Btrfs: remove nr_async_submits and async_submit_draining
authorLiu Bo <bo.li.liu@oracle.com>
Thu, 7 Sep 2017 17:22:22 +0000 (11:22 -0600)
committerDavid Sterba <dsterba@suse.com>
Mon, 30 Oct 2017 11:27:59 +0000 (12:27 +0100)
commit736cd52e0c720103f52ab9da47b6cc3af6b083f6
treeb9918e4c713f73d43d6d702d48099ce241e41829
parent80e03a2c512bd78858c130d6fe3b11982a61c0c5
Btrfs: remove nr_async_submits and async_submit_draining

Now that we have the combo of flushing twice, which can make sure IO
have started since the second flush will wait for page lock which
won't be unlocked unless setting page writeback and queuing ordered
extents, we don't need %async_submit_draining, %async_delalloc_pages
and %nr_async_submits to tell whether the IO has actually started.

Moreover, all the flushers in use are followed by functions that wait
for ordered extents to complete, so %nr_async_submits, which tracks
whether bio's async submit has made progress, doesn't really make
sense.

However, %async_delalloc_pages is still required by shrink_delalloc()
as that function doesn't flush twice in the normal case (just issues a
writeback with WB_REASON_FS_FREE_SPACE).

Signed-off-by: Liu Bo <bo.li.liu@oracle.com>
Signed-off-by: David Sterba <dsterba@suse.com>
fs/btrfs/ctree.h
fs/btrfs/disk-io.c
fs/btrfs/inode.c