Merge tag 'nvme-5.13-2021-04-06' of git://git.infradead.org/nvme into for-5.13/drivers

Pull NVMe updates from Christoph:

"nvme updates for Linux 5.13

- fix handling of very large MDTS values (Bart Van Assche)
- retrigger ANA log update if group descriptor isn't found
(Hannes Reinecke)
- fix locking contexts in nvme-tcp and nvmet-tcp (Sagi Grimberg)
- return proper error code from discovery ctrl (Hou Pu)
- verify the SGLS field in nvmet-tcp and nvmet-fc (Max Gurtovoy)
- disallow passthru cmd from targeting a nsid != nsid of the block dev
(Niklas Cassel)
- do not allow model_number exceed 40 bytes in nvmet (Noam Gottlieb)
- enable optional queue idle period tracking in nvmet-tcp
(Mark Wunderlich)
- various cleanups and optimizations (Chaitanya Kulkarni, Kanchan Joshi)
- expose fast_io_fail_tmo in sysfs (Daniel Wagner)
- implement non-MDTS command limits (Keith Busch)
- reduce warnings for unhandled command effects (Keith Busch)
- allocate storage for the SQE as part of the nvme_request (Keith Busch)"

* tag 'nvme-5.13-2021-04-06' of git://git.infradead.org/nvme: (33 commits)
nvme: fix handling of large MDTS values
nvme: implement non-mdts command limits
nvme: disallow passthru cmd from targeting a nsid != nsid of the block dev
nvme: retrigger ANA log update if group descriptor isn't found
nvme: export fast_io_fail_tmo to sysfs
nvme: remove superfluous else in nvme_ctrl_loss_tmo_store
nvme: use sysfs_emit instead of sprintf
nvme-fc: check sgl supported by target
nvme-tcp: check sgl supported by target
nvmet-tcp: enable optional queue idle period tracking
nvmet-tcp: fix incorrect locking in state_change sk callback
nvme-tcp: block BH in sk state_change sk callback
nvmet: return proper error code from discovery ctrl
nvme: warn of unhandled effects only once
nvme: use driver pdu command for passthrough
nvme-pci: allocate nvme_command within driver pdu
nvmet: do not allow model_number exceed 40 bytes
nvmet: remove unnecessary ctrl parameter
nvmet-fc: update function documentation
nvme-fc: fix the function documentation comment
...

drivers/nvme/host/core.c

···
 
 static inline void nvme_clear_nvme_request(struct request *req)
 {
-	if (!(req->rq_flags & RQF_DONTPREP)) {
-		nvme_req(req)->retries = 0;
-		nvme_req(req)->flags = 0;
-		req->rq_flags |= RQF_DONTPREP;
-	}
+	struct nvme_command *cmd = nvme_req(req)->cmd;
+
+	memset(cmd, 0, sizeof(*cmd));
+	nvme_req(req)->retries = 0;
+	nvme_req(req)->flags = 0;
+	req->rq_flags |= RQF_DONTPREP;
 }
 
 static inline unsigned int nvme_req_op(struct nvme_command *cmd)
···
 	else /* no queuedata implies admin queue */
 		req->timeout = NVME_ADMIN_TIMEOUT;
 
+	/* passthru commands should let the driver set the SGL flags */
+	cmd->common.flags &= ~NVME_CMD_SGL_ALL;
+
 	req->cmd_flags |= REQ_FAILFAST_DRIVER;
 	nvme_clear_nvme_request(req);
-	nvme_req(req)->cmd = cmd;
+	memcpy(nvme_req(req)->cmd, cmd, sizeof(*cmd));
 }
 
 struct request *nvme_alloc_request(struct request_queue *q,
···
 
 	if (streamid < ARRAY_SIZE(req->q->write_hints))
 		req->q->write_hints[streamid] += blk_rq_bytes(req) >> 9;
-}
-
-static void nvme_setup_passthrough(struct request *req,
-		struct nvme_command *cmd)
-{
-	memcpy(cmd, nvme_req(req)->cmd, sizeof(*cmd));
-	/* passthru commands should let the driver set the SGL flags */
-	cmd->common.flags &= ~NVME_CMD_SGL_ALL;
 }
 
 static inline void nvme_setup_flush(struct nvme_ns *ns,
···
 }
 EXPORT_SYMBOL_GPL(nvme_cleanup_cmd);
 
-blk_status_t nvme_setup_cmd(struct nvme_ns *ns, struct request *req,
-		struct nvme_command *cmd)
+blk_status_t nvme_setup_cmd(struct nvme_ns *ns, struct request *req)
 {
+	struct nvme_command *cmd = nvme_req(req)->cmd;
 	blk_status_t ret = BLK_STS_OK;
 
-	nvme_clear_nvme_request(req);
+	if (!(req->rq_flags & RQF_DONTPREP))
+		nvme_clear_nvme_request(req);
 
-	memset(cmd, 0, sizeof(*cmd));
 	switch (req_op(req)) {
 	case REQ_OP_DRV_IN:
 	case REQ_OP_DRV_OUT:
-		nvme_setup_passthrough(req, cmd);
+		/* these are setup prior to execution in nvme_init_request() */
 		break;
 	case REQ_OP_FLUSH:
 		nvme_setup_flush(ns, cmd);
···
 		if (ns->head->effects)
 			effects = le32_to_cpu(ns->head->effects->iocs[opcode]);
 		if (effects & ~(NVME_CMD_EFFECTS_CSUPP | NVME_CMD_EFFECTS_LBCC))
-			dev_warn(ctrl->device,
-				 "IO command:%02x has unhandled effects:%08x\n",
-				 opcode, effects);
+			dev_warn_once(ctrl->device,
+				"IO command:%02x has unhandled effects:%08x\n",
+				opcode, effects);
 		return 0;
 	}
 
···
 		mutex_unlock(&ctrl->scan_lock);
 	}
 	if (effects & NVME_CMD_EFFECTS_CCC)
-		nvme_init_identify(ctrl);
+		nvme_init_ctrl_finish(ctrl);
 	if (effects & (NVME_CMD_EFFECTS_NIC | NVME_CMD_EFFECTS_NCC)) {
 		nvme_queue_scan(ctrl);
 		flush_work(&ctrl->scan_work);
···
 
 	effects = nvme_passthru_start(ctrl, ns, cmd->common.opcode);
 	blk_execute_rq(disk, rq, 0);
-	nvme_passthru_end(ctrl, effects);
+	if (effects) /* nothing to be done for zero cmd effects */
+		nvme_passthru_end(ctrl, effects);
 }
 EXPORT_SYMBOL_NS_GPL(nvme_execute_passthru_rq, NVME_TARGET_PASSTHRU);
 
···
 		return -EFAULT;
 	if (cmd.flags)
 		return -EINVAL;
+	if (ns && cmd.nsid != ns->head->ns_id) {
+		dev_err(ctrl->device,
+			"%s: nsid (%u) in cmd does not match nsid (%u) of namespace\n",
+			current->comm, cmd.nsid, ns->head->ns_id);
+		return -EINVAL;
+	}
 
 	memset(&c, 0, sizeof(c));
 	c.common.opcode = cmd.opcode;
···
 		return -EFAULT;
 	if (cmd.flags)
 		return -EINVAL;
+	if (ns && cmd.nsid != ns->head->ns_id) {
+		dev_err(ctrl->device,
+			"%s: nsid (%u) in cmd does not match nsid (%u) of namespace\n",
+			current->comm, cmd.nsid, ns->head->ns_id);
+		return -EINVAL;
+	}
 
 	memset(&c, 0, sizeof(c));
 	c.common.opcode = cmd.opcode;
···
 	struct request_queue *queue = disk->queue;
 	u32 size = queue_logical_block_size(queue);
 
-	if (!(ctrl->oncs & NVME_CTRL_ONCS_DSM)) {
+	if (ctrl->max_discard_sectors == 0) {
 		blk_queue_flag_clear(QUEUE_FLAG_DISCARD, queue);
 		return;
 	}
···
 	if (blk_queue_flag_test_and_set(QUEUE_FLAG_DISCARD, queue))
 		return;
 
-	blk_queue_max_discard_sectors(queue, UINT_MAX);
-	blk_queue_max_discard_segments(queue, NVME_DSM_MAX_RANGES);
+	blk_queue_max_discard_sectors(queue, ctrl->max_discard_sectors);
+	blk_queue_max_discard_segments(queue, ctrl->max_discard_segments);
 
 	if (ctrl->quirks & NVME_QUIRK_DEALLOCATE_ZEROES)
 		blk_queue_max_write_zeroes_sectors(queue, UINT_MAX);
-}
-
-static void nvme_config_write_zeroes(struct gendisk *disk, struct nvme_ns *ns)
-{
-	u64 max_blocks;
-
-	if (!(ns->ctrl->oncs & NVME_CTRL_ONCS_WRITE_ZEROES) ||
-	    (ns->ctrl->quirks & NVME_QUIRK_DISABLE_WRITE_ZEROES))
-		return;
-	/*
-	 * Even though NVMe spec explicitly states that MDTS is not
-	 * applicable to the write-zeroes:- "The restriction does not apply to
-	 * commands that do not transfer data between the host and the
-	 * controller (e.g., Write Uncorrectable ro Write Zeroes command).".
-	 * In order to be more cautious use controller's max_hw_sectors value
-	 * to configure the maximum sectors for the write-zeroes which is
-	 * configured based on the controller's MDTS field in the
-	 * nvme_init_identify() if available.
-	 */
-	if (ns->ctrl->max_hw_sectors == UINT_MAX)
-		max_blocks = (u64)USHRT_MAX + 1;
-	else
-		max_blocks = ns->ctrl->max_hw_sectors + 1;
-
-	blk_queue_max_write_zeroes_sectors(disk->queue,
-					   nvme_lba_to_sect(ns, max_blocks));
 }
 
 static bool nvme_ns_ids_valid(struct nvme_ns_ids *ids)
···
 	set_capacity_and_notify(disk, capacity);
 
 	nvme_config_discard(disk, ns);
-	nvme_config_write_zeroes(disk, ns);
+	blk_queue_max_write_zeroes_sectors(disk->queue,
+					   ns->ctrl->max_zeroes_sectors);
 
 	set_disk_ro(disk, (id->nsattr & NVME_NS_ATTR_RO) ||
 		test_bit(NVME_NS_FORCE_RO, &ns->flags));
···
 		enum pr_type type, bool abort)
 {
 	u32 cdw10 = nvme_pr_type(type) << 8 | (abort ? 2 : 1);
+
 	return nvme_pr_command(bdev, cdw10, old, new, nvme_cmd_resv_acquire);
 }
 
 static int nvme_pr_clear(struct block_device *bdev, u64 key)
 {
 	u32 cdw10 = 1 | (key ? 1 << 3 : 0);
+
 	return nvme_pr_command(bdev, cdw10, key, 0, nvme_cmd_resv_register);
 }
 
 static int nvme_pr_release(struct block_device *bdev, u64 key, enum pr_type type)
 {
 	u32 cdw10 = nvme_pr_type(type) << 8 | (key ? 1 << 3 : 0);
+
 	return nvme_pr_command(bdev, cdw10, key, 0, nvme_cmd_resv_release);
 }
 
···
 {									\
 	struct nvme_subsystem *subsys =					\
 		container_of(dev, struct nvme_subsystem, dev);		\
-	return sprintf(buf, "%.*s\n",					\
-		       (int)sizeof(subsys->field), subsys->field);	\
+	return sysfs_emit(buf, "%.*s\n",				\
+			   (int)sizeof(subsys->field), subsys->field);	\
 }									\
 static SUBSYS_ATTR_RO(field, S_IRUGO, subsys_##field##_show);
 
···
 	return 0;
 }
 
-/*
- * Initialize the cached copies of the Identify data and various controller
- * register in our nvme_ctrl structure.  This should be called as soon as
- * the admin queue is fully up and running.
- */
-int nvme_init_identify(struct nvme_ctrl *ctrl)
+static inline u32 nvme_mps_to_sectors(struct nvme_ctrl *ctrl, u32 units)
+{
+	u32 page_shift = NVME_CAP_MPSMIN(ctrl->cap) + 12, val;
+
+	if (check_shl_overflow(1U, units + page_shift - 9, &val))
+		return UINT_MAX;
+	return val;
+}
+
+static int nvme_init_non_mdts_limits(struct nvme_ctrl *ctrl)
+{
+	struct nvme_command c = { };
+	struct nvme_id_ctrl_nvm *id;
+	int ret;
+
+	if (ctrl->oncs & NVME_CTRL_ONCS_DSM) {
+		ctrl->max_discard_sectors = UINT_MAX;
+		ctrl->max_discard_segments = NVME_DSM_MAX_RANGES;
+	} else {
+		ctrl->max_discard_sectors = 0;
+		ctrl->max_discard_segments = 0;
+	}
+
+	/*
+	 * Even though NVMe spec explicitly states that MDTS is not applicable
+	 * to the write-zeroes, we are cautious and limit the size to the
+	 * controllers max_hw_sectors value, which is based on the MDTS field
+	 * and possibly other limiting factors.
+	 */
+	if ((ctrl->oncs & NVME_CTRL_ONCS_WRITE_ZEROES) &&
+	    !(ctrl->quirks & NVME_QUIRK_DISABLE_WRITE_ZEROES))
+		ctrl->max_zeroes_sectors = ctrl->max_hw_sectors;
+	else
+		ctrl->max_zeroes_sectors = 0;
+
+	if (nvme_ctrl_limited_cns(ctrl))
+		return 0;
+
+	id = kzalloc(sizeof(*id), GFP_KERNEL);
+	if (!id)
+		return 0;
+
+	c.identify.opcode = nvme_admin_identify;
+	c.identify.cns = NVME_ID_CNS_CS_CTRL;
+	c.identify.csi = NVME_CSI_NVM;
+
+	ret = nvme_submit_sync_cmd(ctrl->admin_q, &c, id, sizeof(*id));
+	if (ret)
+		goto free_data;
+
+	if (id->dmrl)
+		ctrl->max_discard_segments = id->dmrl;
+	if (id->dmrsl)
+		ctrl->max_discard_sectors = le32_to_cpu(id->dmrsl);
+	if (id->wzsl)
+		ctrl->max_zeroes_sectors = nvme_mps_to_sectors(ctrl, id->wzsl);
+
+free_data:
+	kfree(id);
+	return ret;
+}
+
+static int nvme_init_identify(struct nvme_ctrl *ctrl)
 {
 	struct nvme_id_ctrl *id;
-	int ret, page_shift;
 	u32 max_hw_sectors;
 	bool prev_apst_enabled;
-
-	ret = ctrl->ops->reg_read32(ctrl, NVME_REG_VS, &ctrl->vs);
-	if (ret) {
-		dev_err(ctrl->device, "Reading VS failed (%d)\n", ret);
-		return ret;
-	}
-	page_shift = NVME_CAP_MPSMIN(ctrl->cap) + 12;
-	ctrl->sqsize = min_t(u16, NVME_CAP_MQES(ctrl->cap), ctrl->sqsize);
-
-	if (ctrl->vs >= NVME_VS(1, 1, 0))
-		ctrl->subsystem = NVME_CAP_NSSRC(ctrl->cap);
+	int ret;
 
 	ret = nvme_identify_ctrl(ctrl, &id);
 	if (ret) {
···
 		ctrl->cntlid = le16_to_cpu(id->cntlid);
 
 	if (!ctrl->identified) {
-		int i;
+		unsigned int i;
 
 		ret = nvme_init_subsystem(ctrl, id);
 		if (ret)
···
 	atomic_set(&ctrl->abort_limit, id->acl + 1);
 	ctrl->vwc = id->vwc;
 	if (id->mdts)
-		max_hw_sectors = 1 << (id->mdts + page_shift - 9);
+		max_hw_sectors = nvme_mps_to_sectors(ctrl, id->mdts);
 	else
 		max_hw_sectors = UINT_MAX;
 	ctrl->max_hw_sectors =
···
 	}
 
 	ret = nvme_mpath_init(ctrl, id);
-	kfree(id);
-
 	if (ret < 0)
-		return ret;
+		goto out_free;
 
 	if (ctrl->apst_enabled && !prev_apst_enabled)
 		dev_pm_qos_expose_latency_tolerance(ctrl->device);
 	else if (!ctrl->apst_enabled && prev_apst_enabled)
 		dev_pm_qos_hide_latency_tolerance(ctrl->device);
+
+out_free:
+	kfree(id);
+	return ret;
+}
+
+/*
+ * Initialize the cached copies of the Identify data and various controller
+ * register in our nvme_ctrl structure.  This should be called as soon as
+ * the admin queue is fully up and running.
+ */
+int nvme_init_ctrl_finish(struct nvme_ctrl *ctrl)
+{
+	int ret;
+
+	ret = ctrl->ops->reg_read32(ctrl, NVME_REG_VS, &ctrl->vs);
+	if (ret) {
+		dev_err(ctrl->device, "Reading VS failed (%d)\n", ret);
+		return ret;
+	}
+
+	ctrl->sqsize = min_t(u16, NVME_CAP_MQES(ctrl->cap), ctrl->sqsize);
+
+	if (ctrl->vs >= NVME_VS(1, 1, 0))
+		ctrl->subsystem = NVME_CAP_NSSRC(ctrl->cap);
+
+	ret = nvme_init_identify(ctrl);
+	if (ret)
+		return ret;
+
+	ret = nvme_init_non_mdts_limits(ctrl);
+	if (ret < 0)
+		return ret;
 
 	ret = nvme_configure_apst(ctrl);
 	if (ret < 0)
···
 	ctrl->identified = true;
 
 	return 0;
-
-out_free:
-	kfree(id);
-	return ret;
 }
-EXPORT_SYMBOL_GPL(nvme_init_identify);
+EXPORT_SYMBOL_GPL(nvme_init_ctrl_finish);
 
 static int nvme_dev_open(struct inode *inode, struct file *file)
 {
···
 	int model_len = sizeof(subsys->model);
 
 	if (!uuid_is_null(&ids->uuid))
-		return sprintf(buf, "uuid.%pU\n", &ids->uuid);
+		return sysfs_emit(buf, "uuid.%pU\n", &ids->uuid);
 
 	if (memchr_inv(ids->nguid, 0, sizeof(ids->nguid)))
-		return sprintf(buf, "eui.%16phN\n", ids->nguid);
+		return sysfs_emit(buf, "eui.%16phN\n", ids->nguid);
 
 	if (memchr_inv(ids->eui64, 0, sizeof(ids->eui64)))
-		return sprintf(buf, "eui.%8phN\n", ids->eui64);
+		return sysfs_emit(buf, "eui.%8phN\n", ids->eui64);
 
 	while (serial_len > 0 && (subsys->serial[serial_len - 1] == ' ' ||
 				  subsys->serial[serial_len - 1] == '\0'))
···
 				 subsys->model[model_len - 1] == '\0'))
 		model_len--;
 
-	return sprintf(buf, "nvme.%04x-%*phN-%*phN-%08x\n", subsys->vendor_id,
+	return sysfs_emit(buf, "nvme.%04x-%*phN-%*phN-%08x\n", subsys->vendor_id,
 		serial_len, subsys->serial, model_len, subsys->model,
 		head->ns_id);
 }
···
 static ssize_t nguid_show(struct device *dev, struct device_attribute *attr,
 		char *buf)
 {
-	return sprintf(buf, "%pU\n", dev_to_ns_head(dev)->ids.nguid);
+	return sysfs_emit(buf, "%pU\n", dev_to_ns_head(dev)->ids.nguid);
 }
 static DEVICE_ATTR_RO(nguid);
 
···
 	if (uuid_is_null(&ids->uuid)) {
 		printk_ratelimited(KERN_WARNING
 				   "No UUID available providing old NGUID\n");
-		return sprintf(buf, "%pU\n", ids->nguid);
+		return sysfs_emit(buf, "%pU\n", ids->nguid);
 	}
-	return sprintf(buf, "%pU\n", &ids->uuid);
+	return sysfs_emit(buf, "%pU\n", &ids->uuid);
 }
 static DEVICE_ATTR_RO(uuid);
 
 static ssize_t eui_show(struct device *dev, struct device_attribute *attr,
 		char *buf)
 {
-	return sprintf(buf, "%8ph\n", dev_to_ns_head(dev)->ids.eui64);
+	return sysfs_emit(buf, "%8ph\n", dev_to_ns_head(dev)->ids.eui64);
 }
 static DEVICE_ATTR_RO(eui);
 
 static ssize_t nsid_show(struct device *dev, struct device_attribute *attr,
 		char *buf)
 {
-	return sprintf(buf, "%d\n", dev_to_ns_head(dev)->ns_id);
+	return sysfs_emit(buf, "%d\n", dev_to_ns_head(dev)->ns_id);
 }
 static DEVICE_ATTR_RO(nsid);
 
···
 			    struct device_attribute *attr, char *buf)		\
 {										\
         struct nvme_ctrl *ctrl = dev_get_drvdata(dev);				\
-        return sprintf(buf, "%.*s\n",						\
+        return sysfs_emit(buf, "%.*s\n",					\
 		(int)sizeof(ctrl->subsys->field), ctrl->subsys->field);		\
 }										\
 static DEVICE_ATTR(field, S_IRUGO, field##_show, NULL);
···
 			    struct device_attribute *attr, char *buf)		\
 {										\
         struct nvme_ctrl *ctrl = dev_get_drvdata(dev);				\
-        return sprintf(buf, "%d\n", ctrl->field);	\
+        return sysfs_emit(buf, "%d\n", ctrl->field);				\
 }										\
 static DEVICE_ATTR(field, S_IRUGO, field##_show, NULL);
 
···
 
 	if ((unsigned)ctrl->state < ARRAY_SIZE(state_name) &&
 	    state_name[ctrl->state])
-		return sprintf(buf, "%s\n", state_name[ctrl->state]);
+		return sysfs_emit(buf, "%s\n", state_name[ctrl->state]);
 
-	return sprintf(buf, "unknown state\n");
+	return sysfs_emit(buf, "unknown state\n");
 }
 
 static DEVICE_ATTR(state, S_IRUGO, nvme_sysfs_show_state, NULL);
···
 	struct nvmf_ctrl_options *opts = ctrl->opts;
 
 	if (ctrl->opts->max_reconnects == -1)
-		return sprintf(buf, "off\n");
-	return sprintf(buf, "%d\n",
-			opts->max_reconnects * opts->reconnect_delay);
+		return sysfs_emit(buf, "off\n");
+	return sysfs_emit(buf, "%d\n",
+			  opts->max_reconnects * opts->reconnect_delay);
 }
 
 static ssize_t nvme_ctrl_loss_tmo_store(struct device *dev,
···
 	if (err)
 		return -EINVAL;
 
-	else if (ctrl_loss_tmo < 0)
+	if (ctrl_loss_tmo < 0)
 		opts->max_reconnects = -1;
 	else
 		opts->max_reconnects = DIV_ROUND_UP(ctrl_loss_tmo,
···
 	struct nvme_ctrl *ctrl = dev_get_drvdata(dev);
 
 	if (ctrl->opts->reconnect_delay == -1)
-		return sprintf(buf, "off\n");
-	return sprintf(buf, "%d\n", ctrl->opts->reconnect_delay);
+		return sysfs_emit(buf, "off\n");
+	return sysfs_emit(buf, "%d\n", ctrl->opts->reconnect_delay);
 }
 
 static ssize_t nvme_ctrl_reconnect_delay_store(struct device *dev,
···
 }
 static DEVICE_ATTR(reconnect_delay, S_IRUGO | S_IWUSR,
 	nvme_ctrl_reconnect_delay_show, nvme_ctrl_reconnect_delay_store);
+
+static ssize_t nvme_ctrl_fast_io_fail_tmo_show(struct device *dev,
+		struct device_attribute *attr, char *buf)
+{
+	struct nvme_ctrl *ctrl = dev_get_drvdata(dev);
+
+	if (ctrl->opts->fast_io_fail_tmo == -1)
+		return sysfs_emit(buf, "off\n");
+	return sysfs_emit(buf, "%d\n", ctrl->opts->fast_io_fail_tmo);
+}
+
+static ssize_t nvme_ctrl_fast_io_fail_tmo_store(struct device *dev,
+		struct device_attribute *attr, const char *buf, size_t count)
+{
+	struct nvme_ctrl *ctrl = dev_get_drvdata(dev);
+	struct nvmf_ctrl_options *opts = ctrl->opts;
+	int fast_io_fail_tmo, err;
+
+	err = kstrtoint(buf, 10, &fast_io_fail_tmo);
+	if (err)
+		return -EINVAL;
+
+	if (fast_io_fail_tmo < 0)
+		opts->fast_io_fail_tmo = -1;
+	else
+		opts->fast_io_fail_tmo = fast_io_fail_tmo;
+	return count;
+}
+static DEVICE_ATTR(fast_io_fail_tmo, S_IRUGO | S_IWUSR,
+	nvme_ctrl_fast_io_fail_tmo_show, nvme_ctrl_fast_io_fail_tmo_store);
 
 static struct attribute *nvme_dev_attrs[] = {
 	&dev_attr_reset_controller.attr,
···
 	&dev_attr_hostid.attr,
 	&dev_attr_ctrl_loss_tmo.attr,
 	&dev_attr_reconnect_delay.attr,
+	&dev_attr_fast_io_fail_tmo.attr,
 	NULL
 };
 
···
 	BUILD_BUG_ON(sizeof(struct nvme_id_ns) != NVME_IDENTIFY_DATA_SIZE);
 	BUILD_BUG_ON(sizeof(struct nvme_id_ns_zns) != NVME_IDENTIFY_DATA_SIZE);
 	BUILD_BUG_ON(sizeof(struct nvme_id_ctrl_zns) != NVME_IDENTIFY_DATA_SIZE);
+	BUILD_BUG_ON(sizeof(struct nvme_id_ctrl_nvm) != NVME_IDENTIFY_DATA_SIZE);
 	BUILD_BUG_ON(sizeof(struct nvme_lba_range_type) != 64);
 	BUILD_BUG_ON(sizeof(struct nvme_smart_log) != 512);
 	BUILD_BUG_ON(sizeof(struct nvme_dbbuf) != 64);

drivers/nvme/host/fc.c

···
  *
  * If this routine returns error, the LLDD should abort the exchange.
  *
- * @remoteport: pointer to the (registered) remote port that the LS
+ * @portptr:    pointer to the (registered) remote port that the LS
  *              was received from. The remoteport is associated with
  *              a specific localport.
  * @lsrsp:      pointer to a nvmefc_ls_rsp response structure to be
···
 	op->op.fcp_req.first_sgl = op->sgl;
 	op->op.fcp_req.private = &op->priv[0];
 	nvme_req(rq)->ctrl = &ctrl->ctrl;
+	nvme_req(rq)->cmd = &op->op.cmd_iu.sqe;
 	return res;
 }
 
···
 	struct nvme_fc_ctrl *ctrl = queue->ctrl;
 	struct request *rq = bd->rq;
 	struct nvme_fc_fcp_op *op = blk_mq_rq_to_pdu(rq);
-	struct nvme_fc_cmd_iu *cmdiu = &op->cmd_iu;
-	struct nvme_command *sqe = &cmdiu->sqe;
 	enum nvmefc_fcp_datadir	io_dir;
 	bool queue_ready = test_bit(NVME_FC_Q_LIVE, &queue->flags);
 	u32 data_len;
···
 	    !nvmf_check_ready(&queue->ctrl->ctrl, rq, queue_ready))
 		return nvmf_fail_nonready_command(&queue->ctrl->ctrl, rq);
 
-	ret = nvme_setup_cmd(ns, rq, sqe);
+	ret = nvme_setup_cmd(ns, rq);
 	if (ret)
 		return ret;
 
···
 
 	blk_mq_unquiesce_queue(ctrl->ctrl.admin_q);
 
-	ret = nvme_init_identify(&ctrl->ctrl);
+	ret = nvme_init_ctrl_finish(&ctrl->ctrl);
 	if (ret || test_bit(ASSOC_FAILED, &ctrl->flags))
 		goto out_disconnect_admin_queue;
 
···
 	}
 
 	/* FC-NVME supports normal SGL Data Block Descriptors */
+	if (!(ctrl->ctrl.sgls & ((1 << 0) | (1 << 1)))) {
+		dev_err(ctrl->ctrl.device,
+			"Mandatory sgls are not supported!\n");
+		goto out_disconnect_admin_queue;
+	}
 
 	if (opts->queue_size > ctrl->ctrl.maxcmd) {
 		/* warn if maxcmd is lower than queue_size */

drivers/nvme/host/multipath.c

···
 	struct nvme_subsystem *subsys =
 		container_of(dev, struct nvme_subsystem, dev);
 
-	return sprintf(buf, "%s\n",
-			nvme_iopolicy_names[READ_ONCE(subsys->iopolicy)]);
+	return sysfs_emit(buf, "%s\n",
+			  nvme_iopolicy_names[READ_ONCE(subsys->iopolicy)]);
 }
 
 static ssize_t nvme_subsys_iopolicy_store(struct device *dev,
···
 static ssize_t ana_grpid_show(struct device *dev, struct device_attribute *attr,
 		char *buf)
 {
-	return sprintf(buf, "%d\n", nvme_get_ns_from_dev(dev)->ana_grpid);
+	return sysfs_emit(buf, "%d\n", nvme_get_ns_from_dev(dev)->ana_grpid);
 }
 DEVICE_ATTR_RO(ana_grpid);
 
···
 {
 	struct nvme_ns *ns = nvme_get_ns_from_dev(dev);
 
-	return sprintf(buf, "%s\n", nvme_ana_state_names[ns->ana_state]);
+	return sysfs_emit(buf, "%s\n", nvme_ana_state_names[ns->ana_state]);
 }
 DEVICE_ATTR_RO(ana_state);
 
···
 		if (desc.state) {
 			/* found the group desc: update */
 			nvme_update_ns_ana_state(&desc, ns);
+		} else {
+			/* group desc not found: trigger a re-read */
+			set_bit(NVME_NS_ANA_PENDING, &ns->flags);
+			queue_work(nvme_wq, &ns->ctrl->ana_work);
 		}
 	} else {
 		ns->ana_state = NVME_ANA_OPTIMIZED; 

drivers/nvme/host/nvme.h

···
 	u32 max_hw_sectors;
 	u32 max_segments;
 	u32 max_integrity_segments;
+	u32 max_discard_sectors;
+	u32 max_discard_segments;
+	u32 max_zeroes_sectors;
 #ifdef CONFIG_BLK_DEV_ZONED
 	u32 max_zone_append;
 #endif
···
 void nvme_uninit_ctrl(struct nvme_ctrl *ctrl);
 void nvme_start_ctrl(struct nvme_ctrl *ctrl);
 void nvme_stop_ctrl(struct nvme_ctrl *ctrl);
-int nvme_init_identify(struct nvme_ctrl *ctrl);
+int nvme_init_ctrl_finish(struct nvme_ctrl *ctrl);
 
 void nvme_remove_namespaces(struct nvme_ctrl *ctrl);
 
···
 struct request *nvme_alloc_request(struct request_queue *q,
 		struct nvme_command *cmd, blk_mq_req_flags_t flags);
 void nvme_cleanup_cmd(struct request *req);
-blk_status_t nvme_setup_cmd(struct nvme_ns *ns, struct request *req,
-		struct nvme_command *cmd);
+blk_status_t nvme_setup_cmd(struct nvme_ns *ns, struct request *req);
 int nvme_submit_sync_cmd(struct request_queue *q, struct nvme_command *cmd,
 		void *buf, unsigned bufflen);
 int __nvme_submit_sync_cmd(struct request_queue *q, struct nvme_command *cmd,
···
 static inline int nvme_mpath_init(struct nvme_ctrl *ctrl,
 		struct nvme_id_ctrl *id)
 {
-	if (ctrl->subsys->cmic & (1 << 3))
+	if (ctrl->subsys->cmic & NVME_CTRL_CMIC_ANA)
 		dev_warn(ctrl->device,
 "Please enable CONFIG_NVME_MULTIPATH for full support of multi-port devices.\n");
 	return 0;

drivers/nvme/host/pci.c

···
  */
 struct nvme_iod {
 	struct nvme_request req;
+	struct nvme_command cmd;
 	struct nvme_queue *nvmeq;
 	bool use_sgl;
 	int aborted;
···
 	iod->nvmeq = nvmeq;
 
 	nvme_req(req)->ctrl = &dev->ctrl;
+	nvme_req(req)->cmd = &iod->cmd;
 	return 0;
 }
 
···
 	struct nvme_dev *dev = nvmeq->dev;
 	struct request *req = bd->rq;
 	struct nvme_iod *iod = blk_mq_rq_to_pdu(req);
-	struct nvme_command cmnd;
+	struct nvme_command *cmnd = &iod->cmd;
 	blk_status_t ret;
 
 	iod->aborted = 0;
···
 	if (unlikely(!test_bit(NVMEQ_ENABLED, &nvmeq->flags)))
 		return BLK_STS_IOERR;
 
-	ret = nvme_setup_cmd(ns, req, &cmnd);
+	ret = nvme_setup_cmd(ns, req);
 	if (ret)
 		return ret;
 
 	if (blk_rq_nr_phys_segments(req)) {
-		ret = nvme_map_data(dev, req, &cmnd);
+		ret = nvme_map_data(dev, req, cmnd);
 		if (ret)
 			goto out_free_cmd;
 	}
 
 	if (blk_integrity_rq(req)) {
-		ret = nvme_map_metadata(dev, req, &cmnd);
+		ret = nvme_map_metadata(dev, req, cmnd);
 		if (ret)
 			goto out_unmap_data;
 	}
 
 	blk_mq_start_request(req);
-	nvme_submit_cmd(nvmeq, &cmnd, bd->last);
+	nvme_submit_cmd(nvmeq, cmnd, bd->last);
 	return BLK_STS_OK;
 out_unmap_data:
 	nvme_unmap_data(dev, req);
···
 static irqreturn_t nvme_irq(int irq, void *data)
 {
 	struct nvme_queue *nvmeq = data;
-	irqreturn_t ret = IRQ_NONE;
 
-	/*
-	 * The rmb/wmb pair ensures we see all updates from a previous run of
-	 * the irq handler, even if that was on another CPU.
-	 */
-	rmb();
 	if (nvme_process_cq(nvmeq))
-		ret = IRQ_HANDLED;
-	wmb();
-
-	return ret;
+		return IRQ_HANDLED;
+	return IRQ_NONE;
 }
 
 static irqreturn_t nvme_irq_check(int irq, void *data)
···
 	 */
 	dev->ctrl.max_integrity_segments = 1;
 
-	result = nvme_init_identify(&dev->ctrl);
+	result = nvme_init_ctrl_finish(&dev->ctrl);
 	if (result)
 		goto out;
 

drivers/nvme/host/rdma.c

···
 			NVME_RDMA_DATA_SGL_SIZE;
 
 	req->queue = queue;
+	nvme_req(rq)->cmd = req->sqe.data;
 
 	return 0;
 }
···
 
 	blk_mq_unquiesce_queue(ctrl->ctrl.admin_q);
 
-	error = nvme_init_identify(&ctrl->ctrl);
+	error = nvme_init_ctrl_finish(&ctrl->ctrl);
 	if (error)
 		goto out_quiesce_queue;
 
···
 	struct request *rq = bd->rq;
 	struct nvme_rdma_request *req = blk_mq_rq_to_pdu(rq);
 	struct nvme_rdma_qe *sqe = &req->sqe;
-	struct nvme_command *c = sqe->data;
+	struct nvme_command *c = nvme_req(rq)->cmd;
 	struct ib_device *dev;
 	bool queue_ready = test_bit(NVME_RDMA_Q_LIVE, &queue->flags);
 	blk_status_t ret;
···
 	ib_dma_sync_single_for_cpu(dev, sqe->dma,
 			sizeof(struct nvme_command), DMA_TO_DEVICE);
 
-	ret = nvme_setup_cmd(ns, rq, c);
+	ret = nvme_setup_cmd(ns, rq);
 	if (ret)
 		goto unmap_qe;
 

drivers/nvme/host/tcp.c

···
 {
 	struct nvme_tcp_ctrl *ctrl = set->driver_data;
 	struct nvme_tcp_request *req = blk_mq_rq_to_pdu(rq);
+	struct nvme_tcp_cmd_pdu *pdu;
 	int queue_idx = (set == &ctrl->tag_set) ? hctx_idx + 1 : 0;
 	struct nvme_tcp_queue *queue = &ctrl->queues[queue_idx];
 	u8 hdgst = nvme_tcp_hdgst_len(queue);
···
 	if (!req->pdu)
 		return -ENOMEM;
 
+	pdu = req->pdu;
 	req->queue = queue;
 	nvme_req(rq)->ctrl = &ctrl->ctrl;
+	nvme_req(rq)->cmd = &pdu->cmd;
 
 	return 0;
 }
···
 {
 	struct nvme_tcp_queue *queue;
 
-	read_lock(&sk->sk_callback_lock);
+	read_lock_bh(&sk->sk_callback_lock);
 	queue = sk->sk_user_data;
 	if (!queue)
 		goto done;
···
 
 	queue->state_change(sk);
 done:
-	read_unlock(&sk->sk_callback_lock);
+	read_unlock_bh(&sk->sk_callback_lock);
 }
 
 static inline bool nvme_tcp_queue_more(struct nvme_tcp_queue *queue)
···
 
 	blk_mq_unquiesce_queue(ctrl->admin_q);
 
-	error = nvme_init_identify(ctrl);
+	error = nvme_init_ctrl_finish(ctrl);
 	if (error)
 		goto out_quiesce_queue;
 
···
 
 	if (ctrl->icdoff) {
 		dev_err(ctrl->device, "icdoff is not supported!\n");
+		goto destroy_admin;
+	}
+
+	if (!(ctrl->sgls & ((1 << 0) | (1 << 1)))) {
+		dev_err(ctrl->device, "Mandatory sgls are not supported!\n");
 		goto destroy_admin;
 	}
 
···
 	u8 hdgst = nvme_tcp_hdgst_len(queue), ddgst = 0;
 	blk_status_t ret;
 
-	ret = nvme_setup_cmd(ns, rq, &pdu->cmd);
+	ret = nvme_setup_cmd(ns, rq);
 	if (ret)
 		return ret;
 

drivers/nvme/target/admin-cmd.c

···
 	default:
 		id->nuse = id->nsze;
 		break;
-        }
+	}
 
 	if (req->ns->bdev)
 		nvmet_bdev_set_limits(req->ns->bdev, id);
···
 	if (nvmet_req_subsys(req)->type == NVME_NQN_DISC)
 		return nvmet_parse_discovery_cmd(req);
 
-	ret = nvmet_check_ctrl_status(req, cmd);
+	ret = nvmet_check_ctrl_status(req);
 	if (unlikely(ret))
 		return ret;
 

drivers/nvme/target/configfs.c

···
 	if (!len)
 		return -EINVAL;
 
+	if (len > NVMET_MN_MAX_SIZE) {
+		pr_err("Model nubmer size can not exceed %d Bytes\n",
+		       NVMET_MN_MAX_SIZE);
+		return -EINVAL;
+	}
+
 	for (pos = 0; pos < len; pos++) {
 		if (!nvmet_is_ascii(page[pos]))
 			return -EINVAL;

drivers/nvme/target/core.c

···
 
 static u16 nvmet_parse_io_cmd(struct nvmet_req *req)
 {
-	struct nvme_command *cmd = req->cmd;
 	u16 ret;
 
-	ret = nvmet_check_ctrl_status(req, cmd);
+	ret = nvmet_check_ctrl_status(req);
 	if (unlikely(ret))
 		return ret;
 
···
 	ctrl->cap |= NVMET_QUEUE_SIZE - 1;
 }
 
-u16 nvmet_ctrl_find_get(const char *subsysnqn, const char *hostnqn, u16 cntlid,
-		struct nvmet_req *req, struct nvmet_ctrl **ret)
+struct nvmet_ctrl *nvmet_ctrl_find_get(const char *subsysnqn,
+				       const char *hostnqn, u16 cntlid,
+				       struct nvmet_req *req)
 {
+	struct nvmet_ctrl *ctrl = NULL;
 	struct nvmet_subsys *subsys;
-	struct nvmet_ctrl *ctrl;
-	u16 status = 0;
 
 	subsys = nvmet_find_get_subsys(req->port, subsysnqn);
 	if (!subsys) {
 		pr_warn("connect request for invalid subsystem %s!\n",
 			subsysnqn);
 		req->cqe->result.u32 = IPO_IATTR_CONNECT_DATA(subsysnqn);
-		return NVME_SC_CONNECT_INVALID_PARAM | NVME_SC_DNR;
+		goto out;
 	}
 
 	mutex_lock(&subsys->lock);
···
 			if (!kref_get_unless_zero(&ctrl->ref))
 				continue;
 
-			*ret = ctrl;
-			goto out;
+			/* ctrl found */
+			goto found;
 		}
 	}
 
+	ctrl = NULL; /* ctrl not found */
 	pr_warn("could not find controller %d for subsys %s / host %s\n",
 		cntlid, subsysnqn, hostnqn);
 	req->cqe->result.u32 = IPO_IATTR_CONNECT_DATA(cntlid);
-	status = NVME_SC_CONNECT_INVALID_PARAM | NVME_SC_DNR;
 
-out:
+found:
 	mutex_unlock(&subsys->lock);
 	nvmet_subsys_put(subsys);
-	return status;
+out:
+	return ctrl;
 }
 
-u16 nvmet_check_ctrl_status(struct nvmet_req *req, struct nvme_command *cmd)
+u16 nvmet_check_ctrl_status(struct nvmet_req *req)
 {
 	if (unlikely(!(req->sq->ctrl->cc & NVME_CC_ENABLE))) {
 		pr_err("got cmd %d while CC.EN == 0 on qid = %d\n",
-		       cmd->common.opcode, req->sq->qid);
+		       req->cmd->common.opcode, req->sq->qid);
 		return NVME_SC_CMD_SEQ_ERROR | NVME_SC_DNR;
 	}
 
 	if (unlikely(!(req->sq->ctrl->csts & NVME_CSTS_RDY))) {
 		pr_err("got cmd %d while CSTS.RDY == 0 on qid = %d\n",
-		       cmd->common.opcode, req->sq->qid);
+		       req->cmd->common.opcode, req->sq->qid);
 		return NVME_SC_CMD_SEQ_ERROR | NVME_SC_DNR;
 	}
 	return 0;
···
 		pr_warn("connect request for invalid subsystem %s!\n",
 			subsysnqn);
 		req->cqe->result.u32 = IPO_IATTR_CONNECT_DATA(subsysnqn);
+		req->error_loc = offsetof(struct nvme_common_command, dptr);
 		goto out;
 	}
 
-	status = NVME_SC_CONNECT_INVALID_PARAM | NVME_SC_DNR;
 	down_read(&nvmet_config_sem);
 	if (!nvmet_host_allowed(subsys, hostnqn)) {
 		pr_info("connect by host %s for subsystem %s not allowed\n",
···
 		req->cqe->result.u32 = IPO_IATTR_CONNECT_DATA(hostnqn);
 		up_read(&nvmet_config_sem);
 		status = NVME_SC_CONNECT_INVALID_HOST | NVME_SC_DNR;
+		req->error_loc = offsetof(struct nvme_common_command, dptr);
 		goto out_put_subsystem;
 	}
 	up_read(&nvmet_config_sem);

drivers/nvme/target/discovery.c

···
 	if (req->cmd->get_log_page.lid != NVME_LOG_DISC) {
 		req->error_loc =
 			offsetof(struct nvme_get_log_page_command, lid);
-		status = NVME_SC_INVALID_OPCODE | NVME_SC_DNR;
+		status = NVME_SC_INVALID_FIELD | NVME_SC_DNR;
 		goto out;
 	}
 
 	/* Spec requires dword aligned offsets */
 	if (offset & 0x3) {
+		req->error_loc =
+			offsetof(struct nvme_get_log_page_command, lpo);
 		status = NVME_SC_INVALID_FIELD | NVME_SC_DNR;
 		goto out;
 	}
···
 
 	if (req->cmd->identify.cns != NVME_ID_CNS_CTRL) {
 		req->error_loc = offsetof(struct nvme_identify, cns);
-		status = NVME_SC_INVALID_OPCODE | NVME_SC_DNR;
+		status = NVME_SC_INVALID_FIELD | NVME_SC_DNR;
 		goto out;
 	}
 

drivers/nvme/target/fabrics-cmd.c

···
 
 	status = nvmet_alloc_ctrl(d->subsysnqn, d->hostnqn, req,
 				  le32_to_cpu(c->kato), &ctrl);
-	if (status) {
-		if (status == (NVME_SC_INVALID_FIELD | NVME_SC_DNR))
-			req->error_loc =
-				offsetof(struct nvme_common_command, opcode);
+	if (status)
 		goto out;
-	}
 
 	ctrl->pi_support = ctrl->port->pi_enable && ctrl->subsys->pi_support;
 
···
 {
 	struct nvmf_connect_command *c = &req->cmd->connect;
 	struct nvmf_connect_data *d;
-	struct nvmet_ctrl *ctrl = NULL;
+	struct nvmet_ctrl *ctrl;
 	u16 qid = le16_to_cpu(c->qid);
 	u16 status = 0;
 
···
 		goto out;
 	}
 
-	status = nvmet_ctrl_find_get(d->subsysnqn, d->hostnqn,
-				     le16_to_cpu(d->cntlid),
-				     req, &ctrl);
-	if (status)
+	ctrl = nvmet_ctrl_find_get(d->subsysnqn, d->hostnqn,
+				   le16_to_cpu(d->cntlid), req);
+	if (!ctrl) {
+		status = NVME_SC_CONNECT_INVALID_PARAM | NVME_SC_DNR;
 		goto out;
+	}
 
 	if (unlikely(qid > ctrl->subsys->max_qid)) {
 		pr_warn("invalid queue id (%d)\n", qid);

drivers/nvme/target/fc.c

···
  *
  * @target_port: pointer to the (registered) target port the LS was
  *              received on.
+ * @hosthandle: pointer to the host specific data, gets stored in iod.
  * @lsrsp:      pointer to a lsrsp structure to be used to reference
  *              the exchange corresponding to the LS.
  * @lsreqbuf:   pointer to the buffer containing the LS Request

drivers/nvme/target/loop.c

···
 	if (!nvmf_check_ready(&queue->ctrl->ctrl, req, queue_ready))
 		return nvmf_fail_nonready_command(&queue->ctrl->ctrl, req);
 
-	ret = nvme_setup_cmd(ns, req, &iod->cmd);
+	ret = nvme_setup_cmd(ns, req);
 	if (ret)
 		return ret;
 
···
 		unsigned int numa_node)
 {
 	struct nvme_loop_ctrl *ctrl = set->driver_data;
+	struct nvme_loop_iod *iod = blk_mq_rq_to_pdu(req);
 
 	nvme_req(req)->ctrl = &ctrl->ctrl;
+	nvme_req(req)->cmd = &iod->cmd;
 	return nvme_loop_init_iod(ctrl, blk_mq_rq_to_pdu(req),
 			(set == &ctrl->tag_set) ? hctx_idx + 1 : 0);
 }
···
 
 	blk_mq_unquiesce_queue(ctrl->ctrl.admin_q);
 
-	error = nvme_init_identify(&ctrl->ctrl);
+	error = nvme_init_ctrl_finish(&ctrl->ctrl);
 	if (error)
 		goto out_cleanup_queue;
 

drivers/nvme/target/nvmet.h

···
 #define NVMET_ERROR_LOG_SLOTS		128
 #define NVMET_NO_ERROR_LOC		((u16)-1)
 #define NVMET_DEFAULT_CTRL_MODEL	"Linux"
+#define NVMET_MN_MAX_SIZE		40
 
 /*
  * Supported optional AENs:
···
 void nvmet_update_cc(struct nvmet_ctrl *ctrl, u32 new);
 u16 nvmet_alloc_ctrl(const char *subsysnqn, const char *hostnqn,
 		struct nvmet_req *req, u32 kato, struct nvmet_ctrl **ctrlp);
-u16 nvmet_ctrl_find_get(const char *subsysnqn, const char *hostnqn, u16 cntlid,
-		struct nvmet_req *req, struct nvmet_ctrl **ret);
+struct nvmet_ctrl *nvmet_ctrl_find_get(const char *subsysnqn,
+				       const char *hostnqn, u16 cntlid,
+				       struct nvmet_req *req);
 void nvmet_ctrl_put(struct nvmet_ctrl *ctrl);
-u16 nvmet_check_ctrl_status(struct nvmet_req *req, struct nvme_command *cmd);
+u16 nvmet_check_ctrl_status(struct nvmet_req *req);
 
 struct nvmet_subsys *nvmet_subsys_alloc(const char *subsysnqn,
 		enum nvme_subsys_type type);

drivers/nvme/target/tcp.c

···
 module_param(so_priority, int, 0644);
 MODULE_PARM_DESC(so_priority, "nvmet tcp socket optimize priority");
 
+/* Define a time period (in usecs) that io_work() shall sample an activated
+ * queue before determining it to be idle.  This optional module behavior
+ * can enable NIC solutions that support socket optimized packet processing
+ * using advanced interrupt moderation techniques.
+ */
+static int idle_poll_period_usecs;
+module_param(idle_poll_period_usecs, int, 0644);
+MODULE_PARM_DESC(idle_poll_period_usecs,
+		"nvmet tcp io_work poll till idle time period in usecs");
+
 #define NVMET_TCP_RECV_BUDGET		8
 #define NVMET_TCP_SEND_BUDGET		8
 #define NVMET_TCP_IO_WORK_BUDGET	64
···
 	bool			data_digest;
 	struct ahash_request	*snd_hash;
 	struct ahash_request	*rcv_hash;
+
+	unsigned long           poll_end;
 
 	spinlock_t		state_lock;
 	enum nvmet_tcp_queue_state state;
···
 	spin_unlock(&queue->state_lock);
 }
 
+static inline void nvmet_tcp_arm_queue_deadline(struct nvmet_tcp_queue *queue)
+{
+	queue->poll_end = jiffies + usecs_to_jiffies(idle_poll_period_usecs);
+}
+
+static bool nvmet_tcp_check_queue_deadline(struct nvmet_tcp_queue *queue,
+		int ops)
+{
+	if (!idle_poll_period_usecs)
+		return false;
+
+	if (ops)
+		nvmet_tcp_arm_queue_deadline(queue);
+
+	return !time_after(jiffies, queue->poll_end);
+}
+
 static void nvmet_tcp_io_work(struct work_struct *w)
 {
 	struct nvmet_tcp_queue *queue =
···
 	} while (pending && ops < NVMET_TCP_IO_WORK_BUDGET);
 
 	/*
-	 * We exahusted our budget, requeue our selves
+	 * Requeue the worker if idle deadline period is in progress or any
+	 * ops activity was recorded during the do-while loop above.
 	 */
-	if (pending)
+	if (nvmet_tcp_check_queue_deadline(queue, ops) || pending)
 		queue_work_on(queue_cpu(queue), nvmet_tcp_wq, &queue->io_work);
 }
 
···
 {
 	struct nvmet_tcp_queue *queue;
 
-	write_lock_bh(&sk->sk_callback_lock);
+	read_lock_bh(&sk->sk_callback_lock);
 	queue = sk->sk_user_data;
 	if (!queue)
 		goto done;
···
 			queue->idx, sk->sk_state);
 	}
 done:
-	write_unlock_bh(&sk->sk_callback_lock);
+	read_unlock_bh(&sk->sk_callback_lock);
 }
 
 static int nvmet_tcp_set_queue_sock(struct nvmet_tcp_queue *queue)
···
 		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;
+		if (idle_poll_period_usecs)
+			nvmet_tcp_arm_queue_deadline(queue);
 		queue_work_on(queue_cpu(queue), nvmet_tcp_wq, &queue->io_work);
 	}
 	write_unlock_bh(&sock->sk->sk_callback_lock);

include/linux/nvme.h

···
 	__u8	rsvd1[4095];
 };
 
+struct nvme_id_ctrl_nvm {
+	__u8	vsl;
+	__u8	wzsl;
+	__u8	wusl;
+	__u8	dmrl;
+	__le32	dmrsl;
+	__le64	dmsl;
+	__u8	rsvd16[4080];
+};
+
 enum {
 	NVME_ID_CNS_NS			= 0x00,
 	NVME_ID_CNS_CTRL		= 0x01,