Btrfs: don't wait for unrelated IO to finish before relocation
Before the relocation process of a block group starts, it sets the block
group to readonly mode, then flushes all delalloc writes and then finally
it waits for all ordered extents to complete. This last step includes
waiting for ordered extents destinated at extents allocated in other block
groups, making us waste unecessary time.
So improve this by waiting only for ordered extents that fall into the
block group's range.
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: Josef Bacik <jbacik@fb.com>
Reviewed-by: Liu Bo <bo.li.liu@oracle.com>
diff --git a/fs/btrfs/dev-replace.c b/fs/btrfs/dev-replace.c
index 26bcb48..3371f9e 100644
--- a/fs/btrfs/dev-replace.c
+++ b/fs/btrfs/dev-replace.c
@@ -403,7 +403,7 @@
if (ret)
btrfs_err(root->fs_info, "kobj add dev failed %d\n", ret);
- btrfs_wait_ordered_roots(root->fs_info, -1);
+ btrfs_wait_ordered_roots(root->fs_info, -1, 0, (u64)-1);
/* force writing the updated state information to disk */
trans = btrfs_start_transaction(root, 0);
@@ -495,7 +495,7 @@
mutex_unlock(&dev_replace->lock_finishing_cancel_unmount);
return ret;
}
- btrfs_wait_ordered_roots(root->fs_info, -1);
+ btrfs_wait_ordered_roots(root->fs_info, -1, 0, (u64)-1);
trans = btrfs_start_transaction(root, 0);
if (IS_ERR(trans)) {
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index 84e060e..251452a 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -4141,7 +4141,7 @@
if (need_commit > 0) {
btrfs_start_delalloc_roots(fs_info, 0, -1);
- btrfs_wait_ordered_roots(fs_info, -1);
+ btrfs_wait_ordered_roots(fs_info, -1, 0, (u64)-1);
}
trans = btrfs_join_transaction(root);
@@ -4583,7 +4583,8 @@
*/
btrfs_start_delalloc_roots(root->fs_info, 0, nr_items);
if (!current->journal_info)
- btrfs_wait_ordered_roots(root->fs_info, nr_items);
+ btrfs_wait_ordered_roots(root->fs_info, nr_items,
+ 0, (u64)-1);
}
}
@@ -4632,7 +4633,8 @@
if (trans)
return;
if (wait_ordered)
- btrfs_wait_ordered_roots(root->fs_info, items);
+ btrfs_wait_ordered_roots(root->fs_info, items,
+ 0, (u64)-1);
return;
}
@@ -4671,7 +4673,8 @@
loops++;
if (wait_ordered && !trans) {
- btrfs_wait_ordered_roots(root->fs_info, items);
+ btrfs_wait_ordered_roots(root->fs_info, items,
+ 0, (u64)-1);
} else {
time_left = schedule_timeout_killable(1);
if (time_left)
diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c
index 5a23806..697cc33 100644
--- a/fs/btrfs/ioctl.c
+++ b/fs/btrfs/ioctl.c
@@ -681,7 +681,7 @@
if (ret)
goto dec_and_free;
- btrfs_wait_ordered_extents(root, -1);
+ btrfs_wait_ordered_extents(root, -1, 0, (u64)-1);
btrfs_init_block_rsv(&pending_snapshot->block_rsv,
BTRFS_BLOCK_RSV_TEMP);
diff --git a/fs/btrfs/ordered-data.c b/fs/btrfs/ordered-data.c
index 0de7da5..5591704 100644
--- a/fs/btrfs/ordered-data.c
+++ b/fs/btrfs/ordered-data.c
@@ -661,14 +661,15 @@
* 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 nr)
+int btrfs_wait_ordered_extents(struct btrfs_root *root, int nr,
+ const u64 range_start, const u64 range_len)
{
- struct list_head splice, works;
+ LIST_HEAD(splice);
+ LIST_HEAD(skipped);
+ LIST_HEAD(works);
struct btrfs_ordered_extent *ordered, *next;
int count = 0;
-
- INIT_LIST_HEAD(&splice);
- INIT_LIST_HEAD(&works);
+ const u64 range_end = range_start + range_len;
mutex_lock(&root->ordered_extent_mutex);
spin_lock(&root->ordered_extent_lock);
@@ -676,6 +677,14 @@
while (!list_empty(&splice) && nr) {
ordered = list_first_entry(&splice, struct btrfs_ordered_extent,
root_extent_list);
+
+ if (range_end <= ordered->start ||
+ ordered->start + ordered->disk_len <= range_start) {
+ list_move_tail(&ordered->root_extent_list, &skipped);
+ cond_resched_lock(&root->ordered_extent_lock);
+ continue;
+ }
+
list_move_tail(&ordered->root_extent_list,
&root->ordered_extents);
atomic_inc(&ordered->refs);
@@ -694,6 +703,7 @@
nr--;
count++;
}
+ list_splice_tail(&skipped, &root->ordered_extents);
list_splice_tail(&splice, &root->ordered_extents);
spin_unlock(&root->ordered_extent_lock);
@@ -708,7 +718,8 @@
return count;
}
-void btrfs_wait_ordered_roots(struct btrfs_fs_info *fs_info, int nr)
+void btrfs_wait_ordered_roots(struct btrfs_fs_info *fs_info, int nr,
+ const u64 range_start, const u64 range_len)
{
struct btrfs_root *root;
struct list_head splice;
@@ -728,7 +739,8 @@
&fs_info->ordered_roots);
spin_unlock(&fs_info->ordered_root_lock);
- done = btrfs_wait_ordered_extents(root, nr);
+ done = btrfs_wait_ordered_extents(root, nr,
+ range_start, range_len);
btrfs_put_fs_root(root);
spin_lock(&fs_info->ordered_root_lock);
diff --git a/fs/btrfs/ordered-data.h b/fs/btrfs/ordered-data.h
index 23c96059..8ef1262 100644
--- a/fs/btrfs/ordered-data.h
+++ b/fs/btrfs/ordered-data.h
@@ -197,8 +197,10 @@
struct btrfs_ordered_extent *ordered);
int btrfs_find_ordered_sum(struct inode *inode, u64 offset, u64 disk_bytenr,
u32 *sum, int len);
-int btrfs_wait_ordered_extents(struct btrfs_root *root, int nr);
-void btrfs_wait_ordered_roots(struct btrfs_fs_info *fs_info, int nr);
+int btrfs_wait_ordered_extents(struct btrfs_root *root, int nr,
+ const u64 range_start, const u64 range_len);
+void btrfs_wait_ordered_roots(struct btrfs_fs_info *fs_info, int nr,
+ const u64 range_start, const u64 range_len);
void btrfs_get_logged_extents(struct inode *inode,
struct list_head *logged_list,
const loff_t start,
diff --git a/fs/btrfs/relocation.c b/fs/btrfs/relocation.c
index 08ef890..30f77ed 100644
--- a/fs/btrfs/relocation.c
+++ b/fs/btrfs/relocation.c
@@ -4259,7 +4259,9 @@
err = ret;
goto out;
}
- btrfs_wait_ordered_roots(fs_info, -1);
+ btrfs_wait_ordered_roots(fs_info, -1,
+ rc->block_group->key.objectid,
+ rc->block_group->key.offset);
while (1) {
mutex_lock(&fs_info->cleaner_mutex);
diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c
index 00b8f37..89d1347 100644
--- a/fs/btrfs/super.c
+++ b/fs/btrfs/super.c
@@ -1160,7 +1160,7 @@
return 0;
}
- btrfs_wait_ordered_roots(fs_info, -1);
+ btrfs_wait_ordered_roots(fs_info, -1, 0, (u64)-1);
trans = btrfs_attach_transaction_barrier(root);
if (IS_ERR(trans)) {
diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c
index 43885e5..f0bb54a 100644
--- a/fs/btrfs/transaction.c
+++ b/fs/btrfs/transaction.c
@@ -1821,7 +1821,7 @@
static inline void btrfs_wait_delalloc_flush(struct btrfs_fs_info *fs_info)
{
if (btrfs_test_opt(fs_info->tree_root, FLUSHONCOMMIT))
- btrfs_wait_ordered_roots(fs_info, -1);
+ btrfs_wait_ordered_roots(fs_info, -1, 0, (u64)-1);
}
static inline void