[9/9] nvme: implement multipath access to nvme subsystems

Commit Message

Keith Busch Sept. 22, 2017, 3:09 p.m. UTC

On Mon, Sep 18, 2017 at 04:14:53PM -0700, Christoph Hellwig wrote:
> +static void nvme_failover_req(struct request *req)
> +{
> +	struct nvme_ns *ns = req->q->queuedata;
> +	unsigned long flags;
> +
> +	spin_lock_irqsave(&ns->head->requeue_lock, flags);
> +	blk_steal_bios(&ns->head->requeue_list, req);
> +	spin_unlock_irqrestore(&ns->head->requeue_lock, flags);
> +
> +	nvme_reset_ctrl(ns->ctrl);
> +	kblockd_schedule_work(&ns->head->requeue_work);
> +}

Need to call blk_mq_free_req after stealing all its bios to prevent
leaking that entered request.

All together, I needed just two small changes to the entire series to
get everything testing successfully. Here's the diff on top of your set:

---
--

Comments

Christoph Hellwig Sept. 24, 2017, 2:48 p.m. UTC | #1

On Fri, Sep 22, 2017 at 11:09:16AM -0400, Keith Busch wrote:
> On Mon, Sep 18, 2017 at 04:14:53PM -0700, Christoph Hellwig wrote:
> > +static void nvme_failover_req(struct request *req)
> > +{
> > +	struct nvme_ns *ns = req->q->queuedata;
> > +	unsigned long flags;
> > +
> > +	spin_lock_irqsave(&ns->head->requeue_lock, flags);
> > +	blk_steal_bios(&ns->head->requeue_list, req);
> > +	spin_unlock_irqrestore(&ns->head->requeue_lock, flags);
> > +
> > +	nvme_reset_ctrl(ns->ctrl);
> > +	kblockd_schedule_work(&ns->head->requeue_work);
> > +}
> 
> Need to call blk_mq_free_req after stealing all its bios to prevent
> leaking that entered request.

I think this should be a blk_mq_end_request actually.  The difference
is that blk_mq_end_request will get the I/O accounting right, and
treats the case of having and ->end_io handler correctly as well.

Patch

diff --git a/drivers/nvme/host/core.c b/drivers/nvme/host/core.c
index 5449c83..55620ba 100644
--- a/drivers/nvme/host/core.c
+++ b/drivers/nvme/host/core.c
@@ -115,6 +115,8 @@  static void nvme_failover_req(struct request *req)
 	blk_steal_bios(&ns->head->requeue_list, req);
 	spin_unlock_irqrestore(&ns->head->requeue_lock, flags);
 
+	blk_mq_free_request(req);
+
 	nvme_reset_ctrl(ns->ctrl);
 	kblockd_schedule_work(&ns->head->requeue_work);
 }
@@ -1935,6 +1937,9 @@  static int nvme_init_subsystem(struct nvme_ctrl *ctrl, struct nvme_id_ctrl *id)
 {
 	struct nvme_subsystem *subsys, *found;
 
+	if (ctrl->identified)
+		return 0;
+
 	subsys = kzalloc(sizeof(*subsys), GFP_KERNEL);
 	if (!subsys)
 		return -ENOMEM;