Btrfs: merge pending IO for tree log write back
Before applying this patch, we flushed the log tree of the fs/file
tree firstly, and then flushed the log root tree. It is ineffective,
especially on the hard disk. This patch improved this problem by wrapping
the above two flushes by the same blk_plug.
By test, the performance of the sync write went up ~60%(2.9MB/s -> 4.6MB/s)
on my scsi disk whose disk buffer was enabled.
Test step:
# mkfs.btrfs -f -m single <disk>
# mount <disk> <mnt>
# dd if=/dev/zero of=<mnt>/file0 bs=32K count=1024 oflag=sync
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c
index eec8686..8c8b800 100644
--- a/fs/btrfs/transaction.c
+++ b/fs/btrfs/transaction.c
@@ -778,9 +778,7 @@
struct extent_state *cached_state = NULL;
u64 start = 0;
u64 end;
- struct blk_plug plug;
- blk_start_plug(&plug);
while (!find_first_extent_bit(dirty_pages, start, &start, &end,
mark, &cached_state)) {
convert_extent_bit(dirty_pages, start, end, EXTENT_NEED_WAIT,
@@ -794,7 +792,6 @@
}
if (err)
werr = err;
- blk_finish_plug(&plug);
return werr;
}
@@ -839,8 +836,11 @@
{
int ret;
int ret2;
+ struct blk_plug plug;
+ blk_start_plug(&plug);
ret = btrfs_write_marked_extents(root, dirty_pages, mark);
+ blk_finish_plug(&plug);
ret2 = btrfs_wait_marked_extents(root, dirty_pages, mark);
if (ret)
diff --git a/fs/btrfs/tree-log.c b/fs/btrfs/tree-log.c
index 965d5e6..831ddd4 100644
--- a/fs/btrfs/tree-log.c
+++ b/fs/btrfs/tree-log.c
@@ -18,6 +18,7 @@
#include <linux/sched.h>
#include <linux/slab.h>
+#include <linux/blkdev.h>
#include <linux/list_sort.h>
#include "ctree.h"
#include "transaction.h"
@@ -2353,6 +2354,7 @@
struct btrfs_root *log = root->log_root;
struct btrfs_root *log_root_tree = root->fs_info->log_root_tree;
unsigned long log_transid = 0;
+ struct blk_plug plug;
mutex_lock(&root->log_mutex);
log_transid = root->log_transid;
@@ -2396,8 +2398,10 @@
/* we start IO on all the marked extents here, but we don't actually
* wait for them until later.
*/
+ blk_start_plug(&plug);
ret = btrfs_write_marked_extents(log, &log->dirty_log_pages, mark);
if (ret) {
+ blk_finish_plug(&plug);
btrfs_abort_transaction(trans, root, ret);
btrfs_free_logged_extents(log, log_transid);
mutex_unlock(&root->log_mutex);
@@ -2432,6 +2436,7 @@
}
if (ret) {
+ blk_finish_plug(&plug);
if (ret != -ENOSPC) {
btrfs_abort_transaction(trans, root, ret);
mutex_unlock(&log_root_tree->log_mutex);
@@ -2447,6 +2452,7 @@
index2 = log_root_tree->log_transid % 2;
if (atomic_read(&log_root_tree->log_commit[index2])) {
+ blk_finish_plug(&plug);
btrfs_wait_marked_extents(log, &log->dirty_log_pages, mark);
wait_log_commit(trans, log_root_tree,
log_root_tree->log_transid);
@@ -2469,6 +2475,7 @@
* check the full commit flag again
*/
if (root->fs_info->last_trans_log_full_commit == trans->transid) {
+ blk_finish_plug(&plug);
btrfs_wait_marked_extents(log, &log->dirty_log_pages, mark);
btrfs_free_logged_extents(log, log_transid);
mutex_unlock(&log_root_tree->log_mutex);
@@ -2476,9 +2483,10 @@
goto out_wake_log_root;
}
- ret = btrfs_write_and_wait_marked_extents(log_root_tree,
- &log_root_tree->dirty_log_pages,
- EXTENT_DIRTY | EXTENT_NEW);
+ ret = btrfs_write_marked_extents(log_root_tree,
+ &log_root_tree->dirty_log_pages,
+ EXTENT_DIRTY | EXTENT_NEW);
+ blk_finish_plug(&plug);
if (ret) {
btrfs_abort_transaction(trans, root, ret);
btrfs_free_logged_extents(log, log_transid);
@@ -2486,6 +2494,9 @@
goto out_wake_log_root;
}
btrfs_wait_marked_extents(log, &log->dirty_log_pages, mark);
+ btrfs_wait_marked_extents(log_root_tree,
+ &log_root_tree->dirty_log_pages,
+ EXTENT_NEW | EXTENT_DIRTY);
btrfs_wait_logged_extents(log, log_transid);
btrfs_set_super_log_root(root->fs_info->super_for_commit,