTERRUPTIBLE);
+
smp_mb();
if (cur_trans->num_writers > 1 || should_grow)
schedule_timeout(timeout);
ret = commit_cowonly_roots(trans, root);
BUG_ON(ret);
+ btrfs_prepare_extent_commit(trans, root);
+
cur_trans = root->fs_info->running_transaction;
spin_lock(&root->fs_info->new_trans_lock);
root->fs_info->running_transaction = NULL;
memcpy(&root->fs_info->super_for_commit, &root->fs_info->super_copy,
sizeof(root->fs_info->super_copy));
- btrfs_copy_pinned(root, pinned_copy);
-
trans->transaction->blocked = 0;
wake_up(&root->fs_info->transaction_wait);
*/
mutex_unlock(&root->fs_info->tree_log_mutex);
- btrfs_finish_extent_commit(trans, root, pinned_copy);
- kfree(pinned_copy);
-
- /* do the directory inserts of any pending snapshot creations */
- finish_pending_snapshots(trans, root->fs_info);
+ btrfs_finish_extent_commit(trans, root);
mutex_lock(&root->fs_info->trans_mutex);
mutex_unlock(&root->fs_info->trans_mutex);
+ if (current->journal_info == trans)
+ current->journal_info = NULL;
+
kmem_cache_free(btrfs_trans_handle_cachep, trans);
+
+ if (current != root->fs_info->transaction_kthread)
+ btrfs_run_delayed_iputs(root);
+
return ret;
}
while (!list_empty(&list)) {
root = list_entry(list.next, struct btrfs_root, root_list);
- list_del_init(&root->root_list);
- btrfs_drop_snapshot(root, 0);
+ list_del(&root->root_list);
+
+ if (btrfs_header_backref_rev(root->node) <
+ BTRFS_MIXED_BACKREF_REV)
+ btrfs_drop_snapshot(root, NULL, 0);
+ else
+ btrfs_drop_snapshot(root, NULL, 1);
}
return 0;
}