nvmet-tcp: add NVMe over TCP target driver

This patch implements the TCP transport driver for the NVMe over Fabrics
target stack. This allows exporting NVMe over Fabrics functionality over
good old TCP/IP.

The driver implements the TP 8000 of how nvme over fabrics capsules and
data are encapsulated in nvme-tcp pdus and exchaged on top of a TCP byte
stream. nvme-tcp header and data digest are supported as well.

Signed-off-by: Sagi Grimberg <sagi@lightbitslabs.com>
Signed-off-by: Roy Shterman <roys@lightbitslabs.com>
Signed-off-by: Solganik Alexander <sashas@lightbitslabs.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>

3 files changed, 1749 insertions, 0 deletions

diff --git a/drivers/nvme/target/Kconfig b/drivers/nvme/target/Kconfig
index 3c7b61ddb0d1..d94f25cde019 100644
--- a/drivers/nvme/target/Kconfig
+++ b/drivers/nvme/target/Kconfig
@@ -60,3 +60,13 @@ config NVME_TARGET_FCLOOP
 	  to test NVMe-FC transport interfaces.
 
 	  If unsure, say N.
+
+config NVME_TARGET_TCP
+	tristate "NVMe over Fabrics TCP target support"
+	depends on INET
+	depends on NVME_TARGET
+	help
+	  This enables the NVMe TCP target support, which allows exporting NVMe
+	  devices over TCP.
+
+	  If unsure, say N.
diff --git a/drivers/nvme/target/Makefile b/drivers/nvme/target/Makefile
index 8118c93391c6..8c3ad0fb6860 100644
--- a/drivers/nvme/target/Makefile
+++ b/drivers/nvme/target/Makefile
@@ -5,6 +5,7 @@ obj-$(CONFIG_NVME_TARGET_LOOP)		+= nvme-loop.o
 obj-$(CONFIG_NVME_TARGET_RDMA)		+= nvmet-rdma.o
 obj-$(CONFIG_NVME_TARGET_FC)		+= nvmet-fc.o
 obj-$(CONFIG_NVME_TARGET_FCLOOP)	+= nvme-fcloop.o
+obj-$(CONFIG_NVME_TARGET_TCP)		+= nvmet-tcp.o
 
 nvmet-y		+= core.o configfs.o admin-cmd.o fabrics-cmd.o \
 			discovery.o io-cmd-file.o io-cmd-bdev.o
@@ -12,3 +13,4 @@ nvme-loop-y	+= loop.o
 nvmet-rdma-y	+= rdma.o
 nvmet-fc-y	+= fc.o
 nvme-fcloop-y	+= fcloop.o
+nvmet-tcp-y	+= tcp.o
diff --git a/drivers/nvme/target/tcp.c b/drivers/nvme/target/tcp.c
new file mode 100644
index 000000000000..d31bec260160
--- /dev/null
+++ b/drivers/nvme/target/tcp.c
@@ -0,0 +1,1737 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * NVMe over Fabrics TCP target.
+ * Copyright (c) 2018 Lightbits Labs. All rights reserved.
+ */
+#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
+#include <linux/module.h>
+#include <linux/init.h>
+#include <linux/slab.h>
+#include <linux/err.h>
+#include <linux/nvme-tcp.h>
+#include <net/sock.h>
+#include <net/tcp.h>
+#include <linux/inet.h>
+#include <linux/llist.h>
+#include <crypto/hash.h>
+
+#include "nvmet.h"
+
+#define NVMET_TCP_DEF_INLINE_DATA_SIZE	(4 * PAGE_SIZE)
+
+#define NVMET_TCP_RECV_BUDGET		8
+#define NVMET_TCP_SEND_BUDGET		8
+#define NVMET_TCP_IO_WORK_BUDGET	64
+
+enum nvmet_tcp_send_state {
+	NVMET_TCP_SEND_DATA_PDU,
+	NVMET_TCP_SEND_DATA,
+	NVMET_TCP_SEND_R2T,
+	NVMET_TCP_SEND_DDGST,
+	NVMET_TCP_SEND_RESPONSE
+};
+
+enum nvmet_tcp_recv_state {
+	NVMET_TCP_RECV_PDU,
+	NVMET_TCP_RECV_DATA,
+	NVMET_TCP_RECV_DDGST,
+	NVMET_TCP_RECV_ERR,
+};
+
+enum {
+	NVMET_TCP_F_INIT_FAILED = (1 << 0),
+};
+
+struct nvmet_tcp_cmd {
+	struct nvmet_tcp_queue		*queue;
+	struct nvmet_req		req;
+
+	struct nvme_tcp_cmd_pdu		*cmd_pdu;
+	struct nvme_tcp_rsp_pdu		*rsp_pdu;
+	struct nvme_tcp_data_pdu	*data_pdu;
+	struct nvme_tcp_r2t_pdu		*r2t_pdu;
+
+	u32				rbytes_done;
+	u32				wbytes_done;
+
+	u32				pdu_len;
+	u32				pdu_recv;
+	int				sg_idx;
+	int				nr_mapped;
+	struct msghdr			recv_msg;
+	struct kvec			*iov;
+	u32				flags;
+
+	struct list_head		entry;
+	struct llist_node		lentry;
+
+	/* send state */
+	u32				offset;
+	struct scatterlist		*cur_sg;
+	enum nvmet_tcp_send_state	state;
+
+	__le32				exp_ddgst;
+	__le32				recv_ddgst;
+};
+
+enum nvmet_tcp_queue_state {
+	NVMET_TCP_Q_CONNECTING,
+	NVMET_TCP_Q_LIVE,
+	NVMET_TCP_Q_DISCONNECTING,
+};
+
+struct nvmet_tcp_queue {
+	struct socket		*sock;
+	struct nvmet_tcp_port	*port;
+	struct work_struct	io_work;
+	int			cpu;
+	struct nvmet_cq		nvme_cq;
+	struct nvmet_sq		nvme_sq;
+
+	/* send state */
+	struct nvmet_tcp_cmd	*cmds;
+	unsigned int		nr_cmds;
+	struct list_head	free_list;
+	struct llist_head	resp_list;
+	struct list_head	resp_send_list;
+	int			send_list_len;
+	struct nvmet_tcp_cmd	*snd_cmd;
+
+	/* recv state */
+	int			offset;
+	int			left;
+	enum nvmet_tcp_recv_state rcv_state;
+	struct nvmet_tcp_cmd	*cmd;
+	union nvme_tcp_pdu	pdu;
+
+	/* digest state */
+	bool			hdr_digest;
+	bool			data_digest;
+	struct ahash_request	*snd_hash;
+	struct ahash_request	*rcv_hash;
+
+	spinlock_t		state_lock;
+	enum nvmet_tcp_queue_state state;
+
+	struct sockaddr_storage	sockaddr;
+	struct sockaddr_storage	sockaddr_peer;
+	struct work_struct	release_work;
+
+	int			idx;
+	struct list_head	queue_list;
+
+	struct nvmet_tcp_cmd	connect;
+
+	struct page_frag_cache	pf_cache;
+
+	void (*data_ready)(struct sock *);
+	void (*state_change)(struct sock *);
+	void (*write_space)(struct sock *);
+};
+
+struct nvmet_tcp_port {
+	struct socket		*sock;
+	struct work_struct	accept_work;
+	struct nvmet_port	*nport;
+	struct sockaddr_storage addr;
+	int			last_cpu;
+	void (*data_ready)(struct sock *);
+};
+
+static DEFINE_IDA(nvmet_tcp_queue_ida);
+static LIST_HEAD(nvmet_tcp_queue_list);
+static DEFINE_MUTEX(nvmet_tcp_queue_mutex);
+
+static struct workqueue_struct *nvmet_tcp_wq;
+static struct nvmet_fabrics_ops nvmet_tcp_ops;
+static void nvmet_tcp_free_cmd(struct nvmet_tcp_cmd *c);
+static void nvmet_tcp_finish_cmd(struct nvmet_tcp_cmd *cmd);
+
+static inline u16 nvmet_tcp_cmd_tag(struct nvmet_tcp_queue *queue,
+		struct nvmet_tcp_cmd *cmd)
+{
+	return cmd - queue->cmds;
+}
+
+static inline bool nvmet_tcp_has_data_in(struct nvmet_tcp_cmd *cmd)
+{
+	return nvme_is_write(cmd->req.cmd) &&
+		cmd->rbytes_done < cmd->req.transfer_len;
+}
+
+static inline bool nvmet_tcp_need_data_in(struct nvmet_tcp_cmd *cmd)
+{
+	return nvmet_tcp_has_data_in(cmd) && !cmd->req.rsp->status;
+}
+
+static inline bool nvmet_tcp_need_data_out(struct nvmet_tcp_cmd *cmd)
+{
+	return !nvme_is_write(cmd->req.cmd) &&
+		cmd->req.transfer_len > 0 &&
+		!cmd->req.rsp->status;
+}
+
+static inline bool nvmet_tcp_has_inline_data(struct nvmet_tcp_cmd *cmd)
+{
+	return nvme_is_write(cmd->req.cmd) && cmd->pdu_len &&
+		!cmd->rbytes_done;
+}
+
+static inline struct nvmet_tcp_cmd *
+nvmet_tcp_get_cmd(struct nvmet_tcp_queue *queue)
+{
+	struct nvmet_tcp_cmd *cmd;
+
+	cmd = list_first_entry_or_null(&queue->free_list,
+				struct nvmet_tcp_cmd, entry);
+	if (!cmd)
+		return NULL;
+	list_del_init(&cmd->entry);
+
+	cmd->rbytes_done = cmd->wbytes_done = 0;
+	cmd->pdu_len = 0;
+	cmd->pdu_recv = 0;
+	cmd->iov = NULL;
+	cmd->flags = 0;
+	return cmd;
+}
+
+static inline void nvmet_tcp_put_cmd(struct nvmet_tcp_cmd *cmd)
+{
+	if (unlikely(cmd == &cmd->queue->connect))
+		return;
+
+	list_add_tail(&cmd->entry, &cmd->queue->free_list);
+}
+
+static inline u8 nvmet_tcp_hdgst_len(struct nvmet_tcp_queue *queue)
+{
+	return queue->hdr_digest ? NVME_TCP_DIGEST_LENGTH : 0;
+}
+
+static inline u8 nvmet_tcp_ddgst_len(struct nvmet_tcp_queue *queue)
+{
+	return queue->data_digest ? NVME_TCP_DIGEST_LENGTH : 0;
+}
+
+static inline void nvmet_tcp_hdgst(struct ahash_request *hash,
+		void *pdu, size_t len)
+{
+	struct scatterlist sg;
+
+	sg_init_one(&sg, pdu, len);
+	ahash_request_set_crypt(hash, &sg, pdu + len, len);
+	crypto_ahash_digest(hash);
+}
+
+static int nvmet_tcp_verify_hdgst(struct nvmet_tcp_queue *queue,
+	void *pdu, size_t len)
+{
+	struct nvme_tcp_hdr *hdr = pdu;
+	__le32 recv_digest;
+	__le32 exp_digest;
+
+	if (unlikely(!(hdr->flags & NVME_TCP_F_HDGST))) {
+		pr_err("queue %d: header digest enabled but no header digest\n",
+			queue->idx);
+		return -EPROTO;
+	}
+
+	recv_digest = *(__le32 *)(pdu + hdr->hlen);
+	nvmet_tcp_hdgst(queue->rcv_hash, pdu, len);
+	exp_digest = *(__le32 *)(pdu + hdr->hlen);
+	if (recv_digest != exp_digest) {
+		pr_err("queue %d: header digest error: recv %#x expected %#x\n",
+			queue->idx, le32_to_cpu(recv_digest),
+			le32_to_cpu(exp_digest));
+		return -EPROTO;
+	}
+
+	return 0;
+}
+
+static int nvmet_tcp_check_ddgst(struct nvmet_tcp_queue *queue, void *pdu)
+{
+	struct nvme_tcp_hdr *hdr = pdu;
+	u8 digest_len = nvmet_tcp_hdgst_len(queue);
+	u32 len;
+
+	len = le32_to_cpu(hdr->plen) - hdr->hlen -
+		(hdr->flags & NVME_TCP_F_HDGST ? digest_len : 0);
+
+	if (unlikely(len && !(hdr->flags & NVME_TCP_F_DDGST))) {
+		pr_err("queue %d: data digest flag is cleared\n", queue->idx);
+		return -EPROTO;
+	}
+
+	return 0;
+}
+
+static void nvmet_tcp_unmap_pdu_iovec(struct nvmet_tcp_cmd *cmd)
+{
+	struct scatterlist *sg;
+	int i;
+
+	sg = &cmd->req.sg[cmd->sg_idx];
+
+	for (i = 0; i < cmd->nr_mapped; i++)
+		kunmap(sg_page(&sg[i]));
+}
+
+static void nvmet_tcp_map_pdu_iovec(struct nvmet_tcp_cmd *cmd)
+{
+	struct kvec *iov = cmd->iov;
+	struct scatterlist *sg;
+	u32 length, offset, sg_offset;
+
+	length = cmd->pdu_len;
+	cmd->nr_mapped = DIV_ROUND_UP(length, PAGE_SIZE);
+	offset = cmd->rbytes_done;
+	cmd->sg_idx = DIV_ROUND_UP(offset, PAGE_SIZE);
+	sg_offset = offset % PAGE_SIZE;
+	sg = &cmd->req.sg[cmd->sg_idx];
+
+	while (length) {
+		u32 iov_len = min_t(u32, length, sg->length - sg_offset);
+
+		iov->iov_base = kmap(sg_page(sg)) + sg->offset + sg_offset;
+		iov->iov_len = iov_len;
+
+		length -= iov_len;
+		sg = sg_next(sg);
+		iov++;
+	}
+
+	iov_iter_kvec(&cmd->recv_msg.msg_iter, READ, cmd->iov,
+		cmd->nr_mapped, cmd->pdu_len);
+}
+
+static void nvmet_tcp_fatal_error(struct nvmet_tcp_queue *queue)
+{
+	queue->rcv_state = NVMET_TCP_RECV_ERR;
+	if (queue->nvme_sq.ctrl)
+		nvmet_ctrl_fatal_error(queue->nvme_sq.ctrl);
+	else
+		kernel_sock_shutdown(queue->sock, SHUT_RDWR);
+}
+
+static int nvmet_tcp_map_data(struct nvmet_tcp_cmd *cmd)
+{
+	struct nvme_sgl_desc *sgl = &cmd->req.cmd->common.dptr.sgl;
+	u32 len = le32_to_cpu(sgl->length);
+
+	if (!cmd->req.data_len)
+		return 0;
+
+	if (sgl->type == ((NVME_SGL_FMT_DATA_DESC << 4) |
+			  NVME_SGL_FMT_OFFSET)) {
+		if (!nvme_is_write(cmd->req.cmd))
+			return NVME_SC_INVALID_FIELD | NVME_SC_DNR;
+
+		if (len > cmd->req.port->inline_data_size)
+			return NVME_SC_SGL_INVALID_OFFSET | NVME_SC_DNR;
+		cmd->pdu_len = len;
+	}
+	cmd->req.transfer_len += len;
+
+	cmd->req.sg = sgl_alloc(len, GFP_KERNEL, &cmd->req.sg_cnt);
+	if (!cmd->req.sg)
+		return NVME_SC_INTERNAL;
+	cmd->cur_sg = cmd->req.sg;
+
+	if (nvmet_tcp_has_data_in(cmd)) {
+		cmd->iov = kmalloc_array(cmd->req.sg_cnt,
+				sizeof(*cmd->iov), GFP_KERNEL);
+		if (!cmd->iov)
+			goto err;
+	}
+
+	return 0;
+err:
+	sgl_free(cmd->req.sg);
+	return NVME_SC_INTERNAL;
+}
+
+static void nvmet_tcp_ddgst(struct ahash_request *hash,
+		struct nvmet_tcp_cmd *cmd)
+{
+	ahash_request_set_crypt(hash, cmd->req.sg,
+		(void *)&cmd->exp_ddgst, cmd->req.transfer_len);
+	crypto_ahash_digest(hash);
+}
+
+static void nvmet_setup_c2h_data_pdu(struct nvmet_tcp_cmd *cmd)
+{
+	struct nvme_tcp_data_pdu *pdu = cmd->data_pdu;
+	struct nvmet_tcp_queue *queue = cmd->queue;
+	u8 hdgst = nvmet_tcp_hdgst_len(cmd->queue);
+	u8 ddgst = nvmet_tcp_ddgst_len(cmd->queue);
+
+	cmd->offset = 0;
+	cmd->state = NVMET_TCP_SEND_DATA_PDU;
+
+	pdu->hdr.type = nvme_tcp_c2h_data;
+	pdu->hdr.flags = NVME_TCP_F_DATA_LAST;
+	pdu->hdr.hlen = sizeof(*pdu);
+	pdu->hdr.pdo = pdu->hdr.hlen + hdgst;
+	pdu->hdr.plen =
+		cpu_to_le32(pdu->hdr.hlen + hdgst +
+				cmd->req.transfer_len + ddgst);
+	pdu->command_id = cmd->req.rsp->command_id;
+	pdu->data_length = cpu_to_le32(cmd->req.transfer_len);
+	pdu->data_offset = cpu_to_le32(cmd->wbytes_done);
+
+	if (queue->data_digest) {
+		pdu->hdr.flags |= NVME_TCP_F_DDGST;
+		nvmet_tcp_ddgst(queue->snd_hash, cmd);
+	}
+
+	if (cmd->queue->hdr_digest) {
+		pdu->hdr.flags |= NVME_TCP_F_HDGST;
+		nvmet_tcp_hdgst(queue->snd_hash, pdu, sizeof(*pdu));
+	}
+}
+
+static void nvmet_setup_r2t_pdu(struct nvmet_tcp_cmd *cmd)
+{
+	struct nvme_tcp_r2t_pdu *pdu = cmd->r2t_pdu;
+	struct nvmet_tcp_queue *queue = cmd->queue;
+	u8 hdgst = nvmet_tcp_hdgst_len(cmd->queue);
+
+	cmd->offset = 0;
+	cmd->state = NVMET_TCP_SEND_R2T;
+
+	pdu->hdr.type = nvme_tcp_r2t;
+	pdu->hdr.flags = 0;
+	pdu->hdr.hlen = sizeof(*pdu);
+	pdu->hdr.pdo = 0;
+	pdu->hdr.plen = cpu_to_le32(pdu->hdr.hlen + hdgst);
+
+	pdu->command_id = cmd->req.cmd->common.command_id;
+	pdu->ttag = nvmet_tcp_cmd_tag(cmd->queue, cmd);
+	pdu->r2t_length = cpu_to_le32(cmd->req.transfer_len - cmd->rbytes_done);
+	pdu->r2t_offset = cpu_to_le32(cmd->rbytes_done);
+	if (cmd->queue->hdr_digest) {
+		pdu->hdr.flags |= NVME_TCP_F_HDGST;
+		nvmet_tcp_hdgst(queue->snd_hash, pdu, sizeof(*pdu));
+	}
+}
+
+static void nvmet_setup_response_pdu(struct nvmet_tcp_cmd *cmd)
+{
+	struct nvme_tcp_rsp_pdu *pdu = cmd->rsp_pdu;
+	struct nvmet_tcp_queue *queue = cmd->queue;
+	u8 hdgst = nvmet_tcp_hdgst_len(cmd->queue);
+
+	cmd->offset = 0;
+	cmd->state = NVMET_TCP_SEND_RESPONSE;
+
+	pdu->hdr.type = nvme_tcp_rsp;
+	pdu->hdr.flags = 0;
+	pdu->hdr.hlen = sizeof(*pdu);
+	pdu->hdr.pdo = 0;
+	pdu->hdr.plen = cpu_to_le32(pdu->hdr.hlen + hdgst);
+	if (cmd->queue->hdr_digest) {
+		pdu->hdr.flags |= NVME_TCP_F_HDGST;
+		nvmet_tcp_hdgst(queue->snd_hash, pdu, sizeof(*pdu));
+	}
+}
+
+static void nvmet_tcp_process_resp_list(struct nvmet_tcp_queue *queue)
+{
+	struct llist_node *node;
+
+	node = llist_del_all(&queue->resp_list);
+	if (!node)
+		return;
+
+	while (node) {
+		struct nvmet_tcp_cmd *cmd = llist_entry(node,
+					struct nvmet_tcp_cmd, lentry);
+
+		list_add(&cmd->entry, &queue->resp_send_list);
+		node = node->next;
+		queue->send_list_len++;
+	}
+}
+
+static struct nvmet_tcp_cmd *nvmet_tcp_fetch_cmd(struct nvmet_tcp_queue *queue)
+{
+	queue->snd_cmd = list_first_entry_or_null(&queue->resp_send_list,
+				struct nvmet_tcp_cmd, entry);
+	if (!queue->snd_cmd) {
+		nvmet_tcp_process_resp_list(queue);
+		queue->snd_cmd =
+			list_first_entry_or_null(&queue->resp_send_list,
+					struct nvmet_tcp_cmd, entry);
+		if (unlikely(!queue->snd_cmd))
+			return NULL;
+	}
+
+	list_del_init(&queue->snd_cmd->entry);
+	queue->send_list_len--;
+
+	if (nvmet_tcp_need_data_out(queue->snd_cmd))
+		nvmet_setup_c2h_data_pdu(queue->snd_cmd);
+	else if (nvmet_tcp_need_data_in(queue->snd_cmd))
+		nvmet_setup_r2t_pdu(queue->snd_cmd);
+	else
+		nvmet_setup_response_pdu(queue->snd_cmd);
+
+	return queue->snd_cmd;
+}
+
+static void nvmet_tcp_queue_response(struct nvmet_req *req)
+{
+	struct nvmet_tcp_cmd *cmd =
+		container_of(req, struct nvmet_tcp_cmd, req);
+	struct nvmet_tcp_queue	*queue = cmd->queue;
+
+	llist_add(&cmd->lentry, &queue->resp_list);
+	queue_work_on(cmd->queue->cpu, nvmet_tcp_wq, &cmd->queue->io_work);
+}
+
+static int nvmet_try_send_data_pdu(struct nvmet_tcp_cmd *cmd)
+{
+	u8 hdgst = nvmet_tcp_hdgst_len(cmd->queue);
+	int left = sizeof(*cmd->data_pdu) - cmd->offset + hdgst;
+	int ret;
+
+	ret = kernel_sendpage(cmd->queue->sock, virt_to_page(cmd->data_pdu),
+			offset_in_page(cmd->data_pdu) + cmd->offset,
+			left, MSG_DONTWAIT | MSG_MORE);
+	if (ret <= 0)
+		return ret;
+
+	cmd->offset += ret;
+	left -= ret;
+
+	if (left)
+		return -EAGAIN;
+
+	cmd->state = NVMET_TCP_SEND_DATA;
+	cmd->offset  = 0;
+	return 1;
+}
+
+static int nvmet_try_send_data(struct nvmet_tcp_cmd *cmd)
+{
+	struct nvmet_tcp_queue *queue = cmd->queue;
+	int ret;
+
+	while (cmd->cur_sg) {
+		struct page *page = sg_page(cmd->cur_sg);
+		u32 left = cmd->cur_sg->length - cmd->offset;
+
+		ret = kernel_sendpage(cmd->queue->sock, page, cmd->offset,
+					left, MSG_DONTWAIT | MSG_MORE);
+		if (ret <= 0)
+			return ret;
+
+		cmd->offset += ret;
+		cmd->wbytes_done += ret;
+
+		/* Done with sg?*/
+		if (cmd->offset == cmd->cur_sg->length) {
+			cmd->cur_sg = sg_next(cmd->cur_sg);
+			cmd->offset = 0;
+		}
+	}
+
+	if (queue->data_digest) {
+		cmd->state = NVMET_TCP_SEND_DDGST;
+		cmd->offset = 0;
+	} else {
+		nvmet_setup_response_pdu(cmd);
+	}
+	return 1;
+
+}
+
+static int nvmet_try_send_response(struct nvmet_tcp_cmd *cmd,
+		bool last_in_batch)
+{
+	u8 hdgst = nvmet_tcp_hdgst_len(cmd->queue);
+	int left = sizeof(*cmd->rsp_pdu) - cmd->offset + hdgst;
+	int flags = MSG_DONTWAIT;
+	int ret;
+
+	if (!last_in_batch && cmd->queue->send_list_len)
+		flags |= MSG_MORE;
+	else
+		flags |= MSG_EOR;
+
+	ret = kernel_sendpage(cmd->queue->sock, virt_to_page(cmd->rsp_pdu),
+		offset_in_page(cmd->rsp_pdu) + cmd->offset, left, flags);
+	if (ret <= 0)
+		return ret;
+	cmd->offset += ret;
+	left -= ret;
+
+	if (left)
+		return -EAGAIN;
+
+	kfree(cmd->iov);
+	sgl_free(cmd->req.sg);
+	cmd->queue->snd_cmd = NULL;
+	nvmet_tcp_put_cmd(cmd);
+	return 1;
+}
+
+static int nvmet_try_send_r2t(struct nvmet_tcp_cmd *cmd, bool last_in_batch)
+{
+	u8 hdgst = nvmet_tcp_hdgst_len(cmd->queue);
+	int left = sizeof(*cmd->r2t_pdu) - cmd->offset + hdgst;
+	int flags = MSG_DONTWAIT;
+	int ret;
+
+	if (!last_in_batch && cmd->queue->send_list_len)
+		flags |= MSG_MORE;
+	else
+		flags |= MSG_EOR;
+
+	ret = kernel_sendpage(cmd->queue->sock, virt_to_page(cmd->r2t_pdu),
+		offset_in_page(cmd->r2t_pdu) + cmd->offset, left, flags);
+	if (ret <= 0)
+		return ret;
+	cmd->offset += ret;
+	left -= ret;
+
+	if (left)
+		return -EAGAIN;
+
+	cmd->queue->snd_cmd = NULL;
+	return 1;
+}
+
+static int nvmet_try_send_ddgst(struct nvmet_tcp_cmd *cmd)
+{
+	struct nvmet_tcp_queue *queue = cmd->queue;
+	struct msghdr msg = { .msg_flags = MSG_DONTWAIT };
+	struct kvec iov = {
+		.iov_base = &cmd->exp_ddgst + cmd->offset,
+		.iov_len = NVME_TCP_DIGEST_LENGTH - cmd->offset
+	};
+	int ret;
+
+	ret = kernel_sendmsg(queue->sock, &msg, &iov, 1, iov.iov_len);
+	if (unlikely(ret <= 0))
+		return ret;
+
+	cmd->offset += ret;
+	nvmet_setup_response_pdu(cmd);
+	return 1;
+}
+
+static int nvmet_tcp_try_send_one(struct nvmet_tcp_queue *queue,
+		bool last_in_batch)
+{
+	struct nvmet_tcp_cmd *cmd = queue->snd_cmd;
+	int ret = 0;
+
+	if (!cmd || queue->state == NVMET_TCP_Q_DISCONNECTING) {
+		cmd = nvmet_tcp_fetch_cmd(queue);
+		if (unlikely(!cmd))
+			return 0;
+	}
+
+	if (cmd->state == NVMET_TCP_SEND_DATA_PDU) {
+		ret = nvmet_try_send_data_pdu(cmd);
+		if (ret <= 0)
+			goto done_send;
+	}
+
+	if (cmd->state == NVMET_TCP_SEND_DATA) {
+		ret = nvmet_try_send_data(cmd);
+		if (ret <= 0)
+			goto done_send;
+	}
+
+	if (cmd->state == NVMET_TCP_SEND_DDGST) {
+		ret = nvmet_try_send_ddgst(cmd);
+		if (ret <= 0)
+			goto done_send;
+	}
+
+	if (cmd->state == NVMET_TCP_SEND_R2T) {
+		ret = nvmet_try_send_r2t(cmd, last_in_batch);
+		if (ret <= 0)
+			goto done_send;
+	}
+
+	if (cmd->state == NVMET_TCP_SEND_RESPONSE)
+		ret = nvmet_try_send_response(cmd, last_in_batch);
+
+done_send:
+	if (ret < 0) {
+		if (ret == -EAGAIN)
+			return 0;
+		return ret;
+	}
+
+	return 1;
+}
+
+static int nvmet_tcp_try_send(struct nvmet_tcp_queue *queue,
+		int budget, int *sends)
+{
+	int i, ret = 0;
+
+	for (i = 0; i < budget; i++) {
+		ret = nvmet_tcp_try_send_one(queue, i == budget - 1);
+		if (ret <= 0)
+			break;
+		(*sends)++;
+	}
+
+	return ret;
+}
+
+static void nvmet_prepare_receive_pdu(struct nvmet_tcp_queue *queue)
+{
+	queue->offset = 0;
+	queue->left = sizeof(struct nvme_tcp_hdr);
+	queue->cmd = NULL;
+	queue->rcv_state = NVMET_TCP_RECV_PDU;
+}
+
+static void nvmet_tcp_free_crypto(struct nvmet_tcp_queue *queue)
+{
+	struct crypto_ahash *tfm = crypto_ahash_reqtfm(queue->rcv_hash);
+
+	ahash_request_free(queue->rcv_hash);
+	ahash_request_free(queue->snd_hash);
+	crypto_free_ahash(tfm);
+}
+
+static int nvmet_tcp_alloc_crypto(struct nvmet_tcp_queue *queue)
+{
+	struct crypto_ahash *tfm;
+
+	tfm = crypto_alloc_ahash("crc32c", 0, CRYPTO_ALG_ASYNC);
+	if (IS_ERR(tfm))
+		return PTR_ERR(tfm);
+
+	queue->snd_hash = ahash_request_alloc(tfm, GFP_KERNEL);
+	if (!queue->snd_hash)
+		goto free_tfm;
+	ahash_request_set_callback(queue->snd_hash, 0, NULL, NULL);
+
+	queue->rcv_hash = ahash_request_alloc(tfm, GFP_KERNEL);
+	if (!queue->rcv_hash)
+		goto free_snd_hash;
+	ahash_request_set_callback(queue->rcv_hash, 0, NULL, NULL);
+
+	return 0;
+free_snd_hash:
+	ahash_request_free(queue->snd_hash);
+free_tfm:
+	crypto_free_ahash(tfm);
+	return -ENOMEM;
+}
+
+
+static int nvmet_tcp_handle_icreq(struct nvmet_tcp_queue *queue)
+{
+	struct nvme_tcp_icreq_pdu *icreq = &queue->pdu.icreq;
+	struct nvme_tcp_icresp_pdu *icresp = &queue->pdu.icresp;
+	struct msghdr msg = {};
+	struct kvec iov;
+	int ret;
+
+	if (le32_to_cpu(icreq->hdr.plen) != sizeof(struct nvme_tcp_icreq_pdu)) {
+		pr_err("bad nvme-tcp pdu length (%d)\n",
+			le32_to_cpu(icreq->hdr.plen));
+		nvmet_tcp_fatal_error(queue);
+	}
+
+	if (icreq->pfv != NVME_TCP_PFV_1_0) {
+		pr_err("queue %d: bad pfv %d\n", queue->idx, icreq->pfv);
+		return -EPROTO;
+	}
+
+	if (icreq->hpda != 0) {
+		pr_err("queue %d: unsupported hpda %d\n", queue->idx,
+			icreq->hpda);
+		return -EPROTO;
+	}
+
+	if (icreq->maxr2t != 0) {
+		pr_err("queue %d: unsupported maxr2t %d\n", queue->idx,
+			le16_to_cpu(icreq->maxr2t) + 1);
+		return -EPROTO;
+	}
+
+	queue->hdr_digest = !!(icreq->digest & NVME_TCP_HDR_DIGEST_ENABLE);
+	queue->data_digest = !!(icreq->digest & NVME_TCP_DATA_DIGEST_ENABLE);
+	if (queue->hdr_digest || queue->data_digest) {
+		ret = nvmet_tcp_alloc_crypto(queue);
+		if (ret)
+			return ret;
+	}
+
+	memset(icresp, 0, sizeof(*icresp));
+	icresp->hdr.type = nvme_tcp_icresp;
+	icresp->hdr.hlen = sizeof(*icresp);
+	icresp->hdr.pdo = 0;
+	icresp->hdr.plen = cpu_to_le32(icresp->hdr.hlen);
+	icresp->pfv = cpu_to_le16(NVME_TCP_PFV_1_0);
+	icresp->maxdata = 0xffff; /* FIXME: support r2t */
+	icresp->cpda = 0;
+	if (queue->hdr_digest)
+		icresp->digest |= NVME_TCP_HDR_DIGEST_ENABLE;
+	if (queue->data_digest)
+		icresp->digest |= NVME_TCP_DATA_DIGEST_ENABLE;
+
+	iov.iov_base = icresp;
+	iov.iov_len = sizeof(*icresp);
+	ret = kernel_sendmsg(queue->sock, &msg, &iov, 1, iov.iov_len);
+	if (ret < 0)
+		goto free_crypto;
+
+	queue->state = NVMET_TCP_Q_LIVE;
+	nvmet_prepare_receive_pdu(queue);
+	return 0;
+free_crypto:
+	if (queue->hdr_digest || queue->data_digest)
+		nvmet_tcp_free_crypto(queue);
+	return ret;
+}
+
+static void nvmet_tcp_handle_req_failure(struct nvmet_tcp_queue *queue,
+		struct nvmet_tcp_cmd *cmd, struct nvmet_req *req)
+{
+	int ret;
+
+	/* recover the expected data transfer length */
+	req->data_len = le32_to_cpu(req->cmd->common.dptr.sgl.length);
+
+	if (!nvme_is_write(cmd->req.cmd) ||
+	    req->data_len > cmd->req.port->inline_data_size) {
+		nvmet_prepare_receive_pdu(queue);
+		return;
+	}
+
+	ret = nvmet_tcp_map_data(cmd);
+	if (unlikely(ret)) {
+		pr_err("queue %d: failed to map data\n", queue->idx);
+		nvmet_tcp_fatal_error(queue);
+		return;
+	}
+
+	queue->rcv_state = NVMET_TCP_RECV_DATA;
+	nvmet_tcp_map_pdu_iovec(cmd);
+	cmd->flags |= NVMET_TCP_F_INIT_FAILED;
+}
+
+static int nvmet_tcp_handle_h2c_data_pdu(struct nvmet_tcp_queue *queue)
+{
+	struct nvme_tcp_data_pdu *data = &queue->pdu.data;
+	struct nvmet_tcp_cmd *cmd;
+
+	cmd = &queue->cmds[data->ttag];
+
+	if (le32_to_cpu(data->data_offset) != cmd->rbytes_done) {
+		pr_err("ttag %u unexpected data offset %u (expected %u)\n",
+			data->ttag, le32_to_cpu(data->data_offset),
+			cmd->rbytes_done);
+		/* FIXME: use path and transport errors */
+		nvmet_req_complete(&cmd->req,
+			NVME_SC_INVALID_FIELD | NVME_SC_DNR);
+		return -EPROTO;
+	}
+
+	cmd->pdu_len = le32_to_cpu(data->data_length);
+	cmd->pdu_recv = 0;
+	nvmet_tcp_map_pdu_iovec(cmd);
+	queue->cmd = cmd;
+	queue->rcv_state = NVMET_TCP_RECV_DATA;
+
+	return 0;
+}
+
+static int nvmet_tcp_done_recv_pdu(struct nvmet_tcp_queue *queue)
+{
+	struct nvme_tcp_hdr *hdr = &queue->pdu.cmd.hdr;
+	struct nvme_command *nvme_cmd = &queue->pdu.cmd.cmd;
+	struct nvmet_req *req;
+	int ret;
+
+	if (unlikely(queue->state == NVMET_TCP_Q_CONNECTING)) {
+		if (hdr->type != nvme_tcp_icreq) {
+			pr_err("unexpected pdu type (%d) before icreq\n",
+				hdr->type);
+			nvmet_tcp_fatal_error(queue);
+			return -EPROTO;
+		}
+		return nvmet_tcp_handle_icreq(queue);
+	}
+
+	if (hdr->type == nvme_tcp_h2c_data) {
+		ret = nvmet_tcp_handle_h2c_data_pdu(queue);
+		if (unlikely(ret))
+			return ret;
+		return 0;
+	}
+
+	queue->cmd = nvmet_tcp_get_cmd(queue);
+	if (unlikely(!queue->cmd)) {
+		/* This should never happen */
+		pr_err("queue %d: out of commands (%d) send_list_len: %d, opcode: %d",
+			queue->idx, queue->nr_cmds, queue->send_list_len,
+			nvme_cmd->common.opcode);
+		nvmet_tcp_fatal_error(queue);
+		return -ENOMEM;
+	}
+
+	req = &queue->cmd->req;
+	memcpy(req->cmd, nvme_cmd, sizeof(*nvme_cmd));
+
+	if (unlikely(!nvmet_req_init(req, &queue->nvme_cq,
+			&queue->nvme_sq, &nvmet_tcp_ops))) {
+		pr_err("failed cmd %p id %d opcode %d, data_len: %d\n",
+			req->cmd, req->cmd->common.command_id,
+			req->cmd->common.opcode,
+			le32_to_cpu(req->cmd->common.dptr.sgl.length));
+
+		nvmet_tcp_handle_req_failure(queue, queue->cmd, req);
+		return -EAGAIN;
+	}
+
+	ret = nvmet_tcp_map_data(queue->cmd);
+	if (unlikely(ret)) {
+		pr_err("queue %d: failed to map data\n", queue->idx);
+		if (nvmet_tcp_has_inline_data(queue->cmd))
+			nvmet_tcp_fatal_error(queue);
+		else
+			nvmet_req_complete(req, ret);
+		ret = -EAGAIN;
+		goto out;
+	}
+
+	if (nvmet_tcp_need_data_in(queue->cmd)) {
+		if (nvmet_tcp_has_inline_data(queue->cmd)) {
+			queue->rcv_state = NVMET_TCP_RECV_DATA;
+			nvmet_tcp_map_pdu_iovec(queue->cmd);
+			return 0;
+		}
+		/* send back R2T */
+		nvmet_tcp_queue_response(&queue->cmd->req);
+		goto out;
+	}
+
+	nvmet_req_execute(&queue->cmd->req);
+out:
+	nvmet_prepare_receive_pdu(queue);
+	return ret;
+}
+
+static const u8 nvme_tcp_pdu_sizes[] = {
+	[nvme_tcp_icreq]	= sizeof(struct nvme_tcp_icreq_pdu),
+	[nvme_tcp_cmd]		= sizeof(struct nvme_tcp_cmd_pdu),
+	[nvme_tcp_h2c_data]	= sizeof(struct nvme_tcp_data_pdu),
+};
+
+static inline u8 nvmet_tcp_pdu_size(u8 type)
+{
+	size_t idx = type;
+
+	return (idx < ARRAY_SIZE(nvme_tcp_pdu_sizes) &&
+		nvme_tcp_pdu_sizes[idx]) ?
+			nvme_tcp_pdu_sizes[idx] : 0;
+}
+
+static inline bool nvmet_tcp_pdu_valid(u8 type)
+{
+	switch (type) {
+	case nvme_tcp_icreq:
+	case nvme_tcp_cmd:
+	case nvme_tcp_h2c_data:
+		/* fallthru */
+		return true;
+	}
+
+	return false;
+}
+
+static int nvmet_tcp_try_recv_pdu(struct nvmet_tcp_queue *queue)
+{
+	struct nvme_tcp_hdr *hdr = &queue->pdu.cmd.hdr;
+	int len;
+	struct kvec iov;
+	struct msghdr msg = { .msg_flags = MSG_DONTWAIT };
+
+recv:
+	iov.iov_base = (void *)&queue->pdu + queue->offset;
+	iov.iov_len = queue->left;
+	len = kernel_recvmsg(queue->sock, &msg, &iov, 1,
+			iov.iov_len, msg.msg_flags);
+	if (unlikely(len < 0))
+		return len;
+
+	queue->offset += len;
+	queue->left -= len;
+	if (queue->left)
+		return -EAGAIN;
+
+	if (queue->offset == sizeof(struct nvme_tcp_hdr)) {
+		u8 hdgst = nvmet_tcp_hdgst_len(queue);
+
+		if (unlikely(!nvmet_tcp_pdu_valid(hdr->type))) {
+			pr_err("unexpected pdu type %d\n", hdr->type);
+			nvmet_tcp_fatal_error(queue);
+			return -EIO;
+		}
+
+		if (unlikely(hdr->hlen != nvmet_tcp_pdu_size(hdr->type))) {
+			pr_err("pdu %d bad hlen %d\n", hdr->type, hdr->hlen);
+			return -EIO;
+		}
+
+		queue->left = hdr->hlen - queue->offset + hdgst;
+		goto recv;
+	}
+
+	if (queue->hdr_digest &&
+	    nvmet_tcp_verify_hdgst(queue, &queue->pdu, queue->offset)) {
+		nvmet_tcp_fatal_error(queue); /* fatal */
+		return -EPROTO;
+	}
+
+	if (queue->data_digest &&
+	    nvmet_tcp_check_ddgst(queue, &queue->pdu)) {
+		nvmet_tcp_fatal_error(queue); /* fatal */
+		return -EPROTO;
+	}
+
+	return nvmet_tcp_done_recv_pdu(queue);
+}
+
+static void nvmet_tcp_prep_recv_ddgst(struct nvmet_tcp_cmd *cmd)
+{
+	struct nvmet_tcp_queue *queue = cmd->queue;
+
+	nvmet_tcp_ddgst(queue->rcv_hash, cmd);
+	queue->offset = 0;
+	queue->left = NVME_TCP_DIGEST_LENGTH;
+	queue->rcv_state = NVMET_TCP_RECV_DDGST;
+}
+
+static int nvmet_tcp_try_recv_data(struct nvmet_tcp_queue *queue)
+{
+	struct nvmet_tcp_cmd  *cmd = queue->cmd;
+	int ret;
+
+	while (msg_data_left(&cmd->recv_msg)) {
+		ret = sock_recvmsg(cmd->queue->sock, &cmd->recv_msg,
+			cmd->recv_msg.msg_flags);
+		if (ret <= 0)
+			return ret;
+
+		cmd->pdu_recv += ret;
+		cmd->rbytes_done += ret;
+	}
+
+	nvmet_tcp_unmap_pdu_iovec(cmd);
+
+	if (!(cmd->flags & NVMET_TCP_F_INIT_FAILED) &&
+	    cmd->rbytes_done == cmd->req.transfer_len) {
+		if (queue->data_digest) {
+			nvmet_tcp_prep_recv_ddgst(cmd);
+			return 0;
+		}
+		nvmet_req_execute(&cmd->req);
+	}
+
+	nvmet_prepare_receive_pdu(queue);
+	return 0;
+}
+
+static int nvmet_tcp_try_recv_ddgst(struct nvmet_tcp_queue *queue)
+{
+	struct nvmet_tcp_cmd *cmd = queue->cmd;
+	int ret;
+	struct msghdr msg = { .msg_flags = MSG_DONTWAIT };
+	struct kvec iov = {
+		.iov_base = (void *)&cmd->recv_ddgst + queue->offset,
+		.iov_len = queue->left
+	};
+
+	ret = kernel_recvmsg(queue->sock, &msg, &iov, 1,
+			iov.iov_len, msg.msg_flags);
+	if (unlikely(ret < 0))
+		return ret;
+
+	queue->offset += ret;
+	queue->left -= ret;
+	if (queue->left)
+		return -EAGAIN;
+
+	if (queue->data_digest && cmd->exp_ddgst != cmd->recv_ddgst) {
+		pr_err("queue %d: cmd %d pdu (%d) data digest error: recv %#x expected %#x\n",
+			queue->idx, cmd->req.cmd->common.command_id,
+			queue->pdu.cmd.hdr.type, le32_to_cpu(cmd->recv_ddgst),
+			le32_to_cpu(cmd->exp_ddgst));
+		nvmet_tcp_finish_cmd(cmd);
+		nvmet_tcp_fatal_error(queue);
+		ret = -EPROTO;
+		goto out;
+	}
+
+	if (!(cmd->flags & NVMET_TCP_F_INIT_FAILED) &&
+	    cmd->rbytes_done == cmd->req.transfer_len)
+		nvmet_req_execute(&cmd->req);
+	ret = 0;
+out:
+	nvmet_prepare_receive_pdu(queue);
+	return ret;
+}
+
+static int nvmet_tcp_try_recv_one(struct nvmet_tcp_queue *queue)
+{
+	int result;
+
+	if (unlikely(queue->rcv_state == NVMET_TCP_RECV_ERR))
+		return 0;
+
+	if (queue->rcv_state == NVMET_TCP_RECV_PDU) {
+		result = nvmet_tcp_try_recv_pdu(queue);
+		if (result != 0)
+			goto done_recv;
+	}
+
+	if (queue->rcv_state == NVMET_TCP_RECV_DATA) {
+		result = nvmet_tcp_try_recv_data(queue);
+		if (result != 0)
+			goto done_recv;
+	}
+
+	if (queue->rcv_state == NVMET_TCP_RECV_DDGST) {
+		result = nvmet_tcp_try_recv_ddgst(queue);
+		if (result != 0)
+			goto done_recv;
+	}
+
+done_recv:
+	if (result < 0) {
+		if (result == -EAGAIN)
+			return 0;
+		return result;
+	}
+	return 1;
+}
+
+static int nvmet_tcp_try_recv(struct nvmet_tcp_queue *queue,
+		int budget, int *recvs)
+{
+	int i, ret = 0;
+
+	for (i = 0; i < budget; i++) {
+		ret = nvmet_tcp_try_recv_one(queue);
+		if (ret <= 0)
+			break;
+		(*recvs)++;
+	}
+
+	return ret;
+}
+
+static void nvmet_tcp_schedule_release_queue(struct nvmet_tcp_queue *queue)
+{
+	spin_lock(&queue->state_lock);
+	if (queue->state != NVMET_TCP_Q_DISCONNECTING) {
+		queue->state = NVMET_TCP_Q_DISCONNECTING;
+		schedule_work(&queue->release_work);
+	}
+	spin_unlock(&queue->state_lock);
+}
+
+static void nvmet_tcp_io_work(struct work_struct *w)
+{
+	struct nvmet_tcp_queue *queue =
+		container_of(w, struct nvmet_tcp_queue, io_work);
+	bool pending;
+	int ret, ops = 0;
+
+	do {
+		pending = false;
+
+		ret = nvmet_tcp_try_recv(queue, NVMET_TCP_RECV_BUDGET, &ops);
+		if (ret > 0) {
+			pending = true;
+		} else if (ret < 0) {
+			if (ret == -EPIPE || ret == -ECONNRESET)
+				kernel_sock_shutdown(queue->sock, SHUT_RDWR);
+			else
+				nvmet_tcp_fatal_error(queue);
+			return;
+		}
+
+		ret = nvmet_tcp_try_send(queue, NVMET_TCP_SEND_BUDGET, &ops);
+		if (ret > 0) {
+			/* transmitted message/data */
+			pending = true;
+		} else if (ret < 0) {
+			if (ret == -EPIPE || ret == -ECONNRESET)
+				kernel_sock_shutdown(queue->sock, SHUT_RDWR);
+			else
+				nvmet_tcp_fatal_error(queue);
+			return;
+		}
+
+	} while (pending && ops < NVMET_TCP_IO_WORK_BUDGET);
+
+	/*
+	 * We exahusted our budget, requeue our selves
+	 */
+	if (pending)
+		queue_work_on(queue->cpu, nvmet_tcp_wq, &queue->io_work);
+}
+
+static int nvmet_tcp_alloc_cmd(struct nvmet_tcp_queue *queue,
+		struct nvmet_tcp_cmd *c)
+{
+	u8 hdgst = nvmet_tcp_hdgst_len(queue);
+
+	c->queue = queue;
+	c->req.port = queue->port->nport;
+
+	c->cmd_pdu = page_frag_alloc(&queue->pf_cache,
+			sizeof(*c->cmd_pdu) + hdgst, GFP_KERNEL | __GFP_ZERO);
+	if (!c->cmd_pdu)
+		return -ENOMEM;
+	c->req.cmd = &c->cmd_pdu->cmd;
+
+	c->rsp_pdu = page_frag_alloc(&queue->pf_cache,
+			sizeof(*c->rsp_pdu) + hdgst, GFP_KERNEL | __GFP_ZERO);
+	if (!c->rsp_pdu)
+		goto out_free_cmd;
+	c->req.rsp = &c->rsp_pdu->cqe;
+
+	c->data_pdu = page_frag_alloc(&queue->pf_cache,
+			sizeof(*c->data_pdu) + hdgst, GFP_KERNEL | __GFP_ZERO);
+	if (!c->data_pdu)
+		goto out_free_rsp;
+
+	c->r2t_pdu = page_frag_alloc(&queue->pf_cache,
+			sizeof(*c->r2t_pdu) + hdgst, GFP_KERNEL | __GFP_ZERO);
+	if (!c->r2t_pdu)
+		goto out_free_data;
+
+	c->recv_msg.msg_flags = MSG_DONTWAIT | MSG_NOSIGNAL;
+
+	list_add_tail(&c->entry, &queue->free_list);
+
+	return 0;
+out_free_data:
+	page_frag_free(c->data_pdu);
+out_free_rsp:
+	page_frag_free(c->rsp_pdu);
+out_free_cmd:
+	page_frag_free(c->cmd_pdu);
+	return -ENOMEM;
+}
+
+static void nvmet_tcp_free_cmd(struct nvmet_tcp_cmd *c)
+{
+	page_frag_free(c->r2t_pdu);
+	page_frag_free(c->data_pdu);
+	page_frag_free(c->rsp_pdu);
+	page_frag_free(c->cmd_pdu);
+}
+
+static int nvmet_tcp_alloc_cmds(struct nvmet_tcp_queue *queue)
+{
+	struct nvmet_tcp_cmd *cmds;
+	int i, ret = -EINVAL, nr_cmds = queue->nr_cmds;
+
+	cmds = kcalloc(nr_cmds, sizeof(struct nvmet_tcp_cmd), GFP_KERNEL);
+	if (!cmds)
+		goto out;
+
+	for (i = 0; i < nr_cmds; i++) {
+		ret = nvmet_tcp_alloc_cmd(queue, cmds + i);
+		if (ret)
+			goto out_free;
+	}
+
+	queue->cmds = cmds;
+
+	return 0;
+out_free:
+	while (--i >= 0)
+		nvmet_tcp_free_cmd(cmds + i);
+	kfree(cmds);
+out:
+	return ret;
+}
+
+static void nvmet_tcp_free_cmds(struct nvmet_tcp_queue *queue)
+{
+	struct nvmet_tcp_cmd *cmds = queue->cmds;
+	int i;
+
+	for (i = 0; i < queue->nr_cmds; i++)
+		nvmet_tcp_free_cmd(cmds + i);
+
+	nvmet_tcp_free_cmd(&queue->connect);
+	kfree(cmds);
+}
+
+static void nvmet_tcp_restore_socket_callbacks(struct nvmet_tcp_queue *queue)
+{
+	struct socket *sock = queue->sock;
+
+	write_lock_bh(&sock->sk->sk_callback_lock);
+	sock->sk->sk_data_ready =  queue->data_ready;
+	sock->sk->sk_state_change = queue->state_change;
+	sock->sk->sk_write_space = queue->write_space;
+	sock->sk->sk_user_data = NULL;
+	write_unlock_bh(&sock->sk->sk_callback_lock);
+}
+
+static void nvmet_tcp_finish_cmd(struct nvmet_tcp_cmd *cmd)
+{
+	nvmet_req_uninit(&cmd->req);
+	nvmet_tcp_unmap_pdu_iovec(cmd);
+	sgl_free(cmd->req.sg);
+}
+
+static void nvmet_tcp_uninit_data_in_cmds(struct nvmet_tcp_queue *queue)
+{
+	struct nvmet_tcp_cmd *cmd = queue->cmds;
+	int i;
+
+	for (i = 0; i < queue->nr_cmds; i++, cmd++) {
+		if (nvmet_tcp_need_data_in(cmd))
+			nvmet_tcp_finish_cmd(cmd);
+	}
+
+	if (!queue->nr_cmds && nvmet_tcp_need_data_in(&queue->connect)) {
+		/* failed in connect */
+		nvmet_tcp_finish_cmd(&queue->connect);
+	}
+}
+
+static void nvmet_tcp_release_queue_work(struct work_struct *w)
+{
+	struct nvmet_tcp_queue *queue =
+		container_of(w, struct nvmet_tcp_queue, release_work);
+
+	mutex_lock(&nvmet_tcp_queue_mutex);
+	list_del_init(&queue->queue_list);
+	mutex_unlock(&nvmet_tcp_queue_mutex);
+
+	nvmet_tcp_restore_socket_callbacks(queue);
+	flush_work(&queue->io_work);
+
+	nvmet_tcp_uninit_data_in_cmds(queue);
+	nvmet_sq_destroy(&queue->nvme_sq);
+	cancel_work_sync(&queue->io_work);
+	sock_release(queue->sock);
+	nvmet_tcp_free_cmds(queue);
+	if (queue->hdr_digest || queue->data_digest)
+		nvmet_tcp_free_crypto(queue);
+	ida_simple_remove(&nvmet_tcp_queue_ida, queue->idx);
+
+	kfree(queue);
+}
+
+static void nvmet_tcp_data_ready(struct sock *sk)
+{
+	struct nvmet_tcp_queue *queue;
+
+	read_lock_bh(&sk->sk_callback_lock);
+	queue = sk->sk_user_data;
+	if (likely(queue))
+		queue_work_on(queue->cpu, nvmet_tcp_wq, &queue->io_work);
+	read_unlock_bh(&sk->sk_callback_lock);
+}
+
+static void nvmet_tcp_write_space(struct sock *sk)
+{
+	struct nvmet_tcp_queue *queue;
+
+	read_lock_bh(&sk->sk_callback_lock);
+	queue = sk->sk_user_data;
+	if (unlikely(!queue))
+		goto out;
+
+	if (unlikely(queue->state == NVMET_TCP_Q_CONNECTING)) {
+		queue->write_space(sk);
+		goto out;
+	}
+
+	if (sk_stream_is_writeable(sk)) {
+		clear_bit(SOCK_NOSPACE, &sk->sk_socket->flags);
+		queue_work_on(queue->cpu, nvmet_tcp_wq, &queue->io_work);
+	}
+out:
+	read_unlock_bh(&sk->sk_callback_lock);
+}
+
+static void nvmet_tcp_state_change(struct sock *sk)
+{
+	struct nvmet_tcp_queue *queue;
+
+	write_lock_bh(&sk->sk_callback_lock);
+	queue = sk->sk_user_data;
+	if (!queue)
+		goto done;
+
+	switch (sk->sk_state) {
+	case TCP_FIN_WAIT1:
+	case TCP_CLOSE_WAIT:
+	case TCP_CLOSE:
+		/* FALLTHRU */
+		sk->sk_user_data = NULL;
+		nvmet_tcp_schedule_release_queue(queue);
+		break;
+	default:
+		pr_warn("queue %d unhandled state %d\n",
+			queue->idx, sk->sk_state);
+	}
+done:
+	write_unlock_bh(&sk->sk_callback_lock);
+}
+
+static int nvmet_tcp_set_queue_sock(struct nvmet_tcp_queue *queue)
+{
+	struct socket *sock = queue->sock;
+	struct linger sol = { .l_onoff = 1, .l_linger = 0 };
+	int ret;
+
+	ret = kernel_getsockname(sock,
+		(struct sockaddr *)&queue->sockaddr);
+	if (ret < 0)
+		return ret;
+
+	ret = kernel_getpeername(sock,
+		(struct sockaddr *)&queue->sockaddr_peer);
+	if (ret < 0)
+		return ret;
+
+	/*
+	 * Cleanup whatever is sitting in the TCP transmit queue on socket
+	 * close. This is done to prevent stale data from being sent should
+	 * the network connection be restored before TCP times out.
+	 */
+	ret = kernel_setsockopt(sock, SOL_SOCKET, SO_LINGER,
+			(char *)&sol, sizeof(sol));
+	if (ret)
+		return ret;
+
+	write_lock_bh(&sock->sk->sk_callback_lock);
+	sock->sk->sk_user_data = queue;
+	queue->data_ready = sock->sk->sk_data_ready;
+	sock->sk->sk_data_ready = nvmet_tcp_data_ready;
+	queue->state_change = sock->sk->sk_state_change;
+	sock->sk->sk_state_change = nvmet_tcp_state_change;
+	queue->write_space = sock->sk->sk_write_space;
+	sock->sk->sk_write_space = nvmet_tcp_write_space;
+	write_unlock_bh(&sock->sk->sk_callback_lock);
+
+	return 0;
+}
+
+static int nvmet_tcp_alloc_queue(struct nvmet_tcp_port *port,
+		struct socket *newsock)
+{
+	struct nvmet_tcp_queue *queue;
+	int ret;
+
+	queue = kzalloc(sizeof(*queue), GFP_KERNEL);
+	if (!queue)
+		return -ENOMEM;
+
+	INIT_WORK(&queue->release_work, nvmet_tcp_release_queue_work);
+	INIT_WORK(&queue->io_work, nvmet_tcp_io_work);
+	queue->sock = newsock;
+	queue->port = port;
+	queue->nr_cmds = 0;
+	spin_lock_init(&queue->state_lock);
+	queue->state = NVMET_TCP_Q_CONNECTING;
+	INIT_LIST_HEAD(&queue->free_list);
+	init_llist_head(&queue->resp_list);
+	INIT_LIST_HEAD(&queue->resp_send_list);
+
+	queue->idx = ida_simple_get(&nvmet_tcp_queue_ida, 0, 0, GFP_KERNEL);
+	if (queue->idx < 0) {
+		ret = queue->idx;
+		goto out_free_queue;
+	}
+
+	ret = nvmet_tcp_alloc_cmd(queue, &queue->connect);
+	if (ret)
+		goto out_ida_remove;
+
+	ret = nvmet_sq_init(&queue->nvme_sq);
+	if (ret)
+		goto out_free_connect;
+
+	port->last_cpu = cpumask_next_wrap(port->last_cpu,
+				cpu_online_mask, -1, false);
+	queue->cpu = port->last_cpu;
+	nvmet_prepare_receive_pdu(queue);
+
+	mutex_lock(&nvmet_tcp_queue_mutex);
+	list_add_tail(&queue->queue_list, &nvmet_tcp_queue_list);
+	mutex_unlock(&nvmet_tcp_queue_mutex);
+
+	ret = nvmet_tcp_set_queue_sock(queue);
+	if (ret)
+		goto out_destroy_sq;
+
+	queue_work_on(queue->cpu, nvmet_tcp_wq, &queue->io_work);
+
+	return 0;
+out_destroy_sq:
+	mutex_lock(&nvmet_tcp_queue_mutex);
+	list_del_init(&queue->queue_list);
+	mutex_unlock(&nvmet_tcp_queue_mutex);
+	nvmet_sq_destroy(&queue->nvme_sq);
+out_free_connect:
+	nvmet_tcp_free_cmd(&queue->connect);
+out_ida_remove:
+	ida_simple_remove(&nvmet_tcp_queue_ida, queue->idx);
+out_free_queue:
+	kfree(queue);
+	return ret;
+}
+
+static void nvmet_tcp_accept_work(struct work_struct *w)
+{
+	struct nvmet_tcp_port *port =
+		container_of(w, struct nvmet_tcp_port, accept_work);
+	struct socket *newsock;
+	int ret;
+
+	while (true) {
+		ret = kernel_accept(port->sock, &newsock, O_NONBLOCK);
+		if (ret < 0) {
+			if (ret != -EAGAIN)
+				pr_warn("failed to accept err=%d\n", ret);
+			return;
+		}
+		ret = nvmet_tcp_alloc_queue(port, newsock);
+		if (ret) {
+			pr_err("failed to allocate queue\n");
+			sock_release(newsock);
+		}
+	}
+}
+
+static void nvmet_tcp_listen_data_ready(struct sock *sk)
+{
+	struct nvmet_tcp_port *port;
+
+	read_lock_bh(&sk->sk_callback_lock);
+	port = sk->sk_user_data;
+	if (!port)
+		goto out;
+
+	if (sk->sk_state == TCP_LISTEN)
+		schedule_work(&port->accept_work);
+out:
+	read_unlock_bh(&sk->sk_callback_lock);
+}
+
+static int nvmet_tcp_add_port(struct nvmet_port *nport)
+{
+	struct nvmet_tcp_port *port;
+	__kernel_sa_family_t af;
+	int opt, ret;
+
+	port = kzalloc(sizeof(*port), GFP_KERNEL);
+	if (!port)
+		return -ENOMEM;
+
+	switch (nport->disc_addr.adrfam) {
+	case NVMF_ADDR_FAMILY_IP4:
+		af = AF_INET;
+		break;
+	case NVMF_ADDR_FAMILY_IP6:
+		af = AF_INET6;
+		break;
+	default:
+		pr_err("address family %d not supported\n",
+				nport->disc_addr.adrfam);
+		ret = -EINVAL;
+		goto err_port;
+	}
+
+	ret = inet_pton_with_scope(&init_net, af, nport->disc_addr.traddr,
+			nport->disc_addr.trsvcid, &port->addr);
+	if (ret) {
+		pr_err("malformed ip/port passed: %s:%s\n",
+			nport->disc_addr.traddr, nport->disc_addr.trsvcid);
+		goto err_port;
+	}
+
+	port->nport = nport;
+	port->last_cpu = -1;
+	INIT_WORK(&port->accept_work, nvmet_tcp_accept_work);
+	if (port->nport->inline_data_size < 0)
+		port->nport->inline_data_size = NVMET_TCP_DEF_INLINE_DATA_SIZE;
+
+	ret = sock_create(port->addr.ss_family, SOCK_STREAM,
+				IPPROTO_TCP, &port->sock);
+	if (ret) {
+		pr_err("failed to create a socket\n");
+		goto err_port;
+	}
+
+	port->sock->sk->sk_user_data = port;
+	port->data_ready = port->sock->sk->sk_data_ready;
+	port->sock->sk->sk_data_ready = nvmet_tcp_listen_data_ready;
+
+	opt = 1;
+	ret = kernel_setsockopt(port->sock, IPPROTO_TCP,
+			TCP_NODELAY, (char *)&opt, sizeof(opt));
+	if (ret) {
+		pr_err("failed to set TCP_NODELAY sock opt %d\n", ret);
+		goto err_sock;
+	}
+
+	ret = kernel_setsockopt(port->sock, SOL_SOCKET, SO_REUSEADDR,
+			(char *)&opt, sizeof(opt));
+	if (ret) {
+		pr_err("failed to set SO_REUSEADDR sock opt %d\n", ret);
+		goto err_sock;
+	}
+
+	ret = kernel_bind(port->sock, (struct sockaddr *)&port->addr,
+			sizeof(port->addr));
+	if (ret) {
+		pr_err("failed to bind port socket %d\n", ret);
+		goto err_sock;
+	}
+
+	ret = kernel_listen(port->sock, 128);
+	if (ret) {
+		pr_err("failed to listen %d on port sock\n", ret);
+		goto err_sock;
+	}
+
+	nport->priv = port;
+	pr_info("enabling port %d (%pISpc)\n",
+		le16_to_cpu(nport->disc_addr.portid), &port->addr);
+
+	return 0;
+
+err_sock:
+	sock_release(port->sock);
+err_port:
+	kfree(port);
+	return ret;
+}
+
+static void nvmet_tcp_remove_port(struct nvmet_port *nport)
+{
+	struct nvmet_tcp_port *port = nport->priv;
+
+	write_lock_bh(&port->sock->sk->sk_callback_lock);
+	port->sock->sk->sk_data_ready = port->data_ready;
+	port->sock->sk->sk_user_data = NULL;
+	write_unlock_bh(&port->sock->sk->sk_callback_lock);
+	cancel_work_sync(&port->accept_work);
+
+	sock_release(port->sock);
+	kfree(port);
+}
+
+static void nvmet_tcp_delete_ctrl(struct nvmet_ctrl *ctrl)
+{
+	struct nvmet_tcp_queue *queue;
+
+	mutex_lock(&nvmet_tcp_queue_mutex);
+	list_for_each_entry(queue, &nvmet_tcp_queue_list, queue_list)
+		if (queue->nvme_sq.ctrl == ctrl)
+			kernel_sock_shutdown(queue->sock, SHUT_RDWR);
+	mutex_unlock(&nvmet_tcp_queue_mutex);
+}
+
+static u16 nvmet_tcp_install_queue(struct nvmet_sq *sq)
+{
+	struct nvmet_tcp_queue *queue =
+		container_of(sq, struct nvmet_tcp_queue, nvme_sq);
+
+	if (sq->qid == 0) {
+		/* Let inflight controller teardown complete */
+		flush_scheduled_work();
+	}
+
+	queue->nr_cmds = sq->size * 2;
+	if (nvmet_tcp_alloc_cmds(queue))
+		return NVME_SC_INTERNAL;
+	return 0;
+}
+
+static void nvmet_tcp_disc_port_addr(struct nvmet_req *req,
+		struct nvmet_port *nport, char *traddr)
+{
+	struct nvmet_tcp_port *port = nport->priv;
+
+	if (inet_addr_is_any((struct sockaddr *)&port->addr)) {
+		struct nvmet_tcp_cmd *cmd =
+			container_of(req, struct nvmet_tcp_cmd, req);
+		struct nvmet_tcp_queue *queue = cmd->queue;
+
+		sprintf(traddr, "%pISc", (struct sockaddr *)&queue->sockaddr);
+	} else {
+		memcpy(traddr, nport->disc_addr.traddr, NVMF_TRADDR_SIZE);
+	}
+}
+
+static struct nvmet_fabrics_ops nvmet_tcp_ops = {
+	.owner			= THIS_MODULE,
+	.type			= NVMF_TRTYPE_TCP,
+	.msdbd			= 1,
+	.has_keyed_sgls		= 0,
+	.add_port		= nvmet_tcp_add_port,
+	.remove_port		= nvmet_tcp_remove_port,
+	.queue_response		= nvmet_tcp_queue_response,
+	.delete_ctrl		= nvmet_tcp_delete_ctrl,
+	.install_queue		= nvmet_tcp_install_queue,
+	.disc_traddr		= nvmet_tcp_disc_port_addr,
+};
+
+static int __init nvmet_tcp_init(void)
+{
+	int ret;
+
+	nvmet_tcp_wq = alloc_workqueue("nvmet_tcp_wq", WQ_HIGHPRI, 0);
+	if (!nvmet_tcp_wq)
+		return -ENOMEM;
+
+	ret = nvmet_register_transport(&nvmet_tcp_ops);
+	if (ret)
+		goto err;
+
+	return 0;
+err:
+	destroy_workqueue(nvmet_tcp_wq);
+	return ret;
+}
+
+static void __exit nvmet_tcp_exit(void)
+{
+	struct nvmet_tcp_queue *queue;
+
+	nvmet_unregister_transport(&nvmet_tcp_ops);
+
+	flush_scheduled_work();
+	mutex_lock(&nvmet_tcp_queue_mutex);
+	list_for_each_entry(queue, &nvmet_tcp_queue_list, queue_list)
+		kernel_sock_shutdown(queue->sock, SHUT_RDWR);
+	mutex_unlock(&nvmet_tcp_queue_mutex);
+	flush_scheduled_work();
+
+	destroy_workqueue(nvmet_tcp_wq);
+}
+
+module_init(nvmet_tcp_init);
+module_exit(nvmet_tcp_exit);
+
+MODULE_LICENSE("GPL v2");
+MODULE_ALIAS("nvmet-transport-3"); /* 3 == NVMF_TRTYPE_TCP */