Skip to contentSkip to conversation

nvme_fc: add message on final controller removal

PATCH New James Smart 2 participants 3 messages Sep 19, 2017 – Sep 20, 2017

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 nvme_fc: add message on final controller removal

Given disparity between class level delete and transport delete, add an info message when the transport controller object deletes.

Signed-off-by James Smart <james.smart at broadcom.com>
---
 drivers/nvme/host/fc.c | 2 ++
 1 file changed, 2 insertions(+)
drivers/nvme/host/fc.c +2 -0
index 2748a84cec3a..c70aa0975361 100644
@@ -1817,6 +1817,8 @@ nvme_fc_ctrl_free(struct kref *ref)
 		container_of(ref, struct nvme_fc_ctrl, ref);
 	unsigned long flags;
 
+	dev_info(ctrl->dev, "NVME-FC{%d}: Controller Removed\n", ctrl->cnum);
+
 	if (ctrl->ctrl.tagset) {
 		blk_cleanup_queue(ctrl->ctrl.connect_q);
 		blk_mq_free_tag_set(&ctrl->tag_set);
Signature · 1 line
2.13.1
#1 lore raw

PATCH 1/1 nvme_fc: add message on final controller removal

Quoted 2 lines · James Smart

Given disparity between class level delete and transport delete, add an info message when the transport controller object deletes.

Hmm. Why do we need this in the FC code?

#2 lore raw

PATCH 1/1 nvme_fc: add message on final controller removal

Quoted 6 lines · Christoph Hellwig
Quoted 2 lines · James Smart

Given disparity between class level delete and transport delete, add an info message when the transport controller object deletes.

Hmm. Why do we need this in the FC code?

It's a debug aid. I'm finding all the transitions spit out messages saying controller is deleting, which makes the admin think it's deleted. But what really happened was the delete calls were made to the core routines, but those routines are getting hung up in ns teardown. So admins are reporting things as "deleted" when they really aren't yet. This message just book-ends the initial deleting messages.

it could certainly go in the core nvme_free_ctrl() call before it calls the transport free routine.

-- james