Re: [PATCH RFC 0/3] blk-mq/nvme: use blk_mq_alloc_request() for NVMe's connect request

On 2019-11-15 23:17, Ming Lei wrote:

Now blk-mq takes a static queue mapping between CPU and hw queues, given
CPU hotplug may happen any time, so the specified hw queue may become
inactive any time.

Hi Ming,

I can trigger a race between blk_mq_alloc_request_hctx() and
CPU-hotplugging by running blktests. The patch below fixes that race
on my setup. Does this patch also fix the race(s) that you ran into?

Thanks,

Bart.


Subject: [PATCH] blk-mq: Fix a race between blk_mq_alloc_request_hctx() and
 CPU hot-plugging

---
 block/blk-mq.c | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

diff --git a/block/blk-mq.c b/block/blk-mq.c
index 20a71dcdc339..16057aa2307f 100644
--- a/block/blk-mq.c
+++ b/block/blk-mq.c

@@ -442,13 +442,15 @@ struct request *blk_mq_alloc_request_hctx(struct request_queue *q,
 	if (WARN_ON_ONCE(!(flags & BLK_MQ_REQ_NOWAIT)))
 		return ERR_PTR(-EINVAL);

-	if (hctx_idx >= q->nr_hw_queues)
-		return ERR_PTR(-EIO);
-
 	ret = blk_queue_enter(q, flags);
 	if (ret)
 		return ERR_PTR(ret);

+	if (hctx_idx >= q->nr_hw_queues) {
+		blk_queue_exit(q);
+		return ERR_PTR(-EIO);
+	}
+
 	/*
 	 * Check if the hardware context is actually mapped to anything.
 	 * If not tell the caller that it should skip this queue.