NVMe: Use IDA for namespace disk naming

PATCH New Keith Busch 3 participants 8 messages Feb 18, 2016 – Feb 25, 2016 NVMe

Share & tools

Keith Busch

author

2016-02-18 18:18 UTC #0 lore raw

PATCH 1/1 NVMe: Use IDA for namespace disk naming

A namespace may be detached from a controller, but a user may be holding a reference to it. Attaching a new namespace with the same NSID will create duplicate names when using the NSID to name the disk.

This patch uses an IDA that is released only when the last reference is released instead of using the namespace ID.

Signed-off-by Keith Busch <keith.busch at intel.com>

---
 drivers/nvme/host/core.c | 25 ++++++++++++++++++-------
 drivers/nvme/host/nvme.h |  2 ++
 2 files changed, 20 insertions(+), 7 deletions(-)

drivers/nvme/host/core.c +18 -7

index b29fae6..9b1253f 100644
@@ -48,10 +48,6 @@ static void nvme_free_ns(struct kref *kref)
 {
 	struct nvme_ns *ns = container_of(kref, struct nvme_ns, kref);
 
-	mutex_lock(&ns->ctrl->namespaces_mutex);
-	list_del_init(&ns->list);
-	mutex_unlock(&ns->ctrl->namespaces_mutex);
-
 	if (ns->type == NVME_NS_LIGHTNVM)
 		nvme_nvm_unregister(ns->queue, ns->disk->disk_name);

@@ -59,8 +55,9 @@ static void nvme_free_ns(struct kref *kref)
 	ns->disk->private_data = NULL;
 	spin_unlock(&dev_list_lock);
 
-	nvme_put_ctrl(ns->ctrl);
 	put_disk(ns->disk);
+	ida_simple_remove(&ns->ctrl->ns_ida, ns->instance);
+	nvme_put_ctrl(ns->ctrl);
 	kfree(ns);
 }

@@ -1152,9 +1149,13 @@ static void nvme_alloc_ns(struct nvme_ctrl *ctrl, unsigned nsid)
 	if (!ns)
 		return;
 
+	ns->instance = ida_simple_get(&ctrl->ns_ida, 1, 0, GFP_KERNEL);
+	if (ns->instance < 0)
+		goto out_free_ns;
+
 	ns->queue = blk_mq_init_queue(ctrl->tagset);
 	if (IS_ERR(ns->queue))
-		goto out_free_ns;
+		goto out_release_instance;
 	queue_flag_set_unlocked(QUEUE_FLAG_NONROT, ns->queue);
 	ns->queue->queuedata = ns;
 	ns->ctrl = ctrl;

@@ -1187,7 +1188,7 @@ static void nvme_alloc_ns(struct nvme_ctrl *ctrl, unsigned nsid)
 	disk->queue = ns->queue;
 	disk->driverfs_dev = ctrl->device;
 	disk->flags = GENHD_FL_EXT_DEVT;
-	sprintf(disk->disk_name, "nvme%dn%d", ctrl->instance, nsid);
+	sprintf(disk->disk_name, "nvme%dn%d", ctrl->instance, ns->instance);
 
 	if (nvme_revalidate_disk(ns->disk))
 		goto out_free_disk;

@@ -1207,6 +1208,8 @@ static void nvme_alloc_ns(struct nvme_ctrl *ctrl, unsigned nsid)
 	kfree(disk);
  out_free_queue:
 	blk_cleanup_queue(ns->queue);
+ out_release_instance:
+	ida_simple_remove(&ctrl->ns_ida, ns->instance);
  out_free_ns:
 	kfree(ns);
 }

@@ -1225,6 +1228,11 @@ static void nvme_ns_remove(struct nvme_ns *ns)
 		blk_mq_abort_requeue_list(ns->queue);
 		blk_cleanup_queue(ns->queue);
 	}
+
+	mutex_lock(&ns->ctrl->namespaces_mutex);
+	list_del_init(&ns->list);
+	mutex_unlock(&ns->ctrl->namespaces_mutex);
+
 	nvme_put_ns(ns);
 }

@@ -1358,6 +1366,8 @@ void nvme_uninit_ctrl(struct nvme_ctrl *ctrl)
 	spin_lock(&dev_list_lock);
 	list_del(&ctrl->node);
 	spin_unlock(&dev_list_lock);
+
+	ida_destroy(&ctrl->ns_ida);
 }
 
 static void nvme_free_ctrl(struct kref *kref)

@@ -1406,6 +1416,7 @@ int nvme_init_ctrl(struct nvme_ctrl *ctrl, struct device *dev,
 	}
 	get_device(ctrl->device);
 	dev_set_drvdata(ctrl->device, ctrl);
+	ida_init(&ctrl->ns_ida);
 
 	spin_lock(&dev_list_lock);
 	list_add_tail(&ctrl->node, &nvme_ctrl_list);

drivers/nvme/host/nvme.h +2 -0

index b50e591..3b0f84a 100644
@@ -72,6 +72,7 @@ struct nvme_ctrl {
 	struct mutex namespaces_mutex;
 	struct device *device;	/* char device */
 	struct list_head node;
+	struct ida ns_ida;
 
 	char name[12];
 	char serial[20];

@@ -103,6 +104,7 @@ struct nvme_ns {
 	struct request_queue *queue;
 	struct gendisk *disk;
 	struct kref kref;
+	int instance;
 
 	u8 eui[8];
 	u8 uuid[16];

Signature · 1 line

2.6.2.307.g37023ba

Christoph Hellwig

2016-02-21 16:12 UTC #1 lore raw

PATCH 1/1 NVMe: Use IDA for namespace disk naming

Quoted 3 lines · Keith Busch

A namespace may be detached from a controller, but a user may be holding a reference to it. Attaching a new namespace with the same NSID will create duplicate names when using the NSID to name the disk.

This looks sensible, but I fear we might be running into the same issue with the backing_device embedded in the request_queue.

Keith Busch

author

2016-02-23 22:56 UTC #2 lore raw

PATCH 1/1 NVMe: Use IDA for namespace disk naming

Quoted 7 lines · Christoph Hellwig

Quoted 3 lines · Keith Busch
A namespace may be detached from a controller, but a user may be holding a reference to it. Attaching a new namespace with the same NSID will create duplicate names when using the NSID to name the disk.
This looks sensible, but I fear we might be running into the same issue with the backing_device embedded in the request_queue.

I don't think I follow. On rescanning, a new namespace with the same NSID as a previously deleted one will allocate a new request_queue.

Christoph Hellwig

2016-02-24 13:05 UTC #3 lore raw

PATCH 1/1 NVMe: Use IDA for namespace disk naming

Quoted 5 lines · Keith Busch

Quoted 2 lines
This looks sensible, but I fear we might be running into the same issue with the backing_device embedded in the request_queue.
I don't think I follow. On rescanning, a new namespace with the same NSID as a previously deleted one will allocate a new request_queue.

But the bdi name embedds the major/minor of the device. So if we haven't fully torn down the device we might run into sysfs name conflicts.

Keith Busch

author

2016-02-24 14:58 UTC #4 lore raw

PATCH 1/1 NVMe: Use IDA for namespace disk naming

Quoted 3 lines · Christoph Hellwig

But the bdi name embedds the major/minor of the device. So if we haven't fully torn down the device we might run into sysfs name conflicts.

We don't tear down the minor until the final put_disk, fixing the major/minor lifetime issue. That was commit:

https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=2da78092dda13f1efd26edbbf99a567776913750

The namespace's disk_name using the IDA is released in the same path as the final put_disk, so I think we're safe.

Christoph Hellwig

2016-02-24 19:33 UTC #5 lore raw

PATCH 1/1 NVMe: Use IDA for namespace disk naming

Quoted 5 lines · Keith Busch

Quoted 3 lines · Christoph Hellwig
But the bdi name embedds the major/minor of the device. So if we haven't fully torn down the device we might run into sysfs name conflicts.

Quoted 7 lines

We don't tear down the minor until the final put_disk, fixing the major/minor lifetime issue. That was commit:
https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=2da78092dda13f1efd26edbbf99a567776913750
The namespace's disk_name using the IDA is released in the same path as the final put_disk, so I think we're safe.

Ok, we should be fine then:

Reviewed-by Christoph Hellwig <hch at lst.de>

Sagi Grimberg

2016-02-22 08:54 UTC #6 lore raw

PATCH 1/1 NVMe: Use IDA for namespace disk naming

In reply to Keith Busch

This looks good to me,

Reviewed-by Sagi Grimberg <sagig at mellanox.com>

Sagi Grimberg

2016-02-25 08:55 UTC #7 lore raw

PATCH 1/1 NVMe: Use IDA for namespace disk naming

In reply to Keith Busch

Looks fine to me too,

Reviewed-by Sagi Grimberg <sagig at mellanox.com>