blk-merge.c 19.6 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
201 202
		break;
	case REQ_OP_WRITE_SAME:
203
		split = blk_bio_write_same_split(q, *bio, bs, &nsegs);
Adrian Hunter's avatar
Adrian Hunter committed
204 205
		break;
	default:
206
		split = blk_bio_segment_split(q, *bio, q->bio_split, &nsegs);
Adrian Hunter's avatar
Adrian Hunter committed
207 208
		break;
	}
209 210 211 212 213

	/* physical segments can be figured out during splitting */
	res = split ? split : *bio;
	res->bi_phys_segments = nsegs;
	bio_set_flag(res, BIO_SEG_VALID);
214 215

	if (split) {
Ming Lei's avatar
Ming Lei committed
216
		/* there isn't chance to merge the splitted bio */
Jens Axboe's avatar
Jens Axboe committed
217
		split->bi_opf |= REQ_NOMERGE;
Ming Lei's avatar
Ming Lei committed
218

219
		bio_chain(split, *bio);
220
		trace_block_split(q, split, (*bio)->bi_iter.bi_sector);
221 222 223 224 225 226
		generic_make_request(*bio);
		*bio = split;
	}
}
EXPORT_SYMBOL(blk_queue_split);

227
static unsigned int __blk_recalc_rq_segments(struct request_queue *q,
Ming Lei's avatar
Ming Lei committed
228 229
					     struct bio *bio,
					     bool no_sg_merge)
230
{
231
	struct bio_vec bv, bvprv = { NULL };
232
	int cluster, prev = 0;
233
	unsigned int seg_size, nr_phys_segs;
234
	struct bio *fbio, *bbio;
235
	struct bvec_iter iter;
236

237 238
	if (!bio)
		return 0;
239

240 241 242 243
	/*
	 * This should probably be returning 0, but blk_add_request_payload()
	 * (Christoph!!!!)
	 */
Adrian Hunter's avatar
Adrian Hunter committed
244
	if (bio_op(bio) == REQ_OP_DISCARD || bio_op(bio) == REQ_OP_SECURE_ERASE)
245 246
		return 1;

247
	if (bio_op(bio) == REQ_OP_WRITE_SAME)
248 249
		return 1;

250
	fbio = bio;
251
	cluster = blk_queue_cluster(q);
Mikulas Patocka's avatar
Mikulas Patocka committed
252
	seg_size = 0;
253
	nr_phys_segs = 0;
254
	for_each_bio(bio) {
255
		bio_for_each_segment(bv, bio, iter) {
256 257 258 259 260 261 262
			/*
			 * If SG merging is disabled, each bio vector is
			 * a segment
			 */
			if (no_sg_merge)
				goto new_segment;

263
			if (prev && cluster) {
264
				if (seg_size + bv.bv_len
265
				    > queue_max_segment_size(q))
266
					goto new_segment;
267
				if (!BIOVEC_PHYS_MERGEABLE(&bvprv, &bv))
268
					goto new_segment;
269
				if (!BIOVEC_SEG_BOUNDARY(q, &bvprv, &bv))
270
					goto new_segment;
271

272
				seg_size += bv.bv_len;
273 274 275
				bvprv = bv;
				continue;
			}
276
new_segment:
277 278 279
			if (nr_phys_segs == 1 && seg_size >
			    fbio->bi_seg_front_size)
				fbio->bi_seg_front_size = seg_size;
280

281 282
			nr_phys_segs++;
			bvprv = bv;
283
			prev = 1;
284
			seg_size = bv.bv_len;
285
		}
286
		bbio = bio;
287 288
	}

289 290 291 292
	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;
293 294 295 296 297 298

	return nr_phys_segs;
}

void blk_recalc_rq_segments(struct request *rq)
{
Ming Lei's avatar
Ming Lei committed
299 300 301 302 303
	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);
304 305 306 307
}

void blk_recount_segments(struct request_queue *q, struct bio *bio)
{
308 309 310 311 312 313 314
	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;
315

316 317 318
	if (test_bit(QUEUE_FLAG_NO_SG_MERGE, &q->queue_flags) &&
			(seg_cnt < queue_max_segments(q)))
		bio->bi_phys_segments = seg_cnt;
319 320 321 322
	else {
		struct bio *nxt = bio->bi_next;

		bio->bi_next = NULL;
323
		bio->bi_phys_segments = __blk_recalc_rq_segments(q, bio, false);
324 325
		bio->bi_next = nxt;
	}
326

327
	bio_set_flag(bio, BIO_SEG_VALID);
328 329 330 331 332 333
}
EXPORT_SYMBOL(blk_recount_segments);

static int blk_phys_contig_segment(struct request_queue *q, struct bio *bio,
				   struct bio *nxt)
{
334
	struct bio_vec end_bv = { NULL }, nxt_bv;
335

336
	if (!blk_queue_cluster(q))
337 338
		return 0;

339
	if (bio->bi_seg_back_size + nxt->bi_seg_front_size >
340
	    queue_max_segment_size(q))
341 342
		return 0;

343 344 345
	if (!bio_has_data(bio))
		return 1;

346 347
	bio_get_last_bvec(bio, &end_bv);
	bio_get_first_bvec(nxt, &nxt_bv);
348 349

	if (!BIOVEC_PHYS_MERGEABLE(&end_bv, &nxt_bv))
350 351
		return 0;

352
	/*
353
	 * bio and nxt are contiguous in memory; check if the queue allows
354 355
	 * these two to be merged into one
	 */
356
	if (BIOVEC_SEG_BOUNDARY(q, &end_bv, &nxt_bv))
357 358 359 360 361
		return 1;

	return 0;
}

362
static inline void
363
__blk_segment_map_sg(struct request_queue *q, struct bio_vec *bvec,
364
		     struct scatterlist *sglist, struct bio_vec *bvprv,
365 366 367 368 369
		     struct scatterlist **sg, int *nsegs, int *cluster)
{

	int nbytes = bvec->bv_len;

370
	if (*sg && *cluster) {
371 372 373
		if ((*sg)->length + nbytes > queue_max_segment_size(q))
			goto new_segment;

374
		if (!BIOVEC_PHYS_MERGEABLE(bvprv, bvec))
375
			goto new_segment;
376
		if (!BIOVEC_SEG_BOUNDARY(q, bvprv, bvec))
377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394
			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.
			 */
395
			sg_unmark_end(*sg);
396 397 398 399 400 401
			*sg = sg_next(*sg);
		}

		sg_set_page(*sg, bvec->bv_page, nbytes, bvec->bv_offset);
		(*nsegs)++;
	}
402
	*bvprv = *bvec;
403 404
}

405 406 407
static int __blk_bios_map_sg(struct request_queue *q, struct bio *bio,
			     struct scatterlist *sglist,
			     struct scatterlist **sg)
408
{
409
	struct bio_vec bvec, bvprv = { NULL };
410
	struct bvec_iter iter;
411 412 413
	int nsegs, cluster;

	nsegs = 0;
414
	cluster = blk_queue_cluster(q);
415

Adrian Hunter's avatar
Adrian Hunter committed
416 417 418
	switch (bio_op(bio)) {
	case REQ_OP_DISCARD:
	case REQ_OP_SECURE_ERASE:
419 420 421 422 423 424 425
		/*
		 * 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
426 427 428 429
		if (!bio->bi_vcnt)
			return 0;
		/* Fall through */
	case REQ_OP_WRITE_SAME:
430 431 432 433
		*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
434 435
	default:
		break;
436 437 438 439 440 441
	}

	for_each_bio(bio)
		bio_for_each_segment(bvec, bio, iter)
			__blk_segment_map_sg(q, &bvec, sglist, &bvprv, sg,
					     &nsegs, &cluster);
442

443 444 445 446 447 448 449 450 451 452 453 454 455 456 457
	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);
458

459
	if (unlikely(rq->rq_flags & RQF_COPY_USER) &&
460 461 462
	    (blk_rq_bytes(rq) & q->dma_pad_mask)) {
		unsigned int pad_len =
			(q->dma_pad_mask & ~blk_rq_bytes(rq)) + 1;
463 464 465 466 467

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

468
	if (q->dma_drain_size && q->dma_drain_needed(rq)) {
469
		if (op_is_write(req_op(rq)))
470 471
			memset(q->dma_drain_buffer, 0, q->dma_drain_size);

472
		sg_unmark_end(sg);
473 474 475 476 477 478
		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++;
479
		rq->extra_len += q->dma_drain_size;
480 481 482 483 484
	}

	if (sg)
		sg_mark_end(sg);

485 486 487 488 489 490
	/*
	 * 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);

491 492 493 494 495 496 497 498 499 500
	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);

501 502 503
	if (req->nr_phys_segments + nr_phys_segs > queue_max_segments(q))
		goto no_merge;

504
	if (blk_integrity_merge_bio(q, req, bio) == false)
505
		goto no_merge;
506 507 508 509 510 511 512

	/*
	 * This will form the start of a new hw segment.  Bump both
	 * counters.
	 */
	req->nr_phys_segments += nr_phys_segs;
	return 1;
513 514 515 516 517 518

no_merge:
	req->cmd_flags |= REQ_NOMERGE;
	if (req == q->last_merge)
		q->last_merge = NULL;
	return 0;
519 520 521 522 523
}

int ll_back_merge_fn(struct request_queue *q, struct request *req,
		     struct bio *bio)
{
524 525
	if (req_gap_back_merge(req, bio))
		return 0;
526 527 528
	if (blk_integrity_rq(req) &&
	    integrity_req_gap_back_merge(req, bio))
		return 0;
529
	if (blk_rq_sectors(req) + bio_sectors(bio) >
Damien Le Moal's avatar
Damien Le Moal committed
530
	    blk_rq_get_max_sectors(req, blk_rq_pos(req))) {
531 532 533 534 535
		req->cmd_flags |= REQ_NOMERGE;
		if (req == q->last_merge)
			q->last_merge = NULL;
		return 0;
	}
536
	if (!bio_flagged(req->biotail, BIO_SEG_VALID))
537
		blk_recount_segments(q, req->biotail);
538
	if (!bio_flagged(bio, BIO_SEG_VALID))
539 540 541 542 543
		blk_recount_segments(q, bio);

	return ll_new_hw_segment(q, req, bio);
}

544
int ll_front_merge_fn(struct request_queue *q, struct request *req,
545 546
		      struct bio *bio)
{
547 548 549

	if (req_gap_front_merge(req, bio))
		return 0;
550 551 552
	if (blk_integrity_rq(req) &&
	    integrity_req_gap_front_merge(req, bio))
		return 0;
553
	if (blk_rq_sectors(req) + bio_sectors(bio) >
Damien Le Moal's avatar
Damien Le Moal committed
554
	    blk_rq_get_max_sectors(req, bio->bi_iter.bi_sector)) {
555 556 557 558 559
		req->cmd_flags |= REQ_NOMERGE;
		if (req == q->last_merge)
			q->last_merge = NULL;
		return 0;
	}
560
	if (!bio_flagged(bio, BIO_SEG_VALID))
561
		blk_recount_segments(q, bio);
562
	if (!bio_flagged(req->bio, BIO_SEG_VALID))
563 564 565 566 567
		blk_recount_segments(q, req->bio);

	return ll_new_hw_segment(q, req, bio);
}

568 569 570 571 572 573 574 575 576 577 578
/*
 * 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;
}

579 580 581 582
static int ll_merge_requests_fn(struct request_queue *q, struct request *req,
				struct request *next)
{
	int total_phys_segments;
583 584
	unsigned int seg_size =
		req->biotail->bi_seg_back_size + next->bio->bi_seg_front_size;
585 586 587 588 589

	/*
	 * First check if the either of the requests are re-queued
	 * requests.  Can't merge them if they are.
	 */
590
	if (req_no_special_merge(req) || req_no_special_merge(next))
591 592
		return 0;

593
	if (req_gap_back_merge(req, next->bio))
594 595
		return 0;

596 597 598
	/*
	 * Will it become too large?
	 */
599
	if ((blk_rq_sectors(req) + blk_rq_sectors(next)) >
Damien Le Moal's avatar
Damien Le Moal committed
600
	    blk_rq_get_max_sectors(req, blk_rq_pos(req)))
601 602 603
		return 0;

	total_phys_segments = req->nr_phys_segments + next->nr_phys_segments;
604 605 606 607 608
	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;
609
		total_phys_segments--;
610
	}
611

612
	if (total_phys_segments > queue_max_segments(q))
613 614
		return 0;

615
	if (blk_integrity_merge_rq(q, req, next) == false)
616 617
		return 0;

618 619 620 621 622
	/* Merge is OK... */
	req->nr_phys_segments = total_phys_segments;
	return 1;
}

623 624 625 626 627 628 629 630 631 632 633 634 635 636
/**
 * 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;

637
	if (rq->rq_flags & RQF_MIXED_MERGE)
638 639 640 641 642 643 644 645
		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) {
Jens Axboe's avatar
Jens Axboe committed
646 647 648
		WARN_ON_ONCE((bio->bi_opf & REQ_FAILFAST_MASK) &&
			     (bio->bi_opf & REQ_FAILFAST_MASK) != ff);
		bio->bi_opf |= ff;
649
	}
650
	rq->rq_flags |= RQF_MIXED_MERGE;
651 652
}

653 654 655 656 657 658 659
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();
660
		part = req->part;
661 662

		part_round_stats(cpu, part);
663
		part_dec_in_flight(part, rq_data_dir(req));
664

665
		hd_struct_put(part);
666 667 668 669
		part_stat_unlock();
	}
}

670 671 672 673 674 675 676 677 678
/*
 * 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;

679
	if (req_op(req) != req_op(next))
680 681
		return 0;

682 683 684
	/*
	 * not contiguous
	 */
685
	if (blk_rq_pos(req) + blk_rq_sectors(req) != blk_rq_pos(next))
686 687 688 689
		return 0;

	if (rq_data_dir(req) != rq_data_dir(next)
	    || req->rq_disk != next->rq_disk
690
	    || req_no_special_merge(next))
691 692
		return 0;

693
	if (req_op(req) == REQ_OP_WRITE_SAME &&
694 695 696
	    !blk_write_same_mergeable(req->bio, next->bio))
		return 0;

697 698 699 700 701 702 703 704 705
	/*
	 * 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;

706 707 708 709 710 711
	/*
	 * 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.
	 */
712
	if (((req->rq_flags | next->rq_flags) & RQF_MIXED_MERGE) ||
713 714 715 716 717 718
	    (req->cmd_flags & REQ_FAILFAST_MASK) !=
	    (next->cmd_flags & REQ_FAILFAST_MASK)) {
		blk_rq_set_mixed_merge(req);
		blk_rq_set_mixed_merge(next);
	}

719 720 721 722 723 724 725 726 727 728 729 730
	/*
	 * 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;

731
	req->__data_len += blk_rq_bytes(next);
732 733 734

	elv_merge_requests(q, req, next);

735 736 737 738
	/*
	 * 'next' is going away, so update stats accordingly
	 */
	blk_account_io_merge(next);
739 740

	req->ioprio = ioprio_best(req->ioprio, next->ioprio);
741 742
	if (blk_rq_cpu_valid(next))
		req->cpu = next->cpu;
743

744 745
	/* owner-ship of bio passed from next to req */
	next->bio = NULL;
746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768
	__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;
}
769 770 771 772

int blk_attempt_req_merge(struct request_queue *q, struct request *rq,
			  struct request *next)
{
773 774 775 776 777 778
	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;

779 780
	return attempt_merge(q, rq, next);
}
781 782 783

bool blk_rq_merge_ok(struct request *rq, struct bio *bio)
{
784
	if (!rq_mergeable(rq) || !bio_mergeable(bio))
785 786
		return false;

787
	if (req_op(rq) != bio_op(bio))
788 789
		return false;

790 791 792 793 794
	/* 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 */
795
	if (rq->rq_disk != bio->bi_bdev->bd_disk || req_no_special_merge(rq))
796 797 798
		return false;

	/* only merge integrity protected bio into ditto rq */
799
	if (blk_integrity_merge_bio(rq->q, rq, bio) == false)
800 801
		return false;

802
	/* must be using the same buffer */
803
	if (req_op(rq) == REQ_OP_WRITE_SAME &&
804 805 806
	    !blk_write_same_mergeable(rq->bio, bio))
		return false;

807 808 809 810 811
	return true;
}

int blk_try_merge(struct request *rq, struct bio *bio)
{
812
	if (blk_rq_pos(rq) + blk_rq_sectors(rq) == bio->bi_iter.bi_sector)
813
		return ELEVATOR_BACK_MERGE;
814
	else if (blk_rq_pos(rq) - bio_sectors(bio) == bio->bi_iter.bi_sector)
815 816 817
		return ELEVATOR_FRONT_MERGE;
	return ELEVATOR_NO_MERGE;
}