block: loop: convert to per-device workqueue

Documentation/workqueue.txt:
	If there is dependency among multiple work items used
	during memory reclaim, they should be queued to separate
	wq each with WQ_MEM_RECLAIM.

Loop devices can be stacked, so we have to convert to per-device
workqueue. One example is Fedora live CD.

Fixes: b5dd2f6047ca108001328aac0e8588edd15f1778
Cc: stable@vger.kernel.org (v4.0)
Cc: Justin M. Forbes <jforbes@fedoraproject.org>
Signed-off-by: Ming Lei <ming.lei@canonical.com>
Acked-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <axboe@fb.com>
diff --git a/drivers/block/loop.c b/drivers/block/loop.c
index ae3fcb4..3dc1598 100644
--- a/drivers/block/loop.c
+++ b/drivers/block/loop.c
@@ -86,8 +86,6 @@
 static int max_part;
 static int part_shift;
 
-static struct workqueue_struct *loop_wq;
-
 static int transfer_xor(struct loop_device *lo, int cmd,
 			struct page *raw_page, unsigned raw_off,
 			struct page *loop_page, unsigned loop_off,
@@ -725,6 +723,12 @@
 	size = get_loop_size(lo, file);
 	if ((loff_t)(sector_t)size != size)
 		goto out_putf;
+	error = -ENOMEM;
+	lo->wq = alloc_workqueue("kloopd%d",
+			WQ_MEM_RECLAIM | WQ_HIGHPRI | WQ_UNBOUND, 0,
+			lo->lo_number);
+	if (!lo->wq)
+		goto out_putf;
 
 	error = 0;
 
@@ -872,6 +876,8 @@
 	lo->lo_flags = 0;
 	if (!part_shift)
 		lo->lo_disk->flags |= GENHD_FL_NO_PART_SCAN;
+	destroy_workqueue(lo->wq);
+	lo->wq = NULL;
 	mutex_unlock(&lo->lo_ctl_mutex);
 	/*
 	 * Need not hold lo_ctl_mutex to fput backing file.
@@ -1425,9 +1431,13 @@
 		const struct blk_mq_queue_data *bd)
 {
 	struct loop_cmd *cmd = blk_mq_rq_to_pdu(bd->rq);
+	struct loop_device *lo = cmd->rq->q->queuedata;
 
 	blk_mq_start_request(bd->rq);
 
+	if (lo->lo_state != Lo_bound)
+		return -EIO;
+
 	if (cmd->rq->cmd_flags & REQ_WRITE) {
 		struct loop_device *lo = cmd->rq->q->queuedata;
 		bool need_sched = true;
@@ -1441,9 +1451,9 @@
 		spin_unlock_irq(&lo->lo_lock);
 
 		if (need_sched)
-			queue_work(loop_wq, &lo->write_work);
+			queue_work(lo->wq, &lo->write_work);
 	} else {
-		queue_work(loop_wq, &cmd->read_work);
+		queue_work(lo->wq, &cmd->read_work);
 	}
 
 	return BLK_MQ_RQ_QUEUE_OK;
@@ -1455,9 +1465,6 @@
 	struct loop_device *lo = cmd->rq->q->queuedata;
 	int ret = -EIO;
 
-	if (lo->lo_state != Lo_bound)
-		goto failed;
-
 	if (write && (lo->lo_flags & LO_FLAGS_READ_ONLY))
 		goto failed;
 
@@ -1806,13 +1813,6 @@
 		goto misc_out;
 	}
 
-	loop_wq = alloc_workqueue("kloopd",
-			WQ_MEM_RECLAIM | WQ_HIGHPRI | WQ_UNBOUND, 0);
-	if (!loop_wq) {
-		err = -ENOMEM;
-		goto misc_out;
-	}
-
 	blk_register_region(MKDEV(LOOP_MAJOR, 0), range,
 				  THIS_MODULE, loop_probe, NULL, NULL);
 
@@ -1850,8 +1850,6 @@
 	blk_unregister_region(MKDEV(LOOP_MAJOR, 0), range);
 	unregister_blkdev(LOOP_MAJOR, "loop");
 
-	destroy_workqueue(loop_wq);
-
 	misc_deregister(&loop_misc);
 }
 
diff --git a/drivers/block/loop.h b/drivers/block/loop.h
index 301c27f..49564ed 100644
--- a/drivers/block/loop.h
+++ b/drivers/block/loop.h
@@ -54,6 +54,7 @@
 	gfp_t		old_gfp_mask;
 
 	spinlock_t		lo_lock;
+	struct workqueue_struct *wq;
 	struct list_head	write_cmd_head;
 	struct work_struct	write_work;
 	bool			write_started;