[PATCH] cfq-iosched: check busy queues before deciding we are idle

For just one busy queue (like async write out), we often overlooked
that we could queue more io and decided we were idle instead. This causes
us quite a bit of performance loss.

Signed-off-by: Jens Axboe <axboe@suse.de>

diff --git a/block/cfq-iosched.c b/block/cfq-iosched.c
index 11ce6aaf1bd02fdc8063f46566ecfeb2ff389159..d582433441c502f31d238ee2fc974287e2ec5adb 100644 (file)
--- a/block/cfq-iosched.c
+++ b/block/cfq-iosched.c
@@ -878,6 +878,13 @@ static struct cfq_queue *cfq_set_active_queue(struct cfq_data *cfqd)
        if (!list_empty(&cfqd->cur_rr) || cfq_get_next_prio_level(cfqd) != -1)
                cfqq = list_entry_cfqq(cfqd->cur_rr.next);
 
+       /*
+        * If no new queues are available, check if the busy list has some
+        * before falling back to idle io.
+        */
+       if (!cfqq && !list_empty(&cfqd->busy_rr))
+               cfqq = list_entry_cfqq(cfqd->busy_rr.next);
+
        /*
         * if we have idle queues and no rt or be queues had pending
         * requests, either allow immediate service if the grace period