core.c 121.7 KB
Newer Older
1
// SPDX-License-Identifier: GPL-2.0
2 3 4 5 6 7 8
/*
 * NVM Express device driver
 * Copyright (c) 2011-2014, Intel Corporation.
 */

#include <linux/blkdev.h>
#include <linux/blk-mq.h>
9
#include <linux/compat.h>
10
#include <linux/delay.h>
11
#include <linux/errno.h>
12
#include <linux/hdreg.h>
13
#include <linux/kernel.h>
14
#include <linux/module.h>
15
#include <linux/backing-dev.h>
16
#include <linux/list_sort.h>
17 18
#include <linux/slab.h>
#include <linux/types.h>
19 20 21
#include <linux/pr.h>
#include <linux/ptrace.h>
#include <linux/nvme_ioctl.h>
22
#include <linux/pm_qos.h>
23
#include <asm/unaligned.h>
24 25

#include "nvme.h"
S
Sagi Grimberg 已提交
26
#include "fabrics.h"
27

H
Hannes Reinecke 已提交
28 29 30
#define CREATE_TRACE_POINTS
#include "trace.h"

31 32
#define NVME_MINORS		(1U << MINORBITS)

33 34
unsigned int admin_timeout = 60;
module_param(admin_timeout, uint, 0644);
35
MODULE_PARM_DESC(admin_timeout, "timeout in seconds for admin commands");
36
EXPORT_SYMBOL_GPL(admin_timeout);
37

38 39
unsigned int nvme_io_timeout = 30;
module_param_named(io_timeout, nvme_io_timeout, uint, 0644);
40
MODULE_PARM_DESC(io_timeout, "timeout in seconds for I/O");
41
EXPORT_SYMBOL_GPL(nvme_io_timeout);
42

43
static unsigned char shutdown_timeout = 5;
44 45 46
module_param(shutdown_timeout, byte, 0644);
MODULE_PARM_DESC(shutdown_timeout, "timeout in seconds for controller shutdown");

47 48
static u8 nvme_max_retries = 5;
module_param_named(max_retries, nvme_max_retries, byte, 0644);
K
Keith Busch 已提交
49
MODULE_PARM_DESC(max_retries, "max number of retries a command may have");
50

51
static unsigned long default_ps_max_latency_us = 100000;
52 53 54 55
module_param(default_ps_max_latency_us, ulong, 0644);
MODULE_PARM_DESC(default_ps_max_latency_us,
		 "max power saving latency for new devices; use PM QOS to change per device");

56 57 58 59
static bool force_apst;
module_param(force_apst, bool, 0644);
MODULE_PARM_DESC(force_apst, "allow APST for newly enumerated devices even if quirked off");

60 61 62 63
static bool streams;
module_param(streams, bool, 0644);
MODULE_PARM_DESC(streams, "turn on support for Streams write directives");

64 65 66 67 68
/*
 * nvme_wq - hosts nvme related works that are not reset or delete
 * nvme_reset_wq - hosts nvme reset works
 * nvme_delete_wq - hosts nvme delete works
 *
69 70
 * nvme_wq will host works such as scan, aen handling, fw activation,
 * keep-alive, periodic reconnects etc. nvme_reset_wq
71 72 73 74
 * runs reset works which also flush works hosted on nvme_wq for
 * serialization purposes. nvme_delete_wq host controller deletion
 * works which flush reset works for serialization.
 */
75 76 77
struct workqueue_struct *nvme_wq;
EXPORT_SYMBOL_GPL(nvme_wq);

78 79 80 81 82 83
struct workqueue_struct *nvme_reset_wq;
EXPORT_SYMBOL_GPL(nvme_reset_wq);

struct workqueue_struct *nvme_delete_wq;
EXPORT_SYMBOL_GPL(nvme_delete_wq);

C
Christoph Hellwig 已提交
84 85
static LIST_HEAD(nvme_subsystems);
static DEFINE_MUTEX(nvme_subsystems_lock);
86

87
static DEFINE_IDA(nvme_instance_ida);
88
static dev_t nvme_ctrl_base_chr_devt;
89
static struct class *nvme_class;
C
Christoph Hellwig 已提交
90
static struct class *nvme_subsys_class;
91

92
static void nvme_put_subsystem(struct nvme_subsystem *subsys);
93 94 95
static void nvme_remove_invalid_namespaces(struct nvme_ctrl *ctrl,
					   unsigned nsid);

96 97 98 99 100 101 102 103
/*
 * Prepare a queue for teardown.
 *
 * This must forcibly unquiesce queues to avoid blocking dispatch, and only set
 * the capacity to 0 after that to avoid blocking dispatchers that may be
 * holding bd_butex.  This will end buffered writers dirtying pages that can't
 * be synced.
 */
104 105
static void nvme_set_queue_dying(struct nvme_ns *ns)
{
C
Christoph Hellwig 已提交
106
	if (test_and_set_bit(NVME_NS_DEAD, &ns->flags))
107
		return;
108

109 110
	blk_set_queue_dying(ns->queue);
	blk_mq_unquiesce_queue(ns->queue);
111

112
	set_capacity_and_notify(ns->disk, 0);
113
}
114

115 116 117 118 119
static void nvme_queue_scan(struct nvme_ctrl *ctrl)
{
	/*
	 * Only new queue scan work when admin and IO queues are both alive
	 */
K
Keith Busch 已提交
120
	if (ctrl->state == NVME_CTRL_LIVE && ctrl->tagset)
121 122 123
		queue_work(nvme_wq, &ctrl->scan_work);
}

124 125 126 127 128 129
/*
 * Use this function to proceed with scheduling reset_work for a controller
 * that had previously been set to the resetting state. This is intended for
 * code paths that can't be interrupted by other reset attempts. A hot removal
 * may prevent this from succeeding.
 */
130
int nvme_try_sched_reset(struct nvme_ctrl *ctrl)
131 132 133 134 135 136 137
{
	if (ctrl->state != NVME_CTRL_RESETTING)
		return -EBUSY;
	if (!queue_work(nvme_reset_wq, &ctrl->reset_work))
		return -EBUSY;
	return 0;
}
138
EXPORT_SYMBOL_GPL(nvme_try_sched_reset);
139

140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171
static void nvme_failfast_work(struct work_struct *work)
{
	struct nvme_ctrl *ctrl = container_of(to_delayed_work(work),
			struct nvme_ctrl, failfast_work);

	if (ctrl->state != NVME_CTRL_CONNECTING)
		return;

	set_bit(NVME_CTRL_FAILFAST_EXPIRED, &ctrl->flags);
	dev_info(ctrl->device, "failfast expired\n");
	nvme_kick_requeue_lists(ctrl);
}

static inline void nvme_start_failfast_work(struct nvme_ctrl *ctrl)
{
	if (!ctrl->opts || ctrl->opts->fast_io_fail_tmo == -1)
		return;

	schedule_delayed_work(&ctrl->failfast_work,
			      ctrl->opts->fast_io_fail_tmo * HZ);
}

static inline void nvme_stop_failfast_work(struct nvme_ctrl *ctrl)
{
	if (!ctrl->opts)
		return;

	cancel_delayed_work_sync(&ctrl->failfast_work);
	clear_bit(NVME_CTRL_FAILFAST_EXPIRED, &ctrl->flags);
}


172 173 174 175
int nvme_reset_ctrl(struct nvme_ctrl *ctrl)
{
	if (!nvme_change_ctrl_state(ctrl, NVME_CTRL_RESETTING))
		return -EBUSY;
176
	if (!queue_work(nvme_reset_wq, &ctrl->reset_work))
177 178 179 180 181
		return -EBUSY;
	return 0;
}
EXPORT_SYMBOL_GPL(nvme_reset_ctrl);

182
static int nvme_reset_ctrl_sync(struct nvme_ctrl *ctrl)
183 184 185 186
{
	int ret;

	ret = nvme_reset_ctrl(ctrl);
187
	if (!ret) {
188
		flush_work(&ctrl->reset_work);
K
Keith Busch 已提交
189
		if (ctrl->state != NVME_CTRL_LIVE)
190 191 192
			ret = -ENETRESET;
	}

193 194 195
	return ret;
}

196
static void nvme_do_delete_ctrl(struct nvme_ctrl *ctrl)
197
{
198 199 200
	dev_info(ctrl->device,
		 "Removing ctrl: NQN \"%s\"\n", ctrl->opts->subsysnqn);

201
	flush_work(&ctrl->reset_work);
202 203
	nvme_stop_ctrl(ctrl);
	nvme_remove_namespaces(ctrl);
204
	ctrl->ops->delete_ctrl(ctrl);
205
	nvme_uninit_ctrl(ctrl);
206 207
}

208 209 210 211 212 213 214 215
static void nvme_delete_ctrl_work(struct work_struct *work)
{
	struct nvme_ctrl *ctrl =
		container_of(work, struct nvme_ctrl, delete_work);

	nvme_do_delete_ctrl(ctrl);
}

216 217 218 219
int nvme_delete_ctrl(struct nvme_ctrl *ctrl)
{
	if (!nvme_change_ctrl_state(ctrl, NVME_CTRL_DELETING))
		return -EBUSY;
220
	if (!queue_work(nvme_delete_wq, &ctrl->delete_work))
221 222 223 224 225
		return -EBUSY;
	return 0;
}
EXPORT_SYMBOL_GPL(nvme_delete_ctrl);

226
static void nvme_delete_ctrl_sync(struct nvme_ctrl *ctrl)
227 228
{
	/*
229 230
	 * Keep a reference until nvme_do_delete_ctrl() complete,
	 * since ->delete_ctrl can free the controller.
231 232
	 */
	nvme_get_ctrl(ctrl);
233
	if (nvme_change_ctrl_state(ctrl, NVME_CTRL_DELETING))
234
		nvme_do_delete_ctrl(ctrl);
235 236 237
	nvme_put_ctrl(ctrl);
}

238
static blk_status_t nvme_error_status(u16 status)
239
{
240
	switch (status & 0x7ff) {
241
	case NVME_SC_SUCCESS:
242
		return BLK_STS_OK;
243
	case NVME_SC_CAP_EXCEEDED:
244
		return BLK_STS_NOSPC;
245
	case NVME_SC_LBA_RANGE:
246 247
	case NVME_SC_CMD_INTERRUPTED:
	case NVME_SC_NS_NOT_READY:
248 249
		return BLK_STS_TARGET;
	case NVME_SC_BAD_ATTRIBUTES:
250
	case NVME_SC_ONCS_NOT_SUPPORTED:
251 252 253
	case NVME_SC_INVALID_OPCODE:
	case NVME_SC_INVALID_FIELD:
	case NVME_SC_INVALID_NS:
254
		return BLK_STS_NOTSUPP;
255 256 257
	case NVME_SC_WRITE_FAULT:
	case NVME_SC_READ_ERROR:
	case NVME_SC_UNWRITTEN_BLOCK:
258 259
	case NVME_SC_ACCESS_DENIED:
	case NVME_SC_READ_ONLY:
260
	case NVME_SC_COMPARE_FAILED:
261
		return BLK_STS_MEDIUM;
262 263 264 265 266 267 268
	case NVME_SC_GUARD_CHECK:
	case NVME_SC_APPTAG_CHECK:
	case NVME_SC_REFTAG_CHECK:
	case NVME_SC_INVALID_PI:
		return BLK_STS_PROTECTION;
	case NVME_SC_RESERVATION_CONFLICT:
		return BLK_STS_NEXUS;
269 270
	case NVME_SC_HOST_PATH_ERROR:
		return BLK_STS_TRANSPORT;
K
Keith Busch 已提交
271 272 273 274
	case NVME_SC_ZONE_TOO_MANY_ACTIVE:
		return BLK_STS_ZONE_ACTIVE_RESOURCE;
	case NVME_SC_ZONE_TOO_MANY_OPEN:
		return BLK_STS_ZONE_OPEN_RESOURCE;
275 276
	default:
		return BLK_STS_IOERR;
277 278 279
	}
}

280 281 282 283 284 285 286
static void nvme_retry_req(struct request *req)
{
	unsigned long delay = 0;
	u16 crd;

	/* The mask and shift result must be <= 3 */
	crd = (nvme_req(req)->status & NVME_SC_CRD) >> 11;
287 288
	if (crd)
		delay = nvme_req(req)->ctrl->crdt[crd - 1] * 100;
289 290 291 292 293 294

	nvme_req(req)->retries++;
	blk_mq_requeue_request(req, false);
	blk_mq_delay_kick_requeue_list(req->q, delay);
}

295 296 297 298 299 300 301
enum nvme_disposition {
	COMPLETE,
	RETRY,
	FAILOVER,
};

static inline enum nvme_disposition nvme_decide_disposition(struct request *req)
302
{
303 304
	if (likely(nvme_req(req)->status == 0))
		return COMPLETE;
305

306 307 308 309
	if (blk_noretry_request(req) ||
	    (nvme_req(req)->status & NVME_SC_DNR) ||
	    nvme_req(req)->retries >= nvme_max_retries)
		return COMPLETE;
310

311
	if (req->cmd_flags & REQ_NVME_MPATH) {
312 313
		if (nvme_is_path_error(nvme_req(req)->status) ||
		    blk_queue_dying(req->q))
314
			return FAILOVER;
315 316 317
	} else {
		if (blk_queue_dying(req->q))
			return COMPLETE;
318
	}
319

320 321
	return RETRY;
}
322

323 324 325
static inline void nvme_end_req(struct request *req)
{
	blk_status_t status = nvme_error_status(nvme_req(req)->status);
326

327 328
	if (IS_ENABLED(CONFIG_BLK_DEV_ZONED) &&
	    req_op(req) == REQ_OP_ZONE_APPEND)
K
Keith Busch 已提交
329 330
		req->__sector = nvme_lba_to_sect(req->q->queuedata,
			le64_to_cpu(nvme_req(req)->result.u64));
H
Hannes Reinecke 已提交
331

332
	nvme_trace_bio_complete(req);
333
	blk_mq_end_request(req, status);
334
}
335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355

void nvme_complete_rq(struct request *req)
{
	trace_nvme_complete_rq(req);
	nvme_cleanup_cmd(req);

	if (nvme_req(req)->ctrl->kas)
		nvme_req(req)->ctrl->comp_seen = true;

	switch (nvme_decide_disposition(req)) {
	case COMPLETE:
		nvme_end_req(req);
		return;
	case RETRY:
		nvme_retry_req(req);
		return;
	case FAILOVER:
		nvme_failover_req(req);
		return;
	}
}
356 357
EXPORT_SYMBOL_GPL(nvme_complete_rq);

358
bool nvme_cancel_request(struct request *req, void *data, bool reserved)
359 360 361 362
{
	dev_dbg_ratelimited(((struct nvme_ctrl *) data)->device,
				"Cancelling I/O %d", req->tag);

363 364 365 366
	/* don't abort one completed request */
	if (blk_mq_request_completed(req))
		return true;

367
	nvme_req(req)->status = NVME_SC_HOST_ABORTED_CMD;
368
	blk_mq_complete_request(req);
369
	return true;
370 371 372
}
EXPORT_SYMBOL_GPL(nvme_cancel_request);

C
Chao Leng 已提交
373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392
void nvme_cancel_tagset(struct nvme_ctrl *ctrl)
{
	if (ctrl->tagset) {
		blk_mq_tagset_busy_iter(ctrl->tagset,
				nvme_cancel_request, ctrl);
		blk_mq_tagset_wait_completed_request(ctrl->tagset);
	}
}
EXPORT_SYMBOL_GPL(nvme_cancel_tagset);

void nvme_cancel_admin_tagset(struct nvme_ctrl *ctrl)
{
	if (ctrl->admin_tagset) {
		blk_mq_tagset_busy_iter(ctrl->admin_tagset,
				nvme_cancel_request, ctrl);
		blk_mq_tagset_wait_completed_request(ctrl->admin_tagset);
	}
}
EXPORT_SYMBOL_GPL(nvme_cancel_admin_tagset);

393 394 395
bool nvme_change_ctrl_state(struct nvme_ctrl *ctrl,
		enum nvme_ctrl_state new_state)
{
396
	enum nvme_ctrl_state old_state;
397
	unsigned long flags;
398 399
	bool changed = false;

400
	spin_lock_irqsave(&ctrl->lock, flags);
401 402

	old_state = ctrl->state;
403 404 405
	switch (new_state) {
	case NVME_CTRL_LIVE:
		switch (old_state) {
406
		case NVME_CTRL_NEW:
407
		case NVME_CTRL_RESETTING:
408
		case NVME_CTRL_CONNECTING:
409
			changed = true;
410
			fallthrough;
411 412 413 414 415 416 417
		default:
			break;
		}
		break;
	case NVME_CTRL_RESETTING:
		switch (old_state) {
		case NVME_CTRL_NEW:
418 419
		case NVME_CTRL_LIVE:
			changed = true;
420
			fallthrough;
421 422 423 424
		default:
			break;
		}
		break;
425
	case NVME_CTRL_CONNECTING:
426
		switch (old_state) {
427
		case NVME_CTRL_NEW:
428
		case NVME_CTRL_RESETTING:
429
			changed = true;
430
			fallthrough;
431 432 433 434 435 436 437 438
		default:
			break;
		}
		break;
	case NVME_CTRL_DELETING:
		switch (old_state) {
		case NVME_CTRL_LIVE:
		case NVME_CTRL_RESETTING:
439
		case NVME_CTRL_CONNECTING:
440
			changed = true;
441
			fallthrough;
442 443 444 445
		default:
			break;
		}
		break;
446 447 448 449 450
	case NVME_CTRL_DELETING_NOIO:
		switch (old_state) {
		case NVME_CTRL_DELETING:
		case NVME_CTRL_DEAD:
			changed = true;
451
			fallthrough;
452 453 454 455
		default:
			break;
		}
		break;
456 457 458 459
	case NVME_CTRL_DEAD:
		switch (old_state) {
		case NVME_CTRL_DELETING:
			changed = true;
460
			fallthrough;
461 462 463 464
		default:
			break;
		}
		break;
465 466 467 468
	default:
		break;
	}

469
	if (changed) {
470
		ctrl->state = new_state;
471 472
		wake_up_all(&ctrl->state_wq);
	}
473

474
	spin_unlock_irqrestore(&ctrl->lock, flags);
475 476 477 478 479 480
	if (!changed)
		return false;

	if (ctrl->state == NVME_CTRL_LIVE) {
		if (old_state == NVME_CTRL_CONNECTING)
			nvme_stop_failfast_work(ctrl);
481
		nvme_kick_requeue_lists(ctrl);
482 483 484 485
	} else if (ctrl->state == NVME_CTRL_CONNECTING &&
		old_state == NVME_CTRL_RESETTING) {
		nvme_start_failfast_work(ctrl);
	}
486 487 488 489
	return changed;
}
EXPORT_SYMBOL_GPL(nvme_change_ctrl_state);

490 491 492 493 494 495 496 497 498 499 500 501
/*
 * Returns true for sink states that can't ever transition back to live.
 */
static bool nvme_state_terminal(struct nvme_ctrl *ctrl)
{
	switch (ctrl->state) {
	case NVME_CTRL_NEW:
	case NVME_CTRL_LIVE:
	case NVME_CTRL_RESETTING:
	case NVME_CTRL_CONNECTING:
		return false;
	case NVME_CTRL_DELETING:
502
	case NVME_CTRL_DELETING_NOIO:
503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523
	case NVME_CTRL_DEAD:
		return true;
	default:
		WARN_ONCE(1, "Unhandled ctrl state:%d", ctrl->state);
		return true;
	}
}

/*
 * Waits for the controller state to be resetting, or returns false if it is
 * not possible to ever transition to that state.
 */
bool nvme_wait_reset(struct nvme_ctrl *ctrl)
{
	wait_event(ctrl->state_wq,
		   nvme_change_ctrl_state(ctrl, NVME_CTRL_RESETTING) ||
		   nvme_state_terminal(ctrl));
	return ctrl->state == NVME_CTRL_RESETTING;
}
EXPORT_SYMBOL_GPL(nvme_wait_reset);

C
Christoph Hellwig 已提交
524 525 526 527 528
static void nvme_free_ns_head(struct kref *ref)
{
	struct nvme_ns_head *head =
		container_of(ref, struct nvme_ns_head, ref);

529
	nvme_mpath_remove_disk(head);
C
Christoph Hellwig 已提交
530
	ida_simple_remove(&head->subsys->ns_ida, head->instance);
531
	cleanup_srcu_struct(&head->srcu);
532
	nvme_put_subsystem(head->subsys);
C
Christoph Hellwig 已提交
533 534 535 536 537 538 539 540
	kfree(head);
}

static void nvme_put_ns_head(struct nvme_ns_head *head)
{
	kref_put(&head->ref, nvme_free_ns_head);
}

541 542 543 544
static void nvme_free_ns(struct kref *kref)
{
	struct nvme_ns *ns = container_of(kref, struct nvme_ns, kref);

545 546
	if (ns->ndev)
		nvme_nvm_unregister(ns);
547 548

	put_disk(ns->disk);
C
Christoph Hellwig 已提交
549
	nvme_put_ns_head(ns->head);
550
	nvme_put_ctrl(ns->ctrl);
551 552 553
	kfree(ns);
}

554
void nvme_put_ns(struct nvme_ns *ns)
555 556 557
{
	kref_put(&ns->kref, nvme_free_ns);
}
558
EXPORT_SYMBOL_NS_GPL(nvme_put_ns, NVME_TARGET_PASSTHRU);
559

560 561 562 563 564 565 566 567 568
static inline void nvme_clear_nvme_request(struct request *req)
{
	if (!(req->rq_flags & RQF_DONTPREP)) {
		nvme_req(req)->retries = 0;
		nvme_req(req)->flags = 0;
		req->rq_flags |= RQF_DONTPREP;
	}
}

569
static inline unsigned int nvme_req_op(struct nvme_command *cmd)
570
{
571 572
	return nvme_is_write(cmd) ? REQ_OP_DRV_OUT : REQ_OP_DRV_IN;
}
573

574 575 576
static inline void nvme_init_request(struct request *req,
		struct nvme_command *cmd)
{
577 578 579
	if (req->q->queuedata)
		req->timeout = NVME_IO_TIMEOUT;
	else /* no queuedata implies admin queue */
580
		req->timeout = NVME_ADMIN_TIMEOUT;
581 582

	req->cmd_flags |= REQ_FAILFAST_DRIVER;
583
	nvme_clear_nvme_request(req);
584
	nvme_req(req)->cmd = cmd;
585
}
586

587 588 589 590 591 592 593 594
struct request *nvme_alloc_request(struct request_queue *q,
		struct nvme_command *cmd, blk_mq_req_flags_t flags)
{
	struct request *req;

	req = blk_mq_alloc_request(q, nvme_req_op(cmd), flags);
	if (!IS_ERR(req))
		nvme_init_request(req, cmd);
595 596
	return req;
}
597
EXPORT_SYMBOL_GPL(nvme_alloc_request);
598

599
static struct request *nvme_alloc_request_qid(struct request_queue *q,
600 601 602 603 604 605 606 607 608 609 610
		struct nvme_command *cmd, blk_mq_req_flags_t flags, int qid)
{
	struct request *req;

	req = blk_mq_alloc_request_hctx(q, nvme_req_op(cmd), flags,
			qid ? qid - 1 : 0);
	if (!IS_ERR(req))
		nvme_init_request(req, cmd);
	return req;
}

611 612 613 614 615 616 617
static int nvme_toggle_streams(struct nvme_ctrl *ctrl, bool enable)
{
	struct nvme_command c;

	memset(&c, 0, sizeof(c));

	c.directive.opcode = nvme_admin_directive_send;
A
Arnav Dawn 已提交
618
	c.directive.nsid = cpu_to_le32(NVME_NSID_ALL);
619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646
	c.directive.doper = NVME_DIR_SND_ID_OP_ENABLE;
	c.directive.dtype = NVME_DIR_IDENTIFY;
	c.directive.tdtype = NVME_DIR_STREAMS;
	c.directive.endir = enable ? NVME_DIR_ENDIR : 0;

	return nvme_submit_sync_cmd(ctrl->admin_q, &c, NULL, 0);
}

static int nvme_disable_streams(struct nvme_ctrl *ctrl)
{
	return nvme_toggle_streams(ctrl, false);
}

static int nvme_enable_streams(struct nvme_ctrl *ctrl)
{
	return nvme_toggle_streams(ctrl, true);
}

static int nvme_get_stream_params(struct nvme_ctrl *ctrl,
				  struct streams_directive_params *s, u32 nsid)
{
	struct nvme_command c;

	memset(&c, 0, sizeof(c));
	memset(s, 0, sizeof(*s));

	c.directive.opcode = nvme_admin_directive_recv;
	c.directive.nsid = cpu_to_le32(nsid);
K
Keith Busch 已提交
647
	c.directive.numd = cpu_to_le32(nvme_bytes_to_numd(sizeof(*s)));
648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667
	c.directive.doper = NVME_DIR_RCV_ST_OP_PARAM;
	c.directive.dtype = NVME_DIR_STREAMS;

	return nvme_submit_sync_cmd(ctrl->admin_q, &c, s, sizeof(*s));
}

static int nvme_configure_directives(struct nvme_ctrl *ctrl)
{
	struct streams_directive_params s;
	int ret;

	if (!(ctrl->oacs & NVME_CTRL_OACS_DIRECTIVES))
		return 0;
	if (!streams)
		return 0;

	ret = nvme_enable_streams(ctrl);
	if (ret)
		return ret;

A
Arnav Dawn 已提交
668
	ret = nvme_get_stream_params(ctrl, &s, NVME_NSID_ALL);
669
	if (ret)
670
		goto out_disable_stream;
671 672 673 674 675

	ctrl->nssa = le16_to_cpu(s.nssa);
	if (ctrl->nssa < BLK_MAX_WRITE_HINTS - 1) {
		dev_info(ctrl->device, "too few streams (%u) available\n",
					ctrl->nssa);
676
		goto out_disable_stream;
677 678
	}

679
	ctrl->nr_streams = min_t(u16, ctrl->nssa, BLK_MAX_WRITE_HINTS - 1);
680 681
	dev_info(ctrl->device, "Using %u streams\n", ctrl->nr_streams);
	return 0;
682 683 684 685

out_disable_stream:
	nvme_disable_streams(ctrl);
	return ret;
686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712
}

/*
 * Check if 'req' has a write hint associated with it. If it does, assign
 * a valid namespace stream to the write.
 */
static void nvme_assign_write_stream(struct nvme_ctrl *ctrl,
				     struct request *req, u16 *control,
				     u32 *dsmgmt)
{
	enum rw_hint streamid = req->write_hint;

	if (streamid == WRITE_LIFE_NOT_SET || streamid == WRITE_LIFE_NONE)
		streamid = 0;
	else {
		streamid--;
		if (WARN_ON_ONCE(streamid > ctrl->nr_streams))
			return;

		*control |= NVME_RW_DTYPE_STREAMS;
		*dsmgmt |= streamid << 16;
	}

	if (streamid < ARRAY_SIZE(req->q->write_hints))
		req->q->write_hints[streamid] += blk_rq_bytes(req) >> 9;
}

713 714 715 716 717 718 719 720
static void nvme_setup_passthrough(struct request *req,
		struct nvme_command *cmd)
{
	memcpy(cmd, nvme_req(req)->cmd, sizeof(*cmd));
	/* passthru commands should let the driver set the SGL flags */
	cmd->common.flags &= ~NVME_CMD_SGL_ALL;
}

M
Ming Lin 已提交
721 722 723 724
static inline void nvme_setup_flush(struct nvme_ns *ns,
		struct nvme_command *cmnd)
{
	cmnd->common.opcode = nvme_cmd_flush;
C
Christoph Hellwig 已提交
725
	cmnd->common.nsid = cpu_to_le32(ns->head->ns_id);
M
Ming Lin 已提交
726 727
}

728
static blk_status_t nvme_setup_discard(struct nvme_ns *ns, struct request *req,
M
Ming Lin 已提交
729 730
		struct nvme_command *cmnd)
{
731
	unsigned short segments = blk_rq_nr_discard_segments(req), n = 0;
M
Ming Lin 已提交
732
	struct nvme_dsm_range *range;
733
	struct bio *bio;
M
Ming Lin 已提交
734

735 736 737 738 739 740 741 742
	/*
	 * Some devices do not consider the DSM 'Number of Ranges' field when
	 * determining how much data to DMA. Always allocate memory for maximum
	 * number of segments to prevent device reading beyond end of buffer.
	 */
	static const size_t alloc_size = sizeof(*range) * NVME_DSM_MAX_RANGES;

	range = kzalloc(alloc_size, GFP_ATOMIC | __GFP_NOWARN);
743 744 745 746 747 748 749 750 751 752 753
	if (!range) {
		/*
		 * If we fail allocation our range, fallback to the controller
		 * discard page. If that's also busy, it's safe to return
		 * busy, as we know we can make progress once that's freed.
		 */
		if (test_and_set_bit_lock(0, &ns->ctrl->discard_page_busy))
			return BLK_STS_RESOURCE;

		range = page_address(ns->ctrl->discard_page);
	}
M
Ming Lin 已提交
754

755
	__rq_for_each_bio(bio, req) {
756
		u64 slba = nvme_sect_to_lba(ns, bio->bi_iter.bi_sector);
757 758
		u32 nlb = bio->bi_iter.bi_size >> ns->lba_shift;

K
Keith Busch 已提交
759 760 761 762 763
		if (n < segments) {
			range[n].cattr = cpu_to_le32(0);
			range[n].nlb = cpu_to_le32(nlb);
			range[n].slba = cpu_to_le64(slba);
		}
764 765 766 767
		n++;
	}

	if (WARN_ON_ONCE(n != segments)) {
768 769 770 771
		if (virt_to_page(range) == ns->ctrl->discard_page)
			clear_bit_unlock(0, &ns->ctrl->discard_page_busy);
		else
			kfree(range);
772
		return BLK_STS_IOERR;
773
	}
M
Ming Lin 已提交
774 775

	cmnd->dsm.opcode = nvme_cmd_dsm;
C
Christoph Hellwig 已提交
776
	cmnd->dsm.nsid = cpu_to_le32(ns->head->ns_id);
777
	cmnd->dsm.nr = cpu_to_le32(segments - 1);
M
Ming Lin 已提交
778 779
	cmnd->dsm.attributes = cpu_to_le32(NVME_DSMGMT_AD);

780 781
	req->special_vec.bv_page = virt_to_page(range);
	req->special_vec.bv_offset = offset_in_page(range);
782
	req->special_vec.bv_len = alloc_size;
783
	req->rq_flags |= RQF_SPECIAL_PAYLOAD;
M
Ming Lin 已提交
784

785
	return BLK_STS_OK;
M
Ming Lin 已提交
786 787
}

788 789 790 791 792 793 794 795 796
static inline blk_status_t nvme_setup_write_zeroes(struct nvme_ns *ns,
		struct request *req, struct nvme_command *cmnd)
{
	if (ns->ctrl->quirks & NVME_QUIRK_DEALLOCATE_ZEROES)
		return nvme_setup_discard(ns, req, cmnd);

	cmnd->write_zeroes.opcode = nvme_cmd_write_zeroes;
	cmnd->write_zeroes.nsid = cpu_to_le32(ns->head->ns_id);
	cmnd->write_zeroes.slba =
797
		cpu_to_le64(nvme_sect_to_lba(ns, blk_rq_pos(req)));
798 799 800 801 802 803
	cmnd->write_zeroes.length =
		cpu_to_le16((blk_rq_bytes(req) >> ns->lba_shift) - 1);
	cmnd->write_zeroes.control = 0;
	return BLK_STS_OK;
}

804
static inline blk_status_t nvme_setup_rw(struct nvme_ns *ns,
K
Keith Busch 已提交
805 806
		struct request *req, struct nvme_command *cmnd,
		enum nvme_opcode op)
M
Ming Lin 已提交
807
{
808
	struct nvme_ctrl *ctrl = ns->ctrl;
M
Ming Lin 已提交
809 810 811 812 813 814 815 816 817 818 819
	u16 control = 0;
	u32 dsmgmt = 0;

	if (req->cmd_flags & REQ_FUA)
		control |= NVME_RW_FUA;
	if (req->cmd_flags & (REQ_FAILFAST_DEV | REQ_RAHEAD))
		control |= NVME_RW_LR;

	if (req->cmd_flags & REQ_RAHEAD)
		dsmgmt |= NVME_RW_DSM_FREQ_PREFETCH;

K
Keith Busch 已提交
820
	cmnd->rw.opcode = op;
C
Christoph Hellwig 已提交
821
	cmnd->rw.nsid = cpu_to_le32(ns->head->ns_id);
822
	cmnd->rw.slba = cpu_to_le64(nvme_sect_to_lba(ns, blk_rq_pos(req)));
M
Ming Lin 已提交
823 824
	cmnd->rw.length = cpu_to_le16((blk_rq_bytes(req) >> ns->lba_shift) - 1);

825 826 827
	if (req_op(req) == REQ_OP_WRITE && ctrl->nr_streams)
		nvme_assign_write_stream(ctrl, req, &control, &dsmgmt);

M
Ming Lin 已提交
828
	if (ns->ms) {
829 830 831 832 833 834 835 836 837 838 839 840
		/*
		 * If formated with metadata, the block layer always provides a
		 * metadata buffer if CONFIG_BLK_DEV_INTEGRITY is enabled.  Else
		 * we enable the PRACT bit for protection information or set the
		 * namespace capacity to zero to prevent any I/O.
		 */
		if (!blk_integrity_rq(req)) {
			if (WARN_ON_ONCE(!nvme_ns_has_pi(ns)))
				return BLK_STS_NOTSUPP;
			control |= NVME_RW_PRINFO_PRACT;
		}

M
Ming Lin 已提交
841 842 843 844 845 846 847 848
		switch (ns->pi_type) {
		case NVME_NS_DPS_PI_TYPE3:
			control |= NVME_RW_PRINFO_PRCHK_GUARD;
			break;
		case NVME_NS_DPS_PI_TYPE1:
		case NVME_NS_DPS_PI_TYPE2:
			control |= NVME_RW_PRINFO_PRCHK_GUARD |
					NVME_RW_PRINFO_PRCHK_REF;
K
Keith Busch 已提交
849 850
			if (op == nvme_cmd_zone_append)
				control |= NVME_RW_APPEND_PIREMAP;
851
			cmnd->rw.reftag = cpu_to_le32(t10_pi_ref_tag(req));
M
Ming Lin 已提交
852 853 854 855 856 857
			break;
		}
	}

	cmnd->rw.control = cpu_to_le16(control);
	cmnd->rw.dsmgmt = cpu_to_le32(dsmgmt);
858
	return 0;
M
Ming Lin 已提交
859 860
}

861 862 863
void nvme_cleanup_cmd(struct request *req)
{
	if (req->rq_flags & RQF_SPECIAL_PAYLOAD) {
M
Minwoo Im 已提交
864
		struct nvme_ctrl *ctrl = nvme_req(req)->ctrl;
865 866
		struct page *page = req->special_vec.bv_page;

M
Minwoo Im 已提交
867 868
		if (page == ctrl->discard_page)
			clear_bit_unlock(0, &ctrl->discard_page_busy);
869 870
		else
			kfree(page_address(page) + req->special_vec.bv_offset);
871 872 873 874
	}
}
EXPORT_SYMBOL_GPL(nvme_cleanup_cmd);

875
blk_status_t nvme_setup_cmd(struct nvme_ns *ns, struct request *req,
M
Ming Lin 已提交
876 877
		struct nvme_command *cmd)
{
878
	blk_status_t ret = BLK_STS_OK;
M
Ming Lin 已提交
879

880
	nvme_clear_nvme_request(req);
881

882
	memset(cmd, 0, sizeof(*cmd));
883 884 885
	switch (req_op(req)) {
	case REQ_OP_DRV_IN:
	case REQ_OP_DRV_OUT:
886
		nvme_setup_passthrough(req, cmd);
887 888
		break;
	case REQ_OP_FLUSH:
M
Ming Lin 已提交
889
		nvme_setup_flush(ns, cmd);
890
		break;
K
Keith Busch 已提交
891 892 893 894 895 896 897 898 899 900 901 902 903
	case REQ_OP_ZONE_RESET_ALL:
	case REQ_OP_ZONE_RESET:
		ret = nvme_setup_zone_mgmt_send(ns, req, cmd, NVME_ZONE_RESET);
		break;
	case REQ_OP_ZONE_OPEN:
		ret = nvme_setup_zone_mgmt_send(ns, req, cmd, NVME_ZONE_OPEN);
		break;
	case REQ_OP_ZONE_CLOSE:
		ret = nvme_setup_zone_mgmt_send(ns, req, cmd, NVME_ZONE_CLOSE);
		break;
	case REQ_OP_ZONE_FINISH:
		ret = nvme_setup_zone_mgmt_send(ns, req, cmd, NVME_ZONE_FINISH);
		break;
904
	case REQ_OP_WRITE_ZEROES:
905 906
		ret = nvme_setup_write_zeroes(ns, req, cmd);
		break;
907
	case REQ_OP_DISCARD:
M
Ming Lin 已提交
908
		ret = nvme_setup_discard(ns, req, cmd);
909 910
		break;
	case REQ_OP_READ:
K
Keith Busch 已提交
911 912
		ret = nvme_setup_rw(ns, req, cmd, nvme_cmd_read);
		break;
913
	case REQ_OP_WRITE:
K
Keith Busch 已提交
914 915 916 917
		ret = nvme_setup_rw(ns, req, cmd, nvme_cmd_write);
		break;
	case REQ_OP_ZONE_APPEND:
		ret = nvme_setup_rw(ns, req, cmd, nvme_cmd_zone_append);
918 919 920
		break;
	default:
		WARN_ON_ONCE(1);
921
		return BLK_STS_IOERR;
922
	}
M
Ming Lin 已提交
923

924
	cmd->common.command_id = req->tag;
K
Keith Busch 已提交
925
	trace_nvme_setup_cmd(req, cmd);
M
Ming Lin 已提交
926 927 928 929
	return ret;
}
EXPORT_SYMBOL_GPL(nvme_setup_cmd);

930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946
static void nvme_end_sync_rq(struct request *rq, blk_status_t error)
{
	struct completion *waiting = rq->end_io_data;

	rq->end_io_data = NULL;
	complete(waiting);
}

static void nvme_execute_rq_polled(struct request_queue *q,
		struct gendisk *bd_disk, struct request *rq, int at_head)
{
	DECLARE_COMPLETION_ONSTACK(wait);

	WARN_ON_ONCE(!test_bit(QUEUE_FLAG_POLL, &q->queue_flags));

	rq->cmd_flags |= REQ_HIPRI;
	rq->end_io_data = &wait;
947
	blk_execute_rq_nowait(bd_disk, rq, at_head, nvme_end_sync_rq);
948 949 950 951 952 953 954

	while (!completion_done(&wait)) {
		blk_poll(q, request_to_qc_t(rq->mq_hctx, rq), true);
		cond_resched();
	}
}

955 956 957 958 959
/*
 * Returns 0 on success.  If the result is negative, it's a Linux error code;
 * if the result is positive, it's an NVM Express status code
 */
int __nvme_submit_sync_cmd(struct request_queue *q, struct nvme_command *cmd,
960
		union nvme_result *result, void *buffer, unsigned bufflen,
961
		unsigned timeout, int qid, int at_head,
962
		blk_mq_req_flags_t flags, bool poll)
963 964 965 966
{
	struct request *req;
	int ret;

967 968 969 970
	if (qid == NVME_QID_ANY)
		req = nvme_alloc_request(q, cmd, flags);
	else
		req = nvme_alloc_request_qid(q, cmd, flags, qid);
971 972 973
	if (IS_ERR(req))
		return PTR_ERR(req);

974 975
	if (timeout)
		req->timeout = timeout;
976

977 978 979 980
	if (buffer && bufflen) {
		ret = blk_rq_map_kern(q, req, buffer, bufflen, GFP_KERNEL);
		if (ret)
			goto out;
981 982
	}

983 984 985
	if (poll)
		nvme_execute_rq_polled(req->q, NULL, req, at_head);
	else
986
		blk_execute_rq(NULL, req, at_head);
987 988
	if (result)
		*result = nvme_req(req)->result;
989 990 991 992
	if (nvme_req(req)->flags & NVME_REQ_CANCELLED)
		ret = -EINTR;
	else
		ret = nvme_req(req)->status;
993 994 995 996
 out:
	blk_mq_free_request(req);
	return ret;
}
997
EXPORT_SYMBOL_GPL(__nvme_submit_sync_cmd);
998 999 1000 1001

int nvme_submit_sync_cmd(struct request_queue *q, struct nvme_command *cmd,
		void *buffer, unsigned bufflen)
{
1002
	return __nvme_submit_sync_cmd(q, cmd, NULL, buffer, bufflen, 0,
1003
			NVME_QID_ANY, 0, 0, false);
1004
}
1005
EXPORT_SYMBOL_GPL(nvme_submit_sync_cmd);
1006

1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040
static void *nvme_add_user_metadata(struct bio *bio, void __user *ubuf,
		unsigned len, u32 seed, bool write)
{
	struct bio_integrity_payload *bip;
	int ret = -ENOMEM;
	void *buf;

	buf = kmalloc(len, GFP_KERNEL);
	if (!buf)
		goto out;

	ret = -EFAULT;
	if (write && copy_from_user(buf, ubuf, len))
		goto out_free_meta;

	bip = bio_integrity_alloc(bio, GFP_KERNEL, 1);
	if (IS_ERR(bip)) {
		ret = PTR_ERR(bip);
		goto out_free_meta;
	}

	bip->bip_iter.bi_size = len;
	bip->bip_iter.bi_sector = seed;
	ret = bio_integrity_add_page(bio, virt_to_page(buf), len,
			offset_in_page(buf));
	if (ret == len)
		return buf;
	ret = -ENOMEM;
out_free_meta:
	kfree(buf);
out:
	return ERR_PTR(ret);
}

1041 1042 1043 1044
static u32 nvme_known_admin_effects(u8 opcode)
{
	switch (opcode) {
	case nvme_admin_format_nvm:
1045
		return NVME_CMD_EFFECTS_LBCC | NVME_CMD_EFFECTS_NCC |
1046 1047
			NVME_CMD_EFFECTS_CSE_MASK;
	case nvme_admin_sanitize_nvm:
1048
		return NVME_CMD_EFFECTS_LBCC | NVME_CMD_EFFECTS_CSE_MASK;
1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085
	default:
		break;
	}
	return 0;
}

u32 nvme_command_effects(struct nvme_ctrl *ctrl, struct nvme_ns *ns, u8 opcode)
{
	u32 effects = 0;

	if (ns) {
		if (ns->head->effects)
			effects = le32_to_cpu(ns->head->effects->iocs[opcode]);
		if (effects & ~(NVME_CMD_EFFECTS_CSUPP | NVME_CMD_EFFECTS_LBCC))
			dev_warn(ctrl->device,
				 "IO command:%02x has unhandled effects:%08x\n",
				 opcode, effects);
		return 0;
	}

	if (ctrl->effects)
		effects = le32_to_cpu(ctrl->effects->acs[opcode]);
	effects |= nvme_known_admin_effects(opcode);

	return effects;
}
EXPORT_SYMBOL_NS_GPL(nvme_command_effects, NVME_TARGET_PASSTHRU);

static u32 nvme_passthru_start(struct nvme_ctrl *ctrl, struct nvme_ns *ns,
			       u8 opcode)
{
	u32 effects = nvme_command_effects(ctrl, ns, opcode);

	/*
	 * For simplicity, IO to all namespaces is quiesced even if the command
	 * effects say only one namespace is affected.
	 */
1086
	if (effects & NVME_CMD_EFFECTS_CSE_MASK) {
1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098
		mutex_lock(&ctrl->scan_lock);
		mutex_lock(&ctrl->subsys->lock);
		nvme_mpath_start_freeze(ctrl->subsys);
		nvme_mpath_wait_freeze(ctrl->subsys);
		nvme_start_freeze(ctrl);
		nvme_wait_freeze(ctrl);
	}
	return effects;
}

static void nvme_passthru_end(struct nvme_ctrl *ctrl, u32 effects)
{
1099
	if (effects & NVME_CMD_EFFECTS_CSE_MASK) {
1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113
		nvme_unfreeze(ctrl);
		nvme_mpath_unfreeze(ctrl->subsys);
		mutex_unlock(&ctrl->subsys->lock);
		nvme_remove_invalid_namespaces(ctrl, NVME_NSID_ALL);
		mutex_unlock(&ctrl->scan_lock);
	}
	if (effects & NVME_CMD_EFFECTS_CCC)
		nvme_init_identify(ctrl);
	if (effects & (NVME_CMD_EFFECTS_NIC | NVME_CMD_EFFECTS_NCC)) {
		nvme_queue_scan(ctrl);
		flush_work(&ctrl->scan_work);
	}
}

1114 1115 1116 1117 1118 1119 1120 1121 1122
void nvme_execute_passthru_rq(struct request *rq)
{
	struct nvme_command *cmd = nvme_req(rq)->cmd;
	struct nvme_ctrl *ctrl = nvme_req(rq)->ctrl;
	struct nvme_ns *ns = rq->q->queuedata;
	struct gendisk *disk = ns ? ns->disk : NULL;
	u32 effects;

	effects = nvme_passthru_start(ctrl, ns, cmd->common.opcode);
1123
	blk_execute_rq(disk, rq, 0);
1124 1125 1126 1127
	nvme_passthru_end(ctrl, effects);
}
EXPORT_SYMBOL_NS_GPL(nvme_execute_passthru_rq, NVME_TARGET_PASSTHRU);

1128
static int nvme_submit_user_cmd(struct request_queue *q,
1129 1130
		struct nvme_command *cmd, void __user *ubuffer,
		unsigned bufflen, void __user *meta_buffer, unsigned meta_len,
1131
		u32 meta_seed, u64 *result, unsigned timeout)
1132
{
1133
	bool write = nvme_is_write(cmd);
1134
	struct nvme_ns *ns = q->queuedata;
1135
	struct block_device *bdev = ns ? ns->disk->part0 : NULL;
1136
	struct request *req;
1137 1138
	struct bio *bio = NULL;
	void *meta = NULL;
1139 1140
	int ret;

1141
	req = nvme_alloc_request(q, cmd, 0);
1142 1143 1144
	if (IS_ERR(req))
		return PTR_ERR(req);

1145 1146
	if (timeout)
		req->timeout = timeout;
1147
	nvme_req(req)->flags |= NVME_REQ_USERCMD;
1148 1149

	if (ubuffer && bufflen) {
1150 1151 1152 1153 1154
		ret = blk_rq_map_user(q, req, NULL, ubuffer, bufflen,
				GFP_KERNEL);
		if (ret)
			goto out;
		bio = req->bio;
1155
		bio_set_dev(bio, bdev);
1156
		if (bdev && meta_buffer && meta_len) {
1157 1158 1159 1160
			meta = nvme_add_user_metadata(bio, meta_buffer, meta_len,
					meta_seed, write);
			if (IS_ERR(meta)) {
				ret = PTR_ERR(meta);
1161 1162
				goto out_unmap;
			}
1163
			req->cmd_flags |= REQ_INTEGRITY;
1164 1165
		}
	}
1166

1167
	nvme_execute_passthru_rq(req);
1168 1169 1170 1171
	if (nvme_req(req)->flags & NVME_REQ_CANCELLED)
		ret = -EINTR;
	else
		ret = nvme_req(req)->status;
1172
	if (result)
1173
		*result = le64_to_cpu(nvme_req(req)->result.u64);
1174 1175 1176 1177 1178 1179
	if (meta && !ret && !write) {
		if (copy_to_user(meta_buffer, meta, meta_len))
			ret = -EFAULT;
	}
	kfree(meta);
 out_unmap:
1180
	if (bio)
1181
		blk_rq_unmap_user(bio);
1182 1183 1184 1185 1186
 out:
	blk_mq_free_request(req);
	return ret;
}

1187
static void nvme_keep_alive_end_io(struct request *rq, blk_status_t status)
S
Sagi Grimberg 已提交
1188 1189
{
	struct nvme_ctrl *ctrl = rq->end_io_data;
1190 1191
	unsigned long flags;
	bool startka = false;
S
Sagi Grimberg 已提交
1192 1193 1194

	blk_mq_free_request(rq);

1195
	if (status) {
S
Sagi Grimberg 已提交
1196
		dev_err(ctrl->device,
1197 1198
			"failed nvme_keep_alive_end_io error=%d\n",
				status);
S
Sagi Grimberg 已提交
1199 1200 1201
		return;
	}

1202
	ctrl->comp_seen = false;
1203 1204 1205 1206 1207 1208
	spin_lock_irqsave(&ctrl->lock, flags);
	if (ctrl->state == NVME_CTRL_LIVE ||
	    ctrl->state == NVME_CTRL_CONNECTING)
		startka = true;
	spin_unlock_irqrestore(&ctrl->lock, flags);
	if (startka)
1209
		queue_delayed_work(nvme_wq, &ctrl->ka_work, ctrl->kato * HZ);
S
Sagi Grimberg 已提交
1210 1211 1212 1213 1214 1215
}

static int nvme_keep_alive(struct nvme_ctrl *ctrl)
{
	struct request *rq;

1216 1217
	rq = nvme_alloc_request(ctrl->admin_q, &ctrl->ka_cmd,
			BLK_MQ_REQ_RESERVED);
S
Sagi Grimberg 已提交
1218 1219 1220 1221 1222 1223
	if (IS_ERR(rq))
		return PTR_ERR(rq);

	rq->timeout = ctrl->kato * HZ;
	rq->end_io_data = ctrl;

1224
	blk_execute_rq_nowait(NULL, rq, 0, nvme_keep_alive_end_io);
S
Sagi Grimberg 已提交
1225 1226 1227 1228 1229 1230 1231 1232

	return 0;
}

static void nvme_keep_alive_work(struct work_struct *work)
{
	struct nvme_ctrl *ctrl = container_of(to_delayed_work(work),
			struct nvme_ctrl, ka_work);
1233 1234 1235 1236 1237 1238
	bool comp_seen = ctrl->comp_seen;

	if ((ctrl->ctratt & NVME_CTRL_ATTR_TBKAS) && comp_seen) {
		dev_dbg(ctrl->device,
			"reschedule traffic based keep-alive timer\n");
		ctrl->comp_seen = false;
1239
		queue_delayed_work(nvme_wq, &ctrl->ka_work, ctrl->kato * HZ);
1240 1241
		return;
	}
S
Sagi Grimberg 已提交
1242 1243 1244 1245

	if (nvme_keep_alive(ctrl)) {
		/* allocation failure, reset the controller */
		dev_err(ctrl->device, "keep-alive failed\n");
1246
		nvme_reset_ctrl(ctrl);
S
Sagi Grimberg 已提交
1247 1248 1249 1250
		return;
	}
}

1251
static void nvme_start_keep_alive(struct nvme_ctrl *ctrl)
S
Sagi Grimberg 已提交
1252 1253 1254 1255
{
	if (unlikely(ctrl->kato == 0))
		return;

1256
	queue_delayed_work(nvme_wq, &ctrl->ka_work, ctrl->kato * HZ);
S
Sagi Grimberg 已提交
1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267
}

void nvme_stop_keep_alive(struct nvme_ctrl *ctrl)
{
	if (unlikely(ctrl->kato == 0))
		return;

	cancel_delayed_work_sync(&ctrl->ka_work);
}
EXPORT_SYMBOL_GPL(nvme_stop_keep_alive);

1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280
/*
 * In NVMe 1.0 the CNS field was just a binary controller or namespace
 * flag, thus sending any new CNS opcodes has a big chance of not working.
 * Qemu unfortunately had that bug after reporting a 1.1 version compliance
 * (but not for any later version).
 */
static bool nvme_ctrl_limited_cns(struct nvme_ctrl *ctrl)
{
	if (ctrl->quirks & NVME_QUIRK_IDENTIFY_CNS)
		return ctrl->vs < NVME_VS(1, 2, 0);
	return ctrl->vs < NVME_VS(1, 1, 0);
}

K
Keith Busch 已提交
1281
static int nvme_identify_ctrl(struct nvme_ctrl *dev, struct nvme_id_ctrl **id)
1282 1283 1284 1285 1286 1287
{
	struct nvme_command c = { };
	int error;

	/* gcc-4.4.4 (at least) has issues with initializers and anon unions */
	c.identify.opcode = nvme_admin_identify;
1288
	c.identify.cns = NVME_ID_CNS_CTRL;
1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300

	*id = kmalloc(sizeof(struct nvme_id_ctrl), GFP_KERNEL);
	if (!*id)
		return -ENOMEM;

	error = nvme_submit_sync_cmd(dev->admin_q, &c, *id,
			sizeof(struct nvme_id_ctrl));
	if (error)
		kfree(*id);
	return error;
}

1301 1302 1303 1304 1305
static bool nvme_multi_css(struct nvme_ctrl *ctrl)
{
	return (ctrl->ctrl_config & NVME_CC_CSS_MASK) == NVME_CC_CSS_CSI;
}

1306
static int nvme_process_ns_desc(struct nvme_ctrl *ctrl, struct nvme_ns_ids *ids,
1307
		struct nvme_ns_id_desc *cur, bool *csi_seen)
1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336
{
	const char *warn_str = "ctrl returned bogus length:";
	void *data = cur;

	switch (cur->nidt) {
	case NVME_NIDT_EUI64:
		if (cur->nidl != NVME_NIDT_EUI64_LEN) {
			dev_warn(ctrl->device, "%s %d for NVME_NIDT_EUI64\n",
				 warn_str, cur->nidl);
			return -1;
		}
		memcpy(ids->eui64, data + sizeof(*cur), NVME_NIDT_EUI64_LEN);
		return NVME_NIDT_EUI64_LEN;
	case NVME_NIDT_NGUID:
		if (cur->nidl != NVME_NIDT_NGUID_LEN) {
			dev_warn(ctrl->device, "%s %d for NVME_NIDT_NGUID\n",
				 warn_str, cur->nidl);
			return -1;
		}
		memcpy(ids->nguid, data + sizeof(*cur), NVME_NIDT_NGUID_LEN);
		return NVME_NIDT_NGUID_LEN;
	case NVME_NIDT_UUID:
		if (cur->nidl != NVME_NIDT_UUID_LEN) {
			dev_warn(ctrl->device, "%s %d for NVME_NIDT_UUID\n",
				 warn_str, cur->nidl);
			return -1;
		}
		uuid_copy(&ids->uuid, data + sizeof(*cur));
		return NVME_NIDT_UUID_LEN;
1337 1338 1339 1340 1341 1342 1343 1344 1345
	case NVME_NIDT_CSI:
		if (cur->nidl != NVME_NIDT_CSI_LEN) {
			dev_warn(ctrl->device, "%s %d for NVME_NIDT_CSI\n",
				 warn_str, cur->nidl);
			return -1;
		}
		memcpy(&ids->csi, data + sizeof(*cur), NVME_NIDT_CSI_LEN);
		*csi_seen = true;
		return NVME_NIDT_CSI_LEN;
1346 1347 1348 1349 1350 1351
	default:
		/* Skip unknown types */
		return cur->nidl;
	}
}

1352
static int nvme_identify_ns_descs(struct nvme_ctrl *ctrl, unsigned nsid,
1353
		struct nvme_ns_ids *ids)
1354 1355
{
	struct nvme_command c = { };
1356 1357
	bool csi_seen = false;
	int status, pos, len;
1358 1359
	void *data;

1360 1361
	if (ctrl->vs < NVME_VS(1, 3, 0) && !nvme_multi_css(ctrl))
		return 0;
1362 1363 1364
	if (ctrl->quirks & NVME_QUIRK_NO_NS_DESC_LIST)
		return 0;

1365 1366 1367 1368 1369 1370 1371 1372
	c.identify.opcode = nvme_admin_identify;
	c.identify.nsid = cpu_to_le32(nsid);
	c.identify.cns = NVME_ID_CNS_NS_DESC_LIST;

	data = kzalloc(NVME_IDENTIFY_DATA_SIZE, GFP_KERNEL);
	if (!data)
		return -ENOMEM;

1373
	status = nvme_submit_sync_cmd(ctrl->admin_q, &c, data,
1374
				      NVME_IDENTIFY_DATA_SIZE);
1375 1376
	if (status) {
		dev_warn(ctrl->device,
1377 1378
			"Identify Descriptors failed (nsid=%u, status=0x%x)\n",
			nsid, status);
1379
		goto free_data;
1380
	}
1381 1382 1383 1384 1385 1386 1387

	for (pos = 0; pos < NVME_IDENTIFY_DATA_SIZE; pos += len) {
		struct nvme_ns_id_desc *cur = data + pos;

		if (cur->nidl == 0)
			break;

1388
		len = nvme_process_ns_desc(ctrl, ids, cur, &csi_seen);
1389
		if (len < 0)
1390
			break;
1391 1392 1393

		len += sizeof(*cur);
	}
1394 1395 1396 1397 1398 1399 1400

	if (nvme_multi_css(ctrl) && !csi_seen) {
		dev_warn(ctrl->device, "Command set not reported for nsid:%d\n",
			 nsid);
		status = -EINVAL;
	}

1401 1402 1403 1404 1405
free_data:
	kfree(data);
	return status;
}

1406 1407
static int nvme_identify_ns(struct nvme_ctrl *ctrl, unsigned nsid,
			struct nvme_ns_ids *ids, struct nvme_id_ns **id)
1408 1409 1410 1411 1412
{
	struct nvme_command c = { };
	int error;

	/* gcc-4.4.4 (at least) has issues with initializers and anon unions */
1413 1414
	c.identify.opcode = nvme_admin_identify;
	c.identify.nsid = cpu_to_le32(nsid);
1415
	c.identify.cns = NVME_ID_CNS_NS;
1416

1417 1418 1419
	*id = kmalloc(sizeof(**id), GFP_KERNEL);
	if (!*id)
		return -ENOMEM;
1420

1421
	error = nvme_submit_sync_cmd(ctrl->admin_q, &c, *id, sizeof(**id));
1422
	if (error) {
1423
		dev_warn(ctrl->device, "Identify namespace failed (%d)\n", error);
1424
		goto out_free_id;
1425 1426
	}

1427 1428 1429
	error = -ENODEV;
	if ((*id)->ncap == 0) /* namespace not allocated or attached */
		goto out_free_id;
1430 1431 1432 1433 1434 1435 1436 1437

	if (ctrl->vs >= NVME_VS(1, 1, 0) &&
	    !memchr_inv(ids->eui64, 0, sizeof(ids->eui64)))
		memcpy(ids->eui64, (*id)->eui64, sizeof(ids->eui64));
	if (ctrl->vs >= NVME_VS(1, 2, 0) &&
	    !memchr_inv(ids->nguid, 0, sizeof(ids->nguid)))
		memcpy(ids->nguid, (*id)->nguid, sizeof(ids->nguid));

1438 1439 1440 1441
	return 0;

out_free_id:
	kfree(*id);
1442
	return error;
1443 1444
}

K
Keith Busch 已提交
1445 1446
static int nvme_features(struct nvme_ctrl *dev, u8 op, unsigned int fid,
		unsigned int dword11, void *buffer, size_t buflen, u32 *result)
1447
{
1448
	union nvme_result res = { 0 };
1449
	struct nvme_command c;
1450
	int ret;
1451 1452

	memset(&c, 0, sizeof(c));
K
Keith Busch 已提交
1453
	c.features.opcode = op;
1454 1455 1456
	c.features.fid = cpu_to_le32(fid);
	c.features.dword11 = cpu_to_le32(dword11);

1457
	ret = __nvme_submit_sync_cmd(dev->admin_q, &c, &res,
1458
			buffer, buflen, 0, NVME_QID_ANY, 0, 0, false);
1459
	if (ret >= 0 && result)
1460
		*result = le32_to_cpu(res.u32);
1461
	return ret;
1462 1463
}

K
Keith Busch 已提交
1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481
int nvme_set_features(struct nvme_ctrl *dev, unsigned int fid,
		      unsigned int dword11, void *buffer, size_t buflen,
		      u32 *result)
{
	return nvme_features(dev, nvme_admin_set_features, fid, dword11, buffer,
			     buflen, result);
}
EXPORT_SYMBOL_GPL(nvme_set_features);

int nvme_get_features(struct nvme_ctrl *dev, unsigned int fid,
		      unsigned int dword11, void *buffer, size_t buflen,
		      u32 *result)
{
	return nvme_features(dev, nvme_admin_get_features, fid, dword11, buffer,
			     buflen, result);
}
EXPORT_SYMBOL_GPL(nvme_get_features);

C
Christoph Hellwig 已提交
1482 1483 1484 1485 1486 1487
int nvme_set_queue_count(struct nvme_ctrl *ctrl, int *count)
{
	u32 q_count = (*count - 1) | ((*count - 1) << 16);
	u32 result;
	int status, nr_io_queues;

1488
	status = nvme_set_features(ctrl, NVME_FEAT_NUM_QUEUES, q_count, NULL, 0,
C
Christoph Hellwig 已提交
1489
			&result);
1490
	if (status < 0)
C
Christoph Hellwig 已提交
1491 1492
		return status;

1493 1494 1495 1496 1497 1498
	/*
	 * Degraded controllers might return an error when setting the queue
	 * count.  We still want to be able to bring them online and offer
	 * access to the admin queue, as that might be only way to fix them up.
	 */
	if (status > 0) {
1499
		dev_err(ctrl->device, "Could not set queue count (%d)\n", status);
1500 1501 1502 1503 1504 1505
		*count = 0;
	} else {
		nr_io_queues = min(result & 0xffff, result >> 16) + 1;
		*count = min(*count, nr_io_queues);
	}

C
Christoph Hellwig 已提交
1506 1507
	return 0;
}
1508
EXPORT_SYMBOL_GPL(nvme_set_queue_count);
C
Christoph Hellwig 已提交
1509

1510
#define NVME_AEN_SUPPORTED \
1511 1512
	(NVME_AEN_CFG_NS_ATTR | NVME_AEN_CFG_FW_ACT | \
	 NVME_AEN_CFG_ANA_CHANGE | NVME_AEN_CFG_DISC_CHANGE)
1513 1514 1515

static void nvme_enable_aen(struct nvme_ctrl *ctrl)
{
1516
	u32 result, supported_aens = ctrl->oaes & NVME_AEN_SUPPORTED;
1517 1518
	int status;

1519 1520 1521 1522 1523
	if (!supported_aens)
		return;

	status = nvme_set_features(ctrl, NVME_FEAT_ASYNC_EVENT, supported_aens,
			NULL, 0, &result);
1524 1525
	if (status)
		dev_warn(ctrl->device, "Failed to configure AEN (cfg %x)\n",
1526
			 supported_aens);
1527 1528

	queue_work(nvme_wq, &ctrl->async_event_work);
1529 1530
}

1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542
/*
 * Convert integer values from ioctl structures to user pointers, silently
 * ignoring the upper bits in the compat case to match behaviour of 32-bit
 * kernels.
 */
static void __user *nvme_to_user_ptr(uintptr_t ptrval)
{
	if (in_compat_syscall())
		ptrval = (compat_uptr_t)ptrval;
	return (void __user *)ptrval;
}

1543 1544 1545 1546 1547 1548 1549 1550 1551
static int nvme_submit_io(struct nvme_ns *ns, struct nvme_user_io __user *uio)
{
	struct nvme_user_io io;
	struct nvme_command c;
	unsigned length, meta_len;
	void __user *metadata;

	if (copy_from_user(&io, uio, sizeof(io)))
		return -EFAULT;
1552 1553
	if (io.flags)
		return -EINVAL;
1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564

	switch (io.opcode) {
	case nvme_cmd_write:
	case nvme_cmd_read:
	case nvme_cmd_compare:
		break;
	default:
		return -EINVAL;
	}

	length = (io.nblocks + 1) << ns->lba_shift;
1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579

	if ((io.control & NVME_RW_PRINFO_PRACT) &&
	    ns->ms == sizeof(struct t10_pi_tuple)) {
		/*
		 * Protection information is stripped/inserted by the
		 * controller.
		 */
		if (nvme_to_user_ptr(io.metadata))
			return -EINVAL;
		meta_len = 0;
		metadata = NULL;
	} else {
		meta_len = (io.nblocks + 1) * ns->ms;
		metadata = nvme_to_user_ptr(io.metadata);
	}
1580

1581
	if (ns->features & NVME_NS_EXT_LBAS) {
1582 1583 1584 1585 1586 1587 1588 1589 1590 1591
		length += meta_len;
		meta_len = 0;
	} else if (meta_len) {
		if ((io.metadata & 3) || !io.metadata)
			return -EINVAL;
	}

	memset(&c, 0, sizeof(c));
	c.rw.opcode = io.opcode;
	c.rw.flags = io.flags;
C
Christoph Hellwig 已提交
1592
	c.rw.nsid = cpu_to_le32(ns->head->ns_id);
1593 1594 1595 1596 1597 1598 1599 1600
	c.rw.slba = cpu_to_le64(io.slba);
	c.rw.length = cpu_to_le16(io.nblocks);
	c.rw.control = cpu_to_le16(io.control);
	c.rw.dsmgmt = cpu_to_le32(io.dsmgmt);
	c.rw.reftag = cpu_to_le32(io.reftag);
	c.rw.apptag = cpu_to_le16(io.apptag);
	c.rw.appmask = cpu_to_le16(io.appmask);

1601
	return nvme_submit_user_cmd(ns->queue, &c,
1602
			nvme_to_user_ptr(io.addr), length,
1603
			metadata, meta_len, lower_32_bits(io.slba), NULL, 0);
1604 1605
}

1606
static int nvme_user_cmd(struct nvme_ctrl *ctrl, struct nvme_ns *ns,
1607 1608 1609 1610 1611
			struct nvme_passthru_cmd __user *ucmd)
{
	struct nvme_passthru_cmd cmd;
	struct nvme_command c;
	unsigned timeout = 0;
1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638
	u64 result;
	int status;

	if (!capable(CAP_SYS_ADMIN))
		return -EACCES;
	if (copy_from_user(&cmd, ucmd, sizeof(cmd)))
		return -EFAULT;
	if (cmd.flags)
		return -EINVAL;

	memset(&c, 0, sizeof(c));
	c.common.opcode = cmd.opcode;
	c.common.flags = cmd.flags;
	c.common.nsid = cpu_to_le32(cmd.nsid);
	c.common.cdw2[0] = cpu_to_le32(cmd.cdw2);
	c.common.cdw2[1] = cpu_to_le32(cmd.cdw3);
	c.common.cdw10 = cpu_to_le32(cmd.cdw10);
	c.common.cdw11 = cpu_to_le32(cmd.cdw11);
	c.common.cdw12 = cpu_to_le32(cmd.cdw12);
	c.common.cdw13 = cpu_to_le32(cmd.cdw13);
	c.common.cdw14 = cpu_to_le32(cmd.cdw14);
	c.common.cdw15 = cpu_to_le32(cmd.cdw15);

	if (cmd.timeout_ms)
		timeout = msecs_to_jiffies(cmd.timeout_ms);

	status = nvme_submit_user_cmd(ns ? ns->queue : ctrl->admin_q, &c,
1639 1640 1641
			nvme_to_user_ptr(cmd.addr), cmd.data_len,
			nvme_to_user_ptr(cmd.metadata), cmd.metadata_len,
			0, &result, timeout);
1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656

	if (status >= 0) {
		if (put_user(result, &ucmd->result))
			return -EFAULT;
	}

	return status;
}

static int nvme_user_cmd64(struct nvme_ctrl *ctrl, struct nvme_ns *ns,
			struct nvme_passthru_cmd64 __user *ucmd)
{
	struct nvme_passthru_cmd64 cmd;
	struct nvme_command c;
	unsigned timeout = 0;
1657 1658 1659 1660 1661 1662
	int status;

	if (!capable(CAP_SYS_ADMIN))
		return -EACCES;
	if (copy_from_user(&cmd, ucmd, sizeof(cmd)))
		return -EFAULT;
1663 1664
	if (cmd.flags)
		return -EINVAL;
1665 1666 1667 1668 1669 1670 1671

	memset(&c, 0, sizeof(c));
	c.common.opcode = cmd.opcode;
	c.common.flags = cmd.flags;
	c.common.nsid = cpu_to_le32(cmd.nsid);
	c.common.cdw2[0] = cpu_to_le32(cmd.cdw2);
	c.common.cdw2[1] = cpu_to_le32(cmd.cdw3);
1672 1673 1674 1675 1676 1677
	c.common.cdw10 = cpu_to_le32(cmd.cdw10);
	c.common.cdw11 = cpu_to_le32(cmd.cdw11);
	c.common.cdw12 = cpu_to_le32(cmd.cdw12);
	c.common.cdw13 = cpu_to_le32(cmd.cdw13);
	c.common.cdw14 = cpu_to_le32(cmd.cdw14);
	c.common.cdw15 = cpu_to_le32(cmd.cdw15);
1678 1679 1680 1681 1682

	if (cmd.timeout_ms)
		timeout = msecs_to_jiffies(cmd.timeout_ms);

	status = nvme_submit_user_cmd(ns ? ns->queue : ctrl->admin_q, &c,
1683 1684
			nvme_to_user_ptr(cmd.addr), cmd.data_len,
			nvme_to_user_ptr(cmd.metadata), cmd.metadata_len,
1685
			0, &cmd.result, timeout);
1686

1687 1688 1689 1690 1691 1692 1693 1694
	if (status >= 0) {
		if (put_user(cmd.result, &ucmd->result))
			return -EFAULT;
	}

	return status;
}

1695 1696 1697 1698
/*
 * Issue ioctl requests on the first available path.  Note that unlike normal
 * block layer requests we will not retry failed request on another controller.
 */
K
Keith Busch 已提交
1699
struct nvme_ns *nvme_get_ns_from_disk(struct gendisk *disk,
1700
		struct nvme_ns_head **head, int *srcu_idx)
1701
{
1702 1703
#ifdef CONFIG_NVME_MULTIPATH
	if (disk->fops == &nvme_ns_head_ops) {
1704 1705
		struct nvme_ns *ns;

1706 1707
		*head = disk->private_data;
		*srcu_idx = srcu_read_lock(&(*head)->srcu);
1708 1709 1710 1711
		ns = nvme_find_path(*head);
		if (!ns)
			srcu_read_unlock(&(*head)->srcu, *srcu_idx);
		return ns;
1712 1713 1714 1715 1716 1717
	}
#endif
	*head = NULL;
	*srcu_idx = -1;
	return disk->private_data;
}
1718

K
Keith Busch 已提交
1719
void nvme_put_ns_from_disk(struct nvme_ns_head *head, int idx)
1720 1721 1722 1723
{
	if (head)
		srcu_read_unlock(&head->srcu, idx);
}
1724

1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759
static bool is_ctrl_ioctl(unsigned int cmd)
{
	if (cmd == NVME_IOCTL_ADMIN_CMD || cmd == NVME_IOCTL_ADMIN64_CMD)
		return true;
	if (is_sed_ioctl(cmd))
		return true;
	return false;
}

static int nvme_handle_ctrl_ioctl(struct nvme_ns *ns, unsigned int cmd,
				  void __user *argp,
				  struct nvme_ns_head *head,
				  int srcu_idx)
{
	struct nvme_ctrl *ctrl = ns->ctrl;
	int ret;

	nvme_get_ctrl(ns->ctrl);
	nvme_put_ns_from_disk(head, srcu_idx);

	switch (cmd) {
	case NVME_IOCTL_ADMIN_CMD:
		ret = nvme_user_cmd(ctrl, NULL, argp);
		break;
	case NVME_IOCTL_ADMIN64_CMD:
		ret = nvme_user_cmd64(ctrl, NULL, argp);
		break;
	default:
		ret = sed_ioctl(ctrl->opal_dev, cmd, argp);
		break;
	}
	nvme_put_ctrl(ctrl);
	return ret;
}

1760 1761
static int nvme_ioctl(struct block_device *bdev, fmode_t mode,
		unsigned int cmd, unsigned long arg)
1762
{
1763
	struct nvme_ns_head *head = NULL;
1764
	void __user *argp = (void __user *)arg;
1765 1766 1767 1768 1769
	struct nvme_ns *ns;
	int srcu_idx, ret;

	ns = nvme_get_ns_from_disk(bdev->bd_disk, &head, &srcu_idx);
	if (unlikely(!ns))
1770 1771
		return -EWOULDBLOCK;

1772 1773 1774 1775 1776
	/*
	 * Handle ioctls that apply to the controller instead of the namespace
	 * seperately and drop the ns SRCU reference early.  This avoids a
	 * deadlock when deleting namespaces using the passthrough interface.
	 */
1777 1778
	if (is_ctrl_ioctl(cmd))
		return nvme_handle_ctrl_ioctl(ns, cmd, argp, head, srcu_idx);
1779

1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790
	switch (cmd) {
	case NVME_IOCTL_ID:
		force_successful_syscall_return();
		ret = ns->head->ns_id;
		break;
	case NVME_IOCTL_IO_CMD:
		ret = nvme_user_cmd(ns->ctrl, ns, argp);
		break;
	case NVME_IOCTL_SUBMIT_IO:
		ret = nvme_submit_io(ns, argp);
		break;
1791 1792 1793
	case NVME_IOCTL_IO64_CMD:
		ret = nvme_user_cmd64(ns->ctrl, ns, argp);
		break;
1794 1795 1796 1797 1798 1799 1800
	default:
		if (ns->ndev)
			ret = nvme_nvm_ioctl(ns, cmd, arg);
		else
			ret = -ENOTTY;
	}

1801 1802
	nvme_put_ns_from_disk(head, srcu_idx);
	return ret;
1803 1804
}

1805 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 1840 1841 1842 1843 1844 1845
#ifdef CONFIG_COMPAT
struct nvme_user_io32 {
	__u8	opcode;
	__u8	flags;
	__u16	control;
	__u16	nblocks;
	__u16	rsvd;
	__u64	metadata;
	__u64	addr;
	__u64	slba;
	__u32	dsmgmt;
	__u32	reftag;
	__u16	apptag;
	__u16	appmask;
} __attribute__((__packed__));

#define NVME_IOCTL_SUBMIT_IO32	_IOW('N', 0x42, struct nvme_user_io32)

static int nvme_compat_ioctl(struct block_device *bdev, fmode_t mode,
		unsigned int cmd, unsigned long arg)
{
	/*
	 * Corresponds to the difference of NVME_IOCTL_SUBMIT_IO
	 * between 32 bit programs and 64 bit kernel.
	 * The cause is that the results of sizeof(struct nvme_user_io),
	 * which is used to define NVME_IOCTL_SUBMIT_IO,
	 * are not same between 32 bit compiler and 64 bit compiler.
	 * NVME_IOCTL_SUBMIT_IO32 is for 64 bit kernel handling
	 * NVME_IOCTL_SUBMIT_IO issued from 32 bit programs.
	 * Other IOCTL numbers are same between 32 bit and 64 bit.
	 * So there is nothing to do regarding to other IOCTL numbers.
	 */
	if (cmd == NVME_IOCTL_SUBMIT_IO32)
		return nvme_ioctl(bdev, mode, NVME_IOCTL_SUBMIT_IO, arg);

	return nvme_ioctl(bdev, mode, cmd, arg);
}
#else
#define nvme_compat_ioctl	NULL
#endif /* CONFIG_COMPAT */

1846 1847
static int nvme_open(struct block_device *bdev, fmode_t mode)
{
C
Christoph Hellwig 已提交
1848 1849
	struct nvme_ns *ns = bdev->bd_disk->private_data;

1850 1851 1852
#ifdef CONFIG_NVME_MULTIPATH
	/* should never be called due to GENHD_FL_HIDDEN */
	if (WARN_ON_ONCE(ns->head->disk))
1853
		goto fail;
1854
#endif
C
Christoph Hellwig 已提交
1855
	if (!kref_get_unless_zero(&ns->kref))
1856 1857 1858 1859
		goto fail;
	if (!try_module_get(ns->ctrl->ops->module))
		goto fail_put_ns;

C
Christoph Hellwig 已提交
1860
	return 0;
1861 1862 1863 1864 1865

fail_put_ns:
	nvme_put_ns(ns);
fail:
	return -ENXIO;
1866 1867 1868 1869
}

static void nvme_release(struct gendisk *disk, fmode_t mode)
{
1870 1871 1872 1873
	struct nvme_ns *ns = disk->private_data;

	module_put(ns->ctrl->ops->module);
	nvme_put_ns(ns);
1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885
}

static int nvme_getgeo(struct block_device *bdev, struct hd_geometry *geo)
{
	/* some standard values */
	geo->heads = 1 << 6;
	geo->sectors = 1 << 5;
	geo->cylinders = get_capacity(bdev->bd_disk) >> 11;
	return 0;
}

#ifdef CONFIG_BLK_DEV_INTEGRITY
1886 1887
static void nvme_init_integrity(struct gendisk *disk, u16 ms, u8 pi_type,
				u32 max_integrity_segments)
1888 1889 1890
{
	struct blk_integrity integrity;

1891
	memset(&integrity, 0, sizeof(integrity));
1892
	switch (pi_type) {
1893 1894
	case NVME_NS_DPS_PI_TYPE3:
		integrity.profile = &t10_pi_type3_crc;
1895 1896
		integrity.tag_size = sizeof(u16) + sizeof(u32);
		integrity.flags |= BLK_INTEGRITY_DEVICE_CAPABLE;
1897 1898 1899 1900
		break;
	case NVME_NS_DPS_PI_TYPE1:
	case NVME_NS_DPS_PI_TYPE2:
		integrity.profile = &t10_pi_type1_crc;
1901 1902
		integrity.tag_size = sizeof(u16);
		integrity.flags |= BLK_INTEGRITY_DEVICE_CAPABLE;
1903 1904 1905 1906 1907
		break;
	default:
		integrity.profile = NULL;
		break;
	}
1908 1909
	integrity.tuple_size = ms;
	blk_integrity_register(disk, &integrity);
1910
	blk_queue_max_integrity_segments(disk->queue, max_integrity_segments);
1911 1912
}
#else
1913 1914
static void nvme_init_integrity(struct gendisk *disk, u16 ms, u8 pi_type,
				u32 max_integrity_segments)
1915 1916 1917 1918
{
}
#endif /* CONFIG_BLK_DEV_INTEGRITY */

1919
static void nvme_config_discard(struct gendisk *disk, struct nvme_ns *ns)
1920
{
1921
	struct nvme_ctrl *ctrl = ns->ctrl;
1922
	struct request_queue *queue = disk->queue;
1923 1924
	u32 size = queue_logical_block_size(queue);

1925 1926 1927 1928 1929 1930 1931
	if (!(ctrl->oncs & NVME_CTRL_ONCS_DSM)) {
		blk_queue_flag_clear(QUEUE_FLAG_DISCARD, queue);
		return;
	}

	if (ctrl->nr_streams && ns->sws && ns->sgs)
		size *= ns->sws * ns->sgs;
1932

1933 1934 1935
	BUILD_BUG_ON(PAGE_SIZE / sizeof(struct nvme_dsm_range) <
			NVME_DSM_MAX_RANGES);

1936
	queue->limits.discard_alignment = 0;
1937
	queue->limits.discard_granularity = size;
1938

1939 1940 1941 1942
	/* If discard is already enabled, don't reset queue limits */
	if (blk_queue_flag_test_and_set(QUEUE_FLAG_DISCARD, queue))
		return;

1943 1944
	blk_queue_max_discard_sectors(queue, UINT_MAX);
	blk_queue_max_discard_segments(queue, NVME_DSM_MAX_RANGES);
1945 1946

	if (ctrl->quirks & NVME_QUIRK_DEALLOCATE_ZEROES)
1947
		blk_queue_max_write_zeroes_sectors(queue, UINT_MAX);
1948 1949
}

1950
static void nvme_config_write_zeroes(struct gendisk *disk, struct nvme_ns *ns)
1951
{
1952
	u64 max_blocks;
1953

1954 1955
	if (!(ns->ctrl->oncs & NVME_CTRL_ONCS_WRITE_ZEROES) ||
	    (ns->ctrl->quirks & NVME_QUIRK_DISABLE_WRITE_ZEROES))
1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967
		return;
	/*
	 * Even though NVMe spec explicitly states that MDTS is not
	 * applicable to the write-zeroes:- "The restriction does not apply to
	 * commands that do not transfer data between the host and the
	 * controller (e.g., Write Uncorrectable ro Write Zeroes command).".
	 * In order to be more cautious use controller's max_hw_sectors value
	 * to configure the maximum sectors for the write-zeroes which is
	 * configured based on the controller's MDTS field in the
	 * nvme_init_identify() if available.
	 */
	if (ns->ctrl->max_hw_sectors == UINT_MAX)
1968
		max_blocks = (u64)USHRT_MAX + 1;
1969
	else
1970
		max_blocks = ns->ctrl->max_hw_sectors + 1;
1971

1972 1973
	blk_queue_max_write_zeroes_sectors(disk->queue,
					   nvme_lba_to_sect(ns, max_blocks));
1974 1975
}

C
Christoph Hellwig 已提交
1976 1977 1978 1979 1980 1981 1982
static bool nvme_ns_ids_valid(struct nvme_ns_ids *ids)
{
	return !uuid_is_null(&ids->uuid) ||
		memchr_inv(ids->nguid, 0, sizeof(ids->nguid)) ||
		memchr_inv(ids->eui64, 0, sizeof(ids->eui64));
}

1983 1984 1985 1986
static bool nvme_ns_ids_equal(struct nvme_ns_ids *a, struct nvme_ns_ids *b)
{
	return uuid_equal(&a->uuid, &b->uuid) &&
		memcmp(&a->nguid, &b->nguid, sizeof(a->nguid)) == 0 &&
1987 1988
		memcmp(&a->eui64, &b->eui64, sizeof(a->eui64)) == 0 &&
		a->csi == b->csi;
1989 1990
}

K
Keith Busch 已提交
1991 1992
static int nvme_setup_streams_ns(struct nvme_ctrl *ctrl, struct nvme_ns *ns,
				 u32 *phys_bs, u32 *io_opt)
1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007
{
	struct streams_directive_params s;
	int ret;

	if (!ctrl->nr_streams)
		return 0;

	ret = nvme_get_stream_params(ctrl, &s, ns->head->ns_id);
	if (ret)
		return ret;

	ns->sws = le32_to_cpu(s.sws);
	ns->sgs = le16_to_cpu(s.sgs);

	if (ns->sws) {
K
Keith Busch 已提交
2008
		*phys_bs = ns->sws * (1 << ns->lba_shift);
2009
		if (ns->sgs)
K
Keith Busch 已提交
2010
			*io_opt = *phys_bs * ns->sgs;
2011 2012 2013 2014 2015
	}

	return 0;
}

2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059
static int nvme_configure_metadata(struct nvme_ns *ns, struct nvme_id_ns *id)
{
	struct nvme_ctrl *ctrl = ns->ctrl;

	/*
	 * The PI implementation requires the metadata size to be equal to the
	 * t10 pi tuple size.
	 */
	ns->ms = le16_to_cpu(id->lbaf[id->flbas & NVME_NS_FLBAS_LBA_MASK].ms);
	if (ns->ms == sizeof(struct t10_pi_tuple))
		ns->pi_type = id->dps & NVME_NS_DPS_PI_MASK;
	else
		ns->pi_type = 0;

	ns->features &= ~(NVME_NS_METADATA_SUPPORTED | NVME_NS_EXT_LBAS);
	if (!ns->ms || !(ctrl->ops->flags & NVME_F_METADATA_SUPPORTED))
		return 0;
	if (ctrl->ops->flags & NVME_F_FABRICS) {
		/*
		 * The NVMe over Fabrics specification only supports metadata as
		 * part of the extended data LBA.  We rely on HCA/HBA support to
		 * remap the separate metadata buffer from the block layer.
		 */
		if (WARN_ON_ONCE(!(id->flbas & NVME_NS_FLBAS_META_EXT)))
			return -EINVAL;
		if (ctrl->max_integrity_segments)
			ns->features |=
				(NVME_NS_METADATA_SUPPORTED | NVME_NS_EXT_LBAS);
	} else {
		/*
		 * For PCIe controllers, we can't easily remap the separate
		 * metadata buffer from the block layer and thus require a
		 * separate metadata buffer for block layer metadata/PI support.
		 * We allow extended LBAs for the passthrough interface, though.
		 */
		if (id->flbas & NVME_NS_FLBAS_META_EXT)
			ns->features |= NVME_NS_EXT_LBAS;
		else
			ns->features |= NVME_NS_METADATA_SUPPORTED;
	}

	return 0;
}

2060 2061 2062
static void nvme_set_queue_limits(struct nvme_ctrl *ctrl,
		struct request_queue *q)
{
2063
	bool vwc = ctrl->vwc & NVME_CTRL_VWC_PRESENT;
2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077

	if (ctrl->max_hw_sectors) {
		u32 max_segments =
			(ctrl->max_hw_sectors / (NVME_CTRL_PAGE_SIZE >> 9)) + 1;

		max_segments = min_not_zero(max_segments, ctrl->max_segments);
		blk_queue_max_hw_sectors(q, ctrl->max_hw_sectors);
		blk_queue_max_segments(q, min_t(u32, max_segments, USHRT_MAX));
	}
	blk_queue_virt_boundary(q, NVME_CTRL_PAGE_SIZE - 1);
	blk_queue_dma_alignment(q, 7);
	blk_queue_write_cache(q, vwc, vwc);
}

2078 2079 2080
static void nvme_update_disk_info(struct gendisk *disk,
		struct nvme_ns *ns, struct nvme_id_ns *id)
{
2081
	sector_t capacity = nvme_lba_to_sect(ns, le64_to_cpu(id->nsze));
2082
	unsigned short bs = 1 << ns->lba_shift;
D
Damien Le Moal 已提交
2083
	u32 atomic_bs, phys_bs, io_opt = 0;
2084

2085 2086 2087 2088
	/*
	 * The block layer can't support LBA sizes larger than the page size
	 * yet, so catch this early and don't allow block I/O.
	 */
2089
	if (ns->lba_shift > PAGE_SHIFT) {
2090
		capacity = 0;
2091 2092
		bs = (1 << 9);
	}
2093

2094 2095
	blk_integrity_unregister(disk);

D
Damien Le Moal 已提交
2096
	atomic_bs = phys_bs = bs;
K
Keith Busch 已提交
2097
	nvme_setup_streams_ns(ns->ctrl, ns, &phys_bs, &io_opt);
2098 2099 2100 2101 2102 2103
	if (id->nabo == 0) {
		/*
		 * Bit 1 indicates whether NAWUPF is defined for this namespace
		 * and whether it should be used instead of AWUPF. If NAWUPF ==
		 * 0 then AWUPF must be used instead.
		 */
2104
		if (id->nsfeat & NVME_NS_FEAT_ATOMICS && id->nawupf)
2105 2106 2107 2108
			atomic_bs = (1 + le16_to_cpu(id->nawupf)) * bs;
		else
			atomic_bs = (1 + ns->ctrl->subsys->awupf) * bs;
	}
K
Keith Busch 已提交
2109

2110
	if (id->nsfeat & NVME_NS_FEAT_IO_OPT) {
2111
		/* NPWG = Namespace Preferred Write Granularity */
K
Keith Busch 已提交
2112
		phys_bs = bs * (1 + le16_to_cpu(id->npwg));
2113
		/* NOWS = Namespace Optimal Write Size */
K
Keith Busch 已提交
2114
		io_opt = bs * (1 + le16_to_cpu(id->nows));
2115 2116
	}

2117
	blk_queue_logical_block_size(disk->queue, bs);
2118 2119 2120 2121 2122 2123 2124 2125
	/*
	 * Linux filesystems assume writing a single physical block is
	 * an atomic operation. Hence limit the physical block size to the
	 * value of the Atomic Write Unit Power Fail parameter.
	 */
	blk_queue_physical_block_size(disk->queue, min(phys_bs, atomic_bs));
	blk_queue_io_min(disk->queue, phys_bs);
	blk_queue_io_opt(disk->queue, io_opt);
2126

2127 2128 2129 2130 2131 2132 2133 2134 2135
	/*
	 * Register a metadata profile for PI, or the plain non-integrity NVMe
	 * metadata masquerading as Type 0 if supported, otherwise reject block
	 * I/O to namespaces with metadata except when the namespace supports
	 * PI, as it can strip/insert in that case.
	 */
	if (ns->ms) {
		if (IS_ENABLED(CONFIG_BLK_DEV_INTEGRITY) &&
		    (ns->features & NVME_NS_METADATA_SUPPORTED))
2136 2137
			nvme_init_integrity(disk, ns->ms, ns->pi_type,
					    ns->ctrl->max_integrity_segments);
2138 2139 2140 2141
		else if (!nvme_ns_has_pi(ns))
			capacity = 0;
	}

2142
	set_capacity_and_notify(disk, capacity);
2143

2144
	nvme_config_discard(disk, ns);
2145
	nvme_config_write_zeroes(disk, ns);
2146

2147 2148
	set_disk_ro(disk, (id->nsattr & NVME_NS_ATTR_RO) ||
		test_bit(NVME_NS_FORCE_RO, &ns->flags));
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 2179 2180 2181 2182 2183 2184 2185 2186 2187
static inline bool nvme_first_scan(struct gendisk *disk)
{
	/* nvme_alloc_ns() scans the disk prior to adding it */
	return !(disk->flags & GENHD_FL_UP);
}

static void nvme_set_chunk_sectors(struct nvme_ns *ns, struct nvme_id_ns *id)
{
	struct nvme_ctrl *ctrl = ns->ctrl;
	u32 iob;

	if ((ctrl->quirks & NVME_QUIRK_STRIPE_SIZE) &&
	    is_power_of_2(ctrl->max_hw_sectors))
		iob = ctrl->max_hw_sectors;
	else
		iob = nvme_lba_to_sect(ns, le16_to_cpu(id->noiob));

	if (!iob)
		return;

	if (!is_power_of_2(iob)) {
		if (nvme_first_scan(ns->disk))
			pr_warn("%s: ignoring unaligned IO boundary:%u\n",
				ns->disk->disk_name, iob);
		return;
	}

	if (blk_queue_is_zoned(ns->disk->queue)) {
		if (nvme_first_scan(ns->disk))
			pr_warn("%s: ignoring zoned namespace IO boundary\n",
				ns->disk->disk_name);
		return;
	}

	blk_queue_chunk_sectors(ns->queue, iob);
}

2188
static int nvme_update_ns_info(struct nvme_ns *ns, struct nvme_id_ns *id)
2189
{
K
Keith Busch 已提交
2190 2191
	unsigned lbaf = id->flbas & NVME_NS_FLBAS_LBA_MASK;
	int ret;
2192

2193
	blk_mq_freeze_queue(ns->disk->queue);
K
Keith Busch 已提交
2194
	ns->lba_shift = id->lbaf[lbaf].ds;
2195
	nvme_set_queue_limits(ns->ctrl, ns->queue);
2196

2197
	if (ns->head->ids.csi == NVME_CSI_ZNS) {
2198
		ret = nvme_update_zone_info(ns, lbaf);
2199
		if (ret)
2200
			goto out_unfreeze;
2201 2202
	}

2203 2204
	ret = nvme_configure_metadata(ns, id);
	if (ret)
2205
		goto out_unfreeze;
2206
	nvme_set_chunk_sectors(ns, id);
2207
	nvme_update_disk_info(ns->disk, ns, id);
2208
	blk_mq_unfreeze_queue(ns->disk->queue);
2209

2210 2211
	if (blk_queue_is_zoned(ns->queue)) {
		ret = nvme_revalidate_zones(ns);
2212
		if (ret && !nvme_first_scan(ns->disk))
2213
			return ret;
2214 2215
	}

2216
#ifdef CONFIG_NVME_MULTIPATH
2217
	if (ns->head->disk) {
2218
		blk_mq_freeze_queue(ns->head->disk->queue);
2219
		nvme_update_disk_info(ns->head->disk, ns, id);
2220 2221
		blk_stack_limits(&ns->head->disk->queue->limits,
				 &ns->queue->limits, 0);
2222
		blk_queue_update_readahead(ns->head->disk->queue);
2223
		blk_mq_unfreeze_queue(ns->head->disk->queue);
2224
	}
2225
#endif
2226
	return 0;
2227

2228 2229
out_unfreeze:
	blk_mq_unfreeze_queue(ns->disk->queue);
K
Keith Busch 已提交
2230 2231 2232
	return ret;
}

2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255
static char nvme_pr_type(enum pr_type type)
{
	switch (type) {
	case PR_WRITE_EXCLUSIVE:
		return 1;
	case PR_EXCLUSIVE_ACCESS:
		return 2;
	case PR_WRITE_EXCLUSIVE_REG_ONLY:
		return 3;
	case PR_EXCLUSIVE_ACCESS_REG_ONLY:
		return 4;
	case PR_WRITE_EXCLUSIVE_ALL_REGS:
		return 5;
	case PR_EXCLUSIVE_ACCESS_ALL_REGS:
		return 6;
	default:
		return 0;
	}
};

static int nvme_pr_command(struct block_device *bdev, u32 cdw10,
				u64 key, u64 sa_key, u8 op)
{
2256 2257
	struct nvme_ns_head *head = NULL;
	struct nvme_ns *ns;
2258
	struct nvme_command c;
2259
	int srcu_idx, ret;
2260 2261
	u8 data[16] = { 0, };

2262 2263 2264 2265
	ns = nvme_get_ns_from_disk(bdev->bd_disk, &head, &srcu_idx);
	if (unlikely(!ns))
		return -EWOULDBLOCK;

2266 2267 2268 2269 2270
	put_unaligned_le64(key, &data[0]);
	put_unaligned_le64(sa_key, &data[8]);

	memset(&c, 0, sizeof(c));
	c.common.opcode = op;
2271
	c.common.nsid = cpu_to_le32(ns->head->ns_id);
2272
	c.common.cdw10 = cpu_to_le32(cdw10);
2273

2274
	ret = nvme_submit_sync_cmd(ns->queue, &c, data, 16);
2275 2276
	nvme_put_ns_from_disk(head, srcu_idx);
	return ret;
2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308
}

static int nvme_pr_register(struct block_device *bdev, u64 old,
		u64 new, unsigned flags)
{
	u32 cdw10;

	if (flags & ~PR_FL_IGNORE_KEY)
		return -EOPNOTSUPP;

	cdw10 = old ? 2 : 0;
	cdw10 |= (flags & PR_FL_IGNORE_KEY) ? 1 << 3 : 0;
	cdw10 |= (1 << 30) | (1 << 31); /* PTPL=1 */
	return nvme_pr_command(bdev, cdw10, old, new, nvme_cmd_resv_register);
}

static int nvme_pr_reserve(struct block_device *bdev, u64 key,
		enum pr_type type, unsigned flags)
{
	u32 cdw10;

	if (flags & ~PR_FL_IGNORE_KEY)
		return -EOPNOTSUPP;

	cdw10 = nvme_pr_type(type) << 8;
	cdw10 |= ((flags & PR_FL_IGNORE_KEY) ? 1 << 3 : 0);
	return nvme_pr_command(bdev, cdw10, key, 0, nvme_cmd_resv_acquire);
}

static int nvme_pr_preempt(struct block_device *bdev, u64 old, u64 new,
		enum pr_type type, bool abort)
{
2309
	u32 cdw10 = nvme_pr_type(type) << 8 | (abort ? 2 : 1);
2310 2311 2312 2313 2314
	return nvme_pr_command(bdev, cdw10, old, new, nvme_cmd_resv_acquire);
}

static int nvme_pr_clear(struct block_device *bdev, u64 key)
{
2315
	u32 cdw10 = 1 | (key ? 1 << 3 : 0);
2316 2317 2318 2319 2320
	return nvme_pr_command(bdev, cdw10, key, 0, nvme_cmd_resv_register);
}

static int nvme_pr_release(struct block_device *bdev, u64 key, enum pr_type type)
{
2321
	u32 cdw10 = nvme_pr_type(type) << 8 | (key ? 1 << 3 : 0);
2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332
	return nvme_pr_command(bdev, cdw10, key, 0, nvme_cmd_resv_release);
}

static const struct pr_ops nvme_pr_ops = {
	.pr_register	= nvme_pr_register,
	.pr_reserve	= nvme_pr_reserve,
	.pr_release	= nvme_pr_release,
	.pr_preempt	= nvme_pr_preempt,
	.pr_clear	= nvme_pr_clear,
};

2333
#ifdef CONFIG_BLK_SED_OPAL
2334 2335
int nvme_sec_submit(void *data, u16 spsp, u8 secp, void *buffer, size_t len,
		bool send)
2336
{
2337
	struct nvme_ctrl *ctrl = data;
2338 2339 2340 2341 2342 2343 2344 2345
	struct nvme_command cmd;

	memset(&cmd, 0, sizeof(cmd));
	if (send)
		cmd.common.opcode = nvme_admin_security_send;
	else
		cmd.common.opcode = nvme_admin_security_recv;
	cmd.common.nsid = 0;
2346 2347
	cmd.common.cdw10 = cpu_to_le32(((u32)secp) << 24 | ((u32)spsp) << 8);
	cmd.common.cdw11 = cpu_to_le32(len);
2348

2349 2350
	return __nvme_submit_sync_cmd(ctrl->admin_q, &cmd, NULL, buffer, len, 0,
			NVME_QID_ANY, 1, 0, false);
2351 2352 2353 2354
}
EXPORT_SYMBOL_GPL(nvme_sec_submit);
#endif /* CONFIG_BLK_SED_OPAL */

J
Javier González 已提交
2355
static const struct block_device_operations nvme_bdev_ops = {
2356 2357
	.owner		= THIS_MODULE,
	.ioctl		= nvme_ioctl,
2358
	.compat_ioctl	= nvme_compat_ioctl,
2359 2360 2361
	.open		= nvme_open,
	.release	= nvme_release,
	.getgeo		= nvme_getgeo,
K
Keith Busch 已提交
2362
	.report_zones	= nvme_report_zones,
2363 2364 2365
	.pr_ops		= &nvme_pr_ops,
};

2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382
#ifdef CONFIG_NVME_MULTIPATH
static int nvme_ns_head_open(struct block_device *bdev, fmode_t mode)
{
	struct nvme_ns_head *head = bdev->bd_disk->private_data;

	if (!kref_get_unless_zero(&head->ref))
		return -ENXIO;
	return 0;
}

static void nvme_ns_head_release(struct gendisk *disk, fmode_t mode)
{
	nvme_put_ns_head(disk->private_data);
}

const struct block_device_operations nvme_ns_head_ops = {
	.owner		= THIS_MODULE,
2383
	.submit_bio	= nvme_ns_head_submit_bio,
2384 2385 2386
	.open		= nvme_ns_head_open,
	.release	= nvme_ns_head_release,
	.ioctl		= nvme_ioctl,
2387
	.compat_ioctl	= nvme_compat_ioctl,
2388
	.getgeo		= nvme_getgeo,
K
Keith Busch 已提交
2389
	.report_zones	= nvme_report_zones,
2390 2391 2392 2393
	.pr_ops		= &nvme_pr_ops,
};
#endif /* CONFIG_NVME_MULTIPATH */

2394 2395 2396 2397 2398 2399 2400 2401
static int nvme_wait_ready(struct nvme_ctrl *ctrl, u64 cap, bool enabled)
{
	unsigned long timeout =
		((NVME_CAP_TIMEOUT(cap) + 1) * HZ / 2) + jiffies;
	u32 csts, bit = enabled ? NVME_CSTS_RDY : 0;
	int ret;

	while ((ret = ctrl->ops->reg_read32(ctrl, NVME_REG_CSTS, &csts)) == 0) {
K
Keith Busch 已提交
2402 2403
		if (csts == ~0)
			return -ENODEV;
2404 2405 2406
		if ((csts & NVME_CSTS_RDY) == bit)
			break;

2407
		usleep_range(1000, 2000);
2408 2409 2410
		if (fatal_signal_pending(current))
			return -EINTR;
		if (time_after(jiffies, timeout)) {
2411
			dev_err(ctrl->device,
2412 2413
				"Device not ready; aborting %s, CSTS=0x%x\n",
				enabled ? "initialisation" : "reset", csts);
2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426
			return -ENODEV;
		}
	}

	return ret;
}

/*
 * If the device has been passed off to us in an enabled state, just clear
 * the enabled bit.  The spec says we should set the 'shutdown notification
 * bits', but doing so may cause the device to complete commands to the
 * admin queue ... and we don't know what memory that might be pointing at!
 */
2427
int nvme_disable_ctrl(struct nvme_ctrl *ctrl)
2428 2429 2430 2431 2432 2433 2434 2435 2436
{
	int ret;

	ctrl->ctrl_config &= ~NVME_CC_SHN_MASK;
	ctrl->ctrl_config &= ~NVME_CC_ENABLE;

	ret = ctrl->ops->reg_write32(ctrl, NVME_REG_CC, ctrl->ctrl_config);
	if (ret)
		return ret;
2437

2438
	if (ctrl->quirks & NVME_QUIRK_DELAY_BEFORE_CHK_RDY)
2439 2440
		msleep(NVME_QUIRK_DELAY_AMOUNT);

2441
	return nvme_wait_ready(ctrl, ctrl->cap, false);
2442
}
2443
EXPORT_SYMBOL_GPL(nvme_disable_ctrl);
2444

2445
int nvme_enable_ctrl(struct nvme_ctrl *ctrl)
2446
{
2447
	unsigned dev_page_min;
2448 2449
	int ret;

2450 2451 2452 2453 2454 2455 2456
	ret = ctrl->ops->reg_read64(ctrl, NVME_REG_CAP, &ctrl->cap);
	if (ret) {
		dev_err(ctrl->device, "Reading CAP failed (%d)\n", ret);
		return ret;
	}
	dev_page_min = NVME_CAP_MPSMIN(ctrl->cap) + 12;

2457
	if (NVME_CTRL_PAGE_SHIFT < dev_page_min) {
2458
		dev_err(ctrl->device,
2459
			"Minimum device page size %u too large for host (%u)\n",
2460
			1 << dev_page_min, 1 << NVME_CTRL_PAGE_SHIFT);
2461 2462 2463
		return -ENODEV;
	}

2464 2465 2466 2467
	if (NVME_CAP_CSS(ctrl->cap) & NVME_CAP_CSS_CSI)
		ctrl->ctrl_config = NVME_CC_CSS_CSI;
	else
		ctrl->ctrl_config = NVME_CC_CSS_NVM;
2468
	ctrl->ctrl_config |= (NVME_CTRL_PAGE_SHIFT - 12) << NVME_CC_MPS_SHIFT;
2469
	ctrl->ctrl_config |= NVME_CC_AMS_RR | NVME_CC_SHN_NONE;
2470 2471 2472 2473 2474 2475
	ctrl->ctrl_config |= NVME_CC_IOSQES | NVME_CC_IOCQES;
	ctrl->ctrl_config |= NVME_CC_ENABLE;

	ret = ctrl->ops->reg_write32(ctrl, NVME_REG_CC, ctrl->ctrl_config);
	if (ret)
		return ret;
2476
	return nvme_wait_ready(ctrl, ctrl->cap, true);
2477
}
2478
EXPORT_SYMBOL_GPL(nvme_enable_ctrl);
2479 2480 2481

int nvme_shutdown_ctrl(struct nvme_ctrl *ctrl)
{
2482
	unsigned long timeout = jiffies + (ctrl->shutdown_timeout * HZ);
2483 2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500
	u32 csts;
	int ret;

	ctrl->ctrl_config &= ~NVME_CC_SHN_MASK;
	ctrl->ctrl_config |= NVME_CC_SHN_NORMAL;

	ret = ctrl->ops->reg_write32(ctrl, NVME_REG_CC, ctrl->ctrl_config);
	if (ret)
		return ret;

	while ((ret = ctrl->ops->reg_read32(ctrl, NVME_REG_CSTS, &csts)) == 0) {
		if ((csts & NVME_CSTS_SHST_MASK) == NVME_CSTS_SHST_CMPLT)
			break;

		msleep(100);
		if (fatal_signal_pending(current))
			return -EINTR;
		if (time_after(jiffies, timeout)) {
2501
			dev_err(ctrl->device,
2502 2503 2504 2505 2506 2507 2508
				"Device shutdown incomplete; abort shutdown\n");
			return -ENODEV;
		}
	}

	return ret;
}
2509
EXPORT_SYMBOL_GPL(nvme_shutdown_ctrl);
2510

2511 2512 2513 2514 2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527
static int nvme_configure_timestamp(struct nvme_ctrl *ctrl)
{
	__le64 ts;
	int ret;

	if (!(ctrl->oncs & NVME_CTRL_ONCS_TIMESTAMP))
		return 0;

	ts = cpu_to_le64(ktime_to_ms(ktime_get_real()));
	ret = nvme_set_features(ctrl, NVME_FEAT_TIMESTAMP, 0, &ts, sizeof(ts),
			NULL);
	if (ret)
		dev_warn_once(ctrl->device,
			"could not set timestamp (%d)\n", ret);
	return ret;
}

2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547
static int nvme_configure_acre(struct nvme_ctrl *ctrl)
{
	struct nvme_feat_host_behavior *host;
	int ret;

	/* Don't bother enabling the feature if retry delay is not reported */
	if (!ctrl->crdt[0])
		return 0;

	host = kzalloc(sizeof(*host), GFP_KERNEL);
	if (!host)
		return 0;

	host->acre = NVME_ENABLE_ACRE;
	ret = nvme_set_features(ctrl, NVME_FEAT_HOST_BEHAVIOR, 0,
				host, sizeof(*host), NULL);
	kfree(host);
	return ret;
}

2548
static int nvme_configure_apst(struct nvme_ctrl *ctrl)
2549 2550 2551 2552 2553 2554 2555 2556
{
	/*
	 * APST (Autonomous Power State Transition) lets us program a
	 * table of power state transitions that the controller will
	 * perform automatically.  We configure it with a simple
	 * heuristic: we are willing to spend at most 2% of the time
	 * transitioning between power states.  Therefore, when running
	 * in any given state, we will enter the next lower-power
A
Andy Lutomirski 已提交
2557
	 * non-operational state after waiting 50 * (enlat + exlat)
2558
	 * microseconds, as long as that state's exit latency is under
2559 2560 2561 2562 2563 2564 2565 2566 2567
	 * the requested maximum latency.
	 *
	 * We will not autonomously enter any non-operational state for
	 * which the total latency exceeds ps_max_latency_us.  Users
	 * can set ps_max_latency_us to zero to turn off APST.
	 */

	unsigned apste;
	struct nvme_feat_auto_pst *table;
2568 2569
	u64 max_lat_us = 0;
	int max_ps = -1;
2570 2571 2572 2573 2574 2575 2576
	int ret;

	/*
	 * If APST isn't supported or if we haven't been initialized yet,
	 * then don't do anything.
	 */
	if (!ctrl->apsta)
2577
		return 0;
2578 2579 2580

	if (ctrl->npss > 31) {
		dev_warn(ctrl->device, "NPSS is invalid; not using APST\n");
2581
		return 0;
2582 2583 2584 2585
	}

	table = kzalloc(sizeof(*table), GFP_KERNEL);
	if (!table)
2586
		return 0;
2587

2588
	if (!ctrl->apst_enabled || ctrl->ps_max_latency_us == 0) {
2589 2590
		/* Turn off APST. */
		apste = 0;
2591
		dev_dbg(ctrl->device, "APST disabled\n");
2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602
	} else {
		__le64 target = cpu_to_le64(0);
		int state;

		/*
		 * Walk through all states from lowest- to highest-power.
		 * According to the spec, lower-numbered states use more
		 * power.  NPSS, despite the name, is the index of the
		 * lowest-power state, not the number of states.
		 */
		for (state = (int)ctrl->npss; state >= 0; state--) {
2603
			u64 total_latency_us, exit_latency_us, transition_ms;
2604 2605 2606 2607

			if (target)
				table->entries[state] = target;

2608 2609 2610 2611 2612 2613 2614 2615
			/*
			 * Don't allow transitions to the deepest state
			 * if it's quirked off.
			 */
			if (state == ctrl->npss &&
			    (ctrl->quirks & NVME_QUIRK_NO_DEEPEST_PS))
				continue;

2616 2617 2618 2619 2620 2621 2622 2623
			/*
			 * Is this state a useful non-operational state for
			 * higher-power states to autonomously transition to?
			 */
			if (!(ctrl->psd[state].flags &
			      NVME_PS_FLAGS_NON_OP_STATE))
				continue;

2624 2625 2626
			exit_latency_us =
				(u64)le32_to_cpu(ctrl->psd[state].exit_lat);
			if (exit_latency_us > ctrl->ps_max_latency_us)
2627 2628
				continue;

2629 2630 2631 2632
			total_latency_us =
				exit_latency_us +
				le32_to_cpu(ctrl->psd[state].entry_lat);

2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643
			/*
			 * This state is good.  Use it as the APST idle
			 * target for higher power states.
			 */
			transition_ms = total_latency_us + 19;
			do_div(transition_ms, 20);
			if (transition_ms > (1 << 24) - 1)
				transition_ms = (1 << 24) - 1;

			target = cpu_to_le64((state << 3) |
					     (transition_ms << 8));
2644 2645 2646 2647 2648 2649

			if (max_ps == -1)
				max_ps = state;

			if (total_latency_us > max_lat_us)
				max_lat_us = total_latency_us;
2650 2651 2652
		}

		apste = 1;
2653 2654 2655 2656 2657 2658 2659

		if (max_ps == -1) {
			dev_dbg(ctrl->device, "APST enabled but no non-operational states are available\n");
		} else {
			dev_dbg(ctrl->device, "APST enabled: max PS = %d, max round-trip latency = %lluus, table = %*phN\n",
				max_ps, max_lat_us, (int)sizeof(*table), table);
		}
2660 2661 2662 2663 2664 2665 2666 2667
	}

	ret = nvme_set_features(ctrl, NVME_FEAT_AUTO_PST, apste,
				table, sizeof(*table), NULL);
	if (ret)
		dev_err(ctrl->device, "failed to set APST feature (%d)\n", ret);

	kfree(table);
2668
	return ret;
2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691
}

static void nvme_set_latency_tolerance(struct device *dev, s32 val)
{
	struct nvme_ctrl *ctrl = dev_get_drvdata(dev);
	u64 latency;

	switch (val) {
	case PM_QOS_LATENCY_TOLERANCE_NO_CONSTRAINT:
	case PM_QOS_LATENCY_ANY:
		latency = U64_MAX;
		break;

	default:
		latency = val;
	}

	if (ctrl->ps_max_latency_us != latency) {
		ctrl->ps_max_latency_us = latency;
		nvme_configure_apst(ctrl);
	}
}

2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704
struct nvme_core_quirk_entry {
	/*
	 * NVMe model and firmware strings are padded with spaces.  For
	 * simplicity, strings in the quirk table are padded with NULLs
	 * instead.
	 */
	u16 vid;
	const char *mn;
	const char *fr;
	unsigned long quirks;
};

static const struct nvme_core_quirk_entry core_quirks[] = {
2705
	{
2706 2707 2708 2709 2710 2711
		/*
		 * This Toshiba device seems to die using any APST states.  See:
		 * https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1678184/comments/11
		 */
		.vid = 0x1179,
		.mn = "THNSF5256GPUK TOSHIBA",
2712
		.quirks = NVME_QUIRK_NO_APST,
2713 2714 2715 2716 2717 2718 2719 2720 2721 2722
	},
	{
		/*
		 * This LiteON CL1-3D*-Q11 firmware version has a race
		 * condition associated with actions related to suspend to idle
		 * LiteON has resolved the problem in future firmware
		 */
		.vid = 0x14a4,
		.fr = "22301111",
		.quirks = NVME_QUIRK_SIMPLE_SUSPEND,
2723
	}
2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754
};

/* match is null-terminated but idstr is space-padded. */
static bool string_matches(const char *idstr, const char *match, size_t len)
{
	size_t matchlen;

	if (!match)
		return true;

	matchlen = strlen(match);
	WARN_ON_ONCE(matchlen > len);

	if (memcmp(idstr, match, matchlen))
		return false;

	for (; matchlen < len; matchlen++)
		if (idstr[matchlen] != ' ')
			return false;

	return true;
}

static bool quirk_matches(const struct nvme_id_ctrl *id,
			  const struct nvme_core_quirk_entry *q)
{
	return q->vid == le16_to_cpu(id->vid) &&
		string_matches(id->mn, q->mn, sizeof(id->mn)) &&
		string_matches(id->fr, q->fr, sizeof(id->fr));
}

C
Christoph Hellwig 已提交
2755 2756
static void nvme_init_subnqn(struct nvme_subsystem *subsys, struct nvme_ctrl *ctrl,
		struct nvme_id_ctrl *id)
2757 2758 2759 2760
{
	size_t nqnlen;
	int off;

2761 2762 2763 2764 2765 2766
	if(!(ctrl->quirks & NVME_QUIRK_IGNORE_DEV_SUBNQN)) {
		nqnlen = strnlen(id->subnqn, NVMF_NQN_SIZE);
		if (nqnlen > 0 && nqnlen < NVMF_NQN_SIZE) {
			strlcpy(subsys->subnqn, id->subnqn, NVMF_NQN_SIZE);
			return;
		}
2767

2768 2769 2770
		if (ctrl->vs >= NVME_VS(1, 2, 1))
			dev_warn(ctrl->device, "missing or invalid SUBNQN field.\n");
	}
2771 2772

	/* Generate a "fake" NQN per Figure 254 in NVMe 1.3 + ECN 001 */
C
Christoph Hellwig 已提交
2773
	off = snprintf(subsys->subnqn, NVMF_NQN_SIZE,
2774
			"nqn.2014.08.org.nvmexpress:%04x%04x",
2775
			le16_to_cpu(id->vid), le16_to_cpu(id->ssvid));
C
Christoph Hellwig 已提交
2776
	memcpy(subsys->subnqn + off, id->sn, sizeof(id->sn));
2777
	off += sizeof(id->sn);
C
Christoph Hellwig 已提交
2778
	memcpy(subsys->subnqn + off, id->mn, sizeof(id->mn));
2779
	off += sizeof(id->mn);
C
Christoph Hellwig 已提交
2780 2781 2782
	memset(subsys->subnqn + off, 0, sizeof(subsys->subnqn) - off);
}

2783
static void nvme_release_subsystem(struct device *dev)
C
Christoph Hellwig 已提交
2784
{
2785 2786 2787
	struct nvme_subsystem *subsys =
		container_of(dev, struct nvme_subsystem, dev);

2788 2789
	if (subsys->instance >= 0)
		ida_simple_remove(&nvme_instance_ida, subsys->instance);
C
Christoph Hellwig 已提交
2790 2791 2792 2793 2794 2795 2796 2797 2798 2799 2800 2801
	kfree(subsys);
}

static void nvme_destroy_subsystem(struct kref *ref)
{
	struct nvme_subsystem *subsys =
			container_of(ref, struct nvme_subsystem, ref);

	mutex_lock(&nvme_subsystems_lock);
	list_del(&subsys->entry);
	mutex_unlock(&nvme_subsystems_lock);

C
Christoph Hellwig 已提交
2802
	ida_destroy(&subsys->ns_ida);
C
Christoph Hellwig 已提交
2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814 2815 2816 2817
	device_del(&subsys->dev);
	put_device(&subsys->dev);
}

static void nvme_put_subsystem(struct nvme_subsystem *subsys)
{
	kref_put(&subsys->ref, nvme_destroy_subsystem);
}

static struct nvme_subsystem *__nvme_find_get_subsystem(const char *subsysnqn)
{
	struct nvme_subsystem *subsys;

	lockdep_assert_held(&nvme_subsystems_lock);

2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828
	/*
	 * Fail matches for discovery subsystems. This results
	 * in each discovery controller bound to a unique subsystem.
	 * This avoids issues with validating controller values
	 * that can only be true when there is a single unique subsystem.
	 * There may be multiple and completely independent entities
	 * that provide discovery controllers.
	 */
	if (!strcmp(subsysnqn, NVME_DISC_SUBSYS_NAME))
		return NULL;

C
Christoph Hellwig 已提交
2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839
	list_for_each_entry(subsys, &nvme_subsystems, entry) {
		if (strcmp(subsys->subnqn, subsysnqn))
			continue;
		if (!kref_get_unless_zero(&subsys->ref))
			continue;
		return subsys;
	}

	return NULL;
}

2840 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850
#define SUBSYS_ATTR_RO(_name, _mode, _show)			\
	struct device_attribute subsys_attr_##_name = \
		__ATTR(_name, _mode, _show, NULL)

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

2851
	return sysfs_emit(buf, "%s\n", subsys->subnqn);
2852 2853 2854 2855 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 2873 2874
}
static SUBSYS_ATTR_RO(subsysnqn, S_IRUGO, nvme_subsys_show_nqn);

#define nvme_subsys_show_str_function(field)				\
static ssize_t subsys_##field##_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",					\
		       (int)sizeof(subsys->field), subsys->field);	\
}									\
static SUBSYS_ATTR_RO(field, S_IRUGO, subsys_##field##_show);

nvme_subsys_show_str_function(model);
nvme_subsys_show_str_function(serial);
nvme_subsys_show_str_function(firmware_rev);

static struct attribute *nvme_subsys_attrs[] = {
	&subsys_attr_model.attr,
	&subsys_attr_serial.attr,
	&subsys_attr_firmware_rev.attr,
	&subsys_attr_subsysnqn.attr,
2875 2876 2877
#ifdef CONFIG_NVME_MULTIPATH
	&subsys_attr_iopolicy.attr,
#endif
2878 2879 2880
	NULL,
};

2881
static const struct attribute_group nvme_subsys_attrs_group = {
2882 2883 2884 2885 2886 2887 2888 2889
	.attrs = nvme_subsys_attrs,
};

static const struct attribute_group *nvme_subsys_attrs_groups[] = {
	&nvme_subsys_attrs_group,
	NULL,
};

2890 2891 2892 2893 2894
static inline bool nvme_discovery_ctrl(struct nvme_ctrl *ctrl)
{
	return ctrl->opts && ctrl->opts->discovery_nqn;
}

2895 2896
static bool nvme_validate_cntlid(struct nvme_subsystem *subsys,
		struct nvme_ctrl *ctrl, struct nvme_id_ctrl *id)
2897
{
2898
	struct nvme_ctrl *tmp;
2899

2900 2901
	lockdep_assert_held(&nvme_subsystems_lock);

2902
	list_for_each_entry(tmp, &subsys->ctrls, subsys_entry) {
2903
		if (nvme_state_terminal(tmp))
2904 2905 2906 2907 2908 2909 2910 2911
			continue;

		if (tmp->cntlid == ctrl->cntlid) {
			dev_err(ctrl->device,
				"Duplicate cntlid %u with %s, rejecting\n",
				ctrl->cntlid, dev_name(tmp->device));
			return false;
		}
2912

2913
		if ((id->cmic & NVME_CTRL_CMIC_MULTI_CTRL) ||
2914
		    nvme_discovery_ctrl(ctrl))
2915 2916 2917 2918 2919
			continue;

		dev_err(ctrl->device,
			"Subsystem does not support multiple controllers\n");
		return false;
2920 2921
	}

2922
	return true;
2923 2924
}

C
Christoph Hellwig 已提交
2925 2926 2927 2928 2929 2930 2931 2932
static int nvme_init_subsystem(struct nvme_ctrl *ctrl, struct nvme_id_ctrl *id)
{
	struct nvme_subsystem *subsys, *found;
	int ret;

	subsys = kzalloc(sizeof(*subsys), GFP_KERNEL);
	if (!subsys)
		return -ENOMEM;
2933 2934

	subsys->instance = -1;
C
Christoph Hellwig 已提交
2935 2936 2937
	mutex_init(&subsys->lock);
	kref_init(&subsys->ref);
	INIT_LIST_HEAD(&subsys->ctrls);
C
Christoph Hellwig 已提交
2938
	INIT_LIST_HEAD(&subsys->nsheads);
C
Christoph Hellwig 已提交
2939 2940 2941 2942 2943 2944
	nvme_init_subnqn(subsys, ctrl, id);
	memcpy(subsys->serial, id->sn, sizeof(subsys->serial));
	memcpy(subsys->model, id->mn, sizeof(subsys->model));
	memcpy(subsys->firmware_rev, id->fr, sizeof(subsys->firmware_rev));
	subsys->vendor_id = le16_to_cpu(id->vid);
	subsys->cmic = id->cmic;
2945
	subsys->awupf = le16_to_cpu(id->awupf);
2946 2947 2948
#ifdef CONFIG_NVME_MULTIPATH
	subsys->iopolicy = NVME_IOPOLICY_NUMA;
#endif
C
Christoph Hellwig 已提交
2949 2950 2951

	subsys->dev.class = nvme_subsys_class;
	subsys->dev.release = nvme_release_subsystem;
2952
	subsys->dev.groups = nvme_subsys_attrs_groups;
2953
	dev_set_name(&subsys->dev, "nvme-subsys%d", ctrl->instance);
C
Christoph Hellwig 已提交
2954 2955 2956 2957 2958
	device_initialize(&subsys->dev);

	mutex_lock(&nvme_subsystems_lock);
	found = __nvme_find_get_subsystem(subsys->subnqn);
	if (found) {
2959
		put_device(&subsys->dev);
C
Christoph Hellwig 已提交
2960
		subsys = found;
2961

2962
		if (!nvme_validate_cntlid(subsys, ctrl, id)) {
C
Christoph Hellwig 已提交
2963
			ret = -EINVAL;
2964
			goto out_put_subsystem;
C
Christoph Hellwig 已提交
2965 2966 2967 2968 2969 2970
		}
	} else {
		ret = device_add(&subsys->dev);
		if (ret) {
			dev_err(ctrl->device,
				"failed to register subsystem device.\n");
2971
			put_device(&subsys->dev);
C
Christoph Hellwig 已提交
2972 2973
			goto out_unlock;
		}
C
Christoph Hellwig 已提交
2974
		ida_init(&subsys->ns_ida);
C
Christoph Hellwig 已提交
2975 2976 2977
		list_add_tail(&subsys->entry, &nvme_subsystems);
	}

2978 2979 2980
	ret = sysfs_create_link(&subsys->dev.kobj, &ctrl->device->kobj,
				dev_name(ctrl->device));
	if (ret) {
C
Christoph Hellwig 已提交
2981 2982
		dev_err(ctrl->device,
			"failed to create sysfs link from subsystem.\n");
2983
		goto out_put_subsystem;
C
Christoph Hellwig 已提交
2984 2985
	}

2986 2987
	if (!found)
		subsys->instance = ctrl->instance;
2988
	ctrl->subsys = subsys;
C
Christoph Hellwig 已提交
2989
	list_add_tail(&ctrl->subsys_entry, &subsys->ctrls);
2990
	mutex_unlock(&nvme_subsystems_lock);
C
Christoph Hellwig 已提交
2991 2992
	return 0;

2993 2994
out_put_subsystem:
	nvme_put_subsystem(subsys);
C
Christoph Hellwig 已提交
2995 2996 2997
out_unlock:
	mutex_unlock(&nvme_subsystems_lock);
	return ret;
2998 2999
}

3000
int nvme_get_log(struct nvme_ctrl *ctrl, u32 nsid, u8 log_page, u8 lsp, u8 csi,
3001
		void *log, size_t size, u64 offset)
K
Keith Busch 已提交
3002 3003
{
	struct nvme_command c = { };
K
Keith Busch 已提交
3004
	u32 dwlen = nvme_bytes_to_numd(size);
3005 3006

	c.get_log_page.opcode = nvme_admin_get_log_page;
3007
	c.get_log_page.nsid = cpu_to_le32(nsid);
3008
	c.get_log_page.lid = log_page;
3009
	c.get_log_page.lsp = lsp;
3010 3011
	c.get_log_page.numdl = cpu_to_le16(dwlen & ((1 << 16) - 1));
	c.get_log_page.numdu = cpu_to_le16(dwlen >> 16);
3012 3013
	c.get_log_page.lpol = cpu_to_le32(lower_32_bits(offset));
	c.get_log_page.lpou = cpu_to_le32(upper_32_bits(offset));
3014
	c.get_log_page.csi = csi;
K
Keith Busch 已提交
3015 3016 3017 3018

	return nvme_submit_sync_cmd(ctrl->admin_q, &c, log, size);
}

3019 3020
static int nvme_get_effects_log(struct nvme_ctrl *ctrl, u8 csi,
				struct nvme_effects_log **log)
3021
{
3022
	struct nvme_effects_log	*cel = xa_load(&ctrl->cels, csi);
3023 3024
	int ret;

3025 3026
	if (cel)
		goto out;
3027

3028 3029 3030
	cel = kzalloc(sizeof(*cel), GFP_KERNEL);
	if (!cel)
		return -ENOMEM;
3031

3032
	ret = nvme_get_log(ctrl, 0x00, NVME_LOG_CMD_EFFECTS, 0, csi,
3033
			cel, sizeof(*cel), 0);
3034
	if (ret) {
3035 3036
		kfree(cel);
		return ret;
3037
	}
3038

3039
	xa_store(&ctrl->cels, csi, cel, GFP_KERNEL);
3040
out:
3041
	*log = cel;
3042
	return 0;
3043 3044
}

3045 3046 3047 3048 3049 3050 3051 3052 3053
/*
 * Initialize the cached copies of the Identify data and various controller
 * register in our nvme_ctrl structure.  This should be called as soon as
 * the admin queue is fully up and running.
 */
int nvme_init_identify(struct nvme_ctrl *ctrl)
{
	struct nvme_id_ctrl *id;
	int ret, page_shift;
3054
	u32 max_hw_sectors;
3055
	bool prev_apst_enabled;
3056

3057 3058
	ret = ctrl->ops->reg_read32(ctrl, NVME_REG_VS, &ctrl->vs);
	if (ret) {
3059
		dev_err(ctrl->device, "Reading VS failed (%d)\n", ret);
3060 3061
		return ret;
	}
3062
	page_shift = NVME_CAP_MPSMIN(ctrl->cap) + 12;
3063
	ctrl->sqsize = min_t(u16, NVME_CAP_MQES(ctrl->cap), ctrl->sqsize);
3064

3065
	if (ctrl->vs >= NVME_VS(1, 1, 0))
3066
		ctrl->subsystem = NVME_CAP_NSSRC(ctrl->cap);
3067

3068 3069
	ret = nvme_identify_ctrl(ctrl, &id);
	if (ret) {
3070
		dev_err(ctrl->device, "Identify Controller failed (%d)\n", ret);
3071 3072 3073
		return -EIO;
	}

3074
	if (id->lpa & NVME_CTRL_LPA_CMD_EFFECTS_LOG) {
3075
		ret = nvme_get_effects_log(ctrl, NVME_CSI_NVM, &ctrl->effects);
3076
		if (ret < 0)
3077
			goto out_free;
3078
	}
3079

3080 3081 3082
	if (!(ctrl->ops->flags & NVME_F_FABRICS))
		ctrl->cntlid = le16_to_cpu(id->cntlid);

3083
	if (!ctrl->identified) {
C
Christoph Hellwig 已提交
3084 3085 3086 3087 3088 3089
		int i;

		ret = nvme_init_subsystem(ctrl, id);
		if (ret)
			goto out_free;

3090 3091 3092 3093 3094 3095 3096 3097 3098 3099 3100 3101 3102 3103
		/*
		 * Check for quirks.  Quirk can depend on firmware version,
		 * so, in principle, the set of quirks present can change
		 * across a reset.  As a possible future enhancement, we
		 * could re-scan for quirks every time we reinitialize
		 * the device, but we'd have to make sure that the driver
		 * behaves intelligently if the quirks change.
		 */
		for (i = 0; i < ARRAY_SIZE(core_quirks); i++) {
			if (quirk_matches(id, &core_quirks[i]))
				ctrl->quirks |= core_quirks[i].quirks;
		}
	}

3104
	if (force_apst && (ctrl->quirks & NVME_QUIRK_NO_DEEPEST_PS)) {
3105
		dev_warn(ctrl->device, "forcibly allowing all power states due to nvme_core.force_apst -- use at your own risk\n");
3106 3107 3108
		ctrl->quirks &= ~NVME_QUIRK_NO_DEEPEST_PS;
	}

3109 3110 3111 3112
	ctrl->crdt[0] = le16_to_cpu(id->crdt1);
	ctrl->crdt[1] = le16_to_cpu(id->crdt2);
	ctrl->crdt[2] = le16_to_cpu(id->crdt3);

3113
	ctrl->oacs = le16_to_cpu(id->oacs);
3114
	ctrl->oncs = le16_to_cpu(id->oncs);
3115
	ctrl->mtfa = le16_to_cpu(id->mtfa);
3116
	ctrl->oaes = le32_to_cpu(id->oaes);
3117 3118 3119
	ctrl->wctemp = le16_to_cpu(id->wctemp);
	ctrl->cctemp = le16_to_cpu(id->cctemp);

3120
	atomic_set(&ctrl->abort_limit, id->acl + 1);
3121 3122
	ctrl->vwc = id->vwc;
	if (id->mdts)
3123
		max_hw_sectors = 1 << (id->mdts + page_shift - 9);
3124
	else
3125 3126 3127
		max_hw_sectors = UINT_MAX;
	ctrl->max_hw_sectors =
		min_not_zero(ctrl->max_hw_sectors, max_hw_sectors);
3128

3129
	nvme_set_queue_limits(ctrl, ctrl->admin_q);
3130
	ctrl->sgls = le32_to_cpu(id->sgls);
S
Sagi Grimberg 已提交
3131
	ctrl->kas = le16_to_cpu(id->kas);
C
Christoph Hellwig 已提交
3132
	ctrl->max_namespaces = le32_to_cpu(id->mnan);
S
Sagi Grimberg 已提交
3133
	ctrl->ctratt = le32_to_cpu(id->ctratt);
3134

3135 3136
	if (id->rtd3e) {
		/* us -> s */
3137
		u32 transition_time = le32_to_cpu(id->rtd3e) / USEC_PER_SEC;
3138 3139 3140 3141 3142

		ctrl->shutdown_timeout = clamp_t(unsigned int, transition_time,
						 shutdown_timeout, 60);

		if (ctrl->shutdown_timeout != shutdown_timeout)
3143
			dev_info(ctrl->device,
3144 3145 3146 3147 3148
				 "Shutdown timeout set to %u seconds\n",
				 ctrl->shutdown_timeout);
	} else
		ctrl->shutdown_timeout = shutdown_timeout;

3149
	ctrl->npss = id->npss;
3150 3151
	ctrl->apsta = id->apsta;
	prev_apst_enabled = ctrl->apst_enabled;
3152 3153
	if (ctrl->quirks & NVME_QUIRK_NO_APST) {
		if (force_apst && id->apsta) {
3154
			dev_warn(ctrl->device, "forcibly allowing APST due to nvme_core.force_apst -- use at your own risk\n");
3155
			ctrl->apst_enabled = true;
3156
		} else {
3157
			ctrl->apst_enabled = false;
3158 3159
		}
	} else {
3160
		ctrl->apst_enabled = id->apsta;
3161
	}
3162 3163
	memcpy(ctrl->psd, id->psd, sizeof(ctrl->psd));

3164
	if (ctrl->ops->flags & NVME_F_FABRICS) {
3165 3166 3167 3168 3169 3170 3171 3172 3173
		ctrl->icdoff = le16_to_cpu(id->icdoff);
		ctrl->ioccsz = le32_to_cpu(id->ioccsz);
		ctrl->iorcsz = le32_to_cpu(id->iorcsz);
		ctrl->maxcmd = le16_to_cpu(id->maxcmd);

		/*
		 * In fabrics we need to verify the cntlid matches the
		 * admin connect
		 */
3174
		if (ctrl->cntlid != le16_to_cpu(id->cntlid)) {
3175 3176 3177 3178
			dev_err(ctrl->device,
				"Mismatching cntlid: Connect %u vs Identify "
				"%u, rejecting\n",
				ctrl->cntlid, le16_to_cpu(id->cntlid));
3179
			ret = -EINVAL;
3180 3181
			goto out_free;
		}
S
Sagi Grimberg 已提交
3182

3183
		if (!nvme_discovery_ctrl(ctrl) && !ctrl->kas) {
3184
			dev_err(ctrl->device,
S
Sagi Grimberg 已提交
3185 3186
				"keep-alive support is mandatory for fabrics\n");
			ret = -EINVAL;
3187
			goto out_free;
S
Sagi Grimberg 已提交
3188
		}
3189
	} else {
3190 3191
		ctrl->hmpre = le32_to_cpu(id->hmpre);
		ctrl->hmmin = le32_to_cpu(id->hmmin);
3192 3193
		ctrl->hmminds = le32_to_cpu(id->hmminds);
		ctrl->hmmaxd = le16_to_cpu(id->hmmaxd);
3194
	}
3195

C
Christoph Hellwig 已提交
3196
	ret = nvme_mpath_init(ctrl, id);
3197
	kfree(id);
3198

C
Christoph Hellwig 已提交
3199 3200 3201
	if (ret < 0)
		return ret;

3202
	if (ctrl->apst_enabled && !prev_apst_enabled)
3203
		dev_pm_qos_expose_latency_tolerance(ctrl->device);
3204
	else if (!ctrl->apst_enabled && prev_apst_enabled)
3205 3206
		dev_pm_qos_hide_latency_tolerance(ctrl->device);

3207 3208 3209
	ret = nvme_configure_apst(ctrl);
	if (ret < 0)
		return ret;
3210 3211 3212 3213
	
	ret = nvme_configure_timestamp(ctrl);
	if (ret < 0)
		return ret;
3214 3215 3216 3217

	ret = nvme_configure_directives(ctrl);
	if (ret < 0)
		return ret;
3218

3219 3220 3221 3222
	ret = nvme_configure_acre(ctrl);
	if (ret < 0)
		return ret;

3223
	if (!ctrl->identified && !nvme_discovery_ctrl(ctrl)) {
K
Keith Busch 已提交
3224 3225 3226 3227
		ret = nvme_hwmon_init(ctrl);
		if (ret < 0)
			return ret;
	}
3228

3229
	ctrl->identified = true;
3230

3231 3232 3233 3234
	return 0;

out_free:
	kfree(id);
3235
	return ret;
3236
}
3237
EXPORT_SYMBOL_GPL(nvme_init_identify);
3238

3239
static int nvme_dev_open(struct inode *inode, struct file *file)
3240
{
3241 3242
	struct nvme_ctrl *ctrl =
		container_of(inode->i_cdev, struct nvme_ctrl, cdev);
3243

3244 3245 3246 3247
	switch (ctrl->state) {
	case NVME_CTRL_LIVE:
		break;
	default:
3248
		return -EWOULDBLOCK;
3249 3250
	}

3251
	nvme_get_ctrl(ctrl);
3252 3253
	if (!try_module_get(ctrl->ops->module)) {
		nvme_put_ctrl(ctrl);
3254
		return -EINVAL;
3255
	}
3256

3257
	file->private_data = ctrl;
3258 3259 3260
	return 0;
}

3261 3262 3263 3264 3265 3266 3267 3268 3269 3270
static int nvme_dev_release(struct inode *inode, struct file *file)
{
	struct nvme_ctrl *ctrl =
		container_of(inode->i_cdev, struct nvme_ctrl, cdev);

	module_put(ctrl->ops->module);
	nvme_put_ctrl(ctrl);
	return 0;
}

3271 3272 3273 3274 3275
static int nvme_dev_user_cmd(struct nvme_ctrl *ctrl, void __user *argp)
{
	struct nvme_ns *ns;
	int ret;

3276
	down_read(&ctrl->namespaces_rwsem);
3277 3278 3279 3280 3281 3282 3283
	if (list_empty(&ctrl->namespaces)) {
		ret = -ENOTTY;
		goto out_unlock;
	}

	ns = list_first_entry(&ctrl->namespaces, struct nvme_ns, list);
	if (ns != list_last_entry(&ctrl->namespaces, struct nvme_ns, list)) {
3284
		dev_warn(ctrl->device,
3285 3286 3287 3288 3289
			"NVME_IOCTL_IO_CMD not supported when multiple namespaces present!\n");
		ret = -EINVAL;
		goto out_unlock;
	}

3290
	dev_warn(ctrl->device,
3291 3292
		"using deprecated NVME_IOCTL_IO_CMD ioctl on the char device!\n");
	kref_get(&ns->kref);
3293
	up_read(&ctrl->namespaces_rwsem);
3294 3295 3296 3297 3298 3299

	ret = nvme_user_cmd(ctrl, ns, argp);
	nvme_put_ns(ns);
	return ret;

out_unlock:
3300
	up_read(&ctrl->namespaces_rwsem);
3301 3302 3303
	return ret;
}

3304 3305 3306 3307 3308 3309 3310 3311 3312
static long nvme_dev_ioctl(struct file *file, unsigned int cmd,
		unsigned long arg)
{
	struct nvme_ctrl *ctrl = file->private_data;
	void __user *argp = (void __user *)arg;

	switch (cmd) {
	case NVME_IOCTL_ADMIN_CMD:
		return nvme_user_cmd(ctrl, NULL, argp);
3313 3314
	case NVME_IOCTL_ADMIN64_CMD:
		return nvme_user_cmd64(ctrl, NULL, argp);
3315
	case NVME_IOCTL_IO_CMD:
3316
		return nvme_dev_user_cmd(ctrl, argp);
3317
	case NVME_IOCTL_RESET:
3318
		dev_warn(ctrl->device, "resetting controller\n");
3319
		return nvme_reset_ctrl_sync(ctrl);
3320 3321
	case NVME_IOCTL_SUBSYS_RESET:
		return nvme_reset_subsystem(ctrl);
K
Keith Busch 已提交
3322 3323 3324
	case NVME_IOCTL_RESCAN:
		nvme_queue_scan(ctrl);
		return 0;
3325 3326 3327 3328 3329 3330 3331 3332
	default:
		return -ENOTTY;
	}
}

static const struct file_operations nvme_dev_fops = {
	.owner		= THIS_MODULE,
	.open		= nvme_dev_open,
3333
	.release	= nvme_dev_release,
3334
	.unlocked_ioctl	= nvme_dev_ioctl,
3335
	.compat_ioctl	= compat_ptr_ioctl,
3336 3337 3338 3339 3340 3341 3342 3343 3344
};

static ssize_t nvme_sysfs_reset(struct device *dev,
				struct device_attribute *attr, const char *buf,
				size_t count)
{
	struct nvme_ctrl *ctrl = dev_get_drvdata(dev);
	int ret;

3345
	ret = nvme_reset_ctrl_sync(ctrl);
3346 3347 3348
	if (ret < 0)
		return ret;
	return count;
3349
}
3350
static DEVICE_ATTR(reset_controller, S_IWUSR, NULL, nvme_sysfs_reset);
3351

K
Keith Busch 已提交
3352 3353 3354 3355 3356 3357 3358 3359 3360 3361 3362
static ssize_t nvme_sysfs_rescan(struct device *dev,
				struct device_attribute *attr, const char *buf,
				size_t count)
{
	struct nvme_ctrl *ctrl = dev_get_drvdata(dev);

	nvme_queue_scan(ctrl);
	return count;
}
static DEVICE_ATTR(rescan_controller, S_IWUSR, NULL, nvme_sysfs_rescan);

3363 3364 3365 3366
static inline struct nvme_ns_head *dev_to_ns_head(struct device *dev)
{
	struct gendisk *disk = dev_to_disk(dev);

J
Javier González 已提交
3367
	if (disk->fops == &nvme_bdev_ops)
3368 3369 3370 3371 3372
		return nvme_get_ns_from_dev(dev)->head;
	else
		return disk->private_data;
}

3373
static ssize_t wwid_show(struct device *dev, struct device_attribute *attr,
3374
		char *buf)
3375
{
3376 3377 3378
	struct nvme_ns_head *head = dev_to_ns_head(dev);
	struct nvme_ns_ids *ids = &head->ids;
	struct nvme_subsystem *subsys = head->subsys;
C
Christoph Hellwig 已提交
3379 3380
	int serial_len = sizeof(subsys->serial);
	int model_len = sizeof(subsys->model);
3381

3382 3383
	if (!uuid_is_null(&ids->uuid))
		return sprintf(buf, "uuid.%pU\n", &ids->uuid);
3384

3385 3386
	if (memchr_inv(ids->nguid, 0, sizeof(ids->nguid)))
		return sprintf(buf, "eui.%16phN\n", ids->nguid);
3387

3388 3389
	if (memchr_inv(ids->eui64, 0, sizeof(ids->eui64)))
		return sprintf(buf, "eui.%8phN\n", ids->eui64);
3390

C
Christoph Hellwig 已提交
3391 3392
	while (serial_len > 0 && (subsys->serial[serial_len - 1] == ' ' ||
				  subsys->serial[serial_len - 1] == '\0'))
3393
		serial_len--;
C
Christoph Hellwig 已提交
3394 3395
	while (model_len > 0 && (subsys->model[model_len - 1] == ' ' ||
				 subsys->model[model_len - 1] == '\0'))
3396 3397
		model_len--;

C
Christoph Hellwig 已提交
3398 3399
	return sprintf(buf, "nvme.%04x-%*phN-%*phN-%08x\n", subsys->vendor_id,
		serial_len, subsys->serial, model_len, subsys->model,
3400
		head->ns_id);
3401
}
J
Joe Perches 已提交
3402
static DEVICE_ATTR_RO(wwid);
3403

3404
static ssize_t nguid_show(struct device *dev, struct device_attribute *attr,
3405
		char *buf)
3406
{
3407
	return sprintf(buf, "%pU\n", dev_to_ns_head(dev)->ids.nguid);
3408
}
J
Joe Perches 已提交
3409
static DEVICE_ATTR_RO(nguid);
3410

3411
static ssize_t uuid_show(struct device *dev, struct device_attribute *attr,
3412
		char *buf)
3413
{
3414
	struct nvme_ns_ids *ids = &dev_to_ns_head(dev)->ids;
3415 3416 3417 3418

	/* For backward compatibility expose the NGUID to userspace if
	 * we have no UUID set
	 */
3419
	if (uuid_is_null(&ids->uuid)) {
3420 3421
		printk_ratelimited(KERN_WARNING
				   "No UUID available providing old NGUID\n");
3422
		return sprintf(buf, "%pU\n", ids->nguid);
3423
	}
3424
	return sprintf(buf, "%pU\n", &ids->uuid);
3425
}
J
Joe Perches 已提交
3426
static DEVICE_ATTR_RO(uuid);
3427 3428

static ssize_t eui_show(struct device *dev, struct device_attribute *attr,
3429
		char *buf)
3430
{
3431
	return sprintf(buf, "%8ph\n", dev_to_ns_head(dev)->ids.eui64);
3432
}
J
Joe Perches 已提交
3433
static DEVICE_ATTR_RO(eui);
3434 3435

static ssize_t nsid_show(struct device *dev, struct device_attribute *attr,
3436
		char *buf)
3437
{
3438
	return sprintf(buf, "%d\n", dev_to_ns_head(dev)->ns_id);
3439
}
J
Joe Perches 已提交
3440
static DEVICE_ATTR_RO(nsid);
3441

3442
static struct attribute *nvme_ns_id_attrs[] = {
3443
	&dev_attr_wwid.attr,
3444
	&dev_attr_uuid.attr,
3445
	&dev_attr_nguid.attr,
3446 3447
	&dev_attr_eui.attr,
	&dev_attr_nsid.attr,
C
Christoph Hellwig 已提交
3448 3449 3450 3451
#ifdef CONFIG_NVME_MULTIPATH
	&dev_attr_ana_grpid.attr,
	&dev_attr_ana_state.attr,
#endif
3452 3453 3454
	NULL,
};

3455
static umode_t nvme_ns_id_attrs_are_visible(struct kobject *kobj,
3456 3457 3458
		struct attribute *a, int n)
{
	struct device *dev = container_of(kobj, struct device, kobj);
3459
	struct nvme_ns_ids *ids = &dev_to_ns_head(dev)->ids;
3460 3461

	if (a == &dev_attr_uuid.attr) {
3462
		if (uuid_is_null(&ids->uuid) &&
3463
		    !memchr_inv(ids->nguid, 0, sizeof(ids->nguid)))
3464 3465 3466
			return 0;
	}
	if (a == &dev_attr_nguid.attr) {
3467
		if (!memchr_inv(ids->nguid, 0, sizeof(ids->nguid)))
3468 3469 3470
			return 0;
	}
	if (a == &dev_attr_eui.attr) {
3471
		if (!memchr_inv(ids->eui64, 0, sizeof(ids->eui64)))
3472 3473
			return 0;
	}
C
Christoph Hellwig 已提交
3474 3475
#ifdef CONFIG_NVME_MULTIPATH
	if (a == &dev_attr_ana_grpid.attr || a == &dev_attr_ana_state.attr) {
J
Javier González 已提交
3476
		if (dev_to_disk(dev)->fops != &nvme_bdev_ops) /* per-path attr */
C
Christoph Hellwig 已提交
3477 3478 3479 3480 3481
			return 0;
		if (!nvme_ctrl_use_ana(nvme_get_ns_from_dev(dev)->ctrl))
			return 0;
	}
#endif
3482 3483 3484
	return a->mode;
}

3485
static const struct attribute_group nvme_ns_id_attr_group = {
3486 3487
	.attrs		= nvme_ns_id_attrs,
	.is_visible	= nvme_ns_id_attrs_are_visible,
3488 3489
};

3490 3491 3492 3493 3494 3495 3496 3497
const struct attribute_group *nvme_ns_id_attr_groups[] = {
	&nvme_ns_id_attr_group,
#ifdef CONFIG_NVM
	&nvme_nvm_attr_group,
#endif
	NULL,
};

M
Ming Lin 已提交
3498
#define nvme_show_str_function(field)						\
3499 3500 3501 3502
static ssize_t  field##_show(struct device *dev,				\
			    struct device_attribute *attr, char *buf)		\
{										\
        struct nvme_ctrl *ctrl = dev_get_drvdata(dev);				\
C
Christoph Hellwig 已提交
3503 3504
        return sprintf(buf, "%.*s\n",						\
		(int)sizeof(ctrl->subsys->field), ctrl->subsys->field);		\
3505 3506 3507
}										\
static DEVICE_ATTR(field, S_IRUGO, field##_show, NULL);

C
Christoph Hellwig 已提交
3508 3509 3510 3511
nvme_show_str_function(model);
nvme_show_str_function(serial);
nvme_show_str_function(firmware_rev);

M
Ming Lin 已提交
3512 3513 3514 3515 3516 3517 3518 3519 3520 3521
#define nvme_show_int_function(field)						\
static ssize_t  field##_show(struct device *dev,				\
			    struct device_attribute *attr, char *buf)		\
{										\
        struct nvme_ctrl *ctrl = dev_get_drvdata(dev);				\
        return sprintf(buf, "%d\n", ctrl->field);	\
}										\
static DEVICE_ATTR(field, S_IRUGO, field##_show, NULL);

nvme_show_int_function(cntlid);
3522
nvme_show_int_function(numa_node);
3523 3524
nvme_show_int_function(queue_count);
nvme_show_int_function(sqsize);
3525

M
Ming Lin 已提交
3526 3527 3528 3529 3530 3531 3532
static ssize_t nvme_sysfs_delete(struct device *dev,
				struct device_attribute *attr, const char *buf,
				size_t count)
{
	struct nvme_ctrl *ctrl = dev_get_drvdata(dev);

	if (device_remove_file_self(dev, attr))
3533
		nvme_delete_ctrl_sync(ctrl);
M
Ming Lin 已提交
3534 3535 3536 3537 3538 3539 3540 3541 3542 3543
	return count;
}
static DEVICE_ATTR(delete_controller, S_IWUSR, NULL, nvme_sysfs_delete);

static ssize_t nvme_sysfs_show_transport(struct device *dev,
					 struct device_attribute *attr,
					 char *buf)
{
	struct nvme_ctrl *ctrl = dev_get_drvdata(dev);

3544
	return sysfs_emit(buf, "%s\n", ctrl->ops->name);
M
Ming Lin 已提交
3545 3546 3547
}
static DEVICE_ATTR(transport, S_IRUGO, nvme_sysfs_show_transport, NULL);

3548 3549 3550 3551 3552 3553 3554 3555 3556
static ssize_t nvme_sysfs_show_state(struct device *dev,
				     struct device_attribute *attr,
				     char *buf)
{
	struct nvme_ctrl *ctrl = dev_get_drvdata(dev);
	static const char *const state_name[] = {
		[NVME_CTRL_NEW]		= "new",
		[NVME_CTRL_LIVE]	= "live",
		[NVME_CTRL_RESETTING]	= "resetting",
3557
		[NVME_CTRL_CONNECTING]	= "connecting",
3558
		[NVME_CTRL_DELETING]	= "deleting",
3559
		[NVME_CTRL_DELETING_NOIO]= "deleting (no IO)",
3560 3561 3562 3563 3564 3565 3566 3567 3568 3569 3570 3571
		[NVME_CTRL_DEAD]	= "dead",
	};

	if ((unsigned)ctrl->state < ARRAY_SIZE(state_name) &&
	    state_name[ctrl->state])
		return sprintf(buf, "%s\n", state_name[ctrl->state]);

	return sprintf(buf, "unknown state\n");
}

static DEVICE_ATTR(state, S_IRUGO, nvme_sysfs_show_state, NULL);

M
Ming Lin 已提交
3572 3573 3574 3575 3576 3577
static ssize_t nvme_sysfs_show_subsysnqn(struct device *dev,
					 struct device_attribute *attr,
					 char *buf)
{
	struct nvme_ctrl *ctrl = dev_get_drvdata(dev);

3578
	return sysfs_emit(buf, "%s\n", ctrl->subsys->subnqn);
M
Ming Lin 已提交
3579 3580 3581
}
static DEVICE_ATTR(subsysnqn, S_IRUGO, nvme_sysfs_show_subsysnqn, NULL);

3582 3583 3584 3585 3586 3587
static ssize_t nvme_sysfs_show_hostnqn(struct device *dev,
					struct device_attribute *attr,
					char *buf)
{
	struct nvme_ctrl *ctrl = dev_get_drvdata(dev);

3588
	return sysfs_emit(buf, "%s\n", ctrl->opts->host->nqn);
3589 3590 3591
}
static DEVICE_ATTR(hostnqn, S_IRUGO, nvme_sysfs_show_hostnqn, NULL);

3592 3593 3594 3595 3596 3597
static ssize_t nvme_sysfs_show_hostid(struct device *dev,
					struct device_attribute *attr,
					char *buf)
{
	struct nvme_ctrl *ctrl = dev_get_drvdata(dev);

3598
	return sysfs_emit(buf, "%pU\n", &ctrl->opts->host->id);
3599 3600 3601
}
static DEVICE_ATTR(hostid, S_IRUGO, nvme_sysfs_show_hostid, NULL);

M
Ming Lin 已提交
3602 3603 3604 3605 3606 3607 3608 3609 3610 3611
static ssize_t nvme_sysfs_show_address(struct device *dev,
					 struct device_attribute *attr,
					 char *buf)
{
	struct nvme_ctrl *ctrl = dev_get_drvdata(dev);

	return ctrl->ops->get_address(ctrl, buf, PAGE_SIZE);
}
static DEVICE_ATTR(address, S_IRUGO, nvme_sysfs_show_address, NULL);

3612 3613 3614 3615 3616 3617 3618 3619 3620 3621 3622 3623 3624 3625 3626 3627 3628 3629 3630 3631 3632 3633 3634 3635 3636 3637 3638 3639 3640 3641 3642 3643 3644 3645 3646 3647 3648 3649 3650 3651 3652 3653 3654 3655 3656 3657 3658 3659 3660 3661 3662
static ssize_t nvme_ctrl_loss_tmo_show(struct device *dev,
		struct device_attribute *attr, char *buf)
{
	struct nvme_ctrl *ctrl = dev_get_drvdata(dev);
	struct nvmf_ctrl_options *opts = ctrl->opts;

	if (ctrl->opts->max_reconnects == -1)
		return sprintf(buf, "off\n");
	return sprintf(buf, "%d\n",
			opts->max_reconnects * opts->reconnect_delay);
}

static ssize_t nvme_ctrl_loss_tmo_store(struct device *dev,
		struct device_attribute *attr, const char *buf, size_t count)
{
	struct nvme_ctrl *ctrl = dev_get_drvdata(dev);
	struct nvmf_ctrl_options *opts = ctrl->opts;
	int ctrl_loss_tmo, err;

	err = kstrtoint(buf, 10, &ctrl_loss_tmo);
	if (err)
		return -EINVAL;

	else if (ctrl_loss_tmo < 0)
		opts->max_reconnects = -1;
	else
		opts->max_reconnects = DIV_ROUND_UP(ctrl_loss_tmo,
						opts->reconnect_delay);
	return count;
}
static DEVICE_ATTR(ctrl_loss_tmo, S_IRUGO | S_IWUSR,
	nvme_ctrl_loss_tmo_show, nvme_ctrl_loss_tmo_store);

static ssize_t nvme_ctrl_reconnect_delay_show(struct device *dev,
		struct device_attribute *attr, char *buf)
{
	struct nvme_ctrl *ctrl = dev_get_drvdata(dev);

	if (ctrl->opts->reconnect_delay == -1)
		return sprintf(buf, "off\n");
	return sprintf(buf, "%d\n", ctrl->opts->reconnect_delay);
}

static ssize_t nvme_ctrl_reconnect_delay_store(struct device *dev,
		struct device_attribute *attr, const char *buf, size_t count)
{
	struct nvme_ctrl *ctrl = dev_get_drvdata(dev);
	unsigned int v;
	int err;

	err = kstrtou32(buf, 10, &v);
3663 3664
	if (err)
		return err;
3665 3666 3667 3668 3669 3670 3671

	ctrl->opts->reconnect_delay = v;
	return count;
}
static DEVICE_ATTR(reconnect_delay, S_IRUGO | S_IWUSR,
	nvme_ctrl_reconnect_delay_show, nvme_ctrl_reconnect_delay_store);

3672 3673
static struct attribute *nvme_dev_attrs[] = {
	&dev_attr_reset_controller.attr,
K
Keith Busch 已提交
3674
	&dev_attr_rescan_controller.attr,
3675 3676 3677
	&dev_attr_model.attr,
	&dev_attr_serial.attr,
	&dev_attr_firmware_rev.attr,
M
Ming Lin 已提交
3678
	&dev_attr_cntlid.attr,
M
Ming Lin 已提交
3679 3680 3681 3682
	&dev_attr_delete_controller.attr,
	&dev_attr_transport.attr,
	&dev_attr_subsysnqn.attr,
	&dev_attr_address.attr,
3683
	&dev_attr_state.attr,
3684
	&dev_attr_numa_node.attr,
3685 3686
	&dev_attr_queue_count.attr,
	&dev_attr_sqsize.attr,
3687
	&dev_attr_hostnqn.attr,
3688
	&dev_attr_hostid.attr,
3689 3690
	&dev_attr_ctrl_loss_tmo.attr,
	&dev_attr_reconnect_delay.attr,
3691 3692 3693
	NULL
};

M
Ming Lin 已提交
3694 3695 3696 3697 3698 3699
static umode_t nvme_dev_attrs_are_visible(struct kobject *kobj,
		struct attribute *a, int n)
{
	struct device *dev = container_of(kobj, struct device, kobj);
	struct nvme_ctrl *ctrl = dev_get_drvdata(dev);

3700 3701 3702 3703
	if (a == &dev_attr_delete_controller.attr && !ctrl->ops->delete_ctrl)
		return 0;
	if (a == &dev_attr_address.attr && !ctrl->ops->get_address)
		return 0;
3704 3705
	if (a == &dev_attr_hostnqn.attr && !ctrl->opts)
		return 0;
3706 3707
	if (a == &dev_attr_hostid.attr && !ctrl->opts)
		return 0;
3708 3709 3710 3711
	if (a == &dev_attr_ctrl_loss_tmo.attr && !ctrl->opts)
		return 0;
	if (a == &dev_attr_reconnect_delay.attr && !ctrl->opts)
		return 0;
M
Ming Lin 已提交
3712 3713 3714 3715

	return a->mode;
}

3716
static const struct attribute_group nvme_dev_attrs_group = {
M
Ming Lin 已提交
3717 3718
	.attrs		= nvme_dev_attrs,
	.is_visible	= nvme_dev_attrs_are_visible,
3719 3720 3721 3722 3723 3724 3725
};

static const struct attribute_group *nvme_dev_attr_groups[] = {
	&nvme_dev_attrs_group,
	NULL,
};

3726
static struct nvme_ns_head *nvme_find_ns_head(struct nvme_subsystem *subsys,
C
Christoph Hellwig 已提交
3727 3728 3729 3730 3731 3732 3733 3734 3735 3736 3737 3738 3739 3740 3741 3742 3743 3744 3745 3746 3747 3748 3749 3750 3751 3752 3753 3754 3755 3756 3757
		unsigned nsid)
{
	struct nvme_ns_head *h;

	lockdep_assert_held(&subsys->lock);

	list_for_each_entry(h, &subsys->nsheads, entry) {
		if (h->ns_id == nsid && kref_get_unless_zero(&h->ref))
			return h;
	}

	return NULL;
}

static int __nvme_check_ids(struct nvme_subsystem *subsys,
		struct nvme_ns_head *new)
{
	struct nvme_ns_head *h;

	lockdep_assert_held(&subsys->lock);

	list_for_each_entry(h, &subsys->nsheads, entry) {
		if (nvme_ns_ids_valid(&new->ids) &&
		    nvme_ns_ids_equal(&new->ids, &h->ids))
			return -EINVAL;
	}

	return 0;
}

static struct nvme_ns_head *nvme_alloc_ns_head(struct nvme_ctrl *ctrl,
K
Keith Busch 已提交
3758
		unsigned nsid, struct nvme_ns_ids *ids)
C
Christoph Hellwig 已提交
3759 3760
{
	struct nvme_ns_head *head;
3761
	size_t size = sizeof(*head);
C
Christoph Hellwig 已提交
3762 3763
	int ret = -ENOMEM;

3764 3765 3766 3767 3768
#ifdef CONFIG_NVME_MULTIPATH
	size += num_possible_nodes() * sizeof(struct nvme_ns *);
#endif

	head = kzalloc(size, GFP_KERNEL);
C
Christoph Hellwig 已提交
3769 3770 3771 3772 3773 3774 3775
	if (!head)
		goto out;
	ret = ida_simple_get(&ctrl->subsys->ns_ida, 1, 0, GFP_KERNEL);
	if (ret < 0)
		goto out_free_head;
	head->instance = ret;
	INIT_LIST_HEAD(&head->list);
3776 3777 3778
	ret = init_srcu_struct(&head->srcu);
	if (ret)
		goto out_ida_remove;
C
Christoph Hellwig 已提交
3779 3780
	head->subsys = ctrl->subsys;
	head->ns_id = nsid;
3781
	head->ids = *ids;
C
Christoph Hellwig 已提交
3782 3783 3784 3785 3786 3787 3788 3789 3790
	kref_init(&head->ref);

	ret = __nvme_check_ids(ctrl->subsys, head);
	if (ret) {
		dev_err(ctrl->device,
			"duplicate IDs for nsid %d\n", nsid);
		goto out_cleanup_srcu;
	}

3791 3792 3793 3794 3795 3796 3797
	if (head->ids.csi) {
		ret = nvme_get_effects_log(ctrl, head->ids.csi, &head->effects);
		if (ret)
			goto out_cleanup_srcu;
	} else
		head->effects = ctrl->effects;

3798 3799 3800 3801
	ret = nvme_mpath_alloc_disk(ctrl, head);
	if (ret)
		goto out_cleanup_srcu;

C
Christoph Hellwig 已提交
3802
	list_add_tail(&head->entry, &ctrl->subsys->nsheads);
3803 3804 3805

	kref_get(&ctrl->subsys->ref);

C
Christoph Hellwig 已提交
3806 3807 3808
	return head;
out_cleanup_srcu:
	cleanup_srcu_struct(&head->srcu);
3809
out_ida_remove:
C
Christoph Hellwig 已提交
3810 3811 3812 3813
	ida_simple_remove(&ctrl->subsys->ns_ida, head->instance);
out_free_head:
	kfree(head);
out:
3814 3815
	if (ret > 0)
		ret = blk_status_to_errno(nvme_error_status(ret));
C
Christoph Hellwig 已提交
3816 3817 3818 3819
	return ERR_PTR(ret);
}

static int nvme_init_ns_head(struct nvme_ns *ns, unsigned nsid,
3820
		struct nvme_ns_ids *ids, bool is_shared)
C
Christoph Hellwig 已提交
3821 3822 3823 3824 3825 3826
{
	struct nvme_ctrl *ctrl = ns->ctrl;
	struct nvme_ns_head *head = NULL;
	int ret = 0;

	mutex_lock(&ctrl->subsys->lock);
3827
	head = nvme_find_ns_head(ctrl->subsys, nsid);
C
Christoph Hellwig 已提交
3828
	if (!head) {
3829
		head = nvme_alloc_ns_head(ctrl, nsid, ids);
C
Christoph Hellwig 已提交
3830 3831 3832 3833
		if (IS_ERR(head)) {
			ret = PTR_ERR(head);
			goto out_unlock;
		}
3834
		head->shared = is_shared;
C
Christoph Hellwig 已提交
3835
	} else {
3836
		ret = -EINVAL;
3837
		if (!is_shared || !head->shared) {
3838
			dev_err(ctrl->device,
3839 3840
				"Duplicate unshared namespace %d\n", nsid);
			goto out_put_ns_head;
3841
		}
3842
		if (!nvme_ns_ids_equal(&head->ids, ids)) {
C
Christoph Hellwig 已提交
3843 3844 3845
			dev_err(ctrl->device,
				"IDs don't match for shared namespace %d\n",
					nsid);
3846
			goto out_put_ns_head;
C
Christoph Hellwig 已提交
3847 3848 3849 3850 3851
		}
	}

	list_add_tail(&ns->siblings, &head->list);
	ns->head = head;
3852 3853
	mutex_unlock(&ctrl->subsys->lock);
	return 0;
C
Christoph Hellwig 已提交
3854

3855 3856
out_put_ns_head:
	nvme_put_ns_head(head);
C
Christoph Hellwig 已提交
3857 3858 3859 3860 3861
out_unlock:
	mutex_unlock(&ctrl->subsys->lock);
	return ret;
}

3862 3863 3864 3865 3866
static int ns_cmp(void *priv, struct list_head *a, struct list_head *b)
{
	struct nvme_ns *nsa = container_of(a, struct nvme_ns, list);
	struct nvme_ns *nsb = container_of(b, struct nvme_ns, list);

C
Christoph Hellwig 已提交
3867
	return nsa->head->ns_id - nsb->head->ns_id;
3868 3869
}

3870
struct nvme_ns *nvme_find_get_ns(struct nvme_ctrl *ctrl, unsigned nsid)
3871
{
3872
	struct nvme_ns *ns, *ret = NULL;
3873

3874
	down_read(&ctrl->namespaces_rwsem);
3875
	list_for_each_entry(ns, &ctrl->namespaces, list) {
C
Christoph Hellwig 已提交
3876
		if (ns->head->ns_id == nsid) {
3877 3878
			if (!kref_get_unless_zero(&ns->kref))
				continue;
3879 3880 3881
			ret = ns;
			break;
		}
C
Christoph Hellwig 已提交
3882
		if (ns->head->ns_id > nsid)
3883 3884
			break;
	}
3885
	up_read(&ctrl->namespaces_rwsem);
3886
	return ret;
3887
}
3888
EXPORT_SYMBOL_NS_GPL(nvme_find_get_ns, NVME_TARGET_PASSTHRU);
3889

3890 3891
static void nvme_alloc_ns(struct nvme_ctrl *ctrl, unsigned nsid,
		struct nvme_ns_ids *ids)
3892 3893 3894
{
	struct nvme_ns *ns;
	struct gendisk *disk;
3895 3896
	struct nvme_id_ns *id;
	char disk_name[DISK_NAME_LEN];
3897
	int node = ctrl->numa_node, flags = GENHD_FL_EXT_DEVT;
3898

3899
	if (nvme_identify_ns(ctrl, nsid, ids, &id))
3900 3901
		return;

3902 3903
	ns = kzalloc_node(sizeof(*ns), GFP_KERNEL, node);
	if (!ns)
3904
		goto out_free_id;
3905 3906

	ns->queue = blk_mq_init_queue(ctrl->tagset);
3907
	if (IS_ERR(ns->queue))
C
Christoph Hellwig 已提交
3908
		goto out_free_ns;
3909

3910
	if (ctrl->opts && ctrl->opts->data_digest)
3911
		blk_queue_flag_set(QUEUE_FLAG_STABLE_WRITES, ns->queue);
3912

3913
	blk_queue_flag_set(QUEUE_FLAG_NONROT, ns->queue);
3914 3915 3916
	if (ctrl->ops->flags & NVME_F_PCI_P2PDMA)
		blk_queue_flag_set(QUEUE_FLAG_PCI_P2PDMA, ns->queue);

3917 3918 3919 3920
	ns->queue->queuedata = ns;
	ns->ctrl = ctrl;
	kref_init(&ns->kref);

3921
	if (nvme_init_ns_head(ns, nsid, ids, id->nmic & NVME_NS_NMIC_SHARED))
3922
		goto out_free_queue;
3923
	nvme_set_disk_name(disk_name, ns, ctrl, &flags);
3924

3925
	disk = alloc_disk_node(0, node);
3926
	if (!disk)
C
Christoph Hellwig 已提交
3927
		goto out_unlink_ns;
3928

J
Javier González 已提交
3929
	disk->fops = &nvme_bdev_ops;
3930 3931
	disk->private_data = ns;
	disk->queue = ns->queue;
3932
	disk->flags = flags;
3933 3934 3935
	memcpy(disk->disk_name, disk_name, DISK_NAME_LEN);
	ns->disk = disk;

3936
	if (nvme_update_ns_info(ns, id))
3937
		goto out_put_disk;
3938

3939
	if ((ctrl->quirks & NVME_QUIRK_LIGHTNVM) && id->vs[0] == 0x1) {
3940
		if (nvme_nvm_register(ns, disk_name, node)) {
3941 3942 3943 3944 3945
			dev_warn(ctrl->device, "LightNVM init failure\n");
			goto out_put_disk;
		}
	}

3946
	down_write(&ctrl->namespaces_rwsem);
3947
	list_add_tail(&ns->list, &ctrl->namespaces);
3948
	up_write(&ctrl->namespaces_rwsem);
3949

3950
	nvme_get_ctrl(ctrl);
3951

3952
	device_add_disk(ctrl->device, ns->disk, nvme_ns_id_attr_groups);
3953

C
Christoph Hellwig 已提交
3954
	nvme_mpath_add_disk(ns, id);
3955
	nvme_fault_inject_init(&ns->fault_inject, ns->disk->disk_name);
C
Christoph Hellwig 已提交
3956 3957
	kfree(id);

3958
	return;
3959
 out_put_disk:
3960 3961
	/* prevent double queue cleanup */
	ns->disk->queue = NULL;
3962
	put_disk(ns->disk);
C
Christoph Hellwig 已提交
3963 3964 3965
 out_unlink_ns:
	mutex_lock(&ctrl->subsys->lock);
	list_del_rcu(&ns->siblings);
3966 3967
	if (list_empty(&ns->head->list))
		list_del_init(&ns->head->entry);
C
Christoph Hellwig 已提交
3968
	mutex_unlock(&ctrl->subsys->lock);
3969
	nvme_put_ns_head(ns->head);
3970 3971 3972 3973
 out_free_queue:
	blk_cleanup_queue(ns->queue);
 out_free_ns:
	kfree(ns);
3974 3975
 out_free_id:
	kfree(id);
3976 3977 3978 3979
}

static void nvme_ns_remove(struct nvme_ns *ns)
{
3980 3981
	if (test_and_set_bit(NVME_NS_REMOVING, &ns->flags))
		return;
3982

3983
	set_capacity(ns->disk, 0);
3984
	nvme_fault_inject_fini(&ns->fault_inject);
3985 3986 3987

	mutex_lock(&ns->ctrl->subsys->lock);
	list_del_rcu(&ns->siblings);
3988 3989
	if (list_empty(&ns->head->list))
		list_del_init(&ns->head->entry);
3990
	mutex_unlock(&ns->ctrl->subsys->lock);
3991

3992 3993 3994 3995
	synchronize_rcu(); /* guarantee not available in head->list */
	nvme_mpath_clear_current_path(ns);
	synchronize_srcu(&ns->head->srcu); /* wait for concurrent submissions */

C
Christoph Hellwig 已提交
3996
	if (ns->disk->flags & GENHD_FL_UP) {
3997 3998
		del_gendisk(ns->disk);
		blk_cleanup_queue(ns->queue);
3999 4000
		if (blk_get_integrity(ns->disk))
			blk_integrity_unregister(ns->disk);
4001
	}
4002

4003
	down_write(&ns->ctrl->namespaces_rwsem);
4004
	list_del_init(&ns->list);
4005
	up_write(&ns->ctrl->namespaces_rwsem);
4006

4007
	nvme_mpath_check_last_path(ns);
4008 4009 4010
	nvme_put_ns(ns);
}

4011 4012 4013 4014 4015 4016 4017 4018 4019 4020
static void nvme_ns_remove_by_nsid(struct nvme_ctrl *ctrl, u32 nsid)
{
	struct nvme_ns *ns = nvme_find_get_ns(ctrl, nsid);

	if (ns) {
		nvme_ns_remove(ns);
		nvme_put_ns(ns);
	}
}

4021
static void nvme_validate_ns(struct nvme_ns *ns, struct nvme_ns_ids *ids)
C
Christoph Hellwig 已提交
4022 4023
{
	struct nvme_id_ns *id;
4024
	int ret = -ENODEV;
C
Christoph Hellwig 已提交
4025

4026 4027
	if (test_bit(NVME_NS_DEAD, &ns->flags))
		goto out;
C
Christoph Hellwig 已提交
4028

4029
	ret = nvme_identify_ns(ns->ctrl, ns->head->ns_id, ids, &id);
C
Christoph Hellwig 已提交
4030 4031 4032
	if (ret)
		goto out;

4033
	ret = -ENODEV;
C
Christoph Hellwig 已提交
4034
	if (!nvme_ns_ids_equal(&ns->head->ids, ids)) {
4035
		dev_err(ns->ctrl->device,
C
Christoph Hellwig 已提交
4036
			"identifiers changed for nsid %d\n", ns->head->ns_id);
4037
		goto out_free_id;
C
Christoph Hellwig 已提交
4038 4039 4040
	}

	ret = nvme_update_ns_info(ns, id);
4041 4042

out_free_id:
C
Christoph Hellwig 已提交
4043 4044 4045
	kfree(id);
out:
	/*
4046
	 * Only remove the namespace if we got a fatal error back from the
C
Christoph Hellwig 已提交
4047
	 * device, otherwise ignore the error and just move on.
4048 4049
	 *
	 * TODO: we should probably schedule a delayed retry here.
C
Christoph Hellwig 已提交
4050
	 */
4051 4052
	if (ret && ret != -ENOMEM && !(ret > 0 && !(ret & NVME_SC_DNR)))
		nvme_ns_remove(ns);
C
Christoph Hellwig 已提交
4053 4054
}

4055
static void nvme_validate_or_alloc_ns(struct nvme_ctrl *ctrl, unsigned nsid)
4056
{
4057
	struct nvme_ns_ids ids = { };
4058 4059
	struct nvme_ns *ns;

4060 4061
	if (nvme_identify_ns_descs(ctrl, nsid, &ids))
		return;
4062

4063
	ns = nvme_find_get_ns(ctrl, nsid);
4064
	if (ns) {
4065
		nvme_validate_ns(ns, &ids);
4066
		nvme_put_ns(ns);
4067 4068 4069
		return;
	}

4070 4071 4072 4073 4074 4075 4076 4077 4078 4079 4080 4081 4082 4083 4084 4085 4086 4087
	switch (ids.csi) {
	case NVME_CSI_NVM:
		nvme_alloc_ns(ctrl, nsid, &ids);
		break;
	case NVME_CSI_ZNS:
		if (!IS_ENABLED(CONFIG_BLK_DEV_ZONED)) {
			dev_warn(ctrl->device,
				"nsid %u not supported without CONFIG_BLK_DEV_ZONED\n",
				nsid);
			break;
		}
		nvme_alloc_ns(ctrl, nsid, &ids);
		break;
	default:
		dev_warn(ctrl->device, "unknown csi %u for nsid %u\n",
			ids.csi, nsid);
		break;
	}
4088 4089
}

4090 4091 4092 4093
static void nvme_remove_invalid_namespaces(struct nvme_ctrl *ctrl,
					unsigned nsid)
{
	struct nvme_ns *ns, *next;
4094
	LIST_HEAD(rm_list);
4095

4096
	down_write(&ctrl->namespaces_rwsem);
4097
	list_for_each_entry_safe(ns, next, &ctrl->namespaces, list) {
4098
		if (ns->head->ns_id > nsid || test_bit(NVME_NS_DEAD, &ns->flags))
4099
			list_move_tail(&ns->list, &rm_list);
4100
	}
4101
	up_write(&ctrl->namespaces_rwsem);
4102 4103 4104 4105

	list_for_each_entry_safe(ns, next, &rm_list, list)
		nvme_ns_remove(ns);

4106 4107
}

4108
static int nvme_scan_ns_list(struct nvme_ctrl *ctrl)
4109
{
4110
	const int nr_entries = NVME_IDENTIFY_DATA_SIZE / sizeof(__le32);
4111
	__le32 *ns_list;
4112 4113
	u32 prev = 0;
	int ret = 0, i;
4114

4115 4116
	if (nvme_ctrl_limited_cns(ctrl))
		return -EOPNOTSUPP;
4117

4118
	ns_list = kzalloc(NVME_IDENTIFY_DATA_SIZE, GFP_KERNEL);
4119 4120 4121
	if (!ns_list)
		return -ENOMEM;

4122
	for (;;) {
4123 4124 4125 4126 4127 4128 4129 4130
		struct nvme_command cmd = {
			.identify.opcode	= nvme_admin_identify,
			.identify.cns		= NVME_ID_CNS_NS_ACTIVE_LIST,
			.identify.nsid		= cpu_to_le32(prev),
		};

		ret = nvme_submit_sync_cmd(ctrl->admin_q, &cmd, ns_list,
					    NVME_IDENTIFY_DATA_SIZE);
4131 4132 4133
		if (ret) {
			dev_warn(ctrl->device,
				"Identify NS List failed (status=0x%x)\n", ret);
4134
			goto free;
4135
		}
4136

4137
		for (i = 0; i < nr_entries; i++) {
4138
			u32 nsid = le32_to_cpu(ns_list[i]);
4139

4140 4141
			if (!nsid)	/* end of the list? */
				goto out;
4142
			nvme_validate_or_alloc_ns(ctrl, nsid);
4143 4144
			while (++prev < nsid)
				nvme_ns_remove_by_nsid(ctrl, prev);
4145 4146 4147
		}
	}
 out:
4148 4149
	nvme_remove_invalid_namespaces(ctrl, prev);
 free:
4150 4151 4152 4153
	kfree(ns_list);
	return ret;
}

4154
static void nvme_scan_ns_sequential(struct nvme_ctrl *ctrl)
4155
{
4156 4157 4158 4159 4160 4161 4162
	struct nvme_id_ctrl *id;
	u32 nn, i;

	if (nvme_identify_ctrl(ctrl, &id))
		return;
	nn = le32_to_cpu(id->nn);
	kfree(id);
4163

4164
	for (i = 1; i <= nn; i++)
4165
		nvme_validate_or_alloc_ns(ctrl, i);
4166

4167
	nvme_remove_invalid_namespaces(ctrl, nn);
4168 4169
}

4170
static void nvme_clear_changed_ns_log(struct nvme_ctrl *ctrl)
4171 4172 4173
{
	size_t log_size = NVME_MAX_CHANGED_NAMESPACES * sizeof(__le32);
	__le32 *log;
4174
	int error;
4175 4176 4177

	log = kzalloc(log_size, GFP_KERNEL);
	if (!log)
4178
		return;
4179

4180 4181 4182 4183 4184 4185
	/*
	 * We need to read the log to clear the AEN, but we don't want to rely
	 * on it for the changed namespace information as userspace could have
	 * raced with us in reading the log page, which could cause us to miss
	 * updates.
	 */
4186 4187
	error = nvme_get_log(ctrl, NVME_NSID_ALL, NVME_LOG_CHANGED_NS, 0,
			NVME_CSI_NVM, log, log_size, 0);
4188
	if (error)
4189 4190 4191 4192 4193 4194
		dev_warn(ctrl->device,
			"reading changed ns log failed: %d\n", error);

	kfree(log);
}

4195
static void nvme_scan_work(struct work_struct *work)
4196
{
4197 4198
	struct nvme_ctrl *ctrl =
		container_of(work, struct nvme_ctrl, scan_work);
4199

K
Keith Busch 已提交
4200 4201
	/* No tagset on a live ctrl means IO queues could not created */
	if (ctrl->state != NVME_CTRL_LIVE || !ctrl->tagset)
4202 4203
		return;

D
Dan Carpenter 已提交
4204
	if (test_and_clear_bit(NVME_AER_NOTICE_NS_CHANGED, &ctrl->events)) {
4205
		dev_info(ctrl->device, "rescanning namespaces.\n");
4206
		nvme_clear_changed_ns_log(ctrl);
4207 4208
	}

4209
	mutex_lock(&ctrl->scan_lock);
4210 4211
	if (nvme_scan_ns_list(ctrl) != 0)
		nvme_scan_ns_sequential(ctrl);
4212
	mutex_unlock(&ctrl->scan_lock);
4213

4214
	down_write(&ctrl->namespaces_rwsem);
4215
	list_sort(NULL, &ctrl->namespaces, ns_cmp);
4216
	up_write(&ctrl->namespaces_rwsem);
4217
}
4218

4219 4220 4221 4222 4223
/*
 * This function iterates the namespace list unlocked to allow recovery from
 * controller failure. It is up to the caller to ensure the namespace list is
 * not modified by scan work while this function is executing.
 */
4224 4225 4226
void nvme_remove_namespaces(struct nvme_ctrl *ctrl)
{
	struct nvme_ns *ns, *next;
4227
	LIST_HEAD(ns_list);
4228

4229 4230 4231 4232 4233 4234 4235
	/*
	 * make sure to requeue I/O to all namespaces as these
	 * might result from the scan itself and must complete
	 * for the scan_work to make progress
	 */
	nvme_mpath_clear_ctrl_paths(ctrl);

4236 4237 4238
	/* prevent racing with ns scanning */
	flush_work(&ctrl->scan_work);

4239 4240 4241 4242 4243 4244 4245 4246 4247
	/*
	 * The dead states indicates the controller was not gracefully
	 * disconnected. In that case, we won't be able to flush any data while
	 * removing the namespaces' disks; fail all the queues now to avoid
	 * potentially having to clean up the failed sync later.
	 */
	if (ctrl->state == NVME_CTRL_DEAD)
		nvme_kill_queues(ctrl);

4248 4249 4250
	/* this is a no-op when called from the controller reset handler */
	nvme_change_ctrl_state(ctrl, NVME_CTRL_DELETING_NOIO);

4251
	down_write(&ctrl->namespaces_rwsem);
4252
	list_splice_init(&ctrl->namespaces, &ns_list);
4253
	up_write(&ctrl->namespaces_rwsem);
4254 4255

	list_for_each_entry_safe(ns, next, &ns_list, list)
4256 4257
		nvme_ns_remove(ns);
}
4258
EXPORT_SYMBOL_GPL(nvme_remove_namespaces);
4259

4260 4261 4262 4263 4264 4265 4266 4267 4268 4269 4270 4271 4272 4273 4274 4275 4276 4277 4278 4279 4280 4281 4282 4283 4284 4285 4286
static int nvme_class_uevent(struct device *dev, struct kobj_uevent_env *env)
{
	struct nvme_ctrl *ctrl =
		container_of(dev, struct nvme_ctrl, ctrl_device);
	struct nvmf_ctrl_options *opts = ctrl->opts;
	int ret;

	ret = add_uevent_var(env, "NVME_TRTYPE=%s", ctrl->ops->name);
	if (ret)
		return ret;

	if (opts) {
		ret = add_uevent_var(env, "NVME_TRADDR=%s", opts->traddr);
		if (ret)
			return ret;

		ret = add_uevent_var(env, "NVME_TRSVCID=%s",
				opts->trsvcid ?: "none");
		if (ret)
			return ret;

		ret = add_uevent_var(env, "NVME_HOST_TRADDR=%s",
				opts->host_traddr ?: "none");
	}
	return ret;
}

4287 4288 4289 4290 4291 4292 4293 4294 4295 4296 4297 4298 4299 4300 4301 4302
static void nvme_aen_uevent(struct nvme_ctrl *ctrl)
{
	char *envp[2] = { NULL, NULL };
	u32 aen_result = ctrl->aen_result;

	ctrl->aen_result = 0;
	if (!aen_result)
		return;

	envp[0] = kasprintf(GFP_KERNEL, "NVME_AEN=%#08x", aen_result);
	if (!envp[0])
		return;
	kobject_uevent_env(&ctrl->device->kobj, KOBJ_CHANGE, envp);
	kfree(envp[0]);
}

4303 4304 4305 4306 4307
static void nvme_async_event_work(struct work_struct *work)
{
	struct nvme_ctrl *ctrl =
		container_of(work, struct nvme_ctrl, async_event_work);

4308
	nvme_aen_uevent(ctrl);
4309
	ctrl->ops->submit_async_event(ctrl);
4310 4311
}

4312 4313 4314 4315 4316 4317 4318 4319 4320 4321 4322 4323 4324 4325 4326 4327 4328 4329 4330 4331 4332 4333
static bool nvme_ctrl_pp_status(struct nvme_ctrl *ctrl)
{

	u32 csts;

	if (ctrl->ops->reg_read32(ctrl, NVME_REG_CSTS, &csts))
		return false;

	if (csts == ~0)
		return false;

	return ((ctrl->ctrl_config & NVME_CC_ENABLE) && (csts & NVME_CSTS_PP));
}

static void nvme_get_fw_slot_info(struct nvme_ctrl *ctrl)
{
	struct nvme_fw_slot_info_log *log;

	log = kmalloc(sizeof(*log), GFP_KERNEL);
	if (!log)
		return;

4334 4335
	if (nvme_get_log(ctrl, NVME_NSID_ALL, NVME_LOG_FW_SLOT, 0, NVME_CSI_NVM,
			log, sizeof(*log), 0))
4336
		dev_warn(ctrl->device, "Get FW SLOT INFO log error\n");
4337 4338 4339 4340 4341 4342 4343 4344 4345 4346 4347 4348 4349 4350 4351 4352 4353 4354 4355 4356 4357
	kfree(log);
}

static void nvme_fw_act_work(struct work_struct *work)
{
	struct nvme_ctrl *ctrl = container_of(work,
				struct nvme_ctrl, fw_act_work);
	unsigned long fw_act_timeout;

	if (ctrl->mtfa)
		fw_act_timeout = jiffies +
				msecs_to_jiffies(ctrl->mtfa * 100);
	else
		fw_act_timeout = jiffies +
				msecs_to_jiffies(admin_timeout * 1000);

	nvme_stop_queues(ctrl);
	while (nvme_ctrl_pp_status(ctrl)) {
		if (time_after(jiffies, fw_act_timeout)) {
			dev_warn(ctrl->device,
				"Fw activation timeout, reset controller\n");
4358 4359
			nvme_try_sched_reset(ctrl);
			return;
4360 4361 4362 4363
		}
		msleep(100);
	}

4364
	if (!nvme_change_ctrl_state(ctrl, NVME_CTRL_LIVE))
4365 4366 4367
		return;

	nvme_start_queues(ctrl);
4368
	/* read FW slot information to clear the AER */
4369 4370 4371
	nvme_get_fw_slot_info(ctrl);
}

4372 4373
static void nvme_handle_aen_notice(struct nvme_ctrl *ctrl, u32 result)
{
4374 4375
	u32 aer_notice_type = (result & 0xff00) >> 8;

4376 4377
	trace_nvme_async_event(ctrl, aer_notice_type);

4378
	switch (aer_notice_type) {
4379
	case NVME_AER_NOTICE_NS_CHANGED:
D
Dan Carpenter 已提交
4380
		set_bit(NVME_AER_NOTICE_NS_CHANGED, &ctrl->events);
4381 4382 4383
		nvme_queue_scan(ctrl);
		break;
	case NVME_AER_NOTICE_FW_ACT_STARTING:
4384 4385 4386 4387 4388 4389 4390
		/*
		 * We are (ab)using the RESETTING state to prevent subsequent
		 * recovery actions from interfering with the controller's
		 * firmware activation.
		 */
		if (nvme_change_ctrl_state(ctrl, NVME_CTRL_RESETTING))
			queue_work(nvme_wq, &ctrl->fw_act_work);
4391
		break;
C
Christoph Hellwig 已提交
4392 4393 4394 4395 4396 4397 4398
#ifdef CONFIG_NVME_MULTIPATH
	case NVME_AER_NOTICE_ANA:
		if (!ctrl->ana_log_buf)
			break;
		queue_work(nvme_wq, &ctrl->ana_work);
		break;
#endif
4399 4400 4401
	case NVME_AER_NOTICE_DISC_CHANGED:
		ctrl->aen_result = result;
		break;
4402 4403 4404 4405 4406
	default:
		dev_warn(ctrl->device, "async event result %08x\n", result);
	}
}

4407
void nvme_complete_async_event(struct nvme_ctrl *ctrl, __le16 status,
4408
		volatile union nvme_result *res)
4409
{
4410
	u32 result = le32_to_cpu(res->u32);
4411
	u32 aer_type = result & 0x07;
4412

4413
	if (le16_to_cpu(status) >> 1 != NVME_SC_SUCCESS)
4414 4415
		return;

4416
	switch (aer_type) {
4417 4418 4419
	case NVME_AER_NOTICE:
		nvme_handle_aen_notice(ctrl, result);
		break;
4420 4421 4422 4423
	case NVME_AER_ERROR:
	case NVME_AER_SMART:
	case NVME_AER_CSS:
	case NVME_AER_VS:
4424
		trace_nvme_async_event(ctrl, aer_type);
4425
		ctrl->aen_result = result;
4426 4427 4428
		break;
	default:
		break;
4429
	}
4430
	queue_work(nvme_wq, &ctrl->async_event_work);
4431 4432
}
EXPORT_SYMBOL_GPL(nvme_complete_async_event);
4433

4434
void nvme_stop_ctrl(struct nvme_ctrl *ctrl)
4435
{
C
Christoph Hellwig 已提交
4436
	nvme_mpath_stop(ctrl);
4437
	nvme_stop_keep_alive(ctrl);
4438
	nvme_stop_failfast_work(ctrl);
4439
	flush_work(&ctrl->async_event_work);
4440
	cancel_work_sync(&ctrl->fw_act_work);
4441 4442 4443 4444 4445
}
EXPORT_SYMBOL_GPL(nvme_stop_ctrl);

void nvme_start_ctrl(struct nvme_ctrl *ctrl)
{
4446
	nvme_start_keep_alive(ctrl);
4447

4448 4449
	nvme_enable_aen(ctrl);

4450 4451 4452 4453 4454 4455
	if (ctrl->queue_count > 1) {
		nvme_queue_scan(ctrl);
		nvme_start_queues(ctrl);
	}
}
EXPORT_SYMBOL_GPL(nvme_start_ctrl);
4456

4457 4458
void nvme_uninit_ctrl(struct nvme_ctrl *ctrl)
{
4459
	nvme_fault_inject_fini(&ctrl->fault_inject);
4460
	dev_pm_qos_hide_latency_tolerance(ctrl->device);
4461
	cdev_device_del(&ctrl->cdev, ctrl->device);
4462
	nvme_put_ctrl(ctrl);
4463
}
4464
EXPORT_SYMBOL_GPL(nvme_uninit_ctrl);
4465

4466 4467 4468 4469 4470
static void nvme_free_cels(struct nvme_ctrl *ctrl)
{
	struct nvme_effects_log	*cel;
	unsigned long i;

4471
	xa_for_each(&ctrl->cels, i, cel) {
4472 4473 4474 4475 4476 4477 4478
		xa_erase(&ctrl->cels, i);
		kfree(cel);
	}

	xa_destroy(&ctrl->cels);
}

4479
static void nvme_free_ctrl(struct device *dev)
4480
{
4481 4482
	struct nvme_ctrl *ctrl =
		container_of(dev, struct nvme_ctrl, ctrl_device);
C
Christoph Hellwig 已提交
4483
	struct nvme_subsystem *subsys = ctrl->subsys;
4484

K
Keith Busch 已提交
4485
	if (!subsys || ctrl->instance != subsys->instance)
4486 4487
		ida_simple_remove(&nvme_instance_ida, ctrl->instance);

4488
	nvme_free_cels(ctrl);
C
Christoph Hellwig 已提交
4489
	nvme_mpath_uninit(ctrl);
S
Sagi Grimberg 已提交
4490
	__free_page(ctrl->discard_page);
4491

C
Christoph Hellwig 已提交
4492
	if (subsys) {
4493
		mutex_lock(&nvme_subsystems_lock);
C
Christoph Hellwig 已提交
4494 4495
		list_del(&ctrl->subsys_entry);
		sysfs_remove_link(&subsys->dev.kobj, dev_name(ctrl->device));
4496
		mutex_unlock(&nvme_subsystems_lock);
C
Christoph Hellwig 已提交
4497
	}
4498 4499 4500

	ctrl->ops->free_ctrl(ctrl);

C
Christoph Hellwig 已提交
4501 4502
	if (subsys)
		nvme_put_subsystem(subsys);
4503 4504 4505 4506 4507 4508 4509 4510 4511 4512 4513 4514
}

/*
 * Initialize a NVMe controller structures.  This needs to be called during
 * earliest initialization so that we have the initialized structured around
 * during probing.
 */
int nvme_init_ctrl(struct nvme_ctrl *ctrl, struct device *dev,
		const struct nvme_ctrl_ops *ops, unsigned long quirks)
{
	int ret;

4515
	ctrl->state = NVME_CTRL_NEW;
4516
	clear_bit(NVME_CTRL_FAILFAST_EXPIRED, &ctrl->flags);
4517
	spin_lock_init(&ctrl->lock);
4518
	mutex_init(&ctrl->scan_lock);
4519
	INIT_LIST_HEAD(&ctrl->namespaces);
4520
	xa_init(&ctrl->cels);
4521
	init_rwsem(&ctrl->namespaces_rwsem);
4522 4523 4524
	ctrl->dev = dev;
	ctrl->ops = ops;
	ctrl->quirks = quirks;
4525
	ctrl->numa_node = NUMA_NO_NODE;
4526
	INIT_WORK(&ctrl->scan_work, nvme_scan_work);
4527
	INIT_WORK(&ctrl->async_event_work, nvme_async_event_work);
4528
	INIT_WORK(&ctrl->fw_act_work, nvme_fw_act_work);
4529
	INIT_WORK(&ctrl->delete_work, nvme_delete_ctrl_work);
4530
	init_waitqueue_head(&ctrl->state_wq);
4531

4532
	INIT_DELAYED_WORK(&ctrl->ka_work, nvme_keep_alive_work);
4533
	INIT_DELAYED_WORK(&ctrl->failfast_work, nvme_failfast_work);
4534 4535 4536
	memset(&ctrl->ka_cmd, 0, sizeof(ctrl->ka_cmd));
	ctrl->ka_cmd.common.opcode = nvme_admin_keep_alive;

4537 4538 4539 4540 4541 4542 4543 4544
	BUILD_BUG_ON(NVME_DSM_MAX_RANGES * sizeof(struct nvme_dsm_range) >
			PAGE_SIZE);
	ctrl->discard_page = alloc_page(GFP_KERNEL);
	if (!ctrl->discard_page) {
		ret = -ENOMEM;
		goto out;
	}

4545 4546
	ret = ida_simple_get(&nvme_instance_ida, 0, 0, GFP_KERNEL);
	if (ret < 0)
4547
		goto out;
4548
	ctrl->instance = ret;
4549

4550 4551
	device_initialize(&ctrl->ctrl_device);
	ctrl->device = &ctrl->ctrl_device;
4552 4553
	ctrl->device->devt = MKDEV(MAJOR(nvme_ctrl_base_chr_devt),
			ctrl->instance);
4554 4555 4556 4557 4558 4559 4560
	ctrl->device->class = nvme_class;
	ctrl->device->parent = ctrl->dev;
	ctrl->device->groups = nvme_dev_attr_groups;
	ctrl->device->release = nvme_free_ctrl;
	dev_set_drvdata(ctrl->device, ctrl);
	ret = dev_set_name(ctrl->device, "nvme%d", ctrl->instance);
	if (ret)
4561 4562
		goto out_release_instance;

4563
	nvme_get_ctrl(ctrl);
4564 4565 4566
	cdev_init(&ctrl->cdev, &nvme_dev_fops);
	ctrl->cdev.owner = ops->module;
	ret = cdev_device_add(&ctrl->cdev, ctrl->device);
4567 4568
	if (ret)
		goto out_free_name;
4569

4570 4571 4572 4573 4574 4575 4576 4577
	/*
	 * Initialize latency tolerance controls.  The sysfs files won't
	 * be visible to userspace unless the device actually supports APST.
	 */
	ctrl->device->power.set_latency_tolerance = nvme_set_latency_tolerance;
	dev_pm_qos_update_user_latency_tolerance(ctrl->device,
		min(default_ps_max_latency_us, (unsigned long)S32_MAX));

4578 4579
	nvme_fault_inject_init(&ctrl->fault_inject, dev_name(ctrl->device));

4580
	return 0;
4581
out_free_name:
4582
	nvme_put_ctrl(ctrl);
4583
	kfree_const(ctrl->device->kobj.name);
4584
out_release_instance:
4585
	ida_simple_remove(&nvme_instance_ida, ctrl->instance);
4586
out:
4587 4588
	if (ctrl->discard_page)
		__free_page(ctrl->discard_page);
4589 4590
	return ret;
}
4591
EXPORT_SYMBOL_GPL(nvme_init_ctrl);
4592

4593 4594 4595 4596 4597 4598 4599 4600 4601 4602 4603
/**
 * nvme_kill_queues(): Ends all namespace queues
 * @ctrl: the dead controller that needs to end
 *
 * Call this function when the driver determines it is unable to get the
 * controller in a state capable of servicing IO.
 */
void nvme_kill_queues(struct nvme_ctrl *ctrl)
{
	struct nvme_ns *ns;

4604
	down_read(&ctrl->namespaces_rwsem);
M
Ming Lei 已提交
4605

4606
	/* Forcibly unquiesce queues to avoid blocking dispatch */
I
Igor Konopko 已提交
4607
	if (ctrl->admin_q && !blk_queue_dying(ctrl->admin_q))
4608
		blk_mq_unquiesce_queue(ctrl->admin_q);
4609

4610 4611
	list_for_each_entry(ns, &ctrl->namespaces, list)
		nvme_set_queue_dying(ns);
4612

4613
	up_read(&ctrl->namespaces_rwsem);
4614
}
4615
EXPORT_SYMBOL_GPL(nvme_kill_queues);
4616

K
Keith Busch 已提交
4617 4618 4619 4620
void nvme_unfreeze(struct nvme_ctrl *ctrl)
{
	struct nvme_ns *ns;

4621
	down_read(&ctrl->namespaces_rwsem);
K
Keith Busch 已提交
4622 4623
	list_for_each_entry(ns, &ctrl->namespaces, list)
		blk_mq_unfreeze_queue(ns->queue);
4624
	up_read(&ctrl->namespaces_rwsem);
K
Keith Busch 已提交
4625 4626 4627
}
EXPORT_SYMBOL_GPL(nvme_unfreeze);

4628
int nvme_wait_freeze_timeout(struct nvme_ctrl *ctrl, long timeout)
K
Keith Busch 已提交
4629 4630 4631
{
	struct nvme_ns *ns;

4632
	down_read(&ctrl->namespaces_rwsem);
K
Keith Busch 已提交
4633 4634 4635 4636 4637
	list_for_each_entry(ns, &ctrl->namespaces, list) {
		timeout = blk_mq_freeze_queue_wait_timeout(ns->queue, timeout);
		if (timeout <= 0)
			break;
	}
4638
	up_read(&ctrl->namespaces_rwsem);
4639
	return timeout;
K
Keith Busch 已提交
4640 4641 4642 4643 4644 4645 4646
}
EXPORT_SYMBOL_GPL(nvme_wait_freeze_timeout);

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

4647
	down_read(&ctrl->namespaces_rwsem);
K
Keith Busch 已提交
4648 4649
	list_for_each_entry(ns, &ctrl->namespaces, list)
		blk_mq_freeze_queue_wait(ns->queue);
4650
	up_read(&ctrl->namespaces_rwsem);
K
Keith Busch 已提交
4651 4652 4653 4654 4655 4656 4657
}
EXPORT_SYMBOL_GPL(nvme_wait_freeze);

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

4658
	down_read(&ctrl->namespaces_rwsem);
K
Keith Busch 已提交
4659
	list_for_each_entry(ns, &ctrl->namespaces, list)
4660
		blk_freeze_queue_start(ns->queue);
4661
	up_read(&ctrl->namespaces_rwsem);
K
Keith Busch 已提交
4662 4663 4664
}
EXPORT_SYMBOL_GPL(nvme_start_freeze);

4665
void nvme_stop_queues(struct nvme_ctrl *ctrl)
4666 4667 4668
{
	struct nvme_ns *ns;

4669
	down_read(&ctrl->namespaces_rwsem);
4670
	list_for_each_entry(ns, &ctrl->namespaces, list)
4671
		blk_mq_quiesce_queue(ns->queue);
4672
	up_read(&ctrl->namespaces_rwsem);
4673
}
4674
EXPORT_SYMBOL_GPL(nvme_stop_queues);
4675

4676
void nvme_start_queues(struct nvme_ctrl *ctrl)
4677 4678 4679
{
	struct nvme_ns *ns;

4680
	down_read(&ctrl->namespaces_rwsem);
4681
	list_for_each_entry(ns, &ctrl->namespaces, list)
4682
		blk_mq_unquiesce_queue(ns->queue);
4683
	up_read(&ctrl->namespaces_rwsem);
4684
}
4685
EXPORT_SYMBOL_GPL(nvme_start_queues);
4686

C
Chao Leng 已提交
4687
void nvme_sync_io_queues(struct nvme_ctrl *ctrl)
K
Keith Busch 已提交
4688 4689 4690 4691 4692 4693 4694
{
	struct nvme_ns *ns;

	down_read(&ctrl->namespaces_rwsem);
	list_for_each_entry(ns, &ctrl->namespaces, list)
		blk_sync_queue(ns->queue);
	up_read(&ctrl->namespaces_rwsem);
C
Chao Leng 已提交
4695 4696
}
EXPORT_SYMBOL_GPL(nvme_sync_io_queues);
4697

C
Chao Leng 已提交
4698 4699 4700
void nvme_sync_queues(struct nvme_ctrl *ctrl)
{
	nvme_sync_io_queues(ctrl);
4701 4702
	if (ctrl->admin_q)
		blk_sync_queue(ctrl->admin_q);
K
Keith Busch 已提交
4703 4704 4705
}
EXPORT_SYMBOL_GPL(nvme_sync_queues);

4706
struct nvme_ctrl *nvme_ctrl_from_file(struct file *file)
4707
{
4708 4709 4710
	if (file->f_op != &nvme_dev_fops)
		return NULL;
	return file->private_data;
4711
}
4712
EXPORT_SYMBOL_NS_GPL(nvme_ctrl_from_file, NVME_TARGET_PASSTHRU);
4713

4714 4715 4716 4717 4718 4719 4720 4721 4722 4723 4724 4725 4726 4727 4728 4729 4730 4731
/*
 * Check we didn't inadvertently grow the command structure sizes:
 */
static inline void _nvme_check_size(void)
{
	BUILD_BUG_ON(sizeof(struct nvme_common_command) != 64);
	BUILD_BUG_ON(sizeof(struct nvme_rw_command) != 64);
	BUILD_BUG_ON(sizeof(struct nvme_identify) != 64);
	BUILD_BUG_ON(sizeof(struct nvme_features) != 64);
	BUILD_BUG_ON(sizeof(struct nvme_download_firmware) != 64);
	BUILD_BUG_ON(sizeof(struct nvme_format_cmd) != 64);
	BUILD_BUG_ON(sizeof(struct nvme_dsm_cmd) != 64);
	BUILD_BUG_ON(sizeof(struct nvme_write_zeroes_cmd) != 64);
	BUILD_BUG_ON(sizeof(struct nvme_abort_cmd) != 64);
	BUILD_BUG_ON(sizeof(struct nvme_get_log_page_command) != 64);
	BUILD_BUG_ON(sizeof(struct nvme_command) != 64);
	BUILD_BUG_ON(sizeof(struct nvme_id_ctrl) != NVME_IDENTIFY_DATA_SIZE);
	BUILD_BUG_ON(sizeof(struct nvme_id_ns) != NVME_IDENTIFY_DATA_SIZE);
K
Keith Busch 已提交
4732 4733
	BUILD_BUG_ON(sizeof(struct nvme_id_ns_zns) != NVME_IDENTIFY_DATA_SIZE);
	BUILD_BUG_ON(sizeof(struct nvme_id_ctrl_zns) != NVME_IDENTIFY_DATA_SIZE);
4734 4735 4736 4737 4738 4739 4740
	BUILD_BUG_ON(sizeof(struct nvme_lba_range_type) != 64);
	BUILD_BUG_ON(sizeof(struct nvme_smart_log) != 512);
	BUILD_BUG_ON(sizeof(struct nvme_dbbuf) != 64);
	BUILD_BUG_ON(sizeof(struct nvme_directive_cmd) != 64);
}


4741
static int __init nvme_core_init(void)
4742
{
4743
	int result = -ENOMEM;
4744

4745 4746
	_nvme_check_size();

4747 4748 4749
	nvme_wq = alloc_workqueue("nvme-wq",
			WQ_UNBOUND | WQ_MEM_RECLAIM | WQ_SYSFS, 0);
	if (!nvme_wq)
4750 4751 4752 4753 4754 4755 4756 4757 4758 4759 4760
		goto out;

	nvme_reset_wq = alloc_workqueue("nvme-reset-wq",
			WQ_UNBOUND | WQ_MEM_RECLAIM | WQ_SYSFS, 0);
	if (!nvme_reset_wq)
		goto destroy_wq;

	nvme_delete_wq = alloc_workqueue("nvme-delete-wq",
			WQ_UNBOUND | WQ_MEM_RECLAIM | WQ_SYSFS, 0);
	if (!nvme_delete_wq)
		goto destroy_reset_wq;
4761

4762 4763
	result = alloc_chrdev_region(&nvme_ctrl_base_chr_devt, 0,
			NVME_MINORS, "nvme");
4764
	if (result < 0)
4765
		goto destroy_delete_wq;
4766 4767 4768 4769 4770 4771

	nvme_class = class_create(THIS_MODULE, "nvme");
	if (IS_ERR(nvme_class)) {
		result = PTR_ERR(nvme_class);
		goto unregister_chrdev;
	}
4772
	nvme_class->dev_uevent = nvme_class_uevent;
4773

C
Christoph Hellwig 已提交
4774 4775 4776 4777 4778
	nvme_subsys_class = class_create(THIS_MODULE, "nvme-subsystem");
	if (IS_ERR(nvme_subsys_class)) {
		result = PTR_ERR(nvme_subsys_class);
		goto destroy_class;
	}
4779
	return 0;
4780

C
Christoph Hellwig 已提交
4781 4782
destroy_class:
	class_destroy(nvme_class);
4783
unregister_chrdev:
4784
	unregister_chrdev_region(nvme_ctrl_base_chr_devt, NVME_MINORS);
4785 4786 4787 4788
destroy_delete_wq:
	destroy_workqueue(nvme_delete_wq);
destroy_reset_wq:
	destroy_workqueue(nvme_reset_wq);
4789 4790
destroy_wq:
	destroy_workqueue(nvme_wq);
4791
out:
4792
	return result;
4793 4794
}

4795
static void __exit nvme_core_exit(void)
4796
{
C
Christoph Hellwig 已提交
4797
	class_destroy(nvme_subsys_class);
4798
	class_destroy(nvme_class);
4799
	unregister_chrdev_region(nvme_ctrl_base_chr_devt, NVME_MINORS);
4800 4801
	destroy_workqueue(nvme_delete_wq);
	destroy_workqueue(nvme_reset_wq);
4802
	destroy_workqueue(nvme_wq);
M
Max Gurtovoy 已提交
4803
	ida_destroy(&nvme_instance_ida);
4804
}
4805 4806 4807 4808 4809

MODULE_LICENSE("GPL");
MODULE_VERSION("1.0");
module_init(nvme_core_init);
module_exit(nvme_core_exit);