blk-wbt: wake up all when we scale up, not down

Tetsuo brought to my attention that I screwed up the scale_up/scale_down
helpers when I factored out the rq-qos code.  We need to wake up all the
waiters when we add slots for requests to make, not when we shrink the
slots.  Otherwise we'll end up things waiting forever.  This was a
mistake and simply puts everything back the way it was.

cc: stable@vger.kernel.org
Fixes: a790504 ("blk-rq-qos: refactor out common elements of blk-wbt")
eported-by: Tetsuo Handa <penguin-kernel@i-love.sakura.ne.jp>
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block/blk-wbt.c

@@ -310,6 +310,7 @@ static void scale_up(struct rq_wb *rwb)
 	rq_depth_scale_up(&rwb->rq_depth);
 	calc_wb_limits(rwb);
 	rwb->unknown_cnt = 0;
+	rwb_wake_all(rwb);
 	rwb_trace_step(rwb, "scale up");
 }
 
@@ -318,7 +319,6 @@ static void scale_down(struct rq_wb *rwb, bool hard_throttle)
 	rq_depth_scale_down(&rwb->rq_depth, hard_throttle);
 	calc_wb_limits(rwb);
 	rwb->unknown_cnt = 0;
-	rwb_wake_all(rwb);
 	rwb_trace_step(rwb, "scale down");
 }