blk-merge.c 19.7 KB
Newer Older
1 2 3 4 5 6 7 8 9
/*
 * Functions related to segment and merge handling
 */
#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/bio.h>
#include <linux/blkdev.h>
#include <linux/scatterlist.h>

10 11
#include <trace/events/block.h>

12 13
#include "blk.h"

14 15
static struct bio *blk_bio_discard_split(struct request_queue *q,
					 struct bio *bio,
16 17
					 struct bio_set *bs,
					 unsigned *nsegs)
18 19 20 21 22 23
{
	unsigned int max_discard_sectors, granularity;
	int alignment;
	sector_t tmp;
	unsigned split_sectors;

24 25
	*nsegs = 1;

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
	/* Zero-sector (unknown) and one-sector granularities are the same.  */
	granularity = max(q->limits.discard_granularity >> 9, 1U);

	max_discard_sectors = min(q->limits.max_discard_sectors, UINT_MAX >> 9);
	max_discard_sectors -= max_discard_sectors % granularity;

	if (unlikely(!max_discard_sectors)) {
		/* XXX: warn */
		return NULL;
	}

	if (bio_sectors(bio) <= max_discard_sectors)
		return NULL;

	split_sectors = max_discard_sectors;

	/*
	 * If the next starting sector would be misaligned, stop the discard at
	 * the previous aligned sector.
	 */
	alignment = (q->limits.discard_alignment >> 9) % granularity;

	tmp = bio->bi_iter.bi_sector + split_sectors - alignment;
	tmp = sector_div(tmp, granularity);

	if (split_sectors > tmp)
		split_sectors -= tmp;

	return bio_split(bio, split_sectors, GFP_NOIO, bs);
}

static struct bio *blk_bio_write_same_split(struct request_queue *q,
					    struct bio *bio,
59 60
					    struct bio_set *bs,
					    unsigned *nsegs)
61
{
62 63
	*nsegs = 1;

64 65 66 67 68 69 70 71 72
	if (!q->limits.max_write_same_sectors)
		return NULL;

	if (bio_sectors(bio) <= q->limits.max_write_same_sectors)
		return NULL;

	return bio_split(bio, q->limits.max_write_same_sectors, GFP_NOIO, bs);
}

73 74 75 76 77 78 79 80 81 82 83 84
static inline unsigned get_max_io_size(struct request_queue *q,
				       struct bio *bio)
{
	unsigned sectors = blk_max_size_offset(q, bio->bi_iter.bi_sector);
	unsigned mask = queue_logical_block_size(q) - 1;

	/* aligned to logical block size */
	sectors &= ~(mask >> 9);

	return sectors;
}

85 86
static struct bio *blk_bio_segment_split(struct request_queue *q,
					 struct bio *bio,
87 88
					 struct bio_set *bs,
					 unsigned *segs)
89
{
90
	struct bio_vec bv, bvprv, *bvprvp = NULL;
91
	struct bvec_iter iter;
92
	unsigned seg_size = 0, nsegs = 0, sectors = 0;
93 94 95
	unsigned front_seg_size = bio->bi_seg_front_size;
	bool do_split = true;
	struct bio *new = NULL;
96
	const unsigned max_sectors = get_max_io_size(q, bio);
97
	unsigned bvecs = 0;
98 99

	bio_for_each_segment(bv, bio, iter) {
100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120
		/*
		 * With arbitrary bio size, the incoming bio may be very
		 * big. We have to split the bio into small bios so that
		 * each holds at most BIO_MAX_PAGES bvecs because
		 * bio_clone() can fail to allocate big bvecs.
		 *
		 * It should have been better to apply the limit per
		 * request queue in which bio_clone() is involved,
		 * instead of globally. The biggest blocker is the
		 * bio_clone() in bio bounce.
		 *
		 * If bio is splitted by this reason, we should have
		 * allowed to continue bios merging, but don't do
		 * that now for making the change simple.
		 *
		 * TODO: deal with bio bounce's bio_clone() gracefully
		 * and convert the global limit into per-queue limit.
		 */
		if (bvecs++ >= BIO_MAX_PAGES)
			goto split;

121 122 123 124
		/*
		 * If the queue doesn't support SG gaps and adding this
		 * offset would create a gap, disallow it.
		 */
125
		if (bvprvp && bvec_gap_to_prev(q, bvprvp, bv.bv_offset))
126 127
			goto split;

128
		if (sectors + (bv.bv_len >> 9) > max_sectors) {
129 130 131 132 133
			/*
			 * Consider this a new segment if we're splitting in
			 * the middle of this vector.
			 */
			if (nsegs < queue_max_segments(q) &&
134
			    sectors < max_sectors) {
135
				nsegs++;
136
				sectors = max_sectors;
137
			}
138 139 140
			if (sectors)
				goto split;
			/* Make this single bvec as the 1st segment */
141 142
		}

143
		if (bvprvp && blk_queue_cluster(q)) {
144 145
			if (seg_size + bv.bv_len > queue_max_segment_size(q))
				goto new_segment;
146
			if (!BIOVEC_PHYS_MERGEABLE(bvprvp, &bv))
147
				goto new_segment;
148
			if (!BIOVEC_SEG_BOUNDARY(q, bvprvp, &bv))
149 150 151 152
				goto new_segment;

			seg_size += bv.bv_len;
			bvprv = bv;
Ming Lei's avatar
Ming Lei committed
153
			bvprvp = &bvprv;
154
			sectors += bv.bv_len >> 9;
155 156 157

			if (nsegs == 1 && seg_size > front_seg_size)
				front_seg_size = seg_size;
158 159 160 161 162 163 164 165
			continue;
		}
new_segment:
		if (nsegs == queue_max_segments(q))
			goto split;

		nsegs++;
		bvprv = bv;
Ming Lei's avatar
Ming Lei committed
166
		bvprvp = &bvprv;
167
		seg_size = bv.bv_len;
168
		sectors += bv.bv_len >> 9;
169 170 171

		if (nsegs == 1 && seg_size > front_seg_size)
			front_seg_size = seg_size;
172 173
	}

174
	do_split = false;
175
split:
176
	*segs = nsegs;
177 178 179 180 181 182 183 184 185 186 187 188

	if (do_split) {
		new = bio_split(bio, sectors, GFP_NOIO, bs);
		if (new)
			bio = new;
	}

	bio->bi_seg_front_size = front_seg_size;
	if (seg_size > bio->bi_seg_back_size)
		bio->bi_seg_back_size = seg_size;

	return do_split ? new : NULL;
189 190 191 192 193
}

void blk_queue_split(struct request_queue *q, struct bio **bio,
		     struct bio_set *bs)
{
194 195
	struct bio *split, *res;
	unsigned nsegs;
196

Adrian Hunter's avatar
Adrian Hunter committed
197 198 199
	switch (bio_op(*bio)) {
	case REQ_OP_DISCARD:
	case REQ_OP_SECURE_ERASE:
200
		split = blk_bio_discard_split(q, *bio, bs, &nsegs);
Adrian Hunter's avatar
Adrian Hunter committed
201
		break;
202 203 204 205
	case REQ_OP_WRITE_ZEROES:
		split = NULL;
		nsegs = (*bio)->bi_phys_segments;
		break;
Adrian Hunter's avatar
Adrian Hunter committed
206
	case REQ_OP_WRITE_SAME:
207
		split = blk_bio_write_same_split(q, *bio, bs, &nsegs);
Adrian Hunter's avatar
Adrian Hunter committed
208 209
		break;
	default:
210
		split = blk_bio_segment_split(q, *bio, q->bio_split, &nsegs);
Adrian Hunter's avatar
Adrian Hunter committed
211 212
		break;
	}
213 214 215 216 217

	/* physical segments can be figured out during splitting */
	res = split ? split : *bio;
	res->bi_phys_segments = nsegs;
	bio_set_flag(res, BIO_SEG_VALID);
218 219

	if (split) {
220
		/* there isn't chance to merge the splitted bio */
221
		split->bi_opf |= REQ_NOMERGE;
222

223
		bio_chain(split, *bio);
224
		trace_block_split(q, split, (*bio)->bi_iter.bi_sector);
225 226 227 228 229 230
		generic_make_request(*bio);
		*bio = split;
	}
}
EXPORT_SYMBOL(blk_queue_split);

231
static unsigned int __blk_recalc_rq_segments(struct request_queue *q,
232 233
					     struct bio *bio,
					     bool no_sg_merge)
234
{
235
	struct bio_vec bv, bvprv = { NULL };
236
	int cluster, prev = 0;
237
	unsigned int seg_size, nr_phys_segs;
238
	struct bio *fbio, *bbio;
239
	struct bvec_iter iter;
240

241 242
	if (!bio)
		return 0;
243

244 245 246 247
	/*
	 * This should probably be returning 0, but blk_add_request_payload()
	 * (Christoph!!!!)
	 */
248 249 250 251 252
	switch (bio_op(bio)) {
	case REQ_OP_DISCARD:
	case REQ_OP_SECURE_ERASE:
	case REQ_OP_WRITE_SAME:
	case REQ_OP_WRITE_ZEROES:
253
		return 1;
254 255 256
	default:
		break;
	}
257

258
	fbio = bio;
259
	cluster = blk_queue_cluster(q);
Mikulas Patocka's avatar
Mikulas Patocka committed
260
	seg_size = 0;
261
	nr_phys_segs = 0;
262
	for_each_bio(bio) {
263
		bio_for_each_segment(bv, bio, iter) {
264 265 266 267 268 269 270
			/*
			 * If SG merging is disabled, each bio vector is
			 * a segment
			 */
			if (no_sg_merge)
				goto new_segment;

271
			if (prev && cluster) {
272
				if (seg_size + bv.bv_len
273
				    > queue_max_segment_size(q))
274
					goto new_segment;
275
				if (!BIOVEC_PHYS_MERGEABLE(&bvprv, &bv))
276
					goto new_segment;
277
				if (!BIOVEC_SEG_BOUNDARY(q, &bvprv, &bv))
278
					goto new_segment;
279

280
				seg_size += bv.bv_len;
281 282 283
				bvprv = bv;
				continue;
			}
284
new_segment:
285 286 287
			if (nr_phys_segs == 1 && seg_size >
			    fbio->bi_seg_front_size)
				fbio->bi_seg_front_size = seg_size;
288

289 290
			nr_phys_segs++;
			bvprv = bv;
291
			prev = 1;
292
			seg_size = bv.bv_len;
293
		}
294
		bbio = bio;
295 296
	}

297 298 299 300
	if (nr_phys_segs == 1 && seg_size > fbio->bi_seg_front_size)
		fbio->bi_seg_front_size = seg_size;
	if (seg_size > bbio->bi_seg_back_size)
		bbio->bi_seg_back_size = seg_size;
301 302 303 304 305 306

	return nr_phys_segs;
}

void blk_recalc_rq_segments(struct request *rq)
{
307 308 309 310 311
	bool no_sg_merge = !!test_bit(QUEUE_FLAG_NO_SG_MERGE,
			&rq->q->queue_flags);

	rq->nr_phys_segments = __blk_recalc_rq_segments(rq->q, rq->bio,
			no_sg_merge);
312 313 314 315
}

void blk_recount_segments(struct request_queue *q, struct bio *bio)
{
316 317 318 319 320 321 322
	unsigned short seg_cnt;

	/* estimate segment number by bi_vcnt for non-cloned bio */
	if (bio_flagged(bio, BIO_CLONED))
		seg_cnt = bio_segments(bio);
	else
		seg_cnt = bio->bi_vcnt;
323

324 325 326
	if (test_bit(QUEUE_FLAG_NO_SG_MERGE, &q->queue_flags) &&
			(seg_cnt < queue_max_segments(q)))
		bio->bi_phys_segments = seg_cnt;
327 328 329 330
	else {
		struct bio *nxt = bio->bi_next;

		bio->bi_next = NULL;
331
		bio->bi_phys_segments = __blk_recalc_rq_segments(q, bio, false);
332 333
		bio->bi_next = nxt;
	}
334

335
	bio_set_flag(bio, BIO_SEG_VALID);
336 337 338 339 340 341
}
EXPORT_SYMBOL(blk_recount_segments);

static int blk_phys_contig_segment(struct request_queue *q, struct bio *bio,
				   struct bio *nxt)
{
342
	struct bio_vec end_bv = { NULL }, nxt_bv;
343

344
	if (!blk_queue_cluster(q))
345 346
		return 0;

347
	if (bio->bi_seg_back_size + nxt->bi_seg_front_size >
348
	    queue_max_segment_size(q))
349 350
		return 0;

351 352 353
	if (!bio_has_data(bio))
		return 1;

354 355
	bio_get_last_bvec(bio, &end_bv);
	bio_get_first_bvec(nxt, &nxt_bv);
356 357

	if (!BIOVEC_PHYS_MERGEABLE(&end_bv, &nxt_bv))
358 359
		return 0;

360
	/*
361
	 * bio and nxt are contiguous in memory; check if the queue allows
362 363
	 * these two to be merged into one
	 */
364
	if (BIOVEC_SEG_BOUNDARY(q, &end_bv, &nxt_bv))
365 366 367 368 369
		return 1;

	return 0;
}

370
static inline void
371
__blk_segment_map_sg(struct request_queue *q, struct bio_vec *bvec,
372
		     struct scatterlist *sglist, struct bio_vec *bvprv,
373 374 375 376 377
		     struct scatterlist **sg, int *nsegs, int *cluster)
{

	int nbytes = bvec->bv_len;

378
	if (*sg && *cluster) {
379 380 381
		if ((*sg)->length + nbytes > queue_max_segment_size(q))
			goto new_segment;

382
		if (!BIOVEC_PHYS_MERGEABLE(bvprv, bvec))
383
			goto new_segment;
384
		if (!BIOVEC_SEG_BOUNDARY(q, bvprv, bvec))
385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402
			goto new_segment;

		(*sg)->length += nbytes;
	} else {
new_segment:
		if (!*sg)
			*sg = sglist;
		else {
			/*
			 * If the driver previously mapped a shorter
			 * list, we could see a termination bit
			 * prematurely unless it fully inits the sg
			 * table on each mapping. We KNOW that there
			 * must be more entries here or the driver
			 * would be buggy, so force clear the
			 * termination bit to avoid doing a full
			 * sg_init_table() in drivers for each command.
			 */
403
			sg_unmark_end(*sg);
404 405 406 407 408 409
			*sg = sg_next(*sg);
		}

		sg_set_page(*sg, bvec->bv_page, nbytes, bvec->bv_offset);
		(*nsegs)++;
	}
410
	*bvprv = *bvec;
411 412
}

413 414 415
static int __blk_bios_map_sg(struct request_queue *q, struct bio *bio,
			     struct scatterlist *sglist,
			     struct scatterlist **sg)
416
{
417
	struct bio_vec bvec, bvprv = { NULL };
418
	struct bvec_iter iter;
419 420 421
	int nsegs, cluster;

	nsegs = 0;
422
	cluster = blk_queue_cluster(q);
423

Adrian Hunter's avatar
Adrian Hunter committed
424 425 426
	switch (bio_op(bio)) {
	case REQ_OP_DISCARD:
	case REQ_OP_SECURE_ERASE:
427
	case REQ_OP_WRITE_ZEROES:
428 429 430 431 432 433 434
		/*
		 * This is a hack - drivers should be neither modifying the
		 * biovec, nor relying on bi_vcnt - but because of
		 * blk_add_request_payload(), a discard bio may or may not have
		 * a payload we need to set up here (thank you Christoph) and
		 * bi_vcnt is really the only way of telling if we need to.
		 */
Adrian Hunter's avatar
Adrian Hunter committed
435 436 437 438
		if (!bio->bi_vcnt)
			return 0;
		/* Fall through */
	case REQ_OP_WRITE_SAME:
439 440 441 442
		*sg = sglist;
		bvec = bio_iovec(bio);
		sg_set_page(*sg, bvec.bv_page, bvec.bv_len, bvec.bv_offset);
		return 1;
Adrian Hunter's avatar
Adrian Hunter committed
443 444
	default:
		break;
445 446 447 448 449 450
	}

	for_each_bio(bio)
		bio_for_each_segment(bvec, bio, iter)
			__blk_segment_map_sg(q, &bvec, sglist, &bvprv, sg,
					     &nsegs, &cluster);
451

452 453 454 455 456 457 458 459 460 461 462 463 464 465 466
	return nsegs;
}

/*
 * map a request to scatterlist, return number of sg entries setup. Caller
 * must make sure sg can hold rq->nr_phys_segments entries
 */
int blk_rq_map_sg(struct request_queue *q, struct request *rq,
		  struct scatterlist *sglist)
{
	struct scatterlist *sg = NULL;
	int nsegs = 0;

	if (rq->bio)
		nsegs = __blk_bios_map_sg(q, rq->bio, sglist, &sg);
467

468
	if (unlikely(rq->rq_flags & RQF_COPY_USER) &&
469 470 471
	    (blk_rq_bytes(rq) & q->dma_pad_mask)) {
		unsigned int pad_len =
			(q->dma_pad_mask & ~blk_rq_bytes(rq)) + 1;
472 473 474 475 476

		sg->length += pad_len;
		rq->extra_len += pad_len;
	}

477
	if (q->dma_drain_size && q->dma_drain_needed(rq)) {
478
		if (op_is_write(req_op(rq)))
479 480
			memset(q->dma_drain_buffer, 0, q->dma_drain_size);

481
		sg_unmark_end(sg);
482 483 484 485 486 487
		sg = sg_next(sg);
		sg_set_page(sg, virt_to_page(q->dma_drain_buffer),
			    q->dma_drain_size,
			    ((unsigned long)q->dma_drain_buffer) &
			    (PAGE_SIZE - 1));
		nsegs++;
488
		rq->extra_len += q->dma_drain_size;
489 490 491 492 493
	}

	if (sg)
		sg_mark_end(sg);

494 495 496 497 498 499
	/*
	 * Something must have been wrong if the figured number of
	 * segment is bigger than number of req's physical segments
	 */
	WARN_ON(nsegs > rq->nr_phys_segments);

500 501 502 503 504 505 506 507 508 509
	return nsegs;
}
EXPORT_SYMBOL(blk_rq_map_sg);

static inline int ll_new_hw_segment(struct request_queue *q,
				    struct request *req,
				    struct bio *bio)
{
	int nr_phys_segs = bio_phys_segments(q, bio);

510 511 512
	if (req->nr_phys_segments + nr_phys_segs > queue_max_segments(q))
		goto no_merge;

513
	if (blk_integrity_merge_bio(q, req, bio) == false)
514
		goto no_merge;
515 516 517 518 519 520 521

	/*
	 * This will form the start of a new hw segment.  Bump both
	 * counters.
	 */
	req->nr_phys_segments += nr_phys_segs;
	return 1;
522 523 524 525 526 527

no_merge:
	req->cmd_flags |= REQ_NOMERGE;
	if (req == q->last_merge)
		q->last_merge = NULL;
	return 0;
528 529 530 531 532
}

int ll_back_merge_fn(struct request_queue *q, struct request *req,
		     struct bio *bio)
{
533 534
	if (req_gap_back_merge(req, bio))
		return 0;
535 536 537
	if (blk_integrity_rq(req) &&
	    integrity_req_gap_back_merge(req, bio))
		return 0;
538
	if (blk_rq_sectors(req) + bio_sectors(bio) >
539
	    blk_rq_get_max_sectors(req, blk_rq_pos(req))) {
540 541 542 543 544
		req->cmd_flags |= REQ_NOMERGE;
		if (req == q->last_merge)
			q->last_merge = NULL;
		return 0;
	}
545
	if (!bio_flagged(req->biotail, BIO_SEG_VALID))
546
		blk_recount_segments(q, req->biotail);
547
	if (!bio_flagged(bio, BIO_SEG_VALID))
548 549 550 551 552
		blk_recount_segments(q, bio);

	return ll_new_hw_segment(q, req, bio);
}

553
int ll_front_merge_fn(struct request_queue *q, struct request *req,
554 555
		      struct bio *bio)
{
556 557 558

	if (req_gap_front_merge(req, bio))
		return 0;
559 560 561
	if (blk_integrity_rq(req) &&
	    integrity_req_gap_front_merge(req, bio))
		return 0;
562
	if (blk_rq_sectors(req) + bio_sectors(bio) >
563
	    blk_rq_get_max_sectors(req, bio->bi_iter.bi_sector)) {
564 565 566 567 568
		req->cmd_flags |= REQ_NOMERGE;
		if (req == q->last_merge)
			q->last_merge = NULL;
		return 0;
	}
569
	if (!bio_flagged(bio, BIO_SEG_VALID))
570
		blk_recount_segments(q, bio);
571
	if (!bio_flagged(req->bio, BIO_SEG_VALID))
572 573 574 575 576
		blk_recount_segments(q, req->bio);

	return ll_new_hw_segment(q, req, bio);
}

577 578 579 580 581 582 583 584 585 586 587
/*
 * blk-mq uses req->special to carry normal driver per-request payload, it
 * does not indicate a prepared command that we cannot merge with.
 */
static bool req_no_special_merge(struct request *req)
{
	struct request_queue *q = req->q;

	return !q->mq_ops && req->special;
}

588 589 590 591
static int ll_merge_requests_fn(struct request_queue *q, struct request *req,
				struct request *next)
{
	int total_phys_segments;
592 593
	unsigned int seg_size =
		req->biotail->bi_seg_back_size + next->bio->bi_seg_front_size;
594 595 596 597 598

	/*
	 * First check if the either of the requests are re-queued
	 * requests.  Can't merge them if they are.
	 */
599
	if (req_no_special_merge(req) || req_no_special_merge(next))
600 601
		return 0;

602
	if (req_gap_back_merge(req, next->bio))
603 604
		return 0;

605 606 607
	/*
	 * Will it become too large?
	 */
608
	if ((blk_rq_sectors(req) + blk_rq_sectors(next)) >
609
	    blk_rq_get_max_sectors(req, blk_rq_pos(req)))
610 611 612
		return 0;

	total_phys_segments = req->nr_phys_segments + next->nr_phys_segments;
613 614 615 616 617
	if (blk_phys_contig_segment(q, req->biotail, next->bio)) {
		if (req->nr_phys_segments == 1)
			req->bio->bi_seg_front_size = seg_size;
		if (next->nr_phys_segments == 1)
			next->biotail->bi_seg_back_size = seg_size;
618
		total_phys_segments--;
619
	}
620

621
	if (total_phys_segments > queue_max_segments(q))
622 623
		return 0;

624
	if (blk_integrity_merge_rq(q, req, next) == false)
625 626
		return 0;

627 628 629 630 631
	/* Merge is OK... */
	req->nr_phys_segments = total_phys_segments;
	return 1;
}

632 633 634 635 636 637 638 639 640 641 642 643 644 645
/**
 * blk_rq_set_mixed_merge - mark a request as mixed merge
 * @rq: request to mark as mixed merge
 *
 * Description:
 *     @rq is about to be mixed merged.  Make sure the attributes
 *     which can be mixed are set in each bio and mark @rq as mixed
 *     merged.
 */
void blk_rq_set_mixed_merge(struct request *rq)
{
	unsigned int ff = rq->cmd_flags & REQ_FAILFAST_MASK;
	struct bio *bio;

646
	if (rq->rq_flags & RQF_MIXED_MERGE)
647 648 649 650 651 652 653 654
		return;

	/*
	 * @rq will no longer represent mixable attributes for all the
	 * contained bios.  It will just track those of the first one.
	 * Distributes the attributs to each bio.
	 */
	for (bio = rq->bio; bio; bio = bio->bi_next) {
655 656 657
		WARN_ON_ONCE((bio->bi_opf & REQ_FAILFAST_MASK) &&
			     (bio->bi_opf & REQ_FAILFAST_MASK) != ff);
		bio->bi_opf |= ff;
658
	}
659
	rq->rq_flags |= RQF_MIXED_MERGE;
660 661
}

662 663 664 665 666 667 668
static void blk_account_io_merge(struct request *req)
{
	if (blk_do_io_stat(req)) {
		struct hd_struct *part;
		int cpu;

		cpu = part_stat_lock();
669
		part = req->part;
670 671

		part_round_stats(cpu, part);
672
		part_dec_in_flight(part, rq_data_dir(req));
673

674
		hd_struct_put(part);
675 676 677 678
		part_stat_unlock();
	}
}

679 680 681 682 683 684 685 686 687
/*
 * Has to be called with the request spinlock acquired
 */
static int attempt_merge(struct request_queue *q, struct request *req,
			  struct request *next)
{
	if (!rq_mergeable(req) || !rq_mergeable(next))
		return 0;

688
	if (req_op(req) != req_op(next))
689 690
		return 0;

691 692 693
	/*
	 * not contiguous
	 */
694
	if (blk_rq_pos(req) + blk_rq_sectors(req) != blk_rq_pos(next))
695 696 697 698
		return 0;

	if (rq_data_dir(req) != rq_data_dir(next)
	    || req->rq_disk != next->rq_disk
699
	    || req_no_special_merge(next))
700 701
		return 0;

702
	if (req_op(req) == REQ_OP_WRITE_SAME &&
703 704 705
	    !blk_write_same_mergeable(req->bio, next->bio))
		return 0;

706 707 708 709 710 711 712 713 714
	/*
	 * If we are allowed to merge, then append bio list
	 * from next to rq and release next. merge_requests_fn
	 * will have updated segment counts, update sector
	 * counts here.
	 */
	if (!ll_merge_requests_fn(q, req, next))
		return 0;

715 716 717 718 719 720
	/*
	 * If failfast settings disagree or any of the two is already
	 * a mixed merge, mark both as mixed before proceeding.  This
	 * makes sure that all involved bios have mixable attributes
	 * set properly.
	 */
721
	if (((req->rq_flags | next->rq_flags) & RQF_MIXED_MERGE) ||
722 723 724 725 726 727
	    (req->cmd_flags & REQ_FAILFAST_MASK) !=
	    (next->cmd_flags & REQ_FAILFAST_MASK)) {
		blk_rq_set_mixed_merge(req);
		blk_rq_set_mixed_merge(next);
	}

728 729 730 731 732 733 734 735 736 737 738 739
	/*
	 * At this point we have either done a back merge
	 * or front merge. We need the smaller start_time of
	 * the merged requests to be the current request
	 * for accounting purposes.
	 */
	if (time_after(req->start_time, next->start_time))
		req->start_time = next->start_time;

	req->biotail->bi_next = next->bio;
	req->biotail = next->biotail;

740
	req->__data_len += blk_rq_bytes(next);
741 742 743

	elv_merge_requests(q, req, next);

744 745 746 747
	/*
	 * 'next' is going away, so update stats accordingly
	 */
	blk_account_io_merge(next);
748 749

	req->ioprio = ioprio_best(req->ioprio, next->ioprio);
750 751
	if (blk_rq_cpu_valid(next))
		req->cpu = next->cpu;
752

753 754
	/* owner-ship of bio passed from next to req */
	next->bio = NULL;
755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777
	__blk_put_request(q, next);
	return 1;
}

int attempt_back_merge(struct request_queue *q, struct request *rq)
{
	struct request *next = elv_latter_request(q, rq);

	if (next)
		return attempt_merge(q, rq, next);

	return 0;
}

int attempt_front_merge(struct request_queue *q, struct request *rq)
{
	struct request *prev = elv_former_request(q, rq);

	if (prev)
		return attempt_merge(q, prev, rq);

	return 0;
}
778 779 780 781

int blk_attempt_req_merge(struct request_queue *q, struct request *rq,
			  struct request *next)
{
782 783 784 785 786 787
	struct elevator_queue *e = q->elevator;

	if (e->type->ops.elevator_allow_rq_merge_fn)
		if (!e->type->ops.elevator_allow_rq_merge_fn(q, rq, next))
			return 0;

788 789
	return attempt_merge(q, rq, next);
}
790 791 792

bool blk_rq_merge_ok(struct request *rq, struct bio *bio)
{
793
	if (!rq_mergeable(rq) || !bio_mergeable(bio))
794 795
		return false;

796
	if (req_op(rq) != bio_op(bio))
797 798
		return false;

799 800 801 802 803
	/* different data direction or already started, don't merge */
	if (bio_data_dir(bio) != rq_data_dir(rq))
		return false;

	/* must be same device and not a special request */
804
	if (rq->rq_disk != bio->bi_bdev->bd_disk || req_no_special_merge(rq))
805 806 807
		return false;

	/* only merge integrity protected bio into ditto rq */
808
	if (blk_integrity_merge_bio(rq->q, rq, bio) == false)
809 810
		return false;

811
	/* must be using the same buffer */
812
	if (req_op(rq) == REQ_OP_WRITE_SAME &&
813 814 815
	    !blk_write_same_mergeable(rq->bio, bio))
		return false;

816 817 818 819 820
	return true;
}

int blk_try_merge(struct request *rq, struct bio *bio)
{
821
	if (blk_rq_pos(rq) + blk_rq_sectors(rq) == bio->bi_iter.bi_sector)
822
		return ELEVATOR_BACK_MERGE;
823
	else if (blk_rq_pos(rq) - bio_sectors(bio) == bio->bi_iter.bi_sector)
824 825 826
		return ELEVATOR_FRONT_MERGE;
	return ELEVATOR_NO_MERGE;
}