dm thin: wakeup worker only when deferred bios exist

Single thread fio test (read, bs=4k, ioengine=libaio, iodepth=128,
numjobs=1) over dm-thin device has poor performance versus bare nvme
device.

Further investigation with perf indicates that queue_work_on() consumes
over 20% CPU time when doing IO over dm-thin device. The call stack is
as follows.

- 40.57% thin_map
    + 22.07% queue_work_on
    + 9.95% dm_thin_find_block
    + 2.80% cell_defer_no_holder
      1.91% inc_all_io_entry.isra.33.part.34
    + 1.78% bio_detain.isra.35

In cell_defer_no_holder(), wakeup_worker() is always called, no matter
whether the tc->deferred_bio_list list is empty or not. In single thread
IO model, this list is most likely empty. So skip waking up worker thread
if tc->deferred_bio_list list is empty.

Single thread IO performance improves from 448 MiB/s to 646 MiB/s (+44%)
once the needless wake_worker() calls are properly skipped.

Signed-off-by: Jeffle Xu <jefflexu@linux.alibaba.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>

diff --git a/drivers/md/dm-thin.c b/drivers/md/dm-thin.c
index 85ada5ad212115d0cb3dc79f5e9f3a111e52f3d7..5a2c494cb55288e782d0fb4702d4e9efb8cab63a 100644 (file)
--- a/drivers/md/dm-thin.c
+++ b/drivers/md/dm-thin.c
@@ -882,12 +882,15 @@ static void cell_defer_no_holder(struct thin_c *tc, struct dm_bio_prison_cell *c
 {
        struct pool *pool = tc->pool;
        unsigned long flags;
+       int has_work;
 
        spin_lock_irqsave(&tc->lock, flags);
        cell_release_no_holder(pool, cell, &tc->deferred_bio_list);
+       has_work = !bio_list_empty(&tc->deferred_bio_list);
        spin_unlock_irqrestore(&tc->lock, flags);
 
-       wake_worker(pool);
+       if (has_work)
+               wake_worker(pool);
 }
 
 static void thin_defer_bio(struct thin_c *tc, struct bio *bio);