nvme-fabrics: add a generic NVMe over Fabrics library

The NVMe over Fabrics library provides an interface for both transports
and the nvme core to handle fabrics specific commands and attributes
independent of the underlying transport.

In addition, the fabrics library adds a misc device interface that allow
actually creating a fabrics controller, as we can't just autodiscover
it like in the PCI case.  The nvme-cli utility has been enhanced to use
this interface to support fabric connect and discovery.

Signed-off-by: Armen Baloyan <armenx.baloyan@intel.com>,
Signed-off-by: Jay Freyensee <james.p.freyensee@intel.com>,
Signed-off-by: Ming Lin <ming.l@ssi.samsung.com>
Signed-off-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Keith Busch <keith.busch@intel.com>
Signed-off-by: Jens Axboe <axboe@fb.com>

drivers/nvme/host/Kconfig

@@ -24,3 +24,6 @@ config BLK_DEV_NVME_SCSI
 	  to say N here, unless you run a distro that abuses the SCSI
 	  emulation to provide stable device names for mount by id, like
 	  some OpenSuSE and SLES versions.
+
+config NVME_FABRICS
+	tristate

drivers/nvme/host/Makefile

 obj-$(CONFIG_NVME_CORE)			+= nvme-core.o
 obj-$(CONFIG_BLK_DEV_NVME)		+= nvme.o
+obj-$(CONFIG_NVME_FABRICS)		+= nvme-fabrics.o
 
 nvme-core-y				:= core.o
 nvme-core-$(CONFIG_BLK_DEV_NVME_SCSI)	+= scsi.o
 nvme-core-$(CONFIG_NVM)			+= lightnvm.o
 
 nvme-y					+= pci.o
+
+nvme-fabrics-y				+= fabrics.o

drivers/nvme/host/core.c

@@ -1178,9 +1178,26 @@ int nvme_init_identify(struct nvme_ctrl *ctrl)
 	}
 
 	nvme_set_queue_limits(ctrl, ctrl->admin_q);
+	ctrl->sgls = le32_to_cpu(id->sgls);
+
+	if (ctrl->ops->is_fabrics) {
+		ctrl->icdoff = le16_to_cpu(id->icdoff);
+		ctrl->ioccsz = le32_to_cpu(id->ioccsz);
+		ctrl->iorcsz = le32_to_cpu(id->iorcsz);
+		ctrl->maxcmd = le16_to_cpu(id->maxcmd);
+
+		/*
+		 * In fabrics we need to verify the cntlid matches the
+		 * admin connect
+		 */
+		if (ctrl->cntlid != le16_to_cpu(id->cntlid))
+			ret = -EINVAL;
+	} else {
+		ctrl->cntlid = le16_to_cpu(id->cntlid);
+	}
 
 	kfree(id);
-	return 0;
+	return ret;
 }
 EXPORT_SYMBOL_GPL(nvme_init_identify);
 

drivers/nvme/host/fabrics.h

+/*
+ * NVMe over Fabrics common host code.
+ * Copyright (c) 2015-2016 HGST, a Western Digital Company.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ */
+#ifndef _NVME_FABRICS_H
+#define _NVME_FABRICS_H 1
+
+#include <linux/in.h>
+#include <linux/inet.h>
+
+#define NVMF_MIN_QUEUE_SIZE	16
+#define NVMF_MAX_QUEUE_SIZE	1024
+#define NVMF_DEF_QUEUE_SIZE	128
+#define NVMF_DEF_RECONNECT_DELAY	10
+
+/*
+ * Define a host as seen by the target.  We allocate one at boot, but also
+ * allow the override it when creating controllers.  This is both to provide
+ * persistence of the Host NQN over multiple boots, and to allow using
+ * multiple ones, for example in a container scenario.  Because we must not
+ * use different Host NQNs with the same Host ID we generate a Host ID and
+ * use this structure to keep track of the relation between the two.
+ */
+struct nvmf_host {
+	struct kref		ref;
+	struct list_head	list;
+	char			nqn[NVMF_NQN_SIZE];
+	uuid_le			id;
+};
+
+/**
+ * enum nvmf_parsing_opts - used to define the sysfs parsing options used.
+ */
+enum {
+	NVMF_OPT_ERR		= 0,
+	NVMF_OPT_TRANSPORT	= 1 << 0,
+	NVMF_OPT_NQN		= 1 << 1,
+	NVMF_OPT_TRADDR		= 1 << 2,
+	NVMF_OPT_TRSVCID	= 1 << 3,
+	NVMF_OPT_QUEUE_SIZE	= 1 << 4,
+	NVMF_OPT_NR_IO_QUEUES	= 1 << 5,
+	NVMF_OPT_TL_RETRY_COUNT	= 1 << 6,
+	NVMF_OPT_HOSTNQN	= 1 << 8,
+	NVMF_OPT_RECONNECT_DELAY = 1 << 9,
+};
+
+/**
+ * struct nvmf_ctrl_options - Used to hold the options specified
+ *			      with the parsing opts enum.
+ * @mask:	Used by the fabrics library to parse through sysfs options
+ *		on adding a NVMe controller.
+ * @transport:	Holds the fabric transport "technology name" (for a lack of
+ *		better description) that will be used by an NVMe controller
+ *		being added.
+ * @subsysnqn:	Hold the fully qualified NQN subystem name (format defined
+ *		in the NVMe specification, "NVMe Qualified Names").
+ * @traddr:	network address that will be used by the host to communicate
+ *		to the added NVMe controller.
+ * @trsvcid:	network port used for host-controller communication.
+ * @queue_size: Number of IO queue elements.
+ * @nr_io_queues: Number of controller IO queues that will be established.
+ * @tl_retry_count: Number of transport layer retries for a fabric queue before
+ *		     kicking upper layer(s) error recovery.
+ * @reconnect_delay: Time between two consecutive reconnect attempts.
+ * @discovery_nqn: indicates if the subsysnqn is the well-known discovery NQN.
+ * @host:	Virtual NVMe host, contains the NQN and Host ID.
+ */
+struct nvmf_ctrl_options {
+	unsigned		mask;
+	char			*transport;
+	char			*subsysnqn;
+	char			*traddr;
+	char			*trsvcid;
+	size_t			queue_size;
+	unsigned int		nr_io_queues;
+	unsigned short		tl_retry_count;
+	unsigned int		reconnect_delay;
+	bool			discovery_nqn;
+	struct nvmf_host	*host;
+};
+
+/*
+ * struct nvmf_transport_ops - used to register a specific
+ *			       fabric implementation of NVMe fabrics.
+ * @entry:		Used by the fabrics library to add the new
+ *			registration entry to its linked-list internal tree.
+ * @name:		Name of the NVMe fabric driver implementation.
+ * @required_opts:	sysfs command-line options that must be specified
+ *			when adding a new NVMe controller.
+ * @allowed_opts:	sysfs command-line options that can be specified
+ *			when adding a new NVMe controller.
+ * @create_ctrl():	function pointer that points to a non-NVMe
+ *			implementation-specific fabric technology
+ *			that would go into starting up that fabric
+ *			for the purpose of conneciton to an NVMe controller
+ *			using that fabric technology.
+ *
+ * Notes:
+ *	1. At minimum, 'required_opts' and 'allowed_opts' should
+ *	   be set to the same enum parsing options defined earlier.
+ *	2. create_ctrl() must be defined (even if it does nothing)
+ */
+struct nvmf_transport_ops {
+	struct list_head	entry;
+	const char		*name;
+	int			required_opts;
+	int			allowed_opts;
+	struct nvme_ctrl	*(*create_ctrl)(struct device *dev,
+					struct nvmf_ctrl_options *opts);
+};
+
+int nvmf_reg_read32(struct nvme_ctrl *ctrl, u32 off, u32 *val);
+int nvmf_reg_read64(struct nvme_ctrl *ctrl, u32 off, u64 *val);
+int nvmf_reg_write32(struct nvme_ctrl *ctrl, u32 off, u32 val);
+int nvmf_connect_admin_queue(struct nvme_ctrl *ctrl);
+int nvmf_connect_io_queue(struct nvme_ctrl *ctrl, u16 qid);
+void nvmf_register_transport(struct nvmf_transport_ops *ops);
+void nvmf_unregister_transport(struct nvmf_transport_ops *ops);
+void nvmf_free_options(struct nvmf_ctrl_options *opts);
+const char *nvmf_get_subsysnqn(struct nvme_ctrl *ctrl);
+int nvmf_get_address(struct nvme_ctrl *ctrl, char *buf, int size);
+
+#endif /* _NVME_FABRICS_H */

drivers/nvme/host/nvme.h

@@ -80,6 +80,7 @@ struct nvme_ctrl {
 	spinlock_t lock;
 	const struct nvme_ctrl_ops *ops;
 	struct request_queue *admin_q;
+	struct request_queue *connect_q;
 	struct device *dev;
 	struct kref kref;
 	int instance;
@@ -107,10 +108,19 @@ struct nvme_ctrl {
 	u8 event_limit;
 	u8 vwc;
 	u32 vs;
+	u32 sgls;
 	bool subsystem;
 	unsigned long quirks;
 	struct work_struct scan_work;
 	struct work_struct async_event_work;
+
+	/* Fabrics only */
+	u16 sqsize;
+	u32 ioccsz;
+	u32 iorcsz;
+	u16 icdoff;
+	u16 maxcmd;
+	struct nvmf_ctrl_options *opts;
 };
 
 /*
@@ -146,6 +156,7 @@ struct nvme_ns {
 struct nvme_ctrl_ops {
 	const char *name;
 	struct module *module;
+	bool is_fabrics;
 	int (*reg_read32)(struct nvme_ctrl *ctrl, u32 off, u32 *val);
 	int (*reg_write32)(struct nvme_ctrl *ctrl, u32 off, u32 val);
 	int (*reg_read64)(struct nvme_ctrl *ctrl, u32 off, u64 *val);