提交 18e9781d 编写于 作者: J Jens Axboe

blk-mq: blk_mq_requeue_work() doesn't need to save IRQ flags

We know we're in process context, so don't bother using the
IRQ safe versions of the spin lock.
Signed-off-by: NJens Axboe <axboe@kernel.dk>
上级 33027c2b
...@@ -620,11 +620,10 @@ static void blk_mq_requeue_work(struct work_struct *work) ...@@ -620,11 +620,10 @@ static void blk_mq_requeue_work(struct work_struct *work)
container_of(work, struct request_queue, requeue_work.work); container_of(work, struct request_queue, requeue_work.work);
LIST_HEAD(rq_list); LIST_HEAD(rq_list);
struct request *rq, *next; struct request *rq, *next;
unsigned long flags;
spin_lock_irqsave(&q->requeue_lock, flags); spin_lock_irq(&q->requeue_lock);
list_splice_init(&q->requeue_list, &rq_list); list_splice_init(&q->requeue_list, &rq_list);
spin_unlock_irqrestore(&q->requeue_lock, flags); spin_unlock_irq(&q->requeue_lock);
list_for_each_entry_safe(rq, next, &rq_list, queuelist) { list_for_each_entry_safe(rq, next, &rq_list, queuelist) {
if (!(rq->rq_flags & RQF_SOFTBARRIER)) if (!(rq->rq_flags & RQF_SOFTBARRIER))
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册