multipath.c 18.9 KB
Newer Older
1
// SPDX-License-Identifier: GPL-2.0
2
/*
C
Christoph Hellwig 已提交
3
 * Copyright (c) 2017-2018 Christoph Hellwig.
4 5
 */

6
#include <linux/backing-dev.h>
7
#include <linux/moduleparam.h>
8
#include <trace/events/block.h>
9 10 11
#include "nvme.h"

static bool multipath = true;
12
module_param(multipath, bool, 0444);
13 14 15
MODULE_PARM_DESC(multipath,
	"turn on native support for multiple controllers per subsystem");

16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45
void nvme_mpath_unfreeze(struct nvme_subsystem *subsys)
{
	struct nvme_ns_head *h;

	lockdep_assert_held(&subsys->lock);
	list_for_each_entry(h, &subsys->nsheads, entry)
		if (h->disk)
			blk_mq_unfreeze_queue(h->disk->queue);
}

void nvme_mpath_wait_freeze(struct nvme_subsystem *subsys)
{
	struct nvme_ns_head *h;

	lockdep_assert_held(&subsys->lock);
	list_for_each_entry(h, &subsys->nsheads, entry)
		if (h->disk)
			blk_mq_freeze_queue_wait(h->disk->queue);
}

void nvme_mpath_start_freeze(struct nvme_subsystem *subsys)
{
	struct nvme_ns_head *h;

	lockdep_assert_held(&subsys->lock);
	list_for_each_entry(h, &subsys->nsheads, entry)
		if (h->disk)
			blk_freeze_queue_start(h->disk->queue);
}

46 47 48 49 50 51 52 53 54 55 56 57 58 59
/*
 * If multipathing is enabled we need to always use the subsystem instance
 * number for numbering our devices to avoid conflicts between subsystems that
 * have multiple controllers and thus use the multipath-aware subsystem node
 * and those that have a single controller and use the controller node
 * directly.
 */
void nvme_set_disk_name(char *disk_name, struct nvme_ns *ns,
			struct nvme_ctrl *ctrl, int *flags)
{
	if (!multipath) {
		sprintf(disk_name, "nvme%dn%d", ctrl->instance, ns->head->instance);
	} else if (ns->head->disk) {
		sprintf(disk_name, "nvme%dc%dn%d", ctrl->subsys->instance,
60
				ctrl->instance, ns->head->instance);
61 62 63 64 65 66 67
		*flags = GENHD_FL_HIDDEN;
	} else {
		sprintf(disk_name, "nvme%dn%d", ctrl->subsys->instance,
				ns->head->instance);
	}
}

68
bool nvme_failover_req(struct request *req)
69 70
{
	struct nvme_ns *ns = req->q->queuedata;
C
Christoph Hellwig 已提交
71
	u16 status = nvme_req(req)->status;
72 73
	unsigned long flags;

C
Christoph Hellwig 已提交
74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92
	switch (status & 0x7ff) {
	case NVME_SC_ANA_TRANSITION:
	case NVME_SC_ANA_INACCESSIBLE:
	case NVME_SC_ANA_PERSISTENT_LOSS:
		/*
		 * If we got back an ANA error we know the controller is alive,
		 * but not ready to serve this namespaces.  The spec suggests
		 * we should update our general state here, but due to the fact
		 * that the admin and I/O queues are not serialized that is
		 * fundamentally racy.  So instead just clear the current path,
		 * mark the the path as pending and kick of a re-read of the ANA
		 * log page ASAP.
		 */
		nvme_mpath_clear_current_path(ns);
		if (ns->ctrl->ana_log_buf) {
			set_bit(NVME_NS_ANA_PENDING, &ns->flags);
			queue_work(nvme_wq, &ns->ctrl->ana_work);
		}
		break;
93
	case NVME_SC_HOST_PATH_ERROR:
94
	case NVME_SC_HOST_ABORTED_CMD:
95 96 97 98 99 100
		/*
		 * Temporary transport disruption in talking to the controller.
		 * Try to send on a new path.
		 */
		nvme_mpath_clear_current_path(ns);
		break;
C
Christoph Hellwig 已提交
101
	default:
102 103
		/* This was a non-ANA error so follow the normal error path. */
		return false;
C
Christoph Hellwig 已提交
104 105
	}

106 107 108 109 110
	spin_lock_irqsave(&ns->head->requeue_lock, flags);
	blk_steal_bios(&ns->head->requeue_list, req);
	spin_unlock_irqrestore(&ns->head->requeue_lock, flags);
	blk_mq_end_request(req, 0);

111
	kblockd_schedule_work(&ns->head->requeue_work);
112
	return true;
113 114 115 116 117 118
}

void nvme_kick_requeue_lists(struct nvme_ctrl *ctrl)
{
	struct nvme_ns *ns;

119
	down_read(&ctrl->namespaces_rwsem);
120 121 122 123
	list_for_each_entry(ns, &ctrl->namespaces, list) {
		if (ns->head->disk)
			kblockd_schedule_work(&ns->head->requeue_work);
	}
124
	up_read(&ctrl->namespaces_rwsem);
125 126
}

C
Christoph Hellwig 已提交
127 128 129 130 131 132 133 134 135
static const char *nvme_ana_state_names[] = {
	[0]				= "invalid state",
	[NVME_ANA_OPTIMIZED]		= "optimized",
	[NVME_ANA_NONOPTIMIZED]		= "non-optimized",
	[NVME_ANA_INACCESSIBLE]		= "inaccessible",
	[NVME_ANA_PERSISTENT_LOSS]	= "persistent-loss",
	[NVME_ANA_CHANGE]		= "change",
};

136
bool nvme_mpath_clear_current_path(struct nvme_ns *ns)
137
{
138
	struct nvme_ns_head *head = ns->head;
139
	bool changed = false;
140 141 142
	int node;

	if (!head)
143
		goto out;
144 145

	for_each_node(node) {
146
		if (ns == rcu_access_pointer(head->current_path[node])) {
147
			rcu_assign_pointer(head->current_path[node], NULL);
148 149
			changed = true;
		}
150
	}
151 152 153 154 155 156 157 158 159
out:
	return changed;
}

void nvme_mpath_clear_ctrl_paths(struct nvme_ctrl *ctrl)
{
	struct nvme_ns *ns;

	mutex_lock(&ctrl->scan_lock);
160
	down_read(&ctrl->namespaces_rwsem);
161 162 163
	list_for_each_entry(ns, &ctrl->namespaces, list)
		if (nvme_mpath_clear_current_path(ns))
			kblockd_schedule_work(&ns->head->requeue_work);
164
	up_read(&ctrl->namespaces_rwsem);
165
	mutex_unlock(&ctrl->scan_lock);
166 167
}

168 169 170
static bool nvme_path_is_disabled(struct nvme_ns *ns)
{
	return ns->ctrl->state != NVME_CTRL_LIVE ||
171 172
		test_bit(NVME_NS_ANA_PENDING, &ns->flags) ||
		test_bit(NVME_NS_REMOVING, &ns->flags);
173 174
}

175 176 177 178
static struct nvme_ns *__nvme_find_path(struct nvme_ns_head *head, int node)
{
	int found_distance = INT_MAX, fallback_distance = INT_MAX, distance;
	struct nvme_ns *found = NULL, *fallback = NULL, *ns;
179 180

	list_for_each_entry_rcu(ns, &head->list, siblings) {
181
		if (nvme_path_is_disabled(ns))
C
Christoph Hellwig 已提交
182
			continue;
183

184 185 186 187
		if (READ_ONCE(head->subsys->iopolicy) == NVME_IOPOLICY_NUMA)
			distance = node_distance(node, ns->ctrl->numa_node);
		else
			distance = LOCAL_DISTANCE;
188

C
Christoph Hellwig 已提交
189 190
		switch (ns->ana_state) {
		case NVME_ANA_OPTIMIZED:
191 192 193 194 195
			if (distance < found_distance) {
				found_distance = distance;
				found = ns;
			}
			break;
C
Christoph Hellwig 已提交
196
		case NVME_ANA_NONOPTIMIZED:
197 198 199 200
			if (distance < fallback_distance) {
				fallback_distance = distance;
				fallback = ns;
			}
C
Christoph Hellwig 已提交
201 202 203
			break;
		default:
			break;
204 205 206
		}
	}

207 208 209 210 211
	if (!found)
		found = fallback;
	if (found)
		rcu_assign_pointer(head->current_path[node], found);
	return found;
C
Christoph Hellwig 已提交
212 213
}

214 215 216 217 218 219 220 221 222 223 224 225 226 227 228
static struct nvme_ns *nvme_next_ns(struct nvme_ns_head *head,
		struct nvme_ns *ns)
{
	ns = list_next_or_null_rcu(&head->list, &ns->siblings, struct nvme_ns,
			siblings);
	if (ns)
		return ns;
	return list_first_or_null_rcu(&head->list, struct nvme_ns, siblings);
}

static struct nvme_ns *nvme_round_robin_path(struct nvme_ns_head *head,
		int node, struct nvme_ns *old)
{
	struct nvme_ns *ns, *found, *fallback = NULL;

229 230 231
	if (list_is_singular(&head->list)) {
		if (nvme_path_is_disabled(old))
			return NULL;
232
		return old;
233
	}
234 235 236 237

	for (ns = nvme_next_ns(head, old);
	     ns != old;
	     ns = nvme_next_ns(head, ns)) {
238
		if (nvme_path_is_disabled(ns))
239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256
			continue;

		if (ns->ana_state == NVME_ANA_OPTIMIZED) {
			found = ns;
			goto out;
		}
		if (ns->ana_state == NVME_ANA_NONOPTIMIZED)
			fallback = ns;
	}

	if (!fallback)
		return NULL;
	found = fallback;
out:
	rcu_assign_pointer(head->current_path[node], found);
	return found;
}

C
Christoph Hellwig 已提交
257 258 259 260
static inline bool nvme_path_is_optimized(struct nvme_ns *ns)
{
	return ns->ctrl->state == NVME_CTRL_LIVE &&
		ns->ana_state == NVME_ANA_OPTIMIZED;
261 262 263 264
}

inline struct nvme_ns *nvme_find_path(struct nvme_ns_head *head)
{
265 266
	int node = numa_node_id();
	struct nvme_ns *ns;
267

268
	ns = srcu_dereference(head->current_path[node], &head->srcu);
269 270
	if (READ_ONCE(head->subsys->iopolicy) == NVME_IOPOLICY_RR && ns)
		ns = nvme_round_robin_path(head, node, ns);
C
Christoph Hellwig 已提交
271
	if (unlikely(!ns || !nvme_path_is_optimized(ns)))
272
		ns = __nvme_find_path(head, node);
273 274 275
	return ns;
}

276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293
static bool nvme_available_path(struct nvme_ns_head *head)
{
	struct nvme_ns *ns;

	list_for_each_entry_rcu(ns, &head->list, siblings) {
		switch (ns->ctrl->state) {
		case NVME_CTRL_LIVE:
		case NVME_CTRL_RESETTING:
		case NVME_CTRL_CONNECTING:
			/* fallthru */
			return true;
		default:
			break;
		}
	}
	return false;
}

294 295 296
static blk_qc_t nvme_ns_head_make_request(struct request_queue *q,
		struct bio *bio)
{
297
	struct nvme_ns_head *head = bio->bi_disk->private_data;
298 299 300 301 302
	struct device *dev = disk_to_dev(head->disk);
	struct nvme_ns *ns;
	blk_qc_t ret = BLK_QC_T_NONE;
	int srcu_idx;

303 304 305 306 307 308 309 310
	/*
	 * The namespace might be going away and the bio might
	 * be moved to a different queue via blk_steal_bios(),
	 * so we need to use the bio_split pool from the original
	 * queue to allocate the bvecs from.
	 */
	blk_queue_split(q, &bio);

311 312 313 314 315
	srcu_idx = srcu_read_lock(&head->srcu);
	ns = nvme_find_path(head);
	if (likely(ns)) {
		bio->bi_disk = ns->disk;
		bio->bi_opf |= REQ_NVME_MPATH;
316 317 318
		trace_block_bio_remap(bio->bi_disk->queue, bio,
				      disk_devt(ns->head->disk),
				      bio->bi_iter.bi_sector);
319
		ret = direct_make_request(bio);
320 321
	} else if (nvme_available_path(head)) {
		dev_warn_ratelimited(dev, "no usable path - requeuing I/O\n");
322 323 324 325 326

		spin_lock_irq(&head->requeue_lock);
		bio_list_add(&head->requeue_list, bio);
		spin_unlock_irq(&head->requeue_lock);
	} else {
327
		dev_warn_ratelimited(dev, "no available path - failing I/O\n");
328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364

		bio->bi_status = BLK_STS_IOERR;
		bio_endio(bio);
	}

	srcu_read_unlock(&head->srcu, srcu_idx);
	return ret;
}

static void nvme_requeue_work(struct work_struct *work)
{
	struct nvme_ns_head *head =
		container_of(work, struct nvme_ns_head, requeue_work);
	struct bio *bio, *next;

	spin_lock_irq(&head->requeue_lock);
	next = bio_list_get(&head->requeue_list);
	spin_unlock_irq(&head->requeue_lock);

	while ((bio = next) != NULL) {
		next = bio->bi_next;
		bio->bi_next = NULL;

		/*
		 * Reset disk to the mpath node and resubmit to select a new
		 * path.
		 */
		bio->bi_disk = head->disk;
		generic_make_request(bio);
	}
}

int nvme_mpath_alloc_disk(struct nvme_ctrl *ctrl, struct nvme_ns_head *head)
{
	struct request_queue *q;
	bool vwc = false;

C
Christoph Hellwig 已提交
365
	mutex_init(&head->lock);
366 367 368 369 370 371 372 373 374
	bio_list_init(&head->requeue_list);
	spin_lock_init(&head->requeue_lock);
	INIT_WORK(&head->requeue_work, nvme_requeue_work);

	/*
	 * Add a multipath node if the subsystems supports multiple controllers.
	 * We also do this for private namespaces as the namespace sharing data could
	 * change after a rescan.
	 */
375
	if (!(ctrl->subsys->cmic & NVME_CTRL_CMIC_MULTI_CTRL) || !multipath)
376 377
		return 0;

378
	q = blk_alloc_queue(nvme_ns_head_make_request, ctrl->numa_node);
379 380
	if (!q)
		goto out;
381
	blk_queue_flag_set(QUEUE_FLAG_NONROT, q);
382 383
	/* set to a default value for 512 until disk is validated */
	blk_queue_logical_block_size(q, 512);
384
	blk_set_stacking_limits(&q->limits);
385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407

	/* we need to propagate up the VMC settings */
	if (ctrl->vwc & NVME_CTRL_VWC_PRESENT)
		vwc = true;
	blk_queue_write_cache(q, vwc, vwc);

	head->disk = alloc_disk(0);
	if (!head->disk)
		goto out_cleanup_queue;
	head->disk->fops = &nvme_ns_head_ops;
	head->disk->private_data = head;
	head->disk->queue = q;
	head->disk->flags = GENHD_FL_EXT_DEVT;
	sprintf(head->disk->disk_name, "nvme%dn%d",
			ctrl->subsys->instance, head->instance);
	return 0;

out_cleanup_queue:
	blk_cleanup_queue(q);
out:
	return -ENOMEM;
}

C
Christoph Hellwig 已提交
408
static void nvme_mpath_set_live(struct nvme_ns *ns)
409
{
C
Christoph Hellwig 已提交
410 411
	struct nvme_ns_head *head = ns->head;

412 413
	if (!head->disk)
		return;
414

415
	mutex_lock(&head->lock);
416 417 418
	if (!(head->disk->flags & GENHD_FL_UP))
		device_add_disk(&head->subsys->dev, head->disk,
				nvme_ns_id_attr_groups);
C
Christoph Hellwig 已提交
419

420 421 422 423 424 425 426 427
	if (nvme_path_is_optimized(ns)) {
		int node, srcu_idx;

		srcu_idx = srcu_read_lock(&head->srcu);
		for_each_node(node)
			__nvme_find_path(head, node);
		srcu_read_unlock(&head->srcu, srcu_idx);
	}
428
	mutex_unlock(&head->lock);
429

430 431
	synchronize_srcu(&head->srcu);
	kblockd_schedule_work(&head->requeue_work);
C
Christoph Hellwig 已提交
432 433 434 435 436 437 438 439 440 441 442 443 444 445
}

static int nvme_parse_ana_log(struct nvme_ctrl *ctrl, void *data,
		int (*cb)(struct nvme_ctrl *ctrl, struct nvme_ana_group_desc *,
			void *))
{
	void *base = ctrl->ana_log_buf;
	size_t offset = sizeof(struct nvme_ana_rsp_hdr);
	int error, i;

	lockdep_assert_held(&ctrl->ana_lock);

	for (i = 0; i < le16_to_cpu(ctrl->ana_log_buf->ngrps); i++) {
		struct nvme_ana_group_desc *desc = base + offset;
446 447 448 449 450 451 452 453
		u32 nr_nsids;
		size_t nsid_buf_size;

		if (WARN_ON_ONCE(offset > ctrl->ana_log_size - sizeof(*desc)))
			return -EINVAL;

		nr_nsids = le32_to_cpu(desc->nnsids);
		nsid_buf_size = nr_nsids * sizeof(__le32);
C
Christoph Hellwig 已提交
454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489

		if (WARN_ON_ONCE(desc->grpid == 0))
			return -EINVAL;
		if (WARN_ON_ONCE(le32_to_cpu(desc->grpid) > ctrl->anagrpmax))
			return -EINVAL;
		if (WARN_ON_ONCE(desc->state == 0))
			return -EINVAL;
		if (WARN_ON_ONCE(desc->state > NVME_ANA_CHANGE))
			return -EINVAL;

		offset += sizeof(*desc);
		if (WARN_ON_ONCE(offset > ctrl->ana_log_size - nsid_buf_size))
			return -EINVAL;

		error = cb(ctrl, desc, data);
		if (error)
			return error;

		offset += nsid_buf_size;
	}

	return 0;
}

static inline bool nvme_state_is_live(enum nvme_ana_state state)
{
	return state == NVME_ANA_OPTIMIZED || state == NVME_ANA_NONOPTIMIZED;
}

static void nvme_update_ns_ana_state(struct nvme_ana_group_desc *desc,
		struct nvme_ns *ns)
{
	ns->ana_grpid = le32_to_cpu(desc->grpid);
	ns->ana_state = desc->state;
	clear_bit(NVME_NS_ANA_PENDING, &ns->flags);

490
	if (nvme_state_is_live(ns->ana_state))
C
Christoph Hellwig 已提交
491 492 493 494 495 496 497 498 499 500
		nvme_mpath_set_live(ns);
}

static int nvme_update_ana_state(struct nvme_ctrl *ctrl,
		struct nvme_ana_group_desc *desc, void *data)
{
	u32 nr_nsids = le32_to_cpu(desc->nnsids), n = 0;
	unsigned *nr_change_groups = data;
	struct nvme_ns *ns;

501
	dev_dbg(ctrl->device, "ANA group %d: %s.\n",
C
Christoph Hellwig 已提交
502 503 504 505 506 507 508 509 510
			le32_to_cpu(desc->grpid),
			nvme_ana_state_names[desc->state]);

	if (desc->state == NVME_ANA_CHANGE)
		(*nr_change_groups)++;

	if (!nr_nsids)
		return 0;

511
	down_read(&ctrl->namespaces_rwsem);
C
Christoph Hellwig 已提交
512
	list_for_each_entry(ns, &ctrl->namespaces, list) {
513 514 515
		unsigned nsid = le32_to_cpu(desc->nsids[n]);

		if (ns->head->ns_id < nsid)
C
Christoph Hellwig 已提交
516
			continue;
517 518
		if (ns->head->ns_id == nsid)
			nvme_update_ns_ana_state(desc, ns);
C
Christoph Hellwig 已提交
519 520 521
		if (++n == nr_nsids)
			break;
	}
522
	up_read(&ctrl->namespaces_rwsem);
C
Christoph Hellwig 已提交
523 524 525
	return 0;
}

526
static int nvme_read_ana_log(struct nvme_ctrl *ctrl)
C
Christoph Hellwig 已提交
527 528 529 530 531
{
	u32 nr_change_groups = 0;
	int error;

	mutex_lock(&ctrl->ana_lock);
532
	error = nvme_get_log(ctrl, NVME_NSID_ALL, NVME_LOG_ANA, 0,
C
Christoph Hellwig 已提交
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
			ctrl->ana_log_buf, ctrl->ana_log_size, 0);
	if (error) {
		dev_warn(ctrl->device, "Failed to get ANA log: %d\n", error);
		goto out_unlock;
	}

	error = nvme_parse_ana_log(ctrl, &nr_change_groups,
			nvme_update_ana_state);
	if (error)
		goto out_unlock;

	/*
	 * In theory we should have an ANATT timer per group as they might enter
	 * the change state at different times.  But that is a lot of overhead
	 * just to protect against a target that keeps entering new changes
	 * states while never finishing previous ones.  But we'll still
	 * eventually time out once all groups are in change state, so this
	 * isn't a big deal.
	 *
	 * We also double the ANATT value to provide some slack for transports
	 * or AEN processing overhead.
	 */
	if (nr_change_groups)
		mod_timer(&ctrl->anatt_timer, ctrl->anatt * HZ * 2 + jiffies);
	else
		del_timer_sync(&ctrl->anatt_timer);
out_unlock:
	mutex_unlock(&ctrl->ana_lock);
	return error;
}

static void nvme_ana_work(struct work_struct *work)
{
	struct nvme_ctrl *ctrl = container_of(work, struct nvme_ctrl, ana_work);

568
	nvme_read_ana_log(ctrl);
C
Christoph Hellwig 已提交
569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586
}

static void nvme_anatt_timeout(struct timer_list *t)
{
	struct nvme_ctrl *ctrl = from_timer(ctrl, t, anatt_timer);

	dev_info(ctrl->device, "ANATT timeout, resetting controller.\n");
	nvme_reset_ctrl(ctrl);
}

void nvme_mpath_stop(struct nvme_ctrl *ctrl)
{
	if (!nvme_ctrl_use_ana(ctrl))
		return;
	del_timer_sync(&ctrl->anatt_timer);
	cancel_work_sync(&ctrl->ana_work);
}

587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624
#define SUBSYS_ATTR_RW(_name, _mode, _show, _store)  \
	struct device_attribute subsys_attr_##_name =	\
		__ATTR(_name, _mode, _show, _store)

static const char *nvme_iopolicy_names[] = {
	[NVME_IOPOLICY_NUMA]	= "numa",
	[NVME_IOPOLICY_RR]	= "round-robin",
};

static ssize_t nvme_subsys_iopolicy_show(struct device *dev,
		struct device_attribute *attr, char *buf)
{
	struct nvme_subsystem *subsys =
		container_of(dev, struct nvme_subsystem, dev);

	return sprintf(buf, "%s\n",
			nvme_iopolicy_names[READ_ONCE(subsys->iopolicy)]);
}

static ssize_t nvme_subsys_iopolicy_store(struct device *dev,
		struct device_attribute *attr, const char *buf, size_t count)
{
	struct nvme_subsystem *subsys =
		container_of(dev, struct nvme_subsystem, dev);
	int i;

	for (i = 0; i < ARRAY_SIZE(nvme_iopolicy_names); i++) {
		if (sysfs_streq(buf, nvme_iopolicy_names[i])) {
			WRITE_ONCE(subsys->iopolicy, i);
			return count;
		}
	}

	return -EINVAL;
}
SUBSYS_ATTR_RW(iopolicy, S_IRUGO | S_IWUSR,
		      nvme_subsys_iopolicy_show, nvme_subsys_iopolicy_store);

C
Christoph Hellwig 已提交
625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640
static ssize_t ana_grpid_show(struct device *dev, struct device_attribute *attr,
		char *buf)
{
	return sprintf(buf, "%d\n", nvme_get_ns_from_dev(dev)->ana_grpid);
}
DEVICE_ATTR_RO(ana_grpid);

static ssize_t ana_state_show(struct device *dev, struct device_attribute *attr,
		char *buf)
{
	struct nvme_ns *ns = nvme_get_ns_from_dev(dev);

	return sprintf(buf, "%s\n", nvme_ana_state_names[ns->ana_state]);
}
DEVICE_ATTR_RO(ana_state);

641
static int nvme_lookup_ana_group_desc(struct nvme_ctrl *ctrl,
C
Christoph Hellwig 已提交
642 643
		struct nvme_ana_group_desc *desc, void *data)
{
644
	struct nvme_ana_group_desc *dst = data;
C
Christoph Hellwig 已提交
645

646 647
	if (desc->grpid != dst->grpid)
		return 0;
C
Christoph Hellwig 已提交
648

649 650
	*dst = *desc;
	return -ENXIO; /* just break out of the loop */
C
Christoph Hellwig 已提交
651 652 653 654 655
}

void nvme_mpath_add_disk(struct nvme_ns *ns, struct nvme_id_ns *id)
{
	if (nvme_ctrl_use_ana(ns->ctrl)) {
656 657 658 659 660
		struct nvme_ana_group_desc desc = {
			.grpid = id->anagrpid,
			.state = 0,
		};

C
Christoph Hellwig 已提交
661 662
		mutex_lock(&ns->ctrl->ana_lock);
		ns->ana_grpid = le32_to_cpu(id->anagrpid);
663
		nvme_parse_ana_log(ns->ctrl, &desc, nvme_lookup_ana_group_desc);
C
Christoph Hellwig 已提交
664
		mutex_unlock(&ns->ctrl->ana_lock);
665 666 667 668
		if (desc.state) {
			/* found the group desc: update */
			nvme_update_ns_ana_state(&desc, ns);
		}
C
Christoph Hellwig 已提交
669 670 671
	} else {
		ns->ana_state = NVME_ANA_OPTIMIZED; 
		nvme_mpath_set_live(ns);
672
	}
673 674 675 676 677 678 679

	if (bdi_cap_stable_pages_required(ns->queue->backing_dev_info)) {
		struct backing_dev_info *info =
					ns->head->disk->queue->backing_dev_info;

		info->capabilities |= BDI_CAP_STABLE_WRITES;
	}
680 681 682 683 684 685
}

void nvme_mpath_remove_disk(struct nvme_ns_head *head)
{
	if (!head->disk)
		return;
686
	if (head->disk->flags & GENHD_FL_UP)
C
Christoph Hellwig 已提交
687
		del_gendisk(head->disk);
688 689 690 691 692 693 694
	blk_set_queue_dying(head->disk->queue);
	/* make sure all pending bios are cleaned up */
	kblockd_schedule_work(&head->requeue_work);
	flush_work(&head->requeue_work);
	blk_cleanup_queue(head->disk->queue);
	put_disk(head->disk);
}
C
Christoph Hellwig 已提交
695 696 697 698 699

int nvme_mpath_init(struct nvme_ctrl *ctrl, struct nvme_id_ctrl *id)
{
	int error;

700
	/* check if multipath is enabled and we have the capability */
701 702
	if (!multipath || !ctrl->subsys ||
	    !(ctrl->subsys->cmic & NVME_CTRL_CMIC_ANA))
C
Christoph Hellwig 已提交
703 704 705 706 707 708 709 710 711 712 713
		return 0;

	ctrl->anacap = id->anacap;
	ctrl->anatt = id->anatt;
	ctrl->nanagrpid = le32_to_cpu(id->nanagrpid);
	ctrl->anagrpmax = le32_to_cpu(id->anagrpmax);

	mutex_init(&ctrl->ana_lock);
	timer_setup(&ctrl->anatt_timer, nvme_anatt_timeout, 0);
	ctrl->ana_log_size = sizeof(struct nvme_ana_rsp_hdr) +
		ctrl->nanagrpid * sizeof(struct nvme_ana_group_desc);
714
	ctrl->ana_log_size += ctrl->max_namespaces * sizeof(__le32);
C
Christoph Hellwig 已提交
715 716 717 718 719 720 721 722 723 724 725

	if (ctrl->ana_log_size > ctrl->max_hw_sectors << SECTOR_SHIFT) {
		dev_err(ctrl->device,
			"ANA log page size (%zd) larger than MDTS (%d).\n",
			ctrl->ana_log_size,
			ctrl->max_hw_sectors << SECTOR_SHIFT);
		dev_err(ctrl->device, "disabling ANA support.\n");
		return 0;
	}

	INIT_WORK(&ctrl->ana_work, nvme_ana_work);
726
	kfree(ctrl->ana_log_buf);
C
Christoph Hellwig 已提交
727
	ctrl->ana_log_buf = kmalloc(ctrl->ana_log_size, GFP_KERNEL);
728 729
	if (!ctrl->ana_log_buf) {
		error = -ENOMEM;
C
Christoph Hellwig 已提交
730
		goto out;
731
	}
C
Christoph Hellwig 已提交
732

733
	error = nvme_read_ana_log(ctrl);
C
Christoph Hellwig 已提交
734 735 736 737 738
	if (error)
		goto out_free_ana_log_buf;
	return 0;
out_free_ana_log_buf:
	kfree(ctrl->ana_log_buf);
739
	ctrl->ana_log_buf = NULL;
C
Christoph Hellwig 已提交
740
out:
741
	return error;
C
Christoph Hellwig 已提交
742 743 744 745 746
}

void nvme_mpath_uninit(struct nvme_ctrl *ctrl)
{
	kfree(ctrl->ana_log_buf);
747
	ctrl->ana_log_buf = NULL;
C
Christoph Hellwig 已提交
748 749
}