On 8/9/18 2:17 PM, Bart Van Assche wrote:

On Thu, 2018-08-09 at 12:56 -0700, Tejun Heo wrote:

quoted

Hello,

On Thu, Aug 09, 2018 at 07:53:36AM -0700, Bart Van Assche wrote:

quoted

+/**
+ * blkg_lookup - look up blkg for the specified request queue
+ * @q: request_queue of interest
+ *
+ * Lookup blkg for @q at the root level. See also blkg_lookup().
+ */
+static inline struct blkcg_gq *blkg_root_lookup(struct request_queue *q)
+{
+	struct blkcg_gq *blkg;
+
+	rcu_read_lock();
+	blkg = blkg_lookup(&blkcg_root, q);
+	rcu_read_unlock();
+
+	return blkg;
+}

So, root blkg is always available as long as the request_queue is
alive.  Sth like the following would be simpler?

static inline struct blkcg_gq *blk_queue_root_blkg(struct request_queue *q)
{
	return q->root_blkg;
}

That would not only be simpler, it would also avoid that blk_queue_root_blkg()
returns NULL if bypass mode is enabled and the request queue is still associated
with the block cgroup controller. How do you want to proceed with this change?

Do a followup patch?

-- 
Jens Axboe