lightnvm: control life of nvm_dev in driver

LightNVM compatible device drivers does not have a method to expose
LightNVM specific sysfs entries.

To enable LightNVM sysfs entries to be exposed, lightnvm device
drivers require a struct device to attach it to. To allow both the
actual device driver and lightnvm sysfs entries to coexist, the device
driver tracks the lifetime of the nvm_dev structure.

This patch refactors NVMe and null_blk to handle the lifetime of struct
nvm_dev, which eliminates the need for struct gendisk when a lightnvm
compatible device is provided.

Signed-off-by: Matias Bjørling <m@bjorling.me>
Signed-off-by: Jens Axboe <axboe@fb.com>

5 files changed, 74 insertions, 62 deletions

diff --git a/drivers/block/null_blk.c b/drivers/block/null_blk.c
index 895867a..91e1de8 100644
--- a/drivers/block/null_blk.c
+++ b/drivers/block/null_blk.c
@@ -34,6 +34,7 @@ struct nullb {
 	unsigned int index;
 	struct request_queue *q;
 	struct gendisk *disk;
+	struct nvm_dev *ndev;
 	struct blk_mq_tag_set tag_set;
 	struct hrtimer timer;
 	unsigned int queue_depth;
@@ -550,12 +551,29 @@ static struct nvm_dev_ops null_lnvm_dev_ops = {
 
 static int null_nvm_register(struct nullb *nullb)
 {
-	return nvm_register(nullb->q, nullb->disk_name, &null_lnvm_dev_ops);
+	struct nvm_dev *dev;
+	int rv;
+
+	dev = nvm_alloc_dev(0);
+	if (!dev)
+		return -ENOMEM;
+
+	dev->q = nullb->q;
+	memcpy(dev->name, nullb->disk_name, DISK_NAME_LEN);
+	dev->ops = &null_lnvm_dev_ops;
+
+	rv = nvm_register(dev);
+	if (rv) {
+		kfree(dev);
+		return rv;
+	}
+	nullb->ndev = dev;
+	return 0;
 }
 
 static void null_nvm_unregister(struct nullb *nullb)
 {
-	nvm_unregister(nullb->disk_name);
+	nvm_unregister(nullb->ndev);
 }
 #else
 static int null_nvm_register(struct nullb *nullb)
diff --git a/drivers/lightnvm/core.c b/drivers/lightnvm/core.c
index 25c5df9..a99b59d 100644
--- a/drivers/lightnvm/core.c
+++ b/drivers/lightnvm/core.c
@@ -660,22 +660,15 @@ static void nvm_exit(struct nvm_dev *dev)
 	pr_info("nvm: successfully unloaded\n");
 }
 
-int nvm_register(struct request_queue *q, char *disk_name,
-							struct nvm_dev_ops *ops)
+struct nvm_dev *nvm_alloc_dev(int node)
 {
-	struct nvm_dev *dev;
-	int ret;
-
-	if (!ops->identity)
-		return -EINVAL;
-
-	dev = kzalloc(sizeof(struct nvm_dev), GFP_KERNEL);
-	if (!dev)
-		return -ENOMEM;
+	return kzalloc_node(sizeof(struct nvm_dev), GFP_KERNEL, node);
+}
+EXPORT_SYMBOL(nvm_alloc_dev);
 
-	dev->q = q;
-	dev->ops = ops;
-	strncpy(dev->name, disk_name, DISK_NAME_LEN);
+int nvm_register(struct nvm_dev *dev)
+{
+	int ret;
 
 	ret = nvm_init(dev);
 	if (ret)
@@ -714,29 +707,17 @@ int nvm_register(struct request_queue *q, char *disk_name,
 	return 0;
 err_init:
 	kfree(dev->lun_map);
-	kfree(dev);
 	return ret;
 }
 EXPORT_SYMBOL(nvm_register);
 
-void nvm_unregister(char *disk_name)
+void nvm_unregister(struct nvm_dev *dev)
 {
-	struct nvm_dev *dev;
-
 	down_write(&nvm_lock);
-	dev = nvm_find_nvm_dev(disk_name);
-	if (!dev) {
-		pr_err("nvm: could not find device %s to unregister\n",
-								disk_name);
-		up_write(&nvm_lock);
-		return;
-	}
-
 	list_del(&dev->devices);
 	up_write(&nvm_lock);
 
 	nvm_exit(dev);
-	kfree(dev);
 }
 EXPORT_SYMBOL(nvm_unregister);
 
diff --git a/drivers/nvme/host/core.c b/drivers/nvme/host/core.c
index 2c3da33..3c707d8 100644
--- a/drivers/nvme/host/core.c
+++ b/drivers/nvme/host/core.c
@@ -156,12 +156,14 @@ static void nvme_free_ns(struct kref *kref)
 {
 	struct nvme_ns *ns = container_of(kref, struct nvme_ns, kref);
 
-	if (ns->type == NVME_NS_LIGHTNVM)
-		nvme_nvm_unregister(ns->queue, ns->disk->disk_name);
+	if (ns->ndev)
+		nvme_nvm_unregister(ns);
 
-	spin_lock(&dev_list_lock);
-	ns->disk->private_data = NULL;
-	spin_unlock(&dev_list_lock);
+	if (ns->disk) {
+		spin_lock(&dev_list_lock);
+		ns->disk->private_data = NULL;
+		spin_unlock(&dev_list_lock);
+	}
 
 	put_disk(ns->disk);
 	ida_simple_remove(&ns->ctrl->ns_ida, ns->instance);
@@ -891,8 +893,7 @@ static void nvme_config_discard(struct nvme_ns *ns)
 static int nvme_revalidate_ns(struct nvme_ns *ns, struct nvme_id_ns **id)
 {
 	if (nvme_identify_ns(ns->ctrl, ns->ns_id, id)) {
-		dev_warn(disk_to_dev(ns->disk), "%s: Identify failure\n",
-				__func__);
+		dev_warn(ns->ctrl->dev, "%s: Identify failure\n", __func__);
 		return -ENODEV;
 	}
 
@@ -1683,18 +1684,11 @@ static void nvme_alloc_ns(struct nvme_ctrl *ctrl, unsigned nsid)
 		goto out_free_queue;
 
 	if (nvme_nvm_ns_supported(ns, id)) {
-		if (nvme_nvm_register(ns->queue, disk_name)) {
-			dev_warn(ctrl->dev,
-				"%s: LightNVM init failure\n", __func__);
+		if (nvme_nvm_register(ns, disk_name, node)) {
+			dev_warn(ctrl->dev, "%s: LightNVM init failure\n",
+								__func__);
 			goto out_free_id;
 		}
-
-		disk = alloc_disk_node(0, node);
-		if (!disk)
-			goto out_free_id;
-		memcpy(disk->disk_name, disk_name, DISK_NAME_LEN);
-		ns->disk = disk;
-		ns->type = NVME_NS_LIGHTNVM;
 	} else {
 		disk = alloc_disk_node(0, node);
 		if (!disk)
@@ -1718,7 +1712,7 @@ static void nvme_alloc_ns(struct nvme_ctrl *ctrl, unsigned nsid)
 
 	kfree(id);
 
-	if (ns->type == NVME_NS_LIGHTNVM)
+	if (ns->ndev)
 		return;
 
 	device_add_disk(ctrl->device, ns->disk);
@@ -1742,7 +1736,7 @@ static void nvme_ns_remove(struct nvme_ns *ns)
 	if (test_and_set_bit(NVME_NS_REMOVING, &ns->flags))
 		return;
 
-	if (ns->disk->flags & GENHD_FL_UP) {
+	if (ns->disk && ns->disk->flags & GENHD_FL_UP) {
 		if (blk_get_integrity(ns->disk))
 			blk_integrity_unregister(ns->disk);
 		sysfs_remove_group(&disk_to_dev(ns->disk)->kobj,
@@ -1765,7 +1759,7 @@ static void nvme_validate_ns(struct nvme_ctrl *ctrl, unsigned nsid)
 
 	ns = nvme_find_get_ns(ctrl, nsid);
 	if (ns) {
-		if (revalidate_disk(ns->disk))
+		if (ns->disk && revalidate_disk(ns->disk))
 			nvme_ns_remove(ns);
 		nvme_put_ns(ns);
 	} else
@@ -2070,7 +2064,7 @@ void nvme_kill_queues(struct nvme_ctrl *ctrl)
 		 * Revalidating a dead namespace sets capacity to 0. This will
 		 * end buffered writers dirtying pages that can't be synced.
 		 */
-		if (!test_and_set_bit(NVME_NS_DEAD, &ns->flags))
+		if (ns->disk && !test_and_set_bit(NVME_NS_DEAD, &ns->flags))
 			revalidate_disk(ns->disk);
 
 		blk_set_queue_dying(ns->queue);
diff --git a/drivers/nvme/host/lightnvm.c b/drivers/nvme/host/lightnvm.c
index 7268a7a..798fcd9 100644
--- a/drivers/nvme/host/lightnvm.c
+++ b/drivers/nvme/host/lightnvm.c
@@ -474,9 +474,8 @@ static inline void nvme_nvm_rqtocmd(struct request *rq, struct nvm_rq *rqd,
 	c->ph_rw.length = cpu_to_le16(rqd->nr_ppas - 1);
 
 	if (rqd->opcode == NVM_OP_HBWRITE || rqd->opcode == NVM_OP_HBREAD)
-		/* momentarily hardcode the shift configuration. lba_shift from
-		 * nvm_dev will be available in a follow-up patch */
-		c->hb_rw.slba = cpu_to_le64(rqd->bio->bi_iter.bi_sector >> 3);
+		c->hb_rw.slba = cpu_to_le64(nvme_block_nr(ns,
+					rqd->bio->bi_iter.bi_sector));
 }
 
 static void nvme_nvm_end_io(struct request *rq, int error)
@@ -593,14 +592,32 @@ static struct nvm_dev_ops nvme_nvm_dev_ops = {
 	.max_phys_sect		= 64,
 };
 
-int nvme_nvm_register(struct request_queue *q, char *disk_name)
+int nvme_nvm_register(struct nvme_ns *ns, char *disk_name, int node)
 {
-	return nvm_register(q, disk_name, &nvme_nvm_dev_ops);
+	struct request_queue *q = ns->queue;
+	struct nvm_dev *dev;
+	int ret;
+
+	dev = nvm_alloc_dev(node);
+	if (!dev)
+		return -ENOMEM;
+
+	dev->q = q;
+	memcpy(dev->name, disk_name, DISK_NAME_LEN);
+	dev->ops = &nvme_nvm_dev_ops;
+	ns->ndev = dev;
+
+	ret = nvm_register(dev);
+
+	ns->lba_shift = ilog2(dev->sec_size) - 9;
+
+	return ret;
 }
 
-void nvme_nvm_unregister(struct request_queue *q, char *disk_name)
+void nvme_nvm_unregister(struct nvme_ns *ns)
 {
-	nvm_unregister(disk_name);
+	nvm_unregister(ns->ndev);
+	kfree(ns->ndev);
 }
 
 /* move to shared place when used in multiple places. */
diff --git a/drivers/nvme/host/nvme.h b/drivers/nvme/host/nvme.h
index ab18b78..e0535c1 100644
--- a/drivers/nvme/host/nvme.h
+++ b/drivers/nvme/host/nvme.h
@@ -18,6 +18,7 @@
 #include <linux/pci.h>
 #include <linux/kref.h>
 #include <linux/blk-mq.h>
+#include <linux/lightnvm.h>
 
 enum {
 	/*
@@ -154,6 +155,7 @@ struct nvme_ns {
 	struct nvme_ctrl *ctrl;
 	struct request_queue *queue;
 	struct gendisk *disk;
+	struct nvm_dev *ndev;
 	struct kref kref;
 	int instance;
 
@@ -165,7 +167,6 @@ struct nvme_ns {
 	u16 ms;
 	bool ext;
 	u8 pi_type;
-	int type;
 	unsigned long flags;
 
 #define NVME_NS_REMOVING 0
@@ -307,15 +308,16 @@ int nvme_sg_get_version_num(int __user *ip);
 
 #ifdef CONFIG_NVM
 int nvme_nvm_ns_supported(struct nvme_ns *ns, struct nvme_id_ns *id);
-int nvme_nvm_register(struct request_queue *q, char *disk_name);
-void nvme_nvm_unregister(struct request_queue *q, char *disk_name);
+int nvme_nvm_register(struct nvme_ns *ns, char *disk_name, int node);
+void nvme_nvm_unregister(struct nvme_ns *ns);
 #else
-static inline int nvme_nvm_register(struct request_queue *q, char *disk_name)
+static inline int nvme_nvm_register(struct nvme_ns *ns, char *disk_name,
+								int node)
 {
 	return 0;
 }
 
-static inline void nvme_nvm_unregister(struct request_queue *q, char *disk_name) {};
+static inline void nvme_nvm_unregister(struct nvme_ns *ns) {};
 
 static inline int nvme_nvm_ns_supported(struct nvme_ns *ns, struct nvme_id_ns *id)
 {