blk-mq.c 38.4 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75
#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/backing-dev.h>
#include <linux/bio.h>
#include <linux/blkdev.h>
#include <linux/mm.h>
#include <linux/init.h>
#include <linux/slab.h>
#include <linux/workqueue.h>
#include <linux/smp.h>
#include <linux/llist.h>
#include <linux/list_sort.h>
#include <linux/cpu.h>
#include <linux/cache.h>
#include <linux/sched/sysctl.h>
#include <linux/delay.h>

#include <trace/events/block.h>

#include <linux/blk-mq.h>
#include "blk.h"
#include "blk-mq.h"
#include "blk-mq-tag.h"

static DEFINE_MUTEX(all_q_mutex);
static LIST_HEAD(all_q_list);

static void __blk_mq_run_hw_queue(struct blk_mq_hw_ctx *hctx);

static struct blk_mq_ctx *__blk_mq_get_ctx(struct request_queue *q,
					   unsigned int cpu)
{
	return per_cpu_ptr(q->queue_ctx, cpu);
}

/*
 * This assumes per-cpu software queueing queues. They could be per-node
 * as well, for instance. For now this is hardcoded as-is. Note that we don't
 * care about preemption, since we know the ctx's are persistent. This does
 * mean that we can't rely on ctx always matching the currently running CPU.
 */
static struct blk_mq_ctx *blk_mq_get_ctx(struct request_queue *q)
{
	return __blk_mq_get_ctx(q, get_cpu());
}

static void blk_mq_put_ctx(struct blk_mq_ctx *ctx)
{
	put_cpu();
}

/*
 * Check if any of the ctx's have pending work in this hardware queue
 */
static bool blk_mq_hctx_has_pending(struct blk_mq_hw_ctx *hctx)
{
	unsigned int i;

	for (i = 0; i < hctx->nr_ctx_map; i++)
		if (hctx->ctx_map[i])
			return true;

	return false;
}

/*
 * Mark this ctx as having pending work in this hardware queue
 */
static void blk_mq_hctx_mark_pending(struct blk_mq_hw_ctx *hctx,
				     struct blk_mq_ctx *ctx)
{
	if (!test_bit(ctx->index_hw, hctx->ctx_map))
		set_bit(ctx->index_hw, hctx->ctx_map);
}

76
static struct request *__blk_mq_alloc_request(struct blk_mq_hw_ctx *hctx,
77
					      struct blk_mq_ctx *ctx,
78
					      gfp_t gfp, bool reserved)
79 80 81 82
{
	struct request *rq;
	unsigned int tag;

83
	tag = blk_mq_get_tag(hctx->tags, hctx, &ctx->last_tag, gfp, reserved);
84
	if (tag != BLK_MQ_TAG_FAIL) {
85
		rq = hctx->tags->rqs[tag];
86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106
		rq->tag = tag;
		return rq;
	}

	return NULL;
}

static int blk_mq_queue_enter(struct request_queue *q)
{
	int ret;

	__percpu_counter_add(&q->mq_usage_counter, 1, 1000000);
	smp_wmb();
	/* we have problems to freeze the queue if it's initializing */
	if (!blk_queue_bypass(q) || !blk_queue_init_done(q))
		return 0;

	__percpu_counter_add(&q->mq_usage_counter, -1, 1000000);

	spin_lock_irq(q->queue_lock);
	ret = wait_event_interruptible_lock_irq(q->mq_freeze_wq,
107 108
		!blk_queue_bypass(q) || blk_queue_dying(q),
		*q->queue_lock);
109
	/* inc usage with lock hold to avoid freeze_queue runs here */
110
	if (!ret && !blk_queue_dying(q))
111
		__percpu_counter_add(&q->mq_usage_counter, 1, 1000000);
112 113
	else if (blk_queue_dying(q))
		ret = -ENODEV;
114 115 116 117 118 119 120 121 122 123
	spin_unlock_irq(q->queue_lock);

	return ret;
}

static void blk_mq_queue_exit(struct request_queue *q)
{
	__percpu_counter_add(&q->mq_usage_counter, -1, 1000000);
}

124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139
static void __blk_mq_drain_queue(struct request_queue *q)
{
	while (true) {
		s64 count;

		spin_lock_irq(q->queue_lock);
		count = percpu_counter_sum(&q->mq_usage_counter);
		spin_unlock_irq(q->queue_lock);

		if (count == 0)
			break;
		blk_mq_run_queues(q, false);
		msleep(10);
	}
}

140 141 142 143 144 145 146 147 148 149 150 151 152
/*
 * Guarantee no request is in use, so we can change any data structure of
 * the queue afterward.
 */
static void blk_mq_freeze_queue(struct request_queue *q)
{
	bool drain;

	spin_lock_irq(q->queue_lock);
	drain = !q->bypass_depth++;
	queue_flag_set(QUEUE_FLAG_BYPASS, q);
	spin_unlock_irq(q->queue_lock);

153 154 155
	if (drain)
		__blk_mq_drain_queue(q);
}
156

157 158 159
void blk_mq_drain_queue(struct request_queue *q)
{
	__blk_mq_drain_queue(q);
160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182
}

static void blk_mq_unfreeze_queue(struct request_queue *q)
{
	bool wake = false;

	spin_lock_irq(q->queue_lock);
	if (!--q->bypass_depth) {
		queue_flag_clear(QUEUE_FLAG_BYPASS, q);
		wake = true;
	}
	WARN_ON_ONCE(q->bypass_depth < 0);
	spin_unlock_irq(q->queue_lock);
	if (wake)
		wake_up_all(&q->mq_freeze_wq);
}

bool blk_mq_can_queue(struct blk_mq_hw_ctx *hctx)
{
	return blk_mq_has_free_tags(hctx->tags);
}
EXPORT_SYMBOL(blk_mq_can_queue);

183 184
static void blk_mq_rq_ctx_init(struct request_queue *q, struct blk_mq_ctx *ctx,
			       struct request *rq, unsigned int rw_flags)
185
{
186 187 188
	if (blk_queue_io_stat(q))
		rw_flags |= REQ_IO_STAT;

189 190 191
	INIT_LIST_HEAD(&rq->queuelist);
	/* csd/requeue_work/fifo_time is initialized before use */
	rq->q = q;
192 193
	rq->mq_ctx = ctx;
	rq->cmd_flags = rw_flags;
194 195 196 197 198 199 200 201 202 203 204 205
	rq->cmd_type = 0;
	/* do not touch atomic flags, it needs atomic ops against the timer */
	rq->cpu = -1;
	rq->__data_len = 0;
	rq->__sector = (sector_t) -1;
	rq->bio = NULL;
	rq->biotail = NULL;
	INIT_HLIST_NODE(&rq->hash);
	RB_CLEAR_NODE(&rq->rb_node);
	memset(&rq->flush, 0, max(sizeof(rq->flush), sizeof(rq->elv)));
	rq->rq_disk = NULL;
	rq->part = NULL;
206
	rq->start_time = jiffies;
207 208
#ifdef CONFIG_BLK_CGROUP
	rq->rl = NULL;
209
	set_start_time_ns(rq);
210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236
	rq->io_start_time_ns = 0;
#endif
	rq->nr_phys_segments = 0;
#if defined(CONFIG_BLK_DEV_INTEGRITY)
	rq->nr_integrity_segments = 0;
#endif
	rq->ioprio = 0;
	rq->special = NULL;
	/* tag was already set */
	rq->errors = 0;
	memset(rq->__cmd, 0, sizeof(rq->__cmd));
	rq->cmd = rq->__cmd;
	rq->cmd_len = BLK_MAX_CDB;

	rq->extra_len = 0;
	rq->sense_len = 0;
	rq->resid_len = 0;
	rq->sense = NULL;

	rq->deadline = 0;
	INIT_LIST_HEAD(&rq->timeout_list);
	rq->timeout = 0;
	rq->retries = 0;
	rq->end_io = NULL;
	rq->end_io_data = NULL;
	rq->next_rq = NULL;

237 238 239 240 241 242 243 244 245 246 247 248 249
	ctx->rq_dispatched[rw_is_sync(rw_flags)]++;
}

static struct request *blk_mq_alloc_request_pinned(struct request_queue *q,
						   int rw, gfp_t gfp,
						   bool reserved)
{
	struct request *rq;

	do {
		struct blk_mq_ctx *ctx = blk_mq_get_ctx(q);
		struct blk_mq_hw_ctx *hctx = q->mq_ops->map_queue(q, ctx->cpu);

250 251
		rq = __blk_mq_alloc_request(hctx, ctx, gfp & ~__GFP_WAIT,
						reserved);
252
		if (rq) {
253
			blk_mq_rq_ctx_init(q, ctx, rq, rw);
254
			break;
255
		}
256

257 258 259 260 261
		if (gfp & __GFP_WAIT) {
			__blk_mq_run_hw_queue(hctx);
			blk_mq_put_ctx(ctx);
		} else {
			blk_mq_put_ctx(ctx);
262
			break;
263
		}
264

265
		blk_mq_wait_for_tags(hctx->tags, hctx, reserved);
266 267 268 269 270
	} while (1);

	return rq;
}

271
struct request *blk_mq_alloc_request(struct request_queue *q, int rw, gfp_t gfp)
272 273 274 275 276 277
{
	struct request *rq;

	if (blk_mq_queue_enter(q))
		return NULL;

278
	rq = blk_mq_alloc_request_pinned(q, rw, gfp, false);
279 280
	if (rq)
		blk_mq_put_ctx(rq->mq_ctx);
281 282
	return rq;
}
283
EXPORT_SYMBOL(blk_mq_alloc_request);
284 285 286 287 288 289 290 291 292 293

struct request *blk_mq_alloc_reserved_request(struct request_queue *q, int rw,
					      gfp_t gfp)
{
	struct request *rq;

	if (blk_mq_queue_enter(q))
		return NULL;

	rq = blk_mq_alloc_request_pinned(q, rw, gfp, true);
294 295
	if (rq)
		blk_mq_put_ctx(rq->mq_ctx);
296 297 298 299 300 301 302 303 304 305
	return rq;
}
EXPORT_SYMBOL(blk_mq_alloc_reserved_request);

static void __blk_mq_free_request(struct blk_mq_hw_ctx *hctx,
				  struct blk_mq_ctx *ctx, struct request *rq)
{
	const int tag = rq->tag;
	struct request_queue *q = rq->q;

306
	clear_bit(REQ_ATOM_STARTED, &rq->atomic_flags);
307
	blk_mq_put_tag(hctx->tags, tag, &ctx->last_tag);
308 309 310 311 312 313 314 315 316 317 318 319 320 321 322
	blk_mq_queue_exit(q);
}

void blk_mq_free_request(struct request *rq)
{
	struct blk_mq_ctx *ctx = rq->mq_ctx;
	struct blk_mq_hw_ctx *hctx;
	struct request_queue *q = rq->q;

	ctx->rq_completed[rq_is_sync(rq)]++;

	hctx = q->mq_ops->map_queue(q, ctx->cpu);
	__blk_mq_free_request(hctx, ctx, rq);
}

323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342
/*
 * Clone all relevant state from a request that has been put on hold in
 * the flush state machine into the preallocated flush request that hangs
 * off the request queue.
 *
 * For a driver the flush request should be invisible, that's why we are
 * impersonating the original request here.
 */
void blk_mq_clone_flush_request(struct request *flush_rq,
		struct request *orig_rq)
{
	struct blk_mq_hw_ctx *hctx =
		orig_rq->q->mq_ops->map_queue(orig_rq->q, orig_rq->mq_ctx->cpu);

	flush_rq->mq_ctx = orig_rq->mq_ctx;
	flush_rq->tag = orig_rq->tag;
	memcpy(blk_mq_rq_to_pdu(flush_rq), blk_mq_rq_to_pdu(orig_rq),
		hctx->cmd_size);
}

343
inline void __blk_mq_end_io(struct request *rq, int error)
344
{
M
Ming Lei 已提交
345 346
	blk_account_io_done(rq);

C
Christoph Hellwig 已提交
347
	if (rq->end_io) {
348
		rq->end_io(rq, error);
C
Christoph Hellwig 已提交
349 350 351
	} else {
		if (unlikely(blk_bidi_rq(rq)))
			blk_mq_free_request(rq->next_rq);
352
		blk_mq_free_request(rq);
C
Christoph Hellwig 已提交
353
	}
354
}
355 356 357 358 359 360 361 362 363
EXPORT_SYMBOL(__blk_mq_end_io);

void blk_mq_end_io(struct request *rq, int error)
{
	if (blk_update_request(rq, error, blk_rq_bytes(rq)))
		BUG();
	__blk_mq_end_io(rq, error);
}
EXPORT_SYMBOL(blk_mq_end_io);
364

365
static void __blk_mq_complete_request_remote(void *data)
366
{
367
	struct request *rq = data;
368

369
	rq->q->softirq_done_fn(rq);
370 371
}

372
void __blk_mq_complete_request(struct request *rq)
373 374
{
	struct blk_mq_ctx *ctx = rq->mq_ctx;
C
Christoph Hellwig 已提交
375
	bool shared = false;
376 377
	int cpu;

C
Christoph Hellwig 已提交
378
	if (!test_bit(QUEUE_FLAG_SAME_COMP, &rq->q->queue_flags)) {
379 380 381
		rq->q->softirq_done_fn(rq);
		return;
	}
382 383

	cpu = get_cpu();
C
Christoph Hellwig 已提交
384 385 386 387
	if (!test_bit(QUEUE_FLAG_SAME_FORCE, &rq->q->queue_flags))
		shared = cpus_share_cache(cpu, ctx->cpu);

	if (cpu != ctx->cpu && !shared && cpu_online(ctx->cpu)) {
388
		rq->csd.func = __blk_mq_complete_request_remote;
389 390
		rq->csd.info = rq;
		rq->csd.flags = 0;
391
		smp_call_function_single_async(ctx->cpu, &rq->csd);
392
	} else {
393
		rq->q->softirq_done_fn(rq);
394
	}
395 396
	put_cpu();
}
397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413

/**
 * blk_mq_complete_request - end I/O on a request
 * @rq:		the request being processed
 *
 * Description:
 *	Ends all I/O on a request. It does not handle partial completions.
 *	The actual completion happens out-of-order, through a IPI handler.
 **/
void blk_mq_complete_request(struct request *rq)
{
	if (unlikely(blk_should_fake_timeout(rq->q)))
		return;
	if (!blk_mark_rq_complete(rq))
		__blk_mq_complete_request(rq);
}
EXPORT_SYMBOL(blk_mq_complete_request);
414

415
static void blk_mq_start_request(struct request *rq, bool last)
416 417 418 419 420
{
	struct request_queue *q = rq->q;

	trace_block_rq_issue(q, rq);

C
Christoph Hellwig 已提交
421
	rq->resid_len = blk_rq_bytes(rq);
C
Christoph Hellwig 已提交
422 423
	if (unlikely(blk_bidi_rq(rq)))
		rq->next_rq->resid_len = blk_rq_bytes(rq->next_rq);
C
Christoph Hellwig 已提交
424

425 426 427 428 429 430
	/*
	 * Just mark start time and set the started bit. Due to memory
	 * ordering, we know we'll see the correct deadline as long as
	 * REQ_ATOMIC_STARTED is seen.
	 */
	rq->deadline = jiffies + q->rq_timeout;
431 432 433 434 435 436 437

	/*
	 * Mark us as started and clear complete. Complete might have been
	 * set if requeue raced with timeout, which then marked it as
	 * complete. So be sure to clear complete again when we start
	 * the request, otherwise we'll ignore the completion event.
	 */
438
	set_bit(REQ_ATOM_STARTED, &rq->atomic_flags);
439
	clear_bit(REQ_ATOM_COMPLETE, &rq->atomic_flags);
440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458

	if (q->dma_drain_size && blk_rq_bytes(rq)) {
		/*
		 * Make sure space for the drain appears.  We know we can do
		 * this because max_hw_segments has been adjusted to be one
		 * fewer than the device can handle.
		 */
		rq->nr_phys_segments++;
	}

	/*
	 * Flag the last request in the series so that drivers know when IO
	 * should be kicked off, if they don't do it on a per-request basis.
	 *
	 * Note: the flag isn't the only condition drivers should do kick off.
	 * If drive is busy, the last request might not have the bit set.
	 */
	if (last)
		rq->cmd_flags |= REQ_END;
459 460
}

461
static void __blk_mq_requeue_request(struct request *rq)
462 463 464 465 466
{
	struct request_queue *q = rq->q;

	trace_block_rq_requeue(q, rq);
	clear_bit(REQ_ATOM_STARTED, &rq->atomic_flags);
467 468 469 470 471

	rq->cmd_flags &= ~REQ_END;

	if (q->dma_drain_size && blk_rq_bytes(rq))
		rq->nr_phys_segments--;
472 473
}

474 475 476 477 478 479 480 481 482 483
void blk_mq_requeue_request(struct request *rq)
{
	__blk_mq_requeue_request(rq);
	blk_clear_rq_complete(rq);

	BUG_ON(blk_queued_rq(rq));
	blk_mq_insert_request(rq, true, true, false);
}
EXPORT_SYMBOL(blk_mq_requeue_request);

484 485 486 487 488 489
struct request *blk_mq_tag_to_rq(struct blk_mq_tags *tags, unsigned int tag)
{
	return tags->rqs[tag];
}
EXPORT_SYMBOL(blk_mq_tag_to_rq);

490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510
struct blk_mq_timeout_data {
	struct blk_mq_hw_ctx *hctx;
	unsigned long *next;
	unsigned int *next_set;
};

static void blk_mq_timeout_check(void *__data, unsigned long *free_tags)
{
	struct blk_mq_timeout_data *data = __data;
	struct blk_mq_hw_ctx *hctx = data->hctx;
	unsigned int tag;

	 /* It may not be in flight yet (this is where
	 * the REQ_ATOMIC_STARTED flag comes in). The requests are
	 * statically allocated, so we know it's always safe to access the
	 * memory associated with a bit offset into ->rqs[].
	 */
	tag = 0;
	do {
		struct request *rq;

511 512
		tag = find_next_zero_bit(free_tags, hctx->tags->nr_tags, tag);
		if (tag >= hctx->tags->nr_tags)
513 514
			break;

515 516 517
		rq = blk_mq_tag_to_rq(hctx->tags, tag++);
		if (rq->q != hctx->queue)
			continue;
518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541
		if (!test_bit(REQ_ATOM_STARTED, &rq->atomic_flags))
			continue;

		blk_rq_check_expired(rq, data->next, data->next_set);
	} while (1);
}

static void blk_mq_hw_ctx_check_timeout(struct blk_mq_hw_ctx *hctx,
					unsigned long *next,
					unsigned int *next_set)
{
	struct blk_mq_timeout_data data = {
		.hctx		= hctx,
		.next		= next,
		.next_set	= next_set,
	};

	/*
	 * Ask the tagging code to iterate busy requests, so we can
	 * check them for timeout.
	 */
	blk_mq_tag_busy_iter(hctx->tags, blk_mq_timeout_check, &data);
}

542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563
static enum blk_eh_timer_return blk_mq_rq_timed_out(struct request *rq)
{
	struct request_queue *q = rq->q;

	/*
	 * We know that complete is set at this point. If STARTED isn't set
	 * anymore, then the request isn't active and the "timeout" should
	 * just be ignored. This can happen due to the bitflag ordering.
	 * Timeout first checks if STARTED is set, and if it is, assumes
	 * the request is active. But if we race with completion, then
	 * we both flags will get cleared. So check here again, and ignore
	 * a timeout event with a request that isn't active.
	 */
	if (!test_bit(REQ_ATOM_STARTED, &rq->atomic_flags))
		return BLK_EH_NOT_HANDLED;

	if (!q->mq_ops->timeout)
		return BLK_EH_RESET_TIMER;

	return q->mq_ops->timeout(rq);
}

564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630
static void blk_mq_rq_timer(unsigned long data)
{
	struct request_queue *q = (struct request_queue *) data;
	struct blk_mq_hw_ctx *hctx;
	unsigned long next = 0;
	int i, next_set = 0;

	queue_for_each_hw_ctx(q, hctx, i)
		blk_mq_hw_ctx_check_timeout(hctx, &next, &next_set);

	if (next_set)
		mod_timer(&q->timeout, round_jiffies_up(next));
}

/*
 * Reverse check our software queue for entries that we could potentially
 * merge with. Currently includes a hand-wavy stop count of 8, to not spend
 * too much time checking for merges.
 */
static bool blk_mq_attempt_merge(struct request_queue *q,
				 struct blk_mq_ctx *ctx, struct bio *bio)
{
	struct request *rq;
	int checked = 8;

	list_for_each_entry_reverse(rq, &ctx->rq_list, queuelist) {
		int el_ret;

		if (!checked--)
			break;

		if (!blk_rq_merge_ok(rq, bio))
			continue;

		el_ret = blk_try_merge(rq, bio);
		if (el_ret == ELEVATOR_BACK_MERGE) {
			if (bio_attempt_back_merge(q, rq, bio)) {
				ctx->rq_merged++;
				return true;
			}
			break;
		} else if (el_ret == ELEVATOR_FRONT_MERGE) {
			if (bio_attempt_front_merge(q, rq, bio)) {
				ctx->rq_merged++;
				return true;
			}
			break;
		}
	}

	return false;
}

/*
 * Run this hardware queue, pulling any software queues mapped to it in.
 * Note that this function currently has various problems around ordering
 * of IO. In particular, we'd like FIFO behaviour on handling existing
 * items on the hctx->dispatch list. Ignore that for now.
 */
static void __blk_mq_run_hw_queue(struct blk_mq_hw_ctx *hctx)
{
	struct request_queue *q = hctx->queue;
	struct blk_mq_ctx *ctx;
	struct request *rq;
	LIST_HEAD(rq_list);
	int bit, queued;

631
	WARN_ON(!cpumask_test_cpu(raw_smp_processor_id(), hctx->cpumask));
632

633
	if (unlikely(test_bit(BLK_MQ_S_STOPPED, &hctx->state)))
634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674
		return;

	hctx->run++;

	/*
	 * Touch any software queue that has pending entries.
	 */
	for_each_set_bit(bit, hctx->ctx_map, hctx->nr_ctx) {
		clear_bit(bit, hctx->ctx_map);
		ctx = hctx->ctxs[bit];

		spin_lock(&ctx->lock);
		list_splice_tail_init(&ctx->rq_list, &rq_list);
		spin_unlock(&ctx->lock);
	}

	/*
	 * If we have previous entries on our dispatch list, grab them
	 * and stuff them at the front for more fair dispatch.
	 */
	if (!list_empty_careful(&hctx->dispatch)) {
		spin_lock(&hctx->lock);
		if (!list_empty(&hctx->dispatch))
			list_splice_init(&hctx->dispatch, &rq_list);
		spin_unlock(&hctx->lock);
	}

	/*
	 * Delete and return all entries from our dispatch list
	 */
	queued = 0;

	/*
	 * Now process all the entries, sending them to the driver.
	 */
	while (!list_empty(&rq_list)) {
		int ret;

		rq = list_first_entry(&rq_list, struct request, queuelist);
		list_del_init(&rq->queuelist);

675
		blk_mq_start_request(rq, list_empty(&rq_list));
676 677 678 679 680 681 682 683

		ret = q->mq_ops->queue_rq(hctx, rq);
		switch (ret) {
		case BLK_MQ_RQ_QUEUE_OK:
			queued++;
			continue;
		case BLK_MQ_RQ_QUEUE_BUSY:
			list_add(&rq->queuelist, &rq_list);
684
			__blk_mq_requeue_request(rq);
685 686 687 688
			break;
		default:
			pr_err("blk-mq: bad return on queue: %d\n", ret);
		case BLK_MQ_RQ_QUEUE_ERROR:
689
			rq->errors = -EIO;
690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713
			blk_mq_end_io(rq, rq->errors);
			break;
		}

		if (ret == BLK_MQ_RQ_QUEUE_BUSY)
			break;
	}

	if (!queued)
		hctx->dispatched[0]++;
	else if (queued < (1 << (BLK_MQ_MAX_DISPATCH_ORDER - 1)))
		hctx->dispatched[ilog2(queued) + 1]++;

	/*
	 * Any items that need requeuing? Stuff them into hctx->dispatch,
	 * that is where we will continue on next queue run.
	 */
	if (!list_empty(&rq_list)) {
		spin_lock(&hctx->lock);
		list_splice(&rq_list, &hctx->dispatch);
		spin_unlock(&hctx->lock);
	}
}

714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737
/*
 * It'd be great if the workqueue API had a way to pass
 * in a mask and had some smarts for more clever placement.
 * For now we just round-robin here, switching for every
 * BLK_MQ_CPU_WORK_BATCH queued items.
 */
static int blk_mq_hctx_next_cpu(struct blk_mq_hw_ctx *hctx)
{
	int cpu = hctx->next_cpu;

	if (--hctx->next_cpu_batch <= 0) {
		int next_cpu;

		next_cpu = cpumask_next(hctx->next_cpu, hctx->cpumask);
		if (next_cpu >= nr_cpu_ids)
			next_cpu = cpumask_first(hctx->cpumask);

		hctx->next_cpu = next_cpu;
		hctx->next_cpu_batch = BLK_MQ_CPU_WORK_BATCH;
	}

	return cpu;
}

738 739
void blk_mq_run_hw_queue(struct blk_mq_hw_ctx *hctx, bool async)
{
740
	if (unlikely(test_bit(BLK_MQ_S_STOPPED, &hctx->state)))
741 742
		return;

743
	if (!async && cpumask_test_cpu(smp_processor_id(), hctx->cpumask))
744
		__blk_mq_run_hw_queue(hctx);
745
	else if (hctx->queue->nr_hw_queues == 1)
746
		kblockd_schedule_delayed_work(&hctx->run_work, 0);
747 748 749
	else {
		unsigned int cpu;

750
		cpu = blk_mq_hctx_next_cpu(hctx);
751
		kblockd_schedule_delayed_work_on(cpu, &hctx->run_work, 0);
752
	}
753 754 755 756 757 758 759 760 761 762
}

void blk_mq_run_queues(struct request_queue *q, bool async)
{
	struct blk_mq_hw_ctx *hctx;
	int i;

	queue_for_each_hw_ctx(q, hctx, i) {
		if ((!blk_mq_hctx_has_pending(hctx) &&
		    list_empty_careful(&hctx->dispatch)) ||
763
		    test_bit(BLK_MQ_S_STOPPED, &hctx->state))
764 765
			continue;

766
		preempt_disable();
767
		blk_mq_run_hw_queue(hctx, async);
768
		preempt_enable();
769 770 771 772 773 774
	}
}
EXPORT_SYMBOL(blk_mq_run_queues);

void blk_mq_stop_hw_queue(struct blk_mq_hw_ctx *hctx)
{
775 776
	cancel_delayed_work(&hctx->run_work);
	cancel_delayed_work(&hctx->delay_work);
777 778 779 780
	set_bit(BLK_MQ_S_STOPPED, &hctx->state);
}
EXPORT_SYMBOL(blk_mq_stop_hw_queue);

781 782 783 784 785 786 787 788 789 790
void blk_mq_stop_hw_queues(struct request_queue *q)
{
	struct blk_mq_hw_ctx *hctx;
	int i;

	queue_for_each_hw_ctx(q, hctx, i)
		blk_mq_stop_hw_queue(hctx);
}
EXPORT_SYMBOL(blk_mq_stop_hw_queues);

791 792 793
void blk_mq_start_hw_queue(struct blk_mq_hw_ctx *hctx)
{
	clear_bit(BLK_MQ_S_STOPPED, &hctx->state);
794 795

	preempt_disable();
796
	__blk_mq_run_hw_queue(hctx);
797
	preempt_enable();
798 799 800
}
EXPORT_SYMBOL(blk_mq_start_hw_queue);

801 802 803 804 805 806 807 808 809 810 811
void blk_mq_start_hw_queues(struct request_queue *q)
{
	struct blk_mq_hw_ctx *hctx;
	int i;

	queue_for_each_hw_ctx(q, hctx, i)
		blk_mq_start_hw_queue(hctx);
}
EXPORT_SYMBOL(blk_mq_start_hw_queues);


812
void blk_mq_start_stopped_hw_queues(struct request_queue *q, bool async)
813 814 815 816 817 818 819 820 821
{
	struct blk_mq_hw_ctx *hctx;
	int i;

	queue_for_each_hw_ctx(q, hctx, i) {
		if (!test_bit(BLK_MQ_S_STOPPED, &hctx->state))
			continue;

		clear_bit(BLK_MQ_S_STOPPED, &hctx->state);
822
		preempt_disable();
823
		blk_mq_run_hw_queue(hctx, async);
824
		preempt_enable();
825 826 827 828
	}
}
EXPORT_SYMBOL(blk_mq_start_stopped_hw_queues);

829
static void blk_mq_run_work_fn(struct work_struct *work)
830 831 832
{
	struct blk_mq_hw_ctx *hctx;

833
	hctx = container_of(work, struct blk_mq_hw_ctx, run_work.work);
834

835 836 837
	__blk_mq_run_hw_queue(hctx);
}

838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856
static void blk_mq_delay_work_fn(struct work_struct *work)
{
	struct blk_mq_hw_ctx *hctx;

	hctx = container_of(work, struct blk_mq_hw_ctx, delay_work.work);

	if (test_and_clear_bit(BLK_MQ_S_STOPPED, &hctx->state))
		__blk_mq_run_hw_queue(hctx);
}

void blk_mq_delay_queue(struct blk_mq_hw_ctx *hctx, unsigned long msecs)
{
	unsigned long tmo = msecs_to_jiffies(msecs);

	if (hctx->queue->nr_hw_queues == 1)
		kblockd_schedule_delayed_work(&hctx->delay_work, tmo);
	else {
		unsigned int cpu;

857
		cpu = blk_mq_hctx_next_cpu(hctx);
858 859 860 861 862
		kblockd_schedule_delayed_work_on(cpu, &hctx->delay_work, tmo);
	}
}
EXPORT_SYMBOL(blk_mq_delay_queue);

863
static void __blk_mq_insert_request(struct blk_mq_hw_ctx *hctx,
864
				    struct request *rq, bool at_head)
865 866 867
{
	struct blk_mq_ctx *ctx = rq->mq_ctx;

868 869
	trace_block_rq_insert(hctx->queue, rq);

870 871 872 873
	if (at_head)
		list_add(&rq->queuelist, &ctx->rq_list);
	else
		list_add_tail(&rq->queuelist, &ctx->rq_list);
874

875 876 877 878 879
	blk_mq_hctx_mark_pending(hctx, ctx);

	/*
	 * We do this early, to ensure we are on the right CPU.
	 */
880
	blk_add_timer(rq);
881 882
}

883 884
void blk_mq_insert_request(struct request *rq, bool at_head, bool run_queue,
		bool async)
885
{
886
	struct request_queue *q = rq->q;
887
	struct blk_mq_hw_ctx *hctx;
888 889 890 891 892
	struct blk_mq_ctx *ctx = rq->mq_ctx, *current_ctx;

	current_ctx = blk_mq_get_ctx(q);
	if (!cpu_online(ctx->cpu))
		rq->mq_ctx = ctx = current_ctx;
893 894 895

	hctx = q->mq_ops->map_queue(q, ctx->cpu);

896 897
	if (rq->cmd_flags & (REQ_FLUSH | REQ_FUA) &&
	    !(rq->cmd_flags & (REQ_FLUSH_SEQ))) {
898 899 900
		blk_insert_flush(rq);
	} else {
		spin_lock(&ctx->lock);
901
		__blk_mq_insert_request(hctx, rq, at_head);
902 903 904 905 906
		spin_unlock(&ctx->lock);
	}

	if (run_queue)
		blk_mq_run_hw_queue(hctx, async);
907 908

	blk_mq_put_ctx(current_ctx);
909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939
}

static void blk_mq_insert_requests(struct request_queue *q,
				     struct blk_mq_ctx *ctx,
				     struct list_head *list,
				     int depth,
				     bool from_schedule)

{
	struct blk_mq_hw_ctx *hctx;
	struct blk_mq_ctx *current_ctx;

	trace_block_unplug(q, depth, !from_schedule);

	current_ctx = blk_mq_get_ctx(q);

	if (!cpu_online(ctx->cpu))
		ctx = current_ctx;
	hctx = q->mq_ops->map_queue(q, ctx->cpu);

	/*
	 * preemption doesn't flush plug list, so it's possible ctx->cpu is
	 * offline now
	 */
	spin_lock(&ctx->lock);
	while (!list_empty(list)) {
		struct request *rq;

		rq = list_first_entry(list, struct request, queuelist);
		list_del_init(&rq->queuelist);
		rq->mq_ctx = ctx;
940
		__blk_mq_insert_request(hctx, rq, false);
941 942 943 944
	}
	spin_unlock(&ctx->lock);

	blk_mq_run_hw_queue(hctx, from_schedule);
945
	blk_mq_put_ctx(current_ctx);
946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028
}

static int plug_ctx_cmp(void *priv, struct list_head *a, struct list_head *b)
{
	struct request *rqa = container_of(a, struct request, queuelist);
	struct request *rqb = container_of(b, struct request, queuelist);

	return !(rqa->mq_ctx < rqb->mq_ctx ||
		 (rqa->mq_ctx == rqb->mq_ctx &&
		  blk_rq_pos(rqa) < blk_rq_pos(rqb)));
}

void blk_mq_flush_plug_list(struct blk_plug *plug, bool from_schedule)
{
	struct blk_mq_ctx *this_ctx;
	struct request_queue *this_q;
	struct request *rq;
	LIST_HEAD(list);
	LIST_HEAD(ctx_list);
	unsigned int depth;

	list_splice_init(&plug->mq_list, &list);

	list_sort(NULL, &list, plug_ctx_cmp);

	this_q = NULL;
	this_ctx = NULL;
	depth = 0;

	while (!list_empty(&list)) {
		rq = list_entry_rq(list.next);
		list_del_init(&rq->queuelist);
		BUG_ON(!rq->q);
		if (rq->mq_ctx != this_ctx) {
			if (this_ctx) {
				blk_mq_insert_requests(this_q, this_ctx,
							&ctx_list, depth,
							from_schedule);
			}

			this_ctx = rq->mq_ctx;
			this_q = rq->q;
			depth = 0;
		}

		depth++;
		list_add_tail(&rq->queuelist, &ctx_list);
	}

	/*
	 * If 'this_ctx' is set, we know we have entries to complete
	 * on 'ctx_list'. Do those.
	 */
	if (this_ctx) {
		blk_mq_insert_requests(this_q, this_ctx, &ctx_list, depth,
				       from_schedule);
	}
}

static void blk_mq_bio_to_request(struct request *rq, struct bio *bio)
{
	init_request_from_bio(rq, bio);
	blk_account_io_start(rq, 1);
}

static void blk_mq_make_request(struct request_queue *q, struct bio *bio)
{
	struct blk_mq_hw_ctx *hctx;
	struct blk_mq_ctx *ctx;
	const int is_sync = rw_is_sync(bio->bi_rw);
	const int is_flush_fua = bio->bi_rw & (REQ_FLUSH | REQ_FUA);
	int rw = bio_data_dir(bio);
	struct request *rq;
	unsigned int use_plug, request_count = 0;

	/*
	 * If we have multiple hardware queues, just go directly to
	 * one of those for sync IO.
	 */
	use_plug = !is_flush_fua && ((q->nr_hw_queues == 1) || !is_sync);

	blk_queue_bounce(q, &bio);

1029 1030 1031 1032 1033
	if (bio_integrity_enabled(bio) && bio_integrity_prep(bio)) {
		bio_endio(bio, -EIO);
		return;
	}

1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044
	if (use_plug && blk_attempt_plug_merge(q, bio, &request_count))
		return;

	if (blk_mq_queue_enter(q)) {
		bio_endio(bio, -EIO);
		return;
	}

	ctx = blk_mq_get_ctx(q);
	hctx = q->mq_ops->map_queue(q, ctx->cpu);

S
Shaohua Li 已提交
1045 1046
	if (is_sync)
		rw |= REQ_SYNC;
1047
	trace_block_getrq(q, bio, rw);
1048
	rq = __blk_mq_alloc_request(hctx, ctx, GFP_ATOMIC, false);
1049
	if (likely(rq))
1050
		blk_mq_rq_ctx_init(q, ctx, rq, rw);
1051 1052 1053
	else {
		blk_mq_put_ctx(ctx);
		trace_block_sleeprq(q, bio, rw);
1054 1055
		rq = blk_mq_alloc_request_pinned(q, rw, __GFP_WAIT|GFP_ATOMIC,
							false);
1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077
		ctx = rq->mq_ctx;
		hctx = q->mq_ops->map_queue(q, ctx->cpu);
	}

	hctx->queued++;

	if (unlikely(is_flush_fua)) {
		blk_mq_bio_to_request(rq, bio);
		blk_insert_flush(rq);
		goto run_queue;
	}

	/*
	 * A task plug currently exists. Since this is completely lockless,
	 * utilize that to temporarily store requests until the task is
	 * either done or scheduled away.
	 */
	if (use_plug) {
		struct blk_plug *plug = current->plug;

		if (plug) {
			blk_mq_bio_to_request(rq, bio);
S
Shaohua Li 已提交
1078
			if (list_empty(&plug->mq_list))
1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089
				trace_block_plug(q);
			else if (request_count >= BLK_MAX_REQUEST_COUNT) {
				blk_flush_plug_list(plug, false);
				trace_block_plug(q);
			}
			list_add_tail(&rq->queuelist, &plug->mq_list);
			blk_mq_put_ctx(ctx);
			return;
		}
	}

1090 1091
	if (!(hctx->flags & BLK_MQ_F_SHOULD_MERGE)) {
		init_request_from_bio(rq, bio);
1092

1093 1094
		spin_lock(&ctx->lock);
insert_rq:
1095
		__blk_mq_insert_request(hctx, rq, false);
1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106
		spin_unlock(&ctx->lock);
		blk_account_io_start(rq, 1);
	} else {
		spin_lock(&ctx->lock);
		if (!blk_mq_attempt_merge(q, ctx, bio)) {
			init_request_from_bio(rq, bio);
			goto insert_rq;
		}

		spin_unlock(&ctx->lock);
		__blk_mq_free_request(hctx, ctx, rq);
1107 1108 1109 1110 1111 1112 1113 1114 1115 1116
	}


	/*
	 * For a SYNC request, send it to the hardware immediately. For an
	 * ASYNC request, just ensure that we run it later on. The latter
	 * allows for merging opportunities and more efficient dispatching.
	 */
run_queue:
	blk_mq_run_hw_queue(hctx, !is_sync || is_flush_fua);
1117
	blk_mq_put_ctx(ctx);
1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128
}

/*
 * Default mapping to a software queue, since we use one per CPU.
 */
struct blk_mq_hw_ctx *blk_mq_map_queue(struct request_queue *q, const int cpu)
{
	return q->queue_hw_ctx[q->mq_map[cpu]];
}
EXPORT_SYMBOL(blk_mq_map_queue);

1129
struct blk_mq_hw_ctx *blk_mq_alloc_single_hw_queue(struct blk_mq_tag_set *set,
1130 1131
						   unsigned int hctx_index)
{
1132 1133
	return kzalloc_node(sizeof(struct blk_mq_hw_ctx), GFP_KERNEL,
				set->numa_node);
1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147
}
EXPORT_SYMBOL(blk_mq_alloc_single_hw_queue);

void blk_mq_free_single_hw_queue(struct blk_mq_hw_ctx *hctx,
				 unsigned int hctx_index)
{
	kfree(hctx);
}
EXPORT_SYMBOL(blk_mq_free_single_hw_queue);

static void blk_mq_hctx_notify(void *data, unsigned long action,
			       unsigned int cpu)
{
	struct blk_mq_hw_ctx *hctx = data;
1148
	struct request_queue *q = hctx->queue;
1149 1150 1151 1152 1153 1154 1155 1156 1157
	struct blk_mq_ctx *ctx;
	LIST_HEAD(tmp);

	if (action != CPU_DEAD && action != CPU_DEAD_FROZEN)
		return;

	/*
	 * Move ctx entries to new CPU, if this one is going away.
	 */
1158
	ctx = __blk_mq_get_ctx(q, cpu);
1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169

	spin_lock(&ctx->lock);
	if (!list_empty(&ctx->rq_list)) {
		list_splice_init(&ctx->rq_list, &tmp);
		clear_bit(ctx->index_hw, hctx->ctx_map);
	}
	spin_unlock(&ctx->lock);

	if (list_empty(&tmp))
		return;

1170
	ctx = blk_mq_get_ctx(q);
1171 1172 1173 1174 1175 1176 1177 1178 1179 1180
	spin_lock(&ctx->lock);

	while (!list_empty(&tmp)) {
		struct request *rq;

		rq = list_first_entry(&tmp, struct request, queuelist);
		rq->mq_ctx = ctx;
		list_move_tail(&rq->queuelist, &ctx->rq_list);
	}

1181
	hctx = q->mq_ops->map_queue(q, ctx->cpu);
1182 1183 1184
	blk_mq_hctx_mark_pending(hctx, ctx);

	spin_unlock(&ctx->lock);
1185 1186

	blk_mq_run_hw_queue(hctx, true);
1187
	blk_mq_put_ctx(ctx);
1188 1189
}

1190 1191
static void blk_mq_free_rq_map(struct blk_mq_tag_set *set,
		struct blk_mq_tags *tags, unsigned int hctx_idx)
1192
{
1193
	struct page *page;
1194

1195
	if (tags->rqs && set->ops->exit_request) {
1196
		int i;
1197

1198 1199
		for (i = 0; i < tags->nr_tags; i++) {
			if (!tags->rqs[i])
1200
				continue;
1201 1202
			set->ops->exit_request(set->driver_data, tags->rqs[i],
						hctx_idx, i);
1203
		}
1204 1205
	}

1206 1207
	while (!list_empty(&tags->page_list)) {
		page = list_first_entry(&tags->page_list, struct page, lru);
1208
		list_del_init(&page->lru);
1209 1210 1211
		__free_pages(page, page->private);
	}

1212
	kfree(tags->rqs);
1213

1214
	blk_mq_free_tags(tags);
1215 1216 1217 1218
}

static size_t order_to_size(unsigned int order)
{
1219
	return (size_t)PAGE_SIZE << order;
1220 1221
}

1222 1223
static struct blk_mq_tags *blk_mq_init_rq_map(struct blk_mq_tag_set *set,
		unsigned int hctx_idx)
1224
{
1225
	struct blk_mq_tags *tags;
1226 1227 1228
	unsigned int i, j, entries_per_page, max_order = 4;
	size_t rq_size, left;

1229 1230 1231 1232
	tags = blk_mq_init_tags(set->queue_depth, set->reserved_tags,
				set->numa_node);
	if (!tags)
		return NULL;
1233

1234 1235 1236 1237 1238 1239 1240 1241
	INIT_LIST_HEAD(&tags->page_list);

	tags->rqs = kmalloc_node(set->queue_depth * sizeof(struct request *),
					GFP_KERNEL, set->numa_node);
	if (!tags->rqs) {
		blk_mq_free_tags(tags);
		return NULL;
	}
1242 1243 1244 1245 1246

	/*
	 * rq_size is the size of the request plus driver payload, rounded
	 * to the cacheline size
	 */
1247
	rq_size = round_up(sizeof(struct request) + set->cmd_size,
1248
				cache_line_size());
1249
	left = rq_size * set->queue_depth;
1250

1251
	for (i = 0; i < set->queue_depth; ) {
1252 1253 1254 1255 1256 1257 1258 1259 1260
		int this_order = max_order;
		struct page *page;
		int to_do;
		void *p;

		while (left < order_to_size(this_order - 1) && this_order)
			this_order--;

		do {
1261 1262
			page = alloc_pages_node(set->numa_node, GFP_KERNEL,
						this_order);
1263 1264 1265 1266 1267 1268 1269 1270 1271
			if (page)
				break;
			if (!this_order--)
				break;
			if (order_to_size(this_order) < rq_size)
				break;
		} while (1);

		if (!page)
1272
			goto fail;
1273 1274

		page->private = this_order;
1275
		list_add_tail(&page->lru, &tags->page_list);
1276 1277 1278

		p = page_address(page);
		entries_per_page = order_to_size(this_order) / rq_size;
1279
		to_do = min(entries_per_page, set->queue_depth - i);
1280 1281
		left -= to_do * rq_size;
		for (j = 0; j < to_do; j++) {
1282 1283 1284 1285 1286 1287
			tags->rqs[i] = p;
			if (set->ops->init_request) {
				if (set->ops->init_request(set->driver_data,
						tags->rqs[i], hctx_idx, i,
						set->numa_node))
					goto fail;
1288 1289
			}

1290 1291 1292 1293 1294
			p += rq_size;
			i++;
		}
	}

1295
	return tags;
1296

1297 1298 1299 1300
fail:
	pr_warn("%s: failed to allocate requests\n", __func__);
	blk_mq_free_rq_map(set, tags, hctx_idx);
	return NULL;
1301 1302 1303
}

static int blk_mq_init_hw_queues(struct request_queue *q,
1304
		struct blk_mq_tag_set *set)
1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317
{
	struct blk_mq_hw_ctx *hctx;
	unsigned int i, j;

	/*
	 * Initialize hardware queues
	 */
	queue_for_each_hw_ctx(q, hctx, i) {
		unsigned int num_maps;
		int node;

		node = hctx->numa_node;
		if (node == NUMA_NO_NODE)
1318
			node = hctx->numa_node = set->numa_node;
1319

1320 1321
		INIT_DELAYED_WORK(&hctx->run_work, blk_mq_run_work_fn);
		INIT_DELAYED_WORK(&hctx->delay_work, blk_mq_delay_work_fn);
1322 1323 1324 1325
		spin_lock_init(&hctx->lock);
		INIT_LIST_HEAD(&hctx->dispatch);
		hctx->queue = q;
		hctx->queue_num = i;
1326 1327
		hctx->flags = set->flags;
		hctx->cmd_size = set->cmd_size;
1328 1329 1330 1331 1332

		blk_mq_init_cpu_notifier(&hctx->cpu_notifier,
						blk_mq_hctx_notify, hctx);
		blk_mq_register_cpu_notifier(&hctx->cpu_notifier);

1333
		hctx->tags = set->tags[i];
1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352

		/*
		 * Allocate space for all possible cpus to avoid allocation in
		 * runtime
		 */
		hctx->ctxs = kmalloc_node(nr_cpu_ids * sizeof(void *),
						GFP_KERNEL, node);
		if (!hctx->ctxs)
			break;

		num_maps = ALIGN(nr_cpu_ids, BITS_PER_LONG) / BITS_PER_LONG;
		hctx->ctx_map = kzalloc_node(num_maps * sizeof(unsigned long),
						GFP_KERNEL, node);
		if (!hctx->ctx_map)
			break;

		hctx->nr_ctx_map = num_maps;
		hctx->nr_ctx = 0;

1353 1354
		if (set->ops->init_hctx &&
		    set->ops->init_hctx(hctx, set->driver_data, i))
1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367
			break;
	}

	if (i == q->nr_hw_queues)
		return 0;

	/*
	 * Init failed
	 */
	queue_for_each_hw_ctx(q, hctx, j) {
		if (i == j)
			break;

1368 1369
		if (set->ops->exit_hctx)
			set->ops->exit_hctx(hctx, j);
1370 1371 1372

		blk_mq_unregister_cpu_notifier(&hctx->cpu_notifier);
		kfree(hctx->ctxs);
1373
		kfree(hctx->ctx_map);
1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397
	}

	return 1;
}

static void blk_mq_init_cpu_queues(struct request_queue *q,
				   unsigned int nr_hw_queues)
{
	unsigned int i;

	for_each_possible_cpu(i) {
		struct blk_mq_ctx *__ctx = per_cpu_ptr(q->queue_ctx, i);
		struct blk_mq_hw_ctx *hctx;

		memset(__ctx, 0, sizeof(*__ctx));
		__ctx->cpu = i;
		spin_lock_init(&__ctx->lock);
		INIT_LIST_HEAD(&__ctx->rq_list);
		__ctx->queue = q;

		/* If the cpu isn't online, the cpu is mapped to first hctx */
		if (!cpu_online(i))
			continue;

1398 1399 1400 1401
		hctx = q->mq_ops->map_queue(q, i);
		cpumask_set_cpu(i, hctx->cpumask);
		hctx->nr_ctx++;

1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417
		/*
		 * Set local node, IFF we have more than one hw queue. If
		 * not, we remain on the home node of the device
		 */
		if (nr_hw_queues > 1 && hctx->numa_node == NUMA_NO_NODE)
			hctx->numa_node = cpu_to_node(i);
	}
}

static void blk_mq_map_swqueue(struct request_queue *q)
{
	unsigned int i;
	struct blk_mq_hw_ctx *hctx;
	struct blk_mq_ctx *ctx;

	queue_for_each_hw_ctx(q, hctx, i) {
1418
		cpumask_clear(hctx->cpumask);
1419 1420 1421 1422 1423 1424 1425 1426
		hctx->nr_ctx = 0;
	}

	/*
	 * Map software to hardware queues
	 */
	queue_for_each_ctx(q, ctx, i) {
		/* If the cpu isn't online, the cpu is mapped to first hctx */
1427 1428 1429
		if (!cpu_online(i))
			continue;

1430
		hctx = q->mq_ops->map_queue(q, i);
1431
		cpumask_set_cpu(i, hctx->cpumask);
1432 1433 1434
		ctx->index_hw = hctx->nr_ctx;
		hctx->ctxs[hctx->nr_ctx++] = ctx;
	}
1435 1436 1437 1438 1439

	queue_for_each_hw_ctx(q, hctx, i) {
		hctx->next_cpu = cpumask_first(hctx->cpumask);
		hctx->next_cpu_batch = BLK_MQ_CPU_WORK_BATCH;
	}
1440 1441
}

1442
struct request_queue *blk_mq_init_queue(struct blk_mq_tag_set *set)
1443 1444 1445 1446 1447 1448 1449 1450 1451 1452
{
	struct blk_mq_hw_ctx **hctxs;
	struct blk_mq_ctx *ctx;
	struct request_queue *q;
	int i;

	ctx = alloc_percpu(struct blk_mq_ctx);
	if (!ctx)
		return ERR_PTR(-ENOMEM);

1453 1454
	hctxs = kmalloc_node(set->nr_hw_queues * sizeof(*hctxs), GFP_KERNEL,
			set->numa_node);
1455 1456 1457 1458

	if (!hctxs)
		goto err_percpu;

1459 1460
	for (i = 0; i < set->nr_hw_queues; i++) {
		hctxs[i] = set->ops->alloc_hctx(set, i);
1461 1462 1463
		if (!hctxs[i])
			goto err_hctxs;

1464 1465 1466
		if (!zalloc_cpumask_var(&hctxs[i]->cpumask, GFP_KERNEL))
			goto err_hctxs;

1467 1468 1469 1470
		hctxs[i]->numa_node = NUMA_NO_NODE;
		hctxs[i]->queue_num = i;
	}

1471
	q = blk_alloc_queue_node(GFP_KERNEL, set->numa_node);
1472 1473 1474
	if (!q)
		goto err_hctxs;

1475
	q->mq_map = blk_mq_make_queue_map(set);
1476 1477 1478 1479 1480 1481 1482
	if (!q->mq_map)
		goto err_map;

	setup_timer(&q->timeout, blk_mq_rq_timer, (unsigned long) q);
	blk_queue_rq_timeout(q, 30000);

	q->nr_queues = nr_cpu_ids;
1483
	q->nr_hw_queues = set->nr_hw_queues;
1484 1485 1486 1487

	q->queue_ctx = ctx;
	q->queue_hw_ctx = hctxs;

1488
	q->mq_ops = set->ops;
1489
	q->queue_flags |= QUEUE_FLAG_MQ_DEFAULT;
1490

1491 1492
	q->sg_reserved_size = INT_MAX;

1493
	blk_queue_make_request(q, blk_mq_make_request);
1494
	blk_queue_rq_timed_out(q, blk_mq_rq_timed_out);
1495 1496
	if (set->timeout)
		blk_queue_rq_timeout(q, set->timeout);
1497

1498 1499
	if (set->ops->complete)
		blk_queue_softirq_done(q, set->ops->complete);
1500

1501
	blk_mq_init_flush(q);
1502
	blk_mq_init_cpu_queues(q, set->nr_hw_queues);
1503

1504 1505 1506
	q->flush_rq = kzalloc(round_up(sizeof(struct request) +
				set->cmd_size, cache_line_size()),
				GFP_KERNEL);
1507
	if (!q->flush_rq)
1508 1509
		goto err_hw;

1510
	if (blk_mq_init_hw_queues(q, set))
1511 1512
		goto err_flush_rq;

1513 1514 1515 1516 1517 1518 1519
	blk_mq_map_swqueue(q);

	mutex_lock(&all_q_mutex);
	list_add_tail(&q->all_q_node, &all_q_list);
	mutex_unlock(&all_q_mutex);

	return q;
1520 1521 1522

err_flush_rq:
	kfree(q->flush_rq);
1523 1524 1525 1526 1527
err_hw:
	kfree(q->mq_map);
err_map:
	blk_cleanup_queue(q);
err_hctxs:
1528
	for (i = 0; i < set->nr_hw_queues; i++) {
1529 1530
		if (!hctxs[i])
			break;
1531
		free_cpumask_var(hctxs[i]->cpumask);
1532
		set->ops->free_hctx(hctxs[i], i);
1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551
	}
	kfree(hctxs);
err_percpu:
	free_percpu(ctx);
	return ERR_PTR(-ENOMEM);
}
EXPORT_SYMBOL(blk_mq_init_queue);

void blk_mq_free_queue(struct request_queue *q)
{
	struct blk_mq_hw_ctx *hctx;
	int i;

	queue_for_each_hw_ctx(q, hctx, i) {
		kfree(hctx->ctx_map);
		kfree(hctx->ctxs);
		blk_mq_unregister_cpu_notifier(&hctx->cpu_notifier);
		if (q->mq_ops->exit_hctx)
			q->mq_ops->exit_hctx(hctx, i);
1552
		free_cpumask_var(hctx->cpumask);
1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569
		q->mq_ops->free_hctx(hctx, i);
	}

	free_percpu(q->queue_ctx);
	kfree(q->queue_hw_ctx);
	kfree(q->mq_map);

	q->queue_ctx = NULL;
	q->queue_hw_ctx = NULL;
	q->mq_map = NULL;

	mutex_lock(&all_q_mutex);
	list_del_init(&q->all_q_node);
	mutex_unlock(&all_q_mutex);
}

/* Basically redo blk_mq_init_queue with queue frozen */
1570
static void blk_mq_queue_reinit(struct request_queue *q)
1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586
{
	blk_mq_freeze_queue(q);

	blk_mq_update_queue_map(q->mq_map, q->nr_hw_queues);

	/*
	 * redo blk_mq_init_cpu_queues and blk_mq_init_hw_queues. FIXME: maybe
	 * we should change hctx numa_node according to new topology (this
	 * involves free and re-allocate memory, worthy doing?)
	 */

	blk_mq_map_swqueue(q);

	blk_mq_unfreeze_queue(q);
}

1587 1588
static int blk_mq_queue_reinit_notify(struct notifier_block *nb,
				      unsigned long action, void *hcpu)
1589 1590 1591 1592
{
	struct request_queue *q;

	/*
1593 1594 1595 1596
	 * Before new mappings are established, hotadded cpu might already
	 * start handling requests. This doesn't break anything as we map
	 * offline CPUs to first hardware queue. We will re-init the queue
	 * below to get optimal settings.
1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608
	 */
	if (action != CPU_DEAD && action != CPU_DEAD_FROZEN &&
	    action != CPU_ONLINE && action != CPU_ONLINE_FROZEN)
		return NOTIFY_OK;

	mutex_lock(&all_q_mutex);
	list_for_each_entry(q, &all_q_list, all_q_node)
		blk_mq_queue_reinit(q);
	mutex_unlock(&all_q_mutex);
	return NOTIFY_OK;
}

1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625
int blk_mq_alloc_tag_set(struct blk_mq_tag_set *set)
{
	int i;

	if (!set->nr_hw_queues)
		return -EINVAL;
	if (!set->queue_depth || set->queue_depth > BLK_MQ_MAX_DEPTH)
		return -EINVAL;
	if (set->queue_depth < set->reserved_tags + BLK_MQ_TAG_MIN)
		return -EINVAL;

	if (!set->nr_hw_queues ||
	    !set->ops->queue_rq || !set->ops->map_queue ||
	    !set->ops->alloc_hctx || !set->ops->free_hctx)
		return -EINVAL;


M
Ming Lei 已提交
1626 1627
	set->tags = kmalloc_node(set->nr_hw_queues *
				 sizeof(struct blk_mq_tags *),
1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653
				 GFP_KERNEL, set->numa_node);
	if (!set->tags)
		goto out;

	for (i = 0; i < set->nr_hw_queues; i++) {
		set->tags[i] = blk_mq_init_rq_map(set, i);
		if (!set->tags[i])
			goto out_unwind;
	}

	return 0;

out_unwind:
	while (--i >= 0)
		blk_mq_free_rq_map(set, set->tags[i], i);
out:
	return -ENOMEM;
}
EXPORT_SYMBOL(blk_mq_alloc_tag_set);

void blk_mq_free_tag_set(struct blk_mq_tag_set *set)
{
	int i;

	for (i = 0; i < set->nr_hw_queues; i++)
		blk_mq_free_rq_map(set, set->tags[i], i);
M
Ming Lei 已提交
1654
	kfree(set->tags);
1655 1656 1657
}
EXPORT_SYMBOL(blk_mq_free_tag_set);

1658 1659 1660 1661 1662 1663 1664 1665 1666 1667
void blk_mq_disable_hotplug(void)
{
	mutex_lock(&all_q_mutex);
}

void blk_mq_enable_hotplug(void)
{
	mutex_unlock(&all_q_mutex);
}

1668 1669 1670 1671 1672 1673 1674 1675 1676 1677
static int __init blk_mq_init(void)
{
	blk_mq_cpu_init();

	/* Must be called after percpu_counter_hotcpu_callback() */
	hotcpu_notifier(blk_mq_queue_reinit_notify, -10);

	return 0;
}
subsys_initcall(blk_mq_init);