fc.c 92.0 KB
Newer Older
1
// SPDX-License-Identifier: GPL-2.0
2 3 4 5 6 7 8 9
/*
 * Copyright (c) 2016 Avago Technologies.  All rights reserved.
 */
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
#include <linux/module.h>
#include <linux/parser.h>
#include <uapi/scsi/fc/fc_fs.h>
#include <uapi/scsi/fc/fc_els.h>
10
#include <linux/delay.h>
11
#include <linux/overflow.h>
12 13 14 15 16

#include "nvme.h"
#include "fabrics.h"
#include <linux/nvme-fc-driver.h>
#include <linux/nvme-fc.h>
17
#include <scsi/scsi_transport_fc.h>
18 19 20 21 22

/* *************************** Data Structures/Defines ****************** */


enum nvme_fc_queue_flags {
23 24
	NVME_FC_Q_CONNECTED = 0,
	NVME_FC_Q_LIVE,
25 26
};

27 28
#define NVME_FC_DEFAULT_DEV_LOSS_TMO	60	/* seconds */

29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44
struct nvme_fc_queue {
	struct nvme_fc_ctrl	*ctrl;
	struct device		*dev;
	struct blk_mq_hw_ctx	*hctx;
	void			*lldd_handle;
	size_t			cmnd_capsule_len;
	u32			qnum;
	u32			rqcnt;
	u32			seqno;

	u64			connection_id;
	atomic_t		csn;

	unsigned long		flags;
} __aligned(sizeof(u64));	/* alignment for other things alloc'd with */

45 46
enum nvme_fcop_flags {
	FCOP_FLAGS_TERMIO	= (1 << 0),
J
James Smart 已提交
47
	FCOP_FLAGS_AEN		= (1 << 1),
48 49
};

50 51 52
struct nvmefc_ls_req_op {
	struct nvmefc_ls_req	ls_req;

J
James Smart 已提交
53
	struct nvme_fc_rport	*rport;
54 55
	struct nvme_fc_queue	*queue;
	struct request		*rq;
56
	u32			flags;
57 58 59

	int			ls_error;
	struct completion	ls_done;
J
James Smart 已提交
60
	struct list_head	lsreq_list;	/* rport->ls_req_list */
61 62 63 64 65 66 67 68
	bool			req_queued;
};

enum nvme_fcpop_state {
	FCPOP_STATE_UNINIT	= 0,
	FCPOP_STATE_IDLE	= 1,
	FCPOP_STATE_ACTIVE	= 2,
	FCPOP_STATE_ABORTED	= 3,
J
James Smart 已提交
69
	FCPOP_STATE_COMPLETE	= 4,
70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87
};

struct nvme_fc_fcp_op {
	struct nvme_request	nreq;		/*
						 * nvme/host/core.c
						 * requires this to be
						 * the 1st element in the
						 * private structure
						 * associated with the
						 * request.
						 */
	struct nvmefc_fcp_req	fcp_req;

	struct nvme_fc_ctrl	*ctrl;
	struct nvme_fc_queue	*queue;
	struct request		*rq;

	atomic_t		state;
J
James Smart 已提交
88
	u32			flags;
89 90 91 92 93 94 95
	u32			rqno;
	u32			nents;

	struct nvme_fc_cmd_iu	cmd_iu;
	struct nvme_fc_ersp_iu	rsp_iu;
};

96 97 98 99 100 101
struct nvme_fcp_op_w_sgl {
	struct nvme_fc_fcp_op	op;
	struct scatterlist	sgl[SG_CHUNK_SIZE];
	uint8_t			priv[0];
};

102 103 104 105 106 107 108 109 110
struct nvme_fc_lport {
	struct nvme_fc_local_port	localport;

	struct ida			endp_cnt;
	struct list_head		port_list;	/* nvme_fc_port_list */
	struct list_head		endp_list;
	struct device			*dev;	/* physical device for dma */
	struct nvme_fc_port_template	*ops;
	struct kref			ref;
111
	atomic_t                        act_rport_cnt;
112 113 114 115 116 117 118
} __aligned(sizeof(u64));	/* alignment for other things alloc'd with */

struct nvme_fc_rport {
	struct nvme_fc_remote_port	remoteport;

	struct list_head		endp_list; /* for lport->endp_list */
	struct list_head		ctrl_list;
J
James Smart 已提交
119
	struct list_head		ls_req_list;
120
	struct list_head		disc_list;
J
James Smart 已提交
121 122
	struct device			*dev;	/* physical device for dma */
	struct nvme_fc_lport		*lport;
123 124
	spinlock_t			lock;
	struct kref			ref;
125
	atomic_t                        act_ctrl_cnt;
126
	unsigned long			dev_loss_end;
127 128
} __aligned(sizeof(u64));	/* alignment for other things alloc'd with */

129 130
enum nvme_fcctrl_flags {
	FCCTRL_TERMIO		= (1 << 0),
131 132 133 134 135 136 137 138 139 140
};

struct nvme_fc_ctrl {
	spinlock_t		lock;
	struct nvme_fc_queue	*queues;
	struct device		*dev;
	struct nvme_fc_lport	*lport;
	struct nvme_fc_rport	*rport;
	u32			cnum;

141
	bool			ioq_live;
142
	bool			assoc_active;
143
	atomic_t		err_work_active;
144 145 146 147 148 149 150
	u64			association_id;

	struct list_head	ctrl_list;	/* rport->ctrl_list */

	struct blk_mq_tag_set	admin_tag_set;
	struct blk_mq_tag_set	tag_set;

151
	struct delayed_work	connect_work;
152
	struct work_struct	err_work;
153

154
	struct kref		ref;
155 156
	u32			flags;
	u32			iocnt;
157
	wait_queue_head_t	ioabort_wait;
158

K
Keith Busch 已提交
159
	struct nvme_fc_fcp_op	aen_ops[NVME_NR_AEN_COMMANDS];
160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204

	struct nvme_ctrl	ctrl;
};

static inline struct nvme_fc_ctrl *
to_fc_ctrl(struct nvme_ctrl *ctrl)
{
	return container_of(ctrl, struct nvme_fc_ctrl, ctrl);
}

static inline struct nvme_fc_lport *
localport_to_lport(struct nvme_fc_local_port *portptr)
{
	return container_of(portptr, struct nvme_fc_lport, localport);
}

static inline struct nvme_fc_rport *
remoteport_to_rport(struct nvme_fc_remote_port *portptr)
{
	return container_of(portptr, struct nvme_fc_rport, remoteport);
}

static inline struct nvmefc_ls_req_op *
ls_req_to_lsop(struct nvmefc_ls_req *lsreq)
{
	return container_of(lsreq, struct nvmefc_ls_req_op, ls_req);
}

static inline struct nvme_fc_fcp_op *
fcp_req_to_fcp_op(struct nvmefc_fcp_req *fcpreq)
{
	return container_of(fcpreq, struct nvme_fc_fcp_op, fcp_req);
}



/* *************************** Globals **************************** */


static DEFINE_SPINLOCK(nvme_fc_lock);

static LIST_HEAD(nvme_fc_lport_list);
static DEFINE_IDA(nvme_fc_local_port_cnt);
static DEFINE_IDA(nvme_fc_ctrl_cnt);

205
static struct workqueue_struct *nvme_fc_wq;
206

207 208 209 210 211 212
/*
 * These items are short-term. They will eventually be moved into
 * a generic FC class. See comments in module init.
 */
static struct device *fc_udev_device;

213 214 215 216 217 218

/* *********************** FC-NVME Port Management ************************ */

static void __nvme_fc_delete_hw_queue(struct nvme_fc_ctrl *,
			struct nvme_fc_queue *, unsigned int);

219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255
static void
nvme_fc_free_lport(struct kref *ref)
{
	struct nvme_fc_lport *lport =
		container_of(ref, struct nvme_fc_lport, ref);
	unsigned long flags;

	WARN_ON(lport->localport.port_state != FC_OBJSTATE_DELETED);
	WARN_ON(!list_empty(&lport->endp_list));

	/* remove from transport list */
	spin_lock_irqsave(&nvme_fc_lock, flags);
	list_del(&lport->port_list);
	spin_unlock_irqrestore(&nvme_fc_lock, flags);

	ida_simple_remove(&nvme_fc_local_port_cnt, lport->localport.port_num);
	ida_destroy(&lport->endp_cnt);

	put_device(lport->dev);

	kfree(lport);
}

static void
nvme_fc_lport_put(struct nvme_fc_lport *lport)
{
	kref_put(&lport->ref, nvme_fc_free_lport);
}

static int
nvme_fc_lport_get(struct nvme_fc_lport *lport)
{
	return kref_get_unless_zero(&lport->ref);
}


static struct nvme_fc_lport *
256 257 258
nvme_fc_attach_to_unreg_lport(struct nvme_fc_port_info *pinfo,
			struct nvme_fc_port_template *ops,
			struct device *dev)
259 260 261 262 263 264 265 266 267 268 269
{
	struct nvme_fc_lport *lport;
	unsigned long flags;

	spin_lock_irqsave(&nvme_fc_lock, flags);

	list_for_each_entry(lport, &nvme_fc_lport_list, port_list) {
		if (lport->localport.node_name != pinfo->node_name ||
		    lport->localport.port_name != pinfo->port_name)
			continue;

270 271 272 273 274
		if (lport->dev != dev) {
			lport = ERR_PTR(-EXDEV);
			goto out_done;
		}

275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290
		if (lport->localport.port_state != FC_OBJSTATE_DELETED) {
			lport = ERR_PTR(-EEXIST);
			goto out_done;
		}

		if (!nvme_fc_lport_get(lport)) {
			/*
			 * fails if ref cnt already 0. If so,
			 * act as if lport already deleted
			 */
			lport = NULL;
			goto out_done;
		}

		/* resume the lport */

291
		lport->ops = ops;
292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307
		lport->localport.port_role = pinfo->port_role;
		lport->localport.port_id = pinfo->port_id;
		lport->localport.port_state = FC_OBJSTATE_ONLINE;

		spin_unlock_irqrestore(&nvme_fc_lock, flags);

		return lport;
	}

	lport = NULL;

out_done:
	spin_unlock_irqrestore(&nvme_fc_lock, flags);

	return lport;
}
308 309 310 311 312 313 314 315 316

/**
 * nvme_fc_register_localport - transport entry point called by an
 *                              LLDD to register the existence of a NVME
 *                              host FC port.
 * @pinfo:     pointer to information about the port to be registered
 * @template:  LLDD entrypoints and operational parameters for the port
 * @dev:       physical hardware device node port corresponds to. Will be
 *             used for DMA mappings
B
Bart Van Assche 已提交
317
 * @portptr:   pointer to a local port pointer. Upon success, the routine
318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344
 *             will allocate a nvme_fc_local_port structure and place its
 *             address in the local port pointer. Upon failure, local port
 *             pointer will be set to 0.
 *
 * Returns:
 * a completion status. Must be 0 upon success; a negative errno
 * (ex: -ENXIO) upon failure.
 */
int
nvme_fc_register_localport(struct nvme_fc_port_info *pinfo,
			struct nvme_fc_port_template *template,
			struct device *dev,
			struct nvme_fc_local_port **portptr)
{
	struct nvme_fc_lport *newrec;
	unsigned long flags;
	int ret, idx;

	if (!template->localport_delete || !template->remoteport_delete ||
	    !template->ls_req || !template->fcp_io ||
	    !template->ls_abort || !template->fcp_abort ||
	    !template->max_hw_queues || !template->max_sgl_segments ||
	    !template->max_dif_sgl_segments || !template->dma_boundary) {
		ret = -EINVAL;
		goto out_reghost_failed;
	}

345 346 347 348 349 350 351
	/*
	 * look to see if there is already a localport that had been
	 * deregistered and in the process of waiting for all the
	 * references to fully be removed.  If the references haven't
	 * expired, we can simply re-enable the localport. Remoteports
	 * and controller reconnections should resume naturally.
	 */
352
	newrec = nvme_fc_attach_to_unreg_lport(pinfo, template, dev);
353 354 355 356 357 358 359 360 361 362 363 364 365 366

	/* found an lport, but something about its state is bad */
	if (IS_ERR(newrec)) {
		ret = PTR_ERR(newrec);
		goto out_reghost_failed;

	/* found existing lport, which was resumed */
	} else if (newrec) {
		*portptr = &newrec->localport;
		return 0;
	}

	/* nothing found - allocate a new localport struct */

367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387
	newrec = kmalloc((sizeof(*newrec) + template->local_priv_sz),
			 GFP_KERNEL);
	if (!newrec) {
		ret = -ENOMEM;
		goto out_reghost_failed;
	}

	idx = ida_simple_get(&nvme_fc_local_port_cnt, 0, 0, GFP_KERNEL);
	if (idx < 0) {
		ret = -ENOSPC;
		goto out_fail_kfree;
	}

	if (!get_device(dev) && dev) {
		ret = -ENODEV;
		goto out_ida_put;
	}

	INIT_LIST_HEAD(&newrec->port_list);
	INIT_LIST_HEAD(&newrec->endp_list);
	kref_init(&newrec->ref);
388
	atomic_set(&newrec->act_rport_cnt, 0);
389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424
	newrec->ops = template;
	newrec->dev = dev;
	ida_init(&newrec->endp_cnt);
	newrec->localport.private = &newrec[1];
	newrec->localport.node_name = pinfo->node_name;
	newrec->localport.port_name = pinfo->port_name;
	newrec->localport.port_role = pinfo->port_role;
	newrec->localport.port_id = pinfo->port_id;
	newrec->localport.port_state = FC_OBJSTATE_ONLINE;
	newrec->localport.port_num = idx;

	spin_lock_irqsave(&nvme_fc_lock, flags);
	list_add_tail(&newrec->port_list, &nvme_fc_lport_list);
	spin_unlock_irqrestore(&nvme_fc_lock, flags);

	if (dev)
		dma_set_seg_boundary(dev, template->dma_boundary);

	*portptr = &newrec->localport;
	return 0;

out_ida_put:
	ida_simple_remove(&nvme_fc_local_port_cnt, idx);
out_fail_kfree:
	kfree(newrec);
out_reghost_failed:
	*portptr = NULL;

	return ret;
}
EXPORT_SYMBOL_GPL(nvme_fc_register_localport);

/**
 * nvme_fc_unregister_localport - transport entry point called by an
 *                              LLDD to deregister/remove a previously
 *                              registered a NVME host FC port.
B
Bart Van Assche 已提交
425
 * @portptr: pointer to the (registered) local port that is to be deregistered.
426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449
 *
 * Returns:
 * a completion status. Must be 0 upon success; a negative errno
 * (ex: -ENXIO) upon failure.
 */
int
nvme_fc_unregister_localport(struct nvme_fc_local_port *portptr)
{
	struct nvme_fc_lport *lport = localport_to_lport(portptr);
	unsigned long flags;

	if (!portptr)
		return -EINVAL;

	spin_lock_irqsave(&nvme_fc_lock, flags);

	if (portptr->port_state != FC_OBJSTATE_ONLINE) {
		spin_unlock_irqrestore(&nvme_fc_lock, flags);
		return -EINVAL;
	}
	portptr->port_state = FC_OBJSTATE_DELETED;

	spin_unlock_irqrestore(&nvme_fc_lock, flags);

450 451 452
	if (atomic_read(&lport->act_rport_cnt) == 0)
		lport->ops->localport_delete(&lport->localport);

453 454 455 456 457 458
	nvme_fc_lport_put(lport);

	return 0;
}
EXPORT_SYMBOL_GPL(nvme_fc_unregister_localport);

J
James Smart 已提交
459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488
/*
 * TRADDR strings, per FC-NVME are fixed format:
 *   "nn-0x<16hexdigits>:pn-0x<16hexdigits>" - 43 characters
 * udev event will only differ by prefix of what field is
 * being specified:
 *    "NVMEFC_HOST_TRADDR=" or "NVMEFC_TRADDR=" - 19 max characters
 *  19 + 43 + null_fudge = 64 characters
 */
#define FCNVME_TRADDR_LENGTH		64

static void
nvme_fc_signal_discovery_scan(struct nvme_fc_lport *lport,
		struct nvme_fc_rport *rport)
{
	char hostaddr[FCNVME_TRADDR_LENGTH];	/* NVMEFC_HOST_TRADDR=...*/
	char tgtaddr[FCNVME_TRADDR_LENGTH];	/* NVMEFC_TRADDR=...*/
	char *envp[4] = { "FC_EVENT=nvmediscovery", hostaddr, tgtaddr, NULL };

	if (!(rport->remoteport.port_role & FC_PORT_ROLE_NVME_DISCOVERY))
		return;

	snprintf(hostaddr, sizeof(hostaddr),
		"NVMEFC_HOST_TRADDR=nn-0x%016llx:pn-0x%016llx",
		lport->localport.node_name, lport->localport.port_name);
	snprintf(tgtaddr, sizeof(tgtaddr),
		"NVMEFC_TRADDR=nn-0x%016llx:pn-0x%016llx",
		rport->remoteport.node_name, rport->remoteport.port_name);
	kobject_uevent_env(&fc_udev_device->kobj, KOBJ_CHANGE, envp);
}

489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505
static void
nvme_fc_free_rport(struct kref *ref)
{
	struct nvme_fc_rport *rport =
		container_of(ref, struct nvme_fc_rport, ref);
	struct nvme_fc_lport *lport =
			localport_to_lport(rport->remoteport.localport);
	unsigned long flags;

	WARN_ON(rport->remoteport.port_state != FC_OBJSTATE_DELETED);
	WARN_ON(!list_empty(&rport->ctrl_list));

	/* remove from lport list */
	spin_lock_irqsave(&nvme_fc_lock, flags);
	list_del(&rport->endp_list);
	spin_unlock_irqrestore(&nvme_fc_lock, flags);

506
	WARN_ON(!list_empty(&rport->disc_list));
507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525
	ida_simple_remove(&lport->endp_cnt, rport->remoteport.port_num);

	kfree(rport);

	nvme_fc_lport_put(lport);
}

static void
nvme_fc_rport_put(struct nvme_fc_rport *rport)
{
	kref_put(&rport->ref, nvme_fc_free_rport);
}

static int
nvme_fc_rport_get(struct nvme_fc_rport *rport)
{
	return kref_get_unless_zero(&rport->ref);
}

526 527 528 529 530
static void
nvme_fc_resume_controller(struct nvme_fc_ctrl *ctrl)
{
	switch (ctrl->ctrl.state) {
	case NVME_CTRL_NEW:
531
	case NVME_CTRL_CONNECTING:
532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588
		/*
		 * As all reconnects were suppressed, schedule a
		 * connect.
		 */
		dev_info(ctrl->ctrl.device,
			"NVME-FC{%d}: connectivity re-established. "
			"Attempting reconnect\n", ctrl->cnum);

		queue_delayed_work(nvme_wq, &ctrl->connect_work, 0);
		break;

	case NVME_CTRL_RESETTING:
		/*
		 * Controller is already in the process of terminating the
		 * association. No need to do anything further. The reconnect
		 * step will naturally occur after the reset completes.
		 */
		break;

	default:
		/* no action to take - let it delete */
		break;
	}
}

static struct nvme_fc_rport *
nvme_fc_attach_to_suspended_rport(struct nvme_fc_lport *lport,
				struct nvme_fc_port_info *pinfo)
{
	struct nvme_fc_rport *rport;
	struct nvme_fc_ctrl *ctrl;
	unsigned long flags;

	spin_lock_irqsave(&nvme_fc_lock, flags);

	list_for_each_entry(rport, &lport->endp_list, endp_list) {
		if (rport->remoteport.node_name != pinfo->node_name ||
		    rport->remoteport.port_name != pinfo->port_name)
			continue;

		if (!nvme_fc_rport_get(rport)) {
			rport = ERR_PTR(-ENOLCK);
			goto out_done;
		}

		spin_unlock_irqrestore(&nvme_fc_lock, flags);

		spin_lock_irqsave(&rport->lock, flags);

		/* has it been unregistered */
		if (rport->remoteport.port_state != FC_OBJSTATE_DELETED) {
			/* means lldd called us twice */
			spin_unlock_irqrestore(&rport->lock, flags);
			nvme_fc_rport_put(rport);
			return ERR_PTR(-ESTALE);
		}

589 590
		rport->remoteport.port_role = pinfo->port_role;
		rport->remoteport.port_id = pinfo->port_id;
591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623
		rport->remoteport.port_state = FC_OBJSTATE_ONLINE;
		rport->dev_loss_end = 0;

		/*
		 * kick off a reconnect attempt on all associations to the
		 * remote port. A successful reconnects will resume i/o.
		 */
		list_for_each_entry(ctrl, &rport->ctrl_list, ctrl_list)
			nvme_fc_resume_controller(ctrl);

		spin_unlock_irqrestore(&rport->lock, flags);

		return rport;
	}

	rport = NULL;

out_done:
	spin_unlock_irqrestore(&nvme_fc_lock, flags);

	return rport;
}

static inline void
__nvme_fc_set_dev_loss_tmo(struct nvme_fc_rport *rport,
			struct nvme_fc_port_info *pinfo)
{
	if (pinfo->dev_loss_tmo)
		rport->remoteport.dev_loss_tmo = pinfo->dev_loss_tmo;
	else
		rport->remoteport.dev_loss_tmo = NVME_FC_DEFAULT_DEV_LOSS_TMO;
}

624 625 626 627 628 629 630
/**
 * nvme_fc_register_remoteport - transport entry point called by an
 *                              LLDD to register the existence of a NVME
 *                              subsystem FC port on its fabric.
 * @localport: pointer to the (registered) local port that the remote
 *             subsystem port is connected to.
 * @pinfo:     pointer to information about the port to be registered
B
Bart Van Assche 已提交
631
 * @portptr:   pointer to a remote port pointer. Upon success, the routine
632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649
 *             will allocate a nvme_fc_remote_port structure and place its
 *             address in the remote port pointer. Upon failure, remote port
 *             pointer will be set to 0.
 *
 * Returns:
 * a completion status. Must be 0 upon success; a negative errno
 * (ex: -ENXIO) upon failure.
 */
int
nvme_fc_register_remoteport(struct nvme_fc_local_port *localport,
				struct nvme_fc_port_info *pinfo,
				struct nvme_fc_remote_port **portptr)
{
	struct nvme_fc_lport *lport = localport_to_lport(localport);
	struct nvme_fc_rport *newrec;
	unsigned long flags;
	int ret, idx;

650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677
	if (!nvme_fc_lport_get(lport)) {
		ret = -ESHUTDOWN;
		goto out_reghost_failed;
	}

	/*
	 * look to see if there is already a remoteport that is waiting
	 * for a reconnect (within dev_loss_tmo) with the same WWN's.
	 * If so, transition to it and reconnect.
	 */
	newrec = nvme_fc_attach_to_suspended_rport(lport, pinfo);

	/* found an rport, but something about its state is bad */
	if (IS_ERR(newrec)) {
		ret = PTR_ERR(newrec);
		goto out_lport_put;

	/* found existing rport, which was resumed */
	} else if (newrec) {
		nvme_fc_lport_put(lport);
		__nvme_fc_set_dev_loss_tmo(newrec, pinfo);
		nvme_fc_signal_discovery_scan(lport, newrec);
		*portptr = &newrec->remoteport;
		return 0;
	}

	/* nothing found - allocate a new remoteport struct */

678 679 680 681
	newrec = kmalloc((sizeof(*newrec) + lport->ops->remote_priv_sz),
			 GFP_KERNEL);
	if (!newrec) {
		ret = -ENOMEM;
682
		goto out_lport_put;
683 684 685 686 687
	}

	idx = ida_simple_get(&lport->endp_cnt, 0, 0, GFP_KERNEL);
	if (idx < 0) {
		ret = -ENOSPC;
688
		goto out_kfree_rport;
689 690 691 692
	}

	INIT_LIST_HEAD(&newrec->endp_list);
	INIT_LIST_HEAD(&newrec->ctrl_list);
J
James Smart 已提交
693
	INIT_LIST_HEAD(&newrec->ls_req_list);
694
	INIT_LIST_HEAD(&newrec->disc_list);
695
	kref_init(&newrec->ref);
696
	atomic_set(&newrec->act_ctrl_cnt, 0);
697 698
	spin_lock_init(&newrec->lock);
	newrec->remoteport.localport = &lport->localport;
J
James Smart 已提交
699 700
	newrec->dev = lport->dev;
	newrec->lport = lport;
701 702 703 704 705 706 707
	newrec->remoteport.private = &newrec[1];
	newrec->remoteport.port_role = pinfo->port_role;
	newrec->remoteport.node_name = pinfo->node_name;
	newrec->remoteport.port_name = pinfo->port_name;
	newrec->remoteport.port_id = pinfo->port_id;
	newrec->remoteport.port_state = FC_OBJSTATE_ONLINE;
	newrec->remoteport.port_num = idx;
708
	__nvme_fc_set_dev_loss_tmo(newrec, pinfo);
709 710 711 712 713

	spin_lock_irqsave(&nvme_fc_lock, flags);
	list_add_tail(&newrec->endp_list, &lport->endp_list);
	spin_unlock_irqrestore(&nvme_fc_lock, flags);

J
James Smart 已提交
714 715
	nvme_fc_signal_discovery_scan(lport, newrec);

716 717 718 719 720
	*portptr = &newrec->remoteport;
	return 0;

out_kfree_rport:
	kfree(newrec);
721 722
out_lport_put:
	nvme_fc_lport_put(lport);
723 724 725 726 727 728
out_reghost_failed:
	*portptr = NULL;
	return ret;
}
EXPORT_SYMBOL_GPL(nvme_fc_register_remoteport);

729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752
static int
nvme_fc_abort_lsops(struct nvme_fc_rport *rport)
{
	struct nvmefc_ls_req_op *lsop;
	unsigned long flags;

restart:
	spin_lock_irqsave(&rport->lock, flags);

	list_for_each_entry(lsop, &rport->ls_req_list, lsreq_list) {
		if (!(lsop->flags & FCOP_FLAGS_TERMIO)) {
			lsop->flags |= FCOP_FLAGS_TERMIO;
			spin_unlock_irqrestore(&rport->lock, flags);
			rport->lport->ops->ls_abort(&rport->lport->localport,
						&rport->remoteport,
						&lsop->ls_req);
			goto restart;
		}
	}
	spin_unlock_irqrestore(&rport->lock, flags);

	return 0;
}

753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771
static void
nvme_fc_ctrl_connectivity_loss(struct nvme_fc_ctrl *ctrl)
{
	dev_info(ctrl->ctrl.device,
		"NVME-FC{%d}: controller connectivity lost. Awaiting "
		"Reconnect", ctrl->cnum);

	switch (ctrl->ctrl.state) {
	case NVME_CTRL_NEW:
	case NVME_CTRL_LIVE:
		/*
		 * Schedule a controller reset. The reset will terminate the
		 * association and schedule the reconnect timer.  Reconnects
		 * will be attempted until either the ctlr_loss_tmo
		 * (max_retries * connect_delay) expires or the remoteport's
		 * dev_loss_tmo expires.
		 */
		if (nvme_reset_ctrl(&ctrl->ctrl)) {
			dev_warn(ctrl->ctrl.device,
772
				"NVME-FC{%d}: Couldn't schedule reset.\n",
773 774 775 776 777
				ctrl->cnum);
			nvme_delete_ctrl(&ctrl->ctrl);
		}
		break;

778
	case NVME_CTRL_CONNECTING:
779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803
		/*
		 * The association has already been terminated and the
		 * controller is attempting reconnects.  No need to do anything
		 * futher.  Reconnects will be attempted until either the
		 * ctlr_loss_tmo (max_retries * connect_delay) expires or the
		 * remoteport's dev_loss_tmo expires.
		 */
		break;

	case NVME_CTRL_RESETTING:
		/*
		 * Controller is already in the process of terminating the
		 * association.  No need to do anything further. The reconnect
		 * step will kick in naturally after the association is
		 * terminated.
		 */
		break;

	case NVME_CTRL_DELETING:
	default:
		/* no action to take - let it delete */
		break;
	}
}

804 805 806 807
/**
 * nvme_fc_unregister_remoteport - transport entry point called by an
 *                              LLDD to deregister/remove a previously
 *                              registered a NVME subsystem FC port.
B
Bart Van Assche 已提交
808 809
 * @portptr: pointer to the (registered) remote port that is to be
 *           deregistered.
810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832
 *
 * Returns:
 * a completion status. Must be 0 upon success; a negative errno
 * (ex: -ENXIO) upon failure.
 */
int
nvme_fc_unregister_remoteport(struct nvme_fc_remote_port *portptr)
{
	struct nvme_fc_rport *rport = remoteport_to_rport(portptr);
	struct nvme_fc_ctrl *ctrl;
	unsigned long flags;

	if (!portptr)
		return -EINVAL;

	spin_lock_irqsave(&rport->lock, flags);

	if (portptr->port_state != FC_OBJSTATE_ONLINE) {
		spin_unlock_irqrestore(&rport->lock, flags);
		return -EINVAL;
	}
	portptr->port_state = FC_OBJSTATE_DELETED;

833 834 835 836 837 838
	rport->dev_loss_end = jiffies + (portptr->dev_loss_tmo * HZ);

	list_for_each_entry(ctrl, &rport->ctrl_list, ctrl_list) {
		/* if dev_loss_tmo==0, dev loss is immediate */
		if (!portptr->dev_loss_tmo) {
			dev_warn(ctrl->ctrl.device,
839
				"NVME-FC{%d}: controller connectivity lost.\n",
840 841 842 843 844
				ctrl->cnum);
			nvme_delete_ctrl(&ctrl->ctrl);
		} else
			nvme_fc_ctrl_connectivity_loss(ctrl);
	}
845 846 847

	spin_unlock_irqrestore(&rport->lock, flags);

848 849
	nvme_fc_abort_lsops(rport);

850 851 852
	if (atomic_read(&rport->act_ctrl_cnt) == 0)
		rport->lport->ops->remoteport_delete(portptr);

853 854 855 856 857
	/*
	 * release the reference, which will allow, if all controllers
	 * go away, which should only occur after dev_loss_tmo occurs,
	 * for the rport to be torn down.
	 */
858
	nvme_fc_rport_put(rport);
859

860 861 862 863
	return 0;
}
EXPORT_SYMBOL_GPL(nvme_fc_unregister_remoteport);

J
James Smart 已提交
864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880
/**
 * nvme_fc_rescan_remoteport - transport entry point called by an
 *                              LLDD to request a nvme device rescan.
 * @remoteport: pointer to the (registered) remote port that is to be
 *              rescanned.
 *
 * Returns: N/A
 */
void
nvme_fc_rescan_remoteport(struct nvme_fc_remote_port *remoteport)
{
	struct nvme_fc_rport *rport = remoteport_to_rport(remoteport);

	nvme_fc_signal_discovery_scan(rport->lport, rport);
}
EXPORT_SYMBOL_GPL(nvme_fc_rescan_remoteport);

881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903
int
nvme_fc_set_remoteport_devloss(struct nvme_fc_remote_port *portptr,
			u32 dev_loss_tmo)
{
	struct nvme_fc_rport *rport = remoteport_to_rport(portptr);
	unsigned long flags;

	spin_lock_irqsave(&rport->lock, flags);

	if (portptr->port_state != FC_OBJSTATE_ONLINE) {
		spin_unlock_irqrestore(&rport->lock, flags);
		return -EINVAL;
	}

	/* a dev_loss_tmo of 0 (immediate) is allowed to be set */
	rport->remoteport.dev_loss_tmo = dev_loss_tmo;

	spin_unlock_irqrestore(&rport->lock, flags);

	return 0;
}
EXPORT_SYMBOL_GPL(nvme_fc_set_remoteport_devloss);

904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999

/* *********************** FC-NVME DMA Handling **************************** */

/*
 * The fcloop device passes in a NULL device pointer. Real LLD's will
 * pass in a valid device pointer. If NULL is passed to the dma mapping
 * routines, depending on the platform, it may or may not succeed, and
 * may crash.
 *
 * As such:
 * Wrapper all the dma routines and check the dev pointer.
 *
 * If simple mappings (return just a dma address, we'll noop them,
 * returning a dma address of 0.
 *
 * On more complex mappings (dma_map_sg), a pseudo routine fills
 * in the scatter list, setting all dma addresses to 0.
 */

static inline dma_addr_t
fc_dma_map_single(struct device *dev, void *ptr, size_t size,
		enum dma_data_direction dir)
{
	return dev ? dma_map_single(dev, ptr, size, dir) : (dma_addr_t)0L;
}

static inline int
fc_dma_mapping_error(struct device *dev, dma_addr_t dma_addr)
{
	return dev ? dma_mapping_error(dev, dma_addr) : 0;
}

static inline void
fc_dma_unmap_single(struct device *dev, dma_addr_t addr, size_t size,
	enum dma_data_direction dir)
{
	if (dev)
		dma_unmap_single(dev, addr, size, dir);
}

static inline void
fc_dma_sync_single_for_cpu(struct device *dev, dma_addr_t addr, size_t size,
		enum dma_data_direction dir)
{
	if (dev)
		dma_sync_single_for_cpu(dev, addr, size, dir);
}

static inline void
fc_dma_sync_single_for_device(struct device *dev, dma_addr_t addr, size_t size,
		enum dma_data_direction dir)
{
	if (dev)
		dma_sync_single_for_device(dev, addr, size, dir);
}

/* pseudo dma_map_sg call */
static int
fc_map_sg(struct scatterlist *sg, int nents)
{
	struct scatterlist *s;
	int i;

	WARN_ON(nents == 0 || sg[0].length == 0);

	for_each_sg(sg, s, nents, i) {
		s->dma_address = 0L;
#ifdef CONFIG_NEED_SG_DMA_LENGTH
		s->dma_length = s->length;
#endif
	}
	return nents;
}

static inline int
fc_dma_map_sg(struct device *dev, struct scatterlist *sg, int nents,
		enum dma_data_direction dir)
{
	return dev ? dma_map_sg(dev, sg, nents, dir) : fc_map_sg(sg, nents);
}

static inline void
fc_dma_unmap_sg(struct device *dev, struct scatterlist *sg, int nents,
		enum dma_data_direction dir)
{
	if (dev)
		dma_unmap_sg(dev, sg, nents, dir);
}

/* *********************** FC-NVME LS Handling **************************** */

static void nvme_fc_ctrl_put(struct nvme_fc_ctrl *);
static int nvme_fc_ctrl_get(struct nvme_fc_ctrl *);


static void
J
James Smart 已提交
1000
__nvme_fc_finish_ls_req(struct nvmefc_ls_req_op *lsop)
1001
{
J
James Smart 已提交
1002
	struct nvme_fc_rport *rport = lsop->rport;
1003 1004 1005
	struct nvmefc_ls_req *lsreq = &lsop->ls_req;
	unsigned long flags;

J
James Smart 已提交
1006
	spin_lock_irqsave(&rport->lock, flags);
1007 1008

	if (!lsop->req_queued) {
J
James Smart 已提交
1009
		spin_unlock_irqrestore(&rport->lock, flags);
1010 1011 1012 1013 1014 1015 1016
		return;
	}

	list_del(&lsop->lsreq_list);

	lsop->req_queued = false;

J
James Smart 已提交
1017
	spin_unlock_irqrestore(&rport->lock, flags);
1018

J
James Smart 已提交
1019
	fc_dma_unmap_single(rport->dev, lsreq->rqstdma,
1020 1021 1022
				  (lsreq->rqstlen + lsreq->rsplen),
				  DMA_BIDIRECTIONAL);

J
James Smart 已提交
1023
	nvme_fc_rport_put(rport);
1024 1025 1026
}

static int
J
James Smart 已提交
1027
__nvme_fc_send_ls_req(struct nvme_fc_rport *rport,
1028 1029 1030 1031 1032
		struct nvmefc_ls_req_op *lsop,
		void (*done)(struct nvmefc_ls_req *req, int status))
{
	struct nvmefc_ls_req *lsreq = &lsop->ls_req;
	unsigned long flags;
J
James Smart 已提交
1033
	int ret = 0;
1034

J
James Smart 已提交
1035 1036 1037 1038
	if (rport->remoteport.port_state != FC_OBJSTATE_ONLINE)
		return -ECONNREFUSED;

	if (!nvme_fc_rport_get(rport))
1039 1040 1041
		return -ESHUTDOWN;

	lsreq->done = done;
J
James Smart 已提交
1042
	lsop->rport = rport;
1043 1044 1045 1046
	lsop->req_queued = false;
	INIT_LIST_HEAD(&lsop->lsreq_list);
	init_completion(&lsop->ls_done);

J
James Smart 已提交
1047
	lsreq->rqstdma = fc_dma_map_single(rport->dev, lsreq->rqstaddr,
1048 1049
				  lsreq->rqstlen + lsreq->rsplen,
				  DMA_BIDIRECTIONAL);
J
James Smart 已提交
1050 1051 1052
	if (fc_dma_mapping_error(rport->dev, lsreq->rqstdma)) {
		ret = -EFAULT;
		goto out_putrport;
1053 1054 1055
	}
	lsreq->rspdma = lsreq->rqstdma + lsreq->rqstlen;

J
James Smart 已提交
1056
	spin_lock_irqsave(&rport->lock, flags);
1057

J
James Smart 已提交
1058
	list_add_tail(&lsop->lsreq_list, &rport->ls_req_list);
1059 1060 1061

	lsop->req_queued = true;

J
James Smart 已提交
1062
	spin_unlock_irqrestore(&rport->lock, flags);
1063

J
James Smart 已提交
1064 1065
	ret = rport->lport->ops->ls_req(&rport->lport->localport,
					&rport->remoteport, lsreq);
1066
	if (ret)
J
James Smart 已提交
1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081
		goto out_unlink;

	return 0;

out_unlink:
	lsop->ls_error = ret;
	spin_lock_irqsave(&rport->lock, flags);
	lsop->req_queued = false;
	list_del(&lsop->lsreq_list);
	spin_unlock_irqrestore(&rport->lock, flags);
	fc_dma_unmap_single(rport->dev, lsreq->rqstdma,
				  (lsreq->rqstlen + lsreq->rsplen),
				  DMA_BIDIRECTIONAL);
out_putrport:
	nvme_fc_rport_put(rport);
1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095

	return ret;
}

static void
nvme_fc_send_ls_req_done(struct nvmefc_ls_req *lsreq, int status)
{
	struct nvmefc_ls_req_op *lsop = ls_req_to_lsop(lsreq);

	lsop->ls_error = status;
	complete(&lsop->ls_done);
}

static int
J
James Smart 已提交
1096
nvme_fc_send_ls_req(struct nvme_fc_rport *rport, struct nvmefc_ls_req_op *lsop)
1097 1098 1099 1100 1101
{
	struct nvmefc_ls_req *lsreq = &lsop->ls_req;
	struct fcnvme_ls_rjt *rjt = lsreq->rspaddr;
	int ret;

J
James Smart 已提交
1102
	ret = __nvme_fc_send_ls_req(rport, lsop, nvme_fc_send_ls_req_done);
1103

J
James Smart 已提交
1104
	if (!ret) {
1105 1106 1107 1108 1109 1110 1111 1112
		/*
		 * No timeout/not interruptible as we need the struct
		 * to exist until the lldd calls us back. Thus mandate
		 * wait until driver calls back. lldd responsible for
		 * the timeout action
		 */
		wait_for_completion(&lsop->ls_done);

J
James Smart 已提交
1113
		__nvme_fc_finish_ls_req(lsop);
1114

J
James Smart 已提交
1115
		ret = lsop->ls_error;
1116 1117
	}

J
James Smart 已提交
1118 1119 1120
	if (ret)
		return ret;

1121 1122 1123 1124 1125 1126 1127
	/* ACC or RJT payload ? */
	if (rjt->w0.ls_cmd == FCNVME_LS_RJT)
		return -ENXIO;

	return 0;
}

J
James Smart 已提交
1128 1129
static int
nvme_fc_send_ls_req_async(struct nvme_fc_rport *rport,
1130 1131 1132 1133 1134
		struct nvmefc_ls_req_op *lsop,
		void (*done)(struct nvmefc_ls_req *req, int status))
{
	/* don't wait for completion */

J
James Smart 已提交
1135
	return __nvme_fc_send_ls_req(rport, lsop, done);
1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207
}

/* Validation Error indexes into the string table below */
enum {
	VERR_NO_ERROR		= 0,
	VERR_LSACC		= 1,
	VERR_LSDESC_RQST	= 2,
	VERR_LSDESC_RQST_LEN	= 3,
	VERR_ASSOC_ID		= 4,
	VERR_ASSOC_ID_LEN	= 5,
	VERR_CONN_ID		= 6,
	VERR_CONN_ID_LEN	= 7,
	VERR_CR_ASSOC		= 8,
	VERR_CR_ASSOC_ACC_LEN	= 9,
	VERR_CR_CONN		= 10,
	VERR_CR_CONN_ACC_LEN	= 11,
	VERR_DISCONN		= 12,
	VERR_DISCONN_ACC_LEN	= 13,
};

static char *validation_errors[] = {
	"OK",
	"Not LS_ACC",
	"Not LSDESC_RQST",
	"Bad LSDESC_RQST Length",
	"Not Association ID",
	"Bad Association ID Length",
	"Not Connection ID",
	"Bad Connection ID Length",
	"Not CR_ASSOC Rqst",
	"Bad CR_ASSOC ACC Length",
	"Not CR_CONN Rqst",
	"Bad CR_CONN ACC Length",
	"Not Disconnect Rqst",
	"Bad Disconnect ACC Length",
};

static int
nvme_fc_connect_admin_queue(struct nvme_fc_ctrl *ctrl,
	struct nvme_fc_queue *queue, u16 qsize, u16 ersp_ratio)
{
	struct nvmefc_ls_req_op *lsop;
	struct nvmefc_ls_req *lsreq;
	struct fcnvme_ls_cr_assoc_rqst *assoc_rqst;
	struct fcnvme_ls_cr_assoc_acc *assoc_acc;
	int ret, fcret = 0;

	lsop = kzalloc((sizeof(*lsop) +
			 ctrl->lport->ops->lsrqst_priv_sz +
			 sizeof(*assoc_rqst) + sizeof(*assoc_acc)), GFP_KERNEL);
	if (!lsop) {
		ret = -ENOMEM;
		goto out_no_memory;
	}
	lsreq = &lsop->ls_req;

	lsreq->private = (void *)&lsop[1];
	assoc_rqst = (struct fcnvme_ls_cr_assoc_rqst *)
			(lsreq->private + ctrl->lport->ops->lsrqst_priv_sz);
	assoc_acc = (struct fcnvme_ls_cr_assoc_acc *)&assoc_rqst[1];

	assoc_rqst->w0.ls_cmd = FCNVME_LS_CREATE_ASSOCIATION;
	assoc_rqst->desc_list_len =
			cpu_to_be32(sizeof(struct fcnvme_lsdesc_cr_assoc_cmd));

	assoc_rqst->assoc_cmd.desc_tag =
			cpu_to_be32(FCNVME_LSDESC_CREATE_ASSOC_CMD);
	assoc_rqst->assoc_cmd.desc_len =
			fcnvme_lsdesc_len(
				sizeof(struct fcnvme_lsdesc_cr_assoc_cmd));

	assoc_rqst->assoc_cmd.ersp_ratio = cpu_to_be16(ersp_ratio);
J
James Smart 已提交
1208
	assoc_rqst->assoc_cmd.sqsize = cpu_to_be16(qsize - 1);
1209 1210
	/* Linux supports only Dynamic controllers */
	assoc_rqst->assoc_cmd.cntlid = cpu_to_be16(0xffff);
C
Christoph Hellwig 已提交
1211
	uuid_copy(&assoc_rqst->assoc_cmd.hostid, &ctrl->ctrl.opts->host->id);
1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223
	strncpy(assoc_rqst->assoc_cmd.hostnqn, ctrl->ctrl.opts->host->nqn,
		min(FCNVME_ASSOC_HOSTNQN_LEN, NVMF_NQN_SIZE));
	strncpy(assoc_rqst->assoc_cmd.subnqn, ctrl->ctrl.opts->subsysnqn,
		min(FCNVME_ASSOC_SUBNQN_LEN, NVMF_NQN_SIZE));

	lsop->queue = queue;
	lsreq->rqstaddr = assoc_rqst;
	lsreq->rqstlen = sizeof(*assoc_rqst);
	lsreq->rspaddr = assoc_acc;
	lsreq->rsplen = sizeof(*assoc_acc);
	lsreq->timeout = NVME_FC_CONNECT_TIMEOUT_SEC;

J
James Smart 已提交
1224
	ret = nvme_fc_send_ls_req(ctrl->rport, lsop);
1225 1226 1227 1228 1229 1230 1231 1232
	if (ret)
		goto out_free_buffer;

	/* process connect LS completion */

	/* validate the ACC response */
	if (assoc_acc->hdr.w0.ls_cmd != FCNVME_LS_ACC)
		fcret = VERR_LSACC;
J
James Smart 已提交
1233
	else if (assoc_acc->hdr.desc_list_len !=
1234 1235 1236
			fcnvme_lsdesc_len(
				sizeof(struct fcnvme_ls_cr_assoc_acc)))
		fcret = VERR_CR_ASSOC_ACC_LEN;
J
James Smart 已提交
1237 1238
	else if (assoc_acc->hdr.rqst.desc_tag !=
			cpu_to_be32(FCNVME_LSDESC_RQST))
1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322
		fcret = VERR_LSDESC_RQST;
	else if (assoc_acc->hdr.rqst.desc_len !=
			fcnvme_lsdesc_len(sizeof(struct fcnvme_lsdesc_rqst)))
		fcret = VERR_LSDESC_RQST_LEN;
	else if (assoc_acc->hdr.rqst.w0.ls_cmd != FCNVME_LS_CREATE_ASSOCIATION)
		fcret = VERR_CR_ASSOC;
	else if (assoc_acc->associd.desc_tag !=
			cpu_to_be32(FCNVME_LSDESC_ASSOC_ID))
		fcret = VERR_ASSOC_ID;
	else if (assoc_acc->associd.desc_len !=
			fcnvme_lsdesc_len(
				sizeof(struct fcnvme_lsdesc_assoc_id)))
		fcret = VERR_ASSOC_ID_LEN;
	else if (assoc_acc->connectid.desc_tag !=
			cpu_to_be32(FCNVME_LSDESC_CONN_ID))
		fcret = VERR_CONN_ID;
	else if (assoc_acc->connectid.desc_len !=
			fcnvme_lsdesc_len(sizeof(struct fcnvme_lsdesc_conn_id)))
		fcret = VERR_CONN_ID_LEN;

	if (fcret) {
		ret = -EBADF;
		dev_err(ctrl->dev,
			"q %d connect failed: %s\n",
			queue->qnum, validation_errors[fcret]);
	} else {
		ctrl->association_id =
			be64_to_cpu(assoc_acc->associd.association_id);
		queue->connection_id =
			be64_to_cpu(assoc_acc->connectid.connection_id);
		set_bit(NVME_FC_Q_CONNECTED, &queue->flags);
	}

out_free_buffer:
	kfree(lsop);
out_no_memory:
	if (ret)
		dev_err(ctrl->dev,
			"queue %d connect admin queue failed (%d).\n",
			queue->qnum, ret);
	return ret;
}

static int
nvme_fc_connect_queue(struct nvme_fc_ctrl *ctrl, struct nvme_fc_queue *queue,
			u16 qsize, u16 ersp_ratio)
{
	struct nvmefc_ls_req_op *lsop;
	struct nvmefc_ls_req *lsreq;
	struct fcnvme_ls_cr_conn_rqst *conn_rqst;
	struct fcnvme_ls_cr_conn_acc *conn_acc;
	int ret, fcret = 0;

	lsop = kzalloc((sizeof(*lsop) +
			 ctrl->lport->ops->lsrqst_priv_sz +
			 sizeof(*conn_rqst) + sizeof(*conn_acc)), GFP_KERNEL);
	if (!lsop) {
		ret = -ENOMEM;
		goto out_no_memory;
	}
	lsreq = &lsop->ls_req;

	lsreq->private = (void *)&lsop[1];
	conn_rqst = (struct fcnvme_ls_cr_conn_rqst *)
			(lsreq->private + ctrl->lport->ops->lsrqst_priv_sz);
	conn_acc = (struct fcnvme_ls_cr_conn_acc *)&conn_rqst[1];

	conn_rqst->w0.ls_cmd = FCNVME_LS_CREATE_CONNECTION;
	conn_rqst->desc_list_len = cpu_to_be32(
				sizeof(struct fcnvme_lsdesc_assoc_id) +
				sizeof(struct fcnvme_lsdesc_cr_conn_cmd));

	conn_rqst->associd.desc_tag = cpu_to_be32(FCNVME_LSDESC_ASSOC_ID);
	conn_rqst->associd.desc_len =
			fcnvme_lsdesc_len(
				sizeof(struct fcnvme_lsdesc_assoc_id));
	conn_rqst->associd.association_id = cpu_to_be64(ctrl->association_id);
	conn_rqst->connect_cmd.desc_tag =
			cpu_to_be32(FCNVME_LSDESC_CREATE_CONN_CMD);
	conn_rqst->connect_cmd.desc_len =
			fcnvme_lsdesc_len(
				sizeof(struct fcnvme_lsdesc_cr_conn_cmd));
	conn_rqst->connect_cmd.ersp_ratio = cpu_to_be16(ersp_ratio);
	conn_rqst->connect_cmd.qid  = cpu_to_be16(queue->qnum);
J
James Smart 已提交
1323
	conn_rqst->connect_cmd.sqsize = cpu_to_be16(qsize - 1);
1324 1325 1326 1327 1328 1329 1330 1331

	lsop->queue = queue;
	lsreq->rqstaddr = conn_rqst;
	lsreq->rqstlen = sizeof(*conn_rqst);
	lsreq->rspaddr = conn_acc;
	lsreq->rsplen = sizeof(*conn_acc);
	lsreq->timeout = NVME_FC_CONNECT_TIMEOUT_SEC;

J
James Smart 已提交
1332
	ret = nvme_fc_send_ls_req(ctrl->rport, lsop);
1333 1334 1335 1336 1337 1338 1339 1340
	if (ret)
		goto out_free_buffer;

	/* process connect LS completion */

	/* validate the ACC response */
	if (conn_acc->hdr.w0.ls_cmd != FCNVME_LS_ACC)
		fcret = VERR_LSACC;
J
James Smart 已提交
1341
	else if (conn_acc->hdr.desc_list_len !=
1342 1343
			fcnvme_lsdesc_len(sizeof(struct fcnvme_ls_cr_conn_acc)))
		fcret = VERR_CR_CONN_ACC_LEN;
J
James Smart 已提交
1344
	else if (conn_acc->hdr.rqst.desc_tag != cpu_to_be32(FCNVME_LSDESC_RQST))
1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383
		fcret = VERR_LSDESC_RQST;
	else if (conn_acc->hdr.rqst.desc_len !=
			fcnvme_lsdesc_len(sizeof(struct fcnvme_lsdesc_rqst)))
		fcret = VERR_LSDESC_RQST_LEN;
	else if (conn_acc->hdr.rqst.w0.ls_cmd != FCNVME_LS_CREATE_CONNECTION)
		fcret = VERR_CR_CONN;
	else if (conn_acc->connectid.desc_tag !=
			cpu_to_be32(FCNVME_LSDESC_CONN_ID))
		fcret = VERR_CONN_ID;
	else if (conn_acc->connectid.desc_len !=
			fcnvme_lsdesc_len(sizeof(struct fcnvme_lsdesc_conn_id)))
		fcret = VERR_CONN_ID_LEN;

	if (fcret) {
		ret = -EBADF;
		dev_err(ctrl->dev,
			"q %d connect failed: %s\n",
			queue->qnum, validation_errors[fcret]);
	} else {
		queue->connection_id =
			be64_to_cpu(conn_acc->connectid.connection_id);
		set_bit(NVME_FC_Q_CONNECTED, &queue->flags);
	}

out_free_buffer:
	kfree(lsop);
out_no_memory:
	if (ret)
		dev_err(ctrl->dev,
			"queue %d connect command failed (%d).\n",
			queue->qnum, ret);
	return ret;
}

static void
nvme_fc_disconnect_assoc_done(struct nvmefc_ls_req *lsreq, int status)
{
	struct nvmefc_ls_req_op *lsop = ls_req_to_lsop(lsreq);

J
James Smart 已提交
1384
	__nvme_fc_finish_ls_req(lsop);
1385

M
Milan P. Gandhi 已提交
1386
	/* fc-nvme initiator doesn't care about success or failure of cmd */
1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414

	kfree(lsop);
}

/*
 * This routine sends a FC-NVME LS to disconnect (aka terminate)
 * the FC-NVME Association.  Terminating the association also
 * terminates the FC-NVME connections (per queue, both admin and io
 * queues) that are part of the association. E.g. things are torn
 * down, and the related FC-NVME Association ID and Connection IDs
 * become invalid.
 *
 * The behavior of the fc-nvme initiator is such that it's
 * understanding of the association and connections will implicitly
 * be torn down. The action is implicit as it may be due to a loss of
 * connectivity with the fc-nvme target, so you may never get a
 * response even if you tried.  As such, the action of this routine
 * is to asynchronously send the LS, ignore any results of the LS, and
 * continue on with terminating the association. If the fc-nvme target
 * is present and receives the LS, it too can tear down.
 */
static void
nvme_fc_xmt_disconnect_assoc(struct nvme_fc_ctrl *ctrl)
{
	struct fcnvme_ls_disconnect_rqst *discon_rqst;
	struct fcnvme_ls_disconnect_acc *discon_acc;
	struct nvmefc_ls_req_op *lsop;
	struct nvmefc_ls_req *lsreq;
J
James Smart 已提交
1415
	int ret;
1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457

	lsop = kzalloc((sizeof(*lsop) +
			 ctrl->lport->ops->lsrqst_priv_sz +
			 sizeof(*discon_rqst) + sizeof(*discon_acc)),
			GFP_KERNEL);
	if (!lsop)
		/* couldn't sent it... too bad */
		return;

	lsreq = &lsop->ls_req;

	lsreq->private = (void *)&lsop[1];
	discon_rqst = (struct fcnvme_ls_disconnect_rqst *)
			(lsreq->private + ctrl->lport->ops->lsrqst_priv_sz);
	discon_acc = (struct fcnvme_ls_disconnect_acc *)&discon_rqst[1];

	discon_rqst->w0.ls_cmd = FCNVME_LS_DISCONNECT;
	discon_rqst->desc_list_len = cpu_to_be32(
				sizeof(struct fcnvme_lsdesc_assoc_id) +
				sizeof(struct fcnvme_lsdesc_disconn_cmd));

	discon_rqst->associd.desc_tag = cpu_to_be32(FCNVME_LSDESC_ASSOC_ID);
	discon_rqst->associd.desc_len =
			fcnvme_lsdesc_len(
				sizeof(struct fcnvme_lsdesc_assoc_id));

	discon_rqst->associd.association_id = cpu_to_be64(ctrl->association_id);

	discon_rqst->discon_cmd.desc_tag = cpu_to_be32(
						FCNVME_LSDESC_DISCONN_CMD);
	discon_rqst->discon_cmd.desc_len =
			fcnvme_lsdesc_len(
				sizeof(struct fcnvme_lsdesc_disconn_cmd));
	discon_rqst->discon_cmd.scope = FCNVME_DISCONN_ASSOCIATION;
	discon_rqst->discon_cmd.id = cpu_to_be64(ctrl->association_id);

	lsreq->rqstaddr = discon_rqst;
	lsreq->rqstlen = sizeof(*discon_rqst);
	lsreq->rspaddr = discon_acc;
	lsreq->rsplen = sizeof(*discon_acc);
	lsreq->timeout = NVME_FC_CONNECT_TIMEOUT_SEC;

J
James Smart 已提交
1458 1459 1460 1461
	ret = nvme_fc_send_ls_req_async(ctrl->rport, lsop,
				nvme_fc_disconnect_assoc_done);
	if (ret)
		kfree(lsop);
1462 1463 1464 1465 1466 1467 1468 1469

	/* only meaningful part to terminating the association */
	ctrl->association_id = 0;
}


/* *********************** NVME Ctrl Routines **************************** */

1470
static void nvme_fc_error_recovery(struct nvme_fc_ctrl *ctrl, char *errmsg);
1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484

static void
__nvme_fc_exit_request(struct nvme_fc_ctrl *ctrl,
		struct nvme_fc_fcp_op *op)
{
	fc_dma_unmap_single(ctrl->lport->dev, op->fcp_req.rspdma,
				sizeof(op->rsp_iu), DMA_FROM_DEVICE);
	fc_dma_unmap_single(ctrl->lport->dev, op->fcp_req.cmddma,
				sizeof(op->cmd_iu), DMA_TO_DEVICE);

	atomic_set(&op->state, FCPOP_STATE_UNINIT);
}

static void
1485 1486
nvme_fc_exit_request(struct blk_mq_tag_set *set, struct request *rq,
		unsigned int hctx_idx)
1487 1488 1489
{
	struct nvme_fc_fcp_op *op = blk_mq_rq_to_pdu(rq);

1490
	return __nvme_fc_exit_request(set->driver_data, op);
1491 1492
}

J
James Smart 已提交
1493 1494 1495
static int
__nvme_fc_abort_op(struct nvme_fc_ctrl *ctrl, struct nvme_fc_fcp_op *op)
{
1496 1497 1498 1499 1500 1501 1502 1503 1504 1505
	unsigned long flags;
	int opstate;

	spin_lock_irqsave(&ctrl->lock, flags);
	opstate = atomic_xchg(&op->state, FCPOP_STATE_ABORTED);
	if (opstate != FCPOP_STATE_ACTIVE)
		atomic_set(&op->state, opstate);
	else if (ctrl->flags & FCCTRL_TERMIO)
		ctrl->iocnt++;
	spin_unlock_irqrestore(&ctrl->lock, flags);
J
James Smart 已提交
1506

1507
	if (opstate != FCPOP_STATE_ACTIVE)
J
James Smart 已提交
1508 1509 1510 1511 1512 1513 1514 1515 1516 1517
		return -ECANCELED;

	ctrl->lport->ops->fcp_abort(&ctrl->lport->localport,
					&ctrl->rport->remoteport,
					op->queue->lldd_handle,
					&op->fcp_req);

	return 0;
}

1518
static void
J
James Smart 已提交
1519
nvme_fc_abort_aen_ops(struct nvme_fc_ctrl *ctrl)
1520 1521
{
	struct nvme_fc_fcp_op *aen_op = ctrl->aen_ops;
1522
	int i;
J
James Smart 已提交
1523

1524 1525 1526 1527
	/* ensure we've initialized the ops once */
	if (!(aen_op->flags & FCOP_FLAGS_AEN))
		return;

1528 1529
	for (i = 0; i < NVME_NR_AEN_COMMANDS; i++, aen_op++)
		__nvme_fc_abort_op(ctrl, aen_op);
1530 1531
}

J
James Smart 已提交
1532
static inline void
J
James Smart 已提交
1533
__nvme_fc_fcpop_chk_teardowns(struct nvme_fc_ctrl *ctrl,
1534
		struct nvme_fc_fcp_op *op, int opstate)
J
James Smart 已提交
1535 1536 1537
{
	unsigned long flags;

J
James Smart 已提交
1538 1539 1540 1541 1542 1543 1544
	if (opstate == FCPOP_STATE_ABORTED) {
		spin_lock_irqsave(&ctrl->lock, flags);
		if (ctrl->flags & FCCTRL_TERMIO) {
			if (!--ctrl->iocnt)
				wake_up(&ctrl->ioabort_wait);
		}
		spin_unlock_irqrestore(&ctrl->lock, flags);
1545
	}
J
James Smart 已提交
1546 1547
}

1548
static void
1549 1550 1551 1552 1553 1554 1555 1556
nvme_fc_fcpio_done(struct nvmefc_fcp_req *req)
{
	struct nvme_fc_fcp_op *op = fcp_req_to_fcp_op(req);
	struct request *rq = op->rq;
	struct nvmefc_fcp_req *freq = &op->fcp_req;
	struct nvme_fc_ctrl *ctrl = op->ctrl;
	struct nvme_fc_queue *queue = op->queue;
	struct nvme_completion *cqe = &op->rsp_iu.cqe;
J
James Smart 已提交
1557
	struct nvme_command *sqe = &op->cmd_iu.sqe;
1558
	__le16 status = cpu_to_le16(NVME_SC_SUCCESS << 1);
1559
	union nvme_result result;
1560
	bool terminate_assoc = true;
1561
	int opstate;
1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589

	/*
	 * WARNING:
	 * The current linux implementation of a nvme controller
	 * allocates a single tag set for all io queues and sizes
	 * the io queues to fully hold all possible tags. Thus, the
	 * implementation does not reference or care about the sqhd
	 * value as it never needs to use the sqhd/sqtail pointers
	 * for submission pacing.
	 *
	 * This affects the FC-NVME implementation in two ways:
	 * 1) As the value doesn't matter, we don't need to waste
	 *    cycles extracting it from ERSPs and stamping it in the
	 *    cases where the transport fabricates CQEs on successful
	 *    completions.
	 * 2) The FC-NVME implementation requires that delivery of
	 *    ERSP completions are to go back to the nvme layer in order
	 *    relative to the rsn, such that the sqhd value will always
	 *    be "in order" for the nvme layer. As the nvme layer in
	 *    linux doesn't care about sqhd, there's no need to return
	 *    them in order.
	 *
	 * Additionally:
	 * As the core nvme layer in linux currently does not look at
	 * every field in the cqe - in cases where the FC transport must
	 * fabricate a CQE, the following fields will not be set as they
	 * are not referenced:
	 *      cqe.sqid,  cqe.sqhd,  cqe.command_id
1590 1591 1592 1593 1594 1595 1596 1597
	 *
	 * Failure or error of an individual i/o, in a transport
	 * detected fashion unrelated to the nvme completion status,
	 * potentially cause the initiator and target sides to get out
	 * of sync on SQ head/tail (aka outstanding io count allowed).
	 * Per FC-NVME spec, failure of an individual command requires
	 * the connection to be terminated, which in turn requires the
	 * association to be terminated.
1598 1599
	 */

1600 1601
	opstate = atomic_xchg(&op->state, FCPOP_STATE_COMPLETE);

1602 1603 1604
	fc_dma_sync_single_for_cpu(ctrl->lport->dev, op->fcp_req.rspdma,
				sizeof(op->rsp_iu), DMA_FROM_DEVICE);

1605
	if (opstate == FCPOP_STATE_ABORTED)
1606
		status = cpu_to_le16(NVME_SC_ABORT_REQ << 1);
1607
	else if (freq->status)
1608
		status = cpu_to_le16(NVME_SC_INTERNAL << 1);
1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635

	/*
	 * For the linux implementation, if we have an unsuccesful
	 * status, they blk-mq layer can typically be called with the
	 * non-zero status and the content of the cqe isn't important.
	 */
	if (status)
		goto done;

	/*
	 * command completed successfully relative to the wire
	 * protocol. However, validate anything received and
	 * extract the status and result from the cqe (create it
	 * where necessary).
	 */

	switch (freq->rcv_rsplen) {

	case 0:
	case NVME_FC_SIZEOF_ZEROS_RSP:
		/*
		 * No response payload or 12 bytes of payload (which
		 * should all be zeros) are considered successful and
		 * no payload in the CQE by the transport.
		 */
		if (freq->transferred_length !=
			be32_to_cpu(op->cmd_iu.data_len)) {
1636
			status = cpu_to_le16(NVME_SC_INTERNAL << 1);
1637 1638
			goto done;
		}
1639
		result.u64 = 0;
1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650
		break;

	case sizeof(struct nvme_fc_ersp_iu):
		/*
		 * The ERSP IU contains a full completion with CQE.
		 * Validate ERSP IU and look at cqe.
		 */
		if (unlikely(be16_to_cpu(op->rsp_iu.iu_len) !=
					(freq->rcv_rsplen / 4) ||
			     be32_to_cpu(op->rsp_iu.xfrd_len) !=
					freq->transferred_length ||
1651
			     op->rsp_iu.status_code ||
J
James Smart 已提交
1652
			     sqe->common.command_id != cqe->command_id)) {
1653
			status = cpu_to_le16(NVME_SC_INTERNAL << 1);
1654 1655
			goto done;
		}
1656
		result = cqe->result;
1657
		status = cqe->status;
1658 1659 1660
		break;

	default:
1661
		status = cpu_to_le16(NVME_SC_INTERNAL << 1);
1662 1663 1664
		goto done;
	}

1665 1666
	terminate_assoc = false;

1667
done:
J
James Smart 已提交
1668
	if (op->flags & FCOP_FLAGS_AEN) {
1669
		nvme_complete_async_event(&queue->ctrl->ctrl, status, &result);
1670
		__nvme_fc_fcpop_chk_teardowns(ctrl, op, opstate);
J
James Smart 已提交
1671 1672
		atomic_set(&op->state, FCPOP_STATE_IDLE);
		op->flags = FCOP_FLAGS_AEN;	/* clear other flags */
1673
		nvme_fc_ctrl_put(ctrl);
1674
		goto check_error;
1675 1676
	}

J
James Smart 已提交
1677 1678
	__nvme_fc_fcpop_chk_teardowns(ctrl, op, opstate);
	nvme_end_request(rq, status, result);
1679 1680 1681 1682

check_error:
	if (terminate_assoc)
		nvme_fc_error_recovery(ctrl, "transport detected io error");
1683 1684 1685 1686 1687 1688 1689
}

static int
__nvme_fc_init_request(struct nvme_fc_ctrl *ctrl,
		struct nvme_fc_queue *queue, struct nvme_fc_fcp_op *op,
		struct request *rq, u32 rqno)
{
1690 1691
	struct nvme_fcp_op_w_sgl *op_w_sgl =
		container_of(op, typeof(*op_w_sgl), op);
1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733
	struct nvme_fc_cmd_iu *cmdiu = &op->cmd_iu;
	int ret = 0;

	memset(op, 0, sizeof(*op));
	op->fcp_req.cmdaddr = &op->cmd_iu;
	op->fcp_req.cmdlen = sizeof(op->cmd_iu);
	op->fcp_req.rspaddr = &op->rsp_iu;
	op->fcp_req.rsplen = sizeof(op->rsp_iu);
	op->fcp_req.done = nvme_fc_fcpio_done;
	op->ctrl = ctrl;
	op->queue = queue;
	op->rq = rq;
	op->rqno = rqno;

	cmdiu->scsi_id = NVME_CMD_SCSI_ID;
	cmdiu->fc_id = NVME_CMD_FC_ID;
	cmdiu->iu_len = cpu_to_be16(sizeof(*cmdiu) / sizeof(u32));

	op->fcp_req.cmddma = fc_dma_map_single(ctrl->lport->dev,
				&op->cmd_iu, sizeof(op->cmd_iu), DMA_TO_DEVICE);
	if (fc_dma_mapping_error(ctrl->lport->dev, op->fcp_req.cmddma)) {
		dev_err(ctrl->dev,
			"FCP Op failed - cmdiu dma mapping failed.\n");
		ret = EFAULT;
		goto out_on_error;
	}

	op->fcp_req.rspdma = fc_dma_map_single(ctrl->lport->dev,
				&op->rsp_iu, sizeof(op->rsp_iu),
				DMA_FROM_DEVICE);
	if (fc_dma_mapping_error(ctrl->lport->dev, op->fcp_req.rspdma)) {
		dev_err(ctrl->dev,
			"FCP Op failed - rspiu dma mapping failed.\n");
		ret = EFAULT;
	}

	atomic_set(&op->state, FCPOP_STATE_IDLE);
out_on_error:
	return ret;
}

static int
1734 1735
nvme_fc_init_request(struct blk_mq_tag_set *set, struct request *rq,
		unsigned int hctx_idx, unsigned int numa_node)
1736
{
1737
	struct nvme_fc_ctrl *ctrl = set->driver_data;
1738
	struct nvme_fcp_op_w_sgl *op = blk_mq_rq_to_pdu(rq);
1739 1740
	int queue_idx = (set == &ctrl->tag_set) ? hctx_idx + 1 : 0;
	struct nvme_fc_queue *queue = &ctrl->queues[queue_idx];
1741
	int res;
1742

1743 1744 1745 1746
	res = __nvme_fc_init_request(ctrl, queue, &op->op, rq, queue->rqcnt++);
	if (res)
		return res;
	op->op.fcp_req.first_sgl = &op->sgl[0];
1747
	op->op.fcp_req.private = &op->priv[0];
1748
	nvme_req(rq)->ctrl = &ctrl->ctrl;
1749
	return res;
1750 1751 1752 1753 1754 1755 1756 1757
}

static int
nvme_fc_init_aen_ops(struct nvme_fc_ctrl *ctrl)
{
	struct nvme_fc_fcp_op *aen_op;
	struct nvme_fc_cmd_iu *cmdiu;
	struct nvme_command *sqe;
1758
	void *private;
1759 1760 1761
	int i, ret;

	aen_op = ctrl->aen_ops;
K
Keith Busch 已提交
1762
	for (i = 0; i < NVME_NR_AEN_COMMANDS; i++, aen_op++) {
1763 1764 1765 1766 1767
		private = kzalloc(ctrl->lport->ops->fcprqst_priv_sz,
						GFP_KERNEL);
		if (!private)
			return -ENOMEM;

1768 1769 1770 1771
		cmdiu = &aen_op->cmd_iu;
		sqe = &cmdiu->sqe;
		ret = __nvme_fc_init_request(ctrl, &ctrl->queues[0],
				aen_op, (struct request *)NULL,
K
Keith Busch 已提交
1772
				(NVME_AQ_BLK_MQ_DEPTH + i));
1773 1774
		if (ret) {
			kfree(private);
1775
			return ret;
1776
		}
1777

J
James Smart 已提交
1778
		aen_op->flags = FCOP_FLAGS_AEN;
1779
		aen_op->fcp_req.private = private;
J
James Smart 已提交
1780

1781 1782
		memset(sqe, 0, sizeof(*sqe));
		sqe->common.opcode = nvme_admin_async_event;
J
James Smart 已提交
1783
		/* Note: core layer may overwrite the sqe.command_id value */
K
Keith Busch 已提交
1784
		sqe->common.command_id = NVME_AQ_BLK_MQ_DEPTH + i;
1785 1786 1787 1788
	}
	return 0;
}

1789 1790 1791 1792 1793 1794 1795
static void
nvme_fc_term_aen_ops(struct nvme_fc_ctrl *ctrl)
{
	struct nvme_fc_fcp_op *aen_op;
	int i;

	aen_op = ctrl->aen_ops;
K
Keith Busch 已提交
1796
	for (i = 0; i < NVME_NR_AEN_COMMANDS; i++, aen_op++) {
1797 1798 1799 1800 1801 1802 1803 1804 1805
		if (!aen_op->fcp_req.private)
			continue;

		__nvme_fc_exit_request(ctrl, aen_op);

		kfree(aen_op->fcp_req.private);
		aen_op->fcp_req.private = NULL;
	}
}
1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839

static inline void
__nvme_fc_init_hctx(struct blk_mq_hw_ctx *hctx, struct nvme_fc_ctrl *ctrl,
		unsigned int qidx)
{
	struct nvme_fc_queue *queue = &ctrl->queues[qidx];

	hctx->driver_data = queue;
	queue->hctx = hctx;
}

static int
nvme_fc_init_hctx(struct blk_mq_hw_ctx *hctx, void *data,
		unsigned int hctx_idx)
{
	struct nvme_fc_ctrl *ctrl = data;

	__nvme_fc_init_hctx(hctx, ctrl, hctx_idx + 1);

	return 0;
}

static int
nvme_fc_init_admin_hctx(struct blk_mq_hw_ctx *hctx, void *data,
		unsigned int hctx_idx)
{
	struct nvme_fc_ctrl *ctrl = data;

	__nvme_fc_init_hctx(hctx, ctrl, hctx_idx);

	return 0;
}

static void
1840
nvme_fc_init_queue(struct nvme_fc_ctrl *ctrl, int idx)
1841 1842 1843 1844 1845 1846 1847
{
	struct nvme_fc_queue *queue;

	queue = &ctrl->queues[idx];
	memset(queue, 0, sizeof(*queue));
	queue->ctrl = ctrl;
	queue->qnum = idx;
1848
	atomic_set(&queue->csn, 0);
1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881
	queue->dev = ctrl->dev;

	if (idx > 0)
		queue->cmnd_capsule_len = ctrl->ctrl.ioccsz * 16;
	else
		queue->cmnd_capsule_len = sizeof(struct nvme_command);

	/*
	 * Considered whether we should allocate buffers for all SQEs
	 * and CQEs and dma map them - mapping their respective entries
	 * into the request structures (kernel vm addr and dma address)
	 * thus the driver could use the buffers/mappings directly.
	 * It only makes sense if the LLDD would use them for its
	 * messaging api. It's very unlikely most adapter api's would use
	 * a native NVME sqe/cqe. More reasonable if FC-NVME IU payload
	 * structures were used instead.
	 */
}

/*
 * This routine terminates a queue at the transport level.
 * The transport has already ensured that all outstanding ios on
 * the queue have been terminated.
 * The transport will send a Disconnect LS request to terminate
 * the queue's connection. Termination of the admin queue will also
 * terminate the association at the target.
 */
static void
nvme_fc_free_queue(struct nvme_fc_queue *queue)
{
	if (!test_and_clear_bit(NVME_FC_Q_CONNECTED, &queue->flags))
		return;

1882
	clear_bit(NVME_FC_Q_LIVE, &queue->flags);
1883 1884 1885 1886 1887 1888 1889
	/*
	 * Current implementation never disconnects a single queue.
	 * It always terminates a whole association. So there is never
	 * a disconnect(queue) LS sent to the target.
	 */

	queue->connection_id = 0;
1890
	atomic_set(&queue->csn, 0);
1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907
}

static void
__nvme_fc_delete_hw_queue(struct nvme_fc_ctrl *ctrl,
	struct nvme_fc_queue *queue, unsigned int qidx)
{
	if (ctrl->lport->ops->delete_queue)
		ctrl->lport->ops->delete_queue(&ctrl->lport->localport, qidx,
				queue->lldd_handle);
	queue->lldd_handle = NULL;
}

static void
nvme_fc_free_io_queues(struct nvme_fc_ctrl *ctrl)
{
	int i;

1908
	for (i = 1; i < ctrl->ctrl.queue_count; i++)
1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928
		nvme_fc_free_queue(&ctrl->queues[i]);
}

static int
__nvme_fc_create_hw_queue(struct nvme_fc_ctrl *ctrl,
	struct nvme_fc_queue *queue, unsigned int qidx, u16 qsize)
{
	int ret = 0;

	queue->lldd_handle = NULL;
	if (ctrl->lport->ops->create_queue)
		ret = ctrl->lport->ops->create_queue(&ctrl->lport->localport,
				qidx, qsize, &queue->lldd_handle);

	return ret;
}

static void
nvme_fc_delete_hw_io_queues(struct nvme_fc_ctrl *ctrl)
{
1929
	struct nvme_fc_queue *queue = &ctrl->queues[ctrl->ctrl.queue_count - 1];
1930 1931
	int i;

1932
	for (i = ctrl->ctrl.queue_count - 1; i >= 1; i--, queue--)
1933 1934 1935 1936 1937 1938 1939
		__nvme_fc_delete_hw_queue(ctrl, queue, i);
}

static int
nvme_fc_create_hw_io_queues(struct nvme_fc_ctrl *ctrl, u16 qsize)
{
	struct nvme_fc_queue *queue = &ctrl->queues[1];
1940
	int i, ret;
1941

1942
	for (i = 1; i < ctrl->ctrl.queue_count; i++, queue++) {
1943
		ret = __nvme_fc_create_hw_queue(ctrl, queue, i, qsize);
1944 1945
		if (ret)
			goto delete_queues;
1946 1947 1948
	}

	return 0;
1949 1950 1951 1952 1953

delete_queues:
	for (; i >= 0; i--)
		__nvme_fc_delete_hw_queue(ctrl, &ctrl->queues[i], i);
	return ret;
1954 1955 1956 1957 1958 1959 1960
}

static int
nvme_fc_connect_io_queues(struct nvme_fc_ctrl *ctrl, u16 qsize)
{
	int i, ret = 0;

1961
	for (i = 1; i < ctrl->ctrl.queue_count; i++) {
1962 1963 1964 1965
		ret = nvme_fc_connect_queue(ctrl, &ctrl->queues[i], qsize,
					(qsize / 5));
		if (ret)
			break;
1966
		ret = nvmf_connect_io_queue(&ctrl->ctrl, i, false);
1967 1968
		if (ret)
			break;
1969 1970

		set_bit(NVME_FC_Q_LIVE, &ctrl->queues[i].flags);
1971 1972 1973 1974 1975 1976 1977 1978 1979 1980
	}

	return ret;
}

static void
nvme_fc_init_io_queues(struct nvme_fc_ctrl *ctrl)
{
	int i;

1981
	for (i = 1; i < ctrl->ctrl.queue_count; i++)
1982
		nvme_fc_init_queue(ctrl, i);
1983 1984 1985 1986 1987 1988 1989 1990 1991
}

static void
nvme_fc_ctrl_free(struct kref *ref)
{
	struct nvme_fc_ctrl *ctrl =
		container_of(ref, struct nvme_fc_ctrl, ref);
	unsigned long flags;

1992 1993 1994
	if (ctrl->ctrl.tagset) {
		blk_cleanup_queue(ctrl->ctrl.connect_q);
		blk_mq_free_tag_set(&ctrl->tag_set);
1995 1996
	}

1997 1998 1999 2000 2001
	/* remove from rport list */
	spin_lock_irqsave(&ctrl->rport->lock, flags);
	list_del(&ctrl->ctrl_list);
	spin_unlock_irqrestore(&ctrl->rport->lock, flags);

2002
	blk_mq_unquiesce_queue(ctrl->ctrl.admin_q);
2003 2004 2005 2006 2007
	blk_cleanup_queue(ctrl->ctrl.admin_q);
	blk_mq_free_tag_set(&ctrl->admin_tag_set);

	kfree(ctrl->queues);

2008 2009 2010 2011
	put_device(ctrl->dev);
	nvme_fc_rport_put(ctrl->rport);

	ida_simple_remove(&nvme_fc_ctrl_cnt, ctrl->cnum);
2012 2013
	if (ctrl->ctrl.opts)
		nvmf_free_options(ctrl->ctrl.opts);
2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033
	kfree(ctrl);
}

static void
nvme_fc_ctrl_put(struct nvme_fc_ctrl *ctrl)
{
	kref_put(&ctrl->ref, nvme_fc_ctrl_free);
}

static int
nvme_fc_ctrl_get(struct nvme_fc_ctrl *ctrl)
{
	return kref_get_unless_zero(&ctrl->ref);
}

/*
 * All accesses from nvme core layer done - can now free the
 * controller. Called after last nvme_put_ctrl() call
 */
static void
2034
nvme_fc_nvme_ctrl_freed(struct nvme_ctrl *nctrl)
2035 2036 2037 2038 2039
{
	struct nvme_fc_ctrl *ctrl = to_fc_ctrl(nctrl);

	WARN_ON(nctrl != &ctrl->ctrl);

2040 2041
	nvme_fc_ctrl_put(ctrl);
}
2042

2043 2044 2045
static void
nvme_fc_error_recovery(struct nvme_fc_ctrl *ctrl, char *errmsg)
{
2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056
	int active;

	/*
	 * if an error (io timeout, etc) while (re)connecting,
	 * it's an error on creating the new association.
	 * Start the error recovery thread if it hasn't already
	 * been started. It is expected there could be multiple
	 * ios hitting this path before things are cleaned up.
	 */
	if (ctrl->ctrl.state == NVME_CTRL_CONNECTING) {
		active = atomic_xchg(&ctrl->err_work_active, 1);
2057
		if (!active && !queue_work(nvme_fc_wq, &ctrl->err_work)) {
2058 2059 2060 2061 2062 2063 2064
			atomic_set(&ctrl->err_work_active, 0);
			WARN_ON(1);
		}
		return;
	}

	/* Otherwise, only proceed if in LIVE state - e.g. on first error */
2065 2066 2067
	if (ctrl->ctrl.state != NVME_CTRL_LIVE)
		return;

2068 2069 2070
	dev_warn(ctrl->ctrl.device,
		"NVME-FC{%d}: transport association error detected: %s\n",
		ctrl->cnum, errmsg);
2071
	dev_warn(ctrl->ctrl.device,
2072
		"NVME-FC{%d}: resetting controller\n", ctrl->cnum);
2073

2074
	nvme_reset_ctrl(&ctrl->ctrl);
2075 2076
}

2077
static enum blk_eh_timer_return
2078 2079 2080 2081 2082 2083
nvme_fc_timeout(struct request *rq, bool reserved)
{
	struct nvme_fc_fcp_op *op = blk_mq_rq_to_pdu(rq);
	struct nvme_fc_ctrl *ctrl = op->ctrl;

	/*
2084
	 * we can't individually ABTS an io without affecting the queue,
2085
	 * thus killing the queue, and thus the association.
2086 2087 2088
	 * So resolve by performing a controller reset, which will stop
	 * the host/io stack, terminate the association on the link,
	 * and recreate an association on the link.
2089
	 */
2090
	nvme_fc_error_recovery(ctrl, "io timeout error");
2091

J
James Smart 已提交
2092 2093 2094 2095 2096 2097
	/*
	 * the io abort has been initiated. Have the reset timer
	 * restarted and the abort completion will complete the io
	 * shortly. Avoids a synchronous wait while the abort finishes.
	 */
	return BLK_EH_RESET_TIMER;
2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109
}

static int
nvme_fc_map_data(struct nvme_fc_ctrl *ctrl, struct request *rq,
		struct nvme_fc_fcp_op *op)
{
	struct nvmefc_fcp_req *freq = &op->fcp_req;
	enum dma_data_direction dir;
	int ret;

	freq->sg_cnt = 0;

2110
	if (!blk_rq_nr_phys_segments(rq))
2111 2112 2113
		return 0;

	freq->sg_table.sgl = freq->first_sgl;
2114 2115
	ret = sg_alloc_table_chained(&freq->sg_table,
			blk_rq_nr_phys_segments(rq), freq->sg_table.sgl);
2116 2117 2118 2119
	if (ret)
		return -ENOMEM;

	op->nents = blk_rq_map_sg(rq->q, rq, freq->sg_table.sgl);
2120
	WARN_ON(op->nents > blk_rq_nr_phys_segments(rq));
2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178
	dir = (rq_data_dir(rq) == WRITE) ? DMA_TO_DEVICE : DMA_FROM_DEVICE;
	freq->sg_cnt = fc_dma_map_sg(ctrl->lport->dev, freq->sg_table.sgl,
				op->nents, dir);
	if (unlikely(freq->sg_cnt <= 0)) {
		sg_free_table_chained(&freq->sg_table, true);
		freq->sg_cnt = 0;
		return -EFAULT;
	}

	/*
	 * TODO: blk_integrity_rq(rq)  for DIF
	 */
	return 0;
}

static void
nvme_fc_unmap_data(struct nvme_fc_ctrl *ctrl, struct request *rq,
		struct nvme_fc_fcp_op *op)
{
	struct nvmefc_fcp_req *freq = &op->fcp_req;

	if (!freq->sg_cnt)
		return;

	fc_dma_unmap_sg(ctrl->lport->dev, freq->sg_table.sgl, op->nents,
				((rq_data_dir(rq) == WRITE) ?
					DMA_TO_DEVICE : DMA_FROM_DEVICE));

	nvme_cleanup_cmd(rq);

	sg_free_table_chained(&freq->sg_table, true);

	freq->sg_cnt = 0;
}

/*
 * In FC, the queue is a logical thing. At transport connect, the target
 * creates its "queue" and returns a handle that is to be given to the
 * target whenever it posts something to the corresponding SQ.  When an
 * SQE is sent on a SQ, FC effectively considers the SQE, or rather the
 * command contained within the SQE, an io, and assigns a FC exchange
 * to it. The SQE and the associated SQ handle are sent in the initial
 * CMD IU sents on the exchange. All transfers relative to the io occur
 * as part of the exchange.  The CQE is the last thing for the io,
 * which is transferred (explicitly or implicitly) with the RSP IU
 * sent on the exchange. After the CQE is received, the FC exchange is
 * terminaed and the Exchange may be used on a different io.
 *
 * The transport to LLDD api has the transport making a request for a
 * new fcp io request to the LLDD. The LLDD then allocates a FC exchange
 * resource and transfers the command. The LLDD will then process all
 * steps to complete the io. Upon completion, the transport done routine
 * is called.
 *
 * So - while the operation is outstanding to the LLDD, there is a link
 * level FC exchange resource that is also outstanding. This must be
 * considered in all cleanup operations.
 */
2179
static blk_status_t
2180 2181 2182 2183 2184 2185
nvme_fc_start_fcp_op(struct nvme_fc_ctrl *ctrl, struct nvme_fc_queue *queue,
	struct nvme_fc_fcp_op *op, u32 data_len,
	enum nvmefc_fcp_datadir	io_dir)
{
	struct nvme_fc_cmd_iu *cmdiu = &op->cmd_iu;
	struct nvme_command *sqe = &cmdiu->sqe;
2186
	int ret, opstate;
2187

2188 2189 2190 2191 2192
	/*
	 * before attempting to send the io, check to see if we believe
	 * the target device is present
	 */
	if (ctrl->rport->remoteport.port_state != FC_OBJSTATE_ONLINE)
2193
		return BLK_STS_RESOURCE;
2194

2195
	if (!nvme_fc_ctrl_get(ctrl))
2196
		return BLK_STS_IOERR;
2197 2198 2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215

	/* format the FC-NVME CMD IU and fcp_req */
	cmdiu->connection_id = cpu_to_be64(queue->connection_id);
	cmdiu->data_len = cpu_to_be32(data_len);
	switch (io_dir) {
	case NVMEFC_FCP_WRITE:
		cmdiu->flags = FCNVME_CMD_FLAGS_WRITE;
		break;
	case NVMEFC_FCP_READ:
		cmdiu->flags = FCNVME_CMD_FLAGS_READ;
		break;
	case NVMEFC_FCP_NODATA:
		cmdiu->flags = 0;
		break;
	}
	op->fcp_req.payload_length = data_len;
	op->fcp_req.io_dir = io_dir;
	op->fcp_req.transferred_length = 0;
	op->fcp_req.rcv_rsplen = 0;
2216
	op->fcp_req.status = NVME_SC_SUCCESS;
2217 2218 2219 2220 2221 2222 2223 2224 2225 2226
	op->fcp_req.sqid = cpu_to_le16(queue->qnum);

	/*
	 * validate per fabric rules, set fields mandated by fabric spec
	 * as well as those by FC-NVME spec.
	 */
	WARN_ON_ONCE(sqe->common.metadata);
	sqe->common.flags |= NVME_CMD_SGL_METABUF;

	/*
2227 2228 2229 2230 2231
	 * format SQE DPTR field per FC-NVME rules:
	 *    type=0x5     Transport SGL Data Block Descriptor
	 *    subtype=0xA  Transport-specific value
	 *    address=0
	 *    length=length of the data series
2232
	 */
2233 2234
	sqe->rw.dptr.sgl.type = (NVME_TRANSPORT_SGL_DATA_DESC << 4) |
					NVME_SGL_FMT_TRANSPORT_A;
2235 2236 2237
	sqe->rw.dptr.sgl.length = cpu_to_le32(data_len);
	sqe->rw.dptr.sgl.addr = 0;

J
James Smart 已提交
2238
	if (!(op->flags & FCOP_FLAGS_AEN)) {
2239 2240 2241 2242
		ret = nvme_fc_map_data(ctrl, op->rq, op);
		if (ret < 0) {
			nvme_cleanup_cmd(op->rq);
			nvme_fc_ctrl_put(ctrl);
2243 2244 2245
			if (ret == -ENOMEM || ret == -EAGAIN)
				return BLK_STS_RESOURCE;
			return BLK_STS_IOERR;
2246 2247 2248 2249 2250 2251 2252 2253
		}
	}

	fc_dma_sync_single_for_device(ctrl->lport->dev, op->fcp_req.cmddma,
				  sizeof(op->cmd_iu), DMA_TO_DEVICE);

	atomic_set(&op->state, FCPOP_STATE_ACTIVE);

J
James Smart 已提交
2254
	if (!(op->flags & FCOP_FLAGS_AEN))
2255 2256
		blk_mq_start_request(op->rq);

2257
	cmdiu->csn = cpu_to_be32(atomic_inc_return(&queue->csn));
2258 2259 2260 2261 2262
	ret = ctrl->lport->ops->fcp_io(&ctrl->lport->localport,
					&ctrl->rport->remoteport,
					queue->lldd_handle, &op->fcp_req);

	if (ret) {
2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274
		/*
		 * If the lld fails to send the command is there an issue with
		 * the csn value?  If the command that fails is the Connect,
		 * no - as the connection won't be live.  If it is a command
		 * post-connect, it's possible a gap in csn may be created.
		 * Does this matter?  As Linux initiators don't send fused
		 * commands, no.  The gap would exist, but as there's nothing
		 * that depends on csn order to be delivered on the target
		 * side, it shouldn't hurt.  It would be difficult for a
		 * target to even detect the csn gap as it has no idea when the
		 * cmd with the csn was supposed to arrive.
		 */
2275 2276 2277
		opstate = atomic_xchg(&op->state, FCPOP_STATE_COMPLETE);
		__nvme_fc_fcpop_chk_teardowns(ctrl, op, opstate);

2278
		if (!(op->flags & FCOP_FLAGS_AEN))
2279 2280 2281 2282
			nvme_fc_unmap_data(ctrl, op->rq, op);

		nvme_fc_ctrl_put(ctrl);

2283 2284
		if (ctrl->rport->remoteport.port_state == FC_OBJSTATE_ONLINE &&
				ret != -EBUSY)
2285
			return BLK_STS_IOERR;
2286

2287
		return BLK_STS_RESOURCE;
2288 2289
	}

2290
	return BLK_STS_OK;
2291 2292
}

2293
static blk_status_t
2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304
nvme_fc_queue_rq(struct blk_mq_hw_ctx *hctx,
			const struct blk_mq_queue_data *bd)
{
	struct nvme_ns *ns = hctx->queue->queuedata;
	struct nvme_fc_queue *queue = hctx->driver_data;
	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;
2305
	bool queue_ready = test_bit(NVME_FC_Q_LIVE, &queue->flags);
2306
	u32 data_len;
2307
	blk_status_t ret;
2308

2309 2310
	if (ctrl->rport->remoteport.port_state != FC_OBJSTATE_ONLINE ||
	    !nvmf_check_ready(&queue->ctrl->ctrl, rq, queue_ready))
2311
		return nvmf_fail_nonready_command(&queue->ctrl->ctrl, rq);
2312

2313 2314 2315 2316
	ret = nvme_setup_cmd(ns, rq, sqe);
	if (ret)
		return ret;

2317 2318 2319 2320 2321 2322 2323 2324 2325 2326
	/*
	 * nvme core doesn't quite treat the rq opaquely. Commands such
	 * as WRITE ZEROES will return a non-zero rq payload_bytes yet
	 * there is no actual payload to be transferred.
	 * To get it right, key data transmission on there being 1 or
	 * more physical segments in the sg list. If there is no
	 * physical segments, there is no payload.
	 */
	if (blk_rq_nr_phys_segments(rq)) {
		data_len = blk_rq_payload_bytes(rq);
2327 2328
		io_dir = ((rq_data_dir(rq) == WRITE) ?
					NVMEFC_FCP_WRITE : NVMEFC_FCP_READ);
2329 2330
	} else {
		data_len = 0;
2331
		io_dir = NVMEFC_FCP_NODATA;
2332 2333
	}

2334 2335 2336 2337 2338

	return nvme_fc_start_fcp_op(ctrl, queue, op, data_len, io_dir);
}

static void
2339
nvme_fc_submit_async_event(struct nvme_ctrl *arg)
2340 2341 2342
{
	struct nvme_fc_ctrl *ctrl = to_fc_ctrl(arg);
	struct nvme_fc_fcp_op *aen_op;
2343 2344
	unsigned long flags;
	bool terminating = false;
2345
	blk_status_t ret;
2346

2347 2348 2349 2350 2351 2352 2353 2354
	spin_lock_irqsave(&ctrl->lock, flags);
	if (ctrl->flags & FCCTRL_TERMIO)
		terminating = true;
	spin_unlock_irqrestore(&ctrl->lock, flags);

	if (terminating)
		return;

2355
	aen_op = &ctrl->aen_ops[0];
2356 2357 2358 2359 2360

	ret = nvme_fc_start_fcp_op(ctrl, aen_op->queue, aen_op, 0,
					NVMEFC_FCP_NODATA);
	if (ret)
		dev_err(ctrl->ctrl.device,
2361
			"failed async event work\n");
2362 2363 2364
}

static void
J
James Smart 已提交
2365
nvme_fc_complete_rq(struct request *rq)
2366 2367 2368 2369
{
	struct nvme_fc_fcp_op *op = blk_mq_rq_to_pdu(rq);
	struct nvme_fc_ctrl *ctrl = op->ctrl;

J
James Smart 已提交
2370
	atomic_set(&op->state, FCPOP_STATE_IDLE);
2371 2372

	nvme_fc_unmap_data(ctrl, rq, op);
2373
	nvme_complete_rq(rq);
2374
	nvme_fc_ctrl_put(ctrl);
J
James Smart 已提交
2375 2376
}

2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389
/*
 * This routine is used by the transport when it needs to find active
 * io on a queue that is to be terminated. The transport uses
 * blk_mq_tagset_busy_itr() to find the busy requests, which then invoke
 * this routine to kill them on a 1 by 1 basis.
 *
 * As FC allocates FC exchange for each io, the transport must contact
 * the LLDD to terminate the exchange, thus releasing the FC exchange.
 * After terminating the exchange the LLDD will call the transport's
 * normal io done path for the request, but it will have an aborted
 * status. The done path will return the io request back to the block
 * layer with an error status.
 */
2390
static bool
2391 2392 2393 2394 2395 2396
nvme_fc_terminate_exchange(struct request *req, void *data, bool reserved)
{
	struct nvme_ctrl *nctrl = data;
	struct nvme_fc_ctrl *ctrl = to_fc_ctrl(nctrl);
	struct nvme_fc_fcp_op *op = blk_mq_rq_to_pdu(req);

2397
	__nvme_fc_abort_op(ctrl, op);
2398
	return true;
2399 2400
}

J
James Smart 已提交
2401

2402 2403 2404 2405 2406 2407 2408 2409
static const struct blk_mq_ops nvme_fc_mq_ops = {
	.queue_rq	= nvme_fc_queue_rq,
	.complete	= nvme_fc_complete_rq,
	.init_request	= nvme_fc_init_request,
	.exit_request	= nvme_fc_exit_request,
	.init_hctx	= nvme_fc_init_hctx,
	.timeout	= nvme_fc_timeout,
};
2410

2411 2412
static int
nvme_fc_create_io_queues(struct nvme_fc_ctrl *ctrl)
2413
{
2414
	struct nvmf_ctrl_options *opts = ctrl->ctrl.opts;
2415
	unsigned int nr_io_queues;
2416
	int ret;
2417

2418 2419 2420
	nr_io_queues = min(min(opts->nr_io_queues, num_online_cpus()),
				ctrl->lport->ops->max_hw_queues);
	ret = nvme_set_queue_count(&ctrl->ctrl, &nr_io_queues);
2421 2422 2423 2424 2425
	if (ret) {
		dev_info(ctrl->ctrl.device,
			"set_queue_count failed: %d\n", ret);
		return ret;
	}
2426

2427 2428
	ctrl->ctrl.queue_count = nr_io_queues + 1;
	if (!nr_io_queues)
2429
		return 0;
2430

2431
	nvme_fc_init_io_queues(ctrl);
2432

2433 2434 2435 2436
	memset(&ctrl->tag_set, 0, sizeof(ctrl->tag_set));
	ctrl->tag_set.ops = &nvme_fc_mq_ops;
	ctrl->tag_set.queue_depth = ctrl->ctrl.opts->queue_size;
	ctrl->tag_set.reserved_tags = 1; /* fabric connect */
2437
	ctrl->tag_set.numa_node = ctrl->ctrl.numa_node;
2438
	ctrl->tag_set.flags = BLK_MQ_F_SHOULD_MERGE;
2439 2440 2441
	ctrl->tag_set.cmd_size =
		struct_size((struct nvme_fcp_op_w_sgl *)NULL, priv,
			    ctrl->lport->ops->fcprqst_priv_sz);
2442
	ctrl->tag_set.driver_data = ctrl;
2443
	ctrl->tag_set.nr_hw_queues = ctrl->ctrl.queue_count - 1;
2444
	ctrl->tag_set.timeout = NVME_IO_TIMEOUT;
2445

2446 2447 2448
	ret = blk_mq_alloc_tag_set(&ctrl->tag_set);
	if (ret)
		return ret;
2449

2450
	ctrl->ctrl.tagset = &ctrl->tag_set;
2451

2452 2453 2454 2455 2456
	ctrl->ctrl.connect_q = blk_mq_init_queue(&ctrl->tag_set);
	if (IS_ERR(ctrl->ctrl.connect_q)) {
		ret = PTR_ERR(ctrl->ctrl.connect_q);
		goto out_free_tag_set;
	}
2457

J
James Smart 已提交
2458
	ret = nvme_fc_create_hw_io_queues(ctrl, ctrl->ctrl.sqsize + 1);
2459
	if (ret)
2460
		goto out_cleanup_blk_queue;
2461

J
James Smart 已提交
2462
	ret = nvme_fc_connect_io_queues(ctrl, ctrl->ctrl.sqsize + 1);
2463 2464
	if (ret)
		goto out_delete_hw_queues;
2465

2466 2467
	ctrl->ioq_live = true;

2468 2469
	return 0;

2470 2471 2472 2473 2474 2475 2476
out_delete_hw_queues:
	nvme_fc_delete_hw_io_queues(ctrl);
out_cleanup_blk_queue:
	blk_cleanup_queue(ctrl->ctrl.connect_q);
out_free_tag_set:
	blk_mq_free_tag_set(&ctrl->tag_set);
	nvme_fc_free_io_queues(ctrl);
2477

2478 2479 2480 2481 2482
	/* force put free routine to ignore io queues */
	ctrl->ctrl.tagset = NULL;

	return ret;
}
2483 2484

static int
2485
nvme_fc_recreate_io_queues(struct nvme_fc_ctrl *ctrl)
2486 2487
{
	struct nvmf_ctrl_options *opts = ctrl->ctrl.opts;
2488
	u32 prior_ioq_cnt = ctrl->ctrl.queue_count - 1;
2489
	unsigned int nr_io_queues;
2490 2491
	int ret;

2492 2493 2494
	nr_io_queues = min(min(opts->nr_io_queues, num_online_cpus()),
				ctrl->lport->ops->max_hw_queues);
	ret = nvme_set_queue_count(&ctrl->ctrl, &nr_io_queues);
2495 2496 2497 2498 2499 2500
	if (ret) {
		dev_info(ctrl->ctrl.device,
			"set_queue_count failed: %d\n", ret);
		return ret;
	}

2501 2502 2503 2504 2505 2506 2507
	if (!nr_io_queues && prior_ioq_cnt) {
		dev_info(ctrl->ctrl.device,
			"Fail Reconnect: At least 1 io queue "
			"required (was %d)\n", prior_ioq_cnt);
		return -ENOSPC;
	}

2508
	ctrl->ctrl.queue_count = nr_io_queues + 1;
2509
	/* check for io queues existing */
2510
	if (ctrl->ctrl.queue_count == 1)
2511 2512
		return 0;

J
James Smart 已提交
2513
	ret = nvme_fc_create_hw_io_queues(ctrl, ctrl->ctrl.sqsize + 1);
2514
	if (ret)
2515
		goto out_free_io_queues;
2516

J
James Smart 已提交
2517
	ret = nvme_fc_connect_io_queues(ctrl, ctrl->ctrl.sqsize + 1);
2518 2519 2520
	if (ret)
		goto out_delete_hw_queues;

2521 2522 2523 2524
	if (prior_ioq_cnt != nr_io_queues)
		dev_info(ctrl->ctrl.device,
			"reconnect: revising io queue count from %d to %d\n",
			prior_ioq_cnt, nr_io_queues);
2525 2526
	blk_mq_update_nr_hw_queues(&ctrl->tag_set, nr_io_queues);

2527 2528 2529 2530
	return 0;

out_delete_hw_queues:
	nvme_fc_delete_hw_io_queues(ctrl);
2531
out_free_io_queues:
2532
	nvme_fc_free_io_queues(ctrl);
2533 2534
	return ret;
}
2535

2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566 2567 2568 2569 2570 2571 2572 2573 2574 2575 2576 2577 2578 2579 2580 2581 2582 2583 2584 2585 2586 2587 2588 2589 2590
static void
nvme_fc_rport_active_on_lport(struct nvme_fc_rport *rport)
{
	struct nvme_fc_lport *lport = rport->lport;

	atomic_inc(&lport->act_rport_cnt);
}

static void
nvme_fc_rport_inactive_on_lport(struct nvme_fc_rport *rport)
{
	struct nvme_fc_lport *lport = rport->lport;
	u32 cnt;

	cnt = atomic_dec_return(&lport->act_rport_cnt);
	if (cnt == 0 && lport->localport.port_state == FC_OBJSTATE_DELETED)
		lport->ops->localport_delete(&lport->localport);
}

static int
nvme_fc_ctlr_active_on_rport(struct nvme_fc_ctrl *ctrl)
{
	struct nvme_fc_rport *rport = ctrl->rport;
	u32 cnt;

	if (ctrl->assoc_active)
		return 1;

	ctrl->assoc_active = true;
	cnt = atomic_inc_return(&rport->act_ctrl_cnt);
	if (cnt == 1)
		nvme_fc_rport_active_on_lport(rport);

	return 0;
}

static int
nvme_fc_ctlr_inactive_on_rport(struct nvme_fc_ctrl *ctrl)
{
	struct nvme_fc_rport *rport = ctrl->rport;
	struct nvme_fc_lport *lport = rport->lport;
	u32 cnt;

	/* ctrl->assoc_active=false will be set independently */

	cnt = atomic_dec_return(&rport->act_ctrl_cnt);
	if (cnt == 0) {
		if (rport->remoteport.port_state == FC_OBJSTATE_DELETED)
			lport->ops->remoteport_delete(&rport->remoteport);
		nvme_fc_rport_inactive_on_lport(rport);
	}

	return 0;
}

2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601
/*
 * This routine restarts the controller on the host side, and
 * on the link side, recreates the controller association.
 */
static int
nvme_fc_create_association(struct nvme_fc_ctrl *ctrl)
{
	struct nvmf_ctrl_options *opts = ctrl->ctrl.opts;
	int ret;
	bool changed;

2602
	++ctrl->ctrl.nr_reconnects;
2603

2604 2605 2606
	if (ctrl->rport->remoteport.port_state != FC_OBJSTATE_ONLINE)
		return -ENODEV;

2607 2608 2609
	if (nvme_fc_ctlr_active_on_rport(ctrl))
		return -ENOTUNIQ;

2610 2611 2612 2613 2614
	/*
	 * Create the admin queue
	 */

	ret = __nvme_fc_create_hw_queue(ctrl, &ctrl->queues[0], 0,
J
James Smart 已提交
2615
				NVME_AQ_DEPTH);
2616 2617 2618 2619
	if (ret)
		goto out_free_queue;

	ret = nvme_fc_connect_admin_queue(ctrl, &ctrl->queues[0],
J
James Smart 已提交
2620
				NVME_AQ_DEPTH, (NVME_AQ_DEPTH / 4));
2621 2622 2623
	if (ret)
		goto out_delete_hw_queue;

2624
	blk_mq_unquiesce_queue(ctrl->ctrl.admin_q);
2625 2626 2627 2628 2629

	ret = nvmf_connect_admin_queue(&ctrl->ctrl);
	if (ret)
		goto out_disconnect_admin_queue;

2630 2631
	set_bit(NVME_FC_Q_LIVE, &ctrl->queues[0].flags);

2632 2633 2634 2635 2636 2637 2638
	/*
	 * Check controller capabilities
	 *
	 * todo:- add code to check if ctrl attributes changed from
	 * prior connection values
	 */

2639
	ret = nvmf_reg_read64(&ctrl->ctrl, NVME_REG_CAP, &ctrl->ctrl.cap);
2640 2641 2642 2643 2644 2645 2646
	if (ret) {
		dev_err(ctrl->ctrl.device,
			"prop_get NVME_REG_CAP failed\n");
		goto out_disconnect_admin_queue;
	}

	ctrl->ctrl.sqsize =
J
James Smart 已提交
2647
		min_t(int, NVME_CAP_MQES(ctrl->ctrl.cap), ctrl->ctrl.sqsize);
2648

2649
	ret = nvme_enable_ctrl(&ctrl->ctrl, ctrl->ctrl.cap);
2650 2651 2652
	if (ret)
		goto out_disconnect_admin_queue;

J
James Smart 已提交
2653 2654
	ctrl->ctrl.max_hw_sectors =
		(ctrl->lport->ops->max_sgl_segments - 1) << (PAGE_SHIFT - 9);
2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679

	ret = nvme_init_identify(&ctrl->ctrl);
	if (ret)
		goto out_disconnect_admin_queue;

	/* sanity checks */

	/* FC-NVME does not have other data in the capsule */
	if (ctrl->ctrl.icdoff) {
		dev_err(ctrl->ctrl.device, "icdoff %d is not supported!\n",
				ctrl->ctrl.icdoff);
		goto out_disconnect_admin_queue;
	}

	/* FC-NVME supports normal SGL Data Block Descriptors */

	if (opts->queue_size > ctrl->ctrl.maxcmd) {
		/* warn if maxcmd is lower than queue_size */
		dev_warn(ctrl->ctrl.device,
			"queue_size %zu > ctrl maxcmd %u, reducing "
			"to queue_size\n",
			opts->queue_size, ctrl->ctrl.maxcmd);
		opts->queue_size = ctrl->ctrl.maxcmd;
	}

J
James Smart 已提交
2680 2681 2682 2683 2684 2685 2686 2687
	if (opts->queue_size > ctrl->ctrl.sqsize + 1) {
		/* warn if sqsize is lower than queue_size */
		dev_warn(ctrl->ctrl.device,
			"queue_size %zu > ctrl sqsize %u, clamping down\n",
			opts->queue_size, ctrl->ctrl.sqsize + 1);
		opts->queue_size = ctrl->ctrl.sqsize + 1;
	}

2688 2689 2690 2691 2692 2693 2694 2695
	ret = nvme_fc_init_aen_ops(ctrl);
	if (ret)
		goto out_term_aen_ops;

	/*
	 * Create the io queues
	 */

2696
	if (ctrl->ctrl.queue_count > 1) {
2697
		if (!ctrl->ioq_live)
2698 2699
			ret = nvme_fc_create_io_queues(ctrl);
		else
2700
			ret = nvme_fc_recreate_io_queues(ctrl);
2701 2702 2703 2704 2705 2706
		if (ret)
			goto out_term_aen_ops;
	}

	changed = nvme_change_ctrl_state(&ctrl->ctrl, NVME_CTRL_LIVE);

2707
	ctrl->ctrl.nr_reconnects = 0;
2708

2709 2710
	if (changed)
		nvme_start_ctrl(&ctrl->ctrl);
2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722

	return 0;	/* Success */

out_term_aen_ops:
	nvme_fc_term_aen_ops(ctrl);
out_disconnect_admin_queue:
	/* send a Disconnect(association) LS to fc-nvme target */
	nvme_fc_xmt_disconnect_assoc(ctrl);
out_delete_hw_queue:
	__nvme_fc_delete_hw_queue(ctrl, &ctrl->queues[0], 0);
out_free_queue:
	nvme_fc_free_queue(&ctrl->queues[0]);
2723 2724
	ctrl->assoc_active = false;
	nvme_fc_ctlr_inactive_on_rport(ctrl);
2725 2726 2727 2728

	return ret;
}

2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739
/*
 * This routine stops operation of the controller on the host side.
 * On the host os stack side: Admin and IO queues are stopped,
 *   outstanding ios on them terminated via FC ABTS.
 * On the link side: the association is terminated.
 */
static void
nvme_fc_delete_association(struct nvme_fc_ctrl *ctrl)
{
	unsigned long flags;

2740 2741 2742 2743
	if (!ctrl->assoc_active)
		return;
	ctrl->assoc_active = false;

2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760
	spin_lock_irqsave(&ctrl->lock, flags);
	ctrl->flags |= FCCTRL_TERMIO;
	ctrl->iocnt = 0;
	spin_unlock_irqrestore(&ctrl->lock, flags);

	/*
	 * If io queues are present, stop them and terminate all outstanding
	 * ios on them. As FC allocates FC exchange for each io, the
	 * transport must contact the LLDD to terminate the exchange,
	 * thus releasing the FC exchange. We use blk_mq_tagset_busy_itr()
	 * to tell us what io's are busy and invoke a transport routine
	 * to kill them with the LLDD.  After terminating the exchange
	 * the LLDD will call the transport's normal io done path, but it
	 * will have an aborted status. The done path will return the
	 * io requests back to the block layer as part of normal completions
	 * (but with error status).
	 */
2761
	if (ctrl->ctrl.queue_count > 1) {
2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783
		nvme_stop_queues(&ctrl->ctrl);
		blk_mq_tagset_busy_iter(&ctrl->tag_set,
				nvme_fc_terminate_exchange, &ctrl->ctrl);
	}

	/*
	 * Other transports, which don't have link-level contexts bound
	 * to sqe's, would try to gracefully shutdown the controller by
	 * writing the registers for shutdown and polling (call
	 * nvme_shutdown_ctrl()). Given a bunch of i/o was potentially
	 * just aborted and we will wait on those contexts, and given
	 * there was no indication of how live the controlelr is on the
	 * link, don't send more io to create more contexts for the
	 * shutdown. Let the controller fail via keepalive failure if
	 * its still present.
	 */

	/*
	 * clean up the admin queue. Same thing as above.
	 * use blk_mq_tagset_busy_itr() and the transport routine to
	 * terminate the exchanges.
	 */
2784
	blk_mq_quiesce_queue(ctrl->ctrl.admin_q);
2785 2786 2787 2788 2789 2790 2791
	blk_mq_tagset_busy_iter(&ctrl->admin_tag_set,
				nvme_fc_terminate_exchange, &ctrl->ctrl);

	/* kill the aens as they are a separate path */
	nvme_fc_abort_aen_ops(ctrl);

	/* wait for all io that had to be aborted */
J
James Smart 已提交
2792
	spin_lock_irq(&ctrl->lock);
2793
	wait_event_lock_irq(ctrl->ioabort_wait, ctrl->iocnt == 0, ctrl->lock);
2794
	ctrl->flags &= ~FCCTRL_TERMIO;
J
James Smart 已提交
2795
	spin_unlock_irq(&ctrl->lock);
2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814

	nvme_fc_term_aen_ops(ctrl);

	/*
	 * send a Disconnect(association) LS to fc-nvme target
	 * Note: could have been sent at top of process, but
	 * cleaner on link traffic if after the aborts complete.
	 * Note: if association doesn't exist, association_id will be 0
	 */
	if (ctrl->association_id)
		nvme_fc_xmt_disconnect_assoc(ctrl);

	if (ctrl->ctrl.tagset) {
		nvme_fc_delete_hw_io_queues(ctrl);
		nvme_fc_free_io_queues(ctrl);
	}

	__nvme_fc_delete_hw_queue(ctrl, &ctrl->queues[0], 0);
	nvme_fc_free_queue(&ctrl->queues[0]);
2815

2816 2817 2818
	/* re-enable the admin_q so anything new can fast fail */
	blk_mq_unquiesce_queue(ctrl->ctrl.admin_q);

2819 2820 2821
	/* resume the io queues so that things will fast fail */
	nvme_start_queues(&ctrl->ctrl);

2822
	nvme_fc_ctlr_inactive_on_rport(ctrl);
2823 2824 2825
}

static void
2826
nvme_fc_delete_ctrl(struct nvme_ctrl *nctrl)
2827
{
2828
	struct nvme_fc_ctrl *ctrl = to_fc_ctrl(nctrl);
2829

2830
	cancel_work_sync(&ctrl->err_work);
2831 2832 2833 2834 2835 2836 2837 2838
	cancel_delayed_work_sync(&ctrl->connect_work);
	/*
	 * kill the association on the link side.  this will block
	 * waiting for io to terminate
	 */
	nvme_fc_delete_association(ctrl);
}

J
James Smart 已提交
2839 2840 2841
static void
nvme_fc_reconnect_or_delete(struct nvme_fc_ctrl *ctrl, int status)
{
2842 2843 2844 2845
	struct nvme_fc_rport *rport = ctrl->rport;
	struct nvme_fc_remote_port *portptr = &rport->remoteport;
	unsigned long recon_delay = ctrl->ctrl.opts->reconnect_delay * HZ;
	bool recon = true;
J
James Smart 已提交
2846

2847
	if (ctrl->ctrl.state != NVME_CTRL_CONNECTING)
J
James Smart 已提交
2848 2849
		return;

2850
	if (portptr->port_state == FC_OBJSTATE_ONLINE)
J
James Smart 已提交
2851
		dev_info(ctrl->ctrl.device,
2852 2853 2854 2855
			"NVME-FC{%d}: reset: Reconnect attempt failed (%d)\n",
			ctrl->cnum, status);
	else if (time_after_eq(jiffies, rport->dev_loss_end))
		recon = false;
J
James Smart 已提交
2856

2857 2858 2859 2860 2861 2862 2863 2864
	if (recon && nvmf_should_reconnect(&ctrl->ctrl)) {
		if (portptr->port_state == FC_OBJSTATE_ONLINE)
			dev_info(ctrl->ctrl.device,
				"NVME-FC{%d}: Reconnect attempt in %ld "
				"seconds\n",
				ctrl->cnum, recon_delay / HZ);
		else if (time_after(jiffies + recon_delay, rport->dev_loss_end))
			recon_delay = rport->dev_loss_end - jiffies;
2865

2866
		queue_delayed_work(nvme_wq, &ctrl->connect_work, recon_delay);
J
James Smart 已提交
2867
	} else {
2868 2869
		if (portptr->port_state == FC_OBJSTATE_ONLINE)
			dev_warn(ctrl->ctrl.device,
J
James Smart 已提交
2870
				"NVME-FC{%d}: Max reconnect attempts (%d) "
2871
				"reached.\n",
2872
				ctrl->cnum, ctrl->ctrl.nr_reconnects);
2873 2874 2875
		else
			dev_warn(ctrl->ctrl.device,
				"NVME-FC{%d}: dev_loss_tmo (%d) expired "
2876 2877
				"while waiting for remoteport connectivity.\n",
				ctrl->cnum, portptr->dev_loss_tmo);
2878
		WARN_ON(nvme_delete_ctrl(&ctrl->ctrl));
J
James Smart 已提交
2879 2880 2881
	}
}

2882
static void
2883
__nvme_fc_terminate_io(struct nvme_fc_ctrl *ctrl)
2884
{
2885
	nvme_stop_keep_alive(&ctrl->ctrl);
2886

2887 2888 2889
	/* will block will waiting for io to terminate */
	nvme_fc_delete_association(ctrl);

2890 2891
	if (ctrl->ctrl.state != NVME_CTRL_CONNECTING &&
	    !nvme_change_ctrl_state(&ctrl->ctrl, NVME_CTRL_CONNECTING))
2892 2893
		dev_err(ctrl->ctrl.device,
			"NVME-FC{%d}: error_recovery: Couldn't change state "
2894
			"to CONNECTING\n", ctrl->cnum);
2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906
}

static void
nvme_fc_reset_ctrl_work(struct work_struct *work)
{
	struct nvme_fc_ctrl *ctrl =
		container_of(work, struct nvme_fc_ctrl, ctrl.reset_work);
	int ret;

	__nvme_fc_terminate_io(ctrl);

	nvme_stop_ctrl(&ctrl->ctrl);
2907

2908
	if (ctrl->rport->remoteport.port_state == FC_OBJSTATE_ONLINE)
2909
		ret = nvme_fc_create_association(ctrl);
2910 2911 2912
	else
		ret = -ENOTCONN;

J
James Smart 已提交
2913 2914 2915
	if (ret)
		nvme_fc_reconnect_or_delete(ctrl, ret);
	else
2916
		dev_info(ctrl->ctrl.device,
2917 2918
			"NVME-FC{%d}: controller reset complete\n",
			ctrl->cnum);
2919 2920
}

2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 2931 2932 2933 2934 2935 2936 2937 2938
static void
nvme_fc_connect_err_work(struct work_struct *work)
{
	struct nvme_fc_ctrl *ctrl =
			container_of(work, struct nvme_fc_ctrl, err_work);

	__nvme_fc_terminate_io(ctrl);

	atomic_set(&ctrl->err_work_active, 0);

	/*
	 * Rescheduling the connection after recovering
	 * from the io error is left to the reconnect work
	 * item, which is what should have stalled waiting on
	 * the io that had the error that scheduled this work.
	 */
}

2939 2940 2941
static const struct nvme_ctrl_ops nvme_fc_ctrl_ops = {
	.name			= "fc",
	.module			= THIS_MODULE,
2942
	.flags			= NVME_F_FABRICS,
2943 2944 2945 2946 2947
	.reg_read32		= nvmf_reg_read32,
	.reg_read64		= nvmf_reg_read64,
	.reg_write32		= nvmf_reg_write32,
	.free_ctrl		= nvme_fc_nvme_ctrl_freed,
	.submit_async_event	= nvme_fc_submit_async_event,
2948
	.delete_ctrl		= nvme_fc_delete_ctrl,
2949 2950 2951 2952 2953 2954 2955 2956 2957 2958 2959 2960 2961
	.get_address		= nvmf_get_address,
};

static void
nvme_fc_connect_ctrl_work(struct work_struct *work)
{
	int ret;

	struct nvme_fc_ctrl *ctrl =
			container_of(to_delayed_work(work),
				struct nvme_fc_ctrl, connect_work);

	ret = nvme_fc_create_association(ctrl);
J
James Smart 已提交
2962 2963 2964
	if (ret)
		nvme_fc_reconnect_or_delete(ctrl, ret);
	else
2965
		dev_info(ctrl->ctrl.device,
2966
			"NVME-FC{%d}: controller connect complete\n",
2967 2968 2969 2970 2971 2972 2973
			ctrl->cnum);
}


static const struct blk_mq_ops nvme_fc_admin_mq_ops = {
	.queue_rq	= nvme_fc_queue_rq,
	.complete	= nvme_fc_complete_rq,
2974
	.init_request	= nvme_fc_init_request,
2975 2976 2977 2978 2979
	.exit_request	= nvme_fc_exit_request,
	.init_hctx	= nvme_fc_init_admin_hctx,
	.timeout	= nvme_fc_timeout,
};

2980

2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007
/*
 * Fails a controller request if it matches an existing controller
 * (association) with the same tuple:
 * <Host NQN, Host ID, local FC port, remote FC port, SUBSYS NQN>
 *
 * The ports don't need to be compared as they are intrinsically
 * already matched by the port pointers supplied.
 */
static bool
nvme_fc_existing_controller(struct nvme_fc_rport *rport,
		struct nvmf_ctrl_options *opts)
{
	struct nvme_fc_ctrl *ctrl;
	unsigned long flags;
	bool found = false;

	spin_lock_irqsave(&rport->lock, flags);
	list_for_each_entry(ctrl, &rport->ctrl_list, ctrl_list) {
		found = nvmf_ctlr_matches_baseopts(&ctrl->ctrl, opts);
		if (found)
			break;
	}
	spin_unlock_irqrestore(&rport->lock, flags);

	return found;
}

3008
static struct nvme_ctrl *
3009
nvme_fc_init_ctrl(struct device *dev, struct nvmf_ctrl_options *opts,
3010 3011 3012 3013
	struct nvme_fc_lport *lport, struct nvme_fc_rport *rport)
{
	struct nvme_fc_ctrl *ctrl;
	unsigned long flags;
3014
	int ret, idx;
3015

3016 3017 3018 3019 3020 3021
	if (!(rport->remoteport.port_role &
	    (FC_PORT_ROLE_NVME_DISCOVERY | FC_PORT_ROLE_NVME_TARGET))) {
		ret = -EBADR;
		goto out_fail;
	}

3022 3023 3024 3025 3026 3027
	if (!opts->duplicate_connect &&
	    nvme_fc_existing_controller(rport, opts)) {
		ret = -EALREADY;
		goto out_fail;
	}

3028 3029 3030 3031 3032 3033 3034 3035 3036 3037 3038 3039 3040
	ctrl = kzalloc(sizeof(*ctrl), GFP_KERNEL);
	if (!ctrl) {
		ret = -ENOMEM;
		goto out_fail;
	}

	idx = ida_simple_get(&nvme_fc_ctrl_cnt, 0, 0, GFP_KERNEL);
	if (idx < 0) {
		ret = -ENOSPC;
		goto out_free_ctrl;
	}

	ctrl->ctrl.opts = opts;
3041
	ctrl->ctrl.nr_reconnects = 0;
3042 3043 3044 3045
	if (lport->dev)
		ctrl->ctrl.numa_node = dev_to_node(lport->dev);
	else
		ctrl->ctrl.numa_node = NUMA_NO_NODE;
3046 3047 3048 3049 3050
	INIT_LIST_HEAD(&ctrl->ctrl_list);
	ctrl->lport = lport;
	ctrl->rport = rport;
	ctrl->dev = lport->dev;
	ctrl->cnum = idx;
3051
	ctrl->ioq_live = false;
3052
	ctrl->assoc_active = false;
3053
	atomic_set(&ctrl->err_work_active, 0);
J
James Smart 已提交
3054
	init_waitqueue_head(&ctrl->ioabort_wait);
3055 3056 3057 3058

	get_device(ctrl->dev);
	kref_init(&ctrl->ref);

3059
	INIT_WORK(&ctrl->ctrl.reset_work, nvme_fc_reset_ctrl_work);
3060
	INIT_DELAYED_WORK(&ctrl->connect_work, nvme_fc_connect_ctrl_work);
3061
	INIT_WORK(&ctrl->err_work, nvme_fc_connect_err_work);
3062 3063 3064
	spin_lock_init(&ctrl->lock);

	/* io queue count */
3065
	ctrl->ctrl.queue_count = min_t(unsigned int,
3066 3067
				opts->nr_io_queues,
				lport->ops->max_hw_queues);
3068
	ctrl->ctrl.queue_count++;	/* +1 for admin queue */
3069 3070 3071

	ctrl->ctrl.sqsize = opts->queue_size - 1;
	ctrl->ctrl.kato = opts->kato;
3072
	ctrl->ctrl.cntlid = 0xffff;
3073 3074

	ret = -ENOMEM;
3075 3076
	ctrl->queues = kcalloc(ctrl->ctrl.queue_count,
				sizeof(struct nvme_fc_queue), GFP_KERNEL);
3077
	if (!ctrl->queues)
3078
		goto out_free_ida;
3079

3080 3081
	nvme_fc_init_queue(ctrl, 0);

3082 3083
	memset(&ctrl->admin_tag_set, 0, sizeof(ctrl->admin_tag_set));
	ctrl->admin_tag_set.ops = &nvme_fc_admin_mq_ops;
K
Keith Busch 已提交
3084
	ctrl->admin_tag_set.queue_depth = NVME_AQ_MQ_TAG_DEPTH;
3085
	ctrl->admin_tag_set.reserved_tags = 2; /* fabric connect + Keep-Alive */
3086
	ctrl->admin_tag_set.numa_node = ctrl->ctrl.numa_node;
3087 3088 3089
	ctrl->admin_tag_set.cmd_size =
		struct_size((struct nvme_fcp_op_w_sgl *)NULL, priv,
			    ctrl->lport->ops->fcprqst_priv_sz);
3090 3091 3092
	ctrl->admin_tag_set.driver_data = ctrl;
	ctrl->admin_tag_set.nr_hw_queues = 1;
	ctrl->admin_tag_set.timeout = ADMIN_TIMEOUT;
3093
	ctrl->admin_tag_set.flags = BLK_MQ_F_NO_SCHED;
3094

3095
	ret = blk_mq_alloc_tag_set(&ctrl->admin_tag_set);
3096
	if (ret)
3097
		goto out_free_queues;
3098
	ctrl->ctrl.admin_tagset = &ctrl->admin_tag_set;
3099

3100 3101 3102 3103
	ctrl->ctrl.admin_q = blk_mq_init_queue(&ctrl->admin_tag_set);
	if (IS_ERR(ctrl->ctrl.admin_q)) {
		ret = PTR_ERR(ctrl->ctrl.admin_q);
		goto out_free_admin_tag_set;
3104 3105
	}

3106 3107 3108 3109 3110 3111
	/*
	 * Would have been nice to init io queues tag set as well.
	 * However, we require interaction from the controller
	 * for max io queue count before we can do so.
	 * Defer this to the connect path.
	 */
3112

3113 3114 3115
	ret = nvme_init_ctrl(&ctrl->ctrl, dev, &nvme_fc_ctrl_ops, 0);
	if (ret)
		goto out_cleanup_admin_q;
3116

3117
	/* at this point, teardown path changes to ref counting on nvme ctrl */
3118 3119 3120 3121 3122

	spin_lock_irqsave(&rport->lock, flags);
	list_add_tail(&ctrl->ctrl_list, &rport->ctrl_list);
	spin_unlock_irqrestore(&rport->lock, flags);

3123 3124
	if (!nvme_change_ctrl_state(&ctrl->ctrl, NVME_CTRL_RESETTING) ||
	    !nvme_change_ctrl_state(&ctrl->ctrl, NVME_CTRL_CONNECTING)) {
3125
		dev_err(ctrl->ctrl.device,
3126 3127 3128
			"NVME-FC{%d}: failed to init ctrl state\n", ctrl->cnum);
		goto fail_ctrl;
	}
3129

3130
	nvme_get_ctrl(&ctrl->ctrl);
3131

3132
	if (!queue_delayed_work(nvme_wq, &ctrl->connect_work, 0)) {
3133
		nvme_put_ctrl(&ctrl->ctrl);
3134 3135 3136 3137
		dev_err(ctrl->ctrl.device,
			"NVME-FC{%d}: failed to schedule initial connect\n",
			ctrl->cnum);
		goto fail_ctrl;
3138 3139
	}

3140
	flush_delayed_work(&ctrl->connect_work);
3141

3142 3143 3144
	dev_info(ctrl->ctrl.device,
		"NVME-FC{%d}: new ctrl: NQN \"%s\"\n",
		ctrl->cnum, ctrl->ctrl.opts->subsysnqn);
3145

3146
	return &ctrl->ctrl;
3147

3148 3149 3150
fail_ctrl:
	nvme_change_ctrl_state(&ctrl->ctrl, NVME_CTRL_DELETING);
	cancel_work_sync(&ctrl->ctrl.reset_work);
3151
	cancel_work_sync(&ctrl->err_work);
3152 3153 3154 3155 3156 3157 3158 3159 3160 3161 3162 3163 3164 3165 3166 3167 3168 3169 3170 3171 3172
	cancel_delayed_work_sync(&ctrl->connect_work);

	ctrl->ctrl.opts = NULL;

	/* initiate nvme ctrl ref counting teardown */
	nvme_uninit_ctrl(&ctrl->ctrl);

	/* Remove core ctrl ref. */
	nvme_put_ctrl(&ctrl->ctrl);

	/* as we're past the point where we transition to the ref
	 * counting teardown path, if we return a bad pointer here,
	 * the calling routine, thinking it's prior to the
	 * transition, will do an rport put. Since the teardown
	 * path also does a rport put, we do an extra get here to
	 * so proper order/teardown happens.
	 */
	nvme_fc_rport_get(rport);

	return ERR_PTR(-EIO);

3173 3174 3175 3176 3177 3178
out_cleanup_admin_q:
	blk_cleanup_queue(ctrl->ctrl.admin_q);
out_free_admin_tag_set:
	blk_mq_free_tag_set(&ctrl->admin_tag_set);
out_free_queues:
	kfree(ctrl->queues);
3179
out_free_ida:
3180
	put_device(ctrl->dev);
3181 3182 3183 3184 3185 3186 3187 3188 3189 3190 3191 3192 3193 3194 3195
	ida_simple_remove(&nvme_fc_ctrl_cnt, ctrl->cnum);
out_free_ctrl:
	kfree(ctrl);
out_fail:
	/* exit via here doesn't follow ctlr ref points */
	return ERR_PTR(ret);
}


struct nvmet_fc_traddr {
	u64	nn;
	u64	pn;
};

static int
J
James Smart 已提交
3196
__nvme_fc_parse_u64(substring_t *sstr, u64 *val)
3197 3198 3199
{
	u64 token64;

J
James Smart 已提交
3200 3201 3202
	if (match_u64(sstr, &token64))
		return -EINVAL;
	*val = token64;
3203

J
James Smart 已提交
3204 3205
	return 0;
}
3206

J
James Smart 已提交
3207 3208 3209 3210 3211 3212 3213 3214 3215 3216 3217 3218
/*
 * This routine validates and extracts the WWN's from the TRADDR string.
 * As kernel parsers need the 0x to determine number base, universally
 * build string to parse with 0x prefix before parsing name strings.
 */
static int
nvme_fc_parse_traddr(struct nvmet_fc_traddr *traddr, char *buf, size_t blen)
{
	char name[2 + NVME_FC_TRADDR_HEXNAMELEN + 1];
	substring_t wwn = { name, &name[sizeof(name)-1] };
	int nnoffset, pnoffset;

M
Milan P. Gandhi 已提交
3219
	/* validate if string is one of the 2 allowed formats */
J
James Smart 已提交
3220 3221 3222 3223 3224 3225 3226 3227 3228 3229 3230 3231 3232 3233 3234
	if (strnlen(buf, blen) == NVME_FC_TRADDR_MAXLENGTH &&
			!strncmp(buf, "nn-0x", NVME_FC_TRADDR_OXNNLEN) &&
			!strncmp(&buf[NVME_FC_TRADDR_MAX_PN_OFFSET],
				"pn-0x", NVME_FC_TRADDR_OXNNLEN)) {
		nnoffset = NVME_FC_TRADDR_OXNNLEN;
		pnoffset = NVME_FC_TRADDR_MAX_PN_OFFSET +
						NVME_FC_TRADDR_OXNNLEN;
	} else if ((strnlen(buf, blen) == NVME_FC_TRADDR_MINLENGTH &&
			!strncmp(buf, "nn-", NVME_FC_TRADDR_NNLEN) &&
			!strncmp(&buf[NVME_FC_TRADDR_MIN_PN_OFFSET],
				"pn-", NVME_FC_TRADDR_NNLEN))) {
		nnoffset = NVME_FC_TRADDR_NNLEN;
		pnoffset = NVME_FC_TRADDR_MIN_PN_OFFSET + NVME_FC_TRADDR_NNLEN;
	} else
		goto out_einval;
3235

J
James Smart 已提交
3236 3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250 3251 3252
	name[0] = '0';
	name[1] = 'x';
	name[2 + NVME_FC_TRADDR_HEXNAMELEN] = 0;

	memcpy(&name[2], &buf[nnoffset], NVME_FC_TRADDR_HEXNAMELEN);
	if (__nvme_fc_parse_u64(&wwn, &traddr->nn))
		goto out_einval;

	memcpy(&name[2], &buf[pnoffset], NVME_FC_TRADDR_HEXNAMELEN);
	if (__nvme_fc_parse_u64(&wwn, &traddr->pn))
		goto out_einval;

	return 0;

out_einval:
	pr_warn("%s: bad traddr string\n", __func__);
	return -EINVAL;
3253 3254 3255 3256 3257 3258 3259
}

static struct nvme_ctrl *
nvme_fc_create_ctrl(struct device *dev, struct nvmf_ctrl_options *opts)
{
	struct nvme_fc_lport *lport;
	struct nvme_fc_rport *rport;
3260
	struct nvme_ctrl *ctrl;
3261 3262 3263 3264 3265
	struct nvmet_fc_traddr laddr = { 0L, 0L };
	struct nvmet_fc_traddr raddr = { 0L, 0L };
	unsigned long flags;
	int ret;

J
James Smart 已提交
3266
	ret = nvme_fc_parse_traddr(&raddr, opts->traddr, NVMF_TRADDR_SIZE);
3267 3268 3269
	if (ret || !raddr.nn || !raddr.pn)
		return ERR_PTR(-EINVAL);

J
James Smart 已提交
3270
	ret = nvme_fc_parse_traddr(&laddr, opts->host_traddr, NVMF_TRADDR_SIZE);
3271 3272 3273 3274 3275 3276 3277 3278 3279 3280 3281 3282 3283 3284 3285 3286 3287 3288 3289 3290 3291
	if (ret || !laddr.nn || !laddr.pn)
		return ERR_PTR(-EINVAL);

	/* find the host and remote ports to connect together */
	spin_lock_irqsave(&nvme_fc_lock, flags);
	list_for_each_entry(lport, &nvme_fc_lport_list, port_list) {
		if (lport->localport.node_name != laddr.nn ||
		    lport->localport.port_name != laddr.pn)
			continue;

		list_for_each_entry(rport, &lport->endp_list, endp_list) {
			if (rport->remoteport.node_name != raddr.nn ||
			    rport->remoteport.port_name != raddr.pn)
				continue;

			/* if fail to get reference fall through. Will error */
			if (!nvme_fc_rport_get(rport))
				break;

			spin_unlock_irqrestore(&nvme_fc_lock, flags);

3292 3293 3294 3295
			ctrl = nvme_fc_init_ctrl(dev, opts, lport, rport);
			if (IS_ERR(ctrl))
				nvme_fc_rport_put(rport);
			return ctrl;
3296 3297 3298 3299
		}
	}
	spin_unlock_irqrestore(&nvme_fc_lock, flags);

3300 3301
	pr_warn("%s: %s - %s combination not found\n",
		__func__, opts->traddr, opts->host_traddr);
3302 3303 3304 3305 3306 3307
	return ERR_PTR(-ENOENT);
}


static struct nvmf_transport_ops nvme_fc_transport = {
	.name		= "fc",
3308
	.module		= THIS_MODULE,
3309
	.required_opts	= NVMF_OPT_TRADDR | NVMF_OPT_HOST_TRADDR,
J
James Smart 已提交
3310
	.allowed_opts	= NVMF_OPT_RECONNECT_DELAY | NVMF_OPT_CTRL_LOSS_TMO,
3311 3312 3313
	.create_ctrl	= nvme_fc_create_ctrl,
};

3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326 3327 3328 3329 3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340 3341 3342 3343 3344 3345 3346 3347 3348 3349 3350 3351 3352 3353 3354 3355 3356 3357 3358 3359 3360 3361 3362 3363 3364 3365 3366 3367 3368 3369 3370 3371 3372 3373 3374 3375 3376 3377 3378 3379 3380 3381 3382 3383 3384 3385 3386 3387 3388 3389 3390 3391 3392 3393 3394 3395 3396 3397
/* Arbitrary successive failures max. With lots of subsystems could be high */
#define DISCOVERY_MAX_FAIL	20

static ssize_t nvme_fc_nvme_discovery_store(struct device *dev,
		struct device_attribute *attr, const char *buf, size_t count)
{
	unsigned long flags;
	LIST_HEAD(local_disc_list);
	struct nvme_fc_lport *lport;
	struct nvme_fc_rport *rport;
	int failcnt = 0;

	spin_lock_irqsave(&nvme_fc_lock, flags);
restart:
	list_for_each_entry(lport, &nvme_fc_lport_list, port_list) {
		list_for_each_entry(rport, &lport->endp_list, endp_list) {
			if (!nvme_fc_lport_get(lport))
				continue;
			if (!nvme_fc_rport_get(rport)) {
				/*
				 * This is a temporary condition. Upon restart
				 * this rport will be gone from the list.
				 *
				 * Revert the lport put and retry.  Anything
				 * added to the list already will be skipped (as
				 * they are no longer list_empty).  Loops should
				 * resume at rports that were not yet seen.
				 */
				nvme_fc_lport_put(lport);

				if (failcnt++ < DISCOVERY_MAX_FAIL)
					goto restart;

				pr_err("nvme_discovery: too many reference "
				       "failures\n");
				goto process_local_list;
			}
			if (list_empty(&rport->disc_list))
				list_add_tail(&rport->disc_list,
					      &local_disc_list);
		}
	}

process_local_list:
	while (!list_empty(&local_disc_list)) {
		rport = list_first_entry(&local_disc_list,
					 struct nvme_fc_rport, disc_list);
		list_del_init(&rport->disc_list);
		spin_unlock_irqrestore(&nvme_fc_lock, flags);

		lport = rport->lport;
		/* signal discovery. Won't hurt if it repeats */
		nvme_fc_signal_discovery_scan(lport, rport);
		nvme_fc_rport_put(rport);
		nvme_fc_lport_put(lport);

		spin_lock_irqsave(&nvme_fc_lock, flags);
	}
	spin_unlock_irqrestore(&nvme_fc_lock, flags);

	return count;
}
static DEVICE_ATTR(nvme_discovery, 0200, NULL, nvme_fc_nvme_discovery_store);

static struct attribute *nvme_fc_attrs[] = {
	&dev_attr_nvme_discovery.attr,
	NULL
};

static struct attribute_group nvme_fc_attr_group = {
	.attrs = nvme_fc_attrs,
};

static const struct attribute_group *nvme_fc_attr_groups[] = {
	&nvme_fc_attr_group,
	NULL
};

static struct class fc_class = {
	.name = "fc",
	.dev_groups = nvme_fc_attr_groups,
	.owner = THIS_MODULE,
};

3398 3399
static int __init nvme_fc_init_module(void)
{
3400 3401
	int ret;

3402 3403 3404 3405
	nvme_fc_wq = alloc_workqueue("nvme_fc_wq", WQ_MEM_RECLAIM, 0);
	if (!nvme_fc_wq)
		return -ENOMEM;

3406 3407 3408 3409 3410 3411 3412 3413 3414 3415 3416 3417 3418 3419
	/*
	 * NOTE:
	 * It is expected that in the future the kernel will combine
	 * the FC-isms that are currently under scsi and now being
	 * added to by NVME into a new standalone FC class. The SCSI
	 * and NVME protocols and their devices would be under this
	 * new FC class.
	 *
	 * As we need something to post FC-specific udev events to,
	 * specifically for nvme probe events, start by creating the
	 * new device class.  When the new standalone FC class is
	 * put in place, this code will move to a more generic
	 * location for the class.
	 */
3420 3421
	ret = class_register(&fc_class);
	if (ret) {
3422
		pr_err("couldn't register class fc\n");
3423
		goto out_destroy_wq;
3424 3425 3426 3427 3428
	}

	/*
	 * Create a device for the FC-centric udev events
	 */
3429
	fc_udev_device = device_create(&fc_class, NULL, MKDEV(0, 0), NULL,
3430 3431 3432 3433 3434 3435 3436 3437 3438 3439 3440 3441 3442 3443
				"fc_udev_device");
	if (IS_ERR(fc_udev_device)) {
		pr_err("couldn't create fc_udev device!\n");
		ret = PTR_ERR(fc_udev_device);
		goto out_destroy_class;
	}

	ret = nvmf_register_transport(&nvme_fc_transport);
	if (ret)
		goto out_destroy_device;

	return 0;

out_destroy_device:
3444
	device_destroy(&fc_class, MKDEV(0, 0));
3445
out_destroy_class:
3446
	class_unregister(&fc_class);
3447 3448 3449
out_destroy_wq:
	destroy_workqueue(nvme_fc_wq);

3450
	return ret;
3451 3452 3453 3454 3455 3456 3457 3458 3459 3460 3461 3462
}

static void __exit nvme_fc_exit_module(void)
{
	/* sanity check - all lports should be removed */
	if (!list_empty(&nvme_fc_lport_list))
		pr_warn("%s: localport list not empty\n", __func__);

	nvmf_unregister_transport(&nvme_fc_transport);

	ida_destroy(&nvme_fc_local_port_cnt);
	ida_destroy(&nvme_fc_ctrl_cnt);
3463

3464 3465
	device_destroy(&fc_class, MKDEV(0, 0));
	class_unregister(&fc_class);
3466
	destroy_workqueue(nvme_fc_wq);
3467 3468 3469 3470 3471 3472
}

module_init(nvme_fc_init_module);
module_exit(nvme_fc_exit_module);

MODULE_LICENSE("GPL v2");